{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 100.0,
  "global_step": 29007,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 26.71628189086914,
      "learning_rate": 2.2962112514351324e-08,
      "loss": 1.9941,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 16.78429412841797,
      "learning_rate": 4.592422502870265e-08,
      "loss": 1.5063,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.47647476196289,
      "learning_rate": 6.888633754305396e-08,
      "loss": 1.2042,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.848700523376465,
      "learning_rate": 9.18484500574053e-08,
      "loss": 1.1134,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.084993362426758,
      "learning_rate": 1.1481056257175662e-07,
      "loss": 1.2159,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.88464641571045,
      "learning_rate": 1.3777267508610792e-07,
      "loss": 1.1423,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.53444766998291,
      "learning_rate": 1.6073478760045926e-07,
      "loss": 1.0687,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.6861724853515625,
      "learning_rate": 1.836969001148106e-07,
      "loss": 1.0629,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.3049774169921875,
      "learning_rate": 2.066590126291619e-07,
      "loss": 1.0918,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.297573566436768,
      "learning_rate": 2.2962112514351323e-07,
      "loss": 0.9468,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.76364803314209,
      "learning_rate": 2.5258323765786454e-07,
      "loss": 1.0869,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.184765815734863,
      "learning_rate": 2.7554535017221585e-07,
      "loss": 1.1654,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.748183250427246,
      "learning_rate": 2.9850746268656716e-07,
      "loss": 1.015,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.695939540863037,
      "learning_rate": 3.214695752009185e-07,
      "loss": 1.1769,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.036831855773926,
      "learning_rate": 3.444316877152698e-07,
      "loss": 1.1219,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.6727423667907715,
      "learning_rate": 3.673938002296212e-07,
      "loss": 1.0047,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.77660608291626,
      "learning_rate": 3.9035591274397244e-07,
      "loss": 1.1293,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.394479751586914,
      "learning_rate": 4.133180252583238e-07,
      "loss": 0.9037,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.043769836425781,
      "learning_rate": 4.362801377726751e-07,
      "loss": 1.1386,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.649626731872559,
      "learning_rate": 4.5924225028702647e-07,
      "loss": 1.1798,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.381901741027832,
      "learning_rate": 4.822043628013777e-07,
      "loss": 1.0303,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.945267677307129,
      "learning_rate": 5.051664753157291e-07,
      "loss": 1.0209,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.009099006652832,
      "learning_rate": 5.281285878300804e-07,
      "loss": 0.9182,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.6060662269592285,
      "learning_rate": 5.510907003444317e-07,
      "loss": 1.069,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.094696998596191,
      "learning_rate": 5.740528128587831e-07,
      "loss": 0.9135,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.301419734954834,
      "learning_rate": 5.970149253731343e-07,
      "loss": 1.0017,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.629584789276123,
      "learning_rate": 6.199770378874858e-07,
      "loss": 1.1462,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.682167053222656,
      "learning_rate": 6.42939150401837e-07,
      "loss": 0.9178,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.811071872711182,
      "learning_rate": 6.659012629161883e-07,
      "loss": 0.9836,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.814019203186035,
      "learning_rate": 6.888633754305397e-07,
      "loss": 1.1149,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.98969841003418,
      "learning_rate": 7.11825487944891e-07,
      "loss": 1.1128,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.2680463790893555,
      "learning_rate": 7.347876004592424e-07,
      "loss": 1.0309,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.374821662902832,
      "learning_rate": 7.577497129735936e-07,
      "loss": 1.0143,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.766819953918457,
      "learning_rate": 7.807118254879449e-07,
      "loss": 0.9073,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.188901424407959,
      "learning_rate": 8.036739380022963e-07,
      "loss": 1.0775,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.861509799957275,
      "learning_rate": 8.266360505166476e-07,
      "loss": 0.962,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.696706295013428,
      "learning_rate": 8.49598163030999e-07,
      "loss": 1.0072,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.516738414764404,
      "learning_rate": 8.725602755453502e-07,
      "loss": 1.0411,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.007785797119141,
      "learning_rate": 8.955223880597015e-07,
      "loss": 0.9705,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.6335554122924805,
      "learning_rate": 9.184845005740529e-07,
      "loss": 0.8498,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.7292680740356445,
      "learning_rate": 9.414466130884042e-07,
      "loss": 0.9443,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.779412269592285,
      "learning_rate": 9.644087256027554e-07,
      "loss": 0.996,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.594991207122803,
      "learning_rate": 9.87370838117107e-07,
      "loss": 0.9455,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.282378196716309,
      "learning_rate": 1.0103329506314582e-06,
      "loss": 1.035,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.200046539306641,
      "learning_rate": 1.0332950631458096e-06,
      "loss": 0.8628,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.380337715148926,
      "learning_rate": 1.0562571756601609e-06,
      "loss": 0.9745,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.719575881958008,
      "learning_rate": 1.0792192881745121e-06,
      "loss": 0.8293,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.880608081817627,
      "learning_rate": 1.1021814006888634e-06,
      "loss": 0.9528,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.818783283233643,
      "learning_rate": 1.1251435132032149e-06,
      "loss": 0.9331,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.140171051025391,
      "learning_rate": 1.1481056257175661e-06,
      "loss": 1.0833,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 10.210826873779297,
      "learning_rate": 1.1710677382319174e-06,
      "loss": 0.9354,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.898293495178223,
      "learning_rate": 1.1940298507462686e-06,
      "loss": 0.9687,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.7199530601501465,
      "learning_rate": 1.21699196326062e-06,
      "loss": 0.8657,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.751926898956299,
      "learning_rate": 1.2399540757749716e-06,
      "loss": 1.0665,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.927700042724609,
      "learning_rate": 1.2629161882893226e-06,
      "loss": 1.1547,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.7648491859436035,
      "learning_rate": 1.285878300803674e-06,
      "loss": 0.8097,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2445902824401855,
      "learning_rate": 1.3088404133180255e-06,
      "loss": 0.9221,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.738978385925293,
      "learning_rate": 1.3318025258323766e-06,
      "loss": 0.9699,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.977999210357666,
      "learning_rate": 1.354764638346728e-06,
      "loss": 0.9387,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.928709506988525,
      "learning_rate": 1.3777267508610793e-06,
      "loss": 0.9629,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.800056457519531,
      "learning_rate": 1.4006888633754306e-06,
      "loss": 0.8731,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.0732421875,
      "learning_rate": 1.423650975889782e-06,
      "loss": 0.8511,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.649219989776611,
      "learning_rate": 1.4466130884041333e-06,
      "loss": 0.9849,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.85325288772583,
      "learning_rate": 1.4695752009184847e-06,
      "loss": 0.9608,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.762007713317871,
      "learning_rate": 1.4925373134328358e-06,
      "loss": 1.1336,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.1543192863464355,
      "learning_rate": 1.5154994259471873e-06,
      "loss": 0.9327,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.998604774475098,
      "learning_rate": 1.5384615384615387e-06,
      "loss": 0.9572,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.751152038574219,
      "learning_rate": 1.5614236509758898e-06,
      "loss": 0.9537,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.280971527099609,
      "learning_rate": 1.5843857634902412e-06,
      "loss": 1.1008,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.961222171783447,
      "learning_rate": 1.6073478760045927e-06,
      "loss": 0.8409,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.09221076965332,
      "learning_rate": 1.630309988518944e-06,
      "loss": 1.2014,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.257001876831055,
      "learning_rate": 1.6532721010332952e-06,
      "loss": 0.931,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.5343756675720215,
      "learning_rate": 1.6762342135476465e-06,
      "loss": 0.8626,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.760910511016846,
      "learning_rate": 1.699196326061998e-06,
      "loss": 0.9315,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.142362117767334,
      "learning_rate": 1.7221584385763494e-06,
      "loss": 0.8208,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.173382759094238,
      "learning_rate": 1.7451205510907004e-06,
      "loss": 0.7955,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.494527339935303,
      "learning_rate": 1.768082663605052e-06,
      "loss": 0.8514,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.202250003814697,
      "learning_rate": 1.791044776119403e-06,
      "loss": 0.9999,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.876047611236572,
      "learning_rate": 1.8140068886337544e-06,
      "loss": 0.945,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.985454559326172,
      "learning_rate": 1.8369690011481059e-06,
      "loss": 0.8383,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.894712448120117,
      "learning_rate": 1.8599311136624571e-06,
      "loss": 0.7869,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.981006622314453,
      "learning_rate": 1.8828932261768084e-06,
      "loss": 0.8813,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.62688684463501,
      "learning_rate": 1.9058553386911596e-06,
      "loss": 0.7533,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.706407070159912,
      "learning_rate": 1.928817451205511e-06,
      "loss": 0.9266,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.2817511558532715,
      "learning_rate": 1.9517795637198626e-06,
      "loss": 0.8622,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.40199613571167,
      "learning_rate": 1.974741676234214e-06,
      "loss": 0.8388,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.379444122314453,
      "learning_rate": 1.997703788748565e-06,
      "loss": 0.7892,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.459906101226807,
      "learning_rate": 2.0206659012629163e-06,
      "loss": 0.9625,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.7295427322387695,
      "learning_rate": 2.0436280137772676e-06,
      "loss": 1.0147,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.745770454406738,
      "learning_rate": 2.0665901262916193e-06,
      "loss": 0.9609,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.187803268432617,
      "learning_rate": 2.08955223880597e-06,
      "loss": 1.0492,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.139333248138428,
      "learning_rate": 2.1125143513203218e-06,
      "loss": 0.9652,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.724826335906982,
      "learning_rate": 2.135476463834673e-06,
      "loss": 0.8548,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.4161787033081055,
      "learning_rate": 2.1584385763490243e-06,
      "loss": 1.0343,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.0141921043396,
      "learning_rate": 2.1814006888633755e-06,
      "loss": 0.9935,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.008433818817139,
      "learning_rate": 2.204362801377727e-06,
      "loss": 1.0488,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.17753267288208,
      "learning_rate": 2.2273249138920785e-06,
      "loss": 0.9284,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.694668292999268,
      "learning_rate": 2.2502870264064297e-06,
      "loss": 0.6686,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.8419718742370605,
      "learning_rate": 2.273249138920781e-06,
      "loss": 0.9933,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.971826553344727,
      "learning_rate": 2.2962112514351322e-06,
      "loss": 0.764,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.31289005279541,
      "learning_rate": 2.3191733639494835e-06,
      "loss": 0.9285,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.838889122009277,
      "learning_rate": 2.3421354764638347e-06,
      "loss": 0.8421,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.771417140960693,
      "learning_rate": 2.3650975889781864e-06,
      "loss": 0.903,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.961888790130615,
      "learning_rate": 2.3880597014925373e-06,
      "loss": 1.0685,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.045954704284668,
      "learning_rate": 2.411021814006889e-06,
      "loss": 0.8204,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.988632678985596,
      "learning_rate": 2.43398392652124e-06,
      "loss": 0.9573,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.253096103668213,
      "learning_rate": 2.4569460390355914e-06,
      "loss": 0.8041,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.932774543762207,
      "learning_rate": 2.479908151549943e-06,
      "loss": 0.9826,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.466474533081055,
      "learning_rate": 2.5028702640642944e-06,
      "loss": 0.8269,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.720095157623291,
      "learning_rate": 2.525832376578645e-06,
      "loss": 0.9275,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1351776123046875,
      "learning_rate": 2.5487944890929965e-06,
      "loss": 0.9255,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.171075344085693,
      "learning_rate": 2.571756601607348e-06,
      "loss": 0.8903,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.097987174987793,
      "learning_rate": 2.5947187141216994e-06,
      "loss": 1.0064,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.863966941833496,
      "learning_rate": 2.617680826636051e-06,
      "loss": 0.8269,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.3440961837768555,
      "learning_rate": 2.640642939150402e-06,
      "loss": 0.908,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.45919942855835,
      "learning_rate": 2.663605051664753e-06,
      "loss": 0.8261,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.0509934425354,
      "learning_rate": 2.686567164179105e-06,
      "loss": 0.8822,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.999479293823242,
      "learning_rate": 2.709529276693456e-06,
      "loss": 0.8613,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.956490993499756,
      "learning_rate": 2.7324913892078073e-06,
      "loss": 0.9707,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.549967288970947,
      "learning_rate": 2.7554535017221586e-06,
      "loss": 0.8351,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6750874519348145,
      "learning_rate": 2.77841561423651e-06,
      "loss": 0.8373,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.1035637855529785,
      "learning_rate": 2.801377726750861e-06,
      "loss": 0.7914,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.397061824798584,
      "learning_rate": 2.8243398392652128e-06,
      "loss": 0.8062,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.321897029876709,
      "learning_rate": 2.847301951779564e-06,
      "loss": 1.0021,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.947620868682861,
      "learning_rate": 2.870264064293915e-06,
      "loss": 0.7207,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.999775409698486,
      "learning_rate": 2.8932261768082666e-06,
      "loss": 0.9247,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.901172161102295,
      "learning_rate": 2.916188289322618e-06,
      "loss": 0.8538,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.471467018127441,
      "learning_rate": 2.9391504018369695e-06,
      "loss": 0.897,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.691971302032471,
      "learning_rate": 2.9621125143513207e-06,
      "loss": 0.8501,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.745563983917236,
      "learning_rate": 2.9850746268656716e-06,
      "loss": 0.8806,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.993727684020996,
      "learning_rate": 3.0080367393800232e-06,
      "loss": 0.8333,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.664424419403076,
      "learning_rate": 3.0309988518943745e-06,
      "loss": 0.8648,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.089878559112549,
      "learning_rate": 3.0539609644087258e-06,
      "loss": 0.7313,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.998203754425049,
      "learning_rate": 3.0769230769230774e-06,
      "loss": 0.8479,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.614637851715088,
      "learning_rate": 3.0998851894374283e-06,
      "loss": 0.8351,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.965922832489014,
      "learning_rate": 3.1228473019517795e-06,
      "loss": 0.8613,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.175832748413086,
      "learning_rate": 3.145809414466131e-06,
      "loss": 0.989,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.490781784057617,
      "learning_rate": 3.1687715269804825e-06,
      "loss": 1.0627,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.091092586517334,
      "learning_rate": 3.191733639494834e-06,
      "loss": 0.7785,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.73376989364624,
      "learning_rate": 3.2146957520091854e-06,
      "loss": 0.7821,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.7888922691345215,
      "learning_rate": 3.2376578645235362e-06,
      "loss": 0.9042,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.244744300842285,
      "learning_rate": 3.260619977037888e-06,
      "loss": 0.9861,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.625418186187744,
      "learning_rate": 3.283582089552239e-06,
      "loss": 0.7599,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.109224319458008,
      "learning_rate": 3.3065442020665904e-06,
      "loss": 0.8255,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.629339694976807,
      "learning_rate": 3.329506314580942e-06,
      "loss": 0.9326,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.896908760070801,
      "learning_rate": 3.352468427095293e-06,
      "loss": 0.9126,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.903641700744629,
      "learning_rate": 3.375430539609644e-06,
      "loss": 0.788,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.063019752502441,
      "learning_rate": 3.398392652123996e-06,
      "loss": 1.0035,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.56251859664917,
      "learning_rate": 3.421354764638347e-06,
      "loss": 0.7922,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3656487464904785,
      "learning_rate": 3.4443168771526988e-06,
      "loss": 0.9158,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.403493881225586,
      "learning_rate": 3.4672789896670496e-06,
      "loss": 0.9129,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.533061981201172,
      "learning_rate": 3.490241102181401e-06,
      "loss": 0.9737,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.7049665451049805,
      "learning_rate": 3.513203214695752e-06,
      "loss": 0.9111,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.656744480133057,
      "learning_rate": 3.536165327210104e-06,
      "loss": 1.0232,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.676136016845703,
      "learning_rate": 3.559127439724455e-06,
      "loss": 0.7491,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.259581565856934,
      "learning_rate": 3.582089552238806e-06,
      "loss": 0.922,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4081010818481445,
      "learning_rate": 3.6050516647531576e-06,
      "loss": 1.1058,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.518855094909668,
      "learning_rate": 3.628013777267509e-06,
      "loss": 0.8011,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.592240333557129,
      "learning_rate": 3.6509758897818605e-06,
      "loss": 0.841,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.523603439331055,
      "learning_rate": 3.6739380022962117e-06,
      "loss": 0.9007,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 6.88736629486084,
      "learning_rate": 3.6969001148105626e-06,
      "loss": 1.0249,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.3426289558410645,
      "learning_rate": 3.7198622273249143e-06,
      "loss": 0.9513,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.149969577789307,
      "learning_rate": 3.7428243398392655e-06,
      "loss": 0.78,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.328353404998779,
      "learning_rate": 3.7657864523536168e-06,
      "loss": 0.8487,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.221036434173584,
      "learning_rate": 3.7887485648679684e-06,
      "loss": 0.7932,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.867484092712402,
      "learning_rate": 3.8117106773823193e-06,
      "loss": 0.8442,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.915316581726074,
      "learning_rate": 3.834672789896671e-06,
      "loss": 0.9246,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.253307342529297,
      "learning_rate": 3.857634902411022e-06,
      "loss": 0.781,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.59785270690918,
      "learning_rate": 3.8805970149253735e-06,
      "loss": 0.9364,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.70538854598999,
      "learning_rate": 3.903559127439725e-06,
      "loss": 0.9197,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2291388511657715,
      "learning_rate": 3.926521239954076e-06,
      "loss": 0.7859,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.780330181121826,
      "learning_rate": 3.949483352468428e-06,
      "loss": 0.6884,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.220276355743408,
      "learning_rate": 3.9724454649827785e-06,
      "loss": 0.8829,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.149847030639648,
      "learning_rate": 3.99540757749713e-06,
      "loss": 1.0233,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.986118793487549,
      "learning_rate": 4.018369690011482e-06,
      "loss": 0.86,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.297974586486816,
      "learning_rate": 4.041331802525833e-06,
      "loss": 0.8074,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.536306381225586,
      "learning_rate": 4.0642939150401835e-06,
      "loss": 0.9527,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.294879913330078,
      "learning_rate": 4.087256027554535e-06,
      "loss": 0.9811,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.775522708892822,
      "learning_rate": 4.110218140068887e-06,
      "loss": 0.8945,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.862768650054932,
      "learning_rate": 4.1331802525832385e-06,
      "loss": 0.9182,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.271902084350586,
      "learning_rate": 4.156142365097589e-06,
      "loss": 0.9922,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.884993076324463,
      "learning_rate": 4.17910447761194e-06,
      "loss": 0.8928,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.408032417297363,
      "learning_rate": 4.202066590126292e-06,
      "loss": 0.9021,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.629034996032715,
      "learning_rate": 4.2250287026406436e-06,
      "loss": 0.8252,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.753468036651611,
      "learning_rate": 4.247990815154994e-06,
      "loss": 0.7858,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.608772277832031,
      "learning_rate": 4.270952927669346e-06,
      "loss": 0.892,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.784573793411255,
      "learning_rate": 4.293915040183697e-06,
      "loss": 0.7836,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.448291778564453,
      "learning_rate": 4.3168771526980486e-06,
      "loss": 0.8462,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.365347862243652,
      "learning_rate": 4.3398392652124e-06,
      "loss": 0.9829,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.137197017669678,
      "learning_rate": 4.362801377726751e-06,
      "loss": 0.9014,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.410482406616211,
      "learning_rate": 4.385763490241103e-06,
      "loss": 0.87,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.350554466247559,
      "learning_rate": 4.408725602755454e-06,
      "loss": 0.8673,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.982302665710449,
      "learning_rate": 4.431687715269805e-06,
      "loss": 0.8156,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.026127815246582,
      "learning_rate": 4.454649827784157e-06,
      "loss": 0.9487,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.358982563018799,
      "learning_rate": 4.477611940298508e-06,
      "loss": 0.9449,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.539482116699219,
      "learning_rate": 4.5005740528128595e-06,
      "loss": 1.0953,
      "step": 196
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.050193786621094,
      "learning_rate": 4.52353616532721e-06,
      "loss": 0.9903,
      "step": 197
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9400794506073,
      "learning_rate": 4.546498277841562e-06,
      "loss": 0.8377,
      "step": 198
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.1204752922058105,
      "learning_rate": 4.569460390355913e-06,
      "loss": 0.8603,
      "step": 199
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.446455478668213,
      "learning_rate": 4.5924225028702645e-06,
      "loss": 0.8235,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.704043865203857,
      "learning_rate": 4.615384615384616e-06,
      "loss": 0.8228,
      "step": 201
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.129299163818359,
      "learning_rate": 4.638346727898967e-06,
      "loss": 0.8028,
      "step": 202
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.490741729736328,
      "learning_rate": 4.661308840413319e-06,
      "loss": 0.8293,
      "step": 203
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.727370262145996,
      "learning_rate": 4.6842709529276695e-06,
      "loss": 0.9106,
      "step": 204
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.1572394371032715,
      "learning_rate": 4.707233065442021e-06,
      "loss": 0.9754,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.379526615142822,
      "learning_rate": 4.730195177956373e-06,
      "loss": 0.7873,
      "step": 206
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.465531349182129,
      "learning_rate": 4.753157290470724e-06,
      "loss": 0.9116,
      "step": 207
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.214580059051514,
      "learning_rate": 4.7761194029850745e-06,
      "loss": 0.7533,
      "step": 208
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.7737812995910645,
      "learning_rate": 4.799081515499426e-06,
      "loss": 0.9772,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.983648300170898,
      "learning_rate": 4.822043628013778e-06,
      "loss": 0.8276,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.512796878814697,
      "learning_rate": 4.8450057405281295e-06,
      "loss": 0.8174,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4331135749816895,
      "learning_rate": 4.86796785304248e-06,
      "loss": 0.9492,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.959298610687256,
      "learning_rate": 4.890929965556831e-06,
      "loss": 1.0584,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.431134223937988,
      "learning_rate": 4.913892078071183e-06,
      "loss": 0.8452,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.259829521179199,
      "learning_rate": 4.9368541905855346e-06,
      "loss": 0.8851,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.099543571472168,
      "learning_rate": 4.959816303099886e-06,
      "loss": 0.8888,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.6095871925354,
      "learning_rate": 4.982778415614237e-06,
      "loss": 0.6995,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.050583362579346,
      "learning_rate": 5.005740528128589e-06,
      "loss": 0.8205,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.581203937530518,
      "learning_rate": 5.02870264064294e-06,
      "loss": 0.7709,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.021407127380371,
      "learning_rate": 5.05166475315729e-06,
      "loss": 0.7754,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.26349401473999,
      "learning_rate": 5.074626865671642e-06,
      "loss": 0.6679,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.635706901550293,
      "learning_rate": 5.097588978185993e-06,
      "loss": 0.9408,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.711188316345215,
      "learning_rate": 5.1205510907003454e-06,
      "loss": 0.8308,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.771569013595581,
      "learning_rate": 5.143513203214696e-06,
      "loss": 0.8983,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.761629104614258,
      "learning_rate": 5.166475315729047e-06,
      "loss": 0.9287,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.034287929534912,
      "learning_rate": 5.189437428243399e-06,
      "loss": 0.9333,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.9626193046569824,
      "learning_rate": 5.21239954075775e-06,
      "loss": 0.6744,
      "step": 227
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.392879009246826,
      "learning_rate": 5.235361653272102e-06,
      "loss": 0.8884,
      "step": 228
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.553466796875,
      "learning_rate": 5.258323765786453e-06,
      "loss": 0.9811,
      "step": 229
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.400286674499512,
      "learning_rate": 5.281285878300804e-06,
      "loss": 0.8937,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.102355480194092,
      "learning_rate": 5.3042479908151555e-06,
      "loss": 0.9307,
      "step": 231
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.2982635498046875,
      "learning_rate": 5.327210103329506e-06,
      "loss": 0.8029,
      "step": 232
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.660780429840088,
      "learning_rate": 5.350172215843859e-06,
      "loss": 0.7311,
      "step": 233
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.173713207244873,
      "learning_rate": 5.37313432835821e-06,
      "loss": 0.8409,
      "step": 234
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.164562225341797,
      "learning_rate": 5.3960964408725605e-06,
      "loss": 0.8962,
      "step": 235
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.293713092803955,
      "learning_rate": 5.419058553386912e-06,
      "loss": 0.7117,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.452841758728027,
      "learning_rate": 5.442020665901263e-06,
      "loss": 0.8802,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.225895881652832,
      "learning_rate": 5.464982778415615e-06,
      "loss": 0.742,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.624668121337891,
      "learning_rate": 5.487944890929966e-06,
      "loss": 1.131,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.140521049499512,
      "learning_rate": 5.510907003444317e-06,
      "loss": 0.8643,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.707704067230225,
      "learning_rate": 5.533869115958669e-06,
      "loss": 0.7451,
      "step": 241
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.210961818695068,
      "learning_rate": 5.55683122847302e-06,
      "loss": 0.7738,
      "step": 242
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.673533916473389,
      "learning_rate": 5.579793340987371e-06,
      "loss": 0.934,
      "step": 243
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.192202091217041,
      "learning_rate": 5.602755453501722e-06,
      "loss": 0.8731,
      "step": 244
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9897563457489014,
      "learning_rate": 5.625717566016073e-06,
      "loss": 0.9617,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.559494495391846,
      "learning_rate": 5.6486796785304256e-06,
      "loss": 0.9343,
      "step": 246
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.073042869567871,
      "learning_rate": 5.671641791044776e-06,
      "loss": 0.977,
      "step": 247
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.014003276824951,
      "learning_rate": 5.694603903559128e-06,
      "loss": 1.1399,
      "step": 248
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.484326362609863,
      "learning_rate": 5.717566016073479e-06,
      "loss": 0.8292,
      "step": 249
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.6355361938476562,
      "learning_rate": 5.74052812858783e-06,
      "loss": 0.8593,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.324713230133057,
      "learning_rate": 5.763490241102182e-06,
      "loss": 0.8123,
      "step": 251
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.127200126647949,
      "learning_rate": 5.786452353616533e-06,
      "loss": 0.9075,
      "step": 252
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.7245001792907715,
      "learning_rate": 5.809414466130885e-06,
      "loss": 0.7575,
      "step": 253
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.755061149597168,
      "learning_rate": 5.832376578645236e-06,
      "loss": 0.8221,
      "step": 254
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.3579277992248535,
      "learning_rate": 5.8553386911595864e-06,
      "loss": 0.9052,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.452084541320801,
      "learning_rate": 5.878300803673939e-06,
      "loss": 0.8143,
      "step": 256
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.128534317016602,
      "learning_rate": 5.90126291618829e-06,
      "loss": 0.9222,
      "step": 257
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.0851850509643555,
      "learning_rate": 5.9242250287026415e-06,
      "loss": 0.8658,
      "step": 258
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.12392520904541,
      "learning_rate": 5.947187141216992e-06,
      "loss": 1.0442,
      "step": 259
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.922631740570068,
      "learning_rate": 5.970149253731343e-06,
      "loss": 0.8398,
      "step": 260
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.693253040313721,
      "learning_rate": 5.993111366245695e-06,
      "loss": 0.865,
      "step": 261
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.020393371582031,
      "learning_rate": 6.0160734787600465e-06,
      "loss": 0.939,
      "step": 262
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.390758991241455,
      "learning_rate": 6.039035591274398e-06,
      "loss": 0.8212,
      "step": 263
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.256641864776611,
      "learning_rate": 6.061997703788749e-06,
      "loss": 0.8673,
      "step": 264
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.165924549102783,
      "learning_rate": 6.0849598163031e-06,
      "loss": 0.8333,
      "step": 265
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.8947038650512695,
      "learning_rate": 6.1079219288174515e-06,
      "loss": 0.923,
      "step": 266
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.344574928283691,
      "learning_rate": 6.130884041331802e-06,
      "loss": 1.0277,
      "step": 267
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.320878028869629,
      "learning_rate": 6.153846153846155e-06,
      "loss": 0.7437,
      "step": 268
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.014426231384277,
      "learning_rate": 6.176808266360506e-06,
      "loss": 0.7579,
      "step": 269
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.299583435058594,
      "learning_rate": 6.1997703788748565e-06,
      "loss": 0.9362,
      "step": 270
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.040727615356445,
      "learning_rate": 6.222732491389208e-06,
      "loss": 0.6764,
      "step": 271
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.751906156539917,
      "learning_rate": 6.245694603903559e-06,
      "loss": 0.9711,
      "step": 272
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.867489814758301,
      "learning_rate": 6.2686567164179116e-06,
      "loss": 0.741,
      "step": 273
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.4372403621673584,
      "learning_rate": 6.291618828932262e-06,
      "loss": 0.9414,
      "step": 274
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.964564800262451,
      "learning_rate": 6.314580941446614e-06,
      "loss": 0.9309,
      "step": 275
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.395320415496826,
      "learning_rate": 6.337543053960965e-06,
      "loss": 0.9179,
      "step": 276
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.871152639389038,
      "learning_rate": 6.360505166475316e-06,
      "loss": 0.8596,
      "step": 277
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.609108924865723,
      "learning_rate": 6.383467278989668e-06,
      "loss": 0.9033,
      "step": 278
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.133185863494873,
      "learning_rate": 6.406429391504019e-06,
      "loss": 0.9263,
      "step": 279
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.618285179138184,
      "learning_rate": 6.429391504018371e-06,
      "loss": 1.0165,
      "step": 280
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.5369744300842285,
      "learning_rate": 6.452353616532722e-06,
      "loss": 0.8858,
      "step": 281
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.251987457275391,
      "learning_rate": 6.4753157290470724e-06,
      "loss": 0.751,
      "step": 282
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9472572803497314,
      "learning_rate": 6.498277841561424e-06,
      "loss": 0.8184,
      "step": 283
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.692542314529419,
      "learning_rate": 6.521239954075776e-06,
      "loss": 0.8538,
      "step": 284
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.917450904846191,
      "learning_rate": 6.5442020665901275e-06,
      "loss": 1.0737,
      "step": 285
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.303256034851074,
      "learning_rate": 6.567164179104478e-06,
      "loss": 0.945,
      "step": 286
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.026377201080322,
      "learning_rate": 6.590126291618829e-06,
      "loss": 0.8422,
      "step": 287
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.188206195831299,
      "learning_rate": 6.613088404133181e-06,
      "loss": 0.8447,
      "step": 288
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.371993064880371,
      "learning_rate": 6.636050516647532e-06,
      "loss": 0.9322,
      "step": 289
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.168227672576904,
      "learning_rate": 6.659012629161884e-06,
      "loss": 0.8997,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9674065113067627,
      "learning_rate": 6.681974741676235e-06,
      "loss": 0.7986,
      "step": 291
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.828332901000977,
      "learning_rate": 6.704936854190586e-06,
      "loss": 0.7203,
      "step": 292
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.4826748371124268,
      "learning_rate": 6.7278989667049375e-06,
      "loss": 0.8873,
      "step": 293
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.5365898609161377,
      "learning_rate": 6.750861079219288e-06,
      "loss": 0.8745,
      "step": 294
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.899048328399658,
      "learning_rate": 6.773823191733641e-06,
      "loss": 0.9383,
      "step": 295
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.057714939117432,
      "learning_rate": 6.796785304247992e-06,
      "loss": 0.7545,
      "step": 296
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.038323879241943,
      "learning_rate": 6.8197474167623425e-06,
      "loss": 0.829,
      "step": 297
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.000314712524414,
      "learning_rate": 6.842709529276694e-06,
      "loss": 0.798,
      "step": 298
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.319011688232422,
      "learning_rate": 6.865671641791045e-06,
      "loss": 0.7875,
      "step": 299
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.1542487144470215,
      "learning_rate": 6.8886337543053976e-06,
      "loss": 0.9523,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9735119342803955,
      "learning_rate": 6.911595866819748e-06,
      "loss": 0.7142,
      "step": 301
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9490973949432373,
      "learning_rate": 6.934557979334099e-06,
      "loss": 0.862,
      "step": 302
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.536620140075684,
      "learning_rate": 6.957520091848451e-06,
      "loss": 0.8032,
      "step": 303
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.259764671325684,
      "learning_rate": 6.980482204362802e-06,
      "loss": 0.8844,
      "step": 304
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.935363531112671,
      "learning_rate": 7.003444316877153e-06,
      "loss": 0.8892,
      "step": 305
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.347543716430664,
      "learning_rate": 7.026406429391504e-06,
      "loss": 0.9517,
      "step": 306
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.149044036865234,
      "learning_rate": 7.049368541905856e-06,
      "loss": 0.8875,
      "step": 307
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.693485260009766,
      "learning_rate": 7.072330654420208e-06,
      "loss": 0.8191,
      "step": 308
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.980512619018555,
      "learning_rate": 7.0952927669345584e-06,
      "loss": 0.7825,
      "step": 309
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.277537822723389,
      "learning_rate": 7.11825487944891e-06,
      "loss": 0.7197,
      "step": 310
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.829869508743286,
      "learning_rate": 7.141216991963261e-06,
      "loss": 0.8837,
      "step": 311
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.725839376449585,
      "learning_rate": 7.164179104477612e-06,
      "loss": 0.9207,
      "step": 312
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.225294589996338,
      "learning_rate": 7.187141216991964e-06,
      "loss": 0.9568,
      "step": 313
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.069120407104492,
      "learning_rate": 7.210103329506315e-06,
      "loss": 0.9584,
      "step": 314
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.8187780380249023,
      "learning_rate": 7.233065442020667e-06,
      "loss": 0.9636,
      "step": 315
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.584237575531006,
      "learning_rate": 7.256027554535018e-06,
      "loss": 0.8033,
      "step": 316
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.993786811828613,
      "learning_rate": 7.2789896670493685e-06,
      "loss": 0.9901,
      "step": 317
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.696444511413574,
      "learning_rate": 7.301951779563721e-06,
      "loss": 0.8487,
      "step": 318
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.009458541870117,
      "learning_rate": 7.324913892078072e-06,
      "loss": 0.746,
      "step": 319
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.085396766662598,
      "learning_rate": 7.3478760045924235e-06,
      "loss": 1.0337,
      "step": 320
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.088379859924316,
      "learning_rate": 7.370838117106774e-06,
      "loss": 0.9745,
      "step": 321
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.935992240905762,
      "learning_rate": 7.393800229621125e-06,
      "loss": 0.8972,
      "step": 322
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.163352966308594,
      "learning_rate": 7.416762342135478e-06,
      "loss": 0.9017,
      "step": 323
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.02083683013916,
      "learning_rate": 7.4397244546498285e-06,
      "loss": 0.8408,
      "step": 324
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.5164315700531006,
      "learning_rate": 7.46268656716418e-06,
      "loss": 0.9003,
      "step": 325
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.119628429412842,
      "learning_rate": 7.485648679678531e-06,
      "loss": 0.7431,
      "step": 326
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.926002025604248,
      "learning_rate": 7.508610792192882e-06,
      "loss": 1.0917,
      "step": 327
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.14971923828125,
      "learning_rate": 7.5315729047072335e-06,
      "loss": 0.7585,
      "step": 328
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.8495354652404785,
      "learning_rate": 7.554535017221584e-06,
      "loss": 0.9115,
      "step": 329
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.6478772163391113,
      "learning_rate": 7.577497129735937e-06,
      "loss": 0.9272,
      "step": 330
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.454667568206787,
      "learning_rate": 7.600459242250288e-06,
      "loss": 0.899,
      "step": 331
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.349596977233887,
      "learning_rate": 7.6234213547646386e-06,
      "loss": 0.9622,
      "step": 332
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.209454536437988,
      "learning_rate": 7.64638346727899e-06,
      "loss": 1.0481,
      "step": 333
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.015999794006348,
      "learning_rate": 7.669345579793342e-06,
      "loss": 0.9588,
      "step": 334
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.9464075565338135,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.9626,
      "step": 335
    },
    {
      "epoch": 0.03,
      "grad_norm": 3.7841718196868896,
      "learning_rate": 7.715269804822044e-06,
      "loss": 0.7559,
      "step": 336
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.137583255767822,
      "learning_rate": 7.738231917336395e-06,
      "loss": 0.8211,
      "step": 337
    },
    {
      "epoch": 0.03,
      "grad_norm": 4.319131374359131,
      "learning_rate": 7.761194029850747e-06,
      "loss": 0.906,
      "step": 338
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5761184692382812,
      "learning_rate": 7.784156142365099e-06,
      "loss": 0.862,
      "step": 339
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.2184576988220215,
      "learning_rate": 7.80711825487945e-06,
      "loss": 0.7754,
      "step": 340
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.476991176605225,
      "learning_rate": 7.8300803673938e-06,
      "loss": 0.8047,
      "step": 341
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.956390380859375,
      "learning_rate": 7.853042479908152e-06,
      "loss": 0.9686,
      "step": 342
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9638142585754395,
      "learning_rate": 7.876004592422504e-06,
      "loss": 0.8629,
      "step": 343
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4150116443634033,
      "learning_rate": 7.898966704936855e-06,
      "loss": 0.8759,
      "step": 344
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.386812925338745,
      "learning_rate": 7.921928817451207e-06,
      "loss": 0.8732,
      "step": 345
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.935555934906006,
      "learning_rate": 7.944890929965557e-06,
      "loss": 0.8533,
      "step": 346
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.865407466888428,
      "learning_rate": 7.967853042479909e-06,
      "loss": 1.0756,
      "step": 347
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.056333541870117,
      "learning_rate": 7.99081515499426e-06,
      "loss": 0.9058,
      "step": 348
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.121253490447998,
      "learning_rate": 8.013777267508612e-06,
      "loss": 0.8785,
      "step": 349
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.186450958251953,
      "learning_rate": 8.036739380022964e-06,
      "loss": 0.9076,
      "step": 350
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7961554527282715,
      "learning_rate": 8.059701492537314e-06,
      "loss": 1.1256,
      "step": 351
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.071828842163086,
      "learning_rate": 8.082663605051665e-06,
      "loss": 0.9319,
      "step": 352
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.152992248535156,
      "learning_rate": 8.105625717566017e-06,
      "loss": 0.861,
      "step": 353
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.137078762054443,
      "learning_rate": 8.128587830080367e-06,
      "loss": 0.8485,
      "step": 354
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.6234564781188965,
      "learning_rate": 8.15154994259472e-06,
      "loss": 0.9143,
      "step": 355
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.381000518798828,
      "learning_rate": 8.17451205510907e-06,
      "loss": 0.8279,
      "step": 356
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6291046142578125,
      "learning_rate": 8.197474167623422e-06,
      "loss": 0.7429,
      "step": 357
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.5726399421691895,
      "learning_rate": 8.220436280137774e-06,
      "loss": 0.7988,
      "step": 358
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4461419582366943,
      "learning_rate": 8.243398392652124e-06,
      "loss": 0.8242,
      "step": 359
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.436957359313965,
      "learning_rate": 8.266360505166477e-06,
      "loss": 0.8136,
      "step": 360
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4564287662506104,
      "learning_rate": 8.289322617680827e-06,
      "loss": 0.9437,
      "step": 361
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.707491874694824,
      "learning_rate": 8.312284730195179e-06,
      "loss": 0.844,
      "step": 362
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.799055814743042,
      "learning_rate": 8.33524684270953e-06,
      "loss": 0.8709,
      "step": 363
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.526956558227539,
      "learning_rate": 8.35820895522388e-06,
      "loss": 1.0614,
      "step": 364
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.527256965637207,
      "learning_rate": 8.381171067738234e-06,
      "loss": 0.8572,
      "step": 365
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.1697797775268555,
      "learning_rate": 8.404133180252584e-06,
      "loss": 0.6619,
      "step": 366
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.2627854347229,
      "learning_rate": 8.427095292766935e-06,
      "loss": 0.9242,
      "step": 367
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9385406970977783,
      "learning_rate": 8.450057405281287e-06,
      "loss": 0.9186,
      "step": 368
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.20401930809021,
      "learning_rate": 8.473019517795637e-06,
      "loss": 0.951,
      "step": 369
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.899785041809082,
      "learning_rate": 8.495981630309989e-06,
      "loss": 0.7822,
      "step": 370
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.247210502624512,
      "learning_rate": 8.51894374282434e-06,
      "loss": 0.7743,
      "step": 371
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.430152893066406,
      "learning_rate": 8.541905855338692e-06,
      "loss": 0.8939,
      "step": 372
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.840051651000977,
      "learning_rate": 8.564867967853044e-06,
      "loss": 0.7041,
      "step": 373
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9736602306365967,
      "learning_rate": 8.587830080367394e-06,
      "loss": 0.8146,
      "step": 374
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5759263038635254,
      "learning_rate": 8.610792192881745e-06,
      "loss": 1.0012,
      "step": 375
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.19254207611084,
      "learning_rate": 8.633754305396097e-06,
      "loss": 1.0002,
      "step": 376
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.176846027374268,
      "learning_rate": 8.656716417910447e-06,
      "loss": 0.7495,
      "step": 377
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.349353313446045,
      "learning_rate": 8.6796785304248e-06,
      "loss": 0.8806,
      "step": 378
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9658732414245605,
      "learning_rate": 8.70264064293915e-06,
      "loss": 0.8756,
      "step": 379
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.476055145263672,
      "learning_rate": 8.725602755453502e-06,
      "loss": 0.9161,
      "step": 380
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.798128604888916,
      "learning_rate": 8.748564867967854e-06,
      "loss": 0.6806,
      "step": 381
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.30305814743042,
      "learning_rate": 8.771526980482206e-06,
      "loss": 1.0195,
      "step": 382
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.111638307571411,
      "learning_rate": 8.794489092996557e-06,
      "loss": 0.9432,
      "step": 383
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.133058071136475,
      "learning_rate": 8.817451205510907e-06,
      "loss": 1.0681,
      "step": 384
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.1507513523101807,
      "learning_rate": 8.840413318025259e-06,
      "loss": 0.8815,
      "step": 385
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.52451753616333,
      "learning_rate": 8.86337543053961e-06,
      "loss": 0.9604,
      "step": 386
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.024641990661621,
      "learning_rate": 8.886337543053962e-06,
      "loss": 0.7885,
      "step": 387
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.658738613128662,
      "learning_rate": 8.909299655568314e-06,
      "loss": 0.8504,
      "step": 388
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3896985054016113,
      "learning_rate": 8.932261768082664e-06,
      "loss": 0.9437,
      "step": 389
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.020682334899902,
      "learning_rate": 8.955223880597016e-06,
      "loss": 0.9268,
      "step": 390
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.640362024307251,
      "learning_rate": 8.978185993111367e-06,
      "loss": 0.7688,
      "step": 391
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.189111709594727,
      "learning_rate": 9.001148105625719e-06,
      "loss": 0.8667,
      "step": 392
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0973405838012695,
      "learning_rate": 9.024110218140069e-06,
      "loss": 0.6905,
      "step": 393
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8031136989593506,
      "learning_rate": 9.04707233065442e-06,
      "loss": 0.9379,
      "step": 394
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.168731689453125,
      "learning_rate": 9.070034443168772e-06,
      "loss": 0.8147,
      "step": 395
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4128355979919434,
      "learning_rate": 9.092996555683124e-06,
      "loss": 0.8062,
      "step": 396
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9147837162017822,
      "learning_rate": 9.115958668197476e-06,
      "loss": 0.9556,
      "step": 397
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9512898921966553,
      "learning_rate": 9.138920780711826e-06,
      "loss": 1.0267,
      "step": 398
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.071354866027832,
      "learning_rate": 9.161882893226177e-06,
      "loss": 0.9341,
      "step": 399
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.717787265777588,
      "learning_rate": 9.184845005740529e-06,
      "loss": 0.9068,
      "step": 400
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.131740093231201,
      "learning_rate": 9.20780711825488e-06,
      "loss": 0.8533,
      "step": 401
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.508207321166992,
      "learning_rate": 9.230769230769232e-06,
      "loss": 0.9115,
      "step": 402
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.754851818084717,
      "learning_rate": 9.253731343283582e-06,
      "loss": 0.8635,
      "step": 403
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.264218330383301,
      "learning_rate": 9.276693455797934e-06,
      "loss": 0.9506,
      "step": 404
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.41965913772583,
      "learning_rate": 9.299655568312286e-06,
      "loss": 0.7647,
      "step": 405
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3875439167022705,
      "learning_rate": 9.322617680826637e-06,
      "loss": 0.8134,
      "step": 406
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.308156967163086,
      "learning_rate": 9.345579793340989e-06,
      "loss": 0.8552,
      "step": 407
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.104142189025879,
      "learning_rate": 9.368541905855339e-06,
      "loss": 0.8704,
      "step": 408
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7334983348846436,
      "learning_rate": 9.39150401836969e-06,
      "loss": 0.7979,
      "step": 409
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8039512634277344,
      "learning_rate": 9.414466130884042e-06,
      "loss": 0.664,
      "step": 410
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.071091651916504,
      "learning_rate": 9.437428243398394e-06,
      "loss": 0.946,
      "step": 411
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.5400805473327637,
      "learning_rate": 9.460390355912746e-06,
      "loss": 0.8246,
      "step": 412
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.372835397720337,
      "learning_rate": 9.483352468427096e-06,
      "loss": 0.8482,
      "step": 413
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.768386125564575,
      "learning_rate": 9.506314580941447e-06,
      "loss": 0.8137,
      "step": 414
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4911694526672363,
      "learning_rate": 9.529276693455799e-06,
      "loss": 0.8596,
      "step": 415
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7153401374816895,
      "learning_rate": 9.552238805970149e-06,
      "loss": 0.7577,
      "step": 416
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.571103572845459,
      "learning_rate": 9.575200918484502e-06,
      "loss": 0.7231,
      "step": 417
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.9452903270721436,
      "learning_rate": 9.598163030998852e-06,
      "loss": 0.7078,
      "step": 418
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.8598387241363525,
      "learning_rate": 9.621125143513204e-06,
      "loss": 0.8143,
      "step": 419
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.4617202281951904,
      "learning_rate": 9.644087256027556e-06,
      "loss": 0.861,
      "step": 420
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.533492088317871,
      "learning_rate": 9.667049368541906e-06,
      "loss": 0.9601,
      "step": 421
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6831490993499756,
      "learning_rate": 9.690011481056259e-06,
      "loss": 0.6879,
      "step": 422
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.380390167236328,
      "learning_rate": 9.712973593570609e-06,
      "loss": 0.929,
      "step": 423
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.175103664398193,
      "learning_rate": 9.73593570608496e-06,
      "loss": 1.0125,
      "step": 424
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7637226581573486,
      "learning_rate": 9.758897818599312e-06,
      "loss": 0.8377,
      "step": 425
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.029807090759277,
      "learning_rate": 9.781859931113662e-06,
      "loss": 0.6964,
      "step": 426
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.384666919708252,
      "learning_rate": 9.804822043628016e-06,
      "loss": 0.8327,
      "step": 427
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.321525812149048,
      "learning_rate": 9.827784156142366e-06,
      "loss": 0.8164,
      "step": 428
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.20074200630188,
      "learning_rate": 9.850746268656717e-06,
      "loss": 0.8781,
      "step": 429
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.6321041584014893,
      "learning_rate": 9.873708381171069e-06,
      "loss": 0.8353,
      "step": 430
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.3848459720611572,
      "learning_rate": 9.896670493685419e-06,
      "loss": 0.7782,
      "step": 431
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.539113998413086,
      "learning_rate": 9.919632606199772e-06,
      "loss": 0.9857,
      "step": 432
    },
    {
      "epoch": 0.04,
      "grad_norm": 4.394684314727783,
      "learning_rate": 9.942594718714122e-06,
      "loss": 0.8589,
      "step": 433
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.06990909576416,
      "learning_rate": 9.965556831228474e-06,
      "loss": 0.9968,
      "step": 434
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.0745246410369873,
      "learning_rate": 9.988518943742826e-06,
      "loss": 0.9269,
      "step": 435
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.8269810676574707,
      "learning_rate": 1.0011481056257178e-05,
      "loss": 0.9544,
      "step": 436
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.747255325317383,
      "learning_rate": 1.0034443168771527e-05,
      "loss": 1.0492,
      "step": 437
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.046621799468994,
      "learning_rate": 1.005740528128588e-05,
      "loss": 1.0229,
      "step": 438
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1211981773376465,
      "learning_rate": 1.0080367393800229e-05,
      "loss": 0.9221,
      "step": 439
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.5968313217163086,
      "learning_rate": 1.010332950631458e-05,
      "loss": 0.9207,
      "step": 440
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.058807373046875,
      "learning_rate": 1.0126291618828934e-05,
      "loss": 0.8615,
      "step": 441
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.327049732208252,
      "learning_rate": 1.0149253731343284e-05,
      "loss": 0.7708,
      "step": 442
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.22676682472229,
      "learning_rate": 1.0172215843857636e-05,
      "loss": 1.01,
      "step": 443
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7570745944976807,
      "learning_rate": 1.0195177956371986e-05,
      "loss": 0.8702,
      "step": 444
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7596163749694824,
      "learning_rate": 1.0218140068886338e-05,
      "loss": 0.9626,
      "step": 445
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.301237106323242,
      "learning_rate": 1.0241102181400691e-05,
      "loss": 0.9686,
      "step": 446
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.026617527008057,
      "learning_rate": 1.0264064293915041e-05,
      "loss": 0.8502,
      "step": 447
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.100874900817871,
      "learning_rate": 1.0287026406429393e-05,
      "loss": 0.8012,
      "step": 448
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.162534713745117,
      "learning_rate": 1.0309988518943743e-05,
      "loss": 0.9236,
      "step": 449
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.307284116744995,
      "learning_rate": 1.0332950631458094e-05,
      "loss": 0.734,
      "step": 450
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.490355968475342,
      "learning_rate": 1.0355912743972448e-05,
      "loss": 0.8708,
      "step": 451
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4261879920959473,
      "learning_rate": 1.0378874856486798e-05,
      "loss": 0.6721,
      "step": 452
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.580018997192383,
      "learning_rate": 1.040183696900115e-05,
      "loss": 0.6701,
      "step": 453
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6100616455078125,
      "learning_rate": 1.04247990815155e-05,
      "loss": 0.834,
      "step": 454
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.052455425262451,
      "learning_rate": 1.0447761194029851e-05,
      "loss": 0.8662,
      "step": 455
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.31422758102417,
      "learning_rate": 1.0470723306544204e-05,
      "loss": 0.9425,
      "step": 456
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.703728199005127,
      "learning_rate": 1.0493685419058554e-05,
      "loss": 0.8731,
      "step": 457
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.3318328857421875,
      "learning_rate": 1.0516647531572906e-05,
      "loss": 0.787,
      "step": 458
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1770381927490234,
      "learning_rate": 1.0539609644087256e-05,
      "loss": 0.9185,
      "step": 459
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.119440078735352,
      "learning_rate": 1.0562571756601608e-05,
      "loss": 0.9924,
      "step": 460
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4901556968688965,
      "learning_rate": 1.0585533869115961e-05,
      "loss": 0.9891,
      "step": 461
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3573129177093506,
      "learning_rate": 1.0608495981630311e-05,
      "loss": 0.7711,
      "step": 462
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6228442192077637,
      "learning_rate": 1.0631458094144663e-05,
      "loss": 0.9401,
      "step": 463
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.447473049163818,
      "learning_rate": 1.0654420206659013e-05,
      "loss": 1.034,
      "step": 464
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.9034993648529053,
      "learning_rate": 1.0677382319173364e-05,
      "loss": 0.9153,
      "step": 465
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.5602071285247803,
      "learning_rate": 1.0700344431687718e-05,
      "loss": 1.0159,
      "step": 466
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3267831802368164,
      "learning_rate": 1.0723306544202068e-05,
      "loss": 0.9176,
      "step": 467
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7425696849823,
      "learning_rate": 1.074626865671642e-05,
      "loss": 0.8224,
      "step": 468
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.058056831359863,
      "learning_rate": 1.076923076923077e-05,
      "loss": 0.9029,
      "step": 469
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.489707946777344,
      "learning_rate": 1.0792192881745121e-05,
      "loss": 0.9501,
      "step": 470
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.91005277633667,
      "learning_rate": 1.0815154994259474e-05,
      "loss": 0.9856,
      "step": 471
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.307399272918701,
      "learning_rate": 1.0838117106773824e-05,
      "loss": 0.9406,
      "step": 472
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.649885892868042,
      "learning_rate": 1.0861079219288176e-05,
      "loss": 0.9311,
      "step": 473
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.0562267303466797,
      "learning_rate": 1.0884041331802526e-05,
      "loss": 0.7164,
      "step": 474
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.533273935317993,
      "learning_rate": 1.0907003444316878e-05,
      "loss": 0.9525,
      "step": 475
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.620434045791626,
      "learning_rate": 1.092996555683123e-05,
      "loss": 0.8251,
      "step": 476
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.714993476867676,
      "learning_rate": 1.0952927669345581e-05,
      "loss": 0.8089,
      "step": 477
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.351182699203491,
      "learning_rate": 1.0975889781859933e-05,
      "loss": 0.7792,
      "step": 478
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6330106258392334,
      "learning_rate": 1.0998851894374283e-05,
      "loss": 0.8574,
      "step": 479
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.5142009258270264,
      "learning_rate": 1.1021814006888634e-05,
      "loss": 0.805,
      "step": 480
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.936117172241211,
      "learning_rate": 1.1044776119402986e-05,
      "loss": 0.9218,
      "step": 481
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6479907035827637,
      "learning_rate": 1.1067738231917338e-05,
      "loss": 0.7439,
      "step": 482
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.253437042236328,
      "learning_rate": 1.1090700344431688e-05,
      "loss": 0.854,
      "step": 483
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.683920383453369,
      "learning_rate": 1.111366245694604e-05,
      "loss": 0.8139,
      "step": 484
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6228065490722656,
      "learning_rate": 1.1136624569460391e-05,
      "loss": 0.9221,
      "step": 485
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.090435266494751,
      "learning_rate": 1.1159586681974743e-05,
      "loss": 0.9719,
      "step": 486
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4524080753326416,
      "learning_rate": 1.1182548794489094e-05,
      "loss": 0.8157,
      "step": 487
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.8177051544189453,
      "learning_rate": 1.1205510907003444e-05,
      "loss": 0.9161,
      "step": 488
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4604666233062744,
      "learning_rate": 1.1228473019517796e-05,
      "loss": 1.0901,
      "step": 489
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.9055161476135254,
      "learning_rate": 1.1251435132032146e-05,
      "loss": 0.9279,
      "step": 490
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.982961416244507,
      "learning_rate": 1.12743972445465e-05,
      "loss": 0.8378,
      "step": 491
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.440809726715088,
      "learning_rate": 1.1297359357060851e-05,
      "loss": 0.7931,
      "step": 492
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.5206141471862793,
      "learning_rate": 1.1320321469575201e-05,
      "loss": 1.022,
      "step": 493
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3727004528045654,
      "learning_rate": 1.1343283582089553e-05,
      "loss": 0.9525,
      "step": 494
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.365635395050049,
      "learning_rate": 1.1366245694603903e-05,
      "loss": 0.8283,
      "step": 495
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2720632553100586,
      "learning_rate": 1.1389207807118256e-05,
      "loss": 0.9198,
      "step": 496
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7254505157470703,
      "learning_rate": 1.1412169919632608e-05,
      "loss": 0.7387,
      "step": 497
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1786537170410156,
      "learning_rate": 1.1435132032146958e-05,
      "loss": 0.7819,
      "step": 498
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.736795425415039,
      "learning_rate": 1.145809414466131e-05,
      "loss": 0.8152,
      "step": 499
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.610224962234497,
      "learning_rate": 1.148105625717566e-05,
      "loss": 0.8852,
      "step": 500
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.214968681335449,
      "learning_rate": 1.1504018369690013e-05,
      "loss": 0.8991,
      "step": 501
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.486290454864502,
      "learning_rate": 1.1526980482204365e-05,
      "loss": 0.7847,
      "step": 502
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.521998643875122,
      "learning_rate": 1.1549942594718715e-05,
      "loss": 0.8662,
      "step": 503
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.509025812149048,
      "learning_rate": 1.1572904707233066e-05,
      "loss": 0.8146,
      "step": 504
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6503446102142334,
      "learning_rate": 1.1595866819747416e-05,
      "loss": 0.8574,
      "step": 505
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.2747483253479,
      "learning_rate": 1.161882893226177e-05,
      "loss": 0.8526,
      "step": 506
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.742077589035034,
      "learning_rate": 1.1641791044776121e-05,
      "loss": 0.9164,
      "step": 507
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.8833694458007812,
      "learning_rate": 1.1664753157290471e-05,
      "loss": 0.8799,
      "step": 508
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1992568969726562,
      "learning_rate": 1.1687715269804823e-05,
      "loss": 0.8899,
      "step": 509
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6627745628356934,
      "learning_rate": 1.1710677382319173e-05,
      "loss": 0.8822,
      "step": 510
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3951098918914795,
      "learning_rate": 1.1733639494833526e-05,
      "loss": 0.829,
      "step": 511
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.471184253692627,
      "learning_rate": 1.1756601607347878e-05,
      "loss": 0.8717,
      "step": 512
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2254996299743652,
      "learning_rate": 1.1779563719862228e-05,
      "loss": 0.8426,
      "step": 513
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7690932750701904,
      "learning_rate": 1.180252583237658e-05,
      "loss": 0.8299,
      "step": 514
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.9429826736450195,
      "learning_rate": 1.182548794489093e-05,
      "loss": 0.8251,
      "step": 515
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.031096935272217,
      "learning_rate": 1.1848450057405283e-05,
      "loss": 0.9736,
      "step": 516
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.295043468475342,
      "learning_rate": 1.1871412169919635e-05,
      "loss": 0.9574,
      "step": 517
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.110567569732666,
      "learning_rate": 1.1894374282433985e-05,
      "loss": 0.8197,
      "step": 518
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.1688337326049805,
      "learning_rate": 1.1917336394948336e-05,
      "loss": 0.7337,
      "step": 519
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.3539233207702637,
      "learning_rate": 1.1940298507462686e-05,
      "loss": 0.8183,
      "step": 520
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.0996060371398926,
      "learning_rate": 1.196326061997704e-05,
      "loss": 0.852,
      "step": 521
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.6841914653778076,
      "learning_rate": 1.198622273249139e-05,
      "loss": 0.9281,
      "step": 522
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.011679172515869,
      "learning_rate": 1.2009184845005741e-05,
      "loss": 0.6748,
      "step": 523
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.4486563205718994,
      "learning_rate": 1.2032146957520093e-05,
      "loss": 0.9411,
      "step": 524
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.754462957382202,
      "learning_rate": 1.2055109070034443e-05,
      "loss": 0.8198,
      "step": 525
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7498652935028076,
      "learning_rate": 1.2078071182548796e-05,
      "loss": 0.8816,
      "step": 526
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.7777602672576904,
      "learning_rate": 1.2101033295063146e-05,
      "loss": 0.7661,
      "step": 527
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.5019218921661377,
      "learning_rate": 1.2123995407577498e-05,
      "loss": 0.9115,
      "step": 528
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.2724573612213135,
      "learning_rate": 1.2146957520091848e-05,
      "loss": 0.7842,
      "step": 529
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.989933490753174,
      "learning_rate": 1.21699196326062e-05,
      "loss": 0.8781,
      "step": 530
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.503113269805908,
      "learning_rate": 1.2192881745120553e-05,
      "loss": 0.7842,
      "step": 531
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5942482948303223,
      "learning_rate": 1.2215843857634903e-05,
      "loss": 0.9065,
      "step": 532
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.271517038345337,
      "learning_rate": 1.2238805970149255e-05,
      "loss": 1.0145,
      "step": 533
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0260024070739746,
      "learning_rate": 1.2261768082663605e-05,
      "loss": 0.8874,
      "step": 534
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.464106798171997,
      "learning_rate": 1.2284730195177956e-05,
      "loss": 0.8309,
      "step": 535
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8433003425598145,
      "learning_rate": 1.230769230769231e-05,
      "loss": 0.9756,
      "step": 536
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.599761724472046,
      "learning_rate": 1.233065442020666e-05,
      "loss": 0.8708,
      "step": 537
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.7371487617492676,
      "learning_rate": 1.2353616532721011e-05,
      "loss": 0.8066,
      "step": 538
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.342031955718994,
      "learning_rate": 1.2376578645235361e-05,
      "loss": 0.8737,
      "step": 539
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3767552375793457,
      "learning_rate": 1.2399540757749713e-05,
      "loss": 0.9594,
      "step": 540
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.140208721160889,
      "learning_rate": 1.2422502870264066e-05,
      "loss": 0.8405,
      "step": 541
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.193423748016357,
      "learning_rate": 1.2445464982778416e-05,
      "loss": 0.8147,
      "step": 542
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1445810794830322,
      "learning_rate": 1.2468427095292768e-05,
      "loss": 0.8521,
      "step": 543
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.01643705368042,
      "learning_rate": 1.2491389207807118e-05,
      "loss": 0.9437,
      "step": 544
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.43487286567688,
      "learning_rate": 1.2514351320321471e-05,
      "loss": 0.8784,
      "step": 545
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.664339542388916,
      "learning_rate": 1.2537313432835823e-05,
      "loss": 0.8883,
      "step": 546
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.297792673110962,
      "learning_rate": 1.2560275545350173e-05,
      "loss": 1.067,
      "step": 547
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6528899669647217,
      "learning_rate": 1.2583237657864525e-05,
      "loss": 0.984,
      "step": 548
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6728320121765137,
      "learning_rate": 1.2606199770378875e-05,
      "loss": 0.7744,
      "step": 549
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.051248073577881,
      "learning_rate": 1.2629161882893228e-05,
      "loss": 0.8567,
      "step": 550
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.923189401626587,
      "learning_rate": 1.265212399540758e-05,
      "loss": 0.8792,
      "step": 551
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2526495456695557,
      "learning_rate": 1.267508610792193e-05,
      "loss": 0.8537,
      "step": 552
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9394538402557373,
      "learning_rate": 1.2698048220436281e-05,
      "loss": 0.8264,
      "step": 553
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.085975408554077,
      "learning_rate": 1.2721010332950631e-05,
      "loss": 1.0375,
      "step": 554
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.356471538543701,
      "learning_rate": 1.2743972445464985e-05,
      "loss": 0.9673,
      "step": 555
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.113237142562866,
      "learning_rate": 1.2766934557979337e-05,
      "loss": 0.9335,
      "step": 556
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.340766429901123,
      "learning_rate": 1.2789896670493687e-05,
      "loss": 0.8587,
      "step": 557
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2730979919433594,
      "learning_rate": 1.2812858783008038e-05,
      "loss": 0.7372,
      "step": 558
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5382702350616455,
      "learning_rate": 1.2835820895522388e-05,
      "loss": 0.8068,
      "step": 559
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0718743801116943,
      "learning_rate": 1.2858783008036742e-05,
      "loss": 0.8529,
      "step": 560
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6543161869049072,
      "learning_rate": 1.2881745120551093e-05,
      "loss": 1.1398,
      "step": 561
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1366214752197266,
      "learning_rate": 1.2904707233065443e-05,
      "loss": 0.9775,
      "step": 562
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.53302264213562,
      "learning_rate": 1.2927669345579795e-05,
      "loss": 0.7046,
      "step": 563
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6017239093780518,
      "learning_rate": 1.2950631458094145e-05,
      "loss": 0.7603,
      "step": 564
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.652869701385498,
      "learning_rate": 1.2973593570608498e-05,
      "loss": 0.9049,
      "step": 565
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.777569055557251,
      "learning_rate": 1.2996555683122848e-05,
      "loss": 0.9902,
      "step": 566
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.113292694091797,
      "learning_rate": 1.30195177956372e-05,
      "loss": 0.8603,
      "step": 567
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.165821075439453,
      "learning_rate": 1.3042479908151552e-05,
      "loss": 0.7306,
      "step": 568
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.282181739807129,
      "learning_rate": 1.3065442020665902e-05,
      "loss": 0.84,
      "step": 569
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.348649263381958,
      "learning_rate": 1.3088404133180255e-05,
      "loss": 0.9071,
      "step": 570
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.8373281955718994,
      "learning_rate": 1.3111366245694605e-05,
      "loss": 0.8318,
      "step": 571
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2313923835754395,
      "learning_rate": 1.3134328358208957e-05,
      "loss": 0.9233,
      "step": 572
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1371819972991943,
      "learning_rate": 1.3157290470723307e-05,
      "loss": 0.727,
      "step": 573
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1881964206695557,
      "learning_rate": 1.3180252583237658e-05,
      "loss": 0.963,
      "step": 574
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.451094627380371,
      "learning_rate": 1.3203214695752012e-05,
      "loss": 0.8026,
      "step": 575
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.807602643966675,
      "learning_rate": 1.3226176808266362e-05,
      "loss": 0.9823,
      "step": 576
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.964092493057251,
      "learning_rate": 1.3249138920780713e-05,
      "loss": 0.868,
      "step": 577
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0867488384246826,
      "learning_rate": 1.3272101033295063e-05,
      "loss": 0.8998,
      "step": 578
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9775092601776123,
      "learning_rate": 1.3295063145809415e-05,
      "loss": 0.8627,
      "step": 579
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4376680850982666,
      "learning_rate": 1.3318025258323768e-05,
      "loss": 0.8928,
      "step": 580
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.332655668258667,
      "learning_rate": 1.3340987370838118e-05,
      "loss": 0.7401,
      "step": 581
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4109504222869873,
      "learning_rate": 1.336394948335247e-05,
      "loss": 0.9229,
      "step": 582
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.027367353439331,
      "learning_rate": 1.338691159586682e-05,
      "loss": 0.8407,
      "step": 583
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.155247449874878,
      "learning_rate": 1.3409873708381172e-05,
      "loss": 0.8399,
      "step": 584
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1963284015655518,
      "learning_rate": 1.3432835820895525e-05,
      "loss": 0.9076,
      "step": 585
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8136065006256104,
      "learning_rate": 1.3455797933409875e-05,
      "loss": 0.8652,
      "step": 586
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4413435459136963,
      "learning_rate": 1.3478760045924227e-05,
      "loss": 0.9594,
      "step": 587
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4618966579437256,
      "learning_rate": 1.3501722158438577e-05,
      "loss": 0.8044,
      "step": 588
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.300273895263672,
      "learning_rate": 1.3524684270952928e-05,
      "loss": 0.8631,
      "step": 589
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1685702800750732,
      "learning_rate": 1.3547646383467282e-05,
      "loss": 0.8073,
      "step": 590
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1630306243896484,
      "learning_rate": 1.3570608495981632e-05,
      "loss": 0.9865,
      "step": 591
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7342309951782227,
      "learning_rate": 1.3593570608495983e-05,
      "loss": 0.7668,
      "step": 592
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.006671905517578,
      "learning_rate": 1.3616532721010333e-05,
      "loss": 1.0262,
      "step": 593
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4591946601867676,
      "learning_rate": 1.3639494833524685e-05,
      "loss": 0.9892,
      "step": 594
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.8197121620178223,
      "learning_rate": 1.3662456946039038e-05,
      "loss": 0.9456,
      "step": 595
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.859922409057617,
      "learning_rate": 1.3685419058553388e-05,
      "loss": 0.8166,
      "step": 596
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9817922115325928,
      "learning_rate": 1.370838117106774e-05,
      "loss": 0.8261,
      "step": 597
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.399034023284912,
      "learning_rate": 1.373134328358209e-05,
      "loss": 0.8323,
      "step": 598
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6136727333068848,
      "learning_rate": 1.3754305396096442e-05,
      "loss": 0.8431,
      "step": 599
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.323173761367798,
      "learning_rate": 1.3777267508610795e-05,
      "loss": 0.8155,
      "step": 600
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.210742473602295,
      "learning_rate": 1.3800229621125145e-05,
      "loss": 0.8659,
      "step": 601
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.4509479999542236,
      "learning_rate": 1.3823191733639497e-05,
      "loss": 0.7582,
      "step": 602
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.2635579109191895,
      "learning_rate": 1.3846153846153847e-05,
      "loss": 0.9699,
      "step": 603
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.130608558654785,
      "learning_rate": 1.3869115958668198e-05,
      "loss": 0.8212,
      "step": 604
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.284322738647461,
      "learning_rate": 1.389207807118255e-05,
      "loss": 0.7359,
      "step": 605
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3794703483581543,
      "learning_rate": 1.3915040183696902e-05,
      "loss": 0.8844,
      "step": 606
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1055006980895996,
      "learning_rate": 1.3938002296211253e-05,
      "loss": 0.7266,
      "step": 607
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5256075859069824,
      "learning_rate": 1.3960964408725603e-05,
      "loss": 0.8606,
      "step": 608
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.6733860969543457,
      "learning_rate": 1.3983926521239955e-05,
      "loss": 0.792,
      "step": 609
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3777506351470947,
      "learning_rate": 1.4006888633754307e-05,
      "loss": 0.8929,
      "step": 610
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.587770462036133,
      "learning_rate": 1.4029850746268658e-05,
      "loss": 1.0374,
      "step": 611
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.5215954780578613,
      "learning_rate": 1.4052812858783008e-05,
      "loss": 0.9345,
      "step": 612
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.9325110912323,
      "learning_rate": 1.407577497129736e-05,
      "loss": 0.8344,
      "step": 613
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.031860828399658,
      "learning_rate": 1.4098737083811712e-05,
      "loss": 0.956,
      "step": 614
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9552829265594482,
      "learning_rate": 1.4121699196326064e-05,
      "loss": 1.009,
      "step": 615
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.744626760482788,
      "learning_rate": 1.4144661308840415e-05,
      "loss": 0.9103,
      "step": 616
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.655282497406006,
      "learning_rate": 1.4167623421354765e-05,
      "loss": 0.9353,
      "step": 617
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1177048683166504,
      "learning_rate": 1.4190585533869117e-05,
      "loss": 0.8984,
      "step": 618
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.934053659439087,
      "learning_rate": 1.4213547646383467e-05,
      "loss": 0.9246,
      "step": 619
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.7638936042785645,
      "learning_rate": 1.423650975889782e-05,
      "loss": 0.8151,
      "step": 620
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9884917736053467,
      "learning_rate": 1.4259471871412172e-05,
      "loss": 0.859,
      "step": 621
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.82218599319458,
      "learning_rate": 1.4282433983926522e-05,
      "loss": 0.9409,
      "step": 622
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.3481173515319824,
      "learning_rate": 1.4305396096440874e-05,
      "loss": 0.9029,
      "step": 623
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.786447048187256,
      "learning_rate": 1.4328358208955224e-05,
      "loss": 0.6577,
      "step": 624
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.244760751724243,
      "learning_rate": 1.4351320321469577e-05,
      "loss": 0.8001,
      "step": 625
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.182206392288208,
      "learning_rate": 1.4374282433983929e-05,
      "loss": 0.8519,
      "step": 626
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9583280086517334,
      "learning_rate": 1.4397244546498279e-05,
      "loss": 0.8373,
      "step": 627
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.9334590435028076,
      "learning_rate": 1.442020665901263e-05,
      "loss": 0.859,
      "step": 628
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9576547145843506,
      "learning_rate": 1.444316877152698e-05,
      "loss": 0.8328,
      "step": 629
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4299488067626953,
      "learning_rate": 1.4466130884041334e-05,
      "loss": 0.8978,
      "step": 630
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.033498764038086,
      "learning_rate": 1.4489092996555685e-05,
      "loss": 0.8756,
      "step": 631
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.598738670349121,
      "learning_rate": 1.4512055109070035e-05,
      "loss": 0.9293,
      "step": 632
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1022589206695557,
      "learning_rate": 1.4535017221584387e-05,
      "loss": 1.0016,
      "step": 633
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.618626117706299,
      "learning_rate": 1.4557979334098737e-05,
      "loss": 0.9431,
      "step": 634
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.385056734085083,
      "learning_rate": 1.458094144661309e-05,
      "loss": 0.8943,
      "step": 635
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.372645139694214,
      "learning_rate": 1.4603903559127442e-05,
      "loss": 0.9093,
      "step": 636
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1216111183166504,
      "learning_rate": 1.4626865671641792e-05,
      "loss": 0.8378,
      "step": 637
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.395212411880493,
      "learning_rate": 1.4649827784156144e-05,
      "loss": 0.8831,
      "step": 638
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.037611961364746,
      "learning_rate": 1.4672789896670494e-05,
      "loss": 0.9902,
      "step": 639
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.28792142868042,
      "learning_rate": 1.4695752009184847e-05,
      "loss": 0.8186,
      "step": 640
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.1745123863220215,
      "learning_rate": 1.4718714121699199e-05,
      "loss": 0.9336,
      "step": 641
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.5173087120056152,
      "learning_rate": 1.4741676234213549e-05,
      "loss": 0.9703,
      "step": 642
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.888092041015625,
      "learning_rate": 1.47646383467279e-05,
      "loss": 0.7093,
      "step": 643
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.135319471359253,
      "learning_rate": 1.478760045924225e-05,
      "loss": 0.8405,
      "step": 644
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2723138332366943,
      "learning_rate": 1.4810562571756604e-05,
      "loss": 0.8513,
      "step": 645
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.436913251876831,
      "learning_rate": 1.4833524684270955e-05,
      "loss": 0.8786,
      "step": 646
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.7655091285705566,
      "learning_rate": 1.4856486796785305e-05,
      "loss": 0.7788,
      "step": 647
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4136555194854736,
      "learning_rate": 1.4879448909299657e-05,
      "loss": 0.9332,
      "step": 648
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.843653917312622,
      "learning_rate": 1.4902411021814007e-05,
      "loss": 0.8384,
      "step": 649
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.779755115509033,
      "learning_rate": 1.492537313432836e-05,
      "loss": 0.8754,
      "step": 650
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.075176954269409,
      "learning_rate": 1.494833524684271e-05,
      "loss": 0.8164,
      "step": 651
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.828186511993408,
      "learning_rate": 1.4971297359357062e-05,
      "loss": 0.9437,
      "step": 652
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2531182765960693,
      "learning_rate": 1.4994259471871414e-05,
      "loss": 0.8003,
      "step": 653
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.751694679260254,
      "learning_rate": 1.5017221584385764e-05,
      "loss": 0.8954,
      "step": 654
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.410426378250122,
      "learning_rate": 1.5040183696900117e-05,
      "loss": 0.9712,
      "step": 655
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.989579916000366,
      "learning_rate": 1.5063145809414467e-05,
      "loss": 0.9506,
      "step": 656
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5307092666625977,
      "learning_rate": 1.5086107921928819e-05,
      "loss": 0.8615,
      "step": 657
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8520071506500244,
      "learning_rate": 1.5109070034443169e-05,
      "loss": 0.92,
      "step": 658
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.6973531246185303,
      "learning_rate": 1.513203214695752e-05,
      "loss": 0.9668,
      "step": 659
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1771600246429443,
      "learning_rate": 1.5154994259471874e-05,
      "loss": 0.9029,
      "step": 660
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.6042683124542236,
      "learning_rate": 1.5177956371986224e-05,
      "loss": 0.7856,
      "step": 661
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.0088462829589844,
      "learning_rate": 1.5200918484500575e-05,
      "loss": 0.8511,
      "step": 662
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.041168451309204,
      "learning_rate": 1.5223880597014925e-05,
      "loss": 1.0696,
      "step": 663
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1171517372131348,
      "learning_rate": 1.5246842709529277e-05,
      "loss": 1.0291,
      "step": 664
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9869580268859863,
      "learning_rate": 1.526980482204363e-05,
      "loss": 0.9885,
      "step": 665
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1203064918518066,
      "learning_rate": 1.529276693455798e-05,
      "loss": 0.9457,
      "step": 666
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9822795391082764,
      "learning_rate": 1.531572904707233e-05,
      "loss": 0.8641,
      "step": 667
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.965097665786743,
      "learning_rate": 1.5338691159586684e-05,
      "loss": 0.8911,
      "step": 668
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.655259609222412,
      "learning_rate": 1.5361653272101034e-05,
      "loss": 0.8958,
      "step": 669
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9025375843048096,
      "learning_rate": 1.5384615384615387e-05,
      "loss": 0.837,
      "step": 670
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1188805103302,
      "learning_rate": 1.5407577497129737e-05,
      "loss": 0.9108,
      "step": 671
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.926222324371338,
      "learning_rate": 1.5430539609644087e-05,
      "loss": 0.7934,
      "step": 672
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.642127275466919,
      "learning_rate": 1.545350172215844e-05,
      "loss": 0.7775,
      "step": 673
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2506415843963623,
      "learning_rate": 1.547646383467279e-05,
      "loss": 0.753,
      "step": 674
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.238351821899414,
      "learning_rate": 1.5499425947187144e-05,
      "loss": 0.8169,
      "step": 675
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2316341400146484,
      "learning_rate": 1.5522388059701494e-05,
      "loss": 0.7974,
      "step": 676
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.007772445678711,
      "learning_rate": 1.5545350172215844e-05,
      "loss": 0.7189,
      "step": 677
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.577174186706543,
      "learning_rate": 1.5568312284730197e-05,
      "loss": 1.0037,
      "step": 678
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.996863603591919,
      "learning_rate": 1.5591274397244547e-05,
      "loss": 0.723,
      "step": 679
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9003701210021973,
      "learning_rate": 1.56142365097589e-05,
      "loss": 0.871,
      "step": 680
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2805538177490234,
      "learning_rate": 1.563719862227325e-05,
      "loss": 0.828,
      "step": 681
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.104759693145752,
      "learning_rate": 1.56601607347876e-05,
      "loss": 0.8413,
      "step": 682
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8980650901794434,
      "learning_rate": 1.5683122847301954e-05,
      "loss": 0.8912,
      "step": 683
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3528671264648438,
      "learning_rate": 1.5706084959816304e-05,
      "loss": 0.8708,
      "step": 684
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.035257339477539,
      "learning_rate": 1.5729047072330657e-05,
      "loss": 0.8497,
      "step": 685
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.042098045349121,
      "learning_rate": 1.5752009184845007e-05,
      "loss": 0.9104,
      "step": 686
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8078856468200684,
      "learning_rate": 1.5774971297359357e-05,
      "loss": 0.7156,
      "step": 687
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2000670433044434,
      "learning_rate": 1.579793340987371e-05,
      "loss": 0.8836,
      "step": 688
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.483649492263794,
      "learning_rate": 1.582089552238806e-05,
      "loss": 0.9165,
      "step": 689
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2864673137664795,
      "learning_rate": 1.5843857634902414e-05,
      "loss": 1.0243,
      "step": 690
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3800299167633057,
      "learning_rate": 1.5866819747416764e-05,
      "loss": 0.9091,
      "step": 691
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2332241535186768,
      "learning_rate": 1.5889781859931114e-05,
      "loss": 0.9792,
      "step": 692
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.724653482437134,
      "learning_rate": 1.5912743972445467e-05,
      "loss": 0.7275,
      "step": 693
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3619723320007324,
      "learning_rate": 1.5935706084959817e-05,
      "loss": 0.8623,
      "step": 694
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.1155014038085938,
      "learning_rate": 1.595866819747417e-05,
      "loss": 0.923,
      "step": 695
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.607236623764038,
      "learning_rate": 1.598163030998852e-05,
      "loss": 0.9485,
      "step": 696
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.159066915512085,
      "learning_rate": 1.600459242250287e-05,
      "loss": 0.8142,
      "step": 697
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9797682762145996,
      "learning_rate": 1.6027554535017224e-05,
      "loss": 1.0477,
      "step": 698
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.950838804244995,
      "learning_rate": 1.6050516647531574e-05,
      "loss": 1.026,
      "step": 699
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9110193252563477,
      "learning_rate": 1.6073478760045927e-05,
      "loss": 0.794,
      "step": 700
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.567335605621338,
      "learning_rate": 1.6096440872560277e-05,
      "loss": 0.8471,
      "step": 701
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.6483445167541504,
      "learning_rate": 1.6119402985074627e-05,
      "loss": 0.9238,
      "step": 702
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.083828926086426,
      "learning_rate": 1.6142365097588977e-05,
      "loss": 0.9335,
      "step": 703
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.7765557765960693,
      "learning_rate": 1.616532721010333e-05,
      "loss": 0.7853,
      "step": 704
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.869961738586426,
      "learning_rate": 1.6188289322617684e-05,
      "loss": 0.8352,
      "step": 705
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8744349479675293,
      "learning_rate": 1.6211251435132034e-05,
      "loss": 0.6898,
      "step": 706
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.751593589782715,
      "learning_rate": 1.6234213547646384e-05,
      "loss": 0.7854,
      "step": 707
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.969900369644165,
      "learning_rate": 1.6257175660160734e-05,
      "loss": 0.7807,
      "step": 708
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.9137070178985596,
      "learning_rate": 1.6280137772675087e-05,
      "loss": 0.6869,
      "step": 709
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.8694422245025635,
      "learning_rate": 1.630309988518944e-05,
      "loss": 0.9817,
      "step": 710
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6149230003356934,
      "learning_rate": 1.632606199770379e-05,
      "loss": 0.8602,
      "step": 711
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.747627019882202,
      "learning_rate": 1.634902411021814e-05,
      "loss": 0.8514,
      "step": 712
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.6090638637542725,
      "learning_rate": 1.637198622273249e-05,
      "loss": 0.8786,
      "step": 713
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3849434852600098,
      "learning_rate": 1.6394948335246844e-05,
      "loss": 0.9613,
      "step": 714
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.91542387008667,
      "learning_rate": 1.6417910447761197e-05,
      "loss": 0.8558,
      "step": 715
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.5893614292144775,
      "learning_rate": 1.6440872560275547e-05,
      "loss": 0.7401,
      "step": 716
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.755019187927246,
      "learning_rate": 1.6463834672789897e-05,
      "loss": 0.9644,
      "step": 717
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5651957988739014,
      "learning_rate": 1.6486796785304247e-05,
      "loss": 0.9185,
      "step": 718
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.2135043144226074,
      "learning_rate": 1.65097588978186e-05,
      "loss": 0.9656,
      "step": 719
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4224705696105957,
      "learning_rate": 1.6532721010332954e-05,
      "loss": 1.0579,
      "step": 720
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8234059810638428,
      "learning_rate": 1.6555683122847304e-05,
      "loss": 0.9931,
      "step": 721
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.946467399597168,
      "learning_rate": 1.6578645235361654e-05,
      "loss": 1.0368,
      "step": 722
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.993129253387451,
      "learning_rate": 1.6601607347876004e-05,
      "loss": 0.7607,
      "step": 723
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.049433469772339,
      "learning_rate": 1.6624569460390357e-05,
      "loss": 0.9805,
      "step": 724
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.8884007930755615,
      "learning_rate": 1.664753157290471e-05,
      "loss": 0.8522,
      "step": 725
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4711813926696777,
      "learning_rate": 1.667049368541906e-05,
      "loss": 0.9121,
      "step": 726
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.438626289367676,
      "learning_rate": 1.669345579793341e-05,
      "loss": 0.9156,
      "step": 727
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.5187530517578125,
      "learning_rate": 1.671641791044776e-05,
      "loss": 0.8962,
      "step": 728
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.047717571258545,
      "learning_rate": 1.6739380022962114e-05,
      "loss": 0.773,
      "step": 729
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7068443298339844,
      "learning_rate": 1.6762342135476468e-05,
      "loss": 0.9593,
      "step": 730
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7117321491241455,
      "learning_rate": 1.6785304247990818e-05,
      "loss": 0.8744,
      "step": 731
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8715767860412598,
      "learning_rate": 1.6808266360505168e-05,
      "loss": 0.9886,
      "step": 732
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7349157333374023,
      "learning_rate": 1.6831228473019517e-05,
      "loss": 0.7707,
      "step": 733
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4654181003570557,
      "learning_rate": 1.685419058553387e-05,
      "loss": 0.9215,
      "step": 734
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.187586545944214,
      "learning_rate": 1.687715269804822e-05,
      "loss": 0.9885,
      "step": 735
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9722201824188232,
      "learning_rate": 1.6900114810562574e-05,
      "loss": 0.9174,
      "step": 736
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.552947521209717,
      "learning_rate": 1.6923076923076924e-05,
      "loss": 0.9113,
      "step": 737
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.954569101333618,
      "learning_rate": 1.6946039035591274e-05,
      "loss": 0.8495,
      "step": 738
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9934866428375244,
      "learning_rate": 1.6969001148105628e-05,
      "loss": 0.8463,
      "step": 739
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.676614761352539,
      "learning_rate": 1.6991963260619978e-05,
      "loss": 0.9574,
      "step": 740
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0674703121185303,
      "learning_rate": 1.701492537313433e-05,
      "loss": 0.952,
      "step": 741
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.461832046508789,
      "learning_rate": 1.703788748564868e-05,
      "loss": 0.7773,
      "step": 742
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7202866077423096,
      "learning_rate": 1.706084959816303e-05,
      "loss": 0.778,
      "step": 743
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.303269386291504,
      "learning_rate": 1.7083811710677384e-05,
      "loss": 1.1048,
      "step": 744
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.123577356338501,
      "learning_rate": 1.7106773823191734e-05,
      "loss": 0.7001,
      "step": 745
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6915948390960693,
      "learning_rate": 1.7129735935706088e-05,
      "loss": 0.6477,
      "step": 746
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.072223663330078,
      "learning_rate": 1.7152698048220438e-05,
      "loss": 1.0796,
      "step": 747
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8355371952056885,
      "learning_rate": 1.7175660160734788e-05,
      "loss": 0.8476,
      "step": 748
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1912715435028076,
      "learning_rate": 1.7198622273249138e-05,
      "loss": 1.133,
      "step": 749
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5859570503234863,
      "learning_rate": 1.722158438576349e-05,
      "loss": 0.7805,
      "step": 750
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.881028175354004,
      "learning_rate": 1.7244546498277844e-05,
      "loss": 0.6907,
      "step": 751
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.066716194152832,
      "learning_rate": 1.7267508610792194e-05,
      "loss": 0.9178,
      "step": 752
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9769446849823,
      "learning_rate": 1.7290470723306544e-05,
      "loss": 0.9115,
      "step": 753
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7590172290802,
      "learning_rate": 1.7313432835820894e-05,
      "loss": 0.8493,
      "step": 754
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.644439458847046,
      "learning_rate": 1.7336394948335248e-05,
      "loss": 0.8146,
      "step": 755
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8205363750457764,
      "learning_rate": 1.73593570608496e-05,
      "loss": 0.8264,
      "step": 756
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5979411602020264,
      "learning_rate": 1.738231917336395e-05,
      "loss": 0.8643,
      "step": 757
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6825385093688965,
      "learning_rate": 1.74052812858783e-05,
      "loss": 1.0352,
      "step": 758
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7352206707000732,
      "learning_rate": 1.742824339839265e-05,
      "loss": 0.748,
      "step": 759
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.711498737335205,
      "learning_rate": 1.7451205510907004e-05,
      "loss": 0.982,
      "step": 760
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.5104987621307373,
      "learning_rate": 1.7474167623421358e-05,
      "loss": 0.9034,
      "step": 761
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7704765796661377,
      "learning_rate": 1.7497129735935708e-05,
      "loss": 0.8042,
      "step": 762
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8795664310455322,
      "learning_rate": 1.7520091848450058e-05,
      "loss": 0.9413,
      "step": 763
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8637332916259766,
      "learning_rate": 1.754305396096441e-05,
      "loss": 0.8433,
      "step": 764
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0988569259643555,
      "learning_rate": 1.756601607347876e-05,
      "loss": 0.9957,
      "step": 765
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.388356924057007,
      "learning_rate": 1.7588978185993114e-05,
      "loss": 0.8366,
      "step": 766
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.819035053253174,
      "learning_rate": 1.7611940298507464e-05,
      "loss": 0.8894,
      "step": 767
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4355437755584717,
      "learning_rate": 1.7634902411021814e-05,
      "loss": 0.9703,
      "step": 768
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.5701730251312256,
      "learning_rate": 1.7657864523536168e-05,
      "loss": 0.7869,
      "step": 769
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.545642375946045,
      "learning_rate": 1.7680826636050518e-05,
      "loss": 0.7661,
      "step": 770
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.404409170150757,
      "learning_rate": 1.770378874856487e-05,
      "loss": 0.7816,
      "step": 771
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8166613578796387,
      "learning_rate": 1.772675086107922e-05,
      "loss": 0.969,
      "step": 772
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.418102741241455,
      "learning_rate": 1.774971297359357e-05,
      "loss": 0.6645,
      "step": 773
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8022780418395996,
      "learning_rate": 1.7772675086107924e-05,
      "loss": 0.6851,
      "step": 774
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.650477409362793,
      "learning_rate": 1.7795637198622274e-05,
      "loss": 0.8092,
      "step": 775
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.347161054611206,
      "learning_rate": 1.7818599311136628e-05,
      "loss": 1.0091,
      "step": 776
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0893313884735107,
      "learning_rate": 1.7841561423650978e-05,
      "loss": 0.8951,
      "step": 777
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4576048851013184,
      "learning_rate": 1.7864523536165328e-05,
      "loss": 1.0753,
      "step": 778
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.793785810470581,
      "learning_rate": 1.788748564867968e-05,
      "loss": 1.0397,
      "step": 779
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.238753080368042,
      "learning_rate": 1.791044776119403e-05,
      "loss": 1.0435,
      "step": 780
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.972264289855957,
      "learning_rate": 1.7933409873708384e-05,
      "loss": 0.9488,
      "step": 781
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.038844347000122,
      "learning_rate": 1.7956371986222734e-05,
      "loss": 0.8299,
      "step": 782
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8793282508850098,
      "learning_rate": 1.7979334098737084e-05,
      "loss": 0.8206,
      "step": 783
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4721055030822754,
      "learning_rate": 1.8002296211251438e-05,
      "loss": 0.9761,
      "step": 784
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0490221977233887,
      "learning_rate": 1.8025258323765788e-05,
      "loss": 0.9498,
      "step": 785
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6500275135040283,
      "learning_rate": 1.8048220436280138e-05,
      "loss": 1.0299,
      "step": 786
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.738213539123535,
      "learning_rate": 1.807118254879449e-05,
      "loss": 0.8258,
      "step": 787
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8518106937408447,
      "learning_rate": 1.809414466130884e-05,
      "loss": 0.8436,
      "step": 788
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0822482109069824,
      "learning_rate": 1.8117106773823195e-05,
      "loss": 1.0139,
      "step": 789
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9280598163604736,
      "learning_rate": 1.8140068886337545e-05,
      "loss": 1.0396,
      "step": 790
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4894165992736816,
      "learning_rate": 1.8163030998851894e-05,
      "loss": 0.9478,
      "step": 791
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.816270112991333,
      "learning_rate": 1.8185993111366248e-05,
      "loss": 0.921,
      "step": 792
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1545515060424805,
      "learning_rate": 1.8208955223880598e-05,
      "loss": 0.896,
      "step": 793
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.426240921020508,
      "learning_rate": 1.823191733639495e-05,
      "loss": 0.8579,
      "step": 794
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.394425868988037,
      "learning_rate": 1.82548794489093e-05,
      "loss": 0.9621,
      "step": 795
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.4773104190826416,
      "learning_rate": 1.827784156142365e-05,
      "loss": 0.9063,
      "step": 796
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4848878383636475,
      "learning_rate": 1.8300803673938005e-05,
      "loss": 0.912,
      "step": 797
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8063530921936035,
      "learning_rate": 1.8323765786452355e-05,
      "loss": 0.9847,
      "step": 798
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8333656787872314,
      "learning_rate": 1.8346727898966708e-05,
      "loss": 0.9202,
      "step": 799
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.697174549102783,
      "learning_rate": 1.8369690011481058e-05,
      "loss": 0.8504,
      "step": 800
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.7410833835601807,
      "learning_rate": 1.8392652123995408e-05,
      "loss": 0.964,
      "step": 801
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.879204750061035,
      "learning_rate": 1.841561423650976e-05,
      "loss": 0.701,
      "step": 802
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.12770938873291,
      "learning_rate": 1.843857634902411e-05,
      "loss": 1.043,
      "step": 803
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.915311813354492,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 0.9136,
      "step": 804
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.449093818664551,
      "learning_rate": 1.8484500574052815e-05,
      "loss": 0.8353,
      "step": 805
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3794877529144287,
      "learning_rate": 1.8507462686567165e-05,
      "loss": 0.8797,
      "step": 806
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4822747707366943,
      "learning_rate": 1.8530424799081518e-05,
      "loss": 1.1721,
      "step": 807
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.4410579204559326,
      "learning_rate": 1.8553386911595868e-05,
      "loss": 0.7992,
      "step": 808
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0571539402008057,
      "learning_rate": 1.857634902411022e-05,
      "loss": 0.7946,
      "step": 809
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.109219551086426,
      "learning_rate": 1.859931113662457e-05,
      "loss": 0.9786,
      "step": 810
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.786639451980591,
      "learning_rate": 1.862227324913892e-05,
      "loss": 0.8239,
      "step": 811
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9449753761291504,
      "learning_rate": 1.8645235361653275e-05,
      "loss": 0.9092,
      "step": 812
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.096168041229248,
      "learning_rate": 1.8668197474167625e-05,
      "loss": 0.9707,
      "step": 813
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.6220927238464355,
      "learning_rate": 1.8691159586681978e-05,
      "loss": 0.8718,
      "step": 814
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.0937633514404297,
      "learning_rate": 1.8714121699196328e-05,
      "loss": 1.0646,
      "step": 815
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.8486649990081787,
      "learning_rate": 1.8737083811710678e-05,
      "loss": 0.8586,
      "step": 816
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.525486946105957,
      "learning_rate": 1.876004592422503e-05,
      "loss": 0.923,
      "step": 817
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.1844699382781982,
      "learning_rate": 1.878300803673938e-05,
      "loss": 0.9082,
      "step": 818
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.5681328773498535,
      "learning_rate": 1.8805970149253735e-05,
      "loss": 0.8774,
      "step": 819
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9530794620513916,
      "learning_rate": 1.8828932261768085e-05,
      "loss": 0.8255,
      "step": 820
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.830906867980957,
      "learning_rate": 1.8851894374282435e-05,
      "loss": 0.9145,
      "step": 821
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9188671112060547,
      "learning_rate": 1.8874856486796788e-05,
      "loss": 0.8399,
      "step": 822
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3724372386932373,
      "learning_rate": 1.8897818599311138e-05,
      "loss": 0.8748,
      "step": 823
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.826037883758545,
      "learning_rate": 1.892078071182549e-05,
      "loss": 1.1221,
      "step": 824
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.769397735595703,
      "learning_rate": 1.894374282433984e-05,
      "loss": 0.8943,
      "step": 825
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0204813480377197,
      "learning_rate": 1.896670493685419e-05,
      "loss": 0.9641,
      "step": 826
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5326883792877197,
      "learning_rate": 1.8989667049368545e-05,
      "loss": 0.8503,
      "step": 827
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3608124256134033,
      "learning_rate": 1.9012629161882895e-05,
      "loss": 0.9128,
      "step": 828
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5768256187438965,
      "learning_rate": 1.9035591274397248e-05,
      "loss": 0.8589,
      "step": 829
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.609642505645752,
      "learning_rate": 1.9058553386911598e-05,
      "loss": 0.7253,
      "step": 830
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6901865005493164,
      "learning_rate": 1.9081515499425948e-05,
      "loss": 0.9447,
      "step": 831
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.495391368865967,
      "learning_rate": 1.9104477611940298e-05,
      "loss": 0.8323,
      "step": 832
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6523489952087402,
      "learning_rate": 1.912743972445465e-05,
      "loss": 0.8977,
      "step": 833
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.58539080619812,
      "learning_rate": 1.9150401836969005e-05,
      "loss": 0.9524,
      "step": 834
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5684220790863037,
      "learning_rate": 1.9173363949483355e-05,
      "loss": 0.8389,
      "step": 835
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.002803325653076,
      "learning_rate": 1.9196326061997705e-05,
      "loss": 0.9881,
      "step": 836
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7153303623199463,
      "learning_rate": 1.9219288174512055e-05,
      "loss": 0.9207,
      "step": 837
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.3688998222351074,
      "learning_rate": 1.9242250287026408e-05,
      "loss": 1.0025,
      "step": 838
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.504105806350708,
      "learning_rate": 1.926521239954076e-05,
      "loss": 0.864,
      "step": 839
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8823630809783936,
      "learning_rate": 1.928817451205511e-05,
      "loss": 1.1364,
      "step": 840
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0295166969299316,
      "learning_rate": 1.931113662456946e-05,
      "loss": 0.9225,
      "step": 841
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.642953395843506,
      "learning_rate": 1.933409873708381e-05,
      "loss": 0.9252,
      "step": 842
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.652095079421997,
      "learning_rate": 1.9357060849598165e-05,
      "loss": 0.9457,
      "step": 843
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2250101566314697,
      "learning_rate": 1.9380022962112518e-05,
      "loss": 1.011,
      "step": 844
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.71874737739563,
      "learning_rate": 1.9402985074626868e-05,
      "loss": 0.9923,
      "step": 845
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5964443683624268,
      "learning_rate": 1.9425947187141218e-05,
      "loss": 0.9903,
      "step": 846
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8400301933288574,
      "learning_rate": 1.9448909299655568e-05,
      "loss": 0.8522,
      "step": 847
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5018081665039062,
      "learning_rate": 1.947187141216992e-05,
      "loss": 0.8731,
      "step": 848
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5717241764068604,
      "learning_rate": 1.9494833524684275e-05,
      "loss": 0.9693,
      "step": 849
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4790127277374268,
      "learning_rate": 1.9517795637198625e-05,
      "loss": 0.9588,
      "step": 850
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5659632682800293,
      "learning_rate": 1.9540757749712975e-05,
      "loss": 0.9915,
      "step": 851
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9224209785461426,
      "learning_rate": 1.9563719862227325e-05,
      "loss": 0.8784,
      "step": 852
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7787928581237793,
      "learning_rate": 1.9586681974741678e-05,
      "loss": 0.9452,
      "step": 853
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.498107433319092,
      "learning_rate": 1.960964408725603e-05,
      "loss": 0.7679,
      "step": 854
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.813915252685547,
      "learning_rate": 1.963260619977038e-05,
      "loss": 0.9742,
      "step": 855
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.1595685482025146,
      "learning_rate": 1.965556831228473e-05,
      "loss": 0.9689,
      "step": 856
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.645312786102295,
      "learning_rate": 1.967853042479908e-05,
      "loss": 1.0274,
      "step": 857
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7952122688293457,
      "learning_rate": 1.9701492537313435e-05,
      "loss": 0.9673,
      "step": 858
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9935407638549805,
      "learning_rate": 1.9724454649827788e-05,
      "loss": 0.8874,
      "step": 859
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6467511653900146,
      "learning_rate": 1.9747416762342138e-05,
      "loss": 0.9031,
      "step": 860
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.415799856185913,
      "learning_rate": 1.9770378874856488e-05,
      "loss": 0.9134,
      "step": 861
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.9027531147003174,
      "learning_rate": 1.9793340987370838e-05,
      "loss": 0.8948,
      "step": 862
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.47847843170166,
      "learning_rate": 1.981630309988519e-05,
      "loss": 0.7607,
      "step": 863
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1133315563201904,
      "learning_rate": 1.9839265212399545e-05,
      "loss": 0.8953,
      "step": 864
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5125489234924316,
      "learning_rate": 1.9862227324913895e-05,
      "loss": 1.0425,
      "step": 865
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8470327854156494,
      "learning_rate": 1.9885189437428245e-05,
      "loss": 1.0345,
      "step": 866
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2874515056610107,
      "learning_rate": 1.9908151549942595e-05,
      "loss": 0.8007,
      "step": 867
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7260584831237793,
      "learning_rate": 1.9931113662456948e-05,
      "loss": 0.7524,
      "step": 868
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6743409633636475,
      "learning_rate": 1.9954075774971298e-05,
      "loss": 1.0215,
      "step": 869
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6105194091796875,
      "learning_rate": 1.997703788748565e-05,
      "loss": 0.9697,
      "step": 870
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1913399696350098,
      "learning_rate": 2e-05,
      "loss": 0.6974,
      "step": 871
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.723367214202881,
      "learning_rate": 1.9999999937663123e-05,
      "loss": 0.768,
      "step": 872
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.100646734237671,
      "learning_rate": 1.9999999750652496e-05,
      "loss": 0.8491,
      "step": 873
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5527663230895996,
      "learning_rate": 1.9999999438968115e-05,
      "loss": 0.819,
      "step": 874
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.566145896911621,
      "learning_rate": 1.9999999002609992e-05,
      "loss": 0.9068,
      "step": 875
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7661781311035156,
      "learning_rate": 1.9999998441578123e-05,
      "loss": 0.9142,
      "step": 876
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.407264471054077,
      "learning_rate": 1.999999775587252e-05,
      "loss": 0.9476,
      "step": 877
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.1255531311035156,
      "learning_rate": 1.9999996945493196e-05,
      "loss": 0.7707,
      "step": 878
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.654320001602173,
      "learning_rate": 1.9999996010440156e-05,
      "loss": 0.7025,
      "step": 879
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6140458583831787,
      "learning_rate": 1.999999495071341e-05,
      "loss": 0.8249,
      "step": 880
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5260376930236816,
      "learning_rate": 1.9999993766312977e-05,
      "loss": 0.9013,
      "step": 881
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4571969509124756,
      "learning_rate": 1.9999992457238864e-05,
      "loss": 0.8914,
      "step": 882
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6889851093292236,
      "learning_rate": 1.9999991023491094e-05,
      "loss": 0.8869,
      "step": 883
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6033318042755127,
      "learning_rate": 1.9999989465069682e-05,
      "loss": 0.8362,
      "step": 884
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.016779661178589,
      "learning_rate": 1.9999987781974646e-05,
      "loss": 0.8925,
      "step": 885
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8732540607452393,
      "learning_rate": 1.9999985974206014e-05,
      "loss": 0.9514,
      "step": 886
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.0783581733703613,
      "learning_rate": 1.9999984041763798e-05,
      "loss": 0.9419,
      "step": 887
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.43451189994812,
      "learning_rate": 1.999998198464803e-05,
      "loss": 1.0048,
      "step": 888
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8434882164001465,
      "learning_rate": 1.9999979802858737e-05,
      "loss": 0.9706,
      "step": 889
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5891807079315186,
      "learning_rate": 1.9999977496395936e-05,
      "loss": 1.0295,
      "step": 890
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4661381244659424,
      "learning_rate": 1.9999975065259667e-05,
      "loss": 0.9037,
      "step": 891
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4369966983795166,
      "learning_rate": 1.9999972509449953e-05,
      "loss": 0.8565,
      "step": 892
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.069127082824707,
      "learning_rate": 1.999996982896683e-05,
      "loss": 0.815,
      "step": 893
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.537055253982544,
      "learning_rate": 1.9999967023810325e-05,
      "loss": 1.0295,
      "step": 894
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.636227607727051,
      "learning_rate": 1.9999964093980484e-05,
      "loss": 0.8991,
      "step": 895
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7450008392333984,
      "learning_rate": 1.9999961039477334e-05,
      "loss": 0.9637,
      "step": 896
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.607830047607422,
      "learning_rate": 1.9999957860300916e-05,
      "loss": 0.8044,
      "step": 897
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.691972255706787,
      "learning_rate": 1.9999954556451274e-05,
      "loss": 0.8593,
      "step": 898
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4329683780670166,
      "learning_rate": 1.999995112792844e-05,
      "loss": 0.8482,
      "step": 899
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6534605026245117,
      "learning_rate": 1.9999947574732465e-05,
      "loss": 0.9043,
      "step": 900
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2491750717163086,
      "learning_rate": 1.999994389686339e-05,
      "loss": 0.9412,
      "step": 901
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5062406063079834,
      "learning_rate": 1.999994009432126e-05,
      "loss": 0.8919,
      "step": 902
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6822407245635986,
      "learning_rate": 1.9999936167106126e-05,
      "loss": 0.861,
      "step": 903
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.958000659942627,
      "learning_rate": 1.999993211521803e-05,
      "loss": 1.0451,
      "step": 904
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4181530475616455,
      "learning_rate": 1.9999927938657028e-05,
      "loss": 0.9479,
      "step": 905
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.489133358001709,
      "learning_rate": 1.9999923637423177e-05,
      "loss": 0.9515,
      "step": 906
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.803149461746216,
      "learning_rate": 1.9999919211516517e-05,
      "loss": 0.7817,
      "step": 907
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.8150739669799805,
      "learning_rate": 1.9999914660937115e-05,
      "loss": 0.9292,
      "step": 908
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.426208257675171,
      "learning_rate": 1.9999909985685023e-05,
      "loss": 0.9894,
      "step": 909
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.4720308780670166,
      "learning_rate": 1.99999051857603e-05,
      "loss": 1.1513,
      "step": 910
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5060760974884033,
      "learning_rate": 1.9999900261163004e-05,
      "loss": 0.9699,
      "step": 911
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.6651268005371094,
      "learning_rate": 1.99998952118932e-05,
      "loss": 0.822,
      "step": 912
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.2426252365112305,
      "learning_rate": 1.999989003795095e-05,
      "loss": 0.8789,
      "step": 913
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.7411611080169678,
      "learning_rate": 1.9999884739336314e-05,
      "loss": 0.7212,
      "step": 914
    },
    {
      "epoch": 0.09,
      "grad_norm": 3.2448904514312744,
      "learning_rate": 1.9999879316049366e-05,
      "loss": 0.9576,
      "step": 915
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5815165042877197,
      "learning_rate": 1.9999873768090165e-05,
      "loss": 0.9732,
      "step": 916
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5583789348602295,
      "learning_rate": 1.9999868095458787e-05,
      "loss": 1.0862,
      "step": 917
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.405000925064087,
      "learning_rate": 1.99998622981553e-05,
      "loss": 0.85,
      "step": 918
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7782609462738037,
      "learning_rate": 1.9999856376179778e-05,
      "loss": 0.8386,
      "step": 919
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9213449954986572,
      "learning_rate": 1.9999850329532288e-05,
      "loss": 0.9616,
      "step": 920
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4610252380371094,
      "learning_rate": 1.9999844158212912e-05,
      "loss": 0.9993,
      "step": 921
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5463004112243652,
      "learning_rate": 1.999983786222173e-05,
      "loss": 0.9375,
      "step": 922
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.939316749572754,
      "learning_rate": 1.9999831441558813e-05,
      "loss": 0.7862,
      "step": 923
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.0206663608551025,
      "learning_rate": 1.9999824896224244e-05,
      "loss": 0.8375,
      "step": 924
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7214043140411377,
      "learning_rate": 1.9999818226218106e-05,
      "loss": 0.9896,
      "step": 925
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.30002498626709,
      "learning_rate": 1.9999811431540482e-05,
      "loss": 0.9561,
      "step": 926
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.706022262573242,
      "learning_rate": 1.999980451219145e-05,
      "loss": 0.9488,
      "step": 927
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.562495231628418,
      "learning_rate": 1.999979746817111e-05,
      "loss": 0.7511,
      "step": 928
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4366025924682617,
      "learning_rate": 1.9999790299479534e-05,
      "loss": 1.0084,
      "step": 929
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6734261512756348,
      "learning_rate": 1.9999783006116823e-05,
      "loss": 0.875,
      "step": 930
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.390395402908325,
      "learning_rate": 1.9999775588083062e-05,
      "loss": 1.2161,
      "step": 931
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.717597246170044,
      "learning_rate": 1.999976804537835e-05,
      "loss": 0.9102,
      "step": 932
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5338664054870605,
      "learning_rate": 1.9999760378002773e-05,
      "loss": 0.7959,
      "step": 933
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.37174654006958,
      "learning_rate": 1.999975258595643e-05,
      "loss": 0.7999,
      "step": 934
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.311476707458496,
      "learning_rate": 1.999974466923942e-05,
      "loss": 0.9775,
      "step": 935
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9003608226776123,
      "learning_rate": 1.9999736627851835e-05,
      "loss": 0.9991,
      "step": 936
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5780723094940186,
      "learning_rate": 1.9999728461793785e-05,
      "loss": 0.8753,
      "step": 937
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.249624013900757,
      "learning_rate": 1.9999720171065367e-05,
      "loss": 0.9562,
      "step": 938
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.888937473297119,
      "learning_rate": 1.9999711755666686e-05,
      "loss": 0.8172,
      "step": 939
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.569915533065796,
      "learning_rate": 1.999970321559784e-05,
      "loss": 0.9622,
      "step": 940
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4684488773345947,
      "learning_rate": 1.999969455085894e-05,
      "loss": 0.8496,
      "step": 941
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.742236614227295,
      "learning_rate": 1.99996857614501e-05,
      "loss": 0.8406,
      "step": 942
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.578219175338745,
      "learning_rate": 1.9999676847371424e-05,
      "loss": 0.9151,
      "step": 943
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5757062435150146,
      "learning_rate": 1.9999667808623023e-05,
      "loss": 0.8476,
      "step": 944
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0406198501586914,
      "learning_rate": 1.9999658645205006e-05,
      "loss": 0.8007,
      "step": 945
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2870593070983887,
      "learning_rate": 1.9999649357117495e-05,
      "loss": 0.9608,
      "step": 946
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.686136245727539,
      "learning_rate": 1.99996399443606e-05,
      "loss": 0.8828,
      "step": 947
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.372722864151001,
      "learning_rate": 1.9999630406934445e-05,
      "loss": 0.7379,
      "step": 948
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4711430072784424,
      "learning_rate": 1.9999620744839143e-05,
      "loss": 1.1081,
      "step": 949
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.88568115234375,
      "learning_rate": 1.9999610958074814e-05,
      "loss": 0.8568,
      "step": 950
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.571930170059204,
      "learning_rate": 1.999960104664158e-05,
      "loss": 0.7812,
      "step": 951
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4372596740722656,
      "learning_rate": 1.999959101053957e-05,
      "loss": 1.0097,
      "step": 952
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6148781776428223,
      "learning_rate": 1.9999580849768904e-05,
      "loss": 0.8158,
      "step": 953
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.137955665588379,
      "learning_rate": 1.9999570564329713e-05,
      "loss": 0.9236,
      "step": 954
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.448589324951172,
      "learning_rate": 1.9999560154222123e-05,
      "loss": 0.9248,
      "step": 955
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.319563627243042,
      "learning_rate": 1.9999549619446258e-05,
      "loss": 0.7391,
      "step": 956
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.561960458755493,
      "learning_rate": 1.9999538960002258e-05,
      "loss": 0.7623,
      "step": 957
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.778262138366699,
      "learning_rate": 1.9999528175890254e-05,
      "loss": 0.902,
      "step": 958
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4342846870422363,
      "learning_rate": 1.9999517267110377e-05,
      "loss": 0.9109,
      "step": 959
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7331042289733887,
      "learning_rate": 1.9999506233662767e-05,
      "loss": 0.9343,
      "step": 960
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5074474811553955,
      "learning_rate": 1.9999495075547557e-05,
      "loss": 0.9019,
      "step": 961
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.799865245819092,
      "learning_rate": 1.999948379276489e-05,
      "loss": 0.8557,
      "step": 962
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.812520742416382,
      "learning_rate": 1.9999472385314905e-05,
      "loss": 0.9683,
      "step": 963
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3033828735351562,
      "learning_rate": 1.9999460853197746e-05,
      "loss": 1.0048,
      "step": 964
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.607726573944092,
      "learning_rate": 1.9999449196413555e-05,
      "loss": 0.888,
      "step": 965
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5376880168914795,
      "learning_rate": 1.9999437414962476e-05,
      "loss": 0.8595,
      "step": 966
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9395010471343994,
      "learning_rate": 1.9999425508844657e-05,
      "loss": 1.0354,
      "step": 967
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.694248676300049,
      "learning_rate": 1.999941347806025e-05,
      "loss": 0.873,
      "step": 968
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.9424896240234375,
      "learning_rate": 1.99994013226094e-05,
      "loss": 0.8948,
      "step": 969
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.8795652389526367,
      "learning_rate": 1.9999389042492262e-05,
      "loss": 0.9303,
      "step": 970
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.0644118785858154,
      "learning_rate": 1.9999376637708986e-05,
      "loss": 0.9978,
      "step": 971
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7903547286987305,
      "learning_rate": 1.999936410825973e-05,
      "loss": 0.9263,
      "step": 972
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.267378330230713,
      "learning_rate": 1.999935145414465e-05,
      "loss": 0.8585,
      "step": 973
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3255128860473633,
      "learning_rate": 1.9999338675363896e-05,
      "loss": 0.8537,
      "step": 974
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.395404100418091,
      "learning_rate": 1.999932577191764e-05,
      "loss": 0.9058,
      "step": 975
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.452144145965576,
      "learning_rate": 1.9999312743806034e-05,
      "loss": 0.7474,
      "step": 976
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1243772506713867,
      "learning_rate": 1.999929959102924e-05,
      "loss": 0.9506,
      "step": 977
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7852776050567627,
      "learning_rate": 1.9999286313587428e-05,
      "loss": 0.9893,
      "step": 978
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6329469680786133,
      "learning_rate": 1.9999272911480764e-05,
      "loss": 0.8952,
      "step": 979
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4152209758758545,
      "learning_rate": 1.999925938470941e-05,
      "loss": 0.8093,
      "step": 980
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5717620849609375,
      "learning_rate": 1.9999245733273532e-05,
      "loss": 0.7358,
      "step": 981
    },
    {
      "epoch": 0.1,
      "grad_norm": 3.1925976276397705,
      "learning_rate": 1.9999231957173303e-05,
      "loss": 0.9368,
      "step": 982
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3553385734558105,
      "learning_rate": 1.9999218056408904e-05,
      "loss": 0.8904,
      "step": 983
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.394230365753174,
      "learning_rate": 1.9999204030980492e-05,
      "loss": 0.8831,
      "step": 984
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.693147897720337,
      "learning_rate": 1.9999189880888258e-05,
      "loss": 0.8567,
      "step": 985
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.274751901626587,
      "learning_rate": 1.9999175606132364e-05,
      "loss": 0.8455,
      "step": 986
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3059723377227783,
      "learning_rate": 1.9999161206713e-05,
      "loss": 0.8765,
      "step": 987
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.189061403274536,
      "learning_rate": 1.9999146682630336e-05,
      "loss": 0.796,
      "step": 988
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.618983745574951,
      "learning_rate": 1.9999132033884557e-05,
      "loss": 0.8602,
      "step": 989
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6669464111328125,
      "learning_rate": 1.999911726047585e-05,
      "loss": 0.9323,
      "step": 990
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7233095169067383,
      "learning_rate": 1.9999102362404393e-05,
      "loss": 0.9378,
      "step": 991
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.228811264038086,
      "learning_rate": 1.9999087339670377e-05,
      "loss": 0.8665,
      "step": 992
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6293067932128906,
      "learning_rate": 1.9999072192273982e-05,
      "loss": 0.9268,
      "step": 993
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5925099849700928,
      "learning_rate": 1.99990569202154e-05,
      "loss": 0.8671,
      "step": 994
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5729122161865234,
      "learning_rate": 1.9999041523494823e-05,
      "loss": 0.9975,
      "step": 995
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6713266372680664,
      "learning_rate": 1.9999026002112447e-05,
      "loss": 0.9231,
      "step": 996
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.323176145553589,
      "learning_rate": 1.9999010356068457e-05,
      "loss": 0.8479,
      "step": 997
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7096779346466064,
      "learning_rate": 1.9998994585363053e-05,
      "loss": 0.7147,
      "step": 998
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.615011215209961,
      "learning_rate": 1.999897868999643e-05,
      "loss": 0.9633,
      "step": 999
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.6658740043640137,
      "learning_rate": 1.9998962669968786e-05,
      "loss": 1.0357,
      "step": 1000
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.442504405975342,
      "learning_rate": 1.999894652528032e-05,
      "loss": 1.0265,
      "step": 1001
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.398993492126465,
      "learning_rate": 1.999893025593124e-05,
      "loss": 0.8015,
      "step": 1002
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1876471042633057,
      "learning_rate": 1.9998913861921738e-05,
      "loss": 0.8621,
      "step": 1003
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.903266191482544,
      "learning_rate": 1.9998897343252026e-05,
      "loss": 0.7767,
      "step": 1004
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.718898296356201,
      "learning_rate": 1.9998880699922306e-05,
      "loss": 0.8137,
      "step": 1005
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0966720581054688,
      "learning_rate": 1.9998863931932792e-05,
      "loss": 0.8155,
      "step": 1006
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.415025234222412,
      "learning_rate": 1.9998847039283685e-05,
      "loss": 1.0177,
      "step": 1007
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4356281757354736,
      "learning_rate": 1.99988300219752e-05,
      "loss": 0.9098,
      "step": 1008
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0012900829315186,
      "learning_rate": 1.999881288000755e-05,
      "loss": 0.835,
      "step": 1009
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.4588851928710938,
      "learning_rate": 1.9998795613380944e-05,
      "loss": 0.7681,
      "step": 1010
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3957622051239014,
      "learning_rate": 1.9998778222095602e-05,
      "loss": 1.0174,
      "step": 1011
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.7640249729156494,
      "learning_rate": 1.999876070615174e-05,
      "loss": 0.9325,
      "step": 1012
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.5631203651428223,
      "learning_rate": 1.9998743065549572e-05,
      "loss": 0.7005,
      "step": 1013
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.159585952758789,
      "learning_rate": 1.9998725300289325e-05,
      "loss": 0.9409,
      "step": 1014
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.3866236209869385,
      "learning_rate": 1.999870741037121e-05,
      "loss": 0.8857,
      "step": 1015
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6895339488983154,
      "learning_rate": 1.9998689395795462e-05,
      "loss": 0.9085,
      "step": 1016
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5904998779296875,
      "learning_rate": 1.99986712565623e-05,
      "loss": 0.8368,
      "step": 1017
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7750790119171143,
      "learning_rate": 1.999865299267195e-05,
      "loss": 0.908,
      "step": 1018
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.350917339324951,
      "learning_rate": 1.9998634604124642e-05,
      "loss": 0.8816,
      "step": 1019
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7290894985198975,
      "learning_rate": 1.99986160909206e-05,
      "loss": 0.9458,
      "step": 1020
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.831031322479248,
      "learning_rate": 1.999859745306006e-05,
      "loss": 1.0012,
      "step": 1021
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8352346420288086,
      "learning_rate": 1.9998578690543256e-05,
      "loss": 0.9325,
      "step": 1022
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.611283540725708,
      "learning_rate": 1.9998559803370418e-05,
      "loss": 0.9512,
      "step": 1023
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9895169734954834,
      "learning_rate": 1.9998540791541777e-05,
      "loss": 0.8633,
      "step": 1024
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3132975101470947,
      "learning_rate": 1.9998521655057577e-05,
      "loss": 1.0029,
      "step": 1025
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8286421298980713,
      "learning_rate": 1.999850239391806e-05,
      "loss": 0.9406,
      "step": 1026
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5232622623443604,
      "learning_rate": 1.9998483008123454e-05,
      "loss": 0.9502,
      "step": 1027
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.126694917678833,
      "learning_rate": 1.9998463497674012e-05,
      "loss": 0.9559,
      "step": 1028
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.3944413661956787,
      "learning_rate": 1.999844386256997e-05,
      "loss": 0.9656,
      "step": 1029
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.488713502883911,
      "learning_rate": 1.999842410281158e-05,
      "loss": 1.0696,
      "step": 1030
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.955962657928467,
      "learning_rate": 1.999840421839908e-05,
      "loss": 0.9204,
      "step": 1031
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1887598037719727,
      "learning_rate": 1.9998384209332722e-05,
      "loss": 0.986,
      "step": 1032
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5542821884155273,
      "learning_rate": 1.9998364075612754e-05,
      "loss": 0.89,
      "step": 1033
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.245013475418091,
      "learning_rate": 1.9998343817239433e-05,
      "loss": 0.9064,
      "step": 1034
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8431363105773926,
      "learning_rate": 1.9998323434213003e-05,
      "loss": 1.0229,
      "step": 1035
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3743443489074707,
      "learning_rate": 1.999830292653372e-05,
      "loss": 0.9883,
      "step": 1036
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.161116600036621,
      "learning_rate": 1.9998282294201846e-05,
      "loss": 0.9729,
      "step": 1037
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1244313716888428,
      "learning_rate": 1.9998261537217634e-05,
      "loss": 1.0001,
      "step": 1038
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.505965232849121,
      "learning_rate": 1.999824065558134e-05,
      "loss": 1.1393,
      "step": 1039
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3131752014160156,
      "learning_rate": 1.999821964929323e-05,
      "loss": 0.917,
      "step": 1040
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4016125202178955,
      "learning_rate": 1.999819851835356e-05,
      "loss": 0.9498,
      "step": 1041
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6381728649139404,
      "learning_rate": 1.9998177262762598e-05,
      "loss": 1.0213,
      "step": 1042
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2630906105041504,
      "learning_rate": 1.999815588252061e-05,
      "loss": 0.9796,
      "step": 1043
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.395596981048584,
      "learning_rate": 1.9998134377627854e-05,
      "loss": 0.8485,
      "step": 1044
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4095993041992188,
      "learning_rate": 1.999811274808461e-05,
      "loss": 0.9826,
      "step": 1045
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3860652446746826,
      "learning_rate": 1.9998090993891137e-05,
      "loss": 0.7605,
      "step": 1046
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.711824417114258,
      "learning_rate": 1.9998069115047714e-05,
      "loss": 0.7648,
      "step": 1047
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.7195279598236084,
      "learning_rate": 1.9998047111554608e-05,
      "loss": 0.9691,
      "step": 1048
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.159280300140381,
      "learning_rate": 1.9998024983412095e-05,
      "loss": 0.9402,
      "step": 1049
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.243244171142578,
      "learning_rate": 1.9998002730620455e-05,
      "loss": 0.7858,
      "step": 1050
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8792017698287964,
      "learning_rate": 1.9997980353179965e-05,
      "loss": 0.8963,
      "step": 1051
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.333859443664551,
      "learning_rate": 1.9997957851090895e-05,
      "loss": 0.9705,
      "step": 1052
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4171221256256104,
      "learning_rate": 1.9997935224353538e-05,
      "loss": 1.0271,
      "step": 1053
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.063269853591919,
      "learning_rate": 1.9997912472968165e-05,
      "loss": 0.7065,
      "step": 1054
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3938937187194824,
      "learning_rate": 1.999788959693507e-05,
      "loss": 0.8222,
      "step": 1055
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.502145290374756,
      "learning_rate": 1.9997866596254526e-05,
      "loss": 0.9932,
      "step": 1056
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.122746229171753,
      "learning_rate": 1.9997843470926833e-05,
      "loss": 0.7253,
      "step": 1057
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.367216110229492,
      "learning_rate": 1.999782022095227e-05,
      "loss": 1.0482,
      "step": 1058
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4551141262054443,
      "learning_rate": 1.999779684633113e-05,
      "loss": 0.8366,
      "step": 1059
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.191899538040161,
      "learning_rate": 1.9997773347063703e-05,
      "loss": 0.9669,
      "step": 1060
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.286536693572998,
      "learning_rate": 1.9997749723150283e-05,
      "loss": 0.7918,
      "step": 1061
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2694783210754395,
      "learning_rate": 1.9997725974591164e-05,
      "loss": 0.7698,
      "step": 1062
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8688554763793945,
      "learning_rate": 1.9997702101386644e-05,
      "loss": 0.8274,
      "step": 1063
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.112438678741455,
      "learning_rate": 1.9997678103537018e-05,
      "loss": 0.8411,
      "step": 1064
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3570494651794434,
      "learning_rate": 1.999765398104259e-05,
      "loss": 0.7899,
      "step": 1065
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.314091920852661,
      "learning_rate": 1.999762973390366e-05,
      "loss": 0.8585,
      "step": 1066
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.042515754699707,
      "learning_rate": 1.999760536212052e-05,
      "loss": 0.9327,
      "step": 1067
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.660369396209717,
      "learning_rate": 1.9997580865693488e-05,
      "loss": 1.0083,
      "step": 1068
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4432475566864014,
      "learning_rate": 1.9997556244622857e-05,
      "loss": 0.819,
      "step": 1069
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3953611850738525,
      "learning_rate": 1.9997531498908943e-05,
      "loss": 0.868,
      "step": 1070
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8767056465148926,
      "learning_rate": 1.999750662855205e-05,
      "loss": 0.7851,
      "step": 1071
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9155421257019043,
      "learning_rate": 1.999748163355249e-05,
      "loss": 1.0463,
      "step": 1072
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.321767568588257,
      "learning_rate": 1.9997456513910575e-05,
      "loss": 0.8572,
      "step": 1073
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6389636993408203,
      "learning_rate": 1.9997431269626617e-05,
      "loss": 0.9559,
      "step": 1074
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4063451290130615,
      "learning_rate": 1.999740590070093e-05,
      "loss": 0.7777,
      "step": 1075
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.571638345718384,
      "learning_rate": 1.9997380407133835e-05,
      "loss": 0.9511,
      "step": 1076
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8865456581115723,
      "learning_rate": 1.999735478892564e-05,
      "loss": 0.9996,
      "step": 1077
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3738832473754883,
      "learning_rate": 1.9997329046076673e-05,
      "loss": 1.04,
      "step": 1078
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4944751262664795,
      "learning_rate": 1.999730317858725e-05,
      "loss": 0.8896,
      "step": 1079
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.597644090652466,
      "learning_rate": 1.99972771864577e-05,
      "loss": 0.8387,
      "step": 1080
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1737277507781982,
      "learning_rate": 1.9997251069688343e-05,
      "loss": 1.0306,
      "step": 1081
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.8586273193359375,
      "learning_rate": 1.9997224828279504e-05,
      "loss": 0.7666,
      "step": 1082
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1517269611358643,
      "learning_rate": 1.999719846223151e-05,
      "loss": 0.9255,
      "step": 1083
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.039755344390869,
      "learning_rate": 1.999717197154469e-05,
      "loss": 1.013,
      "step": 1084
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3018312454223633,
      "learning_rate": 1.9997145356219376e-05,
      "loss": 1.0037,
      "step": 1085
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2381603717803955,
      "learning_rate": 1.9997118616255898e-05,
      "loss": 0.9363,
      "step": 1086
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3726816177368164,
      "learning_rate": 1.999709175165459e-05,
      "loss": 0.9483,
      "step": 1087
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0487442016601562,
      "learning_rate": 1.9997064762415786e-05,
      "loss": 0.8518,
      "step": 1088
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.161224126815796,
      "learning_rate": 1.9997037648539822e-05,
      "loss": 0.8524,
      "step": 1089
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3803205490112305,
      "learning_rate": 1.9997010410027043e-05,
      "loss": 1.1256,
      "step": 1090
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.39127779006958,
      "learning_rate": 1.9996983046877777e-05,
      "loss": 0.8676,
      "step": 1091
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1663191318511963,
      "learning_rate": 1.9996955559092376e-05,
      "loss": 0.7891,
      "step": 1092
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.387831449508667,
      "learning_rate": 1.9996927946671173e-05,
      "loss": 0.7331,
      "step": 1093
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4758055210113525,
      "learning_rate": 1.999690020961452e-05,
      "loss": 0.9477,
      "step": 1094
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.33884596824646,
      "learning_rate": 1.999687234792276e-05,
      "loss": 1.0405,
      "step": 1095
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.079026460647583,
      "learning_rate": 1.999684436159624e-05,
      "loss": 0.9163,
      "step": 1096
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4745805263519287,
      "learning_rate": 1.9996816250635306e-05,
      "loss": 0.9455,
      "step": 1097
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.040314197540283,
      "learning_rate": 1.9996788015040315e-05,
      "loss": 0.7261,
      "step": 1098
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3390021324157715,
      "learning_rate": 1.9996759654811616e-05,
      "loss": 0.9893,
      "step": 1099
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.305917501449585,
      "learning_rate": 1.999673116994956e-05,
      "loss": 0.9457,
      "step": 1100
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.295480966567993,
      "learning_rate": 1.9996702560454505e-05,
      "loss": 0.9599,
      "step": 1101
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4746947288513184,
      "learning_rate": 1.9996673826326807e-05,
      "loss": 0.7666,
      "step": 1102
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.613006353378296,
      "learning_rate": 1.9996644967566828e-05,
      "loss": 0.8619,
      "step": 1103
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.4676616191864014,
      "learning_rate": 1.999661598417492e-05,
      "loss": 0.9815,
      "step": 1104
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.276594877243042,
      "learning_rate": 1.999658687615145e-05,
      "loss": 0.8772,
      "step": 1105
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.594017744064331,
      "learning_rate": 1.9996557643496778e-05,
      "loss": 0.9679,
      "step": 1106
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.292070150375366,
      "learning_rate": 1.999652828621127e-05,
      "loss": 0.777,
      "step": 1107
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2335686683654785,
      "learning_rate": 1.9996498804295292e-05,
      "loss": 0.9557,
      "step": 1108
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.3808937072753906,
      "learning_rate": 1.9996469197749212e-05,
      "loss": 0.9257,
      "step": 1109
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1796703338623047,
      "learning_rate": 1.99964394665734e-05,
      "loss": 0.8281,
      "step": 1110
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.9500982761383057,
      "learning_rate": 1.9996409610768223e-05,
      "loss": 0.8326,
      "step": 1111
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9038772583007812,
      "learning_rate": 1.9996379630334055e-05,
      "loss": 1.0513,
      "step": 1112
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.141040563583374,
      "learning_rate": 1.999634952527127e-05,
      "loss": 0.8651,
      "step": 1113
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.450162172317505,
      "learning_rate": 1.9996319295580243e-05,
      "loss": 0.9892,
      "step": 1114
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.368626832962036,
      "learning_rate": 1.9996288941261355e-05,
      "loss": 0.9563,
      "step": 1115
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.332430362701416,
      "learning_rate": 1.999625846231498e-05,
      "loss": 0.8312,
      "step": 1116
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.092228651046753,
      "learning_rate": 1.9996227858741494e-05,
      "loss": 0.8192,
      "step": 1117
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3413116931915283,
      "learning_rate": 1.9996197130541288e-05,
      "loss": 0.8542,
      "step": 1118
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0190370082855225,
      "learning_rate": 1.9996166277714738e-05,
      "loss": 0.7633,
      "step": 1119
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5407073497772217,
      "learning_rate": 1.999613530026223e-05,
      "loss": 0.9189,
      "step": 1120
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.101924419403076,
      "learning_rate": 1.9996104198184155e-05,
      "loss": 0.8706,
      "step": 1121
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3709092140197754,
      "learning_rate": 1.9996072971480895e-05,
      "loss": 0.7391,
      "step": 1122
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1639537811279297,
      "learning_rate": 1.9996041620152843e-05,
      "loss": 1.0123,
      "step": 1123
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.7472803592681885,
      "learning_rate": 1.9996010144200387e-05,
      "loss": 0.7435,
      "step": 1124
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.523120164871216,
      "learning_rate": 1.9995978543623922e-05,
      "loss": 1.0704,
      "step": 1125
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3157129287719727,
      "learning_rate": 1.9995946818423835e-05,
      "loss": 0.8622,
      "step": 1126
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2208781242370605,
      "learning_rate": 1.9995914968600533e-05,
      "loss": 0.8255,
      "step": 1127
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0736122131347656,
      "learning_rate": 1.999588299415441e-05,
      "loss": 0.8503,
      "step": 1128
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.188123941421509,
      "learning_rate": 1.9995850895085854e-05,
      "loss": 0.8282,
      "step": 1129
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0180680751800537,
      "learning_rate": 1.999581867139528e-05,
      "loss": 0.8622,
      "step": 1130
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.722723960876465,
      "learning_rate": 1.9995786323083077e-05,
      "loss": 0.8576,
      "step": 1131
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.523627519607544,
      "learning_rate": 1.9995753850149656e-05,
      "loss": 0.9122,
      "step": 1132
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1660544872283936,
      "learning_rate": 1.9995721252595423e-05,
      "loss": 0.81,
      "step": 1133
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8266772031784058,
      "learning_rate": 1.999568853042078e-05,
      "loss": 0.7354,
      "step": 1134
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1019062995910645,
      "learning_rate": 1.9995655683626135e-05,
      "loss": 0.7804,
      "step": 1135
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.04512357711792,
      "learning_rate": 1.9995622712211898e-05,
      "loss": 0.6355,
      "step": 1136
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3918845653533936,
      "learning_rate": 1.999558961617848e-05,
      "loss": 0.7256,
      "step": 1137
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1808316707611084,
      "learning_rate": 1.9995556395526302e-05,
      "loss": 0.8314,
      "step": 1138
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1670897006988525,
      "learning_rate": 1.9995523050255764e-05,
      "loss": 0.8078,
      "step": 1139
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.175096273422241,
      "learning_rate": 1.999548958036729e-05,
      "loss": 0.893,
      "step": 1140
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5602526664733887,
      "learning_rate": 1.9995455985861297e-05,
      "loss": 0.9578,
      "step": 1141
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1048433780670166,
      "learning_rate": 1.99954222667382e-05,
      "loss": 0.5576,
      "step": 1142
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2984049320220947,
      "learning_rate": 1.9995388422998424e-05,
      "loss": 0.8974,
      "step": 1143
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2766976356506348,
      "learning_rate": 1.9995354454642386e-05,
      "loss": 0.7115,
      "step": 1144
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.67962908744812,
      "learning_rate": 1.9995320361670515e-05,
      "loss": 0.9234,
      "step": 1145
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.895491600036621,
      "learning_rate": 1.999528614408323e-05,
      "loss": 0.7609,
      "step": 1146
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5385329723358154,
      "learning_rate": 1.9995251801880966e-05,
      "loss": 0.8621,
      "step": 1147
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3324673175811768,
      "learning_rate": 1.9995217335064144e-05,
      "loss": 0.7656,
      "step": 1148
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1976938247680664,
      "learning_rate": 1.9995182743633198e-05,
      "loss": 0.948,
      "step": 1149
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3729183673858643,
      "learning_rate": 1.9995148027588557e-05,
      "loss": 0.8647,
      "step": 1150
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3955304622650146,
      "learning_rate": 1.999511318693065e-05,
      "loss": 1.0144,
      "step": 1151
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9984513521194458,
      "learning_rate": 1.9995078221659917e-05,
      "loss": 0.8848,
      "step": 1152
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.332023859024048,
      "learning_rate": 1.9995043131776793e-05,
      "loss": 0.8715,
      "step": 1153
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.642184257507324,
      "learning_rate": 1.9995007917281716e-05,
      "loss": 0.8901,
      "step": 1154
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.474250078201294,
      "learning_rate": 1.9994972578175123e-05,
      "loss": 0.8566,
      "step": 1155
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3573272228240967,
      "learning_rate": 1.9994937114457453e-05,
      "loss": 0.7479,
      "step": 1156
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3905577659606934,
      "learning_rate": 1.9994901526129152e-05,
      "loss": 0.9755,
      "step": 1157
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6091725826263428,
      "learning_rate": 1.9994865813190666e-05,
      "loss": 1.0009,
      "step": 1158
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.823626756668091,
      "learning_rate": 1.9994829975642428e-05,
      "loss": 1.1185,
      "step": 1159
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0651965141296387,
      "learning_rate": 1.99947940134849e-05,
      "loss": 0.8221,
      "step": 1160
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.348629951477051,
      "learning_rate": 1.9994757926718523e-05,
      "loss": 0.9896,
      "step": 1161
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.184816837310791,
      "learning_rate": 1.9994721715343747e-05,
      "loss": 0.755,
      "step": 1162
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3199615478515625,
      "learning_rate": 1.999468537936102e-05,
      "loss": 0.7554,
      "step": 1163
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6910786628723145,
      "learning_rate": 1.9994648918770805e-05,
      "loss": 0.761,
      "step": 1164
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2779877185821533,
      "learning_rate": 1.9994612333573548e-05,
      "loss": 0.9801,
      "step": 1165
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0876693725585938,
      "learning_rate": 1.999457562376971e-05,
      "loss": 0.782,
      "step": 1166
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.389376401901245,
      "learning_rate": 1.9994538789359745e-05,
      "loss": 0.7737,
      "step": 1167
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4637858867645264,
      "learning_rate": 1.999450183034411e-05,
      "loss": 0.9369,
      "step": 1168
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.181694746017456,
      "learning_rate": 1.999446474672327e-05,
      "loss": 0.8261,
      "step": 1169
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5079848766326904,
      "learning_rate": 1.9994427538497693e-05,
      "loss": 0.9507,
      "step": 1170
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6165671348571777,
      "learning_rate": 1.999439020566783e-05,
      "loss": 0.786,
      "step": 1171
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1650142669677734,
      "learning_rate": 1.999435274823416e-05,
      "loss": 0.8237,
      "step": 1172
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.170030117034912,
      "learning_rate": 1.9994315166197135e-05,
      "loss": 0.9274,
      "step": 1173
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.113377094268799,
      "learning_rate": 1.9994277459557234e-05,
      "loss": 0.7563,
      "step": 1174
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.361104726791382,
      "learning_rate": 1.9994239628314928e-05,
      "loss": 0.8083,
      "step": 1175
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6916919946670532,
      "learning_rate": 1.9994201672470684e-05,
      "loss": 0.6926,
      "step": 1176
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.43784761428833,
      "learning_rate": 1.9994163592024972e-05,
      "loss": 0.9109,
      "step": 1177
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4354958534240723,
      "learning_rate": 1.9994125386978278e-05,
      "loss": 0.971,
      "step": 1178
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.402939796447754,
      "learning_rate": 1.999408705733107e-05,
      "loss": 0.7376,
      "step": 1179
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.4019370079040527,
      "learning_rate": 1.9994048603083826e-05,
      "loss": 0.7922,
      "step": 1180
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6017708778381348,
      "learning_rate": 1.9994010024237024e-05,
      "loss": 1.1055,
      "step": 1181
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.421807289123535,
      "learning_rate": 1.9993971320791153e-05,
      "loss": 1.0247,
      "step": 1182
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.203589916229248,
      "learning_rate": 1.999393249274669e-05,
      "loss": 0.7655,
      "step": 1183
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.693824291229248,
      "learning_rate": 1.9993893540104117e-05,
      "loss": 1.0346,
      "step": 1184
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5037717819213867,
      "learning_rate": 1.999385446286392e-05,
      "loss": 0.8908,
      "step": 1185
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.286957263946533,
      "learning_rate": 1.9993815261026595e-05,
      "loss": 0.9107,
      "step": 1186
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.9321181774139404,
      "learning_rate": 1.999377593459262e-05,
      "loss": 0.9777,
      "step": 1187
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1486992835998535,
      "learning_rate": 1.999373648356249e-05,
      "loss": 0.7637,
      "step": 1188
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2139029502868652,
      "learning_rate": 1.9993696907936697e-05,
      "loss": 0.8476,
      "step": 1189
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.6424801349639893,
      "learning_rate": 1.9993657207715732e-05,
      "loss": 0.8026,
      "step": 1190
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.23602032661438,
      "learning_rate": 1.9993617382900093e-05,
      "loss": 0.9496,
      "step": 1191
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.258272647857666,
      "learning_rate": 1.9993577433490278e-05,
      "loss": 0.8242,
      "step": 1192
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.8189594745635986,
      "learning_rate": 1.9993537359486775e-05,
      "loss": 0.9772,
      "step": 1193
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3605310916900635,
      "learning_rate": 1.9993497160890096e-05,
      "loss": 0.8541,
      "step": 1194
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.5395846366882324,
      "learning_rate": 1.9993456837700735e-05,
      "loss": 0.8688,
      "step": 1195
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2490997314453125,
      "learning_rate": 1.9993416389919198e-05,
      "loss": 0.9602,
      "step": 1196
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1922554969787598,
      "learning_rate": 1.9993375817545988e-05,
      "loss": 0.8769,
      "step": 1197
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3638525009155273,
      "learning_rate": 1.999333512058161e-05,
      "loss": 0.8492,
      "step": 1198
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2011775970458984,
      "learning_rate": 1.999329429902657e-05,
      "loss": 0.8742,
      "step": 1199
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.1650032997131348,
      "learning_rate": 1.9993253352881385e-05,
      "loss": 0.8476,
      "step": 1200
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3116397857666016,
      "learning_rate": 1.9993212282146556e-05,
      "loss": 0.795,
      "step": 1201
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0883889198303223,
      "learning_rate": 1.99931710868226e-05,
      "loss": 0.9609,
      "step": 1202
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9121267795562744,
      "learning_rate": 1.9993129766910027e-05,
      "loss": 0.7647,
      "step": 1203
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.217317819595337,
      "learning_rate": 1.9993088322409356e-05,
      "loss": 0.7624,
      "step": 1204
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.575193166732788,
      "learning_rate": 1.99930467533211e-05,
      "loss": 1.0172,
      "step": 1205
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.197516441345215,
      "learning_rate": 1.9993005059645784e-05,
      "loss": 1.0394,
      "step": 1206
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0670621395111084,
      "learning_rate": 1.9992963241383917e-05,
      "loss": 0.8489,
      "step": 1207
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2024857997894287,
      "learning_rate": 1.999292129853603e-05,
      "loss": 0.8831,
      "step": 1208
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2186803817749023,
      "learning_rate": 1.9992879231102642e-05,
      "loss": 0.9184,
      "step": 1209
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.789914846420288,
      "learning_rate": 1.999283703908428e-05,
      "loss": 0.9886,
      "step": 1210
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.122426986694336,
      "learning_rate": 1.9992794722481465e-05,
      "loss": 0.7217,
      "step": 1211
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1555368900299072,
      "learning_rate": 1.9992752281294727e-05,
      "loss": 0.8112,
      "step": 1212
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8644723892211914,
      "learning_rate": 1.9992709715524598e-05,
      "loss": 0.9572,
      "step": 1213
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9670331478118896,
      "learning_rate": 1.9992667025171606e-05,
      "loss": 0.7397,
      "step": 1214
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3196709156036377,
      "learning_rate": 1.9992624210236283e-05,
      "loss": 0.8775,
      "step": 1215
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.956457495689392,
      "learning_rate": 1.9992581270719164e-05,
      "loss": 0.7943,
      "step": 1216
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9623984098434448,
      "learning_rate": 1.999253820662078e-05,
      "loss": 0.9427,
      "step": 1217
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9398199319839478,
      "learning_rate": 1.9992495017941677e-05,
      "loss": 0.9173,
      "step": 1218
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.912971258163452,
      "learning_rate": 1.9992451704682384e-05,
      "loss": 0.8817,
      "step": 1219
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2705206871032715,
      "learning_rate": 1.999240826684345e-05,
      "loss": 0.8381,
      "step": 1220
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6191964149475098,
      "learning_rate": 1.999236470442541e-05,
      "loss": 0.8585,
      "step": 1221
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.064499616622925,
      "learning_rate": 1.9992321017428807e-05,
      "loss": 0.8023,
      "step": 1222
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5895400047302246,
      "learning_rate": 1.9992277205854186e-05,
      "loss": 1.0033,
      "step": 1223
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9776784181594849,
      "learning_rate": 1.9992233269702096e-05,
      "loss": 0.8096,
      "step": 1224
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0252063274383545,
      "learning_rate": 1.9992189208973082e-05,
      "loss": 0.8421,
      "step": 1225
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8195230960845947,
      "learning_rate": 1.99921450236677e-05,
      "loss": 0.7353,
      "step": 1226
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1752524375915527,
      "learning_rate": 1.999210071378649e-05,
      "loss": 0.8593,
      "step": 1227
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.101597547531128,
      "learning_rate": 1.9992056279330013e-05,
      "loss": 0.9059,
      "step": 1228
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.304426908493042,
      "learning_rate": 1.999201172029882e-05,
      "loss": 0.929,
      "step": 1229
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8705590963363647,
      "learning_rate": 1.9991967036693465e-05,
      "loss": 0.828,
      "step": 1230
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.054888963699341,
      "learning_rate": 1.9991922228514508e-05,
      "loss": 0.9598,
      "step": 1231
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3977091312408447,
      "learning_rate": 1.9991877295762504e-05,
      "loss": 0.901,
      "step": 1232
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4372196197509766,
      "learning_rate": 1.9991832238438014e-05,
      "loss": 0.9138,
      "step": 1233
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.177037000656128,
      "learning_rate": 1.9991787056541606e-05,
      "loss": 0.8505,
      "step": 1234
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8506920337677002,
      "learning_rate": 1.9991741750073837e-05,
      "loss": 0.9379,
      "step": 1235
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.579043388366699,
      "learning_rate": 1.999169631903527e-05,
      "loss": 1.005,
      "step": 1236
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.080195665359497,
      "learning_rate": 1.999165076342648e-05,
      "loss": 0.9082,
      "step": 1237
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.145644426345825,
      "learning_rate": 1.9991605083248028e-05,
      "loss": 0.8087,
      "step": 1238
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0026628971099854,
      "learning_rate": 1.9991559278500484e-05,
      "loss": 0.7278,
      "step": 1239
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.493753671646118,
      "learning_rate": 1.999151334918442e-05,
      "loss": 0.8792,
      "step": 1240
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5224239826202393,
      "learning_rate": 1.999146729530041e-05,
      "loss": 0.9466,
      "step": 1241
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3716413974761963,
      "learning_rate": 1.999142111684903e-05,
      "loss": 0.9059,
      "step": 1242
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.763392686843872,
      "learning_rate": 1.9991374813830847e-05,
      "loss": 0.8799,
      "step": 1243
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2024548053741455,
      "learning_rate": 1.9991328386246447e-05,
      "loss": 0.7273,
      "step": 1244
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2554163932800293,
      "learning_rate": 1.9991281834096407e-05,
      "loss": 0.898,
      "step": 1245
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3151779174804688,
      "learning_rate": 1.9991235157381302e-05,
      "loss": 0.9295,
      "step": 1246
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3401753902435303,
      "learning_rate": 1.999118835610172e-05,
      "loss": 0.8441,
      "step": 1247
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5479421615600586,
      "learning_rate": 1.9991141430258244e-05,
      "loss": 0.8052,
      "step": 1248
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.803499221801758,
      "learning_rate": 1.999109437985146e-05,
      "loss": 0.812,
      "step": 1249
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2033331394195557,
      "learning_rate": 1.999104720488195e-05,
      "loss": 0.9611,
      "step": 1250
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1682205200195312,
      "learning_rate": 1.9990999905350305e-05,
      "loss": 0.8337,
      "step": 1251
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.257291316986084,
      "learning_rate": 1.999095248125711e-05,
      "loss": 0.8927,
      "step": 1252
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.492009401321411,
      "learning_rate": 1.9990904932602966e-05,
      "loss": 1.0798,
      "step": 1253
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.26436448097229,
      "learning_rate": 1.9990857259388456e-05,
      "loss": 0.7353,
      "step": 1254
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.190441846847534,
      "learning_rate": 1.9990809461614182e-05,
      "loss": 0.8216,
      "step": 1255
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4845664501190186,
      "learning_rate": 1.9990761539280736e-05,
      "loss": 0.9726,
      "step": 1256
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9490392208099365,
      "learning_rate": 1.9990713492388716e-05,
      "loss": 0.7299,
      "step": 1257
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4712843894958496,
      "learning_rate": 1.999066532093872e-05,
      "loss": 1.0159,
      "step": 1258
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9364635944366455,
      "learning_rate": 1.999061702493135e-05,
      "loss": 0.6472,
      "step": 1259
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.312035083770752,
      "learning_rate": 1.9990568604367207e-05,
      "loss": 0.8274,
      "step": 1260
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.27276349067688,
      "learning_rate": 1.9990520059246898e-05,
      "loss": 0.8545,
      "step": 1261
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.259852409362793,
      "learning_rate": 1.999047138957102e-05,
      "loss": 0.8675,
      "step": 1262
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6718065738677979,
      "learning_rate": 1.999042259534019e-05,
      "loss": 0.841,
      "step": 1263
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3096683025360107,
      "learning_rate": 1.9990373676555014e-05,
      "loss": 0.8574,
      "step": 1264
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.871803879737854,
      "learning_rate": 1.9990324633216095e-05,
      "loss": 0.8955,
      "step": 1265
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.015805244445801,
      "learning_rate": 1.999027546532405e-05,
      "loss": 0.8343,
      "step": 1266
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.484341859817505,
      "learning_rate": 1.999022617287949e-05,
      "loss": 0.8493,
      "step": 1267
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1836817264556885,
      "learning_rate": 1.9990176755883032e-05,
      "loss": 0.8069,
      "step": 1268
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.080164670944214,
      "learning_rate": 1.999012721433529e-05,
      "loss": 0.868,
      "step": 1269
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.94775390625,
      "learning_rate": 1.9990077548236885e-05,
      "loss": 0.9259,
      "step": 1270
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8740893602371216,
      "learning_rate": 1.999002775758843e-05,
      "loss": 0.6803,
      "step": 1271
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6312787532806396,
      "learning_rate": 1.9989977842390553e-05,
      "loss": 0.7282,
      "step": 1272
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.04114031791687,
      "learning_rate": 1.998992780264387e-05,
      "loss": 0.8466,
      "step": 1273
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.807124137878418,
      "learning_rate": 1.998987763834901e-05,
      "loss": 0.8043,
      "step": 1274
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.364823579788208,
      "learning_rate": 1.9989827349506594e-05,
      "loss": 0.9014,
      "step": 1275
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2352893352508545,
      "learning_rate": 1.998977693611725e-05,
      "loss": 0.7642,
      "step": 1276
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.335007667541504,
      "learning_rate": 1.9989726398181607e-05,
      "loss": 0.9009,
      "step": 1277
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0655956268310547,
      "learning_rate": 1.9989675735700296e-05,
      "loss": 0.703,
      "step": 1278
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0902650356292725,
      "learning_rate": 1.998962494867395e-05,
      "loss": 0.6969,
      "step": 1279
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6243693828582764,
      "learning_rate": 1.9989574037103202e-05,
      "loss": 0.85,
      "step": 1280
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1588923931121826,
      "learning_rate": 1.9989523000988682e-05,
      "loss": 0.9327,
      "step": 1281
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3479881286621094,
      "learning_rate": 1.998947184033103e-05,
      "loss": 0.9595,
      "step": 1282
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8825600147247314,
      "learning_rate": 1.998942055513088e-05,
      "loss": 1.0046,
      "step": 1283
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9417567253112793,
      "learning_rate": 1.998936914538888e-05,
      "loss": 0.7751,
      "step": 1284
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9090148210525513,
      "learning_rate": 1.998931761110566e-05,
      "loss": 0.9189,
      "step": 1285
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2907557487487793,
      "learning_rate": 1.9989265952281873e-05,
      "loss": 0.8749,
      "step": 1286
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2496883869171143,
      "learning_rate": 1.9989214168918156e-05,
      "loss": 0.7342,
      "step": 1287
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1979219913482666,
      "learning_rate": 1.9989162261015155e-05,
      "loss": 0.9001,
      "step": 1288
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.168403387069702,
      "learning_rate": 1.9989110228573523e-05,
      "loss": 0.8915,
      "step": 1289
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.7078874111175537,
      "learning_rate": 1.99890580715939e-05,
      "loss": 0.9685,
      "step": 1290
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9064253568649292,
      "learning_rate": 1.9989005790076942e-05,
      "loss": 0.7087,
      "step": 1291
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.36386775970459,
      "learning_rate": 1.99889533840233e-05,
      "loss": 0.9049,
      "step": 1292
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.030689239501953,
      "learning_rate": 1.9988900853433627e-05,
      "loss": 1.0477,
      "step": 1293
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.237128257751465,
      "learning_rate": 1.998884819830858e-05,
      "loss": 0.813,
      "step": 1294
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.004497528076172,
      "learning_rate": 1.9988795418648813e-05,
      "loss": 0.832,
      "step": 1295
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.326470136642456,
      "learning_rate": 1.9988742514454983e-05,
      "loss": 0.9228,
      "step": 1296
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2891411781311035,
      "learning_rate": 1.9988689485727747e-05,
      "loss": 0.929,
      "step": 1297
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.14656400680542,
      "learning_rate": 1.9988636332467776e-05,
      "loss": 0.6864,
      "step": 1298
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5103795528411865,
      "learning_rate": 1.9988583054675723e-05,
      "loss": 0.9756,
      "step": 1299
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3438313007354736,
      "learning_rate": 1.9988529652352256e-05,
      "loss": 0.879,
      "step": 1300
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.498013734817505,
      "learning_rate": 1.9988476125498042e-05,
      "loss": 1.0049,
      "step": 1301
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4105942249298096,
      "learning_rate": 1.998842247411375e-05,
      "loss": 0.6814,
      "step": 1302
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9290317296981812,
      "learning_rate": 1.9988368698200042e-05,
      "loss": 0.836,
      "step": 1303
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.4997358322143555,
      "learning_rate": 1.9988314797757593e-05,
      "loss": 0.8554,
      "step": 1304
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0460550785064697,
      "learning_rate": 1.9988260772787078e-05,
      "loss": 0.9339,
      "step": 1305
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1148626804351807,
      "learning_rate": 1.9988206623289163e-05,
      "loss": 0.8209,
      "step": 1306
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2841360569000244,
      "learning_rate": 1.9988152349264527e-05,
      "loss": 0.9111,
      "step": 1307
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8242769241333008,
      "learning_rate": 1.998809795071385e-05,
      "loss": 0.7479,
      "step": 1308
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4010894298553467,
      "learning_rate": 1.9988043427637804e-05,
      "loss": 0.8306,
      "step": 1309
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6726396083831787,
      "learning_rate": 1.9987988780037072e-05,
      "loss": 0.7137,
      "step": 1310
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2197189331054688,
      "learning_rate": 1.9987934007912337e-05,
      "loss": 0.813,
      "step": 1311
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.438638925552368,
      "learning_rate": 1.9987879111264277e-05,
      "loss": 0.7729,
      "step": 1312
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0144519805908203,
      "learning_rate": 1.9987824090093583e-05,
      "loss": 0.6687,
      "step": 1313
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.275131940841675,
      "learning_rate": 1.9987768944400934e-05,
      "loss": 0.8534,
      "step": 1314
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8227113485336304,
      "learning_rate": 1.9987713674187022e-05,
      "loss": 0.8014,
      "step": 1315
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0967156887054443,
      "learning_rate": 1.9987658279452534e-05,
      "loss": 0.7689,
      "step": 1316
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2650306224823,
      "learning_rate": 1.9987602760198162e-05,
      "loss": 0.9266,
      "step": 1317
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.120508909225464,
      "learning_rate": 1.9987547116424597e-05,
      "loss": 0.8579,
      "step": 1318
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0350449085235596,
      "learning_rate": 1.9987491348132537e-05,
      "loss": 0.8647,
      "step": 1319
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.443394899368286,
      "learning_rate": 1.998743545532267e-05,
      "loss": 0.7939,
      "step": 1320
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0967979431152344,
      "learning_rate": 1.99873794379957e-05,
      "loss": 0.9302,
      "step": 1321
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.426034927368164,
      "learning_rate": 1.9987323296152318e-05,
      "loss": 0.874,
      "step": 1322
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.105686664581299,
      "learning_rate": 1.998726702979323e-05,
      "loss": 0.9305,
      "step": 1323
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.024730682373047,
      "learning_rate": 1.9987210638919135e-05,
      "loss": 0.8437,
      "step": 1324
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1677756309509277,
      "learning_rate": 1.9987154123530734e-05,
      "loss": 0.9518,
      "step": 1325
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3360438346862793,
      "learning_rate": 1.9987097483628737e-05,
      "loss": 0.965,
      "step": 1326
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.250674247741699,
      "learning_rate": 1.9987040719213846e-05,
      "loss": 0.7418,
      "step": 1327
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9772491455078125,
      "learning_rate": 1.998698383028677e-05,
      "loss": 0.8186,
      "step": 1328
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8535512685775757,
      "learning_rate": 1.9986926816848218e-05,
      "loss": 0.8907,
      "step": 1329
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.775660276412964,
      "learning_rate": 1.99868696788989e-05,
      "loss": 0.7552,
      "step": 1330
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.565218210220337,
      "learning_rate": 1.9986812416439533e-05,
      "loss": 0.7462,
      "step": 1331
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.359036922454834,
      "learning_rate": 1.9986755029470824e-05,
      "loss": 0.8482,
      "step": 1332
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2826356887817383,
      "learning_rate": 1.9986697517993493e-05,
      "loss": 0.9393,
      "step": 1333
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.196211099624634,
      "learning_rate": 1.9986639882008255e-05,
      "loss": 0.9424,
      "step": 1334
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0020740032196045,
      "learning_rate": 1.9986582121515828e-05,
      "loss": 0.8397,
      "step": 1335
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.143423080444336,
      "learning_rate": 1.9986524236516937e-05,
      "loss": 0.6851,
      "step": 1336
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1188461780548096,
      "learning_rate": 1.9986466227012297e-05,
      "loss": 0.7386,
      "step": 1337
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.279566764831543,
      "learning_rate": 1.9986408093002633e-05,
      "loss": 0.7653,
      "step": 1338
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2763781547546387,
      "learning_rate": 1.9986349834488673e-05,
      "loss": 0.9737,
      "step": 1339
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3548648357391357,
      "learning_rate": 1.9986291451471143e-05,
      "loss": 0.8837,
      "step": 1340
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.156475782394409,
      "learning_rate": 1.9986232943950768e-05,
      "loss": 0.9677,
      "step": 1341
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8727140426635742,
      "learning_rate": 1.9986174311928276e-05,
      "loss": 0.8424,
      "step": 1342
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9229803085327148,
      "learning_rate": 1.9986115555404405e-05,
      "loss": 0.8628,
      "step": 1343
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.151350498199463,
      "learning_rate": 1.9986056674379882e-05,
      "loss": 0.7694,
      "step": 1344
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.328934907913208,
      "learning_rate": 1.998599766885544e-05,
      "loss": 0.7647,
      "step": 1345
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.086602210998535,
      "learning_rate": 1.998593853883182e-05,
      "loss": 0.8014,
      "step": 1346
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.344527244567871,
      "learning_rate": 1.9985879284309753e-05,
      "loss": 0.7452,
      "step": 1347
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.5420079231262207,
      "learning_rate": 1.9985819905289982e-05,
      "loss": 0.8912,
      "step": 1348
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.719214677810669,
      "learning_rate": 1.9985760401773248e-05,
      "loss": 0.9507,
      "step": 1349
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3747925758361816,
      "learning_rate": 1.998570077376029e-05,
      "loss": 0.8897,
      "step": 1350
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.120058298110962,
      "learning_rate": 1.998564102125185e-05,
      "loss": 0.7991,
      "step": 1351
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8508632183074951,
      "learning_rate": 1.998558114424868e-05,
      "loss": 0.8351,
      "step": 1352
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.960957646369934,
      "learning_rate": 1.9985521142751517e-05,
      "loss": 0.8821,
      "step": 1353
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.155815362930298,
      "learning_rate": 1.9985461016761115e-05,
      "loss": 0.8256,
      "step": 1354
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6493427753448486,
      "learning_rate": 1.9985400766278225e-05,
      "loss": 0.8663,
      "step": 1355
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.92195725440979,
      "learning_rate": 1.9985340391303596e-05,
      "loss": 0.7449,
      "step": 1356
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2149507999420166,
      "learning_rate": 1.9985279891837977e-05,
      "loss": 0.7857,
      "step": 1357
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8167805671691895,
      "learning_rate": 1.998521926788213e-05,
      "loss": 0.7324,
      "step": 1358
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.698359251022339,
      "learning_rate": 1.9985158519436804e-05,
      "loss": 0.767,
      "step": 1359
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3762199878692627,
      "learning_rate": 1.998509764650276e-05,
      "loss": 0.8894,
      "step": 1360
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4575605392456055,
      "learning_rate": 1.9985036649080756e-05,
      "loss": 0.8204,
      "step": 1361
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.434906482696533,
      "learning_rate": 1.9984975527171552e-05,
      "loss": 0.9299,
      "step": 1362
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.7537448406219482,
      "learning_rate": 1.998491428077591e-05,
      "loss": 0.8563,
      "step": 1363
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3621740341186523,
      "learning_rate": 1.9984852909894594e-05,
      "loss": 0.8377,
      "step": 1364
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.359677791595459,
      "learning_rate": 1.9984791414528367e-05,
      "loss": 0.7671,
      "step": 1365
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9433538913726807,
      "learning_rate": 1.9984729794678005e-05,
      "loss": 0.7566,
      "step": 1366
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2462754249572754,
      "learning_rate": 1.998466805034426e-05,
      "loss": 0.9627,
      "step": 1367
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.365280866622925,
      "learning_rate": 1.9984606181527917e-05,
      "loss": 0.9407,
      "step": 1368
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.441058874130249,
      "learning_rate": 1.998454418822974e-05,
      "loss": 0.8813,
      "step": 1369
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.360508918762207,
      "learning_rate": 1.99844820704505e-05,
      "loss": 0.7004,
      "step": 1370
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3370184898376465,
      "learning_rate": 1.9984419828190977e-05,
      "loss": 0.8589,
      "step": 1371
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.240863084793091,
      "learning_rate": 1.9984357461451946e-05,
      "loss": 0.9975,
      "step": 1372
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.253857374191284,
      "learning_rate": 1.998429497023418e-05,
      "loss": 0.8637,
      "step": 1373
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1325271129608154,
      "learning_rate": 1.9984232354538463e-05,
      "loss": 0.9472,
      "step": 1374
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9765818119049072,
      "learning_rate": 1.9984169614365573e-05,
      "loss": 0.7485,
      "step": 1375
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.171241521835327,
      "learning_rate": 1.9984106749716292e-05,
      "loss": 0.7962,
      "step": 1376
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.21195912361145,
      "learning_rate": 1.9984043760591403e-05,
      "loss": 0.9751,
      "step": 1377
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3543317317962646,
      "learning_rate": 1.9983980646991696e-05,
      "loss": 1.012,
      "step": 1378
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.071765661239624,
      "learning_rate": 1.9983917408917956e-05,
      "loss": 0.7772,
      "step": 1379
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.290609836578369,
      "learning_rate": 1.998385404637097e-05,
      "loss": 0.9057,
      "step": 1380
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.943160057067871,
      "learning_rate": 1.9983790559351525e-05,
      "loss": 0.6583,
      "step": 1381
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.932578682899475,
      "learning_rate": 1.998372694786042e-05,
      "loss": 0.9148,
      "step": 1382
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3959014415740967,
      "learning_rate": 1.998366321189844e-05,
      "loss": 0.9074,
      "step": 1383
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1262166500091553,
      "learning_rate": 1.9983599351466388e-05,
      "loss": 0.879,
      "step": 1384
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.3297078609466553,
      "learning_rate": 1.9983535366565053e-05,
      "loss": 0.9217,
      "step": 1385
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.965494990348816,
      "learning_rate": 1.9983471257195236e-05,
      "loss": 0.9475,
      "step": 1386
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.244307518005371,
      "learning_rate": 1.9983407023357735e-05,
      "loss": 0.7389,
      "step": 1387
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6303939819335938,
      "learning_rate": 1.998334266505335e-05,
      "loss": 0.8047,
      "step": 1388
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0692999362945557,
      "learning_rate": 1.998327818228289e-05,
      "loss": 0.8889,
      "step": 1389
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.148275136947632,
      "learning_rate": 1.998321357504715e-05,
      "loss": 0.9464,
      "step": 1390
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1133646965026855,
      "learning_rate": 1.9983148843346937e-05,
      "loss": 0.9245,
      "step": 1391
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9247456789016724,
      "learning_rate": 1.9983083987183065e-05,
      "loss": 0.8217,
      "step": 1392
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.107004404067993,
      "learning_rate": 1.9983019006556337e-05,
      "loss": 0.8507,
      "step": 1393
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1168744564056396,
      "learning_rate": 1.998295390146756e-05,
      "loss": 0.8846,
      "step": 1394
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.326822519302368,
      "learning_rate": 1.9982888671917554e-05,
      "loss": 0.7372,
      "step": 1395
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.180107593536377,
      "learning_rate": 1.998282331790713e-05,
      "loss": 0.81,
      "step": 1396
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9312138557434082,
      "learning_rate": 1.9982757839437098e-05,
      "loss": 0.8749,
      "step": 1397
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9578534364700317,
      "learning_rate": 1.9982692236508274e-05,
      "loss": 0.7981,
      "step": 1398
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8636670112609863,
      "learning_rate": 1.9982626509121484e-05,
      "loss": 0.8503,
      "step": 1399
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0360376834869385,
      "learning_rate": 1.9982560657277538e-05,
      "loss": 1.0036,
      "step": 1400
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.082550525665283,
      "learning_rate": 1.9982494680977265e-05,
      "loss": 1.0062,
      "step": 1401
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.035940170288086,
      "learning_rate": 1.9982428580221482e-05,
      "loss": 1.0042,
      "step": 1402
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2140510082244873,
      "learning_rate": 1.9982362355011017e-05,
      "loss": 0.8769,
      "step": 1403
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.038815975189209,
      "learning_rate": 1.998229600534669e-05,
      "loss": 0.8144,
      "step": 1404
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.388134479522705,
      "learning_rate": 1.998222953122934e-05,
      "loss": 0.8864,
      "step": 1405
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0093984603881836,
      "learning_rate": 1.9982162932659778e-05,
      "loss": 1.0397,
      "step": 1406
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9391536712646484,
      "learning_rate": 1.998209620963885e-05,
      "loss": 0.874,
      "step": 1407
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.6670846939086914,
      "learning_rate": 1.998202936216738e-05,
      "loss": 0.8462,
      "step": 1408
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.24808406829834,
      "learning_rate": 1.9981962390246204e-05,
      "loss": 0.7206,
      "step": 1409
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4698925018310547,
      "learning_rate": 1.9981895293876155e-05,
      "loss": 0.8538,
      "step": 1410
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.32609224319458,
      "learning_rate": 1.998182807305807e-05,
      "loss": 0.8572,
      "step": 1411
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.774049997329712,
      "learning_rate": 1.9981760727792792e-05,
      "loss": 0.8516,
      "step": 1412
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.147341728210449,
      "learning_rate": 1.9981693258081154e-05,
      "loss": 0.8435,
      "step": 1413
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3134028911590576,
      "learning_rate": 1.9981625663924e-05,
      "loss": 0.927,
      "step": 1414
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1746859550476074,
      "learning_rate": 1.998155794532217e-05,
      "loss": 0.8676,
      "step": 1415
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1769533157348633,
      "learning_rate": 1.9981490102276516e-05,
      "loss": 1.024,
      "step": 1416
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0068788528442383,
      "learning_rate": 1.9981422134787873e-05,
      "loss": 0.8731,
      "step": 1417
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3592147827148438,
      "learning_rate": 1.9981354042857095e-05,
      "loss": 0.6409,
      "step": 1418
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.207287311553955,
      "learning_rate": 1.9981285826485033e-05,
      "loss": 0.7559,
      "step": 1419
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7811951637268066,
      "learning_rate": 1.998121748567253e-05,
      "loss": 0.7816,
      "step": 1420
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0556726455688477,
      "learning_rate": 1.9981149020420445e-05,
      "loss": 1.0142,
      "step": 1421
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.165717601776123,
      "learning_rate": 1.998108043072963e-05,
      "loss": 0.9162,
      "step": 1422
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1355791091918945,
      "learning_rate": 1.9981011716600936e-05,
      "loss": 0.8815,
      "step": 1423
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9876612424850464,
      "learning_rate": 1.9980942878035226e-05,
      "loss": 0.8043,
      "step": 1424
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5028703212738037,
      "learning_rate": 1.9980873915033352e-05,
      "loss": 0.9204,
      "step": 1425
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0783565044403076,
      "learning_rate": 1.9980804827596175e-05,
      "loss": 0.8348,
      "step": 1426
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4310989379882812,
      "learning_rate": 1.9980735615724566e-05,
      "loss": 0.8614,
      "step": 1427
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2219161987304688,
      "learning_rate": 1.998066627941937e-05,
      "loss": 0.723,
      "step": 1428
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1135261058807373,
      "learning_rate": 1.998059681868147e-05,
      "loss": 0.7884,
      "step": 1429
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9281123876571655,
      "learning_rate": 1.9980527233511717e-05,
      "loss": 0.9495,
      "step": 1430
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8003442287445068,
      "learning_rate": 1.9980457523910988e-05,
      "loss": 0.7041,
      "step": 1431
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.722428798675537,
      "learning_rate": 1.998038768988015e-05,
      "loss": 0.7547,
      "step": 1432
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3156964778900146,
      "learning_rate": 1.998031773142007e-05,
      "loss": 0.8283,
      "step": 1433
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0558958053588867,
      "learning_rate": 1.9980247648531624e-05,
      "loss": 0.6291,
      "step": 1434
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.249913454055786,
      "learning_rate": 1.9980177441215684e-05,
      "loss": 0.8672,
      "step": 1435
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3934271335601807,
      "learning_rate": 1.9980107109473124e-05,
      "loss": 1.0344,
      "step": 1436
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8080096244812012,
      "learning_rate": 1.9980036653304827e-05,
      "loss": 0.8065,
      "step": 1437
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.280529499053955,
      "learning_rate": 1.9979966072711665e-05,
      "loss": 0.7542,
      "step": 1438
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9872018098831177,
      "learning_rate": 1.997989536769452e-05,
      "loss": 0.9138,
      "step": 1439
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.364560842514038,
      "learning_rate": 1.9979824538254276e-05,
      "loss": 0.9201,
      "step": 1440
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.121229648590088,
      "learning_rate": 1.997975358439181e-05,
      "loss": 0.8914,
      "step": 1441
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.203490972518921,
      "learning_rate": 1.9979682506108012e-05,
      "loss": 0.9198,
      "step": 1442
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3255105018615723,
      "learning_rate": 1.997961130340377e-05,
      "loss": 1.0282,
      "step": 1443
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3222501277923584,
      "learning_rate": 1.9979539976279964e-05,
      "loss": 0.8329,
      "step": 1444
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3006534576416016,
      "learning_rate": 1.997946852473749e-05,
      "loss": 0.7859,
      "step": 1445
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0708532333374023,
      "learning_rate": 1.9979396948777233e-05,
      "loss": 0.6724,
      "step": 1446
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9902007579803467,
      "learning_rate": 1.997932524840009e-05,
      "loss": 0.8766,
      "step": 1447
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.9222664833068848,
      "learning_rate": 1.9979253423606952e-05,
      "loss": 0.8837,
      "step": 1448
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.920893669128418,
      "learning_rate": 1.997918147439872e-05,
      "loss": 0.7956,
      "step": 1449
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9585260152816772,
      "learning_rate": 1.9979109400776283e-05,
      "loss": 0.6849,
      "step": 1450
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1277058124542236,
      "learning_rate": 1.9979037202740545e-05,
      "loss": 0.7889,
      "step": 1451
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9129104614257812,
      "learning_rate": 1.9978964880292405e-05,
      "loss": 0.8202,
      "step": 1452
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.5514400005340576,
      "learning_rate": 1.9978892433432764e-05,
      "loss": 0.8675,
      "step": 1453
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0302653312683105,
      "learning_rate": 1.9978819862162526e-05,
      "loss": 0.8428,
      "step": 1454
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.185041904449463,
      "learning_rate": 1.9978747166482595e-05,
      "loss": 0.8756,
      "step": 1455
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2428178787231445,
      "learning_rate": 1.9978674346393878e-05,
      "loss": 0.9379,
      "step": 1456
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2377419471740723,
      "learning_rate": 1.997860140189728e-05,
      "loss": 0.8687,
      "step": 1457
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0421159267425537,
      "learning_rate": 1.997852833299372e-05,
      "loss": 0.7845,
      "step": 1458
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8316705226898193,
      "learning_rate": 1.9978455139684097e-05,
      "loss": 0.86,
      "step": 1459
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.990789532661438,
      "learning_rate": 1.9978381821969327e-05,
      "loss": 0.8266,
      "step": 1460
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.3159685134887695,
      "learning_rate": 1.9978308379850326e-05,
      "loss": 0.9459,
      "step": 1461
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.941540002822876,
      "learning_rate": 1.9978234813328013e-05,
      "loss": 0.767,
      "step": 1462
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.384413480758667,
      "learning_rate": 1.99781611224033e-05,
      "loss": 1.0429,
      "step": 1463
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0302605628967285,
      "learning_rate": 1.9978087307077106e-05,
      "loss": 0.7516,
      "step": 1464
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.15533185005188,
      "learning_rate": 1.9978013367350352e-05,
      "loss": 0.7963,
      "step": 1465
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.167025327682495,
      "learning_rate": 1.997793930322396e-05,
      "loss": 0.7474,
      "step": 1466
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9547181129455566,
      "learning_rate": 1.9977865114698854e-05,
      "loss": 0.8037,
      "step": 1467
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.114837408065796,
      "learning_rate": 1.997779080177596e-05,
      "loss": 0.9167,
      "step": 1468
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2022790908813477,
      "learning_rate": 1.99777163644562e-05,
      "loss": 0.8538,
      "step": 1469
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9499748945236206,
      "learning_rate": 1.997764180274051e-05,
      "loss": 0.7045,
      "step": 1470
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.44927978515625,
      "learning_rate": 1.997756711662981e-05,
      "loss": 0.8015,
      "step": 1471
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.000530958175659,
      "learning_rate": 1.9977492306125037e-05,
      "loss": 0.9305,
      "step": 1472
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9083176851272583,
      "learning_rate": 1.9977417371227123e-05,
      "loss": 0.8144,
      "step": 1473
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1114327907562256,
      "learning_rate": 1.9977342311937004e-05,
      "loss": 0.9612,
      "step": 1474
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.043478488922119,
      "learning_rate": 1.997726712825561e-05,
      "loss": 0.7841,
      "step": 1475
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4438319206237793,
      "learning_rate": 1.9977191820183884e-05,
      "loss": 0.8001,
      "step": 1476
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0242764949798584,
      "learning_rate": 1.9977116387722765e-05,
      "loss": 0.7074,
      "step": 1477
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9934285879135132,
      "learning_rate": 1.9977040830873186e-05,
      "loss": 0.8487,
      "step": 1478
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.071017026901245,
      "learning_rate": 1.9976965149636095e-05,
      "loss": 0.8446,
      "step": 1479
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7480524778366089,
      "learning_rate": 1.9976889344012438e-05,
      "loss": 0.8423,
      "step": 1480
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.222317695617676,
      "learning_rate": 1.9976813414003152e-05,
      "loss": 0.8844,
      "step": 1481
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2737903594970703,
      "learning_rate": 1.9976737359609192e-05,
      "loss": 0.8097,
      "step": 1482
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0549700260162354,
      "learning_rate": 1.9976661180831503e-05,
      "loss": 0.9151,
      "step": 1483
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.4201531410217285,
      "learning_rate": 1.9976584877671038e-05,
      "loss": 0.8123,
      "step": 1484
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.505293846130371,
      "learning_rate": 1.9976508450128737e-05,
      "loss": 1.0033,
      "step": 1485
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.065796136856079,
      "learning_rate": 1.9976431898205564e-05,
      "loss": 0.7549,
      "step": 1486
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.914083480834961,
      "learning_rate": 1.997635522190247e-05,
      "loss": 1.0292,
      "step": 1487
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.141000270843506,
      "learning_rate": 1.997627842122041e-05,
      "loss": 0.9016,
      "step": 1488
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0881497859954834,
      "learning_rate": 1.9976201496160346e-05,
      "loss": 0.9577,
      "step": 1489
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.779639720916748,
      "learning_rate": 1.997612444672323e-05,
      "loss": 1.0561,
      "step": 1490
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0380897521972656,
      "learning_rate": 1.997604727291003e-05,
      "loss": 0.8225,
      "step": 1491
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.246157646179199,
      "learning_rate": 1.9975969974721702e-05,
      "loss": 0.962,
      "step": 1492
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1383588314056396,
      "learning_rate": 1.997589255215921e-05,
      "loss": 0.9498,
      "step": 1493
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8669142723083496,
      "learning_rate": 1.9975815005223524e-05,
      "loss": 0.8635,
      "step": 1494
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9585753679275513,
      "learning_rate": 1.9975737333915604e-05,
      "loss": 0.7811,
      "step": 1495
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.052074670791626,
      "learning_rate": 1.9975659538236428e-05,
      "loss": 0.9773,
      "step": 1496
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2226788997650146,
      "learning_rate": 1.9975581618186956e-05,
      "loss": 0.9199,
      "step": 1497
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.1809725761413574,
      "learning_rate": 1.9975503573768165e-05,
      "loss": 0.931,
      "step": 1498
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.112470865249634,
      "learning_rate": 1.9975425404981025e-05,
      "loss": 0.9754,
      "step": 1499
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1719958782196045,
      "learning_rate": 1.9975347111826516e-05,
      "loss": 0.8075,
      "step": 1500
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.131343364715576,
      "learning_rate": 1.997526869430561e-05,
      "loss": 0.8353,
      "step": 1501
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9498462677001953,
      "learning_rate": 1.9975190152419283e-05,
      "loss": 0.8393,
      "step": 1502
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9012415409088135,
      "learning_rate": 1.9975111486168517e-05,
      "loss": 0.8368,
      "step": 1503
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.412224531173706,
      "learning_rate": 1.9975032695554293e-05,
      "loss": 0.7356,
      "step": 1504
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.983511209487915,
      "learning_rate": 1.9974953780577594e-05,
      "loss": 0.9077,
      "step": 1505
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1442458629608154,
      "learning_rate": 1.99748747412394e-05,
      "loss": 1.0196,
      "step": 1506
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9656306505203247,
      "learning_rate": 1.9974795577540694e-05,
      "loss": 0.8967,
      "step": 1507
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.155992269515991,
      "learning_rate": 1.9974716289482476e-05,
      "loss": 0.737,
      "step": 1508
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.934535264968872,
      "learning_rate": 1.9974636877065722e-05,
      "loss": 0.9868,
      "step": 1509
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1184840202331543,
      "learning_rate": 1.9974557340291423e-05,
      "loss": 0.9645,
      "step": 1510
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1439599990844727,
      "learning_rate": 1.997447767916058e-05,
      "loss": 0.8534,
      "step": 1511
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.524571180343628,
      "learning_rate": 1.9974397893674178e-05,
      "loss": 0.8601,
      "step": 1512
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0072314739227295,
      "learning_rate": 1.9974317983833213e-05,
      "loss": 0.7582,
      "step": 1513
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.811074137687683,
      "learning_rate": 1.997423794963868e-05,
      "loss": 0.7942,
      "step": 1514
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3412044048309326,
      "learning_rate": 1.9974157791091583e-05,
      "loss": 0.8985,
      "step": 1515
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2082974910736084,
      "learning_rate": 1.9974077508192913e-05,
      "loss": 0.9461,
      "step": 1516
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0717062950134277,
      "learning_rate": 1.997399710094368e-05,
      "loss": 0.89,
      "step": 1517
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.333954334259033,
      "learning_rate": 1.9973916569344877e-05,
      "loss": 1.0611,
      "step": 1518
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1079750061035156,
      "learning_rate": 1.9973835913397515e-05,
      "loss": 0.823,
      "step": 1519
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1081717014312744,
      "learning_rate": 1.9973755133102597e-05,
      "loss": 0.8353,
      "step": 1520
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0796396732330322,
      "learning_rate": 1.9973674228461127e-05,
      "loss": 0.7748,
      "step": 1521
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0013504028320312,
      "learning_rate": 1.997359319947412e-05,
      "loss": 0.7808,
      "step": 1522
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.114996910095215,
      "learning_rate": 1.997351204614258e-05,
      "loss": 0.8657,
      "step": 1523
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.065028190612793,
      "learning_rate": 1.9973430768467527e-05,
      "loss": 0.6452,
      "step": 1524
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3170785903930664,
      "learning_rate": 1.9973349366449968e-05,
      "loss": 0.8268,
      "step": 1525
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.6163923740386963,
      "learning_rate": 1.9973267840090916e-05,
      "loss": 0.8524,
      "step": 1526
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.288658857345581,
      "learning_rate": 1.9973186189391392e-05,
      "loss": 0.7807,
      "step": 1527
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.1615724563598633,
      "learning_rate": 1.9973104414352415e-05,
      "loss": 0.9151,
      "step": 1528
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5070459842681885,
      "learning_rate": 1.9973022514975e-05,
      "loss": 0.8533,
      "step": 1529
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.083430767059326,
      "learning_rate": 1.997294049126017e-05,
      "loss": 0.945,
      "step": 1530
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9587806463241577,
      "learning_rate": 1.9972858343208946e-05,
      "loss": 0.9065,
      "step": 1531
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.386260986328125,
      "learning_rate": 1.9972776070822357e-05,
      "loss": 0.8375,
      "step": 1532
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9582428932189941,
      "learning_rate": 1.9972693674101424e-05,
      "loss": 0.7892,
      "step": 1533
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.284496307373047,
      "learning_rate": 1.9972611153047174e-05,
      "loss": 0.8754,
      "step": 1534
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.079176902770996,
      "learning_rate": 1.997252850766064e-05,
      "loss": 0.8333,
      "step": 1535
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0471131801605225,
      "learning_rate": 1.997244573794285e-05,
      "loss": 0.7055,
      "step": 1536
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8659627437591553,
      "learning_rate": 1.9972362843894837e-05,
      "loss": 0.7357,
      "step": 1537
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.923458456993103,
      "learning_rate": 1.997227982551763e-05,
      "loss": 0.6977,
      "step": 1538
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2105214595794678,
      "learning_rate": 1.997219668281227e-05,
      "loss": 0.8291,
      "step": 1539
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.096212387084961,
      "learning_rate": 1.997211341577979e-05,
      "loss": 0.8796,
      "step": 1540
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.020599603652954,
      "learning_rate": 1.997203002442123e-05,
      "loss": 0.7284,
      "step": 1541
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.043600559234619,
      "learning_rate": 1.997194650873763e-05,
      "loss": 0.7537,
      "step": 1542
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2016639709472656,
      "learning_rate": 1.997186286873003e-05,
      "loss": 0.9422,
      "step": 1543
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1568667888641357,
      "learning_rate": 1.997177910439947e-05,
      "loss": 0.721,
      "step": 1544
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.188638210296631,
      "learning_rate": 1.9971695215746996e-05,
      "loss": 0.77,
      "step": 1545
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.792357325553894,
      "learning_rate": 1.9971611202773657e-05,
      "loss": 0.8298,
      "step": 1546
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0890679359436035,
      "learning_rate": 1.99715270654805e-05,
      "loss": 0.7618,
      "step": 1547
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3860585689544678,
      "learning_rate": 1.997144280386857e-05,
      "loss": 0.9008,
      "step": 1548
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0864124298095703,
      "learning_rate": 1.9971358417938923e-05,
      "loss": 0.8361,
      "step": 1549
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.305755615234375,
      "learning_rate": 1.9971273907692604e-05,
      "loss": 0.7833,
      "step": 1550
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.400529384613037,
      "learning_rate": 1.9971189273130672e-05,
      "loss": 1.0391,
      "step": 1551
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.207249641418457,
      "learning_rate": 1.9971104514254184e-05,
      "loss": 1.014,
      "step": 1552
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.158567428588867,
      "learning_rate": 1.997101963106419e-05,
      "loss": 0.89,
      "step": 1553
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3988611698150635,
      "learning_rate": 1.9970934623561754e-05,
      "loss": 0.7914,
      "step": 1554
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4113664627075195,
      "learning_rate": 1.997084949174793e-05,
      "loss": 0.8935,
      "step": 1555
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0521905422210693,
      "learning_rate": 1.9970764235623787e-05,
      "loss": 1.0426,
      "step": 1556
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8569751977920532,
      "learning_rate": 1.997067885519038e-05,
      "loss": 0.7654,
      "step": 1557
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8444029092788696,
      "learning_rate": 1.9970593350448785e-05,
      "loss": 0.8277,
      "step": 1558
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0133392810821533,
      "learning_rate": 1.9970507721400052e-05,
      "loss": 0.9083,
      "step": 1559
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9046534299850464,
      "learning_rate": 1.9970421968045263e-05,
      "loss": 0.7943,
      "step": 1560
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9263949394226074,
      "learning_rate": 1.9970336090385476e-05,
      "loss": 0.8068,
      "step": 1561
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4323818683624268,
      "learning_rate": 1.9970250088421767e-05,
      "loss": 0.9156,
      "step": 1562
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9136974811553955,
      "learning_rate": 1.997016396215521e-05,
      "loss": 0.7421,
      "step": 1563
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3033134937286377,
      "learning_rate": 1.9970077711586878e-05,
      "loss": 1.0224,
      "step": 1564
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.057241916656494,
      "learning_rate": 1.9969991336717842e-05,
      "loss": 0.8309,
      "step": 1565
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.143099784851074,
      "learning_rate": 1.9969904837549184e-05,
      "loss": 0.7673,
      "step": 1566
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6815029382705688,
      "learning_rate": 1.996981821408198e-05,
      "loss": 0.9582,
      "step": 1567
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0251662731170654,
      "learning_rate": 1.996973146631731e-05,
      "loss": 0.7352,
      "step": 1568
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0950872898101807,
      "learning_rate": 1.9969644594256255e-05,
      "loss": 0.8645,
      "step": 1569
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7747372388839722,
      "learning_rate": 1.9969557597899902e-05,
      "loss": 0.8757,
      "step": 1570
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.484999895095825,
      "learning_rate": 1.9969470477249327e-05,
      "loss": 0.9699,
      "step": 1571
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.063124418258667,
      "learning_rate": 1.9969383232305626e-05,
      "loss": 0.8127,
      "step": 1572
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.170994758605957,
      "learning_rate": 1.996929586306988e-05,
      "loss": 0.6838,
      "step": 1573
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.779688596725464,
      "learning_rate": 1.9969208369543182e-05,
      "loss": 0.755,
      "step": 1574
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1615633964538574,
      "learning_rate": 1.9969120751726622e-05,
      "loss": 0.8534,
      "step": 1575
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8941069841384888,
      "learning_rate": 1.9969033009621293e-05,
      "loss": 0.9467,
      "step": 1576
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2525923252105713,
      "learning_rate": 1.9968945143228285e-05,
      "loss": 0.8013,
      "step": 1577
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.554527759552002,
      "learning_rate": 1.9968857152548696e-05,
      "loss": 0.7363,
      "step": 1578
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2785680294036865,
      "learning_rate": 1.9968769037583624e-05,
      "loss": 0.8498,
      "step": 1579
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.874155879020691,
      "learning_rate": 1.9968680798334166e-05,
      "loss": 0.8547,
      "step": 1580
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8598421812057495,
      "learning_rate": 1.9968592434801424e-05,
      "loss": 0.8315,
      "step": 1581
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.868424892425537,
      "learning_rate": 1.99685039469865e-05,
      "loss": 0.8089,
      "step": 1582
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8358720541000366,
      "learning_rate": 1.9968415334890493e-05,
      "loss": 0.5747,
      "step": 1583
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1781578063964844,
      "learning_rate": 1.9968326598514515e-05,
      "loss": 0.7983,
      "step": 1584
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.34885835647583,
      "learning_rate": 1.996823773785966e-05,
      "loss": 1.0063,
      "step": 1585
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7684051990509033,
      "learning_rate": 1.996814875292705e-05,
      "loss": 0.8573,
      "step": 1586
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0124292373657227,
      "learning_rate": 1.9968059643717786e-05,
      "loss": 0.7745,
      "step": 1587
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1922237873077393,
      "learning_rate": 1.996797041023298e-05,
      "loss": 0.7491,
      "step": 1588
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.876099944114685,
      "learning_rate": 1.996788105247375e-05,
      "loss": 0.8716,
      "step": 1589
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.3297762870788574,
      "learning_rate": 1.99677915704412e-05,
      "loss": 0.9191,
      "step": 1590
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8266769647598267,
      "learning_rate": 1.9967701964136453e-05,
      "loss": 0.8506,
      "step": 1591
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.4366962909698486,
      "learning_rate": 1.9967612233560626e-05,
      "loss": 0.7562,
      "step": 1592
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.033644199371338,
      "learning_rate": 1.9967522378714833e-05,
      "loss": 0.9019,
      "step": 1593
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.694018602371216,
      "learning_rate": 1.9967432399600202e-05,
      "loss": 1.0427,
      "step": 1594
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.399831771850586,
      "learning_rate": 1.996734229621785e-05,
      "loss": 0.7988,
      "step": 1595
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4160044193267822,
      "learning_rate": 1.9967252068568896e-05,
      "loss": 0.8325,
      "step": 1596
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2095632553100586,
      "learning_rate": 1.9967161716654472e-05,
      "loss": 0.8562,
      "step": 1597
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.7160778045654297,
      "learning_rate": 1.99670712404757e-05,
      "loss": 0.8241,
      "step": 1598
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0243306159973145,
      "learning_rate": 1.996698064003371e-05,
      "loss": 0.733,
      "step": 1599
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.103377342224121,
      "learning_rate": 1.9966889915329634e-05,
      "loss": 0.9131,
      "step": 1600
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.237752676010132,
      "learning_rate": 1.99667990663646e-05,
      "loss": 0.9726,
      "step": 1601
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.959752082824707,
      "learning_rate": 1.9966708093139738e-05,
      "loss": 0.6809,
      "step": 1602
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4354517459869385,
      "learning_rate": 1.996661699565619e-05,
      "loss": 0.8902,
      "step": 1603
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1217658519744873,
      "learning_rate": 1.9966525773915082e-05,
      "loss": 0.8461,
      "step": 1604
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.019906997680664,
      "learning_rate": 1.996643442791756e-05,
      "loss": 0.9149,
      "step": 1605
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6788800954818726,
      "learning_rate": 1.9966342957664758e-05,
      "loss": 0.8394,
      "step": 1606
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.104144811630249,
      "learning_rate": 1.996625136315782e-05,
      "loss": 0.8356,
      "step": 1607
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.109844923019409,
      "learning_rate": 1.996615964439788e-05,
      "loss": 0.695,
      "step": 1608
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9337471723556519,
      "learning_rate": 1.9966067801386093e-05,
      "loss": 0.7644,
      "step": 1609
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.200258731842041,
      "learning_rate": 1.9965975834123596e-05,
      "loss": 1.0222,
      "step": 1610
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0546882152557373,
      "learning_rate": 1.9965883742611533e-05,
      "loss": 0.9045,
      "step": 1611
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9291082620620728,
      "learning_rate": 1.9965791526851064e-05,
      "loss": 0.921,
      "step": 1612
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0675222873687744,
      "learning_rate": 1.9965699186843327e-05,
      "loss": 0.7651,
      "step": 1613
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.957495093345642,
      "learning_rate": 1.996560672258948e-05,
      "loss": 0.7842,
      "step": 1614
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.057788610458374,
      "learning_rate": 1.9965514134090674e-05,
      "loss": 0.7839,
      "step": 1615
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.113687753677368,
      "learning_rate": 1.996542142134806e-05,
      "loss": 0.9469,
      "step": 1616
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9088232517242432,
      "learning_rate": 1.99653285843628e-05,
      "loss": 0.838,
      "step": 1617
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.420234441757202,
      "learning_rate": 1.9965235623136046e-05,
      "loss": 0.9322,
      "step": 1618
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3346028327941895,
      "learning_rate": 1.996514253766896e-05,
      "loss": 0.8495,
      "step": 1619
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8050408363342285,
      "learning_rate": 1.99650493279627e-05,
      "loss": 0.8429,
      "step": 1620
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9674404859542847,
      "learning_rate": 1.9964955994018432e-05,
      "loss": 0.8087,
      "step": 1621
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.286215305328369,
      "learning_rate": 1.9964862535837317e-05,
      "loss": 0.7993,
      "step": 1622
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9593493938446045,
      "learning_rate": 1.996476895342052e-05,
      "loss": 0.8061,
      "step": 1623
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0241055488586426,
      "learning_rate": 1.9964675246769206e-05,
      "loss": 0.8771,
      "step": 1624
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9804273843765259,
      "learning_rate": 1.996458141588455e-05,
      "loss": 0.9403,
      "step": 1625
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8524878025054932,
      "learning_rate": 1.9964487460767715e-05,
      "loss": 0.8765,
      "step": 1626
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8709931373596191,
      "learning_rate": 1.9964393381419877e-05,
      "loss": 0.7014,
      "step": 1627
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.014432907104492,
      "learning_rate": 1.9964299177842203e-05,
      "loss": 0.9061,
      "step": 1628
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.269115686416626,
      "learning_rate": 1.9964204850035874e-05,
      "loss": 0.9098,
      "step": 1629
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8316842317581177,
      "learning_rate": 1.9964110398002064e-05,
      "loss": 0.9081,
      "step": 1630
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9385055303573608,
      "learning_rate": 1.9964015821741948e-05,
      "loss": 0.8768,
      "step": 1631
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3172242641448975,
      "learning_rate": 1.9963921121256708e-05,
      "loss": 0.7972,
      "step": 1632
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9104719161987305,
      "learning_rate": 1.996382629654752e-05,
      "loss": 0.736,
      "step": 1633
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.777837872505188,
      "learning_rate": 1.9963731347615574e-05,
      "loss": 0.7283,
      "step": 1634
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.431102752685547,
      "learning_rate": 1.9963636274462048e-05,
      "loss": 0.8635,
      "step": 1635
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.476243257522583,
      "learning_rate": 1.9963541077088128e-05,
      "loss": 0.8963,
      "step": 1636
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.839237928390503,
      "learning_rate": 1.9963445755495004e-05,
      "loss": 0.7192,
      "step": 1637
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9772884845733643,
      "learning_rate": 1.996335030968386e-05,
      "loss": 0.8635,
      "step": 1638
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9947389364242554,
      "learning_rate": 1.9963254739655888e-05,
      "loss": 0.671,
      "step": 1639
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3226537704467773,
      "learning_rate": 1.9963159045412282e-05,
      "loss": 0.9452,
      "step": 1640
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0683906078338623,
      "learning_rate": 1.9963063226954232e-05,
      "loss": 0.6774,
      "step": 1641
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9116028547286987,
      "learning_rate": 1.996296728428293e-05,
      "loss": 0.7189,
      "step": 1642
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8927586078643799,
      "learning_rate": 1.996287121739958e-05,
      "loss": 0.8973,
      "step": 1643
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0583927631378174,
      "learning_rate": 1.996277502630537e-05,
      "loss": 0.7861,
      "step": 1644
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.621506929397583,
      "learning_rate": 1.996267871100151e-05,
      "loss": 0.865,
      "step": 1645
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.29880690574646,
      "learning_rate": 1.996258227148919e-05,
      "loss": 1.0447,
      "step": 1646
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1618199348449707,
      "learning_rate": 1.9962485707769617e-05,
      "loss": 0.9416,
      "step": 1647
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.070385456085205,
      "learning_rate": 1.9962389019844e-05,
      "loss": 0.8276,
      "step": 1648
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.141390800476074,
      "learning_rate": 1.9962292207713538e-05,
      "loss": 0.6829,
      "step": 1649
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9873788356781006,
      "learning_rate": 1.9962195271379442e-05,
      "loss": 0.832,
      "step": 1650
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0933568477630615,
      "learning_rate": 1.9962098210842914e-05,
      "loss": 0.8845,
      "step": 1651
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1470015048980713,
      "learning_rate": 1.996200102610517e-05,
      "loss": 0.8782,
      "step": 1652
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.148221254348755,
      "learning_rate": 1.9961903717167422e-05,
      "loss": 0.7784,
      "step": 1653
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.077247381210327,
      "learning_rate": 1.996180628403088e-05,
      "loss": 0.8555,
      "step": 1654
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8312114477157593,
      "learning_rate": 1.996170872669676e-05,
      "loss": 0.8021,
      "step": 1655
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3248348236083984,
      "learning_rate": 1.996161104516628e-05,
      "loss": 0.7469,
      "step": 1656
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.336686611175537,
      "learning_rate": 1.9961513239440653e-05,
      "loss": 0.9682,
      "step": 1657
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1797094345092773,
      "learning_rate": 1.9961415309521105e-05,
      "loss": 1.0028,
      "step": 1658
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.614027261734009,
      "learning_rate": 1.996131725540885e-05,
      "loss": 0.8422,
      "step": 1659
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.2151098251342773,
      "learning_rate": 1.9961219077105118e-05,
      "loss": 0.7344,
      "step": 1660
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9582123756408691,
      "learning_rate": 1.9961120774611124e-05,
      "loss": 1.0315,
      "step": 1661
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1174190044403076,
      "learning_rate": 1.99610223479281e-05,
      "loss": 0.8618,
      "step": 1662
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0579636096954346,
      "learning_rate": 1.9960923797057275e-05,
      "loss": 0.903,
      "step": 1663
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0817837715148926,
      "learning_rate": 1.9960825121999872e-05,
      "loss": 0.9387,
      "step": 1664
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.289825916290283,
      "learning_rate": 1.9960726322757122e-05,
      "loss": 0.8518,
      "step": 1665
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8668832778930664,
      "learning_rate": 1.996062739933026e-05,
      "loss": 0.8925,
      "step": 1666
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.731629490852356,
      "learning_rate": 1.9960528351720517e-05,
      "loss": 0.757,
      "step": 1667
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0218770503997803,
      "learning_rate": 1.996042917992913e-05,
      "loss": 0.8415,
      "step": 1668
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9918311834335327,
      "learning_rate": 1.9960329883957335e-05,
      "loss": 0.6487,
      "step": 1669
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.6210904121398926,
      "learning_rate": 1.9960230463806366e-05,
      "loss": 0.8378,
      "step": 1670
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.03678035736084,
      "learning_rate": 1.9960130919477464e-05,
      "loss": 0.819,
      "step": 1671
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.712074875831604,
      "learning_rate": 1.9960031250971875e-05,
      "loss": 0.9722,
      "step": 1672
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.415595531463623,
      "learning_rate": 1.9959931458290835e-05,
      "loss": 0.8752,
      "step": 1673
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.852091670036316,
      "learning_rate": 1.9959831541435592e-05,
      "loss": 0.9073,
      "step": 1674
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.09562087059021,
      "learning_rate": 1.9959731500407393e-05,
      "loss": 0.8083,
      "step": 1675
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7463313341140747,
      "learning_rate": 1.995963133520748e-05,
      "loss": 0.9239,
      "step": 1676
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9309155941009521,
      "learning_rate": 1.995953104583711e-05,
      "loss": 0.8758,
      "step": 1677
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.926533818244934,
      "learning_rate": 1.995943063229752e-05,
      "loss": 0.8511,
      "step": 1678
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1038458347320557,
      "learning_rate": 1.9959330094589978e-05,
      "loss": 0.9226,
      "step": 1679
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.3474223613739014,
      "learning_rate": 1.995922943271573e-05,
      "loss": 0.8574,
      "step": 1680
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8244690895080566,
      "learning_rate": 1.9959128646676028e-05,
      "loss": 0.9092,
      "step": 1681
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.057917356491089,
      "learning_rate": 1.995902773647213e-05,
      "loss": 0.933,
      "step": 1682
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9969228506088257,
      "learning_rate": 1.9958926702105296e-05,
      "loss": 0.8643,
      "step": 1683
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7440626621246338,
      "learning_rate": 1.995882554357678e-05,
      "loss": 0.8264,
      "step": 1684
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.02559232711792,
      "learning_rate": 1.9958724260887857e-05,
      "loss": 0.8765,
      "step": 1685
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7800719738006592,
      "learning_rate": 1.9958622854039772e-05,
      "loss": 0.9117,
      "step": 1686
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1782307624816895,
      "learning_rate": 1.9958521323033805e-05,
      "loss": 0.9322,
      "step": 1687
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0821802616119385,
      "learning_rate": 1.9958419667871212e-05,
      "loss": 0.8685,
      "step": 1688
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9309961795806885,
      "learning_rate": 1.9958317888553264e-05,
      "loss": 0.9027,
      "step": 1689
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9502907991409302,
      "learning_rate": 1.9958215985081228e-05,
      "loss": 0.815,
      "step": 1690
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8347887992858887,
      "learning_rate": 1.9958113957456374e-05,
      "loss": 0.782,
      "step": 1691
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9059467315673828,
      "learning_rate": 1.9958011805679983e-05,
      "loss": 0.7358,
      "step": 1692
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0002715587615967,
      "learning_rate": 1.9957909529753313e-05,
      "loss": 0.9983,
      "step": 1693
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0898356437683105,
      "learning_rate": 1.995780712967765e-05,
      "loss": 0.6909,
      "step": 1694
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.126542091369629,
      "learning_rate": 1.995770460545427e-05,
      "loss": 0.9369,
      "step": 1695
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7626144886016846,
      "learning_rate": 1.9957601957084448e-05,
      "loss": 0.8028,
      "step": 1696
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2590019702911377,
      "learning_rate": 1.9957499184569466e-05,
      "loss": 0.8254,
      "step": 1697
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8817940950393677,
      "learning_rate": 1.99573962879106e-05,
      "loss": 0.8077,
      "step": 1698
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8049789667129517,
      "learning_rate": 1.995729326710914e-05,
      "loss": 0.7482,
      "step": 1699
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2460010051727295,
      "learning_rate": 1.995719012216637e-05,
      "loss": 0.9327,
      "step": 1700
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0972328186035156,
      "learning_rate": 1.995708685308357e-05,
      "loss": 0.8299,
      "step": 1701
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.099921703338623,
      "learning_rate": 1.9956983459862033e-05,
      "loss": 0.9159,
      "step": 1702
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.847413182258606,
      "learning_rate": 1.995687994250304e-05,
      "loss": 0.9451,
      "step": 1703
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1555047035217285,
      "learning_rate": 1.9956776301007896e-05,
      "loss": 0.8824,
      "step": 1704
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.5327260494232178,
      "learning_rate": 1.9956672535377882e-05,
      "loss": 0.8895,
      "step": 1705
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.355301856994629,
      "learning_rate": 1.9956568645614293e-05,
      "loss": 0.9166,
      "step": 1706
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.174027681350708,
      "learning_rate": 1.9956464631718425e-05,
      "loss": 0.864,
      "step": 1707
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1451001167297363,
      "learning_rate": 1.995636049369158e-05,
      "loss": 0.7936,
      "step": 1708
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9849272966384888,
      "learning_rate": 1.9956256231535044e-05,
      "loss": 0.9067,
      "step": 1709
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9386250972747803,
      "learning_rate": 1.9956151845250132e-05,
      "loss": 0.8746,
      "step": 1710
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7945274114608765,
      "learning_rate": 1.9956047334838137e-05,
      "loss": 0.8194,
      "step": 1711
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5573667287826538,
      "learning_rate": 1.995594270030036e-05,
      "loss": 0.8396,
      "step": 1712
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0124831199645996,
      "learning_rate": 1.995583794163811e-05,
      "loss": 0.7743,
      "step": 1713
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.219546318054199,
      "learning_rate": 1.995573305885269e-05,
      "loss": 0.6573,
      "step": 1714
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.328904390335083,
      "learning_rate": 1.995562805194541e-05,
      "loss": 0.9108,
      "step": 1715
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8249130249023438,
      "learning_rate": 1.9955522920917577e-05,
      "loss": 0.8822,
      "step": 1716
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9469226598739624,
      "learning_rate": 1.9955417665770507e-05,
      "loss": 0.8413,
      "step": 1717
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.320507526397705,
      "learning_rate": 1.9955312286505504e-05,
      "loss": 0.9805,
      "step": 1718
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.077101945877075,
      "learning_rate": 1.995520678312389e-05,
      "loss": 0.7398,
      "step": 1719
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1929304599761963,
      "learning_rate": 1.9955101155626974e-05,
      "loss": 0.8698,
      "step": 1720
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0391845703125,
      "learning_rate": 1.995499540401608e-05,
      "loss": 0.646,
      "step": 1721
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1104376316070557,
      "learning_rate": 1.9954889528292515e-05,
      "loss": 0.8617,
      "step": 1722
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8092806339263916,
      "learning_rate": 1.995478352845761e-05,
      "loss": 0.7967,
      "step": 1723
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2570676803588867,
      "learning_rate": 1.995467740451268e-05,
      "loss": 0.8073,
      "step": 1724
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.159123182296753,
      "learning_rate": 1.995457115645905e-05,
      "loss": 0.8279,
      "step": 1725
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9031134843826294,
      "learning_rate": 1.9954464784298047e-05,
      "loss": 0.711,
      "step": 1726
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.127410411834717,
      "learning_rate": 1.9954358288030997e-05,
      "loss": 0.7566,
      "step": 1727
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0706393718719482,
      "learning_rate": 1.9954251667659222e-05,
      "loss": 0.7369,
      "step": 1728
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.129589080810547,
      "learning_rate": 1.9954144923184058e-05,
      "loss": 0.883,
      "step": 1729
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.23581600189209,
      "learning_rate": 1.995403805460683e-05,
      "loss": 0.8445,
      "step": 1730
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7451573610305786,
      "learning_rate": 1.9953931061928876e-05,
      "loss": 0.771,
      "step": 1731
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4515857696533203,
      "learning_rate": 1.9953823945151528e-05,
      "loss": 0.6855,
      "step": 1732
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4709017276763916,
      "learning_rate": 1.995371670427612e-05,
      "loss": 0.7148,
      "step": 1733
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2094810009002686,
      "learning_rate": 1.9953609339303993e-05,
      "loss": 0.7261,
      "step": 1734
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.123647689819336,
      "learning_rate": 1.9953501850236477e-05,
      "loss": 0.8458,
      "step": 1735
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.186502456665039,
      "learning_rate": 1.9953394237074918e-05,
      "loss": 0.68,
      "step": 1736
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9907976388931274,
      "learning_rate": 1.995328649982066e-05,
      "loss": 1.0303,
      "step": 1737
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9077175855636597,
      "learning_rate": 1.9953178638475045e-05,
      "loss": 0.8096,
      "step": 1738
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8598743677139282,
      "learning_rate": 1.995307065303941e-05,
      "loss": 0.8329,
      "step": 1739
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.231891393661499,
      "learning_rate": 1.9952962543515112e-05,
      "loss": 0.8124,
      "step": 1740
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7994141578674316,
      "learning_rate": 1.9952854309903495e-05,
      "loss": 0.7504,
      "step": 1741
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.174640417098999,
      "learning_rate": 1.9952745952205904e-05,
      "loss": 0.761,
      "step": 1742
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9507066011428833,
      "learning_rate": 1.9952637470423695e-05,
      "loss": 0.8934,
      "step": 1743
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8562194108963013,
      "learning_rate": 1.995252886455822e-05,
      "loss": 0.8515,
      "step": 1744
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8150830268859863,
      "learning_rate": 1.9952420134610832e-05,
      "loss": 0.829,
      "step": 1745
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2376444339752197,
      "learning_rate": 1.9952311280582885e-05,
      "loss": 0.8108,
      "step": 1746
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.031468629837036,
      "learning_rate": 1.995220230247574e-05,
      "loss": 0.721,
      "step": 1747
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0722110271453857,
      "learning_rate": 1.995209320029075e-05,
      "loss": 1.0494,
      "step": 1748
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.823135495185852,
      "learning_rate": 1.9951983974029283e-05,
      "loss": 0.7976,
      "step": 1749
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0300543308258057,
      "learning_rate": 1.9951874623692694e-05,
      "loss": 0.9079,
      "step": 1750
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7559515237808228,
      "learning_rate": 1.9951765149282352e-05,
      "loss": 0.9557,
      "step": 1751
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.171924591064453,
      "learning_rate": 1.9951655550799614e-05,
      "loss": 0.8358,
      "step": 1752
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9448041915893555,
      "learning_rate": 1.9951545828245855e-05,
      "loss": 0.8818,
      "step": 1753
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.811143398284912,
      "learning_rate": 1.9951435981622436e-05,
      "loss": 0.7556,
      "step": 1754
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.01025390625,
      "learning_rate": 1.995132601093073e-05,
      "loss": 0.7694,
      "step": 1755
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.218576669692993,
      "learning_rate": 1.9951215916172107e-05,
      "loss": 0.8437,
      "step": 1756
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9130232334136963,
      "learning_rate": 1.995110569734794e-05,
      "loss": 0.7266,
      "step": 1757
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8732750415802002,
      "learning_rate": 1.9950995354459608e-05,
      "loss": 0.9784,
      "step": 1758
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1131343841552734,
      "learning_rate": 1.9950884887508476e-05,
      "loss": 0.7279,
      "step": 1759
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.4820287227630615,
      "learning_rate": 1.9950774296495932e-05,
      "loss": 0.8845,
      "step": 1760
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.115093946456909,
      "learning_rate": 1.9950663581423345e-05,
      "loss": 0.8594,
      "step": 1761
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9739112854003906,
      "learning_rate": 1.9950552742292103e-05,
      "loss": 0.7918,
      "step": 1762
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.13425350189209,
      "learning_rate": 1.9950441779103585e-05,
      "loss": 0.7397,
      "step": 1763
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7292414903640747,
      "learning_rate": 1.9950330691859174e-05,
      "loss": 0.9272,
      "step": 1764
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9577373266220093,
      "learning_rate": 1.995021948056026e-05,
      "loss": 0.7944,
      "step": 1765
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.6491189002990723,
      "learning_rate": 1.995010814520822e-05,
      "loss": 0.8127,
      "step": 1766
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9753156900405884,
      "learning_rate": 1.9949996685804447e-05,
      "loss": 0.7959,
      "step": 1767
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9165513515472412,
      "learning_rate": 1.9949885102350334e-05,
      "loss": 0.753,
      "step": 1768
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9703686237335205,
      "learning_rate": 1.9949773394847264e-05,
      "loss": 0.9423,
      "step": 1769
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.085176467895508,
      "learning_rate": 1.994966156329664e-05,
      "loss": 0.9617,
      "step": 1770
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6966673135757446,
      "learning_rate": 1.9949549607699844e-05,
      "loss": 0.8004,
      "step": 1771
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1541848182678223,
      "learning_rate": 1.9949437528058287e-05,
      "loss": 0.6689,
      "step": 1772
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6782946586608887,
      "learning_rate": 1.994932532437335e-05,
      "loss": 0.8427,
      "step": 1773
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.909291386604309,
      "learning_rate": 1.9949212996646445e-05,
      "loss": 0.8663,
      "step": 1774
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7885193824768066,
      "learning_rate": 1.9949100544878967e-05,
      "loss": 0.832,
      "step": 1775
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.0433123111724854,
      "learning_rate": 1.9948987969072315e-05,
      "loss": 0.9423,
      "step": 1776
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2459378242492676,
      "learning_rate": 1.9948875269227897e-05,
      "loss": 0.8385,
      "step": 1777
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7150613069534302,
      "learning_rate": 1.9948762445347122e-05,
      "loss": 0.8852,
      "step": 1778
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8599846363067627,
      "learning_rate": 1.9948649497431386e-05,
      "loss": 0.937,
      "step": 1779
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.120145559310913,
      "learning_rate": 1.9948536425482104e-05,
      "loss": 0.8038,
      "step": 1780
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8929753303527832,
      "learning_rate": 1.9948423229500683e-05,
      "loss": 0.8057,
      "step": 1781
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.956886887550354,
      "learning_rate": 1.9948309909488537e-05,
      "loss": 0.9527,
      "step": 1782
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1436922550201416,
      "learning_rate": 1.9948196465447077e-05,
      "loss": 0.8589,
      "step": 1783
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.321552276611328,
      "learning_rate": 1.994808289737772e-05,
      "loss": 0.7515,
      "step": 1784
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.3067996501922607,
      "learning_rate": 1.9947969205281875e-05,
      "loss": 0.7607,
      "step": 1785
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.2418270111083984,
      "learning_rate": 1.994785538916097e-05,
      "loss": 0.6534,
      "step": 1786
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1032567024230957,
      "learning_rate": 1.9947741449016415e-05,
      "loss": 0.9741,
      "step": 1787
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7446526288986206,
      "learning_rate": 1.9947627384849632e-05,
      "loss": 0.8385,
      "step": 1788
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.022207498550415,
      "learning_rate": 1.9947513196662044e-05,
      "loss": 0.9183,
      "step": 1789
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0437848567962646,
      "learning_rate": 1.994739888445508e-05,
      "loss": 0.8045,
      "step": 1790
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7421238422393799,
      "learning_rate": 1.9947284448230158e-05,
      "loss": 0.7524,
      "step": 1791
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.709474802017212,
      "learning_rate": 1.9947169887988704e-05,
      "loss": 0.7941,
      "step": 1792
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8835946321487427,
      "learning_rate": 1.9947055203732157e-05,
      "loss": 0.7831,
      "step": 1793
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.894338607788086,
      "learning_rate": 1.9946940395461932e-05,
      "loss": 0.8122,
      "step": 1794
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.138991355895996,
      "learning_rate": 1.994682546317947e-05,
      "loss": 0.937,
      "step": 1795
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9367986917495728,
      "learning_rate": 1.9946710406886202e-05,
      "loss": 0.714,
      "step": 1796
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0880954265594482,
      "learning_rate": 1.9946595226583563e-05,
      "loss": 0.8518,
      "step": 1797
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.919074296951294,
      "learning_rate": 1.994647992227299e-05,
      "loss": 0.8572,
      "step": 1798
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0121681690216064,
      "learning_rate": 1.994636449395591e-05,
      "loss": 0.7642,
      "step": 1799
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1373507976531982,
      "learning_rate": 1.9946248941633778e-05,
      "loss": 0.8218,
      "step": 1800
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8676258325576782,
      "learning_rate": 1.9946133265308025e-05,
      "loss": 0.7373,
      "step": 1801
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.874926209449768,
      "learning_rate": 1.9946017464980097e-05,
      "loss": 0.7495,
      "step": 1802
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8355424404144287,
      "learning_rate": 1.9945901540651434e-05,
      "loss": 0.7121,
      "step": 1803
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6171507835388184,
      "learning_rate": 1.9945785492323484e-05,
      "loss": 0.7342,
      "step": 1804
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9594964981079102,
      "learning_rate": 1.9945669319997694e-05,
      "loss": 0.7892,
      "step": 1805
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.093165397644043,
      "learning_rate": 1.9945553023675512e-05,
      "loss": 0.8945,
      "step": 1806
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.973748803138733,
      "learning_rate": 1.9945436603358386e-05,
      "loss": 0.7958,
      "step": 1807
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.153601884841919,
      "learning_rate": 1.994532005904777e-05,
      "loss": 0.8915,
      "step": 1808
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8272448778152466,
      "learning_rate": 1.9945203390745116e-05,
      "loss": 0.6582,
      "step": 1809
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.079284429550171,
      "learning_rate": 1.9945086598451876e-05,
      "loss": 0.7468,
      "step": 1810
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0998992919921875,
      "learning_rate": 1.9944969682169512e-05,
      "loss": 0.8266,
      "step": 1811
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.4745891094207764,
      "learning_rate": 1.994485264189948e-05,
      "loss": 0.9339,
      "step": 1812
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.843001127243042,
      "learning_rate": 1.9944735477643232e-05,
      "loss": 0.7753,
      "step": 1813
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9045566320419312,
      "learning_rate": 1.9944618189402237e-05,
      "loss": 0.9442,
      "step": 1814
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9291654825210571,
      "learning_rate": 1.9944500777177956e-05,
      "loss": 0.6097,
      "step": 1815
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2365517616271973,
      "learning_rate": 1.994438324097185e-05,
      "loss": 0.9517,
      "step": 1816
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.823899269104004,
      "learning_rate": 1.9944265580785385e-05,
      "loss": 0.9071,
      "step": 1817
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8751978874206543,
      "learning_rate": 1.9944147796620026e-05,
      "loss": 0.8563,
      "step": 1818
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9668699502944946,
      "learning_rate": 1.9944029888477246e-05,
      "loss": 0.6416,
      "step": 1819
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2405295372009277,
      "learning_rate": 1.9943911856358516e-05,
      "loss": 0.8391,
      "step": 1820
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.940204381942749,
      "learning_rate": 1.9943793700265303e-05,
      "loss": 0.9211,
      "step": 1821
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8297455310821533,
      "learning_rate": 1.994367542019908e-05,
      "loss": 0.9656,
      "step": 1822
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8819434642791748,
      "learning_rate": 1.9943557016161327e-05,
      "loss": 0.6247,
      "step": 1823
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.133601188659668,
      "learning_rate": 1.9943438488153514e-05,
      "loss": 0.755,
      "step": 1824
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6655895709991455,
      "learning_rate": 1.9943319836177117e-05,
      "loss": 0.8788,
      "step": 1825
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.145965099334717,
      "learning_rate": 1.9943201060233626e-05,
      "loss": 0.8143,
      "step": 1826
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1585206985473633,
      "learning_rate": 1.9943082160324512e-05,
      "loss": 0.9278,
      "step": 1827
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1236860752105713,
      "learning_rate": 1.9942963136451262e-05,
      "loss": 0.8727,
      "step": 1828
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2915749549865723,
      "learning_rate": 1.994284398861536e-05,
      "loss": 1.1379,
      "step": 1829
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8819048404693604,
      "learning_rate": 1.9942724716818288e-05,
      "loss": 0.8227,
      "step": 1830
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.918076515197754,
      "learning_rate": 1.994260532106153e-05,
      "loss": 0.8816,
      "step": 1831
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8332206010818481,
      "learning_rate": 1.994248580134659e-05,
      "loss": 0.8089,
      "step": 1832
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7531356811523438,
      "learning_rate": 1.994236615767494e-05,
      "loss": 0.8421,
      "step": 1833
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3178939819335938,
      "learning_rate": 1.9942246390048084e-05,
      "loss": 0.8301,
      "step": 1834
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7625783681869507,
      "learning_rate": 1.994212649846751e-05,
      "loss": 0.9164,
      "step": 1835
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1568188667297363,
      "learning_rate": 1.994200648293471e-05,
      "loss": 0.882,
      "step": 1836
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8137487173080444,
      "learning_rate": 1.9941886343451187e-05,
      "loss": 0.8387,
      "step": 1837
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.010082483291626,
      "learning_rate": 1.9941766080018433e-05,
      "loss": 0.7683,
      "step": 1838
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.873814582824707,
      "learning_rate": 1.994164569263795e-05,
      "loss": 0.7253,
      "step": 1839
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8646878004074097,
      "learning_rate": 1.994152518131124e-05,
      "loss": 0.6566,
      "step": 1840
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9423093795776367,
      "learning_rate": 1.9941404546039806e-05,
      "loss": 0.7368,
      "step": 1841
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9704595804214478,
      "learning_rate": 1.9941283786825146e-05,
      "loss": 0.7381,
      "step": 1842
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0916097164154053,
      "learning_rate": 1.9941162903668773e-05,
      "loss": 0.855,
      "step": 1843
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.011875629425049,
      "learning_rate": 1.9941041896572187e-05,
      "loss": 0.8255,
      "step": 1844
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9183837175369263,
      "learning_rate": 1.9940920765536904e-05,
      "loss": 0.7424,
      "step": 1845
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9447336196899414,
      "learning_rate": 1.9940799510564428e-05,
      "loss": 0.9293,
      "step": 1846
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.329120635986328,
      "learning_rate": 1.9940678131656276e-05,
      "loss": 0.7605,
      "step": 1847
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7346229553222656,
      "learning_rate": 1.9940556628813956e-05,
      "loss": 0.9288,
      "step": 1848
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9928412437438965,
      "learning_rate": 1.994043500203899e-05,
      "loss": 0.7365,
      "step": 1849
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7207950353622437,
      "learning_rate": 1.9940313251332885e-05,
      "loss": 0.9503,
      "step": 1850
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.129836082458496,
      "learning_rate": 1.9940191376697166e-05,
      "loss": 0.7316,
      "step": 1851
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.242732048034668,
      "learning_rate": 1.9940069378133347e-05,
      "loss": 0.9602,
      "step": 1852
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0542514324188232,
      "learning_rate": 1.9939947255642957e-05,
      "loss": 0.8074,
      "step": 1853
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.6415956020355225,
      "learning_rate": 1.9939825009227513e-05,
      "loss": 0.9826,
      "step": 1854
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8948239088058472,
      "learning_rate": 1.9939702638888536e-05,
      "loss": 0.8321,
      "step": 1855
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9779738187789917,
      "learning_rate": 1.993958014462756e-05,
      "loss": 0.8365,
      "step": 1856
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9514789581298828,
      "learning_rate": 1.9939457526446105e-05,
      "loss": 0.9853,
      "step": 1857
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.049711227416992,
      "learning_rate": 1.99393347843457e-05,
      "loss": 0.794,
      "step": 1858
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.200317859649658,
      "learning_rate": 1.9939211918327884e-05,
      "loss": 0.7962,
      "step": 1859
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.107179880142212,
      "learning_rate": 1.993908892839418e-05,
      "loss": 0.835,
      "step": 1860
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.777277946472168,
      "learning_rate": 1.993896581454612e-05,
      "loss": 0.7994,
      "step": 1861
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0518546104431152,
      "learning_rate": 1.9938842576785243e-05,
      "loss": 0.9032,
      "step": 1862
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0084896087646484,
      "learning_rate": 1.9938719215113092e-05,
      "loss": 0.7709,
      "step": 1863
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.3746962547302246,
      "learning_rate": 1.9938595729531194e-05,
      "loss": 1.0073,
      "step": 1864
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0571959018707275,
      "learning_rate": 1.993847212004109e-05,
      "loss": 0.7263,
      "step": 1865
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.234649181365967,
      "learning_rate": 1.993834838664433e-05,
      "loss": 0.746,
      "step": 1866
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6413084268569946,
      "learning_rate": 1.9938224529342446e-05,
      "loss": 0.9166,
      "step": 1867
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.2832186222076416,
      "learning_rate": 1.9938100548136983e-05,
      "loss": 0.7529,
      "step": 1868
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.360795259475708,
      "learning_rate": 1.9937976443029496e-05,
      "loss": 0.7288,
      "step": 1869
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7458109855651855,
      "learning_rate": 1.9937852214021525e-05,
      "loss": 0.7274,
      "step": 1870
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.019641876220703,
      "learning_rate": 1.993772786111462e-05,
      "loss": 0.8084,
      "step": 1871
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.081582546234131,
      "learning_rate": 1.9937603384310336e-05,
      "loss": 0.9391,
      "step": 1872
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8259927034378052,
      "learning_rate": 1.9937478783610217e-05,
      "loss": 0.7207,
      "step": 1873
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1636486053466797,
      "learning_rate": 1.9937354059015822e-05,
      "loss": 0.9353,
      "step": 1874
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.900032877922058,
      "learning_rate": 1.9937229210528703e-05,
      "loss": 0.8596,
      "step": 1875
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8750654458999634,
      "learning_rate": 1.9937104238150415e-05,
      "loss": 0.8869,
      "step": 1876
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7059684991836548,
      "learning_rate": 1.993697914188252e-05,
      "loss": 0.6766,
      "step": 1877
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.730273962020874,
      "learning_rate": 1.9936853921726576e-05,
      "loss": 0.6894,
      "step": 1878
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9054664373397827,
      "learning_rate": 1.993672857768415e-05,
      "loss": 0.6588,
      "step": 1879
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.131502866744995,
      "learning_rate": 1.9936603109756794e-05,
      "loss": 0.724,
      "step": 1880
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9783997535705566,
      "learning_rate": 1.9936477517946077e-05,
      "loss": 0.8567,
      "step": 1881
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.385190963745117,
      "learning_rate": 1.9936351802253565e-05,
      "loss": 0.8444,
      "step": 1882
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8562532663345337,
      "learning_rate": 1.9936225962680828e-05,
      "loss": 0.8989,
      "step": 1883
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5613248348236084,
      "learning_rate": 1.993609999922943e-05,
      "loss": 0.8668,
      "step": 1884
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8455630540847778,
      "learning_rate": 1.993597391190095e-05,
      "loss": 0.7841,
      "step": 1885
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.727325439453125,
      "learning_rate": 1.9935847700696945e-05,
      "loss": 0.7418,
      "step": 1886
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6699252128601074,
      "learning_rate": 1.9935721365619004e-05,
      "loss": 0.9263,
      "step": 1887
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.355448007583618,
      "learning_rate": 1.9935594906668695e-05,
      "loss": 0.8877,
      "step": 1888
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4766833782196045,
      "learning_rate": 1.9935468323847592e-05,
      "loss": 0.659,
      "step": 1889
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.315701723098755,
      "learning_rate": 1.9935341617157276e-05,
      "loss": 0.8114,
      "step": 1890
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.833027958869934,
      "learning_rate": 1.9935214786599332e-05,
      "loss": 0.9168,
      "step": 1891
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0161545276641846,
      "learning_rate": 1.993508783217533e-05,
      "loss": 0.9508,
      "step": 1892
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.219160795211792,
      "learning_rate": 1.9934960753886864e-05,
      "loss": 0.997,
      "step": 1893
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2370429039001465,
      "learning_rate": 1.993483355173551e-05,
      "loss": 0.8548,
      "step": 1894
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.036857843399048,
      "learning_rate": 1.993470622572286e-05,
      "loss": 0.834,
      "step": 1895
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.3264756202697754,
      "learning_rate": 1.9934578775850498e-05,
      "loss": 0.7878,
      "step": 1896
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7943627834320068,
      "learning_rate": 1.9934451202120015e-05,
      "loss": 0.7738,
      "step": 1897
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5504422187805176,
      "learning_rate": 1.9934323504532993e-05,
      "loss": 1.0657,
      "step": 1898
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6859548091888428,
      "learning_rate": 1.9934195683091038e-05,
      "loss": 0.8933,
      "step": 1899
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.480055809020996,
      "learning_rate": 1.9934067737795735e-05,
      "loss": 0.9167,
      "step": 1900
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8961925506591797,
      "learning_rate": 1.9933939668648682e-05,
      "loss": 0.8436,
      "step": 1901
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0554046630859375,
      "learning_rate": 1.9933811475651472e-05,
      "loss": 0.76,
      "step": 1902
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7082198858261108,
      "learning_rate": 1.9933683158805707e-05,
      "loss": 0.8712,
      "step": 1903
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9783533811569214,
      "learning_rate": 1.9933554718112983e-05,
      "loss": 0.8199,
      "step": 1904
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6959195137023926,
      "learning_rate": 1.993342615357491e-05,
      "loss": 0.857,
      "step": 1905
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5555225610733032,
      "learning_rate": 1.993329746519308e-05,
      "loss": 0.7718,
      "step": 1906
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.5554375648498535,
      "learning_rate": 1.9933168652969103e-05,
      "loss": 0.812,
      "step": 1907
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.910657286643982,
      "learning_rate": 1.9933039716904584e-05,
      "loss": 0.7942,
      "step": 1908
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.029705286026001,
      "learning_rate": 1.9932910657001134e-05,
      "loss": 0.9264,
      "step": 1909
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1656806468963623,
      "learning_rate": 1.9932781473260355e-05,
      "loss": 0.6669,
      "step": 1910
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9044324159622192,
      "learning_rate": 1.993265216568386e-05,
      "loss": 0.8026,
      "step": 1911
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.869019627571106,
      "learning_rate": 1.9932522734273265e-05,
      "loss": 0.8279,
      "step": 1912
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1262526512145996,
      "learning_rate": 1.993239317903018e-05,
      "loss": 0.8432,
      "step": 1913
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.121199369430542,
      "learning_rate": 1.9932263499956224e-05,
      "loss": 0.8098,
      "step": 1914
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.204948902130127,
      "learning_rate": 1.9932133697053006e-05,
      "loss": 0.9439,
      "step": 1915
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.007819175720215,
      "learning_rate": 1.9932003770322154e-05,
      "loss": 0.7576,
      "step": 1916
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.941897988319397,
      "learning_rate": 1.9931873719765282e-05,
      "loss": 0.7383,
      "step": 1917
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7854971885681152,
      "learning_rate": 1.993174354538401e-05,
      "loss": 0.7784,
      "step": 1918
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9943994283676147,
      "learning_rate": 1.9931613247179965e-05,
      "loss": 0.7863,
      "step": 1919
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.321317434310913,
      "learning_rate": 1.9931482825154772e-05,
      "loss": 0.7676,
      "step": 1920
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8571966886520386,
      "learning_rate": 1.9931352279310052e-05,
      "loss": 0.8375,
      "step": 1921
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0752639770507812,
      "learning_rate": 1.9931221609647435e-05,
      "loss": 0.9828,
      "step": 1922
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9214048385620117,
      "learning_rate": 1.9931090816168553e-05,
      "loss": 0.7959,
      "step": 1923
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8174916505813599,
      "learning_rate": 1.9930959898875034e-05,
      "loss": 0.8648,
      "step": 1924
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9148505926132202,
      "learning_rate": 1.9930828857768513e-05,
      "loss": 0.9179,
      "step": 1925
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7936087846755981,
      "learning_rate": 1.9930697692850617e-05,
      "loss": 0.8155,
      "step": 1926
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.009693145751953,
      "learning_rate": 1.9930566404122986e-05,
      "loss": 0.9013,
      "step": 1927
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7912704944610596,
      "learning_rate": 1.993043499158726e-05,
      "loss": 0.9966,
      "step": 1928
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2676076889038086,
      "learning_rate": 1.9930303455245067e-05,
      "loss": 0.8548,
      "step": 1929
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.06366229057312,
      "learning_rate": 1.993017179509806e-05,
      "loss": 0.766,
      "step": 1930
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7770260572433472,
      "learning_rate": 1.9930040011147872e-05,
      "loss": 0.7714,
      "step": 1931
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7773255109786987,
      "learning_rate": 1.9929908103396145e-05,
      "loss": 0.8364,
      "step": 1932
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7036548852920532,
      "learning_rate": 1.992977607184453e-05,
      "loss": 0.8777,
      "step": 1933
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.012810468673706,
      "learning_rate": 1.992964391649467e-05,
      "loss": 0.9047,
      "step": 1934
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.631680965423584,
      "learning_rate": 1.9929511637348213e-05,
      "loss": 0.8021,
      "step": 1935
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7582101821899414,
      "learning_rate": 1.9929379234406805e-05,
      "loss": 0.6952,
      "step": 1936
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2953619956970215,
      "learning_rate": 1.99292467076721e-05,
      "loss": 0.6753,
      "step": 1937
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1229639053344727,
      "learning_rate": 1.992911405714575e-05,
      "loss": 0.8587,
      "step": 1938
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.915014386177063,
      "learning_rate": 1.9928981282829407e-05,
      "loss": 0.9114,
      "step": 1939
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6342415809631348,
      "learning_rate": 1.992884838472473e-05,
      "loss": 0.8943,
      "step": 1940
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.844488263130188,
      "learning_rate": 1.9928715362833372e-05,
      "loss": 0.8255,
      "step": 1941
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9399219751358032,
      "learning_rate": 1.992858221715699e-05,
      "loss": 0.8168,
      "step": 1942
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0560970306396484,
      "learning_rate": 1.9928448947697248e-05,
      "loss": 0.964,
      "step": 1943
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2717862129211426,
      "learning_rate": 1.9928315554455812e-05,
      "loss": 0.8047,
      "step": 1944
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9869587421417236,
      "learning_rate": 1.9928182037434337e-05,
      "loss": 0.6011,
      "step": 1945
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.718112826347351,
      "learning_rate": 1.9928048396634488e-05,
      "loss": 0.9038,
      "step": 1946
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9799163341522217,
      "learning_rate": 1.9927914632057934e-05,
      "loss": 0.7525,
      "step": 1947
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0062882900238037,
      "learning_rate": 1.9927780743706343e-05,
      "loss": 0.9961,
      "step": 1948
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2624197006225586,
      "learning_rate": 1.9927646731581382e-05,
      "loss": 0.7179,
      "step": 1949
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7529243230819702,
      "learning_rate": 1.9927512595684724e-05,
      "loss": 0.7381,
      "step": 1950
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.208343267440796,
      "learning_rate": 1.992737833601804e-05,
      "loss": 0.9477,
      "step": 1951
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9179166555404663,
      "learning_rate": 1.9927243952583006e-05,
      "loss": 0.8615,
      "step": 1952
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2561721801757812,
      "learning_rate": 1.992710944538129e-05,
      "loss": 0.6301,
      "step": 1953
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9860801696777344,
      "learning_rate": 1.992697481441458e-05,
      "loss": 0.9064,
      "step": 1954
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.799079179763794,
      "learning_rate": 1.9926840059684543e-05,
      "loss": 0.8635,
      "step": 1955
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9076088666915894,
      "learning_rate": 1.992670518119287e-05,
      "loss": 0.836,
      "step": 1956
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2110228538513184,
      "learning_rate": 1.992657017894124e-05,
      "loss": 0.6713,
      "step": 1957
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7303274869918823,
      "learning_rate": 1.992643505293133e-05,
      "loss": 0.8801,
      "step": 1958
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5778510570526123,
      "learning_rate": 1.992629980316483e-05,
      "loss": 0.8067,
      "step": 1959
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8581809997558594,
      "learning_rate": 1.9926164429643423e-05,
      "loss": 0.708,
      "step": 1960
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0283899307250977,
      "learning_rate": 1.9926028932368798e-05,
      "loss": 0.8346,
      "step": 1961
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8222229480743408,
      "learning_rate": 1.992589331134265e-05,
      "loss": 0.8871,
      "step": 1962
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7978054285049438,
      "learning_rate": 1.9925757566566658e-05,
      "loss": 0.83,
      "step": 1963
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.128608465194702,
      "learning_rate": 1.9925621698042524e-05,
      "loss": 0.9589,
      "step": 1964
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.072519540786743,
      "learning_rate": 1.9925485705771938e-05,
      "loss": 0.8459,
      "step": 1965
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9837517738342285,
      "learning_rate": 1.9925349589756594e-05,
      "loss": 0.8459,
      "step": 1966
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.791979193687439,
      "learning_rate": 1.9925213349998197e-05,
      "loss": 0.8753,
      "step": 1967
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.0996997356414795,
      "learning_rate": 1.9925076986498434e-05,
      "loss": 0.8281,
      "step": 1968
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7694730758666992,
      "learning_rate": 1.9924940499259016e-05,
      "loss": 0.9102,
      "step": 1969
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2077548503875732,
      "learning_rate": 1.9924803888281637e-05,
      "loss": 0.6957,
      "step": 1970
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8085356950759888,
      "learning_rate": 1.9924667153568e-05,
      "loss": 0.8494,
      "step": 1971
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.308974027633667,
      "learning_rate": 1.992453029511982e-05,
      "loss": 0.85,
      "step": 1972
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7460156679153442,
      "learning_rate": 1.9924393312938788e-05,
      "loss": 0.89,
      "step": 1973
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1223983764648438,
      "learning_rate": 1.992425620702662e-05,
      "loss": 0.7044,
      "step": 1974
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4559967517852783,
      "learning_rate": 1.9924118977385033e-05,
      "loss": 0.8439,
      "step": 1975
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.914780855178833,
      "learning_rate": 1.9923981624015724e-05,
      "loss": 0.7746,
      "step": 1976
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.390493869781494,
      "learning_rate": 1.9923844146920414e-05,
      "loss": 0.8625,
      "step": 1977
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8416887521743774,
      "learning_rate": 1.992370654610081e-05,
      "loss": 0.7327,
      "step": 1978
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.135481834411621,
      "learning_rate": 1.992356882155863e-05,
      "loss": 1.002,
      "step": 1979
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9923038482666016,
      "learning_rate": 1.99234309732956e-05,
      "loss": 0.9767,
      "step": 1980
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.1134989261627197,
      "learning_rate": 1.992329300131343e-05,
      "loss": 0.8234,
      "step": 1981
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9018161296844482,
      "learning_rate": 1.9923154905613837e-05,
      "loss": 0.9339,
      "step": 1982
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7626452445983887,
      "learning_rate": 1.992301668619855e-05,
      "loss": 0.7384,
      "step": 1983
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7630864381790161,
      "learning_rate": 1.992287834306929e-05,
      "loss": 0.6948,
      "step": 1984
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6473383903503418,
      "learning_rate": 1.9922739876227782e-05,
      "loss": 0.8839,
      "step": 1985
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1773502826690674,
      "learning_rate": 1.992260128567575e-05,
      "loss": 0.7098,
      "step": 1986
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8686637878417969,
      "learning_rate": 1.9922462571414925e-05,
      "loss": 0.9351,
      "step": 1987
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9876888990402222,
      "learning_rate": 1.9922323733447032e-05,
      "loss": 0.7704,
      "step": 1988
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9399374723434448,
      "learning_rate": 1.9922184771773806e-05,
      "loss": 0.6433,
      "step": 1989
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9633125066757202,
      "learning_rate": 1.992204568639698e-05,
      "loss": 0.947,
      "step": 1990
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0509047508239746,
      "learning_rate": 1.9921906477318285e-05,
      "loss": 0.6602,
      "step": 1991
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8222562074661255,
      "learning_rate": 1.9921767144539456e-05,
      "loss": 0.7794,
      "step": 1992
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.050781488418579,
      "learning_rate": 1.9921627688062237e-05,
      "loss": 0.8285,
      "step": 1993
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9635255336761475,
      "learning_rate": 1.9921488107888358e-05,
      "loss": 0.7195,
      "step": 1994
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9186346530914307,
      "learning_rate": 1.9921348404019564e-05,
      "loss": 0.753,
      "step": 1995
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.916827917098999,
      "learning_rate": 1.9921208576457593e-05,
      "loss": 0.746,
      "step": 1996
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9036606550216675,
      "learning_rate": 1.9921068625204194e-05,
      "loss": 0.8385,
      "step": 1997
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9810590744018555,
      "learning_rate": 1.9920928550261108e-05,
      "loss": 0.7979,
      "step": 1998
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1697959899902344,
      "learning_rate": 1.9920788351630083e-05,
      "loss": 0.7512,
      "step": 1999
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2641496658325195,
      "learning_rate": 1.9920648029312864e-05,
      "loss": 0.7571,
      "step": 2000
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8358745574951172,
      "learning_rate": 1.9920507583311203e-05,
      "loss": 0.7191,
      "step": 2001
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7295222282409668,
      "learning_rate": 1.9920367013626852e-05,
      "loss": 0.679,
      "step": 2002
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6439427137374878,
      "learning_rate": 1.992022632026156e-05,
      "loss": 0.6867,
      "step": 2003
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7015143632888794,
      "learning_rate": 1.9920085503217085e-05,
      "loss": 0.7464,
      "step": 2004
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1032497882843018,
      "learning_rate": 1.991994456249518e-05,
      "loss": 0.7769,
      "step": 2005
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8583122491836548,
      "learning_rate": 1.99198034980976e-05,
      "loss": 0.9114,
      "step": 2006
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9725456237792969,
      "learning_rate": 1.9919662310026108e-05,
      "loss": 0.9316,
      "step": 2007
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.087984323501587,
      "learning_rate": 1.9919520998282463e-05,
      "loss": 0.8358,
      "step": 2008
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8987711668014526,
      "learning_rate": 1.9919379562868427e-05,
      "loss": 0.6164,
      "step": 2009
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7561389207839966,
      "learning_rate": 1.9919238003785764e-05,
      "loss": 0.7919,
      "step": 2010
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8904991149902344,
      "learning_rate": 1.9919096321036237e-05,
      "loss": 0.7444,
      "step": 2011
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.344862937927246,
      "learning_rate": 1.9918954514621614e-05,
      "loss": 0.933,
      "step": 2012
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9100348949432373,
      "learning_rate": 1.991881258454366e-05,
      "loss": 0.8934,
      "step": 2013
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9824289083480835,
      "learning_rate": 1.991867053080415e-05,
      "loss": 0.7607,
      "step": 2014
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9241888523101807,
      "learning_rate": 1.9918528353404845e-05,
      "loss": 0.7549,
      "step": 2015
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9138317108154297,
      "learning_rate": 1.991838605234753e-05,
      "loss": 0.8039,
      "step": 2016
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8029991388320923,
      "learning_rate": 1.9918243627633975e-05,
      "loss": 0.7505,
      "step": 2017
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8420557975769043,
      "learning_rate": 1.991810107926595e-05,
      "loss": 0.7453,
      "step": 2018
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.192249059677124,
      "learning_rate": 1.9917958407245238e-05,
      "loss": 0.9413,
      "step": 2019
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.7962043285369873,
      "learning_rate": 1.9917815611573613e-05,
      "loss": 0.6273,
      "step": 2020
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.393944025039673,
      "learning_rate": 1.991767269225286e-05,
      "loss": 0.9411,
      "step": 2021
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9592512845993042,
      "learning_rate": 1.991752964928476e-05,
      "loss": 0.9004,
      "step": 2022
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.639752149581909,
      "learning_rate": 1.9917386482671095e-05,
      "loss": 0.6983,
      "step": 2023
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7250339984893799,
      "learning_rate": 1.991724319241365e-05,
      "loss": 0.7123,
      "step": 2024
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6760202646255493,
      "learning_rate": 1.991709977851421e-05,
      "loss": 0.9405,
      "step": 2025
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.959006905555725,
      "learning_rate": 1.991695624097457e-05,
      "loss": 0.8559,
      "step": 2026
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.243129014968872,
      "learning_rate": 1.991681257979651e-05,
      "loss": 0.652,
      "step": 2027
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7089152336120605,
      "learning_rate": 1.991666879498183e-05,
      "loss": 0.6789,
      "step": 2028
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8021332025527954,
      "learning_rate": 1.991652488653231e-05,
      "loss": 0.6817,
      "step": 2029
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9406538009643555,
      "learning_rate": 1.9916380854449757e-05,
      "loss": 0.8394,
      "step": 2030
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8667213916778564,
      "learning_rate": 1.9916236698735963e-05,
      "loss": 0.7271,
      "step": 2031
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.785486102104187,
      "learning_rate": 1.9916092419392723e-05,
      "loss": 0.7329,
      "step": 2032
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7660441398620605,
      "learning_rate": 1.991594801642184e-05,
      "loss": 0.8266,
      "step": 2033
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6317542791366577,
      "learning_rate": 1.9915803489825104e-05,
      "loss": 0.9322,
      "step": 2034
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.235821485519409,
      "learning_rate": 1.991565883960433e-05,
      "loss": 0.9312,
      "step": 2035
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7620352506637573,
      "learning_rate": 1.9915514065761315e-05,
      "loss": 0.7021,
      "step": 2036
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9283411502838135,
      "learning_rate": 1.991536916829786e-05,
      "loss": 0.8929,
      "step": 2037
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.191737651824951,
      "learning_rate": 1.991522414721578e-05,
      "loss": 0.8691,
      "step": 2038
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.089974880218506,
      "learning_rate": 1.9915079002516878e-05,
      "loss": 0.8321,
      "step": 2039
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5767979621887207,
      "learning_rate": 1.991493373420297e-05,
      "loss": 0.7954,
      "step": 2040
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7948036193847656,
      "learning_rate": 1.991478834227585e-05,
      "loss": 0.9599,
      "step": 2041
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8770731687545776,
      "learning_rate": 1.9914642826737352e-05,
      "loss": 0.9428,
      "step": 2042
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1288883686065674,
      "learning_rate": 1.991449718758928e-05,
      "loss": 0.7373,
      "step": 2043
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2415285110473633,
      "learning_rate": 1.9914351424833443e-05,
      "loss": 0.8458,
      "step": 2044
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0610954761505127,
      "learning_rate": 1.9914205538471672e-05,
      "loss": 0.771,
      "step": 2045
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.055516481399536,
      "learning_rate": 1.9914059528505778e-05,
      "loss": 0.9128,
      "step": 2046
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8698420524597168,
      "learning_rate": 1.9913913394937583e-05,
      "loss": 0.6375,
      "step": 2047
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.447360634803772,
      "learning_rate": 1.9913767137768908e-05,
      "loss": 0.8585,
      "step": 2048
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9367032051086426,
      "learning_rate": 1.9913620757001578e-05,
      "loss": 0.9315,
      "step": 2049
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1453211307525635,
      "learning_rate": 1.9913474252637416e-05,
      "loss": 0.8317,
      "step": 2050
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7642085552215576,
      "learning_rate": 1.991332762467825e-05,
      "loss": 0.6788,
      "step": 2051
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7114896774291992,
      "learning_rate": 1.991318087312591e-05,
      "loss": 0.8548,
      "step": 2052
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9198087453842163,
      "learning_rate": 1.991303399798222e-05,
      "loss": 0.7238,
      "step": 2053
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.770117163658142,
      "learning_rate": 1.9912886999249012e-05,
      "loss": 0.8544,
      "step": 2054
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7171108722686768,
      "learning_rate": 1.9912739876928128e-05,
      "loss": 0.7776,
      "step": 2055
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.81100594997406,
      "learning_rate": 1.991259263102139e-05,
      "loss": 0.7991,
      "step": 2056
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1037421226501465,
      "learning_rate": 1.991244526153064e-05,
      "loss": 0.8458,
      "step": 2057
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.771275758743286,
      "learning_rate": 1.9912297768457717e-05,
      "loss": 0.8861,
      "step": 2058
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1742568016052246,
      "learning_rate": 1.9912150151804454e-05,
      "loss": 0.9475,
      "step": 2059
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6369812488555908,
      "learning_rate": 1.9912002411572696e-05,
      "loss": 0.8588,
      "step": 2060
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.621869683265686,
      "learning_rate": 1.9911854547764283e-05,
      "loss": 0.8999,
      "step": 2061
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.026409864425659,
      "learning_rate": 1.991170656038106e-05,
      "loss": 0.6986,
      "step": 2062
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6105358600616455,
      "learning_rate": 1.9911558449424873e-05,
      "loss": 0.8342,
      "step": 2063
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.315835952758789,
      "learning_rate": 1.9911410214897566e-05,
      "loss": 0.879,
      "step": 2064
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0364246368408203,
      "learning_rate": 1.9911261856800986e-05,
      "loss": 0.795,
      "step": 2065
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7460386753082275,
      "learning_rate": 1.9911113375136983e-05,
      "loss": 0.8394,
      "step": 2066
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1984024047851562,
      "learning_rate": 1.991096476990741e-05,
      "loss": 0.7071,
      "step": 2067
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.6971702575683594,
      "learning_rate": 1.991081604111412e-05,
      "loss": 0.9903,
      "step": 2068
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7173709869384766,
      "learning_rate": 1.991066718875897e-05,
      "loss": 0.697,
      "step": 2069
    },
    {
      "epoch": 0.21,
      "grad_norm": 3.5401744842529297,
      "learning_rate": 1.991051821284381e-05,
      "loss": 0.7191,
      "step": 2070
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1978909969329834,
      "learning_rate": 1.9910369113370498e-05,
      "loss": 0.9765,
      "step": 2071
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7334545850753784,
      "learning_rate": 1.9910219890340895e-05,
      "loss": 0.7645,
      "step": 2072
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9179364442825317,
      "learning_rate": 1.9910070543756862e-05,
      "loss": 0.802,
      "step": 2073
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1797971725463867,
      "learning_rate": 1.990992107362026e-05,
      "loss": 0.7964,
      "step": 2074
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1385600566864014,
      "learning_rate": 1.990977147993295e-05,
      "loss": 0.8349,
      "step": 2075
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.8414525985717773,
      "learning_rate": 1.9909621762696805e-05,
      "loss": 0.8912,
      "step": 2076
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1414577960968018,
      "learning_rate": 1.9909471921913682e-05,
      "loss": 0.7606,
      "step": 2077
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1811184883117676,
      "learning_rate": 1.990932195758545e-05,
      "loss": 0.9514,
      "step": 2078
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.885715365409851,
      "learning_rate": 1.990917186971399e-05,
      "loss": 0.8365,
      "step": 2079
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.182248592376709,
      "learning_rate": 1.9909021658301158e-05,
      "loss": 0.8095,
      "step": 2080
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.049020767211914,
      "learning_rate": 1.9908871323348836e-05,
      "loss": 0.8696,
      "step": 2081
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.199831247329712,
      "learning_rate": 1.9908720864858897e-05,
      "loss": 0.933,
      "step": 2082
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.854202389717102,
      "learning_rate": 1.9908570282833214e-05,
      "loss": 0.7969,
      "step": 2083
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9232800006866455,
      "learning_rate": 1.990841957727367e-05,
      "loss": 0.7524,
      "step": 2084
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3891141414642334,
      "learning_rate": 1.990826874818214e-05,
      "loss": 1.003,
      "step": 2085
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8980940580368042,
      "learning_rate": 1.99081177955605e-05,
      "loss": 0.8637,
      "step": 2086
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.261415481567383,
      "learning_rate": 1.990796671941064e-05,
      "loss": 0.8815,
      "step": 2087
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8627233505249023,
      "learning_rate": 1.9907815519734436e-05,
      "loss": 0.7677,
      "step": 2088
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6248915195465088,
      "learning_rate": 1.990766419653378e-05,
      "loss": 0.9371,
      "step": 2089
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1352434158325195,
      "learning_rate": 1.9907512749810555e-05,
      "loss": 0.8606,
      "step": 2090
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8389860391616821,
      "learning_rate": 1.990736117956665e-05,
      "loss": 0.7212,
      "step": 2091
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.015244960784912,
      "learning_rate": 1.990720948580396e-05,
      "loss": 0.7765,
      "step": 2092
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9794825315475464,
      "learning_rate": 1.9907057668524362e-05,
      "loss": 0.7399,
      "step": 2093
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.889494776725769,
      "learning_rate": 1.990690572772976e-05,
      "loss": 0.8948,
      "step": 2094
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6866103410720825,
      "learning_rate": 1.990675366342205e-05,
      "loss": 0.8804,
      "step": 2095
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1013529300689697,
      "learning_rate": 1.990660147560312e-05,
      "loss": 1.0876,
      "step": 2096
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9872796535491943,
      "learning_rate": 1.990644916427487e-05,
      "loss": 1.0121,
      "step": 2097
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0229828357696533,
      "learning_rate": 1.9906296729439206e-05,
      "loss": 0.9261,
      "step": 2098
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9623749256134033,
      "learning_rate": 1.990614417109802e-05,
      "loss": 0.7093,
      "step": 2099
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8729890584945679,
      "learning_rate": 1.9905991489253214e-05,
      "loss": 0.8085,
      "step": 2100
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.847267985343933,
      "learning_rate": 1.9905838683906693e-05,
      "loss": 0.8581,
      "step": 2101
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.049931049346924,
      "learning_rate": 1.9905685755060365e-05,
      "loss": 0.8078,
      "step": 2102
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8256869316101074,
      "learning_rate": 1.9905532702716136e-05,
      "loss": 0.8451,
      "step": 2103
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7374293804168701,
      "learning_rate": 1.9905379526875913e-05,
      "loss": 0.7644,
      "step": 2104
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8098464012145996,
      "learning_rate": 1.9905226227541603e-05,
      "loss": 0.8561,
      "step": 2105
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1500508785247803,
      "learning_rate": 1.9905072804715124e-05,
      "loss": 0.7878,
      "step": 2106
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8590757846832275,
      "learning_rate": 1.990491925839838e-05,
      "loss": 0.8415,
      "step": 2107
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.615122675895691,
      "learning_rate": 1.9904765588593292e-05,
      "loss": 0.7286,
      "step": 2108
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9054012298583984,
      "learning_rate": 1.9904611795301775e-05,
      "loss": 0.7804,
      "step": 2109
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.4924612045288086,
      "learning_rate": 1.9904457878525742e-05,
      "loss": 0.8797,
      "step": 2110
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7054641246795654,
      "learning_rate": 1.990430383826712e-05,
      "loss": 0.6195,
      "step": 2111
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.328518867492676,
      "learning_rate": 1.990414967452782e-05,
      "loss": 0.9536,
      "step": 2112
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7498852014541626,
      "learning_rate": 1.990399538730977e-05,
      "loss": 0.9195,
      "step": 2113
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6039657592773438,
      "learning_rate": 1.9903840976614894e-05,
      "loss": 0.8815,
      "step": 2114
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9247539043426514,
      "learning_rate": 1.9903686442445114e-05,
      "loss": 0.7998,
      "step": 2115
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7859625816345215,
      "learning_rate": 1.9903531784802353e-05,
      "loss": 0.9129,
      "step": 2116
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.512420177459717,
      "learning_rate": 1.9903377003688554e-05,
      "loss": 0.8471,
      "step": 2117
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9063235521316528,
      "learning_rate": 1.990322209910563e-05,
      "loss": 0.668,
      "step": 2118
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.210592269897461,
      "learning_rate": 1.990306707105552e-05,
      "loss": 0.6909,
      "step": 2119
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8993234634399414,
      "learning_rate": 1.9902911919540156e-05,
      "loss": 0.936,
      "step": 2120
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7960516214370728,
      "learning_rate": 1.9902756644561473e-05,
      "loss": 0.9457,
      "step": 2121
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8443907499313354,
      "learning_rate": 1.9902601246121406e-05,
      "loss": 0.9081,
      "step": 2122
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.090520143508911,
      "learning_rate": 1.9902445724221892e-05,
      "loss": 0.7215,
      "step": 2123
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.406083345413208,
      "learning_rate": 1.990229007886487e-05,
      "loss": 0.796,
      "step": 2124
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.460015296936035,
      "learning_rate": 1.990213431005228e-05,
      "loss": 0.8423,
      "step": 2125
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2201740741729736,
      "learning_rate": 1.990197841778607e-05,
      "loss": 0.8884,
      "step": 2126
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0132744312286377,
      "learning_rate": 1.9901822402068173e-05,
      "loss": 0.6927,
      "step": 2127
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1650006771087646,
      "learning_rate": 1.990166626290054e-05,
      "loss": 0.7859,
      "step": 2128
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8683929443359375,
      "learning_rate": 1.990151000028512e-05,
      "loss": 0.809,
      "step": 2129
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2768514156341553,
      "learning_rate": 1.9901353614223855e-05,
      "loss": 0.8841,
      "step": 2130
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5555247068405151,
      "learning_rate": 1.99011971047187e-05,
      "loss": 0.7272,
      "step": 2131
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8610453605651855,
      "learning_rate": 1.9901040471771606e-05,
      "loss": 0.7882,
      "step": 2132
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1595964431762695,
      "learning_rate": 1.9900883715384523e-05,
      "loss": 0.6912,
      "step": 2133
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8067100048065186,
      "learning_rate": 1.9900726835559407e-05,
      "loss": 0.5968,
      "step": 2134
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0374608039855957,
      "learning_rate": 1.9900569832298212e-05,
      "loss": 0.899,
      "step": 2135
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0008914470672607,
      "learning_rate": 1.9900412705602897e-05,
      "loss": 0.7899,
      "step": 2136
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0064587593078613,
      "learning_rate": 1.990025545547542e-05,
      "loss": 0.692,
      "step": 2137
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.626063346862793,
      "learning_rate": 1.9900098081917746e-05,
      "loss": 0.9537,
      "step": 2138
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.698777437210083,
      "learning_rate": 1.989994058493183e-05,
      "loss": 0.9442,
      "step": 2139
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.107210397720337,
      "learning_rate": 1.9899782964519642e-05,
      "loss": 0.9001,
      "step": 2140
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5762754678726196,
      "learning_rate": 1.9899625220683142e-05,
      "loss": 0.8087,
      "step": 2141
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0714192390441895,
      "learning_rate": 1.98994673534243e-05,
      "loss": 1.0076,
      "step": 2142
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.200153112411499,
      "learning_rate": 1.989930936274508e-05,
      "loss": 0.9555,
      "step": 2143
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9518722295761108,
      "learning_rate": 1.9899151248647457e-05,
      "loss": 0.7077,
      "step": 2144
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8287770748138428,
      "learning_rate": 1.98989930111334e-05,
      "loss": 0.7832,
      "step": 2145
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.910875678062439,
      "learning_rate": 1.989883465020488e-05,
      "loss": 0.6831,
      "step": 2146
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2311556339263916,
      "learning_rate": 1.989867616586387e-05,
      "loss": 0.8744,
      "step": 2147
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7803906202316284,
      "learning_rate": 1.9898517558112356e-05,
      "loss": 0.8296,
      "step": 2148
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0426197052001953,
      "learning_rate": 1.9898358826952308e-05,
      "loss": 0.7487,
      "step": 2149
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.902571201324463,
      "learning_rate": 1.9898199972385698e-05,
      "loss": 0.8214,
      "step": 2150
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9668489694595337,
      "learning_rate": 1.989804099441452e-05,
      "loss": 0.8798,
      "step": 2151
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7940902709960938,
      "learning_rate": 1.9897881893040747e-05,
      "loss": 0.813,
      "step": 2152
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.053135395050049,
      "learning_rate": 1.9897722668266367e-05,
      "loss": 0.7668,
      "step": 2153
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5833693742752075,
      "learning_rate": 1.9897563320093364e-05,
      "loss": 0.7128,
      "step": 2154
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0340399742126465,
      "learning_rate": 1.989740384852372e-05,
      "loss": 0.7291,
      "step": 2155
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1018624305725098,
      "learning_rate": 1.9897244253559435e-05,
      "loss": 0.7441,
      "step": 2156
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.063577175140381,
      "learning_rate": 1.9897084535202483e-05,
      "loss": 0.75,
      "step": 2157
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1744143962860107,
      "learning_rate": 1.9896924693454867e-05,
      "loss": 0.7624,
      "step": 2158
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6545543670654297,
      "learning_rate": 1.9896764728318576e-05,
      "loss": 0.725,
      "step": 2159
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6812490224838257,
      "learning_rate": 1.9896604639795606e-05,
      "loss": 0.8967,
      "step": 2160
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8000935316085815,
      "learning_rate": 1.9896444427887948e-05,
      "loss": 0.7764,
      "step": 2161
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.203953266143799,
      "learning_rate": 1.9896284092597605e-05,
      "loss": 0.674,
      "step": 2162
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.124861001968384,
      "learning_rate": 1.9896123633926577e-05,
      "loss": 0.7666,
      "step": 2163
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3813250064849854,
      "learning_rate": 1.9895963051876857e-05,
      "loss": 0.8017,
      "step": 2164
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.076467514038086,
      "learning_rate": 1.989580234645045e-05,
      "loss": 0.84,
      "step": 2165
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8103898763656616,
      "learning_rate": 1.9895641517649365e-05,
      "loss": 0.7074,
      "step": 2166
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9396142959594727,
      "learning_rate": 1.9895480565475602e-05,
      "loss": 0.8202,
      "step": 2167
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.622430682182312,
      "learning_rate": 1.9895319489931166e-05,
      "loss": 0.7975,
      "step": 2168
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8788862228393555,
      "learning_rate": 1.989515829101807e-05,
      "loss": 0.7467,
      "step": 2169
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7609659433364868,
      "learning_rate": 1.989499696873832e-05,
      "loss": 0.7107,
      "step": 2170
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.1517245769500732,
      "learning_rate": 1.989483552309393e-05,
      "loss": 0.9055,
      "step": 2171
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7945953607559204,
      "learning_rate": 1.9894673954086915e-05,
      "loss": 0.7644,
      "step": 2172
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9719959497451782,
      "learning_rate": 1.989451226171928e-05,
      "loss": 0.9128,
      "step": 2173
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9052788019180298,
      "learning_rate": 1.9894350445993052e-05,
      "loss": 0.8538,
      "step": 2174
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.0805773735046387,
      "learning_rate": 1.989418850691024e-05,
      "loss": 0.8942,
      "step": 2175
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.648906946182251,
      "learning_rate": 1.9894026444472866e-05,
      "loss": 0.9809,
      "step": 2176
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.263112783432007,
      "learning_rate": 1.989386425868295e-05,
      "loss": 0.8741,
      "step": 2177
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.144439220428467,
      "learning_rate": 1.989370194954252e-05,
      "loss": 0.7808,
      "step": 2178
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.139998435974121,
      "learning_rate": 1.9893539517053592e-05,
      "loss": 0.8308,
      "step": 2179
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7194597721099854,
      "learning_rate": 1.989337696121819e-05,
      "loss": 0.7728,
      "step": 2180
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.533400535583496,
      "learning_rate": 1.9893214282038345e-05,
      "loss": 0.952,
      "step": 2181
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.020662784576416,
      "learning_rate": 1.989305147951609e-05,
      "loss": 0.6952,
      "step": 2182
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6004233360290527,
      "learning_rate": 1.989288855365344e-05,
      "loss": 0.9139,
      "step": 2183
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0539958477020264,
      "learning_rate": 1.9892725504452442e-05,
      "loss": 0.8776,
      "step": 2184
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8169182538986206,
      "learning_rate": 1.9892562331915118e-05,
      "loss": 0.5698,
      "step": 2185
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.914350986480713,
      "learning_rate": 1.9892399036043512e-05,
      "loss": 0.7805,
      "step": 2186
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.73432195186615,
      "learning_rate": 1.9892235616839646e-05,
      "loss": 0.6293,
      "step": 2187
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.262955904006958,
      "learning_rate": 1.989207207430557e-05,
      "loss": 0.8535,
      "step": 2188
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7253910303115845,
      "learning_rate": 1.9891908408443323e-05,
      "loss": 0.7966,
      "step": 2189
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0330705642700195,
      "learning_rate": 1.9891744619254934e-05,
      "loss": 0.933,
      "step": 2190
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.005059003829956,
      "learning_rate": 1.9891580706742454e-05,
      "loss": 0.7837,
      "step": 2191
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.249779462814331,
      "learning_rate": 1.9891416670907927e-05,
      "loss": 0.8172,
      "step": 2192
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.654079794883728,
      "learning_rate": 1.9891252511753392e-05,
      "loss": 0.6172,
      "step": 2193
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7895411252975464,
      "learning_rate": 1.9891088229280906e-05,
      "loss": 0.7973,
      "step": 2194
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8266199827194214,
      "learning_rate": 1.9890923823492505e-05,
      "loss": 0.8662,
      "step": 2195
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.801957130432129,
      "learning_rate": 1.9890759294390245e-05,
      "loss": 0.8129,
      "step": 2196
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.171485424041748,
      "learning_rate": 1.9890594641976177e-05,
      "loss": 0.9877,
      "step": 2197
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.849527359008789,
      "learning_rate": 1.9890429866252353e-05,
      "loss": 0.751,
      "step": 2198
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7918219566345215,
      "learning_rate": 1.989026496722083e-05,
      "loss": 0.8065,
      "step": 2199
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9495075941085815,
      "learning_rate": 1.989009994488366e-05,
      "loss": 0.7558,
      "step": 2200
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.679559350013733,
      "learning_rate": 1.98899347992429e-05,
      "loss": 0.8527,
      "step": 2201
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7863179445266724,
      "learning_rate": 1.9889769530300612e-05,
      "loss": 0.8771,
      "step": 2202
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.703275442123413,
      "learning_rate": 1.9889604138058857e-05,
      "loss": 0.7566,
      "step": 2203
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0503647327423096,
      "learning_rate": 1.9889438622519692e-05,
      "loss": 0.7557,
      "step": 2204
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.729037880897522,
      "learning_rate": 1.9889272983685186e-05,
      "loss": 0.8598,
      "step": 2205
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.708093643188477,
      "learning_rate": 1.98891072215574e-05,
      "loss": 0.9682,
      "step": 2206
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9489692449569702,
      "learning_rate": 1.9888941336138405e-05,
      "loss": 0.8325,
      "step": 2207
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0842761993408203,
      "learning_rate": 1.9888775327430266e-05,
      "loss": 0.9671,
      "step": 2208
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6476935148239136,
      "learning_rate": 1.9888609195435054e-05,
      "loss": 0.7995,
      "step": 2209
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9071111679077148,
      "learning_rate": 1.9888442940154837e-05,
      "loss": 0.6031,
      "step": 2210
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0474443435668945,
      "learning_rate": 1.9888276561591692e-05,
      "loss": 0.712,
      "step": 2211
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.924757719039917,
      "learning_rate": 1.9888110059747692e-05,
      "loss": 0.8878,
      "step": 2212
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.143160581588745,
      "learning_rate": 1.988794343462491e-05,
      "loss": 0.7917,
      "step": 2213
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6845723390579224,
      "learning_rate": 1.988777668622543e-05,
      "loss": 0.636,
      "step": 2214
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.754797339439392,
      "learning_rate": 1.9887609814551325e-05,
      "loss": 0.7294,
      "step": 2215
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0529236793518066,
      "learning_rate": 1.9887442819604676e-05,
      "loss": 1.1073,
      "step": 2216
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6569385528564453,
      "learning_rate": 1.988727570138757e-05,
      "loss": 0.9146,
      "step": 2217
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.7918758392333984,
      "learning_rate": 1.9887108459902085e-05,
      "loss": 0.9324,
      "step": 2218
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8747469186782837,
      "learning_rate": 1.9886941095150304e-05,
      "loss": 0.8667,
      "step": 2219
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9245120286941528,
      "learning_rate": 1.988677360713432e-05,
      "loss": 0.8862,
      "step": 2220
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.106210231781006,
      "learning_rate": 1.988660599585622e-05,
      "loss": 0.8224,
      "step": 2221
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8764290809631348,
      "learning_rate": 1.988643826131809e-05,
      "loss": 0.8382,
      "step": 2222
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1322789192199707,
      "learning_rate": 1.9886270403522028e-05,
      "loss": 0.7896,
      "step": 2223
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.153358221054077,
      "learning_rate": 1.9886102422470118e-05,
      "loss": 0.8163,
      "step": 2224
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8520832061767578,
      "learning_rate": 1.988593431816446e-05,
      "loss": 0.7965,
      "step": 2225
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.015265941619873,
      "learning_rate": 1.9885766090607147e-05,
      "loss": 0.7558,
      "step": 2226
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9271150827407837,
      "learning_rate": 1.9885597739800277e-05,
      "loss": 0.9532,
      "step": 2227
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0688657760620117,
      "learning_rate": 1.9885429265745952e-05,
      "loss": 0.8079,
      "step": 2228
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7520681619644165,
      "learning_rate": 1.988526066844627e-05,
      "loss": 0.7469,
      "step": 2229
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6224627494812012,
      "learning_rate": 1.988509194790333e-05,
      "loss": 0.926,
      "step": 2230
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.1580493450164795,
      "learning_rate": 1.988492310411924e-05,
      "loss": 0.6194,
      "step": 2231
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5365782976150513,
      "learning_rate": 1.9884754137096105e-05,
      "loss": 0.7872,
      "step": 2232
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8219268321990967,
      "learning_rate": 1.9884585046836028e-05,
      "loss": 0.7684,
      "step": 2233
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7781325578689575,
      "learning_rate": 1.988441583334112e-05,
      "loss": 0.7718,
      "step": 2234
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.089327096939087,
      "learning_rate": 1.988424649661349e-05,
      "loss": 0.7907,
      "step": 2235
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.77101469039917,
      "learning_rate": 1.988407703665525e-05,
      "loss": 0.732,
      "step": 2236
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.772105097770691,
      "learning_rate": 1.988390745346851e-05,
      "loss": 0.7773,
      "step": 2237
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8927452564239502,
      "learning_rate": 1.9883737747055387e-05,
      "loss": 0.7122,
      "step": 2238
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0702948570251465,
      "learning_rate": 1.9883567917417997e-05,
      "loss": 0.7486,
      "step": 2239
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6907103061676025,
      "learning_rate": 1.9883397964558452e-05,
      "loss": 0.8399,
      "step": 2240
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9429739713668823,
      "learning_rate": 1.9883227888478876e-05,
      "loss": 0.7882,
      "step": 2241
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8009506464004517,
      "learning_rate": 1.9883057689181393e-05,
      "loss": 0.9123,
      "step": 2242
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7336384057998657,
      "learning_rate": 1.988288736666812e-05,
      "loss": 0.6924,
      "step": 2243
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.5624167919158936,
      "learning_rate": 1.9882716920941175e-05,
      "loss": 0.7404,
      "step": 2244
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.3086366653442383,
      "learning_rate": 1.9882546352002692e-05,
      "loss": 0.9275,
      "step": 2245
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6498301029205322,
      "learning_rate": 1.9882375659854793e-05,
      "loss": 0.8436,
      "step": 2246
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0978970527648926,
      "learning_rate": 1.988220484449961e-05,
      "loss": 0.887,
      "step": 2247
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6363774538040161,
      "learning_rate": 1.9882033905939268e-05,
      "loss": 0.8571,
      "step": 2248
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.987053632736206,
      "learning_rate": 1.98818628441759e-05,
      "loss": 0.8118,
      "step": 2249
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8889193534851074,
      "learning_rate": 1.9881691659211643e-05,
      "loss": 0.7622,
      "step": 2250
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.947100043296814,
      "learning_rate": 1.988152035104862e-05,
      "loss": 0.8358,
      "step": 2251
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9004617929458618,
      "learning_rate": 1.988134891968898e-05,
      "loss": 0.7996,
      "step": 2252
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.991628885269165,
      "learning_rate": 1.9881177365134852e-05,
      "loss": 0.74,
      "step": 2253
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.658215880393982,
      "learning_rate": 1.9881005687388377e-05,
      "loss": 0.8746,
      "step": 2254
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0695533752441406,
      "learning_rate": 1.9880833886451697e-05,
      "loss": 0.9072,
      "step": 2255
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.4996418952941895,
      "learning_rate": 1.988066196232695e-05,
      "loss": 0.7818,
      "step": 2256
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.420750617980957,
      "learning_rate": 1.988048991501628e-05,
      "loss": 0.6299,
      "step": 2257
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.175931930541992,
      "learning_rate": 1.988031774452184e-05,
      "loss": 0.9385,
      "step": 2258
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6104998588562012,
      "learning_rate": 1.9880145450845764e-05,
      "loss": 0.7339,
      "step": 2259
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0842323303222656,
      "learning_rate": 1.9879973033990208e-05,
      "loss": 0.7141,
      "step": 2260
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.777282476425171,
      "learning_rate": 1.9879800493957318e-05,
      "loss": 0.6865,
      "step": 2261
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3971716165542603,
      "learning_rate": 1.9879627830749253e-05,
      "loss": 0.7939,
      "step": 2262
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6652374267578125,
      "learning_rate": 1.9879455044368154e-05,
      "loss": 0.8436,
      "step": 2263
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.219456911087036,
      "learning_rate": 1.9879282134816185e-05,
      "loss": 0.9381,
      "step": 2264
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6671035289764404,
      "learning_rate": 1.9879109102095496e-05,
      "loss": 0.862,
      "step": 2265
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.177830219268799,
      "learning_rate": 1.9878935946208243e-05,
      "loss": 0.8573,
      "step": 2266
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0465145111083984,
      "learning_rate": 1.987876266715659e-05,
      "loss": 1.0019,
      "step": 2267
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0792171955108643,
      "learning_rate": 1.9878589264942697e-05,
      "loss": 0.7537,
      "step": 2268
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.0140960216522217,
      "learning_rate": 1.9878415739568722e-05,
      "loss": 0.9272,
      "step": 2269
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2016677856445312,
      "learning_rate": 1.987824209103683e-05,
      "loss": 0.8691,
      "step": 2270
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.097804546356201,
      "learning_rate": 1.9878068319349186e-05,
      "loss": 0.8998,
      "step": 2271
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8416907787322998,
      "learning_rate": 1.9877894424507958e-05,
      "loss": 0.8732,
      "step": 2272
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.4786126613616943,
      "learning_rate": 1.9877720406515312e-05,
      "loss": 0.6238,
      "step": 2273
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9306983947753906,
      "learning_rate": 1.9877546265373416e-05,
      "loss": 0.7758,
      "step": 2274
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.391683340072632,
      "learning_rate": 1.9877372001084447e-05,
      "loss": 0.8902,
      "step": 2275
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0228092670440674,
      "learning_rate": 1.987719761365057e-05,
      "loss": 0.8572,
      "step": 2276
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.756330966949463,
      "learning_rate": 1.9877023103073965e-05,
      "loss": 0.7799,
      "step": 2277
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7743476629257202,
      "learning_rate": 1.9876848469356806e-05,
      "loss": 0.8757,
      "step": 2278
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9102826118469238,
      "learning_rate": 1.987667371250127e-05,
      "loss": 0.7647,
      "step": 2279
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7642695903778076,
      "learning_rate": 1.9876498832509537e-05,
      "loss": 0.8093,
      "step": 2280
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8910906314849854,
      "learning_rate": 1.9876323829383787e-05,
      "loss": 0.7193,
      "step": 2281
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6103936433792114,
      "learning_rate": 1.9876148703126196e-05,
      "loss": 0.7688,
      "step": 2282
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0486741065979004,
      "learning_rate": 1.9875973453738955e-05,
      "loss": 0.8766,
      "step": 2283
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9243412017822266,
      "learning_rate": 1.9875798081224244e-05,
      "loss": 0.8875,
      "step": 2284
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7445322275161743,
      "learning_rate": 1.9875622585584255e-05,
      "loss": 0.8685,
      "step": 2285
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1025590896606445,
      "learning_rate": 1.9875446966821172e-05,
      "loss": 0.8421,
      "step": 2286
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0528104305267334,
      "learning_rate": 1.9875271224937185e-05,
      "loss": 0.8048,
      "step": 2287
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.870100975036621,
      "learning_rate": 1.9875095359934483e-05,
      "loss": 0.8973,
      "step": 2288
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5844817161560059,
      "learning_rate": 1.9874919371815263e-05,
      "loss": 0.8678,
      "step": 2289
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7373285293579102,
      "learning_rate": 1.9874743260581716e-05,
      "loss": 0.7495,
      "step": 2290
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0007383823394775,
      "learning_rate": 1.9874567026236037e-05,
      "loss": 0.8437,
      "step": 2291
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.656772255897522,
      "learning_rate": 1.9874390668780427e-05,
      "loss": 0.7781,
      "step": 2292
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8992714881896973,
      "learning_rate": 1.987421418821708e-05,
      "loss": 0.8426,
      "step": 2293
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.129793643951416,
      "learning_rate": 1.98740375845482e-05,
      "loss": 0.7363,
      "step": 2294
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6023792028427124,
      "learning_rate": 1.9873860857775985e-05,
      "loss": 0.8252,
      "step": 2295
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.869383454322815,
      "learning_rate": 1.9873684007902643e-05,
      "loss": 0.9555,
      "step": 2296
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8953756093978882,
      "learning_rate": 1.9873507034930376e-05,
      "loss": 0.6824,
      "step": 2297
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9720227718353271,
      "learning_rate": 1.9873329938861393e-05,
      "loss": 0.7358,
      "step": 2298
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.021120309829712,
      "learning_rate": 1.9873152719697898e-05,
      "loss": 0.7174,
      "step": 2299
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.009195566177368,
      "learning_rate": 1.9872975377442102e-05,
      "loss": 0.6201,
      "step": 2300
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7482753992080688,
      "learning_rate": 1.9872797912096217e-05,
      "loss": 0.7898,
      "step": 2301
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5951727628707886,
      "learning_rate": 1.987262032366245e-05,
      "loss": 0.8404,
      "step": 2302
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1334187984466553,
      "learning_rate": 1.9872442612143028e-05,
      "loss": 1.0034,
      "step": 2303
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9778738021850586,
      "learning_rate": 1.9872264777540153e-05,
      "loss": 0.9716,
      "step": 2304
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.704412817955017,
      "learning_rate": 1.987208681985605e-05,
      "loss": 0.8753,
      "step": 2305
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9746133089065552,
      "learning_rate": 1.9871908739092934e-05,
      "loss": 0.8974,
      "step": 2306
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6980184316635132,
      "learning_rate": 1.987173053525303e-05,
      "loss": 0.7352,
      "step": 2307
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.811883568763733,
      "learning_rate": 1.9871552208338548e-05,
      "loss": 0.7908,
      "step": 2308
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0431787967681885,
      "learning_rate": 1.9871373758351725e-05,
      "loss": 0.9354,
      "step": 2309
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7206717729568481,
      "learning_rate": 1.987119518529478e-05,
      "loss": 0.769,
      "step": 2310
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.182011127471924,
      "learning_rate": 1.987101648916994e-05,
      "loss": 0.938,
      "step": 2311
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9287129640579224,
      "learning_rate": 1.9870837669979433e-05,
      "loss": 0.7327,
      "step": 2312
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8750052452087402,
      "learning_rate": 1.9870658727725488e-05,
      "loss": 0.8884,
      "step": 2313
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7463796138763428,
      "learning_rate": 1.9870479662410334e-05,
      "loss": 0.6969,
      "step": 2314
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7788141965866089,
      "learning_rate": 1.9870300474036207e-05,
      "loss": 0.8851,
      "step": 2315
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1851165294647217,
      "learning_rate": 1.9870121162605337e-05,
      "loss": 0.77,
      "step": 2316
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8676153421401978,
      "learning_rate": 1.9869941728119963e-05,
      "loss": 0.8642,
      "step": 2317
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8154371976852417,
      "learning_rate": 1.9869762170582322e-05,
      "loss": 0.6795,
      "step": 2318
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.081228733062744,
      "learning_rate": 1.986958248999465e-05,
      "loss": 0.8558,
      "step": 2319
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8880722522735596,
      "learning_rate": 1.9869402686359188e-05,
      "loss": 0.7485,
      "step": 2320
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0823428630828857,
      "learning_rate": 1.9869222759678178e-05,
      "loss": 0.7133,
      "step": 2321
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1758925914764404,
      "learning_rate": 1.986904270995387e-05,
      "loss": 0.7381,
      "step": 2322
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.25016188621521,
      "learning_rate": 1.9868862537188497e-05,
      "loss": 0.9393,
      "step": 2323
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8855465650558472,
      "learning_rate": 1.986868224138431e-05,
      "loss": 0.8932,
      "step": 2324
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6507364511489868,
      "learning_rate": 1.986850182254356e-05,
      "loss": 0.7375,
      "step": 2325
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.014913320541382,
      "learning_rate": 1.986832128066849e-05,
      "loss": 0.7879,
      "step": 2326
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9193238019943237,
      "learning_rate": 1.9868140615761358e-05,
      "loss": 0.8325,
      "step": 2327
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7166224718093872,
      "learning_rate": 1.9867959827824413e-05,
      "loss": 0.7486,
      "step": 2328
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2042930126190186,
      "learning_rate": 1.9867778916859907e-05,
      "loss": 0.9418,
      "step": 2329
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2443482875823975,
      "learning_rate": 1.98675978828701e-05,
      "loss": 0.8211,
      "step": 2330
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5894591808319092,
      "learning_rate": 1.9867416725857243e-05,
      "loss": 0.7968,
      "step": 2331
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.940802812576294,
      "learning_rate": 1.9867235445823602e-05,
      "loss": 0.7718,
      "step": 2332
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0828118324279785,
      "learning_rate": 1.986705404277143e-05,
      "loss": 0.839,
      "step": 2333
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9010075330734253,
      "learning_rate": 1.9866872516702994e-05,
      "loss": 0.8928,
      "step": 2334
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9439140558242798,
      "learning_rate": 1.9866690867620554e-05,
      "loss": 0.7089,
      "step": 2335
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2307302951812744,
      "learning_rate": 1.9866509095526375e-05,
      "loss": 0.9229,
      "step": 2336
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.868897795677185,
      "learning_rate": 1.9866327200422724e-05,
      "loss": 0.9317,
      "step": 2337
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.051461935043335,
      "learning_rate": 1.9866145182311867e-05,
      "loss": 0.6389,
      "step": 2338
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7394131422042847,
      "learning_rate": 1.986596304119608e-05,
      "loss": 0.63,
      "step": 2339
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9594241380691528,
      "learning_rate": 1.9865780777077625e-05,
      "loss": 0.8196,
      "step": 2340
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2549374103546143,
      "learning_rate": 1.986559838995878e-05,
      "loss": 0.7871,
      "step": 2341
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.03761625289917,
      "learning_rate": 1.9865415879841815e-05,
      "loss": 0.686,
      "step": 2342
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.212421178817749,
      "learning_rate": 1.9865233246729006e-05,
      "loss": 0.9136,
      "step": 2343
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.967837929725647,
      "learning_rate": 1.9865050490622636e-05,
      "loss": 0.8049,
      "step": 2344
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9590404033660889,
      "learning_rate": 1.9864867611524977e-05,
      "loss": 0.7183,
      "step": 2345
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.14086651802063,
      "learning_rate": 1.986468460943831e-05,
      "loss": 0.6528,
      "step": 2346
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9340685606002808,
      "learning_rate": 1.9864501484364917e-05,
      "loss": 0.9268,
      "step": 2347
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.131622552871704,
      "learning_rate": 1.9864318236307083e-05,
      "loss": 0.812,
      "step": 2348
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8949648141860962,
      "learning_rate": 1.9864134865267088e-05,
      "loss": 0.5656,
      "step": 2349
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1110165119171143,
      "learning_rate": 1.9863951371247223e-05,
      "loss": 0.796,
      "step": 2350
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9480600357055664,
      "learning_rate": 1.9863767754249776e-05,
      "loss": 0.836,
      "step": 2351
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5589364767074585,
      "learning_rate": 1.9863584014277032e-05,
      "loss": 0.8358,
      "step": 2352
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6279774904251099,
      "learning_rate": 1.9863400151331286e-05,
      "loss": 0.7175,
      "step": 2353
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6990430355072021,
      "learning_rate": 1.986321616541483e-05,
      "loss": 0.7249,
      "step": 2354
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9201478958129883,
      "learning_rate": 1.9863032056529946e-05,
      "loss": 0.728,
      "step": 2355
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7655003070831299,
      "learning_rate": 1.986284782467895e-05,
      "loss": 0.9489,
      "step": 2356
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7791883945465088,
      "learning_rate": 1.9862663469864122e-05,
      "loss": 0.7946,
      "step": 2357
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6470954418182373,
      "learning_rate": 1.986247899208777e-05,
      "loss": 0.7418,
      "step": 2358
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.702484369277954,
      "learning_rate": 1.9862294391352188e-05,
      "loss": 0.8822,
      "step": 2359
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.894893765449524,
      "learning_rate": 1.9862109667659682e-05,
      "loss": 0.8534,
      "step": 2360
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.889972448348999,
      "learning_rate": 1.9861924821012555e-05,
      "loss": 0.7302,
      "step": 2361
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1445562839508057,
      "learning_rate": 1.986173985141311e-05,
      "loss": 0.85,
      "step": 2362
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.7525297403335571,
      "learning_rate": 1.986155475886365e-05,
      "loss": 0.7535,
      "step": 2363
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.0237672328948975,
      "learning_rate": 1.9861369543366483e-05,
      "loss": 0.7269,
      "step": 2364
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.92788827419281,
      "learning_rate": 1.9861184204923926e-05,
      "loss": 0.8099,
      "step": 2365
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.869275450706482,
      "learning_rate": 1.986099874353828e-05,
      "loss": 0.8027,
      "step": 2366
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9245903491973877,
      "learning_rate": 1.9860813159211864e-05,
      "loss": 0.8901,
      "step": 2367
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9420045614242554,
      "learning_rate": 1.986062745194699e-05,
      "loss": 0.6986,
      "step": 2368
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6623085737228394,
      "learning_rate": 1.986044162174597e-05,
      "loss": 0.9137,
      "step": 2369
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.043455123901367,
      "learning_rate": 1.9860255668611122e-05,
      "loss": 0.8132,
      "step": 2370
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.877676248550415,
      "learning_rate": 1.986006959254477e-05,
      "loss": 0.792,
      "step": 2371
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9689875841140747,
      "learning_rate": 1.985988339354923e-05,
      "loss": 0.824,
      "step": 2372
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.655370831489563,
      "learning_rate": 1.9859697071626818e-05,
      "loss": 0.8585,
      "step": 2373
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.914463996887207,
      "learning_rate": 1.9859510626779867e-05,
      "loss": 0.8305,
      "step": 2374
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9332361221313477,
      "learning_rate": 1.9859324059010692e-05,
      "loss": 0.6477,
      "step": 2375
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6311452388763428,
      "learning_rate": 1.9859137368321627e-05,
      "loss": 0.652,
      "step": 2376
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7411324977874756,
      "learning_rate": 1.9858950554714993e-05,
      "loss": 0.9782,
      "step": 2377
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.129160165786743,
      "learning_rate": 1.985876361819312e-05,
      "loss": 0.6503,
      "step": 2378
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8568105697631836,
      "learning_rate": 1.9858576558758346e-05,
      "loss": 0.7839,
      "step": 2379
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8271574974060059,
      "learning_rate": 1.9858389376412998e-05,
      "loss": 0.77,
      "step": 2380
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9060277938842773,
      "learning_rate": 1.9858202071159405e-05,
      "loss": 0.7011,
      "step": 2381
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6382715702056885,
      "learning_rate": 1.985801464299991e-05,
      "loss": 0.7867,
      "step": 2382
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8675321340560913,
      "learning_rate": 1.9857827091936843e-05,
      "loss": 0.8157,
      "step": 2383
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8346283435821533,
      "learning_rate": 1.985763941797255e-05,
      "loss": 0.7773,
      "step": 2384
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8737187385559082,
      "learning_rate": 1.9857451621109362e-05,
      "loss": 0.8881,
      "step": 2385
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0826447010040283,
      "learning_rate": 1.985726370134963e-05,
      "loss": 0.8305,
      "step": 2386
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6026432514190674,
      "learning_rate": 1.9857075658695685e-05,
      "loss": 0.8517,
      "step": 2387
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9957036972045898,
      "learning_rate": 1.985688749314988e-05,
      "loss": 0.92,
      "step": 2388
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7207789421081543,
      "learning_rate": 1.985669920471456e-05,
      "loss": 0.8253,
      "step": 2389
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9238746166229248,
      "learning_rate": 1.9856510793392075e-05,
      "loss": 0.7582,
      "step": 2390
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1895174980163574,
      "learning_rate": 1.985632225918477e-05,
      "loss": 0.8517,
      "step": 2391
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0023903846740723,
      "learning_rate": 1.985613360209499e-05,
      "loss": 0.8991,
      "step": 2392
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.4256670475006104,
      "learning_rate": 1.9855944822125098e-05,
      "loss": 0.7989,
      "step": 2393
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.862402319908142,
      "learning_rate": 1.985575591927744e-05,
      "loss": 0.7615,
      "step": 2394
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7078310251235962,
      "learning_rate": 1.9855566893554376e-05,
      "loss": 0.8893,
      "step": 2395
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5537142753601074,
      "learning_rate": 1.9855377744958257e-05,
      "loss": 0.7042,
      "step": 2396
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9947959184646606,
      "learning_rate": 1.9855188473491448e-05,
      "loss": 0.8646,
      "step": 2397
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1643483638763428,
      "learning_rate": 1.9854999079156302e-05,
      "loss": 0.7171,
      "step": 2398
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1997036933898926,
      "learning_rate": 1.985480956195519e-05,
      "loss": 0.7255,
      "step": 2399
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.974021077156067,
      "learning_rate": 1.985461992189046e-05,
      "loss": 0.7854,
      "step": 2400
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8185293674468994,
      "learning_rate": 1.985443015896449e-05,
      "loss": 0.8189,
      "step": 2401
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6940891742706299,
      "learning_rate": 1.9854240273179638e-05,
      "loss": 0.6577,
      "step": 2402
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7421940565109253,
      "learning_rate": 1.9854050264538273e-05,
      "loss": 0.7078,
      "step": 2403
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.000591278076172,
      "learning_rate": 1.9853860133042767e-05,
      "loss": 0.8925,
      "step": 2404
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2558202743530273,
      "learning_rate": 1.985366987869549e-05,
      "loss": 0.6347,
      "step": 2405
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9558202028274536,
      "learning_rate": 1.985347950149881e-05,
      "loss": 0.68,
      "step": 2406
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9010872840881348,
      "learning_rate": 1.9853289001455103e-05,
      "loss": 0.9486,
      "step": 2407
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.842438817024231,
      "learning_rate": 1.985309837856674e-05,
      "loss": 0.7585,
      "step": 2408
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8663747310638428,
      "learning_rate": 1.9852907632836106e-05,
      "loss": 0.9265,
      "step": 2409
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9628825187683105,
      "learning_rate": 1.9852716764265574e-05,
      "loss": 0.9454,
      "step": 2410
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.678647756576538,
      "learning_rate": 1.9852525772857522e-05,
      "loss": 0.6921,
      "step": 2411
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7778215408325195,
      "learning_rate": 1.9852334658614337e-05,
      "loss": 0.9,
      "step": 2412
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5723459720611572,
      "learning_rate": 1.985214342153839e-05,
      "loss": 0.7663,
      "step": 2413
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.64043390750885,
      "learning_rate": 1.985195206163208e-05,
      "loss": 0.8399,
      "step": 2414
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9550377130508423,
      "learning_rate": 1.9851760578897786e-05,
      "loss": 0.8498,
      "step": 2415
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7523020505905151,
      "learning_rate": 1.9851568973337893e-05,
      "loss": 0.9421,
      "step": 2416
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9789260625839233,
      "learning_rate": 1.985137724495479e-05,
      "loss": 0.8361,
      "step": 2417
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5871586799621582,
      "learning_rate": 1.9851185393750875e-05,
      "loss": 0.8273,
      "step": 2418
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8951326608657837,
      "learning_rate": 1.985099341972853e-05,
      "loss": 0.908,
      "step": 2419
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5451198816299438,
      "learning_rate": 1.9850801322890155e-05,
      "loss": 0.7215,
      "step": 2420
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7468775510787964,
      "learning_rate": 1.9850609103238138e-05,
      "loss": 0.7623,
      "step": 2421
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5993545055389404,
      "learning_rate": 1.9850416760774882e-05,
      "loss": 0.7516,
      "step": 2422
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9216400384902954,
      "learning_rate": 1.9850224295502786e-05,
      "loss": 0.8092,
      "step": 2423
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7725472450256348,
      "learning_rate": 1.985003170742424e-05,
      "loss": 0.8266,
      "step": 2424
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.777413010597229,
      "learning_rate": 1.984983899654166e-05,
      "loss": 0.8389,
      "step": 2425
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.2250421047210693,
      "learning_rate": 1.9849646162857435e-05,
      "loss": 0.8738,
      "step": 2426
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1739540100097656,
      "learning_rate": 1.9849453206373975e-05,
      "loss": 0.8669,
      "step": 2427
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5263978242874146,
      "learning_rate": 1.9849260127093686e-05,
      "loss": 0.7233,
      "step": 2428
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9901186227798462,
      "learning_rate": 1.9849066925018972e-05,
      "loss": 0.9062,
      "step": 2429
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6900137662887573,
      "learning_rate": 1.984887360015225e-05,
      "loss": 0.7634,
      "step": 2430
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9035093784332275,
      "learning_rate": 1.984868015249592e-05,
      "loss": 0.7765,
      "step": 2431
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0569427013397217,
      "learning_rate": 1.9848486582052398e-05,
      "loss": 0.7302,
      "step": 2432
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8775012493133545,
      "learning_rate": 1.9848292888824095e-05,
      "loss": 0.9006,
      "step": 2433
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.892627239227295,
      "learning_rate": 1.984809907281343e-05,
      "loss": 0.8335,
      "step": 2434
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0438616275787354,
      "learning_rate": 1.984790513402282e-05,
      "loss": 0.7478,
      "step": 2435
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.087486743927002,
      "learning_rate": 1.9847711072454676e-05,
      "loss": 0.8516,
      "step": 2436
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6867858171463013,
      "learning_rate": 1.9847516888111426e-05,
      "loss": 0.8474,
      "step": 2437
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8634113073349,
      "learning_rate": 1.9847322580995486e-05,
      "loss": 0.9353,
      "step": 2438
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9192969799041748,
      "learning_rate": 1.9847128151109277e-05,
      "loss": 0.8242,
      "step": 2439
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7930129766464233,
      "learning_rate": 1.9846933598455227e-05,
      "loss": 0.7641,
      "step": 2440
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9779428243637085,
      "learning_rate": 1.9846738923035758e-05,
      "loss": 1.0519,
      "step": 2441
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.719486951828003,
      "learning_rate": 1.98465441248533e-05,
      "loss": 0.8255,
      "step": 2442
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7027236223220825,
      "learning_rate": 1.984634920391028e-05,
      "loss": 0.9183,
      "step": 2443
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.649320125579834,
      "learning_rate": 1.9846154160209134e-05,
      "loss": 0.7664,
      "step": 2444
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9245389699935913,
      "learning_rate": 1.9845958993752282e-05,
      "loss": 0.9314,
      "step": 2445
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8183033466339111,
      "learning_rate": 1.984576370454217e-05,
      "loss": 0.7139,
      "step": 2446
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1614153385162354,
      "learning_rate": 1.984556829258122e-05,
      "loss": 0.9348,
      "step": 2447
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8913174867630005,
      "learning_rate": 1.984537275787188e-05,
      "loss": 0.8347,
      "step": 2448
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9237751960754395,
      "learning_rate": 1.9845177100416577e-05,
      "loss": 0.6874,
      "step": 2449
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7250696420669556,
      "learning_rate": 1.984498132021776e-05,
      "loss": 0.8806,
      "step": 2450
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1576688289642334,
      "learning_rate": 1.984478541727786e-05,
      "loss": 0.7985,
      "step": 2451
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6022447347640991,
      "learning_rate": 1.984458939159933e-05,
      "loss": 0.715,
      "step": 2452
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0329928398132324,
      "learning_rate": 1.9844393243184613e-05,
      "loss": 0.7651,
      "step": 2453
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1073572635650635,
      "learning_rate": 1.9844196972036146e-05,
      "loss": 0.9094,
      "step": 2454
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.698811411857605,
      "learning_rate": 1.9844000578156377e-05,
      "loss": 0.7924,
      "step": 2455
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6806977987289429,
      "learning_rate": 1.984380406154776e-05,
      "loss": 0.6549,
      "step": 2456
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8722474575042725,
      "learning_rate": 1.9843607422212748e-05,
      "loss": 0.7121,
      "step": 2457
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9227807521820068,
      "learning_rate": 1.9843410660153784e-05,
      "loss": 0.6907,
      "step": 2458
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5018833875656128,
      "learning_rate": 1.9843213775373322e-05,
      "loss": 0.85,
      "step": 2459
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8382426500320435,
      "learning_rate": 1.9843016767873825e-05,
      "loss": 0.9501,
      "step": 2460
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.1978092193603516,
      "learning_rate": 1.984281963765774e-05,
      "loss": 0.7792,
      "step": 2461
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9971177577972412,
      "learning_rate": 1.9842622384727527e-05,
      "loss": 0.8905,
      "step": 2462
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.839276671409607,
      "learning_rate": 1.984242500908565e-05,
      "loss": 0.8796,
      "step": 2463
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.5323357582092285,
      "learning_rate": 1.9842227510734563e-05,
      "loss": 1.0819,
      "step": 2464
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.884098768234253,
      "learning_rate": 1.9842029889676736e-05,
      "loss": 0.872,
      "step": 2465
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2085342407226562,
      "learning_rate": 1.9841832145914623e-05,
      "loss": 0.5985,
      "step": 2466
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9663361310958862,
      "learning_rate": 1.9841634279450697e-05,
      "loss": 0.7032,
      "step": 2467
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2042713165283203,
      "learning_rate": 1.984143629028742e-05,
      "loss": 0.895,
      "step": 2468
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1198344230651855,
      "learning_rate": 1.9841238178427266e-05,
      "loss": 0.8177,
      "step": 2469
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7317477464675903,
      "learning_rate": 1.98410399438727e-05,
      "loss": 0.792,
      "step": 2470
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2637383937835693,
      "learning_rate": 1.9840841586626197e-05,
      "loss": 0.8295,
      "step": 2471
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6308695077896118,
      "learning_rate": 1.984064310669023e-05,
      "loss": 0.8195,
      "step": 2472
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0447747707366943,
      "learning_rate": 1.9840444504067268e-05,
      "loss": 0.8401,
      "step": 2473
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9283910989761353,
      "learning_rate": 1.9840245778759795e-05,
      "loss": 0.9936,
      "step": 2474
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8357629776000977,
      "learning_rate": 1.984004693077028e-05,
      "loss": 0.7647,
      "step": 2475
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.719896674156189,
      "learning_rate": 1.9839847960101208e-05,
      "loss": 0.7603,
      "step": 2476
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1539201736450195,
      "learning_rate": 1.983964886675506e-05,
      "loss": 0.8737,
      "step": 2477
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.923586130142212,
      "learning_rate": 1.9839449650734315e-05,
      "loss": 0.6618,
      "step": 2478
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6788878440856934,
      "learning_rate": 1.9839250312041457e-05,
      "loss": 0.6152,
      "step": 2479
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7419184446334839,
      "learning_rate": 1.9839050850678975e-05,
      "loss": 0.8563,
      "step": 2480
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9076497554779053,
      "learning_rate": 1.983885126664935e-05,
      "loss": 0.7731,
      "step": 2481
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9352165460586548,
      "learning_rate": 1.9838651559955077e-05,
      "loss": 0.923,
      "step": 2482
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8509275913238525,
      "learning_rate": 1.983845173059864e-05,
      "loss": 0.6312,
      "step": 2483
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5597715377807617,
      "learning_rate": 1.983825177858253e-05,
      "loss": 0.881,
      "step": 2484
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0890731811523438,
      "learning_rate": 1.9838051703909246e-05,
      "loss": 0.8132,
      "step": 2485
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.305412769317627,
      "learning_rate": 1.983785150658128e-05,
      "loss": 0.6474,
      "step": 2486
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1215784549713135,
      "learning_rate": 1.9837651186601125e-05,
      "loss": 0.8384,
      "step": 2487
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.020650863647461,
      "learning_rate": 1.983745074397128e-05,
      "loss": 0.9051,
      "step": 2488
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9235976934432983,
      "learning_rate": 1.9837250178694242e-05,
      "loss": 0.9865,
      "step": 2489
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.70766282081604,
      "learning_rate": 1.9837049490772517e-05,
      "loss": 0.7408,
      "step": 2490
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.306028127670288,
      "learning_rate": 1.9836848680208603e-05,
      "loss": 0.7256,
      "step": 2491
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9541860818862915,
      "learning_rate": 1.9836647747005005e-05,
      "loss": 0.7882,
      "step": 2492
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6825820207595825,
      "learning_rate": 1.9836446691164226e-05,
      "loss": 0.7067,
      "step": 2493
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.108759641647339,
      "learning_rate": 1.9836245512688773e-05,
      "loss": 0.817,
      "step": 2494
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7847176790237427,
      "learning_rate": 1.9836044211581158e-05,
      "loss": 0.9042,
      "step": 2495
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.867513656616211,
      "learning_rate": 1.9835842787843886e-05,
      "loss": 0.7862,
      "step": 2496
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.115872859954834,
      "learning_rate": 1.9835641241479473e-05,
      "loss": 0.7449,
      "step": 2497
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7419664859771729,
      "learning_rate": 1.9835439572490426e-05,
      "loss": 0.7861,
      "step": 2498
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6796705722808838,
      "learning_rate": 1.9835237780879265e-05,
      "loss": 0.8245,
      "step": 2499
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.107060194015503,
      "learning_rate": 1.98350358666485e-05,
      "loss": 0.8919,
      "step": 2500
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.13346529006958,
      "learning_rate": 1.9834833829800654e-05,
      "loss": 1.1028,
      "step": 2501
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.877855896949768,
      "learning_rate": 1.9834631670338242e-05,
      "loss": 0.79,
      "step": 2502
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.039473533630371,
      "learning_rate": 1.9834429388263786e-05,
      "loss": 0.7911,
      "step": 2503
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.602362036705017,
      "learning_rate": 1.983422698357981e-05,
      "loss": 0.8438,
      "step": 2504
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9906455278396606,
      "learning_rate": 1.983402445628883e-05,
      "loss": 0.8218,
      "step": 2505
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7668516635894775,
      "learning_rate": 1.983382180639338e-05,
      "loss": 0.7659,
      "step": 2506
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.903620719909668,
      "learning_rate": 1.983361903389598e-05,
      "loss": 0.7388,
      "step": 2507
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.662479043006897,
      "learning_rate": 1.983341613879916e-05,
      "loss": 0.8039,
      "step": 2508
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.4535105228424072,
      "learning_rate": 1.983321312110545e-05,
      "loss": 0.8344,
      "step": 2509
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7881102561950684,
      "learning_rate": 1.9833009980817386e-05,
      "loss": 0.7216,
      "step": 2510
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.328218698501587,
      "learning_rate": 1.9832806717937493e-05,
      "loss": 0.9571,
      "step": 2511
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.093726634979248,
      "learning_rate": 1.9832603332468308e-05,
      "loss": 0.8165,
      "step": 2512
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0118181705474854,
      "learning_rate": 1.9832399824412367e-05,
      "loss": 0.8297,
      "step": 2513
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6413323879241943,
      "learning_rate": 1.983219619377221e-05,
      "loss": 0.9243,
      "step": 2514
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6842800378799438,
      "learning_rate": 1.9831992440550368e-05,
      "loss": 0.8503,
      "step": 2515
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.216088056564331,
      "learning_rate": 1.9831788564749388e-05,
      "loss": 0.9146,
      "step": 2516
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.682508111000061,
      "learning_rate": 1.9831584566371807e-05,
      "loss": 0.8398,
      "step": 2517
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9343138933181763,
      "learning_rate": 1.9831380445420173e-05,
      "loss": 0.6298,
      "step": 2518
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7963446378707886,
      "learning_rate": 1.983117620189703e-05,
      "loss": 0.9885,
      "step": 2519
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8382189273834229,
      "learning_rate": 1.9830971835804924e-05,
      "loss": 0.8598,
      "step": 2520
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8346704244613647,
      "learning_rate": 1.98307673471464e-05,
      "loss": 0.6481,
      "step": 2521
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8107432126998901,
      "learning_rate": 1.983056273592401e-05,
      "loss": 0.7562,
      "step": 2522
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9216278791427612,
      "learning_rate": 1.9830358002140304e-05,
      "loss": 0.702,
      "step": 2523
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.715465784072876,
      "learning_rate": 1.9830153145797836e-05,
      "loss": 0.7437,
      "step": 2524
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7147674560546875,
      "learning_rate": 1.982994816689916e-05,
      "loss": 0.703,
      "step": 2525
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.530118465423584,
      "learning_rate": 1.9829743065446826e-05,
      "loss": 0.7622,
      "step": 2526
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.691676378250122,
      "learning_rate": 1.98295378414434e-05,
      "loss": 0.8701,
      "step": 2527
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.2912650108337402,
      "learning_rate": 1.9829332494891438e-05,
      "loss": 0.8236,
      "step": 2528
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5772113800048828,
      "learning_rate": 1.9829127025793495e-05,
      "loss": 0.9061,
      "step": 2529
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.685059905052185,
      "learning_rate": 1.9828921434152138e-05,
      "loss": 0.8175,
      "step": 2530
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7486543655395508,
      "learning_rate": 1.9828715719969928e-05,
      "loss": 0.8386,
      "step": 2531
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.127267837524414,
      "learning_rate": 1.982850988324943e-05,
      "loss": 0.8967,
      "step": 2532
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8199830055236816,
      "learning_rate": 1.982830392399321e-05,
      "loss": 0.8161,
      "step": 2533
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0095338821411133,
      "learning_rate": 1.9828097842203835e-05,
      "loss": 0.8182,
      "step": 2534
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.99245023727417,
      "learning_rate": 1.982789163788388e-05,
      "loss": 0.8644,
      "step": 2535
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0615668296813965,
      "learning_rate": 1.9827685311035905e-05,
      "loss": 0.8303,
      "step": 2536
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6790751218795776,
      "learning_rate": 1.982747886166249e-05,
      "loss": 0.8958,
      "step": 2537
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.635562539100647,
      "learning_rate": 1.982727228976621e-05,
      "loss": 0.7806,
      "step": 2538
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.0440430641174316,
      "learning_rate": 1.9827065595349636e-05,
      "loss": 1.0271,
      "step": 2539
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1479952335357666,
      "learning_rate": 1.9826858778415347e-05,
      "loss": 0.7194,
      "step": 2540
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.26395320892334,
      "learning_rate": 1.982665183896592e-05,
      "loss": 0.8159,
      "step": 2541
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9995287656784058,
      "learning_rate": 1.982644477700394e-05,
      "loss": 0.7473,
      "step": 2542
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7334598302841187,
      "learning_rate": 1.982623759253198e-05,
      "loss": 0.8752,
      "step": 2543
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8480331897735596,
      "learning_rate": 1.9826030285552633e-05,
      "loss": 0.7162,
      "step": 2544
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7977999448776245,
      "learning_rate": 1.9825822856068474e-05,
      "loss": 0.8603,
      "step": 2545
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7853946685791016,
      "learning_rate": 1.9825615304082097e-05,
      "loss": 0.6621,
      "step": 2546
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7232245206832886,
      "learning_rate": 1.9825407629596084e-05,
      "loss": 0.7746,
      "step": 2547
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.030799627304077,
      "learning_rate": 1.9825199832613026e-05,
      "loss": 0.7218,
      "step": 2548
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7261983156204224,
      "learning_rate": 1.9824991913135513e-05,
      "loss": 0.8779,
      "step": 2549
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5954563617706299,
      "learning_rate": 1.982478387116614e-05,
      "loss": 0.7566,
      "step": 2550
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8194153308868408,
      "learning_rate": 1.9824575706707497e-05,
      "loss": 0.7258,
      "step": 2551
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8983665704727173,
      "learning_rate": 1.9824367419762182e-05,
      "loss": 0.9971,
      "step": 2552
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.3370361328125,
      "learning_rate": 1.9824159010332793e-05,
      "loss": 0.8955,
      "step": 2553
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6769063472747803,
      "learning_rate": 1.9823950478421924e-05,
      "loss": 0.8019,
      "step": 2554
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9792817831039429,
      "learning_rate": 1.9823741824032178e-05,
      "loss": 0.944,
      "step": 2555
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.1873397827148438,
      "learning_rate": 1.9823533047166153e-05,
      "loss": 0.9657,
      "step": 2556
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.896888017654419,
      "learning_rate": 1.9823324147826456e-05,
      "loss": 0.7399,
      "step": 2557
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.117032766342163,
      "learning_rate": 1.982311512601569e-05,
      "loss": 0.898,
      "step": 2558
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6536613702774048,
      "learning_rate": 1.9822905981736465e-05,
      "loss": 0.8054,
      "step": 2559
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5802680253982544,
      "learning_rate": 1.9822696714991378e-05,
      "loss": 0.7757,
      "step": 2560
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9668300151824951,
      "learning_rate": 1.9822487325783047e-05,
      "loss": 0.7632,
      "step": 2561
    },
    {
      "epoch": 0.26,
      "grad_norm": 2.196606397628784,
      "learning_rate": 1.982227781411408e-05,
      "loss": 0.9252,
      "step": 2562
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.692564845085144,
      "learning_rate": 1.982206817998709e-05,
      "loss": 0.6999,
      "step": 2563
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.807031512260437,
      "learning_rate": 1.9821858423404685e-05,
      "loss": 0.8486,
      "step": 2564
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6349140405654907,
      "learning_rate": 1.982164854436949e-05,
      "loss": 0.8022,
      "step": 2565
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7351429462432861,
      "learning_rate": 1.982143854288411e-05,
      "loss": 0.7415,
      "step": 2566
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6653430461883545,
      "learning_rate": 1.9821228418951175e-05,
      "loss": 0.7631,
      "step": 2567
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6779799461364746,
      "learning_rate": 1.9821018172573298e-05,
      "loss": 0.738,
      "step": 2568
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9012587070465088,
      "learning_rate": 1.98208078037531e-05,
      "loss": 0.903,
      "step": 2569
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7321580648422241,
      "learning_rate": 1.9820597312493205e-05,
      "loss": 0.7555,
      "step": 2570
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5875684022903442,
      "learning_rate": 1.9820386698796238e-05,
      "loss": 0.8646,
      "step": 2571
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.12722110748291,
      "learning_rate": 1.9820175962664822e-05,
      "loss": 0.8347,
      "step": 2572
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5378538370132446,
      "learning_rate": 1.9819965104101587e-05,
      "loss": 0.8834,
      "step": 2573
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.832958698272705,
      "learning_rate": 1.981975412310916e-05,
      "loss": 0.872,
      "step": 2574
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.759641408920288,
      "learning_rate": 1.9819543019690175e-05,
      "loss": 0.9255,
      "step": 2575
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8654918670654297,
      "learning_rate": 1.9819331793847264e-05,
      "loss": 0.892,
      "step": 2576
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0722439289093018,
      "learning_rate": 1.9819120445583053e-05,
      "loss": 0.6714,
      "step": 2577
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1071627140045166,
      "learning_rate": 1.9818908974900183e-05,
      "loss": 1.0344,
      "step": 2578
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7685195207595825,
      "learning_rate": 1.9818697381801293e-05,
      "loss": 0.794,
      "step": 2579
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.664614200592041,
      "learning_rate": 1.9818485666289017e-05,
      "loss": 0.5907,
      "step": 2580
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.649149775505066,
      "learning_rate": 1.981827382836599e-05,
      "loss": 0.8191,
      "step": 2581
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8312238454818726,
      "learning_rate": 1.981806186803486e-05,
      "loss": 0.7245,
      "step": 2582
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7742304801940918,
      "learning_rate": 1.9817849785298268e-05,
      "loss": 0.8046,
      "step": 2583
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.745552897453308,
      "learning_rate": 1.981763758015886e-05,
      "loss": 0.795,
      "step": 2584
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7961889505386353,
      "learning_rate": 1.981742525261928e-05,
      "loss": 0.8706,
      "step": 2585
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.69040846824646,
      "learning_rate": 1.981721280268217e-05,
      "loss": 0.5873,
      "step": 2586
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8996422290802002,
      "learning_rate": 1.9817000230350187e-05,
      "loss": 0.8407,
      "step": 2587
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.49149227142334,
      "learning_rate": 1.9816787535625974e-05,
      "loss": 0.887,
      "step": 2588
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.911877989768982,
      "learning_rate": 1.981657471851219e-05,
      "loss": 0.7007,
      "step": 2589
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.036271095275879,
      "learning_rate": 1.981636177901148e-05,
      "loss": 0.8757,
      "step": 2590
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.820878267288208,
      "learning_rate": 1.9816148717126506e-05,
      "loss": 0.8723,
      "step": 2591
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.835601568222046,
      "learning_rate": 1.9815935532859923e-05,
      "loss": 0.7876,
      "step": 2592
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8570889234542847,
      "learning_rate": 1.9815722226214383e-05,
      "loss": 0.8438,
      "step": 2593
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6046518087387085,
      "learning_rate": 1.981550879719255e-05,
      "loss": 0.8013,
      "step": 2594
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8703784942626953,
      "learning_rate": 1.9815295245797092e-05,
      "loss": 0.7576,
      "step": 2595
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8200217485427856,
      "learning_rate": 1.9815081572030657e-05,
      "loss": 0.8584,
      "step": 2596
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6148604154586792,
      "learning_rate": 1.981486777589592e-05,
      "loss": 0.891,
      "step": 2597
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4962860345840454,
      "learning_rate": 1.981465385739554e-05,
      "loss": 0.7994,
      "step": 2598
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9407011270523071,
      "learning_rate": 1.981443981653219e-05,
      "loss": 0.773,
      "step": 2599
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6066639423370361,
      "learning_rate": 1.9814225653308532e-05,
      "loss": 0.756,
      "step": 2600
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.533266305923462,
      "learning_rate": 1.981401136772724e-05,
      "loss": 0.7454,
      "step": 2601
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.57278311252594,
      "learning_rate": 1.9813796959790985e-05,
      "loss": 0.9,
      "step": 2602
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.064927577972412,
      "learning_rate": 1.981358242950244e-05,
      "loss": 0.9093,
      "step": 2603
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6280572414398193,
      "learning_rate": 1.981336777686428e-05,
      "loss": 0.701,
      "step": 2604
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6053566932678223,
      "learning_rate": 1.9813153001879177e-05,
      "loss": 0.7312,
      "step": 2605
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.890709638595581,
      "learning_rate": 1.9812938104549814e-05,
      "loss": 0.7588,
      "step": 2606
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.260756254196167,
      "learning_rate": 1.981272308487887e-05,
      "loss": 0.8871,
      "step": 2607
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8012686967849731,
      "learning_rate": 1.9812507942869023e-05,
      "loss": 0.815,
      "step": 2608
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8428763151168823,
      "learning_rate": 1.981229267852296e-05,
      "loss": 0.8122,
      "step": 2609
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8155173063278198,
      "learning_rate": 1.9812077291843355e-05,
      "loss": 0.931,
      "step": 2610
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.0955088138580322,
      "learning_rate": 1.9811861782832904e-05,
      "loss": 0.9845,
      "step": 2611
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8207682371139526,
      "learning_rate": 1.981164615149429e-05,
      "loss": 0.8899,
      "step": 2612
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6302103996276855,
      "learning_rate": 1.98114303978302e-05,
      "loss": 0.7001,
      "step": 2613
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7934517860412598,
      "learning_rate": 1.9811214521843324e-05,
      "loss": 0.7545,
      "step": 2614
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9377553462982178,
      "learning_rate": 1.9810998523536352e-05,
      "loss": 0.8688,
      "step": 2615
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.127148151397705,
      "learning_rate": 1.9810782402911978e-05,
      "loss": 1.0011,
      "step": 2616
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.93989896774292,
      "learning_rate": 1.9810566159972902e-05,
      "loss": 0.6646,
      "step": 2617
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6061660051345825,
      "learning_rate": 1.9810349794721815e-05,
      "loss": 0.7617,
      "step": 2618
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8276722431182861,
      "learning_rate": 1.9810133307161413e-05,
      "loss": 0.9011,
      "step": 2619
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7520960569381714,
      "learning_rate": 1.9809916697294397e-05,
      "loss": 0.8677,
      "step": 2620
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5886659622192383,
      "learning_rate": 1.9809699965123467e-05,
      "loss": 0.7173,
      "step": 2621
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7439414262771606,
      "learning_rate": 1.9809483110651327e-05,
      "loss": 0.8717,
      "step": 2622
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4324432611465454,
      "learning_rate": 1.9809266133880676e-05,
      "loss": 0.5652,
      "step": 2623
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7034616470336914,
      "learning_rate": 1.980904903481423e-05,
      "loss": 0.9107,
      "step": 2624
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4024039506912231,
      "learning_rate": 1.980883181345468e-05,
      "loss": 0.7307,
      "step": 2625
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9667843580245972,
      "learning_rate": 1.9808614469804744e-05,
      "loss": 0.8708,
      "step": 2626
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.578162431716919,
      "learning_rate": 1.9808397003867132e-05,
      "loss": 0.8504,
      "step": 2627
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6273629665374756,
      "learning_rate": 1.9808179415644556e-05,
      "loss": 0.7263,
      "step": 2628
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8570873737335205,
      "learning_rate": 1.980796170513972e-05,
      "loss": 0.8427,
      "step": 2629
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6732786893844604,
      "learning_rate": 1.9807743872355347e-05,
      "loss": 0.7116,
      "step": 2630
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8901524543762207,
      "learning_rate": 1.980752591729415e-05,
      "loss": 0.9636,
      "step": 2631
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.039903402328491,
      "learning_rate": 1.980730783995884e-05,
      "loss": 0.8014,
      "step": 2632
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1028101444244385,
      "learning_rate": 1.9807089640352153e-05,
      "loss": 0.8859,
      "step": 2633
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7037286758422852,
      "learning_rate": 1.980687131847679e-05,
      "loss": 0.9381,
      "step": 2634
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.823722243309021,
      "learning_rate": 1.9806652874335486e-05,
      "loss": 0.8611,
      "step": 2635
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7718782424926758,
      "learning_rate": 1.9806434307930958e-05,
      "loss": 0.9679,
      "step": 2636
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.665643572807312,
      "learning_rate": 1.9806215619265935e-05,
      "loss": 0.7636,
      "step": 2637
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.893896222114563,
      "learning_rate": 1.9805996808343137e-05,
      "loss": 0.77,
      "step": 2638
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7715051174163818,
      "learning_rate": 1.98057778751653e-05,
      "loss": 0.7956,
      "step": 2639
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7131218910217285,
      "learning_rate": 1.9805558819735146e-05,
      "loss": 0.8097,
      "step": 2640
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8178075551986694,
      "learning_rate": 1.9805339642055414e-05,
      "loss": 0.8021,
      "step": 2641
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.590823769569397,
      "learning_rate": 1.980512034212883e-05,
      "loss": 0.9086,
      "step": 2642
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.899208664894104,
      "learning_rate": 1.980490091995813e-05,
      "loss": 0.851,
      "step": 2643
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9860678911209106,
      "learning_rate": 1.9804681375546052e-05,
      "loss": 0.8922,
      "step": 2644
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9016557931900024,
      "learning_rate": 1.980446170889533e-05,
      "loss": 0.7244,
      "step": 2645
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.882939100265503,
      "learning_rate": 1.9804241920008706e-05,
      "loss": 0.7729,
      "step": 2646
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.041562795639038,
      "learning_rate": 1.9804022008888915e-05,
      "loss": 0.7973,
      "step": 2647
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7700202465057373,
      "learning_rate": 1.9803801975538702e-05,
      "loss": 0.8295,
      "step": 2648
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2425522804260254,
      "learning_rate": 1.9803581819960815e-05,
      "loss": 0.766,
      "step": 2649
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.931870937347412,
      "learning_rate": 1.9803361542157987e-05,
      "loss": 0.7628,
      "step": 2650
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.1471097469329834,
      "learning_rate": 1.9803141142132975e-05,
      "loss": 0.6388,
      "step": 2651
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8605766296386719,
      "learning_rate": 1.980292061988852e-05,
      "loss": 0.7991,
      "step": 2652
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.2796270847320557,
      "learning_rate": 1.9802699975427377e-05,
      "loss": 0.759,
      "step": 2653
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9098293781280518,
      "learning_rate": 1.9802479208752295e-05,
      "loss": 0.9868,
      "step": 2654
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7692694664001465,
      "learning_rate": 1.9802258319866025e-05,
      "loss": 0.7461,
      "step": 2655
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5746451616287231,
      "learning_rate": 1.9802037308771318e-05,
      "loss": 0.7431,
      "step": 2656
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.703604817390442,
      "learning_rate": 1.9801816175470937e-05,
      "loss": 0.8353,
      "step": 2657
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8934389352798462,
      "learning_rate": 1.980159491996763e-05,
      "loss": 0.7646,
      "step": 2658
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6742202043533325,
      "learning_rate": 1.9801373542264163e-05,
      "loss": 0.8421,
      "step": 2659
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1219048500061035,
      "learning_rate": 1.9801152042363294e-05,
      "loss": 0.82,
      "step": 2660
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.062049388885498,
      "learning_rate": 1.9800930420267785e-05,
      "loss": 0.9766,
      "step": 2661
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.77937912940979,
      "learning_rate": 1.9800708675980393e-05,
      "loss": 0.8308,
      "step": 2662
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.235031843185425,
      "learning_rate": 1.9800486809503892e-05,
      "loss": 0.9902,
      "step": 2663
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.545465111732483,
      "learning_rate": 1.980026482084104e-05,
      "loss": 0.754,
      "step": 2664
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8988864421844482,
      "learning_rate": 1.9800042709994612e-05,
      "loss": 0.8482,
      "step": 2665
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5189872980117798,
      "learning_rate": 1.979982047696737e-05,
      "loss": 0.7536,
      "step": 2666
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8116356134414673,
      "learning_rate": 1.9799598121762088e-05,
      "loss": 0.901,
      "step": 2667
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.848343849182129,
      "learning_rate": 1.9799375644381538e-05,
      "loss": 0.676,
      "step": 2668
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.876780390739441,
      "learning_rate": 1.9799153044828497e-05,
      "loss": 0.8242,
      "step": 2669
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.659354329109192,
      "learning_rate": 1.9798930323105735e-05,
      "loss": 0.6978,
      "step": 2670
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.637986183166504,
      "learning_rate": 1.9798707479216032e-05,
      "loss": 0.8978,
      "step": 2671
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7819342613220215,
      "learning_rate": 1.979848451316216e-05,
      "loss": 0.8261,
      "step": 2672
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.978220820426941,
      "learning_rate": 1.979826142494691e-05,
      "loss": 0.7747,
      "step": 2673
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.511189341545105,
      "learning_rate": 1.9798038214573056e-05,
      "loss": 0.6089,
      "step": 2674
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9749114513397217,
      "learning_rate": 1.9797814882043377e-05,
      "loss": 0.639,
      "step": 2675
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6242003440856934,
      "learning_rate": 1.979759142736067e-05,
      "loss": 0.7834,
      "step": 2676
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.851699709892273,
      "learning_rate": 1.9797367850527705e-05,
      "loss": 0.9331,
      "step": 2677
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7752594947814941,
      "learning_rate": 1.9797144151547284e-05,
      "loss": 0.7118,
      "step": 2678
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.719692349433899,
      "learning_rate": 1.9796920330422184e-05,
      "loss": 1.0112,
      "step": 2679
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7845838069915771,
      "learning_rate": 1.9796696387155205e-05,
      "loss": 0.7863,
      "step": 2680
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.708499789237976,
      "learning_rate": 1.9796472321749137e-05,
      "loss": 0.6594,
      "step": 2681
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4757016897201538,
      "learning_rate": 1.979624813420677e-05,
      "loss": 0.71,
      "step": 2682
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8874237537384033,
      "learning_rate": 1.9796023824530898e-05,
      "loss": 0.8126,
      "step": 2683
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.4877564907073975,
      "learning_rate": 1.979579939272432e-05,
      "loss": 0.9156,
      "step": 2684
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.066790819168091,
      "learning_rate": 1.979557483878984e-05,
      "loss": 0.7513,
      "step": 2685
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9654160737991333,
      "learning_rate": 1.9795350162730245e-05,
      "loss": 0.8341,
      "step": 2686
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.864690899848938,
      "learning_rate": 1.979512536454835e-05,
      "loss": 0.7715,
      "step": 2687
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.526118516921997,
      "learning_rate": 1.9794900444246943e-05,
      "loss": 0.6827,
      "step": 2688
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0352561473846436,
      "learning_rate": 1.9794675401828842e-05,
      "loss": 0.8276,
      "step": 2689
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1898744106292725,
      "learning_rate": 1.9794450237296844e-05,
      "loss": 0.8669,
      "step": 2690
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7352526187896729,
      "learning_rate": 1.979422495065376e-05,
      "loss": 0.7967,
      "step": 2691
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0895168781280518,
      "learning_rate": 1.9793999541902397e-05,
      "loss": 0.7574,
      "step": 2692
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7895840406417847,
      "learning_rate": 1.9793774011045563e-05,
      "loss": 0.8077,
      "step": 2693
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8708689212799072,
      "learning_rate": 1.9793548358086073e-05,
      "loss": 0.7737,
      "step": 2694
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6496076583862305,
      "learning_rate": 1.9793322583026743e-05,
      "loss": 0.9476,
      "step": 2695
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0751307010650635,
      "learning_rate": 1.979309668587038e-05,
      "loss": 0.8012,
      "step": 2696
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6554968357086182,
      "learning_rate": 1.979287066661981e-05,
      "loss": 0.7544,
      "step": 2697
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9688612222671509,
      "learning_rate": 1.9792644525277844e-05,
      "loss": 0.7641,
      "step": 2698
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6030980348587036,
      "learning_rate": 1.9792418261847303e-05,
      "loss": 0.8084,
      "step": 2699
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.036508560180664,
      "learning_rate": 1.9792191876331006e-05,
      "loss": 0.8479,
      "step": 2700
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8242896795272827,
      "learning_rate": 1.979196536873178e-05,
      "loss": 0.7956,
      "step": 2701
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9831433296203613,
      "learning_rate": 1.9791738739052445e-05,
      "loss": 0.7573,
      "step": 2702
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.072064161300659,
      "learning_rate": 1.979151198729583e-05,
      "loss": 0.8627,
      "step": 2703
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.2341952323913574,
      "learning_rate": 1.9791285113464756e-05,
      "loss": 0.9113,
      "step": 2704
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8744990825653076,
      "learning_rate": 1.979105811756206e-05,
      "loss": 0.7706,
      "step": 2705
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5086934566497803,
      "learning_rate": 1.9790830999590563e-05,
      "loss": 1.0017,
      "step": 2706
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.697980523109436,
      "learning_rate": 1.9790603759553104e-05,
      "loss": 0.681,
      "step": 2707
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.87692391872406,
      "learning_rate": 1.979037639745251e-05,
      "loss": 0.8454,
      "step": 2708
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8222618103027344,
      "learning_rate": 1.9790148913291625e-05,
      "loss": 0.7527,
      "step": 2709
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7259461879730225,
      "learning_rate": 1.9789921307073275e-05,
      "loss": 0.8015,
      "step": 2710
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.735503077507019,
      "learning_rate": 1.97896935788003e-05,
      "loss": 0.7849,
      "step": 2711
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6723867654800415,
      "learning_rate": 1.9789465728475543e-05,
      "loss": 0.9461,
      "step": 2712
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.701989769935608,
      "learning_rate": 1.978923775610184e-05,
      "loss": 0.8172,
      "step": 2713
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8173327445983887,
      "learning_rate": 1.978900966168204e-05,
      "loss": 0.8912,
      "step": 2714
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8692137002944946,
      "learning_rate": 1.978878144521898e-05,
      "loss": 0.802,
      "step": 2715
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8463890552520752,
      "learning_rate": 1.9788553106715506e-05,
      "loss": 0.8577,
      "step": 2716
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5837429761886597,
      "learning_rate": 1.978832464617447e-05,
      "loss": 0.6852,
      "step": 2717
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.855353593826294,
      "learning_rate": 1.9788096063598714e-05,
      "loss": 0.6626,
      "step": 2718
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8566523790359497,
      "learning_rate": 1.9787867358991093e-05,
      "loss": 0.5853,
      "step": 2719
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1246109008789062,
      "learning_rate": 1.9787638532354455e-05,
      "loss": 0.8408,
      "step": 2720
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7945002317428589,
      "learning_rate": 1.9787409583691652e-05,
      "loss": 0.7857,
      "step": 2721
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7893346548080444,
      "learning_rate": 1.978718051300554e-05,
      "loss": 0.7916,
      "step": 2722
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.814289927482605,
      "learning_rate": 1.9786951320298975e-05,
      "loss": 0.7447,
      "step": 2723
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.835131287574768,
      "learning_rate": 1.9786722005574817e-05,
      "loss": 0.6916,
      "step": 2724
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.015345811843872,
      "learning_rate": 1.9786492568835922e-05,
      "loss": 0.8741,
      "step": 2725
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7174760103225708,
      "learning_rate": 1.978626301008515e-05,
      "loss": 0.7793,
      "step": 2726
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6052125692367554,
      "learning_rate": 1.9786033329325363e-05,
      "loss": 0.6556,
      "step": 2727
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8411307334899902,
      "learning_rate": 1.9785803526559425e-05,
      "loss": 0.765,
      "step": 2728
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5201507806777954,
      "learning_rate": 1.9785573601790206e-05,
      "loss": 0.7803,
      "step": 2729
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7298637628555298,
      "learning_rate": 1.978534355502056e-05,
      "loss": 0.8297,
      "step": 2730
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.008899450302124,
      "learning_rate": 1.978511338625337e-05,
      "loss": 0.7835,
      "step": 2731
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7006717920303345,
      "learning_rate": 1.9784883095491497e-05,
      "loss": 0.7321,
      "step": 2732
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7470529079437256,
      "learning_rate": 1.9784652682737815e-05,
      "loss": 0.7667,
      "step": 2733
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9463635683059692,
      "learning_rate": 1.9784422147995192e-05,
      "loss": 0.8343,
      "step": 2734
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8413500785827637,
      "learning_rate": 1.978419149126651e-05,
      "loss": 1.0254,
      "step": 2735
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.646999478340149,
      "learning_rate": 1.9783960712554635e-05,
      "loss": 0.8096,
      "step": 2736
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7521388530731201,
      "learning_rate": 1.9783729811862456e-05,
      "loss": 0.9087,
      "step": 2737
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7066292762756348,
      "learning_rate": 1.9783498789192837e-05,
      "loss": 0.8252,
      "step": 2738
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0769965648651123,
      "learning_rate": 1.9783267644548675e-05,
      "loss": 0.9968,
      "step": 2739
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.5848817825317383,
      "learning_rate": 1.978303637793284e-05,
      "loss": 0.9462,
      "step": 2740
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6868222951889038,
      "learning_rate": 1.9782804989348215e-05,
      "loss": 0.7689,
      "step": 2741
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.850097417831421,
      "learning_rate": 1.978257347879769e-05,
      "loss": 0.9045,
      "step": 2742
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9537937641143799,
      "learning_rate": 1.9782341846284154e-05,
      "loss": 0.7587,
      "step": 2743
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.642008900642395,
      "learning_rate": 1.9782110091810486e-05,
      "loss": 0.8719,
      "step": 2744
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.80238938331604,
      "learning_rate": 1.9781878215379585e-05,
      "loss": 0.7582,
      "step": 2745
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1554582118988037,
      "learning_rate": 1.978164621699433e-05,
      "loss": 0.7547,
      "step": 2746
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.665912389755249,
      "learning_rate": 1.978141409665762e-05,
      "loss": 0.9308,
      "step": 2747
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0956342220306396,
      "learning_rate": 1.9781181854372357e-05,
      "loss": 0.7986,
      "step": 2748
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0221810340881348,
      "learning_rate": 1.9780949490141422e-05,
      "loss": 0.8339,
      "step": 2749
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9015142917633057,
      "learning_rate": 1.978071700396772e-05,
      "loss": 0.9595,
      "step": 2750
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6732630729675293,
      "learning_rate": 1.978048439585415e-05,
      "loss": 0.9609,
      "step": 2751
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9002783298492432,
      "learning_rate": 1.9780251665803608e-05,
      "loss": 0.7961,
      "step": 2752
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6800936460494995,
      "learning_rate": 1.9780018813819e-05,
      "loss": 0.8185,
      "step": 2753
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.947963833808899,
      "learning_rate": 1.9779785839903226e-05,
      "loss": 0.7248,
      "step": 2754
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.650031328201294,
      "learning_rate": 1.977955274405919e-05,
      "loss": 0.6338,
      "step": 2755
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.100184202194214,
      "learning_rate": 1.97793195262898e-05,
      "loss": 0.8137,
      "step": 2756
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9644230604171753,
      "learning_rate": 1.9779086186597963e-05,
      "loss": 0.6854,
      "step": 2757
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.028930425643921,
      "learning_rate": 1.9778852724986586e-05,
      "loss": 0.7364,
      "step": 2758
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5916411876678467,
      "learning_rate": 1.9778619141458588e-05,
      "loss": 0.749,
      "step": 2759
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9641008377075195,
      "learning_rate": 1.9778385436016866e-05,
      "loss": 0.6593,
      "step": 2760
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7668735980987549,
      "learning_rate": 1.9778151608664346e-05,
      "loss": 0.9385,
      "step": 2761
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7488771677017212,
      "learning_rate": 1.9777917659403944e-05,
      "loss": 0.8815,
      "step": 2762
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.026235818862915,
      "learning_rate": 1.9777683588238567e-05,
      "loss": 0.6866,
      "step": 2763
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.895424485206604,
      "learning_rate": 1.977744939517114e-05,
      "loss": 0.8098,
      "step": 2764
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.831160068511963,
      "learning_rate": 1.9777215080204583e-05,
      "loss": 0.7743,
      "step": 2765
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.05118465423584,
      "learning_rate": 1.9776980643341818e-05,
      "loss": 0.8438,
      "step": 2766
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7593315839767456,
      "learning_rate": 1.977674608458576e-05,
      "loss": 0.8387,
      "step": 2767
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6723461151123047,
      "learning_rate": 1.9776511403939343e-05,
      "loss": 0.703,
      "step": 2768
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.418867588043213,
      "learning_rate": 1.9776276601405488e-05,
      "loss": 0.8184,
      "step": 2769
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6247665882110596,
      "learning_rate": 1.9776041676987123e-05,
      "loss": 0.6976,
      "step": 2770
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8433443307876587,
      "learning_rate": 1.9775806630687178e-05,
      "loss": 0.744,
      "step": 2771
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.244940996170044,
      "learning_rate": 1.9775571462508583e-05,
      "loss": 0.8556,
      "step": 2772
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.831948161125183,
      "learning_rate": 1.9775336172454267e-05,
      "loss": 0.8242,
      "step": 2773
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9444353580474854,
      "learning_rate": 1.9775100760527166e-05,
      "loss": 0.9205,
      "step": 2774
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.775368332862854,
      "learning_rate": 1.977486522673022e-05,
      "loss": 0.6674,
      "step": 2775
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9271056652069092,
      "learning_rate": 1.977462957106635e-05,
      "loss": 0.7265,
      "step": 2776
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7112253904342651,
      "learning_rate": 1.9774393793538513e-05,
      "loss": 0.7728,
      "step": 2777
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7106209993362427,
      "learning_rate": 1.977415789414964e-05,
      "loss": 0.6856,
      "step": 2778
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8071813583374023,
      "learning_rate": 1.9773921872902666e-05,
      "loss": 0.7219,
      "step": 2779
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6674245595932007,
      "learning_rate": 1.9773685729800544e-05,
      "loss": 0.8143,
      "step": 2780
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.617523431777954,
      "learning_rate": 1.9773449464846213e-05,
      "loss": 0.8438,
      "step": 2781
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.060844898223877,
      "learning_rate": 1.9773213078042615e-05,
      "loss": 0.7277,
      "step": 2782
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8462345600128174,
      "learning_rate": 1.9772976569392708e-05,
      "loss": 0.699,
      "step": 2783
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8183976411819458,
      "learning_rate": 1.977273993889943e-05,
      "loss": 0.7445,
      "step": 2784
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6312888860702515,
      "learning_rate": 1.9772503186565738e-05,
      "loss": 0.7764,
      "step": 2785
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8647462129592896,
      "learning_rate": 1.9772266312394578e-05,
      "loss": 0.952,
      "step": 2786
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5457407236099243,
      "learning_rate": 1.977202931638891e-05,
      "loss": 0.9457,
      "step": 2787
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.529544711112976,
      "learning_rate": 1.977179219855168e-05,
      "loss": 0.6914,
      "step": 2788
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8131071329116821,
      "learning_rate": 1.977155495888585e-05,
      "loss": 0.9112,
      "step": 2789
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5152649879455566,
      "learning_rate": 1.977131759739438e-05,
      "loss": 0.6559,
      "step": 2790
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5168907642364502,
      "learning_rate": 1.977108011408022e-05,
      "loss": 0.7517,
      "step": 2791
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8117295503616333,
      "learning_rate": 1.9770842508946342e-05,
      "loss": 0.7827,
      "step": 2792
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9797521829605103,
      "learning_rate": 1.9770604781995702e-05,
      "loss": 0.8077,
      "step": 2793
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9364356994628906,
      "learning_rate": 1.9770366933231262e-05,
      "loss": 0.7795,
      "step": 2794
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7685905694961548,
      "learning_rate": 1.9770128962655993e-05,
      "loss": 0.8843,
      "step": 2795
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4927341938018799,
      "learning_rate": 1.9769890870272856e-05,
      "loss": 0.7849,
      "step": 2796
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5000079870224,
      "learning_rate": 1.976965265608483e-05,
      "loss": 0.6633,
      "step": 2797
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9293720722198486,
      "learning_rate": 1.976941432009487e-05,
      "loss": 0.6452,
      "step": 2798
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8854222297668457,
      "learning_rate": 1.9769175862305956e-05,
      "loss": 0.765,
      "step": 2799
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.717624306678772,
      "learning_rate": 1.9768937282721062e-05,
      "loss": 0.6465,
      "step": 2800
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2052199840545654,
      "learning_rate": 1.976869858134316e-05,
      "loss": 0.6925,
      "step": 2801
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.676418423652649,
      "learning_rate": 1.9768459758175224e-05,
      "loss": 0.7802,
      "step": 2802
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.198270082473755,
      "learning_rate": 1.9768220813220235e-05,
      "loss": 0.8302,
      "step": 2803
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1643259525299072,
      "learning_rate": 1.976798174648117e-05,
      "loss": 0.7844,
      "step": 2804
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9422485828399658,
      "learning_rate": 1.9767742557961015e-05,
      "loss": 0.817,
      "step": 2805
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5542278289794922,
      "learning_rate": 1.9767503247662742e-05,
      "loss": 0.7742,
      "step": 2806
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.768497109413147,
      "learning_rate": 1.976726381558934e-05,
      "loss": 0.7671,
      "step": 2807
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7114721536636353,
      "learning_rate": 1.9767024261743797e-05,
      "loss": 0.8776,
      "step": 2808
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7644461393356323,
      "learning_rate": 1.9766784586129095e-05,
      "loss": 0.5785,
      "step": 2809
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0515003204345703,
      "learning_rate": 1.9766544788748226e-05,
      "loss": 0.5704,
      "step": 2810
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.789812445640564,
      "learning_rate": 1.9766304869604175e-05,
      "loss": 0.961,
      "step": 2811
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.2531328201293945,
      "learning_rate": 1.9766064828699934e-05,
      "loss": 0.82,
      "step": 2812
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8874213695526123,
      "learning_rate": 1.97658246660385e-05,
      "loss": 0.6782,
      "step": 2813
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7533183097839355,
      "learning_rate": 1.9765584381622863e-05,
      "loss": 0.8712,
      "step": 2814
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8526397943496704,
      "learning_rate": 1.976534397545602e-05,
      "loss": 0.6666,
      "step": 2815
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.586181640625,
      "learning_rate": 1.976510344754097e-05,
      "loss": 0.8345,
      "step": 2816
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.1395294666290283,
      "learning_rate": 1.976486279788071e-05,
      "loss": 0.9084,
      "step": 2817
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8297163248062134,
      "learning_rate": 1.9764622026478237e-05,
      "loss": 0.8452,
      "step": 2818
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.826894998550415,
      "learning_rate": 1.976438113333656e-05,
      "loss": 0.9011,
      "step": 2819
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.806146502494812,
      "learning_rate": 1.9764140118458678e-05,
      "loss": 0.684,
      "step": 2820
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8830225467681885,
      "learning_rate": 1.9763898981847596e-05,
      "loss": 0.8133,
      "step": 2821
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.719005823135376,
      "learning_rate": 1.976365772350632e-05,
      "loss": 0.7352,
      "step": 2822
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.4614737033843994,
      "learning_rate": 1.976341634343786e-05,
      "loss": 0.9546,
      "step": 2823
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5792466402053833,
      "learning_rate": 1.976317484164522e-05,
      "loss": 0.7743,
      "step": 2824
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7046562433242798,
      "learning_rate": 1.9762933218131418e-05,
      "loss": 0.8344,
      "step": 2825
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8567501306533813,
      "learning_rate": 1.9762691472899462e-05,
      "loss": 0.738,
      "step": 2826
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.872323751449585,
      "learning_rate": 1.9762449605952368e-05,
      "loss": 0.7812,
      "step": 2827
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6154109239578247,
      "learning_rate": 1.976220761729315e-05,
      "loss": 0.8536,
      "step": 2828
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9149761199951172,
      "learning_rate": 1.9761965506924827e-05,
      "loss": 0.8528,
      "step": 2829
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4830349683761597,
      "learning_rate": 1.9761723274850415e-05,
      "loss": 0.6897,
      "step": 2830
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5737531185150146,
      "learning_rate": 1.9761480921072933e-05,
      "loss": 0.7385,
      "step": 2831
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6613861322402954,
      "learning_rate": 1.9761238445595404e-05,
      "loss": 0.8279,
      "step": 2832
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.635148525238037,
      "learning_rate": 1.9760995848420857e-05,
      "loss": 0.6894,
      "step": 2833
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8204129934310913,
      "learning_rate": 1.9760753129552307e-05,
      "loss": 0.7823,
      "step": 2834
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7909736633300781,
      "learning_rate": 1.9760510288992787e-05,
      "loss": 0.7511,
      "step": 2835
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6909228563308716,
      "learning_rate": 1.976026732674532e-05,
      "loss": 0.7673,
      "step": 2836
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.692775845527649,
      "learning_rate": 1.9760024242812938e-05,
      "loss": 0.6308,
      "step": 2837
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7971491813659668,
      "learning_rate": 1.9759781037198666e-05,
      "loss": 0.7563,
      "step": 2838
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7572640180587769,
      "learning_rate": 1.9759537709905544e-05,
      "loss": 0.677,
      "step": 2839
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6806389093399048,
      "learning_rate": 1.9759294260936603e-05,
      "loss": 0.8944,
      "step": 2840
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8649413585662842,
      "learning_rate": 1.9759050690294876e-05,
      "loss": 0.7801,
      "step": 2841
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4931657314300537,
      "learning_rate": 1.9758806997983398e-05,
      "loss": 0.8152,
      "step": 2842
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8638495206832886,
      "learning_rate": 1.975856318400522e-05,
      "loss": 0.6961,
      "step": 2843
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.960382342338562,
      "learning_rate": 1.975831924836336e-05,
      "loss": 0.727,
      "step": 2844
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9310991764068604,
      "learning_rate": 1.9758075191060883e-05,
      "loss": 0.9446,
      "step": 2845
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6167192459106445,
      "learning_rate": 1.975783101210081e-05,
      "loss": 0.79,
      "step": 2846
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6343120336532593,
      "learning_rate": 1.9757586711486202e-05,
      "loss": 0.814,
      "step": 2847
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9054361581802368,
      "learning_rate": 1.9757342289220094e-05,
      "loss": 0.9055,
      "step": 2848
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9566067457199097,
      "learning_rate": 1.9757097745305537e-05,
      "loss": 0.8061,
      "step": 2849
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.818731427192688,
      "learning_rate": 1.9756853079745582e-05,
      "loss": 0.7597,
      "step": 2850
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6277600526809692,
      "learning_rate": 1.975660829254328e-05,
      "loss": 0.7579,
      "step": 2851
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6168789863586426,
      "learning_rate": 1.9756363383701678e-05,
      "loss": 0.6407,
      "step": 2852
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.237363576889038,
      "learning_rate": 1.9756118353223833e-05,
      "loss": 0.9031,
      "step": 2853
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.81245756149292,
      "learning_rate": 1.97558732011128e-05,
      "loss": 0.7868,
      "step": 2854
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.4087443351745605,
      "learning_rate": 1.975562792737163e-05,
      "loss": 0.9596,
      "step": 2855
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8635741472244263,
      "learning_rate": 1.9755382532003384e-05,
      "loss": 0.8232,
      "step": 2856
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9920425415039062,
      "learning_rate": 1.975513701501113e-05,
      "loss": 0.763,
      "step": 2857
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.634156584739685,
      "learning_rate": 1.9754891376397913e-05,
      "loss": 0.8093,
      "step": 2858
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.647100567817688,
      "learning_rate": 1.9754645616166808e-05,
      "loss": 0.7913,
      "step": 2859
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.968064785003662,
      "learning_rate": 1.9754399734320874e-05,
      "loss": 0.8246,
      "step": 2860
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7926474809646606,
      "learning_rate": 1.9754153730863178e-05,
      "loss": 0.8373,
      "step": 2861
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.715740442276001,
      "learning_rate": 1.9753907605796785e-05,
      "loss": 0.7038,
      "step": 2862
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8488520383834839,
      "learning_rate": 1.975366135912477e-05,
      "loss": 0.7577,
      "step": 2863
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0275909900665283,
      "learning_rate": 1.975341499085019e-05,
      "loss": 0.8267,
      "step": 2864
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7904962301254272,
      "learning_rate": 1.9753168500976127e-05,
      "loss": 0.8416,
      "step": 2865
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4817078113555908,
      "learning_rate": 1.9752921889505652e-05,
      "loss": 0.8134,
      "step": 2866
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8018896579742432,
      "learning_rate": 1.9752675156441842e-05,
      "loss": 0.6944,
      "step": 2867
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7928414344787598,
      "learning_rate": 1.9752428301787767e-05,
      "loss": 0.8704,
      "step": 2868
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8817001581192017,
      "learning_rate": 1.975218132554651e-05,
      "loss": 0.8029,
      "step": 2869
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8367600440979004,
      "learning_rate": 1.9751934227721148e-05,
      "loss": 0.9262,
      "step": 2870
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7008731365203857,
      "learning_rate": 1.975168700831476e-05,
      "loss": 0.6502,
      "step": 2871
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8731071949005127,
      "learning_rate": 1.975143966733043e-05,
      "loss": 0.7885,
      "step": 2872
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8820443153381348,
      "learning_rate": 1.9751192204771245e-05,
      "loss": 0.8962,
      "step": 2873
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7952388525009155,
      "learning_rate": 1.9750944620640283e-05,
      "loss": 0.8782,
      "step": 2874
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5870712995529175,
      "learning_rate": 1.9750696914940635e-05,
      "loss": 0.6704,
      "step": 2875
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6337884664535522,
      "learning_rate": 1.9750449087675388e-05,
      "loss": 0.7521,
      "step": 2876
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7927947044372559,
      "learning_rate": 1.9750201138847635e-05,
      "loss": 0.8172,
      "step": 2877
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0331332683563232,
      "learning_rate": 1.9749953068460463e-05,
      "loss": 0.8342,
      "step": 2878
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9926049709320068,
      "learning_rate": 1.9749704876516967e-05,
      "loss": 0.5746,
      "step": 2879
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7569139003753662,
      "learning_rate": 1.9749456563020243e-05,
      "loss": 0.8262,
      "step": 2880
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8998463153839111,
      "learning_rate": 1.9749208127973386e-05,
      "loss": 1.0089,
      "step": 2881
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8745988607406616,
      "learning_rate": 1.9748959571379486e-05,
      "loss": 0.6897,
      "step": 2882
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7457177639007568,
      "learning_rate": 1.974871089324165e-05,
      "loss": 0.7016,
      "step": 2883
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8814295530319214,
      "learning_rate": 1.974846209356298e-05,
      "loss": 0.6489,
      "step": 2884
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.87429940700531,
      "learning_rate": 1.974821317234657e-05,
      "loss": 0.8946,
      "step": 2885
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6743757724761963,
      "learning_rate": 1.974796412959553e-05,
      "loss": 0.785,
      "step": 2886
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8120537996292114,
      "learning_rate": 1.974771496531296e-05,
      "loss": 0.8163,
      "step": 2887
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4752051830291748,
      "learning_rate": 1.9747465679501968e-05,
      "loss": 0.8836,
      "step": 2888
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7040541172027588,
      "learning_rate": 1.9747216272165665e-05,
      "loss": 0.8636,
      "step": 2889
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.862507939338684,
      "learning_rate": 1.9746966743307162e-05,
      "loss": 0.8636,
      "step": 2890
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.692354679107666,
      "learning_rate": 1.974671709292956e-05,
      "loss": 0.6313,
      "step": 2891
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.5230512619018555,
      "learning_rate": 1.974646732103598e-05,
      "loss": 0.7759,
      "step": 2892
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.824294924736023,
      "learning_rate": 1.9746217427629534e-05,
      "loss": 0.9132,
      "step": 2893
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6473480463027954,
      "learning_rate": 1.9745967412713338e-05,
      "loss": 0.8352,
      "step": 2894
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7142525911331177,
      "learning_rate": 1.9745717276290505e-05,
      "loss": 0.7215,
      "step": 2895
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2721340656280518,
      "learning_rate": 1.974546701836416e-05,
      "loss": 0.8648,
      "step": 2896
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7505725622177124,
      "learning_rate": 1.9745216638937422e-05,
      "loss": 0.7499,
      "step": 2897
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8216915130615234,
      "learning_rate": 1.9744966138013402e-05,
      "loss": 0.7058,
      "step": 2898
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.298537254333496,
      "learning_rate": 1.974471551559524e-05,
      "loss": 0.8335,
      "step": 2899
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.616792917251587,
      "learning_rate": 1.9744464771686048e-05,
      "loss": 0.8334,
      "step": 2900
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8258191347122192,
      "learning_rate": 1.974421390628896e-05,
      "loss": 0.8754,
      "step": 2901
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8318077325820923,
      "learning_rate": 1.9743962919407097e-05,
      "loss": 0.7407,
      "step": 2902
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8545058965682983,
      "learning_rate": 1.974371181104359e-05,
      "loss": 0.716,
      "step": 2903
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7314984798431396,
      "learning_rate": 1.974346058120157e-05,
      "loss": 0.9357,
      "step": 2904
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6473824977874756,
      "learning_rate": 1.974320922988417e-05,
      "loss": 0.8524,
      "step": 2905
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.047386407852173,
      "learning_rate": 1.974295775709453e-05,
      "loss": 0.7987,
      "step": 2906
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7180839776992798,
      "learning_rate": 1.974270616283577e-05,
      "loss": 0.8868,
      "step": 2907
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6023905277252197,
      "learning_rate": 1.974245444711104e-05,
      "loss": 0.852,
      "step": 2908
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.817975640296936,
      "learning_rate": 1.9742202609923476e-05,
      "loss": 0.7494,
      "step": 2909
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.703830361366272,
      "learning_rate": 1.974195065127621e-05,
      "loss": 0.9024,
      "step": 2910
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8152039051055908,
      "learning_rate": 1.9741698571172394e-05,
      "loss": 0.7549,
      "step": 2911
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4947493076324463,
      "learning_rate": 1.974144636961516e-05,
      "loss": 0.8447,
      "step": 2912
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9455150365829468,
      "learning_rate": 1.9741194046607663e-05,
      "loss": 0.792,
      "step": 2913
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7303491830825806,
      "learning_rate": 1.974094160215304e-05,
      "loss": 0.9013,
      "step": 2914
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6653423309326172,
      "learning_rate": 1.9740689036254442e-05,
      "loss": 0.7233,
      "step": 2915
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.721893310546875,
      "learning_rate": 1.9740436348915023e-05,
      "loss": 0.8143,
      "step": 2916
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0263283252716064,
      "learning_rate": 1.9740183540137923e-05,
      "loss": 0.8008,
      "step": 2917
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6120623350143433,
      "learning_rate": 1.97399306099263e-05,
      "loss": 0.751,
      "step": 2918
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8369895219802856,
      "learning_rate": 1.9739677558283306e-05,
      "loss": 0.8992,
      "step": 2919
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6981680393218994,
      "learning_rate": 1.9739424385212097e-05,
      "loss": 0.8472,
      "step": 2920
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5924022197723389,
      "learning_rate": 1.973917109071583e-05,
      "loss": 0.7783,
      "step": 2921
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8723801374435425,
      "learning_rate": 1.973891767479766e-05,
      "loss": 0.7727,
      "step": 2922
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7472656965255737,
      "learning_rate": 1.9738664137460745e-05,
      "loss": 0.8564,
      "step": 2923
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.911554217338562,
      "learning_rate": 1.973841047870825e-05,
      "loss": 0.7814,
      "step": 2924
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7871829271316528,
      "learning_rate": 1.973815669854334e-05,
      "loss": 0.8397,
      "step": 2925
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.590516209602356,
      "learning_rate": 1.973790279696917e-05,
      "loss": 0.7554,
      "step": 2926
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.2011826038360596,
      "learning_rate": 1.9737648773988916e-05,
      "loss": 0.8523,
      "step": 2927
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9736788272857666,
      "learning_rate": 1.9737394629605738e-05,
      "loss": 0.832,
      "step": 2928
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9615586996078491,
      "learning_rate": 1.9737140363822806e-05,
      "loss": 0.8639,
      "step": 2929
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7329647541046143,
      "learning_rate": 1.9736885976643293e-05,
      "loss": 0.7986,
      "step": 2930
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.9032869338989258,
      "learning_rate": 1.9736631468070364e-05,
      "loss": 0.8802,
      "step": 2931
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8375779390335083,
      "learning_rate": 1.97363768381072e-05,
      "loss": 0.7015,
      "step": 2932
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7806909084320068,
      "learning_rate": 1.9736122086756966e-05,
      "loss": 0.8924,
      "step": 2933
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5301276445388794,
      "learning_rate": 1.973586721402285e-05,
      "loss": 0.6758,
      "step": 2934
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8522441387176514,
      "learning_rate": 1.973561221990802e-05,
      "loss": 0.7941,
      "step": 2935
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7040064334869385,
      "learning_rate": 1.973535710441566e-05,
      "loss": 0.7187,
      "step": 2936
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8175921440124512,
      "learning_rate": 1.9735101867548945e-05,
      "loss": 0.7345,
      "step": 2937
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.804065227508545,
      "learning_rate": 1.973484650931106e-05,
      "loss": 0.8569,
      "step": 2938
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0953638553619385,
      "learning_rate": 1.9734591029705196e-05,
      "loss": 0.8182,
      "step": 2939
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0304770469665527,
      "learning_rate": 1.973433542873453e-05,
      "loss": 0.8107,
      "step": 2940
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8228020668029785,
      "learning_rate": 1.9734079706402246e-05,
      "loss": 0.7781,
      "step": 2941
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.209925651550293,
      "learning_rate": 1.9733823862711544e-05,
      "loss": 0.9635,
      "step": 2942
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.628171443939209,
      "learning_rate": 1.9733567897665602e-05,
      "loss": 0.8089,
      "step": 2943
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.85551118850708,
      "learning_rate": 1.9733311811267614e-05,
      "loss": 0.7476,
      "step": 2944
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5979894399642944,
      "learning_rate": 1.9733055603520776e-05,
      "loss": 0.8047,
      "step": 2945
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4923887252807617,
      "learning_rate": 1.9732799274428283e-05,
      "loss": 0.6514,
      "step": 2946
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5818737745285034,
      "learning_rate": 1.973254282399333e-05,
      "loss": 0.7515,
      "step": 2947
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6995186805725098,
      "learning_rate": 1.9732286252219106e-05,
      "loss": 0.6365,
      "step": 2948
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5468400716781616,
      "learning_rate": 1.973202955910882e-05,
      "loss": 0.7915,
      "step": 2949
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5405547618865967,
      "learning_rate": 1.9731772744665668e-05,
      "loss": 0.849,
      "step": 2950
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9113802909851074,
      "learning_rate": 1.9731515808892853e-05,
      "loss": 0.8763,
      "step": 2951
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.068298816680908,
      "learning_rate": 1.973125875179358e-05,
      "loss": 0.7824,
      "step": 2952
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9827187061309814,
      "learning_rate": 1.973100157337105e-05,
      "loss": 0.7509,
      "step": 2953
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8852801322937012,
      "learning_rate": 1.973074427362847e-05,
      "loss": 0.8446,
      "step": 2954
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6701247692108154,
      "learning_rate": 1.9730486852569048e-05,
      "loss": 0.6824,
      "step": 2955
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4908740520477295,
      "learning_rate": 1.9730229310195997e-05,
      "loss": 0.8079,
      "step": 2956
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1876275539398193,
      "learning_rate": 1.9729971646512523e-05,
      "loss": 0.8898,
      "step": 2957
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8695796728134155,
      "learning_rate": 1.972971386152184e-05,
      "loss": 0.7841,
      "step": 2958
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9464763402938843,
      "learning_rate": 1.9729455955227164e-05,
      "loss": 0.6731,
      "step": 2959
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6709498167037964,
      "learning_rate": 1.972919792763171e-05,
      "loss": 0.716,
      "step": 2960
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5133949518203735,
      "learning_rate": 1.972893977873869e-05,
      "loss": 0.7269,
      "step": 2961
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5870754718780518,
      "learning_rate": 1.972868150855133e-05,
      "loss": 0.7904,
      "step": 2962
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.210137367248535,
      "learning_rate": 1.972842311707284e-05,
      "loss": 0.9714,
      "step": 2963
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.742958426475525,
      "learning_rate": 1.9728164604306454e-05,
      "loss": 0.9369,
      "step": 2964
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7265640497207642,
      "learning_rate": 1.972790597025539e-05,
      "loss": 0.8345,
      "step": 2965
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8455003499984741,
      "learning_rate": 1.9727647214922863e-05,
      "loss": 0.8003,
      "step": 2966
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.975825309753418,
      "learning_rate": 1.9727388338312112e-05,
      "loss": 0.8511,
      "step": 2967
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7821764945983887,
      "learning_rate": 1.972712934042636e-05,
      "loss": 0.9293,
      "step": 2968
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8223668336868286,
      "learning_rate": 1.972687022126883e-05,
      "loss": 0.793,
      "step": 2969
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7717041969299316,
      "learning_rate": 1.9726610980842766e-05,
      "loss": 0.716,
      "step": 2970
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3798474073410034,
      "learning_rate": 1.9726351619151388e-05,
      "loss": 0.5712,
      "step": 2971
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.923438310623169,
      "learning_rate": 1.9726092136197933e-05,
      "loss": 0.7601,
      "step": 2972
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8152110576629639,
      "learning_rate": 1.9725832531985644e-05,
      "loss": 0.689,
      "step": 2973
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7143770456314087,
      "learning_rate": 1.972557280651774e-05,
      "loss": 0.7135,
      "step": 2974
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8645702600479126,
      "learning_rate": 1.9725312959797476e-05,
      "loss": 0.7852,
      "step": 2975
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6261523962020874,
      "learning_rate": 1.9725052991828084e-05,
      "loss": 0.7218,
      "step": 2976
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5818004608154297,
      "learning_rate": 1.9724792902612802e-05,
      "loss": 0.6316,
      "step": 2977
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5991036891937256,
      "learning_rate": 1.9724532692154882e-05,
      "loss": 0.7624,
      "step": 2978
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6952545642852783,
      "learning_rate": 1.972427236045756e-05,
      "loss": 0.8255,
      "step": 2979
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9178556203842163,
      "learning_rate": 1.9724011907524086e-05,
      "loss": 0.8154,
      "step": 2980
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6139378547668457,
      "learning_rate": 1.9723751333357706e-05,
      "loss": 0.6924,
      "step": 2981
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7043781280517578,
      "learning_rate": 1.972349063796167e-05,
      "loss": 0.7251,
      "step": 2982
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7612953186035156,
      "learning_rate": 1.9723229821339227e-05,
      "loss": 0.8552,
      "step": 2983
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7569830417633057,
      "learning_rate": 1.9722968883493623e-05,
      "loss": 0.8091,
      "step": 2984
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0181806087493896,
      "learning_rate": 1.972270782442812e-05,
      "loss": 0.7557,
      "step": 2985
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7724518775939941,
      "learning_rate": 1.9722446644145966e-05,
      "loss": 0.8256,
      "step": 2986
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0262951850891113,
      "learning_rate": 1.9722185342650422e-05,
      "loss": 0.7382,
      "step": 2987
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8551321029663086,
      "learning_rate": 1.972192391994475e-05,
      "loss": 0.8015,
      "step": 2988
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6677227020263672,
      "learning_rate": 1.9721662376032197e-05,
      "loss": 0.6949,
      "step": 2989
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.123671054840088,
      "learning_rate": 1.972140071091603e-05,
      "loss": 0.7934,
      "step": 2990
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7019320726394653,
      "learning_rate": 1.9721138924599516e-05,
      "loss": 0.8208,
      "step": 2991
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.017636299133301,
      "learning_rate": 1.972087701708591e-05,
      "loss": 0.7838,
      "step": 2992
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9807765483856201,
      "learning_rate": 1.9720614988378484e-05,
      "loss": 0.8843,
      "step": 2993
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.16713547706604,
      "learning_rate": 1.9720352838480507e-05,
      "loss": 0.8905,
      "step": 2994
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9649841785430908,
      "learning_rate": 1.9720090567395232e-05,
      "loss": 0.9421,
      "step": 2995
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.2649879455566406,
      "learning_rate": 1.971982817512595e-05,
      "loss": 0.7186,
      "step": 2996
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6607991456985474,
      "learning_rate": 1.9719565661675915e-05,
      "loss": 0.8239,
      "step": 2997
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8376497030258179,
      "learning_rate": 1.971930302704841e-05,
      "loss": 0.8363,
      "step": 2998
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5214895009994507,
      "learning_rate": 1.9719040271246704e-05,
      "loss": 0.6271,
      "step": 2999
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.782069206237793,
      "learning_rate": 1.971877739427408e-05,
      "loss": 0.7372,
      "step": 3000
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.86141037940979,
      "learning_rate": 1.9718514396133805e-05,
      "loss": 0.8081,
      "step": 3001
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0443930625915527,
      "learning_rate": 1.9718251276829164e-05,
      "loss": 0.8949,
      "step": 3002
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8615152835845947,
      "learning_rate": 1.971798803636344e-05,
      "loss": 0.7992,
      "step": 3003
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6837289333343506,
      "learning_rate": 1.971772467473991e-05,
      "loss": 0.81,
      "step": 3004
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7311434745788574,
      "learning_rate": 1.9717461191961856e-05,
      "loss": 0.8274,
      "step": 3005
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.954944372177124,
      "learning_rate": 1.971719758803257e-05,
      "loss": 0.8328,
      "step": 3006
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7333757877349854,
      "learning_rate": 1.971693386295533e-05,
      "loss": 0.8846,
      "step": 3007
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6571557521820068,
      "learning_rate": 1.9716670016733436e-05,
      "loss": 0.8651,
      "step": 3008
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6028467416763306,
      "learning_rate": 1.9716406049370165e-05,
      "loss": 0.7728,
      "step": 3009
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5825470685958862,
      "learning_rate": 1.9716141960868813e-05,
      "loss": 0.788,
      "step": 3010
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7805736064910889,
      "learning_rate": 1.971587775123267e-05,
      "loss": 0.8801,
      "step": 3011
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5601176023483276,
      "learning_rate": 1.9715613420465038e-05,
      "loss": 0.7442,
      "step": 3012
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7662566900253296,
      "learning_rate": 1.97153489685692e-05,
      "loss": 0.7491,
      "step": 3013
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.835625410079956,
      "learning_rate": 1.9715084395548465e-05,
      "loss": 0.8142,
      "step": 3014
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8657026290893555,
      "learning_rate": 1.971481970140613e-05,
      "loss": 0.9628,
      "step": 3015
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.286679744720459,
      "learning_rate": 1.9714554886145483e-05,
      "loss": 0.8476,
      "step": 3016
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9821910858154297,
      "learning_rate": 1.9714289949769836e-05,
      "loss": 0.7996,
      "step": 3017
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4932575225830078,
      "learning_rate": 1.9714024892282495e-05,
      "loss": 0.7328,
      "step": 3018
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6122568845748901,
      "learning_rate": 1.9713759713686755e-05,
      "loss": 0.7103,
      "step": 3019
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8020265102386475,
      "learning_rate": 1.9713494413985924e-05,
      "loss": 0.856,
      "step": 3020
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4754986763000488,
      "learning_rate": 1.9713228993183318e-05,
      "loss": 0.7112,
      "step": 3021
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.462268590927124,
      "learning_rate": 1.9712963451282242e-05,
      "loss": 0.6514,
      "step": 3022
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.065854549407959,
      "learning_rate": 1.9712697788286e-05,
      "loss": 0.8297,
      "step": 3023
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1548025608062744,
      "learning_rate": 1.971243200419791e-05,
      "loss": 0.9713,
      "step": 3024
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.741076111793518,
      "learning_rate": 1.9712166099021287e-05,
      "loss": 0.883,
      "step": 3025
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.617133617401123,
      "learning_rate": 1.971190007275944e-05,
      "loss": 0.8767,
      "step": 3026
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8551080226898193,
      "learning_rate": 1.9711633925415695e-05,
      "loss": 0.9363,
      "step": 3027
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.644111156463623,
      "learning_rate": 1.9711367656993362e-05,
      "loss": 0.6629,
      "step": 3028
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.789528489112854,
      "learning_rate": 1.9711101267495764e-05,
      "loss": 0.7392,
      "step": 3029
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8073285818099976,
      "learning_rate": 1.9710834756926217e-05,
      "loss": 0.7424,
      "step": 3030
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7825709581375122,
      "learning_rate": 1.971056812528805e-05,
      "loss": 0.7607,
      "step": 3031
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5592150688171387,
      "learning_rate": 1.9710301372584592e-05,
      "loss": 0.8545,
      "step": 3032
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5230741500854492,
      "learning_rate": 1.9710034498819155e-05,
      "loss": 0.7175,
      "step": 3033
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.1594879627227783,
      "learning_rate": 1.970976750399508e-05,
      "loss": 0.8302,
      "step": 3034
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6955256462097168,
      "learning_rate": 1.9709500388115682e-05,
      "loss": 0.7376,
      "step": 3035
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4523637294769287,
      "learning_rate": 1.9709233151184303e-05,
      "loss": 0.7093,
      "step": 3036
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.72709059715271,
      "learning_rate": 1.9708965793204266e-05,
      "loss": 0.8408,
      "step": 3037
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5601698160171509,
      "learning_rate": 1.970869831417891e-05,
      "loss": 0.8201,
      "step": 3038
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7830803394317627,
      "learning_rate": 1.970843071411157e-05,
      "loss": 0.855,
      "step": 3039
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.0940442085266113,
      "learning_rate": 1.970816299300558e-05,
      "loss": 0.647,
      "step": 3040
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.848454475402832,
      "learning_rate": 1.9707895150864277e-05,
      "loss": 0.7501,
      "step": 3041
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8271241188049316,
      "learning_rate": 1.9707627187691003e-05,
      "loss": 0.8047,
      "step": 3042
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.712241768836975,
      "learning_rate": 1.9707359103489098e-05,
      "loss": 0.826,
      "step": 3043
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.9405711889266968,
      "learning_rate": 1.97070908982619e-05,
      "loss": 0.8744,
      "step": 3044
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5148471593856812,
      "learning_rate": 1.970682257201276e-05,
      "loss": 0.7929,
      "step": 3045
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7429046630859375,
      "learning_rate": 1.9706554124745015e-05,
      "loss": 0.6922,
      "step": 3046
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9355660676956177,
      "learning_rate": 1.9706285556462022e-05,
      "loss": 0.7512,
      "step": 3047
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8281441926956177,
      "learning_rate": 1.9706016867167118e-05,
      "loss": 0.7541,
      "step": 3048
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.89713716506958,
      "learning_rate": 1.9705748056863664e-05,
      "loss": 0.8797,
      "step": 3049
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6541239023208618,
      "learning_rate": 1.9705479125555e-05,
      "loss": 0.6868,
      "step": 3050
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6287329196929932,
      "learning_rate": 1.970521007324449e-05,
      "loss": 0.7248,
      "step": 3051
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4984142780303955,
      "learning_rate": 1.9704940899935482e-05,
      "loss": 0.6449,
      "step": 3052
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.972780466079712,
      "learning_rate": 1.9704671605631334e-05,
      "loss": 0.6573,
      "step": 3053
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2324273586273193,
      "learning_rate": 1.9704402190335404e-05,
      "loss": 0.753,
      "step": 3054
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7453781366348267,
      "learning_rate": 1.9704132654051043e-05,
      "loss": 0.6765,
      "step": 3055
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.2522432804107666,
      "learning_rate": 1.9703862996781625e-05,
      "loss": 0.9827,
      "step": 3056
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9450031518936157,
      "learning_rate": 1.9703593218530502e-05,
      "loss": 0.8661,
      "step": 3057
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.662867546081543,
      "learning_rate": 1.970332331930104e-05,
      "loss": 0.8639,
      "step": 3058
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.734736442565918,
      "learning_rate": 1.9703053299096603e-05,
      "loss": 0.8317,
      "step": 3059
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9848676919937134,
      "learning_rate": 1.9702783157920562e-05,
      "loss": 0.8017,
      "step": 3060
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.322916269302368,
      "learning_rate": 1.9702512895776277e-05,
      "loss": 0.7008,
      "step": 3061
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4161183834075928,
      "learning_rate": 1.9702242512667126e-05,
      "loss": 0.8168,
      "step": 3062
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6611062288284302,
      "learning_rate": 1.9701972008596475e-05,
      "loss": 0.6948,
      "step": 3063
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9912800788879395,
      "learning_rate": 1.9701701383567697e-05,
      "loss": 0.9218,
      "step": 3064
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9923813343048096,
      "learning_rate": 1.9701430637584166e-05,
      "loss": 0.7925,
      "step": 3065
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.663568377494812,
      "learning_rate": 1.970115977064926e-05,
      "loss": 0.835,
      "step": 3066
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0180013179779053,
      "learning_rate": 1.9700888782766352e-05,
      "loss": 0.7556,
      "step": 3067
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.168924570083618,
      "learning_rate": 1.9700617673938822e-05,
      "loss": 0.7037,
      "step": 3068
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9062621593475342,
      "learning_rate": 1.970034644417005e-05,
      "loss": 0.8564,
      "step": 3069
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.861117959022522,
      "learning_rate": 1.970007509346342e-05,
      "loss": 0.735,
      "step": 3070
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1006975173950195,
      "learning_rate": 1.969980362182231e-05,
      "loss": 0.816,
      "step": 3071
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9552007913589478,
      "learning_rate": 1.969953202925011e-05,
      "loss": 0.9332,
      "step": 3072
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8009060621261597,
      "learning_rate": 1.9699260315750207e-05,
      "loss": 0.7006,
      "step": 3073
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.840852975845337,
      "learning_rate": 1.969898848132598e-05,
      "loss": 0.7962,
      "step": 3074
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9432482719421387,
      "learning_rate": 1.9698716525980825e-05,
      "loss": 0.7463,
      "step": 3075
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7827035188674927,
      "learning_rate": 1.969844444971813e-05,
      "loss": 0.8046,
      "step": 3076
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9881911277770996,
      "learning_rate": 1.9698172252541288e-05,
      "loss": 0.7817,
      "step": 3077
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.194566011428833,
      "learning_rate": 1.9697899934453694e-05,
      "loss": 0.7207,
      "step": 3078
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0422778129577637,
      "learning_rate": 1.969762749545874e-05,
      "loss": 0.9322,
      "step": 3079
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7648824453353882,
      "learning_rate": 1.9697354935559826e-05,
      "loss": 0.8538,
      "step": 3080
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8958649635314941,
      "learning_rate": 1.9697082254760347e-05,
      "loss": 0.8849,
      "step": 3081
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9676176309585571,
      "learning_rate": 1.9696809453063702e-05,
      "loss": 0.8458,
      "step": 3082
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7852790355682373,
      "learning_rate": 1.9696536530473294e-05,
      "loss": 0.8142,
      "step": 3083
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9253687858581543,
      "learning_rate": 1.969626348699253e-05,
      "loss": 0.7432,
      "step": 3084
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9674912691116333,
      "learning_rate": 1.9695990322624806e-05,
      "loss": 0.9247,
      "step": 3085
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.768267035484314,
      "learning_rate": 1.969571703737353e-05,
      "loss": 0.8054,
      "step": 3086
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8050336837768555,
      "learning_rate": 1.9695443631242114e-05,
      "loss": 0.9385,
      "step": 3087
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8449900150299072,
      "learning_rate": 1.9695170104233962e-05,
      "loss": 0.8495,
      "step": 3088
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.11079478263855,
      "learning_rate": 1.9694896456352488e-05,
      "loss": 0.8225,
      "step": 3089
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6724408864974976,
      "learning_rate": 1.96946226876011e-05,
      "loss": 0.7533,
      "step": 3090
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.3044471740722656,
      "learning_rate": 1.9694348797983207e-05,
      "loss": 0.8034,
      "step": 3091
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9264733791351318,
      "learning_rate": 1.9694074787502236e-05,
      "loss": 0.7733,
      "step": 3092
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5257816314697266,
      "learning_rate": 1.969380065616159e-05,
      "loss": 0.7735,
      "step": 3093
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6246085166931152,
      "learning_rate": 1.9693526403964695e-05,
      "loss": 0.8227,
      "step": 3094
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7164762020111084,
      "learning_rate": 1.969325203091497e-05,
      "loss": 0.8849,
      "step": 3095
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.700057864189148,
      "learning_rate": 1.9692977537015832e-05,
      "loss": 0.8863,
      "step": 3096
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4674733877182007,
      "learning_rate": 1.9692702922270707e-05,
      "loss": 0.8183,
      "step": 3097
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7329365015029907,
      "learning_rate": 1.9692428186683014e-05,
      "loss": 0.7866,
      "step": 3098
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.536234974861145,
      "learning_rate": 1.969215333025618e-05,
      "loss": 0.9611,
      "step": 3099
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.792055606842041,
      "learning_rate": 1.9691878352993635e-05,
      "loss": 0.7371,
      "step": 3100
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8010929822921753,
      "learning_rate": 1.9691603254898805e-05,
      "loss": 0.9239,
      "step": 3101
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.132768154144287,
      "learning_rate": 1.9691328035975118e-05,
      "loss": 0.848,
      "step": 3102
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5002723932266235,
      "learning_rate": 1.9691052696226006e-05,
      "loss": 0.8503,
      "step": 3103
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8474253416061401,
      "learning_rate": 1.9690777235654905e-05,
      "loss": 0.7043,
      "step": 3104
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4311165809631348,
      "learning_rate": 1.9690501654265247e-05,
      "loss": 0.8162,
      "step": 3105
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5022305250167847,
      "learning_rate": 1.9690225952060467e-05,
      "loss": 0.7173,
      "step": 3106
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6703437566757202,
      "learning_rate": 1.9689950129044002e-05,
      "loss": 0.6867,
      "step": 3107
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5818413496017456,
      "learning_rate": 1.9689674185219293e-05,
      "loss": 0.8366,
      "step": 3108
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0342297554016113,
      "learning_rate": 1.9689398120589776e-05,
      "loss": 0.8648,
      "step": 3109
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.570570707321167,
      "learning_rate": 1.96891219351589e-05,
      "loss": 0.7509,
      "step": 3110
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9674320220947266,
      "learning_rate": 1.96888456289301e-05,
      "loss": 0.7513,
      "step": 3111
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0799641609191895,
      "learning_rate": 1.9688569201906832e-05,
      "loss": 0.7299,
      "step": 3112
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9503211975097656,
      "learning_rate": 1.968829265409253e-05,
      "loss": 0.7835,
      "step": 3113
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6269712448120117,
      "learning_rate": 1.9688015985490646e-05,
      "loss": 0.898,
      "step": 3114
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9143753051757812,
      "learning_rate": 1.9687739196104632e-05,
      "loss": 0.8714,
      "step": 3115
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1146202087402344,
      "learning_rate": 1.968746228593794e-05,
      "loss": 0.9055,
      "step": 3116
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9042385816574097,
      "learning_rate": 1.9687185254994016e-05,
      "loss": 0.8383,
      "step": 3117
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9456671476364136,
      "learning_rate": 1.968690810327632e-05,
      "loss": 0.9246,
      "step": 3118
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.801957368850708,
      "learning_rate": 1.9686630830788306e-05,
      "loss": 0.8999,
      "step": 3119
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5700627565383911,
      "learning_rate": 1.9686353437533422e-05,
      "loss": 0.8309,
      "step": 3120
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7931917905807495,
      "learning_rate": 1.9686075923515143e-05,
      "loss": 0.6621,
      "step": 3121
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5154162645339966,
      "learning_rate": 1.9685798288736915e-05,
      "loss": 0.7142,
      "step": 3122
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4695361852645874,
      "learning_rate": 1.9685520533202207e-05,
      "loss": 0.6352,
      "step": 3123
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8584003448486328,
      "learning_rate": 1.9685242656914475e-05,
      "loss": 0.8303,
      "step": 3124
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5941174030303955,
      "learning_rate": 1.9684964659877193e-05,
      "loss": 0.8081,
      "step": 3125
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9589470624923706,
      "learning_rate": 1.9684686542093815e-05,
      "loss": 0.7658,
      "step": 3126
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7301608324050903,
      "learning_rate": 1.9684408303567822e-05,
      "loss": 0.846,
      "step": 3127
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.9264757633209229,
      "learning_rate": 1.968412994430267e-05,
      "loss": 0.7609,
      "step": 3128
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5748536586761475,
      "learning_rate": 1.9683851464301837e-05,
      "loss": 0.8038,
      "step": 3129
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6628766059875488,
      "learning_rate": 1.9683572863568795e-05,
      "loss": 0.7738,
      "step": 3130
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7463139295578003,
      "learning_rate": 1.9683294142107018e-05,
      "loss": 0.6254,
      "step": 3131
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.518823266029358,
      "learning_rate": 1.9683015299919973e-05,
      "loss": 0.6519,
      "step": 3132
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.00831937789917,
      "learning_rate": 1.9682736337011142e-05,
      "loss": 0.6878,
      "step": 3133
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5248388051986694,
      "learning_rate": 1.9682457253384003e-05,
      "loss": 0.8252,
      "step": 3134
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7632031440734863,
      "learning_rate": 1.9682178049042038e-05,
      "loss": 0.7942,
      "step": 3135
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.935718059539795,
      "learning_rate": 1.9681898723988723e-05,
      "loss": 0.8099,
      "step": 3136
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8886219263076782,
      "learning_rate": 1.9681619278227545e-05,
      "loss": 0.8081,
      "step": 3137
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8533939123153687,
      "learning_rate": 1.9681339711761983e-05,
      "loss": 0.6114,
      "step": 3138
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7354339361190796,
      "learning_rate": 1.9681060024595527e-05,
      "loss": 0.9541,
      "step": 3139
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7270318269729614,
      "learning_rate": 1.968078021673166e-05,
      "loss": 0.8838,
      "step": 3140
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.7411912679672241,
      "learning_rate": 1.9680500288173872e-05,
      "loss": 0.846,
      "step": 3141
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8929684162139893,
      "learning_rate": 1.968022023892566e-05,
      "loss": 0.8028,
      "step": 3142
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7403491735458374,
      "learning_rate": 1.96799400689905e-05,
      "loss": 0.7793,
      "step": 3143
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8378294706344604,
      "learning_rate": 1.96796597783719e-05,
      "loss": 0.7419,
      "step": 3144
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8347924947738647,
      "learning_rate": 1.9679379367073345e-05,
      "loss": 0.7159,
      "step": 3145
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.224592924118042,
      "learning_rate": 1.9679098835098335e-05,
      "loss": 0.8804,
      "step": 3146
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.937103271484375,
      "learning_rate": 1.967881818245037e-05,
      "loss": 0.7373,
      "step": 3147
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8420709371566772,
      "learning_rate": 1.9678537409132942e-05,
      "loss": 0.7211,
      "step": 3148
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8511948585510254,
      "learning_rate": 1.9678256515149556e-05,
      "loss": 0.8457,
      "step": 3149
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6353881359100342,
      "learning_rate": 1.9677975500503717e-05,
      "loss": 0.5965,
      "step": 3150
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9227336645126343,
      "learning_rate": 1.967769436519892e-05,
      "loss": 0.7169,
      "step": 3151
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6199570894241333,
      "learning_rate": 1.967741310923868e-05,
      "loss": 0.6929,
      "step": 3152
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7889677286148071,
      "learning_rate": 1.9677131732626494e-05,
      "loss": 0.8136,
      "step": 3153
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.628126621246338,
      "learning_rate": 1.9676850235365873e-05,
      "loss": 0.7945,
      "step": 3154
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.541437029838562,
      "learning_rate": 1.9676568617460333e-05,
      "loss": 0.5897,
      "step": 3155
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.766094446182251,
      "learning_rate": 1.967628687891338e-05,
      "loss": 0.8979,
      "step": 3156
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.968922734260559,
      "learning_rate": 1.9676005019728522e-05,
      "loss": 0.6908,
      "step": 3157
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8256292343139648,
      "learning_rate": 1.9675723039909284e-05,
      "loss": 0.803,
      "step": 3158
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7307766675949097,
      "learning_rate": 1.967544093945917e-05,
      "loss": 0.9357,
      "step": 3159
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9204020500183105,
      "learning_rate": 1.9675158718381703e-05,
      "loss": 0.8004,
      "step": 3160
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.674965262413025,
      "learning_rate": 1.9674876376680405e-05,
      "loss": 0.8153,
      "step": 3161
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6332346200942993,
      "learning_rate": 1.967459391435879e-05,
      "loss": 0.6947,
      "step": 3162
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.555092215538025,
      "learning_rate": 1.967431133142038e-05,
      "loss": 0.8257,
      "step": 3163
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6603070497512817,
      "learning_rate": 1.9674028627868696e-05,
      "loss": 0.8085,
      "step": 3164
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7119311094284058,
      "learning_rate": 1.967374580370727e-05,
      "loss": 0.9148,
      "step": 3165
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.733095645904541,
      "learning_rate": 1.9673462858939625e-05,
      "loss": 0.8226,
      "step": 3166
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6229273080825806,
      "learning_rate": 1.9673179793569287e-05,
      "loss": 0.7627,
      "step": 3167
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7568830251693726,
      "learning_rate": 1.9672896607599784e-05,
      "loss": 0.8243,
      "step": 3168
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5906846523284912,
      "learning_rate": 1.9672613301034653e-05,
      "loss": 0.8416,
      "step": 3169
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6315747499465942,
      "learning_rate": 1.9672329873877418e-05,
      "loss": 0.7784,
      "step": 3170
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5970549583435059,
      "learning_rate": 1.9672046326131615e-05,
      "loss": 0.6701,
      "step": 3171
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.986319661140442,
      "learning_rate": 1.9671762657800782e-05,
      "loss": 0.6303,
      "step": 3172
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7090332508087158,
      "learning_rate": 1.9671478868888453e-05,
      "loss": 0.8425,
      "step": 3173
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5605995655059814,
      "learning_rate": 1.967119495939817e-05,
      "loss": 0.6979,
      "step": 3174
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6094880104064941,
      "learning_rate": 1.9670910929333466e-05,
      "loss": 0.8282,
      "step": 3175
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1913397312164307,
      "learning_rate": 1.9670626778697885e-05,
      "loss": 0.9295,
      "step": 3176
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8608843088150024,
      "learning_rate": 1.9670342507494974e-05,
      "loss": 0.7475,
      "step": 3177
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7231173515319824,
      "learning_rate": 1.9670058115728267e-05,
      "loss": 0.7754,
      "step": 3178
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6489614248275757,
      "learning_rate": 1.966977360340132e-05,
      "loss": 0.7765,
      "step": 3179
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7525140047073364,
      "learning_rate": 1.966948897051768e-05,
      "loss": 0.7777,
      "step": 3180
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0874600410461426,
      "learning_rate": 1.9669204217080886e-05,
      "loss": 0.788,
      "step": 3181
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0597901344299316,
      "learning_rate": 1.9668919343094496e-05,
      "loss": 0.9216,
      "step": 3182
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8104877471923828,
      "learning_rate": 1.966863434856206e-05,
      "loss": 0.7236,
      "step": 3183
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8313372135162354,
      "learning_rate": 1.9668349233487126e-05,
      "loss": 0.8023,
      "step": 3184
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6867988109588623,
      "learning_rate": 1.966806399787326e-05,
      "loss": 0.8987,
      "step": 3185
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8737941980361938,
      "learning_rate": 1.9667778641724007e-05,
      "loss": 0.8065,
      "step": 3186
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.508273959159851,
      "learning_rate": 1.966749316504293e-05,
      "loss": 0.622,
      "step": 3187
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.605480670928955,
      "learning_rate": 1.9667207567833588e-05,
      "loss": 0.6919,
      "step": 3188
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.186652421951294,
      "learning_rate": 1.966692185009954e-05,
      "loss": 0.7525,
      "step": 3189
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1390185356140137,
      "learning_rate": 1.966663601184435e-05,
      "loss": 0.7885,
      "step": 3190
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.730812907218933,
      "learning_rate": 1.966635005307158e-05,
      "loss": 0.7494,
      "step": 3191
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6365338563919067,
      "learning_rate": 1.9666063973784794e-05,
      "loss": 0.727,
      "step": 3192
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7449599504470825,
      "learning_rate": 1.966577777398756e-05,
      "loss": 0.8623,
      "step": 3193
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5580198764801025,
      "learning_rate": 1.9665491453683452e-05,
      "loss": 0.8029,
      "step": 3194
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9491406679153442,
      "learning_rate": 1.966520501287603e-05,
      "loss": 0.7503,
      "step": 3195
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6567738056182861,
      "learning_rate": 1.9664918451568868e-05,
      "loss": 0.9057,
      "step": 3196
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7245268821716309,
      "learning_rate": 1.9664631769765544e-05,
      "loss": 0.9257,
      "step": 3197
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5068469047546387,
      "learning_rate": 1.9664344967469626e-05,
      "loss": 0.7204,
      "step": 3198
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7011643648147583,
      "learning_rate": 1.9664058044684693e-05,
      "loss": 0.8945,
      "step": 3199
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8350049257278442,
      "learning_rate": 1.966377100141432e-05,
      "loss": 0.8651,
      "step": 3200
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9272756576538086,
      "learning_rate": 1.966348383766209e-05,
      "loss": 0.7087,
      "step": 3201
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6275173425674438,
      "learning_rate": 1.9663196553431572e-05,
      "loss": 0.7413,
      "step": 3202
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6654196977615356,
      "learning_rate": 1.9662909148726364e-05,
      "loss": 0.8685,
      "step": 3203
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8215210437774658,
      "learning_rate": 1.9662621623550034e-05,
      "loss": 0.8813,
      "step": 3204
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6896398067474365,
      "learning_rate": 1.966233397790618e-05,
      "loss": 0.657,
      "step": 3205
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.74263334274292,
      "learning_rate": 1.9662046211798378e-05,
      "loss": 0.7378,
      "step": 3206
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8477424383163452,
      "learning_rate": 1.966175832523022e-05,
      "loss": 0.8697,
      "step": 3207
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5705840587615967,
      "learning_rate": 1.9661470318205294e-05,
      "loss": 0.7647,
      "step": 3208
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8390697240829468,
      "learning_rate": 1.966118219072719e-05,
      "loss": 0.836,
      "step": 3209
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5570085048675537,
      "learning_rate": 1.9660893942799503e-05,
      "loss": 0.7601,
      "step": 3210
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5033553838729858,
      "learning_rate": 1.9660605574425825e-05,
      "loss": 0.7269,
      "step": 3211
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9560163021087646,
      "learning_rate": 1.9660317085609754e-05,
      "loss": 0.8586,
      "step": 3212
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8964481353759766,
      "learning_rate": 1.966002847635488e-05,
      "loss": 0.7686,
      "step": 3213
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.659209132194519,
      "learning_rate": 1.965973974666481e-05,
      "loss": 0.9134,
      "step": 3214
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0005602836608887,
      "learning_rate": 1.9659450896543135e-05,
      "loss": 0.6619,
      "step": 3215
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7723467350006104,
      "learning_rate": 1.965916192599346e-05,
      "loss": 0.7902,
      "step": 3216
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7997571229934692,
      "learning_rate": 1.965887283501939e-05,
      "loss": 0.7561,
      "step": 3217
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.555294156074524,
      "learning_rate": 1.9658583623624526e-05,
      "loss": 0.7636,
      "step": 3218
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9319944381713867,
      "learning_rate": 1.9658294291812474e-05,
      "loss": 0.7761,
      "step": 3219
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3527748584747314,
      "learning_rate": 1.965800483958684e-05,
      "loss": 0.7693,
      "step": 3220
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5663522481918335,
      "learning_rate": 1.965771526695124e-05,
      "loss": 0.8263,
      "step": 3221
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.0621652603149414,
      "learning_rate": 1.9657425573909275e-05,
      "loss": 0.6488,
      "step": 3222
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.3093109130859375,
      "learning_rate": 1.965713576046456e-05,
      "loss": 0.8728,
      "step": 3223
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9368726015090942,
      "learning_rate": 1.965684582662071e-05,
      "loss": 0.7386,
      "step": 3224
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7148586511611938,
      "learning_rate": 1.9656555772381343e-05,
      "loss": 0.6766,
      "step": 3225
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6819090843200684,
      "learning_rate": 1.9656265597750064e-05,
      "loss": 0.9077,
      "step": 3226
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7640976905822754,
      "learning_rate": 1.96559753027305e-05,
      "loss": 0.863,
      "step": 3227
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8635547161102295,
      "learning_rate": 1.965568488732627e-05,
      "loss": 0.7534,
      "step": 3228
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.499746084213257,
      "learning_rate": 1.965539435154099e-05,
      "loss": 0.915,
      "step": 3229
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6881024837493896,
      "learning_rate": 1.9655103695378284e-05,
      "loss": 0.708,
      "step": 3230
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.516518473625183,
      "learning_rate": 1.9654812918841778e-05,
      "loss": 0.631,
      "step": 3231
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6578985452651978,
      "learning_rate": 1.9654522021935094e-05,
      "loss": 0.689,
      "step": 3232
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.916333794593811,
      "learning_rate": 1.965423100466186e-05,
      "loss": 0.8097,
      "step": 3233
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.1931746006011963,
      "learning_rate": 1.9653939867025705e-05,
      "loss": 0.9701,
      "step": 3234
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9710789918899536,
      "learning_rate": 1.965364860903026e-05,
      "loss": 0.9127,
      "step": 3235
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.950551152229309,
      "learning_rate": 1.9653357230679153e-05,
      "loss": 0.8934,
      "step": 3236
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.544278621673584,
      "learning_rate": 1.965306573197602e-05,
      "loss": 0.7646,
      "step": 3237
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.866336464881897,
      "learning_rate": 1.965277411292449e-05,
      "loss": 0.8104,
      "step": 3238
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.639955759048462,
      "learning_rate": 1.9652482373528204e-05,
      "loss": 0.9058,
      "step": 3239
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7551733255386353,
      "learning_rate": 1.9652190513790796e-05,
      "loss": 0.7213,
      "step": 3240
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.981408715248108,
      "learning_rate": 1.9651898533715906e-05,
      "loss": 0.6181,
      "step": 3241
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4351857900619507,
      "learning_rate": 1.9651606433307178e-05,
      "loss": 0.8006,
      "step": 3242
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.59320068359375,
      "learning_rate": 1.9651314212568245e-05,
      "loss": 0.7516,
      "step": 3243
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0092484951019287,
      "learning_rate": 1.9651021871502756e-05,
      "loss": 0.8785,
      "step": 3244
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4660893678665161,
      "learning_rate": 1.9650729410114357e-05,
      "loss": 0.6563,
      "step": 3245
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.672434687614441,
      "learning_rate": 1.9650436828406695e-05,
      "loss": 0.6599,
      "step": 3246
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7095476388931274,
      "learning_rate": 1.965014412638341e-05,
      "loss": 0.7959,
      "step": 3247
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6366535425186157,
      "learning_rate": 1.9649851304048157e-05,
      "loss": 0.693,
      "step": 3248
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.684761643409729,
      "learning_rate": 1.9649558361404585e-05,
      "loss": 0.8759,
      "step": 3249
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7268576622009277,
      "learning_rate": 1.964926529845635e-05,
      "loss": 0.8217,
      "step": 3250
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7288610935211182,
      "learning_rate": 1.96489721152071e-05,
      "loss": 0.8344,
      "step": 3251
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.84888756275177,
      "learning_rate": 1.9648678811660496e-05,
      "loss": 0.8791,
      "step": 3252
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7826762199401855,
      "learning_rate": 1.964838538782019e-05,
      "loss": 0.8614,
      "step": 3253
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8998576402664185,
      "learning_rate": 1.9648091843689845e-05,
      "loss": 0.7656,
      "step": 3254
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7716989517211914,
      "learning_rate": 1.9647798179273114e-05,
      "loss": 0.6792,
      "step": 3255
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9516624212265015,
      "learning_rate": 1.964750439457366e-05,
      "loss": 0.762,
      "step": 3256
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.879123330116272,
      "learning_rate": 1.964721048959515e-05,
      "loss": 0.6981,
      "step": 3257
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8337126970291138,
      "learning_rate": 1.964691646434125e-05,
      "loss": 0.6936,
      "step": 3258
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8314685821533203,
      "learning_rate": 1.9646622318815618e-05,
      "loss": 0.6946,
      "step": 3259
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8007694482803345,
      "learning_rate": 1.9646328053021923e-05,
      "loss": 0.8148,
      "step": 3260
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.050302505493164,
      "learning_rate": 1.9646033666963837e-05,
      "loss": 0.8299,
      "step": 3261
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.765547513961792,
      "learning_rate": 1.964573916064503e-05,
      "loss": 1.1177,
      "step": 3262
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6349173784255981,
      "learning_rate": 1.964544453406917e-05,
      "loss": 0.678,
      "step": 3263
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9904729127883911,
      "learning_rate": 1.9645149787239933e-05,
      "loss": 0.767,
      "step": 3264
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9926297664642334,
      "learning_rate": 1.9644854920160998e-05,
      "loss": 0.7077,
      "step": 3265
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6450257301330566,
      "learning_rate": 1.9644559932836032e-05,
      "loss": 0.7069,
      "step": 3266
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8008019924163818,
      "learning_rate": 1.964426482526872e-05,
      "loss": 0.7975,
      "step": 3267
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5588321685791016,
      "learning_rate": 1.9643969597462735e-05,
      "loss": 0.5821,
      "step": 3268
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9232721328735352,
      "learning_rate": 1.9643674249421765e-05,
      "loss": 1.0066,
      "step": 3269
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6254112720489502,
      "learning_rate": 1.964337878114949e-05,
      "loss": 0.6661,
      "step": 3270
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6739712953567505,
      "learning_rate": 1.9643083192649585e-05,
      "loss": 0.5936,
      "step": 3271
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7611607313156128,
      "learning_rate": 1.964278748392575e-05,
      "loss": 0.9498,
      "step": 3272
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7584021091461182,
      "learning_rate": 1.964249165498166e-05,
      "loss": 0.5777,
      "step": 3273
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5355955362319946,
      "learning_rate": 1.9642195705821008e-05,
      "loss": 0.8461,
      "step": 3274
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.812778353691101,
      "learning_rate": 1.9641899636447485e-05,
      "loss": 0.8739,
      "step": 3275
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4745503664016724,
      "learning_rate": 1.9641603446864778e-05,
      "loss": 0.7312,
      "step": 3276
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6961511373519897,
      "learning_rate": 1.9641307137076582e-05,
      "loss": 0.7695,
      "step": 3277
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6931220293045044,
      "learning_rate": 1.964101070708659e-05,
      "loss": 0.7853,
      "step": 3278
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8803476095199585,
      "learning_rate": 1.9640714156898503e-05,
      "loss": 0.9196,
      "step": 3279
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6133273839950562,
      "learning_rate": 1.9640417486516012e-05,
      "loss": 0.6169,
      "step": 3280
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.817956566810608,
      "learning_rate": 1.9640120695942815e-05,
      "loss": 0.5357,
      "step": 3281
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.829474687576294,
      "learning_rate": 1.963982378518262e-05,
      "loss": 0.6869,
      "step": 3282
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9684362411499023,
      "learning_rate": 1.963952675423912e-05,
      "loss": 0.6554,
      "step": 3283
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1765692234039307,
      "learning_rate": 1.9639229603116023e-05,
      "loss": 0.6105,
      "step": 3284
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5776517391204834,
      "learning_rate": 1.9638932331817033e-05,
      "loss": 0.769,
      "step": 3285
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8289576768875122,
      "learning_rate": 1.9638634940345856e-05,
      "loss": 0.7106,
      "step": 3286
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7456979751586914,
      "learning_rate": 1.96383374287062e-05,
      "loss": 0.8716,
      "step": 3287
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7334327697753906,
      "learning_rate": 1.963803979690177e-05,
      "loss": 0.7599,
      "step": 3288
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3134055137634277,
      "learning_rate": 1.9637742044936282e-05,
      "loss": 0.953,
      "step": 3289
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.736097812652588,
      "learning_rate": 1.963744417281345e-05,
      "loss": 0.9056,
      "step": 3290
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.52665114402771,
      "learning_rate": 1.963714618053698e-05,
      "loss": 0.9976,
      "step": 3291
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5418908596038818,
      "learning_rate": 1.9636848068110593e-05,
      "loss": 0.9085,
      "step": 3292
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.612304925918579,
      "learning_rate": 1.9636549835538007e-05,
      "loss": 0.9151,
      "step": 3293
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6085635423660278,
      "learning_rate": 1.9636251482822932e-05,
      "loss": 0.7478,
      "step": 3294
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.438728928565979,
      "learning_rate": 1.9635953009969094e-05,
      "loss": 0.8615,
      "step": 3295
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5528626441955566,
      "learning_rate": 1.9635654416980215e-05,
      "loss": 0.7706,
      "step": 3296
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.171846866607666,
      "learning_rate": 1.9635355703860017e-05,
      "loss": 0.6934,
      "step": 3297
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.787734031677246,
      "learning_rate": 1.963505687061222e-05,
      "loss": 0.6626,
      "step": 3298
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7869001626968384,
      "learning_rate": 1.9634757917240557e-05,
      "loss": 0.7958,
      "step": 3299
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.832854986190796,
      "learning_rate": 1.9634458843748745e-05,
      "loss": 0.7852,
      "step": 3300
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0300581455230713,
      "learning_rate": 1.9634159650140522e-05,
      "loss": 0.7447,
      "step": 3301
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6049237251281738,
      "learning_rate": 1.9633860336419612e-05,
      "loss": 0.8123,
      "step": 3302
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.5739386081695557,
      "learning_rate": 1.9633560902589752e-05,
      "loss": 0.833,
      "step": 3303
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8352981805801392,
      "learning_rate": 1.9633261348654672e-05,
      "loss": 0.9095,
      "step": 3304
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9169795513153076,
      "learning_rate": 1.9632961674618107e-05,
      "loss": 0.7529,
      "step": 3305
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.056434392929077,
      "learning_rate": 1.9632661880483792e-05,
      "loss": 0.6406,
      "step": 3306
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.804818034172058,
      "learning_rate": 1.9632361966255466e-05,
      "loss": 0.6125,
      "step": 3307
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6907352209091187,
      "learning_rate": 1.9632061931936868e-05,
      "loss": 0.7545,
      "step": 3308
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7805914878845215,
      "learning_rate": 1.963176177753174e-05,
      "loss": 0.843,
      "step": 3309
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.796802043914795,
      "learning_rate": 1.963146150304382e-05,
      "loss": 0.8924,
      "step": 3310
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.930440068244934,
      "learning_rate": 1.9631161108476856e-05,
      "loss": 0.7676,
      "step": 3311
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.318499803543091,
      "learning_rate": 1.9630860593834592e-05,
      "loss": 0.8829,
      "step": 3312
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7987669706344604,
      "learning_rate": 1.9630559959120776e-05,
      "loss": 0.8028,
      "step": 3313
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6245238780975342,
      "learning_rate": 1.9630259204339152e-05,
      "loss": 0.6937,
      "step": 3314
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7240278720855713,
      "learning_rate": 1.962995832949347e-05,
      "loss": 0.6956,
      "step": 3315
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5220990180969238,
      "learning_rate": 1.9629657334587486e-05,
      "loss": 0.817,
      "step": 3316
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8019936084747314,
      "learning_rate": 1.962935621962495e-05,
      "loss": 1.0174,
      "step": 3317
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6924962997436523,
      "learning_rate": 1.9629054984609613e-05,
      "loss": 0.7428,
      "step": 3318
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.629791498184204,
      "learning_rate": 1.9628753629545233e-05,
      "loss": 0.4962,
      "step": 3319
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9608556032180786,
      "learning_rate": 1.9628452154435572e-05,
      "loss": 0.8205,
      "step": 3320
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5189783573150635,
      "learning_rate": 1.962815055928438e-05,
      "loss": 0.7604,
      "step": 3321
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6314704418182373,
      "learning_rate": 1.9627848844095423e-05,
      "loss": 0.6143,
      "step": 3322
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7339245080947876,
      "learning_rate": 1.962754700887246e-05,
      "loss": 0.76,
      "step": 3323
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6642065048217773,
      "learning_rate": 1.9627245053619256e-05,
      "loss": 0.8568,
      "step": 3324
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.81730318069458,
      "learning_rate": 1.9626942978339573e-05,
      "loss": 0.7369,
      "step": 3325
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6815509796142578,
      "learning_rate": 1.962664078303718e-05,
      "loss": 0.6244,
      "step": 3326
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5224835872650146,
      "learning_rate": 1.962633846771584e-05,
      "loss": 0.7223,
      "step": 3327
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5952715873718262,
      "learning_rate": 1.962603603237933e-05,
      "loss": 0.7606,
      "step": 3328
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7873684167861938,
      "learning_rate": 1.9625733477031413e-05,
      "loss": 0.8713,
      "step": 3329
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.852425217628479,
      "learning_rate": 1.9625430801675863e-05,
      "loss": 0.6739,
      "step": 3330
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4672552347183228,
      "learning_rate": 1.962512800631646e-05,
      "loss": 0.7632,
      "step": 3331
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4726265668869019,
      "learning_rate": 1.9624825090956963e-05,
      "loss": 0.6419,
      "step": 3332
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.3635079860687256,
      "learning_rate": 1.962452205560117e-05,
      "loss": 0.7237,
      "step": 3333
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4186371564865112,
      "learning_rate": 1.9624218900252844e-05,
      "loss": 0.7461,
      "step": 3334
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.2246382236480713,
      "learning_rate": 1.9623915624915766e-05,
      "loss": 0.7826,
      "step": 3335
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7183818817138672,
      "learning_rate": 1.962361222959372e-05,
      "loss": 0.7638,
      "step": 3336
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9723581075668335,
      "learning_rate": 1.962330871429049e-05,
      "loss": 1.0377,
      "step": 3337
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.566063404083252,
      "learning_rate": 1.9623005079009856e-05,
      "loss": 0.6418,
      "step": 3338
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.189368963241577,
      "learning_rate": 1.962270132375561e-05,
      "loss": 0.7778,
      "step": 3339
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5164207220077515,
      "learning_rate": 1.9622397448531534e-05,
      "loss": 0.7196,
      "step": 3340
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.652667760848999,
      "learning_rate": 1.9622093453341416e-05,
      "loss": 0.6705,
      "step": 3341
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.4763591289520264,
      "learning_rate": 1.9621789338189047e-05,
      "loss": 0.7703,
      "step": 3342
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9017088413238525,
      "learning_rate": 1.962148510307822e-05,
      "loss": 0.9023,
      "step": 3343
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.541650414466858,
      "learning_rate": 1.9621180748012725e-05,
      "loss": 0.6487,
      "step": 3344
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8777143955230713,
      "learning_rate": 1.9620876272996362e-05,
      "loss": 0.8084,
      "step": 3345
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6398566961288452,
      "learning_rate": 1.962057167803292e-05,
      "loss": 0.7619,
      "step": 3346
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6730637550354004,
      "learning_rate": 1.96202669631262e-05,
      "loss": 0.8561,
      "step": 3347
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.354451298713684,
      "learning_rate": 1.9619962128280004e-05,
      "loss": 0.7066,
      "step": 3348
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6295809745788574,
      "learning_rate": 1.9619657173498128e-05,
      "loss": 0.7838,
      "step": 3349
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.718684196472168,
      "learning_rate": 1.9619352098784373e-05,
      "loss": 0.8847,
      "step": 3350
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6676307916641235,
      "learning_rate": 1.961904690414255e-05,
      "loss": 0.7025,
      "step": 3351
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8389077186584473,
      "learning_rate": 1.9618741589576454e-05,
      "loss": 0.7192,
      "step": 3352
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5548322200775146,
      "learning_rate": 1.96184361550899e-05,
      "loss": 0.9587,
      "step": 3353
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.85719895362854,
      "learning_rate": 1.9618130600686692e-05,
      "loss": 0.7747,
      "step": 3354
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9622972011566162,
      "learning_rate": 1.9617824926370637e-05,
      "loss": 0.7843,
      "step": 3355
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.083423137664795,
      "learning_rate": 1.9617519132145548e-05,
      "loss": 0.8573,
      "step": 3356
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6484272480010986,
      "learning_rate": 1.961721321801524e-05,
      "loss": 0.607,
      "step": 3357
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8859508037567139,
      "learning_rate": 1.9616907183983526e-05,
      "loss": 0.9376,
      "step": 3358
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.614203929901123,
      "learning_rate": 1.961660103005422e-05,
      "loss": 0.7508,
      "step": 3359
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7798181772232056,
      "learning_rate": 1.9616294756231138e-05,
      "loss": 0.7943,
      "step": 3360
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.873949408531189,
      "learning_rate": 1.96159883625181e-05,
      "loss": 0.7558,
      "step": 3361
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7495445013046265,
      "learning_rate": 1.9615681848918928e-05,
      "loss": 0.7827,
      "step": 3362
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8116846084594727,
      "learning_rate": 1.9615375215437443e-05,
      "loss": 0.8246,
      "step": 3363
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.526886224746704,
      "learning_rate": 1.9615068462077463e-05,
      "loss": 0.851,
      "step": 3364
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8319329023361206,
      "learning_rate": 1.9614761588842818e-05,
      "loss": 0.8158,
      "step": 3365
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5130674839019775,
      "learning_rate": 1.9614454595737327e-05,
      "loss": 0.8003,
      "step": 3366
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7643163204193115,
      "learning_rate": 1.9614147482764823e-05,
      "loss": 0.6637,
      "step": 3367
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.933364748954773,
      "learning_rate": 1.9613840249929137e-05,
      "loss": 0.8731,
      "step": 3368
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1047964096069336,
      "learning_rate": 1.961353289723409e-05,
      "loss": 0.7707,
      "step": 3369
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.180872678756714,
      "learning_rate": 1.961322542468353e-05,
      "loss": 0.7111,
      "step": 3370
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8516205549240112,
      "learning_rate": 1.9612917832281275e-05,
      "loss": 0.7824,
      "step": 3371
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0867409706115723,
      "learning_rate": 1.9612610120031165e-05,
      "loss": 0.7746,
      "step": 3372
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4900659322738647,
      "learning_rate": 1.9612302287937037e-05,
      "loss": 0.7441,
      "step": 3373
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9850051403045654,
      "learning_rate": 1.961199433600273e-05,
      "loss": 0.7568,
      "step": 3374
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8580080270767212,
      "learning_rate": 1.961168626423208e-05,
      "loss": 0.7031,
      "step": 3375
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9465515613555908,
      "learning_rate": 1.9611378072628928e-05,
      "loss": 0.8975,
      "step": 3376
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8207257986068726,
      "learning_rate": 1.961106976119712e-05,
      "loss": 0.709,
      "step": 3377
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7907376289367676,
      "learning_rate": 1.96107613299405e-05,
      "loss": 0.689,
      "step": 3378
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6303592920303345,
      "learning_rate": 1.961045277886291e-05,
      "loss": 0.686,
      "step": 3379
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6594947576522827,
      "learning_rate": 1.9610144107968194e-05,
      "loss": 0.8657,
      "step": 3380
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6185487508773804,
      "learning_rate": 1.9609835317260208e-05,
      "loss": 0.8125,
      "step": 3381
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.908667802810669,
      "learning_rate": 1.96095264067428e-05,
      "loss": 0.8446,
      "step": 3382
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5962034463882446,
      "learning_rate": 1.9609217376419817e-05,
      "loss": 0.6845,
      "step": 3383
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9899405241012573,
      "learning_rate": 1.9608908226295114e-05,
      "loss": 0.8739,
      "step": 3384
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9194109439849854,
      "learning_rate": 1.9608598956372546e-05,
      "loss": 0.7965,
      "step": 3385
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6804783344268799,
      "learning_rate": 1.9608289566655968e-05,
      "loss": 0.8541,
      "step": 3386
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7237682342529297,
      "learning_rate": 1.960798005714924e-05,
      "loss": 0.621,
      "step": 3387
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.686549425125122,
      "learning_rate": 1.9607670427856216e-05,
      "loss": 0.807,
      "step": 3388
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8754092454910278,
      "learning_rate": 1.960736067878076e-05,
      "loss": 0.7014,
      "step": 3389
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.867842435836792,
      "learning_rate": 1.960705080992673e-05,
      "loss": 0.7489,
      "step": 3390
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.323524236679077,
      "learning_rate": 1.9606740821297995e-05,
      "loss": 0.904,
      "step": 3391
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9021326303482056,
      "learning_rate": 1.9606430712898414e-05,
      "loss": 0.8409,
      "step": 3392
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3781421184539795,
      "learning_rate": 1.9606120484731856e-05,
      "loss": 0.7215,
      "step": 3393
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9511884450912476,
      "learning_rate": 1.960581013680219e-05,
      "loss": 0.8701,
      "step": 3394
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0458669662475586,
      "learning_rate": 1.9605499669113285e-05,
      "loss": 0.7279,
      "step": 3395
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.0532288551330566,
      "learning_rate": 1.9605189081669006e-05,
      "loss": 0.723,
      "step": 3396
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5406090021133423,
      "learning_rate": 1.960487837447323e-05,
      "loss": 0.6802,
      "step": 3397
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8067890405654907,
      "learning_rate": 1.960456754752983e-05,
      "loss": 0.7618,
      "step": 3398
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7042293548583984,
      "learning_rate": 1.9604256600842687e-05,
      "loss": 0.7752,
      "step": 3399
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.630698323249817,
      "learning_rate": 1.960394553441567e-05,
      "loss": 0.7494,
      "step": 3400
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7335152626037598,
      "learning_rate": 1.9603634348252652e-05,
      "loss": 0.7477,
      "step": 3401
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5188249349594116,
      "learning_rate": 1.9603323042357528e-05,
      "loss": 0.6251,
      "step": 3402
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8041824102401733,
      "learning_rate": 1.9603011616734167e-05,
      "loss": 0.6113,
      "step": 3403
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.374772310256958,
      "learning_rate": 1.9602700071386457e-05,
      "loss": 1.1392,
      "step": 3404
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.382141351699829,
      "learning_rate": 1.9602388406318284e-05,
      "loss": 0.9753,
      "step": 3405
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7861984968185425,
      "learning_rate": 1.9602076621533526e-05,
      "loss": 0.8045,
      "step": 3406
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.3182897567749023,
      "learning_rate": 1.9601764717036078e-05,
      "loss": 0.7302,
      "step": 3407
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7572344541549683,
      "learning_rate": 1.9601452692829824e-05,
      "loss": 0.8171,
      "step": 3408
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7186740636825562,
      "learning_rate": 1.9601140548918655e-05,
      "loss": 0.8608,
      "step": 3409
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7100210189819336,
      "learning_rate": 1.9600828285306464e-05,
      "loss": 0.7829,
      "step": 3410
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3296170234680176,
      "learning_rate": 1.9600515901997143e-05,
      "loss": 0.7524,
      "step": 3411
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6519228219985962,
      "learning_rate": 1.9600203398994586e-05,
      "loss": 0.7607,
      "step": 3412
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5252251625061035,
      "learning_rate": 1.9599890776302692e-05,
      "loss": 0.7709,
      "step": 3413
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5357469320297241,
      "learning_rate": 1.9599578033925354e-05,
      "loss": 0.7248,
      "step": 3414
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5622870922088623,
      "learning_rate": 1.9599265171866476e-05,
      "loss": 0.7774,
      "step": 3415
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7400550842285156,
      "learning_rate": 1.9598952190129957e-05,
      "loss": 0.7788,
      "step": 3416
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8588701486587524,
      "learning_rate": 1.9598639088719697e-05,
      "loss": 0.7721,
      "step": 3417
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.24554181098938,
      "learning_rate": 1.95983258676396e-05,
      "loss": 0.9157,
      "step": 3418
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.848140835762024,
      "learning_rate": 1.959801252689357e-05,
      "loss": 0.7439,
      "step": 3419
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6204519271850586,
      "learning_rate": 1.9597699066485514e-05,
      "loss": 0.7679,
      "step": 3420
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.724379062652588,
      "learning_rate": 1.9597385486419348e-05,
      "loss": 0.599,
      "step": 3421
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.2824416160583496,
      "learning_rate": 1.959707178669897e-05,
      "loss": 0.8854,
      "step": 3422
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7687488794326782,
      "learning_rate": 1.95967579673283e-05,
      "loss": 0.6355,
      "step": 3423
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.9040007591247559,
      "learning_rate": 1.959644402831124e-05,
      "loss": 0.9041,
      "step": 3424
    },
    {
      "epoch": 0.35,
      "grad_norm": 2.1394295692443848,
      "learning_rate": 1.9596129969651717e-05,
      "loss": 0.8714,
      "step": 3425
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6789110898971558,
      "learning_rate": 1.9595815791353637e-05,
      "loss": 0.7963,
      "step": 3426
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7342301607131958,
      "learning_rate": 1.959550149342092e-05,
      "loss": 0.7079,
      "step": 3427
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.508315086364746,
      "learning_rate": 1.9595187075857482e-05,
      "loss": 0.7395,
      "step": 3428
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7058237791061401,
      "learning_rate": 1.959487253866725e-05,
      "loss": 0.7445,
      "step": 3429
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6818969249725342,
      "learning_rate": 1.9594557881854136e-05,
      "loss": 0.7168,
      "step": 3430
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6836248636245728,
      "learning_rate": 1.959424310542207e-05,
      "loss": 0.7318,
      "step": 3431
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6825968027114868,
      "learning_rate": 1.9593928209374976e-05,
      "loss": 0.842,
      "step": 3432
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.828032374382019,
      "learning_rate": 1.9593613193716778e-05,
      "loss": 0.8391,
      "step": 3433
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6695376634597778,
      "learning_rate": 1.95932980584514e-05,
      "loss": 0.7754,
      "step": 3434
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7985514402389526,
      "learning_rate": 1.9592982803582777e-05,
      "loss": 0.7742,
      "step": 3435
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4882051944732666,
      "learning_rate": 1.9592667429114832e-05,
      "loss": 0.7523,
      "step": 3436
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6885838508605957,
      "learning_rate": 1.9592351935051507e-05,
      "loss": 0.7261,
      "step": 3437
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.603201150894165,
      "learning_rate": 1.9592036321396728e-05,
      "loss": 0.7456,
      "step": 3438
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.973539113998413,
      "learning_rate": 1.9591720588154436e-05,
      "loss": 0.7725,
      "step": 3439
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8688150644302368,
      "learning_rate": 1.9591404735328557e-05,
      "loss": 0.7237,
      "step": 3440
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.848151445388794,
      "learning_rate": 1.959108876292304e-05,
      "loss": 0.7483,
      "step": 3441
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5100573301315308,
      "learning_rate": 1.9590772670941817e-05,
      "loss": 0.6954,
      "step": 3442
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5526553392410278,
      "learning_rate": 1.9590456459388832e-05,
      "loss": 0.8165,
      "step": 3443
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.840682864189148,
      "learning_rate": 1.9590140128268026e-05,
      "loss": 0.7352,
      "step": 3444
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.05234694480896,
      "learning_rate": 1.9589823677583345e-05,
      "loss": 0.5401,
      "step": 3445
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8837615251541138,
      "learning_rate": 1.958950710733873e-05,
      "loss": 0.7831,
      "step": 3446
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.169928550720215,
      "learning_rate": 1.9589190417538133e-05,
      "loss": 0.9099,
      "step": 3447
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7528936862945557,
      "learning_rate": 1.95888736081855e-05,
      "loss": 0.8366,
      "step": 3448
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8184678554534912,
      "learning_rate": 1.9588556679284778e-05,
      "loss": 0.8404,
      "step": 3449
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6985522508621216,
      "learning_rate": 1.9588239630839924e-05,
      "loss": 0.8381,
      "step": 3450
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8845183849334717,
      "learning_rate": 1.9587922462854885e-05,
      "loss": 0.692,
      "step": 3451
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7126871347427368,
      "learning_rate": 1.958760517533362e-05,
      "loss": 0.8139,
      "step": 3452
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9382468461990356,
      "learning_rate": 1.9587287768280082e-05,
      "loss": 0.7964,
      "step": 3453
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8518624305725098,
      "learning_rate": 1.958697024169823e-05,
      "loss": 0.5906,
      "step": 3454
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0457262992858887,
      "learning_rate": 1.9586652595592022e-05,
      "loss": 0.8617,
      "step": 3455
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8326528072357178,
      "learning_rate": 1.9586334829965415e-05,
      "loss": 0.8355,
      "step": 3456
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8175634145736694,
      "learning_rate": 1.9586016944822378e-05,
      "loss": 0.7383,
      "step": 3457
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.80479097366333,
      "learning_rate": 1.9585698940166865e-05,
      "loss": 0.838,
      "step": 3458
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.2033047676086426,
      "learning_rate": 1.9585380816002846e-05,
      "loss": 0.7355,
      "step": 3459
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.516717791557312,
      "learning_rate": 1.9585062572334293e-05,
      "loss": 0.8618,
      "step": 3460
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6205657720565796,
      "learning_rate": 1.958474420916516e-05,
      "loss": 0.7781,
      "step": 3461
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7808085680007935,
      "learning_rate": 1.9584425726499424e-05,
      "loss": 0.7267,
      "step": 3462
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7377837896347046,
      "learning_rate": 1.958410712434106e-05,
      "loss": 0.7022,
      "step": 3463
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7232606410980225,
      "learning_rate": 1.9583788402694033e-05,
      "loss": 0.7875,
      "step": 3464
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.006612777709961,
      "learning_rate": 1.9583469561562317e-05,
      "loss": 0.8354,
      "step": 3465
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6363810300827026,
      "learning_rate": 1.958315060094989e-05,
      "loss": 0.7906,
      "step": 3466
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.594934344291687,
      "learning_rate": 1.958283152086073e-05,
      "loss": 0.6886,
      "step": 3467
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.667445421218872,
      "learning_rate": 1.9582512321298808e-05,
      "loss": 0.7633,
      "step": 3468
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1099281311035156,
      "learning_rate": 1.9582193002268108e-05,
      "loss": 0.8566,
      "step": 3469
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7426085472106934,
      "learning_rate": 1.9581873563772616e-05,
      "loss": 0.7989,
      "step": 3470
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8800346851348877,
      "learning_rate": 1.9581554005816307e-05,
      "loss": 0.7629,
      "step": 3471
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.803578495979309,
      "learning_rate": 1.958123432840317e-05,
      "loss": 0.8055,
      "step": 3472
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9935272932052612,
      "learning_rate": 1.958091453153719e-05,
      "loss": 0.8693,
      "step": 3473
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6878007650375366,
      "learning_rate": 1.9580594615222344e-05,
      "loss": 0.8097,
      "step": 3474
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4614936113357544,
      "learning_rate": 1.9580274579462637e-05,
      "loss": 0.6719,
      "step": 3475
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8164435625076294,
      "learning_rate": 1.9579954424262045e-05,
      "loss": 0.9049,
      "step": 3476
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8715310096740723,
      "learning_rate": 1.957963414962457e-05,
      "loss": 0.8772,
      "step": 3477
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5650622844696045,
      "learning_rate": 1.9579313755554202e-05,
      "loss": 0.9778,
      "step": 3478
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.1974551677703857,
      "learning_rate": 1.9578993242054933e-05,
      "loss": 0.8324,
      "step": 3479
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.843766689300537,
      "learning_rate": 1.9578672609130757e-05,
      "loss": 0.6666,
      "step": 3480
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.74419105052948,
      "learning_rate": 1.9578351856785672e-05,
      "loss": 0.7798,
      "step": 3481
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.741714596748352,
      "learning_rate": 1.9578030985023684e-05,
      "loss": 0.802,
      "step": 3482
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7907304763793945,
      "learning_rate": 1.9577709993848787e-05,
      "loss": 0.8342,
      "step": 3483
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.098421096801758,
      "learning_rate": 1.957738888326499e-05,
      "loss": 0.7425,
      "step": 3484
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7943044900894165,
      "learning_rate": 1.9577067653276284e-05,
      "loss": 0.7931,
      "step": 3485
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.735897421836853,
      "learning_rate": 1.957674630388668e-05,
      "loss": 0.7876,
      "step": 3486
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7823528051376343,
      "learning_rate": 1.957642483510019e-05,
      "loss": 0.8523,
      "step": 3487
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.583064317703247,
      "learning_rate": 1.9576103246920816e-05,
      "loss": 0.5716,
      "step": 3488
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6179333925247192,
      "learning_rate": 1.957578153935257e-05,
      "loss": 0.8384,
      "step": 3489
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.565696120262146,
      "learning_rate": 1.957545971239946e-05,
      "loss": 0.7738,
      "step": 3490
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9373128414154053,
      "learning_rate": 1.9575137766065498e-05,
      "loss": 0.948,
      "step": 3491
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5276292562484741,
      "learning_rate": 1.95748157003547e-05,
      "loss": 0.709,
      "step": 3492
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7397792339324951,
      "learning_rate": 1.9574493515271082e-05,
      "loss": 0.6693,
      "step": 3493
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7700676918029785,
      "learning_rate": 1.957417121081866e-05,
      "loss": 0.8764,
      "step": 3494
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6845977306365967,
      "learning_rate": 1.957384878700145e-05,
      "loss": 0.9143,
      "step": 3495
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0145204067230225,
      "learning_rate": 1.9573526243823478e-05,
      "loss": 0.8966,
      "step": 3496
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.031588315963745,
      "learning_rate": 1.9573203581288756e-05,
      "loss": 0.737,
      "step": 3497
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7162351608276367,
      "learning_rate": 1.9572880799401315e-05,
      "loss": 0.8153,
      "step": 3498
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6939202547073364,
      "learning_rate": 1.9572557898165177e-05,
      "loss": 0.7459,
      "step": 3499
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6017637252807617,
      "learning_rate": 1.9572234877584364e-05,
      "loss": 0.8382,
      "step": 3500
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8175877332687378,
      "learning_rate": 1.9571911737662906e-05,
      "loss": 0.8597,
      "step": 3501
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.782923936843872,
      "learning_rate": 1.9571588478404833e-05,
      "loss": 0.8227,
      "step": 3502
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.527007818222046,
      "learning_rate": 1.9571265099814174e-05,
      "loss": 0.7831,
      "step": 3503
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.726800799369812,
      "learning_rate": 1.9570941601894963e-05,
      "loss": 0.6334,
      "step": 3504
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6961171627044678,
      "learning_rate": 1.9570617984651225e-05,
      "loss": 0.7509,
      "step": 3505
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4595881700515747,
      "learning_rate": 1.9570294248087004e-05,
      "loss": 0.6955,
      "step": 3506
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.00007963180542,
      "learning_rate": 1.956997039220633e-05,
      "loss": 0.7982,
      "step": 3507
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9259649515151978,
      "learning_rate": 1.9569646417013245e-05,
      "loss": 0.8655,
      "step": 3508
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6628601551055908,
      "learning_rate": 1.9569322322511785e-05,
      "loss": 0.6788,
      "step": 3509
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7160539627075195,
      "learning_rate": 1.9568998108705996e-05,
      "loss": 0.8711,
      "step": 3510
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6418862342834473,
      "learning_rate": 1.9568673775599912e-05,
      "loss": 0.7396,
      "step": 3511
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4887611865997314,
      "learning_rate": 1.9568349323197582e-05,
      "loss": 0.6365,
      "step": 3512
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7224420309066772,
      "learning_rate": 1.956802475150305e-05,
      "loss": 0.7447,
      "step": 3513
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.711634635925293,
      "learning_rate": 1.956770006052036e-05,
      "loss": 0.6931,
      "step": 3514
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4364651441574097,
      "learning_rate": 1.9567375250253563e-05,
      "loss": 0.8342,
      "step": 3515
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6489412784576416,
      "learning_rate": 1.956705032070671e-05,
      "loss": 0.9092,
      "step": 3516
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7006504535675049,
      "learning_rate": 1.956672527188385e-05,
      "loss": 0.7925,
      "step": 3517
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8703134059906006,
      "learning_rate": 1.956640010378903e-05,
      "loss": 0.8897,
      "step": 3518
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.28598952293396,
      "learning_rate": 1.9566074816426314e-05,
      "loss": 0.7528,
      "step": 3519
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5779094696044922,
      "learning_rate": 1.956574940979975e-05,
      "loss": 0.7506,
      "step": 3520
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8003618717193604,
      "learning_rate": 1.95654238839134e-05,
      "loss": 0.7541,
      "step": 3521
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.0630133152008057,
      "learning_rate": 1.956509823877132e-05,
      "loss": 0.7869,
      "step": 3522
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4102489948272705,
      "learning_rate": 1.956477247437757e-05,
      "loss": 0.745,
      "step": 3523
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.7867660522460938,
      "learning_rate": 1.956444659073621e-05,
      "loss": 0.7435,
      "step": 3524
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.626665711402893,
      "learning_rate": 1.9564120587851304e-05,
      "loss": 0.9113,
      "step": 3525
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.527886152267456,
      "learning_rate": 1.9563794465726918e-05,
      "loss": 0.8323,
      "step": 3526
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8358213901519775,
      "learning_rate": 1.956346822436712e-05,
      "loss": 0.7976,
      "step": 3527
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8673814535140991,
      "learning_rate": 1.956314186377597e-05,
      "loss": 0.8903,
      "step": 3528
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.9281644821166992,
      "learning_rate": 1.9562815383957544e-05,
      "loss": 0.7147,
      "step": 3529
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.595461368560791,
      "learning_rate": 1.9562488784915906e-05,
      "loss": 0.749,
      "step": 3530
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6790870428085327,
      "learning_rate": 1.9562162066655133e-05,
      "loss": 0.8135,
      "step": 3531
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.857643723487854,
      "learning_rate": 1.9561835229179298e-05,
      "loss": 0.7419,
      "step": 3532
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8413972854614258,
      "learning_rate": 1.9561508272492472e-05,
      "loss": 0.792,
      "step": 3533
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3710626363754272,
      "learning_rate": 1.9561181196598732e-05,
      "loss": 0.7611,
      "step": 3534
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8103147745132446,
      "learning_rate": 1.956085400150216e-05,
      "loss": 0.6768,
      "step": 3535
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3973259925842285,
      "learning_rate": 1.9560526687206837e-05,
      "loss": 0.7868,
      "step": 3536
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.94424569606781,
      "learning_rate": 1.956019925371683e-05,
      "loss": 0.8524,
      "step": 3537
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0202484130859375,
      "learning_rate": 1.9559871701036237e-05,
      "loss": 0.703,
      "step": 3538
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7188509702682495,
      "learning_rate": 1.9559544029169136e-05,
      "loss": 0.6966,
      "step": 3539
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6181186437606812,
      "learning_rate": 1.9559216238119608e-05,
      "loss": 0.9039,
      "step": 3540
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7073345184326172,
      "learning_rate": 1.9558888327891743e-05,
      "loss": 0.7886,
      "step": 3541
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5573809146881104,
      "learning_rate": 1.955856029848963e-05,
      "loss": 0.7007,
      "step": 3542
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6037307977676392,
      "learning_rate": 1.955823214991736e-05,
      "loss": 0.7298,
      "step": 3543
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.576297402381897,
      "learning_rate": 1.955790388217902e-05,
      "loss": 0.7256,
      "step": 3544
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5569815635681152,
      "learning_rate": 1.9557575495278706e-05,
      "loss": 0.7325,
      "step": 3545
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6670001745224,
      "learning_rate": 1.955724698922051e-05,
      "loss": 0.7802,
      "step": 3546
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.021949052810669,
      "learning_rate": 1.955691836400853e-05,
      "loss": 0.7391,
      "step": 3547
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.262631893157959,
      "learning_rate": 1.9556589619646863e-05,
      "loss": 0.7549,
      "step": 3548
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7114518880844116,
      "learning_rate": 1.9556260756139605e-05,
      "loss": 0.7526,
      "step": 3549
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6020500659942627,
      "learning_rate": 1.955593177349086e-05,
      "loss": 0.6577,
      "step": 3550
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.657677412033081,
      "learning_rate": 1.955560267170472e-05,
      "loss": 0.8679,
      "step": 3551
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6690412759780884,
      "learning_rate": 1.95552734507853e-05,
      "loss": 0.7762,
      "step": 3552
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.553146481513977,
      "learning_rate": 1.95549441107367e-05,
      "loss": 0.8044,
      "step": 3553
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9631736278533936,
      "learning_rate": 1.955461465156302e-05,
      "loss": 0.9011,
      "step": 3554
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.046109437942505,
      "learning_rate": 1.9554285073268376e-05,
      "loss": 0.8837,
      "step": 3555
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7719228267669678,
      "learning_rate": 1.9553955375856874e-05,
      "loss": 0.8129,
      "step": 3556
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1820850372314453,
      "learning_rate": 1.9553625559332627e-05,
      "loss": 0.8737,
      "step": 3557
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5794342756271362,
      "learning_rate": 1.9553295623699742e-05,
      "loss": 0.7457,
      "step": 3558
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4704620838165283,
      "learning_rate": 1.9552965568962336e-05,
      "loss": 0.7376,
      "step": 3559
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9071348905563354,
      "learning_rate": 1.955263539512452e-05,
      "loss": 0.9857,
      "step": 3560
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.74660325050354,
      "learning_rate": 1.9552305102190415e-05,
      "loss": 0.6687,
      "step": 3561
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6118860244750977,
      "learning_rate": 1.9551974690164133e-05,
      "loss": 0.768,
      "step": 3562
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5074124336242676,
      "learning_rate": 1.9551644159049804e-05,
      "loss": 0.6568,
      "step": 3563
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7617195844650269,
      "learning_rate": 1.955131350885154e-05,
      "loss": 0.7093,
      "step": 3564
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7074482440948486,
      "learning_rate": 1.9550982739573462e-05,
      "loss": 0.8156,
      "step": 3565
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6623785495758057,
      "learning_rate": 1.95506518512197e-05,
      "loss": 0.7837,
      "step": 3566
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8410345315933228,
      "learning_rate": 1.955032084379438e-05,
      "loss": 0.7408,
      "step": 3567
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6873046159744263,
      "learning_rate": 1.954998971730162e-05,
      "loss": 0.9308,
      "step": 3568
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7048488855361938,
      "learning_rate": 1.9549658471745556e-05,
      "loss": 0.642,
      "step": 3569
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.714964747428894,
      "learning_rate": 1.9549327107130315e-05,
      "loss": 0.8534,
      "step": 3570
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.667565941810608,
      "learning_rate": 1.954899562346003e-05,
      "loss": 0.7939,
      "step": 3571
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7801587581634521,
      "learning_rate": 1.9548664020738832e-05,
      "loss": 0.8375,
      "step": 3572
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.623132348060608,
      "learning_rate": 1.9548332298970854e-05,
      "loss": 0.8754,
      "step": 3573
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7618184089660645,
      "learning_rate": 1.9548000458160236e-05,
      "loss": 0.7283,
      "step": 3574
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8589354753494263,
      "learning_rate": 1.954766849831111e-05,
      "loss": 0.6745,
      "step": 3575
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8982077836990356,
      "learning_rate": 1.954733641942762e-05,
      "loss": 0.7965,
      "step": 3576
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5685791969299316,
      "learning_rate": 1.9547004221513904e-05,
      "loss": 0.752,
      "step": 3577
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.717860221862793,
      "learning_rate": 1.95466719045741e-05,
      "loss": 0.7611,
      "step": 3578
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.060321807861328,
      "learning_rate": 1.9546339468612355e-05,
      "loss": 0.7429,
      "step": 3579
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.494253396987915,
      "learning_rate": 1.9546006913632812e-05,
      "loss": 0.5741,
      "step": 3580
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0707027912139893,
      "learning_rate": 1.9545674239639618e-05,
      "loss": 0.7737,
      "step": 3581
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7303638458251953,
      "learning_rate": 1.9545341446636922e-05,
      "loss": 0.5882,
      "step": 3582
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.073326349258423,
      "learning_rate": 1.9545008534628872e-05,
      "loss": 0.9104,
      "step": 3583
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7215455770492554,
      "learning_rate": 1.9544675503619615e-05,
      "loss": 0.9214,
      "step": 3584
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0912389755249023,
      "learning_rate": 1.9544342353613304e-05,
      "loss": 0.7612,
      "step": 3585
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9826691150665283,
      "learning_rate": 1.95440090846141e-05,
      "loss": 0.8715,
      "step": 3586
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8408352136611938,
      "learning_rate": 1.9543675696626153e-05,
      "loss": 0.851,
      "step": 3587
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5691783428192139,
      "learning_rate": 1.9543342189653616e-05,
      "loss": 0.7362,
      "step": 3588
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1131911277770996,
      "learning_rate": 1.954300856370065e-05,
      "loss": 0.8228,
      "step": 3589
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5090680122375488,
      "learning_rate": 1.9542674818771416e-05,
      "loss": 0.735,
      "step": 3590
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5110913515090942,
      "learning_rate": 1.9542340954870074e-05,
      "loss": 0.8516,
      "step": 3591
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.508513331413269,
      "learning_rate": 1.9542006972000783e-05,
      "loss": 0.7796,
      "step": 3592
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3223540782928467,
      "learning_rate": 1.954167287016771e-05,
      "loss": 0.7753,
      "step": 3593
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5273475646972656,
      "learning_rate": 1.954133864937502e-05,
      "loss": 0.7886,
      "step": 3594
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6940513849258423,
      "learning_rate": 1.9541004309626882e-05,
      "loss": 0.7099,
      "step": 3595
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8535274267196655,
      "learning_rate": 1.9540669850927462e-05,
      "loss": 0.8278,
      "step": 3596
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.696101188659668,
      "learning_rate": 1.954033527328093e-05,
      "loss": 0.905,
      "step": 3597
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.003862142562866,
      "learning_rate": 1.9540000576691455e-05,
      "loss": 0.8885,
      "step": 3598
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.70115065574646,
      "learning_rate": 1.953966576116321e-05,
      "loss": 0.7142,
      "step": 3599
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.504999041557312,
      "learning_rate": 1.9539330826700375e-05,
      "loss": 0.8725,
      "step": 3600
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6708253622055054,
      "learning_rate": 1.9538995773307122e-05,
      "loss": 0.9163,
      "step": 3601
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8571611642837524,
      "learning_rate": 1.953866060098763e-05,
      "loss": 0.7556,
      "step": 3602
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7981640100479126,
      "learning_rate": 1.9538325309746075e-05,
      "loss": 0.7548,
      "step": 3603
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5952671766281128,
      "learning_rate": 1.9537989899586637e-05,
      "loss": 0.6629,
      "step": 3604
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5326741933822632,
      "learning_rate": 1.9537654370513497e-05,
      "loss": 0.8609,
      "step": 3605
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6036850214004517,
      "learning_rate": 1.9537318722530843e-05,
      "loss": 0.6669,
      "step": 3606
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.742053747177124,
      "learning_rate": 1.9536982955642856e-05,
      "loss": 0.7284,
      "step": 3607
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6935219764709473,
      "learning_rate": 1.9536647069853723e-05,
      "loss": 0.8407,
      "step": 3608
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7049946784973145,
      "learning_rate": 1.9536311065167633e-05,
      "loss": 0.936,
      "step": 3609
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7729239463806152,
      "learning_rate": 1.953597494158877e-05,
      "loss": 0.816,
      "step": 3610
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.731677770614624,
      "learning_rate": 1.953563869912133e-05,
      "loss": 0.6906,
      "step": 3611
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8294473886489868,
      "learning_rate": 1.9535302337769507e-05,
      "loss": 0.6422,
      "step": 3612
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8960366249084473,
      "learning_rate": 1.9534965857537487e-05,
      "loss": 0.8243,
      "step": 3613
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6597789525985718,
      "learning_rate": 1.9534629258429468e-05,
      "loss": 0.7599,
      "step": 3614
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.531407356262207,
      "learning_rate": 1.953429254044965e-05,
      "loss": 0.7824,
      "step": 3615
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.8597893714904785,
      "learning_rate": 1.953395570360223e-05,
      "loss": 0.8848,
      "step": 3616
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.961867332458496,
      "learning_rate": 1.95336187478914e-05,
      "loss": 0.8876,
      "step": 3617
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.508878231048584,
      "learning_rate": 1.953328167332137e-05,
      "loss": 0.7087,
      "step": 3618
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5929300785064697,
      "learning_rate": 1.953294447989634e-05,
      "loss": 0.7464,
      "step": 3619
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.784256100654602,
      "learning_rate": 1.953260716762051e-05,
      "loss": 0.926,
      "step": 3620
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.773704171180725,
      "learning_rate": 1.9532269736498093e-05,
      "loss": 0.6733,
      "step": 3621
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5198346376419067,
      "learning_rate": 1.9531932186533287e-05,
      "loss": 0.7709,
      "step": 3622
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7040891647338867,
      "learning_rate": 1.9531594517730307e-05,
      "loss": 0.926,
      "step": 3623
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7600619792938232,
      "learning_rate": 1.953125673009336e-05,
      "loss": 0.7793,
      "step": 3624
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4556655883789062,
      "learning_rate": 1.9530918823626658e-05,
      "loss": 0.687,
      "step": 3625
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7918150424957275,
      "learning_rate": 1.9530580798334412e-05,
      "loss": 0.7086,
      "step": 3626
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.101353406906128,
      "learning_rate": 1.953024265422084e-05,
      "loss": 0.686,
      "step": 3627
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7798948287963867,
      "learning_rate": 1.9529904391290158e-05,
      "loss": 0.6431,
      "step": 3628
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.772153377532959,
      "learning_rate": 1.952956600954658e-05,
      "loss": 0.8095,
      "step": 3629
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.663551688194275,
      "learning_rate": 1.9529227508994322e-05,
      "loss": 0.6278,
      "step": 3630
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.813283920288086,
      "learning_rate": 1.9528888889637613e-05,
      "loss": 0.8256,
      "step": 3631
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6965361833572388,
      "learning_rate": 1.9528550151480666e-05,
      "loss": 0.6316,
      "step": 3632
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.078922986984253,
      "learning_rate": 1.9528211294527712e-05,
      "loss": 0.7965,
      "step": 3633
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.3416905403137207,
      "learning_rate": 1.9527872318782967e-05,
      "loss": 0.809,
      "step": 3634
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6976597309112549,
      "learning_rate": 1.9527533224250664e-05,
      "loss": 0.7936,
      "step": 3635
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7770034074783325,
      "learning_rate": 1.9527194010935028e-05,
      "loss": 0.7197,
      "step": 3636
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8718723058700562,
      "learning_rate": 1.9526854678840283e-05,
      "loss": 0.7723,
      "step": 3637
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0374481678009033,
      "learning_rate": 1.952651522797067e-05,
      "loss": 0.8858,
      "step": 3638
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6850471496582031,
      "learning_rate": 1.9526175658330416e-05,
      "loss": 0.8788,
      "step": 3639
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.57758629322052,
      "learning_rate": 1.9525835969923754e-05,
      "loss": 0.658,
      "step": 3640
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9585864543914795,
      "learning_rate": 1.952549616275492e-05,
      "loss": 0.8446,
      "step": 3641
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.405428171157837,
      "learning_rate": 1.9525156236828152e-05,
      "loss": 0.896,
      "step": 3642
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6009299755096436,
      "learning_rate": 1.9524816192147683e-05,
      "loss": 0.7266,
      "step": 3643
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5586974620819092,
      "learning_rate": 1.9524476028717754e-05,
      "loss": 0.8143,
      "step": 3644
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.665805697441101,
      "learning_rate": 1.9524135746542613e-05,
      "loss": 0.6724,
      "step": 3645
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6923155784606934,
      "learning_rate": 1.9523795345626496e-05,
      "loss": 0.8212,
      "step": 3646
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.949000358581543,
      "learning_rate": 1.9523454825973643e-05,
      "loss": 0.8776,
      "step": 3647
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6057289838790894,
      "learning_rate": 1.952311418758831e-05,
      "loss": 0.6965,
      "step": 3648
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.479678988456726,
      "learning_rate": 1.9522773430474732e-05,
      "loss": 0.7416,
      "step": 3649
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4958490133285522,
      "learning_rate": 1.9522432554637166e-05,
      "loss": 0.8028,
      "step": 3650
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9262480735778809,
      "learning_rate": 1.952209156007986e-05,
      "loss": 0.852,
      "step": 3651
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.349868655204773,
      "learning_rate": 1.9521750446807064e-05,
      "loss": 0.7954,
      "step": 3652
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7947930097579956,
      "learning_rate": 1.952140921482303e-05,
      "loss": 0.8147,
      "step": 3653
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.576350450515747,
      "learning_rate": 1.9521067864132017e-05,
      "loss": 0.8164,
      "step": 3654
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8574225902557373,
      "learning_rate": 1.9520726394738274e-05,
      "loss": 0.94,
      "step": 3655
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5601054430007935,
      "learning_rate": 1.9520384806646064e-05,
      "loss": 0.8608,
      "step": 3656
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7659344673156738,
      "learning_rate": 1.9520043099859642e-05,
      "loss": 0.7164,
      "step": 3657
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7130565643310547,
      "learning_rate": 1.951970127438327e-05,
      "loss": 0.7138,
      "step": 3658
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7288206815719604,
      "learning_rate": 1.9519359330221208e-05,
      "loss": 0.7316,
      "step": 3659
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8969687223434448,
      "learning_rate": 1.951901726737772e-05,
      "loss": 0.8326,
      "step": 3660
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.625723123550415,
      "learning_rate": 1.9518675085857074e-05,
      "loss": 0.8106,
      "step": 3661
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6272531747817993,
      "learning_rate": 1.951833278566353e-05,
      "loss": 0.5072,
      "step": 3662
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.745118260383606,
      "learning_rate": 1.9517990366801358e-05,
      "loss": 0.8121,
      "step": 3663
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4336438179016113,
      "learning_rate": 1.951764782927483e-05,
      "loss": 0.6075,
      "step": 3664
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.004446268081665,
      "learning_rate": 1.9517305173088212e-05,
      "loss": 0.7647,
      "step": 3665
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9469176530838013,
      "learning_rate": 1.9516962398245783e-05,
      "loss": 0.7601,
      "step": 3666
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.62281334400177,
      "learning_rate": 1.9516619504751806e-05,
      "loss": 0.84,
      "step": 3667
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5635825395584106,
      "learning_rate": 1.9516276492610563e-05,
      "loss": 0.7269,
      "step": 3668
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8257231712341309,
      "learning_rate": 1.951593336182633e-05,
      "loss": 0.7834,
      "step": 3669
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7192708253860474,
      "learning_rate": 1.9515590112403386e-05,
      "loss": 0.7553,
      "step": 3670
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5513859987258911,
      "learning_rate": 1.951524674434601e-05,
      "loss": 0.8281,
      "step": 3671
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.884653091430664,
      "learning_rate": 1.9514903257658475e-05,
      "loss": 0.7943,
      "step": 3672
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.602116584777832,
      "learning_rate": 1.9514559652345073e-05,
      "loss": 0.794,
      "step": 3673
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.391160011291504,
      "learning_rate": 1.9514215928410084e-05,
      "loss": 0.8162,
      "step": 3674
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0154929161071777,
      "learning_rate": 1.9513872085857794e-05,
      "loss": 0.8136,
      "step": 3675
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.523823857307434,
      "learning_rate": 1.951352812469249e-05,
      "loss": 0.7428,
      "step": 3676
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5589916706085205,
      "learning_rate": 1.951318404491846e-05,
      "loss": 0.7996,
      "step": 3677
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7800358533859253,
      "learning_rate": 1.9512839846539995e-05,
      "loss": 0.7415,
      "step": 3678
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9322350025177002,
      "learning_rate": 1.9512495529561386e-05,
      "loss": 0.7298,
      "step": 3679
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7664772272109985,
      "learning_rate": 1.9512151093986923e-05,
      "loss": 0.7626,
      "step": 3680
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.1225240230560303,
      "learning_rate": 1.9511806539820902e-05,
      "loss": 0.8657,
      "step": 3681
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8251670598983765,
      "learning_rate": 1.951146186706762e-05,
      "loss": 0.7723,
      "step": 3682
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6171706914901733,
      "learning_rate": 1.9511117075731373e-05,
      "loss": 0.7673,
      "step": 3683
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.727776288986206,
      "learning_rate": 1.9510772165816457e-05,
      "loss": 0.8286,
      "step": 3684
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5835068225860596,
      "learning_rate": 1.9510427137327178e-05,
      "loss": 0.8002,
      "step": 3685
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4450149536132812,
      "learning_rate": 1.951008199026783e-05,
      "loss": 0.5775,
      "step": 3686
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7076038122177124,
      "learning_rate": 1.9509736724642723e-05,
      "loss": 0.6642,
      "step": 3687
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.477518081665039,
      "learning_rate": 1.950939134045616e-05,
      "loss": 0.7977,
      "step": 3688
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4546141624450684,
      "learning_rate": 1.9509045837712443e-05,
      "loss": 0.9188,
      "step": 3689
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7495094537734985,
      "learning_rate": 1.9508700216415886e-05,
      "loss": 0.8093,
      "step": 3690
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9595160484313965,
      "learning_rate": 1.950835447657079e-05,
      "loss": 0.8523,
      "step": 3691
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6209111213684082,
      "learning_rate": 1.9508008618181473e-05,
      "loss": 0.8674,
      "step": 3692
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.506331443786621,
      "learning_rate": 1.9507662641252245e-05,
      "loss": 0.6111,
      "step": 3693
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.898935317993164,
      "learning_rate": 1.9507316545787417e-05,
      "loss": 0.7486,
      "step": 3694
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4389801025390625,
      "learning_rate": 1.9506970331791306e-05,
      "loss": 0.7891,
      "step": 3695
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5183345079421997,
      "learning_rate": 1.9506623999268224e-05,
      "loss": 0.5776,
      "step": 3696
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.898126244544983,
      "learning_rate": 1.9506277548222495e-05,
      "loss": 0.7132,
      "step": 3697
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.612593650817871,
      "learning_rate": 1.9505930978658436e-05,
      "loss": 0.9297,
      "step": 3698
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7715497016906738,
      "learning_rate": 1.950558429058037e-05,
      "loss": 0.7773,
      "step": 3699
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7775506973266602,
      "learning_rate": 1.9505237483992612e-05,
      "loss": 0.7771,
      "step": 3700
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9162073135375977,
      "learning_rate": 1.950489055889949e-05,
      "loss": 0.7359,
      "step": 3701
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5994620323181152,
      "learning_rate": 1.9504543515305335e-05,
      "loss": 0.8366,
      "step": 3702
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.4382503032684326,
      "learning_rate": 1.9504196353214465e-05,
      "loss": 0.8927,
      "step": 3703
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.750445008277893,
      "learning_rate": 1.9503849072631214e-05,
      "loss": 0.6902,
      "step": 3704
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7827847003936768,
      "learning_rate": 1.950350167355991e-05,
      "loss": 0.7992,
      "step": 3705
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.269902467727661,
      "learning_rate": 1.950315415600488e-05,
      "loss": 0.8498,
      "step": 3706
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4735208749771118,
      "learning_rate": 1.9502806519970462e-05,
      "loss": 0.7822,
      "step": 3707
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.833630919456482,
      "learning_rate": 1.9502458765460988e-05,
      "loss": 0.8633,
      "step": 3708
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6704274415969849,
      "learning_rate": 1.9502110892480796e-05,
      "loss": 0.7349,
      "step": 3709
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.7890607118606567,
      "learning_rate": 1.9501762901034218e-05,
      "loss": 0.6619,
      "step": 3710
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5121421813964844,
      "learning_rate": 1.95014147911256e-05,
      "loss": 0.7647,
      "step": 3711
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9072104692459106,
      "learning_rate": 1.9501066562759274e-05,
      "loss": 0.7808,
      "step": 3712
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4681261777877808,
      "learning_rate": 1.9500718215939585e-05,
      "loss": 0.7464,
      "step": 3713
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4505103826522827,
      "learning_rate": 1.9500369750670877e-05,
      "loss": 0.9434,
      "step": 3714
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9526385068893433,
      "learning_rate": 1.9500021166957492e-05,
      "loss": 0.791,
      "step": 3715
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8718191385269165,
      "learning_rate": 1.949967246480378e-05,
      "loss": 0.9353,
      "step": 3716
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6000714302062988,
      "learning_rate": 1.9499323644214084e-05,
      "loss": 0.7444,
      "step": 3717
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9573800563812256,
      "learning_rate": 1.9498974705192753e-05,
      "loss": 0.7514,
      "step": 3718
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6410340070724487,
      "learning_rate": 1.949862564774414e-05,
      "loss": 0.7653,
      "step": 3719
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4314979314804077,
      "learning_rate": 1.94982764718726e-05,
      "loss": 0.8429,
      "step": 3720
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.6470434665679932,
      "learning_rate": 1.9497927177582482e-05,
      "loss": 0.6369,
      "step": 3721
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4438239336013794,
      "learning_rate": 1.9497577764878136e-05,
      "loss": 0.6694,
      "step": 3722
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.494306206703186,
      "learning_rate": 1.9497228233763926e-05,
      "loss": 0.8068,
      "step": 3723
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5622406005859375,
      "learning_rate": 1.9496878584244207e-05,
      "loss": 0.6005,
      "step": 3724
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.694959044456482,
      "learning_rate": 1.9496528816323336e-05,
      "loss": 0.8902,
      "step": 3725
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6621819734573364,
      "learning_rate": 1.9496178930005677e-05,
      "loss": 0.6656,
      "step": 3726
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.884511113166809,
      "learning_rate": 1.9495828925295594e-05,
      "loss": 0.6001,
      "step": 3727
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0955259799957275,
      "learning_rate": 1.9495478802197443e-05,
      "loss": 0.7673,
      "step": 3728
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7576624155044556,
      "learning_rate": 1.9495128560715596e-05,
      "loss": 0.6169,
      "step": 3729
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4063280820846558,
      "learning_rate": 1.9494778200854417e-05,
      "loss": 0.8149,
      "step": 3730
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6487579345703125,
      "learning_rate": 1.9494427722618273e-05,
      "loss": 0.6157,
      "step": 3731
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9934477806091309,
      "learning_rate": 1.9494077126011537e-05,
      "loss": 0.8208,
      "step": 3732
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6797831058502197,
      "learning_rate": 1.949372641103858e-05,
      "loss": 0.8823,
      "step": 3733
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8612958192825317,
      "learning_rate": 1.9493375577703772e-05,
      "loss": 0.7069,
      "step": 3734
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9626083374023438,
      "learning_rate": 1.9493024626011484e-05,
      "loss": 0.8271,
      "step": 3735
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7452980279922485,
      "learning_rate": 1.9492673555966094e-05,
      "loss": 0.6232,
      "step": 3736
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7055209875106812,
      "learning_rate": 1.9492322367571984e-05,
      "loss": 0.6655,
      "step": 3737
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6561449766159058,
      "learning_rate": 1.9491971060833527e-05,
      "loss": 0.7059,
      "step": 3738
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0660557746887207,
      "learning_rate": 1.9491619635755104e-05,
      "loss": 0.8583,
      "step": 3739
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8099281787872314,
      "learning_rate": 1.9491268092341096e-05,
      "loss": 0.7786,
      "step": 3740
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0189359188079834,
      "learning_rate": 1.9490916430595884e-05,
      "loss": 0.7128,
      "step": 3741
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.667280673980713,
      "learning_rate": 1.949056465052386e-05,
      "loss": 0.7167,
      "step": 3742
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.776950716972351,
      "learning_rate": 1.94902127521294e-05,
      "loss": 0.8656,
      "step": 3743
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4008303880691528,
      "learning_rate": 1.9489860735416897e-05,
      "loss": 0.5248,
      "step": 3744
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.876052737236023,
      "learning_rate": 1.9489508600390738e-05,
      "loss": 0.8096,
      "step": 3745
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.709682583808899,
      "learning_rate": 1.9489156347055314e-05,
      "loss": 0.8718,
      "step": 3746
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.155280351638794,
      "learning_rate": 1.9488803975415018e-05,
      "loss": 0.6491,
      "step": 3747
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5537289381027222,
      "learning_rate": 1.9488451485474238e-05,
      "loss": 0.8189,
      "step": 3748
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9018328189849854,
      "learning_rate": 1.948809887723737e-05,
      "loss": 0.6519,
      "step": 3749
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.866379737854004,
      "learning_rate": 1.9487746150708817e-05,
      "loss": 0.7852,
      "step": 3750
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.712284803390503,
      "learning_rate": 1.9487393305892972e-05,
      "loss": 0.7269,
      "step": 3751
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7891751527786255,
      "learning_rate": 1.948704034279423e-05,
      "loss": 0.7107,
      "step": 3752
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.76515793800354,
      "learning_rate": 1.9486687261417e-05,
      "loss": 0.9152,
      "step": 3753
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5484302043914795,
      "learning_rate": 1.9486334061765673e-05,
      "loss": 0.8895,
      "step": 3754
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6573588848114014,
      "learning_rate": 1.9485980743844664e-05,
      "loss": 0.9084,
      "step": 3755
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9152947664260864,
      "learning_rate": 1.9485627307658368e-05,
      "loss": 0.7567,
      "step": 3756
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5011855363845825,
      "learning_rate": 1.94852737532112e-05,
      "loss": 0.7898,
      "step": 3757
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7530053853988647,
      "learning_rate": 1.948492008050756e-05,
      "loss": 0.8774,
      "step": 3758
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.552711844444275,
      "learning_rate": 1.9484566289551864e-05,
      "loss": 0.6873,
      "step": 3759
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5226526260375977,
      "learning_rate": 1.9484212380348522e-05,
      "loss": 0.723,
      "step": 3760
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6720492839813232,
      "learning_rate": 1.9483858352901942e-05,
      "loss": 0.6001,
      "step": 3761
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.724550724029541,
      "learning_rate": 1.9483504207216543e-05,
      "loss": 0.8865,
      "step": 3762
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.699721097946167,
      "learning_rate": 1.9483149943296735e-05,
      "loss": 0.7087,
      "step": 3763
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4573028087615967,
      "learning_rate": 1.9482795561146936e-05,
      "loss": 0.689,
      "step": 3764
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.952087163925171,
      "learning_rate": 1.9482441060771565e-05,
      "loss": 0.7593,
      "step": 3765
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4573798179626465,
      "learning_rate": 1.9482086442175047e-05,
      "loss": 0.8438,
      "step": 3766
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.796043872833252,
      "learning_rate": 1.9481731705361793e-05,
      "loss": 0.8077,
      "step": 3767
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.676393747329712,
      "learning_rate": 1.9481376850336234e-05,
      "loss": 0.7606,
      "step": 3768
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6250126361846924,
      "learning_rate": 1.948102187710279e-05,
      "loss": 0.7519,
      "step": 3769
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6628049612045288,
      "learning_rate": 1.948066678566589e-05,
      "loss": 0.7052,
      "step": 3770
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9530186653137207,
      "learning_rate": 1.9480311576029957e-05,
      "loss": 0.6535,
      "step": 3771
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3599882125854492,
      "learning_rate": 1.947995624819942e-05,
      "loss": 0.8813,
      "step": 3772
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6327811479568481,
      "learning_rate": 1.9479600802178713e-05,
      "loss": 0.7053,
      "step": 3773
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5475313663482666,
      "learning_rate": 1.947924523797226e-05,
      "loss": 0.6967,
      "step": 3774
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4604748487472534,
      "learning_rate": 1.9478889555584506e-05,
      "loss": 0.6626,
      "step": 3775
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0489859580993652,
      "learning_rate": 1.9478533755019874e-05,
      "loss": 0.9129,
      "step": 3776
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.045675039291382,
      "learning_rate": 1.9478177836282805e-05,
      "loss": 0.6664,
      "step": 3777
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4862436056137085,
      "learning_rate": 1.9477821799377734e-05,
      "loss": 0.7816,
      "step": 3778
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4097304344177246,
      "learning_rate": 1.94774656443091e-05,
      "loss": 0.7057,
      "step": 3779
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0419602394104004,
      "learning_rate": 1.9477109371081348e-05,
      "loss": 0.9409,
      "step": 3780
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7966177463531494,
      "learning_rate": 1.9476752979698915e-05,
      "loss": 0.7649,
      "step": 3781
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6916205883026123,
      "learning_rate": 1.9476396470166245e-05,
      "loss": 0.6699,
      "step": 3782
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.706567645072937,
      "learning_rate": 1.9476039842487787e-05,
      "loss": 0.8264,
      "step": 3783
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.5021233558654785,
      "learning_rate": 1.9475683096667978e-05,
      "loss": 0.8218,
      "step": 3784
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5416306257247925,
      "learning_rate": 1.9475326232711275e-05,
      "loss": 0.7854,
      "step": 3785
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.644849181175232,
      "learning_rate": 1.947496925062212e-05,
      "loss": 0.6918,
      "step": 3786
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5564817190170288,
      "learning_rate": 1.947461215040497e-05,
      "loss": 0.6994,
      "step": 3787
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9019132852554321,
      "learning_rate": 1.9474254932064272e-05,
      "loss": 0.7606,
      "step": 3788
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4198615550994873,
      "learning_rate": 1.9473897595604487e-05,
      "loss": 0.8322,
      "step": 3789
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5253568887710571,
      "learning_rate": 1.947354014103006e-05,
      "loss": 0.79,
      "step": 3790
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6401580572128296,
      "learning_rate": 1.9473182568345453e-05,
      "loss": 0.6802,
      "step": 3791
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7640761137008667,
      "learning_rate": 1.9472824877555122e-05,
      "loss": 0.7898,
      "step": 3792
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7820245027542114,
      "learning_rate": 1.947246706866353e-05,
      "loss": 0.9602,
      "step": 3793
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1896214485168457,
      "learning_rate": 1.9472109141675135e-05,
      "loss": 0.7581,
      "step": 3794
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4898368120193481,
      "learning_rate": 1.94717510965944e-05,
      "loss": 0.7462,
      "step": 3795
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6355090141296387,
      "learning_rate": 1.9471392933425788e-05,
      "loss": 0.7491,
      "step": 3796
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.558385968208313,
      "learning_rate": 1.9471034652173767e-05,
      "loss": 0.8317,
      "step": 3797
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6718372106552124,
      "learning_rate": 1.94706762528428e-05,
      "loss": 0.7855,
      "step": 3798
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1356468200683594,
      "learning_rate": 1.947031773543736e-05,
      "loss": 0.8671,
      "step": 3799
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.0154340267181396,
      "learning_rate": 1.946995909996191e-05,
      "loss": 0.6777,
      "step": 3800
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.177236557006836,
      "learning_rate": 1.946960034642093e-05,
      "loss": 0.871,
      "step": 3801
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8813437223434448,
      "learning_rate": 1.946924147481889e-05,
      "loss": 0.7888,
      "step": 3802
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7221968173980713,
      "learning_rate": 1.946888248516026e-05,
      "loss": 0.8999,
      "step": 3803
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7699964046478271,
      "learning_rate": 1.9468523377449514e-05,
      "loss": 0.7134,
      "step": 3804
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5712831020355225,
      "learning_rate": 1.946816415169114e-05,
      "loss": 0.7465,
      "step": 3805
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7410995960235596,
      "learning_rate": 1.9467804807889604e-05,
      "loss": 0.8235,
      "step": 3806
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.038494348526001,
      "learning_rate": 1.9467445346049394e-05,
      "loss": 0.7424,
      "step": 3807
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7732561826705933,
      "learning_rate": 1.946708576617499e-05,
      "loss": 0.6555,
      "step": 3808
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7774189710617065,
      "learning_rate": 1.9466726068270875e-05,
      "loss": 0.8194,
      "step": 3809
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9017030000686646,
      "learning_rate": 1.946636625234153e-05,
      "loss": 0.823,
      "step": 3810
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5278891324996948,
      "learning_rate": 1.9466006318391448e-05,
      "loss": 0.7349,
      "step": 3811
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7047686576843262,
      "learning_rate": 1.9465646266425108e-05,
      "loss": 0.7299,
      "step": 3812
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5044527053833008,
      "learning_rate": 1.9465286096447003e-05,
      "loss": 0.7512,
      "step": 3813
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5393927097320557,
      "learning_rate": 1.946492580846163e-05,
      "loss": 0.7357,
      "step": 3814
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.618251919746399,
      "learning_rate": 1.946456540247347e-05,
      "loss": 0.7733,
      "step": 3815
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.2147915363311768,
      "learning_rate": 1.9464204878487018e-05,
      "loss": 0.8134,
      "step": 3816
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.672716736793518,
      "learning_rate": 1.9463844236506776e-05,
      "loss": 0.7196,
      "step": 3817
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.9367929697036743,
      "learning_rate": 1.9463483476537234e-05,
      "loss": 0.7858,
      "step": 3818
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.365797758102417,
      "learning_rate": 1.946312259858289e-05,
      "loss": 0.8158,
      "step": 3819
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8177708387374878,
      "learning_rate": 1.9462761602648248e-05,
      "loss": 0.6567,
      "step": 3820
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9723255634307861,
      "learning_rate": 1.9462400488737806e-05,
      "loss": 0.7439,
      "step": 3821
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4964179992675781,
      "learning_rate": 1.946203925685606e-05,
      "loss": 0.6959,
      "step": 3822
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6986582279205322,
      "learning_rate": 1.9461677907007526e-05,
      "loss": 0.8023,
      "step": 3823
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5929551124572754,
      "learning_rate": 1.9461316439196697e-05,
      "loss": 0.784,
      "step": 3824
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7939634323120117,
      "learning_rate": 1.9460954853428086e-05,
      "loss": 0.7172,
      "step": 3825
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.53040611743927,
      "learning_rate": 1.94605931497062e-05,
      "loss": 0.7095,
      "step": 3826
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.325294017791748,
      "learning_rate": 1.9460231328035547e-05,
      "loss": 0.7084,
      "step": 3827
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.866451382637024,
      "learning_rate": 1.945986938842064e-05,
      "loss": 0.9378,
      "step": 3828
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.628967523574829,
      "learning_rate": 1.9459507330865995e-05,
      "loss": 0.771,
      "step": 3829
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.002526044845581,
      "learning_rate": 1.9459145155376117e-05,
      "loss": 0.7802,
      "step": 3830
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7970554828643799,
      "learning_rate": 1.9458782861955528e-05,
      "loss": 0.8663,
      "step": 3831
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9521613121032715,
      "learning_rate": 1.9458420450608742e-05,
      "loss": 0.7042,
      "step": 3832
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.466181993484497,
      "learning_rate": 1.945805792134028e-05,
      "loss": 0.8319,
      "step": 3833
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8291717767715454,
      "learning_rate": 1.9457695274154656e-05,
      "loss": 0.849,
      "step": 3834
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4288570880889893,
      "learning_rate": 1.9457332509056398e-05,
      "loss": 0.7269,
      "step": 3835
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8699830770492554,
      "learning_rate": 1.9456969626050026e-05,
      "loss": 0.7698,
      "step": 3836
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6707947254180908,
      "learning_rate": 1.9456606625140062e-05,
      "loss": 0.8634,
      "step": 3837
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.092548370361328,
      "learning_rate": 1.9456243506331037e-05,
      "loss": 0.8826,
      "step": 3838
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5814248323440552,
      "learning_rate": 1.9455880269627473e-05,
      "loss": 0.7853,
      "step": 3839
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4871580600738525,
      "learning_rate": 1.9455516915033898e-05,
      "loss": 0.7126,
      "step": 3840
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6587815284729004,
      "learning_rate": 1.9455153442554848e-05,
      "loss": 0.8568,
      "step": 3841
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.722244381904602,
      "learning_rate": 1.945478985219485e-05,
      "loss": 0.7738,
      "step": 3842
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.814063310623169,
      "learning_rate": 1.945442614395844e-05,
      "loss": 0.8289,
      "step": 3843
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6428202390670776,
      "learning_rate": 1.945406231785015e-05,
      "loss": 0.6799,
      "step": 3844
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.000387191772461,
      "learning_rate": 1.9453698373874513e-05,
      "loss": 0.8714,
      "step": 3845
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.762983798980713,
      "learning_rate": 1.9453334312036074e-05,
      "loss": 0.7867,
      "step": 3846
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4930096864700317,
      "learning_rate": 1.9452970132339367e-05,
      "loss": 0.9059,
      "step": 3847
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.255516290664673,
      "learning_rate": 1.945260583478893e-05,
      "loss": 0.7697,
      "step": 3848
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9084057807922363,
      "learning_rate": 1.9452241419389314e-05,
      "loss": 0.7452,
      "step": 3849
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5505834817886353,
      "learning_rate": 1.9451876886145053e-05,
      "loss": 0.7082,
      "step": 3850
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7086405754089355,
      "learning_rate": 1.9451512235060697e-05,
      "loss": 0.811,
      "step": 3851
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3648658990859985,
      "learning_rate": 1.945114746614079e-05,
      "loss": 0.7436,
      "step": 3852
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.495844841003418,
      "learning_rate": 1.945078257938988e-05,
      "loss": 0.8257,
      "step": 3853
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.531083345413208,
      "learning_rate": 1.9450417574812515e-05,
      "loss": 0.7701,
      "step": 3854
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8276780843734741,
      "learning_rate": 1.9450052452413247e-05,
      "loss": 0.7925,
      "step": 3855
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4771496057510376,
      "learning_rate": 1.944968721219663e-05,
      "loss": 0.8168,
      "step": 3856
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.361590027809143,
      "learning_rate": 1.9449321854167215e-05,
      "loss": 0.7216,
      "step": 3857
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9396259784698486,
      "learning_rate": 1.944895637832956e-05,
      "loss": 0.8675,
      "step": 3858
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6905133724212646,
      "learning_rate": 1.9448590784688213e-05,
      "loss": 0.6949,
      "step": 3859
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7999098300933838,
      "learning_rate": 1.9448225073247743e-05,
      "loss": 0.7085,
      "step": 3860
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.696768045425415,
      "learning_rate": 1.9447859244012704e-05,
      "loss": 0.7461,
      "step": 3861
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0414090156555176,
      "learning_rate": 1.9447493296987658e-05,
      "loss": 0.7032,
      "step": 3862
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4212228059768677,
      "learning_rate": 1.9447127232177168e-05,
      "loss": 0.7709,
      "step": 3863
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5281503200531006,
      "learning_rate": 1.9446761049585795e-05,
      "loss": 0.6961,
      "step": 3864
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8598142862319946,
      "learning_rate": 1.944639474921811e-05,
      "loss": 0.9272,
      "step": 3865
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8470940589904785,
      "learning_rate": 1.944602833107867e-05,
      "loss": 0.7513,
      "step": 3866
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8936764001846313,
      "learning_rate": 1.9445661795172053e-05,
      "loss": 0.8093,
      "step": 3867
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.0317797660827637,
      "learning_rate": 1.9445295141502825e-05,
      "loss": 0.7782,
      "step": 3868
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7420421838760376,
      "learning_rate": 1.9444928370075556e-05,
      "loss": 0.6763,
      "step": 3869
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8466025590896606,
      "learning_rate": 1.944456148089482e-05,
      "loss": 0.7471,
      "step": 3870
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6970269680023193,
      "learning_rate": 1.9444194473965193e-05,
      "loss": 0.8079,
      "step": 3871
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.633888840675354,
      "learning_rate": 1.944382734929125e-05,
      "loss": 0.9526,
      "step": 3872
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5231534242630005,
      "learning_rate": 1.9443460106877562e-05,
      "loss": 0.6362,
      "step": 3873
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.782042384147644,
      "learning_rate": 1.9443092746728713e-05,
      "loss": 0.6539,
      "step": 3874
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7282414436340332,
      "learning_rate": 1.9442725268849285e-05,
      "loss": 0.8316,
      "step": 3875
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6596516370773315,
      "learning_rate": 1.9442357673243858e-05,
      "loss": 0.7027,
      "step": 3876
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.734186053276062,
      "learning_rate": 1.9441989959917012e-05,
      "loss": 0.7677,
      "step": 3877
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6618963479995728,
      "learning_rate": 1.9441622128873333e-05,
      "loss": 0.7506,
      "step": 3878
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.655316710472107,
      "learning_rate": 1.9441254180117406e-05,
      "loss": 0.6725,
      "step": 3879
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6863679885864258,
      "learning_rate": 1.9440886113653824e-05,
      "loss": 0.6901,
      "step": 3880
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5966355800628662,
      "learning_rate": 1.9440517929487167e-05,
      "loss": 0.786,
      "step": 3881
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5418552160263062,
      "learning_rate": 1.944014962762203e-05,
      "loss": 0.7521,
      "step": 3882
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7922817468643188,
      "learning_rate": 1.9439781208063006e-05,
      "loss": 0.6703,
      "step": 3883
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.658118486404419,
      "learning_rate": 1.9439412670814683e-05,
      "loss": 0.7088,
      "step": 3884
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8732805252075195,
      "learning_rate": 1.9439044015881663e-05,
      "loss": 0.7749,
      "step": 3885
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6497626304626465,
      "learning_rate": 1.9438675243268536e-05,
      "loss": 0.7553,
      "step": 3886
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.9778746366500854,
      "learning_rate": 1.9438306352979906e-05,
      "loss": 0.6994,
      "step": 3887
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7542976140975952,
      "learning_rate": 1.9437937345020365e-05,
      "loss": 0.865,
      "step": 3888
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1724393367767334,
      "learning_rate": 1.9437568219394518e-05,
      "loss": 0.8272,
      "step": 3889
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.758215308189392,
      "learning_rate": 1.9437198976106964e-05,
      "loss": 0.7657,
      "step": 3890
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7945421934127808,
      "learning_rate": 1.9436829615162312e-05,
      "loss": 0.8842,
      "step": 3891
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3621010780334473,
      "learning_rate": 1.9436460136565157e-05,
      "loss": 0.8118,
      "step": 3892
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7931336164474487,
      "learning_rate": 1.943609054032012e-05,
      "loss": 0.9195,
      "step": 3893
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.8734203577041626,
      "learning_rate": 1.943572082643179e-05,
      "loss": 0.6702,
      "step": 3894
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4650617837905884,
      "learning_rate": 1.9435350994904795e-05,
      "loss": 0.7236,
      "step": 3895
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.6110856533050537,
      "learning_rate": 1.9434981045743737e-05,
      "loss": 0.8275,
      "step": 3896
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.004140853881836,
      "learning_rate": 1.9434610978953226e-05,
      "loss": 0.744,
      "step": 3897
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.976688265800476,
      "learning_rate": 1.943424079453788e-05,
      "loss": 0.7529,
      "step": 3898
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.2010138034820557,
      "learning_rate": 1.9433870492502315e-05,
      "loss": 0.7195,
      "step": 3899
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4773290157318115,
      "learning_rate": 1.943350007285114e-05,
      "loss": 0.783,
      "step": 3900
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7563787698745728,
      "learning_rate": 1.9433129535588983e-05,
      "loss": 0.9297,
      "step": 3901
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.661257028579712,
      "learning_rate": 1.943275888072046e-05,
      "loss": 0.8198,
      "step": 3902
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3226696252822876,
      "learning_rate": 1.943238810825019e-05,
      "loss": 0.7306,
      "step": 3903
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.1046385765075684,
      "learning_rate": 1.9432017218182798e-05,
      "loss": 0.7278,
      "step": 3904
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7371158599853516,
      "learning_rate": 1.943164621052291e-05,
      "loss": 0.7415,
      "step": 3905
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5726269483566284,
      "learning_rate": 1.9431275085275147e-05,
      "loss": 0.6969,
      "step": 3906
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7366745471954346,
      "learning_rate": 1.9430903842444135e-05,
      "loss": 0.7406,
      "step": 3907
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5719796419143677,
      "learning_rate": 1.9430532482034504e-05,
      "loss": 0.6355,
      "step": 3908
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5146458148956299,
      "learning_rate": 1.943016100405089e-05,
      "loss": 0.779,
      "step": 3909
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.050163507461548,
      "learning_rate": 1.9429789408497913e-05,
      "loss": 0.6378,
      "step": 3910
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.676647424697876,
      "learning_rate": 1.9429417695380216e-05,
      "loss": 0.7834,
      "step": 3911
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7135181427001953,
      "learning_rate": 1.942904586470243e-05,
      "loss": 0.6438,
      "step": 3912
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.820215106010437,
      "learning_rate": 1.942867391646919e-05,
      "loss": 0.9043,
      "step": 3913
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7742210626602173,
      "learning_rate": 1.942830185068513e-05,
      "loss": 0.7822,
      "step": 3914
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.758414387702942,
      "learning_rate": 1.9427929667354898e-05,
      "loss": 0.6998,
      "step": 3915
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.692508339881897,
      "learning_rate": 1.9427557366483122e-05,
      "loss": 0.7869,
      "step": 3916
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.656701922416687,
      "learning_rate": 1.9427184948074453e-05,
      "loss": 0.8719,
      "step": 3917
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4219427108764648,
      "learning_rate": 1.942681241213353e-05,
      "loss": 0.7424,
      "step": 3918
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5396867990493774,
      "learning_rate": 1.9426439758665003e-05,
      "loss": 0.7251,
      "step": 3919
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5137581825256348,
      "learning_rate": 1.9426066987673506e-05,
      "loss": 0.7317,
      "step": 3920
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8151662349700928,
      "learning_rate": 1.9425694099163695e-05,
      "loss": 0.7462,
      "step": 3921
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.710797667503357,
      "learning_rate": 1.9425321093140223e-05,
      "loss": 0.8525,
      "step": 3922
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5734120607376099,
      "learning_rate": 1.9424947969607732e-05,
      "loss": 0.7441,
      "step": 3923
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5319576263427734,
      "learning_rate": 1.9424574728570876e-05,
      "loss": 0.8655,
      "step": 3924
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8322142362594604,
      "learning_rate": 1.942420137003431e-05,
      "loss": 0.7812,
      "step": 3925
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7098106145858765,
      "learning_rate": 1.9423827894002692e-05,
      "loss": 0.7469,
      "step": 3926
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.624755859375,
      "learning_rate": 1.942345430048067e-05,
      "loss": 0.6243,
      "step": 3927
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8452062606811523,
      "learning_rate": 1.942308058947291e-05,
      "loss": 0.7657,
      "step": 3928
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.69338059425354,
      "learning_rate": 1.9422706760984065e-05,
      "loss": 0.71,
      "step": 3929
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5840644836425781,
      "learning_rate": 1.9422332815018798e-05,
      "loss": 0.86,
      "step": 3930
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6120421886444092,
      "learning_rate": 1.9421958751581773e-05,
      "loss": 0.763,
      "step": 3931
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6045031547546387,
      "learning_rate": 1.942158457067765e-05,
      "loss": 0.6763,
      "step": 3932
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.1074812412261963,
      "learning_rate": 1.9421210272311098e-05,
      "loss": 0.9165,
      "step": 3933
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0944314002990723,
      "learning_rate": 1.9420835856486778e-05,
      "loss": 0.9021,
      "step": 3934
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7282636165618896,
      "learning_rate": 1.9420461323209366e-05,
      "loss": 0.7276,
      "step": 3935
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5942741632461548,
      "learning_rate": 1.9420086672483522e-05,
      "loss": 0.8193,
      "step": 3936
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4910145998001099,
      "learning_rate": 1.9419711904313926e-05,
      "loss": 0.7779,
      "step": 3937
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4364780187606812,
      "learning_rate": 1.9419337018705242e-05,
      "loss": 0.8446,
      "step": 3938
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8315128087997437,
      "learning_rate": 1.9418962015662152e-05,
      "loss": 0.8577,
      "step": 3939
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7249895334243774,
      "learning_rate": 1.9418586895189327e-05,
      "loss": 0.7254,
      "step": 3940
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8514169454574585,
      "learning_rate": 1.9418211657291438e-05,
      "loss": 0.8257,
      "step": 3941
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5329018831253052,
      "learning_rate": 1.9417836301973176e-05,
      "loss": 0.7924,
      "step": 3942
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6350269317626953,
      "learning_rate": 1.9417460829239213e-05,
      "loss": 0.7706,
      "step": 3943
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.430349588394165,
      "learning_rate": 1.941708523909423e-05,
      "loss": 0.7384,
      "step": 3944
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8461880683898926,
      "learning_rate": 1.941670953154291e-05,
      "loss": 0.8729,
      "step": 3945
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8942387104034424,
      "learning_rate": 1.9416333706589937e-05,
      "loss": 0.7698,
      "step": 3946
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5611521005630493,
      "learning_rate": 1.9415957764239997e-05,
      "loss": 0.7521,
      "step": 3947
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4754915237426758,
      "learning_rate": 1.9415581704497782e-05,
      "loss": 0.8218,
      "step": 3948
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.639549732208252,
      "learning_rate": 1.9415205527367972e-05,
      "loss": 0.6282,
      "step": 3949
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5735175609588623,
      "learning_rate": 1.941482923285526e-05,
      "loss": 0.811,
      "step": 3950
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.748230218887329,
      "learning_rate": 1.9414452820964343e-05,
      "loss": 0.7893,
      "step": 3951
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8819977045059204,
      "learning_rate": 1.9414076291699907e-05,
      "loss": 1.0508,
      "step": 3952
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4697046279907227,
      "learning_rate": 1.9413699645066646e-05,
      "loss": 0.777,
      "step": 3953
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7279118299484253,
      "learning_rate": 1.941332288106926e-05,
      "loss": 0.7371,
      "step": 3954
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6172982454299927,
      "learning_rate": 1.9412945999712443e-05,
      "loss": 0.8132,
      "step": 3955
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6535755395889282,
      "learning_rate": 1.9412569001000898e-05,
      "loss": 0.8099,
      "step": 3956
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5192726850509644,
      "learning_rate": 1.941219188493932e-05,
      "loss": 0.715,
      "step": 3957
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7864052057266235,
      "learning_rate": 1.9411814651532415e-05,
      "loss": 0.9036,
      "step": 3958
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7668230533599854,
      "learning_rate": 1.9411437300784882e-05,
      "loss": 0.8737,
      "step": 3959
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.711465835571289,
      "learning_rate": 1.941105983270143e-05,
      "loss": 0.7038,
      "step": 3960
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6161246299743652,
      "learning_rate": 1.9410682247286764e-05,
      "loss": 0.6965,
      "step": 3961
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.710800051689148,
      "learning_rate": 1.9410304544545586e-05,
      "loss": 0.6944,
      "step": 3962
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5941978693008423,
      "learning_rate": 1.940992672448261e-05,
      "loss": 0.5898,
      "step": 3963
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.548445701599121,
      "learning_rate": 1.9409548787102546e-05,
      "loss": 0.8867,
      "step": 3964
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4913336038589478,
      "learning_rate": 1.9409170732410106e-05,
      "loss": 0.6975,
      "step": 3965
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.596683144569397,
      "learning_rate": 1.9408792560410003e-05,
      "loss": 0.6923,
      "step": 3966
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5200982093811035,
      "learning_rate": 1.9408414271106952e-05,
      "loss": 0.6806,
      "step": 3967
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8269948959350586,
      "learning_rate": 1.940803586450567e-05,
      "loss": 0.7815,
      "step": 3968
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.718766212463379,
      "learning_rate": 1.9407657340610874e-05,
      "loss": 0.6758,
      "step": 3969
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5419249534606934,
      "learning_rate": 1.940727869942728e-05,
      "loss": 0.4897,
      "step": 3970
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.661954641342163,
      "learning_rate": 1.940689994095961e-05,
      "loss": 0.841,
      "step": 3971
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.216703176498413,
      "learning_rate": 1.9406521065212594e-05,
      "loss": 1.0158,
      "step": 3972
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.582375168800354,
      "learning_rate": 1.9406142072190946e-05,
      "loss": 0.7399,
      "step": 3973
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7688695192337036,
      "learning_rate": 1.9405762961899393e-05,
      "loss": 0.6652,
      "step": 3974
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.0347723960876465,
      "learning_rate": 1.9405383734342663e-05,
      "loss": 0.8596,
      "step": 3975
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5546241998672485,
      "learning_rate": 1.9405004389525484e-05,
      "loss": 0.8466,
      "step": 3976
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7195948362350464,
      "learning_rate": 1.9404624927452586e-05,
      "loss": 0.6552,
      "step": 3977
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.100893020629883,
      "learning_rate": 1.94042453481287e-05,
      "loss": 1.0006,
      "step": 3978
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7872626781463623,
      "learning_rate": 1.9403865651558556e-05,
      "loss": 0.8022,
      "step": 3979
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7704479694366455,
      "learning_rate": 1.940348583774689e-05,
      "loss": 0.7212,
      "step": 3980
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.839829683303833,
      "learning_rate": 1.9403105906698437e-05,
      "loss": 0.7449,
      "step": 3981
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8263194561004639,
      "learning_rate": 1.9402725858417936e-05,
      "loss": 0.6923,
      "step": 3982
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5135302543640137,
      "learning_rate": 1.9402345692910117e-05,
      "loss": 0.7269,
      "step": 3983
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9539284706115723,
      "learning_rate": 1.940196541017973e-05,
      "loss": 0.9035,
      "step": 3984
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5605703592300415,
      "learning_rate": 1.9401585010231507e-05,
      "loss": 0.8808,
      "step": 3985
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.394649863243103,
      "learning_rate": 1.9401204493070195e-05,
      "loss": 0.7627,
      "step": 3986
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4775285720825195,
      "learning_rate": 1.940082385870054e-05,
      "loss": 0.9293,
      "step": 3987
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7009408473968506,
      "learning_rate": 1.9400443107127288e-05,
      "loss": 0.774,
      "step": 3988
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4206502437591553,
      "learning_rate": 1.940006223835518e-05,
      "loss": 0.7305,
      "step": 3989
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5754635334014893,
      "learning_rate": 1.939968125238897e-05,
      "loss": 0.6593,
      "step": 3990
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6008775234222412,
      "learning_rate": 1.9399300149233405e-05,
      "loss": 0.772,
      "step": 3991
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.576002597808838,
      "learning_rate": 1.9398918928893234e-05,
      "loss": 0.7426,
      "step": 3992
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5823860168457031,
      "learning_rate": 1.9398537591373215e-05,
      "loss": 0.7158,
      "step": 3993
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.72848379611969,
      "learning_rate": 1.9398156136678103e-05,
      "loss": 0.7449,
      "step": 3994
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.733222484588623,
      "learning_rate": 1.9397774564812646e-05,
      "loss": 0.7969,
      "step": 3995
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7190402746200562,
      "learning_rate": 1.939739287578161e-05,
      "loss": 0.7771,
      "step": 3996
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5513514280319214,
      "learning_rate": 1.939701106958975e-05,
      "loss": 0.7964,
      "step": 3997
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.230821371078491,
      "learning_rate": 1.9396629146241827e-05,
      "loss": 0.8135,
      "step": 3998
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.796710729598999,
      "learning_rate": 1.9396247105742595e-05,
      "loss": 0.8025,
      "step": 3999
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.05084228515625,
      "learning_rate": 1.939586494809683e-05,
      "loss": 0.8216,
      "step": 4000
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4809370040893555,
      "learning_rate": 1.9395482673309287e-05,
      "loss": 0.8185,
      "step": 4001
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.941067099571228,
      "learning_rate": 1.9395100281384738e-05,
      "loss": 0.6952,
      "step": 4002
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8451722860336304,
      "learning_rate": 1.9394717772327944e-05,
      "loss": 0.8141,
      "step": 4003
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.671423077583313,
      "learning_rate": 1.939433514614368e-05,
      "loss": 0.7207,
      "step": 4004
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6135504245758057,
      "learning_rate": 1.939395240283671e-05,
      "loss": 0.786,
      "step": 4005
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.557112216949463,
      "learning_rate": 1.9393569542411814e-05,
      "loss": 0.7924,
      "step": 4006
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5555646419525146,
      "learning_rate": 1.9393186564873758e-05,
      "loss": 0.7947,
      "step": 4007
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4477646350860596,
      "learning_rate": 1.9392803470227322e-05,
      "loss": 0.7223,
      "step": 4008
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.9714456796646118,
      "learning_rate": 1.9392420258477276e-05,
      "loss": 0.7918,
      "step": 4009
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5119116306304932,
      "learning_rate": 1.9392036929628402e-05,
      "loss": 0.7552,
      "step": 4010
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6721264123916626,
      "learning_rate": 1.939165348368548e-05,
      "loss": 0.7193,
      "step": 4011
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6548937559127808,
      "learning_rate": 1.939126992065329e-05,
      "loss": 0.7837,
      "step": 4012
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6184161901474,
      "learning_rate": 1.9390886240536614e-05,
      "loss": 0.8151,
      "step": 4013
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6495046615600586,
      "learning_rate": 1.939050244334023e-05,
      "loss": 0.7524,
      "step": 4014
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5863726139068604,
      "learning_rate": 1.939011852906893e-05,
      "loss": 0.7402,
      "step": 4015
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5191845893859863,
      "learning_rate": 1.9389734497727498e-05,
      "loss": 0.8264,
      "step": 4016
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.037113904953003,
      "learning_rate": 1.938935034932072e-05,
      "loss": 0.8689,
      "step": 4017
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7503973245620728,
      "learning_rate": 1.938896608385339e-05,
      "loss": 0.6177,
      "step": 4018
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7216923236846924,
      "learning_rate": 1.9388581701330294e-05,
      "loss": 0.7045,
      "step": 4019
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6838449239730835,
      "learning_rate": 1.9388197201756225e-05,
      "loss": 0.7859,
      "step": 4020
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4214049577713013,
      "learning_rate": 1.9387812585135978e-05,
      "loss": 0.8468,
      "step": 4021
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6901472806930542,
      "learning_rate": 1.9387427851474348e-05,
      "loss": 0.7491,
      "step": 4022
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6711256504058838,
      "learning_rate": 1.9387043000776133e-05,
      "loss": 0.8255,
      "step": 4023
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5979748964309692,
      "learning_rate": 1.9386658033046127e-05,
      "loss": 0.7967,
      "step": 4024
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.690148115158081,
      "learning_rate": 1.9386272948289134e-05,
      "loss": 0.7174,
      "step": 4025
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9321768283843994,
      "learning_rate": 1.9385887746509955e-05,
      "loss": 0.9553,
      "step": 4026
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6385587453842163,
      "learning_rate": 1.938550242771339e-05,
      "loss": 0.7238,
      "step": 4027
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5040464401245117,
      "learning_rate": 1.938511699190424e-05,
      "loss": 0.8021,
      "step": 4028
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3057918548583984,
      "learning_rate": 1.938473143908732e-05,
      "loss": 0.8122,
      "step": 4029
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.760646939277649,
      "learning_rate": 1.9384345769267425e-05,
      "loss": 0.7079,
      "step": 4030
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5276286602020264,
      "learning_rate": 1.938395998244937e-05,
      "loss": 0.7988,
      "step": 4031
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5776523351669312,
      "learning_rate": 1.9383574078637967e-05,
      "loss": 0.7082,
      "step": 4032
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6169663667678833,
      "learning_rate": 1.9383188057838017e-05,
      "loss": 0.7995,
      "step": 4033
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.636006474494934,
      "learning_rate": 1.9382801920054346e-05,
      "loss": 0.7227,
      "step": 4034
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7755500078201294,
      "learning_rate": 1.938241566529176e-05,
      "loss": 0.7136,
      "step": 4035
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5573091506958008,
      "learning_rate": 1.9382029293555078e-05,
      "loss": 0.8348,
      "step": 4036
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5621434450149536,
      "learning_rate": 1.9381642804849113e-05,
      "loss": 0.7898,
      "step": 4037
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5074886083602905,
      "learning_rate": 1.9381256199178686e-05,
      "loss": 0.6977,
      "step": 4038
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.559183120727539,
      "learning_rate": 1.9380869476548618e-05,
      "loss": 0.726,
      "step": 4039
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9983181953430176,
      "learning_rate": 1.9380482636963727e-05,
      "loss": 0.8181,
      "step": 4040
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6165305376052856,
      "learning_rate": 1.938009568042884e-05,
      "loss": 0.7479,
      "step": 4041
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8742144107818604,
      "learning_rate": 1.9379708606948782e-05,
      "loss": 0.7726,
      "step": 4042
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.638826847076416,
      "learning_rate": 1.937932141652837e-05,
      "loss": 0.6586,
      "step": 4043
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8859741687774658,
      "learning_rate": 1.937893410917244e-05,
      "loss": 0.6154,
      "step": 4044
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.471187949180603,
      "learning_rate": 1.9378546684885823e-05,
      "loss": 0.7274,
      "step": 4045
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9937982559204102,
      "learning_rate": 1.937815914367334e-05,
      "loss": 0.8932,
      "step": 4046
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.531019687652588,
      "learning_rate": 1.9377771485539828e-05,
      "loss": 0.8477,
      "step": 4047
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8043663501739502,
      "learning_rate": 1.937738371049012e-05,
      "loss": 0.93,
      "step": 4048
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4778624773025513,
      "learning_rate": 1.9376995818529046e-05,
      "loss": 0.6595,
      "step": 4049
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6499658823013306,
      "learning_rate": 1.937660780966145e-05,
      "loss": 0.6365,
      "step": 4050
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.029179096221924,
      "learning_rate": 1.937621968389216e-05,
      "loss": 0.9157,
      "step": 4051
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.352662205696106,
      "learning_rate": 1.9375831441226025e-05,
      "loss": 0.6882,
      "step": 4052
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.479119896888733,
      "learning_rate": 1.9375443081667876e-05,
      "loss": 0.8201,
      "step": 4053
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7112963199615479,
      "learning_rate": 1.9375054605222562e-05,
      "loss": 1.0174,
      "step": 4054
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7606515884399414,
      "learning_rate": 1.9374666011894924e-05,
      "loss": 0.6399,
      "step": 4055
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4506698846817017,
      "learning_rate": 1.9374277301689804e-05,
      "loss": 0.7479,
      "step": 4056
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4260482788085938,
      "learning_rate": 1.9373888474612054e-05,
      "loss": 0.5795,
      "step": 4057
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8634650707244873,
      "learning_rate": 1.937349953066651e-05,
      "loss": 0.7482,
      "step": 4058
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7825785875320435,
      "learning_rate": 1.937311046985804e-05,
      "loss": 0.7952,
      "step": 4059
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5576363801956177,
      "learning_rate": 1.9372721292191475e-05,
      "loss": 0.7815,
      "step": 4060
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7844672203063965,
      "learning_rate": 1.937233199767168e-05,
      "loss": 0.7737,
      "step": 4061
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.900876760482788,
      "learning_rate": 1.9371942586303506e-05,
      "loss": 0.8969,
      "step": 4062
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5072859525680542,
      "learning_rate": 1.93715530580918e-05,
      "loss": 0.8579,
      "step": 4063
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.626997947692871,
      "learning_rate": 1.9371163413041428e-05,
      "loss": 0.6217,
      "step": 4064
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.807750940322876,
      "learning_rate": 1.9370773651157243e-05,
      "loss": 0.5965,
      "step": 4065
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9082828760147095,
      "learning_rate": 1.9370383772444107e-05,
      "loss": 0.8243,
      "step": 4066
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6969316005706787,
      "learning_rate": 1.9369993776906883e-05,
      "loss": 0.7396,
      "step": 4067
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6038384437561035,
      "learning_rate": 1.9369603664550425e-05,
      "loss": 0.7592,
      "step": 4068
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.812211036682129,
      "learning_rate": 1.9369213435379603e-05,
      "loss": 0.6078,
      "step": 4069
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5248749256134033,
      "learning_rate": 1.936882308939928e-05,
      "loss": 0.6676,
      "step": 4070
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6458901166915894,
      "learning_rate": 1.9368432626614324e-05,
      "loss": 0.8073,
      "step": 4071
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7764604091644287,
      "learning_rate": 1.93680420470296e-05,
      "loss": 0.9667,
      "step": 4072
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.673642635345459,
      "learning_rate": 1.936765135064998e-05,
      "loss": 0.8813,
      "step": 4073
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.726894497871399,
      "learning_rate": 1.9367260537480336e-05,
      "loss": 0.7512,
      "step": 4074
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8213003873825073,
      "learning_rate": 1.936686960752554e-05,
      "loss": 0.6169,
      "step": 4075
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.721029281616211,
      "learning_rate": 1.9366478560790463e-05,
      "loss": 0.7462,
      "step": 4076
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8750869035720825,
      "learning_rate": 1.9366087397279982e-05,
      "loss": 0.8646,
      "step": 4077
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7695164680480957,
      "learning_rate": 1.9365696116998973e-05,
      "loss": 0.786,
      "step": 4078
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7323496341705322,
      "learning_rate": 1.936530471995232e-05,
      "loss": 0.8553,
      "step": 4079
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6537824869155884,
      "learning_rate": 1.9364913206144892e-05,
      "loss": 0.8766,
      "step": 4080
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5714809894561768,
      "learning_rate": 1.936452157558158e-05,
      "loss": 0.7103,
      "step": 4081
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7566919326782227,
      "learning_rate": 1.936412982826726e-05,
      "loss": 0.8065,
      "step": 4082
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7638317346572876,
      "learning_rate": 1.936373796420682e-05,
      "loss": 0.7181,
      "step": 4083
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9356646537780762,
      "learning_rate": 1.9363345983405145e-05,
      "loss": 0.6878,
      "step": 4084
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.998623251914978,
      "learning_rate": 1.936295388586712e-05,
      "loss": 0.8459,
      "step": 4085
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3894026279449463,
      "learning_rate": 1.9362561671597635e-05,
      "loss": 0.8254,
      "step": 4086
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6500879526138306,
      "learning_rate": 1.936216934060158e-05,
      "loss": 0.8051,
      "step": 4087
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9168169498443604,
      "learning_rate": 1.9361776892883846e-05,
      "loss": 0.8133,
      "step": 4088
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5548352003097534,
      "learning_rate": 1.9361384328449328e-05,
      "loss": 0.652,
      "step": 4089
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.775146484375,
      "learning_rate": 1.9360991647302912e-05,
      "loss": 0.8124,
      "step": 4090
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.476334810256958,
      "learning_rate": 1.9360598849449502e-05,
      "loss": 0.8981,
      "step": 4091
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7327523231506348,
      "learning_rate": 1.9360205934893993e-05,
      "loss": 0.6693,
      "step": 4092
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.0914437770843506,
      "learning_rate": 1.9359812903641284e-05,
      "loss": 0.8085,
      "step": 4093
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.793230652809143,
      "learning_rate": 1.9359419755696275e-05,
      "loss": 0.6437,
      "step": 4094
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9014922380447388,
      "learning_rate": 1.9359026491063865e-05,
      "loss": 0.7786,
      "step": 4095
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9909958839416504,
      "learning_rate": 1.935863310974896e-05,
      "loss": 0.8938,
      "step": 4096
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9367308616638184,
      "learning_rate": 1.9358239611756465e-05,
      "loss": 0.9641,
      "step": 4097
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4130171537399292,
      "learning_rate": 1.9357845997091284e-05,
      "loss": 0.777,
      "step": 4098
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9795866012573242,
      "learning_rate": 1.9357452265758323e-05,
      "loss": 0.6193,
      "step": 4099
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4836516380310059,
      "learning_rate": 1.9357058417762494e-05,
      "loss": 0.6536,
      "step": 4100
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6160871982574463,
      "learning_rate": 1.9356664453108703e-05,
      "loss": 0.6118,
      "step": 4101
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.8215543031692505,
      "learning_rate": 1.9356270371801867e-05,
      "loss": 0.9027,
      "step": 4102
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.584923505783081,
      "learning_rate": 1.93558761738469e-05,
      "loss": 0.6605,
      "step": 4103
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7477866411209106,
      "learning_rate": 1.9355481859248707e-05,
      "loss": 0.6955,
      "step": 4104
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6647318601608276,
      "learning_rate": 1.935508742801221e-05,
      "loss": 0.8279,
      "step": 4105
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5777537822723389,
      "learning_rate": 1.935469288014233e-05,
      "loss": 0.6471,
      "step": 4106
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5935180187225342,
      "learning_rate": 1.935429821564398e-05,
      "loss": 0.7468,
      "step": 4107
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.442557692527771,
      "learning_rate": 1.9353903434522084e-05,
      "loss": 0.6567,
      "step": 4108
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.7790277004241943,
      "learning_rate": 1.9353508536781566e-05,
      "loss": 0.8428,
      "step": 4109
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3926008939743042,
      "learning_rate": 1.935311352242734e-05,
      "loss": 0.7648,
      "step": 4110
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5409256219863892,
      "learning_rate": 1.935271839146434e-05,
      "loss": 0.7715,
      "step": 4111
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.748059868812561,
      "learning_rate": 1.9352323143897495e-05,
      "loss": 0.7664,
      "step": 4112
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8607004880905151,
      "learning_rate": 1.935192777973172e-05,
      "loss": 0.7532,
      "step": 4113
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.625313639640808,
      "learning_rate": 1.9351532298971954e-05,
      "loss": 0.8498,
      "step": 4114
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7391563653945923,
      "learning_rate": 1.9351136701623127e-05,
      "loss": 0.7485,
      "step": 4115
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7402573823928833,
      "learning_rate": 1.9350740987690165e-05,
      "loss": 0.5413,
      "step": 4116
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5388737916946411,
      "learning_rate": 1.9350345157178005e-05,
      "loss": 0.7065,
      "step": 4117
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7637115716934204,
      "learning_rate": 1.9349949210091584e-05,
      "loss": 0.7065,
      "step": 4118
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2747962474822998,
      "learning_rate": 1.934955314643584e-05,
      "loss": 0.684,
      "step": 4119
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.654841661453247,
      "learning_rate": 1.9349156966215705e-05,
      "loss": 0.6337,
      "step": 4120
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.786617398262024,
      "learning_rate": 1.934876066943612e-05,
      "loss": 0.9367,
      "step": 4121
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2144665718078613,
      "learning_rate": 1.934836425610203e-05,
      "loss": 0.9401,
      "step": 4122
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7403833866119385,
      "learning_rate": 1.934796772621837e-05,
      "loss": 0.6929,
      "step": 4123
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.116044521331787,
      "learning_rate": 1.934757107979009e-05,
      "loss": 0.845,
      "step": 4124
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6195369958877563,
      "learning_rate": 1.9347174316822132e-05,
      "loss": 0.878,
      "step": 4125
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8169608116149902,
      "learning_rate": 1.9346777437319444e-05,
      "loss": 0.8384,
      "step": 4126
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.041151762008667,
      "learning_rate": 1.9346380441286974e-05,
      "loss": 0.7007,
      "step": 4127
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.595874547958374,
      "learning_rate": 1.934598332872967e-05,
      "loss": 0.7045,
      "step": 4128
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7411166429519653,
      "learning_rate": 1.9345586099652483e-05,
      "loss": 0.967,
      "step": 4129
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6409512758255005,
      "learning_rate": 1.9345188754060368e-05,
      "loss": 0.6451,
      "step": 4130
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9258809089660645,
      "learning_rate": 1.9344791291958274e-05,
      "loss": 0.6894,
      "step": 4131
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.0567233562469482,
      "learning_rate": 1.9344393713351163e-05,
      "loss": 0.7733,
      "step": 4132
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6789960861206055,
      "learning_rate": 1.9343996018243987e-05,
      "loss": 0.5801,
      "step": 4133
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5027052164077759,
      "learning_rate": 1.9343598206641707e-05,
      "loss": 0.8071,
      "step": 4134
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.977447271347046,
      "learning_rate": 1.9343200278549283e-05,
      "loss": 0.7134,
      "step": 4135
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5867277383804321,
      "learning_rate": 1.934280223397167e-05,
      "loss": 0.7918,
      "step": 4136
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9623619318008423,
      "learning_rate": 1.934240407291384e-05,
      "loss": 0.7702,
      "step": 4137
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6713041067123413,
      "learning_rate": 1.934200579538075e-05,
      "loss": 0.6142,
      "step": 4138
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.973301649093628,
      "learning_rate": 1.934160740137736e-05,
      "loss": 0.833,
      "step": 4139
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.93295419216156,
      "learning_rate": 1.9341208890908653e-05,
      "loss": 0.6794,
      "step": 4140
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6946027278900146,
      "learning_rate": 1.9340810263979583e-05,
      "loss": 0.9322,
      "step": 4141
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.528892993927002,
      "learning_rate": 1.934041152059513e-05,
      "loss": 0.7411,
      "step": 4142
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5780450105667114,
      "learning_rate": 1.9340012660760255e-05,
      "loss": 0.6813,
      "step": 4143
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6099509000778198,
      "learning_rate": 1.9339613684479945e-05,
      "loss": 0.9227,
      "step": 4144
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.754671335220337,
      "learning_rate": 1.933921459175916e-05,
      "loss": 0.6922,
      "step": 4145
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5833840370178223,
      "learning_rate": 1.933881538260288e-05,
      "loss": 0.6343,
      "step": 4146
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.812504768371582,
      "learning_rate": 1.9338416057016086e-05,
      "loss": 0.8389,
      "step": 4147
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6128121614456177,
      "learning_rate": 1.9338016615003756e-05,
      "loss": 0.7009,
      "step": 4148
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7303094863891602,
      "learning_rate": 1.9337617056570867e-05,
      "loss": 0.687,
      "step": 4149
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4446240663528442,
      "learning_rate": 1.9337217381722396e-05,
      "loss": 0.5799,
      "step": 4150
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8932621479034424,
      "learning_rate": 1.9336817590463338e-05,
      "loss": 0.8419,
      "step": 4151
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.693373441696167,
      "learning_rate": 1.9336417682798666e-05,
      "loss": 0.73,
      "step": 4152
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.761362075805664,
      "learning_rate": 1.9336017658733375e-05,
      "loss": 0.8267,
      "step": 4153
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7526593208312988,
      "learning_rate": 1.933561751827245e-05,
      "loss": 0.7849,
      "step": 4154
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8916220664978027,
      "learning_rate": 1.933521726142087e-05,
      "loss": 0.8873,
      "step": 4155
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2359392642974854,
      "learning_rate": 1.9334816888183634e-05,
      "loss": 0.8522,
      "step": 4156
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7944059371948242,
      "learning_rate": 1.9334416398565735e-05,
      "loss": 0.8605,
      "step": 4157
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5168805122375488,
      "learning_rate": 1.933401579257216e-05,
      "loss": 0.8751,
      "step": 4158
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8167704343795776,
      "learning_rate": 1.933361507020791e-05,
      "loss": 0.946,
      "step": 4159
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9539884328842163,
      "learning_rate": 1.9333214231477975e-05,
      "loss": 0.7572,
      "step": 4160
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4172075986862183,
      "learning_rate": 1.9332813276387353e-05,
      "loss": 0.7309,
      "step": 4161
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4346672296524048,
      "learning_rate": 1.9332412204941052e-05,
      "loss": 0.8277,
      "step": 4162
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5727099180221558,
      "learning_rate": 1.933201101714406e-05,
      "loss": 0.9149,
      "step": 4163
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.579258918762207,
      "learning_rate": 1.9331609713001385e-05,
      "loss": 0.8804,
      "step": 4164
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4372646808624268,
      "learning_rate": 1.933120829251803e-05,
      "loss": 0.7515,
      "step": 4165
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6271528005599976,
      "learning_rate": 1.9330806755698995e-05,
      "loss": 0.8238,
      "step": 4166
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.1212193965911865,
      "learning_rate": 1.9330405102549297e-05,
      "loss": 0.7717,
      "step": 4167
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5101172924041748,
      "learning_rate": 1.9330003333073932e-05,
      "loss": 0.8352,
      "step": 4168
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.686665654182434,
      "learning_rate": 1.9329601447277914e-05,
      "loss": 0.791,
      "step": 4169
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8898507356643677,
      "learning_rate": 1.9329199445166255e-05,
      "loss": 0.7957,
      "step": 4170
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8263682126998901,
      "learning_rate": 1.932879732674396e-05,
      "loss": 0.7451,
      "step": 4171
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7388665676116943,
      "learning_rate": 1.9328395092016055e-05,
      "loss": 0.8282,
      "step": 4172
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8343604803085327,
      "learning_rate": 1.9327992740987543e-05,
      "loss": 0.8183,
      "step": 4173
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5495468378067017,
      "learning_rate": 1.9327590273663444e-05,
      "loss": 0.8263,
      "step": 4174
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.663567304611206,
      "learning_rate": 1.9327187690048775e-05,
      "loss": 0.7283,
      "step": 4175
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7343195676803589,
      "learning_rate": 1.9326784990148558e-05,
      "loss": 0.7204,
      "step": 4176
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8790730237960815,
      "learning_rate": 1.932638217396781e-05,
      "loss": 0.8675,
      "step": 4177
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.784278392791748,
      "learning_rate": 1.9325979241511557e-05,
      "loss": 0.6533,
      "step": 4178
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6718217134475708,
      "learning_rate": 1.9325576192784822e-05,
      "loss": 0.7636,
      "step": 4179
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5611703395843506,
      "learning_rate": 1.9325173027792627e-05,
      "loss": 0.6725,
      "step": 4180
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6088948249816895,
      "learning_rate": 1.9324769746539996e-05,
      "loss": 0.809,
      "step": 4181
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7368525266647339,
      "learning_rate": 1.9324366349031965e-05,
      "loss": 0.9163,
      "step": 4182
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6230058670043945,
      "learning_rate": 1.9323962835273558e-05,
      "loss": 0.6198,
      "step": 4183
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9403997659683228,
      "learning_rate": 1.9323559205269805e-05,
      "loss": 0.8378,
      "step": 4184
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6672853231430054,
      "learning_rate": 1.932315545902574e-05,
      "loss": 0.7873,
      "step": 4185
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7999526262283325,
      "learning_rate": 1.93227515965464e-05,
      "loss": 0.7363,
      "step": 4186
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6047937870025635,
      "learning_rate": 1.932234761783681e-05,
      "loss": 0.7243,
      "step": 4187
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4876956939697266,
      "learning_rate": 1.932194352290202e-05,
      "loss": 0.6964,
      "step": 4188
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8292052745819092,
      "learning_rate": 1.932153931174706e-05,
      "loss": 0.7321,
      "step": 4189
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.860658884048462,
      "learning_rate": 1.9321134984376968e-05,
      "loss": 0.6403,
      "step": 4190
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9267117977142334,
      "learning_rate": 1.932073054079679e-05,
      "loss": 0.9225,
      "step": 4191
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6047621965408325,
      "learning_rate": 1.9320325981011564e-05,
      "loss": 0.6833,
      "step": 4192
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5181206464767456,
      "learning_rate": 1.9319921305026338e-05,
      "loss": 0.6793,
      "step": 4193
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5764902830123901,
      "learning_rate": 1.931951651284615e-05,
      "loss": 0.6323,
      "step": 4194
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.689390778541565,
      "learning_rate": 1.931911160447606e-05,
      "loss": 0.7599,
      "step": 4195
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8377835750579834,
      "learning_rate": 1.9318706579921102e-05,
      "loss": 0.8248,
      "step": 4196
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6765233278274536,
      "learning_rate": 1.9318301439186333e-05,
      "loss": 0.743,
      "step": 4197
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9284987449645996,
      "learning_rate": 1.93178961822768e-05,
      "loss": 0.886,
      "step": 4198
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6438905000686646,
      "learning_rate": 1.9317490809197563e-05,
      "loss": 0.8226,
      "step": 4199
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8891545534133911,
      "learning_rate": 1.9317085319953666e-05,
      "loss": 0.9271,
      "step": 4200
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7377700805664062,
      "learning_rate": 1.931667971455017e-05,
      "loss": 0.7136,
      "step": 4201
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7749717235565186,
      "learning_rate": 1.9316273992992133e-05,
      "loss": 0.7386,
      "step": 4202
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.9318699836730957,
      "learning_rate": 1.931586815528461e-05,
      "loss": 0.7452,
      "step": 4203
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7179999351501465,
      "learning_rate": 1.9315462201432666e-05,
      "loss": 0.904,
      "step": 4204
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7318161725997925,
      "learning_rate": 1.9315056131441353e-05,
      "loss": 0.7733,
      "step": 4205
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4778685569763184,
      "learning_rate": 1.9314649945315742e-05,
      "loss": 0.6185,
      "step": 4206
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5766537189483643,
      "learning_rate": 1.931424364306089e-05,
      "loss": 0.7682,
      "step": 4207
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4484020471572876,
      "learning_rate": 1.931383722468187e-05,
      "loss": 0.7241,
      "step": 4208
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6658216714859009,
      "learning_rate": 1.931343069018375e-05,
      "loss": 0.8432,
      "step": 4209
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8325108289718628,
      "learning_rate": 1.9313024039571586e-05,
      "loss": 0.8889,
      "step": 4210
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6251713037490845,
      "learning_rate": 1.931261727285046e-05,
      "loss": 0.7996,
      "step": 4211
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.748856782913208,
      "learning_rate": 1.931221039002544e-05,
      "loss": 0.699,
      "step": 4212
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5808440446853638,
      "learning_rate": 1.9311803391101592e-05,
      "loss": 0.8174,
      "step": 4213
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0031015872955322,
      "learning_rate": 1.9311396276084e-05,
      "loss": 0.7648,
      "step": 4214
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.826385259628296,
      "learning_rate": 1.9310989044977736e-05,
      "loss": 0.9215,
      "step": 4215
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6653473377227783,
      "learning_rate": 1.931058169778788e-05,
      "loss": 0.7484,
      "step": 4216
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5074950456619263,
      "learning_rate": 1.9310174234519502e-05,
      "loss": 0.7587,
      "step": 4217
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4797581434249878,
      "learning_rate": 1.9309766655177687e-05,
      "loss": 0.8078,
      "step": 4218
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5286599397659302,
      "learning_rate": 1.930935895976752e-05,
      "loss": 0.8725,
      "step": 4219
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.2355306148529053,
      "learning_rate": 1.930895114829408e-05,
      "loss": 0.6239,
      "step": 4220
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8217464685440063,
      "learning_rate": 1.9308543220762452e-05,
      "loss": 0.7809,
      "step": 4221
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.434189796447754,
      "learning_rate": 1.9308135177177722e-05,
      "loss": 0.7137,
      "step": 4222
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6217087507247925,
      "learning_rate": 1.9307727017544975e-05,
      "loss": 0.6656,
      "step": 4223
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5273241996765137,
      "learning_rate": 1.9307318741869304e-05,
      "loss": 0.7538,
      "step": 4224
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6756106615066528,
      "learning_rate": 1.9306910350155798e-05,
      "loss": 0.8483,
      "step": 4225
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.695133924484253,
      "learning_rate": 1.9306501842409545e-05,
      "loss": 0.7437,
      "step": 4226
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6355735063552856,
      "learning_rate": 1.930609321863564e-05,
      "loss": 0.5978,
      "step": 4227
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6190440654754639,
      "learning_rate": 1.930568447883918e-05,
      "loss": 0.619,
      "step": 4228
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7099733352661133,
      "learning_rate": 1.9305275623025257e-05,
      "loss": 0.678,
      "step": 4229
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.608973503112793,
      "learning_rate": 1.9304866651198973e-05,
      "loss": 0.7205,
      "step": 4230
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4636712074279785,
      "learning_rate": 1.9304457563365422e-05,
      "loss": 0.7194,
      "step": 4231
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5537151098251343,
      "learning_rate": 1.9304048359529702e-05,
      "loss": 0.8044,
      "step": 4232
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5929607152938843,
      "learning_rate": 1.9303639039696927e-05,
      "loss": 0.6772,
      "step": 4233
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.596651554107666,
      "learning_rate": 1.9303229603872188e-05,
      "loss": 0.8391,
      "step": 4234
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6026662588119507,
      "learning_rate": 1.9302820052060592e-05,
      "loss": 0.7355,
      "step": 4235
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6323928833007812,
      "learning_rate": 1.930241038426725e-05,
      "loss": 0.6387,
      "step": 4236
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8558623790740967,
      "learning_rate": 1.9302000600497265e-05,
      "loss": 0.7997,
      "step": 4237
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9333007335662842,
      "learning_rate": 1.930159070075575e-05,
      "loss": 0.6937,
      "step": 4238
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4357218742370605,
      "learning_rate": 1.930118068504781e-05,
      "loss": 0.8332,
      "step": 4239
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.715769648551941,
      "learning_rate": 1.9300770553378555e-05,
      "loss": 0.7971,
      "step": 4240
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4251182079315186,
      "learning_rate": 1.9300360305753107e-05,
      "loss": 0.6025,
      "step": 4241
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9338622093200684,
      "learning_rate": 1.9299949942176576e-05,
      "loss": 0.7236,
      "step": 4242
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.427950143814087,
      "learning_rate": 1.929953946265408e-05,
      "loss": 0.7737,
      "step": 4243
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.745498538017273,
      "learning_rate": 1.9299128867190732e-05,
      "loss": 0.8137,
      "step": 4244
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.780079960823059,
      "learning_rate": 1.9298718155791654e-05,
      "loss": 0.7625,
      "step": 4245
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9281809329986572,
      "learning_rate": 1.9298307328461972e-05,
      "loss": 0.6787,
      "step": 4246
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.999129295349121,
      "learning_rate": 1.92978963852068e-05,
      "loss": 0.7356,
      "step": 4247
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6758201122283936,
      "learning_rate": 1.929748532603126e-05,
      "loss": 0.6957,
      "step": 4248
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4602477550506592,
      "learning_rate": 1.9297074150940483e-05,
      "loss": 0.6435,
      "step": 4249
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9427319765090942,
      "learning_rate": 1.9296662859939595e-05,
      "loss": 0.7873,
      "step": 4250
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6001898050308228,
      "learning_rate": 1.9296251453033718e-05,
      "loss": 0.5415,
      "step": 4251
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3446807861328125,
      "learning_rate": 1.9295839930227987e-05,
      "loss": 0.7249,
      "step": 4252
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5269511938095093,
      "learning_rate": 1.9295428291527532e-05,
      "loss": 0.7693,
      "step": 4253
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7770133018493652,
      "learning_rate": 1.9295016536937483e-05,
      "loss": 0.7092,
      "step": 4254
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7089279890060425,
      "learning_rate": 1.929460466646297e-05,
      "loss": 0.7403,
      "step": 4255
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9834864139556885,
      "learning_rate": 1.9294192680109137e-05,
      "loss": 0.89,
      "step": 4256
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.714895248413086,
      "learning_rate": 1.9293780577881113e-05,
      "loss": 0.7246,
      "step": 4257
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5225882530212402,
      "learning_rate": 1.929336835978404e-05,
      "loss": 0.7955,
      "step": 4258
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.689738392829895,
      "learning_rate": 1.9292956025823052e-05,
      "loss": 0.7042,
      "step": 4259
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.1721537113189697,
      "learning_rate": 1.92925435760033e-05,
      "loss": 0.6798,
      "step": 4260
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9044746160507202,
      "learning_rate": 1.929213101032991e-05,
      "loss": 0.8595,
      "step": 4261
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7266994714736938,
      "learning_rate": 1.9291718328808042e-05,
      "loss": 0.9475,
      "step": 4262
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7360014915466309,
      "learning_rate": 1.9291305531442833e-05,
      "loss": 0.8476,
      "step": 4263
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8935537338256836,
      "learning_rate": 1.929089261823943e-05,
      "loss": 0.8494,
      "step": 4264
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6070499420166016,
      "learning_rate": 1.9290479589202983e-05,
      "loss": 0.7815,
      "step": 4265
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6195158958435059,
      "learning_rate": 1.9290066444338634e-05,
      "loss": 0.8341,
      "step": 4266
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.610994577407837,
      "learning_rate": 1.9289653183651546e-05,
      "loss": 0.7867,
      "step": 4267
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7105462551116943,
      "learning_rate": 1.9289239807146862e-05,
      "loss": 0.8373,
      "step": 4268
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6837568283081055,
      "learning_rate": 1.928882631482974e-05,
      "loss": 0.7899,
      "step": 4269
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8752832412719727,
      "learning_rate": 1.9288412706705334e-05,
      "loss": 0.7863,
      "step": 4270
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6797475814819336,
      "learning_rate": 1.92879989827788e-05,
      "loss": 0.7044,
      "step": 4271
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3870781660079956,
      "learning_rate": 1.9287585143055297e-05,
      "loss": 0.8098,
      "step": 4272
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.040675640106201,
      "learning_rate": 1.9287171187539984e-05,
      "loss": 0.9744,
      "step": 4273
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.486303448677063,
      "learning_rate": 1.9286757116238022e-05,
      "loss": 0.6206,
      "step": 4274
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.607062578201294,
      "learning_rate": 1.928634292915457e-05,
      "loss": 0.7192,
      "step": 4275
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.620355248451233,
      "learning_rate": 1.92859286262948e-05,
      "loss": 0.8367,
      "step": 4276
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6652250289916992,
      "learning_rate": 1.928551420766387e-05,
      "loss": 0.6633,
      "step": 4277
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.744903564453125,
      "learning_rate": 1.928509967326695e-05,
      "loss": 0.761,
      "step": 4278
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6517136096954346,
      "learning_rate": 1.9284685023109208e-05,
      "loss": 0.6441,
      "step": 4279
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4284940958023071,
      "learning_rate": 1.9284270257195814e-05,
      "loss": 0.6505,
      "step": 4280
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6534533500671387,
      "learning_rate": 1.9283855375531936e-05,
      "loss": 0.6515,
      "step": 4281
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.0277676582336426,
      "learning_rate": 1.928344037812275e-05,
      "loss": 0.7569,
      "step": 4282
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7994216680526733,
      "learning_rate": 1.9283025264973424e-05,
      "loss": 0.75,
      "step": 4283
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8007240295410156,
      "learning_rate": 1.928261003608914e-05,
      "loss": 0.8296,
      "step": 4284
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.745962142944336,
      "learning_rate": 1.9282194691475075e-05,
      "loss": 0.6974,
      "step": 4285
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.63766610622406,
      "learning_rate": 1.9281779231136404e-05,
      "loss": 0.7628,
      "step": 4286
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4246023893356323,
      "learning_rate": 1.9281363655078307e-05,
      "loss": 0.6428,
      "step": 4287
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6700541973114014,
      "learning_rate": 1.9280947963305965e-05,
      "loss": 0.8327,
      "step": 4288
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8145554065704346,
      "learning_rate": 1.9280532155824562e-05,
      "loss": 0.831,
      "step": 4289
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.781805157661438,
      "learning_rate": 1.928011623263928e-05,
      "loss": 0.8401,
      "step": 4290
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4578866958618164,
      "learning_rate": 1.9279700193755305e-05,
      "loss": 0.7014,
      "step": 4291
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4613146781921387,
      "learning_rate": 1.9279284039177828e-05,
      "loss": 0.7267,
      "step": 4292
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6920448541641235,
      "learning_rate": 1.9278867768912033e-05,
      "loss": 0.5971,
      "step": 4293
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.9871375560760498,
      "learning_rate": 1.927845138296311e-05,
      "loss": 0.7693,
      "step": 4294
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.940824031829834,
      "learning_rate": 1.927803488133625e-05,
      "loss": 0.8051,
      "step": 4295
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4524650573730469,
      "learning_rate": 1.9277618264036652e-05,
      "loss": 0.7083,
      "step": 4296
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8585950136184692,
      "learning_rate": 1.9277201531069497e-05,
      "loss": 0.7955,
      "step": 4297
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7593553066253662,
      "learning_rate": 1.9276784682439996e-05,
      "loss": 0.6347,
      "step": 4298
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8024306297302246,
      "learning_rate": 1.9276367718153334e-05,
      "loss": 0.7057,
      "step": 4299
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4369741678237915,
      "learning_rate": 1.9275950638214714e-05,
      "loss": 0.6846,
      "step": 4300
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7978856563568115,
      "learning_rate": 1.9275533442629334e-05,
      "loss": 0.7477,
      "step": 4301
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6581469774246216,
      "learning_rate": 1.92751161314024e-05,
      "loss": 0.6543,
      "step": 4302
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8089838027954102,
      "learning_rate": 1.9274698704539113e-05,
      "loss": 0.7017,
      "step": 4303
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8779633045196533,
      "learning_rate": 1.9274281162044675e-05,
      "loss": 0.8506,
      "step": 4304
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7805774211883545,
      "learning_rate": 1.9273863503924292e-05,
      "loss": 0.7764,
      "step": 4305
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7760742902755737,
      "learning_rate": 1.927344573018317e-05,
      "loss": 0.6981,
      "step": 4306
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.532816767692566,
      "learning_rate": 1.9273027840826524e-05,
      "loss": 0.8202,
      "step": 4307
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.821070909500122,
      "learning_rate": 1.9272609835859556e-05,
      "loss": 0.5732,
      "step": 4308
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.714169979095459,
      "learning_rate": 1.9272191715287483e-05,
      "loss": 0.6886,
      "step": 4309
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.563713788986206,
      "learning_rate": 1.9271773479115517e-05,
      "loss": 0.6167,
      "step": 4310
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.901546835899353,
      "learning_rate": 1.9271355127348866e-05,
      "loss": 0.901,
      "step": 4311
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8754301071166992,
      "learning_rate": 1.9270936659992755e-05,
      "loss": 0.72,
      "step": 4312
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7760637998580933,
      "learning_rate": 1.9270518077052392e-05,
      "loss": 0.822,
      "step": 4313
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.243523597717285,
      "learning_rate": 1.9270099378533008e-05,
      "loss": 0.9793,
      "step": 4314
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.1240170001983643,
      "learning_rate": 1.926968056443981e-05,
      "loss": 0.8896,
      "step": 4315
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.524345874786377,
      "learning_rate": 1.9269261634778026e-05,
      "loss": 0.8222,
      "step": 4316
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7796759605407715,
      "learning_rate": 1.9268842589552877e-05,
      "loss": 0.7708,
      "step": 4317
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4314565658569336,
      "learning_rate": 1.9268423428769595e-05,
      "loss": 0.7058,
      "step": 4318
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5101033449172974,
      "learning_rate": 1.9268004152433394e-05,
      "loss": 0.6153,
      "step": 4319
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4967681169509888,
      "learning_rate": 1.9267584760549506e-05,
      "loss": 0.8048,
      "step": 4320
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7474318742752075,
      "learning_rate": 1.9267165253123164e-05,
      "loss": 0.8593,
      "step": 4321
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4382888078689575,
      "learning_rate": 1.926674563015959e-05,
      "loss": 0.6994,
      "step": 4322
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8536466360092163,
      "learning_rate": 1.9266325891664022e-05,
      "loss": 0.8272,
      "step": 4323
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5299726724624634,
      "learning_rate": 1.926590603764169e-05,
      "loss": 0.7521,
      "step": 4324
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4815322160720825,
      "learning_rate": 1.9265486068097833e-05,
      "loss": 0.8193,
      "step": 4325
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.646371603012085,
      "learning_rate": 1.9265065983037684e-05,
      "loss": 0.7504,
      "step": 4326
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.090384006500244,
      "learning_rate": 1.9264645782466473e-05,
      "loss": 1.0145,
      "step": 4327
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.614535927772522,
      "learning_rate": 1.926422546638945e-05,
      "loss": 0.8264,
      "step": 4328
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7528225183486938,
      "learning_rate": 1.9263805034811852e-05,
      "loss": 0.7973,
      "step": 4329
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4983162879943848,
      "learning_rate": 1.9263384487738917e-05,
      "loss": 0.9285,
      "step": 4330
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.04622745513916,
      "learning_rate": 1.9262963825175896e-05,
      "loss": 0.7299,
      "step": 4331
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5189875364303589,
      "learning_rate": 1.9262543047128024e-05,
      "loss": 0.5382,
      "step": 4332
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.604175090789795,
      "learning_rate": 1.926212215360055e-05,
      "loss": 0.7991,
      "step": 4333
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8527389764785767,
      "learning_rate": 1.9261701144598724e-05,
      "loss": 0.7499,
      "step": 4334
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4751464128494263,
      "learning_rate": 1.9261280020127797e-05,
      "loss": 0.6738,
      "step": 4335
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.736393928527832,
      "learning_rate": 1.926085878019301e-05,
      "loss": 0.7036,
      "step": 4336
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2959150075912476,
      "learning_rate": 1.9260437424799624e-05,
      "loss": 0.6398,
      "step": 4337
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.722982406616211,
      "learning_rate": 1.926001595395289e-05,
      "loss": 0.7712,
      "step": 4338
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.532947301864624,
      "learning_rate": 1.9259594367658057e-05,
      "loss": 0.733,
      "step": 4339
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.407886028289795,
      "learning_rate": 1.925917266592039e-05,
      "loss": 0.6038,
      "step": 4340
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5467206239700317,
      "learning_rate": 1.925875084874514e-05,
      "loss": 0.8306,
      "step": 4341
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.903018593788147,
      "learning_rate": 1.925832891613757e-05,
      "loss": 1.0115,
      "step": 4342
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4464833736419678,
      "learning_rate": 1.9257906868102936e-05,
      "loss": 0.714,
      "step": 4343
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5267258882522583,
      "learning_rate": 1.92574847046465e-05,
      "loss": 0.6461,
      "step": 4344
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6658114194869995,
      "learning_rate": 1.9257062425773534e-05,
      "loss": 0.8074,
      "step": 4345
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.887660026550293,
      "learning_rate": 1.9256640031489294e-05,
      "loss": 0.7855,
      "step": 4346
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.647383451461792,
      "learning_rate": 1.9256217521799045e-05,
      "loss": 0.626,
      "step": 4347
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5245307683944702,
      "learning_rate": 1.925579489670806e-05,
      "loss": 0.7728,
      "step": 4348
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.554071307182312,
      "learning_rate": 1.925537215622161e-05,
      "loss": 0.822,
      "step": 4349
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8054437637329102,
      "learning_rate": 1.9254949300344952e-05,
      "loss": 0.7014,
      "step": 4350
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.818310022354126,
      "learning_rate": 1.9254526329083377e-05,
      "loss": 0.8078,
      "step": 4351
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.734984040260315,
      "learning_rate": 1.9254103242442145e-05,
      "loss": 0.6742,
      "step": 4352
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7020268440246582,
      "learning_rate": 1.925368004042653e-05,
      "loss": 0.8588,
      "step": 4353
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7300639152526855,
      "learning_rate": 1.925325672304182e-05,
      "loss": 0.7876,
      "step": 4354
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7458571195602417,
      "learning_rate": 1.9252833290293282e-05,
      "loss": 0.572,
      "step": 4355
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7957299947738647,
      "learning_rate": 1.9252409742186196e-05,
      "loss": 0.6994,
      "step": 4356
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6708884239196777,
      "learning_rate": 1.925198607872585e-05,
      "loss": 0.83,
      "step": 4357
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5226831436157227,
      "learning_rate": 1.9251562299917517e-05,
      "loss": 0.8655,
      "step": 4358
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.694014072418213,
      "learning_rate": 1.9251138405766488e-05,
      "loss": 0.6905,
      "step": 4359
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.539077639579773,
      "learning_rate": 1.9250714396278042e-05,
      "loss": 0.7028,
      "step": 4360
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.550165057182312,
      "learning_rate": 1.9250290271457468e-05,
      "loss": 0.643,
      "step": 4361
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6725428104400635,
      "learning_rate": 1.9249866031310055e-05,
      "loss": 0.7803,
      "step": 4362
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.574925422668457,
      "learning_rate": 1.9249441675841085e-05,
      "loss": 0.7432,
      "step": 4363
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6019704341888428,
      "learning_rate": 1.9249017205055856e-05,
      "loss": 0.7442,
      "step": 4364
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4834235906600952,
      "learning_rate": 1.9248592618959662e-05,
      "loss": 0.6538,
      "step": 4365
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7430588006973267,
      "learning_rate": 1.924816791755779e-05,
      "loss": 0.875,
      "step": 4366
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9875848293304443,
      "learning_rate": 1.9247743100855535e-05,
      "loss": 0.9342,
      "step": 4367
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4943549633026123,
      "learning_rate": 1.92473181688582e-05,
      "loss": 0.8355,
      "step": 4368
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4715173244476318,
      "learning_rate": 1.9246893121571076e-05,
      "loss": 0.7783,
      "step": 4369
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6954963207244873,
      "learning_rate": 1.9246467958999467e-05,
      "loss": 0.7934,
      "step": 4370
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7270835638046265,
      "learning_rate": 1.924604268114867e-05,
      "loss": 0.7883,
      "step": 4371
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5916668176651,
      "learning_rate": 1.9245617288023986e-05,
      "loss": 0.8948,
      "step": 4372
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3840261697769165,
      "learning_rate": 1.9245191779630723e-05,
      "loss": 0.7275,
      "step": 4373
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4516340494155884,
      "learning_rate": 1.9244766155974185e-05,
      "loss": 0.7104,
      "step": 4374
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.833795189857483,
      "learning_rate": 1.9244340417059674e-05,
      "loss": 0.9235,
      "step": 4375
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.982468843460083,
      "learning_rate": 1.9243914562892505e-05,
      "loss": 0.8071,
      "step": 4376
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3862193822860718,
      "learning_rate": 1.9243488593477985e-05,
      "loss": 0.7884,
      "step": 4377
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.8773871660232544,
      "learning_rate": 1.924306250882142e-05,
      "loss": 0.9087,
      "step": 4378
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4965840578079224,
      "learning_rate": 1.9242636308928128e-05,
      "loss": 0.6846,
      "step": 4379
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.049064874649048,
      "learning_rate": 1.924220999380342e-05,
      "loss": 0.8711,
      "step": 4380
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4160383939743042,
      "learning_rate": 1.9241783563452604e-05,
      "loss": 0.7173,
      "step": 4381
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7033088207244873,
      "learning_rate": 1.9241357017881012e-05,
      "loss": 0.7882,
      "step": 4382
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6741445064544678,
      "learning_rate": 1.924093035709395e-05,
      "loss": 0.7571,
      "step": 4383
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3282028436660767,
      "learning_rate": 1.924050358109674e-05,
      "loss": 0.6223,
      "step": 4384
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3735177516937256,
      "learning_rate": 1.9240076689894705e-05,
      "loss": 0.7021,
      "step": 4385
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5385818481445312,
      "learning_rate": 1.9239649683493164e-05,
      "loss": 0.7207,
      "step": 4386
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6106226444244385,
      "learning_rate": 1.9239222561897443e-05,
      "loss": 0.7747,
      "step": 4387
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3350075483322144,
      "learning_rate": 1.9238795325112867e-05,
      "loss": 0.641,
      "step": 4388
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.086798667907715,
      "learning_rate": 1.9238367973144764e-05,
      "loss": 0.774,
      "step": 4389
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4620777368545532,
      "learning_rate": 1.923794050599846e-05,
      "loss": 0.7082,
      "step": 4390
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7090965509414673,
      "learning_rate": 1.923751292367928e-05,
      "loss": 0.683,
      "step": 4391
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5576971769332886,
      "learning_rate": 1.9237085226192564e-05,
      "loss": 0.8185,
      "step": 4392
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.648476243019104,
      "learning_rate": 1.9236657413543637e-05,
      "loss": 0.9105,
      "step": 4393
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.476611614227295,
      "learning_rate": 1.9236229485737837e-05,
      "loss": 0.6694,
      "step": 4394
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7321451902389526,
      "learning_rate": 1.9235801442780496e-05,
      "loss": 0.7865,
      "step": 4395
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.9075514078140259,
      "learning_rate": 1.923537328467695e-05,
      "loss": 0.6619,
      "step": 4396
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6535717248916626,
      "learning_rate": 1.923494501143254e-05,
      "loss": 0.6857,
      "step": 4397
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.7250429391860962,
      "learning_rate": 1.9234516623052605e-05,
      "loss": 0.8515,
      "step": 4398
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.489861011505127,
      "learning_rate": 1.9234088119542488e-05,
      "loss": 0.6682,
      "step": 4399
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4616965055465698,
      "learning_rate": 1.9233659500907526e-05,
      "loss": 0.7864,
      "step": 4400
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4782752990722656,
      "learning_rate": 1.923323076715306e-05,
      "loss": 0.6623,
      "step": 4401
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5402214527130127,
      "learning_rate": 1.923280191828445e-05,
      "loss": 0.8637,
      "step": 4402
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8295906782150269,
      "learning_rate": 1.9232372954307026e-05,
      "loss": 0.7664,
      "step": 4403
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6188794374465942,
      "learning_rate": 1.9231943875226146e-05,
      "loss": 0.8595,
      "step": 4404
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5646741390228271,
      "learning_rate": 1.9231514681047157e-05,
      "loss": 0.8567,
      "step": 4405
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9435747861862183,
      "learning_rate": 1.923108537177541e-05,
      "loss": 0.6671,
      "step": 4406
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7062342166900635,
      "learning_rate": 1.9230655947416254e-05,
      "loss": 0.7433,
      "step": 4407
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4525104761123657,
      "learning_rate": 1.9230226407975047e-05,
      "loss": 0.7645,
      "step": 4408
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.623658299446106,
      "learning_rate": 1.9229796753457144e-05,
      "loss": 0.7288,
      "step": 4409
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5329029560089111,
      "learning_rate": 1.92293669838679e-05,
      "loss": 0.7687,
      "step": 4410
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5068801641464233,
      "learning_rate": 1.9228937099212674e-05,
      "loss": 0.7698,
      "step": 4411
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6962783336639404,
      "learning_rate": 1.9228507099496823e-05,
      "loss": 0.8778,
      "step": 4412
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7382221221923828,
      "learning_rate": 1.9228076984725716e-05,
      "loss": 0.8156,
      "step": 4413
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.64566171169281,
      "learning_rate": 1.9227646754904702e-05,
      "loss": 0.7927,
      "step": 4414
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5401825904846191,
      "learning_rate": 1.9227216410039158e-05,
      "loss": 0.7137,
      "step": 4415
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6827203035354614,
      "learning_rate": 1.9226785950134444e-05,
      "loss": 0.5918,
      "step": 4416
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9642281532287598,
      "learning_rate": 1.9226355375195924e-05,
      "loss": 0.9303,
      "step": 4417
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7361201047897339,
      "learning_rate": 1.9225924685228967e-05,
      "loss": 0.6353,
      "step": 4418
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5822856426239014,
      "learning_rate": 1.9225493880238944e-05,
      "loss": 0.9457,
      "step": 4419
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.0105154514312744,
      "learning_rate": 1.922506296023123e-05,
      "loss": 0.7334,
      "step": 4420
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9350965023040771,
      "learning_rate": 1.9224631925211192e-05,
      "loss": 1.0209,
      "step": 4421
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6385377645492554,
      "learning_rate": 1.9224200775184206e-05,
      "loss": 0.8511,
      "step": 4422
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.616451382637024,
      "learning_rate": 1.9223769510155644e-05,
      "loss": 0.8443,
      "step": 4423
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.592620849609375,
      "learning_rate": 1.9223338130130885e-05,
      "loss": 0.7817,
      "step": 4424
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5374431610107422,
      "learning_rate": 1.9222906635115314e-05,
      "loss": 0.8777,
      "step": 4425
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.548531174659729,
      "learning_rate": 1.92224750251143e-05,
      "loss": 0.8345,
      "step": 4426
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5025837421417236,
      "learning_rate": 1.9222043300133227e-05,
      "loss": 0.7403,
      "step": 4427
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5442701578140259,
      "learning_rate": 1.9221611460177482e-05,
      "loss": 0.7579,
      "step": 4428
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6068192720413208,
      "learning_rate": 1.9221179505252447e-05,
      "loss": 0.8137,
      "step": 4429
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7998074293136597,
      "learning_rate": 1.92207474353635e-05,
      "loss": 0.8897,
      "step": 4430
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8265748023986816,
      "learning_rate": 1.9220315250516036e-05,
      "loss": 0.8558,
      "step": 4431
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7300522327423096,
      "learning_rate": 1.9219882950715444e-05,
      "loss": 0.8666,
      "step": 4432
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.511511206626892,
      "learning_rate": 1.921945053596711e-05,
      "loss": 0.8685,
      "step": 4433
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6706888675689697,
      "learning_rate": 1.9219018006276422e-05,
      "loss": 0.5522,
      "step": 4434
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8591148853302002,
      "learning_rate": 1.921858536164878e-05,
      "loss": 0.839,
      "step": 4435
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.966213345527649,
      "learning_rate": 1.9218152602089573e-05,
      "loss": 0.678,
      "step": 4436
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9437050819396973,
      "learning_rate": 1.9217719727604194e-05,
      "loss": 0.694,
      "step": 4437
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6239925622940063,
      "learning_rate": 1.921728673819805e-05,
      "loss": 0.73,
      "step": 4438
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.503610372543335,
      "learning_rate": 1.921685363387653e-05,
      "loss": 0.7566,
      "step": 4439
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9102022647857666,
      "learning_rate": 1.921642041464504e-05,
      "loss": 0.7472,
      "step": 4440
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7334895133972168,
      "learning_rate": 1.9215987080508972e-05,
      "loss": 0.6778,
      "step": 4441
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.530027985572815,
      "learning_rate": 1.9215553631473732e-05,
      "loss": 0.7596,
      "step": 4442
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.576404094696045,
      "learning_rate": 1.921512006754473e-05,
      "loss": 0.5921,
      "step": 4443
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4788058996200562,
      "learning_rate": 1.921468638872737e-05,
      "loss": 0.8075,
      "step": 4444
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9081735610961914,
      "learning_rate": 1.921425259502705e-05,
      "loss": 0.7477,
      "step": 4445
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.03943133354187,
      "learning_rate": 1.9213818686449187e-05,
      "loss": 0.7087,
      "step": 4446
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.423834204673767,
      "learning_rate": 1.921338466299919e-05,
      "loss": 0.6997,
      "step": 4447
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5996174812316895,
      "learning_rate": 1.9212950524682464e-05,
      "loss": 0.8963,
      "step": 4448
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7380293607711792,
      "learning_rate": 1.921251627150443e-05,
      "loss": 0.6957,
      "step": 4449
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6846321821212769,
      "learning_rate": 1.92120819034705e-05,
      "loss": 0.8223,
      "step": 4450
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.196873188018799,
      "learning_rate": 1.921164742058608e-05,
      "loss": 0.9998,
      "step": 4451
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5579842329025269,
      "learning_rate": 1.92112128228566e-05,
      "loss": 0.8262,
      "step": 4452
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7856777906417847,
      "learning_rate": 1.921077811028747e-05,
      "loss": 0.7448,
      "step": 4453
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4394347667694092,
      "learning_rate": 1.9210343282884114e-05,
      "loss": 0.5987,
      "step": 4454
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7518701553344727,
      "learning_rate": 1.920990834065195e-05,
      "loss": 0.6032,
      "step": 4455
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7229223251342773,
      "learning_rate": 1.92094732835964e-05,
      "loss": 0.7806,
      "step": 4456
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.577748417854309,
      "learning_rate": 1.9209038111722893e-05,
      "loss": 0.7279,
      "step": 4457
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5720393657684326,
      "learning_rate": 1.920860282503685e-05,
      "loss": 0.7012,
      "step": 4458
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7772233486175537,
      "learning_rate": 1.92081674235437e-05,
      "loss": 0.8515,
      "step": 4459
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5368560552597046,
      "learning_rate": 1.920773190724887e-05,
      "loss": 0.632,
      "step": 4460
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6144230365753174,
      "learning_rate": 1.920729627615779e-05,
      "loss": 0.6859,
      "step": 4461
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3535847663879395,
      "learning_rate": 1.920686053027589e-05,
      "loss": 0.7564,
      "step": 4462
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6725406646728516,
      "learning_rate": 1.9206424669608607e-05,
      "loss": 0.8292,
      "step": 4463
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5289263725280762,
      "learning_rate": 1.9205988694161372e-05,
      "loss": 0.968,
      "step": 4464
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8374124765396118,
      "learning_rate": 1.9205552603939618e-05,
      "loss": 0.7704,
      "step": 4465
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9334192276000977,
      "learning_rate": 1.9205116398948785e-05,
      "loss": 0.6944,
      "step": 4466
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7560275793075562,
      "learning_rate": 1.9204680079194312e-05,
      "loss": 0.6703,
      "step": 4467
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7728705406188965,
      "learning_rate": 1.920424364468164e-05,
      "loss": 0.7306,
      "step": 4468
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3914611339569092,
      "learning_rate": 1.9203807095416205e-05,
      "loss": 0.7641,
      "step": 4469
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7233729362487793,
      "learning_rate": 1.9203370431403453e-05,
      "loss": 0.7985,
      "step": 4470
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6862893104553223,
      "learning_rate": 1.9202933652648826e-05,
      "loss": 0.8212,
      "step": 4471
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.565631628036499,
      "learning_rate": 1.9202496759157775e-05,
      "loss": 0.8035,
      "step": 4472
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.604813814163208,
      "learning_rate": 1.9202059750935736e-05,
      "loss": 0.7543,
      "step": 4473
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7312002182006836,
      "learning_rate": 1.920162262798817e-05,
      "loss": 0.8215,
      "step": 4474
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3783422708511353,
      "learning_rate": 1.920118539032052e-05,
      "loss": 0.5283,
      "step": 4475
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5074965953826904,
      "learning_rate": 1.920074803793824e-05,
      "loss": 0.6552,
      "step": 4476
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.2493979930877686,
      "learning_rate": 1.9200310570846773e-05,
      "loss": 0.8505,
      "step": 4477
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.642016887664795,
      "learning_rate": 1.919987298905159e-05,
      "loss": 0.7567,
      "step": 4478
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6478478908538818,
      "learning_rate": 1.919943529255813e-05,
      "loss": 0.6279,
      "step": 4479
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3903080224990845,
      "learning_rate": 1.9198997481371862e-05,
      "loss": 0.8456,
      "step": 4480
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.9287655353546143,
      "learning_rate": 1.919855955549824e-05,
      "loss": 0.9244,
      "step": 4481
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4876914024353027,
      "learning_rate": 1.9198121514942722e-05,
      "loss": 0.7153,
      "step": 4482
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3381136655807495,
      "learning_rate": 1.919768335971077e-05,
      "loss": 0.5878,
      "step": 4483
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7508797645568848,
      "learning_rate": 1.9197245089807847e-05,
      "loss": 0.7954,
      "step": 4484
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5053629875183105,
      "learning_rate": 1.919680670523942e-05,
      "loss": 0.8572,
      "step": 4485
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2501546144485474,
      "learning_rate": 1.919636820601095e-05,
      "loss": 0.7196,
      "step": 4486
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5024001598358154,
      "learning_rate": 1.9195929592127905e-05,
      "loss": 0.6662,
      "step": 4487
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8791977167129517,
      "learning_rate": 1.9195490863595757e-05,
      "loss": 0.7117,
      "step": 4488
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6197277307510376,
      "learning_rate": 1.919505202041997e-05,
      "loss": 0.7464,
      "step": 4489
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.496887445449829,
      "learning_rate": 1.919461306260602e-05,
      "loss": 0.6947,
      "step": 4490
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.834929347038269,
      "learning_rate": 1.919417399015938e-05,
      "loss": 0.797,
      "step": 4491
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6147007942199707,
      "learning_rate": 1.919373480308552e-05,
      "loss": 0.8395,
      "step": 4492
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7553164958953857,
      "learning_rate": 1.9193295501389916e-05,
      "loss": 0.9555,
      "step": 4493
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7099837064743042,
      "learning_rate": 1.9192856085078046e-05,
      "loss": 0.7568,
      "step": 4494
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6333186626434326,
      "learning_rate": 1.919241655415539e-05,
      "loss": 0.8549,
      "step": 4495
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6774598360061646,
      "learning_rate": 1.9191976908627428e-05,
      "loss": 0.637,
      "step": 4496
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8414032459259033,
      "learning_rate": 1.919153714849964e-05,
      "loss": 0.7594,
      "step": 4497
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6296565532684326,
      "learning_rate": 1.9191097273777506e-05,
      "loss": 0.7293,
      "step": 4498
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.897648572921753,
      "learning_rate": 1.9190657284466513e-05,
      "loss": 0.8665,
      "step": 4499
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5977272987365723,
      "learning_rate": 1.9190217180572145e-05,
      "loss": 0.7999,
      "step": 4500
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4519188404083252,
      "learning_rate": 1.9189776962099895e-05,
      "loss": 0.6361,
      "step": 4501
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3822723627090454,
      "learning_rate": 1.918933662905524e-05,
      "loss": 0.792,
      "step": 4502
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4697096347808838,
      "learning_rate": 1.918889618144368e-05,
      "loss": 0.6989,
      "step": 4503
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6737028360366821,
      "learning_rate": 1.9188455619270702e-05,
      "loss": 0.8859,
      "step": 4504
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5164803266525269,
      "learning_rate": 1.91880149425418e-05,
      "loss": 0.619,
      "step": 4505
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.428580641746521,
      "learning_rate": 1.9187574151262464e-05,
      "loss": 0.7997,
      "step": 4506
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.976205587387085,
      "learning_rate": 1.9187133245438193e-05,
      "loss": 0.9274,
      "step": 4507
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.474961280822754,
      "learning_rate": 1.9186692225074484e-05,
      "loss": 0.7661,
      "step": 4508
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3200725317001343,
      "learning_rate": 1.9186251090176837e-05,
      "loss": 0.7859,
      "step": 4509
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1816773414611816,
      "learning_rate": 1.9185809840750748e-05,
      "loss": 0.7639,
      "step": 4510
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4698147773742676,
      "learning_rate": 1.918536847680172e-05,
      "loss": 0.6463,
      "step": 4511
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4929187297821045,
      "learning_rate": 1.9184926998335255e-05,
      "loss": 0.7806,
      "step": 4512
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5702131986618042,
      "learning_rate": 1.918448540535686e-05,
      "loss": 0.8205,
      "step": 4513
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8936309814453125,
      "learning_rate": 1.918404369787204e-05,
      "loss": 0.7891,
      "step": 4514
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.50968337059021,
      "learning_rate": 1.9183601875886294e-05,
      "loss": 0.7102,
      "step": 4515
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.623363971710205,
      "learning_rate": 1.918315993940514e-05,
      "loss": 0.8027,
      "step": 4516
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3535526990890503,
      "learning_rate": 1.9182717888434083e-05,
      "loss": 0.6009,
      "step": 4517
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4486079216003418,
      "learning_rate": 1.918227572297864e-05,
      "loss": 0.6406,
      "step": 4518
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8395527601242065,
      "learning_rate": 1.9181833443044314e-05,
      "loss": 0.7619,
      "step": 4519
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.108640193939209,
      "learning_rate": 1.9181391048636623e-05,
      "loss": 0.7144,
      "step": 4520
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.361088752746582,
      "learning_rate": 1.9180948539761086e-05,
      "loss": 0.5737,
      "step": 4521
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.718360424041748,
      "learning_rate": 1.918050591642322e-05,
      "loss": 0.7771,
      "step": 4522
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5635228157043457,
      "learning_rate": 1.9180063178628537e-05,
      "loss": 0.9019,
      "step": 4523
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5435675382614136,
      "learning_rate": 1.9179620326382564e-05,
      "loss": 0.7847,
      "step": 4524
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0304622650146484,
      "learning_rate": 1.9179177359690818e-05,
      "loss": 0.6616,
      "step": 4525
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.04359769821167,
      "learning_rate": 1.9178734278558822e-05,
      "loss": 0.909,
      "step": 4526
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0547637939453125,
      "learning_rate": 1.91782910829921e-05,
      "loss": 0.8156,
      "step": 4527
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6635246276855469,
      "learning_rate": 1.917784777299618e-05,
      "loss": 0.6239,
      "step": 4528
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4676775932312012,
      "learning_rate": 1.9177404348576583e-05,
      "loss": 0.7319,
      "step": 4529
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8724161386489868,
      "learning_rate": 1.9176960809738846e-05,
      "loss": 0.8868,
      "step": 4530
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7932047843933105,
      "learning_rate": 1.917651715648849e-05,
      "loss": 0.7964,
      "step": 4531
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9152941703796387,
      "learning_rate": 1.9176073388831054e-05,
      "loss": 0.7277,
      "step": 4532
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.858352780342102,
      "learning_rate": 1.9175629506772066e-05,
      "loss": 0.8077,
      "step": 4533
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.026623010635376,
      "learning_rate": 1.917518551031706e-05,
      "loss": 0.7956,
      "step": 4534
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5299270153045654,
      "learning_rate": 1.9174741399471573e-05,
      "loss": 0.8419,
      "step": 4535
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7836540937423706,
      "learning_rate": 1.9174297174241144e-05,
      "loss": 0.6104,
      "step": 4536
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9275873899459839,
      "learning_rate": 1.9173852834631302e-05,
      "loss": 1.0312,
      "step": 4537
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6778708696365356,
      "learning_rate": 1.91734083806476e-05,
      "loss": 0.7578,
      "step": 4538
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6145910024642944,
      "learning_rate": 1.917296381229557e-05,
      "loss": 0.7398,
      "step": 4539
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5765488147735596,
      "learning_rate": 1.917251912958076e-05,
      "loss": 0.647,
      "step": 4540
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8520177602767944,
      "learning_rate": 1.9172074332508707e-05,
      "loss": 0.7801,
      "step": 4541
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6098588705062866,
      "learning_rate": 1.917162942108496e-05,
      "loss": 0.7573,
      "step": 4542
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4232537746429443,
      "learning_rate": 1.917118439531507e-05,
      "loss": 0.7676,
      "step": 4543
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.531360149383545,
      "learning_rate": 1.917073925520458e-05,
      "loss": 0.8326,
      "step": 4544
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4583923816680908,
      "learning_rate": 1.9170294000759042e-05,
      "loss": 0.749,
      "step": 4545
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6205158233642578,
      "learning_rate": 1.9169848631984006e-05,
      "loss": 0.7907,
      "step": 4546
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7623878717422485,
      "learning_rate": 1.916940314888503e-05,
      "loss": 0.7733,
      "step": 4547
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3386335372924805,
      "learning_rate": 1.9168957551467656e-05,
      "loss": 0.6666,
      "step": 4548
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8106673955917358,
      "learning_rate": 1.916851183973745e-05,
      "loss": 0.8316,
      "step": 4549
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6153322458267212,
      "learning_rate": 1.916806601369997e-05,
      "loss": 0.8981,
      "step": 4550
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6198683977127075,
      "learning_rate": 1.9167620073360765e-05,
      "loss": 0.7895,
      "step": 4551
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7895218133926392,
      "learning_rate": 1.9167174018725402e-05,
      "loss": 0.8017,
      "step": 4552
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0875258445739746,
      "learning_rate": 1.9166727849799436e-05,
      "loss": 0.836,
      "step": 4553
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.537146806716919,
      "learning_rate": 1.9166281566588437e-05,
      "loss": 0.6971,
      "step": 4554
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9306265115737915,
      "learning_rate": 1.9165835169097962e-05,
      "loss": 0.9092,
      "step": 4555
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0640289783477783,
      "learning_rate": 1.916538865733358e-05,
      "loss": 0.9518,
      "step": 4556
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.206890106201172,
      "learning_rate": 1.916494203130086e-05,
      "loss": 0.8253,
      "step": 4557
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4975358247756958,
      "learning_rate": 1.9164495291005368e-05,
      "loss": 0.8477,
      "step": 4558
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6018924713134766,
      "learning_rate": 1.9164048436452672e-05,
      "loss": 0.7566,
      "step": 4559
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.526986837387085,
      "learning_rate": 1.9163601467648344e-05,
      "loss": 0.7643,
      "step": 4560
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.571946144104004,
      "learning_rate": 1.9163154384597956e-05,
      "loss": 0.8367,
      "step": 4561
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6958069801330566,
      "learning_rate": 1.9162707187307087e-05,
      "loss": 0.7498,
      "step": 4562
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.339910626411438,
      "learning_rate": 1.9162259875781305e-05,
      "loss": 0.762,
      "step": 4563
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4125642776489258,
      "learning_rate": 1.9161812450026192e-05,
      "loss": 0.7467,
      "step": 4564
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8139193058013916,
      "learning_rate": 1.9161364910047324e-05,
      "loss": 0.7685,
      "step": 4565
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5798418521881104,
      "learning_rate": 1.916091725585028e-05,
      "loss": 0.7728,
      "step": 4566
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6592448949813843,
      "learning_rate": 1.9160469487440647e-05,
      "loss": 0.5831,
      "step": 4567
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2747101783752441,
      "learning_rate": 1.9160021604823998e-05,
      "loss": 0.5786,
      "step": 4568
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4408715963363647,
      "learning_rate": 1.9159573608005923e-05,
      "loss": 0.7022,
      "step": 4569
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5821086168289185,
      "learning_rate": 1.9159125496992e-05,
      "loss": 0.6549,
      "step": 4570
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6709779500961304,
      "learning_rate": 1.915867727178783e-05,
      "loss": 0.5684,
      "step": 4571
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6077018976211548,
      "learning_rate": 1.915822893239899e-05,
      "loss": 0.7017,
      "step": 4572
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0832979679107666,
      "learning_rate": 1.9157780478831073e-05,
      "loss": 0.8976,
      "step": 4573
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.348098635673523,
      "learning_rate": 1.915733191108967e-05,
      "loss": 0.6243,
      "step": 4574
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7916418313980103,
      "learning_rate": 1.915688322918037e-05,
      "loss": 0.726,
      "step": 4575
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.122117757797241,
      "learning_rate": 1.9156434433108774e-05,
      "loss": 0.8,
      "step": 4576
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4946138858795166,
      "learning_rate": 1.915598552288047e-05,
      "loss": 0.5854,
      "step": 4577
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.916839838027954,
      "learning_rate": 1.9155536498501056e-05,
      "loss": 0.8498,
      "step": 4578
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8643677234649658,
      "learning_rate": 1.9155087359976138e-05,
      "loss": 0.7046,
      "step": 4579
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.9259750843048096,
      "learning_rate": 1.9154638107311304e-05,
      "loss": 0.8917,
      "step": 4580
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8437892198562622,
      "learning_rate": 1.915418874051216e-05,
      "loss": 0.9678,
      "step": 4581
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.79427969455719,
      "learning_rate": 1.9153739259584315e-05,
      "loss": 0.6742,
      "step": 4582
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.3475146293640137,
      "learning_rate": 1.9153289664533365e-05,
      "loss": 0.8485,
      "step": 4583
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7485054731369019,
      "learning_rate": 1.9152839955364918e-05,
      "loss": 0.8209,
      "step": 4584
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.595900535583496,
      "learning_rate": 1.915239013208458e-05,
      "loss": 0.7517,
      "step": 4585
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.562359094619751,
      "learning_rate": 1.9151940194697955e-05,
      "loss": 0.7509,
      "step": 4586
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5221123695373535,
      "learning_rate": 1.9151490143210656e-05,
      "loss": 0.7148,
      "step": 4587
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.7184584140777588,
      "learning_rate": 1.9151039977628297e-05,
      "loss": 0.6029,
      "step": 4588
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5774924755096436,
      "learning_rate": 1.915058969795649e-05,
      "loss": 0.8973,
      "step": 4589
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8053944110870361,
      "learning_rate": 1.9150139304200843e-05,
      "loss": 0.8018,
      "step": 4590
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.643210530281067,
      "learning_rate": 1.9149688796366978e-05,
      "loss": 0.8546,
      "step": 4591
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5407615900039673,
      "learning_rate": 1.9149238174460507e-05,
      "loss": 0.7638,
      "step": 4592
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6215314865112305,
      "learning_rate": 1.914878743848705e-05,
      "loss": 0.7271,
      "step": 4593
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.774770736694336,
      "learning_rate": 1.9148336588452226e-05,
      "loss": 0.7075,
      "step": 4594
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5777020454406738,
      "learning_rate": 1.9147885624361656e-05,
      "loss": 0.5514,
      "step": 4595
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5299785137176514,
      "learning_rate": 1.914743454622096e-05,
      "loss": 0.6908,
      "step": 4596
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.792807698249817,
      "learning_rate": 1.914698335403577e-05,
      "loss": 0.6829,
      "step": 4597
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7184054851531982,
      "learning_rate": 1.9146532047811698e-05,
      "loss": 0.732,
      "step": 4598
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.811234474182129,
      "learning_rate": 1.9146080627554383e-05,
      "loss": 0.6664,
      "step": 4599
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5068774223327637,
      "learning_rate": 1.9145629093269445e-05,
      "loss": 0.7603,
      "step": 4600
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6759271621704102,
      "learning_rate": 1.914517744496252e-05,
      "loss": 0.7145,
      "step": 4601
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7198593616485596,
      "learning_rate": 1.9144725682639236e-05,
      "loss": 0.7841,
      "step": 4602
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.555667519569397,
      "learning_rate": 1.914427380630522e-05,
      "loss": 0.8509,
      "step": 4603
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6674208641052246,
      "learning_rate": 1.9143821815966112e-05,
      "loss": 0.6834,
      "step": 4604
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7862920761108398,
      "learning_rate": 1.9143369711627547e-05,
      "loss": 0.6909,
      "step": 4605
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6959766149520874,
      "learning_rate": 1.914291749329516e-05,
      "loss": 0.7439,
      "step": 4606
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.954657793045044,
      "learning_rate": 1.9142465160974586e-05,
      "loss": 0.7318,
      "step": 4607
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.386237382888794,
      "learning_rate": 1.914201271467147e-05,
      "loss": 0.7309,
      "step": 4608
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4968199729919434,
      "learning_rate": 1.914156015439145e-05,
      "loss": 0.8474,
      "step": 4609
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.645053505897522,
      "learning_rate": 1.9141107480140167e-05,
      "loss": 0.8209,
      "step": 4610
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6284639835357666,
      "learning_rate": 1.9140654691923266e-05,
      "loss": 0.837,
      "step": 4611
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.570553183555603,
      "learning_rate": 1.914020178974639e-05,
      "loss": 0.8946,
      "step": 4612
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4431830644607544,
      "learning_rate": 1.913974877361519e-05,
      "loss": 0.736,
      "step": 4613
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6655322313308716,
      "learning_rate": 1.9139295643535316e-05,
      "loss": 0.5916,
      "step": 4614
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.560081958770752,
      "learning_rate": 1.913884239951241e-05,
      "loss": 0.8336,
      "step": 4615
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9042927026748657,
      "learning_rate": 1.9138389041552122e-05,
      "loss": 0.7529,
      "step": 4616
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6441575288772583,
      "learning_rate": 1.9137935569660107e-05,
      "loss": 0.8314,
      "step": 4617
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7634540796279907,
      "learning_rate": 1.9137481983842025e-05,
      "loss": 0.7825,
      "step": 4618
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4731552600860596,
      "learning_rate": 1.9137028284103524e-05,
      "loss": 0.8083,
      "step": 4619
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5314778089523315,
      "learning_rate": 1.9136574470450262e-05,
      "loss": 0.7449,
      "step": 4620
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8504583835601807,
      "learning_rate": 1.9136120542887898e-05,
      "loss": 0.8679,
      "step": 4621
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4813727140426636,
      "learning_rate": 1.9135666501422086e-05,
      "loss": 0.7156,
      "step": 4622
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2826875448226929,
      "learning_rate": 1.9135212346058494e-05,
      "loss": 0.7688,
      "step": 4623
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5620321035385132,
      "learning_rate": 1.9134758076802778e-05,
      "loss": 0.7967,
      "step": 4624
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6920822858810425,
      "learning_rate": 1.9134303693660606e-05,
      "loss": 0.631,
      "step": 4625
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4748871326446533,
      "learning_rate": 1.9133849196637642e-05,
      "loss": 0.8587,
      "step": 4626
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.717797875404358,
      "learning_rate": 1.9133394585739552e-05,
      "loss": 0.8215,
      "step": 4627
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5074455738067627,
      "learning_rate": 1.9132939860972e-05,
      "loss": 0.6776,
      "step": 4628
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4912091493606567,
      "learning_rate": 1.9132485022340663e-05,
      "loss": 0.6478,
      "step": 4629
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5227669477462769,
      "learning_rate": 1.913203006985121e-05,
      "loss": 0.8568,
      "step": 4630
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.747432827949524,
      "learning_rate": 1.9131575003509305e-05,
      "loss": 0.7297,
      "step": 4631
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4562660455703735,
      "learning_rate": 1.913111982332063e-05,
      "loss": 0.6986,
      "step": 4632
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5173383951187134,
      "learning_rate": 1.9130664529290856e-05,
      "loss": 0.6756,
      "step": 4633
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4138458967208862,
      "learning_rate": 1.913020912142566e-05,
      "loss": 0.5507,
      "step": 4634
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7690075635910034,
      "learning_rate": 1.9129753599730717e-05,
      "loss": 0.7483,
      "step": 4635
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2693932056427002,
      "learning_rate": 1.9129297964211714e-05,
      "loss": 0.8344,
      "step": 4636
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4411733150482178,
      "learning_rate": 1.9128842214874326e-05,
      "loss": 0.6171,
      "step": 4637
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7924857139587402,
      "learning_rate": 1.9128386351724234e-05,
      "loss": 0.8945,
      "step": 4638
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5741864442825317,
      "learning_rate": 1.9127930374767124e-05,
      "loss": 0.8105,
      "step": 4639
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.790489912033081,
      "learning_rate": 1.912747428400868e-05,
      "loss": 0.6563,
      "step": 4640
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9783416986465454,
      "learning_rate": 1.9127018079454585e-05,
      "loss": 0.6493,
      "step": 4641
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6026278734207153,
      "learning_rate": 1.9126561761110532e-05,
      "loss": 0.7975,
      "step": 4642
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7980291843414307,
      "learning_rate": 1.9126105328982213e-05,
      "loss": 0.7959,
      "step": 4643
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7134872674942017,
      "learning_rate": 1.9125648783075305e-05,
      "loss": 0.7007,
      "step": 4644
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4502722024917603,
      "learning_rate": 1.9125192123395515e-05,
      "loss": 0.8409,
      "step": 4645
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3963147401809692,
      "learning_rate": 1.9124735349948527e-05,
      "loss": 0.8207,
      "step": 4646
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8768577575683594,
      "learning_rate": 1.912427846274004e-05,
      "loss": 0.7943,
      "step": 4647
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5134474039077759,
      "learning_rate": 1.9123821461775748e-05,
      "loss": 0.9141,
      "step": 4648
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.588104248046875,
      "learning_rate": 1.9123364347061347e-05,
      "loss": 0.7317,
      "step": 4649
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8175450563430786,
      "learning_rate": 1.912290711860254e-05,
      "loss": 0.9133,
      "step": 4650
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6121699810028076,
      "learning_rate": 1.912244977640503e-05,
      "loss": 0.5901,
      "step": 4651
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.616822600364685,
      "learning_rate": 1.9121992320474508e-05,
      "loss": 0.8763,
      "step": 4652
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.714634895324707,
      "learning_rate": 1.912153475081669e-05,
      "loss": 0.6282,
      "step": 4653
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7417128086090088,
      "learning_rate": 1.912107706743727e-05,
      "loss": 0.7318,
      "step": 4654
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7075650691986084,
      "learning_rate": 1.912061927034196e-05,
      "loss": 0.8832,
      "step": 4655
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5133205652236938,
      "learning_rate": 1.9120161359536467e-05,
      "loss": 0.7986,
      "step": 4656
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4746540784835815,
      "learning_rate": 1.91197033350265e-05,
      "loss": 0.6388,
      "step": 4657
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6041113138198853,
      "learning_rate": 1.9119245196817763e-05,
      "loss": 0.6942,
      "step": 4658
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6817127466201782,
      "learning_rate": 1.911878694491598e-05,
      "loss": 0.7417,
      "step": 4659
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7032394409179688,
      "learning_rate": 1.9118328579326856e-05,
      "loss": 0.8012,
      "step": 4660
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4051374197006226,
      "learning_rate": 1.9117870100056106e-05,
      "loss": 0.8276,
      "step": 4661
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.720507264137268,
      "learning_rate": 1.9117411507109448e-05,
      "loss": 0.8505,
      "step": 4662
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6812529563903809,
      "learning_rate": 1.91169528004926e-05,
      "loss": 0.6993,
      "step": 4663
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4375382661819458,
      "learning_rate": 1.9116493980211273e-05,
      "loss": 0.6637,
      "step": 4664
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7999924421310425,
      "learning_rate": 1.91160350462712e-05,
      "loss": 0.6953,
      "step": 4665
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3946588039398193,
      "learning_rate": 1.9115575998678095e-05,
      "loss": 0.6682,
      "step": 4666
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.418508768081665,
      "learning_rate": 1.9115116837437683e-05,
      "loss": 0.7774,
      "step": 4667
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.55277681350708,
      "learning_rate": 1.9114657562555686e-05,
      "loss": 0.8022,
      "step": 4668
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8316535949707031,
      "learning_rate": 1.9114198174037836e-05,
      "loss": 0.7812,
      "step": 4669
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3288464546203613,
      "learning_rate": 1.9113738671889855e-05,
      "loss": 0.6348,
      "step": 4670
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4783653020858765,
      "learning_rate": 1.9113279056117472e-05,
      "loss": 0.742,
      "step": 4671
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.832775354385376,
      "learning_rate": 1.9112819326726418e-05,
      "loss": 0.7956,
      "step": 4672
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7214183807373047,
      "learning_rate": 1.911235948372243e-05,
      "loss": 0.7231,
      "step": 4673
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0477781295776367,
      "learning_rate": 1.911189952711123e-05,
      "loss": 0.7857,
      "step": 4674
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6725972890853882,
      "learning_rate": 1.911143945689856e-05,
      "loss": 0.653,
      "step": 4675
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.812554121017456,
      "learning_rate": 1.9110979273090154e-05,
      "loss": 0.8566,
      "step": 4676
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.521127700805664,
      "learning_rate": 1.911051897569175e-05,
      "loss": 0.636,
      "step": 4677
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3181582689285278,
      "learning_rate": 1.9110058564709087e-05,
      "loss": 0.8571,
      "step": 4678
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.0259084701538086,
      "learning_rate": 1.9109598040147907e-05,
      "loss": 0.7411,
      "step": 4679
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.2238683700561523,
      "learning_rate": 1.910913740201395e-05,
      "loss": 0.8926,
      "step": 4680
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.693673014640808,
      "learning_rate": 1.910867665031295e-05,
      "loss": 0.9309,
      "step": 4681
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.1392927169799805,
      "learning_rate": 1.9108215785050666e-05,
      "loss": 0.9829,
      "step": 4682
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4340581893920898,
      "learning_rate": 1.9107754806232835e-05,
      "loss": 0.7833,
      "step": 4683
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6923012733459473,
      "learning_rate": 1.9107293713865206e-05,
      "loss": 0.8058,
      "step": 4684
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.740350365638733,
      "learning_rate": 1.910683250795353e-05,
      "loss": 0.8586,
      "step": 4685
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4989404678344727,
      "learning_rate": 1.9106371188503555e-05,
      "loss": 0.7658,
      "step": 4686
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4434380531311035,
      "learning_rate": 1.9105909755521033e-05,
      "loss": 0.7353,
      "step": 4687
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.542209506034851,
      "learning_rate": 1.9105448209011715e-05,
      "loss": 0.84,
      "step": 4688
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.671143889427185,
      "learning_rate": 1.9104986548981355e-05,
      "loss": 0.7494,
      "step": 4689
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.592411756515503,
      "learning_rate": 1.910452477543571e-05,
      "loss": 0.8119,
      "step": 4690
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.58713960647583,
      "learning_rate": 1.910406288838054e-05,
      "loss": 0.717,
      "step": 4691
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5869767665863037,
      "learning_rate": 1.91036008878216e-05,
      "loss": 0.7615,
      "step": 4692
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.829210638999939,
      "learning_rate": 1.910313877376465e-05,
      "loss": 0.6974,
      "step": 4693
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6816647052764893,
      "learning_rate": 1.9102676546215454e-05,
      "loss": 0.7069,
      "step": 4694
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2758456468582153,
      "learning_rate": 1.910221420517977e-05,
      "loss": 0.8252,
      "step": 4695
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6742750406265259,
      "learning_rate": 1.9101751750663366e-05,
      "loss": 0.7683,
      "step": 4696
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5256261825561523,
      "learning_rate": 1.910128918267201e-05,
      "loss": 0.8158,
      "step": 4697
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9209614992141724,
      "learning_rate": 1.910082650121146e-05,
      "loss": 0.6768,
      "step": 4698
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9484059810638428,
      "learning_rate": 1.9100363706287496e-05,
      "loss": 0.9541,
      "step": 4699
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7708332538604736,
      "learning_rate": 1.9099900797905876e-05,
      "loss": 0.896,
      "step": 4700
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.502781629562378,
      "learning_rate": 1.9099437776072383e-05,
      "loss": 0.8186,
      "step": 4701
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.680725336074829,
      "learning_rate": 1.909897464079278e-05,
      "loss": 0.6673,
      "step": 4702
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3881518840789795,
      "learning_rate": 1.9098511392072846e-05,
      "loss": 0.625,
      "step": 4703
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6350221633911133,
      "learning_rate": 1.9098048029918356e-05,
      "loss": 0.902,
      "step": 4704
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4166674613952637,
      "learning_rate": 1.909758455433509e-05,
      "loss": 0.7576,
      "step": 4705
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8226925134658813,
      "learning_rate": 1.909712096532882e-05,
      "loss": 0.7855,
      "step": 4706
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3689007759094238,
      "learning_rate": 1.9096657262905324e-05,
      "loss": 0.7814,
      "step": 4707
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2940049171447754,
      "learning_rate": 1.9096193447070393e-05,
      "loss": 0.6951,
      "step": 4708
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.912441372871399,
      "learning_rate": 1.90957295178298e-05,
      "loss": 0.8002,
      "step": 4709
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7118499279022217,
      "learning_rate": 1.9095265475189337e-05,
      "loss": 0.8852,
      "step": 4710
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4337188005447388,
      "learning_rate": 1.9094801319154782e-05,
      "loss": 0.7319,
      "step": 4711
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6493680477142334,
      "learning_rate": 1.909433704973193e-05,
      "loss": 0.5492,
      "step": 4712
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8475712537765503,
      "learning_rate": 1.909387266692656e-05,
      "loss": 0.8736,
      "step": 4713
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7392486333847046,
      "learning_rate": 1.909340817074447e-05,
      "loss": 0.7419,
      "step": 4714
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.269752860069275,
      "learning_rate": 1.9092943561191443e-05,
      "loss": 0.725,
      "step": 4715
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5002886056900024,
      "learning_rate": 1.9092478838273275e-05,
      "loss": 0.6873,
      "step": 4716
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8061617612838745,
      "learning_rate": 1.9092014001995767e-05,
      "loss": 0.7985,
      "step": 4717
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.48714017868042,
      "learning_rate": 1.9091549052364704e-05,
      "loss": 0.7932,
      "step": 4718
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6662858724594116,
      "learning_rate": 1.9091083989385885e-05,
      "loss": 0.8508,
      "step": 4719
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3531718254089355,
      "learning_rate": 1.9090618813065112e-05,
      "loss": 0.6528,
      "step": 4720
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5333038568496704,
      "learning_rate": 1.909015352340818e-05,
      "loss": 0.6876,
      "step": 4721
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.477439284324646,
      "learning_rate": 1.9089688120420894e-05,
      "loss": 0.7883,
      "step": 4722
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7073262929916382,
      "learning_rate": 1.9089222604109055e-05,
      "loss": 0.8923,
      "step": 4723
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.507111668586731,
      "learning_rate": 1.908875697447847e-05,
      "loss": 0.7153,
      "step": 4724
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5378388166427612,
      "learning_rate": 1.9088291231534935e-05,
      "loss": 0.8411,
      "step": 4725
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.030305862426758,
      "learning_rate": 1.9087825375284265e-05,
      "loss": 0.7118,
      "step": 4726
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.773978352546692,
      "learning_rate": 1.9087359405732262e-05,
      "loss": 0.7696,
      "step": 4727
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3695528507232666,
      "learning_rate": 1.9086893322884737e-05,
      "loss": 0.6592,
      "step": 4728
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0645530223846436,
      "learning_rate": 1.9086427126747508e-05,
      "loss": 1.0041,
      "step": 4729
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5030878782272339,
      "learning_rate": 1.9085960817326377e-05,
      "loss": 0.7913,
      "step": 4730
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.679494023323059,
      "learning_rate": 1.9085494394627164e-05,
      "loss": 0.8393,
      "step": 4731
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5295768976211548,
      "learning_rate": 1.908502785865568e-05,
      "loss": 0.6798,
      "step": 4732
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5092748403549194,
      "learning_rate": 1.9084561209417746e-05,
      "loss": 0.7378,
      "step": 4733
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8615381717681885,
      "learning_rate": 1.908409444691918e-05,
      "loss": 0.7451,
      "step": 4734
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.477599859237671,
      "learning_rate": 1.9083627571165797e-05,
      "loss": 0.8117,
      "step": 4735
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2114479541778564,
      "learning_rate": 1.9083160582163418e-05,
      "loss": 0.7605,
      "step": 4736
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.0913825035095215,
      "learning_rate": 1.908269347991787e-05,
      "loss": 0.7919,
      "step": 4737
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5261822938919067,
      "learning_rate": 1.908222626443497e-05,
      "loss": 0.8718,
      "step": 4738
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3717436790466309,
      "learning_rate": 1.9081758935720545e-05,
      "loss": 0.8571,
      "step": 4739
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7095478773117065,
      "learning_rate": 1.9081291493780427e-05,
      "loss": 0.7559,
      "step": 4740
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5109611749649048,
      "learning_rate": 1.908082393862044e-05,
      "loss": 0.8129,
      "step": 4741
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.395567774772644,
      "learning_rate": 1.9080356270246408e-05,
      "loss": 0.7876,
      "step": 4742
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3406014442443848,
      "learning_rate": 1.907988848866417e-05,
      "loss": 0.713,
      "step": 4743
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5702306032180786,
      "learning_rate": 1.907942059387955e-05,
      "loss": 0.8596,
      "step": 4744
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4759652614593506,
      "learning_rate": 1.907895258589839e-05,
      "loss": 0.7081,
      "step": 4745
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7983295917510986,
      "learning_rate": 1.907848446472652e-05,
      "loss": 0.7968,
      "step": 4746
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8047395944595337,
      "learning_rate": 1.9078016230369773e-05,
      "loss": 0.6416,
      "step": 4747
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.003737449645996,
      "learning_rate": 1.9077547882833994e-05,
      "loss": 0.772,
      "step": 4748
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5334460735321045,
      "learning_rate": 1.9077079422125015e-05,
      "loss": 0.6775,
      "step": 4749
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.612687587738037,
      "learning_rate": 1.9076610848248684e-05,
      "loss": 0.7277,
      "step": 4750
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6018608808517456,
      "learning_rate": 1.9076142161210834e-05,
      "loss": 0.6211,
      "step": 4751
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7395097017288208,
      "learning_rate": 1.9075673361017318e-05,
      "loss": 0.7159,
      "step": 4752
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4138623476028442,
      "learning_rate": 1.9075204447673973e-05,
      "loss": 0.753,
      "step": 4753
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.766363501548767,
      "learning_rate": 1.9074735421186646e-05,
      "loss": 0.723,
      "step": 4754
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4306938648223877,
      "learning_rate": 1.907426628156119e-05,
      "loss": 0.7834,
      "step": 4755
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6794949769973755,
      "learning_rate": 1.907379702880345e-05,
      "loss": 0.6872,
      "step": 4756
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6905838251113892,
      "learning_rate": 1.9073327662919277e-05,
      "loss": 0.772,
      "step": 4757
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7631444931030273,
      "learning_rate": 1.907285818391452e-05,
      "loss": 0.8316,
      "step": 4758
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5775821208953857,
      "learning_rate": 1.9072388591795036e-05,
      "loss": 0.7387,
      "step": 4759
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6927154064178467,
      "learning_rate": 1.9071918886566677e-05,
      "loss": 0.8478,
      "step": 4760
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.536329746246338,
      "learning_rate": 1.9071449068235303e-05,
      "loss": 0.8292,
      "step": 4761
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5569454431533813,
      "learning_rate": 1.9070979136806768e-05,
      "loss": 0.7302,
      "step": 4762
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3543459177017212,
      "learning_rate": 1.9070509092286927e-05,
      "loss": 0.7381,
      "step": 4763
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5439883470535278,
      "learning_rate": 1.907003893468165e-05,
      "loss": 0.688,
      "step": 4764
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5758475065231323,
      "learning_rate": 1.9069568663996795e-05,
      "loss": 0.8104,
      "step": 4765
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4594751596450806,
      "learning_rate": 1.906909828023822e-05,
      "loss": 0.7176,
      "step": 4766
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8993871212005615,
      "learning_rate": 1.906862778341179e-05,
      "loss": 0.8274,
      "step": 4767
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7659058570861816,
      "learning_rate": 1.9068157173523377e-05,
      "loss": 0.8165,
      "step": 4768
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.582292079925537,
      "learning_rate": 1.9067686450578843e-05,
      "loss": 0.766,
      "step": 4769
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9143304824829102,
      "learning_rate": 1.906721561458406e-05,
      "loss": 0.8054,
      "step": 4770
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4782806634902954,
      "learning_rate": 1.9066744665544895e-05,
      "loss": 0.5948,
      "step": 4771
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.710381031036377,
      "learning_rate": 1.9066273603467224e-05,
      "loss": 0.8102,
      "step": 4772
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5856828689575195,
      "learning_rate": 1.9065802428356916e-05,
      "loss": 0.7306,
      "step": 4773
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.496527075767517,
      "learning_rate": 1.9065331140219843e-05,
      "loss": 0.7202,
      "step": 4774
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6627353429794312,
      "learning_rate": 1.906485973906189e-05,
      "loss": 0.844,
      "step": 4775
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7972122430801392,
      "learning_rate": 1.9064388224888925e-05,
      "loss": 0.8537,
      "step": 4776
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6635383367538452,
      "learning_rate": 1.906391659770683e-05,
      "loss": 0.642,
      "step": 4777
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6693075895309448,
      "learning_rate": 1.9063444857521487e-05,
      "loss": 0.7578,
      "step": 4778
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6686880588531494,
      "learning_rate": 1.9062973004338774e-05,
      "loss": 0.7941,
      "step": 4779
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.834830641746521,
      "learning_rate": 1.9062501038164572e-05,
      "loss": 0.7767,
      "step": 4780
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7265536785125732,
      "learning_rate": 1.906202895900477e-05,
      "loss": 0.7568,
      "step": 4781
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.8403817415237427,
      "learning_rate": 1.9061556766865253e-05,
      "loss": 0.7607,
      "step": 4782
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5497575998306274,
      "learning_rate": 1.9061084461751906e-05,
      "loss": 0.9146,
      "step": 4783
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.506365180015564,
      "learning_rate": 1.9060612043670618e-05,
      "loss": 0.6973,
      "step": 4784
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9358443021774292,
      "learning_rate": 1.9060139512627275e-05,
      "loss": 0.8575,
      "step": 4785
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4691827297210693,
      "learning_rate": 1.9059666868627778e-05,
      "loss": 0.7221,
      "step": 4786
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.037694215774536,
      "learning_rate": 1.9059194111678013e-05,
      "loss": 0.7756,
      "step": 4787
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4493597745895386,
      "learning_rate": 1.905872124178387e-05,
      "loss": 0.6439,
      "step": 4788
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3445496559143066,
      "learning_rate": 1.905824825895125e-05,
      "loss": 0.6683,
      "step": 4789
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7588753700256348,
      "learning_rate": 1.9057775163186055e-05,
      "loss": 0.6964,
      "step": 4790
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5898478031158447,
      "learning_rate": 1.9057301954494175e-05,
      "loss": 0.6312,
      "step": 4791
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.594058036804199,
      "learning_rate": 1.905682863288151e-05,
      "loss": 0.8866,
      "step": 4792
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7670918703079224,
      "learning_rate": 1.905635519835396e-05,
      "loss": 0.9286,
      "step": 4793
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5693544149398804,
      "learning_rate": 1.9055881650917437e-05,
      "loss": 0.7324,
      "step": 4794
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.479641079902649,
      "learning_rate": 1.9055407990577835e-05,
      "loss": 0.7485,
      "step": 4795
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.707995891571045,
      "learning_rate": 1.9054934217341064e-05,
      "loss": 0.903,
      "step": 4796
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.744505524635315,
      "learning_rate": 1.9054460331213026e-05,
      "loss": 0.7149,
      "step": 4797
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4847899675369263,
      "learning_rate": 1.9053986332199636e-05,
      "loss": 0.7952,
      "step": 4798
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.606081485748291,
      "learning_rate": 1.90535122203068e-05,
      "loss": 0.7999,
      "step": 4799
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9129416942596436,
      "learning_rate": 1.9053037995540432e-05,
      "loss": 0.8232,
      "step": 4800
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4617125988006592,
      "learning_rate": 1.9052563657906435e-05,
      "loss": 0.7157,
      "step": 4801
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3557405471801758,
      "learning_rate": 1.9052089207410732e-05,
      "loss": 0.8731,
      "step": 4802
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6882514953613281,
      "learning_rate": 1.905161464405924e-05,
      "loss": 0.8261,
      "step": 4803
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2795462608337402,
      "learning_rate": 1.905113996785786e-05,
      "loss": 0.622,
      "step": 4804
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6845693588256836,
      "learning_rate": 1.905066517881253e-05,
      "loss": 0.8386,
      "step": 4805
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.80239999294281,
      "learning_rate": 1.9050190276929157e-05,
      "loss": 0.7765,
      "step": 4806
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5893300771713257,
      "learning_rate": 1.9049715262213664e-05,
      "loss": 0.8796,
      "step": 4807
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.537598729133606,
      "learning_rate": 1.9049240134671977e-05,
      "loss": 0.7349,
      "step": 4808
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8025867938995361,
      "learning_rate": 1.9048764894310014e-05,
      "loss": 0.8586,
      "step": 4809
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3928804397583008,
      "learning_rate": 1.9048289541133705e-05,
      "loss": 0.7219,
      "step": 4810
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7252192497253418,
      "learning_rate": 1.9047814075148972e-05,
      "loss": 0.7802,
      "step": 4811
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.538987398147583,
      "learning_rate": 1.9047338496361748e-05,
      "loss": 0.8146,
      "step": 4812
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.620144248008728,
      "learning_rate": 1.9046862804777955e-05,
      "loss": 0.9695,
      "step": 4813
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7615289688110352,
      "learning_rate": 1.904638700040353e-05,
      "loss": 0.7667,
      "step": 4814
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7669421434402466,
      "learning_rate": 1.9045911083244403e-05,
      "loss": 0.9726,
      "step": 4815
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6640499830245972,
      "learning_rate": 1.9045435053306505e-05,
      "loss": 0.8151,
      "step": 4816
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6696012020111084,
      "learning_rate": 1.9044958910595777e-05,
      "loss": 0.5927,
      "step": 4817
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.631760597229004,
      "learning_rate": 1.904448265511815e-05,
      "loss": 0.7111,
      "step": 4818
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6939616203308105,
      "learning_rate": 1.904400628687956e-05,
      "loss": 0.7762,
      "step": 4819
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7428208589553833,
      "learning_rate": 1.904352980588595e-05,
      "loss": 0.8651,
      "step": 4820
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.660784363746643,
      "learning_rate": 1.9043053212143263e-05,
      "loss": 0.6977,
      "step": 4821
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6112539768218994,
      "learning_rate": 1.9042576505657437e-05,
      "loss": 0.7697,
      "step": 4822
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.551998496055603,
      "learning_rate": 1.9042099686434415e-05,
      "loss": 0.8681,
      "step": 4823
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.580263614654541,
      "learning_rate": 1.9041622754480138e-05,
      "loss": 0.7075,
      "step": 4824
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8126636743545532,
      "learning_rate": 1.904114570980056e-05,
      "loss": 0.6153,
      "step": 4825
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5174670219421387,
      "learning_rate": 1.9040668552401628e-05,
      "loss": 0.7267,
      "step": 4826
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6898612976074219,
      "learning_rate": 1.9040191282289284e-05,
      "loss": 0.8668,
      "step": 4827
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3164809942245483,
      "learning_rate": 1.903971389946948e-05,
      "loss": 0.6099,
      "step": 4828
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6106680631637573,
      "learning_rate": 1.9039236403948177e-05,
      "loss": 0.7961,
      "step": 4829
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5416444540023804,
      "learning_rate": 1.9038758795731316e-05,
      "loss": 0.7701,
      "step": 4830
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7611171007156372,
      "learning_rate": 1.9038281074824856e-05,
      "loss": 0.7323,
      "step": 4831
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5590020418167114,
      "learning_rate": 1.9037803241234757e-05,
      "loss": 0.7329,
      "step": 4832
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4820674657821655,
      "learning_rate": 1.903732529496697e-05,
      "loss": 0.7161,
      "step": 4833
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1928629875183105,
      "learning_rate": 1.9036847236027454e-05,
      "loss": 0.9058,
      "step": 4834
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5442720651626587,
      "learning_rate": 1.9036369064422176e-05,
      "loss": 0.7503,
      "step": 4835
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9365788698196411,
      "learning_rate": 1.9035890780157095e-05,
      "loss": 0.7327,
      "step": 4836
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.627275824546814,
      "learning_rate": 1.9035412383238167e-05,
      "loss": 0.7266,
      "step": 4837
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7047338485717773,
      "learning_rate": 1.9034933873671364e-05,
      "loss": 0.7577,
      "step": 4838
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6253694295883179,
      "learning_rate": 1.9034455251462648e-05,
      "loss": 0.8529,
      "step": 4839
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5773855447769165,
      "learning_rate": 1.903397651661799e-05,
      "loss": 0.6682,
      "step": 4840
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4948773384094238,
      "learning_rate": 1.9033497669143352e-05,
      "loss": 0.6833,
      "step": 4841
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6051133871078491,
      "learning_rate": 1.9033018709044712e-05,
      "loss": 0.6693,
      "step": 4842
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0916237831115723,
      "learning_rate": 1.9032539636328034e-05,
      "loss": 0.7842,
      "step": 4843
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5417873859405518,
      "learning_rate": 1.9032060450999295e-05,
      "loss": 0.8221,
      "step": 4844
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.461725115776062,
      "learning_rate": 1.903158115306447e-05,
      "loss": 0.6762,
      "step": 4845
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.690932035446167,
      "learning_rate": 1.9031101742529533e-05,
      "loss": 0.9021,
      "step": 4846
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9168896675109863,
      "learning_rate": 1.903062221940046e-05,
      "loss": 0.8322,
      "step": 4847
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.599871277809143,
      "learning_rate": 1.903014258368323e-05,
      "loss": 0.8683,
      "step": 4848
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9207916259765625,
      "learning_rate": 1.9029662835383823e-05,
      "loss": 0.7919,
      "step": 4849
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7942900657653809,
      "learning_rate": 1.902918297450822e-05,
      "loss": 0.732,
      "step": 4850
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5981699228286743,
      "learning_rate": 1.9028703001062404e-05,
      "loss": 0.8247,
      "step": 4851
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8295316696166992,
      "learning_rate": 1.9028222915052358e-05,
      "loss": 0.7888,
      "step": 4852
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5646758079528809,
      "learning_rate": 1.902774271648407e-05,
      "loss": 0.6141,
      "step": 4853
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4265923500061035,
      "learning_rate": 1.9027262405363523e-05,
      "loss": 0.7812,
      "step": 4854
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.659633755683899,
      "learning_rate": 1.902678198169671e-05,
      "loss": 0.7448,
      "step": 4855
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4945135116577148,
      "learning_rate": 1.9026301445489617e-05,
      "loss": 0.7359,
      "step": 4856
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1691792011260986,
      "learning_rate": 1.9025820796748235e-05,
      "loss": 0.9773,
      "step": 4857
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4103035926818848,
      "learning_rate": 1.902534003547856e-05,
      "loss": 0.6494,
      "step": 4858
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3956822156906128,
      "learning_rate": 1.902485916168658e-05,
      "loss": 0.7339,
      "step": 4859
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5469025373458862,
      "learning_rate": 1.9024378175378294e-05,
      "loss": 0.7149,
      "step": 4860
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9791146516799927,
      "learning_rate": 1.9023897076559697e-05,
      "loss": 0.7437,
      "step": 4861
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7640997171401978,
      "learning_rate": 1.9023415865236792e-05,
      "loss": 0.863,
      "step": 4862
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6659715175628662,
      "learning_rate": 1.9022934541415576e-05,
      "loss": 0.713,
      "step": 4863
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9284937381744385,
      "learning_rate": 1.9022453105102045e-05,
      "loss": 0.8788,
      "step": 4864
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4467345476150513,
      "learning_rate": 1.9021971556302203e-05,
      "loss": 0.6205,
      "step": 4865
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5631120204925537,
      "learning_rate": 1.9021489895022057e-05,
      "loss": 0.7081,
      "step": 4866
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3724722862243652,
      "learning_rate": 1.902100812126761e-05,
      "loss": 0.7,
      "step": 4867
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8013091087341309,
      "learning_rate": 1.9020526235044873e-05,
      "loss": 0.6833,
      "step": 4868
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.374557375907898,
      "learning_rate": 1.9020044236359846e-05,
      "loss": 0.6406,
      "step": 4869
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8853554725646973,
      "learning_rate": 1.9019562125218537e-05,
      "loss": 0.8268,
      "step": 4870
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7753535509109497,
      "learning_rate": 1.901907990162697e-05,
      "loss": 0.8551,
      "step": 4871
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3459101915359497,
      "learning_rate": 1.9018597565591146e-05,
      "loss": 0.7646,
      "step": 4872
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5332878828048706,
      "learning_rate": 1.901811511711708e-05,
      "loss": 0.745,
      "step": 4873
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6644158363342285,
      "learning_rate": 1.9017632556210788e-05,
      "loss": 0.7543,
      "step": 4874
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.714665412902832,
      "learning_rate": 1.9017149882878286e-05,
      "loss": 0.8734,
      "step": 4875
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.613524079322815,
      "learning_rate": 1.9016667097125595e-05,
      "loss": 0.8282,
      "step": 4876
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4223538637161255,
      "learning_rate": 1.9016184198958728e-05,
      "loss": 0.6755,
      "step": 4877
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6864831447601318,
      "learning_rate": 1.901570118838371e-05,
      "loss": 0.6614,
      "step": 4878
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9329619407653809,
      "learning_rate": 1.901521806540656e-05,
      "loss": 0.8853,
      "step": 4879
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.1876280307769775,
      "learning_rate": 1.90147348300333e-05,
      "loss": 0.9541,
      "step": 4880
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.55497407913208,
      "learning_rate": 1.9014251482269963e-05,
      "loss": 0.8964,
      "step": 4881
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5774544477462769,
      "learning_rate": 1.9013768022122566e-05,
      "loss": 0.6817,
      "step": 4882
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5131853818893433,
      "learning_rate": 1.901328444959714e-05,
      "loss": 0.7765,
      "step": 4883
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9780460596084595,
      "learning_rate": 1.9012800764699717e-05,
      "loss": 0.7838,
      "step": 4884
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2965086698532104,
      "learning_rate": 1.901231696743632e-05,
      "loss": 0.656,
      "step": 4885
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4588861465454102,
      "learning_rate": 1.9011833057812987e-05,
      "loss": 0.7125,
      "step": 4886
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5368648767471313,
      "learning_rate": 1.9011349035835748e-05,
      "loss": 0.8148,
      "step": 4887
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5860306024551392,
      "learning_rate": 1.901086490151064e-05,
      "loss": 0.6194,
      "step": 4888
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6574993133544922,
      "learning_rate": 1.9010380654843694e-05,
      "loss": 0.8046,
      "step": 4889
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.486970067024231,
      "learning_rate": 1.9009896295840953e-05,
      "loss": 0.7299,
      "step": 4890
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7203102111816406,
      "learning_rate": 1.9009411824508452e-05,
      "loss": 0.768,
      "step": 4891
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.705621600151062,
      "learning_rate": 1.9008927240852234e-05,
      "loss": 0.6916,
      "step": 4892
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.658665657043457,
      "learning_rate": 1.9008442544878335e-05,
      "loss": 0.6705,
      "step": 4893
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8117402791976929,
      "learning_rate": 1.9007957736592804e-05,
      "loss": 0.7702,
      "step": 4894
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6684452295303345,
      "learning_rate": 1.9007472816001682e-05,
      "loss": 0.7124,
      "step": 4895
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6415551900863647,
      "learning_rate": 1.9006987783111017e-05,
      "loss": 0.8227,
      "step": 4896
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6815470457077026,
      "learning_rate": 1.9006502637926852e-05,
      "loss": 0.7498,
      "step": 4897
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5684674978256226,
      "learning_rate": 1.900601738045524e-05,
      "loss": 0.8091,
      "step": 4898
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9487297534942627,
      "learning_rate": 1.900553201070223e-05,
      "loss": 0.7981,
      "step": 4899
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4722082614898682,
      "learning_rate": 1.900504652867387e-05,
      "loss": 0.7136,
      "step": 4900
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7895517349243164,
      "learning_rate": 1.9004560934376218e-05,
      "loss": 0.6414,
      "step": 4901
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3706333637237549,
      "learning_rate": 1.9004075227815323e-05,
      "loss": 0.5902,
      "step": 4902
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.020885705947876,
      "learning_rate": 1.9003589408997243e-05,
      "loss": 0.7765,
      "step": 4903
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5744028091430664,
      "learning_rate": 1.9003103477928036e-05,
      "loss": 0.6943,
      "step": 4904
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.661507248878479,
      "learning_rate": 1.9002617434613757e-05,
      "loss": 0.923,
      "step": 4905
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.82991623878479,
      "learning_rate": 1.9002131279060467e-05,
      "loss": 0.8077,
      "step": 4906
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2916542291641235,
      "learning_rate": 1.9001645011274226e-05,
      "loss": 0.711,
      "step": 4907
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8361557722091675,
      "learning_rate": 1.9001158631261105e-05,
      "loss": 0.6777,
      "step": 4908
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.591536045074463,
      "learning_rate": 1.9000672139027152e-05,
      "loss": 0.8738,
      "step": 4909
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.771562099456787,
      "learning_rate": 1.9000185534578448e-05,
      "loss": 0.7687,
      "step": 4910
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5640461444854736,
      "learning_rate": 1.899969881792105e-05,
      "loss": 0.6794,
      "step": 4911
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5879207849502563,
      "learning_rate": 1.899921198906103e-05,
      "loss": 0.8081,
      "step": 4912
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8661599159240723,
      "learning_rate": 1.8998725048004454e-05,
      "loss": 0.7612,
      "step": 4913
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.934577226638794,
      "learning_rate": 1.8998237994757396e-05,
      "loss": 0.6988,
      "step": 4914
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6194727420806885,
      "learning_rate": 1.899775082932593e-05,
      "loss": 0.6424,
      "step": 4915
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.442052960395813,
      "learning_rate": 1.8997263551716124e-05,
      "loss": 0.8382,
      "step": 4916
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4067208766937256,
      "learning_rate": 1.899677616193406e-05,
      "loss": 0.6823,
      "step": 4917
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.592626929283142,
      "learning_rate": 1.8996288659985806e-05,
      "loss": 0.7,
      "step": 4918
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.662056565284729,
      "learning_rate": 1.8995801045877447e-05,
      "loss": 0.7904,
      "step": 4919
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8255362510681152,
      "learning_rate": 1.8995313319615062e-05,
      "loss": 0.7574,
      "step": 4920
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6489503383636475,
      "learning_rate": 1.8994825481204725e-05,
      "loss": 0.6565,
      "step": 4921
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5518946647644043,
      "learning_rate": 1.8994337530652528e-05,
      "loss": 0.8365,
      "step": 4922
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5678858757019043,
      "learning_rate": 1.8993849467964546e-05,
      "loss": 0.696,
      "step": 4923
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3711706399917603,
      "learning_rate": 1.8993361293146865e-05,
      "loss": 0.7543,
      "step": 4924
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5482820272445679,
      "learning_rate": 1.8992873006205575e-05,
      "loss": 0.806,
      "step": 4925
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7782797813415527,
      "learning_rate": 1.8992384607146763e-05,
      "loss": 0.7999,
      "step": 4926
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5859630107879639,
      "learning_rate": 1.8991896095976516e-05,
      "loss": 0.7723,
      "step": 4927
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.552042841911316,
      "learning_rate": 1.8991407472700927e-05,
      "loss": 0.7178,
      "step": 4928
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9638845920562744,
      "learning_rate": 1.8990918737326084e-05,
      "loss": 0.8298,
      "step": 4929
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3914471864700317,
      "learning_rate": 1.8990429889858086e-05,
      "loss": 0.8137,
      "step": 4930
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6291455030441284,
      "learning_rate": 1.898994093030302e-05,
      "loss": 0.6511,
      "step": 4931
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7141528129577637,
      "learning_rate": 1.8989451858666986e-05,
      "loss": 0.7419,
      "step": 4932
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9310115575790405,
      "learning_rate": 1.8988962674956086e-05,
      "loss": 0.7273,
      "step": 4933
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5342402458190918,
      "learning_rate": 1.8988473379176412e-05,
      "loss": 0.7566,
      "step": 4934
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6672744750976562,
      "learning_rate": 1.8987983971334065e-05,
      "loss": 0.7976,
      "step": 4935
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5296777486801147,
      "learning_rate": 1.898749445143515e-05,
      "loss": 0.7003,
      "step": 4936
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7039093971252441,
      "learning_rate": 1.898700481948577e-05,
      "loss": 0.8037,
      "step": 4937
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4780117273330688,
      "learning_rate": 1.8986515075492024e-05,
      "loss": 0.6254,
      "step": 4938
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.620112657546997,
      "learning_rate": 1.8986025219460025e-05,
      "loss": 0.6597,
      "step": 4939
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5184705257415771,
      "learning_rate": 1.8985535251395874e-05,
      "loss": 0.6723,
      "step": 4940
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7605984210968018,
      "learning_rate": 1.8985045171305683e-05,
      "loss": 0.7735,
      "step": 4941
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7069593667984009,
      "learning_rate": 1.898455497919556e-05,
      "loss": 0.653,
      "step": 4942
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3551397323608398,
      "learning_rate": 1.8984064675071623e-05,
      "loss": 0.6714,
      "step": 4943
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5385003089904785,
      "learning_rate": 1.898357425893998e-05,
      "loss": 0.7731,
      "step": 4944
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5152192115783691,
      "learning_rate": 1.8983083730806737e-05,
      "loss": 0.809,
      "step": 4945
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4410099983215332,
      "learning_rate": 1.8982593090678022e-05,
      "loss": 0.6919,
      "step": 4946
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7811506986618042,
      "learning_rate": 1.8982102338559948e-05,
      "loss": 0.5718,
      "step": 4947
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5121040344238281,
      "learning_rate": 1.8981611474458632e-05,
      "loss": 0.7544,
      "step": 4948
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7338236570358276,
      "learning_rate": 1.8981120498380195e-05,
      "loss": 0.7753,
      "step": 4949
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6631964445114136,
      "learning_rate": 1.898062941033076e-05,
      "loss": 0.6967,
      "step": 4950
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7522904872894287,
      "learning_rate": 1.8980138210316443e-05,
      "loss": 0.6221,
      "step": 4951
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5958033800125122,
      "learning_rate": 1.8979646898343377e-05,
      "loss": 0.814,
      "step": 4952
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5091655254364014,
      "learning_rate": 1.8979155474417678e-05,
      "loss": 0.6401,
      "step": 4953
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.925923466682434,
      "learning_rate": 1.897866393854548e-05,
      "loss": 0.9263,
      "step": 4954
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6016424894332886,
      "learning_rate": 1.897817229073291e-05,
      "loss": 0.7385,
      "step": 4955
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.068065643310547,
      "learning_rate": 1.8977680530986094e-05,
      "loss": 0.735,
      "step": 4956
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3808435201644897,
      "learning_rate": 1.8977188659311167e-05,
      "loss": 0.6605,
      "step": 4957
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5359766483306885,
      "learning_rate": 1.8976696675714255e-05,
      "loss": 0.7146,
      "step": 4958
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.460657000541687,
      "learning_rate": 1.89762045802015e-05,
      "loss": 0.7113,
      "step": 4959
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3228740692138672,
      "learning_rate": 1.8975712372779032e-05,
      "loss": 0.6258,
      "step": 4960
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6781922578811646,
      "learning_rate": 1.8975220053452992e-05,
      "loss": 0.6838,
      "step": 4961
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5028475522994995,
      "learning_rate": 1.8974727622229512e-05,
      "loss": 0.5268,
      "step": 4962
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.549438238143921,
      "learning_rate": 1.8974235079114735e-05,
      "loss": 0.7588,
      "step": 4963
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.511612892150879,
      "learning_rate": 1.89737424241148e-05,
      "loss": 0.7331,
      "step": 4964
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7348692417144775,
      "learning_rate": 1.897324965723585e-05,
      "loss": 0.6622,
      "step": 4965
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7463269233703613,
      "learning_rate": 1.8972756778484033e-05,
      "loss": 0.8201,
      "step": 4966
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.548833966255188,
      "learning_rate": 1.8972263787865483e-05,
      "loss": 0.8223,
      "step": 4967
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7292232513427734,
      "learning_rate": 1.8971770685386357e-05,
      "loss": 0.7991,
      "step": 4968
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6146982908248901,
      "learning_rate": 1.89712774710528e-05,
      "loss": 0.6015,
      "step": 4969
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.7406699657440186,
      "learning_rate": 1.8970784144870953e-05,
      "loss": 0.6644,
      "step": 4970
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9367693662643433,
      "learning_rate": 1.8970290706846978e-05,
      "loss": 0.8436,
      "step": 4971
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5979068279266357,
      "learning_rate": 1.8969797156987017e-05,
      "loss": 0.7855,
      "step": 4972
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5633352994918823,
      "learning_rate": 1.8969303495297233e-05,
      "loss": 0.7048,
      "step": 4973
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.795491337776184,
      "learning_rate": 1.8968809721783773e-05,
      "loss": 0.6982,
      "step": 4974
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.758351445198059,
      "learning_rate": 1.89683158364528e-05,
      "loss": 0.888,
      "step": 4975
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.787879228591919,
      "learning_rate": 1.8967821839310466e-05,
      "loss": 0.7421,
      "step": 4976
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4150317907333374,
      "learning_rate": 1.8967327730362928e-05,
      "loss": 0.8037,
      "step": 4977
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5066255331039429,
      "learning_rate": 1.896683350961635e-05,
      "loss": 0.7216,
      "step": 4978
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.9582817554473877,
      "learning_rate": 1.8966339177076895e-05,
      "loss": 0.699,
      "step": 4979
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0643527507781982,
      "learning_rate": 1.8965844732750724e-05,
      "loss": 0.8528,
      "step": 4980
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.427981972694397,
      "learning_rate": 1.8965350176644e-05,
      "loss": 0.8252,
      "step": 4981
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.497399926185608,
      "learning_rate": 1.896485550876289e-05,
      "loss": 0.624,
      "step": 4982
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5034981966018677,
      "learning_rate": 1.8964360729113565e-05,
      "loss": 0.8076,
      "step": 4983
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.66006338596344,
      "learning_rate": 1.8963865837702186e-05,
      "loss": 0.7117,
      "step": 4984
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5657875537872314,
      "learning_rate": 1.8963370834534932e-05,
      "loss": 0.6889,
      "step": 4985
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5173627138137817,
      "learning_rate": 1.8962875719617965e-05,
      "loss": 0.7322,
      "step": 4986
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6350706815719604,
      "learning_rate": 1.8962380492957463e-05,
      "loss": 0.7857,
      "step": 4987
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5604857206344604,
      "learning_rate": 1.89618851545596e-05,
      "loss": 0.8004,
      "step": 4988
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.639543890953064,
      "learning_rate": 1.896138970443055e-05,
      "loss": 0.6502,
      "step": 4989
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3674324750900269,
      "learning_rate": 1.8960894142576496e-05,
      "loss": 0.7153,
      "step": 4990
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4385229349136353,
      "learning_rate": 1.8960398469003607e-05,
      "loss": 0.8305,
      "step": 4991
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5406358242034912,
      "learning_rate": 1.8959902683718066e-05,
      "loss": 0.9047,
      "step": 4992
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3588608503341675,
      "learning_rate": 1.8959406786726058e-05,
      "loss": 0.6621,
      "step": 4993
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.581921935081482,
      "learning_rate": 1.895891077803376e-05,
      "loss": 0.7781,
      "step": 4994
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4703810214996338,
      "learning_rate": 1.895841465764736e-05,
      "loss": 0.721,
      "step": 4995
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5332987308502197,
      "learning_rate": 1.8957918425573045e-05,
      "loss": 0.6645,
      "step": 4996
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4142508506774902,
      "learning_rate": 1.8957422081816995e-05,
      "loss": 0.6509,
      "step": 4997
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3285611867904663,
      "learning_rate": 1.8956925626385404e-05,
      "loss": 0.8407,
      "step": 4998
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.746902346611023,
      "learning_rate": 1.895642905928446e-05,
      "loss": 0.7657,
      "step": 4999
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4408626556396484,
      "learning_rate": 1.8955932380520353e-05,
      "loss": 0.6666,
      "step": 5000
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6792184114456177,
      "learning_rate": 1.8955435590099273e-05,
      "loss": 0.7007,
      "step": 5001
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4665173292160034,
      "learning_rate": 1.895493868802742e-05,
      "loss": 0.7424,
      "step": 5002
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7345553636550903,
      "learning_rate": 1.895444167431098e-05,
      "loss": 0.8461,
      "step": 5003
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5776562690734863,
      "learning_rate": 1.8953944548956156e-05,
      "loss": 0.8584,
      "step": 5004
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5605404376983643,
      "learning_rate": 1.895344731196915e-05,
      "loss": 0.647,
      "step": 5005
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7368539571762085,
      "learning_rate": 1.895294996335615e-05,
      "loss": 0.6953,
      "step": 5006
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.092693328857422,
      "learning_rate": 1.8952452503123365e-05,
      "loss": 0.7958,
      "step": 5007
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5276298522949219,
      "learning_rate": 1.8951954931276996e-05,
      "loss": 0.7,
      "step": 5008
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7926961183547974,
      "learning_rate": 1.8951457247823238e-05,
      "loss": 0.6899,
      "step": 5009
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8976495265960693,
      "learning_rate": 1.8950959452768312e-05,
      "loss": 0.655,
      "step": 5010
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.803798794746399,
      "learning_rate": 1.895046154611841e-05,
      "loss": 0.6355,
      "step": 5011
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.496962070465088,
      "learning_rate": 1.8949963527879744e-05,
      "loss": 0.7467,
      "step": 5012
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8532614707946777,
      "learning_rate": 1.8949465398058522e-05,
      "loss": 0.8623,
      "step": 5013
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4936422109603882,
      "learning_rate": 1.894896715666096e-05,
      "loss": 0.7105,
      "step": 5014
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3250725269317627,
      "learning_rate": 1.8948468803693262e-05,
      "loss": 0.6568,
      "step": 5015
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4812674522399902,
      "learning_rate": 1.894797033916165e-05,
      "loss": 0.8272,
      "step": 5016
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8808248043060303,
      "learning_rate": 1.8947471763072326e-05,
      "loss": 0.5817,
      "step": 5017
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.68411386013031,
      "learning_rate": 1.894697307543152e-05,
      "loss": 0.6523,
      "step": 5018
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7287650108337402,
      "learning_rate": 1.8946474276245438e-05,
      "loss": 0.733,
      "step": 5019
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5805108547210693,
      "learning_rate": 1.8945975365520307e-05,
      "loss": 0.6743,
      "step": 5020
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7097914218902588,
      "learning_rate": 1.894547634326234e-05,
      "loss": 0.7238,
      "step": 5021
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6906242370605469,
      "learning_rate": 1.8944977209477763e-05,
      "loss": 0.6434,
      "step": 5022
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.203904867172241,
      "learning_rate": 1.8944477964172794e-05,
      "loss": 0.7464,
      "step": 5023
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9281797409057617,
      "learning_rate": 1.894397860735367e-05,
      "loss": 0.6192,
      "step": 5024
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.0385448932647705,
      "learning_rate": 1.89434791390266e-05,
      "loss": 0.9602,
      "step": 5025
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.531819224357605,
      "learning_rate": 1.894297955919782e-05,
      "loss": 0.808,
      "step": 5026
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5504848957061768,
      "learning_rate": 1.894247986787356e-05,
      "loss": 0.6463,
      "step": 5027
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5604944229125977,
      "learning_rate": 1.8941980065060046e-05,
      "loss": 0.6368,
      "step": 5028
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5232853889465332,
      "learning_rate": 1.894148015076351e-05,
      "loss": 0.7066,
      "step": 5029
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5053588151931763,
      "learning_rate": 1.8940980124990183e-05,
      "loss": 0.5694,
      "step": 5030
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6852303743362427,
      "learning_rate": 1.8940479987746304e-05,
      "loss": 0.6306,
      "step": 5031
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8074262142181396,
      "learning_rate": 1.8939979739038104e-05,
      "loss": 0.6181,
      "step": 5032
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8007681369781494,
      "learning_rate": 1.893947937887182e-05,
      "loss": 0.8442,
      "step": 5033
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8208621740341187,
      "learning_rate": 1.8938978907253692e-05,
      "loss": 0.7025,
      "step": 5034
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8335095643997192,
      "learning_rate": 1.8938478324189956e-05,
      "loss": 0.7499,
      "step": 5035
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5263910293579102,
      "learning_rate": 1.8937977629686862e-05,
      "loss": 0.7139,
      "step": 5036
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.647306203842163,
      "learning_rate": 1.893747682375064e-05,
      "loss": 0.8457,
      "step": 5037
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5014123916625977,
      "learning_rate": 1.8936975906387544e-05,
      "loss": 0.7072,
      "step": 5038
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6746100187301636,
      "learning_rate": 1.8936474877603816e-05,
      "loss": 0.7171,
      "step": 5039
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6957378387451172,
      "learning_rate": 1.89359737374057e-05,
      "loss": 0.7771,
      "step": 5040
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.768712043762207,
      "learning_rate": 1.8935472485799442e-05,
      "loss": 0.9004,
      "step": 5041
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7030800580978394,
      "learning_rate": 1.8934971122791298e-05,
      "loss": 0.6214,
      "step": 5042
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5057730674743652,
      "learning_rate": 1.8934469648387517e-05,
      "loss": 0.7009,
      "step": 5043
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.635236144065857,
      "learning_rate": 1.8933968062594347e-05,
      "loss": 0.6883,
      "step": 5044
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5048569440841675,
      "learning_rate": 1.8933466365418045e-05,
      "loss": 0.7118,
      "step": 5045
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7390224933624268,
      "learning_rate": 1.8932964556864865e-05,
      "loss": 0.5598,
      "step": 5046
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8220001459121704,
      "learning_rate": 1.893246263694106e-05,
      "loss": 0.7912,
      "step": 5047
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7532315254211426,
      "learning_rate": 1.8931960605652892e-05,
      "loss": 0.7824,
      "step": 5048
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.633011817932129,
      "learning_rate": 1.893145846300662e-05,
      "loss": 0.7199,
      "step": 5049
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5632244348526,
      "learning_rate": 1.8930956209008502e-05,
      "loss": 0.76,
      "step": 5050
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7804038524627686,
      "learning_rate": 1.89304538436648e-05,
      "loss": 0.6199,
      "step": 5051
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.9236475229263306,
      "learning_rate": 1.8929951366981784e-05,
      "loss": 0.8207,
      "step": 5052
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3823784589767456,
      "learning_rate": 1.8929448778965705e-05,
      "loss": 0.8289,
      "step": 5053
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2737112045288086,
      "learning_rate": 1.892894607962284e-05,
      "loss": 0.7344,
      "step": 5054
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5105904340744019,
      "learning_rate": 1.892844326895945e-05,
      "loss": 0.7782,
      "step": 5055
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8124922513961792,
      "learning_rate": 1.892794034698181e-05,
      "loss": 0.6846,
      "step": 5056
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7231929302215576,
      "learning_rate": 1.8927437313696188e-05,
      "loss": 0.8016,
      "step": 5057
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.368844747543335,
      "learning_rate": 1.892693416910885e-05,
      "loss": 0.6329,
      "step": 5058
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.702396035194397,
      "learning_rate": 1.8926430913226076e-05,
      "loss": 0.713,
      "step": 5059
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7641947269439697,
      "learning_rate": 1.8925927546054133e-05,
      "loss": 0.7432,
      "step": 5060
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7073255777359009,
      "learning_rate": 1.8925424067599307e-05,
      "loss": 0.7786,
      "step": 5061
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8924071788787842,
      "learning_rate": 1.8924920477867864e-05,
      "loss": 0.8507,
      "step": 5062
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5811831951141357,
      "learning_rate": 1.892441677686609e-05,
      "loss": 0.7649,
      "step": 5063
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5401155948638916,
      "learning_rate": 1.8923912964600263e-05,
      "loss": 0.7018,
      "step": 5064
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.561429738998413,
      "learning_rate": 1.8923409041076663e-05,
      "loss": 0.7023,
      "step": 5065
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6811736822128296,
      "learning_rate": 1.8922905006301575e-05,
      "loss": 0.731,
      "step": 5066
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6841198205947876,
      "learning_rate": 1.8922400860281278e-05,
      "loss": 0.8409,
      "step": 5067
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5958502292633057,
      "learning_rate": 1.8921896603022062e-05,
      "loss": 0.6288,
      "step": 5068
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5929138660430908,
      "learning_rate": 1.8921392234530212e-05,
      "loss": 0.8469,
      "step": 5069
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4277762174606323,
      "learning_rate": 1.8920887754812016e-05,
      "loss": 0.6393,
      "step": 5070
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4997378587722778,
      "learning_rate": 1.8920383163873766e-05,
      "loss": 0.784,
      "step": 5071
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8144595623016357,
      "learning_rate": 1.891987846172175e-05,
      "loss": 0.9334,
      "step": 5072
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4672905206680298,
      "learning_rate": 1.891937364836226e-05,
      "loss": 0.7272,
      "step": 5073
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4842438697814941,
      "learning_rate": 1.8918868723801592e-05,
      "loss": 0.6313,
      "step": 5074
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5264756679534912,
      "learning_rate": 1.891836368804604e-05,
      "loss": 0.729,
      "step": 5075
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.897550106048584,
      "learning_rate": 1.8917858541101898e-05,
      "loss": 0.708,
      "step": 5076
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6453784704208374,
      "learning_rate": 1.8917353282975474e-05,
      "loss": 0.8051,
      "step": 5077
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4199753999710083,
      "learning_rate": 1.8916847913673054e-05,
      "loss": 0.7078,
      "step": 5078
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5980181694030762,
      "learning_rate": 1.8916342433200942e-05,
      "loss": 0.8288,
      "step": 5079
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9178545475006104,
      "learning_rate": 1.8915836841565446e-05,
      "loss": 0.8081,
      "step": 5080
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.652463436126709,
      "learning_rate": 1.8915331138772863e-05,
      "loss": 0.7669,
      "step": 5081
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.611588478088379,
      "learning_rate": 1.89148253248295e-05,
      "loss": 0.7066,
      "step": 5082
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6439765691757202,
      "learning_rate": 1.891431939974167e-05,
      "loss": 0.8677,
      "step": 5083
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6231071949005127,
      "learning_rate": 1.891381336351567e-05,
      "loss": 0.7621,
      "step": 5084
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8796908855438232,
      "learning_rate": 1.8913307216157814e-05,
      "loss": 0.73,
      "step": 5085
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5264521837234497,
      "learning_rate": 1.891280095767441e-05,
      "loss": 0.9461,
      "step": 5086
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4314236640930176,
      "learning_rate": 1.891229458807177e-05,
      "loss": 0.5825,
      "step": 5087
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6394890546798706,
      "learning_rate": 1.891178810735621e-05,
      "loss": 0.6339,
      "step": 5088
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8068488836288452,
      "learning_rate": 1.8911281515534045e-05,
      "loss": 0.78,
      "step": 5089
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4819458723068237,
      "learning_rate": 1.8910774812611582e-05,
      "loss": 0.7292,
      "step": 5090
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7466689348220825,
      "learning_rate": 1.8910267998595152e-05,
      "loss": 0.7681,
      "step": 5091
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.440396785736084,
      "learning_rate": 1.8909761073491062e-05,
      "loss": 0.665,
      "step": 5092
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.586106300354004,
      "learning_rate": 1.890925403730564e-05,
      "loss": 0.7661,
      "step": 5093
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4472546577453613,
      "learning_rate": 1.89087468900452e-05,
      "loss": 0.7266,
      "step": 5094
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.428511142730713,
      "learning_rate": 1.890823963171607e-05,
      "loss": 0.7713,
      "step": 5095
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.712683916091919,
      "learning_rate": 1.8907732262324575e-05,
      "loss": 0.9364,
      "step": 5096
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6119004487991333,
      "learning_rate": 1.8907224781877038e-05,
      "loss": 0.9119,
      "step": 5097
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7324293851852417,
      "learning_rate": 1.8906717190379785e-05,
      "loss": 0.7359,
      "step": 5098
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4225859642028809,
      "learning_rate": 1.8906209487839146e-05,
      "loss": 0.7022,
      "step": 5099
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.478015422821045,
      "learning_rate": 1.8905701674261452e-05,
      "loss": 0.9204,
      "step": 5100
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4555890560150146,
      "learning_rate": 1.8905193749653033e-05,
      "loss": 0.7762,
      "step": 5101
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6721986532211304,
      "learning_rate": 1.8904685714020217e-05,
      "loss": 0.6821,
      "step": 5102
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.475538730621338,
      "learning_rate": 1.8904177567369348e-05,
      "loss": 0.6009,
      "step": 5103
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5767054557800293,
      "learning_rate": 1.890366930970675e-05,
      "loss": 0.6655,
      "step": 5104
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4646722078323364,
      "learning_rate": 1.8903160941038767e-05,
      "loss": 0.6718,
      "step": 5105
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5254437923431396,
      "learning_rate": 1.8902652461371735e-05,
      "loss": 0.7858,
      "step": 5106
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.444762945175171,
      "learning_rate": 1.890214387071199e-05,
      "loss": 0.5694,
      "step": 5107
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6981244087219238,
      "learning_rate": 1.890163516906588e-05,
      "loss": 0.9096,
      "step": 5108
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6643503904342651,
      "learning_rate": 1.8901126356439744e-05,
      "loss": 0.7395,
      "step": 5109
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6170780658721924,
      "learning_rate": 1.890061743283992e-05,
      "loss": 0.7521,
      "step": 5110
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9178402423858643,
      "learning_rate": 1.890010839827276e-05,
      "loss": 0.6656,
      "step": 5111
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6829679012298584,
      "learning_rate": 1.8899599252744608e-05,
      "loss": 0.7627,
      "step": 5112
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5299053192138672,
      "learning_rate": 1.8899089996261816e-05,
      "loss": 0.7783,
      "step": 5113
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6810067892074585,
      "learning_rate": 1.8898580628830725e-05,
      "loss": 0.7547,
      "step": 5114
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6246120929718018,
      "learning_rate": 1.889807115045769e-05,
      "loss": 0.6936,
      "step": 5115
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.825386643409729,
      "learning_rate": 1.889756156114906e-05,
      "loss": 0.8067,
      "step": 5116
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8822780847549438,
      "learning_rate": 1.8897051860911194e-05,
      "loss": 0.9058,
      "step": 5117
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8032410144805908,
      "learning_rate": 1.889654204975044e-05,
      "loss": 0.7834,
      "step": 5118
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9053089618682861,
      "learning_rate": 1.8896032127673162e-05,
      "loss": 0.7608,
      "step": 5119
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.595353126525879,
      "learning_rate": 1.889552209468571e-05,
      "loss": 0.6749,
      "step": 5120
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2524657249450684,
      "learning_rate": 1.889501195079444e-05,
      "loss": 0.6132,
      "step": 5121
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5239834785461426,
      "learning_rate": 1.8894501696005722e-05,
      "loss": 0.8161,
      "step": 5122
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5233205556869507,
      "learning_rate": 1.8893991330325913e-05,
      "loss": 0.7122,
      "step": 5123
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5123347043991089,
      "learning_rate": 1.8893480853761374e-05,
      "loss": 0.8446,
      "step": 5124
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5084118843078613,
      "learning_rate": 1.8892970266318473e-05,
      "loss": 0.8852,
      "step": 5125
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.316749334335327,
      "learning_rate": 1.889245956800357e-05,
      "loss": 0.6669,
      "step": 5126
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6721092462539673,
      "learning_rate": 1.8891948758823044e-05,
      "loss": 0.7881,
      "step": 5127
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6197028160095215,
      "learning_rate": 1.8891437838783247e-05,
      "loss": 0.7634,
      "step": 5128
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.712034821510315,
      "learning_rate": 1.8890926807890562e-05,
      "loss": 0.849,
      "step": 5129
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.702304720878601,
      "learning_rate": 1.8890415666151357e-05,
      "loss": 0.6626,
      "step": 5130
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5367681980133057,
      "learning_rate": 1.8889904413571996e-05,
      "loss": 0.8574,
      "step": 5131
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7900612354278564,
      "learning_rate": 1.8889393050158865e-05,
      "loss": 0.7791,
      "step": 5132
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9796513319015503,
      "learning_rate": 1.8888881575918334e-05,
      "loss": 0.7557,
      "step": 5133
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.894856333732605,
      "learning_rate": 1.888836999085678e-05,
      "loss": 0.6775,
      "step": 5134
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.379258155822754,
      "learning_rate": 1.888785829498058e-05,
      "loss": 0.6661,
      "step": 5135
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4044814109802246,
      "learning_rate": 1.8887346488296117e-05,
      "loss": 0.6988,
      "step": 5136
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.872663140296936,
      "learning_rate": 1.8886834570809767e-05,
      "loss": 0.8512,
      "step": 5137
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7205660343170166,
      "learning_rate": 1.888632254252792e-05,
      "loss": 0.7485,
      "step": 5138
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6978062391281128,
      "learning_rate": 1.8885810403456947e-05,
      "loss": 0.8272,
      "step": 5139
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.368780493736267,
      "learning_rate": 1.8885298153603243e-05,
      "loss": 0.8423,
      "step": 5140
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.9565509557724,
      "learning_rate": 1.8884785792973192e-05,
      "loss": 0.7703,
      "step": 5141
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4534567594528198,
      "learning_rate": 1.8884273321573183e-05,
      "loss": 0.717,
      "step": 5142
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6284669637680054,
      "learning_rate": 1.8883760739409606e-05,
      "loss": 0.7091,
      "step": 5143
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4946238994598389,
      "learning_rate": 1.8883248046488845e-05,
      "loss": 0.7877,
      "step": 5144
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.912851095199585,
      "learning_rate": 1.88827352428173e-05,
      "loss": 0.7303,
      "step": 5145
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7888046503067017,
      "learning_rate": 1.8882222328401355e-05,
      "loss": 0.7406,
      "step": 5146
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2105988264083862,
      "learning_rate": 1.8881709303247418e-05,
      "loss": 0.8537,
      "step": 5147
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7742866277694702,
      "learning_rate": 1.8881196167361873e-05,
      "loss": 0.6516,
      "step": 5148
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8856185674667358,
      "learning_rate": 1.8880682920751124e-05,
      "loss": 0.7953,
      "step": 5149
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4506611824035645,
      "learning_rate": 1.8880169563421567e-05,
      "loss": 0.716,
      "step": 5150
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.82158362865448,
      "learning_rate": 1.88796560953796e-05,
      "loss": 0.7276,
      "step": 5151
    },
    {
      "epoch": 0.53,
      "grad_norm": 2.0893807411193848,
      "learning_rate": 1.887914251663163e-05,
      "loss": 0.7957,
      "step": 5152
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3877887725830078,
      "learning_rate": 1.8878628827184057e-05,
      "loss": 0.8194,
      "step": 5153
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7052421569824219,
      "learning_rate": 1.8878115027043288e-05,
      "loss": 0.7972,
      "step": 5154
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4832688570022583,
      "learning_rate": 1.8877601116215725e-05,
      "loss": 0.7062,
      "step": 5155
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6116783618927002,
      "learning_rate": 1.8877087094707777e-05,
      "loss": 0.74,
      "step": 5156
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4800893068313599,
      "learning_rate": 1.8876572962525854e-05,
      "loss": 0.673,
      "step": 5157
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4285370111465454,
      "learning_rate": 1.8876058719676362e-05,
      "loss": 0.7548,
      "step": 5158
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6180336475372314,
      "learning_rate": 1.8875544366165715e-05,
      "loss": 0.6605,
      "step": 5159
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2861340045928955,
      "learning_rate": 1.8875029902000324e-05,
      "loss": 0.6519,
      "step": 5160
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5348395109176636,
      "learning_rate": 1.887451532718661e-05,
      "loss": 0.547,
      "step": 5161
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8065637350082397,
      "learning_rate": 1.8874000641730978e-05,
      "loss": 0.6955,
      "step": 5162
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.517734169960022,
      "learning_rate": 1.8873485845639846e-05,
      "loss": 0.6335,
      "step": 5163
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6971497535705566,
      "learning_rate": 1.887297093891964e-05,
      "loss": 0.6521,
      "step": 5164
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.920454978942871,
      "learning_rate": 1.8872455921576772e-05,
      "loss": 0.8296,
      "step": 5165
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.599506139755249,
      "learning_rate": 1.8871940793617668e-05,
      "loss": 0.8661,
      "step": 5166
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8405522108078003,
      "learning_rate": 1.887142555504875e-05,
      "loss": 0.9771,
      "step": 5167
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.605847716331482,
      "learning_rate": 1.8870910205876437e-05,
      "loss": 0.805,
      "step": 5168
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6377278566360474,
      "learning_rate": 1.8870394746107154e-05,
      "loss": 0.6776,
      "step": 5169
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8154261112213135,
      "learning_rate": 1.8869879175747335e-05,
      "loss": 0.5919,
      "step": 5170
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8209259510040283,
      "learning_rate": 1.88693634948034e-05,
      "loss": 0.7536,
      "step": 5171
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5492339134216309,
      "learning_rate": 1.8868847703281786e-05,
      "loss": 0.642,
      "step": 5172
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6599467992782593,
      "learning_rate": 1.8868331801188914e-05,
      "loss": 0.8314,
      "step": 5173
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4329471588134766,
      "learning_rate": 1.8867815788531224e-05,
      "loss": 0.6085,
      "step": 5174
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8323246240615845,
      "learning_rate": 1.8867299665315143e-05,
      "loss": 0.8468,
      "step": 5175
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7437596321105957,
      "learning_rate": 1.8866783431547112e-05,
      "loss": 0.7435,
      "step": 5176
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.652599573135376,
      "learning_rate": 1.8866267087233564e-05,
      "loss": 0.7706,
      "step": 5177
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4354356527328491,
      "learning_rate": 1.8865750632380932e-05,
      "loss": 0.695,
      "step": 5178
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6596887111663818,
      "learning_rate": 1.8865234066995664e-05,
      "loss": 0.6893,
      "step": 5179
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5358061790466309,
      "learning_rate": 1.8864717391084194e-05,
      "loss": 0.7984,
      "step": 5180
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.840928554534912,
      "learning_rate": 1.8864200604652963e-05,
      "loss": 0.8479,
      "step": 5181
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3610259294509888,
      "learning_rate": 1.886368370770842e-05,
      "loss": 0.6837,
      "step": 5182
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8942384719848633,
      "learning_rate": 1.8863166700257004e-05,
      "loss": 0.751,
      "step": 5183
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4215049743652344,
      "learning_rate": 1.8862649582305158e-05,
      "loss": 0.5741,
      "step": 5184
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7213095426559448,
      "learning_rate": 1.8862132353859338e-05,
      "loss": 0.6879,
      "step": 5185
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.331766963005066,
      "learning_rate": 1.8861615014925984e-05,
      "loss": 0.5804,
      "step": 5186
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4701416492462158,
      "learning_rate": 1.8861097565511552e-05,
      "loss": 0.6941,
      "step": 5187
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5047578811645508,
      "learning_rate": 1.8860580005622492e-05,
      "loss": 0.8108,
      "step": 5188
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.552116870880127,
      "learning_rate": 1.886006233526525e-05,
      "loss": 0.5884,
      "step": 5189
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8398743867874146,
      "learning_rate": 1.885954455444629e-05,
      "loss": 0.7758,
      "step": 5190
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3839435577392578,
      "learning_rate": 1.885902666317206e-05,
      "loss": 0.6454,
      "step": 5191
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4992575645446777,
      "learning_rate": 1.8858508661449026e-05,
      "loss": 0.712,
      "step": 5192
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5885120630264282,
      "learning_rate": 1.8857990549283635e-05,
      "loss": 0.7112,
      "step": 5193
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.711678147315979,
      "learning_rate": 1.885747232668235e-05,
      "loss": 0.7603,
      "step": 5194
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7746647596359253,
      "learning_rate": 1.8856953993651634e-05,
      "loss": 0.6341,
      "step": 5195
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6755532026290894,
      "learning_rate": 1.8856435550197946e-05,
      "loss": 0.6637,
      "step": 5196
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5980894565582275,
      "learning_rate": 1.8855916996327756e-05,
      "loss": 0.7237,
      "step": 5197
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6001278162002563,
      "learning_rate": 1.885539833204752e-05,
      "loss": 0.7357,
      "step": 5198
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4394969940185547,
      "learning_rate": 1.8854879557363716e-05,
      "loss": 0.7403,
      "step": 5199
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.587976098060608,
      "learning_rate": 1.8854360672282797e-05,
      "loss": 0.8039,
      "step": 5200
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6885689496994019,
      "learning_rate": 1.8853841676811246e-05,
      "loss": 0.8301,
      "step": 5201
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5282169580459595,
      "learning_rate": 1.8853322570955526e-05,
      "loss": 0.7659,
      "step": 5202
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7535568475723267,
      "learning_rate": 1.885280335472211e-05,
      "loss": 0.7167,
      "step": 5203
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.9149724245071411,
      "learning_rate": 1.8852284028117474e-05,
      "loss": 0.7723,
      "step": 5204
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6706631183624268,
      "learning_rate": 1.8851764591148088e-05,
      "loss": 0.6508,
      "step": 5205
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5190078020095825,
      "learning_rate": 1.8851245043820433e-05,
      "loss": 0.6714,
      "step": 5206
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4015488624572754,
      "learning_rate": 1.885072538614098e-05,
      "loss": 0.7823,
      "step": 5207
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6601308584213257,
      "learning_rate": 1.8850205618116213e-05,
      "loss": 0.7606,
      "step": 5208
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5306479930877686,
      "learning_rate": 1.884968573975261e-05,
      "loss": 0.8172,
      "step": 5209
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4660325050354004,
      "learning_rate": 1.8849165751056654e-05,
      "loss": 0.8151,
      "step": 5210
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.104795217514038,
      "learning_rate": 1.884864565203483e-05,
      "loss": 0.8754,
      "step": 5211
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.77652907371521,
      "learning_rate": 1.8848125442693616e-05,
      "loss": 0.6862,
      "step": 5212
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6718440055847168,
      "learning_rate": 1.8847605123039504e-05,
      "loss": 0.8321,
      "step": 5213
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.376044750213623,
      "learning_rate": 1.8847084693078974e-05,
      "loss": 0.6644,
      "step": 5214
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8247731924057007,
      "learning_rate": 1.884656415281852e-05,
      "loss": 0.859,
      "step": 5215
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.69974946975708,
      "learning_rate": 1.8846043502264635e-05,
      "loss": 0.9824,
      "step": 5216
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8314398527145386,
      "learning_rate": 1.88455227414238e-05,
      "loss": 0.7276,
      "step": 5217
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5582917928695679,
      "learning_rate": 1.8845001870302514e-05,
      "loss": 0.8096,
      "step": 5218
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.797414779663086,
      "learning_rate": 1.8844480888907272e-05,
      "loss": 0.7333,
      "step": 5219
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8333460092544556,
      "learning_rate": 1.8843959797244563e-05,
      "loss": 0.8385,
      "step": 5220
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4072418212890625,
      "learning_rate": 1.8843438595320892e-05,
      "loss": 0.7649,
      "step": 5221
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3785431385040283,
      "learning_rate": 1.884291728314275e-05,
      "loss": 0.7154,
      "step": 5222
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8799482583999634,
      "learning_rate": 1.884239586071664e-05,
      "loss": 0.7865,
      "step": 5223
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7863094806671143,
      "learning_rate": 1.8841874328049063e-05,
      "loss": 0.7353,
      "step": 5224
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.434215784072876,
      "learning_rate": 1.884135268514652e-05,
      "loss": 0.6321,
      "step": 5225
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5821515321731567,
      "learning_rate": 1.8840830932015512e-05,
      "loss": 0.7446,
      "step": 5226
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4409613609313965,
      "learning_rate": 1.884030906866255e-05,
      "loss": 0.8409,
      "step": 5227
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5250756740570068,
      "learning_rate": 1.8839787095094132e-05,
      "loss": 0.8418,
      "step": 5228
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4428189992904663,
      "learning_rate": 1.8839265011316776e-05,
      "loss": 0.6922,
      "step": 5229
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8092008829116821,
      "learning_rate": 1.883874281733698e-05,
      "loss": 0.7054,
      "step": 5230
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6823917627334595,
      "learning_rate": 1.8838220513161264e-05,
      "loss": 0.6834,
      "step": 5231
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5490350723266602,
      "learning_rate": 1.8837698098796133e-05,
      "loss": 0.8078,
      "step": 5232
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.808021903038025,
      "learning_rate": 1.8837175574248102e-05,
      "loss": 0.7684,
      "step": 5233
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6960408687591553,
      "learning_rate": 1.883665293952369e-05,
      "loss": 0.8173,
      "step": 5234
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8840916156768799,
      "learning_rate": 1.8836130194629407e-05,
      "loss": 0.7701,
      "step": 5235
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7279194593429565,
      "learning_rate": 1.8835607339571774e-05,
      "loss": 0.651,
      "step": 5236
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.805235505104065,
      "learning_rate": 1.8835084374357305e-05,
      "loss": 0.666,
      "step": 5237
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8359366655349731,
      "learning_rate": 1.8834561298992528e-05,
      "loss": 0.7974,
      "step": 5238
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5895776748657227,
      "learning_rate": 1.8834038113483953e-05,
      "loss": 0.742,
      "step": 5239
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5986095666885376,
      "learning_rate": 1.883351481783811e-05,
      "loss": 0.8779,
      "step": 5240
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7001869678497314,
      "learning_rate": 1.8832991412061525e-05,
      "loss": 0.7204,
      "step": 5241
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.741682767868042,
      "learning_rate": 1.883246789616072e-05,
      "loss": 0.7341,
      "step": 5242
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.439016580581665,
      "learning_rate": 1.8831944270142224e-05,
      "loss": 0.7518,
      "step": 5243
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6286519765853882,
      "learning_rate": 1.883142053401256e-05,
      "loss": 0.9735,
      "step": 5244
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8495901823043823,
      "learning_rate": 1.8830896687778263e-05,
      "loss": 0.6986,
      "step": 5245
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4597132205963135,
      "learning_rate": 1.8830372731445864e-05,
      "loss": 0.5237,
      "step": 5246
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6086125373840332,
      "learning_rate": 1.8829848665021893e-05,
      "loss": 0.6356,
      "step": 5247
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6793501377105713,
      "learning_rate": 1.8829324488512885e-05,
      "loss": 0.842,
      "step": 5248
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.467831015586853,
      "learning_rate": 1.8828800201925372e-05,
      "loss": 0.7888,
      "step": 5249
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5028825998306274,
      "learning_rate": 1.8828275805265897e-05,
      "loss": 0.7796,
      "step": 5250
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.610846757888794,
      "learning_rate": 1.8827751298540993e-05,
      "loss": 0.8562,
      "step": 5251
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6851190328598022,
      "learning_rate": 1.8827226681757198e-05,
      "loss": 0.7749,
      "step": 5252
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.291918158531189,
      "learning_rate": 1.8826701954921055e-05,
      "loss": 0.6567,
      "step": 5253
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6970264911651611,
      "learning_rate": 1.882617711803911e-05,
      "loss": 0.6505,
      "step": 5254
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7633671760559082,
      "learning_rate": 1.8825652171117898e-05,
      "loss": 0.7666,
      "step": 5255
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3737740516662598,
      "learning_rate": 1.882512711416397e-05,
      "loss": 0.8086,
      "step": 5256
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7545831203460693,
      "learning_rate": 1.8824601947183865e-05,
      "loss": 0.8077,
      "step": 5257
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5777002573013306,
      "learning_rate": 1.8824076670184143e-05,
      "loss": 0.7578,
      "step": 5258
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7544686794281006,
      "learning_rate": 1.882355128317134e-05,
      "loss": 0.8447,
      "step": 5259
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.734057068824768,
      "learning_rate": 1.8823025786152008e-05,
      "loss": 0.6686,
      "step": 5260
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.786339521408081,
      "learning_rate": 1.882250017913271e-05,
      "loss": 0.5943,
      "step": 5261
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5141874551773071,
      "learning_rate": 1.8821974462119986e-05,
      "loss": 0.7426,
      "step": 5262
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6201695203781128,
      "learning_rate": 1.8821448635120395e-05,
      "loss": 0.7377,
      "step": 5263
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.329466462135315,
      "learning_rate": 1.8820922698140494e-05,
      "loss": 0.7585,
      "step": 5264
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6171547174453735,
      "learning_rate": 1.882039665118684e-05,
      "loss": 0.7902,
      "step": 5265
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6758934259414673,
      "learning_rate": 1.8819870494265992e-05,
      "loss": 0.6804,
      "step": 5266
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5989750623703003,
      "learning_rate": 1.8819344227384506e-05,
      "loss": 0.6363,
      "step": 5267
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.319484829902649,
      "learning_rate": 1.8818817850548944e-05,
      "loss": 0.7282,
      "step": 5268
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.931064486503601,
      "learning_rate": 1.881829136376587e-05,
      "loss": 0.8017,
      "step": 5269
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.613527536392212,
      "learning_rate": 1.881776476704185e-05,
      "loss": 0.6599,
      "step": 5270
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.334223747253418,
      "learning_rate": 1.881723806038344e-05,
      "loss": 0.8249,
      "step": 5271
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9572968482971191,
      "learning_rate": 1.8816711243797223e-05,
      "loss": 0.7567,
      "step": 5272
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.839255452156067,
      "learning_rate": 1.8816184317289754e-05,
      "loss": 0.7466,
      "step": 5273
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.706498384475708,
      "learning_rate": 1.8815657280867607e-05,
      "loss": 0.7036,
      "step": 5274
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2959132194519043,
      "learning_rate": 1.881513013453735e-05,
      "loss": 0.7911,
      "step": 5275
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6570614576339722,
      "learning_rate": 1.8814602878305556e-05,
      "loss": 0.8315,
      "step": 5276
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.617505431175232,
      "learning_rate": 1.8814075512178802e-05,
      "loss": 0.7352,
      "step": 5277
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.350177526473999,
      "learning_rate": 1.8813548036163662e-05,
      "loss": 0.7468,
      "step": 5278
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6745985746383667,
      "learning_rate": 1.8813020450266705e-05,
      "loss": 0.7222,
      "step": 5279
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4578156471252441,
      "learning_rate": 1.8812492754494518e-05,
      "loss": 0.7468,
      "step": 5280
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.586411476135254,
      "learning_rate": 1.8811964948853677e-05,
      "loss": 0.6916,
      "step": 5281
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7644915580749512,
      "learning_rate": 1.881143703335076e-05,
      "loss": 0.6759,
      "step": 5282
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7846932411193848,
      "learning_rate": 1.881090900799235e-05,
      "loss": 0.7609,
      "step": 5283
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6709266901016235,
      "learning_rate": 1.881038087278503e-05,
      "loss": 0.7595,
      "step": 5284
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4380658864974976,
      "learning_rate": 1.8809852627735385e-05,
      "loss": 0.5894,
      "step": 5285
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.439538836479187,
      "learning_rate": 1.880932427285e-05,
      "loss": 0.7341,
      "step": 5286
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5128718614578247,
      "learning_rate": 1.8808795808135465e-05,
      "loss": 0.8096,
      "step": 5287
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5653247833251953,
      "learning_rate": 1.8808267233598365e-05,
      "loss": 0.7528,
      "step": 5288
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4123131036758423,
      "learning_rate": 1.880773854924529e-05,
      "loss": 0.7788,
      "step": 5289
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.615679383277893,
      "learning_rate": 1.8807209755082838e-05,
      "loss": 0.7689,
      "step": 5290
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3934006690979004,
      "learning_rate": 1.880668085111759e-05,
      "loss": 0.63,
      "step": 5291
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.766560435295105,
      "learning_rate": 1.880615183735615e-05,
      "loss": 0.7557,
      "step": 5292
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.022122383117676,
      "learning_rate": 1.880562271380511e-05,
      "loss": 0.7229,
      "step": 5293
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.479962944984436,
      "learning_rate": 1.8805093480471065e-05,
      "loss": 0.7565,
      "step": 5294
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6541807651519775,
      "learning_rate": 1.8804564137360613e-05,
      "loss": 0.722,
      "step": 5295
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7666068077087402,
      "learning_rate": 1.880403468448036e-05,
      "loss": 0.8626,
      "step": 5296
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7044689655303955,
      "learning_rate": 1.8803505121836898e-05,
      "loss": 0.9413,
      "step": 5297
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.508827805519104,
      "learning_rate": 1.8802975449436835e-05,
      "loss": 0.83,
      "step": 5298
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.745030164718628,
      "learning_rate": 1.8802445667286774e-05,
      "loss": 0.6155,
      "step": 5299
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.415090560913086,
      "learning_rate": 1.8801915775393314e-05,
      "loss": 0.7856,
      "step": 5300
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6463031768798828,
      "learning_rate": 1.880138577376307e-05,
      "loss": 0.8382,
      "step": 5301
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7940846681594849,
      "learning_rate": 1.880085566240265e-05,
      "loss": 0.7278,
      "step": 5302
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5946993827819824,
      "learning_rate": 1.8800325441318655e-05,
      "loss": 0.8016,
      "step": 5303
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.798819661140442,
      "learning_rate": 1.87997951105177e-05,
      "loss": 0.7269,
      "step": 5304
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4968844652175903,
      "learning_rate": 1.8799264670006395e-05,
      "loss": 0.6118,
      "step": 5305
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4376647472381592,
      "learning_rate": 1.8798734119791357e-05,
      "loss": 0.7451,
      "step": 5306
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3978267908096313,
      "learning_rate": 1.87982034598792e-05,
      "loss": 0.6485,
      "step": 5307
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.54329252243042,
      "learning_rate": 1.8797672690276535e-05,
      "loss": 0.6436,
      "step": 5308
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.611088514328003,
      "learning_rate": 1.879714181098998e-05,
      "loss": 0.7055,
      "step": 5309
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.647961139678955,
      "learning_rate": 1.879661082202616e-05,
      "loss": 0.9331,
      "step": 5310
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.490351676940918,
      "learning_rate": 1.879607972339169e-05,
      "loss": 0.6711,
      "step": 5311
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.363988995552063,
      "learning_rate": 1.8795548515093195e-05,
      "loss": 0.7274,
      "step": 5312
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8537237644195557,
      "learning_rate": 1.879501719713729e-05,
      "loss": 0.7457,
      "step": 5313
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6349440813064575,
      "learning_rate": 1.879448576953061e-05,
      "loss": 0.7019,
      "step": 5314
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7505075931549072,
      "learning_rate": 1.8793954232279776e-05,
      "loss": 0.8851,
      "step": 5315
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.840103268623352,
      "learning_rate": 1.8793422585391408e-05,
      "loss": 0.7299,
      "step": 5316
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5740208625793457,
      "learning_rate": 1.879289082887214e-05,
      "loss": 0.7521,
      "step": 5317
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5370711088180542,
      "learning_rate": 1.8792358962728607e-05,
      "loss": 0.9477,
      "step": 5318
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4231441020965576,
      "learning_rate": 1.879182698696743e-05,
      "loss": 0.6318,
      "step": 5319
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.550729751586914,
      "learning_rate": 1.879129490159525e-05,
      "loss": 0.71,
      "step": 5320
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.582088828086853,
      "learning_rate": 1.8790762706618695e-05,
      "loss": 0.7661,
      "step": 5321
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9609192609786987,
      "learning_rate": 1.8790230402044402e-05,
      "loss": 0.7277,
      "step": 5322
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9242677688598633,
      "learning_rate": 1.878969798787901e-05,
      "loss": 0.8417,
      "step": 5323
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6486186981201172,
      "learning_rate": 1.878916546412915e-05,
      "loss": 0.9117,
      "step": 5324
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5846295356750488,
      "learning_rate": 1.8788632830801464e-05,
      "loss": 0.7083,
      "step": 5325
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6157153844833374,
      "learning_rate": 1.87881000879026e-05,
      "loss": 0.5867,
      "step": 5326
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3325564861297607,
      "learning_rate": 1.8787567235439187e-05,
      "loss": 0.7503,
      "step": 5327
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5073994398117065,
      "learning_rate": 1.8787034273417883e-05,
      "loss": 0.8159,
      "step": 5328
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5602761507034302,
      "learning_rate": 1.878650120184532e-05,
      "loss": 0.7464,
      "step": 5329
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6727573871612549,
      "learning_rate": 1.8785968020728145e-05,
      "loss": 0.8414,
      "step": 5330
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7018918991088867,
      "learning_rate": 1.8785434730073013e-05,
      "loss": 0.7915,
      "step": 5331
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8519796133041382,
      "learning_rate": 1.878490132988657e-05,
      "loss": 0.7841,
      "step": 5332
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4078351259231567,
      "learning_rate": 1.8784367820175463e-05,
      "loss": 0.7671,
      "step": 5333
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.641674518585205,
      "learning_rate": 1.8783834200946343e-05,
      "loss": 0.7304,
      "step": 5334
    },
    {
      "epoch": 0.55,
      "grad_norm": 2.015178680419922,
      "learning_rate": 1.878330047220587e-05,
      "loss": 0.8201,
      "step": 5335
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5996325016021729,
      "learning_rate": 1.878276663396069e-05,
      "loss": 0.8228,
      "step": 5336
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7870835065841675,
      "learning_rate": 1.878223268621746e-05,
      "loss": 0.7726,
      "step": 5337
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5707099437713623,
      "learning_rate": 1.8781698628982843e-05,
      "loss": 0.7861,
      "step": 5338
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5558485984802246,
      "learning_rate": 1.878116446226349e-05,
      "loss": 0.9029,
      "step": 5339
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7236605882644653,
      "learning_rate": 1.8780630186066067e-05,
      "loss": 0.6492,
      "step": 5340
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6257964372634888,
      "learning_rate": 1.878009580039723e-05,
      "loss": 0.8633,
      "step": 5341
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6580928564071655,
      "learning_rate": 1.8779561305263643e-05,
      "loss": 0.7995,
      "step": 5342
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4703072309494019,
      "learning_rate": 1.8779026700671968e-05,
      "loss": 0.6831,
      "step": 5343
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8656258583068848,
      "learning_rate": 1.8778491986628875e-05,
      "loss": 0.7457,
      "step": 5344
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.583504557609558,
      "learning_rate": 1.8777957163141027e-05,
      "loss": 0.7384,
      "step": 5345
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5353294610977173,
      "learning_rate": 1.8777422230215096e-05,
      "loss": 0.6604,
      "step": 5346
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6106985807418823,
      "learning_rate": 1.8776887187857743e-05,
      "loss": 0.7171,
      "step": 5347
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5392951965332031,
      "learning_rate": 1.877635203607564e-05,
      "loss": 0.7359,
      "step": 5348
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5908432006835938,
      "learning_rate": 1.877581677487547e-05,
      "loss": 0.5477,
      "step": 5349
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8264795541763306,
      "learning_rate": 1.8775281404263893e-05,
      "loss": 0.8191,
      "step": 5350
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5611482858657837,
      "learning_rate": 1.877474592424759e-05,
      "loss": 0.6182,
      "step": 5351
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9382888078689575,
      "learning_rate": 1.8774210334833233e-05,
      "loss": 0.8329,
      "step": 5352
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3871331214904785,
      "learning_rate": 1.8773674636027508e-05,
      "loss": 0.5355,
      "step": 5353
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4553998708724976,
      "learning_rate": 1.8773138827837086e-05,
      "loss": 0.8692,
      "step": 5354
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7390177249908447,
      "learning_rate": 1.877260291026865e-05,
      "loss": 0.6864,
      "step": 5355
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5490450859069824,
      "learning_rate": 1.8772066883328878e-05,
      "loss": 0.6649,
      "step": 5356
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9103193283081055,
      "learning_rate": 1.877153074702446e-05,
      "loss": 0.8554,
      "step": 5357
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4513291120529175,
      "learning_rate": 1.8770994501362075e-05,
      "loss": 0.5899,
      "step": 5358
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6742669343948364,
      "learning_rate": 1.8770458146348406e-05,
      "loss": 0.6354,
      "step": 5359
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4390125274658203,
      "learning_rate": 1.8769921681990145e-05,
      "loss": 0.739,
      "step": 5360
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.8006505966186523,
      "learning_rate": 1.876938510829398e-05,
      "loss": 0.8729,
      "step": 5361
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7178375720977783,
      "learning_rate": 1.87688484252666e-05,
      "loss": 0.7165,
      "step": 5362
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3380941152572632,
      "learning_rate": 1.8768311632914695e-05,
      "loss": 0.56,
      "step": 5363
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7950841188430786,
      "learning_rate": 1.8767774731244955e-05,
      "loss": 0.7251,
      "step": 5364
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.845207929611206,
      "learning_rate": 1.876723772026408e-05,
      "loss": 0.895,
      "step": 5365
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6069488525390625,
      "learning_rate": 1.876670059997876e-05,
      "loss": 0.8475,
      "step": 5366
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7598583698272705,
      "learning_rate": 1.8766163370395692e-05,
      "loss": 0.7169,
      "step": 5367
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4895241260528564,
      "learning_rate": 1.876562603152158e-05,
      "loss": 0.7478,
      "step": 5368
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8273049592971802,
      "learning_rate": 1.8765088583363112e-05,
      "loss": 0.7405,
      "step": 5369
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4323699474334717,
      "learning_rate": 1.8764551025926997e-05,
      "loss": 0.86,
      "step": 5370
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.626625657081604,
      "learning_rate": 1.8764013359219938e-05,
      "loss": 0.769,
      "step": 5371
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6202282905578613,
      "learning_rate": 1.876347558324863e-05,
      "loss": 0.9186,
      "step": 5372
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9575496912002563,
      "learning_rate": 1.8762937698019788e-05,
      "loss": 0.8152,
      "step": 5373
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4696245193481445,
      "learning_rate": 1.8762399703540107e-05,
      "loss": 0.4616,
      "step": 5374
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8802833557128906,
      "learning_rate": 1.8761861599816303e-05,
      "loss": 0.9185,
      "step": 5375
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5155476331710815,
      "learning_rate": 1.876132338685508e-05,
      "loss": 0.6076,
      "step": 5376
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.634395718574524,
      "learning_rate": 1.8760785064663153e-05,
      "loss": 0.5707,
      "step": 5377
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5500082969665527,
      "learning_rate": 1.8760246633247228e-05,
      "loss": 0.7964,
      "step": 5378
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.575156569480896,
      "learning_rate": 1.875970809261402e-05,
      "loss": 0.8195,
      "step": 5379
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5488401651382446,
      "learning_rate": 1.8759169442770244e-05,
      "loss": 0.7789,
      "step": 5380
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8135697841644287,
      "learning_rate": 1.8758630683722616e-05,
      "loss": 0.7813,
      "step": 5381
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6507500410079956,
      "learning_rate": 1.875809181547785e-05,
      "loss": 0.7441,
      "step": 5382
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0189473628997803,
      "learning_rate": 1.875755283804267e-05,
      "loss": 0.736,
      "step": 5383
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.773606538772583,
      "learning_rate": 1.8757013751423784e-05,
      "loss": 0.7144,
      "step": 5384
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.706531047821045,
      "learning_rate": 1.875647455562793e-05,
      "loss": 0.7616,
      "step": 5385
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3465721607208252,
      "learning_rate": 1.8755935250661817e-05,
      "loss": 0.6606,
      "step": 5386
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.890619158744812,
      "learning_rate": 1.8755395836532168e-05,
      "loss": 0.6674,
      "step": 5387
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8323928117752075,
      "learning_rate": 1.875485631324572e-05,
      "loss": 0.6732,
      "step": 5388
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8512516021728516,
      "learning_rate": 1.875431668080919e-05,
      "loss": 0.7729,
      "step": 5389
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3182779550552368,
      "learning_rate": 1.8753776939229306e-05,
      "loss": 0.6698,
      "step": 5390
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.735281229019165,
      "learning_rate": 1.8753237088512805e-05,
      "loss": 0.9254,
      "step": 5391
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.584434151649475,
      "learning_rate": 1.8752697128666405e-05,
      "loss": 0.6491,
      "step": 5392
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5019809007644653,
      "learning_rate": 1.8752157059696848e-05,
      "loss": 0.6658,
      "step": 5393
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6573164463043213,
      "learning_rate": 1.8751616881610864e-05,
      "loss": 0.6859,
      "step": 5394
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9774733781814575,
      "learning_rate": 1.8751076594415185e-05,
      "loss": 0.7233,
      "step": 5395
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6767505407333374,
      "learning_rate": 1.8750536198116552e-05,
      "loss": 0.7358,
      "step": 5396
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4477007389068604,
      "learning_rate": 1.87499956927217e-05,
      "loss": 0.7163,
      "step": 5397
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6617157459259033,
      "learning_rate": 1.8749455078237368e-05,
      "loss": 0.745,
      "step": 5398
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4365413188934326,
      "learning_rate": 1.874891435467029e-05,
      "loss": 0.6098,
      "step": 5399
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.665893316268921,
      "learning_rate": 1.874837352202722e-05,
      "loss": 0.6616,
      "step": 5400
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5688947439193726,
      "learning_rate": 1.874783258031489e-05,
      "loss": 0.7802,
      "step": 5401
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.589381217956543,
      "learning_rate": 1.874729152954005e-05,
      "loss": 0.7072,
      "step": 5402
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.067918300628662,
      "learning_rate": 1.8746750369709442e-05,
      "loss": 0.7778,
      "step": 5403
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5721107721328735,
      "learning_rate": 1.8746209100829814e-05,
      "loss": 0.8344,
      "step": 5404
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3195514678955078,
      "learning_rate": 1.8745667722907918e-05,
      "loss": 0.7976,
      "step": 5405
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5751566886901855,
      "learning_rate": 1.8745126235950493e-05,
      "loss": 0.6599,
      "step": 5406
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3099262714385986,
      "learning_rate": 1.87445846399643e-05,
      "loss": 0.719,
      "step": 5407
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6254935264587402,
      "learning_rate": 1.874404293495609e-05,
      "loss": 0.7596,
      "step": 5408
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5679810047149658,
      "learning_rate": 1.8743501120932612e-05,
      "loss": 0.7525,
      "step": 5409
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3483601808547974,
      "learning_rate": 1.8742959197900626e-05,
      "loss": 0.6439,
      "step": 5410
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4259144067764282,
      "learning_rate": 1.8742417165866885e-05,
      "loss": 0.5006,
      "step": 5411
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2437467575073242,
      "learning_rate": 1.8741875024838146e-05,
      "loss": 0.6543,
      "step": 5412
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.277844786643982,
      "learning_rate": 1.8741332774821177e-05,
      "loss": 0.6467,
      "step": 5413
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7748216390609741,
      "learning_rate": 1.8740790415822725e-05,
      "loss": 0.7638,
      "step": 5414
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5583605766296387,
      "learning_rate": 1.8740247947849562e-05,
      "loss": 0.604,
      "step": 5415
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5407257080078125,
      "learning_rate": 1.8739705370908443e-05,
      "loss": 0.7728,
      "step": 5416
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.545350193977356,
      "learning_rate": 1.8739162685006137e-05,
      "loss": 0.764,
      "step": 5417
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4843940734863281,
      "learning_rate": 1.8738619890149414e-05,
      "loss": 0.7796,
      "step": 5418
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9506267309188843,
      "learning_rate": 1.8738076986345032e-05,
      "loss": 0.7866,
      "step": 5419
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8219751119613647,
      "learning_rate": 1.873753397359977e-05,
      "loss": 1.0113,
      "step": 5420
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7576006650924683,
      "learning_rate": 1.8736990851920388e-05,
      "loss": 0.8032,
      "step": 5421
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5415985584259033,
      "learning_rate": 1.8736447621313664e-05,
      "loss": 0.685,
      "step": 5422
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6644320487976074,
      "learning_rate": 1.8735904281786366e-05,
      "loss": 0.6272,
      "step": 5423
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4960016012191772,
      "learning_rate": 1.8735360833345272e-05,
      "loss": 0.5924,
      "step": 5424
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.49301016330719,
      "learning_rate": 1.8734817275997153e-05,
      "loss": 0.7428,
      "step": 5425
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5165762901306152,
      "learning_rate": 1.8734273609748794e-05,
      "loss": 0.5986,
      "step": 5426
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8958191871643066,
      "learning_rate": 1.8733729834606963e-05,
      "loss": 0.6505,
      "step": 5427
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.722542405128479,
      "learning_rate": 1.8733185950578447e-05,
      "loss": 0.7614,
      "step": 5428
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7371412515640259,
      "learning_rate": 1.8732641957670022e-05,
      "loss": 0.7719,
      "step": 5429
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7356334924697876,
      "learning_rate": 1.873209785588847e-05,
      "loss": 0.7121,
      "step": 5430
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.284921407699585,
      "learning_rate": 1.8731553645240582e-05,
      "loss": 0.6773,
      "step": 5431
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5261943340301514,
      "learning_rate": 1.8731009325733132e-05,
      "loss": 0.7241,
      "step": 5432
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.090831995010376,
      "learning_rate": 1.8730464897372916e-05,
      "loss": 0.8805,
      "step": 5433
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.714728593826294,
      "learning_rate": 1.8729920360166717e-05,
      "loss": 0.8927,
      "step": 5434
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.243153691291809,
      "learning_rate": 1.8729375714121323e-05,
      "loss": 0.5943,
      "step": 5435
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7165449857711792,
      "learning_rate": 1.8728830959243523e-05,
      "loss": 0.8428,
      "step": 5436
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9919826984405518,
      "learning_rate": 1.8728286095540116e-05,
      "loss": 0.7252,
      "step": 5437
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.0371131896972656,
      "learning_rate": 1.8727741123017887e-05,
      "loss": 0.7408,
      "step": 5438
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5455563068389893,
      "learning_rate": 1.8727196041683635e-05,
      "loss": 0.7037,
      "step": 5439
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.516432285308838,
      "learning_rate": 1.872665085154415e-05,
      "loss": 0.6544,
      "step": 5440
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4793189764022827,
      "learning_rate": 1.8726105552606243e-05,
      "loss": 0.7525,
      "step": 5441
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4812195301055908,
      "learning_rate": 1.8725560144876695e-05,
      "loss": 0.748,
      "step": 5442
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3238521814346313,
      "learning_rate": 1.8725014628362315e-05,
      "loss": 0.7637,
      "step": 5443
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.519153118133545,
      "learning_rate": 1.8724469003069903e-05,
      "loss": 0.7801,
      "step": 5444
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.637115240097046,
      "learning_rate": 1.8723923269006265e-05,
      "loss": 0.5677,
      "step": 5445
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.889175534248352,
      "learning_rate": 1.8723377426178195e-05,
      "loss": 0.9645,
      "step": 5446
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3220053911209106,
      "learning_rate": 1.8722831474592507e-05,
      "loss": 0.6968,
      "step": 5447
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8856885433197021,
      "learning_rate": 1.8722285414256006e-05,
      "loss": 0.9938,
      "step": 5448
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5373268127441406,
      "learning_rate": 1.87217392451755e-05,
      "loss": 0.666,
      "step": 5449
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.642713189125061,
      "learning_rate": 1.8721192967357794e-05,
      "loss": 0.7516,
      "step": 5450
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.543929934501648,
      "learning_rate": 1.8720646580809703e-05,
      "loss": 0.912,
      "step": 5451
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6272698640823364,
      "learning_rate": 1.8720100085538036e-05,
      "loss": 0.8154,
      "step": 5452
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5765563249588013,
      "learning_rate": 1.8719553481549612e-05,
      "loss": 0.671,
      "step": 5453
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.786473274230957,
      "learning_rate": 1.871900676885124e-05,
      "loss": 0.677,
      "step": 5454
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6305956840515137,
      "learning_rate": 1.871845994744974e-05,
      "loss": 0.809,
      "step": 5455
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6722239255905151,
      "learning_rate": 1.8717913017351925e-05,
      "loss": 0.8883,
      "step": 5456
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.407930850982666,
      "learning_rate": 1.8717365978564617e-05,
      "loss": 0.809,
      "step": 5457
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6682406663894653,
      "learning_rate": 1.8716818831094636e-05,
      "loss": 0.7926,
      "step": 5458
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5451751947402954,
      "learning_rate": 1.8716271574948803e-05,
      "loss": 0.6683,
      "step": 5459
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.502082347869873,
      "learning_rate": 1.871572421013394e-05,
      "loss": 0.6428,
      "step": 5460
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.569222092628479,
      "learning_rate": 1.871517673665687e-05,
      "loss": 0.8265,
      "step": 5461
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4151155948638916,
      "learning_rate": 1.8714629154524425e-05,
      "loss": 0.7432,
      "step": 5462
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7738860845565796,
      "learning_rate": 1.871408146374343e-05,
      "loss": 0.7994,
      "step": 5463
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4852372407913208,
      "learning_rate": 1.8713533664320703e-05,
      "loss": 0.711,
      "step": 5464
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.358262538909912,
      "learning_rate": 1.8712985756263085e-05,
      "loss": 0.8028,
      "step": 5465
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5698792934417725,
      "learning_rate": 1.87124377395774e-05,
      "loss": 0.7804,
      "step": 5466
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5676532983779907,
      "learning_rate": 1.8711889614270488e-05,
      "loss": 0.5695,
      "step": 5467
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5345017910003662,
      "learning_rate": 1.8711341380349175e-05,
      "loss": 0.7267,
      "step": 5468
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5858490467071533,
      "learning_rate": 1.8710793037820304e-05,
      "loss": 0.738,
      "step": 5469
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.879705786705017,
      "learning_rate": 1.8710244586690703e-05,
      "loss": 0.7176,
      "step": 5470
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3191176652908325,
      "learning_rate": 1.8709696026967212e-05,
      "loss": 0.8112,
      "step": 5471
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.802870273590088,
      "learning_rate": 1.8709147358656673e-05,
      "loss": 0.7728,
      "step": 5472
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3541996479034424,
      "learning_rate": 1.8708598581765925e-05,
      "loss": 0.5418,
      "step": 5473
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9134852886199951,
      "learning_rate": 1.870804969630181e-05,
      "loss": 0.8293,
      "step": 5474
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.700335144996643,
      "learning_rate": 1.8707500702271174e-05,
      "loss": 0.8983,
      "step": 5475
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3904392719268799,
      "learning_rate": 1.8706951599680856e-05,
      "loss": 0.739,
      "step": 5476
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.656914472579956,
      "learning_rate": 1.8706402388537702e-05,
      "loss": 0.8755,
      "step": 5477
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6725140810012817,
      "learning_rate": 1.8705853068848564e-05,
      "loss": 0.6706,
      "step": 5478
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5645731687545776,
      "learning_rate": 1.8705303640620288e-05,
      "loss": 0.6366,
      "step": 5479
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5184776782989502,
      "learning_rate": 1.8704754103859722e-05,
      "loss": 0.7809,
      "step": 5480
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4626610279083252,
      "learning_rate": 1.870420445857372e-05,
      "loss": 0.8096,
      "step": 5481
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4489120244979858,
      "learning_rate": 1.8703654704769138e-05,
      "loss": 0.6887,
      "step": 5482
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3304589986801147,
      "learning_rate": 1.8703104842452825e-05,
      "loss": 0.6669,
      "step": 5483
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8291693925857544,
      "learning_rate": 1.8702554871631635e-05,
      "loss": 0.6952,
      "step": 5484
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3300268650054932,
      "learning_rate": 1.870200479231243e-05,
      "loss": 0.5654,
      "step": 5485
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.904367446899414,
      "learning_rate": 1.870145460450206e-05,
      "loss": 0.6633,
      "step": 5486
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5738307237625122,
      "learning_rate": 1.8700904308207395e-05,
      "loss": 0.6915,
      "step": 5487
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5771923065185547,
      "learning_rate": 1.8700353903435285e-05,
      "loss": 0.8461,
      "step": 5488
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5603272914886475,
      "learning_rate": 1.86998033901926e-05,
      "loss": 0.6066,
      "step": 5489
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3335708379745483,
      "learning_rate": 1.86992527684862e-05,
      "loss": 0.7476,
      "step": 5490
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9992783069610596,
      "learning_rate": 1.8698702038322952e-05,
      "loss": 0.749,
      "step": 5491
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4844213724136353,
      "learning_rate": 1.869815119970972e-05,
      "loss": 0.7986,
      "step": 5492
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7640151977539062,
      "learning_rate": 1.8697600252653372e-05,
      "loss": 0.797,
      "step": 5493
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6216986179351807,
      "learning_rate": 1.869704919716078e-05,
      "loss": 0.8131,
      "step": 5494
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.517155647277832,
      "learning_rate": 1.8696498033238807e-05,
      "loss": 0.6813,
      "step": 5495
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5037355422973633,
      "learning_rate": 1.8695946760894332e-05,
      "loss": 0.7868,
      "step": 5496
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8956594467163086,
      "learning_rate": 1.8695395380134225e-05,
      "loss": 0.6607,
      "step": 5497
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6476969718933105,
      "learning_rate": 1.8694843890965355e-05,
      "loss": 0.985,
      "step": 5498
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4107441902160645,
      "learning_rate": 1.8694292293394608e-05,
      "loss": 0.6116,
      "step": 5499
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3184829950332642,
      "learning_rate": 1.8693740587428855e-05,
      "loss": 0.7111,
      "step": 5500
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7129411697387695,
      "learning_rate": 1.8693188773074973e-05,
      "loss": 0.6918,
      "step": 5501
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6922792196273804,
      "learning_rate": 1.8692636850339845e-05,
      "loss": 0.7707,
      "step": 5502
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.830264687538147,
      "learning_rate": 1.869208481923035e-05,
      "loss": 0.7826,
      "step": 5503
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8597416877746582,
      "learning_rate": 1.869153267975337e-05,
      "loss": 0.7512,
      "step": 5504
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8334155082702637,
      "learning_rate": 1.8690980431915794e-05,
      "loss": 0.944,
      "step": 5505
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.904406189918518,
      "learning_rate": 1.8690428075724498e-05,
      "loss": 0.9219,
      "step": 5506
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.843483805656433,
      "learning_rate": 1.8689875611186378e-05,
      "loss": 0.7007,
      "step": 5507
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5735135078430176,
      "learning_rate": 1.8689323038308312e-05,
      "loss": 0.7585,
      "step": 5508
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5981639623641968,
      "learning_rate": 1.86887703570972e-05,
      "loss": 0.8117,
      "step": 5509
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3628910779953003,
      "learning_rate": 1.8688217567559923e-05,
      "loss": 0.8106,
      "step": 5510
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.99089515209198,
      "learning_rate": 1.8687664669703376e-05,
      "loss": 0.7089,
      "step": 5511
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3417307138442993,
      "learning_rate": 1.8687111663534456e-05,
      "loss": 0.7818,
      "step": 5512
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6597163677215576,
      "learning_rate": 1.868655854906005e-05,
      "loss": 0.7483,
      "step": 5513
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5948227643966675,
      "learning_rate": 1.8686005326287063e-05,
      "loss": 0.7197,
      "step": 5514
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3682345151901245,
      "learning_rate": 1.8685451995222386e-05,
      "loss": 0.7677,
      "step": 5515
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6624566316604614,
      "learning_rate": 1.8684898555872916e-05,
      "loss": 0.7859,
      "step": 5516
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5136549472808838,
      "learning_rate": 1.868434500824556e-05,
      "loss": 0.8144,
      "step": 5517
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4963065385818481,
      "learning_rate": 1.868379135234721e-05,
      "loss": 0.8608,
      "step": 5518
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3946272134780884,
      "learning_rate": 1.868323758818478e-05,
      "loss": 0.7432,
      "step": 5519
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3177695274353027,
      "learning_rate": 1.8682683715765165e-05,
      "loss": 0.621,
      "step": 5520
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9800872802734375,
      "learning_rate": 1.868212973509527e-05,
      "loss": 0.7852,
      "step": 5521
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3725725412368774,
      "learning_rate": 1.8681575646182007e-05,
      "loss": 0.7709,
      "step": 5522
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5286749601364136,
      "learning_rate": 1.8681021449032288e-05,
      "loss": 0.6857,
      "step": 5523
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0020134449005127,
      "learning_rate": 1.8680467143653008e-05,
      "loss": 0.8539,
      "step": 5524
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6313819885253906,
      "learning_rate": 1.8679912730051087e-05,
      "loss": 0.864,
      "step": 5525
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.596635341644287,
      "learning_rate": 1.8679358208233443e-05,
      "loss": 0.8477,
      "step": 5526
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.1182315349578857,
      "learning_rate": 1.8678803578206977e-05,
      "loss": 0.9128,
      "step": 5527
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9418420791625977,
      "learning_rate": 1.867824883997861e-05,
      "loss": 0.806,
      "step": 5528
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7970398664474487,
      "learning_rate": 1.8677693993555258e-05,
      "loss": 0.7709,
      "step": 5529
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5358411073684692,
      "learning_rate": 1.8677139038943835e-05,
      "loss": 0.7149,
      "step": 5530
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3770952224731445,
      "learning_rate": 1.867658397615127e-05,
      "loss": 0.6815,
      "step": 5531
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5741026401519775,
      "learning_rate": 1.867602880518447e-05,
      "loss": 0.6244,
      "step": 5532
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4199109077453613,
      "learning_rate": 1.867547352605036e-05,
      "loss": 0.6536,
      "step": 5533
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3684136867523193,
      "learning_rate": 1.867491813875587e-05,
      "loss": 0.6139,
      "step": 5534
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6006078720092773,
      "learning_rate": 1.8674362643307918e-05,
      "loss": 0.8523,
      "step": 5535
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.803787112236023,
      "learning_rate": 1.8673807039713434e-05,
      "loss": 0.8562,
      "step": 5536
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4647204875946045,
      "learning_rate": 1.867325132797934e-05,
      "loss": 0.618,
      "step": 5537
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.749781608581543,
      "learning_rate": 1.8672695508112566e-05,
      "loss": 0.7467,
      "step": 5538
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4184541702270508,
      "learning_rate": 1.8672139580120042e-05,
      "loss": 0.6023,
      "step": 5539
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.590425968170166,
      "learning_rate": 1.86715835440087e-05,
      "loss": 0.7516,
      "step": 5540
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.487403392791748,
      "learning_rate": 1.867102739978547e-05,
      "loss": 0.7828,
      "step": 5541
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9012740850448608,
      "learning_rate": 1.8670471147457287e-05,
      "loss": 0.7627,
      "step": 5542
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4762240648269653,
      "learning_rate": 1.8669914787031085e-05,
      "loss": 0.7773,
      "step": 5543
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6534019708633423,
      "learning_rate": 1.86693583185138e-05,
      "loss": 0.6386,
      "step": 5544
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6445080041885376,
      "learning_rate": 1.8668801741912373e-05,
      "loss": 0.685,
      "step": 5545
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5096800327301025,
      "learning_rate": 1.8668245057233744e-05,
      "loss": 0.8209,
      "step": 5546
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.335362195968628,
      "learning_rate": 1.8667688264484843e-05,
      "loss": 0.6943,
      "step": 5547
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4978116750717163,
      "learning_rate": 1.8667131363672625e-05,
      "loss": 0.7566,
      "step": 5548
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5785712003707886,
      "learning_rate": 1.8666574354804027e-05,
      "loss": 0.8132,
      "step": 5549
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3345658779144287,
      "learning_rate": 1.866601723788599e-05,
      "loss": 0.663,
      "step": 5550
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5315746068954468,
      "learning_rate": 1.866546001292547e-05,
      "loss": 0.6184,
      "step": 5551
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.66499662399292,
      "learning_rate": 1.86649026799294e-05,
      "loss": 0.5864,
      "step": 5552
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3839362859725952,
      "learning_rate": 1.8664345238904738e-05,
      "loss": 0.7654,
      "step": 5553
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1954489946365356,
      "learning_rate": 1.8663787689858435e-05,
      "loss": 0.6883,
      "step": 5554
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8301403522491455,
      "learning_rate": 1.8663230032797437e-05,
      "loss": 0.7641,
      "step": 5555
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5820236206054688,
      "learning_rate": 1.86626722677287e-05,
      "loss": 0.6095,
      "step": 5556
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6174442768096924,
      "learning_rate": 1.8662114394659173e-05,
      "loss": 0.7775,
      "step": 5557
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7151628732681274,
      "learning_rate": 1.8661556413595818e-05,
      "loss": 0.7083,
      "step": 5558
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7497806549072266,
      "learning_rate": 1.8660998324545584e-05,
      "loss": 0.6916,
      "step": 5559
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5912576913833618,
      "learning_rate": 1.8660440127515438e-05,
      "loss": 0.746,
      "step": 5560
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.387764811515808,
      "learning_rate": 1.865988182251233e-05,
      "loss": 0.656,
      "step": 5561
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.135066509246826,
      "learning_rate": 1.865932340954323e-05,
      "loss": 0.7923,
      "step": 5562
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.970717191696167,
      "learning_rate": 1.8658764888615092e-05,
      "loss": 0.8816,
      "step": 5563
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7622851133346558,
      "learning_rate": 1.865820625973488e-05,
      "loss": 0.667,
      "step": 5564
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6850602626800537,
      "learning_rate": 1.8657647522909567e-05,
      "loss": 0.8059,
      "step": 5565
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6633905172348022,
      "learning_rate": 1.8657088678146106e-05,
      "loss": 0.5746,
      "step": 5566
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.390453815460205,
      "learning_rate": 1.8656529725451474e-05,
      "loss": 0.6838,
      "step": 5567
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1060097217559814,
      "learning_rate": 1.865597066483264e-05,
      "loss": 0.7723,
      "step": 5568
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6573066711425781,
      "learning_rate": 1.8655411496296568e-05,
      "loss": 0.8022,
      "step": 5569
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6597685813903809,
      "learning_rate": 1.865485221985023e-05,
      "loss": 0.8222,
      "step": 5570
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6347815990447998,
      "learning_rate": 1.8654292835500604e-05,
      "loss": 0.7838,
      "step": 5571
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7681483030319214,
      "learning_rate": 1.865373334325466e-05,
      "loss": 0.7288,
      "step": 5572
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6072691679000854,
      "learning_rate": 1.865317374311937e-05,
      "loss": 0.7562,
      "step": 5573
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9352208375930786,
      "learning_rate": 1.8652614035101723e-05,
      "loss": 0.7647,
      "step": 5574
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.536231279373169,
      "learning_rate": 1.8652054219208686e-05,
      "loss": 0.7674,
      "step": 5575
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4059288501739502,
      "learning_rate": 1.865149429544724e-05,
      "loss": 0.7574,
      "step": 5576
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5618754625320435,
      "learning_rate": 1.865093426382437e-05,
      "loss": 0.7724,
      "step": 5577
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.371518611907959,
      "learning_rate": 1.865037412434705e-05,
      "loss": 0.5856,
      "step": 5578
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5786443948745728,
      "learning_rate": 1.8649813877022275e-05,
      "loss": 0.6078,
      "step": 5579
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5054888725280762,
      "learning_rate": 1.8649253521857022e-05,
      "loss": 0.6562,
      "step": 5580
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.870177984237671,
      "learning_rate": 1.864869305885828e-05,
      "loss": 0.7418,
      "step": 5581
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4963403940200806,
      "learning_rate": 1.864813248803303e-05,
      "loss": 0.6853,
      "step": 5582
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.503575325012207,
      "learning_rate": 1.8647571809388277e-05,
      "loss": 0.825,
      "step": 5583
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5547711849212646,
      "learning_rate": 1.864701102293099e-05,
      "loss": 0.9091,
      "step": 5584
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.622564673423767,
      "learning_rate": 1.8646450128668176e-05,
      "loss": 0.7104,
      "step": 5585
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6212620735168457,
      "learning_rate": 1.864588912660682e-05,
      "loss": 0.7679,
      "step": 5586
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.669983983039856,
      "learning_rate": 1.8645328016753925e-05,
      "loss": 0.7043,
      "step": 5587
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6216410398483276,
      "learning_rate": 1.8644766799116475e-05,
      "loss": 0.6458,
      "step": 5588
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6369916200637817,
      "learning_rate": 1.8644205473701475e-05,
      "loss": 0.7758,
      "step": 5589
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4355727434158325,
      "learning_rate": 1.864364404051592e-05,
      "loss": 0.7258,
      "step": 5590
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.31391179561615,
      "learning_rate": 1.864308249956681e-05,
      "loss": 0.7314,
      "step": 5591
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8239567279815674,
      "learning_rate": 1.8642520850861148e-05,
      "loss": 0.7442,
      "step": 5592
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5175269842147827,
      "learning_rate": 1.8641959094405935e-05,
      "loss": 0.6433,
      "step": 5593
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9655017852783203,
      "learning_rate": 1.8641397230208172e-05,
      "loss": 0.7838,
      "step": 5594
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8395071029663086,
      "learning_rate": 1.864083525827487e-05,
      "loss": 0.8611,
      "step": 5595
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4284123182296753,
      "learning_rate": 1.864027317861303e-05,
      "loss": 0.6377,
      "step": 5596
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5110316276550293,
      "learning_rate": 1.863971099122966e-05,
      "loss": 0.939,
      "step": 5597
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7826054096221924,
      "learning_rate": 1.8639148696131768e-05,
      "loss": 0.9095,
      "step": 5598
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6445462703704834,
      "learning_rate": 1.863858629332637e-05,
      "loss": 0.7089,
      "step": 5599
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.9580302238464355,
      "learning_rate": 1.8638023782820474e-05,
      "loss": 0.8776,
      "step": 5600
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7605888843536377,
      "learning_rate": 1.8637461164621093e-05,
      "loss": 0.7004,
      "step": 5601
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4798638820648193,
      "learning_rate": 1.863689843873524e-05,
      "loss": 0.7314,
      "step": 5602
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6653226613998413,
      "learning_rate": 1.8636335605169935e-05,
      "loss": 0.7886,
      "step": 5603
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4955835342407227,
      "learning_rate": 1.8635772663932192e-05,
      "loss": 0.9018,
      "step": 5604
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.53203284740448,
      "learning_rate": 1.863520961502903e-05,
      "loss": 0.6751,
      "step": 5605
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.608373761177063,
      "learning_rate": 1.8634646458467467e-05,
      "loss": 0.5644,
      "step": 5606
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5978021621704102,
      "learning_rate": 1.8634083194254526e-05,
      "loss": 0.7572,
      "step": 5607
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5302214622497559,
      "learning_rate": 1.8633519822397234e-05,
      "loss": 0.9105,
      "step": 5608
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5106390714645386,
      "learning_rate": 1.8632956342902605e-05,
      "loss": 0.7238,
      "step": 5609
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.332565426826477,
      "learning_rate": 1.8632392755777666e-05,
      "loss": 0.7158,
      "step": 5610
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4217909574508667,
      "learning_rate": 1.863182906102945e-05,
      "loss": 0.7832,
      "step": 5611
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.601300835609436,
      "learning_rate": 1.8631265258664983e-05,
      "loss": 0.8001,
      "step": 5612
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.0781097412109375,
      "learning_rate": 1.8630701348691295e-05,
      "loss": 0.8493,
      "step": 5613
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.797236442565918,
      "learning_rate": 1.8630137331115405e-05,
      "loss": 0.7295,
      "step": 5614
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8642655611038208,
      "learning_rate": 1.862957320594436e-05,
      "loss": 0.8631,
      "step": 5615
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.47858726978302,
      "learning_rate": 1.8629008973185185e-05,
      "loss": 0.6695,
      "step": 5616
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.622968316078186,
      "learning_rate": 1.8628444632844918e-05,
      "loss": 0.7059,
      "step": 5617
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6726163625717163,
      "learning_rate": 1.8627880184930593e-05,
      "loss": 0.876,
      "step": 5618
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.570600152015686,
      "learning_rate": 1.8627315629449243e-05,
      "loss": 0.7792,
      "step": 5619
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6045278310775757,
      "learning_rate": 1.8626750966407915e-05,
      "loss": 0.7458,
      "step": 5620
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4651695489883423,
      "learning_rate": 1.8626186195813644e-05,
      "loss": 0.7763,
      "step": 5621
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4605218172073364,
      "learning_rate": 1.8625621317673474e-05,
      "loss": 0.8007,
      "step": 5622
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7414920330047607,
      "learning_rate": 1.8625056331994444e-05,
      "loss": 0.6044,
      "step": 5623
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5452539920806885,
      "learning_rate": 1.8624491238783595e-05,
      "loss": 0.8321,
      "step": 5624
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.674017310142517,
      "learning_rate": 1.8623926038047982e-05,
      "loss": 0.8564,
      "step": 5625
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.474761962890625,
      "learning_rate": 1.8623360729794645e-05,
      "loss": 0.6965,
      "step": 5626
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4875367879867554,
      "learning_rate": 1.862279531403063e-05,
      "loss": 0.8311,
      "step": 5627
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.863823413848877,
      "learning_rate": 1.8622229790762992e-05,
      "loss": 0.7243,
      "step": 5628
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5922497510910034,
      "learning_rate": 1.8621664159998784e-05,
      "loss": 0.7412,
      "step": 5629
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.644352674484253,
      "learning_rate": 1.8621098421745048e-05,
      "loss": 0.6588,
      "step": 5630
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3487533330917358,
      "learning_rate": 1.862053257600884e-05,
      "loss": 0.7167,
      "step": 5631
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.007267713546753,
      "learning_rate": 1.861996662279722e-05,
      "loss": 0.8418,
      "step": 5632
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8126670122146606,
      "learning_rate": 1.861940056211724e-05,
      "loss": 0.7513,
      "step": 5633
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.1591386795043945,
      "learning_rate": 1.8618834393975957e-05,
      "loss": 0.7247,
      "step": 5634
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6051843166351318,
      "learning_rate": 1.861826811838043e-05,
      "loss": 0.7536,
      "step": 5635
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.427047848701477,
      "learning_rate": 1.8617701735337724e-05,
      "loss": 0.8325,
      "step": 5636
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6415154933929443,
      "learning_rate": 1.861713524485489e-05,
      "loss": 0.6494,
      "step": 5637
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4643818140029907,
      "learning_rate": 1.8616568646939e-05,
      "loss": 0.6195,
      "step": 5638
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5019257068634033,
      "learning_rate": 1.8616001941597114e-05,
      "loss": 0.8935,
      "step": 5639
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.388340711593628,
      "learning_rate": 1.86154351288363e-05,
      "loss": 0.6628,
      "step": 5640
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.593717336654663,
      "learning_rate": 1.8614868208663618e-05,
      "loss": 0.8511,
      "step": 5641
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4299412965774536,
      "learning_rate": 1.8614301181086146e-05,
      "loss": 0.7323,
      "step": 5642
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8067866563796997,
      "learning_rate": 1.8613734046110946e-05,
      "loss": 0.9077,
      "step": 5643
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5521671772003174,
      "learning_rate": 1.8613166803745088e-05,
      "loss": 0.7137,
      "step": 5644
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5236923694610596,
      "learning_rate": 1.861259945399565e-05,
      "loss": 0.6799,
      "step": 5645
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5473625659942627,
      "learning_rate": 1.8612031996869696e-05,
      "loss": 0.8998,
      "step": 5646
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6557872295379639,
      "learning_rate": 1.861146443237431e-05,
      "loss": 0.8511,
      "step": 5647
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2842357158660889,
      "learning_rate": 1.861089676051657e-05,
      "loss": 0.8427,
      "step": 5648
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7137624025344849,
      "learning_rate": 1.861032898130354e-05,
      "loss": 0.8245,
      "step": 5649
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6625123023986816,
      "learning_rate": 1.860976109474231e-05,
      "loss": 0.6646,
      "step": 5650
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6733688116073608,
      "learning_rate": 1.860919310083996e-05,
      "loss": 0.8516,
      "step": 5651
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.352352261543274,
      "learning_rate": 1.8608624999603567e-05,
      "loss": 0.6538,
      "step": 5652
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4855782985687256,
      "learning_rate": 1.8608056791040212e-05,
      "loss": 0.9113,
      "step": 5653
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5409235954284668,
      "learning_rate": 1.8607488475156984e-05,
      "loss": 0.6849,
      "step": 5654
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5603195428848267,
      "learning_rate": 1.8606920051960966e-05,
      "loss": 0.7186,
      "step": 5655
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.334377408027649,
      "learning_rate": 1.8606351521459247e-05,
      "loss": 0.7055,
      "step": 5656
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7181050777435303,
      "learning_rate": 1.860578288365891e-05,
      "loss": 0.8394,
      "step": 5657
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4276319742202759,
      "learning_rate": 1.860521413856705e-05,
      "loss": 0.5753,
      "step": 5658
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.780235767364502,
      "learning_rate": 1.8604645286190757e-05,
      "loss": 0.8425,
      "step": 5659
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6416094303131104,
      "learning_rate": 1.8604076326537118e-05,
      "loss": 0.7468,
      "step": 5660
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3533999919891357,
      "learning_rate": 1.860350725961323e-05,
      "loss": 0.7334,
      "step": 5661
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7039517164230347,
      "learning_rate": 1.860293808542619e-05,
      "loss": 0.8336,
      "step": 5662
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2694370746612549,
      "learning_rate": 1.8602368803983093e-05,
      "loss": 0.661,
      "step": 5663
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3071601390838623,
      "learning_rate": 1.8601799415291034e-05,
      "loss": 0.674,
      "step": 5664
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4347728490829468,
      "learning_rate": 1.860122991935711e-05,
      "loss": 0.7748,
      "step": 5665
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3261247873306274,
      "learning_rate": 1.860066031618843e-05,
      "loss": 0.6939,
      "step": 5666
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5991299152374268,
      "learning_rate": 1.8600090605792086e-05,
      "loss": 0.8178,
      "step": 5667
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4065883159637451,
      "learning_rate": 1.8599520788175184e-05,
      "loss": 0.6914,
      "step": 5668
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.827972412109375,
      "learning_rate": 1.8598950863344832e-05,
      "loss": 0.7085,
      "step": 5669
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5349602699279785,
      "learning_rate": 1.859838083130813e-05,
      "loss": 0.6838,
      "step": 5670
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8168182373046875,
      "learning_rate": 1.8597810692072187e-05,
      "loss": 0.7297,
      "step": 5671
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7269423007965088,
      "learning_rate": 1.859724044564411e-05,
      "loss": 0.8828,
      "step": 5672
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7241053581237793,
      "learning_rate": 1.859667009203101e-05,
      "loss": 0.7251,
      "step": 5673
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5929951667785645,
      "learning_rate": 1.8596099631239998e-05,
      "loss": 0.7806,
      "step": 5674
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3221385478973389,
      "learning_rate": 1.8595529063278184e-05,
      "loss": 0.6013,
      "step": 5675
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3653626441955566,
      "learning_rate": 1.8594958388152685e-05,
      "loss": 0.6799,
      "step": 5676
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4984790086746216,
      "learning_rate": 1.8594387605870613e-05,
      "loss": 0.7052,
      "step": 5677
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.795450210571289,
      "learning_rate": 1.8593816716439083e-05,
      "loss": 0.8986,
      "step": 5678
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2391482591629028,
      "learning_rate": 1.8593245719865218e-05,
      "loss": 0.6503,
      "step": 5679
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5439447164535522,
      "learning_rate": 1.859267461615613e-05,
      "loss": 0.7584,
      "step": 5680
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5929546356201172,
      "learning_rate": 1.8592103405318946e-05,
      "loss": 0.712,
      "step": 5681
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6405097246170044,
      "learning_rate": 1.859153208736078e-05,
      "loss": 0.7309,
      "step": 5682
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4983609914779663,
      "learning_rate": 1.859096066228876e-05,
      "loss": 0.8365,
      "step": 5683
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3429534435272217,
      "learning_rate": 1.859038913011001e-05,
      "loss": 0.6552,
      "step": 5684
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4316942691802979,
      "learning_rate": 1.8589817490831655e-05,
      "loss": 0.7698,
      "step": 5685
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5825247764587402,
      "learning_rate": 1.858924574446082e-05,
      "loss": 0.7869,
      "step": 5686
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5990355014801025,
      "learning_rate": 1.8588673891004633e-05,
      "loss": 0.9778,
      "step": 5687
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8632758855819702,
      "learning_rate": 1.8588101930470224e-05,
      "loss": 0.6472,
      "step": 5688
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4990154504776,
      "learning_rate": 1.8587529862864732e-05,
      "loss": 0.8382,
      "step": 5689
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4349567890167236,
      "learning_rate": 1.8586957688195278e-05,
      "loss": 0.7437,
      "step": 5690
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4321670532226562,
      "learning_rate": 1.8586385406468998e-05,
      "loss": 0.67,
      "step": 5691
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6568973064422607,
      "learning_rate": 1.8585813017693027e-05,
      "loss": 0.8016,
      "step": 5692
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.059561014175415,
      "learning_rate": 1.8585240521874504e-05,
      "loss": 0.8469,
      "step": 5693
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4120776653289795,
      "learning_rate": 1.8584667919020568e-05,
      "loss": 0.6836,
      "step": 5694
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4005323648452759,
      "learning_rate": 1.8584095209138354e-05,
      "loss": 0.6908,
      "step": 5695
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5396414995193481,
      "learning_rate": 1.8583522392234997e-05,
      "loss": 0.7639,
      "step": 5696
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5593929290771484,
      "learning_rate": 1.8582949468317654e-05,
      "loss": 0.6613,
      "step": 5697
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4544223546981812,
      "learning_rate": 1.8582376437393453e-05,
      "loss": 0.7703,
      "step": 5698
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4486724138259888,
      "learning_rate": 1.8581803299469542e-05,
      "loss": 0.7261,
      "step": 5699
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7281025648117065,
      "learning_rate": 1.8581230054553072e-05,
      "loss": 0.7957,
      "step": 5700
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.521946668624878,
      "learning_rate": 1.8580656702651187e-05,
      "loss": 0.8694,
      "step": 5701
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1582789421081543,
      "learning_rate": 1.858008324377103e-05,
      "loss": 0.8827,
      "step": 5702
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9572468996047974,
      "learning_rate": 1.857950967791976e-05,
      "loss": 0.9615,
      "step": 5703
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3723697662353516,
      "learning_rate": 1.8578936005104516e-05,
      "loss": 0.7555,
      "step": 5704
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7633295059204102,
      "learning_rate": 1.857836222533246e-05,
      "loss": 0.787,
      "step": 5705
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5598807334899902,
      "learning_rate": 1.8577788338610746e-05,
      "loss": 0.796,
      "step": 5706
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.475573182106018,
      "learning_rate": 1.857721434494652e-05,
      "loss": 0.7395,
      "step": 5707
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4494831562042236,
      "learning_rate": 1.8576640244346946e-05,
      "loss": 0.7368,
      "step": 5708
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6156775951385498,
      "learning_rate": 1.8576066036819177e-05,
      "loss": 0.7172,
      "step": 5709
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7993457317352295,
      "learning_rate": 1.857549172237038e-05,
      "loss": 0.8985,
      "step": 5710
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7961174249649048,
      "learning_rate": 1.8574917301007704e-05,
      "loss": 0.8398,
      "step": 5711
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1498616933822632,
      "learning_rate": 1.8574342772738313e-05,
      "loss": 0.7393,
      "step": 5712
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3300855159759521,
      "learning_rate": 1.8573768137569376e-05,
      "loss": 0.838,
      "step": 5713
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5997436046600342,
      "learning_rate": 1.857319339550805e-05,
      "loss": 0.6441,
      "step": 5714
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.672188401222229,
      "learning_rate": 1.8572618546561506e-05,
      "loss": 0.7479,
      "step": 5715
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4633108377456665,
      "learning_rate": 1.8572043590736914e-05,
      "loss": 0.684,
      "step": 5716
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6210943460464478,
      "learning_rate": 1.8571468528041433e-05,
      "loss": 0.8193,
      "step": 5717
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5499175786972046,
      "learning_rate": 1.8570893358482232e-05,
      "loss": 0.5408,
      "step": 5718
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6459578275680542,
      "learning_rate": 1.8570318082066495e-05,
      "loss": 0.7747,
      "step": 5719
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6906918287277222,
      "learning_rate": 1.856974269880138e-05,
      "loss": 0.8075,
      "step": 5720
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5605295896530151,
      "learning_rate": 1.856916720869407e-05,
      "loss": 0.5761,
      "step": 5721
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5574198961257935,
      "learning_rate": 1.856859161175173e-05,
      "loss": 0.7299,
      "step": 5722
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6406238079071045,
      "learning_rate": 1.856801590798155e-05,
      "loss": 0.7396,
      "step": 5723
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.499437689781189,
      "learning_rate": 1.8567440097390697e-05,
      "loss": 0.7509,
      "step": 5724
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.596709966659546,
      "learning_rate": 1.856686417998635e-05,
      "loss": 0.6293,
      "step": 5725
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.027578353881836,
      "learning_rate": 1.8566288155775694e-05,
      "loss": 0.7614,
      "step": 5726
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4843207597732544,
      "learning_rate": 1.856571202476591e-05,
      "loss": 0.8194,
      "step": 5727
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5541037321090698,
      "learning_rate": 1.8565135786964175e-05,
      "loss": 0.8646,
      "step": 5728
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6611344814300537,
      "learning_rate": 1.856455944237768e-05,
      "loss": 0.6108,
      "step": 5729
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.814605951309204,
      "learning_rate": 1.8563982991013608e-05,
      "loss": 0.8301,
      "step": 5730
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4986200332641602,
      "learning_rate": 1.8563406432879144e-05,
      "loss": 0.8719,
      "step": 5731
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.921535849571228,
      "learning_rate": 1.8562829767981477e-05,
      "loss": 0.8952,
      "step": 5732
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4731886386871338,
      "learning_rate": 1.8562252996327802e-05,
      "loss": 0.6716,
      "step": 5733
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.367231011390686,
      "learning_rate": 1.8561676117925302e-05,
      "loss": 0.6896,
      "step": 5734
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.604764699935913,
      "learning_rate": 1.8561099132781174e-05,
      "loss": 0.6822,
      "step": 5735
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.766297698020935,
      "learning_rate": 1.8560522040902608e-05,
      "loss": 0.9521,
      "step": 5736
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9019523859024048,
      "learning_rate": 1.85599448422968e-05,
      "loss": 0.7872,
      "step": 5737
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5079314708709717,
      "learning_rate": 1.8559367536970947e-05,
      "loss": 0.8205,
      "step": 5738
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.375783920288086,
      "learning_rate": 1.8558790124932246e-05,
      "loss": 0.6938,
      "step": 5739
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4467111825942993,
      "learning_rate": 1.85582126061879e-05,
      "loss": 0.8703,
      "step": 5740
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.641951322555542,
      "learning_rate": 1.8557634980745104e-05,
      "loss": 0.6604,
      "step": 5741
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8862535953521729,
      "learning_rate": 1.8557057248611056e-05,
      "loss": 0.7576,
      "step": 5742
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9979528188705444,
      "learning_rate": 1.8556479409792968e-05,
      "loss": 0.7875,
      "step": 5743
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.717760443687439,
      "learning_rate": 1.855590146429804e-05,
      "loss": 0.7031,
      "step": 5744
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.1879093647003174,
      "learning_rate": 1.8555323412133474e-05,
      "loss": 0.6929,
      "step": 5745
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.447007656097412,
      "learning_rate": 1.855474525330648e-05,
      "loss": 0.6956,
      "step": 5746
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.550410509109497,
      "learning_rate": 1.8554166987824266e-05,
      "loss": 0.638,
      "step": 5747
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.583451271057129,
      "learning_rate": 1.8553588615694045e-05,
      "loss": 0.7094,
      "step": 5748
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.353567361831665,
      "learning_rate": 1.8553010136923022e-05,
      "loss": 0.7496,
      "step": 5749
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3046022653579712,
      "learning_rate": 1.855243155151841e-05,
      "loss": 0.6489,
      "step": 5750
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3754029273986816,
      "learning_rate": 1.8551852859487428e-05,
      "loss": 0.7766,
      "step": 5751
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8779793977737427,
      "learning_rate": 1.8551274060837283e-05,
      "loss": 0.8613,
      "step": 5752
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3649333715438843,
      "learning_rate": 1.8550695155575197e-05,
      "loss": 0.7262,
      "step": 5753
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3798056840896606,
      "learning_rate": 1.855011614370838e-05,
      "loss": 0.82,
      "step": 5754
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.487318754196167,
      "learning_rate": 1.8549537025244062e-05,
      "loss": 0.6288,
      "step": 5755
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.604934811592102,
      "learning_rate": 1.8548957800189454e-05,
      "loss": 0.792,
      "step": 5756
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.539963722229004,
      "learning_rate": 1.854837846855178e-05,
      "loss": 0.6188,
      "step": 5757
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.545517086982727,
      "learning_rate": 1.8547799030338267e-05,
      "loss": 0.8206,
      "step": 5758
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.549003005027771,
      "learning_rate": 1.854721948555613e-05,
      "loss": 0.6519,
      "step": 5759
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.259252667427063,
      "learning_rate": 1.8546639834212602e-05,
      "loss": 0.7192,
      "step": 5760
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.501577138900757,
      "learning_rate": 1.8546060076314906e-05,
      "loss": 0.7117,
      "step": 5761
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3008095026016235,
      "learning_rate": 1.854548021187027e-05,
      "loss": 0.6478,
      "step": 5762
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3201427459716797,
      "learning_rate": 1.8544900240885934e-05,
      "loss": 0.6659,
      "step": 5763
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7863061428070068,
      "learning_rate": 1.854432016336911e-05,
      "loss": 0.7693,
      "step": 5764
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.494314432144165,
      "learning_rate": 1.8543739979327042e-05,
      "loss": 0.6547,
      "step": 5765
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8726556301116943,
      "learning_rate": 1.8543159688766962e-05,
      "loss": 0.757,
      "step": 5766
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.694989800453186,
      "learning_rate": 1.8542579291696104e-05,
      "loss": 0.7688,
      "step": 5767
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5700981616973877,
      "learning_rate": 1.85419987881217e-05,
      "loss": 0.9446,
      "step": 5768
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5837080478668213,
      "learning_rate": 1.8541418178050995e-05,
      "loss": 0.6782,
      "step": 5769
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.429356098175049,
      "learning_rate": 1.8540837461491222e-05,
      "loss": 0.7495,
      "step": 5770
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.648228406906128,
      "learning_rate": 1.8540256638449623e-05,
      "loss": 0.8182,
      "step": 5771
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5471266508102417,
      "learning_rate": 1.8539675708933437e-05,
      "loss": 0.5441,
      "step": 5772
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4338390827178955,
      "learning_rate": 1.853909467294991e-05,
      "loss": 0.7537,
      "step": 5773
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5786463022232056,
      "learning_rate": 1.8538513530506287e-05,
      "loss": 0.6368,
      "step": 5774
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.787619948387146,
      "learning_rate": 1.853793228160981e-05,
      "loss": 0.6712,
      "step": 5775
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.590881109237671,
      "learning_rate": 1.8537350926267726e-05,
      "loss": 0.749,
      "step": 5776
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5020030736923218,
      "learning_rate": 1.853676946448728e-05,
      "loss": 0.6766,
      "step": 5777
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.760705828666687,
      "learning_rate": 1.8536187896275728e-05,
      "loss": 0.7385,
      "step": 5778
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4504305124282837,
      "learning_rate": 1.853560622164032e-05,
      "loss": 0.5043,
      "step": 5779
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5769214630126953,
      "learning_rate": 1.8535024440588304e-05,
      "loss": 0.7372,
      "step": 5780
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6273367404937744,
      "learning_rate": 1.8534442553126936e-05,
      "loss": 0.8391,
      "step": 5781
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6991323232650757,
      "learning_rate": 1.8533860559263464e-05,
      "loss": 0.7689,
      "step": 5782
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7521376609802246,
      "learning_rate": 1.8533278459005152e-05,
      "loss": 0.6291,
      "step": 5783
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4856677055358887,
      "learning_rate": 1.8532696252359255e-05,
      "loss": 0.8456,
      "step": 5784
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3670145273208618,
      "learning_rate": 1.8532113939333034e-05,
      "loss": 0.7382,
      "step": 5785
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.484189748764038,
      "learning_rate": 1.8531531519933745e-05,
      "loss": 0.6267,
      "step": 5786
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.709069013595581,
      "learning_rate": 1.8530948994168645e-05,
      "loss": 0.7149,
      "step": 5787
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5743684768676758,
      "learning_rate": 1.853036636204501e-05,
      "loss": 0.7915,
      "step": 5788
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5957083702087402,
      "learning_rate": 1.852978362357009e-05,
      "loss": 0.766,
      "step": 5789
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6887422800064087,
      "learning_rate": 1.852920077875116e-05,
      "loss": 0.7591,
      "step": 5790
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5262531042099,
      "learning_rate": 1.852861782759548e-05,
      "loss": 0.7181,
      "step": 5791
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4765303134918213,
      "learning_rate": 1.852803477011032e-05,
      "loss": 0.836,
      "step": 5792
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3498996496200562,
      "learning_rate": 1.8527451606302952e-05,
      "loss": 0.774,
      "step": 5793
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4400264024734497,
      "learning_rate": 1.852686833618064e-05,
      "loss": 0.6625,
      "step": 5794
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8031415939331055,
      "learning_rate": 1.8526284959750663e-05,
      "loss": 0.8263,
      "step": 5795
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3689727783203125,
      "learning_rate": 1.852570147702029e-05,
      "loss": 0.7668,
      "step": 5796
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3759300708770752,
      "learning_rate": 1.85251178879968e-05,
      "loss": 0.5394,
      "step": 5797
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5215685367584229,
      "learning_rate": 1.852453419268746e-05,
      "loss": 0.762,
      "step": 5798
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3802547454833984,
      "learning_rate": 1.8523950391099558e-05,
      "loss": 0.7511,
      "step": 5799
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5409144163131714,
      "learning_rate": 1.8523366483240367e-05,
      "loss": 0.7507,
      "step": 5800
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3020076751708984,
      "learning_rate": 1.8522782469117162e-05,
      "loss": 0.7981,
      "step": 5801
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2197948694229126,
      "learning_rate": 1.8522198348737235e-05,
      "loss": 0.7329,
      "step": 5802
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5021599531173706,
      "learning_rate": 1.8521614122107857e-05,
      "loss": 0.7512,
      "step": 5803
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6392651796340942,
      "learning_rate": 1.852102978923632e-05,
      "loss": 0.7165,
      "step": 5804
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.453492283821106,
      "learning_rate": 1.8520445350129908e-05,
      "loss": 0.7363,
      "step": 5805
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3447380065917969,
      "learning_rate": 1.8519860804795903e-05,
      "loss": 0.6537,
      "step": 5806
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3218952417373657,
      "learning_rate": 1.8519276153241598e-05,
      "loss": 0.8314,
      "step": 5807
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3590823411941528,
      "learning_rate": 1.8518691395474274e-05,
      "loss": 0.76,
      "step": 5808
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.656175971031189,
      "learning_rate": 1.8518106531501232e-05,
      "loss": 0.8839,
      "step": 5809
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.773988962173462,
      "learning_rate": 1.851752156132976e-05,
      "loss": 0.7721,
      "step": 5810
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2699661254882812,
      "learning_rate": 1.8516936484967146e-05,
      "loss": 0.7711,
      "step": 5811
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.867224097251892,
      "learning_rate": 1.8516351302420687e-05,
      "loss": 0.6915,
      "step": 5812
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5322059392929077,
      "learning_rate": 1.8515766013697685e-05,
      "loss": 0.6539,
      "step": 5813
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7099575996398926,
      "learning_rate": 1.851518061880543e-05,
      "loss": 0.8765,
      "step": 5814
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6770581007003784,
      "learning_rate": 1.851459511775122e-05,
      "loss": 0.6778,
      "step": 5815
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6691347360610962,
      "learning_rate": 1.851400951054236e-05,
      "loss": 0.7738,
      "step": 5816
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5487239360809326,
      "learning_rate": 1.8513423797186143e-05,
      "loss": 0.875,
      "step": 5817
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6140384674072266,
      "learning_rate": 1.8512837977689878e-05,
      "loss": 0.7832,
      "step": 5818
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4483073949813843,
      "learning_rate": 1.851225205206087e-05,
      "loss": 0.8434,
      "step": 5819
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6540889739990234,
      "learning_rate": 1.8511666020306417e-05,
      "loss": 0.8454,
      "step": 5820
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.472469687461853,
      "learning_rate": 1.851107988243383e-05,
      "loss": 0.6755,
      "step": 5821
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.420080542564392,
      "learning_rate": 1.851049363845042e-05,
      "loss": 0.5969,
      "step": 5822
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.844827651977539,
      "learning_rate": 1.8509907288363486e-05,
      "loss": 0.7485,
      "step": 5823
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0844860076904297,
      "learning_rate": 1.8509320832180344e-05,
      "loss": 0.8218,
      "step": 5824
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.42518150806427,
      "learning_rate": 1.850873426990831e-05,
      "loss": 0.7686,
      "step": 5825
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2179887294769287,
      "learning_rate": 1.8508147601554686e-05,
      "loss": 0.6734,
      "step": 5826
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6659634113311768,
      "learning_rate": 1.8507560827126795e-05,
      "loss": 0.7149,
      "step": 5827
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.586142897605896,
      "learning_rate": 1.850697394663195e-05,
      "loss": 0.7126,
      "step": 5828
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1937527656555176,
      "learning_rate": 1.8506386960077468e-05,
      "loss": 0.6505,
      "step": 5829
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3765778541564941,
      "learning_rate": 1.850579986747067e-05,
      "loss": 0.6704,
      "step": 5830
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6818764209747314,
      "learning_rate": 1.850521266881887e-05,
      "loss": 0.602,
      "step": 5831
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.645098328590393,
      "learning_rate": 1.850462536412939e-05,
      "loss": 0.8181,
      "step": 5832
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5839546918869019,
      "learning_rate": 1.8504037953409554e-05,
      "loss": 0.7048,
      "step": 5833
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.590840220451355,
      "learning_rate": 1.8503450436666687e-05,
      "loss": 0.8091,
      "step": 5834
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4879108667373657,
      "learning_rate": 1.850286281390811e-05,
      "loss": 0.619,
      "step": 5835
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8006963729858398,
      "learning_rate": 1.850227508514115e-05,
      "loss": 0.8012,
      "step": 5836
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6505807638168335,
      "learning_rate": 1.8501687250373134e-05,
      "loss": 0.6278,
      "step": 5837
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7831945419311523,
      "learning_rate": 1.8501099309611398e-05,
      "loss": 0.7303,
      "step": 5838
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5819538831710815,
      "learning_rate": 1.8500511262863262e-05,
      "loss": 0.7344,
      "step": 5839
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4981344938278198,
      "learning_rate": 1.849992311013606e-05,
      "loss": 0.6146,
      "step": 5840
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4385907649993896,
      "learning_rate": 1.849933485143713e-05,
      "loss": 0.7066,
      "step": 5841
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.525299072265625,
      "learning_rate": 1.84987464867738e-05,
      "loss": 0.6915,
      "step": 5842
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3304309844970703,
      "learning_rate": 1.849815801615341e-05,
      "loss": 0.7297,
      "step": 5843
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.1521472930908203,
      "learning_rate": 1.8497569439583292e-05,
      "loss": 0.8282,
      "step": 5844
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4611996412277222,
      "learning_rate": 1.8496980757070785e-05,
      "loss": 0.7518,
      "step": 5845
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4025851488113403,
      "learning_rate": 1.8496391968623233e-05,
      "loss": 0.6459,
      "step": 5846
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4828910827636719,
      "learning_rate": 1.8495803074247972e-05,
      "loss": 0.7116,
      "step": 5847
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7367761135101318,
      "learning_rate": 1.8495214073952346e-05,
      "loss": 0.6879,
      "step": 5848
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4974554777145386,
      "learning_rate": 1.8494624967743697e-05,
      "loss": 0.7641,
      "step": 5849
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5388158559799194,
      "learning_rate": 1.849403575562937e-05,
      "loss": 0.5598,
      "step": 5850
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6079468727111816,
      "learning_rate": 1.849344643761671e-05,
      "loss": 0.852,
      "step": 5851
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.659308671951294,
      "learning_rate": 1.8492857013713066e-05,
      "loss": 0.913,
      "step": 5852
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6264170408248901,
      "learning_rate": 1.8492267483925788e-05,
      "loss": 0.8492,
      "step": 5853
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.786021113395691,
      "learning_rate": 1.8491677848262224e-05,
      "loss": 0.9121,
      "step": 5854
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7629992961883545,
      "learning_rate": 1.8491088106729723e-05,
      "loss": 0.6771,
      "step": 5855
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8085542917251587,
      "learning_rate": 1.849049825933564e-05,
      "loss": 0.7076,
      "step": 5856
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4847402572631836,
      "learning_rate": 1.8489908306087328e-05,
      "loss": 0.7103,
      "step": 5857
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4000118970870972,
      "learning_rate": 1.8489318246992143e-05,
      "loss": 0.7813,
      "step": 5858
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.0247175693511963,
      "learning_rate": 1.848872808205744e-05,
      "loss": 0.8384,
      "step": 5859
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4764113426208496,
      "learning_rate": 1.8488137811290577e-05,
      "loss": 0.9085,
      "step": 5860
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4530717134475708,
      "learning_rate": 1.848754743469892e-05,
      "loss": 0.6829,
      "step": 5861
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7390940189361572,
      "learning_rate": 1.848695695228982e-05,
      "loss": 0.7018,
      "step": 5862
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.523048996925354,
      "learning_rate": 1.8486366364070638e-05,
      "loss": 0.8628,
      "step": 5863
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.364375114440918,
      "learning_rate": 1.8485775670048744e-05,
      "loss": 0.692,
      "step": 5864
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9927151203155518,
      "learning_rate": 1.8485184870231503e-05,
      "loss": 0.7237,
      "step": 5865
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5618964433670044,
      "learning_rate": 1.8484593964626274e-05,
      "loss": 0.8279,
      "step": 5866
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5183297395706177,
      "learning_rate": 1.848400295324043e-05,
      "loss": 0.7406,
      "step": 5867
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4617658853530884,
      "learning_rate": 1.8483411836081333e-05,
      "loss": 0.7785,
      "step": 5868
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4285223484039307,
      "learning_rate": 1.848282061315636e-05,
      "loss": 0.7568,
      "step": 5869
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4190634489059448,
      "learning_rate": 1.8482229284472878e-05,
      "loss": 0.6429,
      "step": 5870
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2882827520370483,
      "learning_rate": 1.8481637850038258e-05,
      "loss": 0.617,
      "step": 5871
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6142665147781372,
      "learning_rate": 1.8481046309859876e-05,
      "loss": 0.7479,
      "step": 5872
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7302147150039673,
      "learning_rate": 1.8480454663945107e-05,
      "loss": 0.7138,
      "step": 5873
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.74515962600708,
      "learning_rate": 1.8479862912301327e-05,
      "loss": 0.8263,
      "step": 5874
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8099756240844727,
      "learning_rate": 1.8479271054935913e-05,
      "loss": 0.7391,
      "step": 5875
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5052257776260376,
      "learning_rate": 1.8478679091856243e-05,
      "loss": 0.6796,
      "step": 5876
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.442386507987976,
      "learning_rate": 1.84780870230697e-05,
      "loss": 0.5942,
      "step": 5877
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5085372924804688,
      "learning_rate": 1.8477494848583662e-05,
      "loss": 0.7055,
      "step": 5878
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6300462484359741,
      "learning_rate": 1.8476902568405516e-05,
      "loss": 0.9588,
      "step": 5879
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8784698247909546,
      "learning_rate": 1.8476310182542643e-05,
      "loss": 0.7405,
      "step": 5880
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6858128309249878,
      "learning_rate": 1.847571769100243e-05,
      "loss": 0.6861,
      "step": 5881
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8008636236190796,
      "learning_rate": 1.8475125093792264e-05,
      "loss": 0.8145,
      "step": 5882
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.722834587097168,
      "learning_rate": 1.8474532390919534e-05,
      "loss": 0.8445,
      "step": 5883
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3868646621704102,
      "learning_rate": 1.847393958239162e-05,
      "loss": 0.7445,
      "step": 5884
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2659132480621338,
      "learning_rate": 1.8473346668215927e-05,
      "loss": 0.7947,
      "step": 5885
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6789604425430298,
      "learning_rate": 1.847275364839984e-05,
      "loss": 0.7269,
      "step": 5886
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2655017375946045,
      "learning_rate": 1.8472160522950755e-05,
      "loss": 0.6965,
      "step": 5887
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8729604482650757,
      "learning_rate": 1.8471567291876062e-05,
      "loss": 0.8704,
      "step": 5888
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.457877516746521,
      "learning_rate": 1.847097395518316e-05,
      "loss": 0.7003,
      "step": 5889
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6514874696731567,
      "learning_rate": 1.8470380512879446e-05,
      "loss": 0.6544,
      "step": 5890
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5689728260040283,
      "learning_rate": 1.846978696497232e-05,
      "loss": 0.6624,
      "step": 5891
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.478536605834961,
      "learning_rate": 1.8469193311469184e-05,
      "loss": 0.8021,
      "step": 5892
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6407791376113892,
      "learning_rate": 1.846859955237743e-05,
      "loss": 0.7027,
      "step": 5893
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.890522837638855,
      "learning_rate": 1.8468005687704468e-05,
      "loss": 0.6415,
      "step": 5894
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.799080491065979,
      "learning_rate": 1.8467411717457705e-05,
      "loss": 0.7781,
      "step": 5895
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8069746494293213,
      "learning_rate": 1.8466817641644538e-05,
      "loss": 0.8397,
      "step": 5896
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.9673941135406494,
      "learning_rate": 1.846622346027238e-05,
      "loss": 0.7107,
      "step": 5897
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.691522479057312,
      "learning_rate": 1.8465629173348634e-05,
      "loss": 0.7259,
      "step": 5898
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6305742263793945,
      "learning_rate": 1.8465034780880714e-05,
      "loss": 0.8432,
      "step": 5899
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5865716934204102,
      "learning_rate": 1.8464440282876026e-05,
      "loss": 0.6306,
      "step": 5900
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4383612871170044,
      "learning_rate": 1.8463845679341987e-05,
      "loss": 0.7255,
      "step": 5901
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6620478630065918,
      "learning_rate": 1.8463250970286005e-05,
      "loss": 0.9343,
      "step": 5902
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.441634178161621,
      "learning_rate": 1.8462656155715496e-05,
      "loss": 0.6407,
      "step": 5903
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5187150239944458,
      "learning_rate": 1.846206123563788e-05,
      "loss": 0.6913,
      "step": 5904
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3907642364501953,
      "learning_rate": 1.8461466210060566e-05,
      "loss": 0.7422,
      "step": 5905
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5580475330352783,
      "learning_rate": 1.8460871078990978e-05,
      "loss": 0.7467,
      "step": 5906
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.697208046913147,
      "learning_rate": 1.8460275842436535e-05,
      "loss": 0.7749,
      "step": 5907
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5335304737091064,
      "learning_rate": 1.8459680500404657e-05,
      "loss": 0.5907,
      "step": 5908
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4900275468826294,
      "learning_rate": 1.8459085052902767e-05,
      "loss": 0.8,
      "step": 5909
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.535046100616455,
      "learning_rate": 1.845848949993829e-05,
      "loss": 0.7097,
      "step": 5910
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.250923991203308,
      "learning_rate": 1.845789384151865e-05,
      "loss": 0.7924,
      "step": 5911
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5338022708892822,
      "learning_rate": 1.8457298077651273e-05,
      "loss": 0.708,
      "step": 5912
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6097099781036377,
      "learning_rate": 1.845670220834358e-05,
      "loss": 0.7902,
      "step": 5913
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3609308004379272,
      "learning_rate": 1.8456106233603014e-05,
      "loss": 0.7119,
      "step": 5914
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3337595462799072,
      "learning_rate": 1.8455510153436997e-05,
      "loss": 0.7063,
      "step": 5915
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2451071739196777,
      "learning_rate": 1.845491396785296e-05,
      "loss": 0.5874,
      "step": 5916
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2033007144927979,
      "learning_rate": 1.8454317676858336e-05,
      "loss": 0.7499,
      "step": 5917
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6109910011291504,
      "learning_rate": 1.845372128046056e-05,
      "loss": 0.6469,
      "step": 5918
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6505305767059326,
      "learning_rate": 1.8453124778667072e-05,
      "loss": 0.7448,
      "step": 5919
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.923891544342041,
      "learning_rate": 1.84525281714853e-05,
      "loss": 0.836,
      "step": 5920
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5144386291503906,
      "learning_rate": 1.8451931458922688e-05,
      "loss": 0.6056,
      "step": 5921
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4807947874069214,
      "learning_rate": 1.8451334640986674e-05,
      "loss": 0.6802,
      "step": 5922
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5846328735351562,
      "learning_rate": 1.84507377176847e-05,
      "loss": 0.7319,
      "step": 5923
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.284191608428955,
      "learning_rate": 1.8450140689024207e-05,
      "loss": 0.7629,
      "step": 5924
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.751984715461731,
      "learning_rate": 1.8449543555012638e-05,
      "loss": 0.7805,
      "step": 5925
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4689995050430298,
      "learning_rate": 1.8448946315657437e-05,
      "loss": 0.6993,
      "step": 5926
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3236488103866577,
      "learning_rate": 1.8448348970966053e-05,
      "loss": 0.6282,
      "step": 5927
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5242242813110352,
      "learning_rate": 1.8447751520945933e-05,
      "loss": 0.6562,
      "step": 5928
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5940885543823242,
      "learning_rate": 1.8447153965604522e-05,
      "loss": 0.776,
      "step": 5929
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7665058374404907,
      "learning_rate": 1.844655630494927e-05,
      "loss": 0.7456,
      "step": 5930
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8526860475540161,
      "learning_rate": 1.844595853898763e-05,
      "loss": 0.8285,
      "step": 5931
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.55104660987854,
      "learning_rate": 1.844536066772706e-05,
      "loss": 0.8042,
      "step": 5932
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4333076477050781,
      "learning_rate": 1.8444762691175e-05,
      "loss": 0.8696,
      "step": 5933
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.561957836151123,
      "learning_rate": 1.8444164609338924e-05,
      "loss": 0.5973,
      "step": 5934
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4193004369735718,
      "learning_rate": 1.844356642222627e-05,
      "loss": 0.6104,
      "step": 5935
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.55640709400177,
      "learning_rate": 1.8442968129844508e-05,
      "loss": 0.7013,
      "step": 5936
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5670777559280396,
      "learning_rate": 1.8442369732201094e-05,
      "loss": 0.8362,
      "step": 5937
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5179266929626465,
      "learning_rate": 1.844177122930349e-05,
      "loss": 0.7314,
      "step": 5938
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.53774094581604,
      "learning_rate": 1.8441172621159153e-05,
      "loss": 0.6914,
      "step": 5939
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.51024329662323,
      "learning_rate": 1.8440573907775547e-05,
      "loss": 0.7275,
      "step": 5940
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5373255014419556,
      "learning_rate": 1.843997508916014e-05,
      "loss": 0.7665,
      "step": 5941
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.35696280002594,
      "learning_rate": 1.8439376165320393e-05,
      "loss": 0.5814,
      "step": 5942
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5742807388305664,
      "learning_rate": 1.843877713626378e-05,
      "loss": 0.8071,
      "step": 5943
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6432360410690308,
      "learning_rate": 1.8438178001997763e-05,
      "loss": 0.7976,
      "step": 5944
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4944559335708618,
      "learning_rate": 1.843757876252981e-05,
      "loss": 0.6673,
      "step": 5945
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3565233945846558,
      "learning_rate": 1.84369794178674e-05,
      "loss": 0.8204,
      "step": 5946
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4413783550262451,
      "learning_rate": 1.8436379968018e-05,
      "loss": 0.6968,
      "step": 5947
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.682010293006897,
      "learning_rate": 1.8435780412989087e-05,
      "loss": 0.688,
      "step": 5948
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.876076102256775,
      "learning_rate": 1.8435180752788133e-05,
      "loss": 0.8451,
      "step": 5949
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5706722736358643,
      "learning_rate": 1.8434580987422607e-05,
      "loss": 0.7847,
      "step": 5950
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4978140592575073,
      "learning_rate": 1.84339811169e-05,
      "loss": 0.7875,
      "step": 5951
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5365291833877563,
      "learning_rate": 1.8433381141227785e-05,
      "loss": 0.7752,
      "step": 5952
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5155153274536133,
      "learning_rate": 1.843278106041344e-05,
      "loss": 0.7195,
      "step": 5953
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9370536804199219,
      "learning_rate": 1.8432180874464453e-05,
      "loss": 0.7287,
      "step": 5954
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5688433647155762,
      "learning_rate": 1.8431580583388298e-05,
      "loss": 0.7515,
      "step": 5955
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.670833706855774,
      "learning_rate": 1.8430980187192464e-05,
      "loss": 0.7921,
      "step": 5956
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9744786024093628,
      "learning_rate": 1.8430379685884436e-05,
      "loss": 0.8241,
      "step": 5957
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.690497636795044,
      "learning_rate": 1.8429779079471698e-05,
      "loss": 0.711,
      "step": 5958
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.576248049736023,
      "learning_rate": 1.842917836796174e-05,
      "loss": 0.7287,
      "step": 5959
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.78910231590271,
      "learning_rate": 1.8428577551362055e-05,
      "loss": 0.8036,
      "step": 5960
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4366737604141235,
      "learning_rate": 1.842797662968013e-05,
      "loss": 0.6841,
      "step": 5961
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8158036470413208,
      "learning_rate": 1.8427375602923457e-05,
      "loss": 0.7203,
      "step": 5962
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5984076261520386,
      "learning_rate": 1.8426774471099525e-05,
      "loss": 0.6168,
      "step": 5963
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.619758129119873,
      "learning_rate": 1.8426173234215838e-05,
      "loss": 0.904,
      "step": 5964
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7459243535995483,
      "learning_rate": 1.8425571892279885e-05,
      "loss": 0.885,
      "step": 5965
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5929207801818848,
      "learning_rate": 1.842497044529916e-05,
      "loss": 0.7899,
      "step": 5966
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.203022837638855,
      "learning_rate": 1.842436889328117e-05,
      "loss": 0.6664,
      "step": 5967
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4417250156402588,
      "learning_rate": 1.8423767236233414e-05,
      "loss": 0.7751,
      "step": 5968
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4884687662124634,
      "learning_rate": 1.842316547416339e-05,
      "loss": 0.7061,
      "step": 5969
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4808369874954224,
      "learning_rate": 1.8422563607078597e-05,
      "loss": 0.7148,
      "step": 5970
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6180384159088135,
      "learning_rate": 1.8421961634986544e-05,
      "loss": 0.761,
      "step": 5971
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.373062014579773,
      "learning_rate": 1.842135955789473e-05,
      "loss": 0.7916,
      "step": 5972
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3593990802764893,
      "learning_rate": 1.842075737581067e-05,
      "loss": 0.7221,
      "step": 5973
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4572372436523438,
      "learning_rate": 1.8420155088741867e-05,
      "loss": 0.709,
      "step": 5974
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7662372589111328,
      "learning_rate": 1.841955269669583e-05,
      "loss": 0.8229,
      "step": 5975
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9755935668945312,
      "learning_rate": 1.8418950199680066e-05,
      "loss": 0.5213,
      "step": 5976
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.431085228919983,
      "learning_rate": 1.8418347597702096e-05,
      "loss": 0.6571,
      "step": 5977
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4116908311843872,
      "learning_rate": 1.841774489076942e-05,
      "loss": 0.7257,
      "step": 5978
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5954686403274536,
      "learning_rate": 1.8417142078889564e-05,
      "loss": 0.7788,
      "step": 5979
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.963265061378479,
      "learning_rate": 1.8416539162070036e-05,
      "loss": 0.666,
      "step": 5980
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.572086215019226,
      "learning_rate": 1.841593614031836e-05,
      "loss": 0.7629,
      "step": 5981
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5985668897628784,
      "learning_rate": 1.841533301364204e-05,
      "loss": 0.6829,
      "step": 5982
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6963367462158203,
      "learning_rate": 1.841472978204861e-05,
      "loss": 0.7087,
      "step": 5983
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7839713096618652,
      "learning_rate": 1.841412644554559e-05,
      "loss": 0.7324,
      "step": 5984
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7343242168426514,
      "learning_rate": 1.841352300414049e-05,
      "loss": 0.6608,
      "step": 5985
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8655225038528442,
      "learning_rate": 1.8412919457840844e-05,
      "loss": 0.6947,
      "step": 5986
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.9612114429473877,
      "learning_rate": 1.8412315806654173e-05,
      "loss": 0.6706,
      "step": 5987
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3734058141708374,
      "learning_rate": 1.8411712050588004e-05,
      "loss": 0.7563,
      "step": 5988
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5444450378417969,
      "learning_rate": 1.8411108189649863e-05,
      "loss": 0.6596,
      "step": 5989
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3418885469436646,
      "learning_rate": 1.8410504223847278e-05,
      "loss": 0.6565,
      "step": 5990
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8599013090133667,
      "learning_rate": 1.8409900153187784e-05,
      "loss": 0.8229,
      "step": 5991
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3697565793991089,
      "learning_rate": 1.8409295977678906e-05,
      "loss": 0.6509,
      "step": 5992
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3965532779693604,
      "learning_rate": 1.840869169732818e-05,
      "loss": 0.7378,
      "step": 5993
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4642024040222168,
      "learning_rate": 1.8408087312143136e-05,
      "loss": 0.7034,
      "step": 5994
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.377296805381775,
      "learning_rate": 1.8407482822131314e-05,
      "loss": 0.629,
      "step": 5995
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2203187942504883,
      "learning_rate": 1.8406878227300246e-05,
      "loss": 0.6599,
      "step": 5996
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5302627086639404,
      "learning_rate": 1.840627352765747e-05,
      "loss": 0.7424,
      "step": 5997
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3076180219650269,
      "learning_rate": 1.840566872321053e-05,
      "loss": 0.6713,
      "step": 5998
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5049854516983032,
      "learning_rate": 1.8405063813966963e-05,
      "loss": 0.73,
      "step": 5999
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5979983806610107,
      "learning_rate": 1.8404458799934312e-05,
      "loss": 0.8713,
      "step": 6000
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.927076816558838,
      "learning_rate": 1.8403853681120114e-05,
      "loss": 0.7379,
      "step": 6001
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4152876138687134,
      "learning_rate": 1.840324845753192e-05,
      "loss": 0.7339,
      "step": 6002
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7575266361236572,
      "learning_rate": 1.8402643129177275e-05,
      "loss": 0.7379,
      "step": 6003
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5434508323669434,
      "learning_rate": 1.8402037696063723e-05,
      "loss": 0.573,
      "step": 6004
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3877936601638794,
      "learning_rate": 1.8401432158198816e-05,
      "loss": 0.7376,
      "step": 6005
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.732982873916626,
      "learning_rate": 1.8400826515590096e-05,
      "loss": 0.6709,
      "step": 6006
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4893795251846313,
      "learning_rate": 1.840022076824512e-05,
      "loss": 0.7254,
      "step": 6007
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6948049068450928,
      "learning_rate": 1.8399614916171443e-05,
      "loss": 0.6901,
      "step": 6008
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5074479579925537,
      "learning_rate": 1.839900895937661e-05,
      "loss": 0.8601,
      "step": 6009
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5721063613891602,
      "learning_rate": 1.839840289786818e-05,
      "loss": 0.7564,
      "step": 6010
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7265158891677856,
      "learning_rate": 1.839779673165371e-05,
      "loss": 0.8234,
      "step": 6011
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7046345472335815,
      "learning_rate": 1.8397190460740756e-05,
      "loss": 0.8856,
      "step": 6012
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3833961486816406,
      "learning_rate": 1.839658408513688e-05,
      "loss": 0.5802,
      "step": 6013
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6389714479446411,
      "learning_rate": 1.8395977604849634e-05,
      "loss": 0.7116,
      "step": 6014
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6680231094360352,
      "learning_rate": 1.8395371019886586e-05,
      "loss": 0.749,
      "step": 6015
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4254473447799683,
      "learning_rate": 1.8394764330255294e-05,
      "loss": 0.683,
      "step": 6016
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6515522003173828,
      "learning_rate": 1.839415753596333e-05,
      "loss": 0.7386,
      "step": 6017
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3676929473876953,
      "learning_rate": 1.8393550637018246e-05,
      "loss": 0.8206,
      "step": 6018
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.098027229309082,
      "learning_rate": 1.839294363342762e-05,
      "loss": 0.8145,
      "step": 6019
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4035859107971191,
      "learning_rate": 1.8392336525199013e-05,
      "loss": 0.8104,
      "step": 6020
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.64716637134552,
      "learning_rate": 1.839172931234e-05,
      "loss": 0.6808,
      "step": 6021
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5254188776016235,
      "learning_rate": 1.8391121994858145e-05,
      "loss": 0.6952,
      "step": 6022
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4714434146881104,
      "learning_rate": 1.8390514572761024e-05,
      "loss": 0.785,
      "step": 6023
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4251735210418701,
      "learning_rate": 1.838990704605621e-05,
      "loss": 0.9242,
      "step": 6024
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8220841884613037,
      "learning_rate": 1.8389299414751273e-05,
      "loss": 0.7753,
      "step": 6025
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6434555053710938,
      "learning_rate": 1.8388691678853794e-05,
      "loss": 0.7576,
      "step": 6026
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4023663997650146,
      "learning_rate": 1.8388083838371345e-05,
      "loss": 0.7643,
      "step": 6027
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.423278570175171,
      "learning_rate": 1.838747589331151e-05,
      "loss": 0.7037,
      "step": 6028
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.392786979675293,
      "learning_rate": 1.8386867843681863e-05,
      "loss": 0.6592,
      "step": 6029
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.604302167892456,
      "learning_rate": 1.8386259689489984e-05,
      "loss": 0.6762,
      "step": 6030
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3975872993469238,
      "learning_rate": 1.838565143074346e-05,
      "loss": 0.8506,
      "step": 6031
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4291378259658813,
      "learning_rate": 1.838504306744987e-05,
      "loss": 0.801,
      "step": 6032
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5272741317749023,
      "learning_rate": 1.83844345996168e-05,
      "loss": 0.7838,
      "step": 6033
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8792753219604492,
      "learning_rate": 1.8383826027251845e-05,
      "loss": 0.7515,
      "step": 6034
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.533073902130127,
      "learning_rate": 1.8383217350362575e-05,
      "loss": 0.734,
      "step": 6035
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3910850286483765,
      "learning_rate": 1.8382608568956596e-05,
      "loss": 0.7418,
      "step": 6036
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4953267574310303,
      "learning_rate": 1.8381999683041482e-05,
      "loss": 0.7506,
      "step": 6037
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3651331663131714,
      "learning_rate": 1.8381390692624834e-05,
      "loss": 0.7418,
      "step": 6038
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4065555334091187,
      "learning_rate": 1.8380781597714247e-05,
      "loss": 0.7036,
      "step": 6039
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3606802225112915,
      "learning_rate": 1.8380172398317305e-05,
      "loss": 0.8211,
      "step": 6040
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5019053220748901,
      "learning_rate": 1.837956309444161e-05,
      "loss": 0.7559,
      "step": 6041
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7721529006958008,
      "learning_rate": 1.8378953686094756e-05,
      "loss": 0.6633,
      "step": 6042
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.78330397605896,
      "learning_rate": 1.8378344173284344e-05,
      "loss": 0.7527,
      "step": 6043
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.544240117073059,
      "learning_rate": 1.8377734556017968e-05,
      "loss": 0.6403,
      "step": 6044
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.442259430885315,
      "learning_rate": 1.8377124834303234e-05,
      "loss": 0.8054,
      "step": 6045
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.390701174736023,
      "learning_rate": 1.8376515008147736e-05,
      "loss": 0.8382,
      "step": 6046
    },
    {
      "epoch": 0.63,
      "grad_norm": 51.9072380065918,
      "learning_rate": 1.8375905077559082e-05,
      "loss": 0.6353,
      "step": 6047
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5723059177398682,
      "learning_rate": 1.8375295042544876e-05,
      "loss": 0.8699,
      "step": 6048
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6466118097305298,
      "learning_rate": 1.8374684903112724e-05,
      "loss": 0.7692,
      "step": 6049
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6674600839614868,
      "learning_rate": 1.8374074659270235e-05,
      "loss": 0.854,
      "step": 6050
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8806875944137573,
      "learning_rate": 1.837346431102501e-05,
      "loss": 0.7098,
      "step": 6051
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.357704758644104,
      "learning_rate": 1.8372853858384662e-05,
      "loss": 0.8558,
      "step": 6052
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6580042839050293,
      "learning_rate": 1.8372243301356807e-05,
      "loss": 0.8488,
      "step": 6053
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5013598203659058,
      "learning_rate": 1.837163263994905e-05,
      "loss": 0.8973,
      "step": 6054
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8740839958190918,
      "learning_rate": 1.8371021874169005e-05,
      "loss": 0.8279,
      "step": 6055
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.629590630531311,
      "learning_rate": 1.837041100402429e-05,
      "loss": 0.7707,
      "step": 6056
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5213788747787476,
      "learning_rate": 1.836980002952252e-05,
      "loss": 0.7381,
      "step": 6057
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.489929437637329,
      "learning_rate": 1.8369188950671312e-05,
      "loss": 0.5814,
      "step": 6058
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6277976036071777,
      "learning_rate": 1.836857776747828e-05,
      "loss": 0.7819,
      "step": 6059
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5423243045806885,
      "learning_rate": 1.8367966479951048e-05,
      "loss": 0.5948,
      "step": 6060
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.672681450843811,
      "learning_rate": 1.8367355088097244e-05,
      "loss": 0.6899,
      "step": 6061
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7646510601043701,
      "learning_rate": 1.8366743591924475e-05,
      "loss": 0.7754,
      "step": 6062
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4006471633911133,
      "learning_rate": 1.8366131991440377e-05,
      "loss": 0.828,
      "step": 6063
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4656751155853271,
      "learning_rate": 1.8365520286652575e-05,
      "loss": 0.688,
      "step": 6064
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4908758401870728,
      "learning_rate": 1.8364908477568688e-05,
      "loss": 0.779,
      "step": 6065
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2699685096740723,
      "learning_rate": 1.8364296564196346e-05,
      "loss": 0.6055,
      "step": 6066
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4749687910079956,
      "learning_rate": 1.8363684546543182e-05,
      "loss": 0.6532,
      "step": 6067
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6183936595916748,
      "learning_rate": 1.8363072424616824e-05,
      "loss": 0.6983,
      "step": 6068
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3507076501846313,
      "learning_rate": 1.83624601984249e-05,
      "loss": 0.5736,
      "step": 6069
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7144349813461304,
      "learning_rate": 1.836184786797505e-05,
      "loss": 0.7856,
      "step": 6070
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5503571033477783,
      "learning_rate": 1.83612354332749e-05,
      "loss": 0.5816,
      "step": 6071
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.097271680831909,
      "learning_rate": 1.8360622894332092e-05,
      "loss": 0.8835,
      "step": 6072
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9346987009048462,
      "learning_rate": 1.836001025115426e-05,
      "loss": 0.5996,
      "step": 6073
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.441186785697937,
      "learning_rate": 1.835939750374904e-05,
      "loss": 0.6382,
      "step": 6074
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6646933555603027,
      "learning_rate": 1.8358784652124077e-05,
      "loss": 0.6479,
      "step": 6075
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5939569473266602,
      "learning_rate": 1.835817169628701e-05,
      "loss": 0.6696,
      "step": 6076
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4814060926437378,
      "learning_rate": 1.8357558636245475e-05,
      "loss": 0.6719,
      "step": 6077
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5343313217163086,
      "learning_rate": 1.835694547200712e-05,
      "loss": 0.7802,
      "step": 6078
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.690730094909668,
      "learning_rate": 1.835633220357959e-05,
      "loss": 0.6362,
      "step": 6079
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.347634196281433,
      "learning_rate": 1.8355718830970533e-05,
      "loss": 0.7841,
      "step": 6080
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5114965438842773,
      "learning_rate": 1.835510535418759e-05,
      "loss": 0.7768,
      "step": 6081
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3840006589889526,
      "learning_rate": 1.835449177323841e-05,
      "loss": 0.7975,
      "step": 6082
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9010463953018188,
      "learning_rate": 1.835387808813065e-05,
      "loss": 0.8774,
      "step": 6083
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.443410873413086,
      "learning_rate": 1.8353264298871954e-05,
      "loss": 0.887,
      "step": 6084
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5962331295013428,
      "learning_rate": 1.8352650405469977e-05,
      "loss": 0.784,
      "step": 6085
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3149771690368652,
      "learning_rate": 1.8352036407932373e-05,
      "loss": 0.8998,
      "step": 6086
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4057433605194092,
      "learning_rate": 1.8351422306266797e-05,
      "loss": 0.7257,
      "step": 6087
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5570896863937378,
      "learning_rate": 1.8350808100480902e-05,
      "loss": 0.6731,
      "step": 6088
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4281262159347534,
      "learning_rate": 1.8350193790582352e-05,
      "loss": 0.6853,
      "step": 6089
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4663033485412598,
      "learning_rate": 1.8349579376578797e-05,
      "loss": 0.6471,
      "step": 6090
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.00542950630188,
      "learning_rate": 1.8348964858477906e-05,
      "loss": 0.78,
      "step": 6091
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9228184223175049,
      "learning_rate": 1.8348350236287337e-05,
      "loss": 0.6385,
      "step": 6092
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2419559955596924,
      "learning_rate": 1.834773551001475e-05,
      "loss": 0.6055,
      "step": 6093
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8575336933135986,
      "learning_rate": 1.834712067966781e-05,
      "loss": 0.8499,
      "step": 6094
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6163355112075806,
      "learning_rate": 1.8346505745254185e-05,
      "loss": 0.7386,
      "step": 6095
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5027999877929688,
      "learning_rate": 1.834589070678154e-05,
      "loss": 0.715,
      "step": 6096
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.819066047668457,
      "learning_rate": 1.8345275564257544e-05,
      "loss": 0.8902,
      "step": 6097
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4556145668029785,
      "learning_rate": 1.8344660317689862e-05,
      "loss": 0.6652,
      "step": 6098
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.580581545829773,
      "learning_rate": 1.8344044967086172e-05,
      "loss": 0.7705,
      "step": 6099
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6473582983016968,
      "learning_rate": 1.8343429512454137e-05,
      "loss": 0.6521,
      "step": 6100
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5295186042785645,
      "learning_rate": 1.8342813953801436e-05,
      "loss": 0.7353,
      "step": 6101
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6589906215667725,
      "learning_rate": 1.8342198291135744e-05,
      "loss": 0.6442,
      "step": 6102
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.2221086025238037,
      "learning_rate": 1.8341582524464737e-05,
      "loss": 0.8092,
      "step": 6103
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5585037469863892,
      "learning_rate": 1.8340966653796085e-05,
      "loss": 0.6681,
      "step": 6104
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5759238004684448,
      "learning_rate": 1.8340350679137473e-05,
      "loss": 0.6815,
      "step": 6105
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.830348014831543,
      "learning_rate": 1.833973460049658e-05,
      "loss": 0.6664,
      "step": 6106
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4873765707015991,
      "learning_rate": 1.8339118417881083e-05,
      "loss": 0.6414,
      "step": 6107
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4344502687454224,
      "learning_rate": 1.8338502131298668e-05,
      "loss": 0.5137,
      "step": 6108
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.649643898010254,
      "learning_rate": 1.833788574075702e-05,
      "loss": 0.6379,
      "step": 6109
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8086391687393188,
      "learning_rate": 1.8337269246263815e-05,
      "loss": 0.9143,
      "step": 6110
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.456842064857483,
      "learning_rate": 1.8336652647826748e-05,
      "loss": 0.6708,
      "step": 6111
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6839253902435303,
      "learning_rate": 1.8336035945453505e-05,
      "loss": 0.7122,
      "step": 6112
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3768166303634644,
      "learning_rate": 1.8335419139151768e-05,
      "loss": 0.6395,
      "step": 6113
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.603707194328308,
      "learning_rate": 1.833480222892924e-05,
      "loss": 0.86,
      "step": 6114
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.401505947113037,
      "learning_rate": 1.8334185214793597e-05,
      "loss": 0.6193,
      "step": 6115
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.731063723564148,
      "learning_rate": 1.8333568096752544e-05,
      "loss": 0.6067,
      "step": 6116
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9195537567138672,
      "learning_rate": 1.8332950874813767e-05,
      "loss": 0.8262,
      "step": 6117
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5148202180862427,
      "learning_rate": 1.833233354898496e-05,
      "loss": 0.6602,
      "step": 6118
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.998075246810913,
      "learning_rate": 1.833171611927383e-05,
      "loss": 0.741,
      "step": 6119
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5320336818695068,
      "learning_rate": 1.8331098585688063e-05,
      "loss": 0.6562,
      "step": 6120
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4929379224777222,
      "learning_rate": 1.8330480948235365e-05,
      "loss": 0.7693,
      "step": 6121
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5901739597320557,
      "learning_rate": 1.8329863206923435e-05,
      "loss": 0.6913,
      "step": 6122
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7920109033584595,
      "learning_rate": 1.8329245361759973e-05,
      "loss": 0.6368,
      "step": 6123
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4610515832901,
      "learning_rate": 1.8328627412752684e-05,
      "loss": 0.8745,
      "step": 6124
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5289921760559082,
      "learning_rate": 1.832800935990927e-05,
      "loss": 0.7532,
      "step": 6125
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7796356678009033,
      "learning_rate": 1.8327391203237436e-05,
      "loss": 0.8645,
      "step": 6126
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3974822759628296,
      "learning_rate": 1.8326772942744894e-05,
      "loss": 0.7797,
      "step": 6127
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4763115644454956,
      "learning_rate": 1.8326154578439347e-05,
      "loss": 0.7529,
      "step": 6128
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4626833200454712,
      "learning_rate": 1.8325536110328502e-05,
      "loss": 0.8212,
      "step": 6129
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7399256229400635,
      "learning_rate": 1.832491753842008e-05,
      "loss": 0.7386,
      "step": 6130
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7728794813156128,
      "learning_rate": 1.832429886272178e-05,
      "loss": 0.7833,
      "step": 6131
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3190524578094482,
      "learning_rate": 1.8323680083241323e-05,
      "loss": 0.599,
      "step": 6132
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6504712104797363,
      "learning_rate": 1.8323061199986428e-05,
      "loss": 0.7611,
      "step": 6133
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2652394771575928,
      "learning_rate": 1.83224422129648e-05,
      "loss": 0.6856,
      "step": 6134
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.394720435142517,
      "learning_rate": 1.832182312218416e-05,
      "loss": 0.6974,
      "step": 6135
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5866930484771729,
      "learning_rate": 1.832120392765223e-05,
      "loss": 0.6365,
      "step": 6136
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5649256706237793,
      "learning_rate": 1.8320584629376727e-05,
      "loss": 0.6369,
      "step": 6137
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6677446365356445,
      "learning_rate": 1.8319965227365373e-05,
      "loss": 0.8482,
      "step": 6138
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.348031997680664,
      "learning_rate": 1.831934572162589e-05,
      "loss": 0.6396,
      "step": 6139
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.604977011680603,
      "learning_rate": 1.8318726112166e-05,
      "loss": 0.8762,
      "step": 6140
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5843161344528198,
      "learning_rate": 1.831810639899343e-05,
      "loss": 0.8674,
      "step": 6141
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7809244394302368,
      "learning_rate": 1.8317486582115906e-05,
      "loss": 0.8149,
      "step": 6142
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6907316446304321,
      "learning_rate": 1.8316866661541152e-05,
      "loss": 0.6933,
      "step": 6143
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5344172716140747,
      "learning_rate": 1.83162466372769e-05,
      "loss": 0.6733,
      "step": 6144
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4520012140274048,
      "learning_rate": 1.8315626509330885e-05,
      "loss": 0.7058,
      "step": 6145
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.618518352508545,
      "learning_rate": 1.831500627771083e-05,
      "loss": 0.6581,
      "step": 6146
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3969637155532837,
      "learning_rate": 1.8314385942424467e-05,
      "loss": 0.8252,
      "step": 6147
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5259370803833008,
      "learning_rate": 1.831376550347954e-05,
      "loss": 0.6837,
      "step": 6148
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4323019981384277,
      "learning_rate": 1.8313144960883777e-05,
      "loss": 0.7914,
      "step": 6149
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8767602443695068,
      "learning_rate": 1.8312524314644913e-05,
      "loss": 0.8026,
      "step": 6150
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0139060020446777,
      "learning_rate": 1.8311903564770688e-05,
      "loss": 0.8634,
      "step": 6151
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4523091316223145,
      "learning_rate": 1.8311282711268842e-05,
      "loss": 0.7207,
      "step": 6152
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3704344034194946,
      "learning_rate": 1.8310661754147115e-05,
      "loss": 0.7119,
      "step": 6153
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9063849449157715,
      "learning_rate": 1.831004069341325e-05,
      "loss": 0.9449,
      "step": 6154
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.449473261833191,
      "learning_rate": 1.830941952907499e-05,
      "loss": 0.5063,
      "step": 6155
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7561753988265991,
      "learning_rate": 1.8308798261140076e-05,
      "loss": 0.7407,
      "step": 6156
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5171693563461304,
      "learning_rate": 1.8308176889616254e-05,
      "loss": 0.7035,
      "step": 6157
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3778376579284668,
      "learning_rate": 1.8307555414511274e-05,
      "loss": 0.7289,
      "step": 6158
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8591831922531128,
      "learning_rate": 1.830693383583288e-05,
      "loss": 0.7756,
      "step": 6159
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6796634197235107,
      "learning_rate": 1.830631215358883e-05,
      "loss": 0.8313,
      "step": 6160
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5212733745574951,
      "learning_rate": 1.830569036778686e-05,
      "loss": 0.8482,
      "step": 6161
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4245688915252686,
      "learning_rate": 1.830506847843474e-05,
      "loss": 0.7218,
      "step": 6162
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4394631385803223,
      "learning_rate": 1.8304446485540207e-05,
      "loss": 0.7381,
      "step": 6163
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4625043869018555,
      "learning_rate": 1.8303824389111027e-05,
      "loss": 0.8383,
      "step": 6164
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.59284508228302,
      "learning_rate": 1.830320218915495e-05,
      "loss": 0.9958,
      "step": 6165
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4679392576217651,
      "learning_rate": 1.8302579885679736e-05,
      "loss": 0.7545,
      "step": 6166
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.491166353225708,
      "learning_rate": 1.830195747869314e-05,
      "loss": 0.6659,
      "step": 6167
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5384567975997925,
      "learning_rate": 1.8301334968202925e-05,
      "loss": 0.6253,
      "step": 6168
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4448822736740112,
      "learning_rate": 1.830071235421685e-05,
      "loss": 0.7259,
      "step": 6169
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0517027378082275,
      "learning_rate": 1.830008963674268e-05,
      "loss": 0.8401,
      "step": 6170
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4309008121490479,
      "learning_rate": 1.829946681578818e-05,
      "loss": 0.6319,
      "step": 6171
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.362527847290039,
      "learning_rate": 1.8298843891361108e-05,
      "loss": 0.781,
      "step": 6172
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5673160552978516,
      "learning_rate": 1.8298220863469237e-05,
      "loss": 0.8578,
      "step": 6173
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4016116857528687,
      "learning_rate": 1.829759773212033e-05,
      "loss": 0.6852,
      "step": 6174
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4240984916687012,
      "learning_rate": 1.8296974497322154e-05,
      "loss": 0.7249,
      "step": 6175
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.755946159362793,
      "learning_rate": 1.829635115908249e-05,
      "loss": 0.7172,
      "step": 6176
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9622886180877686,
      "learning_rate": 1.82957277174091e-05,
      "loss": 0.6522,
      "step": 6177
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6247444152832031,
      "learning_rate": 1.8295104172309756e-05,
      "loss": 0.7342,
      "step": 6178
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5657124519348145,
      "learning_rate": 1.8294480523792236e-05,
      "loss": 0.7237,
      "step": 6179
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.921989917755127,
      "learning_rate": 1.8293856771864317e-05,
      "loss": 0.6722,
      "step": 6180
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.476029634475708,
      "learning_rate": 1.829323291653377e-05,
      "loss": 0.7937,
      "step": 6181
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.522572636604309,
      "learning_rate": 1.8292608957808377e-05,
      "loss": 0.7832,
      "step": 6182
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1491196155548096,
      "learning_rate": 1.8291984895695915e-05,
      "loss": 0.687,
      "step": 6183
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5110934972763062,
      "learning_rate": 1.8291360730204162e-05,
      "loss": 0.6593,
      "step": 6184
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4589800834655762,
      "learning_rate": 1.8290736461340905e-05,
      "loss": 0.7804,
      "step": 6185
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7384750843048096,
      "learning_rate": 1.8290112089113926e-05,
      "loss": 0.6863,
      "step": 6186
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5134090185165405,
      "learning_rate": 1.8289487613531008e-05,
      "loss": 0.6039,
      "step": 6187
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.451690673828125,
      "learning_rate": 1.8288863034599936e-05,
      "loss": 0.8062,
      "step": 6188
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2945051193237305,
      "learning_rate": 1.8288238352328497e-05,
      "loss": 0.6914,
      "step": 6189
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3457039594650269,
      "learning_rate": 1.828761356672448e-05,
      "loss": 0.5357,
      "step": 6190
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5791045427322388,
      "learning_rate": 1.8286988677795673e-05,
      "loss": 0.8703,
      "step": 6191
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6263275146484375,
      "learning_rate": 1.8286363685549867e-05,
      "loss": 0.7196,
      "step": 6192
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6268489360809326,
      "learning_rate": 1.828573858999486e-05,
      "loss": 0.6247,
      "step": 6193
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4130686521530151,
      "learning_rate": 1.8285113391138434e-05,
      "loss": 0.8353,
      "step": 6194
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6506391763687134,
      "learning_rate": 1.8284488088988392e-05,
      "loss": 0.6857,
      "step": 6195
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4622864723205566,
      "learning_rate": 1.8283862683552528e-05,
      "loss": 0.7182,
      "step": 6196
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5895824432373047,
      "learning_rate": 1.8283237174838638e-05,
      "loss": 0.6945,
      "step": 6197
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.464690923690796,
      "learning_rate": 1.828261156285452e-05,
      "loss": 0.712,
      "step": 6198
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8049911260604858,
      "learning_rate": 1.8281985847607975e-05,
      "loss": 0.6472,
      "step": 6199
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6047027111053467,
      "learning_rate": 1.8281360029106805e-05,
      "loss": 0.6807,
      "step": 6200
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5795074701309204,
      "learning_rate": 1.8280734107358813e-05,
      "loss": 0.7349,
      "step": 6201
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3948992490768433,
      "learning_rate": 1.82801080823718e-05,
      "loss": 0.7056,
      "step": 6202
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7015138864517212,
      "learning_rate": 1.827948195415357e-05,
      "loss": 0.8247,
      "step": 6203
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4298471212387085,
      "learning_rate": 1.8278855722711932e-05,
      "loss": 0.7295,
      "step": 6204
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5236996412277222,
      "learning_rate": 1.827822938805469e-05,
      "loss": 0.9273,
      "step": 6205
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5310680866241455,
      "learning_rate": 1.827760295018966e-05,
      "loss": 0.6471,
      "step": 6206
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8074588775634766,
      "learning_rate": 1.8276976409124644e-05,
      "loss": 0.8036,
      "step": 6207
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.527084231376648,
      "learning_rate": 1.8276349764867455e-05,
      "loss": 0.6951,
      "step": 6208
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.499497890472412,
      "learning_rate": 1.8275723017425912e-05,
      "loss": 0.7391,
      "step": 6209
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.1740212440490723,
      "learning_rate": 1.8275096166807817e-05,
      "loss": 0.6574,
      "step": 6210
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5416717529296875,
      "learning_rate": 1.8274469213020998e-05,
      "loss": 0.5902,
      "step": 6211
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.991788625717163,
      "learning_rate": 1.8273842156073264e-05,
      "loss": 0.7655,
      "step": 6212
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9418803453445435,
      "learning_rate": 1.8273214995972434e-05,
      "loss": 0.8657,
      "step": 6213
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7511569261550903,
      "learning_rate": 1.8272587732726325e-05,
      "loss": 0.7751,
      "step": 6214
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.399618148803711,
      "learning_rate": 1.8271960366342764e-05,
      "loss": 0.5609,
      "step": 6215
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.62284255027771,
      "learning_rate": 1.8271332896829564e-05,
      "loss": 0.711,
      "step": 6216
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6066330671310425,
      "learning_rate": 1.8270705324194554e-05,
      "loss": 0.7767,
      "step": 6217
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.027541399002075,
      "learning_rate": 1.8270077648445558e-05,
      "loss": 0.717,
      "step": 6218
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4738283157348633,
      "learning_rate": 1.8269449869590403e-05,
      "loss": 0.8027,
      "step": 6219
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.535486102104187,
      "learning_rate": 1.8268821987636904e-05,
      "loss": 0.7353,
      "step": 6220
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2933580875396729,
      "learning_rate": 1.8268194002592903e-05,
      "loss": 0.5655,
      "step": 6221
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.499045491218567,
      "learning_rate": 1.8267565914466225e-05,
      "loss": 0.6516,
      "step": 6222
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5952277183532715,
      "learning_rate": 1.8266937723264697e-05,
      "loss": 0.7126,
      "step": 6223
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6963533163070679,
      "learning_rate": 1.8266309428996153e-05,
      "loss": 0.7615,
      "step": 6224
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6649807691574097,
      "learning_rate": 1.826568103166843e-05,
      "loss": 0.7021,
      "step": 6225
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0373075008392334,
      "learning_rate": 1.8265052531289355e-05,
      "loss": 0.6269,
      "step": 6226
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5319721698760986,
      "learning_rate": 1.8264423927866772e-05,
      "loss": 0.6548,
      "step": 6227
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6283156871795654,
      "learning_rate": 1.8263795221408513e-05,
      "loss": 0.6673,
      "step": 6228
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8085793256759644,
      "learning_rate": 1.8263166411922414e-05,
      "loss": 0.8667,
      "step": 6229
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5368225574493408,
      "learning_rate": 1.826253749941632e-05,
      "loss": 0.6722,
      "step": 6230
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6471785306930542,
      "learning_rate": 1.826190848389807e-05,
      "loss": 0.8071,
      "step": 6231
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.496606707572937,
      "learning_rate": 1.826127936537551e-05,
      "loss": 0.8305,
      "step": 6232
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.0030503273010254,
      "learning_rate": 1.8260650143856474e-05,
      "loss": 0.6571,
      "step": 6233
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.9220857620239258,
      "learning_rate": 1.8260020819348813e-05,
      "loss": 0.6985,
      "step": 6234
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5381511449813843,
      "learning_rate": 1.825939139186037e-05,
      "loss": 0.7629,
      "step": 6235
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4804681539535522,
      "learning_rate": 1.8258761861399e-05,
      "loss": 0.8199,
      "step": 6236
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9564123153686523,
      "learning_rate": 1.8258132227972543e-05,
      "loss": 0.7228,
      "step": 6237
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6739492416381836,
      "learning_rate": 1.825750249158885e-05,
      "loss": 0.823,
      "step": 6238
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6591029167175293,
      "learning_rate": 1.8256872652255778e-05,
      "loss": 0.7642,
      "step": 6239
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4459999799728394,
      "learning_rate": 1.8256242709981175e-05,
      "loss": 0.8713,
      "step": 6240
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6526111364364624,
      "learning_rate": 1.825561266477289e-05,
      "loss": 0.713,
      "step": 6241
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.73427152633667,
      "learning_rate": 1.825498251663879e-05,
      "loss": 0.8262,
      "step": 6242
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.509489893913269,
      "learning_rate": 1.825435226558672e-05,
      "loss": 0.7014,
      "step": 6243
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.60771906375885,
      "learning_rate": 1.8253721911624548e-05,
      "loss": 0.6446,
      "step": 6244
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0092036724090576,
      "learning_rate": 1.8253091454760123e-05,
      "loss": 0.8739,
      "step": 6245
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.311420202255249,
      "learning_rate": 1.8252460895001306e-05,
      "loss": 0.8337,
      "step": 6246
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7783622741699219,
      "learning_rate": 1.8251830232355966e-05,
      "loss": 0.7579,
      "step": 6247
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.443211317062378,
      "learning_rate": 1.825119946683196e-05,
      "loss": 0.6999,
      "step": 6248
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8979755640029907,
      "learning_rate": 1.8250568598437153e-05,
      "loss": 0.8525,
      "step": 6249
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6180826425552368,
      "learning_rate": 1.824993762717941e-05,
      "loss": 0.7129,
      "step": 6250
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6177806854248047,
      "learning_rate": 1.82493065530666e-05,
      "loss": 0.5985,
      "step": 6251
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3570339679718018,
      "learning_rate": 1.8248675376106588e-05,
      "loss": 0.6419,
      "step": 6252
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.710835337638855,
      "learning_rate": 1.8248044096307245e-05,
      "loss": 0.7666,
      "step": 6253
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3419121503829956,
      "learning_rate": 1.824741271367644e-05,
      "loss": 0.7526,
      "step": 6254
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3121548891067505,
      "learning_rate": 1.824678122822204e-05,
      "loss": 0.7451,
      "step": 6255
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9377416372299194,
      "learning_rate": 1.824614963995193e-05,
      "loss": 0.6904,
      "step": 6256
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6268442869186401,
      "learning_rate": 1.8245517948873975e-05,
      "loss": 0.7314,
      "step": 6257
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6860204935073853,
      "learning_rate": 1.8244886154996052e-05,
      "loss": 0.5993,
      "step": 6258
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3633365631103516,
      "learning_rate": 1.8244254258326043e-05,
      "loss": 0.6944,
      "step": 6259
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4612308740615845,
      "learning_rate": 1.8243622258871815e-05,
      "loss": 0.8216,
      "step": 6260
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5521187782287598,
      "learning_rate": 1.824299015664126e-05,
      "loss": 0.6401,
      "step": 6261
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.68271803855896,
      "learning_rate": 1.824235795164225e-05,
      "loss": 0.762,
      "step": 6262
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6667232513427734,
      "learning_rate": 1.824172564388267e-05,
      "loss": 0.7833,
      "step": 6263
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0566020011901855,
      "learning_rate": 1.8241093233370402e-05,
      "loss": 0.8401,
      "step": 6264
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4845821857452393,
      "learning_rate": 1.8240460720113334e-05,
      "loss": 0.7332,
      "step": 6265
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4623769521713257,
      "learning_rate": 1.8239828104119348e-05,
      "loss": 0.8175,
      "step": 6266
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5723426342010498,
      "learning_rate": 1.823919538539633e-05,
      "loss": 0.7726,
      "step": 6267
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.751602292060852,
      "learning_rate": 1.8238562563952173e-05,
      "loss": 0.7933,
      "step": 6268
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.64458429813385,
      "learning_rate": 1.8237929639794766e-05,
      "loss": 0.8292,
      "step": 6269
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5071179866790771,
      "learning_rate": 1.8237296612931995e-05,
      "loss": 0.8964,
      "step": 6270
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4199645519256592,
      "learning_rate": 1.823666348337176e-05,
      "loss": 0.7115,
      "step": 6271
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8336623907089233,
      "learning_rate": 1.8236030251121947e-05,
      "loss": 0.7914,
      "step": 6272
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3951011896133423,
      "learning_rate": 1.823539691619045e-05,
      "loss": 0.707,
      "step": 6273
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.876989722251892,
      "learning_rate": 1.823476347858517e-05,
      "loss": 0.7533,
      "step": 6274
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4557467699050903,
      "learning_rate": 1.8234129938314007e-05,
      "loss": 0.7542,
      "step": 6275
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.546616792678833,
      "learning_rate": 1.8233496295384854e-05,
      "loss": 0.6422,
      "step": 6276
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4185608625411987,
      "learning_rate": 1.823286254980561e-05,
      "loss": 0.8866,
      "step": 6277
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.541373372077942,
      "learning_rate": 1.823222870158418e-05,
      "loss": 0.7491,
      "step": 6278
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.2101097106933594,
      "learning_rate": 1.8231594750728466e-05,
      "loss": 0.8219,
      "step": 6279
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.622296929359436,
      "learning_rate": 1.8230960697246368e-05,
      "loss": 0.7446,
      "step": 6280
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3676360845565796,
      "learning_rate": 1.8230326541145794e-05,
      "loss": 0.7592,
      "step": 6281
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6205670833587646,
      "learning_rate": 1.8229692282434653e-05,
      "loss": 0.6555,
      "step": 6282
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4798097610473633,
      "learning_rate": 1.822905792112085e-05,
      "loss": 0.4861,
      "step": 6283
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.031388521194458,
      "learning_rate": 1.8228423457212287e-05,
      "loss": 0.9504,
      "step": 6284
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.4206323623657227,
      "learning_rate": 1.8227788890716885e-05,
      "loss": 0.6556,
      "step": 6285
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.398790955543518,
      "learning_rate": 1.822715422164255e-05,
      "loss": 0.642,
      "step": 6286
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7860411405563354,
      "learning_rate": 1.8226519449997195e-05,
      "loss": 0.7619,
      "step": 6287
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9795470237731934,
      "learning_rate": 1.8225884575788736e-05,
      "loss": 0.8358,
      "step": 6288
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5983365774154663,
      "learning_rate": 1.8225249599025082e-05,
      "loss": 0.8193,
      "step": 6289
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.926903486251831,
      "learning_rate": 1.8224614519714158e-05,
      "loss": 0.7001,
      "step": 6290
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6065744161605835,
      "learning_rate": 1.8223979337863878e-05,
      "loss": 0.7277,
      "step": 6291
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4983106851577759,
      "learning_rate": 1.822334405348216e-05,
      "loss": 0.5617,
      "step": 6292
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6344577074050903,
      "learning_rate": 1.8222708666576924e-05,
      "loss": 0.8086,
      "step": 6293
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4741076231002808,
      "learning_rate": 1.8222073177156093e-05,
      "loss": 0.6841,
      "step": 6294
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4112244844436646,
      "learning_rate": 1.8221437585227592e-05,
      "loss": 0.7007,
      "step": 6295
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5515828132629395,
      "learning_rate": 1.822080189079934e-05,
      "loss": 0.6278,
      "step": 6296
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.656404972076416,
      "learning_rate": 1.8220166093879266e-05,
      "loss": 0.715,
      "step": 6297
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.3430874347686768,
      "learning_rate": 1.8219530194475298e-05,
      "loss": 0.7242,
      "step": 6298
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5888084173202515,
      "learning_rate": 1.821889419259536e-05,
      "loss": 0.8022,
      "step": 6299
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3811280727386475,
      "learning_rate": 1.821825808824738e-05,
      "loss": 0.7207,
      "step": 6300
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3236255645751953,
      "learning_rate": 1.8217621881439294e-05,
      "loss": 0.7101,
      "step": 6301
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3632175922393799,
      "learning_rate": 1.8216985572179034e-05,
      "loss": 0.7363,
      "step": 6302
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.460252046585083,
      "learning_rate": 1.821634916047453e-05,
      "loss": 0.6371,
      "step": 6303
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1775249242782593,
      "learning_rate": 1.821571264633372e-05,
      "loss": 0.6727,
      "step": 6304
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5089294910430908,
      "learning_rate": 1.821507602976453e-05,
      "loss": 0.7044,
      "step": 6305
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4861465692520142,
      "learning_rate": 1.8214439310774907e-05,
      "loss": 0.7918,
      "step": 6306
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4726365804672241,
      "learning_rate": 1.8213802489372788e-05,
      "loss": 0.6797,
      "step": 6307
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7284327745437622,
      "learning_rate": 1.8213165565566106e-05,
      "loss": 0.6595,
      "step": 6308
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5386524200439453,
      "learning_rate": 1.821252853936281e-05,
      "loss": 0.7132,
      "step": 6309
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3911222219467163,
      "learning_rate": 1.8211891410770836e-05,
      "loss": 0.7267,
      "step": 6310
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5282644033432007,
      "learning_rate": 1.8211254179798132e-05,
      "loss": 0.8757,
      "step": 6311
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5824637413024902,
      "learning_rate": 1.821061684645264e-05,
      "loss": 0.7141,
      "step": 6312
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3700624704360962,
      "learning_rate": 1.8209979410742305e-05,
      "loss": 0.764,
      "step": 6313
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8397859334945679,
      "learning_rate": 1.820934187267507e-05,
      "loss": 0.7595,
      "step": 6314
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6589739322662354,
      "learning_rate": 1.82087042322589e-05,
      "loss": 0.6732,
      "step": 6315
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6752538681030273,
      "learning_rate": 1.8208066489501722e-05,
      "loss": 0.8087,
      "step": 6316
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.834212303161621,
      "learning_rate": 1.8207428644411504e-05,
      "loss": 0.729,
      "step": 6317
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.8095695972442627,
      "learning_rate": 1.820679069699619e-05,
      "loss": 0.6595,
      "step": 6318
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4555593729019165,
      "learning_rate": 1.820615264726374e-05,
      "loss": 0.8088,
      "step": 6319
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.682908058166504,
      "learning_rate": 1.8205514495222104e-05,
      "loss": 0.756,
      "step": 6320
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.55364191532135,
      "learning_rate": 1.8204876240879236e-05,
      "loss": 0.7658,
      "step": 6321
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5345524549484253,
      "learning_rate": 1.82042378842431e-05,
      "loss": 0.7177,
      "step": 6322
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4502824544906616,
      "learning_rate": 1.8203599425321647e-05,
      "loss": 0.7173,
      "step": 6323
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7507671117782593,
      "learning_rate": 1.8202960864122842e-05,
      "loss": 0.7585,
      "step": 6324
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7129005193710327,
      "learning_rate": 1.820232220065465e-05,
      "loss": 0.7189,
      "step": 6325
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7089288234710693,
      "learning_rate": 1.820168343492502e-05,
      "loss": 0.8391,
      "step": 6326
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4900124073028564,
      "learning_rate": 1.820104456694193e-05,
      "loss": 0.7967,
      "step": 6327
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5781188011169434,
      "learning_rate": 1.820040559671334e-05,
      "loss": 0.5389,
      "step": 6328
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3327139616012573,
      "learning_rate": 1.8199766524247212e-05,
      "loss": 0.8063,
      "step": 6329
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.668070912361145,
      "learning_rate": 1.8199127349551518e-05,
      "loss": 0.6803,
      "step": 6330
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6592974662780762,
      "learning_rate": 1.8198488072634223e-05,
      "loss": 0.903,
      "step": 6331
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.0988407135009766,
      "learning_rate": 1.8197848693503304e-05,
      "loss": 0.8137,
      "step": 6332
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7457243204116821,
      "learning_rate": 1.8197209212166727e-05,
      "loss": 0.6815,
      "step": 6333
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3680154085159302,
      "learning_rate": 1.8196569628632468e-05,
      "loss": 0.6928,
      "step": 6334
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7246086597442627,
      "learning_rate": 1.81959299429085e-05,
      "loss": 0.7588,
      "step": 6335
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6021294593811035,
      "learning_rate": 1.8195290155002792e-05,
      "loss": 0.7497,
      "step": 6336
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.607207179069519,
      "learning_rate": 1.819465026492333e-05,
      "loss": 0.7914,
      "step": 6337
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.789633870124817,
      "learning_rate": 1.8194010272678086e-05,
      "loss": 0.7118,
      "step": 6338
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3579424619674683,
      "learning_rate": 1.819337017827504e-05,
      "loss": 0.834,
      "step": 6339
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2725460529327393,
      "learning_rate": 1.8192729981722172e-05,
      "loss": 0.495,
      "step": 6340
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2960041761398315,
      "learning_rate": 1.8192089683027465e-05,
      "loss": 0.7831,
      "step": 6341
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6055669784545898,
      "learning_rate": 1.81914492821989e-05,
      "loss": 0.7287,
      "step": 6342
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.655681848526001,
      "learning_rate": 1.8190808779244467e-05,
      "loss": 0.8972,
      "step": 6343
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5970463752746582,
      "learning_rate": 1.8190168174172142e-05,
      "loss": 0.6742,
      "step": 6344
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4874707460403442,
      "learning_rate": 1.8189527466989918e-05,
      "loss": 0.6951,
      "step": 6345
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2764226198196411,
      "learning_rate": 1.8188886657705782e-05,
      "loss": 0.6695,
      "step": 6346
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7931615114212036,
      "learning_rate": 1.818824574632772e-05,
      "loss": 0.7045,
      "step": 6347
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.332656979560852,
      "learning_rate": 1.8187604732863727e-05,
      "loss": 0.7926,
      "step": 6348
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4937825202941895,
      "learning_rate": 1.8186963617321792e-05,
      "loss": 0.8317,
      "step": 6349
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.358942985534668,
      "learning_rate": 1.8186322399709907e-05,
      "loss": 0.6944,
      "step": 6350
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5357520580291748,
      "learning_rate": 1.818568108003607e-05,
      "loss": 0.7331,
      "step": 6351
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2340824604034424,
      "learning_rate": 1.8185039658308273e-05,
      "loss": 0.6313,
      "step": 6352
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5250296592712402,
      "learning_rate": 1.8184398134534517e-05,
      "loss": 0.6186,
      "step": 6353
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8111038208007812,
      "learning_rate": 1.8183756508722797e-05,
      "loss": 0.8131,
      "step": 6354
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7872484922409058,
      "learning_rate": 1.818311478088111e-05,
      "loss": 0.7717,
      "step": 6355
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5324583053588867,
      "learning_rate": 1.8182472951017464e-05,
      "loss": 0.8311,
      "step": 6356
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.621275782585144,
      "learning_rate": 1.8181831019139852e-05,
      "loss": 0.7591,
      "step": 6357
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7030701637268066,
      "learning_rate": 1.8181188985256285e-05,
      "loss": 0.5608,
      "step": 6358
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6062815189361572,
      "learning_rate": 1.8180546849374762e-05,
      "loss": 0.6208,
      "step": 6359
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4064501523971558,
      "learning_rate": 1.8179904611503293e-05,
      "loss": 0.8198,
      "step": 6360
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.415242075920105,
      "learning_rate": 1.8179262271649883e-05,
      "loss": 0.5465,
      "step": 6361
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8967015743255615,
      "learning_rate": 1.817861982982254e-05,
      "loss": 0.7517,
      "step": 6362
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2749489545822144,
      "learning_rate": 1.8177977286029272e-05,
      "loss": 0.7781,
      "step": 6363
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.824358344078064,
      "learning_rate": 1.8177334640278096e-05,
      "loss": 0.7875,
      "step": 6364
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.631138563156128,
      "learning_rate": 1.8176691892577015e-05,
      "loss": 0.6612,
      "step": 6365
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.083495616912842,
      "learning_rate": 1.817604904293405e-05,
      "loss": 0.8198,
      "step": 6366
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2734732627868652,
      "learning_rate": 1.817540609135721e-05,
      "loss": 0.8394,
      "step": 6367
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.515450358390808,
      "learning_rate": 1.817476303785452e-05,
      "loss": 0.9079,
      "step": 6368
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7034040689468384,
      "learning_rate": 1.8174119882433983e-05,
      "loss": 0.6744,
      "step": 6369
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.632537603378296,
      "learning_rate": 1.817347662510363e-05,
      "loss": 0.6504,
      "step": 6370
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6915123462677002,
      "learning_rate": 1.8172833265871476e-05,
      "loss": 0.7539,
      "step": 6371
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6815353631973267,
      "learning_rate": 1.817218980474554e-05,
      "loss": 0.6773,
      "step": 6372
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5009225606918335,
      "learning_rate": 1.8171546241733845e-05,
      "loss": 0.8265,
      "step": 6373
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5619111061096191,
      "learning_rate": 1.817090257684442e-05,
      "loss": 0.7448,
      "step": 6374
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4901584386825562,
      "learning_rate": 1.8170258810085285e-05,
      "loss": 0.7762,
      "step": 6375
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.700998306274414,
      "learning_rate": 1.8169614941464468e-05,
      "loss": 0.7935,
      "step": 6376
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8172109127044678,
      "learning_rate": 1.816897097098999e-05,
      "loss": 0.8362,
      "step": 6377
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8798516988754272,
      "learning_rate": 1.8168326898669886e-05,
      "loss": 0.8365,
      "step": 6378
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5981587171554565,
      "learning_rate": 1.816768272451219e-05,
      "loss": 0.7769,
      "step": 6379
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3806514739990234,
      "learning_rate": 1.8167038448524926e-05,
      "loss": 0.6953,
      "step": 6380
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.309677243232727,
      "learning_rate": 1.8166394070716124e-05,
      "loss": 0.8101,
      "step": 6381
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4262075424194336,
      "learning_rate": 1.8165749591093823e-05,
      "loss": 0.7497,
      "step": 6382
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8309335708618164,
      "learning_rate": 1.8165105009666062e-05,
      "loss": 0.686,
      "step": 6383
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3857148885726929,
      "learning_rate": 1.8164460326440866e-05,
      "loss": 0.6592,
      "step": 6384
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5603679418563843,
      "learning_rate": 1.816381554142628e-05,
      "loss": 0.7947,
      "step": 6385
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3768339157104492,
      "learning_rate": 1.8163170654630345e-05,
      "loss": 0.7746,
      "step": 6386
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8160808086395264,
      "learning_rate": 1.8162525666061098e-05,
      "loss": 0.7745,
      "step": 6387
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5754272937774658,
      "learning_rate": 1.8161880575726575e-05,
      "loss": 0.7599,
      "step": 6388
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.742780089378357,
      "learning_rate": 1.8161235383634826e-05,
      "loss": 0.737,
      "step": 6389
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7978850603103638,
      "learning_rate": 1.816059008979389e-05,
      "loss": 0.671,
      "step": 6390
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7989952564239502,
      "learning_rate": 1.815994469421182e-05,
      "loss": 0.6676,
      "step": 6391
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7834362983703613,
      "learning_rate": 1.8159299196896652e-05,
      "loss": 0.8043,
      "step": 6392
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.605224847793579,
      "learning_rate": 1.8158653597856438e-05,
      "loss": 0.5431,
      "step": 6393
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4032907485961914,
      "learning_rate": 1.8158007897099232e-05,
      "loss": 0.8601,
      "step": 6394
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4184681177139282,
      "learning_rate": 1.8157362094633075e-05,
      "loss": 0.6024,
      "step": 6395
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4213392734527588,
      "learning_rate": 1.8156716190466026e-05,
      "loss": 0.6258,
      "step": 6396
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4233633279800415,
      "learning_rate": 1.815607018460613e-05,
      "loss": 0.7161,
      "step": 6397
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3870948553085327,
      "learning_rate": 1.8155424077061453e-05,
      "loss": 0.6735,
      "step": 6398
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4378331899642944,
      "learning_rate": 1.815477786784004e-05,
      "loss": 0.7347,
      "step": 6399
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3135794401168823,
      "learning_rate": 1.8154131556949948e-05,
      "loss": 0.7076,
      "step": 6400
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4534358978271484,
      "learning_rate": 1.815348514439924e-05,
      "loss": 0.7856,
      "step": 6401
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.335323691368103,
      "learning_rate": 1.815283863019597e-05,
      "loss": 0.6487,
      "step": 6402
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.562559723854065,
      "learning_rate": 1.81521920143482e-05,
      "loss": 0.6846,
      "step": 6403
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.651015043258667,
      "learning_rate": 1.8151545296863995e-05,
      "loss": 0.8584,
      "step": 6404
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4305052757263184,
      "learning_rate": 1.8150898477751414e-05,
      "loss": 0.7577,
      "step": 6405
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5046508312225342,
      "learning_rate": 1.8150251557018523e-05,
      "loss": 0.7365,
      "step": 6406
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.574226975440979,
      "learning_rate": 1.8149604534673387e-05,
      "loss": 0.7897,
      "step": 6407
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4672449827194214,
      "learning_rate": 1.814895741072407e-05,
      "loss": 0.682,
      "step": 6408
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6384690999984741,
      "learning_rate": 1.8148310185178646e-05,
      "loss": 0.7077,
      "step": 6409
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6252484321594238,
      "learning_rate": 1.814766285804518e-05,
      "loss": 0.751,
      "step": 6410
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3411921262741089,
      "learning_rate": 1.8147015429331743e-05,
      "loss": 0.6676,
      "step": 6411
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8014532327651978,
      "learning_rate": 1.8146367899046407e-05,
      "loss": 0.7533,
      "step": 6412
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7393603324890137,
      "learning_rate": 1.814572026719724e-05,
      "loss": 0.7035,
      "step": 6413
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1915184259414673,
      "learning_rate": 1.8145072533792327e-05,
      "loss": 0.721,
      "step": 6414
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7353956699371338,
      "learning_rate": 1.8144424698839733e-05,
      "loss": 0.7087,
      "step": 6415
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5420961380004883,
      "learning_rate": 1.8143776762347544e-05,
      "loss": 0.5853,
      "step": 6416
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.265135645866394,
      "learning_rate": 1.8143128724323832e-05,
      "loss": 0.6426,
      "step": 6417
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4875260591506958,
      "learning_rate": 1.8142480584776673e-05,
      "loss": 0.6547,
      "step": 6418
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7142889499664307,
      "learning_rate": 1.814183234371416e-05,
      "loss": 0.6833,
      "step": 6419
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5636464357376099,
      "learning_rate": 1.814118400114436e-05,
      "loss": 0.7634,
      "step": 6420
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.742129921913147,
      "learning_rate": 1.814053555707537e-05,
      "loss": 0.6454,
      "step": 6421
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.503066062927246,
      "learning_rate": 1.8139887011515264e-05,
      "loss": 0.8652,
      "step": 6422
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.9097795486450195,
      "learning_rate": 1.8139238364472133e-05,
      "loss": 0.8155,
      "step": 6423
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4470455646514893,
      "learning_rate": 1.8138589615954058e-05,
      "loss": 0.6478,
      "step": 6424
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.59260094165802,
      "learning_rate": 1.8137940765969137e-05,
      "loss": 0.929,
      "step": 6425
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.8239622116088867,
      "learning_rate": 1.813729181452545e-05,
      "loss": 0.8816,
      "step": 6426
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6154288053512573,
      "learning_rate": 1.8136642761631094e-05,
      "loss": 0.8339,
      "step": 6427
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6574527025222778,
      "learning_rate": 1.813599360729416e-05,
      "loss": 0.626,
      "step": 6428
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.733788251876831,
      "learning_rate": 1.8135344351522738e-05,
      "loss": 0.6741,
      "step": 6429
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.542554259300232,
      "learning_rate": 1.8134694994324924e-05,
      "loss": 0.7605,
      "step": 6430
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4305518865585327,
      "learning_rate": 1.8134045535708816e-05,
      "loss": 0.7796,
      "step": 6431
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5750850439071655,
      "learning_rate": 1.8133395975682508e-05,
      "loss": 0.6828,
      "step": 6432
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.741775631904602,
      "learning_rate": 1.81327463142541e-05,
      "loss": 0.8563,
      "step": 6433
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6377997398376465,
      "learning_rate": 1.8132096551431694e-05,
      "loss": 0.7916,
      "step": 6434
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6965339183807373,
      "learning_rate": 1.8131446687223383e-05,
      "loss": 0.6467,
      "step": 6435
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7968339920043945,
      "learning_rate": 1.8130796721637277e-05,
      "loss": 0.8683,
      "step": 6436
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5368667840957642,
      "learning_rate": 1.8130146654681476e-05,
      "loss": 0.8147,
      "step": 6437
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.634216547012329,
      "learning_rate": 1.8129496486364084e-05,
      "loss": 0.7512,
      "step": 6438
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3707990646362305,
      "learning_rate": 1.812884621669321e-05,
      "loss": 0.8131,
      "step": 6439
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9151427745819092,
      "learning_rate": 1.8128195845676956e-05,
      "loss": 0.6141,
      "step": 6440
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.442714810371399,
      "learning_rate": 1.8127545373323438e-05,
      "loss": 0.7333,
      "step": 6441
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5133333206176758,
      "learning_rate": 1.8126894799640757e-05,
      "loss": 0.7218,
      "step": 6442
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7246527671813965,
      "learning_rate": 1.8126244124637032e-05,
      "loss": 0.9436,
      "step": 6443
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9278082847595215,
      "learning_rate": 1.8125593348320367e-05,
      "loss": 0.7557,
      "step": 6444
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.055614948272705,
      "learning_rate": 1.8124942470698883e-05,
      "loss": 0.8884,
      "step": 6445
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4996488094329834,
      "learning_rate": 1.8124291491780692e-05,
      "loss": 0.8043,
      "step": 6446
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5370279550552368,
      "learning_rate": 1.8123640411573912e-05,
      "loss": 0.7096,
      "step": 6447
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.577713131904602,
      "learning_rate": 1.8122989230086653e-05,
      "loss": 0.7358,
      "step": 6448
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.273985505104065,
      "learning_rate": 1.8122337947327038e-05,
      "loss": 0.7213,
      "step": 6449
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.573725700378418,
      "learning_rate": 1.812168656330319e-05,
      "loss": 0.7083,
      "step": 6450
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6324641704559326,
      "learning_rate": 1.8121035078023224e-05,
      "loss": 0.782,
      "step": 6451
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.646737813949585,
      "learning_rate": 1.812038349149527e-05,
      "loss": 0.605,
      "step": 6452
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.197508454322815,
      "learning_rate": 1.8119731803727446e-05,
      "loss": 0.6007,
      "step": 6453
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1480382680892944,
      "learning_rate": 1.8119080014727876e-05,
      "loss": 0.5851,
      "step": 6454
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.162947177886963,
      "learning_rate": 1.811842812450469e-05,
      "loss": 0.6607,
      "step": 6455
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4502354860305786,
      "learning_rate": 1.8117776133066015e-05,
      "loss": 0.5737,
      "step": 6456
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.455335259437561,
      "learning_rate": 1.8117124040419974e-05,
      "loss": 0.9453,
      "step": 6457
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3214497566223145,
      "learning_rate": 1.81164718465747e-05,
      "loss": 0.6862,
      "step": 6458
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8116260766983032,
      "learning_rate": 1.811581955153833e-05,
      "loss": 0.6869,
      "step": 6459
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5469671487808228,
      "learning_rate": 1.811516715531899e-05,
      "loss": 0.7156,
      "step": 6460
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5918529033660889,
      "learning_rate": 1.8114514657924814e-05,
      "loss": 0.7831,
      "step": 6461
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.460286259651184,
      "learning_rate": 1.811386205936394e-05,
      "loss": 0.6127,
      "step": 6462
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3090953826904297,
      "learning_rate": 1.81132093596445e-05,
      "loss": 0.7121,
      "step": 6463
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3183672428131104,
      "learning_rate": 1.8112556558774635e-05,
      "loss": 0.7377,
      "step": 6464
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.551743984222412,
      "learning_rate": 1.811190365676248e-05,
      "loss": 0.6524,
      "step": 6465
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7125567197799683,
      "learning_rate": 1.8111250653616183e-05,
      "loss": 0.6993,
      "step": 6466
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4590318202972412,
      "learning_rate": 1.8110597549343873e-05,
      "loss": 0.7766,
      "step": 6467
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6130056381225586,
      "learning_rate": 1.81099443439537e-05,
      "loss": 0.7171,
      "step": 6468
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.0004141330718994,
      "learning_rate": 1.810929103745381e-05,
      "loss": 0.8226,
      "step": 6469
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4776614904403687,
      "learning_rate": 1.8108637629852347e-05,
      "loss": 0.7541,
      "step": 6470
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.411687970161438,
      "learning_rate": 1.810798412115745e-05,
      "loss": 0.7636,
      "step": 6471
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3649126291275024,
      "learning_rate": 1.8107330511377274e-05,
      "loss": 0.5757,
      "step": 6472
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3287773132324219,
      "learning_rate": 1.8106676800519968e-05,
      "loss": 0.745,
      "step": 6473
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7057486772537231,
      "learning_rate": 1.8106022988593673e-05,
      "loss": 0.7105,
      "step": 6474
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7448076009750366,
      "learning_rate": 1.8105369075606554e-05,
      "loss": 0.7131,
      "step": 6475
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4883708953857422,
      "learning_rate": 1.8104715061566753e-05,
      "loss": 0.7338,
      "step": 6476
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.905068039894104,
      "learning_rate": 1.8104060946482427e-05,
      "loss": 0.8856,
      "step": 6477
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6625754833221436,
      "learning_rate": 1.810340673036173e-05,
      "loss": 0.7059,
      "step": 6478
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7794194221496582,
      "learning_rate": 1.8102752413212824e-05,
      "loss": 0.6786,
      "step": 6479
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6173062324523926,
      "learning_rate": 1.8102097995043863e-05,
      "loss": 0.7163,
      "step": 6480
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4258947372436523,
      "learning_rate": 1.8101443475863e-05,
      "loss": 0.8015,
      "step": 6481
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7574044466018677,
      "learning_rate": 1.8100788855678404e-05,
      "loss": 0.7076,
      "step": 6482
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6326650381088257,
      "learning_rate": 1.8100134134498233e-05,
      "loss": 0.8325,
      "step": 6483
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7693827152252197,
      "learning_rate": 1.8099479312330647e-05,
      "loss": 0.8485,
      "step": 6484
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6024196147918701,
      "learning_rate": 1.8098824389183815e-05,
      "loss": 0.6775,
      "step": 6485
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9710383415222168,
      "learning_rate": 1.80981693650659e-05,
      "loss": 0.6277,
      "step": 6486
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3399741649627686,
      "learning_rate": 1.8097514239985067e-05,
      "loss": 0.7292,
      "step": 6487
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3639477491378784,
      "learning_rate": 1.809685901394948e-05,
      "loss": 0.7456,
      "step": 6488
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5555496215820312,
      "learning_rate": 1.8096203686967318e-05,
      "loss": 0.6693,
      "step": 6489
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5539659261703491,
      "learning_rate": 1.8095548259046746e-05,
      "loss": 0.6502,
      "step": 6490
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3427774906158447,
      "learning_rate": 1.8094892730195933e-05,
      "loss": 0.6964,
      "step": 6491
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4432164430618286,
      "learning_rate": 1.8094237100423052e-05,
      "loss": 0.7174,
      "step": 6492
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.356841802597046,
      "learning_rate": 1.8093581369736285e-05,
      "loss": 0.6358,
      "step": 6493
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5063412189483643,
      "learning_rate": 1.80929255381438e-05,
      "loss": 0.72,
      "step": 6494
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4732894897460938,
      "learning_rate": 1.809226960565377e-05,
      "loss": 0.7937,
      "step": 6495
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6022273302078247,
      "learning_rate": 1.809161357227438e-05,
      "loss": 0.6826,
      "step": 6496
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5319122076034546,
      "learning_rate": 1.8090957438013808e-05,
      "loss": 0.7079,
      "step": 6497
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6320101022720337,
      "learning_rate": 1.8090301202880232e-05,
      "loss": 0.8259,
      "step": 6498
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6709412336349487,
      "learning_rate": 1.8089644866881834e-05,
      "loss": 0.744,
      "step": 6499
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.026352643966675,
      "learning_rate": 1.8088988430026796e-05,
      "loss": 0.7332,
      "step": 6500
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6592159271240234,
      "learning_rate": 1.8088331892323307e-05,
      "loss": 0.6339,
      "step": 6501
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9370986223220825,
      "learning_rate": 1.8087675253779543e-05,
      "loss": 0.7169,
      "step": 6502
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6827932596206665,
      "learning_rate": 1.80870185144037e-05,
      "loss": 0.7362,
      "step": 6503
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.394022822380066,
      "learning_rate": 1.808636167420396e-05,
      "loss": 0.8481,
      "step": 6504
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.777808427810669,
      "learning_rate": 1.8085704733188514e-05,
      "loss": 0.8414,
      "step": 6505
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6853703260421753,
      "learning_rate": 1.8085047691365553e-05,
      "loss": 0.7343,
      "step": 6506
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6351878643035889,
      "learning_rate": 1.808439054874327e-05,
      "loss": 0.6101,
      "step": 6507
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.565271258354187,
      "learning_rate": 1.8083733305329853e-05,
      "loss": 0.7353,
      "step": 6508
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4710092544555664,
      "learning_rate": 1.8083075961133498e-05,
      "loss": 0.6957,
      "step": 6509
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.66428542137146,
      "learning_rate": 1.80824185161624e-05,
      "loss": 0.7457,
      "step": 6510
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9517885446548462,
      "learning_rate": 1.808176097042476e-05,
      "loss": 0.7789,
      "step": 6511
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5500707626342773,
      "learning_rate": 1.808110332392877e-05,
      "loss": 0.7058,
      "step": 6512
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.334593415260315,
      "learning_rate": 1.8080445576682636e-05,
      "loss": 0.7672,
      "step": 6513
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4742995500564575,
      "learning_rate": 1.807978772869455e-05,
      "loss": 0.7392,
      "step": 6514
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.9973719120025635,
      "learning_rate": 1.8079129779972716e-05,
      "loss": 0.6819,
      "step": 6515
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5095661878585815,
      "learning_rate": 1.8078471730525345e-05,
      "loss": 0.723,
      "step": 6516
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5027433633804321,
      "learning_rate": 1.807781358036063e-05,
      "loss": 0.7589,
      "step": 6517
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.484889268875122,
      "learning_rate": 1.807715532948678e-05,
      "loss": 0.6595,
      "step": 6518
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6219843626022339,
      "learning_rate": 1.8076496977912007e-05,
      "loss": 0.8825,
      "step": 6519
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6343637704849243,
      "learning_rate": 1.8075838525644515e-05,
      "loss": 0.6533,
      "step": 6520
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2992132902145386,
      "learning_rate": 1.8075179972692513e-05,
      "loss": 0.7597,
      "step": 6521
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3588123321533203,
      "learning_rate": 1.8074521319064208e-05,
      "loss": 0.8286,
      "step": 6522
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6203663349151611,
      "learning_rate": 1.8073862564767816e-05,
      "loss": 0.6278,
      "step": 6523
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.404577374458313,
      "learning_rate": 1.807320370981155e-05,
      "loss": 0.7889,
      "step": 6524
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8318569660186768,
      "learning_rate": 1.8072544754203622e-05,
      "loss": 0.9171,
      "step": 6525
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.422762393951416,
      "learning_rate": 1.807188569795225e-05,
      "loss": 0.6815,
      "step": 6526
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8124926090240479,
      "learning_rate": 1.8071226541065648e-05,
      "loss": 0.8083,
      "step": 6527
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4912419319152832,
      "learning_rate": 1.8070567283552037e-05,
      "loss": 0.6355,
      "step": 6528
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5677701234817505,
      "learning_rate": 1.8069907925419633e-05,
      "loss": 0.7438,
      "step": 6529
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2050520181655884,
      "learning_rate": 1.806924846667666e-05,
      "loss": 0.6687,
      "step": 6530
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7227888107299805,
      "learning_rate": 1.8068588907331335e-05,
      "loss": 0.7169,
      "step": 6531
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4757673740386963,
      "learning_rate": 1.8067929247391886e-05,
      "loss": 0.7083,
      "step": 6532
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2868891954421997,
      "learning_rate": 1.8067269486866533e-05,
      "loss": 0.7123,
      "step": 6533
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3848073482513428,
      "learning_rate": 1.80666096257635e-05,
      "loss": 0.7865,
      "step": 6534
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.56834077835083,
      "learning_rate": 1.8065949664091022e-05,
      "loss": 0.7326,
      "step": 6535
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.204037070274353,
      "learning_rate": 1.8065289601857324e-05,
      "loss": 0.7864,
      "step": 6536
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7889716625213623,
      "learning_rate": 1.806462943907063e-05,
      "loss": 0.7814,
      "step": 6537
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.43832528591156,
      "learning_rate": 1.8063969175739174e-05,
      "loss": 0.6524,
      "step": 6538
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.475639820098877,
      "learning_rate": 1.806330881187119e-05,
      "loss": 0.731,
      "step": 6539
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8214569091796875,
      "learning_rate": 1.8062648347474904e-05,
      "loss": 0.762,
      "step": 6540
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4866173267364502,
      "learning_rate": 1.806198778255856e-05,
      "loss": 0.6866,
      "step": 6541
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.408902645111084,
      "learning_rate": 1.8061327117130386e-05,
      "loss": 0.6453,
      "step": 6542
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8313747644424438,
      "learning_rate": 1.8060666351198622e-05,
      "loss": 0.6371,
      "step": 6543
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6841278076171875,
      "learning_rate": 1.80600054847715e-05,
      "loss": 0.7118,
      "step": 6544
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4255361557006836,
      "learning_rate": 1.805934451785727e-05,
      "loss": 0.7837,
      "step": 6545
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5688092708587646,
      "learning_rate": 1.8058683450464167e-05,
      "loss": 0.6678,
      "step": 6546
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6778159141540527,
      "learning_rate": 1.805802228260043e-05,
      "loss": 0.8152,
      "step": 6547
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6214954853057861,
      "learning_rate": 1.8057361014274305e-05,
      "loss": 0.5802,
      "step": 6548
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7076292037963867,
      "learning_rate": 1.8056699645494038e-05,
      "loss": 0.7076,
      "step": 6549
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3521264791488647,
      "learning_rate": 1.805603817626787e-05,
      "loss": 0.7101,
      "step": 6550
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3121119737625122,
      "learning_rate": 1.8055376606604052e-05,
      "loss": 0.8073,
      "step": 6551
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5815800428390503,
      "learning_rate": 1.805471493651083e-05,
      "loss": 0.6676,
      "step": 6552
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4980686902999878,
      "learning_rate": 1.8054053165996454e-05,
      "loss": 0.7629,
      "step": 6553
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.895704984664917,
      "learning_rate": 1.8053391295069173e-05,
      "loss": 0.7413,
      "step": 6554
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6946316957473755,
      "learning_rate": 1.8052729323737243e-05,
      "loss": 0.8243,
      "step": 6555
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4570746421813965,
      "learning_rate": 1.805206725200891e-05,
      "loss": 0.6622,
      "step": 6556
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.229232907295227,
      "learning_rate": 1.8051405079892437e-05,
      "loss": 0.5675,
      "step": 6557
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.445134162902832,
      "learning_rate": 1.8050742807396072e-05,
      "loss": 0.6726,
      "step": 6558
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3207536935806274,
      "learning_rate": 1.8050080434528075e-05,
      "loss": 0.7674,
      "step": 6559
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6307142972946167,
      "learning_rate": 1.8049417961296703e-05,
      "loss": 0.7531,
      "step": 6560
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6281497478485107,
      "learning_rate": 1.804875538771022e-05,
      "loss": 0.5267,
      "step": 6561
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4852423667907715,
      "learning_rate": 1.8048092713776876e-05,
      "loss": 0.632,
      "step": 6562
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5780837535858154,
      "learning_rate": 1.8047429939504942e-05,
      "loss": 0.6359,
      "step": 6563
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6635063886642456,
      "learning_rate": 1.804676706490268e-05,
      "loss": 0.7722,
      "step": 6564
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7507696151733398,
      "learning_rate": 1.8046104089978353e-05,
      "loss": 0.7954,
      "step": 6565
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7065476179122925,
      "learning_rate": 1.8045441014740224e-05,
      "loss": 0.931,
      "step": 6566
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3050237894058228,
      "learning_rate": 1.8044777839196562e-05,
      "loss": 0.7054,
      "step": 6567
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4613348245620728,
      "learning_rate": 1.8044114563355637e-05,
      "loss": 0.5648,
      "step": 6568
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7482140064239502,
      "learning_rate": 1.8043451187225717e-05,
      "loss": 0.7751,
      "step": 6569
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5960066318511963,
      "learning_rate": 1.804278771081507e-05,
      "loss": 0.7509,
      "step": 6570
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7468394041061401,
      "learning_rate": 1.8042124134131974e-05,
      "loss": 0.8592,
      "step": 6571
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8164803981781006,
      "learning_rate": 1.8041460457184696e-05,
      "loss": 0.6638,
      "step": 6572
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7077430486679077,
      "learning_rate": 1.804079667998151e-05,
      "loss": 0.7461,
      "step": 6573
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7024174928665161,
      "learning_rate": 1.804013280253069e-05,
      "loss": 0.7792,
      "step": 6574
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6240315437316895,
      "learning_rate": 1.8039468824840526e-05,
      "loss": 0.7642,
      "step": 6575
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.184397578239441,
      "learning_rate": 1.803880474691928e-05,
      "loss": 0.6758,
      "step": 6576
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.0589020252227783,
      "learning_rate": 1.803814056877524e-05,
      "loss": 0.882,
      "step": 6577
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3328747749328613,
      "learning_rate": 1.8037476290416683e-05,
      "loss": 0.7289,
      "step": 6578
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6384207010269165,
      "learning_rate": 1.8036811911851896e-05,
      "loss": 0.6531,
      "step": 6579
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.721506118774414,
      "learning_rate": 1.8036147433089156e-05,
      "loss": 0.6053,
      "step": 6580
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5880284309387207,
      "learning_rate": 1.803548285413675e-05,
      "loss": 0.8395,
      "step": 6581
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4210282564163208,
      "learning_rate": 1.803481817500296e-05,
      "loss": 0.6808,
      "step": 6582
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4253844022750854,
      "learning_rate": 1.8034153395696082e-05,
      "loss": 0.7643,
      "step": 6583
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4229472875595093,
      "learning_rate": 1.8033488516224396e-05,
      "loss": 0.7224,
      "step": 6584
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4739938974380493,
      "learning_rate": 1.8032823536596193e-05,
      "loss": 0.6894,
      "step": 6585
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6588994264602661,
      "learning_rate": 1.8032158456819764e-05,
      "loss": 0.7829,
      "step": 6586
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5120011568069458,
      "learning_rate": 1.80314932769034e-05,
      "loss": 0.7244,
      "step": 6587
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7835841178894043,
      "learning_rate": 1.8030827996855397e-05,
      "loss": 0.6943,
      "step": 6588
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4379103183746338,
      "learning_rate": 1.8030162616684043e-05,
      "loss": 0.6545,
      "step": 6589
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7494573593139648,
      "learning_rate": 1.8029497136397644e-05,
      "loss": 0.7821,
      "step": 6590
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.56136953830719,
      "learning_rate": 1.8028831556004482e-05,
      "loss": 0.6619,
      "step": 6591
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6480159759521484,
      "learning_rate": 1.8028165875512867e-05,
      "loss": 0.8414,
      "step": 6592
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.555359959602356,
      "learning_rate": 1.80275000949311e-05,
      "loss": 0.6805,
      "step": 6593
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4871195554733276,
      "learning_rate": 1.802683421426747e-05,
      "loss": 0.8587,
      "step": 6594
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4988737106323242,
      "learning_rate": 1.8026168233530286e-05,
      "loss": 0.6147,
      "step": 6595
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8043071031570435,
      "learning_rate": 1.802550215272785e-05,
      "loss": 0.7362,
      "step": 6596
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6862108707427979,
      "learning_rate": 1.802483597186847e-05,
      "loss": 0.8671,
      "step": 6597
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5451322793960571,
      "learning_rate": 1.8024169690960443e-05,
      "loss": 0.6728,
      "step": 6598
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3238539695739746,
      "learning_rate": 1.8023503310012086e-05,
      "loss": 0.5719,
      "step": 6599
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.482791543006897,
      "learning_rate": 1.80228368290317e-05,
      "loss": 0.7525,
      "step": 6600
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3903030157089233,
      "learning_rate": 1.8022170248027594e-05,
      "loss": 0.635,
      "step": 6601
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6922082901000977,
      "learning_rate": 1.802150356700808e-05,
      "loss": 0.7214,
      "step": 6602
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6241295337677002,
      "learning_rate": 1.8020836785981473e-05,
      "loss": 0.7702,
      "step": 6603
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6195205450057983,
      "learning_rate": 1.8020169904956085e-05,
      "loss": 0.5998,
      "step": 6604
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7541189193725586,
      "learning_rate": 1.8019502923940223e-05,
      "loss": 0.5648,
      "step": 6605
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5804225206375122,
      "learning_rate": 1.8018835842942212e-05,
      "loss": 0.8017,
      "step": 6606
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.519290566444397,
      "learning_rate": 1.8018168661970362e-05,
      "loss": 0.7923,
      "step": 6607
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4487720727920532,
      "learning_rate": 1.8017501381032995e-05,
      "loss": 0.8476,
      "step": 6608
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4536609649658203,
      "learning_rate": 1.801683400013843e-05,
      "loss": 0.5461,
      "step": 6609
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6130675077438354,
      "learning_rate": 1.8016166519294987e-05,
      "loss": 0.7296,
      "step": 6610
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9231935739517212,
      "learning_rate": 1.8015498938510986e-05,
      "loss": 0.7779,
      "step": 6611
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7982158660888672,
      "learning_rate": 1.801483125779475e-05,
      "loss": 0.8072,
      "step": 6612
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5557576417922974,
      "learning_rate": 1.8014163477154608e-05,
      "loss": 0.8133,
      "step": 6613
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5264428853988647,
      "learning_rate": 1.8013495596598877e-05,
      "loss": 0.8397,
      "step": 6614
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6856008768081665,
      "learning_rate": 1.801282761613589e-05,
      "loss": 0.6575,
      "step": 6615
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8740695714950562,
      "learning_rate": 1.8012159535773975e-05,
      "loss": 0.8379,
      "step": 6616
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4195919036865234,
      "learning_rate": 1.8011491355521463e-05,
      "loss": 0.7905,
      "step": 6617
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5165311098098755,
      "learning_rate": 1.8010823075386678e-05,
      "loss": 0.6193,
      "step": 6618
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5575073957443237,
      "learning_rate": 1.8010154695377952e-05,
      "loss": 0.759,
      "step": 6619
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6908308267593384,
      "learning_rate": 1.8009486215503623e-05,
      "loss": 0.8138,
      "step": 6620
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2592576742172241,
      "learning_rate": 1.8008817635772023e-05,
      "loss": 0.9287,
      "step": 6621
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7264949083328247,
      "learning_rate": 1.8008148956191492e-05,
      "loss": 0.8086,
      "step": 6622
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9579932689666748,
      "learning_rate": 1.8007480176770357e-05,
      "loss": 0.7457,
      "step": 6623
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.654010534286499,
      "learning_rate": 1.800681129751696e-05,
      "loss": 0.7372,
      "step": 6624
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.697377324104309,
      "learning_rate": 1.8006142318439646e-05,
      "loss": 0.678,
      "step": 6625
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.575991153717041,
      "learning_rate": 1.800547323954675e-05,
      "loss": 0.7474,
      "step": 6626
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2633893489837646,
      "learning_rate": 1.8004804060846614e-05,
      "loss": 0.788,
      "step": 6627
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5015758275985718,
      "learning_rate": 1.800413478234758e-05,
      "loss": 0.8409,
      "step": 6628
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.602527141571045,
      "learning_rate": 1.8003465404057995e-05,
      "loss": 0.6551,
      "step": 6629
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7051477432250977,
      "learning_rate": 1.80027959259862e-05,
      "loss": 0.6727,
      "step": 6630
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5263962745666504,
      "learning_rate": 1.8002126348140546e-05,
      "loss": 0.7057,
      "step": 6631
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9173951148986816,
      "learning_rate": 1.800145667052938e-05,
      "loss": 0.7813,
      "step": 6632
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.47641122341156,
      "learning_rate": 1.8000786893161048e-05,
      "loss": 0.72,
      "step": 6633
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4905751943588257,
      "learning_rate": 1.800011701604391e-05,
      "loss": 0.7761,
      "step": 6634
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6539902687072754,
      "learning_rate": 1.7999447039186303e-05,
      "loss": 0.7844,
      "step": 6635
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6919244527816772,
      "learning_rate": 1.799877696259659e-05,
      "loss": 0.7435,
      "step": 6636
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3722904920578003,
      "learning_rate": 1.799810678628312e-05,
      "loss": 0.7488,
      "step": 6637
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6588667631149292,
      "learning_rate": 1.799743651025425e-05,
      "loss": 0.7064,
      "step": 6638
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6517826318740845,
      "learning_rate": 1.7996766134518343e-05,
      "loss": 0.8422,
      "step": 6639
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.461022138595581,
      "learning_rate": 1.7996095659083748e-05,
      "loss": 0.6601,
      "step": 6640
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.777546763420105,
      "learning_rate": 1.7995425083958826e-05,
      "loss": 0.738,
      "step": 6641
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9218288660049438,
      "learning_rate": 1.799475440915194e-05,
      "loss": 0.7753,
      "step": 6642
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.559326410293579,
      "learning_rate": 1.799408363467145e-05,
      "loss": 0.6679,
      "step": 6643
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4254348278045654,
      "learning_rate": 1.799341276052572e-05,
      "loss": 0.7503,
      "step": 6644
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4275822639465332,
      "learning_rate": 1.799274178672311e-05,
      "loss": 0.8033,
      "step": 6645
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3560198545455933,
      "learning_rate": 1.799207071327199e-05,
      "loss": 0.8112,
      "step": 6646
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9190287590026855,
      "learning_rate": 1.7991399540180726e-05,
      "loss": 0.6208,
      "step": 6647
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9016410112380981,
      "learning_rate": 1.7990728267457684e-05,
      "loss": 0.6314,
      "step": 6648
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.847695231437683,
      "learning_rate": 1.7990056895111227e-05,
      "loss": 0.9419,
      "step": 6649
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6843624114990234,
      "learning_rate": 1.798938542314974e-05,
      "loss": 0.751,
      "step": 6650
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6567044258117676,
      "learning_rate": 1.7988713851581584e-05,
      "loss": 0.6963,
      "step": 6651
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4672839641571045,
      "learning_rate": 1.798804218041513e-05,
      "loss": 0.6782,
      "step": 6652
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4616410732269287,
      "learning_rate": 1.7987370409658762e-05,
      "loss": 0.6581,
      "step": 6653
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.509813666343689,
      "learning_rate": 1.7986698539320846e-05,
      "loss": 0.6614,
      "step": 6654
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.429503321647644,
      "learning_rate": 1.7986026569409762e-05,
      "loss": 0.8141,
      "step": 6655
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4140197038650513,
      "learning_rate": 1.798535449993389e-05,
      "loss": 0.7269,
      "step": 6656
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6994835138320923,
      "learning_rate": 1.7984682330901598e-05,
      "loss": 0.5978,
      "step": 6657
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7722874879837036,
      "learning_rate": 1.7984010062321283e-05,
      "loss": 0.7516,
      "step": 6658
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8823970556259155,
      "learning_rate": 1.7983337694201314e-05,
      "loss": 0.8689,
      "step": 6659
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5508025884628296,
      "learning_rate": 1.798266522655008e-05,
      "loss": 0.7476,
      "step": 6660
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.298859715461731,
      "learning_rate": 1.798199265937596e-05,
      "loss": 0.6286,
      "step": 6661
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4200801849365234,
      "learning_rate": 1.798131999268734e-05,
      "loss": 0.7769,
      "step": 6662
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7415682077407837,
      "learning_rate": 1.7980647226492613e-05,
      "loss": 0.8717,
      "step": 6663
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6055206060409546,
      "learning_rate": 1.7979974360800158e-05,
      "loss": 0.6141,
      "step": 6664
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5656729936599731,
      "learning_rate": 1.7979301395618368e-05,
      "loss": 0.803,
      "step": 6665
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5954725742340088,
      "learning_rate": 1.797862833095563e-05,
      "loss": 0.764,
      "step": 6666
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3185685873031616,
      "learning_rate": 1.7977955166820344e-05,
      "loss": 0.7292,
      "step": 6667
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.574997901916504,
      "learning_rate": 1.797728190322089e-05,
      "loss": 0.8138,
      "step": 6668
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5520387887954712,
      "learning_rate": 1.7976608540165675e-05,
      "loss": 0.7486,
      "step": 6669
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6953158378601074,
      "learning_rate": 1.7975935077663083e-05,
      "loss": 0.743,
      "step": 6670
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4174864292144775,
      "learning_rate": 1.7975261515721516e-05,
      "loss": 0.7294,
      "step": 6671
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6430909633636475,
      "learning_rate": 1.797458785434937e-05,
      "loss": 0.855,
      "step": 6672
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3541576862335205,
      "learning_rate": 1.7973914093555046e-05,
      "loss": 0.6783,
      "step": 6673
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.932231068611145,
      "learning_rate": 1.797324023334694e-05,
      "loss": 0.7085,
      "step": 6674
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6872363090515137,
      "learning_rate": 1.7972566273733454e-05,
      "loss": 0.7452,
      "step": 6675
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.318307876586914,
      "learning_rate": 1.7971892214722996e-05,
      "loss": 0.7064,
      "step": 6676
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4074705839157104,
      "learning_rate": 1.797121805632396e-05,
      "loss": 0.6605,
      "step": 6677
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5479602813720703,
      "learning_rate": 1.797054379854476e-05,
      "loss": 0.8166,
      "step": 6678
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6460365056991577,
      "learning_rate": 1.7969869441393803e-05,
      "loss": 0.7905,
      "step": 6679
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2450895309448242,
      "learning_rate": 1.7969194984879483e-05,
      "loss": 0.7671,
      "step": 6680
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4379181861877441,
      "learning_rate": 1.7968520429010225e-05,
      "loss": 0.7183,
      "step": 6681
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4922454357147217,
      "learning_rate": 1.796784577379443e-05,
      "loss": 0.8135,
      "step": 6682
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.544987440109253,
      "learning_rate": 1.7967171019240513e-05,
      "loss": 0.6228,
      "step": 6683
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.585766315460205,
      "learning_rate": 1.7966496165356878e-05,
      "loss": 0.7644,
      "step": 6684
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.488972544670105,
      "learning_rate": 1.7965821212151952e-05,
      "loss": 0.7141,
      "step": 6685
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6541945934295654,
      "learning_rate": 1.7965146159634138e-05,
      "loss": 0.7368,
      "step": 6686
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5924859046936035,
      "learning_rate": 1.7964471007811864e-05,
      "loss": 0.7306,
      "step": 6687
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7570346593856812,
      "learning_rate": 1.7963795756693535e-05,
      "loss": 0.8259,
      "step": 6688
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5819785594940186,
      "learning_rate": 1.796312040628758e-05,
      "loss": 0.8279,
      "step": 6689
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6654181480407715,
      "learning_rate": 1.796244495660241e-05,
      "loss": 0.825,
      "step": 6690
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7077995538711548,
      "learning_rate": 1.796176940764645e-05,
      "loss": 0.6585,
      "step": 6691
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5191730260849,
      "learning_rate": 1.796109375942812e-05,
      "loss": 0.9364,
      "step": 6692
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4193739891052246,
      "learning_rate": 1.796041801195585e-05,
      "loss": 0.6472,
      "step": 6693
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7128126621246338,
      "learning_rate": 1.7959742165238064e-05,
      "loss": 0.6582,
      "step": 6694
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4742494821548462,
      "learning_rate": 1.795906621928318e-05,
      "loss": 0.8622,
      "step": 6695
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3189568519592285,
      "learning_rate": 1.795839017409963e-05,
      "loss": 0.6925,
      "step": 6696
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8025360107421875,
      "learning_rate": 1.7957714029695844e-05,
      "loss": 0.7257,
      "step": 6697
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.439040184020996,
      "learning_rate": 1.795703778608025e-05,
      "loss": 0.7201,
      "step": 6698
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3935216665267944,
      "learning_rate": 1.7956361443261282e-05,
      "loss": 0.7701,
      "step": 6699
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4150177240371704,
      "learning_rate": 1.7955685001247365e-05,
      "loss": 0.7098,
      "step": 6700
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7524940967559814,
      "learning_rate": 1.795500846004694e-05,
      "loss": 0.8865,
      "step": 6701
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6326960325241089,
      "learning_rate": 1.7954331819668437e-05,
      "loss": 0.5902,
      "step": 6702
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6590598821640015,
      "learning_rate": 1.7953655080120295e-05,
      "loss": 0.5996,
      "step": 6703
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2954641580581665,
      "learning_rate": 1.7952978241410948e-05,
      "loss": 0.6783,
      "step": 6704
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.8135546445846558,
      "learning_rate": 1.795230130354884e-05,
      "loss": 0.5835,
      "step": 6705
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9968644380569458,
      "learning_rate": 1.7951624266542402e-05,
      "loss": 0.8214,
      "step": 6706
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4588255882263184,
      "learning_rate": 1.795094713040008e-05,
      "loss": 0.6863,
      "step": 6707
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.414166808128357,
      "learning_rate": 1.7950269895130322e-05,
      "loss": 0.7105,
      "step": 6708
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.62289297580719,
      "learning_rate": 1.794959256074156e-05,
      "loss": 0.5946,
      "step": 6709
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3326054811477661,
      "learning_rate": 1.7948915127242243e-05,
      "loss": 0.7801,
      "step": 6710
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5399134159088135,
      "learning_rate": 1.7948237594640822e-05,
      "loss": 0.7923,
      "step": 6711
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5617356300354004,
      "learning_rate": 1.794755996294574e-05,
      "loss": 0.85,
      "step": 6712
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5755679607391357,
      "learning_rate": 1.794688223216544e-05,
      "loss": 0.6459,
      "step": 6713
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.404183030128479,
      "learning_rate": 1.7946204402308377e-05,
      "loss": 0.6448,
      "step": 6714
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7856651544570923,
      "learning_rate": 1.7945526473383005e-05,
      "loss": 0.8714,
      "step": 6715
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.581648349761963,
      "learning_rate": 1.794484844539777e-05,
      "loss": 0.5923,
      "step": 6716
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.9230520725250244,
      "learning_rate": 1.7944170318361126e-05,
      "loss": 0.8167,
      "step": 6717
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3670899868011475,
      "learning_rate": 1.7943492092281532e-05,
      "loss": 0.689,
      "step": 6718
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2471853494644165,
      "learning_rate": 1.794281376716744e-05,
      "loss": 0.5837,
      "step": 6719
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9360517263412476,
      "learning_rate": 1.7942135343027305e-05,
      "loss": 0.7461,
      "step": 6720
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.579500675201416,
      "learning_rate": 1.794145681986959e-05,
      "loss": 0.6349,
      "step": 6721
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3217021226882935,
      "learning_rate": 1.794077819770275e-05,
      "loss": 0.6056,
      "step": 6722
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3432807922363281,
      "learning_rate": 1.794009947653525e-05,
      "loss": 0.7621,
      "step": 6723
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7679606676101685,
      "learning_rate": 1.793942065637555e-05,
      "loss": 0.7673,
      "step": 6724
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9101642370224,
      "learning_rate": 1.7938741737232107e-05,
      "loss": 0.7418,
      "step": 6725
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3907129764556885,
      "learning_rate": 1.7938062719113398e-05,
      "loss": 0.7004,
      "step": 6726
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4803009033203125,
      "learning_rate": 1.793738360202788e-05,
      "loss": 0.5797,
      "step": 6727
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4996469020843506,
      "learning_rate": 1.793670438598402e-05,
      "loss": 0.7539,
      "step": 6728
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9041533470153809,
      "learning_rate": 1.793602507099029e-05,
      "loss": 0.7166,
      "step": 6729
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.74461030960083,
      "learning_rate": 1.7935345657055154e-05,
      "loss": 0.8762,
      "step": 6730
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4296923875808716,
      "learning_rate": 1.7934666144187085e-05,
      "loss": 0.6125,
      "step": 6731
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4694746732711792,
      "learning_rate": 1.7933986532394556e-05,
      "loss": 0.5565,
      "step": 6732
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6413822174072266,
      "learning_rate": 1.793330682168604e-05,
      "loss": 0.6462,
      "step": 6733
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6446266174316406,
      "learning_rate": 1.793262701207001e-05,
      "loss": 0.6692,
      "step": 6734
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.276573896408081,
      "learning_rate": 1.793194710355494e-05,
      "loss": 0.6054,
      "step": 6735
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.020442485809326,
      "learning_rate": 1.793126709614931e-05,
      "loss": 0.82,
      "step": 6736
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5920871496200562,
      "learning_rate": 1.7930586989861593e-05,
      "loss": 0.8755,
      "step": 6737
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8108817338943481,
      "learning_rate": 1.7929906784700276e-05,
      "loss": 0.7207,
      "step": 6738
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4451581239700317,
      "learning_rate": 1.7929226480673832e-05,
      "loss": 0.8547,
      "step": 6739
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7315462827682495,
      "learning_rate": 1.7928546077790746e-05,
      "loss": 0.8816,
      "step": 6740
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.479337453842163,
      "learning_rate": 1.79278655760595e-05,
      "loss": 0.5761,
      "step": 6741
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.390628695487976,
      "learning_rate": 1.7927184975488577e-05,
      "loss": 0.7701,
      "step": 6742
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5645127296447754,
      "learning_rate": 1.7926504276086465e-05,
      "loss": 0.8058,
      "step": 6743
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7657217979431152,
      "learning_rate": 1.7925823477861647e-05,
      "loss": 0.6517,
      "step": 6744
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5780373811721802,
      "learning_rate": 1.7925142580822615e-05,
      "loss": 0.7667,
      "step": 6745
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3644416332244873,
      "learning_rate": 1.7924461584977858e-05,
      "loss": 0.7383,
      "step": 6746
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.537649393081665,
      "learning_rate": 1.792378049033586e-05,
      "loss": 0.8028,
      "step": 6747
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6731010675430298,
      "learning_rate": 1.792309929690512e-05,
      "loss": 0.8571,
      "step": 6748
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.0811469554901123,
      "learning_rate": 1.7922418004694123e-05,
      "loss": 0.7444,
      "step": 6749
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6548229455947876,
      "learning_rate": 1.7921736613711373e-05,
      "loss": 0.7855,
      "step": 6750
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4744198322296143,
      "learning_rate": 1.7921055123965353e-05,
      "loss": 0.7063,
      "step": 6751
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5110002756118774,
      "learning_rate": 1.792037353546457e-05,
      "loss": 0.7592,
      "step": 6752
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.398354411125183,
      "learning_rate": 1.791969184821752e-05,
      "loss": 0.7246,
      "step": 6753
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6166658401489258,
      "learning_rate": 1.7919010062232693e-05,
      "loss": 0.6357,
      "step": 6754
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4991629123687744,
      "learning_rate": 1.7918328177518594e-05,
      "loss": 0.5131,
      "step": 6755
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5095882415771484,
      "learning_rate": 1.7917646194083734e-05,
      "loss": 0.6056,
      "step": 6756
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5010193586349487,
      "learning_rate": 1.79169641119366e-05,
      "loss": 0.7485,
      "step": 6757
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3740911483764648,
      "learning_rate": 1.7916281931085705e-05,
      "loss": 0.7497,
      "step": 6758
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.418715238571167,
      "learning_rate": 1.791559965153955e-05,
      "loss": 0.8721,
      "step": 6759
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3873778581619263,
      "learning_rate": 1.7914917273306646e-05,
      "loss": 0.8031,
      "step": 6760
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6765676736831665,
      "learning_rate": 1.7914234796395497e-05,
      "loss": 0.715,
      "step": 6761
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5072171688079834,
      "learning_rate": 1.791355222081461e-05,
      "loss": 0.6934,
      "step": 6762
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.583755373954773,
      "learning_rate": 1.7912869546572503e-05,
      "loss": 0.7534,
      "step": 6763
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7282867431640625,
      "learning_rate": 1.7912186773677674e-05,
      "loss": 0.7916,
      "step": 6764
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.443912148475647,
      "learning_rate": 1.791150390213865e-05,
      "loss": 0.7669,
      "step": 6765
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3665549755096436,
      "learning_rate": 1.7910820931963934e-05,
      "loss": 0.6983,
      "step": 6766
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6233798265457153,
      "learning_rate": 1.791013786316204e-05,
      "loss": 0.741,
      "step": 6767
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.59872567653656,
      "learning_rate": 1.7909454695741495e-05,
      "loss": 0.7251,
      "step": 6768
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5744599103927612,
      "learning_rate": 1.7908771429710806e-05,
      "loss": 0.7469,
      "step": 6769
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5021802186965942,
      "learning_rate": 1.7908088065078496e-05,
      "loss": 0.6885,
      "step": 6770
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2977997064590454,
      "learning_rate": 1.7907404601853086e-05,
      "loss": 0.6526,
      "step": 6771
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.667894721031189,
      "learning_rate": 1.7906721040043093e-05,
      "loss": 0.7304,
      "step": 6772
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4233890771865845,
      "learning_rate": 1.790603737965704e-05,
      "loss": 0.7969,
      "step": 6773
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3853799104690552,
      "learning_rate": 1.790535362070345e-05,
      "loss": 0.6944,
      "step": 6774
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3696740865707397,
      "learning_rate": 1.790466976319085e-05,
      "loss": 0.8232,
      "step": 6775
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3345470428466797,
      "learning_rate": 1.7903985807127767e-05,
      "loss": 0.6444,
      "step": 6776
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6710819005966187,
      "learning_rate": 1.7903301752522727e-05,
      "loss": 0.6674,
      "step": 6777
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6376309394836426,
      "learning_rate": 1.7902617599384258e-05,
      "loss": 0.6268,
      "step": 6778
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9968363046646118,
      "learning_rate": 1.7901933347720888e-05,
      "loss": 0.6941,
      "step": 6779
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6467097997665405,
      "learning_rate": 1.790124899754115e-05,
      "loss": 0.6953,
      "step": 6780
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.524309754371643,
      "learning_rate": 1.7900564548853573e-05,
      "loss": 0.7191,
      "step": 6781
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6165529489517212,
      "learning_rate": 1.7899880001666696e-05,
      "loss": 0.6906,
      "step": 6782
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6835650205612183,
      "learning_rate": 1.7899195355989048e-05,
      "loss": 0.7795,
      "step": 6783
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.430525302886963,
      "learning_rate": 1.7898510611829163e-05,
      "loss": 0.7526,
      "step": 6784
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8318891525268555,
      "learning_rate": 1.7897825769195588e-05,
      "loss": 0.8695,
      "step": 6785
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5709173679351807,
      "learning_rate": 1.789714082809685e-05,
      "loss": 0.7645,
      "step": 6786
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4421374797821045,
      "learning_rate": 1.7896455788541494e-05,
      "loss": 0.7613,
      "step": 6787
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7483373880386353,
      "learning_rate": 1.789577065053806e-05,
      "loss": 0.6496,
      "step": 6788
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3757469654083252,
      "learning_rate": 1.789508541409509e-05,
      "loss": 0.6388,
      "step": 6789
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4944403171539307,
      "learning_rate": 1.7894400079221127e-05,
      "loss": 0.6378,
      "step": 6790
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.311126470565796,
      "learning_rate": 1.789371464592472e-05,
      "loss": 0.8111,
      "step": 6791
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4447554349899292,
      "learning_rate": 1.7893029114214403e-05,
      "loss": 0.9859,
      "step": 6792
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5273751020431519,
      "learning_rate": 1.7892343484098732e-05,
      "loss": 0.6388,
      "step": 6793
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4451335668563843,
      "learning_rate": 1.7891657755586253e-05,
      "loss": 0.6337,
      "step": 6794
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5026992559432983,
      "learning_rate": 1.7890971928685513e-05,
      "loss": 0.639,
      "step": 6795
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6550180912017822,
      "learning_rate": 1.7890286003405064e-05,
      "loss": 0.7062,
      "step": 6796
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3469929695129395,
      "learning_rate": 1.788959997975346e-05,
      "loss": 0.7435,
      "step": 6797
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4528522491455078,
      "learning_rate": 1.788891385773925e-05,
      "loss": 0.6306,
      "step": 6798
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4674186706542969,
      "learning_rate": 1.788822763737099e-05,
      "loss": 0.6761,
      "step": 6799
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8992204666137695,
      "learning_rate": 1.7887541318657237e-05,
      "loss": 0.8982,
      "step": 6800
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.572259545326233,
      "learning_rate": 1.7886854901606545e-05,
      "loss": 0.9249,
      "step": 6801
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6022694110870361,
      "learning_rate": 1.7886168386227472e-05,
      "loss": 0.7636,
      "step": 6802
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6372450590133667,
      "learning_rate": 1.7885481772528578e-05,
      "loss": 0.7424,
      "step": 6803
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.357282042503357,
      "learning_rate": 1.7884795060518422e-05,
      "loss": 0.6798,
      "step": 6804
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7216917276382446,
      "learning_rate": 1.7884108250205566e-05,
      "loss": 0.8079,
      "step": 6805
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5833630561828613,
      "learning_rate": 1.788342134159858e-05,
      "loss": 0.7135,
      "step": 6806
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3776565790176392,
      "learning_rate": 1.7882734334706015e-05,
      "loss": 0.7522,
      "step": 6807
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.572827935218811,
      "learning_rate": 1.7882047229536443e-05,
      "loss": 0.6587,
      "step": 6808
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4289591312408447,
      "learning_rate": 1.7881360026098432e-05,
      "loss": 0.7526,
      "step": 6809
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5979903936386108,
      "learning_rate": 1.7880672724400544e-05,
      "loss": 0.7576,
      "step": 6810
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.274006724357605,
      "learning_rate": 1.7879985324451353e-05,
      "loss": 0.8146,
      "step": 6811
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4762425422668457,
      "learning_rate": 1.787929782625943e-05,
      "loss": 0.7954,
      "step": 6812
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.852757215499878,
      "learning_rate": 1.787861022983334e-05,
      "loss": 0.9388,
      "step": 6813
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2081313133239746,
      "learning_rate": 1.787792253518166e-05,
      "loss": 0.694,
      "step": 6814
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.656256079673767,
      "learning_rate": 1.7877234742312965e-05,
      "loss": 0.7849,
      "step": 6815
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6288114786148071,
      "learning_rate": 1.7876546851235823e-05,
      "loss": 0.7832,
      "step": 6816
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4689102172851562,
      "learning_rate": 1.7875858861958818e-05,
      "loss": 0.6335,
      "step": 6817
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3482327461242676,
      "learning_rate": 1.7875170774490524e-05,
      "loss": 0.6322,
      "step": 6818
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6832369565963745,
      "learning_rate": 1.787448258883952e-05,
      "loss": 0.6227,
      "step": 6819
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.360809326171875,
      "learning_rate": 1.7873794305014388e-05,
      "loss": 0.6731,
      "step": 6820
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4675052165985107,
      "learning_rate": 1.7873105923023706e-05,
      "loss": 0.7,
      "step": 6821
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.457337498664856,
      "learning_rate": 1.7872417442876057e-05,
      "loss": 0.7056,
      "step": 6822
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9715542793273926,
      "learning_rate": 1.7871728864580024e-05,
      "loss": 0.6584,
      "step": 6823
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.499213695526123,
      "learning_rate": 1.7871040188144193e-05,
      "loss": 0.6014,
      "step": 6824
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8352843523025513,
      "learning_rate": 1.7870351413577152e-05,
      "loss": 0.8574,
      "step": 6825
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7333132028579712,
      "learning_rate": 1.7869662540887483e-05,
      "loss": 0.6157,
      "step": 6826
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5181899070739746,
      "learning_rate": 1.786897357008378e-05,
      "loss": 0.634,
      "step": 6827
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.597070336341858,
      "learning_rate": 1.786828450117463e-05,
      "loss": 0.7402,
      "step": 6828
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4660564661026,
      "learning_rate": 1.786759533416862e-05,
      "loss": 0.6445,
      "step": 6829
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4369423389434814,
      "learning_rate": 1.7866906069074347e-05,
      "loss": 0.7264,
      "step": 6830
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.358047366142273,
      "learning_rate": 1.7866216705900408e-05,
      "loss": 0.6289,
      "step": 6831
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1844006776809692,
      "learning_rate": 1.7865527244655388e-05,
      "loss": 0.642,
      "step": 6832
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4487093687057495,
      "learning_rate": 1.786483768534789e-05,
      "loss": 0.7195,
      "step": 6833
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3669872283935547,
      "learning_rate": 1.7864148027986506e-05,
      "loss": 0.75,
      "step": 6834
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7459402084350586,
      "learning_rate": 1.786345827257984e-05,
      "loss": 0.7942,
      "step": 6835
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.616363525390625,
      "learning_rate": 1.7862768419136487e-05,
      "loss": 0.686,
      "step": 6836
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3771073818206787,
      "learning_rate": 1.7862078467665047e-05,
      "loss": 0.6581,
      "step": 6837
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6024974584579468,
      "learning_rate": 1.7861388418174127e-05,
      "loss": 0.7036,
      "step": 6838
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3768762350082397,
      "learning_rate": 1.7860698270672324e-05,
      "loss": 0.6763,
      "step": 6839
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.262461543083191,
      "learning_rate": 1.7860008025168246e-05,
      "loss": 0.7061,
      "step": 6840
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3853919506072998,
      "learning_rate": 1.7859317681670503e-05,
      "loss": 0.6425,
      "step": 6841
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5961931943893433,
      "learning_rate": 1.785862724018769e-05,
      "loss": 0.6358,
      "step": 6842
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0069456100463867,
      "learning_rate": 1.7857936700728425e-05,
      "loss": 0.6824,
      "step": 6843
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7608305215835571,
      "learning_rate": 1.785724606330131e-05,
      "loss": 0.7426,
      "step": 6844
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.465441107749939,
      "learning_rate": 1.7856555327914963e-05,
      "loss": 0.8664,
      "step": 6845
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6212629079818726,
      "learning_rate": 1.785586449457799e-05,
      "loss": 0.744,
      "step": 6846
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6282060146331787,
      "learning_rate": 1.7855173563299008e-05,
      "loss": 0.7156,
      "step": 6847
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.560596227645874,
      "learning_rate": 1.7854482534086627e-05,
      "loss": 0.7905,
      "step": 6848
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4642378091812134,
      "learning_rate": 1.785379140694946e-05,
      "loss": 0.6923,
      "step": 6849
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.648970127105713,
      "learning_rate": 1.7853100181896138e-05,
      "loss": 0.754,
      "step": 6850
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.627840518951416,
      "learning_rate": 1.7852408858935262e-05,
      "loss": 0.6179,
      "step": 6851
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.610769271850586,
      "learning_rate": 1.7851717438075456e-05,
      "loss": 0.6018,
      "step": 6852
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6943751573562622,
      "learning_rate": 1.7851025919325343e-05,
      "loss": 0.7547,
      "step": 6853
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3823285102844238,
      "learning_rate": 1.7850334302693546e-05,
      "loss": 0.6305,
      "step": 6854
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3607563972473145,
      "learning_rate": 1.7849642588188682e-05,
      "loss": 0.652,
      "step": 6855
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6052106618881226,
      "learning_rate": 1.784895077581938e-05,
      "loss": 0.7293,
      "step": 6856
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2708698511123657,
      "learning_rate": 1.784825886559426e-05,
      "loss": 0.7106,
      "step": 6857
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.2954490184783936,
      "learning_rate": 1.7847566857521948e-05,
      "loss": 0.697,
      "step": 6858
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5472828149795532,
      "learning_rate": 1.7846874751611083e-05,
      "loss": 0.8902,
      "step": 6859
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6052918434143066,
      "learning_rate": 1.7846182547870278e-05,
      "loss": 0.7362,
      "step": 6860
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3905895948410034,
      "learning_rate": 1.7845490246308172e-05,
      "loss": 0.7852,
      "step": 6861
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3687747716903687,
      "learning_rate": 1.7844797846933398e-05,
      "loss": 0.9529,
      "step": 6862
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2433607578277588,
      "learning_rate": 1.7844105349754582e-05,
      "loss": 0.7438,
      "step": 6863
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3589141368865967,
      "learning_rate": 1.7843412754780362e-05,
      "loss": 0.704,
      "step": 6864
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7043570280075073,
      "learning_rate": 1.784272006201937e-05,
      "loss": 0.7377,
      "step": 6865
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.985557198524475,
      "learning_rate": 1.784202727148024e-05,
      "loss": 0.8265,
      "step": 6866
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.465186595916748,
      "learning_rate": 1.7841334383171614e-05,
      "loss": 0.6522,
      "step": 6867
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5956647396087646,
      "learning_rate": 1.7840641397102133e-05,
      "loss": 0.7578,
      "step": 6868
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3586053848266602,
      "learning_rate": 1.7839948313280428e-05,
      "loss": 0.6701,
      "step": 6869
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4797325134277344,
      "learning_rate": 1.783925513171515e-05,
      "loss": 0.743,
      "step": 6870
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3096429109573364,
      "learning_rate": 1.7838561852414933e-05,
      "loss": 0.7191,
      "step": 6871
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4070440530776978,
      "learning_rate": 1.7837868475388426e-05,
      "loss": 0.6986,
      "step": 6872
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.9208356142044067,
      "learning_rate": 1.7837175000644268e-05,
      "loss": 0.7941,
      "step": 6873
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6646617650985718,
      "learning_rate": 1.7836481428191105e-05,
      "loss": 0.5533,
      "step": 6874
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2890794277191162,
      "learning_rate": 1.783578775803759e-05,
      "loss": 0.6902,
      "step": 6875
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2718243598937988,
      "learning_rate": 1.783509399019237e-05,
      "loss": 0.7039,
      "step": 6876
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.561618447303772,
      "learning_rate": 1.783440012466409e-05,
      "loss": 0.7034,
      "step": 6877
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.658374547958374,
      "learning_rate": 1.78337061614614e-05,
      "loss": 0.7514,
      "step": 6878
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7962803840637207,
      "learning_rate": 1.783301210059296e-05,
      "loss": 0.6832,
      "step": 6879
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4160079956054688,
      "learning_rate": 1.7832317942067417e-05,
      "loss": 0.6382,
      "step": 6880
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3291165828704834,
      "learning_rate": 1.7831623685893427e-05,
      "loss": 0.7933,
      "step": 6881
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6317607164382935,
      "learning_rate": 1.783092933207964e-05,
      "loss": 0.797,
      "step": 6882
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5528730154037476,
      "learning_rate": 1.7830234880634724e-05,
      "loss": 0.7762,
      "step": 6883
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.854719638824463,
      "learning_rate": 1.782954033156733e-05,
      "loss": 0.887,
      "step": 6884
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3903807401657104,
      "learning_rate": 1.7828845684886115e-05,
      "loss": 0.5745,
      "step": 6885
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.565104365348816,
      "learning_rate": 1.7828150940599744e-05,
      "loss": 0.5635,
      "step": 6886
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.61179780960083,
      "learning_rate": 1.7827456098716877e-05,
      "loss": 0.6721,
      "step": 6887
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4777950048446655,
      "learning_rate": 1.782676115924618e-05,
      "loss": 0.7383,
      "step": 6888
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.635390043258667,
      "learning_rate": 1.782606612219631e-05,
      "loss": 0.7326,
      "step": 6889
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6518099308013916,
      "learning_rate": 1.7825370987575935e-05,
      "loss": 0.6853,
      "step": 6890
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6088439226150513,
      "learning_rate": 1.7824675755393727e-05,
      "loss": 0.5512,
      "step": 6891
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4640047550201416,
      "learning_rate": 1.7823980425658345e-05,
      "loss": 0.6539,
      "step": 6892
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4602272510528564,
      "learning_rate": 1.7823284998378465e-05,
      "loss": 0.6559,
      "step": 6893
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4134753942489624,
      "learning_rate": 1.7822589473562756e-05,
      "loss": 0.7812,
      "step": 6894
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4216210842132568,
      "learning_rate": 1.7821893851219886e-05,
      "loss": 0.6679,
      "step": 6895
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7114017009735107,
      "learning_rate": 1.782119813135853e-05,
      "loss": 0.7696,
      "step": 6896
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6477283239364624,
      "learning_rate": 1.782050231398736e-05,
      "loss": 0.6885,
      "step": 6897
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6169825792312622,
      "learning_rate": 1.7819806399115052e-05,
      "loss": 0.8112,
      "step": 6898
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7497371435165405,
      "learning_rate": 1.7819110386750284e-05,
      "loss": 0.8476,
      "step": 6899
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.251387119293213,
      "learning_rate": 1.7818414276901732e-05,
      "loss": 0.646,
      "step": 6900
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4814845323562622,
      "learning_rate": 1.7817718069578074e-05,
      "loss": 0.8973,
      "step": 6901
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3306105136871338,
      "learning_rate": 1.7817021764787993e-05,
      "loss": 0.5899,
      "step": 6902
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6248550415039062,
      "learning_rate": 1.7816325362540167e-05,
      "loss": 0.8048,
      "step": 6903
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4413734674453735,
      "learning_rate": 1.7815628862843274e-05,
      "loss": 0.8274,
      "step": 6904
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.553455114364624,
      "learning_rate": 1.7814932265706008e-05,
      "loss": 0.7657,
      "step": 6905
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.623968482017517,
      "learning_rate": 1.7814235571137046e-05,
      "loss": 0.7622,
      "step": 6906
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2543342113494873,
      "learning_rate": 1.7813538779145078e-05,
      "loss": 0.6971,
      "step": 6907
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2483283281326294,
      "learning_rate": 1.781284188973879e-05,
      "loss": 0.5586,
      "step": 6908
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7538747787475586,
      "learning_rate": 1.7812144902926865e-05,
      "loss": 0.8992,
      "step": 6909
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5583853721618652,
      "learning_rate": 1.7811447818718e-05,
      "loss": 0.6849,
      "step": 6910
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.0807301998138428,
      "learning_rate": 1.7810750637120882e-05,
      "loss": 0.7123,
      "step": 6911
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7572047710418701,
      "learning_rate": 1.7810053358144204e-05,
      "loss": 0.7104,
      "step": 6912
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7286961078643799,
      "learning_rate": 1.7809355981796665e-05,
      "loss": 0.6914,
      "step": 6913
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.607478141784668,
      "learning_rate": 1.7808658508086946e-05,
      "loss": 0.6457,
      "step": 6914
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.498636245727539,
      "learning_rate": 1.7807960937023754e-05,
      "loss": 0.586,
      "step": 6915
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6753631830215454,
      "learning_rate": 1.7807263268615784e-05,
      "loss": 0.8603,
      "step": 6916
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9656002521514893,
      "learning_rate": 1.780656550287173e-05,
      "loss": 0.8561,
      "step": 6917
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7858399152755737,
      "learning_rate": 1.7805867639800297e-05,
      "loss": 0.7854,
      "step": 6918
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8838648796081543,
      "learning_rate": 1.7805169679410177e-05,
      "loss": 0.6927,
      "step": 6919
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3602241277694702,
      "learning_rate": 1.780447162171008e-05,
      "loss": 0.6692,
      "step": 6920
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3203928470611572,
      "learning_rate": 1.7803773466708706e-05,
      "loss": 0.5998,
      "step": 6921
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5622440576553345,
      "learning_rate": 1.780307521441476e-05,
      "loss": 0.6043,
      "step": 6922
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4869391918182373,
      "learning_rate": 1.7802376864836945e-05,
      "loss": 0.7256,
      "step": 6923
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4803048372268677,
      "learning_rate": 1.780167841798397e-05,
      "loss": 0.7219,
      "step": 6924
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8196359872817993,
      "learning_rate": 1.7800979873864542e-05,
      "loss": 0.8004,
      "step": 6925
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.252654790878296,
      "learning_rate": 1.780028123248737e-05,
      "loss": 0.655,
      "step": 6926
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5967384576797485,
      "learning_rate": 1.7799582493861163e-05,
      "loss": 0.6877,
      "step": 6927
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5284415483474731,
      "learning_rate": 1.7798883657994635e-05,
      "loss": 0.6458,
      "step": 6928
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3391193151474,
      "learning_rate": 1.7798184724896493e-05,
      "loss": 0.7233,
      "step": 6929
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1634536981582642,
      "learning_rate": 1.7797485694575463e-05,
      "loss": 0.6274,
      "step": 6930
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6846978664398193,
      "learning_rate": 1.7796786567040247e-05,
      "loss": 0.7578,
      "step": 6931
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5928633213043213,
      "learning_rate": 1.7796087342299565e-05,
      "loss": 0.6826,
      "step": 6932
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.487985372543335,
      "learning_rate": 1.779538802036214e-05,
      "loss": 0.8559,
      "step": 6933
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.753509283065796,
      "learning_rate": 1.7794688601236686e-05,
      "loss": 0.8084,
      "step": 6934
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6079602241516113,
      "learning_rate": 1.779398908493192e-05,
      "loss": 0.6747,
      "step": 6935
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3349987268447876,
      "learning_rate": 1.779328947145657e-05,
      "loss": 0.6825,
      "step": 6936
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2572979927062988,
      "learning_rate": 1.7792589760819355e-05,
      "loss": 0.6498,
      "step": 6937
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5187283754348755,
      "learning_rate": 1.7791889953029e-05,
      "loss": 0.7944,
      "step": 6938
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5278950929641724,
      "learning_rate": 1.7791190048094225e-05,
      "loss": 0.7041,
      "step": 6939
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8143861293792725,
      "learning_rate": 1.779049004602376e-05,
      "loss": 0.8073,
      "step": 6940
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5228297710418701,
      "learning_rate": 1.778978994682633e-05,
      "loss": 0.684,
      "step": 6941
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6044440269470215,
      "learning_rate": 1.7789089750510665e-05,
      "loss": 0.6863,
      "step": 6942
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.1490707397460938,
      "learning_rate": 1.7788389457085498e-05,
      "loss": 0.868,
      "step": 6943
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6564568281173706,
      "learning_rate": 1.7787689066559557e-05,
      "loss": 0.7768,
      "step": 6944
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6202285289764404,
      "learning_rate": 1.778698857894157e-05,
      "loss": 0.5501,
      "step": 6945
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.506731390953064,
      "learning_rate": 1.7786287994240274e-05,
      "loss": 0.684,
      "step": 6946
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3582921028137207,
      "learning_rate": 1.7785587312464405e-05,
      "loss": 0.72,
      "step": 6947
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.779822826385498,
      "learning_rate": 1.7784886533622695e-05,
      "loss": 0.7939,
      "step": 6948
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6754494905471802,
      "learning_rate": 1.7784185657723886e-05,
      "loss": 0.7215,
      "step": 6949
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9857302904129028,
      "learning_rate": 1.7783484684776708e-05,
      "loss": 0.8249,
      "step": 6950
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.514047384262085,
      "learning_rate": 1.778278361478991e-05,
      "loss": 0.6392,
      "step": 6951
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5812870264053345,
      "learning_rate": 1.7782082447772225e-05,
      "loss": 0.7628,
      "step": 6952
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7414305210113525,
      "learning_rate": 1.7781381183732397e-05,
      "loss": 0.8397,
      "step": 6953
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3276110887527466,
      "learning_rate": 1.7780679822679164e-05,
      "loss": 0.7867,
      "step": 6954
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8629571199417114,
      "learning_rate": 1.7779978364621282e-05,
      "loss": 0.6277,
      "step": 6955
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3886569738388062,
      "learning_rate": 1.777927680956749e-05,
      "loss": 0.7362,
      "step": 6956
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.918065071105957,
      "learning_rate": 1.7778575157526533e-05,
      "loss": 0.7676,
      "step": 6957
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3452363014221191,
      "learning_rate": 1.777787340850716e-05,
      "loss": 0.7141,
      "step": 6958
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4918521642684937,
      "learning_rate": 1.7777171562518115e-05,
      "loss": 0.6328,
      "step": 6959
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7082793712615967,
      "learning_rate": 1.777646961956816e-05,
      "loss": 0.558,
      "step": 6960
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4944220781326294,
      "learning_rate": 1.7775767579666035e-05,
      "loss": 0.6726,
      "step": 6961
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.766397476196289,
      "learning_rate": 1.77750654428205e-05,
      "loss": 0.6698,
      "step": 6962
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7964788675308228,
      "learning_rate": 1.7774363209040304e-05,
      "loss": 0.6471,
      "step": 6963
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5174957513809204,
      "learning_rate": 1.7773660878334206e-05,
      "loss": 0.635,
      "step": 6964
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5355015993118286,
      "learning_rate": 1.7772958450710957e-05,
      "loss": 0.6314,
      "step": 6965
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4810376167297363,
      "learning_rate": 1.777225592617932e-05,
      "loss": 0.7864,
      "step": 6966
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5362348556518555,
      "learning_rate": 1.7771553304748047e-05,
      "loss": 0.5857,
      "step": 6967
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5153228044509888,
      "learning_rate": 1.7770850586425906e-05,
      "loss": 0.6454,
      "step": 6968
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5407005548477173,
      "learning_rate": 1.777014777122165e-05,
      "loss": 0.6871,
      "step": 6969
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8413366079330444,
      "learning_rate": 1.776944485914405e-05,
      "loss": 0.7146,
      "step": 6970
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5845258235931396,
      "learning_rate": 1.7768741850201864e-05,
      "loss": 0.7438,
      "step": 6971
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6114914417266846,
      "learning_rate": 1.7768038744403858e-05,
      "loss": 0.8084,
      "step": 6972
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.519805669784546,
      "learning_rate": 1.7767335541758798e-05,
      "loss": 0.824,
      "step": 6973
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.459167718887329,
      "learning_rate": 1.7766632242275446e-05,
      "loss": 0.6591,
      "step": 6974
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.250130534172058,
      "learning_rate": 1.7765928845962577e-05,
      "loss": 0.8202,
      "step": 6975
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.316851258277893,
      "learning_rate": 1.7765225352828958e-05,
      "loss": 0.6928,
      "step": 6976
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5385688543319702,
      "learning_rate": 1.776452176288336e-05,
      "loss": 0.8828,
      "step": 6977
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8780978918075562,
      "learning_rate": 1.7763818076134557e-05,
      "loss": 0.664,
      "step": 6978
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4260777235031128,
      "learning_rate": 1.7763114292591317e-05,
      "loss": 0.6371,
      "step": 6979
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.616344928741455,
      "learning_rate": 1.776241041226242e-05,
      "loss": 0.7927,
      "step": 6980
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5015259981155396,
      "learning_rate": 1.7761706435156637e-05,
      "loss": 0.7884,
      "step": 6981
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0932154655456543,
      "learning_rate": 1.7761002361282747e-05,
      "loss": 0.7354,
      "step": 6982
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6251134872436523,
      "learning_rate": 1.7760298190649526e-05,
      "loss": 0.7237,
      "step": 6983
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4715074300765991,
      "learning_rate": 1.775959392326576e-05,
      "loss": 0.8582,
      "step": 6984
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.512372612953186,
      "learning_rate": 1.775888955914022e-05,
      "loss": 0.6509,
      "step": 6985
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8280282020568848,
      "learning_rate": 1.775818509828169e-05,
      "loss": 0.8638,
      "step": 6986
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4918891191482544,
      "learning_rate": 1.7757480540698958e-05,
      "loss": 0.6691,
      "step": 6987
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9944512844085693,
      "learning_rate": 1.7756775886400803e-05,
      "loss": 0.822,
      "step": 6988
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2778438329696655,
      "learning_rate": 1.7756071135396013e-05,
      "loss": 0.5281,
      "step": 6989
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.9110159873962402,
      "learning_rate": 1.7755366287693368e-05,
      "loss": 0.7988,
      "step": 6990
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7427921295166016,
      "learning_rate": 1.7754661343301665e-05,
      "loss": 0.6683,
      "step": 6991
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6055266857147217,
      "learning_rate": 1.775395630222969e-05,
      "loss": 0.9039,
      "step": 6992
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.590600848197937,
      "learning_rate": 1.775325116448623e-05,
      "loss": 0.7359,
      "step": 6993
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4988170862197876,
      "learning_rate": 1.7752545930080075e-05,
      "loss": 0.7711,
      "step": 6994
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3497366905212402,
      "learning_rate": 1.7751840599020024e-05,
      "loss": 0.7873,
      "step": 6995
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7682870626449585,
      "learning_rate": 1.7751135171314862e-05,
      "loss": 0.7164,
      "step": 6996
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7132525444030762,
      "learning_rate": 1.7750429646973392e-05,
      "loss": 0.853,
      "step": 6997
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4995989799499512,
      "learning_rate": 1.7749724026004406e-05,
      "loss": 0.7232,
      "step": 6998
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3435940742492676,
      "learning_rate": 1.7749018308416702e-05,
      "loss": 0.6758,
      "step": 6999
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6827294826507568,
      "learning_rate": 1.7748312494219076e-05,
      "loss": 0.7903,
      "step": 7000
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6280608177185059,
      "learning_rate": 1.7747606583420335e-05,
      "loss": 0.6862,
      "step": 7001
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3248913288116455,
      "learning_rate": 1.774690057602927e-05,
      "loss": 0.6207,
      "step": 7002
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.401033878326416,
      "learning_rate": 1.774619447205469e-05,
      "loss": 0.7479,
      "step": 7003
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4447779655456543,
      "learning_rate": 1.7745488271505395e-05,
      "loss": 0.6967,
      "step": 7004
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7917547225952148,
      "learning_rate": 1.7744781974390193e-05,
      "loss": 0.6195,
      "step": 7005
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6300876140594482,
      "learning_rate": 1.7744075580717884e-05,
      "loss": 0.8394,
      "step": 7006
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2512632608413696,
      "learning_rate": 1.774336909049728e-05,
      "loss": 0.583,
      "step": 7007
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.738525629043579,
      "learning_rate": 1.7742662503737183e-05,
      "loss": 0.6933,
      "step": 7008
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4450451135635376,
      "learning_rate": 1.7741955820446413e-05,
      "loss": 0.8974,
      "step": 7009
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3153444528579712,
      "learning_rate": 1.774124904063377e-05,
      "loss": 0.7896,
      "step": 7010
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.184746265411377,
      "learning_rate": 1.774054216430807e-05,
      "loss": 0.6349,
      "step": 7011
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3607869148254395,
      "learning_rate": 1.773983519147813e-05,
      "loss": 0.716,
      "step": 7012
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.28643000125885,
      "learning_rate": 1.7739128122152753e-05,
      "loss": 0.6687,
      "step": 7013
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7432405948638916,
      "learning_rate": 1.7738420956340764e-05,
      "loss": 0.8332,
      "step": 7014
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3966652154922485,
      "learning_rate": 1.7737713694050974e-05,
      "loss": 0.6835,
      "step": 7015
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9443281888961792,
      "learning_rate": 1.7737006335292205e-05,
      "loss": 0.7978,
      "step": 7016
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3895864486694336,
      "learning_rate": 1.7736298880073276e-05,
      "loss": 0.7875,
      "step": 7017
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3185242414474487,
      "learning_rate": 1.7735591328403004e-05,
      "loss": 0.6577,
      "step": 7018
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6349858045578003,
      "learning_rate": 1.773488368029021e-05,
      "loss": 0.8756,
      "step": 7019
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6475162506103516,
      "learning_rate": 1.773417593574372e-05,
      "loss": 0.7948,
      "step": 7020
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.365494728088379,
      "learning_rate": 1.7733468094772353e-05,
      "loss": 0.7656,
      "step": 7021
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.610030174255371,
      "learning_rate": 1.7732760157384937e-05,
      "loss": 0.6743,
      "step": 7022
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.593176007270813,
      "learning_rate": 1.77320521235903e-05,
      "loss": 0.6687,
      "step": 7023
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3724071979522705,
      "learning_rate": 1.7731343993397264e-05,
      "loss": 0.7807,
      "step": 7024
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.343837857246399,
      "learning_rate": 1.7730635766814663e-05,
      "loss": 0.6295,
      "step": 7025
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.371742844581604,
      "learning_rate": 1.7729927443851327e-05,
      "loss": 0.8008,
      "step": 7026
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4161790609359741,
      "learning_rate": 1.772921902451608e-05,
      "loss": 0.7593,
      "step": 7027
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.812920331954956,
      "learning_rate": 1.772851050881776e-05,
      "loss": 0.7845,
      "step": 7028
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7271273136138916,
      "learning_rate": 1.7727801896765195e-05,
      "loss": 0.6532,
      "step": 7029
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3255412578582764,
      "learning_rate": 1.7727093188367228e-05,
      "loss": 0.7923,
      "step": 7030
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.754511833190918,
      "learning_rate": 1.772638438363269e-05,
      "loss": 0.9287,
      "step": 7031
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.496191382408142,
      "learning_rate": 1.7725675482570415e-05,
      "loss": 0.6111,
      "step": 7032
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6039531230926514,
      "learning_rate": 1.7724966485189245e-05,
      "loss": 0.8657,
      "step": 7033
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.385822057723999,
      "learning_rate": 1.772425739149802e-05,
      "loss": 0.675,
      "step": 7034
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8179717063903809,
      "learning_rate": 1.7723548201505576e-05,
      "loss": 0.6393,
      "step": 7035
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.46415114402771,
      "learning_rate": 1.7722838915220756e-05,
      "loss": 0.6437,
      "step": 7036
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7278120517730713,
      "learning_rate": 1.772212953265241e-05,
      "loss": 0.6509,
      "step": 7037
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3522930145263672,
      "learning_rate": 1.7721420053809375e-05,
      "loss": 0.762,
      "step": 7038
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.437567949295044,
      "learning_rate": 1.77207104787005e-05,
      "loss": 0.8033,
      "step": 7039
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7685195207595825,
      "learning_rate": 1.7720000807334624e-05,
      "loss": 0.7571,
      "step": 7040
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5725759267807007,
      "learning_rate": 1.7719291039720604e-05,
      "loss": 0.6099,
      "step": 7041
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5198880434036255,
      "learning_rate": 1.7718581175867284e-05,
      "loss": 0.6419,
      "step": 7042
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3046553134918213,
      "learning_rate": 1.7717871215783516e-05,
      "loss": 0.7704,
      "step": 7043
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6878734827041626,
      "learning_rate": 1.771716115947815e-05,
      "loss": 0.658,
      "step": 7044
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6179089546203613,
      "learning_rate": 1.771645100696004e-05,
      "loss": 0.8167,
      "step": 7045
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.528869390487671,
      "learning_rate": 1.771574075823804e-05,
      "loss": 0.7699,
      "step": 7046
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.53611421585083,
      "learning_rate": 1.7715030413321003e-05,
      "loss": 0.6665,
      "step": 7047
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6856931447982788,
      "learning_rate": 1.771431997221779e-05,
      "loss": 0.8001,
      "step": 7048
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5581504106521606,
      "learning_rate": 1.7713609434937246e-05,
      "loss": 0.6496,
      "step": 7049
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4997931718826294,
      "learning_rate": 1.7712898801488246e-05,
      "loss": 0.7286,
      "step": 7050
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7666070461273193,
      "learning_rate": 1.7712188071879636e-05,
      "loss": 0.7661,
      "step": 7051
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4767473936080933,
      "learning_rate": 1.771147724612029e-05,
      "loss": 0.6819,
      "step": 7052
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.315893292427063,
      "learning_rate": 1.7710766324219056e-05,
      "loss": 0.6773,
      "step": 7053
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4602829217910767,
      "learning_rate": 1.7710055306184804e-05,
      "loss": 0.6802,
      "step": 7054
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.9397464990615845,
      "learning_rate": 1.77093441920264e-05,
      "loss": 0.7797,
      "step": 7055
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.624794840812683,
      "learning_rate": 1.770863298175271e-05,
      "loss": 0.7551,
      "step": 7056
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8823554515838623,
      "learning_rate": 1.77079216753726e-05,
      "loss": 0.9182,
      "step": 7057
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4755864143371582,
      "learning_rate": 1.7707210272894933e-05,
      "loss": 0.7464,
      "step": 7058
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6629961729049683,
      "learning_rate": 1.7706498774328588e-05,
      "loss": 0.6442,
      "step": 7059
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4178011417388916,
      "learning_rate": 1.7705787179682424e-05,
      "loss": 0.736,
      "step": 7060
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.614406704902649,
      "learning_rate": 1.7705075488965323e-05,
      "loss": 0.6611,
      "step": 7061
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4113000631332397,
      "learning_rate": 1.7704363702186154e-05,
      "loss": 0.7032,
      "step": 7062
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6892367601394653,
      "learning_rate": 1.770365181935379e-05,
      "loss": 0.8112,
      "step": 7063
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6831387281417847,
      "learning_rate": 1.7702939840477108e-05,
      "loss": 0.7062,
      "step": 7064
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6559247970581055,
      "learning_rate": 1.7702227765564985e-05,
      "loss": 0.6328,
      "step": 7065
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2414016723632812,
      "learning_rate": 1.7701515594626295e-05,
      "loss": 0.7531,
      "step": 7066
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5250955820083618,
      "learning_rate": 1.7700803327669918e-05,
      "loss": 0.8066,
      "step": 7067
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.71602201461792,
      "learning_rate": 1.770009096470474e-05,
      "loss": 0.8368,
      "step": 7068
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.49367356300354,
      "learning_rate": 1.7699378505739637e-05,
      "loss": 0.6942,
      "step": 7069
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6260287761688232,
      "learning_rate": 1.769866595078349e-05,
      "loss": 0.6804,
      "step": 7070
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.422178030014038,
      "learning_rate": 1.7697953299845188e-05,
      "loss": 0.6911,
      "step": 7071
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7710316181182861,
      "learning_rate": 1.769724055293361e-05,
      "loss": 0.6836,
      "step": 7072
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4442906379699707,
      "learning_rate": 1.7696527710057646e-05,
      "loss": 0.7489,
      "step": 7073
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.405421257019043,
      "learning_rate": 1.7695814771226184e-05,
      "loss": 0.7161,
      "step": 7074
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4506993293762207,
      "learning_rate": 1.769510173644811e-05,
      "loss": 0.7224,
      "step": 7075
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.324249505996704,
      "learning_rate": 1.769438860573231e-05,
      "loss": 0.6036,
      "step": 7076
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3900337219238281,
      "learning_rate": 1.7693675379087686e-05,
      "loss": 0.8299,
      "step": 7077
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7810516357421875,
      "learning_rate": 1.7692962056523116e-05,
      "loss": 0.7094,
      "step": 7078
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3989492654800415,
      "learning_rate": 1.7692248638047508e-05,
      "loss": 0.6589,
      "step": 7079
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4982999563217163,
      "learning_rate": 1.7691535123669745e-05,
      "loss": 0.6764,
      "step": 7080
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.491114854812622,
      "learning_rate": 1.769082151339873e-05,
      "loss": 0.6757,
      "step": 7081
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6780897378921509,
      "learning_rate": 1.7690107807243352e-05,
      "loss": 0.7872,
      "step": 7082
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.413025140762329,
      "learning_rate": 1.7689394005212516e-05,
      "loss": 0.7183,
      "step": 7083
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4963762760162354,
      "learning_rate": 1.7688680107315118e-05,
      "loss": 0.7427,
      "step": 7084
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2057785987854004,
      "learning_rate": 1.7687966113560058e-05,
      "loss": 0.6714,
      "step": 7085
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3479595184326172,
      "learning_rate": 1.768725202395624e-05,
      "loss": 0.6692,
      "step": 7086
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4652009010314941,
      "learning_rate": 1.7686537838512564e-05,
      "loss": 0.7251,
      "step": 7087
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.457298994064331,
      "learning_rate": 1.768582355723794e-05,
      "loss": 0.5969,
      "step": 7088
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.576357364654541,
      "learning_rate": 1.7685109180141267e-05,
      "loss": 0.6974,
      "step": 7089
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5456793308258057,
      "learning_rate": 1.7684394707231453e-05,
      "loss": 0.695,
      "step": 7090
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.886030673980713,
      "learning_rate": 1.7683680138517405e-05,
      "loss": 0.9169,
      "step": 7091
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4810949563980103,
      "learning_rate": 1.7682965474008037e-05,
      "loss": 0.7676,
      "step": 7092
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5527347326278687,
      "learning_rate": 1.768225071371225e-05,
      "loss": 0.6365,
      "step": 7093
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5722428560256958,
      "learning_rate": 1.768153585763896e-05,
      "loss": 0.814,
      "step": 7094
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6999973058700562,
      "learning_rate": 1.7680820905797084e-05,
      "loss": 0.7843,
      "step": 7095
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.652327060699463,
      "learning_rate": 1.7680105858195528e-05,
      "loss": 0.6426,
      "step": 7096
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4155166149139404,
      "learning_rate": 1.7679390714843208e-05,
      "loss": 0.72,
      "step": 7097
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.372258186340332,
      "learning_rate": 1.7678675475749044e-05,
      "loss": 0.7218,
      "step": 7098
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3892649412155151,
      "learning_rate": 1.767796014092195e-05,
      "loss": 0.753,
      "step": 7099
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4804731607437134,
      "learning_rate": 1.7677244710370843e-05,
      "loss": 0.7353,
      "step": 7100
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4670103788375854,
      "learning_rate": 1.7676529184104646e-05,
      "loss": 0.7399,
      "step": 7101
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6145840883255005,
      "learning_rate": 1.767581356213228e-05,
      "loss": 0.8037,
      "step": 7102
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5461920499801636,
      "learning_rate": 1.7675097844462664e-05,
      "loss": 0.7314,
      "step": 7103
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.764650583267212,
      "learning_rate": 1.7674382031104726e-05,
      "loss": 0.7376,
      "step": 7104
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3260421752929688,
      "learning_rate": 1.767366612206738e-05,
      "loss": 0.5829,
      "step": 7105
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3801933526992798,
      "learning_rate": 1.767295011735956e-05,
      "loss": 0.7562,
      "step": 7106
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5364283323287964,
      "learning_rate": 1.7672234016990196e-05,
      "loss": 0.6401,
      "step": 7107
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4777615070343018,
      "learning_rate": 1.7671517820968204e-05,
      "loss": 0.6472,
      "step": 7108
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7772849798202515,
      "learning_rate": 1.7670801529302527e-05,
      "loss": 0.7738,
      "step": 7109
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4958821535110474,
      "learning_rate": 1.7670085142002083e-05,
      "loss": 0.822,
      "step": 7110
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2614121437072754,
      "learning_rate": 1.766936865907581e-05,
      "loss": 0.8256,
      "step": 7111
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.425530195236206,
      "learning_rate": 1.766865208053264e-05,
      "loss": 0.6848,
      "step": 7112
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5104517936706543,
      "learning_rate": 1.7667935406381505e-05,
      "loss": 0.6633,
      "step": 7113
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5763438940048218,
      "learning_rate": 1.7667218636631344e-05,
      "loss": 0.5536,
      "step": 7114
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.530341386795044,
      "learning_rate": 1.7666501771291085e-05,
      "loss": 0.6428,
      "step": 7115
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3791550397872925,
      "learning_rate": 1.7665784810369677e-05,
      "loss": 0.8209,
      "step": 7116
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3481745719909668,
      "learning_rate": 1.766506775387605e-05,
      "loss": 0.7172,
      "step": 7117
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3832992315292358,
      "learning_rate": 1.7664350601819145e-05,
      "loss": 0.7532,
      "step": 7118
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5349336862564087,
      "learning_rate": 1.7663633354207906e-05,
      "loss": 0.656,
      "step": 7119
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6703585386276245,
      "learning_rate": 1.7662916011051273e-05,
      "loss": 0.8196,
      "step": 7120
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5513999462127686,
      "learning_rate": 1.766219857235819e-05,
      "loss": 0.7421,
      "step": 7121
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8720623254776,
      "learning_rate": 1.7661481038137602e-05,
      "loss": 0.6515,
      "step": 7122
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5270558595657349,
      "learning_rate": 1.7660763408398454e-05,
      "loss": 0.7258,
      "step": 7123
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.483134388923645,
      "learning_rate": 1.766004568314969e-05,
      "loss": 0.6225,
      "step": 7124
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5124024152755737,
      "learning_rate": 1.7659327862400266e-05,
      "loss": 0.7388,
      "step": 7125
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9640984535217285,
      "learning_rate": 1.7658609946159125e-05,
      "loss": 0.7086,
      "step": 7126
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4382388591766357,
      "learning_rate": 1.7657891934435222e-05,
      "loss": 0.8712,
      "step": 7127
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.434423565864563,
      "learning_rate": 1.7657173827237503e-05,
      "loss": 0.7288,
      "step": 7128
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4846155643463135,
      "learning_rate": 1.7656455624574922e-05,
      "loss": 0.618,
      "step": 7129
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5575512647628784,
      "learning_rate": 1.7655737326456438e-05,
      "loss": 0.725,
      "step": 7130
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5320168733596802,
      "learning_rate": 1.7655018932891e-05,
      "loss": 0.8777,
      "step": 7131
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.362838864326477,
      "learning_rate": 1.765430044388757e-05,
      "loss": 0.7132,
      "step": 7132
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6670845746994019,
      "learning_rate": 1.7653581859455107e-05,
      "loss": 0.5801,
      "step": 7133
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7515772581100464,
      "learning_rate": 1.7652863179602563e-05,
      "loss": 0.7332,
      "step": 7134
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9826537370681763,
      "learning_rate": 1.76521444043389e-05,
      "loss": 0.796,
      "step": 7135
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.779025673866272,
      "learning_rate": 1.765142553367308e-05,
      "loss": 0.7912,
      "step": 7136
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5259159803390503,
      "learning_rate": 1.7650706567614068e-05,
      "loss": 0.6848,
      "step": 7137
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5995365381240845,
      "learning_rate": 1.7649987506170826e-05,
      "loss": 0.7311,
      "step": 7138
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3963289260864258,
      "learning_rate": 1.7649268349352318e-05,
      "loss": 0.767,
      "step": 7139
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6360671520233154,
      "learning_rate": 1.764854909716751e-05,
      "loss": 0.7538,
      "step": 7140
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5667814016342163,
      "learning_rate": 1.764782974962537e-05,
      "loss": 0.7375,
      "step": 7141
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7011363506317139,
      "learning_rate": 1.7647110306734865e-05,
      "loss": 0.7158,
      "step": 7142
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4627647399902344,
      "learning_rate": 1.7646390768504967e-05,
      "loss": 0.6818,
      "step": 7143
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.942669153213501,
      "learning_rate": 1.7645671134944648e-05,
      "loss": 0.5708,
      "step": 7144
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5117965936660767,
      "learning_rate": 1.7644951406062872e-05,
      "loss": 0.6514,
      "step": 7145
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7445170879364014,
      "learning_rate": 1.764423158186862e-05,
      "loss": 0.6199,
      "step": 7146
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7034775018692017,
      "learning_rate": 1.7643511662370866e-05,
      "loss": 0.7426,
      "step": 7147
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6055570840835571,
      "learning_rate": 1.7642791647578577e-05,
      "loss": 0.8247,
      "step": 7148
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6659867763519287,
      "learning_rate": 1.764207153750074e-05,
      "loss": 0.6314,
      "step": 7149
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.600855827331543,
      "learning_rate": 1.7641351332146328e-05,
      "loss": 0.8315,
      "step": 7150
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.519023060798645,
      "learning_rate": 1.7640631031524323e-05,
      "loss": 0.688,
      "step": 7151
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.472499966621399,
      "learning_rate": 1.76399106356437e-05,
      "loss": 0.6385,
      "step": 7152
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3928933143615723,
      "learning_rate": 1.763919014451344e-05,
      "loss": 0.7889,
      "step": 7153
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8096576929092407,
      "learning_rate": 1.7638469558142538e-05,
      "loss": 0.7426,
      "step": 7154
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4085980653762817,
      "learning_rate": 1.7637748876539964e-05,
      "loss": 0.6937,
      "step": 7155
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8393291234970093,
      "learning_rate": 1.763702809971471e-05,
      "loss": 0.7492,
      "step": 7156
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6326299905776978,
      "learning_rate": 1.7636307227675756e-05,
      "loss": 0.7883,
      "step": 7157
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8708566427230835,
      "learning_rate": 1.76355862604321e-05,
      "loss": 0.8415,
      "step": 7158
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.583630084991455,
      "learning_rate": 1.7634865197992722e-05,
      "loss": 0.7406,
      "step": 7159
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5092873573303223,
      "learning_rate": 1.7634144040366614e-05,
      "loss": 0.7529,
      "step": 7160
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2669076919555664,
      "learning_rate": 1.7633422787562764e-05,
      "loss": 0.7036,
      "step": 7161
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1390858888626099,
      "learning_rate": 1.7632701439590168e-05,
      "loss": 0.7335,
      "step": 7162
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4530704021453857,
      "learning_rate": 1.7631979996457822e-05,
      "loss": 0.7894,
      "step": 7163
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4359110593795776,
      "learning_rate": 1.763125845817471e-05,
      "loss": 0.7601,
      "step": 7164
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4905502796173096,
      "learning_rate": 1.763053682474984e-05,
      "loss": 0.656,
      "step": 7165
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4115992784500122,
      "learning_rate": 1.7629815096192202e-05,
      "loss": 0.6075,
      "step": 7166
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7455891370773315,
      "learning_rate": 1.7629093272510795e-05,
      "loss": 0.7275,
      "step": 7167
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3912945985794067,
      "learning_rate": 1.762837135371462e-05,
      "loss": 0.693,
      "step": 7168
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.436492681503296,
      "learning_rate": 1.7627649339812672e-05,
      "loss": 0.735,
      "step": 7169
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.718997836112976,
      "learning_rate": 1.762692723081396e-05,
      "loss": 0.6661,
      "step": 7170
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.673985481262207,
      "learning_rate": 1.7626205026727485e-05,
      "loss": 0.7829,
      "step": 7171
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0389604568481445,
      "learning_rate": 1.762548272756225e-05,
      "loss": 0.769,
      "step": 7172
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.880408763885498,
      "learning_rate": 1.7624760333327256e-05,
      "loss": 0.7477,
      "step": 7173
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.0232551097869873,
      "learning_rate": 1.7624037844031514e-05,
      "loss": 0.7745,
      "step": 7174
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6062493324279785,
      "learning_rate": 1.7623315259684034e-05,
      "loss": 0.8241,
      "step": 7175
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3499845266342163,
      "learning_rate": 1.7622592580293816e-05,
      "loss": 0.7502,
      "step": 7176
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3731086254119873,
      "learning_rate": 1.762186980586988e-05,
      "loss": 0.7593,
      "step": 7177
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8749102354049683,
      "learning_rate": 1.762114693642123e-05,
      "loss": 0.7969,
      "step": 7178
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5376793146133423,
      "learning_rate": 1.762042397195688e-05,
      "loss": 0.7715,
      "step": 7179
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3346920013427734,
      "learning_rate": 1.7619700912485846e-05,
      "loss": 0.7703,
      "step": 7180
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7083622217178345,
      "learning_rate": 1.761897775801714e-05,
      "loss": 0.8141,
      "step": 7181
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4317396879196167,
      "learning_rate": 1.761825450855978e-05,
      "loss": 0.8,
      "step": 7182
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9444106817245483,
      "learning_rate": 1.7617531164122778e-05,
      "loss": 0.7235,
      "step": 7183
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.673534631729126,
      "learning_rate": 1.761680772471516e-05,
      "loss": 0.6046,
      "step": 7184
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5748311281204224,
      "learning_rate": 1.7616084190345937e-05,
      "loss": 0.7652,
      "step": 7185
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.376196026802063,
      "learning_rate": 1.7615360561024136e-05,
      "loss": 0.6486,
      "step": 7186
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.8569550514221191,
      "learning_rate": 1.7614636836758777e-05,
      "loss": 0.7164,
      "step": 7187
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4411722421646118,
      "learning_rate": 1.761391301755888e-05,
      "loss": 0.781,
      "step": 7188
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4050763845443726,
      "learning_rate": 1.7613189103433476e-05,
      "loss": 0.8318,
      "step": 7189
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4184930324554443,
      "learning_rate": 1.7612465094391585e-05,
      "loss": 0.5816,
      "step": 7190
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6103547811508179,
      "learning_rate": 1.761174099044223e-05,
      "loss": 0.8185,
      "step": 7191
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6926225423812866,
      "learning_rate": 1.761101679159445e-05,
      "loss": 0.9271,
      "step": 7192
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.841125726699829,
      "learning_rate": 1.7610292497857263e-05,
      "loss": 0.8252,
      "step": 7193
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6803840398788452,
      "learning_rate": 1.7609568109239702e-05,
      "loss": 0.5921,
      "step": 7194
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5768312215805054,
      "learning_rate": 1.76088436257508e-05,
      "loss": 0.7353,
      "step": 7195
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4860689640045166,
      "learning_rate": 1.760811904739959e-05,
      "loss": 0.6286,
      "step": 7196
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5090874433517456,
      "learning_rate": 1.76073943741951e-05,
      "loss": 0.6911,
      "step": 7197
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4116535186767578,
      "learning_rate": 1.7606669606146373e-05,
      "loss": 0.8198,
      "step": 7198
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.702834963798523,
      "learning_rate": 1.7605944743262443e-05,
      "loss": 0.8282,
      "step": 7199
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4500694274902344,
      "learning_rate": 1.7605219785552342e-05,
      "loss": 0.7323,
      "step": 7200
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5091853141784668,
      "learning_rate": 1.7604494733025112e-05,
      "loss": 0.4962,
      "step": 7201
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5921521186828613,
      "learning_rate": 1.760376958568979e-05,
      "loss": 0.7258,
      "step": 7202
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6180248260498047,
      "learning_rate": 1.760304434355542e-05,
      "loss": 0.7408,
      "step": 7203
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.553188443183899,
      "learning_rate": 1.7602319006631043e-05,
      "loss": 0.8053,
      "step": 7204
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6926765441894531,
      "learning_rate": 1.7601593574925702e-05,
      "loss": 0.7663,
      "step": 7205
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8229721784591675,
      "learning_rate": 1.760086804844844e-05,
      "loss": 0.6532,
      "step": 7206
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6605224609375,
      "learning_rate": 1.7600142427208303e-05,
      "loss": 0.7027,
      "step": 7207
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5956661701202393,
      "learning_rate": 1.7599416711214338e-05,
      "loss": 0.791,
      "step": 7208
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7572509050369263,
      "learning_rate": 1.7598690900475594e-05,
      "loss": 0.8249,
      "step": 7209
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6258299350738525,
      "learning_rate": 1.7597964995001114e-05,
      "loss": 0.7627,
      "step": 7210
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5172789096832275,
      "learning_rate": 1.7597238994799958e-05,
      "loss": 0.8078,
      "step": 7211
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6038470268249512,
      "learning_rate": 1.759651289988117e-05,
      "loss": 0.7263,
      "step": 7212
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.851133108139038,
      "learning_rate": 1.7595786710253803e-05,
      "loss": 0.6568,
      "step": 7213
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.8746581077575684,
      "learning_rate": 1.7595060425926913e-05,
      "loss": 0.7692,
      "step": 7214
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7131999731063843,
      "learning_rate": 1.7594334046909554e-05,
      "loss": 0.6879,
      "step": 7215
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3728710412979126,
      "learning_rate": 1.759360757321078e-05,
      "loss": 0.7691,
      "step": 7216
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.518043875694275,
      "learning_rate": 1.7592881004839655e-05,
      "loss": 0.8057,
      "step": 7217
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9630645513534546,
      "learning_rate": 1.759215434180523e-05,
      "loss": 0.5887,
      "step": 7218
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.778977632522583,
      "learning_rate": 1.7591427584116566e-05,
      "loss": 0.8017,
      "step": 7219
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.566606044769287,
      "learning_rate": 1.7590700731782725e-05,
      "loss": 0.767,
      "step": 7220
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5175729990005493,
      "learning_rate": 1.7589973784812772e-05,
      "loss": 0.7473,
      "step": 7221
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7624105215072632,
      "learning_rate": 1.7589246743215765e-05,
      "loss": 0.7805,
      "step": 7222
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.691689372062683,
      "learning_rate": 1.758851960700077e-05,
      "loss": 0.7675,
      "step": 7223
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6333765983581543,
      "learning_rate": 1.7587792376176852e-05,
      "loss": 0.7701,
      "step": 7224
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5344651937484741,
      "learning_rate": 1.758706505075308e-05,
      "loss": 0.8794,
      "step": 7225
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2380648851394653,
      "learning_rate": 1.758633763073852e-05,
      "loss": 0.8475,
      "step": 7226
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.631584644317627,
      "learning_rate": 1.758561011614224e-05,
      "loss": 0.8928,
      "step": 7227
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4036659002304077,
      "learning_rate": 1.7584882506973314e-05,
      "loss": 0.6565,
      "step": 7228
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4859426021575928,
      "learning_rate": 1.758415480324081e-05,
      "loss": 0.6738,
      "step": 7229
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2977946996688843,
      "learning_rate": 1.75834270049538e-05,
      "loss": 0.6643,
      "step": 7230
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.349250316619873,
      "learning_rate": 1.758269911212136e-05,
      "loss": 0.742,
      "step": 7231
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5856518745422363,
      "learning_rate": 1.7581971124752564e-05,
      "loss": 0.7457,
      "step": 7232
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8147002458572388,
      "learning_rate": 1.758124304285649e-05,
      "loss": 0.6315,
      "step": 7233
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.751473307609558,
      "learning_rate": 1.758051486644221e-05,
      "loss": 0.6508,
      "step": 7234
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9313071966171265,
      "learning_rate": 1.7579786595518808e-05,
      "loss": 0.7061,
      "step": 7235
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7819385528564453,
      "learning_rate": 1.7579058230095363e-05,
      "loss": 0.6314,
      "step": 7236
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6850910186767578,
      "learning_rate": 1.7578329770180957e-05,
      "loss": 0.8097,
      "step": 7237
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5357800722122192,
      "learning_rate": 1.757760121578466e-05,
      "loss": 0.6774,
      "step": 7238
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6012119054794312,
      "learning_rate": 1.757687256691557e-05,
      "loss": 0.9665,
      "step": 7239
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7858192920684814,
      "learning_rate": 1.7576143823582766e-05,
      "loss": 0.7487,
      "step": 7240
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3206764459609985,
      "learning_rate": 1.7575414985795334e-05,
      "loss": 0.7033,
      "step": 7241
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.486962914466858,
      "learning_rate": 1.7574686053562357e-05,
      "loss": 0.6268,
      "step": 7242
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5119520425796509,
      "learning_rate": 1.757395702689293e-05,
      "loss": 0.6476,
      "step": 7243
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7514948844909668,
      "learning_rate": 1.7573227905796136e-05,
      "loss": 0.7495,
      "step": 7244
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3092008829116821,
      "learning_rate": 1.7572498690281067e-05,
      "loss": 0.7203,
      "step": 7245
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5086290836334229,
      "learning_rate": 1.7571769380356816e-05,
      "loss": 0.7767,
      "step": 7246
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8614693880081177,
      "learning_rate": 1.7571039976032472e-05,
      "loss": 0.7572,
      "step": 7247
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4559824466705322,
      "learning_rate": 1.757031047731713e-05,
      "loss": 0.7668,
      "step": 7248
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5755517482757568,
      "learning_rate": 1.756958088421989e-05,
      "loss": 0.6715,
      "step": 7249
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.419250249862671,
      "learning_rate": 1.756885119674984e-05,
      "loss": 0.8059,
      "step": 7250
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5966719388961792,
      "learning_rate": 1.7568121414916083e-05,
      "loss": 0.8368,
      "step": 7251
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3445333242416382,
      "learning_rate": 1.7567391538727716e-05,
      "loss": 0.6029,
      "step": 7252
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6909518241882324,
      "learning_rate": 1.756666156819384e-05,
      "loss": 0.7763,
      "step": 7253
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5643647909164429,
      "learning_rate": 1.756593150332355e-05,
      "loss": 0.8481,
      "step": 7254
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5491682291030884,
      "learning_rate": 1.7565201344125954e-05,
      "loss": 0.6348,
      "step": 7255
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2795263528823853,
      "learning_rate": 1.7564471090610152e-05,
      "loss": 0.6273,
      "step": 7256
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.704532265663147,
      "learning_rate": 1.7563740742785254e-05,
      "loss": 0.7624,
      "step": 7257
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5025596618652344,
      "learning_rate": 1.7563010300660357e-05,
      "loss": 0.801,
      "step": 7258
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.424607276916504,
      "learning_rate": 1.7562279764244573e-05,
      "loss": 0.749,
      "step": 7259
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1889452934265137,
      "learning_rate": 1.7561549133547008e-05,
      "loss": 0.7334,
      "step": 7260
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5757890939712524,
      "learning_rate": 1.7560818408576773e-05,
      "loss": 0.7675,
      "step": 7261
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.293244481086731,
      "learning_rate": 1.756008758934298e-05,
      "loss": 0.6131,
      "step": 7262
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.250439167022705,
      "learning_rate": 1.7559356675854734e-05,
      "loss": 0.6573,
      "step": 7263
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3514654636383057,
      "learning_rate": 1.755862566812115e-05,
      "loss": 0.6307,
      "step": 7264
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9065903425216675,
      "learning_rate": 1.7557894566151347e-05,
      "loss": 0.7502,
      "step": 7265
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9570789337158203,
      "learning_rate": 1.7557163369954434e-05,
      "loss": 0.5703,
      "step": 7266
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.590195655822754,
      "learning_rate": 1.7556432079539525e-05,
      "loss": 0.682,
      "step": 7267
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6202605962753296,
      "learning_rate": 1.7555700694915746e-05,
      "loss": 0.7012,
      "step": 7268
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5283217430114746,
      "learning_rate": 1.7554969216092205e-05,
      "loss": 0.6432,
      "step": 7269
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5314769744873047,
      "learning_rate": 1.7554237643078035e-05,
      "loss": 0.589,
      "step": 7270
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.45774245262146,
      "learning_rate": 1.755350597588234e-05,
      "loss": 0.7817,
      "step": 7271
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6043376922607422,
      "learning_rate": 1.755277421451426e-05,
      "loss": 0.8736,
      "step": 7272
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5963358879089355,
      "learning_rate": 1.7552042358982902e-05,
      "loss": 0.6962,
      "step": 7273
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2399091720581055,
      "learning_rate": 1.75513104092974e-05,
      "loss": 0.7287,
      "step": 7274
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6120142936706543,
      "learning_rate": 1.7550578365466877e-05,
      "loss": 0.6905,
      "step": 7275
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6878339052200317,
      "learning_rate": 1.7549846227500458e-05,
      "loss": 0.7682,
      "step": 7276
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4379390478134155,
      "learning_rate": 1.7549113995407275e-05,
      "loss": 0.6277,
      "step": 7277
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6584653854370117,
      "learning_rate": 1.7548381669196453e-05,
      "loss": 0.8245,
      "step": 7278
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3901646137237549,
      "learning_rate": 1.754764924887712e-05,
      "loss": 0.7008,
      "step": 7279
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3619040250778198,
      "learning_rate": 1.7546916734458416e-05,
      "loss": 0.7462,
      "step": 7280
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4799339771270752,
      "learning_rate": 1.7546184125949467e-05,
      "loss": 0.6966,
      "step": 7281
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4729355573654175,
      "learning_rate": 1.7545451423359406e-05,
      "loss": 0.4971,
      "step": 7282
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6349163055419922,
      "learning_rate": 1.754471862669737e-05,
      "loss": 0.7921,
      "step": 7283
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.482758641242981,
      "learning_rate": 1.75439857359725e-05,
      "loss": 0.66,
      "step": 7284
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.494100570678711,
      "learning_rate": 1.7543252751193924e-05,
      "loss": 0.6759,
      "step": 7285
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2869179248809814,
      "learning_rate": 1.7542519672370788e-05,
      "loss": 0.7435,
      "step": 7286
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3504923582077026,
      "learning_rate": 1.754178649951223e-05,
      "loss": 0.7454,
      "step": 7287
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4773656129837036,
      "learning_rate": 1.7541053232627382e-05,
      "loss": 0.8199,
      "step": 7288
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6418004035949707,
      "learning_rate": 1.7540319871725395e-05,
      "loss": 0.6683,
      "step": 7289
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3708003759384155,
      "learning_rate": 1.7539586416815412e-05,
      "loss": 0.6463,
      "step": 7290
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5243092775344849,
      "learning_rate": 1.7538852867906574e-05,
      "loss": 0.6662,
      "step": 7291
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.74100661277771,
      "learning_rate": 1.7538119225008027e-05,
      "loss": 0.6232,
      "step": 7292
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.415968418121338,
      "learning_rate": 1.753738548812892e-05,
      "loss": 0.6383,
      "step": 7293
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.253904104232788,
      "learning_rate": 1.7536651657278397e-05,
      "loss": 0.6636,
      "step": 7294
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.34884512424469,
      "learning_rate": 1.753591773246561e-05,
      "loss": 0.7411,
      "step": 7295
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3731142282485962,
      "learning_rate": 1.753518371369971e-05,
      "loss": 0.709,
      "step": 7296
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3588752746582031,
      "learning_rate": 1.7534449600989846e-05,
      "loss": 0.7083,
      "step": 7297
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5030460357666016,
      "learning_rate": 1.753371539434517e-05,
      "loss": 0.7304,
      "step": 7298
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7329599857330322,
      "learning_rate": 1.7532981093774836e-05,
      "loss": 0.6377,
      "step": 7299
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5839893817901611,
      "learning_rate": 1.7532246699288e-05,
      "loss": 0.603,
      "step": 7300
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3613708019256592,
      "learning_rate": 1.753151221089382e-05,
      "loss": 0.6912,
      "step": 7301
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3859307765960693,
      "learning_rate": 1.7530777628601445e-05,
      "loss": 0.6965,
      "step": 7302
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2786362171173096,
      "learning_rate": 1.7530042952420043e-05,
      "loss": 0.5532,
      "step": 7303
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5019874572753906,
      "learning_rate": 1.7529308182358764e-05,
      "loss": 0.723,
      "step": 7304
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3904613256454468,
      "learning_rate": 1.752857331842678e-05,
      "loss": 0.746,
      "step": 7305
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5145984888076782,
      "learning_rate": 1.752783836063324e-05,
      "loss": 0.498,
      "step": 7306
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8647016286849976,
      "learning_rate": 1.7527103308987318e-05,
      "loss": 0.82,
      "step": 7307
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3467230796813965,
      "learning_rate": 1.7526368163498177e-05,
      "loss": 0.8326,
      "step": 7308
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5632412433624268,
      "learning_rate": 1.7525632924174973e-05,
      "loss": 0.7084,
      "step": 7309
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4881181716918945,
      "learning_rate": 1.7524897591026882e-05,
      "loss": 0.7474,
      "step": 7310
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7667655944824219,
      "learning_rate": 1.7524162164063065e-05,
      "loss": 0.6457,
      "step": 7311
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5327409505844116,
      "learning_rate": 1.7523426643292697e-05,
      "loss": 0.7012,
      "step": 7312
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.293297529220581,
      "learning_rate": 1.7522691028724947e-05,
      "loss": 0.731,
      "step": 7313
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6161198616027832,
      "learning_rate": 1.7521955320368982e-05,
      "loss": 0.5678,
      "step": 7314
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.37949800491333,
      "learning_rate": 1.7521219518233978e-05,
      "loss": 0.691,
      "step": 7315
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5048418045043945,
      "learning_rate": 1.7520483622329106e-05,
      "loss": 0.6216,
      "step": 7316
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.493194580078125,
      "learning_rate": 1.751974763266354e-05,
      "loss": 0.6365,
      "step": 7317
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3138786554336548,
      "learning_rate": 1.7519011549246463e-05,
      "loss": 0.6234,
      "step": 7318
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5206167697906494,
      "learning_rate": 1.7518275372087045e-05,
      "loss": 0.6748,
      "step": 7319
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8203063011169434,
      "learning_rate": 1.7517539101194467e-05,
      "loss": 0.6854,
      "step": 7320
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7080602645874023,
      "learning_rate": 1.7516802736577903e-05,
      "loss": 0.7367,
      "step": 7321
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5441033840179443,
      "learning_rate": 1.751606627824654e-05,
      "loss": 0.6526,
      "step": 7322
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4182437658309937,
      "learning_rate": 1.751532972620956e-05,
      "loss": 0.739,
      "step": 7323
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.509295105934143,
      "learning_rate": 1.751459308047614e-05,
      "loss": 0.7443,
      "step": 7324
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7203855514526367,
      "learning_rate": 1.751385634105547e-05,
      "loss": 0.7572,
      "step": 7325
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.435446858406067,
      "learning_rate": 1.7513119507956733e-05,
      "loss": 0.8933,
      "step": 7326
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6050622463226318,
      "learning_rate": 1.7512382581189115e-05,
      "loss": 0.6302,
      "step": 7327
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6369472742080688,
      "learning_rate": 1.75116455607618e-05,
      "loss": 0.7476,
      "step": 7328
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4058369398117065,
      "learning_rate": 1.7510908446683984e-05,
      "loss": 0.6057,
      "step": 7329
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8643276691436768,
      "learning_rate": 1.751017123896485e-05,
      "loss": 0.7536,
      "step": 7330
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3744401931762695,
      "learning_rate": 1.75094339376136e-05,
      "loss": 0.7559,
      "step": 7331
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3932913541793823,
      "learning_rate": 1.7508696542639413e-05,
      "loss": 0.6474,
      "step": 7332
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3029251098632812,
      "learning_rate": 1.7507959054051487e-05,
      "loss": 0.6829,
      "step": 7333
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7349472045898438,
      "learning_rate": 1.750722147185902e-05,
      "loss": 0.6413,
      "step": 7334
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5345487594604492,
      "learning_rate": 1.7506483796071203e-05,
      "loss": 0.723,
      "step": 7335
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.324277400970459,
      "learning_rate": 1.7505746026697235e-05,
      "loss": 0.7246,
      "step": 7336
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6465246677398682,
      "learning_rate": 1.7505008163746317e-05,
      "loss": 0.6365,
      "step": 7337
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5963075160980225,
      "learning_rate": 1.7504270207227643e-05,
      "loss": 0.6959,
      "step": 7338
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.522381067276001,
      "learning_rate": 1.7503532157150416e-05,
      "loss": 0.7361,
      "step": 7339
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3729019165039062,
      "learning_rate": 1.7502794013523838e-05,
      "loss": 0.7032,
      "step": 7340
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.454277753829956,
      "learning_rate": 1.750205577635711e-05,
      "loss": 0.8458,
      "step": 7341
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5049047470092773,
      "learning_rate": 1.7501317445659437e-05,
      "loss": 0.7308,
      "step": 7342
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5031591653823853,
      "learning_rate": 1.7500579021440026e-05,
      "loss": 0.7723,
      "step": 7343
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2898521423339844,
      "learning_rate": 1.7499840503708075e-05,
      "loss": 0.6886,
      "step": 7344
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6467742919921875,
      "learning_rate": 1.7499101892472805e-05,
      "loss": 0.6766,
      "step": 7345
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5987212657928467,
      "learning_rate": 1.7498363187743413e-05,
      "loss": 0.7042,
      "step": 7346
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4097108840942383,
      "learning_rate": 1.7497624389529116e-05,
      "loss": 0.6698,
      "step": 7347
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3674228191375732,
      "learning_rate": 1.749688549783912e-05,
      "loss": 0.7369,
      "step": 7348
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3745800256729126,
      "learning_rate": 1.7496146512682635e-05,
      "loss": 0.6023,
      "step": 7349
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6072992086410522,
      "learning_rate": 1.7495407434068882e-05,
      "loss": 0.7871,
      "step": 7350
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5404853820800781,
      "learning_rate": 1.749466826200707e-05,
      "loss": 0.7037,
      "step": 7351
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6064417362213135,
      "learning_rate": 1.7493928996506414e-05,
      "loss": 0.7328,
      "step": 7352
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3604423999786377,
      "learning_rate": 1.7493189637576133e-05,
      "loss": 0.7507,
      "step": 7353
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5570404529571533,
      "learning_rate": 1.7492450185225447e-05,
      "loss": 0.635,
      "step": 7354
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3442174196243286,
      "learning_rate": 1.7491710639463567e-05,
      "loss": 0.6644,
      "step": 7355
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.488574743270874,
      "learning_rate": 1.7490971000299724e-05,
      "loss": 0.6702,
      "step": 7356
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.761513590812683,
      "learning_rate": 1.7490231267743132e-05,
      "loss": 0.8262,
      "step": 7357
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4448292255401611,
      "learning_rate": 1.748949144180302e-05,
      "loss": 0.7495,
      "step": 7358
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4851144552230835,
      "learning_rate": 1.7488751522488598e-05,
      "loss": 0.7085,
      "step": 7359
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.481195330619812,
      "learning_rate": 1.7488011509809106e-05,
      "loss": 0.5826,
      "step": 7360
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6192219257354736,
      "learning_rate": 1.7487271403773764e-05,
      "loss": 0.7236,
      "step": 7361
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5511977672576904,
      "learning_rate": 1.7486531204391795e-05,
      "loss": 0.7178,
      "step": 7362
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3828355073928833,
      "learning_rate": 1.7485790911672434e-05,
      "loss": 0.6804,
      "step": 7363
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6342360973358154,
      "learning_rate": 1.7485050525624913e-05,
      "loss": 0.8434,
      "step": 7364
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3629045486450195,
      "learning_rate": 1.748431004625845e-05,
      "loss": 0.7269,
      "step": 7365
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5077489614486694,
      "learning_rate": 1.7483569473582285e-05,
      "loss": 0.6101,
      "step": 7366
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3805503845214844,
      "learning_rate": 1.7482828807605655e-05,
      "loss": 0.7048,
      "step": 7367
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.361072063446045,
      "learning_rate": 1.748208804833779e-05,
      "loss": 0.7399,
      "step": 7368
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3816118240356445,
      "learning_rate": 1.7481347195787922e-05,
      "loss": 0.7744,
      "step": 7369
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5950472354888916,
      "learning_rate": 1.748060624996529e-05,
      "loss": 0.5948,
      "step": 7370
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3489739894866943,
      "learning_rate": 1.7479865210879135e-05,
      "loss": 0.6654,
      "step": 7371
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7800339460372925,
      "learning_rate": 1.747912407853869e-05,
      "loss": 0.6593,
      "step": 7372
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4626373052597046,
      "learning_rate": 1.74783828529532e-05,
      "loss": 0.8231,
      "step": 7373
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2686917781829834,
      "learning_rate": 1.7477641534131904e-05,
      "loss": 0.6871,
      "step": 7374
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.57474946975708,
      "learning_rate": 1.7476900122084043e-05,
      "loss": 0.7749,
      "step": 7375
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7316187620162964,
      "learning_rate": 1.747615861681886e-05,
      "loss": 0.6952,
      "step": 7376
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.398018717765808,
      "learning_rate": 1.7475417018345607e-05,
      "loss": 0.6629,
      "step": 7377
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7575212717056274,
      "learning_rate": 1.7474675326673522e-05,
      "loss": 0.8524,
      "step": 7378
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4699283838272095,
      "learning_rate": 1.747393354181185e-05,
      "loss": 0.7914,
      "step": 7379
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5274542570114136,
      "learning_rate": 1.7473191663769847e-05,
      "loss": 0.7268,
      "step": 7380
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5386745929718018,
      "learning_rate": 1.747244969255676e-05,
      "loss": 0.7702,
      "step": 7381
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.638643741607666,
      "learning_rate": 1.7471707628181836e-05,
      "loss": 0.804,
      "step": 7382
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.341225504875183,
      "learning_rate": 1.7470965470654328e-05,
      "loss": 0.6322,
      "step": 7383
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5682990550994873,
      "learning_rate": 1.747022321998349e-05,
      "loss": 0.7161,
      "step": 7384
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5431632995605469,
      "learning_rate": 1.7469480876178576e-05,
      "loss": 0.7725,
      "step": 7385
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6528470516204834,
      "learning_rate": 1.746873843924884e-05,
      "loss": 0.7073,
      "step": 7386
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5422422885894775,
      "learning_rate": 1.746799590920354e-05,
      "loss": 0.6485,
      "step": 7387
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3794677257537842,
      "learning_rate": 1.746725328605193e-05,
      "loss": 0.7267,
      "step": 7388
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5629533529281616,
      "learning_rate": 1.746651056980327e-05,
      "loss": 0.8419,
      "step": 7389
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.242156982421875,
      "learning_rate": 1.746576776046682e-05,
      "loss": 0.6045,
      "step": 7390
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.803473711013794,
      "learning_rate": 1.746502485805184e-05,
      "loss": 0.663,
      "step": 7391
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.9527459144592285,
      "learning_rate": 1.74642818625676e-05,
      "loss": 0.6778,
      "step": 7392
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6031713485717773,
      "learning_rate": 1.746353877402335e-05,
      "loss": 0.7742,
      "step": 7393
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7186201810836792,
      "learning_rate": 1.7462795592428363e-05,
      "loss": 0.7613,
      "step": 7394
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5188372135162354,
      "learning_rate": 1.7462052317791904e-05,
      "loss": 0.7567,
      "step": 7395
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5372295379638672,
      "learning_rate": 1.7461308950123234e-05,
      "loss": 0.6333,
      "step": 7396
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3989629745483398,
      "learning_rate": 1.7460565489431627e-05,
      "loss": 0.6148,
      "step": 7397
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6202232837677002,
      "learning_rate": 1.7459821935726353e-05,
      "loss": 0.7665,
      "step": 7398
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.469141960144043,
      "learning_rate": 1.7459078289016674e-05,
      "loss": 0.7485,
      "step": 7399
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6221518516540527,
      "learning_rate": 1.7458334549311866e-05,
      "loss": 0.6184,
      "step": 7400
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5746257305145264,
      "learning_rate": 1.7457590716621203e-05,
      "loss": 0.6577,
      "step": 7401
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5946921110153198,
      "learning_rate": 1.745684679095396e-05,
      "loss": 0.6901,
      "step": 7402
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.73493492603302,
      "learning_rate": 1.745610277231941e-05,
      "loss": 0.7098,
      "step": 7403
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.739919662475586,
      "learning_rate": 1.745535866072682e-05,
      "loss": 0.7859,
      "step": 7404
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3002849817276,
      "learning_rate": 1.7454614456185483e-05,
      "loss": 0.6694,
      "step": 7405
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3805328607559204,
      "learning_rate": 1.7453870158704667e-05,
      "loss": 0.6481,
      "step": 7406
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.383555293083191,
      "learning_rate": 1.7453125768293655e-05,
      "loss": 0.6826,
      "step": 7407
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.537702202796936,
      "learning_rate": 1.7452381284961725e-05,
      "loss": 0.7274,
      "step": 7408
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4171117544174194,
      "learning_rate": 1.745163670871816e-05,
      "loss": 0.6781,
      "step": 7409
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7290191650390625,
      "learning_rate": 1.745089203957225e-05,
      "loss": 0.7716,
      "step": 7410
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5527102947235107,
      "learning_rate": 1.7450147277533265e-05,
      "loss": 0.7748,
      "step": 7411
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5945788621902466,
      "learning_rate": 1.74494024226105e-05,
      "loss": 0.678,
      "step": 7412
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6056089401245117,
      "learning_rate": 1.7448657474813236e-05,
      "loss": 0.6903,
      "step": 7413
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6160510778427124,
      "learning_rate": 1.7447912434150768e-05,
      "loss": 0.7888,
      "step": 7414
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4995568990707397,
      "learning_rate": 1.744716730063238e-05,
      "loss": 0.7982,
      "step": 7415
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5834357738494873,
      "learning_rate": 1.7446422074267357e-05,
      "loss": 0.7541,
      "step": 7416
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7705857753753662,
      "learning_rate": 1.7445676755065002e-05,
      "loss": 0.7864,
      "step": 7417
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4118516445159912,
      "learning_rate": 1.7444931343034596e-05,
      "loss": 0.8004,
      "step": 7418
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.681648850440979,
      "learning_rate": 1.7444185838185438e-05,
      "loss": 0.8864,
      "step": 7419
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7576581239700317,
      "learning_rate": 1.744344024052682e-05,
      "loss": 0.6783,
      "step": 7420
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2470505237579346,
      "learning_rate": 1.744269455006804e-05,
      "loss": 0.6207,
      "step": 7421
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4145972728729248,
      "learning_rate": 1.744194876681839e-05,
      "loss": 0.6592,
      "step": 7422
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.617977499961853,
      "learning_rate": 1.7441202890787177e-05,
      "loss": 0.6703,
      "step": 7423
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5041415691375732,
      "learning_rate": 1.744045692198369e-05,
      "loss": 0.7032,
      "step": 7424
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2822366952896118,
      "learning_rate": 1.7439710860417236e-05,
      "loss": 0.6279,
      "step": 7425
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4693621397018433,
      "learning_rate": 1.7438964706097112e-05,
      "loss": 0.6287,
      "step": 7426
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3841323852539062,
      "learning_rate": 1.7438218459032627e-05,
      "loss": 0.5693,
      "step": 7427
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4072082042694092,
      "learning_rate": 1.7437472119233078e-05,
      "loss": 0.6395,
      "step": 7428
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4986366033554077,
      "learning_rate": 1.7436725686707774e-05,
      "loss": 0.7998,
      "step": 7429
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.457840919494629,
      "learning_rate": 1.7435979161466018e-05,
      "loss": 0.7633,
      "step": 7430
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2617887258529663,
      "learning_rate": 1.7435232543517118e-05,
      "loss": 0.6908,
      "step": 7431
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4634445905685425,
      "learning_rate": 1.7434485832870386e-05,
      "loss": 0.7239,
      "step": 7432
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.351328730583191,
      "learning_rate": 1.743373902953513e-05,
      "loss": 0.7088,
      "step": 7433
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5927684307098389,
      "learning_rate": 1.7432992133520655e-05,
      "loss": 0.6147,
      "step": 7434
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3545153141021729,
      "learning_rate": 1.743224514483628e-05,
      "loss": 0.7827,
      "step": 7435
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6343501806259155,
      "learning_rate": 1.743149806349131e-05,
      "loss": 0.6812,
      "step": 7436
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4840589761734009,
      "learning_rate": 1.743075088949507e-05,
      "loss": 0.7769,
      "step": 7437
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5599809885025024,
      "learning_rate": 1.7430003622856868e-05,
      "loss": 0.6759,
      "step": 7438
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5491418838500977,
      "learning_rate": 1.7429256263586024e-05,
      "loss": 0.4756,
      "step": 7439
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.890052318572998,
      "learning_rate": 1.7428508811691852e-05,
      "loss": 0.7558,
      "step": 7440
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9461487531661987,
      "learning_rate": 1.742776126718367e-05,
      "loss": 0.9499,
      "step": 7441
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.345510721206665,
      "learning_rate": 1.7427013630070804e-05,
      "loss": 0.6932,
      "step": 7442
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.493655800819397,
      "learning_rate": 1.742626590036257e-05,
      "loss": 0.6721,
      "step": 7443
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.725547194480896,
      "learning_rate": 1.7425518078068293e-05,
      "loss": 0.7387,
      "step": 7444
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3381643295288086,
      "learning_rate": 1.7424770163197293e-05,
      "loss": 0.7311,
      "step": 7445
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5630589723587036,
      "learning_rate": 1.7424022155758898e-05,
      "loss": 0.8047,
      "step": 7446
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.357473373413086,
      "learning_rate": 1.742327405576243e-05,
      "loss": 0.6279,
      "step": 7447
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.843988299369812,
      "learning_rate": 1.7422525863217222e-05,
      "loss": 0.7138,
      "step": 7448
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6992050409317017,
      "learning_rate": 1.7421777578132594e-05,
      "loss": 0.7846,
      "step": 7449
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.795265555381775,
      "learning_rate": 1.7421029200517883e-05,
      "loss": 0.6967,
      "step": 7450
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2337915897369385,
      "learning_rate": 1.7420280730382414e-05,
      "loss": 0.581,
      "step": 7451
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.563795566558838,
      "learning_rate": 1.7419532167735518e-05,
      "loss": 0.8516,
      "step": 7452
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8028311729431152,
      "learning_rate": 1.741878351258653e-05,
      "loss": 0.7151,
      "step": 7453
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.577576756477356,
      "learning_rate": 1.7418034764944787e-05,
      "loss": 0.7746,
      "step": 7454
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1728209257125854,
      "learning_rate": 1.7417285924819617e-05,
      "loss": 0.6616,
      "step": 7455
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.625077486038208,
      "learning_rate": 1.7416536992220364e-05,
      "loss": 0.8543,
      "step": 7456
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3579195737838745,
      "learning_rate": 1.7415787967156356e-05,
      "loss": 0.7685,
      "step": 7457
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7348530292510986,
      "learning_rate": 1.7415038849636936e-05,
      "loss": 0.8195,
      "step": 7458
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6188526153564453,
      "learning_rate": 1.7414289639671445e-05,
      "loss": 0.5832,
      "step": 7459
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4515933990478516,
      "learning_rate": 1.7413540337269222e-05,
      "loss": 0.6858,
      "step": 7460
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6561932563781738,
      "learning_rate": 1.7412790942439607e-05,
      "loss": 0.7531,
      "step": 7461
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5306211709976196,
      "learning_rate": 1.741204145519195e-05,
      "loss": 0.7318,
      "step": 7462
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4155715703964233,
      "learning_rate": 1.7411291875535585e-05,
      "loss": 0.7103,
      "step": 7463
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.563236951828003,
      "learning_rate": 1.7410542203479866e-05,
      "loss": 0.5967,
      "step": 7464
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6559727191925049,
      "learning_rate": 1.7409792439034133e-05,
      "loss": 0.5802,
      "step": 7465
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.383172869682312,
      "learning_rate": 1.7409042582207738e-05,
      "loss": 0.6641,
      "step": 7466
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3895803689956665,
      "learning_rate": 1.7408292633010028e-05,
      "loss": 0.647,
      "step": 7467
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6251288652420044,
      "learning_rate": 1.7407542591450355e-05,
      "loss": 0.6388,
      "step": 7468
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3627548217773438,
      "learning_rate": 1.7406792457538066e-05,
      "loss": 0.7249,
      "step": 7469
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4340956211090088,
      "learning_rate": 1.7406042231282516e-05,
      "loss": 0.8361,
      "step": 7470
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.571402668952942,
      "learning_rate": 1.7405291912693057e-05,
      "loss": 0.7546,
      "step": 7471
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.430417537689209,
      "learning_rate": 1.7404541501779046e-05,
      "loss": 0.7183,
      "step": 7472
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6010223627090454,
      "learning_rate": 1.7403790998549835e-05,
      "loss": 0.7138,
      "step": 7473
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6929279565811157,
      "learning_rate": 1.7403040403014785e-05,
      "loss": 0.8081,
      "step": 7474
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.754992127418518,
      "learning_rate": 1.740228971518325e-05,
      "loss": 0.6983,
      "step": 7475
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8319453001022339,
      "learning_rate": 1.740153893506459e-05,
      "loss": 0.7628,
      "step": 7476
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5660070180892944,
      "learning_rate": 1.7400788062668168e-05,
      "loss": 0.5511,
      "step": 7477
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8100742101669312,
      "learning_rate": 1.7400037098003342e-05,
      "loss": 0.762,
      "step": 7478
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5665875673294067,
      "learning_rate": 1.7399286041079478e-05,
      "loss": 0.6047,
      "step": 7479
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4326112270355225,
      "learning_rate": 1.7398534891905936e-05,
      "loss": 0.6909,
      "step": 7480
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3627033233642578,
      "learning_rate": 1.7397783650492082e-05,
      "loss": 0.7076,
      "step": 7481
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5638247728347778,
      "learning_rate": 1.7397032316847284e-05,
      "loss": 0.6424,
      "step": 7482
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.631466269493103,
      "learning_rate": 1.7396280890980906e-05,
      "loss": 0.7379,
      "step": 7483
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.516119122505188,
      "learning_rate": 1.739552937290232e-05,
      "loss": 0.7257,
      "step": 7484
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.448697566986084,
      "learning_rate": 1.7394777762620895e-05,
      "loss": 0.6505,
      "step": 7485
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6387231349945068,
      "learning_rate": 1.7394026060145998e-05,
      "loss": 0.8103,
      "step": 7486
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5417637825012207,
      "learning_rate": 1.7393274265486997e-05,
      "loss": 0.7389,
      "step": 7487
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3918644189834595,
      "learning_rate": 1.739252237865328e-05,
      "loss": 0.7072,
      "step": 7488
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.567844271659851,
      "learning_rate": 1.7391770399654207e-05,
      "loss": 0.6933,
      "step": 7489
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6048678159713745,
      "learning_rate": 1.739101832849916e-05,
      "loss": 0.599,
      "step": 7490
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.211761236190796,
      "learning_rate": 1.7390266165197512e-05,
      "loss": 0.7105,
      "step": 7491
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5315921306610107,
      "learning_rate": 1.7389513909758643e-05,
      "loss": 0.7173,
      "step": 7492
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4888206720352173,
      "learning_rate": 1.738876156219193e-05,
      "loss": 0.6541,
      "step": 7493
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5082881450653076,
      "learning_rate": 1.7388009122506753e-05,
      "loss": 0.636,
      "step": 7494
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.465760588645935,
      "learning_rate": 1.7387256590712495e-05,
      "loss": 0.5607,
      "step": 7495
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9476207494735718,
      "learning_rate": 1.7386503966818532e-05,
      "loss": 0.7594,
      "step": 7496
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6860243082046509,
      "learning_rate": 1.7385751250834257e-05,
      "loss": 0.6089,
      "step": 7497
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5671601295471191,
      "learning_rate": 1.7384998442769046e-05,
      "loss": 0.76,
      "step": 7498
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4796335697174072,
      "learning_rate": 1.738424554263229e-05,
      "loss": 0.6677,
      "step": 7499
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.549772024154663,
      "learning_rate": 1.7383492550433368e-05,
      "loss": 0.7601,
      "step": 7500
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4021340608596802,
      "learning_rate": 1.738273946618168e-05,
      "loss": 0.7035,
      "step": 7501
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6882190704345703,
      "learning_rate": 1.7381986289886603e-05,
      "loss": 0.7386,
      "step": 7502
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4236184358596802,
      "learning_rate": 1.7381233021557533e-05,
      "loss": 0.5427,
      "step": 7503
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4648195505142212,
      "learning_rate": 1.7380479661203863e-05,
      "loss": 0.8408,
      "step": 7504
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5401623249053955,
      "learning_rate": 1.737972620883498e-05,
      "loss": 0.8658,
      "step": 7505
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7930327653884888,
      "learning_rate": 1.7378972664460282e-05,
      "loss": 0.6461,
      "step": 7506
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.431509017944336,
      "learning_rate": 1.7378219028089162e-05,
      "loss": 0.7555,
      "step": 7507
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.767796516418457,
      "learning_rate": 1.7377465299731016e-05,
      "loss": 0.7165,
      "step": 7508
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5640639066696167,
      "learning_rate": 1.737671147939524e-05,
      "loss": 0.6892,
      "step": 7509
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4866441488265991,
      "learning_rate": 1.7375957567091235e-05,
      "loss": 0.6614,
      "step": 7510
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4895930290222168,
      "learning_rate": 1.7375203562828397e-05,
      "loss": 0.728,
      "step": 7511
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4775677919387817,
      "learning_rate": 1.7374449466616127e-05,
      "loss": 0.6223,
      "step": 7512
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3321185111999512,
      "learning_rate": 1.7373695278463832e-05,
      "loss": 0.8093,
      "step": 7513
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3137295246124268,
      "learning_rate": 1.7372940998380906e-05,
      "loss": 0.6559,
      "step": 7514
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5793660879135132,
      "learning_rate": 1.7372186626376763e-05,
      "loss": 0.7668,
      "step": 7515
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4242960214614868,
      "learning_rate": 1.7371432162460796e-05,
      "loss": 0.6636,
      "step": 7516
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7578152418136597,
      "learning_rate": 1.7370677606642422e-05,
      "loss": 0.6667,
      "step": 7517
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2159639596939087,
      "learning_rate": 1.736992295893104e-05,
      "loss": 0.7813,
      "step": 7518
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3983724117279053,
      "learning_rate": 1.7369168219336063e-05,
      "loss": 0.7036,
      "step": 7519
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4337671995162964,
      "learning_rate": 1.73684133878669e-05,
      "loss": 0.8145,
      "step": 7520
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.324746012687683,
      "learning_rate": 1.736765846453296e-05,
      "loss": 0.4878,
      "step": 7521
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3002947568893433,
      "learning_rate": 1.7366903449343663e-05,
      "loss": 0.7587,
      "step": 7522
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4614925384521484,
      "learning_rate": 1.736614834230841e-05,
      "loss": 0.6166,
      "step": 7523
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4110954999923706,
      "learning_rate": 1.7365393143436623e-05,
      "loss": 0.7115,
      "step": 7524
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4609994888305664,
      "learning_rate": 1.7364637852737715e-05,
      "loss": 0.748,
      "step": 7525
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5019646883010864,
      "learning_rate": 1.7363882470221104e-05,
      "loss": 0.554,
      "step": 7526
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4896659851074219,
      "learning_rate": 1.7363126995896204e-05,
      "loss": 0.8258,
      "step": 7527
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3860684633255005,
      "learning_rate": 1.7362371429772434e-05,
      "loss": 0.8159,
      "step": 7528
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.460044264793396,
      "learning_rate": 1.7361615771859218e-05,
      "loss": 0.7204,
      "step": 7529
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6459838151931763,
      "learning_rate": 1.7360860022165974e-05,
      "loss": 0.6987,
      "step": 7530
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4230859279632568,
      "learning_rate": 1.736010418070213e-05,
      "loss": 0.6822,
      "step": 7531
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.182645320892334,
      "learning_rate": 1.73593482474771e-05,
      "loss": 0.6581,
      "step": 7532
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4429954290390015,
      "learning_rate": 1.7358592222500317e-05,
      "loss": 0.7729,
      "step": 7533
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3604143857955933,
      "learning_rate": 1.7357836105781198e-05,
      "loss": 0.67,
      "step": 7534
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4356261491775513,
      "learning_rate": 1.735707989732918e-05,
      "loss": 0.7293,
      "step": 7535
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.565341830253601,
      "learning_rate": 1.7356323597153682e-05,
      "loss": 0.6471,
      "step": 7536
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3582946062088013,
      "learning_rate": 1.7355567205264137e-05,
      "loss": 0.7413,
      "step": 7537
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.737076759338379,
      "learning_rate": 1.7354810721669973e-05,
      "loss": 0.722,
      "step": 7538
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4293537139892578,
      "learning_rate": 1.7354054146380627e-05,
      "loss": 0.6947,
      "step": 7539
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6006038188934326,
      "learning_rate": 1.7353297479405527e-05,
      "loss": 0.6849,
      "step": 7540
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4972270727157593,
      "learning_rate": 1.7352540720754107e-05,
      "loss": 0.8573,
      "step": 7541
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.76937735080719,
      "learning_rate": 1.7351783870435802e-05,
      "loss": 0.7788,
      "step": 7542
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8032363653182983,
      "learning_rate": 1.7351026928460045e-05,
      "loss": 0.7886,
      "step": 7543
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2924282550811768,
      "learning_rate": 1.735026989483628e-05,
      "loss": 0.6459,
      "step": 7544
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.69728684425354,
      "learning_rate": 1.7349512769573942e-05,
      "loss": 0.7178,
      "step": 7545
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4691441059112549,
      "learning_rate": 1.734875555268247e-05,
      "loss": 0.7486,
      "step": 7546
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6613142490386963,
      "learning_rate": 1.73479982441713e-05,
      "loss": 0.6378,
      "step": 7547
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4022939205169678,
      "learning_rate": 1.734724084404988e-05,
      "loss": 0.6109,
      "step": 7548
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9805407524108887,
      "learning_rate": 1.734648335232765e-05,
      "loss": 0.7665,
      "step": 7549
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5995131731033325,
      "learning_rate": 1.7345725769014058e-05,
      "loss": 0.8816,
      "step": 7550
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9492801427841187,
      "learning_rate": 1.734496809411854e-05,
      "loss": 0.7405,
      "step": 7551
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6458139419555664,
      "learning_rate": 1.7344210327650555e-05,
      "loss": 0.5875,
      "step": 7552
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8433640003204346,
      "learning_rate": 1.7343452469619538e-05,
      "loss": 0.7296,
      "step": 7553
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6043570041656494,
      "learning_rate": 1.7342694520034946e-05,
      "loss": 0.5827,
      "step": 7554
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7973392009735107,
      "learning_rate": 1.7341936478906224e-05,
      "loss": 0.7593,
      "step": 7555
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5999867916107178,
      "learning_rate": 1.7341178346242822e-05,
      "loss": 0.8096,
      "step": 7556
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3678914308547974,
      "learning_rate": 1.7340420122054197e-05,
      "loss": 0.6681,
      "step": 7557
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5639896392822266,
      "learning_rate": 1.7339661806349798e-05,
      "loss": 0.7437,
      "step": 7558
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.554978847503662,
      "learning_rate": 1.7338903399139085e-05,
      "loss": 0.7066,
      "step": 7559
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.834757924079895,
      "learning_rate": 1.7338144900431503e-05,
      "loss": 0.7891,
      "step": 7560
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6201421022415161,
      "learning_rate": 1.7337386310236516e-05,
      "loss": 0.6539,
      "step": 7561
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4571502208709717,
      "learning_rate": 1.7336627628563577e-05,
      "loss": 0.8483,
      "step": 7562
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4829448461532593,
      "learning_rate": 1.733586885542215e-05,
      "loss": 0.6052,
      "step": 7563
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5906962156295776,
      "learning_rate": 1.7335109990821693e-05,
      "loss": 0.6679,
      "step": 7564
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6323792934417725,
      "learning_rate": 1.733435103477167e-05,
      "loss": 0.7871,
      "step": 7565
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5111321210861206,
      "learning_rate": 1.7333591987281534e-05,
      "loss": 0.7734,
      "step": 7566
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.7752646207809448,
      "learning_rate": 1.7332832848360758e-05,
      "loss": 0.5671,
      "step": 7567
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3029433488845825,
      "learning_rate": 1.7332073618018802e-05,
      "loss": 0.6752,
      "step": 7568
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8127305507659912,
      "learning_rate": 1.7331314296265128e-05,
      "loss": 0.6472,
      "step": 7569
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.489520788192749,
      "learning_rate": 1.7330554883109213e-05,
      "loss": 0.7008,
      "step": 7570
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4685955047607422,
      "learning_rate": 1.7329795378560516e-05,
      "loss": 0.8514,
      "step": 7571
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.395392894744873,
      "learning_rate": 1.7329035782628508e-05,
      "loss": 0.7382,
      "step": 7572
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.490224003791809,
      "learning_rate": 1.732827609532266e-05,
      "loss": 0.8741,
      "step": 7573
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4708144664764404,
      "learning_rate": 1.7327516316652446e-05,
      "loss": 0.8441,
      "step": 7574
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.729736328125,
      "learning_rate": 1.7326756446627335e-05,
      "loss": 0.6993,
      "step": 7575
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5982813835144043,
      "learning_rate": 1.73259964852568e-05,
      "loss": 0.7586,
      "step": 7576
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2414112091064453,
      "learning_rate": 1.7325236432550317e-05,
      "loss": 0.7691,
      "step": 7577
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5046013593673706,
      "learning_rate": 1.7324476288517363e-05,
      "loss": 0.6725,
      "step": 7578
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.472497582435608,
      "learning_rate": 1.7323716053167413e-05,
      "loss": 0.6786,
      "step": 7579
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.320422649383545,
      "learning_rate": 1.7322955726509947e-05,
      "loss": 0.6116,
      "step": 7580
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8967976570129395,
      "learning_rate": 1.7322195308554443e-05,
      "loss": 0.7054,
      "step": 7581
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6584094762802124,
      "learning_rate": 1.7321434799310382e-05,
      "loss": 0.7226,
      "step": 7582
    },
    {
      "epoch": 0.78,
      "grad_norm": 3.0722732543945312,
      "learning_rate": 1.7320674198787246e-05,
      "loss": 0.7601,
      "step": 7583
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.728208303451538,
      "learning_rate": 1.7319913506994517e-05,
      "loss": 0.7077,
      "step": 7584
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6355856657028198,
      "learning_rate": 1.7319152723941676e-05,
      "loss": 0.8789,
      "step": 7585
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.1058619022369385,
      "learning_rate": 1.7318391849638213e-05,
      "loss": 0.882,
      "step": 7586
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.937831401824951,
      "learning_rate": 1.7317630884093614e-05,
      "loss": 0.6296,
      "step": 7587
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.8744795322418213,
      "learning_rate": 1.7316869827317363e-05,
      "loss": 0.8904,
      "step": 7588
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.578536868095398,
      "learning_rate": 1.7316108679318947e-05,
      "loss": 0.7179,
      "step": 7589
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4796792268753052,
      "learning_rate": 1.731534744010786e-05,
      "loss": 0.6792,
      "step": 7590
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4064273834228516,
      "learning_rate": 1.7314586109693593e-05,
      "loss": 0.7423,
      "step": 7591
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4394421577453613,
      "learning_rate": 1.731382468808563e-05,
      "loss": 0.7773,
      "step": 7592
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6778688430786133,
      "learning_rate": 1.7313063175293473e-05,
      "loss": 0.661,
      "step": 7593
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4468516111373901,
      "learning_rate": 1.7312301571326613e-05,
      "loss": 0.7386,
      "step": 7594
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.353108525276184,
      "learning_rate": 1.7311539876194542e-05,
      "loss": 0.7092,
      "step": 7595
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4411383867263794,
      "learning_rate": 1.7310778089906764e-05,
      "loss": 0.6739,
      "step": 7596
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.426485300064087,
      "learning_rate": 1.731001621247277e-05,
      "loss": 0.6631,
      "step": 7597
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3878858089447021,
      "learning_rate": 1.7309254243902058e-05,
      "loss": 0.6408,
      "step": 7598
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4548466205596924,
      "learning_rate": 1.730849218420413e-05,
      "loss": 0.7309,
      "step": 7599
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6367603540420532,
      "learning_rate": 1.730773003338849e-05,
      "loss": 0.6714,
      "step": 7600
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3898581266403198,
      "learning_rate": 1.730696779146463e-05,
      "loss": 0.7077,
      "step": 7601
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4397984743118286,
      "learning_rate": 1.7306205458442065e-05,
      "loss": 0.907,
      "step": 7602
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.525158166885376,
      "learning_rate": 1.7305443034330292e-05,
      "loss": 0.7872,
      "step": 7603
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5054042339324951,
      "learning_rate": 1.7304680519138822e-05,
      "loss": 0.7674,
      "step": 7604
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5207765102386475,
      "learning_rate": 1.7303917912877157e-05,
      "loss": 0.5385,
      "step": 7605
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4358011484146118,
      "learning_rate": 1.7303155215554805e-05,
      "loss": 0.7288,
      "step": 7606
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4120142459869385,
      "learning_rate": 1.7302392427181277e-05,
      "loss": 0.7141,
      "step": 7607
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4238723516464233,
      "learning_rate": 1.730162954776608e-05,
      "loss": 0.7311,
      "step": 7608
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6485894918441772,
      "learning_rate": 1.7300866577318727e-05,
      "loss": 0.7459,
      "step": 7609
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4629353284835815,
      "learning_rate": 1.7300103515848732e-05,
      "loss": 0.7046,
      "step": 7610
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4631991386413574,
      "learning_rate": 1.7299340363365606e-05,
      "loss": 0.7688,
      "step": 7611
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4419505596160889,
      "learning_rate": 1.7298577119878863e-05,
      "loss": 0.7331,
      "step": 7612
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2790499925613403,
      "learning_rate": 1.729781378539802e-05,
      "loss": 0.6689,
      "step": 7613
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.436774730682373,
      "learning_rate": 1.729705035993259e-05,
      "loss": 0.715,
      "step": 7614
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4411814212799072,
      "learning_rate": 1.72962868434921e-05,
      "loss": 0.6874,
      "step": 7615
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6125010251998901,
      "learning_rate": 1.7295523236086063e-05,
      "loss": 0.6854,
      "step": 7616
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6353684663772583,
      "learning_rate": 1.7294759537723996e-05,
      "loss": 0.6883,
      "step": 7617
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9152064323425293,
      "learning_rate": 1.7293995748415426e-05,
      "loss": 0.5105,
      "step": 7618
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.60254967212677,
      "learning_rate": 1.7293231868169872e-05,
      "loss": 0.8073,
      "step": 7619
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6444454193115234,
      "learning_rate": 1.729246789699686e-05,
      "loss": 0.6937,
      "step": 7620
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4878734350204468,
      "learning_rate": 1.7291703834905913e-05,
      "loss": 0.7258,
      "step": 7621
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.319667100906372,
      "learning_rate": 1.729093968190656e-05,
      "loss": 0.5856,
      "step": 7622
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5870420932769775,
      "learning_rate": 1.7290175438008324e-05,
      "loss": 0.6437,
      "step": 7623
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.641139268875122,
      "learning_rate": 1.7289411103220734e-05,
      "loss": 0.7722,
      "step": 7624
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4786173105239868,
      "learning_rate": 1.7288646677553318e-05,
      "loss": 0.7337,
      "step": 7625
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4752423763275146,
      "learning_rate": 1.728788216101561e-05,
      "loss": 0.6047,
      "step": 7626
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4034178256988525,
      "learning_rate": 1.728711755361714e-05,
      "loss": 0.5937,
      "step": 7627
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2947025299072266,
      "learning_rate": 1.7286352855367436e-05,
      "loss": 0.7377,
      "step": 7628
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4963266849517822,
      "learning_rate": 1.728558806627604e-05,
      "loss": 0.8349,
      "step": 7629
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3917590379714966,
      "learning_rate": 1.7284823186352484e-05,
      "loss": 0.8353,
      "step": 7630
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6844213008880615,
      "learning_rate": 1.7284058215606302e-05,
      "loss": 0.8072,
      "step": 7631
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5433099269866943,
      "learning_rate": 1.728329315404703e-05,
      "loss": 0.6553,
      "step": 7632
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3386586904525757,
      "learning_rate": 1.7282528001684212e-05,
      "loss": 0.7006,
      "step": 7633
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4782906770706177,
      "learning_rate": 1.7281762758527383e-05,
      "loss": 0.6478,
      "step": 7634
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3624546527862549,
      "learning_rate": 1.7280997424586084e-05,
      "loss": 0.7661,
      "step": 7635
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5761772394180298,
      "learning_rate": 1.7280231999869857e-05,
      "loss": 0.5896,
      "step": 7636
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.356408715248108,
      "learning_rate": 1.727946648438825e-05,
      "loss": 0.7249,
      "step": 7637
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5245389938354492,
      "learning_rate": 1.7278700878150798e-05,
      "loss": 0.7891,
      "step": 7638
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0667740106582642,
      "learning_rate": 1.7277935181167047e-05,
      "loss": 0.8063,
      "step": 7639
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.622146725654602,
      "learning_rate": 1.727716939344655e-05,
      "loss": 0.5947,
      "step": 7640
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5974425077438354,
      "learning_rate": 1.727640351499885e-05,
      "loss": 0.7787,
      "step": 7641
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6322219371795654,
      "learning_rate": 1.7275637545833495e-05,
      "loss": 0.7308,
      "step": 7642
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3841240406036377,
      "learning_rate": 1.7274871485960038e-05,
      "loss": 0.5969,
      "step": 7643
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5433577299118042,
      "learning_rate": 1.7274105335388027e-05,
      "loss": 0.6356,
      "step": 7644
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3707765340805054,
      "learning_rate": 1.7273339094127017e-05,
      "loss": 0.698,
      "step": 7645
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6910380125045776,
      "learning_rate": 1.7272572762186556e-05,
      "loss": 0.6812,
      "step": 7646
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.322608232498169,
      "learning_rate": 1.72718063395762e-05,
      "loss": 0.7588,
      "step": 7647
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6847048997879028,
      "learning_rate": 1.7271039826305503e-05,
      "loss": 0.6297,
      "step": 7648
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7055238485336304,
      "learning_rate": 1.7270273222384027e-05,
      "loss": 0.6272,
      "step": 7649
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.517059326171875,
      "learning_rate": 1.7269506527821325e-05,
      "loss": 0.7431,
      "step": 7650
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.544755458831787,
      "learning_rate": 1.7268739742626958e-05,
      "loss": 0.7827,
      "step": 7651
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4658324718475342,
      "learning_rate": 1.7267972866810484e-05,
      "loss": 0.7549,
      "step": 7652
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3043535947799683,
      "learning_rate": 1.7267205900381464e-05,
      "loss": 0.7948,
      "step": 7653
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.9335018396377563,
      "learning_rate": 1.7266438843349457e-05,
      "loss": 0.7501,
      "step": 7654
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3435484170913696,
      "learning_rate": 1.7265671695724034e-05,
      "loss": 0.6237,
      "step": 7655
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.845627784729004,
      "learning_rate": 1.7264904457514752e-05,
      "loss": 0.7725,
      "step": 7656
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6720068454742432,
      "learning_rate": 1.726413712873118e-05,
      "loss": 0.9329,
      "step": 7657
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1832979917526245,
      "learning_rate": 1.7263369709382887e-05,
      "loss": 0.7334,
      "step": 7658
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5882476568222046,
      "learning_rate": 1.7262602199479436e-05,
      "loss": 0.6379,
      "step": 7659
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5883502960205078,
      "learning_rate": 1.7261834599030395e-05,
      "loss": 0.6794,
      "step": 7660
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4784659147262573,
      "learning_rate": 1.7261066908045337e-05,
      "loss": 0.7092,
      "step": 7661
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.673882007598877,
      "learning_rate": 1.7260299126533836e-05,
      "loss": 0.7771,
      "step": 7662
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.33388090133667,
      "learning_rate": 1.7259531254505456e-05,
      "loss": 0.5571,
      "step": 7663
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4686689376831055,
      "learning_rate": 1.7258763291969777e-05,
      "loss": 0.6678,
      "step": 7664
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4179593324661255,
      "learning_rate": 1.7257995238936373e-05,
      "loss": 0.7289,
      "step": 7665
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.502036452293396,
      "learning_rate": 1.7257227095414818e-05,
      "loss": 0.6409,
      "step": 7666
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8496698141098022,
      "learning_rate": 1.7256458861414687e-05,
      "loss": 0.8905,
      "step": 7667
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7825145721435547,
      "learning_rate": 1.7255690536945563e-05,
      "loss": 0.824,
      "step": 7668
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4826068878173828,
      "learning_rate": 1.725492212201702e-05,
      "loss": 0.7843,
      "step": 7669
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7785160541534424,
      "learning_rate": 1.7254153616638637e-05,
      "loss": 0.7639,
      "step": 7670
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5588059425354004,
      "learning_rate": 1.7253385020820004e-05,
      "loss": 0.6428,
      "step": 7671
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8595550060272217,
      "learning_rate": 1.7252616334570694e-05,
      "loss": 0.5942,
      "step": 7672
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5562412738800049,
      "learning_rate": 1.7251847557900295e-05,
      "loss": 0.7726,
      "step": 7673
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6475944519042969,
      "learning_rate": 1.725107869081839e-05,
      "loss": 0.65,
      "step": 7674
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6339647769927979,
      "learning_rate": 1.7250309733334564e-05,
      "loss": 0.7688,
      "step": 7675
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5323066711425781,
      "learning_rate": 1.7249540685458405e-05,
      "loss": 0.7117,
      "step": 7676
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.456489086151123,
      "learning_rate": 1.7248771547199504e-05,
      "loss": 0.8027,
      "step": 7677
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6326091289520264,
      "learning_rate": 1.7248002318567445e-05,
      "loss": 0.6377,
      "step": 7678
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.2388129234313965,
      "learning_rate": 1.724723299957182e-05,
      "loss": 0.7485,
      "step": 7679
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6647948026657104,
      "learning_rate": 1.7246463590222224e-05,
      "loss": 0.7112,
      "step": 7680
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2811532020568848,
      "learning_rate": 1.7245694090528246e-05,
      "loss": 0.7143,
      "step": 7681
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4714652299880981,
      "learning_rate": 1.7244924500499477e-05,
      "loss": 0.8093,
      "step": 7682
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5425410270690918,
      "learning_rate": 1.724415482014552e-05,
      "loss": 0.8613,
      "step": 7683
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.505050778388977,
      "learning_rate": 1.7243385049475964e-05,
      "loss": 0.7727,
      "step": 7684
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4459991455078125,
      "learning_rate": 1.7242615188500407e-05,
      "loss": 0.8191,
      "step": 7685
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.666084885597229,
      "learning_rate": 1.724184523722845e-05,
      "loss": 0.7782,
      "step": 7686
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5660459995269775,
      "learning_rate": 1.7241075195669688e-05,
      "loss": 0.6549,
      "step": 7687
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6924712657928467,
      "learning_rate": 1.7240305063833723e-05,
      "loss": 0.6383,
      "step": 7688
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5928070545196533,
      "learning_rate": 1.7239534841730158e-05,
      "loss": 0.6983,
      "step": 7689
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7390837669372559,
      "learning_rate": 1.7238764529368595e-05,
      "loss": 0.7737,
      "step": 7690
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4633028507232666,
      "learning_rate": 1.723799412675864e-05,
      "loss": 0.7529,
      "step": 7691
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.241044282913208,
      "learning_rate": 1.723722363390989e-05,
      "loss": 0.5544,
      "step": 7692
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4162873029708862,
      "learning_rate": 1.723645305083196e-05,
      "loss": 0.7001,
      "step": 7693
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5299943685531616,
      "learning_rate": 1.7235682377534456e-05,
      "loss": 0.6139,
      "step": 7694
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4039578437805176,
      "learning_rate": 1.723491161402698e-05,
      "loss": 0.6194,
      "step": 7695
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4454396963119507,
      "learning_rate": 1.723414076031915e-05,
      "loss": 0.6331,
      "step": 7696
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4024252891540527,
      "learning_rate": 1.7233369816420565e-05,
      "loss": 0.812,
      "step": 7697
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1397461891174316,
      "learning_rate": 1.723259878234085e-05,
      "loss": 0.6712,
      "step": 7698
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.576901912689209,
      "learning_rate": 1.7231827658089608e-05,
      "loss": 0.6772,
      "step": 7699
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7162387371063232,
      "learning_rate": 1.723105644367646e-05,
      "loss": 0.7915,
      "step": 7700
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9572110176086426,
      "learning_rate": 1.7230285139111012e-05,
      "loss": 0.7621,
      "step": 7701
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.570308804512024,
      "learning_rate": 1.722951374440289e-05,
      "loss": 0.6029,
      "step": 7702
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4802007675170898,
      "learning_rate": 1.7228742259561705e-05,
      "loss": 0.6166,
      "step": 7703
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7532296180725098,
      "learning_rate": 1.7227970684597073e-05,
      "loss": 0.7637,
      "step": 7704
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.725976824760437,
      "learning_rate": 1.7227199019518625e-05,
      "loss": 0.8033,
      "step": 7705
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8347198963165283,
      "learning_rate": 1.7226427264335966e-05,
      "loss": 0.8941,
      "step": 7706
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.435627818107605,
      "learning_rate": 1.7225655419058733e-05,
      "loss": 0.6949,
      "step": 7707
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.827338457107544,
      "learning_rate": 1.7224883483696537e-05,
      "loss": 0.7302,
      "step": 7708
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4221152067184448,
      "learning_rate": 1.722411145825901e-05,
      "loss": 0.678,
      "step": 7709
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4991099834442139,
      "learning_rate": 1.722333934275577e-05,
      "loss": 0.7541,
      "step": 7710
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.698838233947754,
      "learning_rate": 1.722256713719645e-05,
      "loss": 0.6895,
      "step": 7711
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8143279552459717,
      "learning_rate": 1.7221794841590678e-05,
      "loss": 0.7875,
      "step": 7712
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.586456537246704,
      "learning_rate": 1.7221022455948074e-05,
      "loss": 0.8356,
      "step": 7713
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.202039361000061,
      "learning_rate": 1.7220249980278273e-05,
      "loss": 0.6268,
      "step": 7714
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5738799571990967,
      "learning_rate": 1.721947741459091e-05,
      "loss": 0.6482,
      "step": 7715
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7588380575180054,
      "learning_rate": 1.7218704758895607e-05,
      "loss": 0.6163,
      "step": 7716
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5686147212982178,
      "learning_rate": 1.7217932013202005e-05,
      "loss": 0.8959,
      "step": 7717
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2595453262329102,
      "learning_rate": 1.7217159177519735e-05,
      "loss": 0.7411,
      "step": 7718
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7198114395141602,
      "learning_rate": 1.7216386251858432e-05,
      "loss": 0.8552,
      "step": 7719
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7660143375396729,
      "learning_rate": 1.7215613236227735e-05,
      "loss": 0.5358,
      "step": 7720
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.546769142150879,
      "learning_rate": 1.7214840130637278e-05,
      "loss": 0.6942,
      "step": 7721
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6107327938079834,
      "learning_rate": 1.72140669350967e-05,
      "loss": 0.6271,
      "step": 7722
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6344586610794067,
      "learning_rate": 1.7213293649615647e-05,
      "loss": 0.7422,
      "step": 7723
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3732619285583496,
      "learning_rate": 1.7212520274203748e-05,
      "loss": 0.6569,
      "step": 7724
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4285029172897339,
      "learning_rate": 1.7211746808870656e-05,
      "loss": 0.7007,
      "step": 7725
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.502634882926941,
      "learning_rate": 1.721097325362601e-05,
      "loss": 0.8059,
      "step": 7726
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7609628438949585,
      "learning_rate": 1.721019960847945e-05,
      "loss": 0.6899,
      "step": 7727
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5059393644332886,
      "learning_rate": 1.7209425873440626e-05,
      "loss": 0.7734,
      "step": 7728
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5212604999542236,
      "learning_rate": 1.7208652048519184e-05,
      "loss": 0.8777,
      "step": 7729
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2596975564956665,
      "learning_rate": 1.7207878133724772e-05,
      "loss": 0.7323,
      "step": 7730
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5402354001998901,
      "learning_rate": 1.720710412906704e-05,
      "loss": 0.7539,
      "step": 7731
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.449207067489624,
      "learning_rate": 1.7206330034555634e-05,
      "loss": 0.7971,
      "step": 7732
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6997066736221313,
      "learning_rate": 1.7205555850200207e-05,
      "loss": 0.696,
      "step": 7733
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4959311485290527,
      "learning_rate": 1.720478157601041e-05,
      "loss": 0.7889,
      "step": 7734
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8625913858413696,
      "learning_rate": 1.7204007211995896e-05,
      "loss": 0.8115,
      "step": 7735
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5146065950393677,
      "learning_rate": 1.7203232758166322e-05,
      "loss": 0.7272,
      "step": 7736
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3591227531433105,
      "learning_rate": 1.720245821453134e-05,
      "loss": 0.8084,
      "step": 7737
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1370171308517456,
      "learning_rate": 1.720168358110061e-05,
      "loss": 0.5434,
      "step": 7738
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6004011631011963,
      "learning_rate": 1.7200908857883787e-05,
      "loss": 0.6854,
      "step": 7739
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.763463020324707,
      "learning_rate": 1.7200134044890528e-05,
      "loss": 0.8995,
      "step": 7740
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.607579231262207,
      "learning_rate": 1.71993591421305e-05,
      "loss": 0.7464,
      "step": 7741
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3813987970352173,
      "learning_rate": 1.7198584149613355e-05,
      "loss": 0.7118,
      "step": 7742
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7803833484649658,
      "learning_rate": 1.7197809067348763e-05,
      "loss": 0.6909,
      "step": 7743
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3872774839401245,
      "learning_rate": 1.7197033895346384e-05,
      "loss": 0.6408,
      "step": 7744
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5754566192626953,
      "learning_rate": 1.719625863361588e-05,
      "loss": 0.7557,
      "step": 7745
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.527114748954773,
      "learning_rate": 1.7195483282166917e-05,
      "loss": 0.6625,
      "step": 7746
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1837871074676514,
      "learning_rate": 1.7194707841009166e-05,
      "loss": 0.7311,
      "step": 7747
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4400476217269897,
      "learning_rate": 1.719393231015229e-05,
      "loss": 0.6328,
      "step": 7748
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5714505910873413,
      "learning_rate": 1.719315668960596e-05,
      "loss": 0.638,
      "step": 7749
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4397324323654175,
      "learning_rate": 1.7192380979379846e-05,
      "loss": 0.5869,
      "step": 7750
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.521461009979248,
      "learning_rate": 1.7191605179483622e-05,
      "loss": 0.5779,
      "step": 7751
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.368792176246643,
      "learning_rate": 1.7190829289926954e-05,
      "loss": 0.6594,
      "step": 7752
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3704049587249756,
      "learning_rate": 1.7190053310719517e-05,
      "loss": 0.6016,
      "step": 7753
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.628937005996704,
      "learning_rate": 1.718927724187099e-05,
      "loss": 0.6981,
      "step": 7754
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6254115104675293,
      "learning_rate": 1.7188501083391043e-05,
      "loss": 0.9426,
      "step": 7755
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.852684497833252,
      "learning_rate": 1.7187724835289354e-05,
      "loss": 0.7063,
      "step": 7756
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4856605529785156,
      "learning_rate": 1.7186948497575604e-05,
      "loss": 0.736,
      "step": 7757
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4505645036697388,
      "learning_rate": 1.7186172070259467e-05,
      "loss": 0.7267,
      "step": 7758
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5237950086593628,
      "learning_rate": 1.718539555335063e-05,
      "loss": 0.6793,
      "step": 7759
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.74473237991333,
      "learning_rate": 1.7184618946858766e-05,
      "loss": 0.6886,
      "step": 7760
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3785299062728882,
      "learning_rate": 1.7183842250793562e-05,
      "loss": 0.741,
      "step": 7761
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9292572736740112,
      "learning_rate": 1.7183065465164697e-05,
      "loss": 0.7282,
      "step": 7762
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4801721572875977,
      "learning_rate": 1.7182288589981863e-05,
      "loss": 0.811,
      "step": 7763
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.826894998550415,
      "learning_rate": 1.7181511625254742e-05,
      "loss": 0.7916,
      "step": 7764
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3456851243972778,
      "learning_rate": 1.7180734570993017e-05,
      "loss": 0.6902,
      "step": 7765
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7882188558578491,
      "learning_rate": 1.717995742720638e-05,
      "loss": 0.7625,
      "step": 7766
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8051928281784058,
      "learning_rate": 1.717918019390452e-05,
      "loss": 0.7999,
      "step": 7767
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4041813611984253,
      "learning_rate": 1.7178402871097123e-05,
      "loss": 0.6921,
      "step": 7768
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.505035161972046,
      "learning_rate": 1.7177625458793886e-05,
      "loss": 0.7105,
      "step": 7769
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5024000406265259,
      "learning_rate": 1.71768479570045e-05,
      "loss": 0.5999,
      "step": 7770
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3295351266860962,
      "learning_rate": 1.717607036573865e-05,
      "loss": 0.6362,
      "step": 7771
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7131154537200928,
      "learning_rate": 1.7175292685006044e-05,
      "loss": 0.6592,
      "step": 7772
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3904949426651,
      "learning_rate": 1.717451491481637e-05,
      "loss": 0.6659,
      "step": 7773
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8474177122116089,
      "learning_rate": 1.7173737055179323e-05,
      "loss": 0.7548,
      "step": 7774
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5486724376678467,
      "learning_rate": 1.7172959106104606e-05,
      "loss": 0.7267,
      "step": 7775
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6330333948135376,
      "learning_rate": 1.7172181067601916e-05,
      "loss": 0.8055,
      "step": 7776
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5112227201461792,
      "learning_rate": 1.717140293968095e-05,
      "loss": 0.7348,
      "step": 7777
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6959675550460815,
      "learning_rate": 1.717062472235141e-05,
      "loss": 0.7397,
      "step": 7778
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.469622015953064,
      "learning_rate": 1.7169846415623003e-05,
      "loss": 0.7751,
      "step": 7779
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3320305347442627,
      "learning_rate": 1.7169068019505428e-05,
      "loss": 0.6154,
      "step": 7780
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5951036214828491,
      "learning_rate": 1.7168289534008394e-05,
      "loss": 0.7737,
      "step": 7781
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7158969640731812,
      "learning_rate": 1.7167510959141602e-05,
      "loss": 0.838,
      "step": 7782
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4017666578292847,
      "learning_rate": 1.716673229491476e-05,
      "loss": 0.8201,
      "step": 7783
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4405499696731567,
      "learning_rate": 1.716595354133758e-05,
      "loss": 0.7318,
      "step": 7784
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8353164196014404,
      "learning_rate": 1.7165174698419765e-05,
      "loss": 0.7501,
      "step": 7785
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7442901134490967,
      "learning_rate": 1.716439576617103e-05,
      "loss": 0.7368,
      "step": 7786
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3388217687606812,
      "learning_rate": 1.716361674460108e-05,
      "loss": 0.8491,
      "step": 7787
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.717682123184204,
      "learning_rate": 1.7162837633719635e-05,
      "loss": 0.7768,
      "step": 7788
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4112697839736938,
      "learning_rate": 1.7162058433536404e-05,
      "loss": 0.6171,
      "step": 7789
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3984967470169067,
      "learning_rate": 1.71612791440611e-05,
      "loss": 0.675,
      "step": 7790
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5172662734985352,
      "learning_rate": 1.7160499765303443e-05,
      "loss": 0.6324,
      "step": 7791
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.525269627571106,
      "learning_rate": 1.7159720297273147e-05,
      "loss": 0.7556,
      "step": 7792
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2166121006011963,
      "learning_rate": 1.715894073997993e-05,
      "loss": 0.6752,
      "step": 7793
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4835125207901,
      "learning_rate": 1.7158161093433512e-05,
      "loss": 0.7545,
      "step": 7794
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.720661997795105,
      "learning_rate": 1.7157381357643617e-05,
      "loss": 0.5787,
      "step": 7795
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.65348482131958,
      "learning_rate": 1.7156601532619956e-05,
      "loss": 0.6374,
      "step": 7796
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4206748008728027,
      "learning_rate": 1.7155821618372263e-05,
      "loss": 0.6841,
      "step": 7797
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7479883432388306,
      "learning_rate": 1.7155041614910253e-05,
      "loss": 0.8556,
      "step": 7798
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4001692533493042,
      "learning_rate": 1.7154261522243652e-05,
      "loss": 0.6694,
      "step": 7799
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4933278560638428,
      "learning_rate": 1.715348134038219e-05,
      "loss": 0.7448,
      "step": 7800
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3289439678192139,
      "learning_rate": 1.715270106933559e-05,
      "loss": 0.6185,
      "step": 7801
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4812495708465576,
      "learning_rate": 1.7151920709113583e-05,
      "loss": 0.6738,
      "step": 7802
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3329415321350098,
      "learning_rate": 1.7151140259725894e-05,
      "loss": 0.7877,
      "step": 7803
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7125164270401,
      "learning_rate": 1.7150359721182258e-05,
      "loss": 0.6788,
      "step": 7804
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.392474889755249,
      "learning_rate": 1.7149579093492402e-05,
      "loss": 0.6326,
      "step": 7805
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4822982549667358,
      "learning_rate": 1.7148798376666057e-05,
      "loss": 0.8215,
      "step": 7806
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5589776039123535,
      "learning_rate": 1.7148017570712964e-05,
      "loss": 0.7375,
      "step": 7807
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2428579330444336,
      "learning_rate": 1.714723667564285e-05,
      "loss": 0.6819,
      "step": 7808
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5532721281051636,
      "learning_rate": 1.7146455691465456e-05,
      "loss": 0.7472,
      "step": 7809
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.436682939529419,
      "learning_rate": 1.7145674618190516e-05,
      "loss": 0.6424,
      "step": 7810
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4626742601394653,
      "learning_rate": 1.714489345582777e-05,
      "loss": 0.7839,
      "step": 7811
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6964068412780762,
      "learning_rate": 1.714411220438695e-05,
      "loss": 0.8487,
      "step": 7812
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6421610116958618,
      "learning_rate": 1.7143330863877804e-05,
      "loss": 0.5891,
      "step": 7813
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3445369005203247,
      "learning_rate": 1.7142549434310074e-05,
      "loss": 0.7629,
      "step": 7814
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4803576469421387,
      "learning_rate": 1.7141767915693498e-05,
      "loss": 0.6254,
      "step": 7815
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6805226802825928,
      "learning_rate": 1.714098630803782e-05,
      "loss": 0.8918,
      "step": 7816
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4717903137207031,
      "learning_rate": 1.7140204611352786e-05,
      "loss": 0.6741,
      "step": 7817
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6714017391204834,
      "learning_rate": 1.713942282564814e-05,
      "loss": 0.7222,
      "step": 7818
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5759645700454712,
      "learning_rate": 1.7138640950933632e-05,
      "loss": 0.7816,
      "step": 7819
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.500722050666809,
      "learning_rate": 1.7137858987219007e-05,
      "loss": 0.6985,
      "step": 7820
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7031611204147339,
      "learning_rate": 1.7137076934514015e-05,
      "loss": 0.833,
      "step": 7821
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4023412466049194,
      "learning_rate": 1.7136294792828403e-05,
      "loss": 0.5861,
      "step": 7822
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4528639316558838,
      "learning_rate": 1.713551256217193e-05,
      "loss": 0.6646,
      "step": 7823
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4207528829574585,
      "learning_rate": 1.713473024255434e-05,
      "loss": 0.584,
      "step": 7824
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6794085502624512,
      "learning_rate": 1.7133947833985395e-05,
      "loss": 0.6761,
      "step": 7825
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3821759223937988,
      "learning_rate": 1.713316533647484e-05,
      "loss": 0.7758,
      "step": 7826
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1891087293624878,
      "learning_rate": 1.7132382750032438e-05,
      "loss": 0.6846,
      "step": 7827
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5260958671569824,
      "learning_rate": 1.7131600074667944e-05,
      "loss": 0.6898,
      "step": 7828
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4992095232009888,
      "learning_rate": 1.713081731039111e-05,
      "loss": 0.7607,
      "step": 7829
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4892585277557373,
      "learning_rate": 1.7130034457211705e-05,
      "loss": 0.6795,
      "step": 7830
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.629880666732788,
      "learning_rate": 1.712925151513948e-05,
      "loss": 0.8237,
      "step": 7831
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4907199144363403,
      "learning_rate": 1.7128468484184205e-05,
      "loss": 0.696,
      "step": 7832
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.641181468963623,
      "learning_rate": 1.7127685364355638e-05,
      "loss": 0.8243,
      "step": 7833
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5104140043258667,
      "learning_rate": 1.712690215566354e-05,
      "loss": 0.6707,
      "step": 7834
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3698749542236328,
      "learning_rate": 1.7126118858117676e-05,
      "loss": 0.5385,
      "step": 7835
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5807669162750244,
      "learning_rate": 1.7125335471727816e-05,
      "loss": 0.6554,
      "step": 7836
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6698659658432007,
      "learning_rate": 1.7124551996503723e-05,
      "loss": 0.7077,
      "step": 7837
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6094739437103271,
      "learning_rate": 1.712376843245517e-05,
      "loss": 0.6851,
      "step": 7838
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.517339825630188,
      "learning_rate": 1.7122984779591914e-05,
      "loss": 0.7856,
      "step": 7839
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4128179550170898,
      "learning_rate": 1.712220103792374e-05,
      "loss": 0.7578,
      "step": 7840
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3748533725738525,
      "learning_rate": 1.712141720746041e-05,
      "loss": 0.766,
      "step": 7841
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4260655641555786,
      "learning_rate": 1.7120633288211702e-05,
      "loss": 0.6771,
      "step": 7842
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7716645002365112,
      "learning_rate": 1.7119849280187385e-05,
      "loss": 0.9861,
      "step": 7843
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5165925025939941,
      "learning_rate": 1.7119065183397234e-05,
      "loss": 0.7143,
      "step": 7844
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.065183401107788,
      "learning_rate": 1.7118280997851023e-05,
      "loss": 0.8847,
      "step": 7845
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6653543710708618,
      "learning_rate": 1.7117496723558534e-05,
      "loss": 0.5845,
      "step": 7846
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5201480388641357,
      "learning_rate": 1.7116712360529543e-05,
      "loss": 0.7562,
      "step": 7847
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3662887811660767,
      "learning_rate": 1.7115927908773823e-05,
      "loss": 0.7214,
      "step": 7848
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4069318771362305,
      "learning_rate": 1.7115143368301163e-05,
      "loss": 0.7759,
      "step": 7849
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.479990005493164,
      "learning_rate": 1.7114358739121338e-05,
      "loss": 0.8481,
      "step": 7850
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6621572971343994,
      "learning_rate": 1.7113574021244132e-05,
      "loss": 0.6892,
      "step": 7851
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6782536506652832,
      "learning_rate": 1.7112789214679328e-05,
      "loss": 0.7663,
      "step": 7852
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4684257507324219,
      "learning_rate": 1.7112004319436713e-05,
      "loss": 0.8401,
      "step": 7853
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8232628107070923,
      "learning_rate": 1.7111219335526073e-05,
      "loss": 0.7287,
      "step": 7854
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5185917615890503,
      "learning_rate": 1.711043426295719e-05,
      "loss": 0.6314,
      "step": 7855
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.866430640220642,
      "learning_rate": 1.7109649101739855e-05,
      "loss": 0.7573,
      "step": 7856
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3718613386154175,
      "learning_rate": 1.710886385188385e-05,
      "loss": 0.6709,
      "step": 7857
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.484752893447876,
      "learning_rate": 1.710807851339898e-05,
      "loss": 0.8258,
      "step": 7858
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.581272840499878,
      "learning_rate": 1.710729308629502e-05,
      "loss": 0.7289,
      "step": 7859
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4112906455993652,
      "learning_rate": 1.710650757058178e-05,
      "loss": 0.6362,
      "step": 7860
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3721253871917725,
      "learning_rate": 1.7105721966269032e-05,
      "loss": 0.7125,
      "step": 7861
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6120491027832031,
      "learning_rate": 1.7104936273366587e-05,
      "loss": 0.6396,
      "step": 7862
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3872618675231934,
      "learning_rate": 1.7104150491884234e-05,
      "loss": 0.7,
      "step": 7863
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3133004903793335,
      "learning_rate": 1.7103364621831767e-05,
      "loss": 0.5538,
      "step": 7864
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3701893091201782,
      "learning_rate": 1.7102578663218992e-05,
      "loss": 0.6844,
      "step": 7865
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4926488399505615,
      "learning_rate": 1.71017926160557e-05,
      "loss": 0.7841,
      "step": 7866
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6114453077316284,
      "learning_rate": 1.7101006480351694e-05,
      "loss": 0.6513,
      "step": 7867
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5591270923614502,
      "learning_rate": 1.7100220256116774e-05,
      "loss": 0.6804,
      "step": 7868
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3523640632629395,
      "learning_rate": 1.7099433943360744e-05,
      "loss": 0.6898,
      "step": 7869
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5466549396514893,
      "learning_rate": 1.7098647542093407e-05,
      "loss": 0.6262,
      "step": 7870
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6213164329528809,
      "learning_rate": 1.7097861052324567e-05,
      "loss": 0.7395,
      "step": 7871
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3990832567214966,
      "learning_rate": 1.7097074474064027e-05,
      "loss": 0.6187,
      "step": 7872
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3333425521850586,
      "learning_rate": 1.70962878073216e-05,
      "loss": 0.648,
      "step": 7873
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3127690553665161,
      "learning_rate": 1.7095501052107084e-05,
      "loss": 0.5718,
      "step": 7874
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.433628797531128,
      "learning_rate": 1.7094714208430298e-05,
      "loss": 0.8034,
      "step": 7875
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.554095983505249,
      "learning_rate": 1.7093927276301046e-05,
      "loss": 0.5737,
      "step": 7876
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4209232330322266,
      "learning_rate": 1.7093140255729137e-05,
      "loss": 0.7462,
      "step": 7877
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8898274898529053,
      "learning_rate": 1.7092353146724386e-05,
      "loss": 0.6452,
      "step": 7878
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5147781372070312,
      "learning_rate": 1.709156594929661e-05,
      "loss": 0.7052,
      "step": 7879
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8122659921646118,
      "learning_rate": 1.7090778663455617e-05,
      "loss": 0.7017,
      "step": 7880
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5884274244308472,
      "learning_rate": 1.7089991289211227e-05,
      "loss": 0.8074,
      "step": 7881
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5681651830673218,
      "learning_rate": 1.7089203826573253e-05,
      "loss": 0.6195,
      "step": 7882
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.454200267791748,
      "learning_rate": 1.7088416275551515e-05,
      "loss": 0.691,
      "step": 7883
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8885828256607056,
      "learning_rate": 1.708762863615583e-05,
      "loss": 0.6544,
      "step": 7884
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7452735900878906,
      "learning_rate": 1.7086840908396022e-05,
      "loss": 0.6319,
      "step": 7885
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7429273128509521,
      "learning_rate": 1.7086053092281902e-05,
      "loss": 0.6942,
      "step": 7886
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5926508903503418,
      "learning_rate": 1.70852651878233e-05,
      "loss": 0.743,
      "step": 7887
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2530715465545654,
      "learning_rate": 1.708447719503004e-05,
      "loss": 0.7245,
      "step": 7888
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6523343324661255,
      "learning_rate": 1.7083689113911942e-05,
      "loss": 0.5693,
      "step": 7889
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6461552381515503,
      "learning_rate": 1.7082900944478838e-05,
      "loss": 0.8322,
      "step": 7890
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6297954320907593,
      "learning_rate": 1.708211268674054e-05,
      "loss": 0.7661,
      "step": 7891
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.522806167602539,
      "learning_rate": 1.708132434070689e-05,
      "loss": 0.7169,
      "step": 7892
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.51332426071167,
      "learning_rate": 1.7080535906387716e-05,
      "loss": 0.6436,
      "step": 7893
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6488420963287354,
      "learning_rate": 1.707974738379284e-05,
      "loss": 0.6098,
      "step": 7894
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4125045537948608,
      "learning_rate": 1.7078958772932095e-05,
      "loss": 0.6887,
      "step": 7895
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5964698791503906,
      "learning_rate": 1.7078170073815313e-05,
      "loss": 0.6832,
      "step": 7896
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3886858224868774,
      "learning_rate": 1.707738128645233e-05,
      "loss": 0.7299,
      "step": 7897
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5236790180206299,
      "learning_rate": 1.7076592410852975e-05,
      "loss": 0.5886,
      "step": 7898
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6310330629348755,
      "learning_rate": 1.7075803447027094e-05,
      "loss": 0.6857,
      "step": 7899
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5556411743164062,
      "learning_rate": 1.7075014394984508e-05,
      "loss": 0.7346,
      "step": 7900
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4761559963226318,
      "learning_rate": 1.7074225254735064e-05,
      "loss": 0.6052,
      "step": 7901
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.664289116859436,
      "learning_rate": 1.70734360262886e-05,
      "loss": 0.7057,
      "step": 7902
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4178534746170044,
      "learning_rate": 1.7072646709654953e-05,
      "loss": 0.734,
      "step": 7903
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4394923448562622,
      "learning_rate": 1.7071857304843963e-05,
      "loss": 0.692,
      "step": 7904
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.520620584487915,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.7297,
      "step": 7905
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4425287246704102,
      "learning_rate": 1.7070278230729333e-05,
      "loss": 0.6831,
      "step": 7906
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5106338262557983,
      "learning_rate": 1.7069488561445378e-05,
      "loss": 0.6924,
      "step": 7907
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.391392469406128,
      "learning_rate": 1.7068698804023453e-05,
      "loss": 0.9023,
      "step": 7908
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4802889823913574,
      "learning_rate": 1.7067908958473408e-05,
      "loss": 0.747,
      "step": 7909
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4300322532653809,
      "learning_rate": 1.706711902480509e-05,
      "loss": 0.608,
      "step": 7910
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3409186601638794,
      "learning_rate": 1.7066329003028345e-05,
      "loss": 0.598,
      "step": 7911
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3853318691253662,
      "learning_rate": 1.7065538893153026e-05,
      "loss": 0.5675,
      "step": 7912
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.646220326423645,
      "learning_rate": 1.706474869518898e-05,
      "loss": 0.7368,
      "step": 7913
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5835763216018677,
      "learning_rate": 1.706395840914606e-05,
      "loss": 0.6709,
      "step": 7914
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7430323362350464,
      "learning_rate": 1.706316803503412e-05,
      "loss": 0.6184,
      "step": 7915
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4553760290145874,
      "learning_rate": 1.7062377572863017e-05,
      "loss": 0.6572,
      "step": 7916
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5282578468322754,
      "learning_rate": 1.7061587022642597e-05,
      "loss": 0.7021,
      "step": 7917
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4369721412658691,
      "learning_rate": 1.706079638438272e-05,
      "loss": 0.8131,
      "step": 7918
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3438633680343628,
      "learning_rate": 1.706000565809325e-05,
      "loss": 0.7087,
      "step": 7919
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3865094184875488,
      "learning_rate": 1.7059214843784036e-05,
      "loss": 0.6513,
      "step": 7920
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.733942985534668,
      "learning_rate": 1.705842394146494e-05,
      "loss": 0.771,
      "step": 7921
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5516669750213623,
      "learning_rate": 1.7057632951145824e-05,
      "loss": 0.6164,
      "step": 7922
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4739995002746582,
      "learning_rate": 1.705684187283655e-05,
      "loss": 0.7112,
      "step": 7923
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3723176717758179,
      "learning_rate": 1.705605070654698e-05,
      "loss": 0.6279,
      "step": 7924
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3332412242889404,
      "learning_rate": 1.7055259452286975e-05,
      "loss": 0.7108,
      "step": 7925
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4193507432937622,
      "learning_rate": 1.7054468110066405e-05,
      "loss": 0.7017,
      "step": 7926
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.333153009414673,
      "learning_rate": 1.7053676679895136e-05,
      "loss": 0.6149,
      "step": 7927
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4842212200164795,
      "learning_rate": 1.7052885161783025e-05,
      "loss": 0.826,
      "step": 7928
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.316375970840454,
      "learning_rate": 1.7052093555739953e-05,
      "loss": 0.6762,
      "step": 7929
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7945958375930786,
      "learning_rate": 1.705130186177578e-05,
      "loss": 0.7886,
      "step": 7930
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7807692289352417,
      "learning_rate": 1.7050510079900384e-05,
      "loss": 0.6384,
      "step": 7931
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3446632623672485,
      "learning_rate": 1.7049718210123632e-05,
      "loss": 0.7581,
      "step": 7932
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3475220203399658,
      "learning_rate": 1.7048926252455395e-05,
      "loss": 0.5259,
      "step": 7933
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.36685311794281,
      "learning_rate": 1.7048134206905548e-05,
      "loss": 0.5914,
      "step": 7934
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6897408962249756,
      "learning_rate": 1.7047342073483967e-05,
      "loss": 0.7942,
      "step": 7935
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5547817945480347,
      "learning_rate": 1.704654985220053e-05,
      "loss": 0.7609,
      "step": 7936
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4360138177871704,
      "learning_rate": 1.704575754306511e-05,
      "loss": 0.7835,
      "step": 7937
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2764595746994019,
      "learning_rate": 1.704496514608758e-05,
      "loss": 0.5912,
      "step": 7938
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4885073900222778,
      "learning_rate": 1.7044172661277834e-05,
      "loss": 0.6341,
      "step": 7939
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.357409954071045,
      "learning_rate": 1.704338008864574e-05,
      "loss": 0.654,
      "step": 7940
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4717626571655273,
      "learning_rate": 1.7042587428201182e-05,
      "loss": 0.6714,
      "step": 7941
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8982751369476318,
      "learning_rate": 1.7041794679954045e-05,
      "loss": 0.6945,
      "step": 7942
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4637718200683594,
      "learning_rate": 1.704100184391421e-05,
      "loss": 0.7422,
      "step": 7943
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5501312017440796,
      "learning_rate": 1.7040208920091564e-05,
      "loss": 0.6961,
      "step": 7944
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3929129838943481,
      "learning_rate": 1.7039415908495988e-05,
      "loss": 0.8378,
      "step": 7945
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5093753337860107,
      "learning_rate": 1.7038622809137374e-05,
      "loss": 0.6738,
      "step": 7946
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5496817827224731,
      "learning_rate": 1.7037829622025605e-05,
      "loss": 0.6751,
      "step": 7947
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4079781770706177,
      "learning_rate": 1.7037036347170573e-05,
      "loss": 0.8261,
      "step": 7948
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6648082733154297,
      "learning_rate": 1.703624298458217e-05,
      "loss": 0.6039,
      "step": 7949
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.533653974533081,
      "learning_rate": 1.7035449534270284e-05,
      "loss": 0.5904,
      "step": 7950
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6259304285049438,
      "learning_rate": 1.7034655996244806e-05,
      "loss": 0.613,
      "step": 7951
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.585153579711914,
      "learning_rate": 1.703386237051563e-05,
      "loss": 0.6824,
      "step": 7952
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6533749103546143,
      "learning_rate": 1.7033068657092657e-05,
      "loss": 0.6853,
      "step": 7953
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4996368885040283,
      "learning_rate": 1.7032274855985776e-05,
      "loss": 0.6282,
      "step": 7954
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.495766043663025,
      "learning_rate": 1.7031480967204882e-05,
      "loss": 0.8506,
      "step": 7955
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7331717014312744,
      "learning_rate": 1.7030686990759877e-05,
      "loss": 0.6151,
      "step": 7956
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.434190034866333,
      "learning_rate": 1.7029892926660657e-05,
      "loss": 0.6766,
      "step": 7957
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6043927669525146,
      "learning_rate": 1.702909877491713e-05,
      "loss": 0.7037,
      "step": 7958
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6678754091262817,
      "learning_rate": 1.702830453553918e-05,
      "loss": 0.664,
      "step": 7959
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.443748116493225,
      "learning_rate": 1.7027510208536726e-05,
      "loss": 0.6141,
      "step": 7960
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5304299592971802,
      "learning_rate": 1.702671579391966e-05,
      "loss": 0.7088,
      "step": 7961
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.612577199935913,
      "learning_rate": 1.70259212916979e-05,
      "loss": 0.623,
      "step": 7962
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.605276346206665,
      "learning_rate": 1.7025126701881334e-05,
      "loss": 0.5911,
      "step": 7963
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7649555206298828,
      "learning_rate": 1.7024332024479875e-05,
      "loss": 0.7302,
      "step": 7964
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.044251799583435,
      "learning_rate": 1.702353725950344e-05,
      "loss": 0.6058,
      "step": 7965
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8090070486068726,
      "learning_rate": 1.7022742406961927e-05,
      "loss": 0.7804,
      "step": 7966
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4672492742538452,
      "learning_rate": 1.7021947466865244e-05,
      "loss": 0.7162,
      "step": 7967
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3331620693206787,
      "learning_rate": 1.7021152439223308e-05,
      "loss": 0.6781,
      "step": 7968
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5877158641815186,
      "learning_rate": 1.7020357324046033e-05,
      "loss": 0.8308,
      "step": 7969
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3156917095184326,
      "learning_rate": 1.7019562121343326e-05,
      "loss": 0.5384,
      "step": 7970
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3003870248794556,
      "learning_rate": 1.7018766831125103e-05,
      "loss": 0.6995,
      "step": 7971
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.664218783378601,
      "learning_rate": 1.701797145340128e-05,
      "loss": 0.7682,
      "step": 7972
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4372456073760986,
      "learning_rate": 1.7017175988181774e-05,
      "loss": 0.7844,
      "step": 7973
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6517399549484253,
      "learning_rate": 1.7016380435476498e-05,
      "loss": 0.6806,
      "step": 7974
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7727315425872803,
      "learning_rate": 1.7015584795295375e-05,
      "loss": 0.7978,
      "step": 7975
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2364907264709473,
      "learning_rate": 1.7014789067648325e-05,
      "loss": 0.7663,
      "step": 7976
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4109190702438354,
      "learning_rate": 1.7013993252545263e-05,
      "loss": 0.7653,
      "step": 7977
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4409161806106567,
      "learning_rate": 1.7013197349996118e-05,
      "loss": 0.8263,
      "step": 7978
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2952800989151,
      "learning_rate": 1.7012401360010808e-05,
      "loss": 0.7193,
      "step": 7979
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.390620470046997,
      "learning_rate": 1.701160528259926e-05,
      "loss": 0.708,
      "step": 7980
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4933476448059082,
      "learning_rate": 1.7010809117771394e-05,
      "loss": 0.5754,
      "step": 7981
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6100326776504517,
      "learning_rate": 1.7010012865537142e-05,
      "loss": 0.6103,
      "step": 7982
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4293140172958374,
      "learning_rate": 1.7009216525906426e-05,
      "loss": 0.8214,
      "step": 7983
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5348109006881714,
      "learning_rate": 1.700842009888918e-05,
      "loss": 0.7425,
      "step": 7984
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.400858759880066,
      "learning_rate": 1.700762358449533e-05,
      "loss": 0.7345,
      "step": 7985
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4191341400146484,
      "learning_rate": 1.7006826982734804e-05,
      "loss": 0.744,
      "step": 7986
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5049582719802856,
      "learning_rate": 1.700603029361754e-05,
      "loss": 0.5806,
      "step": 7987
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6118285655975342,
      "learning_rate": 1.700523351715346e-05,
      "loss": 0.8035,
      "step": 7988
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4250133037567139,
      "learning_rate": 1.700443665335251e-05,
      "loss": 0.6636,
      "step": 7989
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2220600843429565,
      "learning_rate": 1.7003639702224616e-05,
      "loss": 0.7641,
      "step": 7990
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.314227819442749,
      "learning_rate": 1.7002842663779718e-05,
      "loss": 0.7496,
      "step": 7991
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.353569507598877,
      "learning_rate": 1.7002045538027753e-05,
      "loss": 0.5214,
      "step": 7992
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5223554372787476,
      "learning_rate": 1.7001248324978658e-05,
      "loss": 0.6948,
      "step": 7993
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4898756742477417,
      "learning_rate": 1.7000451024642374e-05,
      "loss": 0.688,
      "step": 7994
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5319161415100098,
      "learning_rate": 1.699965363702884e-05,
      "loss": 0.7988,
      "step": 7995
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6492482423782349,
      "learning_rate": 1.699885616214799e-05,
      "loss": 0.5949,
      "step": 7996
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5764020681381226,
      "learning_rate": 1.6998058600009775e-05,
      "loss": 0.7911,
      "step": 7997
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6895992755889893,
      "learning_rate": 1.6997260950624143e-05,
      "loss": 0.7504,
      "step": 7998
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8729133605957031,
      "learning_rate": 1.699646321400103e-05,
      "loss": 0.6568,
      "step": 7999
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4197169542312622,
      "learning_rate": 1.6995665390150384e-05,
      "loss": 0.6302,
      "step": 8000
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.868575930595398,
      "learning_rate": 1.699486747908215e-05,
      "loss": 0.7866,
      "step": 8001
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7372276782989502,
      "learning_rate": 1.6994069480806278e-05,
      "loss": 0.7548,
      "step": 8002
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6868313550949097,
      "learning_rate": 1.6993271395332715e-05,
      "loss": 0.7115,
      "step": 8003
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4107578992843628,
      "learning_rate": 1.6992473222671416e-05,
      "loss": 0.5493,
      "step": 8004
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7350424528121948,
      "learning_rate": 1.699167496283233e-05,
      "loss": 0.645,
      "step": 8005
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4482606649398804,
      "learning_rate": 1.6990876615825405e-05,
      "loss": 0.679,
      "step": 8006
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8592779636383057,
      "learning_rate": 1.6990078181660596e-05,
      "loss": 0.7156,
      "step": 8007
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.577735424041748,
      "learning_rate": 1.698927966034786e-05,
      "loss": 0.6413,
      "step": 8008
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.237537145614624,
      "learning_rate": 1.6988481051897153e-05,
      "loss": 0.6904,
      "step": 8009
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3875341415405273,
      "learning_rate": 1.698768235631843e-05,
      "loss": 0.6841,
      "step": 8010
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8567416667938232,
      "learning_rate": 1.6986883573621647e-05,
      "loss": 0.8503,
      "step": 8011
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.340397596359253,
      "learning_rate": 1.698608470381676e-05,
      "loss": 0.7381,
      "step": 8012
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4597293138504028,
      "learning_rate": 1.698528574691374e-05,
      "loss": 0.6706,
      "step": 8013
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8329027891159058,
      "learning_rate": 1.6984486702922537e-05,
      "loss": 0.7576,
      "step": 8014
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6120004653930664,
      "learning_rate": 1.6983687571853118e-05,
      "loss": 0.6434,
      "step": 8015
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.595361590385437,
      "learning_rate": 1.6982888353715443e-05,
      "loss": 0.7089,
      "step": 8016
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.281233310699463,
      "learning_rate": 1.6982089048519478e-05,
      "loss": 0.7555,
      "step": 8017
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6985923051834106,
      "learning_rate": 1.698128965627519e-05,
      "loss": 0.6991,
      "step": 8018
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7083715200424194,
      "learning_rate": 1.6980490176992542e-05,
      "loss": 0.7111,
      "step": 8019
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6052454710006714,
      "learning_rate": 1.6979690610681507e-05,
      "loss": 0.6259,
      "step": 8020
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9324671030044556,
      "learning_rate": 1.6978890957352047e-05,
      "loss": 0.6549,
      "step": 8021
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6928470134735107,
      "learning_rate": 1.6978091217014132e-05,
      "loss": 0.9608,
      "step": 8022
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4586968421936035,
      "learning_rate": 1.6977291389677734e-05,
      "loss": 0.8084,
      "step": 8023
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5559052228927612,
      "learning_rate": 1.697649147535283e-05,
      "loss": 0.761,
      "step": 8024
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6452206373214722,
      "learning_rate": 1.697569147404938e-05,
      "loss": 0.607,
      "step": 8025
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6172503232955933,
      "learning_rate": 1.6974891385777376e-05,
      "loss": 0.6948,
      "step": 8026
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.755679965019226,
      "learning_rate": 1.697409121054678e-05,
      "loss": 0.6319,
      "step": 8027
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4533191919326782,
      "learning_rate": 1.697329094836757e-05,
      "loss": 0.7641,
      "step": 8028
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2273329496383667,
      "learning_rate": 1.6972490599249725e-05,
      "loss": 0.6496,
      "step": 8029
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6022025346755981,
      "learning_rate": 1.697169016320322e-05,
      "loss": 0.5416,
      "step": 8030
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.481284260749817,
      "learning_rate": 1.697088964023804e-05,
      "loss": 0.9239,
      "step": 8031
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.608094334602356,
      "learning_rate": 1.697008903036416e-05,
      "loss": 0.7499,
      "step": 8032
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0600768327713013,
      "learning_rate": 1.6969288333591567e-05,
      "loss": 0.4942,
      "step": 8033
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.434135913848877,
      "learning_rate": 1.696848754993024e-05,
      "loss": 0.6431,
      "step": 8034
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5845414400100708,
      "learning_rate": 1.6967686679390164e-05,
      "loss": 0.836,
      "step": 8035
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3524504899978638,
      "learning_rate": 1.696688572198132e-05,
      "loss": 0.5532,
      "step": 8036
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2385140657424927,
      "learning_rate": 1.6966084677713697e-05,
      "loss": 0.7617,
      "step": 8037
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6877644062042236,
      "learning_rate": 1.696528354659728e-05,
      "loss": 0.8052,
      "step": 8038
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6073607206344604,
      "learning_rate": 1.6964482328642066e-05,
      "loss": 0.765,
      "step": 8039
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3154555559158325,
      "learning_rate": 1.696368102385803e-05,
      "loss": 0.6266,
      "step": 8040
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.493801236152649,
      "learning_rate": 1.6962879632255173e-05,
      "loss": 0.7725,
      "step": 8041
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.544339895248413,
      "learning_rate": 1.696207815384348e-05,
      "loss": 0.6195,
      "step": 8042
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6650822162628174,
      "learning_rate": 1.6961276588632944e-05,
      "loss": 0.6963,
      "step": 8043
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4734092950820923,
      "learning_rate": 1.6960474936633563e-05,
      "loss": 0.6307,
      "step": 8044
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5644173622131348,
      "learning_rate": 1.6959673197855325e-05,
      "loss": 0.6683,
      "step": 8045
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.48423433303833,
      "learning_rate": 1.695887137230823e-05,
      "loss": 0.7494,
      "step": 8046
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2983125448226929,
      "learning_rate": 1.6958069460002277e-05,
      "loss": 0.6119,
      "step": 8047
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4820277690887451,
      "learning_rate": 1.6957267460947457e-05,
      "loss": 0.6574,
      "step": 8048
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7483103275299072,
      "learning_rate": 1.6956465375153776e-05,
      "loss": 0.6135,
      "step": 8049
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5367552042007446,
      "learning_rate": 1.6955663202631226e-05,
      "loss": 0.727,
      "step": 8050
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.572947382926941,
      "learning_rate": 1.6954860943389816e-05,
      "loss": 0.6712,
      "step": 8051
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.533460021018982,
      "learning_rate": 1.695405859743954e-05,
      "loss": 0.7633,
      "step": 8052
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5322233438491821,
      "learning_rate": 1.6953256164790408e-05,
      "loss": 0.681,
      "step": 8053
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.419631004333496,
      "learning_rate": 1.695245364545242e-05,
      "loss": 0.7018,
      "step": 8054
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7192933559417725,
      "learning_rate": 1.6951651039435586e-05,
      "loss": 0.8233,
      "step": 8055
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5535932779312134,
      "learning_rate": 1.6950848346749906e-05,
      "loss": 0.6337,
      "step": 8056
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6076644659042358,
      "learning_rate": 1.6950045567405394e-05,
      "loss": 0.7983,
      "step": 8057
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3464072942733765,
      "learning_rate": 1.694924270141205e-05,
      "loss": 0.6211,
      "step": 8058
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4081288576126099,
      "learning_rate": 1.6948439748779895e-05,
      "loss": 0.7191,
      "step": 8059
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2991341352462769,
      "learning_rate": 1.6947636709518932e-05,
      "loss": 0.6773,
      "step": 8060
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.600683331489563,
      "learning_rate": 1.6946833583639174e-05,
      "loss": 0.792,
      "step": 8061
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4030561447143555,
      "learning_rate": 1.694603037115063e-05,
      "loss": 0.8351,
      "step": 8062
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.700113296508789,
      "learning_rate": 1.6945227072063322e-05,
      "loss": 0.7759,
      "step": 8063
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7096763849258423,
      "learning_rate": 1.6944423686387263e-05,
      "loss": 0.7341,
      "step": 8064
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.358407735824585,
      "learning_rate": 1.6943620214132466e-05,
      "loss": 0.6061,
      "step": 8065
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9206262826919556,
      "learning_rate": 1.6942816655308947e-05,
      "loss": 0.8336,
      "step": 8066
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2654889822006226,
      "learning_rate": 1.694201300992673e-05,
      "loss": 0.7042,
      "step": 8067
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5359687805175781,
      "learning_rate": 1.6941209277995828e-05,
      "loss": 0.7413,
      "step": 8068
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5909725427627563,
      "learning_rate": 1.6940405459526272e-05,
      "loss": 0.5714,
      "step": 8069
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3845298290252686,
      "learning_rate": 1.6939601554528068e-05,
      "loss": 0.6874,
      "step": 8070
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2916909456253052,
      "learning_rate": 1.693879756301125e-05,
      "loss": 0.5946,
      "step": 8071
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3557592630386353,
      "learning_rate": 1.6937993484985838e-05,
      "loss": 0.5654,
      "step": 8072
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.557603359222412,
      "learning_rate": 1.693718932046186e-05,
      "loss": 0.5633,
      "step": 8073
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.507951021194458,
      "learning_rate": 1.6936385069449334e-05,
      "loss": 0.7703,
      "step": 8074
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4766201972961426,
      "learning_rate": 1.6935580731958293e-05,
      "loss": 0.7151,
      "step": 8075
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5897718667984009,
      "learning_rate": 1.6934776307998764e-05,
      "loss": 0.6363,
      "step": 8076
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7662897109985352,
      "learning_rate": 1.693397179758078e-05,
      "loss": 0.6943,
      "step": 8077
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.472821593284607,
      "learning_rate": 1.6933167200714366e-05,
      "loss": 0.6763,
      "step": 8078
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4662941694259644,
      "learning_rate": 1.693236251740955e-05,
      "loss": 0.7037,
      "step": 8079
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4426252841949463,
      "learning_rate": 1.693155774767637e-05,
      "loss": 0.7174,
      "step": 8080
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6484613418579102,
      "learning_rate": 1.693075289152486e-05,
      "loss": 0.7752,
      "step": 8081
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.487396240234375,
      "learning_rate": 1.6929947948965052e-05,
      "loss": 0.6913,
      "step": 8082
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3067697286605835,
      "learning_rate": 1.692914292000698e-05,
      "loss": 0.753,
      "step": 8083
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3672226667404175,
      "learning_rate": 1.6928337804660683e-05,
      "loss": 0.8024,
      "step": 8084
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4477226734161377,
      "learning_rate": 1.69275326029362e-05,
      "loss": 0.6864,
      "step": 8085
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6531256437301636,
      "learning_rate": 1.692672731484356e-05,
      "loss": 0.6669,
      "step": 8086
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6594910621643066,
      "learning_rate": 1.692592194039282e-05,
      "loss": 0.7762,
      "step": 8087
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6076809167861938,
      "learning_rate": 1.692511647959401e-05,
      "loss": 0.6923,
      "step": 8088
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7547812461853027,
      "learning_rate": 1.692431093245717e-05,
      "loss": 0.8256,
      "step": 8089
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5691088438034058,
      "learning_rate": 1.692350529899235e-05,
      "loss": 0.7287,
      "step": 8090
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4456944465637207,
      "learning_rate": 1.6922699579209593e-05,
      "loss": 0.6704,
      "step": 8091
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5964089632034302,
      "learning_rate": 1.6921893773118937e-05,
      "loss": 0.7615,
      "step": 8092
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4334697723388672,
      "learning_rate": 1.6921087880730433e-05,
      "loss": 0.684,
      "step": 8093
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4643967151641846,
      "learning_rate": 1.6920281902054133e-05,
      "loss": 0.6486,
      "step": 8094
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6898401975631714,
      "learning_rate": 1.6919475837100078e-05,
      "loss": 0.6584,
      "step": 8095
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.666135311126709,
      "learning_rate": 1.6918669685878322e-05,
      "loss": 0.5897,
      "step": 8096
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.726584553718567,
      "learning_rate": 1.691786344839891e-05,
      "loss": 0.6305,
      "step": 8097
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.575005292892456,
      "learning_rate": 1.6917057124671906e-05,
      "loss": 0.6744,
      "step": 8098
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4085886478424072,
      "learning_rate": 1.691625071470735e-05,
      "loss": 0.6576,
      "step": 8099
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7064261436462402,
      "learning_rate": 1.6915444218515298e-05,
      "loss": 0.7725,
      "step": 8100
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6504089832305908,
      "learning_rate": 1.691463763610581e-05,
      "loss": 0.7102,
      "step": 8101
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5451223850250244,
      "learning_rate": 1.691383096748894e-05,
      "loss": 0.7404,
      "step": 8102
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5184825658798218,
      "learning_rate": 1.691302421267474e-05,
      "loss": 0.7947,
      "step": 8103
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7828950881958008,
      "learning_rate": 1.6912217371673277e-05,
      "loss": 0.7107,
      "step": 8104
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.296836256980896,
      "learning_rate": 1.6911410444494606e-05,
      "loss": 0.8337,
      "step": 8105
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4456679821014404,
      "learning_rate": 1.6910603431148784e-05,
      "loss": 0.6775,
      "step": 8106
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.677411437034607,
      "learning_rate": 1.6909796331645875e-05,
      "loss": 0.7313,
      "step": 8107
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.526764154434204,
      "learning_rate": 1.6908989145995947e-05,
      "loss": 0.6677,
      "step": 8108
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2253139019012451,
      "learning_rate": 1.6908181874209054e-05,
      "loss": 0.6137,
      "step": 8109
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3627851009368896,
      "learning_rate": 1.6907374516295264e-05,
      "loss": 0.7264,
      "step": 8110
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3513333797454834,
      "learning_rate": 1.6906567072264644e-05,
      "loss": 0.6621,
      "step": 8111
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7844711542129517,
      "learning_rate": 1.690575954212726e-05,
      "loss": 0.7048,
      "step": 8112
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.30682373046875,
      "learning_rate": 1.690495192589318e-05,
      "loss": 0.7381,
      "step": 8113
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7773230075836182,
      "learning_rate": 1.6904144223572473e-05,
      "loss": 0.6498,
      "step": 8114
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.398927927017212,
      "learning_rate": 1.6903336435175208e-05,
      "loss": 0.8092,
      "step": 8115
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7369314432144165,
      "learning_rate": 1.6902528560711457e-05,
      "loss": 0.6533,
      "step": 8116
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3342807292938232,
      "learning_rate": 1.6901720600191293e-05,
      "loss": 0.787,
      "step": 8117
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.655876636505127,
      "learning_rate": 1.6900912553624785e-05,
      "loss": 0.8151,
      "step": 8118
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8827388286590576,
      "learning_rate": 1.690010442102201e-05,
      "loss": 0.6537,
      "step": 8119
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.242018699645996,
      "learning_rate": 1.6899296202393046e-05,
      "loss": 0.7042,
      "step": 8120
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.72730553150177,
      "learning_rate": 1.6898487897747966e-05,
      "loss": 0.7367,
      "step": 8121
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7055413722991943,
      "learning_rate": 1.689767950709685e-05,
      "loss": 0.6661,
      "step": 8122
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.725009799003601,
      "learning_rate": 1.689687103044977e-05,
      "loss": 0.6921,
      "step": 8123
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6222363710403442,
      "learning_rate": 1.689606246781681e-05,
      "loss": 0.7361,
      "step": 8124
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2712188959121704,
      "learning_rate": 1.6895253819208056e-05,
      "loss": 0.6259,
      "step": 8125
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3379179239273071,
      "learning_rate": 1.6894445084633582e-05,
      "loss": 0.6278,
      "step": 8126
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3529722690582275,
      "learning_rate": 1.689363626410347e-05,
      "loss": 0.6276,
      "step": 8127
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5932458639144897,
      "learning_rate": 1.6892827357627813e-05,
      "loss": 0.7014,
      "step": 8128
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8396050930023193,
      "learning_rate": 1.6892018365216686e-05,
      "loss": 0.724,
      "step": 8129
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3766013383865356,
      "learning_rate": 1.6891209286880183e-05,
      "loss": 0.8509,
      "step": 8130
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2897050380706787,
      "learning_rate": 1.6890400122628382e-05,
      "loss": 0.6465,
      "step": 8131
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.650744080543518,
      "learning_rate": 1.688959087247138e-05,
      "loss": 0.728,
      "step": 8132
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3078999519348145,
      "learning_rate": 1.6888781536419262e-05,
      "loss": 0.6171,
      "step": 8133
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.721640944480896,
      "learning_rate": 1.6887972114482122e-05,
      "loss": 0.6615,
      "step": 8134
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.429423451423645,
      "learning_rate": 1.6887162606670045e-05,
      "loss": 0.8017,
      "step": 8135
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8387434482574463,
      "learning_rate": 1.688635301299313e-05,
      "loss": 0.6291,
      "step": 8136
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4155664443969727,
      "learning_rate": 1.6885543333461463e-05,
      "loss": 0.9249,
      "step": 8137
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.69516122341156,
      "learning_rate": 1.6884733568085145e-05,
      "loss": 0.6715,
      "step": 8138
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9061779975891113,
      "learning_rate": 1.688392371687427e-05,
      "loss": 0.8223,
      "step": 8139
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.260259747505188,
      "learning_rate": 1.688311377983893e-05,
      "loss": 0.6448,
      "step": 8140
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4257261753082275,
      "learning_rate": 1.6882303756989233e-05,
      "loss": 0.7318,
      "step": 8141
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4678025245666504,
      "learning_rate": 1.688149364833527e-05,
      "loss": 0.5723,
      "step": 8142
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4991087913513184,
      "learning_rate": 1.688068345388714e-05,
      "loss": 0.8621,
      "step": 8143
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4355467557907104,
      "learning_rate": 1.687987317365495e-05,
      "loss": 0.7195,
      "step": 8144
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2873971462249756,
      "learning_rate": 1.6879062807648797e-05,
      "loss": 0.7371,
      "step": 8145
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3111450672149658,
      "learning_rate": 1.6878252355878788e-05,
      "loss": 0.6872,
      "step": 8146
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5604244470596313,
      "learning_rate": 1.687744181835502e-05,
      "loss": 0.7806,
      "step": 8147
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4719454050064087,
      "learning_rate": 1.687663119508761e-05,
      "loss": 0.8576,
      "step": 8148
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5497798919677734,
      "learning_rate": 1.6875820486086655e-05,
      "loss": 0.5681,
      "step": 8149
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.905558705329895,
      "learning_rate": 1.6875009691362266e-05,
      "loss": 0.9309,
      "step": 8150
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5691412687301636,
      "learning_rate": 1.6874198810924552e-05,
      "loss": 0.5775,
      "step": 8151
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3597427606582642,
      "learning_rate": 1.687338784478362e-05,
      "loss": 0.6813,
      "step": 8152
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.332463026046753,
      "learning_rate": 1.687257679294958e-05,
      "loss": 0.8105,
      "step": 8153
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4416102170944214,
      "learning_rate": 1.6871765655432547e-05,
      "loss": 0.8161,
      "step": 8154
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3326020240783691,
      "learning_rate": 1.6870954432242633e-05,
      "loss": 0.5782,
      "step": 8155
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7583893537521362,
      "learning_rate": 1.687014312338995e-05,
      "loss": 0.687,
      "step": 8156
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5700960159301758,
      "learning_rate": 1.6869331728884617e-05,
      "loss": 0.6367,
      "step": 8157
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3634918928146362,
      "learning_rate": 1.6868520248736745e-05,
      "loss": 0.795,
      "step": 8158
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4304141998291016,
      "learning_rate": 1.686770868295645e-05,
      "loss": 0.6549,
      "step": 8159
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.1385674476623535,
      "learning_rate": 1.6866897031553856e-05,
      "loss": 0.898,
      "step": 8160
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4185057878494263,
      "learning_rate": 1.6866085294539082e-05,
      "loss": 0.6713,
      "step": 8161
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6042096614837646,
      "learning_rate": 1.686527347192224e-05,
      "loss": 0.7038,
      "step": 8162
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5881237983703613,
      "learning_rate": 1.6864461563713458e-05,
      "loss": 0.7745,
      "step": 8163
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4247936010360718,
      "learning_rate": 1.6863649569922862e-05,
      "loss": 0.6981,
      "step": 8164
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3582823276519775,
      "learning_rate": 1.686283749056057e-05,
      "loss": 0.6072,
      "step": 8165
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2218060493469238,
      "learning_rate": 1.68620253256367e-05,
      "loss": 0.72,
      "step": 8166
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.356918215751648,
      "learning_rate": 1.686121307516139e-05,
      "loss": 0.7037,
      "step": 8167
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5839612483978271,
      "learning_rate": 1.6860400739144764e-05,
      "loss": 0.6089,
      "step": 8168
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4898761510849,
      "learning_rate": 1.6859588317596942e-05,
      "loss": 0.7175,
      "step": 8169
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7198067903518677,
      "learning_rate": 1.685877581052806e-05,
      "loss": 0.6714,
      "step": 8170
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6615564823150635,
      "learning_rate": 1.6857963217948244e-05,
      "loss": 0.7306,
      "step": 8171
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.36397123336792,
      "learning_rate": 1.6857150539867628e-05,
      "loss": 0.7163,
      "step": 8172
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3978352546691895,
      "learning_rate": 1.6856337776296343e-05,
      "loss": 0.7836,
      "step": 8173
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.310360312461853,
      "learning_rate": 1.6855524927244518e-05,
      "loss": 0.6703,
      "step": 8174
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5519275665283203,
      "learning_rate": 1.6854711992722296e-05,
      "loss": 0.6572,
      "step": 8175
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3740148544311523,
      "learning_rate": 1.6853898972739805e-05,
      "loss": 0.6001,
      "step": 8176
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.472191333770752,
      "learning_rate": 1.685308586730718e-05,
      "loss": 0.6776,
      "step": 8177
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5100702047348022,
      "learning_rate": 1.6852272676434563e-05,
      "loss": 0.7736,
      "step": 8178
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.509912133216858,
      "learning_rate": 1.685145940013209e-05,
      "loss": 0.8043,
      "step": 8179
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.373131275177002,
      "learning_rate": 1.6850646038409905e-05,
      "loss": 0.6432,
      "step": 8180
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5194207429885864,
      "learning_rate": 1.684983259127814e-05,
      "loss": 0.7542,
      "step": 8181
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3659523725509644,
      "learning_rate": 1.6849019058746942e-05,
      "loss": 0.7362,
      "step": 8182
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6740281581878662,
      "learning_rate": 1.6848205440826455e-05,
      "loss": 0.8029,
      "step": 8183
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4422264099121094,
      "learning_rate": 1.6847391737526816e-05,
      "loss": 0.657,
      "step": 8184
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5995922088623047,
      "learning_rate": 1.684657794885818e-05,
      "loss": 0.7569,
      "step": 8185
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6865181922912598,
      "learning_rate": 1.684576407483068e-05,
      "loss": 0.7524,
      "step": 8186
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.57173490524292,
      "learning_rate": 1.6844950115454476e-05,
      "loss": 0.7722,
      "step": 8187
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3468585014343262,
      "learning_rate": 1.6844136070739705e-05,
      "loss": 0.6342,
      "step": 8188
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4131691455841064,
      "learning_rate": 1.6843321940696525e-05,
      "loss": 0.6055,
      "step": 8189
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.432879090309143,
      "learning_rate": 1.684250772533508e-05,
      "loss": 0.6695,
      "step": 8190
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5909265279769897,
      "learning_rate": 1.684169342466552e-05,
      "loss": 0.6819,
      "step": 8191
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4100040197372437,
      "learning_rate": 1.6840879038698004e-05,
      "loss": 0.7307,
      "step": 8192
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3401199579238892,
      "learning_rate": 1.6840064567442682e-05,
      "loss": 0.6793,
      "step": 8193
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.416960597038269,
      "learning_rate": 1.6839250010909706e-05,
      "loss": 0.7286,
      "step": 8194
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4869418144226074,
      "learning_rate": 1.6838435369109234e-05,
      "loss": 0.7097,
      "step": 8195
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7447776794433594,
      "learning_rate": 1.683762064205142e-05,
      "loss": 0.7701,
      "step": 8196
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4765040874481201,
      "learning_rate": 1.6836805829746423e-05,
      "loss": 0.7782,
      "step": 8197
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5127137899398804,
      "learning_rate": 1.6835990932204406e-05,
      "loss": 0.6501,
      "step": 8198
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4971519708633423,
      "learning_rate": 1.6835175949435516e-05,
      "loss": 0.7797,
      "step": 8199
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.385479211807251,
      "learning_rate": 1.683436088144993e-05,
      "loss": 0.6609,
      "step": 8200
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7766259908676147,
      "learning_rate": 1.6833545728257797e-05,
      "loss": 0.6942,
      "step": 8201
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.498945951461792,
      "learning_rate": 1.6832730489869285e-05,
      "loss": 0.7104,
      "step": 8202
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4035024642944336,
      "learning_rate": 1.6831915166294558e-05,
      "loss": 0.7065,
      "step": 8203
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9596736431121826,
      "learning_rate": 1.683109975754378e-05,
      "loss": 0.7606,
      "step": 8204
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7100578546524048,
      "learning_rate": 1.683028426362712e-05,
      "loss": 0.8343,
      "step": 8205
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.618443489074707,
      "learning_rate": 1.682946868455474e-05,
      "loss": 0.7145,
      "step": 8206
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.509575605392456,
      "learning_rate": 1.682865302033681e-05,
      "loss": 0.6836,
      "step": 8207
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6036313772201538,
      "learning_rate": 1.6827837270983502e-05,
      "loss": 0.6815,
      "step": 8208
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5766499042510986,
      "learning_rate": 1.6827021436504983e-05,
      "loss": 0.6034,
      "step": 8209
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5726006031036377,
      "learning_rate": 1.6826205516911423e-05,
      "loss": 0.6546,
      "step": 8210
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3392478227615356,
      "learning_rate": 1.6825389512213e-05,
      "loss": 0.6552,
      "step": 8211
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2751652002334595,
      "learning_rate": 1.6824573422419882e-05,
      "loss": 0.6507,
      "step": 8212
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4639543294906616,
      "learning_rate": 1.6823757247542247e-05,
      "loss": 0.5885,
      "step": 8213
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3219571113586426,
      "learning_rate": 1.682294098759027e-05,
      "loss": 0.8801,
      "step": 8214
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3614416122436523,
      "learning_rate": 1.6822124642574124e-05,
      "loss": 0.755,
      "step": 8215
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3699108362197876,
      "learning_rate": 1.6821308212503988e-05,
      "loss": 0.6966,
      "step": 8216
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3533436059951782,
      "learning_rate": 1.682049169739005e-05,
      "loss": 0.5827,
      "step": 8217
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5090219974517822,
      "learning_rate": 1.6819675097242475e-05,
      "loss": 0.6695,
      "step": 8218
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8778101205825806,
      "learning_rate": 1.6818858412071454e-05,
      "loss": 0.7889,
      "step": 8219
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7244538068771362,
      "learning_rate": 1.6818041641887164e-05,
      "loss": 0.8115,
      "step": 8220
    },
    {
      "epoch": 0.85,
      "grad_norm": 3.0009219646453857,
      "learning_rate": 1.681722478669979e-05,
      "loss": 0.8017,
      "step": 8221
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5436536073684692,
      "learning_rate": 1.681640784651952e-05,
      "loss": 0.7638,
      "step": 8222
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5616941452026367,
      "learning_rate": 1.6815590821356527e-05,
      "loss": 0.6353,
      "step": 8223
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4193472862243652,
      "learning_rate": 1.6814773711221012e-05,
      "loss": 0.5826,
      "step": 8224
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2227160930633545,
      "learning_rate": 1.681395651612315e-05,
      "loss": 0.6549,
      "step": 8225
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7141170501708984,
      "learning_rate": 1.681313923607314e-05,
      "loss": 0.6875,
      "step": 8226
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4561748504638672,
      "learning_rate": 1.6812321871081164e-05,
      "loss": 0.6016,
      "step": 8227
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5362768173217773,
      "learning_rate": 1.6811504421157413e-05,
      "loss": 0.6381,
      "step": 8228
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3391932249069214,
      "learning_rate": 1.681068688631208e-05,
      "loss": 0.6813,
      "step": 8229
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7216472625732422,
      "learning_rate": 1.6809869266555358e-05,
      "loss": 0.628,
      "step": 8230
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7858415842056274,
      "learning_rate": 1.680905156189744e-05,
      "loss": 0.7994,
      "step": 8231
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2337738275527954,
      "learning_rate": 1.680823377234852e-05,
      "loss": 0.6998,
      "step": 8232
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3086169958114624,
      "learning_rate": 1.6807415897918792e-05,
      "loss": 0.743,
      "step": 8233
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4690728187561035,
      "learning_rate": 1.6806597938618457e-05,
      "loss": 0.6323,
      "step": 8234
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.636590838432312,
      "learning_rate": 1.6805779894457712e-05,
      "loss": 0.731,
      "step": 8235
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6252796649932861,
      "learning_rate": 1.6804961765446753e-05,
      "loss": 0.8329,
      "step": 8236
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7785643339157104,
      "learning_rate": 1.6804143551595785e-05,
      "loss": 0.8286,
      "step": 8237
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4687743186950684,
      "learning_rate": 1.6803325252915003e-05,
      "loss": 0.6676,
      "step": 8238
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7142120599746704,
      "learning_rate": 1.680250686941461e-05,
      "loss": 0.7899,
      "step": 8239
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6672190427780151,
      "learning_rate": 1.680168840110481e-05,
      "loss": 0.5646,
      "step": 8240
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4059301614761353,
      "learning_rate": 1.6800869847995813e-05,
      "loss": 0.6316,
      "step": 8241
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4887923002243042,
      "learning_rate": 1.6800051210097813e-05,
      "loss": 0.7447,
      "step": 8242
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5534886121749878,
      "learning_rate": 1.6799232487421028e-05,
      "loss": 0.6053,
      "step": 8243
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4234737157821655,
      "learning_rate": 1.6798413679975655e-05,
      "loss": 0.6188,
      "step": 8244
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.554016351699829,
      "learning_rate": 1.6797594787771907e-05,
      "loss": 0.8956,
      "step": 8245
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.502843976020813,
      "learning_rate": 1.6796775810819995e-05,
      "loss": 0.6353,
      "step": 8246
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2871187925338745,
      "learning_rate": 1.6795956749130127e-05,
      "loss": 0.6774,
      "step": 8247
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3088370561599731,
      "learning_rate": 1.6795137602712517e-05,
      "loss": 0.6683,
      "step": 8248
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1533827781677246,
      "learning_rate": 1.6794318371577378e-05,
      "loss": 0.5856,
      "step": 8249
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7732175588607788,
      "learning_rate": 1.6793499055734915e-05,
      "loss": 0.6679,
      "step": 8250
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4604668617248535,
      "learning_rate": 1.6792679655195355e-05,
      "loss": 0.748,
      "step": 8251
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2710872888565063,
      "learning_rate": 1.6791860169968906e-05,
      "loss": 0.7094,
      "step": 8252
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3138575553894043,
      "learning_rate": 1.6791040600065785e-05,
      "loss": 0.8033,
      "step": 8253
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5161144733428955,
      "learning_rate": 1.6790220945496216e-05,
      "loss": 0.7142,
      "step": 8254
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4009991884231567,
      "learning_rate": 1.6789401206270414e-05,
      "loss": 0.6323,
      "step": 8255
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.600716233253479,
      "learning_rate": 1.6788581382398596e-05,
      "loss": 0.6919,
      "step": 8256
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9167734384536743,
      "learning_rate": 1.6787761473890985e-05,
      "loss": 0.7561,
      "step": 8257
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6015779972076416,
      "learning_rate": 1.6786941480757807e-05,
      "loss": 0.8523,
      "step": 8258
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3453412055969238,
      "learning_rate": 1.678612140300928e-05,
      "loss": 0.7844,
      "step": 8259
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.8968368768692017,
      "learning_rate": 1.6785301240655633e-05,
      "loss": 0.5814,
      "step": 8260
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3302465677261353,
      "learning_rate": 1.6784480993707087e-05,
      "loss": 0.5936,
      "step": 8261
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3808146715164185,
      "learning_rate": 1.6783660662173872e-05,
      "loss": 0.634,
      "step": 8262
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.573520302772522,
      "learning_rate": 1.678284024606621e-05,
      "loss": 0.644,
      "step": 8263
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.468636155128479,
      "learning_rate": 1.6782019745394335e-05,
      "loss": 0.716,
      "step": 8264
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6309518814086914,
      "learning_rate": 1.6781199160168474e-05,
      "loss": 0.5913,
      "step": 8265
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5033327341079712,
      "learning_rate": 1.6780378490398855e-05,
      "loss": 0.5859,
      "step": 8266
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.755429744720459,
      "learning_rate": 1.6779557736095718e-05,
      "loss": 0.7195,
      "step": 8267
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4398530721664429,
      "learning_rate": 1.6778736897269286e-05,
      "loss": 0.6794,
      "step": 8268
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.633407711982727,
      "learning_rate": 1.6777915973929798e-05,
      "loss": 0.4573,
      "step": 8269
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6621654033660889,
      "learning_rate": 1.6777094966087486e-05,
      "loss": 0.5954,
      "step": 8270
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8942534923553467,
      "learning_rate": 1.6776273873752588e-05,
      "loss": 0.8033,
      "step": 8271
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.658281922340393,
      "learning_rate": 1.6775452696935342e-05,
      "loss": 0.8541,
      "step": 8272
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6588072776794434,
      "learning_rate": 1.6774631435645984e-05,
      "loss": 0.7579,
      "step": 8273
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3690999746322632,
      "learning_rate": 1.6773810089894752e-05,
      "loss": 0.6464,
      "step": 8274
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3223319053649902,
      "learning_rate": 1.677298865969189e-05,
      "loss": 0.7205,
      "step": 8275
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7032095193862915,
      "learning_rate": 1.677216714504763e-05,
      "loss": 0.5906,
      "step": 8276
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.420508623123169,
      "learning_rate": 1.6771345545972226e-05,
      "loss": 0.6657,
      "step": 8277
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6231672763824463,
      "learning_rate": 1.677052386247591e-05,
      "loss": 0.7001,
      "step": 8278
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.4627912044525146,
      "learning_rate": 1.6769702094568937e-05,
      "loss": 0.8963,
      "step": 8279
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8023097515106201,
      "learning_rate": 1.6768880242261547e-05,
      "loss": 0.8434,
      "step": 8280
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8879822492599487,
      "learning_rate": 1.6768058305563986e-05,
      "loss": 0.82,
      "step": 8281
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5871435403823853,
      "learning_rate": 1.6767236284486498e-05,
      "loss": 0.6935,
      "step": 8282
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5809985399246216,
      "learning_rate": 1.6766414179039337e-05,
      "loss": 0.6424,
      "step": 8283
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4428431987762451,
      "learning_rate": 1.6765591989232753e-05,
      "loss": 0.6221,
      "step": 8284
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4487977027893066,
      "learning_rate": 1.6764769715076992e-05,
      "loss": 0.6128,
      "step": 8285
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.224963665008545,
      "learning_rate": 1.676394735658231e-05,
      "loss": 0.5735,
      "step": 8286
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6077872514724731,
      "learning_rate": 1.6763124913758958e-05,
      "loss": 0.8298,
      "step": 8287
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4190140962600708,
      "learning_rate": 1.676230238661719e-05,
      "loss": 0.7967,
      "step": 8288
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.507642388343811,
      "learning_rate": 1.6761479775167255e-05,
      "loss": 0.6807,
      "step": 8289
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2860476970672607,
      "learning_rate": 1.6760657079419414e-05,
      "loss": 0.6508,
      "step": 8290
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5252621173858643,
      "learning_rate": 1.675983429938393e-05,
      "loss": 0.7439,
      "step": 8291
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5660669803619385,
      "learning_rate": 1.675901143507105e-05,
      "loss": 0.762,
      "step": 8292
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4842031002044678,
      "learning_rate": 1.6758188486491038e-05,
      "loss": 0.6906,
      "step": 8293
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4425241947174072,
      "learning_rate": 1.6757365453654154e-05,
      "loss": 0.6841,
      "step": 8294
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4555202722549438,
      "learning_rate": 1.6756542336570655e-05,
      "loss": 0.7065,
      "step": 8295
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8940073251724243,
      "learning_rate": 1.675571913525081e-05,
      "loss": 0.8001,
      "step": 8296
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6404979228973389,
      "learning_rate": 1.6754895849704883e-05,
      "loss": 0.6251,
      "step": 8297
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1665996313095093,
      "learning_rate": 1.6754072479943128e-05,
      "loss": 0.7182,
      "step": 8298
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.81267511844635,
      "learning_rate": 1.675324902597582e-05,
      "loss": 0.6033,
      "step": 8299
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4546234607696533,
      "learning_rate": 1.6752425487813217e-05,
      "loss": 0.7238,
      "step": 8300
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6543632745742798,
      "learning_rate": 1.6751601865465597e-05,
      "loss": 0.6479,
      "step": 8301
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2518502473831177,
      "learning_rate": 1.675077815894322e-05,
      "loss": 0.6556,
      "step": 8302
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5344077348709106,
      "learning_rate": 1.6749954368256357e-05,
      "loss": 0.74,
      "step": 8303
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5136367082595825,
      "learning_rate": 1.674913049341528e-05,
      "loss": 0.8195,
      "step": 8304
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3386698961257935,
      "learning_rate": 1.6748306534430258e-05,
      "loss": 0.7716,
      "step": 8305
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.487676739692688,
      "learning_rate": 1.6747482491311567e-05,
      "loss": 0.757,
      "step": 8306
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8131083250045776,
      "learning_rate": 1.674665836406948e-05,
      "loss": 0.8865,
      "step": 8307
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5736699104309082,
      "learning_rate": 1.674583415271427e-05,
      "loss": 0.8146,
      "step": 8308
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.69645357131958,
      "learning_rate": 1.6745009857256212e-05,
      "loss": 0.6665,
      "step": 8309
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2254300117492676,
      "learning_rate": 1.674418547770559e-05,
      "loss": 0.8311,
      "step": 8310
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5901528596878052,
      "learning_rate": 1.6743361014072673e-05,
      "loss": 0.766,
      "step": 8311
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4809705018997192,
      "learning_rate": 1.674253646636774e-05,
      "loss": 0.7829,
      "step": 8312
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.232926607131958,
      "learning_rate": 1.674171183460108e-05,
      "loss": 0.6763,
      "step": 8313
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3333547115325928,
      "learning_rate": 1.6740887118782966e-05,
      "loss": 0.7176,
      "step": 8314
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.501574993133545,
      "learning_rate": 1.674006231892368e-05,
      "loss": 0.7185,
      "step": 8315
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7047536373138428,
      "learning_rate": 1.6739237435033508e-05,
      "loss": 0.774,
      "step": 8316
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2500216960906982,
      "learning_rate": 1.6738412467122734e-05,
      "loss": 0.6179,
      "step": 8317
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4316562414169312,
      "learning_rate": 1.6737587415201644e-05,
      "loss": 0.803,
      "step": 8318
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3063466548919678,
      "learning_rate": 1.673676227928052e-05,
      "loss": 0.7839,
      "step": 8319
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.9743552207946777,
      "learning_rate": 1.6735937059369655e-05,
      "loss": 0.9121,
      "step": 8320
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3658519983291626,
      "learning_rate": 1.6735111755479333e-05,
      "loss": 0.7413,
      "step": 8321
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4963438510894775,
      "learning_rate": 1.6734286367619846e-05,
      "loss": 0.6372,
      "step": 8322
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5639536380767822,
      "learning_rate": 1.6733460895801485e-05,
      "loss": 0.5962,
      "step": 8323
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.587456226348877,
      "learning_rate": 1.6732635340034535e-05,
      "loss": 0.8426,
      "step": 8324
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6111754179000854,
      "learning_rate": 1.6731809700329296e-05,
      "loss": 0.5966,
      "step": 8325
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5192514657974243,
      "learning_rate": 1.6730983976696056e-05,
      "loss": 0.6497,
      "step": 8326
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.402335286140442,
      "learning_rate": 1.6730158169145117e-05,
      "loss": 0.5573,
      "step": 8327
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8544951677322388,
      "learning_rate": 1.6729332277686765e-05,
      "loss": 0.5839,
      "step": 8328
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5633114576339722,
      "learning_rate": 1.6728506302331304e-05,
      "loss": 0.6299,
      "step": 8329
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5636439323425293,
      "learning_rate": 1.672768024308903e-05,
      "loss": 0.5244,
      "step": 8330
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3125349283218384,
      "learning_rate": 1.672685409997024e-05,
      "loss": 0.7151,
      "step": 8331
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5527678728103638,
      "learning_rate": 1.6726027872985236e-05,
      "loss": 0.8313,
      "step": 8332
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3486510515213013,
      "learning_rate": 1.6725201562144318e-05,
      "loss": 0.6652,
      "step": 8333
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7256032228469849,
      "learning_rate": 1.6724375167457787e-05,
      "loss": 0.72,
      "step": 8334
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.489014744758606,
      "learning_rate": 1.672354868893595e-05,
      "loss": 0.817,
      "step": 8335
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5652114152908325,
      "learning_rate": 1.6722722126589106e-05,
      "loss": 0.6971,
      "step": 8336
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6335198879241943,
      "learning_rate": 1.672189548042756e-05,
      "loss": 0.7223,
      "step": 8337
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0745596885681152,
      "learning_rate": 1.672106875046162e-05,
      "loss": 0.8549,
      "step": 8338
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.538541555404663,
      "learning_rate": 1.6720241936701595e-05,
      "loss": 0.8039,
      "step": 8339
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.731515884399414,
      "learning_rate": 1.6719415039157793e-05,
      "loss": 0.7889,
      "step": 8340
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5679324865341187,
      "learning_rate": 1.671858805784052e-05,
      "loss": 0.5885,
      "step": 8341
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2781251668930054,
      "learning_rate": 1.671776099276009e-05,
      "loss": 0.7408,
      "step": 8342
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2222691774368286,
      "learning_rate": 1.671693384392681e-05,
      "loss": 0.687,
      "step": 8343
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5505142211914062,
      "learning_rate": 1.6716106611350994e-05,
      "loss": 0.7901,
      "step": 8344
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3066035509109497,
      "learning_rate": 1.6715279295042956e-05,
      "loss": 0.7174,
      "step": 8345
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.338869333267212,
      "learning_rate": 1.6714451895013013e-05,
      "loss": 0.6382,
      "step": 8346
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2588616609573364,
      "learning_rate": 1.6713624411271475e-05,
      "loss": 0.7034,
      "step": 8347
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3460811376571655,
      "learning_rate": 1.6712796843828665e-05,
      "loss": 0.6401,
      "step": 8348
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.498546838760376,
      "learning_rate": 1.6711969192694893e-05,
      "loss": 0.826,
      "step": 8349
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6489931344985962,
      "learning_rate": 1.6711141457880487e-05,
      "loss": 0.6696,
      "step": 8350
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2377578020095825,
      "learning_rate": 1.6710313639395757e-05,
      "loss": 0.62,
      "step": 8351
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6678309440612793,
      "learning_rate": 1.6709485737251032e-05,
      "loss": 0.6267,
      "step": 8352
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7547612190246582,
      "learning_rate": 1.6708657751456627e-05,
      "loss": 0.5096,
      "step": 8353
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.470347285270691,
      "learning_rate": 1.6707829682022868e-05,
      "loss": 0.6672,
      "step": 8354
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6051141023635864,
      "learning_rate": 1.6707001528960082e-05,
      "loss": 0.7633,
      "step": 8355
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6818649768829346,
      "learning_rate": 1.6706173292278584e-05,
      "loss": 0.648,
      "step": 8356
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.512085199356079,
      "learning_rate": 1.670534497198871e-05,
      "loss": 0.6059,
      "step": 8357
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7415059804916382,
      "learning_rate": 1.6704516568100786e-05,
      "loss": 0.8028,
      "step": 8358
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5414105653762817,
      "learning_rate": 1.6703688080625134e-05,
      "loss": 0.6431,
      "step": 8359
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5799492597579956,
      "learning_rate": 1.670285950957209e-05,
      "loss": 0.7334,
      "step": 8360
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4359065294265747,
      "learning_rate": 1.670203085495198e-05,
      "loss": 0.7525,
      "step": 8361
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5692133903503418,
      "learning_rate": 1.6701202116775134e-05,
      "loss": 0.7382,
      "step": 8362
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6471666097640991,
      "learning_rate": 1.6700373295051885e-05,
      "loss": 0.7799,
      "step": 8363
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.070272922515869,
      "learning_rate": 1.6699544389792567e-05,
      "loss": 0.8016,
      "step": 8364
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5362309217453003,
      "learning_rate": 1.669871540100752e-05,
      "loss": 0.5797,
      "step": 8365
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5605313777923584,
      "learning_rate": 1.6697886328707066e-05,
      "loss": 0.6874,
      "step": 8366
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.700405240058899,
      "learning_rate": 1.6697057172901555e-05,
      "loss": 0.7604,
      "step": 8367
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3536927700042725,
      "learning_rate": 1.6696227933601315e-05,
      "loss": 0.7583,
      "step": 8368
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4868953227996826,
      "learning_rate": 1.6695398610816692e-05,
      "loss": 0.7303,
      "step": 8369
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3905138969421387,
      "learning_rate": 1.669456920455802e-05,
      "loss": 0.8339,
      "step": 8370
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5836422443389893,
      "learning_rate": 1.6693739714835637e-05,
      "loss": 0.7385,
      "step": 8371
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5299407243728638,
      "learning_rate": 1.6692910141659892e-05,
      "loss": 0.7548,
      "step": 8372
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4058221578598022,
      "learning_rate": 1.6692080485041126e-05,
      "loss": 0.6804,
      "step": 8373
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4576895236968994,
      "learning_rate": 1.669125074498968e-05,
      "loss": 0.6919,
      "step": 8374
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6137572526931763,
      "learning_rate": 1.66904209215159e-05,
      "loss": 0.6509,
      "step": 8375
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4197027683258057,
      "learning_rate": 1.6689591014630128e-05,
      "loss": 0.6032,
      "step": 8376
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4596188068389893,
      "learning_rate": 1.6688761024342716e-05,
      "loss": 0.7912,
      "step": 8377
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7490378618240356,
      "learning_rate": 1.6687930950664005e-05,
      "loss": 0.7177,
      "step": 8378
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2155452966690063,
      "learning_rate": 1.6687100793604357e-05,
      "loss": 0.6756,
      "step": 8379
    },
    {
      "epoch": 0.87,
      "grad_norm": 9.335400581359863,
      "learning_rate": 1.6686270553174105e-05,
      "loss": 0.7153,
      "step": 8380
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7395919561386108,
      "learning_rate": 1.6685440229383614e-05,
      "loss": 0.7698,
      "step": 8381
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2748295068740845,
      "learning_rate": 1.668460982224323e-05,
      "loss": 0.7837,
      "step": 8382
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7014139890670776,
      "learning_rate": 1.6683779331763307e-05,
      "loss": 0.7908,
      "step": 8383
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3337788581848145,
      "learning_rate": 1.6682948757954196e-05,
      "loss": 0.6795,
      "step": 8384
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.541327953338623,
      "learning_rate": 1.6682118100826255e-05,
      "loss": 0.7173,
      "step": 8385
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4598957300186157,
      "learning_rate": 1.668128736038984e-05,
      "loss": 0.8196,
      "step": 8386
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.741611123085022,
      "learning_rate": 1.668045653665531e-05,
      "loss": 0.9091,
      "step": 8387
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.444918155670166,
      "learning_rate": 1.6679625629633017e-05,
      "loss": 0.6877,
      "step": 8388
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3980714082717896,
      "learning_rate": 1.6678794639333324e-05,
      "loss": 0.7279,
      "step": 8389
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3764543533325195,
      "learning_rate": 1.6677963565766595e-05,
      "loss": 0.6388,
      "step": 8390
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4041131734848022,
      "learning_rate": 1.6677132408943188e-05,
      "loss": 0.7188,
      "step": 8391
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3767374753952026,
      "learning_rate": 1.667630116887346e-05,
      "loss": 0.6832,
      "step": 8392
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5459957122802734,
      "learning_rate": 1.6675469845567782e-05,
      "loss": 0.7638,
      "step": 8393
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2880630493164062,
      "learning_rate": 1.6674638439036518e-05,
      "loss": 0.6188,
      "step": 8394
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.684656023979187,
      "learning_rate": 1.667380694929003e-05,
      "loss": 0.7188,
      "step": 8395
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4915958642959595,
      "learning_rate": 1.6672975376338686e-05,
      "loss": 0.7513,
      "step": 8396
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.533210039138794,
      "learning_rate": 1.6672143720192854e-05,
      "loss": 0.7134,
      "step": 8397
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.399973750114441,
      "learning_rate": 1.66713119808629e-05,
      "loss": 0.6937,
      "step": 8398
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5610381364822388,
      "learning_rate": 1.6670480158359197e-05,
      "loss": 0.6888,
      "step": 8399
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6608803272247314,
      "learning_rate": 1.6669648252692115e-05,
      "loss": 0.7611,
      "step": 8400
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5256797075271606,
      "learning_rate": 1.6668816263872023e-05,
      "loss": 0.7108,
      "step": 8401
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5639140605926514,
      "learning_rate": 1.6667984191909297e-05,
      "loss": 0.6727,
      "step": 8402
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7239360809326172,
      "learning_rate": 1.6667152036814306e-05,
      "loss": 0.8337,
      "step": 8403
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.670719027519226,
      "learning_rate": 1.6666319798597434e-05,
      "loss": 0.6407,
      "step": 8404
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5449810028076172,
      "learning_rate": 1.6665487477269045e-05,
      "loss": 0.5905,
      "step": 8405
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3496164083480835,
      "learning_rate": 1.6664655072839524e-05,
      "loss": 0.6928,
      "step": 8406
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3371894359588623,
      "learning_rate": 1.666382258531925e-05,
      "loss": 0.722,
      "step": 8407
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3025591373443604,
      "learning_rate": 1.6662990014718596e-05,
      "loss": 0.6991,
      "step": 8408
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5609359741210938,
      "learning_rate": 1.6662157361047943e-05,
      "loss": 0.7772,
      "step": 8409
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.619039535522461,
      "learning_rate": 1.6661324624317673e-05,
      "loss": 0.6765,
      "step": 8410
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.756807565689087,
      "learning_rate": 1.666049180453817e-05,
      "loss": 0.693,
      "step": 8411
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7913979291915894,
      "learning_rate": 1.665965890171982e-05,
      "loss": 0.7836,
      "step": 8412
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.478540301322937,
      "learning_rate": 1.6658825915872998e-05,
      "loss": 0.7269,
      "step": 8413
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4962222576141357,
      "learning_rate": 1.6657992847008096e-05,
      "loss": 0.5981,
      "step": 8414
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8147116899490356,
      "learning_rate": 1.6657159695135496e-05,
      "loss": 0.7168,
      "step": 8415
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5434842109680176,
      "learning_rate": 1.665632646026559e-05,
      "loss": 0.7672,
      "step": 8416
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3450759649276733,
      "learning_rate": 1.6655493142408764e-05,
      "loss": 0.7177,
      "step": 8417
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.790334939956665,
      "learning_rate": 1.6654659741575407e-05,
      "loss": 0.7419,
      "step": 8418
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.591663122177124,
      "learning_rate": 1.6653826257775905e-05,
      "loss": 0.7953,
      "step": 8419
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.799546480178833,
      "learning_rate": 1.665299269102066e-05,
      "loss": 0.7105,
      "step": 8420
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.409460186958313,
      "learning_rate": 1.6652159041320053e-05,
      "loss": 0.7366,
      "step": 8421
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5711270570755005,
      "learning_rate": 1.6651325308684484e-05,
      "loss": 0.719,
      "step": 8422
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5621269941329956,
      "learning_rate": 1.6650491493124345e-05,
      "loss": 0.6533,
      "step": 8423
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4144936800003052,
      "learning_rate": 1.664965759465003e-05,
      "loss": 0.7335,
      "step": 8424
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5288071632385254,
      "learning_rate": 1.6648823613271944e-05,
      "loss": 0.7448,
      "step": 8425
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.338884711265564,
      "learning_rate": 1.6647989549000475e-05,
      "loss": 0.4699,
      "step": 8426
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7050726413726807,
      "learning_rate": 1.6647155401846026e-05,
      "loss": 0.7882,
      "step": 8427
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6895344257354736,
      "learning_rate": 1.6646321171818994e-05,
      "loss": 0.7193,
      "step": 8428
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3817459344863892,
      "learning_rate": 1.6645486858929786e-05,
      "loss": 0.6994,
      "step": 8429
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5731345415115356,
      "learning_rate": 1.6644652463188792e-05,
      "loss": 0.6324,
      "step": 8430
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.418687343597412,
      "learning_rate": 1.6643817984606427e-05,
      "loss": 0.7355,
      "step": 8431
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.817216396331787,
      "learning_rate": 1.6642983423193087e-05,
      "loss": 0.7242,
      "step": 8432
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8407388925552368,
      "learning_rate": 1.6642148778959182e-05,
      "loss": 0.8544,
      "step": 8433
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9503083229064941,
      "learning_rate": 1.6641314051915115e-05,
      "loss": 0.8118,
      "step": 8434
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5047616958618164,
      "learning_rate": 1.6640479242071286e-05,
      "loss": 0.7069,
      "step": 8435
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4957764148712158,
      "learning_rate": 1.663964434943812e-05,
      "loss": 0.7157,
      "step": 8436
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3831045627593994,
      "learning_rate": 1.6638809374026007e-05,
      "loss": 0.7046,
      "step": 8437
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7014901638031006,
      "learning_rate": 1.6637974315845373e-05,
      "loss": 0.7548,
      "step": 8438
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5744884014129639,
      "learning_rate": 1.663713917490662e-05,
      "loss": 0.6362,
      "step": 8439
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3331323862075806,
      "learning_rate": 1.6636303951220155e-05,
      "loss": 0.6575,
      "step": 8440
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4792027473449707,
      "learning_rate": 1.6635468644796407e-05,
      "loss": 0.7043,
      "step": 8441
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8420970439910889,
      "learning_rate": 1.6634633255645777e-05,
      "loss": 0.9499,
      "step": 8442
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6622065305709839,
      "learning_rate": 1.6633797783778683e-05,
      "loss": 0.4975,
      "step": 8443
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4864505529403687,
      "learning_rate": 1.6632962229205544e-05,
      "loss": 0.7979,
      "step": 8444
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3365241289138794,
      "learning_rate": 1.6632126591936772e-05,
      "loss": 0.5832,
      "step": 8445
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.525325894355774,
      "learning_rate": 1.663129087198279e-05,
      "loss": 0.7487,
      "step": 8446
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2493271827697754,
      "learning_rate": 1.663045506935402e-05,
      "loss": 0.7391,
      "step": 8447
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.32066810131073,
      "learning_rate": 1.6629619184060876e-05,
      "loss": 0.5446,
      "step": 8448
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7811423540115356,
      "learning_rate": 1.6628783216113778e-05,
      "loss": 0.5542,
      "step": 8449
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4076470136642456,
      "learning_rate": 1.6627947165523152e-05,
      "loss": 0.7167,
      "step": 8450
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6964129209518433,
      "learning_rate": 1.6627111032299422e-05,
      "loss": 0.6353,
      "step": 8451
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6359748840332031,
      "learning_rate": 1.6626274816453014e-05,
      "loss": 0.5741,
      "step": 8452
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6572949886322021,
      "learning_rate": 1.6625438517994348e-05,
      "loss": 0.7437,
      "step": 8453
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4831331968307495,
      "learning_rate": 1.6624602136933852e-05,
      "loss": 0.6566,
      "step": 8454
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1935250759124756,
      "learning_rate": 1.662376567328196e-05,
      "loss": 0.6384,
      "step": 8455
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6613234281539917,
      "learning_rate": 1.662292912704909e-05,
      "loss": 0.7184,
      "step": 8456
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3498787879943848,
      "learning_rate": 1.662209249824568e-05,
      "loss": 0.8513,
      "step": 8457
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5317187309265137,
      "learning_rate": 1.6621255786882153e-05,
      "loss": 0.7989,
      "step": 8458
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5237164497375488,
      "learning_rate": 1.662041899296895e-05,
      "loss": 0.6784,
      "step": 8459
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3206528425216675,
      "learning_rate": 1.6619582116516494e-05,
      "loss": 0.5586,
      "step": 8460
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3382080793380737,
      "learning_rate": 1.6618745157535227e-05,
      "loss": 0.6933,
      "step": 8461
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4180738925933838,
      "learning_rate": 1.6617908116035576e-05,
      "loss": 0.561,
      "step": 8462
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3949655294418335,
      "learning_rate": 1.6617070992027985e-05,
      "loss": 0.6426,
      "step": 8463
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6894035339355469,
      "learning_rate": 1.6616233785522884e-05,
      "loss": 0.684,
      "step": 8464
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.398346185684204,
      "learning_rate": 1.6615396496530713e-05,
      "loss": 0.7309,
      "step": 8465
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.454471468925476,
      "learning_rate": 1.6614559125061914e-05,
      "loss": 0.7778,
      "step": 8466
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7546595335006714,
      "learning_rate": 1.661372167112692e-05,
      "loss": 0.7582,
      "step": 8467
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.527612328529358,
      "learning_rate": 1.6612884134736177e-05,
      "loss": 0.6703,
      "step": 8468
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6602762937545776,
      "learning_rate": 1.6612046515900123e-05,
      "loss": 0.6788,
      "step": 8469
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4470449686050415,
      "learning_rate": 1.6611208814629203e-05,
      "loss": 0.6027,
      "step": 8470
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.235303521156311,
      "learning_rate": 1.6610371030933867e-05,
      "loss": 0.6329,
      "step": 8471
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8335282802581787,
      "learning_rate": 1.6609533164824552e-05,
      "loss": 0.7467,
      "step": 8472
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.407562255859375,
      "learning_rate": 1.66086952163117e-05,
      "loss": 0.6914,
      "step": 8473
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.630335807800293,
      "learning_rate": 1.660785718540577e-05,
      "loss": 0.7024,
      "step": 8474
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.320635437965393,
      "learning_rate": 1.6607019072117206e-05,
      "loss": 0.6707,
      "step": 8475
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4949626922607422,
      "learning_rate": 1.660618087645645e-05,
      "loss": 0.6835,
      "step": 8476
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7212680578231812,
      "learning_rate": 1.6605342598433964e-05,
      "loss": 0.7028,
      "step": 8477
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9173738956451416,
      "learning_rate": 1.660450423806019e-05,
      "loss": 0.8177,
      "step": 8478
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5155465602874756,
      "learning_rate": 1.660366579534558e-05,
      "loss": 0.7712,
      "step": 8479
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.569700837135315,
      "learning_rate": 1.660282727030059e-05,
      "loss": 0.5155,
      "step": 8480
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7083498239517212,
      "learning_rate": 1.6601988662935678e-05,
      "loss": 0.6962,
      "step": 8481
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3829392194747925,
      "learning_rate": 1.6601149973261294e-05,
      "loss": 0.5429,
      "step": 8482
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2703235149383545,
      "learning_rate": 1.6600311201287894e-05,
      "loss": 0.6138,
      "step": 8483
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2659581899642944,
      "learning_rate": 1.659947234702594e-05,
      "loss": 0.7856,
      "step": 8484
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7670259475708008,
      "learning_rate": 1.6598633410485882e-05,
      "loss": 0.6504,
      "step": 8485
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4149205684661865,
      "learning_rate": 1.6597794391678188e-05,
      "loss": 0.7225,
      "step": 8486
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4274262189865112,
      "learning_rate": 1.6596955290613315e-05,
      "loss": 0.7161,
      "step": 8487
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3633410930633545,
      "learning_rate": 1.6596116107301723e-05,
      "loss": 0.7587,
      "step": 8488
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5658681392669678,
      "learning_rate": 1.6595276841753877e-05,
      "loss": 0.8176,
      "step": 8489
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5870286226272583,
      "learning_rate": 1.659443749398024e-05,
      "loss": 0.7238,
      "step": 8490
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5776002407073975,
      "learning_rate": 1.6593598063991273e-05,
      "loss": 0.7296,
      "step": 8491
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7193931341171265,
      "learning_rate": 1.6592758551797445e-05,
      "loss": 0.7465,
      "step": 8492
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.252908706665039,
      "learning_rate": 1.6591918957409223e-05,
      "loss": 0.5182,
      "step": 8493
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4359118938446045,
      "learning_rate": 1.659107928083707e-05,
      "loss": 0.6651,
      "step": 8494
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8758621215820312,
      "learning_rate": 1.6590239522091463e-05,
      "loss": 0.6815,
      "step": 8495
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6319502592086792,
      "learning_rate": 1.6589399681182864e-05,
      "loss": 0.7713,
      "step": 8496
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7167329788208008,
      "learning_rate": 1.6588559758121742e-05,
      "loss": 0.6007,
      "step": 8497
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.411393642425537,
      "learning_rate": 1.6587719752918577e-05,
      "loss": 0.6538,
      "step": 8498
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4174386262893677,
      "learning_rate": 1.6586879665583838e-05,
      "loss": 0.7362,
      "step": 8499
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6229088306427002,
      "learning_rate": 1.6586039496127994e-05,
      "loss": 0.5709,
      "step": 8500
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4217305183410645,
      "learning_rate": 1.6585199244561525e-05,
      "loss": 0.6972,
      "step": 8501
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5987993478775024,
      "learning_rate": 1.6584358910894907e-05,
      "loss": 0.5549,
      "step": 8502
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.705788016319275,
      "learning_rate": 1.6583518495138612e-05,
      "loss": 0.7982,
      "step": 8503
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3351387977600098,
      "learning_rate": 1.6582677997303124e-05,
      "loss": 0.6811,
      "step": 8504
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6544098854064941,
      "learning_rate": 1.6581837417398916e-05,
      "loss": 0.6769,
      "step": 8505
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3434771299362183,
      "learning_rate": 1.6580996755436473e-05,
      "loss": 0.7359,
      "step": 8506
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6990280151367188,
      "learning_rate": 1.6580156011426275e-05,
      "loss": 0.7405,
      "step": 8507
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5273301601409912,
      "learning_rate": 1.65793151853788e-05,
      "loss": 0.7565,
      "step": 8508
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.470343828201294,
      "learning_rate": 1.6578474277304532e-05,
      "loss": 0.6661,
      "step": 8509
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4077750444412231,
      "learning_rate": 1.6577633287213957e-05,
      "loss": 0.6571,
      "step": 8510
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.676759123802185,
      "learning_rate": 1.6576792215117558e-05,
      "loss": 0.7728,
      "step": 8511
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2319031953811646,
      "learning_rate": 1.657595106102582e-05,
      "loss": 0.7505,
      "step": 8512
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5267035961151123,
      "learning_rate": 1.6575109824949238e-05,
      "loss": 0.5693,
      "step": 8513
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4246678352355957,
      "learning_rate": 1.657426850689829e-05,
      "loss": 0.7212,
      "step": 8514
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.501835584640503,
      "learning_rate": 1.657342710688347e-05,
      "loss": 0.5205,
      "step": 8515
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3745474815368652,
      "learning_rate": 1.6572585624915264e-05,
      "loss": 0.584,
      "step": 8516
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5900593996047974,
      "learning_rate": 1.657174406100417e-05,
      "loss": 0.7058,
      "step": 8517
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6547248363494873,
      "learning_rate": 1.6570902415160676e-05,
      "loss": 0.7922,
      "step": 8518
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3991522789001465,
      "learning_rate": 1.6570060687395277e-05,
      "loss": 0.5588,
      "step": 8519
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5188987255096436,
      "learning_rate": 1.656921887771846e-05,
      "loss": 0.7545,
      "step": 8520
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.537998914718628,
      "learning_rate": 1.6568376986140728e-05,
      "loss": 0.7094,
      "step": 8521
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5834673643112183,
      "learning_rate": 1.6567535012672578e-05,
      "loss": 0.7524,
      "step": 8522
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3151061534881592,
      "learning_rate": 1.65666929573245e-05,
      "loss": 0.7117,
      "step": 8523
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.705989956855774,
      "learning_rate": 1.6565850820106996e-05,
      "loss": 0.6925,
      "step": 8524
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.615989089012146,
      "learning_rate": 1.6565008601030568e-05,
      "loss": 0.754,
      "step": 8525
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4476873874664307,
      "learning_rate": 1.656416630010571e-05,
      "loss": 0.5617,
      "step": 8526
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3319237232208252,
      "learning_rate": 1.656332391734293e-05,
      "loss": 0.6271,
      "step": 8527
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3060823678970337,
      "learning_rate": 1.6562481452752725e-05,
      "loss": 0.7708,
      "step": 8528
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2818379402160645,
      "learning_rate": 1.6561638906345604e-05,
      "loss": 0.775,
      "step": 8529
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4258320331573486,
      "learning_rate": 1.6560796278132065e-05,
      "loss": 0.5742,
      "step": 8530
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1672724485397339,
      "learning_rate": 1.6559953568122618e-05,
      "loss": 0.6176,
      "step": 8531
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4368544816970825,
      "learning_rate": 1.6559110776327766e-05,
      "loss": 0.7484,
      "step": 8532
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4199274778366089,
      "learning_rate": 1.6558267902758022e-05,
      "loss": 0.7601,
      "step": 8533
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4245822429656982,
      "learning_rate": 1.6557424947423886e-05,
      "loss": 0.7467,
      "step": 8534
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4907985925674438,
      "learning_rate": 1.655658191033587e-05,
      "loss": 0.7458,
      "step": 8535
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3294662237167358,
      "learning_rate": 1.6555738791504495e-05,
      "loss": 0.7816,
      "step": 8536
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4367488622665405,
      "learning_rate": 1.6554895590940257e-05,
      "loss": 0.5445,
      "step": 8537
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4120255708694458,
      "learning_rate": 1.6554052308653676e-05,
      "loss": 0.7939,
      "step": 8538
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0733989477157593,
      "learning_rate": 1.6553208944655268e-05,
      "loss": 0.6966,
      "step": 8539
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6410177946090698,
      "learning_rate": 1.655236549895554e-05,
      "loss": 0.638,
      "step": 8540
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.203647255897522,
      "learning_rate": 1.6551521971565015e-05,
      "loss": 0.7629,
      "step": 8541
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2527838945388794,
      "learning_rate": 1.6550678362494208e-05,
      "loss": 0.6677,
      "step": 8542
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4543137550354004,
      "learning_rate": 1.654983467175363e-05,
      "loss": 0.7095,
      "step": 8543
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2627393007278442,
      "learning_rate": 1.654899089935381e-05,
      "loss": 0.583,
      "step": 8544
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.232556700706482,
      "learning_rate": 1.654814704530526e-05,
      "loss": 0.65,
      "step": 8545
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4251935482025146,
      "learning_rate": 1.6547303109618503e-05,
      "loss": 0.6993,
      "step": 8546
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4282475709915161,
      "learning_rate": 1.6546459092304063e-05,
      "loss": 0.5859,
      "step": 8547
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4583446979522705,
      "learning_rate": 1.6545614993372456e-05,
      "loss": 0.6107,
      "step": 8548
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4911800622940063,
      "learning_rate": 1.6544770812834215e-05,
      "loss": 0.8178,
      "step": 8549
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5136109590530396,
      "learning_rate": 1.6543926550699857e-05,
      "loss": 0.765,
      "step": 8550
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.50856351852417,
      "learning_rate": 1.654308220697991e-05,
      "loss": 0.7988,
      "step": 8551
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3936588764190674,
      "learning_rate": 1.65422377816849e-05,
      "loss": 0.8314,
      "step": 8552
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4663336277008057,
      "learning_rate": 1.654139327482536e-05,
      "loss": 0.6675,
      "step": 8553
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2349114418029785,
      "learning_rate": 1.6540548686411814e-05,
      "loss": 0.6837,
      "step": 8554
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3595213890075684,
      "learning_rate": 1.653970401645479e-05,
      "loss": 0.7938,
      "step": 8555
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.34896981716156,
      "learning_rate": 1.653885926496482e-05,
      "loss": 0.7195,
      "step": 8556
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5126672983169556,
      "learning_rate": 1.6538014431952442e-05,
      "loss": 0.6042,
      "step": 8557
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9321869611740112,
      "learning_rate": 1.6537169517428182e-05,
      "loss": 0.6472,
      "step": 8558
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3312164545059204,
      "learning_rate": 1.6536324521402578e-05,
      "loss": 0.7805,
      "step": 8559
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4515131711959839,
      "learning_rate": 1.653547944388616e-05,
      "loss": 0.7734,
      "step": 8560
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5493842363357544,
      "learning_rate": 1.653463428488947e-05,
      "loss": 0.6093,
      "step": 8561
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4300603866577148,
      "learning_rate": 1.653378904442304e-05,
      "loss": 0.6813,
      "step": 8562
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5829322338104248,
      "learning_rate": 1.6532943722497406e-05,
      "loss": 0.6404,
      "step": 8563
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.480019211769104,
      "learning_rate": 1.6532098319123117e-05,
      "loss": 0.5938,
      "step": 8564
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3532713651657104,
      "learning_rate": 1.65312528343107e-05,
      "loss": 0.8706,
      "step": 8565
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4340720176696777,
      "learning_rate": 1.653040726807071e-05,
      "loss": 0.7747,
      "step": 8566
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6691844463348389,
      "learning_rate": 1.6529561620413675e-05,
      "loss": 0.7496,
      "step": 8567
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5921911001205444,
      "learning_rate": 1.652871589135015e-05,
      "loss": 0.8502,
      "step": 8568
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1778912544250488,
      "learning_rate": 1.6527870080890667e-05,
      "loss": 0.6412,
      "step": 8569
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.474295973777771,
      "learning_rate": 1.6527024189045782e-05,
      "loss": 0.6172,
      "step": 8570
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6917780637741089,
      "learning_rate": 1.6526178215826036e-05,
      "loss": 0.8184,
      "step": 8571
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2810825109481812,
      "learning_rate": 1.6525332161241975e-05,
      "loss": 0.6769,
      "step": 8572
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4324523210525513,
      "learning_rate": 1.652448602530415e-05,
      "loss": 0.7098,
      "step": 8573
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3464516401290894,
      "learning_rate": 1.6523639808023108e-05,
      "loss": 0.9159,
      "step": 8574
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3340961933135986,
      "learning_rate": 1.65227935094094e-05,
      "loss": 0.589,
      "step": 8575
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3210502862930298,
      "learning_rate": 1.6521947129473577e-05,
      "loss": 0.6048,
      "step": 8576
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7683323621749878,
      "learning_rate": 1.6521100668226196e-05,
      "loss": 0.697,
      "step": 8577
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3798490762710571,
      "learning_rate": 1.6520254125677798e-05,
      "loss": 0.7362,
      "step": 8578
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.056460380554199,
      "learning_rate": 1.651940750183895e-05,
      "loss": 0.7727,
      "step": 8579
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5060211420059204,
      "learning_rate": 1.65185607967202e-05,
      "loss": 0.7002,
      "step": 8580
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6706011295318604,
      "learning_rate": 1.6517714010332106e-05,
      "loss": 0.583,
      "step": 8581
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.683608055114746,
      "learning_rate": 1.6516867142685225e-05,
      "loss": 0.679,
      "step": 8582
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.558708667755127,
      "learning_rate": 1.6516020193790114e-05,
      "loss": 0.7719,
      "step": 8583
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.668958067893982,
      "learning_rate": 1.6515173163657333e-05,
      "loss": 0.7101,
      "step": 8584
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2953362464904785,
      "learning_rate": 1.6514326052297444e-05,
      "loss": 0.4988,
      "step": 8585
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5396313667297363,
      "learning_rate": 1.651347885972101e-05,
      "loss": 0.7887,
      "step": 8586
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.485666036605835,
      "learning_rate": 1.6512631585938585e-05,
      "loss": 0.7563,
      "step": 8587
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2603662014007568,
      "learning_rate": 1.651178423096074e-05,
      "loss": 0.6744,
      "step": 8588
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2363978624343872,
      "learning_rate": 1.6510936794798038e-05,
      "loss": 0.6393,
      "step": 8589
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4897243976593018,
      "learning_rate": 1.651008927746104e-05,
      "loss": 0.8682,
      "step": 8590
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3303868770599365,
      "learning_rate": 1.6509241678960323e-05,
      "loss": 0.5937,
      "step": 8591
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5718297958374023,
      "learning_rate": 1.650839399930644e-05,
      "loss": 0.6262,
      "step": 8592
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3695576190948486,
      "learning_rate": 1.650754623850997e-05,
      "loss": 0.661,
      "step": 8593
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5042240619659424,
      "learning_rate": 1.6506698396581473e-05,
      "loss": 0.6987,
      "step": 8594
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5462766885757446,
      "learning_rate": 1.6505850473531527e-05,
      "loss": 0.7657,
      "step": 8595
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1760573387145996,
      "learning_rate": 1.6505002469370705e-05,
      "loss": 0.726,
      "step": 8596
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3191876411437988,
      "learning_rate": 1.650415438410957e-05,
      "loss": 0.658,
      "step": 8597
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5077334642410278,
      "learning_rate": 1.6503306217758708e-05,
      "loss": 0.7269,
      "step": 8598
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9770444631576538,
      "learning_rate": 1.650245797032868e-05,
      "loss": 0.7324,
      "step": 8599
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.55971097946167,
      "learning_rate": 1.650160964183007e-05,
      "loss": 0.7234,
      "step": 8600
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5703312158584595,
      "learning_rate": 1.6500761232273452e-05,
      "loss": 0.7737,
      "step": 8601
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8712351322174072,
      "learning_rate": 1.6499912741669405e-05,
      "loss": 0.5441,
      "step": 8602
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2996540069580078,
      "learning_rate": 1.6499064170028506e-05,
      "loss": 0.6801,
      "step": 8603
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6203175783157349,
      "learning_rate": 1.6498215517361336e-05,
      "loss": 0.6434,
      "step": 8604
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4677163362503052,
      "learning_rate": 1.6497366783678474e-05,
      "loss": 0.5544,
      "step": 8605
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3773773908615112,
      "learning_rate": 1.64965179689905e-05,
      "loss": 0.6785,
      "step": 8606
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6040940284729004,
      "learning_rate": 1.6495669073307995e-05,
      "loss": 0.5754,
      "step": 8607
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.948843240737915,
      "learning_rate": 1.6494820096641548e-05,
      "loss": 0.7714,
      "step": 8608
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3471566438674927,
      "learning_rate": 1.6493971039001743e-05,
      "loss": 0.7012,
      "step": 8609
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3945825099945068,
      "learning_rate": 1.6493121900399164e-05,
      "loss": 0.9067,
      "step": 8610
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5913276672363281,
      "learning_rate": 1.6492272680844393e-05,
      "loss": 0.7983,
      "step": 8611
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7719389200210571,
      "learning_rate": 1.6491423380348025e-05,
      "loss": 0.8102,
      "step": 8612
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4749248027801514,
      "learning_rate": 1.6490573998920646e-05,
      "loss": 0.6549,
      "step": 8613
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.360659122467041,
      "learning_rate": 1.6489724536572846e-05,
      "loss": 0.761,
      "step": 8614
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3770318031311035,
      "learning_rate": 1.6488874993315208e-05,
      "loss": 0.7212,
      "step": 8615
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6681370735168457,
      "learning_rate": 1.6488025369158335e-05,
      "loss": 0.7561,
      "step": 8616
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5886482000350952,
      "learning_rate": 1.6487175664112816e-05,
      "loss": 0.7093,
      "step": 8617
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.762343168258667,
      "learning_rate": 1.6486325878189238e-05,
      "loss": 0.7762,
      "step": 8618
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5540114641189575,
      "learning_rate": 1.6485476011398205e-05,
      "loss": 0.734,
      "step": 8619
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2720000743865967,
      "learning_rate": 1.6484626063750308e-05,
      "loss": 0.7582,
      "step": 8620
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7180503606796265,
      "learning_rate": 1.648377603525614e-05,
      "loss": 0.7065,
      "step": 8621
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5296956300735474,
      "learning_rate": 1.6482925925926307e-05,
      "loss": 0.6489,
      "step": 8622
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4189432859420776,
      "learning_rate": 1.64820757357714e-05,
      "loss": 0.6858,
      "step": 8623
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.560888409614563,
      "learning_rate": 1.6481225464802024e-05,
      "loss": 0.6242,
      "step": 8624
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.599793553352356,
      "learning_rate": 1.6480375113028775e-05,
      "loss": 0.7425,
      "step": 8625
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5516120195388794,
      "learning_rate": 1.647952468046226e-05,
      "loss": 0.6539,
      "step": 8626
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.445142149925232,
      "learning_rate": 1.6478674167113076e-05,
      "loss": 0.786,
      "step": 8627
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6850032806396484,
      "learning_rate": 1.6477823572991832e-05,
      "loss": 0.7027,
      "step": 8628
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5174421072006226,
      "learning_rate": 1.6476972898109127e-05,
      "loss": 0.7055,
      "step": 8629
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2903999090194702,
      "learning_rate": 1.6476122142475573e-05,
      "loss": 0.7831,
      "step": 8630
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3359564542770386,
      "learning_rate": 1.647527130610177e-05,
      "loss": 0.6923,
      "step": 8631
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7083395719528198,
      "learning_rate": 1.6474420388998337e-05,
      "loss": 0.6162,
      "step": 8632
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7207945585250854,
      "learning_rate": 1.6473569391175867e-05,
      "loss": 0.7542,
      "step": 8633
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4974843263626099,
      "learning_rate": 1.6472718312644982e-05,
      "loss": 0.7757,
      "step": 8634
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4517321586608887,
      "learning_rate": 1.6471867153416285e-05,
      "loss": 0.7465,
      "step": 8635
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5376875400543213,
      "learning_rate": 1.6471015913500395e-05,
      "loss": 0.7518,
      "step": 8636
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.785265564918518,
      "learning_rate": 1.6470164592907917e-05,
      "loss": 0.7066,
      "step": 8637
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.63120698928833,
      "learning_rate": 1.646931319164947e-05,
      "loss": 0.5907,
      "step": 8638
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5255831480026245,
      "learning_rate": 1.6468461709735666e-05,
      "loss": 0.7864,
      "step": 8639
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4973958730697632,
      "learning_rate": 1.6467610147177123e-05,
      "loss": 0.5933,
      "step": 8640
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6494437456130981,
      "learning_rate": 1.646675850398446e-05,
      "loss": 0.5739,
      "step": 8641
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.850874900817871,
      "learning_rate": 1.646590678016829e-05,
      "loss": 0.7127,
      "step": 8642
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.540158748626709,
      "learning_rate": 1.646505497573923e-05,
      "loss": 0.6449,
      "step": 8643
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.760864496231079,
      "learning_rate": 1.6464203090707903e-05,
      "loss": 0.6005,
      "step": 8644
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5045329332351685,
      "learning_rate": 1.6463351125084935e-05,
      "loss": 0.6921,
      "step": 8645
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4093512296676636,
      "learning_rate": 1.6462499078880937e-05,
      "loss": 0.7956,
      "step": 8646
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.468631386756897,
      "learning_rate": 1.6461646952106538e-05,
      "loss": 0.6388,
      "step": 8647
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4453927278518677,
      "learning_rate": 1.6460794744772367e-05,
      "loss": 0.7256,
      "step": 8648
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6887894868850708,
      "learning_rate": 1.6459942456889036e-05,
      "loss": 0.745,
      "step": 8649
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5196566581726074,
      "learning_rate": 1.6459090088467183e-05,
      "loss": 0.7208,
      "step": 8650
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4062429666519165,
      "learning_rate": 1.6458237639517426e-05,
      "loss": 0.6815,
      "step": 8651
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4371668100357056,
      "learning_rate": 1.6457385110050398e-05,
      "loss": 0.8578,
      "step": 8652
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7393358945846558,
      "learning_rate": 1.6456532500076724e-05,
      "loss": 0.6841,
      "step": 8653
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.424393892288208,
      "learning_rate": 1.6455679809607035e-05,
      "loss": 0.5549,
      "step": 8654
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.374189853668213,
      "learning_rate": 1.6454827038651968e-05,
      "loss": 0.7298,
      "step": 8655
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6840617656707764,
      "learning_rate": 1.645397418722215e-05,
      "loss": 0.6738,
      "step": 8656
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3124409914016724,
      "learning_rate": 1.645312125532821e-05,
      "loss": 0.7176,
      "step": 8657
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5324798822402954,
      "learning_rate": 1.6452268242980783e-05,
      "loss": 0.6282,
      "step": 8658
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8982281684875488,
      "learning_rate": 1.645141515019051e-05,
      "loss": 0.7319,
      "step": 8659
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4064433574676514,
      "learning_rate": 1.645056197696802e-05,
      "loss": 0.6075,
      "step": 8660
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7955199480056763,
      "learning_rate": 1.6449708723323953e-05,
      "loss": 0.7765,
      "step": 8661
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3449609279632568,
      "learning_rate": 1.644885538926895e-05,
      "loss": 0.7658,
      "step": 8662
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6052497625350952,
      "learning_rate": 1.6448001974813643e-05,
      "loss": 0.5292,
      "step": 8663
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7259833812713623,
      "learning_rate": 1.6447148479968677e-05,
      "loss": 0.7977,
      "step": 8664
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3648697137832642,
      "learning_rate": 1.644629490474469e-05,
      "loss": 0.7815,
      "step": 8665
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4424351453781128,
      "learning_rate": 1.6445441249152325e-05,
      "loss": 0.7118,
      "step": 8666
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2332642078399658,
      "learning_rate": 1.6444587513202222e-05,
      "loss": 0.7494,
      "step": 8667
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2444874048233032,
      "learning_rate": 1.6443733696905034e-05,
      "loss": 0.6005,
      "step": 8668
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4482430219650269,
      "learning_rate": 1.6442879800271393e-05,
      "loss": 0.6699,
      "step": 8669
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.398532509803772,
      "learning_rate": 1.6442025823311957e-05,
      "loss": 0.589,
      "step": 8670
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.342828631401062,
      "learning_rate": 1.644117176603736e-05,
      "loss": 0.6742,
      "step": 8671
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.488576889038086,
      "learning_rate": 1.6440317628458266e-05,
      "loss": 0.6348,
      "step": 8672
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.129526376724243,
      "learning_rate": 1.643946341058531e-05,
      "loss": 0.9636,
      "step": 8673
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1889677047729492,
      "learning_rate": 1.6438609112429146e-05,
      "loss": 0.6596,
      "step": 8674
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7195907831192017,
      "learning_rate": 1.6437754734000427e-05,
      "loss": 0.7043,
      "step": 8675
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8563644886016846,
      "learning_rate": 1.6436900275309803e-05,
      "loss": 0.5787,
      "step": 8676
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2442529201507568,
      "learning_rate": 1.6436045736367926e-05,
      "loss": 0.7182,
      "step": 8677
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.221609354019165,
      "learning_rate": 1.6435191117185453e-05,
      "loss": 0.8902,
      "step": 8678
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5360450744628906,
      "learning_rate": 1.6434336417773036e-05,
      "loss": 0.6687,
      "step": 8679
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3337512016296387,
      "learning_rate": 1.643348163814133e-05,
      "loss": 0.641,
      "step": 8680
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7410876750946045,
      "learning_rate": 1.6432626778300998e-05,
      "loss": 0.6585,
      "step": 8681
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4591718912124634,
      "learning_rate": 1.643177183826269e-05,
      "loss": 0.6614,
      "step": 8682
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3932849168777466,
      "learning_rate": 1.643091681803707e-05,
      "loss": 0.758,
      "step": 8683
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.557604432106018,
      "learning_rate": 1.6430061717634797e-05,
      "loss": 0.735,
      "step": 8684
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6570247411727905,
      "learning_rate": 1.642920653706653e-05,
      "loss": 0.7147,
      "step": 8685
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4134008884429932,
      "learning_rate": 1.6428351276342933e-05,
      "loss": 0.5664,
      "step": 8686
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3407292366027832,
      "learning_rate": 1.6427495935474667e-05,
      "loss": 0.6606,
      "step": 8687
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4309831857681274,
      "learning_rate": 1.6426640514472395e-05,
      "loss": 0.5961,
      "step": 8688
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7172036170959473,
      "learning_rate": 1.6425785013346786e-05,
      "loss": 0.7513,
      "step": 8689
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6084296703338623,
      "learning_rate": 1.6424929432108503e-05,
      "loss": 0.8268,
      "step": 8690
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3509780168533325,
      "learning_rate": 1.6424073770768215e-05,
      "loss": 0.5037,
      "step": 8691
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.651047945022583,
      "learning_rate": 1.6423218029336584e-05,
      "loss": 0.8335,
      "step": 8692
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2522722482681274,
      "learning_rate": 1.6422362207824284e-05,
      "loss": 0.642,
      "step": 8693
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.538345217704773,
      "learning_rate": 1.6421506306241985e-05,
      "loss": 0.8251,
      "step": 8694
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6432726383209229,
      "learning_rate": 1.642065032460036e-05,
      "loss": 0.7726,
      "step": 8695
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7412197589874268,
      "learning_rate": 1.641979426291007e-05,
      "loss": 0.61,
      "step": 8696
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.614406943321228,
      "learning_rate": 1.64189381211818e-05,
      "loss": 0.8904,
      "step": 8697
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.9885495901107788,
      "learning_rate": 1.641808189942622e-05,
      "loss": 0.81,
      "step": 8698
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.330090045928955,
      "learning_rate": 1.6417225597654004e-05,
      "loss": 0.6501,
      "step": 8699
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4033873081207275,
      "learning_rate": 1.6416369215875826e-05,
      "loss": 0.6783,
      "step": 8700
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2233562469482422,
      "learning_rate": 1.6415512754102362e-05,
      "loss": 0.7275,
      "step": 8701
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2745959758758545,
      "learning_rate": 1.6414656212344296e-05,
      "loss": 0.601,
      "step": 8702
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.370718240737915,
      "learning_rate": 1.6413799590612302e-05,
      "loss": 0.6181,
      "step": 8703
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7841376066207886,
      "learning_rate": 1.6412942888917065e-05,
      "loss": 0.8248,
      "step": 8704
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.465226173400879,
      "learning_rate": 1.641208610726926e-05,
      "loss": 0.5911,
      "step": 8705
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7393264770507812,
      "learning_rate": 1.6411229245679568e-05,
      "loss": 0.7388,
      "step": 8706
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.36798894405365,
      "learning_rate": 1.6410372304158677e-05,
      "loss": 0.7858,
      "step": 8707
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.431248664855957,
      "learning_rate": 1.6409515282717267e-05,
      "loss": 0.7667,
      "step": 8708
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3482965230941772,
      "learning_rate": 1.6408658181366027e-05,
      "loss": 0.7252,
      "step": 8709
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6879758834838867,
      "learning_rate": 1.6407801000115635e-05,
      "loss": 0.7491,
      "step": 8710
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6474720239639282,
      "learning_rate": 1.640694373897679e-05,
      "loss": 0.6141,
      "step": 8711
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4263936281204224,
      "learning_rate": 1.6406086397960165e-05,
      "loss": 0.8066,
      "step": 8712
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.553968906402588,
      "learning_rate": 1.6405228977076462e-05,
      "loss": 0.7536,
      "step": 8713
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2458542585372925,
      "learning_rate": 1.6404371476336363e-05,
      "loss": 0.5914,
      "step": 8714
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4859564304351807,
      "learning_rate": 1.640351389575056e-05,
      "loss": 0.5818,
      "step": 8715
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4591950178146362,
      "learning_rate": 1.6402656235329752e-05,
      "loss": 0.7498,
      "step": 8716
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5543349981307983,
      "learning_rate": 1.640179849508462e-05,
      "loss": 0.5208,
      "step": 8717
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8981610536575317,
      "learning_rate": 1.6400940675025862e-05,
      "loss": 0.774,
      "step": 8718
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3963744640350342,
      "learning_rate": 1.6400082775164173e-05,
      "loss": 0.7765,
      "step": 8719
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4149786233901978,
      "learning_rate": 1.6399224795510257e-05,
      "loss": 0.7564,
      "step": 8720
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4803217649459839,
      "learning_rate": 1.63983667360748e-05,
      "loss": 0.6476,
      "step": 8721
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.519640326499939,
      "learning_rate": 1.63975085968685e-05,
      "loss": 0.6728,
      "step": 8722
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4562656879425049,
      "learning_rate": 1.6396650377902064e-05,
      "loss": 0.5083,
      "step": 8723
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8012968301773071,
      "learning_rate": 1.6395792079186184e-05,
      "loss": 0.6787,
      "step": 8724
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5574411153793335,
      "learning_rate": 1.6394933700731565e-05,
      "loss": 0.795,
      "step": 8725
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8375778198242188,
      "learning_rate": 1.6394075242548905e-05,
      "loss": 0.8231,
      "step": 8726
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7188490629196167,
      "learning_rate": 1.6393216704648912e-05,
      "loss": 0.6599,
      "step": 8727
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4685990810394287,
      "learning_rate": 1.6392358087042286e-05,
      "loss": 0.7193,
      "step": 8728
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.478766679763794,
      "learning_rate": 1.639149938973973e-05,
      "loss": 0.6362,
      "step": 8729
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.581413745880127,
      "learning_rate": 1.6390640612751957e-05,
      "loss": 0.787,
      "step": 8730
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5444179773330688,
      "learning_rate": 1.6389781756089667e-05,
      "loss": 0.6828,
      "step": 8731
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.483187198638916,
      "learning_rate": 1.6388922819763567e-05,
      "loss": 0.734,
      "step": 8732
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8106790781021118,
      "learning_rate": 1.6388063803784372e-05,
      "loss": 0.805,
      "step": 8733
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4241394996643066,
      "learning_rate": 1.6387204708162782e-05,
      "loss": 0.7322,
      "step": 8734
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8287495374679565,
      "learning_rate": 1.638634553290952e-05,
      "loss": 0.6847,
      "step": 8735
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7441871166229248,
      "learning_rate": 1.6385486278035288e-05,
      "loss": 0.7161,
      "step": 8736
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5808409452438354,
      "learning_rate": 1.6384626943550803e-05,
      "loss": 0.6594,
      "step": 8737
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5398576259613037,
      "learning_rate": 1.6383767529466777e-05,
      "loss": 0.8092,
      "step": 8738
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6115955114364624,
      "learning_rate": 1.6382908035793926e-05,
      "loss": 0.8257,
      "step": 8739
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6554535627365112,
      "learning_rate": 1.6382048462542962e-05,
      "loss": 0.7509,
      "step": 8740
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5880624055862427,
      "learning_rate": 1.6381188809724604e-05,
      "loss": 0.6514,
      "step": 8741
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4709832668304443,
      "learning_rate": 1.6380329077349574e-05,
      "loss": 0.8333,
      "step": 8742
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4671779870986938,
      "learning_rate": 1.6379469265428585e-05,
      "loss": 0.7058,
      "step": 8743
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.673445463180542,
      "learning_rate": 1.637860937397236e-05,
      "loss": 0.6477,
      "step": 8744
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5136899948120117,
      "learning_rate": 1.6377749402991613e-05,
      "loss": 0.732,
      "step": 8745
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7062004804611206,
      "learning_rate": 1.6376889352497073e-05,
      "loss": 0.7313,
      "step": 8746
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6460086107254028,
      "learning_rate": 1.637602922249946e-05,
      "loss": 0.7651,
      "step": 8747
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4100191593170166,
      "learning_rate": 1.6375169013009498e-05,
      "loss": 0.618,
      "step": 8748
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.676900863647461,
      "learning_rate": 1.637430872403791e-05,
      "loss": 0.638,
      "step": 8749
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3146460056304932,
      "learning_rate": 1.6373448355595422e-05,
      "loss": 0.7345,
      "step": 8750
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.548227310180664,
      "learning_rate": 1.6372587907692765e-05,
      "loss": 0.6955,
      "step": 8751
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4967050552368164,
      "learning_rate": 1.637172738034066e-05,
      "loss": 0.6433,
      "step": 8752
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5120034217834473,
      "learning_rate": 1.6370866773549837e-05,
      "loss": 0.5152,
      "step": 8753
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4331912994384766,
      "learning_rate": 1.637000608733103e-05,
      "loss": 0.7139,
      "step": 8754
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6510074138641357,
      "learning_rate": 1.6369145321694967e-05,
      "loss": 0.7202,
      "step": 8755
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5409246683120728,
      "learning_rate": 1.6368284476652375e-05,
      "loss": 0.6922,
      "step": 8756
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3149021863937378,
      "learning_rate": 1.636742355221399e-05,
      "loss": 0.7381,
      "step": 8757
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.41698157787323,
      "learning_rate": 1.636656254839055e-05,
      "loss": 0.539,
      "step": 8758
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4292593002319336,
      "learning_rate": 1.636570146519278e-05,
      "loss": 0.7159,
      "step": 8759
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5763686895370483,
      "learning_rate": 1.636484030263142e-05,
      "loss": 0.6948,
      "step": 8760
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8046133518218994,
      "learning_rate": 1.636397906071721e-05,
      "loss": 0.8931,
      "step": 8761
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7750532627105713,
      "learning_rate": 1.6363117739460883e-05,
      "loss": 0.7544,
      "step": 8762
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.600785493850708,
      "learning_rate": 1.636225633887318e-05,
      "loss": 0.7655,
      "step": 8763
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6935691833496094,
      "learning_rate": 1.6361394858964842e-05,
      "loss": 0.5535,
      "step": 8764
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3304578065872192,
      "learning_rate": 1.63605332997466e-05,
      "loss": 0.4967,
      "step": 8765
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5169225931167603,
      "learning_rate": 1.635967166122921e-05,
      "loss": 0.7092,
      "step": 8766
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5076817274093628,
      "learning_rate": 1.63588099434234e-05,
      "loss": 0.8097,
      "step": 8767
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6896957159042358,
      "learning_rate": 1.635794814633992e-05,
      "loss": 0.7519,
      "step": 8768
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6387072801589966,
      "learning_rate": 1.6357086269989517e-05,
      "loss": 0.6893,
      "step": 8769
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.383470892906189,
      "learning_rate": 1.6356224314382933e-05,
      "loss": 0.6136,
      "step": 8770
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6473990678787231,
      "learning_rate": 1.6355362279530914e-05,
      "loss": 0.7231,
      "step": 8771
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6457639932632446,
      "learning_rate": 1.6354500165444207e-05,
      "loss": 0.7519,
      "step": 8772
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7414592504501343,
      "learning_rate": 1.6353637972133564e-05,
      "loss": 0.7533,
      "step": 8773
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9427241086959839,
      "learning_rate": 1.635277569960973e-05,
      "loss": 0.5936,
      "step": 8774
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6634799242019653,
      "learning_rate": 1.6351913347883456e-05,
      "loss": 0.6549,
      "step": 8775
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.607850193977356,
      "learning_rate": 1.6351050916965495e-05,
      "loss": 0.7652,
      "step": 8776
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7494903802871704,
      "learning_rate": 1.63501884068666e-05,
      "loss": 0.9315,
      "step": 8777
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.607477903366089,
      "learning_rate": 1.634932581759752e-05,
      "loss": 0.6217,
      "step": 8778
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5882563591003418,
      "learning_rate": 1.6348463149169017e-05,
      "loss": 0.7907,
      "step": 8779
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3556221723556519,
      "learning_rate": 1.6347600401591838e-05,
      "loss": 0.7551,
      "step": 8780
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3324726819992065,
      "learning_rate": 1.634673757487674e-05,
      "loss": 0.7276,
      "step": 8781
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5058083534240723,
      "learning_rate": 1.6345874669034484e-05,
      "loss": 0.6683,
      "step": 8782
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.441776156425476,
      "learning_rate": 1.6345011684075826e-05,
      "loss": 0.6167,
      "step": 8783
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3421814441680908,
      "learning_rate": 1.634414862001153e-05,
      "loss": 0.7789,
      "step": 8784
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4765945672988892,
      "learning_rate": 1.6343285476852345e-05,
      "loss": 0.7056,
      "step": 8785
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8312512636184692,
      "learning_rate": 1.6342422254609042e-05,
      "loss": 0.7033,
      "step": 8786
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2464430332183838,
      "learning_rate": 1.634155895329238e-05,
      "loss": 0.8065,
      "step": 8787
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4127788543701172,
      "learning_rate": 1.6340695572913127e-05,
      "loss": 0.4958,
      "step": 8788
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5593456029891968,
      "learning_rate": 1.633983211348204e-05,
      "loss": 0.6308,
      "step": 8789
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2176988124847412,
      "learning_rate": 1.633896857500988e-05,
      "loss": 0.6696,
      "step": 8790
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4332947731018066,
      "learning_rate": 1.633810495750743e-05,
      "loss": 0.6701,
      "step": 8791
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3830782175064087,
      "learning_rate": 1.6337241260985436e-05,
      "loss": 0.7144,
      "step": 8792
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.62802255153656,
      "learning_rate": 1.6336377485454683e-05,
      "loss": 0.826,
      "step": 8793
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.178347110748291,
      "learning_rate": 1.6335513630925933e-05,
      "loss": 0.6866,
      "step": 8794
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6582199335098267,
      "learning_rate": 1.6334649697409956e-05,
      "loss": 0.673,
      "step": 8795
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2657034397125244,
      "learning_rate": 1.6333785684917524e-05,
      "loss": 0.7376,
      "step": 8796
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.496110200881958,
      "learning_rate": 1.633292159345941e-05,
      "loss": 0.7083,
      "step": 8797
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1770093441009521,
      "learning_rate": 1.6332057423046382e-05,
      "loss": 0.6129,
      "step": 8798
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6288390159606934,
      "learning_rate": 1.633119317368922e-05,
      "loss": 0.6761,
      "step": 8799
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6064308881759644,
      "learning_rate": 1.6330328845398694e-05,
      "loss": 0.7234,
      "step": 8800
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7005021572113037,
      "learning_rate": 1.6329464438185585e-05,
      "loss": 0.8286,
      "step": 8801
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4562243223190308,
      "learning_rate": 1.6328599952060668e-05,
      "loss": 0.8305,
      "step": 8802
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.604341983795166,
      "learning_rate": 1.6327735387034716e-05,
      "loss": 0.741,
      "step": 8803
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3063416481018066,
      "learning_rate": 1.6326870743118514e-05,
      "loss": 0.7131,
      "step": 8804
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8719897270202637,
      "learning_rate": 1.632600602032284e-05,
      "loss": 0.7521,
      "step": 8805
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6141619682312012,
      "learning_rate": 1.6325141218658475e-05,
      "loss": 0.7428,
      "step": 8806
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5792407989501953,
      "learning_rate": 1.63242763381362e-05,
      "loss": 0.6469,
      "step": 8807
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.807203769683838,
      "learning_rate": 1.63234113787668e-05,
      "loss": 0.8065,
      "step": 8808
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3932387828826904,
      "learning_rate": 1.6322546340561053e-05,
      "loss": 0.8174,
      "step": 8809
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3677387237548828,
      "learning_rate": 1.632168122352975e-05,
      "loss": 0.6485,
      "step": 8810
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8709712028503418,
      "learning_rate": 1.632081602768367e-05,
      "loss": 0.7277,
      "step": 8811
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2551506757736206,
      "learning_rate": 1.631995075303361e-05,
      "loss": 0.7252,
      "step": 8812
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7702502012252808,
      "learning_rate": 1.631908539959035e-05,
      "loss": 0.7654,
      "step": 8813
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4637596607208252,
      "learning_rate": 1.631821996736468e-05,
      "loss": 0.6698,
      "step": 8814
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9259321689605713,
      "learning_rate": 1.631735445636739e-05,
      "loss": 0.7687,
      "step": 8815
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.533738374710083,
      "learning_rate": 1.6316488866609274e-05,
      "loss": 0.7359,
      "step": 8816
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1622854471206665,
      "learning_rate": 1.6315623198101114e-05,
      "loss": 0.6154,
      "step": 8817
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4080498218536377,
      "learning_rate": 1.6314757450853714e-05,
      "loss": 0.7308,
      "step": 8818
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2244503498077393,
      "learning_rate": 1.6313891624877864e-05,
      "loss": 0.779,
      "step": 8819
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5104622840881348,
      "learning_rate": 1.6313025720184355e-05,
      "loss": 0.6163,
      "step": 8820
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5441138744354248,
      "learning_rate": 1.6312159736783987e-05,
      "loss": 0.553,
      "step": 8821
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4953696727752686,
      "learning_rate": 1.631129367468755e-05,
      "loss": 0.6742,
      "step": 8822
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4737404584884644,
      "learning_rate": 1.631042753390585e-05,
      "loss": 0.8331,
      "step": 8823
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.392256498336792,
      "learning_rate": 1.630956131444968e-05,
      "loss": 0.7465,
      "step": 8824
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5432031154632568,
      "learning_rate": 1.6308695016329838e-05,
      "loss": 0.6475,
      "step": 8825
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.324101448059082,
      "learning_rate": 1.630782863955713e-05,
      "loss": 0.73,
      "step": 8826
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1432541608810425,
      "learning_rate": 1.6306962184142355e-05,
      "loss": 0.6408,
      "step": 8827
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4881006479263306,
      "learning_rate": 1.6306095650096312e-05,
      "loss": 0.7093,
      "step": 8828
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.826633095741272,
      "learning_rate": 1.6305229037429814e-05,
      "loss": 0.8189,
      "step": 8829
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.290669322013855,
      "learning_rate": 1.6304362346153657e-05,
      "loss": 0.6384,
      "step": 8830
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5610779523849487,
      "learning_rate": 1.6303495576278642e-05,
      "loss": 0.7761,
      "step": 8831
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5574406385421753,
      "learning_rate": 1.630262872781559e-05,
      "loss": 0.6769,
      "step": 8832
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4095349311828613,
      "learning_rate": 1.6301761800775296e-05,
      "loss": 0.6976,
      "step": 8833
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5586637258529663,
      "learning_rate": 1.6300894795168573e-05,
      "loss": 0.6641,
      "step": 8834
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5944660902023315,
      "learning_rate": 1.630002771100623e-05,
      "loss": 0.5567,
      "step": 8835
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6529686450958252,
      "learning_rate": 1.6299160548299076e-05,
      "loss": 0.84,
      "step": 8836
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4840638637542725,
      "learning_rate": 1.6298293307057924e-05,
      "loss": 0.6582,
      "step": 8837
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6386001110076904,
      "learning_rate": 1.6297425987293586e-05,
      "loss": 0.7088,
      "step": 8838
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.326830506324768,
      "learning_rate": 1.6296558589016873e-05,
      "loss": 0.6425,
      "step": 8839
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5896366834640503,
      "learning_rate": 1.62956911122386e-05,
      "loss": 0.7385,
      "step": 8840
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5936022996902466,
      "learning_rate": 1.6294823556969586e-05,
      "loss": 0.6827,
      "step": 8841
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4171404838562012,
      "learning_rate": 1.629395592322064e-05,
      "loss": 0.6656,
      "step": 8842
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4524849653244019,
      "learning_rate": 1.6293088211002587e-05,
      "loss": 0.8093,
      "step": 8843
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4752366542816162,
      "learning_rate": 1.629222042032624e-05,
      "loss": 0.662,
      "step": 8844
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2440283298492432,
      "learning_rate": 1.6291352551202418e-05,
      "loss": 0.6173,
      "step": 8845
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9223620891571045,
      "learning_rate": 1.6290484603641945e-05,
      "loss": 0.6655,
      "step": 8846
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5685192346572876,
      "learning_rate": 1.6289616577655638e-05,
      "loss": 0.7705,
      "step": 8847
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5372477769851685,
      "learning_rate": 1.628874847325432e-05,
      "loss": 0.8001,
      "step": 8848
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3145298957824707,
      "learning_rate": 1.6287880290448813e-05,
      "loss": 0.7294,
      "step": 8849
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3696357011795044,
      "learning_rate": 1.6287012029249943e-05,
      "loss": 0.5313,
      "step": 8850
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3686225414276123,
      "learning_rate": 1.6286143689668538e-05,
      "loss": 0.632,
      "step": 8851
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3758785724639893,
      "learning_rate": 1.6285275271715416e-05,
      "loss": 0.6921,
      "step": 8852
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.550468921661377,
      "learning_rate": 1.628440677540141e-05,
      "loss": 0.7469,
      "step": 8853
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6348156929016113,
      "learning_rate": 1.6283538200737347e-05,
      "loss": 0.8547,
      "step": 8854
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4086498022079468,
      "learning_rate": 1.628266954773405e-05,
      "loss": 0.7337,
      "step": 8855
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5706453323364258,
      "learning_rate": 1.628180081640236e-05,
      "loss": 0.7479,
      "step": 8856
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2706124782562256,
      "learning_rate": 1.6280932006753095e-05,
      "loss": 0.559,
      "step": 8857
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5552787780761719,
      "learning_rate": 1.6280063118797098e-05,
      "loss": 0.777,
      "step": 8858
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.425089716911316,
      "learning_rate": 1.6279194152545194e-05,
      "loss": 0.6171,
      "step": 8859
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5995092391967773,
      "learning_rate": 1.6278325108008222e-05,
      "loss": 0.668,
      "step": 8860
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8525811433792114,
      "learning_rate": 1.6277455985197016e-05,
      "loss": 0.7915,
      "step": 8861
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.38750422000885,
      "learning_rate": 1.6276586784122405e-05,
      "loss": 0.683,
      "step": 8862
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.249528408050537,
      "learning_rate": 1.627571750479523e-05,
      "loss": 0.5597,
      "step": 8863
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.487274169921875,
      "learning_rate": 1.6274848147226337e-05,
      "loss": 0.6556,
      "step": 8864
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2987228631973267,
      "learning_rate": 1.6273978711426554e-05,
      "loss": 0.7207,
      "step": 8865
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.922676682472229,
      "learning_rate": 1.627310919740672e-05,
      "loss": 0.8207,
      "step": 8866
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6526583433151245,
      "learning_rate": 1.627223960517768e-05,
      "loss": 0.7984,
      "step": 8867
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6685924530029297,
      "learning_rate": 1.627136993475028e-05,
      "loss": 0.6497,
      "step": 8868
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7199562788009644,
      "learning_rate": 1.627050018613535e-05,
      "loss": 0.8213,
      "step": 8869
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6283228397369385,
      "learning_rate": 1.6269630359343746e-05,
      "loss": 0.7566,
      "step": 8870
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3861992359161377,
      "learning_rate": 1.6268760454386305e-05,
      "loss": 0.7132,
      "step": 8871
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.497977375984192,
      "learning_rate": 1.626789047127387e-05,
      "loss": 0.6734,
      "step": 8872
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3334684371948242,
      "learning_rate": 1.62670204100173e-05,
      "loss": 0.6619,
      "step": 8873
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8174771070480347,
      "learning_rate": 1.626615027062743e-05,
      "loss": 0.769,
      "step": 8874
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4496092796325684,
      "learning_rate": 1.6265280053115112e-05,
      "loss": 0.7531,
      "step": 8875
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5511661767959595,
      "learning_rate": 1.6264409757491198e-05,
      "loss": 0.6485,
      "step": 8876
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1518739461898804,
      "learning_rate": 1.6263539383766536e-05,
      "loss": 0.6079,
      "step": 8877
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.414975881576538,
      "learning_rate": 1.6262668931951976e-05,
      "loss": 0.8192,
      "step": 8878
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3790359497070312,
      "learning_rate": 1.6261798402058376e-05,
      "loss": 0.7688,
      "step": 8879
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4052069187164307,
      "learning_rate": 1.6260927794096582e-05,
      "loss": 0.682,
      "step": 8880
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5998848676681519,
      "learning_rate": 1.6260057108077453e-05,
      "loss": 0.7765,
      "step": 8881
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2502961158752441,
      "learning_rate": 1.6259186344011837e-05,
      "loss": 0.5979,
      "step": 8882
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8123408555984497,
      "learning_rate": 1.6258315501910604e-05,
      "loss": 0.7496,
      "step": 8883
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3710726499557495,
      "learning_rate": 1.6257444581784595e-05,
      "loss": 0.5612,
      "step": 8884
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4693212509155273,
      "learning_rate": 1.625657358364468e-05,
      "loss": 0.6506,
      "step": 8885
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5681049823760986,
      "learning_rate": 1.625570250750171e-05,
      "loss": 0.5757,
      "step": 8886
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7667534351348877,
      "learning_rate": 1.6254831353366554e-05,
      "loss": 0.9152,
      "step": 8887
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.409165620803833,
      "learning_rate": 1.6253960121250064e-05,
      "loss": 0.7616,
      "step": 8888
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7241861820220947,
      "learning_rate": 1.6253088811163107e-05,
      "loss": 0.7002,
      "step": 8889
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4043984413146973,
      "learning_rate": 1.6252217423116542e-05,
      "loss": 0.6378,
      "step": 8890
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9590206146240234,
      "learning_rate": 1.625134595712124e-05,
      "loss": 0.6104,
      "step": 8891
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.316688895225525,
      "learning_rate": 1.625047441318806e-05,
      "loss": 0.6954,
      "step": 8892
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5785161256790161,
      "learning_rate": 1.624960279132787e-05,
      "loss": 0.7138,
      "step": 8893
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4688801765441895,
      "learning_rate": 1.6248731091551536e-05,
      "loss": 0.6432,
      "step": 8894
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5171501636505127,
      "learning_rate": 1.6247859313869927e-05,
      "loss": 0.8122,
      "step": 8895
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4519773721694946,
      "learning_rate": 1.6246987458293908e-05,
      "loss": 0.7322,
      "step": 8896
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3524925708770752,
      "learning_rate": 1.6246115524834355e-05,
      "loss": 0.7016,
      "step": 8897
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2879198789596558,
      "learning_rate": 1.6245243513502133e-05,
      "loss": 0.6097,
      "step": 8898
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4945322275161743,
      "learning_rate": 1.6244371424308118e-05,
      "loss": 0.5326,
      "step": 8899
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3398997783660889,
      "learning_rate": 1.6243499257263176e-05,
      "loss": 0.7334,
      "step": 8900
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5059045553207397,
      "learning_rate": 1.624262701237819e-05,
      "loss": 0.6316,
      "step": 8901
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.620824933052063,
      "learning_rate": 1.624175468966403e-05,
      "loss": 0.8453,
      "step": 8902
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4849717617034912,
      "learning_rate": 1.624088228913157e-05,
      "loss": 0.77,
      "step": 8903
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7244547605514526,
      "learning_rate": 1.624000981079169e-05,
      "loss": 0.692,
      "step": 8904
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2774776220321655,
      "learning_rate": 1.6239137254655265e-05,
      "loss": 0.7303,
      "step": 8905
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1965595483779907,
      "learning_rate": 1.6238264620733172e-05,
      "loss": 0.7181,
      "step": 8906
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3396364450454712,
      "learning_rate": 1.6237391909036294e-05,
      "loss": 0.7591,
      "step": 8907
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5630040168762207,
      "learning_rate": 1.6236519119575507e-05,
      "loss": 0.4575,
      "step": 8908
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.484828233718872,
      "learning_rate": 1.62356462523617e-05,
      "loss": 0.6229,
      "step": 8909
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3669652938842773,
      "learning_rate": 1.6234773307405748e-05,
      "loss": 0.5074,
      "step": 8910
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5792886018753052,
      "learning_rate": 1.623390028471854e-05,
      "loss": 0.7521,
      "step": 8911
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4321839809417725,
      "learning_rate": 1.6233027184310953e-05,
      "loss": 0.5922,
      "step": 8912
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8344621658325195,
      "learning_rate": 1.623215400619388e-05,
      "loss": 0.7226,
      "step": 8913
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5042530298233032,
      "learning_rate": 1.62312807503782e-05,
      "loss": 0.6501,
      "step": 8914
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5484992265701294,
      "learning_rate": 1.6230407416874806e-05,
      "loss": 0.7235,
      "step": 8915
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5670573711395264,
      "learning_rate": 1.6229534005694586e-05,
      "loss": 0.7464,
      "step": 8916
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.444749355316162,
      "learning_rate": 1.6228660516848425e-05,
      "loss": 0.7567,
      "step": 8917
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1508474349975586,
      "learning_rate": 1.622778695034722e-05,
      "loss": 0.745,
      "step": 8918
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6890716552734375,
      "learning_rate": 1.6226913306201852e-05,
      "loss": 0.6497,
      "step": 8919
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4364949464797974,
      "learning_rate": 1.622603958442322e-05,
      "loss": 0.7875,
      "step": 8920
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6622092723846436,
      "learning_rate": 1.622516578502222e-05,
      "loss": 0.7133,
      "step": 8921
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4281798601150513,
      "learning_rate": 1.6224291908009736e-05,
      "loss": 0.6501,
      "step": 8922
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.610709547996521,
      "learning_rate": 1.622341795339667e-05,
      "loss": 0.697,
      "step": 8923
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.499698519706726,
      "learning_rate": 1.6222543921193918e-05,
      "loss": 0.7618,
      "step": 8924
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3120497465133667,
      "learning_rate": 1.6221669811412372e-05,
      "loss": 0.6796,
      "step": 8925
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6846458911895752,
      "learning_rate": 1.622079562406294e-05,
      "loss": 0.7061,
      "step": 8926
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3250468969345093,
      "learning_rate": 1.6219921359156506e-05,
      "loss": 0.5737,
      "step": 8927
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4116877317428589,
      "learning_rate": 1.6219047016703985e-05,
      "loss": 0.5556,
      "step": 8928
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6073309183120728,
      "learning_rate": 1.6218172596716266e-05,
      "loss": 0.6382,
      "step": 8929
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3327577114105225,
      "learning_rate": 1.6217298099204257e-05,
      "loss": 0.6899,
      "step": 8930
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5212972164154053,
      "learning_rate": 1.6216423524178857e-05,
      "loss": 0.6792,
      "step": 8931
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.858411431312561,
      "learning_rate": 1.6215548871650976e-05,
      "loss": 0.5999,
      "step": 8932
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4053789377212524,
      "learning_rate": 1.621467414163151e-05,
      "loss": 0.5928,
      "step": 8933
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5845623016357422,
      "learning_rate": 1.621379933413137e-05,
      "loss": 0.682,
      "step": 8934
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3797539472579956,
      "learning_rate": 1.621292444916146e-05,
      "loss": 0.6575,
      "step": 8935
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6720033884048462,
      "learning_rate": 1.6212049486732695e-05,
      "loss": 0.6005,
      "step": 8936
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5474482774734497,
      "learning_rate": 1.6211174446855972e-05,
      "loss": 0.6902,
      "step": 8937
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4435713291168213,
      "learning_rate": 1.6210299329542207e-05,
      "loss": 0.6991,
      "step": 8938
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.338144063949585,
      "learning_rate": 1.620942413480231e-05,
      "loss": 0.6081,
      "step": 8939
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3921574354171753,
      "learning_rate": 1.620854886264719e-05,
      "loss": 0.7513,
      "step": 8940
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5631721019744873,
      "learning_rate": 1.6207673513087762e-05,
      "loss": 0.6104,
      "step": 8941
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6672544479370117,
      "learning_rate": 1.620679808613494e-05,
      "loss": 0.7093,
      "step": 8942
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3524266481399536,
      "learning_rate": 1.6205922581799637e-05,
      "loss": 0.5663,
      "step": 8943
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5800970792770386,
      "learning_rate": 1.6205047000092764e-05,
      "loss": 0.7554,
      "step": 8944
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3152118921279907,
      "learning_rate": 1.6204171341025247e-05,
      "loss": 0.5826,
      "step": 8945
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4110040664672852,
      "learning_rate": 1.6203295604607992e-05,
      "loss": 0.6579,
      "step": 8946
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4292552471160889,
      "learning_rate": 1.6202419790851923e-05,
      "loss": 0.7726,
      "step": 8947
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6140812635421753,
      "learning_rate": 1.6201543899767956e-05,
      "loss": 0.6721,
      "step": 8948
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4774373769760132,
      "learning_rate": 1.620066793136702e-05,
      "loss": 0.6414,
      "step": 8949
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4182286262512207,
      "learning_rate": 1.6199791885660027e-05,
      "loss": 0.8092,
      "step": 8950
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3615938425064087,
      "learning_rate": 1.6198915762657898e-05,
      "loss": 0.6781,
      "step": 8951
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7027941942214966,
      "learning_rate": 1.6198039562371564e-05,
      "loss": 0.7361,
      "step": 8952
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4491251707077026,
      "learning_rate": 1.6197163284811942e-05,
      "loss": 0.6397,
      "step": 8953
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2458683252334595,
      "learning_rate": 1.6196286929989963e-05,
      "loss": 0.6936,
      "step": 8954
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3152137994766235,
      "learning_rate": 1.6195410497916542e-05,
      "loss": 0.6236,
      "step": 8955
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5495259761810303,
      "learning_rate": 1.6194533988602618e-05,
      "loss": 0.7288,
      "step": 8956
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4951701164245605,
      "learning_rate": 1.6193657402059116e-05,
      "loss": 0.7451,
      "step": 8957
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2170414924621582,
      "learning_rate": 1.6192780738296957e-05,
      "loss": 0.7536,
      "step": 8958
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3066446781158447,
      "learning_rate": 1.619190399732708e-05,
      "loss": 0.6537,
      "step": 8959
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7056149244308472,
      "learning_rate": 1.6191027179160413e-05,
      "loss": 0.6874,
      "step": 8960
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3074345588684082,
      "learning_rate": 1.6190150283807882e-05,
      "loss": 0.7142,
      "step": 8961
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4301116466522217,
      "learning_rate": 1.6189273311280428e-05,
      "loss": 0.7131,
      "step": 8962
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4368680715560913,
      "learning_rate": 1.6188396261588976e-05,
      "loss": 0.5854,
      "step": 8963
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4734866619110107,
      "learning_rate": 1.618751913474447e-05,
      "loss": 0.7191,
      "step": 8964
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4981937408447266,
      "learning_rate": 1.6186641930757835e-05,
      "loss": 0.7268,
      "step": 8965
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.605661392211914,
      "learning_rate": 1.6185764649640018e-05,
      "loss": 0.6501,
      "step": 8966
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3202766180038452,
      "learning_rate": 1.618488729140195e-05,
      "loss": 0.608,
      "step": 8967
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4379792213439941,
      "learning_rate": 1.6184009856054567e-05,
      "loss": 0.6021,
      "step": 8968
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.292108178138733,
      "learning_rate": 1.6183132343608817e-05,
      "loss": 0.7679,
      "step": 8969
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5314406156539917,
      "learning_rate": 1.6182254754075635e-05,
      "loss": 0.6287,
      "step": 8970
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3862634897232056,
      "learning_rate": 1.618137708746596e-05,
      "loss": 0.5152,
      "step": 8971
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4681633710861206,
      "learning_rate": 1.618049934379074e-05,
      "loss": 0.623,
      "step": 8972
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7871567010879517,
      "learning_rate": 1.6179621523060908e-05,
      "loss": 0.7773,
      "step": 8973
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2667367458343506,
      "learning_rate": 1.6178743625287422e-05,
      "loss": 0.6227,
      "step": 8974
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8195037841796875,
      "learning_rate": 1.617786565048122e-05,
      "loss": 0.7193,
      "step": 8975
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5307555198669434,
      "learning_rate": 1.6176987598653243e-05,
      "loss": 0.7687,
      "step": 8976
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6931138038635254,
      "learning_rate": 1.6176109469814448e-05,
      "loss": 0.7616,
      "step": 8977
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7206300497055054,
      "learning_rate": 1.6175231263975777e-05,
      "loss": 0.7679,
      "step": 8978
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4747878313064575,
      "learning_rate": 1.6174352981148178e-05,
      "loss": 0.8503,
      "step": 8979
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7619401216506958,
      "learning_rate": 1.6173474621342603e-05,
      "loss": 0.8319,
      "step": 8980
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3140610456466675,
      "learning_rate": 1.6172596184570008e-05,
      "loss": 0.7959,
      "step": 8981
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.303749680519104,
      "learning_rate": 1.6171717670841332e-05,
      "loss": 0.7112,
      "step": 8982
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.543351411819458,
      "learning_rate": 1.617083908016754e-05,
      "loss": 0.6972,
      "step": 8983
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.391515851020813,
      "learning_rate": 1.616996041255958e-05,
      "loss": 0.6387,
      "step": 8984
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.639105200767517,
      "learning_rate": 1.616908166802841e-05,
      "loss": 0.8579,
      "step": 8985
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8375341892242432,
      "learning_rate": 1.616820284658498e-05,
      "loss": 0.6465,
      "step": 8986
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6230180263519287,
      "learning_rate": 1.616732394824025e-05,
      "loss": 0.7027,
      "step": 8987
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.861241340637207,
      "learning_rate": 1.616644497300518e-05,
      "loss": 0.5077,
      "step": 8988
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4467006921768188,
      "learning_rate": 1.6165565920890725e-05,
      "loss": 0.7167,
      "step": 8989
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6997394561767578,
      "learning_rate": 1.6164686791907843e-05,
      "loss": 0.5003,
      "step": 8990
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3696240186691284,
      "learning_rate": 1.61638075860675e-05,
      "loss": 0.6487,
      "step": 8991
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2235263586044312,
      "learning_rate": 1.6162928303380653e-05,
      "loss": 0.6612,
      "step": 8992
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2947700023651123,
      "learning_rate": 1.6162048943858266e-05,
      "loss": 0.6811,
      "step": 8993
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3490111827850342,
      "learning_rate": 1.6161169507511302e-05,
      "loss": 0.6589,
      "step": 8994
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5043827295303345,
      "learning_rate": 1.6160289994350725e-05,
      "loss": 0.8685,
      "step": 8995
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7771230936050415,
      "learning_rate": 1.6159410404387502e-05,
      "loss": 0.6048,
      "step": 8996
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3830431699752808,
      "learning_rate": 1.6158530737632598e-05,
      "loss": 0.7243,
      "step": 8997
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4041119813919067,
      "learning_rate": 1.6157650994096975e-05,
      "loss": 0.601,
      "step": 8998
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4096907377243042,
      "learning_rate": 1.615677117379161e-05,
      "loss": 0.6771,
      "step": 8999
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4847705364227295,
      "learning_rate": 1.6155891276727467e-05,
      "loss": 0.6013,
      "step": 9000
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6101168394088745,
      "learning_rate": 1.615501130291552e-05,
      "loss": 0.7161,
      "step": 9001
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4353210926055908,
      "learning_rate": 1.6154131252366734e-05,
      "loss": 0.7473,
      "step": 9002
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2967617511749268,
      "learning_rate": 1.615325112509208e-05,
      "loss": 0.7558,
      "step": 9003
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5656260251998901,
      "learning_rate": 1.615237092110254e-05,
      "loss": 0.6671,
      "step": 9004
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6772186756134033,
      "learning_rate": 1.615149064040908e-05,
      "loss": 0.8,
      "step": 9005
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.395031213760376,
      "learning_rate": 1.6150610283022678e-05,
      "loss": 0.6558,
      "step": 9006
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.357071876525879,
      "learning_rate": 1.614972984895431e-05,
      "loss": 0.6641,
      "step": 9007
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5920178890228271,
      "learning_rate": 1.6148849338214948e-05,
      "loss": 0.6861,
      "step": 9008
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4161750078201294,
      "learning_rate": 1.614796875081558e-05,
      "loss": 0.6299,
      "step": 9009
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5760127305984497,
      "learning_rate": 1.6147088086767172e-05,
      "loss": 0.6054,
      "step": 9010
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1963456869125366,
      "learning_rate": 1.6146207346080712e-05,
      "loss": 0.5664,
      "step": 9011
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5129860639572144,
      "learning_rate": 1.614532652876718e-05,
      "loss": 0.6646,
      "step": 9012
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3436095714569092,
      "learning_rate": 1.6144445634837554e-05,
      "loss": 0.7101,
      "step": 9013
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8270654678344727,
      "learning_rate": 1.6143564664302815e-05,
      "loss": 0.8563,
      "step": 9014
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2665249109268188,
      "learning_rate": 1.6142683617173952e-05,
      "loss": 0.7035,
      "step": 9015
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.38361394405365,
      "learning_rate": 1.6141802493461948e-05,
      "loss": 0.6603,
      "step": 9016
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2973992824554443,
      "learning_rate": 1.6140921293177785e-05,
      "loss": 0.668,
      "step": 9017
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7075222730636597,
      "learning_rate": 1.6140040016332453e-05,
      "loss": 0.762,
      "step": 9018
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.616663932800293,
      "learning_rate": 1.6139158662936938e-05,
      "loss": 0.714,
      "step": 9019
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.384337067604065,
      "learning_rate": 1.6138277233002228e-05,
      "loss": 0.562,
      "step": 9020
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2384382486343384,
      "learning_rate": 1.6137395726539307e-05,
      "loss": 0.631,
      "step": 9021
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3825255632400513,
      "learning_rate": 1.6136514143559178e-05,
      "loss": 0.6501,
      "step": 9022
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3250701427459717,
      "learning_rate": 1.6135632484072817e-05,
      "loss": 0.6286,
      "step": 9023
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.274959683418274,
      "learning_rate": 1.6134750748091225e-05,
      "loss": 0.634,
      "step": 9024
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2453858852386475,
      "learning_rate": 1.613386893562539e-05,
      "loss": 0.7246,
      "step": 9025
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3569772243499756,
      "learning_rate": 1.6132987046686315e-05,
      "loss": 0.7828,
      "step": 9026
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.684110403060913,
      "learning_rate": 1.6132105081284986e-05,
      "loss": 0.7146,
      "step": 9027
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6666232347488403,
      "learning_rate": 1.6131223039432398e-05,
      "loss": 0.5917,
      "step": 9028
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3556560277938843,
      "learning_rate": 1.6130340921139555e-05,
      "loss": 0.6219,
      "step": 9029
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.669176697731018,
      "learning_rate": 1.6129458726417448e-05,
      "loss": 0.6872,
      "step": 9030
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6511996984481812,
      "learning_rate": 1.612857645527708e-05,
      "loss": 0.7553,
      "step": 9031
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.581574559211731,
      "learning_rate": 1.6127694107729447e-05,
      "loss": 0.7693,
      "step": 9032
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5196735858917236,
      "learning_rate": 1.6126811683785554e-05,
      "loss": 0.6938,
      "step": 9033
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.412822961807251,
      "learning_rate": 1.61259291834564e-05,
      "loss": 0.7231,
      "step": 9034
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7860740423202515,
      "learning_rate": 1.6125046606752988e-05,
      "loss": 0.8504,
      "step": 9035
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6794878244400024,
      "learning_rate": 1.6124163953686324e-05,
      "loss": 0.6307,
      "step": 9036
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3220512866973877,
      "learning_rate": 1.6123281224267402e-05,
      "loss": 0.7735,
      "step": 9037
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3097196817398071,
      "learning_rate": 1.6122398418507237e-05,
      "loss": 0.5688,
      "step": 9038
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.183472990989685,
      "learning_rate": 1.6121515536416837e-05,
      "loss": 0.6384,
      "step": 9039
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7182323932647705,
      "learning_rate": 1.61206325780072e-05,
      "loss": 0.7963,
      "step": 9040
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.547318696975708,
      "learning_rate": 1.6119749543289343e-05,
      "loss": 0.6726,
      "step": 9041
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7211675643920898,
      "learning_rate": 1.6118866432274272e-05,
      "loss": 0.7308,
      "step": 9042
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3631863594055176,
      "learning_rate": 1.611798324497299e-05,
      "loss": 0.6316,
      "step": 9043
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3116055727005005,
      "learning_rate": 1.6117099981396522e-05,
      "loss": 0.6196,
      "step": 9044
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2382311820983887,
      "learning_rate": 1.611621664155587e-05,
      "loss": 0.6047,
      "step": 9045
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.1172893047332764,
      "learning_rate": 1.6115333225462052e-05,
      "loss": 0.7629,
      "step": 9046
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4898700714111328,
      "learning_rate": 1.6114449733126074e-05,
      "loss": 0.7186,
      "step": 9047
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2883445024490356,
      "learning_rate": 1.611356616455896e-05,
      "loss": 0.7419,
      "step": 9048
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.503781795501709,
      "learning_rate": 1.6112682519771722e-05,
      "loss": 0.5646,
      "step": 9049
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4411394596099854,
      "learning_rate": 1.6111798798775375e-05,
      "loss": 0.6012,
      "step": 9050
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2482255697250366,
      "learning_rate": 1.611091500158094e-05,
      "loss": 0.5728,
      "step": 9051
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4609975814819336,
      "learning_rate": 1.6110031128199434e-05,
      "loss": 0.7974,
      "step": 9052
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6299625635147095,
      "learning_rate": 1.6109147178641874e-05,
      "loss": 0.6697,
      "step": 9053
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.948811650276184,
      "learning_rate": 1.6108263152919285e-05,
      "loss": 0.8502,
      "step": 9054
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3985685110092163,
      "learning_rate": 1.610737905104269e-05,
      "loss": 0.5471,
      "step": 9055
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5584100484848022,
      "learning_rate": 1.6106494873023102e-05,
      "loss": 0.7832,
      "step": 9056
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5233045816421509,
      "learning_rate": 1.6105610618871553e-05,
      "loss": 0.6194,
      "step": 9057
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3422874212265015,
      "learning_rate": 1.6104726288599063e-05,
      "loss": 0.6537,
      "step": 9058
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3303078413009644,
      "learning_rate": 1.610384188221666e-05,
      "loss": 0.5244,
      "step": 9059
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6793485879898071,
      "learning_rate": 1.6102957399735374e-05,
      "loss": 0.7907,
      "step": 9060
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.486824631690979,
      "learning_rate": 1.6102072841166224e-05,
      "loss": 0.7887,
      "step": 9061
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4808979034423828,
      "learning_rate": 1.6101188206520237e-05,
      "loss": 0.6784,
      "step": 9062
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.662269115447998,
      "learning_rate": 1.6100303495808453e-05,
      "loss": 0.6513,
      "step": 9063
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4928779602050781,
      "learning_rate": 1.6099418709041895e-05,
      "loss": 0.6991,
      "step": 9064
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5773521661758423,
      "learning_rate": 1.609853384623159e-05,
      "loss": 0.6648,
      "step": 9065
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5743058919906616,
      "learning_rate": 1.6097648907388576e-05,
      "loss": 0.7573,
      "step": 9066
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3270418643951416,
      "learning_rate": 1.609676389252389e-05,
      "loss": 0.7546,
      "step": 9067
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.584200382232666,
      "learning_rate": 1.6095878801648555e-05,
      "loss": 0.733,
      "step": 9068
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2549690008163452,
      "learning_rate": 1.609499363477361e-05,
      "loss": 0.6742,
      "step": 9069
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.665834903717041,
      "learning_rate": 1.6094108391910097e-05,
      "loss": 0.7201,
      "step": 9070
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5971864461898804,
      "learning_rate": 1.6093223073069043e-05,
      "loss": 0.6745,
      "step": 9071
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3254919052124023,
      "learning_rate": 1.609233767826149e-05,
      "loss": 0.6688,
      "step": 9072
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4822255373001099,
      "learning_rate": 1.6091452207498476e-05,
      "loss": 0.721,
      "step": 9073
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.475172519683838,
      "learning_rate": 1.6090566660791044e-05,
      "loss": 0.6409,
      "step": 9074
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3838759660720825,
      "learning_rate": 1.608968103815023e-05,
      "loss": 0.6504,
      "step": 9075
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.260074496269226,
      "learning_rate": 1.6088795339587076e-05,
      "loss": 0.6485,
      "step": 9076
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5222052335739136,
      "learning_rate": 1.6087909565112627e-05,
      "loss": 0.5543,
      "step": 9077
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3398557901382446,
      "learning_rate": 1.6087023714737924e-05,
      "loss": 0.6674,
      "step": 9078
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.002509593963623,
      "learning_rate": 1.608613778847401e-05,
      "loss": 0.9004,
      "step": 9079
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5240682363510132,
      "learning_rate": 1.6085251786331933e-05,
      "loss": 0.6119,
      "step": 9080
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6218061447143555,
      "learning_rate": 1.608436570832274e-05,
      "loss": 0.7237,
      "step": 9081
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7616268396377563,
      "learning_rate": 1.6083479554457473e-05,
      "loss": 0.8484,
      "step": 9082
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5308449268341064,
      "learning_rate": 1.6082593324747185e-05,
      "loss": 0.657,
      "step": 9083
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4125343561172485,
      "learning_rate": 1.608170701920292e-05,
      "loss": 0.6484,
      "step": 9084
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4411166906356812,
      "learning_rate": 1.6080820637835733e-05,
      "loss": 0.6715,
      "step": 9085
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7586002349853516,
      "learning_rate": 1.6079934180656672e-05,
      "loss": 0.7615,
      "step": 9086
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3301950693130493,
      "learning_rate": 1.6079047647676792e-05,
      "loss": 0.7946,
      "step": 9087
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.549652338027954,
      "learning_rate": 1.607816103890714e-05,
      "loss": 0.6899,
      "step": 9088
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.847257137298584,
      "learning_rate": 1.6077274354358773e-05,
      "loss": 0.7308,
      "step": 9089
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4635883569717407,
      "learning_rate": 1.6076387594042748e-05,
      "loss": 0.8456,
      "step": 9090
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5353912115097046,
      "learning_rate": 1.607550075797012e-05,
      "loss": 0.6287,
      "step": 9091
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.633812665939331,
      "learning_rate": 1.6074613846151937e-05,
      "loss": 0.7826,
      "step": 9092
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3595807552337646,
      "learning_rate": 1.6073726858599266e-05,
      "loss": 0.7003,
      "step": 9093
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4725501537322998,
      "learning_rate": 1.6072839795323164e-05,
      "loss": 0.7083,
      "step": 9094
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4111570119857788,
      "learning_rate": 1.607195265633469e-05,
      "loss": 0.5443,
      "step": 9095
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5405298471450806,
      "learning_rate": 1.6071065441644903e-05,
      "loss": 0.6316,
      "step": 9096
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5529158115386963,
      "learning_rate": 1.6070178151264862e-05,
      "loss": 0.7685,
      "step": 9097
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7182228565216064,
      "learning_rate": 1.6069290785205632e-05,
      "loss": 0.7052,
      "step": 9098
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.387552261352539,
      "learning_rate": 1.606840334347828e-05,
      "loss": 0.8049,
      "step": 9099
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6054953336715698,
      "learning_rate": 1.606751582609386e-05,
      "loss": 0.7243,
      "step": 9100
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6200716495513916,
      "learning_rate": 1.6066628233063443e-05,
      "loss": 0.7307,
      "step": 9101
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7509976625442505,
      "learning_rate": 1.6065740564398097e-05,
      "loss": 0.7383,
      "step": 9102
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5522987842559814,
      "learning_rate": 1.6064852820108888e-05,
      "loss": 0.7015,
      "step": 9103
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3318783044815063,
      "learning_rate": 1.606396500020688e-05,
      "loss": 0.6401,
      "step": 9104
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4449830055236816,
      "learning_rate": 1.6063077104703148e-05,
      "loss": 0.7875,
      "step": 9105
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1384035348892212,
      "learning_rate": 1.6062189133608755e-05,
      "loss": 0.7165,
      "step": 9106
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4699159860610962,
      "learning_rate": 1.6061301086934773e-05,
      "loss": 0.6983,
      "step": 9107
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.807374119758606,
      "learning_rate": 1.6060412964692277e-05,
      "loss": 0.8003,
      "step": 9108
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7638925313949585,
      "learning_rate": 1.605952476689234e-05,
      "loss": 0.815,
      "step": 9109
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.135219931602478,
      "learning_rate": 1.605863649354603e-05,
      "loss": 0.6202,
      "step": 9110
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2460960149765015,
      "learning_rate": 1.605774814466443e-05,
      "loss": 0.6882,
      "step": 9111
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5781272649765015,
      "learning_rate": 1.6056859720258607e-05,
      "loss": 0.6405,
      "step": 9112
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5365488529205322,
      "learning_rate": 1.605597122033964e-05,
      "loss": 0.6957,
      "step": 9113
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.620904564857483,
      "learning_rate": 1.6055082644918606e-05,
      "loss": 0.7734,
      "step": 9114
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3895981311798096,
      "learning_rate": 1.6054193994006584e-05,
      "loss": 0.8184,
      "step": 9115
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.689070463180542,
      "learning_rate": 1.6053305267614657e-05,
      "loss": 0.7694,
      "step": 9116
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6874887943267822,
      "learning_rate": 1.6052416465753896e-05,
      "loss": 0.6884,
      "step": 9117
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5769094228744507,
      "learning_rate": 1.6051527588435392e-05,
      "loss": 0.6875,
      "step": 9118
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3797211647033691,
      "learning_rate": 1.605063863567022e-05,
      "loss": 0.5793,
      "step": 9119
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2340060472488403,
      "learning_rate": 1.6049749607469465e-05,
      "loss": 0.5085,
      "step": 9120
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6154961585998535,
      "learning_rate": 1.6048860503844208e-05,
      "loss": 0.6727,
      "step": 9121
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5539370775222778,
      "learning_rate": 1.604797132480554e-05,
      "loss": 0.8004,
      "step": 9122
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3730357885360718,
      "learning_rate": 1.6047082070364543e-05,
      "loss": 0.7008,
      "step": 9123
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.499200463294983,
      "learning_rate": 1.6046192740532304e-05,
      "loss": 0.6898,
      "step": 9124
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.420112133026123,
      "learning_rate": 1.6045303335319912e-05,
      "loss": 0.6648,
      "step": 9125
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.613187551498413,
      "learning_rate": 1.6044413854738454e-05,
      "loss": 0.765,
      "step": 9126
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.554463267326355,
      "learning_rate": 1.604352429879902e-05,
      "loss": 0.7481,
      "step": 9127
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3476115465164185,
      "learning_rate": 1.6042634667512697e-05,
      "loss": 0.7128,
      "step": 9128
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3841630220413208,
      "learning_rate": 1.6041744960890584e-05,
      "loss": 0.7566,
      "step": 9129
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3911899328231812,
      "learning_rate": 1.6040855178943765e-05,
      "loss": 0.7865,
      "step": 9130
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.74187433719635,
      "learning_rate": 1.603996532168334e-05,
      "loss": 0.7855,
      "step": 9131
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.280556321144104,
      "learning_rate": 1.60390753891204e-05,
      "loss": 0.7291,
      "step": 9132
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.530800223350525,
      "learning_rate": 1.6038185381266037e-05,
      "loss": 0.5402,
      "step": 9133
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.356784701347351,
      "learning_rate": 1.6037295298131355e-05,
      "loss": 0.5949,
      "step": 9134
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5662821531295776,
      "learning_rate": 1.6036405139727444e-05,
      "loss": 0.6766,
      "step": 9135
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2046486139297485,
      "learning_rate": 1.6035514906065405e-05,
      "loss": 0.6652,
      "step": 9136
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2482439279556274,
      "learning_rate": 1.6034624597156337e-05,
      "loss": 0.4939,
      "step": 9137
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2851958274841309,
      "learning_rate": 1.6033734213011336e-05,
      "loss": 0.6128,
      "step": 9138
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6441318988800049,
      "learning_rate": 1.6032843753641508e-05,
      "loss": 0.7296,
      "step": 9139
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.551375389099121,
      "learning_rate": 1.6031953219057953e-05,
      "loss": 0.8152,
      "step": 9140
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3577693700790405,
      "learning_rate": 1.6031062609271774e-05,
      "loss": 0.689,
      "step": 9141
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4269126653671265,
      "learning_rate": 1.6030171924294077e-05,
      "loss": 0.6618,
      "step": 9142
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2698343992233276,
      "learning_rate": 1.6029281164135958e-05,
      "loss": 0.6339,
      "step": 9143
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6374996900558472,
      "learning_rate": 1.6028390328808527e-05,
      "loss": 0.7242,
      "step": 9144
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.741187572479248,
      "learning_rate": 1.6027499418322894e-05,
      "loss": 0.7388,
      "step": 9145
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2757251262664795,
      "learning_rate": 1.6026608432690164e-05,
      "loss": 0.6851,
      "step": 9146
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3997712135314941,
      "learning_rate": 1.602571737192144e-05,
      "loss": 0.4565,
      "step": 9147
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4471238851547241,
      "learning_rate": 1.602482623602784e-05,
      "loss": 0.7644,
      "step": 9148
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7255090475082397,
      "learning_rate": 1.602393502502047e-05,
      "loss": 0.6714,
      "step": 9149
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6083650588989258,
      "learning_rate": 1.6023043738910438e-05,
      "loss": 0.7521,
      "step": 9150
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.462259292602539,
      "learning_rate": 1.6022152377708863e-05,
      "loss": 0.6852,
      "step": 9151
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.606452226638794,
      "learning_rate": 1.602126094142685e-05,
      "loss": 0.9362,
      "step": 9152
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5884156227111816,
      "learning_rate": 1.602036943007552e-05,
      "loss": 0.6906,
      "step": 9153
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4692356586456299,
      "learning_rate": 1.6019477843665986e-05,
      "loss": 0.6805,
      "step": 9154
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6033525466918945,
      "learning_rate": 1.6018586182209358e-05,
      "loss": 0.7094,
      "step": 9155
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7055482864379883,
      "learning_rate": 1.6017694445716758e-05,
      "loss": 0.5488,
      "step": 9156
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.706266164779663,
      "learning_rate": 1.6016802634199305e-05,
      "loss": 0.8959,
      "step": 9157
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7370105981826782,
      "learning_rate": 1.6015910747668113e-05,
      "loss": 0.5344,
      "step": 9158
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6573643684387207,
      "learning_rate": 1.6015018786134305e-05,
      "loss": 0.6863,
      "step": 9159
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8642451763153076,
      "learning_rate": 1.6014126749609e-05,
      "loss": 0.7144,
      "step": 9160
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9192383289337158,
      "learning_rate": 1.601323463810332e-05,
      "loss": 0.7596,
      "step": 9161
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8094550371170044,
      "learning_rate": 1.6012342451628383e-05,
      "loss": 0.7139,
      "step": 9162
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8322092294692993,
      "learning_rate": 1.601145019019532e-05,
      "loss": 0.8484,
      "step": 9163
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5123748779296875,
      "learning_rate": 1.601055785381525e-05,
      "loss": 0.6222,
      "step": 9164
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6426951885223389,
      "learning_rate": 1.60096654424993e-05,
      "loss": 0.658,
      "step": 9165
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2613416910171509,
      "learning_rate": 1.6008772956258596e-05,
      "loss": 0.7052,
      "step": 9166
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9505983591079712,
      "learning_rate": 1.6007880395104263e-05,
      "loss": 0.7173,
      "step": 9167
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3165497779846191,
      "learning_rate": 1.600698775904743e-05,
      "loss": 0.6636,
      "step": 9168
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3978666067123413,
      "learning_rate": 1.6006095048099225e-05,
      "loss": 0.7082,
      "step": 9169
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1252524852752686,
      "learning_rate": 1.600520226227078e-05,
      "loss": 0.5978,
      "step": 9170
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4283946752548218,
      "learning_rate": 1.6004309401573224e-05,
      "loss": 0.7311,
      "step": 9171
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.578310251235962,
      "learning_rate": 1.600341646601769e-05,
      "loss": 0.7561,
      "step": 9172
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5567221641540527,
      "learning_rate": 1.600252345561531e-05,
      "loss": 0.7675,
      "step": 9173
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2937091588974,
      "learning_rate": 1.6001630370377216e-05,
      "loss": 0.7499,
      "step": 9174
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2248051166534424,
      "learning_rate": 1.6000737210314542e-05,
      "loss": 0.7495,
      "step": 9175
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2722069025039673,
      "learning_rate": 1.5999843975438426e-05,
      "loss": 0.6184,
      "step": 9176
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3351457118988037,
      "learning_rate": 1.5998950665760006e-05,
      "loss": 0.624,
      "step": 9177
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.489237904548645,
      "learning_rate": 1.599805728129041e-05,
      "loss": 0.7127,
      "step": 9178
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4775676727294922,
      "learning_rate": 1.5997163822040788e-05,
      "loss": 0.7924,
      "step": 9179
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3620872497558594,
      "learning_rate": 1.5996270288022274e-05,
      "loss": 0.8218,
      "step": 9180
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6718759536743164,
      "learning_rate": 1.5995376679246004e-05,
      "loss": 0.6554,
      "step": 9181
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4907565116882324,
      "learning_rate": 1.5994482995723125e-05,
      "loss": 0.6773,
      "step": 9182
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3970160484313965,
      "learning_rate": 1.5993589237464777e-05,
      "loss": 0.6167,
      "step": 9183
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4755587577819824,
      "learning_rate": 1.59926954044821e-05,
      "loss": 0.6354,
      "step": 9184
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5512633323669434,
      "learning_rate": 1.5991801496786244e-05,
      "loss": 0.6328,
      "step": 9185
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4683494567871094,
      "learning_rate": 1.599090751438835e-05,
      "loss": 0.5484,
      "step": 9186
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3748137950897217,
      "learning_rate": 1.599001345729956e-05,
      "loss": 0.7852,
      "step": 9187
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4914777278900146,
      "learning_rate": 1.5989119325531022e-05,
      "loss": 0.5688,
      "step": 9188
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4154454469680786,
      "learning_rate": 1.598822511909389e-05,
      "loss": 0.6446,
      "step": 9189
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7448453903198242,
      "learning_rate": 1.5987330837999304e-05,
      "loss": 0.8024,
      "step": 9190
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7181921005249023,
      "learning_rate": 1.598643648225842e-05,
      "loss": 0.6192,
      "step": 9191
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.354056715965271,
      "learning_rate": 1.5985542051882387e-05,
      "loss": 0.7235,
      "step": 9192
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.481635332107544,
      "learning_rate": 1.598464754688235e-05,
      "loss": 0.6595,
      "step": 9193
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2197932004928589,
      "learning_rate": 1.598375296726947e-05,
      "loss": 0.582,
      "step": 9194
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4904513359069824,
      "learning_rate": 1.5982858313054894e-05,
      "loss": 0.702,
      "step": 9195
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.921148657798767,
      "learning_rate": 1.5981963584249777e-05,
      "loss": 0.6478,
      "step": 9196
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2817858457565308,
      "learning_rate": 1.5981068780865276e-05,
      "loss": 0.4893,
      "step": 9197
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6092064380645752,
      "learning_rate": 1.5980173902912547e-05,
      "loss": 0.7576,
      "step": 9198
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2526222467422485,
      "learning_rate": 1.5979278950402742e-05,
      "loss": 0.5279,
      "step": 9199
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4512706995010376,
      "learning_rate": 1.5978383923347024e-05,
      "loss": 0.7527,
      "step": 9200
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.568178415298462,
      "learning_rate": 1.597748882175655e-05,
      "loss": 0.6382,
      "step": 9201
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4015419483184814,
      "learning_rate": 1.5976593645642478e-05,
      "loss": 0.8076,
      "step": 9202
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3986419439315796,
      "learning_rate": 1.597569839501597e-05,
      "loss": 0.7094,
      "step": 9203
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3593508005142212,
      "learning_rate": 1.597480306988819e-05,
      "loss": 0.6468,
      "step": 9204
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2940551042556763,
      "learning_rate": 1.5973907670270296e-05,
      "loss": 0.5941,
      "step": 9205
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2408500909805298,
      "learning_rate": 1.597301219617345e-05,
      "loss": 0.7876,
      "step": 9206
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5778807401657104,
      "learning_rate": 1.597211664760882e-05,
      "loss": 0.8472,
      "step": 9207
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4484491348266602,
      "learning_rate": 1.5971221024587572e-05,
      "loss": 0.6772,
      "step": 9208
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4336814880371094,
      "learning_rate": 1.5970325327120872e-05,
      "loss": 0.7198,
      "step": 9209
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7500416040420532,
      "learning_rate": 1.596942955521988e-05,
      "loss": 0.767,
      "step": 9210
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4292001724243164,
      "learning_rate": 1.5968533708895773e-05,
      "loss": 0.5791,
      "step": 9211
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5279998779296875,
      "learning_rate": 1.5967637788159716e-05,
      "loss": 0.6596,
      "step": 9212
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2651387453079224,
      "learning_rate": 1.5966741793022877e-05,
      "loss": 0.7318,
      "step": 9213
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3303966522216797,
      "learning_rate": 1.596584572349643e-05,
      "loss": 0.6692,
      "step": 9214
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4072940349578857,
      "learning_rate": 1.5964949579591545e-05,
      "loss": 0.6842,
      "step": 9215
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8273640871047974,
      "learning_rate": 1.5964053361319393e-05,
      "loss": 0.7344,
      "step": 9216
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5635849237442017,
      "learning_rate": 1.5963157068691152e-05,
      "loss": 0.6884,
      "step": 9217
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3393745422363281,
      "learning_rate": 1.596226070171799e-05,
      "loss": 0.6711,
      "step": 9218
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5494143962860107,
      "learning_rate": 1.596136426041109e-05,
      "loss": 0.8006,
      "step": 9219
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4799003601074219,
      "learning_rate": 1.5960467744781624e-05,
      "loss": 0.7073,
      "step": 9220
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8737550973892212,
      "learning_rate": 1.5959571154840766e-05,
      "loss": 0.5651,
      "step": 9221
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8356828689575195,
      "learning_rate": 1.59586744905997e-05,
      "loss": 0.6217,
      "step": 9222
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5250346660614014,
      "learning_rate": 1.5957777752069602e-05,
      "loss": 0.6179,
      "step": 9223
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5698802471160889,
      "learning_rate": 1.5956880939261654e-05,
      "loss": 0.6832,
      "step": 9224
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.599947452545166,
      "learning_rate": 1.5955984052187035e-05,
      "loss": 0.7732,
      "step": 9225
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5565190315246582,
      "learning_rate": 1.5955087090856928e-05,
      "loss": 0.6562,
      "step": 9226
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.271716594696045,
      "learning_rate": 1.595419005528251e-05,
      "loss": 0.6839,
      "step": 9227
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.0067882537841797,
      "learning_rate": 1.5953292945474977e-05,
      "loss": 0.6251,
      "step": 9228
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8107162714004517,
      "learning_rate": 1.5952395761445502e-05,
      "loss": 0.809,
      "step": 9229
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.701969861984253,
      "learning_rate": 1.5951498503205275e-05,
      "loss": 0.6987,
      "step": 9230
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.712892770767212,
      "learning_rate": 1.5950601170765483e-05,
      "loss": 0.7066,
      "step": 9231
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3863725662231445,
      "learning_rate": 1.594970376413731e-05,
      "loss": 0.6442,
      "step": 9232
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4591028690338135,
      "learning_rate": 1.5948806283331954e-05,
      "loss": 0.6484,
      "step": 9233
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.657221794128418,
      "learning_rate": 1.594790872836059e-05,
      "loss": 0.7315,
      "step": 9234
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.035008668899536,
      "learning_rate": 1.594701109923442e-05,
      "loss": 0.8904,
      "step": 9235
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5070103406906128,
      "learning_rate": 1.594611339596463e-05,
      "loss": 0.5505,
      "step": 9236
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4958736896514893,
      "learning_rate": 1.594521561856241e-05,
      "loss": 0.8074,
      "step": 9237
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6144347190856934,
      "learning_rate": 1.5944317767038958e-05,
      "loss": 0.6788,
      "step": 9238
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5009742975234985,
      "learning_rate": 1.594341984140546e-05,
      "loss": 0.7182,
      "step": 9239
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2730709314346313,
      "learning_rate": 1.5942521841673122e-05,
      "loss": 0.6688,
      "step": 9240
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4073809385299683,
      "learning_rate": 1.594162376785313e-05,
      "loss": 0.7542,
      "step": 9241
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3045551776885986,
      "learning_rate": 1.594072561995669e-05,
      "loss": 0.5974,
      "step": 9242
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9677248001098633,
      "learning_rate": 1.5939827397994988e-05,
      "loss": 0.6547,
      "step": 9243
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1637741327285767,
      "learning_rate": 1.5938929101979233e-05,
      "loss": 0.5791,
      "step": 9244
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3625832796096802,
      "learning_rate": 1.593803073192062e-05,
      "loss": 0.6023,
      "step": 9245
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4449174404144287,
      "learning_rate": 1.5937132287830345e-05,
      "loss": 0.719,
      "step": 9246
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7241389751434326,
      "learning_rate": 1.5936233769719614e-05,
      "loss": 0.7111,
      "step": 9247
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5888514518737793,
      "learning_rate": 1.593533517759963e-05,
      "loss": 0.74,
      "step": 9248
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5238280296325684,
      "learning_rate": 1.5934436511481594e-05,
      "loss": 0.7244,
      "step": 9249
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.736968994140625,
      "learning_rate": 1.5933537771376707e-05,
      "loss": 0.696,
      "step": 9250
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6687426567077637,
      "learning_rate": 1.5932638957296184e-05,
      "loss": 0.691,
      "step": 9251
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4953669309616089,
      "learning_rate": 1.593174006925122e-05,
      "loss": 0.7779,
      "step": 9252
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3720297813415527,
      "learning_rate": 1.593084110725303e-05,
      "loss": 0.7367,
      "step": 9253
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5663366317749023,
      "learning_rate": 1.5929942071312815e-05,
      "loss": 0.819,
      "step": 9254
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.376185417175293,
      "learning_rate": 1.5929042961441787e-05,
      "loss": 0.7395,
      "step": 9255
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5468103885650635,
      "learning_rate": 1.5928143777651156e-05,
      "loss": 0.7196,
      "step": 9256
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0579071044921875,
      "learning_rate": 1.592724451995213e-05,
      "loss": 0.6121,
      "step": 9257
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4049320220947266,
      "learning_rate": 1.592634518835592e-05,
      "loss": 0.638,
      "step": 9258
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.269745945930481,
      "learning_rate": 1.5925445782873744e-05,
      "loss": 0.4744,
      "step": 9259
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3415929079055786,
      "learning_rate": 1.5924546303516807e-05,
      "loss": 0.6517,
      "step": 9260
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9591788053512573,
      "learning_rate": 1.592364675029633e-05,
      "loss": 0.7822,
      "step": 9261
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5843422412872314,
      "learning_rate": 1.592274712322353e-05,
      "loss": 0.8605,
      "step": 9262
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.753039002418518,
      "learning_rate": 1.5921847422309613e-05,
      "loss": 0.7394,
      "step": 9263
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5416609048843384,
      "learning_rate": 1.5920947647565803e-05,
      "loss": 0.7483,
      "step": 9264
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3780567646026611,
      "learning_rate": 1.5920047799003318e-05,
      "loss": 0.6225,
      "step": 9265
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2645349502563477,
      "learning_rate": 1.591914787663337e-05,
      "loss": 0.734,
      "step": 9266
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4702900648117065,
      "learning_rate": 1.591824788046719e-05,
      "loss": 0.6559,
      "step": 9267
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.307447075843811,
      "learning_rate": 1.5917347810515988e-05,
      "loss": 0.6766,
      "step": 9268
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3278735876083374,
      "learning_rate": 1.591644766679099e-05,
      "loss": 0.6615,
      "step": 9269
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4538547992706299,
      "learning_rate": 1.5915547449303425e-05,
      "loss": 0.749,
      "step": 9270
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.999019980430603,
      "learning_rate": 1.5914647158064503e-05,
      "loss": 0.7248,
      "step": 9271
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3323171138763428,
      "learning_rate": 1.5913746793085455e-05,
      "loss": 0.5039,
      "step": 9272
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2260316610336304,
      "learning_rate": 1.5912846354377508e-05,
      "loss": 0.6078,
      "step": 9273
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5481525659561157,
      "learning_rate": 1.5911945841951887e-05,
      "loss": 0.7024,
      "step": 9274
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5960232019424438,
      "learning_rate": 1.5911045255819814e-05,
      "loss": 0.7371,
      "step": 9275
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5129109621047974,
      "learning_rate": 1.5910144595992527e-05,
      "loss": 0.6652,
      "step": 9276
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.630895733833313,
      "learning_rate": 1.5909243862481244e-05,
      "loss": 0.64,
      "step": 9277
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3824483156204224,
      "learning_rate": 1.5908343055297203e-05,
      "loss": 0.59,
      "step": 9278
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4794976711273193,
      "learning_rate": 1.5907442174451627e-05,
      "loss": 0.6891,
      "step": 9279
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.375944972038269,
      "learning_rate": 1.5906541219955757e-05,
      "loss": 0.784,
      "step": 9280
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4401723146438599,
      "learning_rate": 1.5905640191820818e-05,
      "loss": 0.6131,
      "step": 9281
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6000292301177979,
      "learning_rate": 1.5904739090058048e-05,
      "loss": 0.727,
      "step": 9282
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.090409755706787,
      "learning_rate": 1.590383791467868e-05,
      "loss": 0.5135,
      "step": 9283
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5160526037216187,
      "learning_rate": 1.590293666569395e-05,
      "loss": 0.6324,
      "step": 9284
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5382275581359863,
      "learning_rate": 1.5902035343115087e-05,
      "loss": 0.6707,
      "step": 9285
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1981474161148071,
      "learning_rate": 1.5901133946953337e-05,
      "loss": 0.7412,
      "step": 9286
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5731463432312012,
      "learning_rate": 1.5900232477219936e-05,
      "loss": 0.6804,
      "step": 9287
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5095021724700928,
      "learning_rate": 1.589933093392612e-05,
      "loss": 0.7225,
      "step": 9288
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7401556968688965,
      "learning_rate": 1.5898429317083138e-05,
      "loss": 0.6279,
      "step": 9289
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.273559808731079,
      "learning_rate": 1.5897527626702217e-05,
      "loss": 0.6219,
      "step": 9290
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5695457458496094,
      "learning_rate": 1.589662586279461e-05,
      "loss": 0.8198,
      "step": 9291
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4663647413253784,
      "learning_rate": 1.5895724025371547e-05,
      "loss": 0.6616,
      "step": 9292
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.747071385383606,
      "learning_rate": 1.5894822114444288e-05,
      "loss": 0.8086,
      "step": 9293
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4684890508651733,
      "learning_rate": 1.5893920130024065e-05,
      "loss": 0.8032,
      "step": 9294
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.352401614189148,
      "learning_rate": 1.589301807212213e-05,
      "loss": 0.8928,
      "step": 9295
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4375923871994019,
      "learning_rate": 1.5892115940749727e-05,
      "loss": 0.7644,
      "step": 9296
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.0568833351135254,
      "learning_rate": 1.58912137359181e-05,
      "loss": 0.8134,
      "step": 9297
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.300429344177246,
      "learning_rate": 1.5890311457638504e-05,
      "loss": 0.6765,
      "step": 9298
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5827295780181885,
      "learning_rate": 1.588940910592218e-05,
      "loss": 0.7508,
      "step": 9299
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3508989810943604,
      "learning_rate": 1.5888506680780385e-05,
      "loss": 0.812,
      "step": 9300
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4290168285369873,
      "learning_rate": 1.5887604182224364e-05,
      "loss": 0.6678,
      "step": 9301
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2778339385986328,
      "learning_rate": 1.5886701610265376e-05,
      "loss": 0.7346,
      "step": 9302
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8200464248657227,
      "learning_rate": 1.5885798964914666e-05,
      "loss": 0.8544,
      "step": 9303
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4374443292617798,
      "learning_rate": 1.5884896246183496e-05,
      "loss": 0.6871,
      "step": 9304
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5263241529464722,
      "learning_rate": 1.588399345408311e-05,
      "loss": 0.6188,
      "step": 9305
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3420758247375488,
      "learning_rate": 1.5883090588624774e-05,
      "loss": 0.7745,
      "step": 9306
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4219733476638794,
      "learning_rate": 1.588218764981974e-05,
      "loss": 0.6832,
      "step": 9307
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4102431535720825,
      "learning_rate": 1.5881284637679257e-05,
      "loss": 0.7225,
      "step": 9308
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8120661973953247,
      "learning_rate": 1.5880381552214598e-05,
      "loss": 0.7127,
      "step": 9309
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8935903310775757,
      "learning_rate": 1.587947839343701e-05,
      "loss": 0.6792,
      "step": 9310
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2995120286941528,
      "learning_rate": 1.5878575161357763e-05,
      "loss": 0.6317,
      "step": 9311
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5242059230804443,
      "learning_rate": 1.587767185598811e-05,
      "loss": 0.6837,
      "step": 9312
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8726019859313965,
      "learning_rate": 1.5876768477339318e-05,
      "loss": 0.6983,
      "step": 9313
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8197579383850098,
      "learning_rate": 1.5875865025422646e-05,
      "loss": 0.8136,
      "step": 9314
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3358311653137207,
      "learning_rate": 1.587496150024936e-05,
      "loss": 0.7623,
      "step": 9315
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3697388172149658,
      "learning_rate": 1.5874057901830724e-05,
      "loss": 0.6821,
      "step": 9316
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5079323053359985,
      "learning_rate": 1.5873154230178e-05,
      "loss": 0.8267,
      "step": 9317
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5764621496200562,
      "learning_rate": 1.5872250485302462e-05,
      "loss": 0.8275,
      "step": 9318
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5703102350234985,
      "learning_rate": 1.5871346667215375e-05,
      "loss": 0.7725,
      "step": 9319
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4897404909133911,
      "learning_rate": 1.5870442775928002e-05,
      "loss": 0.5809,
      "step": 9320
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.607880711555481,
      "learning_rate": 1.5869538811451616e-05,
      "loss": 0.7962,
      "step": 9321
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4555529356002808,
      "learning_rate": 1.5868634773797485e-05,
      "loss": 0.7988,
      "step": 9322
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3428586721420288,
      "learning_rate": 1.5867730662976885e-05,
      "loss": 0.7123,
      "step": 9323
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3901069164276123,
      "learning_rate": 1.5866826479001082e-05,
      "loss": 0.6243,
      "step": 9324
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2106765508651733,
      "learning_rate": 1.586592222188135e-05,
      "loss": 0.739,
      "step": 9325
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4398550987243652,
      "learning_rate": 1.5865017891628966e-05,
      "loss": 0.6896,
      "step": 9326
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.479638934135437,
      "learning_rate": 1.5864113488255203e-05,
      "loss": 0.7824,
      "step": 9327
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6369678974151611,
      "learning_rate": 1.5863209011771338e-05,
      "loss": 0.6909,
      "step": 9328
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.415558099746704,
      "learning_rate": 1.586230446218864e-05,
      "loss": 0.6369,
      "step": 9329
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.617601752281189,
      "learning_rate": 1.5861399839518394e-05,
      "loss": 0.6702,
      "step": 9330
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4118651151657104,
      "learning_rate": 1.586049514377188e-05,
      "loss": 0.7913,
      "step": 9331
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6947683095932007,
      "learning_rate": 1.585959037496037e-05,
      "loss": 0.5473,
      "step": 9332
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5091623067855835,
      "learning_rate": 1.5858685533095147e-05,
      "loss": 0.713,
      "step": 9333
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2792500257492065,
      "learning_rate": 1.5857780618187492e-05,
      "loss": 0.7076,
      "step": 9334
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4541075229644775,
      "learning_rate": 1.585687563024869e-05,
      "loss": 0.7924,
      "step": 9335
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.732025384902954,
      "learning_rate": 1.5855970569290024e-05,
      "loss": 0.8243,
      "step": 9336
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4483592510223389,
      "learning_rate": 1.585506543532277e-05,
      "loss": 0.5666,
      "step": 9337
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4619684219360352,
      "learning_rate": 1.5854160228358218e-05,
      "loss": 0.6951,
      "step": 9338
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3144264221191406,
      "learning_rate": 1.5853254948407653e-05,
      "loss": 0.7505,
      "step": 9339
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.489968180656433,
      "learning_rate": 1.5852349595482364e-05,
      "loss": 0.8036,
      "step": 9340
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9944069385528564,
      "learning_rate": 1.5851444169593634e-05,
      "loss": 0.921,
      "step": 9341
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.217079997062683,
      "learning_rate": 1.5850538670752754e-05,
      "loss": 0.5911,
      "step": 9342
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6067509651184082,
      "learning_rate": 1.584963309897101e-05,
      "loss": 0.7491,
      "step": 9343
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4500958919525146,
      "learning_rate": 1.5848727454259697e-05,
      "loss": 0.8345,
      "step": 9344
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9589881896972656,
      "learning_rate": 1.5847821736630105e-05,
      "loss": 0.7524,
      "step": 9345
    },
    {
      "epoch": 0.97,
      "grad_norm": 77.05789184570312,
      "learning_rate": 1.5846915946093525e-05,
      "loss": 2.6682,
      "step": 9346
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4527896642684937,
      "learning_rate": 1.5846010082661245e-05,
      "loss": 0.6128,
      "step": 9347
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4520930051803589,
      "learning_rate": 1.5845104146344565e-05,
      "loss": 0.5373,
      "step": 9348
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4040601253509521,
      "learning_rate": 1.5844198137154778e-05,
      "loss": 0.8135,
      "step": 9349
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4711753129959106,
      "learning_rate": 1.5843292055103182e-05,
      "loss": 0.6877,
      "step": 9350
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4342195987701416,
      "learning_rate": 1.5842385900201065e-05,
      "loss": 0.734,
      "step": 9351
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2729568481445312,
      "learning_rate": 1.5841479672459735e-05,
      "loss": 0.6343,
      "step": 9352
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2805966138839722,
      "learning_rate": 1.584057337189048e-05,
      "loss": 0.8014,
      "step": 9353
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6249486207962036,
      "learning_rate": 1.583966699850461e-05,
      "loss": 0.7762,
      "step": 9354
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5109248161315918,
      "learning_rate": 1.583876055231342e-05,
      "loss": 0.7314,
      "step": 9355
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3657569885253906,
      "learning_rate": 1.583785403332821e-05,
      "loss": 0.6642,
      "step": 9356
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2148076295852661,
      "learning_rate": 1.583694744156028e-05,
      "loss": 0.6814,
      "step": 9357
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4737011194229126,
      "learning_rate": 1.5836040777020936e-05,
      "loss": 0.7143,
      "step": 9358
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1455858945846558,
      "learning_rate": 1.583513403972148e-05,
      "loss": 0.7694,
      "step": 9359
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6428463459014893,
      "learning_rate": 1.583422722967322e-05,
      "loss": 0.8408,
      "step": 9360
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3135398626327515,
      "learning_rate": 1.5833320346887457e-05,
      "loss": 0.6994,
      "step": 9361
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3835972547531128,
      "learning_rate": 1.5832413391375503e-05,
      "loss": 0.6964,
      "step": 9362
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2336093187332153,
      "learning_rate": 1.583150636314866e-05,
      "loss": 0.5359,
      "step": 9363
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3505891561508179,
      "learning_rate": 1.583059926221824e-05,
      "loss": 0.637,
      "step": 9364
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2554329633712769,
      "learning_rate": 1.582969208859555e-05,
      "loss": 0.7541,
      "step": 9365
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5028815269470215,
      "learning_rate": 1.5828784842291903e-05,
      "loss": 0.822,
      "step": 9366
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2608333826065063,
      "learning_rate": 1.5827877523318603e-05,
      "loss": 0.5465,
      "step": 9367
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7039021253585815,
      "learning_rate": 1.582697013168697e-05,
      "loss": 0.6622,
      "step": 9368
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4717432260513306,
      "learning_rate": 1.582606266740831e-05,
      "loss": 0.6016,
      "step": 9369
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5430530309677124,
      "learning_rate": 1.5825155130493942e-05,
      "loss": 0.6367,
      "step": 9370
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.476911187171936,
      "learning_rate": 1.582424752095518e-05,
      "loss": 0.7637,
      "step": 9371
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5645663738250732,
      "learning_rate": 1.5823339838803336e-05,
      "loss": 0.7107,
      "step": 9372
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.544690489768982,
      "learning_rate": 1.5822432084049728e-05,
      "loss": 0.6804,
      "step": 9373
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2657666206359863,
      "learning_rate": 1.5821524256705677e-05,
      "loss": 0.6071,
      "step": 9374
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7340401411056519,
      "learning_rate": 1.5820616356782494e-05,
      "loss": 0.718,
      "step": 9375
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1990211009979248,
      "learning_rate": 1.5819708384291505e-05,
      "loss": 0.6946,
      "step": 9376
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4890568256378174,
      "learning_rate": 1.5818800339244026e-05,
      "loss": 0.7132,
      "step": 9377
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4728165864944458,
      "learning_rate": 1.581789222165138e-05,
      "loss": 0.7139,
      "step": 9378
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4651442766189575,
      "learning_rate": 1.5816984031524887e-05,
      "loss": 0.6653,
      "step": 9379
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4859870672225952,
      "learning_rate": 1.5816075768875873e-05,
      "loss": 0.7245,
      "step": 9380
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5890778303146362,
      "learning_rate": 1.5815167433715655e-05,
      "loss": 0.81,
      "step": 9381
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3492170572280884,
      "learning_rate": 1.5814259026055563e-05,
      "loss": 0.6492,
      "step": 9382
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4024384021759033,
      "learning_rate": 1.5813350545906924e-05,
      "loss": 0.654,
      "step": 9383
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4189728498458862,
      "learning_rate": 1.5812441993281063e-05,
      "loss": 0.6591,
      "step": 9384
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.663483738899231,
      "learning_rate": 1.58115333681893e-05,
      "loss": 0.7736,
      "step": 9385
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7331984043121338,
      "learning_rate": 1.5810624670642974e-05,
      "loss": 0.8427,
      "step": 9386
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4278161525726318,
      "learning_rate": 1.5809715900653412e-05,
      "loss": 0.6296,
      "step": 9387
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5571112632751465,
      "learning_rate": 1.5808807058231936e-05,
      "loss": 0.6508,
      "step": 9388
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2363182306289673,
      "learning_rate": 1.5807898143389884e-05,
      "loss": 0.7491,
      "step": 9389
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.580129861831665,
      "learning_rate": 1.5806989156138587e-05,
      "loss": 0.7597,
      "step": 9390
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4619699716567993,
      "learning_rate": 1.5806080096489375e-05,
      "loss": 0.6548,
      "step": 9391
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7127330303192139,
      "learning_rate": 1.580517096445358e-05,
      "loss": 0.7749,
      "step": 9392
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4238356351852417,
      "learning_rate": 1.5804261760042545e-05,
      "loss": 0.7188,
      "step": 9393
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3291910886764526,
      "learning_rate": 1.58033524832676e-05,
      "loss": 0.518,
      "step": 9394
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6479713916778564,
      "learning_rate": 1.580244313414008e-05,
      "loss": 0.6019,
      "step": 9395
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3654608726501465,
      "learning_rate": 1.580153371267132e-05,
      "loss": 0.5457,
      "step": 9396
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4424479007720947,
      "learning_rate": 1.5800624218872668e-05,
      "loss": 0.7069,
      "step": 9397
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.634773850440979,
      "learning_rate": 1.5799714652755453e-05,
      "loss": 0.6288,
      "step": 9398
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2443950176239014,
      "learning_rate": 1.579880501433102e-05,
      "loss": 0.7301,
      "step": 9399
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4470089673995972,
      "learning_rate": 1.579789530361071e-05,
      "loss": 0.6842,
      "step": 9400
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3118149042129517,
      "learning_rate": 1.579698552060586e-05,
      "loss": 0.6358,
      "step": 9401
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3958622217178345,
      "learning_rate": 1.5796075665327813e-05,
      "loss": 0.5389,
      "step": 9402
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2047520875930786,
      "learning_rate": 1.579516573778792e-05,
      "loss": 0.7628,
      "step": 9403
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3212485313415527,
      "learning_rate": 1.579425573799752e-05,
      "loss": 0.7745,
      "step": 9404
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2385774850845337,
      "learning_rate": 1.5793345665967955e-05,
      "loss": 0.6909,
      "step": 9405
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4095112085342407,
      "learning_rate": 1.579243552171058e-05,
      "loss": 0.7048,
      "step": 9406
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.614519715309143,
      "learning_rate": 1.5791525305236737e-05,
      "loss": 0.8235,
      "step": 9407
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5532276630401611,
      "learning_rate": 1.579061501655777e-05,
      "loss": 0.7961,
      "step": 9408
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8082727193832397,
      "learning_rate": 1.578970465568504e-05,
      "loss": 0.7011,
      "step": 9409
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3709636926651,
      "learning_rate": 1.5788794222629878e-05,
      "loss": 0.542,
      "step": 9410
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9749977588653564,
      "learning_rate": 1.5787883717403657e-05,
      "loss": 0.7532,
      "step": 9411
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7301523685455322,
      "learning_rate": 1.578697314001771e-05,
      "loss": 0.6398,
      "step": 9412
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6255077123641968,
      "learning_rate": 1.57860624904834e-05,
      "loss": 0.728,
      "step": 9413
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4685338735580444,
      "learning_rate": 1.5785151768812076e-05,
      "loss": 0.8552,
      "step": 9414
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6577743291854858,
      "learning_rate": 1.5784240975015092e-05,
      "loss": 0.8167,
      "step": 9415
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5173977613449097,
      "learning_rate": 1.5783330109103804e-05,
      "loss": 0.6755,
      "step": 9416
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.379751443862915,
      "learning_rate": 1.5782419171089575e-05,
      "loss": 0.5751,
      "step": 9417
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6427396535873413,
      "learning_rate": 1.5781508160983748e-05,
      "loss": 0.7045,
      "step": 9418
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5292741060256958,
      "learning_rate": 1.5780597078797694e-05,
      "loss": 0.7201,
      "step": 9419
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5611240863800049,
      "learning_rate": 1.5779685924542764e-05,
      "loss": 0.7278,
      "step": 9420
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4928679466247559,
      "learning_rate": 1.577877469823032e-05,
      "loss": 0.7761,
      "step": 9421
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.450667381286621,
      "learning_rate": 1.5777863399871727e-05,
      "loss": 0.6863,
      "step": 9422
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5204933881759644,
      "learning_rate": 1.577695202947834e-05,
      "loss": 0.5892,
      "step": 9423
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.049184799194336,
      "learning_rate": 1.5776040587061523e-05,
      "loss": 0.8905,
      "step": 9424
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.482041835784912,
      "learning_rate": 1.577512907263264e-05,
      "loss": 0.5947,
      "step": 9425
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3068344593048096,
      "learning_rate": 1.5774217486203056e-05,
      "loss": 0.6108,
      "step": 9426
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4807196855545044,
      "learning_rate": 1.5773305827784134e-05,
      "loss": 0.767,
      "step": 9427
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2732465267181396,
      "learning_rate": 1.5772394097387242e-05,
      "loss": 0.6186,
      "step": 9428
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.509397029876709,
      "learning_rate": 1.5771482295023748e-05,
      "loss": 0.6951,
      "step": 9429
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4876596927642822,
      "learning_rate": 1.5770570420705014e-05,
      "loss": 0.7521,
      "step": 9430
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2987923622131348,
      "learning_rate": 1.5769658474442416e-05,
      "loss": 0.6676,
      "step": 9431
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3918907642364502,
      "learning_rate": 1.576874645624732e-05,
      "loss": 0.6829,
      "step": 9432
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4645750522613525,
      "learning_rate": 1.5767834366131097e-05,
      "loss": 0.786,
      "step": 9433
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4931104183197021,
      "learning_rate": 1.5766922204105118e-05,
      "loss": 0.6999,
      "step": 9434
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1692146062850952,
      "learning_rate": 1.5766009970180752e-05,
      "loss": 0.676,
      "step": 9435
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9273905754089355,
      "learning_rate": 1.576509766436938e-05,
      "loss": 0.7302,
      "step": 9436
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4811418056488037,
      "learning_rate": 1.5764185286682366e-05,
      "loss": 0.6823,
      "step": 9437
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5623071193695068,
      "learning_rate": 1.5763272837131097e-05,
      "loss": 0.6742,
      "step": 9438
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2782827615737915,
      "learning_rate": 1.5762360315726937e-05,
      "loss": 0.6383,
      "step": 9439
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2401320934295654,
      "learning_rate": 1.5761447722481272e-05,
      "loss": 0.5768,
      "step": 9440
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.379156470298767,
      "learning_rate": 1.5760535057405474e-05,
      "loss": 0.8032,
      "step": 9441
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3409075736999512,
      "learning_rate": 1.5759622320510922e-05,
      "loss": 0.6348,
      "step": 9442
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7129393815994263,
      "learning_rate": 1.5758709511809e-05,
      "loss": 0.6772,
      "step": 9443
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4252465963363647,
      "learning_rate": 1.5757796631311083e-05,
      "loss": 0.7564,
      "step": 9444
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4251430034637451,
      "learning_rate": 1.5756883679028555e-05,
      "loss": 0.6082,
      "step": 9445
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2300097942352295,
      "learning_rate": 1.5755970654972798e-05,
      "loss": 0.7131,
      "step": 9446
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3763943910598755,
      "learning_rate": 1.5755057559155193e-05,
      "loss": 0.4572,
      "step": 9447
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6190978288650513,
      "learning_rate": 1.5754144391587125e-05,
      "loss": 0.8116,
      "step": 9448
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4376133680343628,
      "learning_rate": 1.5753231152279978e-05,
      "loss": 0.6984,
      "step": 9449
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.53496515750885,
      "learning_rate": 1.5752317841245145e-05,
      "loss": 0.7145,
      "step": 9450
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5615071058273315,
      "learning_rate": 1.5751404458494e-05,
      "loss": 0.7074,
      "step": 9451
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3823521137237549,
      "learning_rate": 1.575049100403794e-05,
      "loss": 0.5835,
      "step": 9452
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5917855501174927,
      "learning_rate": 1.5749577477888348e-05,
      "loss": 0.6214,
      "step": 9453
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.389082670211792,
      "learning_rate": 1.574866388005662e-05,
      "loss": 0.6299,
      "step": 9454
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6185574531555176,
      "learning_rate": 1.5747750210554137e-05,
      "loss": 0.6709,
      "step": 9455
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5123567581176758,
      "learning_rate": 1.5746836469392298e-05,
      "loss": 0.815,
      "step": 9456
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3510172367095947,
      "learning_rate": 1.574592265658249e-05,
      "loss": 0.7647,
      "step": 9457
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3701038360595703,
      "learning_rate": 1.5745008772136106e-05,
      "loss": 0.6204,
      "step": 9458
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7379646301269531,
      "learning_rate": 1.5744094816064543e-05,
      "loss": 0.709,
      "step": 9459
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4059698581695557,
      "learning_rate": 1.5743180788379193e-05,
      "loss": 0.6597,
      "step": 9460
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.781441569328308,
      "learning_rate": 1.5742266689091454e-05,
      "loss": 0.8427,
      "step": 9461
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1927564144134521,
      "learning_rate": 1.574135251821272e-05,
      "loss": 0.6307,
      "step": 9462
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2893942594528198,
      "learning_rate": 1.574043827575439e-05,
      "loss": 0.6739,
      "step": 9463
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6609621047973633,
      "learning_rate": 1.5739523961727863e-05,
      "loss": 0.6633,
      "step": 9464
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.481774091720581,
      "learning_rate": 1.5738609576144533e-05,
      "loss": 0.6952,
      "step": 9465
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.37339186668396,
      "learning_rate": 1.5737695119015806e-05,
      "loss": 0.65,
      "step": 9466
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5387794971466064,
      "learning_rate": 1.573678059035308e-05,
      "loss": 0.695,
      "step": 9467
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5083715915679932,
      "learning_rate": 1.5735865990167757e-05,
      "loss": 0.7275,
      "step": 9468
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3740190267562866,
      "learning_rate": 1.573495131847124e-05,
      "loss": 0.6383,
      "step": 9469
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4584519863128662,
      "learning_rate": 1.5734036575274932e-05,
      "loss": 0.5388,
      "step": 9470
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5115416049957275,
      "learning_rate": 1.5733121760590234e-05,
      "loss": 0.6597,
      "step": 9471
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3051233291625977,
      "learning_rate": 1.5732206874428562e-05,
      "loss": 0.6781,
      "step": 9472
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.273045539855957,
      "learning_rate": 1.573129191680131e-05,
      "loss": 0.7481,
      "step": 9473
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.748082160949707,
      "learning_rate": 1.5730376887719896e-05,
      "loss": 0.5731,
      "step": 9474
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.632460355758667,
      "learning_rate": 1.5729461787195717e-05,
      "loss": 0.6711,
      "step": 9475
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4010355472564697,
      "learning_rate": 1.572854661524019e-05,
      "loss": 0.7437,
      "step": 9476
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5666521787643433,
      "learning_rate": 1.5727631371864723e-05,
      "loss": 0.604,
      "step": 9477
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4975742101669312,
      "learning_rate": 1.5726716057080724e-05,
      "loss": 0.7082,
      "step": 9478
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.357694149017334,
      "learning_rate": 1.5725800670899606e-05,
      "loss": 0.7053,
      "step": 9479
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2482988834381104,
      "learning_rate": 1.5724885213332784e-05,
      "loss": 0.7412,
      "step": 9480
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.368977427482605,
      "learning_rate": 1.572396968439167e-05,
      "loss": 0.8065,
      "step": 9481
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.618240237236023,
      "learning_rate": 1.5723054084087674e-05,
      "loss": 0.7921,
      "step": 9482
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2792221307754517,
      "learning_rate": 1.572213841243222e-05,
      "loss": 0.6347,
      "step": 9483
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6463217735290527,
      "learning_rate": 1.5721222669436713e-05,
      "loss": 0.6632,
      "step": 9484
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6568825244903564,
      "learning_rate": 1.572030685511258e-05,
      "loss": 0.8602,
      "step": 9485
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8877284526824951,
      "learning_rate": 1.571939096947123e-05,
      "loss": 0.7111,
      "step": 9486
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4830894470214844,
      "learning_rate": 1.571847501252409e-05,
      "loss": 0.72,
      "step": 9487
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4290006160736084,
      "learning_rate": 1.5717558984282573e-05,
      "loss": 0.7938,
      "step": 9488
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3578760623931885,
      "learning_rate": 1.5716642884758104e-05,
      "loss": 0.6533,
      "step": 9489
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2789500951766968,
      "learning_rate": 1.57157267139621e-05,
      "loss": 0.728,
      "step": 9490
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4938318729400635,
      "learning_rate": 1.571481047190599e-05,
      "loss": 0.6891,
      "step": 9491
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4857381582260132,
      "learning_rate": 1.571389415860119e-05,
      "loss": 0.6961,
      "step": 9492
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5522711277008057,
      "learning_rate": 1.5712977774059125e-05,
      "loss": 0.9033,
      "step": 9493
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.254264235496521,
      "learning_rate": 1.5712061318291226e-05,
      "loss": 0.7261,
      "step": 9494
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7027251720428467,
      "learning_rate": 1.5711144791308913e-05,
      "loss": 0.6899,
      "step": 9495
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7084918022155762,
      "learning_rate": 1.5710228193123614e-05,
      "loss": 0.6445,
      "step": 9496
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5108494758605957,
      "learning_rate": 1.5709311523746755e-05,
      "loss": 0.7874,
      "step": 9497
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5242459774017334,
      "learning_rate": 1.5708394783189767e-05,
      "loss": 0.5942,
      "step": 9498
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4475274085998535,
      "learning_rate": 1.570747797146408e-05,
      "loss": 0.7078,
      "step": 9499
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.498760461807251,
      "learning_rate": 1.5706561088581126e-05,
      "loss": 0.6943,
      "step": 9500
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.512840747833252,
      "learning_rate": 1.570564413455233e-05,
      "loss": 0.8514,
      "step": 9501
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4541131258010864,
      "learning_rate": 1.5704727109389124e-05,
      "loss": 0.8566,
      "step": 9502
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.458014726638794,
      "learning_rate": 1.570381001310295e-05,
      "loss": 0.731,
      "step": 9503
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3716496229171753,
      "learning_rate": 1.570289284570523e-05,
      "loss": 0.6995,
      "step": 9504
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6296918392181396,
      "learning_rate": 1.5701975607207408e-05,
      "loss": 0.7459,
      "step": 9505
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.387476921081543,
      "learning_rate": 1.5701058297620916e-05,
      "loss": 0.833,
      "step": 9506
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.475502371788025,
      "learning_rate": 1.5700140916957186e-05,
      "loss": 0.8093,
      "step": 9507
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5263408422470093,
      "learning_rate": 1.5699223465227666e-05,
      "loss": 0.7305,
      "step": 9508
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4876415729522705,
      "learning_rate": 1.5698305942443784e-05,
      "loss": 0.6158,
      "step": 9509
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3278117179870605,
      "learning_rate": 1.5697388348616983e-05,
      "loss": 0.6436,
      "step": 9510
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5868905782699585,
      "learning_rate": 1.5696470683758703e-05,
      "loss": 0.6152,
      "step": 9511
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4684383869171143,
      "learning_rate": 1.569555294788039e-05,
      "loss": 0.6769,
      "step": 9512
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.330797791481018,
      "learning_rate": 1.5694635140993472e-05,
      "loss": 0.6897,
      "step": 9513
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4396390914916992,
      "learning_rate": 1.5693717263109404e-05,
      "loss": 0.554,
      "step": 9514
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4421073198318481,
      "learning_rate": 1.569279931423963e-05,
      "loss": 0.6543,
      "step": 9515
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4691510200500488,
      "learning_rate": 1.5691881294395584e-05,
      "loss": 0.8949,
      "step": 9516
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3960734605789185,
      "learning_rate": 1.569096320358872e-05,
      "loss": 0.7104,
      "step": 9517
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6225252151489258,
      "learning_rate": 1.5690045041830485e-05,
      "loss": 0.5868,
      "step": 9518
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2960113286972046,
      "learning_rate": 1.568912680913232e-05,
      "loss": 0.6802,
      "step": 9519
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3892155885696411,
      "learning_rate": 1.5688208505505673e-05,
      "loss": 0.6465,
      "step": 9520
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.9097493886947632,
      "learning_rate": 1.5687290130961997e-05,
      "loss": 0.7112,
      "step": 9521
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.611721158027649,
      "learning_rate": 1.5686371685512743e-05,
      "loss": 0.7794,
      "step": 9522
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.230599284172058,
      "learning_rate": 1.5685453169169358e-05,
      "loss": 0.5226,
      "step": 9523
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.501954197883606,
      "learning_rate": 1.5684534581943292e-05,
      "loss": 0.8751,
      "step": 9524
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4547003507614136,
      "learning_rate": 1.5683615923846002e-05,
      "loss": 0.8133,
      "step": 9525
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3867452144622803,
      "learning_rate": 1.568269719488894e-05,
      "loss": 0.5847,
      "step": 9526
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5698938369750977,
      "learning_rate": 1.5681778395083556e-05,
      "loss": 0.9177,
      "step": 9527
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.577233076095581,
      "learning_rate": 1.568085952444131e-05,
      "loss": 0.6582,
      "step": 9528
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.401933193206787,
      "learning_rate": 1.5679940582973656e-05,
      "loss": 0.7259,
      "step": 9529
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2007542848587036,
      "learning_rate": 1.567902157069205e-05,
      "loss": 0.7268,
      "step": 9530
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.455580472946167,
      "learning_rate": 1.567810248760795e-05,
      "loss": 0.684,
      "step": 9531
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3766200542449951,
      "learning_rate": 1.5677183333732817e-05,
      "loss": 0.6827,
      "step": 9532
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7194154262542725,
      "learning_rate": 1.5676264109078107e-05,
      "loss": 0.6476,
      "step": 9533
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3497834205627441,
      "learning_rate": 1.5675344813655288e-05,
      "loss": 0.6437,
      "step": 9534
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4908066987991333,
      "learning_rate": 1.567442544747581e-05,
      "loss": 0.7434,
      "step": 9535
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2330573797225952,
      "learning_rate": 1.5673506010551135e-05,
      "loss": 0.6657,
      "step": 9536
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5422816276550293,
      "learning_rate": 1.5672586502892735e-05,
      "loss": 0.6565,
      "step": 9537
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5767951011657715,
      "learning_rate": 1.5671666924512076e-05,
      "loss": 0.6345,
      "step": 9538
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3451372385025024,
      "learning_rate": 1.5670747275420607e-05,
      "loss": 0.739,
      "step": 9539
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4683852195739746,
      "learning_rate": 1.566982755562981e-05,
      "loss": 0.671,
      "step": 9540
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5812103748321533,
      "learning_rate": 1.5668907765151144e-05,
      "loss": 0.5742,
      "step": 9541
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2742915153503418,
      "learning_rate": 1.5667987903996073e-05,
      "loss": 0.7592,
      "step": 9542
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6405930519104004,
      "learning_rate": 1.5667067972176072e-05,
      "loss": 0.5313,
      "step": 9543
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2478240728378296,
      "learning_rate": 1.566614796970261e-05,
      "loss": 0.6997,
      "step": 9544
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3820921182632446,
      "learning_rate": 1.566522789658715e-05,
      "loss": 0.8343,
      "step": 9545
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.318107008934021,
      "learning_rate": 1.566430775284117e-05,
      "loss": 0.5947,
      "step": 9546
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6222491264343262,
      "learning_rate": 1.5663387538476142e-05,
      "loss": 0.9035,
      "step": 9547
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7181884050369263,
      "learning_rate": 1.566246725350353e-05,
      "loss": 0.7897,
      "step": 9548
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7226775884628296,
      "learning_rate": 1.5661546897934818e-05,
      "loss": 0.6219,
      "step": 9549
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3052575588226318,
      "learning_rate": 1.5660626471781473e-05,
      "loss": 0.6233,
      "step": 9550
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.736014485359192,
      "learning_rate": 1.5659705975054973e-05,
      "loss": 0.5369,
      "step": 9551
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2524497509002686,
      "learning_rate": 1.5658785407766796e-05,
      "loss": 0.5551,
      "step": 9552
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4695196151733398,
      "learning_rate": 1.565786476992842e-05,
      "loss": 0.857,
      "step": 9553
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5064138174057007,
      "learning_rate": 1.5656944061551316e-05,
      "loss": 0.7278,
      "step": 9554
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4458051919937134,
      "learning_rate": 1.5656023282646966e-05,
      "loss": 0.6436,
      "step": 9555
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.313835859298706,
      "learning_rate": 1.5655102433226853e-05,
      "loss": 0.7397,
      "step": 9556
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5259907245635986,
      "learning_rate": 1.5654181513302455e-05,
      "loss": 0.8116,
      "step": 9557
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.270226240158081,
      "learning_rate": 1.5653260522885255e-05,
      "loss": 0.7375,
      "step": 9558
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3274413347244263,
      "learning_rate": 1.5652339461986734e-05,
      "loss": 0.804,
      "step": 9559
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3700159788131714,
      "learning_rate": 1.5651418330618376e-05,
      "loss": 0.6237,
      "step": 9560
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5592916011810303,
      "learning_rate": 1.5650497128791665e-05,
      "loss": 0.7312,
      "step": 9561
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6563102006912231,
      "learning_rate": 1.564957585651808e-05,
      "loss": 0.932,
      "step": 9562
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.475277304649353,
      "learning_rate": 1.5648654513809117e-05,
      "loss": 0.7703,
      "step": 9563
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4213840961456299,
      "learning_rate": 1.564773310067626e-05,
      "loss": 0.5901,
      "step": 9564
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3247716426849365,
      "learning_rate": 1.564681161713099e-05,
      "loss": 0.6637,
      "step": 9565
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3748040199279785,
      "learning_rate": 1.5645890063184804e-05,
      "loss": 0.6329,
      "step": 9566
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.764640212059021,
      "learning_rate": 1.5644968438849185e-05,
      "loss": 0.669,
      "step": 9567
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4625961780548096,
      "learning_rate": 1.5644046744135626e-05,
      "loss": 0.6667,
      "step": 9568
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.680308222770691,
      "learning_rate": 1.5643124979055617e-05,
      "loss": 0.6901,
      "step": 9569
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.422871470451355,
      "learning_rate": 1.5642203143620652e-05,
      "loss": 0.5564,
      "step": 9570
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3026816844940186,
      "learning_rate": 1.5641281237842224e-05,
      "loss": 0.6612,
      "step": 9571
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4159538745880127,
      "learning_rate": 1.5640359261731823e-05,
      "loss": 0.6837,
      "step": 9572
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8337990045547485,
      "learning_rate": 1.5639437215300945e-05,
      "loss": 0.8274,
      "step": 9573
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.556151270866394,
      "learning_rate": 1.563851509856109e-05,
      "loss": 0.7438,
      "step": 9574
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.453681230545044,
      "learning_rate": 1.5637592911523748e-05,
      "loss": 0.7393,
      "step": 9575
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6188222169876099,
      "learning_rate": 1.5636670654200425e-05,
      "loss": 0.6002,
      "step": 9576
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6779910326004028,
      "learning_rate": 1.5635748326602607e-05,
      "loss": 0.7488,
      "step": 9577
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7236709594726562,
      "learning_rate": 1.56348259287418e-05,
      "loss": 0.6584,
      "step": 9578
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7505830526351929,
      "learning_rate": 1.5633903460629508e-05,
      "loss": 0.5846,
      "step": 9579
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6129169464111328,
      "learning_rate": 1.563298092227722e-05,
      "loss": 0.8463,
      "step": 9580
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4073693752288818,
      "learning_rate": 1.563205831369645e-05,
      "loss": 0.6538,
      "step": 9581
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.661629319190979,
      "learning_rate": 1.5631135634898695e-05,
      "loss": 0.6542,
      "step": 9582
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4714289903640747,
      "learning_rate": 1.5630212885895454e-05,
      "loss": 0.7871,
      "step": 9583
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4303497076034546,
      "learning_rate": 1.562929006669824e-05,
      "loss": 0.7362,
      "step": 9584
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.508842945098877,
      "learning_rate": 1.5628367177318552e-05,
      "loss": 0.6121,
      "step": 9585
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4671210050582886,
      "learning_rate": 1.5627444217767894e-05,
      "loss": 0.7563,
      "step": 9586
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4203767776489258,
      "learning_rate": 1.5626521188057785e-05,
      "loss": 0.6685,
      "step": 9587
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3249850273132324,
      "learning_rate": 1.5625598088199717e-05,
      "loss": 0.6159,
      "step": 9588
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3786731958389282,
      "learning_rate": 1.562467491820521e-05,
      "loss": 0.6335,
      "step": 9589
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6887269020080566,
      "learning_rate": 1.562375167808577e-05,
      "loss": 0.7851,
      "step": 9590
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2955927848815918,
      "learning_rate": 1.5622828367852906e-05,
      "loss": 0.5521,
      "step": 9591
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5746124982833862,
      "learning_rate": 1.562190498751813e-05,
      "loss": 0.6542,
      "step": 9592
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6313090324401855,
      "learning_rate": 1.5620981537092956e-05,
      "loss": 0.6403,
      "step": 9593
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4624301195144653,
      "learning_rate": 1.5620058016588894e-05,
      "loss": 0.6681,
      "step": 9594
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3495361804962158,
      "learning_rate": 1.561913442601746e-05,
      "loss": 0.6886,
      "step": 9595
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.350226879119873,
      "learning_rate": 1.561821076539017e-05,
      "loss": 0.6033,
      "step": 9596
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.46714186668396,
      "learning_rate": 1.5617287034718534e-05,
      "loss": 0.6949,
      "step": 9597
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5929126739501953,
      "learning_rate": 1.5616363234014072e-05,
      "loss": 0.8231,
      "step": 9598
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.43132483959198,
      "learning_rate": 1.5615439363288304e-05,
      "loss": 0.7324,
      "step": 9599
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2335307598114014,
      "learning_rate": 1.5614515422552745e-05,
      "loss": 0.662,
      "step": 9600
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2975727319717407,
      "learning_rate": 1.5613591411818912e-05,
      "loss": 0.6853,
      "step": 9601
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5097006559371948,
      "learning_rate": 1.5612667331098332e-05,
      "loss": 0.6525,
      "step": 9602
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2082303762435913,
      "learning_rate": 1.561174318040252e-05,
      "loss": 0.6018,
      "step": 9603
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.526611089706421,
      "learning_rate": 1.5610818959743e-05,
      "loss": 0.6359,
      "step": 9604
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0973204374313354,
      "learning_rate": 1.5609894669131293e-05,
      "loss": 0.5926,
      "step": 9605
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3838435411453247,
      "learning_rate": 1.5608970308578924e-05,
      "loss": 0.7633,
      "step": 9606
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4439576864242554,
      "learning_rate": 1.560804587809742e-05,
      "loss": 0.89,
      "step": 9607
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6001521348953247,
      "learning_rate": 1.56071213776983e-05,
      "loss": 0.7157,
      "step": 9608
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4394375085830688,
      "learning_rate": 1.560619680739309e-05,
      "loss": 0.7203,
      "step": 9609
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1363816261291504,
      "learning_rate": 1.5605272167193327e-05,
      "loss": 0.6215,
      "step": 9610
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5049891471862793,
      "learning_rate": 1.560434745711053e-05,
      "loss": 0.644,
      "step": 9611
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.634708046913147,
      "learning_rate": 1.5603422677156225e-05,
      "loss": 0.6347,
      "step": 9612
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5964330434799194,
      "learning_rate": 1.5602497827341952e-05,
      "loss": 0.6833,
      "step": 9613
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7360179424285889,
      "learning_rate": 1.5601572907679236e-05,
      "loss": 0.7686,
      "step": 9614
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2685071229934692,
      "learning_rate": 1.5600647918179605e-05,
      "loss": 0.6741,
      "step": 9615
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4517518281936646,
      "learning_rate": 1.5599722858854592e-05,
      "loss": 0.72,
      "step": 9616
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.445407748222351,
      "learning_rate": 1.5598797729715735e-05,
      "loss": 0.7872,
      "step": 9617
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6527440547943115,
      "learning_rate": 1.5597872530774567e-05,
      "loss": 0.5919,
      "step": 9618
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6261250972747803,
      "learning_rate": 1.559694726204262e-05,
      "loss": 0.6953,
      "step": 9619
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.9543640613555908,
      "learning_rate": 1.559602192353143e-05,
      "loss": 0.8266,
      "step": 9620
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8560746908187866,
      "learning_rate": 1.5595096515252535e-05,
      "loss": 0.5859,
      "step": 9621
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5715341567993164,
      "learning_rate": 1.559417103721747e-05,
      "loss": 0.6314,
      "step": 9622
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4546082019805908,
      "learning_rate": 1.559324548943778e-05,
      "loss": 0.7762,
      "step": 9623
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.557654857635498,
      "learning_rate": 1.5592319871924994e-05,
      "loss": 0.652,
      "step": 9624
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5562992095947266,
      "learning_rate": 1.5591394184690657e-05,
      "loss": 0.7459,
      "step": 9625
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.489395022392273,
      "learning_rate": 1.5590468427746314e-05,
      "loss": 0.6699,
      "step": 9626
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.347249984741211,
      "learning_rate": 1.55895426011035e-05,
      "loss": 0.6896,
      "step": 9627
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3348280191421509,
      "learning_rate": 1.5588616704773763e-05,
      "loss": 0.6358,
      "step": 9628
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.489456057548523,
      "learning_rate": 1.558769073876864e-05,
      "loss": 0.6933,
      "step": 9629
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2741410732269287,
      "learning_rate": 1.558676470309968e-05,
      "loss": 0.6726,
      "step": 9630
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3680305480957031,
      "learning_rate": 1.558583859777843e-05,
      "loss": 0.602,
      "step": 9631
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4139037132263184,
      "learning_rate": 1.5584912422816435e-05,
      "loss": 0.8346,
      "step": 9632
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.715531587600708,
      "learning_rate": 1.5583986178225236e-05,
      "loss": 0.658,
      "step": 9633
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4544318914413452,
      "learning_rate": 1.5583059864016386e-05,
      "loss": 0.8102,
      "step": 9634
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5142399072647095,
      "learning_rate": 1.5582133480201434e-05,
      "loss": 0.7214,
      "step": 9635
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8552769422531128,
      "learning_rate": 1.558120702679193e-05,
      "loss": 0.7732,
      "step": 9636
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4093674421310425,
      "learning_rate": 1.5580280503799424e-05,
      "loss": 0.649,
      "step": 9637
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2178161144256592,
      "learning_rate": 1.5579353911235463e-05,
      "loss": 0.7501,
      "step": 9638
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2447078227996826,
      "learning_rate": 1.5578427249111604e-05,
      "loss": 0.7659,
      "step": 9639
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5296478271484375,
      "learning_rate": 1.55775005174394e-05,
      "loss": 0.7183,
      "step": 9640
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.554403305053711,
      "learning_rate": 1.5576573716230405e-05,
      "loss": 0.6576,
      "step": 9641
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7577581405639648,
      "learning_rate": 1.5575646845496168e-05,
      "loss": 0.6488,
      "step": 9642
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4894057512283325,
      "learning_rate": 1.557471990524825e-05,
      "loss": 0.6837,
      "step": 9643
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.516782522201538,
      "learning_rate": 1.5573792895498212e-05,
      "loss": 0.8134,
      "step": 9644
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.678183674812317,
      "learning_rate": 1.55728658162576e-05,
      "loss": 0.7219,
      "step": 9645
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7703238725662231,
      "learning_rate": 1.557193866753798e-05,
      "loss": 0.8216,
      "step": 9646
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1958832740783691,
      "learning_rate": 1.557101144935091e-05,
      "loss": 0.6334,
      "step": 9647
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6767064332962036,
      "learning_rate": 1.557008416170795e-05,
      "loss": 0.7326,
      "step": 9648
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4694641828536987,
      "learning_rate": 1.556915680462066e-05,
      "loss": 0.7377,
      "step": 9649
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6114200353622437,
      "learning_rate": 1.55682293781006e-05,
      "loss": 0.6509,
      "step": 9650
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3802179098129272,
      "learning_rate": 1.556730188215933e-05,
      "loss": 0.6314,
      "step": 9651
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.407871127128601,
      "learning_rate": 1.5566374316808425e-05,
      "loss": 0.6064,
      "step": 9652
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3666599988937378,
      "learning_rate": 1.556544668205944e-05,
      "loss": 0.6747,
      "step": 9653
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4017612934112549,
      "learning_rate": 1.5564518977923945e-05,
      "loss": 0.6625,
      "step": 9654
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5069209337234497,
      "learning_rate": 1.55635912044135e-05,
      "loss": 0.5668,
      "step": 9655
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7063732147216797,
      "learning_rate": 1.5562663361539674e-05,
      "loss": 0.8506,
      "step": 9656
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7122342586517334,
      "learning_rate": 1.5561735449314037e-05,
      "loss": 0.6235,
      "step": 9657
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3421578407287598,
      "learning_rate": 1.5560807467748156e-05,
      "loss": 0.7103,
      "step": 9658
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.38699209690094,
      "learning_rate": 1.5559879416853607e-05,
      "loss": 0.6337,
      "step": 9659
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4509872198104858,
      "learning_rate": 1.5558951296641947e-05,
      "loss": 0.7104,
      "step": 9660
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3353184461593628,
      "learning_rate": 1.5558023107124757e-05,
      "loss": 0.6117,
      "step": 9661
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3370763063430786,
      "learning_rate": 1.5557094848313608e-05,
      "loss": 0.6365,
      "step": 9662
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4229624271392822,
      "learning_rate": 1.555616652022007e-05,
      "loss": 0.7787,
      "step": 9663
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.458281397819519,
      "learning_rate": 1.555523812285572e-05,
      "loss": 0.5591,
      "step": 9664
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.742215633392334,
      "learning_rate": 1.555430965623213e-05,
      "loss": 0.6718,
      "step": 9665
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4890215396881104,
      "learning_rate": 1.555338112036088e-05,
      "loss": 0.699,
      "step": 9666
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.361211895942688,
      "learning_rate": 1.555245251525354e-05,
      "loss": 0.6129,
      "step": 9667
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.454671025276184,
      "learning_rate": 1.555152384092169e-05,
      "loss": 0.8977,
      "step": 9668
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4887065887451172,
      "learning_rate": 1.5550595097376913e-05,
      "loss": 0.8128,
      "step": 9669
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3961786031723022,
      "learning_rate": 1.554966628463078e-05,
      "loss": 0.6141,
      "step": 9670
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.387879729270935,
      "learning_rate": 1.5548737402694873e-05,
      "loss": 0.5383,
      "step": 9671
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5813294649124146,
      "learning_rate": 1.5547808451580778e-05,
      "loss": 0.6699,
      "step": 9672
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3652253150939941,
      "learning_rate": 1.554687943130007e-05,
      "loss": 0.5386,
      "step": 9673
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.527575135231018,
      "learning_rate": 1.5545950341864337e-05,
      "loss": 0.6113,
      "step": 9674
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.412766456604004,
      "learning_rate": 1.5545021183285155e-05,
      "loss": 0.634,
      "step": 9675
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3505624532699585,
      "learning_rate": 1.5544091955574115e-05,
      "loss": 0.5782,
      "step": 9676
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3139848709106445,
      "learning_rate": 1.5543162658742802e-05,
      "loss": 0.5892,
      "step": 9677
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3757296800613403,
      "learning_rate": 1.5542233292802795e-05,
      "loss": 0.5694,
      "step": 9678
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3615195751190186,
      "learning_rate": 1.554130385776569e-05,
      "loss": 0.6733,
      "step": 9679
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7997305393218994,
      "learning_rate": 1.5540374353643066e-05,
      "loss": 0.6622,
      "step": 9680
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5512292385101318,
      "learning_rate": 1.5539444780446518e-05,
      "loss": 0.6681,
      "step": 9681
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4006903171539307,
      "learning_rate": 1.553851513818763e-05,
      "loss": 0.6678,
      "step": 9682
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3602709770202637,
      "learning_rate": 1.5537585426878e-05,
      "loss": 0.5519,
      "step": 9683
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3651024103164673,
      "learning_rate": 1.553665564652921e-05,
      "loss": 0.5538,
      "step": 9684
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5595576763153076,
      "learning_rate": 1.5535725797152857e-05,
      "loss": 0.6111,
      "step": 9685
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.243086576461792,
      "learning_rate": 1.553479587876053e-05,
      "loss": 0.4587,
      "step": 9686
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4523910284042358,
      "learning_rate": 1.553386589136383e-05,
      "loss": 0.6224,
      "step": 9687
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1730763912200928,
      "learning_rate": 1.5532935834974346e-05,
      "loss": 0.5372,
      "step": 9688
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.584510326385498,
      "learning_rate": 1.5532005709603674e-05,
      "loss": 0.4679,
      "step": 9689
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.324876070022583,
      "learning_rate": 1.553107551526341e-05,
      "loss": 0.5684,
      "step": 9690
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.453543782234192,
      "learning_rate": 1.5530145251965148e-05,
      "loss": 0.4983,
      "step": 9691
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5865827798843384,
      "learning_rate": 1.5529214919720495e-05,
      "loss": 0.6961,
      "step": 9692
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2944446802139282,
      "learning_rate": 1.5528284518541042e-05,
      "loss": 0.3656,
      "step": 9693
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2936338186264038,
      "learning_rate": 1.552735404843839e-05,
      "loss": 0.554,
      "step": 9694
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5440900325775146,
      "learning_rate": 1.552642350942414e-05,
      "loss": 0.6443,
      "step": 9695
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7414766550064087,
      "learning_rate": 1.5525492901509895e-05,
      "loss": 0.5276,
      "step": 9696
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7206165790557861,
      "learning_rate": 1.552456222470726e-05,
      "loss": 0.622,
      "step": 9697
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6589319705963135,
      "learning_rate": 1.5523631479027828e-05,
      "loss": 0.5849,
      "step": 9698
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.9685455560684204,
      "learning_rate": 1.552270066448321e-05,
      "loss": 0.6184,
      "step": 9699
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.594920039176941,
      "learning_rate": 1.552176978108501e-05,
      "loss": 0.6214,
      "step": 9700
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5971447229385376,
      "learning_rate": 1.552083882884484e-05,
      "loss": 0.5926,
      "step": 9701
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.384495496749878,
      "learning_rate": 1.5519907807774294e-05,
      "loss": 0.5254,
      "step": 9702
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3960504531860352,
      "learning_rate": 1.5518976717884986e-05,
      "loss": 0.6444,
      "step": 9703
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7985471487045288,
      "learning_rate": 1.5518045559188525e-05,
      "loss": 0.5691,
      "step": 9704
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4226963520050049,
      "learning_rate": 1.551711433169652e-05,
      "loss": 0.568,
      "step": 9705
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6441150903701782,
      "learning_rate": 1.551618303542058e-05,
      "loss": 0.6,
      "step": 9706
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2747455835342407,
      "learning_rate": 1.551525167037231e-05,
      "loss": 0.6481,
      "step": 9707
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2490769624710083,
      "learning_rate": 1.5514320236563332e-05,
      "loss": 0.5272,
      "step": 9708
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6287732124328613,
      "learning_rate": 1.5513388734005254e-05,
      "loss": 0.6206,
      "step": 9709
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.230726718902588,
      "learning_rate": 1.5512457162709686e-05,
      "loss": 0.6154,
      "step": 9710
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5275318622589111,
      "learning_rate": 1.551152552268825e-05,
      "loss": 0.5469,
      "step": 9711
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.370959758758545,
      "learning_rate": 1.5510593813952554e-05,
      "loss": 0.5213,
      "step": 9712
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5600838661193848,
      "learning_rate": 1.5509662036514216e-05,
      "loss": 0.6031,
      "step": 9713
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2874605655670166,
      "learning_rate": 1.5508730190384856e-05,
      "loss": 0.5549,
      "step": 9714
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.578730821609497,
      "learning_rate": 1.5507798275576086e-05,
      "loss": 0.5052,
      "step": 9715
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.318410873413086,
      "learning_rate": 1.5506866292099528e-05,
      "loss": 0.6732,
      "step": 9716
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5480153560638428,
      "learning_rate": 1.5505934239966805e-05,
      "loss": 0.5917,
      "step": 9717
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2832154035568237,
      "learning_rate": 1.5505002119189533e-05,
      "loss": 0.5124,
      "step": 9718
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.608223557472229,
      "learning_rate": 1.550406992977933e-05,
      "loss": 0.5231,
      "step": 9719
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7919937372207642,
      "learning_rate": 1.5503137671747822e-05,
      "loss": 0.5041,
      "step": 9720
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.384486436843872,
      "learning_rate": 1.550220534510663e-05,
      "loss": 0.5426,
      "step": 9721
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6292468309402466,
      "learning_rate": 1.5501272949867382e-05,
      "loss": 0.5278,
      "step": 9722
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2665126323699951,
      "learning_rate": 1.5500340486041695e-05,
      "loss": 0.6142,
      "step": 9723
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.359176754951477,
      "learning_rate": 1.5499407953641204e-05,
      "loss": 0.6454,
      "step": 9724
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.711441159248352,
      "learning_rate": 1.5498475352677525e-05,
      "loss": 0.6775,
      "step": 9725
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5520806312561035,
      "learning_rate": 1.549754268316229e-05,
      "loss": 0.5462,
      "step": 9726
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4762071371078491,
      "learning_rate": 1.5496609945107133e-05,
      "loss": 0.6437,
      "step": 9727
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4567804336547852,
      "learning_rate": 1.5495677138523672e-05,
      "loss": 0.692,
      "step": 9728
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4383454322814941,
      "learning_rate": 1.549474426342354e-05,
      "loss": 0.6289,
      "step": 9729
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5292856693267822,
      "learning_rate": 1.549381131981837e-05,
      "loss": 0.6339,
      "step": 9730
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3883174657821655,
      "learning_rate": 1.5492878307719797e-05,
      "loss": 0.4634,
      "step": 9731
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3815466165542603,
      "learning_rate": 1.5491945227139443e-05,
      "loss": 0.6343,
      "step": 9732
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3790687322616577,
      "learning_rate": 1.549101207808895e-05,
      "loss": 0.564,
      "step": 9733
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4296574592590332,
      "learning_rate": 1.5490078860579947e-05,
      "loss": 0.508,
      "step": 9734
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2872633934020996,
      "learning_rate": 1.548914557462407e-05,
      "loss": 0.5006,
      "step": 9735
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4040101766586304,
      "learning_rate": 1.5488212220232952e-05,
      "loss": 0.541,
      "step": 9736
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3867257833480835,
      "learning_rate": 1.5487278797418236e-05,
      "loss": 0.5761,
      "step": 9737
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.447993278503418,
      "learning_rate": 1.5486345306191556e-05,
      "loss": 0.4949,
      "step": 9738
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4815706014633179,
      "learning_rate": 1.548541174656455e-05,
      "loss": 0.5673,
      "step": 9739
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.402565836906433,
      "learning_rate": 1.5484478118548855e-05,
      "loss": 0.5212,
      "step": 9740
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3106390237808228,
      "learning_rate": 1.548354442215611e-05,
      "loss": 0.6079,
      "step": 9741
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4661847352981567,
      "learning_rate": 1.548261065739796e-05,
      "loss": 0.6115,
      "step": 9742
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.219086766242981,
      "learning_rate": 1.5481676824286052e-05,
      "loss": 0.6119,
      "step": 9743
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5526065826416016,
      "learning_rate": 1.5480742922832015e-05,
      "loss": 0.6315,
      "step": 9744
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4808077812194824,
      "learning_rate": 1.54798089530475e-05,
      "loss": 0.5854,
      "step": 9745
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6510908603668213,
      "learning_rate": 1.547887491494415e-05,
      "loss": 0.6168,
      "step": 9746
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4611607789993286,
      "learning_rate": 1.547794080853361e-05,
      "loss": 0.4872,
      "step": 9747
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2820771932601929,
      "learning_rate": 1.547700663382753e-05,
      "loss": 0.6009,
      "step": 9748
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2903711795806885,
      "learning_rate": 1.5476072390837545e-05,
      "loss": 0.5897,
      "step": 9749
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.518282413482666,
      "learning_rate": 1.5475138079575318e-05,
      "loss": 0.5394,
      "step": 9750
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4707915782928467,
      "learning_rate": 1.547420370005248e-05,
      "loss": 0.5462,
      "step": 9751
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6575983762741089,
      "learning_rate": 1.5473269252280698e-05,
      "loss": 0.5442,
      "step": 9752
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.302717924118042,
      "learning_rate": 1.547233473627161e-05,
      "loss": 0.5279,
      "step": 9753
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4365705251693726,
      "learning_rate": 1.5471400152036873e-05,
      "loss": 0.548,
      "step": 9754
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4959381818771362,
      "learning_rate": 1.5470465499588138e-05,
      "loss": 0.3943,
      "step": 9755
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.448483943939209,
      "learning_rate": 1.5469530778937052e-05,
      "loss": 0.5125,
      "step": 9756
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5892796516418457,
      "learning_rate": 1.5468595990095276e-05,
      "loss": 0.6351,
      "step": 9757
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.8408050537109375,
      "learning_rate": 1.546766113307446e-05,
      "loss": 0.5859,
      "step": 9758
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6488878726959229,
      "learning_rate": 1.546672620788626e-05,
      "loss": 0.549,
      "step": 9759
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4341936111450195,
      "learning_rate": 1.5465791214542335e-05,
      "loss": 0.4452,
      "step": 9760
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2370846271514893,
      "learning_rate": 1.546485615305434e-05,
      "loss": 0.5341,
      "step": 9761
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.523469090461731,
      "learning_rate": 1.5463921023433926e-05,
      "loss": 0.5195,
      "step": 9762
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3359607458114624,
      "learning_rate": 1.5462985825692762e-05,
      "loss": 0.5979,
      "step": 9763
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3766900300979614,
      "learning_rate": 1.5462050559842502e-05,
      "loss": 0.6051,
      "step": 9764
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2748692035675049,
      "learning_rate": 1.5461115225894808e-05,
      "loss": 0.5416,
      "step": 9765
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3105952739715576,
      "learning_rate": 1.5460179823861342e-05,
      "loss": 0.5268,
      "step": 9766
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5092393159866333,
      "learning_rate": 1.5459244353753762e-05,
      "loss": 0.5155,
      "step": 9767
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7419214248657227,
      "learning_rate": 1.5458308815583735e-05,
      "loss": 0.4491,
      "step": 9768
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.444890022277832,
      "learning_rate": 1.5457373209362927e-05,
      "loss": 0.5538,
      "step": 9769
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5052440166473389,
      "learning_rate": 1.5456437535102996e-05,
      "loss": 0.6395,
      "step": 9770
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.191077470779419,
      "learning_rate": 1.545550179281561e-05,
      "loss": 0.4131,
      "step": 9771
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4107508659362793,
      "learning_rate": 1.5454565982512435e-05,
      "loss": 0.4881,
      "step": 9772
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6727228164672852,
      "learning_rate": 1.5453630104205137e-05,
      "loss": 0.622,
      "step": 9773
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7965587377548218,
      "learning_rate": 1.545269415790539e-05,
      "loss": 0.5924,
      "step": 9774
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3242806196212769,
      "learning_rate": 1.5451758143624853e-05,
      "loss": 0.5968,
      "step": 9775
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3301775455474854,
      "learning_rate": 1.5450822061375202e-05,
      "loss": 0.5266,
      "step": 9776
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5027581453323364,
      "learning_rate": 1.5449885911168112e-05,
      "loss": 0.6174,
      "step": 9777
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4743423461914062,
      "learning_rate": 1.5448949693015243e-05,
      "loss": 0.5007,
      "step": 9778
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3498117923736572,
      "learning_rate": 1.5448013406928276e-05,
      "loss": 0.5912,
      "step": 9779
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6318333148956299,
      "learning_rate": 1.544707705291888e-05,
      "loss": 0.6472,
      "step": 9780
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.335598111152649,
      "learning_rate": 1.544614063099873e-05,
      "loss": 0.4857,
      "step": 9781
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.411885142326355,
      "learning_rate": 1.54452041411795e-05,
      "loss": 0.4406,
      "step": 9782
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5643458366394043,
      "learning_rate": 1.5444267583472866e-05,
      "loss": 0.6296,
      "step": 9783
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6649694442749023,
      "learning_rate": 1.5443330957890506e-05,
      "loss": 0.519,
      "step": 9784
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6247307062149048,
      "learning_rate": 1.5442394264444096e-05,
      "loss": 0.5902,
      "step": 9785
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4131321907043457,
      "learning_rate": 1.5441457503145313e-05,
      "loss": 0.5359,
      "step": 9786
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4319686889648438,
      "learning_rate": 1.5440520674005838e-05,
      "loss": 0.4879,
      "step": 9787
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.7792105674743652,
      "learning_rate": 1.5439583777037347e-05,
      "loss": 0.6314,
      "step": 9788
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4958606958389282,
      "learning_rate": 1.5438646812251525e-05,
      "loss": 0.53,
      "step": 9789
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3646297454833984,
      "learning_rate": 1.5437709779660053e-05,
      "loss": 0.5421,
      "step": 9790
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5741477012634277,
      "learning_rate": 1.543677267927461e-05,
      "loss": 0.4974,
      "step": 9791
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.524835228919983,
      "learning_rate": 1.543583551110688e-05,
      "loss": 0.556,
      "step": 9792
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3957923650741577,
      "learning_rate": 1.5434898275168554e-05,
      "loss": 0.5551,
      "step": 9793
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.85386323928833,
      "learning_rate": 1.543396097147131e-05,
      "loss": 0.5811,
      "step": 9794
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3157085180282593,
      "learning_rate": 1.5433023600026833e-05,
      "loss": 0.5287,
      "step": 9795
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4689202308654785,
      "learning_rate": 1.5432086160846813e-05,
      "loss": 0.5619,
      "step": 9796
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.289580225944519,
      "learning_rate": 1.5431148653942934e-05,
      "loss": 0.3871,
      "step": 9797
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.325275182723999,
      "learning_rate": 1.5430211079326885e-05,
      "loss": 0.5588,
      "step": 9798
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6025869846343994,
      "learning_rate": 1.542927343701036e-05,
      "loss": 0.6271,
      "step": 9799
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3046096563339233,
      "learning_rate": 1.5428335727005043e-05,
      "loss": 0.6338,
      "step": 9800
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.444231390953064,
      "learning_rate": 1.542739794932263e-05,
      "loss": 0.5412,
      "step": 9801
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5960272550582886,
      "learning_rate": 1.5426460103974808e-05,
      "loss": 0.6579,
      "step": 9802
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4884148836135864,
      "learning_rate": 1.5425522190973268e-05,
      "loss": 0.5539,
      "step": 9803
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.6494354009628296,
      "learning_rate": 1.5424584210329713e-05,
      "loss": 0.5578,
      "step": 9804
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2587051391601562,
      "learning_rate": 1.5423646162055822e-05,
      "loss": 0.5657,
      "step": 9805
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3475064039230347,
      "learning_rate": 1.5422708046163305e-05,
      "loss": 0.4692,
      "step": 9806
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4025863409042358,
      "learning_rate": 1.542176986266385e-05,
      "loss": 0.5355,
      "step": 9807
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3862648010253906,
      "learning_rate": 1.5420831611569153e-05,
      "loss": 0.5593,
      "step": 9808
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.299672245979309,
      "learning_rate": 1.5419893292890917e-05,
      "loss": 0.594,
      "step": 9809
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.3688359260559082,
      "learning_rate": 1.5418954906640834e-05,
      "loss": 0.4823,
      "step": 9810
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5705982446670532,
      "learning_rate": 1.5418016452830608e-05,
      "loss": 0.5408,
      "step": 9811
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.8935550451278687,
      "learning_rate": 1.541707793147194e-05,
      "loss": 0.7399,
      "step": 9812
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5639764070510864,
      "learning_rate": 1.5416139342576522e-05,
      "loss": 0.5099,
      "step": 9813
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5462085008621216,
      "learning_rate": 1.5415200686156065e-05,
      "loss": 0.6971,
      "step": 9814
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7391688823699951,
      "learning_rate": 1.541426196222227e-05,
      "loss": 0.6407,
      "step": 9815
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4661712646484375,
      "learning_rate": 1.541332317078684e-05,
      "loss": 0.5933,
      "step": 9816
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4485111236572266,
      "learning_rate": 1.5412384311861476e-05,
      "loss": 0.4884,
      "step": 9817
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4562840461730957,
      "learning_rate": 1.5411445385457883e-05,
      "loss": 0.5858,
      "step": 9818
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6502482891082764,
      "learning_rate": 1.541050639158777e-05,
      "loss": 0.4931,
      "step": 9819
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5252306461334229,
      "learning_rate": 1.5409567330262848e-05,
      "loss": 0.5931,
      "step": 9820
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.523521900177002,
      "learning_rate": 1.5408628201494817e-05,
      "loss": 0.5849,
      "step": 9821
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3751857280731201,
      "learning_rate": 1.5407689005295384e-05,
      "loss": 0.5477,
      "step": 9822
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.8746552467346191,
      "learning_rate": 1.540674974167627e-05,
      "loss": 0.6577,
      "step": 9823
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6705302000045776,
      "learning_rate": 1.540581041064917e-05,
      "loss": 0.6343,
      "step": 9824
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.404274344444275,
      "learning_rate": 1.540487101222581e-05,
      "loss": 0.5633,
      "step": 9825
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6194508075714111,
      "learning_rate": 1.5403931546417886e-05,
      "loss": 0.5386,
      "step": 9826
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3957610130310059,
      "learning_rate": 1.5402992013237125e-05,
      "loss": 0.6191,
      "step": 9827
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5128915309906006,
      "learning_rate": 1.5402052412695233e-05,
      "loss": 0.6188,
      "step": 9828
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4980113506317139,
      "learning_rate": 1.5401112744803928e-05,
      "loss": 0.6568,
      "step": 9829
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.48745596408844,
      "learning_rate": 1.540017300957492e-05,
      "loss": 0.5769,
      "step": 9830
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5011770725250244,
      "learning_rate": 1.539923320701993e-05,
      "loss": 0.5568,
      "step": 9831
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2840551137924194,
      "learning_rate": 1.5398293337150672e-05,
      "loss": 0.5166,
      "step": 9832
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4438029527664185,
      "learning_rate": 1.539735339997887e-05,
      "loss": 0.4927,
      "step": 9833
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.535107135772705,
      "learning_rate": 1.539641339551623e-05,
      "loss": 0.5031,
      "step": 9834
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5238380432128906,
      "learning_rate": 1.5395473323774483e-05,
      "loss": 0.5798,
      "step": 9835
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3846336603164673,
      "learning_rate": 1.539453318476534e-05,
      "loss": 0.5546,
      "step": 9836
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2317841053009033,
      "learning_rate": 1.5393592978500534e-05,
      "loss": 0.5158,
      "step": 9837
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4324908256530762,
      "learning_rate": 1.5392652704991774e-05,
      "loss": 0.5209,
      "step": 9838
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5052034854888916,
      "learning_rate": 1.539171236425079e-05,
      "loss": 0.6263,
      "step": 9839
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4006056785583496,
      "learning_rate": 1.5390771956289307e-05,
      "loss": 0.5247,
      "step": 9840
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4130579233169556,
      "learning_rate": 1.5389831481119045e-05,
      "loss": 0.5539,
      "step": 9841
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6120706796646118,
      "learning_rate": 1.538889093875173e-05,
      "loss": 0.5287,
      "step": 9842
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5206083059310913,
      "learning_rate": 1.538795032919909e-05,
      "loss": 0.5351,
      "step": 9843
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4067469835281372,
      "learning_rate": 1.5387009652472852e-05,
      "loss": 0.5315,
      "step": 9844
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6489564180374146,
      "learning_rate": 1.5386068908584742e-05,
      "loss": 0.5573,
      "step": 9845
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5914777517318726,
      "learning_rate": 1.5385128097546492e-05,
      "loss": 0.5829,
      "step": 9846
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3610360622406006,
      "learning_rate": 1.5384187219369823e-05,
      "loss": 0.6267,
      "step": 9847
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6236964464187622,
      "learning_rate": 1.5383246274066473e-05,
      "loss": 0.5359,
      "step": 9848
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7921719551086426,
      "learning_rate": 1.5382305261648173e-05,
      "loss": 0.585,
      "step": 9849
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.365849256515503,
      "learning_rate": 1.5381364182126652e-05,
      "loss": 0.5764,
      "step": 9850
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.355440378189087,
      "learning_rate": 1.5380423035513644e-05,
      "loss": 0.5076,
      "step": 9851
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3445206880569458,
      "learning_rate": 1.537948182182088e-05,
      "loss": 0.6284,
      "step": 9852
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6497323513031006,
      "learning_rate": 1.5378540541060104e-05,
      "loss": 0.6607,
      "step": 9853
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.281129002571106,
      "learning_rate": 1.537759919324304e-05,
      "loss": 0.5028,
      "step": 9854
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.517315149307251,
      "learning_rate": 1.5376657778381425e-05,
      "loss": 0.6265,
      "step": 9855
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2811113595962524,
      "learning_rate": 1.5375716296487003e-05,
      "loss": 0.6172,
      "step": 9856
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4835352897644043,
      "learning_rate": 1.537477474757151e-05,
      "loss": 0.7275,
      "step": 9857
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.479112982749939,
      "learning_rate": 1.537383313164668e-05,
      "loss": 0.5981,
      "step": 9858
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.258629322052002,
      "learning_rate": 1.5372891448724257e-05,
      "loss": 0.5792,
      "step": 9859
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2941733598709106,
      "learning_rate": 1.5371949698815976e-05,
      "loss": 0.622,
      "step": 9860
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3444664478302002,
      "learning_rate": 1.537100788193359e-05,
      "loss": 0.5302,
      "step": 9861
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4452687501907349,
      "learning_rate": 1.5370065998088827e-05,
      "loss": 0.5652,
      "step": 9862
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6305609941482544,
      "learning_rate": 1.5369124047293435e-05,
      "loss": 0.5363,
      "step": 9863
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.364989161491394,
      "learning_rate": 1.5368182029559157e-05,
      "loss": 0.5755,
      "step": 9864
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6489876508712769,
      "learning_rate": 1.5367239944897746e-05,
      "loss": 0.5232,
      "step": 9865
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6003646850585938,
      "learning_rate": 1.5366297793320932e-05,
      "loss": 0.6017,
      "step": 9866
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7650164365768433,
      "learning_rate": 1.5365355574840472e-05,
      "loss": 0.5104,
      "step": 9867
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.420937418937683,
      "learning_rate": 1.5364413289468113e-05,
      "loss": 0.6046,
      "step": 9868
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3972023725509644,
      "learning_rate": 1.5363470937215597e-05,
      "loss": 0.5344,
      "step": 9869
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5963382720947266,
      "learning_rate": 1.5362528518094676e-05,
      "loss": 0.6756,
      "step": 9870
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7995704412460327,
      "learning_rate": 1.53615860321171e-05,
      "loss": 0.5486,
      "step": 9871
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5034751892089844,
      "learning_rate": 1.536064347929462e-05,
      "loss": 0.4521,
      "step": 9872
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.357507586479187,
      "learning_rate": 1.535970085963898e-05,
      "loss": 0.6132,
      "step": 9873
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5700963735580444,
      "learning_rate": 1.5358758173161942e-05,
      "loss": 0.598,
      "step": 9874
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6627169847488403,
      "learning_rate": 1.5357815419875257e-05,
      "loss": 0.6346,
      "step": 9875
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3735439777374268,
      "learning_rate": 1.5356872599790673e-05,
      "loss": 0.6065,
      "step": 9876
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3815193176269531,
      "learning_rate": 1.5355929712919946e-05,
      "loss": 0.5449,
      "step": 9877
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4702590703964233,
      "learning_rate": 1.5354986759274838e-05,
      "loss": 0.5178,
      "step": 9878
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3599201440811157,
      "learning_rate": 1.5354043738867094e-05,
      "loss": 0.5972,
      "step": 9879
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5281710624694824,
      "learning_rate": 1.5353100651708485e-05,
      "loss": 0.6069,
      "step": 9880
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.535468578338623,
      "learning_rate": 1.5352157497810754e-05,
      "loss": 0.6096,
      "step": 9881
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6135690212249756,
      "learning_rate": 1.535121427718567e-05,
      "loss": 0.5907,
      "step": 9882
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5588459968566895,
      "learning_rate": 1.5350270989844988e-05,
      "loss": 0.6573,
      "step": 9883
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4026167392730713,
      "learning_rate": 1.534932763580047e-05,
      "loss": 0.6077,
      "step": 9884
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.494307041168213,
      "learning_rate": 1.5348384215063874e-05,
      "loss": 0.5414,
      "step": 9885
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2902909517288208,
      "learning_rate": 1.5347440727646968e-05,
      "loss": 0.5352,
      "step": 9886
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4074448347091675,
      "learning_rate": 1.534649717356151e-05,
      "loss": 0.6457,
      "step": 9887
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.33078932762146,
      "learning_rate": 1.5345553552819264e-05,
      "loss": 0.4999,
      "step": 9888
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4522590637207031,
      "learning_rate": 1.5344609865431997e-05,
      "loss": 0.508,
      "step": 9889
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5080822706222534,
      "learning_rate": 1.534366611141147e-05,
      "loss": 0.5087,
      "step": 9890
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6119098663330078,
      "learning_rate": 1.5342722290769455e-05,
      "loss": 0.5087,
      "step": 9891
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5301339626312256,
      "learning_rate": 1.5341778403517717e-05,
      "loss": 0.5954,
      "step": 9892
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.223952293395996,
      "learning_rate": 1.534083444966802e-05,
      "loss": 0.5444,
      "step": 9893
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6483585834503174,
      "learning_rate": 1.5339890429232135e-05,
      "loss": 0.4123,
      "step": 9894
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5569772720336914,
      "learning_rate": 1.533894634222183e-05,
      "loss": 0.66,
      "step": 9895
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.7599564790725708,
      "learning_rate": 1.5338002188648884e-05,
      "loss": 0.5986,
      "step": 9896
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2937480211257935,
      "learning_rate": 1.5337057968525056e-05,
      "loss": 0.5867,
      "step": 9897
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3446375131607056,
      "learning_rate": 1.5336113681862125e-05,
      "loss": 0.6256,
      "step": 9898
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4356576204299927,
      "learning_rate": 1.533516932867186e-05,
      "loss": 0.6003,
      "step": 9899
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2831242084503174,
      "learning_rate": 1.5334224908966037e-05,
      "loss": 0.6335,
      "step": 9900
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5127977132797241,
      "learning_rate": 1.533328042275643e-05,
      "loss": 0.5905,
      "step": 9901
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4281445741653442,
      "learning_rate": 1.5332335870054816e-05,
      "loss": 0.4807,
      "step": 9902
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.6186254024505615,
      "learning_rate": 1.5331391250872967e-05,
      "loss": 0.6596,
      "step": 9903
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.493769645690918,
      "learning_rate": 1.5330446565222664e-05,
      "loss": 0.5765,
      "step": 9904
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.8546310663223267,
      "learning_rate": 1.532950181311568e-05,
      "loss": 0.7211,
      "step": 9905
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3622888326644897,
      "learning_rate": 1.53285569945638e-05,
      "loss": 0.5093,
      "step": 9906
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.5574102401733398,
      "learning_rate": 1.53276121095788e-05,
      "loss": 0.5993,
      "step": 9907
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.4995789527893066,
      "learning_rate": 1.5326667158172456e-05,
      "loss": 0.6919,
      "step": 9908
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3820453882217407,
      "learning_rate": 1.532572214035656e-05,
      "loss": 0.5236,
      "step": 9909
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.239852786064148,
      "learning_rate": 1.5324777056142882e-05,
      "loss": 0.6486,
      "step": 9910
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5012493133544922,
      "learning_rate": 1.532383190554321e-05,
      "loss": 0.51,
      "step": 9911
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3071266412734985,
      "learning_rate": 1.532288668856933e-05,
      "loss": 0.5408,
      "step": 9912
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2696171998977661,
      "learning_rate": 1.5321941405233022e-05,
      "loss": 0.5441,
      "step": 9913
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3166406154632568,
      "learning_rate": 1.5320996055546073e-05,
      "loss": 0.5548,
      "step": 9914
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5416887998580933,
      "learning_rate": 1.532005063952027e-05,
      "loss": 0.6092,
      "step": 9915
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.300991415977478,
      "learning_rate": 1.5319105157167394e-05,
      "loss": 0.5818,
      "step": 9916
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4014489650726318,
      "learning_rate": 1.5318159608499243e-05,
      "loss": 0.6197,
      "step": 9917
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3849396705627441,
      "learning_rate": 1.53172139935276e-05,
      "loss": 0.65,
      "step": 9918
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.703783631324768,
      "learning_rate": 1.5316268312264253e-05,
      "loss": 0.6086,
      "step": 9919
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4087804555892944,
      "learning_rate": 1.5315322564720992e-05,
      "loss": 0.5273,
      "step": 9920
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3824559450149536,
      "learning_rate": 1.531437675090961e-05,
      "loss": 0.4437,
      "step": 9921
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3045562505722046,
      "learning_rate": 1.5313430870841902e-05,
      "loss": 0.474,
      "step": 9922
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.776015281677246,
      "learning_rate": 1.5312484924529654e-05,
      "loss": 0.5538,
      "step": 9923
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.8141005039215088,
      "learning_rate": 1.5311538911984665e-05,
      "loss": 0.5725,
      "step": 9924
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5538939237594604,
      "learning_rate": 1.531059283321872e-05,
      "loss": 0.536,
      "step": 9925
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4926700592041016,
      "learning_rate": 1.5309646688243627e-05,
      "loss": 0.4946,
      "step": 9926
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7657406330108643,
      "learning_rate": 1.5308700477071177e-05,
      "loss": 0.6394,
      "step": 9927
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.472289800643921,
      "learning_rate": 1.5307754199713164e-05,
      "loss": 0.5169,
      "step": 9928
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3448081016540527,
      "learning_rate": 1.5306807856181384e-05,
      "loss": 0.5919,
      "step": 9929
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3756990432739258,
      "learning_rate": 1.5305861446487643e-05,
      "loss": 0.5872,
      "step": 9930
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2682026624679565,
      "learning_rate": 1.5304914970643733e-05,
      "loss": 0.5282,
      "step": 9931
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6355055570602417,
      "learning_rate": 1.5303968428661457e-05,
      "loss": 0.6685,
      "step": 9932
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5126827955245972,
      "learning_rate": 1.530302182055262e-05,
      "loss": 0.5505,
      "step": 9933
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3752466440200806,
      "learning_rate": 1.5302075146329013e-05,
      "loss": 0.5104,
      "step": 9934
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3191518783569336,
      "learning_rate": 1.530112840600245e-05,
      "loss": 0.6589,
      "step": 9935
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.671407699584961,
      "learning_rate": 1.5300181599584728e-05,
      "loss": 0.6441,
      "step": 9936
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3136098384857178,
      "learning_rate": 1.529923472708765e-05,
      "loss": 0.5109,
      "step": 9937
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2832202911376953,
      "learning_rate": 1.529828778852303e-05,
      "loss": 0.5783,
      "step": 9938
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5088151693344116,
      "learning_rate": 1.529734078390266e-05,
      "loss": 0.5042,
      "step": 9939
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.9000920057296753,
      "learning_rate": 1.5296393713238356e-05,
      "loss": 0.6269,
      "step": 9940
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6513216495513916,
      "learning_rate": 1.529544657654193e-05,
      "loss": 0.5591,
      "step": 9941
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7783820629119873,
      "learning_rate": 1.5294499373825175e-05,
      "loss": 0.5686,
      "step": 9942
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.393021583557129,
      "learning_rate": 1.5293552105099917e-05,
      "loss": 0.4822,
      "step": 9943
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.635272741317749,
      "learning_rate": 1.5292604770377954e-05,
      "loss": 0.6524,
      "step": 9944
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3587837219238281,
      "learning_rate": 1.52916573696711e-05,
      "loss": 0.5805,
      "step": 9945
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.51191246509552,
      "learning_rate": 1.529070990299117e-05,
      "loss": 0.5969,
      "step": 9946
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6437772512435913,
      "learning_rate": 1.528976237034997e-05,
      "loss": 0.5813,
      "step": 9947
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.525168776512146,
      "learning_rate": 1.5288814771759318e-05,
      "loss": 0.5577,
      "step": 9948
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.660361409187317,
      "learning_rate": 1.528786710723103e-05,
      "loss": 0.4994,
      "step": 9949
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2806357145309448,
      "learning_rate": 1.5286919376776917e-05,
      "loss": 0.5195,
      "step": 9950
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4136466979980469,
      "learning_rate": 1.5285971580408795e-05,
      "loss": 0.6455,
      "step": 9951
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7377371788024902,
      "learning_rate": 1.5285023718138483e-05,
      "loss": 0.6333,
      "step": 9952
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6571649312973022,
      "learning_rate": 1.5284075789977794e-05,
      "loss": 0.542,
      "step": 9953
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4158613681793213,
      "learning_rate": 1.528312779593855e-05,
      "loss": 0.4905,
      "step": 9954
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4765193462371826,
      "learning_rate": 1.528217973603257e-05,
      "loss": 0.5888,
      "step": 9955
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3915044069290161,
      "learning_rate": 1.528123161027167e-05,
      "loss": 0.5663,
      "step": 9956
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5077300071716309,
      "learning_rate": 1.5280283418667678e-05,
      "loss": 0.5781,
      "step": 9957
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7574646472930908,
      "learning_rate": 1.5279335161232407e-05,
      "loss": 0.4994,
      "step": 9958
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2200642824172974,
      "learning_rate": 1.5278386837977685e-05,
      "loss": 0.584,
      "step": 9959
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3266221284866333,
      "learning_rate": 1.527743844891533e-05,
      "loss": 0.5812,
      "step": 9960
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7141817808151245,
      "learning_rate": 1.527648999405717e-05,
      "loss": 0.5435,
      "step": 9961
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.46627676486969,
      "learning_rate": 1.527554147341503e-05,
      "loss": 0.5144,
      "step": 9962
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6696269512176514,
      "learning_rate": 1.5274592887000735e-05,
      "loss": 0.588,
      "step": 9963
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.3856048583984375,
      "learning_rate": 1.527364423482611e-05,
      "loss": 0.5138,
      "step": 9964
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6224212646484375,
      "learning_rate": 1.527269551690298e-05,
      "loss": 0.61,
      "step": 9965
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4434648752212524,
      "learning_rate": 1.527174673324318e-05,
      "loss": 0.4445,
      "step": 9966
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6431457996368408,
      "learning_rate": 1.5270797883858536e-05,
      "loss": 0.5709,
      "step": 9967
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.757528305053711,
      "learning_rate": 1.526984896876087e-05,
      "loss": 0.5393,
      "step": 9968
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6393866539001465,
      "learning_rate": 1.5268899987962024e-05,
      "loss": 0.6527,
      "step": 9969
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4190542697906494,
      "learning_rate": 1.5267950941473824e-05,
      "loss": 0.5711,
      "step": 9970
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3951746225357056,
      "learning_rate": 1.5267001829308103e-05,
      "loss": 0.5714,
      "step": 9971
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5083293914794922,
      "learning_rate": 1.5266052651476694e-05,
      "loss": 0.5674,
      "step": 9972
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6800687313079834,
      "learning_rate": 1.5265103407991422e-05,
      "loss": 0.5666,
      "step": 9973
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4270799160003662,
      "learning_rate": 1.526415409886414e-05,
      "loss": 0.5912,
      "step": 9974
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.8768028020858765,
      "learning_rate": 1.5263204724106665e-05,
      "loss": 0.5993,
      "step": 9975
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4368467330932617,
      "learning_rate": 1.5262255283730848e-05,
      "loss": 0.5605,
      "step": 9976
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.9213335514068604,
      "learning_rate": 1.5261305777748515e-05,
      "loss": 0.6404,
      "step": 9977
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4834320545196533,
      "learning_rate": 1.526035620617151e-05,
      "loss": 0.6872,
      "step": 9978
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.9825561046600342,
      "learning_rate": 1.5259406569011667e-05,
      "loss": 0.7223,
      "step": 9979
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.75397527217865,
      "learning_rate": 1.525845686628083e-05,
      "loss": 0.5987,
      "step": 9980
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.295896530151367,
      "learning_rate": 1.5257507097990836e-05,
      "loss": 0.6285,
      "step": 9981
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3591969013214111,
      "learning_rate": 1.5256557264153533e-05,
      "loss": 0.5432,
      "step": 9982
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.610870361328125,
      "learning_rate": 1.5255607364780752e-05,
      "loss": 0.6447,
      "step": 9983
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7827064990997314,
      "learning_rate": 1.5254657399884343e-05,
      "loss": 0.6127,
      "step": 9984
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.9698257446289062,
      "learning_rate": 1.5253707369476148e-05,
      "loss": 0.6417,
      "step": 9985
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.288222312927246,
      "learning_rate": 1.5252757273568013e-05,
      "loss": 0.527,
      "step": 9986
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.8094313144683838,
      "learning_rate": 1.525180711217178e-05,
      "loss": 0.6268,
      "step": 9987
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6537905931472778,
      "learning_rate": 1.5250856885299295e-05,
      "loss": 0.5472,
      "step": 9988
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6749372482299805,
      "learning_rate": 1.5249906592962407e-05,
      "loss": 0.5829,
      "step": 9989
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4009082317352295,
      "learning_rate": 1.5248956235172966e-05,
      "loss": 0.4243,
      "step": 9990
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3199923038482666,
      "learning_rate": 1.5248005811942816e-05,
      "loss": 0.5855,
      "step": 9991
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4780421257019043,
      "learning_rate": 1.5247055323283806e-05,
      "loss": 0.5125,
      "step": 9992
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.8089557886123657,
      "learning_rate": 1.524610476920779e-05,
      "loss": 0.6562,
      "step": 9993
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.620357871055603,
      "learning_rate": 1.5245154149726616e-05,
      "loss": 0.5155,
      "step": 9994
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5367493629455566,
      "learning_rate": 1.5244203464852136e-05,
      "loss": 0.5738,
      "step": 9995
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4980769157409668,
      "learning_rate": 1.5243252714596204e-05,
      "loss": 0.4739,
      "step": 9996
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.7621209621429443,
      "learning_rate": 1.524230189897067e-05,
      "loss": 0.5776,
      "step": 9997
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5925263166427612,
      "learning_rate": 1.5241351017987396e-05,
      "loss": 0.5678,
      "step": 9998
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.5632673501968384,
      "learning_rate": 1.5240400071658226e-05,
      "loss": 0.5514,
      "step": 9999
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.488749384880066,
      "learning_rate": 1.5239449059995024e-05,
      "loss": 0.5751,
      "step": 10000
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4476791620254517,
      "learning_rate": 1.5238497983009643e-05,
      "loss": 0.632,
      "step": 10001
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.4732959270477295,
      "learning_rate": 1.5237546840713941e-05,
      "loss": 0.6875,
      "step": 10002
    },
    {
      "epoch": 1.03,
      "grad_norm": 2.240980625152588,
      "learning_rate": 1.523659563311978e-05,
      "loss": 0.5809,
      "step": 10003
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3967995643615723,
      "learning_rate": 1.523564436023901e-05,
      "loss": 0.562,
      "step": 10004
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3841556310653687,
      "learning_rate": 1.52346930220835e-05,
      "loss": 0.4532,
      "step": 10005
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6545979976654053,
      "learning_rate": 1.523374161866511e-05,
      "loss": 0.6302,
      "step": 10006
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.6474262475967407,
      "learning_rate": 1.5232790149995695e-05,
      "loss": 0.5559,
      "step": 10007
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.2639447450637817,
      "learning_rate": 1.5231838616087125e-05,
      "loss": 0.5105,
      "step": 10008
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5528000593185425,
      "learning_rate": 1.5230887016951255e-05,
      "loss": 0.5676,
      "step": 10009
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4376264810562134,
      "learning_rate": 1.5229935352599953e-05,
      "loss": 0.4386,
      "step": 10010
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3565860986709595,
      "learning_rate": 1.5228983623045089e-05,
      "loss": 0.552,
      "step": 10011
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4973258972167969,
      "learning_rate": 1.522803182829852e-05,
      "loss": 0.5404,
      "step": 10012
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6555752754211426,
      "learning_rate": 1.5227079968372116e-05,
      "loss": 0.6197,
      "step": 10013
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3532661199569702,
      "learning_rate": 1.5226128043277744e-05,
      "loss": 0.5878,
      "step": 10014
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4480183124542236,
      "learning_rate": 1.5225176053027273e-05,
      "loss": 0.4796,
      "step": 10015
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6237863302230835,
      "learning_rate": 1.522422399763257e-05,
      "loss": 0.5318,
      "step": 10016
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.490073323249817,
      "learning_rate": 1.5223271877105505e-05,
      "loss": 0.5328,
      "step": 10017
    },
    {
      "epoch": 1.04,
      "grad_norm": 2.1296188831329346,
      "learning_rate": 1.522231969145795e-05,
      "loss": 0.6888,
      "step": 10018
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6168757677078247,
      "learning_rate": 1.5221367440701777e-05,
      "loss": 0.5696,
      "step": 10019
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5844802856445312,
      "learning_rate": 1.5220415124848853e-05,
      "loss": 0.5539,
      "step": 10020
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.8011592626571655,
      "learning_rate": 1.5219462743911058e-05,
      "loss": 0.6243,
      "step": 10021
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7094124555587769,
      "learning_rate": 1.521851029790026e-05,
      "loss": 0.7655,
      "step": 10022
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3372042179107666,
      "learning_rate": 1.5217557786828333e-05,
      "loss": 0.5478,
      "step": 10023
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.710193395614624,
      "learning_rate": 1.5216605210707158e-05,
      "loss": 0.5437,
      "step": 10024
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6024999618530273,
      "learning_rate": 1.5215652569548607e-05,
      "loss": 0.6283,
      "step": 10025
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6289116144180298,
      "learning_rate": 1.521469986336456e-05,
      "loss": 0.7016,
      "step": 10026
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.647465467453003,
      "learning_rate": 1.5213747092166888e-05,
      "loss": 0.4687,
      "step": 10027
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4301254749298096,
      "learning_rate": 1.5212794255967476e-05,
      "loss": 0.5288,
      "step": 10028
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.708360195159912,
      "learning_rate": 1.5211841354778201e-05,
      "loss": 0.567,
      "step": 10029
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4188555479049683,
      "learning_rate": 1.5210888388610945e-05,
      "loss": 0.5911,
      "step": 10030
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5048956871032715,
      "learning_rate": 1.5209935357477586e-05,
      "loss": 0.6607,
      "step": 10031
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3385342359542847,
      "learning_rate": 1.5208982261390009e-05,
      "loss": 0.4407,
      "step": 10032
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5849275588989258,
      "learning_rate": 1.5208029100360096e-05,
      "loss": 0.5823,
      "step": 10033
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4901851415634155,
      "learning_rate": 1.5207075874399728e-05,
      "loss": 0.7038,
      "step": 10034
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4216654300689697,
      "learning_rate": 1.5206122583520787e-05,
      "loss": 0.6797,
      "step": 10035
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4015854597091675,
      "learning_rate": 1.5205169227735165e-05,
      "loss": 0.5908,
      "step": 10036
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.410086750984192,
      "learning_rate": 1.5204215807054747e-05,
      "loss": 0.5994,
      "step": 10037
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.8871418237686157,
      "learning_rate": 1.5203262321491416e-05,
      "loss": 0.7692,
      "step": 10038
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5026013851165771,
      "learning_rate": 1.5202308771057062e-05,
      "loss": 0.6812,
      "step": 10039
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7896003723144531,
      "learning_rate": 1.5201355155763567e-05,
      "loss": 0.5218,
      "step": 10040
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6072808504104614,
      "learning_rate": 1.5200401475622827e-05,
      "loss": 0.583,
      "step": 10041
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7067408561706543,
      "learning_rate": 1.519944773064673e-05,
      "loss": 0.5225,
      "step": 10042
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.584093689918518,
      "learning_rate": 1.5198493920847169e-05,
      "loss": 0.5382,
      "step": 10043
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5586048364639282,
      "learning_rate": 1.519754004623603e-05,
      "loss": 0.6017,
      "step": 10044
    },
    {
      "epoch": 1.04,
      "grad_norm": 2.2743470668792725,
      "learning_rate": 1.5196586106825211e-05,
      "loss": 0.5426,
      "step": 10045
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3528168201446533,
      "learning_rate": 1.5195632102626602e-05,
      "loss": 0.488,
      "step": 10046
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.481706976890564,
      "learning_rate": 1.5194678033652097e-05,
      "loss": 0.5412,
      "step": 10047
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3950246572494507,
      "learning_rate": 1.519372389991359e-05,
      "loss": 0.5635,
      "step": 10048
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5209683179855347,
      "learning_rate": 1.5192769701422977e-05,
      "loss": 0.6203,
      "step": 10049
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3653334379196167,
      "learning_rate": 1.5191815438192156e-05,
      "loss": 0.5299,
      "step": 10050
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4214634895324707,
      "learning_rate": 1.5190861110233024e-05,
      "loss": 0.5912,
      "step": 10051
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5855324268341064,
      "learning_rate": 1.5189906717557481e-05,
      "loss": 0.5168,
      "step": 10052
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.2950693368911743,
      "learning_rate": 1.5188952260177418e-05,
      "loss": 0.6851,
      "step": 10053
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6231601238250732,
      "learning_rate": 1.518799773810474e-05,
      "loss": 0.6236,
      "step": 10054
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5330872535705566,
      "learning_rate": 1.518704315135135e-05,
      "loss": 0.5737,
      "step": 10055
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7090436220169067,
      "learning_rate": 1.5186088499929146e-05,
      "loss": 0.6242,
      "step": 10056
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5125807523727417,
      "learning_rate": 1.5185133783850027e-05,
      "loss": 0.6295,
      "step": 10057
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.9652949571609497,
      "learning_rate": 1.5184179003125903e-05,
      "loss": 0.5603,
      "step": 10058
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.2912410497665405,
      "learning_rate": 1.5183224157768673e-05,
      "loss": 0.5491,
      "step": 10059
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5899803638458252,
      "learning_rate": 1.518226924779024e-05,
      "loss": 0.5395,
      "step": 10060
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4738073348999023,
      "learning_rate": 1.5181314273202511e-05,
      "loss": 0.6108,
      "step": 10061
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3772214651107788,
      "learning_rate": 1.5180359234017393e-05,
      "loss": 0.45,
      "step": 10062
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6971168518066406,
      "learning_rate": 1.5179404130246793e-05,
      "loss": 0.5626,
      "step": 10063
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6118468046188354,
      "learning_rate": 1.5178448961902614e-05,
      "loss": 0.5829,
      "step": 10064
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.79864501953125,
      "learning_rate": 1.5177493728996772e-05,
      "loss": 0.6757,
      "step": 10065
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5696303844451904,
      "learning_rate": 1.517653843154117e-05,
      "loss": 0.5723,
      "step": 10066
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.426671028137207,
      "learning_rate": 1.5175583069547721e-05,
      "loss": 0.5924,
      "step": 10067
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6799266338348389,
      "learning_rate": 1.5174627643028337e-05,
      "loss": 0.578,
      "step": 10068
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.503770112991333,
      "learning_rate": 1.5173672151994928e-05,
      "loss": 0.5898,
      "step": 10069
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.266373634338379,
      "learning_rate": 1.5172716596459402e-05,
      "loss": 0.5613,
      "step": 10070
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.584073543548584,
      "learning_rate": 1.5171760976433683e-05,
      "loss": 0.59,
      "step": 10071
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3480043411254883,
      "learning_rate": 1.5170805291929674e-05,
      "loss": 0.6373,
      "step": 10072
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.487061858177185,
      "learning_rate": 1.5169849542959299e-05,
      "loss": 0.5693,
      "step": 10073
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6065624952316284,
      "learning_rate": 1.5168893729534464e-05,
      "loss": 0.4623,
      "step": 10074
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.2449514865875244,
      "learning_rate": 1.5167937851667094e-05,
      "loss": 0.485,
      "step": 10075
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3810979127883911,
      "learning_rate": 1.5166981909369102e-05,
      "loss": 0.5686,
      "step": 10076
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5198190212249756,
      "learning_rate": 1.5166025902652409e-05,
      "loss": 0.5394,
      "step": 10077
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.322383165359497,
      "learning_rate": 1.516506983152893e-05,
      "loss": 0.5211,
      "step": 10078
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7871594429016113,
      "learning_rate": 1.516411369601059e-05,
      "loss": 0.5784,
      "step": 10079
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6741567850112915,
      "learning_rate": 1.5163157496109303e-05,
      "loss": 0.6789,
      "step": 10080
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.619624376296997,
      "learning_rate": 1.5162201231836995e-05,
      "loss": 0.5137,
      "step": 10081
    },
    {
      "epoch": 1.04,
      "grad_norm": 2.0486276149749756,
      "learning_rate": 1.5161244903205586e-05,
      "loss": 0.7489,
      "step": 10082
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.7227282524108887,
      "learning_rate": 1.5160288510226998e-05,
      "loss": 0.6511,
      "step": 10083
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5517616271972656,
      "learning_rate": 1.5159332052913159e-05,
      "loss": 0.5192,
      "step": 10084
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4018198251724243,
      "learning_rate": 1.5158375531275989e-05,
      "loss": 0.5481,
      "step": 10085
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.4837276935577393,
      "learning_rate": 1.5157418945327415e-05,
      "loss": 0.6047,
      "step": 10086
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.9993892908096313,
      "learning_rate": 1.5156462295079366e-05,
      "loss": 0.5295,
      "step": 10087
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.68590247631073,
      "learning_rate": 1.5155505580543764e-05,
      "loss": 0.5811,
      "step": 10088
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3305857181549072,
      "learning_rate": 1.5154548801732541e-05,
      "loss": 0.5751,
      "step": 10089
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.2872340679168701,
      "learning_rate": 1.515359195865762e-05,
      "loss": 0.5174,
      "step": 10090
    },
    {
      "epoch": 1.04,
      "grad_norm": 2.0416903495788574,
      "learning_rate": 1.5152635051330936e-05,
      "loss": 0.5104,
      "step": 10091
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.622467279434204,
      "learning_rate": 1.5151678079764414e-05,
      "loss": 0.6428,
      "step": 10092
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6467145681381226,
      "learning_rate": 1.515072104396999e-05,
      "loss": 0.5162,
      "step": 10093
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.1550873517990112,
      "learning_rate": 1.5149763943959594e-05,
      "loss": 0.5956,
      "step": 10094
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6726680994033813,
      "learning_rate": 1.5148806779745156e-05,
      "loss": 0.6203,
      "step": 10095
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6155037879943848,
      "learning_rate": 1.514784955133861e-05,
      "loss": 0.7181,
      "step": 10096
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.3321462869644165,
      "learning_rate": 1.51468922587519e-05,
      "loss": 0.7042,
      "step": 10097
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.6732019186019897,
      "learning_rate": 1.5145934901996944e-05,
      "loss": 0.5669,
      "step": 10098
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.976831078529358,
      "learning_rate": 1.514497748108569e-05,
      "loss": 0.5478,
      "step": 10099
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5431113243103027,
      "learning_rate": 1.514401999603007e-05,
      "loss": 0.5666,
      "step": 10100
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5744870901107788,
      "learning_rate": 1.5143062446842018e-05,
      "loss": 0.4806,
      "step": 10101
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5730913877487183,
      "learning_rate": 1.5142104833533481e-05,
      "loss": 0.5536,
      "step": 10102
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5034154653549194,
      "learning_rate": 1.5141147156116392e-05,
      "loss": 0.561,
      "step": 10103
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5173383951187134,
      "learning_rate": 1.5140189414602691e-05,
      "loss": 0.668,
      "step": 10104
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7110450267791748,
      "learning_rate": 1.5139231609004323e-05,
      "loss": 0.6752,
      "step": 10105
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3937580585479736,
      "learning_rate": 1.513827373933322e-05,
      "loss": 0.4785,
      "step": 10106
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3968905210494995,
      "learning_rate": 1.5137315805601333e-05,
      "loss": 0.5148,
      "step": 10107
    },
    {
      "epoch": 1.05,
      "grad_norm": 2.062103748321533,
      "learning_rate": 1.5136357807820599e-05,
      "loss": 0.6967,
      "step": 10108
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4510180950164795,
      "learning_rate": 1.5135399746002965e-05,
      "loss": 0.5831,
      "step": 10109
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3637685775756836,
      "learning_rate": 1.5134441620160379e-05,
      "loss": 0.563,
      "step": 10110
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.564922571182251,
      "learning_rate": 1.5133483430304775e-05,
      "loss": 0.5986,
      "step": 10111
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3831454515457153,
      "learning_rate": 1.5132525176448111e-05,
      "loss": 0.5667,
      "step": 10112
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.2265689373016357,
      "learning_rate": 1.5131566858602331e-05,
      "loss": 0.4141,
      "step": 10113
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.338545560836792,
      "learning_rate": 1.5130608476779376e-05,
      "loss": 0.6393,
      "step": 10114
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3425571918487549,
      "learning_rate": 1.5129650030991203e-05,
      "loss": 0.5204,
      "step": 10115
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5066866874694824,
      "learning_rate": 1.5128691521249756e-05,
      "loss": 0.6459,
      "step": 10116
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7940326929092407,
      "learning_rate": 1.5127732947566985e-05,
      "loss": 0.5806,
      "step": 10117
    },
    {
      "epoch": 1.05,
      "grad_norm": 2.266338348388672,
      "learning_rate": 1.5126774309954849e-05,
      "loss": 0.611,
      "step": 10118
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.8033958673477173,
      "learning_rate": 1.5125815608425285e-05,
      "loss": 0.7127,
      "step": 10119
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.916902780532837,
      "learning_rate": 1.5124856842990262e-05,
      "loss": 0.6439,
      "step": 10120
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4836583137512207,
      "learning_rate": 1.512389801366172e-05,
      "loss": 0.5844,
      "step": 10121
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3289047479629517,
      "learning_rate": 1.5122939120451618e-05,
      "loss": 0.457,
      "step": 10122
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5817900896072388,
      "learning_rate": 1.5121980163371914e-05,
      "loss": 0.6014,
      "step": 10123
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.777716040611267,
      "learning_rate": 1.512102114243456e-05,
      "loss": 0.5894,
      "step": 10124
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3625719547271729,
      "learning_rate": 1.5120062057651513e-05,
      "loss": 0.5752,
      "step": 10125
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5653983354568481,
      "learning_rate": 1.511910290903473e-05,
      "loss": 0.4703,
      "step": 10126
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.2361449003219604,
      "learning_rate": 1.511814369659617e-05,
      "loss": 0.5033,
      "step": 10127
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.364729642868042,
      "learning_rate": 1.5117184420347792e-05,
      "loss": 0.63,
      "step": 10128
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6948357820510864,
      "learning_rate": 1.5116225080301554e-05,
      "loss": 0.6644,
      "step": 10129
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3905211687088013,
      "learning_rate": 1.511526567646942e-05,
      "loss": 0.4959,
      "step": 10130
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5786058902740479,
      "learning_rate": 1.5114306208863348e-05,
      "loss": 0.5082,
      "step": 10131
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.54238760471344,
      "learning_rate": 1.5113346677495297e-05,
      "loss": 0.4913,
      "step": 10132
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5620005130767822,
      "learning_rate": 1.511238708237724e-05,
      "loss": 0.5256,
      "step": 10133
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.561631202697754,
      "learning_rate": 1.5111427423521132e-05,
      "loss": 0.6143,
      "step": 10134
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5679587125778198,
      "learning_rate": 1.5110467700938937e-05,
      "loss": 0.6597,
      "step": 10135
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4067022800445557,
      "learning_rate": 1.5109507914642627e-05,
      "loss": 0.5643,
      "step": 10136
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.2562562227249146,
      "learning_rate": 1.5108548064644161e-05,
      "loss": 0.5672,
      "step": 10137
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5972998142242432,
      "learning_rate": 1.510758815095551e-05,
      "loss": 0.5991,
      "step": 10138
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7263067960739136,
      "learning_rate": 1.5106628173588642e-05,
      "loss": 0.6343,
      "step": 10139
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5745958089828491,
      "learning_rate": 1.5105668132555522e-05,
      "loss": 0.6465,
      "step": 10140
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6953741312026978,
      "learning_rate": 1.5104708027868122e-05,
      "loss": 0.6446,
      "step": 10141
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6277363300323486,
      "learning_rate": 1.5103747859538409e-05,
      "loss": 0.4959,
      "step": 10142
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.2744609117507935,
      "learning_rate": 1.5102787627578357e-05,
      "loss": 0.4803,
      "step": 10143
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4572733640670776,
      "learning_rate": 1.510182733199994e-05,
      "loss": 0.5486,
      "step": 10144
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4348515272140503,
      "learning_rate": 1.510086697281512e-05,
      "loss": 0.5312,
      "step": 10145
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.308656930923462,
      "learning_rate": 1.5099906550035881e-05,
      "loss": 0.5344,
      "step": 10146
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5585126876831055,
      "learning_rate": 1.5098946063674194e-05,
      "loss": 0.6036,
      "step": 10147
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5115165710449219,
      "learning_rate": 1.5097985513742028e-05,
      "loss": 0.5247,
      "step": 10148
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5098875761032104,
      "learning_rate": 1.5097024900251369e-05,
      "loss": 0.6604,
      "step": 10149
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3132160902023315,
      "learning_rate": 1.5096064223214182e-05,
      "loss": 0.4826,
      "step": 10150
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4367897510528564,
      "learning_rate": 1.5095103482642452e-05,
      "loss": 0.5255,
      "step": 10151
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6441669464111328,
      "learning_rate": 1.5094142678548157e-05,
      "loss": 0.5567,
      "step": 10152
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4402551651000977,
      "learning_rate": 1.509318181094327e-05,
      "loss": 0.535,
      "step": 10153
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5045537948608398,
      "learning_rate": 1.5092220879839774e-05,
      "loss": 0.4531,
      "step": 10154
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5142464637756348,
      "learning_rate": 1.509125988524965e-05,
      "loss": 0.52,
      "step": 10155
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6809860467910767,
      "learning_rate": 1.5090298827184878e-05,
      "loss": 0.6967,
      "step": 10156
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4156627655029297,
      "learning_rate": 1.508933770565744e-05,
      "loss": 0.5271,
      "step": 10157
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5313352346420288,
      "learning_rate": 1.5088376520679315e-05,
      "loss": 0.5891,
      "step": 10158
    },
    {
      "epoch": 1.05,
      "grad_norm": 2.3587806224823,
      "learning_rate": 1.5087415272262495e-05,
      "loss": 0.7338,
      "step": 10159
    },
    {
      "epoch": 1.05,
      "grad_norm": 2.076472759246826,
      "learning_rate": 1.5086453960418958e-05,
      "loss": 0.6117,
      "step": 10160
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.570677638053894,
      "learning_rate": 1.508549258516069e-05,
      "loss": 0.5249,
      "step": 10161
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4905905723571777,
      "learning_rate": 1.5084531146499677e-05,
      "loss": 0.4716,
      "step": 10162
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3545504808425903,
      "learning_rate": 1.5083569644447904e-05,
      "loss": 0.4307,
      "step": 10163
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4600369930267334,
      "learning_rate": 1.5082608079017363e-05,
      "loss": 0.5888,
      "step": 10164
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7645920515060425,
      "learning_rate": 1.5081646450220038e-05,
      "loss": 0.6845,
      "step": 10165
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6031373739242554,
      "learning_rate": 1.508068475806792e-05,
      "loss": 0.5114,
      "step": 10166
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.340901494026184,
      "learning_rate": 1.5079723002572998e-05,
      "loss": 0.5755,
      "step": 10167
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5019335746765137,
      "learning_rate": 1.507876118374726e-05,
      "loss": 0.5373,
      "step": 10168
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3359016180038452,
      "learning_rate": 1.5077799301602705e-05,
      "loss": 0.557,
      "step": 10169
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6089681386947632,
      "learning_rate": 1.5076837356151319e-05,
      "loss": 0.609,
      "step": 10170
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6606390476226807,
      "learning_rate": 1.5075875347405094e-05,
      "loss": 0.6124,
      "step": 10171
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6440550088882446,
      "learning_rate": 1.5074913275376028e-05,
      "loss": 0.6163,
      "step": 10172
    },
    {
      "epoch": 1.05,
      "grad_norm": 2.069322109222412,
      "learning_rate": 1.5073951140076113e-05,
      "loss": 0.5554,
      "step": 10173
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.2665138244628906,
      "learning_rate": 1.5072988941517341e-05,
      "loss": 0.4373,
      "step": 10174
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.563670039176941,
      "learning_rate": 1.5072026679711717e-05,
      "loss": 0.4365,
      "step": 10175
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.903969168663025,
      "learning_rate": 1.507106435467123e-05,
      "loss": 0.6842,
      "step": 10176
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.472493052482605,
      "learning_rate": 1.507010196640788e-05,
      "loss": 0.5462,
      "step": 10177
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3777462244033813,
      "learning_rate": 1.5069139514933668e-05,
      "loss": 0.5304,
      "step": 10178
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6145319938659668,
      "learning_rate": 1.506817700026059e-05,
      "loss": 0.6212,
      "step": 10179
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.624596357345581,
      "learning_rate": 1.5067214422400647e-05,
      "loss": 0.5936,
      "step": 10180
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3882124423980713,
      "learning_rate": 1.506625178136584e-05,
      "loss": 0.5041,
      "step": 10181
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5896598100662231,
      "learning_rate": 1.5065289077168169e-05,
      "loss": 0.4299,
      "step": 10182
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3481948375701904,
      "learning_rate": 1.506432630981964e-05,
      "loss": 0.5519,
      "step": 10183
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4462523460388184,
      "learning_rate": 1.5063363479332251e-05,
      "loss": 0.6149,
      "step": 10184
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4347691535949707,
      "learning_rate": 1.5062400585718013e-05,
      "loss": 0.5662,
      "step": 10185
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5802892446517944,
      "learning_rate": 1.5061437628988924e-05,
      "loss": 0.649,
      "step": 10186
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.7786110639572144,
      "learning_rate": 1.5060474609156991e-05,
      "loss": 0.5974,
      "step": 10187
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5837210416793823,
      "learning_rate": 1.5059511526234227e-05,
      "loss": 0.4784,
      "step": 10188
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4145472049713135,
      "learning_rate": 1.5058548380232632e-05,
      "loss": 0.5151,
      "step": 10189
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.9627439975738525,
      "learning_rate": 1.505758517116421e-05,
      "loss": 0.5355,
      "step": 10190
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4576244354248047,
      "learning_rate": 1.5056621899040984e-05,
      "loss": 0.6355,
      "step": 10191
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.8357853889465332,
      "learning_rate": 1.5055658563874948e-05,
      "loss": 0.5156,
      "step": 10192
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.4997121095657349,
      "learning_rate": 1.5054695165678123e-05,
      "loss": 0.5947,
      "step": 10193
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.792033314704895,
      "learning_rate": 1.5053731704462516e-05,
      "loss": 0.6131,
      "step": 10194
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.456372618675232,
      "learning_rate": 1.5052768180240136e-05,
      "loss": 0.562,
      "step": 10195
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.8047089576721191,
      "learning_rate": 1.5051804593023e-05,
      "loss": 0.5472,
      "step": 10196
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.3957593441009521,
      "learning_rate": 1.5050840942823122e-05,
      "loss": 0.4955,
      "step": 10197
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.2353029251098633,
      "learning_rate": 1.5049877229652513e-05,
      "loss": 0.5643,
      "step": 10198
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5310639142990112,
      "learning_rate": 1.5048913453523188e-05,
      "loss": 0.5576,
      "step": 10199
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.324836254119873,
      "learning_rate": 1.5047949614447164e-05,
      "loss": 0.5994,
      "step": 10200
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.304413080215454,
      "learning_rate": 1.5046985712436462e-05,
      "loss": 0.5653,
      "step": 10201
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2940272092819214,
      "learning_rate": 1.5046021747503088e-05,
      "loss": 0.5702,
      "step": 10202
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5053844451904297,
      "learning_rate": 1.5045057719659069e-05,
      "loss": 0.5501,
      "step": 10203
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3426077365875244,
      "learning_rate": 1.5044093628916425e-05,
      "loss": 0.4868,
      "step": 10204
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4443414211273193,
      "learning_rate": 1.504312947528717e-05,
      "loss": 0.5086,
      "step": 10205
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3767056465148926,
      "learning_rate": 1.5042165258783328e-05,
      "loss": 0.5007,
      "step": 10206
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4135940074920654,
      "learning_rate": 1.5041200979416917e-05,
      "loss": 0.5979,
      "step": 10207
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.313846230506897,
      "learning_rate": 1.5040236637199963e-05,
      "loss": 0.5463,
      "step": 10208
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4995989799499512,
      "learning_rate": 1.5039272232144489e-05,
      "loss": 0.6006,
      "step": 10209
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4398022890090942,
      "learning_rate": 1.5038307764262513e-05,
      "loss": 0.5554,
      "step": 10210
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4668633937835693,
      "learning_rate": 1.5037343233566066e-05,
      "loss": 0.612,
      "step": 10211
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5628821849822998,
      "learning_rate": 1.5036378640067169e-05,
      "loss": 0.5379,
      "step": 10212
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5203195810317993,
      "learning_rate": 1.5035413983777845e-05,
      "loss": 0.5857,
      "step": 10213
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5082710981369019,
      "learning_rate": 1.503444926471013e-05,
      "loss": 0.4961,
      "step": 10214
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.371150016784668,
      "learning_rate": 1.5033484482876043e-05,
      "loss": 0.5965,
      "step": 10215
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.543701171875,
      "learning_rate": 1.503251963828762e-05,
      "loss": 0.6798,
      "step": 10216
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7397102117538452,
      "learning_rate": 1.5031554730956885e-05,
      "loss": 0.6536,
      "step": 10217
    },
    {
      "epoch": 1.06,
      "grad_norm": 2.109455108642578,
      "learning_rate": 1.5030589760895866e-05,
      "loss": 0.6551,
      "step": 10218
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6148234605789185,
      "learning_rate": 1.5029624728116599e-05,
      "loss": 0.6196,
      "step": 10219
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.524285912513733,
      "learning_rate": 1.502865963263111e-05,
      "loss": 0.557,
      "step": 10220
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6159263849258423,
      "learning_rate": 1.5027694474451435e-05,
      "loss": 0.715,
      "step": 10221
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6502623558044434,
      "learning_rate": 1.5026729253589607e-05,
      "loss": 0.6122,
      "step": 10222
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7953037023544312,
      "learning_rate": 1.5025763970057656e-05,
      "loss": 0.5714,
      "step": 10223
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.9794442653656006,
      "learning_rate": 1.5024798623867623e-05,
      "loss": 0.5069,
      "step": 10224
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.648694396018982,
      "learning_rate": 1.5023833215031538e-05,
      "loss": 0.5168,
      "step": 10225
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4090988636016846,
      "learning_rate": 1.5022867743561435e-05,
      "loss": 0.5406,
      "step": 10226
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5163297653198242,
      "learning_rate": 1.502190220946936e-05,
      "loss": 0.7412,
      "step": 10227
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6873397827148438,
      "learning_rate": 1.5020936612767343e-05,
      "loss": 0.632,
      "step": 10228
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4891531467437744,
      "learning_rate": 1.5019970953467424e-05,
      "loss": 0.5769,
      "step": 10229
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2745310068130493,
      "learning_rate": 1.5019005231581645e-05,
      "loss": 0.5975,
      "step": 10230
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2701170444488525,
      "learning_rate": 1.5018039447122044e-05,
      "loss": 0.496,
      "step": 10231
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.270014762878418,
      "learning_rate": 1.501707360010066e-05,
      "loss": 0.585,
      "step": 10232
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5803710222244263,
      "learning_rate": 1.5016107690529539e-05,
      "loss": 0.508,
      "step": 10233
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7177119255065918,
      "learning_rate": 1.5015141718420717e-05,
      "loss": 0.6082,
      "step": 10234
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5102335214614868,
      "learning_rate": 1.5014175683786243e-05,
      "loss": 0.4985,
      "step": 10235
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6308144330978394,
      "learning_rate": 1.5013209586638158e-05,
      "loss": 0.6031,
      "step": 10236
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2383573055267334,
      "learning_rate": 1.5012243426988507e-05,
      "loss": 0.5408,
      "step": 10237
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7401138544082642,
      "learning_rate": 1.5011277204849335e-05,
      "loss": 0.5325,
      "step": 10238
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3670865297317505,
      "learning_rate": 1.501031092023269e-05,
      "loss": 0.4683,
      "step": 10239
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5037761926651,
      "learning_rate": 1.500934457315062e-05,
      "loss": 0.5667,
      "step": 10240
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4139984846115112,
      "learning_rate": 1.500837816361517e-05,
      "loss": 0.54,
      "step": 10241
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6270930767059326,
      "learning_rate": 1.5007411691638387e-05,
      "loss": 0.6187,
      "step": 10242
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4768304824829102,
      "learning_rate": 1.5006445157232328e-05,
      "loss": 0.4915,
      "step": 10243
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.646274447441101,
      "learning_rate": 1.5005478560409035e-05,
      "loss": 0.5931,
      "step": 10244
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3423793315887451,
      "learning_rate": 1.5004511901180565e-05,
      "loss": 0.5117,
      "step": 10245
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6031137704849243,
      "learning_rate": 1.5003545179558963e-05,
      "loss": 0.5581,
      "step": 10246
    },
    {
      "epoch": 1.06,
      "grad_norm": 2.0666677951812744,
      "learning_rate": 1.5002578395556286e-05,
      "loss": 0.5586,
      "step": 10247
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5937649011611938,
      "learning_rate": 1.500161154918459e-05,
      "loss": 0.6237,
      "step": 10248
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7462693452835083,
      "learning_rate": 1.500064464045592e-05,
      "loss": 0.61,
      "step": 10249
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5460188388824463,
      "learning_rate": 1.499967766938234e-05,
      "loss": 0.5252,
      "step": 10250
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.783673644065857,
      "learning_rate": 1.49987106359759e-05,
      "loss": 0.4859,
      "step": 10251
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.631567120552063,
      "learning_rate": 1.4997743540248659e-05,
      "loss": 0.5522,
      "step": 10252
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4748154878616333,
      "learning_rate": 1.4996776382212675e-05,
      "loss": 0.597,
      "step": 10253
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4996333122253418,
      "learning_rate": 1.4995809161880001e-05,
      "loss": 0.5189,
      "step": 10254
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5306810140609741,
      "learning_rate": 1.49948418792627e-05,
      "loss": 0.6692,
      "step": 10255
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.8872311115264893,
      "learning_rate": 1.4993874534372835e-05,
      "loss": 0.6713,
      "step": 10256
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.465333104133606,
      "learning_rate": 1.4992907127222456e-05,
      "loss": 0.5359,
      "step": 10257
    },
    {
      "epoch": 1.06,
      "grad_norm": 2.012402296066284,
      "learning_rate": 1.4991939657823635e-05,
      "loss": 0.7477,
      "step": 10258
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.883305311203003,
      "learning_rate": 1.4990972126188425e-05,
      "loss": 0.6266,
      "step": 10259
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5350059270858765,
      "learning_rate": 1.4990004532328892e-05,
      "loss": 0.5896,
      "step": 10260
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.782152533531189,
      "learning_rate": 1.4989036876257102e-05,
      "loss": 0.5921,
      "step": 10261
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.409042239189148,
      "learning_rate": 1.4988069157985114e-05,
      "loss": 0.5654,
      "step": 10262
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.583678960800171,
      "learning_rate": 1.4987101377524997e-05,
      "loss": 0.5104,
      "step": 10263
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3167288303375244,
      "learning_rate": 1.4986133534888817e-05,
      "loss": 0.6237,
      "step": 10264
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.402870774269104,
      "learning_rate": 1.4985165630088635e-05,
      "loss": 0.5211,
      "step": 10265
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3219386339187622,
      "learning_rate": 1.4984197663136526e-05,
      "loss": 0.5005,
      "step": 10266
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.614330530166626,
      "learning_rate": 1.498322963404455e-05,
      "loss": 0.7061,
      "step": 10267
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5946370363235474,
      "learning_rate": 1.4982261542824782e-05,
      "loss": 0.5769,
      "step": 10268
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7634106874465942,
      "learning_rate": 1.498129338948929e-05,
      "loss": 0.6087,
      "step": 10269
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5324872732162476,
      "learning_rate": 1.4980325174050144e-05,
      "loss": 0.6103,
      "step": 10270
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.554358720779419,
      "learning_rate": 1.4979356896519413e-05,
      "loss": 0.5436,
      "step": 10271
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4271328449249268,
      "learning_rate": 1.497838855690917e-05,
      "loss": 0.6334,
      "step": 10272
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5477491617202759,
      "learning_rate": 1.4977420155231488e-05,
      "loss": 0.5403,
      "step": 10273
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3390463590621948,
      "learning_rate": 1.4976451691498444e-05,
      "loss": 0.6933,
      "step": 10274
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3825842142105103,
      "learning_rate": 1.4975483165722107e-05,
      "loss": 0.5319,
      "step": 10275
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6227827072143555,
      "learning_rate": 1.4974514577914554e-05,
      "loss": 0.5912,
      "step": 10276
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4425548315048218,
      "learning_rate": 1.497354592808786e-05,
      "loss": 0.6102,
      "step": 10277
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7602579593658447,
      "learning_rate": 1.4972577216254103e-05,
      "loss": 0.5005,
      "step": 10278
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2936489582061768,
      "learning_rate": 1.497160844242536e-05,
      "loss": 0.5591,
      "step": 10279
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3365402221679688,
      "learning_rate": 1.4970639606613708e-05,
      "loss": 0.6256,
      "step": 10280
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.8110371828079224,
      "learning_rate": 1.4969670708831226e-05,
      "loss": 0.7264,
      "step": 10281
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6768862009048462,
      "learning_rate": 1.4968701749089996e-05,
      "loss": 0.5297,
      "step": 10282
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2691696882247925,
      "learning_rate": 1.4967732727402093e-05,
      "loss": 0.5677,
      "step": 10283
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7322039604187012,
      "learning_rate": 1.4966763643779603e-05,
      "loss": 0.7123,
      "step": 10284
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5719112157821655,
      "learning_rate": 1.4965794498234607e-05,
      "loss": 0.5077,
      "step": 10285
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.5545536279678345,
      "learning_rate": 1.4964825290779185e-05,
      "loss": 0.5853,
      "step": 10286
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3148748874664307,
      "learning_rate": 1.4963856021425426e-05,
      "loss": 0.6158,
      "step": 10287
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3505221605300903,
      "learning_rate": 1.4962886690185407e-05,
      "loss": 0.5031,
      "step": 10288
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4043219089508057,
      "learning_rate": 1.496191729707122e-05,
      "loss": 0.5104,
      "step": 10289
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6756855249404907,
      "learning_rate": 1.4960947842094946e-05,
      "loss": 0.5246,
      "step": 10290
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3805092573165894,
      "learning_rate": 1.4959978325268674e-05,
      "loss": 0.6967,
      "step": 10291
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6141533851623535,
      "learning_rate": 1.4959008746604488e-05,
      "loss": 0.5374,
      "step": 10292
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.255578875541687,
      "learning_rate": 1.4958039106114483e-05,
      "loss": 0.5338,
      "step": 10293
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4911423921585083,
      "learning_rate": 1.4957069403810743e-05,
      "loss": 0.5521,
      "step": 10294
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6525408029556274,
      "learning_rate": 1.4956099639705356e-05,
      "loss": 0.658,
      "step": 10295
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.7206259965896606,
      "learning_rate": 1.4955129813810416e-05,
      "loss": 0.583,
      "step": 10296
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.6032347679138184,
      "learning_rate": 1.495415992613801e-05,
      "loss": 0.606,
      "step": 10297
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6289552450180054,
      "learning_rate": 1.4953189976700238e-05,
      "loss": 0.6963,
      "step": 10298
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6224507093429565,
      "learning_rate": 1.4952219965509183e-05,
      "loss": 0.6121,
      "step": 10299
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7683535814285278,
      "learning_rate": 1.4951249892576944e-05,
      "loss": 0.6567,
      "step": 10300
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.578066349029541,
      "learning_rate": 1.4950279757915614e-05,
      "loss": 0.4871,
      "step": 10301
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7276320457458496,
      "learning_rate": 1.4949309561537288e-05,
      "loss": 0.5338,
      "step": 10302
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6536000967025757,
      "learning_rate": 1.4948339303454065e-05,
      "loss": 0.6269,
      "step": 10303
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4408656358718872,
      "learning_rate": 1.4947368983678035e-05,
      "loss": 0.5041,
      "step": 10304
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3146315813064575,
      "learning_rate": 1.4946398602221299e-05,
      "loss": 0.5642,
      "step": 10305
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.854623794555664,
      "learning_rate": 1.4945428159095956e-05,
      "loss": 0.5518,
      "step": 10306
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1735299825668335,
      "learning_rate": 1.4944457654314103e-05,
      "loss": 0.5517,
      "step": 10307
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8955854177474976,
      "learning_rate": 1.4943487087887845e-05,
      "loss": 0.6118,
      "step": 10308
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6391375064849854,
      "learning_rate": 1.4942516459829274e-05,
      "loss": 0.5627,
      "step": 10309
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5409696102142334,
      "learning_rate": 1.4941545770150496e-05,
      "loss": 0.6552,
      "step": 10310
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2837456464767456,
      "learning_rate": 1.4940575018863612e-05,
      "loss": 0.5227,
      "step": 10311
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6199359893798828,
      "learning_rate": 1.4939604205980724e-05,
      "loss": 0.5025,
      "step": 10312
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.146539568901062,
      "learning_rate": 1.4938633331513938e-05,
      "loss": 0.5613,
      "step": 10313
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4570292234420776,
      "learning_rate": 1.4937662395475357e-05,
      "loss": 0.4593,
      "step": 10314
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6495662927627563,
      "learning_rate": 1.4936691397877084e-05,
      "loss": 0.6512,
      "step": 10315
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8536909818649292,
      "learning_rate": 1.4935720338731228e-05,
      "loss": 0.6424,
      "step": 10316
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3740320205688477,
      "learning_rate": 1.4934749218049891e-05,
      "loss": 0.5419,
      "step": 10317
    },
    {
      "epoch": 1.07,
      "grad_norm": 2.339151382446289,
      "learning_rate": 1.4933778035845185e-05,
      "loss": 0.733,
      "step": 10318
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3992130756378174,
      "learning_rate": 1.493280679212922e-05,
      "loss": 0.5557,
      "step": 10319
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4434822797775269,
      "learning_rate": 1.4931835486914095e-05,
      "loss": 0.6337,
      "step": 10320
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.76423978805542,
      "learning_rate": 1.4930864120211932e-05,
      "loss": 0.5219,
      "step": 10321
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.385066032409668,
      "learning_rate": 1.4929892692034831e-05,
      "loss": 0.471,
      "step": 10322
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5792427062988281,
      "learning_rate": 1.4928921202394909e-05,
      "loss": 0.5145,
      "step": 10323
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8531655073165894,
      "learning_rate": 1.4927949651304274e-05,
      "loss": 0.6075,
      "step": 10324
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7523517608642578,
      "learning_rate": 1.4926978038775043e-05,
      "loss": 0.6013,
      "step": 10325
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3752330541610718,
      "learning_rate": 1.492600636481933e-05,
      "loss": 0.587,
      "step": 10326
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6363905668258667,
      "learning_rate": 1.4925034629449241e-05,
      "loss": 0.5919,
      "step": 10327
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5851377248764038,
      "learning_rate": 1.4924062832676902e-05,
      "loss": 0.5482,
      "step": 10328
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7560641765594482,
      "learning_rate": 1.4923090974514423e-05,
      "loss": 0.5906,
      "step": 10329
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3615081310272217,
      "learning_rate": 1.4922119054973918e-05,
      "loss": 0.618,
      "step": 10330
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.588725209236145,
      "learning_rate": 1.4921147074067507e-05,
      "loss": 0.5169,
      "step": 10331
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.45931875705719,
      "learning_rate": 1.4920175031807312e-05,
      "loss": 0.6442,
      "step": 10332
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5181504487991333,
      "learning_rate": 1.4919202928205444e-05,
      "loss": 0.6747,
      "step": 10333
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.323235273361206,
      "learning_rate": 1.4918230763274031e-05,
      "loss": 0.5204,
      "step": 10334
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6398028135299683,
      "learning_rate": 1.4917258537025188e-05,
      "loss": 0.599,
      "step": 10335
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3981989622116089,
      "learning_rate": 1.4916286249471035e-05,
      "loss": 0.6774,
      "step": 10336
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.509944200515747,
      "learning_rate": 1.4915313900623698e-05,
      "loss": 0.5655,
      "step": 10337
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.506559133529663,
      "learning_rate": 1.4914341490495297e-05,
      "loss": 0.5504,
      "step": 10338
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6031060218811035,
      "learning_rate": 1.4913369019097956e-05,
      "loss": 0.5491,
      "step": 10339
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.743245005607605,
      "learning_rate": 1.4912396486443799e-05,
      "loss": 0.531,
      "step": 10340
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3788996934890747,
      "learning_rate": 1.4911423892544951e-05,
      "loss": 0.5643,
      "step": 10341
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.518994688987732,
      "learning_rate": 1.4910451237413538e-05,
      "loss": 0.5589,
      "step": 10342
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6427310705184937,
      "learning_rate": 1.4909478521061689e-05,
      "loss": 0.5619,
      "step": 10343
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5825220346450806,
      "learning_rate": 1.4908505743501525e-05,
      "loss": 0.4748,
      "step": 10344
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5297355651855469,
      "learning_rate": 1.4907532904745181e-05,
      "loss": 0.5465,
      "step": 10345
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6182490587234497,
      "learning_rate": 1.490656000480478e-05,
      "loss": 0.5808,
      "step": 10346
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3533985614776611,
      "learning_rate": 1.4905587043692453e-05,
      "loss": 0.5579,
      "step": 10347
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6612439155578613,
      "learning_rate": 1.4904614021420335e-05,
      "loss": 0.5833,
      "step": 10348
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4369536638259888,
      "learning_rate": 1.4903640938000547e-05,
      "loss": 0.5415,
      "step": 10349
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6897337436676025,
      "learning_rate": 1.4902667793445232e-05,
      "loss": 0.6497,
      "step": 10350
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6357758045196533,
      "learning_rate": 1.4901694587766515e-05,
      "loss": 0.6586,
      "step": 10351
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5252971649169922,
      "learning_rate": 1.4900721320976534e-05,
      "loss": 0.6826,
      "step": 10352
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3621129989624023,
      "learning_rate": 1.4899747993087419e-05,
      "loss": 0.5776,
      "step": 10353
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5770225524902344,
      "learning_rate": 1.4898774604111305e-05,
      "loss": 0.6051,
      "step": 10354
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.451708197593689,
      "learning_rate": 1.4897801154060334e-05,
      "loss": 0.5702,
      "step": 10355
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2616475820541382,
      "learning_rate": 1.4896827642946633e-05,
      "loss": 0.4319,
      "step": 10356
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.72297203540802,
      "learning_rate": 1.4895854070782345e-05,
      "loss": 0.6087,
      "step": 10357
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6309151649475098,
      "learning_rate": 1.4894880437579611e-05,
      "loss": 0.5972,
      "step": 10358
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4313538074493408,
      "learning_rate": 1.489390674335056e-05,
      "loss": 0.5287,
      "step": 10359
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.8366807699203491,
      "learning_rate": 1.489293298810734e-05,
      "loss": 0.6232,
      "step": 10360
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5917876958847046,
      "learning_rate": 1.4891959171862084e-05,
      "loss": 0.6168,
      "step": 10361
    },
    {
      "epoch": 1.07,
      "grad_norm": 2.0222108364105225,
      "learning_rate": 1.489098529462694e-05,
      "loss": 0.6881,
      "step": 10362
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3331995010375977,
      "learning_rate": 1.4890011356414049e-05,
      "loss": 0.4916,
      "step": 10363
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2975975275039673,
      "learning_rate": 1.4889037357235548e-05,
      "loss": 0.6194,
      "step": 10364
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5298793315887451,
      "learning_rate": 1.4888063297103584e-05,
      "loss": 0.489,
      "step": 10365
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5151666402816772,
      "learning_rate": 1.4887089176030299e-05,
      "loss": 0.5885,
      "step": 10366
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2851225137710571,
      "learning_rate": 1.4886114994027839e-05,
      "loss": 0.4843,
      "step": 10367
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4045307636260986,
      "learning_rate": 1.4885140751108348e-05,
      "loss": 0.6236,
      "step": 10368
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6083948612213135,
      "learning_rate": 1.4884166447283977e-05,
      "loss": 0.5358,
      "step": 10369
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4009491205215454,
      "learning_rate": 1.4883192082566869e-05,
      "loss": 0.6274,
      "step": 10370
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4097836017608643,
      "learning_rate": 1.4882217656969172e-05,
      "loss": 0.5452,
      "step": 10371
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2405614852905273,
      "learning_rate": 1.4881243170503035e-05,
      "loss": 0.5991,
      "step": 10372
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5827322006225586,
      "learning_rate": 1.4880268623180608e-05,
      "loss": 0.5076,
      "step": 10373
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6830322742462158,
      "learning_rate": 1.4879294015014037e-05,
      "loss": 0.6095,
      "step": 10374
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5435237884521484,
      "learning_rate": 1.4878319346015478e-05,
      "loss": 0.6644,
      "step": 10375
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.9677163362503052,
      "learning_rate": 1.4877344616197083e-05,
      "loss": 0.6058,
      "step": 10376
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6693259477615356,
      "learning_rate": 1.4876369825571e-05,
      "loss": 0.6466,
      "step": 10377
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4153659343719482,
      "learning_rate": 1.4875394974149386e-05,
      "loss": 0.6964,
      "step": 10378
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1506727933883667,
      "learning_rate": 1.487442006194439e-05,
      "loss": 0.4538,
      "step": 10379
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5418057441711426,
      "learning_rate": 1.487344508896817e-05,
      "loss": 0.5044,
      "step": 10380
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5180377960205078,
      "learning_rate": 1.4872470055232884e-05,
      "loss": 0.5609,
      "step": 10381
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3785521984100342,
      "learning_rate": 1.4871494960750681e-05,
      "loss": 0.5046,
      "step": 10382
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4453010559082031,
      "learning_rate": 1.4870519805533723e-05,
      "loss": 0.4904,
      "step": 10383
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6189279556274414,
      "learning_rate": 1.4869544589594167e-05,
      "loss": 0.6422,
      "step": 10384
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6283501386642456,
      "learning_rate": 1.4868569312944171e-05,
      "loss": 0.5773,
      "step": 10385
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2832518815994263,
      "learning_rate": 1.4867593975595895e-05,
      "loss": 0.5235,
      "step": 10386
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6808804273605347,
      "learning_rate": 1.4866618577561497e-05,
      "loss": 0.5733,
      "step": 10387
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5534698963165283,
      "learning_rate": 1.4865643118853137e-05,
      "loss": 0.4905,
      "step": 10388
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6515799760818481,
      "learning_rate": 1.4864667599482982e-05,
      "loss": 0.5454,
      "step": 10389
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5465569496154785,
      "learning_rate": 1.4863692019463187e-05,
      "loss": 0.461,
      "step": 10390
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2983849048614502,
      "learning_rate": 1.4862716378805919e-05,
      "loss": 0.4762,
      "step": 10391
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5755294561386108,
      "learning_rate": 1.4861740677523341e-05,
      "loss": 0.6749,
      "step": 10392
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.4182277917861938,
      "learning_rate": 1.4860764915627615e-05,
      "loss": 0.5986,
      "step": 10393
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2936807870864868,
      "learning_rate": 1.4859789093130911e-05,
      "loss": 0.5167,
      "step": 10394
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.521372675895691,
      "learning_rate": 1.485881321004539e-05,
      "loss": 0.6012,
      "step": 10395
    },
    {
      "epoch": 1.08,
      "grad_norm": 3.220806360244751,
      "learning_rate": 1.4857837266383224e-05,
      "loss": 0.5889,
      "step": 10396
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5348517894744873,
      "learning_rate": 1.4856861262156577e-05,
      "loss": 0.5402,
      "step": 10397
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5887261629104614,
      "learning_rate": 1.4855885197377615e-05,
      "loss": 0.4593,
      "step": 10398
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7129909992218018,
      "learning_rate": 1.4854909072058513e-05,
      "loss": 0.5825,
      "step": 10399
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.279463529586792,
      "learning_rate": 1.4853932886211436e-05,
      "loss": 0.4444,
      "step": 10400
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7480428218841553,
      "learning_rate": 1.4852956639848554e-05,
      "loss": 0.5632,
      "step": 10401
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5262229442596436,
      "learning_rate": 1.4851980332982043e-05,
      "loss": 0.6165,
      "step": 10402
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5096291303634644,
      "learning_rate": 1.4851003965624069e-05,
      "loss": 0.5416,
      "step": 10403
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4730238914489746,
      "learning_rate": 1.4850027537786811e-05,
      "loss": 0.6469,
      "step": 10404
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2942715883255005,
      "learning_rate": 1.4849051049482437e-05,
      "loss": 0.5724,
      "step": 10405
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5480436086654663,
      "learning_rate": 1.4848074500723123e-05,
      "loss": 0.5271,
      "step": 10406
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.556351900100708,
      "learning_rate": 1.4847097891521048e-05,
      "loss": 0.5423,
      "step": 10407
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5664699077606201,
      "learning_rate": 1.484612122188838e-05,
      "loss": 0.5527,
      "step": 10408
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5577248334884644,
      "learning_rate": 1.4845144491837302e-05,
      "loss": 0.592,
      "step": 10409
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.8352776765823364,
      "learning_rate": 1.484416770137999e-05,
      "loss": 0.4147,
      "step": 10410
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5870161056518555,
      "learning_rate": 1.484319085052862e-05,
      "loss": 0.6041,
      "step": 10411
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.188389539718628,
      "learning_rate": 1.4842213939295373e-05,
      "loss": 0.5451,
      "step": 10412
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5189927816390991,
      "learning_rate": 1.4841236967692424e-05,
      "loss": 0.6421,
      "step": 10413
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5457020998001099,
      "learning_rate": 1.484025993573196e-05,
      "loss": 0.6651,
      "step": 10414
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7595614194869995,
      "learning_rate": 1.4839282843426157e-05,
      "loss": 0.4962,
      "step": 10415
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1635260581970215,
      "learning_rate": 1.48383056907872e-05,
      "loss": 0.5812,
      "step": 10416
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6901099681854248,
      "learning_rate": 1.4837328477827268e-05,
      "loss": 0.4547,
      "step": 10417
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.961705207824707,
      "learning_rate": 1.4836351204558548e-05,
      "loss": 0.6351,
      "step": 10418
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3138877153396606,
      "learning_rate": 1.4835373870993222e-05,
      "loss": 0.4466,
      "step": 10419
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.660884976387024,
      "learning_rate": 1.4834396477143473e-05,
      "loss": 0.6539,
      "step": 10420
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.539903163909912,
      "learning_rate": 1.483341902302149e-05,
      "loss": 0.5793,
      "step": 10421
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3772144317626953,
      "learning_rate": 1.4832441508639459e-05,
      "loss": 0.5065,
      "step": 10422
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5407062768936157,
      "learning_rate": 1.4831463934009564e-05,
      "loss": 0.4931,
      "step": 10423
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2622120380401611,
      "learning_rate": 1.4830486299143997e-05,
      "loss": 0.4748,
      "step": 10424
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3567912578582764,
      "learning_rate": 1.4829508604054942e-05,
      "loss": 0.5551,
      "step": 10425
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4465540647506714,
      "learning_rate": 1.4828530848754592e-05,
      "loss": 0.6122,
      "step": 10426
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.67405104637146,
      "learning_rate": 1.4827553033255133e-05,
      "loss": 0.5009,
      "step": 10427
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7358211278915405,
      "learning_rate": 1.4826575157568761e-05,
      "loss": 0.7258,
      "step": 10428
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2870723009109497,
      "learning_rate": 1.4825597221707666e-05,
      "loss": 0.5472,
      "step": 10429
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2787317037582397,
      "learning_rate": 1.4824619225684038e-05,
      "loss": 0.4472,
      "step": 10430
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5604921579360962,
      "learning_rate": 1.482364116951007e-05,
      "loss": 0.5251,
      "step": 10431
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5155147314071655,
      "learning_rate": 1.4822663053197957e-05,
      "loss": 0.5245,
      "step": 10432
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3292146921157837,
      "learning_rate": 1.4821684876759896e-05,
      "loss": 0.524,
      "step": 10433
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7983644008636475,
      "learning_rate": 1.4820706640208076e-05,
      "loss": 0.564,
      "step": 10434
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4954639673233032,
      "learning_rate": 1.48197283435547e-05,
      "loss": 0.4781,
      "step": 10435
    },
    {
      "epoch": 1.08,
      "grad_norm": 79.95436096191406,
      "learning_rate": 1.4818749986811962e-05,
      "loss": 0.9817,
      "step": 10436
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2514506578445435,
      "learning_rate": 1.4817771569992056e-05,
      "loss": 0.5462,
      "step": 10437
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3484084606170654,
      "learning_rate": 1.4816793093107187e-05,
      "loss": 0.4562,
      "step": 10438
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2583900690078735,
      "learning_rate": 1.481581455616955e-05,
      "loss": 0.5691,
      "step": 10439
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.658092975616455,
      "learning_rate": 1.4814835959191344e-05,
      "loss": 0.6181,
      "step": 10440
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5428792238235474,
      "learning_rate": 1.4813857302184774e-05,
      "loss": 0.5312,
      "step": 10441
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.479854941368103,
      "learning_rate": 1.4812878585162034e-05,
      "loss": 0.5787,
      "step": 10442
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7171905040740967,
      "learning_rate": 1.4811899808135333e-05,
      "loss": 0.5381,
      "step": 10443
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4204721450805664,
      "learning_rate": 1.481092097111687e-05,
      "loss": 0.52,
      "step": 10444
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7792414426803589,
      "learning_rate": 1.4809942074118847e-05,
      "loss": 0.5122,
      "step": 10445
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4310944080352783,
      "learning_rate": 1.4808963117153476e-05,
      "loss": 0.5268,
      "step": 10446
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.521911382675171,
      "learning_rate": 1.4807984100232953e-05,
      "loss": 0.5967,
      "step": 10447
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6626770496368408,
      "learning_rate": 1.480700502336949e-05,
      "loss": 0.6061,
      "step": 10448
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5901941061019897,
      "learning_rate": 1.480602588657529e-05,
      "loss": 0.54,
      "step": 10449
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5529686212539673,
      "learning_rate": 1.480504668986256e-05,
      "loss": 0.4927,
      "step": 10450
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.477419376373291,
      "learning_rate": 1.4804067433243511e-05,
      "loss": 0.5753,
      "step": 10451
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.47996187210083,
      "learning_rate": 1.480308811673035e-05,
      "loss": 0.549,
      "step": 10452
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.617017388343811,
      "learning_rate": 1.4802108740335285e-05,
      "loss": 0.4639,
      "step": 10453
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.817623257637024,
      "learning_rate": 1.4801129304070532e-05,
      "loss": 0.6323,
      "step": 10454
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4018704891204834,
      "learning_rate": 1.4800149807948296e-05,
      "loss": 0.4844,
      "step": 10455
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.565295696258545,
      "learning_rate": 1.479917025198079e-05,
      "loss": 0.5625,
      "step": 10456
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5599173307418823,
      "learning_rate": 1.479819063618023e-05,
      "loss": 0.4908,
      "step": 10457
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7271556854248047,
      "learning_rate": 1.479721096055882e-05,
      "loss": 0.6479,
      "step": 10458
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6518480777740479,
      "learning_rate": 1.4796231225128788e-05,
      "loss": 0.6636,
      "step": 10459
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5440006256103516,
      "learning_rate": 1.4795251429902336e-05,
      "loss": 0.4235,
      "step": 10460
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.486510157585144,
      "learning_rate": 1.4794271574891686e-05,
      "loss": 0.5424,
      "step": 10461
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.6237982511520386,
      "learning_rate": 1.4793291660109052e-05,
      "loss": 0.5274,
      "step": 10462
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2859609127044678,
      "learning_rate": 1.4792311685566652e-05,
      "loss": 0.6089,
      "step": 10463
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3271323442459106,
      "learning_rate": 1.4791331651276704e-05,
      "loss": 0.4728,
      "step": 10464
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5875637531280518,
      "learning_rate": 1.4790351557251428e-05,
      "loss": 0.5034,
      "step": 10465
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.83345365524292,
      "learning_rate": 1.4789371403503037e-05,
      "loss": 0.5364,
      "step": 10466
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4170951843261719,
      "learning_rate": 1.4788391190043759e-05,
      "loss": 0.531,
      "step": 10467
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.416111707687378,
      "learning_rate": 1.4787410916885807e-05,
      "loss": 0.5327,
      "step": 10468
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.834234595298767,
      "learning_rate": 1.4786430584041408e-05,
      "loss": 0.5991,
      "step": 10469
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5044701099395752,
      "learning_rate": 1.4785450191522783e-05,
      "loss": 0.6188,
      "step": 10470
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.7103943824768066,
      "learning_rate": 1.4784469739342153e-05,
      "loss": 0.6519,
      "step": 10471
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3628901243209839,
      "learning_rate": 1.4783489227511747e-05,
      "loss": 0.3935,
      "step": 10472
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.607256531715393,
      "learning_rate": 1.4782508656043782e-05,
      "loss": 0.6758,
      "step": 10473
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4471893310546875,
      "learning_rate": 1.4781528024950484e-05,
      "loss": 0.5474,
      "step": 10474
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.8777899742126465,
      "learning_rate": 1.4780547334244087e-05,
      "loss": 0.6093,
      "step": 10475
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.682255744934082,
      "learning_rate": 1.4779566583936806e-05,
      "loss": 0.5556,
      "step": 10476
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.8227125406265259,
      "learning_rate": 1.4778585774040878e-05,
      "loss": 0.6527,
      "step": 10477
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5574740171432495,
      "learning_rate": 1.4777604904568527e-05,
      "loss": 0.5217,
      "step": 10478
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.473060965538025,
      "learning_rate": 1.4776623975531984e-05,
      "loss": 0.5943,
      "step": 10479
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3479301929473877,
      "learning_rate": 1.4775642986943477e-05,
      "loss": 0.5025,
      "step": 10480
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.636818766593933,
      "learning_rate": 1.4774661938815233e-05,
      "loss": 0.563,
      "step": 10481
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1920229196548462,
      "learning_rate": 1.477368083115949e-05,
      "loss": 0.4892,
      "step": 10482
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.613482117652893,
      "learning_rate": 1.4772699663988474e-05,
      "loss": 0.6073,
      "step": 10483
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.9126715660095215,
      "learning_rate": 1.4771718437314419e-05,
      "loss": 0.584,
      "step": 10484
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4311378002166748,
      "learning_rate": 1.4770737151149564e-05,
      "loss": 0.5928,
      "step": 10485
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4415959119796753,
      "learning_rate": 1.4769755805506134e-05,
      "loss": 0.5332,
      "step": 10486
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5068780183792114,
      "learning_rate": 1.4768774400396368e-05,
      "loss": 0.5981,
      "step": 10487
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2943956851959229,
      "learning_rate": 1.4767792935832506e-05,
      "loss": 0.6312,
      "step": 10488
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4859410524368286,
      "learning_rate": 1.4766811411826776e-05,
      "loss": 0.5985,
      "step": 10489
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.5756323337554932,
      "learning_rate": 1.476582982839142e-05,
      "loss": 0.5366,
      "step": 10490
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.426287055015564,
      "learning_rate": 1.4764848185538677e-05,
      "loss": 0.3945,
      "step": 10491
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.2587600946426392,
      "learning_rate": 1.476386648328078e-05,
      "loss": 0.5023,
      "step": 10492
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6491692066192627,
      "learning_rate": 1.4762884721629974e-05,
      "loss": 0.7068,
      "step": 10493
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1613560914993286,
      "learning_rate": 1.4761902900598495e-05,
      "loss": 0.532,
      "step": 10494
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.618321418762207,
      "learning_rate": 1.4760921020198583e-05,
      "loss": 0.6279,
      "step": 10495
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.2687205076217651,
      "learning_rate": 1.4759939080442484e-05,
      "loss": 0.6538,
      "step": 10496
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4277913570404053,
      "learning_rate": 1.4758957081342437e-05,
      "loss": 0.5897,
      "step": 10497
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4607503414154053,
      "learning_rate": 1.475797502291069e-05,
      "loss": 0.4967,
      "step": 10498
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6877456903457642,
      "learning_rate": 1.475699290515948e-05,
      "loss": 0.6036,
      "step": 10499
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4984264373779297,
      "learning_rate": 1.475601072810105e-05,
      "loss": 0.5571,
      "step": 10500
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6036280393600464,
      "learning_rate": 1.4755028491747654e-05,
      "loss": 0.5973,
      "step": 10501
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5475648641586304,
      "learning_rate": 1.4754046196111531e-05,
      "loss": 0.5747,
      "step": 10502
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3768272399902344,
      "learning_rate": 1.475306384120493e-05,
      "loss": 0.5946,
      "step": 10503
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4972788095474243,
      "learning_rate": 1.4752081427040099e-05,
      "loss": 0.5953,
      "step": 10504
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.895093560218811,
      "learning_rate": 1.4751098953629284e-05,
      "loss": 0.7009,
      "step": 10505
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3865594863891602,
      "learning_rate": 1.4750116420984736e-05,
      "loss": 0.5224,
      "step": 10506
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.8186122179031372,
      "learning_rate": 1.4749133829118703e-05,
      "loss": 0.4633,
      "step": 10507
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3063725233078003,
      "learning_rate": 1.4748151178043437e-05,
      "loss": 0.6194,
      "step": 10508
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4302096366882324,
      "learning_rate": 1.4747168467771186e-05,
      "loss": 0.3825,
      "step": 10509
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7099586725234985,
      "learning_rate": 1.4746185698314206e-05,
      "loss": 0.5958,
      "step": 10510
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.443202018737793,
      "learning_rate": 1.4745202869684749e-05,
      "loss": 0.4307,
      "step": 10511
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5394545793533325,
      "learning_rate": 1.4744219981895062e-05,
      "loss": 0.5295,
      "step": 10512
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7078157663345337,
      "learning_rate": 1.4743237034957406e-05,
      "loss": 0.6221,
      "step": 10513
    },
    {
      "epoch": 1.09,
      "grad_norm": 2.5172085762023926,
      "learning_rate": 1.4742254028884037e-05,
      "loss": 0.6802,
      "step": 10514
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4051628112792969,
      "learning_rate": 1.4741270963687204e-05,
      "loss": 0.4665,
      "step": 10515
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5790694952011108,
      "learning_rate": 1.4740287839379164e-05,
      "loss": 0.646,
      "step": 10516
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1983426809310913,
      "learning_rate": 1.473930465597218e-05,
      "loss": 0.4401,
      "step": 10517
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.502347707748413,
      "learning_rate": 1.4738321413478505e-05,
      "loss": 0.613,
      "step": 10518
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7420237064361572,
      "learning_rate": 1.47373381119104e-05,
      "loss": 0.5486,
      "step": 10519
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5941115617752075,
      "learning_rate": 1.473635475128012e-05,
      "loss": 0.6054,
      "step": 10520
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4813220500946045,
      "learning_rate": 1.4735371331599928e-05,
      "loss": 0.6272,
      "step": 10521
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5069118738174438,
      "learning_rate": 1.4734387852882086e-05,
      "loss": 0.5656,
      "step": 10522
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7086378335952759,
      "learning_rate": 1.473340431513885e-05,
      "loss": 0.4806,
      "step": 10523
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.503248691558838,
      "learning_rate": 1.473242071838249e-05,
      "loss": 0.5484,
      "step": 10524
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4297406673431396,
      "learning_rate": 1.4731437062625263e-05,
      "loss": 0.6202,
      "step": 10525
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3976585865020752,
      "learning_rate": 1.4730453347879432e-05,
      "loss": 0.634,
      "step": 10526
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3660773038864136,
      "learning_rate": 1.4729469574157267e-05,
      "loss": 0.435,
      "step": 10527
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5353367328643799,
      "learning_rate": 1.4728485741471028e-05,
      "loss": 0.4869,
      "step": 10528
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5968867540359497,
      "learning_rate": 1.4727501849832981e-05,
      "loss": 0.5916,
      "step": 10529
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5607448816299438,
      "learning_rate": 1.4726517899255397e-05,
      "loss": 0.5716,
      "step": 10530
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.514723300933838,
      "learning_rate": 1.472553388975054e-05,
      "loss": 0.5492,
      "step": 10531
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6858294010162354,
      "learning_rate": 1.4724549821330678e-05,
      "loss": 0.6289,
      "step": 10532
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4324846267700195,
      "learning_rate": 1.472356569400808e-05,
      "loss": 0.5453,
      "step": 10533
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.690711498260498,
      "learning_rate": 1.4722581507795013e-05,
      "loss": 0.5983,
      "step": 10534
    },
    {
      "epoch": 1.09,
      "grad_norm": 2.1398661136627197,
      "learning_rate": 1.4721597262703755e-05,
      "loss": 0.6837,
      "step": 10535
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.446511149406433,
      "learning_rate": 1.4720612958746568e-05,
      "loss": 0.5702,
      "step": 10536
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.849006175994873,
      "learning_rate": 1.471962859593573e-05,
      "loss": 0.7376,
      "step": 10537
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3136968612670898,
      "learning_rate": 1.4718644174283507e-05,
      "loss": 0.5348,
      "step": 10538
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5376919507980347,
      "learning_rate": 1.4717659693802176e-05,
      "loss": 0.5556,
      "step": 10539
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4523175954818726,
      "learning_rate": 1.4716675154504015e-05,
      "loss": 0.5253,
      "step": 10540
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7354283332824707,
      "learning_rate": 1.4715690556401292e-05,
      "loss": 0.598,
      "step": 10541
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5881394147872925,
      "learning_rate": 1.4714705899506285e-05,
      "loss": 0.4393,
      "step": 10542
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.8452478647232056,
      "learning_rate": 1.4713721183831271e-05,
      "loss": 0.5837,
      "step": 10543
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4488575458526611,
      "learning_rate": 1.4712736409388524e-05,
      "loss": 0.5715,
      "step": 10544
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6999008655548096,
      "learning_rate": 1.4711751576190325e-05,
      "loss": 0.5546,
      "step": 10545
    },
    {
      "epoch": 1.09,
      "grad_norm": 2.094083547592163,
      "learning_rate": 1.471076668424895e-05,
      "loss": 0.5285,
      "step": 10546
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6871665716171265,
      "learning_rate": 1.4709781733576676e-05,
      "loss": 0.6184,
      "step": 10547
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4427869319915771,
      "learning_rate": 1.470879672418579e-05,
      "loss": 0.6076,
      "step": 10548
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.9622888565063477,
      "learning_rate": 1.4707811656088567e-05,
      "loss": 0.6066,
      "step": 10549
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5332396030426025,
      "learning_rate": 1.4706826529297286e-05,
      "loss": 0.6008,
      "step": 10550
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.605741262435913,
      "learning_rate": 1.4705841343824234e-05,
      "loss": 0.4799,
      "step": 10551
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6888681650161743,
      "learning_rate": 1.470485609968169e-05,
      "loss": 0.6396,
      "step": 10552
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5336002111434937,
      "learning_rate": 1.4703870796881942e-05,
      "loss": 0.577,
      "step": 10553
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4442543983459473,
      "learning_rate": 1.4702885435437272e-05,
      "loss": 0.5946,
      "step": 10554
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4081004858016968,
      "learning_rate": 1.470190001535996e-05,
      "loss": 0.5009,
      "step": 10555
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.349104881286621,
      "learning_rate": 1.4700914536662298e-05,
      "loss": 0.3598,
      "step": 10556
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6826012134552002,
      "learning_rate": 1.4699928999356569e-05,
      "loss": 0.637,
      "step": 10557
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.8520010709762573,
      "learning_rate": 1.4698943403455064e-05,
      "loss": 0.5618,
      "step": 10558
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4763818979263306,
      "learning_rate": 1.4697957748970064e-05,
      "loss": 0.6341,
      "step": 10559
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4423351287841797,
      "learning_rate": 1.4696972035913863e-05,
      "loss": 0.5591,
      "step": 10560
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7413420677185059,
      "learning_rate": 1.469598626429875e-05,
      "loss": 0.5591,
      "step": 10561
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.2459688186645508,
      "learning_rate": 1.4695000434137012e-05,
      "loss": 0.5035,
      "step": 10562
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.520601511001587,
      "learning_rate": 1.4694014545440942e-05,
      "loss": 0.5883,
      "step": 10563
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5193594694137573,
      "learning_rate": 1.4693028598222831e-05,
      "loss": 0.5028,
      "step": 10564
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6513721942901611,
      "learning_rate": 1.469204259249497e-05,
      "loss": 0.5496,
      "step": 10565
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.619018316268921,
      "learning_rate": 1.4691056528269653e-05,
      "loss": 0.7149,
      "step": 10566
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.292165994644165,
      "learning_rate": 1.4690070405559175e-05,
      "loss": 0.547,
      "step": 10567
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5876270532608032,
      "learning_rate": 1.4689084224375827e-05,
      "loss": 0.4925,
      "step": 10568
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.564048171043396,
      "learning_rate": 1.4688097984731908e-05,
      "loss": 0.6271,
      "step": 10569
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5621367692947388,
      "learning_rate": 1.4687111686639712e-05,
      "loss": 0.5137,
      "step": 10570
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4457167387008667,
      "learning_rate": 1.4686125330111533e-05,
      "loss": 0.5066,
      "step": 10571
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3200184106826782,
      "learning_rate": 1.4685138915159673e-05,
      "loss": 0.5694,
      "step": 10572
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6664401292800903,
      "learning_rate": 1.4684152441796427e-05,
      "loss": 0.5687,
      "step": 10573
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5265885591506958,
      "learning_rate": 1.4683165910034094e-05,
      "loss": 0.6392,
      "step": 10574
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5619531869888306,
      "learning_rate": 1.4682179319884973e-05,
      "loss": 0.6332,
      "step": 10575
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.593525767326355,
      "learning_rate": 1.4681192671361366e-05,
      "loss": 0.6219,
      "step": 10576
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.8227769136428833,
      "learning_rate": 1.4680205964475574e-05,
      "loss": 0.6209,
      "step": 10577
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.629446268081665,
      "learning_rate": 1.4679219199239893e-05,
      "loss": 0.5431,
      "step": 10578
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.359007716178894,
      "learning_rate": 1.4678232375666634e-05,
      "loss": 0.5025,
      "step": 10579
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.703203558921814,
      "learning_rate": 1.4677245493768096e-05,
      "loss": 0.4986,
      "step": 10580
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3061808347702026,
      "learning_rate": 1.4676258553556582e-05,
      "loss": 0.4821,
      "step": 10581
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.3898085355758667,
      "learning_rate": 1.4675271555044399e-05,
      "loss": 0.5451,
      "step": 10582
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.484761118888855,
      "learning_rate": 1.4674284498243849e-05,
      "loss": 0.493,
      "step": 10583
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4775818586349487,
      "learning_rate": 1.4673297383167242e-05,
      "loss": 0.6564,
      "step": 10584
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.757359266281128,
      "learning_rate": 1.4672310209826879e-05,
      "loss": 0.6185,
      "step": 10585
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.6248129606246948,
      "learning_rate": 1.4671322978235073e-05,
      "loss": 0.6588,
      "step": 10586
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.437945008277893,
      "learning_rate": 1.4670335688404129e-05,
      "loss": 0.5517,
      "step": 10587
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4564169645309448,
      "learning_rate": 1.4669348340346356e-05,
      "loss": 0.5169,
      "step": 10588
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.504501223564148,
      "learning_rate": 1.4668360934074069e-05,
      "loss": 0.6462,
      "step": 10589
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.533326268196106,
      "learning_rate": 1.4667373469599571e-05,
      "loss": 0.5872,
      "step": 10590
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5156419277191162,
      "learning_rate": 1.4666385946935176e-05,
      "loss": 0.5223,
      "step": 10591
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5148576498031616,
      "learning_rate": 1.4665398366093198e-05,
      "loss": 0.5469,
      "step": 10592
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3108794689178467,
      "learning_rate": 1.4664410727085945e-05,
      "loss": 0.5248,
      "step": 10593
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5322750806808472,
      "learning_rate": 1.4663423029925732e-05,
      "loss": 0.5893,
      "step": 10594
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.8167520761489868,
      "learning_rate": 1.4662435274624876e-05,
      "loss": 0.5982,
      "step": 10595
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3672804832458496,
      "learning_rate": 1.4661447461195688e-05,
      "loss": 0.4081,
      "step": 10596
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3914774656295776,
      "learning_rate": 1.4660459589650486e-05,
      "loss": 0.6035,
      "step": 10597
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6521767377853394,
      "learning_rate": 1.4659471660001586e-05,
      "loss": 0.5977,
      "step": 10598
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5907987356185913,
      "learning_rate": 1.46584836722613e-05,
      "loss": 0.7004,
      "step": 10599
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6247316598892212,
      "learning_rate": 1.4657495626441955e-05,
      "loss": 0.5816,
      "step": 10600
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6488896608352661,
      "learning_rate": 1.465650752255586e-05,
      "loss": 0.5477,
      "step": 10601
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4799383878707886,
      "learning_rate": 1.465551936061534e-05,
      "loss": 0.6279,
      "step": 10602
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6055406332015991,
      "learning_rate": 1.465453114063271e-05,
      "loss": 0.5841,
      "step": 10603
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3064614534378052,
      "learning_rate": 1.4653542862620296e-05,
      "loss": 0.4874,
      "step": 10604
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6160372495651245,
      "learning_rate": 1.4652554526590417e-05,
      "loss": 0.5975,
      "step": 10605
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7473481893539429,
      "learning_rate": 1.4651566132555392e-05,
      "loss": 0.7113,
      "step": 10606
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6592458486557007,
      "learning_rate": 1.4650577680527546e-05,
      "loss": 0.6119,
      "step": 10607
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6016056537628174,
      "learning_rate": 1.4649589170519206e-05,
      "loss": 0.62,
      "step": 10608
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6243400573730469,
      "learning_rate": 1.4648600602542688e-05,
      "loss": 0.52,
      "step": 10609
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4001095294952393,
      "learning_rate": 1.4647611976610325e-05,
      "loss": 0.5094,
      "step": 10610
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.489525556564331,
      "learning_rate": 1.4646623292734438e-05,
      "loss": 0.4993,
      "step": 10611
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.64828622341156,
      "learning_rate": 1.4645634550927355e-05,
      "loss": 0.4468,
      "step": 10612
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7733267545700073,
      "learning_rate": 1.46446457512014e-05,
      "loss": 0.5078,
      "step": 10613
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4599415063858032,
      "learning_rate": 1.4643656893568905e-05,
      "loss": 0.6256,
      "step": 10614
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6859711408615112,
      "learning_rate": 1.4642667978042198e-05,
      "loss": 0.6088,
      "step": 10615
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7119314670562744,
      "learning_rate": 1.4641679004633604e-05,
      "loss": 0.4769,
      "step": 10616
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3994885683059692,
      "learning_rate": 1.4640689973355457e-05,
      "loss": 0.6856,
      "step": 10617
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7476822137832642,
      "learning_rate": 1.4639700884220087e-05,
      "loss": 0.5602,
      "step": 10618
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6182302236557007,
      "learning_rate": 1.4638711737239823e-05,
      "loss": 0.5766,
      "step": 10619
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3604679107666016,
      "learning_rate": 1.4637722532426998e-05,
      "loss": 0.5822,
      "step": 10620
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6399191617965698,
      "learning_rate": 1.4636733269793948e-05,
      "loss": 0.5421,
      "step": 10621
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5067375898361206,
      "learning_rate": 1.4635743949353003e-05,
      "loss": 0.4536,
      "step": 10622
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.480992317199707,
      "learning_rate": 1.4634754571116498e-05,
      "loss": 0.6254,
      "step": 10623
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.316903829574585,
      "learning_rate": 1.4633765135096769e-05,
      "loss": 0.5926,
      "step": 10624
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4696228504180908,
      "learning_rate": 1.4632775641306149e-05,
      "loss": 0.6054,
      "step": 10625
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4884867668151855,
      "learning_rate": 1.4631786089756979e-05,
      "loss": 0.5248,
      "step": 10626
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4500672817230225,
      "learning_rate": 1.463079648046159e-05,
      "loss": 0.5745,
      "step": 10627
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7725974321365356,
      "learning_rate": 1.4629806813432326e-05,
      "loss": 0.5736,
      "step": 10628
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3320077657699585,
      "learning_rate": 1.4628817088681522e-05,
      "loss": 0.5802,
      "step": 10629
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6655174493789673,
      "learning_rate": 1.4627827306221516e-05,
      "loss": 0.6181,
      "step": 10630
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3418806791305542,
      "learning_rate": 1.4626837466064654e-05,
      "loss": 0.6531,
      "step": 10631
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3776322603225708,
      "learning_rate": 1.462584756822327e-05,
      "loss": 0.6439,
      "step": 10632
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6997580528259277,
      "learning_rate": 1.4624857612709709e-05,
      "loss": 0.5823,
      "step": 10633
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7018650770187378,
      "learning_rate": 1.4623867599536312e-05,
      "loss": 0.641,
      "step": 10634
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5274288654327393,
      "learning_rate": 1.462287752871542e-05,
      "loss": 0.5999,
      "step": 10635
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.2281720638275146,
      "learning_rate": 1.462188740025938e-05,
      "loss": 0.4516,
      "step": 10636
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4465279579162598,
      "learning_rate": 1.4620897214180537e-05,
      "loss": 0.641,
      "step": 10637
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4763474464416504,
      "learning_rate": 1.4619906970491232e-05,
      "loss": 0.6053,
      "step": 10638
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7368711233139038,
      "learning_rate": 1.4618916669203813e-05,
      "loss": 0.5532,
      "step": 10639
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4363517761230469,
      "learning_rate": 1.4617926310330628e-05,
      "loss": 0.4177,
      "step": 10640
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.361981749534607,
      "learning_rate": 1.4616935893884018e-05,
      "loss": 0.5292,
      "step": 10641
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3724075555801392,
      "learning_rate": 1.4615945419876341e-05,
      "loss": 0.5762,
      "step": 10642
    },
    {
      "epoch": 1.1,
      "grad_norm": 2.113436698913574,
      "learning_rate": 1.4614954888319936e-05,
      "loss": 0.5847,
      "step": 10643
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4586327075958252,
      "learning_rate": 1.4613964299227158e-05,
      "loss": 0.5011,
      "step": 10644
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5111786127090454,
      "learning_rate": 1.4612973652610354e-05,
      "loss": 0.5402,
      "step": 10645
    },
    {
      "epoch": 1.1,
      "grad_norm": 2.1837425231933594,
      "learning_rate": 1.4611982948481877e-05,
      "loss": 0.6392,
      "step": 10646
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7027901411056519,
      "learning_rate": 1.4610992186854078e-05,
      "loss": 0.6113,
      "step": 10647
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.480653166770935,
      "learning_rate": 1.4610001367739308e-05,
      "loss": 0.5962,
      "step": 10648
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5509287118911743,
      "learning_rate": 1.460901049114992e-05,
      "loss": 0.4685,
      "step": 10649
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6398568153381348,
      "learning_rate": 1.460801955709827e-05,
      "loss": 0.6524,
      "step": 10650
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7729878425598145,
      "learning_rate": 1.4607028565596709e-05,
      "loss": 0.607,
      "step": 10651
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3261018991470337,
      "learning_rate": 1.4606037516657594e-05,
      "loss": 0.6216,
      "step": 10652
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5144459009170532,
      "learning_rate": 1.460504641029328e-05,
      "loss": 0.4702,
      "step": 10653
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.616137146949768,
      "learning_rate": 1.4604055246516125e-05,
      "loss": 0.6031,
      "step": 10654
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5103281736373901,
      "learning_rate": 1.4603064025338486e-05,
      "loss": 0.6888,
      "step": 10655
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.2097502946853638,
      "learning_rate": 1.4602072746772717e-05,
      "loss": 0.4595,
      "step": 10656
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5643388032913208,
      "learning_rate": 1.4601081410831182e-05,
      "loss": 0.6422,
      "step": 10657
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5405972003936768,
      "learning_rate": 1.4600090017526239e-05,
      "loss": 0.6897,
      "step": 10658
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.2432963848114014,
      "learning_rate": 1.4599098566870244e-05,
      "loss": 0.4722,
      "step": 10659
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4500517845153809,
      "learning_rate": 1.4598107058875565e-05,
      "loss": 0.5065,
      "step": 10660
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5284157991409302,
      "learning_rate": 1.4597115493554557e-05,
      "loss": 0.6562,
      "step": 10661
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5320085287094116,
      "learning_rate": 1.4596123870919585e-05,
      "loss": 0.4101,
      "step": 10662
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5306532382965088,
      "learning_rate": 1.4595132190983012e-05,
      "loss": 0.6567,
      "step": 10663
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7068397998809814,
      "learning_rate": 1.4594140453757199e-05,
      "loss": 0.4497,
      "step": 10664
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3324177265167236,
      "learning_rate": 1.4593148659254517e-05,
      "loss": 0.5236,
      "step": 10665
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.621752381324768,
      "learning_rate": 1.4592156807487324e-05,
      "loss": 0.6047,
      "step": 10666
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4035964012145996,
      "learning_rate": 1.4591164898467986e-05,
      "loss": 0.5124,
      "step": 10667
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.455064058303833,
      "learning_rate": 1.4590172932208878e-05,
      "loss": 0.6306,
      "step": 10668
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.8174864053726196,
      "learning_rate": 1.4589180908722356e-05,
      "loss": 0.7007,
      "step": 10669
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4216729402542114,
      "learning_rate": 1.4588188828020796e-05,
      "loss": 0.462,
      "step": 10670
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6625556945800781,
      "learning_rate": 1.4587196690116562e-05,
      "loss": 0.5389,
      "step": 10671
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.305529236793518,
      "learning_rate": 1.4586204495022025e-05,
      "loss": 0.5358,
      "step": 10672
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4965471029281616,
      "learning_rate": 1.4585212242749558e-05,
      "loss": 0.6583,
      "step": 10673
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.6117151975631714,
      "learning_rate": 1.4584219933311525e-05,
      "loss": 0.6415,
      "step": 10674
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.913214087486267,
      "learning_rate": 1.4583227566720304e-05,
      "loss": 0.6025,
      "step": 10675
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4196038246154785,
      "learning_rate": 1.4582235142988265e-05,
      "loss": 0.5064,
      "step": 10676
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4946180582046509,
      "learning_rate": 1.458124266212778e-05,
      "loss": 0.6189,
      "step": 10677
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.378512978553772,
      "learning_rate": 1.4580250124151225e-05,
      "loss": 0.6798,
      "step": 10678
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3439732789993286,
      "learning_rate": 1.4579257529070969e-05,
      "loss": 0.5936,
      "step": 10679
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.7418456077575684,
      "learning_rate": 1.457826487689939e-05,
      "loss": 0.6201,
      "step": 10680
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4104442596435547,
      "learning_rate": 1.4577272167648868e-05,
      "loss": 0.5076,
      "step": 10681
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.4270577430725098,
      "learning_rate": 1.4576279401331775e-05,
      "loss": 0.5036,
      "step": 10682
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.8154598474502563,
      "learning_rate": 1.457528657796049e-05,
      "loss": 0.5722,
      "step": 10683
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.8592325448989868,
      "learning_rate": 1.4574293697547388e-05,
      "loss": 0.5951,
      "step": 10684
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6861317157745361,
      "learning_rate": 1.457330076010485e-05,
      "loss": 0.7569,
      "step": 10685
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6597838401794434,
      "learning_rate": 1.4572307765645255e-05,
      "loss": 0.597,
      "step": 10686
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6816062927246094,
      "learning_rate": 1.4571314714180982e-05,
      "loss": 0.516,
      "step": 10687
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.633542537689209,
      "learning_rate": 1.4570321605724413e-05,
      "loss": 0.5965,
      "step": 10688
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3216102123260498,
      "learning_rate": 1.456932844028793e-05,
      "loss": 0.4348,
      "step": 10689
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4095736742019653,
      "learning_rate": 1.4568335217883913e-05,
      "loss": 0.6073,
      "step": 10690
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3490784168243408,
      "learning_rate": 1.4567341938524751e-05,
      "loss": 0.5455,
      "step": 10691
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.405056118965149,
      "learning_rate": 1.4566348602222817e-05,
      "loss": 0.5295,
      "step": 10692
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6113935708999634,
      "learning_rate": 1.4565355208990502e-05,
      "loss": 0.5118,
      "step": 10693
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4962704181671143,
      "learning_rate": 1.4564361758840191e-05,
      "loss": 0.5149,
      "step": 10694
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.1649398803710938,
      "learning_rate": 1.4563368251784268e-05,
      "loss": 0.5699,
      "step": 10695
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.7387984991073608,
      "learning_rate": 1.4562374687835123e-05,
      "loss": 0.5039,
      "step": 10696
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.561487078666687,
      "learning_rate": 1.4561381067005138e-05,
      "loss": 0.5757,
      "step": 10697
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.254757285118103,
      "learning_rate": 1.4560387389306703e-05,
      "loss": 0.5219,
      "step": 10698
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6700278520584106,
      "learning_rate": 1.4559393654752207e-05,
      "loss": 0.5778,
      "step": 10699
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3594493865966797,
      "learning_rate": 1.455839986335404e-05,
      "loss": 0.6594,
      "step": 10700
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2340025901794434,
      "learning_rate": 1.4557406015124588e-05,
      "loss": 0.568,
      "step": 10701
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.7049072980880737,
      "learning_rate": 1.4556412110076251e-05,
      "loss": 0.6293,
      "step": 10702
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.820020318031311,
      "learning_rate": 1.4555418148221408e-05,
      "loss": 0.6035,
      "step": 10703
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6731356382369995,
      "learning_rate": 1.455442412957246e-05,
      "loss": 0.6187,
      "step": 10704
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4442834854125977,
      "learning_rate": 1.4553430054141794e-05,
      "loss": 0.5895,
      "step": 10705
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6195831298828125,
      "learning_rate": 1.4552435921941809e-05,
      "loss": 0.4778,
      "step": 10706
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6296321153640747,
      "learning_rate": 1.4551441732984897e-05,
      "loss": 0.5929,
      "step": 10707
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.359864592552185,
      "learning_rate": 1.4550447487283452e-05,
      "loss": 0.4955,
      "step": 10708
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5176548957824707,
      "learning_rate": 1.454945318484987e-05,
      "loss": 0.5531,
      "step": 10709
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4569296836853027,
      "learning_rate": 1.4548458825696548e-05,
      "loss": 0.6139,
      "step": 10710
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3594944477081299,
      "learning_rate": 1.4547464409835881e-05,
      "loss": 0.6576,
      "step": 10711
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2504085302352905,
      "learning_rate": 1.454646993728027e-05,
      "loss": 0.6458,
      "step": 10712
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5197558403015137,
      "learning_rate": 1.4545475408042112e-05,
      "loss": 0.6652,
      "step": 10713
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3661293983459473,
      "learning_rate": 1.4544480822133804e-05,
      "loss": 0.5128,
      "step": 10714
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3444056510925293,
      "learning_rate": 1.4543486179567751e-05,
      "loss": 0.6021,
      "step": 10715
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5381044149398804,
      "learning_rate": 1.4542491480356349e-05,
      "loss": 0.6242,
      "step": 10716
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6759568452835083,
      "learning_rate": 1.4541496724512e-05,
      "loss": 0.6382,
      "step": 10717
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.667923927307129,
      "learning_rate": 1.4540501912047107e-05,
      "loss": 0.4956,
      "step": 10718
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.196224570274353,
      "learning_rate": 1.4539507042974072e-05,
      "loss": 0.5924,
      "step": 10719
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3348215818405151,
      "learning_rate": 1.45385121173053e-05,
      "loss": 0.4081,
      "step": 10720
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3329169750213623,
      "learning_rate": 1.4537517135053194e-05,
      "loss": 0.6242,
      "step": 10721
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3459722995758057,
      "learning_rate": 1.4536522096230159e-05,
      "loss": 0.4836,
      "step": 10722
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4082274436950684,
      "learning_rate": 1.4535527000848599e-05,
      "loss": 0.5667,
      "step": 10723
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6250401735305786,
      "learning_rate": 1.453453184892092e-05,
      "loss": 0.6189,
      "step": 10724
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2354490756988525,
      "learning_rate": 1.4533536640459535e-05,
      "loss": 0.5761,
      "step": 10725
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5690196752548218,
      "learning_rate": 1.4532541375476843e-05,
      "loss": 0.6579,
      "step": 10726
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5119152069091797,
      "learning_rate": 1.4531546053985257e-05,
      "loss": 0.5422,
      "step": 10727
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2902181148529053,
      "learning_rate": 1.4530550675997187e-05,
      "loss": 0.6076,
      "step": 10728
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4244906902313232,
      "learning_rate": 1.4529555241525042e-05,
      "loss": 0.4614,
      "step": 10729
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3349944353103638,
      "learning_rate": 1.4528559750581232e-05,
      "loss": 0.5925,
      "step": 10730
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4016964435577393,
      "learning_rate": 1.4527564203178165e-05,
      "loss": 0.5433,
      "step": 10731
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4542803764343262,
      "learning_rate": 1.4526568599328257e-05,
      "loss": 0.5442,
      "step": 10732
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6146564483642578,
      "learning_rate": 1.4525572939043921e-05,
      "loss": 0.5889,
      "step": 10733
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2766135931015015,
      "learning_rate": 1.4524577222337565e-05,
      "loss": 0.585,
      "step": 10734
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5697513818740845,
      "learning_rate": 1.452358144922161e-05,
      "loss": 0.6584,
      "step": 10735
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5140546560287476,
      "learning_rate": 1.4522585619708465e-05,
      "loss": 0.5785,
      "step": 10736
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5768356323242188,
      "learning_rate": 1.4521589733810548e-05,
      "loss": 0.5696,
      "step": 10737
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2897692918777466,
      "learning_rate": 1.4520593791540277e-05,
      "loss": 0.5727,
      "step": 10738
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.474185585975647,
      "learning_rate": 1.4519597792910064e-05,
      "loss": 0.3748,
      "step": 10739
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.9289430379867554,
      "learning_rate": 1.4518601737932327e-05,
      "loss": 0.6047,
      "step": 10740
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2705860137939453,
      "learning_rate": 1.451760562661949e-05,
      "loss": 0.4282,
      "step": 10741
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2846873998641968,
      "learning_rate": 1.4516609458983966e-05,
      "loss": 0.5071,
      "step": 10742
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5811896324157715,
      "learning_rate": 1.451561323503818e-05,
      "loss": 0.5311,
      "step": 10743
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.274558186531067,
      "learning_rate": 1.4514616954794543e-05,
      "loss": 0.4732,
      "step": 10744
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4765634536743164,
      "learning_rate": 1.4513620618265485e-05,
      "loss": 0.4408,
      "step": 10745
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4707539081573486,
      "learning_rate": 1.4512624225463426e-05,
      "loss": 0.5585,
      "step": 10746
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4394910335540771,
      "learning_rate": 1.4511627776400785e-05,
      "loss": 0.546,
      "step": 10747
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6220765113830566,
      "learning_rate": 1.451063127108999e-05,
      "loss": 0.7216,
      "step": 10748
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6457774639129639,
      "learning_rate": 1.450963470954346e-05,
      "loss": 0.605,
      "step": 10749
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2805081605911255,
      "learning_rate": 1.4508638091773619e-05,
      "loss": 0.4605,
      "step": 10750
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4071483612060547,
      "learning_rate": 1.4507641417792898e-05,
      "loss": 0.4975,
      "step": 10751
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.1902652978897095,
      "learning_rate": 1.4506644687613718e-05,
      "loss": 0.5092,
      "step": 10752
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.397403597831726,
      "learning_rate": 1.4505647901248505e-05,
      "loss": 0.5996,
      "step": 10753
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3330259323120117,
      "learning_rate": 1.4504651058709693e-05,
      "loss": 0.4957,
      "step": 10754
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5884684324264526,
      "learning_rate": 1.4503654160009703e-05,
      "loss": 0.6767,
      "step": 10755
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6767545938491821,
      "learning_rate": 1.4502657205160968e-05,
      "loss": 0.603,
      "step": 10756
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2536654472351074,
      "learning_rate": 1.4501660194175914e-05,
      "loss": 0.5725,
      "step": 10757
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4858629703521729,
      "learning_rate": 1.4500663127066972e-05,
      "loss": 0.5728,
      "step": 10758
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.655297040939331,
      "learning_rate": 1.4499666003846575e-05,
      "loss": 0.5484,
      "step": 10759
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5008997917175293,
      "learning_rate": 1.4498668824527153e-05,
      "loss": 0.5064,
      "step": 10760
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2791368961334229,
      "learning_rate": 1.4497671589121138e-05,
      "loss": 0.5709,
      "step": 10761
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.8909170627593994,
      "learning_rate": 1.4496674297640962e-05,
      "loss": 0.6645,
      "step": 10762
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3807063102722168,
      "learning_rate": 1.4495676950099063e-05,
      "loss": 0.6592,
      "step": 10763
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6561132669448853,
      "learning_rate": 1.4494679546507871e-05,
      "loss": 0.5527,
      "step": 10764
    },
    {
      "epoch": 1.11,
      "grad_norm": 2.0019876956939697,
      "learning_rate": 1.449368208687982e-05,
      "loss": 0.5863,
      "step": 10765
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6462143659591675,
      "learning_rate": 1.4492684571227346e-05,
      "loss": 0.552,
      "step": 10766
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3826684951782227,
      "learning_rate": 1.4491686999562892e-05,
      "loss": 0.5972,
      "step": 10767
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6801782846450806,
      "learning_rate": 1.4490689371898887e-05,
      "loss": 0.6355,
      "step": 10768
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.433409571647644,
      "learning_rate": 1.4489691688247775e-05,
      "loss": 0.4785,
      "step": 10769
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.560815691947937,
      "learning_rate": 1.4488693948621988e-05,
      "loss": 0.613,
      "step": 10770
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.6933077573776245,
      "learning_rate": 1.4487696153033972e-05,
      "loss": 0.7227,
      "step": 10771
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3745038509368896,
      "learning_rate": 1.4486698301496162e-05,
      "loss": 0.6037,
      "step": 10772
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3081119060516357,
      "learning_rate": 1.4485700394021e-05,
      "loss": 0.6061,
      "step": 10773
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.5835283994674683,
      "learning_rate": 1.448470243062093e-05,
      "loss": 0.4234,
      "step": 10774
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3132774829864502,
      "learning_rate": 1.4483704411308389e-05,
      "loss": 0.5413,
      "step": 10775
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.542412519454956,
      "learning_rate": 1.4482706336095822e-05,
      "loss": 0.706,
      "step": 10776
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4617727994918823,
      "learning_rate": 1.4481708204995674e-05,
      "loss": 0.5031,
      "step": 10777
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2884252071380615,
      "learning_rate": 1.4480710018020388e-05,
      "loss": 0.5331,
      "step": 10778
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.4949631690979004,
      "learning_rate": 1.4479711775182407e-05,
      "loss": 0.6796,
      "step": 10779
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.7978805303573608,
      "learning_rate": 1.447871347649418e-05,
      "loss": 0.5826,
      "step": 10780
    },
    {
      "epoch": 1.12,
      "grad_norm": 2.60860013961792,
      "learning_rate": 1.4477715121968148e-05,
      "loss": 0.5264,
      "step": 10781
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6428157091140747,
      "learning_rate": 1.4476716711616764e-05,
      "loss": 0.5855,
      "step": 10782
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.57090425491333,
      "learning_rate": 1.4475718245452467e-05,
      "loss": 0.6908,
      "step": 10783
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.511209487915039,
      "learning_rate": 1.4474719723487716e-05,
      "loss": 0.5728,
      "step": 10784
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.9438246488571167,
      "learning_rate": 1.4473721145734954e-05,
      "loss": 0.6204,
      "step": 10785
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.506134033203125,
      "learning_rate": 1.447272251220663e-05,
      "loss": 0.5711,
      "step": 10786
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5727392435073853,
      "learning_rate": 1.4471723822915197e-05,
      "loss": 0.3898,
      "step": 10787
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6330690383911133,
      "learning_rate": 1.4470725077873104e-05,
      "loss": 0.6252,
      "step": 10788
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6051371097564697,
      "learning_rate": 1.4469726277092804e-05,
      "loss": 0.6345,
      "step": 10789
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.518656611442566,
      "learning_rate": 1.4468727420586747e-05,
      "loss": 0.5433,
      "step": 10790
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1808507442474365,
      "learning_rate": 1.4467728508367391e-05,
      "loss": 0.4067,
      "step": 10791
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.375430703163147,
      "learning_rate": 1.4466729540447184e-05,
      "loss": 0.6088,
      "step": 10792
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.2239642143249512,
      "learning_rate": 1.4465730516838586e-05,
      "loss": 0.5607,
      "step": 10793
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6478391885757446,
      "learning_rate": 1.4464731437554048e-05,
      "loss": 0.6002,
      "step": 10794
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6022354364395142,
      "learning_rate": 1.4463732302606026e-05,
      "loss": 0.5085,
      "step": 10795
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6817314624786377,
      "learning_rate": 1.446273311200698e-05,
      "loss": 0.6217,
      "step": 10796
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5766687393188477,
      "learning_rate": 1.4461733865769363e-05,
      "loss": 0.4542,
      "step": 10797
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5461245775222778,
      "learning_rate": 1.4460734563905639e-05,
      "loss": 0.697,
      "step": 10798
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.515934944152832,
      "learning_rate": 1.4459735206428259e-05,
      "loss": 0.5154,
      "step": 10799
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4442005157470703,
      "learning_rate": 1.4458735793349688e-05,
      "loss": 0.4444,
      "step": 10800
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3018618822097778,
      "learning_rate": 1.4457736324682383e-05,
      "loss": 0.6335,
      "step": 10801
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6262460947036743,
      "learning_rate": 1.4456736800438808e-05,
      "loss": 0.7364,
      "step": 10802
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4433436393737793,
      "learning_rate": 1.4455737220631423e-05,
      "loss": 0.5912,
      "step": 10803
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6437853574752808,
      "learning_rate": 1.4454737585272689e-05,
      "loss": 0.6055,
      "step": 10804
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5501232147216797,
      "learning_rate": 1.4453737894375067e-05,
      "loss": 0.6026,
      "step": 10805
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.646421194076538,
      "learning_rate": 1.4452738147951027e-05,
      "loss": 0.5846,
      "step": 10806
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5148749351501465,
      "learning_rate": 1.4451738346013028e-05,
      "loss": 0.6109,
      "step": 10807
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5771100521087646,
      "learning_rate": 1.4450738488573535e-05,
      "loss": 0.4719,
      "step": 10808
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5951449871063232,
      "learning_rate": 1.4449738575645016e-05,
      "loss": 0.6577,
      "step": 10809
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3309803009033203,
      "learning_rate": 1.4448738607239934e-05,
      "loss": 0.5925,
      "step": 10810
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.8159387111663818,
      "learning_rate": 1.444773858337076e-05,
      "loss": 0.6217,
      "step": 10811
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4530344009399414,
      "learning_rate": 1.444673850404996e-05,
      "loss": 0.5398,
      "step": 10812
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.252413034439087,
      "learning_rate": 1.4445738369289999e-05,
      "loss": 0.4768,
      "step": 10813
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5259901285171509,
      "learning_rate": 1.4444738179103353e-05,
      "loss": 0.5398,
      "step": 10814
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.8754140138626099,
      "learning_rate": 1.4443737933502483e-05,
      "loss": 0.7372,
      "step": 10815
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5758005380630493,
      "learning_rate": 1.4442737632499869e-05,
      "loss": 0.4813,
      "step": 10816
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.610975980758667,
      "learning_rate": 1.4441737276107976e-05,
      "loss": 0.6019,
      "step": 10817
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5179933309555054,
      "learning_rate": 1.4440736864339276e-05,
      "loss": 0.6571,
      "step": 10818
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.361497402191162,
      "learning_rate": 1.4439736397206245e-05,
      "loss": 0.5093,
      "step": 10819
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3694242238998413,
      "learning_rate": 1.4438735874721352e-05,
      "loss": 0.6533,
      "step": 10820
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5767183303833008,
      "learning_rate": 1.4437735296897072e-05,
      "loss": 0.6071,
      "step": 10821
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7813066244125366,
      "learning_rate": 1.4436734663745883e-05,
      "loss": 0.7166,
      "step": 10822
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4064877033233643,
      "learning_rate": 1.4435733975280258e-05,
      "loss": 0.5352,
      "step": 10823
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.48104727268219,
      "learning_rate": 1.4434733231512673e-05,
      "loss": 0.521,
      "step": 10824
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.576130747795105,
      "learning_rate": 1.44337324324556e-05,
      "loss": 0.4687,
      "step": 10825
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7663171291351318,
      "learning_rate": 1.4432731578121521e-05,
      "loss": 0.6543,
      "step": 10826
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6084145307540894,
      "learning_rate": 1.4431730668522917e-05,
      "loss": 0.6521,
      "step": 10827
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3862874507904053,
      "learning_rate": 1.4430729703672262e-05,
      "loss": 0.6178,
      "step": 10828
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7978463172912598,
      "learning_rate": 1.4429728683582036e-05,
      "loss": 0.5041,
      "step": 10829
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1721516847610474,
      "learning_rate": 1.442872760826472e-05,
      "loss": 0.482,
      "step": 10830
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3825398683547974,
      "learning_rate": 1.4427726477732798e-05,
      "loss": 0.678,
      "step": 10831
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6936571598052979,
      "learning_rate": 1.4426725291998744e-05,
      "loss": 0.5673,
      "step": 10832
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1597923040390015,
      "learning_rate": 1.4425724051075045e-05,
      "loss": 0.5269,
      "step": 10833
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5529720783233643,
      "learning_rate": 1.4424722754974182e-05,
      "loss": 0.4535,
      "step": 10834
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3314440250396729,
      "learning_rate": 1.4423721403708647e-05,
      "loss": 0.5747,
      "step": 10835
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.728963017463684,
      "learning_rate": 1.4422719997290909e-05,
      "loss": 0.4958,
      "step": 10836
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.779690146446228,
      "learning_rate": 1.4421718535733464e-05,
      "loss": 0.7097,
      "step": 10837
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.547210693359375,
      "learning_rate": 1.442071701904879e-05,
      "loss": 0.5917,
      "step": 10838
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.2124123573303223,
      "learning_rate": 1.4419715447249383e-05,
      "loss": 0.5624,
      "step": 10839
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4308463335037231,
      "learning_rate": 1.4418713820347723e-05,
      "loss": 0.6531,
      "step": 10840
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.725348711013794,
      "learning_rate": 1.4417712138356298e-05,
      "loss": 0.6035,
      "step": 10841
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5165715217590332,
      "learning_rate": 1.44167104012876e-05,
      "loss": 0.5296,
      "step": 10842
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6587977409362793,
      "learning_rate": 1.4415708609154115e-05,
      "loss": 0.4456,
      "step": 10843
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5681772232055664,
      "learning_rate": 1.4414706761968331e-05,
      "loss": 0.594,
      "step": 10844
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.422070026397705,
      "learning_rate": 1.4413704859742746e-05,
      "loss": 0.5158,
      "step": 10845
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3381338119506836,
      "learning_rate": 1.4412702902489842e-05,
      "loss": 0.5963,
      "step": 10846
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.491546630859375,
      "learning_rate": 1.4411700890222113e-05,
      "loss": 0.4378,
      "step": 10847
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5909087657928467,
      "learning_rate": 1.4410698822952057e-05,
      "loss": 0.513,
      "step": 10848
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4391379356384277,
      "learning_rate": 1.4409696700692162e-05,
      "loss": 0.5468,
      "step": 10849
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4239308834075928,
      "learning_rate": 1.4408694523454923e-05,
      "loss": 0.5272,
      "step": 10850
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.471731424331665,
      "learning_rate": 1.4407692291252832e-05,
      "loss": 0.547,
      "step": 10851
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4489085674285889,
      "learning_rate": 1.4406690004098389e-05,
      "loss": 0.5359,
      "step": 10852
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3363603353500366,
      "learning_rate": 1.4405687662004088e-05,
      "loss": 0.5386,
      "step": 10853
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.8741477727890015,
      "learning_rate": 1.4404685264982425e-05,
      "loss": 0.7312,
      "step": 10854
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6383202075958252,
      "learning_rate": 1.4403682813045899e-05,
      "loss": 0.4765,
      "step": 10855
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7944551706314087,
      "learning_rate": 1.4402680306207004e-05,
      "loss": 0.5606,
      "step": 10856
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3471676111221313,
      "learning_rate": 1.4401677744478244e-05,
      "loss": 0.4698,
      "step": 10857
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.2794926166534424,
      "learning_rate": 1.4400675127872115e-05,
      "loss": 0.5177,
      "step": 10858
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.8980000019073486,
      "learning_rate": 1.4399672456401115e-05,
      "loss": 0.5982,
      "step": 10859
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5012307167053223,
      "learning_rate": 1.4398669730077748e-05,
      "loss": 0.5411,
      "step": 10860
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.6805049180984497,
      "learning_rate": 1.4397666948914517e-05,
      "loss": 0.5856,
      "step": 10861
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5950076580047607,
      "learning_rate": 1.4396664112923917e-05,
      "loss": 0.445,
      "step": 10862
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.652674913406372,
      "learning_rate": 1.439566122211846e-05,
      "loss": 0.4271,
      "step": 10863
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5006715059280396,
      "learning_rate": 1.4394658276510641e-05,
      "loss": 0.6066,
      "step": 10864
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5089683532714844,
      "learning_rate": 1.439365527611297e-05,
      "loss": 0.5375,
      "step": 10865
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3735793828964233,
      "learning_rate": 1.4392652220937949e-05,
      "loss": 0.5131,
      "step": 10866
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5504510402679443,
      "learning_rate": 1.4391649110998083e-05,
      "loss": 0.5403,
      "step": 10867
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.283484935760498,
      "learning_rate": 1.4390645946305878e-05,
      "loss": 0.5146,
      "step": 10868
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.8743058443069458,
      "learning_rate": 1.4389642726873846e-05,
      "loss": 0.654,
      "step": 10869
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.9945869445800781,
      "learning_rate": 1.4388639452714489e-05,
      "loss": 0.7478,
      "step": 10870
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5746434926986694,
      "learning_rate": 1.4387636123840316e-05,
      "loss": 0.584,
      "step": 10871
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4337408542633057,
      "learning_rate": 1.4386632740263837e-05,
      "loss": 0.584,
      "step": 10872
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.320507526397705,
      "learning_rate": 1.4385629301997562e-05,
      "loss": 0.542,
      "step": 10873
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7520289421081543,
      "learning_rate": 1.4384625809053999e-05,
      "loss": 0.653,
      "step": 10874
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.952513337135315,
      "learning_rate": 1.438362226144566e-05,
      "loss": 0.5631,
      "step": 10875
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.5231566429138184,
      "learning_rate": 1.4382618659185059e-05,
      "loss": 0.5067,
      "step": 10876
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.7164196968078613,
      "learning_rate": 1.4381615002284703e-05,
      "loss": 0.4868,
      "step": 10877
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4004789590835571,
      "learning_rate": 1.438061129075711e-05,
      "loss": 0.5489,
      "step": 10878
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.747488021850586,
      "learning_rate": 1.4379607524614794e-05,
      "loss": 0.5284,
      "step": 10879
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5026341676712036,
      "learning_rate": 1.4378603703870264e-05,
      "loss": 0.5408,
      "step": 10880
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4641331434249878,
      "learning_rate": 1.437759982853604e-05,
      "loss": 0.459,
      "step": 10881
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5609703063964844,
      "learning_rate": 1.4376595898624635e-05,
      "loss": 0.6273,
      "step": 10882
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.427635908126831,
      "learning_rate": 1.4375591914148566e-05,
      "loss": 0.5612,
      "step": 10883
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.483397364616394,
      "learning_rate": 1.4374587875120352e-05,
      "loss": 0.5658,
      "step": 10884
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4781907796859741,
      "learning_rate": 1.4373583781552507e-05,
      "loss": 0.5805,
      "step": 10885
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6496562957763672,
      "learning_rate": 1.4372579633457551e-05,
      "loss": 0.5785,
      "step": 10886
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5633008480072021,
      "learning_rate": 1.4371575430848006e-05,
      "loss": 0.569,
      "step": 10887
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.217275857925415,
      "learning_rate": 1.4370571173736388e-05,
      "loss": 0.538,
      "step": 10888
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6107800006866455,
      "learning_rate": 1.436956686213522e-05,
      "loss": 0.6051,
      "step": 10889
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6725587844848633,
      "learning_rate": 1.4368562496057018e-05,
      "loss": 0.5635,
      "step": 10890
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4366093873977661,
      "learning_rate": 1.436755807551431e-05,
      "loss": 0.5838,
      "step": 10891
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3751071691513062,
      "learning_rate": 1.436655360051962e-05,
      "loss": 0.4815,
      "step": 10892
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4433062076568604,
      "learning_rate": 1.4365549071085462e-05,
      "loss": 0.5245,
      "step": 10893
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2253443002700806,
      "learning_rate": 1.4364544487224366e-05,
      "loss": 0.3698,
      "step": 10894
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7576041221618652,
      "learning_rate": 1.4363539848948857e-05,
      "loss": 0.5878,
      "step": 10895
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6346303224563599,
      "learning_rate": 1.4362535156271457e-05,
      "loss": 0.553,
      "step": 10896
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4878164529800415,
      "learning_rate": 1.4361530409204698e-05,
      "loss": 0.5525,
      "step": 10897
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5627808570861816,
      "learning_rate": 1.4360525607761099e-05,
      "loss": 0.5108,
      "step": 10898
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2621971368789673,
      "learning_rate": 1.435952075195319e-05,
      "loss": 0.4297,
      "step": 10899
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4953182935714722,
      "learning_rate": 1.4358515841793504e-05,
      "loss": 0.6412,
      "step": 10900
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.7216910123825073,
      "learning_rate": 1.435751087729456e-05,
      "loss": 0.6052,
      "step": 10901
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.8098254203796387,
      "learning_rate": 1.4356505858468896e-05,
      "loss": 0.5964,
      "step": 10902
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4570684432983398,
      "learning_rate": 1.4355500785329035e-05,
      "loss": 0.5549,
      "step": 10903
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2254791259765625,
      "learning_rate": 1.4354495657887513e-05,
      "loss": 0.634,
      "step": 10904
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6353530883789062,
      "learning_rate": 1.4353490476156859e-05,
      "loss": 0.7055,
      "step": 10905
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5380330085754395,
      "learning_rate": 1.4352485240149605e-05,
      "loss": 0.5668,
      "step": 10906
    },
    {
      "epoch": 1.13,
      "grad_norm": 18.661828994750977,
      "learning_rate": 1.4351479949878282e-05,
      "loss": 1.2819,
      "step": 10907
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4445637464523315,
      "learning_rate": 1.435047460535543e-05,
      "loss": 0.5956,
      "step": 10908
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4773799180984497,
      "learning_rate": 1.4349469206593574e-05,
      "loss": 0.615,
      "step": 10909
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6345986127853394,
      "learning_rate": 1.4348463753605256e-05,
      "loss": 0.7197,
      "step": 10910
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6751575469970703,
      "learning_rate": 1.4347458246403004e-05,
      "loss": 0.6081,
      "step": 10911
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5263522863388062,
      "learning_rate": 1.4346452684999359e-05,
      "loss": 0.5264,
      "step": 10912
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.8459924459457397,
      "learning_rate": 1.4345447069406859e-05,
      "loss": 0.6813,
      "step": 10913
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6233943700790405,
      "learning_rate": 1.434444139963804e-05,
      "loss": 0.4952,
      "step": 10914
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6046172380447388,
      "learning_rate": 1.434343567570544e-05,
      "loss": 0.6344,
      "step": 10915
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6296225786209106,
      "learning_rate": 1.4342429897621595e-05,
      "loss": 0.721,
      "step": 10916
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.456445574760437,
      "learning_rate": 1.4341424065399045e-05,
      "loss": 0.6462,
      "step": 10917
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3281499147415161,
      "learning_rate": 1.4340418179050337e-05,
      "loss": 0.5771,
      "step": 10918
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.484933614730835,
      "learning_rate": 1.4339412238588002e-05,
      "loss": 0.5254,
      "step": 10919
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6593005657196045,
      "learning_rate": 1.4338406244024588e-05,
      "loss": 0.649,
      "step": 10920
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2806928157806396,
      "learning_rate": 1.4337400195372633e-05,
      "loss": 0.5017,
      "step": 10921
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3461558818817139,
      "learning_rate": 1.4336394092644683e-05,
      "loss": 0.5089,
      "step": 10922
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6709516048431396,
      "learning_rate": 1.4335387935853282e-05,
      "loss": 0.4824,
      "step": 10923
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5658031702041626,
      "learning_rate": 1.4334381725010969e-05,
      "loss": 0.5892,
      "step": 10924
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4292255640029907,
      "learning_rate": 1.4333375460130295e-05,
      "loss": 0.4449,
      "step": 10925
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5221298933029175,
      "learning_rate": 1.4332369141223801e-05,
      "loss": 0.6173,
      "step": 10926
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.388234257698059,
      "learning_rate": 1.4331362768304034e-05,
      "loss": 0.611,
      "step": 10927
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5172395706176758,
      "learning_rate": 1.4330356341383545e-05,
      "loss": 0.517,
      "step": 10928
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.56557035446167,
      "learning_rate": 1.4329349860474875e-05,
      "loss": 0.5107,
      "step": 10929
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5175460577011108,
      "learning_rate": 1.4328343325590575e-05,
      "loss": 0.5334,
      "step": 10930
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6924221515655518,
      "learning_rate": 1.4327336736743199e-05,
      "loss": 0.5787,
      "step": 10931
    },
    {
      "epoch": 1.13,
      "grad_norm": 2.056736469268799,
      "learning_rate": 1.4326330093945287e-05,
      "loss": 0.6048,
      "step": 10932
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3723690509796143,
      "learning_rate": 1.4325323397209395e-05,
      "loss": 0.5267,
      "step": 10933
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5899208784103394,
      "learning_rate": 1.4324316646548073e-05,
      "loss": 0.5778,
      "step": 10934
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2361164093017578,
      "learning_rate": 1.4323309841973872e-05,
      "loss": 0.597,
      "step": 10935
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5422495603561401,
      "learning_rate": 1.4322302983499346e-05,
      "loss": 0.5012,
      "step": 10936
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6432580947875977,
      "learning_rate": 1.4321296071137043e-05,
      "loss": 0.5397,
      "step": 10937
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3842352628707886,
      "learning_rate": 1.4320289104899524e-05,
      "loss": 0.5164,
      "step": 10938
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4894829988479614,
      "learning_rate": 1.4319282084799336e-05,
      "loss": 0.6078,
      "step": 10939
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6274232864379883,
      "learning_rate": 1.4318275010849037e-05,
      "loss": 0.582,
      "step": 10940
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3412946462631226,
      "learning_rate": 1.4317267883061186e-05,
      "loss": 0.5515,
      "step": 10941
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4073377847671509,
      "learning_rate": 1.4316260701448332e-05,
      "loss": 0.561,
      "step": 10942
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.8783658742904663,
      "learning_rate": 1.4315253466023038e-05,
      "loss": 0.5649,
      "step": 10943
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.8051738739013672,
      "learning_rate": 1.4314246176797858e-05,
      "loss": 0.5963,
      "step": 10944
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3591066598892212,
      "learning_rate": 1.4313238833785355e-05,
      "loss": 0.4624,
      "step": 10945
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4280251264572144,
      "learning_rate": 1.431223143699808e-05,
      "loss": 0.52,
      "step": 10946
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.400437831878662,
      "learning_rate": 1.4311223986448601e-05,
      "loss": 0.5018,
      "step": 10947
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3786782026290894,
      "learning_rate": 1.4310216482149471e-05,
      "loss": 0.5429,
      "step": 10948
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4871536493301392,
      "learning_rate": 1.4309208924113258e-05,
      "loss": 0.6864,
      "step": 10949
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3892223834991455,
      "learning_rate": 1.4308201312352516e-05,
      "loss": 0.4901,
      "step": 10950
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2504677772521973,
      "learning_rate": 1.4307193646879814e-05,
      "loss": 0.4247,
      "step": 10951
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.244275689125061,
      "learning_rate": 1.4306185927707712e-05,
      "loss": 0.5564,
      "step": 10952
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.9368265867233276,
      "learning_rate": 1.4305178154848771e-05,
      "loss": 0.5464,
      "step": 10953
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3567129373550415,
      "learning_rate": 1.4304170328315563e-05,
      "loss": 0.6179,
      "step": 10954
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2733649015426636,
      "learning_rate": 1.4303162448120644e-05,
      "loss": 0.5032,
      "step": 10955
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.525853157043457,
      "learning_rate": 1.4302154514276585e-05,
      "loss": 0.5104,
      "step": 10956
    },
    {
      "epoch": 1.13,
      "grad_norm": 2.1106362342834473,
      "learning_rate": 1.430114652679595e-05,
      "loss": 0.7006,
      "step": 10957
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4903490543365479,
      "learning_rate": 1.4300138485691309e-05,
      "loss": 0.5542,
      "step": 10958
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4901916980743408,
      "learning_rate": 1.4299130390975225e-05,
      "loss": 0.4837,
      "step": 10959
    },
    {
      "epoch": 1.13,
      "grad_norm": 2.0019545555114746,
      "learning_rate": 1.429812224266027e-05,
      "loss": 0.5938,
      "step": 10960
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2728605270385742,
      "learning_rate": 1.4297114040759011e-05,
      "loss": 0.5992,
      "step": 10961
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6551276445388794,
      "learning_rate": 1.429610578528402e-05,
      "loss": 0.595,
      "step": 10962
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.674202799797058,
      "learning_rate": 1.4295097476247864e-05,
      "loss": 0.6692,
      "step": 10963
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3825687170028687,
      "learning_rate": 1.4294089113663117e-05,
      "loss": 0.5194,
      "step": 10964
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4866490364074707,
      "learning_rate": 1.4293080697542348e-05,
      "loss": 0.4468,
      "step": 10965
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4997737407684326,
      "learning_rate": 1.4292072227898131e-05,
      "loss": 0.5809,
      "step": 10966
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.4003413915634155,
      "learning_rate": 1.429106370474304e-05,
      "loss": 0.5512,
      "step": 10967
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5861763954162598,
      "learning_rate": 1.4290055128089646e-05,
      "loss": 0.5635,
      "step": 10968
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.240224003791809,
      "learning_rate": 1.4289046497950523e-05,
      "loss": 0.4723,
      "step": 10969
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.601349115371704,
      "learning_rate": 1.4288037814338252e-05,
      "loss": 0.6585,
      "step": 10970
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5488375425338745,
      "learning_rate": 1.4287029077265397e-05,
      "loss": 0.6176,
      "step": 10971
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.298174500465393,
      "learning_rate": 1.4286020286744546e-05,
      "loss": 0.5343,
      "step": 10972
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2024977207183838,
      "learning_rate": 1.4285011442788272e-05,
      "loss": 0.586,
      "step": 10973
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.5047954320907593,
      "learning_rate": 1.4284002545409151e-05,
      "loss": 0.4921,
      "step": 10974
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.398368239402771,
      "learning_rate": 1.4282993594619763e-05,
      "loss": 0.5724,
      "step": 10975
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6025217771530151,
      "learning_rate": 1.4281984590432686e-05,
      "loss": 0.6069,
      "step": 10976
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4336025714874268,
      "learning_rate": 1.4280975532860496e-05,
      "loss": 0.4235,
      "step": 10977
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.2884628772735596,
      "learning_rate": 1.4279966421915784e-05,
      "loss": 0.6339,
      "step": 10978
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.374245047569275,
      "learning_rate": 1.427895725761112e-05,
      "loss": 0.6329,
      "step": 10979
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.487886905670166,
      "learning_rate": 1.4277948039959092e-05,
      "loss": 0.6427,
      "step": 10980
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5990937948226929,
      "learning_rate": 1.4276938768972277e-05,
      "loss": 0.5276,
      "step": 10981
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.2882007360458374,
      "learning_rate": 1.4275929444663264e-05,
      "loss": 0.6001,
      "step": 10982
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.576629638671875,
      "learning_rate": 1.4274920067044636e-05,
      "loss": 0.4996,
      "step": 10983
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3384932279586792,
      "learning_rate": 1.427391063612897e-05,
      "loss": 0.4836,
      "step": 10984
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6250287294387817,
      "learning_rate": 1.4272901151928857e-05,
      "loss": 0.6018,
      "step": 10985
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5775285959243774,
      "learning_rate": 1.4271891614456884e-05,
      "loss": 0.6545,
      "step": 10986
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7016587257385254,
      "learning_rate": 1.4270882023725632e-05,
      "loss": 0.5621,
      "step": 10987
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7604864835739136,
      "learning_rate": 1.426987237974769e-05,
      "loss": 0.6025,
      "step": 10988
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3983395099639893,
      "learning_rate": 1.426886268253565e-05,
      "loss": 0.6026,
      "step": 10989
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6691755056381226,
      "learning_rate": 1.4267852932102094e-05,
      "loss": 0.5027,
      "step": 10990
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4480518102645874,
      "learning_rate": 1.4266843128459615e-05,
      "loss": 0.5505,
      "step": 10991
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5652304887771606,
      "learning_rate": 1.42658332716208e-05,
      "loss": 0.6136,
      "step": 10992
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7176992893218994,
      "learning_rate": 1.426482336159824e-05,
      "loss": 0.615,
      "step": 10993
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3042356967926025,
      "learning_rate": 1.426381339840453e-05,
      "loss": 0.5304,
      "step": 10994
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.49728524684906,
      "learning_rate": 1.4262803382052254e-05,
      "loss": 0.5604,
      "step": 10995
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6284630298614502,
      "learning_rate": 1.4261793312554011e-05,
      "loss": 0.5432,
      "step": 10996
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4734227657318115,
      "learning_rate": 1.4260783189922389e-05,
      "loss": 0.5268,
      "step": 10997
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.643864631652832,
      "learning_rate": 1.4259773014169983e-05,
      "loss": 0.6088,
      "step": 10998
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6544667482376099,
      "learning_rate": 1.4258762785309393e-05,
      "loss": 0.4788,
      "step": 10999
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4948437213897705,
      "learning_rate": 1.4257752503353205e-05,
      "loss": 0.6013,
      "step": 11000
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5145522356033325,
      "learning_rate": 1.4256742168314017e-05,
      "loss": 0.5582,
      "step": 11001
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4762723445892334,
      "learning_rate": 1.4255731780204429e-05,
      "loss": 0.4594,
      "step": 11002
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.8770469427108765,
      "learning_rate": 1.4254721339037034e-05,
      "loss": 0.6629,
      "step": 11003
    },
    {
      "epoch": 1.14,
      "grad_norm": 2.3515477180480957,
      "learning_rate": 1.4253710844824435e-05,
      "loss": 0.5572,
      "step": 11004
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5978789329528809,
      "learning_rate": 1.4252700297579224e-05,
      "loss": 0.6593,
      "step": 11005
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3503414392471313,
      "learning_rate": 1.4251689697314e-05,
      "loss": 0.6279,
      "step": 11006
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6297879219055176,
      "learning_rate": 1.425067904404137e-05,
      "loss": 0.5767,
      "step": 11007
    },
    {
      "epoch": 1.14,
      "grad_norm": 2.007246494293213,
      "learning_rate": 1.4249668337773923e-05,
      "loss": 0.7138,
      "step": 11008
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4827169179916382,
      "learning_rate": 1.4248657578524269e-05,
      "loss": 0.5618,
      "step": 11009
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4767340421676636,
      "learning_rate": 1.4247646766305005e-05,
      "loss": 0.5802,
      "step": 11010
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7197259664535522,
      "learning_rate": 1.4246635901128736e-05,
      "loss": 0.6101,
      "step": 11011
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.506765365600586,
      "learning_rate": 1.4245624983008061e-05,
      "loss": 0.5415,
      "step": 11012
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.903507113456726,
      "learning_rate": 1.4244614011955586e-05,
      "loss": 0.646,
      "step": 11013
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4212777614593506,
      "learning_rate": 1.4243602987983918e-05,
      "loss": 0.4305,
      "step": 11014
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.470687985420227,
      "learning_rate": 1.4242591911105657e-05,
      "loss": 0.6284,
      "step": 11015
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.418397068977356,
      "learning_rate": 1.424158078133341e-05,
      "loss": 0.5724,
      "step": 11016
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4964858293533325,
      "learning_rate": 1.4240569598679783e-05,
      "loss": 0.5367,
      "step": 11017
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3602666854858398,
      "learning_rate": 1.4239558363157384e-05,
      "loss": 0.5968,
      "step": 11018
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5959179401397705,
      "learning_rate": 1.423854707477882e-05,
      "loss": 0.632,
      "step": 11019
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.1495599746704102,
      "learning_rate": 1.4237535733556698e-05,
      "loss": 0.4462,
      "step": 11020
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5391249656677246,
      "learning_rate": 1.4236524339503625e-05,
      "loss": 0.515,
      "step": 11021
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3723350763320923,
      "learning_rate": 1.423551289263222e-05,
      "loss": 0.5717,
      "step": 11022
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3499785661697388,
      "learning_rate": 1.423450139295508e-05,
      "loss": 0.6052,
      "step": 11023
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5477176904678345,
      "learning_rate": 1.4233489840484822e-05,
      "loss": 0.6424,
      "step": 11024
    },
    {
      "epoch": 1.14,
      "grad_norm": 2.0921506881713867,
      "learning_rate": 1.423247823523406e-05,
      "loss": 0.4942,
      "step": 11025
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6159993410110474,
      "learning_rate": 1.42314665772154e-05,
      "loss": 0.69,
      "step": 11026
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.579991340637207,
      "learning_rate": 1.4230454866441462e-05,
      "loss": 0.5639,
      "step": 11027
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3890597820281982,
      "learning_rate": 1.4229443102924851e-05,
      "loss": 0.561,
      "step": 11028
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.72062087059021,
      "learning_rate": 1.4228431286678188e-05,
      "loss": 0.6313,
      "step": 11029
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3985576629638672,
      "learning_rate": 1.4227419417714085e-05,
      "loss": 0.5641,
      "step": 11030
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.750024437904358,
      "learning_rate": 1.4226407496045156e-05,
      "loss": 0.6975,
      "step": 11031
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5010472536087036,
      "learning_rate": 1.4225395521684018e-05,
      "loss": 0.5193,
      "step": 11032
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.337049126625061,
      "learning_rate": 1.4224383494643293e-05,
      "loss": 0.6062,
      "step": 11033
    },
    {
      "epoch": 1.14,
      "grad_norm": 2.175872802734375,
      "learning_rate": 1.4223371414935587e-05,
      "loss": 0.6901,
      "step": 11034
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6109002828598022,
      "learning_rate": 1.4222359282573531e-05,
      "loss": 0.6714,
      "step": 11035
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5169546604156494,
      "learning_rate": 1.4221347097569731e-05,
      "loss": 0.5565,
      "step": 11036
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7277717590332031,
      "learning_rate": 1.4220334859936813e-05,
      "loss": 0.569,
      "step": 11037
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5611475706100464,
      "learning_rate": 1.4219322569687399e-05,
      "loss": 0.531,
      "step": 11038
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.494582176208496,
      "learning_rate": 1.4218310226834105e-05,
      "loss": 0.5447,
      "step": 11039
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.2152223587036133,
      "learning_rate": 1.4217297831389554e-05,
      "loss": 0.4358,
      "step": 11040
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.2240787744522095,
      "learning_rate": 1.421628538336637e-05,
      "loss": 0.5687,
      "step": 11041
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4061394929885864,
      "learning_rate": 1.4215272882777171e-05,
      "loss": 0.5723,
      "step": 11042
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4681692123413086,
      "learning_rate": 1.4214260329634585e-05,
      "loss": 0.6342,
      "step": 11043
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3614908456802368,
      "learning_rate": 1.4213247723951233e-05,
      "loss": 0.5521,
      "step": 11044
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6052416563034058,
      "learning_rate": 1.421223506573974e-05,
      "loss": 0.5592,
      "step": 11045
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7094712257385254,
      "learning_rate": 1.4211222355012732e-05,
      "loss": 0.5964,
      "step": 11046
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.621280550956726,
      "learning_rate": 1.4210209591782833e-05,
      "loss": 0.5754,
      "step": 11047
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.2605032920837402,
      "learning_rate": 1.4209196776062672e-05,
      "loss": 0.4909,
      "step": 11048
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4614641666412354,
      "learning_rate": 1.4208183907864874e-05,
      "loss": 0.6146,
      "step": 11049
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.34462571144104,
      "learning_rate": 1.420717098720207e-05,
      "loss": 0.525,
      "step": 11050
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7681660652160645,
      "learning_rate": 1.4206158014086886e-05,
      "loss": 0.4672,
      "step": 11051
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.684711217880249,
      "learning_rate": 1.4205144988531947e-05,
      "loss": 0.5532,
      "step": 11052
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.2692447900772095,
      "learning_rate": 1.420413191054989e-05,
      "loss": 0.5451,
      "step": 11053
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4114700555801392,
      "learning_rate": 1.4203118780153344e-05,
      "loss": 0.6164,
      "step": 11054
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6458370685577393,
      "learning_rate": 1.4202105597354935e-05,
      "loss": 0.6379,
      "step": 11055
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5330811738967896,
      "learning_rate": 1.4201092362167304e-05,
      "loss": 0.5843,
      "step": 11056
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.508516788482666,
      "learning_rate": 1.420007907460307e-05,
      "loss": 0.5708,
      "step": 11057
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3362675905227661,
      "learning_rate": 1.4199065734674879e-05,
      "loss": 0.5385,
      "step": 11058
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.2364907264709473,
      "learning_rate": 1.4198052342395358e-05,
      "loss": 0.5146,
      "step": 11059
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4911787509918213,
      "learning_rate": 1.4197038897777145e-05,
      "loss": 0.5809,
      "step": 11060
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.8434803485870361,
      "learning_rate": 1.419602540083287e-05,
      "loss": 0.5175,
      "step": 11061
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.395476222038269,
      "learning_rate": 1.419501185157517e-05,
      "loss": 0.4579,
      "step": 11062
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7594900131225586,
      "learning_rate": 1.4193998250016685e-05,
      "loss": 0.6365,
      "step": 11063
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.5592120885849,
      "learning_rate": 1.4192984596170051e-05,
      "loss": 0.5015,
      "step": 11064
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6295521259307861,
      "learning_rate": 1.4191970890047901e-05,
      "loss": 0.49,
      "step": 11065
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.422650694847107,
      "learning_rate": 1.4190957131662878e-05,
      "loss": 0.4909,
      "step": 11066
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.7292680740356445,
      "learning_rate": 1.4189943321027622e-05,
      "loss": 0.5259,
      "step": 11067
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.6118496656417847,
      "learning_rate": 1.4188929458154768e-05,
      "loss": 0.4461,
      "step": 11068
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.3796677589416504,
      "learning_rate": 1.4187915543056957e-05,
      "loss": 0.5697,
      "step": 11069
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.4393751621246338,
      "learning_rate": 1.4186901575746831e-05,
      "loss": 0.5395,
      "step": 11070
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.600975513458252,
      "learning_rate": 1.4185887556237033e-05,
      "loss": 0.5044,
      "step": 11071
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7523363828659058,
      "learning_rate": 1.4184873484540203e-05,
      "loss": 0.6591,
      "step": 11072
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5575147867202759,
      "learning_rate": 1.4183859360668986e-05,
      "loss": 0.5756,
      "step": 11073
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.482579231262207,
      "learning_rate": 1.4182845184636022e-05,
      "loss": 0.5378,
      "step": 11074
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6199719905853271,
      "learning_rate": 1.4181830956453957e-05,
      "loss": 0.6045,
      "step": 11075
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.9591013193130493,
      "learning_rate": 1.4180816676135436e-05,
      "loss": 0.5961,
      "step": 11076
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5573859214782715,
      "learning_rate": 1.4179802343693108e-05,
      "loss": 0.6376,
      "step": 11077
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.604068398475647,
      "learning_rate": 1.4178787959139611e-05,
      "loss": 0.6047,
      "step": 11078
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5920075178146362,
      "learning_rate": 1.4177773522487597e-05,
      "loss": 0.5686,
      "step": 11079
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4722753763198853,
      "learning_rate": 1.4176759033749712e-05,
      "loss": 0.5678,
      "step": 11080
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5608400106430054,
      "learning_rate": 1.4175744492938605e-05,
      "loss": 0.5958,
      "step": 11081
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.3743164539337158,
      "learning_rate": 1.4174729900066926e-05,
      "loss": 0.5879,
      "step": 11082
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.875952124595642,
      "learning_rate": 1.417371525514732e-05,
      "loss": 0.5939,
      "step": 11083
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5929852724075317,
      "learning_rate": 1.4172700558192438e-05,
      "loss": 0.6169,
      "step": 11084
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.3213213682174683,
      "learning_rate": 1.4171685809214936e-05,
      "loss": 0.6494,
      "step": 11085
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.83059561252594,
      "learning_rate": 1.4170671008227458e-05,
      "loss": 0.6245,
      "step": 11086
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4024635553359985,
      "learning_rate": 1.4169656155242661e-05,
      "loss": 0.6166,
      "step": 11087
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5699249505996704,
      "learning_rate": 1.4168641250273197e-05,
      "loss": 0.4654,
      "step": 11088
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5460959672927856,
      "learning_rate": 1.4167626293331713e-05,
      "loss": 0.5966,
      "step": 11089
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.262102484703064,
      "learning_rate": 1.4166611284430873e-05,
      "loss": 0.4439,
      "step": 11090
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4354681968688965,
      "learning_rate": 1.4165596223583324e-05,
      "loss": 0.5006,
      "step": 11091
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.8450682163238525,
      "learning_rate": 1.4164581110801723e-05,
      "loss": 0.5388,
      "step": 11092
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5350393056869507,
      "learning_rate": 1.4163565946098728e-05,
      "loss": 0.6292,
      "step": 11093
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1909540891647339,
      "learning_rate": 1.416255072948699e-05,
      "loss": 0.5461,
      "step": 11094
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.8031938076019287,
      "learning_rate": 1.4161535460979173e-05,
      "loss": 0.5753,
      "step": 11095
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4216692447662354,
      "learning_rate": 1.416052014058793e-05,
      "loss": 0.5923,
      "step": 11096
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.597883939743042,
      "learning_rate": 1.415950476832592e-05,
      "loss": 0.5524,
      "step": 11097
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4204846620559692,
      "learning_rate": 1.4158489344205806e-05,
      "loss": 0.4207,
      "step": 11098
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.47664213180542,
      "learning_rate": 1.415747386824024e-05,
      "loss": 0.4348,
      "step": 11099
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.2769737243652344,
      "learning_rate": 1.4156458340441892e-05,
      "loss": 0.5855,
      "step": 11100
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.2994625568389893,
      "learning_rate": 1.4155442760823413e-05,
      "loss": 0.6033,
      "step": 11101
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5159038305282593,
      "learning_rate": 1.4154427129397473e-05,
      "loss": 0.5914,
      "step": 11102
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5977603197097778,
      "learning_rate": 1.4153411446176728e-05,
      "loss": 0.5395,
      "step": 11103
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5427296161651611,
      "learning_rate": 1.4152395711173845e-05,
      "loss": 0.5203,
      "step": 11104
    },
    {
      "epoch": 1.15,
      "grad_norm": 2.0200507640838623,
      "learning_rate": 1.4151379924401484e-05,
      "loss": 0.4839,
      "step": 11105
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4263854026794434,
      "learning_rate": 1.4150364085872314e-05,
      "loss": 0.5047,
      "step": 11106
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.677051305770874,
      "learning_rate": 1.4149348195598998e-05,
      "loss": 0.6065,
      "step": 11107
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.7243331670761108,
      "learning_rate": 1.41483322535942e-05,
      "loss": 0.6815,
      "step": 11108
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.2748488187789917,
      "learning_rate": 1.4147316259870585e-05,
      "loss": 0.5073,
      "step": 11109
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.290723204612732,
      "learning_rate": 1.414630021444082e-05,
      "loss": 0.5197,
      "step": 11110
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6137139797210693,
      "learning_rate": 1.414528411731758e-05,
      "loss": 0.5568,
      "step": 11111
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.252185344696045,
      "learning_rate": 1.414426796851352e-05,
      "loss": 0.5632,
      "step": 11112
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5388014316558838,
      "learning_rate": 1.4143251768041321e-05,
      "loss": 0.6865,
      "step": 11113
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5164086818695068,
      "learning_rate": 1.4142235515913645e-05,
      "loss": 0.551,
      "step": 11114
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.485050916671753,
      "learning_rate": 1.4141219212143162e-05,
      "loss": 0.4572,
      "step": 11115
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5974195003509521,
      "learning_rate": 1.4140202856742548e-05,
      "loss": 0.5599,
      "step": 11116
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.3240141868591309,
      "learning_rate": 1.413918644972447e-05,
      "loss": 0.5879,
      "step": 11117
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.8371098041534424,
      "learning_rate": 1.41381699911016e-05,
      "loss": 0.5463,
      "step": 11118
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4967710971832275,
      "learning_rate": 1.4137153480886614e-05,
      "loss": 0.5274,
      "step": 11119
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5058445930480957,
      "learning_rate": 1.413613691909218e-05,
      "loss": 0.5714,
      "step": 11120
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6498169898986816,
      "learning_rate": 1.4135120305730978e-05,
      "loss": 0.6851,
      "step": 11121
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.435258388519287,
      "learning_rate": 1.4134103640815675e-05,
      "loss": 0.6042,
      "step": 11122
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5643091201782227,
      "learning_rate": 1.4133086924358951e-05,
      "loss": 0.493,
      "step": 11123
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4094221591949463,
      "learning_rate": 1.4132070156373483e-05,
      "loss": 0.5002,
      "step": 11124
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4763751029968262,
      "learning_rate": 1.4131053336871944e-05,
      "loss": 0.5432,
      "step": 11125
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5419386625289917,
      "learning_rate": 1.4130036465867013e-05,
      "loss": 0.6116,
      "step": 11126
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5184022188186646,
      "learning_rate": 1.412901954337137e-05,
      "loss": 0.5795,
      "step": 11127
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.3669147491455078,
      "learning_rate": 1.4128002569397685e-05,
      "loss": 0.6426,
      "step": 11128
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.865020751953125,
      "learning_rate": 1.4126985543958648e-05,
      "loss": 0.6786,
      "step": 11129
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5340436697006226,
      "learning_rate": 1.4125968467066931e-05,
      "loss": 0.5376,
      "step": 11130
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4821606874465942,
      "learning_rate": 1.4124951338735213e-05,
      "loss": 0.5611,
      "step": 11131
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.742870569229126,
      "learning_rate": 1.4123934158976185e-05,
      "loss": 0.5128,
      "step": 11132
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4819328784942627,
      "learning_rate": 1.4122916927802516e-05,
      "loss": 0.473,
      "step": 11133
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.3625342845916748,
      "learning_rate": 1.41218996452269e-05,
      "loss": 0.501,
      "step": 11134
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.461530327796936,
      "learning_rate": 1.4120882311262007e-05,
      "loss": 0.6075,
      "step": 11135
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.534703254699707,
      "learning_rate": 1.4119864925920532e-05,
      "loss": 0.612,
      "step": 11136
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.9272644519805908,
      "learning_rate": 1.4118847489215156e-05,
      "loss": 0.6785,
      "step": 11137
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4883160591125488,
      "learning_rate": 1.4117830001158559e-05,
      "loss": 0.5715,
      "step": 11138
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.3639215230941772,
      "learning_rate": 1.4116812461763431e-05,
      "loss": 0.6009,
      "step": 11139
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.792538046836853,
      "learning_rate": 1.4115794871042457e-05,
      "loss": 0.5628,
      "step": 11140
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4923946857452393,
      "learning_rate": 1.4114777229008323e-05,
      "loss": 0.602,
      "step": 11141
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5018537044525146,
      "learning_rate": 1.4113759535673719e-05,
      "loss": 0.4238,
      "step": 11142
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5238286256790161,
      "learning_rate": 1.4112741791051327e-05,
      "loss": 0.5234,
      "step": 11143
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.503618836402893,
      "learning_rate": 1.4111723995153841e-05,
      "loss": 0.6474,
      "step": 11144
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5217347145080566,
      "learning_rate": 1.411070614799395e-05,
      "loss": 0.5489,
      "step": 11145
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5786607265472412,
      "learning_rate": 1.4109688249584342e-05,
      "loss": 0.5985,
      "step": 11146
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6589040756225586,
      "learning_rate": 1.410867029993771e-05,
      "loss": 0.5611,
      "step": 11147
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4364110231399536,
      "learning_rate": 1.4107652299066738e-05,
      "loss": 0.5043,
      "step": 11148
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6522005796432495,
      "learning_rate": 1.4106634246984127e-05,
      "loss": 0.4585,
      "step": 11149
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1403735876083374,
      "learning_rate": 1.4105616143702565e-05,
      "loss": 0.5019,
      "step": 11150
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5846612453460693,
      "learning_rate": 1.4104597989234745e-05,
      "loss": 0.5891,
      "step": 11151
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4020216464996338,
      "learning_rate": 1.4103579783593364e-05,
      "loss": 0.4826,
      "step": 11152
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1978378295898438,
      "learning_rate": 1.4102561526791109e-05,
      "loss": 0.4157,
      "step": 11153
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6916284561157227,
      "learning_rate": 1.410154321884068e-05,
      "loss": 0.5081,
      "step": 11154
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5833981037139893,
      "learning_rate": 1.4100524859754776e-05,
      "loss": 0.4675,
      "step": 11155
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5147452354431152,
      "learning_rate": 1.4099506449546088e-05,
      "loss": 0.767,
      "step": 11156
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4722516536712646,
      "learning_rate": 1.4098487988227312e-05,
      "loss": 0.6652,
      "step": 11157
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.559644103050232,
      "learning_rate": 1.4097469475811151e-05,
      "loss": 0.6559,
      "step": 11158
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.293257236480713,
      "learning_rate": 1.40964509123103e-05,
      "loss": 0.6114,
      "step": 11159
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.3236205577850342,
      "learning_rate": 1.4095432297737458e-05,
      "loss": 0.5115,
      "step": 11160
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.507490873336792,
      "learning_rate": 1.4094413632105324e-05,
      "loss": 0.5675,
      "step": 11161
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.670680046081543,
      "learning_rate": 1.4093394915426597e-05,
      "loss": 0.4622,
      "step": 11162
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.6060627698898315,
      "learning_rate": 1.409237614771398e-05,
      "loss": 0.5564,
      "step": 11163
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.476590871810913,
      "learning_rate": 1.4091357328980173e-05,
      "loss": 0.5591,
      "step": 11164
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5502524375915527,
      "learning_rate": 1.4090338459237878e-05,
      "loss": 0.6191,
      "step": 11165
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5013071298599243,
      "learning_rate": 1.40893195384998e-05,
      "loss": 0.6002,
      "step": 11166
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.368250846862793,
      "learning_rate": 1.408830056677864e-05,
      "loss": 0.5118,
      "step": 11167
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5938578844070435,
      "learning_rate": 1.4087281544087098e-05,
      "loss": 0.6749,
      "step": 11168
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4734365940093994,
      "learning_rate": 1.408626247043789e-05,
      "loss": 0.5116,
      "step": 11169
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5371273756027222,
      "learning_rate": 1.4085243345843711e-05,
      "loss": 0.6539,
      "step": 11170
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4794870615005493,
      "learning_rate": 1.4084224170317269e-05,
      "loss": 0.4697,
      "step": 11171
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.320096731185913,
      "learning_rate": 1.4083204943871272e-05,
      "loss": 0.4708,
      "step": 11172
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6728297472000122,
      "learning_rate": 1.4082185666518426e-05,
      "loss": 0.5125,
      "step": 11173
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6493269205093384,
      "learning_rate": 1.408116633827144e-05,
      "loss": 0.4822,
      "step": 11174
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4686815738677979,
      "learning_rate": 1.4080146959143021e-05,
      "loss": 0.4629,
      "step": 11175
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4461678266525269,
      "learning_rate": 1.407912752914588e-05,
      "loss": 0.5236,
      "step": 11176
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5381619930267334,
      "learning_rate": 1.4078108048292723e-05,
      "loss": 0.5918,
      "step": 11177
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.3708821535110474,
      "learning_rate": 1.407708851659626e-05,
      "loss": 0.4804,
      "step": 11178
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6390464305877686,
      "learning_rate": 1.407606893406921e-05,
      "loss": 0.5016,
      "step": 11179
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5277208089828491,
      "learning_rate": 1.4075049300724275e-05,
      "loss": 0.5312,
      "step": 11180
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5632069110870361,
      "learning_rate": 1.407402961657417e-05,
      "loss": 0.559,
      "step": 11181
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.745205283164978,
      "learning_rate": 1.4073009881631612e-05,
      "loss": 0.6058,
      "step": 11182
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.2854082584381104,
      "learning_rate": 1.4071990095909306e-05,
      "loss": 0.5397,
      "step": 11183
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4222476482391357,
      "learning_rate": 1.4070970259419979e-05,
      "loss": 0.5779,
      "step": 11184
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.9208472967147827,
      "learning_rate": 1.406995037217633e-05,
      "loss": 0.6283,
      "step": 11185
    },
    {
      "epoch": 1.16,
      "grad_norm": 2.241781234741211,
      "learning_rate": 1.4068930434191087e-05,
      "loss": 0.5509,
      "step": 11186
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5986831188201904,
      "learning_rate": 1.4067910445476958e-05,
      "loss": 0.5754,
      "step": 11187
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5815783739089966,
      "learning_rate": 1.4066890406046665e-05,
      "loss": 0.505,
      "step": 11188
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5749000310897827,
      "learning_rate": 1.4065870315912921e-05,
      "loss": 0.544,
      "step": 11189
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.573041319847107,
      "learning_rate": 1.4064850175088447e-05,
      "loss": 0.4704,
      "step": 11190
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6611223220825195,
      "learning_rate": 1.4063829983585958e-05,
      "loss": 0.6169,
      "step": 11191
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.818965196609497,
      "learning_rate": 1.4062809741418178e-05,
      "loss": 0.5858,
      "step": 11192
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5112488269805908,
      "learning_rate": 1.4061789448597822e-05,
      "loss": 0.5889,
      "step": 11193
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.371616005897522,
      "learning_rate": 1.4060769105137613e-05,
      "loss": 0.5526,
      "step": 11194
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.336452841758728,
      "learning_rate": 1.4059748711050274e-05,
      "loss": 0.4164,
      "step": 11195
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.431022047996521,
      "learning_rate": 1.405872826634852e-05,
      "loss": 0.554,
      "step": 11196
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6963478326797485,
      "learning_rate": 1.405770777104508e-05,
      "loss": 0.4691,
      "step": 11197
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.9809355735778809,
      "learning_rate": 1.4056687225152672e-05,
      "loss": 0.4485,
      "step": 11198
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.491974949836731,
      "learning_rate": 1.4055666628684024e-05,
      "loss": 0.5519,
      "step": 11199
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4822032451629639,
      "learning_rate": 1.4054645981651857e-05,
      "loss": 0.6045,
      "step": 11200
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.961432695388794,
      "learning_rate": 1.4053625284068896e-05,
      "loss": 0.6942,
      "step": 11201
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4366661310195923,
      "learning_rate": 1.4052604535947869e-05,
      "loss": 0.5054,
      "step": 11202
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6759226322174072,
      "learning_rate": 1.4051583737301499e-05,
      "loss": 0.6167,
      "step": 11203
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.8482611179351807,
      "learning_rate": 1.4050562888142514e-05,
      "loss": 0.4663,
      "step": 11204
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7358729839324951,
      "learning_rate": 1.4049541988483642e-05,
      "loss": 0.5998,
      "step": 11205
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5776989459991455,
      "learning_rate": 1.4048521038337608e-05,
      "loss": 0.6503,
      "step": 11206
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5817102193832397,
      "learning_rate": 1.4047500037717141e-05,
      "loss": 0.6478,
      "step": 11207
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.721869945526123,
      "learning_rate": 1.4046478986634976e-05,
      "loss": 0.595,
      "step": 11208
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.2532548904418945,
      "learning_rate": 1.4045457885103838e-05,
      "loss": 0.4772,
      "step": 11209
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4822860956192017,
      "learning_rate": 1.4044436733136458e-05,
      "loss": 0.6816,
      "step": 11210
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6401252746582031,
      "learning_rate": 1.4043415530745565e-05,
      "loss": 0.6078,
      "step": 11211
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7711042165756226,
      "learning_rate": 1.4042394277943894e-05,
      "loss": 0.581,
      "step": 11212
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6066488027572632,
      "learning_rate": 1.4041372974744178e-05,
      "loss": 0.6018,
      "step": 11213
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.382322907447815,
      "learning_rate": 1.4040351621159146e-05,
      "loss": 0.5409,
      "step": 11214
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5901824235916138,
      "learning_rate": 1.4039330217201536e-05,
      "loss": 0.5965,
      "step": 11215
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.409172534942627,
      "learning_rate": 1.4038308762884079e-05,
      "loss": 0.68,
      "step": 11216
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.3854413032531738,
      "learning_rate": 1.4037287258219512e-05,
      "loss": 0.4723,
      "step": 11217
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.320052981376648,
      "learning_rate": 1.4036265703220569e-05,
      "loss": 0.559,
      "step": 11218
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.3748779296875,
      "learning_rate": 1.4035244097899984e-05,
      "loss": 0.4912,
      "step": 11219
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6579605340957642,
      "learning_rate": 1.4034222442270498e-05,
      "loss": 0.5352,
      "step": 11220
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6618859767913818,
      "learning_rate": 1.4033200736344848e-05,
      "loss": 0.4727,
      "step": 11221
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4836905002593994,
      "learning_rate": 1.4032178980135768e-05,
      "loss": 0.5763,
      "step": 11222
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.550959825515747,
      "learning_rate": 1.4031157173656002e-05,
      "loss": 0.5752,
      "step": 11223
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5108731985092163,
      "learning_rate": 1.4030135316918286e-05,
      "loss": 0.445,
      "step": 11224
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6598682403564453,
      "learning_rate": 1.4029113409935357e-05,
      "loss": 0.6226,
      "step": 11225
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7133437395095825,
      "learning_rate": 1.4028091452719963e-05,
      "loss": 0.5548,
      "step": 11226
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.778364658355713,
      "learning_rate": 1.4027069445284838e-05,
      "loss": 0.4701,
      "step": 11227
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5380421876907349,
      "learning_rate": 1.4026047387642729e-05,
      "loss": 0.5013,
      "step": 11228
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6757622957229614,
      "learning_rate": 1.4025025279806374e-05,
      "loss": 0.5909,
      "step": 11229
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6294959783554077,
      "learning_rate": 1.4024003121788518e-05,
      "loss": 0.6643,
      "step": 11230
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7447023391723633,
      "learning_rate": 1.4022980913601907e-05,
      "loss": 0.5297,
      "step": 11231
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4680365324020386,
      "learning_rate": 1.4021958655259283e-05,
      "loss": 0.5139,
      "step": 11232
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1330856084823608,
      "learning_rate": 1.4020936346773386e-05,
      "loss": 0.621,
      "step": 11233
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.3689223527908325,
      "learning_rate": 1.4019913988156973e-05,
      "loss": 0.5744,
      "step": 11234
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5842623710632324,
      "learning_rate": 1.401889157942278e-05,
      "loss": 0.5782,
      "step": 11235
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.447576642036438,
      "learning_rate": 1.4017869120583558e-05,
      "loss": 0.5479,
      "step": 11236
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.390872597694397,
      "learning_rate": 1.4016846611652053e-05,
      "loss": 0.5614,
      "step": 11237
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4693235158920288,
      "learning_rate": 1.4015824052641013e-05,
      "loss": 0.6014,
      "step": 11238
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.599769115447998,
      "learning_rate": 1.401480144356319e-05,
      "loss": 0.5171,
      "step": 11239
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.2423524856567383,
      "learning_rate": 1.4013778784431327e-05,
      "loss": 0.4875,
      "step": 11240
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5939306020736694,
      "learning_rate": 1.4012756075258183e-05,
      "loss": 0.5597,
      "step": 11241
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.500075101852417,
      "learning_rate": 1.4011733316056497e-05,
      "loss": 0.5329,
      "step": 11242
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.3137882947921753,
      "learning_rate": 1.4010710506839027e-05,
      "loss": 0.5579,
      "step": 11243
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5498063564300537,
      "learning_rate": 1.4009687647618527e-05,
      "loss": 0.5253,
      "step": 11244
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6756641864776611,
      "learning_rate": 1.4008664738407743e-05,
      "loss": 0.5519,
      "step": 11245
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5283299684524536,
      "learning_rate": 1.400764177921943e-05,
      "loss": 0.4437,
      "step": 11246
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6929031610488892,
      "learning_rate": 1.4006618770066346e-05,
      "loss": 0.5704,
      "step": 11247
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.613267183303833,
      "learning_rate": 1.400559571096124e-05,
      "loss": 0.5426,
      "step": 11248
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6106555461883545,
      "learning_rate": 1.400457260191687e-05,
      "loss": 0.574,
      "step": 11249
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.611055612564087,
      "learning_rate": 1.400354944294599e-05,
      "loss": 0.6207,
      "step": 11250
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5389615297317505,
      "learning_rate": 1.4002526234061356e-05,
      "loss": 0.5654,
      "step": 11251
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6189074516296387,
      "learning_rate": 1.4001502975275724e-05,
      "loss": 0.6813,
      "step": 11252
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.606860876083374,
      "learning_rate": 1.4000479666601856e-05,
      "loss": 0.4993,
      "step": 11253
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.741920828819275,
      "learning_rate": 1.3999456308052507e-05,
      "loss": 0.6488,
      "step": 11254
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.2672306299209595,
      "learning_rate": 1.3998432899640428e-05,
      "loss": 0.5955,
      "step": 11255
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5253757238388062,
      "learning_rate": 1.3997409441378391e-05,
      "loss": 0.5563,
      "step": 11256
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5602878332138062,
      "learning_rate": 1.3996385933279149e-05,
      "loss": 0.5282,
      "step": 11257
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.3642929792404175,
      "learning_rate": 1.3995362375355464e-05,
      "loss": 0.5046,
      "step": 11258
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.406686544418335,
      "learning_rate": 1.3994338767620095e-05,
      "loss": 0.4701,
      "step": 11259
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.7537349462509155,
      "learning_rate": 1.3993315110085807e-05,
      "loss": 0.6802,
      "step": 11260
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5749439001083374,
      "learning_rate": 1.399229140276536e-05,
      "loss": 0.6142,
      "step": 11261
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.6375254392623901,
      "learning_rate": 1.3991267645671517e-05,
      "loss": 0.5971,
      "step": 11262
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.4633065462112427,
      "learning_rate": 1.3990243838817044e-05,
      "loss": 0.457,
      "step": 11263
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.5259408950805664,
      "learning_rate": 1.3989219982214701e-05,
      "loss": 0.5316,
      "step": 11264
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.525693655014038,
      "learning_rate": 1.3988196075877259e-05,
      "loss": 0.6553,
      "step": 11265
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.731961965560913,
      "learning_rate": 1.3987172119817476e-05,
      "loss": 0.5265,
      "step": 11266
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3569765090942383,
      "learning_rate": 1.3986148114048124e-05,
      "loss": 0.5721,
      "step": 11267
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5283972024917603,
      "learning_rate": 1.3985124058581964e-05,
      "loss": 0.6129,
      "step": 11268
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.2240251302719116,
      "learning_rate": 1.3984099953431769e-05,
      "loss": 0.5856,
      "step": 11269
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4224668741226196,
      "learning_rate": 1.3983075798610303e-05,
      "loss": 0.4433,
      "step": 11270
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.9353625774383545,
      "learning_rate": 1.398205159413034e-05,
      "loss": 0.4811,
      "step": 11271
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.262602686882019,
      "learning_rate": 1.398102734000464e-05,
      "loss": 0.6067,
      "step": 11272
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4172016382217407,
      "learning_rate": 1.3980003036245981e-05,
      "loss": 0.6245,
      "step": 11273
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6104267835617065,
      "learning_rate": 1.3978978682867128e-05,
      "loss": 0.5228,
      "step": 11274
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6189584732055664,
      "learning_rate": 1.3977954279880858e-05,
      "loss": 0.5917,
      "step": 11275
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8226873874664307,
      "learning_rate": 1.3976929827299937e-05,
      "loss": 0.532,
      "step": 11276
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.9033150672912598,
      "learning_rate": 1.3975905325137137e-05,
      "loss": 0.5353,
      "step": 11277
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7710131406784058,
      "learning_rate": 1.3974880773405238e-05,
      "loss": 0.6099,
      "step": 11278
    },
    {
      "epoch": 1.17,
      "grad_norm": 2.1616568565368652,
      "learning_rate": 1.3973856172117005e-05,
      "loss": 0.7052,
      "step": 11279
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.441229224205017,
      "learning_rate": 1.3972831521285217e-05,
      "loss": 0.5177,
      "step": 11280
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5411430597305298,
      "learning_rate": 1.3971806820922644e-05,
      "loss": 0.6007,
      "step": 11281
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.647226095199585,
      "learning_rate": 1.3970782071042065e-05,
      "loss": 0.5446,
      "step": 11282
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3886021375656128,
      "learning_rate": 1.3969757271656262e-05,
      "loss": 0.4511,
      "step": 11283
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5523804426193237,
      "learning_rate": 1.3968732422778e-05,
      "loss": 0.5363,
      "step": 11284
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5712769031524658,
      "learning_rate": 1.396770752442006e-05,
      "loss": 0.6026,
      "step": 11285
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6934189796447754,
      "learning_rate": 1.3966682576595223e-05,
      "loss": 0.5927,
      "step": 11286
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.32762610912323,
      "learning_rate": 1.3965657579316265e-05,
      "loss": 0.5658,
      "step": 11287
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5647070407867432,
      "learning_rate": 1.3964632532595968e-05,
      "loss": 0.4868,
      "step": 11288
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5072407722473145,
      "learning_rate": 1.3963607436447106e-05,
      "loss": 0.4919,
      "step": 11289
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4235635995864868,
      "learning_rate": 1.3962582290882463e-05,
      "loss": 0.6031,
      "step": 11290
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5833864212036133,
      "learning_rate": 1.3961557095914819e-05,
      "loss": 0.5798,
      "step": 11291
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5871267318725586,
      "learning_rate": 1.3960531851556958e-05,
      "loss": 0.4295,
      "step": 11292
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3574528694152832,
      "learning_rate": 1.3959506557821657e-05,
      "loss": 0.5241,
      "step": 11293
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7792466878890991,
      "learning_rate": 1.3958481214721704e-05,
      "loss": 0.5274,
      "step": 11294
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5734188556671143,
      "learning_rate": 1.3957455822269878e-05,
      "loss": 0.5882,
      "step": 11295
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4457963705062866,
      "learning_rate": 1.3956430380478966e-05,
      "loss": 0.4586,
      "step": 11296
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5620406866073608,
      "learning_rate": 1.3955404889361753e-05,
      "loss": 0.5729,
      "step": 11297
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5470292568206787,
      "learning_rate": 1.395437934893102e-05,
      "loss": 0.4315,
      "step": 11298
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.495862364768982,
      "learning_rate": 1.3953353759199555e-05,
      "loss": 0.5055,
      "step": 11299
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4166465997695923,
      "learning_rate": 1.3952328120180147e-05,
      "loss": 0.5214,
      "step": 11300
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7889379262924194,
      "learning_rate": 1.3951302431885581e-05,
      "loss": 0.5254,
      "step": 11301
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7184958457946777,
      "learning_rate": 1.3950276694328642e-05,
      "loss": 0.5818,
      "step": 11302
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5829416513442993,
      "learning_rate": 1.3949250907522123e-05,
      "loss": 0.5026,
      "step": 11303
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5147833824157715,
      "learning_rate": 1.3948225071478811e-05,
      "loss": 0.5525,
      "step": 11304
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5216681957244873,
      "learning_rate": 1.3947199186211493e-05,
      "loss": 0.6933,
      "step": 11305
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6133949756622314,
      "learning_rate": 1.3946173251732962e-05,
      "loss": 0.5386,
      "step": 11306
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4750027656555176,
      "learning_rate": 1.3945147268056007e-05,
      "loss": 0.6753,
      "step": 11307
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.468977928161621,
      "learning_rate": 1.3944121235193423e-05,
      "loss": 0.5442,
      "step": 11308
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4548664093017578,
      "learning_rate": 1.3943095153157997e-05,
      "loss": 0.4817,
      "step": 11309
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5245928764343262,
      "learning_rate": 1.3942069021962526e-05,
      "loss": 0.4985,
      "step": 11310
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6344237327575684,
      "learning_rate": 1.39410428416198e-05,
      "loss": 0.4582,
      "step": 11311
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8093911409378052,
      "learning_rate": 1.394001661214261e-05,
      "loss": 0.4517,
      "step": 11312
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.531718373298645,
      "learning_rate": 1.3938990333543759e-05,
      "loss": 0.4056,
      "step": 11313
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.405678153038025,
      "learning_rate": 1.3937964005836036e-05,
      "loss": 0.5186,
      "step": 11314
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4339613914489746,
      "learning_rate": 1.3936937629032235e-05,
      "loss": 0.5136,
      "step": 11315
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5812007188796997,
      "learning_rate": 1.3935911203145157e-05,
      "loss": 0.4831,
      "step": 11316
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6303927898406982,
      "learning_rate": 1.3934884728187596e-05,
      "loss": 0.5979,
      "step": 11317
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6422083377838135,
      "learning_rate": 1.3933858204172351e-05,
      "loss": 0.5655,
      "step": 11318
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.599703311920166,
      "learning_rate": 1.393283163111222e-05,
      "loss": 0.7502,
      "step": 11319
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7864995002746582,
      "learning_rate": 1.393180500902e-05,
      "loss": 0.4897,
      "step": 11320
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5227253437042236,
      "learning_rate": 1.3930778337908489e-05,
      "loss": 0.5149,
      "step": 11321
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.721816062927246,
      "learning_rate": 1.3929751617790493e-05,
      "loss": 0.5431,
      "step": 11322
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.473281741142273,
      "learning_rate": 1.3928724848678805e-05,
      "loss": 0.5743,
      "step": 11323
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3952584266662598,
      "learning_rate": 1.3927698030586233e-05,
      "loss": 0.4399,
      "step": 11324
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3960535526275635,
      "learning_rate": 1.3926671163525574e-05,
      "loss": 0.5077,
      "step": 11325
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4952661991119385,
      "learning_rate": 1.392564424750963e-05,
      "loss": 0.455,
      "step": 11326
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5226274728775024,
      "learning_rate": 1.3924617282551208e-05,
      "loss": 0.5287,
      "step": 11327
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5723482370376587,
      "learning_rate": 1.3923590268663107e-05,
      "loss": 0.5627,
      "step": 11328
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4358224868774414,
      "learning_rate": 1.3922563205858136e-05,
      "loss": 0.5042,
      "step": 11329
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3877763748168945,
      "learning_rate": 1.3921536094149097e-05,
      "loss": 0.6762,
      "step": 11330
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4712347984313965,
      "learning_rate": 1.3920508933548793e-05,
      "loss": 0.6214,
      "step": 11331
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.649068832397461,
      "learning_rate": 1.3919481724070036e-05,
      "loss": 0.5552,
      "step": 11332
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.661795735359192,
      "learning_rate": 1.3918454465725628e-05,
      "loss": 0.4895,
      "step": 11333
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.623796820640564,
      "learning_rate": 1.3917427158528376e-05,
      "loss": 0.5685,
      "step": 11334
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7186787128448486,
      "learning_rate": 1.3916399802491095e-05,
      "loss": 0.5667,
      "step": 11335
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4598920345306396,
      "learning_rate": 1.3915372397626584e-05,
      "loss": 0.566,
      "step": 11336
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.448002815246582,
      "learning_rate": 1.391434494394765e-05,
      "loss": 0.583,
      "step": 11337
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.690617322921753,
      "learning_rate": 1.3913317441467115e-05,
      "loss": 0.5517,
      "step": 11338
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5073192119598389,
      "learning_rate": 1.391228989019778e-05,
      "loss": 0.5586,
      "step": 11339
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3717947006225586,
      "learning_rate": 1.391126229015246e-05,
      "loss": 0.6483,
      "step": 11340
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5870046615600586,
      "learning_rate": 1.3910234641343965e-05,
      "loss": 0.6321,
      "step": 11341
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3782380819320679,
      "learning_rate": 1.3909206943785108e-05,
      "loss": 0.6047,
      "step": 11342
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.775698184967041,
      "learning_rate": 1.3908179197488699e-05,
      "loss": 0.5473,
      "step": 11343
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8855806589126587,
      "learning_rate": 1.3907151402467553e-05,
      "loss": 0.757,
      "step": 11344
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.8579100370407104,
      "learning_rate": 1.3906123558734484e-05,
      "loss": 0.5233,
      "step": 11345
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4489128589630127,
      "learning_rate": 1.3905095666302308e-05,
      "loss": 0.5964,
      "step": 11346
    },
    {
      "epoch": 1.17,
      "grad_norm": 2.0395894050598145,
      "learning_rate": 1.3904067725183837e-05,
      "loss": 0.6044,
      "step": 11347
    },
    {
      "epoch": 1.17,
      "grad_norm": 2.0756406784057617,
      "learning_rate": 1.390303973539189e-05,
      "loss": 0.5409,
      "step": 11348
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4345782995224,
      "learning_rate": 1.390201169693928e-05,
      "loss": 0.501,
      "step": 11349
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5246219635009766,
      "learning_rate": 1.3900983609838826e-05,
      "loss": 0.5118,
      "step": 11350
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4240275621414185,
      "learning_rate": 1.3899955474103344e-05,
      "loss": 0.6257,
      "step": 11351
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.4036290645599365,
      "learning_rate": 1.3898927289745654e-05,
      "loss": 0.4413,
      "step": 11352
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.7282441854476929,
      "learning_rate": 1.3897899056778574e-05,
      "loss": 0.5144,
      "step": 11353
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.5876414775848389,
      "learning_rate": 1.3896870775214926e-05,
      "loss": 0.5623,
      "step": 11354
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.6996979713439941,
      "learning_rate": 1.3895842445067524e-05,
      "loss": 0.4611,
      "step": 11355
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.642279028892517,
      "learning_rate": 1.3894814066349196e-05,
      "loss": 0.6026,
      "step": 11356
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3926000595092773,
      "learning_rate": 1.3893785639072755e-05,
      "loss": 0.5589,
      "step": 11357
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.41318941116333,
      "learning_rate": 1.3892757163251027e-05,
      "loss": 0.5781,
      "step": 11358
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.308897614479065,
      "learning_rate": 1.3891728638896838e-05,
      "loss": 0.6786,
      "step": 11359
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.493428349494934,
      "learning_rate": 1.3890700066023004e-05,
      "loss": 0.6206,
      "step": 11360
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.912824034690857,
      "learning_rate": 1.3889671444642354e-05,
      "loss": 0.7302,
      "step": 11361
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4452910423278809,
      "learning_rate": 1.3888642774767709e-05,
      "loss": 0.5564,
      "step": 11362
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7581452131271362,
      "learning_rate": 1.3887614056411893e-05,
      "loss": 0.6569,
      "step": 11363
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4483757019042969,
      "learning_rate": 1.3886585289587738e-05,
      "loss": 0.5853,
      "step": 11364
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4561022520065308,
      "learning_rate": 1.3885556474308062e-05,
      "loss": 0.6467,
      "step": 11365
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4170783758163452,
      "learning_rate": 1.3884527610585697e-05,
      "loss": 0.5694,
      "step": 11366
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7706331014633179,
      "learning_rate": 1.3883498698433468e-05,
      "loss": 0.4372,
      "step": 11367
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3522619009017944,
      "learning_rate": 1.3882469737864202e-05,
      "loss": 0.5111,
      "step": 11368
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4503909349441528,
      "learning_rate": 1.388144072889073e-05,
      "loss": 0.454,
      "step": 11369
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3422895669937134,
      "learning_rate": 1.3880411671525878e-05,
      "loss": 0.5189,
      "step": 11370
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.544748306274414,
      "learning_rate": 1.3879382565782478e-05,
      "loss": 0.6037,
      "step": 11371
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3161911964416504,
      "learning_rate": 1.387835341167336e-05,
      "loss": 0.4773,
      "step": 11372
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2801079750061035,
      "learning_rate": 1.3877324209211352e-05,
      "loss": 0.4875,
      "step": 11373
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1995337009429932,
      "learning_rate": 1.387629495840929e-05,
      "loss": 0.5899,
      "step": 11374
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6778451204299927,
      "learning_rate": 1.3875265659280004e-05,
      "loss": 0.578,
      "step": 11375
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3875571489334106,
      "learning_rate": 1.3874236311836325e-05,
      "loss": 0.5971,
      "step": 11376
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.507550597190857,
      "learning_rate": 1.387320691609109e-05,
      "loss": 0.6456,
      "step": 11377
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.8313246965408325,
      "learning_rate": 1.387217747205713e-05,
      "loss": 0.5585,
      "step": 11378
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.936367392539978,
      "learning_rate": 1.3871147979747278e-05,
      "loss": 0.5876,
      "step": 11379
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7953985929489136,
      "learning_rate": 1.3870118439174371e-05,
      "loss": 0.5113,
      "step": 11380
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.84988534450531,
      "learning_rate": 1.3869088850351246e-05,
      "loss": 0.5782,
      "step": 11381
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3210055828094482,
      "learning_rate": 1.3868059213290741e-05,
      "loss": 0.502,
      "step": 11382
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7137162685394287,
      "learning_rate": 1.3867029528005686e-05,
      "loss": 0.6458,
      "step": 11383
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5663472414016724,
      "learning_rate": 1.3865999794508923e-05,
      "loss": 0.5148,
      "step": 11384
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.562964916229248,
      "learning_rate": 1.3864970012813293e-05,
      "loss": 0.4831,
      "step": 11385
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2962361574172974,
      "learning_rate": 1.3863940182931628e-05,
      "loss": 0.5057,
      "step": 11386
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6145377159118652,
      "learning_rate": 1.3862910304876771e-05,
      "loss": 0.7014,
      "step": 11387
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5459020137786865,
      "learning_rate": 1.3861880378661562e-05,
      "loss": 0.5571,
      "step": 11388
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5180383920669556,
      "learning_rate": 1.3860850404298837e-05,
      "loss": 0.5689,
      "step": 11389
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4456205368041992,
      "learning_rate": 1.3859820381801445e-05,
      "loss": 0.5213,
      "step": 11390
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3772906064987183,
      "learning_rate": 1.3858790311182223e-05,
      "loss": 0.4863,
      "step": 11391
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2923333644866943,
      "learning_rate": 1.3857760192454012e-05,
      "loss": 0.5039,
      "step": 11392
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2178527116775513,
      "learning_rate": 1.3856730025629661e-05,
      "loss": 0.5971,
      "step": 11393
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6872444152832031,
      "learning_rate": 1.3855699810722006e-05,
      "loss": 0.6714,
      "step": 11394
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6922639608383179,
      "learning_rate": 1.3854669547743896e-05,
      "loss": 0.5605,
      "step": 11395
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7791510820388794,
      "learning_rate": 1.3853639236708169e-05,
      "loss": 0.6111,
      "step": 11396
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3988181352615356,
      "learning_rate": 1.385260887762768e-05,
      "loss": 0.6101,
      "step": 11397
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5516377687454224,
      "learning_rate": 1.385157847051527e-05,
      "loss": 0.6271,
      "step": 11398
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2549924850463867,
      "learning_rate": 1.3850548015383782e-05,
      "loss": 0.55,
      "step": 11399
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6357593536376953,
      "learning_rate": 1.3849517512246072e-05,
      "loss": 0.5691,
      "step": 11400
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3530783653259277,
      "learning_rate": 1.3848486961114978e-05,
      "loss": 0.6937,
      "step": 11401
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7547768354415894,
      "learning_rate": 1.3847456362003352e-05,
      "loss": 0.5741,
      "step": 11402
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2229039669036865,
      "learning_rate": 1.3846425714924048e-05,
      "loss": 0.5597,
      "step": 11403
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.441300392150879,
      "learning_rate": 1.3845395019889907e-05,
      "loss": 0.5618,
      "step": 11404
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.298439621925354,
      "learning_rate": 1.3844364276913784e-05,
      "loss": 0.5195,
      "step": 11405
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6996400356292725,
      "learning_rate": 1.384333348600853e-05,
      "loss": 0.5266,
      "step": 11406
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1759930849075317,
      "learning_rate": 1.3842302647186994e-05,
      "loss": 0.614,
      "step": 11407
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5882643461227417,
      "learning_rate": 1.384127176046203e-05,
      "loss": 0.6002,
      "step": 11408
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.471495270729065,
      "learning_rate": 1.3840240825846488e-05,
      "loss": 0.5937,
      "step": 11409
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4144890308380127,
      "learning_rate": 1.383920984335322e-05,
      "loss": 0.436,
      "step": 11410
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.805751919746399,
      "learning_rate": 1.3838178812995085e-05,
      "loss": 0.5602,
      "step": 11411
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.632758378982544,
      "learning_rate": 1.3837147734784932e-05,
      "loss": 0.667,
      "step": 11412
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.371748447418213,
      "learning_rate": 1.383611660873562e-05,
      "loss": 0.5303,
      "step": 11413
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4258663654327393,
      "learning_rate": 1.3835085434860001e-05,
      "loss": 0.5535,
      "step": 11414
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4818413257598877,
      "learning_rate": 1.3834054213170931e-05,
      "loss": 0.683,
      "step": 11415
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.478590726852417,
      "learning_rate": 1.3833022943681271e-05,
      "loss": 0.6102,
      "step": 11416
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2084747552871704,
      "learning_rate": 1.3831991626403873e-05,
      "loss": 0.5095,
      "step": 11417
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3117280006408691,
      "learning_rate": 1.3830960261351598e-05,
      "loss": 0.4414,
      "step": 11418
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.477712869644165,
      "learning_rate": 1.3829928848537306e-05,
      "loss": 0.5034,
      "step": 11419
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.79470694065094,
      "learning_rate": 1.382889738797385e-05,
      "loss": 0.6058,
      "step": 11420
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.8335613012313843,
      "learning_rate": 1.3827865879674097e-05,
      "loss": 0.6234,
      "step": 11421
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.9515708684921265,
      "learning_rate": 1.3826834323650899e-05,
      "loss": 0.5908,
      "step": 11422
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4897844791412354,
      "learning_rate": 1.3825802719917123e-05,
      "loss": 0.6202,
      "step": 11423
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.8745418787002563,
      "learning_rate": 1.382477106848563e-05,
      "loss": 0.6097,
      "step": 11424
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4114216566085815,
      "learning_rate": 1.3823739369369277e-05,
      "loss": 0.4699,
      "step": 11425
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2677544355392456,
      "learning_rate": 1.3822707622580934e-05,
      "loss": 0.4125,
      "step": 11426
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4203208684921265,
      "learning_rate": 1.3821675828133457e-05,
      "loss": 0.6257,
      "step": 11427
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3128058910369873,
      "learning_rate": 1.3820643986039714e-05,
      "loss": 0.4954,
      "step": 11428
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5517771244049072,
      "learning_rate": 1.381961209631257e-05,
      "loss": 0.6033,
      "step": 11429
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5738186836242676,
      "learning_rate": 1.3818580158964888e-05,
      "loss": 0.6388,
      "step": 11430
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4593068361282349,
      "learning_rate": 1.3817548174009532e-05,
      "loss": 0.6124,
      "step": 11431
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5169248580932617,
      "learning_rate": 1.381651614145937e-05,
      "loss": 0.6325,
      "step": 11432
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5933359861373901,
      "learning_rate": 1.381548406132727e-05,
      "loss": 0.5732,
      "step": 11433
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.7750821113586426,
      "learning_rate": 1.3814451933626098e-05,
      "loss": 0.5343,
      "step": 11434
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.8385192155838013,
      "learning_rate": 1.3813419758368722e-05,
      "loss": 0.5633,
      "step": 11435
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.235235333442688,
      "learning_rate": 1.381238753556801e-05,
      "loss": 0.5485,
      "step": 11436
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6921075582504272,
      "learning_rate": 1.3811355265236834e-05,
      "loss": 0.5015,
      "step": 11437
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4172718524932861,
      "learning_rate": 1.381032294738806e-05,
      "loss": 0.4673,
      "step": 11438
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6879671812057495,
      "learning_rate": 1.380929058203456e-05,
      "loss": 0.5736,
      "step": 11439
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4947936534881592,
      "learning_rate": 1.3808258169189203e-05,
      "loss": 0.6196,
      "step": 11440
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.827628254890442,
      "learning_rate": 1.3807225708864862e-05,
      "loss": 0.6637,
      "step": 11441
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5723061561584473,
      "learning_rate": 1.3806193201074411e-05,
      "loss": 0.5448,
      "step": 11442
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2945765256881714,
      "learning_rate": 1.380516064583072e-05,
      "loss": 0.5769,
      "step": 11443
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.8179845809936523,
      "learning_rate": 1.3804128043146661e-05,
      "loss": 0.516,
      "step": 11444
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5944702625274658,
      "learning_rate": 1.3803095393035113e-05,
      "loss": 0.6114,
      "step": 11445
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.585087776184082,
      "learning_rate": 1.3802062695508945e-05,
      "loss": 0.4524,
      "step": 11446
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4995616674423218,
      "learning_rate": 1.3801029950581038e-05,
      "loss": 0.5725,
      "step": 11447
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2709414958953857,
      "learning_rate": 1.379999715826426e-05,
      "loss": 0.5403,
      "step": 11448
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4209716320037842,
      "learning_rate": 1.3798964318571491e-05,
      "loss": 0.5778,
      "step": 11449
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6042786836624146,
      "learning_rate": 1.3797931431515609e-05,
      "loss": 0.5153,
      "step": 11450
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.5014032125473022,
      "learning_rate": 1.3796898497109492e-05,
      "loss": 0.645,
      "step": 11451
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.592453122138977,
      "learning_rate": 1.3795865515366016e-05,
      "loss": 0.6202,
      "step": 11452
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4559338092803955,
      "learning_rate": 1.3794832486298057e-05,
      "loss": 0.5409,
      "step": 11453
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.4121185541152954,
      "learning_rate": 1.3793799409918499e-05,
      "loss": 0.6089,
      "step": 11454
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3541040420532227,
      "learning_rate": 1.379276628624022e-05,
      "loss": 0.6383,
      "step": 11455
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.8710962533950806,
      "learning_rate": 1.3791733115276101e-05,
      "loss": 0.6464,
      "step": 11456
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3859367370605469,
      "learning_rate": 1.379069989703902e-05,
      "loss": 0.6586,
      "step": 11457
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6894208192825317,
      "learning_rate": 1.378966663154186e-05,
      "loss": 0.583,
      "step": 11458
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3113454580307007,
      "learning_rate": 1.3788633318797506e-05,
      "loss": 0.618,
      "step": 11459
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.763638973236084,
      "learning_rate": 1.3787599958818839e-05,
      "loss": 0.4847,
      "step": 11460
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.591401219367981,
      "learning_rate": 1.3786566551618741e-05,
      "loss": 0.6238,
      "step": 11461
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6162225008010864,
      "learning_rate": 1.3785533097210093e-05,
      "loss": 0.579,
      "step": 11462
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5625343322753906,
      "learning_rate": 1.3784499595605786e-05,
      "loss": 0.5818,
      "step": 11463
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7730165719985962,
      "learning_rate": 1.3783466046818702e-05,
      "loss": 0.7268,
      "step": 11464
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7586578130722046,
      "learning_rate": 1.3782432450861729e-05,
      "loss": 0.647,
      "step": 11465
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.355284333229065,
      "learning_rate": 1.3781398807747748e-05,
      "loss": 0.5876,
      "step": 11466
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.537812352180481,
      "learning_rate": 1.3780365117489648e-05,
      "loss": 0.6198,
      "step": 11467
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.37628173828125,
      "learning_rate": 1.3779331380100321e-05,
      "loss": 0.617,
      "step": 11468
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.334542989730835,
      "learning_rate": 1.3778297595592647e-05,
      "loss": 0.4495,
      "step": 11469
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3454071283340454,
      "learning_rate": 1.377726376397952e-05,
      "loss": 0.6273,
      "step": 11470
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5283260345458984,
      "learning_rate": 1.3776229885273828e-05,
      "loss": 0.6526,
      "step": 11471
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6977369785308838,
      "learning_rate": 1.3775195959488463e-05,
      "loss": 0.5726,
      "step": 11472
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3850314617156982,
      "learning_rate": 1.3774161986636312e-05,
      "loss": 0.5257,
      "step": 11473
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3988155126571655,
      "learning_rate": 1.3773127966730264e-05,
      "loss": 0.5053,
      "step": 11474
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4230817556381226,
      "learning_rate": 1.3772093899783214e-05,
      "loss": 0.5252,
      "step": 11475
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3180791139602661,
      "learning_rate": 1.3771059785808056e-05,
      "loss": 0.5738,
      "step": 11476
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7470579147338867,
      "learning_rate": 1.377002562481768e-05,
      "loss": 0.6269,
      "step": 11477
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6530271768569946,
      "learning_rate": 1.376899141682498e-05,
      "loss": 0.5789,
      "step": 11478
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3972018957138062,
      "learning_rate": 1.376795716184285e-05,
      "loss": 0.5917,
      "step": 11479
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.730862021446228,
      "learning_rate": 1.3766922859884177e-05,
      "loss": 0.6186,
      "step": 11480
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5163284540176392,
      "learning_rate": 1.3765888510961872e-05,
      "loss": 0.5737,
      "step": 11481
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6300034523010254,
      "learning_rate": 1.3764854115088815e-05,
      "loss": 0.6196,
      "step": 11482
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4547228813171387,
      "learning_rate": 1.3763819672277911e-05,
      "loss": 0.4862,
      "step": 11483
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5187232494354248,
      "learning_rate": 1.3762785182542055e-05,
      "loss": 0.4331,
      "step": 11484
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.455324649810791,
      "learning_rate": 1.3761750645894143e-05,
      "loss": 0.5952,
      "step": 11485
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3963383436203003,
      "learning_rate": 1.3760716062347075e-05,
      "loss": 0.6529,
      "step": 11486
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7870302200317383,
      "learning_rate": 1.3759681431913748e-05,
      "loss": 0.6691,
      "step": 11487
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.616151213645935,
      "learning_rate": 1.3758646754607057e-05,
      "loss": 0.5701,
      "step": 11488
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4189021587371826,
      "learning_rate": 1.3757612030439913e-05,
      "loss": 0.4556,
      "step": 11489
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.428447961807251,
      "learning_rate": 1.3756577259425204e-05,
      "loss": 0.4998,
      "step": 11490
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2763100862503052,
      "learning_rate": 1.375554244157584e-05,
      "loss": 0.6417,
      "step": 11491
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4542012214660645,
      "learning_rate": 1.3754507576904717e-05,
      "loss": 0.5743,
      "step": 11492
    },
    {
      "epoch": 1.19,
      "grad_norm": 2.9420247077941895,
      "learning_rate": 1.3753472665424738e-05,
      "loss": 0.6095,
      "step": 11493
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.648850679397583,
      "learning_rate": 1.3752437707148809e-05,
      "loss": 0.5716,
      "step": 11494
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5344171524047852,
      "learning_rate": 1.3751402702089829e-05,
      "loss": 0.636,
      "step": 11495
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7480602264404297,
      "learning_rate": 1.3750367650260703e-05,
      "loss": 0.7073,
      "step": 11496
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5661474466323853,
      "learning_rate": 1.3749332551674339e-05,
      "loss": 0.5485,
      "step": 11497
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2416104078292847,
      "learning_rate": 1.3748297406343636e-05,
      "loss": 0.3907,
      "step": 11498
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.673102855682373,
      "learning_rate": 1.3747262214281503e-05,
      "loss": 0.683,
      "step": 11499
    },
    {
      "epoch": 1.19,
      "grad_norm": 2.7349390983581543,
      "learning_rate": 1.3746226975500845e-05,
      "loss": 0.5572,
      "step": 11500
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3176374435424805,
      "learning_rate": 1.3745191690014568e-05,
      "loss": 0.5899,
      "step": 11501
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2356579303741455,
      "learning_rate": 1.3744156357835583e-05,
      "loss": 0.4925,
      "step": 11502
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7023518085479736,
      "learning_rate": 1.3743120978976796e-05,
      "loss": 0.5348,
      "step": 11503
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5481655597686768,
      "learning_rate": 1.3742085553451112e-05,
      "loss": 0.6149,
      "step": 11504
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.8489292860031128,
      "learning_rate": 1.3741050081271445e-05,
      "loss": 0.7028,
      "step": 11505
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4881616830825806,
      "learning_rate": 1.3740014562450701e-05,
      "loss": 0.6267,
      "step": 11506
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4765973091125488,
      "learning_rate": 1.3738978997001792e-05,
      "loss": 0.4714,
      "step": 11507
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5706379413604736,
      "learning_rate": 1.3737943384937628e-05,
      "loss": 0.5834,
      "step": 11508
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2424993515014648,
      "learning_rate": 1.3736907726271123e-05,
      "loss": 0.5046,
      "step": 11509
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.354169487953186,
      "learning_rate": 1.3735872021015189e-05,
      "loss": 0.4495,
      "step": 11510
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4204069375991821,
      "learning_rate": 1.3734836269182732e-05,
      "loss": 0.5987,
      "step": 11511
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.802757740020752,
      "learning_rate": 1.3733800470786669e-05,
      "loss": 0.541,
      "step": 11512
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6887640953063965,
      "learning_rate": 1.3732764625839915e-05,
      "loss": 0.5367,
      "step": 11513
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6104339361190796,
      "learning_rate": 1.3731728734355382e-05,
      "loss": 0.5727,
      "step": 11514
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5002853870391846,
      "learning_rate": 1.373069279634599e-05,
      "loss": 0.568,
      "step": 11515
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6678885221481323,
      "learning_rate": 1.3729656811824647e-05,
      "loss": 0.574,
      "step": 11516
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3018584251403809,
      "learning_rate": 1.3728620780804275e-05,
      "loss": 0.4739,
      "step": 11517
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7290699481964111,
      "learning_rate": 1.3727584703297789e-05,
      "loss": 0.6531,
      "step": 11518
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.301525354385376,
      "learning_rate": 1.3726548579318102e-05,
      "loss": 0.5514,
      "step": 11519
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.7807120084762573,
      "learning_rate": 1.3725512408878137e-05,
      "loss": 0.618,
      "step": 11520
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4530057907104492,
      "learning_rate": 1.3724476191990812e-05,
      "loss": 0.4891,
      "step": 11521
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.177103877067566,
      "learning_rate": 1.3723439928669042e-05,
      "loss": 0.6562,
      "step": 11522
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.521146535873413,
      "learning_rate": 1.372240361892575e-05,
      "loss": 0.5634,
      "step": 11523
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3000253438949585,
      "learning_rate": 1.3721367262773852e-05,
      "loss": 0.4382,
      "step": 11524
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3972303867340088,
      "learning_rate": 1.3720330860226273e-05,
      "loss": 0.6317,
      "step": 11525
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.61903715133667,
      "learning_rate": 1.3719294411295936e-05,
      "loss": 0.6138,
      "step": 11526
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.614773154258728,
      "learning_rate": 1.3718257915995759e-05,
      "loss": 0.6075,
      "step": 11527
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.594019889831543,
      "learning_rate": 1.3717221374338663e-05,
      "loss": 0.5287,
      "step": 11528
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3932188749313354,
      "learning_rate": 1.3716184786337574e-05,
      "loss": 0.5693,
      "step": 11529
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.357923984527588,
      "learning_rate": 1.3715148152005414e-05,
      "loss": 0.5534,
      "step": 11530
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6595793962478638,
      "learning_rate": 1.3714111471355107e-05,
      "loss": 0.4987,
      "step": 11531
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.9761964082717896,
      "learning_rate": 1.371307474439958e-05,
      "loss": 0.5685,
      "step": 11532
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6624101400375366,
      "learning_rate": 1.3712037971151753e-05,
      "loss": 0.5787,
      "step": 11533
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4334650039672852,
      "learning_rate": 1.3711001151624558e-05,
      "loss": 0.6943,
      "step": 11534
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.45167875289917,
      "learning_rate": 1.3709964285830916e-05,
      "loss": 0.4872,
      "step": 11535
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.999113917350769,
      "learning_rate": 1.3708927373783763e-05,
      "loss": 0.5206,
      "step": 11536
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5641733407974243,
      "learning_rate": 1.3707890415496014e-05,
      "loss": 0.5599,
      "step": 11537
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5311291217803955,
      "learning_rate": 1.3706853410980605e-05,
      "loss": 0.5862,
      "step": 11538
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2426092624664307,
      "learning_rate": 1.3705816360250465e-05,
      "loss": 0.5747,
      "step": 11539
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5042697191238403,
      "learning_rate": 1.370477926331852e-05,
      "loss": 0.6441,
      "step": 11540
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4245586395263672,
      "learning_rate": 1.3703742120197703e-05,
      "loss": 0.6309,
      "step": 11541
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.585898756980896,
      "learning_rate": 1.3702704930900943e-05,
      "loss": 0.7277,
      "step": 11542
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5231733322143555,
      "learning_rate": 1.3701667695441168e-05,
      "loss": 0.5539,
      "step": 11543
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3219256401062012,
      "learning_rate": 1.3700630413831315e-05,
      "loss": 0.5713,
      "step": 11544
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5606575012207031,
      "learning_rate": 1.3699593086084313e-05,
      "loss": 0.6441,
      "step": 11545
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4239132404327393,
      "learning_rate": 1.3698555712213093e-05,
      "loss": 0.5753,
      "step": 11546
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.30562162399292,
      "learning_rate": 1.3697518292230597e-05,
      "loss": 0.4741,
      "step": 11547
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6705800294876099,
      "learning_rate": 1.3696480826149745e-05,
      "loss": 0.4496,
      "step": 11548
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2961450815200806,
      "learning_rate": 1.3695443313983486e-05,
      "loss": 0.5825,
      "step": 11549
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.733885407447815,
      "learning_rate": 1.3694405755744745e-05,
      "loss": 0.6835,
      "step": 11550
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.4707800149917603,
      "learning_rate": 1.3693368151446458e-05,
      "loss": 0.4946,
      "step": 11551
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6397783756256104,
      "learning_rate": 1.3692330501101567e-05,
      "loss": 0.4591,
      "step": 11552
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5332798957824707,
      "learning_rate": 1.3691292804723006e-05,
      "loss": 0.4918,
      "step": 11553
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.672200083732605,
      "learning_rate": 1.3690255062323712e-05,
      "loss": 0.6528,
      "step": 11554
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4671589136123657,
      "learning_rate": 1.3689217273916622e-05,
      "loss": 0.505,
      "step": 11555
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5092233419418335,
      "learning_rate": 1.3688179439514674e-05,
      "loss": 0.631,
      "step": 11556
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7740203142166138,
      "learning_rate": 1.3687141559130814e-05,
      "loss": 0.5371,
      "step": 11557
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3616374731063843,
      "learning_rate": 1.3686103632777972e-05,
      "loss": 0.5823,
      "step": 11558
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6949536800384521,
      "learning_rate": 1.3685065660469095e-05,
      "loss": 0.5291,
      "step": 11559
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.8479299545288086,
      "learning_rate": 1.368402764221712e-05,
      "loss": 0.6136,
      "step": 11560
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5349024534225464,
      "learning_rate": 1.3682989578034986e-05,
      "loss": 0.6158,
      "step": 11561
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7477961778640747,
      "learning_rate": 1.3681951467935643e-05,
      "loss": 0.56,
      "step": 11562
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4352465867996216,
      "learning_rate": 1.368091331193203e-05,
      "loss": 0.5464,
      "step": 11563
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3624690771102905,
      "learning_rate": 1.3679875110037087e-05,
      "loss": 0.5309,
      "step": 11564
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.629377841949463,
      "learning_rate": 1.3678836862263759e-05,
      "loss": 0.5222,
      "step": 11565
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4882655143737793,
      "learning_rate": 1.3677798568624992e-05,
      "loss": 0.5956,
      "step": 11566
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5253450870513916,
      "learning_rate": 1.367676022913373e-05,
      "loss": 0.5018,
      "step": 11567
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5809495449066162,
      "learning_rate": 1.3675721843802919e-05,
      "loss": 0.481,
      "step": 11568
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3813350200653076,
      "learning_rate": 1.3674683412645502e-05,
      "loss": 0.577,
      "step": 11569
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.226299524307251,
      "learning_rate": 1.3673644935674431e-05,
      "loss": 0.614,
      "step": 11570
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5215953588485718,
      "learning_rate": 1.3672606412902644e-05,
      "loss": 0.6345,
      "step": 11571
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5196930170059204,
      "learning_rate": 1.36715678443431e-05,
      "loss": 0.5695,
      "step": 11572
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.175148367881775,
      "learning_rate": 1.367052923000874e-05,
      "loss": 0.508,
      "step": 11573
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.8832203149795532,
      "learning_rate": 1.3669490569912513e-05,
      "loss": 0.5547,
      "step": 11574
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.694677710533142,
      "learning_rate": 1.3668451864067373e-05,
      "loss": 0.6116,
      "step": 11575
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5952123403549194,
      "learning_rate": 1.3667413112486263e-05,
      "loss": 0.4594,
      "step": 11576
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4451801776885986,
      "learning_rate": 1.3666374315182138e-05,
      "loss": 0.5474,
      "step": 11577
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4765321016311646,
      "learning_rate": 1.366533547216795e-05,
      "loss": 0.6709,
      "step": 11578
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5381773710250854,
      "learning_rate": 1.366429658345665e-05,
      "loss": 0.6627,
      "step": 11579
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4300601482391357,
      "learning_rate": 1.3663257649061186e-05,
      "loss": 0.5162,
      "step": 11580
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6076087951660156,
      "learning_rate": 1.3662218668994513e-05,
      "loss": 0.5684,
      "step": 11581
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3831835985183716,
      "learning_rate": 1.3661179643269588e-05,
      "loss": 0.4884,
      "step": 11582
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5663840770721436,
      "learning_rate": 1.3660140571899362e-05,
      "loss": 0.6455,
      "step": 11583
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7644912004470825,
      "learning_rate": 1.3659101454896788e-05,
      "loss": 0.5468,
      "step": 11584
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.815451979637146,
      "learning_rate": 1.3658062292274826e-05,
      "loss": 0.5503,
      "step": 11585
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4235742092132568,
      "learning_rate": 1.3657023084046425e-05,
      "loss": 0.5379,
      "step": 11586
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.396209716796875,
      "learning_rate": 1.3655983830224547e-05,
      "loss": 0.5283,
      "step": 11587
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.372294306755066,
      "learning_rate": 1.3654944530822145e-05,
      "loss": 0.5227,
      "step": 11588
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.249925971031189,
      "learning_rate": 1.3653905185852178e-05,
      "loss": 0.4923,
      "step": 11589
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4443455934524536,
      "learning_rate": 1.3652865795327603e-05,
      "loss": 0.6807,
      "step": 11590
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.406597375869751,
      "learning_rate": 1.3651826359261379e-05,
      "loss": 0.5304,
      "step": 11591
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.357837200164795,
      "learning_rate": 1.3650786877666466e-05,
      "loss": 0.6007,
      "step": 11592
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4951280355453491,
      "learning_rate": 1.3649747350555825e-05,
      "loss": 0.4402,
      "step": 11593
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.599998950958252,
      "learning_rate": 1.3648707777942412e-05,
      "loss": 0.5826,
      "step": 11594
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.579448938369751,
      "learning_rate": 1.3647668159839187e-05,
      "loss": 0.5462,
      "step": 11595
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1493133306503296,
      "learning_rate": 1.3646628496259118e-05,
      "loss": 0.5091,
      "step": 11596
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.972394585609436,
      "learning_rate": 1.3645588787215161e-05,
      "loss": 0.6066,
      "step": 11597
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6829779148101807,
      "learning_rate": 1.364454903272028e-05,
      "loss": 0.6132,
      "step": 11598
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5565581321716309,
      "learning_rate": 1.3643509232787438e-05,
      "loss": 0.6142,
      "step": 11599
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3286852836608887,
      "learning_rate": 1.3642469387429598e-05,
      "loss": 0.5926,
      "step": 11600
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3179682493209839,
      "learning_rate": 1.3641429496659728e-05,
      "loss": 0.5459,
      "step": 11601
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4043208360671997,
      "learning_rate": 1.3640389560490787e-05,
      "loss": 0.5356,
      "step": 11602
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.8011722564697266,
      "learning_rate": 1.3639349578935743e-05,
      "loss": 0.5251,
      "step": 11603
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3464534282684326,
      "learning_rate": 1.3638309552007564e-05,
      "loss": 0.5417,
      "step": 11604
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4082467555999756,
      "learning_rate": 1.3637269479719213e-05,
      "loss": 0.4521,
      "step": 11605
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3791056871414185,
      "learning_rate": 1.363622936208366e-05,
      "loss": 0.6152,
      "step": 11606
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5279589891433716,
      "learning_rate": 1.3635189199113867e-05,
      "loss": 0.5285,
      "step": 11607
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.269760012626648,
      "learning_rate": 1.363414899082281e-05,
      "loss": 0.5834,
      "step": 11608
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4096978902816772,
      "learning_rate": 1.363310873722345e-05,
      "loss": 0.5921,
      "step": 11609
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4605058431625366,
      "learning_rate": 1.3632068438328761e-05,
      "loss": 0.5304,
      "step": 11610
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.637215495109558,
      "learning_rate": 1.3631028094151712e-05,
      "loss": 0.688,
      "step": 11611
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5374964475631714,
      "learning_rate": 1.3629987704705272e-05,
      "loss": 0.5584,
      "step": 11612
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7223106622695923,
      "learning_rate": 1.3628947270002412e-05,
      "loss": 0.5289,
      "step": 11613
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.432477593421936,
      "learning_rate": 1.3627906790056107e-05,
      "loss": 0.4877,
      "step": 11614
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4943222999572754,
      "learning_rate": 1.3626866264879324e-05,
      "loss": 0.6633,
      "step": 11615
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.327494740486145,
      "learning_rate": 1.3625825694485037e-05,
      "loss": 0.4524,
      "step": 11616
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.438651204109192,
      "learning_rate": 1.3624785078886224e-05,
      "loss": 0.5024,
      "step": 11617
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3822858333587646,
      "learning_rate": 1.3623744418095854e-05,
      "loss": 0.507,
      "step": 11618
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4021739959716797,
      "learning_rate": 1.3622703712126902e-05,
      "loss": 0.4903,
      "step": 11619
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5042686462402344,
      "learning_rate": 1.3621662960992341e-05,
      "loss": 0.591,
      "step": 11620
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6604487895965576,
      "learning_rate": 1.3620622164705148e-05,
      "loss": 0.6079,
      "step": 11621
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.326846957206726,
      "learning_rate": 1.3619581323278304e-05,
      "loss": 0.6048,
      "step": 11622
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4035837650299072,
      "learning_rate": 1.3618540436724778e-05,
      "loss": 0.539,
      "step": 11623
    },
    {
      "epoch": 1.2,
      "grad_norm": 2.039945125579834,
      "learning_rate": 1.3617499505057552e-05,
      "loss": 0.7124,
      "step": 11624
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7520960569381714,
      "learning_rate": 1.36164585282896e-05,
      "loss": 0.5725,
      "step": 11625
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4647499322891235,
      "learning_rate": 1.3615417506433901e-05,
      "loss": 0.5767,
      "step": 11626
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.895075798034668,
      "learning_rate": 1.3614376439503438e-05,
      "loss": 0.6274,
      "step": 11627
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.2258323431015015,
      "learning_rate": 1.3613335327511186e-05,
      "loss": 0.533,
      "step": 11628
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.472240686416626,
      "learning_rate": 1.3612294170470124e-05,
      "loss": 0.5791,
      "step": 11629
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6489816904067993,
      "learning_rate": 1.3611252968393238e-05,
      "loss": 0.5132,
      "step": 11630
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.2027976512908936,
      "learning_rate": 1.3610211721293506e-05,
      "loss": 0.4678,
      "step": 11631
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3398406505584717,
      "learning_rate": 1.3609170429183908e-05,
      "loss": 0.5828,
      "step": 11632
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.8484776020050049,
      "learning_rate": 1.3608129092077427e-05,
      "loss": 0.6228,
      "step": 11633
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6025727987289429,
      "learning_rate": 1.3607087709987048e-05,
      "loss": 0.6216,
      "step": 11634
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.281003475189209,
      "learning_rate": 1.3606046282925753e-05,
      "loss": 0.6323,
      "step": 11635
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6003824472427368,
      "learning_rate": 1.3605004810906523e-05,
      "loss": 0.5494,
      "step": 11636
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.2962315082550049,
      "learning_rate": 1.3603963293942348e-05,
      "loss": 0.583,
      "step": 11637
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3001865148544312,
      "learning_rate": 1.3602921732046206e-05,
      "loss": 0.5801,
      "step": 11638
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5861060619354248,
      "learning_rate": 1.3601880125231088e-05,
      "loss": 0.6355,
      "step": 11639
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.591296672821045,
      "learning_rate": 1.360083847350998e-05,
      "loss": 0.5085,
      "step": 11640
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.6610066890716553,
      "learning_rate": 1.3599796776895867e-05,
      "loss": 0.5903,
      "step": 11641
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4158776998519897,
      "learning_rate": 1.3598755035401733e-05,
      "loss": 0.5097,
      "step": 11642
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.316112995147705,
      "learning_rate": 1.3597713249040574e-05,
      "loss": 0.5933,
      "step": 11643
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.5426338911056519,
      "learning_rate": 1.3596671417825373e-05,
      "loss": 0.5202,
      "step": 11644
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3108810186386108,
      "learning_rate": 1.3595629541769117e-05,
      "loss": 0.4339,
      "step": 11645
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.729176640510559,
      "learning_rate": 1.3594587620884801e-05,
      "loss": 0.6292,
      "step": 11646
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.7475627660751343,
      "learning_rate": 1.3593545655185407e-05,
      "loss": 0.6448,
      "step": 11647
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3142658472061157,
      "learning_rate": 1.3592503644683936e-05,
      "loss": 0.5743,
      "step": 11648
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3256134986877441,
      "learning_rate": 1.359146158939337e-05,
      "loss": 0.719,
      "step": 11649
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4850177764892578,
      "learning_rate": 1.3590419489326707e-05,
      "loss": 0.4241,
      "step": 11650
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.341120958328247,
      "learning_rate": 1.3589377344496934e-05,
      "loss": 0.7164,
      "step": 11651
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6020890474319458,
      "learning_rate": 1.3588335154917048e-05,
      "loss": 0.559,
      "step": 11652
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.8308944702148438,
      "learning_rate": 1.3587292920600042e-05,
      "loss": 0.5473,
      "step": 11653
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2765189409255981,
      "learning_rate": 1.3586250641558904e-05,
      "loss": 0.4627,
      "step": 11654
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7674421072006226,
      "learning_rate": 1.3585208317806637e-05,
      "loss": 0.6108,
      "step": 11655
    },
    {
      "epoch": 1.21,
      "grad_norm": 2.216545581817627,
      "learning_rate": 1.3584165949356232e-05,
      "loss": 0.6181,
      "step": 11656
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6880155801773071,
      "learning_rate": 1.3583123536220684e-05,
      "loss": 0.5314,
      "step": 11657
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3425559997558594,
      "learning_rate": 1.3582081078412992e-05,
      "loss": 0.5512,
      "step": 11658
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7585047483444214,
      "learning_rate": 1.3581038575946148e-05,
      "loss": 0.5738,
      "step": 11659
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4411814212799072,
      "learning_rate": 1.3579996028833153e-05,
      "loss": 0.6378,
      "step": 11660
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6954988241195679,
      "learning_rate": 1.3578953437087003e-05,
      "loss": 0.5315,
      "step": 11661
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4291422367095947,
      "learning_rate": 1.3577910800720699e-05,
      "loss": 0.5084,
      "step": 11662
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.592502236366272,
      "learning_rate": 1.3576868119747238e-05,
      "loss": 0.6218,
      "step": 11663
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.697911024093628,
      "learning_rate": 1.3575825394179618e-05,
      "loss": 0.513,
      "step": 11664
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2937899827957153,
      "learning_rate": 1.3574782624030843e-05,
      "loss": 0.6203,
      "step": 11665
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3570743799209595,
      "learning_rate": 1.3573739809313908e-05,
      "loss": 0.563,
      "step": 11666
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.360127568244934,
      "learning_rate": 1.3572696950041819e-05,
      "loss": 0.5067,
      "step": 11667
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5637222528457642,
      "learning_rate": 1.3571654046227575e-05,
      "loss": 0.5423,
      "step": 11668
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3264758586883545,
      "learning_rate": 1.3570611097884183e-05,
      "loss": 0.5684,
      "step": 11669
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.297316074371338,
      "learning_rate": 1.356956810502464e-05,
      "loss": 0.5168,
      "step": 11670
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5648021697998047,
      "learning_rate": 1.3568525067661953e-05,
      "loss": 0.5746,
      "step": 11671
    },
    {
      "epoch": 1.21,
      "grad_norm": 2.01241135597229,
      "learning_rate": 1.3567481985809123e-05,
      "loss": 0.6799,
      "step": 11672
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.9395889043807983,
      "learning_rate": 1.3566438859479156e-05,
      "loss": 0.594,
      "step": 11673
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.173011302947998,
      "learning_rate": 1.356539568868506e-05,
      "loss": 0.5485,
      "step": 11674
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4824249744415283,
      "learning_rate": 1.3564352473439832e-05,
      "loss": 0.4894,
      "step": 11675
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6777747869491577,
      "learning_rate": 1.3563309213756488e-05,
      "loss": 0.4324,
      "step": 11676
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.647237777709961,
      "learning_rate": 1.356226590964803e-05,
      "loss": 0.498,
      "step": 11677
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.701441764831543,
      "learning_rate": 1.3561222561127466e-05,
      "loss": 0.5601,
      "step": 11678
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.9296380281448364,
      "learning_rate": 1.3560179168207803e-05,
      "loss": 0.5744,
      "step": 11679
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3695909976959229,
      "learning_rate": 1.355913573090205e-05,
      "loss": 0.5908,
      "step": 11680
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5920169353485107,
      "learning_rate": 1.3558092249223215e-05,
      "loss": 0.6556,
      "step": 11681
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4252911806106567,
      "learning_rate": 1.3557048723184312e-05,
      "loss": 0.5415,
      "step": 11682
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4612315893173218,
      "learning_rate": 1.3556005152798344e-05,
      "loss": 0.4847,
      "step": 11683
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5439872741699219,
      "learning_rate": 1.3554961538078327e-05,
      "loss": 0.6894,
      "step": 11684
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.440175175666809,
      "learning_rate": 1.3553917879037268e-05,
      "loss": 0.5709,
      "step": 11685
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5427969694137573,
      "learning_rate": 1.3552874175688185e-05,
      "loss": 0.6096,
      "step": 11686
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5007140636444092,
      "learning_rate": 1.3551830428044083e-05,
      "loss": 0.5272,
      "step": 11687
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5471493005752563,
      "learning_rate": 1.3550786636117978e-05,
      "loss": 0.6078,
      "step": 11688
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.573964238166809,
      "learning_rate": 1.3549742799922882e-05,
      "loss": 0.6162,
      "step": 11689
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.784518837928772,
      "learning_rate": 1.3548698919471814e-05,
      "loss": 0.6142,
      "step": 11690
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.8329051733016968,
      "learning_rate": 1.3547654994777782e-05,
      "loss": 0.5667,
      "step": 11691
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.37005615234375,
      "learning_rate": 1.3546611025853804e-05,
      "loss": 0.5179,
      "step": 11692
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.659126877784729,
      "learning_rate": 1.3545567012712897e-05,
      "loss": 0.5391,
      "step": 11693
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.436909794807434,
      "learning_rate": 1.354452295536807e-05,
      "loss": 0.6226,
      "step": 11694
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3496726751327515,
      "learning_rate": 1.354347885383235e-05,
      "loss": 0.492,
      "step": 11695
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.466538429260254,
      "learning_rate": 1.3542434708118748e-05,
      "loss": 0.5091,
      "step": 11696
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3470993041992188,
      "learning_rate": 1.3541390518240283e-05,
      "loss": 0.5006,
      "step": 11697
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.814259648323059,
      "learning_rate": 1.3540346284209973e-05,
      "loss": 0.7257,
      "step": 11698
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4842091798782349,
      "learning_rate": 1.3539302006040835e-05,
      "loss": 0.6436,
      "step": 11699
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3048347234725952,
      "learning_rate": 1.3538257683745894e-05,
      "loss": 0.5452,
      "step": 11700
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5746029615402222,
      "learning_rate": 1.3537213317338163e-05,
      "loss": 0.5282,
      "step": 11701
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4456335306167603,
      "learning_rate": 1.3536168906830666e-05,
      "loss": 0.6113,
      "step": 11702
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.8837065696716309,
      "learning_rate": 1.3535124452236428e-05,
      "loss": 0.5106,
      "step": 11703
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.405333399772644,
      "learning_rate": 1.3534079953568464e-05,
      "loss": 0.5433,
      "step": 11704
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4999862909317017,
      "learning_rate": 1.3533035410839799e-05,
      "loss": 0.6183,
      "step": 11705
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5941424369812012,
      "learning_rate": 1.3531990824063455e-05,
      "loss": 0.4979,
      "step": 11706
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3777521848678589,
      "learning_rate": 1.3530946193252454e-05,
      "loss": 0.608,
      "step": 11707
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.8271740674972534,
      "learning_rate": 1.3529901518419826e-05,
      "loss": 0.665,
      "step": 11708
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6156233549118042,
      "learning_rate": 1.3528856799578587e-05,
      "loss": 0.5744,
      "step": 11709
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5590094327926636,
      "learning_rate": 1.3527812036741767e-05,
      "loss": 0.5044,
      "step": 11710
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3843234777450562,
      "learning_rate": 1.3526767229922393e-05,
      "loss": 0.5105,
      "step": 11711
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5394618511199951,
      "learning_rate": 1.3525722379133484e-05,
      "loss": 0.5646,
      "step": 11712
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.595444679260254,
      "learning_rate": 1.3524677484388073e-05,
      "loss": 0.5437,
      "step": 11713
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7425618171691895,
      "learning_rate": 1.3523632545699184e-05,
      "loss": 0.5699,
      "step": 11714
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5819103717803955,
      "learning_rate": 1.3522587563079845e-05,
      "loss": 0.5411,
      "step": 11715
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5896278619766235,
      "learning_rate": 1.3521542536543086e-05,
      "loss": 0.6982,
      "step": 11716
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3197262287139893,
      "learning_rate": 1.3520497466101933e-05,
      "loss": 0.5927,
      "step": 11717
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3470996618270874,
      "learning_rate": 1.3519452351769417e-05,
      "loss": 0.4556,
      "step": 11718
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5370302200317383,
      "learning_rate": 1.3518407193558568e-05,
      "loss": 0.4929,
      "step": 11719
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7161000967025757,
      "learning_rate": 1.3517361991482418e-05,
      "loss": 0.503,
      "step": 11720
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5290825366973877,
      "learning_rate": 1.3516316745553994e-05,
      "loss": 0.5568,
      "step": 11721
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6414440870285034,
      "learning_rate": 1.3515271455786328e-05,
      "loss": 0.5069,
      "step": 11722
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3397021293640137,
      "learning_rate": 1.3514226122192456e-05,
      "loss": 0.562,
      "step": 11723
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3651846647262573,
      "learning_rate": 1.3513180744785406e-05,
      "loss": 0.5576,
      "step": 11724
    },
    {
      "epoch": 1.21,
      "grad_norm": 2.071767568588257,
      "learning_rate": 1.3512135323578212e-05,
      "loss": 0.5559,
      "step": 11725
    },
    {
      "epoch": 1.21,
      "grad_norm": 2.0225725173950195,
      "learning_rate": 1.3511089858583912e-05,
      "loss": 0.6525,
      "step": 11726
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4026390314102173,
      "learning_rate": 1.3510044349815534e-05,
      "loss": 0.5274,
      "step": 11727
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3023971319198608,
      "learning_rate": 1.3508998797286116e-05,
      "loss": 0.5206,
      "step": 11728
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5471996068954468,
      "learning_rate": 1.3507953201008697e-05,
      "loss": 0.6025,
      "step": 11729
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.9109247922897339,
      "learning_rate": 1.3506907560996303e-05,
      "loss": 0.5804,
      "step": 11730
    },
    {
      "epoch": 1.21,
      "grad_norm": 2.854992628097534,
      "learning_rate": 1.3505861877261978e-05,
      "loss": 0.7052,
      "step": 11731
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7699898481369019,
      "learning_rate": 1.350481614981876e-05,
      "loss": 0.763,
      "step": 11732
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.456844449043274,
      "learning_rate": 1.350377037867968e-05,
      "loss": 0.5182,
      "step": 11733
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6035723686218262,
      "learning_rate": 1.3502724563857782e-05,
      "loss": 0.6428,
      "step": 11734
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3364728689193726,
      "learning_rate": 1.3501678705366102e-05,
      "loss": 0.5649,
      "step": 11735
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.60802161693573,
      "learning_rate": 1.3500632803217678e-05,
      "loss": 0.5651,
      "step": 11736
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.4040125608444214,
      "learning_rate": 1.349958685742555e-05,
      "loss": 0.4761,
      "step": 11737
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.505582332611084,
      "learning_rate": 1.3498540868002761e-05,
      "loss": 0.5794,
      "step": 11738
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5909241437911987,
      "learning_rate": 1.3497494834962351e-05,
      "loss": 0.582,
      "step": 11739
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.5363898277282715,
      "learning_rate": 1.3496448758317357e-05,
      "loss": 0.6108,
      "step": 11740
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6303768157958984,
      "learning_rate": 1.3495402638080824e-05,
      "loss": 0.5678,
      "step": 11741
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2663604021072388,
      "learning_rate": 1.3494356474265797e-05,
      "loss": 0.4863,
      "step": 11742
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2030760049819946,
      "learning_rate": 1.3493310266885315e-05,
      "loss": 0.4667,
      "step": 11743
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6529364585876465,
      "learning_rate": 1.3492264015952423e-05,
      "loss": 0.5098,
      "step": 11744
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.7438628673553467,
      "learning_rate": 1.3491217721480166e-05,
      "loss": 0.5579,
      "step": 11745
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.6711386442184448,
      "learning_rate": 1.3490171383481582e-05,
      "loss": 0.5436,
      "step": 11746
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3437321186065674,
      "learning_rate": 1.3489125001969727e-05,
      "loss": 0.484,
      "step": 11747
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6034409999847412,
      "learning_rate": 1.348807857695764e-05,
      "loss": 0.5331,
      "step": 11748
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5310561656951904,
      "learning_rate": 1.3487032108458367e-05,
      "loss": 0.5816,
      "step": 11749
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6310516595840454,
      "learning_rate": 1.3485985596484957e-05,
      "loss": 0.511,
      "step": 11750
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5991239547729492,
      "learning_rate": 1.3484939041050456e-05,
      "loss": 0.5923,
      "step": 11751
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3525876998901367,
      "learning_rate": 1.3483892442167912e-05,
      "loss": 0.5686,
      "step": 11752
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6036913394927979,
      "learning_rate": 1.348284579985037e-05,
      "loss": 0.5657,
      "step": 11753
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8690284490585327,
      "learning_rate": 1.3481799114110885e-05,
      "loss": 0.5045,
      "step": 11754
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.49687922000885,
      "learning_rate": 1.3480752384962505e-05,
      "loss": 0.4261,
      "step": 11755
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.622391700744629,
      "learning_rate": 1.3479705612418276e-05,
      "loss": 0.4765,
      "step": 11756
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5713509321212769,
      "learning_rate": 1.347865879649125e-05,
      "loss": 0.5673,
      "step": 11757
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5346928834915161,
      "learning_rate": 1.3477611937194483e-05,
      "loss": 0.4524,
      "step": 11758
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.604681134223938,
      "learning_rate": 1.3476565034541021e-05,
      "loss": 0.4677,
      "step": 11759
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.556098222732544,
      "learning_rate": 1.347551808854392e-05,
      "loss": 0.6266,
      "step": 11760
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.478628396987915,
      "learning_rate": 1.3474471099216227e-05,
      "loss": 0.5706,
      "step": 11761
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.9562066793441772,
      "learning_rate": 1.3473424066570999e-05,
      "loss": 0.6509,
      "step": 11762
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6721768379211426,
      "learning_rate": 1.3472376990621292e-05,
      "loss": 0.6016,
      "step": 11763
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4257272481918335,
      "learning_rate": 1.3471329871380154e-05,
      "loss": 0.4932,
      "step": 11764
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.449623703956604,
      "learning_rate": 1.3470282708860646e-05,
      "loss": 0.5543,
      "step": 11765
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7685678005218506,
      "learning_rate": 1.346923550307582e-05,
      "loss": 0.5744,
      "step": 11766
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4821873903274536,
      "learning_rate": 1.3468188254038735e-05,
      "loss": 0.517,
      "step": 11767
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7992537021636963,
      "learning_rate": 1.3467140961762443e-05,
      "loss": 0.6446,
      "step": 11768
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.659134864807129,
      "learning_rate": 1.3466093626260004e-05,
      "loss": 0.5879,
      "step": 11769
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5034841299057007,
      "learning_rate": 1.3465046247544476e-05,
      "loss": 0.6721,
      "step": 11770
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4051626920700073,
      "learning_rate": 1.3463998825628917e-05,
      "loss": 0.5443,
      "step": 11771
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3593494892120361,
      "learning_rate": 1.346295136052638e-05,
      "loss": 0.5663,
      "step": 11772
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5226038694381714,
      "learning_rate": 1.346190385224993e-05,
      "loss": 0.6192,
      "step": 11773
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8415557146072388,
      "learning_rate": 1.3460856300812628e-05,
      "loss": 0.5758,
      "step": 11774
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5798490047454834,
      "learning_rate": 1.3459808706227528e-05,
      "loss": 0.5554,
      "step": 11775
    },
    {
      "epoch": 1.22,
      "grad_norm": 2.1368346214294434,
      "learning_rate": 1.3458761068507697e-05,
      "loss": 0.6116,
      "step": 11776
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.2461451292037964,
      "learning_rate": 1.3457713387666194e-05,
      "loss": 0.5715,
      "step": 11777
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8194031715393066,
      "learning_rate": 1.345666566371608e-05,
      "loss": 0.6203,
      "step": 11778
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3016666173934937,
      "learning_rate": 1.3455617896670415e-05,
      "loss": 0.6526,
      "step": 11779
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4171422719955444,
      "learning_rate": 1.3454570086542265e-05,
      "loss": 0.4985,
      "step": 11780
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4498493671417236,
      "learning_rate": 1.3453522233344696e-05,
      "loss": 0.5725,
      "step": 11781
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8596842288970947,
      "learning_rate": 1.3452474337090765e-05,
      "loss": 0.5198,
      "step": 11782
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3449934720993042,
      "learning_rate": 1.3451426397793544e-05,
      "loss": 0.4636,
      "step": 11783
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3767699003219604,
      "learning_rate": 1.3450378415466094e-05,
      "loss": 0.5118,
      "step": 11784
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.316109299659729,
      "learning_rate": 1.344933039012148e-05,
      "loss": 0.5301,
      "step": 11785
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.196569800376892,
      "learning_rate": 1.3448282321772772e-05,
      "loss": 0.4861,
      "step": 11786
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5806339979171753,
      "learning_rate": 1.3447234210433034e-05,
      "loss": 0.555,
      "step": 11787
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.568534016609192,
      "learning_rate": 1.344618605611533e-05,
      "loss": 0.5664,
      "step": 11788
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3226789236068726,
      "learning_rate": 1.3445137858832733e-05,
      "loss": 0.5009,
      "step": 11789
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.53101646900177,
      "learning_rate": 1.3444089618598308e-05,
      "loss": 0.5246,
      "step": 11790
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5302159786224365,
      "learning_rate": 1.3443041335425129e-05,
      "loss": 0.5518,
      "step": 11791
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.614237904548645,
      "learning_rate": 1.3441993009326258e-05,
      "loss": 0.6033,
      "step": 11792
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6245378255844116,
      "learning_rate": 1.3440944640314766e-05,
      "loss": 0.5562,
      "step": 11793
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.591269850730896,
      "learning_rate": 1.3439896228403727e-05,
      "loss": 0.5401,
      "step": 11794
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3788264989852905,
      "learning_rate": 1.3438847773606213e-05,
      "loss": 0.5505,
      "step": 11795
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.2756787538528442,
      "learning_rate": 1.3437799275935293e-05,
      "loss": 0.4519,
      "step": 11796
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3762500286102295,
      "learning_rate": 1.3436750735404038e-05,
      "loss": 0.4518,
      "step": 11797
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.330730676651001,
      "learning_rate": 1.343570215202552e-05,
      "loss": 0.5687,
      "step": 11798
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.407266616821289,
      "learning_rate": 1.3434653525812817e-05,
      "loss": 0.4728,
      "step": 11799
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4551762342453003,
      "learning_rate": 1.3433604856778997e-05,
      "loss": 0.5787,
      "step": 11800
    },
    {
      "epoch": 1.22,
      "grad_norm": 2.125459909439087,
      "learning_rate": 1.3432556144937136e-05,
      "loss": 0.6253,
      "step": 11801
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4773833751678467,
      "learning_rate": 1.3431507390300311e-05,
      "loss": 0.5235,
      "step": 11802
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5843640565872192,
      "learning_rate": 1.3430458592881594e-05,
      "loss": 0.5065,
      "step": 11803
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.285247564315796,
      "learning_rate": 1.3429409752694065e-05,
      "loss": 0.508,
      "step": 11804
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.63517427444458,
      "learning_rate": 1.3428360869750793e-05,
      "loss": 0.4909,
      "step": 11805
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5524730682373047,
      "learning_rate": 1.3427311944064862e-05,
      "loss": 0.5033,
      "step": 11806
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4915834665298462,
      "learning_rate": 1.3426262975649347e-05,
      "loss": 0.5827,
      "step": 11807
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6770306825637817,
      "learning_rate": 1.3425213964517325e-05,
      "loss": 0.5888,
      "step": 11808
    },
    {
      "epoch": 1.22,
      "grad_norm": 2.034597396850586,
      "learning_rate": 1.3424164910681874e-05,
      "loss": 0.6596,
      "step": 11809
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4885159730911255,
      "learning_rate": 1.3423115814156073e-05,
      "loss": 0.6154,
      "step": 11810
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4060771465301514,
      "learning_rate": 1.3422066674953005e-05,
      "loss": 0.5831,
      "step": 11811
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3807896375656128,
      "learning_rate": 1.3421017493085748e-05,
      "loss": 0.5821,
      "step": 11812
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4192085266113281,
      "learning_rate": 1.3419968268567379e-05,
      "loss": 0.4413,
      "step": 11813
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4084060192108154,
      "learning_rate": 1.3418919001410982e-05,
      "loss": 0.5141,
      "step": 11814
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5663563013076782,
      "learning_rate": 1.3417869691629641e-05,
      "loss": 0.6242,
      "step": 11815
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6860697269439697,
      "learning_rate": 1.341682033923643e-05,
      "loss": 0.5667,
      "step": 11816
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5077714920043945,
      "learning_rate": 1.3415770944244444e-05,
      "loss": 0.587,
      "step": 11817
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.834951639175415,
      "learning_rate": 1.3414721506666757e-05,
      "loss": 0.4496,
      "step": 11818
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7104791402816772,
      "learning_rate": 1.3413672026516456e-05,
      "loss": 0.5711,
      "step": 11819
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4499366283416748,
      "learning_rate": 1.3412622503806625e-05,
      "loss": 0.5618,
      "step": 11820
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7046648263931274,
      "learning_rate": 1.3411572938550346e-05,
      "loss": 0.5932,
      "step": 11821
    },
    {
      "epoch": 1.22,
      "grad_norm": 2.141996383666992,
      "learning_rate": 1.3410523330760707e-05,
      "loss": 0.5773,
      "step": 11822
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6973986625671387,
      "learning_rate": 1.3409473680450795e-05,
      "loss": 0.5602,
      "step": 11823
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.500214695930481,
      "learning_rate": 1.3408423987633694e-05,
      "loss": 0.5215,
      "step": 11824
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.198771357536316,
      "learning_rate": 1.3407374252322494e-05,
      "loss": 0.5153,
      "step": 11825
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3407851457595825,
      "learning_rate": 1.340632447453028e-05,
      "loss": 0.5941,
      "step": 11826
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6548126935958862,
      "learning_rate": 1.3405274654270137e-05,
      "loss": 0.5233,
      "step": 11827
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.508718490600586,
      "learning_rate": 1.340422479155516e-05,
      "loss": 0.6818,
      "step": 11828
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4829814434051514,
      "learning_rate": 1.3403174886398432e-05,
      "loss": 0.5542,
      "step": 11829
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4085415601730347,
      "learning_rate": 1.3402124938813049e-05,
      "loss": 0.5173,
      "step": 11830
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3549423217773438,
      "learning_rate": 1.3401074948812094e-05,
      "loss": 0.4839,
      "step": 11831
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1138980388641357,
      "learning_rate": 1.340002491640866e-05,
      "loss": 0.4975,
      "step": 11832
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4003478288650513,
      "learning_rate": 1.3398974841615843e-05,
      "loss": 0.5186,
      "step": 11833
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.667628288269043,
      "learning_rate": 1.3397924724446729e-05,
      "loss": 0.4975,
      "step": 11834
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5215802192687988,
      "learning_rate": 1.3396874564914412e-05,
      "loss": 0.5801,
      "step": 11835
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4803959131240845,
      "learning_rate": 1.3395824363031986e-05,
      "loss": 0.6052,
      "step": 11836
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4225010871887207,
      "learning_rate": 1.339477411881254e-05,
      "loss": 0.5019,
      "step": 11837
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.7579716444015503,
      "learning_rate": 1.3393723832269176e-05,
      "loss": 0.5906,
      "step": 11838
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.696649432182312,
      "learning_rate": 1.3392673503414982e-05,
      "loss": 0.5211,
      "step": 11839
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4389903545379639,
      "learning_rate": 1.339162313226305e-05,
      "loss": 0.6519,
      "step": 11840
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.6464056968688965,
      "learning_rate": 1.3390572718826484e-05,
      "loss": 0.5625,
      "step": 11841
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3680459260940552,
      "learning_rate": 1.3389522263118372e-05,
      "loss": 0.5375,
      "step": 11842
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.8467957973480225,
      "learning_rate": 1.3388471765151815e-05,
      "loss": 0.4584,
      "step": 11843
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.359262228012085,
      "learning_rate": 1.338742122493991e-05,
      "loss": 0.6372,
      "step": 11844
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7209454774856567,
      "learning_rate": 1.338637064249575e-05,
      "loss": 0.6405,
      "step": 11845
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.9122686386108398,
      "learning_rate": 1.3385320017832439e-05,
      "loss": 0.5136,
      "step": 11846
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.320871353149414,
      "learning_rate": 1.338426935096307e-05,
      "loss": 0.5508,
      "step": 11847
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3831429481506348,
      "learning_rate": 1.3383218641900748e-05,
      "loss": 0.5267,
      "step": 11848
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7700296640396118,
      "learning_rate": 1.3382167890658568e-05,
      "loss": 0.6083,
      "step": 11849
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3081666231155396,
      "learning_rate": 1.338111709724963e-05,
      "loss": 0.6022,
      "step": 11850
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5020774602890015,
      "learning_rate": 1.3380066261687039e-05,
      "loss": 0.6651,
      "step": 11851
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5057092905044556,
      "learning_rate": 1.337901538398389e-05,
      "loss": 0.5277,
      "step": 11852
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6047722101211548,
      "learning_rate": 1.337796446415329e-05,
      "loss": 0.5382,
      "step": 11853
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8376463651657104,
      "learning_rate": 1.337691350220834e-05,
      "loss": 0.5806,
      "step": 11854
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6008530855178833,
      "learning_rate": 1.337586249816214e-05,
      "loss": 0.5574,
      "step": 11855
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.574084997177124,
      "learning_rate": 1.3374811452027795e-05,
      "loss": 0.6132,
      "step": 11856
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.466353178024292,
      "learning_rate": 1.3373760363818411e-05,
      "loss": 0.6261,
      "step": 11857
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7382149696350098,
      "learning_rate": 1.3372709233547088e-05,
      "loss": 0.6436,
      "step": 11858
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3638758659362793,
      "learning_rate": 1.3371658061226936e-05,
      "loss": 0.4989,
      "step": 11859
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6230829954147339,
      "learning_rate": 1.3370606846871056e-05,
      "loss": 0.4442,
      "step": 11860
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.768153190612793,
      "learning_rate": 1.3369555590492551e-05,
      "loss": 0.7064,
      "step": 11861
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6036978960037231,
      "learning_rate": 1.336850429210454e-05,
      "loss": 0.5874,
      "step": 11862
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7180145978927612,
      "learning_rate": 1.3367452951720117e-05,
      "loss": 0.523,
      "step": 11863
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6276943683624268,
      "learning_rate": 1.3366401569352392e-05,
      "loss": 0.6188,
      "step": 11864
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6909669637680054,
      "learning_rate": 1.3365350145014479e-05,
      "loss": 0.6335,
      "step": 11865
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.686052918434143,
      "learning_rate": 1.336429867871948e-05,
      "loss": 0.5665,
      "step": 11866
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5085300207138062,
      "learning_rate": 1.3363247170480508e-05,
      "loss": 0.5389,
      "step": 11867
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.463722586631775,
      "learning_rate": 1.3362195620310673e-05,
      "loss": 0.5425,
      "step": 11868
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5616679191589355,
      "learning_rate": 1.3361144028223079e-05,
      "loss": 0.5174,
      "step": 11869
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5103079080581665,
      "learning_rate": 1.3360092394230844e-05,
      "loss": 0.474,
      "step": 11870
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.340976357460022,
      "learning_rate": 1.3359040718347073e-05,
      "loss": 0.562,
      "step": 11871
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.308333158493042,
      "learning_rate": 1.3357989000584886e-05,
      "loss": 0.5195,
      "step": 11872
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.515909194946289,
      "learning_rate": 1.3356937240957384e-05,
      "loss": 0.562,
      "step": 11873
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5868721008300781,
      "learning_rate": 1.3355885439477687e-05,
      "loss": 0.6889,
      "step": 11874
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5716915130615234,
      "learning_rate": 1.335483359615891e-05,
      "loss": 0.6148,
      "step": 11875
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4371894598007202,
      "learning_rate": 1.3353781711014158e-05,
      "loss": 0.4543,
      "step": 11876
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.674864649772644,
      "learning_rate": 1.3352729784056554e-05,
      "loss": 0.5116,
      "step": 11877
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4411754608154297,
      "learning_rate": 1.3351677815299209e-05,
      "loss": 0.6244,
      "step": 11878
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4688528776168823,
      "learning_rate": 1.3350625804755238e-05,
      "loss": 0.506,
      "step": 11879
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3832951784133911,
      "learning_rate": 1.3349573752437759e-05,
      "loss": 0.5865,
      "step": 11880
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3407421112060547,
      "learning_rate": 1.3348521658359885e-05,
      "loss": 0.6401,
      "step": 11881
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.714666724205017,
      "learning_rate": 1.3347469522534734e-05,
      "loss": 0.5562,
      "step": 11882
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5388233661651611,
      "learning_rate": 1.3346417344975426e-05,
      "loss": 0.5103,
      "step": 11883
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7269186973571777,
      "learning_rate": 1.3345365125695078e-05,
      "loss": 0.4643,
      "step": 11884
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3884154558181763,
      "learning_rate": 1.3344312864706806e-05,
      "loss": 0.4465,
      "step": 11885
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4857386350631714,
      "learning_rate": 1.334326056202373e-05,
      "loss": 0.5595,
      "step": 11886
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2834826707839966,
      "learning_rate": 1.3342208217658967e-05,
      "loss": 0.6466,
      "step": 11887
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4615658521652222,
      "learning_rate": 1.3341155831625642e-05,
      "loss": 0.5228,
      "step": 11888
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7236053943634033,
      "learning_rate": 1.3340103403936874e-05,
      "loss": 0.6688,
      "step": 11889
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.463823676109314,
      "learning_rate": 1.3339050934605782e-05,
      "loss": 0.4598,
      "step": 11890
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8354281187057495,
      "learning_rate": 1.333799842364549e-05,
      "loss": 0.552,
      "step": 11891
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3262686729431152,
      "learning_rate": 1.3336945871069117e-05,
      "loss": 0.5347,
      "step": 11892
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6021299362182617,
      "learning_rate": 1.3335893276889791e-05,
      "loss": 0.5654,
      "step": 11893
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6448951959609985,
      "learning_rate": 1.3334840641120628e-05,
      "loss": 0.5577,
      "step": 11894
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8617923259735107,
      "learning_rate": 1.3333787963774755e-05,
      "loss": 0.5531,
      "step": 11895
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4096359014511108,
      "learning_rate": 1.33327352448653e-05,
      "loss": 0.5371,
      "step": 11896
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6647121906280518,
      "learning_rate": 1.333168248440538e-05,
      "loss": 0.5691,
      "step": 11897
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.616963505744934,
      "learning_rate": 1.3330629682408125e-05,
      "loss": 0.5126,
      "step": 11898
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5501461029052734,
      "learning_rate": 1.332957683888666e-05,
      "loss": 0.5458,
      "step": 11899
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8719979524612427,
      "learning_rate": 1.332852395385411e-05,
      "loss": 0.7513,
      "step": 11900
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6264827251434326,
      "learning_rate": 1.3327471027323604e-05,
      "loss": 0.6012,
      "step": 11901
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7572611570358276,
      "learning_rate": 1.3326418059308265e-05,
      "loss": 0.5505,
      "step": 11902
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5352643728256226,
      "learning_rate": 1.3325365049821226e-05,
      "loss": 0.5581,
      "step": 11903
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7177326679229736,
      "learning_rate": 1.3324311998875611e-05,
      "loss": 0.5745,
      "step": 11904
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5936460494995117,
      "learning_rate": 1.3323258906484553e-05,
      "loss": 0.534,
      "step": 11905
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8908051252365112,
      "learning_rate": 1.3322205772661179e-05,
      "loss": 0.6074,
      "step": 11906
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.631467580795288,
      "learning_rate": 1.3321152597418617e-05,
      "loss": 0.5486,
      "step": 11907
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5388661623001099,
      "learning_rate": 1.3320099380769997e-05,
      "loss": 0.6887,
      "step": 11908
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3134407997131348,
      "learning_rate": 1.3319046122728457e-05,
      "loss": 0.64,
      "step": 11909
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5421117544174194,
      "learning_rate": 1.3317992823307117e-05,
      "loss": 0.5176,
      "step": 11910
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5331389904022217,
      "learning_rate": 1.3316939482519123e-05,
      "loss": 0.4496,
      "step": 11911
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.271242618560791,
      "learning_rate": 1.3315886100377591e-05,
      "loss": 0.4363,
      "step": 11912
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2800143957138062,
      "learning_rate": 1.3314832676895666e-05,
      "loss": 0.5053,
      "step": 11913
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.348150610923767,
      "learning_rate": 1.331377921208648e-05,
      "loss": 0.556,
      "step": 11914
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.63484787940979,
      "learning_rate": 1.3312725705963163e-05,
      "loss": 0.5477,
      "step": 11915
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6074590682983398,
      "learning_rate": 1.3311672158538848e-05,
      "loss": 0.6622,
      "step": 11916
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3988614082336426,
      "learning_rate": 1.3310618569826679e-05,
      "loss": 0.5498,
      "step": 11917
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5859211683273315,
      "learning_rate": 1.3309564939839783e-05,
      "loss": 0.4694,
      "step": 11918
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5071790218353271,
      "learning_rate": 1.33085112685913e-05,
      "loss": 0.5394,
      "step": 11919
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3012871742248535,
      "learning_rate": 1.3307457556094364e-05,
      "loss": 0.451,
      "step": 11920
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6286935806274414,
      "learning_rate": 1.3306403802362112e-05,
      "loss": 0.4571,
      "step": 11921
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3944587707519531,
      "learning_rate": 1.3305350007407684e-05,
      "loss": 0.6274,
      "step": 11922
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.406996250152588,
      "learning_rate": 1.3304296171244217e-05,
      "loss": 0.5622,
      "step": 11923
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5514737367630005,
      "learning_rate": 1.330324229388485e-05,
      "loss": 0.4864,
      "step": 11924
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8082256317138672,
      "learning_rate": 1.3302188375342722e-05,
      "loss": 0.642,
      "step": 11925
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7994461059570312,
      "learning_rate": 1.330113441563097e-05,
      "loss": 0.5906,
      "step": 11926
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2409452199935913,
      "learning_rate": 1.3300080414762737e-05,
      "loss": 0.4894,
      "step": 11927
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3655922412872314,
      "learning_rate": 1.3299026372751162e-05,
      "loss": 0.5463,
      "step": 11928
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6097767353057861,
      "learning_rate": 1.329797228960939e-05,
      "loss": 0.5125,
      "step": 11929
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.7428290843963623,
      "learning_rate": 1.3296918165350556e-05,
      "loss": 0.4477,
      "step": 11930
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4629069566726685,
      "learning_rate": 1.3295863999987809e-05,
      "loss": 0.498,
      "step": 11931
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.283477783203125,
      "learning_rate": 1.3294809793534287e-05,
      "loss": 0.5829,
      "step": 11932
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4269721508026123,
      "learning_rate": 1.3293755546003133e-05,
      "loss": 0.5221,
      "step": 11933
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6683250665664673,
      "learning_rate": 1.3292701257407492e-05,
      "loss": 0.6475,
      "step": 11934
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.538465976715088,
      "learning_rate": 1.3291646927760512e-05,
      "loss": 0.4974,
      "step": 11935
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5764633417129517,
      "learning_rate": 1.3290592557075332e-05,
      "loss": 0.6038,
      "step": 11936
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4697133302688599,
      "learning_rate": 1.3289538145365102e-05,
      "loss": 0.6083,
      "step": 11937
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.6344071626663208,
      "learning_rate": 1.3288483692642961e-05,
      "loss": 0.5576,
      "step": 11938
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5647974014282227,
      "learning_rate": 1.3287429198922061e-05,
      "loss": 0.5526,
      "step": 11939
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5832183361053467,
      "learning_rate": 1.328637466421555e-05,
      "loss": 0.5659,
      "step": 11940
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5223629474639893,
      "learning_rate": 1.328532008853657e-05,
      "loss": 0.5966,
      "step": 11941
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4785122871398926,
      "learning_rate": 1.3284265471898272e-05,
      "loss": 0.5379,
      "step": 11942
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5531842708587646,
      "learning_rate": 1.3283210814313805e-05,
      "loss": 0.55,
      "step": 11943
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.2105093002319336,
      "learning_rate": 1.3282156115796315e-05,
      "loss": 0.5822,
      "step": 11944
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5762569904327393,
      "learning_rate": 1.3281101376358953e-05,
      "loss": 0.4981,
      "step": 11945
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.452857494354248,
      "learning_rate": 1.3280046596014868e-05,
      "loss": 0.5651,
      "step": 11946
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5626442432403564,
      "learning_rate": 1.327899177477721e-05,
      "loss": 0.4843,
      "step": 11947
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5773741006851196,
      "learning_rate": 1.3277936912659134e-05,
      "loss": 0.5813,
      "step": 11948
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6853525638580322,
      "learning_rate": 1.3276882009673786e-05,
      "loss": 0.5753,
      "step": 11949
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5352704524993896,
      "learning_rate": 1.3275827065834321e-05,
      "loss": 0.5708,
      "step": 11950
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7733510732650757,
      "learning_rate": 1.327477208115389e-05,
      "loss": 0.4816,
      "step": 11951
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.3147449493408203,
      "learning_rate": 1.3273717055645646e-05,
      "loss": 0.5151,
      "step": 11952
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5658282041549683,
      "learning_rate": 1.3272661989322743e-05,
      "loss": 0.5324,
      "step": 11953
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.3287482261657715,
      "learning_rate": 1.3271606882198336e-05,
      "loss": 0.4254,
      "step": 11954
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7170121669769287,
      "learning_rate": 1.3270551734285575e-05,
      "loss": 0.5042,
      "step": 11955
    },
    {
      "epoch": 1.24,
      "grad_norm": 2.003892421722412,
      "learning_rate": 1.3269496545597623e-05,
      "loss": 0.6398,
      "step": 11956
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.2149808406829834,
      "learning_rate": 1.3268441316147625e-05,
      "loss": 0.4997,
      "step": 11957
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4744772911071777,
      "learning_rate": 1.3267386045948745e-05,
      "loss": 0.5287,
      "step": 11958
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7446644306182861,
      "learning_rate": 1.3266330735014137e-05,
      "loss": 0.5423,
      "step": 11959
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6043208837509155,
      "learning_rate": 1.3265275383356954e-05,
      "loss": 0.6058,
      "step": 11960
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6503134965896606,
      "learning_rate": 1.3264219990990363e-05,
      "loss": 0.6659,
      "step": 11961
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.223563551902771,
      "learning_rate": 1.3263164557927513e-05,
      "loss": 0.4256,
      "step": 11962
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5689096450805664,
      "learning_rate": 1.3262109084181567e-05,
      "loss": 0.5304,
      "step": 11963
    },
    {
      "epoch": 1.24,
      "grad_norm": 2.0042521953582764,
      "learning_rate": 1.3261053569765681e-05,
      "loss": 0.6522,
      "step": 11964
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5479230880737305,
      "learning_rate": 1.3259998014693016e-05,
      "loss": 0.6146,
      "step": 11965
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5225226879119873,
      "learning_rate": 1.3258942418976735e-05,
      "loss": 0.5588,
      "step": 11966
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5033869743347168,
      "learning_rate": 1.3257886782629992e-05,
      "loss": 0.5689,
      "step": 11967
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5802308320999146,
      "learning_rate": 1.3256831105665952e-05,
      "loss": 0.4816,
      "step": 11968
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5560343265533447,
      "learning_rate": 1.325577538809778e-05,
      "loss": 0.592,
      "step": 11969
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5227127075195312,
      "learning_rate": 1.325471962993863e-05,
      "loss": 0.5173,
      "step": 11970
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4100455045700073,
      "learning_rate": 1.3253663831201673e-05,
      "loss": 0.5868,
      "step": 11971
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5501511096954346,
      "learning_rate": 1.3252607991900064e-05,
      "loss": 0.5352,
      "step": 11972
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4314090013504028,
      "learning_rate": 1.325155211204697e-05,
      "loss": 0.5359,
      "step": 11973
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5731300115585327,
      "learning_rate": 1.325049619165556e-05,
      "loss": 0.5665,
      "step": 11974
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.435850977897644,
      "learning_rate": 1.324944023073899e-05,
      "loss": 0.6303,
      "step": 11975
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5411064624786377,
      "learning_rate": 1.3248384229310431e-05,
      "loss": 0.5446,
      "step": 11976
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.87300443649292,
      "learning_rate": 1.3247328187383047e-05,
      "loss": 0.5603,
      "step": 11977
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5102903842926025,
      "learning_rate": 1.3246272104969999e-05,
      "loss": 0.5672,
      "step": 11978
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5001239776611328,
      "learning_rate": 1.3245215982084462e-05,
      "loss": 0.5653,
      "step": 11979
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.592137098312378,
      "learning_rate": 1.3244159818739601e-05,
      "loss": 0.5449,
      "step": 11980
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7341976165771484,
      "learning_rate": 1.3243103614948578e-05,
      "loss": 0.4395,
      "step": 11981
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.639109492301941,
      "learning_rate": 1.3242047370724569e-05,
      "loss": 0.4822,
      "step": 11982
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5069637298583984,
      "learning_rate": 1.3240991086080734e-05,
      "loss": 0.6854,
      "step": 11983
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6961829662322998,
      "learning_rate": 1.3239934761030251e-05,
      "loss": 0.5664,
      "step": 11984
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5782757997512817,
      "learning_rate": 1.3238878395586283e-05,
      "loss": 0.4683,
      "step": 11985
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5311802625656128,
      "learning_rate": 1.3237821989762002e-05,
      "loss": 0.4703,
      "step": 11986
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4782986640930176,
      "learning_rate": 1.323676554357058e-05,
      "loss": 0.5047,
      "step": 11987
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.2576181888580322,
      "learning_rate": 1.3235709057025186e-05,
      "loss": 0.4953,
      "step": 11988
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6736599206924438,
      "learning_rate": 1.3234652530138993e-05,
      "loss": 0.5722,
      "step": 11989
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1732251644134521,
      "learning_rate": 1.3233595962925174e-05,
      "loss": 0.4907,
      "step": 11990
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5172957181930542,
      "learning_rate": 1.3232539355396899e-05,
      "loss": 0.5654,
      "step": 11991
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4485807418823242,
      "learning_rate": 1.3231482707567344e-05,
      "loss": 0.6594,
      "step": 11992
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4479700326919556,
      "learning_rate": 1.323042601944968e-05,
      "loss": 0.5544,
      "step": 11993
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7235608100891113,
      "learning_rate": 1.3229369291057081e-05,
      "loss": 0.6294,
      "step": 11994
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5420485734939575,
      "learning_rate": 1.3228312522402726e-05,
      "loss": 0.6004,
      "step": 11995
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8572967052459717,
      "learning_rate": 1.3227255713499787e-05,
      "loss": 0.6088,
      "step": 11996
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.461823582649231,
      "learning_rate": 1.3226198864361439e-05,
      "loss": 0.605,
      "step": 11997
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.365340232849121,
      "learning_rate": 1.3225141975000857e-05,
      "loss": 0.5359,
      "step": 11998
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.2766658067703247,
      "learning_rate": 1.322408504543122e-05,
      "loss": 0.4935,
      "step": 11999
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4352201223373413,
      "learning_rate": 1.3223028075665707e-05,
      "loss": 0.5102,
      "step": 12000
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.3817229270935059,
      "learning_rate": 1.3221971065717492e-05,
      "loss": 0.6302,
      "step": 12001
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5155071020126343,
      "learning_rate": 1.3220914015599756e-05,
      "loss": 0.5741,
      "step": 12002
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5853495597839355,
      "learning_rate": 1.3219856925325671e-05,
      "loss": 0.5845,
      "step": 12003
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.9145500659942627,
      "learning_rate": 1.3218799794908424e-05,
      "loss": 0.5651,
      "step": 12004
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4256378412246704,
      "learning_rate": 1.3217742624361194e-05,
      "loss": 0.5375,
      "step": 12005
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5613456964492798,
      "learning_rate": 1.3216685413697157e-05,
      "loss": 0.7253,
      "step": 12006
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5763472318649292,
      "learning_rate": 1.3215628162929495e-05,
      "loss": 0.6153,
      "step": 12007
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.3178482055664062,
      "learning_rate": 1.321457087207139e-05,
      "loss": 0.5565,
      "step": 12008
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.3085132837295532,
      "learning_rate": 1.3213513541136025e-05,
      "loss": 0.4341,
      "step": 12009
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4536324739456177,
      "learning_rate": 1.321245617013658e-05,
      "loss": 0.4886,
      "step": 12010
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6104580163955688,
      "learning_rate": 1.3211398759086236e-05,
      "loss": 0.5542,
      "step": 12011
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.361204981803894,
      "learning_rate": 1.321034130799818e-05,
      "loss": 0.5433,
      "step": 12012
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.9228370189666748,
      "learning_rate": 1.3209283816885594e-05,
      "loss": 0.5602,
      "step": 12013
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5970854759216309,
      "learning_rate": 1.3208226285761663e-05,
      "loss": 0.5532,
      "step": 12014
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6421502828598022,
      "learning_rate": 1.3207168714639572e-05,
      "loss": 0.5476,
      "step": 12015
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6271142959594727,
      "learning_rate": 1.3206111103532499e-05,
      "loss": 0.5589,
      "step": 12016
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6641358137130737,
      "learning_rate": 1.3205053452453638e-05,
      "loss": 0.5525,
      "step": 12017
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5157655477523804,
      "learning_rate": 1.3203995761416176e-05,
      "loss": 0.4509,
      "step": 12018
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8867278099060059,
      "learning_rate": 1.3202938030433292e-05,
      "loss": 0.475,
      "step": 12019
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6091312170028687,
      "learning_rate": 1.3201880259518181e-05,
      "loss": 0.5438,
      "step": 12020
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5857845544815063,
      "learning_rate": 1.3200822448684026e-05,
      "loss": 0.5224,
      "step": 12021
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4628397226333618,
      "learning_rate": 1.3199764597944014e-05,
      "loss": 0.5676,
      "step": 12022
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5664669275283813,
      "learning_rate": 1.3198706707311337e-05,
      "loss": 0.5419,
      "step": 12023
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6382596492767334,
      "learning_rate": 1.3197648776799183e-05,
      "loss": 0.5078,
      "step": 12024
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.7863245010375977,
      "learning_rate": 1.3196590806420742e-05,
      "loss": 0.5621,
      "step": 12025
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.423893690109253,
      "learning_rate": 1.3195532796189205e-05,
      "loss": 0.5285,
      "step": 12026
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6057219505310059,
      "learning_rate": 1.3194474746117758e-05,
      "loss": 0.5165,
      "step": 12027
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.9869725704193115,
      "learning_rate": 1.3193416656219598e-05,
      "loss": 0.5334,
      "step": 12028
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6410925388336182,
      "learning_rate": 1.3192358526507914e-05,
      "loss": 0.5915,
      "step": 12029
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.8846371173858643,
      "learning_rate": 1.3191300356995895e-05,
      "loss": 0.6296,
      "step": 12030
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.899013876914978,
      "learning_rate": 1.319024214769674e-05,
      "loss": 0.6013,
      "step": 12031
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6331573724746704,
      "learning_rate": 1.3189183898623634e-05,
      "loss": 0.6017,
      "step": 12032
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6744287014007568,
      "learning_rate": 1.318812560978978e-05,
      "loss": 0.6334,
      "step": 12033
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.3528687953948975,
      "learning_rate": 1.3187067281208368e-05,
      "loss": 0.5231,
      "step": 12034
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6822739839553833,
      "learning_rate": 1.3186008912892587e-05,
      "loss": 0.5649,
      "step": 12035
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.4763303995132446,
      "learning_rate": 1.3184950504855643e-05,
      "loss": 0.6496,
      "step": 12036
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.717268466949463,
      "learning_rate": 1.3183892057110721e-05,
      "loss": 0.5212,
      "step": 12037
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.8178075551986694,
      "learning_rate": 1.3182833569671025e-05,
      "loss": 0.5981,
      "step": 12038
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2882004976272583,
      "learning_rate": 1.3181775042549748e-05,
      "loss": 0.503,
      "step": 12039
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6052746772766113,
      "learning_rate": 1.3180716475760085e-05,
      "loss": 0.4988,
      "step": 12040
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6213165521621704,
      "learning_rate": 1.3179657869315237e-05,
      "loss": 0.7326,
      "step": 12041
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5270256996154785,
      "learning_rate": 1.3178599223228401e-05,
      "loss": 0.426,
      "step": 12042
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4081838130950928,
      "learning_rate": 1.3177540537512775e-05,
      "loss": 0.6261,
      "step": 12043
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.701384425163269,
      "learning_rate": 1.3176481812181556e-05,
      "loss": 0.4899,
      "step": 12044
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5698124170303345,
      "learning_rate": 1.317542304724795e-05,
      "loss": 0.471,
      "step": 12045
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.476596713066101,
      "learning_rate": 1.317436424272515e-05,
      "loss": 0.5411,
      "step": 12046
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5979548692703247,
      "learning_rate": 1.3173305398626363e-05,
      "loss": 0.6267,
      "step": 12047
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3823338747024536,
      "learning_rate": 1.3172246514964783e-05,
      "loss": 0.5104,
      "step": 12048
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4514329433441162,
      "learning_rate": 1.3171187591753615e-05,
      "loss": 0.6159,
      "step": 12049
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5709110498428345,
      "learning_rate": 1.3170128629006066e-05,
      "loss": 0.548,
      "step": 12050
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2724056243896484,
      "learning_rate": 1.3169069626735328e-05,
      "loss": 0.4983,
      "step": 12051
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3058204650878906,
      "learning_rate": 1.3168010584954613e-05,
      "loss": 0.4486,
      "step": 12052
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3217377662658691,
      "learning_rate": 1.316695150367712e-05,
      "loss": 0.5686,
      "step": 12053
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4972553253173828,
      "learning_rate": 1.3165892382916052e-05,
      "loss": 0.4966,
      "step": 12054
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3466987609863281,
      "learning_rate": 1.3164833222684618e-05,
      "loss": 0.5353,
      "step": 12055
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5717531442642212,
      "learning_rate": 1.316377402299602e-05,
      "loss": 0.6533,
      "step": 12056
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7128173112869263,
      "learning_rate": 1.3162714783863461e-05,
      "loss": 0.5592,
      "step": 12057
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4830716848373413,
      "learning_rate": 1.3161655505300154e-05,
      "loss": 0.6062,
      "step": 12058
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1724998950958252,
      "learning_rate": 1.3160596187319298e-05,
      "loss": 0.5533,
      "step": 12059
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4694324731826782,
      "learning_rate": 1.3159536829934105e-05,
      "loss": 0.6963,
      "step": 12060
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5527513027191162,
      "learning_rate": 1.3158477433157779e-05,
      "loss": 0.5064,
      "step": 12061
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.392101526260376,
      "learning_rate": 1.3157417997003529e-05,
      "loss": 0.5326,
      "step": 12062
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6123825311660767,
      "learning_rate": 1.3156358521484567e-05,
      "loss": 0.587,
      "step": 12063
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.9292395114898682,
      "learning_rate": 1.3155299006614096e-05,
      "loss": 0.5181,
      "step": 12064
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3747261762619019,
      "learning_rate": 1.3154239452405329e-05,
      "loss": 0.564,
      "step": 12065
    },
    {
      "epoch": 1.25,
      "grad_norm": 2.2787253856658936,
      "learning_rate": 1.3153179858871472e-05,
      "loss": 0.4528,
      "step": 12066
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6007758378982544,
      "learning_rate": 1.315212022602574e-05,
      "loss": 0.582,
      "step": 12067
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4811114072799683,
      "learning_rate": 1.3151060553881343e-05,
      "loss": 0.5927,
      "step": 12068
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6776728630065918,
      "learning_rate": 1.315000084245149e-05,
      "loss": 0.5311,
      "step": 12069
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.8698933124542236,
      "learning_rate": 1.3148941091749393e-05,
      "loss": 0.5352,
      "step": 12070
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.336511254310608,
      "learning_rate": 1.314788130178827e-05,
      "loss": 0.5838,
      "step": 12071
    },
    {
      "epoch": 1.25,
      "grad_norm": 2.0658881664276123,
      "learning_rate": 1.3146821472581326e-05,
      "loss": 0.4599,
      "step": 12072
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5377769470214844,
      "learning_rate": 1.3145761604141778e-05,
      "loss": 0.5805,
      "step": 12073
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4928525686264038,
      "learning_rate": 1.314470169648284e-05,
      "loss": 0.5052,
      "step": 12074
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5057748556137085,
      "learning_rate": 1.3143641749617724e-05,
      "loss": 0.5524,
      "step": 12075
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3842123746871948,
      "learning_rate": 1.3142581763559648e-05,
      "loss": 0.5617,
      "step": 12076
    },
    {
      "epoch": 1.25,
      "grad_norm": 2.150787115097046,
      "learning_rate": 1.3141521738321825e-05,
      "loss": 0.5058,
      "step": 12077
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.481675386428833,
      "learning_rate": 1.3140461673917472e-05,
      "loss": 0.6435,
      "step": 12078
    },
    {
      "epoch": 1.25,
      "grad_norm": 2.0761501789093018,
      "learning_rate": 1.3139401570359804e-05,
      "loss": 0.5556,
      "step": 12079
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.470861792564392,
      "learning_rate": 1.3138341427662037e-05,
      "loss": 0.6387,
      "step": 12080
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7978323698043823,
      "learning_rate": 1.3137281245837393e-05,
      "loss": 0.4662,
      "step": 12081
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.520319938659668,
      "learning_rate": 1.3136221024899083e-05,
      "loss": 0.6984,
      "step": 12082
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4434807300567627,
      "learning_rate": 1.313516076486033e-05,
      "loss": 0.5469,
      "step": 12083
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4240509271621704,
      "learning_rate": 1.313410046573435e-05,
      "loss": 0.6664,
      "step": 12084
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3060858249664307,
      "learning_rate": 1.3133040127534362e-05,
      "loss": 0.3925,
      "step": 12085
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3374773263931274,
      "learning_rate": 1.3131979750273592e-05,
      "loss": 0.5508,
      "step": 12086
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.818704605102539,
      "learning_rate": 1.3130919333965249e-05,
      "loss": 0.6097,
      "step": 12087
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.429192304611206,
      "learning_rate": 1.3129858878622561e-05,
      "loss": 0.4978,
      "step": 12088
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2469806671142578,
      "learning_rate": 1.312879838425875e-05,
      "loss": 0.4837,
      "step": 12089
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.968144178390503,
      "learning_rate": 1.3127737850887033e-05,
      "loss": 0.6375,
      "step": 12090
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7028130292892456,
      "learning_rate": 1.3126677278520636e-05,
      "loss": 0.5817,
      "step": 12091
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5491570234298706,
      "learning_rate": 1.3125616667172778e-05,
      "loss": 0.6265,
      "step": 12092
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6657960414886475,
      "learning_rate": 1.3124556016856685e-05,
      "loss": 0.6317,
      "step": 12093
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.286767840385437,
      "learning_rate": 1.3123495327585578e-05,
      "loss": 0.6679,
      "step": 12094
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5244652032852173,
      "learning_rate": 1.3122434599372682e-05,
      "loss": 0.525,
      "step": 12095
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4296574592590332,
      "learning_rate": 1.3121373832231223e-05,
      "loss": 0.5211,
      "step": 12096
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5102518796920776,
      "learning_rate": 1.3120313026174427e-05,
      "loss": 0.4583,
      "step": 12097
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3853789567947388,
      "learning_rate": 1.3119252181215513e-05,
      "loss": 0.4589,
      "step": 12098
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.9127873182296753,
      "learning_rate": 1.3118191297367716e-05,
      "loss": 0.5811,
      "step": 12099
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5780816078186035,
      "learning_rate": 1.3117130374644255e-05,
      "loss": 0.4664,
      "step": 12100
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.318179965019226,
      "learning_rate": 1.311606941305836e-05,
      "loss": 0.4971,
      "step": 12101
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4646599292755127,
      "learning_rate": 1.3115008412623258e-05,
      "loss": 0.5484,
      "step": 12102
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3610605001449585,
      "learning_rate": 1.3113947373352176e-05,
      "loss": 0.5012,
      "step": 12103
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.433309555053711,
      "learning_rate": 1.3112886295258344e-05,
      "loss": 0.5394,
      "step": 12104
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3492635488510132,
      "learning_rate": 1.3111825178354992e-05,
      "loss": 0.4432,
      "step": 12105
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4398609399795532,
      "learning_rate": 1.3110764022655344e-05,
      "loss": 0.5538,
      "step": 12106
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7733198404312134,
      "learning_rate": 1.3109702828172637e-05,
      "loss": 0.683,
      "step": 12107
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4006930589675903,
      "learning_rate": 1.3108641594920095e-05,
      "loss": 0.4232,
      "step": 12108
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4889469146728516,
      "learning_rate": 1.3107580322910952e-05,
      "loss": 0.6114,
      "step": 12109
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4814335107803345,
      "learning_rate": 1.310651901215844e-05,
      "loss": 0.5153,
      "step": 12110
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2950105667114258,
      "learning_rate": 1.3105457662675788e-05,
      "loss": 0.5483,
      "step": 12111
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4579331874847412,
      "learning_rate": 1.3104396274476233e-05,
      "loss": 0.5283,
      "step": 12112
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.446308970451355,
      "learning_rate": 1.3103334847573002e-05,
      "loss": 0.5777,
      "step": 12113
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7594306468963623,
      "learning_rate": 1.3102273381979329e-05,
      "loss": 0.576,
      "step": 12114
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3290711641311646,
      "learning_rate": 1.3101211877708451e-05,
      "loss": 0.512,
      "step": 12115
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.570750117301941,
      "learning_rate": 1.3100150334773601e-05,
      "loss": 0.5319,
      "step": 12116
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7541199922561646,
      "learning_rate": 1.3099088753188012e-05,
      "loss": 0.5194,
      "step": 12117
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.454965591430664,
      "learning_rate": 1.309802713296492e-05,
      "loss": 0.5124,
      "step": 12118
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6043686866760254,
      "learning_rate": 1.3096965474117562e-05,
      "loss": 0.5808,
      "step": 12119
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3517074584960938,
      "learning_rate": 1.3095903776659174e-05,
      "loss": 0.4881,
      "step": 12120
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7203301191329956,
      "learning_rate": 1.3094842040602989e-05,
      "loss": 0.6151,
      "step": 12121
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.400872826576233,
      "learning_rate": 1.3093780265962249e-05,
      "loss": 0.4381,
      "step": 12122
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2524909973144531,
      "learning_rate": 1.3092718452750188e-05,
      "loss": 0.4542,
      "step": 12123
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4119176864624023,
      "learning_rate": 1.3091656600980046e-05,
      "loss": 0.5045,
      "step": 12124
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6993837356567383,
      "learning_rate": 1.309059471066506e-05,
      "loss": 0.5746,
      "step": 12125
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.341140866279602,
      "learning_rate": 1.3089532781818471e-05,
      "loss": 0.5302,
      "step": 12126
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.8907917737960815,
      "learning_rate": 1.3088470814453515e-05,
      "loss": 0.6437,
      "step": 12127
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4170117378234863,
      "learning_rate": 1.3087408808583438e-05,
      "loss": 0.5466,
      "step": 12128
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5320576429367065,
      "learning_rate": 1.3086346764221471e-05,
      "loss": 0.5245,
      "step": 12129
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.449240803718567,
      "learning_rate": 1.3085284681380865e-05,
      "loss": 0.5615,
      "step": 12130
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.9711660146713257,
      "learning_rate": 1.3084222560074855e-05,
      "loss": 0.5408,
      "step": 12131
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4254604578018188,
      "learning_rate": 1.3083160400316683e-05,
      "loss": 0.5179,
      "step": 12132
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5269076824188232,
      "learning_rate": 1.3082098202119597e-05,
      "loss": 0.6252,
      "step": 12133
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.4405708312988281,
      "learning_rate": 1.3081035965496834e-05,
      "loss": 0.6193,
      "step": 12134
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6992336511611938,
      "learning_rate": 1.3079973690461637e-05,
      "loss": 0.6867,
      "step": 12135
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.9063515663146973,
      "learning_rate": 1.3078911377027256e-05,
      "loss": 0.5791,
      "step": 12136
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2591321468353271,
      "learning_rate": 1.307784902520693e-05,
      "loss": 0.4446,
      "step": 12137
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.481236219406128,
      "learning_rate": 1.3076786635013906e-05,
      "loss": 0.6254,
      "step": 12138
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3685415983200073,
      "learning_rate": 1.3075724206461425e-05,
      "loss": 0.5938,
      "step": 12139
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2457572221755981,
      "learning_rate": 1.3074661739562737e-05,
      "loss": 0.5033,
      "step": 12140
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.874257206916809,
      "learning_rate": 1.3073599234331089e-05,
      "loss": 0.6889,
      "step": 12141
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.493667483329773,
      "learning_rate": 1.3072536690779723e-05,
      "loss": 0.6188,
      "step": 12142
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5575079917907715,
      "learning_rate": 1.3071474108921892e-05,
      "loss": 0.5802,
      "step": 12143
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.559177041053772,
      "learning_rate": 1.3070411488770836e-05,
      "loss": 0.4732,
      "step": 12144
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.396593451499939,
      "learning_rate": 1.306934883033981e-05,
      "loss": 0.4454,
      "step": 12145
    },
    {
      "epoch": 1.26,
      "grad_norm": 2.445760488510132,
      "learning_rate": 1.3068286133642061e-05,
      "loss": 0.4657,
      "step": 12146
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6161150932312012,
      "learning_rate": 1.3067223398690836e-05,
      "loss": 0.5689,
      "step": 12147
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4711261987686157,
      "learning_rate": 1.3066160625499384e-05,
      "loss": 0.5467,
      "step": 12148
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7435487508773804,
      "learning_rate": 1.306509781408096e-05,
      "loss": 0.6147,
      "step": 12149
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3240107297897339,
      "learning_rate": 1.3064034964448809e-05,
      "loss": 0.633,
      "step": 12150
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4126880168914795,
      "learning_rate": 1.3062972076616187e-05,
      "loss": 0.6077,
      "step": 12151
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.374744176864624,
      "learning_rate": 1.306190915059634e-05,
      "loss": 0.524,
      "step": 12152
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.189992070198059,
      "learning_rate": 1.306084618640252e-05,
      "loss": 0.5467,
      "step": 12153
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4231595993041992,
      "learning_rate": 1.3059783184047988e-05,
      "loss": 0.5137,
      "step": 12154
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7638436555862427,
      "learning_rate": 1.3058720143545985e-05,
      "loss": 0.665,
      "step": 12155
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5658900737762451,
      "learning_rate": 1.3057657064909772e-05,
      "loss": 0.5878,
      "step": 12156
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4515092372894287,
      "learning_rate": 1.30565939481526e-05,
      "loss": 0.5312,
      "step": 12157
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5978410243988037,
      "learning_rate": 1.3055530793287724e-05,
      "loss": 0.5435,
      "step": 12158
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6923844814300537,
      "learning_rate": 1.3054467600328403e-05,
      "loss": 0.6306,
      "step": 12159
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7973852157592773,
      "learning_rate": 1.3053404369287881e-05,
      "loss": 0.626,
      "step": 12160
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.581559658050537,
      "learning_rate": 1.3052341100179424e-05,
      "loss": 0.5651,
      "step": 12161
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4263310432434082,
      "learning_rate": 1.3051277793016286e-05,
      "loss": 0.4721,
      "step": 12162
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7178465127944946,
      "learning_rate": 1.305021444781172e-05,
      "loss": 0.5525,
      "step": 12163
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.710755467414856,
      "learning_rate": 1.304915106457899e-05,
      "loss": 0.5487,
      "step": 12164
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5180201530456543,
      "learning_rate": 1.3048087643331345e-05,
      "loss": 0.6067,
      "step": 12165
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.526455283164978,
      "learning_rate": 1.3047024184082046e-05,
      "loss": 0.5593,
      "step": 12166
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3015789985656738,
      "learning_rate": 1.3045960686844357e-05,
      "loss": 0.5276,
      "step": 12167
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3888509273529053,
      "learning_rate": 1.304489715163153e-05,
      "loss": 0.5213,
      "step": 12168
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8478955030441284,
      "learning_rate": 1.304383357845683e-05,
      "loss": 0.5903,
      "step": 12169
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4076621532440186,
      "learning_rate": 1.3042769967333513e-05,
      "loss": 0.5596,
      "step": 12170
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3480900526046753,
      "learning_rate": 1.3041706318274838e-05,
      "loss": 0.4782,
      "step": 12171
    },
    {
      "epoch": 1.26,
      "grad_norm": 2.1977219581604004,
      "learning_rate": 1.3040642631294074e-05,
      "loss": 0.6257,
      "step": 12172
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8241217136383057,
      "learning_rate": 1.3039578906404473e-05,
      "loss": 0.4985,
      "step": 12173
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4748156070709229,
      "learning_rate": 1.3038515143619303e-05,
      "loss": 0.6357,
      "step": 12174
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.675991415977478,
      "learning_rate": 1.3037451342951824e-05,
      "loss": 0.461,
      "step": 12175
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7503998279571533,
      "learning_rate": 1.3036387504415295e-05,
      "loss": 0.6223,
      "step": 12176
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6292150020599365,
      "learning_rate": 1.3035323628022989e-05,
      "loss": 0.5185,
      "step": 12177
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6711947917938232,
      "learning_rate": 1.3034259713788162e-05,
      "loss": 0.4555,
      "step": 12178
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5061795711517334,
      "learning_rate": 1.3033195761724079e-05,
      "loss": 0.5375,
      "step": 12179
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.663030982017517,
      "learning_rate": 1.3032131771844008e-05,
      "loss": 0.6572,
      "step": 12180
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.569500207901001,
      "learning_rate": 1.303106774416121e-05,
      "loss": 0.534,
      "step": 12181
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6400198936462402,
      "learning_rate": 1.3030003678688956e-05,
      "loss": 0.5881,
      "step": 12182
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4921393394470215,
      "learning_rate": 1.3028939575440505e-05,
      "loss": 0.5689,
      "step": 12183
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6861473321914673,
      "learning_rate": 1.3027875434429129e-05,
      "loss": 0.6484,
      "step": 12184
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4726256132125854,
      "learning_rate": 1.3026811255668095e-05,
      "loss": 0.5916,
      "step": 12185
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3079530000686646,
      "learning_rate": 1.302574703917067e-05,
      "loss": 0.6051,
      "step": 12186
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3097933530807495,
      "learning_rate": 1.3024682784950118e-05,
      "loss": 0.5763,
      "step": 12187
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5062928199768066,
      "learning_rate": 1.3023618493019709e-05,
      "loss": 0.4978,
      "step": 12188
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4228160381317139,
      "learning_rate": 1.3022554163392715e-05,
      "loss": 0.6649,
      "step": 12189
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5067894458770752,
      "learning_rate": 1.3021489796082407e-05,
      "loss": 0.5614,
      "step": 12190
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2350467443466187,
      "learning_rate": 1.3020425391102047e-05,
      "loss": 0.5352,
      "step": 12191
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4654510021209717,
      "learning_rate": 1.3019360948464912e-05,
      "loss": 0.5712,
      "step": 12192
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6536556482315063,
      "learning_rate": 1.3018296468184271e-05,
      "loss": 0.6972,
      "step": 12193
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.62249755859375,
      "learning_rate": 1.3017231950273395e-05,
      "loss": 0.5986,
      "step": 12194
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4920085668563843,
      "learning_rate": 1.3016167394745555e-05,
      "loss": 0.4827,
      "step": 12195
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5698894262313843,
      "learning_rate": 1.3015102801614026e-05,
      "loss": 0.5539,
      "step": 12196
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5730271339416504,
      "learning_rate": 1.3014038170892075e-05,
      "loss": 0.577,
      "step": 12197
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2353698015213013,
      "learning_rate": 1.3012973502592983e-05,
      "loss": 0.4572,
      "step": 12198
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5356098413467407,
      "learning_rate": 1.3011908796730017e-05,
      "loss": 0.6271,
      "step": 12199
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5937119722366333,
      "learning_rate": 1.3010844053316456e-05,
      "loss": 0.59,
      "step": 12200
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.566507339477539,
      "learning_rate": 1.300977927236557e-05,
      "loss": 0.5156,
      "step": 12201
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.77274489402771,
      "learning_rate": 1.3008714453890634e-05,
      "loss": 0.5439,
      "step": 12202
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.373940110206604,
      "learning_rate": 1.300764959790493e-05,
      "loss": 0.582,
      "step": 12203
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4434775114059448,
      "learning_rate": 1.3006584704421728e-05,
      "loss": 0.5883,
      "step": 12204
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8695510625839233,
      "learning_rate": 1.3005519773454305e-05,
      "loss": 0.4287,
      "step": 12205
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4685841798782349,
      "learning_rate": 1.300445480501594e-05,
      "loss": 0.545,
      "step": 12206
    },
    {
      "epoch": 1.26,
      "grad_norm": 2.2168381214141846,
      "learning_rate": 1.3003389799119908e-05,
      "loss": 0.5101,
      "step": 12207
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.438111424446106,
      "learning_rate": 1.3002324755779489e-05,
      "loss": 0.584,
      "step": 12208
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.655977487564087,
      "learning_rate": 1.300125967500796e-05,
      "loss": 0.5461,
      "step": 12209
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3595534563064575,
      "learning_rate": 1.3000194556818599e-05,
      "loss": 0.6054,
      "step": 12210
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.9312344789505005,
      "learning_rate": 1.299912940122469e-05,
      "loss": 0.5153,
      "step": 12211
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4076505899429321,
      "learning_rate": 1.2998064208239508e-05,
      "loss": 0.3531,
      "step": 12212
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3731783628463745,
      "learning_rate": 1.2996998977876331e-05,
      "loss": 0.5012,
      "step": 12213
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7715606689453125,
      "learning_rate": 1.2995933710148448e-05,
      "loss": 0.5317,
      "step": 12214
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3362798690795898,
      "learning_rate": 1.299486840506913e-05,
      "loss": 0.5802,
      "step": 12215
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7575207948684692,
      "learning_rate": 1.2993803062651663e-05,
      "loss": 0.5557,
      "step": 12216
    },
    {
      "epoch": 1.26,
      "grad_norm": 2.0743582248687744,
      "learning_rate": 1.2992737682909336e-05,
      "loss": 0.6065,
      "step": 12217
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6815941333770752,
      "learning_rate": 1.2991672265855419e-05,
      "loss": 0.4374,
      "step": 12218
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6797356605529785,
      "learning_rate": 1.2990606811503205e-05,
      "loss": 0.7381,
      "step": 12219
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.249773621559143,
      "learning_rate": 1.298954131986597e-05,
      "loss": 0.5175,
      "step": 12220
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2277194261550903,
      "learning_rate": 1.2988475790957002e-05,
      "loss": 0.487,
      "step": 12221
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.4742597341537476,
      "learning_rate": 1.2987410224789587e-05,
      "loss": 0.4294,
      "step": 12222
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.725433349609375,
      "learning_rate": 1.2986344621377006e-05,
      "loss": 0.5415,
      "step": 12223
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2982512712478638,
      "learning_rate": 1.2985278980732545e-05,
      "loss": 0.53,
      "step": 12224
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.8408030271530151,
      "learning_rate": 1.298421330286949e-05,
      "loss": 0.5721,
      "step": 12225
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6381535530090332,
      "learning_rate": 1.2983147587801127e-05,
      "loss": 0.589,
      "step": 12226
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7971614599227905,
      "learning_rate": 1.2982081835540747e-05,
      "loss": 0.5679,
      "step": 12227
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.797231912612915,
      "learning_rate": 1.298101604610163e-05,
      "loss": 0.5433,
      "step": 12228
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.6308199167251587,
      "learning_rate": 1.2979950219497068e-05,
      "loss": 0.6004,
      "step": 12229
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.7673227787017822,
      "learning_rate": 1.2978884355740351e-05,
      "loss": 0.57,
      "step": 12230
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.863693356513977,
      "learning_rate": 1.2977818454844762e-05,
      "loss": 0.5949,
      "step": 12231
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5014121532440186,
      "learning_rate": 1.2976752516823595e-05,
      "loss": 0.5846,
      "step": 12232
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.808500051498413,
      "learning_rate": 1.2975686541690133e-05,
      "loss": 0.5365,
      "step": 12233
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.468794822692871,
      "learning_rate": 1.297462052945767e-05,
      "loss": 0.5584,
      "step": 12234
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7215590476989746,
      "learning_rate": 1.2973554480139501e-05,
      "loss": 0.5251,
      "step": 12235
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.518821120262146,
      "learning_rate": 1.2972488393748907e-05,
      "loss": 0.5649,
      "step": 12236
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6124576330184937,
      "learning_rate": 1.297142227029919e-05,
      "loss": 0.5417,
      "step": 12237
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.695726990699768,
      "learning_rate": 1.2970356109803632e-05,
      "loss": 0.5972,
      "step": 12238
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7856525182724,
      "learning_rate": 1.296928991227553e-05,
      "loss": 0.6417,
      "step": 12239
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.2661832571029663,
      "learning_rate": 1.2968223677728178e-05,
      "loss": 0.5065,
      "step": 12240
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.2169277667999268,
      "learning_rate": 1.2967157406174866e-05,
      "loss": 0.4806,
      "step": 12241
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6747950315475464,
      "learning_rate": 1.2966091097628889e-05,
      "loss": 0.5241,
      "step": 12242
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3602073192596436,
      "learning_rate": 1.2965024752103542e-05,
      "loss": 0.6267,
      "step": 12243
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.8937101364135742,
      "learning_rate": 1.2963958369612117e-05,
      "loss": 0.5908,
      "step": 12244
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5708115100860596,
      "learning_rate": 1.2962891950167912e-05,
      "loss": 0.5897,
      "step": 12245
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5576273202896118,
      "learning_rate": 1.2961825493784221e-05,
      "loss": 0.6126,
      "step": 12246
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.26456618309021,
      "learning_rate": 1.2960759000474336e-05,
      "loss": 0.5774,
      "step": 12247
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.442564606666565,
      "learning_rate": 1.2959692470251564e-05,
      "loss": 0.6108,
      "step": 12248
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.2097285985946655,
      "learning_rate": 1.2958625903129192e-05,
      "loss": 0.5474,
      "step": 12249
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3543341159820557,
      "learning_rate": 1.295755929912052e-05,
      "loss": 0.55,
      "step": 12250
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5427658557891846,
      "learning_rate": 1.2956492658238844e-05,
      "loss": 0.5788,
      "step": 12251
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3953194618225098,
      "learning_rate": 1.2955425980497467e-05,
      "loss": 0.4458,
      "step": 12252
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6520535945892334,
      "learning_rate": 1.2954359265909685e-05,
      "loss": 0.5748,
      "step": 12253
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5004926919937134,
      "learning_rate": 1.2953292514488796e-05,
      "loss": 0.4843,
      "step": 12254
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3476811647415161,
      "learning_rate": 1.2952225726248102e-05,
      "loss": 0.6272,
      "step": 12255
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5433443784713745,
      "learning_rate": 1.2951158901200901e-05,
      "loss": 0.5821,
      "step": 12256
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3725465536117554,
      "learning_rate": 1.2950092039360494e-05,
      "loss": 0.4762,
      "step": 12257
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5434110164642334,
      "learning_rate": 1.2949025140740183e-05,
      "loss": 0.5653,
      "step": 12258
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.8663350343704224,
      "learning_rate": 1.2947958205353269e-05,
      "loss": 0.4247,
      "step": 12259
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.475980281829834,
      "learning_rate": 1.2946891233213053e-05,
      "loss": 0.4486,
      "step": 12260
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5440031290054321,
      "learning_rate": 1.2945824224332837e-05,
      "loss": 0.5197,
      "step": 12261
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4094802141189575,
      "learning_rate": 1.2944757178725926e-05,
      "loss": 0.5144,
      "step": 12262
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7497248649597168,
      "learning_rate": 1.2943690096405623e-05,
      "loss": 0.5204,
      "step": 12263
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6411100625991821,
      "learning_rate": 1.2942622977385228e-05,
      "loss": 0.5893,
      "step": 12264
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5331026315689087,
      "learning_rate": 1.294155582167805e-05,
      "loss": 0.6865,
      "step": 12265
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.334823727607727,
      "learning_rate": 1.294048862929739e-05,
      "loss": 0.4375,
      "step": 12266
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7295808792114258,
      "learning_rate": 1.2939421400256555e-05,
      "loss": 0.6682,
      "step": 12267
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5768003463745117,
      "learning_rate": 1.2938354134568848e-05,
      "loss": 0.5645,
      "step": 12268
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4495733976364136,
      "learning_rate": 1.2937286832247581e-05,
      "loss": 0.5123,
      "step": 12269
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.462806224822998,
      "learning_rate": 1.2936219493306058e-05,
      "loss": 0.5666,
      "step": 12270
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5106592178344727,
      "learning_rate": 1.2935152117757581e-05,
      "loss": 0.5013,
      "step": 12271
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3857452869415283,
      "learning_rate": 1.2934084705615462e-05,
      "loss": 0.5769,
      "step": 12272
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5213984251022339,
      "learning_rate": 1.2933017256893005e-05,
      "loss": 0.4417,
      "step": 12273
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7542564868927002,
      "learning_rate": 1.2931949771603525e-05,
      "loss": 0.595,
      "step": 12274
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3696800470352173,
      "learning_rate": 1.2930882249760323e-05,
      "loss": 0.4833,
      "step": 12275
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4911816120147705,
      "learning_rate": 1.2929814691376716e-05,
      "loss": 0.4735,
      "step": 12276
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7900933027267456,
      "learning_rate": 1.2928747096466006e-05,
      "loss": 0.5693,
      "step": 12277
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4854109287261963,
      "learning_rate": 1.2927679465041507e-05,
      "loss": 0.4597,
      "step": 12278
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.554857850074768,
      "learning_rate": 1.2926611797116533e-05,
      "loss": 0.5986,
      "step": 12279
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4016306400299072,
      "learning_rate": 1.2925544092704387e-05,
      "loss": 0.5775,
      "step": 12280
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.791083812713623,
      "learning_rate": 1.2924476351818385e-05,
      "loss": 0.6031,
      "step": 12281
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4477524757385254,
      "learning_rate": 1.2923408574471844e-05,
      "loss": 0.509,
      "step": 12282
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3412078619003296,
      "learning_rate": 1.2922340760678065e-05,
      "loss": 0.5015,
      "step": 12283
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6634761095046997,
      "learning_rate": 1.2921272910450368e-05,
      "loss": 0.6499,
      "step": 12284
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.598130702972412,
      "learning_rate": 1.2920205023802069e-05,
      "loss": 0.5979,
      "step": 12285
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5694806575775146,
      "learning_rate": 1.2919137100746472e-05,
      "loss": 0.4187,
      "step": 12286
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.573853611946106,
      "learning_rate": 1.2918069141296901e-05,
      "loss": 0.5114,
      "step": 12287
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.58036470413208,
      "learning_rate": 1.2917001145466665e-05,
      "loss": 0.6393,
      "step": 12288
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4096641540527344,
      "learning_rate": 1.2915933113269084e-05,
      "loss": 0.5299,
      "step": 12289
    },
    {
      "epoch": 1.27,
      "grad_norm": 2.0256845951080322,
      "learning_rate": 1.2914865044717466e-05,
      "loss": 0.701,
      "step": 12290
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.51055109500885,
      "learning_rate": 1.2913796939825135e-05,
      "loss": 0.5712,
      "step": 12291
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3371970653533936,
      "learning_rate": 1.2912728798605403e-05,
      "loss": 0.4104,
      "step": 12292
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5686137676239014,
      "learning_rate": 1.2911660621071587e-05,
      "loss": 0.5891,
      "step": 12293
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7471563816070557,
      "learning_rate": 1.2910592407237003e-05,
      "loss": 0.5655,
      "step": 12294
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4788386821746826,
      "learning_rate": 1.2909524157114977e-05,
      "loss": 0.5453,
      "step": 12295
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4527592658996582,
      "learning_rate": 1.2908455870718815e-05,
      "loss": 0.6352,
      "step": 12296
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5992997884750366,
      "learning_rate": 1.2907387548061847e-05,
      "loss": 0.5117,
      "step": 12297
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5497632026672363,
      "learning_rate": 1.2906319189157384e-05,
      "loss": 0.5262,
      "step": 12298
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.8411623239517212,
      "learning_rate": 1.2905250794018749e-05,
      "loss": 0.7217,
      "step": 12299
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4444026947021484,
      "learning_rate": 1.2904182362659264e-05,
      "loss": 0.488,
      "step": 12300
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.554212212562561,
      "learning_rate": 1.2903113895092246e-05,
      "loss": 0.5479,
      "step": 12301
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7412011623382568,
      "learning_rate": 1.2902045391331017e-05,
      "loss": 0.5855,
      "step": 12302
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4182515144348145,
      "learning_rate": 1.29009768513889e-05,
      "loss": 0.5009,
      "step": 12303
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.600705623626709,
      "learning_rate": 1.2899908275279213e-05,
      "loss": 0.64,
      "step": 12304
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7193691730499268,
      "learning_rate": 1.2898839663015286e-05,
      "loss": 0.4515,
      "step": 12305
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4693188667297363,
      "learning_rate": 1.2897771014610432e-05,
      "loss": 0.5491,
      "step": 12306
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5249958038330078,
      "learning_rate": 1.289670233007798e-05,
      "loss": 0.5443,
      "step": 12307
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3960129022598267,
      "learning_rate": 1.2895633609431257e-05,
      "loss": 0.6124,
      "step": 12308
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.692811369895935,
      "learning_rate": 1.289456485268358e-05,
      "loss": 0.5728,
      "step": 12309
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.470721960067749,
      "learning_rate": 1.2893496059848279e-05,
      "loss": 0.5237,
      "step": 12310
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.7803881168365479,
      "learning_rate": 1.289242723093867e-05,
      "loss": 0.5704,
      "step": 12311
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4571006298065186,
      "learning_rate": 1.289135836596809e-05,
      "loss": 0.5692,
      "step": 12312
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4829139709472656,
      "learning_rate": 1.2890289464949862e-05,
      "loss": 0.5876,
      "step": 12313
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4292166233062744,
      "learning_rate": 1.2889220527897306e-05,
      "loss": 0.6105,
      "step": 12314
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5825538635253906,
      "learning_rate": 1.2888151554823756e-05,
      "loss": 0.7031,
      "step": 12315
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4133466482162476,
      "learning_rate": 1.2887082545742535e-05,
      "loss": 0.3637,
      "step": 12316
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1528836488723755,
      "learning_rate": 1.2886013500666973e-05,
      "loss": 0.4729,
      "step": 12317
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3679909706115723,
      "learning_rate": 1.2884944419610396e-05,
      "loss": 0.5675,
      "step": 12318
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6895428895950317,
      "learning_rate": 1.2883875302586136e-05,
      "loss": 0.4683,
      "step": 12319
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.2468903064727783,
      "learning_rate": 1.2882806149607519e-05,
      "loss": 0.5875,
      "step": 12320
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.494971752166748,
      "learning_rate": 1.2881736960687876e-05,
      "loss": 0.5671,
      "step": 12321
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4875969886779785,
      "learning_rate": 1.2880667735840535e-05,
      "loss": 0.6154,
      "step": 12322
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.3773694038391113,
      "learning_rate": 1.2879598475078834e-05,
      "loss": 0.4556,
      "step": 12323
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6030793190002441,
      "learning_rate": 1.2878529178416092e-05,
      "loss": 0.5324,
      "step": 12324
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6373652219772339,
      "learning_rate": 1.2877459845865647e-05,
      "loss": 0.5099,
      "step": 12325
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.6738412380218506,
      "learning_rate": 1.2876390477440834e-05,
      "loss": 0.588,
      "step": 12326
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.4188249111175537,
      "learning_rate": 1.2875321073154976e-05,
      "loss": 0.5165,
      "step": 12327
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.475390911102295,
      "learning_rate": 1.2874251633021415e-05,
      "loss": 0.6234,
      "step": 12328
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.3884329795837402,
      "learning_rate": 1.287318215705348e-05,
      "loss": 0.4514,
      "step": 12329
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5851924419403076,
      "learning_rate": 1.2872112645264503e-05,
      "loss": 0.7115,
      "step": 12330
    },
    {
      "epoch": 1.28,
      "grad_norm": 2.393279552459717,
      "learning_rate": 1.2871043097667819e-05,
      "loss": 0.5941,
      "step": 12331
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6592313051223755,
      "learning_rate": 1.2869973514276764e-05,
      "loss": 0.5908,
      "step": 12332
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.590113878250122,
      "learning_rate": 1.2868903895104672e-05,
      "loss": 0.5789,
      "step": 12333
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6680562496185303,
      "learning_rate": 1.286783424016488e-05,
      "loss": 0.4448,
      "step": 12334
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.509209156036377,
      "learning_rate": 1.286676454947072e-05,
      "loss": 0.5947,
      "step": 12335
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4059760570526123,
      "learning_rate": 1.2865694823035533e-05,
      "loss": 0.5237,
      "step": 12336
    },
    {
      "epoch": 1.28,
      "grad_norm": 7.097452640533447,
      "learning_rate": 1.2864625060872648e-05,
      "loss": 0.6557,
      "step": 12337
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4195972681045532,
      "learning_rate": 1.2863555262995409e-05,
      "loss": 0.572,
      "step": 12338
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.575164556503296,
      "learning_rate": 1.2862485429417156e-05,
      "loss": 0.5757,
      "step": 12339
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8206902742385864,
      "learning_rate": 1.2861415560151217e-05,
      "loss": 0.6645,
      "step": 12340
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.3913917541503906,
      "learning_rate": 1.2860345655210941e-05,
      "loss": 0.5216,
      "step": 12341
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8236826658248901,
      "learning_rate": 1.2859275714609658e-05,
      "loss": 0.5591,
      "step": 12342
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6412320137023926,
      "learning_rate": 1.2858205738360714e-05,
      "loss": 0.46,
      "step": 12343
    },
    {
      "epoch": 1.28,
      "grad_norm": 2.0706231594085693,
      "learning_rate": 1.2857135726477448e-05,
      "loss": 0.4905,
      "step": 12344
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8714828491210938,
      "learning_rate": 1.2856065678973196e-05,
      "loss": 0.4608,
      "step": 12345
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4691990613937378,
      "learning_rate": 1.28549955958613e-05,
      "loss": 0.5661,
      "step": 12346
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7978700399398804,
      "learning_rate": 1.2853925477155105e-05,
      "loss": 0.6276,
      "step": 12347
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.413501501083374,
      "learning_rate": 1.2852855322867947e-05,
      "loss": 0.4541,
      "step": 12348
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6300140619277954,
      "learning_rate": 1.2851785133013172e-05,
      "loss": 0.6792,
      "step": 12349
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.48401939868927,
      "learning_rate": 1.2850714907604123e-05,
      "loss": 0.4985,
      "step": 12350
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.9326753616333008,
      "learning_rate": 1.284964464665414e-05,
      "loss": 0.4966,
      "step": 12351
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5850496292114258,
      "learning_rate": 1.2848574350176568e-05,
      "loss": 0.4924,
      "step": 12352
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6445025205612183,
      "learning_rate": 1.2847504018184748e-05,
      "loss": 0.6607,
      "step": 12353
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8253268003463745,
      "learning_rate": 1.2846433650692031e-05,
      "loss": 0.6171,
      "step": 12354
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6014634370803833,
      "learning_rate": 1.2845363247711756e-05,
      "loss": 0.6571,
      "step": 12355
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6171845197677612,
      "learning_rate": 1.2844292809257266e-05,
      "loss": 0.5694,
      "step": 12356
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8234056234359741,
      "learning_rate": 1.2843222335341915e-05,
      "loss": 0.561,
      "step": 12357
    },
    {
      "epoch": 1.28,
      "grad_norm": 2.370985269546509,
      "learning_rate": 1.2842151825979042e-05,
      "loss": 0.751,
      "step": 12358
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4042032957077026,
      "learning_rate": 1.2841081281181993e-05,
      "loss": 0.5082,
      "step": 12359
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5101226568222046,
      "learning_rate": 1.2840010700964121e-05,
      "loss": 0.5624,
      "step": 12360
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1630125045776367,
      "learning_rate": 1.2838940085338767e-05,
      "loss": 0.3709,
      "step": 12361
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.500622272491455,
      "learning_rate": 1.2837869434319283e-05,
      "loss": 0.5802,
      "step": 12362
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.2673494815826416,
      "learning_rate": 1.2836798747919015e-05,
      "loss": 0.5494,
      "step": 12363
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6207319498062134,
      "learning_rate": 1.283572802615131e-05,
      "loss": 0.6649,
      "step": 12364
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5893495082855225,
      "learning_rate": 1.2834657269029524e-05,
      "loss": 0.531,
      "step": 12365
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.656088948249817,
      "learning_rate": 1.2833586476566998e-05,
      "loss": 0.5825,
      "step": 12366
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4584143161773682,
      "learning_rate": 1.2832515648777089e-05,
      "loss": 0.513,
      "step": 12367
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8472838401794434,
      "learning_rate": 1.2831444785673141e-05,
      "loss": 0.5583,
      "step": 12368
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6658943891525269,
      "learning_rate": 1.2830373887268511e-05,
      "loss": 0.4599,
      "step": 12369
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.458068609237671,
      "learning_rate": 1.2829302953576546e-05,
      "loss": 0.5501,
      "step": 12370
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.9663987159729004,
      "learning_rate": 1.2828231984610597e-05,
      "loss": 0.673,
      "step": 12371
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.62440025806427,
      "learning_rate": 1.2827160980384021e-05,
      "loss": 0.5463,
      "step": 12372
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4512883424758911,
      "learning_rate": 1.2826089940910168e-05,
      "loss": 0.5375,
      "step": 12373
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5055471658706665,
      "learning_rate": 1.282501886620239e-05,
      "loss": 0.4807,
      "step": 12374
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6970374584197998,
      "learning_rate": 1.2823947756274042e-05,
      "loss": 0.4475,
      "step": 12375
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4009169340133667,
      "learning_rate": 1.2822876611138478e-05,
      "loss": 0.6035,
      "step": 12376
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7491973638534546,
      "learning_rate": 1.2821805430809051e-05,
      "loss": 0.4821,
      "step": 12377
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4723550081253052,
      "learning_rate": 1.2820734215299116e-05,
      "loss": 0.471,
      "step": 12378
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7807599306106567,
      "learning_rate": 1.281966296462203e-05,
      "loss": 0.5603,
      "step": 12379
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7711988687515259,
      "learning_rate": 1.2818591678791148e-05,
      "loss": 0.5464,
      "step": 12380
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.3735555410385132,
      "learning_rate": 1.2817520357819824e-05,
      "loss": 0.615,
      "step": 12381
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4804240465164185,
      "learning_rate": 1.2816449001721414e-05,
      "loss": 0.4801,
      "step": 12382
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4991252422332764,
      "learning_rate": 1.2815377610509283e-05,
      "loss": 0.5141,
      "step": 12383
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5354347229003906,
      "learning_rate": 1.2814306184196779e-05,
      "loss": 0.5403,
      "step": 12384
    },
    {
      "epoch": 1.28,
      "grad_norm": 2.003401756286621,
      "learning_rate": 1.281323472279726e-05,
      "loss": 0.5359,
      "step": 12385
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5279808044433594,
      "learning_rate": 1.2812163226324094e-05,
      "loss": 0.5178,
      "step": 12386
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8577624559402466,
      "learning_rate": 1.2811091694790632e-05,
      "loss": 0.5681,
      "step": 12387
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5187528133392334,
      "learning_rate": 1.2810020128210235e-05,
      "loss": 0.527,
      "step": 12388
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5585664510726929,
      "learning_rate": 1.2808948526596262e-05,
      "loss": 0.5274,
      "step": 12389
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.707353949546814,
      "learning_rate": 1.2807876889962072e-05,
      "loss": 0.4821,
      "step": 12390
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6222871541976929,
      "learning_rate": 1.280680521832103e-05,
      "loss": 0.6298,
      "step": 12391
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.388695478439331,
      "learning_rate": 1.2805733511686492e-05,
      "loss": 0.5769,
      "step": 12392
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7227669954299927,
      "learning_rate": 1.2804661770071818e-05,
      "loss": 0.4949,
      "step": 12393
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6354625225067139,
      "learning_rate": 1.2803589993490377e-05,
      "loss": 0.532,
      "step": 12394
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5835965871810913,
      "learning_rate": 1.2802518181955526e-05,
      "loss": 0.7008,
      "step": 12395
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5891435146331787,
      "learning_rate": 1.2801446335480628e-05,
      "loss": 0.4835,
      "step": 12396
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.342827558517456,
      "learning_rate": 1.280037445407905e-05,
      "loss": 0.5171,
      "step": 12397
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8280925750732422,
      "learning_rate": 1.2799302537764149e-05,
      "loss": 0.6912,
      "step": 12398
    },
    {
      "epoch": 1.28,
      "grad_norm": 2.028115749359131,
      "learning_rate": 1.2798230586549297e-05,
      "loss": 0.5775,
      "step": 12399
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.49118173122406,
      "learning_rate": 1.279715860044785e-05,
      "loss": 0.5335,
      "step": 12400
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6185591220855713,
      "learning_rate": 1.2796086579473177e-05,
      "loss": 0.5872,
      "step": 12401
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6172832250595093,
      "learning_rate": 1.2795014523638645e-05,
      "loss": 0.4847,
      "step": 12402
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.7540202140808105,
      "learning_rate": 1.2793942432957618e-05,
      "loss": 0.5435,
      "step": 12403
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.3353527784347534,
      "learning_rate": 1.2792870307443461e-05,
      "loss": 0.6675,
      "step": 12404
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6515069007873535,
      "learning_rate": 1.279179814710954e-05,
      "loss": 0.4306,
      "step": 12405
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4305469989776611,
      "learning_rate": 1.2790725951969224e-05,
      "loss": 0.5508,
      "step": 12406
    },
    {
      "epoch": 1.28,
      "grad_norm": 2.149170160293579,
      "learning_rate": 1.2789653722035884e-05,
      "loss": 0.6192,
      "step": 12407
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5155763626098633,
      "learning_rate": 1.2788581457322881e-05,
      "loss": 0.5404,
      "step": 12408
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6660581827163696,
      "learning_rate": 1.2787509157843586e-05,
      "loss": 0.4508,
      "step": 12409
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6385669708251953,
      "learning_rate": 1.278643682361137e-05,
      "loss": 0.5789,
      "step": 12410
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.475635290145874,
      "learning_rate": 1.27853644546396e-05,
      "loss": 0.5541,
      "step": 12411
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1446675062179565,
      "learning_rate": 1.2784292050941644e-05,
      "loss": 0.5036,
      "step": 12412
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6021971702575684,
      "learning_rate": 1.2783219612530875e-05,
      "loss": 0.5808,
      "step": 12413
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5032812356948853,
      "learning_rate": 1.2782147139420662e-05,
      "loss": 0.5024,
      "step": 12414
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.6080602407455444,
      "learning_rate": 1.2781074631624377e-05,
      "loss": 0.6103,
      "step": 12415
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4008554220199585,
      "learning_rate": 1.278000208915539e-05,
      "loss": 0.4958,
      "step": 12416
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.337507724761963,
      "learning_rate": 1.2778929512027076e-05,
      "loss": 0.5265,
      "step": 12417
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4421330690383911,
      "learning_rate": 1.2777856900252801e-05,
      "loss": 0.5236,
      "step": 12418
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.3885279893875122,
      "learning_rate": 1.2776784253845943e-05,
      "loss": 0.5812,
      "step": 12419
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.502480149269104,
      "learning_rate": 1.2775711572819875e-05,
      "loss": 0.6697,
      "step": 12420
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.5106805562973022,
      "learning_rate": 1.2774638857187966e-05,
      "loss": 0.519,
      "step": 12421
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4993330240249634,
      "learning_rate": 1.2773566106963594e-05,
      "loss": 0.592,
      "step": 12422
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.436777949333191,
      "learning_rate": 1.2772493322160136e-05,
      "loss": 0.5933,
      "step": 12423
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.8645050525665283,
      "learning_rate": 1.2771420502790957e-05,
      "loss": 0.5209,
      "step": 12424
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4543943405151367,
      "learning_rate": 1.2770347648869444e-05,
      "loss": 0.5675,
      "step": 12425
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.9201198816299438,
      "learning_rate": 1.2769274760408963e-05,
      "loss": 0.6243,
      "step": 12426
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6249501705169678,
      "learning_rate": 1.2768201837422896e-05,
      "loss": 0.4019,
      "step": 12427
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5501259565353394,
      "learning_rate": 1.2767128879924617e-05,
      "loss": 0.7057,
      "step": 12428
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5560975074768066,
      "learning_rate": 1.2766055887927505e-05,
      "loss": 0.5867,
      "step": 12429
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8447864055633545,
      "learning_rate": 1.2764982861444935e-05,
      "loss": 0.5866,
      "step": 12430
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6272834539413452,
      "learning_rate": 1.2763909800490284e-05,
      "loss": 0.6704,
      "step": 12431
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3507941961288452,
      "learning_rate": 1.2762836705076935e-05,
      "loss": 0.5563,
      "step": 12432
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.9044151306152344,
      "learning_rate": 1.2761763575218264e-05,
      "loss": 0.6621,
      "step": 12433
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8801714181900024,
      "learning_rate": 1.2760690410927647e-05,
      "loss": 0.6189,
      "step": 12434
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4197545051574707,
      "learning_rate": 1.275961721221847e-05,
      "loss": 0.5223,
      "step": 12435
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6011933088302612,
      "learning_rate": 1.2758543979104107e-05,
      "loss": 0.5216,
      "step": 12436
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6047438383102417,
      "learning_rate": 1.2757470711597941e-05,
      "loss": 0.4652,
      "step": 12437
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8788827657699585,
      "learning_rate": 1.2756397409713354e-05,
      "loss": 0.489,
      "step": 12438
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.9639776945114136,
      "learning_rate": 1.2755324073463723e-05,
      "loss": 0.5215,
      "step": 12439
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7889565229415894,
      "learning_rate": 1.2754250702862435e-05,
      "loss": 0.5857,
      "step": 12440
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3226735591888428,
      "learning_rate": 1.2753177297922869e-05,
      "loss": 0.6293,
      "step": 12441
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.483223557472229,
      "learning_rate": 1.275210385865841e-05,
      "loss": 0.5381,
      "step": 12442
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.623923420906067,
      "learning_rate": 1.2751030385082439e-05,
      "loss": 0.4972,
      "step": 12443
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4890862703323364,
      "learning_rate": 1.2749956877208338e-05,
      "loss": 0.5205,
      "step": 12444
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5356346368789673,
      "learning_rate": 1.274888333504949e-05,
      "loss": 0.5981,
      "step": 12445
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6952083110809326,
      "learning_rate": 1.2747809758619287e-05,
      "loss": 0.5644,
      "step": 12446
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4924628734588623,
      "learning_rate": 1.2746736147931105e-05,
      "loss": 0.5989,
      "step": 12447
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6920944452285767,
      "learning_rate": 1.2745662502998335e-05,
      "loss": 0.647,
      "step": 12448
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4005916118621826,
      "learning_rate": 1.2744588823834359e-05,
      "loss": 0.5514,
      "step": 12449
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3373345136642456,
      "learning_rate": 1.2743515110452563e-05,
      "loss": 0.6377,
      "step": 12450
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2943165302276611,
      "learning_rate": 1.2742441362866335e-05,
      "loss": 0.4885,
      "step": 12451
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.576508641242981,
      "learning_rate": 1.2741367581089058e-05,
      "loss": 0.5761,
      "step": 12452
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3959959745407104,
      "learning_rate": 1.2740293765134126e-05,
      "loss": 0.478,
      "step": 12453
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6952229738235474,
      "learning_rate": 1.2739219915014924e-05,
      "loss": 0.4994,
      "step": 12454
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5405631065368652,
      "learning_rate": 1.2738146030744838e-05,
      "loss": 0.5044,
      "step": 12455
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6093827486038208,
      "learning_rate": 1.2737072112337259e-05,
      "loss": 0.4704,
      "step": 12456
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2825604677200317,
      "learning_rate": 1.2735998159805572e-05,
      "loss": 0.5214,
      "step": 12457
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6531658172607422,
      "learning_rate": 1.2734924173163168e-05,
      "loss": 0.513,
      "step": 12458
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.287926197052002,
      "learning_rate": 1.2733850152423442e-05,
      "loss": 0.5675,
      "step": 12459
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3235955238342285,
      "learning_rate": 1.2732776097599777e-05,
      "loss": 0.5903,
      "step": 12460
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4582207202911377,
      "learning_rate": 1.2731702008705572e-05,
      "loss": 0.5906,
      "step": 12461
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8808420896530151,
      "learning_rate": 1.2730627885754205e-05,
      "loss": 0.5638,
      "step": 12462
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6643128395080566,
      "learning_rate": 1.272955372875908e-05,
      "loss": 0.5837,
      "step": 12463
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5025442838668823,
      "learning_rate": 1.2728479537733582e-05,
      "loss": 0.5208,
      "step": 12464
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7724612951278687,
      "learning_rate": 1.2727405312691106e-05,
      "loss": 0.7045,
      "step": 12465
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6719728708267212,
      "learning_rate": 1.2726331053645045e-05,
      "loss": 0.4997,
      "step": 12466
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5148276090621948,
      "learning_rate": 1.2725256760608791e-05,
      "loss": 0.4213,
      "step": 12467
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4598993062973022,
      "learning_rate": 1.2724182433595737e-05,
      "loss": 0.5088,
      "step": 12468
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5126724243164062,
      "learning_rate": 1.2723108072619281e-05,
      "loss": 0.5412,
      "step": 12469
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7144980430603027,
      "learning_rate": 1.272203367769281e-05,
      "loss": 0.637,
      "step": 12470
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5136946439743042,
      "learning_rate": 1.2720959248829727e-05,
      "loss": 0.5613,
      "step": 12471
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5761942863464355,
      "learning_rate": 1.2719884786043425e-05,
      "loss": 0.5241,
      "step": 12472
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4371596574783325,
      "learning_rate": 1.2718810289347297e-05,
      "loss": 0.519,
      "step": 12473
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.43729567527771,
      "learning_rate": 1.271773575875474e-05,
      "loss": 0.5698,
      "step": 12474
    },
    {
      "epoch": 1.29,
      "grad_norm": 2.013697385787964,
      "learning_rate": 1.2716661194279153e-05,
      "loss": 0.6749,
      "step": 12475
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6525026559829712,
      "learning_rate": 1.2715586595933927e-05,
      "loss": 0.4094,
      "step": 12476
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.480474591255188,
      "learning_rate": 1.2714511963732468e-05,
      "loss": 0.7058,
      "step": 12477
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3765909671783447,
      "learning_rate": 1.2713437297688166e-05,
      "loss": 0.6252,
      "step": 12478
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3719733953475952,
      "learning_rate": 1.2712362597814426e-05,
      "loss": 0.5222,
      "step": 12479
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.593347430229187,
      "learning_rate": 1.2711287864124642e-05,
      "loss": 0.6168,
      "step": 12480
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6952489614486694,
      "learning_rate": 1.2710213096632213e-05,
      "loss": 0.592,
      "step": 12481
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5457842350006104,
      "learning_rate": 1.2709138295350545e-05,
      "loss": 0.5817,
      "step": 12482
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.745124101638794,
      "learning_rate": 1.2708063460293028e-05,
      "loss": 0.4383,
      "step": 12483
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5952072143554688,
      "learning_rate": 1.2706988591473069e-05,
      "loss": 0.6137,
      "step": 12484
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4360653162002563,
      "learning_rate": 1.2705913688904068e-05,
      "loss": 0.5599,
      "step": 12485
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.862645149230957,
      "learning_rate": 1.2704838752599424e-05,
      "loss": 0.5695,
      "step": 12486
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4410710334777832,
      "learning_rate": 1.2703763782572543e-05,
      "loss": 0.5838,
      "step": 12487
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6235209703445435,
      "learning_rate": 1.270268877883682e-05,
      "loss": 0.5557,
      "step": 12488
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8847544193267822,
      "learning_rate": 1.2701613741405663e-05,
      "loss": 0.4962,
      "step": 12489
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5886516571044922,
      "learning_rate": 1.2700538670292474e-05,
      "loss": 0.5007,
      "step": 12490
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4407849311828613,
      "learning_rate": 1.2699463565510658e-05,
      "loss": 0.5451,
      "step": 12491
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3265408277511597,
      "learning_rate": 1.2698388427073611e-05,
      "loss": 0.5363,
      "step": 12492
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3782601356506348,
      "learning_rate": 1.2697313254994747e-05,
      "loss": 0.5124,
      "step": 12493
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4850702285766602,
      "learning_rate": 1.2696238049287466e-05,
      "loss": 0.4961,
      "step": 12494
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3612133264541626,
      "learning_rate": 1.2695162809965174e-05,
      "loss": 0.5426,
      "step": 12495
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5943927764892578,
      "learning_rate": 1.2694087537041273e-05,
      "loss": 0.5155,
      "step": 12496
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3215903043746948,
      "learning_rate": 1.2693012230529171e-05,
      "loss": 0.5711,
      "step": 12497
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.9373786449432373,
      "learning_rate": 1.2691936890442278e-05,
      "loss": 0.5361,
      "step": 12498
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.9317058324813843,
      "learning_rate": 1.2690861516793996e-05,
      "loss": 0.6075,
      "step": 12499
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4323304891586304,
      "learning_rate": 1.2689786109597732e-05,
      "loss": 0.6118,
      "step": 12500
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8348125219345093,
      "learning_rate": 1.2688710668866897e-05,
      "loss": 0.5941,
      "step": 12501
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.3850789070129395,
      "learning_rate": 1.2687635194614896e-05,
      "loss": 0.5592,
      "step": 12502
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2859312295913696,
      "learning_rate": 1.268655968685514e-05,
      "loss": 0.4719,
      "step": 12503
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2970249652862549,
      "learning_rate": 1.2685484145601032e-05,
      "loss": 0.5455,
      "step": 12504
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.6334038972854614,
      "learning_rate": 1.2684408570865987e-05,
      "loss": 0.5405,
      "step": 12505
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.7335830926895142,
      "learning_rate": 1.2683332962663416e-05,
      "loss": 0.5253,
      "step": 12506
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.587679386138916,
      "learning_rate": 1.2682257321006721e-05,
      "loss": 0.5562,
      "step": 12507
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.682215929031372,
      "learning_rate": 1.2681181645909322e-05,
      "loss": 0.618,
      "step": 12508
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8748406171798706,
      "learning_rate": 1.2680105937384621e-05,
      "loss": 0.6319,
      "step": 12509
    },
    {
      "epoch": 1.29,
      "grad_norm": 2.3008358478546143,
      "learning_rate": 1.2679030195446034e-05,
      "loss": 0.5507,
      "step": 12510
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2995948791503906,
      "learning_rate": 1.2677954420106975e-05,
      "loss": 0.4491,
      "step": 12511
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5399994850158691,
      "learning_rate": 1.267687861138085e-05,
      "loss": 0.564,
      "step": 12512
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5205035209655762,
      "learning_rate": 1.2675802769281076e-05,
      "loss": 0.6261,
      "step": 12513
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.556679606437683,
      "learning_rate": 1.2674726893821064e-05,
      "loss": 0.5748,
      "step": 12514
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4213248491287231,
      "learning_rate": 1.2673650985014225e-05,
      "loss": 0.5748,
      "step": 12515
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.221956729888916,
      "learning_rate": 1.2672575042873982e-05,
      "loss": 0.4622,
      "step": 12516
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.8384088277816772,
      "learning_rate": 1.2671499067413739e-05,
      "loss": 0.597,
      "step": 12517
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4542651176452637,
      "learning_rate": 1.2670423058646914e-05,
      "loss": 0.497,
      "step": 12518
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.4438953399658203,
      "learning_rate": 1.2669347016586924e-05,
      "loss": 0.5373,
      "step": 12519
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5878559350967407,
      "learning_rate": 1.266827094124718e-05,
      "loss": 0.5855,
      "step": 12520
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.5039153099060059,
      "learning_rate": 1.2667194832641105e-05,
      "loss": 0.5668,
      "step": 12521
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8573803901672363,
      "learning_rate": 1.2666118690782109e-05,
      "loss": 0.4956,
      "step": 12522
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.641326665878296,
      "learning_rate": 1.266504251568361e-05,
      "loss": 0.5505,
      "step": 12523
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6494503021240234,
      "learning_rate": 1.2663966307359025e-05,
      "loss": 0.6608,
      "step": 12524
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5219072103500366,
      "learning_rate": 1.2662890065821773e-05,
      "loss": 0.5345,
      "step": 12525
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5689234733581543,
      "learning_rate": 1.2661813791085273e-05,
      "loss": 0.5203,
      "step": 12526
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3983031511306763,
      "learning_rate": 1.266073748316294e-05,
      "loss": 0.5463,
      "step": 12527
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5583122968673706,
      "learning_rate": 1.2659661142068193e-05,
      "loss": 0.4674,
      "step": 12528
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5127761363983154,
      "learning_rate": 1.2658584767814454e-05,
      "loss": 0.5098,
      "step": 12529
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3535360097885132,
      "learning_rate": 1.2657508360415139e-05,
      "loss": 0.5332,
      "step": 12530
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5974992513656616,
      "learning_rate": 1.2656431919883669e-05,
      "loss": 0.5979,
      "step": 12531
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8076703548431396,
      "learning_rate": 1.2655355446233468e-05,
      "loss": 0.6034,
      "step": 12532
    },
    {
      "epoch": 1.3,
      "grad_norm": 2.0438880920410156,
      "learning_rate": 1.2654278939477954e-05,
      "loss": 0.6833,
      "step": 12533
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4289517402648926,
      "learning_rate": 1.265320239963055e-05,
      "loss": 0.6072,
      "step": 12534
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.285151481628418,
      "learning_rate": 1.265212582670467e-05,
      "loss": 0.5984,
      "step": 12535
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4095447063446045,
      "learning_rate": 1.2651049220713745e-05,
      "loss": 0.5878,
      "step": 12536
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3781328201293945,
      "learning_rate": 1.2649972581671195e-05,
      "loss": 0.4919,
      "step": 12537
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.2764551639556885,
      "learning_rate": 1.264889590959044e-05,
      "loss": 0.533,
      "step": 12538
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.531476616859436,
      "learning_rate": 1.2647819204484909e-05,
      "loss": 0.5574,
      "step": 12539
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8201035261154175,
      "learning_rate": 1.2646742466368018e-05,
      "loss": 0.7273,
      "step": 12540
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4052162170410156,
      "learning_rate": 1.2645665695253197e-05,
      "loss": 0.5743,
      "step": 12541
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6457092761993408,
      "learning_rate": 1.264458889115387e-05,
      "loss": 0.4427,
      "step": 12542
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.396881103515625,
      "learning_rate": 1.264351205408346e-05,
      "loss": 0.5,
      "step": 12543
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.2973397970199585,
      "learning_rate": 1.2642435184055391e-05,
      "loss": 0.5404,
      "step": 12544
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5459758043289185,
      "learning_rate": 1.2641358281083093e-05,
      "loss": 0.7057,
      "step": 12545
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4624402523040771,
      "learning_rate": 1.264028134517999e-05,
      "loss": 0.534,
      "step": 12546
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7234140634536743,
      "learning_rate": 1.2639204376359509e-05,
      "loss": 0.5979,
      "step": 12547
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5212011337280273,
      "learning_rate": 1.2638127374635074e-05,
      "loss": 0.5515,
      "step": 12548
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.744110107421875,
      "learning_rate": 1.2637050340020116e-05,
      "loss": 0.6244,
      "step": 12549
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4084538221359253,
      "learning_rate": 1.2635973272528062e-05,
      "loss": 0.5169,
      "step": 12550
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5358332395553589,
      "learning_rate": 1.2634896172172339e-05,
      "loss": 0.4996,
      "step": 12551
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4554038047790527,
      "learning_rate": 1.263381903896638e-05,
      "loss": 0.5466,
      "step": 12552
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4720439910888672,
      "learning_rate": 1.2632741872923606e-05,
      "loss": 0.4903,
      "step": 12553
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7582049369812012,
      "learning_rate": 1.2631664674057452e-05,
      "loss": 0.5374,
      "step": 12554
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.540428638458252,
      "learning_rate": 1.2630587442381348e-05,
      "loss": 0.5147,
      "step": 12555
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7964754104614258,
      "learning_rate": 1.2629510177908722e-05,
      "loss": 0.4746,
      "step": 12556
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.386803150177002,
      "learning_rate": 1.2628432880653003e-05,
      "loss": 0.4871,
      "step": 12557
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5263264179229736,
      "learning_rate": 1.262735555062763e-05,
      "loss": 0.5663,
      "step": 12558
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3458576202392578,
      "learning_rate": 1.2626278187846025e-05,
      "loss": 0.5424,
      "step": 12559
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.435595154762268,
      "learning_rate": 1.2625200792321628e-05,
      "loss": 0.648,
      "step": 12560
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.950862169265747,
      "learning_rate": 1.2624123364067864e-05,
      "loss": 0.5195,
      "step": 12561
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.9933828115463257,
      "learning_rate": 1.262304590309817e-05,
      "loss": 0.5445,
      "step": 12562
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4431788921356201,
      "learning_rate": 1.262196840942598e-05,
      "loss": 0.5846,
      "step": 12563
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3216272592544556,
      "learning_rate": 1.2620890883064725e-05,
      "loss": 0.5086,
      "step": 12564
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.913368582725525,
      "learning_rate": 1.2619813324027839e-05,
      "loss": 0.574,
      "step": 12565
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6389373540878296,
      "learning_rate": 1.261873573232876e-05,
      "loss": 0.5782,
      "step": 12566
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4576385021209717,
      "learning_rate": 1.2617658107980915e-05,
      "loss": 0.5953,
      "step": 12567
    },
    {
      "epoch": 1.3,
      "grad_norm": 2.057492256164551,
      "learning_rate": 1.2616580450997743e-05,
      "loss": 0.6527,
      "step": 12568
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6601946353912354,
      "learning_rate": 1.2615502761392687e-05,
      "loss": 0.6329,
      "step": 12569
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.50773286819458,
      "learning_rate": 1.261442503917917e-05,
      "loss": 0.5767,
      "step": 12570
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.463542103767395,
      "learning_rate": 1.2613347284370638e-05,
      "loss": 0.4628,
      "step": 12571
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3978300094604492,
      "learning_rate": 1.2612269496980522e-05,
      "loss": 0.5477,
      "step": 12572
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4886081218719482,
      "learning_rate": 1.2611191677022263e-05,
      "loss": 0.4857,
      "step": 12573
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5111743211746216,
      "learning_rate": 1.26101138245093e-05,
      "loss": 0.4766,
      "step": 12574
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8570934534072876,
      "learning_rate": 1.2609035939455065e-05,
      "loss": 0.6556,
      "step": 12575
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.451682686805725,
      "learning_rate": 1.2607958021873002e-05,
      "loss": 0.5175,
      "step": 12576
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5001332759857178,
      "learning_rate": 1.2606880071776543e-05,
      "loss": 0.6265,
      "step": 12577
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4303045272827148,
      "learning_rate": 1.2605802089179135e-05,
      "loss": 0.6019,
      "step": 12578
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4726200103759766,
      "learning_rate": 1.2604724074094214e-05,
      "loss": 0.5706,
      "step": 12579
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8562664985656738,
      "learning_rate": 1.2603646026535221e-05,
      "loss": 0.4722,
      "step": 12580
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6788324117660522,
      "learning_rate": 1.2602567946515593e-05,
      "loss": 0.5473,
      "step": 12581
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8024673461914062,
      "learning_rate": 1.2601489834048777e-05,
      "loss": 0.5578,
      "step": 12582
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.2619866132736206,
      "learning_rate": 1.260041168914821e-05,
      "loss": 0.5529,
      "step": 12583
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3855551481246948,
      "learning_rate": 1.2599333511827334e-05,
      "loss": 0.5148,
      "step": 12584
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6950446367263794,
      "learning_rate": 1.259825530209959e-05,
      "loss": 0.5903,
      "step": 12585
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.445507526397705,
      "learning_rate": 1.2597177059978422e-05,
      "loss": 0.551,
      "step": 12586
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3965178728103638,
      "learning_rate": 1.2596098785477278e-05,
      "loss": 0.5527,
      "step": 12587
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.914878010749817,
      "learning_rate": 1.2595020478609589e-05,
      "loss": 0.6477,
      "step": 12588
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4809120893478394,
      "learning_rate": 1.2593942139388812e-05,
      "loss": 0.4754,
      "step": 12589
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7793511152267456,
      "learning_rate": 1.259286376782838e-05,
      "loss": 0.5267,
      "step": 12590
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3978846073150635,
      "learning_rate": 1.2591785363941744e-05,
      "loss": 0.6152,
      "step": 12591
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.8589980602264404,
      "learning_rate": 1.2590706927742347e-05,
      "loss": 0.4586,
      "step": 12592
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.876495361328125,
      "learning_rate": 1.2589628459243634e-05,
      "loss": 0.6058,
      "step": 12593
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3556780815124512,
      "learning_rate": 1.258854995845905e-05,
      "loss": 0.5614,
      "step": 12594
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3568923473358154,
      "learning_rate": 1.2587471425402045e-05,
      "loss": 0.6284,
      "step": 12595
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.496232032775879,
      "learning_rate": 1.2586392860086059e-05,
      "loss": 0.5552,
      "step": 12596
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5189288854599,
      "learning_rate": 1.2585314262524546e-05,
      "loss": 0.4534,
      "step": 12597
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5872257947921753,
      "learning_rate": 1.2584235632730948e-05,
      "loss": 0.6765,
      "step": 12598
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.2945451736450195,
      "learning_rate": 1.2583156970718712e-05,
      "loss": 0.4919,
      "step": 12599
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6647435426712036,
      "learning_rate": 1.2582078276501292e-05,
      "loss": 0.6005,
      "step": 12600
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4351798295974731,
      "learning_rate": 1.2580999550092132e-05,
      "loss": 0.5506,
      "step": 12601
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4649559259414673,
      "learning_rate": 1.2579920791504684e-05,
      "loss": 0.581,
      "step": 12602
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3179755210876465,
      "learning_rate": 1.2578842000752391e-05,
      "loss": 0.5498,
      "step": 12603
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5127815008163452,
      "learning_rate": 1.2577763177848708e-05,
      "loss": 0.5244,
      "step": 12604
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4994291067123413,
      "learning_rate": 1.2576684322807086e-05,
      "loss": 0.5637,
      "step": 12605
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7917532920837402,
      "learning_rate": 1.2575605435640972e-05,
      "loss": 0.5783,
      "step": 12606
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4449121952056885,
      "learning_rate": 1.2574526516363816e-05,
      "loss": 0.514,
      "step": 12607
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5499346256256104,
      "learning_rate": 1.2573447564989076e-05,
      "loss": 0.5219,
      "step": 12608
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.457335352897644,
      "learning_rate": 1.2572368581530196e-05,
      "loss": 0.5622,
      "step": 12609
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4166792631149292,
      "learning_rate": 1.2571289566000635e-05,
      "loss": 0.5663,
      "step": 12610
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4948608875274658,
      "learning_rate": 1.257021051841384e-05,
      "loss": 0.4641,
      "step": 12611
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.5710376501083374,
      "learning_rate": 1.2569131438783264e-05,
      "loss": 0.5432,
      "step": 12612
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.692484736442566,
      "learning_rate": 1.2568052327122364e-05,
      "loss": 0.4796,
      "step": 12613
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.7378734350204468,
      "learning_rate": 1.2566973183444591e-05,
      "loss": 0.601,
      "step": 12614
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.6255645751953125,
      "learning_rate": 1.2565894007763402e-05,
      "loss": 0.6003,
      "step": 12615
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.405702829360962,
      "learning_rate": 1.2564814800092247e-05,
      "loss": 0.6478,
      "step": 12616
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.388088583946228,
      "learning_rate": 1.2563735560444581e-05,
      "loss": 0.4515,
      "step": 12617
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.536450743675232,
      "learning_rate": 1.2562656288833865e-05,
      "loss": 0.4536,
      "step": 12618
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6025722026824951,
      "learning_rate": 1.2561576985273552e-05,
      "loss": 0.4986,
      "step": 12619
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3995463848114014,
      "learning_rate": 1.2560497649777093e-05,
      "loss": 0.4953,
      "step": 12620
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.308457374572754,
      "learning_rate": 1.2559418282357953e-05,
      "loss": 0.4645,
      "step": 12621
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.9131485223770142,
      "learning_rate": 1.2558338883029581e-05,
      "loss": 0.4266,
      "step": 12622
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7438488006591797,
      "learning_rate": 1.2557259451805442e-05,
      "loss": 0.6667,
      "step": 12623
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5373814105987549,
      "learning_rate": 1.2556179988698987e-05,
      "loss": 0.5801,
      "step": 12624
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.307098388671875,
      "learning_rate": 1.2555100493723678e-05,
      "loss": 0.4624,
      "step": 12625
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7857089042663574,
      "learning_rate": 1.255402096689297e-05,
      "loss": 0.4981,
      "step": 12626
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7114458084106445,
      "learning_rate": 1.2552941408220324e-05,
      "loss": 0.5397,
      "step": 12627
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.668617606163025,
      "learning_rate": 1.2551861817719205e-05,
      "loss": 0.5415,
      "step": 12628
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8529109954833984,
      "learning_rate": 1.255078219540306e-05,
      "loss": 0.6,
      "step": 12629
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.612058401107788,
      "learning_rate": 1.2549702541285358e-05,
      "loss": 0.577,
      "step": 12630
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5234642028808594,
      "learning_rate": 1.2548622855379558e-05,
      "loss": 0.4677,
      "step": 12631
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3943341970443726,
      "learning_rate": 1.2547543137699118e-05,
      "loss": 0.5148,
      "step": 12632
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.62492036819458,
      "learning_rate": 1.2546463388257504e-05,
      "loss": 0.4433,
      "step": 12633
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6271564960479736,
      "learning_rate": 1.2545383607068175e-05,
      "loss": 0.6536,
      "step": 12634
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3814400434494019,
      "learning_rate": 1.2544303794144592e-05,
      "loss": 0.5059,
      "step": 12635
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4370421171188354,
      "learning_rate": 1.2543223949500222e-05,
      "loss": 0.5701,
      "step": 12636
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3457543849945068,
      "learning_rate": 1.254214407314852e-05,
      "loss": 0.4363,
      "step": 12637
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.526653528213501,
      "learning_rate": 1.2541064165102956e-05,
      "loss": 0.5736,
      "step": 12638
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6610161066055298,
      "learning_rate": 1.2539984225376993e-05,
      "loss": 0.5234,
      "step": 12639
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6341341733932495,
      "learning_rate": 1.253890425398409e-05,
      "loss": 0.488,
      "step": 12640
    },
    {
      "epoch": 1.31,
      "grad_norm": 2.2235429286956787,
      "learning_rate": 1.2537824250937716e-05,
      "loss": 0.6363,
      "step": 12641
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.552501916885376,
      "learning_rate": 1.2536744216251335e-05,
      "loss": 0.5856,
      "step": 12642
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4822403192520142,
      "learning_rate": 1.2535664149938409e-05,
      "loss": 0.6219,
      "step": 12643
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5361080169677734,
      "learning_rate": 1.253458405201241e-05,
      "loss": 0.6108,
      "step": 12644
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4583557844161987,
      "learning_rate": 1.25335039224868e-05,
      "loss": 0.5411,
      "step": 12645
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5639512538909912,
      "learning_rate": 1.2532423761375043e-05,
      "loss": 0.5493,
      "step": 12646
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7784732580184937,
      "learning_rate": 1.253134356869061e-05,
      "loss": 0.6476,
      "step": 12647
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5429699420928955,
      "learning_rate": 1.2530263344446968e-05,
      "loss": 0.6404,
      "step": 12648
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3558815717697144,
      "learning_rate": 1.2529183088657582e-05,
      "loss": 0.5327,
      "step": 12649
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3189266920089722,
      "learning_rate": 1.252810280133592e-05,
      "loss": 0.5406,
      "step": 12650
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.2316646575927734,
      "learning_rate": 1.2527022482495453e-05,
      "loss": 0.4478,
      "step": 12651
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5415279865264893,
      "learning_rate": 1.2525942132149647e-05,
      "loss": 0.5916,
      "step": 12652
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5538487434387207,
      "learning_rate": 1.2524861750311974e-05,
      "loss": 0.6193,
      "step": 12653
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5020745992660522,
      "learning_rate": 1.2523781336995902e-05,
      "loss": 0.5397,
      "step": 12654
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5672671794891357,
      "learning_rate": 1.2522700892214899e-05,
      "loss": 0.6718,
      "step": 12655
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8078243732452393,
      "learning_rate": 1.2521620415982436e-05,
      "loss": 0.6435,
      "step": 12656
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5799248218536377,
      "learning_rate": 1.2520539908311988e-05,
      "loss": 0.5894,
      "step": 12657
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6748533248901367,
      "learning_rate": 1.2519459369217022e-05,
      "loss": 0.5904,
      "step": 12658
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3862690925598145,
      "learning_rate": 1.2518378798711009e-05,
      "loss": 0.421,
      "step": 12659
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.2957682609558105,
      "learning_rate": 1.2517298196807423e-05,
      "loss": 0.5512,
      "step": 12660
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5427653789520264,
      "learning_rate": 1.2516217563519738e-05,
      "loss": 0.5873,
      "step": 12661
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5061522722244263,
      "learning_rate": 1.2515136898861421e-05,
      "loss": 0.5414,
      "step": 12662
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3877289295196533,
      "learning_rate": 1.2514056202845948e-05,
      "loss": 0.5341,
      "step": 12663
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6197901964187622,
      "learning_rate": 1.2512975475486795e-05,
      "loss": 0.6201,
      "step": 12664
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.2920629978179932,
      "learning_rate": 1.2511894716797432e-05,
      "loss": 0.4588,
      "step": 12665
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3749971389770508,
      "learning_rate": 1.2510813926791334e-05,
      "loss": 0.4606,
      "step": 12666
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.671097755432129,
      "learning_rate": 1.250973310548198e-05,
      "loss": 0.4716,
      "step": 12667
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.412528157234192,
      "learning_rate": 1.2508652252882838e-05,
      "loss": 0.5565,
      "step": 12668
    },
    {
      "epoch": 1.31,
      "grad_norm": 2.0581328868865967,
      "learning_rate": 1.2507571369007387e-05,
      "loss": 0.6333,
      "step": 12669
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4093257188796997,
      "learning_rate": 1.2506490453869105e-05,
      "loss": 0.4679,
      "step": 12670
    },
    {
      "epoch": 1.31,
      "grad_norm": 2.200328826904297,
      "learning_rate": 1.2505409507481462e-05,
      "loss": 0.674,
      "step": 12671
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.460798978805542,
      "learning_rate": 1.2504328529857941e-05,
      "loss": 0.557,
      "step": 12672
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.9651886224746704,
      "learning_rate": 1.2503247521012017e-05,
      "loss": 0.6384,
      "step": 12673
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5394545793533325,
      "learning_rate": 1.2502166480957163e-05,
      "loss": 0.6593,
      "step": 12674
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5421998500823975,
      "learning_rate": 1.2501085409706865e-05,
      "loss": 0.6424,
      "step": 12675
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.494011402130127,
      "learning_rate": 1.2500004307274593e-05,
      "loss": 0.5553,
      "step": 12676
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4192924499511719,
      "learning_rate": 1.2498923173673829e-05,
      "loss": 0.5238,
      "step": 12677
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.582535982131958,
      "learning_rate": 1.2497842008918055e-05,
      "loss": 0.4979,
      "step": 12678
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5942645072937012,
      "learning_rate": 1.2496760813020743e-05,
      "loss": 0.5804,
      "step": 12679
    },
    {
      "epoch": 1.31,
      "grad_norm": 4.683009624481201,
      "learning_rate": 1.249567958599538e-05,
      "loss": 0.5351,
      "step": 12680
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5564614534378052,
      "learning_rate": 1.249459832785544e-05,
      "loss": 0.468,
      "step": 12681
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7181504964828491,
      "learning_rate": 1.2493517038614409e-05,
      "loss": 0.5541,
      "step": 12682
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5199617147445679,
      "learning_rate": 1.2492435718285765e-05,
      "loss": 0.582,
      "step": 12683
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.53994619846344,
      "learning_rate": 1.2491354366882988e-05,
      "loss": 0.484,
      "step": 12684
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.652772307395935,
      "learning_rate": 1.249027298441956e-05,
      "loss": 0.5275,
      "step": 12685
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.661867380142212,
      "learning_rate": 1.2489191570908967e-05,
      "loss": 0.5776,
      "step": 12686
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7142246961593628,
      "learning_rate": 1.2488110126364687e-05,
      "loss": 0.5631,
      "step": 12687
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7831507921218872,
      "learning_rate": 1.2487028650800205e-05,
      "loss": 0.6205,
      "step": 12688
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5746431350708008,
      "learning_rate": 1.2485947144229003e-05,
      "loss": 0.5897,
      "step": 12689
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4579720497131348,
      "learning_rate": 1.2484865606664562e-05,
      "loss": 0.4496,
      "step": 12690
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3896055221557617,
      "learning_rate": 1.2483784038120372e-05,
      "loss": 0.5373,
      "step": 12691
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.725273847579956,
      "learning_rate": 1.2482702438609913e-05,
      "loss": 0.4743,
      "step": 12692
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3683301210403442,
      "learning_rate": 1.2481620808146673e-05,
      "loss": 0.6401,
      "step": 12693
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7762495279312134,
      "learning_rate": 1.2480539146744133e-05,
      "loss": 0.5017,
      "step": 12694
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.9329332113265991,
      "learning_rate": 1.247945745441578e-05,
      "loss": 0.5507,
      "step": 12695
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5067857503890991,
      "learning_rate": 1.24783757311751e-05,
      "loss": 0.5179,
      "step": 12696
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3802446126937866,
      "learning_rate": 1.247729397703558e-05,
      "loss": 0.4353,
      "step": 12697
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7232341766357422,
      "learning_rate": 1.2476212192010706e-05,
      "loss": 0.4697,
      "step": 12698
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6008307933807373,
      "learning_rate": 1.247513037611397e-05,
      "loss": 0.4632,
      "step": 12699
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5068926811218262,
      "learning_rate": 1.2474048529358847e-05,
      "loss": 0.563,
      "step": 12700
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5851784944534302,
      "learning_rate": 1.2472966651758838e-05,
      "loss": 0.6117,
      "step": 12701
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7820382118225098,
      "learning_rate": 1.2471884743327423e-05,
      "loss": 0.6345,
      "step": 12702
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5326164960861206,
      "learning_rate": 1.247080280407809e-05,
      "loss": 0.56,
      "step": 12703
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3655511140823364,
      "learning_rate": 1.2469720834024337e-05,
      "loss": 0.5748,
      "step": 12704
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5771416425704956,
      "learning_rate": 1.2468638833179646e-05,
      "loss": 0.6085,
      "step": 12705
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5504101514816284,
      "learning_rate": 1.2467556801557507e-05,
      "loss": 0.6432,
      "step": 12706
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4816536903381348,
      "learning_rate": 1.246647473917141e-05,
      "loss": 0.6249,
      "step": 12707
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7342581748962402,
      "learning_rate": 1.2465392646034847e-05,
      "loss": 0.5339,
      "step": 12708
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.7948898077011108,
      "learning_rate": 1.2464310522161312e-05,
      "loss": 0.5449,
      "step": 12709
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3162391185760498,
      "learning_rate": 1.2463228367564289e-05,
      "loss": 0.4604,
      "step": 12710
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5603227615356445,
      "learning_rate": 1.2462146182257273e-05,
      "loss": 0.7054,
      "step": 12711
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5814365148544312,
      "learning_rate": 1.2461063966253758e-05,
      "loss": 0.5667,
      "step": 12712
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4227527379989624,
      "learning_rate": 1.2459981719567234e-05,
      "loss": 0.4335,
      "step": 12713
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.5861380100250244,
      "learning_rate": 1.2458899442211196e-05,
      "loss": 0.5101,
      "step": 12714
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4038017988204956,
      "learning_rate": 1.2457817134199134e-05,
      "loss": 0.6022,
      "step": 12715
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6536989212036133,
      "learning_rate": 1.2456734795544545e-05,
      "loss": 0.6329,
      "step": 12716
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5465047359466553,
      "learning_rate": 1.2455652426260922e-05,
      "loss": 0.6685,
      "step": 12717
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.583635926246643,
      "learning_rate": 1.2454570026361756e-05,
      "loss": 0.4232,
      "step": 12718
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.494676947593689,
      "learning_rate": 1.2453487595860548e-05,
      "loss": 0.4329,
      "step": 12719
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4474742412567139,
      "learning_rate": 1.2452405134770784e-05,
      "loss": 0.5841,
      "step": 12720
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3882085084915161,
      "learning_rate": 1.2451322643105968e-05,
      "loss": 0.5079,
      "step": 12721
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8147568702697754,
      "learning_rate": 1.2450240120879592e-05,
      "loss": 0.5899,
      "step": 12722
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8563644886016846,
      "learning_rate": 1.2449157568105152e-05,
      "loss": 0.6915,
      "step": 12723
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6170717477798462,
      "learning_rate": 1.2448074984796143e-05,
      "loss": 0.6636,
      "step": 12724
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8454935550689697,
      "learning_rate": 1.2446992370966071e-05,
      "loss": 0.5296,
      "step": 12725
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7048553228378296,
      "learning_rate": 1.244590972662842e-05,
      "loss": 0.5519,
      "step": 12726
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.535386085510254,
      "learning_rate": 1.24448270517967e-05,
      "loss": 0.5921,
      "step": 12727
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.48167884349823,
      "learning_rate": 1.2443744346484398e-05,
      "loss": 0.4977,
      "step": 12728
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1639482975006104,
      "learning_rate": 1.2442661610705021e-05,
      "loss": 0.5428,
      "step": 12729
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4444102048873901,
      "learning_rate": 1.2441578844472066e-05,
      "loss": 0.4872,
      "step": 12730
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6593680381774902,
      "learning_rate": 1.244049604779903e-05,
      "loss": 0.5634,
      "step": 12731
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2129912376403809,
      "learning_rate": 1.2439413220699414e-05,
      "loss": 0.5304,
      "step": 12732
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4965386390686035,
      "learning_rate": 1.2438330363186718e-05,
      "loss": 0.5212,
      "step": 12733
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5153579711914062,
      "learning_rate": 1.243724747527444e-05,
      "loss": 0.465,
      "step": 12734
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6604764461517334,
      "learning_rate": 1.2436164556976085e-05,
      "loss": 0.4252,
      "step": 12735
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3418631553649902,
      "learning_rate": 1.2435081608305151e-05,
      "loss": 0.6455,
      "step": 12736
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.74847412109375,
      "learning_rate": 1.243399862927514e-05,
      "loss": 0.5729,
      "step": 12737
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6510547399520874,
      "learning_rate": 1.2432915619899557e-05,
      "loss": 0.6441,
      "step": 12738
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.638464093208313,
      "learning_rate": 1.2431832580191898e-05,
      "loss": 0.5693,
      "step": 12739
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3498404026031494,
      "learning_rate": 1.2430749510165674e-05,
      "loss": 0.5447,
      "step": 12740
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5714510679244995,
      "learning_rate": 1.2429666409834385e-05,
      "loss": 0.7273,
      "step": 12741
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4615427255630493,
      "learning_rate": 1.2428583279211526e-05,
      "loss": 0.4952,
      "step": 12742
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4691072702407837,
      "learning_rate": 1.2427500118310614e-05,
      "loss": 0.6318,
      "step": 12743
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.35648775100708,
      "learning_rate": 1.2426416927145141e-05,
      "loss": 0.4877,
      "step": 12744
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2500180006027222,
      "learning_rate": 1.2425333705728622e-05,
      "loss": 0.5286,
      "step": 12745
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6147453784942627,
      "learning_rate": 1.2424250454074555e-05,
      "loss": 0.414,
      "step": 12746
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5916080474853516,
      "learning_rate": 1.2423167172196448e-05,
      "loss": 0.5635,
      "step": 12747
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5309739112854004,
      "learning_rate": 1.242208386010781e-05,
      "loss": 0.564,
      "step": 12748
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6558856964111328,
      "learning_rate": 1.2421000517822139e-05,
      "loss": 0.5235,
      "step": 12749
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8605579137802124,
      "learning_rate": 1.2419917145352947e-05,
      "loss": 0.6073,
      "step": 12750
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.600948691368103,
      "learning_rate": 1.241883374271374e-05,
      "loss": 0.6103,
      "step": 12751
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.332486629486084,
      "learning_rate": 1.2417750309918024e-05,
      "loss": 0.4962,
      "step": 12752
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6072837114334106,
      "learning_rate": 1.2416666846979306e-05,
      "loss": 0.7262,
      "step": 12753
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4544614553451538,
      "learning_rate": 1.2415583353911098e-05,
      "loss": 0.509,
      "step": 12754
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7197633981704712,
      "learning_rate": 1.2414499830726904e-05,
      "loss": 0.5852,
      "step": 12755
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3527196645736694,
      "learning_rate": 1.2413416277440237e-05,
      "loss": 0.6395,
      "step": 12756
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1603202819824219,
      "learning_rate": 1.2412332694064601e-05,
      "loss": 0.5332,
      "step": 12757
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4514743089675903,
      "learning_rate": 1.2411249080613508e-05,
      "loss": 0.5419,
      "step": 12758
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.399775505065918,
      "learning_rate": 1.241016543710047e-05,
      "loss": 0.5622,
      "step": 12759
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3914482593536377,
      "learning_rate": 1.2409081763538993e-05,
      "loss": 0.57,
      "step": 12760
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5563620328903198,
      "learning_rate": 1.2407998059942592e-05,
      "loss": 0.602,
      "step": 12761
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.638273000717163,
      "learning_rate": 1.240691432632477e-05,
      "loss": 0.6177,
      "step": 12762
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8126506805419922,
      "learning_rate": 1.2405830562699048e-05,
      "loss": 0.5416,
      "step": 12763
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6096038818359375,
      "learning_rate": 1.2404746769078932e-05,
      "loss": 0.5638,
      "step": 12764
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5009819269180298,
      "learning_rate": 1.2403662945477936e-05,
      "loss": 0.6142,
      "step": 12765
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2843831777572632,
      "learning_rate": 1.240257909190957e-05,
      "loss": 0.5696,
      "step": 12766
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4998970031738281,
      "learning_rate": 1.240149520838735e-05,
      "loss": 0.5329,
      "step": 12767
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3552144765853882,
      "learning_rate": 1.2400411294924788e-05,
      "loss": 0.518,
      "step": 12768
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7546461820602417,
      "learning_rate": 1.23993273515354e-05,
      "loss": 0.6797,
      "step": 12769
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6133594512939453,
      "learning_rate": 1.2398243378232693e-05,
      "loss": 0.6307,
      "step": 12770
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4491539001464844,
      "learning_rate": 1.2397159375030188e-05,
      "loss": 0.584,
      "step": 12771
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.454845905303955,
      "learning_rate": 1.2396075341941398e-05,
      "loss": 0.4249,
      "step": 12772
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.477432370185852,
      "learning_rate": 1.2394991278979835e-05,
      "loss": 0.4609,
      "step": 12773
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3196299076080322,
      "learning_rate": 1.2393907186159018e-05,
      "loss": 0.4334,
      "step": 12774
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3866329193115234,
      "learning_rate": 1.239282306349246e-05,
      "loss": 0.5278,
      "step": 12775
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5072484016418457,
      "learning_rate": 1.239173891099368e-05,
      "loss": 0.5222,
      "step": 12776
    },
    {
      "epoch": 1.32,
      "grad_norm": 2.105132818222046,
      "learning_rate": 1.2390654728676197e-05,
      "loss": 0.6288,
      "step": 12777
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.9247640371322632,
      "learning_rate": 1.238957051655352e-05,
      "loss": 0.7343,
      "step": 12778
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4531288146972656,
      "learning_rate": 1.238848627463917e-05,
      "loss": 0.5231,
      "step": 12779
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.668931484222412,
      "learning_rate": 1.2387402002946667e-05,
      "loss": 0.4985,
      "step": 12780
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3550251722335815,
      "learning_rate": 1.2386317701489526e-05,
      "loss": 0.4897,
      "step": 12781
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5442543029785156,
      "learning_rate": 1.238523337028127e-05,
      "loss": 0.449,
      "step": 12782
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4319989681243896,
      "learning_rate": 1.2384149009335409e-05,
      "loss": 0.4434,
      "step": 12783
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6169508695602417,
      "learning_rate": 1.2383064618665468e-05,
      "loss": 0.5036,
      "step": 12784
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2405799627304077,
      "learning_rate": 1.2381980198284971e-05,
      "loss": 0.5462,
      "step": 12785
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4617092609405518,
      "learning_rate": 1.2380895748207426e-05,
      "loss": 0.5491,
      "step": 12786
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4494149684906006,
      "learning_rate": 1.2379811268446367e-05,
      "loss": 0.5793,
      "step": 12787
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.614464282989502,
      "learning_rate": 1.2378726759015302e-05,
      "loss": 0.6331,
      "step": 12788
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3946908712387085,
      "learning_rate": 1.2377642219927761e-05,
      "loss": 0.7017,
      "step": 12789
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4962594509124756,
      "learning_rate": 1.2376557651197261e-05,
      "loss": 0.5199,
      "step": 12790
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6366753578186035,
      "learning_rate": 1.2375473052837326e-05,
      "loss": 0.5895,
      "step": 12791
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.8293628692626953,
      "learning_rate": 1.2374388424861474e-05,
      "loss": 0.5805,
      "step": 12792
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3405171632766724,
      "learning_rate": 1.2373303767283234e-05,
      "loss": 0.4865,
      "step": 12793
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.5387425422668457,
      "learning_rate": 1.2372219080116122e-05,
      "loss": 0.6171,
      "step": 12794
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6733309030532837,
      "learning_rate": 1.2371134363373667e-05,
      "loss": 0.4959,
      "step": 12795
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.488752007484436,
      "learning_rate": 1.2370049617069386e-05,
      "loss": 0.5568,
      "step": 12796
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.9221285581588745,
      "learning_rate": 1.2368964841216809e-05,
      "loss": 0.5627,
      "step": 12797
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2833870649337769,
      "learning_rate": 1.236788003582946e-05,
      "loss": 0.5268,
      "step": 12798
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6523635387420654,
      "learning_rate": 1.236679520092086e-05,
      "loss": 0.6166,
      "step": 12799
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4980053901672363,
      "learning_rate": 1.2365710336504538e-05,
      "loss": 0.511,
      "step": 12800
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7569961547851562,
      "learning_rate": 1.2364625442594014e-05,
      "loss": 0.5719,
      "step": 12801
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6161237955093384,
      "learning_rate": 1.2363540519202817e-05,
      "loss": 0.6539,
      "step": 12802
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4646084308624268,
      "learning_rate": 1.2362455566344476e-05,
      "loss": 0.5948,
      "step": 12803
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4880071878433228,
      "learning_rate": 1.2361370584032513e-05,
      "loss": 0.5864,
      "step": 12804
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2375497817993164,
      "learning_rate": 1.2360285572280457e-05,
      "loss": 0.5283,
      "step": 12805
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.6760774850845337,
      "learning_rate": 1.2359200531101837e-05,
      "loss": 0.5384,
      "step": 12806
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.575046420097351,
      "learning_rate": 1.2358115460510174e-05,
      "loss": 0.6636,
      "step": 12807
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.683032751083374,
      "learning_rate": 1.2357030360519004e-05,
      "loss": 0.6155,
      "step": 12808
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4753670692443848,
      "learning_rate": 1.235594523114185e-05,
      "loss": 0.5351,
      "step": 12809
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7390304803848267,
      "learning_rate": 1.2354860072392243e-05,
      "loss": 0.6145,
      "step": 12810
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4572057723999023,
      "learning_rate": 1.2353774884283714e-05,
      "loss": 0.5682,
      "step": 12811
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6213147640228271,
      "learning_rate": 1.2352689666829787e-05,
      "loss": 0.6345,
      "step": 12812
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6268950700759888,
      "learning_rate": 1.2351604420043997e-05,
      "loss": 0.6029,
      "step": 12813
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4963411092758179,
      "learning_rate": 1.235051914393987e-05,
      "loss": 0.4806,
      "step": 12814
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4623163938522339,
      "learning_rate": 1.2349433838530938e-05,
      "loss": 0.5271,
      "step": 12815
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2707617282867432,
      "learning_rate": 1.2348348503830736e-05,
      "loss": 0.5086,
      "step": 12816
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4106357097625732,
      "learning_rate": 1.2347263139852788e-05,
      "loss": 0.5509,
      "step": 12817
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.0033788681030273,
      "learning_rate": 1.234617774661063e-05,
      "loss": 0.6037,
      "step": 12818
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5481832027435303,
      "learning_rate": 1.2345092324117795e-05,
      "loss": 0.5986,
      "step": 12819
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7790136337280273,
      "learning_rate": 1.2344006872387815e-05,
      "loss": 0.5583,
      "step": 12820
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.9306706190109253,
      "learning_rate": 1.2342921391434219e-05,
      "loss": 0.4707,
      "step": 12821
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4839969873428345,
      "learning_rate": 1.2341835881270543e-05,
      "loss": 0.6027,
      "step": 12822
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6988139152526855,
      "learning_rate": 1.2340750341910321e-05,
      "loss": 0.5522,
      "step": 12823
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.704931378364563,
      "learning_rate": 1.2339664773367086e-05,
      "loss": 0.6233,
      "step": 12824
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3775677680969238,
      "learning_rate": 1.233857917565437e-05,
      "loss": 0.5725,
      "step": 12825
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.1906180381774902,
      "learning_rate": 1.2337493548785715e-05,
      "loss": 0.6287,
      "step": 12826
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5066088438034058,
      "learning_rate": 1.2336407892774644e-05,
      "loss": 0.6402,
      "step": 12827
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4231234788894653,
      "learning_rate": 1.2335322207634701e-05,
      "loss": 0.5005,
      "step": 12828
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4365473985671997,
      "learning_rate": 1.2334236493379424e-05,
      "loss": 0.5464,
      "step": 12829
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.517271637916565,
      "learning_rate": 1.2333150750022337e-05,
      "loss": 0.495,
      "step": 12830
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5206665992736816,
      "learning_rate": 1.2332064977576989e-05,
      "loss": 0.5291,
      "step": 12831
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2825815677642822,
      "learning_rate": 1.2330979176056911e-05,
      "loss": 0.6132,
      "step": 12832
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4428939819335938,
      "learning_rate": 1.2329893345475638e-05,
      "loss": 0.5229,
      "step": 12833
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.9110392332077026,
      "learning_rate": 1.2328807485846715e-05,
      "loss": 0.5841,
      "step": 12834
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.711025595664978,
      "learning_rate": 1.232772159718367e-05,
      "loss": 0.6849,
      "step": 12835
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3441649675369263,
      "learning_rate": 1.232663567950005e-05,
      "loss": 0.591,
      "step": 12836
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.329272747039795,
      "learning_rate": 1.2325549732809387e-05,
      "loss": 0.5193,
      "step": 12837
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3845518827438354,
      "learning_rate": 1.2324463757125225e-05,
      "loss": 0.5166,
      "step": 12838
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.4181478023529053,
      "learning_rate": 1.23233777524611e-05,
      "loss": 0.6201,
      "step": 12839
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5485014915466309,
      "learning_rate": 1.2322291718830553e-05,
      "loss": 0.6374,
      "step": 12840
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.684441089630127,
      "learning_rate": 1.232120565624712e-05,
      "loss": 0.6123,
      "step": 12841
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4712597131729126,
      "learning_rate": 1.2320119564724348e-05,
      "loss": 0.5935,
      "step": 12842
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3547128438949585,
      "learning_rate": 1.2319033444275774e-05,
      "loss": 0.5745,
      "step": 12843
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.340470552444458,
      "learning_rate": 1.2317947294914939e-05,
      "loss": 0.5353,
      "step": 12844
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2659589052200317,
      "learning_rate": 1.2316861116655387e-05,
      "loss": 0.4488,
      "step": 12845
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4957352876663208,
      "learning_rate": 1.2315774909510655e-05,
      "loss": 0.4275,
      "step": 12846
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4989383220672607,
      "learning_rate": 1.231468867349429e-05,
      "loss": 0.5279,
      "step": 12847
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4255670309066772,
      "learning_rate": 1.2313602408619833e-05,
      "loss": 0.5044,
      "step": 12848
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.732867956161499,
      "learning_rate": 1.2312516114900826e-05,
      "loss": 0.5143,
      "step": 12849
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.9375115633010864,
      "learning_rate": 1.231142979235081e-05,
      "loss": 0.5866,
      "step": 12850
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.413326382637024,
      "learning_rate": 1.2310343440983334e-05,
      "loss": 0.5676,
      "step": 12851
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7428895235061646,
      "learning_rate": 1.2309257060811941e-05,
      "loss": 0.4454,
      "step": 12852
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4749131202697754,
      "learning_rate": 1.230817065185017e-05,
      "loss": 0.6134,
      "step": 12853
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3290162086486816,
      "learning_rate": 1.230708421411157e-05,
      "loss": 0.5932,
      "step": 12854
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4284828901290894,
      "learning_rate": 1.2305997747609686e-05,
      "loss": 0.4595,
      "step": 12855
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2987548112869263,
      "learning_rate": 1.230491125235806e-05,
      "loss": 0.6276,
      "step": 12856
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6389721632003784,
      "learning_rate": 1.2303824728370241e-05,
      "loss": 0.5888,
      "step": 12857
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.285840392112732,
      "learning_rate": 1.2302738175659777e-05,
      "loss": 0.52,
      "step": 12858
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6029242277145386,
      "learning_rate": 1.230165159424021e-05,
      "loss": 0.6462,
      "step": 12859
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4299497604370117,
      "learning_rate": 1.230056498412509e-05,
      "loss": 0.521,
      "step": 12860
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5173091888427734,
      "learning_rate": 1.229947834532796e-05,
      "loss": 0.4711,
      "step": 12861
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.1486120223999023,
      "learning_rate": 1.2298391677862371e-05,
      "loss": 0.5074,
      "step": 12862
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.483357548713684,
      "learning_rate": 1.2297304981741873e-05,
      "loss": 0.5511,
      "step": 12863
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4956176280975342,
      "learning_rate": 1.229621825698001e-05,
      "loss": 0.4766,
      "step": 12864
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3126791715621948,
      "learning_rate": 1.2295131503590334e-05,
      "loss": 0.5299,
      "step": 12865
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.480831503868103,
      "learning_rate": 1.2294044721586388e-05,
      "loss": 0.472,
      "step": 12866
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5057107210159302,
      "learning_rate": 1.2292957910981729e-05,
      "loss": 0.4892,
      "step": 12867
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.356747031211853,
      "learning_rate": 1.2291871071789902e-05,
      "loss": 0.4905,
      "step": 12868
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2871124744415283,
      "learning_rate": 1.2290784204024457e-05,
      "loss": 0.5025,
      "step": 12869
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8689247369766235,
      "learning_rate": 1.2289697307698947e-05,
      "loss": 0.5976,
      "step": 12870
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6301357746124268,
      "learning_rate": 1.2288610382826922e-05,
      "loss": 0.5697,
      "step": 12871
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5778145790100098,
      "learning_rate": 1.228752342942193e-05,
      "loss": 0.5205,
      "step": 12872
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5489871501922607,
      "learning_rate": 1.2286436447497528e-05,
      "loss": 0.6425,
      "step": 12873
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4550790786743164,
      "learning_rate": 1.2285349437067262e-05,
      "loss": 0.5199,
      "step": 12874
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.864829659461975,
      "learning_rate": 1.2284262398144686e-05,
      "loss": 0.5837,
      "step": 12875
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6367347240447998,
      "learning_rate": 1.2283175330743355e-05,
      "loss": 0.4656,
      "step": 12876
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5696237087249756,
      "learning_rate": 1.228208823487682e-05,
      "loss": 0.6383,
      "step": 12877
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.386887788772583,
      "learning_rate": 1.2281001110558634e-05,
      "loss": 0.4266,
      "step": 12878
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6391315460205078,
      "learning_rate": 1.2279913957802352e-05,
      "loss": 0.4912,
      "step": 12879
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6439614295959473,
      "learning_rate": 1.2278826776621524e-05,
      "loss": 0.5585,
      "step": 12880
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4254580736160278,
      "learning_rate": 1.2277739567029712e-05,
      "loss": 0.5065,
      "step": 12881
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5038739442825317,
      "learning_rate": 1.2276652329040464e-05,
      "loss": 0.6353,
      "step": 12882
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2844398021697998,
      "learning_rate": 1.2275565062667334e-05,
      "loss": 0.5187,
      "step": 12883
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5801022052764893,
      "learning_rate": 1.2274477767923885e-05,
      "loss": 0.4199,
      "step": 12884
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3965455293655396,
      "learning_rate": 1.2273390444823663e-05,
      "loss": 0.6017,
      "step": 12885
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.871169924736023,
      "learning_rate": 1.2272303093380233e-05,
      "loss": 0.582,
      "step": 12886
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4879956245422363,
      "learning_rate": 1.2271215713607145e-05,
      "loss": 0.5191,
      "step": 12887
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3946382999420166,
      "learning_rate": 1.2270128305517958e-05,
      "loss": 0.5904,
      "step": 12888
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5206916332244873,
      "learning_rate": 1.2269040869126235e-05,
      "loss": 0.533,
      "step": 12889
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4576547145843506,
      "learning_rate": 1.226795340444552e-05,
      "loss": 0.451,
      "step": 12890
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4781179428100586,
      "learning_rate": 1.2266865911489385e-05,
      "loss": 0.4871,
      "step": 12891
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4314980506896973,
      "learning_rate": 1.2265778390271376e-05,
      "loss": 0.4785,
      "step": 12892
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5986520051956177,
      "learning_rate": 1.226469084080506e-05,
      "loss": 0.4562,
      "step": 12893
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.6194838285446167,
      "learning_rate": 1.2263603263103996e-05,
      "loss": 0.6268,
      "step": 12894
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8018654584884644,
      "learning_rate": 1.2262515657181738e-05,
      "loss": 0.4919,
      "step": 12895
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8128209114074707,
      "learning_rate": 1.2261428023051847e-05,
      "loss": 0.5432,
      "step": 12896
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4059633016586304,
      "learning_rate": 1.2260340360727888e-05,
      "loss": 0.511,
      "step": 12897
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.662265419960022,
      "learning_rate": 1.2259252670223412e-05,
      "loss": 0.5313,
      "step": 12898
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3829500675201416,
      "learning_rate": 1.225816495155199e-05,
      "loss": 0.5354,
      "step": 12899
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.223125457763672,
      "learning_rate": 1.2257077204727174e-05,
      "loss": 0.463,
      "step": 12900
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4592853784561157,
      "learning_rate": 1.225598942976253e-05,
      "loss": 0.4899,
      "step": 12901
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3409066200256348,
      "learning_rate": 1.225490162667162e-05,
      "loss": 0.5499,
      "step": 12902
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3742942810058594,
      "learning_rate": 1.2253813795468004e-05,
      "loss": 0.5788,
      "step": 12903
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.4417216777801514,
      "learning_rate": 1.2252725936165249e-05,
      "loss": 0.5252,
      "step": 12904
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.8881171941757202,
      "learning_rate": 1.2251638048776911e-05,
      "loss": 0.4769,
      "step": 12905
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3341864347457886,
      "learning_rate": 1.2250550133316556e-05,
      "loss": 0.5556,
      "step": 12906
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.479587197303772,
      "learning_rate": 1.224946218979775e-05,
      "loss": 0.6014,
      "step": 12907
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5159080028533936,
      "learning_rate": 1.2248374218234051e-05,
      "loss": 0.492,
      "step": 12908
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5397775173187256,
      "learning_rate": 1.224728621863903e-05,
      "loss": 0.6552,
      "step": 12909
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3100959062576294,
      "learning_rate": 1.224619819102625e-05,
      "loss": 0.4964,
      "step": 12910
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.301889181137085,
      "learning_rate": 1.2245110135409268e-05,
      "loss": 0.4428,
      "step": 12911
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.476486086845398,
      "learning_rate": 1.224402205180166e-05,
      "loss": 0.5587,
      "step": 12912
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7529585361480713,
      "learning_rate": 1.2242933940216986e-05,
      "loss": 0.602,
      "step": 12913
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6625032424926758,
      "learning_rate": 1.224184580066881e-05,
      "loss": 0.5858,
      "step": 12914
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6082899570465088,
      "learning_rate": 1.2240757633170706e-05,
      "loss": 0.5414,
      "step": 12915
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.9089020490646362,
      "learning_rate": 1.223966943773623e-05,
      "loss": 0.6465,
      "step": 12916
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6299294233322144,
      "learning_rate": 1.223858121437896e-05,
      "loss": 0.5409,
      "step": 12917
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.524247407913208,
      "learning_rate": 1.2237492963112456e-05,
      "loss": 0.5634,
      "step": 12918
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5800132751464844,
      "learning_rate": 1.2236404683950285e-05,
      "loss": 0.6344,
      "step": 12919
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5422766208648682,
      "learning_rate": 1.2235316376906017e-05,
      "loss": 0.5484,
      "step": 12920
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5080420970916748,
      "learning_rate": 1.2234228041993226e-05,
      "loss": 0.5664,
      "step": 12921
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5731289386749268,
      "learning_rate": 1.223313967922547e-05,
      "loss": 0.6287,
      "step": 12922
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6968872547149658,
      "learning_rate": 1.2232051288616326e-05,
      "loss": 0.5769,
      "step": 12923
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4055026769638062,
      "learning_rate": 1.2230962870179359e-05,
      "loss": 0.4449,
      "step": 12924
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6198137998580933,
      "learning_rate": 1.222987442392814e-05,
      "loss": 0.645,
      "step": 12925
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5478566884994507,
      "learning_rate": 1.2228785949876245e-05,
      "loss": 0.5344,
      "step": 12926
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5038622617721558,
      "learning_rate": 1.2227697448037234e-05,
      "loss": 0.5434,
      "step": 12927
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5272164344787598,
      "learning_rate": 1.2226608918424685e-05,
      "loss": 0.5056,
      "step": 12928
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4209238290786743,
      "learning_rate": 1.2225520361052166e-05,
      "loss": 0.5362,
      "step": 12929
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3759664297103882,
      "learning_rate": 1.2224431775933248e-05,
      "loss": 0.4809,
      "step": 12930
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.668745756149292,
      "learning_rate": 1.2223343163081508e-05,
      "loss": 0.5326,
      "step": 12931
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8930683135986328,
      "learning_rate": 1.222225452251051e-05,
      "loss": 0.5523,
      "step": 12932
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.56662917137146,
      "learning_rate": 1.2221165854233832e-05,
      "loss": 0.4485,
      "step": 12933
    },
    {
      "epoch": 1.34,
      "grad_norm": 2.152266502380371,
      "learning_rate": 1.2220077158265047e-05,
      "loss": 0.4681,
      "step": 12934
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5487195253372192,
      "learning_rate": 1.2218988434617725e-05,
      "loss": 0.5798,
      "step": 12935
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5169132947921753,
      "learning_rate": 1.2217899683305444e-05,
      "loss": 0.6363,
      "step": 12936
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.67305588722229,
      "learning_rate": 1.2216810904341772e-05,
      "loss": 0.5517,
      "step": 12937
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8971483707427979,
      "learning_rate": 1.2215722097740286e-05,
      "loss": 0.5657,
      "step": 12938
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5940965414047241,
      "learning_rate": 1.2214633263514566e-05,
      "loss": 0.5576,
      "step": 12939
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.776892900466919,
      "learning_rate": 1.2213544401678177e-05,
      "loss": 0.587,
      "step": 12940
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3790696859359741,
      "learning_rate": 1.2212455512244704e-05,
      "loss": 0.5762,
      "step": 12941
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.659135103225708,
      "learning_rate": 1.2211366595227711e-05,
      "loss": 0.4816,
      "step": 12942
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4122024774551392,
      "learning_rate": 1.2210277650640785e-05,
      "loss": 0.401,
      "step": 12943
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7457467317581177,
      "learning_rate": 1.2209188678497496e-05,
      "loss": 0.6536,
      "step": 12944
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4587252140045166,
      "learning_rate": 1.2208099678811424e-05,
      "loss": 0.553,
      "step": 12945
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.455306887626648,
      "learning_rate": 1.2207010651596143e-05,
      "loss": 0.5317,
      "step": 12946
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5864899158477783,
      "learning_rate": 1.2205921596865236e-05,
      "loss": 0.645,
      "step": 12947
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4674235582351685,
      "learning_rate": 1.2204832514632272e-05,
      "loss": 0.4806,
      "step": 12948
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3593829870224,
      "learning_rate": 1.2203743404910835e-05,
      "loss": 0.4455,
      "step": 12949
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4400721788406372,
      "learning_rate": 1.2202654267714502e-05,
      "loss": 0.5597,
      "step": 12950
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.577035665512085,
      "learning_rate": 1.2201565103056852e-05,
      "loss": 0.4733,
      "step": 12951
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7377268075942993,
      "learning_rate": 1.2200475910951464e-05,
      "loss": 0.606,
      "step": 12952
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4525278806686401,
      "learning_rate": 1.2199386691411914e-05,
      "loss": 0.5095,
      "step": 12953
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6223312616348267,
      "learning_rate": 1.2198297444451787e-05,
      "loss": 0.5491,
      "step": 12954
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5682259798049927,
      "learning_rate": 1.219720817008466e-05,
      "loss": 0.4921,
      "step": 12955
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3179020881652832,
      "learning_rate": 1.2196118868324112e-05,
      "loss": 0.4962,
      "step": 12956
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3291651010513306,
      "learning_rate": 1.2195029539183729e-05,
      "loss": 0.5314,
      "step": 12957
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8758646249771118,
      "learning_rate": 1.2193940182677086e-05,
      "loss": 0.4882,
      "step": 12958
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.367248296737671,
      "learning_rate": 1.219285079881777e-05,
      "loss": 0.5073,
      "step": 12959
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5203272104263306,
      "learning_rate": 1.219176138761936e-05,
      "loss": 0.6685,
      "step": 12960
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4890333414077759,
      "learning_rate": 1.2190671949095436e-05,
      "loss": 0.6107,
      "step": 12961
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5029046535491943,
      "learning_rate": 1.2189582483259586e-05,
      "loss": 0.4496,
      "step": 12962
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4046180248260498,
      "learning_rate": 1.2188492990125384e-05,
      "loss": 0.5707,
      "step": 12963
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4796277284622192,
      "learning_rate": 1.2187403469706423e-05,
      "loss": 0.5792,
      "step": 12964
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5762087106704712,
      "learning_rate": 1.2186313922016283e-05,
      "loss": 0.5335,
      "step": 12965
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7153425216674805,
      "learning_rate": 1.2185224347068542e-05,
      "loss": 0.4789,
      "step": 12966
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.488242506980896,
      "learning_rate": 1.2184134744876792e-05,
      "loss": 0.4985,
      "step": 12967
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4614659547805786,
      "learning_rate": 1.2183045115454612e-05,
      "loss": 0.4512,
      "step": 12968
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6847507953643799,
      "learning_rate": 1.218195545881559e-05,
      "loss": 0.5801,
      "step": 12969
    },
    {
      "epoch": 1.34,
      "grad_norm": 2.108083486557007,
      "learning_rate": 1.2180865774973313e-05,
      "loss": 0.6371,
      "step": 12970
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3735116720199585,
      "learning_rate": 1.217977606394136e-05,
      "loss": 0.5776,
      "step": 12971
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.501770257949829,
      "learning_rate": 1.2178686325733322e-05,
      "loss": 0.5472,
      "step": 12972
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5229623317718506,
      "learning_rate": 1.2177596560362785e-05,
      "loss": 0.6008,
      "step": 12973
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.569769263267517,
      "learning_rate": 1.2176506767843335e-05,
      "loss": 0.5501,
      "step": 12974
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.9379719495773315,
      "learning_rate": 1.2175416948188557e-05,
      "loss": 0.4843,
      "step": 12975
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7986253499984741,
      "learning_rate": 1.2174327101412038e-05,
      "loss": 0.4693,
      "step": 12976
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3141207695007324,
      "learning_rate": 1.2173237227527368e-05,
      "loss": 0.4849,
      "step": 12977
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.528107762336731,
      "learning_rate": 1.2172147326548133e-05,
      "loss": 0.6476,
      "step": 12978
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.499046802520752,
      "learning_rate": 1.2171057398487924e-05,
      "loss": 0.5008,
      "step": 12979
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6522071361541748,
      "learning_rate": 1.2169967443360328e-05,
      "loss": 0.5162,
      "step": 12980
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.416709065437317,
      "learning_rate": 1.2168877461178931e-05,
      "loss": 0.6009,
      "step": 12981
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.814602255821228,
      "learning_rate": 1.2167787451957324e-05,
      "loss": 0.5619,
      "step": 12982
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5131042003631592,
      "learning_rate": 1.2166697415709102e-05,
      "loss": 0.5497,
      "step": 12983
    },
    {
      "epoch": 1.34,
      "grad_norm": 2.0584142208099365,
      "learning_rate": 1.2165607352447843e-05,
      "loss": 0.649,
      "step": 12984
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5944902896881104,
      "learning_rate": 1.2164517262187148e-05,
      "loss": 0.524,
      "step": 12985
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.428806185722351,
      "learning_rate": 1.2163427144940605e-05,
      "loss": 0.4585,
      "step": 12986
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6942687034606934,
      "learning_rate": 1.2162337000721802e-05,
      "loss": 0.5444,
      "step": 12987
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4852564334869385,
      "learning_rate": 1.2161246829544335e-05,
      "loss": 0.4502,
      "step": 12988
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6434861421585083,
      "learning_rate": 1.2160156631421788e-05,
      "loss": 0.5567,
      "step": 12989
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.596040964126587,
      "learning_rate": 1.215906640636776e-05,
      "loss": 0.5537,
      "step": 12990
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6451970338821411,
      "learning_rate": 1.2157976154395841e-05,
      "loss": 0.6005,
      "step": 12991
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4190237522125244,
      "learning_rate": 1.2156885875519624e-05,
      "loss": 0.5101,
      "step": 12992
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3215559720993042,
      "learning_rate": 1.21557955697527e-05,
      "loss": 0.5506,
      "step": 12993
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7872575521469116,
      "learning_rate": 1.2154705237108663e-05,
      "loss": 0.6109,
      "step": 12994
    },
    {
      "epoch": 1.34,
      "grad_norm": 2.0275449752807617,
      "learning_rate": 1.2153614877601108e-05,
      "loss": 0.5055,
      "step": 12995
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.727331519126892,
      "learning_rate": 1.2152524491243627e-05,
      "loss": 0.5529,
      "step": 12996
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7363237142562866,
      "learning_rate": 1.2151434078049815e-05,
      "loss": 0.5963,
      "step": 12997
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4426064491271973,
      "learning_rate": 1.2150343638033269e-05,
      "loss": 0.5722,
      "step": 12998
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.5499407052993774,
      "learning_rate": 1.214925317120758e-05,
      "loss": 0.4395,
      "step": 12999
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3218590021133423,
      "learning_rate": 1.2148162677586347e-05,
      "loss": 0.5437,
      "step": 13000
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.7360135316848755,
      "learning_rate": 1.2147072157183161e-05,
      "loss": 0.6527,
      "step": 13001
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4740219116210938,
      "learning_rate": 1.214598161001162e-05,
      "loss": 0.5654,
      "step": 13002
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.8533961772918701,
      "learning_rate": 1.2144891036085323e-05,
      "loss": 0.6208,
      "step": 13003
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.792264699935913,
      "learning_rate": 1.2143800435417865e-05,
      "loss": 0.5225,
      "step": 13004
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.453863501548767,
      "learning_rate": 1.2142709808022841e-05,
      "loss": 0.5488,
      "step": 13005
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4502888917922974,
      "learning_rate": 1.2141619153913851e-05,
      "loss": 0.6381,
      "step": 13006
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3613923788070679,
      "learning_rate": 1.214052847310449e-05,
      "loss": 0.4923,
      "step": 13007
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4838346242904663,
      "learning_rate": 1.2139437765608355e-05,
      "loss": 0.6768,
      "step": 13008
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.9518673419952393,
      "learning_rate": 1.2138347031439048e-05,
      "loss": 0.6211,
      "step": 13009
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.467703938484192,
      "learning_rate": 1.2137256270610167e-05,
      "loss": 0.5675,
      "step": 13010
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5848978757858276,
      "learning_rate": 1.213616548313531e-05,
      "loss": 0.5545,
      "step": 13011
    },
    {
      "epoch": 1.35,
      "grad_norm": 2.6725876331329346,
      "learning_rate": 1.2135074669028074e-05,
      "loss": 0.6087,
      "step": 13012
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.632532000541687,
      "learning_rate": 1.2133983828302063e-05,
      "loss": 0.6633,
      "step": 13013
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4320173263549805,
      "learning_rate": 1.2132892960970874e-05,
      "loss": 0.5685,
      "step": 13014
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6191825866699219,
      "learning_rate": 1.2131802067048108e-05,
      "loss": 0.5723,
      "step": 13015
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3972666263580322,
      "learning_rate": 1.2130711146547365e-05,
      "loss": 0.6081,
      "step": 13016
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7122104167938232,
      "learning_rate": 1.2129620199482245e-05,
      "loss": 0.5026,
      "step": 13017
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4080424308776855,
      "learning_rate": 1.2128529225866352e-05,
      "loss": 0.5899,
      "step": 13018
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3435596227645874,
      "learning_rate": 1.2127438225713288e-05,
      "loss": 0.4823,
      "step": 13019
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8951276540756226,
      "learning_rate": 1.2126347199036652e-05,
      "loss": 0.6588,
      "step": 13020
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.50592839717865,
      "learning_rate": 1.2125256145850046e-05,
      "loss": 0.6803,
      "step": 13021
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2205696105957031,
      "learning_rate": 1.2124165066167076e-05,
      "loss": 0.5439,
      "step": 13022
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2802472114562988,
      "learning_rate": 1.2123073960001342e-05,
      "loss": 0.6467,
      "step": 13023
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.683136224746704,
      "learning_rate": 1.2121982827366446e-05,
      "loss": 0.5955,
      "step": 13024
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4005206823349,
      "learning_rate": 1.2120891668275996e-05,
      "loss": 0.4948,
      "step": 13025
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7438135147094727,
      "learning_rate": 1.2119800482743593e-05,
      "loss": 0.5076,
      "step": 13026
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.437921404838562,
      "learning_rate": 1.2118709270782843e-05,
      "loss": 0.6452,
      "step": 13027
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.9231343269348145,
      "learning_rate": 1.2117618032407348e-05,
      "loss": 0.6235,
      "step": 13028
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.42995023727417,
      "learning_rate": 1.2116526767630711e-05,
      "loss": 0.6217,
      "step": 13029
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5439622402191162,
      "learning_rate": 1.2115435476466546e-05,
      "loss": 0.4887,
      "step": 13030
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.347630500793457,
      "learning_rate": 1.211434415892845e-05,
      "loss": 0.5315,
      "step": 13031
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5885887145996094,
      "learning_rate": 1.2113252815030035e-05,
      "loss": 0.4476,
      "step": 13032
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6303932666778564,
      "learning_rate": 1.21121614447849e-05,
      "loss": 0.5599,
      "step": 13033
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4620734453201294,
      "learning_rate": 1.2111070048206654e-05,
      "loss": 0.5524,
      "step": 13034
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7069344520568848,
      "learning_rate": 1.2109978625308911e-05,
      "loss": 0.5412,
      "step": 13035
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.579256534576416,
      "learning_rate": 1.2108887176105268e-05,
      "loss": 0.6443,
      "step": 13036
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6199325323104858,
      "learning_rate": 1.2107795700609337e-05,
      "loss": 0.5307,
      "step": 13037
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2601147890090942,
      "learning_rate": 1.210670419883473e-05,
      "loss": 0.5352,
      "step": 13038
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5814805030822754,
      "learning_rate": 1.2105612670795046e-05,
      "loss": 0.552,
      "step": 13039
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6584105491638184,
      "learning_rate": 1.2104521116503905e-05,
      "loss": 0.4846,
      "step": 13040
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8191330432891846,
      "learning_rate": 1.2103429535974903e-05,
      "loss": 0.5372,
      "step": 13041
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.498746633529663,
      "learning_rate": 1.2102337929221658e-05,
      "loss": 0.4674,
      "step": 13042
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6280823945999146,
      "learning_rate": 1.2101246296257779e-05,
      "loss": 0.4469,
      "step": 13043
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4845174551010132,
      "learning_rate": 1.210015463709687e-05,
      "loss": 0.5308,
      "step": 13044
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3274098634719849,
      "learning_rate": 1.2099062951752547e-05,
      "loss": 0.5615,
      "step": 13045
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.283652901649475,
      "learning_rate": 1.2097971240238418e-05,
      "loss": 0.5565,
      "step": 13046
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5515295267105103,
      "learning_rate": 1.2096879502568094e-05,
      "loss": 0.6845,
      "step": 13047
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4198192358016968,
      "learning_rate": 1.2095787738755185e-05,
      "loss": 0.5853,
      "step": 13048
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.519742488861084,
      "learning_rate": 1.2094695948813305e-05,
      "loss": 0.6026,
      "step": 13049
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6854960918426514,
      "learning_rate": 1.2093604132756065e-05,
      "loss": 0.5813,
      "step": 13050
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6437031030654907,
      "learning_rate": 1.2092512290597075e-05,
      "loss": 0.6847,
      "step": 13051
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3874486684799194,
      "learning_rate": 1.2091420422349949e-05,
      "loss": 0.5964,
      "step": 13052
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.51133394241333,
      "learning_rate": 1.2090328528028301e-05,
      "loss": 0.4922,
      "step": 13053
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.728211760520935,
      "learning_rate": 1.208923660764574e-05,
      "loss": 0.483,
      "step": 13054
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.379629135131836,
      "learning_rate": 1.2088144661215883e-05,
      "loss": 0.5083,
      "step": 13055
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.446236252784729,
      "learning_rate": 1.2087052688752346e-05,
      "loss": 0.4545,
      "step": 13056
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5075933933258057,
      "learning_rate": 1.2085960690268736e-05,
      "loss": 0.6172,
      "step": 13057
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4610918760299683,
      "learning_rate": 1.2084868665778673e-05,
      "loss": 0.5184,
      "step": 13058
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8470723628997803,
      "learning_rate": 1.2083776615295768e-05,
      "loss": 0.672,
      "step": 13059
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.672340750694275,
      "learning_rate": 1.208268453883364e-05,
      "loss": 0.4783,
      "step": 13060
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3728190660476685,
      "learning_rate": 1.2081592436405902e-05,
      "loss": 0.5117,
      "step": 13061
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2908785343170166,
      "learning_rate": 1.2080500308026168e-05,
      "loss": 0.4677,
      "step": 13062
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7644188404083252,
      "learning_rate": 1.2079408153708059e-05,
      "loss": 0.626,
      "step": 13063
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6169949769973755,
      "learning_rate": 1.2078315973465185e-05,
      "loss": 0.623,
      "step": 13064
    },
    {
      "epoch": 1.35,
      "grad_norm": 2.4771511554718018,
      "learning_rate": 1.2077223767311167e-05,
      "loss": 0.6646,
      "step": 13065
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.479282021522522,
      "learning_rate": 1.2076131535259619e-05,
      "loss": 0.5387,
      "step": 13066
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5610712766647339,
      "learning_rate": 1.2075039277324161e-05,
      "loss": 0.5106,
      "step": 13067
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.488221287727356,
      "learning_rate": 1.2073946993518409e-05,
      "loss": 0.4968,
      "step": 13068
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.515466570854187,
      "learning_rate": 1.2072854683855984e-05,
      "loss": 0.5044,
      "step": 13069
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6341633796691895,
      "learning_rate": 1.2071762348350497e-05,
      "loss": 0.596,
      "step": 13070
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7593485116958618,
      "learning_rate": 1.2070669987015575e-05,
      "loss": 0.5702,
      "step": 13071
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8591678142547607,
      "learning_rate": 1.2069577599864833e-05,
      "loss": 0.6967,
      "step": 13072
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.753559947013855,
      "learning_rate": 1.2068485186911886e-05,
      "loss": 0.6146,
      "step": 13073
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6931229829788208,
      "learning_rate": 1.2067392748170364e-05,
      "loss": 0.552,
      "step": 13074
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5283671617507935,
      "learning_rate": 1.2066300283653875e-05,
      "loss": 0.6634,
      "step": 13075
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7217274904251099,
      "learning_rate": 1.206520779337605e-05,
      "loss": 0.7147,
      "step": 13076
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.550514817237854,
      "learning_rate": 1.2064115277350501e-05,
      "loss": 0.4997,
      "step": 13077
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4914265871047974,
      "learning_rate": 1.2063022735590853e-05,
      "loss": 0.5812,
      "step": 13078
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5728497505187988,
      "learning_rate": 1.2061930168110727e-05,
      "loss": 0.4682,
      "step": 13079
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6682745218276978,
      "learning_rate": 1.2060837574923741e-05,
      "loss": 0.6005,
      "step": 13080
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6887075901031494,
      "learning_rate": 1.205974495604352e-05,
      "loss": 0.56,
      "step": 13081
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.8182710409164429,
      "learning_rate": 1.2058652311483687e-05,
      "loss": 0.4758,
      "step": 13082
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6301180124282837,
      "learning_rate": 1.2057559641257864e-05,
      "loss": 0.5517,
      "step": 13083
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4437897205352783,
      "learning_rate": 1.2056466945379672e-05,
      "loss": 0.4652,
      "step": 13084
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.9381775856018066,
      "learning_rate": 1.2055374223862733e-05,
      "loss": 0.5201,
      "step": 13085
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6997061967849731,
      "learning_rate": 1.2054281476720673e-05,
      "loss": 0.4515,
      "step": 13086
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.7736793756484985,
      "learning_rate": 1.2053188703967116e-05,
      "loss": 0.5548,
      "step": 13087
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6342275142669678,
      "learning_rate": 1.2052095905615681e-05,
      "loss": 0.6661,
      "step": 13088
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.6110597848892212,
      "learning_rate": 1.2051003081680002e-05,
      "loss": 0.5368,
      "step": 13089
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2869837284088135,
      "learning_rate": 1.2049910232173695e-05,
      "loss": 0.5331,
      "step": 13090
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4511475563049316,
      "learning_rate": 1.2048817357110385e-05,
      "loss": 0.5068,
      "step": 13091
    },
    {
      "epoch": 1.35,
      "grad_norm": 2.0244812965393066,
      "learning_rate": 1.2047724456503705e-05,
      "loss": 0.5625,
      "step": 13092
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5990145206451416,
      "learning_rate": 1.2046631530367275e-05,
      "loss": 0.5843,
      "step": 13093
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4845885038375854,
      "learning_rate": 1.2045538578714718e-05,
      "loss": 0.5889,
      "step": 13094
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.620572566986084,
      "learning_rate": 1.2044445601559664e-05,
      "loss": 0.3936,
      "step": 13095
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3471839427947998,
      "learning_rate": 1.204335259891574e-05,
      "loss": 0.4666,
      "step": 13096
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.764174461364746,
      "learning_rate": 1.2042259570796573e-05,
      "loss": 0.5682,
      "step": 13097
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5565615892410278,
      "learning_rate": 1.204116651721579e-05,
      "loss": 0.555,
      "step": 13098
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3680129051208496,
      "learning_rate": 1.2040073438187016e-05,
      "loss": 0.513,
      "step": 13099
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.390450119972229,
      "learning_rate": 1.2038980333723884e-05,
      "loss": 0.5293,
      "step": 13100
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5610772371292114,
      "learning_rate": 1.2037887203840015e-05,
      "loss": 0.5478,
      "step": 13101
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.446229338645935,
      "learning_rate": 1.2036794048549043e-05,
      "loss": 0.4818,
      "step": 13102
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5191445350646973,
      "learning_rate": 1.2035700867864597e-05,
      "loss": 0.6264,
      "step": 13103
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6030540466308594,
      "learning_rate": 1.2034607661800302e-05,
      "loss": 0.5868,
      "step": 13104
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3828758001327515,
      "learning_rate": 1.2033514430369788e-05,
      "loss": 0.4724,
      "step": 13105
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3985838890075684,
      "learning_rate": 1.203242117358669e-05,
      "loss": 0.606,
      "step": 13106
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4992053508758545,
      "learning_rate": 1.2031327891464633e-05,
      "loss": 0.5917,
      "step": 13107
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5286184549331665,
      "learning_rate": 1.2030234584017249e-05,
      "loss": 0.5887,
      "step": 13108
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3808599710464478,
      "learning_rate": 1.2029141251258167e-05,
      "loss": 0.4743,
      "step": 13109
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4645274877548218,
      "learning_rate": 1.2028047893201021e-05,
      "loss": 0.5876,
      "step": 13110
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.8015085458755493,
      "learning_rate": 1.202695450985944e-05,
      "loss": 0.5447,
      "step": 13111
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6024174690246582,
      "learning_rate": 1.2025861101247056e-05,
      "loss": 0.6543,
      "step": 13112
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5824592113494873,
      "learning_rate": 1.2024767667377502e-05,
      "loss": 0.4799,
      "step": 13113
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7713862657546997,
      "learning_rate": 1.202367420826441e-05,
      "loss": 0.55,
      "step": 13114
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.016385555267334,
      "learning_rate": 1.2022580723921409e-05,
      "loss": 0.534,
      "step": 13115
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4655203819274902,
      "learning_rate": 1.2021487214362138e-05,
      "loss": 0.4046,
      "step": 13116
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2724535465240479,
      "learning_rate": 1.2020393679600226e-05,
      "loss": 0.4636,
      "step": 13117
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6892646551132202,
      "learning_rate": 1.2019300119649306e-05,
      "loss": 0.6014,
      "step": 13118
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4165054559707642,
      "learning_rate": 1.2018206534523019e-05,
      "loss": 0.4926,
      "step": 13119
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3191163539886475,
      "learning_rate": 1.2017112924234988e-05,
      "loss": 0.4509,
      "step": 13120
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.0479915142059326,
      "learning_rate": 1.2016019288798857e-05,
      "loss": 0.4785,
      "step": 13121
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1974307298660278,
      "learning_rate": 1.2014925628228253e-05,
      "loss": 0.5443,
      "step": 13122
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2184112071990967,
      "learning_rate": 1.2013831942536818e-05,
      "loss": 0.5091,
      "step": 13123
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3247689008712769,
      "learning_rate": 1.2012738231738184e-05,
      "loss": 0.5263,
      "step": 13124
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6809996366500854,
      "learning_rate": 1.2011644495845985e-05,
      "loss": 0.5348,
      "step": 13125
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4654569625854492,
      "learning_rate": 1.2010550734873862e-05,
      "loss": 0.4855,
      "step": 13126
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.9438766241073608,
      "learning_rate": 1.2009456948835446e-05,
      "loss": 0.5343,
      "step": 13127
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.546431303024292,
      "learning_rate": 1.2008363137744376e-05,
      "loss": 0.5704,
      "step": 13128
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.836266040802002,
      "learning_rate": 1.200726930161429e-05,
      "loss": 0.5751,
      "step": 13129
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4106683731079102,
      "learning_rate": 1.2006175440458825e-05,
      "loss": 0.4865,
      "step": 13130
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7316707372665405,
      "learning_rate": 1.2005081554291616e-05,
      "loss": 0.7208,
      "step": 13131
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4797853231430054,
      "learning_rate": 1.2003987643126303e-05,
      "loss": 0.6025,
      "step": 13132
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4521727561950684,
      "learning_rate": 1.2002893706976525e-05,
      "loss": 0.5742,
      "step": 13133
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.673728585243225,
      "learning_rate": 1.2001799745855917e-05,
      "loss": 0.6253,
      "step": 13134
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5239120721817017,
      "learning_rate": 1.2000705759778122e-05,
      "loss": 0.5326,
      "step": 13135
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7203693389892578,
      "learning_rate": 1.1999611748756776e-05,
      "loss": 0.566,
      "step": 13136
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2994537353515625,
      "learning_rate": 1.1998517712805524e-05,
      "loss": 0.4195,
      "step": 13137
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5149738788604736,
      "learning_rate": 1.1997423651937996e-05,
      "loss": 0.5222,
      "step": 13138
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.768035650253296,
      "learning_rate": 1.1996329566167844e-05,
      "loss": 0.5096,
      "step": 13139
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5502334833145142,
      "learning_rate": 1.1995235455508695e-05,
      "loss": 0.6119,
      "step": 13140
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6246225833892822,
      "learning_rate": 1.19941413199742e-05,
      "loss": 0.4869,
      "step": 13141
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5710760354995728,
      "learning_rate": 1.1993047159577997e-05,
      "loss": 0.6774,
      "step": 13142
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4638673067092896,
      "learning_rate": 1.1991952974333727e-05,
      "loss": 0.564,
      "step": 13143
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.8087886571884155,
      "learning_rate": 1.1990858764255029e-05,
      "loss": 0.6424,
      "step": 13144
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3659470081329346,
      "learning_rate": 1.1989764529355553e-05,
      "loss": 0.6061,
      "step": 13145
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6588033437728882,
      "learning_rate": 1.198867026964893e-05,
      "loss": 0.4809,
      "step": 13146
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.421363115310669,
      "learning_rate": 1.1987575985148813e-05,
      "loss": 0.4976,
      "step": 13147
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.298829436302185,
      "learning_rate": 1.1986481675868838e-05,
      "loss": 0.5959,
      "step": 13148
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4725923538208008,
      "learning_rate": 1.198538734182265e-05,
      "loss": 0.421,
      "step": 13149
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3270231485366821,
      "learning_rate": 1.1984292983023893e-05,
      "loss": 0.5329,
      "step": 13150
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6436129808425903,
      "learning_rate": 1.198319859948621e-05,
      "loss": 0.5748,
      "step": 13151
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.427308440208435,
      "learning_rate": 1.1982104191223247e-05,
      "loss": 0.6213,
      "step": 13152
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7394236326217651,
      "learning_rate": 1.1981009758248647e-05,
      "loss": 0.5389,
      "step": 13153
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.727374792098999,
      "learning_rate": 1.1979915300576053e-05,
      "loss": 0.6799,
      "step": 13154
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5276762247085571,
      "learning_rate": 1.1978820818219115e-05,
      "loss": 0.5711,
      "step": 13155
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.179884910583496,
      "learning_rate": 1.1977726311191472e-05,
      "loss": 0.6124,
      "step": 13156
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7845304012298584,
      "learning_rate": 1.1976631779506774e-05,
      "loss": 0.5268,
      "step": 13157
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6914328336715698,
      "learning_rate": 1.1975537223178665e-05,
      "loss": 0.5309,
      "step": 13158
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5136208534240723,
      "learning_rate": 1.1974442642220794e-05,
      "loss": 0.541,
      "step": 13159
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3948745727539062,
      "learning_rate": 1.1973348036646803e-05,
      "loss": 0.6495,
      "step": 13160
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7522441148757935,
      "learning_rate": 1.1972253406470344e-05,
      "loss": 0.5752,
      "step": 13161
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5112849473953247,
      "learning_rate": 1.1971158751705059e-05,
      "loss": 0.5825,
      "step": 13162
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.319907546043396,
      "learning_rate": 1.19700640723646e-05,
      "loss": 0.5466,
      "step": 13163
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4488986730575562,
      "learning_rate": 1.196896936846261e-05,
      "loss": 0.5405,
      "step": 13164
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3580478429794312,
      "learning_rate": 1.1967874640012745e-05,
      "loss": 0.502,
      "step": 13165
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.8430176973342896,
      "learning_rate": 1.1966779887028643e-05,
      "loss": 0.5905,
      "step": 13166
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5828635692596436,
      "learning_rate": 1.1965685109523959e-05,
      "loss": 0.5499,
      "step": 13167
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.440486192703247,
      "learning_rate": 1.1964590307512345e-05,
      "loss": 0.504,
      "step": 13168
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6771138906478882,
      "learning_rate": 1.196349548100744e-05,
      "loss": 0.5926,
      "step": 13169
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4148519039154053,
      "learning_rate": 1.1962400630022904e-05,
      "loss": 0.4938,
      "step": 13170
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.824771761894226,
      "learning_rate": 1.1961305754572383e-05,
      "loss": 0.5977,
      "step": 13171
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5308364629745483,
      "learning_rate": 1.1960210854669525e-05,
      "loss": 0.6256,
      "step": 13172
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3513829708099365,
      "learning_rate": 1.1959115930327985e-05,
      "loss": 0.5875,
      "step": 13173
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3678548336029053,
      "learning_rate": 1.1958020981561406e-05,
      "loss": 0.5862,
      "step": 13174
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4574135541915894,
      "learning_rate": 1.195692600838345e-05,
      "loss": 0.631,
      "step": 13175
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5151973962783813,
      "learning_rate": 1.195583101080776e-05,
      "loss": 0.5755,
      "step": 13176
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6105542182922363,
      "learning_rate": 1.1954735988847992e-05,
      "loss": 0.5329,
      "step": 13177
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4487162828445435,
      "learning_rate": 1.1953640942517799e-05,
      "loss": 0.4714,
      "step": 13178
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7611804008483887,
      "learning_rate": 1.1952545871830827e-05,
      "loss": 0.5074,
      "step": 13179
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2942878007888794,
      "learning_rate": 1.1951450776800735e-05,
      "loss": 0.5252,
      "step": 13180
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7284348011016846,
      "learning_rate": 1.1950355657441173e-05,
      "loss": 0.6048,
      "step": 13181
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2967052459716797,
      "learning_rate": 1.1949260513765794e-05,
      "loss": 0.4827,
      "step": 13182
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3891828060150146,
      "learning_rate": 1.1948165345788254e-05,
      "loss": 0.5927,
      "step": 13183
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.800325870513916,
      "learning_rate": 1.1947070153522207e-05,
      "loss": 0.5406,
      "step": 13184
    },
    {
      "epoch": 1.36,
      "grad_norm": 2.2544517517089844,
      "learning_rate": 1.1945974936981301e-05,
      "loss": 0.6094,
      "step": 13185
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5590118169784546,
      "learning_rate": 1.1944879696179199e-05,
      "loss": 0.6273,
      "step": 13186
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6542725563049316,
      "learning_rate": 1.194378443112955e-05,
      "loss": 0.602,
      "step": 13187
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.795963168144226,
      "learning_rate": 1.1942689141846011e-05,
      "loss": 0.6221,
      "step": 13188
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.6777441501617432,
      "learning_rate": 1.1941593828342239e-05,
      "loss": 0.5356,
      "step": 13189
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.430209755897522,
      "learning_rate": 1.1940498490631886e-05,
      "loss": 0.563,
      "step": 13190
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5841190814971924,
      "learning_rate": 1.1939403128728613e-05,
      "loss": 0.5727,
      "step": 13191
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.7982219457626343,
      "learning_rate": 1.1938307742646068e-05,
      "loss": 0.5331,
      "step": 13192
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.623356819152832,
      "learning_rate": 1.1937212332397917e-05,
      "loss": 0.6258,
      "step": 13193
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4380488395690918,
      "learning_rate": 1.1936116897997813e-05,
      "loss": 0.512,
      "step": 13194
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.721625804901123,
      "learning_rate": 1.1935021439459412e-05,
      "loss": 0.5332,
      "step": 13195
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5648244619369507,
      "learning_rate": 1.193392595679637e-05,
      "loss": 0.5721,
      "step": 13196
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.5442363023757935,
      "learning_rate": 1.193283045002235e-05,
      "loss": 0.6645,
      "step": 13197
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.734409213066101,
      "learning_rate": 1.1931734919151005e-05,
      "loss": 0.558,
      "step": 13198
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3759180307388306,
      "learning_rate": 1.1930639364195999e-05,
      "loss": 0.6837,
      "step": 13199
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5889256000518799,
      "learning_rate": 1.1929543785170984e-05,
      "loss": 0.545,
      "step": 13200
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.634458303451538,
      "learning_rate": 1.1928448182089624e-05,
      "loss": 0.6714,
      "step": 13201
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5014971494674683,
      "learning_rate": 1.192735255496558e-05,
      "loss": 0.4497,
      "step": 13202
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6681090593338013,
      "learning_rate": 1.1926256903812502e-05,
      "loss": 0.5923,
      "step": 13203
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6635953187942505,
      "learning_rate": 1.1925161228644063e-05,
      "loss": 0.7161,
      "step": 13204
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3774490356445312,
      "learning_rate": 1.1924065529473911e-05,
      "loss": 0.4984,
      "step": 13205
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.433705449104309,
      "learning_rate": 1.1922969806315712e-05,
      "loss": 0.4896,
      "step": 13206
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3218053579330444,
      "learning_rate": 1.192187405918313e-05,
      "loss": 0.5878,
      "step": 13207
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4992568492889404,
      "learning_rate": 1.192077828808982e-05,
      "loss": 0.5299,
      "step": 13208
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6837650537490845,
      "learning_rate": 1.1919682493049447e-05,
      "loss": 0.5104,
      "step": 13209
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.343618392944336,
      "learning_rate": 1.1918586674075673e-05,
      "loss": 0.6161,
      "step": 13210
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4018275737762451,
      "learning_rate": 1.1917490831182157e-05,
      "loss": 0.4692,
      "step": 13211
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6766599416732788,
      "learning_rate": 1.1916394964382566e-05,
      "loss": 0.6035,
      "step": 13212
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4304051399230957,
      "learning_rate": 1.1915299073690557e-05,
      "loss": 0.6205,
      "step": 13213
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4283396005630493,
      "learning_rate": 1.1914203159119793e-05,
      "loss": 0.6172,
      "step": 13214
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3613386154174805,
      "learning_rate": 1.1913107220683947e-05,
      "loss": 0.5939,
      "step": 13215
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.47841215133667,
      "learning_rate": 1.1912011258396668e-05,
      "loss": 0.5364,
      "step": 13216
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.686901330947876,
      "learning_rate": 1.1910915272271633e-05,
      "loss": 0.428,
      "step": 13217
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.428723931312561,
      "learning_rate": 1.1909819262322495e-05,
      "loss": 0.4935,
      "step": 13218
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3288993835449219,
      "learning_rate": 1.1908723228562925e-05,
      "loss": 0.5489,
      "step": 13219
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5731096267700195,
      "learning_rate": 1.1907627171006589e-05,
      "loss": 0.4549,
      "step": 13220
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6014095544815063,
      "learning_rate": 1.1906531089667143e-05,
      "loss": 0.602,
      "step": 13221
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5494389533996582,
      "learning_rate": 1.1905434984558261e-05,
      "loss": 0.5362,
      "step": 13222
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4981061220169067,
      "learning_rate": 1.190433885569361e-05,
      "loss": 0.5907,
      "step": 13223
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.290412425994873,
      "learning_rate": 1.1903242703086848e-05,
      "loss": 0.5117,
      "step": 13224
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.9040172100067139,
      "learning_rate": 1.1902146526751646e-05,
      "loss": 0.5445,
      "step": 13225
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.609785795211792,
      "learning_rate": 1.1901050326701668e-05,
      "loss": 0.5754,
      "step": 13226
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4010992050170898,
      "learning_rate": 1.1899954102950582e-05,
      "loss": 0.5951,
      "step": 13227
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.9357765913009644,
      "learning_rate": 1.1898857855512058e-05,
      "loss": 0.6048,
      "step": 13228
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6063669919967651,
      "learning_rate": 1.1897761584399757e-05,
      "loss": 0.6264,
      "step": 13229
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6732457876205444,
      "learning_rate": 1.1896665289627353e-05,
      "loss": 0.6061,
      "step": 13230
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6536310911178589,
      "learning_rate": 1.1895568971208508e-05,
      "loss": 0.4792,
      "step": 13231
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5379563570022583,
      "learning_rate": 1.1894472629156894e-05,
      "loss": 0.594,
      "step": 13232
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5090346336364746,
      "learning_rate": 1.1893376263486181e-05,
      "loss": 0.5875,
      "step": 13233
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5454318523406982,
      "learning_rate": 1.189227987421003e-05,
      "loss": 0.5318,
      "step": 13234
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5632538795471191,
      "learning_rate": 1.189118346134212e-05,
      "loss": 0.6085,
      "step": 13235
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4127063751220703,
      "learning_rate": 1.1890087024896117e-05,
      "loss": 0.5799,
      "step": 13236
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.464735984802246,
      "learning_rate": 1.1888990564885687e-05,
      "loss": 0.3869,
      "step": 13237
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5747787952423096,
      "learning_rate": 1.1887894081324504e-05,
      "loss": 0.5487,
      "step": 13238
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.571622371673584,
      "learning_rate": 1.1886797574226237e-05,
      "loss": 0.5988,
      "step": 13239
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.404835820198059,
      "learning_rate": 1.1885701043604555e-05,
      "loss": 0.6227,
      "step": 13240
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5652366876602173,
      "learning_rate": 1.188460448947313e-05,
      "loss": 0.6131,
      "step": 13241
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.381399154663086,
      "learning_rate": 1.1883507911845635e-05,
      "loss": 0.5731,
      "step": 13242
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.438674807548523,
      "learning_rate": 1.188241131073574e-05,
      "loss": 0.4703,
      "step": 13243
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.2554388046264648,
      "learning_rate": 1.1881314686157115e-05,
      "loss": 0.4981,
      "step": 13244
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.8221794366836548,
      "learning_rate": 1.1880218038123434e-05,
      "loss": 0.4933,
      "step": 13245
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7482967376708984,
      "learning_rate": 1.1879121366648372e-05,
      "loss": 0.5008,
      "step": 13246
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5973318815231323,
      "learning_rate": 1.1878024671745596e-05,
      "loss": 0.5025,
      "step": 13247
    },
    {
      "epoch": 1.37,
      "grad_norm": 2.0825748443603516,
      "learning_rate": 1.1876927953428781e-05,
      "loss": 0.6259,
      "step": 13248
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.8324878215789795,
      "learning_rate": 1.1875831211711601e-05,
      "loss": 0.6523,
      "step": 13249
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6585783958435059,
      "learning_rate": 1.187473444660773e-05,
      "loss": 0.4915,
      "step": 13250
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.9575220346450806,
      "learning_rate": 1.1873637658130843e-05,
      "loss": 0.6829,
      "step": 13251
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.452508807182312,
      "learning_rate": 1.1872540846294609e-05,
      "loss": 0.6449,
      "step": 13252
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6364792585372925,
      "learning_rate": 1.1871444011112705e-05,
      "loss": 0.5506,
      "step": 13253
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3569021224975586,
      "learning_rate": 1.1870347152598809e-05,
      "loss": 0.611,
      "step": 13254
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.31673264503479,
      "learning_rate": 1.186925027076659e-05,
      "loss": 0.545,
      "step": 13255
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3162643909454346,
      "learning_rate": 1.186815336562973e-05,
      "loss": 0.4283,
      "step": 13256
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4446556568145752,
      "learning_rate": 1.1867056437201898e-05,
      "loss": 0.5906,
      "step": 13257
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7948670387268066,
      "learning_rate": 1.1865959485496772e-05,
      "loss": 0.6182,
      "step": 13258
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4696048498153687,
      "learning_rate": 1.186486251052803e-05,
      "loss": 0.5303,
      "step": 13259
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5764515399932861,
      "learning_rate": 1.1863765512309346e-05,
      "loss": 0.4722,
      "step": 13260
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5151804685592651,
      "learning_rate": 1.1862668490854399e-05,
      "loss": 0.5435,
      "step": 13261
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6004728078842163,
      "learning_rate": 1.1861571446176866e-05,
      "loss": 0.4461,
      "step": 13262
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5989090204238892,
      "learning_rate": 1.1860474378290422e-05,
      "loss": 0.5837,
      "step": 13263
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5312515497207642,
      "learning_rate": 1.1859377287208746e-05,
      "loss": 0.4404,
      "step": 13264
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6132714748382568,
      "learning_rate": 1.1858280172945513e-05,
      "loss": 0.592,
      "step": 13265
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.9665626287460327,
      "learning_rate": 1.1857183035514405e-05,
      "loss": 0.522,
      "step": 13266
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4620590209960938,
      "learning_rate": 1.1856085874929102e-05,
      "loss": 0.5511,
      "step": 13267
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.385014295578003,
      "learning_rate": 1.1854988691203276e-05,
      "loss": 0.537,
      "step": 13268
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.588894248008728,
      "learning_rate": 1.1853891484350613e-05,
      "loss": 0.6462,
      "step": 13269
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6232354640960693,
      "learning_rate": 1.1852794254384787e-05,
      "loss": 0.4846,
      "step": 13270
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.289919137954712,
      "learning_rate": 1.1851697001319478e-05,
      "loss": 0.5738,
      "step": 13271
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3609778881072998,
      "learning_rate": 1.1850599725168368e-05,
      "loss": 0.5064,
      "step": 13272
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5560253858566284,
      "learning_rate": 1.184950242594514e-05,
      "loss": 0.6013,
      "step": 13273
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.447052240371704,
      "learning_rate": 1.1848405103663467e-05,
      "loss": 0.5934,
      "step": 13274
    },
    {
      "epoch": 1.37,
      "grad_norm": 2.4089088439941406,
      "learning_rate": 1.1847307758337037e-05,
      "loss": 0.6468,
      "step": 13275
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.752915859222412,
      "learning_rate": 1.1846210389979526e-05,
      "loss": 0.4898,
      "step": 13276
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.7058290243148804,
      "learning_rate": 1.1845112998604618e-05,
      "loss": 0.68,
      "step": 13277
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.40580153465271,
      "learning_rate": 1.1844015584225995e-05,
      "loss": 0.5925,
      "step": 13278
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6248060464859009,
      "learning_rate": 1.1842918146857335e-05,
      "loss": 0.6046,
      "step": 13279
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5923941135406494,
      "learning_rate": 1.1841820686512323e-05,
      "loss": 0.522,
      "step": 13280
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5542898178100586,
      "learning_rate": 1.184072320320464e-05,
      "loss": 0.4505,
      "step": 13281
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5735751390457153,
      "learning_rate": 1.1839625696947971e-05,
      "loss": 0.6289,
      "step": 13282
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.8130042552947998,
      "learning_rate": 1.1838528167756e-05,
      "loss": 0.5645,
      "step": 13283
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3464720249176025,
      "learning_rate": 1.1837430615642406e-05,
      "loss": 0.4104,
      "step": 13284
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5108894109725952,
      "learning_rate": 1.1836333040620875e-05,
      "loss": 0.52,
      "step": 13285
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3437716960906982,
      "learning_rate": 1.1835235442705092e-05,
      "loss": 0.5123,
      "step": 13286
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.302491307258606,
      "learning_rate": 1.183413782190874e-05,
      "loss": 0.5153,
      "step": 13287
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.6524027585983276,
      "learning_rate": 1.1833040178245505e-05,
      "loss": 0.536,
      "step": 13288
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.552892804145813,
      "learning_rate": 1.1831942511729066e-05,
      "loss": 0.6873,
      "step": 13289
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5066022872924805,
      "learning_rate": 1.1830844822373113e-05,
      "loss": 0.583,
      "step": 13290
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3951469659805298,
      "learning_rate": 1.1829747110191332e-05,
      "loss": 0.4725,
      "step": 13291
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3725364208221436,
      "learning_rate": 1.1828649375197408e-05,
      "loss": 0.6207,
      "step": 13292
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.5124520063400269,
      "learning_rate": 1.1827551617405023e-05,
      "loss": 0.4863,
      "step": 13293
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.56387197971344,
      "learning_rate": 1.1826453836827866e-05,
      "loss": 0.6065,
      "step": 13294
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4290295839309692,
      "learning_rate": 1.1825356033479624e-05,
      "loss": 0.5311,
      "step": 13295
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6796667575836182,
      "learning_rate": 1.1824258207373987e-05,
      "loss": 0.6658,
      "step": 13296
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.8213882446289062,
      "learning_rate": 1.1823160358524635e-05,
      "loss": 0.6956,
      "step": 13297
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.8028955459594727,
      "learning_rate": 1.1822062486945258e-05,
      "loss": 0.5154,
      "step": 13298
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4030933380126953,
      "learning_rate": 1.1820964592649547e-05,
      "loss": 0.6875,
      "step": 13299
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4553629159927368,
      "learning_rate": 1.181986667565118e-05,
      "loss": 0.623,
      "step": 13300
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.576267123222351,
      "learning_rate": 1.181876873596386e-05,
      "loss": 0.5197,
      "step": 13301
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4113880395889282,
      "learning_rate": 1.1817670773601264e-05,
      "loss": 0.4918,
      "step": 13302
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4898993968963623,
      "learning_rate": 1.1816572788577082e-05,
      "loss": 0.5906,
      "step": 13303
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5414454936981201,
      "learning_rate": 1.1815474780905012e-05,
      "loss": 0.5074,
      "step": 13304
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5157407522201538,
      "learning_rate": 1.181437675059873e-05,
      "loss": 0.6364,
      "step": 13305
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5086867809295654,
      "learning_rate": 1.1813278697671935e-05,
      "loss": 0.6565,
      "step": 13306
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7036948204040527,
      "learning_rate": 1.1812180622138314e-05,
      "loss": 0.5207,
      "step": 13307
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6817961931228638,
      "learning_rate": 1.1811082524011554e-05,
      "loss": 0.5673,
      "step": 13308
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.826622486114502,
      "learning_rate": 1.1809984403305353e-05,
      "loss": 0.5213,
      "step": 13309
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4626563787460327,
      "learning_rate": 1.1808886260033394e-05,
      "loss": 0.4686,
      "step": 13310
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4503190517425537,
      "learning_rate": 1.1807788094209372e-05,
      "loss": 0.523,
      "step": 13311
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.9867957830429077,
      "learning_rate": 1.1806689905846978e-05,
      "loss": 0.6903,
      "step": 13312
    },
    {
      "epoch": 1.38,
      "grad_norm": 2.0487892627716064,
      "learning_rate": 1.1805591694959899e-05,
      "loss": 0.5368,
      "step": 13313
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6406890153884888,
      "learning_rate": 1.1804493461561837e-05,
      "loss": 0.6014,
      "step": 13314
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.640640377998352,
      "learning_rate": 1.1803395205666473e-05,
      "loss": 0.5231,
      "step": 13315
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7725428342819214,
      "learning_rate": 1.1802296927287504e-05,
      "loss": 0.5352,
      "step": 13316
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6054741144180298,
      "learning_rate": 1.1801198626438626e-05,
      "loss": 0.6966,
      "step": 13317
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4749627113342285,
      "learning_rate": 1.1800100303133526e-05,
      "loss": 0.5793,
      "step": 13318
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4846785068511963,
      "learning_rate": 1.17990019573859e-05,
      "loss": 0.5332,
      "step": 13319
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6977449655532837,
      "learning_rate": 1.1797903589209443e-05,
      "loss": 0.4849,
      "step": 13320
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6196242570877075,
      "learning_rate": 1.1796805198617844e-05,
      "loss": 0.5765,
      "step": 13321
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.2742979526519775,
      "learning_rate": 1.1795706785624804e-05,
      "loss": 0.461,
      "step": 13322
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.524052619934082,
      "learning_rate": 1.1794608350244013e-05,
      "loss": 0.5806,
      "step": 13323
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.9877698421478271,
      "learning_rate": 1.1793509892489166e-05,
      "loss": 0.5989,
      "step": 13324
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.634332299232483,
      "learning_rate": 1.1792411412373954e-05,
      "loss": 0.541,
      "step": 13325
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3758023977279663,
      "learning_rate": 1.179131290991208e-05,
      "loss": 0.5082,
      "step": 13326
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6025818586349487,
      "learning_rate": 1.1790214385117235e-05,
      "loss": 0.4509,
      "step": 13327
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3722753524780273,
      "learning_rate": 1.1789115838003115e-05,
      "loss": 0.4338,
      "step": 13328
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3120789527893066,
      "learning_rate": 1.1788017268583417e-05,
      "loss": 0.5504,
      "step": 13329
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5267409086227417,
      "learning_rate": 1.1786918676871834e-05,
      "loss": 0.5884,
      "step": 13330
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.441410779953003,
      "learning_rate": 1.1785820062882067e-05,
      "loss": 0.6111,
      "step": 13331
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4320471286773682,
      "learning_rate": 1.1784721426627814e-05,
      "loss": 0.5939,
      "step": 13332
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6826024055480957,
      "learning_rate": 1.1783622768122764e-05,
      "loss": 0.5523,
      "step": 13333
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.8154349327087402,
      "learning_rate": 1.1782524087380622e-05,
      "loss": 0.5474,
      "step": 13334
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3045814037322998,
      "learning_rate": 1.1781425384415084e-05,
      "loss": 0.6075,
      "step": 13335
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6405632495880127,
      "learning_rate": 1.1780326659239845e-05,
      "loss": 0.5416,
      "step": 13336
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4444564580917358,
      "learning_rate": 1.1779227911868608e-05,
      "loss": 0.6286,
      "step": 13337
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5680326223373413,
      "learning_rate": 1.1778129142315067e-05,
      "loss": 0.5215,
      "step": 13338
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4737406969070435,
      "learning_rate": 1.1777030350592922e-05,
      "loss": 0.5636,
      "step": 13339
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5069090127944946,
      "learning_rate": 1.1775931536715874e-05,
      "loss": 0.5997,
      "step": 13340
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3033019304275513,
      "learning_rate": 1.1774832700697618e-05,
      "loss": 0.4302,
      "step": 13341
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3958641290664673,
      "learning_rate": 1.1773733842551859e-05,
      "loss": 0.5843,
      "step": 13342
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5407830476760864,
      "learning_rate": 1.1772634962292296e-05,
      "loss": 0.4814,
      "step": 13343
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5671226978302002,
      "learning_rate": 1.1771536059932625e-05,
      "loss": 0.6339,
      "step": 13344
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.567287564277649,
      "learning_rate": 1.1770437135486552e-05,
      "loss": 0.5521,
      "step": 13345
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3173589706420898,
      "learning_rate": 1.1769338188967772e-05,
      "loss": 0.5071,
      "step": 13346
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4424926042556763,
      "learning_rate": 1.1768239220389987e-05,
      "loss": 0.4679,
      "step": 13347
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.702772855758667,
      "learning_rate": 1.1767140229766905e-05,
      "loss": 0.4054,
      "step": 13348
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4069541692733765,
      "learning_rate": 1.1766041217112218e-05,
      "loss": 0.6325,
      "step": 13349
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4187254905700684,
      "learning_rate": 1.1764942182439636e-05,
      "loss": 0.4341,
      "step": 13350
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7576897144317627,
      "learning_rate": 1.1763843125762854e-05,
      "loss": 0.5284,
      "step": 13351
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4485774040222168,
      "learning_rate": 1.1762744047095578e-05,
      "loss": 0.4803,
      "step": 13352
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4801561832427979,
      "learning_rate": 1.1761644946451513e-05,
      "loss": 0.4741,
      "step": 13353
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6177399158477783,
      "learning_rate": 1.1760545823844357e-05,
      "loss": 0.4679,
      "step": 13354
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5207668542861938,
      "learning_rate": 1.1759446679287817e-05,
      "loss": 0.6121,
      "step": 13355
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7159292697906494,
      "learning_rate": 1.1758347512795594e-05,
      "loss": 0.6316,
      "step": 13356
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.2873241901397705,
      "learning_rate": 1.1757248324381392e-05,
      "loss": 0.5713,
      "step": 13357
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.2775007486343384,
      "learning_rate": 1.1756149114058917e-05,
      "loss": 0.5011,
      "step": 13358
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5266757011413574,
      "learning_rate": 1.1755049881841872e-05,
      "loss": 0.6732,
      "step": 13359
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5449097156524658,
      "learning_rate": 1.1753950627743957e-05,
      "loss": 0.5457,
      "step": 13360
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.8851467370986938,
      "learning_rate": 1.1752851351778887e-05,
      "loss": 0.5442,
      "step": 13361
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6960620880126953,
      "learning_rate": 1.1751752053960356e-05,
      "loss": 0.5748,
      "step": 13362
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.496719479560852,
      "learning_rate": 1.1750652734302082e-05,
      "loss": 0.6917,
      "step": 13363
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7836169004440308,
      "learning_rate": 1.1749553392817757e-05,
      "loss": 0.5663,
      "step": 13364
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6585850715637207,
      "learning_rate": 1.1748454029521094e-05,
      "loss": 0.6933,
      "step": 13365
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4884389638900757,
      "learning_rate": 1.1747354644425802e-05,
      "loss": 0.4707,
      "step": 13366
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.40127432346344,
      "learning_rate": 1.1746255237545579e-05,
      "loss": 0.5642,
      "step": 13367
    },
    {
      "epoch": 1.38,
      "grad_norm": 2.0074212551116943,
      "learning_rate": 1.1745155808894138e-05,
      "loss": 0.4343,
      "step": 13368
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5706132650375366,
      "learning_rate": 1.1744056358485187e-05,
      "loss": 0.6473,
      "step": 13369
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5830203294754028,
      "learning_rate": 1.1742956886332429e-05,
      "loss": 0.5761,
      "step": 13370
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.697582721710205,
      "learning_rate": 1.1741857392449574e-05,
      "loss": 0.5536,
      "step": 13371
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.7847139835357666,
      "learning_rate": 1.1740757876850327e-05,
      "loss": 0.6557,
      "step": 13372
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6068142652511597,
      "learning_rate": 1.1739658339548399e-05,
      "loss": 0.5701,
      "step": 13373
    },
    {
      "epoch": 1.38,
      "grad_norm": 2.0939745903015137,
      "learning_rate": 1.1738558780557498e-05,
      "loss": 0.579,
      "step": 13374
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.573265790939331,
      "learning_rate": 1.1737459199891333e-05,
      "loss": 0.66,
      "step": 13375
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3559987545013428,
      "learning_rate": 1.173635959756361e-05,
      "loss": 0.4962,
      "step": 13376
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.594162106513977,
      "learning_rate": 1.1735259973588042e-05,
      "loss": 0.5795,
      "step": 13377
    },
    {
      "epoch": 1.38,
      "grad_norm": 2.163741111755371,
      "learning_rate": 1.1734160327978337e-05,
      "loss": 0.4246,
      "step": 13378
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.5276857614517212,
      "learning_rate": 1.1733060660748203e-05,
      "loss": 0.4364,
      "step": 13379
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4660340547561646,
      "learning_rate": 1.1731960971911351e-05,
      "loss": 0.5495,
      "step": 13380
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3320759534835815,
      "learning_rate": 1.1730861261481491e-05,
      "loss": 0.4944,
      "step": 13381
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4570574760437012,
      "learning_rate": 1.1729761529472337e-05,
      "loss": 0.4283,
      "step": 13382
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4844515323638916,
      "learning_rate": 1.1728661775897594e-05,
      "loss": 0.606,
      "step": 13383
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.436870813369751,
      "learning_rate": 1.1727562000770978e-05,
      "loss": 0.5019,
      "step": 13384
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3543387651443481,
      "learning_rate": 1.1726462204106195e-05,
      "loss": 0.4565,
      "step": 13385
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.414496898651123,
      "learning_rate": 1.1725362385916961e-05,
      "loss": 0.557,
      "step": 13386
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4236295223236084,
      "learning_rate": 1.1724262546216989e-05,
      "loss": 0.5439,
      "step": 13387
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.311493992805481,
      "learning_rate": 1.1723162685019986e-05,
      "loss": 0.459,
      "step": 13388
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.6165212392807007,
      "learning_rate": 1.1722062802339669e-05,
      "loss": 0.5477,
      "step": 13389
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.4005467891693115,
      "learning_rate": 1.1720962898189747e-05,
      "loss": 0.5522,
      "step": 13390
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3514317274093628,
      "learning_rate": 1.1719862972583934e-05,
      "loss": 0.6236,
      "step": 13391
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6078587770462036,
      "learning_rate": 1.1718763025535947e-05,
      "loss": 0.562,
      "step": 13392
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3334144353866577,
      "learning_rate": 1.1717663057059492e-05,
      "loss": 0.5814,
      "step": 13393
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7665550708770752,
      "learning_rate": 1.1716563067168289e-05,
      "loss": 0.5135,
      "step": 13394
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7959226369857788,
      "learning_rate": 1.171546305587605e-05,
      "loss": 0.6064,
      "step": 13395
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5207419395446777,
      "learning_rate": 1.1714363023196488e-05,
      "loss": 0.412,
      "step": 13396
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5298725366592407,
      "learning_rate": 1.1713262969143322e-05,
      "loss": 0.4998,
      "step": 13397
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4069111347198486,
      "learning_rate": 1.171216289373026e-05,
      "loss": 0.4758,
      "step": 13398
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.2538689374923706,
      "learning_rate": 1.1711062796971023e-05,
      "loss": 0.548,
      "step": 13399
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4373654127120972,
      "learning_rate": 1.1709962678879324e-05,
      "loss": 0.6244,
      "step": 13400
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4245595932006836,
      "learning_rate": 1.1708862539468875e-05,
      "loss": 0.4838,
      "step": 13401
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.88699209690094,
      "learning_rate": 1.17077623787534e-05,
      "loss": 0.5697,
      "step": 13402
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7134512662887573,
      "learning_rate": 1.1706662196746608e-05,
      "loss": 0.5312,
      "step": 13403
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4062861204147339,
      "learning_rate": 1.1705561993462215e-05,
      "loss": 0.5634,
      "step": 13404
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6021918058395386,
      "learning_rate": 1.1704461768913946e-05,
      "loss": 0.5795,
      "step": 13405
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3761825561523438,
      "learning_rate": 1.1703361523115509e-05,
      "loss": 0.4235,
      "step": 13406
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.848333477973938,
      "learning_rate": 1.1702261256080625e-05,
      "loss": 0.5708,
      "step": 13407
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6684859991073608,
      "learning_rate": 1.1701160967823013e-05,
      "loss": 0.5849,
      "step": 13408
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5581551790237427,
      "learning_rate": 1.1700060658356384e-05,
      "loss": 0.563,
      "step": 13409
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.8138365745544434,
      "learning_rate": 1.1698960327694467e-05,
      "loss": 0.5981,
      "step": 13410
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7408491373062134,
      "learning_rate": 1.169785997585097e-05,
      "loss": 0.5383,
      "step": 13411
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4561686515808105,
      "learning_rate": 1.1696759602839616e-05,
      "loss": 0.4516,
      "step": 13412
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.435538649559021,
      "learning_rate": 1.1695659208674124e-05,
      "loss": 0.5468,
      "step": 13413
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.9205902814865112,
      "learning_rate": 1.169455879336821e-05,
      "loss": 0.6167,
      "step": 13414
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.2825491428375244,
      "learning_rate": 1.1693458356935599e-05,
      "loss": 0.6829,
      "step": 13415
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.558828592300415,
      "learning_rate": 1.1692357899390003e-05,
      "loss": 0.5774,
      "step": 13416
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6010403633117676,
      "learning_rate": 1.1691257420745148e-05,
      "loss": 0.4863,
      "step": 13417
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5374035835266113,
      "learning_rate": 1.1690156921014753e-05,
      "loss": 0.5993,
      "step": 13418
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.508920669555664,
      "learning_rate": 1.1689056400212535e-05,
      "loss": 0.608,
      "step": 13419
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3855853080749512,
      "learning_rate": 1.1687955858352215e-05,
      "loss": 0.4983,
      "step": 13420
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3403947353363037,
      "learning_rate": 1.1686855295447521e-05,
      "loss": 0.5434,
      "step": 13421
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.244164228439331,
      "learning_rate": 1.1685754711512167e-05,
      "loss": 0.6187,
      "step": 13422
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.662882924079895,
      "learning_rate": 1.1684654106559876e-05,
      "loss": 0.5922,
      "step": 13423
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7575238943099976,
      "learning_rate": 1.168355348060437e-05,
      "loss": 0.5437,
      "step": 13424
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.615657091140747,
      "learning_rate": 1.168245283365937e-05,
      "loss": 0.5826,
      "step": 13425
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4227943420410156,
      "learning_rate": 1.1681352165738601e-05,
      "loss": 0.524,
      "step": 13426
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7506542205810547,
      "learning_rate": 1.1680251476855781e-05,
      "loss": 0.5395,
      "step": 13427
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3375370502471924,
      "learning_rate": 1.1679150767024639e-05,
      "loss": 0.5281,
      "step": 13428
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3144906759262085,
      "learning_rate": 1.167805003625889e-05,
      "loss": 0.5544,
      "step": 13429
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4129866361618042,
      "learning_rate": 1.1676949284572261e-05,
      "loss": 0.5526,
      "step": 13430
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5709733963012695,
      "learning_rate": 1.1675848511978481e-05,
      "loss": 0.5856,
      "step": 13431
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4257694482803345,
      "learning_rate": 1.1674747718491265e-05,
      "loss": 0.588,
      "step": 13432
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7663204669952393,
      "learning_rate": 1.1673646904124339e-05,
      "loss": 0.493,
      "step": 13433
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3352147340774536,
      "learning_rate": 1.1672546068891432e-05,
      "loss": 0.5076,
      "step": 13434
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.2386833429336548,
      "learning_rate": 1.1671445212806263e-05,
      "loss": 0.4007,
      "step": 13435
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.544268012046814,
      "learning_rate": 1.1670344335882561e-05,
      "loss": 0.4555,
      "step": 13436
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5381227731704712,
      "learning_rate": 1.1669243438134047e-05,
      "loss": 0.5366,
      "step": 13437
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4493279457092285,
      "learning_rate": 1.1668142519574451e-05,
      "loss": 0.4623,
      "step": 13438
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6034698486328125,
      "learning_rate": 1.1667041580217497e-05,
      "loss": 0.5799,
      "step": 13439
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4073015451431274,
      "learning_rate": 1.1665940620076905e-05,
      "loss": 0.4871,
      "step": 13440
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6661419868469238,
      "learning_rate": 1.1664839639166412e-05,
      "loss": 0.4883,
      "step": 13441
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4759052991867065,
      "learning_rate": 1.1663738637499736e-05,
      "loss": 0.6315,
      "step": 13442
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5919185876846313,
      "learning_rate": 1.1662637615090603e-05,
      "loss": 0.5491,
      "step": 13443
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.646405816078186,
      "learning_rate": 1.1661536571952747e-05,
      "loss": 0.3952,
      "step": 13444
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.2112984657287598,
      "learning_rate": 1.166043550809989e-05,
      "loss": 0.4356,
      "step": 13445
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.280662178993225,
      "learning_rate": 1.1659334423545759e-05,
      "loss": 0.4046,
      "step": 13446
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6715724468231201,
      "learning_rate": 1.1658233318304084e-05,
      "loss": 0.6385,
      "step": 13447
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7344650030136108,
      "learning_rate": 1.1657132192388592e-05,
      "loss": 0.6179,
      "step": 13448
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.8673831224441528,
      "learning_rate": 1.165603104581301e-05,
      "loss": 0.6779,
      "step": 13449
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7307246923446655,
      "learning_rate": 1.1654929878591067e-05,
      "loss": 0.5569,
      "step": 13450
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4667257070541382,
      "learning_rate": 1.1653828690736493e-05,
      "loss": 0.4526,
      "step": 13451
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6960874795913696,
      "learning_rate": 1.1652727482263016e-05,
      "loss": 0.5424,
      "step": 13452
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4552887678146362,
      "learning_rate": 1.1651626253184365e-05,
      "loss": 0.6,
      "step": 13453
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4608643054962158,
      "learning_rate": 1.1650525003514268e-05,
      "loss": 0.6011,
      "step": 13454
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5159372091293335,
      "learning_rate": 1.1649423733266458e-05,
      "loss": 0.4665,
      "step": 13455
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7378336191177368,
      "learning_rate": 1.1648322442454663e-05,
      "loss": 0.5237,
      "step": 13456
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.853391408920288,
      "learning_rate": 1.1647221131092613e-05,
      "loss": 0.7814,
      "step": 13457
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.531442403793335,
      "learning_rate": 1.164611979919404e-05,
      "loss": 0.5502,
      "step": 13458
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5904167890548706,
      "learning_rate": 1.1645018446772672e-05,
      "loss": 0.5621,
      "step": 13459
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6541197299957275,
      "learning_rate": 1.1643917073842244e-05,
      "loss": 0.4033,
      "step": 13460
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5567517280578613,
      "learning_rate": 1.1642815680416482e-05,
      "loss": 0.5698,
      "step": 13461
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.664271354675293,
      "learning_rate": 1.164171426650912e-05,
      "loss": 0.6367,
      "step": 13462
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3321342468261719,
      "learning_rate": 1.1640612832133893e-05,
      "loss": 0.4694,
      "step": 13463
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6102651357650757,
      "learning_rate": 1.1639511377304529e-05,
      "loss": 0.5326,
      "step": 13464
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4356998205184937,
      "learning_rate": 1.1638409902034762e-05,
      "loss": 0.5604,
      "step": 13465
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4606667757034302,
      "learning_rate": 1.1637308406338322e-05,
      "loss": 0.6028,
      "step": 13466
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.127056837081909,
      "learning_rate": 1.1636206890228943e-05,
      "loss": 0.6375,
      "step": 13467
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.7639559507369995,
      "learning_rate": 1.1635105353720363e-05,
      "loss": 0.5084,
      "step": 13468
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.516008973121643,
      "learning_rate": 1.1634003796826305e-05,
      "loss": 0.5244,
      "step": 13469
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4416639804840088,
      "learning_rate": 1.1632902219560509e-05,
      "loss": 0.6215,
      "step": 13470
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.64388108253479,
      "learning_rate": 1.1631800621936712e-05,
      "loss": 0.6342,
      "step": 13471
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.399945616722107,
      "learning_rate": 1.1630699003968642e-05,
      "loss": 0.546,
      "step": 13472
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.9074422121047974,
      "learning_rate": 1.1629597365670037e-05,
      "loss": 0.4326,
      "step": 13473
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4239225387573242,
      "learning_rate": 1.1628495707054627e-05,
      "loss": 0.5185,
      "step": 13474
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6708424091339111,
      "learning_rate": 1.1627394028136153e-05,
      "loss": 0.5406,
      "step": 13475
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5094228982925415,
      "learning_rate": 1.1626292328928346e-05,
      "loss": 0.6329,
      "step": 13476
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5680837631225586,
      "learning_rate": 1.1625190609444942e-05,
      "loss": 0.447,
      "step": 13477
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4034746885299683,
      "learning_rate": 1.1624088869699674e-05,
      "loss": 0.5167,
      "step": 13478
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.5864149332046509,
      "learning_rate": 1.1622987109706282e-05,
      "loss": 0.4348,
      "step": 13479
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.425150752067566,
      "learning_rate": 1.1621885329478502e-05,
      "loss": 0.622,
      "step": 13480
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4587057828903198,
      "learning_rate": 1.162078352903007e-05,
      "loss": 0.6511,
      "step": 13481
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4880045652389526,
      "learning_rate": 1.1619681708374717e-05,
      "loss": 0.5818,
      "step": 13482
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.778870940208435,
      "learning_rate": 1.1618579867526188e-05,
      "loss": 0.5253,
      "step": 13483
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.6224174499511719,
      "learning_rate": 1.1617478006498215e-05,
      "loss": 0.6954,
      "step": 13484
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3832550048828125,
      "learning_rate": 1.1616376125304535e-05,
      "loss": 0.6694,
      "step": 13485
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4298007488250732,
      "learning_rate": 1.161527422395889e-05,
      "loss": 0.6034,
      "step": 13486
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.4724305868148804,
      "learning_rate": 1.1614172302475013e-05,
      "loss": 0.5826,
      "step": 13487
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3976819515228271,
      "learning_rate": 1.1613070360866646e-05,
      "loss": 0.5384,
      "step": 13488
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.029736042022705,
      "learning_rate": 1.1611968399147525e-05,
      "loss": 0.6067,
      "step": 13489
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4360250234603882,
      "learning_rate": 1.1610866417331386e-05,
      "loss": 0.497,
      "step": 13490
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.823704481124878,
      "learning_rate": 1.1609764415431975e-05,
      "loss": 0.6235,
      "step": 13491
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4145561456680298,
      "learning_rate": 1.1608662393463024e-05,
      "loss": 0.4669,
      "step": 13492
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.683525800704956,
      "learning_rate": 1.1607560351438274e-05,
      "loss": 0.6273,
      "step": 13493
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7178624868392944,
      "learning_rate": 1.160645828937147e-05,
      "loss": 0.6117,
      "step": 13494
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3725169897079468,
      "learning_rate": 1.1605356207276344e-05,
      "loss": 0.5625,
      "step": 13495
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5616308450698853,
      "learning_rate": 1.160425410516664e-05,
      "loss": 0.5472,
      "step": 13496
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.528397560119629,
      "learning_rate": 1.16031519830561e-05,
      "loss": 0.614,
      "step": 13497
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5739563703536987,
      "learning_rate": 1.160204984095846e-05,
      "loss": 0.5289,
      "step": 13498
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3073973655700684,
      "learning_rate": 1.1600947678887464e-05,
      "loss": 0.5828,
      "step": 13499
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.9411559104919434,
      "learning_rate": 1.1599845496856852e-05,
      "loss": 0.5545,
      "step": 13500
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8557357788085938,
      "learning_rate": 1.1598743294880367e-05,
      "loss": 0.4761,
      "step": 13501
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.675676703453064,
      "learning_rate": 1.159764107297175e-05,
      "loss": 0.6357,
      "step": 13502
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3491158485412598,
      "learning_rate": 1.159653883114474e-05,
      "loss": 0.6176,
      "step": 13503
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4991923570632935,
      "learning_rate": 1.1595436569413084e-05,
      "loss": 0.5452,
      "step": 13504
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5478222370147705,
      "learning_rate": 1.1594334287790518e-05,
      "loss": 0.568,
      "step": 13505
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4278076887130737,
      "learning_rate": 1.1593231986290789e-05,
      "loss": 0.5789,
      "step": 13506
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5143059492111206,
      "learning_rate": 1.1592129664927641e-05,
      "loss": 0.5559,
      "step": 13507
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4954478740692139,
      "learning_rate": 1.1591027323714812e-05,
      "loss": 0.5413,
      "step": 13508
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5753287076950073,
      "learning_rate": 1.1589924962666048e-05,
      "loss": 0.4381,
      "step": 13509
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7560824155807495,
      "learning_rate": 1.1588822581795097e-05,
      "loss": 0.6056,
      "step": 13510
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5194125175476074,
      "learning_rate": 1.1587720181115694e-05,
      "loss": 0.6054,
      "step": 13511
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7307437658309937,
      "learning_rate": 1.1586617760641592e-05,
      "loss": 0.661,
      "step": 13512
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5094499588012695,
      "learning_rate": 1.1585515320386527e-05,
      "loss": 0.5316,
      "step": 13513
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8027681112289429,
      "learning_rate": 1.1584412860364246e-05,
      "loss": 0.5423,
      "step": 13514
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5447059869766235,
      "learning_rate": 1.1583310380588499e-05,
      "loss": 0.52,
      "step": 13515
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2719151973724365,
      "learning_rate": 1.1582207881073026e-05,
      "loss": 0.541,
      "step": 13516
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7374707460403442,
      "learning_rate": 1.1581105361831576e-05,
      "loss": 0.6174,
      "step": 13517
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5263582468032837,
      "learning_rate": 1.1580002822877886e-05,
      "loss": 0.6357,
      "step": 13518
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.538019061088562,
      "learning_rate": 1.1578900264225713e-05,
      "loss": 0.557,
      "step": 13519
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.9908347129821777,
      "learning_rate": 1.1577797685888796e-05,
      "loss": 0.5765,
      "step": 13520
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3924058675765991,
      "learning_rate": 1.1576695087880883e-05,
      "loss": 0.4401,
      "step": 13521
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5392433404922485,
      "learning_rate": 1.157559247021572e-05,
      "loss": 0.659,
      "step": 13522
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4292207956314087,
      "learning_rate": 1.1574489832907054e-05,
      "loss": 0.6387,
      "step": 13523
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4910496473312378,
      "learning_rate": 1.1573387175968633e-05,
      "loss": 0.5524,
      "step": 13524
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8335964679718018,
      "learning_rate": 1.1572284499414206e-05,
      "loss": 0.588,
      "step": 13525
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6372469663619995,
      "learning_rate": 1.1571181803257514e-05,
      "loss": 0.6137,
      "step": 13526
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.602698802947998,
      "learning_rate": 1.1570079087512309e-05,
      "loss": 0.5876,
      "step": 13527
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3747491836547852,
      "learning_rate": 1.1568976352192342e-05,
      "loss": 0.5305,
      "step": 13528
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7194573879241943,
      "learning_rate": 1.1567873597311357e-05,
      "loss": 0.527,
      "step": 13529
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8444325923919678,
      "learning_rate": 1.1566770822883101e-05,
      "loss": 0.671,
      "step": 13530
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2898529767990112,
      "learning_rate": 1.1565668028921326e-05,
      "loss": 0.4719,
      "step": 13531
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3026955127716064,
      "learning_rate": 1.1564565215439781e-05,
      "loss": 0.6256,
      "step": 13532
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3978289365768433,
      "learning_rate": 1.1563462382452214e-05,
      "loss": 0.5042,
      "step": 13533
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4147909879684448,
      "learning_rate": 1.1562359529972375e-05,
      "loss": 0.5691,
      "step": 13534
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5175285339355469,
      "learning_rate": 1.1561256658014011e-05,
      "loss": 0.6243,
      "step": 13535
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5827161073684692,
      "learning_rate": 1.1560153766590878e-05,
      "loss": 0.6275,
      "step": 13536
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5455715656280518,
      "learning_rate": 1.155905085571672e-05,
      "loss": 0.4618,
      "step": 13537
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6511108875274658,
      "learning_rate": 1.1557947925405291e-05,
      "loss": 0.6066,
      "step": 13538
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1707416772842407,
      "learning_rate": 1.1556844975670338e-05,
      "loss": 0.5434,
      "step": 13539
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4148263931274414,
      "learning_rate": 1.1555742006525616e-05,
      "loss": 0.4859,
      "step": 13540
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4499846696853638,
      "learning_rate": 1.1554639017984877e-05,
      "loss": 0.4943,
      "step": 13541
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.587479829788208,
      "learning_rate": 1.1553536010061866e-05,
      "loss": 0.5887,
      "step": 13542
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6994903087615967,
      "learning_rate": 1.155243298277034e-05,
      "loss": 0.4954,
      "step": 13543
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7459090948104858,
      "learning_rate": 1.1551329936124049e-05,
      "loss": 0.5761,
      "step": 13544
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3215047121047974,
      "learning_rate": 1.1550226870136744e-05,
      "loss": 0.5095,
      "step": 13545
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.587159514427185,
      "learning_rate": 1.1549123784822183e-05,
      "loss": 0.5587,
      "step": 13546
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3787435293197632,
      "learning_rate": 1.1548020680194113e-05,
      "loss": 0.5659,
      "step": 13547
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3782227039337158,
      "learning_rate": 1.1546917556266284e-05,
      "loss": 0.5443,
      "step": 13548
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6364312171936035,
      "learning_rate": 1.1545814413052456e-05,
      "loss": 0.548,
      "step": 13549
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8211556673049927,
      "learning_rate": 1.1544711250566381e-05,
      "loss": 0.5535,
      "step": 13550
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4686667919158936,
      "learning_rate": 1.154360806882181e-05,
      "loss": 0.5406,
      "step": 13551
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5975275039672852,
      "learning_rate": 1.1542504867832496e-05,
      "loss": 0.6076,
      "step": 13552
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.499246597290039,
      "learning_rate": 1.1541401647612196e-05,
      "loss": 0.4739,
      "step": 13553
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6495071649551392,
      "learning_rate": 1.1540298408174664e-05,
      "loss": 0.6601,
      "step": 13554
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5076627731323242,
      "learning_rate": 1.1539195149533652e-05,
      "loss": 0.5499,
      "step": 13555
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6944999694824219,
      "learning_rate": 1.1538091871702918e-05,
      "loss": 0.572,
      "step": 13556
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6141225099563599,
      "learning_rate": 1.1536988574696215e-05,
      "loss": 0.5815,
      "step": 13557
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5340754985809326,
      "learning_rate": 1.1535885258527298e-05,
      "loss": 0.5876,
      "step": 13558
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5075700283050537,
      "learning_rate": 1.1534781923209923e-05,
      "loss": 0.5728,
      "step": 13559
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5514367818832397,
      "learning_rate": 1.1533678568757848e-05,
      "loss": 0.4716,
      "step": 13560
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5912238359451294,
      "learning_rate": 1.1532575195184823e-05,
      "loss": 0.5572,
      "step": 13561
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8465516567230225,
      "learning_rate": 1.153147180250461e-05,
      "loss": 0.5541,
      "step": 13562
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7461940050125122,
      "learning_rate": 1.1530368390730962e-05,
      "loss": 0.6172,
      "step": 13563
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4409778118133545,
      "learning_rate": 1.1529264959877639e-05,
      "loss": 0.6399,
      "step": 13564
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5209016799926758,
      "learning_rate": 1.1528161509958392e-05,
      "loss": 0.5057,
      "step": 13565
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6389687061309814,
      "learning_rate": 1.1527058040986984e-05,
      "loss": 0.586,
      "step": 13566
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4088091850280762,
      "learning_rate": 1.1525954552977173e-05,
      "loss": 0.5872,
      "step": 13567
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5121636390686035,
      "learning_rate": 1.1524851045942711e-05,
      "loss": 0.5805,
      "step": 13568
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.361893653869629,
      "learning_rate": 1.1523747519897359e-05,
      "loss": 0.5262,
      "step": 13569
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1955897808074951,
      "learning_rate": 1.1522643974854874e-05,
      "loss": 0.5459,
      "step": 13570
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5464065074920654,
      "learning_rate": 1.1521540410829015e-05,
      "loss": 0.5908,
      "step": 13571
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5623993873596191,
      "learning_rate": 1.1520436827833544e-05,
      "loss": 0.4951,
      "step": 13572
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.6187034845352173,
      "learning_rate": 1.1519333225882212e-05,
      "loss": 0.5494,
      "step": 13573
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.8035277128219604,
      "learning_rate": 1.1518229604988782e-05,
      "loss": 0.5301,
      "step": 13574
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.391815185546875,
      "learning_rate": 1.1517125965167018e-05,
      "loss": 0.5602,
      "step": 13575
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3391644954681396,
      "learning_rate": 1.1516022306430672e-05,
      "loss": 0.4515,
      "step": 13576
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.7953295707702637,
      "learning_rate": 1.151491862879351e-05,
      "loss": 0.5842,
      "step": 13577
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4027047157287598,
      "learning_rate": 1.1513814932269285e-05,
      "loss": 0.5195,
      "step": 13578
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.591509222984314,
      "learning_rate": 1.1512711216871762e-05,
      "loss": 0.5306,
      "step": 13579
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.468493938446045,
      "learning_rate": 1.1511607482614702e-05,
      "loss": 0.5212,
      "step": 13580
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.332131266593933,
      "learning_rate": 1.1510503729511863e-05,
      "loss": 0.4919,
      "step": 13581
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3743703365325928,
      "learning_rate": 1.150939995757701e-05,
      "loss": 0.5044,
      "step": 13582
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4964996576309204,
      "learning_rate": 1.1508296166823897e-05,
      "loss": 0.5242,
      "step": 13583
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.434000849723816,
      "learning_rate": 1.150719235726629e-05,
      "loss": 0.6198,
      "step": 13584
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1523644924163818,
      "learning_rate": 1.1506088528917954e-05,
      "loss": 0.4368,
      "step": 13585
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5555592775344849,
      "learning_rate": 1.1504984681792642e-05,
      "loss": 0.5284,
      "step": 13586
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4472671747207642,
      "learning_rate": 1.1503880815904123e-05,
      "loss": 0.5376,
      "step": 13587
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3431637287139893,
      "learning_rate": 1.150277693126616e-05,
      "loss": 0.5534,
      "step": 13588
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.9043488502502441,
      "learning_rate": 1.150167302789251e-05,
      "loss": 0.5419,
      "step": 13589
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.9759529829025269,
      "learning_rate": 1.150056910579694e-05,
      "loss": 0.5997,
      "step": 13590
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7465883493423462,
      "learning_rate": 1.149946516499321e-05,
      "loss": 0.5392,
      "step": 13591
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.451972484588623,
      "learning_rate": 1.1498361205495087e-05,
      "loss": 0.5508,
      "step": 13592
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5401358604431152,
      "learning_rate": 1.1497257227316331e-05,
      "loss": 0.5159,
      "step": 13593
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5406156778335571,
      "learning_rate": 1.1496153230470707e-05,
      "loss": 0.4444,
      "step": 13594
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7473691701889038,
      "learning_rate": 1.1495049214971983e-05,
      "loss": 0.5961,
      "step": 13595
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.696101427078247,
      "learning_rate": 1.1493945180833916e-05,
      "loss": 0.489,
      "step": 13596
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.542833685874939,
      "learning_rate": 1.1492841128070274e-05,
      "loss": 0.6362,
      "step": 13597
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5201812982559204,
      "learning_rate": 1.1491737056694822e-05,
      "loss": 0.6096,
      "step": 13598
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4580786228179932,
      "learning_rate": 1.1490632966721322e-05,
      "loss": 0.491,
      "step": 13599
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.331055998802185,
      "learning_rate": 1.1489528858163544e-05,
      "loss": 0.5654,
      "step": 13600
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.492242693901062,
      "learning_rate": 1.148842473103525e-05,
      "loss": 0.5104,
      "step": 13601
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3003367185592651,
      "learning_rate": 1.1487320585350203e-05,
      "loss": 0.5351,
      "step": 13602
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8194580078125,
      "learning_rate": 1.1486216421122178e-05,
      "loss": 0.5286,
      "step": 13603
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3769941329956055,
      "learning_rate": 1.148511223836493e-05,
      "loss": 0.6349,
      "step": 13604
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.67780601978302,
      "learning_rate": 1.148400803709223e-05,
      "loss": 0.5943,
      "step": 13605
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5224186182022095,
      "learning_rate": 1.148290381731785e-05,
      "loss": 0.5176,
      "step": 13606
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.487869381904602,
      "learning_rate": 1.1481799579055548e-05,
      "loss": 0.5844,
      "step": 13607
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.659472942352295,
      "learning_rate": 1.1480695322319096e-05,
      "loss": 0.5562,
      "step": 13608
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4155998229980469,
      "learning_rate": 1.1479591047122259e-05,
      "loss": 0.469,
      "step": 13609
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6072320938110352,
      "learning_rate": 1.1478486753478804e-05,
      "loss": 0.5644,
      "step": 13610
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7166216373443604,
      "learning_rate": 1.14773824414025e-05,
      "loss": 0.4904,
      "step": 13611
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5615609884262085,
      "learning_rate": 1.1476278110907115e-05,
      "loss": 0.5565,
      "step": 13612
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5619903802871704,
      "learning_rate": 1.1475173762006416e-05,
      "loss": 0.5012,
      "step": 13613
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3856422901153564,
      "learning_rate": 1.1474069394714173e-05,
      "loss": 0.4593,
      "step": 13614
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3167730569839478,
      "learning_rate": 1.1472965009044153e-05,
      "loss": 0.4882,
      "step": 13615
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.9991135597229004,
      "learning_rate": 1.1471860605010128e-05,
      "loss": 0.5291,
      "step": 13616
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4282853603363037,
      "learning_rate": 1.147075618262586e-05,
      "loss": 0.6629,
      "step": 13617
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8448342084884644,
      "learning_rate": 1.1469651741905124e-05,
      "loss": 0.5938,
      "step": 13618
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3280463218688965,
      "learning_rate": 1.1468547282861691e-05,
      "loss": 0.6221,
      "step": 13619
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2743799686431885,
      "learning_rate": 1.1467442805509323e-05,
      "loss": 0.4916,
      "step": 13620
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2989778518676758,
      "learning_rate": 1.1466338309861799e-05,
      "loss": 0.5287,
      "step": 13621
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6355639696121216,
      "learning_rate": 1.1465233795932886e-05,
      "loss": 0.548,
      "step": 13622
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6636152267456055,
      "learning_rate": 1.146412926373635e-05,
      "loss": 0.4391,
      "step": 13623
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4666764736175537,
      "learning_rate": 1.1463024713285966e-05,
      "loss": 0.5762,
      "step": 13624
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6964324712753296,
      "learning_rate": 1.1461920144595501e-05,
      "loss": 0.5124,
      "step": 13625
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3528279066085815,
      "learning_rate": 1.1460815557678733e-05,
      "loss": 0.4371,
      "step": 13626
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4679652452468872,
      "learning_rate": 1.1459710952549428e-05,
      "loss": 0.6039,
      "step": 13627
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7182461023330688,
      "learning_rate": 1.1458606329221358e-05,
      "loss": 0.4637,
      "step": 13628
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3283929824829102,
      "learning_rate": 1.1457501687708294e-05,
      "loss": 0.5185,
      "step": 13629
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6764777898788452,
      "learning_rate": 1.1456397028024012e-05,
      "loss": 0.5764,
      "step": 13630
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6205681562423706,
      "learning_rate": 1.1455292350182279e-05,
      "loss": 0.4657,
      "step": 13631
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2745962142944336,
      "learning_rate": 1.1454187654196873e-05,
      "loss": 0.5377,
      "step": 13632
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.325749158859253,
      "learning_rate": 1.145308294008156e-05,
      "loss": 0.6209,
      "step": 13633
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4822653532028198,
      "learning_rate": 1.1451978207850121e-05,
      "loss": 0.4864,
      "step": 13634
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4102219343185425,
      "learning_rate": 1.1450873457516322e-05,
      "loss": 0.5532,
      "step": 13635
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8176276683807373,
      "learning_rate": 1.1449768689093938e-05,
      "loss": 0.5915,
      "step": 13636
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5424740314483643,
      "learning_rate": 1.1448663902596748e-05,
      "loss": 0.5409,
      "step": 13637
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6701093912124634,
      "learning_rate": 1.1447559098038517e-05,
      "loss": 0.5619,
      "step": 13638
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6481088399887085,
      "learning_rate": 1.1446454275433026e-05,
      "loss": 0.6371,
      "step": 13639
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5924686193466187,
      "learning_rate": 1.1445349434794046e-05,
      "loss": 0.5034,
      "step": 13640
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8074309825897217,
      "learning_rate": 1.1444244576135351e-05,
      "loss": 0.4877,
      "step": 13641
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6029621362686157,
      "learning_rate": 1.1443139699470716e-05,
      "loss": 0.5507,
      "step": 13642
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4841949939727783,
      "learning_rate": 1.144203480481392e-05,
      "loss": 0.4819,
      "step": 13643
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.417667269706726,
      "learning_rate": 1.1440929892178731e-05,
      "loss": 0.5936,
      "step": 13644
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5385973453521729,
      "learning_rate": 1.1439824961578931e-05,
      "loss": 0.5776,
      "step": 13645
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6088311672210693,
      "learning_rate": 1.1438720013028293e-05,
      "loss": 0.531,
      "step": 13646
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.559220790863037,
      "learning_rate": 1.1437615046540591e-05,
      "loss": 0.4776,
      "step": 13647
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3578368425369263,
      "learning_rate": 1.1436510062129604e-05,
      "loss": 0.5003,
      "step": 13648
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3280445337295532,
      "learning_rate": 1.1435405059809104e-05,
      "loss": 0.4744,
      "step": 13649
    },
    {
      "epoch": 1.41,
      "grad_norm": 2.341954231262207,
      "learning_rate": 1.1434300039592875e-05,
      "loss": 0.5835,
      "step": 13650
    },
    {
      "epoch": 1.41,
      "grad_norm": 2.1403937339782715,
      "learning_rate": 1.1433195001494685e-05,
      "loss": 0.6265,
      "step": 13651
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7143669128417969,
      "learning_rate": 1.1432089945528315e-05,
      "loss": 0.5788,
      "step": 13652
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.540626883506775,
      "learning_rate": 1.1430984871707546e-05,
      "loss": 0.4693,
      "step": 13653
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.476664662361145,
      "learning_rate": 1.142987978004615e-05,
      "loss": 0.517,
      "step": 13654
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4676086902618408,
      "learning_rate": 1.1428774670557902e-05,
      "loss": 0.5492,
      "step": 13655
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4285017251968384,
      "learning_rate": 1.142766954325659e-05,
      "loss": 0.517,
      "step": 13656
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6846569776535034,
      "learning_rate": 1.1426564398155982e-05,
      "loss": 0.4724,
      "step": 13657
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3435598611831665,
      "learning_rate": 1.1425459235269863e-05,
      "loss": 0.5697,
      "step": 13658
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2956289052963257,
      "learning_rate": 1.1424354054612007e-05,
      "loss": 0.5283,
      "step": 13659
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7591969966888428,
      "learning_rate": 1.1423248856196193e-05,
      "loss": 0.6591,
      "step": 13660
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.491145133972168,
      "learning_rate": 1.1422143640036202e-05,
      "loss": 0.6176,
      "step": 13661
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6268811225891113,
      "learning_rate": 1.1421038406145812e-05,
      "loss": 0.4526,
      "step": 13662
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7986369132995605,
      "learning_rate": 1.1419933154538807e-05,
      "loss": 0.6144,
      "step": 13663
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4905089139938354,
      "learning_rate": 1.1418827885228958e-05,
      "loss": 0.5626,
      "step": 13664
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6473153829574585,
      "learning_rate": 1.141772259823005e-05,
      "loss": 0.48,
      "step": 13665
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.239462971687317,
      "learning_rate": 1.1416617293555865e-05,
      "loss": 0.5276,
      "step": 13666
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.342325210571289,
      "learning_rate": 1.1415511971220177e-05,
      "loss": 0.565,
      "step": 13667
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.8215680122375488,
      "learning_rate": 1.1414406631236769e-05,
      "loss": 0.6023,
      "step": 13668
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4760764837265015,
      "learning_rate": 1.1413301273619429e-05,
      "loss": 0.5195,
      "step": 13669
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5399452447891235,
      "learning_rate": 1.1412195898381926e-05,
      "loss": 0.5921,
      "step": 13670
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4931833744049072,
      "learning_rate": 1.141109050553805e-05,
      "loss": 0.6103,
      "step": 13671
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4810155630111694,
      "learning_rate": 1.1409985095101574e-05,
      "loss": 0.5211,
      "step": 13672
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.399454116821289,
      "learning_rate": 1.1408879667086287e-05,
      "loss": 0.5673,
      "step": 13673
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5698151588439941,
      "learning_rate": 1.1407774221505972e-05,
      "loss": 0.5346,
      "step": 13674
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.7458053827285767,
      "learning_rate": 1.1406668758374405e-05,
      "loss": 0.6404,
      "step": 13675
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.6965852975845337,
      "learning_rate": 1.140556327770537e-05,
      "loss": 0.5088,
      "step": 13676
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5501549243927002,
      "learning_rate": 1.140445777951265e-05,
      "loss": 0.5795,
      "step": 13677
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.732680082321167,
      "learning_rate": 1.1403352263810026e-05,
      "loss": 0.5929,
      "step": 13678
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5476999282836914,
      "learning_rate": 1.1402246730611286e-05,
      "loss": 0.6068,
      "step": 13679
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.5212172269821167,
      "learning_rate": 1.1401141179930209e-05,
      "loss": 0.506,
      "step": 13680
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.573054313659668,
      "learning_rate": 1.1400035611780576e-05,
      "loss": 0.4171,
      "step": 13681
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4024558067321777,
      "learning_rate": 1.1398930026176178e-05,
      "loss": 0.5014,
      "step": 13682
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.982652187347412,
      "learning_rate": 1.139782442313079e-05,
      "loss": 0.5455,
      "step": 13683
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.9848567247390747,
      "learning_rate": 1.1396718802658204e-05,
      "loss": 0.5918,
      "step": 13684
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5042935609817505,
      "learning_rate": 1.13956131647722e-05,
      "loss": 0.566,
      "step": 13685
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5050665140151978,
      "learning_rate": 1.1394507509486559e-05,
      "loss": 0.5733,
      "step": 13686
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7564266920089722,
      "learning_rate": 1.1393401836815075e-05,
      "loss": 0.4419,
      "step": 13687
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5714800357818604,
      "learning_rate": 1.1392296146771523e-05,
      "loss": 0.5879,
      "step": 13688
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.383363127708435,
      "learning_rate": 1.1391190439369694e-05,
      "loss": 0.4929,
      "step": 13689
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3004753589630127,
      "learning_rate": 1.139008471462337e-05,
      "loss": 0.5432,
      "step": 13690
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5253138542175293,
      "learning_rate": 1.138897897254634e-05,
      "loss": 0.6383,
      "step": 13691
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.644984483718872,
      "learning_rate": 1.138787321315239e-05,
      "loss": 0.5679,
      "step": 13692
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.591651439666748,
      "learning_rate": 1.1386767436455299e-05,
      "loss": 0.5324,
      "step": 13693
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6741420030593872,
      "learning_rate": 1.138566164246886e-05,
      "loss": 0.6305,
      "step": 13694
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.422813057899475,
      "learning_rate": 1.1384555831206858e-05,
      "loss": 0.5769,
      "step": 13695
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3705332279205322,
      "learning_rate": 1.1383450002683077e-05,
      "loss": 0.5563,
      "step": 13696
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.471096158027649,
      "learning_rate": 1.1382344156911307e-05,
      "loss": 0.5416,
      "step": 13697
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4214822053909302,
      "learning_rate": 1.1381238293905335e-05,
      "loss": 0.5166,
      "step": 13698
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3801159858703613,
      "learning_rate": 1.1380132413678942e-05,
      "loss": 0.4236,
      "step": 13699
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5543485879898071,
      "learning_rate": 1.1379026516245926e-05,
      "loss": 0.7315,
      "step": 13700
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4489797353744507,
      "learning_rate": 1.1377920601620063e-05,
      "loss": 0.6244,
      "step": 13701
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5906916856765747,
      "learning_rate": 1.1376814669815151e-05,
      "loss": 0.5904,
      "step": 13702
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7366541624069214,
      "learning_rate": 1.1375708720844971e-05,
      "loss": 0.5847,
      "step": 13703
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5516475439071655,
      "learning_rate": 1.1374602754723317e-05,
      "loss": 0.4569,
      "step": 13704
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4325143098831177,
      "learning_rate": 1.1373496771463972e-05,
      "loss": 0.4495,
      "step": 13705
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5751157999038696,
      "learning_rate": 1.1372390771080728e-05,
      "loss": 0.6476,
      "step": 13706
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4220032691955566,
      "learning_rate": 1.1371284753587373e-05,
      "loss": 0.5627,
      "step": 13707
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5050772428512573,
      "learning_rate": 1.1370178718997698e-05,
      "loss": 0.4569,
      "step": 13708
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6190897226333618,
      "learning_rate": 1.1369072667325486e-05,
      "loss": 0.6554,
      "step": 13709
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2558039426803589,
      "learning_rate": 1.1367966598584537e-05,
      "loss": 0.5,
      "step": 13710
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.392264723777771,
      "learning_rate": 1.136686051278863e-05,
      "loss": 0.566,
      "step": 13711
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.517923355102539,
      "learning_rate": 1.1365754409951564e-05,
      "loss": 0.5912,
      "step": 13712
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4434177875518799,
      "learning_rate": 1.1364648290087123e-05,
      "loss": 0.4099,
      "step": 13713
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6158112287521362,
      "learning_rate": 1.1363542153209099e-05,
      "loss": 0.5944,
      "step": 13714
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6931301355361938,
      "learning_rate": 1.1362435999331288e-05,
      "loss": 0.5119,
      "step": 13715
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5776187181472778,
      "learning_rate": 1.136132982846747e-05,
      "loss": 0.5589,
      "step": 13716
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.321398138999939,
      "learning_rate": 1.1360223640631443e-05,
      "loss": 0.3906,
      "step": 13717
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6426995992660522,
      "learning_rate": 1.1359117435836999e-05,
      "loss": 0.5656,
      "step": 13718
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6660586595535278,
      "learning_rate": 1.1358011214097929e-05,
      "loss": 0.4538,
      "step": 13719
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6107826232910156,
      "learning_rate": 1.135690497542802e-05,
      "loss": 0.5404,
      "step": 13720
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.604763150215149,
      "learning_rate": 1.135579871984107e-05,
      "loss": 0.456,
      "step": 13721
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.477239727973938,
      "learning_rate": 1.1354692447350866e-05,
      "loss": 0.5949,
      "step": 13722
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7735004425048828,
      "learning_rate": 1.1353586157971205e-05,
      "loss": 0.5496,
      "step": 13723
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5434602499008179,
      "learning_rate": 1.1352479851715878e-05,
      "loss": 0.4984,
      "step": 13724
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.35209059715271,
      "learning_rate": 1.1351373528598675e-05,
      "loss": 0.5745,
      "step": 13725
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8725405931472778,
      "learning_rate": 1.1350267188633392e-05,
      "loss": 0.5092,
      "step": 13726
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6060500144958496,
      "learning_rate": 1.134916083183382e-05,
      "loss": 0.56,
      "step": 13727
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5969871282577515,
      "learning_rate": 1.1348054458213757e-05,
      "loss": 0.5763,
      "step": 13728
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3782731294631958,
      "learning_rate": 1.1346948067786989e-05,
      "loss": 0.4583,
      "step": 13729
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7643741369247437,
      "learning_rate": 1.1345841660567315e-05,
      "loss": 0.5496,
      "step": 13730
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7451919317245483,
      "learning_rate": 1.1344735236568529e-05,
      "loss": 0.5652,
      "step": 13731
    },
    {
      "epoch": 1.42,
      "grad_norm": 2.299365758895874,
      "learning_rate": 1.1343628795804424e-05,
      "loss": 0.66,
      "step": 13732
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7207480669021606,
      "learning_rate": 1.1342522338288793e-05,
      "loss": 0.5944,
      "step": 13733
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3961623907089233,
      "learning_rate": 1.1341415864035435e-05,
      "loss": 0.4421,
      "step": 13734
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2202990055084229,
      "learning_rate": 1.1340309373058138e-05,
      "loss": 0.5271,
      "step": 13735
    },
    {
      "epoch": 1.42,
      "grad_norm": 2.055591344833374,
      "learning_rate": 1.1339202865370704e-05,
      "loss": 0.5292,
      "step": 13736
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7137823104858398,
      "learning_rate": 1.1338096340986924e-05,
      "loss": 0.6875,
      "step": 13737
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7038720846176147,
      "learning_rate": 1.1336989799920595e-05,
      "loss": 0.6433,
      "step": 13738
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6447477340698242,
      "learning_rate": 1.1335883242185513e-05,
      "loss": 0.6187,
      "step": 13739
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.6048057079315186,
      "learning_rate": 1.1334776667795473e-05,
      "loss": 0.4307,
      "step": 13740
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.570007085800171,
      "learning_rate": 1.1333670076764271e-05,
      "loss": 0.6799,
      "step": 13741
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5413464307785034,
      "learning_rate": 1.1332563469105702e-05,
      "loss": 0.5068,
      "step": 13742
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.613966703414917,
      "learning_rate": 1.1331456844833566e-05,
      "loss": 0.5486,
      "step": 13743
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4083092212677002,
      "learning_rate": 1.1330350203961657e-05,
      "loss": 0.4566,
      "step": 13744
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3874053955078125,
      "learning_rate": 1.1329243546503771e-05,
      "loss": 0.4528,
      "step": 13745
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.596777081489563,
      "learning_rate": 1.132813687247371e-05,
      "loss": 0.5048,
      "step": 13746
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7389211654663086,
      "learning_rate": 1.1327030181885267e-05,
      "loss": 0.5261,
      "step": 13747
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.759156346321106,
      "learning_rate": 1.132592347475224e-05,
      "loss": 0.6012,
      "step": 13748
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4874372482299805,
      "learning_rate": 1.132481675108843e-05,
      "loss": 0.5718,
      "step": 13749
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4081571102142334,
      "learning_rate": 1.132371001090763e-05,
      "loss": 0.612,
      "step": 13750
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.8529400825500488,
      "learning_rate": 1.1322603254223638e-05,
      "loss": 0.6027,
      "step": 13751
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.31580650806427,
      "learning_rate": 1.1321496481050259e-05,
      "loss": 0.5599,
      "step": 13752
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.350121021270752,
      "learning_rate": 1.1320389691401283e-05,
      "loss": 0.5442,
      "step": 13753
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.708876132965088,
      "learning_rate": 1.131928288529052e-05,
      "loss": 0.6015,
      "step": 13754
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3850253820419312,
      "learning_rate": 1.1318176062731755e-05,
      "loss": 0.6119,
      "step": 13755
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.708587408065796,
      "learning_rate": 1.1317069223738797e-05,
      "loss": 0.464,
      "step": 13756
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2634443044662476,
      "learning_rate": 1.1315962368325444e-05,
      "loss": 0.4675,
      "step": 13757
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2726126909255981,
      "learning_rate": 1.1314855496505492e-05,
      "loss": 0.592,
      "step": 13758
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4051061868667603,
      "learning_rate": 1.1313748608292743e-05,
      "loss": 0.5683,
      "step": 13759
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4636179208755493,
      "learning_rate": 1.1312641703701e-05,
      "loss": 0.6164,
      "step": 13760
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4697046279907227,
      "learning_rate": 1.1311534782744058e-05,
      "loss": 0.4851,
      "step": 13761
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3234386444091797,
      "learning_rate": 1.131042784543572e-05,
      "loss": 0.489,
      "step": 13762
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4814101457595825,
      "learning_rate": 1.1309320891789785e-05,
      "loss": 0.5253,
      "step": 13763
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5018527507781982,
      "learning_rate": 1.1308213921820058e-05,
      "loss": 0.5246,
      "step": 13764
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4957154989242554,
      "learning_rate": 1.1307106935540335e-05,
      "loss": 0.6162,
      "step": 13765
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3370420932769775,
      "learning_rate": 1.1305999932964418e-05,
      "loss": 0.4653,
      "step": 13766
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7068678140640259,
      "learning_rate": 1.1304892914106114e-05,
      "loss": 0.6276,
      "step": 13767
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.9999858140945435,
      "learning_rate": 1.1303785878979214e-05,
      "loss": 0.5245,
      "step": 13768
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5279854536056519,
      "learning_rate": 1.1302678827597528e-05,
      "loss": 0.5113,
      "step": 13769
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.44048011302948,
      "learning_rate": 1.1301571759974861e-05,
      "loss": 0.5667,
      "step": 13770
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4120457172393799,
      "learning_rate": 1.1300464676125005e-05,
      "loss": 0.513,
      "step": 13771
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.7728989124298096,
      "learning_rate": 1.1299357576061769e-05,
      "loss": 0.4177,
      "step": 13772
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5110275745391846,
      "learning_rate": 1.1298250459798957e-05,
      "loss": 0.5142,
      "step": 13773
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5544185638427734,
      "learning_rate": 1.1297143327350367e-05,
      "loss": 0.4575,
      "step": 13774
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4965260028839111,
      "learning_rate": 1.1296036178729805e-05,
      "loss": 0.5918,
      "step": 13775
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5650486946105957,
      "learning_rate": 1.1294929013951073e-05,
      "loss": 0.5273,
      "step": 13776
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2091566324234009,
      "learning_rate": 1.1293821833027973e-05,
      "loss": 0.5377,
      "step": 13777
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5063347816467285,
      "learning_rate": 1.1292714635974313e-05,
      "loss": 0.5761,
      "step": 13778
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5807832479476929,
      "learning_rate": 1.1291607422803895e-05,
      "loss": 0.5237,
      "step": 13779
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6616652011871338,
      "learning_rate": 1.1290500193530523e-05,
      "loss": 0.6298,
      "step": 13780
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.662674903869629,
      "learning_rate": 1.1289392948167997e-05,
      "loss": 0.5485,
      "step": 13781
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4606205224990845,
      "learning_rate": 1.1288285686730128e-05,
      "loss": 0.5804,
      "step": 13782
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5388474464416504,
      "learning_rate": 1.1287178409230717e-05,
      "loss": 0.5583,
      "step": 13783
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7396559715270996,
      "learning_rate": 1.128607111568357e-05,
      "loss": 0.5696,
      "step": 13784
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5953021049499512,
      "learning_rate": 1.1284963806102489e-05,
      "loss": 0.5476,
      "step": 13785
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4530220031738281,
      "learning_rate": 1.1283856480501284e-05,
      "loss": 0.5882,
      "step": 13786
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3190877437591553,
      "learning_rate": 1.1282749138893761e-05,
      "loss": 0.5587,
      "step": 13787
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3244330883026123,
      "learning_rate": 1.1281641781293722e-05,
      "loss": 0.4452,
      "step": 13788
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4191418886184692,
      "learning_rate": 1.1280534407714972e-05,
      "loss": 0.5882,
      "step": 13789
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3502418994903564,
      "learning_rate": 1.1279427018171318e-05,
      "loss": 0.5808,
      "step": 13790
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5585758686065674,
      "learning_rate": 1.127831961267657e-05,
      "loss": 0.5459,
      "step": 13791
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8266327381134033,
      "learning_rate": 1.127721219124453e-05,
      "loss": 0.6388,
      "step": 13792
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5908700227737427,
      "learning_rate": 1.1276104753889008e-05,
      "loss": 0.6966,
      "step": 13793
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.458865761756897,
      "learning_rate": 1.1274997300623807e-05,
      "loss": 0.5824,
      "step": 13794
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7439903020858765,
      "learning_rate": 1.1273889831462734e-05,
      "loss": 0.5403,
      "step": 13795
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5486199855804443,
      "learning_rate": 1.1272782346419603e-05,
      "loss": 0.4661,
      "step": 13796
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5088804960250854,
      "learning_rate": 1.1271674845508217e-05,
      "loss": 0.5314,
      "step": 13797
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3429040908813477,
      "learning_rate": 1.1270567328742379e-05,
      "loss": 0.4287,
      "step": 13798
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2914174795150757,
      "learning_rate": 1.1269459796135904e-05,
      "loss": 0.4041,
      "step": 13799
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4272364377975464,
      "learning_rate": 1.1268352247702596e-05,
      "loss": 0.5038,
      "step": 13800
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5878338813781738,
      "learning_rate": 1.1267244683456267e-05,
      "loss": 0.5528,
      "step": 13801
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7221081256866455,
      "learning_rate": 1.1266137103410722e-05,
      "loss": 0.524,
      "step": 13802
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6293091773986816,
      "learning_rate": 1.1265029507579768e-05,
      "loss": 0.5708,
      "step": 13803
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2723677158355713,
      "learning_rate": 1.126392189597722e-05,
      "loss": 0.4905,
      "step": 13804
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.434753656387329,
      "learning_rate": 1.126281426861688e-05,
      "loss": 0.5294,
      "step": 13805
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.388523817062378,
      "learning_rate": 1.126170662551256e-05,
      "loss": 0.5579,
      "step": 13806
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.422754168510437,
      "learning_rate": 1.1260598966678074e-05,
      "loss": 0.5718,
      "step": 13807
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6944719552993774,
      "learning_rate": 1.1259491292127223e-05,
      "loss": 0.4801,
      "step": 13808
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5462974309921265,
      "learning_rate": 1.1258383601873824e-05,
      "loss": 0.5551,
      "step": 13809
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3729302883148193,
      "learning_rate": 1.1257275895931683e-05,
      "loss": 0.5375,
      "step": 13810
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.9342091083526611,
      "learning_rate": 1.1256168174314612e-05,
      "loss": 0.5409,
      "step": 13811
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5984957218170166,
      "learning_rate": 1.1255060437036423e-05,
      "loss": 0.534,
      "step": 13812
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.677979588508606,
      "learning_rate": 1.1253952684110923e-05,
      "loss": 0.4906,
      "step": 13813
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4146149158477783,
      "learning_rate": 1.1252844915551921e-05,
      "loss": 0.606,
      "step": 13814
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5041024684906006,
      "learning_rate": 1.1251737131373237e-05,
      "loss": 0.6598,
      "step": 13815
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7664315700531006,
      "learning_rate": 1.1250629331588672e-05,
      "loss": 0.533,
      "step": 13816
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4192447662353516,
      "learning_rate": 1.1249521516212045e-05,
      "loss": 0.5151,
      "step": 13817
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5168386697769165,
      "learning_rate": 1.124841368525716e-05,
      "loss": 0.5109,
      "step": 13818
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5944982767105103,
      "learning_rate": 1.1247305838737837e-05,
      "loss": 0.4857,
      "step": 13819
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4898920059204102,
      "learning_rate": 1.1246197976667883e-05,
      "loss": 0.6198,
      "step": 13820
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4151638746261597,
      "learning_rate": 1.1245090099061109e-05,
      "loss": 0.5041,
      "step": 13821
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.530335783958435,
      "learning_rate": 1.124398220593133e-05,
      "loss": 0.6301,
      "step": 13822
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6971694231033325,
      "learning_rate": 1.1242874297292362e-05,
      "loss": 0.5605,
      "step": 13823
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.738187313079834,
      "learning_rate": 1.1241766373158007e-05,
      "loss": 0.6307,
      "step": 13824
    },
    {
      "epoch": 1.43,
      "grad_norm": 2.188944101333618,
      "learning_rate": 1.124065843354209e-05,
      "loss": 0.616,
      "step": 13825
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7862639427185059,
      "learning_rate": 1.1239550478458417e-05,
      "loss": 0.5291,
      "step": 13826
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3482729196548462,
      "learning_rate": 1.1238442507920802e-05,
      "loss": 0.5196,
      "step": 13827
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6442458629608154,
      "learning_rate": 1.123733452194306e-05,
      "loss": 0.6751,
      "step": 13828
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6104768514633179,
      "learning_rate": 1.1236226520539004e-05,
      "loss": 0.6081,
      "step": 13829
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.493287205696106,
      "learning_rate": 1.1235118503722448e-05,
      "loss": 0.4901,
      "step": 13830
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4220632314682007,
      "learning_rate": 1.1234010471507206e-05,
      "loss": 0.4613,
      "step": 13831
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7105364799499512,
      "learning_rate": 1.123290242390709e-05,
      "loss": 0.4813,
      "step": 13832
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.339840054512024,
      "learning_rate": 1.1231794360935921e-05,
      "loss": 0.5474,
      "step": 13833
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8418558835983276,
      "learning_rate": 1.1230686282607509e-05,
      "loss": 0.4857,
      "step": 13834
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.162237524986267,
      "learning_rate": 1.1229578188935666e-05,
      "loss": 0.5327,
      "step": 13835
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4229308366775513,
      "learning_rate": 1.1228470079934214e-05,
      "loss": 0.4971,
      "step": 13836
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6756869554519653,
      "learning_rate": 1.122736195561696e-05,
      "loss": 0.6074,
      "step": 13837
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7291945219039917,
      "learning_rate": 1.1226253815997727e-05,
      "loss": 0.6251,
      "step": 13838
    },
    {
      "epoch": 1.43,
      "grad_norm": 2.2187347412109375,
      "learning_rate": 1.1225145661090327e-05,
      "loss": 0.5219,
      "step": 13839
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6696078777313232,
      "learning_rate": 1.1224037490908574e-05,
      "loss": 0.5357,
      "step": 13840
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.581323266029358,
      "learning_rate": 1.122292930546629e-05,
      "loss": 0.5621,
      "step": 13841
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.205946445465088,
      "learning_rate": 1.1221821104777283e-05,
      "loss": 0.6476,
      "step": 13842
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3644022941589355,
      "learning_rate": 1.1220712888855379e-05,
      "loss": 0.4832,
      "step": 13843
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5719257593154907,
      "learning_rate": 1.1219604657714385e-05,
      "loss": 0.4264,
      "step": 13844
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6908608675003052,
      "learning_rate": 1.1218496411368122e-05,
      "loss": 0.71,
      "step": 13845
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6688677072525024,
      "learning_rate": 1.1217388149830409e-05,
      "loss": 0.5994,
      "step": 13846
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4759148359298706,
      "learning_rate": 1.1216279873115059e-05,
      "loss": 0.5669,
      "step": 13847
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.370484471321106,
      "learning_rate": 1.1215171581235894e-05,
      "loss": 0.5622,
      "step": 13848
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.645739197731018,
      "learning_rate": 1.1214063274206725e-05,
      "loss": 0.5392,
      "step": 13849
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4813599586486816,
      "learning_rate": 1.1212954952041377e-05,
      "loss": 0.4874,
      "step": 13850
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2376329898834229,
      "learning_rate": 1.1211846614753663e-05,
      "loss": 0.519,
      "step": 13851
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6220593452453613,
      "learning_rate": 1.1210738262357401e-05,
      "loss": 0.4297,
      "step": 13852
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.402962327003479,
      "learning_rate": 1.1209629894866412e-05,
      "loss": 0.5168,
      "step": 13853
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4353939294815063,
      "learning_rate": 1.1208521512294514e-05,
      "loss": 0.557,
      "step": 13854
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6060370206832886,
      "learning_rate": 1.1207413114655523e-05,
      "loss": 0.5087,
      "step": 13855
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.774982213973999,
      "learning_rate": 1.1206304701963262e-05,
      "loss": 0.5623,
      "step": 13856
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6557236909866333,
      "learning_rate": 1.1205196274231546e-05,
      "loss": 0.5476,
      "step": 13857
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7106537818908691,
      "learning_rate": 1.1204087831474195e-05,
      "loss": 0.6215,
      "step": 13858
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.6030299663543701,
      "learning_rate": 1.120297937370503e-05,
      "loss": 0.6027,
      "step": 13859
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2788113355636597,
      "learning_rate": 1.1201870900937868e-05,
      "loss": 0.4797,
      "step": 13860
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3646409511566162,
      "learning_rate": 1.1200762413186533e-05,
      "loss": 0.4149,
      "step": 13861
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4631011486053467,
      "learning_rate": 1.1199653910464841e-05,
      "loss": 0.5603,
      "step": 13862
    },
    {
      "epoch": 1.43,
      "grad_norm": 2.0236282348632812,
      "learning_rate": 1.1198545392786612e-05,
      "loss": 0.5653,
      "step": 13863
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.7332837581634521,
      "learning_rate": 1.1197436860165672e-05,
      "loss": 0.4624,
      "step": 13864
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.8982839584350586,
      "learning_rate": 1.1196328312615832e-05,
      "loss": 0.5524,
      "step": 13865
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5857079029083252,
      "learning_rate": 1.1195219750150919e-05,
      "loss": 0.5577,
      "step": 13866
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3028594255447388,
      "learning_rate": 1.1194111172784757e-05,
      "loss": 0.4783,
      "step": 13867
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4587546586990356,
      "learning_rate": 1.1193002580531159e-05,
      "loss": 0.6433,
      "step": 13868
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4770492315292358,
      "learning_rate": 1.119189397340395e-05,
      "loss": 0.4605,
      "step": 13869
    },
    {
      "epoch": 1.43,
      "grad_norm": 2.274184226989746,
      "learning_rate": 1.1190785351416953e-05,
      "loss": 0.4801,
      "step": 13870
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.453084111213684,
      "learning_rate": 1.1189676714583986e-05,
      "loss": 0.4966,
      "step": 13871
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4290266036987305,
      "learning_rate": 1.1188568062918877e-05,
      "loss": 0.4516,
      "step": 13872
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.412133812904358,
      "learning_rate": 1.1187459396435438e-05,
      "loss": 0.4038,
      "step": 13873
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4440220594406128,
      "learning_rate": 1.11863507151475e-05,
      "loss": 0.5019,
      "step": 13874
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4960967302322388,
      "learning_rate": 1.1185242019068884e-05,
      "loss": 0.5985,
      "step": 13875
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4328861236572266,
      "learning_rate": 1.1184133308213408e-05,
      "loss": 0.4217,
      "step": 13876
    },
    {
      "epoch": 1.44,
      "grad_norm": 2.088010311126709,
      "learning_rate": 1.11830245825949e-05,
      "loss": 0.5962,
      "step": 13877
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7391761541366577,
      "learning_rate": 1.118191584222718e-05,
      "loss": 0.4466,
      "step": 13878
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.894446611404419,
      "learning_rate": 1.118080708712407e-05,
      "loss": 0.5694,
      "step": 13879
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5101780891418457,
      "learning_rate": 1.1179698317299396e-05,
      "loss": 0.5391,
      "step": 13880
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6687679290771484,
      "learning_rate": 1.117858953276698e-05,
      "loss": 0.5472,
      "step": 13881
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5617426633834839,
      "learning_rate": 1.1177480733540647e-05,
      "loss": 0.3843,
      "step": 13882
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.868323802947998,
      "learning_rate": 1.1176371919634219e-05,
      "loss": 0.5095,
      "step": 13883
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8993068933486938,
      "learning_rate": 1.1175263091061519e-05,
      "loss": 0.6355,
      "step": 13884
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8764169216156006,
      "learning_rate": 1.1174154247836375e-05,
      "loss": 0.4699,
      "step": 13885
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.3790199756622314,
      "learning_rate": 1.117304538997261e-05,
      "loss": 0.5535,
      "step": 13886
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7567687034606934,
      "learning_rate": 1.1171936517484046e-05,
      "loss": 0.4313,
      "step": 13887
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5413833856582642,
      "learning_rate": 1.117082763038451e-05,
      "loss": 0.6989,
      "step": 13888
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.27377188205719,
      "learning_rate": 1.1169718728687826e-05,
      "loss": 0.3906,
      "step": 13889
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.665527582168579,
      "learning_rate": 1.1168609812407822e-05,
      "loss": 0.667,
      "step": 13890
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.646955966949463,
      "learning_rate": 1.1167500881558318e-05,
      "loss": 0.6348,
      "step": 13891
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6790399551391602,
      "learning_rate": 1.1166391936153142e-05,
      "loss": 0.5472,
      "step": 13892
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5119950771331787,
      "learning_rate": 1.1165282976206124e-05,
      "loss": 0.4107,
      "step": 13893
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.844207763671875,
      "learning_rate": 1.116417400173108e-05,
      "loss": 0.4459,
      "step": 13894
    },
    {
      "epoch": 1.44,
      "grad_norm": 2.151643753051758,
      "learning_rate": 1.1163065012741848e-05,
      "loss": 0.5549,
      "step": 13895
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6986171007156372,
      "learning_rate": 1.1161956009252243e-05,
      "loss": 0.4059,
      "step": 13896
    },
    {
      "epoch": 1.44,
      "grad_norm": 2.060576915740967,
      "learning_rate": 1.1160846991276098e-05,
      "loss": 0.4595,
      "step": 13897
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6756011247634888,
      "learning_rate": 1.115973795882724e-05,
      "loss": 0.477,
      "step": 13898
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.3493355512619019,
      "learning_rate": 1.115862891191949e-05,
      "loss": 0.5744,
      "step": 13899
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.574689507484436,
      "learning_rate": 1.1157519850566679e-05,
      "loss": 0.6572,
      "step": 13900
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4254076480865479,
      "learning_rate": 1.1156410774782636e-05,
      "loss": 0.6084,
      "step": 13901
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6902215480804443,
      "learning_rate": 1.1155301684581184e-05,
      "loss": 0.5325,
      "step": 13902
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.34285306930542,
      "learning_rate": 1.1154192579976153e-05,
      "loss": 0.4909,
      "step": 13903
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.524972915649414,
      "learning_rate": 1.115308346098137e-05,
      "loss": 0.5046,
      "step": 13904
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5651676654815674,
      "learning_rate": 1.1151974327610661e-05,
      "loss": 0.6125,
      "step": 13905
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6612635850906372,
      "learning_rate": 1.115086517987786e-05,
      "loss": 0.6755,
      "step": 13906
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5569350719451904,
      "learning_rate": 1.1149756017796785e-05,
      "loss": 0.5516,
      "step": 13907
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7066621780395508,
      "learning_rate": 1.1148646841381274e-05,
      "loss": 0.6375,
      "step": 13908
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5194379091262817,
      "learning_rate": 1.114753765064515e-05,
      "loss": 0.526,
      "step": 13909
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2545725107192993,
      "learning_rate": 1.1146428445602245e-05,
      "loss": 0.4861,
      "step": 13910
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7921714782714844,
      "learning_rate": 1.1145319226266385e-05,
      "loss": 0.5772,
      "step": 13911
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5147634744644165,
      "learning_rate": 1.11442099926514e-05,
      "loss": 0.598,
      "step": 13912
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.3948917388916016,
      "learning_rate": 1.114310074477112e-05,
      "loss": 0.3972,
      "step": 13913
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5477395057678223,
      "learning_rate": 1.1141991482639373e-05,
      "loss": 0.5611,
      "step": 13914
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.570523738861084,
      "learning_rate": 1.1140882206269991e-05,
      "loss": 0.5085,
      "step": 13915
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6776654720306396,
      "learning_rate": 1.1139772915676805e-05,
      "loss": 0.5641,
      "step": 13916
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.563647985458374,
      "learning_rate": 1.1138663610873637e-05,
      "loss": 0.5213,
      "step": 13917
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1723453998565674,
      "learning_rate": 1.1137554291874324e-05,
      "loss": 0.4298,
      "step": 13918
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.313125729560852,
      "learning_rate": 1.1136444958692699e-05,
      "loss": 0.6144,
      "step": 13919
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.217953085899353,
      "learning_rate": 1.1135335611342581e-05,
      "loss": 0.493,
      "step": 13920
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4315084218978882,
      "learning_rate": 1.1134226249837815e-05,
      "loss": 0.5898,
      "step": 13921
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.3443065881729126,
      "learning_rate": 1.1133116874192217e-05,
      "loss": 0.6486,
      "step": 13922
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5826724767684937,
      "learning_rate": 1.1132007484419632e-05,
      "loss": 0.5263,
      "step": 13923
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2326606512069702,
      "learning_rate": 1.1130898080533883e-05,
      "loss": 0.5169,
      "step": 13924
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4222948551177979,
      "learning_rate": 1.1129788662548802e-05,
      "loss": 0.4776,
      "step": 13925
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5973098278045654,
      "learning_rate": 1.112867923047822e-05,
      "loss": 0.4687,
      "step": 13926
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.611505150794983,
      "learning_rate": 1.1127569784335974e-05,
      "loss": 0.585,
      "step": 13927
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6106330156326294,
      "learning_rate": 1.112646032413589e-05,
      "loss": 0.5229,
      "step": 13928
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9049898386001587,
      "learning_rate": 1.1125350849891806e-05,
      "loss": 0.5411,
      "step": 13929
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6930229663848877,
      "learning_rate": 1.1124241361617547e-05,
      "loss": 0.6122,
      "step": 13930
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9996365308761597,
      "learning_rate": 1.1123131859326948e-05,
      "loss": 0.4799,
      "step": 13931
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.743190884590149,
      "learning_rate": 1.1122022343033847e-05,
      "loss": 0.5328,
      "step": 13932
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.345185399055481,
      "learning_rate": 1.1120912812752067e-05,
      "loss": 0.5571,
      "step": 13933
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7203540802001953,
      "learning_rate": 1.1119803268495451e-05,
      "loss": 0.6125,
      "step": 13934
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.398485541343689,
      "learning_rate": 1.1118693710277823e-05,
      "loss": 0.4482,
      "step": 13935
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5518851280212402,
      "learning_rate": 1.1117584138113025e-05,
      "loss": 0.582,
      "step": 13936
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5552393198013306,
      "learning_rate": 1.1116474552014883e-05,
      "loss": 0.4266,
      "step": 13937
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7315483093261719,
      "learning_rate": 1.1115364951997235e-05,
      "loss": 0.6336,
      "step": 13938
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.7645257711410522,
      "learning_rate": 1.1114255338073914e-05,
      "loss": 0.5664,
      "step": 13939
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.8948392868041992,
      "learning_rate": 1.1113145710258753e-05,
      "loss": 0.6142,
      "step": 13940
    },
    {
      "epoch": 1.44,
      "grad_norm": 2.015204429626465,
      "learning_rate": 1.1112036068565586e-05,
      "loss": 0.5943,
      "step": 13941
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9460232257843018,
      "learning_rate": 1.1110926413008249e-05,
      "loss": 0.536,
      "step": 13942
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.837126612663269,
      "learning_rate": 1.1109816743600574e-05,
      "loss": 0.6785,
      "step": 13943
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.402710199356079,
      "learning_rate": 1.1108707060356397e-05,
      "loss": 0.4301,
      "step": 13944
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4668480157852173,
      "learning_rate": 1.1107597363289555e-05,
      "loss": 0.4915,
      "step": 13945
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5928919315338135,
      "learning_rate": 1.1106487652413877e-05,
      "loss": 0.5769,
      "step": 13946
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.3143028020858765,
      "learning_rate": 1.1105377927743208e-05,
      "loss": 0.4416,
      "step": 13947
    },
    {
      "epoch": 1.44,
      "grad_norm": 2.4415786266326904,
      "learning_rate": 1.1104268189291375e-05,
      "loss": 0.7221,
      "step": 13948
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4678024053573608,
      "learning_rate": 1.1103158437072212e-05,
      "loss": 0.5108,
      "step": 13949
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.290365219116211,
      "learning_rate": 1.1102048671099563e-05,
      "loss": 0.5842,
      "step": 13950
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.543824315071106,
      "learning_rate": 1.1100938891387258e-05,
      "loss": 0.4847,
      "step": 13951
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.46340811252594,
      "learning_rate": 1.1099829097949138e-05,
      "loss": 0.5447,
      "step": 13952
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.578884482383728,
      "learning_rate": 1.1098719290799031e-05,
      "loss": 0.5372,
      "step": 13953
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.3939796686172485,
      "learning_rate": 1.109760946995078e-05,
      "loss": 0.5325,
      "step": 13954
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6677507162094116,
      "learning_rate": 1.109649963541822e-05,
      "loss": 0.5283,
      "step": 13955
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4405455589294434,
      "learning_rate": 1.1095389787215188e-05,
      "loss": 0.4841,
      "step": 13956
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.9732286930084229,
      "learning_rate": 1.109427992535552e-05,
      "loss": 0.6695,
      "step": 13957
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.330275058746338,
      "learning_rate": 1.1093170049853055e-05,
      "loss": 0.4514,
      "step": 13958
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5988142490386963,
      "learning_rate": 1.1092060160721624e-05,
      "loss": 0.6071,
      "step": 13959
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.447356939315796,
      "learning_rate": 1.1090950257975074e-05,
      "loss": 0.4502,
      "step": 13960
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.348067283630371,
      "learning_rate": 1.1089840341627235e-05,
      "loss": 0.6042,
      "step": 13961
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.3582613468170166,
      "learning_rate": 1.1088730411691948e-05,
      "loss": 0.5576,
      "step": 13962
    },
    {
      "epoch": 1.44,
      "grad_norm": 2.048539876937866,
      "learning_rate": 1.1087620468183052e-05,
      "loss": 0.5656,
      "step": 13963
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6300549507141113,
      "learning_rate": 1.1086510511114382e-05,
      "loss": 0.5339,
      "step": 13964
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.699532389640808,
      "learning_rate": 1.1085400540499778e-05,
      "loss": 0.5264,
      "step": 13965
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.5682915449142456,
      "learning_rate": 1.1084290556353077e-05,
      "loss": 0.5058,
      "step": 13966
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2873891592025757,
      "learning_rate": 1.1083180558688119e-05,
      "loss": 0.5407,
      "step": 13967
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.378633975982666,
      "learning_rate": 1.1082070547518744e-05,
      "loss": 0.4981,
      "step": 13968
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6269117593765259,
      "learning_rate": 1.1080960522858786e-05,
      "loss": 0.6516,
      "step": 13969
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6828601360321045,
      "learning_rate": 1.1079850484722087e-05,
      "loss": 0.5875,
      "step": 13970
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.6612979173660278,
      "learning_rate": 1.1078740433122492e-05,
      "loss": 0.5777,
      "step": 13971
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4419037103652954,
      "learning_rate": 1.1077630368073833e-05,
      "loss": 0.5423,
      "step": 13972
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5956614017486572,
      "learning_rate": 1.107652028958995e-05,
      "loss": 0.6257,
      "step": 13973
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3710083961486816,
      "learning_rate": 1.1075410197684688e-05,
      "loss": 0.4578,
      "step": 13974
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.359061360359192,
      "learning_rate": 1.1074300092371879e-05,
      "loss": 0.5501,
      "step": 13975
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3647760152816772,
      "learning_rate": 1.1073189973665373e-05,
      "loss": 0.5205,
      "step": 13976
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4549959897994995,
      "learning_rate": 1.1072079841579e-05,
      "loss": 0.5771,
      "step": 13977
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.669639229774475,
      "learning_rate": 1.1070969696126606e-05,
      "loss": 0.606,
      "step": 13978
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3262202739715576,
      "learning_rate": 1.1069859537322032e-05,
      "loss": 0.5475,
      "step": 13979
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6291708946228027,
      "learning_rate": 1.1068749365179116e-05,
      "loss": 0.5482,
      "step": 13980
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6863099336624146,
      "learning_rate": 1.10676391797117e-05,
      "loss": 0.5246,
      "step": 13981
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6554001569747925,
      "learning_rate": 1.1066528980933628e-05,
      "loss": 0.6334,
      "step": 13982
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.8628448247909546,
      "learning_rate": 1.1065418768858738e-05,
      "loss": 0.5589,
      "step": 13983
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3646302223205566,
      "learning_rate": 1.1064308543500875e-05,
      "loss": 0.4586,
      "step": 13984
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5144468545913696,
      "learning_rate": 1.1063198304873872e-05,
      "loss": 0.4914,
      "step": 13985
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6575247049331665,
      "learning_rate": 1.106208805299158e-05,
      "loss": 0.5498,
      "step": 13986
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4843904972076416,
      "learning_rate": 1.106097778786784e-05,
      "loss": 0.5469,
      "step": 13987
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.9864048957824707,
      "learning_rate": 1.1059867509516486e-05,
      "loss": 0.5788,
      "step": 13988
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5361915826797485,
      "learning_rate": 1.1058757217951373e-05,
      "loss": 0.5936,
      "step": 13989
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3901348114013672,
      "learning_rate": 1.105764691318633e-05,
      "loss": 0.5072,
      "step": 13990
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3706035614013672,
      "learning_rate": 1.1056536595235207e-05,
      "loss": 0.5141,
      "step": 13991
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3697859048843384,
      "learning_rate": 1.1055426264111851e-05,
      "loss": 0.5865,
      "step": 13992
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.428130030632019,
      "learning_rate": 1.1054315919830094e-05,
      "loss": 0.5618,
      "step": 13993
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.946335792541504,
      "learning_rate": 1.1053205562403786e-05,
      "loss": 0.5864,
      "step": 13994
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5593479871749878,
      "learning_rate": 1.105209519184677e-05,
      "loss": 0.5727,
      "step": 13995
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6673275232315063,
      "learning_rate": 1.1050984808172886e-05,
      "loss": 0.5269,
      "step": 13996
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6589572429656982,
      "learning_rate": 1.1049874411395984e-05,
      "loss": 0.5718,
      "step": 13997
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3890290260314941,
      "learning_rate": 1.10487640015299e-05,
      "loss": 0.6546,
      "step": 13998
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6663174629211426,
      "learning_rate": 1.1047653578588481e-05,
      "loss": 0.5768,
      "step": 13999
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3383485078811646,
      "learning_rate": 1.1046543142585577e-05,
      "loss": 0.3848,
      "step": 14000
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5073553323745728,
      "learning_rate": 1.104543269353502e-05,
      "loss": 0.6242,
      "step": 14001
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4716485738754272,
      "learning_rate": 1.1044322231450667e-05,
      "loss": 0.505,
      "step": 14002
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6263824701309204,
      "learning_rate": 1.1043211756346355e-05,
      "loss": 0.6321,
      "step": 14003
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4946024417877197,
      "learning_rate": 1.1042101268235927e-05,
      "loss": 0.4437,
      "step": 14004
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5983747243881226,
      "learning_rate": 1.1040990767133236e-05,
      "loss": 0.4824,
      "step": 14005
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.780511498451233,
      "learning_rate": 1.103988025305212e-05,
      "loss": 0.4115,
      "step": 14006
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6029785871505737,
      "learning_rate": 1.1038769726006427e-05,
      "loss": 0.5951,
      "step": 14007
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.255586862564087,
      "learning_rate": 1.1037659186010003e-05,
      "loss": 0.3893,
      "step": 14008
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3322656154632568,
      "learning_rate": 1.1036548633076692e-05,
      "loss": 0.5248,
      "step": 14009
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.581548810005188,
      "learning_rate": 1.1035438067220342e-05,
      "loss": 0.5693,
      "step": 14010
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7696353197097778,
      "learning_rate": 1.1034327488454792e-05,
      "loss": 0.6473,
      "step": 14011
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4264410734176636,
      "learning_rate": 1.1033216896793895e-05,
      "loss": 0.484,
      "step": 14012
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3321707248687744,
      "learning_rate": 1.1032106292251497e-05,
      "loss": 0.4753,
      "step": 14013
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3965989351272583,
      "learning_rate": 1.103099567484144e-05,
      "loss": 0.4261,
      "step": 14014
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7995023727416992,
      "learning_rate": 1.1029885044577575e-05,
      "loss": 0.6202,
      "step": 14015
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5538581609725952,
      "learning_rate": 1.1028774401473745e-05,
      "loss": 0.5905,
      "step": 14016
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.462377667427063,
      "learning_rate": 1.1027663745543799e-05,
      "loss": 0.4954,
      "step": 14017
    },
    {
      "epoch": 1.45,
      "grad_norm": 2.3678665161132812,
      "learning_rate": 1.1026553076801585e-05,
      "loss": 0.7117,
      "step": 14018
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.526375651359558,
      "learning_rate": 1.1025442395260946e-05,
      "loss": 0.5246,
      "step": 14019
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7699291706085205,
      "learning_rate": 1.1024331700935732e-05,
      "loss": 0.4557,
      "step": 14020
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5882940292358398,
      "learning_rate": 1.1023220993839794e-05,
      "loss": 0.6452,
      "step": 14021
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.367897391319275,
      "learning_rate": 1.102211027398697e-05,
      "loss": 0.4925,
      "step": 14022
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1811400651931763,
      "learning_rate": 1.1020999541391119e-05,
      "loss": 0.4424,
      "step": 14023
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3605992794036865,
      "learning_rate": 1.101988879606608e-05,
      "loss": 0.4839,
      "step": 14024
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6721068620681763,
      "learning_rate": 1.1018778038025705e-05,
      "loss": 0.5006,
      "step": 14025
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.538289189338684,
      "learning_rate": 1.1017667267283843e-05,
      "loss": 0.6803,
      "step": 14026
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3938716650009155,
      "learning_rate": 1.1016556483854339e-05,
      "loss": 0.4611,
      "step": 14027
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5878679752349854,
      "learning_rate": 1.1015445687751047e-05,
      "loss": 0.6041,
      "step": 14028
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.30923330783844,
      "learning_rate": 1.1014334878987811e-05,
      "loss": 0.5095,
      "step": 14029
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3769760131835938,
      "learning_rate": 1.101322405757848e-05,
      "loss": 0.5052,
      "step": 14030
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.526550531387329,
      "learning_rate": 1.1012113223536907e-05,
      "loss": 0.5887,
      "step": 14031
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.374505639076233,
      "learning_rate": 1.1011002376876936e-05,
      "loss": 0.4914,
      "step": 14032
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.650429368019104,
      "learning_rate": 1.1009891517612418e-05,
      "loss": 0.6395,
      "step": 14033
    },
    {
      "epoch": 1.45,
      "grad_norm": 2.3466508388519287,
      "learning_rate": 1.1008780645757206e-05,
      "loss": 0.547,
      "step": 14034
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3851745128631592,
      "learning_rate": 1.1007669761325148e-05,
      "loss": 0.4093,
      "step": 14035
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.7485194206237793,
      "learning_rate": 1.1006558864330093e-05,
      "loss": 0.6432,
      "step": 14036
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.429714322090149,
      "learning_rate": 1.1005447954785885e-05,
      "loss": 0.5288,
      "step": 14037
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.707736611366272,
      "learning_rate": 1.1004337032706385e-05,
      "loss": 0.5682,
      "step": 14038
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6991982460021973,
      "learning_rate": 1.1003226098105439e-05,
      "loss": 0.6223,
      "step": 14039
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4758832454681396,
      "learning_rate": 1.1002115150996894e-05,
      "loss": 0.557,
      "step": 14040
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.317834496498108,
      "learning_rate": 1.1001004191394607e-05,
      "loss": 0.5658,
      "step": 14041
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.9220620393753052,
      "learning_rate": 1.0999893219312421e-05,
      "loss": 0.5772,
      "step": 14042
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4833134412765503,
      "learning_rate": 1.0998782234764192e-05,
      "loss": 0.5346,
      "step": 14043
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5413414239883423,
      "learning_rate": 1.0997671237763772e-05,
      "loss": 0.522,
      "step": 14044
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4001801013946533,
      "learning_rate": 1.0996560228325007e-05,
      "loss": 0.494,
      "step": 14045
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2910090684890747,
      "learning_rate": 1.0995449206461752e-05,
      "loss": 0.6413,
      "step": 14046
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.468070387840271,
      "learning_rate": 1.0994338172187859e-05,
      "loss": 0.5673,
      "step": 14047
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5985915660858154,
      "learning_rate": 1.0993227125517178e-05,
      "loss": 0.5765,
      "step": 14048
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3798737525939941,
      "learning_rate": 1.0992116066463565e-05,
      "loss": 0.4054,
      "step": 14049
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.9938161373138428,
      "learning_rate": 1.0991004995040867e-05,
      "loss": 0.6184,
      "step": 14050
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3453665971755981,
      "learning_rate": 1.0989893911262934e-05,
      "loss": 0.5009,
      "step": 14051
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.46235191822052,
      "learning_rate": 1.0988782815143629e-05,
      "loss": 0.5539,
      "step": 14052
    },
    {
      "epoch": 1.45,
      "grad_norm": 16.1573486328125,
      "learning_rate": 1.0987671706696792e-05,
      "loss": 1.1162,
      "step": 14053
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.793624758720398,
      "learning_rate": 1.0986560585936283e-05,
      "loss": 0.5304,
      "step": 14054
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.578055739402771,
      "learning_rate": 1.098544945287595e-05,
      "loss": 0.6054,
      "step": 14055
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.443251371383667,
      "learning_rate": 1.0984338307529652e-05,
      "loss": 0.4953,
      "step": 14056
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.620962381362915,
      "learning_rate": 1.098322714991124e-05,
      "loss": 0.5459,
      "step": 14057
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.6040370464324951,
      "learning_rate": 1.0982115980034564e-05,
      "loss": 0.5465,
      "step": 14058
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5593293905258179,
      "learning_rate": 1.098100479791348e-05,
      "loss": 0.5154,
      "step": 14059
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.595638632774353,
      "learning_rate": 1.0979893603561841e-05,
      "loss": 0.5136,
      "step": 14060
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3941359519958496,
      "learning_rate": 1.0978782396993498e-05,
      "loss": 0.5538,
      "step": 14061
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.661199927330017,
      "learning_rate": 1.0977671178222312e-05,
      "loss": 0.497,
      "step": 14062
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3204501867294312,
      "learning_rate": 1.0976559947262131e-05,
      "loss": 0.5049,
      "step": 14063
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3790479898452759,
      "learning_rate": 1.0975448704126808e-05,
      "loss": 0.6461,
      "step": 14064
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3607850074768066,
      "learning_rate": 1.0974337448830203e-05,
      "loss": 0.5503,
      "step": 14065
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4762085676193237,
      "learning_rate": 1.0973226181386168e-05,
      "loss": 0.6146,
      "step": 14066
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4451133012771606,
      "learning_rate": 1.0972114901808555e-05,
      "loss": 0.4782,
      "step": 14067
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3185912370681763,
      "learning_rate": 1.0971003610111219e-05,
      "loss": 0.5023,
      "step": 14068
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1598109006881714,
      "learning_rate": 1.0969892306308018e-05,
      "loss": 0.4898,
      "step": 14069
    },
    {
      "epoch": 1.46,
      "grad_norm": 2.0945353507995605,
      "learning_rate": 1.0968780990412809e-05,
      "loss": 0.6705,
      "step": 14070
    },
    {
      "epoch": 1.46,
      "grad_norm": 2.0269272327423096,
      "learning_rate": 1.0967669662439437e-05,
      "loss": 0.6477,
      "step": 14071
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3519693613052368,
      "learning_rate": 1.0966558322401768e-05,
      "loss": 0.5404,
      "step": 14072
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.775635838508606,
      "learning_rate": 1.0965446970313655e-05,
      "loss": 0.5608,
      "step": 14073
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5002479553222656,
      "learning_rate": 1.096433560618895e-05,
      "loss": 0.4644,
      "step": 14074
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5798927545547485,
      "learning_rate": 1.0963224230041513e-05,
      "loss": 0.4772,
      "step": 14075
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.86756432056427,
      "learning_rate": 1.0962112841885195e-05,
      "loss": 0.6444,
      "step": 14076
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8101133108139038,
      "learning_rate": 1.0961001441733857e-05,
      "loss": 0.6158,
      "step": 14077
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5467408895492554,
      "learning_rate": 1.0959890029601354e-05,
      "loss": 0.4797,
      "step": 14078
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4067203998565674,
      "learning_rate": 1.095877860550154e-05,
      "loss": 0.5247,
      "step": 14079
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2741880416870117,
      "learning_rate": 1.0957667169448277e-05,
      "loss": 0.5653,
      "step": 14080
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5426452159881592,
      "learning_rate": 1.0956555721455413e-05,
      "loss": 0.5362,
      "step": 14081
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5507491827011108,
      "learning_rate": 1.095544426153681e-05,
      "loss": 0.6512,
      "step": 14082
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5300871133804321,
      "learning_rate": 1.0954332789706329e-05,
      "loss": 0.7138,
      "step": 14083
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6094805002212524,
      "learning_rate": 1.095322130597782e-05,
      "loss": 0.5929,
      "step": 14084
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.455283522605896,
      "learning_rate": 1.0952109810365141e-05,
      "loss": 0.6459,
      "step": 14085
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6182303428649902,
      "learning_rate": 1.0950998302882157e-05,
      "loss": 0.5152,
      "step": 14086
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6414204835891724,
      "learning_rate": 1.0949886783542715e-05,
      "loss": 0.5032,
      "step": 14087
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2992628812789917,
      "learning_rate": 1.0948775252360684e-05,
      "loss": 0.5769,
      "step": 14088
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3857088088989258,
      "learning_rate": 1.0947663709349912e-05,
      "loss": 0.5425,
      "step": 14089
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.9308851957321167,
      "learning_rate": 1.094655215452426e-05,
      "loss": 0.595,
      "step": 14090
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.53200101852417,
      "learning_rate": 1.094544058789759e-05,
      "loss": 0.5914,
      "step": 14091
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.455861210823059,
      "learning_rate": 1.0944329009483754e-05,
      "loss": 0.5173,
      "step": 14092
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.673690676689148,
      "learning_rate": 1.0943217419296617e-05,
      "loss": 0.6249,
      "step": 14093
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.374544620513916,
      "learning_rate": 1.0942105817350031e-05,
      "loss": 0.4883,
      "step": 14094
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3705358505249023,
      "learning_rate": 1.0940994203657857e-05,
      "loss": 0.4215,
      "step": 14095
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6535792350769043,
      "learning_rate": 1.0939882578233961e-05,
      "loss": 0.5898,
      "step": 14096
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3971790075302124,
      "learning_rate": 1.093877094109219e-05,
      "loss": 0.5486,
      "step": 14097
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4108507633209229,
      "learning_rate": 1.0937659292246414e-05,
      "loss": 0.468,
      "step": 14098
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.511530876159668,
      "learning_rate": 1.0936547631710489e-05,
      "loss": 0.56,
      "step": 14099
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0863782167434692,
      "learning_rate": 1.0935435959498268e-05,
      "loss": 0.5527,
      "step": 14100
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4222135543823242,
      "learning_rate": 1.093432427562362e-05,
      "loss": 0.5976,
      "step": 14101
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6980162858963013,
      "learning_rate": 1.0933212580100396e-05,
      "loss": 0.5192,
      "step": 14102
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.285682201385498,
      "learning_rate": 1.0932100872942464e-05,
      "loss": 0.5596,
      "step": 14103
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4013687372207642,
      "learning_rate": 1.093098915416368e-05,
      "loss": 0.4995,
      "step": 14104
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4403702020645142,
      "learning_rate": 1.0929877423777904e-05,
      "loss": 0.6358,
      "step": 14105
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.453505516052246,
      "learning_rate": 1.0928765681798998e-05,
      "loss": 0.5294,
      "step": 14106
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.582499623298645,
      "learning_rate": 1.092765392824082e-05,
      "loss": 0.6469,
      "step": 14107
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2919220924377441,
      "learning_rate": 1.0926542163117234e-05,
      "loss": 0.5314,
      "step": 14108
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5673308372497559,
      "learning_rate": 1.09254303864421e-05,
      "loss": 0.6571,
      "step": 14109
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.8761608600616455,
      "learning_rate": 1.0924318598229276e-05,
      "loss": 0.661,
      "step": 14110
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.318750262260437,
      "learning_rate": 1.0923206798492625e-05,
      "loss": 0.6344,
      "step": 14111
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.923781394958496,
      "learning_rate": 1.092209498724601e-05,
      "loss": 0.6625,
      "step": 14112
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5908581018447876,
      "learning_rate": 1.092098316450329e-05,
      "loss": 0.6139,
      "step": 14113
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4569816589355469,
      "learning_rate": 1.0919871330278327e-05,
      "loss": 0.5906,
      "step": 14114
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4754289388656616,
      "learning_rate": 1.0918759484584982e-05,
      "loss": 0.5437,
      "step": 14115
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0905650854110718,
      "learning_rate": 1.0917647627437119e-05,
      "loss": 0.4605,
      "step": 14116
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6292363405227661,
      "learning_rate": 1.0916535758848598e-05,
      "loss": 0.6823,
      "step": 14117
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.665361762046814,
      "learning_rate": 1.091542387883328e-05,
      "loss": 0.5718,
      "step": 14118
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.474243402481079,
      "learning_rate": 1.0914311987405033e-05,
      "loss": 0.6192,
      "step": 14119
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5136134624481201,
      "learning_rate": 1.0913200084577711e-05,
      "loss": 0.5272,
      "step": 14120
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4800864458084106,
      "learning_rate": 1.0912088170365182e-05,
      "loss": 0.5269,
      "step": 14121
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4374562501907349,
      "learning_rate": 1.091097624478131e-05,
      "loss": 0.5246,
      "step": 14122
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3195544481277466,
      "learning_rate": 1.0909864307839952e-05,
      "loss": 0.5607,
      "step": 14123
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6195366382598877,
      "learning_rate": 1.0908752359554974e-05,
      "loss": 0.5783,
      "step": 14124
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4111804962158203,
      "learning_rate": 1.0907640399940245e-05,
      "loss": 0.5556,
      "step": 14125
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.277597427368164,
      "learning_rate": 1.0906528429009616e-05,
      "loss": 0.5745,
      "step": 14126
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5575439929962158,
      "learning_rate": 1.0905416446776962e-05,
      "loss": 0.4681,
      "step": 14127
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5244871377944946,
      "learning_rate": 1.0904304453256137e-05,
      "loss": 0.5678,
      "step": 14128
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4291627407073975,
      "learning_rate": 1.0903192448461008e-05,
      "loss": 0.5886,
      "step": 14129
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4712297916412354,
      "learning_rate": 1.0902080432405445e-05,
      "loss": 0.6262,
      "step": 14130
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4382792711257935,
      "learning_rate": 1.0900968405103302e-05,
      "loss": 0.4641,
      "step": 14131
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.641193151473999,
      "learning_rate": 1.0899856366568451e-05,
      "loss": 0.5545,
      "step": 14132
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4429583549499512,
      "learning_rate": 1.089874431681475e-05,
      "loss": 0.5367,
      "step": 14133
    },
    {
      "epoch": 1.46,
      "grad_norm": 2.389289140701294,
      "learning_rate": 1.0897632255856068e-05,
      "loss": 0.6786,
      "step": 14134
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5618730783462524,
      "learning_rate": 1.089652018370627e-05,
      "loss": 0.5763,
      "step": 14135
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3142907619476318,
      "learning_rate": 1.0895408100379214e-05,
      "loss": 0.4527,
      "step": 14136
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6236776113510132,
      "learning_rate": 1.0894296005888769e-05,
      "loss": 0.5472,
      "step": 14137
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5300753116607666,
      "learning_rate": 1.0893183900248805e-05,
      "loss": 0.4902,
      "step": 14138
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5342442989349365,
      "learning_rate": 1.0892071783473176e-05,
      "loss": 0.6207,
      "step": 14139
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4175570011138916,
      "learning_rate": 1.0890959655575758e-05,
      "loss": 0.5994,
      "step": 14140
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5922619104385376,
      "learning_rate": 1.0889847516570411e-05,
      "loss": 0.5943,
      "step": 14141
    },
    {
      "epoch": 1.46,
      "grad_norm": 2.0719656944274902,
      "learning_rate": 1.0888735366470999e-05,
      "loss": 0.5687,
      "step": 14142
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5299901962280273,
      "learning_rate": 1.0887623205291391e-05,
      "loss": 0.506,
      "step": 14143
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6093443632125854,
      "learning_rate": 1.088651103304545e-05,
      "loss": 0.5929,
      "step": 14144
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.6832401752471924,
      "learning_rate": 1.0885398849747046e-05,
      "loss": 0.5426,
      "step": 14145
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5696958303451538,
      "learning_rate": 1.0884286655410039e-05,
      "loss": 0.6154,
      "step": 14146
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5563719272613525,
      "learning_rate": 1.08831744500483e-05,
      "loss": 0.6487,
      "step": 14147
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2893409729003906,
      "learning_rate": 1.0882062233675694e-05,
      "loss": 0.5503,
      "step": 14148
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4268962144851685,
      "learning_rate": 1.0880950006306089e-05,
      "loss": 0.5679,
      "step": 14149
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1611629724502563,
      "learning_rate": 1.0879837767953345e-05,
      "loss": 0.4844,
      "step": 14150
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2013933658599854,
      "learning_rate": 1.0878725518631336e-05,
      "loss": 0.4621,
      "step": 14151
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5720951557159424,
      "learning_rate": 1.0877613258353925e-05,
      "loss": 0.6567,
      "step": 14152
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5938265323638916,
      "learning_rate": 1.0876500987134978e-05,
      "loss": 0.5741,
      "step": 14153
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5773379802703857,
      "learning_rate": 1.087538870498837e-05,
      "loss": 0.5946,
      "step": 14154
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5028581619262695,
      "learning_rate": 1.0874276411927958e-05,
      "loss": 0.569,
      "step": 14155
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.493665099143982,
      "learning_rate": 1.0873164107967618e-05,
      "loss": 0.4725,
      "step": 14156
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.5524356365203857,
      "learning_rate": 1.0872051793121207e-05,
      "loss": 0.5753,
      "step": 14157
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.860514760017395,
      "learning_rate": 1.0870939467402604e-05,
      "loss": 0.533,
      "step": 14158
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.361001968383789,
      "learning_rate": 1.086982713082567e-05,
      "loss": 0.4707,
      "step": 14159
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4806102514266968,
      "learning_rate": 1.0868714783404274e-05,
      "loss": 0.5597,
      "step": 14160
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.9072474241256714,
      "learning_rate": 1.0867602425152285e-05,
      "loss": 0.4857,
      "step": 14161
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2872499227523804,
      "learning_rate": 1.0866490056083574e-05,
      "loss": 0.4332,
      "step": 14162
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.4286003112792969,
      "learning_rate": 1.0865377676212e-05,
      "loss": 0.5681,
      "step": 14163
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.30471670627594,
      "learning_rate": 1.0864265285551445e-05,
      "loss": 0.5185,
      "step": 14164
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.68453049659729,
      "learning_rate": 1.0863152884115767e-05,
      "loss": 0.552,
      "step": 14165
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4424220323562622,
      "learning_rate": 1.0862040471918836e-05,
      "loss": 0.4873,
      "step": 14166
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4058934450149536,
      "learning_rate": 1.0860928048974526e-05,
      "loss": 0.5031,
      "step": 14167
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8712313175201416,
      "learning_rate": 1.08598156152967e-05,
      "loss": 0.5737,
      "step": 14168
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7784169912338257,
      "learning_rate": 1.0858703170899232e-05,
      "loss": 0.5679,
      "step": 14169
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.60132896900177,
      "learning_rate": 1.0857590715795989e-05,
      "loss": 0.4824,
      "step": 14170
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.49937105178833,
      "learning_rate": 1.085647825000084e-05,
      "loss": 0.5771,
      "step": 14171
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5674179792404175,
      "learning_rate": 1.0855365773527657e-05,
      "loss": 0.584,
      "step": 14172
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5441612005233765,
      "learning_rate": 1.0854253286390307e-05,
      "loss": 0.5157,
      "step": 14173
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.559401512145996,
      "learning_rate": 1.0853140788602657e-05,
      "loss": 0.6324,
      "step": 14174
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.731961727142334,
      "learning_rate": 1.0852028280178586e-05,
      "loss": 0.6351,
      "step": 14175
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6988664865493774,
      "learning_rate": 1.0850915761131956e-05,
      "loss": 0.5574,
      "step": 14176
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7451062202453613,
      "learning_rate": 1.0849803231476641e-05,
      "loss": 0.638,
      "step": 14177
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4121925830841064,
      "learning_rate": 1.0848690691226506e-05,
      "loss": 0.6728,
      "step": 14178
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8380615711212158,
      "learning_rate": 1.0847578140395429e-05,
      "loss": 0.5327,
      "step": 14179
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4977588653564453,
      "learning_rate": 1.0846465578997278e-05,
      "loss": 0.5509,
      "step": 14180
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4715160131454468,
      "learning_rate": 1.0845353007045923e-05,
      "loss": 0.5306,
      "step": 14181
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5473041534423828,
      "learning_rate": 1.0844240424555234e-05,
      "loss": 0.5041,
      "step": 14182
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5946993827819824,
      "learning_rate": 1.0843127831539082e-05,
      "loss": 0.5183,
      "step": 14183
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5750389099121094,
      "learning_rate": 1.084201522801134e-05,
      "loss": 0.5139,
      "step": 14184
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7508375644683838,
      "learning_rate": 1.0840902613985878e-05,
      "loss": 0.5729,
      "step": 14185
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.424108862876892,
      "learning_rate": 1.0839789989476566e-05,
      "loss": 0.507,
      "step": 14186
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.041287899017334,
      "learning_rate": 1.0838677354497279e-05,
      "loss": 0.6772,
      "step": 14187
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6665211915969849,
      "learning_rate": 1.0837564709061886e-05,
      "loss": 0.5579,
      "step": 14188
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.273473024368286,
      "learning_rate": 1.0836452053184256e-05,
      "loss": 0.619,
      "step": 14189
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7407519817352295,
      "learning_rate": 1.083533938687827e-05,
      "loss": 0.3767,
      "step": 14190
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.530432939529419,
      "learning_rate": 1.0834226710157791e-05,
      "loss": 0.5356,
      "step": 14191
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4961092472076416,
      "learning_rate": 1.0833114023036695e-05,
      "loss": 0.5029,
      "step": 14192
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6049199104309082,
      "learning_rate": 1.0832001325528853e-05,
      "loss": 0.6232,
      "step": 14193
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.435860276222229,
      "learning_rate": 1.0830888617648142e-05,
      "loss": 0.5073,
      "step": 14194
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.566455364227295,
      "learning_rate": 1.0829775899408427e-05,
      "loss": 0.5982,
      "step": 14195
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5526231527328491,
      "learning_rate": 1.0828663170823586e-05,
      "loss": 0.7013,
      "step": 14196
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4476169347763062,
      "learning_rate": 1.0827550431907488e-05,
      "loss": 0.5618,
      "step": 14197
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3631315231323242,
      "learning_rate": 1.0826437682674012e-05,
      "loss": 0.6827,
      "step": 14198
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4126194715499878,
      "learning_rate": 1.0825324923137021e-05,
      "loss": 0.5495,
      "step": 14199
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7311532497406006,
      "learning_rate": 1.08242121533104e-05,
      "loss": 0.5412,
      "step": 14200
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5204362869262695,
      "learning_rate": 1.0823099373208014e-05,
      "loss": 0.4434,
      "step": 14201
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4191982746124268,
      "learning_rate": 1.082198658284374e-05,
      "loss": 0.5911,
      "step": 14202
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4946187734603882,
      "learning_rate": 1.0820873782231453e-05,
      "loss": 0.6203,
      "step": 14203
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4661892652511597,
      "learning_rate": 1.0819760971385022e-05,
      "loss": 0.6427,
      "step": 14204
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7304915189743042,
      "learning_rate": 1.0818648150318322e-05,
      "loss": 0.4217,
      "step": 14205
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.656505823135376,
      "learning_rate": 1.081753531904523e-05,
      "loss": 0.5601,
      "step": 14206
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.526693344116211,
      "learning_rate": 1.081642247757962e-05,
      "loss": 0.5464,
      "step": 14207
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6547231674194336,
      "learning_rate": 1.0815309625935363e-05,
      "loss": 0.5708,
      "step": 14208
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6822768449783325,
      "learning_rate": 1.0814196764126334e-05,
      "loss": 0.4763,
      "step": 14209
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6220355033874512,
      "learning_rate": 1.081308389216641e-05,
      "loss": 0.514,
      "step": 14210
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3441177606582642,
      "learning_rate": 1.0811971010069461e-05,
      "loss": 0.611,
      "step": 14211
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.25127911567688,
      "learning_rate": 1.0810858117849368e-05,
      "loss": 0.7007,
      "step": 14212
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6922136545181274,
      "learning_rate": 1.0809745215520002e-05,
      "loss": 0.5681,
      "step": 14213
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8254072666168213,
      "learning_rate": 1.0808632303095237e-05,
      "loss": 0.6168,
      "step": 14214
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5809507369995117,
      "learning_rate": 1.0807519380588949e-05,
      "loss": 0.6193,
      "step": 14215
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3079196214675903,
      "learning_rate": 1.0806406448015016e-05,
      "loss": 0.5484,
      "step": 14216
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5532469749450684,
      "learning_rate": 1.080529350538731e-05,
      "loss": 0.5621,
      "step": 14217
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.758715271949768,
      "learning_rate": 1.0804180552719707e-05,
      "loss": 0.457,
      "step": 14218
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.445806860923767,
      "learning_rate": 1.0803067590026085e-05,
      "loss": 0.5466,
      "step": 14219
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5140540599822998,
      "learning_rate": 1.0801954617320316e-05,
      "loss": 0.5341,
      "step": 14220
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6910220384597778,
      "learning_rate": 1.080084163461628e-05,
      "loss": 0.5387,
      "step": 14221
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5736956596374512,
      "learning_rate": 1.0799728641927849e-05,
      "loss": 0.533,
      "step": 14222
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3563058376312256,
      "learning_rate": 1.0798615639268899e-05,
      "loss": 0.5802,
      "step": 14223
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.803815245628357,
      "learning_rate": 1.0797502626653313e-05,
      "loss": 0.6158,
      "step": 14224
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4023243188858032,
      "learning_rate": 1.0796389604094959e-05,
      "loss": 0.5374,
      "step": 14225
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5222511291503906,
      "learning_rate": 1.0795276571607719e-05,
      "loss": 0.4801,
      "step": 14226
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5903321504592896,
      "learning_rate": 1.0794163529205463e-05,
      "loss": 0.5147,
      "step": 14227
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6465880870819092,
      "learning_rate": 1.0793050476902075e-05,
      "loss": 0.5518,
      "step": 14228
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4659171104431152,
      "learning_rate": 1.0791937414711431e-05,
      "loss": 0.5136,
      "step": 14229
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5405611991882324,
      "learning_rate": 1.0790824342647403e-05,
      "loss": 0.5354,
      "step": 14230
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.326185703277588,
      "learning_rate": 1.0789711260723874e-05,
      "loss": 0.6212,
      "step": 14231
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8333704471588135,
      "learning_rate": 1.0788598168954718e-05,
      "loss": 0.4914,
      "step": 14232
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.243233561515808,
      "learning_rate": 1.0787485067353809e-05,
      "loss": 0.6401,
      "step": 14233
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4967249631881714,
      "learning_rate": 1.0786371955935033e-05,
      "loss": 0.6905,
      "step": 14234
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8644708395004272,
      "learning_rate": 1.078525883471226e-05,
      "loss": 0.4795,
      "step": 14235
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4191486835479736,
      "learning_rate": 1.0784145703699369e-05,
      "loss": 0.4751,
      "step": 14236
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.52672278881073,
      "learning_rate": 1.078303256291024e-05,
      "loss": 0.5991,
      "step": 14237
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7384147644042969,
      "learning_rate": 1.0781919412358752e-05,
      "loss": 0.5196,
      "step": 14238
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.8353725671768188,
      "learning_rate": 1.0780806252058782e-05,
      "loss": 0.5107,
      "step": 14239
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5924965143203735,
      "learning_rate": 1.0779693082024203e-05,
      "loss": 0.5327,
      "step": 14240
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6735188961029053,
      "learning_rate": 1.07785799022689e-05,
      "loss": 0.6147,
      "step": 14241
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.826831579208374,
      "learning_rate": 1.077746671280675e-05,
      "loss": 0.5381,
      "step": 14242
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.630475401878357,
      "learning_rate": 1.077635351365163e-05,
      "loss": 0.5685,
      "step": 14243
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4752991199493408,
      "learning_rate": 1.0775240304817418e-05,
      "loss": 0.43,
      "step": 14244
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.514119267463684,
      "learning_rate": 1.0774127086317998e-05,
      "loss": 0.5572,
      "step": 14245
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.262024164199829,
      "learning_rate": 1.077301385816724e-05,
      "loss": 0.5581,
      "step": 14246
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6520369052886963,
      "learning_rate": 1.0771900620379032e-05,
      "loss": 0.6068,
      "step": 14247
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7937512397766113,
      "learning_rate": 1.0770787372967248e-05,
      "loss": 0.5952,
      "step": 14248
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5045255422592163,
      "learning_rate": 1.0769674115945768e-05,
      "loss": 0.5386,
      "step": 14249
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4102247953414917,
      "learning_rate": 1.0768560849328474e-05,
      "loss": 0.4793,
      "step": 14250
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.6751433610916138,
      "learning_rate": 1.076744757312924e-05,
      "loss": 0.5617,
      "step": 14251
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4895706176757812,
      "learning_rate": 1.0766334287361952e-05,
      "loss": 0.3816,
      "step": 14252
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5027711391448975,
      "learning_rate": 1.0765220992040485e-05,
      "loss": 0.5521,
      "step": 14253
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5720596313476562,
      "learning_rate": 1.076410768717872e-05,
      "loss": 0.5044,
      "step": 14254
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3485603332519531,
      "learning_rate": 1.076299437279054e-05,
      "loss": 0.5228,
      "step": 14255
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4556291103363037,
      "learning_rate": 1.0761881048889821e-05,
      "loss": 0.509,
      "step": 14256
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7037503719329834,
      "learning_rate": 1.0760767715490445e-05,
      "loss": 0.4982,
      "step": 14257
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3481320142745972,
      "learning_rate": 1.0759654372606294e-05,
      "loss": 0.5825,
      "step": 14258
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3672735691070557,
      "learning_rate": 1.0758541020251246e-05,
      "loss": 0.5933,
      "step": 14259
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3869017362594604,
      "learning_rate": 1.0757427658439183e-05,
      "loss": 0.5344,
      "step": 14260
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.5678675174713135,
      "learning_rate": 1.0756314287183984e-05,
      "loss": 0.5905,
      "step": 14261
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6360000371932983,
      "learning_rate": 1.075520090649953e-05,
      "loss": 0.5529,
      "step": 14262
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4413306713104248,
      "learning_rate": 1.0754087516399705e-05,
      "loss": 0.5534,
      "step": 14263
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5971872806549072,
      "learning_rate": 1.0752974116898388e-05,
      "loss": 0.6675,
      "step": 14264
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.524680256843567,
      "learning_rate": 1.0751860708009459e-05,
      "loss": 0.5052,
      "step": 14265
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7233338356018066,
      "learning_rate": 1.0750747289746799e-05,
      "loss": 0.5351,
      "step": 14266
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5361484289169312,
      "learning_rate": 1.074963386212429e-05,
      "loss": 0.613,
      "step": 14267
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8832417726516724,
      "learning_rate": 1.074852042515582e-05,
      "loss": 0.6853,
      "step": 14268
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6935045719146729,
      "learning_rate": 1.0747406978855261e-05,
      "loss": 0.625,
      "step": 14269
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2930134534835815,
      "learning_rate": 1.0746293523236497e-05,
      "loss": 0.5192,
      "step": 14270
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6097930669784546,
      "learning_rate": 1.0745180058313416e-05,
      "loss": 0.4757,
      "step": 14271
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6746642589569092,
      "learning_rate": 1.0744066584099893e-05,
      "loss": 0.5852,
      "step": 14272
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6146782636642456,
      "learning_rate": 1.0742953100609811e-05,
      "loss": 0.5644,
      "step": 14273
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5016615390777588,
      "learning_rate": 1.0741839607857056e-05,
      "loss": 0.6327,
      "step": 14274
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.790229082107544,
      "learning_rate": 1.0740726105855508e-05,
      "loss": 0.5338,
      "step": 14275
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7720099687576294,
      "learning_rate": 1.0739612594619051e-05,
      "loss": 0.5058,
      "step": 14276
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.725406289100647,
      "learning_rate": 1.0738499074161564e-05,
      "loss": 0.4861,
      "step": 14277
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6230840682983398,
      "learning_rate": 1.0737385544496931e-05,
      "loss": 0.4222,
      "step": 14278
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4996908903121948,
      "learning_rate": 1.0736272005639036e-05,
      "loss": 0.6375,
      "step": 14279
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.494950532913208,
      "learning_rate": 1.0735158457601764e-05,
      "loss": 0.5327,
      "step": 14280
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6560699939727783,
      "learning_rate": 1.0734044900398994e-05,
      "loss": 0.5929,
      "step": 14281
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.316231369972229,
      "learning_rate": 1.0732931334044612e-05,
      "loss": 0.6124,
      "step": 14282
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.000671625137329,
      "learning_rate": 1.0731817758552496e-05,
      "loss": 0.5862,
      "step": 14283
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2629941701889038,
      "learning_rate": 1.073070417393654e-05,
      "loss": 0.4729,
      "step": 14284
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.15235435962677,
      "learning_rate": 1.0729590580210614e-05,
      "loss": 0.4583,
      "step": 14285
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3750309944152832,
      "learning_rate": 1.0728476977388612e-05,
      "loss": 0.4827,
      "step": 14286
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3295339345932007,
      "learning_rate": 1.0727363365484414e-05,
      "loss": 0.5243,
      "step": 14287
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6159319877624512,
      "learning_rate": 1.0726249744511903e-05,
      "loss": 0.5807,
      "step": 14288
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.750501036643982,
      "learning_rate": 1.0725136114484965e-05,
      "loss": 0.536,
      "step": 14289
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7164853811264038,
      "learning_rate": 1.0724022475417482e-05,
      "loss": 0.5461,
      "step": 14290
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5204942226409912,
      "learning_rate": 1.072290882732334e-05,
      "loss": 0.5439,
      "step": 14291
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6553802490234375,
      "learning_rate": 1.072179517021642e-05,
      "loss": 0.6342,
      "step": 14292
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4142723083496094,
      "learning_rate": 1.0720681504110608e-05,
      "loss": 0.5415,
      "step": 14293
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5303443670272827,
      "learning_rate": 1.0719567829019795e-05,
      "loss": 0.5368,
      "step": 14294
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.780000925064087,
      "learning_rate": 1.0718454144957854e-05,
      "loss": 0.5346,
      "step": 14295
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6435596942901611,
      "learning_rate": 1.0717340451938679e-05,
      "loss": 0.5809,
      "step": 14296
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.35862398147583,
      "learning_rate": 1.0716226749976148e-05,
      "loss": 0.6606,
      "step": 14297
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4679713249206543,
      "learning_rate": 1.0715113039084152e-05,
      "loss": 0.5954,
      "step": 14298
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7263299226760864,
      "learning_rate": 1.0713999319276574e-05,
      "loss": 0.6341,
      "step": 14299
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4119536876678467,
      "learning_rate": 1.0712885590567296e-05,
      "loss": 0.5785,
      "step": 14300
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5556550025939941,
      "learning_rate": 1.0711771852970205e-05,
      "loss": 0.6275,
      "step": 14301
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5077805519104004,
      "learning_rate": 1.0710658106499191e-05,
      "loss": 0.6207,
      "step": 14302
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.649566650390625,
      "learning_rate": 1.0709544351168134e-05,
      "loss": 0.6092,
      "step": 14303
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.518407940864563,
      "learning_rate": 1.0708430586990921e-05,
      "loss": 0.5517,
      "step": 14304
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5513653755187988,
      "learning_rate": 1.0707316813981439e-05,
      "loss": 0.6211,
      "step": 14305
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2558305263519287,
      "learning_rate": 1.070620303215357e-05,
      "loss": 0.49,
      "step": 14306
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.249509334564209,
      "learning_rate": 1.0705089241521204e-05,
      "loss": 0.4872,
      "step": 14307
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4785451889038086,
      "learning_rate": 1.0703975442098226e-05,
      "loss": 0.6107,
      "step": 14308
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4196045398712158,
      "learning_rate": 1.0702861633898523e-05,
      "loss": 0.5175,
      "step": 14309
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3361258506774902,
      "learning_rate": 1.070174781693598e-05,
      "loss": 0.5084,
      "step": 14310
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8956621885299683,
      "learning_rate": 1.0700633991224483e-05,
      "loss": 0.4983,
      "step": 14311
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4693551063537598,
      "learning_rate": 1.0699520156777919e-05,
      "loss": 0.5038,
      "step": 14312
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3845455646514893,
      "learning_rate": 1.0698406313610175e-05,
      "loss": 0.5515,
      "step": 14313
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5846168994903564,
      "learning_rate": 1.0697292461735135e-05,
      "loss": 0.6049,
      "step": 14314
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4838353395462036,
      "learning_rate": 1.0696178601166692e-05,
      "loss": 0.5168,
      "step": 14315
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5704762935638428,
      "learning_rate": 1.0695064731918725e-05,
      "loss": 0.6022,
      "step": 14316
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4330288171768188,
      "learning_rate": 1.069395085400513e-05,
      "loss": 0.5317,
      "step": 14317
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4868170022964478,
      "learning_rate": 1.0692836967439786e-05,
      "loss": 0.5196,
      "step": 14318
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6356629133224487,
      "learning_rate": 1.0691723072236582e-05,
      "loss": 0.4961,
      "step": 14319
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5542384386062622,
      "learning_rate": 1.0690609168409412e-05,
      "loss": 0.5346,
      "step": 14320
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5228602886199951,
      "learning_rate": 1.0689495255972153e-05,
      "loss": 0.4931,
      "step": 14321
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3821520805358887,
      "learning_rate": 1.06883813349387e-05,
      "loss": 0.3964,
      "step": 14322
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6066033840179443,
      "learning_rate": 1.068726740532294e-05,
      "loss": 0.5836,
      "step": 14323
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6278828382492065,
      "learning_rate": 1.0686153467138758e-05,
      "loss": 0.4942,
      "step": 14324
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.2950332164764404,
      "learning_rate": 1.0685039520400043e-05,
      "loss": 0.6807,
      "step": 14325
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8006150722503662,
      "learning_rate": 1.0683925565120685e-05,
      "loss": 0.5247,
      "step": 14326
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5540865659713745,
      "learning_rate": 1.0682811601314567e-05,
      "loss": 0.6624,
      "step": 14327
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6547694206237793,
      "learning_rate": 1.0681697628995584e-05,
      "loss": 0.5648,
      "step": 14328
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4963666200637817,
      "learning_rate": 1.0680583648177618e-05,
      "loss": 0.5201,
      "step": 14329
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3789513111114502,
      "learning_rate": 1.067946965887456e-05,
      "loss": 0.4819,
      "step": 14330
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.057943820953369,
      "learning_rate": 1.06783556611003e-05,
      "loss": 0.6203,
      "step": 14331
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2824398279190063,
      "learning_rate": 1.0677241654868724e-05,
      "loss": 0.4792,
      "step": 14332
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.691933512687683,
      "learning_rate": 1.0676127640193724e-05,
      "loss": 0.4934,
      "step": 14333
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6739037036895752,
      "learning_rate": 1.0675013617089187e-05,
      "loss": 0.6005,
      "step": 14334
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.609131932258606,
      "learning_rate": 1.0673899585569001e-05,
      "loss": 0.5555,
      "step": 14335
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4195266962051392,
      "learning_rate": 1.0672785545647055e-05,
      "loss": 0.5173,
      "step": 14336
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7154678106307983,
      "learning_rate": 1.0671671497337239e-05,
      "loss": 0.6515,
      "step": 14337
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.614233374595642,
      "learning_rate": 1.0670557440653444e-05,
      "loss": 0.5037,
      "step": 14338
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4001121520996094,
      "learning_rate": 1.0669443375609558e-05,
      "loss": 0.605,
      "step": 14339
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4174507856369019,
      "learning_rate": 1.0668329302219469e-05,
      "loss": 0.5672,
      "step": 14340
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.225250244140625,
      "learning_rate": 1.0667215220497068e-05,
      "loss": 0.4513,
      "step": 14341
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.9333258867263794,
      "learning_rate": 1.0666101130456244e-05,
      "loss": 0.5289,
      "step": 14342
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.670411467552185,
      "learning_rate": 1.0664987032110887e-05,
      "loss": 0.5434,
      "step": 14343
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3990322351455688,
      "learning_rate": 1.066387292547489e-05,
      "loss": 0.5782,
      "step": 14344
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.499780535697937,
      "learning_rate": 1.0662758810562135e-05,
      "loss": 0.574,
      "step": 14345
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8087431192398071,
      "learning_rate": 1.0661644687386519e-05,
      "loss": 0.5732,
      "step": 14346
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5506263971328735,
      "learning_rate": 1.0660530555961933e-05,
      "loss": 0.5554,
      "step": 14347
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6254531145095825,
      "learning_rate": 1.0659416416302262e-05,
      "loss": 0.4599,
      "step": 14348
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.8181636333465576,
      "learning_rate": 1.0658302268421402e-05,
      "loss": 0.5942,
      "step": 14349
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4106214046478271,
      "learning_rate": 1.0657188112333236e-05,
      "loss": 0.4962,
      "step": 14350
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2922191619873047,
      "learning_rate": 1.0656073948051663e-05,
      "loss": 0.5916,
      "step": 14351
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4868738651275635,
      "learning_rate": 1.0654959775590568e-05,
      "loss": 0.4703,
      "step": 14352
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6292575597763062,
      "learning_rate": 1.0653845594963843e-05,
      "loss": 0.551,
      "step": 14353
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3049426078796387,
      "learning_rate": 1.0652731406185382e-05,
      "loss": 0.4798,
      "step": 14354
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.448813796043396,
      "learning_rate": 1.0651617209269071e-05,
      "loss": 0.4898,
      "step": 14355
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4720497131347656,
      "learning_rate": 1.0650503004228804e-05,
      "loss": 0.4287,
      "step": 14356
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5401345491409302,
      "learning_rate": 1.0649388791078475e-05,
      "loss": 0.4942,
      "step": 14357
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.439738392829895,
      "learning_rate": 1.0648274569831969e-05,
      "loss": 0.5364,
      "step": 14358
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6555248498916626,
      "learning_rate": 1.064716034050318e-05,
      "loss": 0.6567,
      "step": 14359
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5105055570602417,
      "learning_rate": 1.0646046103106004e-05,
      "loss": 0.5444,
      "step": 14360
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.752280831336975,
      "learning_rate": 1.0644931857654326e-05,
      "loss": 0.5457,
      "step": 14361
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3531733751296997,
      "learning_rate": 1.0643817604162041e-05,
      "loss": 0.5032,
      "step": 14362
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.0131661891937256,
      "learning_rate": 1.0642703342643039e-05,
      "loss": 0.6509,
      "step": 14363
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.9659358263015747,
      "learning_rate": 1.0641589073111213e-05,
      "loss": 0.5944,
      "step": 14364
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5390688180923462,
      "learning_rate": 1.0640474795580459e-05,
      "loss": 0.5643,
      "step": 14365
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5667057037353516,
      "learning_rate": 1.0639360510064663e-05,
      "loss": 0.4916,
      "step": 14366
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2639497518539429,
      "learning_rate": 1.0638246216577719e-05,
      "loss": 0.6104,
      "step": 14367
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5423787832260132,
      "learning_rate": 1.063713191513352e-05,
      "loss": 0.5884,
      "step": 14368
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.46620774269104,
      "learning_rate": 1.0636017605745959e-05,
      "loss": 0.5417,
      "step": 14369
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.714786171913147,
      "learning_rate": 1.0634903288428929e-05,
      "loss": 0.6152,
      "step": 14370
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.815518856048584,
      "learning_rate": 1.0633788963196318e-05,
      "loss": 0.5803,
      "step": 14371
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3231819868087769,
      "learning_rate": 1.0632674630062024e-05,
      "loss": 0.5452,
      "step": 14372
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4350796937942505,
      "learning_rate": 1.0631560289039938e-05,
      "loss": 0.5347,
      "step": 14373
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5758122205734253,
      "learning_rate": 1.0630445940143953e-05,
      "loss": 0.5742,
      "step": 14374
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.441247582435608,
      "learning_rate": 1.062933158338796e-05,
      "loss": 0.4727,
      "step": 14375
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6027594804763794,
      "learning_rate": 1.0628217218785857e-05,
      "loss": 0.575,
      "step": 14376
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.480544090270996,
      "learning_rate": 1.0627102846351532e-05,
      "loss": 0.4605,
      "step": 14377
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.531367540359497,
      "learning_rate": 1.0625988466098882e-05,
      "loss": 0.6072,
      "step": 14378
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7167630195617676,
      "learning_rate": 1.0624874078041795e-05,
      "loss": 0.6078,
      "step": 14379
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3833693265914917,
      "learning_rate": 1.0623759682194175e-05,
      "loss": 0.4132,
      "step": 14380
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.425179123878479,
      "learning_rate": 1.0622645278569902e-05,
      "loss": 0.5659,
      "step": 14381
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4472206830978394,
      "learning_rate": 1.0621530867182879e-05,
      "loss": 0.4502,
      "step": 14382
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8036980628967285,
      "learning_rate": 1.0620416448047e-05,
      "loss": 0.5802,
      "step": 14383
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.510331630706787,
      "learning_rate": 1.0619302021176155e-05,
      "loss": 0.4935,
      "step": 14384
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8253240585327148,
      "learning_rate": 1.0618187586584237e-05,
      "loss": 0.6195,
      "step": 14385
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2278577089309692,
      "learning_rate": 1.0617073144285144e-05,
      "loss": 0.5404,
      "step": 14386
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2591665983200073,
      "learning_rate": 1.0615958694292768e-05,
      "loss": 0.4827,
      "step": 14387
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.374799132347107,
      "learning_rate": 1.0614844236621008e-05,
      "loss": 0.5224,
      "step": 14388
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5848853588104248,
      "learning_rate": 1.0613729771283747e-05,
      "loss": 0.5857,
      "step": 14389
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.257279396057129,
      "learning_rate": 1.061261529829489e-05,
      "loss": 0.5297,
      "step": 14390
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.522370457649231,
      "learning_rate": 1.061150081766833e-05,
      "loss": 0.5329,
      "step": 14391
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2492659091949463,
      "learning_rate": 1.0610386329417956e-05,
      "loss": 0.5807,
      "step": 14392
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4768551588058472,
      "learning_rate": 1.060927183355767e-05,
      "loss": 0.5066,
      "step": 14393
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8964735269546509,
      "learning_rate": 1.060815733010136e-05,
      "loss": 0.5367,
      "step": 14394
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.584464430809021,
      "learning_rate": 1.0607042819062925e-05,
      "loss": 0.4572,
      "step": 14395
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.490524172782898,
      "learning_rate": 1.0605928300456264e-05,
      "loss": 0.5465,
      "step": 14396
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6775377988815308,
      "learning_rate": 1.0604813774295261e-05,
      "loss": 0.6431,
      "step": 14397
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.727016806602478,
      "learning_rate": 1.060369924059382e-05,
      "loss": 0.5155,
      "step": 14398
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6586297750473022,
      "learning_rate": 1.0602584699365837e-05,
      "loss": 0.596,
      "step": 14399
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.595440149307251,
      "learning_rate": 1.0601470150625199e-05,
      "loss": 0.4852,
      "step": 14400
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1789084672927856,
      "learning_rate": 1.0600355594385812e-05,
      "loss": 0.483,
      "step": 14401
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3751885890960693,
      "learning_rate": 1.0599241030661562e-05,
      "loss": 0.4165,
      "step": 14402
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6162004470825195,
      "learning_rate": 1.0598126459466352e-05,
      "loss": 0.5526,
      "step": 14403
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4862102270126343,
      "learning_rate": 1.0597011880814077e-05,
      "loss": 0.4895,
      "step": 14404
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7856519222259521,
      "learning_rate": 1.0595897294718627e-05,
      "loss": 0.5735,
      "step": 14405
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8148514032363892,
      "learning_rate": 1.0594782701193902e-05,
      "loss": 0.5569,
      "step": 14406
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5921598672866821,
      "learning_rate": 1.05936681002538e-05,
      "loss": 0.5537,
      "step": 14407
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4551136493682861,
      "learning_rate": 1.0592553491912211e-05,
      "loss": 0.4952,
      "step": 14408
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.738966464996338,
      "learning_rate": 1.059143887618304e-05,
      "loss": 0.5633,
      "step": 14409
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6709696054458618,
      "learning_rate": 1.0590324253080175e-05,
      "loss": 0.6339,
      "step": 14410
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5463651418685913,
      "learning_rate": 1.0589209622617517e-05,
      "loss": 0.5254,
      "step": 14411
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.516855001449585,
      "learning_rate": 1.0588094984808963e-05,
      "loss": 0.4827,
      "step": 14412
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5972882509231567,
      "learning_rate": 1.0586980339668407e-05,
      "loss": 0.4162,
      "step": 14413
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8760367631912231,
      "learning_rate": 1.0585865687209747e-05,
      "loss": 0.6527,
      "step": 14414
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4348289966583252,
      "learning_rate": 1.0584751027446878e-05,
      "loss": 0.576,
      "step": 14415
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5961121320724487,
      "learning_rate": 1.0583636360393698e-05,
      "loss": 0.5552,
      "step": 14416
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.643925428390503,
      "learning_rate": 1.0582521686064109e-05,
      "loss": 0.5466,
      "step": 14417
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5725080966949463,
      "learning_rate": 1.0581407004472e-05,
      "loss": 0.5122,
      "step": 14418
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4964210987091064,
      "learning_rate": 1.0580292315631274e-05,
      "loss": 0.5672,
      "step": 14419
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4358479976654053,
      "learning_rate": 1.0579177619555823e-05,
      "loss": 0.4895,
      "step": 14420
    },
    {
      "epoch": 1.49,
      "grad_norm": 2.227267026901245,
      "learning_rate": 1.0578062916259548e-05,
      "loss": 0.6102,
      "step": 14421
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3416601419448853,
      "learning_rate": 1.0576948205756348e-05,
      "loss": 0.5352,
      "step": 14422
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.556626319885254,
      "learning_rate": 1.0575833488060115e-05,
      "loss": 0.5794,
      "step": 14423
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5021154880523682,
      "learning_rate": 1.057471876318475e-05,
      "loss": 0.5443,
      "step": 14424
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.734513282775879,
      "learning_rate": 1.0573604031144154e-05,
      "loss": 0.5824,
      "step": 14425
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6270784139633179,
      "learning_rate": 1.0572489291952217e-05,
      "loss": 0.6312,
      "step": 14426
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5978542566299438,
      "learning_rate": 1.0571374545622843e-05,
      "loss": 0.5024,
      "step": 14427
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.364072561264038,
      "learning_rate": 1.0570259792169928e-05,
      "loss": 0.4712,
      "step": 14428
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6711969375610352,
      "learning_rate": 1.0569145031607371e-05,
      "loss": 0.6356,
      "step": 14429
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5357989072799683,
      "learning_rate": 1.056803026394907e-05,
      "loss": 0.4969,
      "step": 14430
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6040488481521606,
      "learning_rate": 1.0566915489208918e-05,
      "loss": 0.6874,
      "step": 14431
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.601341962814331,
      "learning_rate": 1.0565800707400823e-05,
      "loss": 0.5534,
      "step": 14432
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.601556420326233,
      "learning_rate": 1.0564685918538675e-05,
      "loss": 0.5757,
      "step": 14433
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6124556064605713,
      "learning_rate": 1.0563571122636375e-05,
      "loss": 0.544,
      "step": 14434
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8263115882873535,
      "learning_rate": 1.0562456319707826e-05,
      "loss": 0.5952,
      "step": 14435
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8526527881622314,
      "learning_rate": 1.056134150976692e-05,
      "loss": 0.6347,
      "step": 14436
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8580271005630493,
      "learning_rate": 1.0560226692827557e-05,
      "loss": 0.4853,
      "step": 14437
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4256399869918823,
      "learning_rate": 1.0559111868903642e-05,
      "loss": 0.4354,
      "step": 14438
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.9554334878921509,
      "learning_rate": 1.0557997038009067e-05,
      "loss": 0.5261,
      "step": 14439
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5964608192443848,
      "learning_rate": 1.0556882200157736e-05,
      "loss": 0.3596,
      "step": 14440
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.8678699731826782,
      "learning_rate": 1.0555767355363542e-05,
      "loss": 0.6206,
      "step": 14441
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.357017159461975,
      "learning_rate": 1.0554652503640391e-05,
      "loss": 0.6024,
      "step": 14442
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5301361083984375,
      "learning_rate": 1.0553537645002177e-05,
      "loss": 0.5844,
      "step": 14443
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3076602220535278,
      "learning_rate": 1.05524227794628e-05,
      "loss": 0.6184,
      "step": 14444
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4172781705856323,
      "learning_rate": 1.0551307907036166e-05,
      "loss": 0.5002,
      "step": 14445
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7282484769821167,
      "learning_rate": 1.0550193027736164e-05,
      "loss": 0.6652,
      "step": 14446
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4875138998031616,
      "learning_rate": 1.05490781415767e-05,
      "loss": 0.5145,
      "step": 14447
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.523672103881836,
      "learning_rate": 1.0547963248571678e-05,
      "loss": 0.582,
      "step": 14448
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2880038022994995,
      "learning_rate": 1.0546848348734986e-05,
      "loss": 0.4784,
      "step": 14449
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7472161054611206,
      "learning_rate": 1.0545733442080532e-05,
      "loss": 0.5115,
      "step": 14450
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4198048114776611,
      "learning_rate": 1.0544618528622217e-05,
      "loss": 0.6061,
      "step": 14451
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4913406372070312,
      "learning_rate": 1.0543503608373935e-05,
      "loss": 0.5766,
      "step": 14452
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.562882661819458,
      "learning_rate": 1.0542388681349594e-05,
      "loss": 0.7039,
      "step": 14453
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4620251655578613,
      "learning_rate": 1.0541273747563086e-05,
      "loss": 0.5172,
      "step": 14454
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6162604093551636,
      "learning_rate": 1.0540158807028313e-05,
      "loss": 0.5554,
      "step": 14455
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.493710994720459,
      "learning_rate": 1.0539043859759181e-05,
      "loss": 0.5525,
      "step": 14456
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.456642985343933,
      "learning_rate": 1.0537928905769588e-05,
      "loss": 0.4781,
      "step": 14457
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4287185668945312,
      "learning_rate": 1.0536813945073432e-05,
      "loss": 0.4559,
      "step": 14458
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5232747793197632,
      "learning_rate": 1.0535698977684614e-05,
      "loss": 0.5732,
      "step": 14459
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5825703144073486,
      "learning_rate": 1.0534584003617034e-05,
      "loss": 0.6075,
      "step": 14460
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5773530006408691,
      "learning_rate": 1.05334690228846e-05,
      "loss": 0.6441,
      "step": 14461
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9960185289382935,
      "learning_rate": 1.0532354035501202e-05,
      "loss": 0.6303,
      "step": 14462
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1939562559127808,
      "learning_rate": 1.053123904148075e-05,
      "loss": 0.5248,
      "step": 14463
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3624765872955322,
      "learning_rate": 1.0530124040837139e-05,
      "loss": 0.531,
      "step": 14464
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7145427465438843,
      "learning_rate": 1.0529009033584274e-05,
      "loss": 0.5479,
      "step": 14465
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.377819538116455,
      "learning_rate": 1.0527894019736057e-05,
      "loss": 0.3749,
      "step": 14466
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5191099643707275,
      "learning_rate": 1.0526778999306385e-05,
      "loss": 0.4204,
      "step": 14467
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.285637617111206,
      "learning_rate": 1.052566397230916e-05,
      "loss": 0.4544,
      "step": 14468
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5937565565109253,
      "learning_rate": 1.0524548938758288e-05,
      "loss": 0.649,
      "step": 14469
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6726120710372925,
      "learning_rate": 1.0523433898667663e-05,
      "loss": 0.5151,
      "step": 14470
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3234554529190063,
      "learning_rate": 1.0522318852051196e-05,
      "loss": 0.4587,
      "step": 14471
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.494098424911499,
      "learning_rate": 1.052120379892278e-05,
      "loss": 0.5742,
      "step": 14472
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6253165006637573,
      "learning_rate": 1.0520088739296323e-05,
      "loss": 0.5609,
      "step": 14473
    },
    {
      "epoch": 1.5,
      "grad_norm": 2.1044728755950928,
      "learning_rate": 1.0518973673185726e-05,
      "loss": 0.6579,
      "step": 14474
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6387995481491089,
      "learning_rate": 1.0517858600604886e-05,
      "loss": 0.6445,
      "step": 14475
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3164256811141968,
      "learning_rate": 1.051674352156771e-05,
      "loss": 0.4777,
      "step": 14476
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.8473292589187622,
      "learning_rate": 1.0515628436088099e-05,
      "loss": 0.6402,
      "step": 14477
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1848070621490479,
      "learning_rate": 1.0514513344179952e-05,
      "loss": 0.4778,
      "step": 14478
    },
    {
      "epoch": 1.5,
      "grad_norm": 2.343928098678589,
      "learning_rate": 1.0513398245857177e-05,
      "loss": 0.5102,
      "step": 14479
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4202890396118164,
      "learning_rate": 1.0512283141133673e-05,
      "loss": 0.4889,
      "step": 14480
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5010002851486206,
      "learning_rate": 1.051116803002334e-05,
      "loss": 0.5649,
      "step": 14481
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.782700777053833,
      "learning_rate": 1.0510052912540087e-05,
      "loss": 0.6503,
      "step": 14482
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9322789907455444,
      "learning_rate": 1.050893778869781e-05,
      "loss": 0.5991,
      "step": 14483
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.506070852279663,
      "learning_rate": 1.0507822658510418e-05,
      "loss": 0.6052,
      "step": 14484
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1626529693603516,
      "learning_rate": 1.0506707521991807e-05,
      "loss": 0.5913,
      "step": 14485
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5327785015106201,
      "learning_rate": 1.0505592379155884e-05,
      "loss": 0.4796,
      "step": 14486
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7115205526351929,
      "learning_rate": 1.0504477230016555e-05,
      "loss": 0.6868,
      "step": 14487
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4272944927215576,
      "learning_rate": 1.0503362074587713e-05,
      "loss": 0.5527,
      "step": 14488
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.8501371145248413,
      "learning_rate": 1.0502246912883269e-05,
      "loss": 0.6898,
      "step": 14489
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.589698314666748,
      "learning_rate": 1.0501131744917125e-05,
      "loss": 0.4995,
      "step": 14490
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6243436336517334,
      "learning_rate": 1.0500016570703185e-05,
      "loss": 0.6442,
      "step": 14491
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3158698081970215,
      "learning_rate": 1.049890139025535e-05,
      "loss": 0.5115,
      "step": 14492
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.546298623085022,
      "learning_rate": 1.0497786203587521e-05,
      "loss": 0.5853,
      "step": 14493
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3216830492019653,
      "learning_rate": 1.0496671010713606e-05,
      "loss": 0.5092,
      "step": 14494
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4859824180603027,
      "learning_rate": 1.049555581164751e-05,
      "loss": 0.5452,
      "step": 14495
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.261572003364563,
      "learning_rate": 1.0494440606403132e-05,
      "loss": 0.5159,
      "step": 14496
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.8189443349838257,
      "learning_rate": 1.0493325394994376e-05,
      "loss": 0.5721,
      "step": 14497
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.622491717338562,
      "learning_rate": 1.0492210177435147e-05,
      "loss": 0.6149,
      "step": 14498
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4326168298721313,
      "learning_rate": 1.049109495373935e-05,
      "loss": 0.5098,
      "step": 14499
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5182204246520996,
      "learning_rate": 1.0489979723920891e-05,
      "loss": 0.5053,
      "step": 14500
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4042514562606812,
      "learning_rate": 1.0488864487993669e-05,
      "loss": 0.5772,
      "step": 14501
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6311651468276978,
      "learning_rate": 1.0487749245971585e-05,
      "loss": 0.4683,
      "step": 14502
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7902988195419312,
      "learning_rate": 1.0486633997868554e-05,
      "loss": 0.5049,
      "step": 14503
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4787428379058838,
      "learning_rate": 1.0485518743698472e-05,
      "loss": 0.5495,
      "step": 14504
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3907897472381592,
      "learning_rate": 1.0484403483475245e-05,
      "loss": 0.5461,
      "step": 14505
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4528084993362427,
      "learning_rate": 1.0483288217212779e-05,
      "loss": 0.5227,
      "step": 14506
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6345096826553345,
      "learning_rate": 1.0482172944924978e-05,
      "loss": 0.6046,
      "step": 14507
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7803385257720947,
      "learning_rate": 1.0481057666625745e-05,
      "loss": 0.5764,
      "step": 14508
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3781379461288452,
      "learning_rate": 1.0479942382328984e-05,
      "loss": 0.5024,
      "step": 14509
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.55264151096344,
      "learning_rate": 1.0478827092048603e-05,
      "loss": 0.3614,
      "step": 14510
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6792019605636597,
      "learning_rate": 1.0477711795798506e-05,
      "loss": 0.603,
      "step": 14511
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.391355037689209,
      "learning_rate": 1.0476596493592595e-05,
      "loss": 0.6181,
      "step": 14512
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6138412952423096,
      "learning_rate": 1.0475481185444778e-05,
      "loss": 0.6606,
      "step": 14513
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1984914541244507,
      "learning_rate": 1.0474365871368956e-05,
      "loss": 0.4991,
      "step": 14514
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7836651802062988,
      "learning_rate": 1.0473250551379038e-05,
      "loss": 0.5461,
      "step": 14515
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.8014029264450073,
      "learning_rate": 1.0472135225488929e-05,
      "loss": 0.4478,
      "step": 14516
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5806511640548706,
      "learning_rate": 1.047101989371253e-05,
      "loss": 0.5791,
      "step": 14517
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.529930830001831,
      "learning_rate": 1.046990455606375e-05,
      "loss": 0.5389,
      "step": 14518
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5092122554779053,
      "learning_rate": 1.0468789212556496e-05,
      "loss": 0.4099,
      "step": 14519
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.523768663406372,
      "learning_rate": 1.0467673863204669e-05,
      "loss": 0.5006,
      "step": 14520
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3322077989578247,
      "learning_rate": 1.0466558508022178e-05,
      "loss": 0.4775,
      "step": 14521
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2734425067901611,
      "learning_rate": 1.0465443147022925e-05,
      "loss": 0.4325,
      "step": 14522
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5356154441833496,
      "learning_rate": 1.0464327780220818e-05,
      "loss": 0.4623,
      "step": 14523
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4049367904663086,
      "learning_rate": 1.0463212407629764e-05,
      "loss": 0.435,
      "step": 14524
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7015209197998047,
      "learning_rate": 1.0462097029263665e-05,
      "loss": 0.6114,
      "step": 14525
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7858355045318604,
      "learning_rate": 1.0460981645136432e-05,
      "loss": 0.6711,
      "step": 14526
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4229917526245117,
      "learning_rate": 1.0459866255261964e-05,
      "loss": 0.4143,
      "step": 14527
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4885733127593994,
      "learning_rate": 1.0458750859654172e-05,
      "loss": 0.5399,
      "step": 14528
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4169503450393677,
      "learning_rate": 1.0457635458326962e-05,
      "loss": 0.4849,
      "step": 14529
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6297224760055542,
      "learning_rate": 1.0456520051294237e-05,
      "loss": 0.5234,
      "step": 14530
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4287176132202148,
      "learning_rate": 1.0455404638569907e-05,
      "loss": 0.4604,
      "step": 14531
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5550700426101685,
      "learning_rate": 1.0454289220167875e-05,
      "loss": 0.5459,
      "step": 14532
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.7299774885177612,
      "learning_rate": 1.045317379610205e-05,
      "loss": 0.7328,
      "step": 14533
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4641141891479492,
      "learning_rate": 1.0452058366386337e-05,
      "loss": 0.495,
      "step": 14534
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3087999820709229,
      "learning_rate": 1.0450942931034643e-05,
      "loss": 0.5705,
      "step": 14535
    },
    {
      "epoch": 1.5,
      "grad_norm": 2.006072521209717,
      "learning_rate": 1.0449827490060871e-05,
      "loss": 0.6452,
      "step": 14536
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.726684808731079,
      "learning_rate": 1.0448712043478935e-05,
      "loss": 0.6123,
      "step": 14537
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6391615867614746,
      "learning_rate": 1.0447596591302735e-05,
      "loss": 0.4758,
      "step": 14538
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4635156393051147,
      "learning_rate": 1.0446481133546183e-05,
      "loss": 0.5,
      "step": 14539
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6967517137527466,
      "learning_rate": 1.044536567022318e-05,
      "loss": 0.6536,
      "step": 14540
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.744762659072876,
      "learning_rate": 1.0444250201347637e-05,
      "loss": 0.6257,
      "step": 14541
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3824598789215088,
      "learning_rate": 1.0443134726933462e-05,
      "loss": 0.4776,
      "step": 14542
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.43975031375885,
      "learning_rate": 1.0442019246994556e-05,
      "loss": 0.4901,
      "step": 14543
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5591721534729004,
      "learning_rate": 1.0440903761544834e-05,
      "loss": 0.5174,
      "step": 14544
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.968446969985962,
      "learning_rate": 1.0439788270598199e-05,
      "loss": 0.5934,
      "step": 14545
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.522132158279419,
      "learning_rate": 1.0438672774168556e-05,
      "loss": 0.554,
      "step": 14546
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5633385181427002,
      "learning_rate": 1.0437557272269819e-05,
      "loss": 0.57,
      "step": 14547
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.9186997413635254,
      "learning_rate": 1.0436441764915887e-05,
      "loss": 0.6367,
      "step": 14548
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.386482834815979,
      "learning_rate": 1.0435326252120672e-05,
      "loss": 0.5335,
      "step": 14549
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.5733784437179565,
      "learning_rate": 1.0434210733898085e-05,
      "loss": 0.5489,
      "step": 14550
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.8905553817749023,
      "learning_rate": 1.0433095210262026e-05,
      "loss": 0.6015,
      "step": 14551
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5632708072662354,
      "learning_rate": 1.0431979681226409e-05,
      "loss": 0.5425,
      "step": 14552
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4431798458099365,
      "learning_rate": 1.0430864146805137e-05,
      "loss": 0.3903,
      "step": 14553
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4092705249786377,
      "learning_rate": 1.042974860701212e-05,
      "loss": 0.4842,
      "step": 14554
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5572822093963623,
      "learning_rate": 1.0428633061861268e-05,
      "loss": 0.5661,
      "step": 14555
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8702905178070068,
      "learning_rate": 1.0427517511366485e-05,
      "loss": 0.6137,
      "step": 14556
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.287410855293274,
      "learning_rate": 1.0426401955541681e-05,
      "loss": 0.4422,
      "step": 14557
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.56645929813385,
      "learning_rate": 1.0425286394400765e-05,
      "loss": 0.5971,
      "step": 14558
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6988632678985596,
      "learning_rate": 1.0424170827957642e-05,
      "loss": 0.534,
      "step": 14559
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6252715587615967,
      "learning_rate": 1.0423055256226224e-05,
      "loss": 0.4435,
      "step": 14560
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6674787998199463,
      "learning_rate": 1.0421939679220417e-05,
      "loss": 0.5061,
      "step": 14561
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6175589561462402,
      "learning_rate": 1.0420824096954128e-05,
      "loss": 0.6096,
      "step": 14562
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.456963300704956,
      "learning_rate": 1.041970850944127e-05,
      "loss": 0.4731,
      "step": 14563
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5678690671920776,
      "learning_rate": 1.0418592916695744e-05,
      "loss": 0.5698,
      "step": 14564
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.28415846824646,
      "learning_rate": 1.0417477318731466e-05,
      "loss": 0.5296,
      "step": 14565
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7174081802368164,
      "learning_rate": 1.041636171556234e-05,
      "loss": 0.5489,
      "step": 14566
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5754345655441284,
      "learning_rate": 1.0415246107202278e-05,
      "loss": 0.5474,
      "step": 14567
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8695086240768433,
      "learning_rate": 1.0414130493665187e-05,
      "loss": 0.5663,
      "step": 14568
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5738393068313599,
      "learning_rate": 1.0413014874964973e-05,
      "loss": 0.4933,
      "step": 14569
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.400685429573059,
      "learning_rate": 1.041189925111555e-05,
      "loss": 0.4838,
      "step": 14570
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7008497714996338,
      "learning_rate": 1.0410783622130825e-05,
      "loss": 0.5236,
      "step": 14571
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.482379674911499,
      "learning_rate": 1.0409667988024703e-05,
      "loss": 0.5839,
      "step": 14572
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5366836786270142,
      "learning_rate": 1.0408552348811101e-05,
      "loss": 0.5082,
      "step": 14573
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.597274661064148,
      "learning_rate": 1.040743670450392e-05,
      "loss": 0.6004,
      "step": 14574
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6525850296020508,
      "learning_rate": 1.0406321055117072e-05,
      "loss": 0.4878,
      "step": 14575
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8347011804580688,
      "learning_rate": 1.040520540066447e-05,
      "loss": 0.5907,
      "step": 14576
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5831385850906372,
      "learning_rate": 1.040408974116002e-05,
      "loss": 0.5974,
      "step": 14577
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.943841576576233,
      "learning_rate": 1.040297407661763e-05,
      "loss": 0.6523,
      "step": 14578
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5674349069595337,
      "learning_rate": 1.0401858407051211e-05,
      "loss": 0.5415,
      "step": 14579
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6575965881347656,
      "learning_rate": 1.0400742732474674e-05,
      "loss": 0.5536,
      "step": 14580
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.874060034751892,
      "learning_rate": 1.0399627052901926e-05,
      "loss": 0.4397,
      "step": 14581
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4048479795455933,
      "learning_rate": 1.0398511368346875e-05,
      "loss": 0.7127,
      "step": 14582
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5598571300506592,
      "learning_rate": 1.0397395678823437e-05,
      "loss": 0.524,
      "step": 14583
    },
    {
      "epoch": 1.51,
      "grad_norm": 2.057971954345703,
      "learning_rate": 1.0396279984345516e-05,
      "loss": 0.6376,
      "step": 14584
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.715898871421814,
      "learning_rate": 1.0395164284927025e-05,
      "loss": 0.6282,
      "step": 14585
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5622309446334839,
      "learning_rate": 1.0394048580581873e-05,
      "loss": 0.5045,
      "step": 14586
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.9836457967758179,
      "learning_rate": 1.0392932871323966e-05,
      "loss": 0.5833,
      "step": 14587
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5903679132461548,
      "learning_rate": 1.0391817157167218e-05,
      "loss": 0.5495,
      "step": 14588
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5926886796951294,
      "learning_rate": 1.0390701438125542e-05,
      "loss": 0.5832,
      "step": 14589
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3590373992919922,
      "learning_rate": 1.0389585714212842e-05,
      "loss": 0.5903,
      "step": 14590
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3228927850723267,
      "learning_rate": 1.0388469985443033e-05,
      "loss": 0.5887,
      "step": 14591
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8372727632522583,
      "learning_rate": 1.038735425183002e-05,
      "loss": 0.626,
      "step": 14592
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.77985417842865,
      "learning_rate": 1.0386238513387716e-05,
      "loss": 0.5181,
      "step": 14593
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2507145404815674,
      "learning_rate": 1.0385122770130033e-05,
      "loss": 0.4999,
      "step": 14594
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8480969667434692,
      "learning_rate": 1.0384007022070878e-05,
      "loss": 0.5177,
      "step": 14595
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.37943434715271,
      "learning_rate": 1.0382891269224164e-05,
      "loss": 0.4876,
      "step": 14596
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5479803085327148,
      "learning_rate": 1.0381775511603805e-05,
      "loss": 0.6381,
      "step": 14597
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4622598886489868,
      "learning_rate": 1.0380659749223703e-05,
      "loss": 0.6673,
      "step": 14598
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.486026644706726,
      "learning_rate": 1.0379543982097776e-05,
      "loss": 0.3614,
      "step": 14599
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3506590127944946,
      "learning_rate": 1.0378428210239927e-05,
      "loss": 0.4889,
      "step": 14600
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3246761560440063,
      "learning_rate": 1.0377312433664075e-05,
      "loss": 0.5144,
      "step": 14601
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4543395042419434,
      "learning_rate": 1.0376196652384127e-05,
      "loss": 0.5118,
      "step": 14602
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4050098657608032,
      "learning_rate": 1.0375080866413993e-05,
      "loss": 0.5802,
      "step": 14603
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7117066383361816,
      "learning_rate": 1.0373965075767587e-05,
      "loss": 0.4966,
      "step": 14604
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6445491313934326,
      "learning_rate": 1.0372849280458817e-05,
      "loss": 0.6263,
      "step": 14605
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7693787813186646,
      "learning_rate": 1.0371733480501594e-05,
      "loss": 0.5966,
      "step": 14606
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3827526569366455,
      "learning_rate": 1.0370617675909834e-05,
      "loss": 0.5059,
      "step": 14607
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7410107851028442,
      "learning_rate": 1.0369501866697438e-05,
      "loss": 0.5435,
      "step": 14608
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.570697546005249,
      "learning_rate": 1.0368386052878328e-05,
      "loss": 0.5942,
      "step": 14609
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.497389554977417,
      "learning_rate": 1.0367270234466409e-05,
      "loss": 0.4652,
      "step": 14610
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7424713373184204,
      "learning_rate": 1.0366154411475595e-05,
      "loss": 0.6455,
      "step": 14611
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5935335159301758,
      "learning_rate": 1.0365038583919798e-05,
      "loss": 0.6016,
      "step": 14612
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.306512475013733,
      "learning_rate": 1.0363922751812925e-05,
      "loss": 0.4811,
      "step": 14613
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7463635206222534,
      "learning_rate": 1.036280691516889e-05,
      "loss": 0.5304,
      "step": 14614
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4198535680770874,
      "learning_rate": 1.036169107400161e-05,
      "loss": 0.5137,
      "step": 14615
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5013219118118286,
      "learning_rate": 1.0360575228324987e-05,
      "loss": 0.5797,
      "step": 14616
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4694491624832153,
      "learning_rate": 1.035945937815294e-05,
      "loss": 0.605,
      "step": 14617
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.8187181949615479,
      "learning_rate": 1.0358343523499377e-05,
      "loss": 0.4343,
      "step": 14618
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.7968683242797852,
      "learning_rate": 1.0357227664378209e-05,
      "loss": 0.5793,
      "step": 14619
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6808720827102661,
      "learning_rate": 1.0356111800803352e-05,
      "loss": 0.5293,
      "step": 14620
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6478983163833618,
      "learning_rate": 1.0354995932788714e-05,
      "loss": 0.6326,
      "step": 14621
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.458383560180664,
      "learning_rate": 1.0353880060348209e-05,
      "loss": 0.4877,
      "step": 14622
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3265966176986694,
      "learning_rate": 1.035276418349575e-05,
      "loss": 0.5418,
      "step": 14623
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2739624977111816,
      "learning_rate": 1.0351648302245246e-05,
      "loss": 0.605,
      "step": 14624
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2886788845062256,
      "learning_rate": 1.035053241661061e-05,
      "loss": 0.511,
      "step": 14625
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5024755001068115,
      "learning_rate": 1.0349416526605755e-05,
      "loss": 0.4543,
      "step": 14626
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5651967525482178,
      "learning_rate": 1.0348300632244593e-05,
      "loss": 0.501,
      "step": 14627
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4576239585876465,
      "learning_rate": 1.0347184733541038e-05,
      "loss": 0.4544,
      "step": 14628
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4817248582839966,
      "learning_rate": 1.0346068830509e-05,
      "loss": 0.5142,
      "step": 14629
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4883545637130737,
      "learning_rate": 1.0344952923162392e-05,
      "loss": 0.5152,
      "step": 14630
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5716594457626343,
      "learning_rate": 1.0343837011515125e-05,
      "loss": 0.6065,
      "step": 14631
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.687215805053711,
      "learning_rate": 1.0342721095581115e-05,
      "loss": 0.497,
      "step": 14632
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.622140884399414,
      "learning_rate": 1.0341605175374272e-05,
      "loss": 0.5813,
      "step": 14633
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.4230002164840698,
      "learning_rate": 1.0340489250908505e-05,
      "loss": 0.5685,
      "step": 14634
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5553795099258423,
      "learning_rate": 1.0339373322197734e-05,
      "loss": 0.5474,
      "step": 14635
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5451974868774414,
      "learning_rate": 1.0338257389255871e-05,
      "loss": 0.5411,
      "step": 14636
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5292174816131592,
      "learning_rate": 1.0337141452096822e-05,
      "loss": 0.4836,
      "step": 14637
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3709176778793335,
      "learning_rate": 1.0336025510734507e-05,
      "loss": 0.5148,
      "step": 14638
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6417467594146729,
      "learning_rate": 1.0334909565182834e-05,
      "loss": 0.4925,
      "step": 14639
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.800808310508728,
      "learning_rate": 1.0333793615455717e-05,
      "loss": 0.4586,
      "step": 14640
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5607364177703857,
      "learning_rate": 1.0332677661567073e-05,
      "loss": 0.6578,
      "step": 14641
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.438599705696106,
      "learning_rate": 1.0331561703530808e-05,
      "loss": 0.566,
      "step": 14642
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.488464117050171,
      "learning_rate": 1.0330445741360842e-05,
      "loss": 0.5721,
      "step": 14643
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.483892560005188,
      "learning_rate": 1.0329329775071082e-05,
      "loss": 0.5269,
      "step": 14644
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2931938171386719,
      "learning_rate": 1.0328213804675445e-05,
      "loss": 0.5742,
      "step": 14645
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.730608344078064,
      "learning_rate": 1.0327097830187844e-05,
      "loss": 0.5079,
      "step": 14646
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6251896619796753,
      "learning_rate": 1.032598185162219e-05,
      "loss": 0.541,
      "step": 14647
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5987533330917358,
      "learning_rate": 1.03248658689924e-05,
      "loss": 0.5666,
      "step": 14648
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5496861934661865,
      "learning_rate": 1.0323749882312386e-05,
      "loss": 0.4856,
      "step": 14649
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3503817319869995,
      "learning_rate": 1.0322633891596059e-05,
      "loss": 0.5025,
      "step": 14650
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7086261510849,
      "learning_rate": 1.0321517896857335e-05,
      "loss": 0.4718,
      "step": 14651
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.8152300119400024,
      "learning_rate": 1.0320401898110126e-05,
      "loss": 0.5116,
      "step": 14652
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5726884603500366,
      "learning_rate": 1.0319285895368346e-05,
      "loss": 0.6611,
      "step": 14653
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7489937543869019,
      "learning_rate": 1.031816988864591e-05,
      "loss": 0.6323,
      "step": 14654
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5024220943450928,
      "learning_rate": 1.0317053877956729e-05,
      "loss": 0.578,
      "step": 14655
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3406877517700195,
      "learning_rate": 1.031593786331472e-05,
      "loss": 0.5086,
      "step": 14656
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4954371452331543,
      "learning_rate": 1.0314821844733794e-05,
      "loss": 0.5588,
      "step": 14657
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.9990732669830322,
      "learning_rate": 1.0313705822227865e-05,
      "loss": 0.6088,
      "step": 14658
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6033473014831543,
      "learning_rate": 1.031258979581085e-05,
      "loss": 0.4864,
      "step": 14659
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3780862092971802,
      "learning_rate": 1.0311473765496659e-05,
      "loss": 0.4669,
      "step": 14660
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2920148372650146,
      "learning_rate": 1.0310357731299208e-05,
      "loss": 0.4669,
      "step": 14661
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.455907940864563,
      "learning_rate": 1.0309241693232413e-05,
      "loss": 0.4998,
      "step": 14662
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3919988870620728,
      "learning_rate": 1.0308125651310181e-05,
      "loss": 0.5825,
      "step": 14663
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4837229251861572,
      "learning_rate": 1.0307009605546437e-05,
      "loss": 0.5975,
      "step": 14664
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6813297271728516,
      "learning_rate": 1.0305893555955082e-05,
      "loss": 0.5152,
      "step": 14665
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4725772142410278,
      "learning_rate": 1.030477750255004e-05,
      "loss": 0.4721,
      "step": 14666
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3337717056274414,
      "learning_rate": 1.0303661445345225e-05,
      "loss": 0.5152,
      "step": 14667
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4246035814285278,
      "learning_rate": 1.0302545384354547e-05,
      "loss": 0.5232,
      "step": 14668
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.624308705329895,
      "learning_rate": 1.0301429319591923e-05,
      "loss": 0.5833,
      "step": 14669
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.220031976699829,
      "learning_rate": 1.0300313251071266e-05,
      "loss": 0.6432,
      "step": 14670
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1541858911514282,
      "learning_rate": 1.0299197178806488e-05,
      "loss": 0.4755,
      "step": 14671
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4979947805404663,
      "learning_rate": 1.0298081102811513e-05,
      "loss": 0.4586,
      "step": 14672
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3478058576583862,
      "learning_rate": 1.0296965023100243e-05,
      "loss": 0.4983,
      "step": 14673
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6789116859436035,
      "learning_rate": 1.0295848939686598e-05,
      "loss": 0.6142,
      "step": 14674
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5037689208984375,
      "learning_rate": 1.0294732852584499e-05,
      "loss": 0.5851,
      "step": 14675
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.365552306175232,
      "learning_rate": 1.0293616761807854e-05,
      "loss": 0.5448,
      "step": 14676
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3564568758010864,
      "learning_rate": 1.0292500667370574e-05,
      "loss": 0.5264,
      "step": 14677
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.503658413887024,
      "learning_rate": 1.0291384569286584e-05,
      "loss": 0.5649,
      "step": 14678
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6032865047454834,
      "learning_rate": 1.0290268467569788e-05,
      "loss": 0.4633,
      "step": 14679
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2992026805877686,
      "learning_rate": 1.028915236223411e-05,
      "loss": 0.6119,
      "step": 14680
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4480973482131958,
      "learning_rate": 1.0288036253293458e-05,
      "loss": 0.457,
      "step": 14681
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4719170331954956,
      "learning_rate": 1.028692014076175e-05,
      "loss": 0.5186,
      "step": 14682
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5366202592849731,
      "learning_rate": 1.0285804024652904e-05,
      "loss": 0.6087,
      "step": 14683
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.610205888748169,
      "learning_rate": 1.028468790498083e-05,
      "loss": 0.5292,
      "step": 14684
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3860572576522827,
      "learning_rate": 1.0283571781759443e-05,
      "loss": 0.5487,
      "step": 14685
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4964768886566162,
      "learning_rate": 1.0282455655002664e-05,
      "loss": 0.4989,
      "step": 14686
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4192289113998413,
      "learning_rate": 1.0281339524724401e-05,
      "loss": 0.5073,
      "step": 14687
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2611652612686157,
      "learning_rate": 1.0280223390938575e-05,
      "loss": 0.5722,
      "step": 14688
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5444761514663696,
      "learning_rate": 1.0279107253659097e-05,
      "loss": 0.4874,
      "step": 14689
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6193511486053467,
      "learning_rate": 1.0277991112899886e-05,
      "loss": 0.4787,
      "step": 14690
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4456478357315063,
      "learning_rate": 1.0276874968674855e-05,
      "loss": 0.5586,
      "step": 14691
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4411665201187134,
      "learning_rate": 1.0275758820997918e-05,
      "loss": 0.5221,
      "step": 14692
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4282275438308716,
      "learning_rate": 1.0274642669882994e-05,
      "loss": 0.5551,
      "step": 14693
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5421961545944214,
      "learning_rate": 1.0273526515343997e-05,
      "loss": 0.447,
      "step": 14694
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5449823141098022,
      "learning_rate": 1.027241035739484e-05,
      "loss": 0.5975,
      "step": 14695
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.358635663986206,
      "learning_rate": 1.0271294196049445e-05,
      "loss": 0.5116,
      "step": 14696
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4917696714401245,
      "learning_rate": 1.0270178031321718e-05,
      "loss": 0.5626,
      "step": 14697
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3394278287887573,
      "learning_rate": 1.0269061863225583e-05,
      "loss": 0.5194,
      "step": 14698
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.44684898853302,
      "learning_rate": 1.0267945691774954e-05,
      "loss": 0.4195,
      "step": 14699
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.489026665687561,
      "learning_rate": 1.0266829516983743e-05,
      "loss": 0.5658,
      "step": 14700
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3889193534851074,
      "learning_rate": 1.026571333886587e-05,
      "loss": 0.444,
      "step": 14701
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7039934396743774,
      "learning_rate": 1.0264597157435248e-05,
      "loss": 0.643,
      "step": 14702
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.615668535232544,
      "learning_rate": 1.0263480972705792e-05,
      "loss": 0.5495,
      "step": 14703
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5152347087860107,
      "learning_rate": 1.0262364784691426e-05,
      "loss": 0.6878,
      "step": 14704
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6257177591323853,
      "learning_rate": 1.0261248593406055e-05,
      "loss": 0.51,
      "step": 14705
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2400383949279785,
      "learning_rate": 1.0260132398863603e-05,
      "loss": 0.4663,
      "step": 14706
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4005601406097412,
      "learning_rate": 1.0259016201077978e-05,
      "loss": 0.4549,
      "step": 14707
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2702275514602661,
      "learning_rate": 1.0257900000063103e-05,
      "loss": 0.4217,
      "step": 14708
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4794195890426636,
      "learning_rate": 1.0256783795832892e-05,
      "loss": 0.6737,
      "step": 14709
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5769680738449097,
      "learning_rate": 1.0255667588401261e-05,
      "loss": 0.6925,
      "step": 14710
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4433735609054565,
      "learning_rate": 1.0254551377782125e-05,
      "loss": 0.5801,
      "step": 14711
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.605246663093567,
      "learning_rate": 1.0253435163989405e-05,
      "loss": 0.597,
      "step": 14712
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3115477561950684,
      "learning_rate": 1.025231894703701e-05,
      "loss": 0.5458,
      "step": 14713
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2879222631454468,
      "learning_rate": 1.0251202726938862e-05,
      "loss": 0.4451,
      "step": 14714
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5844446420669556,
      "learning_rate": 1.0250086503708876e-05,
      "loss": 0.578,
      "step": 14715
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4208818674087524,
      "learning_rate": 1.0248970277360965e-05,
      "loss": 0.5806,
      "step": 14716
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.528542399406433,
      "learning_rate": 1.0247854047909052e-05,
      "loss": 0.5025,
      "step": 14717
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5272231101989746,
      "learning_rate": 1.0246737815367045e-05,
      "loss": 0.5384,
      "step": 14718
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7889186143875122,
      "learning_rate": 1.024562157974887e-05,
      "loss": 0.5089,
      "step": 14719
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.8155410289764404,
      "learning_rate": 1.0244505341068433e-05,
      "loss": 0.557,
      "step": 14720
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.534915566444397,
      "learning_rate": 1.0243389099339657e-05,
      "loss": 0.6099,
      "step": 14721
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6501030921936035,
      "learning_rate": 1.0242272854576462e-05,
      "loss": 0.5815,
      "step": 14722
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.2803540229797363,
      "learning_rate": 1.0241156606792759e-05,
      "loss": 0.5899,
      "step": 14723
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5136364698410034,
      "learning_rate": 1.0240040356002464e-05,
      "loss": 0.5062,
      "step": 14724
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1457651853561401,
      "learning_rate": 1.02389241022195e-05,
      "loss": 0.5504,
      "step": 14725
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6855319738388062,
      "learning_rate": 1.0237807845457772e-05,
      "loss": 0.5844,
      "step": 14726
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.565740704536438,
      "learning_rate": 1.0236691585731211e-05,
      "loss": 0.5859,
      "step": 14727
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6611605882644653,
      "learning_rate": 1.0235575323053723e-05,
      "loss": 0.5127,
      "step": 14728
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.8015779256820679,
      "learning_rate": 1.0234459057439232e-05,
      "loss": 0.7003,
      "step": 14729
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4925848245620728,
      "learning_rate": 1.0233342788901653e-05,
      "loss": 0.5212,
      "step": 14730
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5229336023330688,
      "learning_rate": 1.0232226517454897e-05,
      "loss": 0.5825,
      "step": 14731
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4768909215927124,
      "learning_rate": 1.0231110243112891e-05,
      "loss": 0.5742,
      "step": 14732
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4658035039901733,
      "learning_rate": 1.0229993965889545e-05,
      "loss": 0.6443,
      "step": 14733
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5673980712890625,
      "learning_rate": 1.0228877685798776e-05,
      "loss": 0.6124,
      "step": 14734
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4173181056976318,
      "learning_rate": 1.0227761402854503e-05,
      "loss": 0.4603,
      "step": 14735
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4076478481292725,
      "learning_rate": 1.0226645117070646e-05,
      "loss": 0.4932,
      "step": 14736
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4076844453811646,
      "learning_rate": 1.0225528828461115e-05,
      "loss": 0.5153,
      "step": 14737
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.023233413696289,
      "learning_rate": 1.0224412537039837e-05,
      "loss": 0.6256,
      "step": 14738
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5631821155548096,
      "learning_rate": 1.0223296242820719e-05,
      "loss": 0.5425,
      "step": 14739
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7337924242019653,
      "learning_rate": 1.0222179945817688e-05,
      "loss": 0.5247,
      "step": 14740
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6327579021453857,
      "learning_rate": 1.0221063646044651e-05,
      "loss": 0.5676,
      "step": 14741
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.6766878366470337,
      "learning_rate": 1.021994734351553e-05,
      "loss": 0.4893,
      "step": 14742
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.364302158355713,
      "learning_rate": 1.0218831038244248e-05,
      "loss": 0.4829,
      "step": 14743
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.630418062210083,
      "learning_rate": 1.0217714730244714e-05,
      "loss": 0.5574,
      "step": 14744
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.164778709411621,
      "learning_rate": 1.0216598419530853e-05,
      "loss": 0.5819,
      "step": 14745
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5553525686264038,
      "learning_rate": 1.0215482106116572e-05,
      "loss": 0.5723,
      "step": 14746
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7491419315338135,
      "learning_rate": 1.0214365790015796e-05,
      "loss": 0.5861,
      "step": 14747
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3571323156356812,
      "learning_rate": 1.0213249471242443e-05,
      "loss": 0.5668,
      "step": 14748
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.464023470878601,
      "learning_rate": 1.0212133149810428e-05,
      "loss": 0.5317,
      "step": 14749
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4304094314575195,
      "learning_rate": 1.021101682573367e-05,
      "loss": 0.5566,
      "step": 14750
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7390780448913574,
      "learning_rate": 1.0209900499026087e-05,
      "loss": 0.5879,
      "step": 14751
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4675514698028564,
      "learning_rate": 1.0208784169701593e-05,
      "loss": 0.694,
      "step": 14752
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.578377366065979,
      "learning_rate": 1.0207667837774113e-05,
      "loss": 0.5059,
      "step": 14753
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2140241861343384,
      "learning_rate": 1.0206551503257555e-05,
      "loss": 0.5583,
      "step": 14754
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5885281562805176,
      "learning_rate": 1.0205435166165844e-05,
      "loss": 0.5006,
      "step": 14755
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.587722659111023,
      "learning_rate": 1.0204318826512897e-05,
      "loss": 0.5718,
      "step": 14756
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7206026315689087,
      "learning_rate": 1.0203202484312629e-05,
      "loss": 0.5082,
      "step": 14757
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3467293977737427,
      "learning_rate": 1.020208613957896e-05,
      "loss": 0.6295,
      "step": 14758
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6699479818344116,
      "learning_rate": 1.0200969792325806e-05,
      "loss": 0.6903,
      "step": 14759
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.745695948600769,
      "learning_rate": 1.0199853442567087e-05,
      "loss": 0.5038,
      "step": 14760
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6407136917114258,
      "learning_rate": 1.0198737090316722e-05,
      "loss": 0.525,
      "step": 14761
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4886995553970337,
      "learning_rate": 1.0197620735588623e-05,
      "loss": 0.5912,
      "step": 14762
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4605121612548828,
      "learning_rate": 1.0196504378396711e-05,
      "loss": 0.4902,
      "step": 14763
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.835286021232605,
      "learning_rate": 1.0195388018754913e-05,
      "loss": 0.5503,
      "step": 14764
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8604234457015991,
      "learning_rate": 1.0194271656677133e-05,
      "loss": 0.6047,
      "step": 14765
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4841361045837402,
      "learning_rate": 1.01931552921773e-05,
      "loss": 0.5279,
      "step": 14766
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3425596952438354,
      "learning_rate": 1.019203892526932e-05,
      "loss": 0.4991,
      "step": 14767
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4170727729797363,
      "learning_rate": 1.0190922555967125e-05,
      "loss": 0.4964,
      "step": 14768
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.738410472869873,
      "learning_rate": 1.0189806184284625e-05,
      "loss": 0.6612,
      "step": 14769
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.61298668384552,
      "learning_rate": 1.0188689810235738e-05,
      "loss": 0.5107,
      "step": 14770
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3760933876037598,
      "learning_rate": 1.018757343383439e-05,
      "loss": 0.5925,
      "step": 14771
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6385891437530518,
      "learning_rate": 1.0186457055094488e-05,
      "loss": 0.4621,
      "step": 14772
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5104597806930542,
      "learning_rate": 1.0185340674029958e-05,
      "loss": 0.4395,
      "step": 14773
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.677979826927185,
      "learning_rate": 1.0184224290654716e-05,
      "loss": 0.5655,
      "step": 14774
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4691691398620605,
      "learning_rate": 1.0183107904982681e-05,
      "loss": 0.5733,
      "step": 14775
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.431006908416748,
      "learning_rate": 1.018199151702777e-05,
      "loss": 0.4819,
      "step": 14776
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.715174913406372,
      "learning_rate": 1.0180875126803905e-05,
      "loss": 0.4777,
      "step": 14777
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.482430100440979,
      "learning_rate": 1.0179758734324999e-05,
      "loss": 0.5722,
      "step": 14778
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3529446125030518,
      "learning_rate": 1.0178642339604974e-05,
      "loss": 0.5372,
      "step": 14779
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4749200344085693,
      "learning_rate": 1.0177525942657748e-05,
      "loss": 0.4623,
      "step": 14780
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5816679000854492,
      "learning_rate": 1.0176409543497238e-05,
      "loss": 0.4715,
      "step": 14781
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.596707820892334,
      "learning_rate": 1.0175293142137368e-05,
      "loss": 0.514,
      "step": 14782
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6412302255630493,
      "learning_rate": 1.017417673859205e-05,
      "loss": 0.6406,
      "step": 14783
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.9124698638916016,
      "learning_rate": 1.0173060332875207e-05,
      "loss": 0.5643,
      "step": 14784
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5857824087142944,
      "learning_rate": 1.0171943925000754e-05,
      "loss": 0.5196,
      "step": 14785
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.413901448249817,
      "learning_rate": 1.0170827514982611e-05,
      "loss": 0.509,
      "step": 14786
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6827428340911865,
      "learning_rate": 1.0169711102834701e-05,
      "loss": 0.5026,
      "step": 14787
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.518827199935913,
      "learning_rate": 1.0168594688570935e-05,
      "loss": 0.5393,
      "step": 14788
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5243537425994873,
      "learning_rate": 1.0167478272205236e-05,
      "loss": 0.5885,
      "step": 14789
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5530483722686768,
      "learning_rate": 1.0166361853751526e-05,
      "loss": 0.4582,
      "step": 14790
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5130560398101807,
      "learning_rate": 1.0165245433223715e-05,
      "loss": 0.5991,
      "step": 14791
    },
    {
      "epoch": 1.53,
      "grad_norm": 2.0841012001037598,
      "learning_rate": 1.0164129010635733e-05,
      "loss": 0.5459,
      "step": 14792
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.64549720287323,
      "learning_rate": 1.0163012586001488e-05,
      "loss": 0.479,
      "step": 14793
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6718988418579102,
      "learning_rate": 1.0161896159334906e-05,
      "loss": 0.6067,
      "step": 14794
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6527552604675293,
      "learning_rate": 1.0160779730649904e-05,
      "loss": 0.4788,
      "step": 14795
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3255482912063599,
      "learning_rate": 1.01596632999604e-05,
      "loss": 0.5999,
      "step": 14796
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6904549598693848,
      "learning_rate": 1.0158546867280315e-05,
      "loss": 0.5839,
      "step": 14797
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.434031367301941,
      "learning_rate": 1.0157430432623564e-05,
      "loss": 0.5089,
      "step": 14798
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6131738424301147,
      "learning_rate": 1.0156313996004073e-05,
      "loss": 0.4916,
      "step": 14799
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6881145238876343,
      "learning_rate": 1.0155197557435755e-05,
      "loss": 0.5804,
      "step": 14800
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3613694906234741,
      "learning_rate": 1.0154081116932529e-05,
      "loss": 0.4969,
      "step": 14801
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4109305143356323,
      "learning_rate": 1.0152964674508317e-05,
      "loss": 0.5882,
      "step": 14802
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.9546632766723633,
      "learning_rate": 1.0151848230177037e-05,
      "loss": 0.6361,
      "step": 14803
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4749910831451416,
      "learning_rate": 1.0150731783952607e-05,
      "loss": 0.4994,
      "step": 14804
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3859469890594482,
      "learning_rate": 1.014961533584895e-05,
      "loss": 0.4906,
      "step": 14805
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6682243347167969,
      "learning_rate": 1.0148498885879981e-05,
      "loss": 0.5802,
      "step": 14806
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3853247165679932,
      "learning_rate": 1.0147382434059618e-05,
      "loss": 0.5367,
      "step": 14807
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6070244312286377,
      "learning_rate": 1.0146265980401789e-05,
      "loss": 0.6157,
      "step": 14808
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4925541877746582,
      "learning_rate": 1.0145149524920401e-05,
      "loss": 0.5471,
      "step": 14809
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.401266098022461,
      "learning_rate": 1.0144033067629384e-05,
      "loss": 0.5817,
      "step": 14810
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6329177618026733,
      "learning_rate": 1.014291660854265e-05,
      "loss": 0.6915,
      "step": 14811
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8966331481933594,
      "learning_rate": 1.0141800147674121e-05,
      "loss": 0.5954,
      "step": 14812
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3027498722076416,
      "learning_rate": 1.0140683685037718e-05,
      "loss": 0.539,
      "step": 14813
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.508969783782959,
      "learning_rate": 1.0139567220647358e-05,
      "loss": 0.5118,
      "step": 14814
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.481682300567627,
      "learning_rate": 1.0138450754516962e-05,
      "loss": 0.4291,
      "step": 14815
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6202037334442139,
      "learning_rate": 1.0137334286660446e-05,
      "loss": 0.4433,
      "step": 14816
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.425329327583313,
      "learning_rate": 1.0136217817091731e-05,
      "loss": 0.5002,
      "step": 14817
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3522006273269653,
      "learning_rate": 1.013510134582474e-05,
      "loss": 0.4619,
      "step": 14818
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4458874464035034,
      "learning_rate": 1.0133984872873391e-05,
      "loss": 0.4579,
      "step": 14819
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6983736753463745,
      "learning_rate": 1.0132868398251598e-05,
      "loss": 0.546,
      "step": 14820
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.8641467094421387,
      "learning_rate": 1.0131751921973289e-05,
      "loss": 0.6334,
      "step": 14821
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4848747253417969,
      "learning_rate": 1.0130635444052377e-05,
      "loss": 0.5299,
      "step": 14822
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6067888736724854,
      "learning_rate": 1.0129518964502788e-05,
      "loss": 0.5977,
      "step": 14823
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.917284369468689,
      "learning_rate": 1.0128402483338431e-05,
      "loss": 0.5038,
      "step": 14824
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2006149291992188,
      "learning_rate": 1.0127286000573235e-05,
      "loss": 0.5027,
      "step": 14825
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5468077659606934,
      "learning_rate": 1.0126169516221118e-05,
      "loss": 0.6317,
      "step": 14826
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6821303367614746,
      "learning_rate": 1.0125053030295997e-05,
      "loss": 0.507,
      "step": 14827
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.738659381866455,
      "learning_rate": 1.0123936542811794e-05,
      "loss": 0.4523,
      "step": 14828
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.6193761825561523,
      "learning_rate": 1.0122820053782425e-05,
      "loss": 0.5707,
      "step": 14829
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5645145177841187,
      "learning_rate": 1.0121703563221814e-05,
      "loss": 0.5333,
      "step": 14830
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4543780088424683,
      "learning_rate": 1.012058707114388e-05,
      "loss": 0.4533,
      "step": 14831
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4804812669754028,
      "learning_rate": 1.011947057756254e-05,
      "loss": 0.538,
      "step": 14832
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5836340188980103,
      "learning_rate": 1.0118354082491714e-05,
      "loss": 0.5872,
      "step": 14833
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5413657426834106,
      "learning_rate": 1.0117237585945327e-05,
      "loss": 0.5039,
      "step": 14834
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7280685901641846,
      "learning_rate": 1.0116121087937292e-05,
      "loss": 0.5315,
      "step": 14835
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4953252077102661,
      "learning_rate": 1.0115004588481536e-05,
      "loss": 0.4949,
      "step": 14836
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5186189413070679,
      "learning_rate": 1.011388808759197e-05,
      "loss": 0.5319,
      "step": 14837
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7251324653625488,
      "learning_rate": 1.0112771585282519e-05,
      "loss": 0.6499,
      "step": 14838
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.7128050327301025,
      "learning_rate": 1.0111655081567105e-05,
      "loss": 0.4695,
      "step": 14839
    },
    {
      "epoch": 1.53,
      "grad_norm": 2.1087920665740967,
      "learning_rate": 1.0110538576459643e-05,
      "loss": 0.672,
      "step": 14840
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5232728719711304,
      "learning_rate": 1.0109422069974058e-05,
      "loss": 0.5335,
      "step": 14841
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8048325777053833,
      "learning_rate": 1.0108305562124264e-05,
      "loss": 0.586,
      "step": 14842
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3804571628570557,
      "learning_rate": 1.0107189052924182e-05,
      "loss": 0.555,
      "step": 14843
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.51128351688385,
      "learning_rate": 1.0106072542387739e-05,
      "loss": 0.5526,
      "step": 14844
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6172807216644287,
      "learning_rate": 1.0104956030528848e-05,
      "loss": 0.6265,
      "step": 14845
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6999995708465576,
      "learning_rate": 1.0103839517361429e-05,
      "loss": 0.5921,
      "step": 14846
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.551133394241333,
      "learning_rate": 1.0102723002899406e-05,
      "loss": 0.6003,
      "step": 14847
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8523049354553223,
      "learning_rate": 1.0101606487156695e-05,
      "loss": 0.5511,
      "step": 14848
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8534752130508423,
      "learning_rate": 1.0100489970147218e-05,
      "loss": 0.6229,
      "step": 14849
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.679398536682129,
      "learning_rate": 1.0099373451884896e-05,
      "loss": 0.5397,
      "step": 14850
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8514602184295654,
      "learning_rate": 1.0098256932383645e-05,
      "loss": 0.5947,
      "step": 14851
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4144729375839233,
      "learning_rate": 1.0097140411657394e-05,
      "loss": 0.7065,
      "step": 14852
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5015629529953003,
      "learning_rate": 1.0096023889720053e-05,
      "loss": 0.5666,
      "step": 14853
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7712095975875854,
      "learning_rate": 1.0094907366585544e-05,
      "loss": 0.4652,
      "step": 14854
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.416089415550232,
      "learning_rate": 1.0093790842267795e-05,
      "loss": 0.541,
      "step": 14855
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3589890003204346,
      "learning_rate": 1.0092674316780715e-05,
      "loss": 0.4723,
      "step": 14856
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5991960763931274,
      "learning_rate": 1.0091557790138231e-05,
      "loss": 0.5494,
      "step": 14857
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.598056435585022,
      "learning_rate": 1.0090441262354262e-05,
      "loss": 0.5578,
      "step": 14858
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.474508285522461,
      "learning_rate": 1.0089324733442726e-05,
      "loss": 0.5074,
      "step": 14859
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6348499059677124,
      "learning_rate": 1.008820820341755e-05,
      "loss": 0.4756,
      "step": 14860
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2161006927490234,
      "learning_rate": 1.0087091672292646e-05,
      "loss": 0.408,
      "step": 14861
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2201918363571167,
      "learning_rate": 1.0085975140081936e-05,
      "loss": 0.506,
      "step": 14862
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3223612308502197,
      "learning_rate": 1.0084858606799344e-05,
      "loss": 0.4815,
      "step": 14863
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5011612176895142,
      "learning_rate": 1.0083742072458786e-05,
      "loss": 0.6006,
      "step": 14864
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5993175506591797,
      "learning_rate": 1.0082625537074187e-05,
      "loss": 0.6435,
      "step": 14865
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5999068021774292,
      "learning_rate": 1.0081509000659462e-05,
      "loss": 0.6537,
      "step": 14866
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2759437561035156,
      "learning_rate": 1.0080392463228535e-05,
      "loss": 0.5933,
      "step": 14867
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.203858494758606,
      "learning_rate": 1.0079275924795324e-05,
      "loss": 0.5645,
      "step": 14868
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3567713499069214,
      "learning_rate": 1.007815938537375e-05,
      "loss": 0.5363,
      "step": 14869
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4579213857650757,
      "learning_rate": 1.0077042844977734e-05,
      "loss": 0.5338,
      "step": 14870
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5763546228408813,
      "learning_rate": 1.0075926303621197e-05,
      "loss": 0.6518,
      "step": 14871
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.898892879486084,
      "learning_rate": 1.0074809761318056e-05,
      "loss": 0.5788,
      "step": 14872
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.194376826286316,
      "learning_rate": 1.0073693218082236e-05,
      "loss": 0.4466,
      "step": 14873
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2737329006195068,
      "learning_rate": 1.0072576673927655e-05,
      "loss": 0.5521,
      "step": 14874
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4367326498031616,
      "learning_rate": 1.0071460128868232e-05,
      "loss": 0.5674,
      "step": 14875
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4991838932037354,
      "learning_rate": 1.007034358291789e-05,
      "loss": 0.5144,
      "step": 14876
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.614194393157959,
      "learning_rate": 1.0069227036090545e-05,
      "loss": 0.488,
      "step": 14877
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6824511289596558,
      "learning_rate": 1.0068110488400124e-05,
      "loss": 0.4703,
      "step": 14878
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3743152618408203,
      "learning_rate": 1.0066993939860543e-05,
      "loss": 0.4641,
      "step": 14879
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3298090696334839,
      "learning_rate": 1.0065877390485723e-05,
      "loss": 0.5871,
      "step": 14880
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.375844120979309,
      "learning_rate": 1.0064760840289587e-05,
      "loss": 0.4608,
      "step": 14881
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4003502130508423,
      "learning_rate": 1.006364428928605e-05,
      "loss": 0.6005,
      "step": 14882
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7666966915130615,
      "learning_rate": 1.0062527737489037e-05,
      "loss": 0.4915,
      "step": 14883
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.489467740058899,
      "learning_rate": 1.0061411184912468e-05,
      "loss": 0.5717,
      "step": 14884
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5019170045852661,
      "learning_rate": 1.0060294631570262e-05,
      "loss": 0.4827,
      "step": 14885
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5717939138412476,
      "learning_rate": 1.0059178077476342e-05,
      "loss": 0.5329,
      "step": 14886
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.441367745399475,
      "learning_rate": 1.0058061522644622e-05,
      "loss": 0.4312,
      "step": 14887
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3329097032546997,
      "learning_rate": 1.0056944967089031e-05,
      "loss": 0.5147,
      "step": 14888
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6961932182312012,
      "learning_rate": 1.0055828410823484e-05,
      "loss": 0.5742,
      "step": 14889
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7598758935928345,
      "learning_rate": 1.0054711853861905e-05,
      "loss": 0.5172,
      "step": 14890
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7900334596633911,
      "learning_rate": 1.0053595296218212e-05,
      "loss": 0.4117,
      "step": 14891
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7119030952453613,
      "learning_rate": 1.0052478737906325e-05,
      "loss": 0.59,
      "step": 14892
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3577470779418945,
      "learning_rate": 1.0051362178940167e-05,
      "loss": 0.5642,
      "step": 14893
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8802350759506226,
      "learning_rate": 1.0050245619333661e-05,
      "loss": 0.5942,
      "step": 14894
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.538910150527954,
      "learning_rate": 1.004912905910072e-05,
      "loss": 0.5481,
      "step": 14895
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.652912974357605,
      "learning_rate": 1.0048012498255267e-05,
      "loss": 0.6035,
      "step": 14896
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4076038599014282,
      "learning_rate": 1.0046895936811228e-05,
      "loss": 0.5186,
      "step": 14897
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5616437196731567,
      "learning_rate": 1.0045779374782518e-05,
      "loss": 0.5677,
      "step": 14898
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.40513277053833,
      "learning_rate": 1.0044662812183059e-05,
      "loss": 0.5765,
      "step": 14899
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3946404457092285,
      "learning_rate": 1.0043546249026772e-05,
      "loss": 0.4559,
      "step": 14900
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5352675914764404,
      "learning_rate": 1.0042429685327578e-05,
      "loss": 0.4914,
      "step": 14901
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5743460655212402,
      "learning_rate": 1.0041313121099398e-05,
      "loss": 0.5575,
      "step": 14902
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5902979373931885,
      "learning_rate": 1.004019655635615e-05,
      "loss": 0.6549,
      "step": 14903
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5070263147354126,
      "learning_rate": 1.003907999111176e-05,
      "loss": 0.5191,
      "step": 14904
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3617994785308838,
      "learning_rate": 1.0037963425380143e-05,
      "loss": 0.4786,
      "step": 14905
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6683553457260132,
      "learning_rate": 1.0036846859175218e-05,
      "loss": 0.4732,
      "step": 14906
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.505199670791626,
      "learning_rate": 1.0035730292510914e-05,
      "loss": 0.6371,
      "step": 14907
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3752024173736572,
      "learning_rate": 1.0034613725401145e-05,
      "loss": 0.5741,
      "step": 14908
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4299120903015137,
      "learning_rate": 1.0033497157859833e-05,
      "loss": 0.5424,
      "step": 14909
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8472354412078857,
      "learning_rate": 1.0032380589900903e-05,
      "loss": 0.5174,
      "step": 14910
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6942036151885986,
      "learning_rate": 1.0031264021538269e-05,
      "loss": 0.5023,
      "step": 14911
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5155853033065796,
      "learning_rate": 1.0030147452785857e-05,
      "loss": 0.5629,
      "step": 14912
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.9765490293502808,
      "learning_rate": 1.0029030883657582e-05,
      "loss": 0.5084,
      "step": 14913
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5154470205307007,
      "learning_rate": 1.0027914314167369e-05,
      "loss": 0.5431,
      "step": 14914
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5472242832183838,
      "learning_rate": 1.0026797744329139e-05,
      "loss": 0.4508,
      "step": 14915
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7711409330368042,
      "learning_rate": 1.0025681174156812e-05,
      "loss": 0.4158,
      "step": 14916
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5706785917282104,
      "learning_rate": 1.0024564603664309e-05,
      "loss": 0.524,
      "step": 14917
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4553546905517578,
      "learning_rate": 1.0023448032865547e-05,
      "loss": 0.6142,
      "step": 14918
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.298425316810608,
      "learning_rate": 1.002233146177445e-05,
      "loss": 0.4324,
      "step": 14919
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.389518141746521,
      "learning_rate": 1.002121489040494e-05,
      "loss": 0.4713,
      "step": 14920
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.719871997833252,
      "learning_rate": 1.0020098318770935e-05,
      "loss": 0.5761,
      "step": 14921
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4083727598190308,
      "learning_rate": 1.0018981746886356e-05,
      "loss": 0.6492,
      "step": 14922
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.6312447786331177,
      "learning_rate": 1.0017865174765127e-05,
      "loss": 0.5674,
      "step": 14923
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7922601699829102,
      "learning_rate": 1.0016748602421162e-05,
      "loss": 0.638,
      "step": 14924
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4587924480438232,
      "learning_rate": 1.001563202986839e-05,
      "loss": 0.5419,
      "step": 14925
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8969556093215942,
      "learning_rate": 1.0014515457120725e-05,
      "loss": 0.4849,
      "step": 14926
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.7656066417694092,
      "learning_rate": 1.0013398884192089e-05,
      "loss": 0.4912,
      "step": 14927
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3864362239837646,
      "learning_rate": 1.0012282311096409e-05,
      "loss": 0.5196,
      "step": 14928
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2794992923736572,
      "learning_rate": 1.0011165737847598e-05,
      "loss": 0.6197,
      "step": 14929
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.638010025024414,
      "learning_rate": 1.0010049164459581e-05,
      "loss": 0.6003,
      "step": 14930
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4419426918029785,
      "learning_rate": 1.0008932590946274e-05,
      "loss": 0.5081,
      "step": 14931
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.380935788154602,
      "learning_rate": 1.0007816017321603e-05,
      "loss": 0.5072,
      "step": 14932
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3900885581970215,
      "learning_rate": 1.0006699443599488e-05,
      "loss": 0.578,
      "step": 14933
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4534274339675903,
      "learning_rate": 1.0005582869793846e-05,
      "loss": 0.6478,
      "step": 14934
    },
    {
      "epoch": 1.54,
      "grad_norm": 2.0073094367980957,
      "learning_rate": 1.00044662959186e-05,
      "loss": 0.5083,
      "step": 14935
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5272051095962524,
      "learning_rate": 1.0003349721987674e-05,
      "loss": 0.6142,
      "step": 14936
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4882371425628662,
      "learning_rate": 1.0002233148014983e-05,
      "loss": 0.7015,
      "step": 14937
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3101367950439453,
      "learning_rate": 1.0001116574014454e-05,
      "loss": 0.4162,
      "step": 14938
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5099470615386963,
      "learning_rate": 1e-05,
      "loss": 0.5147,
      "step": 14939
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5032291412353516,
      "learning_rate": 9.998883425985548e-06,
      "loss": 0.523,
      "step": 14940
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.525474190711975,
      "learning_rate": 9.997766851985018e-06,
      "loss": 0.4486,
      "step": 14941
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3577985763549805,
      "learning_rate": 9.99665027801233e-06,
      "loss": 0.5299,
      "step": 14942
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.282275676727295,
      "learning_rate": 9.995533704081402e-06,
      "loss": 0.5224,
      "step": 14943
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6946302652359009,
      "learning_rate": 9.994417130206157e-06,
      "loss": 0.5361,
      "step": 14944
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.362176537513733,
      "learning_rate": 9.993300556400517e-06,
      "loss": 0.5386,
      "step": 14945
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5520628690719604,
      "learning_rate": 9.9921839826784e-06,
      "loss": 0.4622,
      "step": 14946
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5513187646865845,
      "learning_rate": 9.991067409053728e-06,
      "loss": 0.5645,
      "step": 14947
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2377468347549438,
      "learning_rate": 9.989950835540424e-06,
      "loss": 0.5865,
      "step": 14948
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6105104684829712,
      "learning_rate": 9.988834262152406e-06,
      "loss": 0.5228,
      "step": 14949
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8152167797088623,
      "learning_rate": 9.987717688903594e-06,
      "loss": 0.6325,
      "step": 14950
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8264611959457397,
      "learning_rate": 9.986601115807913e-06,
      "loss": 0.4921,
      "step": 14951
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3216081857681274,
      "learning_rate": 9.985484542879278e-06,
      "loss": 0.4328,
      "step": 14952
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3320707082748413,
      "learning_rate": 9.984367970131614e-06,
      "loss": 0.4801,
      "step": 14953
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6470904350280762,
      "learning_rate": 9.98325139757884e-06,
      "loss": 0.5687,
      "step": 14954
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4972773790359497,
      "learning_rate": 9.982134825234876e-06,
      "loss": 0.4614,
      "step": 14955
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.189720630645752,
      "learning_rate": 9.981018253113647e-06,
      "loss": 0.4169,
      "step": 14956
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4849283695220947,
      "learning_rate": 9.979901681229069e-06,
      "loss": 0.613,
      "step": 14957
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7533819675445557,
      "learning_rate": 9.978785109595063e-06,
      "loss": 0.4818,
      "step": 14958
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8565274477005005,
      "learning_rate": 9.977668538225553e-06,
      "loss": 0.4863,
      "step": 14959
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9703078269958496,
      "learning_rate": 9.976551967134455e-06,
      "loss": 0.6085,
      "step": 14960
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.62348473072052,
      "learning_rate": 9.975435396335695e-06,
      "loss": 0.5343,
      "step": 14961
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5992976427078247,
      "learning_rate": 9.974318825843191e-06,
      "loss": 0.5121,
      "step": 14962
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6049067974090576,
      "learning_rate": 9.973202255670863e-06,
      "loss": 0.5804,
      "step": 14963
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3686758279800415,
      "learning_rate": 9.972085685832633e-06,
      "loss": 0.4385,
      "step": 14964
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.665121078491211,
      "learning_rate": 9.97096911634242e-06,
      "loss": 0.539,
      "step": 14965
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6369515657424927,
      "learning_rate": 9.969852547214146e-06,
      "loss": 0.4375,
      "step": 14966
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5605409145355225,
      "learning_rate": 9.968735978461734e-06,
      "loss": 0.5825,
      "step": 14967
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4207261800765991,
      "learning_rate": 9.9676194100991e-06,
      "loss": 0.5993,
      "step": 14968
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4405524730682373,
      "learning_rate": 9.966502842140169e-06,
      "loss": 0.5619,
      "step": 14969
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7923147678375244,
      "learning_rate": 9.965386274598857e-06,
      "loss": 0.5864,
      "step": 14970
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5253201723098755,
      "learning_rate": 9.964269707489088e-06,
      "loss": 0.5685,
      "step": 14971
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8995064496994019,
      "learning_rate": 9.963153140824784e-06,
      "loss": 0.4443,
      "step": 14972
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6133642196655273,
      "learning_rate": 9.962036574619863e-06,
      "loss": 0.5566,
      "step": 14973
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.074535846710205,
      "learning_rate": 9.960920008888245e-06,
      "loss": 0.5423,
      "step": 14974
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5178354978561401,
      "learning_rate": 9.959803443643852e-06,
      "loss": 0.5082,
      "step": 14975
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5939112901687622,
      "learning_rate": 9.958686878900604e-06,
      "loss": 0.5459,
      "step": 14976
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5261735916137695,
      "learning_rate": 9.957570314672424e-06,
      "loss": 0.7126,
      "step": 14977
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8864200115203857,
      "learning_rate": 9.956453750973231e-06,
      "loss": 0.6798,
      "step": 14978
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7348734140396118,
      "learning_rate": 9.955337187816943e-06,
      "loss": 0.4929,
      "step": 14979
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3058242797851562,
      "learning_rate": 9.954220625217485e-06,
      "loss": 0.4192,
      "step": 14980
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.481980800628662,
      "learning_rate": 9.953104063188774e-06,
      "loss": 0.5012,
      "step": 14981
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7299178838729858,
      "learning_rate": 9.951987501744735e-06,
      "loss": 0.5888,
      "step": 14982
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.683821439743042,
      "learning_rate": 9.950870940899285e-06,
      "loss": 0.5133,
      "step": 14983
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5523223876953125,
      "learning_rate": 9.949754380666342e-06,
      "loss": 0.5485,
      "step": 14984
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5419769287109375,
      "learning_rate": 9.948637821059835e-06,
      "loss": 0.6406,
      "step": 14985
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3087626695632935,
      "learning_rate": 9.947521262093676e-06,
      "loss": 0.601,
      "step": 14986
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5159531831741333,
      "learning_rate": 9.94640470378179e-06,
      "loss": 0.5995,
      "step": 14987
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3798285722732544,
      "learning_rate": 9.945288146138098e-06,
      "loss": 0.5113,
      "step": 14988
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4760019779205322,
      "learning_rate": 9.944171589176517e-06,
      "loss": 0.5701,
      "step": 14989
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4927929639816284,
      "learning_rate": 9.943055032910972e-06,
      "loss": 0.507,
      "step": 14990
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5854016542434692,
      "learning_rate": 9.94193847735538e-06,
      "loss": 0.5688,
      "step": 14991
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2311513423919678,
      "learning_rate": 9.940821922523663e-06,
      "loss": 0.5516,
      "step": 14992
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5562390089035034,
      "learning_rate": 9.939705368429741e-06,
      "loss": 0.4346,
      "step": 14993
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.57435941696167,
      "learning_rate": 9.938588815087534e-06,
      "loss": 0.5468,
      "step": 14994
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6693406105041504,
      "learning_rate": 9.937472262510966e-06,
      "loss": 0.4802,
      "step": 14995
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5165151357650757,
      "learning_rate": 9.936355710713954e-06,
      "loss": 0.5011,
      "step": 14996
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.0553457736968994,
      "learning_rate": 9.935239159710416e-06,
      "loss": 0.6739,
      "step": 14997
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0749475955963135,
      "learning_rate": 9.934122609514279e-06,
      "loss": 0.4885,
      "step": 14998
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.411821722984314,
      "learning_rate": 9.933006060139459e-06,
      "loss": 0.5902,
      "step": 14999
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.51690673828125,
      "learning_rate": 9.931889511599878e-06,
      "loss": 0.6147,
      "step": 15000
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.904423475265503,
      "learning_rate": 9.930772963909457e-06,
      "loss": 0.524,
      "step": 15001
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.408906102180481,
      "learning_rate": 9.929656417082114e-06,
      "loss": 0.6208,
      "step": 15002
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5844522714614868,
      "learning_rate": 9.928539871131771e-06,
      "loss": 0.5088,
      "step": 15003
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.600614070892334,
      "learning_rate": 9.927423326072349e-06,
      "loss": 0.4992,
      "step": 15004
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.8731300830841064,
      "learning_rate": 9.926306781917765e-06,
      "loss": 0.6919,
      "step": 15005
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.924698829650879,
      "learning_rate": 9.925190238681947e-06,
      "loss": 0.512,
      "step": 15006
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4789888858795166,
      "learning_rate": 9.924073696378805e-06,
      "loss": 0.5344,
      "step": 15007
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5189977884292603,
      "learning_rate": 9.92295715502227e-06,
      "loss": 0.5436,
      "step": 15008
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4279935359954834,
      "learning_rate": 9.921840614626253e-06,
      "loss": 0.5022,
      "step": 15009
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.505113959312439,
      "learning_rate": 9.920724075204678e-06,
      "loss": 0.5176,
      "step": 15010
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5131359100341797,
      "learning_rate": 9.91960753677147e-06,
      "loss": 0.4786,
      "step": 15011
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6858125925064087,
      "learning_rate": 9.918490999340541e-06,
      "loss": 0.5515,
      "step": 15012
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4207818508148193,
      "learning_rate": 9.917374462925815e-06,
      "loss": 0.4209,
      "step": 15013
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.926186203956604,
      "learning_rate": 9.916257927541216e-06,
      "loss": 0.5419,
      "step": 15014
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.373183250427246,
      "learning_rate": 9.915141393200658e-06,
      "loss": 0.5599,
      "step": 15015
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7310316562652588,
      "learning_rate": 9.914024859918066e-06,
      "loss": 0.5378,
      "step": 15016
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3806982040405273,
      "learning_rate": 9.912908327707359e-06,
      "loss": 0.5415,
      "step": 15017
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.450266718864441,
      "learning_rate": 9.911791796582453e-06,
      "loss": 0.5772,
      "step": 15018
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9462028741836548,
      "learning_rate": 9.910675266557275e-06,
      "loss": 0.4731,
      "step": 15019
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.9864529371261597,
      "learning_rate": 9.90955873764574e-06,
      "loss": 0.6411,
      "step": 15020
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.4244043827056885,
      "learning_rate": 9.908442209861772e-06,
      "loss": 0.513,
      "step": 15021
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2658203840255737,
      "learning_rate": 9.907325683219288e-06,
      "loss": 0.4948,
      "step": 15022
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3235894441604614,
      "learning_rate": 9.90620915773221e-06,
      "loss": 0.5163,
      "step": 15023
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6110167503356934,
      "learning_rate": 9.905092633414458e-06,
      "loss": 0.4738,
      "step": 15024
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5032927989959717,
      "learning_rate": 9.90397611027995e-06,
      "loss": 0.5193,
      "step": 15025
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4594924449920654,
      "learning_rate": 9.902859588342608e-06,
      "loss": 0.503,
      "step": 15026
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4074597358703613,
      "learning_rate": 9.901743067616355e-06,
      "loss": 0.5057,
      "step": 15027
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3617757558822632,
      "learning_rate": 9.900626548115106e-06,
      "loss": 0.531,
      "step": 15028
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5926586389541626,
      "learning_rate": 9.899510029852783e-06,
      "loss": 0.5422,
      "step": 15029
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.315588355064392,
      "learning_rate": 9.898393512843307e-06,
      "loss": 0.506,
      "step": 15030
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.7146347761154175,
      "learning_rate": 9.897276997100596e-06,
      "loss": 0.5503,
      "step": 15031
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6834266185760498,
      "learning_rate": 9.896160482638573e-06,
      "loss": 0.6455,
      "step": 15032
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5153928995132446,
      "learning_rate": 9.895043969471153e-06,
      "loss": 0.6068,
      "step": 15033
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4724441766738892,
      "learning_rate": 9.893927457612263e-06,
      "loss": 0.4456,
      "step": 15034
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.674514889717102,
      "learning_rate": 9.892810947075817e-06,
      "loss": 0.6094,
      "step": 15035
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5563946962356567,
      "learning_rate": 9.891694437875738e-06,
      "loss": 0.636,
      "step": 15036
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3817484378814697,
      "learning_rate": 9.890577930025945e-06,
      "loss": 0.527,
      "step": 15037
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.544144630432129,
      "learning_rate": 9.889461423540357e-06,
      "loss": 0.6335,
      "step": 15038
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6537153720855713,
      "learning_rate": 9.888344918432895e-06,
      "loss": 0.5752,
      "step": 15039
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5386276245117188,
      "learning_rate": 9.887228414717481e-06,
      "loss": 0.6288,
      "step": 15040
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4021828174591064,
      "learning_rate": 9.886111912408031e-06,
      "loss": 0.4568,
      "step": 15041
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4167776107788086,
      "learning_rate": 9.884995411518468e-06,
      "loss": 0.5373,
      "step": 15042
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6003817319869995,
      "learning_rate": 9.883878912062708e-06,
      "loss": 0.529,
      "step": 15043
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6271673440933228,
      "learning_rate": 9.882762414054672e-06,
      "loss": 0.5378,
      "step": 15044
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3695693016052246,
      "learning_rate": 9.881645917508286e-06,
      "loss": 0.5307,
      "step": 15045
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4676438570022583,
      "learning_rate": 9.880529422437461e-06,
      "loss": 0.595,
      "step": 15046
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6945617198944092,
      "learning_rate": 9.879412928856123e-06,
      "loss": 0.4437,
      "step": 15047
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5355201959609985,
      "learning_rate": 9.878296436778188e-06,
      "loss": 0.4764,
      "step": 15048
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4650609493255615,
      "learning_rate": 9.877179946217575e-06,
      "loss": 0.5432,
      "step": 15049
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8880853652954102,
      "learning_rate": 9.876063457188209e-06,
      "loss": 0.5005,
      "step": 15050
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.774837851524353,
      "learning_rate": 9.874946969704004e-06,
      "loss": 0.5028,
      "step": 15051
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.726534366607666,
      "learning_rate": 9.873830483778882e-06,
      "loss": 0.6484,
      "step": 15052
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6301310062408447,
      "learning_rate": 9.872713999426767e-06,
      "loss": 0.5825,
      "step": 15053
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.665102481842041,
      "learning_rate": 9.87159751666157e-06,
      "loss": 0.6797,
      "step": 15054
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2197434902191162,
      "learning_rate": 9.870481035497217e-06,
      "loss": 0.5083,
      "step": 15055
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9700467586517334,
      "learning_rate": 9.869364555947624e-06,
      "loss": 0.4684,
      "step": 15056
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5394526720046997,
      "learning_rate": 9.868248078026711e-06,
      "loss": 0.5381,
      "step": 15057
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.596616268157959,
      "learning_rate": 9.867131601748402e-06,
      "loss": 0.5645,
      "step": 15058
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5395931005477905,
      "learning_rate": 9.86601512712661e-06,
      "loss": 0.6084,
      "step": 15059
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6508915424346924,
      "learning_rate": 9.864898654175262e-06,
      "loss": 0.6048,
      "step": 15060
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.505240797996521,
      "learning_rate": 9.863782182908269e-06,
      "loss": 0.557,
      "step": 15061
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5113534927368164,
      "learning_rate": 9.862665713339556e-06,
      "loss": 0.5642,
      "step": 15062
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3491684198379517,
      "learning_rate": 9.861549245483043e-06,
      "loss": 0.5156,
      "step": 15063
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3579990863800049,
      "learning_rate": 9.860432779352644e-06,
      "loss": 0.5057,
      "step": 15064
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.621111273765564,
      "learning_rate": 9.859316314962282e-06,
      "loss": 0.5747,
      "step": 15065
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4207488298416138,
      "learning_rate": 9.85819985232588e-06,
      "loss": 0.4916,
      "step": 15066
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3172334432601929,
      "learning_rate": 9.85708339145735e-06,
      "loss": 0.4781,
      "step": 15067
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.632843255996704,
      "learning_rate": 9.85596693237062e-06,
      "loss": 0.5976,
      "step": 15068
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4146449565887451,
      "learning_rate": 9.854850475079599e-06,
      "loss": 0.5381,
      "step": 15069
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.535841464996338,
      "learning_rate": 9.853734019598213e-06,
      "loss": 0.5428,
      "step": 15070
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6634957790374756,
      "learning_rate": 9.85261756594038e-06,
      "loss": 0.4094,
      "step": 15071
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9711743593215942,
      "learning_rate": 9.85150111412002e-06,
      "loss": 0.5525,
      "step": 15072
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6901119947433472,
      "learning_rate": 9.850384664151052e-06,
      "loss": 0.6677,
      "step": 15073
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4611209630966187,
      "learning_rate": 9.849268216047393e-06,
      "loss": 0.5477,
      "step": 15074
    },
    {
      "epoch": 1.56,
      "grad_norm": 2.052076816558838,
      "learning_rate": 9.848151769822963e-06,
      "loss": 0.5897,
      "step": 15075
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7827337980270386,
      "learning_rate": 9.847035325491685e-06,
      "loss": 0.6134,
      "step": 15076
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2872587442398071,
      "learning_rate": 9.845918883067473e-06,
      "loss": 0.4377,
      "step": 15077
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7174556255340576,
      "learning_rate": 9.844802442564247e-06,
      "loss": 0.5432,
      "step": 15078
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4625251293182373,
      "learning_rate": 9.843686003995929e-06,
      "loss": 0.5402,
      "step": 15079
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4433876276016235,
      "learning_rate": 9.842569567376436e-06,
      "loss": 0.5704,
      "step": 15080
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5245726108551025,
      "learning_rate": 9.841453132719687e-06,
      "loss": 0.5749,
      "step": 15081
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8858152627944946,
      "learning_rate": 9.840336700039602e-06,
      "loss": 0.6162,
      "step": 15082
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7773866653442383,
      "learning_rate": 9.839220269350096e-06,
      "loss": 0.5544,
      "step": 15083
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6368544101715088,
      "learning_rate": 9.838103840665096e-06,
      "loss": 0.5433,
      "step": 15084
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.665073275566101,
      "learning_rate": 9.836987413998514e-06,
      "loss": 0.5715,
      "step": 15085
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3152800798416138,
      "learning_rate": 9.83587098936427e-06,
      "loss": 0.5365,
      "step": 15086
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6062796115875244,
      "learning_rate": 9.834754566776286e-06,
      "loss": 0.6398,
      "step": 15087
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9130115509033203,
      "learning_rate": 9.833638146248476e-06,
      "loss": 0.6637,
      "step": 15088
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6586518287658691,
      "learning_rate": 9.832521727794764e-06,
      "loss": 0.5688,
      "step": 15089
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5377578735351562,
      "learning_rate": 9.831405311429067e-06,
      "loss": 0.5508,
      "step": 15090
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3505042791366577,
      "learning_rate": 9.8302888971653e-06,
      "loss": 0.4975,
      "step": 15091
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.523370385169983,
      "learning_rate": 9.82917248501739e-06,
      "loss": 0.4836,
      "step": 15092
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8103774785995483,
      "learning_rate": 9.828056074999248e-06,
      "loss": 0.5865,
      "step": 15093
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3149722814559937,
      "learning_rate": 9.826939667124797e-06,
      "loss": 0.5886,
      "step": 15094
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5668137073516846,
      "learning_rate": 9.82582326140795e-06,
      "loss": 0.541,
      "step": 15095
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8996284008026123,
      "learning_rate": 9.824706857862632e-06,
      "loss": 0.5864,
      "step": 15096
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7754302024841309,
      "learning_rate": 9.823590456502762e-06,
      "loss": 0.5801,
      "step": 15097
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3488572835922241,
      "learning_rate": 9.822474057342254e-06,
      "loss": 0.5538,
      "step": 15098
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3677655458450317,
      "learning_rate": 9.821357660395027e-06,
      "loss": 0.592,
      "step": 15099
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3270494937896729,
      "learning_rate": 9.820241265675003e-06,
      "loss": 0.4867,
      "step": 15100
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3252543210983276,
      "learning_rate": 9.819124873196097e-06,
      "loss": 0.5641,
      "step": 15101
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4799894094467163,
      "learning_rate": 9.81800848297223e-06,
      "loss": 0.5832,
      "step": 15102
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7499606609344482,
      "learning_rate": 9.81689209501732e-06,
      "loss": 0.3998,
      "step": 15103
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3261492252349854,
      "learning_rate": 9.815775709345284e-06,
      "loss": 0.5604,
      "step": 15104
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7587858438491821,
      "learning_rate": 9.814659325970044e-06,
      "loss": 0.5528,
      "step": 15105
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.397518515586853,
      "learning_rate": 9.813542944905512e-06,
      "loss": 0.5762,
      "step": 15106
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5841448307037354,
      "learning_rate": 9.812426566165613e-06,
      "loss": 0.6426,
      "step": 15107
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.765877366065979,
      "learning_rate": 9.811310189764262e-06,
      "loss": 0.5455,
      "step": 15108
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3850606679916382,
      "learning_rate": 9.810193815715375e-06,
      "loss": 0.5205,
      "step": 15109
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4148085117340088,
      "learning_rate": 9.809077444032876e-06,
      "loss": 0.5403,
      "step": 15110
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5652990341186523,
      "learning_rate": 9.80796107473068e-06,
      "loss": 0.545,
      "step": 15111
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3928452730178833,
      "learning_rate": 9.806844707822704e-06,
      "loss": 0.6105,
      "step": 15112
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.380752682685852,
      "learning_rate": 9.805728343322869e-06,
      "loss": 0.4436,
      "step": 15113
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2761543989181519,
      "learning_rate": 9.804611981245088e-06,
      "loss": 0.4347,
      "step": 15114
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6270509958267212,
      "learning_rate": 9.80349562160329e-06,
      "loss": 0.518,
      "step": 15115
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.897689700126648,
      "learning_rate": 9.802379264411382e-06,
      "loss": 0.4734,
      "step": 15116
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5213494300842285,
      "learning_rate": 9.801262909683285e-06,
      "loss": 0.5102,
      "step": 15117
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6453211307525635,
      "learning_rate": 9.800146557432917e-06,
      "loss": 0.4856,
      "step": 15118
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7226684093475342,
      "learning_rate": 9.799030207674199e-06,
      "loss": 0.5331,
      "step": 15119
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6206204891204834,
      "learning_rate": 9.797913860421045e-06,
      "loss": 0.5396,
      "step": 15120
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4361134767532349,
      "learning_rate": 9.796797515687378e-06,
      "loss": 0.4697,
      "step": 15121
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.619386911392212,
      "learning_rate": 9.795681173487108e-06,
      "loss": 0.4647,
      "step": 15122
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.6252230405807495,
      "learning_rate": 9.794564833834159e-06,
      "loss": 0.4659,
      "step": 15123
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7023431062698364,
      "learning_rate": 9.79344849674245e-06,
      "loss": 0.5617,
      "step": 15124
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.9057791233062744,
      "learning_rate": 9.792332162225893e-06,
      "loss": 0.5357,
      "step": 15125
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4321584701538086,
      "learning_rate": 9.79121583029841e-06,
      "loss": 0.5402,
      "step": 15126
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7239656448364258,
      "learning_rate": 9.790099500973917e-06,
      "loss": 0.5548,
      "step": 15127
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3878483772277832,
      "learning_rate": 9.788983174266332e-06,
      "loss": 0.5723,
      "step": 15128
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.8394173383712769,
      "learning_rate": 9.787866850189575e-06,
      "loss": 0.5601,
      "step": 15129
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5014533996582031,
      "learning_rate": 9.78675052875756e-06,
      "loss": 0.4938,
      "step": 15130
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.5664602518081665,
      "learning_rate": 9.785634209984206e-06,
      "loss": 0.5544,
      "step": 15131
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9809916019439697,
      "learning_rate": 9.784517893883433e-06,
      "loss": 0.535,
      "step": 15132
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4109708070755005,
      "learning_rate": 9.783401580469154e-06,
      "loss": 0.4241,
      "step": 15133
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5668660402297974,
      "learning_rate": 9.782285269755291e-06,
      "loss": 0.5544,
      "step": 15134
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.469221591949463,
      "learning_rate": 9.781168961755757e-06,
      "loss": 0.4336,
      "step": 15135
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3492759466171265,
      "learning_rate": 9.780052656484471e-06,
      "loss": 0.5983,
      "step": 15136
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6346192359924316,
      "learning_rate": 9.778936353955354e-06,
      "loss": 0.5329,
      "step": 15137
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8195096254348755,
      "learning_rate": 9.777820054182319e-06,
      "loss": 0.5963,
      "step": 15138
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5495902299880981,
      "learning_rate": 9.776703757179286e-06,
      "loss": 0.5128,
      "step": 15139
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4296246767044067,
      "learning_rate": 9.775587462960168e-06,
      "loss": 0.474,
      "step": 15140
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.644376516342163,
      "learning_rate": 9.774471171538886e-06,
      "loss": 0.4909,
      "step": 15141
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3745054006576538,
      "learning_rate": 9.77335488292936e-06,
      "loss": 0.4898,
      "step": 15142
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5419386625289917,
      "learning_rate": 9.7722385971455e-06,
      "loss": 0.4668,
      "step": 15143
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.510841727256775,
      "learning_rate": 9.771122314201228e-06,
      "loss": 0.5556,
      "step": 15144
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.511744737625122,
      "learning_rate": 9.770006034110462e-06,
      "loss": 0.4887,
      "step": 15145
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8255780935287476,
      "learning_rate": 9.768889756887114e-06,
      "loss": 0.5937,
      "step": 15146
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6141961812973022,
      "learning_rate": 9.767773482545108e-06,
      "loss": 0.5939,
      "step": 15147
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.595832347869873,
      "learning_rate": 9.766657211098354e-06,
      "loss": 0.5468,
      "step": 15148
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5520108938217163,
      "learning_rate": 9.765540942560772e-06,
      "loss": 0.54,
      "step": 15149
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8151427507400513,
      "learning_rate": 9.76442467694628e-06,
      "loss": 0.5903,
      "step": 15150
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2840328216552734,
      "learning_rate": 9.763308414268794e-06,
      "loss": 0.5399,
      "step": 15151
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2948375940322876,
      "learning_rate": 9.762192154542231e-06,
      "loss": 0.5603,
      "step": 15152
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4460300207138062,
      "learning_rate": 9.761075897780508e-06,
      "loss": 0.559,
      "step": 15153
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4931038618087769,
      "learning_rate": 9.75995964399754e-06,
      "loss": 0.4866,
      "step": 15154
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5976423025131226,
      "learning_rate": 9.758843393207248e-06,
      "loss": 0.5868,
      "step": 15155
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7178866863250732,
      "learning_rate": 9.757727145423543e-06,
      "loss": 0.5052,
      "step": 15156
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6513499021530151,
      "learning_rate": 9.756610900660344e-06,
      "loss": 0.5232,
      "step": 15157
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5300806760787964,
      "learning_rate": 9.755494658931572e-06,
      "loss": 0.4645,
      "step": 15158
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6835310459136963,
      "learning_rate": 9.754378420251135e-06,
      "loss": 0.6346,
      "step": 15159
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.386888861656189,
      "learning_rate": 9.75326218463296e-06,
      "loss": 0.5432,
      "step": 15160
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7441596984863281,
      "learning_rate": 9.752145952090954e-06,
      "loss": 0.5103,
      "step": 15161
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.724626898765564,
      "learning_rate": 9.751029722639038e-06,
      "loss": 0.5651,
      "step": 15162
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.464409589767456,
      "learning_rate": 9.74991349629113e-06,
      "loss": 0.6216,
      "step": 15163
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3305859565734863,
      "learning_rate": 9.748797273061141e-06,
      "loss": 0.5551,
      "step": 15164
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5029067993164062,
      "learning_rate": 9.747681052962994e-06,
      "loss": 0.4719,
      "step": 15165
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7000709772109985,
      "learning_rate": 9.7465648360106e-06,
      "loss": 0.5594,
      "step": 15166
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4905213117599487,
      "learning_rate": 9.745448622217877e-06,
      "loss": 0.4605,
      "step": 15167
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3120301961898804,
      "learning_rate": 9.744332411598744e-06,
      "loss": 0.4707,
      "step": 15168
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7591341733932495,
      "learning_rate": 9.743216204167111e-06,
      "loss": 0.4888,
      "step": 15169
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.733765959739685,
      "learning_rate": 9.7420999999369e-06,
      "loss": 0.5586,
      "step": 15170
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5346828699111938,
      "learning_rate": 9.740983798922027e-06,
      "loss": 0.4862,
      "step": 15171
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4744497537612915,
      "learning_rate": 9.739867601136404e-06,
      "loss": 0.4883,
      "step": 15172
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.402785062789917,
      "learning_rate": 9.73875140659395e-06,
      "loss": 0.5479,
      "step": 15173
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.567764163017273,
      "learning_rate": 9.73763521530858e-06,
      "loss": 0.5291,
      "step": 15174
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6443432569503784,
      "learning_rate": 9.73651902729421e-06,
      "loss": 0.537,
      "step": 15175
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.9230796098709106,
      "learning_rate": 9.735402842564757e-06,
      "loss": 0.5134,
      "step": 15176
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3637359142303467,
      "learning_rate": 9.734286661134134e-06,
      "loss": 0.5227,
      "step": 15177
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4015034437179565,
      "learning_rate": 9.733170483016262e-06,
      "loss": 0.4142,
      "step": 15178
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5507214069366455,
      "learning_rate": 9.732054308225053e-06,
      "loss": 0.4948,
      "step": 15179
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.245376706123352,
      "learning_rate": 9.73093813677442e-06,
      "loss": 0.5553,
      "step": 15180
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4859774112701416,
      "learning_rate": 9.729821968678287e-06,
      "loss": 0.6371,
      "step": 15181
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8081016540527344,
      "learning_rate": 9.72870580395056e-06,
      "loss": 0.6981,
      "step": 15182
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4053200483322144,
      "learning_rate": 9.727589642605164e-06,
      "loss": 0.5663,
      "step": 15183
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3167510032653809,
      "learning_rate": 9.72647348465601e-06,
      "loss": 0.5511,
      "step": 15184
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3971521854400635,
      "learning_rate": 9.725357330117011e-06,
      "loss": 0.4292,
      "step": 15185
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.405882716178894,
      "learning_rate": 9.724241179002087e-06,
      "loss": 0.4861,
      "step": 15186
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.713363528251648,
      "learning_rate": 9.72312503132515e-06,
      "loss": 0.7131,
      "step": 15187
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.467092752456665,
      "learning_rate": 9.722008887100118e-06,
      "loss": 0.4468,
      "step": 15188
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7097209692001343,
      "learning_rate": 9.720892746340908e-06,
      "loss": 0.5885,
      "step": 15189
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3472713232040405,
      "learning_rate": 9.71977660906143e-06,
      "loss": 0.5333,
      "step": 15190
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4327752590179443,
      "learning_rate": 9.718660475275604e-06,
      "loss": 0.4999,
      "step": 15191
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6894279718399048,
      "learning_rate": 9.717544344997341e-06,
      "loss": 0.5845,
      "step": 15192
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6311285495758057,
      "learning_rate": 9.71642821824056e-06,
      "loss": 0.5353,
      "step": 15193
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.541811466217041,
      "learning_rate": 9.715312095019176e-06,
      "loss": 0.5992,
      "step": 15194
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8577580451965332,
      "learning_rate": 9.714195975347101e-06,
      "loss": 0.6305,
      "step": 15195
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6149929761886597,
      "learning_rate": 9.713079859238253e-06,
      "loss": 0.4879,
      "step": 15196
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6441704034805298,
      "learning_rate": 9.711963746706547e-06,
      "loss": 0.4914,
      "step": 15197
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.689734697341919,
      "learning_rate": 9.710847637765894e-06,
      "loss": 0.5547,
      "step": 15198
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5940808057785034,
      "learning_rate": 9.709731532430217e-06,
      "loss": 0.5157,
      "step": 15199
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4804341793060303,
      "learning_rate": 9.708615430713421e-06,
      "loss": 0.4561,
      "step": 15200
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.682965874671936,
      "learning_rate": 9.707499332629427e-06,
      "loss": 0.6328,
      "step": 15201
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7725828886032104,
      "learning_rate": 9.706383238192152e-06,
      "loss": 0.4916,
      "step": 15202
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4657564163208008,
      "learning_rate": 9.705267147415503e-06,
      "loss": 0.4596,
      "step": 15203
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.525791049003601,
      "learning_rate": 9.704151060313403e-06,
      "loss": 0.5772,
      "step": 15204
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7062373161315918,
      "learning_rate": 9.70303497689976e-06,
      "loss": 0.4548,
      "step": 15205
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5371073484420776,
      "learning_rate": 9.70191889718849e-06,
      "loss": 0.5539,
      "step": 15206
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.498227596282959,
      "learning_rate": 9.700802821193513e-06,
      "loss": 0.5066,
      "step": 15207
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2871372699737549,
      "learning_rate": 9.699686748928737e-06,
      "loss": 0.5105,
      "step": 15208
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4838588237762451,
      "learning_rate": 9.698570680408079e-06,
      "loss": 0.6002,
      "step": 15209
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.8820644617080688,
      "learning_rate": 9.697454615645456e-06,
      "loss": 0.5547,
      "step": 15210
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4788239002227783,
      "learning_rate": 9.696338554654776e-06,
      "loss": 0.5208,
      "step": 15211
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5401380062103271,
      "learning_rate": 9.695222497449961e-06,
      "loss": 0.4666,
      "step": 15212
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3970763683319092,
      "learning_rate": 9.69410644404492e-06,
      "loss": 0.4537,
      "step": 15213
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7589818239212036,
      "learning_rate": 9.692990394453568e-06,
      "loss": 0.6651,
      "step": 15214
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6196924448013306,
      "learning_rate": 9.691874348689822e-06,
      "loss": 0.4558,
      "step": 15215
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.733913779258728,
      "learning_rate": 9.690758306767592e-06,
      "loss": 0.5802,
      "step": 15216
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6550806760787964,
      "learning_rate": 9.689642268700795e-06,
      "loss": 0.5312,
      "step": 15217
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.650492787361145,
      "learning_rate": 9.688526234503343e-06,
      "loss": 0.5019,
      "step": 15218
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4015413522720337,
      "learning_rate": 9.687410204189153e-06,
      "loss": 0.5136,
      "step": 15219
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4787757396697998,
      "learning_rate": 9.686294177772137e-06,
      "loss": 0.6002,
      "step": 15220
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5962024927139282,
      "learning_rate": 9.68517815526621e-06,
      "loss": 0.6246,
      "step": 15221
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.623540997505188,
      "learning_rate": 9.684062136685282e-06,
      "loss": 0.5202,
      "step": 15222
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.6194475889205933,
      "learning_rate": 9.682946122043274e-06,
      "loss": 0.5851,
      "step": 15223
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4984265565872192,
      "learning_rate": 9.681830111354094e-06,
      "loss": 0.5605,
      "step": 15224
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.7672228813171387,
      "learning_rate": 9.680714104631658e-06,
      "loss": 0.594,
      "step": 15225
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4657871723175049,
      "learning_rate": 9.679598101889877e-06,
      "loss": 0.6177,
      "step": 15226
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4777675867080688,
      "learning_rate": 9.678482103142668e-06,
      "loss": 0.5322,
      "step": 15227
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.5864801406860352,
      "learning_rate": 9.677366108403944e-06,
      "loss": 0.602,
      "step": 15228
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3764524459838867,
      "learning_rate": 9.676250117687617e-06,
      "loss": 0.5632,
      "step": 15229
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5133147239685059,
      "learning_rate": 9.675134131007602e-06,
      "loss": 0.5129,
      "step": 15230
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3137998580932617,
      "learning_rate": 9.67401814837781e-06,
      "loss": 0.5989,
      "step": 15231
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6092358827590942,
      "learning_rate": 9.672902169812158e-06,
      "loss": 0.5873,
      "step": 15232
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.590272307395935,
      "learning_rate": 9.671786195324557e-06,
      "loss": 0.6028,
      "step": 15233
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7263789176940918,
      "learning_rate": 9.67067022492892e-06,
      "loss": 0.5819,
      "step": 15234
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2901986837387085,
      "learning_rate": 9.669554258639161e-06,
      "loss": 0.4896,
      "step": 15235
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.518331527709961,
      "learning_rate": 9.668438296469195e-06,
      "loss": 0.5347,
      "step": 15236
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2605516910552979,
      "learning_rate": 9.66732233843293e-06,
      "loss": 0.4078,
      "step": 15237
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.555533528327942,
      "learning_rate": 9.666206384544285e-06,
      "loss": 0.5607,
      "step": 15238
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3295329809188843,
      "learning_rate": 9.66509043481717e-06,
      "loss": 0.459,
      "step": 15239
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3494831323623657,
      "learning_rate": 9.663974489265495e-06,
      "loss": 0.4999,
      "step": 15240
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4984630346298218,
      "learning_rate": 9.662858547903181e-06,
      "loss": 0.6984,
      "step": 15241
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8955731391906738,
      "learning_rate": 9.661742610744132e-06,
      "loss": 0.4747,
      "step": 15242
    },
    {
      "epoch": 1.58,
      "grad_norm": 2.3533387184143066,
      "learning_rate": 9.660626677802268e-06,
      "loss": 0.5639,
      "step": 15243
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.858980655670166,
      "learning_rate": 9.659510749091497e-06,
      "loss": 0.5156,
      "step": 15244
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8859422206878662,
      "learning_rate": 9.658394824625732e-06,
      "loss": 0.4559,
      "step": 15245
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3826935291290283,
      "learning_rate": 9.65727890441889e-06,
      "loss": 0.523,
      "step": 15246
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4873874187469482,
      "learning_rate": 9.656162988484878e-06,
      "loss": 0.5683,
      "step": 15247
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.866539478302002,
      "learning_rate": 9.655047076837611e-06,
      "loss": 0.5516,
      "step": 15248
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4583135843276978,
      "learning_rate": 9.653931169491004e-06,
      "loss": 0.4864,
      "step": 15249
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4858373403549194,
      "learning_rate": 9.652815266458965e-06,
      "loss": 0.5072,
      "step": 15250
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.879502773284912,
      "learning_rate": 9.651699367755409e-06,
      "loss": 0.5547,
      "step": 15251
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5027904510498047,
      "learning_rate": 9.650583473394246e-06,
      "loss": 0.4673,
      "step": 15252
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.658578872680664,
      "learning_rate": 9.649467583389393e-06,
      "loss": 0.4977,
      "step": 15253
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7650262117385864,
      "learning_rate": 9.648351697754759e-06,
      "loss": 0.4422,
      "step": 15254
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6496033668518066,
      "learning_rate": 9.647235816504254e-06,
      "loss": 0.5525,
      "step": 15255
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5368231534957886,
      "learning_rate": 9.646119939651795e-06,
      "loss": 0.4915,
      "step": 15256
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.632189154624939,
      "learning_rate": 9.64500406721129e-06,
      "loss": 0.6445,
      "step": 15257
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5328572988510132,
      "learning_rate": 9.64388819919665e-06,
      "loss": 0.4896,
      "step": 15258
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4507794380187988,
      "learning_rate": 9.642772335621795e-06,
      "loss": 0.4714,
      "step": 15259
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.479459285736084,
      "learning_rate": 9.641656476500627e-06,
      "loss": 0.4883,
      "step": 15260
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8545961380004883,
      "learning_rate": 9.640540621847063e-06,
      "loss": 0.5614,
      "step": 15261
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5299040079116821,
      "learning_rate": 9.639424771675016e-06,
      "loss": 0.6304,
      "step": 15262
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6286412477493286,
      "learning_rate": 9.638308925998392e-06,
      "loss": 0.5865,
      "step": 15263
    },
    {
      "epoch": 1.58,
      "grad_norm": 2.0990042686462402,
      "learning_rate": 9.637193084831111e-06,
      "loss": 0.6624,
      "step": 15264
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.442568302154541,
      "learning_rate": 9.636077248187079e-06,
      "loss": 0.6007,
      "step": 15265
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3847287893295288,
      "learning_rate": 9.634961416080205e-06,
      "loss": 0.558,
      "step": 15266
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5211107730865479,
      "learning_rate": 9.633845588524408e-06,
      "loss": 0.491,
      "step": 15267
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.614728331565857,
      "learning_rate": 9.632729765533593e-06,
      "loss": 0.5531,
      "step": 15268
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.474961757659912,
      "learning_rate": 9.631613947121676e-06,
      "loss": 0.5183,
      "step": 15269
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.803682804107666,
      "learning_rate": 9.630498133302564e-06,
      "loss": 0.5605,
      "step": 15270
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4836782217025757,
      "learning_rate": 9.629382324090171e-06,
      "loss": 0.4,
      "step": 15271
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8571780920028687,
      "learning_rate": 9.62826651949841e-06,
      "loss": 0.58,
      "step": 15272
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3720855712890625,
      "learning_rate": 9.627150719541187e-06,
      "loss": 0.6461,
      "step": 15273
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5877363681793213,
      "learning_rate": 9.626034924232415e-06,
      "loss": 0.4456,
      "step": 15274
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8100452423095703,
      "learning_rate": 9.624919133586009e-06,
      "loss": 0.5284,
      "step": 15275
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4820772409439087,
      "learning_rate": 9.623803347615875e-06,
      "loss": 0.4951,
      "step": 15276
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7449572086334229,
      "learning_rate": 9.622687566335928e-06,
      "loss": 0.5596,
      "step": 15277
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5595310926437378,
      "learning_rate": 9.621571789760074e-06,
      "loss": 0.6982,
      "step": 15278
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3104864358901978,
      "learning_rate": 9.620456017902228e-06,
      "loss": 0.5297,
      "step": 15279
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4667538404464722,
      "learning_rate": 9.6193402507763e-06,
      "loss": 0.5738,
      "step": 15280
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6724810600280762,
      "learning_rate": 9.618224488396198e-06,
      "loss": 0.5217,
      "step": 15281
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5416536331176758,
      "learning_rate": 9.617108730775837e-06,
      "loss": 0.59,
      "step": 15282
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6871625185012817,
      "learning_rate": 9.615992977929123e-06,
      "loss": 0.5283,
      "step": 15283
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3795039653778076,
      "learning_rate": 9.614877229869968e-06,
      "loss": 0.5304,
      "step": 15284
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5176876783370972,
      "learning_rate": 9.613761486612288e-06,
      "loss": 0.5005,
      "step": 15285
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7395498752593994,
      "learning_rate": 9.612645748169984e-06,
      "loss": 0.56,
      "step": 15286
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.369924783706665,
      "learning_rate": 9.61153001455697e-06,
      "loss": 0.4789,
      "step": 15287
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2936033010482788,
      "learning_rate": 9.61041428578716e-06,
      "loss": 0.5763,
      "step": 15288
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5456469058990479,
      "learning_rate": 9.60929856187446e-06,
      "loss": 0.5538,
      "step": 15289
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2432525157928467,
      "learning_rate": 9.608182842832783e-06,
      "loss": 0.4438,
      "step": 15290
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2954500913619995,
      "learning_rate": 9.607067128676036e-06,
      "loss": 0.4701,
      "step": 15291
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5063978433609009,
      "learning_rate": 9.605951419418132e-06,
      "loss": 0.552,
      "step": 15292
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.9964940547943115,
      "learning_rate": 9.604835715072979e-06,
      "loss": 0.7083,
      "step": 15293
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5150442123413086,
      "learning_rate": 9.603720015654486e-06,
      "loss": 0.554,
      "step": 15294
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3121695518493652,
      "learning_rate": 9.602604321176567e-06,
      "loss": 0.4008,
      "step": 15295
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.717655897140503,
      "learning_rate": 9.601488631653126e-06,
      "loss": 0.5148,
      "step": 15296
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7932409048080444,
      "learning_rate": 9.600372947098076e-06,
      "loss": 0.5858,
      "step": 15297
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.492701530456543,
      "learning_rate": 9.59925726752533e-06,
      "loss": 0.4844,
      "step": 15298
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.466356873512268,
      "learning_rate": 9.59814159294879e-06,
      "loss": 0.4427,
      "step": 15299
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.5207452774047852,
      "learning_rate": 9.597025923382371e-06,
      "loss": 0.6432,
      "step": 15300
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6486070156097412,
      "learning_rate": 9.595910258839984e-06,
      "loss": 0.4501,
      "step": 15301
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3796496391296387,
      "learning_rate": 9.594794599335531e-06,
      "loss": 0.4249,
      "step": 15302
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4695793390274048,
      "learning_rate": 9.593678944882931e-06,
      "loss": 0.5013,
      "step": 15303
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8378498554229736,
      "learning_rate": 9.592563295496084e-06,
      "loss": 0.6269,
      "step": 15304
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4005634784698486,
      "learning_rate": 9.591447651188902e-06,
      "loss": 0.5124,
      "step": 15305
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.492321491241455,
      "learning_rate": 9.590332011975299e-06,
      "loss": 0.5842,
      "step": 15306
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.9919819831848145,
      "learning_rate": 9.589216377869179e-06,
      "loss": 0.5525,
      "step": 15307
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6377320289611816,
      "learning_rate": 9.588100748884453e-06,
      "loss": 0.4832,
      "step": 15308
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6025464534759521,
      "learning_rate": 9.58698512503503e-06,
      "loss": 0.626,
      "step": 15309
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4321175813674927,
      "learning_rate": 9.585869506334816e-06,
      "loss": 0.4858,
      "step": 15310
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4861668348312378,
      "learning_rate": 9.584753892797726e-06,
      "loss": 0.6311,
      "step": 15311
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8417370319366455,
      "learning_rate": 9.583638284437661e-06,
      "loss": 0.5676,
      "step": 15312
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7423828840255737,
      "learning_rate": 9.582522681268537e-06,
      "loss": 0.5653,
      "step": 15313
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7688326835632324,
      "learning_rate": 9.581407083304259e-06,
      "loss": 0.6187,
      "step": 15314
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.571132779121399,
      "learning_rate": 9.580291490558736e-06,
      "loss": 0.5265,
      "step": 15315
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.7263224124908447,
      "learning_rate": 9.579175903045877e-06,
      "loss": 0.5843,
      "step": 15316
    },
    {
      "epoch": 1.58,
      "grad_norm": 2.1642048358917236,
      "learning_rate": 9.578060320779588e-06,
      "loss": 0.578,
      "step": 15317
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.56540048122406,
      "learning_rate": 9.57694474377378e-06,
      "loss": 0.4449,
      "step": 15318
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6437073945999146,
      "learning_rate": 9.575829172042361e-06,
      "loss": 0.4952,
      "step": 15319
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.641849160194397,
      "learning_rate": 9.574713605599239e-06,
      "loss": 0.627,
      "step": 15320
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1911125183105469,
      "learning_rate": 9.573598044458324e-06,
      "loss": 0.4,
      "step": 15321
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4679203033447266,
      "learning_rate": 9.572482488633518e-06,
      "loss": 0.4655,
      "step": 15322
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4727038145065308,
      "learning_rate": 9.571366938138734e-06,
      "loss": 0.4545,
      "step": 15323
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3220876455307007,
      "learning_rate": 9.570251392987883e-06,
      "loss": 0.4519,
      "step": 15324
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.6426782608032227,
      "learning_rate": 9.569135853194867e-06,
      "loss": 0.541,
      "step": 15325
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8298324346542358,
      "learning_rate": 9.568020318773594e-06,
      "loss": 0.5896,
      "step": 15326
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8827098608016968,
      "learning_rate": 9.566904789737978e-06,
      "loss": 0.5945,
      "step": 15327
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6813242435455322,
      "learning_rate": 9.56578926610192e-06,
      "loss": 0.4035,
      "step": 15328
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.802542805671692,
      "learning_rate": 9.564673747879331e-06,
      "loss": 0.5104,
      "step": 15329
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5037930011749268,
      "learning_rate": 9.563558235084116e-06,
      "loss": 0.5931,
      "step": 15330
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8015050888061523,
      "learning_rate": 9.562442727730185e-06,
      "loss": 0.5244,
      "step": 15331
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9859868288040161,
      "learning_rate": 9.561327225831445e-06,
      "loss": 0.6716,
      "step": 15332
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5710405111312866,
      "learning_rate": 9.560211729401805e-06,
      "loss": 0.5113,
      "step": 15333
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.712549090385437,
      "learning_rate": 9.559096238455169e-06,
      "loss": 0.5754,
      "step": 15334
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7492939233779907,
      "learning_rate": 9.557980753005445e-06,
      "loss": 0.5574,
      "step": 15335
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6192463636398315,
      "learning_rate": 9.556865273066542e-06,
      "loss": 0.5444,
      "step": 15336
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.487620234489441,
      "learning_rate": 9.555749798652365e-06,
      "loss": 0.6262,
      "step": 15337
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9491750001907349,
      "learning_rate": 9.554634329776821e-06,
      "loss": 0.5568,
      "step": 15338
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6799139976501465,
      "learning_rate": 9.553518866453819e-06,
      "loss": 0.5645,
      "step": 15339
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6740652322769165,
      "learning_rate": 9.552403408697266e-06,
      "loss": 0.5313,
      "step": 15340
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3585845232009888,
      "learning_rate": 9.551287956521067e-06,
      "loss": 0.4085,
      "step": 15341
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.422486424446106,
      "learning_rate": 9.55017250993913e-06,
      "loss": 0.4328,
      "step": 15342
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2817776203155518,
      "learning_rate": 9.54905706896536e-06,
      "loss": 0.6247,
      "step": 15343
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.576491355895996,
      "learning_rate": 9.547941633613665e-06,
      "loss": 0.5888,
      "step": 15344
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5857294797897339,
      "learning_rate": 9.546826203897953e-06,
      "loss": 0.5293,
      "step": 15345
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5179842710494995,
      "learning_rate": 9.545710779832126e-06,
      "loss": 0.5288,
      "step": 15346
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4814845323562622,
      "learning_rate": 9.544595361430097e-06,
      "loss": 0.4903,
      "step": 15347
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4062763452529907,
      "learning_rate": 9.543479948705766e-06,
      "loss": 0.487,
      "step": 15348
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.431883454322815,
      "learning_rate": 9.542364541673042e-06,
      "loss": 0.4606,
      "step": 15349
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5267101526260376,
      "learning_rate": 9.541249140345831e-06,
      "loss": 0.5493,
      "step": 15350
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4078763723373413,
      "learning_rate": 9.54013374473804e-06,
      "loss": 0.5709,
      "step": 15351
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4464362859725952,
      "learning_rate": 9.539018354863571e-06,
      "loss": 0.5419,
      "step": 15352
    },
    {
      "epoch": 1.59,
      "grad_norm": 2.081601858139038,
      "learning_rate": 9.537902970736339e-06,
      "loss": 0.5011,
      "step": 15353
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9023964405059814,
      "learning_rate": 9.53678759237024e-06,
      "loss": 0.6361,
      "step": 15354
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4355250597000122,
      "learning_rate": 9.535672219779186e-06,
      "loss": 0.6363,
      "step": 15355
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3832781314849854,
      "learning_rate": 9.534556852977077e-06,
      "loss": 0.4526,
      "step": 15356
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4993133544921875,
      "learning_rate": 9.533441491977823e-06,
      "loss": 0.5069,
      "step": 15357
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5945287942886353,
      "learning_rate": 9.532326136795333e-06,
      "loss": 0.5975,
      "step": 15358
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3690769672393799,
      "learning_rate": 9.531210787443507e-06,
      "loss": 0.5636,
      "step": 15359
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3578400611877441,
      "learning_rate": 9.530095443936253e-06,
      "loss": 0.4685,
      "step": 15360
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2271473407745361,
      "learning_rate": 9.528980106287471e-06,
      "loss": 0.4432,
      "step": 15361
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3807927370071411,
      "learning_rate": 9.527864774511074e-06,
      "loss": 0.6403,
      "step": 15362
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.570696473121643,
      "learning_rate": 9.526749448620965e-06,
      "loss": 0.5959,
      "step": 15363
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6996537446975708,
      "learning_rate": 9.525634128631047e-06,
      "loss": 0.5762,
      "step": 15364
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.496503233909607,
      "learning_rate": 9.524518814555227e-06,
      "loss": 0.5838,
      "step": 15365
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5052658319473267,
      "learning_rate": 9.52340350640741e-06,
      "loss": 0.4523,
      "step": 15366
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.775233507156372,
      "learning_rate": 9.522288204201498e-06,
      "loss": 0.5523,
      "step": 15367
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9256685972213745,
      "learning_rate": 9.5211729079514e-06,
      "loss": 0.606,
      "step": 15368
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.446248173713684,
      "learning_rate": 9.520057617671017e-06,
      "loss": 0.563,
      "step": 15369
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1665432453155518,
      "learning_rate": 9.518942333374258e-06,
      "loss": 0.4308,
      "step": 15370
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.9252405166625977,
      "learning_rate": 9.517827055075027e-06,
      "loss": 0.4924,
      "step": 15371
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6926039457321167,
      "learning_rate": 9.516711782787223e-06,
      "loss": 0.6261,
      "step": 15372
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6147644519805908,
      "learning_rate": 9.515596516524757e-06,
      "loss": 0.4797,
      "step": 15373
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6416152715682983,
      "learning_rate": 9.514481256301531e-06,
      "loss": 0.6656,
      "step": 15374
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5970720052719116,
      "learning_rate": 9.513366002131448e-06,
      "loss": 0.568,
      "step": 15375
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5339100360870361,
      "learning_rate": 9.512250754028417e-06,
      "loss": 0.4783,
      "step": 15376
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8808830976486206,
      "learning_rate": 9.511135512006336e-06,
      "loss": 0.5878,
      "step": 15377
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.697309970855713,
      "learning_rate": 9.510020276079112e-06,
      "loss": 0.5993,
      "step": 15378
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4331163167953491,
      "learning_rate": 9.50890504626065e-06,
      "loss": 0.4754,
      "step": 15379
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6622611284255981,
      "learning_rate": 9.507789822564853e-06,
      "loss": 0.4993,
      "step": 15380
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4524743556976318,
      "learning_rate": 9.506674605005626e-06,
      "loss": 0.5679,
      "step": 15381
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.537793755531311,
      "learning_rate": 9.50555939359687e-06,
      "loss": 0.4815,
      "step": 15382
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3032296895980835,
      "learning_rate": 9.504444188352491e-06,
      "loss": 0.5396,
      "step": 15383
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6314170360565186,
      "learning_rate": 9.503328989286394e-06,
      "loss": 0.6013,
      "step": 15384
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4337623119354248,
      "learning_rate": 9.502213796412479e-06,
      "loss": 0.5094,
      "step": 15385
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8840420246124268,
      "learning_rate": 9.501098609744654e-06,
      "loss": 0.5858,
      "step": 15386
    },
    {
      "epoch": 1.59,
      "grad_norm": 2.006347894668579,
      "learning_rate": 9.499983429296817e-06,
      "loss": 0.5285,
      "step": 15387
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7357944250106812,
      "learning_rate": 9.498868255082873e-06,
      "loss": 0.6034,
      "step": 15388
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7229260206222534,
      "learning_rate": 9.497753087116731e-06,
      "loss": 0.5869,
      "step": 15389
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3051470518112183,
      "learning_rate": 9.496637925412287e-06,
      "loss": 0.5904,
      "step": 15390
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8122601509094238,
      "learning_rate": 9.495522769983447e-06,
      "loss": 0.5937,
      "step": 15391
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4591166973114014,
      "learning_rate": 9.494407620844116e-06,
      "loss": 0.4655,
      "step": 15392
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2951446771621704,
      "learning_rate": 9.493292478008193e-06,
      "loss": 0.4176,
      "step": 15393
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7658467292785645,
      "learning_rate": 9.492177341489584e-06,
      "loss": 0.4913,
      "step": 15394
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.793585181236267,
      "learning_rate": 9.491062211302188e-06,
      "loss": 0.4945,
      "step": 15395
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8259742259979248,
      "learning_rate": 9.489947087459913e-06,
      "loss": 0.6281,
      "step": 15396
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4888272285461426,
      "learning_rate": 9.48883196997666e-06,
      "loss": 0.5376,
      "step": 15397
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2041879892349243,
      "learning_rate": 9.487716858866329e-06,
      "loss": 0.5547,
      "step": 15398
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6625438928604126,
      "learning_rate": 9.486601754142824e-06,
      "loss": 0.5881,
      "step": 15399
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5693707466125488,
      "learning_rate": 9.48548665582005e-06,
      "loss": 0.497,
      "step": 15400
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3213071823120117,
      "learning_rate": 9.484371563911903e-06,
      "loss": 0.525,
      "step": 15401
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.613709568977356,
      "learning_rate": 9.483256478432292e-06,
      "loss": 0.4461,
      "step": 15402
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3206921815872192,
      "learning_rate": 9.482141399395116e-06,
      "loss": 0.5515,
      "step": 15403
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4830408096313477,
      "learning_rate": 9.481026326814275e-06,
      "loss": 0.49,
      "step": 15404
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7884279489517212,
      "learning_rate": 9.479911260703679e-06,
      "loss": 0.3961,
      "step": 15405
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5680214166641235,
      "learning_rate": 9.47879620107722e-06,
      "loss": 0.6477,
      "step": 15406
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4348095655441284,
      "learning_rate": 9.477681147948806e-06,
      "loss": 0.5344,
      "step": 15407
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4821385145187378,
      "learning_rate": 9.476566101332336e-06,
      "loss": 0.525,
      "step": 15408
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.62265944480896,
      "learning_rate": 9.475451061241714e-06,
      "loss": 0.5596,
      "step": 15409
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6387969255447388,
      "learning_rate": 9.474336027690842e-06,
      "loss": 0.5563,
      "step": 15410
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5608240365982056,
      "learning_rate": 9.473221000693617e-06,
      "loss": 0.6084,
      "step": 15411
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4665902853012085,
      "learning_rate": 9.472105980263946e-06,
      "loss": 0.4866,
      "step": 15412
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4554539918899536,
      "learning_rate": 9.470990966415726e-06,
      "loss": 0.4091,
      "step": 15413
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4845857620239258,
      "learning_rate": 9.46987595916286e-06,
      "loss": 0.5968,
      "step": 15414
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3766120672225952,
      "learning_rate": 9.468760958519251e-06,
      "loss": 0.4932,
      "step": 15415
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7206978797912598,
      "learning_rate": 9.467645964498797e-06,
      "loss": 0.6349,
      "step": 15416
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6523323059082031,
      "learning_rate": 9.4665309771154e-06,
      "loss": 0.6512,
      "step": 15417
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4631385803222656,
      "learning_rate": 9.465415996382966e-06,
      "loss": 0.6821,
      "step": 15418
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7660118341445923,
      "learning_rate": 9.464301022315387e-06,
      "loss": 0.5692,
      "step": 15419
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7018836736679077,
      "learning_rate": 9.463186054926571e-06,
      "loss": 0.5837,
      "step": 15420
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.267613410949707,
      "learning_rate": 9.462071094230414e-06,
      "loss": 0.5217,
      "step": 15421
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4291772842407227,
      "learning_rate": 9.460956140240819e-06,
      "loss": 0.4564,
      "step": 15422
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5800799131393433,
      "learning_rate": 9.459841192971687e-06,
      "loss": 0.5463,
      "step": 15423
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.9787564277648926,
      "learning_rate": 9.458726252436915e-06,
      "loss": 0.5948,
      "step": 15424
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6343976259231567,
      "learning_rate": 9.45761131865041e-06,
      "loss": 0.5224,
      "step": 15425
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4250223636627197,
      "learning_rate": 9.456496391626065e-06,
      "loss": 0.4505,
      "step": 15426
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7612675428390503,
      "learning_rate": 9.455381471377783e-06,
      "loss": 0.5687,
      "step": 15427
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.574897050857544,
      "learning_rate": 9.454266557919468e-06,
      "loss": 0.5599,
      "step": 15428
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5674619674682617,
      "learning_rate": 9.453151651265014e-06,
      "loss": 0.6199,
      "step": 15429
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6305737495422363,
      "learning_rate": 9.452036751428325e-06,
      "loss": 0.5825,
      "step": 15430
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5480852127075195,
      "learning_rate": 9.4509218584233e-06,
      "loss": 0.6791,
      "step": 15431
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4308823347091675,
      "learning_rate": 9.449806972263835e-06,
      "loss": 0.5552,
      "step": 15432
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.617470145225525,
      "learning_rate": 9.448692092963837e-06,
      "loss": 0.4523,
      "step": 15433
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.707767367362976,
      "learning_rate": 9.4475772205372e-06,
      "loss": 0.6503,
      "step": 15434
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5711969137191772,
      "learning_rate": 9.446462354997823e-06,
      "loss": 0.4899,
      "step": 15435
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.422481894493103,
      "learning_rate": 9.44534749635961e-06,
      "loss": 0.5452,
      "step": 15436
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.037900686264038,
      "learning_rate": 9.444232644636457e-06,
      "loss": 0.5925,
      "step": 15437
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4537014961242676,
      "learning_rate": 9.443117799842266e-06,
      "loss": 0.5295,
      "step": 15438
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6434723138809204,
      "learning_rate": 9.442002961990933e-06,
      "loss": 0.5261,
      "step": 15439
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4470912218093872,
      "learning_rate": 9.440888131096357e-06,
      "loss": 0.5445,
      "step": 15440
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.54597806930542,
      "learning_rate": 9.439773307172441e-06,
      "loss": 0.5242,
      "step": 15441
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6069397926330566,
      "learning_rate": 9.438658490233081e-06,
      "loss": 0.4896,
      "step": 15442
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.569003939628601,
      "learning_rate": 9.437543680292177e-06,
      "loss": 0.5113,
      "step": 15443
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.792303442955017,
      "learning_rate": 9.436428877363625e-06,
      "loss": 0.6368,
      "step": 15444
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5896544456481934,
      "learning_rate": 9.435314081461326e-06,
      "loss": 0.5576,
      "step": 15445
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.557416319847107,
      "learning_rate": 9.434199292599179e-06,
      "loss": 0.5967,
      "step": 15446
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2505106925964355,
      "learning_rate": 9.433084510791082e-06,
      "loss": 0.4517,
      "step": 15447
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3257485628128052,
      "learning_rate": 9.431969736050932e-06,
      "loss": 0.4881,
      "step": 15448
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.472559928894043,
      "learning_rate": 9.43085496839263e-06,
      "loss": 0.4734,
      "step": 15449
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6749258041381836,
      "learning_rate": 9.429740207830072e-06,
      "loss": 0.4734,
      "step": 15450
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5271146297454834,
      "learning_rate": 9.428625454377158e-06,
      "loss": 0.5689,
      "step": 15451
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6037803888320923,
      "learning_rate": 9.427510708047783e-06,
      "loss": 0.5501,
      "step": 15452
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4093793630599976,
      "learning_rate": 9.426395968855848e-06,
      "loss": 0.541,
      "step": 15453
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.491081714630127,
      "learning_rate": 9.425281236815251e-06,
      "loss": 0.5211,
      "step": 15454
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7064533233642578,
      "learning_rate": 9.424166511939887e-06,
      "loss": 0.5842,
      "step": 15455
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.916064977645874,
      "learning_rate": 9.423051794243656e-06,
      "loss": 0.5662,
      "step": 15456
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5592669248580933,
      "learning_rate": 9.421937083740452e-06,
      "loss": 0.5696,
      "step": 15457
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4456887245178223,
      "learning_rate": 9.420822380444177e-06,
      "loss": 0.4388,
      "step": 15458
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2477606534957886,
      "learning_rate": 9.419707684368728e-06,
      "loss": 0.4306,
      "step": 15459
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6065424680709839,
      "learning_rate": 9.418592995528e-06,
      "loss": 0.5501,
      "step": 15460
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4838076829910278,
      "learning_rate": 9.417478313935891e-06,
      "loss": 0.515,
      "step": 15461
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.577239990234375,
      "learning_rate": 9.4163636396063e-06,
      "loss": 0.4829,
      "step": 15462
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6349234580993652,
      "learning_rate": 9.415248972553122e-06,
      "loss": 0.5581,
      "step": 15463
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2592495679855347,
      "learning_rate": 9.414134312790255e-06,
      "loss": 0.4256,
      "step": 15464
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6856039762496948,
      "learning_rate": 9.413019660331595e-06,
      "loss": 0.5445,
      "step": 15465
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3745841979980469,
      "learning_rate": 9.411905015191037e-06,
      "loss": 0.5312,
      "step": 15466
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3936223983764648,
      "learning_rate": 9.410790377382486e-06,
      "loss": 0.4804,
      "step": 15467
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7445341348648071,
      "learning_rate": 9.40967574691983e-06,
      "loss": 0.4982,
      "step": 15468
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2264528274536133,
      "learning_rate": 9.408561123816966e-06,
      "loss": 0.5058,
      "step": 15469
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5944232940673828,
      "learning_rate": 9.40744650808779e-06,
      "loss": 0.5436,
      "step": 15470
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7348936796188354,
      "learning_rate": 9.406331899746206e-06,
      "loss": 0.4719,
      "step": 15471
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.31900954246521,
      "learning_rate": 9.405217298806103e-06,
      "loss": 0.4784,
      "step": 15472
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5025447607040405,
      "learning_rate": 9.40410270528138e-06,
      "loss": 0.5885,
      "step": 15473
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.716500997543335,
      "learning_rate": 9.40298811918593e-06,
      "loss": 0.5886,
      "step": 15474
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6974256038665771,
      "learning_rate": 9.401873540533652e-06,
      "loss": 0.4718,
      "step": 15475
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6381028890609741,
      "learning_rate": 9.400758969338442e-06,
      "loss": 0.4833,
      "step": 15476
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.283685326576233,
      "learning_rate": 9.399644405614193e-06,
      "loss": 0.4076,
      "step": 15477
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8283092975616455,
      "learning_rate": 9.398529849374806e-06,
      "loss": 0.5479,
      "step": 15478
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7846566438674927,
      "learning_rate": 9.39741530063417e-06,
      "loss": 0.6335,
      "step": 15479
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8839834928512573,
      "learning_rate": 9.396300759406183e-06,
      "loss": 0.5077,
      "step": 15480
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.541835904121399,
      "learning_rate": 9.395186225704744e-06,
      "loss": 0.557,
      "step": 15481
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6540392637252808,
      "learning_rate": 9.394071699543743e-06,
      "loss": 0.5559,
      "step": 15482
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6303805112838745,
      "learning_rate": 9.392957180937077e-06,
      "loss": 0.4524,
      "step": 15483
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8555717468261719,
      "learning_rate": 9.391842669898644e-06,
      "loss": 0.6375,
      "step": 15484
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4905343055725098,
      "learning_rate": 9.390728166442336e-06,
      "loss": 0.5344,
      "step": 15485
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5700427293777466,
      "learning_rate": 9.38961367058205e-06,
      "loss": 0.6504,
      "step": 15486
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3419432640075684,
      "learning_rate": 9.388499182331676e-06,
      "loss": 0.493,
      "step": 15487
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5260202884674072,
      "learning_rate": 9.387384701705113e-06,
      "loss": 0.4804,
      "step": 15488
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6870746612548828,
      "learning_rate": 9.386270228716258e-06,
      "loss": 0.5099,
      "step": 15489
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5545942783355713,
      "learning_rate": 9.385155763378999e-06,
      "loss": 0.5449,
      "step": 15490
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5211516618728638,
      "learning_rate": 9.384041305707236e-06,
      "loss": 0.5292,
      "step": 15491
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.423647403717041,
      "learning_rate": 9.38292685571486e-06,
      "loss": 0.6285,
      "step": 15492
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6253005266189575,
      "learning_rate": 9.381812413415767e-06,
      "loss": 0.5897,
      "step": 15493
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6400887966156006,
      "learning_rate": 9.380697978823852e-06,
      "loss": 0.5473,
      "step": 15494
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6445890665054321,
      "learning_rate": 9.379583551953005e-06,
      "loss": 0.4933,
      "step": 15495
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5033822059631348,
      "learning_rate": 9.378469132817123e-06,
      "loss": 0.5766,
      "step": 15496
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3849533796310425,
      "learning_rate": 9.377354721430103e-06,
      "loss": 0.5506,
      "step": 15497
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8731123208999634,
      "learning_rate": 9.376240317805831e-06,
      "loss": 0.4195,
      "step": 15498
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4474904537200928,
      "learning_rate": 9.37512592195821e-06,
      "loss": 0.4888,
      "step": 15499
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2826018333435059,
      "learning_rate": 9.374011533901123e-06,
      "loss": 0.4779,
      "step": 15500
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7526293992996216,
      "learning_rate": 9.372897153648473e-06,
      "loss": 0.6495,
      "step": 15501
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.500489592552185,
      "learning_rate": 9.371782781214148e-06,
      "loss": 0.5728,
      "step": 15502
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.529118299484253,
      "learning_rate": 9.370668416612043e-06,
      "loss": 0.452,
      "step": 15503
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5126115083694458,
      "learning_rate": 9.369554059856053e-06,
      "loss": 0.4676,
      "step": 15504
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.554992914199829,
      "learning_rate": 9.368439710960067e-06,
      "loss": 0.5689,
      "step": 15505
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5053163766860962,
      "learning_rate": 9.36732536993798e-06,
      "loss": 0.5885,
      "step": 15506
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7001779079437256,
      "learning_rate": 9.366211036803687e-06,
      "loss": 0.59,
      "step": 15507
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6141881942749023,
      "learning_rate": 9.365096711571076e-06,
      "loss": 0.5914,
      "step": 15508
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8634264469146729,
      "learning_rate": 9.363982394254045e-06,
      "loss": 0.6598,
      "step": 15509
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5625096559524536,
      "learning_rate": 9.362868084866485e-06,
      "loss": 0.5437,
      "step": 15510
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3721206188201904,
      "learning_rate": 9.361753783422285e-06,
      "loss": 0.5665,
      "step": 15511
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7792881727218628,
      "learning_rate": 9.360639489935342e-06,
      "loss": 0.6116,
      "step": 15512
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.7627133131027222,
      "learning_rate": 9.359525204419546e-06,
      "loss": 0.4644,
      "step": 15513
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6762562990188599,
      "learning_rate": 9.358410926888788e-06,
      "loss": 0.5218,
      "step": 15514
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4764577150344849,
      "learning_rate": 9.357296657356966e-06,
      "loss": 0.5799,
      "step": 15515
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.551945686340332,
      "learning_rate": 9.356182395837964e-06,
      "loss": 0.5268,
      "step": 15516
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3715946674346924,
      "learning_rate": 9.35506814234568e-06,
      "loss": 0.4612,
      "step": 15517
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.9677364826202393,
      "learning_rate": 9.353953896894001e-06,
      "loss": 0.4111,
      "step": 15518
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7645916938781738,
      "learning_rate": 9.352839659496822e-06,
      "loss": 0.5754,
      "step": 15519
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6855113506317139,
      "learning_rate": 9.351725430168036e-06,
      "loss": 0.5179,
      "step": 15520
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.055534601211548,
      "learning_rate": 9.35061120892153e-06,
      "loss": 0.5775,
      "step": 15521
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.820249080657959,
      "learning_rate": 9.349496995771197e-06,
      "loss": 0.5139,
      "step": 15522
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.011221170425415,
      "learning_rate": 9.348382790730932e-06,
      "loss": 0.6539,
      "step": 15523
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5570614337921143,
      "learning_rate": 9.347268593814623e-06,
      "loss": 0.5892,
      "step": 15524
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6874967813491821,
      "learning_rate": 9.346154405036162e-06,
      "loss": 0.5646,
      "step": 15525
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4151467084884644,
      "learning_rate": 9.345040224409437e-06,
      "loss": 0.4611,
      "step": 15526
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7422252893447876,
      "learning_rate": 9.34392605194834e-06,
      "loss": 0.6485,
      "step": 15527
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2710996866226196,
      "learning_rate": 9.342811887666768e-06,
      "loss": 0.5319,
      "step": 15528
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5780205726623535,
      "learning_rate": 9.341697731578603e-06,
      "loss": 0.5405,
      "step": 15529
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6936694383621216,
      "learning_rate": 9.340583583697743e-06,
      "loss": 0.4584,
      "step": 15530
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4013278484344482,
      "learning_rate": 9.339469444038072e-06,
      "loss": 0.4887,
      "step": 15531
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.9564173221588135,
      "learning_rate": 9.338355312613483e-06,
      "loss": 0.5635,
      "step": 15532
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6386208534240723,
      "learning_rate": 9.337241189437868e-06,
      "loss": 0.5675,
      "step": 15533
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3930813074111938,
      "learning_rate": 9.336127074525117e-06,
      "loss": 0.592,
      "step": 15534
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.0373473167419434,
      "learning_rate": 9.335012967889117e-06,
      "loss": 0.6487,
      "step": 15535
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.795980453491211,
      "learning_rate": 9.333898869543761e-06,
      "loss": 0.5912,
      "step": 15536
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6002628803253174,
      "learning_rate": 9.332784779502937e-06,
      "loss": 0.4999,
      "step": 15537
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.568035364151001,
      "learning_rate": 9.331670697780536e-06,
      "loss": 0.3966,
      "step": 15538
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.380426287651062,
      "learning_rate": 9.330556624390447e-06,
      "loss": 0.5309,
      "step": 15539
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.284745216369629,
      "learning_rate": 9.32944255934656e-06,
      "loss": 0.5216,
      "step": 15540
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.962192177772522,
      "learning_rate": 9.328328502662765e-06,
      "loss": 0.5075,
      "step": 15541
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4434901475906372,
      "learning_rate": 9.327214454352948e-06,
      "loss": 0.5872,
      "step": 15542
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.482681155204773,
      "learning_rate": 9.326100414431005e-06,
      "loss": 0.6086,
      "step": 15543
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5211032629013062,
      "learning_rate": 9.324986382910818e-06,
      "loss": 0.5488,
      "step": 15544
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.394534707069397,
      "learning_rate": 9.32387235980628e-06,
      "loss": 0.5373,
      "step": 15545
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5427284240722656,
      "learning_rate": 9.322758345131279e-06,
      "loss": 0.6284,
      "step": 15546
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5039052963256836,
      "learning_rate": 9.321644338899704e-06,
      "loss": 0.5349,
      "step": 15547
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.8183369636535645,
      "learning_rate": 9.320530341125444e-06,
      "loss": 0.541,
      "step": 15548
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6174917221069336,
      "learning_rate": 9.319416351822388e-06,
      "loss": 0.6089,
      "step": 15549
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4608224630355835,
      "learning_rate": 9.318302371004421e-06,
      "loss": 0.5712,
      "step": 15550
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.8524280786514282,
      "learning_rate": 9.317188398685439e-06,
      "loss": 0.6417,
      "step": 15551
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3205695152282715,
      "learning_rate": 9.31607443487932e-06,
      "loss": 0.5147,
      "step": 15552
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.993765950202942,
      "learning_rate": 9.31496047959996e-06,
      "loss": 0.626,
      "step": 15553
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3141154050827026,
      "learning_rate": 9.313846532861246e-06,
      "loss": 0.5278,
      "step": 15554
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7230089902877808,
      "learning_rate": 9.312732594677062e-06,
      "loss": 0.5991,
      "step": 15555
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.465556263923645,
      "learning_rate": 9.311618665061303e-06,
      "loss": 0.6392,
      "step": 15556
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3850737810134888,
      "learning_rate": 9.31050474402785e-06,
      "loss": 0.503,
      "step": 15557
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.714020013809204,
      "learning_rate": 9.309390831590592e-06,
      "loss": 0.5088,
      "step": 15558
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.028488874435425,
      "learning_rate": 9.30827692776342e-06,
      "loss": 0.5362,
      "step": 15559
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4536645412445068,
      "learning_rate": 9.307163032560218e-06,
      "loss": 0.5117,
      "step": 15560
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.539557933807373,
      "learning_rate": 9.306049145994873e-06,
      "loss": 0.4411,
      "step": 15561
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.527279257774353,
      "learning_rate": 9.304935268081276e-06,
      "loss": 0.5426,
      "step": 15562
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6873670816421509,
      "learning_rate": 9.303821398833312e-06,
      "loss": 0.4729,
      "step": 15563
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.409637689590454,
      "learning_rate": 9.302707538264868e-06,
      "loss": 0.4044,
      "step": 15564
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5638821125030518,
      "learning_rate": 9.301593686389828e-06,
      "loss": 0.5762,
      "step": 15565
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4245339632034302,
      "learning_rate": 9.300479843222083e-06,
      "loss": 0.523,
      "step": 15566
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6822574138641357,
      "learning_rate": 9.29936600877552e-06,
      "loss": 0.6095,
      "step": 15567
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4579657316207886,
      "learning_rate": 9.298252183064022e-06,
      "loss": 0.5567,
      "step": 15568
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6053595542907715,
      "learning_rate": 9.29713836610148e-06,
      "loss": 0.4842,
      "step": 15569
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5858722925186157,
      "learning_rate": 9.296024557901775e-06,
      "loss": 0.5018,
      "step": 15570
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6736814975738525,
      "learning_rate": 9.294910758478797e-06,
      "loss": 0.5205,
      "step": 15571
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5713765621185303,
      "learning_rate": 9.293796967846434e-06,
      "loss": 0.4437,
      "step": 15572
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6745631694793701,
      "learning_rate": 9.292683186018566e-06,
      "loss": 0.5387,
      "step": 15573
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6997458934783936,
      "learning_rate": 9.29156941300908e-06,
      "loss": 0.5203,
      "step": 15574
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.490356683731079,
      "learning_rate": 9.29045564883187e-06,
      "loss": 0.475,
      "step": 15575
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6706854104995728,
      "learning_rate": 9.289341893500812e-06,
      "loss": 0.6233,
      "step": 15576
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.8673039674758911,
      "learning_rate": 9.288228147029797e-06,
      "loss": 0.5233,
      "step": 15577
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4746958017349243,
      "learning_rate": 9.287114409432707e-06,
      "loss": 0.4875,
      "step": 15578
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6983263492584229,
      "learning_rate": 9.28600068072343e-06,
      "loss": 0.5039,
      "step": 15579
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6216425895690918,
      "learning_rate": 9.284886960915851e-06,
      "loss": 0.5365,
      "step": 15580
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4313814640045166,
      "learning_rate": 9.283773250023853e-06,
      "loss": 0.4657,
      "step": 15581
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4705332517623901,
      "learning_rate": 9.282659548061326e-06,
      "loss": 0.5666,
      "step": 15582
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3694137334823608,
      "learning_rate": 9.28154585504215e-06,
      "loss": 0.5736,
      "step": 15583
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6007286310195923,
      "learning_rate": 9.28043217098021e-06,
      "loss": 0.5753,
      "step": 15584
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5937234163284302,
      "learning_rate": 9.279318495889393e-06,
      "loss": 0.5009,
      "step": 15585
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.521582841873169,
      "learning_rate": 9.278204829783582e-06,
      "loss": 0.6049,
      "step": 15586
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3798874616622925,
      "learning_rate": 9.277091172676664e-06,
      "loss": 0.5534,
      "step": 15587
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3211926221847534,
      "learning_rate": 9.275977524582521e-06,
      "loss": 0.5037,
      "step": 15588
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.596468210220337,
      "learning_rate": 9.274863885515038e-06,
      "loss": 0.487,
      "step": 15589
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.8730173110961914,
      "learning_rate": 9.2737502554881e-06,
      "loss": 0.5828,
      "step": 15590
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2768789529800415,
      "learning_rate": 9.27263663451559e-06,
      "loss": 0.4919,
      "step": 15591
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3849005699157715,
      "learning_rate": 9.27152302261139e-06,
      "loss": 0.3857,
      "step": 15592
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4704195261001587,
      "learning_rate": 9.270409419789388e-06,
      "loss": 0.509,
      "step": 15593
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.437315821647644,
      "learning_rate": 9.269295826063465e-06,
      "loss": 0.5386,
      "step": 15594
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3984806537628174,
      "learning_rate": 9.268182241447507e-06,
      "loss": 0.5098,
      "step": 15595
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3985404968261719,
      "learning_rate": 9.267068665955393e-06,
      "loss": 0.5666,
      "step": 15596
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5085746049880981,
      "learning_rate": 9.265955099601008e-06,
      "loss": 0.5895,
      "step": 15597
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.097374677658081,
      "learning_rate": 9.26484154239824e-06,
      "loss": 0.6706,
      "step": 15598
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6313862800598145,
      "learning_rate": 9.263727994360965e-06,
      "loss": 0.5041,
      "step": 15599
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.780603051185608,
      "learning_rate": 9.26261445550307e-06,
      "loss": 0.4642,
      "step": 15600
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.7957019805908203,
      "learning_rate": 9.26150092583844e-06,
      "loss": 0.4744,
      "step": 15601
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6289466619491577,
      "learning_rate": 9.260387405380952e-06,
      "loss": 0.4957,
      "step": 15602
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3464488983154297,
      "learning_rate": 9.259273894144495e-06,
      "loss": 0.4388,
      "step": 15603
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4520275592803955,
      "learning_rate": 9.258160392142947e-06,
      "loss": 0.6128,
      "step": 15604
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5987260341644287,
      "learning_rate": 9.25704689939019e-06,
      "loss": 0.3603,
      "step": 15605
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.4319875240325928,
      "learning_rate": 9.255933415900112e-06,
      "loss": 0.7424,
      "step": 15606
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4210962057113647,
      "learning_rate": 9.254819941686587e-06,
      "loss": 0.5588,
      "step": 15607
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.793929934501648,
      "learning_rate": 9.253706476763505e-06,
      "loss": 0.5998,
      "step": 15608
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.857401967048645,
      "learning_rate": 9.252593021144742e-06,
      "loss": 0.5435,
      "step": 15609
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4317054748535156,
      "learning_rate": 9.251479574844184e-06,
      "loss": 0.5338,
      "step": 15610
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6295145750045776,
      "learning_rate": 9.25036613787571e-06,
      "loss": 0.5006,
      "step": 15611
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5449739694595337,
      "learning_rate": 9.249252710253203e-06,
      "loss": 0.5587,
      "step": 15612
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.681240200996399,
      "learning_rate": 9.248139291990543e-06,
      "loss": 0.5637,
      "step": 15613
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4177753925323486,
      "learning_rate": 9.247025883101616e-06,
      "loss": 0.4638,
      "step": 15614
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4466770887374878,
      "learning_rate": 9.245912483600297e-06,
      "loss": 0.576,
      "step": 15615
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5271611213684082,
      "learning_rate": 9.244799093500473e-06,
      "loss": 0.4611,
      "step": 15616
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6440033912658691,
      "learning_rate": 9.24368571281602e-06,
      "loss": 0.5868,
      "step": 15617
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6845228672027588,
      "learning_rate": 9.242572341560819e-06,
      "loss": 0.5636,
      "step": 15618
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4062448740005493,
      "learning_rate": 9.241458979748758e-06,
      "loss": 0.4375,
      "step": 15619
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.438558578491211,
      "learning_rate": 9.24034562739371e-06,
      "loss": 0.4108,
      "step": 15620
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2573959827423096,
      "learning_rate": 9.239232284509557e-06,
      "loss": 0.5547,
      "step": 15621
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.581470012664795,
      "learning_rate": 9.238118951110182e-06,
      "loss": 0.4693,
      "step": 15622
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.757352352142334,
      "learning_rate": 9.237005627209463e-06,
      "loss": 0.452,
      "step": 15623
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4096263647079468,
      "learning_rate": 9.235892312821282e-06,
      "loss": 0.5889,
      "step": 15624
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7179498672485352,
      "learning_rate": 9.234779007959517e-06,
      "loss": 0.5063,
      "step": 15625
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3702616691589355,
      "learning_rate": 9.233665712638052e-06,
      "loss": 0.4404,
      "step": 15626
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4690580368041992,
      "learning_rate": 9.232552426870764e-06,
      "loss": 0.4933,
      "step": 15627
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8541909456253052,
      "learning_rate": 9.23143915067153e-06,
      "loss": 0.4782,
      "step": 15628
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.317216157913208,
      "learning_rate": 9.230325884054235e-06,
      "loss": 0.446,
      "step": 15629
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.537847638130188,
      "learning_rate": 9.229212627032755e-06,
      "loss": 0.5765,
      "step": 15630
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8132959604263306,
      "learning_rate": 9.22809937962097e-06,
      "loss": 0.5685,
      "step": 15631
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8741440773010254,
      "learning_rate": 9.226986141832762e-06,
      "loss": 0.5693,
      "step": 15632
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.448771595954895,
      "learning_rate": 9.225872913682005e-06,
      "loss": 0.5539,
      "step": 15633
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.898818850517273,
      "learning_rate": 9.224759695182584e-06,
      "loss": 0.6154,
      "step": 15634
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3671215772628784,
      "learning_rate": 9.223646486348373e-06,
      "loss": 0.5284,
      "step": 15635
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3899282217025757,
      "learning_rate": 9.222533287193252e-06,
      "loss": 0.5147,
      "step": 15636
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.61601722240448,
      "learning_rate": 9.221420097731103e-06,
      "loss": 0.5729,
      "step": 15637
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8012514114379883,
      "learning_rate": 9.220306917975799e-06,
      "loss": 0.6021,
      "step": 15638
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4671447277069092,
      "learning_rate": 9.219193747941221e-06,
      "loss": 0.4899,
      "step": 15639
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6961437463760376,
      "learning_rate": 9.218080587641251e-06,
      "loss": 0.5002,
      "step": 15640
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.731881022453308,
      "learning_rate": 9.216967437089761e-06,
      "loss": 0.5633,
      "step": 15641
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.2326745986938477,
      "learning_rate": 9.215854296300634e-06,
      "loss": 0.5727,
      "step": 15642
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2989718914031982,
      "learning_rate": 9.214741165287743e-06,
      "loss": 0.4684,
      "step": 15643
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4342024326324463,
      "learning_rate": 9.21362804406497e-06,
      "loss": 0.4684,
      "step": 15644
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.764148235321045,
      "learning_rate": 9.212514932646193e-06,
      "loss": 0.5031,
      "step": 15645
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5854393243789673,
      "learning_rate": 9.211401831045286e-06,
      "loss": 0.6152,
      "step": 15646
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3299018144607544,
      "learning_rate": 9.210288739276129e-06,
      "loss": 0.4379,
      "step": 15647
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5268486738204956,
      "learning_rate": 9.209175657352599e-06,
      "loss": 0.5984,
      "step": 15648
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5696337223052979,
      "learning_rate": 9.20806258528857e-06,
      "loss": 0.5281,
      "step": 15649
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.42780339717865,
      "learning_rate": 9.206949523097926e-06,
      "loss": 0.403,
      "step": 15650
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4481174945831299,
      "learning_rate": 9.205836470794538e-06,
      "loss": 0.4913,
      "step": 15651
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.347712516784668,
      "learning_rate": 9.204723428392285e-06,
      "loss": 0.4517,
      "step": 15652
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.368830919265747,
      "learning_rate": 9.203610395905045e-06,
      "loss": 0.5459,
      "step": 15653
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4403098821640015,
      "learning_rate": 9.20249737334669e-06,
      "loss": 0.4655,
      "step": 15654
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7238520383834839,
      "learning_rate": 9.201384360731103e-06,
      "loss": 0.5922,
      "step": 15655
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4554126262664795,
      "learning_rate": 9.200271358072155e-06,
      "loss": 0.5116,
      "step": 15656
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3822412490844727,
      "learning_rate": 9.199158365383723e-06,
      "loss": 0.5575,
      "step": 15657
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6266040802001953,
      "learning_rate": 9.198045382679688e-06,
      "loss": 0.5439,
      "step": 15658
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7210057973861694,
      "learning_rate": 9.196932409973917e-06,
      "loss": 0.5391,
      "step": 15659
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.575336217880249,
      "learning_rate": 9.195819447280296e-06,
      "loss": 0.5458,
      "step": 15660
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6129251718521118,
      "learning_rate": 9.194706494612692e-06,
      "loss": 0.62,
      "step": 15661
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.553742527961731,
      "learning_rate": 9.193593551984985e-06,
      "loss": 0.5449,
      "step": 15662
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7843430042266846,
      "learning_rate": 9.192480619411055e-06,
      "loss": 0.5548,
      "step": 15663
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4122411012649536,
      "learning_rate": 9.191367696904766e-06,
      "loss": 0.4161,
      "step": 15664
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3203518390655518,
      "learning_rate": 9.190254784480001e-06,
      "loss": 0.5744,
      "step": 15665
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.641871452331543,
      "learning_rate": 9.189141882150635e-06,
      "loss": 0.5946,
      "step": 15666
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4895470142364502,
      "learning_rate": 9.18802898993054e-06,
      "loss": 0.5649,
      "step": 15667
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7091857194900513,
      "learning_rate": 9.186916107833595e-06,
      "loss": 0.535,
      "step": 15668
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3594743013381958,
      "learning_rate": 9.185803235873669e-06,
      "loss": 0.4727,
      "step": 15669
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3535549640655518,
      "learning_rate": 9.18469037406464e-06,
      "loss": 0.6011,
      "step": 15670
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7799969911575317,
      "learning_rate": 9.183577522420384e-06,
      "loss": 0.5459,
      "step": 15671
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5013442039489746,
      "learning_rate": 9.18246468095477e-06,
      "loss": 0.5307,
      "step": 15672
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.50554358959198,
      "learning_rate": 9.18135184968168e-06,
      "loss": 0.5254,
      "step": 15673
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6461318731307983,
      "learning_rate": 9.180239028614981e-06,
      "loss": 0.5797,
      "step": 15674
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7033201456069946,
      "learning_rate": 9.17912621776855e-06,
      "loss": 0.5462,
      "step": 15675
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.405218482017517,
      "learning_rate": 9.178013417156262e-06,
      "loss": 0.4831,
      "step": 15676
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6078248023986816,
      "learning_rate": 9.176900626791987e-06,
      "loss": 0.5305,
      "step": 15677
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.9877464771270752,
      "learning_rate": 9.175787846689601e-06,
      "loss": 0.4621,
      "step": 15678
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5486619472503662,
      "learning_rate": 9.17467507686298e-06,
      "loss": 0.5886,
      "step": 15679
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3048434257507324,
      "learning_rate": 9.173562317325992e-06,
      "loss": 0.4573,
      "step": 15680
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.915683627128601,
      "learning_rate": 9.172449568092515e-06,
      "loss": 0.5983,
      "step": 15681
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4777657985687256,
      "learning_rate": 9.171336829176419e-06,
      "loss": 0.5298,
      "step": 15682
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5947308540344238,
      "learning_rate": 9.170224100591576e-06,
      "loss": 0.4875,
      "step": 15683
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6173745393753052,
      "learning_rate": 9.169111382351863e-06,
      "loss": 0.4053,
      "step": 15684
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4984228610992432,
      "learning_rate": 9.167998674471149e-06,
      "loss": 0.5753,
      "step": 15685
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3220617771148682,
      "learning_rate": 9.16688597696331e-06,
      "loss": 0.5659,
      "step": 15686
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6116042137145996,
      "learning_rate": 9.165773289842212e-06,
      "loss": 0.6075,
      "step": 15687
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5266227722167969,
      "learning_rate": 9.164660613121734e-06,
      "loss": 0.4075,
      "step": 15688
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3929136991500854,
      "learning_rate": 9.163547946815746e-06,
      "loss": 0.4924,
      "step": 15689
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4996285438537598,
      "learning_rate": 9.162435290938117e-06,
      "loss": 0.5844,
      "step": 15690
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5150656700134277,
      "learning_rate": 9.161322645502726e-06,
      "loss": 0.5854,
      "step": 15691
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5878996849060059,
      "learning_rate": 9.160210010523437e-06,
      "loss": 0.4644,
      "step": 15692
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.468340516090393,
      "learning_rate": 9.159097386014126e-06,
      "loss": 0.5023,
      "step": 15693
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6744475364685059,
      "learning_rate": 9.157984771988665e-06,
      "loss": 0.609,
      "step": 15694
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3836454153060913,
      "learning_rate": 9.15687216846092e-06,
      "loss": 0.5149,
      "step": 15695
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.528191328048706,
      "learning_rate": 9.155759575444768e-06,
      "loss": 0.567,
      "step": 15696
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4845229387283325,
      "learning_rate": 9.15464699295408e-06,
      "loss": 0.5743,
      "step": 15697
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.826988935470581,
      "learning_rate": 9.153534421002723e-06,
      "loss": 0.6326,
      "step": 15698
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.681878685951233,
      "learning_rate": 9.152421859604573e-06,
      "loss": 0.5885,
      "step": 15699
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7756702899932861,
      "learning_rate": 9.151309308773495e-06,
      "loss": 0.4587,
      "step": 15700
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.546974539756775,
      "learning_rate": 9.150196768523362e-06,
      "loss": 0.5553,
      "step": 15701
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5950721502304077,
      "learning_rate": 9.149084238868048e-06,
      "loss": 0.5342,
      "step": 15702
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7316876649856567,
      "learning_rate": 9.147971719821418e-06,
      "loss": 0.5089,
      "step": 15703
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.8847770690917969,
      "learning_rate": 9.146859211397345e-06,
      "loss": 0.494,
      "step": 15704
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.1208767890930176,
      "learning_rate": 9.145746713609698e-06,
      "loss": 0.5195,
      "step": 15705
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3881365060806274,
      "learning_rate": 9.144634226472346e-06,
      "loss": 0.4765,
      "step": 15706
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6999529600143433,
      "learning_rate": 9.143521749999164e-06,
      "loss": 0.4773,
      "step": 15707
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.674681544303894,
      "learning_rate": 9.142409284204015e-06,
      "loss": 0.5278,
      "step": 15708
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.9315356016159058,
      "learning_rate": 9.14129682910077e-06,
      "loss": 0.3754,
      "step": 15709
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.7000019550323486,
      "learning_rate": 9.140184384703304e-06,
      "loss": 0.5486,
      "step": 15710
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5653053522109985,
      "learning_rate": 9.139071951025477e-06,
      "loss": 0.5396,
      "step": 15711
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.555695652961731,
      "learning_rate": 9.137959528081167e-06,
      "loss": 0.501,
      "step": 15712
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5762934684753418,
      "learning_rate": 9.136847115884238e-06,
      "loss": 0.5315,
      "step": 15713
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.046337842941284,
      "learning_rate": 9.135734714448558e-06,
      "loss": 0.591,
      "step": 15714
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.362225890159607,
      "learning_rate": 9.134622323788001e-06,
      "loss": 0.5498,
      "step": 15715
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5707260370254517,
      "learning_rate": 9.13350994391643e-06,
      "loss": 0.5305,
      "step": 15716
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4852559566497803,
      "learning_rate": 9.132397574847718e-06,
      "loss": 0.4226,
      "step": 15717
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3703900575637817,
      "learning_rate": 9.13128521659573e-06,
      "loss": 0.4699,
      "step": 15718
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9323252439498901,
      "learning_rate": 9.130172869174332e-06,
      "loss": 0.52,
      "step": 15719
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.363741159439087,
      "learning_rate": 9.1290605325974e-06,
      "loss": 0.5627,
      "step": 15720
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3542828559875488,
      "learning_rate": 9.127948206878794e-06,
      "loss": 0.4654,
      "step": 15721
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4839938879013062,
      "learning_rate": 9.126835892032387e-06,
      "loss": 0.5074,
      "step": 15722
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6173909902572632,
      "learning_rate": 9.125723588072044e-06,
      "loss": 0.6106,
      "step": 15723
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9791423082351685,
      "learning_rate": 9.124611295011632e-06,
      "loss": 0.6141,
      "step": 15724
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5208743810653687,
      "learning_rate": 9.123499012865023e-06,
      "loss": 0.7003,
      "step": 15725
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6448956727981567,
      "learning_rate": 9.122386741646078e-06,
      "loss": 0.5488,
      "step": 15726
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8890674114227295,
      "learning_rate": 9.121274481368665e-06,
      "loss": 0.531,
      "step": 15727
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8163833618164062,
      "learning_rate": 9.120162232046657e-06,
      "loss": 0.5294,
      "step": 15728
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6775765419006348,
      "learning_rate": 9.119049993693915e-06,
      "loss": 0.6144,
      "step": 15729
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8220391273498535,
      "learning_rate": 9.117937766324308e-06,
      "loss": 0.5513,
      "step": 15730
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4368361234664917,
      "learning_rate": 9.116825549951702e-06,
      "loss": 0.485,
      "step": 15731
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8242082595825195,
      "learning_rate": 9.115713344589961e-06,
      "loss": 0.503,
      "step": 15732
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2996010780334473,
      "learning_rate": 9.114601150252957e-06,
      "loss": 0.4804,
      "step": 15733
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.467695951461792,
      "learning_rate": 9.11348896695455e-06,
      "loss": 0.4908,
      "step": 15734
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4289915561676025,
      "learning_rate": 9.112376794708609e-06,
      "loss": 0.5679,
      "step": 15735
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5955506563186646,
      "learning_rate": 9.111264633529003e-06,
      "loss": 0.5694,
      "step": 15736
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7390683889389038,
      "learning_rate": 9.11015248342959e-06,
      "loss": 0.4677,
      "step": 15737
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9822412729263306,
      "learning_rate": 9.109040344424243e-06,
      "loss": 0.5547,
      "step": 15738
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.2605602741241455,
      "learning_rate": 9.107928216526824e-06,
      "loss": 0.5985,
      "step": 15739
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5774588584899902,
      "learning_rate": 9.106816099751197e-06,
      "loss": 0.6104,
      "step": 15740
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3714187145233154,
      "learning_rate": 9.10570399411123e-06,
      "loss": 0.5249,
      "step": 15741
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.908508539199829,
      "learning_rate": 9.104591899620787e-06,
      "loss": 0.6553,
      "step": 15742
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3336074352264404,
      "learning_rate": 9.103479816293735e-06,
      "loss": 0.4674,
      "step": 15743
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5213452577590942,
      "learning_rate": 9.102367744143932e-06,
      "loss": 0.5426,
      "step": 15744
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5631359815597534,
      "learning_rate": 9.10125568318525e-06,
      "loss": 0.482,
      "step": 15745
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6119199991226196,
      "learning_rate": 9.100143633431552e-06,
      "loss": 0.478,
      "step": 15746
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3715910911560059,
      "learning_rate": 9.099031594896698e-06,
      "loss": 0.539,
      "step": 15747
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5687378644943237,
      "learning_rate": 9.097919567594557e-06,
      "loss": 0.4991,
      "step": 15748
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.245915412902832,
      "learning_rate": 9.096807551538991e-06,
      "loss": 0.474,
      "step": 15749
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7528330087661743,
      "learning_rate": 9.095695546743865e-06,
      "loss": 0.5791,
      "step": 15750
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3841516971588135,
      "learning_rate": 9.094583553223042e-06,
      "loss": 0.52,
      "step": 15751
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.418596863746643,
      "learning_rate": 9.093471570990384e-06,
      "loss": 0.5452,
      "step": 15752
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7853903770446777,
      "learning_rate": 9.092359600059757e-06,
      "loss": 0.5263,
      "step": 15753
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6484159231185913,
      "learning_rate": 9.091247640445024e-06,
      "loss": 0.5493,
      "step": 15754
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2057812213897705,
      "learning_rate": 9.090135692160048e-06,
      "loss": 0.4801,
      "step": 15755
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6413328647613525,
      "learning_rate": 9.089023755218692e-06,
      "loss": 0.5658,
      "step": 15756
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9173810482025146,
      "learning_rate": 9.087911829634818e-06,
      "loss": 0.5434,
      "step": 15757
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6500946283340454,
      "learning_rate": 9.086799915422289e-06,
      "loss": 0.5581,
      "step": 15758
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6094105243682861,
      "learning_rate": 9.08568801259497e-06,
      "loss": 0.6256,
      "step": 15759
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3921198844909668,
      "learning_rate": 9.08457612116672e-06,
      "loss": 0.5344,
      "step": 15760
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6004472970962524,
      "learning_rate": 9.083464241151403e-06,
      "loss": 0.5171,
      "step": 15761
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2714598178863525,
      "learning_rate": 9.082352372562883e-06,
      "loss": 0.5753,
      "step": 15762
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7828096151351929,
      "learning_rate": 9.081240515415018e-06,
      "loss": 0.5169,
      "step": 15763
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4860568046569824,
      "learning_rate": 9.080128669721677e-06,
      "loss": 0.5579,
      "step": 15764
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.501150369644165,
      "learning_rate": 9.079016835496712e-06,
      "loss": 0.5119,
      "step": 15765
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2925375699996948,
      "learning_rate": 9.077905012753992e-06,
      "loss": 0.4086,
      "step": 15766
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.609365463256836,
      "learning_rate": 9.076793201507376e-06,
      "loss": 0.5752,
      "step": 15767
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7255523204803467,
      "learning_rate": 9.075681401770725e-06,
      "loss": 0.4984,
      "step": 15768
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7249821424484253,
      "learning_rate": 9.074569613557903e-06,
      "loss": 0.5374,
      "step": 15769
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.678310513496399,
      "learning_rate": 9.073457836882768e-06,
      "loss": 0.4718,
      "step": 15770
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4393213987350464,
      "learning_rate": 9.07234607175918e-06,
      "loss": 0.5127,
      "step": 15771
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.182692527770996,
      "learning_rate": 9.071234318201005e-06,
      "loss": 0.5948,
      "step": 15772
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8028701543807983,
      "learning_rate": 9.070122576222097e-06,
      "loss": 0.5542,
      "step": 15773
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.9792604446411133,
      "learning_rate": 9.069010845836322e-06,
      "loss": 0.5049,
      "step": 15774
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6585144996643066,
      "learning_rate": 9.067899127057538e-06,
      "loss": 0.5503,
      "step": 15775
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5055310726165771,
      "learning_rate": 9.066787419899604e-06,
      "loss": 0.6063,
      "step": 15776
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6430647373199463,
      "learning_rate": 9.065675724376383e-06,
      "loss": 0.5536,
      "step": 15777
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.0733277797698975,
      "learning_rate": 9.064564040501734e-06,
      "loss": 0.4511,
      "step": 15778
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5695807933807373,
      "learning_rate": 9.063452368289513e-06,
      "loss": 0.6481,
      "step": 15779
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3574670553207397,
      "learning_rate": 9.062340707753587e-06,
      "loss": 0.5656,
      "step": 15780
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.157824993133545,
      "learning_rate": 9.061229058907807e-06,
      "loss": 0.6171,
      "step": 15781
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8234597444534302,
      "learning_rate": 9.06011742176604e-06,
      "loss": 0.5606,
      "step": 15782
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4802887439727783,
      "learning_rate": 9.059005796342141e-06,
      "loss": 0.5282,
      "step": 15783
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.383028507232666,
      "learning_rate": 9.05789418264997e-06,
      "loss": 0.5458,
      "step": 15784
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4042974710464478,
      "learning_rate": 9.056782580703386e-06,
      "loss": 0.4468,
      "step": 15785
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8560549020767212,
      "learning_rate": 9.055670990516247e-06,
      "loss": 0.5665,
      "step": 15786
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5773000717163086,
      "learning_rate": 9.054559412102412e-06,
      "loss": 0.482,
      "step": 15787
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6306538581848145,
      "learning_rate": 9.05344784547574e-06,
      "loss": 0.5194,
      "step": 15788
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.900206446647644,
      "learning_rate": 9.05233629065009e-06,
      "loss": 0.554,
      "step": 15789
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8260432481765747,
      "learning_rate": 9.05122474763932e-06,
      "loss": 0.5512,
      "step": 15790
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.561233639717102,
      "learning_rate": 9.050113216457285e-06,
      "loss": 0.5894,
      "step": 15791
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4508216381072998,
      "learning_rate": 9.049001697117843e-06,
      "loss": 0.5481,
      "step": 15792
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5661534070968628,
      "learning_rate": 9.047890189634859e-06,
      "loss": 0.6107,
      "step": 15793
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5002801418304443,
      "learning_rate": 9.04677869402218e-06,
      "loss": 0.5055,
      "step": 15794
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4432812929153442,
      "learning_rate": 9.045667210293675e-06,
      "loss": 0.5247,
      "step": 15795
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.733784794807434,
      "learning_rate": 9.04455573846319e-06,
      "loss": 0.5887,
      "step": 15796
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3340500593185425,
      "learning_rate": 9.043444278544588e-06,
      "loss": 0.4827,
      "step": 15797
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7749379873275757,
      "learning_rate": 9.042332830551728e-06,
      "loss": 0.5319,
      "step": 15798
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6321357488632202,
      "learning_rate": 9.041221394498461e-06,
      "loss": 0.5505,
      "step": 15799
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.343214988708496,
      "learning_rate": 9.040109970398647e-06,
      "loss": 0.4795,
      "step": 15800
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5116766691207886,
      "learning_rate": 9.038998558266145e-06,
      "loss": 0.4424,
      "step": 15801
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.51511812210083,
      "learning_rate": 9.037887158114805e-06,
      "loss": 0.4853,
      "step": 15802
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6082391738891602,
      "learning_rate": 9.03677576995849e-06,
      "loss": 0.4731,
      "step": 15803
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4202343225479126,
      "learning_rate": 9.03566439381105e-06,
      "loss": 0.5511,
      "step": 15804
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.7096554040908813,
      "learning_rate": 9.034553029686346e-06,
      "loss": 0.6266,
      "step": 15805
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6831114292144775,
      "learning_rate": 9.033441677598232e-06,
      "loss": 0.5179,
      "step": 15806
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4290558099746704,
      "learning_rate": 9.032330337560561e-06,
      "loss": 0.4811,
      "step": 15807
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5786839723587036,
      "learning_rate": 9.031219009587196e-06,
      "loss": 0.5281,
      "step": 15808
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4238460063934326,
      "learning_rate": 9.030107693691984e-06,
      "loss": 0.5583,
      "step": 15809
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5990231037139893,
      "learning_rate": 9.028996389888781e-06,
      "loss": 0.6433,
      "step": 15810
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4631831645965576,
      "learning_rate": 9.027885098191448e-06,
      "loss": 0.6392,
      "step": 15811
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5621838569641113,
      "learning_rate": 9.026773818613834e-06,
      "loss": 0.5165,
      "step": 15812
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3723421096801758,
      "learning_rate": 9.025662551169797e-06,
      "loss": 0.5212,
      "step": 15813
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.605668306350708,
      "learning_rate": 9.024551295873192e-06,
      "loss": 0.6497,
      "step": 15814
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6525626182556152,
      "learning_rate": 9.023440052737869e-06,
      "loss": 0.5628,
      "step": 15815
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5462929010391235,
      "learning_rate": 9.02232882177769e-06,
      "loss": 0.5885,
      "step": 15816
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6534357070922852,
      "learning_rate": 9.021217603006502e-06,
      "loss": 0.5222,
      "step": 15817
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4553136825561523,
      "learning_rate": 9.02010639643816e-06,
      "loss": 0.433,
      "step": 15818
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1948096752166748,
      "learning_rate": 9.018995202086524e-06,
      "loss": 0.4677,
      "step": 15819
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7418824434280396,
      "learning_rate": 9.01788401996544e-06,
      "loss": 0.5515,
      "step": 15820
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6862138509750366,
      "learning_rate": 9.016772850088764e-06,
      "loss": 0.4844,
      "step": 15821
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6523172855377197,
      "learning_rate": 9.01566169247035e-06,
      "loss": 0.5647,
      "step": 15822
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7459080219268799,
      "learning_rate": 9.014550547124053e-06,
      "loss": 0.4581,
      "step": 15823
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2351102828979492,
      "learning_rate": 9.013439414063722e-06,
      "loss": 0.4863,
      "step": 15824
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4784797430038452,
      "learning_rate": 9.012328293303213e-06,
      "loss": 0.5272,
      "step": 15825
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7791396379470825,
      "learning_rate": 9.011217184856378e-06,
      "loss": 0.4799,
      "step": 15826
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4671454429626465,
      "learning_rate": 9.010106088737067e-06,
      "loss": 0.5098,
      "step": 15827
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5505506992340088,
      "learning_rate": 9.008995004959138e-06,
      "loss": 0.5178,
      "step": 15828
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3317092657089233,
      "learning_rate": 9.00788393353644e-06,
      "loss": 0.5046,
      "step": 15829
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.741216778755188,
      "learning_rate": 9.006772874482825e-06,
      "loss": 0.5289,
      "step": 15830
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.353622555732727,
      "learning_rate": 9.005661827812144e-06,
      "loss": 0.5068,
      "step": 15831
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3490482568740845,
      "learning_rate": 9.004550793538251e-06,
      "loss": 0.5294,
      "step": 15832
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5167441368103027,
      "learning_rate": 9.003439771674998e-06,
      "loss": 0.5321,
      "step": 15833
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5553995370864868,
      "learning_rate": 9.002328762236235e-06,
      "loss": 0.4195,
      "step": 15834
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7825177907943726,
      "learning_rate": 9.001217765235812e-06,
      "loss": 0.6206,
      "step": 15835
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6600122451782227,
      "learning_rate": 9.000106780687584e-06,
      "loss": 0.6885,
      "step": 15836
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5751971006393433,
      "learning_rate": 8.998995808605398e-06,
      "loss": 0.5292,
      "step": 15837
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4806854724884033,
      "learning_rate": 8.99788484900311e-06,
      "loss": 0.5308,
      "step": 15838
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3526115417480469,
      "learning_rate": 8.996773901894566e-06,
      "loss": 0.4155,
      "step": 15839
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4295868873596191,
      "learning_rate": 8.995662967293616e-06,
      "loss": 0.4847,
      "step": 15840
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3806071281433105,
      "learning_rate": 8.994552045214118e-06,
      "loss": 0.4503,
      "step": 15841
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.1072733402252197,
      "learning_rate": 8.993441135669914e-06,
      "loss": 0.7274,
      "step": 15842
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7668483257293701,
      "learning_rate": 8.992330238674857e-06,
      "loss": 0.5723,
      "step": 15843
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4778085947036743,
      "learning_rate": 8.991219354242797e-06,
      "loss": 0.566,
      "step": 15844
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6573199033737183,
      "learning_rate": 8.990108482387583e-06,
      "loss": 0.5632,
      "step": 15845
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.71305513381958,
      "learning_rate": 8.988997623123069e-06,
      "loss": 0.5325,
      "step": 15846
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6505147218704224,
      "learning_rate": 8.987886776463098e-06,
      "loss": 0.538,
      "step": 15847
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.625179409980774,
      "learning_rate": 8.986775942421524e-06,
      "loss": 0.5771,
      "step": 15848
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6203910112380981,
      "learning_rate": 8.985665121012194e-06,
      "loss": 0.4941,
      "step": 15849
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3103286027908325,
      "learning_rate": 8.984554312248958e-06,
      "loss": 0.489,
      "step": 15850
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.657408595085144,
      "learning_rate": 8.983443516145666e-06,
      "loss": 0.5114,
      "step": 15851
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5988322496414185,
      "learning_rate": 8.98233273271616e-06,
      "loss": 0.4714,
      "step": 15852
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4310747385025024,
      "learning_rate": 8.981221961974297e-06,
      "loss": 0.547,
      "step": 15853
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7697484493255615,
      "learning_rate": 8.980111203933925e-06,
      "loss": 0.6383,
      "step": 15854
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5351710319519043,
      "learning_rate": 8.979000458608886e-06,
      "loss": 0.4806,
      "step": 15855
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3809312582015991,
      "learning_rate": 8.977889726013034e-06,
      "loss": 0.5727,
      "step": 15856
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5488481521606445,
      "learning_rate": 8.976779006160213e-06,
      "loss": 0.5863,
      "step": 15857
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4653204679489136,
      "learning_rate": 8.975668299064271e-06,
      "loss": 0.533,
      "step": 15858
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6382968425750732,
      "learning_rate": 8.974557604739059e-06,
      "loss": 0.5034,
      "step": 15859
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4722707271575928,
      "learning_rate": 8.97344692319842e-06,
      "loss": 0.4928,
      "step": 15860
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7607309818267822,
      "learning_rate": 8.972336254456203e-06,
      "loss": 0.4589,
      "step": 15861
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6294766664505005,
      "learning_rate": 8.971225598526259e-06,
      "loss": 0.5433,
      "step": 15862
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.9637413024902344,
      "learning_rate": 8.970114955422428e-06,
      "loss": 0.4394,
      "step": 15863
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6086647510528564,
      "learning_rate": 8.969004325158565e-06,
      "loss": 0.595,
      "step": 15864
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.0163962841033936,
      "learning_rate": 8.967893707748508e-06,
      "loss": 0.7218,
      "step": 15865
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3484458923339844,
      "learning_rate": 8.966783103206108e-06,
      "loss": 0.5226,
      "step": 15866
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3775672912597656,
      "learning_rate": 8.965672511545213e-06,
      "loss": 0.548,
      "step": 15867
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5721772909164429,
      "learning_rate": 8.964561932779665e-06,
      "loss": 0.496,
      "step": 15868
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2853715419769287,
      "learning_rate": 8.963451366923313e-06,
      "loss": 0.5175,
      "step": 15869
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5251834392547607,
      "learning_rate": 8.962340813990002e-06,
      "loss": 0.4541,
      "step": 15870
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5827044248580933,
      "learning_rate": 8.961230273993575e-06,
      "loss": 0.5318,
      "step": 15871
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4392997026443481,
      "learning_rate": 8.960119746947884e-06,
      "loss": 0.5,
      "step": 15872
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5214203596115112,
      "learning_rate": 8.959009232866769e-06,
      "loss": 0.5177,
      "step": 15873
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3941733837127686,
      "learning_rate": 8.957898731764075e-06,
      "loss": 0.4991,
      "step": 15874
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7845308780670166,
      "learning_rate": 8.956788243653651e-06,
      "loss": 0.5822,
      "step": 15875
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.651060938835144,
      "learning_rate": 8.955677768549338e-06,
      "loss": 0.5076,
      "step": 15876
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4358994960784912,
      "learning_rate": 8.954567306464983e-06,
      "loss": 0.6177,
      "step": 15877
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4533207416534424,
      "learning_rate": 8.953456857414428e-06,
      "loss": 0.5269,
      "step": 15878
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6637263298034668,
      "learning_rate": 8.95234642141152e-06,
      "loss": 0.5781,
      "step": 15879
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.438638687133789,
      "learning_rate": 8.951235998470104e-06,
      "loss": 0.4576,
      "step": 15880
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7244282960891724,
      "learning_rate": 8.950125588604021e-06,
      "loss": 0.5333,
      "step": 15881
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7055414915084839,
      "learning_rate": 8.949015191827119e-06,
      "loss": 0.4888,
      "step": 15882
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.433529257774353,
      "learning_rate": 8.947904808153234e-06,
      "loss": 0.4206,
      "step": 15883
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.1891584396362305,
      "learning_rate": 8.946794437596219e-06,
      "loss": 0.7023,
      "step": 15884
    },
    {
      "epoch": 1.64,
      "grad_norm": 4.012714385986328,
      "learning_rate": 8.945684080169911e-06,
      "loss": 0.4244,
      "step": 15885
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7160063982009888,
      "learning_rate": 8.944573735888156e-06,
      "loss": 0.593,
      "step": 15886
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.496182918548584,
      "learning_rate": 8.943463404764794e-06,
      "loss": 0.4949,
      "step": 15887
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6792902946472168,
      "learning_rate": 8.942353086813675e-06,
      "loss": 0.4278,
      "step": 15888
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.516236662864685,
      "learning_rate": 8.941242782048634e-06,
      "loss": 0.5449,
      "step": 15889
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7809807062149048,
      "learning_rate": 8.940132490483517e-06,
      "loss": 0.5255,
      "step": 15890
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7685407400131226,
      "learning_rate": 8.939022212132165e-06,
      "loss": 0.5664,
      "step": 15891
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5985651016235352,
      "learning_rate": 8.937911947008424e-06,
      "loss": 0.5146,
      "step": 15892
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.7255128622055054,
      "learning_rate": 8.936801695126131e-06,
      "loss": 0.5312,
      "step": 15893
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.624864101409912,
      "learning_rate": 8.935691456499132e-06,
      "loss": 0.5499,
      "step": 15894
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3420506715774536,
      "learning_rate": 8.934581231141267e-06,
      "loss": 0.5298,
      "step": 15895
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6296268701553345,
      "learning_rate": 8.933471019066377e-06,
      "loss": 0.4881,
      "step": 15896
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.633637547492981,
      "learning_rate": 8.932360820288302e-06,
      "loss": 0.4944,
      "step": 15897
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.5532763004302979,
      "learning_rate": 8.931250634820889e-06,
      "loss": 0.5485,
      "step": 15898
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4419257640838623,
      "learning_rate": 8.930140462677974e-06,
      "loss": 0.5641,
      "step": 15899
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.8654230833053589,
      "learning_rate": 8.929030303873398e-06,
      "loss": 0.4949,
      "step": 15900
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4181904792785645,
      "learning_rate": 8.927920158421005e-06,
      "loss": 0.5374,
      "step": 15901
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.520288109779358,
      "learning_rate": 8.926810026334633e-06,
      "loss": 0.454,
      "step": 15902
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4709947109222412,
      "learning_rate": 8.925699907628125e-06,
      "loss": 0.5816,
      "step": 15903
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.474865436553955,
      "learning_rate": 8.924589802315316e-06,
      "loss": 0.3642,
      "step": 15904
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.581155776977539,
      "learning_rate": 8.92347971041005e-06,
      "loss": 0.598,
      "step": 15905
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.44467294216156,
      "learning_rate": 8.92236963192617e-06,
      "loss": 0.4975,
      "step": 15906
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.397207498550415,
      "learning_rate": 8.92125956687751e-06,
      "loss": 0.497,
      "step": 15907
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.46156907081604,
      "learning_rate": 8.920149515277915e-06,
      "loss": 0.6254,
      "step": 15908
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4751454591751099,
      "learning_rate": 8.919039477141217e-06,
      "loss": 0.5486,
      "step": 15909
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6981425285339355,
      "learning_rate": 8.917929452481259e-06,
      "loss": 0.5915,
      "step": 15910
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6826379299163818,
      "learning_rate": 8.916819441311885e-06,
      "loss": 0.5333,
      "step": 15911
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.387908697128296,
      "learning_rate": 8.915709443646926e-06,
      "loss": 0.6086,
      "step": 15912
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.616810917854309,
      "learning_rate": 8.914599459500225e-06,
      "loss": 0.5987,
      "step": 15913
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.775451898574829,
      "learning_rate": 8.913489488885623e-06,
      "loss": 0.6133,
      "step": 15914
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4730019569396973,
      "learning_rate": 8.912379531816951e-06,
      "loss": 0.4912,
      "step": 15915
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5969696044921875,
      "learning_rate": 8.911269588308056e-06,
      "loss": 0.4117,
      "step": 15916
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5093227624893188,
      "learning_rate": 8.910159658372768e-06,
      "loss": 0.4825,
      "step": 15917
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2375164031982422,
      "learning_rate": 8.909049742024929e-06,
      "loss": 0.4337,
      "step": 15918
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6983329057693481,
      "learning_rate": 8.907939839278377e-06,
      "loss": 0.5332,
      "step": 15919
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4162335395812988,
      "learning_rate": 8.906829950146949e-06,
      "loss": 0.508,
      "step": 15920
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5340688228607178,
      "learning_rate": 8.905720074644484e-06,
      "loss": 0.5334,
      "step": 15921
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4814138412475586,
      "learning_rate": 8.904610212784814e-06,
      "loss": 0.5751,
      "step": 15922
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.0121936798095703,
      "learning_rate": 8.903500364581782e-06,
      "loss": 0.5247,
      "step": 15923
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5136650800704956,
      "learning_rate": 8.902390530049223e-06,
      "loss": 0.4299,
      "step": 15924
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5183467864990234,
      "learning_rate": 8.90128070920097e-06,
      "loss": 0.5194,
      "step": 15925
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3803472518920898,
      "learning_rate": 8.900170902050866e-06,
      "loss": 0.4931,
      "step": 15926
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.15299391746521,
      "learning_rate": 8.899061108612744e-06,
      "loss": 0.6902,
      "step": 15927
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2826201915740967,
      "learning_rate": 8.89795132890044e-06,
      "loss": 0.5319,
      "step": 15928
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.687013030052185,
      "learning_rate": 8.89684156292779e-06,
      "loss": 0.63,
      "step": 15929
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.901827096939087,
      "learning_rate": 8.89573181070863e-06,
      "loss": 0.621,
      "step": 15930
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3534067869186401,
      "learning_rate": 8.894622072256796e-06,
      "loss": 0.3905,
      "step": 15931
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3419392108917236,
      "learning_rate": 8.893512347586124e-06,
      "loss": 0.4877,
      "step": 15932
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7130179405212402,
      "learning_rate": 8.892402636710448e-06,
      "loss": 0.5861,
      "step": 15933
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5780209302902222,
      "learning_rate": 8.891292939643607e-06,
      "loss": 0.5019,
      "step": 15934
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.00374174118042,
      "learning_rate": 8.89018325639943e-06,
      "loss": 0.5194,
      "step": 15935
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3613964319229126,
      "learning_rate": 8.889073586991754e-06,
      "loss": 0.4778,
      "step": 15936
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.377073884010315,
      "learning_rate": 8.887963931434418e-06,
      "loss": 0.516,
      "step": 15937
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4396753311157227,
      "learning_rate": 8.88685428974125e-06,
      "loss": 0.3976,
      "step": 15938
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.415863037109375,
      "learning_rate": 8.885744661926088e-06,
      "loss": 0.5276,
      "step": 15939
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.716611385345459,
      "learning_rate": 8.884635048002767e-06,
      "loss": 0.6015,
      "step": 15940
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.640318751335144,
      "learning_rate": 8.883525447985118e-06,
      "loss": 0.5262,
      "step": 15941
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.576130986213684,
      "learning_rate": 8.882415861886979e-06,
      "loss": 0.4836,
      "step": 15942
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5771198272705078,
      "learning_rate": 8.881306289722179e-06,
      "loss": 0.5494,
      "step": 15943
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.7388139963150024,
      "learning_rate": 8.88019673150455e-06,
      "loss": 0.5822,
      "step": 15944
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6434611082077026,
      "learning_rate": 8.879087187247935e-06,
      "loss": 0.508,
      "step": 15945
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4876686334609985,
      "learning_rate": 8.877977656966156e-06,
      "loss": 0.4625,
      "step": 15946
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.65024733543396,
      "learning_rate": 8.876868140673054e-06,
      "loss": 0.5273,
      "step": 15947
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3792316913604736,
      "learning_rate": 8.875758638382456e-06,
      "loss": 0.5697,
      "step": 15948
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8640917539596558,
      "learning_rate": 8.874649150108198e-06,
      "loss": 0.4933,
      "step": 15949
    },
    {
      "epoch": 1.65,
      "grad_norm": 7.552230358123779,
      "learning_rate": 8.873539675864112e-06,
      "loss": 0.7642,
      "step": 15950
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1857693195343018,
      "learning_rate": 8.872430215664028e-06,
      "loss": 0.5114,
      "step": 15951
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.0173354148864746,
      "learning_rate": 8.871320769521783e-06,
      "loss": 0.521,
      "step": 15952
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.650738000869751,
      "learning_rate": 8.870211337451203e-06,
      "loss": 0.6881,
      "step": 15953
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4648799896240234,
      "learning_rate": 8.869101919466122e-06,
      "loss": 0.4506,
      "step": 15954
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.581607460975647,
      "learning_rate": 8.867992515580373e-06,
      "loss": 0.5154,
      "step": 15955
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6985266208648682,
      "learning_rate": 8.866883125807784e-06,
      "loss": 0.6074,
      "step": 15956
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.9580893516540527,
      "learning_rate": 8.865773750162189e-06,
      "loss": 0.6042,
      "step": 15957
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4510451555252075,
      "learning_rate": 8.86466438865742e-06,
      "loss": 0.5735,
      "step": 15958
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5063706636428833,
      "learning_rate": 8.863555041307306e-06,
      "loss": 0.4949,
      "step": 15959
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.9117735624313354,
      "learning_rate": 8.862445708125678e-06,
      "loss": 0.7029,
      "step": 15960
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6165251731872559,
      "learning_rate": 8.861336389126365e-06,
      "loss": 0.4927,
      "step": 15961
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4770894050598145,
      "learning_rate": 8.860227084323198e-06,
      "loss": 0.4677,
      "step": 15962
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6576118469238281,
      "learning_rate": 8.85911779373001e-06,
      "loss": 0.5824,
      "step": 15963
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3238108158111572,
      "learning_rate": 8.858008517360629e-06,
      "loss": 0.5175,
      "step": 15964
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5528697967529297,
      "learning_rate": 8.856899255228883e-06,
      "loss": 0.5298,
      "step": 15965
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.228380560874939,
      "learning_rate": 8.855790007348602e-06,
      "loss": 0.4359,
      "step": 15966
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5601264238357544,
      "learning_rate": 8.854680773733617e-06,
      "loss": 0.5095,
      "step": 15967
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4288278818130493,
      "learning_rate": 8.85357155439776e-06,
      "loss": 0.5898,
      "step": 15968
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3606115579605103,
      "learning_rate": 8.852462349354853e-06,
      "loss": 0.5962,
      "step": 15969
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6780751943588257,
      "learning_rate": 8.851353158618728e-06,
      "loss": 0.6313,
      "step": 15970
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3513983488082886,
      "learning_rate": 8.850243982203219e-06,
      "loss": 0.5651,
      "step": 15971
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2214699983596802,
      "learning_rate": 8.849134820122146e-06,
      "loss": 0.4076,
      "step": 15972
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.537604808807373,
      "learning_rate": 8.848025672389342e-06,
      "loss": 0.6243,
      "step": 15973
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.677998661994934,
      "learning_rate": 8.846916539018632e-06,
      "loss": 0.5881,
      "step": 15974
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8310681581497192,
      "learning_rate": 8.845807420023849e-06,
      "loss": 0.6359,
      "step": 15975
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.9994699954986572,
      "learning_rate": 8.84469831541882e-06,
      "loss": 0.6909,
      "step": 15976
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.532823920249939,
      "learning_rate": 8.843589225217368e-06,
      "loss": 0.6041,
      "step": 15977
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4775938987731934,
      "learning_rate": 8.842480149433323e-06,
      "loss": 0.4915,
      "step": 15978
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.559955358505249,
      "learning_rate": 8.841371088080512e-06,
      "loss": 0.5316,
      "step": 15979
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.8041014671325684,
      "learning_rate": 8.840262041172764e-06,
      "loss": 0.5131,
      "step": 15980
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5592539310455322,
      "learning_rate": 8.839153008723905e-06,
      "loss": 0.602,
      "step": 15981
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.0513734817504883,
      "learning_rate": 8.838043990747759e-06,
      "loss": 0.6362,
      "step": 15982
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3508168458938599,
      "learning_rate": 8.836934987258155e-06,
      "loss": 0.5057,
      "step": 15983
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5892471075057983,
      "learning_rate": 8.835825998268921e-06,
      "loss": 0.5464,
      "step": 15984
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.51898992061615,
      "learning_rate": 8.834717023793879e-06,
      "loss": 0.4335,
      "step": 15985
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5739213228225708,
      "learning_rate": 8.83360806384686e-06,
      "loss": 0.6171,
      "step": 15986
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.632322072982788,
      "learning_rate": 8.832499118441685e-06,
      "loss": 0.498,
      "step": 15987
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4718958139419556,
      "learning_rate": 8.831390187592181e-06,
      "loss": 0.5133,
      "step": 15988
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2610957622528076,
      "learning_rate": 8.830281271312177e-06,
      "loss": 0.5032,
      "step": 15989
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3948649168014526,
      "learning_rate": 8.829172369615493e-06,
      "loss": 0.544,
      "step": 15990
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6986843347549438,
      "learning_rate": 8.828063482515959e-06,
      "loss": 0.6402,
      "step": 15991
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2572110891342163,
      "learning_rate": 8.826954610027394e-06,
      "loss": 0.5527,
      "step": 15992
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4972795248031616,
      "learning_rate": 8.825845752163627e-06,
      "loss": 0.536,
      "step": 15993
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.355247974395752,
      "learning_rate": 8.824736908938485e-06,
      "loss": 0.5463,
      "step": 15994
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.538903832435608,
      "learning_rate": 8.823628080365784e-06,
      "loss": 0.5105,
      "step": 15995
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6543340682983398,
      "learning_rate": 8.822519266459354e-06,
      "loss": 0.4391,
      "step": 15996
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6458503007888794,
      "learning_rate": 8.821410467233022e-06,
      "loss": 0.5461,
      "step": 15997
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4764370918273926,
      "learning_rate": 8.820301682700606e-06,
      "loss": 0.56,
      "step": 15998
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5214605331420898,
      "learning_rate": 8.819192912875934e-06,
      "loss": 0.503,
      "step": 15999
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5634942054748535,
      "learning_rate": 8.818084157772823e-06,
      "loss": 0.5341,
      "step": 16000
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.791863203048706,
      "learning_rate": 8.816975417405103e-06,
      "loss": 0.5714,
      "step": 16001
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.4737250804901123,
      "learning_rate": 8.815866691786594e-06,
      "loss": 0.47,
      "step": 16002
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.389891505241394,
      "learning_rate": 8.814757980931119e-06,
      "loss": 0.5204,
      "step": 16003
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4362154006958008,
      "learning_rate": 8.813649284852502e-06,
      "loss": 0.4723,
      "step": 16004
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.526418685913086,
      "learning_rate": 8.812540603564565e-06,
      "loss": 0.4561,
      "step": 16005
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3227235078811646,
      "learning_rate": 8.811431937081128e-06,
      "loss": 0.4864,
      "step": 16006
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3826104402542114,
      "learning_rate": 8.810323285416018e-06,
      "loss": 0.5515,
      "step": 16007
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6344200372695923,
      "learning_rate": 8.809214648583052e-06,
      "loss": 0.493,
      "step": 16008
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.741066813468933,
      "learning_rate": 8.808106026596051e-06,
      "loss": 0.5432,
      "step": 16009
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.9993098974227905,
      "learning_rate": 8.806997419468846e-06,
      "loss": 0.6165,
      "step": 16010
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7098543643951416,
      "learning_rate": 8.805888827215248e-06,
      "loss": 0.5371,
      "step": 16011
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3844778537750244,
      "learning_rate": 8.804780249849083e-06,
      "loss": 0.4935,
      "step": 16012
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7349141836166382,
      "learning_rate": 8.803671687384171e-06,
      "loss": 0.4929,
      "step": 16013
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8591610193252563,
      "learning_rate": 8.802563139834333e-06,
      "loss": 0.5547,
      "step": 16014
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.9018216133117676,
      "learning_rate": 8.801454607213391e-06,
      "loss": 0.5013,
      "step": 16015
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.622372031211853,
      "learning_rate": 8.800346089535162e-06,
      "loss": 0.4953,
      "step": 16016
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5566405057907104,
      "learning_rate": 8.799237586813472e-06,
      "loss": 0.6396,
      "step": 16017
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6295053958892822,
      "learning_rate": 8.798129099062134e-06,
      "loss": 0.5641,
      "step": 16018
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5330564975738525,
      "learning_rate": 8.797020626294972e-06,
      "loss": 0.46,
      "step": 16019
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7701520919799805,
      "learning_rate": 8.795912168525808e-06,
      "loss": 0.5411,
      "step": 16020
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.320989966392517,
      "learning_rate": 8.794803725768457e-06,
      "loss": 0.4646,
      "step": 16021
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4888156652450562,
      "learning_rate": 8.79369529803674e-06,
      "loss": 0.5144,
      "step": 16022
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.0845866203308105,
      "learning_rate": 8.792586885344479e-06,
      "loss": 0.5798,
      "step": 16023
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5036250352859497,
      "learning_rate": 8.791478487705488e-06,
      "loss": 0.4872,
      "step": 16024
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.9103962182998657,
      "learning_rate": 8.79037010513359e-06,
      "loss": 0.6342,
      "step": 16025
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.523521900177002,
      "learning_rate": 8.7892617376426e-06,
      "loss": 0.4941,
      "step": 16026
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.488533616065979,
      "learning_rate": 8.788153385246339e-06,
      "loss": 0.5045,
      "step": 16027
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7672039270401,
      "learning_rate": 8.787045047958627e-06,
      "loss": 0.6514,
      "step": 16028
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3131897449493408,
      "learning_rate": 8.785936725793276e-06,
      "loss": 0.5347,
      "step": 16029
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8689948320388794,
      "learning_rate": 8.784828418764112e-06,
      "loss": 0.5336,
      "step": 16030
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2992520332336426,
      "learning_rate": 8.783720126884944e-06,
      "loss": 0.4737,
      "step": 16031
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5370078086853027,
      "learning_rate": 8.782611850169593e-06,
      "loss": 0.4893,
      "step": 16032
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.903165340423584,
      "learning_rate": 8.781503588631881e-06,
      "loss": 0.56,
      "step": 16033
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.680420160293579,
      "learning_rate": 8.780395342285619e-06,
      "loss": 0.5315,
      "step": 16034
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5250484943389893,
      "learning_rate": 8.779287111144624e-06,
      "loss": 0.5566,
      "step": 16035
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.885912537574768,
      "learning_rate": 8.778178895222718e-06,
      "loss": 0.6445,
      "step": 16036
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4148441553115845,
      "learning_rate": 8.777070694533714e-06,
      "loss": 0.5795,
      "step": 16037
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3162099123001099,
      "learning_rate": 8.77596250909143e-06,
      "loss": 0.4745,
      "step": 16038
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7855229377746582,
      "learning_rate": 8.774854338909677e-06,
      "loss": 0.4674,
      "step": 16039
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6266130208969116,
      "learning_rate": 8.773746184002274e-06,
      "loss": 0.5464,
      "step": 16040
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3606808185577393,
      "learning_rate": 8.772638044383043e-06,
      "loss": 0.4709,
      "step": 16041
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3861573934555054,
      "learning_rate": 8.77152992006579e-06,
      "loss": 0.6386,
      "step": 16042
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7762985229492188,
      "learning_rate": 8.770421811064338e-06,
      "loss": 0.5297,
      "step": 16043
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7589552402496338,
      "learning_rate": 8.769313717392496e-06,
      "loss": 0.519,
      "step": 16044
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5262072086334229,
      "learning_rate": 8.76820563906408e-06,
      "loss": 0.4511,
      "step": 16045
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3878140449523926,
      "learning_rate": 8.767097576092911e-06,
      "loss": 0.3508,
      "step": 16046
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5650075674057007,
      "learning_rate": 8.765989528492796e-06,
      "loss": 0.6653,
      "step": 16047
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3381149768829346,
      "learning_rate": 8.764881496277554e-06,
      "loss": 0.512,
      "step": 16048
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5842899084091187,
      "learning_rate": 8.763773479461e-06,
      "loss": 0.5299,
      "step": 16049
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7954961061477661,
      "learning_rate": 8.762665478056943e-06,
      "loss": 0.4872,
      "step": 16050
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.545806646347046,
      "learning_rate": 8.761557492079203e-06,
      "loss": 0.411,
      "step": 16051
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6276072263717651,
      "learning_rate": 8.760449521541588e-06,
      "loss": 0.5677,
      "step": 16052
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6162781715393066,
      "learning_rate": 8.759341566457913e-06,
      "loss": 0.4667,
      "step": 16053
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.384552001953125,
      "learning_rate": 8.758233626841995e-06,
      "loss": 0.615,
      "step": 16054
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.707877278327942,
      "learning_rate": 8.757125702707643e-06,
      "loss": 0.5881,
      "step": 16055
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8175183534622192,
      "learning_rate": 8.756017794068674e-06,
      "loss": 0.4263,
      "step": 16056
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6485239267349243,
      "learning_rate": 8.754909900938894e-06,
      "loss": 0.5087,
      "step": 16057
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2271946668624878,
      "learning_rate": 8.75380202333212e-06,
      "loss": 0.4265,
      "step": 16058
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4038951396942139,
      "learning_rate": 8.752694161262168e-06,
      "loss": 0.5512,
      "step": 16059
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4507132768630981,
      "learning_rate": 8.751586314742841e-06,
      "loss": 0.4932,
      "step": 16060
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5326464176177979,
      "learning_rate": 8.750478483787959e-06,
      "loss": 0.5246,
      "step": 16061
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4020135402679443,
      "learning_rate": 8.749370668411331e-06,
      "loss": 0.5266,
      "step": 16062
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6962597370147705,
      "learning_rate": 8.748262868626766e-06,
      "loss": 0.5561,
      "step": 16063
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6639972925186157,
      "learning_rate": 8.74715508444808e-06,
      "loss": 0.5045,
      "step": 16064
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.363008737564087,
      "learning_rate": 8.746047315889082e-06,
      "loss": 0.4516,
      "step": 16065
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.838789463043213,
      "learning_rate": 8.744939562963579e-06,
      "loss": 0.5634,
      "step": 16066
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.0130419731140137,
      "learning_rate": 8.74383182568539e-06,
      "loss": 0.5158,
      "step": 16067
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4424808025360107,
      "learning_rate": 8.742724104068319e-06,
      "loss": 0.6838,
      "step": 16068
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4672123193740845,
      "learning_rate": 8.74161639812618e-06,
      "loss": 0.538,
      "step": 16069
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3694188594818115,
      "learning_rate": 8.740508707872778e-06,
      "loss": 0.5166,
      "step": 16070
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4181394577026367,
      "learning_rate": 8.73940103332193e-06,
      "loss": 0.4884,
      "step": 16071
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5128358602523804,
      "learning_rate": 8.738293374487443e-06,
      "loss": 0.6056,
      "step": 16072
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.739221453666687,
      "learning_rate": 8.737185731383125e-06,
      "loss": 0.6441,
      "step": 16073
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7390024662017822,
      "learning_rate": 8.736078104022784e-06,
      "loss": 0.5107,
      "step": 16074
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.017556667327881,
      "learning_rate": 8.734970492420235e-06,
      "loss": 0.6287,
      "step": 16075
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4450253248214722,
      "learning_rate": 8.733862896589283e-06,
      "loss": 0.5837,
      "step": 16076
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.728149175643921,
      "learning_rate": 8.732755316543738e-06,
      "loss": 0.465,
      "step": 16077
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5291873216629028,
      "learning_rate": 8.731647752297406e-06,
      "loss": 0.5754,
      "step": 16078
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5991721153259277,
      "learning_rate": 8.730540203864098e-06,
      "loss": 0.6761,
      "step": 16079
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4068341255187988,
      "learning_rate": 8.729432671257623e-06,
      "loss": 0.4169,
      "step": 16080
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6849892139434814,
      "learning_rate": 8.728325154491787e-06,
      "loss": 0.5223,
      "step": 16081
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4426130056381226,
      "learning_rate": 8.7272176535804e-06,
      "loss": 0.5618,
      "step": 16082
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7080472707748413,
      "learning_rate": 8.726110168537266e-06,
      "loss": 0.5786,
      "step": 16083
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5265438556671143,
      "learning_rate": 8.725002699376194e-06,
      "loss": 0.6222,
      "step": 16084
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5934548377990723,
      "learning_rate": 8.723895246110996e-06,
      "loss": 0.5053,
      "step": 16085
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5070688724517822,
      "learning_rate": 8.72278780875547e-06,
      "loss": 0.4187,
      "step": 16086
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7578113079071045,
      "learning_rate": 8.72168038732343e-06,
      "loss": 0.6474,
      "step": 16087
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7920432090759277,
      "learning_rate": 8.720572981828684e-06,
      "loss": 0.3879,
      "step": 16088
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5462846755981445,
      "learning_rate": 8.71946559228503e-06,
      "loss": 0.4885,
      "step": 16089
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4822413921356201,
      "learning_rate": 8.718358218706282e-06,
      "loss": 0.5204,
      "step": 16090
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3531084060668945,
      "learning_rate": 8.71725086110624e-06,
      "loss": 0.4281,
      "step": 16091
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.749033808708191,
      "learning_rate": 8.716143519498714e-06,
      "loss": 0.4902,
      "step": 16092
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.8164786100387573,
      "learning_rate": 8.715036193897511e-06,
      "loss": 0.4784,
      "step": 16093
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5642220973968506,
      "learning_rate": 8.713928884316432e-06,
      "loss": 0.5625,
      "step": 16094
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6302800178527832,
      "learning_rate": 8.712821590769285e-06,
      "loss": 0.4972,
      "step": 16095
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4864414930343628,
      "learning_rate": 8.711714313269874e-06,
      "loss": 0.6005,
      "step": 16096
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6831374168395996,
      "learning_rate": 8.710607051832003e-06,
      "loss": 0.5561,
      "step": 16097
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.7365314960479736,
      "learning_rate": 8.709499806469482e-06,
      "loss": 0.5226,
      "step": 16098
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4791712760925293,
      "learning_rate": 8.708392577196107e-06,
      "loss": 0.5358,
      "step": 16099
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5999201536178589,
      "learning_rate": 8.707285364025687e-06,
      "loss": 0.5448,
      "step": 16100
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6760187149047852,
      "learning_rate": 8.706178166972028e-06,
      "loss": 0.525,
      "step": 16101
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5276154279708862,
      "learning_rate": 8.705070986048928e-06,
      "loss": 0.5275,
      "step": 16102
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6774942874908447,
      "learning_rate": 8.703963821270198e-06,
      "loss": 0.5716,
      "step": 16103
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5249115228652954,
      "learning_rate": 8.702856672649635e-06,
      "loss": 0.4953,
      "step": 16104
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3680939674377441,
      "learning_rate": 8.701749540201045e-06,
      "loss": 0.4641,
      "step": 16105
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3017736673355103,
      "learning_rate": 8.700642423938231e-06,
      "loss": 0.5732,
      "step": 16106
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6304398775100708,
      "learning_rate": 8.699535323874995e-06,
      "loss": 0.5954,
      "step": 16107
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5941904783248901,
      "learning_rate": 8.698428240025144e-06,
      "loss": 0.622,
      "step": 16108
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7314971685409546,
      "learning_rate": 8.69732117240247e-06,
      "loss": 0.6137,
      "step": 16109
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4173169136047363,
      "learning_rate": 8.696214121020786e-06,
      "loss": 0.5694,
      "step": 16110
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6503583192825317,
      "learning_rate": 8.695107085893891e-06,
      "loss": 0.5772,
      "step": 16111
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7479654550552368,
      "learning_rate": 8.694000067035582e-06,
      "loss": 0.5258,
      "step": 16112
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.64906644821167,
      "learning_rate": 8.692893064459665e-06,
      "loss": 0.5304,
      "step": 16113
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3243272304534912,
      "learning_rate": 8.691786078179944e-06,
      "loss": 0.5478,
      "step": 16114
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4324990510940552,
      "learning_rate": 8.690679108210213e-06,
      "loss": 0.5326,
      "step": 16115
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5433276891708374,
      "learning_rate": 8.689572154564283e-06,
      "loss": 0.5781,
      "step": 16116
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5454187393188477,
      "learning_rate": 8.688465217255942e-06,
      "loss": 0.5632,
      "step": 16117
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4140775203704834,
      "learning_rate": 8.687358296299e-06,
      "loss": 0.5071,
      "step": 16118
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.366824984550476,
      "learning_rate": 8.686251391707257e-06,
      "loss": 0.513,
      "step": 16119
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3453222513198853,
      "learning_rate": 8.685144503494508e-06,
      "loss": 0.5048,
      "step": 16120
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6165634393692017,
      "learning_rate": 8.684037631674558e-06,
      "loss": 0.5479,
      "step": 16121
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4543384313583374,
      "learning_rate": 8.682930776261203e-06,
      "loss": 0.4556,
      "step": 16122
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3728662729263306,
      "learning_rate": 8.681823937268245e-06,
      "loss": 0.4633,
      "step": 16123
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6345468759536743,
      "learning_rate": 8.680717114709484e-06,
      "loss": 0.592,
      "step": 16124
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6564725637435913,
      "learning_rate": 8.679610308598715e-06,
      "loss": 0.5793,
      "step": 16125
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7322883605957031,
      "learning_rate": 8.678503518949741e-06,
      "loss": 0.4806,
      "step": 16126
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4069085121154785,
      "learning_rate": 8.677396745776364e-06,
      "loss": 0.4737,
      "step": 16127
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3904670476913452,
      "learning_rate": 8.676289989092372e-06,
      "loss": 0.4918,
      "step": 16128
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.483195185661316,
      "learning_rate": 8.675183248911573e-06,
      "loss": 0.5703,
      "step": 16129
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2731164693832397,
      "learning_rate": 8.674076525247761e-06,
      "loss": 0.3827,
      "step": 16130
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6957226991653442,
      "learning_rate": 8.672969818114734e-06,
      "loss": 0.5178,
      "step": 16131
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3994097709655762,
      "learning_rate": 8.671863127526292e-06,
      "loss": 0.5435,
      "step": 16132
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7582182884216309,
      "learning_rate": 8.670756453496229e-06,
      "loss": 0.5906,
      "step": 16133
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2575860023498535,
      "learning_rate": 8.669649796038345e-06,
      "loss": 0.388,
      "step": 16134
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4600567817687988,
      "learning_rate": 8.668543155166435e-06,
      "loss": 0.5573,
      "step": 16135
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4213614463806152,
      "learning_rate": 8.667436530894298e-06,
      "loss": 0.4939,
      "step": 16136
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7834057807922363,
      "learning_rate": 8.666329923235732e-06,
      "loss": 0.581,
      "step": 16137
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2703137397766113,
      "learning_rate": 8.665223332204529e-06,
      "loss": 0.5378,
      "step": 16138
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5123164653778076,
      "learning_rate": 8.664116757814487e-06,
      "loss": 0.5123,
      "step": 16139
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7586774826049805,
      "learning_rate": 8.663010200079406e-06,
      "loss": 0.5514,
      "step": 16140
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7329990863800049,
      "learning_rate": 8.661903659013076e-06,
      "loss": 0.6143,
      "step": 16141
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.671134114265442,
      "learning_rate": 8.660797134629297e-06,
      "loss": 0.55,
      "step": 16142
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6424802541732788,
      "learning_rate": 8.659690626941862e-06,
      "loss": 0.4957,
      "step": 16143
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7920796871185303,
      "learning_rate": 8.658584135964567e-06,
      "loss": 0.5359,
      "step": 16144
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.850799798965454,
      "learning_rate": 8.657477661711209e-06,
      "loss": 0.642,
      "step": 16145
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6621140241622925,
      "learning_rate": 8.656371204195577e-06,
      "loss": 0.5333,
      "step": 16146
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3404500484466553,
      "learning_rate": 8.655264763431473e-06,
      "loss": 0.5136,
      "step": 16147
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.576261043548584,
      "learning_rate": 8.654158339432686e-06,
      "loss": 0.5063,
      "step": 16148
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1817898750305176,
      "learning_rate": 8.653051932213011e-06,
      "loss": 0.5321,
      "step": 16149
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.8561804294586182,
      "learning_rate": 8.651945541786247e-06,
      "loss": 0.583,
      "step": 16150
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.613370656967163,
      "learning_rate": 8.65083916816618e-06,
      "loss": 0.5298,
      "step": 16151
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.670294165611267,
      "learning_rate": 8.649732811366608e-06,
      "loss": 0.6121,
      "step": 16152
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6228655576705933,
      "learning_rate": 8.648626471401326e-06,
      "loss": 0.6211,
      "step": 16153
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3462049961090088,
      "learning_rate": 8.647520148284123e-06,
      "loss": 0.4786,
      "step": 16154
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3495835065841675,
      "learning_rate": 8.646413842028797e-06,
      "loss": 0.5598,
      "step": 16155
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.605398178100586,
      "learning_rate": 8.645307552649134e-06,
      "loss": 0.5787,
      "step": 16156
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6274926662445068,
      "learning_rate": 8.644201280158931e-06,
      "loss": 0.371,
      "step": 16157
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5898414850234985,
      "learning_rate": 8.643095024571982e-06,
      "loss": 0.5804,
      "step": 16158
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6419806480407715,
      "learning_rate": 8.641988785902073e-06,
      "loss": 0.5473,
      "step": 16159
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4552210569381714,
      "learning_rate": 8.640882564163003e-06,
      "loss": 0.5158,
      "step": 16160
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2900383472442627,
      "learning_rate": 8.639776359368559e-06,
      "loss": 0.5104,
      "step": 16161
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5358835458755493,
      "learning_rate": 8.63867017153253e-06,
      "loss": 0.5229,
      "step": 16162
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4315296411514282,
      "learning_rate": 8.637564000668717e-06,
      "loss": 0.5126,
      "step": 16163
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6431137323379517,
      "learning_rate": 8.636457846790901e-06,
      "loss": 0.5543,
      "step": 16164
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.694234848022461,
      "learning_rate": 8.635351709912877e-06,
      "loss": 0.6328,
      "step": 16165
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.410438060760498,
      "learning_rate": 8.634245590048437e-06,
      "loss": 0.5714,
      "step": 16166
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.634473204612732,
      "learning_rate": 8.63313948721137e-06,
      "loss": 0.5765,
      "step": 16167
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.490009069442749,
      "learning_rate": 8.632033401415466e-06,
      "loss": 0.5323,
      "step": 16168
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5262638330459595,
      "learning_rate": 8.630927332674514e-06,
      "loss": 0.5663,
      "step": 16169
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7419917583465576,
      "learning_rate": 8.629821281002309e-06,
      "loss": 0.486,
      "step": 16170
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.336063265800476,
      "learning_rate": 8.628715246412632e-06,
      "loss": 0.4978,
      "step": 16171
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4832727909088135,
      "learning_rate": 8.627609228919277e-06,
      "loss": 0.4596,
      "step": 16172
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.690565586090088,
      "learning_rate": 8.626503228536031e-06,
      "loss": 0.4772,
      "step": 16173
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6856034994125366,
      "learning_rate": 8.625397245276688e-06,
      "loss": 0.4497,
      "step": 16174
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4665037393569946,
      "learning_rate": 8.624291279155032e-06,
      "loss": 0.5517,
      "step": 16175
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6912128925323486,
      "learning_rate": 8.623185330184854e-06,
      "loss": 0.4818,
      "step": 16176
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6492496728897095,
      "learning_rate": 8.622079398379942e-06,
      "loss": 0.4898,
      "step": 16177
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.8922111988067627,
      "learning_rate": 8.620973483754081e-06,
      "loss": 0.5384,
      "step": 16178
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.1142313480377197,
      "learning_rate": 8.61986758632106e-06,
      "loss": 0.5614,
      "step": 16179
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.071131944656372,
      "learning_rate": 8.618761706094672e-06,
      "loss": 0.4903,
      "step": 16180
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7757346630096436,
      "learning_rate": 8.617655843088697e-06,
      "loss": 0.5413,
      "step": 16181
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6687284708023071,
      "learning_rate": 8.616549997316928e-06,
      "loss": 0.5205,
      "step": 16182
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.660899043083191,
      "learning_rate": 8.615444168793147e-06,
      "loss": 0.6066,
      "step": 16183
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.722428560256958,
      "learning_rate": 8.614338357531144e-06,
      "loss": 0.5409,
      "step": 16184
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4873379468917847,
      "learning_rate": 8.613232563544706e-06,
      "loss": 0.4574,
      "step": 16185
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4163823127746582,
      "learning_rate": 8.612126786847617e-06,
      "loss": 0.4677,
      "step": 16186
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.52342689037323,
      "learning_rate": 8.611021027453663e-06,
      "loss": 0.4265,
      "step": 16187
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5618692636489868,
      "learning_rate": 8.609915285376633e-06,
      "loss": 0.478,
      "step": 16188
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5509097576141357,
      "learning_rate": 8.60880956063031e-06,
      "loss": 0.4679,
      "step": 16189
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.420871615409851,
      "learning_rate": 8.607703853228482e-06,
      "loss": 0.5811,
      "step": 16190
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4237233400344849,
      "learning_rate": 8.606598163184932e-06,
      "loss": 0.5018,
      "step": 16191
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5441865921020508,
      "learning_rate": 8.605492490513443e-06,
      "loss": 0.4889,
      "step": 16192
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.7433518171310425,
      "learning_rate": 8.604386835227807e-06,
      "loss": 0.6361,
      "step": 16193
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5697040557861328,
      "learning_rate": 8.603281197341801e-06,
      "loss": 0.5044,
      "step": 16194
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.5900297164916992,
      "learning_rate": 8.602175576869214e-06,
      "loss": 0.4908,
      "step": 16195
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4186886548995972,
      "learning_rate": 8.601069973823828e-06,
      "loss": 0.5198,
      "step": 16196
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6425442695617676,
      "learning_rate": 8.599964388219425e-06,
      "loss": 0.5831,
      "step": 16197
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.505854606628418,
      "learning_rate": 8.598858820069798e-06,
      "loss": 0.5376,
      "step": 16198
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.403403401374817,
      "learning_rate": 8.59775326938872e-06,
      "loss": 0.5067,
      "step": 16199
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7536640167236328,
      "learning_rate": 8.596647736189975e-06,
      "loss": 0.4129,
      "step": 16200
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4678351879119873,
      "learning_rate": 8.595542220487356e-06,
      "loss": 0.536,
      "step": 16201
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5224030017852783,
      "learning_rate": 8.594436722294634e-06,
      "loss": 0.5856,
      "step": 16202
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4335272312164307,
      "learning_rate": 8.593331241625602e-06,
      "loss": 0.5521,
      "step": 16203
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7411733865737915,
      "learning_rate": 8.592225778494033e-06,
      "loss": 0.5902,
      "step": 16204
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6959432363510132,
      "learning_rate": 8.591120332913714e-06,
      "loss": 0.5596,
      "step": 16205
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6341004371643066,
      "learning_rate": 8.59001490489843e-06,
      "loss": 0.5709,
      "step": 16206
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.361434817314148,
      "learning_rate": 8.588909494461957e-06,
      "loss": 0.5081,
      "step": 16207
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2980811595916748,
      "learning_rate": 8.58780410161808e-06,
      "loss": 0.4767,
      "step": 16208
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3612990379333496,
      "learning_rate": 8.586698726380578e-06,
      "loss": 0.5385,
      "step": 16209
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7226619720458984,
      "learning_rate": 8.585593368763233e-06,
      "loss": 0.6124,
      "step": 16210
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5427910089492798,
      "learning_rate": 8.584488028779828e-06,
      "loss": 0.4657,
      "step": 16211
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.440689206123352,
      "learning_rate": 8.58338270644414e-06,
      "loss": 0.5441,
      "step": 16212
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6641968488693237,
      "learning_rate": 8.582277401769954e-06,
      "loss": 0.4499,
      "step": 16213
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4876664876937866,
      "learning_rate": 8.581172114771047e-06,
      "loss": 0.4941,
      "step": 16214
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4731943607330322,
      "learning_rate": 8.5800668454612e-06,
      "loss": 0.4761,
      "step": 16215
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5579966306686401,
      "learning_rate": 8.578961593854191e-06,
      "loss": 0.5093,
      "step": 16216
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5803046226501465,
      "learning_rate": 8.577856359963803e-06,
      "loss": 0.5356,
      "step": 16217
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4194045066833496,
      "learning_rate": 8.57675114380381e-06,
      "loss": 0.4353,
      "step": 16218
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7635325193405151,
      "learning_rate": 8.575645945388e-06,
      "loss": 0.5581,
      "step": 16219
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.835292935371399,
      "learning_rate": 8.574540764730142e-06,
      "loss": 0.5552,
      "step": 16220
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3921962976455688,
      "learning_rate": 8.573435601844023e-06,
      "loss": 0.5236,
      "step": 16221
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3888188600540161,
      "learning_rate": 8.572330456743416e-06,
      "loss": 0.5472,
      "step": 16222
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.746593952178955,
      "learning_rate": 8.5712253294421e-06,
      "loss": 0.5226,
      "step": 16223
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.539506435394287,
      "learning_rate": 8.570120219953857e-06,
      "loss": 0.5083,
      "step": 16224
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4742428064346313,
      "learning_rate": 8.569015128292459e-06,
      "loss": 0.5273,
      "step": 16225
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.683134913444519,
      "learning_rate": 8.567910054471688e-06,
      "loss": 0.5147,
      "step": 16226
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.594088077545166,
      "learning_rate": 8.56680499850532e-06,
      "loss": 0.4897,
      "step": 16227
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6319586038589478,
      "learning_rate": 8.56569996040713e-06,
      "loss": 0.4983,
      "step": 16228
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4307607412338257,
      "learning_rate": 8.564594940190901e-06,
      "loss": 0.5497,
      "step": 16229
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.534480094909668,
      "learning_rate": 8.563489937870402e-06,
      "loss": 0.5246,
      "step": 16230
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5582634210586548,
      "learning_rate": 8.562384953459412e-06,
      "loss": 0.4594,
      "step": 16231
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.702644944190979,
      "learning_rate": 8.561279986971714e-06,
      "loss": 0.5541,
      "step": 16232
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4500516653060913,
      "learning_rate": 8.560175038421072e-06,
      "loss": 0.501,
      "step": 16233
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5442774295806885,
      "learning_rate": 8.559070107821274e-06,
      "loss": 0.4727,
      "step": 16234
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6167984008789062,
      "learning_rate": 8.557965195186086e-06,
      "loss": 0.4304,
      "step": 16235
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8734606504440308,
      "learning_rate": 8.556860300529287e-06,
      "loss": 0.5978,
      "step": 16236
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4509656429290771,
      "learning_rate": 8.555755423864654e-06,
      "loss": 0.4998,
      "step": 16237
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7393786907196045,
      "learning_rate": 8.55465056520596e-06,
      "loss": 0.523,
      "step": 16238
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7929723262786865,
      "learning_rate": 8.55354572456698e-06,
      "loss": 0.4976,
      "step": 16239
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6913037300109863,
      "learning_rate": 8.552440901961488e-06,
      "loss": 0.5762,
      "step": 16240
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.468829870223999,
      "learning_rate": 8.551336097403259e-06,
      "loss": 0.4247,
      "step": 16241
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6150318384170532,
      "learning_rate": 8.550231310906065e-06,
      "loss": 0.4596,
      "step": 16242
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2583626508712769,
      "learning_rate": 8.549126542483683e-06,
      "loss": 0.5605,
      "step": 16243
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5295335054397583,
      "learning_rate": 8.548021792149884e-06,
      "loss": 0.5075,
      "step": 16244
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6159368753433228,
      "learning_rate": 8.546917059918444e-06,
      "loss": 0.5948,
      "step": 16245
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4217361211776733,
      "learning_rate": 8.545812345803132e-06,
      "loss": 0.508,
      "step": 16246
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6661880016326904,
      "learning_rate": 8.544707649817725e-06,
      "loss": 0.4896,
      "step": 16247
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5840452909469604,
      "learning_rate": 8.543602971975993e-06,
      "loss": 0.4868,
      "step": 16248
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.897609829902649,
      "learning_rate": 8.54249831229171e-06,
      "loss": 0.5653,
      "step": 16249
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3288307189941406,
      "learning_rate": 8.541393670778649e-06,
      "loss": 0.5125,
      "step": 16250
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8330703973770142,
      "learning_rate": 8.540289047450577e-06,
      "loss": 0.5594,
      "step": 16251
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5094510316848755,
      "learning_rate": 8.539184442321273e-06,
      "loss": 0.5223,
      "step": 16252
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4228943586349487,
      "learning_rate": 8.538079855404502e-06,
      "loss": 0.4578,
      "step": 16253
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7726469039916992,
      "learning_rate": 8.536975286714039e-06,
      "loss": 0.5959,
      "step": 16254
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6861902475357056,
      "learning_rate": 8.535870736263656e-06,
      "loss": 0.5556,
      "step": 16255
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5601156949996948,
      "learning_rate": 8.53476620406712e-06,
      "loss": 0.591,
      "step": 16256
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3985874652862549,
      "learning_rate": 8.533661690138203e-06,
      "loss": 0.5571,
      "step": 16257
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6288436651229858,
      "learning_rate": 8.532557194490678e-06,
      "loss": 0.4228,
      "step": 16258
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6267850399017334,
      "learning_rate": 8.531452717138312e-06,
      "loss": 0.4507,
      "step": 16259
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.9116919040679932,
      "learning_rate": 8.530348258094878e-06,
      "loss": 0.5168,
      "step": 16260
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2698490619659424,
      "learning_rate": 8.529243817374142e-06,
      "loss": 0.6217,
      "step": 16261
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8925682306289673,
      "learning_rate": 8.528139394989877e-06,
      "loss": 0.5268,
      "step": 16262
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.576332449913025,
      "learning_rate": 8.527034990955849e-06,
      "loss": 0.508,
      "step": 16263
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.670310616493225,
      "learning_rate": 8.525930605285829e-06,
      "loss": 0.4878,
      "step": 16264
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2819167375564575,
      "learning_rate": 8.524826237993588e-06,
      "loss": 0.5388,
      "step": 16265
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.770601749420166,
      "learning_rate": 8.523721889092889e-06,
      "loss": 0.5522,
      "step": 16266
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5473544597625732,
      "learning_rate": 8.522617558597502e-06,
      "loss": 0.5391,
      "step": 16267
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8715758323669434,
      "learning_rate": 8.521513246521201e-06,
      "loss": 0.5455,
      "step": 16268
    },
    {
      "epoch": 1.68,
      "grad_norm": 2.18243670463562,
      "learning_rate": 8.520408952877745e-06,
      "loss": 0.5778,
      "step": 16269
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6130824089050293,
      "learning_rate": 8.519304677680906e-06,
      "loss": 0.4313,
      "step": 16270
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7744507789611816,
      "learning_rate": 8.518200420944454e-06,
      "loss": 0.5042,
      "step": 16271
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3583635091781616,
      "learning_rate": 8.517096182682153e-06,
      "loss": 0.5259,
      "step": 16272
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5448555946350098,
      "learning_rate": 8.515991962907771e-06,
      "loss": 0.5201,
      "step": 16273
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7219113111495972,
      "learning_rate": 8.514887761635072e-06,
      "loss": 0.5633,
      "step": 16274
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6090478897094727,
      "learning_rate": 8.513783578877825e-06,
      "loss": 0.49,
      "step": 16275
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.96156907081604,
      "learning_rate": 8.512679414649799e-06,
      "loss": 0.5685,
      "step": 16276
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5244560241699219,
      "learning_rate": 8.511575268964753e-06,
      "loss": 0.4678,
      "step": 16277
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.411370873451233,
      "learning_rate": 8.51047114183646e-06,
      "loss": 0.4882,
      "step": 16278
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5632213354110718,
      "learning_rate": 8.509367033278681e-06,
      "loss": 0.5572,
      "step": 16279
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2952426671981812,
      "learning_rate": 8.508262943305181e-06,
      "loss": 0.5277,
      "step": 16280
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8045629262924194,
      "learning_rate": 8.50715887192973e-06,
      "loss": 0.4435,
      "step": 16281
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.835508108139038,
      "learning_rate": 8.506054819166087e-06,
      "loss": 0.5421,
      "step": 16282
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6766470670700073,
      "learning_rate": 8.50495078502802e-06,
      "loss": 0.5132,
      "step": 16283
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.65109384059906,
      "learning_rate": 8.503846769529294e-06,
      "loss": 0.5027,
      "step": 16284
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.930251955986023,
      "learning_rate": 8.50274277268367e-06,
      "loss": 0.5548,
      "step": 16285
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.7167352437973022,
      "learning_rate": 8.501638794504916e-06,
      "loss": 0.6544,
      "step": 16286
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.5741385221481323,
      "learning_rate": 8.500534835006791e-06,
      "loss": 0.5842,
      "step": 16287
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6435495615005493,
      "learning_rate": 8.499430894203063e-06,
      "loss": 0.5887,
      "step": 16288
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.386022925376892,
      "learning_rate": 8.498326972107493e-06,
      "loss": 0.5126,
      "step": 16289
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.686048984527588,
      "learning_rate": 8.497223068733844e-06,
      "loss": 0.5168,
      "step": 16290
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8109208345413208,
      "learning_rate": 8.496119184095879e-06,
      "loss": 0.6183,
      "step": 16291
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.6032968759536743,
      "learning_rate": 8.49501531820736e-06,
      "loss": 0.5394,
      "step": 16292
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4908770322799683,
      "learning_rate": 8.49391147108205e-06,
      "loss": 0.5264,
      "step": 16293
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.605396032333374,
      "learning_rate": 8.492807642733713e-06,
      "loss": 0.5429,
      "step": 16294
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.51799476146698,
      "learning_rate": 8.491703833176105e-06,
      "loss": 0.5248,
      "step": 16295
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5016096830368042,
      "learning_rate": 8.490600042422994e-06,
      "loss": 0.6736,
      "step": 16296
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7273918390274048,
      "learning_rate": 8.48949627048814e-06,
      "loss": 0.6295,
      "step": 16297
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6443736553192139,
      "learning_rate": 8.4883925173853e-06,
      "loss": 0.5895,
      "step": 16298
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.696746826171875,
      "learning_rate": 8.48728878312824e-06,
      "loss": 0.42,
      "step": 16299
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.49894118309021,
      "learning_rate": 8.486185067730717e-06,
      "loss": 0.6001,
      "step": 16300
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.570528268814087,
      "learning_rate": 8.485081371206495e-06,
      "loss": 0.6421,
      "step": 16301
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5416473150253296,
      "learning_rate": 8.483977693569331e-06,
      "loss": 0.5108,
      "step": 16302
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4484481811523438,
      "learning_rate": 8.482874034832985e-06,
      "loss": 0.5154,
      "step": 16303
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7332977056503296,
      "learning_rate": 8.48177039501122e-06,
      "loss": 0.5699,
      "step": 16304
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.286077618598938,
      "learning_rate": 8.480666774117792e-06,
      "loss": 0.5158,
      "step": 16305
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3203595876693726,
      "learning_rate": 8.479563172166461e-06,
      "loss": 0.4715,
      "step": 16306
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6279863119125366,
      "learning_rate": 8.478459589170988e-06,
      "loss": 0.4711,
      "step": 16307
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9223114252090454,
      "learning_rate": 8.47735602514513e-06,
      "loss": 0.6032,
      "step": 16308
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.059837579727173,
      "learning_rate": 8.476252480102644e-06,
      "loss": 0.5251,
      "step": 16309
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4364804029464722,
      "learning_rate": 8.475148954057292e-06,
      "loss": 0.6112,
      "step": 16310
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4580990076065063,
      "learning_rate": 8.47404544702283e-06,
      "loss": 0.4522,
      "step": 16311
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9711549282073975,
      "learning_rate": 8.472941959013017e-06,
      "loss": 0.5721,
      "step": 16312
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3906761407852173,
      "learning_rate": 8.47183849004161e-06,
      "loss": 0.4994,
      "step": 16313
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5772379636764526,
      "learning_rate": 8.470735040122364e-06,
      "loss": 0.4512,
      "step": 16314
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8649436235427856,
      "learning_rate": 8.46963160926904e-06,
      "loss": 0.5218,
      "step": 16315
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.990596890449524,
      "learning_rate": 8.468528197495393e-06,
      "loss": 0.622,
      "step": 16316
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4541676044464111,
      "learning_rate": 8.46742480481518e-06,
      "loss": 0.4502,
      "step": 16317
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6680113077163696,
      "learning_rate": 8.466321431242156e-06,
      "loss": 0.501,
      "step": 16318
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.163553476333618,
      "learning_rate": 8.465218076790078e-06,
      "loss": 0.5325,
      "step": 16319
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6747997999191284,
      "learning_rate": 8.464114741472704e-06,
      "loss": 0.5495,
      "step": 16320
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.482163429260254,
      "learning_rate": 8.463011425303788e-06,
      "loss": 0.6121,
      "step": 16321
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4707573652267456,
      "learning_rate": 8.461908128297083e-06,
      "loss": 0.436,
      "step": 16322
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5647342205047607,
      "learning_rate": 8.46080485046635e-06,
      "loss": 0.5047,
      "step": 16323
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.380258560180664,
      "learning_rate": 8.459701591825338e-06,
      "loss": 0.4249,
      "step": 16324
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4453932046890259,
      "learning_rate": 8.458598352387807e-06,
      "loss": 0.559,
      "step": 16325
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4913444519042969,
      "learning_rate": 8.457495132167505e-06,
      "loss": 0.4875,
      "step": 16326
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5839227437973022,
      "learning_rate": 8.456391931178193e-06,
      "loss": 0.5044,
      "step": 16327
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5553582906723022,
      "learning_rate": 8.455288749433624e-06,
      "loss": 0.5093,
      "step": 16328
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8054038286209106,
      "learning_rate": 8.454185586947546e-06,
      "loss": 0.5332,
      "step": 16329
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5153087377548218,
      "learning_rate": 8.453082443733718e-06,
      "loss": 0.4528,
      "step": 16330
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.0801351070404053,
      "learning_rate": 8.451979319805892e-06,
      "loss": 0.5374,
      "step": 16331
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3851045370101929,
      "learning_rate": 8.45087621517782e-06,
      "loss": 0.408,
      "step": 16332
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.1619374752044678,
      "learning_rate": 8.449773129863257e-06,
      "loss": 0.4828,
      "step": 16333
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8246407508850098,
      "learning_rate": 8.448670063875953e-06,
      "loss": 0.6035,
      "step": 16334
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4695923328399658,
      "learning_rate": 8.447567017229661e-06,
      "loss": 0.51,
      "step": 16335
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4457194805145264,
      "learning_rate": 8.446463989938136e-06,
      "loss": 0.5403,
      "step": 16336
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.078270196914673,
      "learning_rate": 8.445360982015126e-06,
      "loss": 0.6486,
      "step": 16337
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.578858494758606,
      "learning_rate": 8.444257993474387e-06,
      "loss": 0.4654,
      "step": 16338
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9698691368103027,
      "learning_rate": 8.443155024329664e-06,
      "loss": 0.5912,
      "step": 16339
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4834707975387573,
      "learning_rate": 8.44205207459471e-06,
      "loss": 0.5128,
      "step": 16340
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.572852373123169,
      "learning_rate": 8.440949144283284e-06,
      "loss": 0.5615,
      "step": 16341
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4285032749176025,
      "learning_rate": 8.439846233409125e-06,
      "loss": 0.439,
      "step": 16342
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7201634645462036,
      "learning_rate": 8.43874334198599e-06,
      "loss": 0.4976,
      "step": 16343
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7478666305541992,
      "learning_rate": 8.437640470027628e-06,
      "loss": 0.492,
      "step": 16344
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5606111288070679,
      "learning_rate": 8.436537617547787e-06,
      "loss": 0.5088,
      "step": 16345
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6507587432861328,
      "learning_rate": 8.43543478456022e-06,
      "loss": 0.4839,
      "step": 16346
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6329821348190308,
      "learning_rate": 8.434331971078676e-06,
      "loss": 0.487,
      "step": 16347
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.484275460243225,
      "learning_rate": 8.4332291771169e-06,
      "loss": 0.4927,
      "step": 16348
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7018470764160156,
      "learning_rate": 8.432126402688648e-06,
      "loss": 0.5275,
      "step": 16349
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2379164695739746,
      "learning_rate": 8.43102364780766e-06,
      "loss": 0.5158,
      "step": 16350
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.790552020072937,
      "learning_rate": 8.429920912487693e-06,
      "loss": 0.5687,
      "step": 16351
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5355504751205444,
      "learning_rate": 8.428818196742488e-06,
      "loss": 0.4948,
      "step": 16352
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7275681495666504,
      "learning_rate": 8.427715500585798e-06,
      "loss": 0.508,
      "step": 16353
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7396653890609741,
      "learning_rate": 8.426612824031369e-06,
      "loss": 0.4907,
      "step": 16354
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.623897910118103,
      "learning_rate": 8.425510167092948e-06,
      "loss": 0.4568,
      "step": 16355
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7186976671218872,
      "learning_rate": 8.424407529784285e-06,
      "loss": 0.4456,
      "step": 16356
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7173420190811157,
      "learning_rate": 8.423304912119122e-06,
      "loss": 0.3939,
      "step": 16357
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3524119853973389,
      "learning_rate": 8.422202314111207e-06,
      "loss": 0.5015,
      "step": 16358
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5792299509048462,
      "learning_rate": 8.421099735774292e-06,
      "loss": 0.4918,
      "step": 16359
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5231130123138428,
      "learning_rate": 8.419997177122115e-06,
      "loss": 0.4974,
      "step": 16360
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7421029806137085,
      "learning_rate": 8.418894638168428e-06,
      "loss": 0.446,
      "step": 16361
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.660550832748413,
      "learning_rate": 8.417792118926977e-06,
      "loss": 0.6458,
      "step": 16362
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.8222028017044067,
      "learning_rate": 8.416689619411503e-06,
      "loss": 0.5902,
      "step": 16363
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4635882377624512,
      "learning_rate": 8.415587139635755e-06,
      "loss": 0.5469,
      "step": 16364
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9081661701202393,
      "learning_rate": 8.414484679613477e-06,
      "loss": 0.5011,
      "step": 16365
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5739469528198242,
      "learning_rate": 8.413382239358411e-06,
      "loss": 0.5597,
      "step": 16366
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5555020570755005,
      "learning_rate": 8.41227981888431e-06,
      "loss": 0.5532,
      "step": 16367
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7244815826416016,
      "learning_rate": 8.411177418204907e-06,
      "loss": 0.553,
      "step": 16368
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.231213092803955,
      "learning_rate": 8.410075037333954e-06,
      "loss": 0.5528,
      "step": 16369
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4384946823120117,
      "learning_rate": 8.408972676285192e-06,
      "loss": 0.5621,
      "step": 16370
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4531410932540894,
      "learning_rate": 8.407870335072362e-06,
      "loss": 0.5051,
      "step": 16371
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4696060419082642,
      "learning_rate": 8.406768013709214e-06,
      "loss": 0.572,
      "step": 16372
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9273079633712769,
      "learning_rate": 8.405665712209486e-06,
      "loss": 0.5466,
      "step": 16373
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7765448093414307,
      "learning_rate": 8.40456343058692e-06,
      "loss": 0.5249,
      "step": 16374
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4128774404525757,
      "learning_rate": 8.403461168855263e-06,
      "loss": 0.4387,
      "step": 16375
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6487491130828857,
      "learning_rate": 8.402358927028253e-06,
      "loss": 0.6228,
      "step": 16376
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7014394998550415,
      "learning_rate": 8.401256705119636e-06,
      "loss": 0.6162,
      "step": 16377
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6916743516921997,
      "learning_rate": 8.40015450314315e-06,
      "loss": 0.554,
      "step": 16378
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4286162853240967,
      "learning_rate": 8.399052321112538e-06,
      "loss": 0.5212,
      "step": 16379
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.677122712135315,
      "learning_rate": 8.397950159041544e-06,
      "loss": 0.5274,
      "step": 16380
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4844861030578613,
      "learning_rate": 8.396848016943903e-06,
      "loss": 0.5229,
      "step": 16381
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.798083782196045,
      "learning_rate": 8.395745894833364e-06,
      "loss": 0.4764,
      "step": 16382
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6650218963623047,
      "learning_rate": 8.39464379272366e-06,
      "loss": 0.5365,
      "step": 16383
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5692906379699707,
      "learning_rate": 8.393541710628534e-06,
      "loss": 0.5527,
      "step": 16384
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5052704811096191,
      "learning_rate": 8.39243964856173e-06,
      "loss": 0.4815,
      "step": 16385
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.359217643737793,
      "learning_rate": 8.39133760653698e-06,
      "loss": 0.4842,
      "step": 16386
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7747373580932617,
      "learning_rate": 8.390235584568029e-06,
      "loss": 0.4901,
      "step": 16387
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5132514238357544,
      "learning_rate": 8.389133582668616e-06,
      "loss": 0.5853,
      "step": 16388
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6280328035354614,
      "learning_rate": 8.388031600852479e-06,
      "loss": 0.5657,
      "step": 16389
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6198623180389404,
      "learning_rate": 8.386929639133359e-06,
      "loss": 0.5632,
      "step": 16390
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.587132215499878,
      "learning_rate": 8.385827697524989e-06,
      "loss": 0.4687,
      "step": 16391
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3946490287780762,
      "learning_rate": 8.384725776041111e-06,
      "loss": 0.6124,
      "step": 16392
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8251229524612427,
      "learning_rate": 8.383623874695466e-06,
      "loss": 0.5237,
      "step": 16393
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3526266813278198,
      "learning_rate": 8.382521993501787e-06,
      "loss": 0.5677,
      "step": 16394
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.327960729598999,
      "learning_rate": 8.381420132473817e-06,
      "loss": 0.528,
      "step": 16395
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.130293369293213,
      "learning_rate": 8.380318291625284e-06,
      "loss": 0.607,
      "step": 16396
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7754359245300293,
      "learning_rate": 8.379216470969932e-06,
      "loss": 0.5833,
      "step": 16397
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6158729791641235,
      "learning_rate": 8.378114670521502e-06,
      "loss": 0.4688,
      "step": 16398
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7247792482376099,
      "learning_rate": 8.37701289029372e-06,
      "loss": 0.5464,
      "step": 16399
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3802869319915771,
      "learning_rate": 8.375911130300327e-06,
      "loss": 0.5774,
      "step": 16400
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3934431076049805,
      "learning_rate": 8.374809390555063e-06,
      "loss": 0.4958,
      "step": 16401
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.620031714439392,
      "learning_rate": 8.373707671071658e-06,
      "loss": 0.5839,
      "step": 16402
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6018790006637573,
      "learning_rate": 8.372605971863852e-06,
      "loss": 0.5761,
      "step": 16403
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4148237705230713,
      "learning_rate": 8.371504292945374e-06,
      "loss": 0.4593,
      "step": 16404
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4912785291671753,
      "learning_rate": 8.370402634329966e-06,
      "loss": 0.5481,
      "step": 16405
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6504048109054565,
      "learning_rate": 8.369300996031362e-06,
      "loss": 0.5856,
      "step": 16406
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9902454614639282,
      "learning_rate": 8.368199378063291e-06,
      "loss": 0.5816,
      "step": 16407
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6118190288543701,
      "learning_rate": 8.367097780439493e-06,
      "loss": 0.5627,
      "step": 16408
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4820401668548584,
      "learning_rate": 8.365996203173698e-06,
      "loss": 0.5381,
      "step": 16409
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5278571844100952,
      "learning_rate": 8.364894646279642e-06,
      "loss": 0.4193,
      "step": 16410
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.775039792060852,
      "learning_rate": 8.36379310977106e-06,
      "loss": 0.5146,
      "step": 16411
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4895955324172974,
      "learning_rate": 8.362691593661682e-06,
      "loss": 0.4558,
      "step": 16412
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4164490699768066,
      "learning_rate": 8.361590097965243e-06,
      "loss": 0.4712,
      "step": 16413
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4268912076950073,
      "learning_rate": 8.360488622695474e-06,
      "loss": 0.5766,
      "step": 16414
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.2046685218811035,
      "learning_rate": 8.35938716786611e-06,
      "loss": 0.5369,
      "step": 16415
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9537644386291504,
      "learning_rate": 8.358285733490883e-06,
      "loss": 0.5525,
      "step": 16416
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5734978914260864,
      "learning_rate": 8.357184319583522e-06,
      "loss": 0.5138,
      "step": 16417
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6209495067596436,
      "learning_rate": 8.35608292615776e-06,
      "loss": 0.4931,
      "step": 16418
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.708251953125,
      "learning_rate": 8.354981553227332e-06,
      "loss": 0.6119,
      "step": 16419
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3202340602874756,
      "learning_rate": 8.353880200805962e-06,
      "loss": 0.5058,
      "step": 16420
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.627252221107483,
      "learning_rate": 8.35277886890739e-06,
      "loss": 0.5676,
      "step": 16421
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7770287990570068,
      "learning_rate": 8.35167755754534e-06,
      "loss": 0.5061,
      "step": 16422
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8072372674942017,
      "learning_rate": 8.350576266733545e-06,
      "loss": 0.5409,
      "step": 16423
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5574839115142822,
      "learning_rate": 8.349474996485735e-06,
      "loss": 0.5741,
      "step": 16424
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3086270093917847,
      "learning_rate": 8.348373746815638e-06,
      "loss": 0.5505,
      "step": 16425
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.45003342628479,
      "learning_rate": 8.347272517736985e-06,
      "loss": 0.522,
      "step": 16426
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5114201307296753,
      "learning_rate": 8.346171309263512e-06,
      "loss": 0.4595,
      "step": 16427
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7575963735580444,
      "learning_rate": 8.345070121408936e-06,
      "loss": 0.4526,
      "step": 16428
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6703599691390991,
      "learning_rate": 8.343968954186993e-06,
      "loss": 0.554,
      "step": 16429
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.0104517936706543,
      "learning_rate": 8.342867807611411e-06,
      "loss": 0.5228,
      "step": 16430
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5601683855056763,
      "learning_rate": 8.341766681695918e-06,
      "loss": 0.5175,
      "step": 16431
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5692843198776245,
      "learning_rate": 8.340665576454243e-06,
      "loss": 0.4924,
      "step": 16432
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8364102840423584,
      "learning_rate": 8.339564491900112e-06,
      "loss": 0.4853,
      "step": 16433
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5557291507720947,
      "learning_rate": 8.338463428047255e-06,
      "loss": 0.4142,
      "step": 16434
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4455031156539917,
      "learning_rate": 8.337362384909398e-06,
      "loss": 0.4297,
      "step": 16435
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7720839977264404,
      "learning_rate": 8.336261362500266e-06,
      "loss": 0.5181,
      "step": 16436
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.416849970817566,
      "learning_rate": 8.335160360833591e-06,
      "loss": 0.5395,
      "step": 16437
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4176185131072998,
      "learning_rate": 8.334059379923094e-06,
      "loss": 0.5254,
      "step": 16438
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8742636442184448,
      "learning_rate": 8.332958419782505e-06,
      "loss": 0.5741,
      "step": 16439
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.573723554611206,
      "learning_rate": 8.331857480425549e-06,
      "loss": 0.5239,
      "step": 16440
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5426857471466064,
      "learning_rate": 8.330756561865952e-06,
      "loss": 0.5661,
      "step": 16441
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6648335456848145,
      "learning_rate": 8.32965566411744e-06,
      "loss": 0.5962,
      "step": 16442
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2923533916473389,
      "learning_rate": 8.328554787193738e-06,
      "loss": 0.3619,
      "step": 16443
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6602978706359863,
      "learning_rate": 8.32745393110857e-06,
      "loss": 0.5391,
      "step": 16444
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7330752611160278,
      "learning_rate": 8.326353095875663e-06,
      "loss": 0.5539,
      "step": 16445
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.446399211883545,
      "learning_rate": 8.325252281508737e-06,
      "loss": 0.5641,
      "step": 16446
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5420668125152588,
      "learning_rate": 8.324151488021524e-06,
      "loss": 0.5911,
      "step": 16447
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.434910774230957,
      "learning_rate": 8.323050715427739e-06,
      "loss": 0.5296,
      "step": 16448
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.512136697769165,
      "learning_rate": 8.321949963741112e-06,
      "loss": 0.4677,
      "step": 16449
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7048884630203247,
      "learning_rate": 8.320849232975366e-06,
      "loss": 0.5042,
      "step": 16450
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.251084804534912,
      "learning_rate": 8.31974852314422e-06,
      "loss": 0.4599,
      "step": 16451
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.792465329170227,
      "learning_rate": 8.3186478342614e-06,
      "loss": 0.4718,
      "step": 16452
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6006344556808472,
      "learning_rate": 8.317547166340632e-06,
      "loss": 0.5554,
      "step": 16453
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6467667818069458,
      "learning_rate": 8.316446519395631e-06,
      "loss": 0.4706,
      "step": 16454
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4763089418411255,
      "learning_rate": 8.315345893440125e-06,
      "loss": 0.5509,
      "step": 16455
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.602569818496704,
      "learning_rate": 8.314245288487835e-06,
      "loss": 0.5094,
      "step": 16456
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6852017641067505,
      "learning_rate": 8.313144704552479e-06,
      "loss": 0.6042,
      "step": 16457
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5952050685882568,
      "learning_rate": 8.312044141647783e-06,
      "loss": 0.5164,
      "step": 16458
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7970261573791504,
      "learning_rate": 8.310943599787467e-06,
      "loss": 0.5669,
      "step": 16459
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7736101150512695,
      "learning_rate": 8.30984307898525e-06,
      "loss": 0.5301,
      "step": 16460
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.723444938659668,
      "learning_rate": 8.308742579254854e-06,
      "loss": 0.4767,
      "step": 16461
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4998008012771606,
      "learning_rate": 8.307642100609997e-06,
      "loss": 0.4815,
      "step": 16462
    },
    {
      "epoch": 1.7,
      "grad_norm": 2.0395870208740234,
      "learning_rate": 8.306541643064405e-06,
      "loss": 0.6076,
      "step": 16463
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.365195393562317,
      "learning_rate": 8.30544120663179e-06,
      "loss": 0.5517,
      "step": 16464
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.34413480758667,
      "learning_rate": 8.304340791325878e-06,
      "loss": 0.521,
      "step": 16465
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4717293977737427,
      "learning_rate": 8.303240397160386e-06,
      "loss": 0.5327,
      "step": 16466
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5121562480926514,
      "learning_rate": 8.302140024149032e-06,
      "loss": 0.4422,
      "step": 16467
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.509957194328308,
      "learning_rate": 8.301039672305536e-06,
      "loss": 0.4593,
      "step": 16468
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4957019090652466,
      "learning_rate": 8.299939341643614e-06,
      "loss": 0.5017,
      "step": 16469
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6014360189437866,
      "learning_rate": 8.298839032176989e-06,
      "loss": 0.5005,
      "step": 16470
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.672355055809021,
      "learning_rate": 8.297738743919377e-06,
      "loss": 0.4966,
      "step": 16471
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5875718593597412,
      "learning_rate": 8.296638476884493e-06,
      "loss": 0.6362,
      "step": 16472
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.697228193283081,
      "learning_rate": 8.295538231086056e-06,
      "loss": 0.5476,
      "step": 16473
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5727795362472534,
      "learning_rate": 8.294438006537784e-06,
      "loss": 0.5817,
      "step": 16474
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5644053220748901,
      "learning_rate": 8.293337803253394e-06,
      "loss": 0.4652,
      "step": 16475
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5231083631515503,
      "learning_rate": 8.292237621246603e-06,
      "loss": 0.551,
      "step": 16476
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4585157632827759,
      "learning_rate": 8.291137460531124e-06,
      "loss": 0.4966,
      "step": 16477
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.407841444015503,
      "learning_rate": 8.290037321120678e-06,
      "loss": 0.518,
      "step": 16478
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5698539018630981,
      "learning_rate": 8.288937203028979e-06,
      "loss": 0.5353,
      "step": 16479
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.7832236289978027,
      "learning_rate": 8.28783710626974e-06,
      "loss": 0.4365,
      "step": 16480
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.489071249961853,
      "learning_rate": 8.286737030856681e-06,
      "loss": 0.5766,
      "step": 16481
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.517622709274292,
      "learning_rate": 8.285636976803512e-06,
      "loss": 0.5146,
      "step": 16482
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5863161087036133,
      "learning_rate": 8.28453694412395e-06,
      "loss": 0.565,
      "step": 16483
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8553189039230347,
      "learning_rate": 8.283436932831713e-06,
      "loss": 0.5947,
      "step": 16484
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6415308713912964,
      "learning_rate": 8.282336942940508e-06,
      "loss": 0.5988,
      "step": 16485
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4744030237197876,
      "learning_rate": 8.281236974464058e-06,
      "loss": 0.5401,
      "step": 16486
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6378381252288818,
      "learning_rate": 8.280137027416066e-06,
      "loss": 0.4394,
      "step": 16487
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7330310344696045,
      "learning_rate": 8.279037101810254e-06,
      "loss": 0.4672,
      "step": 16488
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5462995767593384,
      "learning_rate": 8.277937197660334e-06,
      "loss": 0.6166,
      "step": 16489
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.78671133518219,
      "learning_rate": 8.276837314980016e-06,
      "loss": 0.5857,
      "step": 16490
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8889554738998413,
      "learning_rate": 8.275737453783011e-06,
      "loss": 0.538,
      "step": 16491
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5324276685714722,
      "learning_rate": 8.274637614083039e-06,
      "loss": 0.6308,
      "step": 16492
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8758296966552734,
      "learning_rate": 8.273537795893805e-06,
      "loss": 0.5644,
      "step": 16493
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.336192011833191,
      "learning_rate": 8.272437999229025e-06,
      "loss": 0.5364,
      "step": 16494
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.473708152770996,
      "learning_rate": 8.271338224102406e-06,
      "loss": 0.5826,
      "step": 16495
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4859780073165894,
      "learning_rate": 8.270238470527664e-06,
      "loss": 0.3453,
      "step": 16496
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6274302005767822,
      "learning_rate": 8.26913873851851e-06,
      "loss": 0.4168,
      "step": 16497
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5942827463150024,
      "learning_rate": 8.268039028088649e-06,
      "loss": 0.5862,
      "step": 16498
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7189199924468994,
      "learning_rate": 8.2669393392518e-06,
      "loss": 0.6499,
      "step": 16499
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.516837477684021,
      "learning_rate": 8.265839672021664e-06,
      "loss": 0.5782,
      "step": 16500
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3585550785064697,
      "learning_rate": 8.264740026411958e-06,
      "loss": 0.6031,
      "step": 16501
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5044491291046143,
      "learning_rate": 8.263640402436391e-06,
      "loss": 0.5456,
      "step": 16502
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6401734352111816,
      "learning_rate": 8.262540800108667e-06,
      "loss": 0.5765,
      "step": 16503
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5736123323440552,
      "learning_rate": 8.2614412194425e-06,
      "loss": 0.5657,
      "step": 16504
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.0407609939575195,
      "learning_rate": 8.260341660451601e-06,
      "loss": 0.6301,
      "step": 16505
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.570230484008789,
      "learning_rate": 8.259242123149673e-06,
      "loss": 0.4898,
      "step": 16506
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4626222848892212,
      "learning_rate": 8.25814260755043e-06,
      "loss": 0.5731,
      "step": 16507
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4600095748901367,
      "learning_rate": 8.257043113667573e-06,
      "loss": 0.5485,
      "step": 16508
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4571009874343872,
      "learning_rate": 8.255943641514815e-06,
      "loss": 0.4942,
      "step": 16509
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.496310830116272,
      "learning_rate": 8.254844191105862e-06,
      "loss": 0.5549,
      "step": 16510
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3818247318267822,
      "learning_rate": 8.253744762454421e-06,
      "loss": 0.5023,
      "step": 16511
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6379776000976562,
      "learning_rate": 8.252645355574202e-06,
      "loss": 0.5299,
      "step": 16512
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5498498678207397,
      "learning_rate": 8.251545970478908e-06,
      "loss": 0.5668,
      "step": 16513
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.62161123752594,
      "learning_rate": 8.250446607182243e-06,
      "loss": 0.6148,
      "step": 16514
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6894792318344116,
      "learning_rate": 8.249347265697923e-06,
      "loss": 0.5111,
      "step": 16515
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2798792123794556,
      "learning_rate": 8.248247946039642e-06,
      "loss": 0.5526,
      "step": 16516
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4601387977600098,
      "learning_rate": 8.247148648221113e-06,
      "loss": 0.5255,
      "step": 16517
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4904841184616089,
      "learning_rate": 8.246049372256043e-06,
      "loss": 0.64,
      "step": 16518
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.34828782081604,
      "learning_rate": 8.24495011815813e-06,
      "loss": 0.5567,
      "step": 16519
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9915586709976196,
      "learning_rate": 8.243850885941086e-06,
      "loss": 0.5441,
      "step": 16520
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4719549417495728,
      "learning_rate": 8.24275167561861e-06,
      "loss": 0.5924,
      "step": 16521
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.4258618354797363,
      "learning_rate": 8.241652487204411e-06,
      "loss": 0.7378,
      "step": 16522
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6950992345809937,
      "learning_rate": 8.240553320712188e-06,
      "loss": 0.5552,
      "step": 16523
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6724705696105957,
      "learning_rate": 8.239454176155649e-06,
      "loss": 0.4896,
      "step": 16524
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9606823921203613,
      "learning_rate": 8.238355053548492e-06,
      "loss": 0.5208,
      "step": 16525
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6205209493637085,
      "learning_rate": 8.237255952904426e-06,
      "loss": 0.5073,
      "step": 16526
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4624494314193726,
      "learning_rate": 8.23615687423715e-06,
      "loss": 0.4912,
      "step": 16527
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7131009101867676,
      "learning_rate": 8.23505781756037e-06,
      "loss": 0.4613,
      "step": 16528
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6915123462677002,
      "learning_rate": 8.233958782887787e-06,
      "loss": 0.4731,
      "step": 16529
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.705057978630066,
      "learning_rate": 8.232859770233102e-06,
      "loss": 0.5254,
      "step": 16530
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.278896689414978,
      "learning_rate": 8.231760779610016e-06,
      "loss": 0.5057,
      "step": 16531
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2765491008758545,
      "learning_rate": 8.230661811032235e-06,
      "loss": 0.4259,
      "step": 16532
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6335127353668213,
      "learning_rate": 8.229562864513453e-06,
      "loss": 0.5084,
      "step": 16533
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6966958045959473,
      "learning_rate": 8.22846394006738e-06,
      "loss": 0.6257,
      "step": 16534
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8078726530075073,
      "learning_rate": 8.227365037707709e-06,
      "loss": 0.5706,
      "step": 16535
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3656494617462158,
      "learning_rate": 8.226266157448144e-06,
      "loss": 0.4512,
      "step": 16536
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2825429439544678,
      "learning_rate": 8.225167299302385e-06,
      "loss": 0.6169,
      "step": 16537
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4007576704025269,
      "learning_rate": 8.224068463284132e-06,
      "loss": 0.5913,
      "step": 16538
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5778084993362427,
      "learning_rate": 8.222969649407085e-06,
      "loss": 0.5077,
      "step": 16539
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5186798572540283,
      "learning_rate": 8.22187085768494e-06,
      "loss": 0.4738,
      "step": 16540
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.413482666015625,
      "learning_rate": 8.220772088131397e-06,
      "loss": 0.4237,
      "step": 16541
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.453521490097046,
      "learning_rate": 8.219673340760158e-06,
      "loss": 0.4192,
      "step": 16542
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.422166109085083,
      "learning_rate": 8.218574615584921e-06,
      "loss": 0.5138,
      "step": 16543
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.643855333328247,
      "learning_rate": 8.217475912619381e-06,
      "loss": 0.6621,
      "step": 16544
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.503359317779541,
      "learning_rate": 8.21637723187724e-06,
      "loss": 0.4893,
      "step": 16545
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7938992977142334,
      "learning_rate": 8.215278573372191e-06,
      "loss": 0.5553,
      "step": 16546
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.923967957496643,
      "learning_rate": 8.214179937117936e-06,
      "loss": 0.5949,
      "step": 16547
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6576814651489258,
      "learning_rate": 8.213081323128169e-06,
      "loss": 0.429,
      "step": 16548
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6096196174621582,
      "learning_rate": 8.211982731416588e-06,
      "loss": 0.5628,
      "step": 16549
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7251578569412231,
      "learning_rate": 8.21088416199689e-06,
      "loss": 0.4905,
      "step": 16550
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3659404516220093,
      "learning_rate": 8.20978561488277e-06,
      "loss": 0.5524,
      "step": 16551
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8158403635025024,
      "learning_rate": 8.208687090087925e-06,
      "loss": 0.5639,
      "step": 16552
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.756304144859314,
      "learning_rate": 8.207588587626049e-06,
      "loss": 0.5696,
      "step": 16553
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4802112579345703,
      "learning_rate": 8.20649010751084e-06,
      "loss": 0.5549,
      "step": 16554
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.8118294477462769,
      "learning_rate": 8.205391649755994e-06,
      "loss": 0.506,
      "step": 16555
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4436376094818115,
      "learning_rate": 8.2042932143752e-06,
      "loss": 0.4872,
      "step": 16556
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.592248558998108,
      "learning_rate": 8.203194801382159e-06,
      "loss": 0.5106,
      "step": 16557
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3431655168533325,
      "learning_rate": 8.202096410790562e-06,
      "loss": 0.5001,
      "step": 16558
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3812624216079712,
      "learning_rate": 8.200998042614103e-06,
      "loss": 0.5118,
      "step": 16559
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6991519927978516,
      "learning_rate": 8.19989969686648e-06,
      "loss": 0.4458,
      "step": 16560
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5288375616073608,
      "learning_rate": 8.198801373561379e-06,
      "loss": 0.4961,
      "step": 16561
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.459729790687561,
      "learning_rate": 8.197703072712499e-06,
      "loss": 0.5068,
      "step": 16562
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7557954788208008,
      "learning_rate": 8.196604794333532e-06,
      "loss": 0.5592,
      "step": 16563
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.3068838119506836,
      "learning_rate": 8.19550653843817e-06,
      "loss": 0.5548,
      "step": 16564
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9339187145233154,
      "learning_rate": 8.194408305040104e-06,
      "loss": 0.5555,
      "step": 16565
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9778276681900024,
      "learning_rate": 8.193310094153029e-06,
      "loss": 0.5369,
      "step": 16566
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5534892082214355,
      "learning_rate": 8.192211905790633e-06,
      "loss": 0.586,
      "step": 16567
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4574118852615356,
      "learning_rate": 8.191113739966611e-06,
      "loss": 0.4062,
      "step": 16568
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5332640409469604,
      "learning_rate": 8.190015596694652e-06,
      "loss": 0.5349,
      "step": 16569
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.942733645439148,
      "learning_rate": 8.188917475988449e-06,
      "loss": 0.5943,
      "step": 16570
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3676073551177979,
      "learning_rate": 8.187819377861693e-06,
      "loss": 0.6452,
      "step": 16571
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.368971586227417,
      "learning_rate": 8.18672130232807e-06,
      "loss": 0.5367,
      "step": 16572
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3800636529922485,
      "learning_rate": 8.185623249401275e-06,
      "loss": 0.4402,
      "step": 16573
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5452873706817627,
      "learning_rate": 8.184525219094995e-06,
      "loss": 0.4841,
      "step": 16574
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.7881357669830322,
      "learning_rate": 8.18342721142292e-06,
      "loss": 0.5701,
      "step": 16575
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5239218473434448,
      "learning_rate": 8.182329226398741e-06,
      "loss": 0.4641,
      "step": 16576
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5739989280700684,
      "learning_rate": 8.181231264036145e-06,
      "loss": 0.4485,
      "step": 16577
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6465439796447754,
      "learning_rate": 8.180133324348823e-06,
      "loss": 0.5372,
      "step": 16578
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5334383249282837,
      "learning_rate": 8.17903540735046e-06,
      "loss": 0.5431,
      "step": 16579
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.9991347789764404,
      "learning_rate": 8.177937513054746e-06,
      "loss": 0.5547,
      "step": 16580
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.292516827583313,
      "learning_rate": 8.17683964147537e-06,
      "loss": 0.4671,
      "step": 16581
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.359190583229065,
      "learning_rate": 8.175741792626018e-06,
      "loss": 0.5058,
      "step": 16582
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.9472371339797974,
      "learning_rate": 8.174643966520377e-06,
      "loss": 0.5306,
      "step": 16583
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5914995670318604,
      "learning_rate": 8.173546163172137e-06,
      "loss": 0.4907,
      "step": 16584
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6740355491638184,
      "learning_rate": 8.17244838259498e-06,
      "loss": 0.5612,
      "step": 16585
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.633673071861267,
      "learning_rate": 8.171350624802599e-06,
      "loss": 0.5897,
      "step": 16586
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7740224599838257,
      "learning_rate": 8.170252889808673e-06,
      "loss": 0.5126,
      "step": 16587
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4732017517089844,
      "learning_rate": 8.16915517762689e-06,
      "loss": 0.5614,
      "step": 16588
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5446219444274902,
      "learning_rate": 8.16805748827094e-06,
      "loss": 0.5464,
      "step": 16589
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5323623418807983,
      "learning_rate": 8.1669598217545e-06,
      "loss": 0.4809,
      "step": 16590
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5332776308059692,
      "learning_rate": 8.165862178091264e-06,
      "loss": 0.5303,
      "step": 16591
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.686600923538208,
      "learning_rate": 8.164764557294911e-06,
      "loss": 0.5691,
      "step": 16592
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4493337869644165,
      "learning_rate": 8.163666959379128e-06,
      "loss": 0.4341,
      "step": 16593
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4700613021850586,
      "learning_rate": 8.162569384357599e-06,
      "loss": 0.4723,
      "step": 16594
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8380794525146484,
      "learning_rate": 8.161471832244005e-06,
      "loss": 0.5406,
      "step": 16595
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.430037021636963,
      "learning_rate": 8.160374303052032e-06,
      "loss": 0.4343,
      "step": 16596
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5197570323944092,
      "learning_rate": 8.159276796795363e-06,
      "loss": 0.5123,
      "step": 16597
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8108713626861572,
      "learning_rate": 8.158179313487682e-06,
      "loss": 0.4514,
      "step": 16598
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7369890213012695,
      "learning_rate": 8.157081853142672e-06,
      "loss": 0.6165,
      "step": 16599
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4748233556747437,
      "learning_rate": 8.155984415774012e-06,
      "loss": 0.4933,
      "step": 16600
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.415308952331543,
      "learning_rate": 8.154887001395387e-06,
      "loss": 0.5632,
      "step": 16601
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.750951886177063,
      "learning_rate": 8.153789610020479e-06,
      "loss": 0.4685,
      "step": 16602
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1963852643966675,
      "learning_rate": 8.152692241662968e-06,
      "loss": 0.3804,
      "step": 16603
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1548975706100464,
      "learning_rate": 8.151594896336536e-06,
      "loss": 0.4002,
      "step": 16604
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.254384160041809,
      "learning_rate": 8.150497574054865e-06,
      "loss": 0.4383,
      "step": 16605
    },
    {
      "epoch": 1.72,
      "grad_norm": 2.0395185947418213,
      "learning_rate": 8.149400274831634e-06,
      "loss": 0.5313,
      "step": 16606
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.639874815940857,
      "learning_rate": 8.148302998680525e-06,
      "loss": 0.5857,
      "step": 16607
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7091671228408813,
      "learning_rate": 8.147205745615218e-06,
      "loss": 0.5872,
      "step": 16608
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4886581897735596,
      "learning_rate": 8.14610851564939e-06,
      "loss": 0.545,
      "step": 16609
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4122228622436523,
      "learning_rate": 8.145011308796728e-06,
      "loss": 0.417,
      "step": 16610
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5435205698013306,
      "learning_rate": 8.143914125070901e-06,
      "loss": 0.5215,
      "step": 16611
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.59722101688385,
      "learning_rate": 8.142816964485598e-06,
      "loss": 0.4081,
      "step": 16612
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5470528602600098,
      "learning_rate": 8.14171982705449e-06,
      "loss": 0.5844,
      "step": 16613
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.507860541343689,
      "learning_rate": 8.140622712791257e-06,
      "loss": 0.5637,
      "step": 16614
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6230802536010742,
      "learning_rate": 8.139525621709582e-06,
      "loss": 0.4739,
      "step": 16615
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.795111060142517,
      "learning_rate": 8.138428553823136e-06,
      "loss": 0.6436,
      "step": 16616
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1773630380630493,
      "learning_rate": 8.137331509145605e-06,
      "loss": 0.5515,
      "step": 16617
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3494094610214233,
      "learning_rate": 8.136234487690656e-06,
      "loss": 0.5468,
      "step": 16618
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4956107139587402,
      "learning_rate": 8.135137489471972e-06,
      "loss": 0.4576,
      "step": 16619
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7474833726882935,
      "learning_rate": 8.134040514503231e-06,
      "loss": 0.5451,
      "step": 16620
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.732667088508606,
      "learning_rate": 8.132943562798105e-06,
      "loss": 0.4533,
      "step": 16621
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7248144149780273,
      "learning_rate": 8.131846634370272e-06,
      "loss": 0.5266,
      "step": 16622
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.350786805152893,
      "learning_rate": 8.130749729233413e-06,
      "loss": 0.4543,
      "step": 16623
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7506555318832397,
      "learning_rate": 8.129652847401193e-06,
      "loss": 0.5823,
      "step": 16624
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6683963537216187,
      "learning_rate": 8.128555988887297e-06,
      "loss": 0.4824,
      "step": 16625
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5736671686172485,
      "learning_rate": 8.127459153705395e-06,
      "loss": 0.5682,
      "step": 16626
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.9090418815612793,
      "learning_rate": 8.12636234186916e-06,
      "loss": 0.5085,
      "step": 16627
    },
    {
      "epoch": 1.72,
      "grad_norm": 2.0835113525390625,
      "learning_rate": 8.125265553392272e-06,
      "loss": 0.564,
      "step": 16628
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.659552812576294,
      "learning_rate": 8.1241687882884e-06,
      "loss": 0.603,
      "step": 16629
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8174506425857544,
      "learning_rate": 8.123072046571224e-06,
      "loss": 0.5527,
      "step": 16630
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4464783668518066,
      "learning_rate": 8.121975328254408e-06,
      "loss": 0.51,
      "step": 16631
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5791900157928467,
      "learning_rate": 8.12087863335163e-06,
      "loss": 0.5384,
      "step": 16632
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4608708620071411,
      "learning_rate": 8.119781961876569e-06,
      "loss": 0.5643,
      "step": 16633
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.9387136697769165,
      "learning_rate": 8.118685313842886e-06,
      "loss": 0.5524,
      "step": 16634
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4572237730026245,
      "learning_rate": 8.117588689264262e-06,
      "loss": 0.5868,
      "step": 16635
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7084453105926514,
      "learning_rate": 8.116492088154369e-06,
      "loss": 0.5313,
      "step": 16636
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5668355226516724,
      "learning_rate": 8.115395510526871e-06,
      "loss": 0.509,
      "step": 16637
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3238040208816528,
      "learning_rate": 8.114298956395449e-06,
      "loss": 0.5164,
      "step": 16638
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5986912250518799,
      "learning_rate": 8.113202425773767e-06,
      "loss": 0.5283,
      "step": 16639
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7715725898742676,
      "learning_rate": 8.112105918675499e-06,
      "loss": 0.6166,
      "step": 16640
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6807007789611816,
      "learning_rate": 8.111009435114317e-06,
      "loss": 0.5042,
      "step": 16641
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.433977484703064,
      "learning_rate": 8.109912975103886e-06,
      "loss": 0.5387,
      "step": 16642
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7536507844924927,
      "learning_rate": 8.108816538657882e-06,
      "loss": 0.5674,
      "step": 16643
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4976685047149658,
      "learning_rate": 8.107720125789971e-06,
      "loss": 0.5748,
      "step": 16644
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8654065132141113,
      "learning_rate": 8.106623736513822e-06,
      "loss": 0.5487,
      "step": 16645
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.459977149963379,
      "learning_rate": 8.10552737084311e-06,
      "loss": 0.4645,
      "step": 16646
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8500713109970093,
      "learning_rate": 8.104431028791495e-06,
      "loss": 0.5492,
      "step": 16647
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.768229603767395,
      "learning_rate": 8.10333471037265e-06,
      "loss": 0.583,
      "step": 16648
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6460622549057007,
      "learning_rate": 8.102238415600245e-06,
      "loss": 0.6012,
      "step": 16649
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7157715559005737,
      "learning_rate": 8.101142144487945e-06,
      "loss": 0.5242,
      "step": 16650
    },
    {
      "epoch": 1.72,
      "grad_norm": 2.128619432449341,
      "learning_rate": 8.10004589704942e-06,
      "loss": 0.4974,
      "step": 16651
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5169750452041626,
      "learning_rate": 8.098949673298335e-06,
      "loss": 0.374,
      "step": 16652
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5298056602478027,
      "learning_rate": 8.097853473248355e-06,
      "loss": 0.5325,
      "step": 16653
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.9396008253097534,
      "learning_rate": 8.096757296913155e-06,
      "loss": 0.6261,
      "step": 16654
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4212535619735718,
      "learning_rate": 8.095661144306392e-06,
      "loss": 0.4442,
      "step": 16655
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7158340215682983,
      "learning_rate": 8.09456501544174e-06,
      "loss": 0.5409,
      "step": 16656
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5298430919647217,
      "learning_rate": 8.093468910332858e-06,
      "loss": 0.601,
      "step": 16657
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8051925897598267,
      "learning_rate": 8.092372828993416e-06,
      "loss": 0.5542,
      "step": 16658
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6307708024978638,
      "learning_rate": 8.091276771437077e-06,
      "loss": 0.5847,
      "step": 16659
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7002593278884888,
      "learning_rate": 8.090180737677507e-06,
      "loss": 0.5289,
      "step": 16660
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5126416683197021,
      "learning_rate": 8.089084727728372e-06,
      "loss": 0.5249,
      "step": 16661
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4978373050689697,
      "learning_rate": 8.087988741603334e-06,
      "loss": 0.5529,
      "step": 16662
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6699824333190918,
      "learning_rate": 8.086892779316058e-06,
      "loss": 0.4407,
      "step": 16663
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5303441286087036,
      "learning_rate": 8.085796840880209e-06,
      "loss": 0.4567,
      "step": 16664
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6001029014587402,
      "learning_rate": 8.084700926309448e-06,
      "loss": 0.5466,
      "step": 16665
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.9189175367355347,
      "learning_rate": 8.083605035617438e-06,
      "loss": 0.5396,
      "step": 16666
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6722102165222168,
      "learning_rate": 8.082509168817846e-06,
      "loss": 0.497,
      "step": 16667
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6505153179168701,
      "learning_rate": 8.08141332592433e-06,
      "loss": 0.5514,
      "step": 16668
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8527238368988037,
      "learning_rate": 8.080317506950557e-06,
      "loss": 0.5148,
      "step": 16669
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5727804899215698,
      "learning_rate": 8.079221711910182e-06,
      "loss": 0.5656,
      "step": 16670
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.342119812965393,
      "learning_rate": 8.078125940816872e-06,
      "loss": 0.5225,
      "step": 16671
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.392577886581421,
      "learning_rate": 8.07703019368429e-06,
      "loss": 0.4927,
      "step": 16672
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.445152759552002,
      "learning_rate": 8.075934470526092e-06,
      "loss": 0.6328,
      "step": 16673
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5214972496032715,
      "learning_rate": 8.07483877135594e-06,
      "loss": 0.5191,
      "step": 16674
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7127877473831177,
      "learning_rate": 8.073743096187499e-06,
      "loss": 0.4494,
      "step": 16675
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.8828388452529907,
      "learning_rate": 8.072647445034424e-06,
      "loss": 0.5023,
      "step": 16676
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.727919340133667,
      "learning_rate": 8.071551817910377e-06,
      "loss": 0.5054,
      "step": 16677
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.9989367723464966,
      "learning_rate": 8.070456214829017e-06,
      "loss": 0.5287,
      "step": 16678
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.5385164022445679,
      "learning_rate": 8.069360635804003e-06,
      "loss": 0.5396,
      "step": 16679
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3797601461410522,
      "learning_rate": 8.068265080848996e-06,
      "loss": 0.489,
      "step": 16680
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.196882963180542,
      "learning_rate": 8.067169549977652e-06,
      "loss": 0.4849,
      "step": 16681
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7060546875,
      "learning_rate": 8.066074043203633e-06,
      "loss": 0.5627,
      "step": 16682
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.529770851135254,
      "learning_rate": 8.064978560540592e-06,
      "loss": 0.4856,
      "step": 16683
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6565805673599243,
      "learning_rate": 8.06388310200219e-06,
      "loss": 0.5988,
      "step": 16684
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7626228332519531,
      "learning_rate": 8.062787667602085e-06,
      "loss": 0.6333,
      "step": 16685
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5305862426757812,
      "learning_rate": 8.061692257353933e-06,
      "loss": 0.5176,
      "step": 16686
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5248719453811646,
      "learning_rate": 8.06059687127139e-06,
      "loss": 0.5014,
      "step": 16687
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5333718061447144,
      "learning_rate": 8.059501509368115e-06,
      "loss": 0.6125,
      "step": 16688
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.288784384727478,
      "learning_rate": 8.058406171657763e-06,
      "loss": 0.4835,
      "step": 16689
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4609451293945312,
      "learning_rate": 8.057310858153992e-06,
      "loss": 0.5188,
      "step": 16690
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.163752794265747,
      "learning_rate": 8.056215568870451e-06,
      "loss": 0.583,
      "step": 16691
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8007932901382446,
      "learning_rate": 8.055120303820803e-06,
      "loss": 0.5944,
      "step": 16692
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8890061378479004,
      "learning_rate": 8.054025063018702e-06,
      "loss": 0.5214,
      "step": 16693
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3030074834823608,
      "learning_rate": 8.052929846477797e-06,
      "loss": 0.5316,
      "step": 16694
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9056446552276611,
      "learning_rate": 8.05183465421175e-06,
      "loss": 0.6532,
      "step": 16695
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.0100808143615723,
      "learning_rate": 8.050739486234207e-06,
      "loss": 0.6312,
      "step": 16696
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2837547063827515,
      "learning_rate": 8.049644342558828e-06,
      "loss": 0.4389,
      "step": 16697
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.774450659751892,
      "learning_rate": 8.048549223199268e-06,
      "loss": 0.5346,
      "step": 16698
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.0519793033599854,
      "learning_rate": 8.047454128169175e-06,
      "loss": 0.5541,
      "step": 16699
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4777477979660034,
      "learning_rate": 8.046359057482205e-06,
      "loss": 0.517,
      "step": 16700
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.574693202972412,
      "learning_rate": 8.04526401115201e-06,
      "loss": 0.5224,
      "step": 16701
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2539544105529785,
      "learning_rate": 8.044168989192241e-06,
      "loss": 0.5,
      "step": 16702
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5808392763137817,
      "learning_rate": 8.043073991616554e-06,
      "loss": 0.4687,
      "step": 16703
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5994948148727417,
      "learning_rate": 8.041979018438596e-06,
      "loss": 0.5086,
      "step": 16704
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5054572820663452,
      "learning_rate": 8.04088406967202e-06,
      "loss": 0.4497,
      "step": 16705
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7117664813995361,
      "learning_rate": 8.03978914533048e-06,
      "loss": 0.5056,
      "step": 16706
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.525098204612732,
      "learning_rate": 8.03869424542762e-06,
      "loss": 0.677,
      "step": 16707
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6052390336990356,
      "learning_rate": 8.0375993699771e-06,
      "loss": 0.5517,
      "step": 16708
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7718747854232788,
      "learning_rate": 8.036504518992561e-06,
      "loss": 0.6021,
      "step": 16709
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2991811037063599,
      "learning_rate": 8.035409692487658e-06,
      "loss": 0.6049,
      "step": 16710
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.651335597038269,
      "learning_rate": 8.034314890476043e-06,
      "loss": 0.573,
      "step": 16711
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6924744844436646,
      "learning_rate": 8.03322011297136e-06,
      "loss": 0.5647,
      "step": 16712
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5998902320861816,
      "learning_rate": 8.032125359987258e-06,
      "loss": 0.4863,
      "step": 16713
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5909099578857422,
      "learning_rate": 8.031030631537391e-06,
      "loss": 0.5358,
      "step": 16714
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.858647346496582,
      "learning_rate": 8.029935927635403e-06,
      "loss": 0.5936,
      "step": 16715
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7162630558013916,
      "learning_rate": 8.028841248294944e-06,
      "loss": 0.5585,
      "step": 16716
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4925391674041748,
      "learning_rate": 8.02774659352966e-06,
      "loss": 0.6123,
      "step": 16717
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4357041120529175,
      "learning_rate": 8.026651963353198e-06,
      "loss": 0.5733,
      "step": 16718
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3487908840179443,
      "learning_rate": 8.02555735777921e-06,
      "loss": 0.5798,
      "step": 16719
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4440932273864746,
      "learning_rate": 8.024462776821336e-06,
      "loss": 0.5634,
      "step": 16720
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4474738836288452,
      "learning_rate": 8.02336822049323e-06,
      "loss": 0.5966,
      "step": 16721
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4261773824691772,
      "learning_rate": 8.02227368880853e-06,
      "loss": 0.5355,
      "step": 16722
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9168906211853027,
      "learning_rate": 8.021179181780888e-06,
      "loss": 0.5504,
      "step": 16723
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2747528553009033,
      "learning_rate": 8.02008469942395e-06,
      "loss": 0.4168,
      "step": 16724
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3505879640579224,
      "learning_rate": 8.018990241751357e-06,
      "loss": 0.4872,
      "step": 16725
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2979379892349243,
      "learning_rate": 8.017895808776756e-06,
      "loss": 0.4517,
      "step": 16726
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6874409914016724,
      "learning_rate": 8.016801400513793e-06,
      "loss": 0.5122,
      "step": 16727
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4657524824142456,
      "learning_rate": 8.01570701697611e-06,
      "loss": 0.4786,
      "step": 16728
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5347881317138672,
      "learning_rate": 8.014612658177354e-06,
      "loss": 0.5289,
      "step": 16729
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6336026191711426,
      "learning_rate": 8.013518324131167e-06,
      "loss": 0.5285,
      "step": 16730
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6710395812988281,
      "learning_rate": 8.01242401485119e-06,
      "loss": 0.5128,
      "step": 16731
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4581058025360107,
      "learning_rate": 8.011329730351072e-06,
      "loss": 0.4907,
      "step": 16732
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2232388257980347,
      "learning_rate": 8.01023547064445e-06,
      "loss": 0.4462,
      "step": 16733
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7025407552719116,
      "learning_rate": 8.009141235744973e-06,
      "loss": 0.5259,
      "step": 16734
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5641053915023804,
      "learning_rate": 8.008047025666276e-06,
      "loss": 0.5064,
      "step": 16735
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.492324709892273,
      "learning_rate": 8.006952840422006e-06,
      "loss": 0.41,
      "step": 16736
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9273309707641602,
      "learning_rate": 8.005858680025803e-06,
      "loss": 0.5476,
      "step": 16737
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8869152069091797,
      "learning_rate": 8.004764544491306e-06,
      "loss": 0.6094,
      "step": 16738
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7073959112167358,
      "learning_rate": 8.00367043383216e-06,
      "loss": 0.5149,
      "step": 16739
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.581708312034607,
      "learning_rate": 8.002576348062006e-06,
      "loss": 0.5181,
      "step": 16740
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5710967779159546,
      "learning_rate": 8.00148228719448e-06,
      "loss": 0.5418,
      "step": 16741
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.8581523895263672,
      "learning_rate": 8.000388251243226e-06,
      "loss": 0.5302,
      "step": 16742
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5576037168502808,
      "learning_rate": 7.99929424022188e-06,
      "loss": 0.5477,
      "step": 16743
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.412160873413086,
      "learning_rate": 7.998200254144085e-06,
      "loss": 0.4523,
      "step": 16744
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6798696517944336,
      "learning_rate": 7.997106293023479e-06,
      "loss": 0.5866,
      "step": 16745
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.49453866481781,
      "learning_rate": 7.996012356873699e-06,
      "loss": 0.4349,
      "step": 16746
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5539827346801758,
      "learning_rate": 7.994918445708389e-06,
      "loss": 0.5205,
      "step": 16747
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5779260396957397,
      "learning_rate": 7.99382455954118e-06,
      "loss": 0.5386,
      "step": 16748
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5096096992492676,
      "learning_rate": 7.992730698385712e-06,
      "loss": 0.5908,
      "step": 16749
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5337724685668945,
      "learning_rate": 7.991636862255627e-06,
      "loss": 0.5099,
      "step": 16750
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6836459636688232,
      "learning_rate": 7.990543051164556e-06,
      "loss": 0.591,
      "step": 16751
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6618422269821167,
      "learning_rate": 7.989449265126141e-06,
      "loss": 0.6427,
      "step": 16752
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5976455211639404,
      "learning_rate": 7.988355504154017e-06,
      "loss": 0.4699,
      "step": 16753
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7737606763839722,
      "learning_rate": 7.98726176826182e-06,
      "loss": 0.555,
      "step": 16754
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3500293493270874,
      "learning_rate": 7.986168057463187e-06,
      "loss": 0.5429,
      "step": 16755
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9620678424835205,
      "learning_rate": 7.985074371771749e-06,
      "loss": 0.5678,
      "step": 16756
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.585078477859497,
      "learning_rate": 7.983980711201145e-06,
      "loss": 0.53,
      "step": 16757
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6193068027496338,
      "learning_rate": 7.982887075765014e-06,
      "loss": 0.5452,
      "step": 16758
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2595281600952148,
      "learning_rate": 7.981793465476984e-06,
      "loss": 0.4759,
      "step": 16759
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6701582670211792,
      "learning_rate": 7.980699880350695e-06,
      "loss": 0.6508,
      "step": 16760
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4051393270492554,
      "learning_rate": 7.979606320399776e-06,
      "loss": 0.5782,
      "step": 16761
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.565732479095459,
      "learning_rate": 7.978512785637864e-06,
      "loss": 0.5291,
      "step": 16762
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7051358222961426,
      "learning_rate": 7.977419276078595e-06,
      "loss": 0.5314,
      "step": 16763
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.662755012512207,
      "learning_rate": 7.976325791735594e-06,
      "loss": 0.4777,
      "step": 16764
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4293193817138672,
      "learning_rate": 7.9752323326225e-06,
      "loss": 0.499,
      "step": 16765
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3833422660827637,
      "learning_rate": 7.974138898752948e-06,
      "loss": 0.5217,
      "step": 16766
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5019503831863403,
      "learning_rate": 7.973045490140563e-06,
      "loss": 0.5709,
      "step": 16767
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7080508470535278,
      "learning_rate": 7.971952106798984e-06,
      "loss": 0.5093,
      "step": 16768
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9974666833877563,
      "learning_rate": 7.970858748741837e-06,
      "loss": 0.6241,
      "step": 16769
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4864094257354736,
      "learning_rate": 7.969765415982755e-06,
      "loss": 0.5258,
      "step": 16770
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.457797884941101,
      "learning_rate": 7.968672108535372e-06,
      "loss": 0.5546,
      "step": 16771
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.43479585647583,
      "learning_rate": 7.967578826413312e-06,
      "loss": 0.5043,
      "step": 16772
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.6293597221374512,
      "learning_rate": 7.966485569630214e-06,
      "loss": 0.5094,
      "step": 16773
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.0669658184051514,
      "learning_rate": 7.965392338199702e-06,
      "loss": 0.5787,
      "step": 16774
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.845319390296936,
      "learning_rate": 7.964299132135405e-06,
      "loss": 0.561,
      "step": 16775
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7124853134155273,
      "learning_rate": 7.96320595145096e-06,
      "loss": 0.5626,
      "step": 16776
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.524402141571045,
      "learning_rate": 7.962112796159987e-06,
      "loss": 0.4641,
      "step": 16777
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8194186687469482,
      "learning_rate": 7.961019666276119e-06,
      "loss": 0.5715,
      "step": 16778
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3535970449447632,
      "learning_rate": 7.959926561812986e-06,
      "loss": 0.5443,
      "step": 16779
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.503013253211975,
      "learning_rate": 7.958833482784212e-06,
      "loss": 0.5041,
      "step": 16780
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8098810911178589,
      "learning_rate": 7.95774042920343e-06,
      "loss": 0.6293,
      "step": 16781
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4560431241989136,
      "learning_rate": 7.956647401084262e-06,
      "loss": 0.5045,
      "step": 16782
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9661258459091187,
      "learning_rate": 7.955554398440337e-06,
      "loss": 0.5772,
      "step": 16783
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4484606981277466,
      "learning_rate": 7.954461421285287e-06,
      "loss": 0.4061,
      "step": 16784
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4963982105255127,
      "learning_rate": 7.95336846963273e-06,
      "loss": 0.6521,
      "step": 16785
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6718145608901978,
      "learning_rate": 7.9522755434963e-06,
      "loss": 0.4572,
      "step": 16786
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8673858642578125,
      "learning_rate": 7.951182642889615e-06,
      "loss": 0.4758,
      "step": 16787
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.354217767715454,
      "learning_rate": 7.950089767826307e-06,
      "loss": 0.5323,
      "step": 16788
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.93147611618042,
      "learning_rate": 7.948996918320001e-06,
      "loss": 0.5902,
      "step": 16789
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6196328401565552,
      "learning_rate": 7.947904094384317e-06,
      "loss": 0.5142,
      "step": 16790
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.973081111907959,
      "learning_rate": 7.946811296032885e-06,
      "loss": 0.6717,
      "step": 16791
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4016985893249512,
      "learning_rate": 7.945718523279328e-06,
      "loss": 0.5192,
      "step": 16792
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3467223644256592,
      "learning_rate": 7.944625776137267e-06,
      "loss": 0.5065,
      "step": 16793
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5978333950042725,
      "learning_rate": 7.94353305462033e-06,
      "loss": 0.5538,
      "step": 16794
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.583035945892334,
      "learning_rate": 7.942440358742138e-06,
      "loss": 0.5103,
      "step": 16795
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4404934644699097,
      "learning_rate": 7.941347688516311e-06,
      "loss": 0.5161,
      "step": 16796
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5523673295974731,
      "learning_rate": 7.94025504395648e-06,
      "loss": 0.4548,
      "step": 16797
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6839778423309326,
      "learning_rate": 7.939162425076259e-06,
      "loss": 0.4523,
      "step": 16798
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4540199041366577,
      "learning_rate": 7.938069831889277e-06,
      "loss": 0.513,
      "step": 16799
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3143119812011719,
      "learning_rate": 7.936977264409148e-06,
      "loss": 0.5186,
      "step": 16800
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.444583773612976,
      "learning_rate": 7.935884722649499e-06,
      "loss": 0.4662,
      "step": 16801
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5757615566253662,
      "learning_rate": 7.934792206623952e-06,
      "loss": 0.4719,
      "step": 16802
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9608547687530518,
      "learning_rate": 7.933699716346123e-06,
      "loss": 0.6728,
      "step": 16803
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6373361349105835,
      "learning_rate": 7.932607251829638e-06,
      "loss": 0.6783,
      "step": 16804
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5470335483551025,
      "learning_rate": 7.931514813088114e-06,
      "loss": 0.463,
      "step": 16805
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4671732187271118,
      "learning_rate": 7.930422400135169e-06,
      "loss": 0.4896,
      "step": 16806
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4986897706985474,
      "learning_rate": 7.929330012984428e-06,
      "loss": 0.5393,
      "step": 16807
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6052181720733643,
      "learning_rate": 7.928237651649503e-06,
      "loss": 0.5879,
      "step": 16808
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4525866508483887,
      "learning_rate": 7.927145316144018e-06,
      "loss": 0.3919,
      "step": 16809
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.536327838897705,
      "learning_rate": 7.926053006481593e-06,
      "loss": 0.5148,
      "step": 16810
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.491702675819397,
      "learning_rate": 7.924960722675839e-06,
      "loss": 0.4578,
      "step": 16811
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6440685987472534,
      "learning_rate": 7.923868464740383e-06,
      "loss": 0.4977,
      "step": 16812
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.473980188369751,
      "learning_rate": 7.922776232688835e-06,
      "loss": 0.4573,
      "step": 16813
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5224719047546387,
      "learning_rate": 7.921684026534817e-06,
      "loss": 0.4379,
      "step": 16814
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.059410333633423,
      "learning_rate": 7.920591846291945e-06,
      "loss": 0.5005,
      "step": 16815
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4230459928512573,
      "learning_rate": 7.919499691973834e-06,
      "loss": 0.4223,
      "step": 16816
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4357500076293945,
      "learning_rate": 7.9184075635941e-06,
      "loss": 0.5999,
      "step": 16817
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3786864280700684,
      "learning_rate": 7.917315461166362e-06,
      "loss": 0.5667,
      "step": 16818
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.691908359527588,
      "learning_rate": 7.916223384704231e-06,
      "loss": 0.5667,
      "step": 16819
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7282218933105469,
      "learning_rate": 7.915131334221328e-06,
      "loss": 0.585,
      "step": 16820
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6368825435638428,
      "learning_rate": 7.914039309731266e-06,
      "loss": 0.483,
      "step": 16821
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5111379623413086,
      "learning_rate": 7.912947311247656e-06,
      "loss": 0.4394,
      "step": 16822
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.245200514793396,
      "learning_rate": 7.911855338784119e-06,
      "loss": 0.4573,
      "step": 16823
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4460917711257935,
      "learning_rate": 7.91076339235426e-06,
      "loss": 0.5693,
      "step": 16824
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5711636543273926,
      "learning_rate": 7.909671471971702e-06,
      "loss": 0.5104,
      "step": 16825
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7216591835021973,
      "learning_rate": 7.908579577650053e-06,
      "loss": 0.6346,
      "step": 16826
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5055049657821655,
      "learning_rate": 7.907487709402925e-06,
      "loss": 0.5537,
      "step": 16827
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4574037790298462,
      "learning_rate": 7.906395867243938e-06,
      "loss": 0.4988,
      "step": 16828
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.9137506484985352,
      "learning_rate": 7.905304051186697e-06,
      "loss": 0.5398,
      "step": 16829
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7942754030227661,
      "learning_rate": 7.904212261244815e-06,
      "loss": 0.7024,
      "step": 16830
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5714352130889893,
      "learning_rate": 7.903120497431908e-06,
      "loss": 0.5546,
      "step": 16831
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.1209981441497803,
      "learning_rate": 7.902028759761584e-06,
      "loss": 0.4681,
      "step": 16832
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7514386177062988,
      "learning_rate": 7.900937048247456e-06,
      "loss": 0.5475,
      "step": 16833
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6645807027816772,
      "learning_rate": 7.899845362903131e-06,
      "loss": 0.6805,
      "step": 16834
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6525503396987915,
      "learning_rate": 7.898753703742223e-06,
      "loss": 0.6097,
      "step": 16835
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7501682043075562,
      "learning_rate": 7.897662070778344e-06,
      "loss": 0.5375,
      "step": 16836
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.602634310722351,
      "learning_rate": 7.896570464025097e-06,
      "loss": 0.5968,
      "step": 16837
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5356024503707886,
      "learning_rate": 7.8954788834961e-06,
      "loss": 0.522,
      "step": 16838
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7226073741912842,
      "learning_rate": 7.894387329204953e-06,
      "loss": 0.5396,
      "step": 16839
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.814054012298584,
      "learning_rate": 7.89329580116527e-06,
      "loss": 0.586,
      "step": 16840
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.580679178237915,
      "learning_rate": 7.892204299390663e-06,
      "loss": 0.6716,
      "step": 16841
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5908629894256592,
      "learning_rate": 7.891112823894734e-06,
      "loss": 0.4987,
      "step": 16842
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4924713373184204,
      "learning_rate": 7.89002137469109e-06,
      "loss": 0.6017,
      "step": 16843
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.8130897283554077,
      "learning_rate": 7.888929951793346e-06,
      "loss": 0.5458,
      "step": 16844
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4738986492156982,
      "learning_rate": 7.887838555215101e-06,
      "loss": 0.549,
      "step": 16845
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.244552493095398,
      "learning_rate": 7.88674718496997e-06,
      "loss": 0.4165,
      "step": 16846
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.150167226791382,
      "learning_rate": 7.885655841071551e-06,
      "loss": 0.583,
      "step": 16847
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.557313084602356,
      "learning_rate": 7.884564523533454e-06,
      "loss": 0.5809,
      "step": 16848
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7108409404754639,
      "learning_rate": 7.883473232369289e-06,
      "loss": 0.5566,
      "step": 16849
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3275688886642456,
      "learning_rate": 7.882381967592653e-06,
      "loss": 0.536,
      "step": 16850
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.835868000984192,
      "learning_rate": 7.88129072921716e-06,
      "loss": 0.6205,
      "step": 16851
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4446630477905273,
      "learning_rate": 7.880199517256409e-06,
      "loss": 0.4087,
      "step": 16852
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1700546741485596,
      "learning_rate": 7.879108331724003e-06,
      "loss": 0.409,
      "step": 16853
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4770493507385254,
      "learning_rate": 7.878017172633554e-06,
      "loss": 0.354,
      "step": 16854
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4700922966003418,
      "learning_rate": 7.87692603999866e-06,
      "loss": 0.5268,
      "step": 16855
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7499724626541138,
      "learning_rate": 7.875834933832924e-06,
      "loss": 0.4838,
      "step": 16856
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.535724401473999,
      "learning_rate": 7.874743854149956e-06,
      "loss": 0.4865,
      "step": 16857
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5980888605117798,
      "learning_rate": 7.87365280096335e-06,
      "loss": 0.5341,
      "step": 16858
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2406314611434937,
      "learning_rate": 7.872561774286713e-06,
      "loss": 0.5161,
      "step": 16859
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5078229904174805,
      "learning_rate": 7.871470774133647e-06,
      "loss": 0.4576,
      "step": 16860
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5075907707214355,
      "learning_rate": 7.870379800517753e-06,
      "loss": 0.4434,
      "step": 16861
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6832071542739868,
      "learning_rate": 7.869288853452637e-06,
      "loss": 0.5727,
      "step": 16862
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5068997144699097,
      "learning_rate": 7.868197932951894e-06,
      "loss": 0.4408,
      "step": 16863
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4046446084976196,
      "learning_rate": 7.867107039029129e-06,
      "loss": 0.5187,
      "step": 16864
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.505126953125,
      "learning_rate": 7.866016171697938e-06,
      "loss": 0.5976,
      "step": 16865
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5733556747436523,
      "learning_rate": 7.864925330971925e-06,
      "loss": 0.5966,
      "step": 16866
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5533709526062012,
      "learning_rate": 7.863834516864692e-06,
      "loss": 0.5461,
      "step": 16867
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3520258665084839,
      "learning_rate": 7.862743729389833e-06,
      "loss": 0.5218,
      "step": 16868
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5059627294540405,
      "learning_rate": 7.86165296856095e-06,
      "loss": 0.5106,
      "step": 16869
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5590816736221313,
      "learning_rate": 7.860562234391645e-06,
      "loss": 0.5268,
      "step": 16870
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.6066814661026,
      "learning_rate": 7.859471526895512e-06,
      "loss": 0.5106,
      "step": 16871
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4958250522613525,
      "learning_rate": 7.858380846086152e-06,
      "loss": 0.5459,
      "step": 16872
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8341716527938843,
      "learning_rate": 7.85729019197716e-06,
      "loss": 0.5326,
      "step": 16873
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.3083419799804688,
      "learning_rate": 7.85619956458214e-06,
      "loss": 0.5272,
      "step": 16874
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6417229175567627,
      "learning_rate": 7.855108963914679e-06,
      "loss": 0.5617,
      "step": 16875
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4000446796417236,
      "learning_rate": 7.854018389988384e-06,
      "loss": 0.386,
      "step": 16876
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.574859619140625,
      "learning_rate": 7.852927842816844e-06,
      "loss": 0.5788,
      "step": 16877
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.789099097251892,
      "learning_rate": 7.85183732241366e-06,
      "loss": 0.5754,
      "step": 16878
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9048960208892822,
      "learning_rate": 7.850746828792425e-06,
      "loss": 0.5942,
      "step": 16879
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.500856637954712,
      "learning_rate": 7.849656361966735e-06,
      "loss": 0.5319,
      "step": 16880
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6083720922470093,
      "learning_rate": 7.848565921950188e-06,
      "loss": 0.5281,
      "step": 16881
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4226325750350952,
      "learning_rate": 7.847475508756376e-06,
      "loss": 0.5678,
      "step": 16882
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.668514609336853,
      "learning_rate": 7.846385122398896e-06,
      "loss": 0.4823,
      "step": 16883
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2980684041976929,
      "learning_rate": 7.845294762891342e-06,
      "loss": 0.5618,
      "step": 16884
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9318914413452148,
      "learning_rate": 7.844204430247304e-06,
      "loss": 0.5862,
      "step": 16885
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.781843900680542,
      "learning_rate": 7.843114124480382e-06,
      "loss": 0.518,
      "step": 16886
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.495226263999939,
      "learning_rate": 7.842023845604164e-06,
      "loss": 0.5781,
      "step": 16887
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9649237394332886,
      "learning_rate": 7.840933593632243e-06,
      "loss": 0.5198,
      "step": 16888
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2132312059402466,
      "learning_rate": 7.839843368578217e-06,
      "loss": 0.3912,
      "step": 16889
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.592976450920105,
      "learning_rate": 7.83875317045567e-06,
      "loss": 0.5234,
      "step": 16890
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5383625030517578,
      "learning_rate": 7.837662999278203e-06,
      "loss": 0.5773,
      "step": 16891
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5092641115188599,
      "learning_rate": 7.836572855059398e-06,
      "loss": 0.4939,
      "step": 16892
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4571081399917603,
      "learning_rate": 7.835482737812854e-06,
      "loss": 0.4839,
      "step": 16893
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.359910249710083,
      "learning_rate": 7.83439264755216e-06,
      "loss": 0.4174,
      "step": 16894
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5303963422775269,
      "learning_rate": 7.833302584290905e-06,
      "loss": 0.6627,
      "step": 16895
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.0466110706329346,
      "learning_rate": 7.832212548042679e-06,
      "loss": 0.5597,
      "step": 16896
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4523895978927612,
      "learning_rate": 7.831122538821074e-06,
      "loss": 0.5872,
      "step": 16897
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6317030191421509,
      "learning_rate": 7.830032556639677e-06,
      "loss": 0.494,
      "step": 16898
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5086826086044312,
      "learning_rate": 7.828942601512081e-06,
      "loss": 0.4655,
      "step": 16899
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5853033065795898,
      "learning_rate": 7.82785267345187e-06,
      "loss": 0.5937,
      "step": 16900
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4295403957366943,
      "learning_rate": 7.826762772472635e-06,
      "loss": 0.5463,
      "step": 16901
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7006231546401978,
      "learning_rate": 7.825672898587967e-06,
      "loss": 0.5311,
      "step": 16902
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.438908338546753,
      "learning_rate": 7.824583051811448e-06,
      "loss": 0.5426,
      "step": 16903
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6789987087249756,
      "learning_rate": 7.823493232156672e-06,
      "loss": 0.527,
      "step": 16904
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5919556617736816,
      "learning_rate": 7.82240343963722e-06,
      "loss": 0.411,
      "step": 16905
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6566963195800781,
      "learning_rate": 7.821313674266681e-06,
      "loss": 0.5852,
      "step": 16906
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6585949659347534,
      "learning_rate": 7.820223936058644e-06,
      "loss": 0.5532,
      "step": 16907
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4241819381713867,
      "learning_rate": 7.819134225026692e-06,
      "loss": 0.5522,
      "step": 16908
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7819695472717285,
      "learning_rate": 7.818044541184413e-06,
      "loss": 0.4592,
      "step": 16909
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.513702392578125,
      "learning_rate": 7.816954884545393e-06,
      "loss": 0.4911,
      "step": 16910
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.3348913192749023,
      "learning_rate": 7.815865255123213e-06,
      "loss": 0.6496,
      "step": 16911
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6251890659332275,
      "learning_rate": 7.814775652931463e-06,
      "loss": 0.4736,
      "step": 16912
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4056191444396973,
      "learning_rate": 7.813686077983724e-06,
      "loss": 0.556,
      "step": 16913
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.432217001914978,
      "learning_rate": 7.81259653029358e-06,
      "loss": 0.521,
      "step": 16914
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6829051971435547,
      "learning_rate": 7.811507009874619e-06,
      "loss": 0.4282,
      "step": 16915
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5075308084487915,
      "learning_rate": 7.810417516740419e-06,
      "loss": 0.5252,
      "step": 16916
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8162648677825928,
      "learning_rate": 7.80932805090457e-06,
      "loss": 0.5174,
      "step": 16917
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.154163122177124,
      "learning_rate": 7.808238612380645e-06,
      "loss": 0.5326,
      "step": 16918
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6512550115585327,
      "learning_rate": 7.807149201182233e-06,
      "loss": 0.5337,
      "step": 16919
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7160550355911255,
      "learning_rate": 7.806059817322917e-06,
      "loss": 0.4536,
      "step": 16920
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5081151723861694,
      "learning_rate": 7.804970460816274e-06,
      "loss": 0.414,
      "step": 16921
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.3477884531021118,
      "learning_rate": 7.80388113167589e-06,
      "loss": 0.6527,
      "step": 16922
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.606650710105896,
      "learning_rate": 7.802791829915346e-06,
      "loss": 0.5726,
      "step": 16923
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7154167890548706,
      "learning_rate": 7.801702555548218e-06,
      "loss": 0.5451,
      "step": 16924
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5700230598449707,
      "learning_rate": 7.800613308588092e-06,
      "loss": 0.5744,
      "step": 16925
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7032791376113892,
      "learning_rate": 7.799524089048543e-06,
      "loss": 0.5542,
      "step": 16926
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7384109497070312,
      "learning_rate": 7.798434896943152e-06,
      "loss": 0.6812,
      "step": 16927
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.0017917156219482,
      "learning_rate": 7.797345732285503e-06,
      "loss": 0.5671,
      "step": 16928
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4668453931808472,
      "learning_rate": 7.796256595089168e-06,
      "loss": 0.4786,
      "step": 16929
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4520586729049683,
      "learning_rate": 7.795167485367733e-06,
      "loss": 0.5238,
      "step": 16930
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6688194274902344,
      "learning_rate": 7.79407840313477e-06,
      "loss": 0.487,
      "step": 16931
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5353419780731201,
      "learning_rate": 7.792989348403858e-06,
      "loss": 0.5552,
      "step": 16932
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6005743741989136,
      "learning_rate": 7.79190032118858e-06,
      "loss": 0.561,
      "step": 16933
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6520601511001587,
      "learning_rate": 7.790811321502507e-06,
      "loss": 0.563,
      "step": 16934
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4299805164337158,
      "learning_rate": 7.78972234935922e-06,
      "loss": 0.5077,
      "step": 16935
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.742250919342041,
      "learning_rate": 7.788633404772294e-06,
      "loss": 0.6291,
      "step": 16936
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7822974920272827,
      "learning_rate": 7.787544487755303e-06,
      "loss": 0.5985,
      "step": 16937
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4786291122436523,
      "learning_rate": 7.786455598321828e-06,
      "loss": 0.5895,
      "step": 16938
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6916205883026123,
      "learning_rate": 7.785366736485441e-06,
      "loss": 0.4443,
      "step": 16939
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4601807594299316,
      "learning_rate": 7.784277902259715e-06,
      "loss": 0.6022,
      "step": 16940
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8535845279693604,
      "learning_rate": 7.783189095658233e-06,
      "loss": 0.6197,
      "step": 16941
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6456066370010376,
      "learning_rate": 7.782100316694561e-06,
      "loss": 0.5248,
      "step": 16942
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6136088371276855,
      "learning_rate": 7.78101156538228e-06,
      "loss": 0.5491,
      "step": 16943
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8880442380905151,
      "learning_rate": 7.779922841734958e-06,
      "loss": 0.6165,
      "step": 16944
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8017330169677734,
      "learning_rate": 7.778834145766171e-06,
      "loss": 0.6057,
      "step": 16945
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4428210258483887,
      "learning_rate": 7.777745477489496e-06,
      "loss": 0.5523,
      "step": 16946
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4593971967697144,
      "learning_rate": 7.776656836918497e-06,
      "loss": 0.5801,
      "step": 16947
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.7157572507858276,
      "learning_rate": 7.775568224066755e-06,
      "loss": 0.5739,
      "step": 16948
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5331593751907349,
      "learning_rate": 7.774479638947839e-06,
      "loss": 0.5512,
      "step": 16949
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.735896348953247,
      "learning_rate": 7.77339108157532e-06,
      "loss": 0.4371,
      "step": 16950
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4021388292312622,
      "learning_rate": 7.772302551962771e-06,
      "loss": 0.5629,
      "step": 16951
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.424554705619812,
      "learning_rate": 7.77121405012376e-06,
      "loss": 0.6196,
      "step": 16952
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.624415397644043,
      "learning_rate": 7.770125576071861e-06,
      "loss": 0.5767,
      "step": 16953
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8814479112625122,
      "learning_rate": 7.769037129820644e-06,
      "loss": 0.6248,
      "step": 16954
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6710283756256104,
      "learning_rate": 7.767948711383679e-06,
      "loss": 0.5809,
      "step": 16955
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.305412769317627,
      "learning_rate": 7.766860320774536e-06,
      "loss": 0.4232,
      "step": 16956
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.507940411567688,
      "learning_rate": 7.765771958006781e-06,
      "loss": 0.5799,
      "step": 16957
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.377177119255066,
      "learning_rate": 7.764683623093984e-06,
      "loss": 0.5176,
      "step": 16958
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6931606531143188,
      "learning_rate": 7.76359531604972e-06,
      "loss": 0.4783,
      "step": 16959
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.0600388050079346,
      "learning_rate": 7.76250703688755e-06,
      "loss": 0.4368,
      "step": 16960
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.8784445524215698,
      "learning_rate": 7.761418785621043e-06,
      "loss": 0.3694,
      "step": 16961
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5378241539001465,
      "learning_rate": 7.760330562263772e-06,
      "loss": 0.4927,
      "step": 16962
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.0450494289398193,
      "learning_rate": 7.759242366829299e-06,
      "loss": 0.541,
      "step": 16963
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2745006084442139,
      "learning_rate": 7.758154199331193e-06,
      "loss": 0.5011,
      "step": 16964
    },
    {
      "epoch": 1.75,
      "grad_norm": 2.0903818607330322,
      "learning_rate": 7.757066059783019e-06,
      "loss": 0.5273,
      "step": 16965
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.424524188041687,
      "learning_rate": 7.755977948198343e-06,
      "loss": 0.5223,
      "step": 16966
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.3737682104110718,
      "learning_rate": 7.754889864590735e-06,
      "loss": 0.5291,
      "step": 16967
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.679172158241272,
      "learning_rate": 7.753801808973756e-06,
      "loss": 0.6235,
      "step": 16968
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.6984012126922607,
      "learning_rate": 7.752713781360973e-06,
      "loss": 0.4735,
      "step": 16969
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3774017095565796,
      "learning_rate": 7.75162578176595e-06,
      "loss": 0.5611,
      "step": 16970
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4960507154464722,
      "learning_rate": 7.750537810202253e-06,
      "loss": 0.5046,
      "step": 16971
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.325339674949646,
      "learning_rate": 7.749449866683447e-06,
      "loss": 0.5373,
      "step": 16972
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5806784629821777,
      "learning_rate": 7.748361951223092e-06,
      "loss": 0.4464,
      "step": 16973
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.332785964012146,
      "learning_rate": 7.747274063834754e-06,
      "loss": 0.533,
      "step": 16974
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4211630821228027,
      "learning_rate": 7.746186204531998e-06,
      "loss": 0.4785,
      "step": 16975
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6772360801696777,
      "learning_rate": 7.745098373328382e-06,
      "loss": 0.5001,
      "step": 16976
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9339239597320557,
      "learning_rate": 7.744010570237474e-06,
      "loss": 0.5584,
      "step": 16977
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7738052606582642,
      "learning_rate": 7.742922795272829e-06,
      "loss": 0.4443,
      "step": 16978
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6641358137130737,
      "learning_rate": 7.741835048448015e-06,
      "loss": 0.5858,
      "step": 16979
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.474837303161621,
      "learning_rate": 7.740747329776591e-06,
      "loss": 0.5286,
      "step": 16980
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.308000087738037,
      "learning_rate": 7.739659639272117e-06,
      "loss": 0.5905,
      "step": 16981
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5319998264312744,
      "learning_rate": 7.738571976948157e-06,
      "loss": 0.4514,
      "step": 16982
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5506337881088257,
      "learning_rate": 7.737484342818265e-06,
      "loss": 0.4544,
      "step": 16983
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.606358289718628,
      "learning_rate": 7.736396736896008e-06,
      "loss": 0.4157,
      "step": 16984
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.512359380722046,
      "learning_rate": 7.735309159194941e-06,
      "loss": 0.4446,
      "step": 16985
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4405676126480103,
      "learning_rate": 7.734221609728626e-06,
      "loss": 0.5097,
      "step": 16986
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.520884394645691,
      "learning_rate": 7.733134088510618e-06,
      "loss": 0.5075,
      "step": 16987
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5939658880233765,
      "learning_rate": 7.732046595554482e-06,
      "loss": 0.489,
      "step": 16988
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8933578729629517,
      "learning_rate": 7.73095913087377e-06,
      "loss": 0.5243,
      "step": 16989
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7993440628051758,
      "learning_rate": 7.729871694482043e-06,
      "loss": 0.4905,
      "step": 16990
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2119373083114624,
      "learning_rate": 7.728784286392857e-06,
      "loss": 0.3854,
      "step": 16991
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4621226787567139,
      "learning_rate": 7.72769690661977e-06,
      "loss": 0.5814,
      "step": 16992
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4627560377120972,
      "learning_rate": 7.726609555176339e-06,
      "loss": 0.6295,
      "step": 16993
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.9195542335510254,
      "learning_rate": 7.725522232076119e-06,
      "loss": 0.4551,
      "step": 16994
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6569445133209229,
      "learning_rate": 7.724434937332668e-06,
      "loss": 0.3566,
      "step": 16995
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3240078687667847,
      "learning_rate": 7.72334767095954e-06,
      "loss": 0.4494,
      "step": 16996
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5617667436599731,
      "learning_rate": 7.72226043297029e-06,
      "loss": 0.5014,
      "step": 16997
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7885522842407227,
      "learning_rate": 7.721173223378478e-06,
      "loss": 0.565,
      "step": 16998
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5300153493881226,
      "learning_rate": 7.720086042197651e-06,
      "loss": 0.4999,
      "step": 16999
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.699491262435913,
      "learning_rate": 7.718998889441367e-06,
      "loss": 0.5514,
      "step": 17000
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4913944005966187,
      "learning_rate": 7.717911765123182e-06,
      "loss": 0.4928,
      "step": 17001
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2731879949569702,
      "learning_rate": 7.716824669256647e-06,
      "loss": 0.4613,
      "step": 17002
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5379183292388916,
      "learning_rate": 7.715737601855318e-06,
      "loss": 0.4726,
      "step": 17003
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.637677788734436,
      "learning_rate": 7.714650562932743e-06,
      "loss": 0.6398,
      "step": 17004
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4047942161560059,
      "learning_rate": 7.713563552502475e-06,
      "loss": 0.5449,
      "step": 17005
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.244468331336975,
      "learning_rate": 7.712476570578072e-06,
      "loss": 0.4623,
      "step": 17006
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6780805587768555,
      "learning_rate": 7.711389617173081e-06,
      "loss": 0.6221,
      "step": 17007
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7509328126907349,
      "learning_rate": 7.710302692301056e-06,
      "loss": 0.6371,
      "step": 17008
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7014639377593994,
      "learning_rate": 7.709215795975544e-06,
      "loss": 0.6391,
      "step": 17009
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.0014750957489014,
      "learning_rate": 7.7081289282101e-06,
      "loss": 0.5104,
      "step": 17010
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8679380416870117,
      "learning_rate": 7.707042089018274e-06,
      "loss": 0.452,
      "step": 17011
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.462425708770752,
      "learning_rate": 7.705955278413613e-06,
      "loss": 0.5578,
      "step": 17012
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5356290340423584,
      "learning_rate": 7.704868496409668e-06,
      "loss": 0.6174,
      "step": 17013
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5930510759353638,
      "learning_rate": 7.703781743019992e-06,
      "loss": 0.476,
      "step": 17014
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.823953628540039,
      "learning_rate": 7.702695018258129e-06,
      "loss": 0.4624,
      "step": 17015
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.733577847480774,
      "learning_rate": 7.70160832213763e-06,
      "loss": 0.4948,
      "step": 17016
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8350516557693481,
      "learning_rate": 7.700521654672042e-06,
      "loss": 0.5739,
      "step": 17017
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.062316656112671,
      "learning_rate": 7.699435015874914e-06,
      "loss": 0.6059,
      "step": 17018
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4261455535888672,
      "learning_rate": 7.698348405759793e-06,
      "loss": 0.5011,
      "step": 17019
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5274006128311157,
      "learning_rate": 7.697261824340225e-06,
      "loss": 0.4864,
      "step": 17020
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6415899991989136,
      "learning_rate": 7.696175271629762e-06,
      "loss": 0.5067,
      "step": 17021
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5641318559646606,
      "learning_rate": 7.695088747641943e-06,
      "loss": 0.5139,
      "step": 17022
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3150224685668945,
      "learning_rate": 7.694002252390318e-06,
      "loss": 0.5098,
      "step": 17023
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4901751279830933,
      "learning_rate": 7.692915785888436e-06,
      "loss": 0.6628,
      "step": 17024
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5846935510635376,
      "learning_rate": 7.691829348149834e-06,
      "loss": 0.5433,
      "step": 17025
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8611574172973633,
      "learning_rate": 7.690742939188062e-06,
      "loss": 0.5513,
      "step": 17026
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6636021137237549,
      "learning_rate": 7.689656559016668e-06,
      "loss": 0.5204,
      "step": 17027
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5253825187683105,
      "learning_rate": 7.688570207649191e-06,
      "loss": 0.5081,
      "step": 17028
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7543295621871948,
      "learning_rate": 7.687483885099178e-06,
      "loss": 0.5283,
      "step": 17029
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7260291576385498,
      "learning_rate": 7.68639759138017e-06,
      "loss": 0.5117,
      "step": 17030
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.326360821723938,
      "learning_rate": 7.685311326505711e-06,
      "loss": 0.4139,
      "step": 17031
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.860840916633606,
      "learning_rate": 7.684225090489348e-06,
      "loss": 0.5149,
      "step": 17032
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7225736379623413,
      "learning_rate": 7.683138883344616e-06,
      "loss": 0.4598,
      "step": 17033
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5427411794662476,
      "learning_rate": 7.682052705085063e-06,
      "loss": 0.517,
      "step": 17034
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7254737615585327,
      "learning_rate": 7.680966555724229e-06,
      "loss": 0.5391,
      "step": 17035
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6324094533920288,
      "learning_rate": 7.679880435275654e-06,
      "loss": 0.5122,
      "step": 17036
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3995821475982666,
      "learning_rate": 7.678794343752882e-06,
      "loss": 0.5026,
      "step": 17037
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5502022504806519,
      "learning_rate": 7.67770828116945e-06,
      "loss": 0.5203,
      "step": 17038
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5536541938781738,
      "learning_rate": 7.676622247538903e-06,
      "loss": 0.6345,
      "step": 17039
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.640518307685852,
      "learning_rate": 7.67553624287478e-06,
      "loss": 0.4873,
      "step": 17040
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3074054718017578,
      "learning_rate": 7.674450267190615e-06,
      "loss": 0.4178,
      "step": 17041
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4024901390075684,
      "learning_rate": 7.673364320499954e-06,
      "loss": 0.5406,
      "step": 17042
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5230858325958252,
      "learning_rate": 7.672278402816331e-06,
      "loss": 0.4311,
      "step": 17043
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6498769521713257,
      "learning_rate": 7.671192514153287e-06,
      "loss": 0.5016,
      "step": 17044
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.445146918296814,
      "learning_rate": 7.670106654524364e-06,
      "loss": 0.5916,
      "step": 17045
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.654604196548462,
      "learning_rate": 7.669020823943092e-06,
      "loss": 0.4948,
      "step": 17046
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3167698383331299,
      "learning_rate": 7.667935022423014e-06,
      "loss": 0.4444,
      "step": 17047
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2766554355621338,
      "learning_rate": 7.666849249977664e-06,
      "loss": 0.4713,
      "step": 17048
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.379989743232727,
      "learning_rate": 7.665763506620581e-06,
      "loss": 0.6485,
      "step": 17049
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4588356018066406,
      "learning_rate": 7.6646777923653e-06,
      "loss": 0.5324,
      "step": 17050
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5081055164337158,
      "learning_rate": 7.663592107225357e-06,
      "loss": 0.5517,
      "step": 17051
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.8864752054214478,
      "learning_rate": 7.662506451214288e-06,
      "loss": 0.5836,
      "step": 17052
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.111848831176758,
      "learning_rate": 7.661420824345631e-06,
      "loss": 0.514,
      "step": 17053
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7917393445968628,
      "learning_rate": 7.660335226632916e-06,
      "loss": 0.6376,
      "step": 17054
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.438482642173767,
      "learning_rate": 7.659249658089682e-06,
      "loss": 0.4996,
      "step": 17055
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2969279289245605,
      "learning_rate": 7.658164118729459e-06,
      "loss": 0.4498,
      "step": 17056
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5537861585617065,
      "learning_rate": 7.657078608565783e-06,
      "loss": 0.6073,
      "step": 17057
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.601586937904358,
      "learning_rate": 7.65599312761219e-06,
      "loss": 0.6113,
      "step": 17058
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5012620687484741,
      "learning_rate": 7.654907675882206e-06,
      "loss": 0.5151,
      "step": 17059
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4485576152801514,
      "learning_rate": 7.653822253389372e-06,
      "loss": 0.6276,
      "step": 17060
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5335228443145752,
      "learning_rate": 7.652736860147215e-06,
      "loss": 0.5128,
      "step": 17061
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3067644834518433,
      "learning_rate": 7.651651496169267e-06,
      "loss": 0.4375,
      "step": 17062
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5133576393127441,
      "learning_rate": 7.650566161469065e-06,
      "loss": 0.6054,
      "step": 17063
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5595862865447998,
      "learning_rate": 7.649480856060134e-06,
      "loss": 0.6103,
      "step": 17064
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3532260656356812,
      "learning_rate": 7.648395579956008e-06,
      "loss": 0.4785,
      "step": 17065
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5887033939361572,
      "learning_rate": 7.647310333170218e-06,
      "loss": 0.6006,
      "step": 17066
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.465482234954834,
      "learning_rate": 7.64622511571629e-06,
      "loss": 0.5352,
      "step": 17067
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.405733346939087,
      "learning_rate": 7.64513992760776e-06,
      "loss": 0.5258,
      "step": 17068
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3337435722351074,
      "learning_rate": 7.644054768858152e-06,
      "loss": 0.4012,
      "step": 17069
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6718064546585083,
      "learning_rate": 7.642969639480998e-06,
      "loss": 0.561,
      "step": 17070
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.9140489101409912,
      "learning_rate": 7.64188453948983e-06,
      "loss": 0.51,
      "step": 17071
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7177703380584717,
      "learning_rate": 7.640799468898168e-06,
      "loss": 0.6121,
      "step": 17072
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3306682109832764,
      "learning_rate": 7.639714427719547e-06,
      "loss": 0.5502,
      "step": 17073
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6089993715286255,
      "learning_rate": 7.63862941596749e-06,
      "loss": 0.4743,
      "step": 17074
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7281522750854492,
      "learning_rate": 7.637544433655527e-06,
      "loss": 0.3887,
      "step": 17075
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3569703102111816,
      "learning_rate": 7.636459480797185e-06,
      "loss": 0.5786,
      "step": 17076
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4092615842819214,
      "learning_rate": 7.63537455740599e-06,
      "loss": 0.481,
      "step": 17077
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.406386137008667,
      "learning_rate": 7.634289663495467e-06,
      "loss": 0.4802,
      "step": 17078
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6318086385726929,
      "learning_rate": 7.633204799079144e-06,
      "loss": 0.505,
      "step": 17079
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3317508697509766,
      "learning_rate": 7.632119964170543e-06,
      "loss": 0.4356,
      "step": 17080
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6176339387893677,
      "learning_rate": 7.631035158783193e-06,
      "loss": 0.5711,
      "step": 17081
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.3878633975982666,
      "learning_rate": 7.629950382930617e-06,
      "loss": 0.6118,
      "step": 17082
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.556673526763916,
      "learning_rate": 7.628865636626337e-06,
      "loss": 0.4356,
      "step": 17083
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7767703533172607,
      "learning_rate": 7.627780919883881e-06,
      "loss": 0.5488,
      "step": 17084
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.945898175239563,
      "learning_rate": 7.62669623271677e-06,
      "loss": 0.5436,
      "step": 17085
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.061861276626587,
      "learning_rate": 7.625611575138529e-06,
      "loss": 0.4983,
      "step": 17086
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.9433125257492065,
      "learning_rate": 7.624526947162678e-06,
      "loss": 0.643,
      "step": 17087
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6409662961959839,
      "learning_rate": 7.623442348802741e-06,
      "loss": 0.5257,
      "step": 17088
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4673703908920288,
      "learning_rate": 7.622357780072242e-06,
      "loss": 0.5357,
      "step": 17089
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6944411993026733,
      "learning_rate": 7.621273240984699e-06,
      "loss": 0.4535,
      "step": 17090
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8811447620391846,
      "learning_rate": 7.620188731553636e-06,
      "loss": 0.5301,
      "step": 17091
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.654378890991211,
      "learning_rate": 7.619104251792575e-06,
      "loss": 0.4588,
      "step": 17092
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.098454236984253,
      "learning_rate": 7.618019801715033e-06,
      "loss": 0.5474,
      "step": 17093
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7094979286193848,
      "learning_rate": 7.616935381334533e-06,
      "loss": 0.6008,
      "step": 17094
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5342406034469604,
      "learning_rate": 7.615850990664594e-06,
      "loss": 0.4917,
      "step": 17095
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.761589527130127,
      "learning_rate": 7.614766629718735e-06,
      "loss": 0.5556,
      "step": 17096
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6347736120224,
      "learning_rate": 7.6136822985104775e-06,
      "loss": 0.4751,
      "step": 17097
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3995022773742676,
      "learning_rate": 7.612597997053336e-06,
      "loss": 0.4743,
      "step": 17098
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6651332378387451,
      "learning_rate": 7.611513725360833e-06,
      "loss": 0.5867,
      "step": 17099
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.355580449104309,
      "learning_rate": 7.610429483446484e-06,
      "loss": 0.5686,
      "step": 17100
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3599510192871094,
      "learning_rate": 7.609345271323807e-06,
      "loss": 0.5495,
      "step": 17101
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.362457036972046,
      "learning_rate": 7.608261089006322e-06,
      "loss": 0.3929,
      "step": 17102
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6085189580917358,
      "learning_rate": 7.607176936507541e-06,
      "loss": 0.5325,
      "step": 17103
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6799911260604858,
      "learning_rate": 7.606092813840984e-06,
      "loss": 0.4583,
      "step": 17104
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1229360103607178,
      "learning_rate": 7.6050087210201685e-06,
      "loss": 0.4474,
      "step": 17105
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3602956533432007,
      "learning_rate": 7.603924658058606e-06,
      "loss": 0.5185,
      "step": 17106
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.214357852935791,
      "learning_rate": 7.602840624969816e-06,
      "loss": 0.5684,
      "step": 17107
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.42812979221344,
      "learning_rate": 7.601756621767309e-06,
      "loss": 0.4607,
      "step": 17108
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5873943567276,
      "learning_rate": 7.600672648464603e-06,
      "loss": 0.4968,
      "step": 17109
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.519628643989563,
      "learning_rate": 7.599588705075214e-06,
      "loss": 0.4937,
      "step": 17110
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.220988154411316,
      "learning_rate": 7.598504791612651e-06,
      "loss": 0.487,
      "step": 17111
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.054396390914917,
      "learning_rate": 7.597420908090433e-06,
      "loss": 0.556,
      "step": 17112
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.9566808938980103,
      "learning_rate": 7.5963370545220674e-06,
      "loss": 0.7318,
      "step": 17113
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8217676877975464,
      "learning_rate": 7.59525323092107e-06,
      "loss": 0.4336,
      "step": 17114
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5238248109817505,
      "learning_rate": 7.594169437300956e-06,
      "loss": 0.478,
      "step": 17115
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5378679037094116,
      "learning_rate": 7.593085673675231e-06,
      "loss": 0.4876,
      "step": 17116
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4312869310379028,
      "learning_rate": 7.592001940057412e-06,
      "loss": 0.5932,
      "step": 17117
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.8815134763717651,
      "learning_rate": 7.5909182364610105e-06,
      "loss": 0.4633,
      "step": 17118
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.576738953590393,
      "learning_rate": 7.589834562899532e-06,
      "loss": 0.625,
      "step": 17119
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6027218103408813,
      "learning_rate": 7.588750919386494e-06,
      "loss": 0.4428,
      "step": 17120
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7074189186096191,
      "learning_rate": 7.587667305935401e-06,
      "loss": 0.3888,
      "step": 17121
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.9129165410995483,
      "learning_rate": 7.586583722559765e-06,
      "loss": 0.5278,
      "step": 17122
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3140560388565063,
      "learning_rate": 7.5855001692730975e-06,
      "loss": 0.5264,
      "step": 17123
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.966792345046997,
      "learning_rate": 7.584416646088903e-06,
      "loss": 0.6441,
      "step": 17124
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5367169380187988,
      "learning_rate": 7.583333153020697e-06,
      "loss": 0.4865,
      "step": 17125
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5561555624008179,
      "learning_rate": 7.58224969008198e-06,
      "loss": 0.5879,
      "step": 17126
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5376769304275513,
      "learning_rate": 7.5811662572862635e-06,
      "loss": 0.5121,
      "step": 17127
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7253777980804443,
      "learning_rate": 7.580082854647057e-06,
      "loss": 0.4705,
      "step": 17128
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4834895133972168,
      "learning_rate": 7.5789994821778644e-06,
      "loss": 0.5157,
      "step": 17129
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7078403234481812,
      "learning_rate": 7.577916139892194e-06,
      "loss": 0.4626,
      "step": 17130
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4298343658447266,
      "learning_rate": 7.5768328278035545e-06,
      "loss": 0.5148,
      "step": 17131
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6672929525375366,
      "learning_rate": 7.575749545925447e-06,
      "loss": 0.4914,
      "step": 17132
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.545422077178955,
      "learning_rate": 7.574666294271382e-06,
      "loss": 0.6022,
      "step": 17133
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5425843000411987,
      "learning_rate": 7.573583072854861e-06,
      "loss": 0.5908,
      "step": 17134
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4757359027862549,
      "learning_rate": 7.572499881689391e-06,
      "loss": 0.577,
      "step": 17135
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.979335069656372,
      "learning_rate": 7.571416720788476e-06,
      "loss": 0.5203,
      "step": 17136
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5843621492385864,
      "learning_rate": 7.57033359016562e-06,
      "loss": 0.414,
      "step": 17137
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.865090012550354,
      "learning_rate": 7.569250489834327e-06,
      "loss": 0.4844,
      "step": 17138
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6314722299575806,
      "learning_rate": 7.568167419808101e-06,
      "loss": 0.5533,
      "step": 17139
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6752502918243408,
      "learning_rate": 7.567084380100444e-06,
      "loss": 0.426,
      "step": 17140
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6660120487213135,
      "learning_rate": 7.56600137072486e-06,
      "loss": 0.537,
      "step": 17141
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7222161293029785,
      "learning_rate": 7.56491839169485e-06,
      "loss": 0.5215,
      "step": 17142
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4873013496398926,
      "learning_rate": 7.563835443023916e-06,
      "loss": 0.4278,
      "step": 17143
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6239280700683594,
      "learning_rate": 7.562752524725562e-06,
      "loss": 0.5325,
      "step": 17144
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5315783023834229,
      "learning_rate": 7.561669636813285e-06,
      "loss": 0.5563,
      "step": 17145
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2844654321670532,
      "learning_rate": 7.56058677930059e-06,
      "loss": 0.4904,
      "step": 17146
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5472395420074463,
      "learning_rate": 7.559503952200971e-06,
      "loss": 0.5391,
      "step": 17147
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.476322054862976,
      "learning_rate": 7.558421155527934e-06,
      "loss": 0.4943,
      "step": 17148
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.357174277305603,
      "learning_rate": 7.557338389294979e-06,
      "loss": 0.4701,
      "step": 17149
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.958686351776123,
      "learning_rate": 7.556255653515601e-06,
      "loss": 0.4402,
      "step": 17150
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.6913145780563354,
      "learning_rate": 7.555172948203303e-06,
      "loss": 0.6617,
      "step": 17151
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.517109990119934,
      "learning_rate": 7.554090273371579e-06,
      "loss": 0.555,
      "step": 17152
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4272254705429077,
      "learning_rate": 7.55300762903393e-06,
      "loss": 0.5125,
      "step": 17153
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3100453615188599,
      "learning_rate": 7.551925015203856e-06,
      "loss": 0.5262,
      "step": 17154
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5309749841690063,
      "learning_rate": 7.550842431894849e-06,
      "loss": 0.6108,
      "step": 17155
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5458474159240723,
      "learning_rate": 7.549759879120408e-06,
      "loss": 0.5816,
      "step": 17156
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3505339622497559,
      "learning_rate": 7.548677356894034e-06,
      "loss": 0.4168,
      "step": 17157
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.65963876247406,
      "learning_rate": 7.547594865229216e-06,
      "loss": 0.3825,
      "step": 17158
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3163200616836548,
      "learning_rate": 7.5465124041394565e-06,
      "loss": 0.5361,
      "step": 17159
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5749340057373047,
      "learning_rate": 7.545429973638244e-06,
      "loss": 0.3866,
      "step": 17160
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.555665135383606,
      "learning_rate": 7.544347573739079e-06,
      "loss": 0.558,
      "step": 17161
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.7971134185791016,
      "learning_rate": 7.543265204455455e-06,
      "loss": 0.5832,
      "step": 17162
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7131842374801636,
      "learning_rate": 7.542182865800866e-06,
      "loss": 0.4741,
      "step": 17163
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6803644895553589,
      "learning_rate": 7.541100557788805e-06,
      "loss": 0.496,
      "step": 17164
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.384442925453186,
      "learning_rate": 7.540018280432766e-06,
      "loss": 0.5311,
      "step": 17165
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6952954530715942,
      "learning_rate": 7.5389360337462425e-06,
      "loss": 0.5339,
      "step": 17166
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7882198095321655,
      "learning_rate": 7.537853817742728e-06,
      "loss": 0.5141,
      "step": 17167
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.442612886428833,
      "learning_rate": 7.536771632435713e-06,
      "loss": 0.5533,
      "step": 17168
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5124419927597046,
      "learning_rate": 7.53568947783869e-06,
      "loss": 0.4705,
      "step": 17169
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5880281925201416,
      "learning_rate": 7.534607353965153e-06,
      "loss": 0.5186,
      "step": 17170
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6006600856781006,
      "learning_rate": 7.533525260828591e-06,
      "loss": 0.5418,
      "step": 17171
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3679232597351074,
      "learning_rate": 7.532443198442496e-06,
      "loss": 0.5281,
      "step": 17172
    },
    {
      "epoch": 1.78,
      "grad_norm": 2.124131441116333,
      "learning_rate": 7.531361166820357e-06,
      "loss": 0.5866,
      "step": 17173
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3194689750671387,
      "learning_rate": 7.530279165975664e-06,
      "loss": 0.5329,
      "step": 17174
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.510501503944397,
      "learning_rate": 7.529197195921909e-06,
      "loss": 0.5218,
      "step": 17175
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5428675413131714,
      "learning_rate": 7.528115256672579e-06,
      "loss": 0.551,
      "step": 17176
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6567749977111816,
      "learning_rate": 7.527033348241166e-06,
      "loss": 0.5817,
      "step": 17177
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5012249946594238,
      "learning_rate": 7.525951470641153e-06,
      "loss": 0.4492,
      "step": 17178
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.525665521621704,
      "learning_rate": 7.524869623886033e-06,
      "loss": 0.6051,
      "step": 17179
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.496302604675293,
      "learning_rate": 7.523787807989293e-06,
      "loss": 0.5285,
      "step": 17180
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.480017900466919,
      "learning_rate": 7.52270602296442e-06,
      "loss": 0.5236,
      "step": 17181
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4992084503173828,
      "learning_rate": 7.5216242688249e-06,
      "loss": 0.4914,
      "step": 17182
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4701541662216187,
      "learning_rate": 7.520542545584222e-06,
      "loss": 0.5475,
      "step": 17183
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7396049499511719,
      "learning_rate": 7.519460853255869e-06,
      "loss": 0.5642,
      "step": 17184
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5119519233703613,
      "learning_rate": 7.51837919185333e-06,
      "loss": 0.4957,
      "step": 17185
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5002140998840332,
      "learning_rate": 7.5172975613900866e-06,
      "loss": 0.4935,
      "step": 17186
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6938992738723755,
      "learning_rate": 7.516215961879627e-06,
      "loss": 0.5685,
      "step": 17187
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6503199338912964,
      "learning_rate": 7.515134393335438e-06,
      "loss": 0.4943,
      "step": 17188
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5459051132202148,
      "learning_rate": 7.514052855770999e-06,
      "loss": 0.4696,
      "step": 17189
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5477898120880127,
      "learning_rate": 7.512971349199798e-06,
      "loss": 0.5362,
      "step": 17190
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5789998769760132,
      "learning_rate": 7.511889873635314e-06,
      "loss": 0.527,
      "step": 17191
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5368881225585938,
      "learning_rate": 7.510808429091033e-06,
      "loss": 0.5824,
      "step": 17192
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5535879135131836,
      "learning_rate": 7.50972701558044e-06,
      "loss": 0.532,
      "step": 17193
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.380549430847168,
      "learning_rate": 7.508645633117014e-06,
      "loss": 0.4912,
      "step": 17194
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5526630878448486,
      "learning_rate": 7.507564281714238e-06,
      "loss": 0.3799,
      "step": 17195
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4204394817352295,
      "learning_rate": 7.506482961385593e-06,
      "loss": 0.5168,
      "step": 17196
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4762418270111084,
      "learning_rate": 7.505401672144559e-06,
      "loss": 0.5332,
      "step": 17197
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6252743005752563,
      "learning_rate": 7.504320414004622e-06,
      "loss": 0.5157,
      "step": 17198
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.733656883239746,
      "learning_rate": 7.503239186979257e-06,
      "loss": 0.5424,
      "step": 17199
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5988458395004272,
      "learning_rate": 7.502157991081947e-06,
      "loss": 0.612,
      "step": 17200
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5822185277938843,
      "learning_rate": 7.501076826326172e-06,
      "loss": 0.5579,
      "step": 17201
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4758973121643066,
      "learning_rate": 7.499995692725408e-06,
      "loss": 0.4927,
      "step": 17202
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7573902606964111,
      "learning_rate": 7.498914590293138e-06,
      "loss": 0.5236,
      "step": 17203
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.663033366203308,
      "learning_rate": 7.497833519042837e-06,
      "loss": 0.4752,
      "step": 17204
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8449777364730835,
      "learning_rate": 7.496752478987984e-06,
      "loss": 0.5101,
      "step": 17205
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.84988272190094,
      "learning_rate": 7.495671470142061e-06,
      "loss": 0.4907,
      "step": 17206
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.380393385887146,
      "learning_rate": 7.4945904925185375e-06,
      "loss": 0.5152,
      "step": 17207
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8707374334335327,
      "learning_rate": 7.493509546130898e-06,
      "loss": 0.6753,
      "step": 17208
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6086596250534058,
      "learning_rate": 7.4924286309926145e-06,
      "loss": 0.4105,
      "step": 17209
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9381335973739624,
      "learning_rate": 7.491347747117163e-06,
      "loss": 0.5755,
      "step": 17210
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9167506694793701,
      "learning_rate": 7.490266894518024e-06,
      "loss": 0.5791,
      "step": 17211
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6114122867584229,
      "learning_rate": 7.489186073208665e-06,
      "loss": 0.5734,
      "step": 17212
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4642853736877441,
      "learning_rate": 7.488105283202568e-06,
      "loss": 0.5775,
      "step": 17213
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4843363761901855,
      "learning_rate": 7.487024524513206e-06,
      "loss": 0.4813,
      "step": 17214
    },
    {
      "epoch": 1.78,
      "grad_norm": 2.0534026622772217,
      "learning_rate": 7.485943797154052e-06,
      "loss": 0.568,
      "step": 17215
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7989903688430786,
      "learning_rate": 7.484863101138582e-06,
      "loss": 0.4368,
      "step": 17216
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7025731801986694,
      "learning_rate": 7.483782436480265e-06,
      "loss": 0.4192,
      "step": 17217
    },
    {
      "epoch": 1.78,
      "grad_norm": 2.094028949737549,
      "learning_rate": 7.4827018031925766e-06,
      "loss": 0.4768,
      "step": 17218
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.544787883758545,
      "learning_rate": 7.481621201288993e-06,
      "loss": 0.5237,
      "step": 17219
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6611459255218506,
      "learning_rate": 7.4805406307829795e-06,
      "loss": 0.4474,
      "step": 17220
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5831209421157837,
      "learning_rate": 7.479460091688015e-06,
      "loss": 0.5309,
      "step": 17221
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7305011749267578,
      "learning_rate": 7.478379584017564e-06,
      "loss": 0.5885,
      "step": 17222
    },
    {
      "epoch": 1.78,
      "grad_norm": 2.3566977977752686,
      "learning_rate": 7.477299107785103e-06,
      "loss": 0.6202,
      "step": 17223
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.634485125541687,
      "learning_rate": 7.476218663004102e-06,
      "loss": 0.5741,
      "step": 17224
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7458760738372803,
      "learning_rate": 7.475138249688028e-06,
      "loss": 0.6391,
      "step": 17225
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5076396465301514,
      "learning_rate": 7.4740578678503574e-06,
      "loss": 0.5421,
      "step": 17226
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6838351488113403,
      "learning_rate": 7.472977517504551e-06,
      "loss": 0.4349,
      "step": 17227
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5656379461288452,
      "learning_rate": 7.471897198664085e-06,
      "loss": 0.6722,
      "step": 17228
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6090022325515747,
      "learning_rate": 7.470816911342422e-06,
      "loss": 0.6639,
      "step": 17229
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.370062232017517,
      "learning_rate": 7.469736655553038e-06,
      "loss": 0.5683,
      "step": 17230
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5391095876693726,
      "learning_rate": 7.468656431309393e-06,
      "loss": 0.5203,
      "step": 17231
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4369133710861206,
      "learning_rate": 7.46757623862496e-06,
      "loss": 0.5127,
      "step": 17232
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6511099338531494,
      "learning_rate": 7.466496077513206e-06,
      "loss": 0.5069,
      "step": 17233
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6019022464752197,
      "learning_rate": 7.465415947987593e-06,
      "loss": 0.4861,
      "step": 17234
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6238563060760498,
      "learning_rate": 7.464335850061593e-06,
      "loss": 0.5104,
      "step": 17235
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4762600660324097,
      "learning_rate": 7.463255783748671e-06,
      "loss": 0.5212,
      "step": 17236
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3551268577575684,
      "learning_rate": 7.462175749062288e-06,
      "loss": 0.5447,
      "step": 17237
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6442514657974243,
      "learning_rate": 7.461095746015915e-06,
      "loss": 0.5306,
      "step": 17238
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5305639505386353,
      "learning_rate": 7.460015774623013e-06,
      "loss": 0.5024,
      "step": 17239
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8854628801345825,
      "learning_rate": 7.458935834897048e-06,
      "loss": 0.5989,
      "step": 17240
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.817547082901001,
      "learning_rate": 7.457855926851484e-06,
      "loss": 0.4232,
      "step": 17241
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.9139341115951538,
      "learning_rate": 7.456776050499783e-06,
      "loss": 0.6039,
      "step": 17242
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.8685083389282227,
      "learning_rate": 7.455696205855412e-06,
      "loss": 0.59,
      "step": 17243
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6959404945373535,
      "learning_rate": 7.4546163929318285e-06,
      "loss": 0.4898,
      "step": 17244
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.431037187576294,
      "learning_rate": 7.453536611742499e-06,
      "loss": 0.5191,
      "step": 17245
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.414008617401123,
      "learning_rate": 7.452456862300885e-06,
      "loss": 0.5005,
      "step": 17246
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5655455589294434,
      "learning_rate": 7.451377144620446e-06,
      "loss": 0.5725,
      "step": 17247
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4149023294448853,
      "learning_rate": 7.450297458714646e-06,
      "loss": 0.4864,
      "step": 17248
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7120558023452759,
      "learning_rate": 7.449217804596945e-06,
      "loss": 0.5338,
      "step": 17249
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5335944890975952,
      "learning_rate": 7.448138182280802e-06,
      "loss": 0.5148,
      "step": 17250
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4930368661880493,
      "learning_rate": 7.44705859177968e-06,
      "loss": 0.5487,
      "step": 17251
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4334495067596436,
      "learning_rate": 7.445979033107034e-06,
      "loss": 0.5855,
      "step": 17252
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3540129661560059,
      "learning_rate": 7.444899506276326e-06,
      "loss": 0.479,
      "step": 17253
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.63748037815094,
      "learning_rate": 7.443820011301017e-06,
      "loss": 0.5072,
      "step": 17254
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.409396767616272,
      "learning_rate": 7.442740548194563e-06,
      "loss": 0.4757,
      "step": 17255
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4047092199325562,
      "learning_rate": 7.441661116970422e-06,
      "loss": 0.4201,
      "step": 17256
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.7113006114959717,
      "learning_rate": 7.440581717642053e-06,
      "loss": 0.5371,
      "step": 17257
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3135709762573242,
      "learning_rate": 7.439502350222909e-06,
      "loss": 0.4798,
      "step": 17258
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6876959800720215,
      "learning_rate": 7.438423014726455e-06,
      "loss": 0.5584,
      "step": 17259
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5119240283966064,
      "learning_rate": 7.4373437111661385e-06,
      "loss": 0.5086,
      "step": 17260
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5866987705230713,
      "learning_rate": 7.43626443955542e-06,
      "loss": 0.6134,
      "step": 17261
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.4067912101745605,
      "learning_rate": 7.43518519990776e-06,
      "loss": 0.439,
      "step": 17262
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.388648509979248,
      "learning_rate": 7.4341059922366045e-06,
      "loss": 0.481,
      "step": 17263
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5352885723114014,
      "learning_rate": 7.4330268165554144e-06,
      "loss": 0.4903,
      "step": 17264
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6154862642288208,
      "learning_rate": 7.43194767287764e-06,
      "loss": 0.4773,
      "step": 17265
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7569278478622437,
      "learning_rate": 7.43086856121674e-06,
      "loss": 0.5296,
      "step": 17266
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7272608280181885,
      "learning_rate": 7.429789481586166e-06,
      "loss": 0.5609,
      "step": 17267
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.35018789768219,
      "learning_rate": 7.428710433999369e-06,
      "loss": 0.5326,
      "step": 17268
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7031750679016113,
      "learning_rate": 7.427631418469807e-06,
      "loss": 0.6018,
      "step": 17269
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5265485048294067,
      "learning_rate": 7.426552435010927e-06,
      "loss": 0.5503,
      "step": 17270
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4007481336593628,
      "learning_rate": 7.425473483636185e-06,
      "loss": 0.5237,
      "step": 17271
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5751253366470337,
      "learning_rate": 7.424394564359033e-06,
      "loss": 0.5551,
      "step": 17272
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.729984998703003,
      "learning_rate": 7.423315677192918e-06,
      "loss": 0.4976,
      "step": 17273
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6376631259918213,
      "learning_rate": 7.422236822151294e-06,
      "loss": 0.4714,
      "step": 17274
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6667975187301636,
      "learning_rate": 7.421157999247613e-06,
      "loss": 0.5718,
      "step": 17275
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.45997953414917,
      "learning_rate": 7.4200792084953214e-06,
      "loss": 0.5119,
      "step": 17276
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4592431783676147,
      "learning_rate": 7.419000449907873e-06,
      "loss": 0.6059,
      "step": 17277
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3847012519836426,
      "learning_rate": 7.417921723498711e-06,
      "loss": 0.5405,
      "step": 17278
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6475162506103516,
      "learning_rate": 7.4168430292812895e-06,
      "loss": 0.5053,
      "step": 17279
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7754749059677124,
      "learning_rate": 7.4157643672690586e-06,
      "loss": 0.6023,
      "step": 17280
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5243672132492065,
      "learning_rate": 7.414685737475459e-06,
      "loss": 0.4776,
      "step": 17281
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.257049560546875,
      "learning_rate": 7.413607139913945e-06,
      "loss": 0.5665,
      "step": 17282
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5260558128356934,
      "learning_rate": 7.412528574597961e-06,
      "loss": 0.5206,
      "step": 17283
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7861926555633545,
      "learning_rate": 7.411450041540953e-06,
      "loss": 0.5009,
      "step": 17284
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6594072580337524,
      "learning_rate": 7.4103715407563715e-06,
      "loss": 0.5711,
      "step": 17285
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4918304681777954,
      "learning_rate": 7.409293072257657e-06,
      "loss": 0.5574,
      "step": 17286
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6272451877593994,
      "learning_rate": 7.4082146360582596e-06,
      "loss": 0.5889,
      "step": 17287
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5181139707565308,
      "learning_rate": 7.407136232171625e-06,
      "loss": 0.4328,
      "step": 17288
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6959820985794067,
      "learning_rate": 7.406057860611193e-06,
      "loss": 0.4975,
      "step": 17289
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.48244047164917,
      "learning_rate": 7.404979521390414e-06,
      "loss": 0.5548,
      "step": 17290
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6118593215942383,
      "learning_rate": 7.4039012145227285e-06,
      "loss": 0.5981,
      "step": 17291
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5537495613098145,
      "learning_rate": 7.402822940021579e-06,
      "loss": 0.494,
      "step": 17292
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2893110513687134,
      "learning_rate": 7.4017446979004134e-06,
      "loss": 0.4489,
      "step": 17293
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4914575815200806,
      "learning_rate": 7.40066648817267e-06,
      "loss": 0.5109,
      "step": 17294
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4453953504562378,
      "learning_rate": 7.399588310851795e-06,
      "loss": 0.475,
      "step": 17295
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8334771394729614,
      "learning_rate": 7.398510165951227e-06,
      "loss": 0.58,
      "step": 17296
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.525718331336975,
      "learning_rate": 7.3974320534844095e-06,
      "loss": 0.5308,
      "step": 17297
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.406323790550232,
      "learning_rate": 7.396353973464785e-06,
      "loss": 0.4697,
      "step": 17298
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6469593048095703,
      "learning_rate": 7.395275925905789e-06,
      "loss": 0.48,
      "step": 17299
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5492264032363892,
      "learning_rate": 7.394197910820869e-06,
      "loss": 0.4652,
      "step": 17300
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.548207402229309,
      "learning_rate": 7.3931199282234616e-06,
      "loss": 0.6024,
      "step": 17301
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.10019588470459,
      "learning_rate": 7.392041978127005e-06,
      "loss": 0.5835,
      "step": 17302
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3136134147644043,
      "learning_rate": 7.390964060544941e-06,
      "loss": 0.4424,
      "step": 17303
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3995788097381592,
      "learning_rate": 7.389886175490706e-06,
      "loss": 0.4481,
      "step": 17304
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6398766040802002,
      "learning_rate": 7.388808322977739e-06,
      "loss": 0.6221,
      "step": 17305
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.452818512916565,
      "learning_rate": 7.387730503019482e-06,
      "loss": 0.5678,
      "step": 17306
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6560136079788208,
      "learning_rate": 7.386652715629368e-06,
      "loss": 0.3792,
      "step": 17307
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.9927705526351929,
      "learning_rate": 7.385574960820836e-06,
      "loss": 0.467,
      "step": 17308
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.0561940670013428,
      "learning_rate": 7.38449723860732e-06,
      "loss": 0.6471,
      "step": 17309
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4128313064575195,
      "learning_rate": 7.38341954900226e-06,
      "loss": 0.4784,
      "step": 17310
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7559136152267456,
      "learning_rate": 7.382341892019092e-06,
      "loss": 0.5319,
      "step": 17311
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6082783937454224,
      "learning_rate": 7.3812642676712465e-06,
      "loss": 0.4703,
      "step": 17312
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5401313304901123,
      "learning_rate": 7.380186675972163e-06,
      "loss": 0.5529,
      "step": 17313
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5807442665100098,
      "learning_rate": 7.379109116935279e-06,
      "loss": 0.6102,
      "step": 17314
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5312834978103638,
      "learning_rate": 7.3780315905740215e-06,
      "loss": 0.5996,
      "step": 17315
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6366866827011108,
      "learning_rate": 7.376954096901832e-06,
      "loss": 0.5207,
      "step": 17316
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.40715491771698,
      "learning_rate": 7.375876635932138e-06,
      "loss": 0.5583,
      "step": 17317
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7451380491256714,
      "learning_rate": 7.374799207678374e-06,
      "loss": 0.4899,
      "step": 17318
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5982404947280884,
      "learning_rate": 7.373721812153976e-06,
      "loss": 0.5184,
      "step": 17319
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.52171790599823,
      "learning_rate": 7.372644449372373e-06,
      "loss": 0.5188,
      "step": 17320
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4432519674301147,
      "learning_rate": 7.371567119346999e-06,
      "loss": 0.4046,
      "step": 17321
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.514496088027954,
      "learning_rate": 7.370489822091283e-06,
      "loss": 0.4618,
      "step": 17322
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4665392637252808,
      "learning_rate": 7.369412557618655e-06,
      "loss": 0.5224,
      "step": 17323
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.63388192653656,
      "learning_rate": 7.3683353259425515e-06,
      "loss": 0.5491,
      "step": 17324
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3330005407333374,
      "learning_rate": 7.367258127076397e-06,
      "loss": 0.52,
      "step": 17325
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5369863510131836,
      "learning_rate": 7.366180961033624e-06,
      "loss": 0.6089,
      "step": 17326
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3419321775436401,
      "learning_rate": 7.3651038278276634e-06,
      "loss": 0.476,
      "step": 17327
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4926822185516357,
      "learning_rate": 7.36402672747194e-06,
      "loss": 0.6708,
      "step": 17328
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5152106285095215,
      "learning_rate": 7.362949659979888e-06,
      "loss": 0.5045,
      "step": 17329
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2700802087783813,
      "learning_rate": 7.361872625364929e-06,
      "loss": 0.5508,
      "step": 17330
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2566635608673096,
      "learning_rate": 7.360795623640494e-06,
      "loss": 0.4653,
      "step": 17331
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.7130687236785889,
      "learning_rate": 7.359718654820014e-06,
      "loss": 0.5367,
      "step": 17332
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3240419626235962,
      "learning_rate": 7.358641718916909e-06,
      "loss": 0.4613,
      "step": 17333
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8203130960464478,
      "learning_rate": 7.357564815944611e-06,
      "loss": 0.5436,
      "step": 17334
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3661024570465088,
      "learning_rate": 7.356487945916544e-06,
      "loss": 0.6356,
      "step": 17335
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2998073101043701,
      "learning_rate": 7.355411108846133e-06,
      "loss": 0.4188,
      "step": 17336
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6390098333358765,
      "learning_rate": 7.354334304746806e-06,
      "loss": 0.4398,
      "step": 17337
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6535861492156982,
      "learning_rate": 7.353257533631984e-06,
      "loss": 0.6743,
      "step": 17338
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3638715744018555,
      "learning_rate": 7.352180795515095e-06,
      "loss": 0.5116,
      "step": 17339
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5838873386383057,
      "learning_rate": 7.3511040904095625e-06,
      "loss": 0.5959,
      "step": 17340
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.656142234802246,
      "learning_rate": 7.350027418328809e-06,
      "loss": 0.462,
      "step": 17341
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5316787958145142,
      "learning_rate": 7.348950779286259e-06,
      "loss": 0.4911,
      "step": 17342
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6352596282958984,
      "learning_rate": 7.347874173295333e-06,
      "loss": 0.5681,
      "step": 17343
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4637004137039185,
      "learning_rate": 7.346797600369455e-06,
      "loss": 0.417,
      "step": 17344
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6116821765899658,
      "learning_rate": 7.34572106052205e-06,
      "loss": 0.5685,
      "step": 17345
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.509359359741211,
      "learning_rate": 7.344644553766534e-06,
      "loss": 0.5805,
      "step": 17346
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.1676928997039795,
      "learning_rate": 7.343568080116332e-06,
      "loss": 0.5958,
      "step": 17347
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6760677099227905,
      "learning_rate": 7.342491639584863e-06,
      "loss": 0.5407,
      "step": 17348
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.428902506828308,
      "learning_rate": 7.341415232185549e-06,
      "loss": 0.4823,
      "step": 17349
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.145237922668457,
      "learning_rate": 7.3403388579318104e-06,
      "loss": 0.5829,
      "step": 17350
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8634248971939087,
      "learning_rate": 7.3392625168370635e-06,
      "loss": 0.4502,
      "step": 17351
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.6495500802993774,
      "learning_rate": 7.338186208914729e-06,
      "loss": 0.5364,
      "step": 17352
    },
    {
      "epoch": 1.79,
      "grad_norm": 2.012561559677124,
      "learning_rate": 7.337109934178228e-06,
      "loss": 0.5346,
      "step": 17353
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.544094204902649,
      "learning_rate": 7.336033692640976e-06,
      "loss": 0.5442,
      "step": 17354
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.426676630973816,
      "learning_rate": 7.334957484316394e-06,
      "loss": 0.4284,
      "step": 17355
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1730889081954956,
      "learning_rate": 7.333881309217894e-06,
      "loss": 0.5123,
      "step": 17356
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5819714069366455,
      "learning_rate": 7.332805167358897e-06,
      "loss": 0.5655,
      "step": 17357
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6597394943237305,
      "learning_rate": 7.3317290587528215e-06,
      "loss": 0.6046,
      "step": 17358
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.27599036693573,
      "learning_rate": 7.330652983413078e-06,
      "loss": 0.5405,
      "step": 17359
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8251839876174927,
      "learning_rate": 7.329576941353089e-06,
      "loss": 0.5715,
      "step": 17360
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6978039741516113,
      "learning_rate": 7.328500932586264e-06,
      "loss": 0.4875,
      "step": 17361
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5066686868667603,
      "learning_rate": 7.327424957126022e-06,
      "loss": 0.4294,
      "step": 17362
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4171122312545776,
      "learning_rate": 7.326349014985777e-06,
      "loss": 0.4659,
      "step": 17363
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4161819219589233,
      "learning_rate": 7.3252731061789396e-06,
      "loss": 0.5311,
      "step": 17364
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5433430671691895,
      "learning_rate": 7.324197230718927e-06,
      "loss": 0.4154,
      "step": 17365
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2762727737426758,
      "learning_rate": 7.323121388619154e-06,
      "loss": 0.5055,
      "step": 17366
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3883614540100098,
      "learning_rate": 7.322045579893029e-06,
      "loss": 0.5958,
      "step": 17367
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8621342182159424,
      "learning_rate": 7.320969804553969e-06,
      "loss": 0.5643,
      "step": 17368
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5069106817245483,
      "learning_rate": 7.319894062615383e-06,
      "loss": 0.5068,
      "step": 17369
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.714169979095459,
      "learning_rate": 7.318818354090682e-06,
      "loss": 0.5484,
      "step": 17370
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7745521068572998,
      "learning_rate": 7.317742678993281e-06,
      "loss": 0.5653,
      "step": 17371
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5363906621932983,
      "learning_rate": 7.3166670373365875e-06,
      "loss": 0.5348,
      "step": 17372
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7648661136627197,
      "learning_rate": 7.315591429134016e-06,
      "loss": 0.527,
      "step": 17373
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.563212513923645,
      "learning_rate": 7.3145158543989704e-06,
      "loss": 0.4909,
      "step": 17374
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4607317447662354,
      "learning_rate": 7.313440313144864e-06,
      "loss": 0.5557,
      "step": 17375
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4238687753677368,
      "learning_rate": 7.312364805385108e-06,
      "loss": 0.5141,
      "step": 17376
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6207095384597778,
      "learning_rate": 7.311289331133107e-06,
      "loss": 0.5641,
      "step": 17377
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2639131546020508,
      "learning_rate": 7.310213890402269e-06,
      "loss": 0.4558,
      "step": 17378
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8587079048156738,
      "learning_rate": 7.3091384832060086e-06,
      "loss": 0.5812,
      "step": 17379
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4269086122512817,
      "learning_rate": 7.308063109557725e-06,
      "loss": 0.5922,
      "step": 17380
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.362113356590271,
      "learning_rate": 7.306987769470832e-06,
      "loss": 0.4667,
      "step": 17381
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4551246166229248,
      "learning_rate": 7.305912462958731e-06,
      "loss": 0.4866,
      "step": 17382
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.677130103111267,
      "learning_rate": 7.30483719003483e-06,
      "loss": 0.4708,
      "step": 17383
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.426619291305542,
      "learning_rate": 7.303761950712538e-06,
      "loss": 0.6287,
      "step": 17384
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8685548305511475,
      "learning_rate": 7.302686745005254e-06,
      "loss": 0.6284,
      "step": 17385
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6846745014190674,
      "learning_rate": 7.30161157292639e-06,
      "loss": 0.5414,
      "step": 17386
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7639564275741577,
      "learning_rate": 7.300536434489347e-06,
      "loss": 0.5548,
      "step": 17387
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6320395469665527,
      "learning_rate": 7.299461329707527e-06,
      "loss": 0.5368,
      "step": 17388
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5220547914505005,
      "learning_rate": 7.2983862585943395e-06,
      "loss": 0.413,
      "step": 17389
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4970693588256836,
      "learning_rate": 7.297311221163182e-06,
      "loss": 0.5106,
      "step": 17390
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.734263300895691,
      "learning_rate": 7.29623621742746e-06,
      "loss": 0.5152,
      "step": 17391
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5675628185272217,
      "learning_rate": 7.2951612474005794e-06,
      "loss": 0.4693,
      "step": 17392
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6165813207626343,
      "learning_rate": 7.294086311095935e-06,
      "loss": 0.4761,
      "step": 17393
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6454216241836548,
      "learning_rate": 7.293011408526934e-06,
      "loss": 0.4821,
      "step": 17394
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7194700241088867,
      "learning_rate": 7.2919365397069745e-06,
      "loss": 0.6182,
      "step": 17395
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4637523889541626,
      "learning_rate": 7.290861704649459e-06,
      "loss": 0.5126,
      "step": 17396
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4967007637023926,
      "learning_rate": 7.289786903367788e-06,
      "loss": 0.4913,
      "step": 17397
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.0563645362854004,
      "learning_rate": 7.28871213587536e-06,
      "loss": 0.6225,
      "step": 17398
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2252404689788818,
      "learning_rate": 7.2876374021855776e-06,
      "loss": 0.5426,
      "step": 17399
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7395838499069214,
      "learning_rate": 7.2865627023118354e-06,
      "loss": 0.62,
      "step": 17400
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4788066148757935,
      "learning_rate": 7.285488036267534e-06,
      "loss": 0.4751,
      "step": 17401
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4917488098144531,
      "learning_rate": 7.284413404066075e-06,
      "loss": 0.5228,
      "step": 17402
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5523018836975098,
      "learning_rate": 7.283338805720852e-06,
      "loss": 0.6318,
      "step": 17403
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6355623006820679,
      "learning_rate": 7.2822642412452625e-06,
      "loss": 0.5868,
      "step": 17404
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.664971947669983,
      "learning_rate": 7.2811897106527075e-06,
      "loss": 0.5201,
      "step": 17405
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3893884420394897,
      "learning_rate": 7.280115213956578e-06,
      "loss": 0.5367,
      "step": 17406
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4563522338867188,
      "learning_rate": 7.279040751170274e-06,
      "loss": 0.3977,
      "step": 17407
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.615366816520691,
      "learning_rate": 7.277966322307191e-06,
      "loss": 0.5505,
      "step": 17408
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5569911003112793,
      "learning_rate": 7.276891927380722e-06,
      "loss": 0.3689,
      "step": 17409
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4190131425857544,
      "learning_rate": 7.275817566404265e-06,
      "loss": 0.505,
      "step": 17410
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7635833024978638,
      "learning_rate": 7.274743239391211e-06,
      "loss": 0.4961,
      "step": 17411
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4466466903686523,
      "learning_rate": 7.2736689463549595e-06,
      "loss": 0.431,
      "step": 17412
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4466668367385864,
      "learning_rate": 7.272594687308896e-06,
      "loss": 0.4503,
      "step": 17413
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.447630524635315,
      "learning_rate": 7.27152046226642e-06,
      "loss": 0.4474,
      "step": 17414
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6596167087554932,
      "learning_rate": 7.2704462712409255e-06,
      "loss": 0.5531,
      "step": 17415
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6975542306900024,
      "learning_rate": 7.269372114245798e-06,
      "loss": 0.6049,
      "step": 17416
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.501340627670288,
      "learning_rate": 7.268297991294433e-06,
      "loss": 0.5456,
      "step": 17417
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.536062479019165,
      "learning_rate": 7.267223902400224e-06,
      "loss": 0.5356,
      "step": 17418
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3224074840545654,
      "learning_rate": 7.26614984757656e-06,
      "loss": 0.5254,
      "step": 17419
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6244837045669556,
      "learning_rate": 7.265075826836834e-06,
      "loss": 0.6155,
      "step": 17420
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6642173528671265,
      "learning_rate": 7.2640018401944315e-06,
      "loss": 0.5895,
      "step": 17421
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.079956293106079,
      "learning_rate": 7.262927887662745e-06,
      "loss": 0.5614,
      "step": 17422
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.585973858833313,
      "learning_rate": 7.261853969255166e-06,
      "loss": 0.5913,
      "step": 17423
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4886621236801147,
      "learning_rate": 7.260780084985078e-06,
      "loss": 0.5286,
      "step": 17424
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6878670454025269,
      "learning_rate": 7.259706234865876e-06,
      "loss": 0.5461,
      "step": 17425
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5209823846817017,
      "learning_rate": 7.258632418910942e-06,
      "loss": 0.4896,
      "step": 17426
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3167816400527954,
      "learning_rate": 7.257558637133668e-06,
      "loss": 0.5186,
      "step": 17427
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6187325716018677,
      "learning_rate": 7.256484889547441e-06,
      "loss": 0.6003,
      "step": 17428
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.8966212272644043,
      "learning_rate": 7.255411176165646e-06,
      "loss": 0.5365,
      "step": 17429
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5575448274612427,
      "learning_rate": 7.254337497001668e-06,
      "loss": 0.5779,
      "step": 17430
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.420130491256714,
      "learning_rate": 7.253263852068897e-06,
      "loss": 0.6677,
      "step": 17431
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.888698935508728,
      "learning_rate": 7.252190241380715e-06,
      "loss": 0.6338,
      "step": 17432
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.099024534225464,
      "learning_rate": 7.251116664950512e-06,
      "loss": 0.5912,
      "step": 17433
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.130483865737915,
      "learning_rate": 7.250043122791665e-06,
      "loss": 0.4637,
      "step": 17434
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6398282051086426,
      "learning_rate": 7.248969614917565e-06,
      "loss": 0.5611,
      "step": 17435
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.0869174003601074,
      "learning_rate": 7.247896141341593e-06,
      "loss": 0.5134,
      "step": 17436
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.365454912185669,
      "learning_rate": 7.246822702077132e-06,
      "loss": 0.5304,
      "step": 17437
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5246703624725342,
      "learning_rate": 7.245749297137568e-06,
      "loss": 0.4905,
      "step": 17438
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3370286226272583,
      "learning_rate": 7.244675926536279e-06,
      "loss": 0.5198,
      "step": 17439
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5348464250564575,
      "learning_rate": 7.243602590286649e-06,
      "loss": 0.6269,
      "step": 17440
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2900335788726807,
      "learning_rate": 7.242529288402063e-06,
      "loss": 0.5308,
      "step": 17441
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4336280822753906,
      "learning_rate": 7.241456020895896e-06,
      "loss": 0.5666,
      "step": 17442
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2825747728347778,
      "learning_rate": 7.240382787781532e-06,
      "loss": 0.6058,
      "step": 17443
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7338675260543823,
      "learning_rate": 7.2393095890723555e-06,
      "loss": 0.5274,
      "step": 17444
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7071813344955444,
      "learning_rate": 7.238236424781739e-06,
      "loss": 0.5739,
      "step": 17445
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4572052955627441,
      "learning_rate": 7.2371632949230685e-06,
      "loss": 0.5395,
      "step": 17446
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.7464244365692139,
      "learning_rate": 7.236090199509718e-06,
      "loss": 0.5579,
      "step": 17447
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4064067602157593,
      "learning_rate": 7.235017138555067e-06,
      "loss": 0.4942,
      "step": 17448
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5757478475570679,
      "learning_rate": 7.2339441120724995e-06,
      "loss": 0.5077,
      "step": 17449
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5758661031723022,
      "learning_rate": 7.232871120075384e-06,
      "loss": 0.4742,
      "step": 17450
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.055187702178955,
      "learning_rate": 7.231798162577108e-06,
      "loss": 0.5704,
      "step": 17451
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.386284589767456,
      "learning_rate": 7.230725239591039e-06,
      "loss": 0.3971,
      "step": 17452
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.517746090888977,
      "learning_rate": 7.229652351130559e-06,
      "loss": 0.4131,
      "step": 17453
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.56294584274292,
      "learning_rate": 7.228579497209045e-06,
      "loss": 0.5875,
      "step": 17454
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.692751169204712,
      "learning_rate": 7.227506677839868e-06,
      "loss": 0.4567,
      "step": 17455
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3978785276412964,
      "learning_rate": 7.226433893036409e-06,
      "loss": 0.5099,
      "step": 17456
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3372396230697632,
      "learning_rate": 7.225361142812036e-06,
      "loss": 0.3285,
      "step": 17457
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9351787567138672,
      "learning_rate": 7.224288427180128e-06,
      "loss": 0.5641,
      "step": 17458
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3773045539855957,
      "learning_rate": 7.22321574615406e-06,
      "loss": 0.5256,
      "step": 17459
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3241535425186157,
      "learning_rate": 7.222143099747201e-06,
      "loss": 0.4958,
      "step": 17460
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5947891473770142,
      "learning_rate": 7.221070487972928e-06,
      "loss": 0.5304,
      "step": 17461
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5171124935150146,
      "learning_rate": 7.219997910844613e-06,
      "loss": 0.4952,
      "step": 17462
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5486832857131958,
      "learning_rate": 7.218925368375625e-06,
      "loss": 0.6146,
      "step": 17463
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2738616466522217,
      "learning_rate": 7.217852860579342e-06,
      "loss": 0.5573,
      "step": 17464
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.677441954612732,
      "learning_rate": 7.216780387469128e-06,
      "loss": 0.394,
      "step": 17465
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6735813617706299,
      "learning_rate": 7.215707949058358e-06,
      "loss": 0.4427,
      "step": 17466
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9613239765167236,
      "learning_rate": 7.214635545360404e-06,
      "loss": 0.6365,
      "step": 17467
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5179611444473267,
      "learning_rate": 7.213563176388632e-06,
      "loss": 0.4798,
      "step": 17468
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.597028374671936,
      "learning_rate": 7.212490842156417e-06,
      "loss": 0.4478,
      "step": 17469
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5528535842895508,
      "learning_rate": 7.211418542677122e-06,
      "loss": 0.4836,
      "step": 17470
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5013337135314941,
      "learning_rate": 7.210346277964119e-06,
      "loss": 0.5124,
      "step": 17471
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6225340366363525,
      "learning_rate": 7.209274048030777e-06,
      "loss": 0.5101,
      "step": 17472
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6112003326416016,
      "learning_rate": 7.208201852890461e-06,
      "loss": 0.634,
      "step": 17473
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.630710244178772,
      "learning_rate": 7.207129692556543e-06,
      "loss": 0.7041,
      "step": 17474
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4231200218200684,
      "learning_rate": 7.206057567042386e-06,
      "loss": 0.5521,
      "step": 17475
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.415740728378296,
      "learning_rate": 7.204985476361357e-06,
      "loss": 0.5641,
      "step": 17476
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6321265697479248,
      "learning_rate": 7.203913420526827e-06,
      "loss": 0.5497,
      "step": 17477
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.722583293914795,
      "learning_rate": 7.202841399552154e-06,
      "loss": 0.5207,
      "step": 17478
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6454981565475464,
      "learning_rate": 7.201769413450708e-06,
      "loss": 0.5624,
      "step": 17479
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3812353610992432,
      "learning_rate": 7.200697462235853e-06,
      "loss": 0.4998,
      "step": 17480
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8464325666427612,
      "learning_rate": 7.199625545920954e-06,
      "loss": 0.5416,
      "step": 17481
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7223880290985107,
      "learning_rate": 7.198553664519374e-06,
      "loss": 0.554,
      "step": 17482
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4557496309280396,
      "learning_rate": 7.197481818044477e-06,
      "loss": 0.3919,
      "step": 17483
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5207563638687134,
      "learning_rate": 7.196410006509625e-06,
      "loss": 0.5143,
      "step": 17484
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5792866945266724,
      "learning_rate": 7.195338229928184e-06,
      "loss": 0.4841,
      "step": 17485
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.387547254562378,
      "learning_rate": 7.194266488313513e-06,
      "loss": 0.5014,
      "step": 17486
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4768856763839722,
      "learning_rate": 7.193194781678974e-06,
      "loss": 0.5332,
      "step": 17487
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4650107622146606,
      "learning_rate": 7.19212311003793e-06,
      "loss": 0.6176,
      "step": 17488
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7786195278167725,
      "learning_rate": 7.191051473403741e-06,
      "loss": 0.4402,
      "step": 17489
    },
    {
      "epoch": 1.81,
      "grad_norm": 2.1082863807678223,
      "learning_rate": 7.189979871789768e-06,
      "loss": 0.536,
      "step": 17490
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.565754771232605,
      "learning_rate": 7.188908305209369e-06,
      "loss": 0.5313,
      "step": 17491
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.853879451751709,
      "learning_rate": 7.1878367736759055e-06,
      "loss": 0.5829,
      "step": 17492
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4786112308502197,
      "learning_rate": 7.186765277202739e-06,
      "loss": 0.4656,
      "step": 17493
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4718111753463745,
      "learning_rate": 7.185693815803223e-06,
      "loss": 0.4411,
      "step": 17494
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.642259955406189,
      "learning_rate": 7.1846223894907205e-06,
      "loss": 0.5271,
      "step": 17495
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.367835283279419,
      "learning_rate": 7.183550998278585e-06,
      "loss": 0.5302,
      "step": 17496
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4040080308914185,
      "learning_rate": 7.182479642180178e-06,
      "loss": 0.4844,
      "step": 17497
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7385329008102417,
      "learning_rate": 7.181408321208856e-06,
      "loss": 0.5594,
      "step": 17498
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6845788955688477,
      "learning_rate": 7.180337035377971e-06,
      "loss": 0.5233,
      "step": 17499
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4579236507415771,
      "learning_rate": 7.179265784700884e-06,
      "loss": 0.4608,
      "step": 17500
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4129252433776855,
      "learning_rate": 7.17819456919095e-06,
      "loss": 0.513,
      "step": 17501
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4823651313781738,
      "learning_rate": 7.177123388861522e-06,
      "loss": 0.4981,
      "step": 17502
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4961681365966797,
      "learning_rate": 7.176052243725959e-06,
      "loss": 0.5946,
      "step": 17503
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8659989833831787,
      "learning_rate": 7.17498113379761e-06,
      "loss": 0.5616,
      "step": 17504
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6645004749298096,
      "learning_rate": 7.173910059089832e-06,
      "loss": 0.5509,
      "step": 17505
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6623563766479492,
      "learning_rate": 7.1728390196159805e-06,
      "loss": 0.5604,
      "step": 17506
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.742677927017212,
      "learning_rate": 7.1717680153894024e-06,
      "loss": 0.5546,
      "step": 17507
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6111310720443726,
      "learning_rate": 7.1706970464234575e-06,
      "loss": 0.5016,
      "step": 17508
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7212772369384766,
      "learning_rate": 7.1696261127314915e-06,
      "loss": 0.4747,
      "step": 17509
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.722430944442749,
      "learning_rate": 7.168555214326859e-06,
      "loss": 0.4465,
      "step": 17510
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7898578643798828,
      "learning_rate": 7.167484351222913e-06,
      "loss": 0.5355,
      "step": 17511
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3458094596862793,
      "learning_rate": 7.166413523433002e-06,
      "loss": 0.5142,
      "step": 17512
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7913157939910889,
      "learning_rate": 7.1653427309704774e-06,
      "loss": 0.4754,
      "step": 17513
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5544548034667969,
      "learning_rate": 7.16427197384869e-06,
      "loss": 0.4938,
      "step": 17514
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9262126684188843,
      "learning_rate": 7.163201252080987e-06,
      "loss": 0.6033,
      "step": 17515
    },
    {
      "epoch": 1.81,
      "grad_norm": 2.4704537391662598,
      "learning_rate": 7.162130565680719e-06,
      "loss": 0.5765,
      "step": 17516
    },
    {
      "epoch": 1.81,
      "grad_norm": 2.1171207427978516,
      "learning_rate": 7.161059914661234e-06,
      "loss": 0.5497,
      "step": 17517
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4486624002456665,
      "learning_rate": 7.15998929903588e-06,
      "loss": 0.5483,
      "step": 17518
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.8103885650634766,
      "learning_rate": 7.158918718818007e-06,
      "loss": 0.5357,
      "step": 17519
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4072909355163574,
      "learning_rate": 7.157848174020961e-06,
      "loss": 0.51,
      "step": 17520
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4388123750686646,
      "learning_rate": 7.156777664658088e-06,
      "loss": 0.463,
      "step": 17521
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.50906240940094,
      "learning_rate": 7.1557071907427335e-06,
      "loss": 0.4751,
      "step": 17522
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.66567862033844,
      "learning_rate": 7.154636752288244e-06,
      "loss": 0.5335,
      "step": 17523
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7696149349212646,
      "learning_rate": 7.1535663493079695e-06,
      "loss": 0.6075,
      "step": 17524
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5859707593917847,
      "learning_rate": 7.152495981815251e-06,
      "loss": 0.5429,
      "step": 17525
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4596483707427979,
      "learning_rate": 7.151425649823431e-06,
      "loss": 0.5321,
      "step": 17526
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5774509906768799,
      "learning_rate": 7.150355353345862e-06,
      "loss": 0.5651,
      "step": 17527
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7191925048828125,
      "learning_rate": 7.149285092395877e-06,
      "loss": 0.4614,
      "step": 17528
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4687342643737793,
      "learning_rate": 7.148214866986829e-06,
      "loss": 0.5056,
      "step": 17529
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.801900029182434,
      "learning_rate": 7.147144677132054e-06,
      "loss": 0.4649,
      "step": 17530
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2671221494674683,
      "learning_rate": 7.1460745228448965e-06,
      "loss": 0.5002,
      "step": 17531
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.357782244682312,
      "learning_rate": 7.145004404138702e-06,
      "loss": 0.3526,
      "step": 17532
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.9084858894348145,
      "learning_rate": 7.143934321026807e-06,
      "loss": 0.4725,
      "step": 17533
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4437593221664429,
      "learning_rate": 7.142864273522556e-06,
      "loss": 0.4366,
      "step": 17534
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.592318058013916,
      "learning_rate": 7.1417942616392865e-06,
      "loss": 0.5733,
      "step": 17535
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7867895364761353,
      "learning_rate": 7.1407242853903416e-06,
      "loss": 0.5451,
      "step": 17536
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7328294515609741,
      "learning_rate": 7.139654344789062e-06,
      "loss": 0.4581,
      "step": 17537
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7733839750289917,
      "learning_rate": 7.138584439848783e-06,
      "loss": 0.592,
      "step": 17538
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.301340937614441,
      "learning_rate": 7.137514570582847e-06,
      "loss": 0.485,
      "step": 17539
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5186002254486084,
      "learning_rate": 7.136444737004591e-06,
      "loss": 0.5677,
      "step": 17540
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6176514625549316,
      "learning_rate": 7.1353749391273516e-06,
      "loss": 0.5312,
      "step": 17541
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6354411840438843,
      "learning_rate": 7.134305176964471e-06,
      "loss": 0.5045,
      "step": 17542
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.757375955581665,
      "learning_rate": 7.1332354505292815e-06,
      "loss": 0.445,
      "step": 17543
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6325184106826782,
      "learning_rate": 7.132165759835121e-06,
      "loss": 0.5285,
      "step": 17544
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4892809391021729,
      "learning_rate": 7.1310961048953296e-06,
      "loss": 0.5032,
      "step": 17545
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6785693168640137,
      "learning_rate": 7.130026485723236e-06,
      "loss": 0.4814,
      "step": 17546
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6218334436416626,
      "learning_rate": 7.128956902332182e-06,
      "loss": 0.5665,
      "step": 17547
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7533031702041626,
      "learning_rate": 7.127887354735499e-06,
      "loss": 0.5266,
      "step": 17548
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.378507375717163,
      "learning_rate": 7.1268178429465214e-06,
      "loss": 0.4784,
      "step": 17549
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6708322763442993,
      "learning_rate": 7.125748366978586e-06,
      "loss": 0.5775,
      "step": 17550
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6457561254501343,
      "learning_rate": 7.1246789268450235e-06,
      "loss": 0.5922,
      "step": 17551
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5618789196014404,
      "learning_rate": 7.123609522559167e-06,
      "loss": 0.5409,
      "step": 17552
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.9159214496612549,
      "learning_rate": 7.122540154134353e-06,
      "loss": 0.5913,
      "step": 17553
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4374668598175049,
      "learning_rate": 7.121470821583909e-06,
      "loss": 0.5893,
      "step": 17554
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.636881709098816,
      "learning_rate": 7.12040152492117e-06,
      "loss": 0.4961,
      "step": 17555
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8538681268692017,
      "learning_rate": 7.119332264159464e-06,
      "loss": 0.5349,
      "step": 17556
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.376196265220642,
      "learning_rate": 7.118263039312124e-06,
      "loss": 0.3987,
      "step": 17557
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6495625972747803,
      "learning_rate": 7.1171938503924835e-06,
      "loss": 0.6639,
      "step": 17558
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4866700172424316,
      "learning_rate": 7.116124697413865e-06,
      "loss": 0.3786,
      "step": 17559
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.699857473373413,
      "learning_rate": 7.115055580389606e-06,
      "loss": 0.542,
      "step": 17560
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7794692516326904,
      "learning_rate": 7.11398649933303e-06,
      "loss": 0.5454,
      "step": 17561
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5737777948379517,
      "learning_rate": 7.112917454257467e-06,
      "loss": 0.4278,
      "step": 17562
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.670973300933838,
      "learning_rate": 7.111848445176247e-06,
      "loss": 0.4895,
      "step": 17563
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5271742343902588,
      "learning_rate": 7.110779472102696e-06,
      "loss": 0.5288,
      "step": 17564
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8096635341644287,
      "learning_rate": 7.109710535050141e-06,
      "loss": 0.6046,
      "step": 17565
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8251609802246094,
      "learning_rate": 7.108641634031911e-06,
      "loss": 0.5377,
      "step": 17566
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6924606561660767,
      "learning_rate": 7.1075727690613296e-06,
      "loss": 0.5266,
      "step": 17567
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3891701698303223,
      "learning_rate": 7.106503940151726e-06,
      "loss": 0.5438,
      "step": 17568
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5155143737792969,
      "learning_rate": 7.105435147316422e-06,
      "loss": 0.5384,
      "step": 17569
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4100010395050049,
      "learning_rate": 7.104366390568744e-06,
      "loss": 0.501,
      "step": 17570
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.110912322998047,
      "learning_rate": 7.103297669922019e-06,
      "loss": 0.5933,
      "step": 17571
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6935762166976929,
      "learning_rate": 7.102228985389567e-06,
      "loss": 0.4598,
      "step": 17572
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5917766094207764,
      "learning_rate": 7.101160336984717e-06,
      "loss": 0.5683,
      "step": 17573
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6395940780639648,
      "learning_rate": 7.100091724720787e-06,
      "loss": 0.5931,
      "step": 17574
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.9462839365005493,
      "learning_rate": 7.099023148611102e-06,
      "loss": 0.6685,
      "step": 17575
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7202775478363037,
      "learning_rate": 7.097954608668984e-06,
      "loss": 0.5799,
      "step": 17576
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.9654179811477661,
      "learning_rate": 7.0968861049077595e-06,
      "loss": 0.5098,
      "step": 17577
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6410962343215942,
      "learning_rate": 7.09581763734074e-06,
      "loss": 0.5623,
      "step": 17578
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.375360369682312,
      "learning_rate": 7.094749205981254e-06,
      "loss": 0.4692,
      "step": 17579
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4908808469772339,
      "learning_rate": 7.093680810842622e-06,
      "loss": 0.497,
      "step": 17580
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3602036237716675,
      "learning_rate": 7.092612451938159e-06,
      "loss": 0.3939,
      "step": 17581
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.384606122970581,
      "learning_rate": 7.09154412928119e-06,
      "loss": 0.4913,
      "step": 17582
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.359883189201355,
      "learning_rate": 7.0904758428850295e-06,
      "loss": 0.4626,
      "step": 17583
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4987008571624756,
      "learning_rate": 7.089407592762999e-06,
      "loss": 0.4899,
      "step": 17584
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3089096546173096,
      "learning_rate": 7.088339378928419e-06,
      "loss": 0.4667,
      "step": 17585
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3688840866088867,
      "learning_rate": 7.087271201394602e-06,
      "loss": 0.4861,
      "step": 17586
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4312310218811035,
      "learning_rate": 7.086203060174869e-06,
      "loss": 0.5499,
      "step": 17587
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6456358432769775,
      "learning_rate": 7.085134955282538e-06,
      "loss": 0.4442,
      "step": 17588
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.466672420501709,
      "learning_rate": 7.084066886730922e-06,
      "loss": 0.4522,
      "step": 17589
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4741615056991577,
      "learning_rate": 7.0829988545333394e-06,
      "loss": 0.5393,
      "step": 17590
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7815656661987305,
      "learning_rate": 7.081930858703103e-06,
      "loss": 0.5142,
      "step": 17591
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4616895914077759,
      "learning_rate": 7.080862899253531e-06,
      "loss": 0.4738,
      "step": 17592
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3930178880691528,
      "learning_rate": 7.079794976197938e-06,
      "loss": 0.5154,
      "step": 17593
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3298734426498413,
      "learning_rate": 7.0787270895496354e-06,
      "loss": 0.4557,
      "step": 17594
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3956223726272583,
      "learning_rate": 7.077659239321941e-06,
      "loss": 0.5185,
      "step": 17595
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4953553676605225,
      "learning_rate": 7.076591425528163e-06,
      "loss": 0.5672,
      "step": 17596
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4085533618927002,
      "learning_rate": 7.075523648181617e-06,
      "loss": 0.497,
      "step": 17597
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3374587297439575,
      "learning_rate": 7.074455907295618e-06,
      "loss": 0.5088,
      "step": 17598
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8446210622787476,
      "learning_rate": 7.073388202883473e-06,
      "loss": 0.5582,
      "step": 17599
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4304770231246948,
      "learning_rate": 7.072320534958494e-06,
      "loss": 0.5247,
      "step": 17600
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.1126883029937744,
      "learning_rate": 7.071252903533999e-06,
      "loss": 0.6296,
      "step": 17601
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4625375270843506,
      "learning_rate": 7.070185308623289e-06,
      "loss": 0.4939,
      "step": 17602
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.27805495262146,
      "learning_rate": 7.0691177502396815e-06,
      "loss": 0.4488,
      "step": 17603
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6048541069030762,
      "learning_rate": 7.0680502283964805e-06,
      "loss": 0.5864,
      "step": 17604
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4309132099151611,
      "learning_rate": 7.066982743106998e-06,
      "loss": 0.5108,
      "step": 17605
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5097002983093262,
      "learning_rate": 7.065915294384545e-06,
      "loss": 0.4797,
      "step": 17606
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.645813226699829,
      "learning_rate": 7.064847882242425e-06,
      "loss": 0.5882,
      "step": 17607
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4744218587875366,
      "learning_rate": 7.063780506693949e-06,
      "loss": 0.573,
      "step": 17608
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6849935054779053,
      "learning_rate": 7.062713167752422e-06,
      "loss": 0.6211,
      "step": 17609
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.661716103553772,
      "learning_rate": 7.061645865431153e-06,
      "loss": 0.4712,
      "step": 17610
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4945886135101318,
      "learning_rate": 7.06057859974345e-06,
      "loss": 0.4824,
      "step": 17611
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6796265840530396,
      "learning_rate": 7.059511370702614e-06,
      "loss": 0.5209,
      "step": 17612
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5699117183685303,
      "learning_rate": 7.058444178321955e-06,
      "loss": 0.414,
      "step": 17613
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4984134435653687,
      "learning_rate": 7.057377022614777e-06,
      "loss": 0.4053,
      "step": 17614
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.085649013519287,
      "learning_rate": 7.056309903594383e-06,
      "loss": 0.5002,
      "step": 17615
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.0522356033325195,
      "learning_rate": 7.0552428212740796e-06,
      "loss": 0.5051,
      "step": 17616
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.294840931892395,
      "learning_rate": 7.054175775667167e-06,
      "loss": 0.4142,
      "step": 17617
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.630104660987854,
      "learning_rate": 7.053108766786951e-06,
      "loss": 0.5152,
      "step": 17618
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8179662227630615,
      "learning_rate": 7.052041794646735e-06,
      "loss": 0.5844,
      "step": 17619
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3004895448684692,
      "learning_rate": 7.05097485925982e-06,
      "loss": 0.3754,
      "step": 17620
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7300031185150146,
      "learning_rate": 7.049907960639511e-06,
      "loss": 0.7027,
      "step": 17621
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5666148662567139,
      "learning_rate": 7.048841098799104e-06,
      "loss": 0.488,
      "step": 17622
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.898128867149353,
      "learning_rate": 7.047774273751902e-06,
      "loss": 0.5654,
      "step": 17623
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5608372688293457,
      "learning_rate": 7.046707485511209e-06,
      "loss": 0.449,
      "step": 17624
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1715195178985596,
      "learning_rate": 7.04564073409032e-06,
      "loss": 0.4759,
      "step": 17625
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.333432912826538,
      "learning_rate": 7.0445740195025355e-06,
      "loss": 0.5474,
      "step": 17626
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.7457228899002075,
      "learning_rate": 7.04350734176116e-06,
      "loss": 0.4769,
      "step": 17627
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4285672903060913,
      "learning_rate": 7.042440700879485e-06,
      "loss": 0.5421,
      "step": 17628
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.624189853668213,
      "learning_rate": 7.041374096870814e-06,
      "loss": 0.4248,
      "step": 17629
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5495493412017822,
      "learning_rate": 7.0403075297484415e-06,
      "loss": 0.4807,
      "step": 17630
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.438593864440918,
      "learning_rate": 7.039240999525665e-06,
      "loss": 0.5629,
      "step": 17631
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3839638233184814,
      "learning_rate": 7.0381745062157845e-06,
      "loss": 0.5428,
      "step": 17632
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5472654104232788,
      "learning_rate": 7.037108049832092e-06,
      "loss": 0.509,
      "step": 17633
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5639705657958984,
      "learning_rate": 7.036041630387887e-06,
      "loss": 0.5005,
      "step": 17634
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.602747917175293,
      "learning_rate": 7.0349752478964625e-06,
      "loss": 0.4581,
      "step": 17635
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3378233909606934,
      "learning_rate": 7.033908902371114e-06,
      "loss": 0.5454,
      "step": 17636
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.0246686935424805,
      "learning_rate": 7.0328425938251385e-06,
      "loss": 0.6006,
      "step": 17637
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.537358283996582,
      "learning_rate": 7.031776322271826e-06,
      "loss": 0.5968,
      "step": 17638
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2436041831970215,
      "learning_rate": 7.0307100877244725e-06,
      "loss": 0.5386,
      "step": 17639
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.8465378284454346,
      "learning_rate": 7.029643890196373e-06,
      "loss": 0.448,
      "step": 17640
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.6654927730560303,
      "learning_rate": 7.0285777297008155e-06,
      "loss": 0.4738,
      "step": 17641
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3735806941986084,
      "learning_rate": 7.027511606251096e-06,
      "loss": 0.4277,
      "step": 17642
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4714076519012451,
      "learning_rate": 7.026445519860504e-06,
      "loss": 0.4839,
      "step": 17643
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3699896335601807,
      "learning_rate": 7.0253794705423314e-06,
      "loss": 0.4836,
      "step": 17644
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4409103393554688,
      "learning_rate": 7.024313458309873e-06,
      "loss": 0.3964,
      "step": 17645
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6151543855667114,
      "learning_rate": 7.023247483176412e-06,
      "loss": 0.4993,
      "step": 17646
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5099495649337769,
      "learning_rate": 7.022181545155243e-06,
      "loss": 0.4519,
      "step": 17647
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6056036949157715,
      "learning_rate": 7.021115644259654e-06,
      "loss": 0.4917,
      "step": 17648
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.853005290031433,
      "learning_rate": 7.020049780502933e-06,
      "loss": 0.4952,
      "step": 17649
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.081782341003418,
      "learning_rate": 7.018983953898373e-06,
      "loss": 0.6024,
      "step": 17650
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.43402099609375,
      "learning_rate": 7.017918164459257e-06,
      "loss": 0.5275,
      "step": 17651
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5129449367523193,
      "learning_rate": 7.0168524121988734e-06,
      "loss": 0.4709,
      "step": 17652
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3688430786132812,
      "learning_rate": 7.0157866971305135e-06,
      "loss": 0.6251,
      "step": 17653
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7195215225219727,
      "learning_rate": 7.014721019267458e-06,
      "loss": 0.5879,
      "step": 17654
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3636436462402344,
      "learning_rate": 7.013655378623e-06,
      "loss": 0.4017,
      "step": 17655
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4924813508987427,
      "learning_rate": 7.012589775210418e-06,
      "loss": 0.5918,
      "step": 17656
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5867863893508911,
      "learning_rate": 7.011524209043e-06,
      "loss": 0.5502,
      "step": 17657
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4956378936767578,
      "learning_rate": 7.010458680134034e-06,
      "loss": 0.4183,
      "step": 17658
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.0856058597564697,
      "learning_rate": 7.0093931884968005e-06,
      "loss": 0.5221,
      "step": 17659
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5685114860534668,
      "learning_rate": 7.008327734144585e-06,
      "loss": 0.5521,
      "step": 17660
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.706228494644165,
      "learning_rate": 7.007262317090668e-06,
      "loss": 0.4071,
      "step": 17661
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.623849630355835,
      "learning_rate": 7.006196937348337e-06,
      "loss": 0.5201,
      "step": 17662
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7440459728240967,
      "learning_rate": 7.005131594930874e-06,
      "loss": 0.5174,
      "step": 17663
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.591803789138794,
      "learning_rate": 7.0040662898515565e-06,
      "loss": 0.5506,
      "step": 17664
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5402910709381104,
      "learning_rate": 7.00300102212367e-06,
      "loss": 0.5348,
      "step": 17665
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5883172750473022,
      "learning_rate": 7.001935791760497e-06,
      "loss": 0.524,
      "step": 17666
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5582607984542847,
      "learning_rate": 7.0008705987753134e-06,
      "loss": 0.5147,
      "step": 17667
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3665491342544556,
      "learning_rate": 6.999805443181402e-06,
      "loss": 0.506,
      "step": 17668
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5323342084884644,
      "learning_rate": 6.998740324992043e-06,
      "loss": 0.5706,
      "step": 17669
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6481431722640991,
      "learning_rate": 6.9976752442205135e-06,
      "loss": 0.5459,
      "step": 17670
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7136459350585938,
      "learning_rate": 6.996610200880095e-06,
      "loss": 0.4844,
      "step": 17671
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7267954349517822,
      "learning_rate": 6.995545194984064e-06,
      "loss": 0.4875,
      "step": 17672
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7906309366226196,
      "learning_rate": 6.994480226545699e-06,
      "loss": 0.5075,
      "step": 17673
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4600563049316406,
      "learning_rate": 6.993415295578276e-06,
      "loss": 0.5197,
      "step": 17674
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7334966659545898,
      "learning_rate": 6.992350402095073e-06,
      "loss": 0.4534,
      "step": 17675
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5164772272109985,
      "learning_rate": 6.991285546109367e-06,
      "loss": 0.6203,
      "step": 17676
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4626067876815796,
      "learning_rate": 6.9902207276344335e-06,
      "loss": 0.5555,
      "step": 17677
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7415108680725098,
      "learning_rate": 6.989155946683548e-06,
      "loss": 0.561,
      "step": 17678
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4537017345428467,
      "learning_rate": 6.988091203269985e-06,
      "loss": 0.5309,
      "step": 17679
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4187077283859253,
      "learning_rate": 6.987026497407019e-06,
      "loss": 0.5073,
      "step": 17680
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5773881673812866,
      "learning_rate": 6.9859618291079264e-06,
      "loss": 0.4682,
      "step": 17681
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5360080003738403,
      "learning_rate": 6.9848971983859785e-06,
      "loss": 0.4512,
      "step": 17682
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.488680362701416,
      "learning_rate": 6.983832605254446e-06,
      "loss": 0.4434,
      "step": 17683
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6987920999526978,
      "learning_rate": 6.9827680497266085e-06,
      "loss": 0.6016,
      "step": 17684
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5101642608642578,
      "learning_rate": 6.9817035318157315e-06,
      "loss": 0.5324,
      "step": 17685
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.30704665184021,
      "learning_rate": 6.980639051535092e-06,
      "loss": 0.4863,
      "step": 17686
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6327565908432007,
      "learning_rate": 6.979574608897955e-06,
      "loss": 0.6026,
      "step": 17687
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6067306995391846,
      "learning_rate": 6.978510203917597e-06,
      "loss": 0.5076,
      "step": 17688
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3609338998794556,
      "learning_rate": 6.977445836607287e-06,
      "loss": 0.5436,
      "step": 17689
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5258897542953491,
      "learning_rate": 6.9763815069802935e-06,
      "loss": 0.4476,
      "step": 17690
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6014151573181152,
      "learning_rate": 6.975317215049886e-06,
      "loss": 0.5433,
      "step": 17691
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8267053365707397,
      "learning_rate": 6.974252960829335e-06,
      "loss": 0.5128,
      "step": 17692
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5553513765335083,
      "learning_rate": 6.973188744331907e-06,
      "loss": 0.5218,
      "step": 17693
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4774975776672363,
      "learning_rate": 6.972124565570874e-06,
      "loss": 0.5709,
      "step": 17694
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.9029145240783691,
      "learning_rate": 6.971060424559497e-06,
      "loss": 0.5381,
      "step": 17695
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4701502323150635,
      "learning_rate": 6.969996321311048e-06,
      "loss": 0.4751,
      "step": 17696
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8221515417099,
      "learning_rate": 6.9689322558387914e-06,
      "loss": 0.5797,
      "step": 17697
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8054529428482056,
      "learning_rate": 6.967868228155994e-06,
      "loss": 0.4584,
      "step": 17698
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6781947612762451,
      "learning_rate": 6.966804238275924e-06,
      "loss": 0.5573,
      "step": 17699
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3561980724334717,
      "learning_rate": 6.965740286211842e-06,
      "loss": 0.531,
      "step": 17700
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5615109205245972,
      "learning_rate": 6.964676371977013e-06,
      "loss": 0.609,
      "step": 17701
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4099260568618774,
      "learning_rate": 6.963612495584706e-06,
      "loss": 0.5722,
      "step": 17702
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.316428542137146,
      "learning_rate": 6.9625486570481805e-06,
      "loss": 0.4281,
      "step": 17703
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2333502769470215,
      "learning_rate": 6.9614848563807e-06,
      "loss": 0.4997,
      "step": 17704
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1703580617904663,
      "learning_rate": 6.960421093595531e-06,
      "loss": 0.449,
      "step": 17705
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3580124378204346,
      "learning_rate": 6.95935736870593e-06,
      "loss": 0.4774,
      "step": 17706
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.693696141242981,
      "learning_rate": 6.958293681725165e-06,
      "loss": 0.3981,
      "step": 17707
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.604716181755066,
      "learning_rate": 6.957230032666491e-06,
      "loss": 0.6101,
      "step": 17708
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3483829498291016,
      "learning_rate": 6.956166421543173e-06,
      "loss": 0.5304,
      "step": 17709
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5475636720657349,
      "learning_rate": 6.955102848368474e-06,
      "loss": 0.5286,
      "step": 17710
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5360770225524902,
      "learning_rate": 6.954039313155646e-06,
      "loss": 0.5869,
      "step": 17711
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.9675251245498657,
      "learning_rate": 6.952975815917957e-06,
      "loss": 0.5318,
      "step": 17712
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5340696573257446,
      "learning_rate": 6.9519123566686595e-06,
      "loss": 0.61,
      "step": 17713
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.9204788208007812,
      "learning_rate": 6.950848935421015e-06,
      "loss": 0.6491,
      "step": 17714
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5878260135650635,
      "learning_rate": 6.949785552188283e-06,
      "loss": 0.5138,
      "step": 17715
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.332891583442688,
      "learning_rate": 6.948722206983718e-06,
      "loss": 0.6001,
      "step": 17716
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5391287803649902,
      "learning_rate": 6.94765889982058e-06,
      "loss": 0.5067,
      "step": 17717
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4377853870391846,
      "learning_rate": 6.946595630712121e-06,
      "loss": 0.548,
      "step": 17718
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6577602624893188,
      "learning_rate": 6.9455323996716016e-06,
      "loss": 0.4596,
      "step": 17719
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7287794351577759,
      "learning_rate": 6.944469206712279e-06,
      "loss": 0.6367,
      "step": 17720
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.0141804218292236,
      "learning_rate": 6.943406051847403e-06,
      "loss": 0.536,
      "step": 17721
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.468910813331604,
      "learning_rate": 6.942342935090229e-06,
      "loss": 0.4383,
      "step": 17722
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.9962782859802246,
      "learning_rate": 6.941279856454019e-06,
      "loss": 0.5395,
      "step": 17723
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.8663252592086792,
      "learning_rate": 6.9402168159520165e-06,
      "loss": 0.5928,
      "step": 17724
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5965485572814941,
      "learning_rate": 6.939153813597481e-06,
      "loss": 0.4209,
      "step": 17725
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5274039506912231,
      "learning_rate": 6.9380908494036636e-06,
      "loss": 0.5796,
      "step": 17726
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6837207078933716,
      "learning_rate": 6.9370279233838165e-06,
      "loss": 0.5106,
      "step": 17727
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2064518928527832,
      "learning_rate": 6.935965035551193e-06,
      "loss": 0.5223,
      "step": 17728
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5592674016952515,
      "learning_rate": 6.934902185919041e-06,
      "loss": 0.4947,
      "step": 17729
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2855994701385498,
      "learning_rate": 6.933839374500617e-06,
      "loss": 0.4343,
      "step": 17730
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.500294804573059,
      "learning_rate": 6.932776601309167e-06,
      "loss": 0.5348,
      "step": 17731
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7462971210479736,
      "learning_rate": 6.931713866357941e-06,
      "loss": 0.5609,
      "step": 17732
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5946846008300781,
      "learning_rate": 6.930651169660193e-06,
      "loss": 0.4523,
      "step": 17733
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5820716619491577,
      "learning_rate": 6.9295885112291665e-06,
      "loss": 0.5103,
      "step": 17734
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5226229429244995,
      "learning_rate": 6.9285258910781115e-06,
      "loss": 0.485,
      "step": 17735
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4042361974716187,
      "learning_rate": 6.92746330922028e-06,
      "loss": 0.5514,
      "step": 17736
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.400696039199829,
      "learning_rate": 6.926400765668915e-06,
      "loss": 0.5313,
      "step": 17737
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.6097701787948608,
      "learning_rate": 6.925338260437266e-06,
      "loss": 0.4689,
      "step": 17738
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.489614725112915,
      "learning_rate": 6.924275793538577e-06,
      "loss": 0.5637,
      "step": 17739
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.321904182434082,
      "learning_rate": 6.923213364986098e-06,
      "loss": 0.4956,
      "step": 17740
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.1233363151550293,
      "learning_rate": 6.922150974793073e-06,
      "loss": 0.6856,
      "step": 17741
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5950218439102173,
      "learning_rate": 6.9210886229727455e-06,
      "loss": 0.4888,
      "step": 17742
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5409220457077026,
      "learning_rate": 6.920026309538364e-06,
      "loss": 0.5389,
      "step": 17743
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3361377716064453,
      "learning_rate": 6.918964034503168e-06,
      "loss": 0.4454,
      "step": 17744
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3615307807922363,
      "learning_rate": 6.917901797880404e-06,
      "loss": 0.4547,
      "step": 17745
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7417229413986206,
      "learning_rate": 6.916839599683318e-06,
      "loss": 0.5727,
      "step": 17746
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5038726329803467,
      "learning_rate": 6.915777439925148e-06,
      "loss": 0.5623,
      "step": 17747
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.304211974143982,
      "learning_rate": 6.914715318619136e-06,
      "loss": 0.5106,
      "step": 17748
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.9326798915863037,
      "learning_rate": 6.91365323577853e-06,
      "loss": 0.53,
      "step": 17749
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6036629676818848,
      "learning_rate": 6.912591191416567e-06,
      "loss": 0.5735,
      "step": 17750
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.331040859222412,
      "learning_rate": 6.911529185546488e-06,
      "loss": 0.5126,
      "step": 17751
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5724271535873413,
      "learning_rate": 6.910467218181532e-06,
      "loss": 0.5082,
      "step": 17752
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.746508002281189,
      "learning_rate": 6.909405289334941e-06,
      "loss": 0.5377,
      "step": 17753
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.755042552947998,
      "learning_rate": 6.908343399019957e-06,
      "loss": 0.6214,
      "step": 17754
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5004873275756836,
      "learning_rate": 6.907281547249814e-06,
      "loss": 0.4654,
      "step": 17755
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8492259979248047,
      "learning_rate": 6.906219734037755e-06,
      "loss": 0.6379,
      "step": 17756
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4916552305221558,
      "learning_rate": 6.905157959397013e-06,
      "loss": 0.5137,
      "step": 17757
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5059349536895752,
      "learning_rate": 6.904096223340828e-06,
      "loss": 0.5215,
      "step": 17758
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6690412759780884,
      "learning_rate": 6.90303452588244e-06,
      "loss": 0.5362,
      "step": 17759
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.559099793434143,
      "learning_rate": 6.901972867035082e-06,
      "loss": 0.5555,
      "step": 17760
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4790955781936646,
      "learning_rate": 6.90091124681199e-06,
      "loss": 0.4613,
      "step": 17761
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5520896911621094,
      "learning_rate": 6.899849665226402e-06,
      "loss": 0.5534,
      "step": 17762
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6020557880401611,
      "learning_rate": 6.898788122291551e-06,
      "loss": 0.4871,
      "step": 17763
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8204514980316162,
      "learning_rate": 6.897726618020675e-06,
      "loss": 0.5009,
      "step": 17764
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.440083384513855,
      "learning_rate": 6.896665152427002e-06,
      "loss": 0.5685,
      "step": 17765
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6688507795333862,
      "learning_rate": 6.89560372552377e-06,
      "loss": 0.5273,
      "step": 17766
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6451916694641113,
      "learning_rate": 6.894542337324214e-06,
      "loss": 0.6049,
      "step": 17767
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3700333833694458,
      "learning_rate": 6.893480987841562e-06,
      "loss": 0.5553,
      "step": 17768
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6733829975128174,
      "learning_rate": 6.892419677089051e-06,
      "loss": 0.4086,
      "step": 17769
    },
    {
      "epoch": 1.84,
      "grad_norm": 2.012996196746826,
      "learning_rate": 6.891358405079906e-06,
      "loss": 0.5065,
      "step": 17770
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6175000667572021,
      "learning_rate": 6.890297171827365e-06,
      "loss": 0.5278,
      "step": 17771
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4578269720077515,
      "learning_rate": 6.889235977344657e-06,
      "loss": 0.5396,
      "step": 17772
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6404603719711304,
      "learning_rate": 6.888174821645011e-06,
      "loss": 0.6472,
      "step": 17773
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6815358400344849,
      "learning_rate": 6.887113704741657e-06,
      "loss": 0.4626,
      "step": 17774
    },
    {
      "epoch": 1.84,
      "grad_norm": 2.0746779441833496,
      "learning_rate": 6.886052626647827e-06,
      "loss": 0.5564,
      "step": 17775
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6461421251296997,
      "learning_rate": 6.884991587376744e-06,
      "loss": 0.6005,
      "step": 17776
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.9434622526168823,
      "learning_rate": 6.883930586941644e-06,
      "loss": 0.5093,
      "step": 17777
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5895919799804688,
      "learning_rate": 6.882869625355748e-06,
      "loss": 0.5903,
      "step": 17778
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7294126749038696,
      "learning_rate": 6.881808702632287e-06,
      "loss": 0.508,
      "step": 17779
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6080608367919922,
      "learning_rate": 6.8807478187844885e-06,
      "loss": 0.5008,
      "step": 17780
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5648468732833862,
      "learning_rate": 6.879686973825576e-06,
      "loss": 0.5437,
      "step": 17781
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5362920761108398,
      "learning_rate": 6.878626167768779e-06,
      "loss": 0.5636,
      "step": 17782
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4877386093139648,
      "learning_rate": 6.877565400627319e-06,
      "loss": 0.5515,
      "step": 17783
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8304591178894043,
      "learning_rate": 6.8765046724144246e-06,
      "loss": 0.633,
      "step": 17784
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5565052032470703,
      "learning_rate": 6.87544398314332e-06,
      "loss": 0.5507,
      "step": 17785
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7282482385635376,
      "learning_rate": 6.874383332827226e-06,
      "loss": 0.4677,
      "step": 17786
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6860227584838867,
      "learning_rate": 6.873322721479368e-06,
      "loss": 0.6097,
      "step": 17787
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3547353744506836,
      "learning_rate": 6.8722621491129705e-06,
      "loss": 0.4763,
      "step": 17788
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4707720279693604,
      "learning_rate": 6.871201615741253e-06,
      "loss": 0.6137,
      "step": 17789
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.458551049232483,
      "learning_rate": 6.870141121377441e-06,
      "loss": 0.423,
      "step": 17790
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5816736221313477,
      "learning_rate": 6.869080666034753e-06,
      "loss": 0.5203,
      "step": 17791
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6588503122329712,
      "learning_rate": 6.868020249726412e-06,
      "loss": 0.5138,
      "step": 17792
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6884158849716187,
      "learning_rate": 6.866959872465639e-06,
      "loss": 0.5624,
      "step": 17793
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.61923086643219,
      "learning_rate": 6.865899534265652e-06,
      "loss": 0.5433,
      "step": 17794
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7194215059280396,
      "learning_rate": 6.864839235139674e-06,
      "loss": 0.4831,
      "step": 17795
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.414524793624878,
      "learning_rate": 6.86377897510092e-06,
      "loss": 0.5082,
      "step": 17796
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3746187686920166,
      "learning_rate": 6.86271875416261e-06,
      "loss": 0.4459,
      "step": 17797
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.860978126525879,
      "learning_rate": 6.861658572337966e-06,
      "loss": 0.4279,
      "step": 17798
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2928087711334229,
      "learning_rate": 6.8605984296401995e-06,
      "loss": 0.4591,
      "step": 17799
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5946162939071655,
      "learning_rate": 6.85953832608253e-06,
      "loss": 0.6092,
      "step": 17800
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6335194110870361,
      "learning_rate": 6.858478261678179e-06,
      "loss": 0.4131,
      "step": 17801
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5561801195144653,
      "learning_rate": 6.857418236440354e-06,
      "loss": 0.5506,
      "step": 17802
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7149144411087036,
      "learning_rate": 6.856358250382279e-06,
      "loss": 0.5636,
      "step": 17803
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8088228702545166,
      "learning_rate": 6.855298303517164e-06,
      "loss": 0.6254,
      "step": 17804
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6981536149978638,
      "learning_rate": 6.854238395858224e-06,
      "loss": 0.5387,
      "step": 17805
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4310600757598877,
      "learning_rate": 6.853178527418678e-06,
      "loss": 0.4802,
      "step": 17806
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4992477893829346,
      "learning_rate": 6.8521186982117335e-06,
      "loss": 0.4588,
      "step": 17807
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.399964690208435,
      "learning_rate": 6.85105890825061e-06,
      "loss": 0.5193,
      "step": 17808
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5535911321640015,
      "learning_rate": 6.849999157548512e-06,
      "loss": 0.4463,
      "step": 17809
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.9137275218963623,
      "learning_rate": 6.84893944611866e-06,
      "loss": 0.4413,
      "step": 17810
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.948064923286438,
      "learning_rate": 6.847879773974263e-06,
      "loss": 0.6406,
      "step": 17811
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7267858982086182,
      "learning_rate": 6.846820141128531e-06,
      "loss": 0.523,
      "step": 17812
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.533947467803955,
      "learning_rate": 6.8457605475946755e-06,
      "loss": 0.4625,
      "step": 17813
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6970010995864868,
      "learning_rate": 6.844700993385909e-06,
      "loss": 0.6169,
      "step": 17814
    },
    {
      "epoch": 1.84,
      "grad_norm": 2.210942506790161,
      "learning_rate": 6.843641478515437e-06,
      "loss": 0.609,
      "step": 17815
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7787768840789795,
      "learning_rate": 6.842582002996474e-06,
      "loss": 0.5569,
      "step": 17816
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5689200162887573,
      "learning_rate": 6.841522566842224e-06,
      "loss": 0.5014,
      "step": 17817
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5649034976959229,
      "learning_rate": 6.840463170065897e-06,
      "loss": 0.5493,
      "step": 17818
    },
    {
      "epoch": 1.84,
      "grad_norm": 2.01597261428833,
      "learning_rate": 6.839403812680705e-06,
      "loss": 0.4803,
      "step": 17819
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4632847309112549,
      "learning_rate": 6.838344494699849e-06,
      "loss": 0.5944,
      "step": 17820
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.469102382659912,
      "learning_rate": 6.83728521613654e-06,
      "loss": 0.4553,
      "step": 17821
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7371002435684204,
      "learning_rate": 6.836225977003983e-06,
      "loss": 0.5776,
      "step": 17822
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5726332664489746,
      "learning_rate": 6.835166777315383e-06,
      "loss": 0.5334,
      "step": 17823
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4333138465881348,
      "learning_rate": 6.834107617083949e-06,
      "loss": 0.5311,
      "step": 17824
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7905129194259644,
      "learning_rate": 6.833048496322882e-06,
      "loss": 0.5294,
      "step": 17825
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6790142059326172,
      "learning_rate": 6.831989415045388e-06,
      "loss": 0.4997,
      "step": 17826
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.9489147663116455,
      "learning_rate": 6.830930373264673e-06,
      "loss": 0.6056,
      "step": 17827
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6931476593017578,
      "learning_rate": 6.829871370993937e-06,
      "loss": 0.5384,
      "step": 17828
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.645395278930664,
      "learning_rate": 6.828812408246386e-06,
      "loss": 0.5174,
      "step": 17829
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3561125993728638,
      "learning_rate": 6.827753485035219e-06,
      "loss": 0.4668,
      "step": 17830
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5752242803573608,
      "learning_rate": 6.82669460137364e-06,
      "loss": 0.4886,
      "step": 17831
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5396521091461182,
      "learning_rate": 6.825635757274852e-06,
      "loss": 0.626,
      "step": 17832
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3696730136871338,
      "learning_rate": 6.8245769527520535e-06,
      "loss": 0.4648,
      "step": 17833
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7246402502059937,
      "learning_rate": 6.823518187818446e-06,
      "loss": 0.5657,
      "step": 17834
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.8009872436523438,
      "learning_rate": 6.8224594624872295e-06,
      "loss": 0.6033,
      "step": 17835
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.41391921043396,
      "learning_rate": 6.821400776771603e-06,
      "loss": 0.5192,
      "step": 17836
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.497444748878479,
      "learning_rate": 6.8203421306847676e-06,
      "loss": 0.5284,
      "step": 17837
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.558343529701233,
      "learning_rate": 6.819283524239919e-06,
      "loss": 0.5314,
      "step": 17838
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6617563962936401,
      "learning_rate": 6.818224957450256e-06,
      "loss": 0.4657,
      "step": 17839
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6291812658309937,
      "learning_rate": 6.817166430328978e-06,
      "loss": 0.6263,
      "step": 17840
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.381223201751709,
      "learning_rate": 6.81610794288928e-06,
      "loss": 0.5706,
      "step": 17841
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.638520359992981,
      "learning_rate": 6.815049495144361e-06,
      "loss": 0.5292,
      "step": 17842
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4846514463424683,
      "learning_rate": 6.813991087107412e-06,
      "loss": 0.5387,
      "step": 17843
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.548271894454956,
      "learning_rate": 6.8129327187916335e-06,
      "loss": 0.4825,
      "step": 17844
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7701791524887085,
      "learning_rate": 6.811874390210221e-06,
      "loss": 0.4574,
      "step": 17845
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.557843565940857,
      "learning_rate": 6.8108161013763655e-06,
      "loss": 0.477,
      "step": 17846
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7030326128005981,
      "learning_rate": 6.809757852303264e-06,
      "loss": 0.5516,
      "step": 17847
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.68565833568573,
      "learning_rate": 6.808699643004107e-06,
      "loss": 0.6921,
      "step": 17848
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3470443487167358,
      "learning_rate": 6.807641473492088e-06,
      "loss": 0.4281,
      "step": 17849
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7803773880004883,
      "learning_rate": 6.806583343780404e-06,
      "loss": 0.4903,
      "step": 17850
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6475266218185425,
      "learning_rate": 6.805525253882243e-06,
      "loss": 0.5127,
      "step": 17851
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.377282977104187,
      "learning_rate": 6.804467203810796e-06,
      "loss": 0.5188,
      "step": 17852
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4408342838287354,
      "learning_rate": 6.803409193579259e-06,
      "loss": 0.5152,
      "step": 17853
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2242276668548584,
      "learning_rate": 6.802351223200816e-06,
      "loss": 0.4842,
      "step": 17854
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9876458644866943,
      "learning_rate": 6.801293292688664e-06,
      "loss": 0.5184,
      "step": 17855
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7317758798599243,
      "learning_rate": 6.800235402055987e-06,
      "loss": 0.4976,
      "step": 17856
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5111652612686157,
      "learning_rate": 6.799177551315975e-06,
      "loss": 0.566,
      "step": 17857
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8407483100891113,
      "learning_rate": 6.798119740481821e-06,
      "loss": 0.5162,
      "step": 17858
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7104291915893555,
      "learning_rate": 6.7970619695667075e-06,
      "loss": 0.5142,
      "step": 17859
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.944705843925476,
      "learning_rate": 6.796004238583827e-06,
      "loss": 0.6225,
      "step": 17860
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.844652771949768,
      "learning_rate": 6.794946547546362e-06,
      "loss": 0.4948,
      "step": 17861
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7885828018188477,
      "learning_rate": 6.793888896467501e-06,
      "loss": 0.4585,
      "step": 17862
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4778311252593994,
      "learning_rate": 6.792831285360433e-06,
      "loss": 0.5169,
      "step": 17863
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7628077268600464,
      "learning_rate": 6.791773714238339e-06,
      "loss": 0.5192,
      "step": 17864
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4364560842514038,
      "learning_rate": 6.790716183114405e-06,
      "loss": 0.4612,
      "step": 17865
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4832552671432495,
      "learning_rate": 6.789658692001821e-06,
      "loss": 0.4511,
      "step": 17866
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.483392596244812,
      "learning_rate": 6.788601240913764e-06,
      "loss": 0.4892,
      "step": 17867
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.638267159461975,
      "learning_rate": 6.787543829863423e-06,
      "loss": 0.5422,
      "step": 17868
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.749042272567749,
      "learning_rate": 6.786486458863977e-06,
      "loss": 0.5231,
      "step": 17869
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3585578203201294,
      "learning_rate": 6.78542912792861e-06,
      "loss": 0.59,
      "step": 17870
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4981658458709717,
      "learning_rate": 6.784371837070507e-06,
      "loss": 0.5379,
      "step": 17871
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7305561304092407,
      "learning_rate": 6.783314586302845e-06,
      "loss": 0.4734,
      "step": 17872
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9801273345947266,
      "learning_rate": 6.782257375638809e-06,
      "loss": 0.5525,
      "step": 17873
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3115649223327637,
      "learning_rate": 6.781200205091576e-06,
      "loss": 0.4734,
      "step": 17874
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5924859046936035,
      "learning_rate": 6.780143074674328e-06,
      "loss": 0.492,
      "step": 17875
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5595557689666748,
      "learning_rate": 6.7790859844002485e-06,
      "loss": 0.4237,
      "step": 17876
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3288823366165161,
      "learning_rate": 6.77802893428251e-06,
      "loss": 0.5363,
      "step": 17877
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6735094785690308,
      "learning_rate": 6.776971924334293e-06,
      "loss": 0.4241,
      "step": 17878
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6273378133773804,
      "learning_rate": 6.7759149545687805e-06,
      "loss": 0.5525,
      "step": 17879
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3812177181243896,
      "learning_rate": 6.774858024999143e-06,
      "loss": 0.5354,
      "step": 17880
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4978532791137695,
      "learning_rate": 6.7738011356385636e-06,
      "loss": 0.5876,
      "step": 17881
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4446122646331787,
      "learning_rate": 6.772744286500215e-06,
      "loss": 0.4992,
      "step": 17882
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.448068618774414,
      "learning_rate": 6.771687477597273e-06,
      "loss": 0.5954,
      "step": 17883
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3152681589126587,
      "learning_rate": 6.770630708942918e-06,
      "loss": 0.459,
      "step": 17884
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4652960300445557,
      "learning_rate": 6.76957398055032e-06,
      "loss": 0.5207,
      "step": 17885
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3952429294586182,
      "learning_rate": 6.768517292432658e-06,
      "loss": 0.4295,
      "step": 17886
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6508013010025024,
      "learning_rate": 6.767460644603101e-06,
      "loss": 0.4463,
      "step": 17887
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4431066513061523,
      "learning_rate": 6.766404037074826e-06,
      "loss": 0.5034,
      "step": 17888
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.1190195083618164,
      "learning_rate": 6.765347469861008e-06,
      "loss": 0.5555,
      "step": 17889
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3816368579864502,
      "learning_rate": 6.764290942974814e-06,
      "loss": 0.4237,
      "step": 17890
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9385420083999634,
      "learning_rate": 6.763234456429421e-06,
      "loss": 0.4384,
      "step": 17891
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.217179536819458,
      "learning_rate": 6.762178010237998e-06,
      "loss": 0.4128,
      "step": 17892
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4131258726119995,
      "learning_rate": 6.761121604413718e-06,
      "loss": 0.5394,
      "step": 17893
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3173539638519287,
      "learning_rate": 6.760065238969752e-06,
      "loss": 0.4406,
      "step": 17894
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6951507329940796,
      "learning_rate": 6.759008913919266e-06,
      "loss": 0.452,
      "step": 17895
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.799347162246704,
      "learning_rate": 6.757952629275433e-06,
      "loss": 0.5623,
      "step": 17896
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.490025281906128,
      "learning_rate": 6.756896385051422e-06,
      "loss": 0.4869,
      "step": 17897
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8513731956481934,
      "learning_rate": 6.7558401812604005e-06,
      "loss": 0.596,
      "step": 17898
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5029304027557373,
      "learning_rate": 6.7547840179155385e-06,
      "loss": 0.5027,
      "step": 17899
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5526809692382812,
      "learning_rate": 6.75372789503e-06,
      "loss": 0.5374,
      "step": 17900
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.434542179107666,
      "learning_rate": 6.7526718126169555e-06,
      "loss": 0.5851,
      "step": 17901
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4109333753585815,
      "learning_rate": 6.7516157706895705e-06,
      "loss": 0.4145,
      "step": 17902
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.477337121963501,
      "learning_rate": 6.75055976926101e-06,
      "loss": 0.4919,
      "step": 17903
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9044562578201294,
      "learning_rate": 6.749503808344441e-06,
      "loss": 0.5541,
      "step": 17904
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.626989483833313,
      "learning_rate": 6.748447887953029e-06,
      "loss": 0.4822,
      "step": 17905
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3971915245056152,
      "learning_rate": 6.747392008099935e-06,
      "loss": 0.4327,
      "step": 17906
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8034007549285889,
      "learning_rate": 6.746336168798329e-06,
      "loss": 0.5241,
      "step": 17907
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5996249914169312,
      "learning_rate": 6.7452803700613686e-06,
      "loss": 0.4975,
      "step": 17908
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6057865619659424,
      "learning_rate": 6.7442246119022216e-06,
      "loss": 0.5318,
      "step": 17909
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4452742338180542,
      "learning_rate": 6.743168894334048e-06,
      "loss": 0.5263,
      "step": 17910
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5052248239517212,
      "learning_rate": 6.742113217370008e-06,
      "loss": 0.5431,
      "step": 17911
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9885220527648926,
      "learning_rate": 6.741057581023268e-06,
      "loss": 0.4746,
      "step": 17912
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.018500804901123,
      "learning_rate": 6.740001985306986e-06,
      "loss": 0.5997,
      "step": 17913
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7280064821243286,
      "learning_rate": 6.73894643023432e-06,
      "loss": 0.5375,
      "step": 17914
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9733624458312988,
      "learning_rate": 6.737890915818436e-06,
      "loss": 0.6709,
      "step": 17915
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4205162525177002,
      "learning_rate": 6.736835442072488e-06,
      "loss": 0.5055,
      "step": 17916
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6123311519622803,
      "learning_rate": 6.735780009009638e-06,
      "loss": 0.4573,
      "step": 17917
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3642799854278564,
      "learning_rate": 6.734724616643045e-06,
      "loss": 0.5586,
      "step": 17918
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6849170923233032,
      "learning_rate": 6.733669264985865e-06,
      "loss": 0.5117,
      "step": 17919
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8035824298858643,
      "learning_rate": 6.732613954051257e-06,
      "loss": 0.5636,
      "step": 17920
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5033555030822754,
      "learning_rate": 6.731558683852376e-06,
      "loss": 0.5437,
      "step": 17921
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.658599615097046,
      "learning_rate": 6.73050345440238e-06,
      "loss": 0.4882,
      "step": 17922
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5203806161880493,
      "learning_rate": 6.729448265714425e-06,
      "loss": 0.5199,
      "step": 17923
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.8948174715042114,
      "learning_rate": 6.728393117801666e-06,
      "loss": 0.51,
      "step": 17924
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5914605855941772,
      "learning_rate": 6.727338010677258e-06,
      "loss": 0.5062,
      "step": 17925
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.45012629032135,
      "learning_rate": 6.726282944354354e-06,
      "loss": 0.4714,
      "step": 17926
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9396015405654907,
      "learning_rate": 6.725227918846111e-06,
      "loss": 0.5071,
      "step": 17927
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7110013961791992,
      "learning_rate": 6.724172934165681e-06,
      "loss": 0.3717,
      "step": 17928
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.743390440940857,
      "learning_rate": 6.723117990326219e-06,
      "loss": 0.4359,
      "step": 17929
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6936776638031006,
      "learning_rate": 6.722063087340871e-06,
      "loss": 0.5347,
      "step": 17930
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6647456884384155,
      "learning_rate": 6.721008225222792e-06,
      "loss": 0.5564,
      "step": 17931
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.59043550491333,
      "learning_rate": 6.719953403985138e-06,
      "loss": 0.6347,
      "step": 17932
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.9237819910049438,
      "learning_rate": 6.718898623641052e-06,
      "loss": 0.4795,
      "step": 17933
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4076006412506104,
      "learning_rate": 6.717843884203691e-06,
      "loss": 0.4976,
      "step": 17934
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.7001413106918335,
      "learning_rate": 6.716789185686201e-06,
      "loss": 0.5092,
      "step": 17935
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.438856840133667,
      "learning_rate": 6.715734528101732e-06,
      "loss": 0.5224,
      "step": 17936
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3879072666168213,
      "learning_rate": 6.714679911463435e-06,
      "loss": 0.3697,
      "step": 17937
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3328242301940918,
      "learning_rate": 6.713625335784455e-06,
      "loss": 0.4527,
      "step": 17938
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5186851024627686,
      "learning_rate": 6.712570801077941e-06,
      "loss": 0.4781,
      "step": 17939
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.318814992904663,
      "learning_rate": 6.711516307357043e-06,
      "loss": 0.5193,
      "step": 17940
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.515446662902832,
      "learning_rate": 6.710461854634904e-06,
      "loss": 0.5424,
      "step": 17941
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5070134401321411,
      "learning_rate": 6.709407442924674e-06,
      "loss": 0.5046,
      "step": 17942
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5957356691360474,
      "learning_rate": 6.708353072239493e-06,
      "loss": 0.6065,
      "step": 17943
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5982948541641235,
      "learning_rate": 6.70729874259251e-06,
      "loss": 0.4919,
      "step": 17944
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.62679123878479,
      "learning_rate": 6.706244453996873e-06,
      "loss": 0.4012,
      "step": 17945
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7537133693695068,
      "learning_rate": 6.705190206465719e-06,
      "loss": 0.485,
      "step": 17946
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3732237815856934,
      "learning_rate": 6.704136000012198e-06,
      "loss": 0.3603,
      "step": 17947
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.865558385848999,
      "learning_rate": 6.703081834649448e-06,
      "loss": 0.588,
      "step": 17948
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5191619396209717,
      "learning_rate": 6.7020277103906155e-06,
      "loss": 0.5106,
      "step": 17949
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6399556398391724,
      "learning_rate": 6.700973627248842e-06,
      "loss": 0.5239,
      "step": 17950
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6230486631393433,
      "learning_rate": 6.699919585237267e-06,
      "loss": 0.5987,
      "step": 17951
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.365464210510254,
      "learning_rate": 6.6988655843690345e-06,
      "loss": 0.4406,
      "step": 17952
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7066750526428223,
      "learning_rate": 6.697811624657284e-06,
      "loss": 0.5745,
      "step": 17953
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6758602857589722,
      "learning_rate": 6.696757706115154e-06,
      "loss": 0.5234,
      "step": 17954
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7138649225234985,
      "learning_rate": 6.695703828755787e-06,
      "loss": 0.5513,
      "step": 17955
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7682737112045288,
      "learning_rate": 6.6946499925923195e-06,
      "loss": 0.4558,
      "step": 17956
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.602230429649353,
      "learning_rate": 6.693596197637891e-06,
      "loss": 0.4784,
      "step": 17957
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.589866280555725,
      "learning_rate": 6.692542443905641e-06,
      "loss": 0.5062,
      "step": 17958
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5406720638275146,
      "learning_rate": 6.691488731408704e-06,
      "loss": 0.4819,
      "step": 17959
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3838502168655396,
      "learning_rate": 6.690435060160222e-06,
      "loss": 0.5431,
      "step": 17960
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7925835847854614,
      "learning_rate": 6.6893814301733254e-06,
      "loss": 0.547,
      "step": 17961
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6047853231430054,
      "learning_rate": 6.688327841461153e-06,
      "loss": 0.5846,
      "step": 17962
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.419124722480774,
      "learning_rate": 6.687274294036843e-06,
      "loss": 0.4296,
      "step": 17963
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5881545543670654,
      "learning_rate": 6.686220787913525e-06,
      "loss": 0.5268,
      "step": 17964
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.670598030090332,
      "learning_rate": 6.685167323104336e-06,
      "loss": 0.566,
      "step": 17965
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4019627571105957,
      "learning_rate": 6.684113899622413e-06,
      "loss": 0.4618,
      "step": 17966
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8722354173660278,
      "learning_rate": 6.683060517480884e-06,
      "loss": 0.5845,
      "step": 17967
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5828551054000854,
      "learning_rate": 6.682007176692886e-06,
      "loss": 0.6424,
      "step": 17968
    },
    {
      "epoch": 1.86,
      "grad_norm": 2.052157402038574,
      "learning_rate": 6.68095387727155e-06,
      "loss": 0.7464,
      "step": 17969
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3745356798171997,
      "learning_rate": 6.679900619230005e-06,
      "loss": 0.4806,
      "step": 17970
    },
    {
      "epoch": 1.86,
      "grad_norm": 2.001201629638672,
      "learning_rate": 6.67884740258139e-06,
      "loss": 0.5539,
      "step": 17971
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.441185712814331,
      "learning_rate": 6.677794227338826e-06,
      "loss": 0.4765,
      "step": 17972
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5118895769119263,
      "learning_rate": 6.676741093515453e-06,
      "loss": 0.5865,
      "step": 17973
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7657217979431152,
      "learning_rate": 6.675688001124392e-06,
      "loss": 0.5706,
      "step": 17974
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6383662223815918,
      "learning_rate": 6.674634950178778e-06,
      "loss": 0.5666,
      "step": 17975
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6328115463256836,
      "learning_rate": 6.673581940691739e-06,
      "loss": 0.5109,
      "step": 17976
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8348277807235718,
      "learning_rate": 6.672528972676402e-06,
      "loss": 0.5858,
      "step": 17977
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6723634004592896,
      "learning_rate": 6.671476046145895e-06,
      "loss": 0.5894,
      "step": 17978
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5762923955917358,
      "learning_rate": 6.670423161113345e-06,
      "loss": 0.572,
      "step": 17979
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7787202596664429,
      "learning_rate": 6.6693703175918785e-06,
      "loss": 0.5751,
      "step": 17980
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5805913209915161,
      "learning_rate": 6.668317515594624e-06,
      "loss": 0.5229,
      "step": 17981
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6561315059661865,
      "learning_rate": 6.667264755134706e-06,
      "loss": 0.5226,
      "step": 17982
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7485865354537964,
      "learning_rate": 6.666212036225247e-06,
      "loss": 0.4849,
      "step": 17983
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7815476655960083,
      "learning_rate": 6.665159358879376e-06,
      "loss": 0.4786,
      "step": 17984
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6781672239303589,
      "learning_rate": 6.664106723110213e-06,
      "loss": 0.5687,
      "step": 17985
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.527112364768982,
      "learning_rate": 6.663054128930886e-06,
      "loss": 0.5218,
      "step": 17986
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2900704145431519,
      "learning_rate": 6.6620015763545135e-06,
      "loss": 0.4467,
      "step": 17987
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4296579360961914,
      "learning_rate": 6.66094906539422e-06,
      "loss": 0.5599,
      "step": 17988
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8117433786392212,
      "learning_rate": 6.65989659606313e-06,
      "loss": 0.4928,
      "step": 17989
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6096583604812622,
      "learning_rate": 6.658844168374361e-06,
      "loss": 0.6308,
      "step": 17990
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5927507877349854,
      "learning_rate": 6.6577917823410376e-06,
      "loss": 0.5636,
      "step": 17991
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5656863451004028,
      "learning_rate": 6.6567394379762766e-06,
      "loss": 0.4473,
      "step": 17992
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5087517499923706,
      "learning_rate": 6.6556871352932e-06,
      "loss": 0.4699,
      "step": 17993
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2986301183700562,
      "learning_rate": 6.654634874304928e-06,
      "loss": 0.5011,
      "step": 17994
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7969168424606323,
      "learning_rate": 6.653582655024577e-06,
      "loss": 0.589,
      "step": 17995
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4838333129882812,
      "learning_rate": 6.652530477465269e-06,
      "loss": 0.5823,
      "step": 17996
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6179630756378174,
      "learning_rate": 6.65147834164012e-06,
      "loss": 0.496,
      "step": 17997
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4880684614181519,
      "learning_rate": 6.650426247562245e-06,
      "loss": 0.5644,
      "step": 17998
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3971377611160278,
      "learning_rate": 6.649374195244766e-06,
      "loss": 0.5006,
      "step": 17999
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.410212755203247,
      "learning_rate": 6.648322184700795e-06,
      "loss": 0.4941,
      "step": 18000
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6317205429077148,
      "learning_rate": 6.647270215943449e-06,
      "loss": 0.5472,
      "step": 18001
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4057080745697021,
      "learning_rate": 6.646218288985846e-06,
      "loss": 0.5229,
      "step": 18002
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.256557822227478,
      "learning_rate": 6.645166403841096e-06,
      "loss": 0.4208,
      "step": 18003
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4299805164337158,
      "learning_rate": 6.644114560522317e-06,
      "loss": 0.5959,
      "step": 18004
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5153839588165283,
      "learning_rate": 6.64306275904262e-06,
      "loss": 0.5073,
      "step": 18005
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.374397873878479,
      "learning_rate": 6.6420109994151206e-06,
      "loss": 0.4393,
      "step": 18006
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6290031671524048,
      "learning_rate": 6.64095928165293e-06,
      "loss": 0.5431,
      "step": 18007
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4786272048950195,
      "learning_rate": 6.63990760576916e-06,
      "loss": 0.4102,
      "step": 18008
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4505795240402222,
      "learning_rate": 6.638855971776923e-06,
      "loss": 0.4202,
      "step": 18009
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3329116106033325,
      "learning_rate": 6.637804379689333e-06,
      "loss": 0.5193,
      "step": 18010
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.9416707754135132,
      "learning_rate": 6.636752829519495e-06,
      "loss": 0.6233,
      "step": 18011
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.536415696144104,
      "learning_rate": 6.6357013212805245e-06,
      "loss": 0.4493,
      "step": 18012
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.472428321838379,
      "learning_rate": 6.634649854985525e-06,
      "loss": 0.5123,
      "step": 18013
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6985132694244385,
      "learning_rate": 6.63359843064761e-06,
      "loss": 0.5427,
      "step": 18014
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.698939323425293,
      "learning_rate": 6.632547048279889e-06,
      "loss": 0.4216,
      "step": 18015
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5036602020263672,
      "learning_rate": 6.631495707895465e-06,
      "loss": 0.5258,
      "step": 18016
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8421443700790405,
      "learning_rate": 6.63044440950745e-06,
      "loss": 0.5509,
      "step": 18017
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.299862027168274,
      "learning_rate": 6.629393153128949e-06,
      "loss": 0.5109,
      "step": 18018
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4470593929290771,
      "learning_rate": 6.6283419387730675e-06,
      "loss": 0.4824,
      "step": 18019
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.812028408050537,
      "learning_rate": 6.627290766452915e-06,
      "loss": 0.4883,
      "step": 18020
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5293608903884888,
      "learning_rate": 6.626239636181593e-06,
      "loss": 0.442,
      "step": 18021
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6396310329437256,
      "learning_rate": 6.6251885479722055e-06,
      "loss": 0.3374,
      "step": 18022
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5491520166397095,
      "learning_rate": 6.624137501837863e-06,
      "loss": 0.4964,
      "step": 18023
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4439995288848877,
      "learning_rate": 6.623086497791663e-06,
      "loss": 0.3914,
      "step": 18024
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8649479150772095,
      "learning_rate": 6.622035535846714e-06,
      "loss": 0.4404,
      "step": 18025
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5945390462875366,
      "learning_rate": 6.620984616016112e-06,
      "loss": 0.6123,
      "step": 18026
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.7152572870254517,
      "learning_rate": 6.6199337383129636e-06,
      "loss": 0.5242,
      "step": 18027
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3311901092529297,
      "learning_rate": 6.618882902750372e-06,
      "loss": 0.5315,
      "step": 18028
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6416383981704712,
      "learning_rate": 6.617832109341435e-06,
      "loss": 0.4356,
      "step": 18029
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5845574140548706,
      "learning_rate": 6.6167813580992556e-06,
      "loss": 0.5211,
      "step": 18030
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3623186349868774,
      "learning_rate": 6.615730649036931e-06,
      "loss": 0.5656,
      "step": 18031
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.8294767141342163,
      "learning_rate": 6.614679982167563e-06,
      "loss": 0.56,
      "step": 18032
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6958390474319458,
      "learning_rate": 6.6136293575042524e-06,
      "loss": 0.4906,
      "step": 18033
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4757074117660522,
      "learning_rate": 6.612578775060094e-06,
      "loss": 0.4115,
      "step": 18034
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.860568642616272,
      "learning_rate": 6.6115282348481865e-06,
      "loss": 0.579,
      "step": 18035
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8004237413406372,
      "learning_rate": 6.610477736881631e-06,
      "loss": 0.6095,
      "step": 18036
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.507239580154419,
      "learning_rate": 6.60942728117352e-06,
      "loss": 0.4353,
      "step": 18037
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4477726221084595,
      "learning_rate": 6.608376867736952e-06,
      "loss": 0.4552,
      "step": 18038
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8078774213790894,
      "learning_rate": 6.607326496585023e-06,
      "loss": 0.5067,
      "step": 18039
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5267019271850586,
      "learning_rate": 6.606276167730826e-06,
      "loss": 0.5087,
      "step": 18040
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.538577675819397,
      "learning_rate": 6.60522588118746e-06,
      "loss": 0.4729,
      "step": 18041
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5276968479156494,
      "learning_rate": 6.604175636968015e-06,
      "loss": 0.411,
      "step": 18042
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5090895891189575,
      "learning_rate": 6.603125435085591e-06,
      "loss": 0.5145,
      "step": 18043
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6789219379425049,
      "learning_rate": 6.6020752755532735e-06,
      "loss": 0.5698,
      "step": 18044
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7101534605026245,
      "learning_rate": 6.601025158384159e-06,
      "loss": 0.4452,
      "step": 18045
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6812138557434082,
      "learning_rate": 6.5999750835913415e-06,
      "loss": 0.4111,
      "step": 18046
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6129659414291382,
      "learning_rate": 6.59892505118791e-06,
      "loss": 0.5519,
      "step": 18047
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.679173469543457,
      "learning_rate": 6.597875061186955e-06,
      "loss": 0.5303,
      "step": 18048
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.9219926595687866,
      "learning_rate": 6.5968251136015705e-06,
      "loss": 0.5302,
      "step": 18049
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.548999547958374,
      "learning_rate": 6.595775208444843e-06,
      "loss": 0.4184,
      "step": 18050
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.579777479171753,
      "learning_rate": 6.594725345729867e-06,
      "loss": 0.4976,
      "step": 18051
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.33252215385437,
      "learning_rate": 6.593675525469725e-06,
      "loss": 0.4881,
      "step": 18052
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.1823697090148926,
      "learning_rate": 6.592625747677508e-06,
      "loss": 0.5362,
      "step": 18053
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5441792011260986,
      "learning_rate": 6.591576012366307e-06,
      "loss": 0.531,
      "step": 18054
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4992964267730713,
      "learning_rate": 6.590526319549206e-06,
      "loss": 0.5042,
      "step": 18055
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4680900573730469,
      "learning_rate": 6.589476669239296e-06,
      "loss": 0.5957,
      "step": 18056
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6833653450012207,
      "learning_rate": 6.588427061449657e-06,
      "loss": 0.4939,
      "step": 18057
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6521073579788208,
      "learning_rate": 6.587377496193379e-06,
      "loss": 0.4896,
      "step": 18058
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5303949117660522,
      "learning_rate": 6.586327973483549e-06,
      "loss": 0.545,
      "step": 18059
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.1029601097106934,
      "learning_rate": 6.585278493333245e-06,
      "loss": 0.508,
      "step": 18060
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5923136472702026,
      "learning_rate": 6.584229055755558e-06,
      "loss": 0.5107,
      "step": 18061
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6737102270126343,
      "learning_rate": 6.58317966076357e-06,
      "loss": 0.4822,
      "step": 18062
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.342826008796692,
      "learning_rate": 6.582130308370362e-06,
      "loss": 0.4782,
      "step": 18063
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6121550798416138,
      "learning_rate": 6.581080998589022e-06,
      "loss": 0.5567,
      "step": 18064
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7393815517425537,
      "learning_rate": 6.580031731432624e-06,
      "loss": 0.4504,
      "step": 18065
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5998501777648926,
      "learning_rate": 6.578982506914255e-06,
      "loss": 0.4941,
      "step": 18066
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7511019706726074,
      "learning_rate": 6.577933325046998e-06,
      "loss": 0.4523,
      "step": 18067
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.962764024734497,
      "learning_rate": 6.576884185843928e-06,
      "loss": 0.4707,
      "step": 18068
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6367992162704468,
      "learning_rate": 6.57583508931813e-06,
      "loss": 0.4547,
      "step": 18069
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.1294965744018555,
      "learning_rate": 6.574786035482679e-06,
      "loss": 0.5972,
      "step": 18070
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8767684698104858,
      "learning_rate": 6.573737024350655e-06,
      "loss": 0.4285,
      "step": 18071
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7781850099563599,
      "learning_rate": 6.5726880559351415e-06,
      "loss": 0.713,
      "step": 18072
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5757994651794434,
      "learning_rate": 6.5716391302492074e-06,
      "loss": 0.5726,
      "step": 18073
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.9866551160812378,
      "learning_rate": 6.570590247305938e-06,
      "loss": 0.4711,
      "step": 18074
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3267674446105957,
      "learning_rate": 6.569541407118408e-06,
      "loss": 0.4734,
      "step": 18075
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6262502670288086,
      "learning_rate": 6.56849260969969e-06,
      "loss": 0.4581,
      "step": 18076
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3407344818115234,
      "learning_rate": 6.567443855062867e-06,
      "loss": 0.5516,
      "step": 18077
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3057897090911865,
      "learning_rate": 6.566395143221006e-06,
      "loss": 0.4402,
      "step": 18078
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.910568356513977,
      "learning_rate": 6.565346474187186e-06,
      "loss": 0.5707,
      "step": 18079
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7260828018188477,
      "learning_rate": 6.564297847974482e-06,
      "loss": 0.599,
      "step": 18080
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3924140930175781,
      "learning_rate": 6.563249264595966e-06,
      "loss": 0.4537,
      "step": 18081
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4642913341522217,
      "learning_rate": 6.562200724064711e-06,
      "loss": 0.4444,
      "step": 18082
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.47362220287323,
      "learning_rate": 6.56115222639379e-06,
      "loss": 0.5275,
      "step": 18083
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7904236316680908,
      "learning_rate": 6.560103771596273e-06,
      "loss": 0.4535,
      "step": 18084
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8581174612045288,
      "learning_rate": 6.559055359685237e-06,
      "loss": 0.4473,
      "step": 18085
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5860658884048462,
      "learning_rate": 6.558006990673747e-06,
      "loss": 0.4135,
      "step": 18086
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.671023964881897,
      "learning_rate": 6.556958664574875e-06,
      "loss": 0.552,
      "step": 18087
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.463106393814087,
      "learning_rate": 6.555910381401695e-06,
      "loss": 0.522,
      "step": 18088
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.532028317451477,
      "learning_rate": 6.554862141167269e-06,
      "loss": 0.4564,
      "step": 18089
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.0545175075531006,
      "learning_rate": 6.553813943884673e-06,
      "loss": 0.4455,
      "step": 18090
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.586168646812439,
      "learning_rate": 6.5527657895669705e-06,
      "loss": 0.5022,
      "step": 18091
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.41256582736969,
      "learning_rate": 6.5517176782272296e-06,
      "loss": 0.509,
      "step": 18092
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8541767597198486,
      "learning_rate": 6.550669609878521e-06,
      "loss": 0.5515,
      "step": 18093
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4882177114486694,
      "learning_rate": 6.549621584533907e-06,
      "loss": 0.5132,
      "step": 18094
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6105984449386597,
      "learning_rate": 6.548573602206458e-06,
      "loss": 0.4757,
      "step": 18095
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4680730104446411,
      "learning_rate": 6.547525662909235e-06,
      "loss": 0.6828,
      "step": 18096
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.281693935394287,
      "learning_rate": 6.546477766655307e-06,
      "loss": 0.5038,
      "step": 18097
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7654792070388794,
      "learning_rate": 6.545429913457737e-06,
      "loss": 0.4896,
      "step": 18098
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3423895835876465,
      "learning_rate": 6.5443821033295885e-06,
      "loss": 0.5045,
      "step": 18099
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.461510419845581,
      "learning_rate": 6.543334336283924e-06,
      "loss": 0.4596,
      "step": 18100
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5430809259414673,
      "learning_rate": 6.54228661233381e-06,
      "loss": 0.5683,
      "step": 18101
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5649288892745972,
      "learning_rate": 6.541238931492305e-06,
      "loss": 0.507,
      "step": 18102
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5680433511734009,
      "learning_rate": 6.540191293772475e-06,
      "loss": 0.4328,
      "step": 18103
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.875060796737671,
      "learning_rate": 6.5391436991873756e-06,
      "loss": 0.5894,
      "step": 18104
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.8098628520965576,
      "learning_rate": 6.53809614775007e-06,
      "loss": 0.5535,
      "step": 18105
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.629239797592163,
      "learning_rate": 6.537048639473622e-06,
      "loss": 0.4992,
      "step": 18106
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7430474758148193,
      "learning_rate": 6.536001174371087e-06,
      "loss": 0.4728,
      "step": 18107
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7479585409164429,
      "learning_rate": 6.534953752455528e-06,
      "loss": 0.5475,
      "step": 18108
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6941601037979126,
      "learning_rate": 6.533906373739998e-06,
      "loss": 0.5865,
      "step": 18109
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.731427550315857,
      "learning_rate": 6.532859038237558e-06,
      "loss": 0.5231,
      "step": 18110
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4045159816741943,
      "learning_rate": 6.5318117459612685e-06,
      "loss": 0.4939,
      "step": 18111
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7295242547988892,
      "learning_rate": 6.53076449692418e-06,
      "loss": 0.5149,
      "step": 18112
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5946452617645264,
      "learning_rate": 6.529717291139354e-06,
      "loss": 0.4919,
      "step": 18113
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.437224268913269,
      "learning_rate": 6.528670128619848e-06,
      "loss": 0.4559,
      "step": 18114
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6858329772949219,
      "learning_rate": 6.527623009378712e-06,
      "loss": 0.4464,
      "step": 18115
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6573456525802612,
      "learning_rate": 6.5265759334290046e-06,
      "loss": 0.4932,
      "step": 18116
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.835825800895691,
      "learning_rate": 6.525528900783776e-06,
      "loss": 0.5758,
      "step": 18117
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6966303586959839,
      "learning_rate": 6.524481911456084e-06,
      "loss": 0.5756,
      "step": 18118
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.264052391052246,
      "learning_rate": 6.5234349654589815e-06,
      "loss": 0.5251,
      "step": 18119
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5634609460830688,
      "learning_rate": 6.522388062805518e-06,
      "loss": 0.5389,
      "step": 18120
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6851937770843506,
      "learning_rate": 6.521341203508751e-06,
      "loss": 0.526,
      "step": 18121
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7354488372802734,
      "learning_rate": 6.520294387581727e-06,
      "loss": 0.5575,
      "step": 18122
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.449479579925537,
      "learning_rate": 6.519247615037498e-06,
      "loss": 0.5699,
      "step": 18123
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6493468284606934,
      "learning_rate": 6.518200885889118e-06,
      "loss": 0.4648,
      "step": 18124
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.495078444480896,
      "learning_rate": 6.517154200149632e-06,
      "loss": 0.5219,
      "step": 18125
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7501575946807861,
      "learning_rate": 6.516107557832092e-06,
      "loss": 0.4692,
      "step": 18126
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.9379597902297974,
      "learning_rate": 6.515060958949548e-06,
      "loss": 0.6784,
      "step": 18127
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5113807916641235,
      "learning_rate": 6.514014403515046e-06,
      "loss": 0.5047,
      "step": 18128
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.569612979888916,
      "learning_rate": 6.512967891541638e-06,
      "loss": 0.559,
      "step": 18129
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5243803262710571,
      "learning_rate": 6.511921423042363e-06,
      "loss": 0.5499,
      "step": 18130
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3813916444778442,
      "learning_rate": 6.510874998030275e-06,
      "loss": 0.5276,
      "step": 18131
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8231018781661987,
      "learning_rate": 6.5098286165184184e-06,
      "loss": 0.3922,
      "step": 18132
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6201587915420532,
      "learning_rate": 6.508782278519838e-06,
      "loss": 0.5608,
      "step": 18133
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4960992336273193,
      "learning_rate": 6.507735984047581e-06,
      "loss": 0.5817,
      "step": 18134
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7033978700637817,
      "learning_rate": 6.506689733114688e-06,
      "loss": 0.5262,
      "step": 18135
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4069870710372925,
      "learning_rate": 6.505643525734205e-06,
      "loss": 0.5492,
      "step": 18136
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4624024629592896,
      "learning_rate": 6.5045973619191786e-06,
      "loss": 0.411,
      "step": 18137
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4326372146606445,
      "learning_rate": 6.503551241682646e-06,
      "loss": 0.4377,
      "step": 18138
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7685033082962036,
      "learning_rate": 6.502505165037652e-06,
      "loss": 0.4195,
      "step": 18139
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4933561086654663,
      "learning_rate": 6.501459131997242e-06,
      "loss": 0.5305,
      "step": 18140
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3678686618804932,
      "learning_rate": 6.500413142574452e-06,
      "loss": 0.4095,
      "step": 18141
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6276319026947021,
      "learning_rate": 6.499367196782327e-06,
      "loss": 0.5858,
      "step": 18142
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6027913093566895,
      "learning_rate": 6.498321294633902e-06,
      "loss": 0.4863,
      "step": 18143
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7201107740402222,
      "learning_rate": 6.497275436142221e-06,
      "loss": 0.4487,
      "step": 18144
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8506920337677002,
      "learning_rate": 6.496229621320323e-06,
      "loss": 0.7023,
      "step": 18145
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.736898422241211,
      "learning_rate": 6.495183850181244e-06,
      "loss": 0.5264,
      "step": 18146
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4939417839050293,
      "learning_rate": 6.494138122738024e-06,
      "loss": 0.4313,
      "step": 18147
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7851719856262207,
      "learning_rate": 6.493092439003699e-06,
      "loss": 0.6083,
      "step": 18148
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5713838338851929,
      "learning_rate": 6.492046798991307e-06,
      "loss": 0.5736,
      "step": 18149
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6997461318969727,
      "learning_rate": 6.491001202713885e-06,
      "loss": 0.5328,
      "step": 18150
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3426976203918457,
      "learning_rate": 6.489955650184467e-06,
      "loss": 0.4384,
      "step": 18151
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6825830936431885,
      "learning_rate": 6.488910141416091e-06,
      "loss": 0.5186,
      "step": 18152
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7474586963653564,
      "learning_rate": 6.487864676421789e-06,
      "loss": 0.5185,
      "step": 18153
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5623966455459595,
      "learning_rate": 6.486819255214596e-06,
      "loss": 0.5858,
      "step": 18154
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8181893825531006,
      "learning_rate": 6.485773877807548e-06,
      "loss": 0.5234,
      "step": 18155
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6357884407043457,
      "learning_rate": 6.484728544213674e-06,
      "loss": 0.4907,
      "step": 18156
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4607585668563843,
      "learning_rate": 6.483683254446009e-06,
      "loss": 0.5528,
      "step": 18157
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6572104692459106,
      "learning_rate": 6.482638008517586e-06,
      "loss": 0.5743,
      "step": 18158
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7576824426651,
      "learning_rate": 6.481592806441432e-06,
      "loss": 0.481,
      "step": 18159
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5191062688827515,
      "learning_rate": 6.480547648230586e-06,
      "loss": 0.5897,
      "step": 18160
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.490473985671997,
      "learning_rate": 6.47950253389807e-06,
      "loss": 0.5083,
      "step": 18161
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.672804594039917,
      "learning_rate": 6.478457463456916e-06,
      "loss": 0.4813,
      "step": 18162
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6578874588012695,
      "learning_rate": 6.477412436920159e-06,
      "loss": 0.5448,
      "step": 18163
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6214390993118286,
      "learning_rate": 6.476367454300819e-06,
      "loss": 0.4419,
      "step": 18164
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4590684175491333,
      "learning_rate": 6.475322515611929e-06,
      "loss": 0.464,
      "step": 18165
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5881386995315552,
      "learning_rate": 6.4742776208665195e-06,
      "loss": 0.512,
      "step": 18166
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7908635139465332,
      "learning_rate": 6.473232770077611e-06,
      "loss": 0.6031,
      "step": 18167
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4081021547317505,
      "learning_rate": 6.472187963258236e-06,
      "loss": 0.5793,
      "step": 18168
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6410388946533203,
      "learning_rate": 6.471143200421415e-06,
      "loss": 0.4549,
      "step": 18169
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5572413206100464,
      "learning_rate": 6.470098481580176e-06,
      "loss": 0.5108,
      "step": 18170
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4588303565979004,
      "learning_rate": 6.469053806747547e-06,
      "loss": 0.5588,
      "step": 18171
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6587814092636108,
      "learning_rate": 6.468009175936547e-06,
      "loss": 0.5474,
      "step": 18172
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.9108508825302124,
      "learning_rate": 6.466964589160205e-06,
      "loss": 0.4916,
      "step": 18173
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.9016430377960205,
      "learning_rate": 6.46592004643154e-06,
      "loss": 0.5695,
      "step": 18174
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8675448894500732,
      "learning_rate": 6.464875547763574e-06,
      "loss": 0.5493,
      "step": 18175
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6139681339263916,
      "learning_rate": 6.4638310931693345e-06,
      "loss": 0.5529,
      "step": 18176
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4485454559326172,
      "learning_rate": 6.462786682661839e-06,
      "loss": 0.5454,
      "step": 18177
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5332494974136353,
      "learning_rate": 6.461742316254109e-06,
      "loss": 0.4695,
      "step": 18178
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4171929359436035,
      "learning_rate": 6.460697993959167e-06,
      "loss": 0.4978,
      "step": 18179
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.928246259689331,
      "learning_rate": 6.459653715790031e-06,
      "loss": 0.4144,
      "step": 18180
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.861674427986145,
      "learning_rate": 6.458609481759721e-06,
      "loss": 0.6766,
      "step": 18181
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.951879858970642,
      "learning_rate": 6.457565291881255e-06,
      "loss": 0.4912,
      "step": 18182
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4356452226638794,
      "learning_rate": 6.456521146167652e-06,
      "loss": 0.5738,
      "step": 18183
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.025017499923706,
      "learning_rate": 6.455477044631931e-06,
      "loss": 0.5443,
      "step": 18184
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6549301147460938,
      "learning_rate": 6.454432987287107e-06,
      "loss": 0.567,
      "step": 18185
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4341068267822266,
      "learning_rate": 6.4533889741461995e-06,
      "loss": 0.4706,
      "step": 18186
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6468039751052856,
      "learning_rate": 6.4523450052222205e-06,
      "loss": 0.4621,
      "step": 18187
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4858641624450684,
      "learning_rate": 6.451301080528189e-06,
      "loss": 0.5111,
      "step": 18188
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4370993375778198,
      "learning_rate": 6.4502572000771194e-06,
      "loss": 0.4786,
      "step": 18189
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4463675022125244,
      "learning_rate": 6.449213363882024e-06,
      "loss": 0.4562,
      "step": 18190
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6137619018554688,
      "learning_rate": 6.448169571955919e-06,
      "loss": 0.4151,
      "step": 18191
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5972763299942017,
      "learning_rate": 6.447125824311819e-06,
      "loss": 0.6271,
      "step": 18192
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7527133226394653,
      "learning_rate": 6.446082120962732e-06,
      "loss": 0.5124,
      "step": 18193
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3455257415771484,
      "learning_rate": 6.445038461921676e-06,
      "loss": 0.5286,
      "step": 18194
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8676624298095703,
      "learning_rate": 6.4439948472016575e-06,
      "loss": 0.5084,
      "step": 18195
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6305512189865112,
      "learning_rate": 6.44295127681569e-06,
      "loss": 0.6742,
      "step": 18196
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5417144298553467,
      "learning_rate": 6.441907750776786e-06,
      "loss": 0.5275,
      "step": 18197
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5210093259811401,
      "learning_rate": 6.44086426909795e-06,
      "loss": 0.5964,
      "step": 18198
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.0146594047546387,
      "learning_rate": 6.439820831792201e-06,
      "loss": 0.5004,
      "step": 18199
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.9317740201950073,
      "learning_rate": 6.438777438872537e-06,
      "loss": 0.4486,
      "step": 18200
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7739112377166748,
      "learning_rate": 6.437734090351971e-06,
      "loss": 0.3419,
      "step": 18201
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2768714427947998,
      "learning_rate": 6.4366907862435135e-06,
      "loss": 0.4443,
      "step": 18202
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4694584608078003,
      "learning_rate": 6.435647526560169e-06,
      "loss": 0.5033,
      "step": 18203
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7918922901153564,
      "learning_rate": 6.434604311314942e-06,
      "loss": 0.5521,
      "step": 18204
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4331072568893433,
      "learning_rate": 6.433561140520845e-06,
      "loss": 0.5176,
      "step": 18205
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.662569284439087,
      "learning_rate": 6.432518014190878e-06,
      "loss": 0.5738,
      "step": 18206
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.096299648284912,
      "learning_rate": 6.4314749323380494e-06,
      "loss": 0.5668,
      "step": 18207
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.766864538192749,
      "learning_rate": 6.430431894975361e-06,
      "loss": 0.5623,
      "step": 18208
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4928619861602783,
      "learning_rate": 6.429388902115817e-06,
      "loss": 0.5389,
      "step": 18209
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3720242977142334,
      "learning_rate": 6.428345953772424e-06,
      "loss": 0.5146,
      "step": 18210
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.9532841444015503,
      "learning_rate": 6.427303049958181e-06,
      "loss": 0.5102,
      "step": 18211
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.693745493888855,
      "learning_rate": 6.426260190686094e-06,
      "loss": 0.6045,
      "step": 18212
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5599766969680786,
      "learning_rate": 6.42521737596916e-06,
      "loss": 0.5217,
      "step": 18213
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.546061635017395,
      "learning_rate": 6.424174605820382e-06,
      "loss": 0.4942,
      "step": 18214
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8295292854309082,
      "learning_rate": 6.423131880252764e-06,
      "loss": 0.4398,
      "step": 18215
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5467768907546997,
      "learning_rate": 6.4220891992793025e-06,
      "loss": 0.6901,
      "step": 18216
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6746132373809814,
      "learning_rate": 6.421046562912996e-06,
      "loss": 0.5399,
      "step": 18217
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5529524087905884,
      "learning_rate": 6.420003971166848e-06,
      "loss": 0.5298,
      "step": 18218
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4640005826950073,
      "learning_rate": 6.418961424053852e-06,
      "loss": 0.5809,
      "step": 18219
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.109652519226074,
      "learning_rate": 6.417918921587011e-06,
      "loss": 0.5891,
      "step": 18220
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8661110401153564,
      "learning_rate": 6.416876463779315e-06,
      "loss": 0.5039,
      "step": 18221
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5750945806503296,
      "learning_rate": 6.415834050643768e-06,
      "loss": 0.4185,
      "step": 18222
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.487568974494934,
      "learning_rate": 6.414791682193364e-06,
      "loss": 0.422,
      "step": 18223
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4074673652648926,
      "learning_rate": 6.413749358441095e-06,
      "loss": 0.4334,
      "step": 18224
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.8371952772140503,
      "learning_rate": 6.412707079399963e-06,
      "loss": 0.5565,
      "step": 18225
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6640257835388184,
      "learning_rate": 6.4116648450829536e-06,
      "loss": 0.588,
      "step": 18226
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.248720407485962,
      "learning_rate": 6.410622655503067e-06,
      "loss": 0.4863,
      "step": 18227
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5914520025253296,
      "learning_rate": 6.409580510673296e-06,
      "loss": 0.6269,
      "step": 18228
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2400391101837158,
      "learning_rate": 6.4085384106066305e-06,
      "loss": 0.4937,
      "step": 18229
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3107683658599854,
      "learning_rate": 6.407496355316065e-06,
      "loss": 0.4748,
      "step": 18230
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.397815465927124,
      "learning_rate": 6.406454344814592e-06,
      "loss": 0.5237,
      "step": 18231
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.346192717552185,
      "learning_rate": 6.405412379115201e-06,
      "loss": 0.4639,
      "step": 18232
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6670355796813965,
      "learning_rate": 6.4043704582308845e-06,
      "loss": 0.5592,
      "step": 18233
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.302411437034607,
      "learning_rate": 6.4033285821746284e-06,
      "loss": 0.4878,
      "step": 18234
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5565096139907837,
      "learning_rate": 6.402286750959426e-06,
      "loss": 0.5839,
      "step": 18235
    },
    {
      "epoch": 1.89,
      "grad_norm": 2.034907102584839,
      "learning_rate": 6.401244964598265e-06,
      "loss": 0.5552,
      "step": 18236
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.565919280052185,
      "learning_rate": 6.400203223104133e-06,
      "loss": 0.5925,
      "step": 18237
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6803711652755737,
      "learning_rate": 6.399161526490022e-06,
      "loss": 0.5723,
      "step": 18238
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6744633913040161,
      "learning_rate": 6.3981198747689125e-06,
      "loss": 0.5638,
      "step": 18239
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.489435076713562,
      "learning_rate": 6.397078267953793e-06,
      "loss": 0.5151,
      "step": 18240
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.9189343452453613,
      "learning_rate": 6.396036706057655e-06,
      "loss": 0.5115,
      "step": 18241
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5105783939361572,
      "learning_rate": 6.394995189093478e-06,
      "loss": 0.4802,
      "step": 18242
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.53994619846344,
      "learning_rate": 6.393953717074248e-06,
      "loss": 0.4557,
      "step": 18243
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6521095037460327,
      "learning_rate": 6.392912290012953e-06,
      "loss": 0.5859,
      "step": 18244
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5956015586853027,
      "learning_rate": 6.391870907922572e-06,
      "loss": 0.5076,
      "step": 18245
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.601304054260254,
      "learning_rate": 6.3908295708160936e-06,
      "loss": 0.5525,
      "step": 18246
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.590905785560608,
      "learning_rate": 6.389788278706494e-06,
      "loss": 0.5602,
      "step": 18247
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7113754749298096,
      "learning_rate": 6.388747031606761e-06,
      "loss": 0.482,
      "step": 18248
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8542574644088745,
      "learning_rate": 6.3877058295298756e-06,
      "loss": 0.5178,
      "step": 18249
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8340883255004883,
      "learning_rate": 6.386664672488815e-06,
      "loss": 0.6546,
      "step": 18250
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.49977445602417,
      "learning_rate": 6.385623560496564e-06,
      "loss": 0.5305,
      "step": 18251
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7673099040985107,
      "learning_rate": 6.384582493566099e-06,
      "loss": 0.4973,
      "step": 18252
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8866944313049316,
      "learning_rate": 6.383541471710402e-06,
      "loss": 0.5124,
      "step": 18253
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7150133848190308,
      "learning_rate": 6.382500494942451e-06,
      "loss": 0.5571,
      "step": 18254
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3796190023422241,
      "learning_rate": 6.381459563275223e-06,
      "loss": 0.5659,
      "step": 18255
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5290356874465942,
      "learning_rate": 6.380418676721696e-06,
      "loss": 0.5803,
      "step": 18256
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5767912864685059,
      "learning_rate": 6.379377835294851e-06,
      "loss": 0.4436,
      "step": 18257
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6915102005004883,
      "learning_rate": 6.3783370390076595e-06,
      "loss": 0.5895,
      "step": 18258
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.72359299659729,
      "learning_rate": 6.3772962878731e-06,
      "loss": 0.4723,
      "step": 18259
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6575448513031006,
      "learning_rate": 6.3762555819041475e-06,
      "loss": 0.4358,
      "step": 18260
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3489640951156616,
      "learning_rate": 6.375214921113775e-06,
      "loss": 0.466,
      "step": 18261
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8416818380355835,
      "learning_rate": 6.3741743055149615e-06,
      "loss": 0.5114,
      "step": 18262
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6046321392059326,
      "learning_rate": 6.373133735120676e-06,
      "loss": 0.5191,
      "step": 18263
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.31761634349823,
      "learning_rate": 6.372093209943895e-06,
      "loss": 0.4792,
      "step": 18264
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4601314067840576,
      "learning_rate": 6.371052729997589e-06,
      "loss": 0.472,
      "step": 18265
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8300901651382446,
      "learning_rate": 6.370012295294729e-06,
      "loss": 0.5567,
      "step": 18266
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.9904769659042358,
      "learning_rate": 6.36897190584829e-06,
      "loss": 0.5681,
      "step": 18267
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4770312309265137,
      "learning_rate": 6.36793156167124e-06,
      "loss": 0.4835,
      "step": 18268
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6389645338058472,
      "learning_rate": 6.366891262776551e-06,
      "loss": 0.5562,
      "step": 18269
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5964229106903076,
      "learning_rate": 6.365851009177193e-06,
      "loss": 0.4849,
      "step": 18270
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5460456609725952,
      "learning_rate": 6.3648108008861335e-06,
      "loss": 0.4151,
      "step": 18271
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.9308264255523682,
      "learning_rate": 6.363770637916343e-06,
      "loss": 0.5816,
      "step": 18272
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8458443880081177,
      "learning_rate": 6.362730520280788e-06,
      "loss": 0.5678,
      "step": 18273
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.682642936706543,
      "learning_rate": 6.3616904479924365e-06,
      "loss": 0.521,
      "step": 18274
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4012233018875122,
      "learning_rate": 6.360650421064257e-06,
      "loss": 0.515,
      "step": 18275
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3770725727081299,
      "learning_rate": 6.359610439509214e-06,
      "loss": 0.4684,
      "step": 18276
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3328415155410767,
      "learning_rate": 6.358570503340274e-06,
      "loss": 0.4596,
      "step": 18277
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7199305295944214,
      "learning_rate": 6.3575306125704015e-06,
      "loss": 0.5282,
      "step": 18278
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3895692825317383,
      "learning_rate": 6.356490767212562e-06,
      "loss": 0.4731,
      "step": 18279
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6211148500442505,
      "learning_rate": 6.3554509672797225e-06,
      "loss": 0.5182,
      "step": 18280
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.746343970298767,
      "learning_rate": 6.354411212784846e-06,
      "loss": 0.5235,
      "step": 18281
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7145240306854248,
      "learning_rate": 6.3533715037408875e-06,
      "loss": 0.6269,
      "step": 18282
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4576308727264404,
      "learning_rate": 6.352331840160815e-06,
      "loss": 0.5749,
      "step": 18283
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7223787307739258,
      "learning_rate": 6.3512922220575945e-06,
      "loss": 0.4879,
      "step": 18284
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4641159772872925,
      "learning_rate": 6.3502526494441795e-06,
      "loss": 0.5096,
      "step": 18285
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7446444034576416,
      "learning_rate": 6.349213122333537e-06,
      "loss": 0.5434,
      "step": 18286
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7210181951522827,
      "learning_rate": 6.348173640738624e-06,
      "loss": 0.5117,
      "step": 18287
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.624802589416504,
      "learning_rate": 6.347134204672401e-06,
      "loss": 0.5525,
      "step": 18288
    },
    {
      "epoch": 1.89,
      "grad_norm": 2.1177852153778076,
      "learning_rate": 6.346094814147828e-06,
      "loss": 0.6152,
      "step": 18289
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.496203899383545,
      "learning_rate": 6.345055469177859e-06,
      "loss": 0.5777,
      "step": 18290
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5788596868515015,
      "learning_rate": 6.344016169775459e-06,
      "loss": 0.3997,
      "step": 18291
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5985772609710693,
      "learning_rate": 6.342976915953578e-06,
      "loss": 0.5147,
      "step": 18292
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.393140435218811,
      "learning_rate": 6.341937707725179e-06,
      "loss": 0.6437,
      "step": 18293
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8285503387451172,
      "learning_rate": 6.340898545103216e-06,
      "loss": 0.5432,
      "step": 18294
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.581567406654358,
      "learning_rate": 6.3398594281006435e-06,
      "loss": 0.412,
      "step": 18295
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7280733585357666,
      "learning_rate": 6.338820356730415e-06,
      "loss": 0.5352,
      "step": 18296
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8601160049438477,
      "learning_rate": 6.33778133100549e-06,
      "loss": 0.5258,
      "step": 18297
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6372919082641602,
      "learning_rate": 6.336742350938818e-06,
      "loss": 0.554,
      "step": 18298
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4087432622909546,
      "learning_rate": 6.335703416543357e-06,
      "loss": 0.5363,
      "step": 18299
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.526759147644043,
      "learning_rate": 6.334664527832055e-06,
      "loss": 0.5562,
      "step": 18300
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.449252963066101,
      "learning_rate": 6.333625684817866e-06,
      "loss": 0.4691,
      "step": 18301
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5624816417694092,
      "learning_rate": 6.332586887513742e-06,
      "loss": 0.4534,
      "step": 18302
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8840196132659912,
      "learning_rate": 6.3315481359326325e-06,
      "loss": 0.5774,
      "step": 18303
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.574867606163025,
      "learning_rate": 6.330509430087493e-06,
      "loss": 0.5008,
      "step": 18304
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6861640214920044,
      "learning_rate": 6.329470769991265e-06,
      "loss": 0.5781,
      "step": 18305
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5234376192092896,
      "learning_rate": 6.328432155656904e-06,
      "loss": 0.4421,
      "step": 18306
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4118821620941162,
      "learning_rate": 6.327393587097359e-06,
      "loss": 0.4886,
      "step": 18307
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5692510604858398,
      "learning_rate": 6.3263550643255755e-06,
      "loss": 0.5143,
      "step": 18308
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8700553178787231,
      "learning_rate": 6.325316587354502e-06,
      "loss": 0.5465,
      "step": 18309
    },
    {
      "epoch": 1.89,
      "grad_norm": 2.229707956314087,
      "learning_rate": 6.3242781561970865e-06,
      "loss": 0.5131,
      "step": 18310
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6516605615615845,
      "learning_rate": 6.323239770866274e-06,
      "loss": 0.4898,
      "step": 18311
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.25836980342865,
      "learning_rate": 6.322201431375012e-06,
      "loss": 0.4363,
      "step": 18312
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5135631561279297,
      "learning_rate": 6.321163137736246e-06,
      "loss": 0.5654,
      "step": 18313
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.9176974296569824,
      "learning_rate": 6.320124889962918e-06,
      "loss": 0.6239,
      "step": 18314
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5061570405960083,
      "learning_rate": 6.319086688067977e-06,
      "loss": 0.4542,
      "step": 18315
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7455710172653198,
      "learning_rate": 6.318048532064361e-06,
      "loss": 0.4609,
      "step": 18316
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5531054735183716,
      "learning_rate": 6.317010421965017e-06,
      "loss": 0.5564,
      "step": 18317
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5213814973831177,
      "learning_rate": 6.315972357782887e-06,
      "loss": 0.5524,
      "step": 18318
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.474078893661499,
      "learning_rate": 6.3149343395309105e-06,
      "loss": 0.5427,
      "step": 18319
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4866690635681152,
      "learning_rate": 6.3138963672220335e-06,
      "loss": 0.5372,
      "step": 18320
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5310369729995728,
      "learning_rate": 6.3128584408691915e-06,
      "loss": 0.589,
      "step": 18321
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3925687074661255,
      "learning_rate": 6.311820560485328e-06,
      "loss": 0.5258,
      "step": 18322
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7984274625778198,
      "learning_rate": 6.310782726083383e-06,
      "loss": 0.5268,
      "step": 18323
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4553231000900269,
      "learning_rate": 6.309744937676292e-06,
      "loss": 0.5059,
      "step": 18324
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6186517477035522,
      "learning_rate": 6.3087071952769995e-06,
      "loss": 0.5187,
      "step": 18325
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.733834981918335,
      "learning_rate": 6.307669498898437e-06,
      "loss": 0.5135,
      "step": 18326
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5326917171478271,
      "learning_rate": 6.306631848553545e-06,
      "loss": 0.5594,
      "step": 18327
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.230280637741089,
      "learning_rate": 6.305594244255262e-06,
      "loss": 0.5549,
      "step": 18328
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6552897691726685,
      "learning_rate": 6.304556686016521e-06,
      "loss": 0.5571,
      "step": 18329
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6421735286712646,
      "learning_rate": 6.303519173850259e-06,
      "loss": 0.5732,
      "step": 18330
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.295128345489502,
      "learning_rate": 6.30248170776941e-06,
      "loss": 0.6272,
      "step": 18331
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6549055576324463,
      "learning_rate": 6.301444287786909e-06,
      "loss": 0.5716,
      "step": 18332
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.738612174987793,
      "learning_rate": 6.300406913915693e-06,
      "loss": 0.5064,
      "step": 18333
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4425110816955566,
      "learning_rate": 6.299369586168689e-06,
      "loss": 0.4291,
      "step": 18334
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5467277765274048,
      "learning_rate": 6.298332304558835e-06,
      "loss": 0.5266,
      "step": 18335
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.615444540977478,
      "learning_rate": 6.297295069099063e-06,
      "loss": 0.5313,
      "step": 18336
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6677724123001099,
      "learning_rate": 6.296257879802301e-06,
      "loss": 0.5062,
      "step": 18337
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4889498949050903,
      "learning_rate": 6.295220736681485e-06,
      "loss": 0.5235,
      "step": 18338
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4399292469024658,
      "learning_rate": 6.294183639749538e-06,
      "loss": 0.4936,
      "step": 18339
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6936614513397217,
      "learning_rate": 6.2931465890193965e-06,
      "loss": 0.4382,
      "step": 18340
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4490900039672852,
      "learning_rate": 6.29210958450399e-06,
      "loss": 0.5884,
      "step": 18341
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.709684133529663,
      "learning_rate": 6.2910726262162435e-06,
      "loss": 0.5081,
      "step": 18342
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3230152130126953,
      "learning_rate": 6.290035714169087e-06,
      "loss": 0.4806,
      "step": 18343
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4664947986602783,
      "learning_rate": 6.288998848375447e-06,
      "loss": 0.5546,
      "step": 18344
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4555772542953491,
      "learning_rate": 6.2879620288482515e-06,
      "loss": 0.6196,
      "step": 18345
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4209518432617188,
      "learning_rate": 6.286925255600427e-06,
      "loss": 0.5548,
      "step": 18346
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7348103523254395,
      "learning_rate": 6.285888528644898e-06,
      "loss": 0.5647,
      "step": 18347
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9783830642700195,
      "learning_rate": 6.28485184799459e-06,
      "loss": 0.4522,
      "step": 18348
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6218998432159424,
      "learning_rate": 6.283815213662432e-06,
      "loss": 0.5179,
      "step": 18349
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.484794020652771,
      "learning_rate": 6.282778625661341e-06,
      "loss": 0.6457,
      "step": 18350
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.646871566772461,
      "learning_rate": 6.281742084004247e-06,
      "loss": 0.4842,
      "step": 18351
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5076370239257812,
      "learning_rate": 6.280705588704068e-06,
      "loss": 0.4939,
      "step": 18352
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8741050958633423,
      "learning_rate": 6.2796691397737275e-06,
      "loss": 0.4835,
      "step": 18353
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5818049907684326,
      "learning_rate": 6.278632737226151e-06,
      "loss": 0.4982,
      "step": 18354
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.084131956100464,
      "learning_rate": 6.277596381074256e-06,
      "loss": 0.6319,
      "step": 18355
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7709276676177979,
      "learning_rate": 6.276560071330964e-06,
      "loss": 0.4541,
      "step": 18356
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.608748435974121,
      "learning_rate": 6.275523808009194e-06,
      "loss": 0.4772,
      "step": 18357
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5765389204025269,
      "learning_rate": 6.274487591121867e-06,
      "loss": 0.4525,
      "step": 18358
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9012842178344727,
      "learning_rate": 6.273451420681902e-06,
      "loss": 0.6008,
      "step": 18359
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.564379096031189,
      "learning_rate": 6.2724152967022164e-06,
      "loss": 0.452,
      "step": 18360
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8375755548477173,
      "learning_rate": 6.271379219195728e-06,
      "loss": 0.5539,
      "step": 18361
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3430875539779663,
      "learning_rate": 6.2703431881753565e-06,
      "loss": 0.526,
      "step": 18362
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8587685823440552,
      "learning_rate": 6.269307203654014e-06,
      "loss": 0.5485,
      "step": 18363
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4805817604064941,
      "learning_rate": 6.268271265644621e-06,
      "loss": 0.5084,
      "step": 18364
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7510377168655396,
      "learning_rate": 6.267235374160089e-06,
      "loss": 0.58,
      "step": 18365
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.9490272998809814,
      "learning_rate": 6.266199529213335e-06,
      "loss": 0.5427,
      "step": 18366
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5980850458145142,
      "learning_rate": 6.265163730817274e-06,
      "loss": 0.515,
      "step": 18367
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4422540664672852,
      "learning_rate": 6.264127978984818e-06,
      "loss": 0.4441,
      "step": 18368
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5176308155059814,
      "learning_rate": 6.263092273728882e-06,
      "loss": 0.4805,
      "step": 18369
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4536794424057007,
      "learning_rate": 6.2620566150623725e-06,
      "loss": 0.5219,
      "step": 18370
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6520923376083374,
      "learning_rate": 6.2610210029982084e-06,
      "loss": 0.5242,
      "step": 18371
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5055450201034546,
      "learning_rate": 6.2599854375493015e-06,
      "loss": 0.5276,
      "step": 18372
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5802725553512573,
      "learning_rate": 6.258949918728559e-06,
      "loss": 0.5926,
      "step": 18373
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2282708883285522,
      "learning_rate": 6.257914446548889e-06,
      "loss": 0.4661,
      "step": 18374
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5243370532989502,
      "learning_rate": 6.256879021023208e-06,
      "loss": 0.5431,
      "step": 18375
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5471835136413574,
      "learning_rate": 6.2558436421644184e-06,
      "loss": 0.4752,
      "step": 18376
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5267704725265503,
      "learning_rate": 6.254808309985434e-06,
      "loss": 0.5295,
      "step": 18377
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5931682586669922,
      "learning_rate": 6.253773024499158e-06,
      "loss": 0.4744,
      "step": 18378
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2797902822494507,
      "learning_rate": 6.252737785718499e-06,
      "loss": 0.5229,
      "step": 18379
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8199152946472168,
      "learning_rate": 6.251702593656368e-06,
      "loss": 0.5269,
      "step": 18380
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4384071826934814,
      "learning_rate": 6.2506674483256645e-06,
      "loss": 0.4687,
      "step": 18381
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5941253900527954,
      "learning_rate": 6.249632349739299e-06,
      "loss": 0.4906,
      "step": 18382
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7870014905929565,
      "learning_rate": 6.248597297910173e-06,
      "loss": 0.5131,
      "step": 18383
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7870527505874634,
      "learning_rate": 6.247562292851193e-06,
      "loss": 0.5875,
      "step": 18384
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4939868450164795,
      "learning_rate": 6.246527334575264e-06,
      "loss": 0.4407,
      "step": 18385
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4552732706069946,
      "learning_rate": 6.245492423095285e-06,
      "loss": 0.4927,
      "step": 18386
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.374367117881775,
      "learning_rate": 6.244457558424162e-06,
      "loss": 0.5378,
      "step": 18387
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8585922718048096,
      "learning_rate": 6.243422740574798e-06,
      "loss": 0.6066,
      "step": 18388
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6472413539886475,
      "learning_rate": 6.242387969560091e-06,
      "loss": 0.391,
      "step": 18389
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.606827735900879,
      "learning_rate": 6.241353245392945e-06,
      "loss": 0.5335,
      "step": 18390
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5676006078720093,
      "learning_rate": 6.2403185680862566e-06,
      "loss": 0.54,
      "step": 18391
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8660557270050049,
      "learning_rate": 6.239283937652928e-06,
      "loss": 0.5131,
      "step": 18392
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5194584131240845,
      "learning_rate": 6.238249354105861e-06,
      "loss": 0.4419,
      "step": 18393
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3570222854614258,
      "learning_rate": 6.237214817457948e-06,
      "loss": 0.3087,
      "step": 18394
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7005077600479126,
      "learning_rate": 6.236180327722092e-06,
      "loss": 0.6025,
      "step": 18395
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6183726787567139,
      "learning_rate": 6.235145884911187e-06,
      "loss": 0.4852,
      "step": 18396
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4567670822143555,
      "learning_rate": 6.234111489038132e-06,
      "loss": 0.5656,
      "step": 18397
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5220001935958862,
      "learning_rate": 6.233077140115824e-06,
      "loss": 0.5232,
      "step": 18398
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3400068283081055,
      "learning_rate": 6.2320428381571555e-06,
      "loss": 0.5343,
      "step": 18399
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4895119667053223,
      "learning_rate": 6.231008583175022e-06,
      "loss": 0.5123,
      "step": 18400
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8719940185546875,
      "learning_rate": 6.2299743751823225e-06,
      "loss": 0.534,
      "step": 18401
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.813209891319275,
      "learning_rate": 6.2289402141919455e-06,
      "loss": 0.492,
      "step": 18402
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8481167554855347,
      "learning_rate": 6.227906100216788e-06,
      "loss": 0.434,
      "step": 18403
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5672944784164429,
      "learning_rate": 6.226872033269738e-06,
      "loss": 0.4188,
      "step": 18404
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7514227628707886,
      "learning_rate": 6.225838013363692e-06,
      "loss": 0.6241,
      "step": 18405
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7380269765853882,
      "learning_rate": 6.2248040405115405e-06,
      "loss": 0.5346,
      "step": 18406
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5428879261016846,
      "learning_rate": 6.2237701147261735e-06,
      "loss": 0.4751,
      "step": 18407
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.581122875213623,
      "learning_rate": 6.222736236020483e-06,
      "loss": 0.5097,
      "step": 18408
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2453303337097168,
      "learning_rate": 6.221702404407355e-06,
      "loss": 0.4134,
      "step": 18409
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.1651527881622314,
      "learning_rate": 6.220668619899682e-06,
      "loss": 0.529,
      "step": 18410
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5940930843353271,
      "learning_rate": 6.219634882510355e-06,
      "loss": 0.522,
      "step": 18411
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7534860372543335,
      "learning_rate": 6.218601192252255e-06,
      "loss": 0.5502,
      "step": 18412
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3342618942260742,
      "learning_rate": 6.217567549138273e-06,
      "loss": 0.5977,
      "step": 18413
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.006378173828125,
      "learning_rate": 6.216533953181299e-06,
      "loss": 0.4852,
      "step": 18414
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7728344202041626,
      "learning_rate": 6.2155004043942144e-06,
      "loss": 0.4598,
      "step": 18415
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6547106504440308,
      "learning_rate": 6.214466902789908e-06,
      "loss": 0.4307,
      "step": 18416
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4308615922927856,
      "learning_rate": 6.213433448381263e-06,
      "loss": 0.4894,
      "step": 18417
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.7458021640777588,
      "learning_rate": 6.212400041181164e-06,
      "loss": 0.3877,
      "step": 18418
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.476081132888794,
      "learning_rate": 6.211366681202498e-06,
      "loss": 0.542,
      "step": 18419
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3726531267166138,
      "learning_rate": 6.210333368458141e-06,
      "loss": 0.5986,
      "step": 18420
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7565289735794067,
      "learning_rate": 6.209300102960985e-06,
      "loss": 0.5452,
      "step": 18421
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.765254020690918,
      "learning_rate": 6.208266884723903e-06,
      "loss": 0.5069,
      "step": 18422
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7534044981002808,
      "learning_rate": 6.207233713759781e-06,
      "loss": 0.5146,
      "step": 18423
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4645026922225952,
      "learning_rate": 6.206200590081504e-06,
      "loss": 0.4731,
      "step": 18424
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6331857442855835,
      "learning_rate": 6.205167513701945e-06,
      "loss": 0.5752,
      "step": 18425
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8150359392166138,
      "learning_rate": 6.204134484633988e-06,
      "loss": 0.4786,
      "step": 18426
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4420126676559448,
      "learning_rate": 6.203101502890512e-06,
      "loss": 0.5144,
      "step": 18427
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.323732852935791,
      "learning_rate": 6.202068568484392e-06,
      "loss": 0.5258,
      "step": 18428
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7504971027374268,
      "learning_rate": 6.201035681428512e-06,
      "loss": 0.6433,
      "step": 18429
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7398406267166138,
      "learning_rate": 6.200002841735744e-06,
      "loss": 0.5599,
      "step": 18430
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4667062759399414,
      "learning_rate": 6.198970049418967e-06,
      "loss": 0.4897,
      "step": 18431
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.1436309814453125,
      "learning_rate": 6.197937304491057e-06,
      "loss": 0.5687,
      "step": 18432
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8406838178634644,
      "learning_rate": 6.19690460696489e-06,
      "loss": 0.5058,
      "step": 18433
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3771699666976929,
      "learning_rate": 6.1958719568533406e-06,
      "loss": 0.5201,
      "step": 18434
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.715743899345398,
      "learning_rate": 6.194839354169284e-06,
      "loss": 0.5031,
      "step": 18435
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6745158433914185,
      "learning_rate": 6.193806798925591e-06,
      "loss": 0.5216,
      "step": 18436
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8286120891571045,
      "learning_rate": 6.192774291135141e-06,
      "loss": 0.5987,
      "step": 18437
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6723759174346924,
      "learning_rate": 6.1917418308108e-06,
      "loss": 0.5879,
      "step": 18438
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5115375518798828,
      "learning_rate": 6.1907094179654435e-06,
      "loss": 0.5057,
      "step": 18439
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8838995695114136,
      "learning_rate": 6.1896770526119444e-06,
      "loss": 0.5952,
      "step": 18440
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5236852169036865,
      "learning_rate": 6.188644734763168e-06,
      "loss": 0.6374,
      "step": 18441
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.460509181022644,
      "learning_rate": 6.187612464431991e-06,
      "loss": 0.4207,
      "step": 18442
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6684768199920654,
      "learning_rate": 6.186580241631279e-06,
      "loss": 0.5363,
      "step": 18443
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9393718242645264,
      "learning_rate": 6.185548066373903e-06,
      "loss": 0.5721,
      "step": 18444
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.569092869758606,
      "learning_rate": 6.184515938672733e-06,
      "loss": 0.5325,
      "step": 18445
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.41563081741333,
      "learning_rate": 6.183483858540631e-06,
      "loss": 0.4593,
      "step": 18446
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.300129771232605,
      "learning_rate": 6.182451825990472e-06,
      "loss": 0.5695,
      "step": 18447
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.972163438796997,
      "learning_rate": 6.181419841035116e-06,
      "loss": 0.5035,
      "step": 18448
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4792826175689697,
      "learning_rate": 6.1803879036874325e-06,
      "loss": 0.4716,
      "step": 18449
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.336216688156128,
      "learning_rate": 6.179356013960289e-06,
      "loss": 0.496,
      "step": 18450
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7170045375823975,
      "learning_rate": 6.1783241718665455e-06,
      "loss": 0.5318,
      "step": 18451
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3609771728515625,
      "learning_rate": 6.177292377419069e-06,
      "loss": 0.4976,
      "step": 18452
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7015070915222168,
      "learning_rate": 6.176260630630726e-06,
      "loss": 0.5153,
      "step": 18453
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8216665983200073,
      "learning_rate": 6.175228931514374e-06,
      "loss": 0.4515,
      "step": 18454
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9756438732147217,
      "learning_rate": 6.174197280082881e-06,
      "loss": 0.4937,
      "step": 18455
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6316449642181396,
      "learning_rate": 6.173165676349103e-06,
      "loss": 0.6378,
      "step": 18456
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.71989905834198,
      "learning_rate": 6.172134120325907e-06,
      "loss": 0.5157,
      "step": 18457
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3681703805923462,
      "learning_rate": 6.171102612026152e-06,
      "loss": 0.5232,
      "step": 18458
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7372390031814575,
      "learning_rate": 6.170071151462697e-06,
      "loss": 0.5816,
      "step": 18459
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4680869579315186,
      "learning_rate": 6.169039738648403e-06,
      "loss": 0.4984,
      "step": 18460
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9619834423065186,
      "learning_rate": 6.168008373596128e-06,
      "loss": 0.5204,
      "step": 18461
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4553464651107788,
      "learning_rate": 6.16697705631873e-06,
      "loss": 0.4922,
      "step": 18462
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3597301244735718,
      "learning_rate": 6.16594578682907e-06,
      "loss": 0.3881,
      "step": 18463
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7303082942962646,
      "learning_rate": 6.164914565140002e-06,
      "loss": 0.5444,
      "step": 18464
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3593591451644897,
      "learning_rate": 6.163883391264383e-06,
      "loss": 0.4663,
      "step": 18465
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5760124921798706,
      "learning_rate": 6.162852265215072e-06,
      "loss": 0.3481,
      "step": 18466
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5931692123413086,
      "learning_rate": 6.161821187004919e-06,
      "loss": 0.4543,
      "step": 18467
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.500570297241211,
      "learning_rate": 6.160790156646783e-06,
      "loss": 0.5135,
      "step": 18468
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5531574487686157,
      "learning_rate": 6.1597591741535165e-06,
      "loss": 0.5203,
      "step": 18469
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.406427025794983,
      "learning_rate": 6.158728239537974e-06,
      "loss": 0.4754,
      "step": 18470
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.405517816543579,
      "learning_rate": 6.15769735281301e-06,
      "loss": 0.4967,
      "step": 18471
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.510240077972412,
      "learning_rate": 6.156666513991473e-06,
      "loss": 0.5295,
      "step": 18472
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8260406255722046,
      "learning_rate": 6.155635723086219e-06,
      "loss": 0.5375,
      "step": 18473
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.757177472114563,
      "learning_rate": 6.154604980110095e-06,
      "loss": 0.4861,
      "step": 18474
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.831484317779541,
      "learning_rate": 6.153574285075955e-06,
      "loss": 0.4789,
      "step": 18475
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4102879762649536,
      "learning_rate": 6.15254363799665e-06,
      "loss": 0.445,
      "step": 18476
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7005003690719604,
      "learning_rate": 6.1515130388850245e-06,
      "loss": 0.4935,
      "step": 18477
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4387872219085693,
      "learning_rate": 6.150482487753932e-06,
      "loss": 0.5506,
      "step": 18478
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8495436906814575,
      "learning_rate": 6.1494519846162195e-06,
      "loss": 0.4459,
      "step": 18479
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5843138694763184,
      "learning_rate": 6.148421529484734e-06,
      "loss": 0.3839,
      "step": 18480
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6499334573745728,
      "learning_rate": 6.147391122372324e-06,
      "loss": 0.5,
      "step": 18481
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4059028625488281,
      "learning_rate": 6.146360763291832e-06,
      "loss": 0.6017,
      "step": 18482
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.590193748474121,
      "learning_rate": 6.145330452256108e-06,
      "loss": 0.4815,
      "step": 18483
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5198687314987183,
      "learning_rate": 6.144300189277998e-06,
      "loss": 0.5392,
      "step": 18484
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6978991031646729,
      "learning_rate": 6.143269974370343e-06,
      "loss": 0.4755,
      "step": 18485
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.544412612915039,
      "learning_rate": 6.14223980754599e-06,
      "loss": 0.3962,
      "step": 18486
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.723179817199707,
      "learning_rate": 6.14120968881778e-06,
      "loss": 0.4978,
      "step": 18487
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.481965184211731,
      "learning_rate": 6.140179618198556e-06,
      "loss": 0.4065,
      "step": 18488
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.8028819561004639,
      "learning_rate": 6.139149595701165e-06,
      "loss": 0.6382,
      "step": 18489
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7707117795944214,
      "learning_rate": 6.138119621338442e-06,
      "loss": 0.6104,
      "step": 18490
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6575661897659302,
      "learning_rate": 6.137089695123232e-06,
      "loss": 0.4341,
      "step": 18491
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6214373111724854,
      "learning_rate": 6.136059817068376e-06,
      "loss": 0.485,
      "step": 18492
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.531952977180481,
      "learning_rate": 6.135029987186712e-06,
      "loss": 0.4229,
      "step": 18493
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9021787643432617,
      "learning_rate": 6.13400020549108e-06,
      "loss": 0.5319,
      "step": 18494
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3934292793273926,
      "learning_rate": 6.1329704719943164e-06,
      "loss": 0.521,
      "step": 18495
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5678467750549316,
      "learning_rate": 6.131940786709261e-06,
      "loss": 0.4557,
      "step": 18496
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5841768980026245,
      "learning_rate": 6.130911149648755e-06,
      "loss": 0.5181,
      "step": 18497
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5523972511291504,
      "learning_rate": 6.1298815608256305e-06,
      "loss": 0.5739,
      "step": 18498
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7135425806045532,
      "learning_rate": 6.128852020252726e-06,
      "loss": 0.4208,
      "step": 18499
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5355181694030762,
      "learning_rate": 6.127822527942875e-06,
      "loss": 0.4431,
      "step": 18500
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.502506971359253,
      "learning_rate": 6.126793083908913e-06,
      "loss": 0.4739,
      "step": 18501
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.681086540222168,
      "learning_rate": 6.1257636881636775e-06,
      "loss": 0.5487,
      "step": 18502
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5794222354888916,
      "learning_rate": 6.124734340719999e-06,
      "loss": 0.5294,
      "step": 18503
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7037012577056885,
      "learning_rate": 6.123705041590711e-06,
      "loss": 0.5437,
      "step": 18504
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5361775159835815,
      "learning_rate": 6.122675790788651e-06,
      "loss": 0.5296,
      "step": 18505
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.339150071144104,
      "learning_rate": 6.121646588326643e-06,
      "loss": 0.3858,
      "step": 18506
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7870441675186157,
      "learning_rate": 6.120617434217526e-06,
      "loss": 0.4697,
      "step": 18507
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6705344915390015,
      "learning_rate": 6.119588328474124e-06,
      "loss": 0.5999,
      "step": 18508
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6740630865097046,
      "learning_rate": 6.118559271109273e-06,
      "loss": 0.4512,
      "step": 18509
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6454788446426392,
      "learning_rate": 6.1175302621358016e-06,
      "loss": 0.4394,
      "step": 18510
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.9625060558319092,
      "learning_rate": 6.116501301566536e-06,
      "loss": 0.5487,
      "step": 18511
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6892483234405518,
      "learning_rate": 6.1154723894143065e-06,
      "loss": 0.5716,
      "step": 18512
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5623829364776611,
      "learning_rate": 6.11444352569194e-06,
      "loss": 0.4687,
      "step": 18513
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.7306973934173584,
      "learning_rate": 6.1134147104122645e-06,
      "loss": 0.5482,
      "step": 18514
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5931941270828247,
      "learning_rate": 6.112385943588108e-06,
      "loss": 0.5327,
      "step": 18515
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.6491738557815552,
      "learning_rate": 6.1113572252322935e-06,
      "loss": 0.4511,
      "step": 18516
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5347447395324707,
      "learning_rate": 6.110328555357649e-06,
      "loss": 0.591,
      "step": 18517
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6818926334381104,
      "learning_rate": 6.109299933976999e-06,
      "loss": 0.5398,
      "step": 18518
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7473204135894775,
      "learning_rate": 6.1082713611031654e-06,
      "loss": 0.5596,
      "step": 18519
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4196418523788452,
      "learning_rate": 6.107242836748976e-06,
      "loss": 0.571,
      "step": 18520
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8126299381256104,
      "learning_rate": 6.106214360927248e-06,
      "loss": 0.4885,
      "step": 18521
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3325068950653076,
      "learning_rate": 6.105185933650808e-06,
      "loss": 0.4741,
      "step": 18522
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6779839992523193,
      "learning_rate": 6.104157554932479e-06,
      "loss": 0.6235,
      "step": 18523
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6812798976898193,
      "learning_rate": 6.103129224785077e-06,
      "loss": 0.5386,
      "step": 18524
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9234614372253418,
      "learning_rate": 6.102100943221428e-06,
      "loss": 0.4388,
      "step": 18525
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.418999433517456,
      "learning_rate": 6.101072710254347e-06,
      "loss": 0.5249,
      "step": 18526
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7286776304244995,
      "learning_rate": 6.100044525896658e-06,
      "loss": 0.5056,
      "step": 18527
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.530179500579834,
      "learning_rate": 6.0990163901611784e-06,
      "loss": 0.4491,
      "step": 18528
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3350236415863037,
      "learning_rate": 6.097988303060723e-06,
      "loss": 0.6026,
      "step": 18529
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7906721830368042,
      "learning_rate": 6.096960264608114e-06,
      "loss": 0.5417,
      "step": 18530
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.167295455932617,
      "learning_rate": 6.0959322748161666e-06,
      "loss": 0.5574,
      "step": 18531
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4356365203857422,
      "learning_rate": 6.094904333697695e-06,
      "loss": 0.553,
      "step": 18532
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5376014709472656,
      "learning_rate": 6.093876441265518e-06,
      "loss": 0.4149,
      "step": 18533
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.734802007675171,
      "learning_rate": 6.09284859753245e-06,
      "loss": 0.4361,
      "step": 18534
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.545907735824585,
      "learning_rate": 6.0918208025113035e-06,
      "loss": 0.4978,
      "step": 18535
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8464686870574951,
      "learning_rate": 6.090793056214896e-06,
      "loss": 0.5218,
      "step": 18536
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6124155521392822,
      "learning_rate": 6.089765358656037e-06,
      "loss": 0.5564,
      "step": 18537
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7517952919006348,
      "learning_rate": 6.088737709847543e-06,
      "loss": 0.4408,
      "step": 18538
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.783842921257019,
      "learning_rate": 6.087710109802221e-06,
      "loss": 0.494,
      "step": 18539
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3891767263412476,
      "learning_rate": 6.0866825585328855e-06,
      "loss": 0.4284,
      "step": 18540
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.9301257133483887,
      "learning_rate": 6.0856550560523505e-06,
      "loss": 0.5932,
      "step": 18541
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.541330099105835,
      "learning_rate": 6.0846276023734205e-06,
      "loss": 0.5162,
      "step": 18542
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5985677242279053,
      "learning_rate": 6.083600197508908e-06,
      "loss": 0.4482,
      "step": 18543
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7774057388305664,
      "learning_rate": 6.082572841471624e-06,
      "loss": 0.5715,
      "step": 18544
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5177068710327148,
      "learning_rate": 6.081545534274372e-06,
      "loss": 0.5401,
      "step": 18545
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5670603513717651,
      "learning_rate": 6.080518275929965e-06,
      "loss": 0.4833,
      "step": 18546
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6940277814865112,
      "learning_rate": 6.079491066451206e-06,
      "loss": 0.5562,
      "step": 18547
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7736469507217407,
      "learning_rate": 6.078463905850904e-06,
      "loss": 0.5303,
      "step": 18548
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4631707668304443,
      "learning_rate": 6.077436794141864e-06,
      "loss": 0.4513,
      "step": 18549
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4859604835510254,
      "learning_rate": 6.076409731336892e-06,
      "loss": 0.5285,
      "step": 18550
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.705392599105835,
      "learning_rate": 6.075382717448795e-06,
      "loss": 0.6162,
      "step": 18551
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4509910345077515,
      "learning_rate": 6.0743557524903705e-06,
      "loss": 0.4593,
      "step": 18552
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7052561044692993,
      "learning_rate": 6.073328836474428e-06,
      "loss": 0.5376,
      "step": 18553
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.456005334854126,
      "learning_rate": 6.07230196941377e-06,
      "loss": 0.4472,
      "step": 18554
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2645564079284668,
      "learning_rate": 6.071275151321196e-06,
      "loss": 0.4078,
      "step": 18555
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5273009538650513,
      "learning_rate": 6.070248382209508e-06,
      "loss": 0.515,
      "step": 18556
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.0154342651367188,
      "learning_rate": 6.069221662091511e-06,
      "loss": 0.5216,
      "step": 18557
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4195668697357178,
      "learning_rate": 6.068194990980002e-06,
      "loss": 0.5018,
      "step": 18558
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7504059076309204,
      "learning_rate": 6.0671683688877834e-06,
      "loss": 0.5139,
      "step": 18559
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7327229976654053,
      "learning_rate": 6.066141795827651e-06,
      "loss": 0.5543,
      "step": 18560
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4553941488265991,
      "learning_rate": 6.065115271812404e-06,
      "loss": 0.4219,
      "step": 18561
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6516485214233398,
      "learning_rate": 6.064088796854845e-06,
      "loss": 0.4568,
      "step": 18562
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7527846097946167,
      "learning_rate": 6.063062370967765e-06,
      "loss": 0.5129,
      "step": 18563
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4445115327835083,
      "learning_rate": 6.062035994163968e-06,
      "loss": 0.6117,
      "step": 18564
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.660918951034546,
      "learning_rate": 6.0610096664562435e-06,
      "loss": 0.4894,
      "step": 18565
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8369388580322266,
      "learning_rate": 6.059983387857389e-06,
      "loss": 0.4687,
      "step": 18566
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6817944049835205,
      "learning_rate": 6.058957158380203e-06,
      "loss": 0.5838,
      "step": 18567
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4018104076385498,
      "learning_rate": 6.0579309780374765e-06,
      "loss": 0.5294,
      "step": 18568
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3854126930236816,
      "learning_rate": 6.056904846842002e-06,
      "loss": 0.5477,
      "step": 18569
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.702025055885315,
      "learning_rate": 6.055878764806579e-06,
      "loss": 0.6253,
      "step": 18570
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7722413539886475,
      "learning_rate": 6.054852731943992e-06,
      "loss": 0.5568,
      "step": 18571
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7509058713912964,
      "learning_rate": 6.05382674826704e-06,
      "loss": 0.4279,
      "step": 18572
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4654808044433594,
      "learning_rate": 6.052800813788508e-06,
      "loss": 0.4926,
      "step": 18573
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3730453252792358,
      "learning_rate": 6.05177492852119e-06,
      "loss": 0.4727,
      "step": 18574
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4991778135299683,
      "learning_rate": 6.050749092477879e-06,
      "loss": 0.5261,
      "step": 18575
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8068252801895142,
      "learning_rate": 6.0497233056713575e-06,
      "loss": 0.4222,
      "step": 18576
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7298436164855957,
      "learning_rate": 6.048697568114422e-06,
      "loss": 0.5132,
      "step": 18577
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.467429280281067,
      "learning_rate": 6.047671879819854e-06,
      "loss": 0.5911,
      "step": 18578
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3986034393310547,
      "learning_rate": 6.046646240800444e-06,
      "loss": 0.5112,
      "step": 18579
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8161993026733398,
      "learning_rate": 6.045620651068982e-06,
      "loss": 0.6187,
      "step": 18580
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7757508754730225,
      "learning_rate": 6.0445951106382494e-06,
      "loss": 0.5194,
      "step": 18581
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5357081890106201,
      "learning_rate": 6.043569619521034e-06,
      "loss": 0.4561,
      "step": 18582
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6847847700119019,
      "learning_rate": 6.042544177730123e-06,
      "loss": 0.5014,
      "step": 18583
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4772698879241943,
      "learning_rate": 6.041518785278297e-06,
      "loss": 0.5152,
      "step": 18584
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3741669654846191,
      "learning_rate": 6.040493442178343e-06,
      "loss": 0.5276,
      "step": 18585
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6789237260818481,
      "learning_rate": 6.0394681484430436e-06,
      "loss": 0.4779,
      "step": 18586
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.629377007484436,
      "learning_rate": 6.03844290408518e-06,
      "loss": 0.5222,
      "step": 18587
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5574768781661987,
      "learning_rate": 6.037417709117539e-06,
      "loss": 0.6461,
      "step": 18588
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8330870866775513,
      "learning_rate": 6.036392563552895e-06,
      "loss": 0.5654,
      "step": 18589
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5153343677520752,
      "learning_rate": 6.035367467404035e-06,
      "loss": 0.4667,
      "step": 18590
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4347593784332275,
      "learning_rate": 6.034342420683735e-06,
      "loss": 0.4624,
      "step": 18591
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.844128131866455,
      "learning_rate": 6.033317423404776e-06,
      "loss": 0.5814,
      "step": 18592
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8495784997940063,
      "learning_rate": 6.032292475579942e-06,
      "loss": 0.5417,
      "step": 18593
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.522065281867981,
      "learning_rate": 6.031267577222002e-06,
      "loss": 0.5749,
      "step": 18594
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.312211275100708,
      "learning_rate": 6.030242728343741e-06,
      "loss": 0.5399,
      "step": 18595
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6733076572418213,
      "learning_rate": 6.029217928957934e-06,
      "loss": 0.5037,
      "step": 18596
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6194989681243896,
      "learning_rate": 6.028193179077356e-06,
      "loss": 0.4927,
      "step": 18597
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.101998805999756,
      "learning_rate": 6.027168478714786e-06,
      "loss": 0.5193,
      "step": 18598
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.740857720375061,
      "learning_rate": 6.026143827882998e-06,
      "loss": 0.5338,
      "step": 18599
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.978041172027588,
      "learning_rate": 6.025119226594765e-06,
      "loss": 0.518,
      "step": 18600
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6173458099365234,
      "learning_rate": 6.0240946748628636e-06,
      "loss": 0.5074,
      "step": 18601
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.526016116142273,
      "learning_rate": 6.023070172700064e-06,
      "loss": 0.51,
      "step": 18602
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8939247131347656,
      "learning_rate": 6.022045720119144e-06,
      "loss": 0.5482,
      "step": 18603
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5764080286026,
      "learning_rate": 6.021021317132871e-06,
      "loss": 0.4512,
      "step": 18604
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.51603102684021,
      "learning_rate": 6.019996963754019e-06,
      "loss": 0.3498,
      "step": 18605
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7665210962295532,
      "learning_rate": 6.0189726599953615e-06,
      "loss": 0.5951,
      "step": 18606
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4649854898452759,
      "learning_rate": 6.017948405869663e-06,
      "loss": 0.5125,
      "step": 18607
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3970317840576172,
      "learning_rate": 6.016924201389695e-06,
      "loss": 0.5024,
      "step": 18608
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4645819664001465,
      "learning_rate": 6.015900046568233e-06,
      "loss": 0.5175,
      "step": 18609
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4770262241363525,
      "learning_rate": 6.014875941418037e-06,
      "loss": 0.5911,
      "step": 18610
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5793901681900024,
      "learning_rate": 6.01385188595188e-06,
      "loss": 0.5877,
      "step": 18611
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.8179126977920532,
      "learning_rate": 6.012827880182526e-06,
      "loss": 0.6059,
      "step": 18612
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5210860967636108,
      "learning_rate": 6.0118039241227436e-06,
      "loss": 0.3821,
      "step": 18613
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6464170217514038,
      "learning_rate": 6.0107800177852994e-06,
      "loss": 0.602,
      "step": 18614
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.054133892059326,
      "learning_rate": 6.009756161182957e-06,
      "loss": 0.5368,
      "step": 18615
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8707664012908936,
      "learning_rate": 6.008732354328484e-06,
      "loss": 0.4382,
      "step": 18616
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.538372278213501,
      "learning_rate": 6.007708597234641e-06,
      "loss": 0.4386,
      "step": 18617
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2286453247070312,
      "learning_rate": 6.0066848899141935e-06,
      "loss": 0.4979,
      "step": 18618
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.298879623413086,
      "learning_rate": 6.005661232379907e-06,
      "loss": 0.5587,
      "step": 18619
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4514679908752441,
      "learning_rate": 6.004637624644538e-06,
      "loss": 0.466,
      "step": 18620
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6787080764770508,
      "learning_rate": 6.003614066720851e-06,
      "loss": 0.5431,
      "step": 18621
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8557465076446533,
      "learning_rate": 6.00259055862161e-06,
      "loss": 0.5497,
      "step": 18622
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.29094398021698,
      "learning_rate": 6.001567100359571e-06,
      "loss": 0.4892,
      "step": 18623
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4172725677490234,
      "learning_rate": 6.000543691947498e-06,
      "loss": 0.4287,
      "step": 18624
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3970516920089722,
      "learning_rate": 5.999520333398145e-06,
      "loss": 0.4718,
      "step": 18625
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6310126781463623,
      "learning_rate": 5.998497024724275e-06,
      "loss": 0.5887,
      "step": 18626
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4827678203582764,
      "learning_rate": 5.997473765938645e-06,
      "loss": 0.5024,
      "step": 18627
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6835070848464966,
      "learning_rate": 5.9964505570540096e-06,
      "loss": 0.5547,
      "step": 18628
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5224120616912842,
      "learning_rate": 5.995427398083131e-06,
      "loss": 0.4796,
      "step": 18629
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7139676809310913,
      "learning_rate": 5.99440428903876e-06,
      "loss": 0.5365,
      "step": 18630
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7698827981948853,
      "learning_rate": 5.993381229933654e-06,
      "loss": 0.6201,
      "step": 18631
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4463697671890259,
      "learning_rate": 5.992358220780572e-06,
      "loss": 0.62,
      "step": 18632
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5124666690826416,
      "learning_rate": 5.991335261592263e-06,
      "loss": 0.5037,
      "step": 18633
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5577878952026367,
      "learning_rate": 5.990312352381479e-06,
      "loss": 0.3637,
      "step": 18634
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.307663917541504,
      "learning_rate": 5.9892894931609755e-06,
      "loss": 0.4828,
      "step": 18635
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9163881540298462,
      "learning_rate": 5.988266683943509e-06,
      "loss": 0.4151,
      "step": 18636
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4012773036956787,
      "learning_rate": 5.9872439247418235e-06,
      "loss": 0.3548,
      "step": 18637
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5264235734939575,
      "learning_rate": 5.9862212155686775e-06,
      "loss": 0.5216,
      "step": 18638
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7741384506225586,
      "learning_rate": 5.985198556436815e-06,
      "loss": 0.5583,
      "step": 18639
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2107141017913818,
      "learning_rate": 5.98417594735899e-06,
      "loss": 0.4182,
      "step": 18640
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6375757455825806,
      "learning_rate": 5.983153388347951e-06,
      "loss": 0.5302,
      "step": 18641
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.471852421760559,
      "learning_rate": 5.982130879416446e-06,
      "loss": 0.3793,
      "step": 18642
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2805614471435547,
      "learning_rate": 5.981108420577225e-06,
      "loss": 0.553,
      "step": 18643
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4387614727020264,
      "learning_rate": 5.9800860118430315e-06,
      "loss": 0.5474,
      "step": 18644
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2772953510284424,
      "learning_rate": 5.9790636532266154e-06,
      "loss": 0.4417,
      "step": 18645
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5043652057647705,
      "learning_rate": 5.978041344740724e-06,
      "loss": 0.4598,
      "step": 18646
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5625704526901245,
      "learning_rate": 5.977019086398097e-06,
      "loss": 0.3877,
      "step": 18647
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.435388207435608,
      "learning_rate": 5.975996878211484e-06,
      "loss": 0.3843,
      "step": 18648
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4257450103759766,
      "learning_rate": 5.974974720193631e-06,
      "loss": 0.5882,
      "step": 18649
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.558224081993103,
      "learning_rate": 5.973952612357276e-06,
      "loss": 0.511,
      "step": 18650
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.1008284091949463,
      "learning_rate": 5.972930554715167e-06,
      "loss": 0.5044,
      "step": 18651
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6455957889556885,
      "learning_rate": 5.971908547280043e-06,
      "loss": 0.5967,
      "step": 18652
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4861857891082764,
      "learning_rate": 5.970886590064646e-06,
      "loss": 0.5099,
      "step": 18653
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5954079627990723,
      "learning_rate": 5.969864683081721e-06,
      "loss": 0.4754,
      "step": 18654
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5525521039962769,
      "learning_rate": 5.9688428263440034e-06,
      "loss": 0.5732,
      "step": 18655
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3834420442581177,
      "learning_rate": 5.967821019864237e-06,
      "loss": 0.4277,
      "step": 18656
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5737224817276,
      "learning_rate": 5.966799263655158e-06,
      "loss": 0.4823,
      "step": 18657
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.719159722328186,
      "learning_rate": 5.965777557729505e-06,
      "loss": 0.5076,
      "step": 18658
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4625909328460693,
      "learning_rate": 5.964755902100021e-06,
      "loss": 0.3502,
      "step": 18659
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5997241735458374,
      "learning_rate": 5.963734296779436e-06,
      "loss": 0.5104,
      "step": 18660
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.637249231338501,
      "learning_rate": 5.962712741780492e-06,
      "loss": 0.4968,
      "step": 18661
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7032214403152466,
      "learning_rate": 5.961691237115925e-06,
      "loss": 0.4543,
      "step": 18662
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7537713050842285,
      "learning_rate": 5.960669782798467e-06,
      "loss": 0.4958,
      "step": 18663
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5241385698318481,
      "learning_rate": 5.959648378840858e-06,
      "loss": 0.5901,
      "step": 18664
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7427501678466797,
      "learning_rate": 5.958627025255825e-06,
      "loss": 0.5588,
      "step": 18665
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7267049551010132,
      "learning_rate": 5.957605722056108e-06,
      "loss": 0.4561,
      "step": 18666
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5235352516174316,
      "learning_rate": 5.956584469254439e-06,
      "loss": 0.5535,
      "step": 18667
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4617373943328857,
      "learning_rate": 5.9555632668635464e-06,
      "loss": 0.5408,
      "step": 18668
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.256334066390991,
      "learning_rate": 5.954542114896167e-06,
      "loss": 0.521,
      "step": 18669
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4369380474090576,
      "learning_rate": 5.953521013365028e-06,
      "loss": 0.6011,
      "step": 18670
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.222482681274414,
      "learning_rate": 5.9524999622828606e-06,
      "loss": 0.4572,
      "step": 18671
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4943113327026367,
      "learning_rate": 5.951478961662398e-06,
      "loss": 0.4907,
      "step": 18672
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4270576238632202,
      "learning_rate": 5.9504580115163634e-06,
      "loss": 0.5099,
      "step": 18673
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6177160739898682,
      "learning_rate": 5.94943711185749e-06,
      "loss": 0.4721,
      "step": 18674
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4279661178588867,
      "learning_rate": 5.948416262698506e-06,
      "loss": 0.4565,
      "step": 18675
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6902940273284912,
      "learning_rate": 5.9473954640521345e-06,
      "loss": 0.5173,
      "step": 18676
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.841726303100586,
      "learning_rate": 5.946374715931108e-06,
      "loss": 0.5718,
      "step": 18677
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4714689254760742,
      "learning_rate": 5.945354018348147e-06,
      "loss": 0.4743,
      "step": 18678
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.092573404312134,
      "learning_rate": 5.94433337131598e-06,
      "loss": 0.5364,
      "step": 18679
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4602105617523193,
      "learning_rate": 5.943312774847331e-06,
      "loss": 0.6099,
      "step": 18680
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4105675220489502,
      "learning_rate": 5.942292228954925e-06,
      "loss": 0.4499,
      "step": 18681
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3971933126449585,
      "learning_rate": 5.941271733651484e-06,
      "loss": 0.5018,
      "step": 18682
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4526976346969604,
      "learning_rate": 5.940251288949732e-06,
      "loss": 0.5265,
      "step": 18683
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6916799545288086,
      "learning_rate": 5.939230894862389e-06,
      "loss": 0.5546,
      "step": 18684
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.0180938243865967,
      "learning_rate": 5.938210551402181e-06,
      "loss": 0.5207,
      "step": 18685
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4892728328704834,
      "learning_rate": 5.937190258581827e-06,
      "loss": 0.5439,
      "step": 18686
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6687718629837036,
      "learning_rate": 5.936170016414045e-06,
      "loss": 0.4891,
      "step": 18687
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6965287923812866,
      "learning_rate": 5.935149824911559e-06,
      "loss": 0.5335,
      "step": 18688
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6470261812210083,
      "learning_rate": 5.934129684087084e-06,
      "loss": 0.5953,
      "step": 18689
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9111220836639404,
      "learning_rate": 5.933109593953341e-06,
      "loss": 0.482,
      "step": 18690
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8823102712631226,
      "learning_rate": 5.932089554523046e-06,
      "loss": 0.6227,
      "step": 18691
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.867898941040039,
      "learning_rate": 5.931069565808918e-06,
      "loss": 0.4698,
      "step": 18692
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5977293252944946,
      "learning_rate": 5.930049627823674e-06,
      "loss": 0.545,
      "step": 18693
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4333630800247192,
      "learning_rate": 5.929029740580028e-06,
      "loss": 0.5891,
      "step": 18694
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.325599193572998,
      "learning_rate": 5.928009904090697e-06,
      "loss": 0.5396,
      "step": 18695
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4558860063552856,
      "learning_rate": 5.926990118368394e-06,
      "loss": 0.5072,
      "step": 18696
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3754526376724243,
      "learning_rate": 5.925970383425832e-06,
      "loss": 0.5074,
      "step": 18697
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7956271171569824,
      "learning_rate": 5.924950699275731e-06,
      "loss": 0.5822,
      "step": 18698
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.618414878845215,
      "learning_rate": 5.923931065930795e-06,
      "loss": 0.481,
      "step": 18699
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.79752779006958,
      "learning_rate": 5.9229114834037416e-06,
      "loss": 0.4999,
      "step": 18700
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.9436959028244019,
      "learning_rate": 5.921891951707283e-06,
      "loss": 0.585,
      "step": 18701
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2754027843475342,
      "learning_rate": 5.920872470854126e-06,
      "loss": 0.5205,
      "step": 18702
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6115052700042725,
      "learning_rate": 5.9198530408569845e-06,
      "loss": 0.58,
      "step": 18703
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2931350469589233,
      "learning_rate": 5.918833661728564e-06,
      "loss": 0.4897,
      "step": 18704
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6127063035964966,
      "learning_rate": 5.917814333481578e-06,
      "loss": 0.5126,
      "step": 18705
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8579198122024536,
      "learning_rate": 5.916795056128733e-06,
      "loss": 0.5639,
      "step": 18706
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5389317274093628,
      "learning_rate": 5.915775829682735e-06,
      "loss": 0.5279,
      "step": 18707
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.721508264541626,
      "learning_rate": 5.914756654156295e-06,
      "loss": 0.5143,
      "step": 18708
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7419325113296509,
      "learning_rate": 5.913737529562115e-06,
      "loss": 0.5926,
      "step": 18709
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6962571144104004,
      "learning_rate": 5.912718455912902e-06,
      "loss": 0.483,
      "step": 18710
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5958033800125122,
      "learning_rate": 5.911699433221365e-06,
      "loss": 0.511,
      "step": 18711
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6470346450805664,
      "learning_rate": 5.9106804615002045e-06,
      "loss": 0.5257,
      "step": 18712
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6825934648513794,
      "learning_rate": 5.909661540762125e-06,
      "loss": 0.4917,
      "step": 18713
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5172194242477417,
      "learning_rate": 5.9086426710198315e-06,
      "loss": 0.5824,
      "step": 18714
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5314245223999023,
      "learning_rate": 5.907623852286024e-06,
      "loss": 0.5621,
      "step": 18715
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5401562452316284,
      "learning_rate": 5.906605084573409e-06,
      "loss": 0.5436,
      "step": 18716
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.587249755859375,
      "learning_rate": 5.905586367894682e-06,
      "loss": 0.5254,
      "step": 18717
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5620001554489136,
      "learning_rate": 5.904567702262546e-06,
      "loss": 0.5332,
      "step": 18718
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3949060440063477,
      "learning_rate": 5.9035490876897035e-06,
      "loss": 0.5656,
      "step": 18719
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.475516676902771,
      "learning_rate": 5.902530524188852e-06,
      "loss": 0.4746,
      "step": 18720
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5448569059371948,
      "learning_rate": 5.9015120117726895e-06,
      "loss": 0.5498,
      "step": 18721
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5814629793167114,
      "learning_rate": 5.900493550453915e-06,
      "loss": 0.4743,
      "step": 18722
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5040236711502075,
      "learning_rate": 5.899475140245225e-06,
      "loss": 0.4675,
      "step": 18723
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5030012130737305,
      "learning_rate": 5.898456781159321e-06,
      "loss": 0.4929,
      "step": 18724
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6755532026290894,
      "learning_rate": 5.897438473208893e-06,
      "loss": 0.533,
      "step": 18725
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8017356395721436,
      "learning_rate": 5.89642021640664e-06,
      "loss": 0.5607,
      "step": 18726
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8213255405426025,
      "learning_rate": 5.895402010765257e-06,
      "loss": 0.5128,
      "step": 18727
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.362488031387329,
      "learning_rate": 5.894383856297436e-06,
      "loss": 0.5311,
      "step": 18728
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4645925760269165,
      "learning_rate": 5.893365753015876e-06,
      "loss": 0.5355,
      "step": 18729
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6374934911727905,
      "learning_rate": 5.892347700933263e-06,
      "loss": 0.5893,
      "step": 18730
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6544991731643677,
      "learning_rate": 5.891329700062294e-06,
      "loss": 0.4627,
      "step": 18731
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3652735948562622,
      "learning_rate": 5.89031175041566e-06,
      "loss": 0.46,
      "step": 18732
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3950963020324707,
      "learning_rate": 5.889293852006052e-06,
      "loss": 0.4051,
      "step": 18733
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8748950958251953,
      "learning_rate": 5.88827600484616e-06,
      "loss": 0.4905,
      "step": 18734
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4324519634246826,
      "learning_rate": 5.887258208948674e-06,
      "loss": 0.4177,
      "step": 18735
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4839001893997192,
      "learning_rate": 5.886240464326285e-06,
      "loss": 0.4603,
      "step": 18736
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4802826642990112,
      "learning_rate": 5.88522277099168e-06,
      "loss": 0.5012,
      "step": 18737
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4355900287628174,
      "learning_rate": 5.884205128957546e-06,
      "loss": 0.5871,
      "step": 18738
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.0699028968811035,
      "learning_rate": 5.88318753823657e-06,
      "loss": 0.6473,
      "step": 18739
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7462635040283203,
      "learning_rate": 5.882169998841443e-06,
      "loss": 0.4651,
      "step": 18740
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.984411597251892,
      "learning_rate": 5.881152510784848e-06,
      "loss": 0.5328,
      "step": 18741
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.443429708480835,
      "learning_rate": 5.88013507407947e-06,
      "loss": 0.4699,
      "step": 18742
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.659567952156067,
      "learning_rate": 5.879117688737994e-06,
      "loss": 0.5604,
      "step": 18743
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4796768426895142,
      "learning_rate": 5.878100354773104e-06,
      "loss": 0.4898,
      "step": 18744
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4090509414672852,
      "learning_rate": 5.877083072197486e-06,
      "loss": 0.52,
      "step": 18745
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7586544752120972,
      "learning_rate": 5.876065841023819e-06,
      "loss": 0.539,
      "step": 18746
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5525970458984375,
      "learning_rate": 5.875048661264788e-06,
      "loss": 0.6257,
      "step": 18747
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4382317066192627,
      "learning_rate": 5.874031532933073e-06,
      "loss": 0.5445,
      "step": 18748
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8592199087142944,
      "learning_rate": 5.873014456041355e-06,
      "loss": 0.427,
      "step": 18749
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3669923543930054,
      "learning_rate": 5.871997430602317e-06,
      "loss": 0.4748,
      "step": 18750
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5925078392028809,
      "learning_rate": 5.8709804566286344e-06,
      "loss": 0.4377,
      "step": 18751
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5275872945785522,
      "learning_rate": 5.869963534132987e-06,
      "loss": 0.5139,
      "step": 18752
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5010753870010376,
      "learning_rate": 5.868946663128058e-06,
      "loss": 0.5487,
      "step": 18753
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.021331548690796,
      "learning_rate": 5.867929843626519e-06,
      "loss": 0.4718,
      "step": 18754
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.603044867515564,
      "learning_rate": 5.866913075641051e-06,
      "loss": 0.4661,
      "step": 18755
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4496272802352905,
      "learning_rate": 5.865896359184326e-06,
      "loss": 0.5442,
      "step": 18756
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3721647262573242,
      "learning_rate": 5.864879694269026e-06,
      "loss": 0.4901,
      "step": 18757
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.786852240562439,
      "learning_rate": 5.863863080907823e-06,
      "loss": 0.5179,
      "step": 18758
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.732095718383789,
      "learning_rate": 5.862846519113389e-06,
      "loss": 0.5594,
      "step": 18759
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.232905626296997,
      "learning_rate": 5.861830008898401e-06,
      "loss": 0.4749,
      "step": 18760
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.0231964588165283,
      "learning_rate": 5.860813550275532e-06,
      "loss": 0.6885,
      "step": 18761
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5311686992645264,
      "learning_rate": 5.859797143257453e-06,
      "loss": 0.5344,
      "step": 18762
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6316606998443604,
      "learning_rate": 5.8587807878568395e-06,
      "loss": 0.4746,
      "step": 18763
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7590727806091309,
      "learning_rate": 5.857764484086359e-06,
      "loss": 0.4604,
      "step": 18764
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.581560730934143,
      "learning_rate": 5.856748231958682e-06,
      "loss": 0.5322,
      "step": 18765
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.61281156539917,
      "learning_rate": 5.855732031486482e-06,
      "loss": 0.4773,
      "step": 18766
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.698089361190796,
      "learning_rate": 5.854715882682425e-06,
      "loss": 0.4878,
      "step": 18767
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6490600109100342,
      "learning_rate": 5.8536997855591814e-06,
      "loss": 0.5608,
      "step": 18768
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8083138465881348,
      "learning_rate": 5.852683740129419e-06,
      "loss": 0.5728,
      "step": 18769
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6265720129013062,
      "learning_rate": 5.851667746405803e-06,
      "loss": 0.5226,
      "step": 18770
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5548820495605469,
      "learning_rate": 5.850651804401006e-06,
      "loss": 0.4421,
      "step": 18771
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4377119541168213,
      "learning_rate": 5.849635914127687e-06,
      "loss": 0.5411,
      "step": 18772
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5650231838226318,
      "learning_rate": 5.848620075598518e-06,
      "loss": 0.4884,
      "step": 18773
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4983149766921997,
      "learning_rate": 5.847604288826159e-06,
      "loss": 0.525,
      "step": 18774
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6698700189590454,
      "learning_rate": 5.846588553823274e-06,
      "loss": 0.5696,
      "step": 18775
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5845279693603516,
      "learning_rate": 5.845572870602531e-06,
      "loss": 0.469,
      "step": 18776
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5627374649047852,
      "learning_rate": 5.844557239176589e-06,
      "loss": 0.506,
      "step": 18777
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4167513847351074,
      "learning_rate": 5.843541659558112e-06,
      "loss": 0.5694,
      "step": 18778
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4538178443908691,
      "learning_rate": 5.842526131759763e-06,
      "loss": 0.4474,
      "step": 18779
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6257827281951904,
      "learning_rate": 5.841510655794198e-06,
      "loss": 0.5387,
      "step": 18780
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6647120714187622,
      "learning_rate": 5.840495231674083e-06,
      "loss": 0.4871,
      "step": 18781
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.610993504524231,
      "learning_rate": 5.8394798594120735e-06,
      "loss": 0.5186,
      "step": 18782
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8589884042739868,
      "learning_rate": 5.83846453902083e-06,
      "loss": 0.5695,
      "step": 18783
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5413107872009277,
      "learning_rate": 5.837449270513013e-06,
      "loss": 0.515,
      "step": 18784
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6022183895111084,
      "learning_rate": 5.836434053901279e-06,
      "loss": 0.5444,
      "step": 18785
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3989454507827759,
      "learning_rate": 5.835418889198279e-06,
      "loss": 0.4965,
      "step": 18786
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.537001371383667,
      "learning_rate": 5.83440377641668e-06,
      "loss": 0.463,
      "step": 18787
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5529204607009888,
      "learning_rate": 5.833388715569131e-06,
      "loss": 0.6127,
      "step": 18788
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6231129169464111,
      "learning_rate": 5.8323737066682865e-06,
      "loss": 0.5701,
      "step": 18789
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3307005167007446,
      "learning_rate": 5.831358749726808e-06,
      "loss": 0.4394,
      "step": 18790
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.075413703918457,
      "learning_rate": 5.8303438447573424e-06,
      "loss": 0.5912,
      "step": 18791
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5239311456680298,
      "learning_rate": 5.829328991772542e-06,
      "loss": 0.5168,
      "step": 18792
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4485819339752197,
      "learning_rate": 5.828314190785067e-06,
      "loss": 0.4359,
      "step": 18793
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3431518077850342,
      "learning_rate": 5.827299441807564e-06,
      "loss": 0.5055,
      "step": 18794
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6061389446258545,
      "learning_rate": 5.8262847448526845e-06,
      "loss": 0.4734,
      "step": 18795
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5264322757720947,
      "learning_rate": 5.825270099933077e-06,
      "loss": 0.5262,
      "step": 18796
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5940132141113281,
      "learning_rate": 5.824255507061398e-06,
      "loss": 0.5682,
      "step": 18797
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4178533554077148,
      "learning_rate": 5.823240966250292e-06,
      "loss": 0.4409,
      "step": 18798
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3893234729766846,
      "learning_rate": 5.822226477512405e-06,
      "loss": 0.4753,
      "step": 18799
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.539350986480713,
      "learning_rate": 5.821212040860393e-06,
      "loss": 0.5631,
      "step": 18800
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5558357238769531,
      "learning_rate": 5.820197656306898e-06,
      "loss": 0.4153,
      "step": 18801
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8605003356933594,
      "learning_rate": 5.819183323864565e-06,
      "loss": 0.5691,
      "step": 18802
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4007774591445923,
      "learning_rate": 5.818169043546045e-06,
      "loss": 0.4675,
      "step": 18803
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3661500215530396,
      "learning_rate": 5.817154815363982e-06,
      "loss": 0.4592,
      "step": 18804
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4552959203720093,
      "learning_rate": 5.816140639331016e-06,
      "loss": 0.5428,
      "step": 18805
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7542486190795898,
      "learning_rate": 5.815126515459799e-06,
      "loss": 0.5916,
      "step": 18806
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5990923643112183,
      "learning_rate": 5.81411244376297e-06,
      "loss": 0.5123,
      "step": 18807
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.646459698677063,
      "learning_rate": 5.813098424253174e-06,
      "loss": 0.4858,
      "step": 18808
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7483614683151245,
      "learning_rate": 5.812084456943046e-06,
      "loss": 0.5544,
      "step": 18809
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8876092433929443,
      "learning_rate": 5.811070541845236e-06,
      "loss": 0.6632,
      "step": 18810
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.9183415174484253,
      "learning_rate": 5.810056678972383e-06,
      "loss": 0.524,
      "step": 18811
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6758872270584106,
      "learning_rate": 5.809042868337122e-06,
      "loss": 0.5007,
      "step": 18812
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7333967685699463,
      "learning_rate": 5.8080291099521e-06,
      "loss": 0.6225,
      "step": 18813
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6039676666259766,
      "learning_rate": 5.807015403829954e-06,
      "loss": 0.5859,
      "step": 18814
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7650660276412964,
      "learning_rate": 5.806001749983316e-06,
      "loss": 0.532,
      "step": 18815
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4402015209197998,
      "learning_rate": 5.80498814842483e-06,
      "loss": 0.4938,
      "step": 18816
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6051828861236572,
      "learning_rate": 5.8039745991671345e-06,
      "loss": 0.5752,
      "step": 18817
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8311498165130615,
      "learning_rate": 5.802961102222857e-06,
      "loss": 0.5593,
      "step": 18818
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7530478239059448,
      "learning_rate": 5.801947657604643e-06,
      "loss": 0.6123,
      "step": 18819
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2497713565826416,
      "learning_rate": 5.800934265325124e-06,
      "loss": 0.4552,
      "step": 18820
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6359347105026245,
      "learning_rate": 5.799920925396934e-06,
      "loss": 0.4683,
      "step": 18821
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3589341640472412,
      "learning_rate": 5.7989076378327004e-06,
      "loss": 0.5746,
      "step": 18822
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4432164430618286,
      "learning_rate": 5.797894402645067e-06,
      "loss": 0.4837,
      "step": 18823
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.464736819267273,
      "learning_rate": 5.796881219846662e-06,
      "loss": 0.4893,
      "step": 18824
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8455003499984741,
      "learning_rate": 5.79586808945011e-06,
      "loss": 0.4881,
      "step": 18825
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8064478635787964,
      "learning_rate": 5.794855011468055e-06,
      "loss": 0.4854,
      "step": 18826
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5372750759124756,
      "learning_rate": 5.79384198591312e-06,
      "loss": 0.4791,
      "step": 18827
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3485374450683594,
      "learning_rate": 5.792829012797932e-06,
      "loss": 0.4912,
      "step": 18828
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.602486491203308,
      "learning_rate": 5.791816092135128e-06,
      "loss": 0.5648,
      "step": 18829
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6654322147369385,
      "learning_rate": 5.790803223937331e-06,
      "loss": 0.5344,
      "step": 18830
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5182461738586426,
      "learning_rate": 5.7897904082171665e-06,
      "loss": 0.5274,
      "step": 18831
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.3972744941711426,
      "learning_rate": 5.7887776449872704e-06,
      "loss": 0.5589,
      "step": 18832
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5730559825897217,
      "learning_rate": 5.787764934260263e-06,
      "loss": 0.4797,
      "step": 18833
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.0525944232940674,
      "learning_rate": 5.786752276048771e-06,
      "loss": 0.5696,
      "step": 18834
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5365437269210815,
      "learning_rate": 5.7857396703654156e-06,
      "loss": 0.5407,
      "step": 18835
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5610718727111816,
      "learning_rate": 5.7847271172228305e-06,
      "loss": 0.5033,
      "step": 18836
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.157241106033325,
      "learning_rate": 5.783714616633635e-06,
      "loss": 0.4708,
      "step": 18837
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3817845582962036,
      "learning_rate": 5.782702168610446e-06,
      "loss": 0.4982,
      "step": 18838
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5784300565719604,
      "learning_rate": 5.7816897731658975e-06,
      "loss": 0.4271,
      "step": 18839
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6470941305160522,
      "learning_rate": 5.780677430312604e-06,
      "loss": 0.5361,
      "step": 18840
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6852262020111084,
      "learning_rate": 5.779665140063186e-06,
      "loss": 0.6598,
      "step": 18841
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4699398279190063,
      "learning_rate": 5.778652902430273e-06,
      "loss": 0.5794,
      "step": 18842
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7826077938079834,
      "learning_rate": 5.777640717426476e-06,
      "loss": 0.6587,
      "step": 18843
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3703587055206299,
      "learning_rate": 5.776628585064412e-06,
      "loss": 0.5506,
      "step": 18844
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4864692687988281,
      "learning_rate": 5.775616505356711e-06,
      "loss": 0.4864,
      "step": 18845
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4202872514724731,
      "learning_rate": 5.774604478315984e-06,
      "loss": 0.5374,
      "step": 18846
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.004056930541992,
      "learning_rate": 5.773592503954848e-06,
      "loss": 0.5568,
      "step": 18847
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4462120532989502,
      "learning_rate": 5.772580582285917e-06,
      "loss": 0.5912,
      "step": 18848
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5214952230453491,
      "learning_rate": 5.771568713321816e-06,
      "loss": 0.6279,
      "step": 18849
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3044663667678833,
      "learning_rate": 5.770556897075152e-06,
      "loss": 0.4527,
      "step": 18850
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.481624960899353,
      "learning_rate": 5.769545133558541e-06,
      "loss": 0.5586,
      "step": 18851
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5356993675231934,
      "learning_rate": 5.768533422784601e-06,
      "loss": 0.5181,
      "step": 18852
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.41579270362854,
      "learning_rate": 5.767521764765944e-06,
      "loss": 0.5632,
      "step": 18853
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3679578304290771,
      "learning_rate": 5.766510159515179e-06,
      "loss": 0.4443,
      "step": 18854
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6873602867126465,
      "learning_rate": 5.765498607044924e-06,
      "loss": 0.4979,
      "step": 18855
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7262760400772095,
      "learning_rate": 5.764487107367786e-06,
      "loss": 0.534,
      "step": 18856
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4953209161758423,
      "learning_rate": 5.763475660496373e-06,
      "loss": 0.5635,
      "step": 18857
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6805306673049927,
      "learning_rate": 5.762464266443304e-06,
      "loss": 0.5506,
      "step": 18858
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7773971557617188,
      "learning_rate": 5.761452925221184e-06,
      "loss": 0.5233,
      "step": 18859
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.344653844833374,
      "learning_rate": 5.76044163684262e-06,
      "loss": 0.5052,
      "step": 18860
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5431427955627441,
      "learning_rate": 5.759430401320218e-06,
      "loss": 0.4624,
      "step": 18861
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.580585241317749,
      "learning_rate": 5.758419218666594e-06,
      "loss": 0.5327,
      "step": 18862
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8757623434066772,
      "learning_rate": 5.757408088894347e-06,
      "loss": 0.4644,
      "step": 18863
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5809298753738403,
      "learning_rate": 5.756397012016084e-06,
      "loss": 0.532,
      "step": 18864
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.687660813331604,
      "learning_rate": 5.755385988044415e-06,
      "loss": 0.5847,
      "step": 18865
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6027641296386719,
      "learning_rate": 5.754375016991942e-06,
      "loss": 0.5431,
      "step": 18866
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.702579379081726,
      "learning_rate": 5.7533640988712655e-06,
      "loss": 0.4413,
      "step": 18867
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.1926519870758057,
      "learning_rate": 5.752353233694998e-06,
      "loss": 0.5463,
      "step": 18868
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7477306127548218,
      "learning_rate": 5.751342421475735e-06,
      "loss": 0.6263,
      "step": 18869
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.612351655960083,
      "learning_rate": 5.750331662226079e-06,
      "loss": 0.4195,
      "step": 18870
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5887221097946167,
      "learning_rate": 5.749320955958635e-06,
      "loss": 0.5633,
      "step": 18871
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5367002487182617,
      "learning_rate": 5.748310302686002e-06,
      "loss": 0.5206,
      "step": 18872
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.664710521697998,
      "learning_rate": 5.747299702420781e-06,
      "loss": 0.544,
      "step": 18873
    },
    {
      "epoch": 1.95,
      "grad_norm": 2.0107967853546143,
      "learning_rate": 5.746289155175568e-06,
      "loss": 0.4772,
      "step": 18874
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5283610820770264,
      "learning_rate": 5.745278660962967e-06,
      "loss": 0.5016,
      "step": 18875
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3387285470962524,
      "learning_rate": 5.744268219795574e-06,
      "loss": 0.4684,
      "step": 18876
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2693244218826294,
      "learning_rate": 5.743257831685985e-06,
      "loss": 0.4876,
      "step": 18877
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5696433782577515,
      "learning_rate": 5.742247496646799e-06,
      "loss": 0.5115,
      "step": 18878
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4836522340774536,
      "learning_rate": 5.741237214690614e-06,
      "loss": 0.5315,
      "step": 18879
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5005557537078857,
      "learning_rate": 5.740226985830017e-06,
      "loss": 0.5852,
      "step": 18880
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5076324939727783,
      "learning_rate": 5.7392168100776134e-06,
      "loss": 0.4923,
      "step": 18881
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5737590789794922,
      "learning_rate": 5.738206687445994e-06,
      "loss": 0.4453,
      "step": 18882
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5064399242401123,
      "learning_rate": 5.737196617947747e-06,
      "loss": 0.4875,
      "step": 18883
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4006065130233765,
      "learning_rate": 5.736186601595474e-06,
      "loss": 0.4639,
      "step": 18884
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6096055507659912,
      "learning_rate": 5.735176638401763e-06,
      "loss": 0.4529,
      "step": 18885
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8725775480270386,
      "learning_rate": 5.7341667283792045e-06,
      "loss": 0.3699,
      "step": 18886
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5092358589172363,
      "learning_rate": 5.733156871540389e-06,
      "loss": 0.5433,
      "step": 18887
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5934873819351196,
      "learning_rate": 5.7321470678979094e-06,
      "loss": 0.503,
      "step": 18888
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3529529571533203,
      "learning_rate": 5.731137317464355e-06,
      "loss": 0.4139,
      "step": 18889
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.350996971130371,
      "learning_rate": 5.73012762025231e-06,
      "loss": 0.4371,
      "step": 18890
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3364975452423096,
      "learning_rate": 5.729117976274372e-06,
      "loss": 0.4416,
      "step": 18891
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7765504121780396,
      "learning_rate": 5.728108385543122e-06,
      "loss": 0.5186,
      "step": 18892
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.540269374847412,
      "learning_rate": 5.727098848071145e-06,
      "loss": 0.5006,
      "step": 18893
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3250391483306885,
      "learning_rate": 5.7260893638710325e-06,
      "loss": 0.5305,
      "step": 18894
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.9035922288894653,
      "learning_rate": 5.725079932955371e-06,
      "loss": 0.5697,
      "step": 18895
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3312628269195557,
      "learning_rate": 5.724070555336736e-06,
      "loss": 0.4458,
      "step": 18896
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7422987222671509,
      "learning_rate": 5.7230612310277225e-06,
      "loss": 0.518,
      "step": 18897
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8083065748214722,
      "learning_rate": 5.722051960040911e-06,
      "loss": 0.5051,
      "step": 18898
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5823633670806885,
      "learning_rate": 5.721042742388882e-06,
      "loss": 0.5213,
      "step": 18899
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.697758436203003,
      "learning_rate": 5.720033578084216e-06,
      "loss": 0.4384,
      "step": 18900
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7995964288711548,
      "learning_rate": 5.719024467139503e-06,
      "loss": 0.5126,
      "step": 18901
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5242745876312256,
      "learning_rate": 5.718015409567318e-06,
      "loss": 0.4773,
      "step": 18902
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6110419034957886,
      "learning_rate": 5.717006405380238e-06,
      "loss": 0.4433,
      "step": 18903
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8053573369979858,
      "learning_rate": 5.715997454590849e-06,
      "loss": 0.4859,
      "step": 18904
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.464184045791626,
      "learning_rate": 5.7149885572117305e-06,
      "loss": 0.5391,
      "step": 18905
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8265314102172852,
      "learning_rate": 5.713979713255452e-06,
      "loss": 0.5543,
      "step": 18906
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3650201559066772,
      "learning_rate": 5.7129709227346016e-06,
      "loss": 0.5533,
      "step": 18907
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8728482723236084,
      "learning_rate": 5.711962185661752e-06,
      "loss": 0.5897,
      "step": 18908
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7236733436584473,
      "learning_rate": 5.710953502049475e-06,
      "loss": 0.6133,
      "step": 18909
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5698678493499756,
      "learning_rate": 5.7099448719103554e-06,
      "loss": 0.5722,
      "step": 18910
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7673574686050415,
      "learning_rate": 5.708936295256962e-06,
      "loss": 0.5439,
      "step": 18911
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5446258783340454,
      "learning_rate": 5.707927772101871e-06,
      "loss": 0.5527,
      "step": 18912
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7662750482559204,
      "learning_rate": 5.706919302457652e-06,
      "loss": 0.5817,
      "step": 18913
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5424467325210571,
      "learning_rate": 5.705910886336885e-06,
      "loss": 0.4866,
      "step": 18914
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5784801244735718,
      "learning_rate": 5.704902523752138e-06,
      "loss": 0.4953,
      "step": 18915
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3967255353927612,
      "learning_rate": 5.70389421471598e-06,
      "loss": 0.4348,
      "step": 18916
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6965138912200928,
      "learning_rate": 5.702885959240989e-06,
      "loss": 0.4705,
      "step": 18917
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5690394639968872,
      "learning_rate": 5.701877757339733e-06,
      "loss": 0.4979,
      "step": 18918
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8981672525405884,
      "learning_rate": 5.700869609024775e-06,
      "loss": 0.4036,
      "step": 18919
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6835945844650269,
      "learning_rate": 5.699861514308693e-06,
      "loss": 0.5843,
      "step": 18920
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.570218801498413,
      "learning_rate": 5.698853473204051e-06,
      "loss": 0.4844,
      "step": 18921
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.662132740020752,
      "learning_rate": 5.697845485723414e-06,
      "loss": 0.534,
      "step": 18922
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4312055110931396,
      "learning_rate": 5.696837551879357e-06,
      "loss": 0.5558,
      "step": 18923
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7225301265716553,
      "learning_rate": 5.695829671684441e-06,
      "loss": 0.5496,
      "step": 18924
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4555374383926392,
      "learning_rate": 5.6948218451512306e-06,
      "loss": 0.5,
      "step": 18925
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8154512643814087,
      "learning_rate": 5.693814072292289e-06,
      "loss": 0.5357,
      "step": 18926
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.743255853652954,
      "learning_rate": 5.692806353120187e-06,
      "loss": 0.4559,
      "step": 18927
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.472744107246399,
      "learning_rate": 5.691798687647485e-06,
      "loss": 0.477,
      "step": 18928
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6482585668563843,
      "learning_rate": 5.690791075886743e-06,
      "loss": 0.4793,
      "step": 18929
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7347173690795898,
      "learning_rate": 5.689783517850529e-06,
      "loss": 0.4802,
      "step": 18930
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7675130367279053,
      "learning_rate": 5.688776013551402e-06,
      "loss": 0.491,
      "step": 18931
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7132680416107178,
      "learning_rate": 5.687768563001919e-06,
      "loss": 0.4163,
      "step": 18932
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8925577402114868,
      "learning_rate": 5.686761166214647e-06,
      "loss": 0.5222,
      "step": 18933
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5660219192504883,
      "learning_rate": 5.6857538232021424e-06,
      "loss": 0.551,
      "step": 18934
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6288642883300781,
      "learning_rate": 5.684746533976961e-06,
      "loss": 0.4637,
      "step": 18935
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5985405445098877,
      "learning_rate": 5.683739298551667e-06,
      "loss": 0.4617,
      "step": 18936
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5961652994155884,
      "learning_rate": 5.6827321169388164e-06,
      "loss": 0.4233,
      "step": 18937
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8614171743392944,
      "learning_rate": 5.681724989150965e-06,
      "loss": 0.6729,
      "step": 18938
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7784677743911743,
      "learning_rate": 5.680717915200664e-06,
      "loss": 0.4842,
      "step": 18939
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.559495210647583,
      "learning_rate": 5.679710895100479e-06,
      "loss": 0.4951,
      "step": 18940
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7298307418823242,
      "learning_rate": 5.678703928862958e-06,
      "loss": 0.4952,
      "step": 18941
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.0283148288726807,
      "learning_rate": 5.677697016500656e-06,
      "loss": 0.5173,
      "step": 18942
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5387650728225708,
      "learning_rate": 5.67669015802613e-06,
      "loss": 0.5612,
      "step": 18943
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2740036249160767,
      "learning_rate": 5.675683353451931e-06,
      "loss": 0.3482,
      "step": 18944
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5217384099960327,
      "learning_rate": 5.674676602790605e-06,
      "loss": 0.4524,
      "step": 18945
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7741049528121948,
      "learning_rate": 5.673669906054715e-06,
      "loss": 0.5353,
      "step": 18946
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.566219449043274,
      "learning_rate": 5.672663263256805e-06,
      "loss": 0.4827,
      "step": 18947
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.517594575881958,
      "learning_rate": 5.671656674409422e-06,
      "loss": 0.3665,
      "step": 18948
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7046457529067993,
      "learning_rate": 5.670650139525126e-06,
      "loss": 0.4186,
      "step": 18949
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5485061407089233,
      "learning_rate": 5.669643658616458e-06,
      "loss": 0.4714,
      "step": 18950
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8174293041229248,
      "learning_rate": 5.668637231695968e-06,
      "loss": 0.4994,
      "step": 18951
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6515617370605469,
      "learning_rate": 5.6676308587761995e-06,
      "loss": 0.5337,
      "step": 18952
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.548134446144104,
      "learning_rate": 5.6666245398697074e-06,
      "loss": 0.4835,
      "step": 18953
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2350045442581177,
      "learning_rate": 5.665618274989033e-06,
      "loss": 0.3637,
      "step": 18954
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4589518308639526,
      "learning_rate": 5.664612064146719e-06,
      "loss": 0.4562,
      "step": 18955
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7978096008300781,
      "learning_rate": 5.663605907355318e-06,
      "loss": 0.5298,
      "step": 18956
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3324270248413086,
      "learning_rate": 5.662599804627369e-06,
      "loss": 0.4478,
      "step": 18957
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7080399990081787,
      "learning_rate": 5.661593755975412e-06,
      "loss": 0.5193,
      "step": 18958
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5016931295394897,
      "learning_rate": 5.660587761412e-06,
      "loss": 0.4389,
      "step": 18959
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6308481693267822,
      "learning_rate": 5.659581820949667e-06,
      "loss": 0.5137,
      "step": 18960
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.1023707389831543,
      "learning_rate": 5.658575934600953e-06,
      "loss": 0.6068,
      "step": 18961
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.0552377700805664,
      "learning_rate": 5.657570102378407e-06,
      "loss": 0.4996,
      "step": 18962
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.1207430362701416,
      "learning_rate": 5.656564324294563e-06,
      "loss": 0.564,
      "step": 18963
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7796542644500732,
      "learning_rate": 5.655558600361964e-06,
      "loss": 0.4627,
      "step": 18964
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.696291208267212,
      "learning_rate": 5.65455293059314e-06,
      "loss": 0.5587,
      "step": 18965
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5309288501739502,
      "learning_rate": 5.65354731500064e-06,
      "loss": 0.5718,
      "step": 18966
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.741984248161316,
      "learning_rate": 5.652541753596999e-06,
      "loss": 0.4657,
      "step": 18967
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4588125944137573,
      "learning_rate": 5.6515362463947466e-06,
      "loss": 0.5464,
      "step": 18968
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7716376781463623,
      "learning_rate": 5.650530793406428e-06,
      "loss": 0.5691,
      "step": 18969
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4917463064193726,
      "learning_rate": 5.6495253946445746e-06,
      "loss": 0.4605,
      "step": 18970
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.42253839969635,
      "learning_rate": 5.648520050121716e-06,
      "loss": 0.4851,
      "step": 18971
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3676254749298096,
      "learning_rate": 5.647514759850397e-06,
      "loss": 0.4175,
      "step": 18972
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5239135026931763,
      "learning_rate": 5.646509523843144e-06,
      "loss": 0.5718,
      "step": 18973
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.704281210899353,
      "learning_rate": 5.645504342112486e-06,
      "loss": 0.3773,
      "step": 18974
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7241648435592651,
      "learning_rate": 5.644499214670965e-06,
      "loss": 0.4323,
      "step": 18975
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5047987699508667,
      "learning_rate": 5.6434941415311075e-06,
      "loss": 0.477,
      "step": 18976
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6834266185760498,
      "learning_rate": 5.642489122705444e-06,
      "loss": 0.4475,
      "step": 18977
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.9910932779312134,
      "learning_rate": 5.641484158206499e-06,
      "loss": 0.4559,
      "step": 18978
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4096150398254395,
      "learning_rate": 5.640479248046811e-06,
      "loss": 0.4648,
      "step": 18979
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3861196041107178,
      "learning_rate": 5.639474392238904e-06,
      "loss": 0.4981,
      "step": 18980
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5909371376037598,
      "learning_rate": 5.6384695907953024e-06,
      "loss": 0.4278,
      "step": 18981
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5872548818588257,
      "learning_rate": 5.637464843728542e-06,
      "loss": 0.472,
      "step": 18982
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4932351112365723,
      "learning_rate": 5.636460151051144e-06,
      "loss": 0.4629,
      "step": 18983
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7233996391296387,
      "learning_rate": 5.635455512775636e-06,
      "loss": 0.4208,
      "step": 18984
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4448400735855103,
      "learning_rate": 5.634450928914543e-06,
      "loss": 0.5536,
      "step": 18985
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6974098682403564,
      "learning_rate": 5.6334463994803875e-06,
      "loss": 0.5126,
      "step": 18986
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6489304304122925,
      "learning_rate": 5.632441924485692e-06,
      "loss": 0.5987,
      "step": 18987
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6213772296905518,
      "learning_rate": 5.631437503942985e-06,
      "loss": 0.5269,
      "step": 18988
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4283041954040527,
      "learning_rate": 5.630433137864787e-06,
      "loss": 0.5368,
      "step": 18989
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6057827472686768,
      "learning_rate": 5.6294288262636144e-06,
      "loss": 0.5192,
      "step": 18990
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.53158438205719,
      "learning_rate": 5.628424569151998e-06,
      "loss": 0.453,
      "step": 18991
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.701587438583374,
      "learning_rate": 5.6274203665424535e-06,
      "loss": 0.4913,
      "step": 18992
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5584206581115723,
      "learning_rate": 5.626416218447495e-06,
      "loss": 0.5237,
      "step": 18993
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6566089391708374,
      "learning_rate": 5.625412124879652e-06,
      "loss": 0.5762,
      "step": 18994
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6648633480072021,
      "learning_rate": 5.624408085851439e-06,
      "loss": 0.5706,
      "step": 18995
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6236926317214966,
      "learning_rate": 5.6234041013753675e-06,
      "loss": 0.4873,
      "step": 18996
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3041503429412842,
      "learning_rate": 5.6224001714639645e-06,
      "loss": 0.4695,
      "step": 18997
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.8244034051895142,
      "learning_rate": 5.621396296129739e-06,
      "loss": 0.6396,
      "step": 18998
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.667293906211853,
      "learning_rate": 5.620392475385212e-06,
      "loss": 0.4326,
      "step": 18999
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4847840070724487,
      "learning_rate": 5.619388709242892e-06,
      "loss": 0.405,
      "step": 19000
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.609462857246399,
      "learning_rate": 5.618384997715301e-06,
      "loss": 0.4175,
      "step": 19001
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6031746864318848,
      "learning_rate": 5.617381340814947e-06,
      "loss": 0.5694,
      "step": 19002
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.691291332244873,
      "learning_rate": 5.616377738554343e-06,
      "loss": 0.4778,
      "step": 19003
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7781636714935303,
      "learning_rate": 5.615374190946006e-06,
      "loss": 0.5563,
      "step": 19004
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4411653280258179,
      "learning_rate": 5.614370698002445e-06,
      "loss": 0.392,
      "step": 19005
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7265089750289917,
      "learning_rate": 5.613367259736167e-06,
      "loss": 0.4643,
      "step": 19006
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.324773907661438,
      "learning_rate": 5.612363876159687e-06,
      "loss": 0.4239,
      "step": 19007
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.613253116607666,
      "learning_rate": 5.611360547285516e-06,
      "loss": 0.4906,
      "step": 19008
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9114292860031128,
      "learning_rate": 5.610357273126157e-06,
      "loss": 0.5653,
      "step": 19009
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.536231279373169,
      "learning_rate": 5.609354053694124e-06,
      "loss": 0.5568,
      "step": 19010
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4502801895141602,
      "learning_rate": 5.6083508890019215e-06,
      "loss": 0.4837,
      "step": 19011
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7360950708389282,
      "learning_rate": 5.607347779062058e-06,
      "loss": 0.4326,
      "step": 19012
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7590547800064087,
      "learning_rate": 5.606344723887034e-06,
      "loss": 0.5798,
      "step": 19013
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.805344581604004,
      "learning_rate": 5.605341723489363e-06,
      "loss": 0.6785,
      "step": 19014
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6248213052749634,
      "learning_rate": 5.604338777881547e-06,
      "loss": 0.5134,
      "step": 19015
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5100947618484497,
      "learning_rate": 5.603335887076084e-06,
      "loss": 0.5197,
      "step": 19016
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5368844270706177,
      "learning_rate": 5.602333051085489e-06,
      "loss": 0.4639,
      "step": 19017
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4385571479797363,
      "learning_rate": 5.601330269922257e-06,
      "loss": 0.4222,
      "step": 19018
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9785972833633423,
      "learning_rate": 5.600327543598888e-06,
      "loss": 0.5249,
      "step": 19019
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2843681573867798,
      "learning_rate": 5.5993248721278905e-06,
      "loss": 0.4432,
      "step": 19020
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9836807250976562,
      "learning_rate": 5.598322255521762e-06,
      "loss": 0.5605,
      "step": 19021
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6605137586593628,
      "learning_rate": 5.597319693792997e-06,
      "loss": 0.5063,
      "step": 19022
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7199230194091797,
      "learning_rate": 5.596317186954106e-06,
      "loss": 0.5898,
      "step": 19023
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4322623014450073,
      "learning_rate": 5.595314735017579e-06,
      "loss": 0.5269,
      "step": 19024
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7405633926391602,
      "learning_rate": 5.594312337995918e-06,
      "loss": 0.596,
      "step": 19025
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6920336484909058,
      "learning_rate": 5.5933099959016135e-06,
      "loss": 0.4388,
      "step": 19026
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.55344557762146,
      "learning_rate": 5.592307708747171e-06,
      "loss": 0.4478,
      "step": 19027
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.788880705833435,
      "learning_rate": 5.591305476545084e-06,
      "loss": 0.6506,
      "step": 19028
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3612867593765259,
      "learning_rate": 5.590303299307842e-06,
      "loss": 0.4617,
      "step": 19029
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.1027064323425293,
      "learning_rate": 5.5893011770479475e-06,
      "loss": 0.4892,
      "step": 19030
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7029104232788086,
      "learning_rate": 5.588299109777892e-06,
      "loss": 0.5044,
      "step": 19031
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.638060450553894,
      "learning_rate": 5.587297097510163e-06,
      "loss": 0.5037,
      "step": 19032
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.1896328926086426,
      "learning_rate": 5.58629514025726e-06,
      "loss": 0.535,
      "step": 19033
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6122206449508667,
      "learning_rate": 5.585293238031673e-06,
      "loss": 0.4529,
      "step": 19034
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.622741937637329,
      "learning_rate": 5.584291390845888e-06,
      "loss": 0.6293,
      "step": 19035
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7967662811279297,
      "learning_rate": 5.583289598712404e-06,
      "loss": 0.5192,
      "step": 19036
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.664892315864563,
      "learning_rate": 5.582287861643705e-06,
      "loss": 0.5452,
      "step": 19037
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3022825717926025,
      "learning_rate": 5.581286179652283e-06,
      "loss": 0.5156,
      "step": 19038
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7361894845962524,
      "learning_rate": 5.58028455275062e-06,
      "loss": 0.562,
      "step": 19039
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9993866682052612,
      "learning_rate": 5.579282980951213e-06,
      "loss": 0.5412,
      "step": 19040
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7110648155212402,
      "learning_rate": 5.578281464266543e-06,
      "loss": 0.5003,
      "step": 19041
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4946165084838867,
      "learning_rate": 5.5772800027090945e-06,
      "loss": 0.5276,
      "step": 19042
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.70534086227417,
      "learning_rate": 5.57627859629136e-06,
      "loss": 0.5783,
      "step": 19043
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5764628648757935,
      "learning_rate": 5.575277245025821e-06,
      "loss": 0.3813,
      "step": 19044
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.591460943222046,
      "learning_rate": 5.574275948924956e-06,
      "loss": 0.379,
      "step": 19045
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7888084650039673,
      "learning_rate": 5.573274708001259e-06,
      "loss": 0.5447,
      "step": 19046
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7317180633544922,
      "learning_rate": 5.572273522267209e-06,
      "loss": 0.4725,
      "step": 19047
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6133445501327515,
      "learning_rate": 5.57127239173528e-06,
      "loss": 0.5679,
      "step": 19048
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.413755178451538,
      "learning_rate": 5.570271316417967e-06,
      "loss": 0.5305,
      "step": 19049
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.874586582183838,
      "learning_rate": 5.5692702963277425e-06,
      "loss": 0.5631,
      "step": 19050
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3628596067428589,
      "learning_rate": 5.56826933147709e-06,
      "loss": 0.485,
      "step": 19051
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4847151041030884,
      "learning_rate": 5.567268421878481e-06,
      "loss": 0.4895,
      "step": 19052
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6590886116027832,
      "learning_rate": 5.566267567544405e-06,
      "loss": 0.4118,
      "step": 19053
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8110525608062744,
      "learning_rate": 5.565266768487337e-06,
      "loss": 0.4636,
      "step": 19054
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5402491092681885,
      "learning_rate": 5.564266024719747e-06,
      "loss": 0.5323,
      "step": 19055
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.589232325553894,
      "learning_rate": 5.563265336254121e-06,
      "loss": 0.4861,
      "step": 19056
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4580714702606201,
      "learning_rate": 5.562264703102932e-06,
      "loss": 0.535,
      "step": 19057
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6497342586517334,
      "learning_rate": 5.561264125278652e-06,
      "loss": 0.5436,
      "step": 19058
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7726887464523315,
      "learning_rate": 5.5602636027937595e-06,
      "loss": 0.4012,
      "step": 19059
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6862808465957642,
      "learning_rate": 5.559263135660728e-06,
      "loss": 0.5582,
      "step": 19060
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6910713911056519,
      "learning_rate": 5.558262723892026e-06,
      "loss": 0.5886,
      "step": 19061
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.543924331665039,
      "learning_rate": 5.557262367500134e-06,
      "loss": 0.4624,
      "step": 19062
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8426034450531006,
      "learning_rate": 5.55626206649752e-06,
      "loss": 0.4742,
      "step": 19063
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3731426000595093,
      "learning_rate": 5.555261820896654e-06,
      "loss": 0.4475,
      "step": 19064
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7400776147842407,
      "learning_rate": 5.554261630710003e-06,
      "loss": 0.551,
      "step": 19065
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4291340112686157,
      "learning_rate": 5.553261495950045e-06,
      "loss": 0.5779,
      "step": 19066
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.672755241394043,
      "learning_rate": 5.5522614166292455e-06,
      "loss": 0.4728,
      "step": 19067
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.716630220413208,
      "learning_rate": 5.5512613927600675e-06,
      "loss": 0.4972,
      "step": 19068
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6243512630462646,
      "learning_rate": 5.550261424354988e-06,
      "loss": 0.4803,
      "step": 19069
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5376267433166504,
      "learning_rate": 5.5492615114264694e-06,
      "loss": 0.5408,
      "step": 19070
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5746641159057617,
      "learning_rate": 5.548261653986975e-06,
      "loss": 0.4547,
      "step": 19071
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.499817132949829,
      "learning_rate": 5.547261852048977e-06,
      "loss": 0.4416,
      "step": 19072
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5852748155593872,
      "learning_rate": 5.546262105624936e-06,
      "loss": 0.5349,
      "step": 19073
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8264799118041992,
      "learning_rate": 5.545262414727313e-06,
      "loss": 0.5553,
      "step": 19074
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6002334356307983,
      "learning_rate": 5.544262779368579e-06,
      "loss": 0.5073,
      "step": 19075
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8769656419754028,
      "learning_rate": 5.543263199561195e-06,
      "loss": 0.4821,
      "step": 19076
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4020427465438843,
      "learning_rate": 5.54226367531762e-06,
      "loss": 0.5522,
      "step": 19077
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.611116647720337,
      "learning_rate": 5.541264206650314e-06,
      "loss": 0.427,
      "step": 19078
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4992389678955078,
      "learning_rate": 5.540264793571744e-06,
      "loss": 0.435,
      "step": 19079
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5406302213668823,
      "learning_rate": 5.539265436094368e-06,
      "loss": 0.4915,
      "step": 19080
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6093251705169678,
      "learning_rate": 5.538266134230639e-06,
      "loss": 0.4329,
      "step": 19081
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9131735563278198,
      "learning_rate": 5.5372668879930235e-06,
      "loss": 0.5457,
      "step": 19082
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.0153796672821045,
      "learning_rate": 5.536267697393978e-06,
      "loss": 0.5362,
      "step": 19083
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4493426084518433,
      "learning_rate": 5.535268562445955e-06,
      "loss": 0.572,
      "step": 19084
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6298619508743286,
      "learning_rate": 5.534269483161418e-06,
      "loss": 0.4518,
      "step": 19085
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7442448139190674,
      "learning_rate": 5.53327045955282e-06,
      "loss": 0.4781,
      "step": 19086
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.557955026626587,
      "learning_rate": 5.53227149163261e-06,
      "loss": 0.4201,
      "step": 19087
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6454780101776123,
      "learning_rate": 5.5312725794132534e-06,
      "loss": 0.4815,
      "step": 19088
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.882946252822876,
      "learning_rate": 5.5302737229072e-06,
      "loss": 0.4,
      "step": 19089
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8543174266815186,
      "learning_rate": 5.529274922126901e-06,
      "loss": 0.5561,
      "step": 19090
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3986576795578003,
      "learning_rate": 5.5282761770848045e-06,
      "loss": 0.4551,
      "step": 19091
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8436354398727417,
      "learning_rate": 5.527277487793372e-06,
      "loss": 0.4809,
      "step": 19092
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.8123611211776733,
      "learning_rate": 5.526278854265051e-06,
      "loss": 0.6042,
      "step": 19093
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5103029012680054,
      "learning_rate": 5.525280276512285e-06,
      "loss": 0.4337,
      "step": 19094
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.774756908416748,
      "learning_rate": 5.524281754547533e-06,
      "loss": 0.5286,
      "step": 19095
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5419542789459229,
      "learning_rate": 5.523283288383241e-06,
      "loss": 0.6086,
      "step": 19096
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8633427619934082,
      "learning_rate": 5.5222848780318536e-06,
      "loss": 0.547,
      "step": 19097
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4780019521713257,
      "learning_rate": 5.521286523505825e-06,
      "loss": 0.5141,
      "step": 19098
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7187420129776,
      "learning_rate": 5.520288224817597e-06,
      "loss": 0.524,
      "step": 19099
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.871317744255066,
      "learning_rate": 5.519289981979615e-06,
      "loss": 0.4726,
      "step": 19100
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9480782747268677,
      "learning_rate": 5.518291795004328e-06,
      "loss": 0.5541,
      "step": 19101
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6326550245285034,
      "learning_rate": 5.517293663904181e-06,
      "loss": 0.4897,
      "step": 19102
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8037830591201782,
      "learning_rate": 5.516295588691616e-06,
      "loss": 0.4538,
      "step": 19103
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7682602405548096,
      "learning_rate": 5.515297569379072e-06,
      "loss": 0.463,
      "step": 19104
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8859058618545532,
      "learning_rate": 5.514299605979002e-06,
      "loss": 0.6409,
      "step": 19105
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5874247550964355,
      "learning_rate": 5.513301698503841e-06,
      "loss": 0.5946,
      "step": 19106
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.535530686378479,
      "learning_rate": 5.5123038469660296e-06,
      "loss": 0.4531,
      "step": 19107
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7901912927627563,
      "learning_rate": 5.511306051378014e-06,
      "loss": 0.5578,
      "step": 19108
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9331555366516113,
      "learning_rate": 5.510308311752229e-06,
      "loss": 0.6169,
      "step": 19109
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8238483667373657,
      "learning_rate": 5.509310628101114e-06,
      "loss": 0.5618,
      "step": 19110
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.921939730644226,
      "learning_rate": 5.508313000437111e-06,
      "loss": 0.5854,
      "step": 19111
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8737996816635132,
      "learning_rate": 5.507315428772656e-06,
      "loss": 0.4367,
      "step": 19112
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3899083137512207,
      "learning_rate": 5.5063179131201815e-06,
      "loss": 0.5793,
      "step": 19113
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7083147764205933,
      "learning_rate": 5.505320453492134e-06,
      "loss": 0.517,
      "step": 19114
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6724153757095337,
      "learning_rate": 5.5043230499009416e-06,
      "loss": 0.4697,
      "step": 19115
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6191946268081665,
      "learning_rate": 5.503325702359042e-06,
      "loss": 0.5336,
      "step": 19116
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6707381010055542,
      "learning_rate": 5.502328410878863e-06,
      "loss": 0.3718,
      "step": 19117
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4675995111465454,
      "learning_rate": 5.5013311754728505e-06,
      "loss": 0.6254,
      "step": 19118
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4972538948059082,
      "learning_rate": 5.500333996153429e-06,
      "loss": 0.5799,
      "step": 19119
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6277672052383423,
      "learning_rate": 5.499336872933029e-06,
      "loss": 0.5672,
      "step": 19120
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5311750173568726,
      "learning_rate": 5.498339805824089e-06,
      "loss": 0.5224,
      "step": 19121
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9467052221298218,
      "learning_rate": 5.497342794839037e-06,
      "loss": 0.4206,
      "step": 19122
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3374594449996948,
      "learning_rate": 5.496345839990298e-06,
      "loss": 0.4981,
      "step": 19123
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6290363073349,
      "learning_rate": 5.49534894129031e-06,
      "loss": 0.5309,
      "step": 19124
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.54652738571167,
      "learning_rate": 5.494352098751496e-06,
      "loss": 0.4603,
      "step": 19125
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.666723608970642,
      "learning_rate": 5.493355312386284e-06,
      "loss": 0.4894,
      "step": 19126
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3194174766540527,
      "learning_rate": 5.492358582207105e-06,
      "loss": 0.5036,
      "step": 19127
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5565956830978394,
      "learning_rate": 5.491361908226385e-06,
      "loss": 0.4301,
      "step": 19128
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7024260759353638,
      "learning_rate": 5.4903652904565474e-06,
      "loss": 0.4799,
      "step": 19129
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4186006784439087,
      "learning_rate": 5.489368728910014e-06,
      "loss": 0.5645,
      "step": 19130
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6247878074645996,
      "learning_rate": 5.488372223599218e-06,
      "loss": 0.4614,
      "step": 19131
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4587981700897217,
      "learning_rate": 5.487375774536578e-06,
      "loss": 0.4698,
      "step": 19132
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.80910062789917,
      "learning_rate": 5.486379381734516e-06,
      "loss": 0.6526,
      "step": 19133
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4014670848846436,
      "learning_rate": 5.485383045205459e-06,
      "loss": 0.4196,
      "step": 19134
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2788560390472412,
      "learning_rate": 5.4843867649618265e-06,
      "loss": 0.4289,
      "step": 19135
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9136229753494263,
      "learning_rate": 5.4833905410160345e-06,
      "loss": 0.507,
      "step": 19136
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.486575722694397,
      "learning_rate": 5.482394373380513e-06,
      "loss": 0.608,
      "step": 19137
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.74774169921875,
      "learning_rate": 5.481398262067675e-06,
      "loss": 0.5104,
      "step": 19138
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4905468225479126,
      "learning_rate": 5.480402207089937e-06,
      "loss": 0.4068,
      "step": 19139
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7269495725631714,
      "learning_rate": 5.479406208459726e-06,
      "loss": 0.4458,
      "step": 19140
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.656863808631897,
      "learning_rate": 5.478410266189455e-06,
      "loss": 0.5093,
      "step": 19141
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5166490077972412,
      "learning_rate": 5.477414380291539e-06,
      "loss": 0.4889,
      "step": 19142
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.313191294670105,
      "learning_rate": 5.476418550778393e-06,
      "loss": 0.418,
      "step": 19143
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7061775922775269,
      "learning_rate": 5.4754227776624366e-06,
      "loss": 0.5265,
      "step": 19144
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6879985332489014,
      "learning_rate": 5.474427060956084e-06,
      "loss": 0.547,
      "step": 19145
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8056926727294922,
      "learning_rate": 5.473431400671744e-06,
      "loss": 0.5335,
      "step": 19146
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5973031520843506,
      "learning_rate": 5.472435796821838e-06,
      "loss": 0.5471,
      "step": 19147
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7271417379379272,
      "learning_rate": 5.4714402494187735e-06,
      "loss": 0.5734,
      "step": 19148
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7687313556671143,
      "learning_rate": 5.470444758474959e-06,
      "loss": 0.4973,
      "step": 19149
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.301655650138855,
      "learning_rate": 5.469449324002813e-06,
      "loss": 0.5065,
      "step": 19150
    },
    {
      "epoch": 1.98,
      "grad_norm": 2.341494083404541,
      "learning_rate": 5.468453946014744e-06,
      "loss": 0.5364,
      "step": 19151
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3122918605804443,
      "learning_rate": 5.467458624523157e-06,
      "loss": 0.4346,
      "step": 19152
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.509295105934143,
      "learning_rate": 5.466463359540468e-06,
      "loss": 0.5048,
      "step": 19153
    },
    {
      "epoch": 1.98,
      "grad_norm": 2.008495330810547,
      "learning_rate": 5.465468151079082e-06,
      "loss": 0.6334,
      "step": 19154
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.782629132270813,
      "learning_rate": 5.464472999151407e-06,
      "loss": 0.4816,
      "step": 19155
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5263628959655762,
      "learning_rate": 5.463477903769844e-06,
      "loss": 0.5665,
      "step": 19156
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5099564790725708,
      "learning_rate": 5.4624828649468095e-06,
      "loss": 0.5638,
      "step": 19157
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.579620361328125,
      "learning_rate": 5.461487882694705e-06,
      "loss": 0.5406,
      "step": 19158
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5074487924575806,
      "learning_rate": 5.460492957025929e-06,
      "loss": 0.4697,
      "step": 19159
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7643892765045166,
      "learning_rate": 5.4594980879528955e-06,
      "loss": 0.4099,
      "step": 19160
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5792016983032227,
      "learning_rate": 5.458503275488004e-06,
      "loss": 0.4336,
      "step": 19161
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.542088508605957,
      "learning_rate": 5.457508519643653e-06,
      "loss": 0.497,
      "step": 19162
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5769323110580444,
      "learning_rate": 5.456513820432252e-06,
      "loss": 0.5431,
      "step": 19163
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9829126596450806,
      "learning_rate": 5.455519177866199e-06,
      "loss": 0.507,
      "step": 19164
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.615315318107605,
      "learning_rate": 5.454524591957889e-06,
      "loss": 0.5327,
      "step": 19165
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4263159036636353,
      "learning_rate": 5.453530062719732e-06,
      "loss": 0.5348,
      "step": 19166
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.697152853012085,
      "learning_rate": 5.452535590164121e-06,
      "loss": 0.4817,
      "step": 19167
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8578517436981201,
      "learning_rate": 5.451541174303456e-06,
      "loss": 0.5591,
      "step": 19168
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.375942349433899,
      "learning_rate": 5.450546815150131e-06,
      "loss": 0.5791,
      "step": 19169
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3463609218597412,
      "learning_rate": 5.449552512716552e-06,
      "loss": 0.5724,
      "step": 19170
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.715728759765625,
      "learning_rate": 5.448558267015107e-06,
      "loss": 0.4986,
      "step": 19171
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5241786241531372,
      "learning_rate": 5.447564078058192e-06,
      "loss": 0.601,
      "step": 19172
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5987677574157715,
      "learning_rate": 5.446569945858206e-06,
      "loss": 0.5047,
      "step": 19173
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7009875774383545,
      "learning_rate": 5.445575870427544e-06,
      "loss": 0.5552,
      "step": 19174
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.670732021331787,
      "learning_rate": 5.444581851778592e-06,
      "loss": 0.537,
      "step": 19175
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8699028491973877,
      "learning_rate": 5.443587889923752e-06,
      "loss": 0.5004,
      "step": 19176
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6198540925979614,
      "learning_rate": 5.442593984875413e-06,
      "loss": 0.4437,
      "step": 19177
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6340689659118652,
      "learning_rate": 5.441600136645961e-06,
      "loss": 0.5222,
      "step": 19178
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5172975063323975,
      "learning_rate": 5.440606345247794e-06,
      "loss": 0.4502,
      "step": 19179
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7887003421783447,
      "learning_rate": 5.439612610693299e-06,
      "loss": 0.5164,
      "step": 19180
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3517438173294067,
      "learning_rate": 5.438618932994866e-06,
      "loss": 0.4722,
      "step": 19181
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5929772853851318,
      "learning_rate": 5.437625312164879e-06,
      "loss": 0.4779,
      "step": 19182
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.57509183883667,
      "learning_rate": 5.4366317482157335e-06,
      "loss": 0.5204,
      "step": 19183
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4741737842559814,
      "learning_rate": 5.435638241159813e-06,
      "loss": 0.5023,
      "step": 19184
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5004609823226929,
      "learning_rate": 5.4346447910094995e-06,
      "loss": 0.4761,
      "step": 19185
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4872548580169678,
      "learning_rate": 5.4336513977771866e-06,
      "loss": 0.4889,
      "step": 19186
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.885420560836792,
      "learning_rate": 5.432658061475256e-06,
      "loss": 0.7014,
      "step": 19187
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.484750509262085,
      "learning_rate": 5.431664782116086e-06,
      "loss": 0.5199,
      "step": 19188
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7636873722076416,
      "learning_rate": 5.430671559712072e-06,
      "loss": 0.485,
      "step": 19189
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.9645962715148926,
      "learning_rate": 5.429678394275591e-06,
      "loss": 0.4731,
      "step": 19190
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2734464406967163,
      "learning_rate": 5.428685285819018e-06,
      "loss": 0.433,
      "step": 19191
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5284606218338013,
      "learning_rate": 5.427692234354748e-06,
      "loss": 0.4114,
      "step": 19192
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.556829571723938,
      "learning_rate": 5.426699239895155e-06,
      "loss": 0.5097,
      "step": 19193
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7213799953460693,
      "learning_rate": 5.425706302452618e-06,
      "loss": 0.4845,
      "step": 19194
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6216970682144165,
      "learning_rate": 5.424713422039514e-06,
      "loss": 0.6679,
      "step": 19195
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5742623805999756,
      "learning_rate": 5.4237205986682275e-06,
      "loss": 0.426,
      "step": 19196
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6054824590682983,
      "learning_rate": 5.422727832351136e-06,
      "loss": 0.5057,
      "step": 19197
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.9304797649383545,
      "learning_rate": 5.421735123100609e-06,
      "loss": 0.5079,
      "step": 19198
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7184991836547852,
      "learning_rate": 5.420742470929035e-06,
      "loss": 0.4674,
      "step": 19199
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5925382375717163,
      "learning_rate": 5.419749875848781e-06,
      "loss": 0.5263,
      "step": 19200
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8386286497116089,
      "learning_rate": 5.4187573378722206e-06,
      "loss": 0.468,
      "step": 19201
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5221362113952637,
      "learning_rate": 5.417764857011737e-06,
      "loss": 0.5151,
      "step": 19202
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7266733646392822,
      "learning_rate": 5.416772433279698e-06,
      "loss": 0.4847,
      "step": 19203
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.256931185722351,
      "learning_rate": 5.4157800666884745e-06,
      "loss": 0.4206,
      "step": 19204
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5327776670455933,
      "learning_rate": 5.4147877572504445e-06,
      "loss": 0.4467,
      "step": 19205
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.031808614730835,
      "learning_rate": 5.413795504977977e-06,
      "loss": 0.6105,
      "step": 19206
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.467703938484192,
      "learning_rate": 5.412803309883442e-06,
      "loss": 0.3938,
      "step": 19207
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0810887813568115,
      "learning_rate": 5.411811171979207e-06,
      "loss": 0.5557,
      "step": 19208
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.902409315109253,
      "learning_rate": 5.410819091277646e-06,
      "loss": 0.4666,
      "step": 19209
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6458219289779663,
      "learning_rate": 5.4098270677911265e-06,
      "loss": 0.5246,
      "step": 19210
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4691296815872192,
      "learning_rate": 5.408835101532013e-06,
      "loss": 0.414,
      "step": 19211
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.9280693531036377,
      "learning_rate": 5.4078431925126795e-06,
      "loss": 0.4271,
      "step": 19212
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7908954620361328,
      "learning_rate": 5.406851340745487e-06,
      "loss": 0.4669,
      "step": 19213
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4528193473815918,
      "learning_rate": 5.4058595462428e-06,
      "loss": 0.4613,
      "step": 19214
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2671887874603271,
      "learning_rate": 5.404867809016992e-06,
      "loss": 0.5141,
      "step": 19215
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6546504497528076,
      "learning_rate": 5.403876129080418e-06,
      "loss": 0.5451,
      "step": 19216
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5615980625152588,
      "learning_rate": 5.4028845064454435e-06,
      "loss": 0.4414,
      "step": 19217
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.480386257171631,
      "learning_rate": 5.401892941124438e-06,
      "loss": 0.4606,
      "step": 19218
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4363080263137817,
      "learning_rate": 5.400901433129757e-06,
      "loss": 0.4984,
      "step": 19219
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5156546831130981,
      "learning_rate": 5.399909982473766e-06,
      "loss": 0.4961,
      "step": 19220
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7002652883529663,
      "learning_rate": 5.398918589168819e-06,
      "loss": 0.5178,
      "step": 19221
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.698434829711914,
      "learning_rate": 5.3979272532272844e-06,
      "loss": 0.4666,
      "step": 19222
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6285234689712524,
      "learning_rate": 5.396935974661519e-06,
      "loss": 0.5736,
      "step": 19223
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7362529039382935,
      "learning_rate": 5.395944753483876e-06,
      "loss": 0.5376,
      "step": 19224
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5617486238479614,
      "learning_rate": 5.394953589706722e-06,
      "loss": 0.5644,
      "step": 19225
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8206087350845337,
      "learning_rate": 5.39396248334241e-06,
      "loss": 0.5616,
      "step": 19226
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.423213243484497,
      "learning_rate": 5.392971434403292e-06,
      "loss": 0.5343,
      "step": 19227
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8712115287780762,
      "learning_rate": 5.3919804429017345e-06,
      "loss": 0.5555,
      "step": 19228
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4808024168014526,
      "learning_rate": 5.390989508850084e-06,
      "loss": 0.4949,
      "step": 19229
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5753796100616455,
      "learning_rate": 5.389998632260693e-06,
      "loss": 0.5308,
      "step": 19230
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.714613437652588,
      "learning_rate": 5.389007813145926e-06,
      "loss": 0.5289,
      "step": 19231
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6472917795181274,
      "learning_rate": 5.388017051518126e-06,
      "loss": 0.5576,
      "step": 19232
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.716764211654663,
      "learning_rate": 5.38702634738965e-06,
      "loss": 0.4909,
      "step": 19233
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3223272562026978,
      "learning_rate": 5.386035700772845e-06,
      "loss": 0.4817,
      "step": 19234
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.1012275218963623,
      "learning_rate": 5.3850451116800675e-06,
      "loss": 0.5816,
      "step": 19235
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4956302642822266,
      "learning_rate": 5.384054580123664e-06,
      "loss": 0.6006,
      "step": 19236
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0768110752105713,
      "learning_rate": 5.383064106115982e-06,
      "loss": 0.4926,
      "step": 19237
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4690924882888794,
      "learning_rate": 5.382073689669376e-06,
      "loss": 0.5127,
      "step": 19238
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5467809438705444,
      "learning_rate": 5.381083330796191e-06,
      "loss": 0.4717,
      "step": 19239
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.9183026552200317,
      "learning_rate": 5.380093029508769e-06,
      "loss": 0.6048,
      "step": 19240
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3696649074554443,
      "learning_rate": 5.3791027858194655e-06,
      "loss": 0.402,
      "step": 19241
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.349289894104004,
      "learning_rate": 5.378112599740623e-06,
      "loss": 0.5939,
      "step": 19242
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5619583129882812,
      "learning_rate": 5.377122471284581e-06,
      "loss": 0.4111,
      "step": 19243
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7170690298080444,
      "learning_rate": 5.376132400463692e-06,
      "loss": 0.413,
      "step": 19244
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6573448181152344,
      "learning_rate": 5.375142387290295e-06,
      "loss": 0.4925,
      "step": 19245
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.590085744857788,
      "learning_rate": 5.374152431776734e-06,
      "loss": 0.61,
      "step": 19246
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7325389385223389,
      "learning_rate": 5.373162533935348e-06,
      "loss": 0.4941,
      "step": 19247
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.559198021888733,
      "learning_rate": 5.372172693778484e-06,
      "loss": 0.4836,
      "step": 19248
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.018860340118408,
      "learning_rate": 5.371182911318482e-06,
      "loss": 0.5737,
      "step": 19249
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0528101921081543,
      "learning_rate": 5.370193186567673e-06,
      "loss": 0.5106,
      "step": 19250
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5493215322494507,
      "learning_rate": 5.3692035195384105e-06,
      "loss": 0.534,
      "step": 19251
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7785847187042236,
      "learning_rate": 5.368213910243024e-06,
      "loss": 0.4856,
      "step": 19252
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.9531476497650146,
      "learning_rate": 5.36722435869385e-06,
      "loss": 0.5082,
      "step": 19253
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5889554023742676,
      "learning_rate": 5.3662348649032325e-06,
      "loss": 0.4693,
      "step": 19254
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7504045963287354,
      "learning_rate": 5.365245428883504e-06,
      "loss": 0.4211,
      "step": 19255
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5394314527511597,
      "learning_rate": 5.364256050646997e-06,
      "loss": 0.4604,
      "step": 19256
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.583687424659729,
      "learning_rate": 5.363266730206051e-06,
      "loss": 0.5534,
      "step": 19257
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.9056475162506104,
      "learning_rate": 5.3622774675730025e-06,
      "loss": 0.4931,
      "step": 19258
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2562525272369385,
      "learning_rate": 5.3612882627601805e-06,
      "loss": 0.4463,
      "step": 19259
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.0896472930908203,
      "learning_rate": 5.3602991157799145e-06,
      "loss": 0.5602,
      "step": 19260
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3810080289840698,
      "learning_rate": 5.359310026644544e-06,
      "loss": 0.453,
      "step": 19261
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6120413541793823,
      "learning_rate": 5.358320995366397e-06,
      "loss": 0.4594,
      "step": 19262
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6874867677688599,
      "learning_rate": 5.3573320219578015e-06,
      "loss": 0.5862,
      "step": 19263
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8587826490402222,
      "learning_rate": 5.356343106431094e-06,
      "loss": 0.5391,
      "step": 19264
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8451913595199585,
      "learning_rate": 5.355354248798601e-06,
      "loss": 0.5078,
      "step": 19265
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.771916151046753,
      "learning_rate": 5.354365449072645e-06,
      "loss": 0.5426,
      "step": 19266
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7993155717849731,
      "learning_rate": 5.353376707265563e-06,
      "loss": 0.4816,
      "step": 19267
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5990126132965088,
      "learning_rate": 5.352388023389677e-06,
      "loss": 0.5153,
      "step": 19268
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6421605348587036,
      "learning_rate": 5.35139939745731e-06,
      "loss": 0.5761,
      "step": 19269
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6808350086212158,
      "learning_rate": 5.350410829480795e-06,
      "loss": 0.516,
      "step": 19270
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.691627025604248,
      "learning_rate": 5.349422319472455e-06,
      "loss": 0.5463,
      "step": 19271
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2690738439559937,
      "learning_rate": 5.3484338674446114e-06,
      "loss": 0.5258,
      "step": 19272
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.913933277130127,
      "learning_rate": 5.3474454734095836e-06,
      "loss": 0.6052,
      "step": 19273
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3565417528152466,
      "learning_rate": 5.346457137379705e-06,
      "loss": 0.5637,
      "step": 19274
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3556498289108276,
      "learning_rate": 5.345468859367291e-06,
      "loss": 0.4467,
      "step": 19275
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.7337498664855957,
      "learning_rate": 5.3444806393846605e-06,
      "loss": 0.493,
      "step": 19276
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.546311855316162,
      "learning_rate": 5.343492477444141e-06,
      "loss": 0.5371,
      "step": 19277
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5197703838348389,
      "learning_rate": 5.342504373558049e-06,
      "loss": 0.444,
      "step": 19278
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.639312982559204,
      "learning_rate": 5.341516327738698e-06,
      "loss": 0.532,
      "step": 19279
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6623728275299072,
      "learning_rate": 5.340528339998415e-06,
      "loss": 0.6457,
      "step": 19280
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.509076714515686,
      "learning_rate": 5.3395404103495156e-06,
      "loss": 0.5671,
      "step": 19281
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3115943670272827,
      "learning_rate": 5.33855253880431e-06,
      "loss": 0.4595,
      "step": 19282
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.777591586112976,
      "learning_rate": 5.337564725375124e-06,
      "loss": 0.4732,
      "step": 19283
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.701385498046875,
      "learning_rate": 5.3365769700742695e-06,
      "loss": 0.4583,
      "step": 19284
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4647455215454102,
      "learning_rate": 5.335589272914059e-06,
      "loss": 0.4207,
      "step": 19285
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.552425742149353,
      "learning_rate": 5.334601633906804e-06,
      "loss": 0.3797,
      "step": 19286
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.753374695777893,
      "learning_rate": 5.333614053064826e-06,
      "loss": 0.5835,
      "step": 19287
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.726517677307129,
      "learning_rate": 5.332626530400433e-06,
      "loss": 0.4547,
      "step": 19288
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.4543373584747314,
      "learning_rate": 5.331639065925932e-06,
      "loss": 0.586,
      "step": 19289
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8088853359222412,
      "learning_rate": 5.330651659653643e-06,
      "loss": 0.5185,
      "step": 19290
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5061372518539429,
      "learning_rate": 5.329664311595873e-06,
      "loss": 0.4535,
      "step": 19291
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.0141701698303223,
      "learning_rate": 5.328677021764927e-06,
      "loss": 0.455,
      "step": 19292
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.721236228942871,
      "learning_rate": 5.327689790173123e-06,
      "loss": 0.531,
      "step": 19293
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5414543151855469,
      "learning_rate": 5.326702616832762e-06,
      "loss": 0.5362,
      "step": 19294
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3122320175170898,
      "learning_rate": 5.3257155017561504e-06,
      "loss": 0.4959,
      "step": 19295
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.678504467010498,
      "learning_rate": 5.3247284449556025e-06,
      "loss": 0.4459,
      "step": 19296
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.381744146347046,
      "learning_rate": 5.323741446443419e-06,
      "loss": 0.4091,
      "step": 19297
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.35597562789917,
      "learning_rate": 5.322754506231907e-06,
      "loss": 0.5025,
      "step": 19298
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7174113988876343,
      "learning_rate": 5.321767624333365e-06,
      "loss": 0.5365,
      "step": 19299
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6694616079330444,
      "learning_rate": 5.320780800760106e-06,
      "loss": 0.5249,
      "step": 19300
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8963392972946167,
      "learning_rate": 5.31979403552443e-06,
      "loss": 0.49,
      "step": 19301
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9821369647979736,
      "learning_rate": 5.318807328638634e-06,
      "loss": 0.5751,
      "step": 19302
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4618521928787231,
      "learning_rate": 5.3178206801150285e-06,
      "loss": 0.4992,
      "step": 19303
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8245458602905273,
      "learning_rate": 5.31683408996591e-06,
      "loss": 0.6237,
      "step": 19304
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3780957460403442,
      "learning_rate": 5.315847558203573e-06,
      "loss": 0.502,
      "step": 19305
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.0509285926818848,
      "learning_rate": 5.314861084840328e-06,
      "loss": 0.6862,
      "step": 19306
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7246673107147217,
      "learning_rate": 5.313874669888468e-06,
      "loss": 0.5031,
      "step": 19307
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5354300737380981,
      "learning_rate": 5.312888313360287e-06,
      "loss": 0.5325,
      "step": 19308
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8025609254837036,
      "learning_rate": 5.311902015268091e-06,
      "loss": 0.4785,
      "step": 19309
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5928043127059937,
      "learning_rate": 5.310915775624172e-06,
      "loss": 0.5124,
      "step": 19310
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3997950553894043,
      "learning_rate": 5.309929594440828e-06,
      "loss": 0.395,
      "step": 19311
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.720573902130127,
      "learning_rate": 5.308943471730346e-06,
      "loss": 0.5028,
      "step": 19312
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.732404112815857,
      "learning_rate": 5.307957407505031e-06,
      "loss": 0.463,
      "step": 19313
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3507423400878906,
      "learning_rate": 5.306971401777172e-06,
      "loss": 0.4306,
      "step": 19314
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5791127681732178,
      "learning_rate": 5.3059854545590575e-06,
      "loss": 0.522,
      "step": 19315
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9226897954940796,
      "learning_rate": 5.304999565862989e-06,
      "loss": 0.5453,
      "step": 19316
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.437403917312622,
      "learning_rate": 5.3040137357012526e-06,
      "loss": 0.4769,
      "step": 19317
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.716134786605835,
      "learning_rate": 5.303027964086136e-06,
      "loss": 0.5905,
      "step": 19318
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9441351890563965,
      "learning_rate": 5.302042251029936e-06,
      "loss": 0.5263,
      "step": 19319
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5243949890136719,
      "learning_rate": 5.30105659654494e-06,
      "loss": 0.4674,
      "step": 19320
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.1898720264434814,
      "learning_rate": 5.30007100064343e-06,
      "loss": 0.48,
      "step": 19321
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.719123363494873,
      "learning_rate": 5.299085463337703e-06,
      "loss": 0.5112,
      "step": 19322
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.9746367931365967,
      "learning_rate": 5.298099984640042e-06,
      "loss": 0.5554,
      "step": 19323
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4600354433059692,
      "learning_rate": 5.2971145645627335e-06,
      "loss": 0.4213,
      "step": 19324
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.346709966659546,
      "learning_rate": 5.296129203118058e-06,
      "loss": 0.4825,
      "step": 19325
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5009765625,
      "learning_rate": 5.295143900318309e-06,
      "loss": 0.5955,
      "step": 19326
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.721747636795044,
      "learning_rate": 5.294158656175769e-06,
      "loss": 0.5173,
      "step": 19327
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.633995771408081,
      "learning_rate": 5.293173470702714e-06,
      "loss": 0.4908,
      "step": 19328
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3112425804138184,
      "learning_rate": 5.292188343911435e-06,
      "loss": 0.5255,
      "step": 19329
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.898134469985962,
      "learning_rate": 5.2912032758142125e-06,
      "loss": 0.6065,
      "step": 19330
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5259677171707153,
      "learning_rate": 5.290218266423321e-06,
      "loss": 0.5242,
      "step": 19331
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7244479656219482,
      "learning_rate": 5.289233315751052e-06,
      "loss": 0.4954,
      "step": 19332
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6076029539108276,
      "learning_rate": 5.2882484238096776e-06,
      "loss": 0.5302,
      "step": 19333
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6986970901489258,
      "learning_rate": 5.287263590611475e-06,
      "loss": 0.4305,
      "step": 19334
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6612751483917236,
      "learning_rate": 5.2862788161687306e-06,
      "loss": 0.5302,
      "step": 19335
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7056230306625366,
      "learning_rate": 5.2852941004937205e-06,
      "loss": 0.4531,
      "step": 19336
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5405107736587524,
      "learning_rate": 5.284309443598712e-06,
      "loss": 0.5234,
      "step": 19337
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5747618675231934,
      "learning_rate": 5.2833248454959896e-06,
      "loss": 0.4705,
      "step": 19338
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3059520721435547,
      "learning_rate": 5.2823403061978285e-06,
      "loss": 0.3611,
      "step": 19339
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4240792989730835,
      "learning_rate": 5.281355825716497e-06,
      "loss": 0.3062,
      "step": 19340
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4610663652420044,
      "learning_rate": 5.2803714040642774e-06,
      "loss": 0.4281,
      "step": 19341
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2108091115951538,
      "learning_rate": 5.279387041253439e-06,
      "loss": 0.3736,
      "step": 19342
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0916023254394531,
      "learning_rate": 5.278402737296253e-06,
      "loss": 0.345,
      "step": 19343
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.193647861480713,
      "learning_rate": 5.277418492204989e-06,
      "loss": 0.3695,
      "step": 19344
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3849290609359741,
      "learning_rate": 5.276434305991926e-06,
      "loss": 0.314,
      "step": 19345
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4063774347305298,
      "learning_rate": 5.275450178669328e-06,
      "loss": 0.4187,
      "step": 19346
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3938093185424805,
      "learning_rate": 5.274466110249463e-06,
      "loss": 0.3272,
      "step": 19347
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1536122560501099,
      "learning_rate": 5.273482100744607e-06,
      "loss": 0.3689,
      "step": 19348
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3434635400772095,
      "learning_rate": 5.272498150167022e-06,
      "loss": 0.3002,
      "step": 19349
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5277844667434692,
      "learning_rate": 5.271514258528975e-06,
      "loss": 0.3341,
      "step": 19350
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2104421854019165,
      "learning_rate": 5.270530425842737e-06,
      "loss": 0.2535,
      "step": 19351
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.221388578414917,
      "learning_rate": 5.269546652120572e-06,
      "loss": 0.3592,
      "step": 19352
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.371690273284912,
      "learning_rate": 5.268562937374741e-06,
      "loss": 0.3385,
      "step": 19353
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.400183081626892,
      "learning_rate": 5.267579281617514e-06,
      "loss": 0.3975,
      "step": 19354
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.448101282119751,
      "learning_rate": 5.266595684861154e-06,
      "loss": 0.3706,
      "step": 19355
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2597938776016235,
      "learning_rate": 5.265612147117922e-06,
      "loss": 0.3439,
      "step": 19356
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3070075511932373,
      "learning_rate": 5.2646286684000755e-06,
      "loss": 0.3374,
      "step": 19357
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2929528951644897,
      "learning_rate": 5.263645248719885e-06,
      "loss": 0.3292,
      "step": 19358
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6335197687149048,
      "learning_rate": 5.262661888089607e-06,
      "loss": 0.355,
      "step": 19359
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4611146450042725,
      "learning_rate": 5.261678586521498e-06,
      "loss": 0.3494,
      "step": 19360
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5133304595947266,
      "learning_rate": 5.260695344027824e-06,
      "loss": 0.3813,
      "step": 19361
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3924379348754883,
      "learning_rate": 5.25971216062084e-06,
      "loss": 0.3402,
      "step": 19362
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2297677993774414,
      "learning_rate": 5.2587290363128e-06,
      "loss": 0.3313,
      "step": 19363
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7557744979858398,
      "learning_rate": 5.257745971115969e-06,
      "loss": 0.4267,
      "step": 19364
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.289005994796753,
      "learning_rate": 5.256762965042597e-06,
      "loss": 0.3208,
      "step": 19365
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5392875671386719,
      "learning_rate": 5.255780018104939e-06,
      "loss": 0.3227,
      "step": 19366
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5528080463409424,
      "learning_rate": 5.254797130315257e-06,
      "loss": 0.2759,
      "step": 19367
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6595122814178467,
      "learning_rate": 5.253814301685798e-06,
      "loss": 0.3468,
      "step": 19368
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7541024684906006,
      "learning_rate": 5.2528315322288196e-06,
      "loss": 0.3463,
      "step": 19369
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5928914546966553,
      "learning_rate": 5.251848821956566e-06,
      "loss": 0.3085,
      "step": 19370
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.647227168083191,
      "learning_rate": 5.250866170881301e-06,
      "loss": 0.2926,
      "step": 19371
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6992748975753784,
      "learning_rate": 5.249883579015269e-06,
      "loss": 0.2904,
      "step": 19372
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7930018901824951,
      "learning_rate": 5.248901046370718e-06,
      "loss": 0.3905,
      "step": 19373
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7831165790557861,
      "learning_rate": 5.2479185729599045e-06,
      "loss": 0.3129,
      "step": 19374
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.541562557220459,
      "learning_rate": 5.246936158795075e-06,
      "loss": 0.3514,
      "step": 19375
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8803240060806274,
      "learning_rate": 5.245953803888472e-06,
      "loss": 0.3962,
      "step": 19376
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.936108112335205,
      "learning_rate": 5.24497150825235e-06,
      "loss": 0.3991,
      "step": 19377
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.825557827949524,
      "learning_rate": 5.243989271898954e-06,
      "loss": 0.3149,
      "step": 19378
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5730502605438232,
      "learning_rate": 5.2430070948405245e-06,
      "loss": 0.3241,
      "step": 19379
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7015352249145508,
      "learning_rate": 5.242024977089315e-06,
      "loss": 0.3437,
      "step": 19380
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6213538646697998,
      "learning_rate": 5.241042918657566e-06,
      "loss": 0.406,
      "step": 19381
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5103791952133179,
      "learning_rate": 5.2400609195575216e-06,
      "loss": 0.346,
      "step": 19382
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5838921070098877,
      "learning_rate": 5.23907897980142e-06,
      "loss": 0.3762,
      "step": 19383
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5756129026412964,
      "learning_rate": 5.238097099401511e-06,
      "loss": 0.3704,
      "step": 19384
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8438782691955566,
      "learning_rate": 5.237115278370033e-06,
      "loss": 0.2886,
      "step": 19385
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5624383687973022,
      "learning_rate": 5.236133516719224e-06,
      "loss": 0.3978,
      "step": 19386
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4469914436340332,
      "learning_rate": 5.235151814461328e-06,
      "loss": 0.3152,
      "step": 19387
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7401024103164673,
      "learning_rate": 5.234170171608585e-06,
      "loss": 0.3549,
      "step": 19388
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4882808923721313,
      "learning_rate": 5.233188588173226e-06,
      "loss": 0.3384,
      "step": 19389
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.22806978225708,
      "learning_rate": 5.232207064167498e-06,
      "loss": 0.3839,
      "step": 19390
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5596016645431519,
      "learning_rate": 5.231225599603636e-06,
      "loss": 0.3934,
      "step": 19391
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6379852294921875,
      "learning_rate": 5.230244194493867e-06,
      "loss": 0.3603,
      "step": 19392
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6095335483551025,
      "learning_rate": 5.2292628488504405e-06,
      "loss": 0.2384,
      "step": 19393
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8921853303909302,
      "learning_rate": 5.2282815626855845e-06,
      "loss": 0.3451,
      "step": 19394
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.9285696744918823,
      "learning_rate": 5.227300336011532e-06,
      "loss": 0.3544,
      "step": 19395
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4833381175994873,
      "learning_rate": 5.2263191688405145e-06,
      "loss": 0.3372,
      "step": 19396
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5985661745071411,
      "learning_rate": 5.2253380611847706e-06,
      "loss": 0.3456,
      "step": 19397
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.1521129608154297,
      "learning_rate": 5.22435701305653e-06,
      "loss": 0.4325,
      "step": 19398
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7021758556365967,
      "learning_rate": 5.223376024468019e-06,
      "loss": 0.4379,
      "step": 19399
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5765571594238281,
      "learning_rate": 5.222395095431476e-06,
      "loss": 0.2581,
      "step": 19400
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6463897228240967,
      "learning_rate": 5.2214142259591275e-06,
      "loss": 0.3606,
      "step": 19401
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4401520490646362,
      "learning_rate": 5.2204334160631956e-06,
      "loss": 0.2561,
      "step": 19402
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.743117094039917,
      "learning_rate": 5.21945266575592e-06,
      "loss": 0.3767,
      "step": 19403
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6307787895202637,
      "learning_rate": 5.218471975049521e-06,
      "loss": 0.3663,
      "step": 19404
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5502663850784302,
      "learning_rate": 5.217491343956222e-06,
      "loss": 0.3431,
      "step": 19405
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.089667797088623,
      "learning_rate": 5.216510772488259e-06,
      "loss": 0.3491,
      "step": 19406
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3788553476333618,
      "learning_rate": 5.21553026065785e-06,
      "loss": 0.3013,
      "step": 19407
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6782668828964233,
      "learning_rate": 5.214549808477223e-06,
      "loss": 0.3371,
      "step": 19408
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.707875370979309,
      "learning_rate": 5.213569415958595e-06,
      "loss": 0.3406,
      "step": 19409
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4709912538528442,
      "learning_rate": 5.212589083114197e-06,
      "loss": 0.377,
      "step": 19410
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5298610925674438,
      "learning_rate": 5.211608809956248e-06,
      "loss": 0.4027,
      "step": 19411
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6182196140289307,
      "learning_rate": 5.210628596496965e-06,
      "loss": 0.3851,
      "step": 19412
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.3317787647247314,
      "learning_rate": 5.209648442748578e-06,
      "loss": 0.3409,
      "step": 19413
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7631311416625977,
      "learning_rate": 5.208668348723301e-06,
      "loss": 0.431,
      "step": 19414
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8783411979675293,
      "learning_rate": 5.20768831443335e-06,
      "loss": 0.3634,
      "step": 19415
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6090017557144165,
      "learning_rate": 5.206708339890951e-06,
      "loss": 0.2768,
      "step": 19416
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4645928144454956,
      "learning_rate": 5.2057284251083195e-06,
      "loss": 0.3423,
      "step": 19417
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7949767112731934,
      "learning_rate": 5.204748570097666e-06,
      "loss": 0.3152,
      "step": 19418
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5275415182113647,
      "learning_rate": 5.203768774871217e-06,
      "loss": 0.3789,
      "step": 19419
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.631489634513855,
      "learning_rate": 5.202789039441183e-06,
      "loss": 0.3546,
      "step": 19420
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6660538911819458,
      "learning_rate": 5.201809363819777e-06,
      "loss": 0.3751,
      "step": 19421
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5617772340774536,
      "learning_rate": 5.200829748019212e-06,
      "loss": 0.3202,
      "step": 19422
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5212775468826294,
      "learning_rate": 5.199850192051708e-06,
      "loss": 0.3535,
      "step": 19423
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.760450005531311,
      "learning_rate": 5.1988706959294725e-06,
      "loss": 0.3499,
      "step": 19424
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6825226545333862,
      "learning_rate": 5.197891259664714e-06,
      "loss": 0.3039,
      "step": 19425
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3955631256103516,
      "learning_rate": 5.196911883269652e-06,
      "loss": 0.2878,
      "step": 19426
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.942821979522705,
      "learning_rate": 5.195932566756492e-06,
      "loss": 0.3214,
      "step": 19427
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.529815673828125,
      "learning_rate": 5.194953310137441e-06,
      "loss": 0.3494,
      "step": 19428
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.432286262512207,
      "learning_rate": 5.1939741134247135e-06,
      "loss": 0.2981,
      "step": 19429
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6041510105133057,
      "learning_rate": 5.192994976630514e-06,
      "loss": 0.3777,
      "step": 19430
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5001572370529175,
      "learning_rate": 5.192015899767047e-06,
      "loss": 0.3837,
      "step": 19431
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.678411602973938,
      "learning_rate": 5.191036882846528e-06,
      "loss": 0.4407,
      "step": 19432
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2525736093521118,
      "learning_rate": 5.190057925881154e-06,
      "loss": 0.2897,
      "step": 19433
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.9294313192367554,
      "learning_rate": 5.189079028883135e-06,
      "loss": 0.4141,
      "step": 19434
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4868971109390259,
      "learning_rate": 5.18810019186467e-06,
      "loss": 0.306,
      "step": 19435
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4271557331085205,
      "learning_rate": 5.18712141483797e-06,
      "loss": 0.306,
      "step": 19436
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7605177164077759,
      "learning_rate": 5.186142697815233e-06,
      "loss": 0.3304,
      "step": 19437
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5551400184631348,
      "learning_rate": 5.185164040808656e-06,
      "loss": 0.2594,
      "step": 19438
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.507781744003296,
      "learning_rate": 5.184185443830454e-06,
      "loss": 0.37,
      "step": 19439
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.9556758403778076,
      "learning_rate": 5.183206906892817e-06,
      "loss": 0.2806,
      "step": 19440
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4739184379577637,
      "learning_rate": 5.182228430007944e-06,
      "loss": 0.3106,
      "step": 19441
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8570189476013184,
      "learning_rate": 5.181250013188042e-06,
      "loss": 0.4163,
      "step": 19442
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8190977573394775,
      "learning_rate": 5.180271656445304e-06,
      "loss": 0.324,
      "step": 19443
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.677794098854065,
      "learning_rate": 5.179293359791924e-06,
      "loss": 0.326,
      "step": 19444
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7471294403076172,
      "learning_rate": 5.1783151232401075e-06,
      "loss": 0.3122,
      "step": 19445
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7418512105941772,
      "learning_rate": 5.1773369468020465e-06,
      "loss": 0.3371,
      "step": 19446
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5426310300827026,
      "learning_rate": 5.176358830489935e-06,
      "loss": 0.3163,
      "step": 19447
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8829306364059448,
      "learning_rate": 5.175380774315966e-06,
      "loss": 0.3743,
      "step": 19448
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5837945938110352,
      "learning_rate": 5.174402778292338e-06,
      "loss": 0.34,
      "step": 19449
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5490261316299438,
      "learning_rate": 5.173424842431243e-06,
      "loss": 0.2884,
      "step": 19450
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7552361488342285,
      "learning_rate": 5.172446966744867e-06,
      "loss": 0.282,
      "step": 19451
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.069873571395874,
      "learning_rate": 5.171469151245412e-06,
      "loss": 0.4092,
      "step": 19452
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5273126363754272,
      "learning_rate": 5.170491395945062e-06,
      "loss": 0.3367,
      "step": 19453
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.9067546129226685,
      "learning_rate": 5.169513700856005e-06,
      "loss": 0.396,
      "step": 19454
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.589015007019043,
      "learning_rate": 5.168536065990437e-06,
      "loss": 0.3048,
      "step": 19455
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6875468492507935,
      "learning_rate": 5.167558491360544e-06,
      "loss": 0.2698,
      "step": 19456
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5883110761642456,
      "learning_rate": 5.1665809769785105e-06,
      "loss": 0.3429,
      "step": 19457
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8008983135223389,
      "learning_rate": 5.165603522856528e-06,
      "loss": 0.3949,
      "step": 19458
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5269291400909424,
      "learning_rate": 5.164626129006782e-06,
      "loss": 0.3811,
      "step": 19459
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8536936044692993,
      "learning_rate": 5.1636487954414576e-06,
      "loss": 0.3095,
      "step": 19460
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7334885597229004,
      "learning_rate": 5.162671522172733e-06,
      "loss": 0.3533,
      "step": 19461
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8279043436050415,
      "learning_rate": 5.161694309212804e-06,
      "loss": 0.3122,
      "step": 19462
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8275535106658936,
      "learning_rate": 5.1607171565738466e-06,
      "loss": 0.295,
      "step": 19463
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.1790530681610107,
      "learning_rate": 5.1597400642680414e-06,
      "loss": 0.4314,
      "step": 19464
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6723272800445557,
      "learning_rate": 5.158763032307577e-06,
      "loss": 0.2948,
      "step": 19465
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5032308101654053,
      "learning_rate": 5.157786060704633e-06,
      "loss": 0.3275,
      "step": 19466
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7132657766342163,
      "learning_rate": 5.156809149471381e-06,
      "loss": 0.2763,
      "step": 19467
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5217077732086182,
      "learning_rate": 5.155832298620013e-06,
      "loss": 0.3787,
      "step": 19468
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.039792537689209,
      "learning_rate": 5.1548555081627e-06,
      "loss": 0.4026,
      "step": 19469
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.098702907562256,
      "learning_rate": 5.153878778111621e-06,
      "loss": 0.3181,
      "step": 19470
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.0949528217315674,
      "learning_rate": 5.1529021084789555e-06,
      "loss": 0.3262,
      "step": 19471
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.8749417066574097,
      "learning_rate": 5.15192549927688e-06,
      "loss": 0.3447,
      "step": 19472
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.736799955368042,
      "learning_rate": 5.1509489505175685e-06,
      "loss": 0.3885,
      "step": 19473
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7705961465835571,
      "learning_rate": 5.1499724622131915e-06,
      "loss": 0.3601,
      "step": 19474
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7113556861877441,
      "learning_rate": 5.1489960343759335e-06,
      "loss": 0.3189,
      "step": 19475
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6852977275848389,
      "learning_rate": 5.148019667017963e-06,
      "loss": 0.3828,
      "step": 19476
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4758334159851074,
      "learning_rate": 5.147043360151447e-06,
      "loss": 0.3022,
      "step": 19477
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2498424053192139,
      "learning_rate": 5.146067113788568e-06,
      "loss": 0.1775,
      "step": 19478
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7193070650100708,
      "learning_rate": 5.145090927941491e-06,
      "loss": 0.3118,
      "step": 19479
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.6658962965011597,
      "learning_rate": 5.144114802622385e-06,
      "loss": 0.2956,
      "step": 19480
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3784427642822266,
      "learning_rate": 5.1431387378434264e-06,
      "loss": 0.2276,
      "step": 19481
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.64175283908844,
      "learning_rate": 5.1421627336167805e-06,
      "loss": 0.3027,
      "step": 19482
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4861124753952026,
      "learning_rate": 5.141186789954609e-06,
      "loss": 0.3038,
      "step": 19483
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.617293357849121,
      "learning_rate": 5.140210906869091e-06,
      "loss": 0.3205,
      "step": 19484
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6730642318725586,
      "learning_rate": 5.139235084372387e-06,
      "loss": 0.2443,
      "step": 19485
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6835689544677734,
      "learning_rate": 5.1382593224766644e-06,
      "loss": 0.3662,
      "step": 19486
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6697074174880981,
      "learning_rate": 5.137283621194082e-06,
      "loss": 0.2707,
      "step": 19487
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.4227242469787598,
      "learning_rate": 5.1363079805368164e-06,
      "loss": 0.3676,
      "step": 19488
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.798263430595398,
      "learning_rate": 5.135332400517023e-06,
      "loss": 0.3253,
      "step": 19489
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6797529458999634,
      "learning_rate": 5.134356881146862e-06,
      "loss": 0.289,
      "step": 19490
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.057555675506592,
      "learning_rate": 5.133381422438506e-06,
      "loss": 0.2591,
      "step": 19491
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5018235445022583,
      "learning_rate": 5.1324060244041084e-06,
      "loss": 0.292,
      "step": 19492
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.384838819503784,
      "learning_rate": 5.131430687055829e-06,
      "loss": 0.3831,
      "step": 19493
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9109684228897095,
      "learning_rate": 5.130455410405835e-06,
      "loss": 0.4196,
      "step": 19494
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7886604070663452,
      "learning_rate": 5.1294801944662795e-06,
      "loss": 0.2995,
      "step": 19495
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7149138450622559,
      "learning_rate": 5.128505039249319e-06,
      "loss": 0.4225,
      "step": 19496
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7707147598266602,
      "learning_rate": 5.12752994476712e-06,
      "loss": 0.247,
      "step": 19497
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6966809034347534,
      "learning_rate": 5.126554911031834e-06,
      "loss": 0.4025,
      "step": 19498
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8672621250152588,
      "learning_rate": 5.125579938055615e-06,
      "loss": 0.3368,
      "step": 19499
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7430979013442993,
      "learning_rate": 5.1246050258506175e-06,
      "loss": 0.3022,
      "step": 19500
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8712571859359741,
      "learning_rate": 5.123630174429003e-06,
      "loss": 0.3193,
      "step": 19501
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.662677526473999,
      "learning_rate": 5.122655383802922e-06,
      "loss": 0.3154,
      "step": 19502
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7425559759140015,
      "learning_rate": 5.121680653984522e-06,
      "loss": 0.3232,
      "step": 19503
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.1096038818359375,
      "learning_rate": 5.1207059849859655e-06,
      "loss": 0.3951,
      "step": 19504
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.046900749206543,
      "learning_rate": 5.119731376819398e-06,
      "loss": 0.3226,
      "step": 19505
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4917224645614624,
      "learning_rate": 5.118756829496967e-06,
      "loss": 0.3256,
      "step": 19506
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5801502466201782,
      "learning_rate": 5.117782343030831e-06,
      "loss": 0.3884,
      "step": 19507
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8253177404403687,
      "learning_rate": 5.116807917433135e-06,
      "loss": 0.3136,
      "step": 19508
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7887569665908813,
      "learning_rate": 5.115833552716024e-06,
      "loss": 0.398,
      "step": 19509
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.685176968574524,
      "learning_rate": 5.1148592488916524e-06,
      "loss": 0.3433,
      "step": 19510
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.556503176689148,
      "learning_rate": 5.113885005972166e-06,
      "loss": 0.34,
      "step": 19511
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8119401931762695,
      "learning_rate": 5.112910823969707e-06,
      "loss": 0.3343,
      "step": 19512
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7302669286727905,
      "learning_rate": 5.11193670289642e-06,
      "loss": 0.3382,
      "step": 19513
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7951631546020508,
      "learning_rate": 5.110962642764456e-06,
      "loss": 0.3784,
      "step": 19514
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9858503341674805,
      "learning_rate": 5.109988643585957e-06,
      "loss": 0.3478,
      "step": 19515
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.701864242553711,
      "learning_rate": 5.1090147053730596e-06,
      "loss": 0.3207,
      "step": 19516
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8162431716918945,
      "learning_rate": 5.108040828137917e-06,
      "loss": 0.3849,
      "step": 19517
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5179164409637451,
      "learning_rate": 5.107067011892665e-06,
      "loss": 0.2103,
      "step": 19518
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6695107221603394,
      "learning_rate": 5.106093256649441e-06,
      "loss": 0.387,
      "step": 19519
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8051815032958984,
      "learning_rate": 5.105119562420393e-06,
      "loss": 0.2597,
      "step": 19520
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9023274183273315,
      "learning_rate": 5.104145929217656e-06,
      "loss": 0.3102,
      "step": 19521
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9341156482696533,
      "learning_rate": 5.103172357053367e-06,
      "loss": 0.3068,
      "step": 19522
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4756274223327637,
      "learning_rate": 5.1021988459396695e-06,
      "loss": 0.3813,
      "step": 19523
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5845118761062622,
      "learning_rate": 5.101225395888697e-06,
      "loss": 0.3344,
      "step": 19524
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5773862600326538,
      "learning_rate": 5.100252006912587e-06,
      "loss": 0.2522,
      "step": 19525
    },
    {
      "epoch": 2.02,
      "grad_norm": 27.164785385131836,
      "learning_rate": 5.0992786790234695e-06,
      "loss": 1.3314,
      "step": 19526
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.089481830596924,
      "learning_rate": 5.0983054122334886e-06,
      "loss": 0.357,
      "step": 19527
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.1056456565856934,
      "learning_rate": 5.097332206554772e-06,
      "loss": 0.3537,
      "step": 19528
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4230519533157349,
      "learning_rate": 5.0963590619994545e-06,
      "loss": 0.3462,
      "step": 19529
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.267788052558899,
      "learning_rate": 5.09538597857967e-06,
      "loss": 0.2828,
      "step": 19530
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5250487327575684,
      "learning_rate": 5.094412956307551e-06,
      "loss": 0.3758,
      "step": 19531
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7942652702331543,
      "learning_rate": 5.093439995195222e-06,
      "loss": 0.3964,
      "step": 19532
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.343531608581543,
      "learning_rate": 5.092467095254823e-06,
      "loss": 0.3092,
      "step": 19533
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5823771953582764,
      "learning_rate": 5.091494256498477e-06,
      "loss": 0.3475,
      "step": 19534
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7407385110855103,
      "learning_rate": 5.090521478938313e-06,
      "loss": 0.4005,
      "step": 19535
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4141978025436401,
      "learning_rate": 5.0895487625864625e-06,
      "loss": 0.2674,
      "step": 19536
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4287837743759155,
      "learning_rate": 5.088576107455051e-06,
      "loss": 0.3725,
      "step": 19537
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.0362672805786133,
      "learning_rate": 5.087603513556206e-06,
      "loss": 0.2996,
      "step": 19538
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7960158586502075,
      "learning_rate": 5.086630980902045e-06,
      "loss": 0.3815,
      "step": 19539
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8924325704574585,
      "learning_rate": 5.085658509504706e-06,
      "loss": 0.3864,
      "step": 19540
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9333710670471191,
      "learning_rate": 5.084686099376306e-06,
      "loss": 0.3441,
      "step": 19541
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7124234437942505,
      "learning_rate": 5.083713750528967e-06,
      "loss": 0.3931,
      "step": 19542
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3543137311935425,
      "learning_rate": 5.082741462974816e-06,
      "loss": 0.288,
      "step": 19543
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7261992692947388,
      "learning_rate": 5.081769236725973e-06,
      "loss": 0.2696,
      "step": 19544
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.0050857067108154,
      "learning_rate": 5.080797071794556e-06,
      "loss": 0.3128,
      "step": 19545
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7578554153442383,
      "learning_rate": 5.079824968192691e-06,
      "loss": 0.3408,
      "step": 19546
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7842142581939697,
      "learning_rate": 5.078852925932496e-06,
      "loss": 0.2974,
      "step": 19547
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6894479990005493,
      "learning_rate": 5.0778809450260835e-06,
      "loss": 0.2527,
      "step": 19548
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8187335729599,
      "learning_rate": 5.076909025485582e-06,
      "loss": 0.3407,
      "step": 19549
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9178434610366821,
      "learning_rate": 5.075937167323101e-06,
      "loss": 0.3607,
      "step": 19550
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9103994369506836,
      "learning_rate": 5.074965370550762e-06,
      "loss": 0.3008,
      "step": 19551
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7867186069488525,
      "learning_rate": 5.073993635180674e-06,
      "loss": 0.3252,
      "step": 19552
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5849697589874268,
      "learning_rate": 5.07302196122496e-06,
      "loss": 0.3649,
      "step": 19553
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.828818917274475,
      "learning_rate": 5.072050348695729e-06,
      "loss": 0.3509,
      "step": 19554
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9116932153701782,
      "learning_rate": 5.071078797605094e-06,
      "loss": 0.3902,
      "step": 19555
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.974937915802002,
      "learning_rate": 5.070107307965173e-06,
      "loss": 0.3374,
      "step": 19556
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4810616970062256,
      "learning_rate": 5.069135879788074e-06,
      "loss": 0.252,
      "step": 19557
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.3519930839538574,
      "learning_rate": 5.068164513085905e-06,
      "loss": 0.3858,
      "step": 19558
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7673436403274536,
      "learning_rate": 5.067193207870785e-06,
      "loss": 0.3299,
      "step": 19559
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.616153597831726,
      "learning_rate": 5.066221964154817e-06,
      "loss": 0.3357,
      "step": 19560
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.672141194343567,
      "learning_rate": 5.065250781950108e-06,
      "loss": 0.3414,
      "step": 19561
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6981548070907593,
      "learning_rate": 5.064279661268775e-06,
      "loss": 0.3573,
      "step": 19562
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8821884393692017,
      "learning_rate": 5.06330860212292e-06,
      "loss": 0.2957,
      "step": 19563
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8601933717727661,
      "learning_rate": 5.062337604524647e-06,
      "loss": 0.4015,
      "step": 19564
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7689366340637207,
      "learning_rate": 5.061366668486063e-06,
      "loss": 0.2853,
      "step": 19565
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5139304399490356,
      "learning_rate": 5.060395794019278e-06,
      "loss": 0.329,
      "step": 19566
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7917969226837158,
      "learning_rate": 5.059424981136392e-06,
      "loss": 0.3194,
      "step": 19567
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5795328617095947,
      "learning_rate": 5.058454229849505e-06,
      "loss": 0.355,
      "step": 19568
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7733988761901855,
      "learning_rate": 5.057483540170729e-06,
      "loss": 0.3713,
      "step": 19569
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8033534288406372,
      "learning_rate": 5.056512912112161e-06,
      "loss": 0.2791,
      "step": 19570
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.202378034591675,
      "learning_rate": 5.055542345685896e-06,
      "loss": 0.2995,
      "step": 19571
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.9199025630950928,
      "learning_rate": 5.054571840904045e-06,
      "loss": 0.321,
      "step": 19572
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7311012744903564,
      "learning_rate": 5.0536013977787046e-06,
      "loss": 0.3759,
      "step": 19573
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7528460025787354,
      "learning_rate": 5.052631016321967e-06,
      "loss": 0.3464,
      "step": 19574
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5339405536651611,
      "learning_rate": 5.051660696545939e-06,
      "loss": 0.3199,
      "step": 19575
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6878716945648193,
      "learning_rate": 5.0506904384627155e-06,
      "loss": 0.3353,
      "step": 19576
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.825561761856079,
      "learning_rate": 5.04972024208439e-06,
      "loss": 0.3973,
      "step": 19577
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8875116109848022,
      "learning_rate": 5.048750107423059e-06,
      "loss": 0.3147,
      "step": 19578
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7258893251419067,
      "learning_rate": 5.047780034490821e-06,
      "loss": 0.343,
      "step": 19579
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7527364492416382,
      "learning_rate": 5.0468100232997685e-06,
      "loss": 0.3887,
      "step": 19580
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7517274618148804,
      "learning_rate": 5.045840073861991e-06,
      "loss": 0.3935,
      "step": 19581
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7881940603256226,
      "learning_rate": 5.044870186189588e-06,
      "loss": 0.303,
      "step": 19582
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6679010391235352,
      "learning_rate": 5.043900360294649e-06,
      "loss": 0.2512,
      "step": 19583
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7707927227020264,
      "learning_rate": 5.042930596189259e-06,
      "loss": 0.454,
      "step": 19584
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8389631509780884,
      "learning_rate": 5.041960893885519e-06,
      "loss": 0.4209,
      "step": 19585
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8944820165634155,
      "learning_rate": 5.040991253395513e-06,
      "loss": 0.341,
      "step": 19586
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.4333269596099854,
      "learning_rate": 5.040021674731327e-06,
      "loss": 0.3955,
      "step": 19587
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9248520135879517,
      "learning_rate": 5.039052157905055e-06,
      "loss": 0.3421,
      "step": 19588
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7864023447036743,
      "learning_rate": 5.038082702928784e-06,
      "loss": 0.2856,
      "step": 19589
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.229402780532837,
      "learning_rate": 5.037113309814596e-06,
      "loss": 0.369,
      "step": 19590
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.022153377532959,
      "learning_rate": 5.036143978574577e-06,
      "loss": 0.4047,
      "step": 19591
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4141868352890015,
      "learning_rate": 5.035174709220818e-06,
      "loss": 0.3127,
      "step": 19592
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.410317063331604,
      "learning_rate": 5.034205501765398e-06,
      "loss": 0.2458,
      "step": 19593
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6530178785324097,
      "learning_rate": 5.033236356220399e-06,
      "loss": 0.2968,
      "step": 19594
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.743299126625061,
      "learning_rate": 5.03226727259791e-06,
      "loss": 0.4265,
      "step": 19595
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.552245855331421,
      "learning_rate": 5.03129825091001e-06,
      "loss": 0.3684,
      "step": 19596
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.180450677871704,
      "learning_rate": 5.030329291168775e-06,
      "loss": 0.2993,
      "step": 19597
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4062578678131104,
      "learning_rate": 5.029360393386295e-06,
      "loss": 0.2775,
      "step": 19598
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.865506887435913,
      "learning_rate": 5.028391557574643e-06,
      "loss": 0.3524,
      "step": 19599
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8262029886245728,
      "learning_rate": 5.027422783745897e-06,
      "loss": 0.2751,
      "step": 19600
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.859188199043274,
      "learning_rate": 5.026454071912141e-06,
      "loss": 0.3895,
      "step": 19601
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8042864799499512,
      "learning_rate": 5.025485422085447e-06,
      "loss": 0.3746,
      "step": 19602
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4495117664337158,
      "learning_rate": 5.024516834277897e-06,
      "loss": 0.2569,
      "step": 19603
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.57664954662323,
      "learning_rate": 5.023548308501557e-06,
      "loss": 0.2241,
      "step": 19604
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6428345441818237,
      "learning_rate": 5.0225798447685114e-06,
      "loss": 0.3236,
      "step": 19605
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6033940315246582,
      "learning_rate": 5.021611443090833e-06,
      "loss": 0.3032,
      "step": 19606
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0771210193634033,
      "learning_rate": 5.020643103480588e-06,
      "loss": 0.3122,
      "step": 19607
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6917654275894165,
      "learning_rate": 5.0196748259498585e-06,
      "loss": 0.3046,
      "step": 19608
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8914507627487183,
      "learning_rate": 5.0187066105107125e-06,
      "loss": 0.3041,
      "step": 19609
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8962546586990356,
      "learning_rate": 5.0177384571752176e-06,
      "loss": 0.2483,
      "step": 19610
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6886839866638184,
      "learning_rate": 5.01677036595545e-06,
      "loss": 0.3676,
      "step": 19611
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0349416732788086,
      "learning_rate": 5.0158023368634776e-06,
      "loss": 0.3712,
      "step": 19612
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8432878255844116,
      "learning_rate": 5.014834369911364e-06,
      "loss": 0.3589,
      "step": 19613
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.681040644645691,
      "learning_rate": 5.013866465111184e-06,
      "loss": 0.387,
      "step": 19614
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4785737991333008,
      "learning_rate": 5.012898622475004e-06,
      "loss": 0.2673,
      "step": 19615
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0435144901275635,
      "learning_rate": 5.011930842014888e-06,
      "loss": 0.3695,
      "step": 19616
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.258676290512085,
      "learning_rate": 5.010963123742899e-06,
      "loss": 0.4476,
      "step": 19617
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9108656644821167,
      "learning_rate": 5.009995467671109e-06,
      "loss": 0.3844,
      "step": 19618
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7803568840026855,
      "learning_rate": 5.009027873811578e-06,
      "loss": 0.3745,
      "step": 19619
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0343968868255615,
      "learning_rate": 5.0080603421763665e-06,
      "loss": 0.4248,
      "step": 19620
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6171619892120361,
      "learning_rate": 5.0070928727775435e-06,
      "loss": 0.3545,
      "step": 19621
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8157498836517334,
      "learning_rate": 5.006125465627168e-06,
      "loss": 0.3893,
      "step": 19622
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5719648599624634,
      "learning_rate": 5.005158120737298e-06,
      "loss": 0.2752,
      "step": 19623
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5084162950515747,
      "learning_rate": 5.004190838119999e-06,
      "loss": 0.38,
      "step": 19624
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4876147508621216,
      "learning_rate": 5.003223617787328e-06,
      "loss": 0.2889,
      "step": 19625
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.1700570583343506,
      "learning_rate": 5.00225645975134e-06,
      "loss": 0.2993,
      "step": 19626
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6698099374771118,
      "learning_rate": 5.0012893640241005e-06,
      "loss": 0.3042,
      "step": 19627
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6931531429290771,
      "learning_rate": 5.000322330617663e-06,
      "loss": 0.3247,
      "step": 19628
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7760288715362549,
      "learning_rate": 4.9993553595440835e-06,
      "loss": 0.3416,
      "step": 19629
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9680583477020264,
      "learning_rate": 4.998388450815414e-06,
      "loss": 0.3614,
      "step": 19630
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7828192710876465,
      "learning_rate": 4.9974216044437155e-06,
      "loss": 0.4012,
      "step": 19631
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5682069063186646,
      "learning_rate": 4.99645482044104e-06,
      "loss": 0.2934,
      "step": 19632
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6997437477111816,
      "learning_rate": 4.995488098819436e-06,
      "loss": 0.3439,
      "step": 19633
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6378517150878906,
      "learning_rate": 4.994521439590966e-06,
      "loss": 0.3008,
      "step": 19634
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6288518905639648,
      "learning_rate": 4.993554842767675e-06,
      "loss": 0.2951,
      "step": 19635
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.670087218284607,
      "learning_rate": 4.99258830836161e-06,
      "loss": 0.3383,
      "step": 19636
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.552992343902588,
      "learning_rate": 4.991621836384831e-06,
      "loss": 0.2475,
      "step": 19637
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6076629161834717,
      "learning_rate": 4.990655426849382e-06,
      "loss": 0.2962,
      "step": 19638
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6322240829467773,
      "learning_rate": 4.989689079767308e-06,
      "loss": 0.3622,
      "step": 19639
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.2036163806915283,
      "learning_rate": 4.988722795150666e-06,
      "loss": 0.374,
      "step": 19640
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6853405237197876,
      "learning_rate": 4.987756573011495e-06,
      "loss": 0.3226,
      "step": 19641
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7227799892425537,
      "learning_rate": 4.986790413361846e-06,
      "loss": 0.314,
      "step": 19642
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0759952068328857,
      "learning_rate": 4.985824316213758e-06,
      "loss": 0.2994,
      "step": 19643
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6677170991897583,
      "learning_rate": 4.984858281579286e-06,
      "loss": 0.3975,
      "step": 19644
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6517235040664673,
      "learning_rate": 4.9838923094704664e-06,
      "loss": 0.2949,
      "step": 19645
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.775470495223999,
      "learning_rate": 4.9829263998993405e-06,
      "loss": 0.2731,
      "step": 19646
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.041630983352661,
      "learning_rate": 4.981960552877958e-06,
      "loss": 0.3608,
      "step": 19647
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6026825904846191,
      "learning_rate": 4.980994768418357e-06,
      "loss": 0.3299,
      "step": 19648
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.9543485641479492,
      "learning_rate": 4.980029046532574e-06,
      "loss": 0.3601,
      "step": 19649
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6615512371063232,
      "learning_rate": 4.979063387232658e-06,
      "loss": 0.3284,
      "step": 19650
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8596632480621338,
      "learning_rate": 4.978097790530641e-06,
      "loss": 0.3066,
      "step": 19651
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6399927139282227,
      "learning_rate": 4.977132256438562e-06,
      "loss": 0.3276,
      "step": 19652
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.3009629249572754,
      "learning_rate": 4.976166784968463e-06,
      "loss": 0.4236,
      "step": 19653
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.2348005771636963,
      "learning_rate": 4.9752013761323795e-06,
      "loss": 0.3357,
      "step": 19654
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0548601150512695,
      "learning_rate": 4.974236029942345e-06,
      "loss": 0.2799,
      "step": 19655
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7765923738479614,
      "learning_rate": 4.973270746410393e-06,
      "loss": 0.3139,
      "step": 19656
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.0587425231933594,
      "learning_rate": 4.972305525548565e-06,
      "loss": 0.4241,
      "step": 19657
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7550582885742188,
      "learning_rate": 4.971340367368892e-06,
      "loss": 0.4113,
      "step": 19658
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8501664400100708,
      "learning_rate": 4.970375271883402e-06,
      "loss": 0.3476,
      "step": 19659
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.917840838432312,
      "learning_rate": 4.969410239104134e-06,
      "loss": 0.3453,
      "step": 19660
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7075064182281494,
      "learning_rate": 4.9684452690431186e-06,
      "loss": 0.3163,
      "step": 19661
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.1969451904296875,
      "learning_rate": 4.967480361712379e-06,
      "loss": 0.4004,
      "step": 19662
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8537179231643677,
      "learning_rate": 4.966515517123956e-06,
      "loss": 0.348,
      "step": 19663
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.102586030960083,
      "learning_rate": 4.965550735289872e-06,
      "loss": 0.3309,
      "step": 19664
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7659287452697754,
      "learning_rate": 4.964586016222153e-06,
      "loss": 0.2892,
      "step": 19665
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7592978477478027,
      "learning_rate": 4.963621359932834e-06,
      "loss": 0.332,
      "step": 19666
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7229058742523193,
      "learning_rate": 4.9626567664339375e-06,
      "loss": 0.3771,
      "step": 19667
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8763220310211182,
      "learning_rate": 4.961692235737491e-06,
      "loss": 0.3515,
      "step": 19668
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.7239207029342651,
      "learning_rate": 4.960727767855513e-06,
      "loss": 0.2689,
      "step": 19669
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6623579263687134,
      "learning_rate": 4.959763362800039e-06,
      "loss": 0.2973,
      "step": 19670
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5429006814956665,
      "learning_rate": 4.958799020583085e-06,
      "loss": 0.2978,
      "step": 19671
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6032419204711914,
      "learning_rate": 4.957834741216673e-06,
      "loss": 0.3162,
      "step": 19672
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.8996473550796509,
      "learning_rate": 4.956870524712831e-06,
      "loss": 0.3715,
      "step": 19673
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.775716781616211,
      "learning_rate": 4.955906371083577e-06,
      "loss": 0.343,
      "step": 19674
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.629660964012146,
      "learning_rate": 4.954942280340929e-06,
      "loss": 0.423,
      "step": 19675
    },
    {
      "epoch": 2.03,
      "grad_norm": 2.067326068878174,
      "learning_rate": 4.9539782524969125e-06,
      "loss": 0.356,
      "step": 19676
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8148128986358643,
      "learning_rate": 4.9530142875635426e-06,
      "loss": 0.306,
      "step": 19677
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.705180287361145,
      "learning_rate": 4.952050385552834e-06,
      "loss": 0.3203,
      "step": 19678
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7580827474594116,
      "learning_rate": 4.951086546476812e-06,
      "loss": 0.3231,
      "step": 19679
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7834935188293457,
      "learning_rate": 4.95012277034749e-06,
      "loss": 0.4068,
      "step": 19680
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5025417804718018,
      "learning_rate": 4.949159057176881e-06,
      "loss": 0.291,
      "step": 19681
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8077706098556519,
      "learning_rate": 4.948195406976999e-06,
      "loss": 0.423,
      "step": 19682
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.058173179626465,
      "learning_rate": 4.947231819759865e-06,
      "loss": 0.3771,
      "step": 19683
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9367679357528687,
      "learning_rate": 4.9462682955374876e-06,
      "loss": 0.3534,
      "step": 19684
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.715469479560852,
      "learning_rate": 4.945304834321877e-06,
      "loss": 0.2881,
      "step": 19685
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8751041889190674,
      "learning_rate": 4.944341436125052e-06,
      "loss": 0.3238,
      "step": 19686
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9173818826675415,
      "learning_rate": 4.94337810095902e-06,
      "loss": 0.2853,
      "step": 19687
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.30662202835083,
      "learning_rate": 4.942414828835791e-06,
      "loss": 0.3615,
      "step": 19688
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7948561906814575,
      "learning_rate": 4.9414516197673746e-06,
      "loss": 0.2555,
      "step": 19689
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6648764610290527,
      "learning_rate": 4.940488473765779e-06,
      "loss": 0.3223,
      "step": 19690
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6760609149932861,
      "learning_rate": 4.939525390843009e-06,
      "loss": 0.3706,
      "step": 19691
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.0324931144714355,
      "learning_rate": 4.93856237101108e-06,
      "loss": 0.2965,
      "step": 19692
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4785465002059937,
      "learning_rate": 4.937599414281993e-06,
      "loss": 0.3838,
      "step": 19693
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9159473180770874,
      "learning_rate": 4.936636520667751e-06,
      "loss": 0.3469,
      "step": 19694
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.71379017829895,
      "learning_rate": 4.935673690180365e-06,
      "loss": 0.3304,
      "step": 19695
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7673041820526123,
      "learning_rate": 4.9347109228318355e-06,
      "loss": 0.322,
      "step": 19696
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7587167024612427,
      "learning_rate": 4.9337482186341635e-06,
      "loss": 0.3532,
      "step": 19697
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8175721168518066,
      "learning_rate": 4.932785577599357e-06,
      "loss": 0.2883,
      "step": 19698
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.279306173324585,
      "learning_rate": 4.931822999739415e-06,
      "loss": 0.3866,
      "step": 19699
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7687342166900635,
      "learning_rate": 4.930860485066335e-06,
      "loss": 0.3949,
      "step": 19700
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8800479173660278,
      "learning_rate": 4.929898033592122e-06,
      "loss": 0.3456,
      "step": 19701
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5076453685760498,
      "learning_rate": 4.928935645328774e-06,
      "loss": 0.2509,
      "step": 19702
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.608319878578186,
      "learning_rate": 4.927973320288289e-06,
      "loss": 0.4024,
      "step": 19703
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.577296495437622,
      "learning_rate": 4.92701105848266e-06,
      "loss": 0.3311,
      "step": 19704
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.455801248550415,
      "learning_rate": 4.926048859923892e-06,
      "loss": 0.338,
      "step": 19705
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.057140827178955,
      "learning_rate": 4.925086724623977e-06,
      "loss": 0.2867,
      "step": 19706
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6148892641067505,
      "learning_rate": 4.9241246525949086e-06,
      "loss": 0.303,
      "step": 19707
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6925439834594727,
      "learning_rate": 4.923162643848685e-06,
      "loss": 0.3748,
      "step": 19708
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7089412212371826,
      "learning_rate": 4.922200698397301e-06,
      "loss": 0.3173,
      "step": 19709
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7090502977371216,
      "learning_rate": 4.92123881625274e-06,
      "loss": 0.3588,
      "step": 19710
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9627904891967773,
      "learning_rate": 4.920276997427006e-06,
      "loss": 0.3489,
      "step": 19711
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.736271858215332,
      "learning_rate": 4.9193152419320846e-06,
      "loss": 0.3897,
      "step": 19712
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7097411155700684,
      "learning_rate": 4.918353549779964e-06,
      "loss": 0.3526,
      "step": 19713
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4941346645355225,
      "learning_rate": 4.9173919209826404e-06,
      "loss": 0.3547,
      "step": 19714
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7368923425674438,
      "learning_rate": 4.9164303555521e-06,
      "loss": 0.3128,
      "step": 19715
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6843751668930054,
      "learning_rate": 4.915468853500329e-06,
      "loss": 0.283,
      "step": 19716
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.0839731693267822,
      "learning_rate": 4.914507414839313e-06,
      "loss": 0.3431,
      "step": 19717
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.149771213531494,
      "learning_rate": 4.913546039581046e-06,
      "loss": 0.3908,
      "step": 19718
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7860521078109741,
      "learning_rate": 4.91258472773751e-06,
      "loss": 0.3168,
      "step": 19719
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.0351758003234863,
      "learning_rate": 4.9116234793206865e-06,
      "loss": 0.4155,
      "step": 19720
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.2921011447906494,
      "learning_rate": 4.910662294342566e-06,
      "loss": 0.311,
      "step": 19721
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.919693112373352,
      "learning_rate": 4.909701172815128e-06,
      "loss": 0.3885,
      "step": 19722
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.896547555923462,
      "learning_rate": 4.908740114750353e-06,
      "loss": 0.4279,
      "step": 19723
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.872544288635254,
      "learning_rate": 4.90777912016023e-06,
      "loss": 0.4514,
      "step": 19724
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.980411171913147,
      "learning_rate": 4.906818189056736e-06,
      "loss": 0.373,
      "step": 19725
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5598363876342773,
      "learning_rate": 4.905857321451847e-06,
      "loss": 0.3078,
      "step": 19726
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7478655576705933,
      "learning_rate": 4.904896517357551e-06,
      "loss": 0.3198,
      "step": 19727
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5581721067428589,
      "learning_rate": 4.903935776785821e-06,
      "loss": 0.2844,
      "step": 19728
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.906369686126709,
      "learning_rate": 4.902975099748638e-06,
      "loss": 0.41,
      "step": 19729
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7584742307662964,
      "learning_rate": 4.902014486257973e-06,
      "loss": 0.3342,
      "step": 19730
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6243486404418945,
      "learning_rate": 4.901053936325811e-06,
      "loss": 0.3624,
      "step": 19731
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.650831699371338,
      "learning_rate": 4.9000934499641236e-06,
      "loss": 0.376,
      "step": 19732
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7924251556396484,
      "learning_rate": 4.899133027184882e-06,
      "loss": 0.362,
      "step": 19733
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5351184606552124,
      "learning_rate": 4.898172668000066e-06,
      "loss": 0.3432,
      "step": 19734
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9802827835083008,
      "learning_rate": 4.897212372421647e-06,
      "loss": 0.3939,
      "step": 19735
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5954476594924927,
      "learning_rate": 4.896252140461593e-06,
      "loss": 0.31,
      "step": 19736
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5675369501113892,
      "learning_rate": 4.895291972131883e-06,
      "loss": 0.3048,
      "step": 19737
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.2123546600341797,
      "learning_rate": 4.894331867444483e-06,
      "loss": 0.3068,
      "step": 19738
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.3011233806610107,
      "learning_rate": 4.893371826411362e-06,
      "loss": 0.3361,
      "step": 19739
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.711864948272705,
      "learning_rate": 4.892411849044493e-06,
      "loss": 0.3339,
      "step": 19740
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.994752287864685,
      "learning_rate": 4.891451935355843e-06,
      "loss": 0.3392,
      "step": 19741
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.0789740085601807,
      "learning_rate": 4.89049208535738e-06,
      "loss": 0.4313,
      "step": 19742
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.140385627746582,
      "learning_rate": 4.889532299061065e-06,
      "loss": 0.3808,
      "step": 19743
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5916234254837036,
      "learning_rate": 4.888572576478874e-06,
      "loss": 0.3595,
      "step": 19744
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9357737302780151,
      "learning_rate": 4.887612917622766e-06,
      "loss": 0.2839,
      "step": 19745
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5499190092086792,
      "learning_rate": 4.886653322504704e-06,
      "loss": 0.3463,
      "step": 19746
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5348504781723022,
      "learning_rate": 4.8856937911366574e-06,
      "loss": 0.2493,
      "step": 19747
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7937564849853516,
      "learning_rate": 4.884734323530586e-06,
      "loss": 0.3441,
      "step": 19748
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.0844554901123047,
      "learning_rate": 4.883774919698447e-06,
      "loss": 0.3279,
      "step": 19749
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6181471347808838,
      "learning_rate": 4.882815579652211e-06,
      "loss": 0.3134,
      "step": 19750
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6246808767318726,
      "learning_rate": 4.881856303403835e-06,
      "loss": 0.278,
      "step": 19751
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7410569190979004,
      "learning_rate": 4.880897090965272e-06,
      "loss": 0.3624,
      "step": 19752
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.521332025527954,
      "learning_rate": 4.879937942348491e-06,
      "loss": 0.3459,
      "step": 19753
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7973552942276,
      "learning_rate": 4.878978857565445e-06,
      "loss": 0.2699,
      "step": 19754
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.3699471950531006,
      "learning_rate": 4.878019836628092e-06,
      "loss": 0.3304,
      "step": 19755
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.1033356189727783,
      "learning_rate": 4.877060879548384e-06,
      "loss": 0.3266,
      "step": 19756
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3235574960708618,
      "learning_rate": 4.8761019863382855e-06,
      "loss": 0.2395,
      "step": 19757
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7880886793136597,
      "learning_rate": 4.875143157009745e-06,
      "loss": 0.364,
      "step": 19758
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9678683280944824,
      "learning_rate": 4.874184391574716e-06,
      "loss": 0.3326,
      "step": 19759
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6745771169662476,
      "learning_rate": 4.873225690045158e-06,
      "loss": 0.4332,
      "step": 19760
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9251205921173096,
      "learning_rate": 4.872267052433019e-06,
      "loss": 0.3239,
      "step": 19761
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9702948331832886,
      "learning_rate": 4.8713084787502475e-06,
      "loss": 0.4016,
      "step": 19762
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7987654209136963,
      "learning_rate": 4.870349969008802e-06,
      "loss": 0.3459,
      "step": 19763
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.741072416305542,
      "learning_rate": 4.869391523220629e-06,
      "loss": 0.3472,
      "step": 19764
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9628257751464844,
      "learning_rate": 4.868433141397672e-06,
      "loss": 0.2968,
      "step": 19765
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.001293897628784,
      "learning_rate": 4.867474823551892e-06,
      "loss": 0.3993,
      "step": 19766
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7331690788269043,
      "learning_rate": 4.866516569695228e-06,
      "loss": 0.3131,
      "step": 19767
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8210291862487793,
      "learning_rate": 4.8655583798396286e-06,
      "loss": 0.3448,
      "step": 19768
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7032586336135864,
      "learning_rate": 4.864600253997037e-06,
      "loss": 0.3074,
      "step": 19769
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.698530912399292,
      "learning_rate": 4.863642192179405e-06,
      "loss": 0.3501,
      "step": 19770
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.9957101345062256,
      "learning_rate": 4.862684194398674e-06,
      "loss": 0.4443,
      "step": 19771
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.8029999732971191,
      "learning_rate": 4.861726260666782e-06,
      "loss": 0.3691,
      "step": 19772
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5828020572662354,
      "learning_rate": 4.860768390995683e-06,
      "loss": 0.3918,
      "step": 19773
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4768271446228027,
      "learning_rate": 4.859810585397312e-06,
      "loss": 0.2916,
      "step": 19774
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7060682773590088,
      "learning_rate": 4.858852843883609e-06,
      "loss": 0.3331,
      "step": 19775
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9136087894439697,
      "learning_rate": 4.8578951664665205e-06,
      "loss": 0.3892,
      "step": 19776
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7117921113967896,
      "learning_rate": 4.856937553157983e-06,
      "loss": 0.3275,
      "step": 19777
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.0945634841918945,
      "learning_rate": 4.8559800039699325e-06,
      "loss": 0.3254,
      "step": 19778
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4960781335830688,
      "learning_rate": 4.855022518914313e-06,
      "loss": 0.3248,
      "step": 19779
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8120603561401367,
      "learning_rate": 4.854065098003058e-06,
      "loss": 0.2653,
      "step": 19780
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6563823223114014,
      "learning_rate": 4.853107741248106e-06,
      "loss": 0.3903,
      "step": 19781
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.828639030456543,
      "learning_rate": 4.852150448661388e-06,
      "loss": 0.2726,
      "step": 19782
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.113926410675049,
      "learning_rate": 4.851193220254847e-06,
      "loss": 0.4002,
      "step": 19783
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7735965251922607,
      "learning_rate": 4.85023605604041e-06,
      "loss": 0.2991,
      "step": 19784
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5401376485824585,
      "learning_rate": 4.849278956030011e-06,
      "loss": 0.3014,
      "step": 19785
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8065050840377808,
      "learning_rate": 4.848321920235588e-06,
      "loss": 0.3107,
      "step": 19786
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6173368692398071,
      "learning_rate": 4.84736494866907e-06,
      "loss": 0.2885,
      "step": 19787
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9294909238815308,
      "learning_rate": 4.846408041342382e-06,
      "loss": 0.2671,
      "step": 19788
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9241604804992676,
      "learning_rate": 4.845451198267463e-06,
      "loss": 0.2883,
      "step": 19789
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.875146508216858,
      "learning_rate": 4.84449441945624e-06,
      "loss": 0.2484,
      "step": 19790
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4936541318893433,
      "learning_rate": 4.843537704920636e-06,
      "loss": 0.367,
      "step": 19791
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.1844332218170166,
      "learning_rate": 4.842581054672586e-06,
      "loss": 0.3246,
      "step": 19792
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7805490493774414,
      "learning_rate": 4.841624468724015e-06,
      "loss": 0.3062,
      "step": 19793
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.90134859085083,
      "learning_rate": 4.840667947086847e-06,
      "loss": 0.3,
      "step": 19794
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9824135303497314,
      "learning_rate": 4.839711489773003e-06,
      "loss": 0.343,
      "step": 19795
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.602763295173645,
      "learning_rate": 4.8387550967944185e-06,
      "loss": 0.3074,
      "step": 19796
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7782803773880005,
      "learning_rate": 4.83779876816301e-06,
      "loss": 0.3748,
      "step": 19797
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5877124071121216,
      "learning_rate": 4.8368425038907e-06,
      "loss": 0.2601,
      "step": 19798
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.0565807819366455,
      "learning_rate": 4.835886303989415e-06,
      "loss": 0.2901,
      "step": 19799
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5075567960739136,
      "learning_rate": 4.8349301684710725e-06,
      "loss": 0.3596,
      "step": 19800
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6062781810760498,
      "learning_rate": 4.833974097347592e-06,
      "loss": 0.3048,
      "step": 19801
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.118182420730591,
      "learning_rate": 4.8330180906308984e-06,
      "loss": 0.3502,
      "step": 19802
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9374301433563232,
      "learning_rate": 4.832062148332909e-06,
      "loss": 0.3209,
      "step": 19803
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.799462914466858,
      "learning_rate": 4.8311062704655355e-06,
      "loss": 0.275,
      "step": 19804
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7518845796585083,
      "learning_rate": 4.830150457040704e-06,
      "loss": 0.3194,
      "step": 19805
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6619571447372437,
      "learning_rate": 4.8291947080703284e-06,
      "loss": 0.3177,
      "step": 19806
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.779056429862976,
      "learning_rate": 4.8282390235663216e-06,
      "loss": 0.2418,
      "step": 19807
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.418849468231201,
      "learning_rate": 4.827283403540597e-06,
      "loss": 0.3981,
      "step": 19808
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8988189697265625,
      "learning_rate": 4.8263278480050755e-06,
      "loss": 0.3056,
      "step": 19809
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5520875453948975,
      "learning_rate": 4.825372356971666e-06,
      "loss": 0.3053,
      "step": 19810
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9701290130615234,
      "learning_rate": 4.8244169304522785e-06,
      "loss": 0.3562,
      "step": 19811
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.615909218788147,
      "learning_rate": 4.823461568458832e-06,
      "loss": 0.3674,
      "step": 19812
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.625347375869751,
      "learning_rate": 4.822506271003231e-06,
      "loss": 0.3204,
      "step": 19813
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.036677837371826,
      "learning_rate": 4.821551038097386e-06,
      "loss": 0.3627,
      "step": 19814
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6382468938827515,
      "learning_rate": 4.820595869753211e-06,
      "loss": 0.3384,
      "step": 19815
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.761315941810608,
      "learning_rate": 4.819640765982611e-06,
      "loss": 0.3219,
      "step": 19816
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.588348627090454,
      "learning_rate": 4.818685726797489e-06,
      "loss": 0.2637,
      "step": 19817
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9619462490081787,
      "learning_rate": 4.817730752209763e-06,
      "loss": 0.3769,
      "step": 19818
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8997501134872437,
      "learning_rate": 4.816775842231332e-06,
      "loss": 0.4327,
      "step": 19819
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6963151693344116,
      "learning_rate": 4.815820996874101e-06,
      "loss": 0.208,
      "step": 19820
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.0606424808502197,
      "learning_rate": 4.814866216149973e-06,
      "loss": 0.2642,
      "step": 19821
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.208216905593872,
      "learning_rate": 4.813911500070857e-06,
      "loss": 0.3344,
      "step": 19822
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.938366413116455,
      "learning_rate": 4.812956848648652e-06,
      "loss": 0.3038,
      "step": 19823
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5006061792373657,
      "learning_rate": 4.812002261895259e-06,
      "loss": 0.2534,
      "step": 19824
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.789384126663208,
      "learning_rate": 4.811047739822584e-06,
      "loss": 0.2943,
      "step": 19825
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.097536563873291,
      "learning_rate": 4.810093282442524e-06,
      "loss": 0.3184,
      "step": 19826
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.5035319328308105,
      "learning_rate": 4.809138889766976e-06,
      "loss": 0.3373,
      "step": 19827
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7366732358932495,
      "learning_rate": 4.808184561807845e-06,
      "loss": 0.2362,
      "step": 19828
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7665647268295288,
      "learning_rate": 4.807230298577026e-06,
      "loss": 0.3559,
      "step": 19829
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.925719976425171,
      "learning_rate": 4.806276100086412e-06,
      "loss": 0.3885,
      "step": 19830
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5305856466293335,
      "learning_rate": 4.805321966347907e-06,
      "loss": 0.3392,
      "step": 19831
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5556411743164062,
      "learning_rate": 4.804367897373402e-06,
      "loss": 0.2333,
      "step": 19832
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.201481580734253,
      "learning_rate": 4.8034138931747945e-06,
      "loss": 0.2886,
      "step": 19833
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9418046474456787,
      "learning_rate": 4.802459953763971e-06,
      "loss": 0.3214,
      "step": 19834
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5311882495880127,
      "learning_rate": 4.801506079152835e-06,
      "loss": 0.2595,
      "step": 19835
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6749773025512695,
      "learning_rate": 4.800552269353273e-06,
      "loss": 0.3344,
      "step": 19836
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8361155986785889,
      "learning_rate": 4.7995985243771735e-06,
      "loss": 0.3231,
      "step": 19837
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.314287781715393,
      "learning_rate": 4.798644844236435e-06,
      "loss": 0.2898,
      "step": 19838
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.1537013053894043,
      "learning_rate": 4.797691228942945e-06,
      "loss": 0.3085,
      "step": 19839
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.84871244430542,
      "learning_rate": 4.796737678508585e-06,
      "loss": 0.2835,
      "step": 19840
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5780385732650757,
      "learning_rate": 4.795784192945254e-06,
      "loss": 0.3396,
      "step": 19841
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.235961437225342,
      "learning_rate": 4.7948307722648355e-06,
      "loss": 0.3728,
      "step": 19842
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9436519145965576,
      "learning_rate": 4.793877416479212e-06,
      "loss": 0.3914,
      "step": 19843
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.525929570198059,
      "learning_rate": 4.792924125600275e-06,
      "loss": 0.318,
      "step": 19844
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9406026601791382,
      "learning_rate": 4.7919708996399084e-06,
      "loss": 0.3681,
      "step": 19845
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.221099853515625,
      "learning_rate": 4.791017738609994e-06,
      "loss": 0.4343,
      "step": 19846
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8491084575653076,
      "learning_rate": 4.790064642522414e-06,
      "loss": 0.1898,
      "step": 19847
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4375287294387817,
      "learning_rate": 4.789111611389058e-06,
      "loss": 0.2183,
      "step": 19848
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5686427354812622,
      "learning_rate": 4.788158645221802e-06,
      "loss": 0.3197,
      "step": 19849
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7988431453704834,
      "learning_rate": 4.787205744032526e-06,
      "loss": 0.346,
      "step": 19850
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.40167498588562,
      "learning_rate": 4.786252907833114e-06,
      "loss": 0.4177,
      "step": 19851
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.1497933864593506,
      "learning_rate": 4.785300136635446e-06,
      "loss": 0.3986,
      "step": 19852
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.1474130153656006,
      "learning_rate": 4.784347430451394e-06,
      "loss": 0.4072,
      "step": 19853
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.172492027282715,
      "learning_rate": 4.783394789292844e-06,
      "loss": 0.3759,
      "step": 19854
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.261366605758667,
      "learning_rate": 4.78244221317167e-06,
      "loss": 0.3025,
      "step": 19855
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6517647504806519,
      "learning_rate": 4.781489702099742e-06,
      "loss": 0.3297,
      "step": 19856
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7186752557754517,
      "learning_rate": 4.780537256088945e-06,
      "loss": 0.236,
      "step": 19857
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.9880573749542236,
      "learning_rate": 4.77958487515115e-06,
      "loss": 0.339,
      "step": 19858
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5770950317382812,
      "learning_rate": 4.778632559298229e-06,
      "loss": 0.3033,
      "step": 19859
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.696919560432434,
      "learning_rate": 4.777680308542051e-06,
      "loss": 0.3169,
      "step": 19860
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.613603949546814,
      "learning_rate": 4.776728122894497e-06,
      "loss": 0.2783,
      "step": 19861
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6315430402755737,
      "learning_rate": 4.7757760023674345e-06,
      "loss": 0.246,
      "step": 19862
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8661030530929565,
      "learning_rate": 4.77482394697273e-06,
      "loss": 0.2975,
      "step": 19863
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.8857420682907104,
      "learning_rate": 4.773871956722259e-06,
      "loss": 0.3551,
      "step": 19864
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.3559117317199707,
      "learning_rate": 4.772920031627889e-06,
      "loss": 0.4055,
      "step": 19865
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.0861282348632812,
      "learning_rate": 4.771968171701482e-06,
      "loss": 0.4016,
      "step": 19866
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.240748167037964,
      "learning_rate": 4.771016376954914e-06,
      "loss": 0.4221,
      "step": 19867
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.464252233505249,
      "learning_rate": 4.770064647400049e-06,
      "loss": 0.2652,
      "step": 19868
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.14536190032959,
      "learning_rate": 4.769112983048746e-06,
      "loss": 0.3818,
      "step": 19869
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7374787330627441,
      "learning_rate": 4.768161383912879e-06,
      "loss": 0.2519,
      "step": 19870
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.6785606145858765,
      "learning_rate": 4.767209850004307e-06,
      "loss": 0.3454,
      "step": 19871
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.641884207725525,
      "learning_rate": 4.766258381334895e-06,
      "loss": 0.3462,
      "step": 19872
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.3276467323303223,
      "learning_rate": 4.7653069779165e-06,
      "loss": 0.3541,
      "step": 19873
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.108283519744873,
      "learning_rate": 4.76435563976099e-06,
      "loss": 0.3252,
      "step": 19874
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7670549154281616,
      "learning_rate": 4.763404366880226e-06,
      "loss": 0.2391,
      "step": 19875
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0256967544555664,
      "learning_rate": 4.762453159286059e-06,
      "loss": 0.3381,
      "step": 19876
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.577581763267517,
      "learning_rate": 4.761502016990359e-06,
      "loss": 0.3375,
      "step": 19877
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7822962999343872,
      "learning_rate": 4.76055094000498e-06,
      "loss": 0.2929,
      "step": 19878
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9439573287963867,
      "learning_rate": 4.759599928341775e-06,
      "loss": 0.3298,
      "step": 19879
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8572168350219727,
      "learning_rate": 4.758648982012608e-06,
      "loss": 0.3108,
      "step": 19880
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.125620126724243,
      "learning_rate": 4.757698101029332e-06,
      "loss": 0.3309,
      "step": 19881
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0937302112579346,
      "learning_rate": 4.7567472854037976e-06,
      "loss": 0.3254,
      "step": 19882
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9814467430114746,
      "learning_rate": 4.755796535147867e-06,
      "loss": 0.3063,
      "step": 19883
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5169495344161987,
      "learning_rate": 4.754845850273387e-06,
      "loss": 0.3416,
      "step": 19884
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.1771676540374756,
      "learning_rate": 4.753895230792215e-06,
      "loss": 0.3445,
      "step": 19885
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9031096696853638,
      "learning_rate": 4.752944676716196e-06,
      "loss": 0.3462,
      "step": 19886
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0336482524871826,
      "learning_rate": 4.7519941880571875e-06,
      "loss": 0.309,
      "step": 19887
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9232343435287476,
      "learning_rate": 4.751043764827038e-06,
      "loss": 0.3254,
      "step": 19888
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7422382831573486,
      "learning_rate": 4.750093407037593e-06,
      "loss": 0.3496,
      "step": 19889
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9113575220108032,
      "learning_rate": 4.7491431147007075e-06,
      "loss": 0.3078,
      "step": 19890
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.9684321880340576,
      "learning_rate": 4.748192887828224e-06,
      "loss": 0.3404,
      "step": 19891
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0323400497436523,
      "learning_rate": 4.747242726431989e-06,
      "loss": 0.3859,
      "step": 19892
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.208548069000244,
      "learning_rate": 4.746292630523855e-06,
      "loss": 0.3749,
      "step": 19893
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.920196294784546,
      "learning_rate": 4.74534260011566e-06,
      "loss": 0.3359,
      "step": 19894
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.029956102371216,
      "learning_rate": 4.74439263521925e-06,
      "loss": 0.3618,
      "step": 19895
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.6478726863861084,
      "learning_rate": 4.74344273584647e-06,
      "loss": 0.3052,
      "step": 19896
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9401370286941528,
      "learning_rate": 4.7424929020091655e-06,
      "loss": 0.2775,
      "step": 19897
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.4711406230926514,
      "learning_rate": 4.741543133719174e-06,
      "loss": 0.338,
      "step": 19898
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.1955809593200684,
      "learning_rate": 4.740593430988335e-06,
      "loss": 0.3569,
      "step": 19899
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8362412452697754,
      "learning_rate": 4.739643793828494e-06,
      "loss": 0.3116,
      "step": 19900
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.273170232772827,
      "learning_rate": 4.73869422225149e-06,
      "loss": 0.333,
      "step": 19901
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0255393981933594,
      "learning_rate": 4.737744716269156e-06,
      "loss": 0.3233,
      "step": 19902
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9114445447921753,
      "learning_rate": 4.736795275893337e-06,
      "loss": 0.3534,
      "step": 19903
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0477540493011475,
      "learning_rate": 4.735845901135866e-06,
      "loss": 0.309,
      "step": 19904
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.445328950881958,
      "learning_rate": 4.734896592008577e-06,
      "loss": 0.4294,
      "step": 19905
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7971528768539429,
      "learning_rate": 4.733947348523311e-06,
      "loss": 0.3134,
      "step": 19906
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.2151598930358887,
      "learning_rate": 4.732998170691902e-06,
      "loss": 0.3531,
      "step": 19907
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.3650615215301514,
      "learning_rate": 4.732049058526176e-06,
      "loss": 0.4129,
      "step": 19908
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5419710874557495,
      "learning_rate": 4.731100012037977e-06,
      "loss": 0.2269,
      "step": 19909
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.3108885288238525,
      "learning_rate": 4.730151031239131e-06,
      "loss": 0.3924,
      "step": 19910
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.623767375946045,
      "learning_rate": 4.729202116141469e-06,
      "loss": 0.354,
      "step": 19911
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.490479588508606,
      "learning_rate": 4.72825326675682e-06,
      "loss": 0.3296,
      "step": 19912
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.812085747718811,
      "learning_rate": 4.727304483097021e-06,
      "loss": 0.3044,
      "step": 19913
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.263852596282959,
      "learning_rate": 4.726355765173895e-06,
      "loss": 0.3573,
      "step": 19914
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.6176756620407104,
      "learning_rate": 4.725407112999266e-06,
      "loss": 0.3434,
      "step": 19915
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4197248220443726,
      "learning_rate": 4.724458526584972e-06,
      "loss": 0.2531,
      "step": 19916
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.050637722015381,
      "learning_rate": 4.723510005942833e-06,
      "loss": 0.3686,
      "step": 19917
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.904910922050476,
      "learning_rate": 4.722561551084671e-06,
      "loss": 0.3547,
      "step": 19918
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9216452836990356,
      "learning_rate": 4.721613162022318e-06,
      "loss": 0.3682,
      "step": 19919
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.851480484008789,
      "learning_rate": 4.720664838767597e-06,
      "loss": 0.255,
      "step": 19920
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7654861211776733,
      "learning_rate": 4.719716581332324e-06,
      "loss": 0.3525,
      "step": 19921
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8920425176620483,
      "learning_rate": 4.71876838972833e-06,
      "loss": 0.3396,
      "step": 19922
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9193778038024902,
      "learning_rate": 4.717820263967433e-06,
      "loss": 0.3798,
      "step": 19923
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5361824035644531,
      "learning_rate": 4.716872204061453e-06,
      "loss": 0.2823,
      "step": 19924
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8524495363235474,
      "learning_rate": 4.715924210022207e-06,
      "loss": 0.3414,
      "step": 19925
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4621222019195557,
      "learning_rate": 4.714976281861519e-06,
      "loss": 0.2467,
      "step": 19926
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5392754077911377,
      "learning_rate": 4.714028419591208e-06,
      "loss": 0.2275,
      "step": 19927
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.6370129585266113,
      "learning_rate": 4.713080623223084e-06,
      "loss": 0.3516,
      "step": 19928
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.245347261428833,
      "learning_rate": 4.712132892768972e-06,
      "loss": 0.286,
      "step": 19929
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.2308125495910645,
      "learning_rate": 4.711185228240684e-06,
      "loss": 0.3679,
      "step": 19930
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.659122109413147,
      "learning_rate": 4.710237629650031e-06,
      "loss": 0.2888,
      "step": 19931
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8012446165084839,
      "learning_rate": 4.709290097008834e-06,
      "loss": 0.2937,
      "step": 19932
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8386869430541992,
      "learning_rate": 4.708342630328904e-06,
      "loss": 0.3282,
      "step": 19933
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.323624849319458,
      "learning_rate": 4.707395229622048e-06,
      "loss": 0.3256,
      "step": 19934
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7617546319961548,
      "learning_rate": 4.706447894900087e-06,
      "loss": 0.3827,
      "step": 19935
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4929386377334595,
      "learning_rate": 4.705500626174827e-06,
      "loss": 0.3633,
      "step": 19936
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.8385298252105713,
      "learning_rate": 4.704553423458077e-06,
      "loss": 0.318,
      "step": 19937
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0287299156188965,
      "learning_rate": 4.703606286761644e-06,
      "loss": 0.318,
      "step": 19938
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.1333212852478027,
      "learning_rate": 4.702659216097343e-06,
      "loss": 0.4459,
      "step": 19939
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5732767581939697,
      "learning_rate": 4.7017122114769775e-06,
      "loss": 0.2964,
      "step": 19940
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8575025796890259,
      "learning_rate": 4.700765272912351e-06,
      "loss": 0.3629,
      "step": 19941
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5702310800552368,
      "learning_rate": 4.699818400415276e-06,
      "loss": 0.2892,
      "step": 19942
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7684208154678345,
      "learning_rate": 4.698871593997555e-06,
      "loss": 0.334,
      "step": 19943
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.6535794734954834,
      "learning_rate": 4.697924853670988e-06,
      "loss": 0.3909,
      "step": 19944
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9390779733657837,
      "learning_rate": 4.696978179447385e-06,
      "loss": 0.2615,
      "step": 19945
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.1215436458587646,
      "learning_rate": 4.696031571338545e-06,
      "loss": 0.3588,
      "step": 19946
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9299721717834473,
      "learning_rate": 4.695085029356268e-06,
      "loss": 0.2728,
      "step": 19947
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7029160261154175,
      "learning_rate": 4.694138553512359e-06,
      "loss": 0.3347,
      "step": 19948
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8348466157913208,
      "learning_rate": 4.693192143818618e-06,
      "loss": 0.3566,
      "step": 19949
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.676260232925415,
      "learning_rate": 4.692245800286841e-06,
      "loss": 0.359,
      "step": 19950
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0662882328033447,
      "learning_rate": 4.691299522928825e-06,
      "loss": 0.3942,
      "step": 19951
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.38895583152771,
      "learning_rate": 4.690353311756373e-06,
      "loss": 0.3292,
      "step": 19952
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8968980312347412,
      "learning_rate": 4.68940716678128e-06,
      "loss": 0.3553,
      "step": 19953
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7399654388427734,
      "learning_rate": 4.688461088015337e-06,
      "loss": 0.3324,
      "step": 19954
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5166751146316528,
      "learning_rate": 4.6875150754703466e-06,
      "loss": 0.2947,
      "step": 19955
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7541420459747314,
      "learning_rate": 4.686569129158101e-06,
      "loss": 0.3629,
      "step": 19956
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9952850341796875,
      "learning_rate": 4.685623249090387e-06,
      "loss": 0.333,
      "step": 19957
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.3976144790649414,
      "learning_rate": 4.684677435279008e-06,
      "loss": 0.4093,
      "step": 19958
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.6642237901687622,
      "learning_rate": 4.68373168773575e-06,
      "loss": 0.3276,
      "step": 19959
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0050902366638184,
      "learning_rate": 4.682786006472404e-06,
      "loss": 0.3388,
      "step": 19960
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0933752059936523,
      "learning_rate": 4.681840391500757e-06,
      "loss": 0.3216,
      "step": 19961
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8536760807037354,
      "learning_rate": 4.680894842832606e-06,
      "loss": 0.3996,
      "step": 19962
    },
    {
      "epoch": 2.06,
      "grad_norm": 3.0164573192596436,
      "learning_rate": 4.679949360479735e-06,
      "loss": 0.3237,
      "step": 19963
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7865188121795654,
      "learning_rate": 4.6790039444539295e-06,
      "loss": 0.3447,
      "step": 19964
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.0677103996276855,
      "learning_rate": 4.678058594766981e-06,
      "loss": 0.3662,
      "step": 19965
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.7589867115020752,
      "learning_rate": 4.677113311430674e-06,
      "loss": 0.3366,
      "step": 19966
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8463644981384277,
      "learning_rate": 4.67616809445679e-06,
      "loss": 0.3614,
      "step": 19967
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8369699716567993,
      "learning_rate": 4.675222943857121e-06,
      "loss": 0.2881,
      "step": 19968
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9242773056030273,
      "learning_rate": 4.674277859643445e-06,
      "loss": 0.3313,
      "step": 19969
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.036940336227417,
      "learning_rate": 4.6733328418275425e-06,
      "loss": 0.3751,
      "step": 19970
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8704198598861694,
      "learning_rate": 4.672387890421202e-06,
      "loss": 0.3931,
      "step": 19971
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6165735721588135,
      "learning_rate": 4.6714430054362015e-06,
      "loss": 0.3136,
      "step": 19972
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8665531873703003,
      "learning_rate": 4.6704981868843214e-06,
      "loss": 0.3267,
      "step": 19973
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7060490846633911,
      "learning_rate": 4.669553434777336e-06,
      "loss": 0.3486,
      "step": 19974
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4227418899536133,
      "learning_rate": 4.668608749127034e-06,
      "loss": 0.3029,
      "step": 19975
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7478225231170654,
      "learning_rate": 4.667664129945187e-06,
      "loss": 0.3296,
      "step": 19976
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7767716646194458,
      "learning_rate": 4.666719577243569e-06,
      "loss": 0.3508,
      "step": 19977
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9095261096954346,
      "learning_rate": 4.665775091033964e-06,
      "loss": 0.3899,
      "step": 19978
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9825454950332642,
      "learning_rate": 4.664830671328142e-06,
      "loss": 0.2777,
      "step": 19979
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8457647562026978,
      "learning_rate": 4.663886318137876e-06,
      "loss": 0.3248,
      "step": 19980
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.651003122329712,
      "learning_rate": 4.662942031474945e-06,
      "loss": 0.3004,
      "step": 19981
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9667497873306274,
      "learning_rate": 4.661997811351119e-06,
      "loss": 0.2637,
      "step": 19982
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.866520881652832,
      "learning_rate": 4.661053657778166e-06,
      "loss": 0.3612,
      "step": 19983
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7188743352890015,
      "learning_rate": 4.660109570767865e-06,
      "loss": 0.3126,
      "step": 19984
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.660182237625122,
      "learning_rate": 4.659165550331982e-06,
      "loss": 0.3666,
      "step": 19985
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.688570499420166,
      "learning_rate": 4.6582215964822876e-06,
      "loss": 0.3065,
      "step": 19986
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4965168237686157,
      "learning_rate": 4.657277709230544e-06,
      "loss": 0.2862,
      "step": 19987
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6737606525421143,
      "learning_rate": 4.65633388858853e-06,
      "loss": 0.4007,
      "step": 19988
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8595308065414429,
      "learning_rate": 4.655390134568006e-06,
      "loss": 0.3274,
      "step": 19989
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6535826921463013,
      "learning_rate": 4.654446447180736e-06,
      "loss": 0.3557,
      "step": 19990
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.713039517402649,
      "learning_rate": 4.6535028264384915e-06,
      "loss": 0.3089,
      "step": 19991
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9052764177322388,
      "learning_rate": 4.6525592723530355e-06,
      "loss": 0.2975,
      "step": 19992
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4765862226486206,
      "learning_rate": 4.651615784936126e-06,
      "loss": 0.2974,
      "step": 19993
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4547313451766968,
      "learning_rate": 4.6506723641995325e-06,
      "loss": 0.3164,
      "step": 19994
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9256104230880737,
      "learning_rate": 4.649729010155015e-06,
      "loss": 0.2927,
      "step": 19995
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8316781520843506,
      "learning_rate": 4.648785722814331e-06,
      "loss": 0.3075,
      "step": 19996
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9721856117248535,
      "learning_rate": 4.647842502189247e-06,
      "loss": 0.3,
      "step": 19997
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8007348775863647,
      "learning_rate": 4.646899348291519e-06,
      "loss": 0.3329,
      "step": 19998
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7911468744277954,
      "learning_rate": 4.645956261132907e-06,
      "loss": 0.3098,
      "step": 19999
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5581936836242676,
      "learning_rate": 4.645013240725164e-06,
      "loss": 0.3706,
      "step": 20000
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7395151853561401,
      "learning_rate": 4.6440702870800535e-06,
      "loss": 0.3189,
      "step": 20001
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.664904236793518,
      "learning_rate": 4.64312740020933e-06,
      "loss": 0.3649,
      "step": 20002
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.879777431488037,
      "learning_rate": 4.642184580124743e-06,
      "loss": 0.3304,
      "step": 20003
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4378920793533325,
      "learning_rate": 4.641241826838057e-06,
      "loss": 0.2943,
      "step": 20004
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5796864032745361,
      "learning_rate": 4.64029914036102e-06,
      "loss": 0.2123,
      "step": 20005
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7523444890975952,
      "learning_rate": 4.639356520705382e-06,
      "loss": 0.3704,
      "step": 20006
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.3663198947906494,
      "learning_rate": 4.638413967882901e-06,
      "loss": 0.3667,
      "step": 20007
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9409035444259644,
      "learning_rate": 4.637471481905326e-06,
      "loss": 0.38,
      "step": 20008
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9724409580230713,
      "learning_rate": 4.636529062784403e-06,
      "loss": 0.375,
      "step": 20009
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.722888708114624,
      "learning_rate": 4.635586710531889e-06,
      "loss": 0.2659,
      "step": 20010
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9007991552352905,
      "learning_rate": 4.634644425159529e-06,
      "loss": 0.3094,
      "step": 20011
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.1250905990600586,
      "learning_rate": 4.6337022066790705e-06,
      "loss": 0.3287,
      "step": 20012
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.020019769668579,
      "learning_rate": 4.632760055102258e-06,
      "loss": 0.3506,
      "step": 20013
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6298060417175293,
      "learning_rate": 4.6318179704408426e-06,
      "loss": 0.3392,
      "step": 20014
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.223043203353882,
      "learning_rate": 4.630875952706569e-06,
      "loss": 0.2822,
      "step": 20015
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.0958993434906006,
      "learning_rate": 4.629934001911175e-06,
      "loss": 0.451,
      "step": 20016
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6882579326629639,
      "learning_rate": 4.628992118066415e-06,
      "loss": 0.3476,
      "step": 20017
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6619094610214233,
      "learning_rate": 4.628050301184024e-06,
      "loss": 0.3921,
      "step": 20018
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.2350144386291504,
      "learning_rate": 4.627108551275743e-06,
      "loss": 0.2825,
      "step": 20019
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7619107961654663,
      "learning_rate": 4.62616686835332e-06,
      "loss": 0.3154,
      "step": 20020
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8167375326156616,
      "learning_rate": 4.625225252428493e-06,
      "loss": 0.3309,
      "step": 20021
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7744803428649902,
      "learning_rate": 4.624283703512995e-06,
      "loss": 0.4134,
      "step": 20022
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9459916353225708,
      "learning_rate": 4.623342221618574e-06,
      "loss": 0.3109,
      "step": 20023
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.961662769317627,
      "learning_rate": 4.6224008067569634e-06,
      "loss": 0.3124,
      "step": 20024
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7374625205993652,
      "learning_rate": 4.621459458939901e-06,
      "loss": 0.234,
      "step": 20025
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.227703809738159,
      "learning_rate": 4.6205181781791175e-06,
      "loss": 0.3663,
      "step": 20026
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7830886840820312,
      "learning_rate": 4.6195769644863575e-06,
      "loss": 0.2573,
      "step": 20027
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8464304208755493,
      "learning_rate": 4.618635817873352e-06,
      "loss": 0.3829,
      "step": 20028
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8129671812057495,
      "learning_rate": 4.617694738351828e-06,
      "loss": 0.3426,
      "step": 20029
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5946135520935059,
      "learning_rate": 4.616753725933528e-06,
      "loss": 0.3388,
      "step": 20030
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.2040956020355225,
      "learning_rate": 4.61581278063018e-06,
      "loss": 0.366,
      "step": 20031
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8622502088546753,
      "learning_rate": 4.61487190245351e-06,
      "loss": 0.3288,
      "step": 20032
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.0506653785705566,
      "learning_rate": 4.613931091415259e-06,
      "loss": 0.367,
      "step": 20033
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8315463066101074,
      "learning_rate": 4.61299034752715e-06,
      "loss": 0.2146,
      "step": 20034
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4393938779830933,
      "learning_rate": 4.612049670800909e-06,
      "loss": 0.2732,
      "step": 20035
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.467402219772339,
      "learning_rate": 4.61110906124827e-06,
      "loss": 0.4069,
      "step": 20036
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8841643333435059,
      "learning_rate": 4.6101685188809565e-06,
      "loss": 0.2584,
      "step": 20037
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7576807737350464,
      "learning_rate": 4.609228043710696e-06,
      "loss": 0.2757,
      "step": 20038
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5189107656478882,
      "learning_rate": 4.608287635749212e-06,
      "loss": 0.4525,
      "step": 20039
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8422397375106812,
      "learning_rate": 4.607347295008231e-06,
      "loss": 0.3436,
      "step": 20040
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7750731706619263,
      "learning_rate": 4.60640702149947e-06,
      "loss": 0.2968,
      "step": 20041
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.331883668899536,
      "learning_rate": 4.605466815234663e-06,
      "loss": 0.3662,
      "step": 20042
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7182056903839111,
      "learning_rate": 4.6045266762255246e-06,
      "loss": 0.3595,
      "step": 20043
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9098641872406006,
      "learning_rate": 4.603586604483773e-06,
      "loss": 0.3871,
      "step": 20044
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8408952951431274,
      "learning_rate": 4.602646600021138e-06,
      "loss": 0.4016,
      "step": 20045
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8374134302139282,
      "learning_rate": 4.601706662849332e-06,
      "loss": 0.3273,
      "step": 20046
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5955915451049805,
      "learning_rate": 4.600766792980076e-06,
      "loss": 0.2853,
      "step": 20047
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.604943037033081,
      "learning_rate": 4.599826990425083e-06,
      "loss": 0.3561,
      "step": 20048
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7241178750991821,
      "learning_rate": 4.598887255196077e-06,
      "loss": 0.2892,
      "step": 20049
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.1467580795288086,
      "learning_rate": 4.597947587304772e-06,
      "loss": 0.3689,
      "step": 20050
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.2611236572265625,
      "learning_rate": 4.597007986762878e-06,
      "loss": 0.3028,
      "step": 20051
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7095078229904175,
      "learning_rate": 4.596068453582116e-06,
      "loss": 0.3785,
      "step": 20052
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6208924055099487,
      "learning_rate": 4.595128987774198e-06,
      "loss": 0.3306,
      "step": 20053
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5077825784683228,
      "learning_rate": 4.594189589350831e-06,
      "loss": 0.2838,
      "step": 20054
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9592413902282715,
      "learning_rate": 4.593250258323737e-06,
      "loss": 0.3323,
      "step": 20055
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.9741731882095337,
      "learning_rate": 4.5923109947046195e-06,
      "loss": 0.3068,
      "step": 20056
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.545757532119751,
      "learning_rate": 4.5913717985051876e-06,
      "loss": 0.3203,
      "step": 20057
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6733616590499878,
      "learning_rate": 4.5904326697371575e-06,
      "loss": 0.3159,
      "step": 20058
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7812554836273193,
      "learning_rate": 4.589493608412232e-06,
      "loss": 0.2648,
      "step": 20059
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4844510555267334,
      "learning_rate": 4.588554614542123e-06,
      "loss": 0.2308,
      "step": 20060
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.768227458000183,
      "learning_rate": 4.587615688138529e-06,
      "loss": 0.2772,
      "step": 20061
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2889389991760254,
      "learning_rate": 4.586676829213166e-06,
      "loss": 0.2998,
      "step": 20062
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7845333814620972,
      "learning_rate": 4.585738037777735e-06,
      "loss": 0.3016,
      "step": 20063
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7139495611190796,
      "learning_rate": 4.584799313843936e-06,
      "loss": 0.39,
      "step": 20064
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7387953996658325,
      "learning_rate": 4.58386065742348e-06,
      "loss": 0.3779,
      "step": 20065
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5153746604919434,
      "learning_rate": 4.582922068528066e-06,
      "loss": 0.2753,
      "step": 20066
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7929792404174805,
      "learning_rate": 4.581983547169393e-06,
      "loss": 0.3649,
      "step": 20067
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9416695833206177,
      "learning_rate": 4.581045093359168e-06,
      "loss": 0.3119,
      "step": 20068
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8466246128082275,
      "learning_rate": 4.580106707109088e-06,
      "loss": 0.3135,
      "step": 20069
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7830811738967896,
      "learning_rate": 4.579168388430848e-06,
      "loss": 0.3207,
      "step": 20070
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9647139310836792,
      "learning_rate": 4.5782301373361535e-06,
      "loss": 0.3564,
      "step": 20071
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7618685960769653,
      "learning_rate": 4.5772919538367e-06,
      "loss": 0.3373,
      "step": 20072
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8785895109176636,
      "learning_rate": 4.576353837944182e-06,
      "loss": 0.4153,
      "step": 20073
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7158708572387695,
      "learning_rate": 4.5754157896702934e-06,
      "loss": 0.3149,
      "step": 20074
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5887235403060913,
      "learning_rate": 4.574477809026735e-06,
      "loss": 0.3288,
      "step": 20075
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7107956409454346,
      "learning_rate": 4.573539896025199e-06,
      "loss": 0.2811,
      "step": 20076
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7954891920089722,
      "learning_rate": 4.572602050677374e-06,
      "loss": 0.4186,
      "step": 20077
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9154627323150635,
      "learning_rate": 4.57166427299496e-06,
      "loss": 0.3861,
      "step": 20078
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6634135246276855,
      "learning_rate": 4.570726562989645e-06,
      "loss": 0.2443,
      "step": 20079
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6744319200515747,
      "learning_rate": 4.569788920673116e-06,
      "loss": 0.3035,
      "step": 20080
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.802362084388733,
      "learning_rate": 4.56885134605707e-06,
      "loss": 0.4209,
      "step": 20081
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6122807264328003,
      "learning_rate": 4.567913839153193e-06,
      "loss": 0.3069,
      "step": 20082
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6230521202087402,
      "learning_rate": 4.56697639997317e-06,
      "loss": 0.3107,
      "step": 20083
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.1096441745758057,
      "learning_rate": 4.566039028528694e-06,
      "loss": 0.2917,
      "step": 20084
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.2277421951293945,
      "learning_rate": 4.56510172483145e-06,
      "loss": 0.264,
      "step": 20085
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7287617921829224,
      "learning_rate": 4.564164488893121e-06,
      "loss": 0.3604,
      "step": 20086
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.79881751537323,
      "learning_rate": 4.563227320725392e-06,
      "loss": 0.3652,
      "step": 20087
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8242172002792358,
      "learning_rate": 4.562290220339951e-06,
      "loss": 0.3441,
      "step": 20088
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9811500310897827,
      "learning_rate": 4.56135318774848e-06,
      "loss": 0.3169,
      "step": 20089
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0134682655334473,
      "learning_rate": 4.560416222962655e-06,
      "loss": 0.3027,
      "step": 20090
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5473331212997437,
      "learning_rate": 4.559479325994167e-06,
      "loss": 0.3372,
      "step": 20091
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.893041729927063,
      "learning_rate": 4.558542496854692e-06,
      "loss": 0.3263,
      "step": 20092
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6149309873580933,
      "learning_rate": 4.557605735555906e-06,
      "loss": 0.3001,
      "step": 20093
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6068967580795288,
      "learning_rate": 4.556669042109497e-06,
      "loss": 0.2632,
      "step": 20094
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.975128412246704,
      "learning_rate": 4.5557324165271375e-06,
      "loss": 0.2895,
      "step": 20095
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0076613426208496,
      "learning_rate": 4.554795858820502e-06,
      "loss": 0.3581,
      "step": 20096
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8288848400115967,
      "learning_rate": 4.553859369001273e-06,
      "loss": 0.2898,
      "step": 20097
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.378019332885742,
      "learning_rate": 4.552922947081125e-06,
      "loss": 0.3986,
      "step": 20098
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.972760558128357,
      "learning_rate": 4.55198659307173e-06,
      "loss": 0.3171,
      "step": 20099
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8411096334457397,
      "learning_rate": 4.55105030698476e-06,
      "loss": 0.2605,
      "step": 20100
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8310002088546753,
      "learning_rate": 4.550114088831894e-06,
      "loss": 0.3134,
      "step": 20101
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6719043254852295,
      "learning_rate": 4.549177938624802e-06,
      "loss": 0.3381,
      "step": 20102
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.083665132522583,
      "learning_rate": 4.548241856375149e-06,
      "loss": 0.3975,
      "step": 20103
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.985382318496704,
      "learning_rate": 4.5473058420946156e-06,
      "loss": 0.3362,
      "step": 20104
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0247180461883545,
      "learning_rate": 4.5463698957948675e-06,
      "loss": 0.355,
      "step": 20105
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6897532939910889,
      "learning_rate": 4.545434017487569e-06,
      "loss": 0.2674,
      "step": 20106
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0525946617126465,
      "learning_rate": 4.544498207184396e-06,
      "loss": 0.3623,
      "step": 20107
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4965640306472778,
      "learning_rate": 4.54356246489701e-06,
      "loss": 0.3459,
      "step": 20108
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.895466923713684,
      "learning_rate": 4.542626790637077e-06,
      "loss": 0.3438,
      "step": 20109
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.656382441520691,
      "learning_rate": 4.541691184416266e-06,
      "loss": 0.3104,
      "step": 20110
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8158613443374634,
      "learning_rate": 4.540755646246241e-06,
      "loss": 0.345,
      "step": 20111
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9653421640396118,
      "learning_rate": 4.539820176138664e-06,
      "loss": 0.3189,
      "step": 20112
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9785248041152954,
      "learning_rate": 4.538884774105195e-06,
      "loss": 0.3208,
      "step": 20113
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5346899032592773,
      "learning_rate": 4.5379494401575015e-06,
      "loss": 0.3286,
      "step": 20114
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8908647298812866,
      "learning_rate": 4.537014174307244e-06,
      "loss": 0.4158,
      "step": 20115
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4184284210205078,
      "learning_rate": 4.536078976566076e-06,
      "loss": 0.2795,
      "step": 20116
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5296186208724976,
      "learning_rate": 4.535143846945667e-06,
      "loss": 0.1395,
      "step": 20117
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5926316976547241,
      "learning_rate": 4.534208785457671e-06,
      "loss": 0.3573,
      "step": 20118
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.169384717941284,
      "learning_rate": 4.533273792113741e-06,
      "loss": 0.3888,
      "step": 20119
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6803101301193237,
      "learning_rate": 4.532338866925544e-06,
      "loss": 0.358,
      "step": 20120
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.762150526046753,
      "learning_rate": 4.5314040099047286e-06,
      "loss": 0.3544,
      "step": 20121
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6887600421905518,
      "learning_rate": 4.53046922106295e-06,
      "loss": 0.3164,
      "step": 20122
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6180362701416016,
      "learning_rate": 4.529534500411866e-06,
      "loss": 0.2626,
      "step": 20123
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.3028223514556885,
      "learning_rate": 4.528599847963131e-06,
      "loss": 0.3194,
      "step": 20124
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9367351531982422,
      "learning_rate": 4.527665263728394e-06,
      "loss": 0.3179,
      "step": 20125
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6819349527359009,
      "learning_rate": 4.526730747719305e-06,
      "loss": 0.2833,
      "step": 20126
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9483301639556885,
      "learning_rate": 4.52579629994752e-06,
      "loss": 0.3484,
      "step": 20127
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0373458862304688,
      "learning_rate": 4.524861920424689e-06,
      "loss": 0.3092,
      "step": 20128
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7969815731048584,
      "learning_rate": 4.523927609162456e-06,
      "loss": 0.3375,
      "step": 20129
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7852097749710083,
      "learning_rate": 4.522993366172476e-06,
      "loss": 0.3436,
      "step": 20130
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.143704652786255,
      "learning_rate": 4.522059191466393e-06,
      "loss": 0.3434,
      "step": 20131
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7493009567260742,
      "learning_rate": 4.521125085055851e-06,
      "loss": 0.3003,
      "step": 20132
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8863577842712402,
      "learning_rate": 4.520191046952502e-06,
      "loss": 0.31,
      "step": 20133
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0729706287384033,
      "learning_rate": 4.519257077167989e-06,
      "loss": 0.3905,
      "step": 20134
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6151602268218994,
      "learning_rate": 4.51832317571395e-06,
      "loss": 0.2988,
      "step": 20135
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.521293044090271,
      "learning_rate": 4.517389342602039e-06,
      "loss": 0.3196,
      "step": 20136
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.085714817047119,
      "learning_rate": 4.516455577843891e-06,
      "loss": 0.4188,
      "step": 20137
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.2713358402252197,
      "learning_rate": 4.515521881451151e-06,
      "loss": 0.3486,
      "step": 20138
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6168925762176514,
      "learning_rate": 4.514588253435453e-06,
      "loss": 0.3119,
      "step": 20139
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9392907619476318,
      "learning_rate": 4.513654693808448e-06,
      "loss": 0.3542,
      "step": 20140
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9477664232254028,
      "learning_rate": 4.512721202581768e-06,
      "loss": 0.3126,
      "step": 20141
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.186584711074829,
      "learning_rate": 4.511787779767048e-06,
      "loss": 0.315,
      "step": 20142
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8283475637435913,
      "learning_rate": 4.510854425375935e-06,
      "loss": 0.346,
      "step": 20143
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7853286266326904,
      "learning_rate": 4.509921139420058e-06,
      "loss": 0.2803,
      "step": 20144
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.989516019821167,
      "learning_rate": 4.508987921911052e-06,
      "loss": 0.3616,
      "step": 20145
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8023300170898438,
      "learning_rate": 4.508054772860558e-06,
      "loss": 0.3278,
      "step": 20146
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.1522462368011475,
      "learning_rate": 4.507121692280208e-06,
      "loss": 0.3247,
      "step": 20147
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.794507622718811,
      "learning_rate": 4.506188680181628e-06,
      "loss": 0.2977,
      "step": 20148
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0992400646209717,
      "learning_rate": 4.50525573657646e-06,
      "loss": 0.355,
      "step": 20149
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0626730918884277,
      "learning_rate": 4.504322861476332e-06,
      "loss": 0.3251,
      "step": 20150
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9087790250778198,
      "learning_rate": 4.503390054892873e-06,
      "loss": 0.3573,
      "step": 20151
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.8533512353897095,
      "learning_rate": 4.502457316837709e-06,
      "loss": 0.3221,
      "step": 20152
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0837676525115967,
      "learning_rate": 4.5015246473224775e-06,
      "loss": 0.3382,
      "step": 20153
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.753539800643921,
      "learning_rate": 4.500592046358801e-06,
      "loss": 0.3198,
      "step": 20154
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9205940961837769,
      "learning_rate": 4.499659513958305e-06,
      "loss": 0.3335,
      "step": 20155
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.935768723487854,
      "learning_rate": 4.498727050132622e-06,
      "loss": 0.2989,
      "step": 20156
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.604086995124817,
      "learning_rate": 4.497794654893374e-06,
      "loss": 0.3297,
      "step": 20157
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9531261920928955,
      "learning_rate": 4.496862328252181e-06,
      "loss": 0.2546,
      "step": 20158
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0543549060821533,
      "learning_rate": 4.495930070220673e-06,
      "loss": 0.3944,
      "step": 20159
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9041602611541748,
      "learning_rate": 4.4949978808104726e-06,
      "loss": 0.3088,
      "step": 20160
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.796781301498413,
      "learning_rate": 4.494065760033196e-06,
      "loss": 0.2708,
      "step": 20161
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.6627675294876099,
      "learning_rate": 4.493133707900471e-06,
      "loss": 0.2997,
      "step": 20162
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8013548851013184,
      "learning_rate": 4.492201724423916e-06,
      "loss": 0.3084,
      "step": 20163
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7347171306610107,
      "learning_rate": 4.491269809615149e-06,
      "loss": 0.3082,
      "step": 20164
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0590744018554688,
      "learning_rate": 4.490337963485785e-06,
      "loss": 0.3368,
      "step": 20165
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0483274459838867,
      "learning_rate": 4.489406186047449e-06,
      "loss": 0.2473,
      "step": 20166
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7410261631011963,
      "learning_rate": 4.488474477311755e-06,
      "loss": 0.3863,
      "step": 20167
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.1650898456573486,
      "learning_rate": 4.487542837290314e-06,
      "loss": 0.3359,
      "step": 20168
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0601587295532227,
      "learning_rate": 4.4866112659947505e-06,
      "loss": 0.3534,
      "step": 20169
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4885337352752686,
      "learning_rate": 4.485679763436672e-06,
      "loss": 0.271,
      "step": 20170
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5087437629699707,
      "learning_rate": 4.48474832962769e-06,
      "loss": 0.2682,
      "step": 20171
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8672057390213013,
      "learning_rate": 4.483816964579426e-06,
      "loss": 0.3101,
      "step": 20172
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.083101511001587,
      "learning_rate": 4.482885668303486e-06,
      "loss": 0.4511,
      "step": 20173
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0303587913513184,
      "learning_rate": 4.4819544408114766e-06,
      "loss": 0.32,
      "step": 20174
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.888732671737671,
      "learning_rate": 4.4810232821150156e-06,
      "loss": 0.3744,
      "step": 20175
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.901113510131836,
      "learning_rate": 4.48009219222571e-06,
      "loss": 0.3337,
      "step": 20176
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.6972527503967285,
      "learning_rate": 4.479161171155166e-06,
      "loss": 0.3629,
      "step": 20177
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.2282533645629883,
      "learning_rate": 4.478230218914989e-06,
      "loss": 0.3435,
      "step": 20178
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8668017387390137,
      "learning_rate": 4.477299335516792e-06,
      "loss": 0.3603,
      "step": 20179
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8307514190673828,
      "learning_rate": 4.476368520972176e-06,
      "loss": 0.3289,
      "step": 20180
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.105903148651123,
      "learning_rate": 4.475437775292744e-06,
      "loss": 0.386,
      "step": 20181
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8949958086013794,
      "learning_rate": 4.474507098490106e-06,
      "loss": 0.3283,
      "step": 20182
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4335108995437622,
      "learning_rate": 4.473576490575863e-06,
      "loss": 0.2592,
      "step": 20183
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.544053077697754,
      "learning_rate": 4.472645951561611e-06,
      "loss": 0.3997,
      "step": 20184
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.800826907157898,
      "learning_rate": 4.471715481458961e-06,
      "loss": 0.237,
      "step": 20185
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7140220403671265,
      "learning_rate": 4.4707850802795096e-06,
      "loss": 0.3045,
      "step": 20186
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.011655569076538,
      "learning_rate": 4.469854748034851e-06,
      "loss": 0.2873,
      "step": 20187
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7474538087844849,
      "learning_rate": 4.468924484736593e-06,
      "loss": 0.346,
      "step": 20188
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.741209626197815,
      "learning_rate": 4.467994290396331e-06,
      "loss": 0.2883,
      "step": 20189
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0135231018066406,
      "learning_rate": 4.467064165025659e-06,
      "loss": 0.3628,
      "step": 20190
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8687885999679565,
      "learning_rate": 4.466134108636171e-06,
      "loss": 0.3888,
      "step": 20191
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5538866519927979,
      "learning_rate": 4.465204121239471e-06,
      "loss": 0.2746,
      "step": 20192
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.6805305480957031,
      "learning_rate": 4.4642742028471464e-06,
      "loss": 0.3654,
      "step": 20193
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8885273933410645,
      "learning_rate": 4.463344353470791e-06,
      "loss": 0.3486,
      "step": 20194
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9429205656051636,
      "learning_rate": 4.462414573122003e-06,
      "loss": 0.2781,
      "step": 20195
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.035202741622925,
      "learning_rate": 4.461484861812372e-06,
      "loss": 0.3106,
      "step": 20196
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.341815710067749,
      "learning_rate": 4.460555219553483e-06,
      "loss": 0.3439,
      "step": 20197
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.120997905731201,
      "learning_rate": 4.459625646356935e-06,
      "loss": 0.3207,
      "step": 20198
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.6721934080123901,
      "learning_rate": 4.458696142234315e-06,
      "loss": 0.3638,
      "step": 20199
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.820886254310608,
      "learning_rate": 4.457766707197204e-06,
      "loss": 0.3532,
      "step": 20200
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.673379898071289,
      "learning_rate": 4.456837341257201e-06,
      "loss": 0.3407,
      "step": 20201
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.59940767288208,
      "learning_rate": 4.455908044425887e-06,
      "loss": 0.3344,
      "step": 20202
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5815829038619995,
      "learning_rate": 4.454978816714849e-06,
      "loss": 0.3029,
      "step": 20203
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8648020029067993,
      "learning_rate": 4.454049658135666e-06,
      "loss": 0.3242,
      "step": 20204
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.472124695777893,
      "learning_rate": 4.453120568699932e-06,
      "loss": 0.3204,
      "step": 20205
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.878175973892212,
      "learning_rate": 4.452191548419227e-06,
      "loss": 0.3569,
      "step": 20206
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5232375860214233,
      "learning_rate": 4.451262597305127e-06,
      "loss": 0.2701,
      "step": 20207
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9473214149475098,
      "learning_rate": 4.450333715369224e-06,
      "loss": 0.3333,
      "step": 20208
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0990915298461914,
      "learning_rate": 4.449404902623092e-06,
      "loss": 0.286,
      "step": 20209
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.2934298515319824,
      "learning_rate": 4.44847615907831e-06,
      "loss": 0.4077,
      "step": 20210
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.323503017425537,
      "learning_rate": 4.447547484746463e-06,
      "loss": 0.3216,
      "step": 20211
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9246562719345093,
      "learning_rate": 4.446618879639124e-06,
      "loss": 0.3066,
      "step": 20212
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3841781616210938,
      "learning_rate": 4.44569034376787e-06,
      "loss": 0.3213,
      "step": 20213
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9158885478973389,
      "learning_rate": 4.444761877144281e-06,
      "loss": 0.2892,
      "step": 20214
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0171573162078857,
      "learning_rate": 4.443833479779932e-06,
      "loss": 0.3667,
      "step": 20215
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8782460689544678,
      "learning_rate": 4.4429051516863965e-06,
      "loss": 0.3785,
      "step": 20216
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.852843999862671,
      "learning_rate": 4.441976892875245e-06,
      "loss": 0.3496,
      "step": 20217
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7754698991775513,
      "learning_rate": 4.441048703358057e-06,
      "loss": 0.3038,
      "step": 20218
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0306155681610107,
      "learning_rate": 4.4401205831464e-06,
      "loss": 0.2894,
      "step": 20219
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.548018217086792,
      "learning_rate": 4.439192532251845e-06,
      "loss": 0.3101,
      "step": 20220
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7512695789337158,
      "learning_rate": 4.438264550685965e-06,
      "loss": 0.2916,
      "step": 20221
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.3993427753448486,
      "learning_rate": 4.437336638460331e-06,
      "loss": 0.4291,
      "step": 20222
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8498321771621704,
      "learning_rate": 4.4364087955865045e-06,
      "loss": 0.2745,
      "step": 20223
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7181849479675293,
      "learning_rate": 4.43548102207606e-06,
      "loss": 0.2431,
      "step": 20224
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7781181335449219,
      "learning_rate": 4.434553317940563e-06,
      "loss": 0.3219,
      "step": 20225
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9852566719055176,
      "learning_rate": 4.4336256831915754e-06,
      "loss": 0.3323,
      "step": 20226
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9242926836013794,
      "learning_rate": 4.432698117840669e-06,
      "loss": 0.3157,
      "step": 20227
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8143783807754517,
      "learning_rate": 4.431770621899406e-06,
      "loss": 0.3254,
      "step": 20228
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.548356771469116,
      "learning_rate": 4.430843195379347e-06,
      "loss": 0.3294,
      "step": 20229
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8501378297805786,
      "learning_rate": 4.429915838292053e-06,
      "loss": 0.3473,
      "step": 20230
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9524732828140259,
      "learning_rate": 4.428988550649093e-06,
      "loss": 0.3079,
      "step": 20231
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8204240798950195,
      "learning_rate": 4.428061332462023e-06,
      "loss": 0.3006,
      "step": 20232
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8820594549179077,
      "learning_rate": 4.427134183742401e-06,
      "loss": 0.3216,
      "step": 20233
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0211422443389893,
      "learning_rate": 4.426207104501792e-06,
      "loss": 0.3873,
      "step": 20234
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9196839332580566,
      "learning_rate": 4.425280094751751e-06,
      "loss": 0.316,
      "step": 20235
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.092210292816162,
      "learning_rate": 4.424353154503832e-06,
      "loss": 0.4031,
      "step": 20236
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9242095947265625,
      "learning_rate": 4.4234262837695985e-06,
      "loss": 0.3193,
      "step": 20237
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.469065546989441,
      "learning_rate": 4.422499482560603e-06,
      "loss": 0.3454,
      "step": 20238
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.860624074935913,
      "learning_rate": 4.421572750888395e-06,
      "loss": 0.2766,
      "step": 20239
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7788732051849365,
      "learning_rate": 4.420646088764538e-06,
      "loss": 0.2833,
      "step": 20240
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9047359228134155,
      "learning_rate": 4.41971949620058e-06,
      "loss": 0.3725,
      "step": 20241
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.772360920906067,
      "learning_rate": 4.418792973208073e-06,
      "loss": 0.209,
      "step": 20242
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9241833686828613,
      "learning_rate": 4.417866519798566e-06,
      "loss": 0.2955,
      "step": 20243
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9030171632766724,
      "learning_rate": 4.416940135983616e-06,
      "loss": 0.4134,
      "step": 20244
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.0888352394104004,
      "learning_rate": 4.416013821774768e-06,
      "loss": 0.3667,
      "step": 20245
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8823325634002686,
      "learning_rate": 4.415087577183569e-06,
      "loss": 0.3502,
      "step": 20246
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7237823009490967,
      "learning_rate": 4.414161402221573e-06,
      "loss": 0.2641,
      "step": 20247
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.589748740196228,
      "learning_rate": 4.413235296900322e-06,
      "loss": 0.3195,
      "step": 20248
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5238327980041504,
      "learning_rate": 4.412309261231361e-06,
      "loss": 0.3442,
      "step": 20249
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9056535959243774,
      "learning_rate": 4.4113832952262415e-06,
      "loss": 0.3498,
      "step": 20250
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7879447937011719,
      "learning_rate": 4.410457398896504e-06,
      "loss": 0.3144,
      "step": 20251
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.6188353300094604,
      "learning_rate": 4.409531572253688e-06,
      "loss": 0.2956,
      "step": 20252
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.923567771911621,
      "learning_rate": 4.408605815309343e-06,
      "loss": 0.3543,
      "step": 20253
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.6195948123931885,
      "learning_rate": 4.407680128075009e-06,
      "loss": 0.3107,
      "step": 20254
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.8963853120803833,
      "learning_rate": 4.406754510562227e-06,
      "loss": 0.3241,
      "step": 20255
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5896275043487549,
      "learning_rate": 4.40582896278253e-06,
      "loss": 0.3421,
      "step": 20256
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5470694303512573,
      "learning_rate": 4.404903484747468e-06,
      "loss": 0.3495,
      "step": 20257
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.828136920928955,
      "learning_rate": 4.403978076468574e-06,
      "loss": 0.3563,
      "step": 20258
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0115718841552734,
      "learning_rate": 4.403052737957382e-06,
      "loss": 0.3832,
      "step": 20259
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9769151210784912,
      "learning_rate": 4.4021274692254355e-06,
      "loss": 0.2535,
      "step": 20260
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8585267066955566,
      "learning_rate": 4.401202270284267e-06,
      "loss": 0.3579,
      "step": 20261
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.515926718711853,
      "learning_rate": 4.4002771411454085e-06,
      "loss": 0.3968,
      "step": 20262
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8582240343093872,
      "learning_rate": 4.3993520818204e-06,
      "loss": 0.2118,
      "step": 20263
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.5693624019622803,
      "learning_rate": 4.39842709232077e-06,
      "loss": 0.3806,
      "step": 20264
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.1704561710357666,
      "learning_rate": 4.397502172658049e-06,
      "loss": 0.468,
      "step": 20265
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.73916757106781,
      "learning_rate": 4.396577322843776e-06,
      "loss": 0.2492,
      "step": 20266
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7637654542922974,
      "learning_rate": 4.3956525428894745e-06,
      "loss": 0.2961,
      "step": 20267
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.134570598602295,
      "learning_rate": 4.3947278328066775e-06,
      "loss": 0.3464,
      "step": 20268
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8766504526138306,
      "learning_rate": 4.393803192606909e-06,
      "loss": 0.2565,
      "step": 20269
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7948487997055054,
      "learning_rate": 4.392878622301704e-06,
      "loss": 0.2663,
      "step": 20270
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.367699384689331,
      "learning_rate": 4.391954121902586e-06,
      "loss": 0.3603,
      "step": 20271
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9013532400131226,
      "learning_rate": 4.391029691421077e-06,
      "loss": 0.3187,
      "step": 20272
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0700061321258545,
      "learning_rate": 4.3901053308687095e-06,
      "loss": 0.3416,
      "step": 20273
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6664918661117554,
      "learning_rate": 4.3891810402570034e-06,
      "loss": 0.3133,
      "step": 20274
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8215820789337158,
      "learning_rate": 4.38825681959748e-06,
      "loss": 0.2917,
      "step": 20275
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8060225248336792,
      "learning_rate": 4.387332668901671e-06,
      "loss": 0.3658,
      "step": 20276
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.104062795639038,
      "learning_rate": 4.38640858818109e-06,
      "loss": 0.2723,
      "step": 20277
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4962809085845947,
      "learning_rate": 4.385484577447258e-06,
      "loss": 0.3193,
      "step": 20278
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.184403657913208,
      "learning_rate": 4.3845606367116985e-06,
      "loss": 0.3134,
      "step": 20279
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.549445629119873,
      "learning_rate": 4.383636765985931e-06,
      "loss": 0.2911,
      "step": 20280
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9865038394927979,
      "learning_rate": 4.3827129652814715e-06,
      "loss": 0.369,
      "step": 20281
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6516860723495483,
      "learning_rate": 4.381789234609835e-06,
      "loss": 0.298,
      "step": 20282
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6252728700637817,
      "learning_rate": 4.380865573982543e-06,
      "loss": 0.3246,
      "step": 20283
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7009203433990479,
      "learning_rate": 4.37994198341111e-06,
      "loss": 0.3177,
      "step": 20284
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0076541900634766,
      "learning_rate": 4.379018462907045e-06,
      "loss": 0.2699,
      "step": 20285
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4748620986938477,
      "learning_rate": 4.378095012481871e-06,
      "loss": 0.2936,
      "step": 20286
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.5709266662597656,
      "learning_rate": 4.377171632147097e-06,
      "loss": 0.3169,
      "step": 20287
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9753539562225342,
      "learning_rate": 4.376248321914231e-06,
      "loss": 0.377,
      "step": 20288
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6301612854003906,
      "learning_rate": 4.3753250817947914e-06,
      "loss": 0.3625,
      "step": 20289
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7079740762710571,
      "learning_rate": 4.374401911800284e-06,
      "loss": 0.3239,
      "step": 20290
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9617345333099365,
      "learning_rate": 4.373478811942217e-06,
      "loss": 0.3187,
      "step": 20291
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5822954177856445,
      "learning_rate": 4.372555782232105e-06,
      "loss": 0.2206,
      "step": 20292
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.200789213180542,
      "learning_rate": 4.371632822681452e-06,
      "loss": 0.35,
      "step": 20293
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.3353888988494873,
      "learning_rate": 4.3707099333017655e-06,
      "loss": 0.3654,
      "step": 20294
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7008041143417358,
      "learning_rate": 4.3697871141045464e-06,
      "loss": 0.41,
      "step": 20295
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7600234746932983,
      "learning_rate": 4.368864365101309e-06,
      "loss": 0.3424,
      "step": 20296
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.391983985900879,
      "learning_rate": 4.367941686303554e-06,
      "loss": 0.3539,
      "step": 20297
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.026855945587158,
      "learning_rate": 4.3670190777227795e-06,
      "loss": 0.3585,
      "step": 20298
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.803699254989624,
      "learning_rate": 4.3660965393704954e-06,
      "loss": 0.3252,
      "step": 20299
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9234189987182617,
      "learning_rate": 4.365174071258202e-06,
      "loss": 0.3386,
      "step": 20300
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8384089469909668,
      "learning_rate": 4.364251673397394e-06,
      "loss": 0.3245,
      "step": 20301
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.884681224822998,
      "learning_rate": 4.363329345799578e-06,
      "loss": 0.283,
      "step": 20302
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.992915153503418,
      "learning_rate": 4.362407088476253e-06,
      "loss": 0.362,
      "step": 20303
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8557630777359009,
      "learning_rate": 4.3614849014389094e-06,
      "loss": 0.2623,
      "step": 20304
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.488856077194214,
      "learning_rate": 4.360562784699054e-06,
      "loss": 0.4335,
      "step": 20305
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8713667392730713,
      "learning_rate": 4.359640738268179e-06,
      "loss": 0.3047,
      "step": 20306
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7347499132156372,
      "learning_rate": 4.35871876215778e-06,
      "loss": 0.3873,
      "step": 20307
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.2197744846343994,
      "learning_rate": 4.357796856379348e-06,
      "loss": 0.2735,
      "step": 20308
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8842406272888184,
      "learning_rate": 4.356875020944384e-06,
      "loss": 0.3068,
      "step": 20309
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.766938328742981,
      "learning_rate": 4.355953255864377e-06,
      "loss": 0.2883,
      "step": 20310
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7396594285964966,
      "learning_rate": 4.355031561150816e-06,
      "loss": 0.4214,
      "step": 20311
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7814627885818481,
      "learning_rate": 4.354109936815198e-06,
      "loss": 0.3334,
      "step": 20312
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6783071756362915,
      "learning_rate": 4.3531883828690114e-06,
      "loss": 0.3851,
      "step": 20313
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7197757959365845,
      "learning_rate": 4.352266899323741e-06,
      "loss": 0.3371,
      "step": 20314
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.842104196548462,
      "learning_rate": 4.3513454861908825e-06,
      "loss": 0.3031,
      "step": 20315
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.2869467735290527,
      "learning_rate": 4.350424143481919e-06,
      "loss": 0.4456,
      "step": 20316
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.18975830078125,
      "learning_rate": 4.349502871208337e-06,
      "loss": 0.3529,
      "step": 20317
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6840659379959106,
      "learning_rate": 4.348581669381625e-06,
      "loss": 0.3311,
      "step": 20318
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6408709287643433,
      "learning_rate": 4.347660538013268e-06,
      "loss": 0.2402,
      "step": 20319
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5451407432556152,
      "learning_rate": 4.346739477114746e-06,
      "loss": 0.3212,
      "step": 20320
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7160329818725586,
      "learning_rate": 4.345818486697544e-06,
      "loss": 0.3588,
      "step": 20321
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9523183107376099,
      "learning_rate": 4.3448975667731475e-06,
      "loss": 0.3642,
      "step": 20322
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5456438064575195,
      "learning_rate": 4.343976717353035e-06,
      "loss": 0.2771,
      "step": 20323
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.5383687019348145,
      "learning_rate": 4.343055938448686e-06,
      "loss": 0.4229,
      "step": 20324
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8202041387557983,
      "learning_rate": 4.342135230071583e-06,
      "loss": 0.2506,
      "step": 20325
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6704158782958984,
      "learning_rate": 4.341214592233206e-06,
      "loss": 0.3214,
      "step": 20326
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9234551191329956,
      "learning_rate": 4.3402940249450254e-06,
      "loss": 0.3782,
      "step": 20327
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4262288808822632,
      "learning_rate": 4.339373528218528e-06,
      "loss": 0.2931,
      "step": 20328
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0672478675842285,
      "learning_rate": 4.3384531020651855e-06,
      "loss": 0.3019,
      "step": 20329
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8171136379241943,
      "learning_rate": 4.337532746496469e-06,
      "loss": 0.2897,
      "step": 20330
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7219876050949097,
      "learning_rate": 4.336612461523861e-06,
      "loss": 0.3499,
      "step": 20331
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0660128593444824,
      "learning_rate": 4.335692247158832e-06,
      "loss": 0.3108,
      "step": 20332
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5703662633895874,
      "learning_rate": 4.334772103412852e-06,
      "loss": 0.2747,
      "step": 20333
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.896148920059204,
      "learning_rate": 4.3338520302973915e-06,
      "loss": 0.4439,
      "step": 20334
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6529347896575928,
      "learning_rate": 4.332932027823928e-06,
      "loss": 0.2796,
      "step": 20335
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6607760190963745,
      "learning_rate": 4.332012096003928e-06,
      "loss": 0.248,
      "step": 20336
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8137694597244263,
      "learning_rate": 4.331092234848858e-06,
      "loss": 0.2961,
      "step": 20337
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8192671537399292,
      "learning_rate": 4.330172444370192e-06,
      "loss": 0.2764,
      "step": 20338
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.722506046295166,
      "learning_rate": 4.329252724579394e-06,
      "loss": 0.33,
      "step": 20339
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.835414171218872,
      "learning_rate": 4.328333075487926e-06,
      "loss": 0.3476,
      "step": 20340
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0143723487854004,
      "learning_rate": 4.327413497107264e-06,
      "loss": 0.2935,
      "step": 20341
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7878236770629883,
      "learning_rate": 4.326493989448866e-06,
      "loss": 0.3621,
      "step": 20342
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.04066801071167,
      "learning_rate": 4.325574552524193e-06,
      "loss": 0.2995,
      "step": 20343
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.623031497001648,
      "learning_rate": 4.324655186344716e-06,
      "loss": 0.3766,
      "step": 20344
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6947205066680908,
      "learning_rate": 4.323735890921893e-06,
      "loss": 0.3629,
      "step": 20345
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.0386769771575928,
      "learning_rate": 4.322816666267184e-06,
      "loss": 0.3116,
      "step": 20346
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5716873407363892,
      "learning_rate": 4.321897512392048e-06,
      "loss": 0.2579,
      "step": 20347
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.2231149673461914,
      "learning_rate": 4.32097842930795e-06,
      "loss": 0.3127,
      "step": 20348
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5735963582992554,
      "learning_rate": 4.320059417026347e-06,
      "loss": 0.2761,
      "step": 20349
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5994277000427246,
      "learning_rate": 4.31914047555869e-06,
      "loss": 0.3067,
      "step": 20350
    },
    {
      "epoch": 2.1,
      "grad_norm": 2.2080960273742676,
      "learning_rate": 4.318221604916445e-06,
      "loss": 0.2372,
      "step": 20351
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.739106297492981,
      "learning_rate": 4.317302805111063e-06,
      "loss": 0.3101,
      "step": 20352
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9457014799118042,
      "learning_rate": 4.3163840761539975e-06,
      "loss": 0.3102,
      "step": 20353
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9098849296569824,
      "learning_rate": 4.315465418056708e-06,
      "loss": 0.3529,
      "step": 20354
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9048453569412231,
      "learning_rate": 4.314546830830645e-06,
      "loss": 0.4234,
      "step": 20355
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8456288576126099,
      "learning_rate": 4.313628314487257e-06,
      "loss": 0.257,
      "step": 20356
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.6512296199798584,
      "learning_rate": 4.312709869038002e-06,
      "loss": 0.2755,
      "step": 20357
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.836044430732727,
      "learning_rate": 4.311791494494328e-06,
      "loss": 0.3695,
      "step": 20358
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9170995950698853,
      "learning_rate": 4.3108731908676845e-06,
      "loss": 0.2989,
      "step": 20359
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7591850757598877,
      "learning_rate": 4.309954958169518e-06,
      "loss": 0.352,
      "step": 20360
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9040313959121704,
      "learning_rate": 4.30903679641128e-06,
      "loss": 0.3421,
      "step": 20361
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.1522648334503174,
      "learning_rate": 4.308118705604418e-06,
      "loss": 0.3487,
      "step": 20362
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.0823211669921875,
      "learning_rate": 4.307200685760372e-06,
      "loss": 0.2896,
      "step": 20363
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.1866414546966553,
      "learning_rate": 4.306282736890596e-06,
      "loss": 0.3705,
      "step": 20364
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7392148971557617,
      "learning_rate": 4.30536485900653e-06,
      "loss": 0.2863,
      "step": 20365
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6207528114318848,
      "learning_rate": 4.304447052119614e-06,
      "loss": 0.2559,
      "step": 20366
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.056617021560669,
      "learning_rate": 4.303529316241297e-06,
      "loss": 0.3341,
      "step": 20367
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.2632768154144287,
      "learning_rate": 4.30261165138302e-06,
      "loss": 0.3667,
      "step": 20368
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9225058555603027,
      "learning_rate": 4.3016940575562175e-06,
      "loss": 0.283,
      "step": 20369
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.655293583869934,
      "learning_rate": 4.300776534772336e-06,
      "loss": 0.3187,
      "step": 20370
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5623502731323242,
      "learning_rate": 4.299859083042814e-06,
      "loss": 0.2892,
      "step": 20371
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.883854866027832,
      "learning_rate": 4.298941702379089e-06,
      "loss": 0.3391,
      "step": 20372
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.440737009048462,
      "learning_rate": 4.298024392792594e-06,
      "loss": 0.3662,
      "step": 20373
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8480216264724731,
      "learning_rate": 4.29710715429477e-06,
      "loss": 0.377,
      "step": 20374
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6662514209747314,
      "learning_rate": 4.296189986897055e-06,
      "loss": 0.3178,
      "step": 20375
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9007527828216553,
      "learning_rate": 4.295272890610875e-06,
      "loss": 0.3849,
      "step": 20376
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.810433030128479,
      "learning_rate": 4.294355865447672e-06,
      "loss": 0.3459,
      "step": 20377
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.767530083656311,
      "learning_rate": 4.293438911418878e-06,
      "loss": 0.3424,
      "step": 20378
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.736768126487732,
      "learning_rate": 4.2925220285359184e-06,
      "loss": 0.3252,
      "step": 20379
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7739269733428955,
      "learning_rate": 4.291605216810232e-06,
      "loss": 0.3807,
      "step": 20380
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.879082202911377,
      "learning_rate": 4.290688476253246e-06,
      "loss": 0.2481,
      "step": 20381
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.1455793380737305,
      "learning_rate": 4.2897718068763864e-06,
      "loss": 0.2127,
      "step": 20382
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.792234182357788,
      "learning_rate": 4.288855208691088e-06,
      "loss": 0.2322,
      "step": 20383
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6227107048034668,
      "learning_rate": 4.287938681708776e-06,
      "loss": 0.2848,
      "step": 20384
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9734658002853394,
      "learning_rate": 4.287022225940877e-06,
      "loss": 0.3375,
      "step": 20385
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.2369236946105957,
      "learning_rate": 4.286105841398811e-06,
      "loss": 0.3339,
      "step": 20386
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8976387977600098,
      "learning_rate": 4.285189528094012e-06,
      "loss": 0.3286,
      "step": 20387
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7203370332717896,
      "learning_rate": 4.284273286037901e-06,
      "loss": 0.2613,
      "step": 20388
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8237559795379639,
      "learning_rate": 4.283357115241896e-06,
      "loss": 0.3348,
      "step": 20389
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.810097098350525,
      "learning_rate": 4.282441015717428e-06,
      "loss": 0.3643,
      "step": 20390
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9559862613677979,
      "learning_rate": 4.281524987475914e-06,
      "loss": 0.4091,
      "step": 20391
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6276217699050903,
      "learning_rate": 4.280609030528773e-06,
      "loss": 0.3101,
      "step": 20392
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9416033029556274,
      "learning_rate": 4.279693144887427e-06,
      "loss": 0.3555,
      "step": 20393
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.1346802711486816,
      "learning_rate": 4.278777330563293e-06,
      "loss": 0.2861,
      "step": 20394
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9769622087478638,
      "learning_rate": 4.2778615875677855e-06,
      "loss": 0.3827,
      "step": 20395
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.057220935821533,
      "learning_rate": 4.276945915912329e-06,
      "loss": 0.3391,
      "step": 20396
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9602100849151611,
      "learning_rate": 4.276030315608336e-06,
      "loss": 0.2886,
      "step": 20397
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6894406080245972,
      "learning_rate": 4.275114786667218e-06,
      "loss": 0.26,
      "step": 20398
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8992513418197632,
      "learning_rate": 4.274199329100397e-06,
      "loss": 0.4515,
      "step": 20399
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.869924783706665,
      "learning_rate": 4.27328394291928e-06,
      "loss": 0.3247,
      "step": 20400
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.598381757736206,
      "learning_rate": 4.2723686281352806e-06,
      "loss": 0.3041,
      "step": 20401
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.048975706100464,
      "learning_rate": 4.271453384759814e-06,
      "loss": 0.3156,
      "step": 20402
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8547008037567139,
      "learning_rate": 4.270538212804287e-06,
      "loss": 0.3298,
      "step": 20403
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.109694242477417,
      "learning_rate": 4.269623112280108e-06,
      "loss": 0.3608,
      "step": 20404
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6312121152877808,
      "learning_rate": 4.268708083198691e-06,
      "loss": 0.3275,
      "step": 20405
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7457209825515747,
      "learning_rate": 4.2677931255714436e-06,
      "loss": 0.2745,
      "step": 20406
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9204413890838623,
      "learning_rate": 4.266878239409769e-06,
      "loss": 0.3877,
      "step": 20407
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5731335878372192,
      "learning_rate": 4.265963424725073e-06,
      "loss": 0.2954,
      "step": 20408
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.966583490371704,
      "learning_rate": 4.265048681528765e-06,
      "loss": 0.3691,
      "step": 20409
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5327047109603882,
      "learning_rate": 4.264134009832249e-06,
      "loss": 0.272,
      "step": 20410
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9895384311676025,
      "learning_rate": 4.263219409646924e-06,
      "loss": 0.2694,
      "step": 20411
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7041854858398438,
      "learning_rate": 4.262304880984199e-06,
      "loss": 0.3097,
      "step": 20412
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7431602478027344,
      "learning_rate": 4.261390423855471e-06,
      "loss": 0.3216,
      "step": 20413
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6872855424880981,
      "learning_rate": 4.26047603827214e-06,
      "loss": 0.3235,
      "step": 20414
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.745653510093689,
      "learning_rate": 4.259561724245612e-06,
      "loss": 0.2635,
      "step": 20415
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.931983232498169,
      "learning_rate": 4.258647481787283e-06,
      "loss": 0.2984,
      "step": 20416
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4319602251052856,
      "learning_rate": 4.257733310908547e-06,
      "loss": 0.2445,
      "step": 20417
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.0136475563049316,
      "learning_rate": 4.256819211620808e-06,
      "loss": 0.3734,
      "step": 20418
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.0809552669525146,
      "learning_rate": 4.2559051839354615e-06,
      "loss": 0.3765,
      "step": 20419
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.534011721611023,
      "learning_rate": 4.2549912278638995e-06,
      "loss": 0.2721,
      "step": 20420
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8372249603271484,
      "learning_rate": 4.2540773434175145e-06,
      "loss": 0.371,
      "step": 20421
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8911075592041016,
      "learning_rate": 4.253163530607708e-06,
      "loss": 0.3407,
      "step": 20422
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5273122787475586,
      "learning_rate": 4.252249789445868e-06,
      "loss": 0.3053,
      "step": 20423
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6799821853637695,
      "learning_rate": 4.251336119943384e-06,
      "loss": 0.2855,
      "step": 20424
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.99332594871521,
      "learning_rate": 4.250422522111654e-06,
      "loss": 0.4189,
      "step": 20425
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.906201958656311,
      "learning_rate": 4.249508995962065e-06,
      "loss": 0.4179,
      "step": 20426
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7383509874343872,
      "learning_rate": 4.248595541506002e-06,
      "loss": 0.307,
      "step": 20427
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.855647087097168,
      "learning_rate": 4.24768215875486e-06,
      "loss": 0.2944,
      "step": 20428
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6781692504882812,
      "learning_rate": 4.2467688477200245e-06,
      "loss": 0.2625,
      "step": 20429
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.870563268661499,
      "learning_rate": 4.245855608412876e-06,
      "loss": 0.4075,
      "step": 20430
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.9027132987976074,
      "learning_rate": 4.24494244084481e-06,
      "loss": 0.2951,
      "step": 20431
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.0842065811157227,
      "learning_rate": 4.244029345027207e-06,
      "loss": 0.3991,
      "step": 20432
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.480526328086853,
      "learning_rate": 4.24311632097145e-06,
      "loss": 0.3176,
      "step": 20433
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8299028873443604,
      "learning_rate": 4.24220336868892e-06,
      "loss": 0.3363,
      "step": 20434
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.046988010406494,
      "learning_rate": 4.241290488191005e-06,
      "loss": 0.3602,
      "step": 20435
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6819626092910767,
      "learning_rate": 4.240377679489082e-06,
      "loss": 0.2882,
      "step": 20436
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7503941059112549,
      "learning_rate": 4.239464942594529e-06,
      "loss": 0.3101,
      "step": 20437
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.8566962480545044,
      "learning_rate": 4.238552277518731e-06,
      "loss": 0.3617,
      "step": 20438
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6795991659164429,
      "learning_rate": 4.237639684273067e-06,
      "loss": 0.307,
      "step": 20439
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6908016204833984,
      "learning_rate": 4.236727162868906e-06,
      "loss": 0.3889,
      "step": 20440
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.0357136726379395,
      "learning_rate": 4.235814713317635e-06,
      "loss": 0.3075,
      "step": 20441
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6995950937271118,
      "learning_rate": 4.234902335630627e-06,
      "loss": 0.324,
      "step": 20442
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7826452255249023,
      "learning_rate": 4.233990029819249e-06,
      "loss": 0.3172,
      "step": 20443
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.884931206703186,
      "learning_rate": 4.233077795894888e-06,
      "loss": 0.3264,
      "step": 20444
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.929863452911377,
      "learning_rate": 4.2321656338689085e-06,
      "loss": 0.2969,
      "step": 20445
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5452220439910889,
      "learning_rate": 4.2312535437526855e-06,
      "loss": 0.2557,
      "step": 20446
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.697568655014038,
      "learning_rate": 4.230341525557586e-06,
      "loss": 0.2361,
      "step": 20447
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.513690710067749,
      "learning_rate": 4.229429579294989e-06,
      "loss": 0.3953,
      "step": 20448
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.256185531616211,
      "learning_rate": 4.228517704976258e-06,
      "loss": 0.3886,
      "step": 20449
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.1027960777282715,
      "learning_rate": 4.22760590261276e-06,
      "loss": 0.3479,
      "step": 20450
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8752087354660034,
      "learning_rate": 4.226694172215869e-06,
      "loss": 0.3897,
      "step": 20451
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6325411796569824,
      "learning_rate": 4.22578251379695e-06,
      "loss": 0.3322,
      "step": 20452
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7960611581802368,
      "learning_rate": 4.2248709273673625e-06,
      "loss": 0.2692,
      "step": 20453
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7182836532592773,
      "learning_rate": 4.223959412938481e-06,
      "loss": 0.3047,
      "step": 20454
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8937945365905762,
      "learning_rate": 4.223047970521666e-06,
      "loss": 0.4054,
      "step": 20455
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4455199241638184,
      "learning_rate": 4.222136600128276e-06,
      "loss": 0.2654,
      "step": 20456
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8657501935958862,
      "learning_rate": 4.22122530176968e-06,
      "loss": 0.365,
      "step": 20457
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.911769151687622,
      "learning_rate": 4.220314075457239e-06,
      "loss": 0.4329,
      "step": 20458
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2987772226333618,
      "learning_rate": 4.219402921202311e-06,
      "loss": 0.2586,
      "step": 20459
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0168025493621826,
      "learning_rate": 4.218491839016253e-06,
      "loss": 0.4166,
      "step": 20460
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8320631980895996,
      "learning_rate": 4.217580828910431e-06,
      "loss": 0.227,
      "step": 20461
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8896379470825195,
      "learning_rate": 4.2166698908962e-06,
      "loss": 0.3381,
      "step": 20462
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.794325828552246,
      "learning_rate": 4.215759024984911e-06,
      "loss": 0.3361,
      "step": 20463
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.5922768115997314,
      "learning_rate": 4.21484823118793e-06,
      "loss": 0.3185,
      "step": 20464
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8907281160354614,
      "learning_rate": 4.213937509516607e-06,
      "loss": 0.3249,
      "step": 20465
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8162078857421875,
      "learning_rate": 4.213026859982293e-06,
      "loss": 0.3867,
      "step": 20466
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.740849256515503,
      "learning_rate": 4.212116282596349e-06,
      "loss": 0.3193,
      "step": 20467
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7083289623260498,
      "learning_rate": 4.2112057773701234e-06,
      "loss": 0.368,
      "step": 20468
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.1547482013702393,
      "learning_rate": 4.210295344314965e-06,
      "loss": 0.3457,
      "step": 20469
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8026267290115356,
      "learning_rate": 4.209384983442231e-06,
      "loss": 0.3284,
      "step": 20470
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9290376901626587,
      "learning_rate": 4.2084746947632685e-06,
      "loss": 0.3629,
      "step": 20471
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6888583898544312,
      "learning_rate": 4.207564478289425e-06,
      "loss": 0.3179,
      "step": 20472
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6734932661056519,
      "learning_rate": 4.206654334032045e-06,
      "loss": 0.2848,
      "step": 20473
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8713361024856567,
      "learning_rate": 4.205744262002484e-06,
      "loss": 0.3297,
      "step": 20474
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.930220603942871,
      "learning_rate": 4.204834262212085e-06,
      "loss": 0.3703,
      "step": 20475
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9771636724472046,
      "learning_rate": 4.2039243346721885e-06,
      "loss": 0.2944,
      "step": 20476
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9302659034729004,
      "learning_rate": 4.2030144793941455e-06,
      "loss": 0.3213,
      "step": 20477
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8513576984405518,
      "learning_rate": 4.202104696389297e-06,
      "loss": 0.3309,
      "step": 20478
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6836775541305542,
      "learning_rate": 4.201194985668983e-06,
      "loss": 0.2717,
      "step": 20479
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.374661922454834,
      "learning_rate": 4.200285347244549e-06,
      "loss": 0.3412,
      "step": 20480
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.517897367477417,
      "learning_rate": 4.199375781127336e-06,
      "loss": 0.3302,
      "step": 20481
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.939706802368164,
      "learning_rate": 4.198466287328683e-06,
      "loss": 0.3572,
      "step": 20482
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6508368253707886,
      "learning_rate": 4.197556865859923e-06,
      "loss": 0.3353,
      "step": 20483
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7250765562057495,
      "learning_rate": 4.196647516732404e-06,
      "loss": 0.3404,
      "step": 20484
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9531840085983276,
      "learning_rate": 4.195738239957459e-06,
      "loss": 0.3117,
      "step": 20485
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9362584352493286,
      "learning_rate": 4.19482903554642e-06,
      "loss": 0.3088,
      "step": 20486
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7401188611984253,
      "learning_rate": 4.193919903510629e-06,
      "loss": 0.291,
      "step": 20487
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7905656099319458,
      "learning_rate": 4.193010843861418e-06,
      "loss": 0.2918,
      "step": 20488
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9989204406738281,
      "learning_rate": 4.192101856610118e-06,
      "loss": 0.353,
      "step": 20489
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.124579668045044,
      "learning_rate": 4.191192941768066e-06,
      "loss": 0.3533,
      "step": 20490
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8323097229003906,
      "learning_rate": 4.1902840993465936e-06,
      "loss": 0.2972,
      "step": 20491
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5522648096084595,
      "learning_rate": 4.189375329357025e-06,
      "loss": 0.2605,
      "step": 20492
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.2534072399139404,
      "learning_rate": 4.188466631810699e-06,
      "loss": 0.3884,
      "step": 20493
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8923197984695435,
      "learning_rate": 4.187558006718941e-06,
      "loss": 0.2194,
      "step": 20494
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7485878467559814,
      "learning_rate": 4.1866494540930795e-06,
      "loss": 0.2832,
      "step": 20495
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.84028160572052,
      "learning_rate": 4.1857409739444365e-06,
      "loss": 0.4258,
      "step": 20496
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.3150548934936523,
      "learning_rate": 4.184832566284348e-06,
      "loss": 0.3083,
      "step": 20497
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.724783182144165,
      "learning_rate": 4.183924231124134e-06,
      "loss": 0.3036,
      "step": 20498
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.76469886302948,
      "learning_rate": 4.183015968475115e-06,
      "loss": 0.3703,
      "step": 20499
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.4444358348846436,
      "learning_rate": 4.182107778348624e-06,
      "loss": 0.2229,
      "step": 20500
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.82680344581604,
      "learning_rate": 4.181199660755979e-06,
      "loss": 0.2777,
      "step": 20501
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7589447498321533,
      "learning_rate": 4.1802916157084975e-06,
      "loss": 0.2808,
      "step": 20502
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.03875732421875,
      "learning_rate": 4.179383643217508e-06,
      "loss": 0.3897,
      "step": 20503
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.097912549972534,
      "learning_rate": 4.178475743294329e-06,
      "loss": 0.4108,
      "step": 20504
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.666831612586975,
      "learning_rate": 4.177567915950274e-06,
      "loss": 0.3465,
      "step": 20505
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8400555849075317,
      "learning_rate": 4.1766601611966675e-06,
      "loss": 0.3188,
      "step": 20506
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.56309175491333,
      "learning_rate": 4.175752479044824e-06,
      "loss": 0.3332,
      "step": 20507
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.588010311126709,
      "learning_rate": 4.174844869506063e-06,
      "loss": 0.287,
      "step": 20508
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.1895017623901367,
      "learning_rate": 4.173937332591692e-06,
      "loss": 0.3404,
      "step": 20509
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7754324674606323,
      "learning_rate": 4.173029868313035e-06,
      "loss": 0.3169,
      "step": 20510
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7372431755065918,
      "learning_rate": 4.172122476681402e-06,
      "loss": 0.3468,
      "step": 20511
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.259066581726074,
      "learning_rate": 4.171215157708101e-06,
      "loss": 0.3569,
      "step": 20512
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8689117431640625,
      "learning_rate": 4.1703079114044524e-06,
      "loss": 0.3001,
      "step": 20513
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.732094645500183,
      "learning_rate": 4.169400737781764e-06,
      "loss": 0.3789,
      "step": 20514
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7180207967758179,
      "learning_rate": 4.1684936368513395e-06,
      "loss": 0.2826,
      "step": 20515
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9407304525375366,
      "learning_rate": 4.167586608624499e-06,
      "loss": 0.3642,
      "step": 20516
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.942366123199463,
      "learning_rate": 4.1666796531125445e-06,
      "loss": 0.4482,
      "step": 20517
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8088966608047485,
      "learning_rate": 4.165772770326781e-06,
      "loss": 0.3199,
      "step": 20518
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.66592276096344,
      "learning_rate": 4.164865960278521e-06,
      "loss": 0.2784,
      "step": 20519
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9439665079116821,
      "learning_rate": 4.163959222979068e-06,
      "loss": 0.3252,
      "step": 20520
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.956289529800415,
      "learning_rate": 4.163052558439726e-06,
      "loss": 0.3279,
      "step": 20521
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.7258925437927246,
      "learning_rate": 4.162145966671794e-06,
      "loss": 0.3259,
      "step": 20522
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.367098569869995,
      "learning_rate": 4.161239447686585e-06,
      "loss": 0.3478,
      "step": 20523
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9764809608459473,
      "learning_rate": 4.1603330014953935e-06,
      "loss": 0.3018,
      "step": 20524
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0531697273254395,
      "learning_rate": 4.15942662810952e-06,
      "loss": 0.298,
      "step": 20525
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.5016841888427734,
      "learning_rate": 4.158520327540269e-06,
      "loss": 0.4213,
      "step": 20526
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8307278156280518,
      "learning_rate": 4.157614099798939e-06,
      "loss": 0.2772,
      "step": 20527
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5481294393539429,
      "learning_rate": 4.156707944896822e-06,
      "loss": 0.3302,
      "step": 20528
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7646559476852417,
      "learning_rate": 4.155801862845225e-06,
      "loss": 0.2889,
      "step": 20529
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.426678419113159,
      "learning_rate": 4.154895853655438e-06,
      "loss": 0.3551,
      "step": 20530
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6373435258865356,
      "learning_rate": 4.1539899173387565e-06,
      "loss": 0.2519,
      "step": 20531
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9594656229019165,
      "learning_rate": 4.15308405390648e-06,
      "loss": 0.393,
      "step": 20532
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4690630435943604,
      "learning_rate": 4.1521782633698996e-06,
      "loss": 0.2451,
      "step": 20533
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9740240573883057,
      "learning_rate": 4.151272545740306e-06,
      "loss": 0.2915,
      "step": 20534
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7969685792922974,
      "learning_rate": 4.150366901028991e-06,
      "loss": 0.2862,
      "step": 20535
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.858385682106018,
      "learning_rate": 4.149461329247249e-06,
      "loss": 0.2486,
      "step": 20536
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6162843704223633,
      "learning_rate": 4.148555830406371e-06,
      "loss": 0.3412,
      "step": 20537
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7216273546218872,
      "learning_rate": 4.147650404517639e-06,
      "loss": 0.3338,
      "step": 20538
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7822335958480835,
      "learning_rate": 4.146745051592349e-06,
      "loss": 0.351,
      "step": 20539
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.763408899307251,
      "learning_rate": 4.145839771641786e-06,
      "loss": 0.3668,
      "step": 20540
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.841086983680725,
      "learning_rate": 4.144934564677232e-06,
      "loss": 0.3684,
      "step": 20541
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.1550722122192383,
      "learning_rate": 4.14402943070998e-06,
      "loss": 0.3608,
      "step": 20542
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8636503219604492,
      "learning_rate": 4.143124369751312e-06,
      "loss": 0.2953,
      "step": 20543
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.130831241607666,
      "learning_rate": 4.142219381812507e-06,
      "loss": 0.3765,
      "step": 20544
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.9344269037246704,
      "learning_rate": 4.141314466904854e-06,
      "loss": 0.3197,
      "step": 20545
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5970686674118042,
      "learning_rate": 4.140409625039633e-06,
      "loss": 0.3283,
      "step": 20546
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.284156322479248,
      "learning_rate": 4.139504856228125e-06,
      "loss": 0.3364,
      "step": 20547
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9756600856781006,
      "learning_rate": 4.138600160481606e-06,
      "loss": 0.3838,
      "step": 20548
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8573731184005737,
      "learning_rate": 4.1376955378113614e-06,
      "loss": 0.389,
      "step": 20549
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9133199453353882,
      "learning_rate": 4.136790988228668e-06,
      "loss": 0.3169,
      "step": 20550
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7480684518814087,
      "learning_rate": 4.135886511744797e-06,
      "loss": 0.2874,
      "step": 20551
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.2019948959350586,
      "learning_rate": 4.134982108371035e-06,
      "loss": 0.2673,
      "step": 20552
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6947228908538818,
      "learning_rate": 4.134077778118653e-06,
      "loss": 0.3095,
      "step": 20553
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8441284894943237,
      "learning_rate": 4.13317352099892e-06,
      "loss": 0.3126,
      "step": 20554
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7420432567596436,
      "learning_rate": 4.132269337023118e-06,
      "loss": 0.2129,
      "step": 20555
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7701048851013184,
      "learning_rate": 4.131365226202517e-06,
      "loss": 0.3225,
      "step": 20556
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.650674819946289,
      "learning_rate": 4.130461188548386e-06,
      "loss": 0.3515,
      "step": 20557
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.018794059753418,
      "learning_rate": 4.129557224072e-06,
      "loss": 0.2669,
      "step": 20558
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.473272442817688,
      "learning_rate": 4.128653332784629e-06,
      "loss": 0.3502,
      "step": 20559
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7092461585998535,
      "learning_rate": 4.12774951469754e-06,
      "loss": 0.3637,
      "step": 20560
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8957176208496094,
      "learning_rate": 4.126845769821998e-06,
      "loss": 0.2948,
      "step": 20561
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4295017719268799,
      "learning_rate": 4.1259420981692785e-06,
      "loss": 0.2481,
      "step": 20562
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.738620400428772,
      "learning_rate": 4.125038499750644e-06,
      "loss": 0.3006,
      "step": 20563
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8393315076828003,
      "learning_rate": 4.1241349745773555e-06,
      "loss": 0.3009,
      "step": 20564
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.3087375164031982,
      "learning_rate": 4.123231522660684e-06,
      "loss": 0.3174,
      "step": 20565
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6423453092575073,
      "learning_rate": 4.122328144011893e-06,
      "loss": 0.3088,
      "step": 20566
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.928061842918396,
      "learning_rate": 4.121424838642239e-06,
      "loss": 0.3416,
      "step": 20567
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.0456111431121826,
      "learning_rate": 4.12052160656299e-06,
      "loss": 0.2829,
      "step": 20568
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.2087085247039795,
      "learning_rate": 4.119618447785407e-06,
      "loss": 0.3506,
      "step": 20569
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.1114602088928223,
      "learning_rate": 4.118715362320743e-06,
      "loss": 0.3377,
      "step": 20570
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.148792028427124,
      "learning_rate": 4.1178123501802654e-06,
      "loss": 0.3466,
      "step": 20571
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.0039305686950684,
      "learning_rate": 4.11690941137523e-06,
      "loss": 0.3268,
      "step": 20572
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9265947341918945,
      "learning_rate": 4.116006545916893e-06,
      "loss": 0.3482,
      "step": 20573
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.237950086593628,
      "learning_rate": 4.115103753816507e-06,
      "loss": 0.2961,
      "step": 20574
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6869077682495117,
      "learning_rate": 4.114201035085336e-06,
      "loss": 0.322,
      "step": 20575
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7626198530197144,
      "learning_rate": 4.113298389734628e-06,
      "loss": 0.2979,
      "step": 20576
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.363337755203247,
      "learning_rate": 4.112395817775635e-06,
      "loss": 0.3575,
      "step": 20577
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9653240442276,
      "learning_rate": 4.111493319219618e-06,
      "loss": 0.3052,
      "step": 20578
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.643217086791992,
      "learning_rate": 4.110590894077824e-06,
      "loss": 0.3861,
      "step": 20579
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9982507228851318,
      "learning_rate": 4.109688542361498e-06,
      "loss": 0.3708,
      "step": 20580
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8317736387252808,
      "learning_rate": 4.108786264081902e-06,
      "loss": 0.3044,
      "step": 20581
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.0564815998077393,
      "learning_rate": 4.107884059250278e-06,
      "loss": 0.347,
      "step": 20582
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.865134596824646,
      "learning_rate": 4.106981927877871e-06,
      "loss": 0.3853,
      "step": 20583
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7865887880325317,
      "learning_rate": 4.106079869975936e-06,
      "loss": 0.3424,
      "step": 20584
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6859424114227295,
      "learning_rate": 4.105177885555716e-06,
      "loss": 0.3354,
      "step": 20585
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.0451934337615967,
      "learning_rate": 4.1042759746284545e-06,
      "loss": 0.3844,
      "step": 20586
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.204944133758545,
      "learning_rate": 4.1033741372053945e-06,
      "loss": 0.3555,
      "step": 20587
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7733731269836426,
      "learning_rate": 4.102472373297787e-06,
      "loss": 0.3295,
      "step": 20588
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.806494951248169,
      "learning_rate": 4.101570682916867e-06,
      "loss": 0.3075,
      "step": 20589
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.422715187072754,
      "learning_rate": 4.100669066073879e-06,
      "loss": 0.2928,
      "step": 20590
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.833665370941162,
      "learning_rate": 4.099767522780066e-06,
      "loss": 0.2724,
      "step": 20591
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8552019596099854,
      "learning_rate": 4.098866053046666e-06,
      "loss": 0.3487,
      "step": 20592
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6994749307632446,
      "learning_rate": 4.097964656884914e-06,
      "loss": 0.2956,
      "step": 20593
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.0586764812469482,
      "learning_rate": 4.097063334306055e-06,
      "loss": 0.3955,
      "step": 20594
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.738379955291748,
      "learning_rate": 4.096162085321325e-06,
      "loss": 0.3071,
      "step": 20595
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7042691707611084,
      "learning_rate": 4.095260909941952e-06,
      "loss": 0.2882,
      "step": 20596
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5465295314788818,
      "learning_rate": 4.094359808179182e-06,
      "loss": 0.3259,
      "step": 20597
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.772933840751648,
      "learning_rate": 4.093458780044246e-06,
      "loss": 0.3212,
      "step": 20598
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.458108901977539,
      "learning_rate": 4.092557825548375e-06,
      "loss": 0.4425,
      "step": 20599
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.0014097690582275,
      "learning_rate": 4.0916569447028e-06,
      "loss": 0.3487,
      "step": 20600
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8747971057891846,
      "learning_rate": 4.090756137518759e-06,
      "loss": 0.3929,
      "step": 20601
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9707059860229492,
      "learning_rate": 4.089855404007479e-06,
      "loss": 0.3569,
      "step": 20602
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7968690395355225,
      "learning_rate": 4.088954744180187e-06,
      "loss": 0.3199,
      "step": 20603
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6680563688278198,
      "learning_rate": 4.088054158048118e-06,
      "loss": 0.3247,
      "step": 20604
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7968851327896118,
      "learning_rate": 4.087153645622497e-06,
      "loss": 0.2922,
      "step": 20605
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.976036787033081,
      "learning_rate": 4.086253206914547e-06,
      "loss": 0.3071,
      "step": 20606
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.997406244277954,
      "learning_rate": 4.085352841935501e-06,
      "loss": 0.3122,
      "step": 20607
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.2003164291381836,
      "learning_rate": 4.084452550696582e-06,
      "loss": 0.3456,
      "step": 20608
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7975083589553833,
      "learning_rate": 4.083552333209009e-06,
      "loss": 0.3426,
      "step": 20609
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4768668413162231,
      "learning_rate": 4.082652189484013e-06,
      "loss": 0.243,
      "step": 20610
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5858869552612305,
      "learning_rate": 4.0817521195328135e-06,
      "loss": 0.3268,
      "step": 20611
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.5493412017822266,
      "learning_rate": 4.080852123366632e-06,
      "loss": 0.3845,
      "step": 20612
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.10913348197937,
      "learning_rate": 4.079952200996685e-06,
      "loss": 0.3367,
      "step": 20613
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.3472936153411865,
      "learning_rate": 4.0790523524342e-06,
      "loss": 0.4339,
      "step": 20614
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.663503885269165,
      "learning_rate": 4.078152577690391e-06,
      "loss": 0.2665,
      "step": 20615
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6304303407669067,
      "learning_rate": 4.077252876776473e-06,
      "loss": 0.3105,
      "step": 20616
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8626251220703125,
      "learning_rate": 4.076353249703671e-06,
      "loss": 0.2855,
      "step": 20617
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9483745098114014,
      "learning_rate": 4.075453696483195e-06,
      "loss": 0.3615,
      "step": 20618
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9330575466156006,
      "learning_rate": 4.074554217126259e-06,
      "loss": 0.3562,
      "step": 20619
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7020512819290161,
      "learning_rate": 4.073654811644082e-06,
      "loss": 0.2918,
      "step": 20620
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9374598264694214,
      "learning_rate": 4.072755480047876e-06,
      "loss": 0.291,
      "step": 20621
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7054370641708374,
      "learning_rate": 4.071856222348848e-06,
      "loss": 0.4098,
      "step": 20622
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.174116611480713,
      "learning_rate": 4.070957038558217e-06,
      "loss": 0.3909,
      "step": 20623
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8789446353912354,
      "learning_rate": 4.07005792868719e-06,
      "loss": 0.3555,
      "step": 20624
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.603979468345642,
      "learning_rate": 4.069158892746976e-06,
      "loss": 0.3429,
      "step": 20625
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7340636253356934,
      "learning_rate": 4.068259930748781e-06,
      "loss": 0.3381,
      "step": 20626
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8838900327682495,
      "learning_rate": 4.06736104270382e-06,
      "loss": 0.3186,
      "step": 20627
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.468104600906372,
      "learning_rate": 4.066462228623296e-06,
      "loss": 0.3201,
      "step": 20628
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7230304479599,
      "learning_rate": 4.065563488518409e-06,
      "loss": 0.3592,
      "step": 20629
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.8285937309265137,
      "learning_rate": 4.064664822400375e-06,
      "loss": 0.2964,
      "step": 20630
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7935103178024292,
      "learning_rate": 4.06376623028039e-06,
      "loss": 0.3267,
      "step": 20631
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7263766527175903,
      "learning_rate": 4.0628677121696584e-06,
      "loss": 0.402,
      "step": 20632
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9256398677825928,
      "learning_rate": 4.061969268079387e-06,
      "loss": 0.2877,
      "step": 20633
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6792551279067993,
      "learning_rate": 4.061070898020771e-06,
      "loss": 0.3146,
      "step": 20634
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.113100290298462,
      "learning_rate": 4.060172602005011e-06,
      "loss": 0.3617,
      "step": 20635
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5727354288101196,
      "learning_rate": 4.059274380043313e-06,
      "loss": 0.2323,
      "step": 20636
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9958552122116089,
      "learning_rate": 4.05837623214687e-06,
      "loss": 0.3816,
      "step": 20637
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6575013399124146,
      "learning_rate": 4.057478158326882e-06,
      "loss": 0.3233,
      "step": 20638
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.674545407295227,
      "learning_rate": 4.05658015859454e-06,
      "loss": 0.3782,
      "step": 20639
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6667547225952148,
      "learning_rate": 4.055682232961047e-06,
      "loss": 0.2911,
      "step": 20640
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7395758628845215,
      "learning_rate": 4.054784381437594e-06,
      "loss": 0.3182,
      "step": 20641
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9523745775222778,
      "learning_rate": 4.053886604035373e-06,
      "loss": 0.2572,
      "step": 20642
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.501217246055603,
      "learning_rate": 4.052988900765583e-06,
      "loss": 0.3103,
      "step": 20643
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7040050029754639,
      "learning_rate": 4.052091271639413e-06,
      "loss": 0.2988,
      "step": 20644
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7410671710968018,
      "learning_rate": 4.051193716668049e-06,
      "loss": 0.2757,
      "step": 20645
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.3224503993988037,
      "learning_rate": 4.05029623586269e-06,
      "loss": 0.3506,
      "step": 20646
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6583712100982666,
      "learning_rate": 4.049398829234521e-06,
      "loss": 0.3348,
      "step": 20647
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9298893213272095,
      "learning_rate": 4.048501496794727e-06,
      "loss": 0.3211,
      "step": 20648
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9068471193313599,
      "learning_rate": 4.047604238554501e-06,
      "loss": 0.3067,
      "step": 20649
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.1575586795806885,
      "learning_rate": 4.046707054525028e-06,
      "loss": 0.3328,
      "step": 20650
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.1643199920654297,
      "learning_rate": 4.045809944717493e-06,
      "loss": 0.372,
      "step": 20651
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.904004693031311,
      "learning_rate": 4.044912909143076e-06,
      "loss": 0.3112,
      "step": 20652
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6471153497695923,
      "learning_rate": 4.044015947812968e-06,
      "loss": 0.3395,
      "step": 20653
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.467763662338257,
      "learning_rate": 4.04311906073835e-06,
      "loss": 0.2841,
      "step": 20654
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.277517080307007,
      "learning_rate": 4.042222247930399e-06,
      "loss": 0.3326,
      "step": 20655
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.2811577320098877,
      "learning_rate": 4.041325509400301e-06,
      "loss": 0.3195,
      "step": 20656
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.1779749393463135,
      "learning_rate": 4.040428845159235e-06,
      "loss": 0.3879,
      "step": 20657
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9631381034851074,
      "learning_rate": 4.039532255218378e-06,
      "loss": 0.3413,
      "step": 20658
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9384167194366455,
      "learning_rate": 4.0386357395889106e-06,
      "loss": 0.3076,
      "step": 20659
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.2050654888153076,
      "learning_rate": 4.037739298282011e-06,
      "loss": 0.3878,
      "step": 20660
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.265977144241333,
      "learning_rate": 4.0368429313088486e-06,
      "loss": 0.3201,
      "step": 20661
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8850572109222412,
      "learning_rate": 4.035946638680607e-06,
      "loss": 0.3267,
      "step": 20662
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8094723224639893,
      "learning_rate": 4.035050420408457e-06,
      "loss": 0.2739,
      "step": 20663
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4882677793502808,
      "learning_rate": 4.034154276503574e-06,
      "loss": 0.3524,
      "step": 20664
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8767035007476807,
      "learning_rate": 4.033258206977123e-06,
      "loss": 0.333,
      "step": 20665
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.736092448234558,
      "learning_rate": 4.032362211840286e-06,
      "loss": 0.3966,
      "step": 20666
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.1136295795440674,
      "learning_rate": 4.03146629110423e-06,
      "loss": 0.3452,
      "step": 20667
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7760597467422485,
      "learning_rate": 4.03057044478012e-06,
      "loss": 0.2976,
      "step": 20668
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6807886362075806,
      "learning_rate": 4.0296746728791314e-06,
      "loss": 0.3046,
      "step": 20669
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7984446287155151,
      "learning_rate": 4.0287789754124304e-06,
      "loss": 0.3548,
      "step": 20670
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.186453104019165,
      "learning_rate": 4.027883352391179e-06,
      "loss": 0.3374,
      "step": 20671
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6977896690368652,
      "learning_rate": 4.026987803826552e-06,
      "loss": 0.3262,
      "step": 20672
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9345730543136597,
      "learning_rate": 4.026092329729709e-06,
      "loss": 0.2486,
      "step": 20673
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7370591163635254,
      "learning_rate": 4.025196930111811e-06,
      "loss": 0.2593,
      "step": 20674
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8451154232025146,
      "learning_rate": 4.02430160498403e-06,
      "loss": 0.2878,
      "step": 20675
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.079017162322998,
      "learning_rate": 4.023406354357524e-06,
      "loss": 0.3045,
      "step": 20676
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9431447982788086,
      "learning_rate": 4.022511178243453e-06,
      "loss": 0.3075,
      "step": 20677
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6699633598327637,
      "learning_rate": 4.021616076652977e-06,
      "loss": 0.3211,
      "step": 20678
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8132190704345703,
      "learning_rate": 4.020721049597259e-06,
      "loss": 0.2549,
      "step": 20679
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.152064800262451,
      "learning_rate": 4.019826097087457e-06,
      "loss": 0.347,
      "step": 20680
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.057978868484497,
      "learning_rate": 4.0189312191347244e-06,
      "loss": 0.3453,
      "step": 20681
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7349905967712402,
      "learning_rate": 4.018036415750224e-06,
      "loss": 0.2951,
      "step": 20682
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.968307375907898,
      "learning_rate": 4.017141686945108e-06,
      "loss": 0.3554,
      "step": 20683
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8841408491134644,
      "learning_rate": 4.0162470327305305e-06,
      "loss": 0.2317,
      "step": 20684
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.716692328453064,
      "learning_rate": 4.01535245311765e-06,
      "loss": 0.2893,
      "step": 20685
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.174360513687134,
      "learning_rate": 4.014457948117616e-06,
      "loss": 0.3213,
      "step": 20686
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8886990547180176,
      "learning_rate": 4.013563517741579e-06,
      "loss": 0.2875,
      "step": 20687
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.5940213203430176,
      "learning_rate": 4.012669162000695e-06,
      "loss": 0.2546,
      "step": 20688
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.491432547569275,
      "learning_rate": 4.011774880906112e-06,
      "loss": 0.2615,
      "step": 20689
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6991215944290161,
      "learning_rate": 4.01088067446898e-06,
      "loss": 0.3044,
      "step": 20690
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.091898202896118,
      "learning_rate": 4.009986542700442e-06,
      "loss": 0.3393,
      "step": 20691
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5313224792480469,
      "learning_rate": 4.009092485611654e-06,
      "loss": 0.2419,
      "step": 20692
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5761926174163818,
      "learning_rate": 4.008198503213759e-06,
      "loss": 0.3221,
      "step": 20693
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7757171392440796,
      "learning_rate": 4.007304595517898e-06,
      "loss": 0.3254,
      "step": 20694
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.949733018875122,
      "learning_rate": 4.006410762535224e-06,
      "loss": 0.3085,
      "step": 20695
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7644728422164917,
      "learning_rate": 4.005517004276877e-06,
      "loss": 0.2896,
      "step": 20696
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.1588573455810547,
      "learning_rate": 4.0046233207539945e-06,
      "loss": 0.3559,
      "step": 20697
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8495352268218994,
      "learning_rate": 4.003729711977728e-06,
      "loss": 0.3441,
      "step": 20698
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.860885739326477,
      "learning_rate": 4.002836177959213e-06,
      "loss": 0.3304,
      "step": 20699
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.0553646087646484,
      "learning_rate": 4.001942718709587e-06,
      "loss": 0.3499,
      "step": 20700
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9738739728927612,
      "learning_rate": 4.001049334239997e-06,
      "loss": 0.3241,
      "step": 20701
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.2378084659576416,
      "learning_rate": 4.000156024561575e-06,
      "loss": 0.3881,
      "step": 20702
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.0686275959014893,
      "learning_rate": 3.999262789685461e-06,
      "loss": 0.33,
      "step": 20703
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8580741882324219,
      "learning_rate": 3.998369629622786e-06,
      "loss": 0.3355,
      "step": 20704
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.703035593032837,
      "learning_rate": 3.997476544384693e-06,
      "loss": 0.3156,
      "step": 20705
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7327519655227661,
      "learning_rate": 3.996583533982314e-06,
      "loss": 0.4103,
      "step": 20706
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6595921516418457,
      "learning_rate": 3.9956905984267756e-06,
      "loss": 0.3119,
      "step": 20707
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8755189180374146,
      "learning_rate": 3.994797737729222e-06,
      "loss": 0.3507,
      "step": 20708
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.652032494544983,
      "learning_rate": 3.993904951900778e-06,
      "loss": 0.2266,
      "step": 20709
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.579540729522705,
      "learning_rate": 3.993012240952571e-06,
      "loss": 0.3246,
      "step": 20710
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8237217664718628,
      "learning_rate": 3.992119604895739e-06,
      "loss": 0.3934,
      "step": 20711
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.866457462310791,
      "learning_rate": 3.991227043741407e-06,
      "loss": 0.3962,
      "step": 20712
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.056412935256958,
      "learning_rate": 3.9903345575006994e-06,
      "loss": 0.3734,
      "step": 20713
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8512324094772339,
      "learning_rate": 3.98944214618475e-06,
      "loss": 0.3387,
      "step": 20714
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6015121936798096,
      "learning_rate": 3.988549809804682e-06,
      "loss": 0.3094,
      "step": 20715
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5673305988311768,
      "learning_rate": 3.9876575483716185e-06,
      "loss": 0.2736,
      "step": 20716
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.832275629043579,
      "learning_rate": 3.9867653618966815e-06,
      "loss": 0.3289,
      "step": 20717
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8114086389541626,
      "learning_rate": 3.985873250391003e-06,
      "loss": 0.3107,
      "step": 20718
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7696809768676758,
      "learning_rate": 3.9849812138656985e-06,
      "loss": 0.3393,
      "step": 20719
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8143410682678223,
      "learning_rate": 3.984089252331887e-06,
      "loss": 0.3228,
      "step": 20720
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.771276831626892,
      "learning_rate": 3.983197365800698e-06,
      "loss": 0.3101,
      "step": 20721
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.869263768196106,
      "learning_rate": 3.982305554283245e-06,
      "loss": 0.4004,
      "step": 20722
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6309599876403809,
      "learning_rate": 3.981413817790643e-06,
      "loss": 0.3045,
      "step": 20723
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7514539957046509,
      "learning_rate": 3.980522156334018e-06,
      "loss": 0.3033,
      "step": 20724
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.593762993812561,
      "learning_rate": 3.979630569924482e-06,
      "loss": 0.3544,
      "step": 20725
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.1135640144348145,
      "learning_rate": 3.978739058573149e-06,
      "loss": 0.2674,
      "step": 20726
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.828047752380371,
      "learning_rate": 3.977847622291139e-06,
      "loss": 0.258,
      "step": 20727
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5948312282562256,
      "learning_rate": 3.9769562610895626e-06,
      "loss": 0.4066,
      "step": 20728
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5169999599456787,
      "learning_rate": 3.9760649749795345e-06,
      "loss": 0.2455,
      "step": 20729
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6087758541107178,
      "learning_rate": 3.975173763972161e-06,
      "loss": 0.253,
      "step": 20730
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.569727897644043,
      "learning_rate": 3.974282628078561e-06,
      "loss": 0.2093,
      "step": 20731
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.8522696495056152,
      "learning_rate": 3.973391567309841e-06,
      "loss": 0.3132,
      "step": 20732
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9693443775177002,
      "learning_rate": 3.972500581677107e-06,
      "loss": 0.3011,
      "step": 20733
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.560470700263977,
      "learning_rate": 3.971609671191475e-06,
      "loss": 0.3087,
      "step": 20734
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.9134039878845215,
      "learning_rate": 3.970718835864046e-06,
      "loss": 0.3526,
      "step": 20735
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7337634563446045,
      "learning_rate": 3.969828075705926e-06,
      "loss": 0.2644,
      "step": 20736
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7571957111358643,
      "learning_rate": 3.968937390728226e-06,
      "loss": 0.2732,
      "step": 20737
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.150052070617676,
      "learning_rate": 3.968046780942048e-06,
      "loss": 0.2887,
      "step": 20738
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.2626125812530518,
      "learning_rate": 3.96715624635849e-06,
      "loss": 0.3201,
      "step": 20739
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.0969083309173584,
      "learning_rate": 3.9662657869886634e-06,
      "loss": 0.284,
      "step": 20740
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1759140491485596,
      "learning_rate": 3.965375402843666e-06,
      "loss": 0.3857,
      "step": 20741
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.912907600402832,
      "learning_rate": 3.964485093934598e-06,
      "loss": 0.3382,
      "step": 20742
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9117631912231445,
      "learning_rate": 3.963594860272561e-06,
      "loss": 0.3298,
      "step": 20743
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7429792881011963,
      "learning_rate": 3.962704701868651e-06,
      "loss": 0.2986,
      "step": 20744
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9940897226333618,
      "learning_rate": 3.9618146187339644e-06,
      "loss": 0.3567,
      "step": 20745
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8604964017868042,
      "learning_rate": 3.960924610879605e-06,
      "loss": 0.3567,
      "step": 20746
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9892175197601318,
      "learning_rate": 3.960034678316666e-06,
      "loss": 0.2889,
      "step": 20747
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9043242931365967,
      "learning_rate": 3.959144821056238e-06,
      "loss": 0.4471,
      "step": 20748
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9642834663391113,
      "learning_rate": 3.958255039109421e-06,
      "loss": 0.3578,
      "step": 20749
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.2393062114715576,
      "learning_rate": 3.957365332487307e-06,
      "loss": 0.3624,
      "step": 20750
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.013087034225464,
      "learning_rate": 3.9564757012009865e-06,
      "loss": 0.3716,
      "step": 20751
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6372027397155762,
      "learning_rate": 3.955586145261549e-06,
      "loss": 0.2438,
      "step": 20752
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.073209524154663,
      "learning_rate": 3.954696664680092e-06,
      "loss": 0.3301,
      "step": 20753
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.711751937866211,
      "learning_rate": 3.9538072594677e-06,
      "loss": 0.2962,
      "step": 20754
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8499670028686523,
      "learning_rate": 3.952917929635459e-06,
      "loss": 0.2719,
      "step": 20755
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.4404799938201904,
      "learning_rate": 3.952028675194464e-06,
      "loss": 0.3996,
      "step": 20756
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1266512870788574,
      "learning_rate": 3.951139496155796e-06,
      "loss": 0.3079,
      "step": 20757
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7397810220718384,
      "learning_rate": 3.950250392530539e-06,
      "loss": 0.2585,
      "step": 20758
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7592116594314575,
      "learning_rate": 3.949361364329785e-06,
      "loss": 0.3335,
      "step": 20759
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.195134401321411,
      "learning_rate": 3.948472411564615e-06,
      "loss": 0.3806,
      "step": 20760
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7057985067367554,
      "learning_rate": 3.947583534246105e-06,
      "loss": 0.3102,
      "step": 20761
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.630049228668213,
      "learning_rate": 3.946694732385348e-06,
      "loss": 0.3211,
      "step": 20762
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1192753314971924,
      "learning_rate": 3.945806005993419e-06,
      "loss": 0.3113,
      "step": 20763
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.801783561706543,
      "learning_rate": 3.944917355081399e-06,
      "loss": 0.3634,
      "step": 20764
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.3880414962768555,
      "learning_rate": 3.944028779660364e-06,
      "loss": 0.3598,
      "step": 20765
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.992231011390686,
      "learning_rate": 3.943140279741398e-06,
      "loss": 0.3317,
      "step": 20766
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.6555590629577637,
      "learning_rate": 3.942251855335576e-06,
      "loss": 0.3879,
      "step": 20767
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.126410722732544,
      "learning_rate": 3.94136350645397e-06,
      "loss": 0.3563,
      "step": 20768
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.388050079345703,
      "learning_rate": 3.940475233107663e-06,
      "loss": 0.3083,
      "step": 20769
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8949580192565918,
      "learning_rate": 3.939587035307726e-06,
      "loss": 0.3289,
      "step": 20770
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4856963157653809,
      "learning_rate": 3.938698913065229e-06,
      "loss": 0.2915,
      "step": 20771
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5767295360565186,
      "learning_rate": 3.93781086639125e-06,
      "loss": 0.2276,
      "step": 20772
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6695114374160767,
      "learning_rate": 3.936922895296859e-06,
      "loss": 0.2993,
      "step": 20773
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.422645330429077,
      "learning_rate": 3.9360349997931215e-06,
      "loss": 0.2955,
      "step": 20774
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.958682894706726,
      "learning_rate": 3.935147179891116e-06,
      "loss": 0.2792,
      "step": 20775
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.0321075916290283,
      "learning_rate": 3.934259435601906e-06,
      "loss": 0.274,
      "step": 20776
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8056013584136963,
      "learning_rate": 3.9333717669365615e-06,
      "loss": 0.3348,
      "step": 20777
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.4179065227508545,
      "learning_rate": 3.932484173906143e-06,
      "loss": 0.3591,
      "step": 20778
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.173862934112549,
      "learning_rate": 3.931596656521728e-06,
      "loss": 0.3839,
      "step": 20779
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9000004529953003,
      "learning_rate": 3.930709214794373e-06,
      "loss": 0.4417,
      "step": 20780
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.403515338897705,
      "learning_rate": 3.929821848735141e-06,
      "loss": 0.3695,
      "step": 20781
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4847220182418823,
      "learning_rate": 3.928934558355103e-06,
      "loss": 0.3363,
      "step": 20782
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8485273122787476,
      "learning_rate": 3.928047343665315e-06,
      "loss": 0.3574,
      "step": 20783
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6002446413040161,
      "learning_rate": 3.927160204676837e-06,
      "loss": 0.3508,
      "step": 20784
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.094302177429199,
      "learning_rate": 3.926273141400736e-06,
      "loss": 0.28,
      "step": 20785
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.057051181793213,
      "learning_rate": 3.925386153848068e-06,
      "loss": 0.3342,
      "step": 20786
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9820135831832886,
      "learning_rate": 3.924499242029886e-06,
      "loss": 0.3598,
      "step": 20787
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1056880950927734,
      "learning_rate": 3.923612405957256e-06,
      "loss": 0.3623,
      "step": 20788
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8279117345809937,
      "learning_rate": 3.922725645641231e-06,
      "loss": 0.358,
      "step": 20789
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.168346881866455,
      "learning_rate": 3.921838961092867e-06,
      "loss": 0.3333,
      "step": 20790
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5667465925216675,
      "learning_rate": 3.920952352323213e-06,
      "loss": 0.2655,
      "step": 20791
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.032297372817993,
      "learning_rate": 3.920065819343333e-06,
      "loss": 0.3956,
      "step": 20792
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.615056037902832,
      "learning_rate": 3.919179362164273e-06,
      "loss": 0.2521,
      "step": 20793
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.922651767730713,
      "learning_rate": 3.918292980797084e-06,
      "loss": 0.3401,
      "step": 20794
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9220279455184937,
      "learning_rate": 3.917406675252822e-06,
      "loss": 0.3371,
      "step": 20795
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.008039951324463,
      "learning_rate": 3.916520445542533e-06,
      "loss": 0.2796,
      "step": 20796
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8389978408813477,
      "learning_rate": 3.915634291677265e-06,
      "loss": 0.342,
      "step": 20797
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.75919508934021,
      "learning_rate": 3.91474821366807e-06,
      "loss": 0.3777,
      "step": 20798
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.893666386604309,
      "learning_rate": 3.913862211525995e-06,
      "loss": 0.3141,
      "step": 20799
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.0397744178771973,
      "learning_rate": 3.912976285262079e-06,
      "loss": 0.314,
      "step": 20800
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9243122339248657,
      "learning_rate": 3.912090434887376e-06,
      "loss": 0.3234,
      "step": 20801
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9532679319381714,
      "learning_rate": 3.911204660412927e-06,
      "loss": 0.3203,
      "step": 20802
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.0013279914855957,
      "learning_rate": 3.910318961849775e-06,
      "loss": 0.3284,
      "step": 20803
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8559072017669678,
      "learning_rate": 3.909433339208959e-06,
      "loss": 0.2989,
      "step": 20804
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.150667905807495,
      "learning_rate": 3.908547792501527e-06,
      "loss": 0.1621,
      "step": 20805
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8962961435317993,
      "learning_rate": 3.907662321738514e-06,
      "loss": 0.3165,
      "step": 20806
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8891749382019043,
      "learning_rate": 3.906776926930961e-06,
      "loss": 0.3266,
      "step": 20807
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.116729259490967,
      "learning_rate": 3.905891608089909e-06,
      "loss": 0.4052,
      "step": 20808
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1500396728515625,
      "learning_rate": 3.905006365226393e-06,
      "loss": 0.28,
      "step": 20809
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6288719177246094,
      "learning_rate": 3.904121198351448e-06,
      "loss": 0.2652,
      "step": 20810
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8618521690368652,
      "learning_rate": 3.903236107476115e-06,
      "loss": 0.2979,
      "step": 20811
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5293947458267212,
      "learning_rate": 3.902351092611426e-06,
      "loss": 0.3308,
      "step": 20812
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7439453601837158,
      "learning_rate": 3.9014661537684115e-06,
      "loss": 0.3386,
      "step": 20813
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7165530920028687,
      "learning_rate": 3.900581290958111e-06,
      "loss": 0.2698,
      "step": 20814
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.261915445327759,
      "learning_rate": 3.8996965041915516e-06,
      "loss": 0.3824,
      "step": 20815
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.9427919387817383,
      "learning_rate": 3.898811793479767e-06,
      "loss": 0.3155,
      "step": 20816
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.950627326965332,
      "learning_rate": 3.897927158833781e-06,
      "loss": 0.4044,
      "step": 20817
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.456395387649536,
      "learning_rate": 3.897042600264632e-06,
      "loss": 0.2783,
      "step": 20818
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1005430221557617,
      "learning_rate": 3.896158117783343e-06,
      "loss": 0.3831,
      "step": 20819
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.616896629333496,
      "learning_rate": 3.895273711400939e-06,
      "loss": 0.2675,
      "step": 20820
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7626653909683228,
      "learning_rate": 3.894389381128451e-06,
      "loss": 0.3401,
      "step": 20821
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8942819833755493,
      "learning_rate": 3.893505126976903e-06,
      "loss": 0.2455,
      "step": 20822
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.52092707157135,
      "learning_rate": 3.892620948957315e-06,
      "loss": 0.2614,
      "step": 20823
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.474886894226074,
      "learning_rate": 3.891736847080718e-06,
      "loss": 0.353,
      "step": 20824
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.03942608833313,
      "learning_rate": 3.89085282135813e-06,
      "loss": 0.3521,
      "step": 20825
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4569944143295288,
      "learning_rate": 3.889968871800569e-06,
      "loss": 0.257,
      "step": 20826
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.1449179649353027,
      "learning_rate": 3.889084998419063e-06,
      "loss": 0.3767,
      "step": 20827
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7781360149383545,
      "learning_rate": 3.888201201224629e-06,
      "loss": 0.3262,
      "step": 20828
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6908390522003174,
      "learning_rate": 3.887317480228283e-06,
      "loss": 0.2817,
      "step": 20829
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.895801067352295,
      "learning_rate": 3.886433835441041e-06,
      "loss": 0.3233,
      "step": 20830
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7839032411575317,
      "learning_rate": 3.8855502668739274e-06,
      "loss": 0.2815,
      "step": 20831
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8268672227859497,
      "learning_rate": 3.884666774537954e-06,
      "loss": 0.3166,
      "step": 20832
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.877055287361145,
      "learning_rate": 3.883783358444131e-06,
      "loss": 0.3231,
      "step": 20833
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.686283826828003,
      "learning_rate": 3.88290001860348e-06,
      "loss": 0.327,
      "step": 20834
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.000722646713257,
      "learning_rate": 3.88201675502701e-06,
      "loss": 0.3466,
      "step": 20835
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8779020309448242,
      "learning_rate": 3.881133567725731e-06,
      "loss": 0.3022,
      "step": 20836
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7618361711502075,
      "learning_rate": 3.880250456710659e-06,
      "loss": 0.3149,
      "step": 20837
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9269453287124634,
      "learning_rate": 3.879367421992802e-06,
      "loss": 0.388,
      "step": 20838
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.121617555618286,
      "learning_rate": 3.8784844635831655e-06,
      "loss": 0.3245,
      "step": 20839
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7197396755218506,
      "learning_rate": 3.877601581492764e-06,
      "loss": 0.2484,
      "step": 20840
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9516189098358154,
      "learning_rate": 3.876718775732601e-06,
      "loss": 0.3397,
      "step": 20841
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.033278465270996,
      "learning_rate": 3.875836046313683e-06,
      "loss": 0.2557,
      "step": 20842
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7299916744232178,
      "learning_rate": 3.8749533932470126e-06,
      "loss": 0.2916,
      "step": 20843
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9779239892959595,
      "learning_rate": 3.874070816543601e-06,
      "loss": 0.34,
      "step": 20844
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.039238214492798,
      "learning_rate": 3.8731883162144476e-06,
      "loss": 0.2512,
      "step": 20845
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8358347415924072,
      "learning_rate": 3.8723058922705524e-06,
      "loss": 0.299,
      "step": 20846
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.058684825897217,
      "learning_rate": 3.871423544722922e-06,
      "loss": 0.3439,
      "step": 20847
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.631047248840332,
      "learning_rate": 3.870541273582555e-06,
      "loss": 0.2382,
      "step": 20848
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8399732112884521,
      "learning_rate": 3.869659078860447e-06,
      "loss": 0.348,
      "step": 20849
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9626096487045288,
      "learning_rate": 3.868776960567604e-06,
      "loss": 0.2721,
      "step": 20850
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.304926872253418,
      "learning_rate": 3.86789491871502e-06,
      "loss": 0.2703,
      "step": 20851
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8720388412475586,
      "learning_rate": 3.867012953313687e-06,
      "loss": 0.3603,
      "step": 20852
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.544329285621643,
      "learning_rate": 3.86613106437461e-06,
      "loss": 0.2561,
      "step": 20853
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.748771071434021,
      "learning_rate": 3.865249251908779e-06,
      "loss": 0.3091,
      "step": 20854
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6477437019348145,
      "learning_rate": 3.8643675159271875e-06,
      "loss": 0.2609,
      "step": 20855
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7979280948638916,
      "learning_rate": 3.863485856440826e-06,
      "loss": 0.3894,
      "step": 20856
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.045313596725464,
      "learning_rate": 3.862604273460694e-06,
      "loss": 0.3146,
      "step": 20857
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.0444841384887695,
      "learning_rate": 3.861722766997777e-06,
      "loss": 0.3266,
      "step": 20858
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7195745706558228,
      "learning_rate": 3.860841337063064e-06,
      "loss": 0.3181,
      "step": 20859
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.075995683670044,
      "learning_rate": 3.859959983667548e-06,
      "loss": 0.3781,
      "step": 20860
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6124769449234009,
      "learning_rate": 3.859078706822217e-06,
      "loss": 0.2555,
      "step": 20861
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.098745346069336,
      "learning_rate": 3.858197506538053e-06,
      "loss": 0.2143,
      "step": 20862
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.037548303604126,
      "learning_rate": 3.857316382826048e-06,
      "loss": 0.2913,
      "step": 20863
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8380578756332397,
      "learning_rate": 3.856435335697186e-06,
      "loss": 0.3292,
      "step": 20864
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.1295359134674072,
      "learning_rate": 3.8555543651624486e-06,
      "loss": 0.2514,
      "step": 20865
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7113467454910278,
      "learning_rate": 3.854673471232823e-06,
      "loss": 0.2334,
      "step": 20866
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7148652076721191,
      "learning_rate": 3.85379265391929e-06,
      "loss": 0.2909,
      "step": 20867
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8636744022369385,
      "learning_rate": 3.852911913232831e-06,
      "loss": 0.3418,
      "step": 20868
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.2599596977233887,
      "learning_rate": 3.852031249184423e-06,
      "loss": 0.3264,
      "step": 20869
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.913116693496704,
      "learning_rate": 3.851150661785052e-06,
      "loss": 0.3581,
      "step": 20870
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.984972357749939,
      "learning_rate": 3.850270151045694e-06,
      "loss": 0.2466,
      "step": 20871
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9913345575332642,
      "learning_rate": 3.849389716977323e-06,
      "loss": 0.3652,
      "step": 20872
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8766100406646729,
      "learning_rate": 3.848509359590922e-06,
      "loss": 0.3545,
      "step": 20873
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.5069680213928223,
      "learning_rate": 3.847629078897464e-06,
      "loss": 0.2859,
      "step": 20874
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9661719799041748,
      "learning_rate": 3.846748874907919e-06,
      "loss": 0.3418,
      "step": 20875
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5471229553222656,
      "learning_rate": 3.84586874763327e-06,
      "loss": 0.2643,
      "step": 20876
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.784160852432251,
      "learning_rate": 3.844988697084485e-06,
      "loss": 0.3216,
      "step": 20877
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7468669414520264,
      "learning_rate": 3.844108723272533e-06,
      "loss": 0.3157,
      "step": 20878
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.869014859199524,
      "learning_rate": 3.8432288262083905e-06,
      "loss": 0.3085,
      "step": 20879
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7870924472808838,
      "learning_rate": 3.8423490059030255e-06,
      "loss": 0.3184,
      "step": 20880
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6543000936508179,
      "learning_rate": 3.841469262367407e-06,
      "loss": 0.3593,
      "step": 20881
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.812767744064331,
      "learning_rate": 3.8405895956124994e-06,
      "loss": 0.2653,
      "step": 20882
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4204806089401245,
      "learning_rate": 3.839710005649277e-06,
      "loss": 0.2995,
      "step": 20883
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9877923727035522,
      "learning_rate": 3.838830492488701e-06,
      "loss": 0.3082,
      "step": 20884
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.1330063343048096,
      "learning_rate": 3.837951056141735e-06,
      "loss": 0.3043,
      "step": 20885
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5856870412826538,
      "learning_rate": 3.837071696619349e-06,
      "loss": 0.2916,
      "step": 20886
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9883818626403809,
      "learning_rate": 3.836192413932502e-06,
      "loss": 0.326,
      "step": 20887
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.675607442855835,
      "learning_rate": 3.835313208092156e-06,
      "loss": 0.291,
      "step": 20888
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9919819831848145,
      "learning_rate": 3.834434079109278e-06,
      "loss": 0.3243,
      "step": 20889
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.5796000957489014,
      "learning_rate": 3.833555026994824e-06,
      "loss": 0.3928,
      "step": 20890
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9482839107513428,
      "learning_rate": 3.83267605175975e-06,
      "loss": 0.3214,
      "step": 20891
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.017892360687256,
      "learning_rate": 3.831797153415022e-06,
      "loss": 0.2997,
      "step": 20892
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8231110572814941,
      "learning_rate": 3.830918331971594e-06,
      "loss": 0.3437,
      "step": 20893
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.2529890537261963,
      "learning_rate": 3.830039587440423e-06,
      "loss": 0.3507,
      "step": 20894
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9118446111679077,
      "learning_rate": 3.8291609198324605e-06,
      "loss": 0.2803,
      "step": 20895
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9404600858688354,
      "learning_rate": 3.828282329158669e-06,
      "loss": 0.2885,
      "step": 20896
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9759860038757324,
      "learning_rate": 3.827403815429998e-06,
      "loss": 0.3447,
      "step": 20897
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7640411853790283,
      "learning_rate": 3.826525378657397e-06,
      "loss": 0.2789,
      "step": 20898
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9842462539672852,
      "learning_rate": 3.825647018851823e-06,
      "loss": 0.2996,
      "step": 20899
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.965088129043579,
      "learning_rate": 3.824768736024227e-06,
      "loss": 0.2986,
      "step": 20900
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.7908363342285156,
      "learning_rate": 3.823890530185554e-06,
      "loss": 0.337,
      "step": 20901
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4930466413497925,
      "learning_rate": 3.8230124013467576e-06,
      "loss": 0.233,
      "step": 20902
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.967390537261963,
      "learning_rate": 3.822134349518785e-06,
      "loss": 0.3229,
      "step": 20903
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6967674493789673,
      "learning_rate": 3.821256374712579e-06,
      "loss": 0.2652,
      "step": 20904
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.665799856185913,
      "learning_rate": 3.820378476939092e-06,
      "loss": 0.3477,
      "step": 20905
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8140873908996582,
      "learning_rate": 3.819500656209265e-06,
      "loss": 0.2944,
      "step": 20906
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.0996217727661133,
      "learning_rate": 3.818622912534044e-06,
      "loss": 0.3081,
      "step": 20907
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.403721332550049,
      "learning_rate": 3.817745245924368e-06,
      "loss": 0.3538,
      "step": 20908
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.7807823419570923,
      "learning_rate": 3.8168676563911855e-06,
      "loss": 0.4089,
      "step": 20909
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.1420388221740723,
      "learning_rate": 3.815990143945435e-06,
      "loss": 0.3435,
      "step": 20910
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.031743049621582,
      "learning_rate": 3.815112708598052e-06,
      "loss": 0.3495,
      "step": 20911
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.505212664604187,
      "learning_rate": 3.814235350359985e-06,
      "loss": 0.302,
      "step": 20912
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5613813400268555,
      "learning_rate": 3.813358069242168e-06,
      "loss": 0.334,
      "step": 20913
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9605029821395874,
      "learning_rate": 3.8124808652555336e-06,
      "loss": 0.2902,
      "step": 20914
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.2941982746124268,
      "learning_rate": 3.811603738411026e-06,
      "loss": 0.3867,
      "step": 20915
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.1687023639678955,
      "learning_rate": 3.810726688719577e-06,
      "loss": 0.3451,
      "step": 20916
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.1538987159729004,
      "learning_rate": 3.8098497161921188e-06,
      "loss": 0.3922,
      "step": 20917
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5613253116607666,
      "learning_rate": 3.8089728208395904e-06,
      "loss": 0.3082,
      "step": 20918
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6595107316970825,
      "learning_rate": 3.808096002672923e-06,
      "loss": 0.2199,
      "step": 20919
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6725690364837646,
      "learning_rate": 3.807219261703046e-06,
      "loss": 0.3163,
      "step": 20920
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6510183811187744,
      "learning_rate": 3.8063425979408864e-06,
      "loss": 0.3107,
      "step": 20921
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6196378469467163,
      "learning_rate": 3.805466011397383e-06,
      "loss": 0.2532,
      "step": 20922
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.585372805595398,
      "learning_rate": 3.804589502083459e-06,
      "loss": 0.3091,
      "step": 20923
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.027650833129883,
      "learning_rate": 3.80371307001004e-06,
      "loss": 0.3041,
      "step": 20924
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.8482574224472046,
      "learning_rate": 3.8028367151880597e-06,
      "loss": 0.3348,
      "step": 20925
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.113832712173462,
      "learning_rate": 3.8019604376284393e-06,
      "loss": 0.3259,
      "step": 20926
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.751585841178894,
      "learning_rate": 3.8010842373421007e-06,
      "loss": 0.4015,
      "step": 20927
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.0426604747772217,
      "learning_rate": 3.800208114339976e-06,
      "loss": 0.3817,
      "step": 20928
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9275892972946167,
      "learning_rate": 3.799332068632984e-06,
      "loss": 0.2781,
      "step": 20929
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.714124083518982,
      "learning_rate": 3.798456100232042e-06,
      "loss": 0.3295,
      "step": 20930
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6332035064697266,
      "learning_rate": 3.7975802091480795e-06,
      "loss": 0.3008,
      "step": 20931
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5683501958847046,
      "learning_rate": 3.7967043953920123e-06,
      "loss": 0.3036,
      "step": 20932
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.2060792446136475,
      "learning_rate": 3.7958286589747593e-06,
      "loss": 0.3412,
      "step": 20933
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7246731519699097,
      "learning_rate": 3.7949529999072366e-06,
      "loss": 0.3216,
      "step": 20934
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6671333312988281,
      "learning_rate": 3.794077418200367e-06,
      "loss": 0.336,
      "step": 20935
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.078937530517578,
      "learning_rate": 3.7932019138650633e-06,
      "loss": 0.3075,
      "step": 20936
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.740457534790039,
      "learning_rate": 3.792326486912238e-06,
      "loss": 0.2904,
      "step": 20937
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7351704835891724,
      "learning_rate": 3.7914511373528105e-06,
      "loss": 0.2896,
      "step": 20938
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.045194387435913,
      "learning_rate": 3.790575865197693e-06,
      "loss": 0.237,
      "step": 20939
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0161256790161133,
      "learning_rate": 3.7897006704577934e-06,
      "loss": 0.3779,
      "step": 20940
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9330568313598633,
      "learning_rate": 3.78882555314403e-06,
      "loss": 0.2454,
      "step": 20941
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6297540664672852,
      "learning_rate": 3.7879505132673087e-06,
      "loss": 0.2894,
      "step": 20942
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6217083930969238,
      "learning_rate": 3.7870755508385382e-06,
      "loss": 0.2974,
      "step": 20943
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8451303243637085,
      "learning_rate": 3.786200665868631e-06,
      "loss": 0.3013,
      "step": 20944
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.656498670578003,
      "learning_rate": 3.7853258583684928e-06,
      "loss": 0.3083,
      "step": 20945
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.6618874073028564,
      "learning_rate": 3.7844511283490293e-06,
      "loss": 0.3608,
      "step": 20946
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9234154224395752,
      "learning_rate": 3.7835764758211434e-06,
      "loss": 0.2981,
      "step": 20947
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.734075903892517,
      "learning_rate": 3.7827019007957467e-06,
      "loss": 0.2984,
      "step": 20948
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0163755416870117,
      "learning_rate": 3.7818274032837378e-06,
      "loss": 0.3662,
      "step": 20949
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.2007391452789307,
      "learning_rate": 3.780952983296018e-06,
      "loss": 0.3351,
      "step": 20950
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.221331834793091,
      "learning_rate": 3.780078640843494e-06,
      "loss": 0.3135,
      "step": 20951
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6463385820388794,
      "learning_rate": 3.7792043759370654e-06,
      "loss": 0.323,
      "step": 20952
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7917022705078125,
      "learning_rate": 3.7783301885876267e-06,
      "loss": 0.3519,
      "step": 20953
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7137062549591064,
      "learning_rate": 3.7774560788060844e-06,
      "loss": 0.2462,
      "step": 20954
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7164024114608765,
      "learning_rate": 3.776582046603333e-06,
      "loss": 0.3335,
      "step": 20955
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.093248128890991,
      "learning_rate": 3.7757080919902657e-06,
      "loss": 0.3139,
      "step": 20956
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7148184776306152,
      "learning_rate": 3.774834214977784e-06,
      "loss": 0.3176,
      "step": 20957
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9015796184539795,
      "learning_rate": 3.773960415576782e-06,
      "loss": 0.311,
      "step": 20958
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0677971839904785,
      "learning_rate": 3.7730866937981514e-06,
      "loss": 0.3549,
      "step": 20959
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8316503763198853,
      "learning_rate": 3.7722130496527844e-06,
      "loss": 0.2527,
      "step": 20960
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.200894594192505,
      "learning_rate": 3.771339483151576e-06,
      "loss": 0.3142,
      "step": 20961
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0167763233184814,
      "learning_rate": 3.770465994305418e-06,
      "loss": 0.3397,
      "step": 20962
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6896780729293823,
      "learning_rate": 3.7695925831251946e-06,
      "loss": 0.2656,
      "step": 20963
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7425881624221802,
      "learning_rate": 3.7687192496218017e-06,
      "loss": 0.3765,
      "step": 20964
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.375901699066162,
      "learning_rate": 3.767845993806125e-06,
      "loss": 0.2888,
      "step": 20965
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.88779616355896,
      "learning_rate": 3.7669728156890484e-06,
      "loss": 0.3104,
      "step": 20966
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.883052110671997,
      "learning_rate": 3.766099715281465e-06,
      "loss": 0.3189,
      "step": 20967
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7616420984268188,
      "learning_rate": 3.765226692594255e-06,
      "loss": 0.2726,
      "step": 20968
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.713566780090332,
      "learning_rate": 3.764353747638302e-06,
      "loss": 0.2589,
      "step": 20969
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7650102376937866,
      "learning_rate": 3.7634808804244937e-06,
      "loss": 0.3663,
      "step": 20970
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0110480785369873,
      "learning_rate": 3.7626080909637097e-06,
      "loss": 0.3324,
      "step": 20971
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.252230644226074,
      "learning_rate": 3.761735379266833e-06,
      "loss": 0.3322,
      "step": 20972
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.896767497062683,
      "learning_rate": 3.7608627453447387e-06,
      "loss": 0.2509,
      "step": 20973
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6968575716018677,
      "learning_rate": 3.7599901892083137e-06,
      "loss": 0.3246,
      "step": 20974
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6329503059387207,
      "learning_rate": 3.759117710868433e-06,
      "loss": 0.3822,
      "step": 20975
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8807426691055298,
      "learning_rate": 3.758245310335972e-06,
      "loss": 0.249,
      "step": 20976
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.1249594688415527,
      "learning_rate": 3.7573729876218113e-06,
      "loss": 0.325,
      "step": 20977
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8228904008865356,
      "learning_rate": 3.756500742736825e-06,
      "loss": 0.3138,
      "step": 20978
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6890631914138794,
      "learning_rate": 3.755628575691884e-06,
      "loss": 0.3397,
      "step": 20979
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8748371601104736,
      "learning_rate": 3.754756486497869e-06,
      "loss": 0.27,
      "step": 20980
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0671257972717285,
      "learning_rate": 3.753884475165649e-06,
      "loss": 0.3181,
      "step": 20981
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0611000061035156,
      "learning_rate": 3.7530125417060926e-06,
      "loss": 0.3357,
      "step": 20982
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.930404782295227,
      "learning_rate": 3.752140686130076e-06,
      "loss": 0.32,
      "step": 20983
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6113077402114868,
      "learning_rate": 3.7512689084484665e-06,
      "loss": 0.1989,
      "step": 20984
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0395278930664062,
      "learning_rate": 3.750397208672134e-06,
      "loss": 0.3809,
      "step": 20985
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.1336510181427,
      "learning_rate": 3.7495255868119405e-06,
      "loss": 0.3053,
      "step": 20986
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9507834911346436,
      "learning_rate": 3.748654042878762e-06,
      "loss": 0.353,
      "step": 20987
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0163767337799072,
      "learning_rate": 3.74778257688346e-06,
      "loss": 0.395,
      "step": 20988
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7435302734375,
      "learning_rate": 3.746911188836895e-06,
      "loss": 0.3339,
      "step": 20989
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9105799198150635,
      "learning_rate": 3.7460398787499387e-06,
      "loss": 0.2513,
      "step": 20990
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.9726808071136475,
      "learning_rate": 3.745168646633451e-06,
      "loss": 0.3732,
      "step": 20991
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0130319595336914,
      "learning_rate": 3.74429749249829e-06,
      "loss": 0.3381,
      "step": 20992
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8912463188171387,
      "learning_rate": 3.7434264163553237e-06,
      "loss": 0.3958,
      "step": 20993
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6704113483428955,
      "learning_rate": 3.7425554182154088e-06,
      "loss": 0.3198,
      "step": 20994
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.3234729766845703,
      "learning_rate": 3.7416844980893997e-06,
      "loss": 0.335,
      "step": 20995
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.064055919647217,
      "learning_rate": 3.740813655988164e-06,
      "loss": 0.3688,
      "step": 20996
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.502954125404358,
      "learning_rate": 3.739942891922552e-06,
      "loss": 0.2713,
      "step": 20997
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9801161289215088,
      "learning_rate": 3.7390722059034222e-06,
      "loss": 0.381,
      "step": 20998
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6599161624908447,
      "learning_rate": 3.7382015979416264e-06,
      "loss": 0.2783,
      "step": 20999
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6301405429840088,
      "learning_rate": 3.7373310680480236e-06,
      "loss": 0.263,
      "step": 21000
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5833696126937866,
      "learning_rate": 3.736460616233466e-06,
      "loss": 0.2775,
      "step": 21001
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0231385231018066,
      "learning_rate": 3.7355902425088007e-06,
      "loss": 0.3187,
      "step": 21002
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.310592532157898,
      "learning_rate": 3.7347199468848883e-06,
      "loss": 0.2315,
      "step": 21003
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.095836639404297,
      "learning_rate": 3.7338497293725717e-06,
      "loss": 0.3501,
      "step": 21004
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.051990509033203,
      "learning_rate": 3.7329795899827006e-06,
      "loss": 0.4414,
      "step": 21005
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9072062969207764,
      "learning_rate": 3.732109528726128e-06,
      "loss": 0.3516,
      "step": 21006
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8412195444107056,
      "learning_rate": 3.7312395456136984e-06,
      "loss": 0.3617,
      "step": 21007
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8564813137054443,
      "learning_rate": 3.730369640656255e-06,
      "loss": 0.3122,
      "step": 21008
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9545120000839233,
      "learning_rate": 3.72949981386465e-06,
      "loss": 0.274,
      "step": 21009
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8580477237701416,
      "learning_rate": 3.7286300652497242e-06,
      "loss": 0.2365,
      "step": 21010
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7686800956726074,
      "learning_rate": 3.7277603948223206e-06,
      "loss": 0.3023,
      "step": 21011
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7584412097930908,
      "learning_rate": 3.7268908025932802e-06,
      "loss": 0.3766,
      "step": 21012
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.960461974143982,
      "learning_rate": 3.726021288573449e-06,
      "loss": 0.2796,
      "step": 21013
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.810799479484558,
      "learning_rate": 3.725151852773666e-06,
      "loss": 0.288,
      "step": 21014
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8206857442855835,
      "learning_rate": 3.724282495204766e-06,
      "loss": 0.27,
      "step": 21015
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6496659517288208,
      "learning_rate": 3.723413215877596e-06,
      "loss": 0.3625,
      "step": 21016
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5326817035675049,
      "learning_rate": 3.722544014802989e-06,
      "loss": 0.3355,
      "step": 21017
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.168008327484131,
      "learning_rate": 3.7216748919917777e-06,
      "loss": 0.343,
      "step": 21018
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.43009352684021,
      "learning_rate": 3.720805847454806e-06,
      "loss": 0.2303,
      "step": 21019
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.9740667343139648,
      "learning_rate": 3.7199368812029046e-06,
      "loss": 0.2533,
      "step": 21020
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8710867166519165,
      "learning_rate": 3.719067993246904e-06,
      "loss": 0.3981,
      "step": 21021
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.623993158340454,
      "learning_rate": 3.7181991835976427e-06,
      "loss": 0.3291,
      "step": 21022
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.202812671661377,
      "learning_rate": 3.7173304522659513e-06,
      "loss": 0.3297,
      "step": 21023
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6291258335113525,
      "learning_rate": 3.7164617992626585e-06,
      "loss": 0.2969,
      "step": 21024
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.218733310699463,
      "learning_rate": 3.7155932245985914e-06,
      "loss": 0.3112,
      "step": 21025
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7561092376708984,
      "learning_rate": 3.7147247282845854e-06,
      "loss": 0.3432,
      "step": 21026
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7701046466827393,
      "learning_rate": 3.7138563103314663e-06,
      "loss": 0.3458,
      "step": 21027
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.483044385910034,
      "learning_rate": 3.712987970750056e-06,
      "loss": 0.4008,
      "step": 21028
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.008138418197632,
      "learning_rate": 3.7121197095511883e-06,
      "loss": 0.3605,
      "step": 21029
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.8772025108337402,
      "learning_rate": 3.711251526745684e-06,
      "loss": 0.3401,
      "step": 21030
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7940638065338135,
      "learning_rate": 3.7103834223443637e-06,
      "loss": 0.3148,
      "step": 21031
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.118321180343628,
      "learning_rate": 3.7095153963580575e-06,
      "loss": 0.294,
      "step": 21032
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5013056993484497,
      "learning_rate": 3.708647448797584e-06,
      "loss": 0.2585,
      "step": 21033
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6894882917404175,
      "learning_rate": 3.7077795796737605e-06,
      "loss": 0.2932,
      "step": 21034
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.0044925212860107,
      "learning_rate": 3.706911788997414e-06,
      "loss": 0.3036,
      "step": 21035
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7183232307434082,
      "learning_rate": 3.7060440767793605e-06,
      "loss": 0.2868,
      "step": 21036
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9765141010284424,
      "learning_rate": 3.705176443030418e-06,
      "loss": 0.3244,
      "step": 21037
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.070889711380005,
      "learning_rate": 3.7043088877613996e-06,
      "loss": 0.3513,
      "step": 21038
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.167677402496338,
      "learning_rate": 3.7034414109831296e-06,
      "loss": 0.2906,
      "step": 21039
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8701834678649902,
      "learning_rate": 3.702574012706418e-06,
      "loss": 0.303,
      "step": 21040
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7151284217834473,
      "learning_rate": 3.7017066929420766e-06,
      "loss": 0.2995,
      "step": 21041
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9433445930480957,
      "learning_rate": 3.700839451700925e-06,
      "loss": 0.3036,
      "step": 21042
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.125474452972412,
      "learning_rate": 3.699972288993773e-06,
      "loss": 0.3414,
      "step": 21043
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5426316261291504,
      "learning_rate": 3.699105204831427e-06,
      "loss": 0.2979,
      "step": 21044
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.684358835220337,
      "learning_rate": 3.6982381992247053e-06,
      "loss": 0.3305,
      "step": 21045
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5793696641921997,
      "learning_rate": 3.6973712721844126e-06,
      "loss": 0.2901,
      "step": 21046
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5553914308547974,
      "learning_rate": 3.696504423721354e-06,
      "loss": 0.3057,
      "step": 21047
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8234453201293945,
      "learning_rate": 3.6956376538463455e-06,
      "loss": 0.3481,
      "step": 21048
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.997511863708496,
      "learning_rate": 3.694770962570189e-06,
      "loss": 0.2832,
      "step": 21049
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.982399582862854,
      "learning_rate": 3.6939043499036875e-06,
      "loss": 0.3086,
      "step": 21050
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.574739933013916,
      "learning_rate": 3.693037815857645e-06,
      "loss": 0.2025,
      "step": 21051
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.800121307373047,
      "learning_rate": 3.692171360442871e-06,
      "loss": 0.3472,
      "step": 21052
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.2320773601531982,
      "learning_rate": 3.691304983670164e-06,
      "loss": 0.2892,
      "step": 21053
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7775774002075195,
      "learning_rate": 3.6904386855503214e-06,
      "loss": 0.3727,
      "step": 21054
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8434571027755737,
      "learning_rate": 3.6895724660941524e-06,
      "loss": 0.2838,
      "step": 21055
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7456371784210205,
      "learning_rate": 3.688706325312452e-06,
      "loss": 0.3149,
      "step": 21056
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.3416171073913574,
      "learning_rate": 3.6878402632160147e-06,
      "loss": 0.2981,
      "step": 21057
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.799953818321228,
      "learning_rate": 3.686974279815646e-06,
      "loss": 0.3267,
      "step": 21058
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7597970962524414,
      "learning_rate": 3.686108375122138e-06,
      "loss": 0.2832,
      "step": 21059
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6105191707611084,
      "learning_rate": 3.685242549146284e-06,
      "loss": 0.3381,
      "step": 21060
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.914158821105957,
      "learning_rate": 3.6843768018988835e-06,
      "loss": 0.339,
      "step": 21061
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8276633024215698,
      "learning_rate": 3.683511133390729e-06,
      "loss": 0.3374,
      "step": 21062
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.680192470550537,
      "learning_rate": 3.6826455436326115e-06,
      "loss": 0.266,
      "step": 21063
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.4644925594329834,
      "learning_rate": 3.6817800326353192e-06,
      "loss": 0.3457,
      "step": 21064
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8872162103652954,
      "learning_rate": 3.6809146004096508e-06,
      "loss": 0.3413,
      "step": 21065
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.018338441848755,
      "learning_rate": 3.680049246966392e-06,
      "loss": 0.3685,
      "step": 21066
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.095484972000122,
      "learning_rate": 3.679183972316328e-06,
      "loss": 0.2599,
      "step": 21067
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.3053691387176514,
      "learning_rate": 3.6783187764702523e-06,
      "loss": 0.3673,
      "step": 21068
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9066836833953857,
      "learning_rate": 3.67745365943895e-06,
      "loss": 0.3606,
      "step": 21069
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6392139196395874,
      "learning_rate": 3.6765886212332024e-06,
      "loss": 0.2827,
      "step": 21070
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6827417612075806,
      "learning_rate": 3.6757236618638016e-06,
      "loss": 0.306,
      "step": 21071
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.2375988960266113,
      "learning_rate": 3.674858781341527e-06,
      "loss": 0.3279,
      "step": 21072
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7042491436004639,
      "learning_rate": 3.6739939796771594e-06,
      "loss": 0.2768,
      "step": 21073
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.050875663757324,
      "learning_rate": 3.673129256881486e-06,
      "loss": 0.3318,
      "step": 21074
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9334096908569336,
      "learning_rate": 3.672264612965285e-06,
      "loss": 0.2906,
      "step": 21075
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8890291452407837,
      "learning_rate": 3.6714000479393362e-06,
      "loss": 0.3516,
      "step": 21076
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6768646240234375,
      "learning_rate": 3.6705355618144145e-06,
      "loss": 0.3668,
      "step": 21077
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8787198066711426,
      "learning_rate": 3.6696711546013065e-06,
      "loss": 0.355,
      "step": 21078
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5402426719665527,
      "learning_rate": 3.6688068263107822e-06,
      "loss": 0.3243,
      "step": 21079
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.052607774734497,
      "learning_rate": 3.667942576953617e-06,
      "loss": 0.2737,
      "step": 21080
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1072304248809814,
      "learning_rate": 3.667078406540592e-06,
      "loss": 0.36,
      "step": 21081
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.2149007320404053,
      "learning_rate": 3.6662143150824782e-06,
      "loss": 0.3641,
      "step": 21082
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6890771389007568,
      "learning_rate": 3.6653503025900427e-06,
      "loss": 0.2833,
      "step": 21083
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.083376407623291,
      "learning_rate": 3.6644863690740673e-06,
      "loss": 0.3116,
      "step": 21084
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.0193958282470703,
      "learning_rate": 3.6636225145453174e-06,
      "loss": 0.3088,
      "step": 21085
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9895931482315063,
      "learning_rate": 3.662758739014561e-06,
      "loss": 0.3054,
      "step": 21086
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6336796283721924,
      "learning_rate": 3.661895042492574e-06,
      "loss": 0.2638,
      "step": 21087
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.2453339099884033,
      "learning_rate": 3.661031424990119e-06,
      "loss": 0.2536,
      "step": 21088
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9468719959259033,
      "learning_rate": 3.6601678865179656e-06,
      "loss": 0.3322,
      "step": 21089
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.3239729404449463,
      "learning_rate": 3.659304427086875e-06,
      "loss": 0.3206,
      "step": 21090
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8421332836151123,
      "learning_rate": 3.6584410467076194e-06,
      "loss": 0.2942,
      "step": 21091
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5643483400344849,
      "learning_rate": 3.657577745390959e-06,
      "loss": 0.2944,
      "step": 21092
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7557367086410522,
      "learning_rate": 3.6567145231476543e-06,
      "loss": 0.3248,
      "step": 21093
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8581219911575317,
      "learning_rate": 3.6558513799884776e-06,
      "loss": 0.3359,
      "step": 21094
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.894478678703308,
      "learning_rate": 3.6549883159241797e-06,
      "loss": 0.316,
      "step": 21095
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9288313388824463,
      "learning_rate": 3.6541253309655202e-06,
      "loss": 0.3209,
      "step": 21096
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.885872721672058,
      "learning_rate": 3.653262425123265e-06,
      "loss": 0.2846,
      "step": 21097
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6849807500839233,
      "learning_rate": 3.6523995984081695e-06,
      "loss": 0.3487,
      "step": 21098
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6180970668792725,
      "learning_rate": 3.6515368508309878e-06,
      "loss": 0.283,
      "step": 21099
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3578615188598633,
      "learning_rate": 3.6506741824024827e-06,
      "loss": 0.2803,
      "step": 21100
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8187729120254517,
      "learning_rate": 3.6498115931334045e-06,
      "loss": 0.3341,
      "step": 21101
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7483240365982056,
      "learning_rate": 3.6489490830345066e-06,
      "loss": 0.3304,
      "step": 21102
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.42421817779541,
      "learning_rate": 3.648086652116546e-06,
      "loss": 0.294,
      "step": 21103
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.645715594291687,
      "learning_rate": 3.6472243003902743e-06,
      "loss": 0.3268,
      "step": 21104
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.129115343093872,
      "learning_rate": 3.646362027866438e-06,
      "loss": 0.3463,
      "step": 21105
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8509373664855957,
      "learning_rate": 3.6454998345557948e-06,
      "loss": 0.2704,
      "step": 21106
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1309802532196045,
      "learning_rate": 3.6446377204690896e-06,
      "loss": 0.3671,
      "step": 21107
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.045780658721924,
      "learning_rate": 3.6437756856170723e-06,
      "loss": 0.3933,
      "step": 21108
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.660770297050476,
      "learning_rate": 3.6429137300104853e-06,
      "loss": 0.2437,
      "step": 21109
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.776024580001831,
      "learning_rate": 3.6420518536600826e-06,
      "loss": 0.3805,
      "step": 21110
    },
    {
      "epoch": 2.18,
      "grad_norm": 7.176438331604004,
      "learning_rate": 3.6411900565766057e-06,
      "loss": 0.5032,
      "step": 21111
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.841003656387329,
      "learning_rate": 3.6403283387707943e-06,
      "loss": 0.2757,
      "step": 21112
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7774600982666016,
      "learning_rate": 3.6394667002534013e-06,
      "loss": 0.3192,
      "step": 21113
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.1167314052581787,
      "learning_rate": 3.6386051410351643e-06,
      "loss": 0.3094,
      "step": 21114
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.0382683277130127,
      "learning_rate": 3.6377436611268214e-06,
      "loss": 0.3545,
      "step": 21115
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.145282745361328,
      "learning_rate": 3.6368822605391186e-06,
      "loss": 0.3207,
      "step": 21116
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9789369106292725,
      "learning_rate": 3.636020939282794e-06,
      "loss": 0.3274,
      "step": 21117
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8455743789672852,
      "learning_rate": 3.635159697368581e-06,
      "loss": 0.3914,
      "step": 21118
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.0122838020324707,
      "learning_rate": 3.6342985348072246e-06,
      "loss": 0.3375,
      "step": 21119
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9730101823806763,
      "learning_rate": 3.6334374516094582e-06,
      "loss": 0.4022,
      "step": 21120
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.626850962638855,
      "learning_rate": 3.632576447786016e-06,
      "loss": 0.3114,
      "step": 21121
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.934311866760254,
      "learning_rate": 3.63171552334763e-06,
      "loss": 0.4133,
      "step": 21122
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5508800745010376,
      "learning_rate": 3.6308546783050402e-06,
      "loss": 0.2379,
      "step": 21123
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.502462387084961,
      "learning_rate": 3.6299939126689755e-06,
      "loss": 0.2418,
      "step": 21124
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.962799072265625,
      "learning_rate": 3.6291332264501643e-06,
      "loss": 0.3399,
      "step": 21125
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9473768472671509,
      "learning_rate": 3.6282726196593444e-06,
      "loss": 0.3168,
      "step": 21126
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.2830967903137207,
      "learning_rate": 3.6274120923072396e-06,
      "loss": 0.3125,
      "step": 21127
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.006466865539551,
      "learning_rate": 3.6265516444045777e-06,
      "loss": 0.2641,
      "step": 21128
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9146859645843506,
      "learning_rate": 3.625691275962093e-06,
      "loss": 0.3701,
      "step": 21129
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8908536434173584,
      "learning_rate": 3.6248309869905063e-06,
      "loss": 0.2952,
      "step": 21130
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.75742769241333,
      "learning_rate": 3.6239707775005416e-06,
      "loss": 0.3143,
      "step": 21131
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.188926935195923,
      "learning_rate": 3.62311064750293e-06,
      "loss": 0.3335,
      "step": 21132
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7833280563354492,
      "learning_rate": 3.622250597008391e-06,
      "loss": 0.3039,
      "step": 21133
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7296830415725708,
      "learning_rate": 3.6213906260276487e-06,
      "loss": 0.3229,
      "step": 21134
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7546281814575195,
      "learning_rate": 3.6205307345714192e-06,
      "loss": 0.3232,
      "step": 21135
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0265157222747803,
      "learning_rate": 3.619670922650431e-06,
      "loss": 0.3225,
      "step": 21136
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.112516164779663,
      "learning_rate": 3.6188111902753998e-06,
      "loss": 0.3011,
      "step": 21137
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.373697519302368,
      "learning_rate": 3.617951537457042e-06,
      "loss": 0.3463,
      "step": 21138
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.15169620513916,
      "learning_rate": 3.6170919642060807e-06,
      "loss": 0.4317,
      "step": 21139
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6694341897964478,
      "learning_rate": 3.6162324705332297e-06,
      "loss": 0.3418,
      "step": 21140
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5467498302459717,
      "learning_rate": 3.615373056449201e-06,
      "loss": 0.2929,
      "step": 21141
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.022676944732666,
      "learning_rate": 3.6145137219647163e-06,
      "loss": 0.3397,
      "step": 21142
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.901826024055481,
      "learning_rate": 3.613654467090486e-06,
      "loss": 0.3571,
      "step": 21143
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8993889093399048,
      "learning_rate": 3.6127952918372187e-06,
      "loss": 0.4544,
      "step": 21144
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.515457034111023,
      "learning_rate": 3.611936196215633e-06,
      "loss": 0.2748,
      "step": 21145
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.791151762008667,
      "learning_rate": 3.6110771802364377e-06,
      "loss": 0.3033,
      "step": 21146
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0700769424438477,
      "learning_rate": 3.6102182439103393e-06,
      "loss": 0.3512,
      "step": 21147
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9631479978561401,
      "learning_rate": 3.609359387248047e-06,
      "loss": 0.3411,
      "step": 21148
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.82664155960083,
      "learning_rate": 3.6085006102602713e-06,
      "loss": 0.3312,
      "step": 21149
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.2229785919189453,
      "learning_rate": 3.607641912957719e-06,
      "loss": 0.3475,
      "step": 21150
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.2786810398101807,
      "learning_rate": 3.6067832953510905e-06,
      "loss": 0.341,
      "step": 21151
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.705666184425354,
      "learning_rate": 3.605924757451097e-06,
      "loss": 0.3469,
      "step": 21152
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.1695032119750977,
      "learning_rate": 3.6050662992684402e-06,
      "loss": 0.2572,
      "step": 21153
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.2403039932250977,
      "learning_rate": 3.604207920813818e-06,
      "loss": 0.2999,
      "step": 21154
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7298393249511719,
      "learning_rate": 3.6033496220979404e-06,
      "loss": 0.2901,
      "step": 21155
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9832850694656372,
      "learning_rate": 3.6024914031315028e-06,
      "loss": 0.3158,
      "step": 21156
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8444124460220337,
      "learning_rate": 3.601633263925204e-06,
      "loss": 0.306,
      "step": 21157
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9433503150939941,
      "learning_rate": 3.6007752044897483e-06,
      "loss": 0.3596,
      "step": 21158
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6115449666976929,
      "learning_rate": 3.5999172248358282e-06,
      "loss": 0.2009,
      "step": 21159
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5462669134140015,
      "learning_rate": 3.5990593249741436e-06,
      "loss": 0.267,
      "step": 21160
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.109854221343994,
      "learning_rate": 3.598201504915386e-06,
      "loss": 0.3216,
      "step": 21161
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6720384359359741,
      "learning_rate": 3.5973437646702557e-06,
      "loss": 0.3356,
      "step": 21162
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.823306918144226,
      "learning_rate": 3.5964861042494438e-06,
      "loss": 0.2777,
      "step": 21163
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7383240461349487,
      "learning_rate": 3.5956285236636403e-06,
      "loss": 0.2823,
      "step": 21164
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.4193062782287598,
      "learning_rate": 3.594771022923542e-06,
      "loss": 0.312,
      "step": 21165
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9272186756134033,
      "learning_rate": 3.593913602039838e-06,
      "loss": 0.3523,
      "step": 21166
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8748458623886108,
      "learning_rate": 3.5930562610232146e-06,
      "loss": 0.3534,
      "step": 21167
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5657507181167603,
      "learning_rate": 3.5921989998843663e-06,
      "loss": 0.2312,
      "step": 21168
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5873934030532837,
      "learning_rate": 3.591341818633979e-06,
      "loss": 0.2825,
      "step": 21169
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7638407945632935,
      "learning_rate": 3.590484717282735e-06,
      "loss": 0.299,
      "step": 21170
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7858178615570068,
      "learning_rate": 3.5896276958413266e-06,
      "loss": 0.3061,
      "step": 21171
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0645909309387207,
      "learning_rate": 3.5887707543204362e-06,
      "loss": 0.2477,
      "step": 21172
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8375320434570312,
      "learning_rate": 3.5879138927307476e-06,
      "loss": 0.3098,
      "step": 21173
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8463280200958252,
      "learning_rate": 3.587057111082939e-06,
      "loss": 0.292,
      "step": 21174
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.095384120941162,
      "learning_rate": 3.5862004093877003e-06,
      "loss": 0.4318,
      "step": 21175
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8396137952804565,
      "learning_rate": 3.5853437876557084e-06,
      "loss": 0.3262,
      "step": 21176
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.155668258666992,
      "learning_rate": 3.584487245897639e-06,
      "loss": 0.3604,
      "step": 21177
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9797611236572266,
      "learning_rate": 3.583630784124179e-06,
      "loss": 0.3875,
      "step": 21178
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.14876127243042,
      "learning_rate": 3.582774402346003e-06,
      "loss": 0.2883,
      "step": 21179
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8745054006576538,
      "learning_rate": 3.581918100573782e-06,
      "loss": 0.3211,
      "step": 21180
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.355149030685425,
      "learning_rate": 3.581061878818203e-06,
      "loss": 0.3276,
      "step": 21181
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.1552412509918213,
      "learning_rate": 3.5802057370899324e-06,
      "loss": 0.33,
      "step": 21182
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7733142375946045,
      "learning_rate": 3.5793496753996438e-06,
      "loss": 0.3464,
      "step": 21183
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7386033535003662,
      "learning_rate": 3.578493693758016e-06,
      "loss": 0.3307,
      "step": 21184
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.046902894973755,
      "learning_rate": 3.577637792175719e-06,
      "loss": 0.3973,
      "step": 21185
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8122302293777466,
      "learning_rate": 3.576781970663421e-06,
      "loss": 0.3201,
      "step": 21186
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9683289527893066,
      "learning_rate": 3.5759262292317886e-06,
      "loss": 0.4191,
      "step": 21187
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8607550859451294,
      "learning_rate": 3.5750705678915e-06,
      "loss": 0.2689,
      "step": 21188
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.11314058303833,
      "learning_rate": 3.5742149866532182e-06,
      "loss": 0.3452,
      "step": 21189
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.410377025604248,
      "learning_rate": 3.5733594855276057e-06,
      "loss": 0.3683,
      "step": 21190
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9121153354644775,
      "learning_rate": 3.572504064525336e-06,
      "loss": 0.3427,
      "step": 21191
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7304867506027222,
      "learning_rate": 3.571648723657072e-06,
      "loss": 0.2772,
      "step": 21192
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0537796020507812,
      "learning_rate": 3.5707934629334705e-06,
      "loss": 0.2893,
      "step": 21193
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6922870874404907,
      "learning_rate": 3.569938282365205e-06,
      "loss": 0.2631,
      "step": 21194
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.77425479888916,
      "learning_rate": 3.569083181962932e-06,
      "loss": 0.3351,
      "step": 21195
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.132267475128174,
      "learning_rate": 3.5682281617373095e-06,
      "loss": 0.2731,
      "step": 21196
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8546737432479858,
      "learning_rate": 3.5673732216990043e-06,
      "loss": 0.3364,
      "step": 21197
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8214298486709595,
      "learning_rate": 3.566518361858671e-06,
      "loss": 0.3947,
      "step": 21198
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0043232440948486,
      "learning_rate": 3.5656635822269682e-06,
      "loss": 0.3377,
      "step": 21199
    },
    {
      "epoch": 2.19,
      "grad_norm": 3.650667428970337,
      "learning_rate": 3.5648088828145488e-06,
      "loss": 0.374,
      "step": 21200
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.763114333152771,
      "learning_rate": 3.5639542636320766e-06,
      "loss": 0.3021,
      "step": 21201
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8636515140533447,
      "learning_rate": 3.563099724690201e-06,
      "loss": 0.3486,
      "step": 21202
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8205941915512085,
      "learning_rate": 3.5622452659995744e-06,
      "loss": 0.3416,
      "step": 21203
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6025724411010742,
      "learning_rate": 3.5613908875708557e-06,
      "loss": 0.277,
      "step": 21204
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.290804624557495,
      "learning_rate": 3.5605365894146947e-06,
      "loss": 0.3555,
      "step": 21205
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6362637281417847,
      "learning_rate": 3.559682371541736e-06,
      "loss": 0.327,
      "step": 21206
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6653891801834106,
      "learning_rate": 3.558828233962639e-06,
      "loss": 0.2828,
      "step": 21207
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8130582571029663,
      "learning_rate": 3.557974176688047e-06,
      "loss": 0.2842,
      "step": 21208
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8437514305114746,
      "learning_rate": 3.5571201997286054e-06,
      "loss": 0.3103,
      "step": 21209
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0876922607421875,
      "learning_rate": 3.5562663030949685e-06,
      "loss": 0.2978,
      "step": 21210
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8517658710479736,
      "learning_rate": 3.555412486797778e-06,
      "loss": 0.3773,
      "step": 21211
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.7565299272537231,
      "learning_rate": 3.55455875084768e-06,
      "loss": 0.3158,
      "step": 21212
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.620108962059021,
      "learning_rate": 3.553705095255312e-06,
      "loss": 0.3365,
      "step": 21213
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.975354790687561,
      "learning_rate": 3.5528515200313265e-06,
      "loss": 0.3559,
      "step": 21214
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0240912437438965,
      "learning_rate": 3.5519980251863608e-06,
      "loss": 0.3085,
      "step": 21215
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4996159076690674,
      "learning_rate": 3.551144610731052e-06,
      "loss": 0.2057,
      "step": 21216
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8404414653778076,
      "learning_rate": 3.550291276676049e-06,
      "loss": 0.2572,
      "step": 21217
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9616012573242188,
      "learning_rate": 3.5494380230319838e-06,
      "loss": 0.2802,
      "step": 21218
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.9671056270599365,
      "learning_rate": 3.5485848498094922e-06,
      "loss": 0.3754,
      "step": 21219
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.0354528427124023,
      "learning_rate": 3.5477317570192194e-06,
      "loss": 0.3648,
      "step": 21220
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.020094394683838,
      "learning_rate": 3.5468787446717957e-06,
      "loss": 0.3008,
      "step": 21221
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6123803853988647,
      "learning_rate": 3.5460258127778537e-06,
      "loss": 0.2596,
      "step": 21222
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.2430198192596436,
      "learning_rate": 3.5451729613480336e-06,
      "loss": 0.3417,
      "step": 21223
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9492731094360352,
      "learning_rate": 3.5443201903929647e-06,
      "loss": 0.2882,
      "step": 21224
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7913103103637695,
      "learning_rate": 3.54346749992328e-06,
      "loss": 0.2557,
      "step": 21225
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.0326521396636963,
      "learning_rate": 3.5426148899496047e-06,
      "loss": 0.1756,
      "step": 21226
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.814437747001648,
      "learning_rate": 3.541762360482578e-06,
      "loss": 0.3356,
      "step": 21227
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.539818525314331,
      "learning_rate": 3.5409099115328226e-06,
      "loss": 0.2905,
      "step": 21228
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.094618558883667,
      "learning_rate": 3.540057543110965e-06,
      "loss": 0.2526,
      "step": 21229
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8470731973648071,
      "learning_rate": 3.539205255227638e-06,
      "loss": 0.2667,
      "step": 21230
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7643463611602783,
      "learning_rate": 3.5383530478934634e-06,
      "loss": 0.2587,
      "step": 21231
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.054307222366333,
      "learning_rate": 3.5375009211190635e-06,
      "loss": 0.3408,
      "step": 21232
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5695531368255615,
      "learning_rate": 3.5366488749150694e-06,
      "loss": 0.2683,
      "step": 21233
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6275851726531982,
      "learning_rate": 3.5357969092920995e-06,
      "loss": 0.3022,
      "step": 21234
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.531391978263855,
      "learning_rate": 3.534945024260772e-06,
      "loss": 0.3392,
      "step": 21235
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8109440803527832,
      "learning_rate": 3.5340932198317146e-06,
      "loss": 0.3013,
      "step": 21236
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9178470373153687,
      "learning_rate": 3.533241496015545e-06,
      "loss": 0.2507,
      "step": 21237
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.6288185119628906,
      "learning_rate": 3.5323898528228797e-06,
      "loss": 0.2754,
      "step": 21238
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1207077503204346,
      "learning_rate": 3.5315382902643346e-06,
      "loss": 0.3249,
      "step": 21239
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8236560821533203,
      "learning_rate": 3.530686808350533e-06,
      "loss": 0.2972,
      "step": 21240
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5833837985992432,
      "learning_rate": 3.529835407092087e-06,
      "loss": 0.3259,
      "step": 21241
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9528918266296387,
      "learning_rate": 3.5289840864996085e-06,
      "loss": 0.3833,
      "step": 21242
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.006108522415161,
      "learning_rate": 3.528132846583717e-06,
      "loss": 0.3212,
      "step": 21243
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.812593698501587,
      "learning_rate": 3.527281687355023e-06,
      "loss": 0.3281,
      "step": 21244
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.6676154136657715,
      "learning_rate": 3.526430608824134e-06,
      "loss": 0.3082,
      "step": 21245
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.3348960876464844,
      "learning_rate": 3.5255796110016682e-06,
      "loss": 0.3886,
      "step": 21246
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7937949895858765,
      "learning_rate": 3.5247286938982307e-06,
      "loss": 0.3058,
      "step": 21247
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.822715163230896,
      "learning_rate": 3.523877857524428e-06,
      "loss": 0.2466,
      "step": 21248
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.107801914215088,
      "learning_rate": 3.5230271018908736e-06,
      "loss": 0.3923,
      "step": 21249
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.41803240776062,
      "learning_rate": 3.5221764270081714e-06,
      "loss": 0.3775,
      "step": 21250
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.040194511413574,
      "learning_rate": 3.5213258328869284e-06,
      "loss": 0.2557,
      "step": 21251
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.885129690170288,
      "learning_rate": 3.5204753195377426e-06,
      "loss": 0.3645,
      "step": 21252
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.0655343532562256,
      "learning_rate": 3.519624886971228e-06,
      "loss": 0.2915,
      "step": 21253
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.340522050857544,
      "learning_rate": 3.518774535197982e-06,
      "loss": 0.2716,
      "step": 21254
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.080246686935425,
      "learning_rate": 3.5179242642286025e-06,
      "loss": 0.3074,
      "step": 21255
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.35101580619812,
      "learning_rate": 3.5170740740736973e-06,
      "loss": 0.4088,
      "step": 21256
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.8363335132598877,
      "learning_rate": 3.5162239647438635e-06,
      "loss": 0.3506,
      "step": 21257
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.2646260261535645,
      "learning_rate": 3.515373936249696e-06,
      "loss": 0.2783,
      "step": 21258
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6698740720748901,
      "learning_rate": 3.514523988601799e-06,
      "loss": 0.3323,
      "step": 21259
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8934406042099,
      "learning_rate": 3.513674121810765e-06,
      "loss": 0.3322,
      "step": 21260
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7183924913406372,
      "learning_rate": 3.5128243358871873e-06,
      "loss": 0.3128,
      "step": 21261
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.96479070186615,
      "learning_rate": 3.511974630841666e-06,
      "loss": 0.2497,
      "step": 21262
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.049118995666504,
      "learning_rate": 3.5111250066847935e-06,
      "loss": 0.3034,
      "step": 21263
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.163210153579712,
      "learning_rate": 3.5102754634271597e-06,
      "loss": 0.3225,
      "step": 21264
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7247729301452637,
      "learning_rate": 3.5094260010793556e-06,
      "loss": 0.3359,
      "step": 21265
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.2281553745269775,
      "learning_rate": 3.5085766196519754e-06,
      "loss": 0.3852,
      "step": 21266
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6801362037658691,
      "learning_rate": 3.5077273191556084e-06,
      "loss": 0.3054,
      "step": 21267
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8391590118408203,
      "learning_rate": 3.506878099600838e-06,
      "loss": 0.3447,
      "step": 21268
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8140112161636353,
      "learning_rate": 3.506028960998259e-06,
      "loss": 0.3541,
      "step": 21269
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.752677321434021,
      "learning_rate": 3.5051799033584533e-06,
      "loss": 0.2821,
      "step": 21270
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7304433584213257,
      "learning_rate": 3.5043309266920045e-06,
      "loss": 0.3699,
      "step": 21271
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.2917985916137695,
      "learning_rate": 3.5034820310095043e-06,
      "loss": 0.3691,
      "step": 21272
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1444034576416016,
      "learning_rate": 3.5026332163215315e-06,
      "loss": 0.2518,
      "step": 21273
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6526463031768799,
      "learning_rate": 3.501784482638666e-06,
      "loss": 0.3153,
      "step": 21274
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8024096488952637,
      "learning_rate": 3.5009358299714948e-06,
      "loss": 0.3958,
      "step": 21275
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8734792470932007,
      "learning_rate": 3.500087258330597e-06,
      "loss": 0.3137,
      "step": 21276
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.204838752746582,
      "learning_rate": 3.4992387677265506e-06,
      "loss": 0.3326,
      "step": 21277
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9183748960494995,
      "learning_rate": 3.4983903581699307e-06,
      "loss": 0.2525,
      "step": 21278
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8476520776748657,
      "learning_rate": 3.4975420296713226e-06,
      "loss": 0.3742,
      "step": 21279
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8377751111984253,
      "learning_rate": 3.496693782241298e-06,
      "loss": 0.3261,
      "step": 21280
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7590160369873047,
      "learning_rate": 3.49584561589043e-06,
      "loss": 0.3187,
      "step": 21281
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.828212022781372,
      "learning_rate": 3.4949975306292995e-06,
      "loss": 0.3052,
      "step": 21282
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1060540676116943,
      "learning_rate": 3.4941495264684753e-06,
      "loss": 0.3206,
      "step": 21283
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9827507734298706,
      "learning_rate": 3.4933016034185284e-06,
      "loss": 0.32,
      "step": 21284
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7016135454177856,
      "learning_rate": 3.4924537614900356e-06,
      "loss": 0.2783,
      "step": 21285
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9824315309524536,
      "learning_rate": 3.491606000693565e-06,
      "loss": 0.2936,
      "step": 21286
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.189899444580078,
      "learning_rate": 3.490758321039681e-06,
      "loss": 0.3169,
      "step": 21287
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9105342626571655,
      "learning_rate": 3.4899107225389594e-06,
      "loss": 0.327,
      "step": 21288
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6851414442062378,
      "learning_rate": 3.4890632052019644e-06,
      "loss": 0.3204,
      "step": 21289
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7112853527069092,
      "learning_rate": 3.4882157690392628e-06,
      "loss": 0.3331,
      "step": 21290
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6898354291915894,
      "learning_rate": 3.487368414061416e-06,
      "loss": 0.3572,
      "step": 21291
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.833715558052063,
      "learning_rate": 3.4865211402789945e-06,
      "loss": 0.313,
      "step": 21292
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9402598142623901,
      "learning_rate": 3.485673947702559e-06,
      "loss": 0.3783,
      "step": 21293
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9202271699905396,
      "learning_rate": 3.484826836342667e-06,
      "loss": 0.2393,
      "step": 21294
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.53300940990448,
      "learning_rate": 3.4839798062098895e-06,
      "loss": 0.3006,
      "step": 21295
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6904839277267456,
      "learning_rate": 3.4831328573147805e-06,
      "loss": 0.3063,
      "step": 21296
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1667582988739014,
      "learning_rate": 3.482285989667896e-06,
      "loss": 0.283,
      "step": 21297
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9341881275177002,
      "learning_rate": 3.4814392032798028e-06,
      "loss": 0.3225,
      "step": 21298
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1392290592193604,
      "learning_rate": 3.480592498161054e-06,
      "loss": 0.287,
      "step": 21299
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8356897830963135,
      "learning_rate": 3.4797458743222013e-06,
      "loss": 0.2942,
      "step": 21300
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.008244514465332,
      "learning_rate": 3.4788993317738074e-06,
      "loss": 0.274,
      "step": 21301
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8305668830871582,
      "learning_rate": 3.478052870526424e-06,
      "loss": 0.3705,
      "step": 21302
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9090834856033325,
      "learning_rate": 3.477206490590602e-06,
      "loss": 0.3186,
      "step": 21303
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.067558526992798,
      "learning_rate": 3.476360191976893e-06,
      "loss": 0.3476,
      "step": 21304
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.40022873878479,
      "learning_rate": 3.4755139746958533e-06,
      "loss": 0.3214,
      "step": 21305
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5275567770004272,
      "learning_rate": 3.4746678387580292e-06,
      "loss": 0.2391,
      "step": 21306
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9278355836868286,
      "learning_rate": 3.4738217841739665e-06,
      "loss": 0.3214,
      "step": 21307
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9167276620864868,
      "learning_rate": 3.472975810954221e-06,
      "loss": 0.3579,
      "step": 21308
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.106019973754883,
      "learning_rate": 3.4721299191093363e-06,
      "loss": 0.2698,
      "step": 21309
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.2699272632598877,
      "learning_rate": 3.471284108649854e-06,
      "loss": 0.4012,
      "step": 21310
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.9742794036865234,
      "learning_rate": 3.470438379586327e-06,
      "loss": 0.263,
      "step": 21311
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.646592378616333,
      "learning_rate": 3.4695927319292953e-06,
      "loss": 0.2265,
      "step": 21312
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.0669608116149902,
      "learning_rate": 3.468747165689299e-06,
      "loss": 0.3665,
      "step": 21313
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5321999788284302,
      "learning_rate": 3.4679016808768862e-06,
      "loss": 0.3131,
      "step": 21314
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.3312389850616455,
      "learning_rate": 3.467056277502595e-06,
      "loss": 0.3871,
      "step": 21315
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.0820417404174805,
      "learning_rate": 3.4662109555769654e-06,
      "loss": 0.2979,
      "step": 21316
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.806349754333496,
      "learning_rate": 3.4653657151105324e-06,
      "loss": 0.2632,
      "step": 21317
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.1829466819763184,
      "learning_rate": 3.4645205561138417e-06,
      "loss": 0.3147,
      "step": 21318
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.212656259536743,
      "learning_rate": 3.4636754785974256e-06,
      "loss": 0.293,
      "step": 21319
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.004765748977661,
      "learning_rate": 3.4628304825718183e-06,
      "loss": 0.3124,
      "step": 21320
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.033653736114502,
      "learning_rate": 3.46198556804756e-06,
      "loss": 0.3435,
      "step": 21321
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.134077548980713,
      "learning_rate": 3.461140735035181e-06,
      "loss": 0.2862,
      "step": 21322
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6893585920333862,
      "learning_rate": 3.460295983545211e-06,
      "loss": 0.3212,
      "step": 21323
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6649399995803833,
      "learning_rate": 3.4594513135881902e-06,
      "loss": 0.2429,
      "step": 21324
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8647080659866333,
      "learning_rate": 3.4586067251746437e-06,
      "loss": 0.3327,
      "step": 21325
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6451284885406494,
      "learning_rate": 3.4577622183150993e-06,
      "loss": 0.275,
      "step": 21326
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.640828847885132,
      "learning_rate": 3.4569177930200924e-06,
      "loss": 0.3966,
      "step": 21327
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.658810019493103,
      "learning_rate": 3.456073449300147e-06,
      "loss": 0.2547,
      "step": 21328
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9445379972457886,
      "learning_rate": 3.4552291871657907e-06,
      "loss": 0.2913,
      "step": 21329
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8371106386184692,
      "learning_rate": 3.4543850066275455e-06,
      "loss": 0.2272,
      "step": 21330
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9013712406158447,
      "learning_rate": 3.4535409076959412e-06,
      "loss": 0.329,
      "step": 21331
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7502225637435913,
      "learning_rate": 3.4526968903815006e-06,
      "loss": 0.2769,
      "step": 21332
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.819288730621338,
      "learning_rate": 3.451852954694741e-06,
      "loss": 0.2478,
      "step": 21333
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9935688972473145,
      "learning_rate": 3.451009100646192e-06,
      "loss": 0.2926,
      "step": 21334
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.62534499168396,
      "learning_rate": 3.4501653282463723e-06,
      "loss": 0.3032,
      "step": 21335
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8718732595443726,
      "learning_rate": 3.449321637505795e-06,
      "loss": 0.2962,
      "step": 21336
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9209955930709839,
      "learning_rate": 3.448478028434986e-06,
      "loss": 0.3498,
      "step": 21337
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.76715087890625,
      "learning_rate": 3.4476345010444624e-06,
      "loss": 0.3241,
      "step": 21338
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9457985162734985,
      "learning_rate": 3.4467910553447335e-06,
      "loss": 0.3262,
      "step": 21339
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.2084860801696777,
      "learning_rate": 3.4459476913463253e-06,
      "loss": 0.3369,
      "step": 21340
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.2659738063812256,
      "learning_rate": 3.4451044090597464e-06,
      "loss": 0.3025,
      "step": 21341
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9663273096084595,
      "learning_rate": 3.4442612084955106e-06,
      "loss": 0.2599,
      "step": 21342
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9693129062652588,
      "learning_rate": 3.443418089664129e-06,
      "loss": 0.3139,
      "step": 21343
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.0786399841308594,
      "learning_rate": 3.442575052576117e-06,
      "loss": 0.2905,
      "step": 21344
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8717001676559448,
      "learning_rate": 3.441732097241983e-06,
      "loss": 0.3225,
      "step": 21345
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5859599113464355,
      "learning_rate": 3.4408892236722345e-06,
      "loss": 0.268,
      "step": 21346
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.063771963119507,
      "learning_rate": 3.440046431877384e-06,
      "loss": 0.3271,
      "step": 21347
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.109497308731079,
      "learning_rate": 3.4392037218679376e-06,
      "loss": 0.2738,
      "step": 21348
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7915334701538086,
      "learning_rate": 3.4383610936543975e-06,
      "loss": 0.3487,
      "step": 21349
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8303037881851196,
      "learning_rate": 3.437518547247275e-06,
      "loss": 0.3756,
      "step": 21350
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6122543811798096,
      "learning_rate": 3.4366760826570734e-06,
      "loss": 0.3943,
      "step": 21351
    },
    {
      "epoch": 2.21,
      "grad_norm": 3.2830922603607178,
      "learning_rate": 3.43583369989429e-06,
      "loss": 0.3323,
      "step": 21352
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7412298917770386,
      "learning_rate": 3.434991398969435e-06,
      "loss": 0.3625,
      "step": 21353
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.07376766204834,
      "learning_rate": 3.434149179893007e-06,
      "loss": 0.3927,
      "step": 21354
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.866274356842041,
      "learning_rate": 3.4333070426755043e-06,
      "loss": 0.3867,
      "step": 21355
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7879563570022583,
      "learning_rate": 3.432464987327425e-06,
      "loss": 0.3088,
      "step": 21356
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.055577278137207,
      "learning_rate": 3.4316230138592733e-06,
      "loss": 0.2967,
      "step": 21357
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.0159871578216553,
      "learning_rate": 3.4307811222815425e-06,
      "loss": 0.2399,
      "step": 21358
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.1127572059631348,
      "learning_rate": 3.4299393126047263e-06,
      "loss": 0.3685,
      "step": 21359
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.927379846572876,
      "learning_rate": 3.429097584839325e-06,
      "loss": 0.3189,
      "step": 21360
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.1274359226226807,
      "learning_rate": 3.4282559389958315e-06,
      "loss": 0.3356,
      "step": 21361
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.556122064590454,
      "learning_rate": 3.427414375084733e-06,
      "loss": 0.2864,
      "step": 21362
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.154219388961792,
      "learning_rate": 3.426572893116531e-06,
      "loss": 0.375,
      "step": 21363
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.849134087562561,
      "learning_rate": 3.4257314931017116e-06,
      "loss": 0.3032,
      "step": 21364
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9729670286178589,
      "learning_rate": 3.4248901750507623e-06,
      "loss": 0.336,
      "step": 21365
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8526993989944458,
      "learning_rate": 3.424048938974178e-06,
      "loss": 0.3535,
      "step": 21366
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8177413940429688,
      "learning_rate": 3.423207784882444e-06,
      "loss": 0.3222,
      "step": 21367
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8154895305633545,
      "learning_rate": 3.422366712786047e-06,
      "loss": 0.2693,
      "step": 21368
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.113185405731201,
      "learning_rate": 3.4215257226954692e-06,
      "loss": 0.3464,
      "step": 21369
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.802639365196228,
      "learning_rate": 3.4206848146212037e-06,
      "loss": 0.2845,
      "step": 21370
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9506582021713257,
      "learning_rate": 3.419843988573729e-06,
      "loss": 0.3623,
      "step": 21371
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7096562385559082,
      "learning_rate": 3.4190032445635256e-06,
      "loss": 0.2582,
      "step": 21372
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9535261392593384,
      "learning_rate": 3.4181625826010835e-06,
      "loss": 0.3294,
      "step": 21373
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7681045532226562,
      "learning_rate": 3.417322002696878e-06,
      "loss": 0.3852,
      "step": 21374
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7932466268539429,
      "learning_rate": 3.4164815048613864e-06,
      "loss": 0.2637,
      "step": 21375
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8929814100265503,
      "learning_rate": 3.415641089105095e-06,
      "loss": 0.2725,
      "step": 21376
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.727314829826355,
      "learning_rate": 3.4148007554384765e-06,
      "loss": 0.2597,
      "step": 21377
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.897506833076477,
      "learning_rate": 3.413960503872006e-06,
      "loss": 0.3433,
      "step": 21378
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.098569869995117,
      "learning_rate": 3.4131203344161645e-06,
      "loss": 0.3646,
      "step": 21379
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.1693992614746094,
      "learning_rate": 3.4122802470814252e-06,
      "loss": 0.3493,
      "step": 21380
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.1110880374908447,
      "learning_rate": 3.41144024187826e-06,
      "loss": 0.3702,
      "step": 21381
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9516539573669434,
      "learning_rate": 3.4106003188171387e-06,
      "loss": 0.3077,
      "step": 21382
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.3029842376708984,
      "learning_rate": 3.4097604779085403e-06,
      "loss": 0.3613,
      "step": 21383
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7508643865585327,
      "learning_rate": 3.4089207191629314e-06,
      "loss": 0.3189,
      "step": 21384
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.710710883140564,
      "learning_rate": 3.4080810425907785e-06,
      "loss": 0.2499,
      "step": 21385
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7785104513168335,
      "learning_rate": 3.4072414482025563e-06,
      "loss": 0.3056,
      "step": 21386
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.018763542175293,
      "learning_rate": 3.40640193600873e-06,
      "loss": 0.3076,
      "step": 21387
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9128549098968506,
      "learning_rate": 3.4055625060197615e-06,
      "loss": 0.3088,
      "step": 21388
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9054839611053467,
      "learning_rate": 3.4047231582461238e-06,
      "loss": 0.3477,
      "step": 21389
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.844428300857544,
      "learning_rate": 3.403883892698279e-06,
      "loss": 0.3204,
      "step": 21390
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.0828230381011963,
      "learning_rate": 3.4030447093866847e-06,
      "loss": 0.3125,
      "step": 21391
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.414330244064331,
      "learning_rate": 3.4022056083218126e-06,
      "loss": 0.3678,
      "step": 21392
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5821079015731812,
      "learning_rate": 3.401366589514119e-06,
      "loss": 0.2554,
      "step": 21393
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6644692420959473,
      "learning_rate": 3.400527652974065e-06,
      "loss": 0.3025,
      "step": 21394
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8258723020553589,
      "learning_rate": 3.399688798712106e-06,
      "loss": 0.3462,
      "step": 21395
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.865308165550232,
      "learning_rate": 3.398850026738708e-06,
      "loss": 0.2721,
      "step": 21396
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.2100675106048584,
      "learning_rate": 3.3980113370643252e-06,
      "loss": 0.3504,
      "step": 21397
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.0224945545196533,
      "learning_rate": 3.3971727296994086e-06,
      "loss": 0.3553,
      "step": 21398
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.646183967590332,
      "learning_rate": 3.396334204654421e-06,
      "loss": 0.2975,
      "step": 21399
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.99757981300354,
      "learning_rate": 3.3954957619398145e-06,
      "loss": 0.3689,
      "step": 21400
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.0203776359558105,
      "learning_rate": 3.3946574015660372e-06,
      "loss": 0.3903,
      "step": 21401
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.021028518676758,
      "learning_rate": 3.393819123543548e-06,
      "loss": 0.3236,
      "step": 21402
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.1483662128448486,
      "learning_rate": 3.3929809278827963e-06,
      "loss": 0.3617,
      "step": 21403
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.2461092472076416,
      "learning_rate": 3.392142814594227e-06,
      "loss": 0.3577,
      "step": 21404
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.3015201091766357,
      "learning_rate": 3.3913047836882974e-06,
      "loss": 0.3228,
      "step": 21405
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6539826393127441,
      "learning_rate": 3.3904668351754523e-06,
      "loss": 0.3484,
      "step": 21406
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9557172060012817,
      "learning_rate": 3.3896289690661364e-06,
      "loss": 0.299,
      "step": 21407
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8670451641082764,
      "learning_rate": 3.3887911853707943e-06,
      "loss": 0.2371,
      "step": 21408
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.7334368228912354,
      "learning_rate": 3.3879534840998773e-06,
      "loss": 0.3738,
      "step": 21409
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6303561925888062,
      "learning_rate": 3.387115865263827e-06,
      "loss": 0.243,
      "step": 21410
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.643243670463562,
      "learning_rate": 3.3862783288730803e-06,
      "loss": 0.3035,
      "step": 21411
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8282781839370728,
      "learning_rate": 3.3854408749380886e-06,
      "loss": 0.3414,
      "step": 21412
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.5330376625061035,
      "learning_rate": 3.3846035034692883e-06,
      "loss": 0.2783,
      "step": 21413
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.957282543182373,
      "learning_rate": 3.3837662144771156e-06,
      "loss": 0.3817,
      "step": 21414
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4694037437438965,
      "learning_rate": 3.3829290079720158e-06,
      "loss": 0.3225,
      "step": 21415
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.029094696044922,
      "learning_rate": 3.3820918839644233e-06,
      "loss": 0.2735,
      "step": 21416
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6285297870635986,
      "learning_rate": 3.381254842464773e-06,
      "loss": 0.3307,
      "step": 21417
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0547127723693848,
      "learning_rate": 3.3804178834835042e-06,
      "loss": 0.3238,
      "step": 21418
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.583086371421814,
      "learning_rate": 3.379581007031052e-06,
      "loss": 0.2853,
      "step": 21419
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7544323205947876,
      "learning_rate": 3.3787442131178482e-06,
      "loss": 0.2732,
      "step": 21420
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7211419343948364,
      "learning_rate": 3.3779075017543217e-06,
      "loss": 0.3011,
      "step": 21421
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.64695143699646,
      "learning_rate": 3.377070872950912e-06,
      "loss": 0.234,
      "step": 21422
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9107155799865723,
      "learning_rate": 3.376234326718044e-06,
      "loss": 0.3489,
      "step": 21423
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9368704557418823,
      "learning_rate": 3.3753978630661464e-06,
      "loss": 0.3688,
      "step": 21424
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7573322057724,
      "learning_rate": 3.3745614820056528e-06,
      "loss": 0.3616,
      "step": 21425
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7045186758041382,
      "learning_rate": 3.3737251835469884e-06,
      "loss": 0.3725,
      "step": 21426
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7373027801513672,
      "learning_rate": 3.372888967700576e-06,
      "loss": 0.3623,
      "step": 21427
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0865399837493896,
      "learning_rate": 3.372052834476848e-06,
      "loss": 0.3356,
      "step": 21428
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.4667603969573975,
      "learning_rate": 3.3712167838862243e-06,
      "loss": 0.2763,
      "step": 21429
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8421145677566528,
      "learning_rate": 3.370380815939126e-06,
      "loss": 0.3904,
      "step": 21430
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.120022773742676,
      "learning_rate": 3.3695449306459814e-06,
      "loss": 0.3169,
      "step": 21431
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0434274673461914,
      "learning_rate": 3.368709128017209e-06,
      "loss": 0.4211,
      "step": 21432
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.1543049812316895,
      "learning_rate": 3.367873408063229e-06,
      "loss": 0.3176,
      "step": 21433
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5872384309768677,
      "learning_rate": 3.3670377707944577e-06,
      "loss": 0.3463,
      "step": 21434
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6987649202346802,
      "learning_rate": 3.3662022162213183e-06,
      "loss": 0.2797,
      "step": 21435
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5469955205917358,
      "learning_rate": 3.365366744354227e-06,
      "loss": 0.3343,
      "step": 21436
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8898038864135742,
      "learning_rate": 3.364531355203594e-06,
      "loss": 0.3126,
      "step": 21437
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0292177200317383,
      "learning_rate": 3.363696048779843e-06,
      "loss": 0.3813,
      "step": 21438
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4772089719772339,
      "learning_rate": 3.3628608250933846e-06,
      "loss": 0.2425,
      "step": 21439
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7982591390609741,
      "learning_rate": 3.3620256841546286e-06,
      "loss": 0.2669,
      "step": 21440
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.14046049118042,
      "learning_rate": 3.3611906259739925e-06,
      "loss": 0.4083,
      "step": 21441
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.1003777980804443,
      "learning_rate": 3.360355650561884e-06,
      "loss": 0.3313,
      "step": 21442
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.1817150115966797,
      "learning_rate": 3.3595207579287115e-06,
      "loss": 0.3657,
      "step": 21443
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.017024517059326,
      "learning_rate": 3.3586859480848888e-06,
      "loss": 0.2956,
      "step": 21444
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.80078125,
      "learning_rate": 3.357851221040821e-06,
      "loss": 0.3201,
      "step": 21445
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.441122531890869,
      "learning_rate": 3.357016576806915e-06,
      "loss": 0.3955,
      "step": 21446
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.3549773693084717,
      "learning_rate": 3.3561820153935775e-06,
      "loss": 0.4001,
      "step": 21447
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0311737060546875,
      "learning_rate": 3.355347536811212e-06,
      "loss": 0.293,
      "step": 21448
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.2061831951141357,
      "learning_rate": 3.3545131410702194e-06,
      "loss": 0.3732,
      "step": 21449
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6687729358673096,
      "learning_rate": 3.3536788281810084e-06,
      "loss": 0.2842,
      "step": 21450
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9382147789001465,
      "learning_rate": 3.3528445981539793e-06,
      "loss": 0.3236,
      "step": 21451
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8147116899490356,
      "learning_rate": 3.3520104509995275e-06,
      "loss": 0.2709,
      "step": 21452
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8659650087356567,
      "learning_rate": 3.35117638672806e-06,
      "loss": 0.2915,
      "step": 21453
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7512568235397339,
      "learning_rate": 3.3503424053499723e-06,
      "loss": 0.2155,
      "step": 21454
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.448274850845337,
      "learning_rate": 3.349508506875662e-06,
      "loss": 0.3109,
      "step": 21455
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.841231107711792,
      "learning_rate": 3.348674691315521e-06,
      "loss": 0.3345,
      "step": 21456
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7831743955612183,
      "learning_rate": 3.3478409586799532e-06,
      "loss": 0.3734,
      "step": 21457
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6016511917114258,
      "learning_rate": 3.3470073089793485e-06,
      "loss": 0.3225,
      "step": 21458
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.972355842590332,
      "learning_rate": 3.346173742224098e-06,
      "loss": 0.3446,
      "step": 21459
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0915420055389404,
      "learning_rate": 3.3453402584246e-06,
      "loss": 0.3434,
      "step": 21460
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8064390420913696,
      "learning_rate": 3.3445068575912422e-06,
      "loss": 0.3442,
      "step": 21461
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.193572759628296,
      "learning_rate": 3.3436735397344124e-06,
      "loss": 0.3892,
      "step": 21462
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.2323944568634033,
      "learning_rate": 3.3428403048645065e-06,
      "loss": 0.3102,
      "step": 21463
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0954272747039795,
      "learning_rate": 3.3420071529919095e-06,
      "loss": 0.279,
      "step": 21464
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7715933322906494,
      "learning_rate": 3.3411740841270044e-06,
      "loss": 0.3747,
      "step": 21465
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9369750022888184,
      "learning_rate": 3.3403410982801843e-06,
      "loss": 0.2848,
      "step": 21466
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9782170057296753,
      "learning_rate": 3.3395081954618326e-06,
      "loss": 0.2816,
      "step": 21467
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.196545124053955,
      "learning_rate": 3.338675375682331e-06,
      "loss": 0.3596,
      "step": 21468
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.38750958442688,
      "learning_rate": 3.337842638952061e-06,
      "loss": 0.3489,
      "step": 21469
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7151597738265991,
      "learning_rate": 3.33700998528141e-06,
      "loss": 0.3047,
      "step": 21470
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9573707580566406,
      "learning_rate": 3.336177414680757e-06,
      "loss": 0.3007,
      "step": 21471
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.924456238746643,
      "learning_rate": 3.3353449271604777e-06,
      "loss": 0.3046,
      "step": 21472
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6967096328735352,
      "learning_rate": 3.3345125227309584e-06,
      "loss": 0.3308,
      "step": 21473
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.1969656944274902,
      "learning_rate": 3.3336802014025728e-06,
      "loss": 0.3221,
      "step": 21474
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5672141313552856,
      "learning_rate": 3.332847963185696e-06,
      "loss": 0.2567,
      "step": 21475
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.3699123859405518,
      "learning_rate": 3.3320158080907085e-06,
      "loss": 0.3941,
      "step": 21476
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9094951152801514,
      "learning_rate": 3.331183736127982e-06,
      "loss": 0.3001,
      "step": 21477
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.663468360900879,
      "learning_rate": 3.3303517473078883e-06,
      "loss": 0.317,
      "step": 21478
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.719746708869934,
      "learning_rate": 3.329519841640807e-06,
      "loss": 0.2579,
      "step": 21479
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.307548761367798,
      "learning_rate": 3.328688019137104e-06,
      "loss": 0.3116,
      "step": 21480
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8832206726074219,
      "learning_rate": 3.3278562798071524e-06,
      "loss": 0.2174,
      "step": 21481
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8475563526153564,
      "learning_rate": 3.327024623661317e-06,
      "loss": 0.3306,
      "step": 21482
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8383923768997192,
      "learning_rate": 3.3261930507099737e-06,
      "loss": 0.2908,
      "step": 21483
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.014336109161377,
      "learning_rate": 3.325361560963487e-06,
      "loss": 0.3176,
      "step": 21484
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4542603492736816,
      "learning_rate": 3.3245301544322185e-06,
      "loss": 0.2554,
      "step": 21485
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.017418384552002,
      "learning_rate": 3.3236988311265426e-06,
      "loss": 0.3606,
      "step": 21486
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.526136875152588,
      "learning_rate": 3.322867591056819e-06,
      "loss": 0.3906,
      "step": 21487
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6539756059646606,
      "learning_rate": 3.322036434233408e-06,
      "loss": 0.3581,
      "step": 21488
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9884082078933716,
      "learning_rate": 3.3212053606666772e-06,
      "loss": 0.368,
      "step": 21489
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.060178279876709,
      "learning_rate": 3.3203743703669876e-06,
      "loss": 0.3308,
      "step": 21490
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.403998374938965,
      "learning_rate": 3.3195434633446945e-06,
      "loss": 0.2806,
      "step": 21491
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9248318672180176,
      "learning_rate": 3.318712639610163e-06,
      "loss": 0.3048,
      "step": 21492
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.439474582672119,
      "learning_rate": 3.3178818991737494e-06,
      "loss": 0.3633,
      "step": 21493
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8479547500610352,
      "learning_rate": 3.317051242045809e-06,
      "loss": 0.3166,
      "step": 21494
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0443594455718994,
      "learning_rate": 3.316220668236697e-06,
      "loss": 0.3165,
      "step": 21495
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6849629878997803,
      "learning_rate": 3.315390177756773e-06,
      "loss": 0.3551,
      "step": 21496
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9048455953598022,
      "learning_rate": 3.31455977061639e-06,
      "loss": 0.3945,
      "step": 21497
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8263200521469116,
      "learning_rate": 3.3137294468258953e-06,
      "loss": 0.2906,
      "step": 21498
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.3353922367095947,
      "learning_rate": 3.312899206395649e-06,
      "loss": 0.2943,
      "step": 21499
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7479671239852905,
      "learning_rate": 3.3120690493359975e-06,
      "loss": 0.2996,
      "step": 21500
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8681823015213013,
      "learning_rate": 3.3112389756572884e-06,
      "loss": 0.3747,
      "step": 21501
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0715463161468506,
      "learning_rate": 3.3104089853698772e-06,
      "loss": 0.3718,
      "step": 21502
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8064457178115845,
      "learning_rate": 3.3095790784841075e-06,
      "loss": 0.3287,
      "step": 21503
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8761388063430786,
      "learning_rate": 3.308749255010324e-06,
      "loss": 0.3619,
      "step": 21504
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.2417190074920654,
      "learning_rate": 3.3079195149588773e-06,
      "loss": 0.3407,
      "step": 21505
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0649592876434326,
      "learning_rate": 3.307089858340111e-06,
      "loss": 0.3253,
      "step": 21506
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7857240438461304,
      "learning_rate": 3.3062602851643666e-06,
      "loss": 0.4021,
      "step": 21507
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8622082471847534,
      "learning_rate": 3.305430795441985e-06,
      "loss": 0.3402,
      "step": 21508
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3601354360580444,
      "learning_rate": 3.3046013891833127e-06,
      "loss": 0.1693,
      "step": 21509
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.8947792053222656,
      "learning_rate": 3.3037720663986884e-06,
      "loss": 0.2773,
      "step": 21510
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6663837432861328,
      "learning_rate": 3.3029428270984487e-06,
      "loss": 0.3584,
      "step": 21511
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.818065881729126,
      "learning_rate": 3.302113671292937e-06,
      "loss": 0.2615,
      "step": 21512
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.761183500289917,
      "learning_rate": 3.301284598992487e-06,
      "loss": 0.3218,
      "step": 21513
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6718066930770874,
      "learning_rate": 3.3004556102074347e-06,
      "loss": 0.3675,
      "step": 21514
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7604390382766724,
      "learning_rate": 3.299626704948119e-06,
      "loss": 0.2945,
      "step": 21515
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8245360851287842,
      "learning_rate": 3.2987978832248726e-06,
      "loss": 0.3437,
      "step": 21516
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9360343217849731,
      "learning_rate": 3.2979691450480244e-06,
      "loss": 0.3067,
      "step": 21517
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1863434314727783,
      "learning_rate": 3.2971404904279146e-06,
      "loss": 0.3938,
      "step": 21518
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1022305488586426,
      "learning_rate": 3.2963119193748695e-06,
      "loss": 0.3224,
      "step": 21519
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8643566370010376,
      "learning_rate": 3.2954834318992193e-06,
      "loss": 0.2897,
      "step": 21520
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0623323917388916,
      "learning_rate": 3.294655028011291e-06,
      "loss": 0.3346,
      "step": 21521
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9010258913040161,
      "learning_rate": 3.293826707721418e-06,
      "loss": 0.3953,
      "step": 21522
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7843014001846313,
      "learning_rate": 3.2929984710399255e-06,
      "loss": 0.3906,
      "step": 21523
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.025944948196411,
      "learning_rate": 3.292170317977135e-06,
      "loss": 0.3197,
      "step": 21524
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9238592386245728,
      "learning_rate": 3.2913422485433777e-06,
      "loss": 0.3457,
      "step": 21525
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9968316555023193,
      "learning_rate": 3.290514262748974e-06,
      "loss": 0.3597,
      "step": 21526
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0035135746002197,
      "learning_rate": 3.289686360604245e-06,
      "loss": 0.2679,
      "step": 21527
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7503275871276855,
      "learning_rate": 3.2888585421195173e-06,
      "loss": 0.3603,
      "step": 21528
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.5794663429260254,
      "learning_rate": 3.288030807305109e-06,
      "loss": 0.3267,
      "step": 21529
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7661865949630737,
      "learning_rate": 3.287203156171338e-06,
      "loss": 0.287,
      "step": 21530
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6306148767471313,
      "learning_rate": 3.2863755887285266e-06,
      "loss": 0.2714,
      "step": 21531
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.666102647781372,
      "learning_rate": 3.285548104986992e-06,
      "loss": 0.3475,
      "step": 21532
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9495266675949097,
      "learning_rate": 3.2847207049570483e-06,
      "loss": 0.4173,
      "step": 21533
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6214168071746826,
      "learning_rate": 3.2838933886490087e-06,
      "loss": 0.3322,
      "step": 21534
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9206011295318604,
      "learning_rate": 3.2830661560731947e-06,
      "loss": 0.2878,
      "step": 21535
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8961337804794312,
      "learning_rate": 3.2822390072399156e-06,
      "loss": 0.325,
      "step": 21536
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5324609279632568,
      "learning_rate": 3.2814119421594814e-06,
      "loss": 0.2751,
      "step": 21537
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8535116910934448,
      "learning_rate": 3.280584960842209e-06,
      "loss": 0.3711,
      "step": 21538
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8059401512145996,
      "learning_rate": 3.279758063298407e-06,
      "loss": 0.2654,
      "step": 21539
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7015941143035889,
      "learning_rate": 3.2789312495383795e-06,
      "loss": 0.3026,
      "step": 21540
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.183701515197754,
      "learning_rate": 3.2781045195724414e-06,
      "loss": 0.2492,
      "step": 21541
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.720115065574646,
      "learning_rate": 3.277277873410899e-06,
      "loss": 0.2862,
      "step": 21542
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9580751657485962,
      "learning_rate": 3.2764513110640518e-06,
      "loss": 0.281,
      "step": 21543
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8071893453598022,
      "learning_rate": 3.2756248325422136e-06,
      "loss": 0.326,
      "step": 21544
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.879230260848999,
      "learning_rate": 3.2747984378556842e-06,
      "loss": 0.3709,
      "step": 21545
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.2648086547851562,
      "learning_rate": 3.2739721270147673e-06,
      "loss": 0.3723,
      "step": 21546
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8837199211120605,
      "learning_rate": 3.2731459000297615e-06,
      "loss": 0.3383,
      "step": 21547
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5014541149139404,
      "learning_rate": 3.2723197569109734e-06,
      "loss": 0.2884,
      "step": 21548
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.93682861328125,
      "learning_rate": 3.2714936976686993e-06,
      "loss": 0.3239,
      "step": 21549
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.927819848060608,
      "learning_rate": 3.2706677223132365e-06,
      "loss": 0.3655,
      "step": 21550
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7106987237930298,
      "learning_rate": 3.2698418308548874e-06,
      "loss": 0.3417,
      "step": 21551
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8257559537887573,
      "learning_rate": 3.269016023303947e-06,
      "loss": 0.303,
      "step": 21552
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6022292375564575,
      "learning_rate": 3.268190299670707e-06,
      "loss": 0.3065,
      "step": 21553
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8003647327423096,
      "learning_rate": 3.2673646599654673e-06,
      "loss": 0.3429,
      "step": 21554
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.289184808731079,
      "learning_rate": 3.2665391041985208e-06,
      "loss": 0.2561,
      "step": 21555
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8791064023971558,
      "learning_rate": 3.2657136323801554e-06,
      "loss": 0.3158,
      "step": 21556
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7411521673202515,
      "learning_rate": 3.2648882445206678e-06,
      "loss": 0.3873,
      "step": 21557
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9962135553359985,
      "learning_rate": 3.264062940630347e-06,
      "loss": 0.3427,
      "step": 21558
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9361549615859985,
      "learning_rate": 3.263237720719482e-06,
      "loss": 0.3484,
      "step": 21559
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6201257705688477,
      "learning_rate": 3.2624125847983567e-06,
      "loss": 0.338,
      "step": 21560
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7451592683792114,
      "learning_rate": 3.2615875328772673e-06,
      "loss": 0.3349,
      "step": 21561
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7342190742492676,
      "learning_rate": 3.260762564966494e-06,
      "loss": 0.3291,
      "step": 21562
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7345716953277588,
      "learning_rate": 3.2599376810763208e-06,
      "loss": 0.2905,
      "step": 21563
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.62442147731781,
      "learning_rate": 3.259112881217037e-06,
      "loss": 0.374,
      "step": 21564
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7115163803100586,
      "learning_rate": 3.258288165398924e-06,
      "loss": 0.2733,
      "step": 21565
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8592805862426758,
      "learning_rate": 3.257463533632259e-06,
      "loss": 0.3405,
      "step": 21566
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6233710050582886,
      "learning_rate": 3.2566389859273297e-06,
      "loss": 0.2831,
      "step": 21567
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8468382358551025,
      "learning_rate": 3.2558145222944136e-06,
      "loss": 0.3276,
      "step": 21568
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7665671110153198,
      "learning_rate": 3.2549901427437858e-06,
      "loss": 0.3305,
      "step": 21569
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1882076263427734,
      "learning_rate": 3.254165847285731e-06,
      "loss": 0.3723,
      "step": 21570
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.099071979522705,
      "learning_rate": 3.2533416359305226e-06,
      "loss": 0.2907,
      "step": 21571
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.252376079559326,
      "learning_rate": 3.2525175086884364e-06,
      "loss": 0.3077,
      "step": 21572
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.161632537841797,
      "learning_rate": 3.2516934655697428e-06,
      "loss": 0.3326,
      "step": 21573
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6261982917785645,
      "learning_rate": 3.250869506584724e-06,
      "loss": 0.3082,
      "step": 21574
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.192976713180542,
      "learning_rate": 3.2500456317436478e-06,
      "loss": 0.354,
      "step": 21575
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.4084346294403076,
      "learning_rate": 3.2492218410567822e-06,
      "loss": 0.3756,
      "step": 21576
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8084356784820557,
      "learning_rate": 3.248398134534406e-06,
      "loss": 0.3391,
      "step": 21577
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.2364816665649414,
      "learning_rate": 3.2475745121867842e-06,
      "loss": 0.3287,
      "step": 21578
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.153024196624756,
      "learning_rate": 3.246750974024182e-06,
      "loss": 0.2835,
      "step": 21579
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6744334697723389,
      "learning_rate": 3.2459275200568728e-06,
      "loss": 0.3337,
      "step": 21580
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.803045392036438,
      "learning_rate": 3.245104150295122e-06,
      "loss": 0.3604,
      "step": 21581
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.135481834411621,
      "learning_rate": 3.2442808647491884e-06,
      "loss": 0.3708,
      "step": 21582
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5411672592163086,
      "learning_rate": 3.2434576634293447e-06,
      "loss": 0.3066,
      "step": 21583
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1728622913360596,
      "learning_rate": 3.2426345463458497e-06,
      "loss": 0.319,
      "step": 21584
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.606656551361084,
      "learning_rate": 3.2418115135089666e-06,
      "loss": 0.277,
      "step": 21585
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.280036449432373,
      "learning_rate": 3.240988564928953e-06,
      "loss": 0.341,
      "step": 21586
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.017448663711548,
      "learning_rate": 3.2401657006160746e-06,
      "loss": 0.401,
      "step": 21587
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0572924613952637,
      "learning_rate": 3.239342920580587e-06,
      "loss": 0.3722,
      "step": 21588
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7633020877838135,
      "learning_rate": 3.238520224832746e-06,
      "loss": 0.2803,
      "step": 21589
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0076231956481934,
      "learning_rate": 3.2376976133828153e-06,
      "loss": 0.3922,
      "step": 21590
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.23683500289917,
      "learning_rate": 3.236875086241046e-06,
      "loss": 0.3741,
      "step": 21591
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8853083848953247,
      "learning_rate": 3.23605264341769e-06,
      "loss": 0.3393,
      "step": 21592
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7109028100967407,
      "learning_rate": 3.2352302849230077e-06,
      "loss": 0.237,
      "step": 21593
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7383784055709839,
      "learning_rate": 3.234408010767249e-06,
      "loss": 0.3042,
      "step": 21594
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0606653690338135,
      "learning_rate": 3.233585820960662e-06,
      "loss": 0.2862,
      "step": 21595
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.1467909812927246,
      "learning_rate": 3.2327637155135027e-06,
      "loss": 0.3982,
      "step": 21596
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6750109195709229,
      "learning_rate": 3.2319416944360184e-06,
      "loss": 0.2481,
      "step": 21597
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7700016498565674,
      "learning_rate": 3.2311197577384567e-06,
      "loss": 0.3604,
      "step": 21598
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9347610473632812,
      "learning_rate": 3.2302979054310637e-06,
      "loss": 0.343,
      "step": 21599
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8593180179595947,
      "learning_rate": 3.2294761375240892e-06,
      "loss": 0.3241,
      "step": 21600
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.015901803970337,
      "learning_rate": 3.2286544540277787e-06,
      "loss": 0.2844,
      "step": 21601
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9103904962539673,
      "learning_rate": 3.2278328549523697e-06,
      "loss": 0.3143,
      "step": 21602
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.970797061920166,
      "learning_rate": 3.227011340308115e-06,
      "loss": 0.3329,
      "step": 21603
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0765185356140137,
      "learning_rate": 3.2261899101052508e-06,
      "loss": 0.3074,
      "step": 21604
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3796027898788452,
      "learning_rate": 3.2253685643540165e-06,
      "loss": 0.2944,
      "step": 21605
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9235399961471558,
      "learning_rate": 3.224547303064659e-06,
      "loss": 0.3219,
      "step": 21606
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.12332820892334,
      "learning_rate": 3.223726126247413e-06,
      "loss": 0.3637,
      "step": 21607
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9038208723068237,
      "learning_rate": 3.2229050339125133e-06,
      "loss": 0.2708,
      "step": 21608
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.397568702697754,
      "learning_rate": 3.2220840260702045e-06,
      "loss": 0.2733,
      "step": 21609
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.168689012527466,
      "learning_rate": 3.2212631027307173e-06,
      "loss": 0.4228,
      "step": 21610
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6942071914672852,
      "learning_rate": 3.2204422639042875e-06,
      "loss": 0.2712,
      "step": 21611
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.903180718421936,
      "learning_rate": 3.2196215096011453e-06,
      "loss": 0.2557,
      "step": 21612
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9331856966018677,
      "learning_rate": 3.2188008398315295e-06,
      "loss": 0.3804,
      "step": 21613
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9587013721466064,
      "learning_rate": 3.2179802546056693e-06,
      "loss": 0.378,
      "step": 21614
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1248254776000977,
      "learning_rate": 3.217159753933792e-06,
      "loss": 0.3244,
      "step": 21615
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6497143507003784,
      "learning_rate": 3.216339337826132e-06,
      "loss": 0.295,
      "step": 21616
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9440410137176514,
      "learning_rate": 3.2155190062929164e-06,
      "loss": 0.3475,
      "step": 21617
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0351343154907227,
      "learning_rate": 3.2146987593443678e-06,
      "loss": 0.4108,
      "step": 21618
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6997193098068237,
      "learning_rate": 3.2138785969907204e-06,
      "loss": 0.2627,
      "step": 21619
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9384552240371704,
      "learning_rate": 3.213058519242196e-06,
      "loss": 0.3174,
      "step": 21620
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8689833879470825,
      "learning_rate": 3.212238526109015e-06,
      "loss": 0.2932,
      "step": 21621
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7780957221984863,
      "learning_rate": 3.211418617601406e-06,
      "loss": 0.3353,
      "step": 21622
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.573331594467163,
      "learning_rate": 3.21059879372959e-06,
      "loss": 0.3054,
      "step": 21623
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6632925271987915,
      "learning_rate": 3.209779054503788e-06,
      "loss": 0.3021,
      "step": 21624
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1857171058654785,
      "learning_rate": 3.208959399934215e-06,
      "loss": 0.4005,
      "step": 21625
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.2300360202789307,
      "learning_rate": 3.2081398300310972e-06,
      "loss": 0.3303,
      "step": 21626
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6687039136886597,
      "learning_rate": 3.20732034480465e-06,
      "loss": 0.2344,
      "step": 21627
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5022034645080566,
      "learning_rate": 3.2065009442650863e-06,
      "loss": 0.2744,
      "step": 21628
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.3395307064056396,
      "learning_rate": 3.205681628422628e-06,
      "loss": 0.3831,
      "step": 21629
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8473658561706543,
      "learning_rate": 3.204862397287487e-06,
      "loss": 0.3639,
      "step": 21630
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8546981811523438,
      "learning_rate": 3.2040432508698737e-06,
      "loss": 0.3175,
      "step": 21631
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.946524739265442,
      "learning_rate": 3.2032241891800077e-06,
      "loss": 0.2861,
      "step": 21632
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.726582646369934,
      "learning_rate": 3.2024052122280957e-06,
      "loss": 0.3106,
      "step": 21633
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0939109325408936,
      "learning_rate": 3.201586320024347e-06,
      "loss": 0.3692,
      "step": 21634
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0473310947418213,
      "learning_rate": 3.200767512578976e-06,
      "loss": 0.3359,
      "step": 21635
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1967427730560303,
      "learning_rate": 3.1999487899021895e-06,
      "loss": 0.3972,
      "step": 21636
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6815509796142578,
      "learning_rate": 3.199130152004193e-06,
      "loss": 0.2517,
      "step": 21637
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9652620553970337,
      "learning_rate": 3.198311598895191e-06,
      "loss": 0.3124,
      "step": 21638
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.7928099632263184,
      "learning_rate": 3.197493130585393e-06,
      "loss": 0.3931,
      "step": 21639
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.2419559955596924,
      "learning_rate": 3.1966747470850025e-06,
      "loss": 0.3949,
      "step": 21640
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6494488716125488,
      "learning_rate": 3.1958564484042174e-06,
      "loss": 0.279,
      "step": 21641
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8886796236038208,
      "learning_rate": 3.195038234553247e-06,
      "loss": 0.3152,
      "step": 21642
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9182746410369873,
      "learning_rate": 3.1942201055422906e-06,
      "loss": 0.3392,
      "step": 21643
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.979871392250061,
      "learning_rate": 3.193402061381542e-06,
      "loss": 0.357,
      "step": 21644
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8845044374465942,
      "learning_rate": 3.1925841020812075e-06,
      "loss": 0.225,
      "step": 21645
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.751362919807434,
      "learning_rate": 3.1917662276514836e-06,
      "loss": 0.361,
      "step": 21646
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7132564783096313,
      "learning_rate": 3.1909484381025613e-06,
      "loss": 0.2612,
      "step": 21647
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9111170768737793,
      "learning_rate": 3.190130733444644e-06,
      "loss": 0.275,
      "step": 21648
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7806040048599243,
      "learning_rate": 3.189313113687923e-06,
      "loss": 0.2725,
      "step": 21649
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9164235591888428,
      "learning_rate": 3.188495578842591e-06,
      "loss": 0.2794,
      "step": 21650
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8421261310577393,
      "learning_rate": 3.1876781289188387e-06,
      "loss": 0.3073,
      "step": 21651
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9723873138427734,
      "learning_rate": 3.1868607639268633e-06,
      "loss": 0.2939,
      "step": 21652
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.4487504959106445,
      "learning_rate": 3.186043483876852e-06,
      "loss": 0.3503,
      "step": 21653
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.2647476196289062,
      "learning_rate": 3.1852262887789896e-06,
      "loss": 0.3891,
      "step": 21654
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9629191160202026,
      "learning_rate": 3.184409178643474e-06,
      "loss": 0.3368,
      "step": 21655
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8503074645996094,
      "learning_rate": 3.183592153480486e-06,
      "loss": 0.3608,
      "step": 21656
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7964519262313843,
      "learning_rate": 3.182775213300211e-06,
      "loss": 0.3031,
      "step": 21657
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8447496891021729,
      "learning_rate": 3.181958358112839e-06,
      "loss": 0.3008,
      "step": 21658
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.220597505569458,
      "learning_rate": 3.1811415879285502e-06,
      "loss": 0.3593,
      "step": 21659
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.2162137031555176,
      "learning_rate": 3.1803249027575255e-06,
      "loss": 0.262,
      "step": 21660
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7640366554260254,
      "learning_rate": 3.179508302609954e-06,
      "loss": 0.2554,
      "step": 21661
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9922010898590088,
      "learning_rate": 3.178691787496012e-06,
      "loss": 0.2348,
      "step": 21662
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8183053731918335,
      "learning_rate": 3.17787535742588e-06,
      "loss": 0.2912,
      "step": 21663
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7583342790603638,
      "learning_rate": 3.1770590124097336e-06,
      "loss": 0.2977,
      "step": 21664
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6242871284484863,
      "learning_rate": 3.1762427524577568e-06,
      "loss": 0.304,
      "step": 21665
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6561543941497803,
      "learning_rate": 3.1754265775801217e-06,
      "loss": 0.3193,
      "step": 21666
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7547001838684082,
      "learning_rate": 3.174610487787002e-06,
      "loss": 0.3232,
      "step": 21667
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0706748962402344,
      "learning_rate": 3.1737944830885793e-06,
      "loss": 0.2937,
      "step": 21668
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8811113834381104,
      "learning_rate": 3.1729785634950215e-06,
      "loss": 0.3207,
      "step": 21669
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9376124143600464,
      "learning_rate": 3.1721627290165003e-06,
      "loss": 0.2777,
      "step": 21670
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.308321714401245,
      "learning_rate": 3.1713469796631924e-06,
      "loss": 0.3173,
      "step": 21671
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8962955474853516,
      "learning_rate": 3.170531315445264e-06,
      "loss": 0.3485,
      "step": 21672
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9016051292419434,
      "learning_rate": 3.1697157363728824e-06,
      "loss": 0.2776,
      "step": 21673
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5789768695831299,
      "learning_rate": 3.168900242456221e-06,
      "loss": 0.2922,
      "step": 21674
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8405598402023315,
      "learning_rate": 3.1680848337054448e-06,
      "loss": 0.3285,
      "step": 21675
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.071636199951172,
      "learning_rate": 3.1672695101307195e-06,
      "loss": 0.3534,
      "step": 21676
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.217343807220459,
      "learning_rate": 3.1664542717422053e-06,
      "loss": 0.3348,
      "step": 21677
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9814412593841553,
      "learning_rate": 3.1656391185500747e-06,
      "loss": 0.342,
      "step": 21678
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1029529571533203,
      "learning_rate": 3.164824050564487e-06,
      "loss": 0.3562,
      "step": 21679
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.835524082183838,
      "learning_rate": 3.164009067795599e-06,
      "loss": 0.2883,
      "step": 21680
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1195595264434814,
      "learning_rate": 3.1631941702535795e-06,
      "loss": 0.2993,
      "step": 21681
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.272327423095703,
      "learning_rate": 3.162379357948584e-06,
      "loss": 0.3158,
      "step": 21682
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.1352336406707764,
      "learning_rate": 3.161564630890769e-06,
      "loss": 0.3126,
      "step": 21683
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.671330213546753,
      "learning_rate": 3.160749989090296e-06,
      "loss": 0.2298,
      "step": 21684
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8443403244018555,
      "learning_rate": 3.1599354325573216e-06,
      "loss": 0.3278,
      "step": 21685
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7267884016036987,
      "learning_rate": 3.1591209613019957e-06,
      "loss": 0.2419,
      "step": 21686
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0117380619049072,
      "learning_rate": 3.15830657533448e-06,
      "loss": 0.2882,
      "step": 21687
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.003023862838745,
      "learning_rate": 3.157492274664924e-06,
      "loss": 0.3142,
      "step": 21688
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.065471649169922,
      "learning_rate": 3.1566780593034795e-06,
      "loss": 0.2812,
      "step": 21689
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.572055697441101,
      "learning_rate": 3.1558639292602954e-06,
      "loss": 0.2913,
      "step": 21690
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.01654052734375,
      "learning_rate": 3.155049884545527e-06,
      "loss": 0.4328,
      "step": 21691
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7946925163269043,
      "learning_rate": 3.1542359251693223e-06,
      "loss": 0.3133,
      "step": 21692
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.877958059310913,
      "learning_rate": 3.1534220511418236e-06,
      "loss": 0.3662,
      "step": 21693
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.025920867919922,
      "learning_rate": 3.152608262473186e-06,
      "loss": 0.3287,
      "step": 21694
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8220248222351074,
      "learning_rate": 3.1517945591735497e-06,
      "loss": 0.2768,
      "step": 21695
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7766748666763306,
      "learning_rate": 3.150980941253059e-06,
      "loss": 0.3089,
      "step": 21696
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7663989067077637,
      "learning_rate": 3.1501674087218625e-06,
      "loss": 0.2985,
      "step": 21697
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7244982719421387,
      "learning_rate": 3.1493539615901004e-06,
      "loss": 0.2298,
      "step": 21698
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0235977172851562,
      "learning_rate": 3.1485405998679098e-06,
      "loss": 0.2168,
      "step": 21699
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.094972848892212,
      "learning_rate": 3.147727323565438e-06,
      "loss": 0.3029,
      "step": 21700
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7845492362976074,
      "learning_rate": 3.1469141326928233e-06,
      "loss": 0.2685,
      "step": 21701
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7305108308792114,
      "learning_rate": 3.146101027260201e-06,
      "loss": 0.2404,
      "step": 21702
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9740327596664429,
      "learning_rate": 3.145288007277706e-06,
      "loss": 0.3493,
      "step": 21703
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8001874685287476,
      "learning_rate": 3.144475072755483e-06,
      "loss": 0.2651,
      "step": 21704
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.0447611808776855,
      "learning_rate": 3.1436622237036607e-06,
      "loss": 0.3493,
      "step": 21705
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.63558030128479,
      "learning_rate": 3.1428494601323735e-06,
      "loss": 0.3465,
      "step": 21706
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1307120323181152,
      "learning_rate": 3.142036782051757e-06,
      "loss": 0.2923,
      "step": 21707
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5704731941223145,
      "learning_rate": 3.1412241894719444e-06,
      "loss": 0.2456,
      "step": 21708
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0094757080078125,
      "learning_rate": 3.1404116824030596e-06,
      "loss": 0.3198,
      "step": 21709
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.915264368057251,
      "learning_rate": 3.1395992608552397e-06,
      "loss": 0.3717,
      "step": 21710
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9774169921875,
      "learning_rate": 3.1387869248386117e-06,
      "loss": 0.3071,
      "step": 21711
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9915519952774048,
      "learning_rate": 3.137974674363301e-06,
      "loss": 0.2692,
      "step": 21712
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.673045039176941,
      "learning_rate": 3.1371625094394344e-06,
      "loss": 0.2684,
      "step": 21713
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7457813024520874,
      "learning_rate": 3.1363504300771407e-06,
      "loss": 0.3303,
      "step": 21714
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.881357192993164,
      "learning_rate": 3.1355384362865427e-06,
      "loss": 0.3395,
      "step": 21715
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0272128582000732,
      "learning_rate": 3.1347265280777597e-06,
      "loss": 0.3585,
      "step": 21716
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.750868558883667,
      "learning_rate": 3.133914705460922e-06,
      "loss": 0.3526,
      "step": 21717
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.836868166923523,
      "learning_rate": 3.1331029684461457e-06,
      "loss": 0.3435,
      "step": 21718
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6552702188491821,
      "learning_rate": 3.13229131704355e-06,
      "loss": 0.2677,
      "step": 21719
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.05078125,
      "learning_rate": 3.1314797512632578e-06,
      "loss": 0.3191,
      "step": 21720
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6337857246398926,
      "learning_rate": 3.1306682711153868e-06,
      "loss": 0.2618,
      "step": 21721
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0796186923980713,
      "learning_rate": 3.129856876610049e-06,
      "loss": 0.3752,
      "step": 21722
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6736412048339844,
      "learning_rate": 3.1290455677573684e-06,
      "loss": 0.2558,
      "step": 21723
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9483380317687988,
      "learning_rate": 3.1282343445674545e-06,
      "loss": 0.3366,
      "step": 21724
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8352224826812744,
      "learning_rate": 3.127423207050423e-06,
      "loss": 0.3305,
      "step": 21725
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7318352460861206,
      "learning_rate": 3.1266121552163818e-06,
      "loss": 0.4178,
      "step": 21726
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0754270553588867,
      "learning_rate": 3.12580118907545e-06,
      "loss": 0.3202,
      "step": 21727
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7439353466033936,
      "learning_rate": 3.1249903086377363e-06,
      "loss": 0.2925,
      "step": 21728
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.662348747253418,
      "learning_rate": 3.124179513913345e-06,
      "loss": 0.2411,
      "step": 21729
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1002535820007324,
      "learning_rate": 3.1233688049123912e-06,
      "loss": 0.4138,
      "step": 21730
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6274383068084717,
      "learning_rate": 3.122558181644979e-06,
      "loss": 0.2466,
      "step": 21731
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6060312986373901,
      "learning_rate": 3.1217476441212135e-06,
      "loss": 0.3079,
      "step": 21732
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.103287696838379,
      "learning_rate": 3.1209371923512034e-06,
      "loss": 0.3456,
      "step": 21733
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8012892007827759,
      "learning_rate": 3.1201268263450525e-06,
      "loss": 0.2371,
      "step": 21734
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6403696537017822,
      "learning_rate": 3.1193165461128595e-06,
      "loss": 0.2805,
      "step": 21735
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.963111162185669,
      "learning_rate": 3.118506351664733e-06,
      "loss": 0.3282,
      "step": 21736
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0327463150024414,
      "learning_rate": 3.1176962430107694e-06,
      "loss": 0.31,
      "step": 21737
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8914273977279663,
      "learning_rate": 3.1168862201610707e-06,
      "loss": 0.3317,
      "step": 21738
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.192235231399536,
      "learning_rate": 3.116076283125732e-06,
      "loss": 0.3179,
      "step": 21739
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1014392375946045,
      "learning_rate": 3.1152664319148574e-06,
      "loss": 0.3007,
      "step": 21740
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9070022106170654,
      "learning_rate": 3.11445666653854e-06,
      "loss": 0.3566,
      "step": 21741
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6979080438613892,
      "learning_rate": 3.113646987006872e-06,
      "loss": 0.3155,
      "step": 21742
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1685054302215576,
      "learning_rate": 3.112837393329956e-06,
      "loss": 0.3618,
      "step": 21743
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.054821491241455,
      "learning_rate": 3.112027885517881e-06,
      "loss": 0.3592,
      "step": 21744
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.346322774887085,
      "learning_rate": 3.1112184635807352e-06,
      "loss": 0.3598,
      "step": 21745
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5647540092468262,
      "learning_rate": 3.1104091275286185e-06,
      "loss": 0.2263,
      "step": 21746
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7043155431747437,
      "learning_rate": 3.109599877371617e-06,
      "loss": 0.2977,
      "step": 21747
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9186389446258545,
      "learning_rate": 3.108790713119818e-06,
      "loss": 0.3137,
      "step": 21748
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7567436695098877,
      "learning_rate": 3.1079816347833134e-06,
      "loss": 0.3354,
      "step": 21749
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9422998428344727,
      "learning_rate": 3.1071726423721894e-06,
      "loss": 0.311,
      "step": 21750
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7855044603347778,
      "learning_rate": 3.1063637358965313e-06,
      "loss": 0.3308,
      "step": 21751
    },
    {
      "epoch": 2.25,
      "grad_norm": 3.0066120624542236,
      "learning_rate": 3.10555491536642e-06,
      "loss": 0.4246,
      "step": 21752
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.462994933128357,
      "learning_rate": 3.1047461807919465e-06,
      "loss": 0.2649,
      "step": 21753
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.2981488704681396,
      "learning_rate": 3.103937532183191e-06,
      "loss": 0.3608,
      "step": 21754
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8720927238464355,
      "learning_rate": 3.103128969550231e-06,
      "loss": 0.315,
      "step": 21755
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1855201721191406,
      "learning_rate": 3.1023204929031545e-06,
      "loss": 0.3504,
      "step": 21756
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8454536199569702,
      "learning_rate": 3.1015121022520367e-06,
      "loss": 0.3297,
      "step": 21757
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5836033821105957,
      "learning_rate": 3.100703797606954e-06,
      "loss": 0.3043,
      "step": 21758
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.629932165145874,
      "learning_rate": 3.0998955789779893e-06,
      "loss": 0.2789,
      "step": 21759
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.342822790145874,
      "learning_rate": 3.0990874463752175e-06,
      "loss": 0.3215,
      "step": 21760
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1365246772766113,
      "learning_rate": 3.098279399808708e-06,
      "loss": 0.3281,
      "step": 21761
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.5282206535339355,
      "learning_rate": 3.097471439288544e-06,
      "loss": 0.3353,
      "step": 21762
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.787479281425476,
      "learning_rate": 3.096663564824793e-06,
      "loss": 0.3249,
      "step": 21763
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.385188102722168,
      "learning_rate": 3.0958557764275297e-06,
      "loss": 0.4276,
      "step": 21764
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0011329650878906,
      "learning_rate": 3.095048074106821e-06,
      "loss": 0.3496,
      "step": 21765
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7426016330718994,
      "learning_rate": 3.0942404578727413e-06,
      "loss": 0.2617,
      "step": 21766
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.820753812789917,
      "learning_rate": 3.0934329277353593e-06,
      "loss": 0.2564,
      "step": 21767
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0389530658721924,
      "learning_rate": 3.0926254837047365e-06,
      "loss": 0.3182,
      "step": 21768
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.5621540546417236,
      "learning_rate": 3.0918181257909485e-06,
      "loss": 0.2799,
      "step": 21769
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9690667390823364,
      "learning_rate": 3.091010854004057e-06,
      "loss": 0.3287,
      "step": 21770
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0268235206604004,
      "learning_rate": 3.0902036683541227e-06,
      "loss": 0.3026,
      "step": 21771
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7953226566314697,
      "learning_rate": 3.0893965688512163e-06,
      "loss": 0.2891,
      "step": 21772
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.660086989402771,
      "learning_rate": 3.088589555505397e-06,
      "loss": 0.2979,
      "step": 21773
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.848823070526123,
      "learning_rate": 3.087782628326722e-06,
      "loss": 0.2981,
      "step": 21774
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.3058717250823975,
      "learning_rate": 3.0869757873252583e-06,
      "loss": 0.3351,
      "step": 21775
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8895446062088013,
      "learning_rate": 3.0861690325110625e-06,
      "loss": 0.2315,
      "step": 21776
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.2646329402923584,
      "learning_rate": 3.0853623638941933e-06,
      "loss": 0.3321,
      "step": 21777
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.1261260509490967,
      "learning_rate": 3.084555781484703e-06,
      "loss": 0.2852,
      "step": 21778
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9791104793548584,
      "learning_rate": 3.0837492852926543e-06,
      "loss": 0.3144,
      "step": 21779
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7964667081832886,
      "learning_rate": 3.0829428753280986e-06,
      "loss": 0.3121,
      "step": 21780
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9956793785095215,
      "learning_rate": 3.0821365516010883e-06,
      "loss": 0.3294,
      "step": 21781
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9900904893875122,
      "learning_rate": 3.0813303141216798e-06,
      "loss": 0.2789,
      "step": 21782
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.2473437786102295,
      "learning_rate": 3.080524162899925e-06,
      "loss": 0.3631,
      "step": 21783
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9431833028793335,
      "learning_rate": 3.0797180979458684e-06,
      "loss": 0.3308,
      "step": 21784
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9721256494522095,
      "learning_rate": 3.078912119269567e-06,
      "loss": 0.287,
      "step": 21785
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.198509693145752,
      "learning_rate": 3.0781062268810656e-06,
      "loss": 0.227,
      "step": 21786
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.945101022720337,
      "learning_rate": 3.077300420790409e-06,
      "loss": 0.3179,
      "step": 21787
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.10219407081604,
      "learning_rate": 3.0764947010076494e-06,
      "loss": 0.3473,
      "step": 21788
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.785895586013794,
      "learning_rate": 3.07568906754283e-06,
      "loss": 0.234,
      "step": 21789
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6522170305252075,
      "learning_rate": 3.0748835204059934e-06,
      "loss": 0.3092,
      "step": 21790
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7953466176986694,
      "learning_rate": 3.0740780596071795e-06,
      "loss": 0.2982,
      "step": 21791
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.837417483329773,
      "learning_rate": 3.0732726851564374e-06,
      "loss": 0.3362,
      "step": 21792
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9646250009536743,
      "learning_rate": 3.0724673970638053e-06,
      "loss": 0.2936,
      "step": 21793
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.0463662147521973,
      "learning_rate": 3.071662195339318e-06,
      "loss": 0.2551,
      "step": 21794
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.100543260574341,
      "learning_rate": 3.070857079993023e-06,
      "loss": 0.3536,
      "step": 21795
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.369459867477417,
      "learning_rate": 3.070052051034952e-06,
      "loss": 0.3632,
      "step": 21796
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.4408230781555176,
      "learning_rate": 3.0692471084751406e-06,
      "loss": 0.3014,
      "step": 21797
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6399078369140625,
      "learning_rate": 3.0684422523236336e-06,
      "loss": 0.2776,
      "step": 21798
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.974969506263733,
      "learning_rate": 3.0676374825904555e-06,
      "loss": 0.3736,
      "step": 21799
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8343158960342407,
      "learning_rate": 3.0668327992856395e-06,
      "loss": 0.3079,
      "step": 21800
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.417754888534546,
      "learning_rate": 3.0660282024192245e-06,
      "loss": 0.3804,
      "step": 21801
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.9954402446746826,
      "learning_rate": 3.065223692001238e-06,
      "loss": 0.2343,
      "step": 21802
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.2782533168792725,
      "learning_rate": 3.0644192680417086e-06,
      "loss": 0.3761,
      "step": 21803
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0215203762054443,
      "learning_rate": 3.063614930550669e-06,
      "loss": 0.3674,
      "step": 21804
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.3983075618743896,
      "learning_rate": 3.0628106795381463e-06,
      "loss": 0.3726,
      "step": 21805
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9907283782958984,
      "learning_rate": 3.0620065150141633e-06,
      "loss": 0.3589,
      "step": 21806
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7309759855270386,
      "learning_rate": 3.061202436988753e-06,
      "loss": 0.3155,
      "step": 21807
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.17968487739563,
      "learning_rate": 3.060398445471936e-06,
      "loss": 0.3459,
      "step": 21808
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.918628454208374,
      "learning_rate": 3.059594540473733e-06,
      "loss": 0.3321,
      "step": 21809
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2712819576263428,
      "learning_rate": 3.0587907220041723e-06,
      "loss": 0.368,
      "step": 21810
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4517621994018555,
      "learning_rate": 3.057986990073274e-06,
      "loss": 0.2015,
      "step": 21811
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.211714744567871,
      "learning_rate": 3.057183344691057e-06,
      "loss": 0.3956,
      "step": 21812
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.776517629623413,
      "learning_rate": 3.0563797858675383e-06,
      "loss": 0.3846,
      "step": 21813
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.194021224975586,
      "learning_rate": 3.0555763136127415e-06,
      "loss": 0.3827,
      "step": 21814
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8533214330673218,
      "learning_rate": 3.054772927936681e-06,
      "loss": 0.2861,
      "step": 21815
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.836745023727417,
      "learning_rate": 3.0539696288493714e-06,
      "loss": 0.2945,
      "step": 21816
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9134268760681152,
      "learning_rate": 3.053166416360832e-06,
      "loss": 0.3498,
      "step": 21817
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5960675477981567,
      "learning_rate": 3.052363290481074e-06,
      "loss": 0.3083,
      "step": 21818
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5839977264404297,
      "learning_rate": 3.051560251220108e-06,
      "loss": 0.3576,
      "step": 21819
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9822170734405518,
      "learning_rate": 3.050757298587951e-06,
      "loss": 0.3171,
      "step": 21820
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.584841012954712,
      "learning_rate": 3.049954432594612e-06,
      "loss": 0.289,
      "step": 21821
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.3698136806488037,
      "learning_rate": 3.049151653250095e-06,
      "loss": 0.3169,
      "step": 21822
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.3895392417907715,
      "learning_rate": 3.0483489605644177e-06,
      "loss": 0.3696,
      "step": 21823
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7747563123703003,
      "learning_rate": 3.0475463545475825e-06,
      "loss": 0.3419,
      "step": 21824
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9474645853042603,
      "learning_rate": 3.046743835209597e-06,
      "loss": 0.2656,
      "step": 21825
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.761041283607483,
      "learning_rate": 3.0459414025604618e-06,
      "loss": 0.322,
      "step": 21826
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5540155172348022,
      "learning_rate": 3.045139056610189e-06,
      "loss": 0.2483,
      "step": 21827
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.1689090728759766,
      "learning_rate": 3.0443367973687776e-06,
      "loss": 0.3534,
      "step": 21828
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.492771029472351,
      "learning_rate": 3.0435346248462274e-06,
      "loss": 0.2626,
      "step": 21829
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6334301233291626,
      "learning_rate": 3.0427325390525454e-06,
      "loss": 0.3185,
      "step": 21830
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.3312432765960693,
      "learning_rate": 3.0419305399977273e-06,
      "loss": 0.3246,
      "step": 21831
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.031240224838257,
      "learning_rate": 3.0411286276917707e-06,
      "loss": 0.3443,
      "step": 21832
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.073664665222168,
      "learning_rate": 3.0403268021446774e-06,
      "loss": 0.3703,
      "step": 21833
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.1683263778686523,
      "learning_rate": 3.0395250633664432e-06,
      "loss": 0.3076,
      "step": 21834
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6386256217956543,
      "learning_rate": 3.0387234113670585e-06,
      "loss": 0.372,
      "step": 21835
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9962968826293945,
      "learning_rate": 3.0379218461565253e-06,
      "loss": 0.3305,
      "step": 21836
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.005681037902832,
      "learning_rate": 3.0371203677448337e-06,
      "loss": 0.3481,
      "step": 21837
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5278569459915161,
      "learning_rate": 3.0363189761419753e-06,
      "loss": 0.2459,
      "step": 21838
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.820203185081482,
      "learning_rate": 3.0355176713579394e-06,
      "loss": 0.3218,
      "step": 21839
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6976070404052734,
      "learning_rate": 3.0347164534027216e-06,
      "loss": 0.2987,
      "step": 21840
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9052163362503052,
      "learning_rate": 3.0339153222863083e-06,
      "loss": 0.3254,
      "step": 21841
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.145902156829834,
      "learning_rate": 3.033114278018684e-06,
      "loss": 0.3497,
      "step": 21842
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0393755435943604,
      "learning_rate": 3.032313320609842e-06,
      "loss": 0.3382,
      "step": 21843
    },
    {
      "epoch": 2.26,
      "grad_norm": 3.1375679969787598,
      "learning_rate": 3.0315124500697656e-06,
      "loss": 0.4394,
      "step": 21844
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9887254238128662,
      "learning_rate": 3.0307116664084357e-06,
      "loss": 0.2928,
      "step": 21845
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0725369453430176,
      "learning_rate": 3.029910969635842e-06,
      "loss": 0.315,
      "step": 21846
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.286360740661621,
      "learning_rate": 3.029110359761964e-06,
      "loss": 0.3573,
      "step": 21847
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7794386148452759,
      "learning_rate": 3.028309836796781e-06,
      "loss": 0.2456,
      "step": 21848
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.900456190109253,
      "learning_rate": 3.0275094007502802e-06,
      "loss": 0.3455,
      "step": 21849
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2716543674468994,
      "learning_rate": 3.0267090516324347e-06,
      "loss": 0.3727,
      "step": 21850
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2295567989349365,
      "learning_rate": 3.0259087894532267e-06,
      "loss": 0.3644,
      "step": 21851
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.940286636352539,
      "learning_rate": 3.025108614222627e-06,
      "loss": 0.3208,
      "step": 21852
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0654032230377197,
      "learning_rate": 3.02430852595062e-06,
      "loss": 0.266,
      "step": 21853
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8121541738510132,
      "learning_rate": 3.0235085246471763e-06,
      "loss": 0.3524,
      "step": 21854
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2725751399993896,
      "learning_rate": 3.022708610322267e-06,
      "loss": 0.34,
      "step": 21855
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8360953330993652,
      "learning_rate": 3.021908782985872e-06,
      "loss": 0.328,
      "step": 21856
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.7808518409729004,
      "learning_rate": 3.0211090426479596e-06,
      "loss": 0.4399,
      "step": 21857
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2277798652648926,
      "learning_rate": 3.0203093893184964e-06,
      "loss": 0.3258,
      "step": 21858
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7053831815719604,
      "learning_rate": 3.019509823007459e-06,
      "loss": 0.3032,
      "step": 21859
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.7677557468414307,
      "learning_rate": 3.0187103437248132e-06,
      "loss": 0.369,
      "step": 21860
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2151241302490234,
      "learning_rate": 3.0179109514805217e-06,
      "loss": 0.3162,
      "step": 21861
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2136800289154053,
      "learning_rate": 3.0171116462845594e-06,
      "loss": 0.3793,
      "step": 21862
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0767550468444824,
      "learning_rate": 3.0163124281468857e-06,
      "loss": 0.3154,
      "step": 21863
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0497961044311523,
      "learning_rate": 3.015513297077468e-06,
      "loss": 0.3638,
      "step": 21864
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.882745623588562,
      "learning_rate": 3.0147142530862627e-06,
      "loss": 0.355,
      "step": 21865
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8947755098342896,
      "learning_rate": 3.013915296183241e-06,
      "loss": 0.396,
      "step": 21866
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4107731580734253,
      "learning_rate": 3.0131164263783587e-06,
      "loss": 0.2407,
      "step": 21867
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6823947429656982,
      "learning_rate": 3.0123176436815736e-06,
      "loss": 0.3554,
      "step": 21868
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5836148262023926,
      "learning_rate": 3.0115189481028506e-06,
      "loss": 0.2562,
      "step": 21869
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6221022605895996,
      "learning_rate": 3.0107203396521434e-06,
      "loss": 0.317,
      "step": 21870
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8327356576919556,
      "learning_rate": 3.009921818339405e-06,
      "loss": 0.26,
      "step": 21871
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.564511299133301,
      "learning_rate": 3.009123384174599e-06,
      "loss": 0.363,
      "step": 21872
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9628514051437378,
      "learning_rate": 3.0083250371676762e-06,
      "loss": 0.4018,
      "step": 21873
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5615254640579224,
      "learning_rate": 3.0075267773285853e-06,
      "loss": 0.2952,
      "step": 21874
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.816908359527588,
      "learning_rate": 3.0067286046672874e-06,
      "loss": 0.3041,
      "step": 21875
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.705304741859436,
      "learning_rate": 3.005930519193727e-06,
      "loss": 0.3015,
      "step": 21876
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.15228271484375,
      "learning_rate": 3.0051325209178572e-06,
      "loss": 0.2551,
      "step": 21877
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0571062564849854,
      "learning_rate": 3.0043346098496217e-06,
      "loss": 0.2341,
      "step": 21878
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.431232452392578,
      "learning_rate": 3.003536785998975e-06,
      "loss": 0.311,
      "step": 21879
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.139911413192749,
      "learning_rate": 3.0027390493758626e-06,
      "loss": 0.3633,
      "step": 21880
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8961169719696045,
      "learning_rate": 3.0019413999902247e-06,
      "loss": 0.2844,
      "step": 21881
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.4955081939697266,
      "learning_rate": 3.001143837852013e-06,
      "loss": 0.417,
      "step": 21882
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.353767156600952,
      "learning_rate": 3.000346362971169e-06,
      "loss": 0.3392,
      "step": 21883
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.070512294769287,
      "learning_rate": 2.99954897535763e-06,
      "loss": 0.3338,
      "step": 21884
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.844254493713379,
      "learning_rate": 2.9987516750213454e-06,
      "loss": 0.3153,
      "step": 21885
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.825640082359314,
      "learning_rate": 2.9979544619722513e-06,
      "loss": 0.3153,
      "step": 21886
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.52535080909729,
      "learning_rate": 2.9971573362202833e-06,
      "loss": 0.2172,
      "step": 21887
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.927131175994873,
      "learning_rate": 2.9963602977753868e-06,
      "loss": 0.3074,
      "step": 21888
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.2857675552368164,
      "learning_rate": 2.9955633466474953e-06,
      "loss": 0.3206,
      "step": 21889
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.0569872856140137,
      "learning_rate": 2.994766482846544e-06,
      "loss": 0.2679,
      "step": 21890
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.1643989086151123,
      "learning_rate": 2.993969706382466e-06,
      "loss": 0.3535,
      "step": 21891
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9493592977523804,
      "learning_rate": 2.9931730172651995e-06,
      "loss": 0.2478,
      "step": 21892
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.808256149291992,
      "learning_rate": 2.9923764155046754e-06,
      "loss": 0.2856,
      "step": 21893
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.409576654434204,
      "learning_rate": 2.9915799011108217e-06,
      "loss": 0.3093,
      "step": 21894
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.1902995109558105,
      "learning_rate": 2.990783474093575e-06,
      "loss": 0.3411,
      "step": 21895
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8996717929840088,
      "learning_rate": 2.989987134462862e-06,
      "loss": 0.3123,
      "step": 21896
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4895317554473877,
      "learning_rate": 2.9891908822286064e-06,
      "loss": 0.2496,
      "step": 21897
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.1077656745910645,
      "learning_rate": 2.9883947174007434e-06,
      "loss": 0.3987,
      "step": 21898
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7316181659698486,
      "learning_rate": 2.987598639989194e-06,
      "loss": 0.3125,
      "step": 21899
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.913881778717041,
      "learning_rate": 2.986802650003883e-06,
      "loss": 0.2856,
      "step": 21900
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.024062395095825,
      "learning_rate": 2.986006747454737e-06,
      "loss": 0.2888,
      "step": 21901
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8488843441009521,
      "learning_rate": 2.985210932351679e-06,
      "loss": 0.259,
      "step": 21902
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9137569665908813,
      "learning_rate": 2.9844152047046283e-06,
      "loss": 0.3322,
      "step": 21903
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.1544268131256104,
      "learning_rate": 2.983619564523503e-06,
      "loss": 0.2403,
      "step": 21904
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.2473597526550293,
      "learning_rate": 2.9828240118182305e-06,
      "loss": 0.3212,
      "step": 21905
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.406489610671997,
      "learning_rate": 2.9820285465987244e-06,
      "loss": 0.4002,
      "step": 21906
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7641749382019043,
      "learning_rate": 2.9812331688748986e-06,
      "loss": 0.3082,
      "step": 21907
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.153658390045166,
      "learning_rate": 2.980437878656677e-06,
      "loss": 0.3548,
      "step": 21908
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.858789086341858,
      "learning_rate": 2.979642675953971e-06,
      "loss": 0.3216,
      "step": 21909
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.153254508972168,
      "learning_rate": 2.978847560776691e-06,
      "loss": 0.2849,
      "step": 21910
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.2629196643829346,
      "learning_rate": 2.9780525331347574e-06,
      "loss": 0.2512,
      "step": 21911
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0745458602905273,
      "learning_rate": 2.977257593038079e-06,
      "loss": 0.3766,
      "step": 21912
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.7521917819976807,
      "learning_rate": 2.976462740496563e-06,
      "loss": 0.37,
      "step": 21913
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9488459825515747,
      "learning_rate": 2.9756679755201245e-06,
      "loss": 0.2979,
      "step": 21914
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6668026447296143,
      "learning_rate": 2.97487329811867e-06,
      "loss": 0.2609,
      "step": 21915
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7344098091125488,
      "learning_rate": 2.9740787083021073e-06,
      "loss": 0.2781,
      "step": 21916
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.955275297164917,
      "learning_rate": 2.973284206080339e-06,
      "loss": 0.2444,
      "step": 21917
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4938726425170898,
      "learning_rate": 2.9724897914632765e-06,
      "loss": 0.2662,
      "step": 21918
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7968511581420898,
      "learning_rate": 2.971695464460822e-06,
      "loss": 0.2726,
      "step": 21919
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4527335166931152,
      "learning_rate": 2.970901225082875e-06,
      "loss": 0.2919,
      "step": 21920
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.4234421253204346,
      "learning_rate": 2.9701070733393446e-06,
      "loss": 0.3266,
      "step": 21921
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6644890308380127,
      "learning_rate": 2.969313009240127e-06,
      "loss": 0.2888,
      "step": 21922
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4951319694519043,
      "learning_rate": 2.9685190327951196e-06,
      "loss": 0.3293,
      "step": 21923
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8464359045028687,
      "learning_rate": 2.9677251440142286e-06,
      "loss": 0.254,
      "step": 21924
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8560436964035034,
      "learning_rate": 2.9669313429073467e-06,
      "loss": 0.3821,
      "step": 21925
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.684891939163208,
      "learning_rate": 2.9661376294843693e-06,
      "loss": 0.3685,
      "step": 21926
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.1276662349700928,
      "learning_rate": 2.9653440037551962e-06,
      "loss": 0.3081,
      "step": 21927
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.03366756439209,
      "learning_rate": 2.9645504657297207e-06,
      "loss": 0.2618,
      "step": 21928
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.193845748901367,
      "learning_rate": 2.9637570154178354e-06,
      "loss": 0.2383,
      "step": 21929
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.2355332374572754,
      "learning_rate": 2.962963652829428e-06,
      "loss": 0.386,
      "step": 21930
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8116375207901,
      "learning_rate": 2.962170377974398e-06,
      "loss": 0.2877,
      "step": 21931
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6847678422927856,
      "learning_rate": 2.961377190862631e-06,
      "loss": 0.2984,
      "step": 21932
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0509531497955322,
      "learning_rate": 2.960584091504014e-06,
      "loss": 0.3446,
      "step": 21933
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0702121257781982,
      "learning_rate": 2.9597910799084407e-06,
      "loss": 0.3681,
      "step": 21934
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8387320041656494,
      "learning_rate": 2.9589981560857926e-06,
      "loss": 0.3439,
      "step": 21935
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.026484966278076,
      "learning_rate": 2.958205320045956e-06,
      "loss": 0.3545,
      "step": 21936
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0453784465789795,
      "learning_rate": 2.957412571798819e-06,
      "loss": 0.3436,
      "step": 21937
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.228679895401001,
      "learning_rate": 2.9566199113542628e-06,
      "loss": 0.301,
      "step": 21938
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8437508344650269,
      "learning_rate": 2.955827338722167e-06,
      "loss": 0.3161,
      "step": 21939
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7774394750595093,
      "learning_rate": 2.9550348539124186e-06,
      "loss": 0.28,
      "step": 21940
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.765611171722412,
      "learning_rate": 2.9542424569348947e-06,
      "loss": 0.3171,
      "step": 21941
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6548972129821777,
      "learning_rate": 2.953450147799475e-06,
      "loss": 0.3357,
      "step": 21942
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.7184832096099854,
      "learning_rate": 2.952657926516034e-06,
      "loss": 0.4503,
      "step": 21943
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.385768175125122,
      "learning_rate": 2.951865793094455e-06,
      "loss": 0.3133,
      "step": 21944
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5401901006698608,
      "learning_rate": 2.9510737475446093e-06,
      "loss": 0.2935,
      "step": 21945
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.295675039291382,
      "learning_rate": 2.9502817898763715e-06,
      "loss": 0.3677,
      "step": 21946
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9713233709335327,
      "learning_rate": 2.9494899200996184e-06,
      "loss": 0.3519,
      "step": 21947
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.857925534248352,
      "learning_rate": 2.948698138224222e-06,
      "loss": 0.2884,
      "step": 21948
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.677579402923584,
      "learning_rate": 2.947906444260049e-06,
      "loss": 0.2666,
      "step": 21949
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.009064197540283,
      "learning_rate": 2.9471148382169755e-06,
      "loss": 0.3016,
      "step": 21950
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8619943857192993,
      "learning_rate": 2.9463233201048703e-06,
      "loss": 0.2797,
      "step": 21951
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.654962420463562,
      "learning_rate": 2.945531889933595e-06,
      "loss": 0.301,
      "step": 21952
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.69646954536438,
      "learning_rate": 2.9447405477130253e-06,
      "loss": 0.3488,
      "step": 21953
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7294236421585083,
      "learning_rate": 2.9439492934530236e-06,
      "loss": 0.3132,
      "step": 21954
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.2451281547546387,
      "learning_rate": 2.9431581271634535e-06,
      "loss": 0.3113,
      "step": 21955
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6303157806396484,
      "learning_rate": 2.9423670488541767e-06,
      "loss": 0.3122,
      "step": 21956
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9705572128295898,
      "learning_rate": 2.9415760585350628e-06,
      "loss": 0.2875,
      "step": 21957
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.789886236190796,
      "learning_rate": 2.9407851562159682e-06,
      "loss": 0.3375,
      "step": 21958
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.669811725616455,
      "learning_rate": 2.9399943419067534e-06,
      "loss": 0.3712,
      "step": 21959
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.634185552597046,
      "learning_rate": 2.939203615617281e-06,
      "loss": 0.2796,
      "step": 21960
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.7147135734558105,
      "learning_rate": 2.9384129773574065e-06,
      "loss": 0.3704,
      "step": 21961
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9484210014343262,
      "learning_rate": 2.9376224271369857e-06,
      "loss": 0.3244,
      "step": 21962
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8986841440200806,
      "learning_rate": 2.9368319649658805e-06,
      "loss": 0.3049,
      "step": 21963
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.999527096748352,
      "learning_rate": 2.936041590853942e-06,
      "loss": 0.3217,
      "step": 21964
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.3986728191375732,
      "learning_rate": 2.935251304811021e-06,
      "loss": 0.3305,
      "step": 21965
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.004978895187378,
      "learning_rate": 2.9344611068469764e-06,
      "loss": 0.252,
      "step": 21966
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7122457027435303,
      "learning_rate": 2.933670996971657e-06,
      "loss": 0.3528,
      "step": 21967
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8949151039123535,
      "learning_rate": 2.9328809751949137e-06,
      "loss": 0.4258,
      "step": 21968
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.817379355430603,
      "learning_rate": 2.9320910415265935e-06,
      "loss": 0.3286,
      "step": 21969
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.882151484489441,
      "learning_rate": 2.9313011959765494e-06,
      "loss": 0.2709,
      "step": 21970
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.1021358966827393,
      "learning_rate": 2.9305114385546274e-06,
      "loss": 0.3328,
      "step": 21971
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.995863914489746,
      "learning_rate": 2.9297217692706683e-06,
      "loss": 0.3961,
      "step": 21972
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5760469436645508,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.3108,
      "step": 21973
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.272700071334839,
      "learning_rate": 2.9281426951560387e-06,
      "loss": 0.3288,
      "step": 21974
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.1348507404327393,
      "learning_rate": 2.9273532903450485e-06,
      "loss": 0.2731,
      "step": 21975
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0050039291381836,
      "learning_rate": 2.9265639737114028e-06,
      "loss": 0.3266,
      "step": 21976
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.3350558280944824,
      "learning_rate": 2.925774745264939e-06,
      "loss": 0.2855,
      "step": 21977
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9684739112854004,
      "learning_rate": 2.9249856050154936e-06,
      "loss": 0.2561,
      "step": 21978
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8148961067199707,
      "learning_rate": 2.9241965529729098e-06,
      "loss": 0.2633,
      "step": 21979
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0179603099823,
      "learning_rate": 2.9234075891470247e-06,
      "loss": 0.3353,
      "step": 21980
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.883992314338684,
      "learning_rate": 2.9226187135476735e-06,
      "loss": 0.2782,
      "step": 21981
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.796999216079712,
      "learning_rate": 2.921829926184687e-06,
      "loss": 0.3788,
      "step": 21982
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.0260469913482666,
      "learning_rate": 2.9210412270679077e-06,
      "loss": 0.3306,
      "step": 21983
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7586884498596191,
      "learning_rate": 2.9202526162071642e-06,
      "loss": 0.3223,
      "step": 21984
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.814638614654541,
      "learning_rate": 2.9194640936122852e-06,
      "loss": 0.2616,
      "step": 21985
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8994451761245728,
      "learning_rate": 2.9186756592931086e-06,
      "loss": 0.2875,
      "step": 21986
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.765250325202942,
      "learning_rate": 2.91788731325946e-06,
      "loss": 0.2859,
      "step": 21987
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.618424654006958,
      "learning_rate": 2.9170990555211655e-06,
      "loss": 0.3217,
      "step": 21988
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5905487537384033,
      "learning_rate": 2.9163108860880586e-06,
      "loss": 0.2116,
      "step": 21989
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.1201088428497314,
      "learning_rate": 2.9155228049699624e-06,
      "loss": 0.2597,
      "step": 21990
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.259852647781372,
      "learning_rate": 2.9147348121766994e-06,
      "loss": 0.3624,
      "step": 21991
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.9657553434371948,
      "learning_rate": 2.9139469077180995e-06,
      "loss": 0.3564,
      "step": 21992
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8232200145721436,
      "learning_rate": 2.9131590916039843e-06,
      "loss": 0.295,
      "step": 21993
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.259521245956421,
      "learning_rate": 2.912371363844174e-06,
      "loss": 0.3283,
      "step": 21994
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.2279934883117676,
      "learning_rate": 2.9115837244484866e-06,
      "loss": 0.3395,
      "step": 21995
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.8588424921035767,
      "learning_rate": 2.910796173426749e-06,
      "loss": 0.3617,
      "step": 21996
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9495971202850342,
      "learning_rate": 2.9100087107887763e-06,
      "loss": 0.3179,
      "step": 21997
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9712183475494385,
      "learning_rate": 2.9092213365443833e-06,
      "loss": 0.2874,
      "step": 21998
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.693562626838684,
      "learning_rate": 2.9084340507033927e-06,
      "loss": 0.3017,
      "step": 21999
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7934460639953613,
      "learning_rate": 2.907646853275615e-06,
      "loss": 0.2533,
      "step": 22000
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7270033359527588,
      "learning_rate": 2.9068597442708648e-06,
      "loss": 0.3079,
      "step": 22001
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9342968463897705,
      "learning_rate": 2.9060727236989583e-06,
      "loss": 0.2771,
      "step": 22002
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.16796612739563,
      "learning_rate": 2.9052857915697065e-06,
      "loss": 0.306,
      "step": 22003
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8196465969085693,
      "learning_rate": 2.9044989478929163e-06,
      "loss": 0.2677,
      "step": 22004
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.819750428199768,
      "learning_rate": 2.903712192678404e-06,
      "loss": 0.2935,
      "step": 22005
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6420975923538208,
      "learning_rate": 2.9029255259359747e-06,
      "loss": 0.2489,
      "step": 22006
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1807029247283936,
      "learning_rate": 2.902138947675438e-06,
      "loss": 0.2631,
      "step": 22007
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.934807300567627,
      "learning_rate": 2.9013524579065943e-06,
      "loss": 0.2728,
      "step": 22008
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.07568621635437,
      "learning_rate": 2.900566056639258e-06,
      "loss": 0.3123,
      "step": 22009
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5491174459457397,
      "learning_rate": 2.8997797438832297e-06,
      "loss": 0.2855,
      "step": 22010
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.631412386894226,
      "learning_rate": 2.898993519648309e-06,
      "loss": 0.2929,
      "step": 22011
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9455939531326294,
      "learning_rate": 2.8982073839443035e-06,
      "loss": 0.3751,
      "step": 22012
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1699697971343994,
      "learning_rate": 2.8974213367810134e-06,
      "loss": 0.4018,
      "step": 22013
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0102851390838623,
      "learning_rate": 2.896635378168233e-06,
      "loss": 0.3566,
      "step": 22014
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.075205087661743,
      "learning_rate": 2.8958495081157702e-06,
      "loss": 0.3582,
      "step": 22015
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7460061311721802,
      "learning_rate": 2.895063726633417e-06,
      "loss": 0.2934,
      "step": 22016
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7616056203842163,
      "learning_rate": 2.894278033730967e-06,
      "loss": 0.2961,
      "step": 22017
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6875683069229126,
      "learning_rate": 2.893492429418224e-06,
      "loss": 0.2765,
      "step": 22018
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1260344982147217,
      "learning_rate": 2.892706913704979e-06,
      "loss": 0.3024,
      "step": 22019
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9697831869125366,
      "learning_rate": 2.8919214866010237e-06,
      "loss": 0.3243,
      "step": 22020
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6729018688201904,
      "learning_rate": 2.891136148116148e-06,
      "loss": 0.3548,
      "step": 22021
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1038074493408203,
      "learning_rate": 2.890350898260149e-06,
      "loss": 0.3722,
      "step": 22022
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.177426338195801,
      "learning_rate": 2.8895657370428145e-06,
      "loss": 0.3436,
      "step": 22023
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.206301689147949,
      "learning_rate": 2.8887806644739293e-06,
      "loss": 0.3341,
      "step": 22024
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8533233404159546,
      "learning_rate": 2.887995680563288e-06,
      "loss": 0.2591,
      "step": 22025
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8666677474975586,
      "learning_rate": 2.8872107853206732e-06,
      "loss": 0.3083,
      "step": 22026
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6284022331237793,
      "learning_rate": 2.886425978755869e-06,
      "loss": 0.2732,
      "step": 22027
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0915348529815674,
      "learning_rate": 2.8856412608786642e-06,
      "loss": 0.3043,
      "step": 22028
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2315196990966797,
      "learning_rate": 2.884856631698841e-06,
      "loss": 0.3558,
      "step": 22029
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9986506700515747,
      "learning_rate": 2.884072091226178e-06,
      "loss": 0.3339,
      "step": 22030
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.137627601623535,
      "learning_rate": 2.8832876394704613e-06,
      "loss": 0.3563,
      "step": 22031
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6356199979782104,
      "learning_rate": 2.8825032764414684e-06,
      "loss": 0.2809,
      "step": 22032
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.252971649169922,
      "learning_rate": 2.8817190021489795e-06,
      "loss": 0.2766,
      "step": 22033
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.278188467025757,
      "learning_rate": 2.8809348166027684e-06,
      "loss": 0.3431,
      "step": 22034
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8392740488052368,
      "learning_rate": 2.8801507198126176e-06,
      "loss": 0.3203,
      "step": 22035
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9184209108352661,
      "learning_rate": 2.8793667117883005e-06,
      "loss": 0.3783,
      "step": 22036
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.66342830657959,
      "learning_rate": 2.878582792539588e-06,
      "loss": 0.3538,
      "step": 22037
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.4268345832824707,
      "learning_rate": 2.87779896207626e-06,
      "loss": 0.3279,
      "step": 22038
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9297691583633423,
      "learning_rate": 2.877015220408086e-06,
      "loss": 0.3342,
      "step": 22039
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8022503852844238,
      "learning_rate": 2.8762315675448327e-06,
      "loss": 0.3442,
      "step": 22040
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8734227418899536,
      "learning_rate": 2.8754480034962772e-06,
      "loss": 0.3479,
      "step": 22041
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.3758864402770996,
      "learning_rate": 2.8746645282721863e-06,
      "loss": 0.3735,
      "step": 22042
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9729102849960327,
      "learning_rate": 2.8738811418823244e-06,
      "loss": 0.2961,
      "step": 22043
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2038824558258057,
      "learning_rate": 2.8730978443364634e-06,
      "loss": 0.3493,
      "step": 22044
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.329514265060425,
      "learning_rate": 2.8723146356443666e-06,
      "loss": 0.2376,
      "step": 22045
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.210782766342163,
      "learning_rate": 2.8715315158157986e-06,
      "loss": 0.4455,
      "step": 22046
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6000198125839233,
      "learning_rate": 2.8707484848605194e-06,
      "loss": 0.2601,
      "step": 22047
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.4320316314697266,
      "learning_rate": 2.8699655427882978e-06,
      "loss": 0.4022,
      "step": 22048
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.041149139404297,
      "learning_rate": 2.8691826896088924e-06,
      "loss": 0.3358,
      "step": 22049
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7548117637634277,
      "learning_rate": 2.8683999253320593e-06,
      "loss": 0.2361,
      "step": 22050
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6984103918075562,
      "learning_rate": 2.8676172499675645e-06,
      "loss": 0.3065,
      "step": 22051
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9670839309692383,
      "learning_rate": 2.8668346635251633e-06,
      "loss": 0.4119,
      "step": 22052
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6547870635986328,
      "learning_rate": 2.866052166014607e-06,
      "loss": 0.277,
      "step": 22053
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8826630115509033,
      "learning_rate": 2.8652697574456602e-06,
      "loss": 0.3529,
      "step": 22054
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8907155990600586,
      "learning_rate": 2.864487437828073e-06,
      "loss": 0.2736,
      "step": 22055
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0388906002044678,
      "learning_rate": 2.8637052071715955e-06,
      "loss": 0.2577,
      "step": 22056
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2680037021636963,
      "learning_rate": 2.8629230654859876e-06,
      "loss": 0.266,
      "step": 22057
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0175716876983643,
      "learning_rate": 2.8621410127809955e-06,
      "loss": 0.2251,
      "step": 22058
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8176954984664917,
      "learning_rate": 2.861359049066371e-06,
      "loss": 0.3252,
      "step": 22059
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.295219898223877,
      "learning_rate": 2.8605771743518606e-06,
      "loss": 0.3051,
      "step": 22060
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8518903255462646,
      "learning_rate": 2.859795388647216e-06,
      "loss": 0.3531,
      "step": 22061
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7610310316085815,
      "learning_rate": 2.8590136919621836e-06,
      "loss": 0.2836,
      "step": 22062
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1924586296081543,
      "learning_rate": 2.858232084306504e-06,
      "loss": 0.2393,
      "step": 22063
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8871853351593018,
      "learning_rate": 2.8574505656899277e-06,
      "loss": 0.3322,
      "step": 22064
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0710177421569824,
      "learning_rate": 2.8566691361221967e-06,
      "loss": 0.3227,
      "step": 22065
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0531532764434814,
      "learning_rate": 2.8558877956130506e-06,
      "loss": 0.2613,
      "step": 22066
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8239630460739136,
      "learning_rate": 2.855106544172235e-06,
      "loss": 0.2955,
      "step": 22067
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5800901651382446,
      "learning_rate": 2.8543253818094874e-06,
      "loss": 0.3196,
      "step": 22068
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0571136474609375,
      "learning_rate": 2.853544308534545e-06,
      "loss": 0.2975,
      "step": 22069
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1605725288391113,
      "learning_rate": 2.8527633243571506e-06,
      "loss": 0.2975,
      "step": 22070
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9298568964004517,
      "learning_rate": 2.8519824292870392e-06,
      "loss": 0.2942,
      "step": 22071
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6621763706207275,
      "learning_rate": 2.8512016233339444e-06,
      "loss": 0.3442,
      "step": 22072
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.698747992515564,
      "learning_rate": 2.8504209065076007e-06,
      "loss": 0.3438,
      "step": 22073
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.0069355964660645,
      "learning_rate": 2.849640278817746e-06,
      "loss": 0.3876,
      "step": 22074
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1986472606658936,
      "learning_rate": 2.848859740274108e-06,
      "loss": 0.3543,
      "step": 22075
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.991504430770874,
      "learning_rate": 2.8480792908864186e-06,
      "loss": 0.3823,
      "step": 22076
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.987453579902649,
      "learning_rate": 2.8472989306644105e-06,
      "loss": 0.2767,
      "step": 22077
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8296372890472412,
      "learning_rate": 2.846518659617813e-06,
      "loss": 0.3752,
      "step": 22078
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.879597783088684,
      "learning_rate": 2.8457384777563477e-06,
      "loss": 0.2888,
      "step": 22079
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.2890775203704834,
      "learning_rate": 2.84495838508975e-06,
      "loss": 0.2842,
      "step": 22080
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8545506000518799,
      "learning_rate": 2.8441783816277413e-06,
      "loss": 0.3116,
      "step": 22081
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1230733394622803,
      "learning_rate": 2.843398467380044e-06,
      "loss": 0.3587,
      "step": 22082
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5401734113693237,
      "learning_rate": 2.8426186423563863e-06,
      "loss": 0.3258,
      "step": 22083
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.462104558944702,
      "learning_rate": 2.841838906566489e-06,
      "loss": 0.3448,
      "step": 22084
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8562731742858887,
      "learning_rate": 2.841059260020073e-06,
      "loss": 0.33,
      "step": 22085
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.274751663208008,
      "learning_rate": 2.8402797027268546e-06,
      "loss": 0.2979,
      "step": 22086
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.972316026687622,
      "learning_rate": 2.8395002346965595e-06,
      "loss": 0.3095,
      "step": 22087
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.3902008533477783,
      "learning_rate": 2.8387208559389034e-06,
      "loss": 0.3068,
      "step": 22088
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.8553075790405273,
      "learning_rate": 2.837941566463599e-06,
      "loss": 0.2919,
      "step": 22089
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6171027421951294,
      "learning_rate": 2.8371623662803673e-06,
      "loss": 0.3665,
      "step": 22090
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.997519850730896,
      "learning_rate": 2.8363832553989224e-06,
      "loss": 0.386,
      "step": 22091
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5449119806289673,
      "learning_rate": 2.8356042338289712e-06,
      "loss": 0.2952,
      "step": 22092
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9736595153808594,
      "learning_rate": 2.8348253015802363e-06,
      "loss": 0.2984,
      "step": 22093
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.259948492050171,
      "learning_rate": 2.8340464586624218e-06,
      "loss": 0.3958,
      "step": 22094
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1464643478393555,
      "learning_rate": 2.8332677050852375e-06,
      "loss": 0.2933,
      "step": 22095
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.094864845275879,
      "learning_rate": 2.8324890408583985e-06,
      "loss": 0.362,
      "step": 22096
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7331187725067139,
      "learning_rate": 2.831710465991607e-06,
      "loss": 0.288,
      "step": 22097
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0191752910614014,
      "learning_rate": 2.8309319804945735e-06,
      "loss": 0.2862,
      "step": 22098
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9656825065612793,
      "learning_rate": 2.8301535843769976e-06,
      "loss": 0.3144,
      "step": 22099
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4023363590240479,
      "learning_rate": 2.8293752776485917e-06,
      "loss": 0.2139,
      "step": 22100
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8412537574768066,
      "learning_rate": 2.828597060319056e-06,
      "loss": 0.309,
      "step": 22101
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9136559963226318,
      "learning_rate": 2.827818932398089e-06,
      "loss": 0.3321,
      "step": 22102
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.907543659210205,
      "learning_rate": 2.8270408938953975e-06,
      "loss": 0.3141,
      "step": 22103
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.240086317062378,
      "learning_rate": 2.8262629448206803e-06,
      "loss": 0.3509,
      "step": 22104
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.3688390254974365,
      "learning_rate": 2.8254850851836324e-06,
      "loss": 0.3323,
      "step": 22105
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7864508628845215,
      "learning_rate": 2.8247073149939575e-06,
      "loss": 0.2971,
      "step": 22106
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1945581436157227,
      "learning_rate": 2.823929634261351e-06,
      "loss": 0.3163,
      "step": 22107
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.4442784786224365,
      "learning_rate": 2.823152042995503e-06,
      "loss": 0.3164,
      "step": 22108
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9349243640899658,
      "learning_rate": 2.822374541206114e-06,
      "loss": 0.3276,
      "step": 22109
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0659735202789307,
      "learning_rate": 2.821597128902878e-06,
      "loss": 0.3281,
      "step": 22110
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.045602321624756,
      "learning_rate": 2.8208198060954837e-06,
      "loss": 0.266,
      "step": 22111
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8157532215118408,
      "learning_rate": 2.820042572793621e-06,
      "loss": 0.2723,
      "step": 22112
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.670589804649353,
      "learning_rate": 2.8192654290069843e-06,
      "loss": 0.3775,
      "step": 22113
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.454991102218628,
      "learning_rate": 2.8184883747452617e-06,
      "loss": 0.3188,
      "step": 22114
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.3099606037139893,
      "learning_rate": 2.817711410018137e-06,
      "loss": 0.4323,
      "step": 22115
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6980575323104858,
      "learning_rate": 2.8169345348353026e-06,
      "loss": 0.3194,
      "step": 22116
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8627076148986816,
      "learning_rate": 2.8161577492064416e-06,
      "loss": 0.3341,
      "step": 22117
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6816868782043457,
      "learning_rate": 2.815381053141235e-06,
      "loss": 0.3685,
      "step": 22118
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.076535940170288,
      "learning_rate": 2.8146044466493727e-06,
      "loss": 0.3111,
      "step": 22119
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8177216053009033,
      "learning_rate": 2.8138279297405347e-06,
      "loss": 0.3917,
      "step": 22120
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.538146734237671,
      "learning_rate": 2.8130515024243964e-06,
      "loss": 0.4354,
      "step": 22121
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8172792196273804,
      "learning_rate": 2.812275164710646e-06,
      "loss": 0.3078,
      "step": 22122
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8408563137054443,
      "learning_rate": 2.81149891660896e-06,
      "loss": 0.2713,
      "step": 22123
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9996651411056519,
      "learning_rate": 2.8107227581290144e-06,
      "loss": 0.3259,
      "step": 22124
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.900905728340149,
      "learning_rate": 2.809946689280483e-06,
      "loss": 0.3891,
      "step": 22125
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5033079385757446,
      "learning_rate": 2.809170710073048e-06,
      "loss": 0.3002,
      "step": 22126
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8814424276351929,
      "learning_rate": 2.8083948205163813e-06,
      "loss": 0.3028,
      "step": 22127
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1931607723236084,
      "learning_rate": 2.8076190206201526e-06,
      "loss": 0.2552,
      "step": 22128
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.571418285369873,
      "learning_rate": 2.8068433103940407e-06,
      "loss": 0.2833,
      "step": 22129
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6196951866149902,
      "learning_rate": 2.8060676898477124e-06,
      "loss": 0.302,
      "step": 22130
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8248021602630615,
      "learning_rate": 2.8052921589908356e-06,
      "loss": 0.3109,
      "step": 22131
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9290282726287842,
      "learning_rate": 2.8045167178330844e-06,
      "loss": 0.291,
      "step": 22132
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7676615715026855,
      "learning_rate": 2.803741366384124e-06,
      "loss": 0.3355,
      "step": 22133
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.965635061264038,
      "learning_rate": 2.8029661046536184e-06,
      "loss": 0.343,
      "step": 22134
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.88848078250885,
      "learning_rate": 2.8021909326512385e-06,
      "loss": 0.2647,
      "step": 22135
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0714938640594482,
      "learning_rate": 2.801415850386646e-06,
      "loss": 0.2575,
      "step": 22136
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.482496976852417,
      "learning_rate": 2.8006408578695045e-06,
      "loss": 0.333,
      "step": 22137
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7391620874404907,
      "learning_rate": 2.799865955109471e-06,
      "loss": 0.3552,
      "step": 22138
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8316917419433594,
      "learning_rate": 2.7990911421162157e-06,
      "loss": 0.3371,
      "step": 22139
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.185904026031494,
      "learning_rate": 2.7983164188993927e-06,
      "loss": 0.307,
      "step": 22140
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5433428287506104,
      "learning_rate": 2.7975417854686593e-06,
      "loss": 0.2136,
      "step": 22141
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9630917310714722,
      "learning_rate": 2.79676724183368e-06,
      "loss": 0.3903,
      "step": 22142
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6212023496627808,
      "learning_rate": 2.795992788004106e-06,
      "loss": 0.2819,
      "step": 22143
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.006685495376587,
      "learning_rate": 2.7952184239895907e-06,
      "loss": 0.3103,
      "step": 22144
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9472160339355469,
      "learning_rate": 2.7944441497997954e-06,
      "loss": 0.3081,
      "step": 22145
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.198279619216919,
      "learning_rate": 2.793669965444369e-06,
      "loss": 0.3058,
      "step": 22146
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1323394775390625,
      "learning_rate": 2.7928958709329602e-06,
      "loss": 0.3492,
      "step": 22147
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0100326538085938,
      "learning_rate": 2.792121866275227e-06,
      "loss": 0.2985,
      "step": 22148
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8470103740692139,
      "learning_rate": 2.7913479514808194e-06,
      "loss": 0.2876,
      "step": 22149
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.943474531173706,
      "learning_rate": 2.7905741265593767e-06,
      "loss": 0.2871,
      "step": 22150
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.836946725845337,
      "learning_rate": 2.789800391520554e-06,
      "loss": 0.2986,
      "step": 22151
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.899900197982788,
      "learning_rate": 2.7890267463739974e-06,
      "loss": 0.3158,
      "step": 22152
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.539760708808899,
      "learning_rate": 2.7882531911293476e-06,
      "loss": 0.3167,
      "step": 22153
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.855026364326477,
      "learning_rate": 2.787479725796255e-06,
      "loss": 0.2951,
      "step": 22154
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7561163902282715,
      "learning_rate": 2.7867063503843593e-06,
      "loss": 0.3023,
      "step": 22155
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5508697032928467,
      "learning_rate": 2.7859330649033013e-06,
      "loss": 0.2449,
      "step": 22156
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.067209482192993,
      "learning_rate": 2.7851598693627245e-06,
      "loss": 0.3204,
      "step": 22157
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6773388385772705,
      "learning_rate": 2.7843867637722687e-06,
      "loss": 0.2695,
      "step": 22158
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8073278665542603,
      "learning_rate": 2.7836137481415716e-06,
      "loss": 0.3883,
      "step": 22159
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5673507452011108,
      "learning_rate": 2.782840822480267e-06,
      "loss": 0.2483,
      "step": 22160
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7145761251449585,
      "learning_rate": 2.7820679867979983e-06,
      "loss": 0.3043,
      "step": 22161
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.2444589138031006,
      "learning_rate": 2.781295241104397e-06,
      "loss": 0.3781,
      "step": 22162
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0559000968933105,
      "learning_rate": 2.7805225854090944e-06,
      "loss": 0.2819,
      "step": 22163
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.183115243911743,
      "learning_rate": 2.7797500197217285e-06,
      "loss": 0.3066,
      "step": 22164
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1874232292175293,
      "learning_rate": 2.7789775440519306e-06,
      "loss": 0.3382,
      "step": 22165
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.113861083984375,
      "learning_rate": 2.7782051584093263e-06,
      "loss": 0.3785,
      "step": 22166
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.3351125717163086,
      "learning_rate": 2.7774328628035516e-06,
      "loss": 0.3538,
      "step": 22167
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.443556070327759,
      "learning_rate": 2.776660657244232e-06,
      "loss": 0.3868,
      "step": 22168
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.604231834411621,
      "learning_rate": 2.775888541740993e-06,
      "loss": 0.3126,
      "step": 22169
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.296205759048462,
      "learning_rate": 2.7751165163034666e-06,
      "loss": 0.3217,
      "step": 22170
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9519379138946533,
      "learning_rate": 2.7743445809412728e-06,
      "loss": 0.3218,
      "step": 22171
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8314085006713867,
      "learning_rate": 2.7735727356640386e-06,
      "loss": 0.3221,
      "step": 22172
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9098076820373535,
      "learning_rate": 2.7728009804813816e-06,
      "loss": 0.3551,
      "step": 22173
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.145261287689209,
      "learning_rate": 2.77202931540293e-06,
      "loss": 0.3872,
      "step": 22174
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.090475082397461,
      "learning_rate": 2.7712577404383023e-06,
      "loss": 0.3894,
      "step": 22175
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.1179559230804443,
      "learning_rate": 2.7704862555971144e-06,
      "loss": 0.3944,
      "step": 22176
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8978580236434937,
      "learning_rate": 2.769714860888991e-06,
      "loss": 0.3215,
      "step": 22177
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.840333104133606,
      "learning_rate": 2.7689435563235466e-06,
      "loss": 0.3068,
      "step": 22178
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8252596855163574,
      "learning_rate": 2.7681723419103935e-06,
      "loss": 0.4006,
      "step": 22179
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7727874517440796,
      "learning_rate": 2.7674012176591535e-06,
      "loss": 0.3469,
      "step": 22180
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.793511986732483,
      "learning_rate": 2.7666301835794373e-06,
      "loss": 0.2819,
      "step": 22181
    },
    {
      "epoch": 2.29,
      "grad_norm": 4.038781642913818,
      "learning_rate": 2.7658592396808537e-06,
      "loss": 0.3327,
      "step": 22182
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8708417415618896,
      "learning_rate": 2.7650883859730216e-06,
      "loss": 0.2347,
      "step": 22183
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.0198206901550293,
      "learning_rate": 2.764317622465549e-06,
      "loss": 0.3293,
      "step": 22184
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.473087787628174,
      "learning_rate": 2.7635469491680435e-06,
      "loss": 0.3335,
      "step": 22185
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8472565412521362,
      "learning_rate": 2.762776366090111e-06,
      "loss": 0.3414,
      "step": 22186
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8327183723449707,
      "learning_rate": 2.762005873241366e-06,
      "loss": 0.2831,
      "step": 22187
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.2603578567504883,
      "learning_rate": 2.7612354706314093e-06,
      "loss": 0.2925,
      "step": 22188
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.042438507080078,
      "learning_rate": 2.760465158269845e-06,
      "loss": 0.2169,
      "step": 22189
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.200615167617798,
      "learning_rate": 2.7596949361662805e-06,
      "loss": 0.2951,
      "step": 22190
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.046086311340332,
      "learning_rate": 2.7589248043303184e-06,
      "loss": 0.3196,
      "step": 22191
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8093876838684082,
      "learning_rate": 2.7581547627715544e-06,
      "loss": 0.3104,
      "step": 22192
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.4208927154541016,
      "learning_rate": 2.757384811499597e-06,
      "loss": 0.3811,
      "step": 22193
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.279493570327759,
      "learning_rate": 2.7566149505240413e-06,
      "loss": 0.3239,
      "step": 22194
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1905479431152344,
      "learning_rate": 2.7558451798544818e-06,
      "loss": 0.397,
      "step": 22195
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.024324655532837,
      "learning_rate": 2.7550754995005235e-06,
      "loss": 0.3189,
      "step": 22196
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7632430791854858,
      "learning_rate": 2.754305909471758e-06,
      "loss": 0.2988,
      "step": 22197
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.021703004837036,
      "learning_rate": 2.7535364097777806e-06,
      "loss": 0.3216,
      "step": 22198
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7602057456970215,
      "learning_rate": 2.7527670004281805e-06,
      "loss": 0.3327,
      "step": 22199
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8323619365692139,
      "learning_rate": 2.7519976814325577e-06,
      "loss": 0.2904,
      "step": 22200
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4061745405197144,
      "learning_rate": 2.7512284528005007e-06,
      "loss": 0.2455,
      "step": 22201
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9034554958343506,
      "learning_rate": 2.750459314541597e-06,
      "loss": 0.3079,
      "step": 22202
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9213794469833374,
      "learning_rate": 2.7496902666654403e-06,
      "loss": 0.3092,
      "step": 22203
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7120492458343506,
      "learning_rate": 2.748921309181616e-06,
      "loss": 0.3159,
      "step": 22204
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6866395473480225,
      "learning_rate": 2.7481524420997083e-06,
      "loss": 0.2989,
      "step": 22205
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.33902645111084,
      "learning_rate": 2.7473836654293094e-06,
      "loss": 0.3197,
      "step": 22206
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.324108362197876,
      "learning_rate": 2.746614979180001e-06,
      "loss": 0.2805,
      "step": 22207
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.054009199142456,
      "learning_rate": 2.7458463833613623e-06,
      "loss": 0.3907,
      "step": 22208
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.969314694404602,
      "learning_rate": 2.7450778779829825e-06,
      "loss": 0.3892,
      "step": 22209
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.703671932220459,
      "learning_rate": 2.7443094630544408e-06,
      "loss": 0.237,
      "step": 22210
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8902522325515747,
      "learning_rate": 2.743541138585315e-06,
      "loss": 0.3911,
      "step": 22211
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6365941762924194,
      "learning_rate": 2.7427729045851835e-06,
      "loss": 0.2642,
      "step": 22212
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7762559652328491,
      "learning_rate": 2.742004761063628e-06,
      "loss": 0.3327,
      "step": 22213
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.2287116050720215,
      "learning_rate": 2.741236708030225e-06,
      "loss": 0.3085,
      "step": 22214
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6371042728424072,
      "learning_rate": 2.7404687454945443e-06,
      "loss": 0.3306,
      "step": 22215
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.031233072280884,
      "learning_rate": 2.7397008734661666e-06,
      "loss": 0.3354,
      "step": 22216
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.963435411453247,
      "learning_rate": 2.7389330919546654e-06,
      "loss": 0.3439,
      "step": 22217
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0184872150421143,
      "learning_rate": 2.7381654009696056e-06,
      "loss": 0.338,
      "step": 22218
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8437304496765137,
      "learning_rate": 2.737397800520568e-06,
      "loss": 0.3518,
      "step": 22219
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9231024980545044,
      "learning_rate": 2.736630290617117e-06,
      "loss": 0.3195,
      "step": 22220
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7672830820083618,
      "learning_rate": 2.7358628712688195e-06,
      "loss": 0.3061,
      "step": 22221
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.069047212600708,
      "learning_rate": 2.7350955424852487e-06,
      "loss": 0.3659,
      "step": 22222
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.4914498329162598,
      "learning_rate": 2.7343283042759693e-06,
      "loss": 0.4541,
      "step": 22223
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8974322080612183,
      "learning_rate": 2.733561156650546e-06,
      "loss": 0.3372,
      "step": 22224
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1193225383758545,
      "learning_rate": 2.7327940996185397e-06,
      "loss": 0.3563,
      "step": 22225
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.472078561782837,
      "learning_rate": 2.7320271331895197e-06,
      "loss": 0.3597,
      "step": 22226
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.960337519645691,
      "learning_rate": 2.731260257373046e-06,
      "loss": 0.2728,
      "step": 22227
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8326332569122314,
      "learning_rate": 2.7304934721786757e-06,
      "loss": 0.3254,
      "step": 22228
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.970099687576294,
      "learning_rate": 2.7297267776159754e-06,
      "loss": 0.3176,
      "step": 22229
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7162965536117554,
      "learning_rate": 2.728960173694499e-06,
      "loss": 0.3696,
      "step": 22230
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9725499153137207,
      "learning_rate": 2.7281936604238035e-06,
      "loss": 0.3732,
      "step": 22231
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.830864429473877,
      "learning_rate": 2.7274272378134493e-06,
      "loss": 0.3126,
      "step": 22232
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.09328031539917,
      "learning_rate": 2.7266609058729886e-06,
      "loss": 0.3689,
      "step": 22233
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6673712730407715,
      "learning_rate": 2.7258946646119743e-06,
      "loss": 0.2354,
      "step": 22234
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.840322494506836,
      "learning_rate": 2.725128514039964e-06,
      "loss": 0.3192,
      "step": 22235
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.596047282218933,
      "learning_rate": 2.724362454166508e-06,
      "loss": 0.3141,
      "step": 22236
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.2490813732147217,
      "learning_rate": 2.7235964850011553e-06,
      "loss": 0.3615,
      "step": 22237
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4688481092453003,
      "learning_rate": 2.7228306065534537e-06,
      "loss": 0.2165,
      "step": 22238
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7251168489456177,
      "learning_rate": 2.7220648188329568e-06,
      "loss": 0.269,
      "step": 22239
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.037820816040039,
      "learning_rate": 2.721299121849209e-06,
      "loss": 0.2899,
      "step": 22240
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6830980777740479,
      "learning_rate": 2.7205335156117553e-06,
      "loss": 0.2357,
      "step": 22241
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.233832359313965,
      "learning_rate": 2.719768000130145e-06,
      "loss": 0.2759,
      "step": 22242
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.40816330909729,
      "learning_rate": 2.7190025754139194e-06,
      "loss": 0.4315,
      "step": 22243
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.323011636734009,
      "learning_rate": 2.718237241472619e-06,
      "loss": 0.3497,
      "step": 22244
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.114476442337036,
      "learning_rate": 2.7174719983157905e-06,
      "loss": 0.3688,
      "step": 22245
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.751619577407837,
      "learning_rate": 2.716706845952972e-06,
      "loss": 0.2557,
      "step": 22246
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9208662509918213,
      "learning_rate": 2.715941784393703e-06,
      "loss": 0.365,
      "step": 22247
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.253788471221924,
      "learning_rate": 2.715176813647519e-06,
      "loss": 0.2834,
      "step": 22248
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8222039937973022,
      "learning_rate": 2.7144119337239626e-06,
      "loss": 0.334,
      "step": 22249
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.544272541999817,
      "learning_rate": 2.7136471446325672e-06,
      "loss": 0.3061,
      "step": 22250
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.557999849319458,
      "learning_rate": 2.712882446382865e-06,
      "loss": 0.3418,
      "step": 22251
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.03688383102417,
      "learning_rate": 2.712117838984395e-06,
      "loss": 0.2999,
      "step": 22252
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0021615028381348,
      "learning_rate": 2.711353322446688e-06,
      "loss": 0.3528,
      "step": 22253
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9332911968231201,
      "learning_rate": 2.7105888967792715e-06,
      "loss": 0.2969,
      "step": 22254
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1129088401794434,
      "learning_rate": 2.709824561991682e-06,
      "loss": 0.3219,
      "step": 22255
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8138433694839478,
      "learning_rate": 2.7090603180934462e-06,
      "loss": 0.3527,
      "step": 22256
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1624486446380615,
      "learning_rate": 2.7082961650940888e-06,
      "loss": 0.3308,
      "step": 22257
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7103567123413086,
      "learning_rate": 2.707532103003142e-06,
      "loss": 0.3042,
      "step": 22258
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8832290172576904,
      "learning_rate": 2.706768131830131e-06,
      "loss": 0.3604,
      "step": 22259
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9043577909469604,
      "learning_rate": 2.7060042515845784e-06,
      "loss": 0.4026,
      "step": 22260
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8428635597229004,
      "learning_rate": 2.7052404622760065e-06,
      "loss": 0.3026,
      "step": 22261
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0532188415527344,
      "learning_rate": 2.704476763913941e-06,
      "loss": 0.3212,
      "step": 22262
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8765454292297363,
      "learning_rate": 2.7037131565079033e-06,
      "loss": 0.3324,
      "step": 22263
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.764877438545227,
      "learning_rate": 2.7029496400674095e-06,
      "loss": 0.3365,
      "step": 22264
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7927286624908447,
      "learning_rate": 2.702186214601984e-06,
      "loss": 0.3267,
      "step": 22265
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.127997875213623,
      "learning_rate": 2.701422880121142e-06,
      "loss": 0.3517,
      "step": 22266
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.87723708152771,
      "learning_rate": 2.7006596366343963e-06,
      "loss": 0.3609,
      "step": 22267
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7582170963287354,
      "learning_rate": 2.699896484151271e-06,
      "loss": 0.3375,
      "step": 22268
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5308984518051147,
      "learning_rate": 2.699133422681276e-06,
      "loss": 0.2896,
      "step": 22269
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0277159214019775,
      "learning_rate": 2.6983704522339215e-06,
      "loss": 0.3212,
      "step": 22270
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7837929725646973,
      "learning_rate": 2.697607572818727e-06,
      "loss": 0.3074,
      "step": 22271
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.511249303817749,
      "learning_rate": 2.6968447844451984e-06,
      "loss": 0.3904,
      "step": 22272
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.111691951751709,
      "learning_rate": 2.6960820871228476e-06,
      "loss": 0.3406,
      "step": 22273
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.8591798543930054,
      "learning_rate": 2.6953194808611803e-06,
      "loss": 0.2821,
      "step": 22274
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0030524730682373,
      "learning_rate": 2.694556965669709e-06,
      "loss": 0.3016,
      "step": 22275
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.250976085662842,
      "learning_rate": 2.693794541557938e-06,
      "loss": 0.2619,
      "step": 22276
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.810550332069397,
      "learning_rate": 2.6930322085353712e-06,
      "loss": 0.2974,
      "step": 22277
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6542960405349731,
      "learning_rate": 2.692269966611516e-06,
      "loss": 0.3028,
      "step": 22278
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1059367656707764,
      "learning_rate": 2.691507815795874e-06,
      "loss": 0.3233,
      "step": 22279
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.2096478939056396,
      "learning_rate": 2.6907457560979444e-06,
      "loss": 0.2884,
      "step": 22280
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.084296941757202,
      "learning_rate": 2.6899837875272338e-06,
      "loss": 0.3488,
      "step": 22281
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9406217336654663,
      "learning_rate": 2.68922191009324e-06,
      "loss": 0.3451,
      "step": 22282
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.397174119949341,
      "learning_rate": 2.6884601238054564e-06,
      "loss": 0.2879,
      "step": 22283
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.40407657623291,
      "learning_rate": 2.6876984286733886e-06,
      "loss": 0.4136,
      "step": 22284
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1219189167022705,
      "learning_rate": 2.6869368247065286e-06,
      "loss": 0.3546,
      "step": 22285
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9025981426239014,
      "learning_rate": 2.6861753119143718e-06,
      "loss": 0.2897,
      "step": 22286
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.429985761642456,
      "learning_rate": 2.6854138903064107e-06,
      "loss": 0.3567,
      "step": 22287
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7947665452957153,
      "learning_rate": 2.6846525598921424e-06,
      "loss": 0.2778,
      "step": 22288
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8169370889663696,
      "learning_rate": 2.6838913206810557e-06,
      "loss": 0.3702,
      "step": 22289
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.060335397720337,
      "learning_rate": 2.68313017268264e-06,
      "loss": 0.3538,
      "step": 22290
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6053698062896729,
      "learning_rate": 2.682369115906389e-06,
      "loss": 0.2684,
      "step": 22291
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7316471338272095,
      "learning_rate": 2.6816081503617885e-06,
      "loss": 0.3301,
      "step": 22292
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.985373854637146,
      "learning_rate": 2.680847276058324e-06,
      "loss": 0.2985,
      "step": 22293
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.742370367050171,
      "learning_rate": 2.680086493005486e-06,
      "loss": 0.2914,
      "step": 22294
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9275074005126953,
      "learning_rate": 2.6793258012127567e-06,
      "loss": 0.3384,
      "step": 22295
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0333070755004883,
      "learning_rate": 2.678565200689618e-06,
      "loss": 0.3966,
      "step": 22296
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.750469446182251,
      "learning_rate": 2.677804691445558e-06,
      "loss": 0.2882,
      "step": 22297
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8999143838882446,
      "learning_rate": 2.6770442734900547e-06,
      "loss": 0.312,
      "step": 22298
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.11145281791687,
      "learning_rate": 2.6762839468325894e-06,
      "loss": 0.3311,
      "step": 22299
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8582168817520142,
      "learning_rate": 2.6755237114826383e-06,
      "loss": 0.322,
      "step": 22300
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1548287868499756,
      "learning_rate": 2.6747635674496854e-06,
      "loss": 0.3865,
      "step": 22301
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9439573287963867,
      "learning_rate": 2.674003514743204e-06,
      "loss": 0.31,
      "step": 22302
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.082001209259033,
      "learning_rate": 2.6732435533726674e-06,
      "loss": 0.3157,
      "step": 22303
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.703298568725586,
      "learning_rate": 2.6724836833475563e-06,
      "loss": 0.3565,
      "step": 22304
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3798902034759521,
      "learning_rate": 2.6717239046773414e-06,
      "loss": 0.2816,
      "step": 22305
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9665288925170898,
      "learning_rate": 2.670964217371492e-06,
      "loss": 0.3731,
      "step": 22306
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5296151638031006,
      "learning_rate": 2.6702046214394873e-06,
      "loss": 0.269,
      "step": 22307
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.2489283084869385,
      "learning_rate": 2.669445116890791e-06,
      "loss": 0.3052,
      "step": 22308
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7294881343841553,
      "learning_rate": 2.6686857037348713e-06,
      "loss": 0.3685,
      "step": 22309
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7121473550796509,
      "learning_rate": 2.667926381981202e-06,
      "loss": 0.3053,
      "step": 22310
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.019075393676758,
      "learning_rate": 2.667167151639245e-06,
      "loss": 0.2766,
      "step": 22311
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.993753433227539,
      "learning_rate": 2.666408012718469e-06,
      "loss": 0.3765,
      "step": 22312
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.292353868484497,
      "learning_rate": 2.6656489652283334e-06,
      "loss": 0.31,
      "step": 22313
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5909323692321777,
      "learning_rate": 2.6648900091783082e-06,
      "loss": 0.3285,
      "step": 22314
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9172803163528442,
      "learning_rate": 2.6641311445778517e-06,
      "loss": 0.3111,
      "step": 22315
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7829809188842773,
      "learning_rate": 2.6633723714364236e-06,
      "loss": 0.2807,
      "step": 22316
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7217861413955688,
      "learning_rate": 2.662613689763488e-06,
      "loss": 0.2507,
      "step": 22317
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7038817405700684,
      "learning_rate": 2.661855099568501e-06,
      "loss": 0.2796,
      "step": 22318
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8935223817825317,
      "learning_rate": 2.6610966008609194e-06,
      "loss": 0.262,
      "step": 22319
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.427689552307129,
      "learning_rate": 2.6603381936502025e-06,
      "loss": 0.3398,
      "step": 22320
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.642037034034729,
      "learning_rate": 2.6595798779458047e-06,
      "loss": 0.3275,
      "step": 22321
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7942311763763428,
      "learning_rate": 2.6588216537571775e-06,
      "loss": 0.2738,
      "step": 22322
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.724450945854187,
      "learning_rate": 2.658063521093779e-06,
      "loss": 0.2666,
      "step": 22323
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9525256156921387,
      "learning_rate": 2.6573054799650577e-06,
      "loss": 0.3919,
      "step": 22324
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7449666261672974,
      "learning_rate": 2.6565475303804645e-06,
      "loss": 0.2804,
      "step": 22325
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8396950960159302,
      "learning_rate": 2.655789672349447e-06,
      "loss": 0.2895,
      "step": 22326
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0560455322265625,
      "learning_rate": 2.65503190588146e-06,
      "loss": 0.3019,
      "step": 22327
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.4602742195129395,
      "learning_rate": 2.654274230985946e-06,
      "loss": 0.3338,
      "step": 22328
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7983454465866089,
      "learning_rate": 2.6535166476723494e-06,
      "loss": 0.3147,
      "step": 22329
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9157624244689941,
      "learning_rate": 2.652759155950122e-06,
      "loss": 0.3155,
      "step": 22330
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7619078159332275,
      "learning_rate": 2.6520017558287026e-06,
      "loss": 0.3075,
      "step": 22331
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7652508020401,
      "learning_rate": 2.6512444473175337e-06,
      "loss": 0.3809,
      "step": 22332
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0563454627990723,
      "learning_rate": 2.65048723042606e-06,
      "loss": 0.3037,
      "step": 22333
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7144232988357544,
      "learning_rate": 2.6497301051637214e-06,
      "loss": 0.3184,
      "step": 22334
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6838968992233276,
      "learning_rate": 2.6489730715399532e-06,
      "loss": 0.2956,
      "step": 22335
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8189270496368408,
      "learning_rate": 2.6482161295642005e-06,
      "loss": 0.2404,
      "step": 22336
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7793467044830322,
      "learning_rate": 2.647459279245895e-06,
      "loss": 0.3408,
      "step": 22337
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.2906007766723633,
      "learning_rate": 2.6467025205944763e-06,
      "loss": 0.2747,
      "step": 22338
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.928665041923523,
      "learning_rate": 2.645945853619374e-06,
      "loss": 0.3034,
      "step": 22339
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.015467882156372,
      "learning_rate": 2.645189278330027e-06,
      "loss": 0.3767,
      "step": 22340
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9590208530426025,
      "learning_rate": 2.644432794735866e-06,
      "loss": 0.2604,
      "step": 22341
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.276123285293579,
      "learning_rate": 2.6436764028463193e-06,
      "loss": 0.3206,
      "step": 22342
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.599277138710022,
      "learning_rate": 2.642920102670823e-06,
      "loss": 0.2315,
      "step": 22343
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9290777444839478,
      "learning_rate": 2.6421638942188034e-06,
      "loss": 0.3401,
      "step": 22344
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.9767998456954956,
      "learning_rate": 2.641407777499685e-06,
      "loss": 0.343,
      "step": 22345
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8479856252670288,
      "learning_rate": 2.6406517525229004e-06,
      "loss": 0.3642,
      "step": 22346
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8686773777008057,
      "learning_rate": 2.639895819297873e-06,
      "loss": 0.3206,
      "step": 22347
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.803368330001831,
      "learning_rate": 2.639139977834024e-06,
      "loss": 0.3825,
      "step": 22348
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8832051753997803,
      "learning_rate": 2.6383842281407824e-06,
      "loss": 0.2995,
      "step": 22349
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5282529592514038,
      "learning_rate": 2.637628570227568e-06,
      "loss": 0.2785,
      "step": 22350
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0050721168518066,
      "learning_rate": 2.6368730041038014e-06,
      "loss": 0.3409,
      "step": 22351
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.101034641265869,
      "learning_rate": 2.6361175297788997e-06,
      "loss": 0.2631,
      "step": 22352
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.388345241546631,
      "learning_rate": 2.6353621472622882e-06,
      "loss": 0.3427,
      "step": 22353
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.020235776901245,
      "learning_rate": 2.6346068565633807e-06,
      "loss": 0.3777,
      "step": 22354
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8307862281799316,
      "learning_rate": 2.633851657691592e-06,
      "loss": 0.296,
      "step": 22355
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8289072513580322,
      "learning_rate": 2.633096550656341e-06,
      "loss": 0.3406,
      "step": 22356
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8558484315872192,
      "learning_rate": 2.63234153546704e-06,
      "loss": 0.2972,
      "step": 22357
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8440220355987549,
      "learning_rate": 2.631586612133101e-06,
      "loss": 0.3177,
      "step": 22358
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.4826033115386963,
      "learning_rate": 2.6308317806639395e-06,
      "loss": 0.3372,
      "step": 22359
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8535425662994385,
      "learning_rate": 2.6300770410689634e-06,
      "loss": 0.2827,
      "step": 22360
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.013310194015503,
      "learning_rate": 2.629322393357582e-06,
      "loss": 0.338,
      "step": 22361
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.018519163131714,
      "learning_rate": 2.6285678375392066e-06,
      "loss": 0.3441,
      "step": 22362
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7611573934555054,
      "learning_rate": 2.6278133736232424e-06,
      "loss": 0.293,
      "step": 22363
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1452066898345947,
      "learning_rate": 2.627059001619097e-06,
      "loss": 0.3499,
      "step": 22364
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6933902502059937,
      "learning_rate": 2.626304721536169e-06,
      "loss": 0.2723,
      "step": 22365
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.7484779357910156,
      "learning_rate": 2.6255505333838727e-06,
      "loss": 0.3653,
      "step": 22366
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8130402565002441,
      "learning_rate": 2.6247964371716063e-06,
      "loss": 0.3667,
      "step": 22367
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.3115525245666504,
      "learning_rate": 2.624042432908767e-06,
      "loss": 0.3858,
      "step": 22368
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.666580319404602,
      "learning_rate": 2.623288520604762e-06,
      "loss": 0.3116,
      "step": 22369
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8538516759872437,
      "learning_rate": 2.622534700268987e-06,
      "loss": 0.2892,
      "step": 22370
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1354873180389404,
      "learning_rate": 2.621780971910839e-06,
      "loss": 0.4031,
      "step": 22371
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6303761005401611,
      "learning_rate": 2.62102733553972e-06,
      "loss": 0.2263,
      "step": 22372
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1171746253967285,
      "learning_rate": 2.6202737911650224e-06,
      "loss": 0.3016,
      "step": 22373
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5035682916641235,
      "learning_rate": 2.619520338796139e-06,
      "loss": 0.2941,
      "step": 22374
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7068135738372803,
      "learning_rate": 2.6187669784424672e-06,
      "loss": 0.2688,
      "step": 22375
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.2062315940856934,
      "learning_rate": 2.6180137101133996e-06,
      "loss": 0.2897,
      "step": 22376
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0415232181549072,
      "learning_rate": 2.6172605338183255e-06,
      "loss": 0.3095,
      "step": 22377
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7507314682006836,
      "learning_rate": 2.616507449566632e-06,
      "loss": 0.3093,
      "step": 22378
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8288328647613525,
      "learning_rate": 2.6157544573677142e-06,
      "loss": 0.3333,
      "step": 22379
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.338515043258667,
      "learning_rate": 2.615001557230957e-06,
      "loss": 0.3351,
      "step": 22380
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.6976637840270996,
      "learning_rate": 2.614248749165744e-06,
      "loss": 0.2826,
      "step": 22381
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.600620985031128,
      "learning_rate": 2.6134960331814683e-06,
      "loss": 0.2288,
      "step": 22382
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4585270881652832,
      "learning_rate": 2.612743409287509e-06,
      "loss": 0.251,
      "step": 22383
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.175264835357666,
      "learning_rate": 2.611990877493248e-06,
      "loss": 0.3201,
      "step": 22384
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.893262267112732,
      "learning_rate": 2.6112384378080725e-06,
      "loss": 0.3628,
      "step": 22385
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.5413925647735596,
      "learning_rate": 2.61048609024136e-06,
      "loss": 0.2546,
      "step": 22386
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7336465120315552,
      "learning_rate": 2.6097338348024882e-06,
      "loss": 0.3212,
      "step": 22387
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8980870246887207,
      "learning_rate": 2.608981671500842e-06,
      "loss": 0.399,
      "step": 22388
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8229485750198364,
      "learning_rate": 2.6082296003457954e-06,
      "loss": 0.3199,
      "step": 22389
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6994285583496094,
      "learning_rate": 2.6074776213467245e-06,
      "loss": 0.3278,
      "step": 22390
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9209064245224,
      "learning_rate": 2.6067257345130016e-06,
      "loss": 0.2593,
      "step": 22391
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7341539859771729,
      "learning_rate": 2.6059739398540076e-06,
      "loss": 0.2087,
      "step": 22392
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5587393045425415,
      "learning_rate": 2.60522223737911e-06,
      "loss": 0.3326,
      "step": 22393
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9302692413330078,
      "learning_rate": 2.6044706270976816e-06,
      "loss": 0.3012,
      "step": 22394
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7318891286849976,
      "learning_rate": 2.603719109019095e-06,
      "loss": 0.2889,
      "step": 22395
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.424466133117676,
      "learning_rate": 2.6029676831527197e-06,
      "loss": 0.3374,
      "step": 22396
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2623226642608643,
      "learning_rate": 2.602216349507919e-06,
      "loss": 0.3228,
      "step": 22397
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8349452018737793,
      "learning_rate": 2.601465108094067e-06,
      "loss": 0.2933,
      "step": 22398
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7155660390853882,
      "learning_rate": 2.6007139589205253e-06,
      "loss": 0.2566,
      "step": 22399
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.1935064792633057,
      "learning_rate": 2.5999629019966576e-06,
      "loss": 0.3024,
      "step": 22400
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7113890647888184,
      "learning_rate": 2.5992119373318337e-06,
      "loss": 0.2816,
      "step": 22401
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8018007278442383,
      "learning_rate": 2.5984610649354115e-06,
      "loss": 0.2926,
      "step": 22402
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8459155559539795,
      "learning_rate": 2.5977102848167534e-06,
      "loss": 0.3005,
      "step": 22403
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.955920934677124,
      "learning_rate": 2.596959596985217e-06,
      "loss": 0.3358,
      "step": 22404
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9444031715393066,
      "learning_rate": 2.596209001450166e-06,
      "loss": 0.3413,
      "step": 22405
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6490176916122437,
      "learning_rate": 2.5954584982209574e-06,
      "loss": 0.2759,
      "step": 22406
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8272700309753418,
      "learning_rate": 2.594708087306943e-06,
      "loss": 0.2944,
      "step": 22407
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0710465908050537,
      "learning_rate": 2.5939577687174854e-06,
      "loss": 0.4142,
      "step": 22408
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.838567852973938,
      "learning_rate": 2.5932075424619362e-06,
      "loss": 0.3413,
      "step": 22409
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6961315870285034,
      "learning_rate": 2.5924574085496467e-06,
      "loss": 0.2632,
      "step": 22410
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.1576309204101562,
      "learning_rate": 2.5917073669899727e-06,
      "loss": 0.3216,
      "step": 22411
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.151244640350342,
      "learning_rate": 2.5909574177922636e-06,
      "loss": 0.4051,
      "step": 22412
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9206511974334717,
      "learning_rate": 2.590207560965867e-06,
      "loss": 0.3243,
      "step": 22413
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9916508197784424,
      "learning_rate": 2.589457796520136e-06,
      "loss": 0.2549,
      "step": 22414
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.962181806564331,
      "learning_rate": 2.5887081244644164e-06,
      "loss": 0.2679,
      "step": 22415
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8666298389434814,
      "learning_rate": 2.5879585448080546e-06,
      "loss": 0.3505,
      "step": 22416
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7712153196334839,
      "learning_rate": 2.5872090575603925e-06,
      "loss": 0.3589,
      "step": 22417
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0507845878601074,
      "learning_rate": 2.5864596627307804e-06,
      "loss": 0.263,
      "step": 22418
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2337100505828857,
      "learning_rate": 2.585710360328558e-06,
      "loss": 0.3286,
      "step": 22419
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6171269416809082,
      "learning_rate": 2.5849611503630656e-06,
      "loss": 0.2788,
      "step": 22420
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.452899932861328,
      "learning_rate": 2.584212032843647e-06,
      "loss": 0.3963,
      "step": 22421
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.361487627029419,
      "learning_rate": 2.583463007779642e-06,
      "loss": 0.3637,
      "step": 22422
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0952141284942627,
      "learning_rate": 2.5827140751803835e-06,
      "loss": 0.3256,
      "step": 22423
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7128859758377075,
      "learning_rate": 2.581965235055215e-06,
      "loss": 0.2465,
      "step": 22424
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5398513078689575,
      "learning_rate": 2.5812164874134717e-06,
      "loss": 0.2252,
      "step": 22425
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7744840383529663,
      "learning_rate": 2.5804678322644825e-06,
      "loss": 0.3475,
      "step": 22426
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0429680347442627,
      "learning_rate": 2.5797192696175886e-06,
      "loss": 0.3275,
      "step": 22427
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7900726795196533,
      "learning_rate": 2.5789707994821213e-06,
      "loss": 0.3357,
      "step": 22428
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0331919193267822,
      "learning_rate": 2.578222421867409e-06,
      "loss": 0.3436,
      "step": 22429
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8345774412155151,
      "learning_rate": 2.5774741367827805e-06,
      "loss": 0.3355,
      "step": 22430
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8968439102172852,
      "learning_rate": 2.5767259442375715e-06,
      "loss": 0.3078,
      "step": 22431
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7582746744155884,
      "learning_rate": 2.5759778442411067e-06,
      "loss": 0.2911,
      "step": 22432
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8778868913650513,
      "learning_rate": 2.5752298368027083e-06,
      "loss": 0.2578,
      "step": 22433
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0253915786743164,
      "learning_rate": 2.574481921931711e-06,
      "loss": 0.335,
      "step": 22434
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7654930353164673,
      "learning_rate": 2.5737340996374327e-06,
      "loss": 0.2872,
      "step": 22435
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.1927125453948975,
      "learning_rate": 2.572986369929197e-06,
      "loss": 0.3804,
      "step": 22436
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.3362464904785156,
      "learning_rate": 2.57223873281633e-06,
      "loss": 0.3737,
      "step": 22437
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8675711154937744,
      "learning_rate": 2.5714911883081515e-06,
      "loss": 0.3288,
      "step": 22438
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8138104677200317,
      "learning_rate": 2.5707437364139777e-06,
      "loss": 0.318,
      "step": 22439
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.963687539100647,
      "learning_rate": 2.5699963771431323e-06,
      "loss": 0.3153,
      "step": 22440
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.1011853218078613,
      "learning_rate": 2.569249110504931e-06,
      "loss": 0.392,
      "step": 22441
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7360268831253052,
      "learning_rate": 2.5685019365086906e-06,
      "loss": 0.3223,
      "step": 22442
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8296843767166138,
      "learning_rate": 2.5677548551637235e-06,
      "loss": 0.3083,
      "step": 22443
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9686914682388306,
      "learning_rate": 2.5670078664793475e-06,
      "loss": 0.2801,
      "step": 22444
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9447444677352905,
      "learning_rate": 2.5662609704648754e-06,
      "loss": 0.3044,
      "step": 22445
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.757794737815857,
      "learning_rate": 2.5655141671296148e-06,
      "loss": 0.2683,
      "step": 22446
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.590890884399414,
      "learning_rate": 2.5647674564828817e-06,
      "loss": 0.2614,
      "step": 22447
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0766801834106445,
      "learning_rate": 2.5640208385339847e-06,
      "loss": 0.2987,
      "step": 22448
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.497423768043518,
      "learning_rate": 2.5632743132922268e-06,
      "loss": 0.2841,
      "step": 22449
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6753789186477661,
      "learning_rate": 2.562527880766923e-06,
      "loss": 0.3444,
      "step": 22450
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8856773376464844,
      "learning_rate": 2.5617815409673753e-06,
      "loss": 0.3455,
      "step": 22451
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2636685371398926,
      "learning_rate": 2.561035293902886e-06,
      "loss": 0.4211,
      "step": 22452
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.828178882598877,
      "learning_rate": 2.5602891395827654e-06,
      "loss": 0.2661,
      "step": 22453
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.1058406829833984,
      "learning_rate": 2.559543078016312e-06,
      "loss": 0.3156,
      "step": 22454
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.808620810508728,
      "learning_rate": 2.558797109212827e-06,
      "loss": 0.3859,
      "step": 22455
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.2418787479400635,
      "learning_rate": 2.55805123318161e-06,
      "loss": 0.3272,
      "step": 22456
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9849096536636353,
      "learning_rate": 2.557305449931964e-06,
      "loss": 0.3044,
      "step": 22457
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.797467827796936,
      "learning_rate": 2.556559759473184e-06,
      "loss": 0.3422,
      "step": 22458
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.092150926589966,
      "learning_rate": 2.555814161814564e-06,
      "loss": 0.3829,
      "step": 22459
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8779622316360474,
      "learning_rate": 2.555068656965406e-06,
      "loss": 0.3575,
      "step": 22460
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9573522806167603,
      "learning_rate": 2.554323244935002e-06,
      "loss": 0.2033,
      "step": 22461
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.714058756828308,
      "learning_rate": 2.553577925732641e-06,
      "loss": 0.2852,
      "step": 22462
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9755038022994995,
      "learning_rate": 2.5528326993676234e-06,
      "loss": 0.3568,
      "step": 22463
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.910614013671875,
      "learning_rate": 2.5520875658492337e-06,
      "loss": 0.3283,
      "step": 22464
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.5894265174865723,
      "learning_rate": 2.551342525186762e-06,
      "loss": 0.2693,
      "step": 22465
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0158350467681885,
      "learning_rate": 2.550597577389502e-06,
      "loss": 0.3505,
      "step": 22466
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.4435648918151855,
      "learning_rate": 2.549852722466737e-06,
      "loss": 0.3504,
      "step": 22467
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7922816276550293,
      "learning_rate": 2.549107960427756e-06,
      "loss": 0.3757,
      "step": 22468
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.5255582332611084,
      "learning_rate": 2.548363291281838e-06,
      "loss": 0.3396,
      "step": 22469
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.84433114528656,
      "learning_rate": 2.547618715038276e-06,
      "loss": 0.3174,
      "step": 22470
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8035579919815063,
      "learning_rate": 2.546874231706348e-06,
      "loss": 0.336,
      "step": 22471
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8614622354507446,
      "learning_rate": 2.5461298412953327e-06,
      "loss": 0.2753,
      "step": 22472
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9518446922302246,
      "learning_rate": 2.545385543814518e-06,
      "loss": 0.3642,
      "step": 22473
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6667273044586182,
      "learning_rate": 2.5446413392731807e-06,
      "loss": 0.3023,
      "step": 22474
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.67483389377594,
      "learning_rate": 2.543897227680594e-06,
      "loss": 0.3018,
      "step": 22475
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.1343882083892822,
      "learning_rate": 2.5431532090460422e-06,
      "loss": 0.3094,
      "step": 22476
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6719094514846802,
      "learning_rate": 2.5424092833787985e-06,
      "loss": 0.3019,
      "step": 22477
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.731567144393921,
      "learning_rate": 2.5416654506881334e-06,
      "loss": 0.3212,
      "step": 22478
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.1885688304901123,
      "learning_rate": 2.5409217109833285e-06,
      "loss": 0.3398,
      "step": 22479
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6017028093338013,
      "learning_rate": 2.5401780642736517e-06,
      "loss": 0.3289,
      "step": 22480
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1381425857543945,
      "learning_rate": 2.539434510568376e-06,
      "loss": 0.3914,
      "step": 22481
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.177476167678833,
      "learning_rate": 2.538691049876766e-06,
      "loss": 0.3784,
      "step": 22482
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1912853717803955,
      "learning_rate": 2.5379476822080993e-06,
      "loss": 0.3598,
      "step": 22483
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.781518578529358,
      "learning_rate": 2.5372044075716397e-06,
      "loss": 0.2608,
      "step": 22484
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9414668083190918,
      "learning_rate": 2.5364612259766498e-06,
      "loss": 0.3338,
      "step": 22485
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.057058811187744,
      "learning_rate": 2.5357181374324024e-06,
      "loss": 0.3008,
      "step": 22486
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5993722677230835,
      "learning_rate": 2.5349751419481596e-06,
      "loss": 0.2788,
      "step": 22487
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8172390460968018,
      "learning_rate": 2.5342322395331797e-06,
      "loss": 0.2899,
      "step": 22488
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.648437738418579,
      "learning_rate": 2.533489430196732e-06,
      "loss": 0.2996,
      "step": 22489
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5471562147140503,
      "learning_rate": 2.5327467139480732e-06,
      "loss": 0.3149,
      "step": 22490
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1021413803100586,
      "learning_rate": 2.5320040907964616e-06,
      "loss": 0.3115,
      "step": 22491
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.4097025394439697,
      "learning_rate": 2.5312615607511616e-06,
      "loss": 0.4466,
      "step": 22492
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.803341031074524,
      "learning_rate": 2.5305191238214256e-06,
      "loss": 0.2599,
      "step": 22493
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.3494701385498047,
      "learning_rate": 2.529776780016512e-06,
      "loss": 0.2881,
      "step": 22494
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9938328266143799,
      "learning_rate": 2.529034529345672e-06,
      "loss": 0.2973,
      "step": 22495
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6027525663375854,
      "learning_rate": 2.528292371818166e-06,
      "loss": 0.2491,
      "step": 22496
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7517400979995728,
      "learning_rate": 2.527550307443243e-06,
      "loss": 0.349,
      "step": 22497
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6352843046188354,
      "learning_rate": 2.5268083362301523e-06,
      "loss": 0.2112,
      "step": 22498
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8292269706726074,
      "learning_rate": 2.52606645818815e-06,
      "loss": 0.2569,
      "step": 22499
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1349904537200928,
      "learning_rate": 2.5253246733264823e-06,
      "loss": 0.3653,
      "step": 22500
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0451574325561523,
      "learning_rate": 2.5245829816543977e-06,
      "loss": 0.3171,
      "step": 22501
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.026074171066284,
      "learning_rate": 2.5238413831811424e-06,
      "loss": 0.3287,
      "step": 22502
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8552602529525757,
      "learning_rate": 2.5230998779159623e-06,
      "loss": 0.2638,
      "step": 22503
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7624717950820923,
      "learning_rate": 2.5223584658681e-06,
      "loss": 0.2886,
      "step": 22504
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.3218932151794434,
      "learning_rate": 2.5216171470468043e-06,
      "loss": 0.3129,
      "step": 22505
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9917083978652954,
      "learning_rate": 2.520875921461314e-06,
      "loss": 0.2903,
      "step": 22506
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9330031871795654,
      "learning_rate": 2.5201347891208683e-06,
      "loss": 0.3013,
      "step": 22507
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0285449028015137,
      "learning_rate": 2.519393750034713e-06,
      "loss": 0.2196,
      "step": 22508
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.9454405307769775,
      "learning_rate": 2.5186528042120827e-06,
      "loss": 0.3318,
      "step": 22509
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8246454000473022,
      "learning_rate": 2.517911951662214e-06,
      "loss": 0.2878,
      "step": 22510
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0534849166870117,
      "learning_rate": 2.5171711923943475e-06,
      "loss": 0.3289,
      "step": 22511
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.972536325454712,
      "learning_rate": 2.516430526417717e-06,
      "loss": 0.2872,
      "step": 22512
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.27848219871521,
      "learning_rate": 2.5156899537415526e-06,
      "loss": 0.3562,
      "step": 22513
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.204831123352051,
      "learning_rate": 2.5149494743750924e-06,
      "loss": 0.3659,
      "step": 22514
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5400828123092651,
      "learning_rate": 2.5142090883275683e-06,
      "loss": 0.2897,
      "step": 22515
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.143035411834717,
      "learning_rate": 2.5134687956082083e-06,
      "loss": 0.2777,
      "step": 22516
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0240161418914795,
      "learning_rate": 2.5127285962262405e-06,
      "loss": 0.3433,
      "step": 22517
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7933495044708252,
      "learning_rate": 2.5119884901908977e-06,
      "loss": 0.2729,
      "step": 22518
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.069448947906494,
      "learning_rate": 2.5112484775114056e-06,
      "loss": 0.2736,
      "step": 22519
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.831559181213379,
      "learning_rate": 2.510508558196986e-06,
      "loss": 0.2017,
      "step": 22520
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1942214965820312,
      "learning_rate": 2.509768732256871e-06,
      "loss": 0.3054,
      "step": 22521
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.4598822593688965,
      "learning_rate": 2.5090289997002793e-06,
      "loss": 0.2918,
      "step": 22522
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1003148555755615,
      "learning_rate": 2.508289360536432e-06,
      "loss": 0.3102,
      "step": 22523
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.941097617149353,
      "learning_rate": 2.5075498147745568e-06,
      "loss": 0.2955,
      "step": 22524
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8860249519348145,
      "learning_rate": 2.5068103624238705e-06,
      "loss": 0.334,
      "step": 22525
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.2000749111175537,
      "learning_rate": 2.5060710034935885e-06,
      "loss": 0.383,
      "step": 22526
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.769087553024292,
      "learning_rate": 2.5053317379929354e-06,
      "loss": 0.2827,
      "step": 22527
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6650267839431763,
      "learning_rate": 2.5045925659311244e-06,
      "loss": 0.2819,
      "step": 22528
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.3015646934509277,
      "learning_rate": 2.5038534873173703e-06,
      "loss": 0.363,
      "step": 22529
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.887591004371643,
      "learning_rate": 2.503114502160886e-06,
      "loss": 0.2905,
      "step": 22530
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9736957550048828,
      "learning_rate": 2.5023756104708895e-06,
      "loss": 0.3265,
      "step": 22531
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8509595394134521,
      "learning_rate": 2.5016368122565913e-06,
      "loss": 0.3502,
      "step": 22532
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.170542001724243,
      "learning_rate": 2.5008981075271976e-06,
      "loss": 0.3225,
      "step": 22533
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9695618152618408,
      "learning_rate": 2.5001594962919263e-06,
      "loss": 0.2398,
      "step": 22534
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1022613048553467,
      "learning_rate": 2.49942097855998e-06,
      "loss": 0.328,
      "step": 22535
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.2101902961730957,
      "learning_rate": 2.4986825543405648e-06,
      "loss": 0.3828,
      "step": 22536
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8879092931747437,
      "learning_rate": 2.497944223642893e-06,
      "loss": 0.2891,
      "step": 22537
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.910274624824524,
      "learning_rate": 2.4972059864761667e-06,
      "loss": 0.2899,
      "step": 22538
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.034681558609009,
      "learning_rate": 2.496467842849586e-06,
      "loss": 0.3424,
      "step": 22539
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9893628358840942,
      "learning_rate": 2.49572979277236e-06,
      "loss": 0.2509,
      "step": 22540
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7317357063293457,
      "learning_rate": 2.4949918362536875e-06,
      "loss": 0.271,
      "step": 22541
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9846243858337402,
      "learning_rate": 2.4942539733027683e-06,
      "loss": 0.3232,
      "step": 22542
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.895186424255371,
      "learning_rate": 2.4935162039287995e-06,
      "loss": 0.3454,
      "step": 22543
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9696273803710938,
      "learning_rate": 2.4927785281409845e-06,
      "loss": 0.2939,
      "step": 22544
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9870566129684448,
      "learning_rate": 2.4920409459485175e-06,
      "loss": 0.3601,
      "step": 22545
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1420650482177734,
      "learning_rate": 2.491303457360591e-06,
      "loss": 0.3422,
      "step": 22546
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7434614896774292,
      "learning_rate": 2.490566062386405e-06,
      "loss": 0.2706,
      "step": 22547
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.2568821907043457,
      "learning_rate": 2.4898287610351513e-06,
      "loss": 0.2816,
      "step": 22548
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.266666889190674,
      "learning_rate": 2.4890915533160176e-06,
      "loss": 0.2791,
      "step": 22549
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0934228897094727,
      "learning_rate": 2.488354439238202e-06,
      "loss": 0.3627,
      "step": 22550
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7648569345474243,
      "learning_rate": 2.4876174188108905e-06,
      "loss": 0.2695,
      "step": 22551
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.0980632305145264,
      "learning_rate": 2.48688049204327e-06,
      "loss": 0.3532,
      "step": 22552
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9099942445755005,
      "learning_rate": 2.486143658944533e-06,
      "loss": 0.251,
      "step": 22553
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6530494689941406,
      "learning_rate": 2.485406919523863e-06,
      "loss": 0.2865,
      "step": 22554
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.926295280456543,
      "learning_rate": 2.484670273790446e-06,
      "loss": 0.2984,
      "step": 22555
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.717512607574463,
      "learning_rate": 2.483933721753462e-06,
      "loss": 0.3565,
      "step": 22556
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9481104612350464,
      "learning_rate": 2.4831972634221004e-06,
      "loss": 0.3256,
      "step": 22557
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9702304601669312,
      "learning_rate": 2.48246089880554e-06,
      "loss": 0.2987,
      "step": 22558
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.95381760597229,
      "learning_rate": 2.481724627912958e-06,
      "loss": 0.3044,
      "step": 22559
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.247507095336914,
      "learning_rate": 2.4809884507535407e-06,
      "loss": 0.3978,
      "step": 22560
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8669681549072266,
      "learning_rate": 2.480252367336462e-06,
      "loss": 0.2961,
      "step": 22561
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.599899411201477,
      "learning_rate": 2.4795163776708964e-06,
      "loss": 0.2826,
      "step": 22562
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7084262371063232,
      "learning_rate": 2.4787804817660254e-06,
      "loss": 0.3241,
      "step": 22563
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1836817264556885,
      "learning_rate": 2.478044679631022e-06,
      "loss": 0.2855,
      "step": 22564
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.164699077606201,
      "learning_rate": 2.4773089712750553e-06,
      "loss": 0.399,
      "step": 22565
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9580941200256348,
      "learning_rate": 2.4765733567073035e-06,
      "loss": 0.2763,
      "step": 22566
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5854063034057617,
      "learning_rate": 2.4758378359369372e-06,
      "loss": 0.2936,
      "step": 22567
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.975225567817688,
      "learning_rate": 2.4751024089731236e-06,
      "loss": 0.2649,
      "step": 22568
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.949885606765747,
      "learning_rate": 2.4743670758250292e-06,
      "loss": 0.3002,
      "step": 22569
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9640737771987915,
      "learning_rate": 2.4736318365018287e-06,
      "loss": 0.2971,
      "step": 22570
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.013859510421753,
      "learning_rate": 2.472896691012685e-06,
      "loss": 0.3237,
      "step": 22571
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.718120813369751,
      "learning_rate": 2.4721616393667603e-06,
      "loss": 0.3404,
      "step": 22572
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7058273553848267,
      "learning_rate": 2.4714266815732256e-06,
      "loss": 0.3209,
      "step": 22573
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7495718002319336,
      "learning_rate": 2.4706918176412397e-06,
      "loss": 0.3323,
      "step": 22574
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7497265338897705,
      "learning_rate": 2.4699570475799616e-06,
      "loss": 0.3206,
      "step": 22575
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9143518209457397,
      "learning_rate": 2.4692223713985586e-06,
      "loss": 0.3235,
      "step": 22576
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7790772914886475,
      "learning_rate": 2.4684877891061877e-06,
      "loss": 0.3538,
      "step": 22577
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8273341655731201,
      "learning_rate": 2.4677533007120027e-06,
      "loss": 0.3156,
      "step": 22578
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.905052900314331,
      "learning_rate": 2.4670189062251672e-06,
      "loss": 0.3522,
      "step": 22579
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.002815008163452,
      "learning_rate": 2.466284605654834e-06,
      "loss": 0.2263,
      "step": 22580
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1231307983398438,
      "learning_rate": 2.465550399010158e-06,
      "loss": 0.3304,
      "step": 22581
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6574400663375854,
      "learning_rate": 2.464816286300291e-06,
      "loss": 0.3612,
      "step": 22582
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.124706506729126,
      "learning_rate": 2.464082267534391e-06,
      "loss": 0.3559,
      "step": 22583
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1805355548858643,
      "learning_rate": 2.4633483427216056e-06,
      "loss": 0.2414,
      "step": 22584
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1786909103393555,
      "learning_rate": 2.462614511871082e-06,
      "loss": 0.2673,
      "step": 22585
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7774566411972046,
      "learning_rate": 2.4618807749919747e-06,
      "loss": 0.2278,
      "step": 22586
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0246148109436035,
      "learning_rate": 2.4611471320934298e-06,
      "loss": 0.2388,
      "step": 22587
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0344431400299072,
      "learning_rate": 2.46041358318459e-06,
      "loss": 0.3373,
      "step": 22588
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.658820629119873,
      "learning_rate": 2.4596801282746075e-06,
      "loss": 0.3797,
      "step": 22589
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.837205410003662,
      "learning_rate": 2.458946767372622e-06,
      "loss": 0.3121,
      "step": 22590
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7582783699035645,
      "learning_rate": 2.458213500487776e-06,
      "loss": 0.3304,
      "step": 22591
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.3959059715270996,
      "learning_rate": 2.4574803276292147e-06,
      "loss": 0.3573,
      "step": 22592
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.3556087017059326,
      "learning_rate": 2.456747248806077e-06,
      "loss": 0.3252,
      "step": 22593
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1393322944641113,
      "learning_rate": 2.456014264027503e-06,
      "loss": 0.2897,
      "step": 22594
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8959091901779175,
      "learning_rate": 2.455281373302628e-06,
      "loss": 0.227,
      "step": 22595
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8513227701187134,
      "learning_rate": 2.4545485766405953e-06,
      "loss": 0.3601,
      "step": 22596
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0456690788269043,
      "learning_rate": 2.453815874050537e-06,
      "loss": 0.2741,
      "step": 22597
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.109724521636963,
      "learning_rate": 2.453083265541586e-06,
      "loss": 0.4554,
      "step": 22598
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9027358293533325,
      "learning_rate": 2.4523507511228805e-06,
      "loss": 0.3722,
      "step": 22599
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9663950204849243,
      "learning_rate": 2.4516183308035524e-06,
      "loss": 0.2971,
      "step": 22600
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.900930404663086,
      "learning_rate": 2.4508860045927284e-06,
      "loss": 0.3575,
      "step": 22601
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.159343957901001,
      "learning_rate": 2.450153772499544e-06,
      "loss": 0.326,
      "step": 22602
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9368102550506592,
      "learning_rate": 2.4494216345331277e-06,
      "loss": 0.306,
      "step": 22603
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.130849599838257,
      "learning_rate": 2.448689590702602e-06,
      "loss": 0.3649,
      "step": 22604
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1161279678344727,
      "learning_rate": 2.4479576410171004e-06,
      "loss": 0.3985,
      "step": 22605
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.092451572418213,
      "learning_rate": 2.447225785485746e-06,
      "loss": 0.3404,
      "step": 22606
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6152222156524658,
      "learning_rate": 2.4464940241176615e-06,
      "loss": 0.2773,
      "step": 22607
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.106196403503418,
      "learning_rate": 2.445762356921969e-06,
      "loss": 0.2697,
      "step": 22608
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1557211875915527,
      "learning_rate": 2.445030783907796e-06,
      "loss": 0.3376,
      "step": 22609
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0046229362487793,
      "learning_rate": 2.444299305084259e-06,
      "loss": 0.2948,
      "step": 22610
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.58864426612854,
      "learning_rate": 2.4435679204604758e-06,
      "loss": 0.2756,
      "step": 22611
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7051315307617188,
      "learning_rate": 2.4428366300455706e-06,
      "loss": 0.2829,
      "step": 22612
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.2526350021362305,
      "learning_rate": 2.4421054338486584e-06,
      "loss": 0.336,
      "step": 22613
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1393094062805176,
      "learning_rate": 2.44137433187885e-06,
      "loss": 0.2736,
      "step": 22614
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.2221012115478516,
      "learning_rate": 2.4406433241452687e-06,
      "loss": 0.3368,
      "step": 22615
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.427353024482727,
      "learning_rate": 2.4399124106570236e-06,
      "loss": 0.2602,
      "step": 22616
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8551452159881592,
      "learning_rate": 2.4391815914232265e-06,
      "loss": 0.2793,
      "step": 22617
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0799739360809326,
      "learning_rate": 2.4384508664529927e-06,
      "loss": 0.3715,
      "step": 22618
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.844267725944519,
      "learning_rate": 2.4377202357554295e-06,
      "loss": 0.3545,
      "step": 22619
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.116971254348755,
      "learning_rate": 2.436989699339647e-06,
      "loss": 0.281,
      "step": 22620
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9315252304077148,
      "learning_rate": 2.4362592572147493e-06,
      "loss": 0.3022,
      "step": 22621
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.572906494140625,
      "learning_rate": 2.4355289093898494e-06,
      "loss": 0.2065,
      "step": 22622
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8334040641784668,
      "learning_rate": 2.4347986558740498e-06,
      "loss": 0.2301,
      "step": 22623
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9242546558380127,
      "learning_rate": 2.4340684966764526e-06,
      "loss": 0.3372,
      "step": 22624
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1826698780059814,
      "learning_rate": 2.4333384318061647e-06,
      "loss": 0.3716,
      "step": 22625
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.084294319152832,
      "learning_rate": 2.432608461272288e-06,
      "loss": 0.3319,
      "step": 22626
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9899133443832397,
      "learning_rate": 2.431878585083919e-06,
      "loss": 0.3875,
      "step": 22627
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7022656202316284,
      "learning_rate": 2.4311488032501627e-06,
      "loss": 0.3561,
      "step": 22628
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9488838911056519,
      "learning_rate": 2.430419115780115e-06,
      "loss": 0.3181,
      "step": 22629
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.1977195739746094,
      "learning_rate": 2.429689522682871e-06,
      "loss": 0.386,
      "step": 22630
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8234094381332397,
      "learning_rate": 2.4289600239675313e-06,
      "loss": 0.2748,
      "step": 22631
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8852061033248901,
      "learning_rate": 2.4282306196431895e-06,
      "loss": 0.3233,
      "step": 22632
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.4167487621307373,
      "learning_rate": 2.427501309718937e-06,
      "loss": 0.31,
      "step": 22633
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7717489004135132,
      "learning_rate": 2.4267720942038664e-06,
      "loss": 0.3554,
      "step": 22634
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8747159242630005,
      "learning_rate": 2.4260429731070735e-06,
      "loss": 0.364,
      "step": 22635
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8068190813064575,
      "learning_rate": 2.4253139464376453e-06,
      "loss": 0.2772,
      "step": 22636
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7642924785614014,
      "learning_rate": 2.424585014204668e-06,
      "loss": 0.325,
      "step": 22637
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7447385787963867,
      "learning_rate": 2.4238561764172354e-06,
      "loss": 0.2968,
      "step": 22638
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7872995138168335,
      "learning_rate": 2.4231274330844323e-06,
      "loss": 0.3134,
      "step": 22639
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.3035781383514404,
      "learning_rate": 2.4223987842153396e-06,
      "loss": 0.3278,
      "step": 22640
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8784692287445068,
      "learning_rate": 2.4216702298190486e-06,
      "loss": 0.3095,
      "step": 22641
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0960922241210938,
      "learning_rate": 2.42094176990464e-06,
      "loss": 0.3029,
      "step": 22642
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8241304159164429,
      "learning_rate": 2.420213404481191e-06,
      "loss": 0.326,
      "step": 22643
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.270512342453003,
      "learning_rate": 2.4194851335577895e-06,
      "loss": 0.37,
      "step": 22644
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.3835630416870117,
      "learning_rate": 2.4187569571435133e-06,
      "loss": 0.3592,
      "step": 22645
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9375239610671997,
      "learning_rate": 2.418028875247439e-06,
      "loss": 0.3406,
      "step": 22646
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6902741193771362,
      "learning_rate": 2.417300887878641e-06,
      "loss": 0.2913,
      "step": 22647
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8301684856414795,
      "learning_rate": 2.4165729950462025e-06,
      "loss": 0.3161,
      "step": 22648
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.577369451522827,
      "learning_rate": 2.4158451967591946e-06,
      "loss": 0.3367,
      "step": 22649
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9518145322799683,
      "learning_rate": 2.4151174930266873e-06,
      "loss": 0.3301,
      "step": 22650
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.469550132751465,
      "learning_rate": 2.414389883857762e-06,
      "loss": 0.3718,
      "step": 22651
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8827303647994995,
      "learning_rate": 2.4136623692614834e-06,
      "loss": 0.317,
      "step": 22652
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6745846271514893,
      "learning_rate": 2.4129349492469214e-06,
      "loss": 0.3246,
      "step": 22653
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.23486065864563,
      "learning_rate": 2.4122076238231493e-06,
      "loss": 0.353,
      "step": 22654
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7915585041046143,
      "learning_rate": 2.4114803929992335e-06,
      "loss": 0.3338,
      "step": 22655
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9700467586517334,
      "learning_rate": 2.4107532567842374e-06,
      "loss": 0.2619,
      "step": 22656
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5208302736282349,
      "learning_rate": 2.4100262151872313e-06,
      "loss": 0.2477,
      "step": 22657
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.895003080368042,
      "learning_rate": 2.4092992682172765e-06,
      "loss": 0.2881,
      "step": 22658
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.634935736656189,
      "learning_rate": 2.4085724158834377e-06,
      "loss": 0.3466,
      "step": 22659
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7649312019348145,
      "learning_rate": 2.407845658194773e-06,
      "loss": 0.2767,
      "step": 22660
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8331916332244873,
      "learning_rate": 2.407118995160349e-06,
      "loss": 0.3556,
      "step": 22661
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.467200517654419,
      "learning_rate": 2.4063924267892223e-06,
      "loss": 0.2638,
      "step": 22662
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0257983207702637,
      "learning_rate": 2.405665953090448e-06,
      "loss": 0.3066,
      "step": 22663
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.639017939567566,
      "learning_rate": 2.4049395740730894e-06,
      "loss": 0.3346,
      "step": 22664
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7393701076507568,
      "learning_rate": 2.4042132897462013e-06,
      "loss": 0.2778,
      "step": 22665
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8789705038070679,
      "learning_rate": 2.403487100118833e-06,
      "loss": 0.2808,
      "step": 22666
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0554370880126953,
      "learning_rate": 2.402761005200045e-06,
      "loss": 0.3556,
      "step": 22667
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.022686004638672,
      "learning_rate": 2.4020350049988873e-06,
      "loss": 0.2901,
      "step": 22668
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0060057640075684,
      "learning_rate": 2.401309099524408e-06,
      "loss": 0.3784,
      "step": 22669
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.783614158630371,
      "learning_rate": 2.400583288785664e-06,
      "loss": 0.3051,
      "step": 22670
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.049710273742676,
      "learning_rate": 2.3998575727916996e-06,
      "loss": 0.2779,
      "step": 22671
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.0639090538024902,
      "learning_rate": 2.3991319515515633e-06,
      "loss": 0.3207,
      "step": 22672
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.120171308517456,
      "learning_rate": 2.3984064250743e-06,
      "loss": 0.3605,
      "step": 22673
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6885459423065186,
      "learning_rate": 2.3976809933689594e-06,
      "loss": 0.2936,
      "step": 22674
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.034886121749878,
      "learning_rate": 2.3969556564445828e-06,
      "loss": 0.3438,
      "step": 22675
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9081130027770996,
      "learning_rate": 2.396230414310211e-06,
      "loss": 0.2906,
      "step": 22676
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8535422086715698,
      "learning_rate": 2.3955052669748913e-06,
      "loss": 0.2736,
      "step": 22677
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8460725545883179,
      "learning_rate": 2.394780214447662e-06,
      "loss": 0.328,
      "step": 22678
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.209627628326416,
      "learning_rate": 2.39405525673756e-06,
      "loss": 0.3582,
      "step": 22679
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.075881004333496,
      "learning_rate": 2.393330393853628e-06,
      "loss": 0.3142,
      "step": 22680
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.761842966079712,
      "learning_rate": 2.3926056258049013e-06,
      "loss": 0.2725,
      "step": 22681
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.29231858253479,
      "learning_rate": 2.3918809526004115e-06,
      "loss": 0.3146,
      "step": 22682
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7579880952835083,
      "learning_rate": 2.3911563742492018e-06,
      "loss": 0.3057,
      "step": 22683
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8355602025985718,
      "learning_rate": 2.390431890760301e-06,
      "loss": 0.319,
      "step": 22684
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.0129802227020264,
      "learning_rate": 2.3897075021427416e-06,
      "loss": 0.3219,
      "step": 22685
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7631347179412842,
      "learning_rate": 2.388983208405553e-06,
      "loss": 0.2703,
      "step": 22686
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8050363063812256,
      "learning_rate": 2.3882590095577707e-06,
      "loss": 0.2648,
      "step": 22687
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1448557376861572,
      "learning_rate": 2.3875349056084194e-06,
      "loss": 0.3008,
      "step": 22688
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6307003498077393,
      "learning_rate": 2.3868108965665247e-06,
      "loss": 0.2618,
      "step": 22689
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.104921340942383,
      "learning_rate": 2.3860869824411203e-06,
      "loss": 0.3933,
      "step": 22690
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7061198949813843,
      "learning_rate": 2.3853631632412256e-06,
      "loss": 0.2952,
      "step": 22691
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.178090810775757,
      "learning_rate": 2.3846394389758642e-06,
      "loss": 0.2997,
      "step": 22692
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.6590049266815186,
      "learning_rate": 2.3839158096540647e-06,
      "loss": 0.3185,
      "step": 22693
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8430083990097046,
      "learning_rate": 2.383192275284845e-06,
      "loss": 0.3351,
      "step": 22694
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.213514566421509,
      "learning_rate": 2.382468835877223e-06,
      "loss": 0.2218,
      "step": 22695
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5831435918807983,
      "learning_rate": 2.3817454914402237e-06,
      "loss": 0.2538,
      "step": 22696
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5556362867355347,
      "learning_rate": 2.3810222419828634e-06,
      "loss": 0.259,
      "step": 22697
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8130027055740356,
      "learning_rate": 2.3802990875141575e-06,
      "loss": 0.2895,
      "step": 22698
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.363668441772461,
      "learning_rate": 2.379576028043121e-06,
      "loss": 0.2794,
      "step": 22699
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2077128887176514,
      "learning_rate": 2.3788530635787733e-06,
      "loss": 0.3826,
      "step": 22700
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.25717830657959,
      "learning_rate": 2.378130194130124e-06,
      "loss": 0.3373,
      "step": 22701
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7837507724761963,
      "learning_rate": 2.3774074197061835e-06,
      "loss": 0.2765,
      "step": 22702
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.19179630279541,
      "learning_rate": 2.3766847403159697e-06,
      "loss": 0.3353,
      "step": 22703
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6415365934371948,
      "learning_rate": 2.3759621559684876e-06,
      "loss": 0.3488,
      "step": 22704
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.711186170578003,
      "learning_rate": 2.375239666672745e-06,
      "loss": 0.2607,
      "step": 22705
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.840852975845337,
      "learning_rate": 2.374517272437753e-06,
      "loss": 0.3719,
      "step": 22706
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2643380165100098,
      "learning_rate": 2.3737949732725175e-06,
      "loss": 0.3211,
      "step": 22707
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.74526047706604,
      "learning_rate": 2.373072769186039e-06,
      "loss": 0.2933,
      "step": 22708
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6942769289016724,
      "learning_rate": 2.372350660187327e-06,
      "loss": 0.2654,
      "step": 22709
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5577466487884521,
      "learning_rate": 2.3716286462853834e-06,
      "loss": 0.2148,
      "step": 22710
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8465526103973389,
      "learning_rate": 2.3709067274892085e-06,
      "loss": 0.3058,
      "step": 22711
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7829170227050781,
      "learning_rate": 2.3701849038077994e-06,
      "loss": 0.3461,
      "step": 22712
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4957646131515503,
      "learning_rate": 2.3694631752501617e-06,
      "loss": 0.236,
      "step": 22713
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.7368855476379395,
      "learning_rate": 2.3687415418252914e-06,
      "loss": 0.2582,
      "step": 22714
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.613869309425354,
      "learning_rate": 2.368020003542182e-06,
      "loss": 0.3106,
      "step": 22715
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8847651481628418,
      "learning_rate": 2.367298560409833e-06,
      "loss": 0.235,
      "step": 22716
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.104149341583252,
      "learning_rate": 2.3665772124372387e-06,
      "loss": 0.3498,
      "step": 22717
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.746091604232788,
      "learning_rate": 2.3658559596333885e-06,
      "loss": 0.2987,
      "step": 22718
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.050151824951172,
      "learning_rate": 2.3651348020072805e-06,
      "loss": 0.4513,
      "step": 22719
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9995696544647217,
      "learning_rate": 2.3644137395679023e-06,
      "loss": 0.3209,
      "step": 22720
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5542371273040771,
      "learning_rate": 2.3636927723242408e-06,
      "loss": 0.2814,
      "step": 22721
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8232293128967285,
      "learning_rate": 2.3629719002852915e-06,
      "loss": 0.3979,
      "step": 22722
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8590807914733887,
      "learning_rate": 2.3622511234600376e-06,
      "loss": 0.3071,
      "step": 22723
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8967918157577515,
      "learning_rate": 2.3615304418574645e-06,
      "loss": 0.3383,
      "step": 22724
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2421674728393555,
      "learning_rate": 2.3608098554865567e-06,
      "loss": 0.2803,
      "step": 22725
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.4549880027770996,
      "learning_rate": 2.3600893643563027e-06,
      "loss": 0.3662,
      "step": 22726
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6439266204833984,
      "learning_rate": 2.3593689684756817e-06,
      "loss": 0.2502,
      "step": 22727
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9804940223693848,
      "learning_rate": 2.3586486678536726e-06,
      "loss": 0.3191,
      "step": 22728
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7995716333389282,
      "learning_rate": 2.3579284624992616e-06,
      "loss": 0.3203,
      "step": 22729
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2087693214416504,
      "learning_rate": 2.357208352421425e-06,
      "loss": 0.3099,
      "step": 22730
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.102750301361084,
      "learning_rate": 2.356488337629137e-06,
      "loss": 0.4012,
      "step": 22731
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9177902936935425,
      "learning_rate": 2.355768418131381e-06,
      "loss": 0.3331,
      "step": 22732
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2047934532165527,
      "learning_rate": 2.3550485939371305e-06,
      "loss": 0.3186,
      "step": 22733
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.3486738204956055,
      "learning_rate": 2.354328865055354e-06,
      "loss": 0.3493,
      "step": 22734
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.897559642791748,
      "learning_rate": 2.353609231495033e-06,
      "loss": 0.2933,
      "step": 22735
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6977837085723877,
      "learning_rate": 2.352889693265136e-06,
      "loss": 0.2916,
      "step": 22736
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.0228164196014404,
      "learning_rate": 2.3521702503746325e-06,
      "loss": 0.2972,
      "step": 22737
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7911102771759033,
      "learning_rate": 2.3514509028324905e-06,
      "loss": 0.3014,
      "step": 22738
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.11761474609375,
      "learning_rate": 2.350731650647683e-06,
      "loss": 0.3688,
      "step": 22739
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1739232540130615,
      "learning_rate": 2.3500124938291756e-06,
      "loss": 0.3056,
      "step": 22740
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.856974720954895,
      "learning_rate": 2.349293432385932e-06,
      "loss": 0.2724,
      "step": 22741
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2461862564086914,
      "learning_rate": 2.3485744663269206e-06,
      "loss": 0.3214,
      "step": 22742
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.19944429397583,
      "learning_rate": 2.347855595661103e-06,
      "loss": 0.3318,
      "step": 22743
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8980209827423096,
      "learning_rate": 2.34713682039744e-06,
      "loss": 0.3718,
      "step": 22744
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8862831592559814,
      "learning_rate": 2.346418140544896e-06,
      "loss": 0.334,
      "step": 22745
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8524545431137085,
      "learning_rate": 2.345699556112431e-06,
      "loss": 0.3161,
      "step": 22746
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8940678834915161,
      "learning_rate": 2.3449810671089988e-06,
      "loss": 0.3177,
      "step": 22747
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1722986698150635,
      "learning_rate": 2.3442626735435637e-06,
      "loss": 0.2846,
      "step": 22748
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8397167921066284,
      "learning_rate": 2.34354437542508e-06,
      "loss": 0.3478,
      "step": 22749
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.808140516281128,
      "learning_rate": 2.3428261727625023e-06,
      "loss": 0.3054,
      "step": 22750
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8400275707244873,
      "learning_rate": 2.342108065564781e-06,
      "loss": 0.2823,
      "step": 22751
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7980374097824097,
      "learning_rate": 2.3413900538408762e-06,
      "loss": 0.2696,
      "step": 22752
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9592641592025757,
      "learning_rate": 2.3406721375997365e-06,
      "loss": 0.285,
      "step": 22753
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.382493257522583,
      "learning_rate": 2.339954316850309e-06,
      "loss": 0.3833,
      "step": 22754
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5519540309906006,
      "learning_rate": 2.339236591601549e-06,
      "loss": 0.2769,
      "step": 22755
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7530337572097778,
      "learning_rate": 2.3385189618624017e-06,
      "loss": 0.2853,
      "step": 22756
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8571789264678955,
      "learning_rate": 2.3378014276418113e-06,
      "loss": 0.2362,
      "step": 22757
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1666460037231445,
      "learning_rate": 2.3370839889487285e-06,
      "loss": 0.3552,
      "step": 22758
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9351500272750854,
      "learning_rate": 2.3363666457920963e-06,
      "loss": 0.3368,
      "step": 22759
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6421535015106201,
      "learning_rate": 2.335649398180856e-06,
      "loss": 0.3058,
      "step": 22760
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8558380603790283,
      "learning_rate": 2.334932246123952e-06,
      "loss": 0.3057,
      "step": 22761
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9592276811599731,
      "learning_rate": 2.3342151896303267e-06,
      "loss": 0.224,
      "step": 22762
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1109251976013184,
      "learning_rate": 2.333498228708917e-06,
      "loss": 0.3514,
      "step": 22763
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8626668453216553,
      "learning_rate": 2.332781363368659e-06,
      "loss": 0.3376,
      "step": 22764
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.1204700469970703,
      "learning_rate": 2.332064593618497e-06,
      "loss": 0.3122,
      "step": 22765
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6277334690093994,
      "learning_rate": 2.331347919467364e-06,
      "loss": 0.294,
      "step": 22766
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.2262139320373535,
      "learning_rate": 2.330631340924191e-06,
      "loss": 0.3093,
      "step": 22767
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7865866422653198,
      "learning_rate": 2.329914857997919e-06,
      "loss": 0.3434,
      "step": 22768
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.9526880979537964,
      "learning_rate": 2.329198470697478e-06,
      "loss": 0.3296,
      "step": 22769
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5254961252212524,
      "learning_rate": 2.3284821790317947e-06,
      "loss": 0.2629,
      "step": 22770
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7091604471206665,
      "learning_rate": 2.3277659830098077e-06,
      "loss": 0.3195,
      "step": 22771
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.64808189868927,
      "learning_rate": 2.327049882640441e-06,
      "loss": 0.3011,
      "step": 22772
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9711099863052368,
      "learning_rate": 2.3263338779326206e-06,
      "loss": 0.3279,
      "step": 22773
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.109821319580078,
      "learning_rate": 2.3256179688952784e-06,
      "loss": 0.2781,
      "step": 22774
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.001577615737915,
      "learning_rate": 2.3249021555373385e-06,
      "loss": 0.3456,
      "step": 22775
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7424875497817993,
      "learning_rate": 2.3241864378677225e-06,
      "loss": 0.2296,
      "step": 22776
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.2747771739959717,
      "learning_rate": 2.3234708158953545e-06,
      "loss": 0.3258,
      "step": 22777
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4308950901031494,
      "learning_rate": 2.3227552896291582e-06,
      "loss": 0.2741,
      "step": 22778
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.162332534790039,
      "learning_rate": 2.3220398590780547e-06,
      "loss": 0.3614,
      "step": 22779
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0273678302764893,
      "learning_rate": 2.3213245242509574e-06,
      "loss": 0.3466,
      "step": 22780
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.427319288253784,
      "learning_rate": 2.3206092851567942e-06,
      "loss": 0.3494,
      "step": 22781
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.802415609359741,
      "learning_rate": 2.319894141804476e-06,
      "loss": 0.4124,
      "step": 22782
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9722431898117065,
      "learning_rate": 2.319179094202918e-06,
      "loss": 0.352,
      "step": 22783
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7916340827941895,
      "learning_rate": 2.31846414236104e-06,
      "loss": 0.3346,
      "step": 22784
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.916353464126587,
      "learning_rate": 2.3177492862877516e-06,
      "loss": 0.2995,
      "step": 22785
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7802221775054932,
      "learning_rate": 2.3170345259919647e-06,
      "loss": 0.2702,
      "step": 22786
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.08412504196167,
      "learning_rate": 2.316319861482594e-06,
      "loss": 0.2573,
      "step": 22787
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6446001529693604,
      "learning_rate": 2.315605292768549e-06,
      "loss": 0.2829,
      "step": 22788
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9018328189849854,
      "learning_rate": 2.3148908198587363e-06,
      "loss": 0.2781,
      "step": 22789
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0738210678100586,
      "learning_rate": 2.3141764427620604e-06,
      "loss": 0.2875,
      "step": 22790
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0410826206207275,
      "learning_rate": 2.3134621614874355e-06,
      "loss": 0.2711,
      "step": 22791
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6568654775619507,
      "learning_rate": 2.3127479760437622e-06,
      "loss": 0.217,
      "step": 22792
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.3773257732391357,
      "learning_rate": 2.312033886439943e-06,
      "loss": 0.3778,
      "step": 22793
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9977543354034424,
      "learning_rate": 2.311319892684886e-06,
      "loss": 0.2987,
      "step": 22794
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0962321758270264,
      "learning_rate": 2.310605994787488e-06,
      "loss": 0.2798,
      "step": 22795
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8187986612319946,
      "learning_rate": 2.3098921927566496e-06,
      "loss": 0.3847,
      "step": 22796
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8886219263076782,
      "learning_rate": 2.3091784866012743e-06,
      "loss": 0.2629,
      "step": 22797
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.022449493408203,
      "learning_rate": 2.3084648763302586e-06,
      "loss": 0.3444,
      "step": 22798
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8780111074447632,
      "learning_rate": 2.307751361952494e-06,
      "loss": 0.3367,
      "step": 22799
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.2518913745880127,
      "learning_rate": 2.307037943476883e-06,
      "loss": 0.3624,
      "step": 22800
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.820488691329956,
      "learning_rate": 2.3063246209123182e-06,
      "loss": 0.3024,
      "step": 22801
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8950542211532593,
      "learning_rate": 2.3056113942676917e-06,
      "loss": 0.2854,
      "step": 22802
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.701765775680542,
      "learning_rate": 2.3048982635518935e-06,
      "loss": 0.2865,
      "step": 22803
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7928802967071533,
      "learning_rate": 2.3041852287738185e-06,
      "loss": 0.2533,
      "step": 22804
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.1613690853118896,
      "learning_rate": 2.3034722899423566e-06,
      "loss": 0.2954,
      "step": 22805
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.294741153717041,
      "learning_rate": 2.3027594470663906e-06,
      "loss": 0.2725,
      "step": 22806
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6985372304916382,
      "learning_rate": 2.302046700154814e-06,
      "loss": 0.2852,
      "step": 22807
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8744958639144897,
      "learning_rate": 2.3013340492165114e-06,
      "loss": 0.2789,
      "step": 22808
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8553435802459717,
      "learning_rate": 2.3006214942603634e-06,
      "loss": 0.3491,
      "step": 22809
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0572102069854736,
      "learning_rate": 2.29990903529526e-06,
      "loss": 0.3072,
      "step": 22810
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.6741530895233154,
      "learning_rate": 2.299196672330081e-06,
      "loss": 0.341,
      "step": 22811
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9307314157485962,
      "learning_rate": 2.2984844053737043e-06,
      "loss": 0.3239,
      "step": 22812
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5748809576034546,
      "learning_rate": 2.2977722344350163e-06,
      "loss": 0.3308,
      "step": 22813
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9045125246047974,
      "learning_rate": 2.2970601595228925e-06,
      "loss": 0.2857,
      "step": 22814
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9109808206558228,
      "learning_rate": 2.296348180646212e-06,
      "loss": 0.2584,
      "step": 22815
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.863730549812317,
      "learning_rate": 2.2956362978138457e-06,
      "loss": 0.314,
      "step": 22816
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.236248016357422,
      "learning_rate": 2.2949245110346774e-06,
      "loss": 0.3134,
      "step": 22817
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9035190343856812,
      "learning_rate": 2.294212820317577e-06,
      "loss": 0.2205,
      "step": 22818
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5653034448623657,
      "learning_rate": 2.293501225671415e-06,
      "loss": 0.3746,
      "step": 22819
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7100439071655273,
      "learning_rate": 2.292789727105068e-06,
      "loss": 0.3425,
      "step": 22820
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9510823488235474,
      "learning_rate": 2.2920783246274047e-06,
      "loss": 0.2986,
      "step": 22821
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0028889179229736,
      "learning_rate": 2.29136701824729e-06,
      "loss": 0.3625,
      "step": 22822
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6784124374389648,
      "learning_rate": 2.2906558079736007e-06,
      "loss": 0.3036,
      "step": 22823
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.185497760772705,
      "learning_rate": 2.289944693815198e-06,
      "loss": 0.4396,
      "step": 22824
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.189220428466797,
      "learning_rate": 2.2892336757809452e-06,
      "loss": 0.3308,
      "step": 22825
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.6248762607574463,
      "learning_rate": 2.288522753879714e-06,
      "loss": 0.2706,
      "step": 22826
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9699735641479492,
      "learning_rate": 2.2878119281203647e-06,
      "loss": 0.2771,
      "step": 22827
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.051327705383301,
      "learning_rate": 2.2871011985117574e-06,
      "loss": 0.2829,
      "step": 22828
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.234426498413086,
      "learning_rate": 2.2863905650627527e-06,
      "loss": 0.2784,
      "step": 22829
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7149677276611328,
      "learning_rate": 2.2856800277822142e-06,
      "loss": 0.2814,
      "step": 22830
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7689504623413086,
      "learning_rate": 2.284969586678998e-06,
      "loss": 0.2406,
      "step": 22831
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8386017084121704,
      "learning_rate": 2.2842592417619592e-06,
      "loss": 0.3771,
      "step": 22832
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.647847294807434,
      "learning_rate": 2.283548993039959e-06,
      "loss": 0.2581,
      "step": 22833
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0402870178222656,
      "learning_rate": 2.2828388405218505e-06,
      "loss": 0.2808,
      "step": 22834
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.118011713027954,
      "learning_rate": 2.282128784216483e-06,
      "loss": 0.311,
      "step": 22835
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7839473485946655,
      "learning_rate": 2.281418824132716e-06,
      "loss": 0.2771,
      "step": 22836
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.3799397945404053,
      "learning_rate": 2.2807089602793976e-06,
      "loss": 0.3136,
      "step": 22837
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.650925874710083,
      "learning_rate": 2.279999192665375e-06,
      "loss": 0.3203,
      "step": 22838
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7484530210494995,
      "learning_rate": 2.2792895212995036e-06,
      "loss": 0.2381,
      "step": 22839
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9135462045669556,
      "learning_rate": 2.2785799461906264e-06,
      "loss": 0.2968,
      "step": 22840
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8334670066833496,
      "learning_rate": 2.2778704673475914e-06,
      "loss": 0.3157,
      "step": 22841
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0520355701446533,
      "learning_rate": 2.2771610847792414e-06,
      "loss": 0.3548,
      "step": 22842
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.336859941482544,
      "learning_rate": 2.276451798494426e-06,
      "loss": 0.2551,
      "step": 22843
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7529443502426147,
      "learning_rate": 2.275742608501984e-06,
      "loss": 0.3192,
      "step": 22844
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.152921676635742,
      "learning_rate": 2.2750335148107563e-06,
      "loss": 0.3437,
      "step": 22845
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.680159330368042,
      "learning_rate": 2.274324517429587e-06,
      "loss": 0.3182,
      "step": 22846
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.0974907875061035,
      "learning_rate": 2.273615616367314e-06,
      "loss": 0.3142,
      "step": 22847
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.279747724533081,
      "learning_rate": 2.2729068116327723e-06,
      "loss": 0.3838,
      "step": 22848
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8641905784606934,
      "learning_rate": 2.272198103234804e-06,
      "loss": 0.3086,
      "step": 22849
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8395185470581055,
      "learning_rate": 2.271489491182243e-06,
      "loss": 0.3039,
      "step": 22850
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9260051250457764,
      "learning_rate": 2.2707809754839206e-06,
      "loss": 0.3075,
      "step": 22851
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5148106813430786,
      "learning_rate": 2.270072556148676e-06,
      "loss": 0.2094,
      "step": 22852
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8117514848709106,
      "learning_rate": 2.2693642331853405e-06,
      "loss": 0.3448,
      "step": 22853
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.877450942993164,
      "learning_rate": 2.268656006602736e-06,
      "loss": 0.3143,
      "step": 22854
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.103525400161743,
      "learning_rate": 2.267947876409703e-06,
      "loss": 0.3125,
      "step": 22855
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9728819131851196,
      "learning_rate": 2.2672398426150666e-06,
      "loss": 0.3334,
      "step": 22856
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9139037132263184,
      "learning_rate": 2.2665319052276492e-06,
      "loss": 0.2935,
      "step": 22857
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7602607011795044,
      "learning_rate": 2.2658240642562847e-06,
      "loss": 0.2828,
      "step": 22858
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.208667039871216,
      "learning_rate": 2.2651163197097947e-06,
      "loss": 0.3363,
      "step": 22859
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9408479928970337,
      "learning_rate": 2.2644086715970016e-06,
      "loss": 0.3038,
      "step": 22860
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.3070669174194336,
      "learning_rate": 2.2637011199267277e-06,
      "loss": 0.3211,
      "step": 22861
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9144996404647827,
      "learning_rate": 2.2629936647077977e-06,
      "loss": 0.3229,
      "step": 22862
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.1349329948425293,
      "learning_rate": 2.26228630594903e-06,
      "loss": 0.3181,
      "step": 22863
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.301330327987671,
      "learning_rate": 2.26157904365924e-06,
      "loss": 0.4207,
      "step": 22864
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.2529444694519043,
      "learning_rate": 2.2608718778472517e-06,
      "loss": 0.3153,
      "step": 22865
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.454045295715332,
      "learning_rate": 2.2601648085218775e-06,
      "loss": 0.2586,
      "step": 22866
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7294721603393555,
      "learning_rate": 2.2594578356919317e-06,
      "loss": 0.2849,
      "step": 22867
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.0477914810180664,
      "learning_rate": 2.2587509593662337e-06,
      "loss": 0.274,
      "step": 22868
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.4325506687164307,
      "learning_rate": 2.2580441795535924e-06,
      "loss": 0.3584,
      "step": 22869
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7714911699295044,
      "learning_rate": 2.257337496262817e-06,
      "loss": 0.2555,
      "step": 22870
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.4905498027801514,
      "learning_rate": 2.256630909502725e-06,
      "loss": 0.2878,
      "step": 22871
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.643073320388794,
      "learning_rate": 2.2559244192821205e-06,
      "loss": 0.29,
      "step": 22872
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9927239418029785,
      "learning_rate": 2.2552180256098143e-06,
      "loss": 0.3076,
      "step": 22873
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.020395517349243,
      "learning_rate": 2.2545117284946083e-06,
      "loss": 0.2425,
      "step": 22874
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.893653392791748,
      "learning_rate": 2.2538055279453143e-06,
      "loss": 0.2676,
      "step": 22875
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.202238082885742,
      "learning_rate": 2.2530994239707347e-06,
      "loss": 0.3233,
      "step": 22876
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8398531675338745,
      "learning_rate": 2.252393416579669e-06,
      "loss": 0.3189,
      "step": 22877
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6384356021881104,
      "learning_rate": 2.251687505780926e-06,
      "loss": 0.3241,
      "step": 22878
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6317826509475708,
      "learning_rate": 2.2509816915833028e-06,
      "loss": 0.2985,
      "step": 22879
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.87297523021698,
      "learning_rate": 2.250275973995596e-06,
      "loss": 0.3227,
      "step": 22880
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.18296480178833,
      "learning_rate": 2.249570353026611e-06,
      "loss": 0.3487,
      "step": 22881
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.239001512527466,
      "learning_rate": 2.2488648286851413e-06,
      "loss": 0.3575,
      "step": 22882
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.338050127029419,
      "learning_rate": 2.248159400979979e-06,
      "loss": 0.3451,
      "step": 22883
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9717636108398438,
      "learning_rate": 2.2474540699199276e-06,
      "loss": 0.2418,
      "step": 22884
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8945691585540771,
      "learning_rate": 2.2467488355137744e-06,
      "loss": 0.301,
      "step": 22885
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9622745513916016,
      "learning_rate": 2.246043697770315e-06,
      "loss": 0.2891,
      "step": 22886
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.933513879776001,
      "learning_rate": 2.245338656698336e-06,
      "loss": 0.2988,
      "step": 22887
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1764607429504395,
      "learning_rate": 2.244633712306633e-06,
      "loss": 0.3248,
      "step": 22888
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9836734533309937,
      "learning_rate": 2.2439288646039924e-06,
      "loss": 0.287,
      "step": 22889
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7270034551620483,
      "learning_rate": 2.2432241135991992e-06,
      "loss": 0.2308,
      "step": 22890
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.208845853805542,
      "learning_rate": 2.2425194593010458e-06,
      "loss": 0.3208,
      "step": 22891
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9456417560577393,
      "learning_rate": 2.241814901718313e-06,
      "loss": 0.305,
      "step": 22892
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8544166088104248,
      "learning_rate": 2.241110440859784e-06,
      "loss": 0.2723,
      "step": 22893
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1665196418762207,
      "learning_rate": 2.240406076734245e-06,
      "loss": 0.3267,
      "step": 22894
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.5760622024536133,
      "learning_rate": 2.2397018093504764e-06,
      "loss": 0.348,
      "step": 22895
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9348933696746826,
      "learning_rate": 2.238997638717255e-06,
      "loss": 0.4457,
      "step": 22896
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.2381131649017334,
      "learning_rate": 2.2382935648433655e-06,
      "loss": 0.2544,
      "step": 22897
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.183581590652466,
      "learning_rate": 2.237589587737584e-06,
      "loss": 0.3754,
      "step": 22898
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7257829904556274,
      "learning_rate": 2.236885707408687e-06,
      "loss": 0.3346,
      "step": 22899
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8363410234451294,
      "learning_rate": 2.236181923865446e-06,
      "loss": 0.2658,
      "step": 22900
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7675992250442505,
      "learning_rate": 2.2354782371166417e-06,
      "loss": 0.2853,
      "step": 22901
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.345519542694092,
      "learning_rate": 2.2347746471710454e-06,
      "loss": 0.3636,
      "step": 22902
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7860645055770874,
      "learning_rate": 2.234071154037425e-06,
      "loss": 0.2712,
      "step": 22903
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.075144052505493,
      "learning_rate": 2.233367757724557e-06,
      "loss": 0.335,
      "step": 22904
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7783576250076294,
      "learning_rate": 2.232664458241208e-06,
      "loss": 0.2701,
      "step": 22905
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8249354362487793,
      "learning_rate": 2.2319612555961445e-06,
      "loss": 0.3302,
      "step": 22906
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.935524582862854,
      "learning_rate": 2.231258149798138e-06,
      "loss": 0.3256,
      "step": 22907
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9868934154510498,
      "learning_rate": 2.2305551408559524e-06,
      "loss": 0.3339,
      "step": 22908
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.876413106918335,
      "learning_rate": 2.2298522287783487e-06,
      "loss": 0.246,
      "step": 22909
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8232572078704834,
      "learning_rate": 2.2291494135740964e-06,
      "loss": 0.3142,
      "step": 22910
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.607316255569458,
      "learning_rate": 2.2284466952519557e-06,
      "loss": 0.2921,
      "step": 22911
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.898391604423523,
      "learning_rate": 2.227744073820687e-06,
      "loss": 0.3237,
      "step": 22912
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.3026373386383057,
      "learning_rate": 2.227041549289046e-06,
      "loss": 0.3538,
      "step": 22913
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.0480120182037354,
      "learning_rate": 2.2263391216658e-06,
      "loss": 0.3358,
      "step": 22914
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.214399576187134,
      "learning_rate": 2.225636790959701e-06,
      "loss": 0.3378,
      "step": 22915
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7157584428787231,
      "learning_rate": 2.224934557179503e-06,
      "loss": 0.2057,
      "step": 22916
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5582891702651978,
      "learning_rate": 2.2242324203339683e-06,
      "loss": 0.234,
      "step": 22917
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.0571401119232178,
      "learning_rate": 2.2235303804318454e-06,
      "loss": 0.3425,
      "step": 22918
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8036385774612427,
      "learning_rate": 2.2228284374818854e-06,
      "loss": 0.3137,
      "step": 22919
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8793374300003052,
      "learning_rate": 2.222126591492846e-06,
      "loss": 0.2524,
      "step": 22920
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.177109479904175,
      "learning_rate": 2.221424842473472e-06,
      "loss": 0.2726,
      "step": 22921
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1076924800872803,
      "learning_rate": 2.220723190432512e-06,
      "loss": 0.3417,
      "step": 22922
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6199967861175537,
      "learning_rate": 2.2200216353787196e-06,
      "loss": 0.3196,
      "step": 22923
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6891899108886719,
      "learning_rate": 2.219320177320836e-06,
      "loss": 0.3194,
      "step": 22924
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7202645540237427,
      "learning_rate": 2.2186188162676094e-06,
      "loss": 0.2879,
      "step": 22925
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.307070732116699,
      "learning_rate": 2.2179175522277796e-06,
      "loss": 0.2866,
      "step": 22926
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.559009075164795,
      "learning_rate": 2.217216385210095e-06,
      "loss": 0.4425,
      "step": 22927
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.0197525024414062,
      "learning_rate": 2.2165153152232953e-06,
      "loss": 0.2895,
      "step": 22928
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.765870213508606,
      "learning_rate": 2.215814342276117e-06,
      "loss": 0.3211,
      "step": 22929
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1481103897094727,
      "learning_rate": 2.2151134663773067e-06,
      "loss": 0.3125,
      "step": 22930
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9822757244110107,
      "learning_rate": 2.2144126875355987e-06,
      "loss": 0.323,
      "step": 22931
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.117877244949341,
      "learning_rate": 2.213712005759726e-06,
      "loss": 0.2779,
      "step": 22932
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.5490827560424805,
      "learning_rate": 2.2130114210584318e-06,
      "loss": 0.3544,
      "step": 22933
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.898737907409668,
      "learning_rate": 2.212310933440448e-06,
      "loss": 0.2996,
      "step": 22934
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8929328918457031,
      "learning_rate": 2.211610542914503e-06,
      "loss": 0.2511,
      "step": 22935
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9245734214782715,
      "learning_rate": 2.2109102494893353e-06,
      "loss": 0.1769,
      "step": 22936
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9940829277038574,
      "learning_rate": 2.2102100531736726e-06,
      "loss": 0.2652,
      "step": 22937
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7819608449935913,
      "learning_rate": 2.2095099539762455e-06,
      "loss": 0.2876,
      "step": 22938
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8278543949127197,
      "learning_rate": 2.2088099519057795e-06,
      "loss": 0.3073,
      "step": 22939
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.006718635559082,
      "learning_rate": 2.208110046971006e-06,
      "loss": 0.2581,
      "step": 22940
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1031367778778076,
      "learning_rate": 2.2074102391806495e-06,
      "loss": 0.317,
      "step": 22941
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.0964314937591553,
      "learning_rate": 2.206710528543432e-06,
      "loss": 0.388,
      "step": 22942
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.4819209575653076,
      "learning_rate": 2.206010915068082e-06,
      "loss": 0.4074,
      "step": 22943
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6053295135498047,
      "learning_rate": 2.20531139876332e-06,
      "loss": 0.2703,
      "step": 22944
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.644493818283081,
      "learning_rate": 2.204611979637863e-06,
      "loss": 0.3219,
      "step": 22945
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.7056503295898438,
      "learning_rate": 2.203912657700437e-06,
      "loss": 0.2888,
      "step": 22946
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.3296940326690674,
      "learning_rate": 2.2032134329597577e-06,
      "loss": 0.3058,
      "step": 22947
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.678781270980835,
      "learning_rate": 2.2025143054245403e-06,
      "loss": 0.3504,
      "step": 22948
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.464223861694336,
      "learning_rate": 2.2018152751035073e-06,
      "loss": 0.4165,
      "step": 22949
    },
    {
      "epoch": 2.37,
      "grad_norm": 3.210137128829956,
      "learning_rate": 2.2011163420053695e-06,
      "loss": 0.3911,
      "step": 22950
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.8422157764434814,
      "learning_rate": 2.200417506138841e-06,
      "loss": 0.2732,
      "step": 22951
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.09934401512146,
      "learning_rate": 2.199718767512633e-06,
      "loss": 0.2883,
      "step": 22952
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7580119371414185,
      "learning_rate": 2.1990201261354617e-06,
      "loss": 0.3348,
      "step": 22953
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1275486946105957,
      "learning_rate": 2.1983215820160343e-06,
      "loss": 0.2969,
      "step": 22954
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.0762877464294434,
      "learning_rate": 2.1976231351630573e-06,
      "loss": 0.3734,
      "step": 22955
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9162907600402832,
      "learning_rate": 2.196924785585244e-06,
      "loss": 0.3325,
      "step": 22956
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.026763439178467,
      "learning_rate": 2.196226533291297e-06,
      "loss": 0.308,
      "step": 22957
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.574744701385498,
      "learning_rate": 2.195528378289922e-06,
      "loss": 0.2984,
      "step": 22958
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.1308107376098633,
      "learning_rate": 2.1948303205898246e-06,
      "loss": 0.2901,
      "step": 22959
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.2594692707061768,
      "learning_rate": 2.1941323601997087e-06,
      "loss": 0.299,
      "step": 22960
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7884641885757446,
      "learning_rate": 2.193434497128272e-06,
      "loss": 0.3011,
      "step": 22961
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.861728549003601,
      "learning_rate": 2.19273673138422e-06,
      "loss": 0.3452,
      "step": 22962
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9697483777999878,
      "learning_rate": 2.192039062976249e-06,
      "loss": 0.2464,
      "step": 22963
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.006666660308838,
      "learning_rate": 2.191341491913057e-06,
      "loss": 0.3088,
      "step": 22964
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9211806058883667,
      "learning_rate": 2.1906440182033396e-06,
      "loss": 0.282,
      "step": 22965
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.486706256866455,
      "learning_rate": 2.189946641855797e-06,
      "loss": 0.3522,
      "step": 22966
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.2276291847229004,
      "learning_rate": 2.189249362879121e-06,
      "loss": 0.347,
      "step": 22967
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.002336025238037,
      "learning_rate": 2.188552181282002e-06,
      "loss": 0.3726,
      "step": 22968
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.108997344970703,
      "learning_rate": 2.1878550970731382e-06,
      "loss": 0.2548,
      "step": 22969
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.579967975616455,
      "learning_rate": 2.187158110261217e-06,
      "loss": 0.2809,
      "step": 22970
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8669135570526123,
      "learning_rate": 2.1864612208549253e-06,
      "loss": 0.3492,
      "step": 22971
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0833163261413574,
      "learning_rate": 2.1857644288629563e-06,
      "loss": 0.4153,
      "step": 22972
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7068203687667847,
      "learning_rate": 2.185067734293995e-06,
      "loss": 0.2954,
      "step": 22973
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.033841848373413,
      "learning_rate": 2.1843711371567257e-06,
      "loss": 0.3225,
      "step": 22974
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.13331937789917,
      "learning_rate": 2.1836746374598374e-06,
      "loss": 0.336,
      "step": 22975
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8423789739608765,
      "learning_rate": 2.182978235212011e-06,
      "loss": 0.3229,
      "step": 22976
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.3142385482788086,
      "learning_rate": 2.182281930421929e-06,
      "loss": 0.3064,
      "step": 22977
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6044386625289917,
      "learning_rate": 2.1815857230982694e-06,
      "loss": 0.286,
      "step": 22978
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8637892007827759,
      "learning_rate": 2.180889613249718e-06,
      "loss": 0.3253,
      "step": 22979
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.5992119312286377,
      "learning_rate": 2.180193600884951e-06,
      "loss": 0.3467,
      "step": 22980
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.562912106513977,
      "learning_rate": 2.1794976860126427e-06,
      "loss": 0.2911,
      "step": 22981
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.299891948699951,
      "learning_rate": 2.1788018686414748e-06,
      "loss": 0.3281,
      "step": 22982
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.67681086063385,
      "learning_rate": 2.1781061487801193e-06,
      "loss": 0.3283,
      "step": 22983
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8419196605682373,
      "learning_rate": 2.177410526437247e-06,
      "loss": 0.2763,
      "step": 22984
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8449815511703491,
      "learning_rate": 2.176715001621538e-06,
      "loss": 0.3144,
      "step": 22985
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.6359527111053467,
      "learning_rate": 2.176019574341658e-06,
      "loss": 0.334,
      "step": 22986
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8172905445098877,
      "learning_rate": 2.175324244606276e-06,
      "loss": 0.2704,
      "step": 22987
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.4015188217163086,
      "learning_rate": 2.1746290124240667e-06,
      "loss": 0.3231,
      "step": 22988
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.629889488220215,
      "learning_rate": 2.1739338778036945e-06,
      "loss": 0.3035,
      "step": 22989
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.224331855773926,
      "learning_rate": 2.1732388407538264e-06,
      "loss": 0.3215,
      "step": 22990
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.3077900409698486,
      "learning_rate": 2.172543901283124e-06,
      "loss": 0.4134,
      "step": 22991
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0378189086914062,
      "learning_rate": 2.1718490594002585e-06,
      "loss": 0.3517,
      "step": 22992
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.902299404144287,
      "learning_rate": 2.1711543151138883e-06,
      "loss": 0.3627,
      "step": 22993
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9313701391220093,
      "learning_rate": 2.1704596684326717e-06,
      "loss": 0.2895,
      "step": 22994
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7890019416809082,
      "learning_rate": 2.169765119365278e-06,
      "loss": 0.2806,
      "step": 22995
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9458965063095093,
      "learning_rate": 2.16907066792036e-06,
      "loss": 0.2884,
      "step": 22996
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.041529655456543,
      "learning_rate": 2.168376314106575e-06,
      "loss": 0.2813,
      "step": 22997
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.273397445678711,
      "learning_rate": 2.1676820579325854e-06,
      "loss": 0.4271,
      "step": 22998
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8839950561523438,
      "learning_rate": 2.1669878994070425e-06,
      "loss": 0.3376,
      "step": 22999
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8094134330749512,
      "learning_rate": 2.166293838538599e-06,
      "loss": 0.1953,
      "step": 23000
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5265439748764038,
      "learning_rate": 2.165599875335913e-06,
      "loss": 0.2309,
      "step": 23001
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.83170747756958,
      "learning_rate": 2.1649060098076335e-06,
      "loss": 0.2887,
      "step": 23002
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8622111082077026,
      "learning_rate": 2.164212241962412e-06,
      "loss": 0.3097,
      "step": 23003
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.353545665740967,
      "learning_rate": 2.163518571808896e-06,
      "loss": 0.2878,
      "step": 23004
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1653690338134766,
      "learning_rate": 2.162824999355737e-06,
      "loss": 0.3493,
      "step": 23005
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7475875616073608,
      "learning_rate": 2.1621315246115803e-06,
      "loss": 0.3294,
      "step": 23006
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8332427740097046,
      "learning_rate": 2.1614381475850687e-06,
      "loss": 0.2578,
      "step": 23007
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.731397271156311,
      "learning_rate": 2.160744868284853e-06,
      "loss": 0.2684,
      "step": 23008
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8256680965423584,
      "learning_rate": 2.1600516867195733e-06,
      "loss": 0.3519,
      "step": 23009
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.7404980659484863,
      "learning_rate": 2.1593586028978685e-06,
      "loss": 0.3136,
      "step": 23010
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.804401159286499,
      "learning_rate": 2.158665616828386e-06,
      "loss": 0.2439,
      "step": 23011
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1161110401153564,
      "learning_rate": 2.157972728519763e-06,
      "loss": 0.2508,
      "step": 23012
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5271823406219482,
      "learning_rate": 2.1572799379806342e-06,
      "loss": 0.2978,
      "step": 23013
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9250998497009277,
      "learning_rate": 2.1565872452196424e-06,
      "loss": 0.3432,
      "step": 23014
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.967541217803955,
      "learning_rate": 2.1558946502454213e-06,
      "loss": 0.2983,
      "step": 23015
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7120527029037476,
      "learning_rate": 2.1552021530666067e-06,
      "loss": 0.3285,
      "step": 23016
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9496325254440308,
      "learning_rate": 2.154509753691828e-06,
      "loss": 0.2724,
      "step": 23017
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.941705584526062,
      "learning_rate": 2.1538174521297238e-06,
      "loss": 0.3129,
      "step": 23018
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1007723808288574,
      "learning_rate": 2.153125248388922e-06,
      "loss": 0.3682,
      "step": 23019
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7953966856002808,
      "learning_rate": 2.152433142478051e-06,
      "loss": 0.2577,
      "step": 23020
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.4588332176208496,
      "learning_rate": 2.1517411344057436e-06,
      "loss": 0.312,
      "step": 23021
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9790891408920288,
      "learning_rate": 2.1510492241806248e-06,
      "loss": 0.2362,
      "step": 23022
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5811424255371094,
      "learning_rate": 2.150357411811319e-06,
      "loss": 0.2474,
      "step": 23023
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5902026891708374,
      "learning_rate": 2.149665697306457e-06,
      "loss": 0.2611,
      "step": 23024
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.284355640411377,
      "learning_rate": 2.1489740806746584e-06,
      "loss": 0.2959,
      "step": 23025
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.695404291152954,
      "learning_rate": 2.148282561924544e-06,
      "loss": 0.2919,
      "step": 23026
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.2137887477874756,
      "learning_rate": 2.1475911410647408e-06,
      "loss": 0.4027,
      "step": 23027
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.07364559173584,
      "learning_rate": 2.146899818103867e-06,
      "loss": 0.323,
      "step": 23028
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.733868956565857,
      "learning_rate": 2.1462085930505395e-06,
      "loss": 0.3145,
      "step": 23029
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7648766040802002,
      "learning_rate": 2.145517465913375e-06,
      "loss": 0.3212,
      "step": 23030
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1466853618621826,
      "learning_rate": 2.144826436700995e-06,
      "loss": 0.3556,
      "step": 23031
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.697165846824646,
      "learning_rate": 2.1441355054220124e-06,
      "loss": 0.3207,
      "step": 23032
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.3100528717041016,
      "learning_rate": 2.143444672085038e-06,
      "loss": 0.3839,
      "step": 23033
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.423985242843628,
      "learning_rate": 2.1427539366986906e-06,
      "loss": 0.3317,
      "step": 23034
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8133759498596191,
      "learning_rate": 2.1420632992715797e-06,
      "loss": 0.3689,
      "step": 23035
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.753350853919983,
      "learning_rate": 2.1413727598123113e-06,
      "loss": 0.3171,
      "step": 23036
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8254456520080566,
      "learning_rate": 2.140682318329501e-06,
      "loss": 0.2818,
      "step": 23037
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7946529388427734,
      "learning_rate": 2.1399919748317543e-06,
      "loss": 0.3235,
      "step": 23038
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9970762729644775,
      "learning_rate": 2.1393017293276754e-06,
      "loss": 0.353,
      "step": 23039
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7160197496414185,
      "learning_rate": 2.1386115818258746e-06,
      "loss": 0.2867,
      "step": 23040
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.672054648399353,
      "learning_rate": 2.1379215323349546e-06,
      "loss": 0.2746,
      "step": 23041
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6979924440383911,
      "learning_rate": 2.137231580863517e-06,
      "loss": 0.3006,
      "step": 23042
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.974941611289978,
      "learning_rate": 2.1365417274201614e-06,
      "loss": 0.3071,
      "step": 23043
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.057415723800659,
      "learning_rate": 2.135851972013495e-06,
      "loss": 0.3106,
      "step": 23044
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.1312968730926514,
      "learning_rate": 2.135162314652114e-06,
      "loss": 0.3889,
      "step": 23045
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.019766092300415,
      "learning_rate": 2.1344727553446133e-06,
      "loss": 0.3068,
      "step": 23046
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8923499584197998,
      "learning_rate": 2.1337832940995962e-06,
      "loss": 0.2748,
      "step": 23047
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9016300439834595,
      "learning_rate": 2.1330939309256547e-06,
      "loss": 0.3359,
      "step": 23048
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8630828857421875,
      "learning_rate": 2.1324046658313815e-06,
      "loss": 0.2997,
      "step": 23049
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.060455799102783,
      "learning_rate": 2.1317154988253743e-06,
      "loss": 0.2621,
      "step": 23050
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0399117469787598,
      "learning_rate": 2.131026429916223e-06,
      "loss": 0.3484,
      "step": 23051
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8630950450897217,
      "learning_rate": 2.130337459112517e-06,
      "loss": 0.3035,
      "step": 23052
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.722640037536621,
      "learning_rate": 2.12964858642285e-06,
      "loss": 0.3005,
      "step": 23053
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.3587303161621094,
      "learning_rate": 2.1289598118558085e-06,
      "loss": 0.3786,
      "step": 23054
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8859814405441284,
      "learning_rate": 2.1282711354199793e-06,
      "loss": 0.2616,
      "step": 23055
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8089152574539185,
      "learning_rate": 2.127582557123945e-06,
      "loss": 0.3563,
      "step": 23056
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.9100655317306519,
      "learning_rate": 2.126894076976297e-06,
      "loss": 0.234,
      "step": 23057
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.408432722091675,
      "learning_rate": 2.1262056949856158e-06,
      "loss": 0.3681,
      "step": 23058
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.0251402854919434,
      "learning_rate": 2.1255174111604805e-06,
      "loss": 0.331,
      "step": 23059
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.644033670425415,
      "learning_rate": 2.124829225509477e-06,
      "loss": 0.2404,
      "step": 23060
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8568474054336548,
      "learning_rate": 2.124141138041185e-06,
      "loss": 0.3318,
      "step": 23061
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.89429771900177,
      "learning_rate": 2.1234531487641774e-06,
      "loss": 0.307,
      "step": 23062
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.318312168121338,
      "learning_rate": 2.1227652576870393e-06,
      "loss": 0.3728,
      "step": 23063
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8343145847320557,
      "learning_rate": 2.122077464818343e-06,
      "loss": 0.2943,
      "step": 23064
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.997204065322876,
      "learning_rate": 2.121389770166662e-06,
      "loss": 0.299,
      "step": 23065
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.885879635810852,
      "learning_rate": 2.1207021737405732e-06,
      "loss": 0.3394,
      "step": 23066
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.3832297325134277,
      "learning_rate": 2.1200146755486482e-06,
      "loss": 0.344,
      "step": 23067
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.349623203277588,
      "learning_rate": 2.119327275599459e-06,
      "loss": 0.3949,
      "step": 23068
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9980782270431519,
      "learning_rate": 2.1186399739015704e-06,
      "loss": 0.2573,
      "step": 23069
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7889982461929321,
      "learning_rate": 2.1179527704635594e-06,
      "loss": 0.2182,
      "step": 23070
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6707754135131836,
      "learning_rate": 2.117265665293988e-06,
      "loss": 0.2885,
      "step": 23071
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9348727464675903,
      "learning_rate": 2.1165786584014225e-06,
      "loss": 0.297,
      "step": 23072
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0843145847320557,
      "learning_rate": 2.115891749794433e-06,
      "loss": 0.3838,
      "step": 23073
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.950161099433899,
      "learning_rate": 2.11520493948158e-06,
      "loss": 0.3342,
      "step": 23074
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8097480535507202,
      "learning_rate": 2.1145182274714225e-06,
      "loss": 0.342,
      "step": 23075
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8510538339614868,
      "learning_rate": 2.1138316137725303e-06,
      "loss": 0.2926,
      "step": 23076
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.99164879322052,
      "learning_rate": 2.1131450983934587e-06,
      "loss": 0.249,
      "step": 23077
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8843189477920532,
      "learning_rate": 2.112458681342765e-06,
      "loss": 0.2732,
      "step": 23078
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0058717727661133,
      "learning_rate": 2.111772362629011e-06,
      "loss": 0.3049,
      "step": 23079
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8442074060440063,
      "learning_rate": 2.111086142260753e-06,
      "loss": 0.2835,
      "step": 23080
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.883163332939148,
      "learning_rate": 2.110400020246545e-06,
      "loss": 0.258,
      "step": 23081
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9317189455032349,
      "learning_rate": 2.1097139965949376e-06,
      "loss": 0.3253,
      "step": 23082
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0557684898376465,
      "learning_rate": 2.1090280713144905e-06,
      "loss": 0.308,
      "step": 23083
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0293936729431152,
      "learning_rate": 2.108342244413751e-06,
      "loss": 0.2824,
      "step": 23084
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.940277099609375,
      "learning_rate": 2.10765651590127e-06,
      "loss": 0.3403,
      "step": 23085
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7073637247085571,
      "learning_rate": 2.1069708857856e-06,
      "loss": 0.2211,
      "step": 23086
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7747559547424316,
      "learning_rate": 2.106285354075286e-06,
      "loss": 0.2892,
      "step": 23087
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0767455101013184,
      "learning_rate": 2.1055999207788713e-06,
      "loss": 0.3037,
      "step": 23088
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.137275218963623,
      "learning_rate": 2.10491458590491e-06,
      "loss": 0.4135,
      "step": 23089
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.134347438812256,
      "learning_rate": 2.1042293494619414e-06,
      "loss": 0.288,
      "step": 23090
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7936052083969116,
      "learning_rate": 2.103544211458506e-06,
      "loss": 0.2916,
      "step": 23091
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.846005916595459,
      "learning_rate": 2.102859171903151e-06,
      "loss": 0.3075,
      "step": 23092
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8971965312957764,
      "learning_rate": 2.1021742308044156e-06,
      "loss": 0.2892,
      "step": 23093
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.366168260574341,
      "learning_rate": 2.1014893881708387e-06,
      "loss": 0.2746,
      "step": 23094
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.616777777671814,
      "learning_rate": 2.1008046440109553e-06,
      "loss": 0.3006,
      "step": 23095
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7582472562789917,
      "learning_rate": 2.1001199983333074e-06,
      "loss": 0.2835,
      "step": 23096
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9952903985977173,
      "learning_rate": 2.099435451146429e-06,
      "loss": 0.371,
      "step": 23097
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8403091430664062,
      "learning_rate": 2.098751002458851e-06,
      "loss": 0.2949,
      "step": 23098
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.2967116832733154,
      "learning_rate": 2.0980666522791126e-06,
      "loss": 0.3458,
      "step": 23099
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7310303449630737,
      "learning_rate": 2.097382400615744e-06,
      "loss": 0.3346,
      "step": 23100
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0549778938293457,
      "learning_rate": 2.0966982474772724e-06,
      "loss": 0.341,
      "step": 23101
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.50630259513855,
      "learning_rate": 2.0960141928722323e-06,
      "loss": 0.3882,
      "step": 23102
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8073776960372925,
      "learning_rate": 2.09533023680915e-06,
      "loss": 0.3085,
      "step": 23103
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8929129838943481,
      "learning_rate": 2.094646379296549e-06,
      "loss": 0.3399,
      "step": 23104
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.2381434440612793,
      "learning_rate": 2.093962620342963e-06,
      "loss": 0.3598,
      "step": 23105
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.2198829650878906,
      "learning_rate": 2.0932789599569113e-06,
      "loss": 0.3569,
      "step": 23106
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4777708053588867,
      "learning_rate": 2.0925953981469193e-06,
      "loss": 0.2469,
      "step": 23107
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.623978853225708,
      "learning_rate": 2.0919119349215046e-06,
      "loss": 0.2572,
      "step": 23108
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7729063034057617,
      "learning_rate": 2.0912285702891953e-06,
      "loss": 0.2841,
      "step": 23109
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9691886901855469,
      "learning_rate": 2.0905453042585087e-06,
      "loss": 0.2778,
      "step": 23110
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8339418172836304,
      "learning_rate": 2.089862136837959e-06,
      "loss": 0.298,
      "step": 23111
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.860032558441162,
      "learning_rate": 2.089179068036069e-06,
      "loss": 0.335,
      "step": 23112
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0176517963409424,
      "learning_rate": 2.0884960978613543e-06,
      "loss": 0.2393,
      "step": 23113
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9933667182922363,
      "learning_rate": 2.087813226322325e-06,
      "loss": 0.2984,
      "step": 23114
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6069318056106567,
      "learning_rate": 2.0871304534274994e-06,
      "loss": 0.2671,
      "step": 23115
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.1834182739257812,
      "learning_rate": 2.08644777918539e-06,
      "loss": 0.3198,
      "step": 23116
    },
    {
      "epoch": 2.39,
      "grad_norm": 3.2698779106140137,
      "learning_rate": 2.085765203604503e-06,
      "loss": 0.3684,
      "step": 23117
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0301733016967773,
      "learning_rate": 2.085082726693355e-06,
      "loss": 0.2531,
      "step": 23118
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9135175943374634,
      "learning_rate": 2.0844003484604504e-06,
      "loss": 0.2934,
      "step": 23119
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7484222650527954,
      "learning_rate": 2.0837180689142988e-06,
      "loss": 0.2845,
      "step": 23120
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.030496835708618,
      "learning_rate": 2.0830358880634016e-06,
      "loss": 0.325,
      "step": 23121
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.069885015487671,
      "learning_rate": 2.0823538059162705e-06,
      "loss": 0.347,
      "step": 23122
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6814262866973877,
      "learning_rate": 2.081671822481407e-06,
      "loss": 0.3501,
      "step": 23123
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8410301208496094,
      "learning_rate": 2.0809899377673092e-06,
      "loss": 0.313,
      "step": 23124
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8478890657424927,
      "learning_rate": 2.080308151782485e-06,
      "loss": 0.2785,
      "step": 23125
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.2467596530914307,
      "learning_rate": 2.0796264645354326e-06,
      "loss": 0.3522,
      "step": 23126
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.3390085697174072,
      "learning_rate": 2.078944876034645e-06,
      "loss": 0.3717,
      "step": 23127
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.285391092300415,
      "learning_rate": 2.078263386288629e-06,
      "loss": 0.3829,
      "step": 23128
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.2523727416992188,
      "learning_rate": 2.0775819953058774e-06,
      "loss": 0.3331,
      "step": 23129
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0058162212371826,
      "learning_rate": 2.0769007030948805e-06,
      "loss": 0.3491,
      "step": 23130
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6893764734268188,
      "learning_rate": 2.0762195096641404e-06,
      "loss": 0.2807,
      "step": 23131
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0025670528411865,
      "learning_rate": 2.075538415022145e-06,
      "loss": 0.2833,
      "step": 23132
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.1907355785369873,
      "learning_rate": 2.074857419177386e-06,
      "loss": 0.3396,
      "step": 23133
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.372366189956665,
      "learning_rate": 2.0741765221383515e-06,
      "loss": 0.3258,
      "step": 23134
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8700352907180786,
      "learning_rate": 2.073495723913537e-06,
      "loss": 0.3222,
      "step": 23135
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9784642457962036,
      "learning_rate": 2.0728150245114255e-06,
      "loss": 0.2988,
      "step": 23136
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5683668851852417,
      "learning_rate": 2.072134423940502e-06,
      "loss": 0.2013,
      "step": 23137
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.894290804862976,
      "learning_rate": 2.071453922209257e-06,
      "loss": 0.333,
      "step": 23138
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6094799041748047,
      "learning_rate": 2.070773519326171e-06,
      "loss": 0.3312,
      "step": 23139
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5790261030197144,
      "learning_rate": 2.0700932152997268e-06,
      "loss": 0.2036,
      "step": 23140
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6703777313232422,
      "learning_rate": 2.069413010138408e-06,
      "loss": 0.3366,
      "step": 23141
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0135996341705322,
      "learning_rate": 2.068732903850694e-06,
      "loss": 0.3551,
      "step": 23142
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.541320562362671,
      "learning_rate": 2.068052896445062e-06,
      "loss": 0.3237,
      "step": 23143
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.242608070373535,
      "learning_rate": 2.067372987929993e-06,
      "loss": 0.3618,
      "step": 23144
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4882134199142456,
      "learning_rate": 2.066693178313963e-06,
      "loss": 0.302,
      "step": 23145
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.489328145980835,
      "learning_rate": 2.0660134676054478e-06,
      "loss": 0.3199,
      "step": 23146
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.910031795501709,
      "learning_rate": 2.0653338558129166e-06,
      "loss": 0.2937,
      "step": 23147
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.123812198638916,
      "learning_rate": 2.0646543429448497e-06,
      "loss": 0.2922,
      "step": 23148
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.1428396701812744,
      "learning_rate": 2.0639749290097155e-06,
      "loss": 0.3856,
      "step": 23149
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8991769552230835,
      "learning_rate": 2.063295614015981e-06,
      "loss": 0.3557,
      "step": 23150
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.966391921043396,
      "learning_rate": 2.0626163979721224e-06,
      "loss": 0.3298,
      "step": 23151
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.245614767074585,
      "learning_rate": 2.0619372808866045e-06,
      "loss": 0.3114,
      "step": 23152
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.109172821044922,
      "learning_rate": 2.0612582627678913e-06,
      "loss": 0.3256,
      "step": 23153
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.97562837600708,
      "learning_rate": 2.060579343624454e-06,
      "loss": 0.2779,
      "step": 23154
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.0681769847869873,
      "learning_rate": 2.059900523464753e-06,
      "loss": 0.2112,
      "step": 23155
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.6749484539031982,
      "learning_rate": 2.05922180229725e-06,
      "loss": 0.3424,
      "step": 23156
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6519572734832764,
      "learning_rate": 2.0585431801304124e-06,
      "loss": 0.2379,
      "step": 23157
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.4715993404388428,
      "learning_rate": 2.057864656972698e-06,
      "loss": 0.3863,
      "step": 23158
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.626262664794922,
      "learning_rate": 2.057186232832565e-06,
      "loss": 0.2943,
      "step": 23159
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.191220283508301,
      "learning_rate": 2.0565079077184704e-06,
      "loss": 0.2533,
      "step": 23160
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9050134420394897,
      "learning_rate": 2.0558296816388757e-06,
      "loss": 0.3724,
      "step": 23161
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.765863060951233,
      "learning_rate": 2.0551515546022337e-06,
      "loss": 0.3008,
      "step": 23162
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9051185846328735,
      "learning_rate": 2.054473526616997e-06,
      "loss": 0.3329,
      "step": 23163
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7916661500930786,
      "learning_rate": 2.0537955976916233e-06,
      "loss": 0.3198,
      "step": 23164
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8350526094436646,
      "learning_rate": 2.0531177678345626e-06,
      "loss": 0.2484,
      "step": 23165
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8249945640563965,
      "learning_rate": 2.052440037054263e-06,
      "loss": 0.2973,
      "step": 23166
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8616474866867065,
      "learning_rate": 2.051762405359179e-06,
      "loss": 0.364,
      "step": 23167
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8877363204956055,
      "learning_rate": 2.051084872757757e-06,
      "loss": 0.2633,
      "step": 23168
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7948087453842163,
      "learning_rate": 2.0504074392584394e-06,
      "loss": 0.2966,
      "step": 23169
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.3098814487457275,
      "learning_rate": 2.049730104869679e-06,
      "loss": 0.3213,
      "step": 23170
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.980963945388794,
      "learning_rate": 2.049052869599919e-06,
      "loss": 0.3298,
      "step": 23171
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.2100279331207275,
      "learning_rate": 2.0483757334575995e-06,
      "loss": 0.3669,
      "step": 23172
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.798702359199524,
      "learning_rate": 2.0476986964511626e-06,
      "loss": 0.3347,
      "step": 23173
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7961030006408691,
      "learning_rate": 2.047021758589053e-06,
      "loss": 0.345,
      "step": 23174
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0643975734710693,
      "learning_rate": 2.046344919879708e-06,
      "loss": 0.2998,
      "step": 23175
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7536070346832275,
      "learning_rate": 2.0456681803315638e-06,
      "loss": 0.2166,
      "step": 23176
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7138015031814575,
      "learning_rate": 2.044991539953062e-06,
      "loss": 0.3302,
      "step": 23177
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.386662483215332,
      "learning_rate": 2.0443149987526367e-06,
      "loss": 0.3859,
      "step": 23178
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.4348785877227783,
      "learning_rate": 2.0436385567387206e-06,
      "loss": 0.3739,
      "step": 23179
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9794609546661377,
      "learning_rate": 2.0429622139197514e-06,
      "loss": 0.367,
      "step": 23180
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7208893299102783,
      "learning_rate": 2.0422859703041577e-06,
      "loss": 0.2845,
      "step": 23181
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.3932945728302,
      "learning_rate": 2.0416098259003702e-06,
      "loss": 0.3215,
      "step": 23182
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.3524038791656494,
      "learning_rate": 2.0409337807168215e-06,
      "loss": 0.3156,
      "step": 23183
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8686381578445435,
      "learning_rate": 2.0402578347619395e-06,
      "loss": 0.3201,
      "step": 23184
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0005104541778564,
      "learning_rate": 2.0395819880441514e-06,
      "loss": 0.3288,
      "step": 23185
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.102992534637451,
      "learning_rate": 2.0389062405718784e-06,
      "loss": 0.328,
      "step": 23186
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.930871605873108,
      "learning_rate": 2.0382305923535526e-06,
      "loss": 0.3165,
      "step": 23187
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0398828983306885,
      "learning_rate": 2.037555043397594e-06,
      "loss": 0.2956,
      "step": 23188
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.792348027229309,
      "learning_rate": 2.0368795937124232e-06,
      "loss": 0.3064,
      "step": 23189
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8645591735839844,
      "learning_rate": 2.0362042433064653e-06,
      "loss": 0.3016,
      "step": 23190
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.960425853729248,
      "learning_rate": 2.0355289921881395e-06,
      "loss": 0.2894,
      "step": 23191
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8784329891204834,
      "learning_rate": 2.034853840365859e-06,
      "loss": 0.2818,
      "step": 23192
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0835301876068115,
      "learning_rate": 2.034178787848049e-06,
      "loss": 0.2967,
      "step": 23193
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.483116626739502,
      "learning_rate": 2.0335038346431214e-06,
      "loss": 0.3445,
      "step": 23194
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.2080349922180176,
      "learning_rate": 2.032828980759489e-06,
      "loss": 0.2982,
      "step": 23195
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6809971332550049,
      "learning_rate": 2.0321542262055715e-06,
      "loss": 0.2337,
      "step": 23196
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6409577131271362,
      "learning_rate": 2.0314795709897772e-06,
      "loss": 0.2751,
      "step": 23197
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7237330675125122,
      "learning_rate": 2.0308050151205173e-06,
      "loss": 0.2832,
      "step": 23198
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9030548334121704,
      "learning_rate": 2.0301305586062002e-06,
      "loss": 0.3869,
      "step": 23199
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.8864662647247314,
      "learning_rate": 2.02945620145524e-06,
      "loss": 0.3846,
      "step": 23200
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7256876230239868,
      "learning_rate": 2.0287819436760405e-06,
      "loss": 0.2735,
      "step": 23201
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.08219838142395,
      "learning_rate": 2.028107785277006e-06,
      "loss": 0.2972,
      "step": 23202
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.9117681980133057,
      "learning_rate": 2.027433726266547e-06,
      "loss": 0.2474,
      "step": 23203
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.253213405609131,
      "learning_rate": 2.0267597666530626e-06,
      "loss": 0.3799,
      "step": 23204
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.084636688232422,
      "learning_rate": 2.0260859064449577e-06,
      "loss": 0.294,
      "step": 23205
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9179582595825195,
      "learning_rate": 2.0254121456506327e-06,
      "loss": 0.2987,
      "step": 23206
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8032076358795166,
      "learning_rate": 2.024738484278488e-06,
      "loss": 0.2735,
      "step": 23207
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8920198678970337,
      "learning_rate": 2.024064922336919e-06,
      "loss": 0.3457,
      "step": 23208
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9654624462127686,
      "learning_rate": 2.0233914598343296e-06,
      "loss": 0.2672,
      "step": 23209
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8992266654968262,
      "learning_rate": 2.0227180967791116e-06,
      "loss": 0.3008,
      "step": 23210
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7104637622833252,
      "learning_rate": 2.0220448331796593e-06,
      "loss": 0.2692,
      "step": 23211
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9242494106292725,
      "learning_rate": 2.021371669044371e-06,
      "loss": 0.3729,
      "step": 23212
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0201637744903564,
      "learning_rate": 2.0206986043816366e-06,
      "loss": 0.2017,
      "step": 23213
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.1136929988861084,
      "learning_rate": 2.0200256391998452e-06,
      "loss": 0.3018,
      "step": 23214
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0539252758026123,
      "learning_rate": 2.0193527735073916e-06,
      "loss": 0.298,
      "step": 23215
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0608034133911133,
      "learning_rate": 2.0186800073126634e-06,
      "loss": 0.2969,
      "step": 23216
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7675158977508545,
      "learning_rate": 2.018007340624043e-06,
      "loss": 0.2713,
      "step": 23217
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.889601707458496,
      "learning_rate": 2.017334773449925e-06,
      "loss": 0.2956,
      "step": 23218
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.2949576377868652,
      "learning_rate": 2.01666230579869e-06,
      "loss": 0.3479,
      "step": 23219
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8199609518051147,
      "learning_rate": 2.015989937678723e-06,
      "loss": 0.2557,
      "step": 23220
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.381523609161377,
      "learning_rate": 2.0153176690984032e-06,
      "loss": 0.2264,
      "step": 23221
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.0533971786499023,
      "learning_rate": 2.014645500066117e-06,
      "loss": 0.331,
      "step": 23222
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.055391311645508,
      "learning_rate": 2.0139734305902436e-06,
      "loss": 0.3643,
      "step": 23223
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9217031002044678,
      "learning_rate": 2.0133014606791566e-06,
      "loss": 0.2548,
      "step": 23224
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.092578411102295,
      "learning_rate": 2.012629590341243e-06,
      "loss": 0.3766,
      "step": 23225
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6152492761611938,
      "learning_rate": 2.011957819584872e-06,
      "loss": 0.2974,
      "step": 23226
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8424659967422485,
      "learning_rate": 2.0112861484184197e-06,
      "loss": 0.266,
      "step": 23227
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.2282543182373047,
      "learning_rate": 2.0106145768502637e-06,
      "loss": 0.2884,
      "step": 23228
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7402945756912231,
      "learning_rate": 2.0099431048887753e-06,
      "loss": 0.3183,
      "step": 23229
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.061455488204956,
      "learning_rate": 2.0092717325423217e-06,
      "loss": 0.281,
      "step": 23230
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.218801259994507,
      "learning_rate": 2.008600459819279e-06,
      "loss": 0.2799,
      "step": 23231
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8049894571304321,
      "learning_rate": 2.0079292867280142e-06,
      "loss": 0.2362,
      "step": 23232
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.1501975059509277,
      "learning_rate": 2.007258213276895e-06,
      "loss": 0.3287,
      "step": 23233
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.39654278755188,
      "learning_rate": 2.0065872394742838e-06,
      "loss": 0.4265,
      "step": 23234
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6781142950057983,
      "learning_rate": 2.0059163653285528e-06,
      "loss": 0.3984,
      "step": 23235
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5150154829025269,
      "learning_rate": 2.005245590848064e-06,
      "loss": 0.2518,
      "step": 23236
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.755526065826416,
      "learning_rate": 2.0045749160411754e-06,
      "loss": 0.2792,
      "step": 23237
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.592774510383606,
      "learning_rate": 2.0039043409162564e-06,
      "loss": 0.2551,
      "step": 23238
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.316977024078369,
      "learning_rate": 2.003233865481662e-06,
      "loss": 0.3427,
      "step": 23239
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.661521553993225,
      "learning_rate": 2.002563489745749e-06,
      "loss": 0.254,
      "step": 23240
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8202629089355469,
      "learning_rate": 2.0018932137168835e-06,
      "loss": 0.3161,
      "step": 23241
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.0502686500549316,
      "learning_rate": 2.001223037403416e-06,
      "loss": 0.3945,
      "step": 23242
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.477400064468384,
      "learning_rate": 2.0005529608136997e-06,
      "loss": 0.401,
      "step": 23243
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.1056652069091797,
      "learning_rate": 1.9998829839560964e-06,
      "loss": 0.3107,
      "step": 23244
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.200044631958008,
      "learning_rate": 1.999213106838954e-06,
      "loss": 0.274,
      "step": 23245
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.4859862327575684,
      "learning_rate": 1.9985433294706246e-06,
      "loss": 0.3506,
      "step": 23246
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7325375080108643,
      "learning_rate": 1.9978736518594566e-06,
      "loss": 0.3606,
      "step": 23247
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.810577392578125,
      "learning_rate": 1.997204074013803e-06,
      "loss": 0.2524,
      "step": 23248
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.1949660778045654,
      "learning_rate": 1.99653459594201e-06,
      "loss": 0.3232,
      "step": 23249
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.9466979503631592,
      "learning_rate": 1.995865217652423e-06,
      "loss": 0.3425,
      "step": 23250
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6455681324005127,
      "learning_rate": 1.9951959391533894e-06,
      "loss": 0.3007,
      "step": 23251
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.2053022384643555,
      "learning_rate": 1.9945267604532535e-06,
      "loss": 0.3062,
      "step": 23252
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.701898217201233,
      "learning_rate": 1.993857681560355e-06,
      "loss": 0.295,
      "step": 23253
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7305707931518555,
      "learning_rate": 1.9931887024830397e-06,
      "loss": 0.2753,
      "step": 23254
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9020220041275024,
      "learning_rate": 1.9925198232296463e-06,
      "loss": 0.2943,
      "step": 23255
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2120752334594727,
      "learning_rate": 1.9918510438085115e-06,
      "loss": 0.2994,
      "step": 23256
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.096949577331543,
      "learning_rate": 1.991182364227977e-06,
      "loss": 0.2672,
      "step": 23257
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0324692726135254,
      "learning_rate": 1.9905137844963785e-06,
      "loss": 0.2879,
      "step": 23258
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1558074951171875,
      "learning_rate": 1.9898453046220513e-06,
      "loss": 0.3011,
      "step": 23259
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8127994537353516,
      "learning_rate": 1.9891769246133265e-06,
      "loss": 0.2326,
      "step": 23260
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.262826919555664,
      "learning_rate": 1.988508644478543e-06,
      "loss": 0.3653,
      "step": 23261
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3188486099243164,
      "learning_rate": 1.9878404642260274e-06,
      "loss": 0.2305,
      "step": 23262
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1139044761657715,
      "learning_rate": 1.987172383864111e-06,
      "loss": 0.2517,
      "step": 23263
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.5345253944396973,
      "learning_rate": 1.986504403401126e-06,
      "loss": 0.333,
      "step": 23264
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.833113670349121,
      "learning_rate": 1.985836522845398e-06,
      "loss": 0.3335,
      "step": 23265
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1442222595214844,
      "learning_rate": 1.985168742205252e-06,
      "loss": 0.3517,
      "step": 23266
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8066917657852173,
      "learning_rate": 1.984501061489018e-06,
      "loss": 0.2401,
      "step": 23267
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1792664527893066,
      "learning_rate": 1.9838334807050176e-06,
      "loss": 0.3074,
      "step": 23268
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.209033250808716,
      "learning_rate": 1.9831659998615713e-06,
      "loss": 0.3685,
      "step": 23269
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0559189319610596,
      "learning_rate": 1.982498618967007e-06,
      "loss": 0.3494,
      "step": 23270
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.15417218208313,
      "learning_rate": 1.9818313380296404e-06,
      "loss": 0.3433,
      "step": 23271
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.924845814704895,
      "learning_rate": 1.981164157057793e-06,
      "loss": 0.2965,
      "step": 23272
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.478191614151001,
      "learning_rate": 1.9804970760597784e-06,
      "loss": 0.3149,
      "step": 23273
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2859675884246826,
      "learning_rate": 1.9798300950439196e-06,
      "loss": 0.3302,
      "step": 23274
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6219717264175415,
      "learning_rate": 1.9791632140185303e-06,
      "loss": 0.3036,
      "step": 23275
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.437701940536499,
      "learning_rate": 1.97849643299192e-06,
      "loss": 0.3623,
      "step": 23276
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.970798969268799,
      "learning_rate": 1.9778297519724086e-06,
      "loss": 0.3697,
      "step": 23277
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7655612230300903,
      "learning_rate": 1.9771631709683047e-06,
      "loss": 0.2523,
      "step": 23278
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0411269664764404,
      "learning_rate": 1.9764966899879155e-06,
      "loss": 0.2059,
      "step": 23279
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8574600219726562,
      "learning_rate": 1.975830309039557e-06,
      "loss": 0.274,
      "step": 23280
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8241485357284546,
      "learning_rate": 1.9751640281315333e-06,
      "loss": 0.3174,
      "step": 23281
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9030178785324097,
      "learning_rate": 1.974497847272149e-06,
      "loss": 0.2874,
      "step": 23282
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5797151327133179,
      "learning_rate": 1.973831766469715e-06,
      "loss": 0.2133,
      "step": 23283
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.223533868789673,
      "learning_rate": 1.9731657857325326e-06,
      "loss": 0.3516,
      "step": 23284
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.473848581314087,
      "learning_rate": 1.9724999050689054e-06,
      "loss": 0.2608,
      "step": 23285
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6053067445755005,
      "learning_rate": 1.971834124487133e-06,
      "loss": 0.3503,
      "step": 23286
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9602928161621094,
      "learning_rate": 1.9711684439955194e-06,
      "loss": 0.3209,
      "step": 23287
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6527425050735474,
      "learning_rate": 1.9705028636023627e-06,
      "loss": 0.3007,
      "step": 23288
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0445122718811035,
      "learning_rate": 1.969837383315958e-06,
      "loss": 0.2465,
      "step": 23289
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5616472959518433,
      "learning_rate": 1.969172003144607e-06,
      "loss": 0.2885,
      "step": 23290
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5707049369812012,
      "learning_rate": 1.968506723096604e-06,
      "loss": 0.2637,
      "step": 23291
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8132894039154053,
      "learning_rate": 1.9678415431802377e-06,
      "loss": 0.2655,
      "step": 23292
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.3950698375701904,
      "learning_rate": 1.9671764634038092e-06,
      "loss": 0.3344,
      "step": 23293
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5793237686157227,
      "learning_rate": 1.9665114837756073e-06,
      "loss": 0.3541,
      "step": 23294
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.241657257080078,
      "learning_rate": 1.965846604303919e-06,
      "loss": 0.3051,
      "step": 23295
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9611395597457886,
      "learning_rate": 1.965181824997039e-06,
      "loss": 0.307,
      "step": 23296
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.125131368637085,
      "learning_rate": 1.9645171458632526e-06,
      "loss": 0.235,
      "step": 23297
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5777653455734253,
      "learning_rate": 1.9638525669108476e-06,
      "loss": 0.2415,
      "step": 23298
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5559158325195312,
      "learning_rate": 1.9631880881481068e-06,
      "loss": 0.2406,
      "step": 23299
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9822643995285034,
      "learning_rate": 1.9625237095833183e-06,
      "loss": 0.3359,
      "step": 23300
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6078708171844482,
      "learning_rate": 1.961859431224764e-06,
      "loss": 0.2881,
      "step": 23301
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.119711399078369,
      "learning_rate": 1.9611952530807222e-06,
      "loss": 0.3151,
      "step": 23302
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.892724871635437,
      "learning_rate": 1.960531175159479e-06,
      "loss": 0.3342,
      "step": 23303
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.500761032104492,
      "learning_rate": 1.9598671974693104e-06,
      "loss": 0.3573,
      "step": 23304
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.0202300548553467,
      "learning_rate": 1.9592033200184945e-06,
      "loss": 0.3503,
      "step": 23305
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0055510997772217,
      "learning_rate": 1.95853954281531e-06,
      "loss": 0.3431,
      "step": 23306
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9031598567962646,
      "learning_rate": 1.957875865868032e-06,
      "loss": 0.3162,
      "step": 23307
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.297976493835449,
      "learning_rate": 1.957212289184931e-06,
      "loss": 0.3488,
      "step": 23308
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8039718866348267,
      "learning_rate": 1.956548812774286e-06,
      "loss": 0.3062,
      "step": 23309
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2194113731384277,
      "learning_rate": 1.955885436644366e-06,
      "loss": 0.3544,
      "step": 23310
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.206059455871582,
      "learning_rate": 1.9552221608034407e-06,
      "loss": 0.4174,
      "step": 23311
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.3839569091796875,
      "learning_rate": 1.954558985259778e-06,
      "loss": 0.3295,
      "step": 23312
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8705216646194458,
      "learning_rate": 1.9538959100216513e-06,
      "loss": 0.2648,
      "step": 23313
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.092423915863037,
      "learning_rate": 1.953232935097323e-06,
      "loss": 0.2855,
      "step": 23314
    },
    {
      "epoch": 2.41,
      "grad_norm": 3.061312675476074,
      "learning_rate": 1.9525700604950594e-06,
      "loss": 0.3291,
      "step": 23315
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6159199476242065,
      "learning_rate": 1.951907286223127e-06,
      "loss": 0.2393,
      "step": 23316
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7798997163772583,
      "learning_rate": 1.9512446122897866e-06,
      "loss": 0.3594,
      "step": 23317
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1608521938323975,
      "learning_rate": 1.9505820387032994e-06,
      "loss": 0.2725,
      "step": 23318
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2845118045806885,
      "learning_rate": 1.949919565471928e-06,
      "loss": 0.2846,
      "step": 23319
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7324968576431274,
      "learning_rate": 1.949257192603932e-06,
      "loss": 0.3183,
      "step": 23320
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9363648891448975,
      "learning_rate": 1.9485949201075648e-06,
      "loss": 0.3841,
      "step": 23321
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.147732973098755,
      "learning_rate": 1.9479327479910903e-06,
      "loss": 0.431,
      "step": 23322
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.219963550567627,
      "learning_rate": 1.94727067626276e-06,
      "loss": 0.3919,
      "step": 23323
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.036008834838867,
      "learning_rate": 1.946608704930829e-06,
      "loss": 0.3264,
      "step": 23324
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.148815870285034,
      "learning_rate": 1.945946834003547e-06,
      "loss": 0.4018,
      "step": 23325
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9380046129226685,
      "learning_rate": 1.9452850634891716e-06,
      "loss": 0.3165,
      "step": 23326
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7430161237716675,
      "learning_rate": 1.944623393395951e-06,
      "loss": 0.2501,
      "step": 23327
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6289516687393188,
      "learning_rate": 1.9439618237321313e-06,
      "loss": 0.3028,
      "step": 23328
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6223526000976562,
      "learning_rate": 1.943300354505965e-06,
      "loss": 0.2732,
      "step": 23329
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.961571455001831,
      "learning_rate": 1.9426389857256977e-06,
      "loss": 0.2645,
      "step": 23330
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.4908103942871094,
      "learning_rate": 1.9419777173995714e-06,
      "loss": 0.3222,
      "step": 23331
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6669443845748901,
      "learning_rate": 1.9413165495358366e-06,
      "loss": 0.2843,
      "step": 23332
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7373499870300293,
      "learning_rate": 1.9406554821427327e-06,
      "loss": 0.2552,
      "step": 23333
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7834265232086182,
      "learning_rate": 1.939994515228499e-06,
      "loss": 0.3719,
      "step": 23334
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9290980100631714,
      "learning_rate": 1.9393336488013824e-06,
      "loss": 0.3598,
      "step": 23335
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.5344290733337402,
      "learning_rate": 1.9386728828696177e-06,
      "loss": 0.2766,
      "step": 23336
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.396070718765259,
      "learning_rate": 1.9380122174414453e-06,
      "loss": 0.3636,
      "step": 23337
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0164802074432373,
      "learning_rate": 1.9373516525250967e-06,
      "loss": 0.3939,
      "step": 23338
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8490982055664062,
      "learning_rate": 1.936691188128814e-06,
      "loss": 0.3335,
      "step": 23339
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8318120241165161,
      "learning_rate": 1.9360308242608283e-06,
      "loss": 0.2262,
      "step": 23340
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9254094362258911,
      "learning_rate": 1.9353705609293714e-06,
      "loss": 0.3258,
      "step": 23341
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.014528274536133,
      "learning_rate": 1.9347103981426786e-06,
      "loss": 0.3016,
      "step": 23342
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.6763980388641357,
      "learning_rate": 1.934050335908979e-06,
      "loss": 0.2786,
      "step": 23343
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.1955573558807373,
      "learning_rate": 1.933390374236498e-06,
      "loss": 0.3003,
      "step": 23344
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6653192043304443,
      "learning_rate": 1.9327305131334706e-06,
      "loss": 0.2719,
      "step": 23345
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9791173934936523,
      "learning_rate": 1.9320707526081183e-06,
      "loss": 0.2908,
      "step": 23346
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6553423404693604,
      "learning_rate": 1.9314110926686667e-06,
      "loss": 0.2941,
      "step": 23347
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.2385342121124268,
      "learning_rate": 1.930751533323344e-06,
      "loss": 0.3099,
      "step": 23348
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.553191065788269,
      "learning_rate": 1.93009207458037e-06,
      "loss": 0.2559,
      "step": 23349
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7258708477020264,
      "learning_rate": 1.9294327164479676e-06,
      "loss": 0.2255,
      "step": 23350
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7318323850631714,
      "learning_rate": 1.9287734589343533e-06,
      "loss": 0.2112,
      "step": 23351
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4442617893218994,
      "learning_rate": 1.9281143020477534e-06,
      "loss": 0.3521,
      "step": 23352
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.3008790016174316,
      "learning_rate": 1.927455245796381e-06,
      "loss": 0.368,
      "step": 23353
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.342864513397217,
      "learning_rate": 1.926796290188452e-06,
      "loss": 0.3172,
      "step": 23354
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.3709752559661865,
      "learning_rate": 1.926137435232187e-06,
      "loss": 0.3289,
      "step": 23355
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0213208198547363,
      "learning_rate": 1.9254786809357963e-06,
      "loss": 0.274,
      "step": 23356
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8566036224365234,
      "learning_rate": 1.9248200273074913e-06,
      "loss": 0.3495,
      "step": 23357
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9938058853149414,
      "learning_rate": 1.9241614743554882e-06,
      "loss": 0.3098,
      "step": 23358
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7901051044464111,
      "learning_rate": 1.9235030220879945e-06,
      "loss": 0.2892,
      "step": 23359
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7373645305633545,
      "learning_rate": 1.9228446705132186e-06,
      "loss": 0.268,
      "step": 23360
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5614484548568726,
      "learning_rate": 1.9221864196393713e-06,
      "loss": 0.3141,
      "step": 23361
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4050405025482178,
      "learning_rate": 1.9215282694746596e-06,
      "loss": 0.2906,
      "step": 23362
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.748022198677063,
      "learning_rate": 1.9208702200272856e-06,
      "loss": 0.3678,
      "step": 23363
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7788243293762207,
      "learning_rate": 1.9202122713054528e-06,
      "loss": 0.3087,
      "step": 23364
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8607892990112305,
      "learning_rate": 1.919554423317369e-06,
      "loss": 0.3025,
      "step": 23365
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4176957607269287,
      "learning_rate": 1.9188966760712323e-06,
      "loss": 0.3092,
      "step": 23366
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.263357639312744,
      "learning_rate": 1.918239029575242e-06,
      "loss": 0.339,
      "step": 23367
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.694295048713684,
      "learning_rate": 1.9175814838376016e-06,
      "loss": 0.3444,
      "step": 23368
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6585931777954102,
      "learning_rate": 1.9169240388665066e-06,
      "loss": 0.293,
      "step": 23369
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.3841888904571533,
      "learning_rate": 1.9162666946701503e-06,
      "loss": 0.345,
      "step": 23370
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0565185546875,
      "learning_rate": 1.915609451256735e-06,
      "loss": 0.2839,
      "step": 23371
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7097582817077637,
      "learning_rate": 1.9149523086344502e-06,
      "loss": 0.2863,
      "step": 23372
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9107764959335327,
      "learning_rate": 1.914295266811487e-06,
      "loss": 0.265,
      "step": 23373
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8130218982696533,
      "learning_rate": 1.913638325796042e-06,
      "loss": 0.3111,
      "step": 23374
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9733538627624512,
      "learning_rate": 1.912981485596304e-06,
      "loss": 0.2403,
      "step": 23375
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6930053234100342,
      "learning_rate": 1.91232474622046e-06,
      "loss": 0.3603,
      "step": 23376
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8217229843139648,
      "learning_rate": 1.911668107676696e-06,
      "loss": 0.3188,
      "step": 23377
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0590410232543945,
      "learning_rate": 1.9110115699732056e-06,
      "loss": 0.3253,
      "step": 23378
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.5432393550872803,
      "learning_rate": 1.91035513311817e-06,
      "loss": 0.285,
      "step": 23379
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6832456588745117,
      "learning_rate": 1.90969879711977e-06,
      "loss": 0.2944,
      "step": 23380
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7696491479873657,
      "learning_rate": 1.909042561986194e-06,
      "loss": 0.2649,
      "step": 23381
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7288466691970825,
      "learning_rate": 1.9083864277256224e-06,
      "loss": 0.2851,
      "step": 23382
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1133689880371094,
      "learning_rate": 1.907730394346231e-06,
      "loss": 0.3496,
      "step": 23383
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0982446670532227,
      "learning_rate": 1.9070744618562043e-06,
      "loss": 0.3522,
      "step": 23384
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5848195552825928,
      "learning_rate": 1.9064186302637177e-06,
      "loss": 0.2525,
      "step": 23385
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9759063720703125,
      "learning_rate": 1.9057628995769461e-06,
      "loss": 0.3411,
      "step": 23386
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4110608100891113,
      "learning_rate": 1.9051072698040685e-06,
      "loss": 0.2866,
      "step": 23387
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1390628814697266,
      "learning_rate": 1.904451740953257e-06,
      "loss": 0.3948,
      "step": 23388
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5475046634674072,
      "learning_rate": 1.9037963130326853e-06,
      "loss": 0.244,
      "step": 23389
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9368089437484741,
      "learning_rate": 1.90314098605052e-06,
      "loss": 0.2664,
      "step": 23390
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7244755029678345,
      "learning_rate": 1.9024857600149372e-06,
      "loss": 0.2808,
      "step": 23391
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7706910371780396,
      "learning_rate": 1.9018306349341053e-06,
      "loss": 0.3048,
      "step": 23392
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.257505178451538,
      "learning_rate": 1.901175610816186e-06,
      "loss": 0.3382,
      "step": 23393
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9961225986480713,
      "learning_rate": 1.9005206876693538e-06,
      "loss": 0.3413,
      "step": 23394
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.012019157409668,
      "learning_rate": 1.8998658655017698e-06,
      "loss": 0.3535,
      "step": 23395
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.033358573913574,
      "learning_rate": 1.8992111443215965e-06,
      "loss": 0.3456,
      "step": 23396
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6071120500564575,
      "learning_rate": 1.8985565241370008e-06,
      "loss": 0.2458,
      "step": 23397
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6929062604904175,
      "learning_rate": 1.8979020049561413e-06,
      "loss": 0.2473,
      "step": 23398
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9667021036148071,
      "learning_rate": 1.8972475867871753e-06,
      "loss": 0.3154,
      "step": 23399
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9795069694519043,
      "learning_rate": 1.8965932696382684e-06,
      "loss": 0.2709,
      "step": 23400
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.985343098640442,
      "learning_rate": 1.8959390535175749e-06,
      "loss": 0.3255,
      "step": 23401
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8901188373565674,
      "learning_rate": 1.8952849384332507e-06,
      "loss": 0.2531,
      "step": 23402
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.173654556274414,
      "learning_rate": 1.894630924393448e-06,
      "loss": 0.3695,
      "step": 23403
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6234561204910278,
      "learning_rate": 1.8939770114063271e-06,
      "loss": 0.25,
      "step": 23404
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1757872104644775,
      "learning_rate": 1.8933231994800372e-06,
      "loss": 0.3075,
      "step": 23405
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1137940883636475,
      "learning_rate": 1.8926694886227259e-06,
      "loss": 0.3281,
      "step": 23406
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8237922191619873,
      "learning_rate": 1.8920158788425513e-06,
      "loss": 0.3355,
      "step": 23407
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0547938346862793,
      "learning_rate": 1.891362370147657e-06,
      "loss": 0.401,
      "step": 23408
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5443590879440308,
      "learning_rate": 1.8907089625461894e-06,
      "loss": 0.2375,
      "step": 23409
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0936601161956787,
      "learning_rate": 1.890055656046299e-06,
      "loss": 0.3215,
      "step": 23410
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5816826820373535,
      "learning_rate": 1.8894024506561294e-06,
      "loss": 0.2722,
      "step": 23411
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7900381088256836,
      "learning_rate": 1.8887493463838201e-06,
      "loss": 0.278,
      "step": 23412
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.186249256134033,
      "learning_rate": 1.888096343237521e-06,
      "loss": 0.3173,
      "step": 23413
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.876712679862976,
      "learning_rate": 1.8874434412253684e-06,
      "loss": 0.2807,
      "step": 23414
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.6789915561676025,
      "learning_rate": 1.8867906403555036e-06,
      "loss": 0.3716,
      "step": 23415
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5472846031188965,
      "learning_rate": 1.8861379406360625e-06,
      "loss": 0.3128,
      "step": 23416
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8616706132888794,
      "learning_rate": 1.8854853420751883e-06,
      "loss": 0.3327,
      "step": 23417
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1744682788848877,
      "learning_rate": 1.8848328446810137e-06,
      "loss": 0.3267,
      "step": 23418
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8642526865005493,
      "learning_rate": 1.884180448461671e-06,
      "loss": 0.3237,
      "step": 23419
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.720702886581421,
      "learning_rate": 1.883528153425299e-06,
      "loss": 0.2801,
      "step": 23420
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.099637985229492,
      "learning_rate": 1.8828759595800284e-06,
      "loss": 0.3025,
      "step": 23421
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5923001766204834,
      "learning_rate": 1.8822238669339876e-06,
      "loss": 0.2928,
      "step": 23422
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0344889163970947,
      "learning_rate": 1.8815718754953117e-06,
      "loss": 0.3218,
      "step": 23423
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.070831298828125,
      "learning_rate": 1.880919985272125e-06,
      "loss": 0.3324,
      "step": 23424
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1618165969848633,
      "learning_rate": 1.8802681962725545e-06,
      "loss": 0.3117,
      "step": 23425
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0057244300842285,
      "learning_rate": 1.8796165085047314e-06,
      "loss": 0.3619,
      "step": 23426
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8566430807113647,
      "learning_rate": 1.8789649219767758e-06,
      "loss": 0.2979,
      "step": 23427
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8897193670272827,
      "learning_rate": 1.8783134366968126e-06,
      "loss": 0.2136,
      "step": 23428
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.232651472091675,
      "learning_rate": 1.8776620526729628e-06,
      "loss": 0.2814,
      "step": 23429
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8944181203842163,
      "learning_rate": 1.8770107699133499e-06,
      "loss": 0.3222,
      "step": 23430
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.4047818183898926,
      "learning_rate": 1.8763595884260943e-06,
      "loss": 0.2478,
      "step": 23431
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1335766315460205,
      "learning_rate": 1.8757085082193094e-06,
      "loss": 0.3237,
      "step": 23432
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8396086692810059,
      "learning_rate": 1.8750575293011175e-06,
      "loss": 0.399,
      "step": 23433
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.2258148193359375,
      "learning_rate": 1.8744066516796343e-06,
      "loss": 0.3014,
      "step": 23434
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.2578017711639404,
      "learning_rate": 1.8737558753629702e-06,
      "loss": 0.2943,
      "step": 23435
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.2149555683135986,
      "learning_rate": 1.8731052003592442e-06,
      "loss": 0.2934,
      "step": 23436
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9793379306793213,
      "learning_rate": 1.8724546266765653e-06,
      "loss": 0.2948,
      "step": 23437
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9530949592590332,
      "learning_rate": 1.8718041543230436e-06,
      "loss": 0.3398,
      "step": 23438
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.171578884124756,
      "learning_rate": 1.8711537833067927e-06,
      "loss": 0.3191,
      "step": 23439
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0096678733825684,
      "learning_rate": 1.8705035136359185e-06,
      "loss": 0.295,
      "step": 23440
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.3608744144439697,
      "learning_rate": 1.8698533453185286e-06,
      "loss": 0.3493,
      "step": 23441
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8978826999664307,
      "learning_rate": 1.8692032783627246e-06,
      "loss": 0.2645,
      "step": 23442
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5786118507385254,
      "learning_rate": 1.8685533127766198e-06,
      "loss": 0.3125,
      "step": 23443
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.793956995010376,
      "learning_rate": 1.8679034485683112e-06,
      "loss": 0.2839,
      "step": 23444
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8150341510772705,
      "learning_rate": 1.8672536857459012e-06,
      "loss": 0.3215,
      "step": 23445
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.9153624773025513,
      "learning_rate": 1.8666040243174932e-06,
      "loss": 0.2584,
      "step": 23446
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7733988761901855,
      "learning_rate": 1.8659544642911864e-06,
      "loss": 0.3008,
      "step": 23447
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.936020851135254,
      "learning_rate": 1.8653050056750755e-06,
      "loss": 0.3449,
      "step": 23448
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.146099805831909,
      "learning_rate": 1.8646556484772638e-06,
      "loss": 0.3094,
      "step": 23449
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8629834651947021,
      "learning_rate": 1.8640063927058427e-06,
      "loss": 0.3326,
      "step": 23450
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.0314853191375732,
      "learning_rate": 1.8633572383689058e-06,
      "loss": 0.284,
      "step": 23451
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.209036111831665,
      "learning_rate": 1.86270818547455e-06,
      "loss": 0.3421,
      "step": 23452
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.528063416481018,
      "learning_rate": 1.862059234030865e-06,
      "loss": 0.2312,
      "step": 23453
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9831106662750244,
      "learning_rate": 1.861410384045943e-06,
      "loss": 0.3252,
      "step": 23454
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9867626428604126,
      "learning_rate": 1.8607616355278702e-06,
      "loss": 0.2478,
      "step": 23455
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9971468448638916,
      "learning_rate": 1.860112988484739e-06,
      "loss": 0.2978,
      "step": 23456
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.618224859237671,
      "learning_rate": 1.8594644429246344e-06,
      "loss": 0.2588,
      "step": 23457
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.2278635501861572,
      "learning_rate": 1.8588159988556398e-06,
      "loss": 0.3081,
      "step": 23458
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.2185068130493164,
      "learning_rate": 1.858167656285843e-06,
      "loss": 0.3941,
      "step": 23459
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6802330017089844,
      "learning_rate": 1.8575194152233279e-06,
      "loss": 0.2996,
      "step": 23460
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6109066009521484,
      "learning_rate": 1.8568712756761708e-06,
      "loss": 0.2975,
      "step": 23461
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8283674716949463,
      "learning_rate": 1.856223237652458e-06,
      "loss": 0.2902,
      "step": 23462
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.44944167137146,
      "learning_rate": 1.8555753011602683e-06,
      "loss": 0.2955,
      "step": 23463
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.022754430770874,
      "learning_rate": 1.8549274662076744e-06,
      "loss": 0.301,
      "step": 23464
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.231464385986328,
      "learning_rate": 1.8542797328027595e-06,
      "loss": 0.3401,
      "step": 23465
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.66543447971344,
      "learning_rate": 1.8536321009535974e-06,
      "loss": 0.2717,
      "step": 23466
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.98613703250885,
      "learning_rate": 1.8529845706682603e-06,
      "loss": 0.2749,
      "step": 23467
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.941970705986023,
      "learning_rate": 1.8523371419548208e-06,
      "loss": 0.2652,
      "step": 23468
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9278976917266846,
      "learning_rate": 1.8516898148213546e-06,
      "loss": 0.3115,
      "step": 23469
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.17116641998291,
      "learning_rate": 1.8510425892759298e-06,
      "loss": 0.2786,
      "step": 23470
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.7805233001708984,
      "learning_rate": 1.8503954653266132e-06,
      "loss": 0.3629,
      "step": 23471
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.927001953125,
      "learning_rate": 1.8497484429814783e-06,
      "loss": 0.2687,
      "step": 23472
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.1951534748077393,
      "learning_rate": 1.8491015222485875e-06,
      "loss": 0.2861,
      "step": 23473
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.301680326461792,
      "learning_rate": 1.8484547031360056e-06,
      "loss": 0.3029,
      "step": 23474
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7154555320739746,
      "learning_rate": 1.8478079856518005e-06,
      "loss": 0.2948,
      "step": 23475
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8531434535980225,
      "learning_rate": 1.8471613698040336e-06,
      "loss": 0.2918,
      "step": 23476
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.398010015487671,
      "learning_rate": 1.8465148556007651e-06,
      "loss": 0.3563,
      "step": 23477
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7202602624893188,
      "learning_rate": 1.8458684430500552e-06,
      "loss": 0.3257,
      "step": 23478
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.1501028537750244,
      "learning_rate": 1.845222132159965e-06,
      "loss": 0.3431,
      "step": 23479
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8729751110076904,
      "learning_rate": 1.8445759229385519e-06,
      "loss": 0.2679,
      "step": 23480
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7799586057662964,
      "learning_rate": 1.843929815393869e-06,
      "loss": 0.2743,
      "step": 23481
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9792542457580566,
      "learning_rate": 1.8432838095339766e-06,
      "loss": 0.293,
      "step": 23482
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7631831169128418,
      "learning_rate": 1.8426379053669274e-06,
      "loss": 0.2384,
      "step": 23483
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9133440256118774,
      "learning_rate": 1.8419921029007702e-06,
      "loss": 0.3261,
      "step": 23484
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9525967836380005,
      "learning_rate": 1.8413464021435624e-06,
      "loss": 0.3502,
      "step": 23485
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.257143974304199,
      "learning_rate": 1.8407008031033502e-06,
      "loss": 0.3261,
      "step": 23486
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7415680885314941,
      "learning_rate": 1.840055305788182e-06,
      "loss": 0.2407,
      "step": 23487
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.019705057144165,
      "learning_rate": 1.8394099102061092e-06,
      "loss": 0.2753,
      "step": 23488
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.715818166732788,
      "learning_rate": 1.838764616365175e-06,
      "loss": 0.2247,
      "step": 23489
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8340469598770142,
      "learning_rate": 1.8381194242734269e-06,
      "loss": 0.2867,
      "step": 23490
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.5014290809631348,
      "learning_rate": 1.8374743339389044e-06,
      "loss": 0.3095,
      "step": 23491
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.295377254486084,
      "learning_rate": 1.8368293453696562e-06,
      "loss": 0.3509,
      "step": 23492
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.2035140991210938,
      "learning_rate": 1.8361844585737198e-06,
      "loss": 0.3318,
      "step": 23493
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6416021585464478,
      "learning_rate": 1.8355396735591335e-06,
      "loss": 0.2839,
      "step": 23494
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.298591136932373,
      "learning_rate": 1.8348949903339409e-06,
      "loss": 0.3045,
      "step": 23495
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.1289238929748535,
      "learning_rate": 1.8342504089061775e-06,
      "loss": 0.326,
      "step": 23496
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9040440320968628,
      "learning_rate": 1.833605929283877e-06,
      "loss": 0.3496,
      "step": 23497
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7023015022277832,
      "learning_rate": 1.8329615514750775e-06,
      "loss": 0.2781,
      "step": 23498
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8324804306030273,
      "learning_rate": 1.8323172754878138e-06,
      "loss": 0.308,
      "step": 23499
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8587592840194702,
      "learning_rate": 1.8316731013301126e-06,
      "loss": 0.335,
      "step": 23500
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.2501060962677,
      "learning_rate": 1.8310290290100107e-06,
      "loss": 0.269,
      "step": 23501
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8980220556259155,
      "learning_rate": 1.8303850585355365e-06,
      "loss": 0.2608,
      "step": 23502
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.207979202270508,
      "learning_rate": 1.8297411899147188e-06,
      "loss": 0.3339,
      "step": 23503
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9291245937347412,
      "learning_rate": 1.8290974231555814e-06,
      "loss": 0.3009,
      "step": 23504
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.2185616493225098,
      "learning_rate": 1.8284537582661555e-06,
      "loss": 0.3473,
      "step": 23505
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7422235012054443,
      "learning_rate": 1.8278101952544637e-06,
      "loss": 0.3575,
      "step": 23506
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.198040246963501,
      "learning_rate": 1.8271667341285272e-06,
      "loss": 0.3876,
      "step": 23507
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.0956568717956543,
      "learning_rate": 1.826523374896373e-06,
      "loss": 0.3416,
      "step": 23508
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9398711919784546,
      "learning_rate": 1.8258801175660191e-06,
      "loss": 0.4006,
      "step": 23509
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9134070873260498,
      "learning_rate": 1.8252369621454836e-06,
      "loss": 0.2622,
      "step": 23510
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.2860116958618164,
      "learning_rate": 1.82459390864279e-06,
      "loss": 0.3589,
      "step": 23511
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.806628704071045,
      "learning_rate": 1.823950957065953e-06,
      "loss": 0.295,
      "step": 23512
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.4367763996124268,
      "learning_rate": 1.8233081074229852e-06,
      "loss": 0.3467,
      "step": 23513
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8119149208068848,
      "learning_rate": 1.8226653597219069e-06,
      "loss": 0.27,
      "step": 23514
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.3651115894317627,
      "learning_rate": 1.8220227139707281e-06,
      "loss": 0.2242,
      "step": 23515
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9346203804016113,
      "learning_rate": 1.8213801701774624e-06,
      "loss": 0.2592,
      "step": 23516
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.065977096557617,
      "learning_rate": 1.820737728350117e-06,
      "loss": 0.3635,
      "step": 23517
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.12964129447937,
      "learning_rate": 1.8200953884967077e-06,
      "loss": 0.3773,
      "step": 23518
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.0894429683685303,
      "learning_rate": 1.8194531506252388e-06,
      "loss": 0.3592,
      "step": 23519
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.3541407585144043,
      "learning_rate": 1.8188110147437155e-06,
      "loss": 0.337,
      "step": 23520
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.6212596893310547,
      "learning_rate": 1.8181689808601477e-06,
      "loss": 0.3928,
      "step": 23521
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.270371913909912,
      "learning_rate": 1.8175270489825392e-06,
      "loss": 0.312,
      "step": 23522
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6566849946975708,
      "learning_rate": 1.816885219118889e-06,
      "loss": 0.2036,
      "step": 23523
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.0080032348632812,
      "learning_rate": 1.8162434912772053e-06,
      "loss": 0.3446,
      "step": 23524
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8499921560287476,
      "learning_rate": 1.815601865465485e-06,
      "loss": 0.3031,
      "step": 23525
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9065005779266357,
      "learning_rate": 1.814960341691726e-06,
      "loss": 0.3025,
      "step": 23526
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.197068214416504,
      "learning_rate": 1.814318919963931e-06,
      "loss": 0.3181,
      "step": 23527
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.144009828567505,
      "learning_rate": 1.8136776002900948e-06,
      "loss": 0.329,
      "step": 23528
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8695757389068604,
      "learning_rate": 1.8130363826782115e-06,
      "loss": 0.3233,
      "step": 23529
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.352630138397217,
      "learning_rate": 1.8123952671362744e-06,
      "loss": 0.3164,
      "step": 23530
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.575927734375,
      "learning_rate": 1.8117542536722821e-06,
      "loss": 0.3143,
      "step": 23531
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.78476083278656,
      "learning_rate": 1.8111133422942218e-06,
      "loss": 0.3201,
      "step": 23532
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.0612990856170654,
      "learning_rate": 1.8104725330100825e-06,
      "loss": 0.3085,
      "step": 23533
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.4893782138824463,
      "learning_rate": 1.809831825827859e-06,
      "loss": 0.318,
      "step": 23534
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.1159114837646484,
      "learning_rate": 1.8091912207555361e-06,
      "loss": 0.3079,
      "step": 23535
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8168491125106812,
      "learning_rate": 1.8085507178010985e-06,
      "loss": 0.2858,
      "step": 23536
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.3392174243927,
      "learning_rate": 1.8079103169725353e-06,
      "loss": 0.3077,
      "step": 23537
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8959119319915771,
      "learning_rate": 1.807270018277829e-06,
      "loss": 0.2606,
      "step": 23538
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.0002429485321045,
      "learning_rate": 1.8066298217249612e-06,
      "loss": 0.3014,
      "step": 23539
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.9318300485610962,
      "learning_rate": 1.8059897273219151e-06,
      "loss": 0.4238,
      "step": 23540
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.1783368587493896,
      "learning_rate": 1.8053497350766724e-06,
      "loss": 0.3403,
      "step": 23541
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.0459232330322266,
      "learning_rate": 1.80470984499721e-06,
      "loss": 0.3335,
      "step": 23542
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.747152328491211,
      "learning_rate": 1.8040700570915026e-06,
      "loss": 0.2858,
      "step": 23543
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6080129146575928,
      "learning_rate": 1.8034303713675328e-06,
      "loss": 0.2699,
      "step": 23544
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.264974594116211,
      "learning_rate": 1.802790787833273e-06,
      "loss": 0.2848,
      "step": 23545
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8081128597259521,
      "learning_rate": 1.8021513064966955e-06,
      "loss": 0.2793,
      "step": 23546
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9255578517913818,
      "learning_rate": 1.8015119273657766e-06,
      "loss": 0.2562,
      "step": 23547
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.360729455947876,
      "learning_rate": 1.8008726504484852e-06,
      "loss": 0.3284,
      "step": 23548
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.970847725868225,
      "learning_rate": 1.8002334757527894e-06,
      "loss": 0.2454,
      "step": 23549
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8380563259124756,
      "learning_rate": 1.7995944032866641e-06,
      "loss": 0.3475,
      "step": 23550
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.983770489692688,
      "learning_rate": 1.7989554330580726e-06,
      "loss": 0.2862,
      "step": 23551
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8417918682098389,
      "learning_rate": 1.798316565074979e-06,
      "loss": 0.218,
      "step": 23552
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.701111078262329,
      "learning_rate": 1.7976777993453542e-06,
      "loss": 0.2809,
      "step": 23553
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.893062114715576,
      "learning_rate": 1.7970391358771588e-06,
      "loss": 0.3707,
      "step": 23554
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9218963384628296,
      "learning_rate": 1.7964005746783553e-06,
      "loss": 0.2954,
      "step": 23555
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0016930103302,
      "learning_rate": 1.795762115756905e-06,
      "loss": 0.2932,
      "step": 23556
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.003962755203247,
      "learning_rate": 1.795123759120767e-06,
      "loss": 0.2794,
      "step": 23557
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.843271255493164,
      "learning_rate": 1.7944855047778997e-06,
      "loss": 0.2907,
      "step": 23558
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.748515009880066,
      "learning_rate": 1.7938473527362632e-06,
      "loss": 0.3268,
      "step": 23559
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0641233921051025,
      "learning_rate": 1.793209303003811e-06,
      "loss": 0.3455,
      "step": 23560
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.139490842819214,
      "learning_rate": 1.792571355588497e-06,
      "loss": 0.3203,
      "step": 23561
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.681028962135315,
      "learning_rate": 1.7919335104982794e-06,
      "loss": 0.229,
      "step": 23562
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.163787364959717,
      "learning_rate": 1.791295767741107e-06,
      "loss": 0.3599,
      "step": 23563
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8936885595321655,
      "learning_rate": 1.7906581273249313e-06,
      "loss": 0.3346,
      "step": 23564
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8543530702590942,
      "learning_rate": 1.7900205892576995e-06,
      "loss": 0.359,
      "step": 23565
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8964914083480835,
      "learning_rate": 1.789383153547365e-06,
      "loss": 0.3314,
      "step": 23566
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.00637149810791,
      "learning_rate": 1.7887458202018726e-06,
      "loss": 0.3434,
      "step": 23567
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.932451605796814,
      "learning_rate": 1.788108589229165e-06,
      "loss": 0.3368,
      "step": 23568
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7076464891433716,
      "learning_rate": 1.7874714606371935e-06,
      "loss": 0.2817,
      "step": 23569
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9057267904281616,
      "learning_rate": 1.7868344344338962e-06,
      "loss": 0.3279,
      "step": 23570
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0260000228881836,
      "learning_rate": 1.7861975106272156e-06,
      "loss": 0.3142,
      "step": 23571
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.204639196395874,
      "learning_rate": 1.7855606892250953e-06,
      "loss": 0.3332,
      "step": 23572
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8443200588226318,
      "learning_rate": 1.7849239702354737e-06,
      "loss": 0.3613,
      "step": 23573
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.903350830078125,
      "learning_rate": 1.7842873536662852e-06,
      "loss": 0.2911,
      "step": 23574
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.129530191421509,
      "learning_rate": 1.7836508395254726e-06,
      "loss": 0.2957,
      "step": 23575
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8569594621658325,
      "learning_rate": 1.7830144278209694e-06,
      "loss": 0.2651,
      "step": 23576
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7843937873840332,
      "learning_rate": 1.7823781185607082e-06,
      "loss": 0.315,
      "step": 23577
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.051191806793213,
      "learning_rate": 1.7817419117526213e-06,
      "loss": 0.2747,
      "step": 23578
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.84977126121521,
      "learning_rate": 1.781105807404646e-06,
      "loss": 0.3637,
      "step": 23579
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9316809177398682,
      "learning_rate": 1.7804698055247094e-06,
      "loss": 0.3121,
      "step": 23580
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0252633094787598,
      "learning_rate": 1.779833906120737e-06,
      "loss": 0.3799,
      "step": 23581
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.132699966430664,
      "learning_rate": 1.779198109200665e-06,
      "loss": 0.3207,
      "step": 23582
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6921883821487427,
      "learning_rate": 1.7785624147724145e-06,
      "loss": 0.3333,
      "step": 23583
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.983875036239624,
      "learning_rate": 1.7779268228439096e-06,
      "loss": 0.2949,
      "step": 23584
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8546559810638428,
      "learning_rate": 1.7772913334230802e-06,
      "loss": 0.3302,
      "step": 23585
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.1141204833984375,
      "learning_rate": 1.7766559465178456e-06,
      "loss": 0.3473,
      "step": 23586
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.928503394126892,
      "learning_rate": 1.7760206621361254e-06,
      "loss": 0.4207,
      "step": 23587
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2343342304229736,
      "learning_rate": 1.775385480285845e-06,
      "loss": 0.2157,
      "step": 23588
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2825405597686768,
      "learning_rate": 1.7747504009749205e-06,
      "loss": 0.3476,
      "step": 23589
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9176554679870605,
      "learning_rate": 1.77411542421127e-06,
      "loss": 0.3352,
      "step": 23590
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0443179607391357,
      "learning_rate": 1.7734805500028073e-06,
      "loss": 0.37,
      "step": 23591
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9167215824127197,
      "learning_rate": 1.7728457783574537e-06,
      "loss": 0.3208,
      "step": 23592
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.156442642211914,
      "learning_rate": 1.7722111092831186e-06,
      "loss": 0.3297,
      "step": 23593
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8021847009658813,
      "learning_rate": 1.7715765427877141e-06,
      "loss": 0.2939,
      "step": 23594
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7841025590896606,
      "learning_rate": 1.7709420788791554e-06,
      "loss": 0.3167,
      "step": 23595
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.034191608428955,
      "learning_rate": 1.7703077175653504e-06,
      "loss": 0.3887,
      "step": 23596
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7950949668884277,
      "learning_rate": 1.7696734588542063e-06,
      "loss": 0.2738,
      "step": 23597
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.339587926864624,
      "learning_rate": 1.7690393027536334e-06,
      "loss": 0.3653,
      "step": 23598
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.2556955814361572,
      "learning_rate": 1.7684052492715386e-06,
      "loss": 0.2769,
      "step": 23599
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.410792112350464,
      "learning_rate": 1.7677712984158213e-06,
      "loss": 0.2945,
      "step": 23600
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.176870584487915,
      "learning_rate": 1.7671374501943916e-06,
      "loss": 0.3562,
      "step": 23601
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7496609687805176,
      "learning_rate": 1.76650370461515e-06,
      "loss": 0.1989,
      "step": 23602
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7189496755599976,
      "learning_rate": 1.765870061685997e-06,
      "loss": 0.231,
      "step": 23603
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6480462551116943,
      "learning_rate": 1.7652365214148304e-06,
      "loss": 0.3064,
      "step": 23604
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9693989753723145,
      "learning_rate": 1.7646030838095529e-06,
      "loss": 0.3374,
      "step": 23605
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8019005060195923,
      "learning_rate": 1.7639697488780595e-06,
      "loss": 0.3283,
      "step": 23606
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0509989261627197,
      "learning_rate": 1.7633365166282446e-06,
      "loss": 0.3137,
      "step": 23607
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.997462272644043,
      "learning_rate": 1.7627033870680066e-06,
      "loss": 0.3583,
      "step": 23608
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8923307657241821,
      "learning_rate": 1.762070360205238e-06,
      "loss": 0.2383,
      "step": 23609
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0185179710388184,
      "learning_rate": 1.7614374360478271e-06,
      "loss": 0.2697,
      "step": 23610
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7884527444839478,
      "learning_rate": 1.7608046146036706e-06,
      "loss": 0.3077,
      "step": 23611
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7220755815505981,
      "learning_rate": 1.7601718958806558e-06,
      "loss": 0.2727,
      "step": 23612
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5642822980880737,
      "learning_rate": 1.7595392798866683e-06,
      "loss": 0.288,
      "step": 23613
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5250407457351685,
      "learning_rate": 1.7589067666296e-06,
      "loss": 0.2709,
      "step": 23614
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8812167644500732,
      "learning_rate": 1.7582743561173332e-06,
      "loss": 0.3025,
      "step": 23615
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0474956035614014,
      "learning_rate": 1.757642048357755e-06,
      "loss": 0.3365,
      "step": 23616
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.1107161045074463,
      "learning_rate": 1.7570098433587435e-06,
      "loss": 0.3113,
      "step": 23617
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7787898778915405,
      "learning_rate": 1.7563777411281869e-06,
      "loss": 0.2706,
      "step": 23618
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.096191883087158,
      "learning_rate": 1.7557457416739631e-06,
      "loss": 0.277,
      "step": 23619
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.094252347946167,
      "learning_rate": 1.7551138450039485e-06,
      "loss": 0.3188,
      "step": 23620
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.7440083026885986,
      "learning_rate": 1.7544820511260274e-06,
      "loss": 0.3639,
      "step": 23621
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.100005626678467,
      "learning_rate": 1.7538503600480727e-06,
      "loss": 0.2408,
      "step": 23622
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8409358263015747,
      "learning_rate": 1.7532187717779591e-06,
      "loss": 0.3049,
      "step": 23623
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6211895942687988,
      "learning_rate": 1.7525872863235649e-06,
      "loss": 0.3395,
      "step": 23624
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.344122886657715,
      "learning_rate": 1.7519559036927592e-06,
      "loss": 0.3825,
      "step": 23625
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7109687328338623,
      "learning_rate": 1.7513246238934134e-06,
      "loss": 0.3091,
      "step": 23626
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.170010805130005,
      "learning_rate": 1.7506934469334025e-06,
      "loss": 0.2975,
      "step": 23627
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.001607656478882,
      "learning_rate": 1.7500623728205922e-06,
      "loss": 0.2961,
      "step": 23628
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7834044694900513,
      "learning_rate": 1.749431401562851e-06,
      "loss": 0.3089,
      "step": 23629
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7224931716918945,
      "learning_rate": 1.7488005331680425e-06,
      "loss": 0.3206,
      "step": 23630
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.8840267658233643,
      "learning_rate": 1.7481697676440379e-06,
      "loss": 0.3679,
      "step": 23631
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.123333215713501,
      "learning_rate": 1.7475391049986967e-06,
      "loss": 0.3041,
      "step": 23632
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7306216955184937,
      "learning_rate": 1.7469085452398816e-06,
      "loss": 0.3289,
      "step": 23633
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.882979393005371,
      "learning_rate": 1.7462780883754571e-06,
      "loss": 0.4741,
      "step": 23634
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.5097508430480957,
      "learning_rate": 1.7456477344132816e-06,
      "loss": 0.3508,
      "step": 23635
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.418330669403076,
      "learning_rate": 1.7450174833612121e-06,
      "loss": 0.3402,
      "step": 23636
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.9579732418060303,
      "learning_rate": 1.7443873352271102e-06,
      "loss": 0.2945,
      "step": 23637
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.060065269470215,
      "learning_rate": 1.7437572900188304e-06,
      "loss": 0.338,
      "step": 23638
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.120237350463867,
      "learning_rate": 1.7431273477442246e-06,
      "loss": 0.3584,
      "step": 23639
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0871708393096924,
      "learning_rate": 1.7424975084111518e-06,
      "loss": 0.3308,
      "step": 23640
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.157111644744873,
      "learning_rate": 1.7418677720274612e-06,
      "loss": 0.3765,
      "step": 23641
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7955394983291626,
      "learning_rate": 1.7412381386010046e-06,
      "loss": 0.2829,
      "step": 23642
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0449581146240234,
      "learning_rate": 1.74060860813963e-06,
      "loss": 0.3285,
      "step": 23643
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.834053874015808,
      "learning_rate": 1.7399791806511912e-06,
      "loss": 0.3413,
      "step": 23644
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9212441444396973,
      "learning_rate": 1.7393498561435319e-06,
      "loss": 0.2863,
      "step": 23645
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0464096069335938,
      "learning_rate": 1.7387206346244944e-06,
      "loss": 0.3534,
      "step": 23646
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.114863872528076,
      "learning_rate": 1.7380915161019318e-06,
      "loss": 0.3546,
      "step": 23647
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8858660459518433,
      "learning_rate": 1.7374625005836832e-06,
      "loss": 0.2588,
      "step": 23648
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9662282466888428,
      "learning_rate": 1.7368335880775877e-06,
      "loss": 0.2642,
      "step": 23649
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.943854570388794,
      "learning_rate": 1.7362047785914915e-06,
      "loss": 0.3247,
      "step": 23650
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0282227993011475,
      "learning_rate": 1.7355760721332327e-06,
      "loss": 0.3345,
      "step": 23651
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9063138961791992,
      "learning_rate": 1.734947468710646e-06,
      "loss": 0.3432,
      "step": 23652
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.1279044151306152,
      "learning_rate": 1.7343189683315743e-06,
      "loss": 0.3423,
      "step": 23653
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.045806884765625,
      "learning_rate": 1.7336905710038498e-06,
      "loss": 0.3051,
      "step": 23654
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.847951054573059,
      "learning_rate": 1.7330622767353077e-06,
      "loss": 0.302,
      "step": 23655
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0260567665100098,
      "learning_rate": 1.7324340855337796e-06,
      "loss": 0.3268,
      "step": 23656
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.730203866958618,
      "learning_rate": 1.7318059974071e-06,
      "loss": 0.3174,
      "step": 23657
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.330449104309082,
      "learning_rate": 1.7311780123630994e-06,
      "loss": 0.2982,
      "step": 23658
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9441348314285278,
      "learning_rate": 1.7305501304096028e-06,
      "loss": 0.2632,
      "step": 23659
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9365800619125366,
      "learning_rate": 1.729922351554445e-06,
      "loss": 0.3646,
      "step": 23660
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.1205477714538574,
      "learning_rate": 1.7292946758054485e-06,
      "loss": 0.2618,
      "step": 23661
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9170045852661133,
      "learning_rate": 1.7286671031704373e-06,
      "loss": 0.3089,
      "step": 23662
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.2714054584503174,
      "learning_rate": 1.7280396336572403e-06,
      "loss": 0.2845,
      "step": 23663
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.3041679859161377,
      "learning_rate": 1.7274122672736781e-06,
      "loss": 0.329,
      "step": 23664
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.721495270729065,
      "learning_rate": 1.7267850040275702e-06,
      "loss": 0.2459,
      "step": 23665
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.3094711303710938,
      "learning_rate": 1.7261578439267402e-06,
      "loss": 0.3664,
      "step": 23666
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9199224710464478,
      "learning_rate": 1.725530786979006e-06,
      "loss": 0.3634,
      "step": 23667
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.1761488914489746,
      "learning_rate": 1.7249038331921863e-06,
      "loss": 0.4223,
      "step": 23668
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.1179721355438232,
      "learning_rate": 1.7242769825740934e-06,
      "loss": 0.3053,
      "step": 23669
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.978410243988037,
      "learning_rate": 1.7236502351325479e-06,
      "loss": 0.2862,
      "step": 23670
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9733452796936035,
      "learning_rate": 1.7230235908753612e-06,
      "loss": 0.2849,
      "step": 23671
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9727016687393188,
      "learning_rate": 1.7223970498103437e-06,
      "loss": 0.2961,
      "step": 23672
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9695854187011719,
      "learning_rate": 1.7217706119453114e-06,
      "loss": 0.3239,
      "step": 23673
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.5631937980651855,
      "learning_rate": 1.7211442772880726e-06,
      "loss": 0.3618,
      "step": 23674
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.250183582305908,
      "learning_rate": 1.7205180458464322e-06,
      "loss": 0.3188,
      "step": 23675
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.7582435607910156,
      "learning_rate": 1.7198919176282035e-06,
      "loss": 0.3672,
      "step": 23676
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.2610373497009277,
      "learning_rate": 1.719265892641191e-06,
      "loss": 0.2885,
      "step": 23677
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.911367654800415,
      "learning_rate": 1.7186399708931955e-06,
      "loss": 0.217,
      "step": 23678
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0864717960357666,
      "learning_rate": 1.7180141523920257e-06,
      "loss": 0.3795,
      "step": 23679
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.810066819190979,
      "learning_rate": 1.717388437145483e-06,
      "loss": 0.373,
      "step": 23680
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8272935152053833,
      "learning_rate": 1.7167628251613665e-06,
      "loss": 0.3502,
      "step": 23681
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0337069034576416,
      "learning_rate": 1.7161373164474748e-06,
      "loss": 0.2686,
      "step": 23682
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5173858404159546,
      "learning_rate": 1.7155119110116114e-06,
      "loss": 0.2333,
      "step": 23683
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.035661220550537,
      "learning_rate": 1.7148866088615702e-06,
      "loss": 0.2778,
      "step": 23684
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9344472885131836,
      "learning_rate": 1.7142614100051447e-06,
      "loss": 0.3021,
      "step": 23685
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9903607368469238,
      "learning_rate": 1.7136363144501345e-06,
      "loss": 0.3599,
      "step": 23686
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.202511787414551,
      "learning_rate": 1.713011322204331e-06,
      "loss": 0.3656,
      "step": 23687
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7193174362182617,
      "learning_rate": 1.7123864332755224e-06,
      "loss": 0.3144,
      "step": 23688
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.483172655105591,
      "learning_rate": 1.711761647671506e-06,
      "loss": 0.328,
      "step": 23689
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.727630615234375,
      "learning_rate": 1.7111369654000676e-06,
      "loss": 0.271,
      "step": 23690
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.924214482307434,
      "learning_rate": 1.7105123864689943e-06,
      "loss": 0.315,
      "step": 23691
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.197488784790039,
      "learning_rate": 1.7098879108860756e-06,
      "loss": 0.4004,
      "step": 23692
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.3466105461120605,
      "learning_rate": 1.7092635386590962e-06,
      "loss": 0.2235,
      "step": 23693
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8062567710876465,
      "learning_rate": 1.708639269795841e-06,
      "loss": 0.269,
      "step": 23694
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6591460704803467,
      "learning_rate": 1.7080151043040882e-06,
      "loss": 0.2825,
      "step": 23695
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8836050033569336,
      "learning_rate": 1.7073910421916262e-06,
      "loss": 0.3478,
      "step": 23696
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6578898429870605,
      "learning_rate": 1.706767083466233e-06,
      "loss": 0.2952,
      "step": 23697
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7802493572235107,
      "learning_rate": 1.706143228135685e-06,
      "loss": 0.3013,
      "step": 23698
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.112671375274658,
      "learning_rate": 1.7055194762077653e-06,
      "loss": 0.3244,
      "step": 23699
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0836896896362305,
      "learning_rate": 1.704895827690246e-06,
      "loss": 0.3577,
      "step": 23700
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.3005659580230713,
      "learning_rate": 1.704272282590903e-06,
      "loss": 0.3576,
      "step": 23701
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7875014543533325,
      "learning_rate": 1.7036488409175122e-06,
      "loss": 0.2798,
      "step": 23702
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0312304496765137,
      "learning_rate": 1.7030255026778464e-06,
      "loss": 0.3394,
      "step": 23703
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.918989896774292,
      "learning_rate": 1.7024022678796726e-06,
      "loss": 0.3566,
      "step": 23704
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.659464955329895,
      "learning_rate": 1.7017791365307668e-06,
      "loss": 0.272,
      "step": 23705
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.591353416442871,
      "learning_rate": 1.701156108638895e-06,
      "loss": 0.2893,
      "step": 23706
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.165536403656006,
      "learning_rate": 1.7005331842118245e-06,
      "loss": 0.3127,
      "step": 23707
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.00296950340271,
      "learning_rate": 1.69991036325732e-06,
      "loss": 0.2185,
      "step": 23708
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0026543140411377,
      "learning_rate": 1.699287645783151e-06,
      "loss": 0.2713,
      "step": 23709
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.815908432006836,
      "learning_rate": 1.6986650317970777e-06,
      "loss": 0.3176,
      "step": 23710
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.843616008758545,
      "learning_rate": 1.698042521306862e-06,
      "loss": 0.3433,
      "step": 23711
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7505558729171753,
      "learning_rate": 1.6974201143202673e-06,
      "loss": 0.2983,
      "step": 23712
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.16919207572937,
      "learning_rate": 1.6967978108450533e-06,
      "loss": 0.3447,
      "step": 23713
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.1544389724731445,
      "learning_rate": 1.6961756108889748e-06,
      "loss": 0.3291,
      "step": 23714
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.146507978439331,
      "learning_rate": 1.6955535144597935e-06,
      "loss": 0.346,
      "step": 23715
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.3104982376098633,
      "learning_rate": 1.694931521565264e-06,
      "loss": 0.325,
      "step": 23716
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.5670993328094482,
      "learning_rate": 1.694309632213138e-06,
      "loss": 0.3695,
      "step": 23717
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.3400673866271973,
      "learning_rate": 1.6936878464111739e-06,
      "loss": 0.3145,
      "step": 23718
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4769446849822998,
      "learning_rate": 1.6930661641671209e-06,
      "loss": 0.2504,
      "step": 23719
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7102210521697998,
      "learning_rate": 1.6924445854887295e-06,
      "loss": 0.2952,
      "step": 23720
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.04695200920105,
      "learning_rate": 1.6918231103837479e-06,
      "loss": 0.2843,
      "step": 23721
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.426896810531616,
      "learning_rate": 1.6912017388599277e-06,
      "loss": 0.3194,
      "step": 23722
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9040087461471558,
      "learning_rate": 1.690580470925014e-06,
      "loss": 0.3552,
      "step": 23723
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.190831422805786,
      "learning_rate": 1.6899593065867504e-06,
      "loss": 0.3276,
      "step": 23724
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9932007789611816,
      "learning_rate": 1.689338245852885e-06,
      "loss": 0.3502,
      "step": 23725
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.980868935585022,
      "learning_rate": 1.6887172887311597e-06,
      "loss": 0.2748,
      "step": 23726
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.449110507965088,
      "learning_rate": 1.6880964352293129e-06,
      "loss": 0.3322,
      "step": 23727
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.917248010635376,
      "learning_rate": 1.68747568535509e-06,
      "loss": 0.3004,
      "step": 23728
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.042989492416382,
      "learning_rate": 1.6868550391162275e-06,
      "loss": 0.2678,
      "step": 23729
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.9663927555084229,
      "learning_rate": 1.6862344965204613e-06,
      "loss": 0.2618,
      "step": 23730
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.229559898376465,
      "learning_rate": 1.685614057575532e-06,
      "loss": 0.3469,
      "step": 23731
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.2635982036590576,
      "learning_rate": 1.6849937222891733e-06,
      "loss": 0.2631,
      "step": 23732
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.611104965209961,
      "learning_rate": 1.6843734906691189e-06,
      "loss": 0.3577,
      "step": 23733
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.2993528842926025,
      "learning_rate": 1.6837533627230996e-06,
      "loss": 0.3752,
      "step": 23734
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.6151928901672363,
      "learning_rate": 1.6831333384588499e-06,
      "loss": 0.332,
      "step": 23735
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.031679153442383,
      "learning_rate": 1.6825134178840985e-06,
      "loss": 0.3189,
      "step": 23736
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8797569274902344,
      "learning_rate": 1.6818936010065723e-06,
      "loss": 0.3352,
      "step": 23737
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.035659074783325,
      "learning_rate": 1.681273887834003e-06,
      "loss": 0.2862,
      "step": 23738
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.3524718284606934,
      "learning_rate": 1.6806542783741141e-06,
      "loss": 0.3784,
      "step": 23739
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8420650959014893,
      "learning_rate": 1.6800347726346277e-06,
      "loss": 0.3228,
      "step": 23740
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.194298267364502,
      "learning_rate": 1.6794153706232742e-06,
      "loss": 0.3182,
      "step": 23741
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7578637599945068,
      "learning_rate": 1.6787960723477725e-06,
      "loss": 0.2839,
      "step": 23742
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9852901697158813,
      "learning_rate": 1.6781768778158403e-06,
      "loss": 0.3269,
      "step": 23743
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.155165195465088,
      "learning_rate": 1.6775577870352032e-06,
      "loss": 0.2813,
      "step": 23744
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.6044209003448486,
      "learning_rate": 1.6769388000135766e-06,
      "loss": 0.322,
      "step": 23745
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8945341110229492,
      "learning_rate": 1.6763199167586785e-06,
      "loss": 0.2678,
      "step": 23746
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0130362510681152,
      "learning_rate": 1.6757011372782206e-06,
      "loss": 0.3789,
      "step": 23747
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9578382968902588,
      "learning_rate": 1.6750824615799244e-06,
      "loss": 0.3144,
      "step": 23748
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7828881740570068,
      "learning_rate": 1.6744638896714993e-06,
      "loss": 0.2264,
      "step": 23749
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7872049808502197,
      "learning_rate": 1.6738454215606559e-06,
      "loss": 0.3336,
      "step": 23750
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0481746196746826,
      "learning_rate": 1.673227057255109e-06,
      "loss": 0.2624,
      "step": 23751
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6803699731826782,
      "learning_rate": 1.672608796762566e-06,
      "loss": 0.26,
      "step": 23752
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7356137037277222,
      "learning_rate": 1.6719906400907316e-06,
      "loss": 0.2332,
      "step": 23753
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.3759660720825195,
      "learning_rate": 1.6713725872473186e-06,
      "loss": 0.3472,
      "step": 23754
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5530864000320435,
      "learning_rate": 1.6707546382400285e-06,
      "loss": 0.2105,
      "step": 23755
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.2789266109466553,
      "learning_rate": 1.6701367930765655e-06,
      "loss": 0.309,
      "step": 23756
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8217570781707764,
      "learning_rate": 1.6695190517646353e-06,
      "loss": 0.3807,
      "step": 23757
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9299434423446655,
      "learning_rate": 1.6689014143119386e-06,
      "loss": 0.3172,
      "step": 23758
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9226104021072388,
      "learning_rate": 1.6682838807261748e-06,
      "loss": 0.2623,
      "step": 23759
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5968881845474243,
      "learning_rate": 1.6676664510150398e-06,
      "loss": 0.3031,
      "step": 23760
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6272492408752441,
      "learning_rate": 1.6670491251862374e-06,
      "loss": 0.2669,
      "step": 23761
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.024944305419922,
      "learning_rate": 1.6664319032474607e-06,
      "loss": 0.3356,
      "step": 23762
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.135629415512085,
      "learning_rate": 1.6658147852064043e-06,
      "loss": 0.2825,
      "step": 23763
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8832130432128906,
      "learning_rate": 1.6651977710707645e-06,
      "loss": 0.2899,
      "step": 23764
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.4397425651550293,
      "learning_rate": 1.6645808608482327e-06,
      "loss": 0.3154,
      "step": 23765
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7910324335098267,
      "learning_rate": 1.6639640545464975e-06,
      "loss": 0.2909,
      "step": 23766
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6828281879425049,
      "learning_rate": 1.6633473521732534e-06,
      "loss": 0.221,
      "step": 23767
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6489449739456177,
      "learning_rate": 1.6627307537361869e-06,
      "loss": 0.2166,
      "step": 23768
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.030747652053833,
      "learning_rate": 1.662114259242983e-06,
      "loss": 0.309,
      "step": 23769
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.13307523727417,
      "learning_rate": 1.6614978687013327e-06,
      "loss": 0.3162,
      "step": 23770
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8950344324111938,
      "learning_rate": 1.6608815821189183e-06,
      "loss": 0.333,
      "step": 23771
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6485943794250488,
      "learning_rate": 1.6602653995034234e-06,
      "loss": 0.3389,
      "step": 23772
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8553212881088257,
      "learning_rate": 1.6596493208625274e-06,
      "loss": 0.225,
      "step": 23773
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6556843519210815,
      "learning_rate": 1.6590333462039166e-06,
      "loss": 0.3351,
      "step": 23774
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9886661767959595,
      "learning_rate": 1.6584174755352666e-06,
      "loss": 0.2396,
      "step": 23775
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.3736305236816406,
      "learning_rate": 1.6578017088642551e-06,
      "loss": 0.2625,
      "step": 23776
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9394583702087402,
      "learning_rate": 1.6571860461985634e-06,
      "loss": 0.2874,
      "step": 23777
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.054502010345459,
      "learning_rate": 1.6565704875458644e-06,
      "loss": 0.3625,
      "step": 23778
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.1117420196533203,
      "learning_rate": 1.6559550329138296e-06,
      "loss": 0.2682,
      "step": 23779
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0911290645599365,
      "learning_rate": 1.6553396823101386e-06,
      "loss": 0.3371,
      "step": 23780
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.513808250427246,
      "learning_rate": 1.6547244357424598e-06,
      "loss": 0.2876,
      "step": 23781
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5903764963150024,
      "learning_rate": 1.6541092932184611e-06,
      "loss": 0.2898,
      "step": 23782
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8597657680511475,
      "learning_rate": 1.6534942547458166e-06,
      "loss": 0.279,
      "step": 23783
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7658240795135498,
      "learning_rate": 1.6528793203321924e-06,
      "loss": 0.3001,
      "step": 23784
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5396208763122559,
      "learning_rate": 1.6522644899852546e-06,
      "loss": 0.2649,
      "step": 23785
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6522727012634277,
      "learning_rate": 1.651649763712666e-06,
      "loss": 0.255,
      "step": 23786
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.175727367401123,
      "learning_rate": 1.6510351415220949e-06,
      "loss": 0.3411,
      "step": 23787
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6820447444915771,
      "learning_rate": 1.650420623421204e-06,
      "loss": 0.2631,
      "step": 23788
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.997501254081726,
      "learning_rate": 1.6498062094176493e-06,
      "loss": 0.2951,
      "step": 23789
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9330536127090454,
      "learning_rate": 1.6491918995190982e-06,
      "loss": 0.3537,
      "step": 23790
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9552485942840576,
      "learning_rate": 1.6485776937332053e-06,
      "loss": 0.3523,
      "step": 23791
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7431795597076416,
      "learning_rate": 1.6479635920676263e-06,
      "loss": 0.3081,
      "step": 23792
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7604089975357056,
      "learning_rate": 1.6473495945300232e-06,
      "loss": 0.296,
      "step": 23793
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.209585666656494,
      "learning_rate": 1.646735701128047e-06,
      "loss": 0.3939,
      "step": 23794
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.696474313735962,
      "learning_rate": 1.6461219118693505e-06,
      "loss": 0.332,
      "step": 23795
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.755825400352478,
      "learning_rate": 1.6455082267615897e-06,
      "loss": 0.2738,
      "step": 23796
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8937832117080688,
      "learning_rate": 1.6448946458124127e-06,
      "loss": 0.2793,
      "step": 23797
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.104238510131836,
      "learning_rate": 1.6442811690294713e-06,
      "loss": 0.3319,
      "step": 23798
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.2903387546539307,
      "learning_rate": 1.6436677964204095e-06,
      "loss": 0.3774,
      "step": 23799
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.231675863265991,
      "learning_rate": 1.6430545279928811e-06,
      "loss": 0.3357,
      "step": 23800
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.1589787006378174,
      "learning_rate": 1.6424413637545278e-06,
      "loss": 0.3324,
      "step": 23801
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0960280895233154,
      "learning_rate": 1.6418283037129933e-06,
      "loss": 0.2958,
      "step": 23802
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.047348976135254,
      "learning_rate": 1.6412153478759241e-06,
      "loss": 0.3095,
      "step": 23803
    },
    {
      "epoch": 2.46,
      "grad_norm": 5.3017258644104,
      "learning_rate": 1.6406024962509604e-06,
      "loss": 0.3464,
      "step": 23804
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9918322563171387,
      "learning_rate": 1.6399897488457417e-06,
      "loss": 0.3171,
      "step": 23805
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.915855884552002,
      "learning_rate": 1.6393771056679098e-06,
      "loss": 0.3238,
      "step": 23806
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.058932304382324,
      "learning_rate": 1.6387645667251018e-06,
      "loss": 0.3257,
      "step": 23807
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9238908290863037,
      "learning_rate": 1.6381521320249515e-06,
      "loss": 0.3295,
      "step": 23808
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9843205213546753,
      "learning_rate": 1.637539801575101e-06,
      "loss": 0.2869,
      "step": 23809
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6497418880462646,
      "learning_rate": 1.6369275753831792e-06,
      "loss": 0.3234,
      "step": 23810
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.558788776397705,
      "learning_rate": 1.6363154534568205e-06,
      "loss": 0.3097,
      "step": 23811
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.1374948024749756,
      "learning_rate": 1.6357034358036538e-06,
      "loss": 0.3415,
      "step": 23812
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.095586061477661,
      "learning_rate": 1.6350915224313147e-06,
      "loss": 0.403,
      "step": 23813
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6965607404708862,
      "learning_rate": 1.6344797133474287e-06,
      "loss": 0.3407,
      "step": 23814
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.5955305099487305,
      "learning_rate": 1.6338680085596215e-06,
      "loss": 0.3454,
      "step": 23815
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9915744066238403,
      "learning_rate": 1.6332564080755254e-06,
      "loss": 0.344,
      "step": 23816
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.15155291557312,
      "learning_rate": 1.632644911902761e-06,
      "loss": 0.3256,
      "step": 23817
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9873305559158325,
      "learning_rate": 1.63203352004895e-06,
      "loss": 0.3474,
      "step": 23818
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7683604955673218,
      "learning_rate": 1.6314222325217222e-06,
      "loss": 0.3369,
      "step": 23819
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8157800436019897,
      "learning_rate": 1.6308110493286933e-06,
      "loss": 0.2853,
      "step": 23820
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7448006868362427,
      "learning_rate": 1.6301999704774818e-06,
      "loss": 0.2979,
      "step": 23821
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.0706021785736084,
      "learning_rate": 1.6295889959757117e-06,
      "loss": 0.3148,
      "step": 23822
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9857549667358398,
      "learning_rate": 1.6289781258309966e-06,
      "loss": 0.243,
      "step": 23823
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.274390459060669,
      "learning_rate": 1.6283673600509542e-06,
      "loss": 0.3338,
      "step": 23824
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.1382460594177246,
      "learning_rate": 1.6277566986431948e-06,
      "loss": 0.2895,
      "step": 23825
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.119818925857544,
      "learning_rate": 1.6271461416153377e-06,
      "loss": 0.2644,
      "step": 23826
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.784257173538208,
      "learning_rate": 1.6265356889749918e-06,
      "loss": 0.2712,
      "step": 23827
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.3600332736968994,
      "learning_rate": 1.625925340729767e-06,
      "loss": 0.3185,
      "step": 23828
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.1092472076416016,
      "learning_rate": 1.6253150968872755e-06,
      "loss": 0.2895,
      "step": 23829
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.4510059356689453,
      "learning_rate": 1.6247049574551243e-06,
      "loss": 0.3842,
      "step": 23830
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.748983383178711,
      "learning_rate": 1.6240949224409186e-06,
      "loss": 0.2647,
      "step": 23831
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9680825471878052,
      "learning_rate": 1.6234849918522666e-06,
      "loss": 0.3296,
      "step": 23832
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.099252939224243,
      "learning_rate": 1.6228751656967712e-06,
      "loss": 0.3571,
      "step": 23833
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.76056969165802,
      "learning_rate": 1.6222654439820329e-06,
      "loss": 0.3041,
      "step": 23834
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.113119602203369,
      "learning_rate": 1.6216558267156578e-06,
      "loss": 0.2604,
      "step": 23835
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9350981712341309,
      "learning_rate": 1.6210463139052457e-06,
      "loss": 0.3639,
      "step": 23836
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5989265441894531,
      "learning_rate": 1.6204369055583935e-06,
      "loss": 0.2802,
      "step": 23837
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8754314184188843,
      "learning_rate": 1.6198276016826963e-06,
      "loss": 0.3535,
      "step": 23838
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9982162714004517,
      "learning_rate": 1.619218402285757e-06,
      "loss": 0.216,
      "step": 23839
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9587689638137817,
      "learning_rate": 1.6186093073751674e-06,
      "loss": 0.3221,
      "step": 23840
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0196478366851807,
      "learning_rate": 1.618000316958519e-06,
      "loss": 0.4316,
      "step": 23841
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2280309200286865,
      "learning_rate": 1.617391431043408e-06,
      "loss": 0.273,
      "step": 23842
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2525289058685303,
      "learning_rate": 1.6167826496374262e-06,
      "loss": 0.3629,
      "step": 23843
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.087675094604492,
      "learning_rate": 1.6161739727481585e-06,
      "loss": 0.3766,
      "step": 23844
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7728980779647827,
      "learning_rate": 1.6155654003831989e-06,
      "loss": 0.2578,
      "step": 23845
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.047860622406006,
      "learning_rate": 1.6149569325501325e-06,
      "loss": 0.2704,
      "step": 23846
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4210114479064941,
      "learning_rate": 1.614348569256542e-06,
      "loss": 0.2982,
      "step": 23847
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.1269779205322266,
      "learning_rate": 1.613740310510019e-06,
      "loss": 0.3125,
      "step": 23848
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8215694427490234,
      "learning_rate": 1.6131321563181424e-06,
      "loss": 0.3063,
      "step": 23849
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7668122053146362,
      "learning_rate": 1.6125241066884955e-06,
      "loss": 0.3173,
      "step": 23850
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.130963087081909,
      "learning_rate": 1.611916161628656e-06,
      "loss": 0.2818,
      "step": 23851
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0504493713378906,
      "learning_rate": 1.6113083211462088e-06,
      "loss": 0.312,
      "step": 23852
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6396816968917847,
      "learning_rate": 1.6107005852487289e-06,
      "loss": 0.2945,
      "step": 23853
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.190767765045166,
      "learning_rate": 1.6100929539437914e-06,
      "loss": 0.2917,
      "step": 23854
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.142432451248169,
      "learning_rate": 1.609485427238977e-06,
      "loss": 0.2807,
      "step": 23855
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.3152849674224854,
      "learning_rate": 1.6088780051418561e-06,
      "loss": 0.2071,
      "step": 23856
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.263397693634033,
      "learning_rate": 1.6082706876600007e-06,
      "loss": 0.3161,
      "step": 23857
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7514766454696655,
      "learning_rate": 1.6076634748009867e-06,
      "loss": 0.3043,
      "step": 23858
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.67857027053833,
      "learning_rate": 1.6070563665723827e-06,
      "loss": 0.3015,
      "step": 23859
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0436291694641113,
      "learning_rate": 1.6064493629817545e-06,
      "loss": 0.3165,
      "step": 23860
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9391334056854248,
      "learning_rate": 1.6058424640366742e-06,
      "loss": 0.3138,
      "step": 23861
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2656986713409424,
      "learning_rate": 1.605235669744707e-06,
      "loss": 0.3313,
      "step": 23862
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.118730068206787,
      "learning_rate": 1.6046289801134174e-06,
      "loss": 0.2989,
      "step": 23863
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.1024110317230225,
      "learning_rate": 1.6040223951503675e-06,
      "loss": 0.3065,
      "step": 23864
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7351378202438354,
      "learning_rate": 1.6034159148631235e-06,
      "loss": 0.2671,
      "step": 23865
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.249699831008911,
      "learning_rate": 1.6028095392592458e-06,
      "loss": 0.3358,
      "step": 23866
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7412419319152832,
      "learning_rate": 1.6022032683462896e-06,
      "loss": 0.3008,
      "step": 23867
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7518483400344849,
      "learning_rate": 1.60159710213182e-06,
      "loss": 0.2981,
      "step": 23868
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.717682957649231,
      "learning_rate": 1.600991040623392e-06,
      "loss": 0.3579,
      "step": 23869
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2500176429748535,
      "learning_rate": 1.6003850838285585e-06,
      "loss": 0.3131,
      "step": 23870
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8408104181289673,
      "learning_rate": 1.5997792317548799e-06,
      "loss": 0.2956,
      "step": 23871
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.922482967376709,
      "learning_rate": 1.599173484409906e-06,
      "loss": 0.3329,
      "step": 23872
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8393033742904663,
      "learning_rate": 1.5985678418011864e-06,
      "loss": 0.3461,
      "step": 23873
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.3581202030181885,
      "learning_rate": 1.597962303936278e-06,
      "loss": 0.3341,
      "step": 23874
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.005157232284546,
      "learning_rate": 1.597356870822727e-06,
      "loss": 0.2901,
      "step": 23875
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8049356937408447,
      "learning_rate": 1.5967515424680813e-06,
      "loss": 0.3243,
      "step": 23876
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.3985652923583984,
      "learning_rate": 1.5961463188798865e-06,
      "loss": 0.3208,
      "step": 23877
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.1089205741882324,
      "learning_rate": 1.5955412000656922e-06,
      "loss": 0.3795,
      "step": 23878
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.066535472869873,
      "learning_rate": 1.5949361860330393e-06,
      "loss": 0.3194,
      "step": 23879
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2161576747894287,
      "learning_rate": 1.5943312767894704e-06,
      "loss": 0.3348,
      "step": 23880
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.116042137145996,
      "learning_rate": 1.5937264723425306e-06,
      "loss": 0.3088,
      "step": 23881
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0918586254119873,
      "learning_rate": 1.5931217726997572e-06,
      "loss": 0.2902,
      "step": 23882
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0074968338012695,
      "learning_rate": 1.5925171778686888e-06,
      "loss": 0.2963,
      "step": 23883
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6567058563232422,
      "learning_rate": 1.5919126878568659e-06,
      "loss": 0.2858,
      "step": 23884
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.098501205444336,
      "learning_rate": 1.5913083026718235e-06,
      "loss": 0.3719,
      "step": 23885
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7832046747207642,
      "learning_rate": 1.5907040223210946e-06,
      "loss": 0.2984,
      "step": 23886
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7130142450332642,
      "learning_rate": 1.5900998468122175e-06,
      "loss": 0.2691,
      "step": 23887
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.251246690750122,
      "learning_rate": 1.5894957761527219e-06,
      "loss": 0.344,
      "step": 23888
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9145561456680298,
      "learning_rate": 1.5888918103501383e-06,
      "loss": 0.251,
      "step": 23889
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8513615131378174,
      "learning_rate": 1.5882879494119963e-06,
      "loss": 0.2644,
      "step": 23890
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7361266613006592,
      "learning_rate": 1.5876841933458277e-06,
      "loss": 0.2652,
      "step": 23891
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.062197685241699,
      "learning_rate": 1.5870805421591585e-06,
      "loss": 0.2462,
      "step": 23892
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8900340795516968,
      "learning_rate": 1.5864769958595104e-06,
      "loss": 0.3274,
      "step": 23893
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6761324405670166,
      "learning_rate": 1.5858735544544134e-06,
      "loss": 0.2609,
      "step": 23894
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.437159776687622,
      "learning_rate": 1.5852702179513901e-06,
      "loss": 0.2761,
      "step": 23895
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.680070400238037,
      "learning_rate": 1.5846669863579588e-06,
      "loss": 0.3166,
      "step": 23896
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7088472843170166,
      "learning_rate": 1.584063859681645e-06,
      "loss": 0.2899,
      "step": 23897
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0994009971618652,
      "learning_rate": 1.5834608379299654e-06,
      "loss": 0.3591,
      "step": 23898
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0366415977478027,
      "learning_rate": 1.5828579211104366e-06,
      "loss": 0.3334,
      "step": 23899
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6386176347732544,
      "learning_rate": 1.5822551092305793e-06,
      "loss": 0.273,
      "step": 23900
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7086081504821777,
      "learning_rate": 1.5816524022979074e-06,
      "loss": 0.2966,
      "step": 23901
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.3582427501678467,
      "learning_rate": 1.5810498003199347e-06,
      "loss": 0.2564,
      "step": 23902
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8137820959091187,
      "learning_rate": 1.5804473033041722e-06,
      "loss": 0.298,
      "step": 23903
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.936967134475708,
      "learning_rate": 1.5798449112581348e-06,
      "loss": 0.3051,
      "step": 23904
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9869052171707153,
      "learning_rate": 1.579242624189332e-06,
      "loss": 0.3774,
      "step": 23905
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9064186811447144,
      "learning_rate": 1.5786404421052692e-06,
      "loss": 0.3383,
      "step": 23906
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8078240156173706,
      "learning_rate": 1.578038365013459e-06,
      "loss": 0.2934,
      "step": 23907
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.083191156387329,
      "learning_rate": 1.5774363929214065e-06,
      "loss": 0.3696,
      "step": 23908
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5545265674591064,
      "learning_rate": 1.5768345258366158e-06,
      "loss": 0.2446,
      "step": 23909
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0527114868164062,
      "learning_rate": 1.5762327637665898e-06,
      "loss": 0.2673,
      "step": 23910
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.872015118598938,
      "learning_rate": 1.5756311067188313e-06,
      "loss": 0.307,
      "step": 23911
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9395097494125366,
      "learning_rate": 1.57502955470084e-06,
      "loss": 0.3394,
      "step": 23912
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.061793327331543,
      "learning_rate": 1.5744281077201196e-06,
      "loss": 0.2693,
      "step": 23913
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.022536039352417,
      "learning_rate": 1.5738267657841667e-06,
      "loss": 0.2934,
      "step": 23914
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7130072116851807,
      "learning_rate": 1.5732255289004762e-06,
      "loss": 0.3329,
      "step": 23915
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.1051180362701416,
      "learning_rate": 1.5726243970765475e-06,
      "loss": 0.2981,
      "step": 23916
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0704753398895264,
      "learning_rate": 1.572023370319874e-06,
      "loss": 0.3326,
      "step": 23917
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.339104652404785,
      "learning_rate": 1.5714224486379458e-06,
      "loss": 0.2993,
      "step": 23918
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.1855969429016113,
      "learning_rate": 1.5708216320382596e-06,
      "loss": 0.3232,
      "step": 23919
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.0385689735412598,
      "learning_rate": 1.5702209205283047e-06,
      "loss": 0.2823,
      "step": 23920
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.013486862182617,
      "learning_rate": 1.5696203141155664e-06,
      "loss": 0.3373,
      "step": 23921
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7278143167495728,
      "learning_rate": 1.5690198128075396e-06,
      "loss": 0.2803,
      "step": 23922
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9146718978881836,
      "learning_rate": 1.5684194166117062e-06,
      "loss": 0.2867,
      "step": 23923
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9250408411026,
      "learning_rate": 1.5678191255355523e-06,
      "loss": 0.3606,
      "step": 23924
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.2668893337249756,
      "learning_rate": 1.56721893958656e-06,
      "loss": 0.3341,
      "step": 23925
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9972517490386963,
      "learning_rate": 1.5666188587722175e-06,
      "loss": 0.3421,
      "step": 23926
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5080373287200928,
      "learning_rate": 1.5660188831000023e-06,
      "loss": 0.2265,
      "step": 23927
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7901802062988281,
      "learning_rate": 1.5654190125773927e-06,
      "loss": 0.2021,
      "step": 23928
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.4549858570098877,
      "learning_rate": 1.5648192472118729e-06,
      "loss": 0.3163,
      "step": 23929
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.554645538330078,
      "learning_rate": 1.5642195870109178e-06,
      "loss": 0.3122,
      "step": 23930
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.901335597038269,
      "learning_rate": 1.5636200319820005e-06,
      "loss": 0.3649,
      "step": 23931
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7977092266082764,
      "learning_rate": 1.5630205821326018e-06,
      "loss": 0.3235,
      "step": 23932
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5652450323104858,
      "learning_rate": 1.562421237470192e-06,
      "loss": 0.2616,
      "step": 23933
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9406731128692627,
      "learning_rate": 1.5618219980022408e-06,
      "loss": 0.258,
      "step": 23934
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8845152854919434,
      "learning_rate": 1.5612228637362237e-06,
      "loss": 0.3265,
      "step": 23935
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1428887844085693,
      "learning_rate": 1.5606238346796099e-06,
      "loss": 0.3368,
      "step": 23936
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2515969276428223,
      "learning_rate": 1.5600249108398658e-06,
      "loss": 0.2446,
      "step": 23937
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7763054370880127,
      "learning_rate": 1.5594260922244564e-06,
      "loss": 0.3541,
      "step": 23938
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.589399814605713,
      "learning_rate": 1.5588273788408526e-06,
      "loss": 0.3129,
      "step": 23939
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.389557361602783,
      "learning_rate": 1.558228770696515e-06,
      "loss": 0.3678,
      "step": 23940
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8887081146240234,
      "learning_rate": 1.5576302677989064e-06,
      "loss": 0.2774,
      "step": 23941
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.267045497894287,
      "learning_rate": 1.557031870155493e-06,
      "loss": 0.3055,
      "step": 23942
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2840826511383057,
      "learning_rate": 1.5564335777737315e-06,
      "loss": 0.4079,
      "step": 23943
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7606394290924072,
      "learning_rate": 1.555835390661079e-06,
      "loss": 0.3325,
      "step": 23944
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.120943546295166,
      "learning_rate": 1.5552373088249995e-06,
      "loss": 0.3155,
      "step": 23945
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0676217079162598,
      "learning_rate": 1.5546393322729458e-06,
      "loss": 0.3244,
      "step": 23946
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.050534248352051,
      "learning_rate": 1.5540414610123699e-06,
      "loss": 0.2863,
      "step": 23947
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8102308511734009,
      "learning_rate": 1.5534436950507326e-06,
      "loss": 0.2252,
      "step": 23948
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9081515073776245,
      "learning_rate": 1.5528460343954832e-06,
      "loss": 0.3704,
      "step": 23949
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1296486854553223,
      "learning_rate": 1.5522484790540726e-06,
      "loss": 0.266,
      "step": 23950
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.245222806930542,
      "learning_rate": 1.551651029033948e-06,
      "loss": 0.29,
      "step": 23951
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8979030847549438,
      "learning_rate": 1.5510536843425638e-06,
      "loss": 0.3365,
      "step": 23952
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.851045846939087,
      "learning_rate": 1.550456444987365e-06,
      "loss": 0.3239,
      "step": 23953
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.005579948425293,
      "learning_rate": 1.5498593109757942e-06,
      "loss": 0.2686,
      "step": 23954
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2513587474823,
      "learning_rate": 1.5492622823153014e-06,
      "loss": 0.3003,
      "step": 23955
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5877057313919067,
      "learning_rate": 1.5486653590133282e-06,
      "loss": 0.2167,
      "step": 23956
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1619887351989746,
      "learning_rate": 1.5480685410773123e-06,
      "loss": 0.3065,
      "step": 23957
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.156888484954834,
      "learning_rate": 1.5474718285147018e-06,
      "loss": 0.3403,
      "step": 23958
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.334362030029297,
      "learning_rate": 1.546875221332932e-06,
      "loss": 0.358,
      "step": 23959
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.124894618988037,
      "learning_rate": 1.5462787195394402e-06,
      "loss": 0.2691,
      "step": 23960
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.061410665512085,
      "learning_rate": 1.5456823231416663e-06,
      "loss": 0.269,
      "step": 23961
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9434664249420166,
      "learning_rate": 1.545086032147043e-06,
      "loss": 0.3001,
      "step": 23962
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8817312717437744,
      "learning_rate": 1.5444898465630065e-06,
      "loss": 0.3035,
      "step": 23963
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8770619630813599,
      "learning_rate": 1.5438937663969865e-06,
      "loss": 0.285,
      "step": 23964
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.778530478477478,
      "learning_rate": 1.5432977916564197e-06,
      "loss": 0.3146,
      "step": 23965
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.5875749588012695,
      "learning_rate": 1.5427019223487326e-06,
      "loss": 0.3126,
      "step": 23966
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.374931812286377,
      "learning_rate": 1.5421061584813524e-06,
      "loss": 0.2889,
      "step": 23967
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7676969766616821,
      "learning_rate": 1.5415105000617124e-06,
      "loss": 0.2402,
      "step": 23968
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9456403255462646,
      "learning_rate": 1.5409149470972352e-06,
      "loss": 0.3187,
      "step": 23969
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2534570693969727,
      "learning_rate": 1.540319499595344e-06,
      "loss": 0.3264,
      "step": 23970
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.79874849319458,
      "learning_rate": 1.539724157563468e-06,
      "loss": 0.2969,
      "step": 23971
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0197415351867676,
      "learning_rate": 1.5391289210090253e-06,
      "loss": 0.3694,
      "step": 23972
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9413189888000488,
      "learning_rate": 1.538533789939437e-06,
      "loss": 0.3435,
      "step": 23973
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.741850733757019,
      "learning_rate": 1.5379387643621247e-06,
      "loss": 0.2863,
      "step": 23974
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1751043796539307,
      "learning_rate": 1.5373438442845068e-06,
      "loss": 0.3469,
      "step": 23975
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.442812919616699,
      "learning_rate": 1.5367490297139998e-06,
      "loss": 0.3506,
      "step": 23976
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9439210891723633,
      "learning_rate": 1.5361543206580166e-06,
      "loss": 0.394,
      "step": 23977
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.059969186782837,
      "learning_rate": 1.5355597171239756e-06,
      "loss": 0.334,
      "step": 23978
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2028543949127197,
      "learning_rate": 1.5349652191192898e-06,
      "loss": 0.3223,
      "step": 23979
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7645375728607178,
      "learning_rate": 1.5343708266513668e-06,
      "loss": 0.309,
      "step": 23980
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6450163125991821,
      "learning_rate": 1.5337765397276229e-06,
      "loss": 0.3042,
      "step": 23981
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.796745777130127,
      "learning_rate": 1.533182358355465e-06,
      "loss": 0.3222,
      "step": 23982
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2105507850646973,
      "learning_rate": 1.5325882825422978e-06,
      "loss": 0.2721,
      "step": 23983
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.085914134979248,
      "learning_rate": 1.5319943122955328e-06,
      "loss": 0.3448,
      "step": 23984
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.5166373252868652,
      "learning_rate": 1.531400447622573e-06,
      "loss": 0.3816,
      "step": 23985
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8346306085586548,
      "learning_rate": 1.5308066885308204e-06,
      "loss": 0.3332,
      "step": 23986
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8528501987457275,
      "learning_rate": 1.5302130350276812e-06,
      "loss": 0.2917,
      "step": 23987
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9480594396591187,
      "learning_rate": 1.529619487120556e-06,
      "loss": 0.2809,
      "step": 23988
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.065474033355713,
      "learning_rate": 1.5290260448168438e-06,
      "loss": 0.2739,
      "step": 23989
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7692842483520508,
      "learning_rate": 1.5284327081239403e-06,
      "loss": 0.3277,
      "step": 23990
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.317781686782837,
      "learning_rate": 1.5278394770492488e-06,
      "loss": 0.4093,
      "step": 23991
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0038442611694336,
      "learning_rate": 1.5272463516001623e-06,
      "loss": 0.3904,
      "step": 23992
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.25191330909729,
      "learning_rate": 1.5266533317840738e-06,
      "loss": 0.3422,
      "step": 23993
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0059967041015625,
      "learning_rate": 1.5260604176083803e-06,
      "loss": 0.235,
      "step": 23994
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0773301124572754,
      "learning_rate": 1.5254676090804721e-06,
      "loss": 0.2745,
      "step": 23995
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2968339920043945,
      "learning_rate": 1.5248749062077385e-06,
      "loss": 0.3066,
      "step": 23996
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.528017044067383,
      "learning_rate": 1.5242823089975712e-06,
      "loss": 0.3122,
      "step": 23997
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.100841760635376,
      "learning_rate": 1.523689817457359e-06,
      "loss": 0.3061,
      "step": 23998
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7534210681915283,
      "learning_rate": 1.5230974315944869e-06,
      "loss": 0.2748,
      "step": 23999
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.3795156478881836,
      "learning_rate": 1.5225051514163391e-06,
      "loss": 0.3171,
      "step": 24000
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2306671142578125,
      "learning_rate": 1.5219129769303031e-06,
      "loss": 0.3834,
      "step": 24001
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9032319784164429,
      "learning_rate": 1.5213209081437597e-06,
      "loss": 0.2621,
      "step": 24002
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.014385461807251,
      "learning_rate": 1.5207289450640894e-06,
      "loss": 0.3271,
      "step": 24003
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.64815354347229,
      "learning_rate": 1.5201370876986764e-06,
      "loss": 0.3243,
      "step": 24004
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1071512699127197,
      "learning_rate": 1.5195453360548961e-06,
      "loss": 0.2699,
      "step": 24005
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.0762975215911865,
      "learning_rate": 1.5189536901401258e-06,
      "loss": 0.3554,
      "step": 24006
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.200561761856079,
      "learning_rate": 1.518362149961745e-06,
      "loss": 0.2306,
      "step": 24007
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.129537343978882,
      "learning_rate": 1.517770715527127e-06,
      "loss": 0.2791,
      "step": 24008
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7056688070297241,
      "learning_rate": 1.5171793868436423e-06,
      "loss": 0.2987,
      "step": 24009
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8274683952331543,
      "learning_rate": 1.5165881639186686e-06,
      "loss": 0.2824,
      "step": 24010
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.005162477493286,
      "learning_rate": 1.515997046759574e-06,
      "loss": 0.3071,
      "step": 24011
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.1422600746154785,
      "learning_rate": 1.5154060353737299e-06,
      "loss": 0.3291,
      "step": 24012
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9809529781341553,
      "learning_rate": 1.5148151297684999e-06,
      "loss": 0.3751,
      "step": 24013
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9611765146255493,
      "learning_rate": 1.514224329951257e-06,
      "loss": 0.2872,
      "step": 24014
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.605414867401123,
      "learning_rate": 1.5136336359293647e-06,
      "loss": 0.3216,
      "step": 24015
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.328627586364746,
      "learning_rate": 1.5130430477101843e-06,
      "loss": 0.3688,
      "step": 24016
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.828829288482666,
      "learning_rate": 1.5124525653010845e-06,
      "loss": 0.3174,
      "step": 24017
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9486644268035889,
      "learning_rate": 1.5118621887094243e-06,
      "loss": 0.2504,
      "step": 24018
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8183997869491577,
      "learning_rate": 1.5112719179425617e-06,
      "loss": 0.2882,
      "step": 24019
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6772640943527222,
      "learning_rate": 1.5106817530078598e-06,
      "loss": 0.2786,
      "step": 24020
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6894582509994507,
      "learning_rate": 1.5100916939126763e-06,
      "loss": 0.2662,
      "step": 24021
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7915680408477783,
      "learning_rate": 1.509501740664363e-06,
      "loss": 0.3492,
      "step": 24022
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9204838275909424,
      "learning_rate": 1.5089118932702807e-06,
      "loss": 0.3052,
      "step": 24023
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8462756872177124,
      "learning_rate": 1.50832215173778e-06,
      "loss": 0.3262,
      "step": 24024
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8518965244293213,
      "learning_rate": 1.507732516074215e-06,
      "loss": 0.3174,
      "step": 24025
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8214882612228394,
      "learning_rate": 1.5071429862869348e-06,
      "loss": 0.3568,
      "step": 24026
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7536797523498535,
      "learning_rate": 1.5065535623832917e-06,
      "loss": 0.3201,
      "step": 24027
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7170264720916748,
      "learning_rate": 1.5059642443706334e-06,
      "loss": 0.2536,
      "step": 24028
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9579832553863525,
      "learning_rate": 1.5053750322563055e-06,
      "loss": 0.2931,
      "step": 24029
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1796069145202637,
      "learning_rate": 1.5047859260476572e-06,
      "loss": 0.2686,
      "step": 24030
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.023696184158325,
      "learning_rate": 1.5041969257520307e-06,
      "loss": 0.3658,
      "step": 24031
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7852965593338013,
      "learning_rate": 1.5036080313767686e-06,
      "loss": 0.3198,
      "step": 24032
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6890521049499512,
      "learning_rate": 1.5030192429292157e-06,
      "loss": 0.2696,
      "step": 24033
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7672960758209229,
      "learning_rate": 1.502430560416711e-06,
      "loss": 0.3099,
      "step": 24034
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5105150938034058,
      "learning_rate": 1.5018419838465926e-06,
      "loss": 0.2463,
      "step": 24035
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0278728008270264,
      "learning_rate": 1.5012535132262018e-06,
      "loss": 0.3595,
      "step": 24036
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.2622759342193604,
      "learning_rate": 1.5006651485628732e-06,
      "loss": 0.364,
      "step": 24037
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1941752433776855,
      "learning_rate": 1.5000768898639418e-06,
      "loss": 0.2484,
      "step": 24038
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0280892848968506,
      "learning_rate": 1.4994887371367407e-06,
      "loss": 0.2061,
      "step": 24039
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9997713565826416,
      "learning_rate": 1.4989006903886061e-06,
      "loss": 0.3109,
      "step": 24040
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9720370769500732,
      "learning_rate": 1.4983127496268678e-06,
      "loss": 0.3062,
      "step": 24041
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9533886909484863,
      "learning_rate": 1.4977249148588525e-06,
      "loss": 0.2953,
      "step": 24042
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.848056435585022,
      "learning_rate": 1.4971371860918938e-06,
      "loss": 0.3401,
      "step": 24043
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7708910703659058,
      "learning_rate": 1.4965495633333182e-06,
      "loss": 0.3178,
      "step": 24044
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8070590496063232,
      "learning_rate": 1.495962046590448e-06,
      "loss": 0.3301,
      "step": 24045
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.870093822479248,
      "learning_rate": 1.4953746358706134e-06,
      "loss": 0.2519,
      "step": 24046
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7748937606811523,
      "learning_rate": 1.4947873311811346e-06,
      "loss": 0.2325,
      "step": 24047
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6062668561935425,
      "learning_rate": 1.4942001325293332e-06,
      "loss": 0.279,
      "step": 24048
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8979792594909668,
      "learning_rate": 1.4936130399225323e-06,
      "loss": 0.2929,
      "step": 24049
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7689316272735596,
      "learning_rate": 1.4930260533680518e-06,
      "loss": 0.3129,
      "step": 24050
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8461982011795044,
      "learning_rate": 1.4924391728732079e-06,
      "loss": 0.337,
      "step": 24051
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.100689649581909,
      "learning_rate": 1.491852398445316e-06,
      "loss": 0.3678,
      "step": 24052
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0942652225494385,
      "learning_rate": 1.4912657300916956e-06,
      "loss": 0.3366,
      "step": 24053
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.448559045791626,
      "learning_rate": 1.4906791678196586e-06,
      "loss": 0.2835,
      "step": 24054
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.5473179817199707,
      "learning_rate": 1.490092711636516e-06,
      "loss": 0.328,
      "step": 24055
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1661176681518555,
      "learning_rate": 1.4895063615495852e-06,
      "loss": 0.3446,
      "step": 24056
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8642948865890503,
      "learning_rate": 1.4889201175661715e-06,
      "loss": 0.2076,
      "step": 24057
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.519411325454712,
      "learning_rate": 1.4883339796935835e-06,
      "loss": 0.2851,
      "step": 24058
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8936537504196167,
      "learning_rate": 1.4877479479391333e-06,
      "loss": 0.3485,
      "step": 24059
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9833475351333618,
      "learning_rate": 1.4871620223101236e-06,
      "loss": 0.3866,
      "step": 24060
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8528002500534058,
      "learning_rate": 1.4865762028138576e-06,
      "loss": 0.284,
      "step": 24061
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1891961097717285,
      "learning_rate": 1.4859904894576449e-06,
      "loss": 0.2755,
      "step": 24062
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.146876096725464,
      "learning_rate": 1.4854048822487832e-06,
      "loss": 0.3515,
      "step": 24063
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.736401915550232,
      "learning_rate": 1.4848193811945754e-06,
      "loss": 0.1928,
      "step": 24064
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7631187438964844,
      "learning_rate": 1.484233986302318e-06,
      "loss": 0.3514,
      "step": 24065
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.009399175643921,
      "learning_rate": 1.483648697579314e-06,
      "loss": 0.2888,
      "step": 24066
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.147247791290283,
      "learning_rate": 1.4830635150328587e-06,
      "loss": 0.3199,
      "step": 24067
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0011909008026123,
      "learning_rate": 1.482478438670244e-06,
      "loss": 0.2815,
      "step": 24068
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8163188695907593,
      "learning_rate": 1.4818934684987707e-06,
      "loss": 0.3004,
      "step": 24069
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.3874118328094482,
      "learning_rate": 1.4813086045257286e-06,
      "loss": 0.311,
      "step": 24070
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7203178405761719,
      "learning_rate": 1.4807238467584062e-06,
      "loss": 0.2809,
      "step": 24071
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.80382239818573,
      "learning_rate": 1.4801391952041e-06,
      "loss": 0.3123,
      "step": 24072
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9286214113235474,
      "learning_rate": 1.4795546498700974e-06,
      "loss": 0.3834,
      "step": 24073
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8854719400405884,
      "learning_rate": 1.4789702107636816e-06,
      "loss": 0.2922,
      "step": 24074
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7400790452957153,
      "learning_rate": 1.4783858778921445e-06,
      "loss": 0.273,
      "step": 24075
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0413763523101807,
      "learning_rate": 1.4778016512627702e-06,
      "loss": 0.3228,
      "step": 24076
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.953099250793457,
      "learning_rate": 1.4772175308828407e-06,
      "loss": 0.4011,
      "step": 24077
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.011221408843994,
      "learning_rate": 1.476633516759637e-06,
      "loss": 0.3183,
      "step": 24078
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6185837984085083,
      "learning_rate": 1.476049608900444e-06,
      "loss": 0.2824,
      "step": 24079
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.2325496673583984,
      "learning_rate": 1.4754658073125406e-06,
      "loss": 0.3379,
      "step": 24080
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9720122814178467,
      "learning_rate": 1.474882112003202e-06,
      "loss": 0.3612,
      "step": 24081
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.829721450805664,
      "learning_rate": 1.4742985229797102e-06,
      "loss": 0.2799,
      "step": 24082
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.2619476318359375,
      "learning_rate": 1.4737150402493395e-06,
      "loss": 0.3143,
      "step": 24083
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.2097651958465576,
      "learning_rate": 1.4731316638193605e-06,
      "loss": 0.3184,
      "step": 24084
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9332987070083618,
      "learning_rate": 1.472548393697052e-06,
      "loss": 0.3167,
      "step": 24085
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.969180941581726,
      "learning_rate": 1.4719652298896826e-06,
      "loss": 0.2899,
      "step": 24086
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.332267999649048,
      "learning_rate": 1.4713821724045219e-06,
      "loss": 0.4371,
      "step": 24087
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.788564920425415,
      "learning_rate": 1.470799221248843e-06,
      "loss": 0.2429,
      "step": 24088
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6575441360473633,
      "learning_rate": 1.4702163764299116e-06,
      "loss": 0.2352,
      "step": 24089
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.358639717102051,
      "learning_rate": 1.4696336379549947e-06,
      "loss": 0.252,
      "step": 24090
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6656426191329956,
      "learning_rate": 1.4690510058313545e-06,
      "loss": 0.2993,
      "step": 24091
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7226719856262207,
      "learning_rate": 1.4684684800662585e-06,
      "loss": 0.2825,
      "step": 24092
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.6883418560028076,
      "learning_rate": 1.4678860606669698e-06,
      "loss": 0.362,
      "step": 24093
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.058086395263672,
      "learning_rate": 1.4673037476407447e-06,
      "loss": 0.3396,
      "step": 24094
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8637220859527588,
      "learning_rate": 1.4667215409948487e-06,
      "loss": 0.2882,
      "step": 24095
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9485701322555542,
      "learning_rate": 1.466139440736538e-06,
      "loss": 0.2824,
      "step": 24096
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0081191062927246,
      "learning_rate": 1.4655574468730682e-06,
      "loss": 0.3235,
      "step": 24097
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0084915161132812,
      "learning_rate": 1.4649755594116999e-06,
      "loss": 0.3796,
      "step": 24098
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.326509952545166,
      "learning_rate": 1.464393778359684e-06,
      "loss": 0.3422,
      "step": 24099
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0940232276916504,
      "learning_rate": 1.4638121037242714e-06,
      "loss": 0.3154,
      "step": 24100
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8813835382461548,
      "learning_rate": 1.4632305355127208e-06,
      "loss": 0.277,
      "step": 24101
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9058434963226318,
      "learning_rate": 1.4626490737322784e-06,
      "loss": 0.2717,
      "step": 24102
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7804358005523682,
      "learning_rate": 1.4620677183901954e-06,
      "loss": 0.2696,
      "step": 24103
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6748430728912354,
      "learning_rate": 1.4614864694937159e-06,
      "loss": 0.268,
      "step": 24104
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8275703191757202,
      "learning_rate": 1.4609053270500916e-06,
      "loss": 0.2987,
      "step": 24105
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8668780326843262,
      "learning_rate": 1.4603242910665661e-06,
      "loss": 0.3668,
      "step": 24106
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.094982147216797,
      "learning_rate": 1.4597433615503798e-06,
      "loss": 0.2293,
      "step": 24107
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1369073390960693,
      "learning_rate": 1.4591625385087816e-06,
      "loss": 0.3913,
      "step": 24108
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0276646614074707,
      "learning_rate": 1.4585818219490088e-06,
      "loss": 0.3136,
      "step": 24109
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.64827561378479,
      "learning_rate": 1.4580012118783004e-06,
      "loss": 0.2383,
      "step": 24110
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8243823051452637,
      "learning_rate": 1.4574207083039006e-06,
      "loss": 0.2434,
      "step": 24111
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9614912271499634,
      "learning_rate": 1.456840311233042e-06,
      "loss": 0.2907,
      "step": 24112
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9414139986038208,
      "learning_rate": 1.4562600206729594e-06,
      "loss": 0.3536,
      "step": 24113
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0376250743865967,
      "learning_rate": 1.4556798366308922e-06,
      "loss": 0.241,
      "step": 24114
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9951609373092651,
      "learning_rate": 1.4550997591140713e-06,
      "loss": 0.3128,
      "step": 24115
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.07148814201355,
      "learning_rate": 1.454519788129729e-06,
      "loss": 0.319,
      "step": 24116
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8080381155014038,
      "learning_rate": 1.453939923685095e-06,
      "loss": 0.362,
      "step": 24117
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.0270888805389404,
      "learning_rate": 1.4533601657873997e-06,
      "loss": 0.3577,
      "step": 24118
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.653204321861267,
      "learning_rate": 1.4527805144438723e-06,
      "loss": 0.288,
      "step": 24119
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.940340280532837,
      "learning_rate": 1.4522009696617357e-06,
      "loss": 0.3067,
      "step": 24120
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.22356915473938,
      "learning_rate": 1.4516215314482207e-06,
      "loss": 0.3453,
      "step": 24121
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9855732917785645,
      "learning_rate": 1.4510421998105483e-06,
      "loss": 0.3242,
      "step": 24122
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.185995101928711,
      "learning_rate": 1.4504629747559395e-06,
      "loss": 0.3206,
      "step": 24123
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7964332103729248,
      "learning_rate": 1.4498838562916194e-06,
      "loss": 0.3053,
      "step": 24124
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8981313705444336,
      "learning_rate": 1.449304844424807e-06,
      "loss": 0.3145,
      "step": 24125
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.4304299354553223,
      "learning_rate": 1.4487259391627184e-06,
      "loss": 0.396,
      "step": 24126
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6435832977294922,
      "learning_rate": 1.4481471405125746e-06,
      "loss": 0.2833,
      "step": 24127
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1076717376708984,
      "learning_rate": 1.447568448481591e-06,
      "loss": 0.3491,
      "step": 24128
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6101136207580566,
      "learning_rate": 1.4469898630769807e-06,
      "loss": 0.2359,
      "step": 24129
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9547721147537231,
      "learning_rate": 1.4464113843059568e-06,
      "loss": 0.3198,
      "step": 24130
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9219001531600952,
      "learning_rate": 1.4458330121757336e-06,
      "loss": 0.2861,
      "step": 24131
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.013725757598877,
      "learning_rate": 1.445254746693522e-06,
      "loss": 0.3856,
      "step": 24132
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.6401376724243164,
      "learning_rate": 1.444676587866528e-06,
      "loss": 0.3178,
      "step": 24133
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7160592079162598,
      "learning_rate": 1.4440985357019643e-06,
      "loss": 0.2706,
      "step": 24134
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.528064250946045,
      "learning_rate": 1.443520590207035e-06,
      "loss": 0.2552,
      "step": 24135
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7748898267745972,
      "learning_rate": 1.442942751388945e-06,
      "loss": 0.3266,
      "step": 24136
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6620417833328247,
      "learning_rate": 1.4423650192549e-06,
      "loss": 0.2524,
      "step": 24137
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.001558780670166,
      "learning_rate": 1.441787393812104e-06,
      "loss": 0.2246,
      "step": 24138
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8708783388137817,
      "learning_rate": 1.4412098750677527e-06,
      "loss": 0.3421,
      "step": 24139
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.3257014751434326,
      "learning_rate": 1.4406324630290547e-06,
      "loss": 0.3334,
      "step": 24140
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7623220682144165,
      "learning_rate": 1.440055157703203e-06,
      "loss": 0.227,
      "step": 24141
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9494915008544922,
      "learning_rate": 1.4394779590973973e-06,
      "loss": 0.3797,
      "step": 24142
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8022350072860718,
      "learning_rate": 1.4389008672188299e-06,
      "loss": 0.2672,
      "step": 24143
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0899407863616943,
      "learning_rate": 1.4383238820747014e-06,
      "loss": 0.3136,
      "step": 24144
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.949581265449524,
      "learning_rate": 1.437747003672202e-06,
      "loss": 0.2948,
      "step": 24145
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0961830615997314,
      "learning_rate": 1.4371702320185233e-06,
      "loss": 0.3421,
      "step": 24146
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8087043762207031,
      "learning_rate": 1.4365935671208587e-06,
      "loss": 0.2446,
      "step": 24147
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.805871844291687,
      "learning_rate": 1.4360170089863957e-06,
      "loss": 0.3519,
      "step": 24148
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0680618286132812,
      "learning_rate": 1.4354405576223218e-06,
      "loss": 0.3113,
      "step": 24149
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.77800452709198,
      "learning_rate": 1.434864213035827e-06,
      "loss": 0.3207,
      "step": 24150
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.717689037322998,
      "learning_rate": 1.4342879752340944e-06,
      "loss": 0.255,
      "step": 24151
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8062468767166138,
      "learning_rate": 1.433711844224307e-06,
      "loss": 0.3423,
      "step": 24152
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8125765323638916,
      "learning_rate": 1.4331358200136514e-06,
      "loss": 0.2807,
      "step": 24153
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8109779357910156,
      "learning_rate": 1.4325599026093073e-06,
      "loss": 0.3337,
      "step": 24154
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9309076070785522,
      "learning_rate": 1.4319840920184536e-06,
      "loss": 0.3525,
      "step": 24155
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.859153151512146,
      "learning_rate": 1.4314083882482689e-06,
      "loss": 0.3312,
      "step": 24156
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6708475351333618,
      "learning_rate": 1.4308327913059339e-06,
      "loss": 0.2893,
      "step": 24157
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0504422187805176,
      "learning_rate": 1.4302573011986221e-06,
      "loss": 0.318,
      "step": 24158
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.671553373336792,
      "learning_rate": 1.4296819179335076e-06,
      "loss": 0.2985,
      "step": 24159
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.038444757461548,
      "learning_rate": 1.4291066415177678e-06,
      "loss": 0.2905,
      "step": 24160
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1818103790283203,
      "learning_rate": 1.4285314719585718e-06,
      "loss": 0.3768,
      "step": 24161
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8892954587936401,
      "learning_rate": 1.4279564092630894e-06,
      "loss": 0.2602,
      "step": 24162
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.5210397243499756,
      "learning_rate": 1.4273814534384933e-06,
      "loss": 0.3965,
      "step": 24163
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.2478435039520264,
      "learning_rate": 1.4268066044919515e-06,
      "loss": 0.3462,
      "step": 24164
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9828035831451416,
      "learning_rate": 1.426231862430626e-06,
      "loss": 0.2282,
      "step": 24165
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.2302801609039307,
      "learning_rate": 1.4256572272616886e-06,
      "loss": 0.3009,
      "step": 24166
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1333582401275635,
      "learning_rate": 1.4250826989923017e-06,
      "loss": 0.2935,
      "step": 24167
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1994619369506836,
      "learning_rate": 1.424508277629626e-06,
      "loss": 0.2912,
      "step": 24168
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.189472198486328,
      "learning_rate": 1.4239339631808224e-06,
      "loss": 0.2794,
      "step": 24169
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7201730012893677,
      "learning_rate": 1.4233597556530565e-06,
      "loss": 0.2394,
      "step": 24170
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.992746114730835,
      "learning_rate": 1.4227856550534824e-06,
      "loss": 0.2901,
      "step": 24171
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1054468154907227,
      "learning_rate": 1.4222116613892567e-06,
      "loss": 0.3243,
      "step": 24172
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.700951337814331,
      "learning_rate": 1.4216377746675402e-06,
      "loss": 0.2416,
      "step": 24173
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.838205099105835,
      "learning_rate": 1.4210639948954852e-06,
      "loss": 0.2519,
      "step": 24174
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9786262512207031,
      "learning_rate": 1.4204903220802435e-06,
      "loss": 0.3799,
      "step": 24175
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0574116706848145,
      "learning_rate": 1.4199167562289705e-06,
      "loss": 0.2734,
      "step": 24176
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8089808225631714,
      "learning_rate": 1.4193432973488164e-06,
      "loss": 0.3243,
      "step": 24177
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8285706043243408,
      "learning_rate": 1.4187699454469285e-06,
      "loss": 0.2778,
      "step": 24178
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.023484468460083,
      "learning_rate": 1.4181967005304575e-06,
      "loss": 0.3041,
      "step": 24179
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.4026143550872803,
      "learning_rate": 1.4176235626065494e-06,
      "loss": 0.3848,
      "step": 24180
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.2736222743988037,
      "learning_rate": 1.4170505316823502e-06,
      "loss": 0.2908,
      "step": 24181
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0466115474700928,
      "learning_rate": 1.416477607765001e-06,
      "loss": 0.3656,
      "step": 24182
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9384151697158813,
      "learning_rate": 1.4159047908616496e-06,
      "loss": 0.3545,
      "step": 24183
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1706316471099854,
      "learning_rate": 1.4153320809794358e-06,
      "loss": 0.2192,
      "step": 24184
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1044318675994873,
      "learning_rate": 1.414759478125496e-06,
      "loss": 0.333,
      "step": 24185
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0630457401275635,
      "learning_rate": 1.4141869823069742e-06,
      "loss": 0.3214,
      "step": 24186
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1436822414398193,
      "learning_rate": 1.4136145935310054e-06,
      "loss": 0.258,
      "step": 24187
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.5779030323028564,
      "learning_rate": 1.4130423118047255e-06,
      "loss": 0.2835,
      "step": 24188
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.8094587326049805,
      "learning_rate": 1.4124701371352712e-06,
      "loss": 0.287,
      "step": 24189
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.7769157886505127,
      "learning_rate": 1.4118980695297758e-06,
      "loss": 0.2871,
      "step": 24190
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8438667058944702,
      "learning_rate": 1.4113261089953666e-06,
      "loss": 0.3258,
      "step": 24191
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.6723031997680664,
      "learning_rate": 1.410754255539183e-06,
      "loss": 0.3482,
      "step": 24192
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.752519965171814,
      "learning_rate": 1.4101825091683485e-06,
      "loss": 0.2189,
      "step": 24193
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.4903812408447266,
      "learning_rate": 1.4096108698899925e-06,
      "loss": 0.4129,
      "step": 24194
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7837064266204834,
      "learning_rate": 1.409039337711241e-06,
      "loss": 0.3394,
      "step": 24195
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.2045187950134277,
      "learning_rate": 1.408467912639222e-06,
      "loss": 0.3125,
      "step": 24196
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.114147663116455,
      "learning_rate": 1.4078965946810586e-06,
      "loss": 0.3194,
      "step": 24197
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9438579082489014,
      "learning_rate": 1.4073253838438705e-06,
      "loss": 0.3265,
      "step": 24198
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.306283473968506,
      "learning_rate": 1.4067542801347845e-06,
      "loss": 0.335,
      "step": 24199
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.053354501724243,
      "learning_rate": 1.406183283560919e-06,
      "loss": 0.2777,
      "step": 24200
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5971804857254028,
      "learning_rate": 1.4056123941293887e-06,
      "loss": 0.2732,
      "step": 24201
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7652943134307861,
      "learning_rate": 1.4050416118473175e-06,
      "loss": 0.2954,
      "step": 24202
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.9992015361785889,
      "learning_rate": 1.4044709367218178e-06,
      "loss": 0.3426,
      "step": 24203
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8292598724365234,
      "learning_rate": 1.403900368760004e-06,
      "loss": 0.2633,
      "step": 24204
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.433985948562622,
      "learning_rate": 1.403329907968992e-06,
      "loss": 0.3187,
      "step": 24205
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.3016860485076904,
      "learning_rate": 1.4027595543558925e-06,
      "loss": 0.2955,
      "step": 24206
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1297965049743652,
      "learning_rate": 1.402189307927817e-06,
      "loss": 0.3904,
      "step": 24207
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8611600399017334,
      "learning_rate": 1.4016191686918723e-06,
      "loss": 0.312,
      "step": 24208
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5914525985717773,
      "learning_rate": 1.4010491366551715e-06,
      "loss": 0.2665,
      "step": 24209
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.876916766166687,
      "learning_rate": 1.4004792118248179e-06,
      "loss": 0.3226,
      "step": 24210
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1333959102630615,
      "learning_rate": 1.3999093942079156e-06,
      "loss": 0.2748,
      "step": 24211
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0918431282043457,
      "learning_rate": 1.3993396838115725e-06,
      "loss": 0.3126,
      "step": 24212
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.1318366527557373,
      "learning_rate": 1.3987700806428905e-06,
      "loss": 0.324,
      "step": 24213
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.303140163421631,
      "learning_rate": 1.3982005847089674e-06,
      "loss": 0.3653,
      "step": 24214
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0442845821380615,
      "learning_rate": 1.3976311960169087e-06,
      "loss": 0.3513,
      "step": 24215
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.543872356414795,
      "learning_rate": 1.3970619145738107e-06,
      "loss": 0.2271,
      "step": 24216
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.054194927215576,
      "learning_rate": 1.396492740386769e-06,
      "loss": 0.3403,
      "step": 24217
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5356826782226562,
      "learning_rate": 1.3959236734628834e-06,
      "loss": 0.2665,
      "step": 24218
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.3245158195495605,
      "learning_rate": 1.3953547138092461e-06,
      "loss": 0.3144,
      "step": 24219
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8924263715744019,
      "learning_rate": 1.3947858614329525e-06,
      "loss": 0.2988,
      "step": 24220
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.014972448348999,
      "learning_rate": 1.39421711634109e-06,
      "loss": 0.296,
      "step": 24221
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7835073471069336,
      "learning_rate": 1.3936484785407555e-06,
      "loss": 0.3116,
      "step": 24222
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.933890700340271,
      "learning_rate": 1.3930799480390357e-06,
      "loss": 0.2674,
      "step": 24223
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.749082088470459,
      "learning_rate": 1.3925115248430166e-06,
      "loss": 0.3309,
      "step": 24224
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0161983966827393,
      "learning_rate": 1.3919432089597885e-06,
      "loss": 0.2901,
      "step": 24225
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.043433904647827,
      "learning_rate": 1.3913750003964365e-06,
      "loss": 0.2926,
      "step": 24226
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9132782220840454,
      "learning_rate": 1.390806899160041e-06,
      "loss": 0.3219,
      "step": 24227
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6269009113311768,
      "learning_rate": 1.3902389052576882e-06,
      "loss": 0.1865,
      "step": 24228
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0606534481048584,
      "learning_rate": 1.38967101869646e-06,
      "loss": 0.2629,
      "step": 24229
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8801634311676025,
      "learning_rate": 1.3891032394834326e-06,
      "loss": 0.2439,
      "step": 24230
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9430999755859375,
      "learning_rate": 1.3885355676256885e-06,
      "loss": 0.3857,
      "step": 24231
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.776431918144226,
      "learning_rate": 1.3879680031303045e-06,
      "loss": 0.3128,
      "step": 24232
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8774486780166626,
      "learning_rate": 1.387400546004355e-06,
      "loss": 0.2525,
      "step": 24233
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.048916816711426,
      "learning_rate": 1.386833196254913e-06,
      "loss": 0.2963,
      "step": 24234
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.197331428527832,
      "learning_rate": 1.3862659538890577e-06,
      "loss": 0.3821,
      "step": 24235
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9477111101150513,
      "learning_rate": 1.3856988189138576e-06,
      "loss": 0.3718,
      "step": 24236
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5822843313217163,
      "learning_rate": 1.3851317913363815e-06,
      "loss": 0.2884,
      "step": 24237
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6169534921646118,
      "learning_rate": 1.3845648711637027e-06,
      "loss": 0.3167,
      "step": 24238
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0821609497070312,
      "learning_rate": 1.3839980584028867e-06,
      "loss": 0.2669,
      "step": 24239
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.3205807209014893,
      "learning_rate": 1.383431353061e-06,
      "loss": 0.3308,
      "step": 24240
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7017385959625244,
      "learning_rate": 1.3828647551451102e-06,
      "loss": 0.3157,
      "step": 24241
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8981786966323853,
      "learning_rate": 1.3822982646622795e-06,
      "loss": 0.2804,
      "step": 24242
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0497055053710938,
      "learning_rate": 1.381731881619569e-06,
      "loss": 0.2997,
      "step": 24243
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.010728120803833,
      "learning_rate": 1.381165606024044e-06,
      "loss": 0.2886,
      "step": 24244
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5910828113555908,
      "learning_rate": 1.3805994378827624e-06,
      "loss": 0.2357,
      "step": 24245
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.1709558963775635,
      "learning_rate": 1.380033377202783e-06,
      "loss": 0.3556,
      "step": 24246
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.1030027866363525,
      "learning_rate": 1.3794674239911598e-06,
      "loss": 0.3309,
      "step": 24247
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0173425674438477,
      "learning_rate": 1.3789015782549553e-06,
      "loss": 0.2849,
      "step": 24248
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.094364881515503,
      "learning_rate": 1.3783358400012204e-06,
      "loss": 0.3691,
      "step": 24249
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.237380266189575,
      "learning_rate": 1.377770209237006e-06,
      "loss": 0.3384,
      "step": 24250
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9511133432388306,
      "learning_rate": 1.377204685969369e-06,
      "loss": 0.2637,
      "step": 24251
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0638225078582764,
      "learning_rate": 1.3766392702053566e-06,
      "loss": 0.3144,
      "step": 24252
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9771685600280762,
      "learning_rate": 1.376073961952018e-06,
      "loss": 0.2672,
      "step": 24253
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.987465739250183,
      "learning_rate": 1.3755087612164041e-06,
      "loss": 0.2744,
      "step": 24254
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8354921340942383,
      "learning_rate": 1.3749436680055605e-06,
      "loss": 0.2822,
      "step": 24255
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6519925594329834,
      "learning_rate": 1.3743786823265281e-06,
      "loss": 0.3007,
      "step": 24256
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.72371506690979,
      "learning_rate": 1.373813804186357e-06,
      "loss": 0.3477,
      "step": 24257
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0086443424224854,
      "learning_rate": 1.3732490335920868e-06,
      "loss": 0.3014,
      "step": 24258
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8845688104629517,
      "learning_rate": 1.3726843705507576e-06,
      "loss": 0.3506,
      "step": 24259
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0520153045654297,
      "learning_rate": 1.3721198150694115e-06,
      "loss": 0.3243,
      "step": 24260
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8378715515136719,
      "learning_rate": 1.3715553671550863e-06,
      "loss": 0.3071,
      "step": 24261
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.811596155166626,
      "learning_rate": 1.3709910268148164e-06,
      "loss": 0.3917,
      "step": 24262
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9624862670898438,
      "learning_rate": 1.3704267940556427e-06,
      "loss": 0.3606,
      "step": 24263
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6205135583877563,
      "learning_rate": 1.3698626688845973e-06,
      "loss": 0.2624,
      "step": 24264
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8193355798721313,
      "learning_rate": 1.3692986513087103e-06,
      "loss": 0.2817,
      "step": 24265
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.865349769592285,
      "learning_rate": 1.3687347413350194e-06,
      "loss": 0.3547,
      "step": 24266
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7004669904708862,
      "learning_rate": 1.368170938970551e-06,
      "loss": 0.3161,
      "step": 24267
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0196008682250977,
      "learning_rate": 1.3676072442223365e-06,
      "loss": 0.3182,
      "step": 24268
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8443255424499512,
      "learning_rate": 1.3670436570974e-06,
      "loss": 0.3239,
      "step": 24269
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9639421701431274,
      "learning_rate": 1.3664801776027725e-06,
      "loss": 0.2727,
      "step": 24270
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.123131513595581,
      "learning_rate": 1.3659168057454764e-06,
      "loss": 0.3323,
      "step": 24271
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.396178722381592,
      "learning_rate": 1.3653535415325347e-06,
      "loss": 0.3423,
      "step": 24272
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0566985607147217,
      "learning_rate": 1.3647903849709732e-06,
      "loss": 0.3222,
      "step": 24273
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.2775402069091797,
      "learning_rate": 1.3642273360678116e-06,
      "loss": 0.2336,
      "step": 24274
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.394193410873413,
      "learning_rate": 1.3636643948300664e-06,
      "loss": 0.3245,
      "step": 24275
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5530699491500854,
      "learning_rate": 1.3631015612647613e-06,
      "loss": 0.1861,
      "step": 24276
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7838486433029175,
      "learning_rate": 1.3625388353789115e-06,
      "loss": 0.3037,
      "step": 24277
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0292718410491943,
      "learning_rate": 1.361976217179528e-06,
      "loss": 0.3209,
      "step": 24278
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8610143661499023,
      "learning_rate": 1.3614137066736332e-06,
      "loss": 0.3465,
      "step": 24279
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0452260971069336,
      "learning_rate": 1.3608513038682346e-06,
      "loss": 0.3004,
      "step": 24280
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.808653712272644,
      "learning_rate": 1.3602890087703458e-06,
      "loss": 0.2953,
      "step": 24281
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0294859409332275,
      "learning_rate": 1.3597268213869751e-06,
      "loss": 0.3106,
      "step": 24282
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9724351167678833,
      "learning_rate": 1.3591647417251353e-06,
      "loss": 0.2364,
      "step": 24283
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7148908376693726,
      "learning_rate": 1.3586027697918314e-06,
      "loss": 0.2385,
      "step": 24284
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0183329582214355,
      "learning_rate": 1.358040905594068e-06,
      "loss": 0.3304,
      "step": 24285
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.298961877822876,
      "learning_rate": 1.357479149138855e-06,
      "loss": 0.2454,
      "step": 24286
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.425990343093872,
      "learning_rate": 1.3569175004331924e-06,
      "loss": 0.3169,
      "step": 24287
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9588606357574463,
      "learning_rate": 1.3563559594840813e-06,
      "loss": 0.297,
      "step": 24288
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.4351413249969482,
      "learning_rate": 1.3557945262985273e-06,
      "loss": 0.3531,
      "step": 24289
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.1551287174224854,
      "learning_rate": 1.355233200883528e-06,
      "loss": 0.3419,
      "step": 24290
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8749797344207764,
      "learning_rate": 1.3546719832460786e-06,
      "loss": 0.2801,
      "step": 24291
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.2127788066864014,
      "learning_rate": 1.3541108733931796e-06,
      "loss": 0.308,
      "step": 24292
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0771901607513428,
      "learning_rate": 1.3535498713318262e-06,
      "loss": 0.3289,
      "step": 24293
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7095686197280884,
      "learning_rate": 1.352988977069013e-06,
      "loss": 0.2551,
      "step": 24294
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.950899600982666,
      "learning_rate": 1.3524281906117276e-06,
      "loss": 0.3843,
      "step": 24295
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5036404132843018,
      "learning_rate": 1.351867511966969e-06,
      "loss": 0.2429,
      "step": 24296
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.1310746669769287,
      "learning_rate": 1.3513069411417245e-06,
      "loss": 0.3533,
      "step": 24297
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8336703777313232,
      "learning_rate": 1.350746478142979e-06,
      "loss": 0.2607,
      "step": 24298
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.1762208938598633,
      "learning_rate": 1.3501861229777268e-06,
      "loss": 0.3152,
      "step": 24299
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.781325340270996,
      "learning_rate": 1.34962587565295e-06,
      "loss": 0.3084,
      "step": 24300
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7784055471420288,
      "learning_rate": 1.3490657361756333e-06,
      "loss": 0.3042,
      "step": 24301
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8023725748062134,
      "learning_rate": 1.3485057045527617e-06,
      "loss": 0.3235,
      "step": 24302
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.2213776111602783,
      "learning_rate": 1.3479457807913178e-06,
      "loss": 0.3247,
      "step": 24303
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.145219564437866,
      "learning_rate": 1.347385964898279e-06,
      "loss": 0.3385,
      "step": 24304
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.2811105251312256,
      "learning_rate": 1.346826256880629e-06,
      "loss": 0.3584,
      "step": 24305
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.247664451599121,
      "learning_rate": 1.3462666567453431e-06,
      "loss": 0.3049,
      "step": 24306
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.070448398590088,
      "learning_rate": 1.3457071644994003e-06,
      "loss": 0.3428,
      "step": 24307
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0841732025146484,
      "learning_rate": 1.3451477801497715e-06,
      "loss": 0.3631,
      "step": 24308
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8836443424224854,
      "learning_rate": 1.3445885037034356e-06,
      "loss": 0.3116,
      "step": 24309
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.851325511932373,
      "learning_rate": 1.344029335167365e-06,
      "loss": 0.3385,
      "step": 24310
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9612253904342651,
      "learning_rate": 1.3434702745485261e-06,
      "loss": 0.3338,
      "step": 24311
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.198803424835205,
      "learning_rate": 1.3429113218538947e-06,
      "loss": 0.3219,
      "step": 24312
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.3208324909210205,
      "learning_rate": 1.3423524770904372e-06,
      "loss": 0.3433,
      "step": 24313
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.0150747299194336,
      "learning_rate": 1.3417937402651194e-06,
      "loss": 0.3049,
      "step": 24314
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7392401695251465,
      "learning_rate": 1.3412351113849097e-06,
      "loss": 0.2462,
      "step": 24315
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.923166275024414,
      "learning_rate": 1.340676590456773e-06,
      "loss": 0.3402,
      "step": 24316
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.966926097869873,
      "learning_rate": 1.3401181774876693e-06,
      "loss": 0.2314,
      "step": 24317
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.682478427886963,
      "learning_rate": 1.3395598724845638e-06,
      "loss": 0.2373,
      "step": 24318
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7770870923995972,
      "learning_rate": 1.3390016754544167e-06,
      "loss": 0.3574,
      "step": 24319
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0895493030548096,
      "learning_rate": 1.3384435864041867e-06,
      "loss": 0.3209,
      "step": 24320
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.81528902053833,
      "learning_rate": 1.3378856053408284e-06,
      "loss": 0.3565,
      "step": 24321
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8808619976043701,
      "learning_rate": 1.3373277322713041e-06,
      "loss": 0.355,
      "step": 24322
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8891949653625488,
      "learning_rate": 1.336769967202567e-06,
      "loss": 0.2322,
      "step": 24323
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8189222812652588,
      "learning_rate": 1.336212310141567e-06,
      "loss": 0.3184,
      "step": 24324
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.2304816246032715,
      "learning_rate": 1.3356547610952631e-06,
      "loss": 0.3247,
      "step": 24325
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.2078657150268555,
      "learning_rate": 1.335097320070603e-06,
      "loss": 0.3113,
      "step": 24326
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7333592176437378,
      "learning_rate": 1.3345399870745347e-06,
      "loss": 0.3188,
      "step": 24327
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.9819624423980713,
      "learning_rate": 1.333982762114011e-06,
      "loss": 0.3868,
      "step": 24328
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9608039855957031,
      "learning_rate": 1.3334256451959771e-06,
      "loss": 0.2564,
      "step": 24329
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.3437607288360596,
      "learning_rate": 1.3328686363273758e-06,
      "loss": 0.3356,
      "step": 24330
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1137444972991943,
      "learning_rate": 1.3323117355151572e-06,
      "loss": 0.2972,
      "step": 24331
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.2712972164154053,
      "learning_rate": 1.3317549427662613e-06,
      "loss": 0.4148,
      "step": 24332
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9645320177078247,
      "learning_rate": 1.3311982580876293e-06,
      "loss": 0.2993,
      "step": 24333
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.379807472229004,
      "learning_rate": 1.330641681486201e-06,
      "loss": 0.345,
      "step": 24334
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.3552117347717285,
      "learning_rate": 1.330085212968919e-06,
      "loss": 0.2804,
      "step": 24335
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0668935775756836,
      "learning_rate": 1.3295288525427186e-06,
      "loss": 0.318,
      "step": 24336
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8212809562683105,
      "learning_rate": 1.3289726002145331e-06,
      "loss": 0.303,
      "step": 24337
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8943984508514404,
      "learning_rate": 1.3284164559913048e-06,
      "loss": 0.3098,
      "step": 24338
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.213674306869507,
      "learning_rate": 1.3278604198799617e-06,
      "loss": 0.3711,
      "step": 24339
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9922353029251099,
      "learning_rate": 1.3273044918874367e-06,
      "loss": 0.3162,
      "step": 24340
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.796887755393982,
      "learning_rate": 1.3267486720206635e-06,
      "loss": 0.3072,
      "step": 24341
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.718062400817871,
      "learning_rate": 1.3261929602865697e-06,
      "loss": 0.2519,
      "step": 24342
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7680028676986694,
      "learning_rate": 1.325637356692082e-06,
      "loss": 0.3814,
      "step": 24343
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.37388277053833,
      "learning_rate": 1.3250818612441308e-06,
      "loss": 0.3516,
      "step": 24344
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.184136152267456,
      "learning_rate": 1.324526473949641e-06,
      "loss": 0.2457,
      "step": 24345
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0824356079101562,
      "learning_rate": 1.3239711948155355e-06,
      "loss": 0.3466,
      "step": 24346
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1017892360687256,
      "learning_rate": 1.3234160238487347e-06,
      "loss": 0.3614,
      "step": 24347
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.041489839553833,
      "learning_rate": 1.322860961056166e-06,
      "loss": 0.2442,
      "step": 24348
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.269684076309204,
      "learning_rate": 1.3223060064447467e-06,
      "loss": 0.2917,
      "step": 24349
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1474740505218506,
      "learning_rate": 1.3217511600213927e-06,
      "loss": 0.3269,
      "step": 24350
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.5348076820373535,
      "learning_rate": 1.3211964217930261e-06,
      "loss": 0.3345,
      "step": 24351
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.843517780303955,
      "learning_rate": 1.3206417917665614e-06,
      "loss": 0.3498,
      "step": 24352
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.136948347091675,
      "learning_rate": 1.3200872699489121e-06,
      "loss": 0.3303,
      "step": 24353
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1295480728149414,
      "learning_rate": 1.3195328563469934e-06,
      "loss": 0.242,
      "step": 24354
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0859687328338623,
      "learning_rate": 1.3189785509677177e-06,
      "loss": 0.4324,
      "step": 24355
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8308426141738892,
      "learning_rate": 1.3184243538179919e-06,
      "loss": 0.2683,
      "step": 24356
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7676814794540405,
      "learning_rate": 1.3178702649047303e-06,
      "loss": 0.2706,
      "step": 24357
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.211344003677368,
      "learning_rate": 1.3173162842348397e-06,
      "loss": 0.3607,
      "step": 24358
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6623411178588867,
      "learning_rate": 1.3167624118152255e-06,
      "loss": 0.2446,
      "step": 24359
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.305588483810425,
      "learning_rate": 1.3162086476527902e-06,
      "loss": 0.3224,
      "step": 24360
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.870652437210083,
      "learning_rate": 1.3156549917544447e-06,
      "loss": 0.296,
      "step": 24361
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0036582946777344,
      "learning_rate": 1.315101444127087e-06,
      "loss": 0.3303,
      "step": 24362
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7200731039047241,
      "learning_rate": 1.3145480047776183e-06,
      "loss": 0.2889,
      "step": 24363
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9702017307281494,
      "learning_rate": 1.3139946737129405e-06,
      "loss": 0.3439,
      "step": 24364
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.398857593536377,
      "learning_rate": 1.3134414509399518e-06,
      "loss": 0.3384,
      "step": 24365
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0238280296325684,
      "learning_rate": 1.3128883364655465e-06,
      "loss": 0.3228,
      "step": 24366
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.009519338607788,
      "learning_rate": 1.3123353302966258e-06,
      "loss": 0.2734,
      "step": 24367
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.3064746856689453,
      "learning_rate": 1.3117824324400807e-06,
      "loss": 0.3326,
      "step": 24368
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9392317533493042,
      "learning_rate": 1.3112296429028026e-06,
      "loss": 0.3374,
      "step": 24369
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.028520345687866,
      "learning_rate": 1.3106769616916881e-06,
      "loss": 0.2901,
      "step": 24370
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.389677047729492,
      "learning_rate": 1.310124388813626e-06,
      "loss": 0.2736,
      "step": 24371
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6438151597976685,
      "learning_rate": 1.3095719242755046e-06,
      "loss": 0.2692,
      "step": 24372
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.4113638401031494,
      "learning_rate": 1.3090195680842088e-06,
      "loss": 0.2785,
      "step": 24373
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8413090705871582,
      "learning_rate": 1.3084673202466313e-06,
      "loss": 0.3423,
      "step": 24374
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0808560848236084,
      "learning_rate": 1.3079151807696533e-06,
      "loss": 0.2853,
      "step": 24375
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5028669834136963,
      "learning_rate": 1.307363149660157e-06,
      "loss": 0.2631,
      "step": 24376
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9790314435958862,
      "learning_rate": 1.3068112269250298e-06,
      "loss": 0.3539,
      "step": 24377
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5901670455932617,
      "learning_rate": 1.3062594125711493e-06,
      "loss": 0.2553,
      "step": 24378
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9016259908676147,
      "learning_rate": 1.3057077066053936e-06,
      "loss": 0.3148,
      "step": 24379
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9533616304397583,
      "learning_rate": 1.3051561090346455e-06,
      "loss": 0.2929,
      "step": 24380
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.028318166732788,
      "learning_rate": 1.3046046198657802e-06,
      "loss": 0.2978,
      "step": 24381
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8450332880020142,
      "learning_rate": 1.3040532391056704e-06,
      "loss": 0.3331,
      "step": 24382
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8905366659164429,
      "learning_rate": 1.3035019667611949e-06,
      "loss": 0.2975,
      "step": 24383
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9545749425888062,
      "learning_rate": 1.3029508028392234e-06,
      "loss": 0.2836,
      "step": 24384
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.029935598373413,
      "learning_rate": 1.3023997473466299e-06,
      "loss": 0.3286,
      "step": 24385
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.995221734046936,
      "learning_rate": 1.3018488002902807e-06,
      "loss": 0.2966,
      "step": 24386
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8857015371322632,
      "learning_rate": 1.3012979616770494e-06,
      "loss": 0.3301,
      "step": 24387
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9407808780670166,
      "learning_rate": 1.3007472315138015e-06,
      "loss": 0.3139,
      "step": 24388
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7391490936279297,
      "learning_rate": 1.3001966098074004e-06,
      "loss": 0.3458,
      "step": 24389
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6796138286590576,
      "learning_rate": 1.2996460965647162e-06,
      "loss": 0.2844,
      "step": 24390
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1400492191314697,
      "learning_rate": 1.2990956917926089e-06,
      "loss": 0.3712,
      "step": 24391
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1032605171203613,
      "learning_rate": 1.29854539549794e-06,
      "loss": 0.3107,
      "step": 24392
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9381635189056396,
      "learning_rate": 1.2979952076875735e-06,
      "loss": 0.2705,
      "step": 24393
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8101428747177124,
      "learning_rate": 1.2974451283683675e-06,
      "loss": 0.2756,
      "step": 24394
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0843212604522705,
      "learning_rate": 1.2968951575471766e-06,
      "loss": 0.3966,
      "step": 24395
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.860602617263794,
      "learning_rate": 1.2963452952308631e-06,
      "loss": 0.2286,
      "step": 24396
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.34220552444458,
      "learning_rate": 1.295795541426279e-06,
      "loss": 0.3441,
      "step": 24397
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.020124912261963,
      "learning_rate": 1.2952458961402792e-06,
      "loss": 0.36,
      "step": 24398
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.771436095237732,
      "learning_rate": 1.2946963593797135e-06,
      "loss": 0.2832,
      "step": 24399
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.4480090141296387,
      "learning_rate": 1.2941469311514375e-06,
      "loss": 0.3355,
      "step": 24400
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.112191677093506,
      "learning_rate": 1.2935976114623005e-06,
      "loss": 0.3381,
      "step": 24401
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8444432020187378,
      "learning_rate": 1.2930484003191467e-06,
      "loss": 0.2702,
      "step": 24402
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9160144329071045,
      "learning_rate": 1.2924992977288297e-06,
      "loss": 0.2911,
      "step": 24403
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.82430100440979,
      "learning_rate": 1.2919503036981918e-06,
      "loss": 0.2388,
      "step": 24404
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.8197267055511475,
      "learning_rate": 1.2914014182340751e-06,
      "loss": 0.293,
      "step": 24405
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.2671263217926025,
      "learning_rate": 1.2908526413433288e-06,
      "loss": 0.3591,
      "step": 24406
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7291563749313354,
      "learning_rate": 1.2903039730327904e-06,
      "loss": 0.2926,
      "step": 24407
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.802851915359497,
      "learning_rate": 1.2897554133093005e-06,
      "loss": 0.3508,
      "step": 24408
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.095184803009033,
      "learning_rate": 1.289206962179701e-06,
      "loss": 0.2819,
      "step": 24409
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.8817123174667358,
      "learning_rate": 1.2886586196508266e-06,
      "loss": 0.1843,
      "step": 24410
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.0019760131835938,
      "learning_rate": 1.2881103857295163e-06,
      "loss": 0.3006,
      "step": 24411
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9773203134536743,
      "learning_rate": 1.2875622604226e-06,
      "loss": 0.3392,
      "step": 24412
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.07443904876709,
      "learning_rate": 1.2870142437369193e-06,
      "loss": 0.3988,
      "step": 24413
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1064674854278564,
      "learning_rate": 1.2864663356793006e-06,
      "loss": 0.2855,
      "step": 24414
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8824443817138672,
      "learning_rate": 1.2859185362565751e-06,
      "loss": 0.2616,
      "step": 24415
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6956971883773804,
      "learning_rate": 1.2853708454755764e-06,
      "loss": 0.2727,
      "step": 24416
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.141878604888916,
      "learning_rate": 1.284823263343129e-06,
      "loss": 0.3364,
      "step": 24417
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.02952241897583,
      "learning_rate": 1.2842757898660608e-06,
      "loss": 0.3046,
      "step": 24418
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6829556226730347,
      "learning_rate": 1.2837284250511982e-06,
      "loss": 0.2567,
      "step": 24419
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8349112272262573,
      "learning_rate": 1.2831811689053653e-06,
      "loss": 0.3066,
      "step": 24420
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.009408950805664,
      "learning_rate": 1.2826340214353828e-06,
      "loss": 0.2825,
      "step": 24421
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7530349493026733,
      "learning_rate": 1.2820869826480763e-06,
      "loss": 0.2322,
      "step": 24422
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0692975521087646,
      "learning_rate": 1.281540052550263e-06,
      "loss": 0.2527,
      "step": 24423
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.366267681121826,
      "learning_rate": 1.2809932311487616e-06,
      "loss": 0.3164,
      "step": 24424
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7382997274398804,
      "learning_rate": 1.2804465184503888e-06,
      "loss": 0.2944,
      "step": 24425
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.1437668800354004,
      "learning_rate": 1.2798999144619639e-06,
      "loss": 0.3168,
      "step": 24426
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8245588541030884,
      "learning_rate": 1.2793534191903e-06,
      "loss": 0.2648,
      "step": 24427
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9697834253311157,
      "learning_rate": 1.2788070326422086e-06,
      "loss": 0.2807,
      "step": 24428
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.1716437339782715,
      "learning_rate": 1.278260754824504e-06,
      "loss": 0.2948,
      "step": 24429
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7261708974838257,
      "learning_rate": 1.2777145857439977e-06,
      "loss": 0.2787,
      "step": 24430
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.3461203575134277,
      "learning_rate": 1.2771685254074939e-06,
      "loss": 0.2897,
      "step": 24431
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6765329837799072,
      "learning_rate": 1.2766225738218075e-06,
      "loss": 0.2554,
      "step": 24432
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6701514720916748,
      "learning_rate": 1.2760767309937406e-06,
      "loss": 0.3523,
      "step": 24433
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7408740520477295,
      "learning_rate": 1.2755309969300978e-06,
      "loss": 0.3177,
      "step": 24434
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.70602285861969,
      "learning_rate": 1.274985371637687e-06,
      "loss": 0.2384,
      "step": 24435
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.695189118385315,
      "learning_rate": 1.2744398551233084e-06,
      "loss": 0.2647,
      "step": 24436
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7150170803070068,
      "learning_rate": 1.273894447393763e-06,
      "loss": 0.3358,
      "step": 24437
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8638851642608643,
      "learning_rate": 1.273349148455848e-06,
      "loss": 0.2829,
      "step": 24438
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.703768014907837,
      "learning_rate": 1.2728039583163675e-06,
      "loss": 0.3055,
      "step": 24439
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.2141222953796387,
      "learning_rate": 1.2722588769821154e-06,
      "loss": 0.2906,
      "step": 24440
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9910709857940674,
      "learning_rate": 1.2717139044598858e-06,
      "loss": 0.3239,
      "step": 24441
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9227426052093506,
      "learning_rate": 1.271169040756477e-06,
      "loss": 0.2871,
      "step": 24442
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.149013042449951,
      "learning_rate": 1.2706242858786799e-06,
      "loss": 0.3062,
      "step": 24443
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.910717248916626,
      "learning_rate": 1.2700796398332848e-06,
      "loss": 0.2366,
      "step": 24444
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8405587673187256,
      "learning_rate": 1.2695351026270853e-06,
      "loss": 0.3428,
      "step": 24445
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7243385314941406,
      "learning_rate": 1.2689906742668678e-06,
      "loss": 0.3019,
      "step": 24446
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9695323705673218,
      "learning_rate": 1.2684463547594195e-06,
      "loss": 0.3428,
      "step": 24447
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9251688718795776,
      "learning_rate": 1.2679021441115292e-06,
      "loss": 0.2663,
      "step": 24448
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9353257417678833,
      "learning_rate": 1.2673580423299803e-06,
      "loss": 0.3148,
      "step": 24449
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.010684013366699,
      "learning_rate": 1.2668140494215564e-06,
      "loss": 0.3221,
      "step": 24450
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.855926513671875,
      "learning_rate": 1.2662701653930376e-06,
      "loss": 0.2473,
      "step": 24451
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0080251693725586,
      "learning_rate": 1.2657263902512097e-06,
      "loss": 0.3132,
      "step": 24452
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5734484195709229,
      "learning_rate": 1.2651827240028481e-06,
      "loss": 0.304,
      "step": 24453
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8635305166244507,
      "learning_rate": 1.2646391666547308e-06,
      "loss": 0.2996,
      "step": 24454
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.05155348777771,
      "learning_rate": 1.264095718213637e-06,
      "loss": 0.2987,
      "step": 24455
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.2433242797851562,
      "learning_rate": 1.2635523786863414e-06,
      "loss": 0.3074,
      "step": 24456
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8424962759017944,
      "learning_rate": 1.2630091480796147e-06,
      "loss": 0.2536,
      "step": 24457
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7958409786224365,
      "learning_rate": 1.2624660264002343e-06,
      "loss": 0.3596,
      "step": 24458
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6834980249404907,
      "learning_rate": 1.2619230136549698e-06,
      "loss": 0.3039,
      "step": 24459
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8517746925354004,
      "learning_rate": 1.2613801098505884e-06,
      "loss": 0.2601,
      "step": 24460
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.2088632583618164,
      "learning_rate": 1.2608373149938635e-06,
      "loss": 0.3268,
      "step": 24461
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.193432092666626,
      "learning_rate": 1.2602946290915597e-06,
      "loss": 0.2771,
      "step": 24462
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.953330397605896,
      "learning_rate": 1.2597520521504436e-06,
      "loss": 0.3035,
      "step": 24463
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9904117584228516,
      "learning_rate": 1.2592095841772768e-06,
      "loss": 0.2697,
      "step": 24464
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6059496402740479,
      "learning_rate": 1.2586672251788268e-06,
      "loss": 0.2575,
      "step": 24465
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.392195224761963,
      "learning_rate": 1.258124975161854e-06,
      "loss": 0.3935,
      "step": 24466
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.1644949913024902,
      "learning_rate": 1.2575828341331153e-06,
      "loss": 0.2844,
      "step": 24467
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.538443922996521,
      "learning_rate": 1.2570408020993752e-06,
      "loss": 0.1892,
      "step": 24468
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.157291889190674,
      "learning_rate": 1.2564988790673893e-06,
      "loss": 0.3479,
      "step": 24469
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9394900798797607,
      "learning_rate": 1.2559570650439112e-06,
      "loss": 0.3124,
      "step": 24470
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7741543054580688,
      "learning_rate": 1.2554153600357e-06,
      "loss": 0.336,
      "step": 24471
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9865095615386963,
      "learning_rate": 1.254873764049509e-06,
      "loss": 0.2849,
      "step": 24472
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.015254497528076,
      "learning_rate": 1.2543322770920863e-06,
      "loss": 0.281,
      "step": 24473
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9566441774368286,
      "learning_rate": 1.2537908991701874e-06,
      "loss": 0.323,
      "step": 24474
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8803677558898926,
      "learning_rate": 1.2532496302905606e-06,
      "loss": 0.3116,
      "step": 24475
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9547218084335327,
      "learning_rate": 1.2527084704599534e-06,
      "loss": 0.3662,
      "step": 24476
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7764427661895752,
      "learning_rate": 1.2521674196851109e-06,
      "loss": 0.3087,
      "step": 24477
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0294482707977295,
      "learning_rate": 1.2516264779727826e-06,
      "loss": 0.2895,
      "step": 24478
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.272949457168579,
      "learning_rate": 1.2510856453297104e-06,
      "loss": 0.3123,
      "step": 24479
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0658299922943115,
      "learning_rate": 1.2505449217626354e-06,
      "loss": 0.3497,
      "step": 24480
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.2067418098449707,
      "learning_rate": 1.250004307278303e-06,
      "loss": 0.2796,
      "step": 24481
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.155229091644287,
      "learning_rate": 1.2494638018834504e-06,
      "loss": 0.3019,
      "step": 24482
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.174053192138672,
      "learning_rate": 1.2489234055848154e-06,
      "loss": 0.3297,
      "step": 24483
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.3740358352661133,
      "learning_rate": 1.2483831183891394e-06,
      "loss": 0.3529,
      "step": 24484
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.931910514831543,
      "learning_rate": 1.2478429403031555e-06,
      "loss": 0.2593,
      "step": 24485
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8923189640045166,
      "learning_rate": 1.2473028713335966e-06,
      "loss": 0.2491,
      "step": 24486
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.092592716217041,
      "learning_rate": 1.2467629114871993e-06,
      "loss": 0.3865,
      "step": 24487
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.904474139213562,
      "learning_rate": 1.246223060770696e-06,
      "loss": 0.3019,
      "step": 24488
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0745606422424316,
      "learning_rate": 1.2456833191908136e-06,
      "loss": 0.2968,
      "step": 24489
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.1459579467773438,
      "learning_rate": 1.245143686754282e-06,
      "loss": 0.2515,
      "step": 24490
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9548395872116089,
      "learning_rate": 1.2446041634678318e-06,
      "loss": 0.3314,
      "step": 24491
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9180846214294434,
      "learning_rate": 1.2440647493381874e-06,
      "loss": 0.3437,
      "step": 24492
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.285548210144043,
      "learning_rate": 1.2435254443720734e-06,
      "loss": 0.3444,
      "step": 24493
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.3298468589782715,
      "learning_rate": 1.2429862485762156e-06,
      "loss": 0.3601,
      "step": 24494
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8183766603469849,
      "learning_rate": 1.2424471619573342e-06,
      "loss": 0.325,
      "step": 24495
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6782598495483398,
      "learning_rate": 1.2419081845221502e-06,
      "loss": 0.3346,
      "step": 24496
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9774867296218872,
      "learning_rate": 1.2413693162773855e-06,
      "loss": 0.2962,
      "step": 24497
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.4463887214660645,
      "learning_rate": 1.2408305572297575e-06,
      "loss": 0.3869,
      "step": 24498
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6998668909072876,
      "learning_rate": 1.2402919073859798e-06,
      "loss": 0.2241,
      "step": 24499
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7542842626571655,
      "learning_rate": 1.2397533667527739e-06,
      "loss": 0.3035,
      "step": 24500
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.077263116836548,
      "learning_rate": 1.23921493533685e-06,
      "loss": 0.3632,
      "step": 24501
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0955312252044678,
      "learning_rate": 1.2386766131449213e-06,
      "loss": 0.3133,
      "step": 24502
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.4052236080169678,
      "learning_rate": 1.2381384001836982e-06,
      "loss": 0.3148,
      "step": 24503
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0854997634887695,
      "learning_rate": 1.2376002964598942e-06,
      "loss": 0.2746,
      "step": 24504
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.1672887802124023,
      "learning_rate": 1.2370623019802163e-06,
      "loss": 0.309,
      "step": 24505
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.941602349281311,
      "learning_rate": 1.23652441675137e-06,
      "loss": 0.3036,
      "step": 24506
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0529792308807373,
      "learning_rate": 1.2359866407800648e-06,
      "loss": 0.338,
      "step": 24507
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.023185968399048,
      "learning_rate": 1.2354489740730035e-06,
      "loss": 0.2931,
      "step": 24508
    },
    {
      "epoch": 2.53,
      "grad_norm": 2.0965256690979004,
      "learning_rate": 1.234911416636888e-06,
      "loss": 0.3467,
      "step": 24509
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.9915194511413574,
      "learning_rate": 1.2343739684784228e-06,
      "loss": 0.3166,
      "step": 24510
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0047707557678223,
      "learning_rate": 1.233836629604308e-06,
      "loss": 0.2916,
      "step": 24511
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.2142980098724365,
      "learning_rate": 1.2332994000212407e-06,
      "loss": 0.2894,
      "step": 24512
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8463990688323975,
      "learning_rate": 1.2327622797359219e-06,
      "loss": 0.3529,
      "step": 24513
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0883853435516357,
      "learning_rate": 1.2322252687550452e-06,
      "loss": 0.3563,
      "step": 24514
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.616889238357544,
      "learning_rate": 1.2316883670853085e-06,
      "loss": 0.3541,
      "step": 24515
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9379534721374512,
      "learning_rate": 1.2311515747334012e-06,
      "loss": 0.2748,
      "step": 24516
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.059126377105713,
      "learning_rate": 1.2306148917060213e-06,
      "loss": 0.3679,
      "step": 24517
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.3183488845825195,
      "learning_rate": 1.2300783180098563e-06,
      "loss": 0.3398,
      "step": 24518
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.3180830478668213,
      "learning_rate": 1.2295418536515946e-06,
      "loss": 0.2819,
      "step": 24519
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8646659851074219,
      "learning_rate": 1.2290054986379285e-06,
      "loss": 0.338,
      "step": 24520
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.6848089694976807,
      "learning_rate": 1.2284692529755427e-06,
      "loss": 0.3113,
      "step": 24521
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0594818592071533,
      "learning_rate": 1.2279331166711218e-06,
      "loss": 0.3245,
      "step": 24522
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0107815265655518,
      "learning_rate": 1.2273970897313526e-06,
      "loss": 0.2849,
      "step": 24523
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.4367737770080566,
      "learning_rate": 1.2268611721629164e-06,
      "loss": 0.3561,
      "step": 24524
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0269582271575928,
      "learning_rate": 1.2263253639724925e-06,
      "loss": 0.2198,
      "step": 24525
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8345104455947876,
      "learning_rate": 1.2257896651667667e-06,
      "loss": 0.2571,
      "step": 24526
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.081538677215576,
      "learning_rate": 1.2252540757524133e-06,
      "loss": 0.2923,
      "step": 24527
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.060652732849121,
      "learning_rate": 1.2247185957361107e-06,
      "loss": 0.3314,
      "step": 24528
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9377652406692505,
      "learning_rate": 1.2241832251245345e-06,
      "loss": 0.3449,
      "step": 24529
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9562411308288574,
      "learning_rate": 1.2236479639243603e-06,
      "loss": 0.3059,
      "step": 24530
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.646732807159424,
      "learning_rate": 1.223112812142262e-06,
      "loss": 0.3856,
      "step": 24531
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.243502616882324,
      "learning_rate": 1.2225777697849085e-06,
      "loss": 0.3581,
      "step": 24532
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0107173919677734,
      "learning_rate": 1.2220428368589743e-06,
      "loss": 0.2501,
      "step": 24533
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8177080154418945,
      "learning_rate": 1.2215080133711267e-06,
      "loss": 0.3,
      "step": 24534
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9391794204711914,
      "learning_rate": 1.220973299328031e-06,
      "loss": 0.332,
      "step": 24535
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.17360258102417,
      "learning_rate": 1.220438694736359e-06,
      "loss": 0.3068,
      "step": 24536
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9609084129333496,
      "learning_rate": 1.2199041996027728e-06,
      "loss": 0.2646,
      "step": 24537
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.03550386428833,
      "learning_rate": 1.2193698139339339e-06,
      "loss": 0.3028,
      "step": 24538
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9425731897354126,
      "learning_rate": 1.2188355377365101e-06,
      "loss": 0.3655,
      "step": 24539
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1839053630828857,
      "learning_rate": 1.2183013710171599e-06,
      "loss": 0.3436,
      "step": 24540
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.123370885848999,
      "learning_rate": 1.217767313782542e-06,
      "loss": 0.2943,
      "step": 24541
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5399177074432373,
      "learning_rate": 1.2172333660393122e-06,
      "loss": 0.2502,
      "step": 24542
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5842347145080566,
      "learning_rate": 1.216699527794134e-06,
      "loss": 0.3239,
      "step": 24543
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8160003423690796,
      "learning_rate": 1.2161657990536591e-06,
      "loss": 0.2897,
      "step": 24544
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8910034894943237,
      "learning_rate": 1.2156321798245396e-06,
      "loss": 0.2779,
      "step": 24545
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.347395420074463,
      "learning_rate": 1.2150986701134337e-06,
      "loss": 0.3933,
      "step": 24546
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8071149587631226,
      "learning_rate": 1.2145652699269894e-06,
      "loss": 0.3307,
      "step": 24547
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9627666473388672,
      "learning_rate": 1.2140319792718546e-06,
      "loss": 0.3039,
      "step": 24548
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.015526533126831,
      "learning_rate": 1.213498798154684e-06,
      "loss": 0.3678,
      "step": 24549
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1964588165283203,
      "learning_rate": 1.2129657265821215e-06,
      "loss": 0.2926,
      "step": 24550
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.678876280784607,
      "learning_rate": 1.2124327645608113e-06,
      "loss": 0.3067,
      "step": 24551
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.84548819065094,
      "learning_rate": 1.211899912097403e-06,
      "loss": 0.2512,
      "step": 24552
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7696059942245483,
      "learning_rate": 1.2113671691985352e-06,
      "loss": 0.2702,
      "step": 24553
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8077846765518188,
      "learning_rate": 1.210834535870853e-06,
      "loss": 0.2833,
      "step": 24554
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.854062795639038,
      "learning_rate": 1.210302012120993e-06,
      "loss": 0.3233,
      "step": 24555
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1182048320770264,
      "learning_rate": 1.209769597955599e-06,
      "loss": 0.2975,
      "step": 24556
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6691100597381592,
      "learning_rate": 1.2092372933813068e-06,
      "loss": 0.2478,
      "step": 24557
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.5049803256988525,
      "learning_rate": 1.2087050984047499e-06,
      "loss": 0.3779,
      "step": 24558
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8824751377105713,
      "learning_rate": 1.2081730130325697e-06,
      "loss": 0.3057,
      "step": 24559
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.217542886734009,
      "learning_rate": 1.2076410372713953e-06,
      "loss": 0.2731,
      "step": 24560
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8569549322128296,
      "learning_rate": 1.2071091711278581e-06,
      "loss": 0.2927,
      "step": 24561
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8137344121932983,
      "learning_rate": 1.206577414608594e-06,
      "loss": 0.2837,
      "step": 24562
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.171466112136841,
      "learning_rate": 1.2060457677202297e-06,
      "loss": 0.3562,
      "step": 24563
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8539820909500122,
      "learning_rate": 1.2055142304693912e-06,
      "loss": 0.3444,
      "step": 24564
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.058861494064331,
      "learning_rate": 1.2049828028627098e-06,
      "loss": 0.2465,
      "step": 24565
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.904831051826477,
      "learning_rate": 1.204451484906809e-06,
      "loss": 0.3061,
      "step": 24566
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.795100212097168,
      "learning_rate": 1.2039202766083125e-06,
      "loss": 0.3745,
      "step": 24567
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.009803533554077,
      "learning_rate": 1.2033891779738416e-06,
      "loss": 0.3504,
      "step": 24568
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.061971426010132,
      "learning_rate": 1.202858189010021e-06,
      "loss": 0.3315,
      "step": 24569
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.371800661087036,
      "learning_rate": 1.20232730972347e-06,
      "loss": 0.3926,
      "step": 24570
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.136779546737671,
      "learning_rate": 1.2017965401208032e-06,
      "loss": 0.286,
      "step": 24571
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.4726407527923584,
      "learning_rate": 1.2012658802086451e-06,
      "loss": 0.3171,
      "step": 24572
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.159909725189209,
      "learning_rate": 1.2007353299936064e-06,
      "loss": 0.3043,
      "step": 24573
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9498307704925537,
      "learning_rate": 1.2002048894823014e-06,
      "loss": 0.3137,
      "step": 24574
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.4076929092407227,
      "learning_rate": 1.1996745586813473e-06,
      "loss": 0.3582,
      "step": 24575
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.247713327407837,
      "learning_rate": 1.1991443375973533e-06,
      "loss": 0.3455,
      "step": 24576
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8184701204299927,
      "learning_rate": 1.1986142262369282e-06,
      "loss": 0.2862,
      "step": 24577
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1922318935394287,
      "learning_rate": 1.1980842246066848e-06,
      "loss": 0.3383,
      "step": 24578
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1801371574401855,
      "learning_rate": 1.1975543327132288e-06,
      "loss": 0.3654,
      "step": 24579
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0359866619110107,
      "learning_rate": 1.1970245505631673e-06,
      "loss": 0.2538,
      "step": 24580
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.054468870162964,
      "learning_rate": 1.1964948781631025e-06,
      "loss": 0.3142,
      "step": 24581
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8722318410873413,
      "learning_rate": 1.1959653155196428e-06,
      "loss": 0.337,
      "step": 24582
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.3390512466430664,
      "learning_rate": 1.195435862639387e-06,
      "loss": 0.3192,
      "step": 24583
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1242992877960205,
      "learning_rate": 1.1949065195289355e-06,
      "loss": 0.3179,
      "step": 24584
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6864780187606812,
      "learning_rate": 1.194377286194892e-06,
      "loss": 0.3113,
      "step": 24585
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8465213775634766,
      "learning_rate": 1.1938481626438515e-06,
      "loss": 0.2905,
      "step": 24586
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0108344554901123,
      "learning_rate": 1.1933191488824091e-06,
      "loss": 0.3778,
      "step": 24587
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5836008787155151,
      "learning_rate": 1.1927902449171647e-06,
      "loss": 0.2969,
      "step": 24588
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.274726629257202,
      "learning_rate": 1.1922614507547092e-06,
      "loss": 0.3354,
      "step": 24589
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9062037467956543,
      "learning_rate": 1.1917327664016342e-06,
      "loss": 0.3069,
      "step": 24590
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.578908085823059,
      "learning_rate": 1.1912041918645356e-06,
      "loss": 0.2263,
      "step": 24591
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.9288455247879028,
      "learning_rate": 1.1906757271500002e-06,
      "loss": 0.3464,
      "step": 24592
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.2288308143615723,
      "learning_rate": 1.1901473722646173e-06,
      "loss": 0.3324,
      "step": 24593
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.2106974124908447,
      "learning_rate": 1.1896191272149704e-06,
      "loss": 0.2599,
      "step": 24594
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.269437074661255,
      "learning_rate": 1.189090992007652e-06,
      "loss": 0.3446,
      "step": 24595
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8875318765640259,
      "learning_rate": 1.1885629666492426e-06,
      "loss": 0.2887,
      "step": 24596
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5719146728515625,
      "learning_rate": 1.1880350511463246e-06,
      "loss": 0.1909,
      "step": 24597
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.2201852798461914,
      "learning_rate": 1.1875072455054826e-06,
      "loss": 0.3634,
      "step": 24598
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8121579885482788,
      "learning_rate": 1.186979549733296e-06,
      "loss": 0.3958,
      "step": 24599
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.08726167678833,
      "learning_rate": 1.1864519638363403e-06,
      "loss": 0.244,
      "step": 24600
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.086832046508789,
      "learning_rate": 1.1859244878211984e-06,
      "loss": 0.3096,
      "step": 24601
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.046058177947998,
      "learning_rate": 1.1853971216944448e-06,
      "loss": 0.37,
      "step": 24602
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.035964250564575,
      "learning_rate": 1.1848698654626512e-06,
      "loss": 0.3084,
      "step": 24603
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0341970920562744,
      "learning_rate": 1.1843427191323954e-06,
      "loss": 0.3997,
      "step": 24604
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.3310861587524414,
      "learning_rate": 1.1838156827102476e-06,
      "loss": 0.2924,
      "step": 24605
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.1828856468200684,
      "learning_rate": 1.1832887562027795e-06,
      "loss": 0.2811,
      "step": 24606
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8841371536254883,
      "learning_rate": 1.1827619396165568e-06,
      "loss": 0.2285,
      "step": 24607
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.476125955581665,
      "learning_rate": 1.1822352329581533e-06,
      "loss": 0.3379,
      "step": 24608
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1573753356933594,
      "learning_rate": 1.1817086362341324e-06,
      "loss": 0.3121,
      "step": 24609
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.938222885131836,
      "learning_rate": 1.1811821494510577e-06,
      "loss": 0.2601,
      "step": 24610
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.962878942489624,
      "learning_rate": 1.1806557726155e-06,
      "loss": 0.2611,
      "step": 24611
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.890322208404541,
      "learning_rate": 1.1801295057340135e-06,
      "loss": 0.3353,
      "step": 24612
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9008762836456299,
      "learning_rate": 1.179603348813162e-06,
      "loss": 0.3024,
      "step": 24613
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.3188724517822266,
      "learning_rate": 1.179077301859507e-06,
      "loss": 0.3647,
      "step": 24614
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.067840814590454,
      "learning_rate": 1.1785513648796077e-06,
      "loss": 0.3464,
      "step": 24615
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.5282914638519287,
      "learning_rate": 1.1780255378800166e-06,
      "loss": 0.359,
      "step": 24616
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.91683828830719,
      "learning_rate": 1.177499820867294e-06,
      "loss": 0.2092,
      "step": 24617
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0651257038116455,
      "learning_rate": 1.1769742138479933e-06,
      "loss": 0.3046,
      "step": 24618
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.932375192642212,
      "learning_rate": 1.1764487168286642e-06,
      "loss": 0.2758,
      "step": 24619
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1848232746124268,
      "learning_rate": 1.1759233298158624e-06,
      "loss": 0.3206,
      "step": 24620
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.214373826980591,
      "learning_rate": 1.175398052816137e-06,
      "loss": 0.2676,
      "step": 24621
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.812367558479309,
      "learning_rate": 1.1748728858360348e-06,
      "loss": 0.3283,
      "step": 24622
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.286515474319458,
      "learning_rate": 1.1743478288821054e-06,
      "loss": 0.297,
      "step": 24623
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.194535732269287,
      "learning_rate": 1.1738228819608955e-06,
      "loss": 0.3003,
      "step": 24624
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.846942663192749,
      "learning_rate": 1.1732980450789478e-06,
      "loss": 0.2641,
      "step": 24625
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1648757457733154,
      "learning_rate": 1.1727733182428048e-06,
      "loss": 0.2823,
      "step": 24626
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.35005259513855,
      "learning_rate": 1.172248701459011e-06,
      "loss": 0.3556,
      "step": 24627
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8184478282928467,
      "learning_rate": 1.1717241947341074e-06,
      "loss": 0.3181,
      "step": 24628
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8341983556747437,
      "learning_rate": 1.1711997980746293e-06,
      "loss": 0.3462,
      "step": 24629
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0120630264282227,
      "learning_rate": 1.1706755114871181e-06,
      "loss": 0.2841,
      "step": 24630
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.951495885848999,
      "learning_rate": 1.1701513349781101e-06,
      "loss": 0.3299,
      "step": 24631
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1219851970672607,
      "learning_rate": 1.1696272685541377e-06,
      "loss": 0.2991,
      "step": 24632
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.195990562438965,
      "learning_rate": 1.1691033122217387e-06,
      "loss": 0.3947,
      "step": 24633
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.811208486557007,
      "learning_rate": 1.1685794659874427e-06,
      "loss": 0.3934,
      "step": 24634
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5322020053863525,
      "learning_rate": 1.16805572985778e-06,
      "loss": 0.261,
      "step": 24635
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.302407741546631,
      "learning_rate": 1.1675321038392827e-06,
      "loss": 0.3937,
      "step": 24636
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9928357601165771,
      "learning_rate": 1.1670085879384785e-06,
      "loss": 0.349,
      "step": 24637
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0161046981811523,
      "learning_rate": 1.1664851821618928e-06,
      "loss": 0.2937,
      "step": 24638
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.092069625854492,
      "learning_rate": 1.1659618865160505e-06,
      "loss": 0.302,
      "step": 24639
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1433279514312744,
      "learning_rate": 1.1654387010074787e-06,
      "loss": 0.3715,
      "step": 24640
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0878548622131348,
      "learning_rate": 1.1649156256426986e-06,
      "loss": 0.3903,
      "step": 24641
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6792792081832886,
      "learning_rate": 1.1643926604282297e-06,
      "loss": 0.2911,
      "step": 24642
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.7351653575897217,
      "learning_rate": 1.1638698053705954e-06,
      "loss": 0.3661,
      "step": 24643
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0617594718933105,
      "learning_rate": 1.163347060476313e-06,
      "loss": 0.2636,
      "step": 24644
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.3268320560455322,
      "learning_rate": 1.1628244257518973e-06,
      "loss": 0.2742,
      "step": 24645
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9922220706939697,
      "learning_rate": 1.1623019012038695e-06,
      "loss": 0.3718,
      "step": 24646
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.276007890701294,
      "learning_rate": 1.16177948683874e-06,
      "loss": 0.2833,
      "step": 24647
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8512284755706787,
      "learning_rate": 1.1612571826630204e-06,
      "loss": 0.3446,
      "step": 24648
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1588144302368164,
      "learning_rate": 1.1607349886832287e-06,
      "loss": 0.2877,
      "step": 24649
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.280247449874878,
      "learning_rate": 1.1602129049058697e-06,
      "loss": 0.2899,
      "step": 24650
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8625105619430542,
      "learning_rate": 1.159690931337456e-06,
      "loss": 0.2966,
      "step": 24651
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8014806509017944,
      "learning_rate": 1.1591690679844903e-06,
      "loss": 0.285,
      "step": 24652
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0409250259399414,
      "learning_rate": 1.158647314853485e-06,
      "loss": 0.32,
      "step": 24653
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1112961769104004,
      "learning_rate": 1.1581256719509415e-06,
      "loss": 0.3088,
      "step": 24654
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9951519966125488,
      "learning_rate": 1.1576041392833616e-06,
      "loss": 0.401,
      "step": 24655
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.5276007652282715,
      "learning_rate": 1.157082716857253e-06,
      "loss": 0.4223,
      "step": 24656
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6414135694503784,
      "learning_rate": 1.1565614046791119e-06,
      "loss": 0.2292,
      "step": 24657
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.571244239807129,
      "learning_rate": 1.1560402027554386e-06,
      "loss": 0.3212,
      "step": 24658
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8231252431869507,
      "learning_rate": 1.1555191110927323e-06,
      "loss": 0.3072,
      "step": 24659
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9026747941970825,
      "learning_rate": 1.154998129697489e-06,
      "loss": 0.2986,
      "step": 24660
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8803917169570923,
      "learning_rate": 1.154477258576202e-06,
      "loss": 0.3994,
      "step": 24661
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9552756547927856,
      "learning_rate": 1.153956497735369e-06,
      "loss": 0.346,
      "step": 24662
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.637140154838562,
      "learning_rate": 1.153435847181481e-06,
      "loss": 0.2849,
      "step": 24663
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.773187279701233,
      "learning_rate": 1.1529153069210285e-06,
      "loss": 0.2585,
      "step": 24664
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.908342957496643,
      "learning_rate": 1.1523948769604998e-06,
      "loss": 0.2497,
      "step": 24665
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9864500761032104,
      "learning_rate": 1.151874557306386e-06,
      "loss": 0.3613,
      "step": 24666
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8884638547897339,
      "learning_rate": 1.1513543479651746e-06,
      "loss": 0.36,
      "step": 24667
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.77974271774292,
      "learning_rate": 1.1508342489433465e-06,
      "loss": 0.2348,
      "step": 24668
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.019237995147705,
      "learning_rate": 1.1503142602473915e-06,
      "loss": 0.3582,
      "step": 24669
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9714549779891968,
      "learning_rate": 1.1497943818837908e-06,
      "loss": 0.3204,
      "step": 24670
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1638410091400146,
      "learning_rate": 1.1492746138590228e-06,
      "loss": 0.3127,
      "step": 24671
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8902714252471924,
      "learning_rate": 1.1487549561795718e-06,
      "loss": 0.3166,
      "step": 24672
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9113787412643433,
      "learning_rate": 1.1482354088519154e-06,
      "loss": 0.2841,
      "step": 24673
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7636631727218628,
      "learning_rate": 1.1477159718825292e-06,
      "loss": 0.2508,
      "step": 24674
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1455652713775635,
      "learning_rate": 1.1471966452778916e-06,
      "loss": 0.3196,
      "step": 24675
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8731578588485718,
      "learning_rate": 1.146677429044477e-06,
      "loss": 0.2851,
      "step": 24676
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6732205152511597,
      "learning_rate": 1.1461583231887573e-06,
      "loss": 0.2627,
      "step": 24677
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8613146543502808,
      "learning_rate": 1.1456393277172028e-06,
      "loss": 0.295,
      "step": 24678
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9275587797164917,
      "learning_rate": 1.1451204426362893e-06,
      "loss": 0.2719,
      "step": 24679
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8297603130340576,
      "learning_rate": 1.1446016679524818e-06,
      "loss": 0.3094,
      "step": 24680
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.179015874862671,
      "learning_rate": 1.1440830036722473e-06,
      "loss": 0.3272,
      "step": 24681
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.242985963821411,
      "learning_rate": 1.143564449802056e-06,
      "loss": 0.3303,
      "step": 24682
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.426356077194214,
      "learning_rate": 1.1430460063483706e-06,
      "loss": 0.2933,
      "step": 24683
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9368726015090942,
      "learning_rate": 1.1425276733176528e-06,
      "loss": 0.2631,
      "step": 24684
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8740551471710205,
      "learning_rate": 1.1420094507163692e-06,
      "loss": 0.2912,
      "step": 24685
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.015160322189331,
      "learning_rate": 1.1414913385509785e-06,
      "loss": 0.366,
      "step": 24686
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9037245512008667,
      "learning_rate": 1.1409733368279385e-06,
      "loss": 0.3126,
      "step": 24687
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9915746450424194,
      "learning_rate": 1.140455445553711e-06,
      "loss": 0.2837,
      "step": 24688
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.7654616832733154,
      "learning_rate": 1.1399376647347504e-06,
      "loss": 0.3625,
      "step": 24689
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.110924243927002,
      "learning_rate": 1.139419994377513e-06,
      "loss": 0.3012,
      "step": 24690
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.1275436878204346,
      "learning_rate": 1.1389024344884502e-06,
      "loss": 0.3168,
      "step": 24691
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8481101989746094,
      "learning_rate": 1.138384985074018e-06,
      "loss": 0.3126,
      "step": 24692
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.995320200920105,
      "learning_rate": 1.1378676461406669e-06,
      "loss": 0.3526,
      "step": 24693
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.672919273376465,
      "learning_rate": 1.1373504176948435e-06,
      "loss": 0.2573,
      "step": 24694
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.0443131923675537,
      "learning_rate": 1.1368332997430008e-06,
      "loss": 0.3069,
      "step": 24695
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.177881956100464,
      "learning_rate": 1.136316292291585e-06,
      "loss": 0.293,
      "step": 24696
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7923376560211182,
      "learning_rate": 1.135799395347038e-06,
      "loss": 0.297,
      "step": 24697
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6036523580551147,
      "learning_rate": 1.1352826089158099e-06,
      "loss": 0.2863,
      "step": 24698
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9339905977249146,
      "learning_rate": 1.1347659330043403e-06,
      "loss": 0.2552,
      "step": 24699
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8755724430084229,
      "learning_rate": 1.1342493676190692e-06,
      "loss": 0.2559,
      "step": 24700
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.326991558074951,
      "learning_rate": 1.133732912766441e-06,
      "loss": 0.2656,
      "step": 24701
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.2106218338012695,
      "learning_rate": 1.1332165684528918e-06,
      "loss": 0.2324,
      "step": 24702
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7896696329116821,
      "learning_rate": 1.1327003346848597e-06,
      "loss": 0.29,
      "step": 24703
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.167443037033081,
      "learning_rate": 1.1321842114687797e-06,
      "loss": 0.3006,
      "step": 24704
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1853160858154297,
      "learning_rate": 1.1316681988110888e-06,
      "loss": 0.3978,
      "step": 24705
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0967342853546143,
      "learning_rate": 1.1311522967182198e-06,
      "loss": 0.2436,
      "step": 24706
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.431431293487549,
      "learning_rate": 1.1306365051966006e-06,
      "loss": 0.3707,
      "step": 24707
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1267266273498535,
      "learning_rate": 1.1301208242526684e-06,
      "loss": 0.2972,
      "step": 24708
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9441516399383545,
      "learning_rate": 1.1296052538928482e-06,
      "loss": 0.2422,
      "step": 24709
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.270504951477051,
      "learning_rate": 1.1290897941235668e-06,
      "loss": 0.2965,
      "step": 24710
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9698554277420044,
      "learning_rate": 1.1285744449512548e-06,
      "loss": 0.2855,
      "step": 24711
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1410322189331055,
      "learning_rate": 1.1280592063823349e-06,
      "loss": 0.3125,
      "step": 24712
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9293228387832642,
      "learning_rate": 1.1275440784232293e-06,
      "loss": 0.2762,
      "step": 24713
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5861719846725464,
      "learning_rate": 1.1270290610803636e-06,
      "loss": 0.2955,
      "step": 24714
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.174266815185547,
      "learning_rate": 1.1265141543601565e-06,
      "loss": 0.2912,
      "step": 24715
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.2967233657836914,
      "learning_rate": 1.1259993582690277e-06,
      "loss": 0.3313,
      "step": 24716
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8957631587982178,
      "learning_rate": 1.1254846728133938e-06,
      "loss": 0.3457,
      "step": 24717
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.084500312805176,
      "learning_rate": 1.1249700979996758e-06,
      "loss": 0.2997,
      "step": 24718
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8073914051055908,
      "learning_rate": 1.124455633834287e-06,
      "loss": 0.229,
      "step": 24719
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8270198106765747,
      "learning_rate": 1.1239412803236394e-06,
      "loss": 0.3751,
      "step": 24720
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7550113201141357,
      "learning_rate": 1.1234270374741485e-06,
      "loss": 0.2981,
      "step": 24721
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.128681182861328,
      "learning_rate": 1.1229129052922249e-06,
      "loss": 0.2766,
      "step": 24722
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7030168771743774,
      "learning_rate": 1.1223988837842758e-06,
      "loss": 0.2649,
      "step": 24723
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.926697850227356,
      "learning_rate": 1.1218849729567139e-06,
      "loss": 0.308,
      "step": 24724
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.956040859222412,
      "learning_rate": 1.1213711728159437e-06,
      "loss": 0.1983,
      "step": 24725
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8888496160507202,
      "learning_rate": 1.1208574833683706e-06,
      "loss": 0.2635,
      "step": 24726
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.120135545730591,
      "learning_rate": 1.1203439046204011e-06,
      "loss": 0.2927,
      "step": 24727
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7404416799545288,
      "learning_rate": 1.1198304365784375e-06,
      "loss": 0.2873,
      "step": 24728
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0253467559814453,
      "learning_rate": 1.1193170792488806e-06,
      "loss": 0.2986,
      "step": 24729
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.255810022354126,
      "learning_rate": 1.118803832638129e-06,
      "loss": 0.3612,
      "step": 24730
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1378333568573,
      "learning_rate": 1.1182906967525853e-06,
      "loss": 0.3128,
      "step": 24731
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.126952648162842,
      "learning_rate": 1.1177776715986455e-06,
      "loss": 0.3079,
      "step": 24732
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0709807872772217,
      "learning_rate": 1.1172647571827034e-06,
      "loss": 0.2788,
      "step": 24733
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.355924129486084,
      "learning_rate": 1.116751953511157e-06,
      "loss": 0.3101,
      "step": 24734
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.2394275665283203,
      "learning_rate": 1.1162392605903992e-06,
      "loss": 0.293,
      "step": 24735
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.405428647994995,
      "learning_rate": 1.1157266784268184e-06,
      "loss": 0.3253,
      "step": 24736
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0696444511413574,
      "learning_rate": 1.1152142070268091e-06,
      "loss": 0.3102,
      "step": 24737
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7393295764923096,
      "learning_rate": 1.1147018463967596e-06,
      "loss": 0.2901,
      "step": 24738
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8373489379882812,
      "learning_rate": 1.1141895965430549e-06,
      "loss": 0.2442,
      "step": 24739
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.3147521018981934,
      "learning_rate": 1.1136774574720865e-06,
      "loss": 0.3233,
      "step": 24740
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0476365089416504,
      "learning_rate": 1.113165429190235e-06,
      "loss": 0.2632,
      "step": 24741
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.994583010673523,
      "learning_rate": 1.1126535117038873e-06,
      "loss": 0.3002,
      "step": 24742
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.008496046066284,
      "learning_rate": 1.1121417050194204e-06,
      "loss": 0.295,
      "step": 24743
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8693796396255493,
      "learning_rate": 1.111630009143222e-06,
      "loss": 0.3674,
      "step": 24744
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8818860054016113,
      "learning_rate": 1.1111184240816685e-06,
      "loss": 0.2151,
      "step": 24745
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8182427883148193,
      "learning_rate": 1.1106069498411354e-06,
      "loss": 0.2753,
      "step": 24746
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.950354814529419,
      "learning_rate": 1.1100955864280039e-06,
      "loss": 0.3227,
      "step": 24747
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8832451105117798,
      "learning_rate": 1.1095843338486478e-06,
      "loss": 0.2846,
      "step": 24748
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5511515140533447,
      "learning_rate": 1.1090731921094388e-06,
      "loss": 0.1962,
      "step": 24749
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8818906545639038,
      "learning_rate": 1.108562161216753e-06,
      "loss": 0.3142,
      "step": 24750
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1529123783111572,
      "learning_rate": 1.1080512411769605e-06,
      "loss": 0.2463,
      "step": 24751
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.218872308731079,
      "learning_rate": 1.1075404319964278e-06,
      "loss": 0.3321,
      "step": 24752
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8258885145187378,
      "learning_rate": 1.1070297336815294e-06,
      "loss": 0.3158,
      "step": 24753
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9491081237792969,
      "learning_rate": 1.1065191462386281e-06,
      "loss": 0.3032,
      "step": 24754
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.516383647918701,
      "learning_rate": 1.1060086696740913e-06,
      "loss": 0.2987,
      "step": 24755
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8948372602462769,
      "learning_rate": 1.1054983039942802e-06,
      "loss": 0.2937,
      "step": 24756
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6289825439453125,
      "learning_rate": 1.104988049205562e-06,
      "loss": 0.2158,
      "step": 24757
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.3071208000183105,
      "learning_rate": 1.1044779053142973e-06,
      "loss": 0.3599,
      "step": 24758
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1114141941070557,
      "learning_rate": 1.1039678723268421e-06,
      "loss": 0.3301,
      "step": 24759
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8756402730941772,
      "learning_rate": 1.1034579502495614e-06,
      "loss": 0.2757,
      "step": 24760
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.013206720352173,
      "learning_rate": 1.1029481390888098e-06,
      "loss": 0.3256,
      "step": 24761
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5926958322525024,
      "learning_rate": 1.1024384388509402e-06,
      "loss": 0.2626,
      "step": 24762
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.680079698562622,
      "learning_rate": 1.101928849542313e-06,
      "loss": 0.327,
      "step": 24763
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9928520917892456,
      "learning_rate": 1.1014193711692779e-06,
      "loss": 0.3152,
      "step": 24764
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.062013864517212,
      "learning_rate": 1.100910003738186e-06,
      "loss": 0.2956,
      "step": 24765
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9720128774642944,
      "learning_rate": 1.1004007472553912e-06,
      "loss": 0.3189,
      "step": 24766
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9364995956420898,
      "learning_rate": 1.099891601727241e-06,
      "loss": 0.2914,
      "step": 24767
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1352033615112305,
      "learning_rate": 1.0993825671600822e-06,
      "loss": 0.3682,
      "step": 24768
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0023770332336426,
      "learning_rate": 1.0988736435602598e-06,
      "loss": 0.3277,
      "step": 24769
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1014230251312256,
      "learning_rate": 1.0983648309341222e-06,
      "loss": 0.2428,
      "step": 24770
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.710129976272583,
      "learning_rate": 1.0978561292880118e-06,
      "loss": 0.2579,
      "step": 24771
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.052967071533203,
      "learning_rate": 1.0973475386282684e-06,
      "loss": 0.2965,
      "step": 24772
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6364600658416748,
      "learning_rate": 1.0968390589612365e-06,
      "loss": 0.2878,
      "step": 24773
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7183201313018799,
      "learning_rate": 1.0963306902932535e-06,
      "loss": 0.3256,
      "step": 24774
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.764473557472229,
      "learning_rate": 1.0958224326306554e-06,
      "loss": 0.2279,
      "step": 24775
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.2291033267974854,
      "learning_rate": 1.095314285979784e-06,
      "loss": 0.3158,
      "step": 24776
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6660889387130737,
      "learning_rate": 1.0948062503469715e-06,
      "loss": 0.2953,
      "step": 24777
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1979479789733887,
      "learning_rate": 1.0942983257385498e-06,
      "loss": 0.3371,
      "step": 24778
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.680945873260498,
      "learning_rate": 1.093790512160855e-06,
      "loss": 0.2488,
      "step": 24779
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.431816339492798,
      "learning_rate": 1.0932828096202174e-06,
      "loss": 0.3918,
      "step": 24780
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7426351308822632,
      "learning_rate": 1.0927752181229655e-06,
      "loss": 0.3095,
      "step": 24781
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7795653343200684,
      "learning_rate": 1.0922677376754265e-06,
      "loss": 0.2816,
      "step": 24782
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6541225910186768,
      "learning_rate": 1.0917603682839306e-06,
      "loss": 0.283,
      "step": 24783
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1666338443756104,
      "learning_rate": 1.0912531099548019e-06,
      "loss": 0.3193,
      "step": 24784
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9260742664337158,
      "learning_rate": 1.0907459626943627e-06,
      "loss": 0.2971,
      "step": 24785
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1033923625946045,
      "learning_rate": 1.0902389265089397e-06,
      "loss": 0.2773,
      "step": 24786
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.077817916870117,
      "learning_rate": 1.0897320014048517e-06,
      "loss": 0.2721,
      "step": 24787
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0520007610321045,
      "learning_rate": 1.0892251873884174e-06,
      "loss": 0.3015,
      "step": 24788
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.0999906063079834,
      "learning_rate": 1.0887184844659594e-06,
      "loss": 0.3372,
      "step": 24789
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9377702474594116,
      "learning_rate": 1.0882118926437924e-06,
      "loss": 0.3541,
      "step": 24790
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6878814697265625,
      "learning_rate": 1.0877054119282315e-06,
      "loss": 0.2547,
      "step": 24791
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9888875484466553,
      "learning_rate": 1.087199042325594e-06,
      "loss": 0.3189,
      "step": 24792
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.291469097137451,
      "learning_rate": 1.0866927838421903e-06,
      "loss": 0.3063,
      "step": 24793
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.943298101425171,
      "learning_rate": 1.0861866364843344e-06,
      "loss": 0.2754,
      "step": 24794
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9906262159347534,
      "learning_rate": 1.0856806002583332e-06,
      "loss": 0.35,
      "step": 24795
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6392558813095093,
      "learning_rate": 1.0851746751704994e-06,
      "loss": 0.3279,
      "step": 24796
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.2509851455688477,
      "learning_rate": 1.0846688612271384e-06,
      "loss": 0.3201,
      "step": 24797
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.246077299118042,
      "learning_rate": 1.0841631584345558e-06,
      "loss": 0.3396,
      "step": 24798
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1911118030548096,
      "learning_rate": 1.08365756679906e-06,
      "loss": 0.3116,
      "step": 24799
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.060957908630371,
      "learning_rate": 1.0831520863269506e-06,
      "loss": 0.3423,
      "step": 24800
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.575415015220642,
      "learning_rate": 1.0826467170245303e-06,
      "loss": 0.243,
      "step": 24801
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7887332439422607,
      "learning_rate": 1.0821414588981016e-06,
      "loss": 0.3185,
      "step": 24802
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7517808675765991,
      "learning_rate": 1.081636311953963e-06,
      "loss": 0.3068,
      "step": 24803
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.169639825820923,
      "learning_rate": 1.0811312761984095e-06,
      "loss": 0.3226,
      "step": 24804
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0576696395874023,
      "learning_rate": 1.0806263516377413e-06,
      "loss": 0.3418,
      "step": 24805
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9062052965164185,
      "learning_rate": 1.0801215382782537e-06,
      "loss": 0.2909,
      "step": 24806
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0089735984802246,
      "learning_rate": 1.0796168361262383e-06,
      "loss": 0.3734,
      "step": 24807
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8692305088043213,
      "learning_rate": 1.0791122451879854e-06,
      "loss": 0.3201,
      "step": 24808
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0367753505706787,
      "learning_rate": 1.0786077654697901e-06,
      "loss": 0.3547,
      "step": 24809
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.9316306114196777,
      "learning_rate": 1.0781033969779408e-06,
      "loss": 0.315,
      "step": 24810
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0927553176879883,
      "learning_rate": 1.0775991397187236e-06,
      "loss": 0.3145,
      "step": 24811
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.3401057720184326,
      "learning_rate": 1.0770949936984288e-06,
      "loss": 0.3961,
      "step": 24812
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9062650203704834,
      "learning_rate": 1.0765909589233392e-06,
      "loss": 0.327,
      "step": 24813
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7077683210372925,
      "learning_rate": 1.0760870353997378e-06,
      "loss": 0.2315,
      "step": 24814
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8492833375930786,
      "learning_rate": 1.0755832231339114e-06,
      "loss": 0.3136,
      "step": 24815
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.175854206085205,
      "learning_rate": 1.0750795221321375e-06,
      "loss": 0.2743,
      "step": 24816
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.3063433170318604,
      "learning_rate": 1.0745759324006955e-06,
      "loss": 0.3133,
      "step": 24817
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8983268737792969,
      "learning_rate": 1.074072453945868e-06,
      "loss": 0.2301,
      "step": 24818
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.2729384899139404,
      "learning_rate": 1.073569086773928e-06,
      "loss": 0.3002,
      "step": 24819
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.094003438949585,
      "learning_rate": 1.0730658308911535e-06,
      "loss": 0.3157,
      "step": 24820
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7695484161376953,
      "learning_rate": 1.0725626863038164e-06,
      "loss": 0.2996,
      "step": 24821
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.3685922622680664,
      "learning_rate": 1.0720596530181925e-06,
      "loss": 0.2909,
      "step": 24822
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6067832708358765,
      "learning_rate": 1.0715567310405516e-06,
      "loss": 0.2321,
      "step": 24823
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.209599018096924,
      "learning_rate": 1.0710539203771619e-06,
      "loss": 0.2859,
      "step": 24824
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.5677037239074707,
      "learning_rate": 1.0705512210342972e-06,
      "loss": 0.2996,
      "step": 24825
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.731809377670288,
      "learning_rate": 1.0700486330182214e-06,
      "loss": 0.294,
      "step": 24826
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8060921430587769,
      "learning_rate": 1.0695461563351993e-06,
      "loss": 0.2861,
      "step": 24827
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.12199068069458,
      "learning_rate": 1.0690437909914996e-06,
      "loss": 0.2795,
      "step": 24828
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0667781829833984,
      "learning_rate": 1.0685415369933828e-06,
      "loss": 0.2456,
      "step": 24829
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.7382380962371826,
      "learning_rate": 1.068039394347108e-06,
      "loss": 0.3748,
      "step": 24830
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.2948312759399414,
      "learning_rate": 1.0675373630589415e-06,
      "loss": 0.3115,
      "step": 24831
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.318140745162964,
      "learning_rate": 1.0670354431351393e-06,
      "loss": 0.3056,
      "step": 24832
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.177417516708374,
      "learning_rate": 1.0665336345819588e-06,
      "loss": 0.2058,
      "step": 24833
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.000716209411621,
      "learning_rate": 1.0660319374056548e-06,
      "loss": 0.2964,
      "step": 24834
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8785500526428223,
      "learning_rate": 1.065530351612486e-06,
      "loss": 0.3265,
      "step": 24835
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8464231491088867,
      "learning_rate": 1.0650288772087036e-06,
      "loss": 0.3226,
      "step": 24836
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.145221710205078,
      "learning_rate": 1.0645275142005573e-06,
      "loss": 0.3424,
      "step": 24837
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.795607328414917,
      "learning_rate": 1.0640262625943032e-06,
      "loss": 0.3193,
      "step": 24838
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.133984088897705,
      "learning_rate": 1.0635251223961873e-06,
      "loss": 0.3028,
      "step": 24839
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9595059156417847,
      "learning_rate": 1.063024093612456e-06,
      "loss": 0.2238,
      "step": 24840
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6740199327468872,
      "learning_rate": 1.0625231762493593e-06,
      "loss": 0.3029,
      "step": 24841
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9862343072891235,
      "learning_rate": 1.0620223703131405e-06,
      "loss": 0.313,
      "step": 24842
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.2614450454711914,
      "learning_rate": 1.0615216758100423e-06,
      "loss": 0.4214,
      "step": 24843
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.097466230392456,
      "learning_rate": 1.0610210927463095e-06,
      "loss": 0.432,
      "step": 24844
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.3809173107147217,
      "learning_rate": 1.0605206211281826e-06,
      "loss": 0.3187,
      "step": 24845
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8117917776107788,
      "learning_rate": 1.0600202609619004e-06,
      "loss": 0.2595,
      "step": 24846
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.2301418781280518,
      "learning_rate": 1.0595200122536986e-06,
      "loss": 0.2806,
      "step": 24847
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1631886959075928,
      "learning_rate": 1.0590198750098192e-06,
      "loss": 0.2697,
      "step": 24848
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.4442238807678223,
      "learning_rate": 1.0585198492364946e-06,
      "loss": 0.3422,
      "step": 24849
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0258405208587646,
      "learning_rate": 1.0580199349399567e-06,
      "loss": 0.2954,
      "step": 24850
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.6337976455688477,
      "learning_rate": 1.0575201321264428e-06,
      "loss": 0.3364,
      "step": 24851
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6834867000579834,
      "learning_rate": 1.0570204408021822e-06,
      "loss": 0.3352,
      "step": 24852
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1817357540130615,
      "learning_rate": 1.0565208609734012e-06,
      "loss": 0.3556,
      "step": 24853
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8302935361862183,
      "learning_rate": 1.0560213926463347e-06,
      "loss": 0.295,
      "step": 24854
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.987762451171875,
      "learning_rate": 1.0555220358272055e-06,
      "loss": 0.3341,
      "step": 24855
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5934325456619263,
      "learning_rate": 1.0550227905222387e-06,
      "loss": 0.2641,
      "step": 24856
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6634712219238281,
      "learning_rate": 1.0545236567376626e-06,
      "loss": 0.2826,
      "step": 24857
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.134225368499756,
      "learning_rate": 1.0540246344796978e-06,
      "loss": 0.3325,
      "step": 24858
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8119462728500366,
      "learning_rate": 1.0535257237545649e-06,
      "loss": 0.3105,
      "step": 24859
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0299837589263916,
      "learning_rate": 1.0530269245684832e-06,
      "loss": 0.3066,
      "step": 24860
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.265408754348755,
      "learning_rate": 1.0525282369276745e-06,
      "loss": 0.2793,
      "step": 24861
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.215402126312256,
      "learning_rate": 1.0520296608383552e-06,
      "loss": 0.3461,
      "step": 24862
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8426709175109863,
      "learning_rate": 1.0515311963067376e-06,
      "loss": 0.3056,
      "step": 24863
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8134665489196777,
      "learning_rate": 1.0510328433390427e-06,
      "loss": 0.2698,
      "step": 24864
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7373842000961304,
      "learning_rate": 1.0505346019414786e-06,
      "loss": 0.2942,
      "step": 24865
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8857648372650146,
      "learning_rate": 1.0500364721202582e-06,
      "loss": 0.3025,
      "step": 24866
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.119269371032715,
      "learning_rate": 1.0495384538815933e-06,
      "loss": 0.3464,
      "step": 24867
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1021335124969482,
      "learning_rate": 1.0490405472316922e-06,
      "loss": 0.355,
      "step": 24868
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7495043277740479,
      "learning_rate": 1.04854275217676e-06,
      "loss": 0.3049,
      "step": 24869
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0729281902313232,
      "learning_rate": 1.0480450687230082e-06,
      "loss": 0.3209,
      "step": 24870
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.4367334842681885,
      "learning_rate": 1.0475474968766374e-06,
      "loss": 0.3727,
      "step": 24871
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9930405616760254,
      "learning_rate": 1.0470500366438519e-06,
      "loss": 0.3295,
      "step": 24872
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1757383346557617,
      "learning_rate": 1.046552688030853e-06,
      "loss": 0.3148,
      "step": 24873
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0367634296417236,
      "learning_rate": 1.0460554510438436e-06,
      "loss": 0.3169,
      "step": 24874
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.677585482597351,
      "learning_rate": 1.0455583256890222e-06,
      "loss": 0.3384,
      "step": 24875
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.3031158447265625,
      "learning_rate": 1.0450613119725839e-06,
      "loss": 0.3469,
      "step": 24876
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0832858085632324,
      "learning_rate": 1.044564409900729e-06,
      "loss": 0.3045,
      "step": 24877
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6412127017974854,
      "learning_rate": 1.0440676194796507e-06,
      "loss": 0.3182,
      "step": 24878
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7506093978881836,
      "learning_rate": 1.0435709407155413e-06,
      "loss": 0.2495,
      "step": 24879
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.099698305130005,
      "learning_rate": 1.0430743736145964e-06,
      "loss": 0.308,
      "step": 24880
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.185023069381714,
      "learning_rate": 1.0425779181830054e-06,
      "loss": 0.389,
      "step": 24881
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7105146646499634,
      "learning_rate": 1.0420815744269552e-06,
      "loss": 0.3184,
      "step": 24882
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8703876733779907,
      "learning_rate": 1.041585342352639e-06,
      "loss": 0.3106,
      "step": 24883
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1152260303497314,
      "learning_rate": 1.0410892219662406e-06,
      "loss": 0.3097,
      "step": 24884
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7686935663223267,
      "learning_rate": 1.040593213273945e-06,
      "loss": 0.3063,
      "step": 24885
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9551293849945068,
      "learning_rate": 1.040097316281935e-06,
      "loss": 0.2906,
      "step": 24886
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.4332563877105713,
      "learning_rate": 1.0396015309963958e-06,
      "loss": 0.2837,
      "step": 24887
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.850508451461792,
      "learning_rate": 1.0391058574235081e-06,
      "loss": 0.2966,
      "step": 24888
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.4709455966949463,
      "learning_rate": 1.0386102955694489e-06,
      "loss": 0.3683,
      "step": 24889
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1648528575897217,
      "learning_rate": 1.038114845440401e-06,
      "loss": 0.299,
      "step": 24890
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.003255605697632,
      "learning_rate": 1.0376195070425388e-06,
      "loss": 0.2394,
      "step": 24891
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0127363204956055,
      "learning_rate": 1.037124280382037e-06,
      "loss": 0.3371,
      "step": 24892
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8591549396514893,
      "learning_rate": 1.0366291654650717e-06,
      "loss": 0.3501,
      "step": 24893
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6304905414581299,
      "learning_rate": 1.0361341622978148e-06,
      "loss": 0.2925,
      "step": 24894
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7405924797058105,
      "learning_rate": 1.0356392708864372e-06,
      "loss": 0.3014,
      "step": 24895
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0669679641723633,
      "learning_rate": 1.0351444912371101e-06,
      "loss": 0.2831,
      "step": 24896
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8340998888015747,
      "learning_rate": 1.034649823356002e-06,
      "loss": 0.2879,
      "step": 24897
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9027923345565796,
      "learning_rate": 1.0341552672492795e-06,
      "loss": 0.2753,
      "step": 24898
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.2711803913116455,
      "learning_rate": 1.033660822923106e-06,
      "loss": 0.3633,
      "step": 24899
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8793889284133911,
      "learning_rate": 1.0331664903836514e-06,
      "loss": 0.3354,
      "step": 24900
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5238391160964966,
      "learning_rate": 1.032672269637075e-06,
      "loss": 0.2095,
      "step": 24901
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.084178924560547,
      "learning_rate": 1.0321781606895376e-06,
      "loss": 0.284,
      "step": 24902
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7263739109039307,
      "learning_rate": 1.031684163547203e-06,
      "loss": 0.272,
      "step": 24903
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9819371700286865,
      "learning_rate": 1.0311902782162275e-06,
      "loss": 0.3497,
      "step": 24904
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.174779176712036,
      "learning_rate": 1.0306965047027673e-06,
      "loss": 0.3236,
      "step": 24905
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9162832498550415,
      "learning_rate": 1.0302028430129828e-06,
      "loss": 0.2685,
      "step": 24906
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.087934970855713,
      "learning_rate": 1.029709293153025e-06,
      "loss": 0.3547,
      "step": 24907
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.138028860092163,
      "learning_rate": 1.0292158551290476e-06,
      "loss": 0.3813,
      "step": 24908
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9407356977462769,
      "learning_rate": 1.0287225289472047e-06,
      "loss": 0.3001,
      "step": 24909
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7131706476211548,
      "learning_rate": 1.0282293146136457e-06,
      "loss": 0.356,
      "step": 24910
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.0374205112457275,
      "learning_rate": 1.0277362121345192e-06,
      "loss": 0.2877,
      "step": 24911
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9690511226654053,
      "learning_rate": 1.02724322151597e-06,
      "loss": 0.3602,
      "step": 24912
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.125626802444458,
      "learning_rate": 1.02675034276415e-06,
      "loss": 0.2189,
      "step": 24913
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.279937982559204,
      "learning_rate": 1.026257575885201e-06,
      "loss": 0.356,
      "step": 24914
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.4855570793151855,
      "learning_rate": 1.0257649208852661e-06,
      "loss": 0.3498,
      "step": 24915
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.83339262008667,
      "learning_rate": 1.0252723777704898e-06,
      "loss": 0.279,
      "step": 24916
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9607402086257935,
      "learning_rate": 1.024779946547011e-06,
      "loss": 0.3431,
      "step": 24917
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.317415952682495,
      "learning_rate": 1.0242876272209667e-06,
      "loss": 0.3655,
      "step": 24918
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7928364276885986,
      "learning_rate": 1.0237954197985012e-06,
      "loss": 0.2228,
      "step": 24919
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8672674894332886,
      "learning_rate": 1.023303324285746e-06,
      "loss": 0.2131,
      "step": 24920
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8220369815826416,
      "learning_rate": 1.0228113406888362e-06,
      "loss": 0.3249,
      "step": 24921
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9340945482254028,
      "learning_rate": 1.0223194690139082e-06,
      "loss": 0.3577,
      "step": 24922
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.3526792526245117,
      "learning_rate": 1.0218277092670936e-06,
      "loss": 0.2679,
      "step": 24923
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1296486854553223,
      "learning_rate": 1.021336061454522e-06,
      "loss": 0.2922,
      "step": 24924
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.8415253162384033,
      "learning_rate": 1.020844525582323e-06,
      "loss": 0.2653,
      "step": 24925
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.331260919570923,
      "learning_rate": 1.020353101656626e-06,
      "loss": 0.2843,
      "step": 24926
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.309703826904297,
      "learning_rate": 1.0198617896835583e-06,
      "loss": 0.2438,
      "step": 24927
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.134674310684204,
      "learning_rate": 1.0193705896692419e-06,
      "loss": 0.3442,
      "step": 24928
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9038923978805542,
      "learning_rate": 1.0188795016198049e-06,
      "loss": 0.311,
      "step": 24929
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0459625720977783,
      "learning_rate": 1.0183885255413685e-06,
      "loss": 0.2881,
      "step": 24930
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6894878149032593,
      "learning_rate": 1.0178976614400515e-06,
      "loss": 0.273,
      "step": 24931
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1964569091796875,
      "learning_rate": 1.0174069093219773e-06,
      "loss": 0.2743,
      "step": 24932
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9558913707733154,
      "learning_rate": 1.0169162691932633e-06,
      "loss": 0.3325,
      "step": 24933
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.267333984375,
      "learning_rate": 1.016425741060023e-06,
      "loss": 0.3142,
      "step": 24934
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7794564962387085,
      "learning_rate": 1.0159353249283777e-06,
      "loss": 0.3175,
      "step": 24935
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.3257977962493896,
      "learning_rate": 1.0154450208044387e-06,
      "loss": 0.3613,
      "step": 24936
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9640487432479858,
      "learning_rate": 1.0149548286943178e-06,
      "loss": 0.3763,
      "step": 24937
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1886813640594482,
      "learning_rate": 1.014464748604127e-06,
      "loss": 0.3798,
      "step": 24938
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1197566986083984,
      "learning_rate": 1.013974780539978e-06,
      "loss": 0.2799,
      "step": 24939
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7717199325561523,
      "learning_rate": 1.013484924507978e-06,
      "loss": 0.2548,
      "step": 24940
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.088062286376953,
      "learning_rate": 1.0129951805142324e-06,
      "loss": 0.3293,
      "step": 24941
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1604421138763428,
      "learning_rate": 1.0125055485648516e-06,
      "loss": 0.3655,
      "step": 24942
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.92119300365448,
      "learning_rate": 1.0120160286659364e-06,
      "loss": 0.3118,
      "step": 24943
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0285699367523193,
      "learning_rate": 1.0115266208235896e-06,
      "loss": 0.2484,
      "step": 24944
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1268420219421387,
      "learning_rate": 1.0110373250439165e-06,
      "loss": 0.3245,
      "step": 24945
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9566117525100708,
      "learning_rate": 1.0105481413330143e-06,
      "loss": 0.3003,
      "step": 24946
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5162179470062256,
      "learning_rate": 1.0100590696969804e-06,
      "loss": 0.3134,
      "step": 24947
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.3867263793945312,
      "learning_rate": 1.0095701101419164e-06,
      "loss": 0.351,
      "step": 24948
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9490602016448975,
      "learning_rate": 1.0090812626739166e-06,
      "loss": 0.2858,
      "step": 24949
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.822414755821228,
      "learning_rate": 1.008592527299076e-06,
      "loss": 0.2514,
      "step": 24950
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.5787785053253174,
      "learning_rate": 1.008103904023484e-06,
      "loss": 0.4058,
      "step": 24951
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7045111656188965,
      "learning_rate": 1.0076153928532383e-06,
      "loss": 0.2347,
      "step": 24952
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.095486879348755,
      "learning_rate": 1.007126993794426e-06,
      "loss": 0.35,
      "step": 24953
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0017340183258057,
      "learning_rate": 1.0066387068531359e-06,
      "loss": 0.2938,
      "step": 24954
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0783212184906006,
      "learning_rate": 1.006150532035457e-06,
      "loss": 0.3338,
      "step": 24955
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8563874959945679,
      "learning_rate": 1.0056624693474759e-06,
      "loss": 0.3643,
      "step": 24956
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.020756244659424,
      "learning_rate": 1.0051745187952754e-06,
      "loss": 0.3619,
      "step": 24957
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.3096847534179688,
      "learning_rate": 1.004686680384941e-06,
      "loss": 0.3177,
      "step": 24958
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.6148905754089355,
      "learning_rate": 1.004198954122555e-06,
      "loss": 0.2402,
      "step": 24959
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9874927997589111,
      "learning_rate": 1.003711340014195e-06,
      "loss": 0.3643,
      "step": 24960
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1956491470336914,
      "learning_rate": 1.0032238380659443e-06,
      "loss": 0.3216,
      "step": 24961
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.2591583728790283,
      "learning_rate": 1.0027364482838787e-06,
      "loss": 0.2706,
      "step": 24962
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9468252658843994,
      "learning_rate": 1.0022491706740745e-06,
      "loss": 0.2758,
      "step": 24963
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6799707412719727,
      "learning_rate": 1.001762005242607e-06,
      "loss": 0.2591,
      "step": 24964
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.578774929046631,
      "learning_rate": 1.00127495199555e-06,
      "loss": 0.2778,
      "step": 24965
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0825610160827637,
      "learning_rate": 1.0007880109389745e-06,
      "loss": 0.2787,
      "step": 24966
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7234539985656738,
      "learning_rate": 1.0003011820789543e-06,
      "loss": 0.2557,
      "step": 24967
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8769266605377197,
      "learning_rate": 9.998144654215569e-07,
      "loss": 0.2183,
      "step": 24968
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0182437896728516,
      "learning_rate": 9.993278609728484e-07,
      "loss": 0.3287,
      "step": 24969
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.2175095081329346,
      "learning_rate": 9.988413687388999e-07,
      "loss": 0.3162,
      "step": 24970
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9012370109558105,
      "learning_rate": 9.98354988725775e-07,
      "loss": 0.2725,
      "step": 24971
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.447709083557129,
      "learning_rate": 9.978687209395366e-07,
      "loss": 0.3429,
      "step": 24972
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8277530670166016,
      "learning_rate": 9.973825653862457e-07,
      "loss": 0.2912,
      "step": 24973
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.237004041671753,
      "learning_rate": 9.968965220719673e-07,
      "loss": 0.2935,
      "step": 24974
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7991657257080078,
      "learning_rate": 9.9641059100276e-07,
      "loss": 0.2856,
      "step": 24975
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9874255657196045,
      "learning_rate": 9.959247721846787e-07,
      "loss": 0.2708,
      "step": 24976
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.770527720451355,
      "learning_rate": 9.954390656237845e-07,
      "loss": 0.253,
      "step": 24977
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.138066053390503,
      "learning_rate": 9.94953471326131e-07,
      "loss": 0.2394,
      "step": 24978
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5130647420883179,
      "learning_rate": 9.944679892977715e-07,
      "loss": 0.3328,
      "step": 24979
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7014925479888916,
      "learning_rate": 9.93982619544761e-07,
      "loss": 0.2549,
      "step": 24980
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.9605448246002197,
      "learning_rate": 9.934973620731492e-07,
      "loss": 0.3104,
      "step": 24981
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.665151357650757,
      "learning_rate": 9.930122168889856e-07,
      "loss": 0.3506,
      "step": 24982
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.248033285140991,
      "learning_rate": 9.925271839983197e-07,
      "loss": 0.2768,
      "step": 24983
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5740487575531006,
      "learning_rate": 9.920422634071991e-07,
      "loss": 0.2823,
      "step": 24984
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5595699548721313,
      "learning_rate": 9.91557455121669e-07,
      "loss": 0.2454,
      "step": 24985
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0442521572113037,
      "learning_rate": 9.910727591477699e-07,
      "loss": 0.2141,
      "step": 24986
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.564744234085083,
      "learning_rate": 9.905881754915513e-07,
      "loss": 0.3794,
      "step": 24987
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7734590768814087,
      "learning_rate": 9.901037041590511e-07,
      "loss": 0.267,
      "step": 24988
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6262681484222412,
      "learning_rate": 9.896193451563074e-07,
      "loss": 0.2676,
      "step": 24989
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.2776410579681396,
      "learning_rate": 9.891350984893644e-07,
      "loss": 0.2342,
      "step": 24990
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.0042994022369385,
      "learning_rate": 9.886509641642549e-07,
      "loss": 0.2886,
      "step": 24991
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8169331550598145,
      "learning_rate": 9.881669421870144e-07,
      "loss": 0.2482,
      "step": 24992
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7723525762557983,
      "learning_rate": 9.876830325636822e-07,
      "loss": 0.3651,
      "step": 24993
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.1994802951812744,
      "learning_rate": 9.871992353002869e-07,
      "loss": 0.3182,
      "step": 24994
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0859854221343994,
      "learning_rate": 9.867155504028602e-07,
      "loss": 0.3892,
      "step": 24995
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.932011365890503,
      "learning_rate": 9.862319778774353e-07,
      "loss": 0.2442,
      "step": 24996
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.402003526687622,
      "learning_rate": 9.857485177300396e-07,
      "loss": 0.3625,
      "step": 24997
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7837330102920532,
      "learning_rate": 9.852651699667014e-07,
      "loss": 0.312,
      "step": 24998
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.6818838119506836,
      "learning_rate": 9.84781934593443e-07,
      "loss": 0.4256,
      "step": 24999
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8138774633407593,
      "learning_rate": 9.842988116162945e-07,
      "loss": 0.3649,
      "step": 25000
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7528316974639893,
      "learning_rate": 9.838158010412768e-07,
      "loss": 0.2886,
      "step": 25001
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8015148639678955,
      "learning_rate": 9.833329028744087e-07,
      "loss": 0.3266,
      "step": 25002
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8390644788742065,
      "learning_rate": 9.828501171217165e-07,
      "loss": 0.2597,
      "step": 25003
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.28916597366333,
      "learning_rate": 9.82367443789215e-07,
      "loss": 0.3526,
      "step": 25004
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.5316989421844482,
      "learning_rate": 9.81884882882923e-07,
      "loss": 0.3571,
      "step": 25005
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.263859748840332,
      "learning_rate": 9.81402434408858e-07,
      "loss": 0.3136,
      "step": 25006
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.034405469894409,
      "learning_rate": 9.809200983730338e-07,
      "loss": 0.3908,
      "step": 25007
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.2630083560943604,
      "learning_rate": 9.804378747814626e-07,
      "loss": 0.3043,
      "step": 25008
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.097886800765991,
      "learning_rate": 9.79955763640159e-07,
      "loss": 0.3841,
      "step": 25009
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1175742149353027,
      "learning_rate": 9.794737649551322e-07,
      "loss": 0.2756,
      "step": 25010
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7834726572036743,
      "learning_rate": 9.789918787323926e-07,
      "loss": 0.3288,
      "step": 25011
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7427409887313843,
      "learning_rate": 9.785101049779445e-07,
      "loss": 0.2988,
      "step": 25012
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9218369722366333,
      "learning_rate": 9.780284436977994e-07,
      "loss": 0.3149,
      "step": 25013
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.004459857940674,
      "learning_rate": 9.775468948979604e-07,
      "loss": 0.3009,
      "step": 25014
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.3221113681793213,
      "learning_rate": 9.770654585844275e-07,
      "loss": 0.3062,
      "step": 25015
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6248853206634521,
      "learning_rate": 9.765841347632088e-07,
      "loss": 0.2892,
      "step": 25016
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.5555672645568848,
      "learning_rate": 9.76102923440303e-07,
      "loss": 0.4403,
      "step": 25017
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.107604503631592,
      "learning_rate": 9.756218246217065e-07,
      "loss": 0.3663,
      "step": 25018
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.93174409866333,
      "learning_rate": 9.75140838313422e-07,
      "loss": 0.2275,
      "step": 25019
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.053598165512085,
      "learning_rate": 9.746599645214439e-07,
      "loss": 0.2356,
      "step": 25020
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.99397611618042,
      "learning_rate": 9.741792032517661e-07,
      "loss": 0.3479,
      "step": 25021
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.4259867668151855,
      "learning_rate": 9.73698554510385e-07,
      "loss": 0.2772,
      "step": 25022
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8455182313919067,
      "learning_rate": 9.732180183032924e-07,
      "loss": 0.2903,
      "step": 25023
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9211266040802002,
      "learning_rate": 9.727375946364792e-07,
      "loss": 0.2837,
      "step": 25024
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6140018701553345,
      "learning_rate": 9.722572835159328e-07,
      "loss": 0.266,
      "step": 25025
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.5697760581970215,
      "learning_rate": 9.717770849476448e-07,
      "loss": 0.3328,
      "step": 25026
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.185943126678467,
      "learning_rate": 9.712969989375999e-07,
      "loss": 0.2573,
      "step": 25027
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.140522003173828,
      "learning_rate": 9.708170254917826e-07,
      "loss": 0.297,
      "step": 25028
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.118114948272705,
      "learning_rate": 9.703371646161809e-07,
      "loss": 0.361,
      "step": 25029
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7568928003311157,
      "learning_rate": 9.69857416316775e-07,
      "loss": 0.3365,
      "step": 25030
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.516835927963257,
      "learning_rate": 9.693777805995431e-07,
      "loss": 0.319,
      "step": 25031
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7252558469772339,
      "learning_rate": 9.688982574704708e-07,
      "loss": 0.2079,
      "step": 25032
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.945489525794983,
      "learning_rate": 9.684188469355327e-07,
      "loss": 0.2581,
      "step": 25033
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0647802352905273,
      "learning_rate": 9.679395490007059e-07,
      "loss": 0.3049,
      "step": 25034
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.2675082683563232,
      "learning_rate": 9.674603636719681e-07,
      "loss": 0.3296,
      "step": 25035
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.180577516555786,
      "learning_rate": 9.669812909552923e-07,
      "loss": 0.2741,
      "step": 25036
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0846009254455566,
      "learning_rate": 9.665023308566512e-07,
      "loss": 0.3942,
      "step": 25037
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6855735778808594,
      "learning_rate": 9.660234833820137e-07,
      "loss": 0.3309,
      "step": 25038
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.58037269115448,
      "learning_rate": 9.655447485373548e-07,
      "loss": 0.2495,
      "step": 25039
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6201142072677612,
      "learning_rate": 9.6506612632864e-07,
      "loss": 0.2389,
      "step": 25040
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.202826738357544,
      "learning_rate": 9.64587616761835e-07,
      "loss": 0.3245,
      "step": 25041
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1338789463043213,
      "learning_rate": 9.641092198429103e-07,
      "loss": 0.2981,
      "step": 25042
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.466726303100586,
      "learning_rate": 9.636309355778262e-07,
      "loss": 0.3026,
      "step": 25043
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1206789016723633,
      "learning_rate": 9.631527639725457e-07,
      "loss": 0.3339,
      "step": 25044
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.57646107673645,
      "learning_rate": 9.626747050330332e-07,
      "loss": 0.3603,
      "step": 25045
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9099406003952026,
      "learning_rate": 9.62196758765248e-07,
      "loss": 0.2777,
      "step": 25046
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.536142587661743,
      "learning_rate": 9.617189251751458e-07,
      "loss": 0.3233,
      "step": 25047
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8168985843658447,
      "learning_rate": 9.612412042686869e-07,
      "loss": 0.3177,
      "step": 25048
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8011671304702759,
      "learning_rate": 9.60763596051828e-07,
      "loss": 0.2837,
      "step": 25049
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.582360029220581,
      "learning_rate": 9.602861005305208e-07,
      "loss": 0.3919,
      "step": 25050
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1562609672546387,
      "learning_rate": 9.598087177107185e-07,
      "loss": 0.2802,
      "step": 25051
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.397249698638916,
      "learning_rate": 9.593314475983762e-07,
      "loss": 0.3358,
      "step": 25052
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.139284372329712,
      "learning_rate": 9.588542901994414e-07,
      "loss": 0.2806,
      "step": 25053
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.4033455848693848,
      "learning_rate": 9.583772455198626e-07,
      "loss": 0.3263,
      "step": 25054
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.909469723701477,
      "learning_rate": 9.579003135655895e-07,
      "loss": 0.3474,
      "step": 25055
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.020500898361206,
      "learning_rate": 9.574234943425675e-07,
      "loss": 0.2883,
      "step": 25056
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.590329170227051,
      "learning_rate": 9.569467878567384e-07,
      "loss": 0.3655,
      "step": 25057
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6144715547561646,
      "learning_rate": 9.564701941140498e-07,
      "loss": 0.2596,
      "step": 25058
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.11838698387146,
      "learning_rate": 9.559937131204422e-07,
      "loss": 0.3567,
      "step": 25059
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7316360473632812,
      "learning_rate": 9.555173448818532e-07,
      "loss": 0.3165,
      "step": 25060
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1164402961730957,
      "learning_rate": 9.550410894042262e-07,
      "loss": 0.3695,
      "step": 25061
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9840096235275269,
      "learning_rate": 9.54564946693497e-07,
      "loss": 0.2343,
      "step": 25062
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.126462936401367,
      "learning_rate": 9.540889167556012e-07,
      "loss": 0.3394,
      "step": 25063
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9814192056655884,
      "learning_rate": 9.536129995964727e-07,
      "loss": 0.2425,
      "step": 25064
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1913132667541504,
      "learning_rate": 9.531371952220481e-07,
      "loss": 0.3036,
      "step": 25065
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.02889347076416,
      "learning_rate": 9.526615036382569e-07,
      "loss": 0.3177,
      "step": 25066
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.18353009223938,
      "learning_rate": 9.521859248510301e-07,
      "loss": 0.3056,
      "step": 25067
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.112217664718628,
      "learning_rate": 9.517104588662985e-07,
      "loss": 0.3522,
      "step": 25068
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9195475578308105,
      "learning_rate": 9.512351056899893e-07,
      "loss": 0.2612,
      "step": 25069
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.538146734237671,
      "learning_rate": 9.507598653280259e-07,
      "loss": 0.3284,
      "step": 25070
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8656806945800781,
      "learning_rate": 9.502847377863378e-07,
      "loss": 0.2521,
      "step": 25071
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.071876049041748,
      "learning_rate": 9.498097230708458e-07,
      "loss": 0.2198,
      "step": 25072
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.682478427886963,
      "learning_rate": 9.49334821187472e-07,
      "loss": 0.2941,
      "step": 25073
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.009615659713745,
      "learning_rate": 9.488600321421393e-07,
      "loss": 0.3517,
      "step": 25074
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1521596908569336,
      "learning_rate": 9.483853559407663e-07,
      "loss": 0.331,
      "step": 25075
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8758903741836548,
      "learning_rate": 9.479107925892706e-07,
      "loss": 0.2254,
      "step": 25076
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.948777198791504,
      "learning_rate": 9.474363420935662e-07,
      "loss": 0.2995,
      "step": 25077
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9025670289993286,
      "learning_rate": 9.469620044595729e-07,
      "loss": 0.3156,
      "step": 25078
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9837394952774048,
      "learning_rate": 9.464877796932015e-07,
      "loss": 0.3064,
      "step": 25079
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7525945901870728,
      "learning_rate": 9.460136678003639e-07,
      "loss": 0.288,
      "step": 25080
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0990452766418457,
      "learning_rate": 9.455396687869745e-07,
      "loss": 0.3014,
      "step": 25081
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.965071201324463,
      "learning_rate": 9.450657826589394e-07,
      "loss": 0.3199,
      "step": 25082
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0864105224609375,
      "learning_rate": 9.445920094221672e-07,
      "loss": 0.2942,
      "step": 25083
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6404837369918823,
      "learning_rate": 9.441183490825667e-07,
      "loss": 0.2737,
      "step": 25084
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8413881063461304,
      "learning_rate": 9.436448016460409e-07,
      "loss": 0.2984,
      "step": 25085
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.0198683738708496,
      "learning_rate": 9.431713671184927e-07,
      "loss": 0.2996,
      "step": 25086
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1473703384399414,
      "learning_rate": 9.426980455058299e-07,
      "loss": 0.3157,
      "step": 25087
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1228654384613037,
      "learning_rate": 9.422248368139486e-07,
      "loss": 0.355,
      "step": 25088
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.094998836517334,
      "learning_rate": 9.417517410487509e-07,
      "loss": 0.282,
      "step": 25089
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7544723749160767,
      "learning_rate": 9.41278758216131e-07,
      "loss": 0.2949,
      "step": 25090
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.857932448387146,
      "learning_rate": 9.408058883219917e-07,
      "loss": 0.2911,
      "step": 25091
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1204724311828613,
      "learning_rate": 9.403331313722253e-07,
      "loss": 0.3261,
      "step": 25092
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0828540325164795,
      "learning_rate": 9.398604873727246e-07,
      "loss": 0.4202,
      "step": 25093
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.2079520225524902,
      "learning_rate": 9.393879563293862e-07,
      "loss": 0.2965,
      "step": 25094
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1883480548858643,
      "learning_rate": 9.389155382480975e-07,
      "loss": 0.4145,
      "step": 25095
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0818564891815186,
      "learning_rate": 9.384432331347493e-07,
      "loss": 0.2824,
      "step": 25096
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.4730913639068604,
      "learning_rate": 9.379710409952314e-07,
      "loss": 0.3941,
      "step": 25097
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.360530138015747,
      "learning_rate": 9.374989618354303e-07,
      "loss": 0.3819,
      "step": 25098
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.977190375328064,
      "learning_rate": 9.370269956612288e-07,
      "loss": 0.3212,
      "step": 25099
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.939412236213684,
      "learning_rate": 9.365551424785158e-07,
      "loss": 0.2776,
      "step": 25100
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.5999159812927246,
      "learning_rate": 9.360834022931709e-07,
      "loss": 0.3176,
      "step": 25101
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.939839482307434,
      "learning_rate": 9.356117751110772e-07,
      "loss": 0.3065,
      "step": 25102
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7652987241744995,
      "learning_rate": 9.351402609381111e-07,
      "loss": 0.263,
      "step": 25103
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.3875956535339355,
      "learning_rate": 9.346688597801556e-07,
      "loss": 0.3203,
      "step": 25104
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0174472332000732,
      "learning_rate": 9.341975716430873e-07,
      "loss": 0.2763,
      "step": 25105
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.146955966949463,
      "learning_rate": 9.337263965327769e-07,
      "loss": 0.2772,
      "step": 25106
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.2339913845062256,
      "learning_rate": 9.332553344551054e-07,
      "loss": 0.3734,
      "step": 25107
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.632724642753601,
      "learning_rate": 9.327843854159424e-07,
      "loss": 0.2783,
      "step": 25108
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9082646369934082,
      "learning_rate": 9.323135494211577e-07,
      "loss": 0.322,
      "step": 25109
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.915174961090088,
      "learning_rate": 9.318428264766255e-07,
      "loss": 0.3432,
      "step": 25110
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8929414749145508,
      "learning_rate": 9.313722165882122e-07,
      "loss": 0.3301,
      "step": 25111
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0123631954193115,
      "learning_rate": 9.309017197617842e-07,
      "loss": 0.3433,
      "step": 25112
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8892775774002075,
      "learning_rate": 9.304313360032091e-07,
      "loss": 0.3634,
      "step": 25113
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.3551831245422363,
      "learning_rate": 9.299610653183511e-07,
      "loss": 0.3521,
      "step": 25114
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9155315160751343,
      "learning_rate": 9.294909077130732e-07,
      "loss": 0.3067,
      "step": 25115
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0092551708221436,
      "learning_rate": 9.29020863193234e-07,
      "loss": 0.2627,
      "step": 25116
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.4471070766448975,
      "learning_rate": 9.28550931764699e-07,
      "loss": 0.3688,
      "step": 25117
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.981798768043518,
      "learning_rate": 9.280811134333245e-07,
      "loss": 0.299,
      "step": 25118
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7163360118865967,
      "learning_rate": 9.276114082049647e-07,
      "loss": 0.2786,
      "step": 25119
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6546919345855713,
      "learning_rate": 9.271418160854817e-07,
      "loss": 0.2635,
      "step": 25120
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.380782127380371,
      "learning_rate": 9.266723370807273e-07,
      "loss": 0.2898,
      "step": 25121
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.363218069076538,
      "learning_rate": 9.262029711965515e-07,
      "loss": 0.3118,
      "step": 25122
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0513737201690674,
      "learning_rate": 9.257337184388116e-07,
      "loss": 0.2872,
      "step": 25123
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.786436676979065,
      "learning_rate": 9.252645788133552e-07,
      "loss": 0.226,
      "step": 25124
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.228090763092041,
      "learning_rate": 9.247955523260288e-07,
      "loss": 0.2852,
      "step": 25125
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0750465393066406,
      "learning_rate": 9.243266389826855e-07,
      "loss": 0.2318,
      "step": 25126
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.224773406982422,
      "learning_rate": 9.238578387891673e-07,
      "loss": 0.2664,
      "step": 25127
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8806641101837158,
      "learning_rate": 9.233891517513205e-07,
      "loss": 0.3282,
      "step": 25128
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.964802861213684,
      "learning_rate": 9.229205778749861e-07,
      "loss": 0.2883,
      "step": 25129
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.385399103164673,
      "learning_rate": 9.224521171660095e-07,
      "loss": 0.3687,
      "step": 25130
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8733159303665161,
      "learning_rate": 9.219837696302303e-07,
      "loss": 0.2782,
      "step": 25131
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0368926525115967,
      "learning_rate": 9.215155352734839e-07,
      "loss": 0.3236,
      "step": 25132
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0417141914367676,
      "learning_rate": 9.210474141016135e-07,
      "loss": 0.3225,
      "step": 25133
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.3410823345184326,
      "learning_rate": 9.205794061204521e-07,
      "loss": 0.4011,
      "step": 25134
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.892311453819275,
      "learning_rate": 9.201115113358328e-07,
      "loss": 0.2625,
      "step": 25135
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7873468399047852,
      "learning_rate": 9.196437297535943e-07,
      "loss": 0.3021,
      "step": 25136
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.856884479522705,
      "learning_rate": 9.191760613795653e-07,
      "loss": 0.3624,
      "step": 25137
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9033924341201782,
      "learning_rate": 9.187085062195744e-07,
      "loss": 0.3972,
      "step": 25138
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.898054599761963,
      "learning_rate": 9.182410642794548e-07,
      "loss": 0.2431,
      "step": 25139
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7205063104629517,
      "learning_rate": 9.17773735565033e-07,
      "loss": 0.2517,
      "step": 25140
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6793925762176514,
      "learning_rate": 9.173065200821351e-07,
      "loss": 0.3483,
      "step": 25141
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7290507555007935,
      "learning_rate": 9.168394178365836e-07,
      "loss": 0.2708,
      "step": 25142
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1728100776672363,
      "learning_rate": 9.163724288342058e-07,
      "loss": 0.2956,
      "step": 25143
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1975741386413574,
      "learning_rate": 9.159055530808225e-07,
      "loss": 0.3409,
      "step": 25144
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8067700862884521,
      "learning_rate": 9.154387905822537e-07,
      "loss": 0.2787,
      "step": 25145
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.901528239250183,
      "learning_rate": 9.149721413443191e-07,
      "loss": 0.2461,
      "step": 25146
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.913170337677002,
      "learning_rate": 9.145056053728385e-07,
      "loss": 0.2919,
      "step": 25147
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.076364040374756,
      "learning_rate": 9.140391826736239e-07,
      "loss": 0.3384,
      "step": 25148
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1006507873535156,
      "learning_rate": 9.135728732524951e-07,
      "loss": 0.313,
      "step": 25149
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9676109552383423,
      "learning_rate": 9.131066771152641e-07,
      "loss": 0.3198,
      "step": 25150
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9372838735580444,
      "learning_rate": 9.126405942677408e-07,
      "loss": 0.2874,
      "step": 25151
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.347654342651367,
      "learning_rate": 9.121746247157392e-07,
      "loss": 0.3261,
      "step": 25152
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.032585859298706,
      "learning_rate": 9.117087684650682e-07,
      "loss": 0.2831,
      "step": 25153
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9498099088668823,
      "learning_rate": 9.112430255215354e-07,
      "loss": 0.2847,
      "step": 25154
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1285629272460938,
      "learning_rate": 9.107773958909449e-07,
      "loss": 0.2962,
      "step": 25155
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.051703453063965,
      "learning_rate": 9.103118795791055e-07,
      "loss": 0.3097,
      "step": 25156
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.135770797729492,
      "learning_rate": 9.098464765918202e-07,
      "loss": 0.298,
      "step": 25157
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.028721809387207,
      "learning_rate": 9.09381186934889e-07,
      "loss": 0.3357,
      "step": 25158
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8523435592651367,
      "learning_rate": 9.08916010614116e-07,
      "loss": 0.3268,
      "step": 25159
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.954392433166504,
      "learning_rate": 9.084509476352998e-07,
      "loss": 0.3147,
      "step": 25160
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1222341060638428,
      "learning_rate": 9.079859980042361e-07,
      "loss": 0.2953,
      "step": 25161
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1203701496124268,
      "learning_rate": 9.075211617267243e-07,
      "loss": 0.2797,
      "step": 25162
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9555611610412598,
      "learning_rate": 9.070564388085601e-07,
      "loss": 0.2833,
      "step": 25163
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.837524652481079,
      "learning_rate": 9.065918292555342e-07,
      "loss": 0.3048,
      "step": 25164
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.463877558708191,
      "learning_rate": 9.061273330734421e-07,
      "loss": 0.2614,
      "step": 25165
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.170414686203003,
      "learning_rate": 9.056629502680747e-07,
      "loss": 0.2977,
      "step": 25166
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1622138023376465,
      "learning_rate": 9.051986808452206e-07,
      "loss": 0.3776,
      "step": 25167
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8495980501174927,
      "learning_rate": 9.047345248106665e-07,
      "loss": 0.2794,
      "step": 25168
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.719504952430725,
      "learning_rate": 9.04270482170202e-07,
      "loss": 0.2673,
      "step": 25169
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.062593460083008,
      "learning_rate": 9.038065529296114e-07,
      "loss": 0.341,
      "step": 25170
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.408677816390991,
      "learning_rate": 9.033427370946779e-07,
      "loss": 0.2536,
      "step": 25171
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.2598118782043457,
      "learning_rate": 9.028790346711858e-07,
      "loss": 0.2431,
      "step": 25172
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7880147695541382,
      "learning_rate": 9.024154456649148e-07,
      "loss": 0.262,
      "step": 25173
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.315044403076172,
      "learning_rate": 9.019519700816437e-07,
      "loss": 0.2427,
      "step": 25174
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9487364292144775,
      "learning_rate": 9.014886079271546e-07,
      "loss": 0.2761,
      "step": 25175
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.043363571166992,
      "learning_rate": 9.010253592072215e-07,
      "loss": 0.3342,
      "step": 25176
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.861926555633545,
      "learning_rate": 9.005622239276179e-07,
      "loss": 0.2719,
      "step": 25177
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6202672719955444,
      "learning_rate": 9.000992020941235e-07,
      "loss": 0.3503,
      "step": 25178
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.1186487674713135,
      "learning_rate": 8.996362937125069e-07,
      "loss": 0.35,
      "step": 25179
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.865794062614441,
      "learning_rate": 8.991734987885414e-07,
      "loss": 0.3681,
      "step": 25180
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4854674339294434,
      "learning_rate": 8.987108173279935e-07,
      "loss": 0.2538,
      "step": 25181
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.4895741939544678,
      "learning_rate": 8.982482493366351e-07,
      "loss": 0.3506,
      "step": 25182
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0290634632110596,
      "learning_rate": 8.977857948202329e-07,
      "loss": 0.3073,
      "step": 25183
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.301618814468384,
      "learning_rate": 8.973234537845487e-07,
      "loss": 0.2275,
      "step": 25184
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.853481650352478,
      "learning_rate": 8.968612262353516e-07,
      "loss": 0.2744,
      "step": 25185
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.9838882684707642,
      "learning_rate": 8.963991121784032e-07,
      "loss": 0.4467,
      "step": 25186
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.2256925106048584,
      "learning_rate": 8.959371116194615e-07,
      "loss": 0.2911,
      "step": 25187
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.1356256008148193,
      "learning_rate": 8.954752245642906e-07,
      "loss": 0.279,
      "step": 25188
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0231120586395264,
      "learning_rate": 8.950134510186482e-07,
      "loss": 0.3742,
      "step": 25189
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7923732995986938,
      "learning_rate": 8.945517909882884e-07,
      "loss": 0.287,
      "step": 25190
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.076695203781128,
      "learning_rate": 8.9409024447897e-07,
      "loss": 0.2964,
      "step": 25191
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3781347274780273,
      "learning_rate": 8.936288114964475e-07,
      "loss": 0.3326,
      "step": 25192
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.787188172340393,
      "learning_rate": 8.931674920464717e-07,
      "loss": 0.2878,
      "step": 25193
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0179972648620605,
      "learning_rate": 8.927062861347935e-07,
      "loss": 0.3534,
      "step": 25194
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.1862590312957764,
      "learning_rate": 8.922451937671661e-07,
      "loss": 0.2879,
      "step": 25195
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.764944076538086,
      "learning_rate": 8.917842149493361e-07,
      "loss": 0.2574,
      "step": 25196
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7056140899658203,
      "learning_rate": 8.913233496870488e-07,
      "loss": 0.2857,
      "step": 25197
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9545797109603882,
      "learning_rate": 8.908625979860541e-07,
      "loss": 0.3195,
      "step": 25198
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.539994478225708,
      "learning_rate": 8.904019598520952e-07,
      "loss": 0.3292,
      "step": 25199
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.006575107574463,
      "learning_rate": 8.899414352909119e-07,
      "loss": 0.3043,
      "step": 25200
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.6777825355529785,
      "learning_rate": 8.894810243082497e-07,
      "loss": 0.3196,
      "step": 25201
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8816659450531006,
      "learning_rate": 8.890207269098472e-07,
      "loss": 0.2966,
      "step": 25202
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.5212862491607666,
      "learning_rate": 8.885605431014411e-07,
      "loss": 0.2805,
      "step": 25203
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6988835334777832,
      "learning_rate": 8.881004728887721e-07,
      "loss": 0.2527,
      "step": 25204
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.908004879951477,
      "learning_rate": 8.876405162775747e-07,
      "loss": 0.3578,
      "step": 25205
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.5176589488983154,
      "learning_rate": 8.871806732735844e-07,
      "loss": 0.3885,
      "step": 25206
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0134966373443604,
      "learning_rate": 8.867209438825297e-07,
      "loss": 0.3221,
      "step": 25207
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3956196308135986,
      "learning_rate": 8.862613281101473e-07,
      "loss": 0.3865,
      "step": 25208
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.814550518989563,
      "learning_rate": 8.85801825962167e-07,
      "loss": 0.2896,
      "step": 25209
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9846431016921997,
      "learning_rate": 8.853424374443132e-07,
      "loss": 0.3315,
      "step": 25210
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9418375492095947,
      "learning_rate": 8.848831625623189e-07,
      "loss": 0.2726,
      "step": 25211
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3744704723358154,
      "learning_rate": 8.844240013219063e-07,
      "loss": 0.3349,
      "step": 25212
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.748355746269226,
      "learning_rate": 8.839649537288009e-07,
      "loss": 0.2879,
      "step": 25213
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8942945003509521,
      "learning_rate": 8.835060197887268e-07,
      "loss": 0.3301,
      "step": 25214
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.742937445640564,
      "learning_rate": 8.830471995074052e-07,
      "loss": 0.3063,
      "step": 25215
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.2283191680908203,
      "learning_rate": 8.825884928905537e-07,
      "loss": 0.3392,
      "step": 25216
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6459624767303467,
      "learning_rate": 8.821298999438954e-07,
      "loss": 0.2163,
      "step": 25217
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.054914951324463,
      "learning_rate": 8.81671420673147e-07,
      "loss": 0.371,
      "step": 25218
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.1300013065338135,
      "learning_rate": 8.812130550840226e-07,
      "loss": 0.3399,
      "step": 25219
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.5578908920288086,
      "learning_rate": 8.807548031822355e-07,
      "loss": 0.254,
      "step": 25220
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.2092111110687256,
      "learning_rate": 8.802966649735034e-07,
      "loss": 0.3127,
      "step": 25221
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8581607341766357,
      "learning_rate": 8.798386404635361e-07,
      "loss": 0.2806,
      "step": 25222
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3081324100494385,
      "learning_rate": 8.793807296580414e-07,
      "loss": 0.3556,
      "step": 25223
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9439202547073364,
      "learning_rate": 8.789229325627313e-07,
      "loss": 0.3311,
      "step": 25224
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9995394945144653,
      "learning_rate": 8.784652491833135e-07,
      "loss": 0.2574,
      "step": 25225
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.992376685142517,
      "learning_rate": 8.78007679525491e-07,
      "loss": 0.3552,
      "step": 25226
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9239741563796997,
      "learning_rate": 8.775502235949729e-07,
      "loss": 0.2928,
      "step": 25227
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.946677565574646,
      "learning_rate": 8.7709288139746e-07,
      "loss": 0.3045,
      "step": 25228
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9499115943908691,
      "learning_rate": 8.766356529386543e-07,
      "loss": 0.2655,
      "step": 25229
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0705902576446533,
      "learning_rate": 8.761785382242538e-07,
      "loss": 0.3466,
      "step": 25230
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3337748050689697,
      "learning_rate": 8.757215372599626e-07,
      "loss": 0.3726,
      "step": 25231
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5552432537078857,
      "learning_rate": 8.75264650051475e-07,
      "loss": 0.2106,
      "step": 25232
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7014678716659546,
      "learning_rate": 8.748078766044865e-07,
      "loss": 0.2859,
      "step": 25233
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.59907865524292,
      "learning_rate": 8.743512169246949e-07,
      "loss": 0.2796,
      "step": 25234
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.1364922523498535,
      "learning_rate": 8.73894671017792e-07,
      "loss": 0.3384,
      "step": 25235
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.973325490951538,
      "learning_rate": 8.734382388894668e-07,
      "loss": 0.3323,
      "step": 25236
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7508479356765747,
      "learning_rate": 8.72981920545416e-07,
      "loss": 0.2433,
      "step": 25237
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.1439356803894043,
      "learning_rate": 8.725257159913236e-07,
      "loss": 0.3241,
      "step": 25238
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7974859476089478,
      "learning_rate": 8.720696252328786e-07,
      "loss": 0.384,
      "step": 25239
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6703938245773315,
      "learning_rate": 8.716136482757686e-07,
      "loss": 0.239,
      "step": 25240
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8475350141525269,
      "learning_rate": 8.711577851256781e-07,
      "loss": 0.3107,
      "step": 25241
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.4911882877349854,
      "learning_rate": 8.707020357882889e-07,
      "loss": 0.391,
      "step": 25242
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.127272844314575,
      "learning_rate": 8.702464002692834e-07,
      "loss": 0.369,
      "step": 25243
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7328391075134277,
      "learning_rate": 8.697908785743437e-07,
      "loss": 0.2772,
      "step": 25244
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.000333070755005,
      "learning_rate": 8.693354707091484e-07,
      "loss": 0.2921,
      "step": 25245
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6824554204940796,
      "learning_rate": 8.688801766793731e-07,
      "loss": 0.291,
      "step": 25246
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8767142295837402,
      "learning_rate": 8.684249964906977e-07,
      "loss": 0.3266,
      "step": 25247
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0463123321533203,
      "learning_rate": 8.679699301487954e-07,
      "loss": 0.2757,
      "step": 25248
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9894026517868042,
      "learning_rate": 8.675149776593372e-07,
      "loss": 0.2651,
      "step": 25249
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.18481183052063,
      "learning_rate": 8.670601390279998e-07,
      "loss": 0.2975,
      "step": 25250
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9031816720962524,
      "learning_rate": 8.666054142604507e-07,
      "loss": 0.2856,
      "step": 25251
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3614282608032227,
      "learning_rate": 8.661508033623589e-07,
      "loss": 0.4184,
      "step": 25252
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.727264165878296,
      "learning_rate": 8.656963063393953e-07,
      "loss": 0.3395,
      "step": 25253
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.714365005493164,
      "learning_rate": 8.652419231972242e-07,
      "loss": 0.2911,
      "step": 25254
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.845123291015625,
      "learning_rate": 8.647876539415101e-07,
      "loss": 0.2797,
      "step": 25255
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9267709255218506,
      "learning_rate": 8.643334985779162e-07,
      "loss": 0.3291,
      "step": 25256
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.336324691772461,
      "learning_rate": 8.638794571121067e-07,
      "loss": 0.301,
      "step": 25257
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9562733173370361,
      "learning_rate": 8.634255295497407e-07,
      "loss": 0.2938,
      "step": 25258
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.598489999771118,
      "learning_rate": 8.629717158964767e-07,
      "loss": 0.3601,
      "step": 25259
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7231223583221436,
      "learning_rate": 8.62518016157976e-07,
      "loss": 0.3481,
      "step": 25260
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.048412799835205,
      "learning_rate": 8.620644303398918e-07,
      "loss": 0.3431,
      "step": 25261
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8036839962005615,
      "learning_rate": 8.616109584478793e-07,
      "loss": 0.3037,
      "step": 25262
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.055168628692627,
      "learning_rate": 8.611576004875943e-07,
      "loss": 0.2795,
      "step": 25263
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0165200233459473,
      "learning_rate": 8.607043564646878e-07,
      "loss": 0.2995,
      "step": 25264
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.1198627948760986,
      "learning_rate": 8.602512263848084e-07,
      "loss": 0.3036,
      "step": 25265
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7743351459503174,
      "learning_rate": 8.597982102536096e-07,
      "loss": 0.3073,
      "step": 25266
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7183464765548706,
      "learning_rate": 8.593453080767366e-07,
      "loss": 0.2978,
      "step": 25267
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7422125339508057,
      "learning_rate": 8.588925198598364e-07,
      "loss": 0.2629,
      "step": 25268
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.061354875564575,
      "learning_rate": 8.58439845608553e-07,
      "loss": 0.2863,
      "step": 25269
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.29711651802063,
      "learning_rate": 8.579872853285332e-07,
      "loss": 0.3064,
      "step": 25270
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0957906246185303,
      "learning_rate": 8.575348390254157e-07,
      "loss": 0.2989,
      "step": 25271
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.543860673904419,
      "learning_rate": 8.570825067048427e-07,
      "loss": 0.3086,
      "step": 25272
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.490269660949707,
      "learning_rate": 8.566302883724554e-07,
      "loss": 0.3623,
      "step": 25273
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9154704809188843,
      "learning_rate": 8.56178184033889e-07,
      "loss": 0.2641,
      "step": 25274
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0196423530578613,
      "learning_rate": 8.557261936947803e-07,
      "loss": 0.2983,
      "step": 25275
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.389120578765869,
      "learning_rate": 8.552743173607669e-07,
      "loss": 0.3355,
      "step": 25276
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.0003607273101807,
      "learning_rate": 8.548225550374811e-07,
      "loss": 0.3314,
      "step": 25277
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7135934829711914,
      "learning_rate": 8.543709067305527e-07,
      "loss": 0.254,
      "step": 25278
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8436330556869507,
      "learning_rate": 8.539193724456174e-07,
      "loss": 0.2826,
      "step": 25279
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7121134996414185,
      "learning_rate": 8.534679521883016e-07,
      "loss": 0.3186,
      "step": 25280
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.1219942569732666,
      "learning_rate": 8.530166459642342e-07,
      "loss": 0.3721,
      "step": 25281
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.812922716140747,
      "learning_rate": 8.525654537790395e-07,
      "loss": 0.367,
      "step": 25282
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.810962200164795,
      "learning_rate": 8.521143756383465e-07,
      "loss": 0.3389,
      "step": 25283
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.3018250465393066,
      "learning_rate": 8.516634115477773e-07,
      "loss": 0.3003,
      "step": 25284
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9469660520553589,
      "learning_rate": 8.512125615129519e-07,
      "loss": 0.2814,
      "step": 25285
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.5960071086883545,
      "learning_rate": 8.507618255394956e-07,
      "loss": 0.3374,
      "step": 25286
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7768681049346924,
      "learning_rate": 8.503112036330252e-07,
      "loss": 0.2077,
      "step": 25287
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7049295902252197,
      "learning_rate": 8.498606957991573e-07,
      "loss": 0.2904,
      "step": 25288
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8610559701919556,
      "learning_rate": 8.494103020435129e-07,
      "loss": 0.317,
      "step": 25289
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.865620493888855,
      "learning_rate": 8.489600223717043e-07,
      "loss": 0.337,
      "step": 25290
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0977134704589844,
      "learning_rate": 8.485098567893446e-07,
      "loss": 0.3265,
      "step": 25291
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.2556653022766113,
      "learning_rate": 8.480598053020483e-07,
      "loss": 0.2501,
      "step": 25292
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.8676016330718994,
      "learning_rate": 8.476098679154266e-07,
      "loss": 0.3346,
      "step": 25293
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.33842396736145,
      "learning_rate": 8.471600446350869e-07,
      "loss": 0.2815,
      "step": 25294
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9414974451065063,
      "learning_rate": 8.467103354666373e-07,
      "loss": 0.3458,
      "step": 25295
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0814712047576904,
      "learning_rate": 8.462607404156864e-07,
      "loss": 0.2556,
      "step": 25296
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.3219969272613525,
      "learning_rate": 8.458112594878387e-07,
      "loss": 0.3311,
      "step": 25297
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.921419620513916,
      "learning_rate": 8.453618926886964e-07,
      "loss": 0.2528,
      "step": 25298
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0695974826812744,
      "learning_rate": 8.449126400238661e-07,
      "loss": 0.3346,
      "step": 25299
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9917031526565552,
      "learning_rate": 8.444635014989455e-07,
      "loss": 0.2977,
      "step": 25300
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6810508966445923,
      "learning_rate": 8.440144771195324e-07,
      "loss": 0.2777,
      "step": 25301
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1220643520355225,
      "learning_rate": 8.435655668912302e-07,
      "loss": 0.3269,
      "step": 25302
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9266891479492188,
      "learning_rate": 8.431167708196319e-07,
      "loss": 0.2529,
      "step": 25303
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8331239223480225,
      "learning_rate": 8.426680889103322e-07,
      "loss": 0.2818,
      "step": 25304
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0242297649383545,
      "learning_rate": 8.422195211689288e-07,
      "loss": 0.2587,
      "step": 25305
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0971882343292236,
      "learning_rate": 8.417710676010116e-07,
      "loss": 0.357,
      "step": 25306
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0554254055023193,
      "learning_rate": 8.413227282121717e-07,
      "loss": 0.3472,
      "step": 25307
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0869669914245605,
      "learning_rate": 8.408745030079979e-07,
      "loss": 0.2826,
      "step": 25308
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.758788824081421,
      "learning_rate": 8.404263919940802e-07,
      "loss": 0.3354,
      "step": 25309
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9355162382125854,
      "learning_rate": 8.399783951760054e-07,
      "loss": 0.2787,
      "step": 25310
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8910516500473022,
      "learning_rate": 8.395305125593556e-07,
      "loss": 0.3189,
      "step": 25311
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.257188081741333,
      "learning_rate": 8.390827441497196e-07,
      "loss": 0.3769,
      "step": 25312
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5221595764160156,
      "learning_rate": 8.386350899526774e-07,
      "loss": 0.2702,
      "step": 25313
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0403616428375244,
      "learning_rate": 8.38187549973808e-07,
      "loss": 0.2549,
      "step": 25314
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9897195100784302,
      "learning_rate": 8.377401242186966e-07,
      "loss": 0.2867,
      "step": 25315
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0587854385375977,
      "learning_rate": 8.372928126929158e-07,
      "loss": 0.3553,
      "step": 25316
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8760734796524048,
      "learning_rate": 8.368456154020444e-07,
      "loss": 0.2767,
      "step": 25317
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7671486139297485,
      "learning_rate": 8.363985323516588e-07,
      "loss": 0.2926,
      "step": 25318
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.009939193725586,
      "learning_rate": 8.359515635473314e-07,
      "loss": 0.2693,
      "step": 25319
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8115547895431519,
      "learning_rate": 8.355047089946344e-07,
      "loss": 0.3444,
      "step": 25320
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.6093220710754395,
      "learning_rate": 8.350579686991422e-07,
      "loss": 0.3858,
      "step": 25321
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.288120746612549,
      "learning_rate": 8.346113426664216e-07,
      "loss": 0.3652,
      "step": 25322
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7856336832046509,
      "learning_rate": 8.3416483090204e-07,
      "loss": 0.318,
      "step": 25323
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.477708339691162,
      "learning_rate": 8.337184334115678e-07,
      "loss": 0.2715,
      "step": 25324
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1781394481658936,
      "learning_rate": 8.332721502005669e-07,
      "loss": 0.2616,
      "step": 25325
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9202581644058228,
      "learning_rate": 8.32825981274602e-07,
      "loss": 0.3049,
      "step": 25326
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8929710388183594,
      "learning_rate": 8.323799266392385e-07,
      "loss": 0.3136,
      "step": 25327
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.7097179889678955,
      "learning_rate": 8.319339863000353e-07,
      "loss": 0.3165,
      "step": 25328
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7329027652740479,
      "learning_rate": 8.314881602625513e-07,
      "loss": 0.3029,
      "step": 25329
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.963819980621338,
      "learning_rate": 8.310424485323454e-07,
      "loss": 0.3535,
      "step": 25330
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.003530502319336,
      "learning_rate": 8.305968511149754e-07,
      "loss": 0.323,
      "step": 25331
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8962310552597046,
      "learning_rate": 8.301513680159956e-07,
      "loss": 0.3313,
      "step": 25332
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.972956895828247,
      "learning_rate": 8.297059992409595e-07,
      "loss": 0.3586,
      "step": 25333
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.341257095336914,
      "learning_rate": 8.292607447954226e-07,
      "loss": 0.3937,
      "step": 25334
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9081809520721436,
      "learning_rate": 8.288156046849338e-07,
      "loss": 0.3006,
      "step": 25335
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7242357730865479,
      "learning_rate": 8.283705789150409e-07,
      "loss": 0.2853,
      "step": 25336
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1403982639312744,
      "learning_rate": 8.279256674912972e-07,
      "loss": 0.3208,
      "step": 25337
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0421500205993652,
      "learning_rate": 8.27480870419246e-07,
      "loss": 0.3716,
      "step": 25338
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.684291958808899,
      "learning_rate": 8.270361877044319e-07,
      "loss": 0.2851,
      "step": 25339
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.5928807258605957,
      "learning_rate": 8.265916193524026e-07,
      "loss": 0.3392,
      "step": 25340
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9077125787734985,
      "learning_rate": 8.261471653686993e-07,
      "loss": 0.3004,
      "step": 25341
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9036058187484741,
      "learning_rate": 8.257028257588618e-07,
      "loss": 0.3439,
      "step": 25342
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.8067610263824463,
      "learning_rate": 8.252586005284292e-07,
      "loss": 0.2975,
      "step": 25343
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.956011176109314,
      "learning_rate": 8.248144896829424e-07,
      "loss": 0.3309,
      "step": 25344
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1281416416168213,
      "learning_rate": 8.243704932279373e-07,
      "loss": 0.2871,
      "step": 25345
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.8238158226013184,
      "learning_rate": 8.239266111689482e-07,
      "loss": 0.2813,
      "step": 25346
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7414751052856445,
      "learning_rate": 8.234828435115116e-07,
      "loss": 0.2882,
      "step": 25347
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8545424938201904,
      "learning_rate": 8.230391902611579e-07,
      "loss": 0.2973,
      "step": 25348
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.160006523132324,
      "learning_rate": 8.225956514234179e-07,
      "loss": 0.3819,
      "step": 25349
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8416860103607178,
      "learning_rate": 8.22152227003824e-07,
      "loss": 0.3488,
      "step": 25350
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.783752679824829,
      "learning_rate": 8.21708917007904e-07,
      "loss": 0.3044,
      "step": 25351
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.7305984497070312,
      "learning_rate": 8.212657214411812e-07,
      "loss": 0.282,
      "step": 25352
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.852555751800537,
      "learning_rate": 8.208226403091857e-07,
      "loss": 0.2984,
      "step": 25353
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.281439781188965,
      "learning_rate": 8.203796736174397e-07,
      "loss": 0.3502,
      "step": 25354
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1647422313690186,
      "learning_rate": 8.199368213714654e-07,
      "loss": 0.3695,
      "step": 25355
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.425185203552246,
      "learning_rate": 8.194940835767828e-07,
      "loss": 0.299,
      "step": 25356
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.025052309036255,
      "learning_rate": 8.190514602389155e-07,
      "loss": 0.3263,
      "step": 25357
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7302027940750122,
      "learning_rate": 8.186089513633788e-07,
      "loss": 0.289,
      "step": 25358
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1434643268585205,
      "learning_rate": 8.181665569556896e-07,
      "loss": 0.3898,
      "step": 25359
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.270690441131592,
      "learning_rate": 8.177242770213656e-07,
      "loss": 0.3004,
      "step": 25360
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.6842808723449707,
      "learning_rate": 8.172821115659191e-07,
      "loss": 0.3594,
      "step": 25361
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6963675022125244,
      "learning_rate": 8.168400605948612e-07,
      "loss": 0.3265,
      "step": 25362
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8503719568252563,
      "learning_rate": 8.163981241137075e-07,
      "loss": 0.2841,
      "step": 25363
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1231038570404053,
      "learning_rate": 8.159563021279649e-07,
      "loss": 0.3398,
      "step": 25364
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.191210985183716,
      "learning_rate": 8.155145946431409e-07,
      "loss": 0.3756,
      "step": 25365
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7049254179000854,
      "learning_rate": 8.150730016647446e-07,
      "loss": 0.3169,
      "step": 25366
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0387277603149414,
      "learning_rate": 8.146315231982815e-07,
      "loss": 0.4006,
      "step": 25367
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.006183624267578,
      "learning_rate": 8.14190159249254e-07,
      "loss": 0.3076,
      "step": 25368
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6361401081085205,
      "learning_rate": 8.137489098231643e-07,
      "loss": 0.3194,
      "step": 25369
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8185151815414429,
      "learning_rate": 8.13307774925517e-07,
      "loss": 0.3115,
      "step": 25370
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7559499740600586,
      "learning_rate": 8.128667545618086e-07,
      "loss": 0.3194,
      "step": 25371
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.16961669921875,
      "learning_rate": 8.124258487375381e-07,
      "loss": 0.2982,
      "step": 25372
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.588510036468506,
      "learning_rate": 8.119850574582044e-07,
      "loss": 0.4208,
      "step": 25373
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.2047526836395264,
      "learning_rate": 8.115443807293011e-07,
      "loss": 0.4057,
      "step": 25374
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.1244144439697266,
      "learning_rate": 8.111038185563213e-07,
      "loss": 0.3215,
      "step": 25375
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.0645904541015625,
      "learning_rate": 8.106633709447609e-07,
      "loss": 0.3037,
      "step": 25376
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5833971500396729,
      "learning_rate": 8.102230379001097e-07,
      "loss": 0.2821,
      "step": 25377
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.742984414100647,
      "learning_rate": 8.097828194278546e-07,
      "loss": 0.2819,
      "step": 25378
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.684533953666687,
      "learning_rate": 8.093427155334887e-07,
      "loss": 0.3368,
      "step": 25379
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8822916746139526,
      "learning_rate": 8.089027262224969e-07,
      "loss": 0.2912,
      "step": 25380
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.9714587926864624,
      "learning_rate": 8.084628515003645e-07,
      "loss": 0.2832,
      "step": 25381
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0279128551483154,
      "learning_rate": 8.08023091372574e-07,
      "loss": 0.2793,
      "step": 25382
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.569706916809082,
      "learning_rate": 8.075834458446108e-07,
      "loss": 0.2923,
      "step": 25383
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7740222215652466,
      "learning_rate": 8.071439149219562e-07,
      "loss": 0.2799,
      "step": 25384
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6580747365951538,
      "learning_rate": 8.067044986100858e-07,
      "loss": 0.3411,
      "step": 25385
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7479256391525269,
      "learning_rate": 8.062651969144842e-07,
      "loss": 0.2967,
      "step": 25386
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.054847002029419,
      "learning_rate": 8.058260098406234e-07,
      "loss": 0.3041,
      "step": 25387
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7379075288772583,
      "learning_rate": 8.053869373939804e-07,
      "loss": 0.2584,
      "step": 25388
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9208104610443115,
      "learning_rate": 8.049479795800308e-07,
      "loss": 0.2916,
      "step": 25389
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0150344371795654,
      "learning_rate": 8.045091364042456e-07,
      "loss": 0.3184,
      "step": 25390
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8692584037780762,
      "learning_rate": 8.040704078720951e-07,
      "loss": 0.2817,
      "step": 25391
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.82503080368042,
      "learning_rate": 8.036317939890514e-07,
      "loss": 0.3253,
      "step": 25392
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9872063398361206,
      "learning_rate": 8.031932947605825e-07,
      "loss": 0.3068,
      "step": 25393
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9812620878219604,
      "learning_rate": 8.02754910192155e-07,
      "loss": 0.3389,
      "step": 25394
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9685511589050293,
      "learning_rate": 8.023166402892313e-07,
      "loss": 0.3446,
      "step": 25395
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9398680925369263,
      "learning_rate": 8.018784850572803e-07,
      "loss": 0.2293,
      "step": 25396
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9187251329421997,
      "learning_rate": 8.014404445017632e-07,
      "loss": 0.3618,
      "step": 25397
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6449047327041626,
      "learning_rate": 8.01002518628139e-07,
      "loss": 0.3003,
      "step": 25398
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.3447623252868652,
      "learning_rate": 8.005647074418699e-07,
      "loss": 0.3695,
      "step": 25399
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0710747241973877,
      "learning_rate": 8.00127010948415e-07,
      "loss": 0.3871,
      "step": 25400
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9668676853179932,
      "learning_rate": 7.996894291532265e-07,
      "loss": 0.2804,
      "step": 25401
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.5834643840789795,
      "learning_rate": 7.992519620617656e-07,
      "loss": 0.3074,
      "step": 25402
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9470487833023071,
      "learning_rate": 7.988146096794835e-07,
      "loss": 0.2883,
      "step": 25403
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.912279486656189,
      "learning_rate": 7.983773720118315e-07,
      "loss": 0.2982,
      "step": 25404
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.268946886062622,
      "learning_rate": 7.979402490642641e-07,
      "loss": 0.3383,
      "step": 25405
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9261146783828735,
      "learning_rate": 7.9750324084223e-07,
      "loss": 0.368,
      "step": 25406
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0554051399230957,
      "learning_rate": 7.970663473511764e-07,
      "loss": 0.256,
      "step": 25407
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9815874099731445,
      "learning_rate": 7.966295685965498e-07,
      "loss": 0.3222,
      "step": 25408
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.48710560798645,
      "learning_rate": 7.96192904583798e-07,
      "loss": 0.2368,
      "step": 25409
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.2736079692840576,
      "learning_rate": 7.957563553183634e-07,
      "loss": 0.318,
      "step": 25410
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8674118518829346,
      "learning_rate": 7.953199208056883e-07,
      "loss": 0.3129,
      "step": 25411
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.542560577392578,
      "learning_rate": 7.948836010512151e-07,
      "loss": 0.3835,
      "step": 25412
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8891428709030151,
      "learning_rate": 7.944473960603838e-07,
      "loss": 0.3138,
      "step": 25413
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.5460922718048096,
      "learning_rate": 7.940113058386312e-07,
      "loss": 0.3506,
      "step": 25414
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7402724027633667,
      "learning_rate": 7.935753303913951e-07,
      "loss": 0.2572,
      "step": 25415
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.96442711353302,
      "learning_rate": 7.931394697241113e-07,
      "loss": 0.3382,
      "step": 25416
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7229244709014893,
      "learning_rate": 7.92703723842212e-07,
      "loss": 0.3215,
      "step": 25417
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0556695461273193,
      "learning_rate": 7.922680927511328e-07,
      "loss": 0.3144,
      "step": 25418
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8681284189224243,
      "learning_rate": 7.918325764563039e-07,
      "loss": 0.1938,
      "step": 25419
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.079805612564087,
      "learning_rate": 7.913971749631533e-07,
      "loss": 0.3456,
      "step": 25420
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9568382501602173,
      "learning_rate": 7.909618882771097e-07,
      "loss": 0.3112,
      "step": 25421
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.837836503982544,
      "learning_rate": 7.905267164036013e-07,
      "loss": 0.2915,
      "step": 25422
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0470190048217773,
      "learning_rate": 7.900916593480534e-07,
      "loss": 0.3073,
      "step": 25423
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.05832576751709,
      "learning_rate": 7.896567171158887e-07,
      "loss": 0.3139,
      "step": 25424
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.786206603050232,
      "learning_rate": 7.892218897125314e-07,
      "loss": 0.2584,
      "step": 25425
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6699665784835815,
      "learning_rate": 7.88787177143403e-07,
      "loss": 0.2433,
      "step": 25426
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8662301301956177,
      "learning_rate": 7.88352579413919e-07,
      "loss": 0.236,
      "step": 25427
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.2303433418273926,
      "learning_rate": 7.879180965295041e-07,
      "loss": 0.2952,
      "step": 25428
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.2370548248291016,
      "learning_rate": 7.874837284955717e-07,
      "loss": 0.3702,
      "step": 25429
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0944314002990723,
      "learning_rate": 7.87049475317535e-07,
      "loss": 0.3385,
      "step": 25430
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.061053514480591,
      "learning_rate": 7.866153370008123e-07,
      "loss": 0.2889,
      "step": 25431
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1199424266815186,
      "learning_rate": 7.861813135508145e-07,
      "loss": 0.3121,
      "step": 25432
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9973881244659424,
      "learning_rate": 7.857474049729519e-07,
      "loss": 0.299,
      "step": 25433
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0143797397613525,
      "learning_rate": 7.853136112726346e-07,
      "loss": 0.384,
      "step": 25434
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9901841878890991,
      "learning_rate": 7.848799324552714e-07,
      "loss": 0.2389,
      "step": 25435
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.753228187561035,
      "learning_rate": 7.844463685262693e-07,
      "loss": 0.4422,
      "step": 25436
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1737029552459717,
      "learning_rate": 7.840129194910318e-07,
      "loss": 0.3156,
      "step": 25437
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6475162506103516,
      "learning_rate": 7.835795853549655e-07,
      "loss": 0.2728,
      "step": 25438
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6004714965820312,
      "learning_rate": 7.831463661234729e-07,
      "loss": 0.2823,
      "step": 25439
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7991713285446167,
      "learning_rate": 7.827132618019506e-07,
      "loss": 0.2486,
      "step": 25440
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1006219387054443,
      "learning_rate": 7.822802723958045e-07,
      "loss": 0.2411,
      "step": 25441
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.3454315662384033,
      "learning_rate": 7.818473979104301e-07,
      "loss": 0.3854,
      "step": 25442
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9845044612884521,
      "learning_rate": 7.814146383512222e-07,
      "loss": 0.2718,
      "step": 25443
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7471704483032227,
      "learning_rate": 7.809819937235796e-07,
      "loss": 0.2177,
      "step": 25444
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7050065994262695,
      "learning_rate": 7.805494640328948e-07,
      "loss": 0.2553,
      "step": 25445
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1099724769592285,
      "learning_rate": 7.8011704928456e-07,
      "loss": 0.3043,
      "step": 25446
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0698635578155518,
      "learning_rate": 7.796847494839643e-07,
      "loss": 0.2806,
      "step": 25447
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.446477174758911,
      "learning_rate": 7.792525646365024e-07,
      "loss": 0.4044,
      "step": 25448
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8353235721588135,
      "learning_rate": 7.788204947475586e-07,
      "loss": 0.313,
      "step": 25449
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.79518723487854,
      "learning_rate": 7.7838853982252e-07,
      "loss": 0.3336,
      "step": 25450
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9360665082931519,
      "learning_rate": 7.779566998667743e-07,
      "loss": 0.264,
      "step": 25451
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8289040327072144,
      "learning_rate": 7.775249748857028e-07,
      "loss": 0.2889,
      "step": 25452
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.943320631980896,
      "learning_rate": 7.770933648846879e-07,
      "loss": 0.34,
      "step": 25453
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.945496916770935,
      "learning_rate": 7.766618698691142e-07,
      "loss": 0.4002,
      "step": 25454
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7355539798736572,
      "learning_rate": 7.762304898443573e-07,
      "loss": 0.3697,
      "step": 25455
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8329733610153198,
      "learning_rate": 7.757992248157964e-07,
      "loss": 0.3391,
      "step": 25456
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0316739082336426,
      "learning_rate": 7.753680747888104e-07,
      "loss": 0.2862,
      "step": 25457
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6585177183151245,
      "learning_rate": 7.749370397687728e-07,
      "loss": 0.2698,
      "step": 25458
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5608227252960205,
      "learning_rate": 7.74506119761057e-07,
      "loss": 0.2564,
      "step": 25459
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9135371446609497,
      "learning_rate": 7.740753147710345e-07,
      "loss": 0.2421,
      "step": 25460
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9105706214904785,
      "learning_rate": 7.736446248040797e-07,
      "loss": 0.2746,
      "step": 25461
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1904213428497314,
      "learning_rate": 7.732140498655605e-07,
      "loss": 0.3052,
      "step": 25462
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8746567964553833,
      "learning_rate": 7.727835899608438e-07,
      "loss": 0.3549,
      "step": 25463
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.5118751525878906,
      "learning_rate": 7.723532450952986e-07,
      "loss": 0.3169,
      "step": 25464
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.2573773860931396,
      "learning_rate": 7.719230152742885e-07,
      "loss": 0.3111,
      "step": 25465
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7266358137130737,
      "learning_rate": 7.714929005031769e-07,
      "loss": 0.281,
      "step": 25466
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8288516998291016,
      "learning_rate": 7.710629007873283e-07,
      "loss": 0.3215,
      "step": 25467
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1917853355407715,
      "learning_rate": 7.706330161321019e-07,
      "loss": 0.2889,
      "step": 25468
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0117695331573486,
      "learning_rate": 7.702032465428566e-07,
      "loss": 0.2589,
      "step": 25469
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1225075721740723,
      "learning_rate": 7.697735920249539e-07,
      "loss": 0.3211,
      "step": 25470
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8187658786773682,
      "learning_rate": 7.693440525837481e-07,
      "loss": 0.3374,
      "step": 25471
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.098180055618286,
      "learning_rate": 7.68914628224594e-07,
      "loss": 0.2677,
      "step": 25472
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.9578343629837036,
      "learning_rate": 7.684853189528452e-07,
      "loss": 0.2503,
      "step": 25473
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6698336601257324,
      "learning_rate": 7.680561247738561e-07,
      "loss": 0.2794,
      "step": 25474
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1403191089630127,
      "learning_rate": 7.676270456929758e-07,
      "loss": 0.3594,
      "step": 25475
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.1010000705718994,
      "learning_rate": 7.671980817155533e-07,
      "loss": 0.3413,
      "step": 25476
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.180194616317749,
      "learning_rate": 7.667692328469389e-07,
      "loss": 0.3512,
      "step": 25477
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.2970902919769287,
      "learning_rate": 7.663404990924783e-07,
      "loss": 0.3924,
      "step": 25478
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6351693868637085,
      "learning_rate": 7.659118804575149e-07,
      "loss": 0.3715,
      "step": 25479
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.612786889076233,
      "learning_rate": 7.654833769473957e-07,
      "loss": 0.3141,
      "step": 25480
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.887473225593567,
      "learning_rate": 7.650549885674619e-07,
      "loss": 0.278,
      "step": 25481
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.289820432662964,
      "learning_rate": 7.646267153230514e-07,
      "loss": 0.3479,
      "step": 25482
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9774978160858154,
      "learning_rate": 7.641985572195076e-07,
      "loss": 0.3003,
      "step": 25483
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.046637535095215,
      "learning_rate": 7.637705142621665e-07,
      "loss": 0.3178,
      "step": 25484
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.3177096843719482,
      "learning_rate": 7.63342586456366e-07,
      "loss": 0.2997,
      "step": 25485
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.5815536975860596,
      "learning_rate": 7.629147738074382e-07,
      "loss": 0.3146,
      "step": 25486
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7759896516799927,
      "learning_rate": 7.624870763207204e-07,
      "loss": 0.2596,
      "step": 25487
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1911628246307373,
      "learning_rate": 7.620594940015435e-07,
      "loss": 0.2848,
      "step": 25488
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.712814211845398,
      "learning_rate": 7.616320268552368e-07,
      "loss": 0.2641,
      "step": 25489
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.85934317111969,
      "learning_rate": 7.612046748871327e-07,
      "loss": 0.2874,
      "step": 25490
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5945684909820557,
      "learning_rate": 7.607774381025579e-07,
      "loss": 0.2117,
      "step": 25491
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7751117944717407,
      "learning_rate": 7.60350316506836e-07,
      "loss": 0.2429,
      "step": 25492
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.207181215286255,
      "learning_rate": 7.599233101052972e-07,
      "loss": 0.3188,
      "step": 25493
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.92872154712677,
      "learning_rate": 7.594964189032627e-07,
      "loss": 0.3685,
      "step": 25494
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.415982484817505,
      "learning_rate": 7.590696429060529e-07,
      "loss": 0.2755,
      "step": 25495
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9742146730422974,
      "learning_rate": 7.586429821189912e-07,
      "loss": 0.2978,
      "step": 25496
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8092522621154785,
      "learning_rate": 7.582164365473965e-07,
      "loss": 0.2747,
      "step": 25497
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.027653217315674,
      "learning_rate": 7.577900061965871e-07,
      "loss": 0.3058,
      "step": 25498
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.618255376815796,
      "learning_rate": 7.573636910718762e-07,
      "loss": 0.2726,
      "step": 25499
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.566041946411133,
      "learning_rate": 7.56937491178582e-07,
      "loss": 0.2721,
      "step": 25500
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0297563076019287,
      "learning_rate": 7.565114065220192e-07,
      "loss": 0.3006,
      "step": 25501
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0245778560638428,
      "learning_rate": 7.560854371074955e-07,
      "loss": 0.3189,
      "step": 25502
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0329360961914062,
      "learning_rate": 7.556595829403257e-07,
      "loss": 0.2969,
      "step": 25503
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8963674306869507,
      "learning_rate": 7.552338440258178e-07,
      "loss": 0.2299,
      "step": 25504
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9159754514694214,
      "learning_rate": 7.548082203692775e-07,
      "loss": 0.308,
      "step": 25505
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.34260630607605,
      "learning_rate": 7.54382711976016e-07,
      "loss": 0.3368,
      "step": 25506
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.402604818344116,
      "learning_rate": 7.539573188513338e-07,
      "loss": 0.2886,
      "step": 25507
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.739569902420044,
      "learning_rate": 7.535320410005354e-07,
      "loss": 0.2821,
      "step": 25508
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.980068325996399,
      "learning_rate": 7.531068784289253e-07,
      "loss": 0.3096,
      "step": 25509
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8376277685165405,
      "learning_rate": 7.526818311418027e-07,
      "loss": 0.3789,
      "step": 25510
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8358523845672607,
      "learning_rate": 7.522568991444656e-07,
      "loss": 0.3142,
      "step": 25511
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8649593591690063,
      "learning_rate": 7.51832082442212e-07,
      "loss": 0.309,
      "step": 25512
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0357882976531982,
      "learning_rate": 7.5140738104034e-07,
      "loss": 0.3431,
      "step": 25513
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7683340311050415,
      "learning_rate": 7.509827949441439e-07,
      "loss": 0.2717,
      "step": 25514
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.149937152862549,
      "learning_rate": 7.505583241589154e-07,
      "loss": 0.3113,
      "step": 25515
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.763386845588684,
      "learning_rate": 7.501339686899489e-07,
      "loss": 0.3291,
      "step": 25516
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.825559616088867,
      "learning_rate": 7.497097285425336e-07,
      "loss": 0.2811,
      "step": 25517
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1893091201782227,
      "learning_rate": 7.492856037219587e-07,
      "loss": 0.3062,
      "step": 25518
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.104037046432495,
      "learning_rate": 7.488615942335132e-07,
      "loss": 0.2647,
      "step": 25519
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.515913963317871,
      "learning_rate": 7.484377000824828e-07,
      "loss": 0.3896,
      "step": 25520
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8989697694778442,
      "learning_rate": 7.480139212741499e-07,
      "loss": 0.3247,
      "step": 25521
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9056951999664307,
      "learning_rate": 7.475902578138028e-07,
      "loss": 0.276,
      "step": 25522
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6618820428848267,
      "learning_rate": 7.471667097067203e-07,
      "loss": 0.2444,
      "step": 25523
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1174135208129883,
      "learning_rate": 7.467432769581828e-07,
      "loss": 0.2545,
      "step": 25524
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9431782960891724,
      "learning_rate": 7.463199595734682e-07,
      "loss": 0.3069,
      "step": 25525
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.9986063241958618,
      "learning_rate": 7.458967575578579e-07,
      "loss": 0.3384,
      "step": 25526
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5562078952789307,
      "learning_rate": 7.454736709166265e-07,
      "loss": 0.2473,
      "step": 25527
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0446133613586426,
      "learning_rate": 7.450506996550466e-07,
      "loss": 0.2774,
      "step": 25528
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.107511043548584,
      "learning_rate": 7.446278437783949e-07,
      "loss": 0.2948,
      "step": 25529
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.307936191558838,
      "learning_rate": 7.442051032919418e-07,
      "loss": 0.3508,
      "step": 25530
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.2101500034332275,
      "learning_rate": 7.43782478200955e-07,
      "loss": 0.2721,
      "step": 25531
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.604022979736328,
      "learning_rate": 7.433599685107096e-07,
      "loss": 0.3111,
      "step": 25532
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8803939819335938,
      "learning_rate": 7.429375742264688e-07,
      "loss": 0.28,
      "step": 25533
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7607026100158691,
      "learning_rate": 7.425152953534986e-07,
      "loss": 0.2654,
      "step": 25534
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0097105503082275,
      "learning_rate": 7.420931318970659e-07,
      "loss": 0.2832,
      "step": 25535
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.9669504165649414,
      "learning_rate": 7.41671083862433e-07,
      "loss": 0.4148,
      "step": 25536
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0032243728637695,
      "learning_rate": 7.412491512548626e-07,
      "loss": 0.2918,
      "step": 25537
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.952128529548645,
      "learning_rate": 7.408273340796113e-07,
      "loss": 0.3415,
      "step": 25538
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8236711025238037,
      "learning_rate": 7.40405632341944e-07,
      "loss": 0.3006,
      "step": 25539
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8325029611587524,
      "learning_rate": 7.399840460471141e-07,
      "loss": 0.3475,
      "step": 25540
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.348775863647461,
      "learning_rate": 7.395625752003777e-07,
      "loss": 0.3051,
      "step": 25541
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6638363599777222,
      "learning_rate": 7.391412198069913e-07,
      "loss": 0.2868,
      "step": 25542
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7675739526748657,
      "learning_rate": 7.387199798722078e-07,
      "loss": 0.3562,
      "step": 25543
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8331049680709839,
      "learning_rate": 7.38298855401276e-07,
      "loss": 0.3265,
      "step": 25544
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5744436979293823,
      "learning_rate": 7.378778463994518e-07,
      "loss": 0.2517,
      "step": 25545
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0788450241088867,
      "learning_rate": 7.374569528719799e-07,
      "loss": 0.2452,
      "step": 25546
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7143731117248535,
      "learning_rate": 7.370361748241073e-07,
      "loss": 0.3032,
      "step": 25547
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8255280256271362,
      "learning_rate": 7.36615512261083e-07,
      "loss": 0.2911,
      "step": 25548
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.670472502708435,
      "learning_rate": 7.361949651881495e-07,
      "loss": 0.2068,
      "step": 25549
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.7667667865753174,
      "learning_rate": 7.357745336105515e-07,
      "loss": 0.2989,
      "step": 25550
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.196833610534668,
      "learning_rate": 7.353542175335271e-07,
      "loss": 0.3635,
      "step": 25551
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8885356187820435,
      "learning_rate": 7.349340169623209e-07,
      "loss": 0.2385,
      "step": 25552
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8509268760681152,
      "learning_rate": 7.345139319021699e-07,
      "loss": 0.2143,
      "step": 25553
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.700124740600586,
      "learning_rate": 7.340939623583099e-07,
      "loss": 0.3229,
      "step": 25554
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.186284303665161,
      "learning_rate": 7.336741083359789e-07,
      "loss": 0.3297,
      "step": 25555
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7201632261276245,
      "learning_rate": 7.332543698404126e-07,
      "loss": 0.2155,
      "step": 25556
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7762730121612549,
      "learning_rate": 7.328347468768393e-07,
      "loss": 0.3322,
      "step": 25557
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0704805850982666,
      "learning_rate": 7.324152394504957e-07,
      "loss": 0.2879,
      "step": 25558
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.688138484954834,
      "learning_rate": 7.319958475666089e-07,
      "loss": 0.2756,
      "step": 25559
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.0370261669158936,
      "learning_rate": 7.315765712304079e-07,
      "loss": 0.3211,
      "step": 25560
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4702826738357544,
      "learning_rate": 7.31157410447122e-07,
      "loss": 0.2281,
      "step": 25561
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1429224014282227,
      "learning_rate": 7.307383652219746e-07,
      "loss": 0.3066,
      "step": 25562
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1162493228912354,
      "learning_rate": 7.303194355601917e-07,
      "loss": 0.3037,
      "step": 25563
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7070984840393066,
      "learning_rate": 7.299006214669946e-07,
      "loss": 0.1952,
      "step": 25564
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.083094835281372,
      "learning_rate": 7.294819229476069e-07,
      "loss": 0.299,
      "step": 25565
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1590359210968018,
      "learning_rate": 7.290633400072478e-07,
      "loss": 0.2889,
      "step": 25566
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.192248821258545,
      "learning_rate": 7.286448726511342e-07,
      "loss": 0.3481,
      "step": 25567
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7702292203903198,
      "learning_rate": 7.282265208844863e-07,
      "loss": 0.2589,
      "step": 25568
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8899996280670166,
      "learning_rate": 7.27808284712519e-07,
      "loss": 0.3068,
      "step": 25569
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8088958263397217,
      "learning_rate": 7.273901641404435e-07,
      "loss": 0.2725,
      "step": 25570
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.869592308998108,
      "learning_rate": 7.269721591734769e-07,
      "loss": 0.3371,
      "step": 25571
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.4390292167663574,
      "learning_rate": 7.265542698168293e-07,
      "loss": 0.3542,
      "step": 25572
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1952507495880127,
      "learning_rate": 7.261364960757089e-07,
      "loss": 0.3284,
      "step": 25573
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.1937332153320312,
      "learning_rate": 7.257188379553271e-07,
      "loss": 0.282,
      "step": 25574
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.4230728149414062,
      "learning_rate": 7.253012954608884e-07,
      "loss": 0.3237,
      "step": 25575
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.085188627243042,
      "learning_rate": 7.248838685976012e-07,
      "loss": 0.2934,
      "step": 25576
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9943093061447144,
      "learning_rate": 7.244665573706655e-07,
      "loss": 0.2429,
      "step": 25577
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7649856805801392,
      "learning_rate": 7.240493617852884e-07,
      "loss": 0.2421,
      "step": 25578
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8587805032730103,
      "learning_rate": 7.236322818466701e-07,
      "loss": 0.287,
      "step": 25579
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6004338264465332,
      "learning_rate": 7.232153175600077e-07,
      "loss": 0.2141,
      "step": 25580
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0918500423431396,
      "learning_rate": 7.227984689305034e-07,
      "loss": 0.2545,
      "step": 25581
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7838265895843506,
      "learning_rate": 7.223817359633534e-07,
      "loss": 0.3728,
      "step": 25582
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0134952068328857,
      "learning_rate": 7.2196511866375e-07,
      "loss": 0.3653,
      "step": 25583
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.152726411819458,
      "learning_rate": 7.215486170368913e-07,
      "loss": 0.3049,
      "step": 25584
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.985404372215271,
      "learning_rate": 7.211322310879698e-07,
      "loss": 0.3536,
      "step": 25585
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.3137996196746826,
      "learning_rate": 7.207159608221726e-07,
      "loss": 0.2422,
      "step": 25586
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9313410520553589,
      "learning_rate": 7.202998062446942e-07,
      "loss": 0.2286,
      "step": 25587
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.2536630630493164,
      "learning_rate": 7.198837673607218e-07,
      "loss": 0.3077,
      "step": 25588
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.053417444229126,
      "learning_rate": 7.19467844175441e-07,
      "loss": 0.3182,
      "step": 25589
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5182745456695557,
      "learning_rate": 7.190520366940368e-07,
      "loss": 0.2793,
      "step": 25590
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7032482624053955,
      "learning_rate": 7.186363449216959e-07,
      "loss": 0.322,
      "step": 25591
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.2623960971832275,
      "learning_rate": 7.182207688635989e-07,
      "loss": 0.3566,
      "step": 25592
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.847146987915039,
      "learning_rate": 7.178053085249259e-07,
      "loss": 0.3115,
      "step": 25593
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.85475492477417,
      "learning_rate": 7.173899639108594e-07,
      "loss": 0.3305,
      "step": 25594
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7961713075637817,
      "learning_rate": 7.169747350265777e-07,
      "loss": 0.3587,
      "step": 25595
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.044809341430664,
      "learning_rate": 7.165596218772531e-07,
      "loss": 0.3084,
      "step": 25596
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.024625539779663,
      "learning_rate": 7.161446244680659e-07,
      "loss": 0.3278,
      "step": 25597
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1594858169555664,
      "learning_rate": 7.157297428041887e-07,
      "loss": 0.2805,
      "step": 25598
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8572919368743896,
      "learning_rate": 7.153149768907919e-07,
      "loss": 0.2361,
      "step": 25599
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.429098606109619,
      "learning_rate": 7.149003267330501e-07,
      "loss": 0.3117,
      "step": 25600
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.773781657218933,
      "learning_rate": 7.144857923361303e-07,
      "loss": 0.2909,
      "step": 25601
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.493504285812378,
      "learning_rate": 7.140713737052018e-07,
      "loss": 0.3251,
      "step": 25602
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.63501238822937,
      "learning_rate": 7.136570708454282e-07,
      "loss": 0.3128,
      "step": 25603
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.899078130722046,
      "learning_rate": 7.132428837619799e-07,
      "loss": 0.2762,
      "step": 25604
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.024505853652954,
      "learning_rate": 7.12828812460018e-07,
      "loss": 0.285,
      "step": 25605
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.2437214851379395,
      "learning_rate": 7.124148569447043e-07,
      "loss": 0.3499,
      "step": 25606
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8873504400253296,
      "learning_rate": 7.120010172212011e-07,
      "loss": 0.2824,
      "step": 25607
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8581546545028687,
      "learning_rate": 7.115872932946677e-07,
      "loss": 0.3035,
      "step": 25608
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1607635021209717,
      "learning_rate": 7.1117368517026e-07,
      "loss": 0.3697,
      "step": 25609
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.975284457206726,
      "learning_rate": 7.107601928531382e-07,
      "loss": 0.2445,
      "step": 25610
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1075844764709473,
      "learning_rate": 7.103468163484551e-07,
      "loss": 0.3491,
      "step": 25611
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.5652618408203125,
      "learning_rate": 7.09933555661364e-07,
      "loss": 0.4312,
      "step": 25612
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1602683067321777,
      "learning_rate": 7.0952041079702e-07,
      "loss": 0.2821,
      "step": 25613
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.041813850402832,
      "learning_rate": 7.091073817605721e-07,
      "loss": 0.2698,
      "step": 25614
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.446720600128174,
      "learning_rate": 7.086944685571695e-07,
      "loss": 0.3131,
      "step": 25615
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.232332229614258,
      "learning_rate": 7.082816711919583e-07,
      "loss": 0.327,
      "step": 25616
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8396053314208984,
      "learning_rate": 7.078689896700886e-07,
      "loss": 0.3117,
      "step": 25617
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.2663652896881104,
      "learning_rate": 7.074564239967041e-07,
      "loss": 0.3175,
      "step": 25618
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.3939011096954346,
      "learning_rate": 7.070439741769464e-07,
      "loss": 0.363,
      "step": 25619
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6520198583602905,
      "learning_rate": 7.066316402159612e-07,
      "loss": 0.2739,
      "step": 25620
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7137587070465088,
      "learning_rate": 7.062194221188878e-07,
      "loss": 0.2824,
      "step": 25621
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.3842110633850098,
      "learning_rate": 7.058073198908633e-07,
      "loss": 0.3355,
      "step": 25622
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.4016449451446533,
      "learning_rate": 7.053953335370289e-07,
      "loss": 0.3442,
      "step": 25623
    },
    {
      "epoch": 2.65,
      "grad_norm": 3.6483840942382812,
      "learning_rate": 7.049834630625207e-07,
      "loss": 0.352,
      "step": 25624
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.104597568511963,
      "learning_rate": 7.04571708472469e-07,
      "loss": 0.3183,
      "step": 25625
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.173391103744507,
      "learning_rate": 7.04160069772013e-07,
      "loss": 0.3185,
      "step": 25626
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.174130916595459,
      "learning_rate": 7.037485469662831e-07,
      "loss": 0.3322,
      "step": 25627
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0034947395324707,
      "learning_rate": 7.033371400604083e-07,
      "loss": 0.3426,
      "step": 25628
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7836756706237793,
      "learning_rate": 7.029258490595181e-07,
      "loss": 0.3806,
      "step": 25629
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8251186609268188,
      "learning_rate": 7.025146739687416e-07,
      "loss": 0.323,
      "step": 25630
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.898386836051941,
      "learning_rate": 7.021036147932047e-07,
      "loss": 0.3184,
      "step": 25631
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.6280100345611572,
      "learning_rate": 7.016926715380301e-07,
      "loss": 0.3226,
      "step": 25632
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.333921432495117,
      "learning_rate": 7.012818442083447e-07,
      "loss": 0.2641,
      "step": 25633
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7107267379760742,
      "learning_rate": 7.008711328092688e-07,
      "loss": 0.2653,
      "step": 25634
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.329785108566284,
      "learning_rate": 7.004605373459217e-07,
      "loss": 0.3032,
      "step": 25635
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6389528512954712,
      "learning_rate": 7.000500578234248e-07,
      "loss": 0.286,
      "step": 25636
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9090425968170166,
      "learning_rate": 6.996396942468942e-07,
      "loss": 0.333,
      "step": 25637
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8118748664855957,
      "learning_rate": 6.992294466214445e-07,
      "loss": 0.2601,
      "step": 25638
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9634225368499756,
      "learning_rate": 6.988193149521938e-07,
      "loss": 0.3224,
      "step": 25639
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.857176661491394,
      "learning_rate": 6.984092992442537e-07,
      "loss": 0.2861,
      "step": 25640
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8482023477554321,
      "learning_rate": 6.979993995027368e-07,
      "loss": 0.2514,
      "step": 25641
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.218747854232788,
      "learning_rate": 6.9758961573275e-07,
      "loss": 0.3414,
      "step": 25642
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.6720566749572754,
      "learning_rate": 6.971799479394081e-07,
      "loss": 0.3316,
      "step": 25643
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9395074844360352,
      "learning_rate": 6.967703961278138e-07,
      "loss": 0.2679,
      "step": 25644
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7016139030456543,
      "learning_rate": 6.963609603030752e-07,
      "loss": 0.26,
      "step": 25645
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.2685444355010986,
      "learning_rate": 6.95951640470296e-07,
      "loss": 0.3799,
      "step": 25646
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1039559841156006,
      "learning_rate": 6.955424366345809e-07,
      "loss": 0.3274,
      "step": 25647
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8396120071411133,
      "learning_rate": 6.951333488010292e-07,
      "loss": 0.2783,
      "step": 25648
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6983543634414673,
      "learning_rate": 6.947243769747436e-07,
      "loss": 0.2554,
      "step": 25649
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1586344242095947,
      "learning_rate": 6.943155211608222e-07,
      "loss": 0.3268,
      "step": 25650
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0486021041870117,
      "learning_rate": 6.939067813643607e-07,
      "loss": 0.2411,
      "step": 25651
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.029902219772339,
      "learning_rate": 6.934981575904565e-07,
      "loss": 0.3353,
      "step": 25652
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.017392158508301,
      "learning_rate": 6.930896498442053e-07,
      "loss": 0.2466,
      "step": 25653
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.916276454925537,
      "learning_rate": 6.926812581306974e-07,
      "loss": 0.356,
      "step": 25654
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8608989715576172,
      "learning_rate": 6.922729824550245e-07,
      "loss": 0.3127,
      "step": 25655
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.902897596359253,
      "learning_rate": 6.918648228222802e-07,
      "loss": 0.3301,
      "step": 25656
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.713829755783081,
      "learning_rate": 6.914567792375504e-07,
      "loss": 0.289,
      "step": 25657
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8271934986114502,
      "learning_rate": 6.910488517059211e-07,
      "loss": 0.3065,
      "step": 25658
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.816753625869751,
      "learning_rate": 6.906410402324814e-07,
      "loss": 0.3512,
      "step": 25659
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9705256223678589,
      "learning_rate": 6.90233344822313e-07,
      "loss": 0.3675,
      "step": 25660
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.2852044105529785,
      "learning_rate": 6.898257654804996e-07,
      "loss": 0.287,
      "step": 25661
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1097159385681152,
      "learning_rate": 6.894183022121236e-07,
      "loss": 0.3099,
      "step": 25662
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5239789485931396,
      "learning_rate": 6.890109550222646e-07,
      "loss": 0.265,
      "step": 25663
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8033536672592163,
      "learning_rate": 6.886037239159992e-07,
      "loss": 0.2848,
      "step": 25664
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9947847127914429,
      "learning_rate": 6.881966088984071e-07,
      "loss": 0.2651,
      "step": 25665
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.9715622663497925,
      "learning_rate": 6.877896099745651e-07,
      "loss": 0.3806,
      "step": 25666
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.2519748210906982,
      "learning_rate": 6.873827271495426e-07,
      "loss": 0.3211,
      "step": 25667
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7711516618728638,
      "learning_rate": 6.869759604284154e-07,
      "loss": 0.2684,
      "step": 25668
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.1784043312072754,
      "learning_rate": 6.865693098162551e-07,
      "loss": 0.2769,
      "step": 25669
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.3703815937042236,
      "learning_rate": 6.861627753181288e-07,
      "loss": 0.4228,
      "step": 25670
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7981650829315186,
      "learning_rate": 6.857563569391091e-07,
      "loss": 0.3217,
      "step": 25671
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9232876300811768,
      "learning_rate": 6.853500546842607e-07,
      "loss": 0.3132,
      "step": 25672
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9865071773529053,
      "learning_rate": 6.849438685586473e-07,
      "loss": 0.2841,
      "step": 25673
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.1887710094451904,
      "learning_rate": 6.845377985673373e-07,
      "loss": 0.3282,
      "step": 25674
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.781543493270874,
      "learning_rate": 6.84131844715391e-07,
      "loss": 0.2853,
      "step": 25675
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.2008464336395264,
      "learning_rate": 6.837260070078688e-07,
      "loss": 0.2866,
      "step": 25676
    },
    {
      "epoch": 2.66,
      "grad_norm": 3.5430283546447754,
      "learning_rate": 6.8332028544983e-07,
      "loss": 0.3139,
      "step": 25677
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.412569761276245,
      "learning_rate": 6.82914680046336e-07,
      "loss": 0.3024,
      "step": 25678
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9648162126541138,
      "learning_rate": 6.825091908024418e-07,
      "loss": 0.3131,
      "step": 25679
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.287396192550659,
      "learning_rate": 6.821038177232009e-07,
      "loss": 0.289,
      "step": 25680
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.3749210834503174,
      "learning_rate": 6.816985608136706e-07,
      "loss": 0.2694,
      "step": 25681
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8831630945205688,
      "learning_rate": 6.812934200789012e-07,
      "loss": 0.2476,
      "step": 25682
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.033240556716919,
      "learning_rate": 6.808883955239432e-07,
      "loss": 0.2465,
      "step": 25683
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.854607343673706,
      "learning_rate": 6.80483487153849e-07,
      "loss": 0.2937,
      "step": 25684
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0401957035064697,
      "learning_rate": 6.800786949736659e-07,
      "loss": 0.3687,
      "step": 25685
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.072927474975586,
      "learning_rate": 6.796740189884377e-07,
      "loss": 0.2649,
      "step": 25686
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.2342803478240967,
      "learning_rate": 6.792694592032134e-07,
      "loss": 0.3784,
      "step": 25687
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.701549768447876,
      "learning_rate": 6.788650156230348e-07,
      "loss": 0.3313,
      "step": 25688
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.170241117477417,
      "learning_rate": 6.784606882529443e-07,
      "loss": 0.2966,
      "step": 25689
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7501887083053589,
      "learning_rate": 6.780564770979826e-07,
      "loss": 0.2537,
      "step": 25690
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.553045630455017,
      "learning_rate": 6.776523821631897e-07,
      "loss": 0.2398,
      "step": 25691
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.271049737930298,
      "learning_rate": 6.772484034536053e-07,
      "loss": 0.3504,
      "step": 25692
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9431453943252563,
      "learning_rate": 6.768445409742619e-07,
      "loss": 0.3111,
      "step": 25693
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9288865327835083,
      "learning_rate": 6.764407947301988e-07,
      "loss": 0.3438,
      "step": 25694
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.171710729598999,
      "learning_rate": 6.760371647264474e-07,
      "loss": 0.2581,
      "step": 25695
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9123497009277344,
      "learning_rate": 6.756336509680383e-07,
      "loss": 0.2671,
      "step": 25696
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0110602378845215,
      "learning_rate": 6.752302534600064e-07,
      "loss": 0.3812,
      "step": 25697
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7018163204193115,
      "learning_rate": 6.748269722073785e-07,
      "loss": 0.2888,
      "step": 25698
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0849382877349854,
      "learning_rate": 6.744238072151821e-07,
      "loss": 0.3266,
      "step": 25699
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.2588000297546387,
      "learning_rate": 6.740207584884451e-07,
      "loss": 0.3821,
      "step": 25700
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5855481624603271,
      "learning_rate": 6.736178260321912e-07,
      "loss": 0.2729,
      "step": 25701
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6395461559295654,
      "learning_rate": 6.732150098514457e-07,
      "loss": 0.3039,
      "step": 25702
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7930649518966675,
      "learning_rate": 6.728123099512273e-07,
      "loss": 0.3162,
      "step": 25703
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9751849174499512,
      "learning_rate": 6.724097263365593e-07,
      "loss": 0.2859,
      "step": 25704
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9627512693405151,
      "learning_rate": 6.720072590124616e-07,
      "loss": 0.2987,
      "step": 25705
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.075058937072754,
      "learning_rate": 6.716049079839482e-07,
      "loss": 0.3846,
      "step": 25706
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8377772569656372,
      "learning_rate": 6.712026732560395e-07,
      "loss": 0.3438,
      "step": 25707
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8904674053192139,
      "learning_rate": 6.708005548337482e-07,
      "loss": 0.2776,
      "step": 25708
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8061307668685913,
      "learning_rate": 6.703985527220869e-07,
      "loss": 0.3156,
      "step": 25709
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.7497129440307617,
      "learning_rate": 6.699966669260694e-07,
      "loss": 0.2829,
      "step": 25710
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.052159070968628,
      "learning_rate": 6.69594897450706e-07,
      "loss": 0.3086,
      "step": 25711
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8899507522583008,
      "learning_rate": 6.691932443010041e-07,
      "loss": 0.3026,
      "step": 25712
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.1047959327697754,
      "learning_rate": 6.687917074819727e-07,
      "loss": 0.3457,
      "step": 25713
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0380287170410156,
      "learning_rate": 6.683902869986181e-07,
      "loss": 0.3008,
      "step": 25714
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.2128427028656006,
      "learning_rate": 6.679889828559438e-07,
      "loss": 0.3795,
      "step": 25715
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.875412106513977,
      "learning_rate": 6.675877950589515e-07,
      "loss": 0.2756,
      "step": 25716
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.813632845878601,
      "learning_rate": 6.671867236126472e-07,
      "loss": 0.302,
      "step": 25717
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8440918922424316,
      "learning_rate": 6.667857685220291e-07,
      "loss": 0.2943,
      "step": 25718
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.016007423400879,
      "learning_rate": 6.663849297920933e-07,
      "loss": 0.248,
      "step": 25719
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.1148781776428223,
      "learning_rate": 6.659842074278422e-07,
      "loss": 0.3016,
      "step": 25720
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.2184207439422607,
      "learning_rate": 6.655836014342698e-07,
      "loss": 0.3285,
      "step": 25721
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.6745660305023193,
      "learning_rate": 6.651831118163676e-07,
      "loss": 0.3193,
      "step": 25722
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9063125848770142,
      "learning_rate": 6.647827385791339e-07,
      "loss": 0.2865,
      "step": 25723
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.846078395843506,
      "learning_rate": 6.643824817275569e-07,
      "loss": 0.2971,
      "step": 25724
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.210167407989502,
      "learning_rate": 6.639823412666268e-07,
      "loss": 0.3322,
      "step": 25725
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0182409286499023,
      "learning_rate": 6.635823172013334e-07,
      "loss": 0.305,
      "step": 25726
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0250940322875977,
      "learning_rate": 6.631824095366646e-07,
      "loss": 0.2178,
      "step": 25727
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.964741826057434,
      "learning_rate": 6.627826182776042e-07,
      "loss": 0.2469,
      "step": 25728
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7543278932571411,
      "learning_rate": 6.623829434291362e-07,
      "loss": 0.2456,
      "step": 25729
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.076765775680542,
      "learning_rate": 6.619833849962465e-07,
      "loss": 0.3956,
      "step": 25730
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.1354000568389893,
      "learning_rate": 6.615839429839155e-07,
      "loss": 0.2803,
      "step": 25731
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.519987106323242,
      "learning_rate": 6.611846173971204e-07,
      "loss": 0.3237,
      "step": 25732
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8069995641708374,
      "learning_rate": 6.607854082408427e-07,
      "loss": 0.3087,
      "step": 25733
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.524249792098999,
      "learning_rate": 6.603863155200596e-07,
      "loss": 0.359,
      "step": 25734
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9911479949951172,
      "learning_rate": 6.599873392397437e-07,
      "loss": 0.3393,
      "step": 25735
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.4148499965667725,
      "learning_rate": 6.595884794048735e-07,
      "loss": 0.3961,
      "step": 25736
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.055990219116211,
      "learning_rate": 6.591897360204191e-07,
      "loss": 0.3484,
      "step": 25737
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.760558843612671,
      "learning_rate": 6.587911090913502e-07,
      "loss": 0.253,
      "step": 25738
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.2539682388305664,
      "learning_rate": 6.583925986226402e-07,
      "loss": 0.3245,
      "step": 25739
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0255227088928223,
      "learning_rate": 6.579942046192567e-07,
      "loss": 0.2993,
      "step": 25740
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8334362506866455,
      "learning_rate": 6.575959270861654e-07,
      "loss": 0.2659,
      "step": 25741
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.157291889190674,
      "learning_rate": 6.571977660283313e-07,
      "loss": 0.3695,
      "step": 25742
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.6352860927581787,
      "learning_rate": 6.567997214507205e-07,
      "loss": 0.3349,
      "step": 25743
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5799968242645264,
      "learning_rate": 6.564017933582944e-07,
      "loss": 0.2585,
      "step": 25744
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.3466811180114746,
      "learning_rate": 6.560039817560127e-07,
      "loss": 0.3064,
      "step": 25745
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6840471029281616,
      "learning_rate": 6.556062866488378e-07,
      "loss": 0.3256,
      "step": 25746
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8142603635787964,
      "learning_rate": 6.552087080417257e-07,
      "loss": 0.3342,
      "step": 25747
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0801143646240234,
      "learning_rate": 6.548112459396339e-07,
      "loss": 0.2919,
      "step": 25748
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.15034556388855,
      "learning_rate": 6.544139003475191e-07,
      "loss": 0.2848,
      "step": 25749
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.663992166519165,
      "learning_rate": 6.540166712703333e-07,
      "loss": 0.2277,
      "step": 25750
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7894831895828247,
      "learning_rate": 6.536195587130289e-07,
      "loss": 0.3094,
      "step": 25751
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9482834339141846,
      "learning_rate": 6.532225626805588e-07,
      "loss": 0.3478,
      "step": 25752
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.0939254760742188,
      "learning_rate": 6.528256831778712e-07,
      "loss": 0.2751,
      "step": 25753
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.4120569229125977,
      "learning_rate": 6.524289202099143e-07,
      "loss": 0.3064,
      "step": 25754
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7368669509887695,
      "learning_rate": 6.52032273781632e-07,
      "loss": 0.2818,
      "step": 25755
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.2214672565460205,
      "learning_rate": 6.516357438979748e-07,
      "loss": 0.2992,
      "step": 25756
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8186631202697754,
      "learning_rate": 6.512393305638831e-07,
      "loss": 0.3075,
      "step": 25757
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.4073257446289062,
      "learning_rate": 6.508430337842975e-07,
      "loss": 0.3965,
      "step": 25758
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7631100416183472,
      "learning_rate": 6.504468535641628e-07,
      "loss": 0.2495,
      "step": 25759
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.530742883682251,
      "learning_rate": 6.500507899084163e-07,
      "loss": 0.3702,
      "step": 25760
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8413081169128418,
      "learning_rate": 6.496548428219951e-07,
      "loss": 0.2479,
      "step": 25761
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.1314752101898193,
      "learning_rate": 6.492590123098375e-07,
      "loss": 0.2514,
      "step": 25762
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7466726303100586,
      "learning_rate": 6.488632983768773e-07,
      "loss": 0.2807,
      "step": 25763
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.069456100463867,
      "learning_rate": 6.484677010280483e-07,
      "loss": 0.3136,
      "step": 25764
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8913735151290894,
      "learning_rate": 6.48072220268281e-07,
      "loss": 0.2964,
      "step": 25765
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.77116858959198,
      "learning_rate": 6.476768561025093e-07,
      "loss": 0.3567,
      "step": 25766
    },
    {
      "epoch": 2.66,
      "grad_norm": 2.3160011768341064,
      "learning_rate": 6.472816085356604e-07,
      "loss": 0.3263,
      "step": 25767
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0370583534240723,
      "learning_rate": 6.468864775726602e-07,
      "loss": 0.2994,
      "step": 25768
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.25875186920166,
      "learning_rate": 6.464914632184383e-07,
      "loss": 0.2742,
      "step": 25769
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.105026960372925,
      "learning_rate": 6.460965654779183e-07,
      "loss": 0.291,
      "step": 25770
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0522398948669434,
      "learning_rate": 6.457017843560221e-07,
      "loss": 0.3541,
      "step": 25771
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9869085550308228,
      "learning_rate": 6.453071198576743e-07,
      "loss": 0.2706,
      "step": 25772
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1427347660064697,
      "learning_rate": 6.449125719877925e-07,
      "loss": 0.3729,
      "step": 25773
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.098954439163208,
      "learning_rate": 6.445181407512968e-07,
      "loss": 0.3668,
      "step": 25774
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9203898906707764,
      "learning_rate": 6.441238261531058e-07,
      "loss": 0.3191,
      "step": 25775
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.2882986068725586,
      "learning_rate": 6.437296281981354e-07,
      "loss": 0.3276,
      "step": 25776
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.953415036201477,
      "learning_rate": 6.433355468912983e-07,
      "loss": 0.3881,
      "step": 25777
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8399101495742798,
      "learning_rate": 6.429415822375084e-07,
      "loss": 0.312,
      "step": 25778
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.606765031814575,
      "learning_rate": 6.425477342416786e-07,
      "loss": 0.2823,
      "step": 25779
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.838873267173767,
      "learning_rate": 6.421540029087193e-07,
      "loss": 0.3486,
      "step": 25780
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7963049411773682,
      "learning_rate": 6.417603882435352e-07,
      "loss": 0.2937,
      "step": 25781
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9935095310211182,
      "learning_rate": 6.413668902510395e-07,
      "loss": 0.2572,
      "step": 25782
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7116875648498535,
      "learning_rate": 6.409735089361358e-07,
      "loss": 0.2763,
      "step": 25783
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9544371366500854,
      "learning_rate": 6.405802443037257e-07,
      "loss": 0.318,
      "step": 25784
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8309506177902222,
      "learning_rate": 6.401870963587165e-07,
      "loss": 0.2477,
      "step": 25785
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.080207586288452,
      "learning_rate": 6.397940651060086e-07,
      "loss": 0.2703,
      "step": 25786
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.20600962638855,
      "learning_rate": 6.394011505504983e-07,
      "loss": 0.2993,
      "step": 25787
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1815574169158936,
      "learning_rate": 6.390083526970903e-07,
      "loss": 0.3886,
      "step": 25788
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4707887172698975,
      "learning_rate": 6.386156715506775e-07,
      "loss": 0.3446,
      "step": 25789
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.281033515930176,
      "learning_rate": 6.382231071161571e-07,
      "loss": 0.3132,
      "step": 25790
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.235612154006958,
      "learning_rate": 6.378306593984218e-07,
      "loss": 0.3209,
      "step": 25791
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.744618535041809,
      "learning_rate": 6.374383284023666e-07,
      "loss": 0.2763,
      "step": 25792
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8112261295318604,
      "learning_rate": 6.370461141328832e-07,
      "loss": 0.3119,
      "step": 25793
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7709875106811523,
      "learning_rate": 6.366540165948576e-07,
      "loss": 0.2328,
      "step": 25794
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9938582181930542,
      "learning_rate": 6.362620357931826e-07,
      "loss": 0.3376,
      "step": 25795
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9775621891021729,
      "learning_rate": 6.358701717327431e-07,
      "loss": 0.2243,
      "step": 25796
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.973549485206604,
      "learning_rate": 6.354784244184231e-07,
      "loss": 0.2914,
      "step": 25797
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1850504875183105,
      "learning_rate": 6.350867938551097e-07,
      "loss": 0.3223,
      "step": 25798
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1611666679382324,
      "learning_rate": 6.346952800476858e-07,
      "loss": 0.3211,
      "step": 25799
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8846317529678345,
      "learning_rate": 6.343038830010284e-07,
      "loss": 0.2977,
      "step": 25800
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.2825369834899902,
      "learning_rate": 6.339126027200204e-07,
      "loss": 0.293,
      "step": 25801
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1005122661590576,
      "learning_rate": 6.335214392095412e-07,
      "loss": 0.293,
      "step": 25802
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7710134983062744,
      "learning_rate": 6.331303924744647e-07,
      "loss": 0.303,
      "step": 25803
    },
    {
      "epoch": 2.67,
      "grad_norm": 3.0643341541290283,
      "learning_rate": 6.327394625196659e-07,
      "loss": 0.2957,
      "step": 25804
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0105650424957275,
      "learning_rate": 6.323486493500219e-07,
      "loss": 0.2524,
      "step": 25805
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4083211421966553,
      "learning_rate": 6.319579529704034e-07,
      "loss": 0.3195,
      "step": 25806
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9325698614120483,
      "learning_rate": 6.315673733856798e-07,
      "loss": 0.3435,
      "step": 25807
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.892948031425476,
      "learning_rate": 6.311769106007226e-07,
      "loss": 0.2458,
      "step": 25808
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0030970573425293,
      "learning_rate": 6.307865646204003e-07,
      "loss": 0.2844,
      "step": 25809
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.000722646713257,
      "learning_rate": 6.303963354495768e-07,
      "loss": 0.2638,
      "step": 25810
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1659750938415527,
      "learning_rate": 6.300062230931203e-07,
      "loss": 0.3345,
      "step": 25811
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8742597103118896,
      "learning_rate": 6.296162275558937e-07,
      "loss": 0.2199,
      "step": 25812
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0774388313293457,
      "learning_rate": 6.292263488427564e-07,
      "loss": 0.3213,
      "step": 25813
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.989444613456726,
      "learning_rate": 6.288365869585733e-07,
      "loss": 0.2953,
      "step": 25814
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6081161499023438,
      "learning_rate": 6.284469419082007e-07,
      "loss": 0.2806,
      "step": 25815
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6216449737548828,
      "learning_rate": 6.28057413696499e-07,
      "loss": 0.2159,
      "step": 25816
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.034230947494507,
      "learning_rate": 6.27668002328321e-07,
      "loss": 0.2845,
      "step": 25817
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.8442752361297607,
      "learning_rate": 6.272787078085252e-07,
      "loss": 0.361,
      "step": 25818
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8732715845108032,
      "learning_rate": 6.268895301419642e-07,
      "loss": 0.2705,
      "step": 25819
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9227244853973389,
      "learning_rate": 6.265004693334875e-07,
      "loss": 0.277,
      "step": 25820
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.177489757537842,
      "learning_rate": 6.261115253879502e-07,
      "loss": 0.3728,
      "step": 25821
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1005859375,
      "learning_rate": 6.257226983101972e-07,
      "loss": 0.3169,
      "step": 25822
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8847389221191406,
      "learning_rate": 6.253339881050779e-07,
      "loss": 0.3779,
      "step": 25823
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0865814685821533,
      "learning_rate": 6.249453947774387e-07,
      "loss": 0.2439,
      "step": 25824
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4778685569763184,
      "learning_rate": 6.245569183321243e-07,
      "loss": 0.2594,
      "step": 25825
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7238727807998657,
      "learning_rate": 6.241685587739765e-07,
      "loss": 0.3429,
      "step": 25826
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0966567993164062,
      "learning_rate": 6.237803161078404e-07,
      "loss": 0.3142,
      "step": 25827
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.110107660293579,
      "learning_rate": 6.233921903385543e-07,
      "loss": 0.3592,
      "step": 25828
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.744244933128357,
      "learning_rate": 6.230041814709564e-07,
      "loss": 0.2856,
      "step": 25829
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0897045135498047,
      "learning_rate": 6.226162895098831e-07,
      "loss": 0.3795,
      "step": 25830
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.9833266735076904,
      "learning_rate": 6.222285144601747e-07,
      "loss": 0.3423,
      "step": 25831
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.301649570465088,
      "learning_rate": 6.218408563266631e-07,
      "loss": 0.2993,
      "step": 25832
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.110133171081543,
      "learning_rate": 6.214533151141788e-07,
      "loss": 0.2855,
      "step": 25833
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.031967878341675,
      "learning_rate": 6.21065890827559e-07,
      "loss": 0.326,
      "step": 25834
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6694889068603516,
      "learning_rate": 6.2067858347163e-07,
      "loss": 0.3354,
      "step": 25835
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0335845947265625,
      "learning_rate": 6.202913930512211e-07,
      "loss": 0.2868,
      "step": 25836
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9007796049118042,
      "learning_rate": 6.199043195711608e-07,
      "loss": 0.2828,
      "step": 25837
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9135252237319946,
      "learning_rate": 6.195173630362738e-07,
      "loss": 0.2576,
      "step": 25838
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9469847679138184,
      "learning_rate": 6.191305234513833e-07,
      "loss": 0.3375,
      "step": 25839
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7492269277572632,
      "learning_rate": 6.187438008213154e-07,
      "loss": 0.2812,
      "step": 25840
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7663993835449219,
      "learning_rate": 6.183571951508893e-07,
      "loss": 0.284,
      "step": 25841
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0795414447784424,
      "learning_rate": 6.179707064449259e-07,
      "loss": 0.3436,
      "step": 25842
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4795165061950684,
      "learning_rate": 6.17584334708241e-07,
      "loss": 0.3627,
      "step": 25843
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8404732942581177,
      "learning_rate": 6.171980799456557e-07,
      "loss": 0.3045,
      "step": 25844
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4222168922424316,
      "learning_rate": 6.168119421619834e-07,
      "loss": 0.355,
      "step": 25845
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.0387914180755615,
      "learning_rate": 6.164259213620372e-07,
      "loss": 0.3178,
      "step": 25846
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.263322114944458,
      "learning_rate": 6.16040017550632e-07,
      "loss": 0.4015,
      "step": 25847
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9406375885009766,
      "learning_rate": 6.156542307325785e-07,
      "loss": 0.3027,
      "step": 25848
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4044508934020996,
      "learning_rate": 6.152685609126841e-07,
      "loss": 0.3042,
      "step": 25849
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7702733278274536,
      "learning_rate": 6.148830080957613e-07,
      "loss": 0.263,
      "step": 25850
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.167163848876953,
      "learning_rate": 6.144975722866131e-07,
      "loss": 0.2611,
      "step": 25851
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.047130584716797,
      "learning_rate": 6.141122534900467e-07,
      "loss": 0.3556,
      "step": 25852
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.4558074474334717,
      "learning_rate": 6.137270517108662e-07,
      "loss": 0.327,
      "step": 25853
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6636285781860352,
      "learning_rate": 6.133419669538731e-07,
      "loss": 0.3387,
      "step": 25854
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7953039407730103,
      "learning_rate": 6.129569992238693e-07,
      "loss": 0.3037,
      "step": 25855
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.044329881668091,
      "learning_rate": 6.12572148525653e-07,
      "loss": 0.3202,
      "step": 25856
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.02935528755188,
      "learning_rate": 6.121874148640239e-07,
      "loss": 0.3122,
      "step": 25857
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7186517715454102,
      "learning_rate": 6.11802798243779e-07,
      "loss": 0.27,
      "step": 25858
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6786589622497559,
      "learning_rate": 6.114182986697092e-07,
      "loss": 0.2533,
      "step": 25859
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8814854621887207,
      "learning_rate": 6.110339161466139e-07,
      "loss": 0.2894,
      "step": 25860
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.496996283531189,
      "learning_rate": 6.106496506792826e-07,
      "loss": 0.2985,
      "step": 25861
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.123420000076294,
      "learning_rate": 6.102655022725046e-07,
      "loss": 0.287,
      "step": 25862
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1959965229034424,
      "learning_rate": 6.09881470931073e-07,
      "loss": 0.4017,
      "step": 25863
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.838036060333252,
      "learning_rate": 6.094975566597727e-07,
      "loss": 0.2989,
      "step": 25864
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0445034503936768,
      "learning_rate": 6.0911375946339e-07,
      "loss": 0.2949,
      "step": 25865
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1198208332061768,
      "learning_rate": 6.087300793467121e-07,
      "loss": 0.284,
      "step": 25866
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7063326835632324,
      "learning_rate": 6.083465163145208e-07,
      "loss": 0.2878,
      "step": 25867
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0734829902648926,
      "learning_rate": 6.079630703715999e-07,
      "loss": 0.3326,
      "step": 25868
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.2180023193359375,
      "learning_rate": 6.075797415227258e-07,
      "loss": 0.3148,
      "step": 25869
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8467482328414917,
      "learning_rate": 6.071965297726823e-07,
      "loss": 0.3012,
      "step": 25870
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.686042308807373,
      "learning_rate": 6.068134351262444e-07,
      "loss": 0.3262,
      "step": 25871
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4309611320495605,
      "learning_rate": 6.064304575881885e-07,
      "loss": 0.249,
      "step": 25872
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6076397895812988,
      "learning_rate": 6.060475971632907e-07,
      "loss": 0.2388,
      "step": 25873
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.481607437133789,
      "learning_rate": 6.056648538563237e-07,
      "loss": 0.3628,
      "step": 25874
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6630868911743164,
      "learning_rate": 6.052822276720582e-07,
      "loss": 0.3282,
      "step": 25875
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.18457293510437,
      "learning_rate": 6.04899718615266e-07,
      "loss": 0.3292,
      "step": 25876
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9070430994033813,
      "learning_rate": 6.045173266907156e-07,
      "loss": 0.2901,
      "step": 25877
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0011563301086426,
      "learning_rate": 6.041350519031719e-07,
      "loss": 0.4395,
      "step": 25878
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.4374780654907227,
      "learning_rate": 6.037528942574055e-07,
      "loss": 0.3533,
      "step": 25879
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8886823654174805,
      "learning_rate": 6.033708537581784e-07,
      "loss": 0.2586,
      "step": 25880
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8109875917434692,
      "learning_rate": 6.029889304102532e-07,
      "loss": 0.2978,
      "step": 25881
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.078756093978882,
      "learning_rate": 6.026071242183906e-07,
      "loss": 0.2868,
      "step": 25882
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.425128936767578,
      "learning_rate": 6.022254351873547e-07,
      "loss": 0.3135,
      "step": 25883
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8331279754638672,
      "learning_rate": 6.018438633219003e-07,
      "loss": 0.3363,
      "step": 25884
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.07045841217041,
      "learning_rate": 6.014624086267851e-07,
      "loss": 0.2794,
      "step": 25885
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.000842332839966,
      "learning_rate": 6.010810711067671e-07,
      "loss": 0.3197,
      "step": 25886
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.224919080734253,
      "learning_rate": 6.006998507665996e-07,
      "loss": 0.2382,
      "step": 25887
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.981968879699707,
      "learning_rate": 6.003187476110329e-07,
      "loss": 0.3402,
      "step": 25888
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.001258373260498,
      "learning_rate": 5.999377616448221e-07,
      "loss": 0.3399,
      "step": 25889
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.181396722793579,
      "learning_rate": 5.995568928727147e-07,
      "loss": 0.2914,
      "step": 25890
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.3187801837921143,
      "learning_rate": 5.99176141299459e-07,
      "loss": 0.2915,
      "step": 25891
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8855259418487549,
      "learning_rate": 5.987955069298046e-07,
      "loss": 0.2478,
      "step": 25892
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1473236083984375,
      "learning_rate": 5.984149897684954e-07,
      "loss": 0.3162,
      "step": 25893
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8484833240509033,
      "learning_rate": 5.980345898202744e-07,
      "loss": 0.3054,
      "step": 25894
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6336201429367065,
      "learning_rate": 5.976543070898843e-07,
      "loss": 0.255,
      "step": 25895
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.2013566493988037,
      "learning_rate": 5.972741415820682e-07,
      "loss": 0.3482,
      "step": 25896
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.042081117630005,
      "learning_rate": 5.968940933015643e-07,
      "loss": 0.3314,
      "step": 25897
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.2308051586151123,
      "learning_rate": 5.965141622531101e-07,
      "loss": 0.3447,
      "step": 25898
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0847091674804688,
      "learning_rate": 5.961343484414439e-07,
      "loss": 0.2558,
      "step": 25899
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0049891471862793,
      "learning_rate": 5.957546518713009e-07,
      "loss": 0.2538,
      "step": 25900
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0111074447631836,
      "learning_rate": 5.953750725474139e-07,
      "loss": 0.3064,
      "step": 25901
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.860276222229004,
      "learning_rate": 5.949956104745158e-07,
      "loss": 0.3543,
      "step": 25902
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.038212537765503,
      "learning_rate": 5.946162656573385e-07,
      "loss": 0.3289,
      "step": 25903
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.37367844581604,
      "learning_rate": 5.94237038100608e-07,
      "loss": 0.3579,
      "step": 25904
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8327796459197998,
      "learning_rate": 5.938579278090573e-07,
      "loss": 0.3104,
      "step": 25905
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9334704875946045,
      "learning_rate": 5.934789347874093e-07,
      "loss": 0.3125,
      "step": 25906
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8917189836502075,
      "learning_rate": 5.931000590403901e-07,
      "loss": 0.2671,
      "step": 25907
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8745554685592651,
      "learning_rate": 5.927213005727217e-07,
      "loss": 0.2741,
      "step": 25908
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.100440263748169,
      "learning_rate": 5.923426593891302e-07,
      "loss": 0.3505,
      "step": 25909
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.236560106277466,
      "learning_rate": 5.919641354943328e-07,
      "loss": 0.3065,
      "step": 25910
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.900909185409546,
      "learning_rate": 5.915857288930482e-07,
      "loss": 0.2737,
      "step": 25911
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7725948095321655,
      "learning_rate": 5.912074395899981e-07,
      "loss": 0.2878,
      "step": 25912
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.4742517471313477,
      "learning_rate": 5.908292675898952e-07,
      "loss": 0.2264,
      "step": 25913
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8878118991851807,
      "learning_rate": 5.904512128974538e-07,
      "loss": 0.2687,
      "step": 25914
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8664162158966064,
      "learning_rate": 5.90073275517391e-07,
      "loss": 0.3058,
      "step": 25915
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8232481479644775,
      "learning_rate": 5.896954554544165e-07,
      "loss": 0.2487,
      "step": 25916
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8877108097076416,
      "learning_rate": 5.893177527132399e-07,
      "loss": 0.3251,
      "step": 25917
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0913360118865967,
      "learning_rate": 5.889401672985717e-07,
      "loss": 0.2986,
      "step": 25918
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.343148946762085,
      "learning_rate": 5.885626992151194e-07,
      "loss": 0.2533,
      "step": 25919
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9838734865188599,
      "learning_rate": 5.881853484675882e-07,
      "loss": 0.2728,
      "step": 25920
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.3612935543060303,
      "learning_rate": 5.878081150606807e-07,
      "loss": 0.3331,
      "step": 25921
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0757532119750977,
      "learning_rate": 5.874309989991034e-07,
      "loss": 0.3099,
      "step": 25922
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.165329933166504,
      "learning_rate": 5.87054000287558e-07,
      "loss": 0.338,
      "step": 25923
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.0639102458953857,
      "learning_rate": 5.866771189307408e-07,
      "loss": 0.2715,
      "step": 25924
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.184006690979004,
      "learning_rate": 5.863003549333557e-07,
      "loss": 0.3409,
      "step": 25925
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0114188194274902,
      "learning_rate": 5.859237083000968e-07,
      "loss": 0.2892,
      "step": 25926
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.025034189224243,
      "learning_rate": 5.855471790356592e-07,
      "loss": 0.2605,
      "step": 25927
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9762284755706787,
      "learning_rate": 5.851707671447393e-07,
      "loss": 0.2947,
      "step": 25928
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8645964860916138,
      "learning_rate": 5.847944726320298e-07,
      "loss": 0.272,
      "step": 25929
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.260540008544922,
      "learning_rate": 5.844182955022193e-07,
      "loss": 0.2668,
      "step": 25930
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9075126647949219,
      "learning_rate": 5.840422357600029e-07,
      "loss": 0.3572,
      "step": 25931
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.271024465560913,
      "learning_rate": 5.836662934100646e-07,
      "loss": 0.2967,
      "step": 25932
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0351569652557373,
      "learning_rate": 5.83290468457094e-07,
      "loss": 0.2874,
      "step": 25933
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.240601062774658,
      "learning_rate": 5.829147609057728e-07,
      "loss": 0.3234,
      "step": 25934
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.4180006980895996,
      "learning_rate": 5.825391707607908e-07,
      "loss": 0.3419,
      "step": 25935
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8452122211456299,
      "learning_rate": 5.821636980268264e-07,
      "loss": 0.3033,
      "step": 25936
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.075666904449463,
      "learning_rate": 5.817883427085613e-07,
      "loss": 0.2989,
      "step": 25937
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.03484845161438,
      "learning_rate": 5.814131048106775e-07,
      "loss": 0.2285,
      "step": 25938
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9780826568603516,
      "learning_rate": 5.810379843378511e-07,
      "loss": 0.2566,
      "step": 25939
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1007232666015625,
      "learning_rate": 5.806629812947584e-07,
      "loss": 0.3462,
      "step": 25940
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9526904821395874,
      "learning_rate": 5.802880956860768e-07,
      "loss": 0.2665,
      "step": 25941
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.980963945388794,
      "learning_rate": 5.799133275164792e-07,
      "loss": 0.2641,
      "step": 25942
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.237173557281494,
      "learning_rate": 5.795386767906364e-07,
      "loss": 0.2813,
      "step": 25943
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1019985675811768,
      "learning_rate": 5.791641435132223e-07,
      "loss": 0.3613,
      "step": 25944
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6299493312835693,
      "learning_rate": 5.787897276889054e-07,
      "loss": 0.2733,
      "step": 25945
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7736225128173828,
      "learning_rate": 5.78415429322352e-07,
      "loss": 0.2776,
      "step": 25946
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0175976753234863,
      "learning_rate": 5.780412484182285e-07,
      "loss": 0.3205,
      "step": 25947
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.3305959701538086,
      "learning_rate": 5.776671849812032e-07,
      "loss": 0.325,
      "step": 25948
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.0550525188446045,
      "learning_rate": 5.772932390159369e-07,
      "loss": 0.2878,
      "step": 25949
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1204051971435547,
      "learning_rate": 5.769194105270915e-07,
      "loss": 0.2767,
      "step": 25950
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.490775227546692,
      "learning_rate": 5.765456995193297e-07,
      "loss": 0.2467,
      "step": 25951
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.5514261722564697,
      "learning_rate": 5.761721059973103e-07,
      "loss": 0.3588,
      "step": 25952
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.1527318954467773,
      "learning_rate": 5.757986299656881e-07,
      "loss": 0.2849,
      "step": 25953
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8330297470092773,
      "learning_rate": 5.754252714291242e-07,
      "loss": 0.285,
      "step": 25954
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.6928884983062744,
      "learning_rate": 5.750520303922702e-07,
      "loss": 0.2421,
      "step": 25955
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.924667477607727,
      "learning_rate": 5.74678906859778e-07,
      "loss": 0.3436,
      "step": 25956
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7247741222381592,
      "learning_rate": 5.743059008363039e-07,
      "loss": 0.2691,
      "step": 25957
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.5717854499816895,
      "learning_rate": 5.739330123264952e-07,
      "loss": 0.2778,
      "step": 25958
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7176165580749512,
      "learning_rate": 5.735602413350028e-07,
      "loss": 0.2588,
      "step": 25959
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.074491262435913,
      "learning_rate": 5.731875878664706e-07,
      "loss": 0.284,
      "step": 25960
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.83770751953125,
      "learning_rate": 5.728150519255483e-07,
      "loss": 0.2881,
      "step": 25961
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9167594909667969,
      "learning_rate": 5.724426335168786e-07,
      "loss": 0.293,
      "step": 25962
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6654417514801025,
      "learning_rate": 5.720703326451049e-07,
      "loss": 0.271,
      "step": 25963
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1287055015563965,
      "learning_rate": 5.716981493148699e-07,
      "loss": 0.2983,
      "step": 25964
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.222200632095337,
      "learning_rate": 5.713260835308132e-07,
      "loss": 0.3547,
      "step": 25965
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3311712741851807,
      "learning_rate": 5.709541352975711e-07,
      "loss": 0.4468,
      "step": 25966
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1317636966705322,
      "learning_rate": 5.705823046197844e-07,
      "loss": 0.3578,
      "step": 25967
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1882808208465576,
      "learning_rate": 5.702105915020872e-07,
      "loss": 0.3074,
      "step": 25968
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0291295051574707,
      "learning_rate": 5.698389959491135e-07,
      "loss": 0.3005,
      "step": 25969
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9113788604736328,
      "learning_rate": 5.694675179654962e-07,
      "loss": 0.3338,
      "step": 25970
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.441246271133423,
      "learning_rate": 5.690961575558685e-07,
      "loss": 0.287,
      "step": 25971
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3459510803222656,
      "learning_rate": 5.687249147248586e-07,
      "loss": 0.3479,
      "step": 25972
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1986331939697266,
      "learning_rate": 5.68353789477093e-07,
      "loss": 0.3166,
      "step": 25973
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9918994903564453,
      "learning_rate": 5.679827818172024e-07,
      "loss": 0.2756,
      "step": 25974
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8363245725631714,
      "learning_rate": 5.676118917498108e-07,
      "loss": 0.2582,
      "step": 25975
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9089962244033813,
      "learning_rate": 5.672411192795402e-07,
      "loss": 0.2892,
      "step": 25976
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.4367830753326416,
      "learning_rate": 5.668704644110167e-07,
      "loss": 0.3754,
      "step": 25977
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.221925735473633,
      "learning_rate": 5.664999271488603e-07,
      "loss": 0.3151,
      "step": 25978
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9141144752502441,
      "learning_rate": 5.66129507497688e-07,
      "loss": 0.2663,
      "step": 25979
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.157670259475708,
      "learning_rate": 5.657592054621219e-07,
      "loss": 0.2562,
      "step": 25980
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.921740174293518,
      "learning_rate": 5.65389021046776e-07,
      "loss": 0.3005,
      "step": 25981
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.047682762145996,
      "learning_rate": 5.650189542562656e-07,
      "loss": 0.3456,
      "step": 25982
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4329453706741333,
      "learning_rate": 5.646490050952058e-07,
      "loss": 0.2083,
      "step": 25983
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.4002695083618164,
      "learning_rate": 5.642791735682096e-07,
      "loss": 0.3688,
      "step": 25984
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.973716378211975,
      "learning_rate": 5.639094596798855e-07,
      "loss": 0.2991,
      "step": 25985
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6791949272155762,
      "learning_rate": 5.635398634348421e-07,
      "loss": 0.2088,
      "step": 25986
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8005969524383545,
      "learning_rate": 5.631703848376912e-07,
      "loss": 0.3497,
      "step": 25987
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8804899454116821,
      "learning_rate": 5.62801023893037e-07,
      "loss": 0.3286,
      "step": 25988
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9186526536941528,
      "learning_rate": 5.624317806054835e-07,
      "loss": 0.2698,
      "step": 25989
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.212660312652588,
      "learning_rate": 5.62062654979636e-07,
      "loss": 0.3245,
      "step": 25990
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3919544219970703,
      "learning_rate": 5.616936470200962e-07,
      "loss": 0.272,
      "step": 25991
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1918039321899414,
      "learning_rate": 5.613247567314628e-07,
      "loss": 0.2585,
      "step": 25992
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7360258102416992,
      "learning_rate": 5.609559841183376e-07,
      "loss": 0.2498,
      "step": 25993
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.282205820083618,
      "learning_rate": 5.60587329185317e-07,
      "loss": 0.3868,
      "step": 25994
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3784077167510986,
      "learning_rate": 5.602187919369961e-07,
      "loss": 0.3515,
      "step": 25995
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.2255592346191406,
      "learning_rate": 5.598503723779713e-07,
      "loss": 0.2702,
      "step": 25996
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8667336702346802,
      "learning_rate": 5.594820705128356e-07,
      "loss": 0.3039,
      "step": 25997
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.221998691558838,
      "learning_rate": 5.591138863461798e-07,
      "loss": 0.3093,
      "step": 25998
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3771936893463135,
      "learning_rate": 5.587458198825946e-07,
      "loss": 0.2996,
      "step": 25999
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.191943883895874,
      "learning_rate": 5.583778711266685e-07,
      "loss": 0.2765,
      "step": 26000
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9064050912857056,
      "learning_rate": 5.580100400829902e-07,
      "loss": 0.2727,
      "step": 26001
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.103804349899292,
      "learning_rate": 5.576423267561437e-07,
      "loss": 0.3344,
      "step": 26002
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9267324209213257,
      "learning_rate": 5.572747311507154e-07,
      "loss": 0.2744,
      "step": 26003
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1723036766052246,
      "learning_rate": 5.569072532712871e-07,
      "loss": 0.2706,
      "step": 26004
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.205122947692871,
      "learning_rate": 5.565398931224386e-07,
      "loss": 0.3384,
      "step": 26005
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.157958984375,
      "learning_rate": 5.561726507087539e-07,
      "loss": 0.3649,
      "step": 26006
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0393269062042236,
      "learning_rate": 5.558055260348094e-07,
      "loss": 0.3413,
      "step": 26007
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.816036343574524,
      "learning_rate": 5.554385191051803e-07,
      "loss": 0.2538,
      "step": 26008
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.895261526107788,
      "learning_rate": 5.550716299244452e-07,
      "loss": 0.3807,
      "step": 26009
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.126821517944336,
      "learning_rate": 5.54704858497177e-07,
      "loss": 0.2892,
      "step": 26010
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9423636198043823,
      "learning_rate": 5.543382048279489e-07,
      "loss": 0.3343,
      "step": 26011
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.22512149810791,
      "learning_rate": 5.539716689213304e-07,
      "loss": 0.3898,
      "step": 26012
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8769530057907104,
      "learning_rate": 5.536052507818945e-07,
      "loss": 0.3889,
      "step": 26013
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9951382875442505,
      "learning_rate": 5.532389504142066e-07,
      "loss": 0.3548,
      "step": 26014
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.838529348373413,
      "learning_rate": 5.528727678228329e-07,
      "loss": 0.3004,
      "step": 26015
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.6139254570007324,
      "learning_rate": 5.525067030123421e-07,
      "loss": 0.3078,
      "step": 26016
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9008722305297852,
      "learning_rate": 5.521407559872971e-07,
      "loss": 0.3322,
      "step": 26017
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9880545139312744,
      "learning_rate": 5.517749267522576e-07,
      "loss": 0.3123,
      "step": 26018
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8496534824371338,
      "learning_rate": 5.514092153117878e-07,
      "loss": 0.2478,
      "step": 26019
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7526551485061646,
      "learning_rate": 5.510436216704452e-07,
      "loss": 0.2798,
      "step": 26020
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.100802421569824,
      "learning_rate": 5.50678145832787e-07,
      "loss": 0.3511,
      "step": 26021
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.399049758911133,
      "learning_rate": 5.503127878033721e-07,
      "loss": 0.2911,
      "step": 26022
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0353569984436035,
      "learning_rate": 5.499475475867544e-07,
      "loss": 0.3512,
      "step": 26023
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0640878677368164,
      "learning_rate": 5.49582425187487e-07,
      "loss": 0.3183,
      "step": 26024
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.904741644859314,
      "learning_rate": 5.492174206101231e-07,
      "loss": 0.3406,
      "step": 26025
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0072414875030518,
      "learning_rate": 5.488525338592132e-07,
      "loss": 0.2683,
      "step": 26026
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.146178722381592,
      "learning_rate": 5.48487764939305e-07,
      "loss": 0.3079,
      "step": 26027
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.939468502998352,
      "learning_rate": 5.481231138549481e-07,
      "loss": 0.2641,
      "step": 26028
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7964675426483154,
      "learning_rate": 5.477585806106889e-07,
      "loss": 0.2624,
      "step": 26029
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.2808310985565186,
      "learning_rate": 5.473941652110692e-07,
      "loss": 0.3627,
      "step": 26030
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7687774896621704,
      "learning_rate": 5.470298676606356e-07,
      "loss": 0.3424,
      "step": 26031
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6671913862228394,
      "learning_rate": 5.466656879639287e-07,
      "loss": 0.2395,
      "step": 26032
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.318908452987671,
      "learning_rate": 5.463016261254895e-07,
      "loss": 0.3055,
      "step": 26033
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.029665231704712,
      "learning_rate": 5.459376821498541e-07,
      "loss": 0.2642,
      "step": 26034
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8704060316085815,
      "learning_rate": 5.455738560415635e-07,
      "loss": 0.3648,
      "step": 26035
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9941482543945312,
      "learning_rate": 5.45210147805153e-07,
      "loss": 0.2407,
      "step": 26036
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1431429386138916,
      "learning_rate": 5.448465574451544e-07,
      "loss": 0.3147,
      "step": 26037
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9677321910858154,
      "learning_rate": 5.444830849661043e-07,
      "loss": 0.3307,
      "step": 26038
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1268773078918457,
      "learning_rate": 5.441197303725321e-07,
      "loss": 0.3205,
      "step": 26039
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5945768356323242,
      "learning_rate": 5.437564936689665e-07,
      "loss": 0.2633,
      "step": 26040
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1679859161376953,
      "learning_rate": 5.433933748599407e-07,
      "loss": 0.3637,
      "step": 26041
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.998618245124817,
      "learning_rate": 5.430303739499787e-07,
      "loss": 0.3544,
      "step": 26042
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9158027172088623,
      "learning_rate": 5.426674909436047e-07,
      "loss": 0.2529,
      "step": 26043
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0597240924835205,
      "learning_rate": 5.423047258453463e-07,
      "loss": 0.2707,
      "step": 26044
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.735958456993103,
      "learning_rate": 5.41942078659724e-07,
      "loss": 0.3186,
      "step": 26045
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.3071744441986084,
      "learning_rate": 5.415795493912612e-07,
      "loss": 0.3063,
      "step": 26046
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1037724018096924,
      "learning_rate": 5.412171380444742e-07,
      "loss": 0.3143,
      "step": 26047
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.371403455734253,
      "learning_rate": 5.408548446238848e-07,
      "loss": 0.299,
      "step": 26048
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.5032846927642822,
      "learning_rate": 5.404926691340085e-07,
      "loss": 0.3927,
      "step": 26049
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.129206418991089,
      "learning_rate": 5.401306115793592e-07,
      "loss": 0.3461,
      "step": 26050
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.7861063480377197,
      "learning_rate": 5.397686719644535e-07,
      "loss": 0.2837,
      "step": 26051
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.1798746585845947,
      "learning_rate": 5.394068502938021e-07,
      "loss": 0.2789,
      "step": 26052
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.032752513885498,
      "learning_rate": 5.390451465719149e-07,
      "loss": 0.2344,
      "step": 26053
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0061376094818115,
      "learning_rate": 5.386835608033048e-07,
      "loss": 0.3614,
      "step": 26054
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0052366256713867,
      "learning_rate": 5.383220929924781e-07,
      "loss": 0.2642,
      "step": 26055
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.0215539932250977,
      "learning_rate": 5.379607431439394e-07,
      "loss": 0.2938,
      "step": 26056
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.571738839149475,
      "learning_rate": 5.375995112621979e-07,
      "loss": 0.2362,
      "step": 26057
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.898613452911377,
      "learning_rate": 5.372383973517537e-07,
      "loss": 0.2395,
      "step": 26058
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9065254926681519,
      "learning_rate": 5.368774014171108e-07,
      "loss": 0.2848,
      "step": 26059
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.11159348487854,
      "learning_rate": 5.365165234627679e-07,
      "loss": 0.3462,
      "step": 26060
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.064547061920166,
      "learning_rate": 5.361557634932257e-07,
      "loss": 0.3208,
      "step": 26061
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.4074506759643555,
      "learning_rate": 5.357951215129831e-07,
      "loss": 0.3165,
      "step": 26062
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.936100721359253,
      "learning_rate": 5.35434597526533e-07,
      "loss": 0.2881,
      "step": 26063
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8671332597732544,
      "learning_rate": 5.350741915383739e-07,
      "loss": 0.317,
      "step": 26064
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.782252311706543,
      "learning_rate": 5.347139035529969e-07,
      "loss": 0.3956,
      "step": 26065
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9434086084365845,
      "learning_rate": 5.343537335748927e-07,
      "loss": 0.2587,
      "step": 26066
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.088768482208252,
      "learning_rate": 5.339936816085556e-07,
      "loss": 0.2627,
      "step": 26067
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0024073123931885,
      "learning_rate": 5.336337476584719e-07,
      "loss": 0.3183,
      "step": 26068
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9724236726760864,
      "learning_rate": 5.332739317291269e-07,
      "loss": 0.3087,
      "step": 26069
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9307736158370972,
      "learning_rate": 5.329142338250115e-07,
      "loss": 0.3522,
      "step": 26070
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.74666166305542,
      "learning_rate": 5.325546539506076e-07,
      "loss": 0.2543,
      "step": 26071
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.422363042831421,
      "learning_rate": 5.321951921103985e-07,
      "loss": 0.3284,
      "step": 26072
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9769223928451538,
      "learning_rate": 5.318358483088637e-07,
      "loss": 0.3806,
      "step": 26073
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.023404121398926,
      "learning_rate": 5.314766225504864e-07,
      "loss": 0.3537,
      "step": 26074
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.135446548461914,
      "learning_rate": 5.31117514839744e-07,
      "loss": 0.299,
      "step": 26075
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9772236347198486,
      "learning_rate": 5.30758525181112e-07,
      "loss": 0.3862,
      "step": 26076
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5657681226730347,
      "learning_rate": 5.303996535790701e-07,
      "loss": 0.2563,
      "step": 26077
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.463913679122925,
      "learning_rate": 5.300409000380891e-07,
      "loss": 0.3281,
      "step": 26078
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.955920696258545,
      "learning_rate": 5.296822645626409e-07,
      "loss": 0.2562,
      "step": 26079
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8609052896499634,
      "learning_rate": 5.293237471572011e-07,
      "loss": 0.2448,
      "step": 26080
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8684427738189697,
      "learning_rate": 5.289653478262358e-07,
      "loss": 0.2433,
      "step": 26081
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.935400366783142,
      "learning_rate": 5.286070665742127e-07,
      "loss": 0.3037,
      "step": 26082
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9462474584579468,
      "learning_rate": 5.282489034056026e-07,
      "loss": 0.2851,
      "step": 26083
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9648221731185913,
      "learning_rate": 5.278908583248677e-07,
      "loss": 0.2813,
      "step": 26084
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7833712100982666,
      "learning_rate": 5.27532931336473e-07,
      "loss": 0.3189,
      "step": 26085
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.327699661254883,
      "learning_rate": 5.271751224448796e-07,
      "loss": 0.3702,
      "step": 26086
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8658487796783447,
      "learning_rate": 5.268174316545504e-07,
      "loss": 0.2577,
      "step": 26087
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9837111234664917,
      "learning_rate": 5.264598589699443e-07,
      "loss": 0.2731,
      "step": 26088
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8386554718017578,
      "learning_rate": 5.261024043955165e-07,
      "loss": 0.2269,
      "step": 26089
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7819288969039917,
      "learning_rate": 5.25745067935729e-07,
      "loss": 0.3374,
      "step": 26090
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.040999412536621,
      "learning_rate": 5.253878495950327e-07,
      "loss": 0.2699,
      "step": 26091
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8997570276260376,
      "learning_rate": 5.250307493778806e-07,
      "loss": 0.2838,
      "step": 26092
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.685242772102356,
      "learning_rate": 5.246737672887281e-07,
      "loss": 0.2416,
      "step": 26093
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0987603664398193,
      "learning_rate": 5.24316903332025e-07,
      "loss": 0.2532,
      "step": 26094
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9518625736236572,
      "learning_rate": 5.239601575122166e-07,
      "loss": 0.3404,
      "step": 26095
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0607640743255615,
      "learning_rate": 5.236035298337561e-07,
      "loss": 0.377,
      "step": 26096
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0385560989379883,
      "learning_rate": 5.232470203010875e-07,
      "loss": 0.2393,
      "step": 26097
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5255767107009888,
      "learning_rate": 5.228906289186541e-07,
      "loss": 0.2538,
      "step": 26098
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7845889329910278,
      "learning_rate": 5.225343556909002e-07,
      "loss": 0.3325,
      "step": 26099
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0264031887054443,
      "learning_rate": 5.221782006222687e-07,
      "loss": 0.3,
      "step": 26100
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.4959568977355957,
      "learning_rate": 5.218221637171983e-07,
      "loss": 0.3323,
      "step": 26101
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.092841863632202,
      "learning_rate": 5.214662449801277e-07,
      "loss": 0.3339,
      "step": 26102
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.408280372619629,
      "learning_rate": 5.211104444154968e-07,
      "loss": 0.2968,
      "step": 26103
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0233771800994873,
      "learning_rate": 5.207547620277398e-07,
      "loss": 0.2981,
      "step": 26104
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0627264976501465,
      "learning_rate": 5.203991978212885e-07,
      "loss": 0.2782,
      "step": 26105
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.463205337524414,
      "learning_rate": 5.200437518005808e-07,
      "loss": 0.2834,
      "step": 26106
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.415506601333618,
      "learning_rate": 5.196884239700462e-07,
      "loss": 0.3248,
      "step": 26107
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.1061625480651855,
      "learning_rate": 5.193332143341123e-07,
      "loss": 0.3614,
      "step": 26108
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8069936037063599,
      "learning_rate": 5.189781228972102e-07,
      "loss": 0.2492,
      "step": 26109
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.924398422241211,
      "learning_rate": 5.186231496637672e-07,
      "loss": 0.3484,
      "step": 26110
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.912227749824524,
      "learning_rate": 5.182682946382089e-07,
      "loss": 0.3568,
      "step": 26111
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7179137468338013,
      "learning_rate": 5.179135578249561e-07,
      "loss": 0.2868,
      "step": 26112
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5693963766098022,
      "learning_rate": 5.175589392284364e-07,
      "loss": 0.2823,
      "step": 26113
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8489762544631958,
      "learning_rate": 5.172044388530672e-07,
      "loss": 0.3337,
      "step": 26114
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8461270332336426,
      "learning_rate": 5.168500567032686e-07,
      "loss": 0.31,
      "step": 26115
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5098406076431274,
      "learning_rate": 5.164957927834613e-07,
      "loss": 0.2985,
      "step": 26116
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.010850667953491,
      "learning_rate": 5.161416470980607e-07,
      "loss": 0.2868,
      "step": 26117
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.822882056236267,
      "learning_rate": 5.157876196514799e-07,
      "loss": 0.35,
      "step": 26118
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7759770154953003,
      "learning_rate": 5.154337104481366e-07,
      "loss": 0.3464,
      "step": 26119
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9778589010238647,
      "learning_rate": 5.150799194924406e-07,
      "loss": 0.1896,
      "step": 26120
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8362237215042114,
      "learning_rate": 5.147262467888025e-07,
      "loss": 0.3025,
      "step": 26121
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.930024266242981,
      "learning_rate": 5.143726923416336e-07,
      "loss": 0.3177,
      "step": 26122
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.4316494464874268,
      "learning_rate": 5.140192561553403e-07,
      "loss": 0.2868,
      "step": 26123
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.6874117851257324,
      "learning_rate": 5.13665938234329e-07,
      "loss": 0.34,
      "step": 26124
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9637752771377563,
      "learning_rate": 5.13312738583004e-07,
      "loss": 0.3368,
      "step": 26125
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7811769247055054,
      "learning_rate": 5.129596572057716e-07,
      "loss": 0.3091,
      "step": 26126
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.7756268978118896,
      "learning_rate": 5.126066941070318e-07,
      "loss": 0.3535,
      "step": 26127
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.3553006649017334,
      "learning_rate": 5.122538492911843e-07,
      "loss": 0.2779,
      "step": 26128
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9386816024780273,
      "learning_rate": 5.119011227626291e-07,
      "loss": 0.3207,
      "step": 26129
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.926419973373413,
      "learning_rate": 5.11548514525766e-07,
      "loss": 0.317,
      "step": 26130
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.150937557220459,
      "learning_rate": 5.111960245849857e-07,
      "loss": 0.2692,
      "step": 26131
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0643310546875,
      "learning_rate": 5.108436529446881e-07,
      "loss": 0.2999,
      "step": 26132
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.949520230293274,
      "learning_rate": 5.104913996092642e-07,
      "loss": 0.2485,
      "step": 26133
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0896856784820557,
      "learning_rate": 5.101392645831049e-07,
      "loss": 0.3042,
      "step": 26134
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.4719207286834717,
      "learning_rate": 5.097872478706023e-07,
      "loss": 0.3822,
      "step": 26135
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.1458475589752197,
      "learning_rate": 5.09435349476144e-07,
      "loss": 0.2787,
      "step": 26136
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7387447357177734,
      "learning_rate": 5.090835694041174e-07,
      "loss": 0.3909,
      "step": 26137
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8781968355178833,
      "learning_rate": 5.08731907658907e-07,
      "loss": 0.319,
      "step": 26138
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9002091884613037,
      "learning_rate": 5.083803642448992e-07,
      "loss": 0.2074,
      "step": 26139
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.846786618232727,
      "learning_rate": 5.080289391664761e-07,
      "loss": 0.3072,
      "step": 26140
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0672800540924072,
      "learning_rate": 5.076776324280175e-07,
      "loss": 0.3039,
      "step": 26141
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.3435745239257812,
      "learning_rate": 5.073264440339065e-07,
      "loss": 0.3564,
      "step": 26142
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9694321155548096,
      "learning_rate": 5.069753739885197e-07,
      "loss": 0.2497,
      "step": 26143
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6819570064544678,
      "learning_rate": 5.066244222962324e-07,
      "loss": 0.2334,
      "step": 26144
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.1191067695617676,
      "learning_rate": 5.062735889614234e-07,
      "loss": 0.3121,
      "step": 26145
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.598421335220337,
      "learning_rate": 5.059228739884636e-07,
      "loss": 0.3023,
      "step": 26146
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.3007595539093018,
      "learning_rate": 5.055722773817262e-07,
      "loss": 0.3159,
      "step": 26147
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8518080711364746,
      "learning_rate": 5.052217991455844e-07,
      "loss": 0.3133,
      "step": 26148
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6349818706512451,
      "learning_rate": 5.048714392844067e-07,
      "loss": 0.2722,
      "step": 26149
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6529247760772705,
      "learning_rate": 5.045211978025599e-07,
      "loss": 0.2278,
      "step": 26150
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.4063310623168945,
      "learning_rate": 5.041710747044104e-07,
      "loss": 0.3598,
      "step": 26151
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6174917221069336,
      "learning_rate": 5.038210699943247e-07,
      "loss": 0.284,
      "step": 26152
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9375247955322266,
      "learning_rate": 5.03471183676667e-07,
      "loss": 0.3378,
      "step": 26153
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.086672782897949,
      "learning_rate": 5.031214157557962e-07,
      "loss": 0.3522,
      "step": 26154
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7434043884277344,
      "learning_rate": 5.027717662360776e-07,
      "loss": 0.3179,
      "step": 26155
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0040576457977295,
      "learning_rate": 5.024222351218666e-07,
      "loss": 0.3344,
      "step": 26156
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1815099716186523,
      "learning_rate": 5.020728224175219e-07,
      "loss": 0.3793,
      "step": 26157
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.734520196914673,
      "learning_rate": 5.017235281274024e-07,
      "loss": 0.2902,
      "step": 26158
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.3019144535064697,
      "learning_rate": 5.013743522558589e-07,
      "loss": 0.3547,
      "step": 26159
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1961593627929688,
      "learning_rate": 5.01025294807247e-07,
      "loss": 0.3452,
      "step": 26160
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0910117626190186,
      "learning_rate": 5.006763557859185e-07,
      "loss": 0.1683,
      "step": 26161
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.814603567123413,
      "learning_rate": 5.003275351962234e-07,
      "loss": 0.2901,
      "step": 26162
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6324946880340576,
      "learning_rate": 4.999788330425104e-07,
      "loss": 0.3171,
      "step": 26163
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.075532913208008,
      "learning_rate": 4.996302493291261e-07,
      "loss": 0.3966,
      "step": 26164
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9630625247955322,
      "learning_rate": 4.99281784060418e-07,
      "loss": 0.3628,
      "step": 26165
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8006024360656738,
      "learning_rate": 4.989334372407306e-07,
      "loss": 0.2485,
      "step": 26166
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.6199235916137695,
      "learning_rate": 4.985852088744037e-07,
      "loss": 0.3149,
      "step": 26167
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8518314361572266,
      "learning_rate": 4.982370989657836e-07,
      "loss": 0.3138,
      "step": 26168
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8192367553710938,
      "learning_rate": 4.978891075192072e-07,
      "loss": 0.241,
      "step": 26169
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.526435375213623,
      "learning_rate": 4.975412345390129e-07,
      "loss": 0.2961,
      "step": 26170
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.036404848098755,
      "learning_rate": 4.971934800295397e-07,
      "loss": 0.2815,
      "step": 26171
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.809212327003479,
      "learning_rate": 4.968458439951229e-07,
      "loss": 0.277,
      "step": 26172
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.487107992172241,
      "learning_rate": 4.964983264400935e-07,
      "loss": 0.2946,
      "step": 26173
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1179680824279785,
      "learning_rate": 4.961509273687881e-07,
      "loss": 0.3844,
      "step": 26174
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7563880681991577,
      "learning_rate": 4.958036467855365e-07,
      "loss": 0.3413,
      "step": 26175
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6798802614212036,
      "learning_rate": 4.954564846946685e-07,
      "loss": 0.2351,
      "step": 26176
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.2075939178466797,
      "learning_rate": 4.951094411005098e-07,
      "loss": 0.3776,
      "step": 26177
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6697274446487427,
      "learning_rate": 4.94762516007391e-07,
      "loss": 0.3169,
      "step": 26178
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7417285442352295,
      "learning_rate": 4.944157094196356e-07,
      "loss": 0.2685,
      "step": 26179
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.3302395343780518,
      "learning_rate": 4.940690213415655e-07,
      "loss": 0.3036,
      "step": 26180
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.308882474899292,
      "learning_rate": 4.937224517775063e-07,
      "loss": 0.2914,
      "step": 26181
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7930359840393066,
      "learning_rate": 4.933760007317778e-07,
      "loss": 0.1975,
      "step": 26182
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0594277381896973,
      "learning_rate": 4.930296682086966e-07,
      "loss": 0.261,
      "step": 26183
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8263726234436035,
      "learning_rate": 4.926834542125858e-07,
      "loss": 0.3052,
      "step": 26184
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.548901081085205,
      "learning_rate": 4.923373587477575e-07,
      "loss": 0.3054,
      "step": 26185
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.799456238746643,
      "learning_rate": 4.919913818185273e-07,
      "loss": 0.3532,
      "step": 26186
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0771307945251465,
      "learning_rate": 4.916455234292094e-07,
      "loss": 0.2951,
      "step": 26187
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5890216827392578,
      "learning_rate": 4.912997835841171e-07,
      "loss": 0.3053,
      "step": 26188
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.01218318939209,
      "learning_rate": 4.909541622875579e-07,
      "loss": 0.2052,
      "step": 26189
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.794913649559021,
      "learning_rate": 4.906086595438419e-07,
      "loss": 0.3391,
      "step": 26190
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8625110387802124,
      "learning_rate": 4.902632753572778e-07,
      "loss": 0.2457,
      "step": 26191
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.522730827331543,
      "learning_rate": 4.899180097321721e-07,
      "loss": 0.375,
      "step": 26192
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8674416542053223,
      "learning_rate": 4.895728626728247e-07,
      "loss": 0.274,
      "step": 26193
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8582861423492432,
      "learning_rate": 4.892278341835444e-07,
      "loss": 0.297,
      "step": 26194
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7400038242340088,
      "learning_rate": 4.888829242686311e-07,
      "loss": 0.3093,
      "step": 26195
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0543501377105713,
      "learning_rate": 4.885381329323824e-07,
      "loss": 0.3358,
      "step": 26196
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.5947670936584473,
      "learning_rate": 4.881934601790994e-07,
      "loss": 0.321,
      "step": 26197
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1713225841522217,
      "learning_rate": 4.878489060130797e-07,
      "loss": 0.284,
      "step": 26198
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7272199392318726,
      "learning_rate": 4.875044704386156e-07,
      "loss": 0.3167,
      "step": 26199
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8863706588745117,
      "learning_rate": 4.871601534600056e-07,
      "loss": 0.2577,
      "step": 26200
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.110142946243286,
      "learning_rate": 4.868159550815398e-07,
      "loss": 0.3219,
      "step": 26201
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.7441418170928955,
      "learning_rate": 4.864718753075115e-07,
      "loss": 0.3281,
      "step": 26202
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7566124200820923,
      "learning_rate": 4.86127914142207e-07,
      "loss": 0.3323,
      "step": 26203
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0679147243499756,
      "learning_rate": 4.857840715899176e-07,
      "loss": 0.2794,
      "step": 26204
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9228626489639282,
      "learning_rate": 4.854403476549297e-07,
      "loss": 0.3368,
      "step": 26205
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9210352897644043,
      "learning_rate": 4.850967423415265e-07,
      "loss": 0.3489,
      "step": 26206
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.5138771533966064,
      "learning_rate": 4.847532556539958e-07,
      "loss": 0.3305,
      "step": 26207
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.244365692138672,
      "learning_rate": 4.844098875966164e-07,
      "loss": 0.408,
      "step": 26208
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8786892890930176,
      "learning_rate": 4.840666381736703e-07,
      "loss": 0.2639,
      "step": 26209
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8339217901229858,
      "learning_rate": 4.837235073894375e-07,
      "loss": 0.3794,
      "step": 26210
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8261638879776,
      "learning_rate": 4.833804952481958e-07,
      "loss": 0.3276,
      "step": 26211
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.178149461746216,
      "learning_rate": 4.830376017542205e-07,
      "loss": 0.3153,
      "step": 26212
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.7257325649261475,
      "learning_rate": 4.826948269117881e-07,
      "loss": 0.4254,
      "step": 26213
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7418391704559326,
      "learning_rate": 4.823521707251722e-07,
      "loss": 0.3291,
      "step": 26214
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1103925704956055,
      "learning_rate": 4.820096331986434e-07,
      "loss": 0.2822,
      "step": 26215
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0310282707214355,
      "learning_rate": 4.81667214336472e-07,
      "loss": 0.2957,
      "step": 26216
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8324600458145142,
      "learning_rate": 4.813249141429288e-07,
      "loss": 0.3581,
      "step": 26217
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8036460876464844,
      "learning_rate": 4.809827326222816e-07,
      "loss": 0.3276,
      "step": 26218
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.146456718444824,
      "learning_rate": 4.806406697787936e-07,
      "loss": 0.3107,
      "step": 26219
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8692446947097778,
      "learning_rate": 4.802987256167324e-07,
      "loss": 0.268,
      "step": 26220
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.2323079109191895,
      "learning_rate": 4.799569001403603e-07,
      "loss": 0.3701,
      "step": 26221
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8480312824249268,
      "learning_rate": 4.796151933539373e-07,
      "loss": 0.2859,
      "step": 26222
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9165430068969727,
      "learning_rate": 4.792736052617275e-07,
      "loss": 0.2487,
      "step": 26223
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.143601894378662,
      "learning_rate": 4.789321358679854e-07,
      "loss": 0.3176,
      "step": 26224
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.339905261993408,
      "learning_rate": 4.785907851769699e-07,
      "loss": 0.3835,
      "step": 26225
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1120729446411133,
      "learning_rate": 4.782495531929376e-07,
      "loss": 0.3063,
      "step": 26226
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7091668844223022,
      "learning_rate": 4.779084399201417e-07,
      "loss": 0.2336,
      "step": 26227
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.177339553833008,
      "learning_rate": 4.775674453628365e-07,
      "loss": 0.3386,
      "step": 26228
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.932974934577942,
      "learning_rate": 4.772265695252698e-07,
      "loss": 0.3603,
      "step": 26229
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9492127895355225,
      "learning_rate": 4.768858124116949e-07,
      "loss": 0.3101,
      "step": 26230
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1276087760925293,
      "learning_rate": 4.765451740263594e-07,
      "loss": 0.2983,
      "step": 26231
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.5652377605438232,
      "learning_rate": 4.7620465437350774e-07,
      "loss": 0.3118,
      "step": 26232
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8077778816223145,
      "learning_rate": 4.7586425345738874e-07,
      "loss": 0.2127,
      "step": 26233
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.1238315105438232,
      "learning_rate": 4.755239712822457e-07,
      "loss": 0.2481,
      "step": 26234
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0772972106933594,
      "learning_rate": 4.751838078523174e-07,
      "loss": 0.236,
      "step": 26235
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.3682029247283936,
      "learning_rate": 4.7484376317185057e-07,
      "loss": 0.2323,
      "step": 26236
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.3603787422180176,
      "learning_rate": 4.7450383724508053e-07,
      "loss": 0.2971,
      "step": 26237
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.215991497039795,
      "learning_rate": 4.741640300762451e-07,
      "loss": 0.2586,
      "step": 26238
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.905869483947754,
      "learning_rate": 4.738243416695842e-07,
      "loss": 0.346,
      "step": 26239
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0793039798736572,
      "learning_rate": 4.734847720293301e-07,
      "loss": 0.2983,
      "step": 26240
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.250600814819336,
      "learning_rate": 4.731453211597181e-07,
      "loss": 0.2622,
      "step": 26241
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.3373169898986816,
      "learning_rate": 4.728059890649761e-07,
      "loss": 0.2969,
      "step": 26242
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.313188076019287,
      "learning_rate": 4.724667757493406e-07,
      "loss": 0.3256,
      "step": 26243
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6945586204528809,
      "learning_rate": 4.7212768121703725e-07,
      "loss": 0.2723,
      "step": 26244
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0851407051086426,
      "learning_rate": 4.717887054722925e-07,
      "loss": 0.3556,
      "step": 26245
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.9082382917404175,
      "learning_rate": 4.7144984851933637e-07,
      "loss": 0.3602,
      "step": 26246
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5858635902404785,
      "learning_rate": 4.7111111036239e-07,
      "loss": 0.313,
      "step": 26247
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.094728708267212,
      "learning_rate": 4.7077249100567767e-07,
      "loss": 0.3434,
      "step": 26248
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7319819927215576,
      "learning_rate": 4.7043399045342275e-07,
      "loss": 0.3088,
      "step": 26249
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.059382200241089,
      "learning_rate": 4.7009560870984406e-07,
      "loss": 0.3097,
      "step": 26250
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.0457570552825928,
      "learning_rate": 4.6975734577915935e-07,
      "loss": 0.2836,
      "step": 26251
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1556737422943115,
      "learning_rate": 4.694192016655874e-07,
      "loss": 0.2883,
      "step": 26252
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.225621223449707,
      "learning_rate": 4.6908117637334385e-07,
      "loss": 0.3265,
      "step": 26253
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.820481300354004,
      "learning_rate": 4.687432699066419e-07,
      "loss": 0.3312,
      "step": 26254
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0698933601379395,
      "learning_rate": 4.6840548226969483e-07,
      "loss": 0.3054,
      "step": 26255
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.199164390563965,
      "learning_rate": 4.6806781346671493e-07,
      "loss": 0.2752,
      "step": 26256
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.833587646484375,
      "learning_rate": 4.67730263501911e-07,
      "loss": 0.3432,
      "step": 26257
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8696309328079224,
      "learning_rate": 4.673928323794907e-07,
      "loss": 0.2273,
      "step": 26258
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9914348125457764,
      "learning_rate": 4.6705552010366307e-07,
      "loss": 0.2622,
      "step": 26259
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6982916593551636,
      "learning_rate": 4.6671832667863237e-07,
      "loss": 0.3185,
      "step": 26260
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.095820188522339,
      "learning_rate": 4.6638125210860086e-07,
      "loss": 0.3511,
      "step": 26261
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6133487224578857,
      "learning_rate": 4.6604429639777406e-07,
      "loss": 0.2809,
      "step": 26262
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0053136348724365,
      "learning_rate": 4.6570745955035193e-07,
      "loss": 0.233,
      "step": 26263
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0653152465820312,
      "learning_rate": 4.653707415705322e-07,
      "loss": 0.3119,
      "step": 26264
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9173524379730225,
      "learning_rate": 4.650341424625149e-07,
      "loss": 0.3053,
      "step": 26265
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9133614301681519,
      "learning_rate": 4.646976622304955e-07,
      "loss": 0.2693,
      "step": 26266
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1936185359954834,
      "learning_rate": 4.643613008786696e-07,
      "loss": 0.3661,
      "step": 26267
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.3706576824188232,
      "learning_rate": 4.6402505841122933e-07,
      "loss": 0.3404,
      "step": 26268
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0680172443389893,
      "learning_rate": 4.636889348323692e-07,
      "loss": 0.2737,
      "step": 26269
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9822698831558228,
      "learning_rate": 4.633529301462791e-07,
      "loss": 0.3109,
      "step": 26270
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8246325254440308,
      "learning_rate": 4.6301704435714466e-07,
      "loss": 0.3657,
      "step": 26271
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.573438882827759,
      "learning_rate": 4.6268127746915915e-07,
      "loss": 0.3038,
      "step": 26272
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.963319182395935,
      "learning_rate": 4.6234562948650474e-07,
      "loss": 0.2576,
      "step": 26273
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.458125352859497,
      "learning_rate": 4.6201010041336593e-07,
      "loss": 0.2826,
      "step": 26274
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.977442741394043,
      "learning_rate": 4.616746902539293e-07,
      "loss": 0.279,
      "step": 26275
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1497297286987305,
      "learning_rate": 4.613393990123738e-07,
      "loss": 0.2884,
      "step": 26276
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7195446491241455,
      "learning_rate": 4.6100422669287823e-07,
      "loss": 0.2411,
      "step": 26277
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8100396394729614,
      "learning_rate": 4.60669173299626e-07,
      "loss": 0.3067,
      "step": 26278
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8283860683441162,
      "learning_rate": 4.6033423883679043e-07,
      "loss": 0.2859,
      "step": 26279
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0434577465057373,
      "learning_rate": 4.599994233085492e-07,
      "loss": 0.294,
      "step": 26280
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.8797736167907715,
      "learning_rate": 4.5966472671907456e-07,
      "loss": 0.2627,
      "step": 26281
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1729037761688232,
      "learning_rate": 4.5933014907254105e-07,
      "loss": 0.3524,
      "step": 26282
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.4659907817840576,
      "learning_rate": 4.5899569037312077e-07,
      "loss": 0.4084,
      "step": 26283
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5712393522262573,
      "learning_rate": 4.586613506249804e-07,
      "loss": 0.2948,
      "step": 26284
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3849024772644043,
      "learning_rate": 4.583271298322911e-07,
      "loss": 0.3366,
      "step": 26285
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9555176496505737,
      "learning_rate": 4.579930279992195e-07,
      "loss": 0.259,
      "step": 26286
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7574056386947632,
      "learning_rate": 4.57659045129929e-07,
      "loss": 0.2691,
      "step": 26287
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.498058319091797,
      "learning_rate": 4.5732518122858616e-07,
      "loss": 0.3899,
      "step": 26288
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9411776065826416,
      "learning_rate": 4.5699143629935106e-07,
      "loss": 0.317,
      "step": 26289
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7961184978485107,
      "learning_rate": 4.5665781034638476e-07,
      "loss": 0.3237,
      "step": 26290
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2155914306640625,
      "learning_rate": 4.563243033738496e-07,
      "loss": 0.3361,
      "step": 26291
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2451274394989014,
      "learning_rate": 4.559909153858999e-07,
      "loss": 0.347,
      "step": 26292
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3217899799346924,
      "learning_rate": 4.556576463866946e-07,
      "loss": 0.2835,
      "step": 26293
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2377500534057617,
      "learning_rate": 4.55324496380386e-07,
      "loss": 0.3181,
      "step": 26294
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6311548948287964,
      "learning_rate": 4.549914653711318e-07,
      "loss": 0.2197,
      "step": 26295
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.841763973236084,
      "learning_rate": 4.5465855336308095e-07,
      "loss": 0.3128,
      "step": 26296
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2232933044433594,
      "learning_rate": 4.5432576036038233e-07,
      "loss": 0.2481,
      "step": 26297
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8352755308151245,
      "learning_rate": 4.539930863671904e-07,
      "loss": 0.234,
      "step": 26298
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2104904651641846,
      "learning_rate": 4.536605313876485e-07,
      "loss": 0.3225,
      "step": 26299
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5034408569335938,
      "learning_rate": 4.5332809542590227e-07,
      "loss": 0.2635,
      "step": 26300
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.4349868297576904,
      "learning_rate": 4.5299577848609943e-07,
      "loss": 0.2949,
      "step": 26301
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3001275062561035,
      "learning_rate": 4.5266358057238224e-07,
      "loss": 0.327,
      "step": 26302
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1044631004333496,
      "learning_rate": 4.5233150168888964e-07,
      "loss": 0.3426,
      "step": 26303
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.378901958465576,
      "learning_rate": 4.5199954183976493e-07,
      "loss": 0.3015,
      "step": 26304
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.939913272857666,
      "learning_rate": 4.5166770102914703e-07,
      "loss": 0.3261,
      "step": 26305
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6163593530654907,
      "learning_rate": 4.513359792611705e-07,
      "loss": 0.2724,
      "step": 26306
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2678329944610596,
      "learning_rate": 4.5100437653997074e-07,
      "loss": 0.4072,
      "step": 26307
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.434558868408203,
      "learning_rate": 4.5067289286968574e-07,
      "loss": 0.3738,
      "step": 26308
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9432010650634766,
      "learning_rate": 4.5034152825444545e-07,
      "loss": 0.372,
      "step": 26309
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6688475608825684,
      "learning_rate": 4.500102826983799e-07,
      "loss": 0.2752,
      "step": 26310
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8543627262115479,
      "learning_rate": 4.4967915620562243e-07,
      "loss": 0.2094,
      "step": 26311
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.992722749710083,
      "learning_rate": 4.4934814878029976e-07,
      "loss": 0.3143,
      "step": 26312
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6647686958312988,
      "learning_rate": 4.4901726042653746e-07,
      "loss": 0.2898,
      "step": 26313
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8209648132324219,
      "learning_rate": 4.486864911484623e-07,
      "loss": 0.3253,
      "step": 26314
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6032774448394775,
      "learning_rate": 4.4835584095019756e-07,
      "loss": 0.2706,
      "step": 26315
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1443355083465576,
      "learning_rate": 4.4802530983586445e-07,
      "loss": 0.2881,
      "step": 26316
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.6474292278289795,
      "learning_rate": 4.476948978095874e-07,
      "loss": 0.3792,
      "step": 26317
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.0124499797821045,
      "learning_rate": 4.4736460487548206e-07,
      "loss": 0.2348,
      "step": 26318
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.822577714920044,
      "learning_rate": 4.4703443103766843e-07,
      "loss": 0.3398,
      "step": 26319
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.157618284225464,
      "learning_rate": 4.467043763002599e-07,
      "loss": 0.261,
      "step": 26320
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1262993812561035,
      "learning_rate": 4.4637444066737536e-07,
      "loss": 0.3137,
      "step": 26321
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.32331919670105,
      "learning_rate": 4.460446241431271e-07,
      "loss": 0.3175,
      "step": 26322
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6409488916397095,
      "learning_rate": 4.457149267316241e-07,
      "loss": 0.2712,
      "step": 26323
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9096351861953735,
      "learning_rate": 4.4538534843698187e-07,
      "loss": 0.3058,
      "step": 26324
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7779725790023804,
      "learning_rate": 4.45055889263305e-07,
      "loss": 0.362,
      "step": 26325
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.929126262664795,
      "learning_rate": 4.447265492147024e-07,
      "loss": 0.258,
      "step": 26326
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9915697574615479,
      "learning_rate": 4.443973282952818e-07,
      "loss": 0.2408,
      "step": 26327
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5963661670684814,
      "learning_rate": 4.4406822650914563e-07,
      "loss": 0.2605,
      "step": 26328
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9861891269683838,
      "learning_rate": 4.4373924386039715e-07,
      "loss": 0.2936,
      "step": 26329
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.092942953109741,
      "learning_rate": 4.4341038035313865e-07,
      "loss": 0.3154,
      "step": 26330
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.0968751907348633,
      "learning_rate": 4.4308163599147025e-07,
      "loss": 0.3402,
      "step": 26331
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1120970249176025,
      "learning_rate": 4.4275301077949084e-07,
      "loss": 0.361,
      "step": 26332
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8458627462387085,
      "learning_rate": 4.424245047212949e-07,
      "loss": 0.3161,
      "step": 26333
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.11726713180542,
      "learning_rate": 4.420961178209804e-07,
      "loss": 0.3358,
      "step": 26334
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8911000490188599,
      "learning_rate": 4.417678500826428e-07,
      "loss": 0.3267,
      "step": 26335
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1181154251098633,
      "learning_rate": 4.4143970151036995e-07,
      "loss": 0.2919,
      "step": 26336
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7888177633285522,
      "learning_rate": 4.411116721082587e-07,
      "loss": 0.3133,
      "step": 26337
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.3689370155334473,
      "learning_rate": 4.407837618803956e-07,
      "loss": 0.3308,
      "step": 26338
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.906790018081665,
      "learning_rate": 4.4045597083086754e-07,
      "loss": 0.282,
      "step": 26339
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8888510465621948,
      "learning_rate": 4.4012829896376455e-07,
      "loss": 0.2342,
      "step": 26340
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9157180786132812,
      "learning_rate": 4.3980074628316996e-07,
      "loss": 0.3681,
      "step": 26341
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9983006715774536,
      "learning_rate": 4.3947331279316716e-07,
      "loss": 0.215,
      "step": 26342
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9314428567886353,
      "learning_rate": 4.391459984978397e-07,
      "loss": 0.4172,
      "step": 26343
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.9886102676391602,
      "learning_rate": 4.3881880340126747e-07,
      "loss": 0.2679,
      "step": 26344
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.1851882934570312,
      "learning_rate": 4.3849172750752956e-07,
      "loss": 0.3497,
      "step": 26345
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7194632291793823,
      "learning_rate": 4.3816477082070377e-07,
      "loss": 0.3456,
      "step": 26346
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8513237237930298,
      "learning_rate": 4.378379333448679e-07,
      "loss": 0.3069,
      "step": 26347
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7182201147079468,
      "learning_rate": 4.3751121508409435e-07,
      "loss": 0.2542,
      "step": 26348
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.35870623588562,
      "learning_rate": 4.3718461604245754e-07,
      "loss": 0.3102,
      "step": 26349
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8077949285507202,
      "learning_rate": 4.3685813622402986e-07,
      "loss": 0.3047,
      "step": 26350
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0887258052825928,
      "learning_rate": 4.365317756328824e-07,
      "loss": 0.3353,
      "step": 26351
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9658153057098389,
      "learning_rate": 4.362055342730809e-07,
      "loss": 0.3406,
      "step": 26352
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.514807939529419,
      "learning_rate": 4.3587941214869536e-07,
      "loss": 0.2922,
      "step": 26353
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1171388626098633,
      "learning_rate": 4.355534092637914e-07,
      "loss": 0.3505,
      "step": 26354
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.829376459121704,
      "learning_rate": 4.3522752562243145e-07,
      "loss": 0.3263,
      "step": 26355
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.467142939567566,
      "learning_rate": 4.349017612286821e-07,
      "loss": 0.3024,
      "step": 26356
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.547434091567993,
      "learning_rate": 4.3457611608660135e-07,
      "loss": 0.3329,
      "step": 26357
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.302868127822876,
      "learning_rate": 4.342505902002514e-07,
      "loss": 0.3444,
      "step": 26358
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0730957984924316,
      "learning_rate": 4.3392518357368797e-07,
      "loss": 0.2338,
      "step": 26359
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2677688598632812,
      "learning_rate": 4.335998962109711e-07,
      "loss": 0.2867,
      "step": 26360
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9499386548995972,
      "learning_rate": 4.3327472811615533e-07,
      "loss": 0.3014,
      "step": 26361
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.100860357284546,
      "learning_rate": 4.329496792932919e-07,
      "loss": 0.3175,
      "step": 26362
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0229389667510986,
      "learning_rate": 4.3262474974643867e-07,
      "loss": 0.3036,
      "step": 26363
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2084004878997803,
      "learning_rate": 4.322999394796423e-07,
      "loss": 0.3155,
      "step": 26364
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.174456834793091,
      "learning_rate": 4.3197524849695307e-07,
      "loss": 0.3421,
      "step": 26365
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.5904626846313477,
      "learning_rate": 4.3165067680241977e-07,
      "loss": 0.3268,
      "step": 26366
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.46236252784729,
      "learning_rate": 4.313262244000904e-07,
      "loss": 0.3089,
      "step": 26367
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.323355197906494,
      "learning_rate": 4.3100189129400615e-07,
      "loss": 0.3045,
      "step": 26368
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2305359840393066,
      "learning_rate": 4.306776774882149e-07,
      "loss": 0.2348,
      "step": 26369
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6415115594863892,
      "learning_rate": 4.3035358298675776e-07,
      "loss": 0.2918,
      "step": 26370
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.972733497619629,
      "learning_rate": 4.3002960779367167e-07,
      "loss": 0.2461,
      "step": 26371
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0130250453948975,
      "learning_rate": 4.297057519129999e-07,
      "loss": 0.2824,
      "step": 26372
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.4469125270843506,
      "learning_rate": 4.293820153487782e-07,
      "loss": 0.315,
      "step": 26373
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6884210109710693,
      "learning_rate": 4.290583981050422e-07,
      "loss": 0.2948,
      "step": 26374
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.005974292755127,
      "learning_rate": 4.2873490018582875e-07,
      "loss": 0.3335,
      "step": 26375
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8749874830245972,
      "learning_rate": 4.28411521595169e-07,
      "loss": 0.2174,
      "step": 26376
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7398960590362549,
      "learning_rate": 4.280882623370963e-07,
      "loss": 0.2649,
      "step": 26377
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2839443683624268,
      "learning_rate": 4.277651224156387e-07,
      "loss": 0.3226,
      "step": 26378
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9375276565551758,
      "learning_rate": 4.2744210183482627e-07,
      "loss": 0.3854,
      "step": 26379
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.116692543029785,
      "learning_rate": 4.2711920059868684e-07,
      "loss": 0.3335,
      "step": 26380
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9064496755599976,
      "learning_rate": 4.26796418711245e-07,
      "loss": 0.2759,
      "step": 26381
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.3177857398986816,
      "learning_rate": 4.264737561765253e-07,
      "loss": 0.2742,
      "step": 26382
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9251372814178467,
      "learning_rate": 4.261512129985512e-07,
      "loss": 0.2425,
      "step": 26383
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9875500202178955,
      "learning_rate": 4.2582878918134176e-07,
      "loss": 0.3058,
      "step": 26384
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.968375563621521,
      "learning_rate": 4.255064847289203e-07,
      "loss": 0.2588,
      "step": 26385
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0939958095550537,
      "learning_rate": 4.251842996453026e-07,
      "loss": 0.3163,
      "step": 26386
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9558967351913452,
      "learning_rate": 4.248622339345043e-07,
      "loss": 0.3446,
      "step": 26387
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7802180051803589,
      "learning_rate": 4.2454028760054445e-07,
      "loss": 0.3308,
      "step": 26388
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.319776773452759,
      "learning_rate": 4.2421846064743423e-07,
      "loss": 0.302,
      "step": 26389
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.166379451751709,
      "learning_rate": 4.238967530791871e-07,
      "loss": 0.2918,
      "step": 26390
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0994598865509033,
      "learning_rate": 4.235751648998121e-07,
      "loss": 0.3751,
      "step": 26391
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1026523113250732,
      "learning_rate": 4.2325369611332046e-07,
      "loss": 0.3076,
      "step": 26392
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.264835834503174,
      "learning_rate": 4.2293234672372007e-07,
      "loss": 0.2659,
      "step": 26393
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.923464059829712,
      "learning_rate": 4.2261111673501556e-07,
      "loss": 0.2973,
      "step": 26394
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2431938648223877,
      "learning_rate": 4.2229000615121473e-07,
      "loss": 0.3558,
      "step": 26395
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9439659118652344,
      "learning_rate": 4.219690149763189e-07,
      "loss": 0.2895,
      "step": 26396
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9253191947937012,
      "learning_rate": 4.2164814321432825e-07,
      "loss": 0.2989,
      "step": 26397
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1956775188446045,
      "learning_rate": 4.213273908692472e-07,
      "loss": 0.3287,
      "step": 26398
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.506009101867676,
      "learning_rate": 4.2100675794507275e-07,
      "loss": 0.4022,
      "step": 26399
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5682871341705322,
      "learning_rate": 4.206862444458015e-07,
      "loss": 0.2767,
      "step": 26400
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.078462839126587,
      "learning_rate": 4.2036585037543156e-07,
      "loss": 0.3233,
      "step": 26401
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7066068649291992,
      "learning_rate": 4.200455757379551e-07,
      "loss": 0.2336,
      "step": 26402
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.3711252212524414,
      "learning_rate": 4.197254205373669e-07,
      "loss": 0.3865,
      "step": 26403
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1130833625793457,
      "learning_rate": 4.1940538477765693e-07,
      "loss": 0.3047,
      "step": 26404
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2243363857269287,
      "learning_rate": 4.1908546846281653e-07,
      "loss": 0.3281,
      "step": 26405
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.5664398670196533,
      "learning_rate": 4.1876567159683355e-07,
      "loss": 0.3372,
      "step": 26406
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.964614987373352,
      "learning_rate": 4.184459941836949e-07,
      "loss": 0.3157,
      "step": 26407
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7273285388946533,
      "learning_rate": 4.1812643622738625e-07,
      "loss": 0.3205,
      "step": 26408
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.396552324295044,
      "learning_rate": 4.1780699773189324e-07,
      "loss": 0.3913,
      "step": 26409
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.2370660305023193,
      "learning_rate": 4.174876787011939e-07,
      "loss": 0.3885,
      "step": 26410
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0418381690979004,
      "learning_rate": 4.1716847913927494e-07,
      "loss": 0.2766,
      "step": 26411
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.730389952659607,
      "learning_rate": 4.1684939905011324e-07,
      "loss": 0.3326,
      "step": 26412
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.678581953048706,
      "learning_rate": 4.1653043843768557e-07,
      "loss": 0.2866,
      "step": 26413
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.678540587425232,
      "learning_rate": 4.1621159730597103e-07,
      "loss": 0.2929,
      "step": 26414
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9066218137741089,
      "learning_rate": 4.158928756589431e-07,
      "loss": 0.3086,
      "step": 26415
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7551459074020386,
      "learning_rate": 4.1557427350057635e-07,
      "loss": 0.3032,
      "step": 26416
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1630046367645264,
      "learning_rate": 4.1525579083484203e-07,
      "loss": 0.3584,
      "step": 26417
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.03971266746521,
      "learning_rate": 4.1493742766571143e-07,
      "loss": 0.2463,
      "step": 26418
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6904044151306152,
      "learning_rate": 4.146191839971547e-07,
      "loss": 0.2334,
      "step": 26419
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9785386323928833,
      "learning_rate": 4.143010598331354e-07,
      "loss": 0.3294,
      "step": 26420
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.028441905975342,
      "learning_rate": 4.1398305517762585e-07,
      "loss": 0.29,
      "step": 26421
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.28408145904541,
      "learning_rate": 4.136651700345862e-07,
      "loss": 0.3938,
      "step": 26422
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0907795429229736,
      "learning_rate": 4.133474044079799e-07,
      "loss": 0.2986,
      "step": 26423
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0501551628112793,
      "learning_rate": 4.130297583017717e-07,
      "loss": 0.3034,
      "step": 26424
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.015824794769287,
      "learning_rate": 4.127122317199195e-07,
      "loss": 0.3186,
      "step": 26425
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7689849138259888,
      "learning_rate": 4.1239482466638e-07,
      "loss": 0.2778,
      "step": 26426
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.3654239177703857,
      "learning_rate": 4.120775371451158e-07,
      "loss": 0.3079,
      "step": 26427
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.1584532260894775,
      "learning_rate": 4.117603691600791e-07,
      "loss": 0.2495,
      "step": 26428
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0928194522857666,
      "learning_rate": 4.114433207152235e-07,
      "loss": 0.3154,
      "step": 26429
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.3101093769073486,
      "learning_rate": 4.1112639181450253e-07,
      "loss": 0.4003,
      "step": 26430
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.676604986190796,
      "learning_rate": 4.108095824618696e-07,
      "loss": 0.3139,
      "step": 26431
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7290948629379272,
      "learning_rate": 4.1049289266127277e-07,
      "loss": 0.313,
      "step": 26432
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7384587526321411,
      "learning_rate": 4.101763224166588e-07,
      "loss": 0.272,
      "step": 26433
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9146238565444946,
      "learning_rate": 4.098598717319768e-07,
      "loss": 0.3735,
      "step": 26434
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.827216625213623,
      "learning_rate": 4.095435406111714e-07,
      "loss": 0.2517,
      "step": 26435
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7197197675704956,
      "learning_rate": 4.092273290581861e-07,
      "loss": 0.2872,
      "step": 26436
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7029451131820679,
      "learning_rate": 4.0891123707696434e-07,
      "loss": 0.2844,
      "step": 26437
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6101731061935425,
      "learning_rate": 4.0859526467144526e-07,
      "loss": 0.3183,
      "step": 26438
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7427527904510498,
      "learning_rate": 4.08279411845568e-07,
      "loss": 0.2902,
      "step": 26439
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7434005737304688,
      "learning_rate": 4.079636786032737e-07,
      "loss": 0.3479,
      "step": 26440
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9252171516418457,
      "learning_rate": 4.0764806494849483e-07,
      "loss": 0.3117,
      "step": 26441
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9853968620300293,
      "learning_rate": 4.0733257088516943e-07,
      "loss": 0.3596,
      "step": 26442
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.9116097688674927,
      "learning_rate": 4.0701719641722757e-07,
      "loss": 0.3485,
      "step": 26443
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.0183019638061523,
      "learning_rate": 4.06701941548604e-07,
      "loss": 0.3278,
      "step": 26444
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.2285139560699463,
      "learning_rate": 4.063868062832277e-07,
      "loss": 0.3353,
      "step": 26445
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9917094707489014,
      "learning_rate": 4.060717906250267e-07,
      "loss": 0.2537,
      "step": 26446
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.858188509941101,
      "learning_rate": 4.057568945779311e-07,
      "loss": 0.2402,
      "step": 26447
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.4733376502990723,
      "learning_rate": 4.0544211814586565e-07,
      "loss": 0.3398,
      "step": 26448
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.680964469909668,
      "learning_rate": 4.051274613327527e-07,
      "loss": 0.231,
      "step": 26449
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.9198999404907227,
      "learning_rate": 4.0481292414251917e-07,
      "loss": 0.3788,
      "step": 26450
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.314746379852295,
      "learning_rate": 4.0449850657908406e-07,
      "loss": 0.2876,
      "step": 26451
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9649300575256348,
      "learning_rate": 4.041842086463654e-07,
      "loss": 0.2601,
      "step": 26452
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.7556490898132324,
      "learning_rate": 4.0387003034828563e-07,
      "loss": 0.3093,
      "step": 26453
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1557412147521973,
      "learning_rate": 4.035559716887605e-07,
      "loss": 0.3273,
      "step": 26454
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.347791910171509,
      "learning_rate": 4.032420326717046e-07,
      "loss": 0.2646,
      "step": 26455
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8558001518249512,
      "learning_rate": 4.029282133010315e-07,
      "loss": 0.2414,
      "step": 26456
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.261298894882202,
      "learning_rate": 4.0261451358065473e-07,
      "loss": 0.2486,
      "step": 26457
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.4651713371276855,
      "learning_rate": 4.0230093351448563e-07,
      "loss": 0.3228,
      "step": 26458
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0812532901763916,
      "learning_rate": 4.0198747310643213e-07,
      "loss": 0.2467,
      "step": 26459
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0360777378082275,
      "learning_rate": 4.0167413236040455e-07,
      "loss": 0.3432,
      "step": 26460
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.047696113586426,
      "learning_rate": 4.013609112803074e-07,
      "loss": 0.2853,
      "step": 26461
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0352182388305664,
      "learning_rate": 4.0104780987004657e-07,
      "loss": 0.3707,
      "step": 26462
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0768113136291504,
      "learning_rate": 4.007348281335255e-07,
      "loss": 0.3079,
      "step": 26463
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0620243549346924,
      "learning_rate": 4.0042196607464665e-07,
      "loss": 0.3301,
      "step": 26464
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.893426537513733,
      "learning_rate": 4.001092236973092e-07,
      "loss": 0.3025,
      "step": 26465
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.3919506072998047,
      "learning_rate": 3.997966010054144e-07,
      "loss": 0.3597,
      "step": 26466
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9981211423873901,
      "learning_rate": 3.994840980028591e-07,
      "loss": 0.3072,
      "step": 26467
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0628275871276855,
      "learning_rate": 3.991717146935381e-07,
      "loss": 0.3167,
      "step": 26468
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7129175662994385,
      "learning_rate": 3.9885945108134595e-07,
      "loss": 0.2958,
      "step": 26469
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.974871039390564,
      "learning_rate": 3.985473071701784e-07,
      "loss": 0.316,
      "step": 26470
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.804726243019104,
      "learning_rate": 3.9823528296392464e-07,
      "loss": 0.2936,
      "step": 26471
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9401850700378418,
      "learning_rate": 3.979233784664749e-07,
      "loss": 0.2913,
      "step": 26472
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.873257040977478,
      "learning_rate": 3.976115936817193e-07,
      "loss": 0.2935,
      "step": 26473
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.061776876449585,
      "learning_rate": 3.972999286135437e-07,
      "loss": 0.2729,
      "step": 26474
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9297376871109009,
      "learning_rate": 3.969883832658339e-07,
      "loss": 0.3337,
      "step": 26475
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7996673583984375,
      "learning_rate": 3.966769576424745e-07,
      "loss": 0.296,
      "step": 26476
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0020196437835693,
      "learning_rate": 3.9636565174734796e-07,
      "loss": 0.2818,
      "step": 26477
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.040419340133667,
      "learning_rate": 3.9605446558433457e-07,
      "loss": 0.2267,
      "step": 26478
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1361570358276367,
      "learning_rate": 3.9574339915731675e-07,
      "loss": 0.3379,
      "step": 26479
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8461673259735107,
      "learning_rate": 3.9543245247017024e-07,
      "loss": 0.2862,
      "step": 26480
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.019580841064453,
      "learning_rate": 3.9512162552677204e-07,
      "loss": 0.2674,
      "step": 26481
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.3769052028656006,
      "learning_rate": 3.9481091833099673e-07,
      "loss": 0.3955,
      "step": 26482
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8340984582901,
      "learning_rate": 3.945003308867201e-07,
      "loss": 0.2897,
      "step": 26483
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0489797592163086,
      "learning_rate": 3.941898631978125e-07,
      "loss": 0.2865,
      "step": 26484
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.108043909072876,
      "learning_rate": 3.938795152681452e-07,
      "loss": 0.3329,
      "step": 26485
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.4031009674072266,
      "learning_rate": 3.9356928710158726e-07,
      "loss": 0.3445,
      "step": 26486
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8012523651123047,
      "learning_rate": 3.9325917870200793e-07,
      "loss": 0.2479,
      "step": 26487
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0880351066589355,
      "learning_rate": 3.929491900732707e-07,
      "loss": 0.2883,
      "step": 26488
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.027609348297119,
      "learning_rate": 3.9263932121924254e-07,
      "loss": 0.2451,
      "step": 26489
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8938981294631958,
      "learning_rate": 3.9232957214378586e-07,
      "loss": 0.311,
      "step": 26490
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9410215616226196,
      "learning_rate": 3.9201994285076205e-07,
      "loss": 0.2698,
      "step": 26491
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.168735980987549,
      "learning_rate": 3.917104333440325e-07,
      "loss": 0.2506,
      "step": 26492
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.2740988731384277,
      "learning_rate": 3.914010436274562e-07,
      "loss": 0.3142,
      "step": 26493
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8742297887802124,
      "learning_rate": 3.91091773704888e-07,
      "loss": 0.2616,
      "step": 26494
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.18929123878479,
      "learning_rate": 3.907826235801848e-07,
      "loss": 0.3403,
      "step": 26495
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.995370864868164,
      "learning_rate": 3.9047359325720345e-07,
      "loss": 0.2469,
      "step": 26496
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9472942352294922,
      "learning_rate": 3.901646827397931e-07,
      "loss": 0.3176,
      "step": 26497
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.35147762298584,
      "learning_rate": 3.898558920318063e-07,
      "loss": 0.2709,
      "step": 26498
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.626311779022217,
      "learning_rate": 3.895472211370943e-07,
      "loss": 0.3205,
      "step": 26499
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7478011846542358,
      "learning_rate": 3.892386700595041e-07,
      "loss": 0.2578,
      "step": 26500
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.830566167831421,
      "learning_rate": 3.889302388028804e-07,
      "loss": 0.2518,
      "step": 26501
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0297250747680664,
      "learning_rate": 3.886219273710734e-07,
      "loss": 0.3603,
      "step": 26502
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8640135526657104,
      "learning_rate": 3.883137357679234e-07,
      "loss": 0.3326,
      "step": 26503
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1900901794433594,
      "learning_rate": 3.88005663997274e-07,
      "loss": 0.3429,
      "step": 26504
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.130537509918213,
      "learning_rate": 3.876977120629655e-07,
      "loss": 0.3619,
      "step": 26505
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8613110780715942,
      "learning_rate": 3.8738987996883805e-07,
      "loss": 0.2226,
      "step": 26506
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7465039491653442,
      "learning_rate": 3.8708216771872865e-07,
      "loss": 0.2257,
      "step": 26507
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1352479457855225,
      "learning_rate": 3.867745753164731e-07,
      "loss": 0.2637,
      "step": 26508
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.124565839767456,
      "learning_rate": 3.8646710276590727e-07,
      "loss": 0.3159,
      "step": 26509
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1681625843048096,
      "learning_rate": 3.861597500708658e-07,
      "loss": 0.2995,
      "step": 26510
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8691445589065552,
      "learning_rate": 3.858525172351768e-07,
      "loss": 0.3012,
      "step": 26511
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.073357105255127,
      "learning_rate": 3.8554540426267493e-07,
      "loss": 0.2941,
      "step": 26512
    },
    {
      "epoch": 2.74,
      "grad_norm": 3.1841306686401367,
      "learning_rate": 3.8523841115718717e-07,
      "loss": 0.332,
      "step": 26513
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7123045921325684,
      "learning_rate": 3.8493153792253933e-07,
      "loss": 0.2468,
      "step": 26514
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1749444007873535,
      "learning_rate": 3.846247845625606e-07,
      "loss": 0.4055,
      "step": 26515
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.738765239715576,
      "learning_rate": 3.843181510810723e-07,
      "loss": 0.2713,
      "step": 26516
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.222987413406372,
      "learning_rate": 3.840116374818992e-07,
      "loss": 0.2053,
      "step": 26517
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.124314308166504,
      "learning_rate": 3.8370524376886265e-07,
      "loss": 0.3935,
      "step": 26518
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8420329093933105,
      "learning_rate": 3.8339896994578187e-07,
      "loss": 0.2716,
      "step": 26519
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.1584701538085938,
      "learning_rate": 3.83092816016476e-07,
      "loss": 0.3345,
      "step": 26520
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0513699054718018,
      "learning_rate": 3.8278678198476085e-07,
      "loss": 0.2867,
      "step": 26521
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.105093002319336,
      "learning_rate": 3.8248086785445337e-07,
      "loss": 0.3127,
      "step": 26522
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8287845849990845,
      "learning_rate": 3.8217507362936615e-07,
      "loss": 0.3177,
      "step": 26523
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.5789151191711426,
      "learning_rate": 3.8186939931331157e-07,
      "loss": 0.3614,
      "step": 26524
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9228966236114502,
      "learning_rate": 3.8156384491010223e-07,
      "loss": 0.3652,
      "step": 26525
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7319732904434204,
      "learning_rate": 3.812584104235473e-07,
      "loss": 0.2849,
      "step": 26526
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.038026809692383,
      "learning_rate": 3.8095309585745255e-07,
      "loss": 0.3084,
      "step": 26527
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7181962728500366,
      "learning_rate": 3.8064790121562613e-07,
      "loss": 0.243,
      "step": 26528
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.248170852661133,
      "learning_rate": 3.803428265018738e-07,
      "loss": 0.3319,
      "step": 26529
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7916361093521118,
      "learning_rate": 3.800378717199971e-07,
      "loss": 0.2484,
      "step": 26530
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8371689319610596,
      "learning_rate": 3.797330368737995e-07,
      "loss": 0.3233,
      "step": 26531
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.231624126434326,
      "learning_rate": 3.7942832196708134e-07,
      "loss": 0.3603,
      "step": 26532
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9109779596328735,
      "learning_rate": 3.7912372700364075e-07,
      "loss": 0.2243,
      "step": 26533
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.868552327156067,
      "learning_rate": 3.7881925198727464e-07,
      "loss": 0.305,
      "step": 26534
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.292816162109375,
      "learning_rate": 3.785148969217811e-07,
      "loss": 0.3144,
      "step": 26535
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8018875122070312,
      "learning_rate": 3.782106618109538e-07,
      "loss": 0.3093,
      "step": 26536
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.978912591934204,
      "learning_rate": 3.7790654665858515e-07,
      "loss": 0.2709,
      "step": 26537
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.069175958633423,
      "learning_rate": 3.776025514684678e-07,
      "loss": 0.3616,
      "step": 26538
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.052335500717163,
      "learning_rate": 3.7729867624439085e-07,
      "loss": 0.3391,
      "step": 26539
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.69602108001709,
      "learning_rate": 3.7699492099014244e-07,
      "loss": 0.3957,
      "step": 26540
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.273118019104004,
      "learning_rate": 3.7669128570951066e-07,
      "loss": 0.3585,
      "step": 26541
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0527117252349854,
      "learning_rate": 3.7638777040628127e-07,
      "loss": 0.3134,
      "step": 26542
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.238600254058838,
      "learning_rate": 3.760843750842358e-07,
      "loss": 0.3075,
      "step": 26543
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9252536296844482,
      "learning_rate": 3.7578109974716004e-07,
      "loss": 0.2948,
      "step": 26544
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0349972248077393,
      "learning_rate": 3.754779443988332e-07,
      "loss": 0.2839,
      "step": 26545
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6631293296813965,
      "learning_rate": 3.7517490904303564e-07,
      "loss": 0.2641,
      "step": 26546
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7814698219299316,
      "learning_rate": 3.748719936835443e-07,
      "loss": 0.3206,
      "step": 26547
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9114065170288086,
      "learning_rate": 3.745691983241373e-07,
      "loss": 0.3112,
      "step": 26548
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1044089794158936,
      "learning_rate": 3.7426652296858933e-07,
      "loss": 0.3873,
      "step": 26549
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1637043952941895,
      "learning_rate": 3.739639676206719e-07,
      "loss": 0.3015,
      "step": 26550
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9786992073059082,
      "learning_rate": 3.7366153228415967e-07,
      "loss": 0.3407,
      "step": 26551
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7924901247024536,
      "learning_rate": 3.7335921696282195e-07,
      "loss": 0.2979,
      "step": 26552
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3486599922180176,
      "learning_rate": 3.730570216604279e-07,
      "loss": 0.2663,
      "step": 26553
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.064568281173706,
      "learning_rate": 3.727549463807456e-07,
      "loss": 0.3176,
      "step": 26554
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9139230251312256,
      "learning_rate": 3.7245299112754207e-07,
      "loss": 0.2955,
      "step": 26555
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.733803153038025,
      "learning_rate": 3.7215115590457763e-07,
      "loss": 0.2645,
      "step": 26556
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.136650562286377,
      "learning_rate": 3.7184944071562147e-07,
      "loss": 0.2862,
      "step": 26557
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9491742849349976,
      "learning_rate": 3.715478455644306e-07,
      "loss": 0.3229,
      "step": 26558
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1242947578430176,
      "learning_rate": 3.7124637045476755e-07,
      "loss": 0.3094,
      "step": 26559
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7701411247253418,
      "learning_rate": 3.7094501539038816e-07,
      "loss": 0.304,
      "step": 26560
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1775763034820557,
      "learning_rate": 3.70643780375054e-07,
      "loss": 0.2537,
      "step": 26561
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.255244493484497,
      "learning_rate": 3.703426654125164e-07,
      "loss": 0.2696,
      "step": 26562
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8691699504852295,
      "learning_rate": 3.700416705065313e-07,
      "loss": 0.3029,
      "step": 26563
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.946044921875,
      "learning_rate": 3.697407956608512e-07,
      "loss": 0.3006,
      "step": 26564
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8913638591766357,
      "learning_rate": 3.694400408792276e-07,
      "loss": 0.2637,
      "step": 26565
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9841763973236084,
      "learning_rate": 3.6913940616540854e-07,
      "loss": 0.3328,
      "step": 26566
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.124884843826294,
      "learning_rate": 3.6883889152314445e-07,
      "loss": 0.3489,
      "step": 26567
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0891847610473633,
      "learning_rate": 3.685384969561812e-07,
      "loss": 0.2896,
      "step": 26568
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7287770509719849,
      "learning_rate": 3.6823822246826234e-07,
      "loss": 0.2958,
      "step": 26569
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.033694267272949,
      "learning_rate": 3.6793806806313395e-07,
      "loss": 0.3016,
      "step": 26570
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.636876106262207,
      "learning_rate": 3.6763803374453623e-07,
      "loss": 0.3233,
      "step": 26571
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.874885082244873,
      "learning_rate": 3.673381195162107e-07,
      "loss": 0.2638,
      "step": 26572
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7724027633666992,
      "learning_rate": 3.6703832538189544e-07,
      "loss": 0.2898,
      "step": 26573
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.861047625541687,
      "learning_rate": 3.667386513453308e-07,
      "loss": 0.3325,
      "step": 26574
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.740260124206543,
      "learning_rate": 3.664390974102505e-07,
      "loss": 0.3183,
      "step": 26575
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.754984736442566,
      "learning_rate": 3.661396635803882e-07,
      "loss": 0.2405,
      "step": 26576
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.023021697998047,
      "learning_rate": 3.658403498594798e-07,
      "loss": 0.341,
      "step": 26577
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.844970464706421,
      "learning_rate": 3.6554115625125675e-07,
      "loss": 0.2959,
      "step": 26578
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9715516567230225,
      "learning_rate": 3.652420827594461e-07,
      "loss": 0.2722,
      "step": 26579
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.22697377204895,
      "learning_rate": 3.649431293877803e-07,
      "loss": 0.2693,
      "step": 26580
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.2338054180145264,
      "learning_rate": 3.646442961399854e-07,
      "loss": 0.3186,
      "step": 26581
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.146959066390991,
      "learning_rate": 3.6434558301978396e-07,
      "loss": 0.3559,
      "step": 26582
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3472540378570557,
      "learning_rate": 3.640469900309052e-07,
      "loss": 0.2645,
      "step": 26583
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.889775037765503,
      "learning_rate": 3.6374851717706936e-07,
      "loss": 0.3076,
      "step": 26584
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.093775987625122,
      "learning_rate": 3.6345016446199697e-07,
      "loss": 0.3281,
      "step": 26585
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5951762199401855,
      "learning_rate": 3.6315193188940724e-07,
      "loss": 0.3034,
      "step": 26586
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9832489490509033,
      "learning_rate": 3.6285381946302043e-07,
      "loss": 0.3034,
      "step": 26587
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.180842638015747,
      "learning_rate": 3.625558271865526e-07,
      "loss": 0.3288,
      "step": 26588
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3691258430480957,
      "learning_rate": 3.622579550637173e-07,
      "loss": 0.3115,
      "step": 26589
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6547800302505493,
      "learning_rate": 3.6196020309823056e-07,
      "loss": 0.2807,
      "step": 26590
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.928521752357483,
      "learning_rate": 3.616625712938038e-07,
      "loss": 0.3088,
      "step": 26591
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7063589096069336,
      "learning_rate": 3.6136505965414627e-07,
      "loss": 0.2495,
      "step": 26592
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8708305358886719,
      "learning_rate": 3.6106766818296833e-07,
      "loss": 0.3166,
      "step": 26593
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.4307749271392822,
      "learning_rate": 3.607703968839793e-07,
      "loss": 0.3621,
      "step": 26594
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8602964878082275,
      "learning_rate": 3.604732457608817e-07,
      "loss": 0.3129,
      "step": 26595
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.686179757118225,
      "learning_rate": 3.6017621481738263e-07,
      "loss": 0.2436,
      "step": 26596
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.875773549079895,
      "learning_rate": 3.598793040571857e-07,
      "loss": 0.2831,
      "step": 26597
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0203678607940674,
      "learning_rate": 3.595825134839914e-07,
      "loss": 0.3123,
      "step": 26598
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.2799644470214844,
      "learning_rate": 3.5928584310150006e-07,
      "loss": 0.3514,
      "step": 26599
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.042494535446167,
      "learning_rate": 3.589892929134109e-07,
      "loss": 0.2874,
      "step": 26600
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1490373611450195,
      "learning_rate": 3.5869286292342097e-07,
      "loss": 0.3076,
      "step": 26601
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.047536611557007,
      "learning_rate": 3.5839655313522513e-07,
      "loss": 0.3804,
      "step": 26602
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9535753726959229,
      "learning_rate": 3.581003635525182e-07,
      "loss": 0.2562,
      "step": 26603
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6246237754821777,
      "learning_rate": 3.578042941789939e-07,
      "loss": 0.2413,
      "step": 26604
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3543453216552734,
      "learning_rate": 3.575083450183414e-07,
      "loss": 0.2681,
      "step": 26605
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9092785120010376,
      "learning_rate": 3.572125160742534e-07,
      "loss": 0.2305,
      "step": 26606
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3845512866973877,
      "learning_rate": 3.569168073504148e-07,
      "loss": 0.344,
      "step": 26607
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1612865924835205,
      "learning_rate": 3.566212188505136e-07,
      "loss": 0.209,
      "step": 26608
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.151092529296875,
      "learning_rate": 3.5632575057823583e-07,
      "loss": 0.314,
      "step": 26609
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7426091432571411,
      "learning_rate": 3.560304025372652e-07,
      "loss": 0.297,
      "step": 26610
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8686848878860474,
      "learning_rate": 3.5573517473128314e-07,
      "loss": 0.2763,
      "step": 26611
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7969375848770142,
      "learning_rate": 3.5544006716396905e-07,
      "loss": 0.2406,
      "step": 26612
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8501522541046143,
      "learning_rate": 3.551450798390044e-07,
      "loss": 0.2703,
      "step": 26613
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.814124345779419,
      "learning_rate": 3.548502127600662e-07,
      "loss": 0.2384,
      "step": 26614
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7774988412857056,
      "learning_rate": 3.545554659308303e-07,
      "loss": 0.3211,
      "step": 26615
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8732237815856934,
      "learning_rate": 3.542608393549718e-07,
      "loss": 0.26,
      "step": 26616
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8891206979751587,
      "learning_rate": 3.5396633303616426e-07,
      "loss": 0.2677,
      "step": 26617
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0874555110931396,
      "learning_rate": 3.5367194697807803e-07,
      "loss": 0.3358,
      "step": 26618
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.109546184539795,
      "learning_rate": 3.5337768118438474e-07,
      "loss": 0.2938,
      "step": 26619
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3610758781433105,
      "learning_rate": 3.5308353565875366e-07,
      "loss": 0.2676,
      "step": 26620
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.1403682231903076,
      "learning_rate": 3.5278951040484844e-07,
      "loss": 0.2584,
      "step": 26621
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.60771644115448,
      "learning_rate": 3.5249560542633953e-07,
      "loss": 0.2571,
      "step": 26622
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3747811317443848,
      "learning_rate": 3.5220182072688845e-07,
      "loss": 0.3834,
      "step": 26623
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9751074314117432,
      "learning_rate": 3.5190815631015897e-07,
      "loss": 0.3532,
      "step": 26624
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7564276456832886,
      "learning_rate": 3.516146121798103e-07,
      "loss": 0.3509,
      "step": 26625
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.013375997543335,
      "learning_rate": 3.513211883395051e-07,
      "loss": 0.2744,
      "step": 26626
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0118978023529053,
      "learning_rate": 3.510278847928994e-07,
      "loss": 0.3619,
      "step": 26627
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0044937133789062,
      "learning_rate": 3.507347015436502e-07,
      "loss": 0.273,
      "step": 26628
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.2006826400756836,
      "learning_rate": 3.504416385954146e-07,
      "loss": 0.3645,
      "step": 26629
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.9743703603744507,
      "learning_rate": 3.501486959518441e-07,
      "loss": 0.2538,
      "step": 26630
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.0486960411071777,
      "learning_rate": 3.4985587361659133e-07,
      "loss": 0.3031,
      "step": 26631
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.016491174697876,
      "learning_rate": 3.4956317159330786e-07,
      "loss": 0.2821,
      "step": 26632
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.273737668991089,
      "learning_rate": 3.49270589885643e-07,
      "loss": 0.2948,
      "step": 26633
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.755456566810608,
      "learning_rate": 3.489781284972427e-07,
      "loss": 0.2679,
      "step": 26634
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.9461734294891357,
      "learning_rate": 3.486857874317551e-07,
      "loss": 0.2695,
      "step": 26635
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6727192401885986,
      "learning_rate": 3.4839356669282507e-07,
      "loss": 0.2899,
      "step": 26636
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.020423173904419,
      "learning_rate": 3.4810146628409424e-07,
      "loss": 0.2638,
      "step": 26637
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.48573637008667,
      "learning_rate": 3.478094862092052e-07,
      "loss": 0.2359,
      "step": 26638
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.84589684009552,
      "learning_rate": 3.475176264717983e-07,
      "loss": 0.3313,
      "step": 26639
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.3776776790618896,
      "learning_rate": 3.4722588707551185e-07,
      "loss": 0.2784,
      "step": 26640
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.3642184734344482,
      "learning_rate": 3.469342680239829e-07,
      "loss": 0.3628,
      "step": 26641
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0066285133361816,
      "learning_rate": 3.466427693208485e-07,
      "loss": 0.3154,
      "step": 26642
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9888328313827515,
      "learning_rate": 3.4635139096974244e-07,
      "loss": 0.2967,
      "step": 26643
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.243915557861328,
      "learning_rate": 3.460601329742952e-07,
      "loss": 0.3635,
      "step": 26644
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9317916631698608,
      "learning_rate": 3.4576899533814044e-07,
      "loss": 0.242,
      "step": 26645
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.780299186706543,
      "learning_rate": 3.4547797806490757e-07,
      "loss": 0.2942,
      "step": 26646
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1108779907226562,
      "learning_rate": 3.451870811582236e-07,
      "loss": 0.3309,
      "step": 26647
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7960412502288818,
      "learning_rate": 3.4489630462171685e-07,
      "loss": 0.2589,
      "step": 26648
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8333417177200317,
      "learning_rate": 3.4460564845901213e-07,
      "loss": 0.2975,
      "step": 26649
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9921976327896118,
      "learning_rate": 3.443151126737332e-07,
      "loss": 0.2712,
      "step": 26650
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6291611194610596,
      "learning_rate": 3.440246972695005e-07,
      "loss": 0.2564,
      "step": 26651
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1443393230438232,
      "learning_rate": 3.437344022499367e-07,
      "loss": 0.2115,
      "step": 26652
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8339372873306274,
      "learning_rate": 3.4344422761866116e-07,
      "loss": 0.2674,
      "step": 26653
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.031510829925537,
      "learning_rate": 3.431541733792887e-07,
      "loss": 0.2616,
      "step": 26654
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0426430702209473,
      "learning_rate": 3.4286423953543977e-07,
      "loss": 0.309,
      "step": 26655
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.093007802963257,
      "learning_rate": 3.425744260907271e-07,
      "loss": 0.3051,
      "step": 26656
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0700080394744873,
      "learning_rate": 3.422847330487622e-07,
      "loss": 0.3254,
      "step": 26657
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.227323293685913,
      "learning_rate": 3.4199516041315994e-07,
      "loss": 0.3126,
      "step": 26658
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8516641855239868,
      "learning_rate": 3.417057081875286e-07,
      "loss": 0.2795,
      "step": 26659
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0174829959869385,
      "learning_rate": 3.414163763754763e-07,
      "loss": 0.2914,
      "step": 26660
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7167816162109375,
      "learning_rate": 3.411271649806125e-07,
      "loss": 0.3159,
      "step": 26661
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0359063148498535,
      "learning_rate": 3.4083807400654203e-07,
      "loss": 0.3346,
      "step": 26662
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.2858755588531494,
      "learning_rate": 3.4054910345686865e-07,
      "loss": 0.2556,
      "step": 26663
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8696528673171997,
      "learning_rate": 3.402602533351929e-07,
      "loss": 0.3117,
      "step": 26664
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9657411575317383,
      "learning_rate": 3.3997152364512063e-07,
      "loss": 0.3569,
      "step": 26665
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7678849697113037,
      "learning_rate": 3.3968291439024917e-07,
      "loss": 0.344,
      "step": 26666
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.188070774078369,
      "learning_rate": 3.3939442557417434e-07,
      "loss": 0.3424,
      "step": 26667
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0527172088623047,
      "learning_rate": 3.3910605720049674e-07,
      "loss": 0.2929,
      "step": 26668
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7753493785858154,
      "learning_rate": 3.3881780927281116e-07,
      "loss": 0.3071,
      "step": 26669
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9508509635925293,
      "learning_rate": 3.3852968179470813e-07,
      "loss": 0.2655,
      "step": 26670
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6077518463134766,
      "learning_rate": 3.382416747697825e-07,
      "loss": 0.2671,
      "step": 26671
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.4305994510650635,
      "learning_rate": 3.379537882016248e-07,
      "loss": 0.3414,
      "step": 26672
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7876648902893066,
      "learning_rate": 3.3766602209382325e-07,
      "loss": 0.2855,
      "step": 26673
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.2943367958068848,
      "learning_rate": 3.373783764499661e-07,
      "loss": 0.2761,
      "step": 26674
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7419874668121338,
      "learning_rate": 3.3709085127364037e-07,
      "loss": 0.292,
      "step": 26675
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7729344367980957,
      "learning_rate": 3.368034465684289e-07,
      "loss": 0.2603,
      "step": 26676
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0268759727478027,
      "learning_rate": 3.365161623379154e-07,
      "loss": 0.2917,
      "step": 26677
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1411638259887695,
      "learning_rate": 3.362289985856826e-07,
      "loss": 0.3438,
      "step": 26678
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0408167839050293,
      "learning_rate": 3.3594195531531095e-07,
      "loss": 0.3467,
      "step": 26679
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.4437572956085205,
      "learning_rate": 3.3565503253037646e-07,
      "loss": 0.2875,
      "step": 26680
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1760916709899902,
      "learning_rate": 3.353682302344585e-07,
      "loss": 0.283,
      "step": 26681
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6690131425857544,
      "learning_rate": 3.3508154843113316e-07,
      "loss": 0.2759,
      "step": 26682
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.29308819770813,
      "learning_rate": 3.3479498712397193e-07,
      "loss": 0.2904,
      "step": 26683
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9331849813461304,
      "learning_rate": 3.3450854631655094e-07,
      "loss": 0.2797,
      "step": 26684
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.014220952987671,
      "learning_rate": 3.342222260124406e-07,
      "loss": 0.3354,
      "step": 26685
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1814136505126953,
      "learning_rate": 3.3393602621520695e-07,
      "loss": 0.2855,
      "step": 26686
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.031032085418701,
      "learning_rate": 3.336499469284227e-07,
      "loss": 0.332,
      "step": 26687
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.16379714012146,
      "learning_rate": 3.3336398815565276e-07,
      "loss": 0.2897,
      "step": 26688
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9344241619110107,
      "learning_rate": 3.3307814990046206e-07,
      "loss": 0.2667,
      "step": 26689
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.6645281314849854,
      "learning_rate": 3.327924321664133e-07,
      "loss": 0.316,
      "step": 26690
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.3774354457855225,
      "learning_rate": 3.325068349570715e-07,
      "loss": 0.3244,
      "step": 26691
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9437557458877563,
      "learning_rate": 3.322213582759948e-07,
      "loss": 0.276,
      "step": 26692
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0223662853240967,
      "learning_rate": 3.319360021267426e-07,
      "loss": 0.3297,
      "step": 26693
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9825745820999146,
      "learning_rate": 3.316507665128732e-07,
      "loss": 0.3824,
      "step": 26694
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.250774383544922,
      "learning_rate": 3.3136565143794376e-07,
      "loss": 0.2877,
      "step": 26695
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.132431745529175,
      "learning_rate": 3.310806569055058e-07,
      "loss": 0.338,
      "step": 26696
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.448293924331665,
      "learning_rate": 3.3079578291911553e-07,
      "loss": 0.2541,
      "step": 26697
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.978514552116394,
      "learning_rate": 3.3051102948232326e-07,
      "loss": 0.2878,
      "step": 26698
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.258500099182129,
      "learning_rate": 3.302263965986785e-07,
      "loss": 0.3388,
      "step": 26699
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.702109932899475,
      "learning_rate": 3.2994188427173167e-07,
      "loss": 0.2927,
      "step": 26700
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8428794145584106,
      "learning_rate": 3.2965749250503e-07,
      "loss": 0.2567,
      "step": 26701
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.015122413635254,
      "learning_rate": 3.2937322130211614e-07,
      "loss": 0.2551,
      "step": 26702
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9561258554458618,
      "learning_rate": 3.2908907066653616e-07,
      "loss": 0.2843,
      "step": 26703
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0707366466522217,
      "learning_rate": 3.2880504060183395e-07,
      "loss": 0.2968,
      "step": 26704
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.3830764293670654,
      "learning_rate": 3.2852113111154773e-07,
      "loss": 0.3332,
      "step": 26705
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7731873989105225,
      "learning_rate": 3.282373421992191e-07,
      "loss": 0.2887,
      "step": 26706
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8898981809616089,
      "learning_rate": 3.279536738683864e-07,
      "loss": 0.3283,
      "step": 26707
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6191648244857788,
      "learning_rate": 3.276701261225845e-07,
      "loss": 0.2546,
      "step": 26708
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.18790602684021,
      "learning_rate": 3.273866989653496e-07,
      "loss": 0.3535,
      "step": 26709
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.141814947128296,
      "learning_rate": 3.2710339240021537e-07,
      "loss": 0.3044,
      "step": 26710
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9103399515151978,
      "learning_rate": 3.268202064307147e-07,
      "loss": 0.3271,
      "step": 26711
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.90857994556427,
      "learning_rate": 3.2653714106037573e-07,
      "loss": 0.2848,
      "step": 26712
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.205051898956299,
      "learning_rate": 3.2625419629273017e-07,
      "loss": 0.3421,
      "step": 26713
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8612983226776123,
      "learning_rate": 3.2597137213130403e-07,
      "loss": 0.302,
      "step": 26714
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7978476285934448,
      "learning_rate": 3.256886685796246e-07,
      "loss": 0.3434,
      "step": 26715
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8925098180770874,
      "learning_rate": 3.2540608564121335e-07,
      "loss": 0.3356,
      "step": 26716
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.2834455966949463,
      "learning_rate": 3.2512362331959755e-07,
      "loss": 0.2972,
      "step": 26717
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.181892156600952,
      "learning_rate": 3.2484128161829665e-07,
      "loss": 0.3028,
      "step": 26718
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.665431022644043,
      "learning_rate": 3.2455906054083e-07,
      "loss": 0.2878,
      "step": 26719
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.01704740524292,
      "learning_rate": 3.2427696009071916e-07,
      "loss": 0.3762,
      "step": 26720
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.249433755874634,
      "learning_rate": 3.2399498027147813e-07,
      "loss": 0.2808,
      "step": 26721
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.835982084274292,
      "learning_rate": 3.237131210866229e-07,
      "loss": 0.3341,
      "step": 26722
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.304938793182373,
      "learning_rate": 3.234313825396684e-07,
      "loss": 0.3445,
      "step": 26723
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.170875310897827,
      "learning_rate": 3.231497646341275e-07,
      "loss": 0.3349,
      "step": 26724
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.0897109508514404,
      "learning_rate": 3.228682673735084e-07,
      "loss": 0.3144,
      "step": 26725
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8930293321609497,
      "learning_rate": 3.22586890761325e-07,
      "loss": 0.2976,
      "step": 26726
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.1331255435943604,
      "learning_rate": 3.2230563480108223e-07,
      "loss": 0.3283,
      "step": 26727
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.528862953186035,
      "learning_rate": 3.2202449949628733e-07,
      "loss": 0.3532,
      "step": 26728
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9985491037368774,
      "learning_rate": 3.217434848504453e-07,
      "loss": 0.2169,
      "step": 26729
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7141876220703125,
      "learning_rate": 3.21462590867061e-07,
      "loss": 0.3099,
      "step": 26730
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.9290653467178345,
      "learning_rate": 3.2118181754963396e-07,
      "loss": 0.305,
      "step": 26731
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8750498294830322,
      "learning_rate": 3.209011649016669e-07,
      "loss": 0.2567,
      "step": 26732
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7597101926803589,
      "learning_rate": 3.206206329266581e-07,
      "loss": 0.2553,
      "step": 26733
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.86320424079895,
      "learning_rate": 3.2034022162810265e-07,
      "loss": 0.2821,
      "step": 26734
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.5371086597442627,
      "learning_rate": 3.2005993100950094e-07,
      "loss": 0.2972,
      "step": 26735
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9033012390136719,
      "learning_rate": 3.1977976107434583e-07,
      "loss": 0.2733,
      "step": 26736
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.155092716217041,
      "learning_rate": 3.1949971182612895e-07,
      "loss": 0.2689,
      "step": 26737
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8238245248794556,
      "learning_rate": 3.192197832683419e-07,
      "loss": 0.3013,
      "step": 26738
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.869126558303833,
      "learning_rate": 3.1893997540447643e-07,
      "loss": 0.3326,
      "step": 26739
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4733175039291382,
      "learning_rate": 3.1866028823801965e-07,
      "loss": 0.2474,
      "step": 26740
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.945312261581421,
      "learning_rate": 3.183807217724577e-07,
      "loss": 0.2931,
      "step": 26741
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8201944828033447,
      "learning_rate": 3.181012760112789e-07,
      "loss": 0.2732,
      "step": 26742
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1846301555633545,
      "learning_rate": 3.17821950957965e-07,
      "loss": 0.3618,
      "step": 26743
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1398465633392334,
      "learning_rate": 3.175427466159975e-07,
      "loss": 0.3437,
      "step": 26744
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.914973497390747,
      "learning_rate": 3.172636629888604e-07,
      "loss": 0.2811,
      "step": 26745
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.33543062210083,
      "learning_rate": 3.1698470008003077e-07,
      "loss": 0.3059,
      "step": 26746
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9021649360656738,
      "learning_rate": 3.167058578929871e-07,
      "loss": 0.3509,
      "step": 26747
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.993317723274231,
      "learning_rate": 3.164271364312066e-07,
      "loss": 0.2934,
      "step": 26748
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1854512691497803,
      "learning_rate": 3.1614853569816306e-07,
      "loss": 0.2878,
      "step": 26749
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.259608268737793,
      "learning_rate": 3.1587005569733154e-07,
      "loss": 0.2946,
      "step": 26750
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7644188404083252,
      "learning_rate": 3.1559169643218034e-07,
      "loss": 0.2837,
      "step": 26751
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.6039183139801025,
      "learning_rate": 3.153134579061845e-07,
      "loss": 0.2769,
      "step": 26752
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1153624057769775,
      "learning_rate": 3.1503534012281123e-07,
      "loss": 0.3002,
      "step": 26753
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9092655181884766,
      "learning_rate": 3.1475734308552554e-07,
      "loss": 0.3034,
      "step": 26754
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1428656578063965,
      "learning_rate": 3.1447946679779684e-07,
      "loss": 0.2968,
      "step": 26755
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.020948886871338,
      "learning_rate": 3.1420171126308796e-07,
      "loss": 0.2781,
      "step": 26756
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8883873224258423,
      "learning_rate": 3.1392407648486057e-07,
      "loss": 0.2964,
      "step": 26757
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9149668216705322,
      "learning_rate": 3.1364656246657743e-07,
      "loss": 0.3536,
      "step": 26758
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8535351753234863,
      "learning_rate": 3.133691692116991e-07,
      "loss": 0.3034,
      "step": 26759
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.140291690826416,
      "learning_rate": 3.1309189672368176e-07,
      "loss": 0.29,
      "step": 26760
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2511093616485596,
      "learning_rate": 3.1281474500598483e-07,
      "loss": 0.2918,
      "step": 26761
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.20156192779541,
      "learning_rate": 3.1253771406206336e-07,
      "loss": 0.2975,
      "step": 26762
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8949613571166992,
      "learning_rate": 3.1226080389536895e-07,
      "loss": 0.3189,
      "step": 26763
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.142057180404663,
      "learning_rate": 3.1198401450935444e-07,
      "loss": 0.3594,
      "step": 26764
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7452164888381958,
      "learning_rate": 3.1170734590747265e-07,
      "loss": 0.2603,
      "step": 26765
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.961912751197815,
      "learning_rate": 3.114307980931719e-07,
      "loss": 0.2813,
      "step": 26766
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8283816576004028,
      "learning_rate": 3.1115437106989833e-07,
      "loss": 0.238,
      "step": 26767
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8280918598175049,
      "learning_rate": 3.108780648411014e-07,
      "loss": 0.2829,
      "step": 26768
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.245450735092163,
      "learning_rate": 3.106018794102239e-07,
      "loss": 0.3017,
      "step": 26769
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9036685228347778,
      "learning_rate": 3.103258147807087e-07,
      "loss": 0.2955,
      "step": 26770
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2063045501708984,
      "learning_rate": 3.100498709559996e-07,
      "loss": 0.2997,
      "step": 26771
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.723662257194519,
      "learning_rate": 3.0977404793953616e-07,
      "loss": 0.3682,
      "step": 26772
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2917637825012207,
      "learning_rate": 3.0949834573475554e-07,
      "loss": 0.3491,
      "step": 26773
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9128984212875366,
      "learning_rate": 3.0922276434509624e-07,
      "loss": 0.2855,
      "step": 26774
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.45973539352417,
      "learning_rate": 3.089473037739954e-07,
      "loss": 0.2603,
      "step": 26775
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.289813995361328,
      "learning_rate": 3.0867196402488477e-07,
      "loss": 0.326,
      "step": 26776
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6494934558868408,
      "learning_rate": 3.0839674510119824e-07,
      "loss": 0.246,
      "step": 26777
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.592567801475525,
      "learning_rate": 3.081216470063675e-07,
      "loss": 0.2373,
      "step": 26778
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.998302936553955,
      "learning_rate": 3.0784666974382207e-07,
      "loss": 0.2411,
      "step": 26779
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1393680572509766,
      "learning_rate": 3.0757181331698917e-07,
      "loss": 0.2471,
      "step": 26780
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.013265371322632,
      "learning_rate": 3.0729707772929716e-07,
      "loss": 0.3669,
      "step": 26781
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.100440502166748,
      "learning_rate": 3.0702246298417117e-07,
      "loss": 0.2951,
      "step": 26782
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8212506771087646,
      "learning_rate": 3.067479690850317e-07,
      "loss": 0.2226,
      "step": 26783
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.309405565261841,
      "learning_rate": 3.06473596035306e-07,
      "loss": 0.3481,
      "step": 26784
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.158013105392456,
      "learning_rate": 3.0619934383841145e-07,
      "loss": 0.2805,
      "step": 26785
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.331156015396118,
      "learning_rate": 3.059252124977674e-07,
      "loss": 0.3368,
      "step": 26786
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.150956869125366,
      "learning_rate": 3.0565120201679344e-07,
      "loss": 0.2994,
      "step": 26787
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.930784821510315,
      "learning_rate": 3.0537731239890454e-07,
      "loss": 0.3233,
      "step": 26788
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8786498308181763,
      "learning_rate": 3.051035436475158e-07,
      "loss": 0.2492,
      "step": 26789
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1931326389312744,
      "learning_rate": 3.048298957660389e-07,
      "loss": 0.3128,
      "step": 26790
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.902427315711975,
      "learning_rate": 3.0455636875788673e-07,
      "loss": 0.3245,
      "step": 26791
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9348913431167603,
      "learning_rate": 3.042829626264698e-07,
      "loss": 0.3078,
      "step": 26792
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.4175403118133545,
      "learning_rate": 3.040096773751955e-07,
      "loss": 0.2846,
      "step": 26793
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1562414169311523,
      "learning_rate": 3.037365130074732e-07,
      "loss": 0.2042,
      "step": 26794
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.218254566192627,
      "learning_rate": 3.0346346952670693e-07,
      "loss": 0.3351,
      "step": 26795
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1096363067626953,
      "learning_rate": 3.0319054693629945e-07,
      "loss": 0.342,
      "step": 26796
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7528482675552368,
      "learning_rate": 3.02917745239657e-07,
      "loss": 0.2717,
      "step": 26797
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.764210820198059,
      "learning_rate": 3.0264506444017794e-07,
      "loss": 0.235,
      "step": 26798
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2976529598236084,
      "learning_rate": 3.023725045412618e-07,
      "loss": 0.3231,
      "step": 26799
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0219764709472656,
      "learning_rate": 3.02100065546308e-07,
      "loss": 0.3261,
      "step": 26800
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8670097589492798,
      "learning_rate": 3.018277474587139e-07,
      "loss": 0.2704,
      "step": 26801
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2488529682159424,
      "learning_rate": 3.0155555028187235e-07,
      "loss": 0.3055,
      "step": 26802
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.4080970287323,
      "learning_rate": 3.0128347401917726e-07,
      "loss": 0.3547,
      "step": 26803
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1397080421447754,
      "learning_rate": 3.0101151867402257e-07,
      "loss": 0.2439,
      "step": 26804
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6754719018936157,
      "learning_rate": 3.0073968424979785e-07,
      "loss": 0.2273,
      "step": 26805
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.6961872577667236,
      "learning_rate": 3.004679707498903e-07,
      "loss": 0.3054,
      "step": 26806
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.039170026779175,
      "learning_rate": 3.0019637817769067e-07,
      "loss": 0.3713,
      "step": 26807
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7721234560012817,
      "learning_rate": 2.999249065365828e-07,
      "loss": 0.2823,
      "step": 26808
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.199589490890503,
      "learning_rate": 2.996535558299507e-07,
      "loss": 0.3738,
      "step": 26809
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1415605545043945,
      "learning_rate": 2.9938232606118055e-07,
      "loss": 0.313,
      "step": 26810
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1071219444274902,
      "learning_rate": 2.9911121723365187e-07,
      "loss": 0.3385,
      "step": 26811
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8293216228485107,
      "learning_rate": 2.988402293507431e-07,
      "loss": 0.2881,
      "step": 26812
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.977323055267334,
      "learning_rate": 2.985693624158359e-07,
      "loss": 0.3161,
      "step": 26813
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8717917203903198,
      "learning_rate": 2.982986164323054e-07,
      "loss": 0.2315,
      "step": 26814
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6842083930969238,
      "learning_rate": 2.9802799140352776e-07,
      "loss": 0.3289,
      "step": 26815
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.432227373123169,
      "learning_rate": 2.9775748733287593e-07,
      "loss": 0.3183,
      "step": 26816
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.897636890411377,
      "learning_rate": 2.974871042237237e-07,
      "loss": 0.3093,
      "step": 26817
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0171778202056885,
      "learning_rate": 2.972168420794419e-07,
      "loss": 0.3088,
      "step": 26818
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9643161296844482,
      "learning_rate": 2.969467009033977e-07,
      "loss": 0.2251,
      "step": 26819
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.105865716934204,
      "learning_rate": 2.966766806989629e-07,
      "loss": 0.3358,
      "step": 26820
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1076297760009766,
      "learning_rate": 2.9640678146950154e-07,
      "loss": 0.214,
      "step": 26821
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.1399755477905273,
      "learning_rate": 2.9613700321837747e-07,
      "loss": 0.3938,
      "step": 26822
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7694473266601562,
      "learning_rate": 2.958673459489569e-07,
      "loss": 0.2415,
      "step": 26823
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8594897985458374,
      "learning_rate": 2.955978096646006e-07,
      "loss": 0.3042,
      "step": 26824
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0474164485931396,
      "learning_rate": 2.953283943686669e-07,
      "loss": 0.3525,
      "step": 26825
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0351154804229736,
      "learning_rate": 2.9505910006451867e-07,
      "loss": 0.2625,
      "step": 26826
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.749489665031433,
      "learning_rate": 2.94789926755511e-07,
      "loss": 0.3197,
      "step": 26827
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7613484859466553,
      "learning_rate": 2.945208744450001e-07,
      "loss": 0.2725,
      "step": 26828
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.3093111515045166,
      "learning_rate": 2.942519431363389e-07,
      "loss": 0.3527,
      "step": 26829
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.0723073482513428,
      "learning_rate": 2.939831328328835e-07,
      "loss": 0.3031,
      "step": 26830
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.166407346725464,
      "learning_rate": 2.937144435379824e-07,
      "loss": 0.3483,
      "step": 26831
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3132004737854004,
      "learning_rate": 2.9344587525498623e-07,
      "loss": 0.2939,
      "step": 26832
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.324171781539917,
      "learning_rate": 2.9317742798724344e-07,
      "loss": 0.2841,
      "step": 26833
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.705460786819458,
      "learning_rate": 2.929091017381025e-07,
      "loss": 0.3002,
      "step": 26834
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.910398244857788,
      "learning_rate": 2.926408965109051e-07,
      "loss": 0.3183,
      "step": 26835
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3453118801116943,
      "learning_rate": 2.923728123089986e-07,
      "loss": 0.3358,
      "step": 26836
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.2825284004211426,
      "learning_rate": 2.921048491357237e-07,
      "loss": 0.2345,
      "step": 26837
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0109121799468994,
      "learning_rate": 2.9183700699442096e-07,
      "loss": 0.3535,
      "step": 26838
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.157958745956421,
      "learning_rate": 2.9156928588843005e-07,
      "loss": 0.3318,
      "step": 26839
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.055574417114258,
      "learning_rate": 2.913016858210893e-07,
      "loss": 0.2303,
      "step": 26840
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8882492780685425,
      "learning_rate": 2.9103420679573503e-07,
      "loss": 0.2803,
      "step": 26841
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0600414276123047,
      "learning_rate": 2.9076684881570004e-07,
      "loss": 0.3112,
      "step": 26842
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3406076431274414,
      "learning_rate": 2.9049961188431953e-07,
      "loss": 0.3068,
      "step": 26843
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.5735602378845215,
      "learning_rate": 2.902324960049252e-07,
      "loss": 0.319,
      "step": 26844
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7456308603286743,
      "learning_rate": 2.8996550118084554e-07,
      "loss": 0.2304,
      "step": 26845
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9093841314315796,
      "learning_rate": 2.896986274154112e-07,
      "loss": 0.3262,
      "step": 26846
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.013383150100708,
      "learning_rate": 2.8943187471194844e-07,
      "loss": 0.3739,
      "step": 26847
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.2853453159332275,
      "learning_rate": 2.8916524307378344e-07,
      "loss": 0.3155,
      "step": 26848
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9385309219360352,
      "learning_rate": 2.8889873250424025e-07,
      "loss": 0.3822,
      "step": 26849
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.710745930671692,
      "learning_rate": 2.886323430066418e-07,
      "loss": 0.3138,
      "step": 26850
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0494632720947266,
      "learning_rate": 2.8836607458430755e-07,
      "loss": 0.3143,
      "step": 26851
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.2143514156341553,
      "learning_rate": 2.8809992724056047e-07,
      "loss": 0.2904,
      "step": 26852
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8077232837677002,
      "learning_rate": 2.878339009787168e-07,
      "loss": 0.253,
      "step": 26853
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0432238578796387,
      "learning_rate": 2.875679958020927e-07,
      "loss": 0.337,
      "step": 26854
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8526219129562378,
      "learning_rate": 2.873022117140023e-07,
      "loss": 0.2666,
      "step": 26855
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.404226064682007,
      "learning_rate": 2.870365487177629e-07,
      "loss": 0.269,
      "step": 26856
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8039507865905762,
      "learning_rate": 2.8677100681668404e-07,
      "loss": 0.3,
      "step": 26857
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.2365036010742188,
      "learning_rate": 2.865055860140753e-07,
      "loss": 0.3525,
      "step": 26858
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.15971302986145,
      "learning_rate": 2.8624028631324854e-07,
      "loss": 0.3633,
      "step": 26859
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0005719661712646,
      "learning_rate": 2.859751077175099e-07,
      "loss": 0.3044,
      "step": 26860
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3000495433807373,
      "learning_rate": 2.8571005023016464e-07,
      "loss": 0.3237,
      "step": 26861
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7480649948120117,
      "learning_rate": 2.8544511385452e-07,
      "loss": 0.2992,
      "step": 26862
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.688975214958191,
      "learning_rate": 2.8518029859387676e-07,
      "loss": 0.2085,
      "step": 26863
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.014829397201538,
      "learning_rate": 2.8491560445153556e-07,
      "loss": 0.3131,
      "step": 26864
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3505935668945312,
      "learning_rate": 2.846510314308004e-07,
      "loss": 0.3179,
      "step": 26865
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8330146074295044,
      "learning_rate": 2.8438657953496653e-07,
      "loss": 0.2906,
      "step": 26866
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5958530902862549,
      "learning_rate": 2.841222487673312e-07,
      "loss": 0.2696,
      "step": 26867
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.922302007675171,
      "learning_rate": 2.838580391311896e-07,
      "loss": 0.366,
      "step": 26868
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8087471723556519,
      "learning_rate": 2.835939506298391e-07,
      "loss": 0.2787,
      "step": 26869
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9057166576385498,
      "learning_rate": 2.833299832665681e-07,
      "loss": 0.276,
      "step": 26870
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7200785875320435,
      "learning_rate": 2.8306613704466857e-07,
      "loss": 0.2665,
      "step": 26871
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.612045168876648,
      "learning_rate": 2.828024119674322e-07,
      "loss": 0.1851,
      "step": 26872
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6012638807296753,
      "learning_rate": 2.825388080381453e-07,
      "loss": 0.2789,
      "step": 26873
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0975935459136963,
      "learning_rate": 2.8227532526009295e-07,
      "loss": 0.3186,
      "step": 26874
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8030096292495728,
      "learning_rate": 2.8201196363656257e-07,
      "loss": 0.2236,
      "step": 26875
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.916638731956482,
      "learning_rate": 2.8174872317083714e-07,
      "loss": 0.2975,
      "step": 26876
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8840497732162476,
      "learning_rate": 2.8148560386619617e-07,
      "loss": 0.3137,
      "step": 26877
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.1479780673980713,
      "learning_rate": 2.8122260572592375e-07,
      "loss": 0.3062,
      "step": 26878
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9535876512527466,
      "learning_rate": 2.809597287532961e-07,
      "loss": 0.2617,
      "step": 26879
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7426456212997437,
      "learning_rate": 2.806969729515918e-07,
      "loss": 0.2204,
      "step": 26880
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.046865224838257,
      "learning_rate": 2.8043433832408597e-07,
      "loss": 0.3283,
      "step": 26881
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.4202635288238525,
      "learning_rate": 2.801718248740537e-07,
      "loss": 0.3528,
      "step": 26882
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8917618989944458,
      "learning_rate": 2.799094326047669e-07,
      "loss": 0.2745,
      "step": 26883
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8397786617279053,
      "learning_rate": 2.7964716151949734e-07,
      "loss": 0.3468,
      "step": 26884
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.121378183364868,
      "learning_rate": 2.7938501162151574e-07,
      "loss": 0.3309,
      "step": 26885
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9846241474151611,
      "learning_rate": 2.791229829140907e-07,
      "loss": 0.3297,
      "step": 26886
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.165189266204834,
      "learning_rate": 2.78861075400485e-07,
      "loss": 0.2391,
      "step": 26887
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6121762990951538,
      "learning_rate": 2.785992890839695e-07,
      "loss": 0.304,
      "step": 26888
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.208507537841797,
      "learning_rate": 2.7833762396780486e-07,
      "loss": 0.3487,
      "step": 26889
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.108548641204834,
      "learning_rate": 2.78076080055254e-07,
      "loss": 0.2806,
      "step": 26890
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.193634033203125,
      "learning_rate": 2.778146573495777e-07,
      "loss": 0.319,
      "step": 26891
    },
    {
      "epoch": 2.78,
      "grad_norm": 3.1176366806030273,
      "learning_rate": 2.775533558540344e-07,
      "loss": 0.3102,
      "step": 26892
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6061432361602783,
      "learning_rate": 2.772921755718838e-07,
      "loss": 0.2454,
      "step": 26893
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0034313201904297,
      "learning_rate": 2.770311165063799e-07,
      "loss": 0.2496,
      "step": 26894
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.030988931655884,
      "learning_rate": 2.7677017866077905e-07,
      "loss": 0.2454,
      "step": 26895
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8898931741714478,
      "learning_rate": 2.765093620383341e-07,
      "loss": 0.2898,
      "step": 26896
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9779584407806396,
      "learning_rate": 2.762486666422948e-07,
      "loss": 0.3405,
      "step": 26897
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0131523609161377,
      "learning_rate": 2.759880924759151e-07,
      "loss": 0.3331,
      "step": 26898
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.169480562210083,
      "learning_rate": 2.757276395424413e-07,
      "loss": 0.3363,
      "step": 26899
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.836934208869934,
      "learning_rate": 2.7546730784511864e-07,
      "loss": 0.2998,
      "step": 26900
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8892757892608643,
      "learning_rate": 2.7520709738719786e-07,
      "loss": 0.2615,
      "step": 26901
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8085490465164185,
      "learning_rate": 2.7494700817191964e-07,
      "loss": 0.2959,
      "step": 26902
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7633098363876343,
      "learning_rate": 2.746870402025259e-07,
      "loss": 0.3058,
      "step": 26903
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9744110107421875,
      "learning_rate": 2.744271934822607e-07,
      "loss": 0.282,
      "step": 26904
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7305543422698975,
      "learning_rate": 2.7416746801436134e-07,
      "loss": 0.2442,
      "step": 26905
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.995739459991455,
      "learning_rate": 2.739078638020665e-07,
      "loss": 0.3192,
      "step": 26906
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.654761791229248,
      "learning_rate": 2.7364838084861236e-07,
      "loss": 0.2397,
      "step": 26907
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.0640218257904053,
      "learning_rate": 2.7338901915723523e-07,
      "loss": 0.319,
      "step": 26908
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9364933967590332,
      "learning_rate": 2.731297787311682e-07,
      "loss": 0.2724,
      "step": 26909
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.1788723468780518,
      "learning_rate": 2.728706595736419e-07,
      "loss": 0.3353,
      "step": 26910
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5730364322662354,
      "learning_rate": 2.726116616878893e-07,
      "loss": 0.2562,
      "step": 26911
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.104055404663086,
      "learning_rate": 2.7235278507713793e-07,
      "loss": 0.293,
      "step": 26912
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9611340761184692,
      "learning_rate": 2.7209402974461507e-07,
      "loss": 0.3196,
      "step": 26913
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.882582664489746,
      "learning_rate": 2.718353956935471e-07,
      "loss": 0.281,
      "step": 26914
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.855294942855835,
      "learning_rate": 2.715768829271592e-07,
      "loss": 0.3463,
      "step": 26915
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.044829845428467,
      "learning_rate": 2.713184914486733e-07,
      "loss": 0.2294,
      "step": 26916
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.816497802734375,
      "learning_rate": 2.7106022126131115e-07,
      "loss": 0.2751,
      "step": 26917
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.1231141090393066,
      "learning_rate": 2.708020723682936e-07,
      "loss": 0.2656,
      "step": 26918
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9938827753067017,
      "learning_rate": 2.7054404477283804e-07,
      "loss": 0.2435,
      "step": 26919
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9755367040634155,
      "learning_rate": 2.702861384781608e-07,
      "loss": 0.3316,
      "step": 26920
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.807224988937378,
      "learning_rate": 2.7002835348747926e-07,
      "loss": 0.3541,
      "step": 26921
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9343022108078003,
      "learning_rate": 2.6977068980400645e-07,
      "loss": 0.2871,
      "step": 26922
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.130122423171997,
      "learning_rate": 2.695131474309531e-07,
      "loss": 0.3088,
      "step": 26923
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6083171367645264,
      "learning_rate": 2.692557263715334e-07,
      "loss": 0.2924,
      "step": 26924
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8724018335342407,
      "learning_rate": 2.689984266289536e-07,
      "loss": 0.2585,
      "step": 26925
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.1942617893218994,
      "learning_rate": 2.6874124820642223e-07,
      "loss": 0.2328,
      "step": 26926
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.879579782485962,
      "learning_rate": 2.6848419110714786e-07,
      "loss": 0.279,
      "step": 26927
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8710075616836548,
      "learning_rate": 2.682272553343335e-07,
      "loss": 0.3008,
      "step": 26928
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1715526580810547,
      "learning_rate": 2.67970440891181e-07,
      "loss": 0.2645,
      "step": 26929
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8653762340545654,
      "learning_rate": 2.6771374778089555e-07,
      "loss": 0.3068,
      "step": 26930
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1796369552612305,
      "learning_rate": 2.6745717600667463e-07,
      "loss": 0.3724,
      "step": 26931
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.91786527633667,
      "learning_rate": 2.6720072557171907e-07,
      "loss": 0.3093,
      "step": 26932
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.245919942855835,
      "learning_rate": 2.6694439647922397e-07,
      "loss": 0.2484,
      "step": 26933
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.6425423622131348,
      "learning_rate": 2.666881887323869e-07,
      "loss": 0.4347,
      "step": 26934
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.14839768409729,
      "learning_rate": 2.6643210233440187e-07,
      "loss": 0.3601,
      "step": 26935
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.817021608352661,
      "learning_rate": 2.661761372884597e-07,
      "loss": 0.2656,
      "step": 26936
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5964003801345825,
      "learning_rate": 2.6592029359775453e-07,
      "loss": 0.3027,
      "step": 26937
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9632388353347778,
      "learning_rate": 2.6566457126547375e-07,
      "loss": 0.3203,
      "step": 26938
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7535218000411987,
      "learning_rate": 2.6540897029480597e-07,
      "loss": 0.2691,
      "step": 26939
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.897047996520996,
      "learning_rate": 2.6515349068893967e-07,
      "loss": 0.3846,
      "step": 26940
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.337090492248535,
      "learning_rate": 2.64898132451058e-07,
      "loss": 0.3171,
      "step": 26941
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.614193320274353,
      "learning_rate": 2.64642895584345e-07,
      "loss": 0.2214,
      "step": 26942
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1794228553771973,
      "learning_rate": 2.6438778009198364e-07,
      "loss": 0.3017,
      "step": 26943
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6992921829223633,
      "learning_rate": 2.6413278597715363e-07,
      "loss": 0.2455,
      "step": 26944
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7538591623306274,
      "learning_rate": 2.638779132430358e-07,
      "loss": 0.2204,
      "step": 26945
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.4437594413757324,
      "learning_rate": 2.6362316189280423e-07,
      "loss": 0.2702,
      "step": 26946
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9713023900985718,
      "learning_rate": 2.6336853192963753e-07,
      "loss": 0.2765,
      "step": 26947
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.030390501022339,
      "learning_rate": 2.631140233567109e-07,
      "loss": 0.3768,
      "step": 26948
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9129136800765991,
      "learning_rate": 2.628596361771951e-07,
      "loss": 0.2094,
      "step": 26949
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0231385231018066,
      "learning_rate": 2.6260537039426327e-07,
      "loss": 0.2917,
      "step": 26950
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6846784353256226,
      "learning_rate": 2.623512260110861e-07,
      "loss": 0.2536,
      "step": 26951
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.021650552749634,
      "learning_rate": 2.6209720303082884e-07,
      "loss": 0.3633,
      "step": 26952
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7519192695617676,
      "learning_rate": 2.618433014566635e-07,
      "loss": 0.2624,
      "step": 26953
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.254765272140503,
      "learning_rate": 2.6158952129175077e-07,
      "loss": 0.2612,
      "step": 26954
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.3970510959625244,
      "learning_rate": 2.61335862539257e-07,
      "loss": 0.3238,
      "step": 26955
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.860460877418518,
      "learning_rate": 2.6108232520234534e-07,
      "loss": 0.2504,
      "step": 26956
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8648695945739746,
      "learning_rate": 2.6082890928417537e-07,
      "loss": 0.311,
      "step": 26957
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.074537992477417,
      "learning_rate": 2.605756147879068e-07,
      "loss": 0.3079,
      "step": 26958
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.390127420425415,
      "learning_rate": 2.603224417166961e-07,
      "loss": 0.2964,
      "step": 26959
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5691943168640137,
      "learning_rate": 2.600693900737028e-07,
      "loss": 0.2247,
      "step": 26960
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.5176804065704346,
      "learning_rate": 2.5981645986208004e-07,
      "loss": 0.251,
      "step": 26961
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.943996787071228,
      "learning_rate": 2.5956365108498084e-07,
      "loss": 0.2561,
      "step": 26962
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8321802616119385,
      "learning_rate": 2.593109637455582e-07,
      "loss": 0.3245,
      "step": 26963
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.4228243827819824,
      "learning_rate": 2.5905839784696074e-07,
      "loss": 0.3185,
      "step": 26964
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.2829577922821045,
      "learning_rate": 2.5880595339233927e-07,
      "loss": 0.2973,
      "step": 26965
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1423447132110596,
      "learning_rate": 2.5855363038484014e-07,
      "loss": 0.3345,
      "step": 26966
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9659769535064697,
      "learning_rate": 2.5830142882760865e-07,
      "loss": 0.2955,
      "step": 26967
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8947349786758423,
      "learning_rate": 2.5804934872379005e-07,
      "loss": 0.2464,
      "step": 26968
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.018411874771118,
      "learning_rate": 2.5779739007652625e-07,
      "loss": 0.3229,
      "step": 26969
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.067760467529297,
      "learning_rate": 2.575455528889603e-07,
      "loss": 0.3733,
      "step": 26970
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.000394344329834,
      "learning_rate": 2.5729383716422973e-07,
      "loss": 0.3707,
      "step": 26971
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.2251296043395996,
      "learning_rate": 2.570422429054731e-07,
      "loss": 0.3014,
      "step": 26972
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.19858455657959,
      "learning_rate": 2.56790770115829e-07,
      "loss": 0.3388,
      "step": 26973
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8001106977462769,
      "learning_rate": 2.5653941879843046e-07,
      "loss": 0.3303,
      "step": 26974
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.865024447441101,
      "learning_rate": 2.562881889564117e-07,
      "loss": 0.3437,
      "step": 26975
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1313812732696533,
      "learning_rate": 2.5603708059290576e-07,
      "loss": 0.3445,
      "step": 26976
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5800251960754395,
      "learning_rate": 2.5578609371104345e-07,
      "loss": 0.301,
      "step": 26977
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9845548868179321,
      "learning_rate": 2.5553522831395226e-07,
      "loss": 0.27,
      "step": 26978
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.05264949798584,
      "learning_rate": 2.5528448440476084e-07,
      "loss": 0.2632,
      "step": 26979
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.990736961364746,
      "learning_rate": 2.550338619865966e-07,
      "loss": 0.3135,
      "step": 26980
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0096640586853027,
      "learning_rate": 2.547833610625816e-07,
      "loss": 0.2573,
      "step": 26981
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8370312452316284,
      "learning_rate": 2.5453298163583995e-07,
      "loss": 0.3874,
      "step": 26982
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6688487529754639,
      "learning_rate": 2.542827237094947e-07,
      "loss": 0.2238,
      "step": 26983
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.033092737197876,
      "learning_rate": 2.5403258728666445e-07,
      "loss": 0.2586,
      "step": 26984
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.142509937286377,
      "learning_rate": 2.537825723704668e-07,
      "loss": 0.2711,
      "step": 26985
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0919077396392822,
      "learning_rate": 2.535326789640213e-07,
      "loss": 0.3057,
      "step": 26986
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.953593373298645,
      "learning_rate": 2.5328290707044123e-07,
      "loss": 0.385,
      "step": 26987
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7897757291793823,
      "learning_rate": 2.5303325669284065e-07,
      "loss": 0.2806,
      "step": 26988
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.842208981513977,
      "learning_rate": 2.527837278343348e-07,
      "loss": 0.3283,
      "step": 26989
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1751577854156494,
      "learning_rate": 2.525343204980313e-07,
      "loss": 0.3124,
      "step": 26990
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1432244777679443,
      "learning_rate": 2.522850346870409e-07,
      "loss": 0.254,
      "step": 26991
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7447078227996826,
      "learning_rate": 2.520358704044723e-07,
      "loss": 0.303,
      "step": 26992
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0292351245880127,
      "learning_rate": 2.5178682765343185e-07,
      "loss": 0.3086,
      "step": 26993
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.857102394104004,
      "learning_rate": 2.515379064370238e-07,
      "loss": 0.3208,
      "step": 26994
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.719360113143921,
      "learning_rate": 2.5128910675835004e-07,
      "loss": 0.3364,
      "step": 26995
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.646560549736023,
      "learning_rate": 2.510404286205159e-07,
      "loss": 0.275,
      "step": 26996
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9432729482650757,
      "learning_rate": 2.507918720266189e-07,
      "loss": 0.2999,
      "step": 26997
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0962510108947754,
      "learning_rate": 2.505434369797577e-07,
      "loss": 0.2951,
      "step": 26998
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.2381432056427,
      "learning_rate": 2.502951234830331e-07,
      "loss": 0.2984,
      "step": 26999
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.928703784942627,
      "learning_rate": 2.500469315395371e-07,
      "loss": 0.316,
      "step": 27000
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.3831233978271484,
      "learning_rate": 2.4979886115236606e-07,
      "loss": 0.2953,
      "step": 27001
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9471794366836548,
      "learning_rate": 2.49550912324612e-07,
      "loss": 0.4116,
      "step": 27002
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9435750246047974,
      "learning_rate": 2.4930308505936696e-07,
      "loss": 0.3361,
      "step": 27003
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8201324939727783,
      "learning_rate": 2.490553793597195e-07,
      "loss": 0.3097,
      "step": 27004
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6713988780975342,
      "learning_rate": 2.488077952287582e-07,
      "loss": 0.2685,
      "step": 27005
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7120225429534912,
      "learning_rate": 2.485603326695707e-07,
      "loss": 0.2773,
      "step": 27006
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1675641536712646,
      "learning_rate": 2.4831299168524227e-07,
      "loss": 0.2793,
      "step": 27007
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.0274667739868164,
      "learning_rate": 2.480657722788549e-07,
      "loss": 0.3184,
      "step": 27008
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.904526710510254,
      "learning_rate": 2.478186744534916e-07,
      "loss": 0.3927,
      "step": 27009
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.054804563522339,
      "learning_rate": 2.475716982122345e-07,
      "loss": 0.3426,
      "step": 27010
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.032528877258301,
      "learning_rate": 2.473248435581599e-07,
      "loss": 0.3323,
      "step": 27011
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9608802795410156,
      "learning_rate": 2.470781104943487e-07,
      "loss": 0.2747,
      "step": 27012
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.119791030883789,
      "learning_rate": 2.46831499023874e-07,
      "loss": 0.3326,
      "step": 27013
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.49285626411438,
      "learning_rate": 2.465850091498123e-07,
      "loss": 0.3591,
      "step": 27014
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7870317697525024,
      "learning_rate": 2.463386408752355e-07,
      "loss": 0.2842,
      "step": 27015
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.619594693183899,
      "learning_rate": 2.460923942032167e-07,
      "loss": 0.2686,
      "step": 27016
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.150461196899414,
      "learning_rate": 2.458462691368235e-07,
      "loss": 0.3317,
      "step": 27017
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.902803659439087,
      "learning_rate": 2.4560026567912787e-07,
      "loss": 0.259,
      "step": 27018
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.083411931991577,
      "learning_rate": 2.45354383833194e-07,
      "loss": 0.366,
      "step": 27019
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7590916156768799,
      "learning_rate": 2.451086236020894e-07,
      "loss": 0.2728,
      "step": 27020
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.2434041500091553,
      "learning_rate": 2.448629849888751e-07,
      "loss": 0.2984,
      "step": 27021
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.827326774597168,
      "learning_rate": 2.446174679966162e-07,
      "loss": 0.2928,
      "step": 27022
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.021963119506836,
      "learning_rate": 2.443720726283727e-07,
      "loss": 0.3198,
      "step": 27023
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.268139362335205,
      "learning_rate": 2.4412679888720426e-07,
      "loss": 0.3201,
      "step": 27024
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2694640159606934,
      "learning_rate": 2.4388164677616956e-07,
      "loss": 0.2822,
      "step": 27025
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9227523803710938,
      "learning_rate": 2.4363661629832393e-07,
      "loss": 0.2595,
      "step": 27026
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9318351745605469,
      "learning_rate": 2.433917074567205e-07,
      "loss": 0.3233,
      "step": 27027
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1715216636657715,
      "learning_rate": 2.431469202544168e-07,
      "loss": 0.3308,
      "step": 27028
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2892606258392334,
      "learning_rate": 2.4290225469446263e-07,
      "loss": 0.3014,
      "step": 27029
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8323967456817627,
      "learning_rate": 2.4265771077990663e-07,
      "loss": 0.3128,
      "step": 27030
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.183936357498169,
      "learning_rate": 2.4241328851380084e-07,
      "loss": 0.269,
      "step": 27031
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7441474199295044,
      "learning_rate": 2.421689878991906e-07,
      "loss": 0.3526,
      "step": 27032
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2144365310668945,
      "learning_rate": 2.419248089391224e-07,
      "loss": 0.2972,
      "step": 27033
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2836127281188965,
      "learning_rate": 2.4168075163663817e-07,
      "loss": 0.3528,
      "step": 27034
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7641619443893433,
      "learning_rate": 2.414368159947844e-07,
      "loss": 0.3161,
      "step": 27035
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.550053596496582,
      "learning_rate": 2.411930020166009e-07,
      "loss": 0.2446,
      "step": 27036
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.203444004058838,
      "learning_rate": 2.409493097051263e-07,
      "loss": 0.2866,
      "step": 27037
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.167214870452881,
      "learning_rate": 2.4070573906339935e-07,
      "loss": 0.2602,
      "step": 27038
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.4028866291046143,
      "learning_rate": 2.4046229009445753e-07,
      "loss": 0.3456,
      "step": 27039
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2549514770507812,
      "learning_rate": 2.4021896280133516e-07,
      "loss": 0.3335,
      "step": 27040
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8480138778686523,
      "learning_rate": 2.3997575718706646e-07,
      "loss": 0.2494,
      "step": 27041
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0470573902130127,
      "learning_rate": 2.3973267325468343e-07,
      "loss": 0.2997,
      "step": 27042
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6753078699111938,
      "learning_rate": 2.3948971100721587e-07,
      "loss": 0.2242,
      "step": 27043
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8160820007324219,
      "learning_rate": 2.3924687044769355e-07,
      "loss": 0.2986,
      "step": 27044
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.426616907119751,
      "learning_rate": 2.3900415157914526e-07,
      "loss": 0.3111,
      "step": 27045
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7747368812561035,
      "learning_rate": 2.3876155440459516e-07,
      "loss": 0.315,
      "step": 27046
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0082950592041016,
      "learning_rate": 2.385190789270675e-07,
      "loss": 0.3056,
      "step": 27047
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9927196502685547,
      "learning_rate": 2.3827672514958765e-07,
      "loss": 0.3487,
      "step": 27048
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9710348844528198,
      "learning_rate": 2.3803449307517546e-07,
      "loss": 0.2859,
      "step": 27049
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9121267795562744,
      "learning_rate": 2.377923827068507e-07,
      "loss": 0.2906,
      "step": 27050
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0989160537719727,
      "learning_rate": 2.3755039404763314e-07,
      "loss": 0.3151,
      "step": 27051
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9928483963012695,
      "learning_rate": 2.3730852710053932e-07,
      "loss": 0.264,
      "step": 27052
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0344736576080322,
      "learning_rate": 2.3706678186858344e-07,
      "loss": 0.3084,
      "step": 27053
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.3814356327056885,
      "learning_rate": 2.3682515835478094e-07,
      "loss": 0.3077,
      "step": 27054
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7379789352416992,
      "learning_rate": 2.3658365656214378e-07,
      "loss": 0.292,
      "step": 27055
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1790497303009033,
      "learning_rate": 2.363422764936818e-07,
      "loss": 0.2547,
      "step": 27056
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.916533350944519,
      "learning_rate": 2.3610101815240594e-07,
      "loss": 0.286,
      "step": 27057
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.838029146194458,
      "learning_rate": 2.358598815413238e-07,
      "loss": 0.2522,
      "step": 27058
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6043503284454346,
      "learning_rate": 2.3561886666344182e-07,
      "loss": 0.2209,
      "step": 27059
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.806121826171875,
      "learning_rate": 2.3537797352176206e-07,
      "loss": 0.241,
      "step": 27060
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9958064556121826,
      "learning_rate": 2.3513720211929214e-07,
      "loss": 0.2923,
      "step": 27061
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.147385597229004,
      "learning_rate": 2.3489655245903186e-07,
      "loss": 0.345,
      "step": 27062
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8264822959899902,
      "learning_rate": 2.3465602454397995e-07,
      "loss": 0.2755,
      "step": 27063
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7388944625854492,
      "learning_rate": 2.3441561837713845e-07,
      "loss": 0.2574,
      "step": 27064
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6033315658569336,
      "learning_rate": 2.341753339615016e-07,
      "loss": 0.2706,
      "step": 27065
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.317091226577759,
      "learning_rate": 2.3393517130006594e-07,
      "loss": 0.2917,
      "step": 27066
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7090121507644653,
      "learning_rate": 2.3369513039582792e-07,
      "loss": 0.3156,
      "step": 27067
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1145386695861816,
      "learning_rate": 2.3345521125177739e-07,
      "loss": 0.3232,
      "step": 27068
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2225475311279297,
      "learning_rate": 2.332154138709053e-07,
      "loss": 0.2792,
      "step": 27069
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.260716199874878,
      "learning_rate": 2.3297573825620368e-07,
      "loss": 0.2755,
      "step": 27070
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.299870729446411,
      "learning_rate": 2.3273618441066016e-07,
      "loss": 0.2636,
      "step": 27071
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1842920780181885,
      "learning_rate": 2.324967523372601e-07,
      "loss": 0.3172,
      "step": 27072
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0466573238372803,
      "learning_rate": 2.3225744203898893e-07,
      "loss": 0.3116,
      "step": 27073
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.5194709300994873,
      "learning_rate": 2.320182535188309e-07,
      "loss": 0.3469,
      "step": 27074
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.808368444442749,
      "learning_rate": 2.3177918677976584e-07,
      "loss": 0.2772,
      "step": 27075
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.956711769104004,
      "learning_rate": 2.3154024182477808e-07,
      "loss": 0.2774,
      "step": 27076
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.225402355194092,
      "learning_rate": 2.3130141865684408e-07,
      "loss": 0.3457,
      "step": 27077
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9493353366851807,
      "learning_rate": 2.3106271727894035e-07,
      "loss": 0.2971,
      "step": 27078
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9147342443466187,
      "learning_rate": 2.3082413769404566e-07,
      "loss": 0.2354,
      "step": 27079
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1119048595428467,
      "learning_rate": 2.3058567990513313e-07,
      "loss": 0.308,
      "step": 27080
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2910120487213135,
      "learning_rate": 2.30347343915176e-07,
      "loss": 0.3407,
      "step": 27081
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8822274208068848,
      "learning_rate": 2.3010912972714407e-07,
      "loss": 0.3456,
      "step": 27082
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9060571193695068,
      "learning_rate": 2.298710373440105e-07,
      "loss": 0.2068,
      "step": 27083
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.365828275680542,
      "learning_rate": 2.2963306676874074e-07,
      "loss": 0.3736,
      "step": 27084
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9580703973770142,
      "learning_rate": 2.2939521800430242e-07,
      "loss": 0.2981,
      "step": 27085
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8882027864456177,
      "learning_rate": 2.2915749105366202e-07,
      "loss": 0.3092,
      "step": 27086
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.784433603286743,
      "learning_rate": 2.2891988591978164e-07,
      "loss": 0.3637,
      "step": 27087
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7476770877838135,
      "learning_rate": 2.2868240260562447e-07,
      "loss": 0.3035,
      "step": 27088
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.858337640762329,
      "learning_rate": 2.2844504111415256e-07,
      "loss": 0.2983,
      "step": 27089
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7476567029953003,
      "learning_rate": 2.2820780144832355e-07,
      "loss": 0.3184,
      "step": 27090
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8666996955871582,
      "learning_rate": 2.2797068361109397e-07,
      "loss": 0.3708,
      "step": 27091
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.04732084274292,
      "learning_rate": 2.277336876054237e-07,
      "loss": 0.2778,
      "step": 27092
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0917415618896484,
      "learning_rate": 2.2749681343426476e-07,
      "loss": 0.2649,
      "step": 27093
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0261218547821045,
      "learning_rate": 2.2726006110057152e-07,
      "loss": 0.3269,
      "step": 27094
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1335558891296387,
      "learning_rate": 2.270234306072938e-07,
      "loss": 0.2489,
      "step": 27095
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9639471769332886,
      "learning_rate": 2.2678692195738372e-07,
      "loss": 0.2791,
      "step": 27096
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1533451080322266,
      "learning_rate": 2.2655053515378998e-07,
      "loss": 0.2529,
      "step": 27097
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6339550018310547,
      "learning_rate": 2.2631427019945807e-07,
      "loss": 0.2114,
      "step": 27098
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.7974202632904053,
      "learning_rate": 2.2607812709733555e-07,
      "loss": 0.3745,
      "step": 27099
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9301141500473022,
      "learning_rate": 2.258421058503646e-07,
      "loss": 0.236,
      "step": 27100
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0690677165985107,
      "learning_rate": 2.2560620646148835e-07,
      "loss": 0.2922,
      "step": 27101
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.047395944595337,
      "learning_rate": 2.2537042893365e-07,
      "loss": 0.3701,
      "step": 27102
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6800243854522705,
      "learning_rate": 2.2513477326978617e-07,
      "loss": 0.2466,
      "step": 27103
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.067441940307617,
      "learning_rate": 2.2489923947283555e-07,
      "loss": 0.3238,
      "step": 27104
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.948936939239502,
      "learning_rate": 2.246638275457358e-07,
      "loss": 0.2989,
      "step": 27105
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.1009817123413086,
      "learning_rate": 2.2442853749142124e-07,
      "loss": 0.2715,
      "step": 27106
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.444457530975342,
      "learning_rate": 2.2419336931282509e-07,
      "loss": 0.3631,
      "step": 27107
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0976767539978027,
      "learning_rate": 2.2395832301287835e-07,
      "loss": 0.3408,
      "step": 27108
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.4163079261779785,
      "learning_rate": 2.237233985945142e-07,
      "loss": 0.229,
      "step": 27109
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9402661323547363,
      "learning_rate": 2.234885960606592e-07,
      "loss": 0.2683,
      "step": 27110
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.159916400909424,
      "learning_rate": 2.2325391541424102e-07,
      "loss": 0.3053,
      "step": 27111
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.851747751235962,
      "learning_rate": 2.230193566581862e-07,
      "loss": 0.2536,
      "step": 27112
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.0207345485687256,
      "learning_rate": 2.2278491979541906e-07,
      "loss": 0.3555,
      "step": 27113
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.426161289215088,
      "learning_rate": 2.2255060482886058e-07,
      "loss": 0.3262,
      "step": 27114
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.076575517654419,
      "learning_rate": 2.2231641176143516e-07,
      "loss": 0.2794,
      "step": 27115
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9221245050430298,
      "learning_rate": 2.2208234059606037e-07,
      "loss": 0.3421,
      "step": 27116
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7517458200454712,
      "learning_rate": 2.2184839133565394e-07,
      "loss": 0.3721,
      "step": 27117
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.834229826927185,
      "learning_rate": 2.2161456398313576e-07,
      "loss": 0.2952,
      "step": 27118
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.2706973552703857,
      "learning_rate": 2.2138085854141788e-07,
      "loss": 0.3414,
      "step": 27119
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9330666065216064,
      "learning_rate": 2.2114727501341582e-07,
      "loss": 0.2853,
      "step": 27120
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7880914211273193,
      "learning_rate": 2.2091381340203942e-07,
      "loss": 0.2828,
      "step": 27121
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.5082595348358154,
      "learning_rate": 2.2068047371020196e-07,
      "loss": 0.3029,
      "step": 27122
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.4313981533050537,
      "learning_rate": 2.204472559408122e-07,
      "loss": 0.382,
      "step": 27123
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7422187328338623,
      "learning_rate": 2.2021416009677554e-07,
      "loss": 0.284,
      "step": 27124
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.238414764404297,
      "learning_rate": 2.1998118618100194e-07,
      "loss": 0.283,
      "step": 27125
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5896556377410889,
      "learning_rate": 2.197483341963924e-07,
      "loss": 0.2382,
      "step": 27126
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2405240535736084,
      "learning_rate": 2.1951560414585016e-07,
      "loss": 0.3705,
      "step": 27127
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.365095853805542,
      "learning_rate": 2.1928299603227953e-07,
      "loss": 0.4174,
      "step": 27128
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0426783561706543,
      "learning_rate": 2.1905050985857823e-07,
      "loss": 0.3575,
      "step": 27129
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1467950344085693,
      "learning_rate": 2.1881814562764503e-07,
      "loss": 0.3249,
      "step": 27130
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8369431495666504,
      "learning_rate": 2.1858590334237761e-07,
      "loss": 0.3465,
      "step": 27131
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.1177589893341064,
      "learning_rate": 2.1835378300567145e-07,
      "loss": 0.3349,
      "step": 27132
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2423040866851807,
      "learning_rate": 2.1812178462041977e-07,
      "loss": 0.3362,
      "step": 27133
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.233186721801758,
      "learning_rate": 2.178899081895147e-07,
      "loss": 0.3794,
      "step": 27134
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8510160446166992,
      "learning_rate": 2.1765815371584843e-07,
      "loss": 0.2698,
      "step": 27135
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.6394448280334473,
      "learning_rate": 2.1742652120230968e-07,
      "loss": 0.3613,
      "step": 27136
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.923614501953125,
      "learning_rate": 2.1719501065178615e-07,
      "loss": 0.309,
      "step": 27137
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7844059467315674,
      "learning_rate": 2.1696362206716449e-07,
      "loss": 0.2684,
      "step": 27138
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6843622922897339,
      "learning_rate": 2.16732355451329e-07,
      "loss": 0.2844,
      "step": 27139
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.952624797821045,
      "learning_rate": 2.165012108071618e-07,
      "loss": 0.373,
      "step": 27140
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.223543643951416,
      "learning_rate": 2.1627018813754842e-07,
      "loss": 0.3013,
      "step": 27141
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8170956373214722,
      "learning_rate": 2.160392874453654e-07,
      "loss": 0.2667,
      "step": 27142
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.311192035675049,
      "learning_rate": 2.158085087334927e-07,
      "loss": 0.3229,
      "step": 27143
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8676881790161133,
      "learning_rate": 2.1557785200480907e-07,
      "loss": 0.2877,
      "step": 27144
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2485318183898926,
      "learning_rate": 2.1534731726218784e-07,
      "loss": 0.2367,
      "step": 27145
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8958463668823242,
      "learning_rate": 2.1511690450850554e-07,
      "loss": 0.2183,
      "step": 27146
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9628649950027466,
      "learning_rate": 2.14886613746631e-07,
      "loss": 0.3204,
      "step": 27147
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8561042547225952,
      "learning_rate": 2.1465644497943972e-07,
      "loss": 0.3592,
      "step": 27148
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.6317925453186035,
      "learning_rate": 2.1442639820979826e-07,
      "loss": 0.3195,
      "step": 27149
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.251469135284424,
      "learning_rate": 2.1419647344057547e-07,
      "loss": 0.3091,
      "step": 27150
    },
    {
      "epoch": 2.81,
      "grad_norm": 4.247159957885742,
      "learning_rate": 2.1396667067463905e-07,
      "loss": 0.3197,
      "step": 27151
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0706257820129395,
      "learning_rate": 2.1373698991485336e-07,
      "loss": 0.2815,
      "step": 27152
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9041374921798706,
      "learning_rate": 2.1350743116408056e-07,
      "loss": 0.3231,
      "step": 27153
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.517245888710022,
      "learning_rate": 2.1327799442518504e-07,
      "loss": 0.3116,
      "step": 27154
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0413222312927246,
      "learning_rate": 2.130486797010256e-07,
      "loss": 0.2412,
      "step": 27155
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8855808973312378,
      "learning_rate": 2.128194869944622e-07,
      "loss": 0.3286,
      "step": 27156
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0005931854248047,
      "learning_rate": 2.125904163083514e-07,
      "loss": 0.301,
      "step": 27157
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2266507148742676,
      "learning_rate": 2.1236146764554877e-07,
      "loss": 0.2732,
      "step": 27158
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.568864345550537,
      "learning_rate": 2.1213264100891085e-07,
      "loss": 0.4245,
      "step": 27159
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.997119665145874,
      "learning_rate": 2.119039364012876e-07,
      "loss": 0.3712,
      "step": 27160
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.5919113159179688,
      "learning_rate": 2.116753538255323e-07,
      "loss": 0.3106,
      "step": 27161
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.0584137439727783,
      "learning_rate": 2.1144689328449487e-07,
      "loss": 0.367,
      "step": 27162
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0499355792999268,
      "learning_rate": 2.1121855478102193e-07,
      "loss": 0.3167,
      "step": 27163
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.659221887588501,
      "learning_rate": 2.1099033831796235e-07,
      "loss": 0.2487,
      "step": 27164
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0563485622406006,
      "learning_rate": 2.1076224389816046e-07,
      "loss": 0.2862,
      "step": 27165
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7966535091400146,
      "learning_rate": 2.1053427152445848e-07,
      "loss": 0.2454,
      "step": 27166
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.139348030090332,
      "learning_rate": 2.103064211997008e-07,
      "loss": 0.3534,
      "step": 27167
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9464879035949707,
      "learning_rate": 2.1007869292672846e-07,
      "loss": 0.3437,
      "step": 27168
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.061572790145874,
      "learning_rate": 2.0985108670837806e-07,
      "loss": 0.3408,
      "step": 27169
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6825425624847412,
      "learning_rate": 2.096236025474896e-07,
      "loss": 0.244,
      "step": 27170
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7581613063812256,
      "learning_rate": 2.0939624044689855e-07,
      "loss": 0.2632,
      "step": 27171
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.985023021697998,
      "learning_rate": 2.0916900040943933e-07,
      "loss": 0.2833,
      "step": 27172
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2091431617736816,
      "learning_rate": 2.089418824379441e-07,
      "loss": 0.3272,
      "step": 27173
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.012768268585205,
      "learning_rate": 2.087148865352462e-07,
      "loss": 0.3011,
      "step": 27174
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1956675052642822,
      "learning_rate": 2.084880127041744e-07,
      "loss": 0.2344,
      "step": 27175
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2944905757904053,
      "learning_rate": 2.082612609475576e-07,
      "loss": 0.2719,
      "step": 27176
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.2390542030334473,
      "learning_rate": 2.0803463126822355e-07,
      "loss": 0.318,
      "step": 27177
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0573291778564453,
      "learning_rate": 2.078081236689966e-07,
      "loss": 0.2499,
      "step": 27178
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9006850719451904,
      "learning_rate": 2.075817381527001e-07,
      "loss": 0.2558,
      "step": 27179
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8071496486663818,
      "learning_rate": 2.0735547472215956e-07,
      "loss": 0.3258,
      "step": 27180
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8661909103393555,
      "learning_rate": 2.071293333801927e-07,
      "loss": 0.3095,
      "step": 27181
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.330317258834839,
      "learning_rate": 2.069033141296195e-07,
      "loss": 0.3714,
      "step": 27182
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1312334537506104,
      "learning_rate": 2.0667741697325993e-07,
      "loss": 0.2921,
      "step": 27183
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6914551258087158,
      "learning_rate": 2.064516419139273e-07,
      "loss": 0.2776,
      "step": 27184
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7968240976333618,
      "learning_rate": 2.0622598895443934e-07,
      "loss": 0.2511,
      "step": 27185
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.431464433670044,
      "learning_rate": 2.0600045809760605e-07,
      "loss": 0.2905,
      "step": 27186
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3388727903366089,
      "learning_rate": 2.0577504934624292e-07,
      "loss": 0.2241,
      "step": 27187
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.328176975250244,
      "learning_rate": 2.0554976270315775e-07,
      "loss": 0.3404,
      "step": 27188
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.160341501235962,
      "learning_rate": 2.0532459817115936e-07,
      "loss": 0.3011,
      "step": 27189
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1359429359436035,
      "learning_rate": 2.0509955575305663e-07,
      "loss": 0.3146,
      "step": 27190
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9918501377105713,
      "learning_rate": 2.0487463545165288e-07,
      "loss": 0.2831,
      "step": 27191
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.4398577213287354,
      "learning_rate": 2.0464983726975362e-07,
      "loss": 0.3663,
      "step": 27192
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.712947964668274,
      "learning_rate": 2.044251612101622e-07,
      "loss": 0.2966,
      "step": 27193
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9875860214233398,
      "learning_rate": 2.042006072756786e-07,
      "loss": 0.2742,
      "step": 27194
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9474031925201416,
      "learning_rate": 2.0397617546910275e-07,
      "loss": 0.2708,
      "step": 27195
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7482175827026367,
      "learning_rate": 2.0375186579323248e-07,
      "loss": 0.215,
      "step": 27196
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.2126924991607666,
      "learning_rate": 2.035276782508655e-07,
      "loss": 0.2135,
      "step": 27197
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.386188268661499,
      "learning_rate": 2.0330361284479517e-07,
      "loss": 0.3616,
      "step": 27198
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7595124244689941,
      "learning_rate": 2.030796695778159e-07,
      "loss": 0.2226,
      "step": 27199
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.1123626232147217,
      "learning_rate": 2.0285584845271987e-07,
      "loss": 0.2825,
      "step": 27200
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.989148736000061,
      "learning_rate": 2.0263214947229604e-07,
      "loss": 0.2682,
      "step": 27201
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7570362091064453,
      "learning_rate": 2.0240857263933544e-07,
      "loss": 0.2967,
      "step": 27202
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8673088550567627,
      "learning_rate": 2.0218511795662478e-07,
      "loss": 0.3072,
      "step": 27203
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8548392057418823,
      "learning_rate": 2.0196178542694956e-07,
      "loss": 0.2665,
      "step": 27204
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.803903341293335,
      "learning_rate": 2.0173857505309313e-07,
      "loss": 0.3453,
      "step": 27205
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8883678913116455,
      "learning_rate": 2.0151548683784106e-07,
      "loss": 0.3308,
      "step": 27206
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7703644037246704,
      "learning_rate": 2.0129252078397222e-07,
      "loss": 0.2629,
      "step": 27207
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6836576461791992,
      "learning_rate": 2.0106967689426772e-07,
      "loss": 0.2819,
      "step": 27208
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.220313787460327,
      "learning_rate": 2.0084695517150532e-07,
      "loss": 0.371,
      "step": 27209
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.0506036281585693,
      "learning_rate": 2.0062435561846283e-07,
      "loss": 0.3776,
      "step": 27210
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7173205614089966,
      "learning_rate": 2.0040187823791356e-07,
      "loss": 0.2935,
      "step": 27211
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.894629716873169,
      "learning_rate": 2.00179523032632e-07,
      "loss": 0.294,
      "step": 27212
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.661861538887024,
      "learning_rate": 1.9995729000539143e-07,
      "loss": 0.2005,
      "step": 27213
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.036412477493286,
      "learning_rate": 1.997351791589608e-07,
      "loss": 0.3047,
      "step": 27214
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7959905862808228,
      "learning_rate": 1.9951319049611008e-07,
      "loss": 0.2893,
      "step": 27215
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.943225622177124,
      "learning_rate": 1.9929132401960706e-07,
      "loss": 0.294,
      "step": 27216
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.759133219718933,
      "learning_rate": 1.990695797322184e-07,
      "loss": 0.2361,
      "step": 27217
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9566439390182495,
      "learning_rate": 1.988479576367075e-07,
      "loss": 0.2791,
      "step": 27218
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9602893590927124,
      "learning_rate": 1.9862645773583768e-07,
      "loss": 0.2605,
      "step": 27219
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.974141001701355,
      "learning_rate": 1.9840508003237115e-07,
      "loss": 0.3041,
      "step": 27220
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6398476362228394,
      "learning_rate": 1.981838245290657e-07,
      "loss": 0.2933,
      "step": 27221
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1698269844055176,
      "learning_rate": 1.9796269122868362e-07,
      "loss": 0.3244,
      "step": 27222
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0102663040161133,
      "learning_rate": 1.9774168013397822e-07,
      "loss": 0.2687,
      "step": 27223
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.060671329498291,
      "learning_rate": 1.975207912477073e-07,
      "loss": 0.2917,
      "step": 27224
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.118453025817871,
      "learning_rate": 1.973000245726231e-07,
      "loss": 0.3461,
      "step": 27225
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9013806581497192,
      "learning_rate": 1.970793801114801e-07,
      "loss": 0.2616,
      "step": 27226
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.965408444404602,
      "learning_rate": 1.9685885786702717e-07,
      "loss": 0.346,
      "step": 27227
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.3368475437164307,
      "learning_rate": 1.966384578420133e-07,
      "loss": 0.3497,
      "step": 27228
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9168295860290527,
      "learning_rate": 1.9641818003918956e-07,
      "loss": 0.2335,
      "step": 27229
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.2670979499816895,
      "learning_rate": 1.9619802446129822e-07,
      "loss": 0.4049,
      "step": 27230
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7928615808486938,
      "learning_rate": 1.9597799111108594e-07,
      "loss": 0.3599,
      "step": 27231
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7512469291687012,
      "learning_rate": 1.9575807999129615e-07,
      "loss": 0.2749,
      "step": 27232
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.75943124294281,
      "learning_rate": 1.9553829110467105e-07,
      "loss": 0.267,
      "step": 27233
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.8585245609283447,
      "learning_rate": 1.9531862445394845e-07,
      "loss": 0.2609,
      "step": 27234
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0777933597564697,
      "learning_rate": 1.950990800418706e-07,
      "loss": 0.3564,
      "step": 27235
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1014230251312256,
      "learning_rate": 1.948796578711709e-07,
      "loss": 0.2867,
      "step": 27236
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1978485584259033,
      "learning_rate": 1.9466035794458827e-07,
      "loss": 0.2775,
      "step": 27237
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9473187923431396,
      "learning_rate": 1.9444118026485381e-07,
      "loss": 0.3503,
      "step": 27238
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0086493492126465,
      "learning_rate": 1.9422212483470203e-07,
      "loss": 0.3316,
      "step": 27239
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.03320050239563,
      "learning_rate": 1.940031916568641e-07,
      "loss": 0.2826,
      "step": 27240
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9382108449935913,
      "learning_rate": 1.937843807340678e-07,
      "loss": 0.3783,
      "step": 27241
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6824091672897339,
      "learning_rate": 1.9356569206904319e-07,
      "loss": 0.2725,
      "step": 27242
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.750722885131836,
      "learning_rate": 1.9334712566451584e-07,
      "loss": 0.3137,
      "step": 27243
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0586957931518555,
      "learning_rate": 1.9312868152320917e-07,
      "loss": 0.34,
      "step": 27244
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.8340498208999634,
      "learning_rate": 1.9291035964784987e-07,
      "loss": 0.331,
      "step": 27245
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9583158493041992,
      "learning_rate": 1.9269216004115798e-07,
      "loss": 0.2568,
      "step": 27246
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.4805431365966797,
      "learning_rate": 1.9247408270585245e-07,
      "loss": 0.27,
      "step": 27247
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7318339347839355,
      "learning_rate": 1.922561276446555e-07,
      "loss": 0.2984,
      "step": 27248
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9696153402328491,
      "learning_rate": 1.920382948602817e-07,
      "loss": 0.269,
      "step": 27249
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7817691564559937,
      "learning_rate": 1.918205843554477e-07,
      "loss": 0.2807,
      "step": 27250
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.954237699508667,
      "learning_rate": 1.9160299613286803e-07,
      "loss": 0.3162,
      "step": 27251
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.371825933456421,
      "learning_rate": 1.9138553019525497e-07,
      "loss": 0.2622,
      "step": 27252
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.106316566467285,
      "learning_rate": 1.9116818654532076e-07,
      "loss": 0.3302,
      "step": 27253
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.146897792816162,
      "learning_rate": 1.9095096518577328e-07,
      "loss": 0.2826,
      "step": 27254
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.8973677158355713,
      "learning_rate": 1.9073386611932255e-07,
      "loss": 0.2392,
      "step": 27255
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.311999797821045,
      "learning_rate": 1.905168893486753e-07,
      "loss": 0.2854,
      "step": 27256
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.8126899003982544,
      "learning_rate": 1.903000348765338e-07,
      "loss": 0.3193,
      "step": 27257
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.2144927978515625,
      "learning_rate": 1.9008330270560481e-07,
      "loss": 0.3849,
      "step": 27258
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.838870644569397,
      "learning_rate": 1.898666928385895e-07,
      "loss": 0.2433,
      "step": 27259
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.123955488204956,
      "learning_rate": 1.896502052781879e-07,
      "loss": 0.3482,
      "step": 27260
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.810797095298767,
      "learning_rate": 1.8943384002709897e-07,
      "loss": 0.2395,
      "step": 27261
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.2661898136138916,
      "learning_rate": 1.8921759708802167e-07,
      "loss": 0.2726,
      "step": 27262
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.8783502578735352,
      "learning_rate": 1.8900147646365053e-07,
      "loss": 0.3046,
      "step": 27263
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7643057107925415,
      "learning_rate": 1.8878547815668002e-07,
      "loss": 0.2687,
      "step": 27264
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0104873180389404,
      "learning_rate": 1.8856960216980357e-07,
      "loss": 0.3292,
      "step": 27265
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9841006994247437,
      "learning_rate": 1.8835384850571348e-07,
      "loss": 0.2904,
      "step": 27266
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.2419381141662598,
      "learning_rate": 1.8813821716709647e-07,
      "loss": 0.2904,
      "step": 27267
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6863362789154053,
      "learning_rate": 1.8792270815664483e-07,
      "loss": 0.3036,
      "step": 27268
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.754089117050171,
      "learning_rate": 1.877073214770442e-07,
      "loss": 0.349,
      "step": 27269
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0279769897460938,
      "learning_rate": 1.8749205713097685e-07,
      "loss": 0.2351,
      "step": 27270
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.026310920715332,
      "learning_rate": 1.8727691512113066e-07,
      "loss": 0.308,
      "step": 27271
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0259199142456055,
      "learning_rate": 1.8706189545018682e-07,
      "loss": 0.3453,
      "step": 27272
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9257323741912842,
      "learning_rate": 1.868469981208243e-07,
      "loss": 0.3825,
      "step": 27273
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9708847999572754,
      "learning_rate": 1.8663222313572315e-07,
      "loss": 0.3297,
      "step": 27274
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0727059841156006,
      "learning_rate": 1.8641757049756237e-07,
      "loss": 0.2958,
      "step": 27275
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.3225603103637695,
      "learning_rate": 1.8620304020901758e-07,
      "loss": 0.2847,
      "step": 27276
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.307060480117798,
      "learning_rate": 1.859886322727611e-07,
      "loss": 0.3313,
      "step": 27277
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1478657722473145,
      "learning_rate": 1.8577434669146965e-07,
      "loss": 0.3363,
      "step": 27278
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.3135716915130615,
      "learning_rate": 1.8556018346781225e-07,
      "loss": 0.2782,
      "step": 27279
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9779287576675415,
      "learning_rate": 1.8534614260446003e-07,
      "loss": 0.2723,
      "step": 27280
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.788779616355896,
      "learning_rate": 1.8513222410408093e-07,
      "loss": 0.2987,
      "step": 27281
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.4067509174346924,
      "learning_rate": 1.8491842796934279e-07,
      "loss": 0.35,
      "step": 27282
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.2079286575317383,
      "learning_rate": 1.8470475420291012e-07,
      "loss": 0.3338,
      "step": 27283
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.016392707824707,
      "learning_rate": 1.8449120280744747e-07,
      "loss": 0.2908,
      "step": 27284
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0418994426727295,
      "learning_rate": 1.8427777378561718e-07,
      "loss": 0.3373,
      "step": 27285
    },
    {
      "epoch": 2.82,
      "grad_norm": 3.616788148880005,
      "learning_rate": 1.8406446714007932e-07,
      "loss": 0.3022,
      "step": 27286
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0750558376312256,
      "learning_rate": 1.8385128287349506e-07,
      "loss": 0.3241,
      "step": 27287
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.004696846008301,
      "learning_rate": 1.8363822098852125e-07,
      "loss": 0.3123,
      "step": 27288
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.699161171913147,
      "learning_rate": 1.8342528148781348e-07,
      "loss": 0.2972,
      "step": 27289
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.735082745552063,
      "learning_rate": 1.832124643740274e-07,
      "loss": 0.2891,
      "step": 27290
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.162041664123535,
      "learning_rate": 1.8299976964981646e-07,
      "loss": 0.3251,
      "step": 27291
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.1904449462890625,
      "learning_rate": 1.827871973178319e-07,
      "loss": 0.3903,
      "step": 27292
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.522295355796814,
      "learning_rate": 1.8257474738072377e-07,
      "loss": 0.3293,
      "step": 27293
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.080738067626953,
      "learning_rate": 1.8236241984114222e-07,
      "loss": 0.2944,
      "step": 27294
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.824196457862854,
      "learning_rate": 1.821502147017329e-07,
      "loss": 0.3169,
      "step": 27295
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0000250339508057,
      "learning_rate": 1.819381319651403e-07,
      "loss": 0.266,
      "step": 27296
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6743719577789307,
      "learning_rate": 1.817261716340113e-07,
      "loss": 0.3204,
      "step": 27297
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9166690111160278,
      "learning_rate": 1.8151433371098814e-07,
      "loss": 0.3136,
      "step": 27298
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6189568042755127,
      "learning_rate": 1.8130261819870875e-07,
      "loss": 0.2277,
      "step": 27299
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9612504243850708,
      "learning_rate": 1.810910250998166e-07,
      "loss": 0.3203,
      "step": 27300
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9422794580459595,
      "learning_rate": 1.8087955441694838e-07,
      "loss": 0.2787,
      "step": 27301
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.3456501960754395,
      "learning_rate": 1.8066820615273983e-07,
      "loss": 0.2551,
      "step": 27302
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0662126541137695,
      "learning_rate": 1.804569803098255e-07,
      "loss": 0.317,
      "step": 27303
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7073391675949097,
      "learning_rate": 1.8024587689083994e-07,
      "loss": 0.2743,
      "step": 27304
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.8698015213012695,
      "learning_rate": 1.800348958984155e-07,
      "loss": 0.3014,
      "step": 27305
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.125028133392334,
      "learning_rate": 1.7982403733518005e-07,
      "loss": 0.3094,
      "step": 27306
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.240967035293579,
      "learning_rate": 1.7961330120376597e-07,
      "loss": 0.2397,
      "step": 27307
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.8744723796844482,
      "learning_rate": 1.794026875067989e-07,
      "loss": 0.2582,
      "step": 27308
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9989107847213745,
      "learning_rate": 1.7919219624690344e-07,
      "loss": 0.3133,
      "step": 27309
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9789890050888062,
      "learning_rate": 1.7898182742670522e-07,
      "loss": 0.2775,
      "step": 27310
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.0731279850006104,
      "learning_rate": 1.7877158104882774e-07,
      "loss": 0.3096,
      "step": 27311
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9514119625091553,
      "learning_rate": 1.785614571158889e-07,
      "loss": 0.2587,
      "step": 27312
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.273442029953003,
      "learning_rate": 1.7835145563051326e-07,
      "loss": 0.2765,
      "step": 27313
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.899085283279419,
      "learning_rate": 1.7814157659531538e-07,
      "loss": 0.2819,
      "step": 27314
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7367945909500122,
      "learning_rate": 1.779318200129132e-07,
      "loss": 0.2055,
      "step": 27315
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7642357349395752,
      "learning_rate": 1.7772218588592128e-07,
      "loss": 0.2804,
      "step": 27316
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.839868187904358,
      "learning_rate": 1.7751267421695305e-07,
      "loss": 0.3442,
      "step": 27317
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9840421676635742,
      "learning_rate": 1.7730328500862204e-07,
      "loss": 0.3837,
      "step": 27318
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9530092477798462,
      "learning_rate": 1.7709401826353723e-07,
      "loss": 0.3115,
      "step": 27319
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9431227445602417,
      "learning_rate": 1.7688487398430875e-07,
      "loss": 0.2452,
      "step": 27320
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9961479902267456,
      "learning_rate": 1.7667585217354345e-07,
      "loss": 0.3409,
      "step": 27321
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8049185276031494,
      "learning_rate": 1.7646695283384586e-07,
      "loss": 0.2958,
      "step": 27322
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.826263666152954,
      "learning_rate": 1.762581759678239e-07,
      "loss": 0.2967,
      "step": 27323
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1274073123931885,
      "learning_rate": 1.7604952157807775e-07,
      "loss": 0.3767,
      "step": 27324
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8389348983764648,
      "learning_rate": 1.7584098966720863e-07,
      "loss": 0.3372,
      "step": 27325
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0014638900756836,
      "learning_rate": 1.7563258023781893e-07,
      "loss": 0.2733,
      "step": 27326
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.374070644378662,
      "learning_rate": 1.7542429329250433e-07,
      "loss": 0.3722,
      "step": 27327
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.249546766281128,
      "learning_rate": 1.7521612883386274e-07,
      "loss": 0.269,
      "step": 27328
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.169257402420044,
      "learning_rate": 1.7500808686448878e-07,
      "loss": 0.3431,
      "step": 27329
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.311645984649658,
      "learning_rate": 1.7480016738697703e-07,
      "loss": 0.335,
      "step": 27330
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1079423427581787,
      "learning_rate": 1.7459237040391986e-07,
      "loss": 0.3189,
      "step": 27331
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7706471681594849,
      "learning_rate": 1.743846959179063e-07,
      "loss": 0.2469,
      "step": 27332
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.3129446506500244,
      "learning_rate": 1.7417714393152763e-07,
      "loss": 0.3328,
      "step": 27333
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.3633971214294434,
      "learning_rate": 1.7396971444737065e-07,
      "loss": 0.4031,
      "step": 27334
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9506701231002808,
      "learning_rate": 1.7376240746801997e-07,
      "loss": 0.3124,
      "step": 27335
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9238511323928833,
      "learning_rate": 1.7355522299606242e-07,
      "loss": 0.2404,
      "step": 27336
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.064608335494995,
      "learning_rate": 1.7334816103408037e-07,
      "loss": 0.3181,
      "step": 27337
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0593645572662354,
      "learning_rate": 1.7314122158465397e-07,
      "loss": 0.2677,
      "step": 27338
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7653497457504272,
      "learning_rate": 1.729344046503656e-07,
      "loss": 0.3,
      "step": 27339
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.4822754859924316,
      "learning_rate": 1.727277102337921e-07,
      "loss": 0.2745,
      "step": 27340
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.828883171081543,
      "learning_rate": 1.7252113833751138e-07,
      "loss": 0.2932,
      "step": 27341
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7448724508285522,
      "learning_rate": 1.7231468896409697e-07,
      "loss": 0.3124,
      "step": 27342
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9433759450912476,
      "learning_rate": 1.7210836211612458e-07,
      "loss": 0.3298,
      "step": 27343
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.497262716293335,
      "learning_rate": 1.719021577961666e-07,
      "loss": 0.3279,
      "step": 27344
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6532620191574097,
      "learning_rate": 1.7169607600679205e-07,
      "loss": 0.3207,
      "step": 27345
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9868041276931763,
      "learning_rate": 1.7149011675057226e-07,
      "loss": 0.329,
      "step": 27346
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.878653883934021,
      "learning_rate": 1.7128428003007403e-07,
      "loss": 0.2505,
      "step": 27347
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.2642390727996826,
      "learning_rate": 1.710785658478631e-07,
      "loss": 0.29,
      "step": 27348
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8883376121520996,
      "learning_rate": 1.7087297420650627e-07,
      "loss": 0.3166,
      "step": 27349
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.065753221511841,
      "learning_rate": 1.7066750510856488e-07,
      "loss": 0.271,
      "step": 27350
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.28507399559021,
      "learning_rate": 1.704621585566002e-07,
      "loss": 0.3067,
      "step": 27351
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8990354537963867,
      "learning_rate": 1.702569345531735e-07,
      "loss": 0.3587,
      "step": 27352
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.400045156478882,
      "learning_rate": 1.7005183310084382e-07,
      "loss": 0.3115,
      "step": 27353
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1048412322998047,
      "learning_rate": 1.6984685420216696e-07,
      "loss": 0.3882,
      "step": 27354
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.370079278945923,
      "learning_rate": 1.6964199785969749e-07,
      "loss": 0.2967,
      "step": 27355
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0913145542144775,
      "learning_rate": 1.6943726407599338e-07,
      "loss": 0.3885,
      "step": 27356
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1581430435180664,
      "learning_rate": 1.6923265285360367e-07,
      "loss": 0.3254,
      "step": 27357
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0733773708343506,
      "learning_rate": 1.6902816419507972e-07,
      "loss": 0.3282,
      "step": 27358
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.971982479095459,
      "learning_rate": 1.6882379810297166e-07,
      "loss": 0.3183,
      "step": 27359
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9840871095657349,
      "learning_rate": 1.686195545798286e-07,
      "loss": 0.2752,
      "step": 27360
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0547890663146973,
      "learning_rate": 1.6841543362819402e-07,
      "loss": 0.3386,
      "step": 27361
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0870909690856934,
      "learning_rate": 1.6821143525061477e-07,
      "loss": 0.2981,
      "step": 27362
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8711878061294556,
      "learning_rate": 1.6800755944963444e-07,
      "loss": 0.3563,
      "step": 27363
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1913998126983643,
      "learning_rate": 1.6780380622779312e-07,
      "loss": 0.2858,
      "step": 27364
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7761443853378296,
      "learning_rate": 1.676001755876333e-07,
      "loss": 0.2962,
      "step": 27365
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5812808275222778,
      "learning_rate": 1.6739666753169291e-07,
      "loss": 0.2445,
      "step": 27366
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9712426662445068,
      "learning_rate": 1.6719328206250772e-07,
      "loss": 0.2309,
      "step": 27367
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0389888286590576,
      "learning_rate": 1.669900191826146e-07,
      "loss": 0.3226,
      "step": 27368
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.7864317893981934,
      "learning_rate": 1.6678687889454814e-07,
      "loss": 0.4193,
      "step": 27369
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1082968711853027,
      "learning_rate": 1.665838612008397e-07,
      "loss": 0.2959,
      "step": 27370
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0884790420532227,
      "learning_rate": 1.6638096610402054e-07,
      "loss": 0.3154,
      "step": 27371
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8508238792419434,
      "learning_rate": 1.6617819360662203e-07,
      "loss": 0.2755,
      "step": 27372
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.919632911682129,
      "learning_rate": 1.6597554371117097e-07,
      "loss": 0.2745,
      "step": 27373
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8685349225997925,
      "learning_rate": 1.6577301642019316e-07,
      "loss": 0.2879,
      "step": 27374
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.245093584060669,
      "learning_rate": 1.6557061173621548e-07,
      "loss": 0.2966,
      "step": 27375
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8218876123428345,
      "learning_rate": 1.6536832966175919e-07,
      "loss": 0.3444,
      "step": 27376
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9490879774093628,
      "learning_rate": 1.651661701993468e-07,
      "loss": 0.1888,
      "step": 27377
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.032552480697632,
      "learning_rate": 1.6496413335149954e-07,
      "loss": 0.3077,
      "step": 27378
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.894802451133728,
      "learning_rate": 1.6476221912073653e-07,
      "loss": 0.3173,
      "step": 27379
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1027016639709473,
      "learning_rate": 1.6456042750957468e-07,
      "loss": 0.2282,
      "step": 27380
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9751644134521484,
      "learning_rate": 1.6435875852052863e-07,
      "loss": 0.3062,
      "step": 27381
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8544418811798096,
      "learning_rate": 1.6415721215611413e-07,
      "loss": 0.259,
      "step": 27382
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0342042446136475,
      "learning_rate": 1.6395578841884362e-07,
      "loss": 0.2569,
      "step": 27383
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.260850429534912,
      "learning_rate": 1.6375448731122623e-07,
      "loss": 0.2836,
      "step": 27384
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1982192993164062,
      "learning_rate": 1.6355330883577547e-07,
      "loss": 0.2996,
      "step": 27385
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.3300671577453613,
      "learning_rate": 1.6335225299499712e-07,
      "loss": 0.2944,
      "step": 27386
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.843451738357544,
      "learning_rate": 1.631513197913981e-07,
      "loss": 0.3345,
      "step": 27387
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0525918006896973,
      "learning_rate": 1.6295050922748412e-07,
      "loss": 0.2751,
      "step": 27388
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.4370203018188477,
      "learning_rate": 1.627498213057588e-07,
      "loss": 0.2394,
      "step": 27389
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9868336915969849,
      "learning_rate": 1.625492560287223e-07,
      "loss": 0.2604,
      "step": 27390
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.3154566287994385,
      "learning_rate": 1.6234881339887708e-07,
      "loss": 0.365,
      "step": 27391
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8043266534805298,
      "learning_rate": 1.6214849341872229e-07,
      "loss": 0.2776,
      "step": 27392
    },
    {
      "epoch": 2.83,
      "grad_norm": 3.3507115840911865,
      "learning_rate": 1.619482960907548e-07,
      "loss": 0.3387,
      "step": 27393
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.186674118041992,
      "learning_rate": 1.6174822141746927e-07,
      "loss": 0.2845,
      "step": 27394
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8702478408813477,
      "learning_rate": 1.615482694013626e-07,
      "loss": 0.2956,
      "step": 27395
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.712603211402893,
      "learning_rate": 1.613484400449261e-07,
      "loss": 0.2539,
      "step": 27396
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7630302906036377,
      "learning_rate": 1.6114873335065006e-07,
      "loss": 0.251,
      "step": 27397
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7582861185073853,
      "learning_rate": 1.609491493210269e-07,
      "loss": 0.3603,
      "step": 27398
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.287426471710205,
      "learning_rate": 1.6074968795854352e-07,
      "loss": 0.3296,
      "step": 27399
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.4622802734375,
      "learning_rate": 1.605503492656868e-07,
      "loss": 0.2967,
      "step": 27400
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.1562108993530273,
      "learning_rate": 1.6035113324494256e-07,
      "loss": 0.3195,
      "step": 27401
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9668314456939697,
      "learning_rate": 1.6015203989879325e-07,
      "loss": 0.2582,
      "step": 27402
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.828765869140625,
      "learning_rate": 1.599530692297213e-07,
      "loss": 0.2212,
      "step": 27403
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.3296542167663574,
      "learning_rate": 1.597542212402081e-07,
      "loss": 0.3172,
      "step": 27404
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.4216132164001465,
      "learning_rate": 1.5955549593273277e-07,
      "loss": 0.3894,
      "step": 27405
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9565789699554443,
      "learning_rate": 1.5935689330977223e-07,
      "loss": 0.3567,
      "step": 27406
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6956522464752197,
      "learning_rate": 1.5915841337380333e-07,
      "loss": 0.2331,
      "step": 27407
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.0827994346618652,
      "learning_rate": 1.589600561272997e-07,
      "loss": 0.3165,
      "step": 27408
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7548201084136963,
      "learning_rate": 1.587618215727349e-07,
      "loss": 0.2631,
      "step": 27409
    },
    {
      "epoch": 2.83,
      "grad_norm": 2.309548854827881,
      "learning_rate": 1.5856370971257917e-07,
      "loss": 0.3205,
      "step": 27410
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9897652864456177,
      "learning_rate": 1.58365720549305e-07,
      "loss": 0.37,
      "step": 27411
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.0819222927093506,
      "learning_rate": 1.5816785408537927e-07,
      "loss": 0.3505,
      "step": 27412
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1552412509918213,
      "learning_rate": 1.579701103232678e-07,
      "loss": 0.2478,
      "step": 27413
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.3122596740722656,
      "learning_rate": 1.5777248926543754e-07,
      "loss": 0.3478,
      "step": 27414
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.263221263885498,
      "learning_rate": 1.5757499091435203e-07,
      "loss": 0.3619,
      "step": 27415
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.3258485794067383,
      "learning_rate": 1.5737761527247263e-07,
      "loss": 0.2975,
      "step": 27416
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9317615032196045,
      "learning_rate": 1.5718036234226187e-07,
      "loss": 0.2588,
      "step": 27417
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9085547924041748,
      "learning_rate": 1.5698323212617773e-07,
      "loss": 0.2978,
      "step": 27418
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9398281574249268,
      "learning_rate": 1.5678622462667826e-07,
      "loss": 0.2746,
      "step": 27419
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6022595167160034,
      "learning_rate": 1.5658933984621817e-07,
      "loss": 0.2998,
      "step": 27420
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7978293895721436,
      "learning_rate": 1.5639257778725436e-07,
      "loss": 0.3458,
      "step": 27421
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6857682466506958,
      "learning_rate": 1.5619593845223935e-07,
      "loss": 0.2397,
      "step": 27422
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1758530139923096,
      "learning_rate": 1.5599942184362339e-07,
      "loss": 0.2233,
      "step": 27423
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9874212741851807,
      "learning_rate": 1.5580302796385782e-07,
      "loss": 0.3346,
      "step": 27424
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1647861003875732,
      "learning_rate": 1.5560675681539072e-07,
      "loss": 0.2827,
      "step": 27425
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.095003366470337,
      "learning_rate": 1.5541060840066903e-07,
      "loss": 0.2708,
      "step": 27426
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8664929866790771,
      "learning_rate": 1.5521458272213852e-07,
      "loss": 0.2657,
      "step": 27427
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1578447818756104,
      "learning_rate": 1.5501867978224395e-07,
      "loss": 0.2893,
      "step": 27428
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.2839441299438477,
      "learning_rate": 1.5482289958342445e-07,
      "loss": 0.1529,
      "step": 27429
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.439877986907959,
      "learning_rate": 1.5462724212812475e-07,
      "loss": 0.3244,
      "step": 27430
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9193068742752075,
      "learning_rate": 1.5443170741878178e-07,
      "loss": 0.2933,
      "step": 27431
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.10353684425354,
      "learning_rate": 1.5423629545783358e-07,
      "loss": 0.3145,
      "step": 27432
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.2041311264038086,
      "learning_rate": 1.5404100624771822e-07,
      "loss": 0.374,
      "step": 27433
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9534701108932495,
      "learning_rate": 1.538458397908693e-07,
      "loss": 0.281,
      "step": 27434
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9736175537109375,
      "learning_rate": 1.536507960897182e-07,
      "loss": 0.2925,
      "step": 27435
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1040518283843994,
      "learning_rate": 1.534558751466997e-07,
      "loss": 0.2736,
      "step": 27436
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.4261183738708496,
      "learning_rate": 1.5326107696424285e-07,
      "loss": 0.3355,
      "step": 27437
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.818314552307129,
      "learning_rate": 1.530664015447747e-07,
      "loss": 0.277,
      "step": 27438
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7337405681610107,
      "learning_rate": 1.5287184889072436e-07,
      "loss": 0.2675,
      "step": 27439
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.015465259552002,
      "learning_rate": 1.5267741900451773e-07,
      "loss": 0.3063,
      "step": 27440
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.0401721000671387,
      "learning_rate": 1.524831118885761e-07,
      "loss": 0.3196,
      "step": 27441
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.3479065895080566,
      "learning_rate": 1.522889275453243e-07,
      "loss": 0.3064,
      "step": 27442
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.895019769668579,
      "learning_rate": 1.520948659771826e-07,
      "loss": 0.3609,
      "step": 27443
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8409489393234253,
      "learning_rate": 1.5190092718657124e-07,
      "loss": 0.3005,
      "step": 27444
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.085308790206909,
      "learning_rate": 1.517071111759061e-07,
      "loss": 0.2905,
      "step": 27445
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7496342658996582,
      "learning_rate": 1.515134179476052e-07,
      "loss": 0.2535,
      "step": 27446
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.507662057876587,
      "learning_rate": 1.5131984750408334e-07,
      "loss": 0.2886,
      "step": 27447
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.036317825317383,
      "learning_rate": 1.5112639984775302e-07,
      "loss": 0.2683,
      "step": 27448
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5550905466079712,
      "learning_rate": 1.5093307498102673e-07,
      "loss": 0.2208,
      "step": 27449
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7417166233062744,
      "learning_rate": 1.5073987290631474e-07,
      "loss": 0.2614,
      "step": 27450
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1798157691955566,
      "learning_rate": 1.505467936260252e-07,
      "loss": 0.3572,
      "step": 27451
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.2128357887268066,
      "learning_rate": 1.5035383714256612e-07,
      "loss": 0.3429,
      "step": 27452
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.910640835762024,
      "learning_rate": 1.501610034583423e-07,
      "loss": 0.3153,
      "step": 27453
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7208892107009888,
      "learning_rate": 1.499682925757584e-07,
      "loss": 0.3189,
      "step": 27454
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9831492900848389,
      "learning_rate": 1.49775704497217e-07,
      "loss": 0.2882,
      "step": 27455
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.823203206062317,
      "learning_rate": 1.4958323922511842e-07,
      "loss": 0.2871,
      "step": 27456
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.040508508682251,
      "learning_rate": 1.4939089676186403e-07,
      "loss": 0.3527,
      "step": 27457
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.2926290035247803,
      "learning_rate": 1.4919867710984858e-07,
      "loss": 0.3262,
      "step": 27458
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.896472454071045,
      "learning_rate": 1.490065802714724e-07,
      "loss": 0.2897,
      "step": 27459
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8141909837722778,
      "learning_rate": 1.4881460624912803e-07,
      "loss": 0.2564,
      "step": 27460
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9174681901931763,
      "learning_rate": 1.4862275504520906e-07,
      "loss": 0.2637,
      "step": 27461
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.0169670581817627,
      "learning_rate": 1.4843102666210917e-07,
      "loss": 0.2998,
      "step": 27462
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.144454002380371,
      "learning_rate": 1.4823942110221644e-07,
      "loss": 0.3196,
      "step": 27463
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7972958087921143,
      "learning_rate": 1.4804793836792008e-07,
      "loss": 0.3171,
      "step": 27464
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9984887838363647,
      "learning_rate": 1.4785657846160816e-07,
      "loss": 0.2765,
      "step": 27465
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8640309572219849,
      "learning_rate": 1.4766534138566658e-07,
      "loss": 0.3591,
      "step": 27466
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.2025301456451416,
      "learning_rate": 1.4747422714247895e-07,
      "loss": 0.2212,
      "step": 27467
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8825159072875977,
      "learning_rate": 1.4728323573442782e-07,
      "loss": 0.3525,
      "step": 27468
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8404823541641235,
      "learning_rate": 1.4709236716389464e-07,
      "loss": 0.2887,
      "step": 27469
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8233716487884521,
      "learning_rate": 1.469016214332597e-07,
      "loss": 0.3095,
      "step": 27470
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1256637573242188,
      "learning_rate": 1.4671099854490002e-07,
      "loss": 0.3908,
      "step": 27471
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.16412353515625,
      "learning_rate": 1.4652049850119254e-07,
      "loss": 0.3146,
      "step": 27472
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1025571823120117,
      "learning_rate": 1.463301213045132e-07,
      "loss": 0.3409,
      "step": 27473
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.86641263961792,
      "learning_rate": 1.4613986695723336e-07,
      "loss": 0.2532,
      "step": 27474
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8299156427383423,
      "learning_rate": 1.4594973546172674e-07,
      "loss": 0.2647,
      "step": 27475
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1583967208862305,
      "learning_rate": 1.4575972682036365e-07,
      "loss": 0.2932,
      "step": 27476
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8403871059417725,
      "learning_rate": 1.455698410355122e-07,
      "loss": 0.2929,
      "step": 27477
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6760108470916748,
      "learning_rate": 1.4538007810954157e-07,
      "loss": 0.2785,
      "step": 27478
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9622447490692139,
      "learning_rate": 1.4519043804481547e-07,
      "loss": 0.3234,
      "step": 27479
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8816324472427368,
      "learning_rate": 1.4500092084369865e-07,
      "loss": 0.267,
      "step": 27480
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.4729886054992676,
      "learning_rate": 1.4481152650855478e-07,
      "loss": 0.291,
      "step": 27481
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.055044651031494,
      "learning_rate": 1.446222550417442e-07,
      "loss": 0.365,
      "step": 27482
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9307249784469604,
      "learning_rate": 1.4443310644562726e-07,
      "loss": 0.2829,
      "step": 27483
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7338268756866455,
      "learning_rate": 1.4424408072256201e-07,
      "loss": 0.2398,
      "step": 27484
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.909946322441101,
      "learning_rate": 1.4405517787490553e-07,
      "loss": 0.346,
      "step": 27485
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.128411293029785,
      "learning_rate": 1.4386639790501144e-07,
      "loss": 0.3047,
      "step": 27486
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.8585281372070312,
      "learning_rate": 1.4367774081523454e-07,
      "loss": 0.3931,
      "step": 27487
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.619896650314331,
      "learning_rate": 1.4348920660792743e-07,
      "loss": 0.3032,
      "step": 27488
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9301412105560303,
      "learning_rate": 1.4330079528543927e-07,
      "loss": 0.2864,
      "step": 27489
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7516242265701294,
      "learning_rate": 1.4311250685011935e-07,
      "loss": 0.2668,
      "step": 27490
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3243322372436523,
      "learning_rate": 1.4292434130431687e-07,
      "loss": 0.1833,
      "step": 27491
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8796812295913696,
      "learning_rate": 1.4273629865037552e-07,
      "loss": 0.2903,
      "step": 27492
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.973550796508789,
      "learning_rate": 1.4254837889064012e-07,
      "loss": 0.2676,
      "step": 27493
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9773905277252197,
      "learning_rate": 1.4236058202745428e-07,
      "loss": 0.275,
      "step": 27494
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.0107035636901855,
      "learning_rate": 1.4217290806315843e-07,
      "loss": 0.3148,
      "step": 27495
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.0860302448272705,
      "learning_rate": 1.4198535700009398e-07,
      "loss": 0.288,
      "step": 27496
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8918803930282593,
      "learning_rate": 1.4179792884059683e-07,
      "loss": 0.3354,
      "step": 27497
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8103770017623901,
      "learning_rate": 1.416106235870063e-07,
      "loss": 0.2701,
      "step": 27498
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.914425015449524,
      "learning_rate": 1.4142344124165598e-07,
      "loss": 0.3293,
      "step": 27499
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9045674800872803,
      "learning_rate": 1.412363818068796e-07,
      "loss": 0.2723,
      "step": 27500
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.0658090114593506,
      "learning_rate": 1.4104944528500973e-07,
      "loss": 0.2957,
      "step": 27501
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8337746858596802,
      "learning_rate": 1.4086263167837677e-07,
      "loss": 0.2956,
      "step": 27502
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.1841201782226562,
      "learning_rate": 1.4067594098930992e-07,
      "loss": 0.2002,
      "step": 27503
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.9810165166854858,
      "learning_rate": 1.4048937322013734e-07,
      "loss": 0.2663,
      "step": 27504
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8427382707595825,
      "learning_rate": 1.4030292837318383e-07,
      "loss": 0.265,
      "step": 27505
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8148179054260254,
      "learning_rate": 1.401166064507742e-07,
      "loss": 0.2594,
      "step": 27506
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.2134857177734375,
      "learning_rate": 1.3993040745523212e-07,
      "loss": 0.2771,
      "step": 27507
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7583777904510498,
      "learning_rate": 1.39744331388878e-07,
      "loss": 0.2517,
      "step": 27508
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.408524751663208,
      "learning_rate": 1.3955837825403217e-07,
      "loss": 0.2874,
      "step": 27509
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.2336387634277344,
      "learning_rate": 1.3937254805301282e-07,
      "loss": 0.3083,
      "step": 27510
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.2281923294067383,
      "learning_rate": 1.3918684078813805e-07,
      "loss": 0.2907,
      "step": 27511
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.755429983139038,
      "learning_rate": 1.3900125646172157e-07,
      "loss": 0.2987,
      "step": 27512
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.169816017150879,
      "learning_rate": 1.3881579507607713e-07,
      "loss": 0.4211,
      "step": 27513
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0068752765655518,
      "learning_rate": 1.3863045663351725e-07,
      "loss": 0.3535,
      "step": 27514
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.152543306350708,
      "learning_rate": 1.384452411363535e-07,
      "loss": 0.3516,
      "step": 27515
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8598066568374634,
      "learning_rate": 1.3826014858689508e-07,
      "loss": 0.3148,
      "step": 27516
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8933815956115723,
      "learning_rate": 1.3807517898744682e-07,
      "loss": 0.3104,
      "step": 27517
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7989658117294312,
      "learning_rate": 1.3789033234031802e-07,
      "loss": 0.2597,
      "step": 27518
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.162588119506836,
      "learning_rate": 1.3770560864781234e-07,
      "loss": 0.3328,
      "step": 27519
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.873429298400879,
      "learning_rate": 1.3752100791223133e-07,
      "loss": 0.3357,
      "step": 27520
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.390622615814209,
      "learning_rate": 1.3733653013587867e-07,
      "loss": 0.2716,
      "step": 27521
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.379257917404175,
      "learning_rate": 1.3715217532105363e-07,
      "loss": 0.2816,
      "step": 27522
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.623237133026123,
      "learning_rate": 1.3696794347005328e-07,
      "loss": 0.2359,
      "step": 27523
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0460312366485596,
      "learning_rate": 1.3678383458517686e-07,
      "loss": 0.2781,
      "step": 27524
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8664445877075195,
      "learning_rate": 1.36599848668717e-07,
      "loss": 0.3007,
      "step": 27525
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9293949604034424,
      "learning_rate": 1.3641598572296967e-07,
      "loss": 0.2775,
      "step": 27526
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8315593004226685,
      "learning_rate": 1.362322457502263e-07,
      "loss": 0.3037,
      "step": 27527
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8864668607711792,
      "learning_rate": 1.3604862875277735e-07,
      "loss": 0.2745,
      "step": 27528
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9719098806381226,
      "learning_rate": 1.358651347329132e-07,
      "loss": 0.3168,
      "step": 27529
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.1431572437286377,
      "learning_rate": 1.3568176369291975e-07,
      "loss": 0.3227,
      "step": 27530
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.469639539718628,
      "learning_rate": 1.3549851563508522e-07,
      "loss": 0.3219,
      "step": 27531
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8945235013961792,
      "learning_rate": 1.353153905616933e-07,
      "loss": 0.3199,
      "step": 27532
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9084550142288208,
      "learning_rate": 1.3513238847502553e-07,
      "loss": 0.2655,
      "step": 27533
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.243307113647461,
      "learning_rate": 1.3494950937736672e-07,
      "loss": 0.3336,
      "step": 27534
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.398186683654785,
      "learning_rate": 1.3476675327099399e-07,
      "loss": 0.3368,
      "step": 27535
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.04886531829834,
      "learning_rate": 1.3458412015818655e-07,
      "loss": 0.2565,
      "step": 27536
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8031320571899414,
      "learning_rate": 1.3440161004122266e-07,
      "loss": 0.2658,
      "step": 27537
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0037448406219482,
      "learning_rate": 1.3421922292237711e-07,
      "loss": 0.3642,
      "step": 27538
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.078380584716797,
      "learning_rate": 1.3403695880392144e-07,
      "loss": 0.3073,
      "step": 27539
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9232877492904663,
      "learning_rate": 1.3385481768813158e-07,
      "loss": 0.2185,
      "step": 27540
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.120934247970581,
      "learning_rate": 1.3367279957727796e-07,
      "loss": 0.3198,
      "step": 27541
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.001265287399292,
      "learning_rate": 1.3349090447362767e-07,
      "loss": 0.3475,
      "step": 27542
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.058793306350708,
      "learning_rate": 1.3330913237944887e-07,
      "loss": 0.3188,
      "step": 27543
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.1502225399017334,
      "learning_rate": 1.3312748329700864e-07,
      "loss": 0.2932,
      "step": 27544
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.1008450984954834,
      "learning_rate": 1.3294595722857185e-07,
      "loss": 0.3258,
      "step": 27545
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.557318091392517,
      "learning_rate": 1.327645541764011e-07,
      "loss": 0.2369,
      "step": 27546
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9640880823135376,
      "learning_rate": 1.3258327414275796e-07,
      "loss": 0.3505,
      "step": 27547
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.180417060852051,
      "learning_rate": 1.324021171299039e-07,
      "loss": 0.2367,
      "step": 27548
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.946253776550293,
      "learning_rate": 1.3222108314009497e-07,
      "loss": 0.2543,
      "step": 27549
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7507522106170654,
      "learning_rate": 1.320401721755904e-07,
      "loss": 0.2662,
      "step": 27550
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.435345411300659,
      "learning_rate": 1.3185938423864397e-07,
      "loss": 0.3193,
      "step": 27551
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.6436562538146973,
      "learning_rate": 1.3167871933151055e-07,
      "loss": 0.3097,
      "step": 27552
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0298409461975098,
      "learning_rate": 1.3149817745644388e-07,
      "loss": 0.3899,
      "step": 27553
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9953275918960571,
      "learning_rate": 1.313177586156922e-07,
      "loss": 0.2926,
      "step": 27554
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0177857875823975,
      "learning_rate": 1.3113746281150696e-07,
      "loss": 0.2451,
      "step": 27555
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.904862403869629,
      "learning_rate": 1.309572900461331e-07,
      "loss": 0.2071,
      "step": 27556
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.22670841217041,
      "learning_rate": 1.3077724032182104e-07,
      "loss": 0.3517,
      "step": 27557
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.932671308517456,
      "learning_rate": 1.3059731364081342e-07,
      "loss": 0.2917,
      "step": 27558
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7234864234924316,
      "learning_rate": 1.3041751000535173e-07,
      "loss": 0.2931,
      "step": 27559
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9341124296188354,
      "learning_rate": 1.3023782941768094e-07,
      "loss": 0.3705,
      "step": 27560
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.1153738498687744,
      "learning_rate": 1.3005827188003916e-07,
      "loss": 0.312,
      "step": 27561
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9177392721176147,
      "learning_rate": 1.2987883739466466e-07,
      "loss": 0.2572,
      "step": 27562
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.237800121307373,
      "learning_rate": 1.2969952596379565e-07,
      "loss": 0.2299,
      "step": 27563
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.4952774047851562,
      "learning_rate": 1.295203375896681e-07,
      "loss": 0.3659,
      "step": 27564
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9037784337997437,
      "learning_rate": 1.2934127227451464e-07,
      "loss": 0.2746,
      "step": 27565
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.919419288635254,
      "learning_rate": 1.29162330020568e-07,
      "loss": 0.3778,
      "step": 27566
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.3712151050567627,
      "learning_rate": 1.2898351083006077e-07,
      "loss": 0.3489,
      "step": 27567
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.298146963119507,
      "learning_rate": 1.288048147052201e-07,
      "loss": 0.2745,
      "step": 27568
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.522629737854004,
      "learning_rate": 1.286262416482742e-07,
      "loss": 0.2496,
      "step": 27569
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.051104784011841,
      "learning_rate": 1.2844779166145128e-07,
      "loss": 0.3005,
      "step": 27570
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9410066604614258,
      "learning_rate": 1.282694647469751e-07,
      "loss": 0.3398,
      "step": 27571
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.1088476181030273,
      "learning_rate": 1.2809126090706725e-07,
      "loss": 0.2536,
      "step": 27572
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.5427491664886475,
      "learning_rate": 1.2791318014395148e-07,
      "loss": 0.3918,
      "step": 27573
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9494061470031738,
      "learning_rate": 1.2773522245984827e-07,
      "loss": 0.3728,
      "step": 27574
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.038527488708496,
      "learning_rate": 1.275573878569747e-07,
      "loss": 0.351,
      "step": 27575
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7514598369598389,
      "learning_rate": 1.27379676337549e-07,
      "loss": 0.3235,
      "step": 27576
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6923240423202515,
      "learning_rate": 1.272020879037861e-07,
      "loss": 0.282,
      "step": 27577
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9894155263900757,
      "learning_rate": 1.2702462255789973e-07,
      "loss": 0.3326,
      "step": 27578
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.043046474456787,
      "learning_rate": 1.268472803021048e-07,
      "loss": 0.2679,
      "step": 27579
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.319202423095703,
      "learning_rate": 1.2667006113860957e-07,
      "loss": 0.2646,
      "step": 27580
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9253419637680054,
      "learning_rate": 1.264929650696245e-07,
      "loss": 0.3901,
      "step": 27581
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8085488080978394,
      "learning_rate": 1.2631599209735778e-07,
      "loss": 0.2495,
      "step": 27582
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.906732201576233,
      "learning_rate": 1.2613914222401547e-07,
      "loss": 0.2938,
      "step": 27583
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9328432083129883,
      "learning_rate": 1.2596241545180242e-07,
      "loss": 0.2727,
      "step": 27584
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0623865127563477,
      "learning_rate": 1.2578581178292138e-07,
      "loss": 0.3074,
      "step": 27585
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.3359792232513428,
      "learning_rate": 1.256093312195761e-07,
      "loss": 0.299,
      "step": 27586
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7770956754684448,
      "learning_rate": 1.254329737639648e-07,
      "loss": 0.3026,
      "step": 27587
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.2018158435821533,
      "learning_rate": 1.2525673941828687e-07,
      "loss": 0.3278,
      "step": 27588
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0471200942993164,
      "learning_rate": 1.2508062818473943e-07,
      "loss": 0.3258,
      "step": 27589
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7480443716049194,
      "learning_rate": 1.249046400655185e-07,
      "loss": 0.2495,
      "step": 27590
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.180421829223633,
      "learning_rate": 1.2472877506281788e-07,
      "loss": 0.3129,
      "step": 27591
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6614148616790771,
      "learning_rate": 1.2455303317883029e-07,
      "loss": 0.194,
      "step": 27592
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.2446141242980957,
      "learning_rate": 1.2437741441574613e-07,
      "loss": 0.3284,
      "step": 27593
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8835593461990356,
      "learning_rate": 1.2420191877575593e-07,
      "loss": 0.2684,
      "step": 27594
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.4005589485168457,
      "learning_rate": 1.2402654626104682e-07,
      "loss": 0.3656,
      "step": 27595
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.2338578701019287,
      "learning_rate": 1.2385129687380592e-07,
      "loss": 0.2823,
      "step": 27596
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.119022846221924,
      "learning_rate": 1.2367617061621706e-07,
      "loss": 0.2715,
      "step": 27597
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9855502843856812,
      "learning_rate": 1.2350116749046515e-07,
      "loss": 0.3373,
      "step": 27598
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8889578580856323,
      "learning_rate": 1.2332628749873065e-07,
      "loss": 0.3475,
      "step": 27599
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.1443216800689697,
      "learning_rate": 1.231515306431952e-07,
      "loss": 0.2881,
      "step": 27600
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0769619941711426,
      "learning_rate": 1.2297689692603475e-07,
      "loss": 0.3339,
      "step": 27601
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0523908138275146,
      "learning_rate": 1.22802386349431e-07,
      "loss": 0.2916,
      "step": 27602
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.160552740097046,
      "learning_rate": 1.2262799891555653e-07,
      "loss": 0.3036,
      "step": 27603
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.028109073638916,
      "learning_rate": 1.2245373462658527e-07,
      "loss": 0.3611,
      "step": 27604
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.16342830657959,
      "learning_rate": 1.2227959348469098e-07,
      "loss": 0.3312,
      "step": 27605
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9963997602462769,
      "learning_rate": 1.2210557549204415e-07,
      "loss": 0.3282,
      "step": 27606
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9070290327072144,
      "learning_rate": 1.219316806508164e-07,
      "loss": 0.2837,
      "step": 27607
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.359083890914917,
      "learning_rate": 1.2175790896317152e-07,
      "loss": 0.2343,
      "step": 27608
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9977003335952759,
      "learning_rate": 1.2158426043128e-07,
      "loss": 0.3777,
      "step": 27609
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9042750597000122,
      "learning_rate": 1.2141073505730572e-07,
      "loss": 0.3201,
      "step": 27610
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0622124671936035,
      "learning_rate": 1.2123733284341022e-07,
      "loss": 0.2691,
      "step": 27611
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.2172486782073975,
      "learning_rate": 1.2106405379175734e-07,
      "loss": 0.333,
      "step": 27612
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8030428886413574,
      "learning_rate": 1.2089089790450648e-07,
      "loss": 0.3144,
      "step": 27613
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1153178215026855,
      "learning_rate": 1.20717865183817e-07,
      "loss": 0.3172,
      "step": 27614
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0794553756713867,
      "learning_rate": 1.205449556318461e-07,
      "loss": 0.2586,
      "step": 27615
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0265514850616455,
      "learning_rate": 1.203721692507498e-07,
      "loss": 0.2757,
      "step": 27616
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.5543551445007324,
      "learning_rate": 1.2019950604268084e-07,
      "loss": 0.3652,
      "step": 27617
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9589641094207764,
      "learning_rate": 1.2002696600979303e-07,
      "loss": 0.2658,
      "step": 27618
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8531486988067627,
      "learning_rate": 1.19854549154238e-07,
      "loss": 0.2528,
      "step": 27619
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7741693258285522,
      "learning_rate": 1.1968225547816513e-07,
      "loss": 0.2503,
      "step": 27620
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.053004503250122,
      "learning_rate": 1.195100849837205e-07,
      "loss": 0.254,
      "step": 27621
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.037763833999634,
      "learning_rate": 1.1933803767305353e-07,
      "loss": 0.3476,
      "step": 27622
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.3770744800567627,
      "learning_rate": 1.1916611354830687e-07,
      "loss": 0.2622,
      "step": 27623
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0848731994628906,
      "learning_rate": 1.1899431261162442e-07,
      "loss": 0.3113,
      "step": 27624
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9472556114196777,
      "learning_rate": 1.1882263486515e-07,
      "loss": 0.3192,
      "step": 27625
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8297463655471802,
      "learning_rate": 1.186510803110219e-07,
      "loss": 0.3144,
      "step": 27626
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6656250953674316,
      "learning_rate": 1.1847964895137954e-07,
      "loss": 0.2869,
      "step": 27627
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1690168380737305,
      "learning_rate": 1.1830834078836007e-07,
      "loss": 0.2949,
      "step": 27628
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.3346810340881348,
      "learning_rate": 1.1813715582410068e-07,
      "loss": 0.2856,
      "step": 27629
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8140778541564941,
      "learning_rate": 1.17966094060733e-07,
      "loss": 0.3116,
      "step": 27630
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.093320608139038,
      "learning_rate": 1.17795155500392e-07,
      "loss": 0.2464,
      "step": 27631
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0309665203094482,
      "learning_rate": 1.1762434014520707e-07,
      "loss": 0.3164,
      "step": 27632
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7549278736114502,
      "learning_rate": 1.1745364799730984e-07,
      "loss": 0.3272,
      "step": 27633
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.811963677406311,
      "learning_rate": 1.1728307905882641e-07,
      "loss": 0.2775,
      "step": 27634
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.020932197570801,
      "learning_rate": 1.1711263333188394e-07,
      "loss": 0.3148,
      "step": 27635
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.2503387928009033,
      "learning_rate": 1.1694231081860852e-07,
      "loss": 0.3208,
      "step": 27636
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.403960704803467,
      "learning_rate": 1.1677211152112289e-07,
      "loss": 0.3313,
      "step": 27637
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7548246383666992,
      "learning_rate": 1.1660203544154758e-07,
      "loss": 0.3013,
      "step": 27638
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9760299921035767,
      "learning_rate": 1.1643208258200534e-07,
      "loss": 0.3375,
      "step": 27639
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9434937238693237,
      "learning_rate": 1.1626225294461335e-07,
      "loss": 0.365,
      "step": 27640
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.2759206295013428,
      "learning_rate": 1.1609254653149105e-07,
      "loss": 0.3035,
      "step": 27641
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8633087873458862,
      "learning_rate": 1.1592296334475117e-07,
      "loss": 0.3011,
      "step": 27642
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.964659333229065,
      "learning_rate": 1.1575350338650982e-07,
      "loss": 0.351,
      "step": 27643
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1940953731536865,
      "learning_rate": 1.1558416665887972e-07,
      "loss": 0.2604,
      "step": 27644
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.846994161605835,
      "learning_rate": 1.1541495316397255e-07,
      "loss": 0.3702,
      "step": 27645
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7108111381530762,
      "learning_rate": 1.1524586290389661e-07,
      "loss": 0.242,
      "step": 27646
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.693974733352661,
      "learning_rate": 1.150768958807602e-07,
      "loss": 0.2689,
      "step": 27647
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9392056465148926,
      "learning_rate": 1.1490805209667055e-07,
      "loss": 0.2633,
      "step": 27648
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.4054834842681885,
      "learning_rate": 1.1473933155373262e-07,
      "loss": 0.223,
      "step": 27649
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.21535587310791,
      "learning_rate": 1.1457073425404919e-07,
      "loss": 0.3318,
      "step": 27650
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8764574527740479,
      "learning_rate": 1.1440226019972301e-07,
      "loss": 0.3015,
      "step": 27651
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.205613613128662,
      "learning_rate": 1.1423390939285461e-07,
      "loss": 0.3294,
      "step": 27652
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.446627140045166,
      "learning_rate": 1.1406568183554234e-07,
      "loss": 0.3499,
      "step": 27653
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1575942039489746,
      "learning_rate": 1.138975775298834e-07,
      "loss": 0.2999,
      "step": 27654
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8111035823822021,
      "learning_rate": 1.1372959647797499e-07,
      "loss": 0.3283,
      "step": 27655
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.2844839096069336,
      "learning_rate": 1.1356173868190878e-07,
      "loss": 0.3702,
      "step": 27656
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.864343285560608,
      "learning_rate": 1.1339400414378088e-07,
      "loss": 0.2912,
      "step": 27657
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.6652519702911377,
      "learning_rate": 1.132263928656796e-07,
      "loss": 0.3763,
      "step": 27658
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.119248628616333,
      "learning_rate": 1.1305890484969662e-07,
      "loss": 0.3303,
      "step": 27659
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7518444061279297,
      "learning_rate": 1.1289154009791803e-07,
      "loss": 0.3172,
      "step": 27660
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9677740335464478,
      "learning_rate": 1.1272429861243328e-07,
      "loss": 0.2464,
      "step": 27661
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.2685422897338867,
      "learning_rate": 1.1255718039532404e-07,
      "loss": 0.323,
      "step": 27662
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.262446641921997,
      "learning_rate": 1.1239018544867641e-07,
      "loss": 0.305,
      "step": 27663
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0591509342193604,
      "learning_rate": 1.1222331377457097e-07,
      "loss": 0.3443,
      "step": 27664
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.3908705711364746,
      "learning_rate": 1.1205656537508936e-07,
      "loss": 0.279,
      "step": 27665
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.147536277770996,
      "learning_rate": 1.1188994025230992e-07,
      "loss": 0.3416,
      "step": 27666
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.294888734817505,
      "learning_rate": 1.1172343840830879e-07,
      "loss": 0.326,
      "step": 27667
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.006490707397461,
      "learning_rate": 1.115570598451643e-07,
      "loss": 0.2466,
      "step": 27668
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0752861499786377,
      "learning_rate": 1.1139080456494922e-07,
      "loss": 0.2794,
      "step": 27669
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8157107830047607,
      "learning_rate": 1.1122467256973635e-07,
      "loss": 0.2524,
      "step": 27670
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1579253673553467,
      "learning_rate": 1.1105866386159736e-07,
      "loss": 0.3051,
      "step": 27671
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8387279510498047,
      "learning_rate": 1.108927784426006e-07,
      "loss": 0.2411,
      "step": 27672
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1621763706207275,
      "learning_rate": 1.1072701631481553e-07,
      "loss": 0.3575,
      "step": 27673
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.3021700382232666,
      "learning_rate": 1.1056137748030937e-07,
      "loss": 0.2581,
      "step": 27674
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8041962385177612,
      "learning_rate": 1.1039586194114715e-07,
      "loss": 0.2968,
      "step": 27675
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1743416786193848,
      "learning_rate": 1.1023046969938944e-07,
      "loss": 0.3421,
      "step": 27676
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8706032037734985,
      "learning_rate": 1.1006520075710236e-07,
      "loss": 0.2429,
      "step": 27677
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.4068148136138916,
      "learning_rate": 1.0990005511634316e-07,
      "loss": 0.3654,
      "step": 27678
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8483601808547974,
      "learning_rate": 1.097350327791724e-07,
      "loss": 0.3669,
      "step": 27679
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.928755760192871,
      "learning_rate": 1.0957013374764736e-07,
      "loss": 0.273,
      "step": 27680
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7473026514053345,
      "learning_rate": 1.0940535802382413e-07,
      "loss": 0.2369,
      "step": 27681
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.063472032546997,
      "learning_rate": 1.0924070560975552e-07,
      "loss": 0.2945,
      "step": 27682
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.8723416328430176,
      "learning_rate": 1.0907617650749658e-07,
      "loss": 0.4397,
      "step": 27683
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.788228988647461,
      "learning_rate": 1.0891177071909675e-07,
      "loss": 0.2764,
      "step": 27684
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8492664098739624,
      "learning_rate": 1.0874748824660775e-07,
      "loss": 0.3536,
      "step": 27685
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7709356546401978,
      "learning_rate": 1.0858332909207459e-07,
      "loss": 0.2726,
      "step": 27686
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9124999046325684,
      "learning_rate": 1.0841929325754564e-07,
      "loss": 0.2803,
      "step": 27687
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.166031837463379,
      "learning_rate": 1.0825538074506703e-07,
      "loss": 0.3415,
      "step": 27688
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8380389213562012,
      "learning_rate": 1.0809159155668048e-07,
      "loss": 0.3279,
      "step": 27689
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.073096752166748,
      "learning_rate": 1.079279256944299e-07,
      "loss": 0.3047,
      "step": 27690
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9414684772491455,
      "learning_rate": 1.0776438316035364e-07,
      "loss": 0.3139,
      "step": 27691
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7612531185150146,
      "learning_rate": 1.0760096395649234e-07,
      "loss": 0.2936,
      "step": 27692
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.027521848678589,
      "learning_rate": 1.074376680848821e-07,
      "loss": 0.252,
      "step": 27693
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.1748743057250977,
      "learning_rate": 1.072744955475602e-07,
      "loss": 0.3186,
      "step": 27694
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.286337375640869,
      "learning_rate": 1.0711144634655946e-07,
      "loss": 0.2958,
      "step": 27695
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.2484724521636963,
      "learning_rate": 1.069485204839149e-07,
      "loss": 0.336,
      "step": 27696
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9355019330978394,
      "learning_rate": 1.0678571796165605e-07,
      "loss": 0.376,
      "step": 27697
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.917015790939331,
      "learning_rate": 1.0662303878181234e-07,
      "loss": 0.3159,
      "step": 27698
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8096683025360107,
      "learning_rate": 1.064604829464122e-07,
      "loss": 0.2787,
      "step": 27699
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.0129923820495605,
      "learning_rate": 1.0629805045748286e-07,
      "loss": 0.2895,
      "step": 27700
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.007385492324829,
      "learning_rate": 1.061357413170494e-07,
      "loss": 0.3123,
      "step": 27701
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7829302549362183,
      "learning_rate": 1.0597355552713462e-07,
      "loss": 0.2681,
      "step": 27702
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.073410749435425,
      "learning_rate": 1.0581149308976246e-07,
      "loss": 0.304,
      "step": 27703
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0302298069000244,
      "learning_rate": 1.0564955400695131e-07,
      "loss": 0.2353,
      "step": 27704
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.268979549407959,
      "learning_rate": 1.0548773828072067e-07,
      "loss": 0.3805,
      "step": 27705
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.022296905517578,
      "learning_rate": 1.0532604591308781e-07,
      "loss": 0.3316,
      "step": 27706
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.158907175064087,
      "learning_rate": 1.0516447690607001e-07,
      "loss": 0.3202,
      "step": 27707
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0416367053985596,
      "learning_rate": 1.050030312616801e-07,
      "loss": 0.2843,
      "step": 27708
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.565253973007202,
      "learning_rate": 1.0484170898193092e-07,
      "loss": 0.2927,
      "step": 27709
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.010280132293701,
      "learning_rate": 1.0468051006883529e-07,
      "loss": 0.3136,
      "step": 27710
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8400667905807495,
      "learning_rate": 1.045194345244005e-07,
      "loss": 0.2953,
      "step": 27711
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4404433965682983,
      "learning_rate": 1.0435848235063605e-07,
      "loss": 0.1745,
      "step": 27712
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9359757900238037,
      "learning_rate": 1.0419765354955036e-07,
      "loss": 0.3183,
      "step": 27713
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7507274150848389,
      "learning_rate": 1.0403694812314513e-07,
      "loss": 0.3035,
      "step": 27714
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.678085446357727,
      "learning_rate": 1.0387636607342544e-07,
      "loss": 0.2324,
      "step": 27715
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7916276454925537,
      "learning_rate": 1.0371590740239413e-07,
      "loss": 0.2532,
      "step": 27716
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.046612501144409,
      "learning_rate": 1.035555721120518e-07,
      "loss": 0.2101,
      "step": 27717
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9217194318771362,
      "learning_rate": 1.0339536020439578e-07,
      "loss": 0.2375,
      "step": 27718
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.975276231765747,
      "learning_rate": 1.0323527168142444e-07,
      "loss": 0.3373,
      "step": 27719
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.760764241218567,
      "learning_rate": 1.0307530654513287e-07,
      "loss": 0.2832,
      "step": 27720
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.217808961868286,
      "learning_rate": 1.0291546479751724e-07,
      "loss": 0.264,
      "step": 27721
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.6220953464508057,
      "learning_rate": 1.0275574644056819e-07,
      "loss": 0.3711,
      "step": 27722
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7204797267913818,
      "learning_rate": 1.0259615147627966e-07,
      "loss": 0.177,
      "step": 27723
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0750584602355957,
      "learning_rate": 1.0243667990663897e-07,
      "loss": 0.3811,
      "step": 27724
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.6841254234313965,
      "learning_rate": 1.0227733173363453e-07,
      "loss": 0.2612,
      "step": 27725
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0564587116241455,
      "learning_rate": 1.0211810695925362e-07,
      "loss": 0.3664,
      "step": 27726
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9524043798446655,
      "learning_rate": 1.0195900558548245e-07,
      "loss": 0.2928,
      "step": 27727
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0251502990722656,
      "learning_rate": 1.0180002761430164e-07,
      "loss": 0.3515,
      "step": 27728
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6912037134170532,
      "learning_rate": 1.0164117304769628e-07,
      "loss": 0.2733,
      "step": 27729
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1897695064544678,
      "learning_rate": 1.014824418876459e-07,
      "loss": 0.3582,
      "step": 27730
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.801592230796814,
      "learning_rate": 1.0132383413612778e-07,
      "loss": 0.2491,
      "step": 27731
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8419373035430908,
      "learning_rate": 1.0116534979512149e-07,
      "loss": 0.2891,
      "step": 27732
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9278613328933716,
      "learning_rate": 1.010069888666032e-07,
      "loss": 0.2229,
      "step": 27733
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.4365272521972656,
      "learning_rate": 1.0084875135254468e-07,
      "loss": 0.3472,
      "step": 27734
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.2356700897216797,
      "learning_rate": 1.00690637254921e-07,
      "loss": 0.2419,
      "step": 27735
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.3320915699005127,
      "learning_rate": 1.0053264657570283e-07,
      "loss": 0.3526,
      "step": 27736
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.799515962600708,
      "learning_rate": 1.0037477931685969e-07,
      "loss": 0.2334,
      "step": 27737
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8504360914230347,
      "learning_rate": 1.0021703548035999e-07,
      "loss": 0.3171,
      "step": 27738
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0059056282043457,
      "learning_rate": 1.0005941506816996e-07,
      "loss": 0.2529,
      "step": 27739
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8258999586105347,
      "learning_rate": 9.99019180822558e-08,
      "loss": 0.2654,
      "step": 27740
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.028362274169922,
      "learning_rate": 9.974454452457927e-08,
      "loss": 0.2761,
      "step": 27741
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8332304954528809,
      "learning_rate": 9.958729439710324e-08,
      "loss": 0.262,
      "step": 27742
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.38095760345459,
      "learning_rate": 9.943016770178948e-08,
      "loss": 0.2972,
      "step": 27743
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9919734001159668,
      "learning_rate": 9.927316444059531e-08,
      "loss": 0.3004,
      "step": 27744
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9693368673324585,
      "learning_rate": 9.911628461547806e-08,
      "loss": 0.2756,
      "step": 27745
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7443387508392334,
      "learning_rate": 9.895952822839505e-08,
      "loss": 0.2992,
      "step": 27746
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.3443405628204346,
      "learning_rate": 9.880289528130027e-08,
      "loss": 0.2183,
      "step": 27747
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7366920709609985,
      "learning_rate": 9.864638577614549e-08,
      "loss": 0.2717,
      "step": 27748
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.077159881591797,
      "learning_rate": 9.848999971488249e-08,
      "loss": 0.2862,
      "step": 27749
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.956039547920227,
      "learning_rate": 9.833373709946082e-08,
      "loss": 0.2755,
      "step": 27750
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0479795932769775,
      "learning_rate": 9.81775979318289e-08,
      "loss": 0.2503,
      "step": 27751
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.3232576847076416,
      "learning_rate": 9.80215822139341e-08,
      "loss": 0.3492,
      "step": 27752
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.3413891792297363,
      "learning_rate": 9.786568994772038e-08,
      "loss": 0.3474,
      "step": 27753
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.2368388175964355,
      "learning_rate": 9.770992113513178e-08,
      "loss": 0.3157,
      "step": 27754
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9063280820846558,
      "learning_rate": 9.755427577811005e-08,
      "loss": 0.3562,
      "step": 27755
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.209869861602783,
      "learning_rate": 9.739875387859698e-08,
      "loss": 0.3108,
      "step": 27756
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1709697246551514,
      "learning_rate": 9.72433554385288e-08,
      "loss": 0.3577,
      "step": 27757
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6552654504776,
      "learning_rate": 9.708808045984508e-08,
      "loss": 0.2354,
      "step": 27758
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9698214530944824,
      "learning_rate": 9.693292894448203e-08,
      "loss": 0.3178,
      "step": 27759
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.4236226081848145,
      "learning_rate": 9.677790089437256e-08,
      "loss": 0.3487,
      "step": 27760
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.933260202407837,
      "learning_rate": 9.662299631144956e-08,
      "loss": 0.2838,
      "step": 27761
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.240339994430542,
      "learning_rate": 9.646821519764593e-08,
      "loss": 0.3177,
      "step": 27762
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5401570796966553,
      "learning_rate": 9.631355755489014e-08,
      "loss": 0.2403,
      "step": 27763
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.410228729248047,
      "learning_rate": 9.615902338510841e-08,
      "loss": 0.3146,
      "step": 27764
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.3099143505096436,
      "learning_rate": 9.600461269023143e-08,
      "loss": 0.2857,
      "step": 27765
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.00295352935791,
      "learning_rate": 9.585032547218209e-08,
      "loss": 0.316,
      "step": 27766
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.311835289001465,
      "learning_rate": 9.569616173288332e-08,
      "loss": 0.4123,
      "step": 27767
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.0097663402557373,
      "learning_rate": 9.554212147425801e-08,
      "loss": 0.2819,
      "step": 27768
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.90593683719635,
      "learning_rate": 9.538820469822797e-08,
      "loss": 0.2769,
      "step": 27769
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.4301764965057373,
      "learning_rate": 9.523441140670831e-08,
      "loss": 0.3638,
      "step": 27770
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.093188762664795,
      "learning_rate": 9.508074160162084e-08,
      "loss": 0.3203,
      "step": 27771
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.844196081161499,
      "learning_rate": 9.492719528487848e-08,
      "loss": 0.394,
      "step": 27772
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.137866973876953,
      "learning_rate": 9.477377245839747e-08,
      "loss": 0.348,
      "step": 27773
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8410896062850952,
      "learning_rate": 9.46204731240885e-08,
      "loss": 0.2441,
      "step": 27774
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1818506717681885,
      "learning_rate": 9.44672972838645e-08,
      "loss": 0.2667,
      "step": 27775
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.4223592281341553,
      "learning_rate": 9.431424493963504e-08,
      "loss": 0.3665,
      "step": 27776
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.117842197418213,
      "learning_rate": 9.416131609330748e-08,
      "loss": 0.2478,
      "step": 27777
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.133500337600708,
      "learning_rate": 9.400851074678918e-08,
      "loss": 0.2914,
      "step": 27778
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4511222839355469,
      "learning_rate": 9.38558289019842e-08,
      "loss": 0.2398,
      "step": 27779
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8582732677459717,
      "learning_rate": 9.370327056079653e-08,
      "loss": 0.3489,
      "step": 27780
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.2380542755126953,
      "learning_rate": 9.355083572512913e-08,
      "loss": 0.3142,
      "step": 27781
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6775200366973877,
      "learning_rate": 9.339852439688157e-08,
      "loss": 0.3063,
      "step": 27782
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.283867359161377,
      "learning_rate": 9.324633657795234e-08,
      "loss": 0.3428,
      "step": 27783
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7860928773880005,
      "learning_rate": 9.309427227023993e-08,
      "loss": 0.2376,
      "step": 27784
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.3151793479919434,
      "learning_rate": 9.294233147563948e-08,
      "loss": 0.3788,
      "step": 27785
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.10103440284729,
      "learning_rate": 9.2790514196045e-08,
      "loss": 0.2679,
      "step": 27786
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.508033037185669,
      "learning_rate": 9.26388204333506e-08,
      "loss": 0.3889,
      "step": 27787
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.85341477394104,
      "learning_rate": 9.248725018944693e-08,
      "loss": 0.3475,
      "step": 27788
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.2330496311187744,
      "learning_rate": 9.23358034662214e-08,
      "loss": 0.2718,
      "step": 27789
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9327656030654907,
      "learning_rate": 9.218448026556581e-08,
      "loss": 0.2531,
      "step": 27790
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7715119123458862,
      "learning_rate": 9.203328058936534e-08,
      "loss": 0.2144,
      "step": 27791
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.253389835357666,
      "learning_rate": 9.188220443950291e-08,
      "loss": 0.2375,
      "step": 27792
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.017428159713745,
      "learning_rate": 9.173125181786591e-08,
      "loss": 0.2981,
      "step": 27793
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.1375346183776855,
      "learning_rate": 9.158042272633283e-08,
      "loss": 0.3025,
      "step": 27794
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.015362501144409,
      "learning_rate": 9.142971716678661e-08,
      "loss": 0.2839,
      "step": 27795
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9222888946533203,
      "learning_rate": 9.127913514110464e-08,
      "loss": 0.268,
      "step": 27796
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7774982452392578,
      "learning_rate": 9.112867665116543e-08,
      "loss": 0.3174,
      "step": 27797
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9589877128601074,
      "learning_rate": 9.097834169884412e-08,
      "loss": 0.3108,
      "step": 27798
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.082789659500122,
      "learning_rate": 9.082813028601368e-08,
      "loss": 0.3159,
      "step": 27799
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.305931568145752,
      "learning_rate": 9.067804241454925e-08,
      "loss": 0.3674,
      "step": 27800
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.738481879234314,
      "learning_rate": 9.052807808632158e-08,
      "loss": 0.2939,
      "step": 27801
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.283308982849121,
      "learning_rate": 9.037823730319805e-08,
      "loss": 0.3183,
      "step": 27802
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.6348636150360107,
      "learning_rate": 9.022852006704941e-08,
      "loss": 0.3589,
      "step": 27803
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.774327039718628,
      "learning_rate": 9.007892637974192e-08,
      "loss": 0.3319,
      "step": 27804
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8916723728179932,
      "learning_rate": 8.992945624313854e-08,
      "loss": 0.3299,
      "step": 27805
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.348928451538086,
      "learning_rate": 8.978010965910556e-08,
      "loss": 0.3299,
      "step": 27806
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9622721672058105,
      "learning_rate": 8.96308866295037e-08,
      "loss": 0.3677,
      "step": 27807
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8663448095321655,
      "learning_rate": 8.94817871561926e-08,
      "loss": 0.2862,
      "step": 27808
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1259260177612305,
      "learning_rate": 8.933281124103188e-08,
      "loss": 0.3087,
      "step": 27809
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.013944387435913,
      "learning_rate": 8.918395888587894e-08,
      "loss": 0.2883,
      "step": 27810
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9036799669265747,
      "learning_rate": 8.903523009259007e-08,
      "loss": 0.2999,
      "step": 27811
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.888886570930481,
      "learning_rate": 8.888662486301824e-08,
      "loss": 0.302,
      "step": 27812
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.051349401473999,
      "learning_rate": 8.873814319901752e-08,
      "loss": 0.3017,
      "step": 27813
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.006678819656372,
      "learning_rate": 8.858978510243755e-08,
      "loss": 0.3978,
      "step": 27814
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.3912160396575928,
      "learning_rate": 8.844155057512905e-08,
      "loss": 0.3697,
      "step": 27815
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6819876432418823,
      "learning_rate": 8.829343961894054e-08,
      "loss": 0.3267,
      "step": 27816
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.137057304382324,
      "learning_rate": 8.814545223571725e-08,
      "loss": 0.3199,
      "step": 27817
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.234999418258667,
      "learning_rate": 8.799758842730433e-08,
      "loss": 0.3343,
      "step": 27818
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.2758407592773438,
      "learning_rate": 8.784984819554698e-08,
      "loss": 0.3349,
      "step": 27819
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.361140012741089,
      "learning_rate": 8.770223154228597e-08,
      "loss": 0.3295,
      "step": 27820
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0372087955474854,
      "learning_rate": 8.755473846936091e-08,
      "loss": 0.2988,
      "step": 27821
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0011606216430664,
      "learning_rate": 8.740736897861146e-08,
      "loss": 0.3247,
      "step": 27822
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.30902099609375,
      "learning_rate": 8.726012307187614e-08,
      "loss": 0.3796,
      "step": 27823
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9636870622634888,
      "learning_rate": 8.711300075098905e-08,
      "loss": 0.3297,
      "step": 27824
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.871232509613037,
      "learning_rate": 8.696600201778316e-08,
      "loss": 0.3092,
      "step": 27825
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.051149845123291,
      "learning_rate": 8.681912687409477e-08,
      "loss": 0.2965,
      "step": 27826
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.310971260070801,
      "learning_rate": 8.667237532175243e-08,
      "loss": 0.2656,
      "step": 27827
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9454859495162964,
      "learning_rate": 8.652574736258579e-08,
      "loss": 0.29,
      "step": 27828
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.752912998199463,
      "learning_rate": 8.637924299842448e-08,
      "loss": 0.3389,
      "step": 27829
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7763615846633911,
      "learning_rate": 8.623286223109373e-08,
      "loss": 0.322,
      "step": 27830
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9754678010940552,
      "learning_rate": 8.608660506241872e-08,
      "loss": 0.3217,
      "step": 27831
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1575753688812256,
      "learning_rate": 8.594047149422357e-08,
      "loss": 0.2646,
      "step": 27832
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.176677942276001,
      "learning_rate": 8.579446152832904e-08,
      "loss": 0.3038,
      "step": 27833
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.2365975379943848,
      "learning_rate": 8.56485751665559e-08,
      "loss": 0.3527,
      "step": 27834
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7796530723571777,
      "learning_rate": 8.550281241072378e-08,
      "loss": 0.2983,
      "step": 27835
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.647180438041687,
      "learning_rate": 8.535717326265014e-08,
      "loss": 0.2686,
      "step": 27836
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.20863938331604,
      "learning_rate": 8.521165772414908e-08,
      "loss": 0.2828,
      "step": 27837
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.450892686843872,
      "learning_rate": 8.50662657970358e-08,
      "loss": 0.2588,
      "step": 27838
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1481947898864746,
      "learning_rate": 8.492099748312221e-08,
      "loss": 0.2604,
      "step": 27839
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.028235673904419,
      "learning_rate": 8.47758527842213e-08,
      "loss": 0.3596,
      "step": 27840
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5518549680709839,
      "learning_rate": 8.463083170213937e-08,
      "loss": 0.2795,
      "step": 27841
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0466015338897705,
      "learning_rate": 8.448593423868834e-08,
      "loss": 0.3156,
      "step": 27842
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9986804723739624,
      "learning_rate": 8.43411603956723e-08,
      "loss": 0.309,
      "step": 27843
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.6841659545898438,
      "learning_rate": 8.419651017489649e-08,
      "loss": 0.3028,
      "step": 27844
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7237457036972046,
      "learning_rate": 8.405198357816502e-08,
      "loss": 0.2115,
      "step": 27845
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8252127170562744,
      "learning_rate": 8.390758060727865e-08,
      "loss": 0.3151,
      "step": 27846
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.831377625465393,
      "learning_rate": 8.37633012640382e-08,
      "loss": 0.2177,
      "step": 27847
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8598520755767822,
      "learning_rate": 8.361914555024331e-08,
      "loss": 0.304,
      "step": 27848
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9554532766342163,
      "learning_rate": 8.347511346769032e-08,
      "loss": 0.337,
      "step": 27849
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7018948793411255,
      "learning_rate": 8.33312050181756e-08,
      "loss": 0.2533,
      "step": 27850
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.426095485687256,
      "learning_rate": 8.318742020349103e-08,
      "loss": 0.3252,
      "step": 27851
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.275968313217163,
      "learning_rate": 8.304375902543293e-08,
      "loss": 0.3794,
      "step": 27852
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1306402683258057,
      "learning_rate": 8.290022148578991e-08,
      "loss": 0.3178,
      "step": 27853
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0479393005371094,
      "learning_rate": 8.275680758635163e-08,
      "loss": 0.2822,
      "step": 27854
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1976540088653564,
      "learning_rate": 8.261351732890666e-08,
      "loss": 0.2623,
      "step": 27855
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7973661422729492,
      "learning_rate": 8.247035071524245e-08,
      "loss": 0.2973,
      "step": 27856
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9704352617263794,
      "learning_rate": 8.232730774714093e-08,
      "loss": 0.2996,
      "step": 27857
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.170276165008545,
      "learning_rate": 8.218438842638843e-08,
      "loss": 0.3332,
      "step": 27858
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9096943140029907,
      "learning_rate": 8.204159275476576e-08,
      "loss": 0.3129,
      "step": 27859
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.3621907234191895,
      "learning_rate": 8.18989207340526e-08,
      "loss": 0.2776,
      "step": 27860
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.152627944946289,
      "learning_rate": 8.175637236602862e-08,
      "loss": 0.3063,
      "step": 27861
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.2223117351531982,
      "learning_rate": 8.161394765247133e-08,
      "loss": 0.2728,
      "step": 27862
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6486667394638062,
      "learning_rate": 8.147164659515483e-08,
      "loss": 0.2908,
      "step": 27863
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.026505708694458,
      "learning_rate": 8.132946919585327e-08,
      "loss": 0.3039,
      "step": 27864
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.4051713943481445,
      "learning_rate": 8.11874154563408e-08,
      "loss": 0.3545,
      "step": 27865
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1691479682922363,
      "learning_rate": 8.10454853783882e-08,
      "loss": 0.2712,
      "step": 27866
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9290653467178345,
      "learning_rate": 8.090367896376406e-08,
      "loss": 0.2656,
      "step": 27867
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0902843475341797,
      "learning_rate": 8.076199621423698e-08,
      "loss": 0.3204,
      "step": 27868
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8274067640304565,
      "learning_rate": 8.062043713157331e-08,
      "loss": 0.3669,
      "step": 27869
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8119986057281494,
      "learning_rate": 8.047900171753719e-08,
      "loss": 0.2871,
      "step": 27870
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0048439502716064,
      "learning_rate": 8.033768997389279e-08,
      "loss": 0.3193,
      "step": 27871
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1378583908081055,
      "learning_rate": 8.019650190240091e-08,
      "loss": 0.321,
      "step": 27872
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0141210556030273,
      "learning_rate": 8.005543750482348e-08,
      "loss": 0.2851,
      "step": 27873
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.3279354572296143,
      "learning_rate": 7.991449678291796e-08,
      "loss": 0.3079,
      "step": 27874
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.1646862030029297,
      "learning_rate": 7.977367973844185e-08,
      "loss": 0.2702,
      "step": 27875
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.130121946334839,
      "learning_rate": 7.963298637315042e-08,
      "loss": 0.3628,
      "step": 27876
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.322122812271118,
      "learning_rate": 7.949241668879782e-08,
      "loss": 0.3224,
      "step": 27877
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.223248243331909,
      "learning_rate": 7.935197068713707e-08,
      "loss": 0.3757,
      "step": 27878
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.6136467456817627,
      "learning_rate": 7.921164836991901e-08,
      "loss": 0.2886,
      "step": 27879
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0979740619659424,
      "learning_rate": 7.907144973889336e-08,
      "loss": 0.3729,
      "step": 27880
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0474448204040527,
      "learning_rate": 7.89313747958076e-08,
      "loss": 0.351,
      "step": 27881
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9947459697723389,
      "learning_rate": 7.879142354240699e-08,
      "loss": 0.3801,
      "step": 27882
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.3285882472991943,
      "learning_rate": 7.865159598043792e-08,
      "loss": 0.2464,
      "step": 27883
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7170106172561646,
      "learning_rate": 7.851189211164346e-08,
      "loss": 0.3069,
      "step": 27884
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7425609827041626,
      "learning_rate": 7.837231193776551e-08,
      "loss": 0.2869,
      "step": 27885
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.576793670654297,
      "learning_rate": 7.823285546054382e-08,
      "loss": 0.3355,
      "step": 27886
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.125014305114746,
      "learning_rate": 7.8093522681717e-08,
      "loss": 0.3386,
      "step": 27887
    },
    {
      "epoch": 2.88,
      "grad_norm": 3.036172866821289,
      "learning_rate": 7.795431360302252e-08,
      "loss": 0.279,
      "step": 27888
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.850268006324768,
      "learning_rate": 7.781522822619458e-08,
      "loss": 0.2462,
      "step": 27889
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.9247689247131348,
      "learning_rate": 7.767626655296956e-08,
      "loss": 0.2945,
      "step": 27890
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.875920057296753,
      "learning_rate": 7.753742858507829e-08,
      "loss": 0.2752,
      "step": 27891
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8131641149520874,
      "learning_rate": 7.739871432425272e-08,
      "loss": 0.307,
      "step": 27892
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.117004871368408,
      "learning_rate": 7.726012377222036e-08,
      "loss": 0.3161,
      "step": 27893
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.0586469173431396,
      "learning_rate": 7.712165693071094e-08,
      "loss": 0.3442,
      "step": 27894
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.763680338859558,
      "learning_rate": 7.698331380145085e-08,
      "loss": 0.211,
      "step": 27895
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8867907524108887,
      "learning_rate": 7.684509438616316e-08,
      "loss": 0.2872,
      "step": 27896
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0414042472839355,
      "learning_rate": 7.670699868657316e-08,
      "loss": 0.2951,
      "step": 27897
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7645080089569092,
      "learning_rate": 7.65690267044017e-08,
      "loss": 0.2755,
      "step": 27898
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.4246575832366943,
      "learning_rate": 7.643117844136849e-08,
      "loss": 0.3674,
      "step": 27899
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.2001497745513916,
      "learning_rate": 7.629345389919219e-08,
      "loss": 0.2685,
      "step": 27900
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0268945693969727,
      "learning_rate": 7.615585307959028e-08,
      "loss": 0.2703,
      "step": 27901
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.955381155014038,
      "learning_rate": 7.601837598427919e-08,
      "loss": 0.329,
      "step": 27902
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.1867239475250244,
      "learning_rate": 7.588102261496977e-08,
      "loss": 0.2955,
      "step": 27903
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0430314540863037,
      "learning_rate": 7.574379297337841e-08,
      "loss": 0.2654,
      "step": 27904
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.4927453994750977,
      "learning_rate": 7.560668706121377e-08,
      "loss": 0.3265,
      "step": 27905
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.4564926624298096,
      "learning_rate": 7.546970488018445e-08,
      "loss": 0.3251,
      "step": 27906
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.431169033050537,
      "learning_rate": 7.533284643200022e-08,
      "loss": 0.25,
      "step": 27907
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6290653944015503,
      "learning_rate": 7.519611171836638e-08,
      "loss": 0.3084,
      "step": 27908
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8699157238006592,
      "learning_rate": 7.505950074098711e-08,
      "loss": 0.1949,
      "step": 27909
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0187764167785645,
      "learning_rate": 7.492301350156661e-08,
      "loss": 0.3535,
      "step": 27910
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.226797103881836,
      "learning_rate": 7.478665000180685e-08,
      "loss": 0.3054,
      "step": 27911
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0789976119995117,
      "learning_rate": 7.465041024340647e-08,
      "loss": 0.3201,
      "step": 27912
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.1443729400634766,
      "learning_rate": 7.451429422806522e-08,
      "loss": 0.2836,
      "step": 27913
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8114312887191772,
      "learning_rate": 7.437830195747953e-08,
      "loss": 0.2516,
      "step": 27914
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.208315372467041,
      "learning_rate": 7.42424334333447e-08,
      "loss": 0.2931,
      "step": 27915
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7341258525848389,
      "learning_rate": 7.410668865735494e-08,
      "loss": 0.2938,
      "step": 27916
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8653552532196045,
      "learning_rate": 7.397106763120331e-08,
      "loss": 0.3149,
      "step": 27917
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.109001636505127,
      "learning_rate": 7.383557035657962e-08,
      "loss": 0.2973,
      "step": 27918
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8763840198516846,
      "learning_rate": 7.370019683517249e-08,
      "loss": 0.3011,
      "step": 27919
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.6978087425231934,
      "learning_rate": 7.356494706867168e-08,
      "loss": 0.3446,
      "step": 27920
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6300331354141235,
      "learning_rate": 7.34298210587625e-08,
      "loss": 0.3281,
      "step": 27921
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7873204946517944,
      "learning_rate": 7.329481880712918e-08,
      "loss": 0.2487,
      "step": 27922
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.748342514038086,
      "learning_rate": 7.315994031545592e-08,
      "loss": 0.3115,
      "step": 27923
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8911094665527344,
      "learning_rate": 7.302518558542248e-08,
      "loss": 0.3772,
      "step": 27924
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.753281831741333,
      "learning_rate": 7.289055461871086e-08,
      "loss": 0.2879,
      "step": 27925
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9525574445724487,
      "learning_rate": 7.27560474169986e-08,
      "loss": 0.299,
      "step": 27926
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.778092384338379,
      "learning_rate": 7.262166398196324e-08,
      "loss": 0.187,
      "step": 27927
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.030517816543579,
      "learning_rate": 7.2487404315279e-08,
      "loss": 0.3809,
      "step": 27928
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.166123390197754,
      "learning_rate": 7.235326841862011e-08,
      "loss": 0.2884,
      "step": 27929
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6955721378326416,
      "learning_rate": 7.221925629365966e-08,
      "loss": 0.2623,
      "step": 27930
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6909937858581543,
      "learning_rate": 7.208536794206856e-08,
      "loss": 0.2902,
      "step": 27931
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6971081495285034,
      "learning_rate": 7.195160336551433e-08,
      "loss": 0.239,
      "step": 27932
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.056013822555542,
      "learning_rate": 7.181796256566676e-08,
      "loss": 0.3005,
      "step": 27933
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.187483549118042,
      "learning_rate": 7.168444554419008e-08,
      "loss": 0.2721,
      "step": 27934
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.726155400276184,
      "learning_rate": 7.155105230275072e-08,
      "loss": 0.2716,
      "step": 27935
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9621853828430176,
      "learning_rate": 7.141778284301071e-08,
      "loss": 0.2987,
      "step": 27936
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9534589052200317,
      "learning_rate": 7.128463716663203e-08,
      "loss": 0.3267,
      "step": 27937
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.7096784114837646,
      "learning_rate": 7.115161527527337e-08,
      "loss": 0.3734,
      "step": 27938
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.051848888397217,
      "learning_rate": 7.101871717059561e-08,
      "loss": 0.3282,
      "step": 27939
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.390275716781616,
      "learning_rate": 7.088594285425298e-08,
      "loss": 0.3162,
      "step": 27940
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.391744613647461,
      "learning_rate": 7.075329232790306e-08,
      "loss": 0.2573,
      "step": 27941
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8664077520370483,
      "learning_rate": 7.062076559319786e-08,
      "loss": 0.2399,
      "step": 27942
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.720336675643921,
      "learning_rate": 7.048836265179048e-08,
      "loss": 0.3283,
      "step": 27943
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.3332324028015137,
      "learning_rate": 7.035608350533186e-08,
      "loss": 0.3464,
      "step": 27944
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9030095338821411,
      "learning_rate": 7.022392815547063e-08,
      "loss": 0.3507,
      "step": 27945
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.426476001739502,
      "learning_rate": 7.009189660385552e-08,
      "loss": 0.3277,
      "step": 27946
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.50952410697937,
      "learning_rate": 6.995998885213073e-08,
      "loss": 0.2857,
      "step": 27947
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7659274339675903,
      "learning_rate": 6.982820490194275e-08,
      "loss": 0.3511,
      "step": 27948
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.6583573818206787,
      "learning_rate": 6.969654475493359e-08,
      "loss": 0.3354,
      "step": 27949
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.27327036857605,
      "learning_rate": 6.956500841274416e-08,
      "loss": 0.299,
      "step": 27950
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7560182809829712,
      "learning_rate": 6.943359587701537e-08,
      "loss": 0.3061,
      "step": 27951
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.689931869506836,
      "learning_rate": 6.930230714938479e-08,
      "loss": 0.2685,
      "step": 27952
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.511561393737793,
      "learning_rate": 6.917114223149113e-08,
      "loss": 0.3295,
      "step": 27953
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9097577333450317,
      "learning_rate": 6.90401011249675e-08,
      "loss": 0.2768,
      "step": 27954
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.3017754554748535,
      "learning_rate": 6.890918383144707e-08,
      "loss": 0.3758,
      "step": 27955
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0694210529327393,
      "learning_rate": 6.877839035256518e-08,
      "loss": 0.2719,
      "step": 27956
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8379662036895752,
      "learning_rate": 6.86477206899494e-08,
      "loss": 0.3392,
      "step": 27957
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6771777868270874,
      "learning_rate": 6.851717484523069e-08,
      "loss": 0.2866,
      "step": 27958
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8564975261688232,
      "learning_rate": 6.838675282003548e-08,
      "loss": 0.2785,
      "step": 27959
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6154035329818726,
      "learning_rate": 6.82564546159914e-08,
      "loss": 0.3067,
      "step": 27960
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.221322536468506,
      "learning_rate": 6.812628023472046e-08,
      "loss": 0.3427,
      "step": 27961
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8981235027313232,
      "learning_rate": 6.799622967784802e-08,
      "loss": 0.3072,
      "step": 27962
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0765280723571777,
      "learning_rate": 6.786630294699392e-08,
      "loss": 0.3072,
      "step": 27963
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9337633848190308,
      "learning_rate": 6.773650004377908e-08,
      "loss": 0.2921,
      "step": 27964
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.062260389328003,
      "learning_rate": 6.760682096982108e-08,
      "loss": 0.2751,
      "step": 27965
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.272773027420044,
      "learning_rate": 6.747726572673641e-08,
      "loss": 0.2866,
      "step": 27966
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0100207328796387,
      "learning_rate": 6.734783431614044e-08,
      "loss": 0.3136,
      "step": 27967
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.191300392150879,
      "learning_rate": 6.721852673964746e-08,
      "loss": 0.3258,
      "step": 27968
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.299525499343872,
      "learning_rate": 6.708934299886949e-08,
      "loss": 0.3137,
      "step": 27969
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9632666110992432,
      "learning_rate": 6.696028309541746e-08,
      "loss": 0.3082,
      "step": 27970
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8737930059432983,
      "learning_rate": 6.683134703089788e-08,
      "loss": 0.2387,
      "step": 27971
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.05127215385437,
      "learning_rate": 6.670253480692168e-08,
      "loss": 0.2966,
      "step": 27972
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.5269079208374023,
      "learning_rate": 6.657384642509313e-08,
      "loss": 0.2341,
      "step": 27973
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0597872734069824,
      "learning_rate": 6.644528188701649e-08,
      "loss": 0.3741,
      "step": 27974
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5923091173171997,
      "learning_rate": 6.631684119429605e-08,
      "loss": 0.1939,
      "step": 27975
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8852747678756714,
      "learning_rate": 6.618852434853052e-08,
      "loss": 0.2921,
      "step": 27976
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7670646905899048,
      "learning_rate": 6.606033135132194e-08,
      "loss": 0.2377,
      "step": 27977
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0140116214752197,
      "learning_rate": 6.593226220426685e-08,
      "loss": 0.3364,
      "step": 27978
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8569178581237793,
      "learning_rate": 6.580431690896394e-08,
      "loss": 0.3015,
      "step": 27979
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9264835119247437,
      "learning_rate": 6.56764954670075e-08,
      "loss": 0.2827,
      "step": 27980
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.4566593170166016,
      "learning_rate": 6.554879787998957e-08,
      "loss": 0.2822,
      "step": 27981
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.043645143508911,
      "learning_rate": 6.542122414950447e-08,
      "loss": 0.3329,
      "step": 27982
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.064406633377075,
      "learning_rate": 6.529377427714201e-08,
      "loss": 0.2667,
      "step": 27983
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.918149471282959,
      "learning_rate": 6.516644826448981e-08,
      "loss": 0.2879,
      "step": 27984
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.0595993995666504,
      "learning_rate": 6.503924611313772e-08,
      "loss": 0.2661,
      "step": 27985
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.1739039421081543,
      "learning_rate": 6.491216782467002e-08,
      "loss": 0.3796,
      "step": 27986
    },
    {
      "epoch": 2.89,
      "grad_norm": 3.121635675430298,
      "learning_rate": 6.478521340067101e-08,
      "loss": 0.2984,
      "step": 27987
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.417933940887451,
      "learning_rate": 6.465838284272386e-08,
      "loss": 0.2989,
      "step": 27988
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7962100505828857,
      "learning_rate": 6.453167615240951e-08,
      "loss": 0.3191,
      "step": 27989
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.1765434741973877,
      "learning_rate": 6.440509333130784e-08,
      "loss": 0.2749,
      "step": 27990
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.1017658710479736,
      "learning_rate": 6.427863438099758e-08,
      "loss": 0.294,
      "step": 27991
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.064607858657837,
      "learning_rate": 6.415229930305411e-08,
      "loss": 0.3056,
      "step": 27992
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8697596788406372,
      "learning_rate": 6.402608809905397e-08,
      "loss": 0.3379,
      "step": 27993
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.063749074935913,
      "learning_rate": 6.390000077056924e-08,
      "loss": 0.3239,
      "step": 27994
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.183558702468872,
      "learning_rate": 6.377403731917309e-08,
      "loss": 0.369,
      "step": 27995
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.674130916595459,
      "learning_rate": 6.364819774643539e-08,
      "loss": 0.3813,
      "step": 27996
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.4477155208587646,
      "learning_rate": 6.352248205392486e-08,
      "loss": 0.2535,
      "step": 27997
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7710014581680298,
      "learning_rate": 6.339689024320916e-08,
      "loss": 0.3251,
      "step": 27998
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7634977102279663,
      "learning_rate": 6.327142231585371e-08,
      "loss": 0.2128,
      "step": 27999
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.909626007080078,
      "learning_rate": 6.31460782734239e-08,
      "loss": 0.3651,
      "step": 28000
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0286686420440674,
      "learning_rate": 6.302085811748071e-08,
      "loss": 0.3557,
      "step": 28001
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.3893301486968994,
      "learning_rate": 6.289576184958624e-08,
      "loss": 0.225,
      "step": 28002
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.5387935638427734,
      "learning_rate": 6.277078947130033e-08,
      "loss": 0.303,
      "step": 28003
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9317476749420166,
      "learning_rate": 6.264594098418064e-08,
      "loss": 0.2923,
      "step": 28004
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.649427056312561,
      "learning_rate": 6.252121638978481e-08,
      "loss": 0.3716,
      "step": 28005
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1526052951812744,
      "learning_rate": 6.239661568966604e-08,
      "loss": 0.2879,
      "step": 28006
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.841696858406067,
      "learning_rate": 6.227213888537865e-08,
      "loss": 0.3165,
      "step": 28007
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7967807054519653,
      "learning_rate": 6.214778597847471e-08,
      "loss": 0.2797,
      "step": 28008
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4939030408859253,
      "learning_rate": 6.202355697050411e-08,
      "loss": 0.3071,
      "step": 28009
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8802696466445923,
      "learning_rate": 6.18994518630156e-08,
      "loss": 0.2692,
      "step": 28010
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5954391956329346,
      "learning_rate": 6.177547065755685e-08,
      "loss": 0.3327,
      "step": 28011
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.066206693649292,
      "learning_rate": 6.165161335567438e-08,
      "loss": 0.2432,
      "step": 28012
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8512829542160034,
      "learning_rate": 6.15278799589103e-08,
      "loss": 0.2202,
      "step": 28013
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0566420555114746,
      "learning_rate": 6.140427046880893e-08,
      "loss": 0.2534,
      "step": 28014
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.993392825126648,
      "learning_rate": 6.128078488691125e-08,
      "loss": 0.3073,
      "step": 28015
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.82917058467865,
      "learning_rate": 6.115742321475604e-08,
      "loss": 0.2993,
      "step": 28016
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0341238975524902,
      "learning_rate": 6.103418545388096e-08,
      "loss": 0.3313,
      "step": 28017
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8969173431396484,
      "learning_rate": 6.091107160582365e-08,
      "loss": 0.3139,
      "step": 28018
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.630410075187683,
      "learning_rate": 6.078808167211958e-08,
      "loss": 0.3212,
      "step": 28019
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.24072003364563,
      "learning_rate": 6.066521565429973e-08,
      "loss": 0.2911,
      "step": 28020
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1132700443267822,
      "learning_rate": 6.054247355389731e-08,
      "loss": 0.3156,
      "step": 28021
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7815396785736084,
      "learning_rate": 6.041985537244332e-08,
      "loss": 0.3011,
      "step": 28022
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8377476930618286,
      "learning_rate": 6.029736111146433e-08,
      "loss": 0.2163,
      "step": 28023
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7037205696105957,
      "learning_rate": 6.017499077249022e-08,
      "loss": 0.2752,
      "step": 28024
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.137176990509033,
      "learning_rate": 6.005274435704534e-08,
      "loss": 0.255,
      "step": 28025
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.283236265182495,
      "learning_rate": 5.993062186665178e-08,
      "loss": 0.3421,
      "step": 28026
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0276458263397217,
      "learning_rate": 5.98086233028361e-08,
      "loss": 0.3691,
      "step": 28027
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9125157594680786,
      "learning_rate": 5.96867486671171e-08,
      "loss": 0.2819,
      "step": 28028
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1807007789611816,
      "learning_rate": 5.956499796101245e-08,
      "loss": 0.3248,
      "step": 28029
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6050281524658203,
      "learning_rate": 5.944337118604426e-08,
      "loss": 0.2072,
      "step": 28030
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.161947727203369,
      "learning_rate": 5.9321868343725776e-08,
      "loss": 0.2711,
      "step": 28031
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.565921664237976,
      "learning_rate": 5.920048943557244e-08,
      "loss": 0.2664,
      "step": 28032
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0753464698791504,
      "learning_rate": 5.907923446309749e-08,
      "loss": 0.3318,
      "step": 28033
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.505335569381714,
      "learning_rate": 5.895810342781305e-08,
      "loss": 0.2909,
      "step": 28034
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1997311115264893,
      "learning_rate": 5.883709633123014e-08,
      "loss": 0.291,
      "step": 28035
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8747459650039673,
      "learning_rate": 5.871621317485532e-08,
      "loss": 0.278,
      "step": 28036
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8818563222885132,
      "learning_rate": 5.859545396019739e-08,
      "loss": 0.2902,
      "step": 28037
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7330571413040161,
      "learning_rate": 5.8474818688760706e-08,
      "loss": 0.2726,
      "step": 28038
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7732008695602417,
      "learning_rate": 5.835430736204961e-08,
      "loss": 0.2767,
      "step": 28039
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0726540088653564,
      "learning_rate": 5.823391998156846e-08,
      "loss": 0.289,
      "step": 28040
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9745423793792725,
      "learning_rate": 5.811365654881496e-08,
      "loss": 0.3074,
      "step": 28041
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.027768135070801,
      "learning_rate": 5.7993517065290105e-08,
      "loss": 0.3241,
      "step": 28042
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.185438632965088,
      "learning_rate": 5.787350153249272e-08,
      "loss": 0.3055,
      "step": 28043
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7570972442626953,
      "learning_rate": 5.775360995191826e-08,
      "loss": 0.2968,
      "step": 28044
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2559890747070312,
      "learning_rate": 5.763384232505997e-08,
      "loss": 0.282,
      "step": 28045
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2339835166931152,
      "learning_rate": 5.751419865341223e-08,
      "loss": 0.2415,
      "step": 28046
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8871761560440063,
      "learning_rate": 5.739467893846828e-08,
      "loss": 0.3001,
      "step": 28047
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.093451499938965,
      "learning_rate": 5.727528318171583e-08,
      "loss": 0.2442,
      "step": 28048
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.3761043548583984,
      "learning_rate": 5.715601138464366e-08,
      "loss": 0.2889,
      "step": 28049
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.17828631401062,
      "learning_rate": 5.70368635487395e-08,
      "loss": 0.295,
      "step": 28050
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.414283514022827,
      "learning_rate": 5.6917839675489915e-08,
      "loss": 0.3136,
      "step": 28051
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2785897254943848,
      "learning_rate": 5.679893976637596e-08,
      "loss": 0.249,
      "step": 28052
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.022611379623413,
      "learning_rate": 5.66801638228831e-08,
      "loss": 0.2906,
      "step": 28053
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.3455188274383545,
      "learning_rate": 5.656151184649017e-08,
      "loss": 0.3838,
      "step": 28054
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0019588470458984,
      "learning_rate": 5.644298383867708e-08,
      "loss": 0.2804,
      "step": 28055
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0865306854248047,
      "learning_rate": 5.6324579800921544e-08,
      "loss": 0.2936,
      "step": 28056
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.279879570007324,
      "learning_rate": 5.620629973469904e-08,
      "loss": 0.3534,
      "step": 28057
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.400385856628418,
      "learning_rate": 5.6088143641486183e-08,
      "loss": 0.2979,
      "step": 28058
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2560534477233887,
      "learning_rate": 5.5970111522754e-08,
      "loss": 0.3196,
      "step": 28059
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.041560649871826,
      "learning_rate": 5.585220337997466e-08,
      "loss": 0.3687,
      "step": 28060
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.986958622932434,
      "learning_rate": 5.5734419214618084e-08,
      "loss": 0.3273,
      "step": 28061
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1932272911071777,
      "learning_rate": 5.561675902815311e-08,
      "loss": 0.2663,
      "step": 28062
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.266909122467041,
      "learning_rate": 5.549922282204634e-08,
      "loss": 0.307,
      "step": 28063
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2411231994628906,
      "learning_rate": 5.538181059776437e-08,
      "loss": 0.2788,
      "step": 28064
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1218581199645996,
      "learning_rate": 5.526452235676827e-08,
      "loss": 0.3657,
      "step": 28065
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.6801745891571045,
      "learning_rate": 5.514735810052352e-08,
      "loss": 0.3859,
      "step": 28066
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6618348360061646,
      "learning_rate": 5.5030317830488953e-08,
      "loss": 0.2693,
      "step": 28067
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0468459129333496,
      "learning_rate": 5.491340154812341e-08,
      "loss": 0.3226,
      "step": 28068
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.249656915664673,
      "learning_rate": 5.479660925488572e-08,
      "loss": 0.3773,
      "step": 28069
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.9531986713409424,
      "learning_rate": 5.467994095223139e-08,
      "loss": 0.2996,
      "step": 28070
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1507530212402344,
      "learning_rate": 5.456339664161592e-08,
      "loss": 0.3459,
      "step": 28071
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.961421251296997,
      "learning_rate": 5.4446976324490365e-08,
      "loss": 0.3186,
      "step": 28072
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1715922355651855,
      "learning_rate": 5.433068000230801e-08,
      "loss": 0.3294,
      "step": 28073
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.096578598022461,
      "learning_rate": 5.42145076765177e-08,
      "loss": 0.3075,
      "step": 28074
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.970274567604065,
      "learning_rate": 5.409845934856717e-08,
      "loss": 0.3292,
      "step": 28075
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.2024550437927246,
      "learning_rate": 5.398253501990525e-08,
      "loss": 0.3541,
      "step": 28076
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.756347894668579,
      "learning_rate": 5.386673469197634e-08,
      "loss": 0.2674,
      "step": 28077
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1442275047302246,
      "learning_rate": 5.375105836622374e-08,
      "loss": 0.3549,
      "step": 28078
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.720759391784668,
      "learning_rate": 5.3635506044089626e-08,
      "loss": 0.2356,
      "step": 28079
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.0719473361968994,
      "learning_rate": 5.352007772701506e-08,
      "loss": 0.2732,
      "step": 28080
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.800423264503479,
      "learning_rate": 5.34047734164389e-08,
      "loss": 0.3078,
      "step": 28081
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.387571334838867,
      "learning_rate": 5.3289593113798885e-08,
      "loss": 0.3237,
      "step": 28082
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.64143705368042,
      "learning_rate": 5.3174536820531644e-08,
      "loss": 0.3413,
      "step": 28083
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.751417636871338,
      "learning_rate": 5.3059604538070466e-08,
      "loss": 0.3086,
      "step": 28084
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.4831583499908447,
      "learning_rate": 5.2944796267847544e-08,
      "loss": 0.2878,
      "step": 28085
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.3763351440429688,
      "learning_rate": 5.283011201129618e-08,
      "loss": 0.3145,
      "step": 28086
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6563084125518799,
      "learning_rate": 5.2715551769845217e-08,
      "loss": 0.2174,
      "step": 28087
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.943570613861084,
      "learning_rate": 5.260111554492242e-08,
      "loss": 0.2677,
      "step": 28088
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.8241236209869385,
      "learning_rate": 5.2486803337956636e-08,
      "loss": 0.3362,
      "step": 28089
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7922464609146118,
      "learning_rate": 5.237261515037007e-08,
      "loss": 0.3143,
      "step": 28090
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.903077483177185,
      "learning_rate": 5.225855098358823e-08,
      "loss": 0.2471,
      "step": 28091
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.064439535140991,
      "learning_rate": 5.2144610839033326e-08,
      "loss": 0.2557,
      "step": 28092
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0531668663024902,
      "learning_rate": 5.203079471812533e-08,
      "loss": 0.2909,
      "step": 28093
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6306171417236328,
      "learning_rate": 5.1917102622282e-08,
      "loss": 0.2305,
      "step": 28094
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8283582925796509,
      "learning_rate": 5.1803534552923304e-08,
      "loss": 0.3544,
      "step": 28095
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.769407868385315,
      "learning_rate": 5.169009051146368e-08,
      "loss": 0.2587,
      "step": 28096
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.040278911590576,
      "learning_rate": 5.157677049931753e-08,
      "loss": 0.2968,
      "step": 28097
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.646022915840149,
      "learning_rate": 5.1463574517898185e-08,
      "loss": 0.2841,
      "step": 28098
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8610296249389648,
      "learning_rate": 5.135050256861562e-08,
      "loss": 0.2761,
      "step": 28099
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8674739599227905,
      "learning_rate": 5.123755465288094e-08,
      "loss": 0.3982,
      "step": 28100
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6723090410232544,
      "learning_rate": 5.1124730772101894e-08,
      "loss": 0.2939,
      "step": 28101
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8825836181640625,
      "learning_rate": 5.101203092768403e-08,
      "loss": 0.2243,
      "step": 28102
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0673906803131104,
      "learning_rate": 5.089945512103511e-08,
      "loss": 0.3545,
      "step": 28103
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.897865891456604,
      "learning_rate": 5.078700335355513e-08,
      "loss": 0.2391,
      "step": 28104
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.4741785526275635,
      "learning_rate": 5.0674675626649626e-08,
      "loss": 0.3693,
      "step": 28105
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7208843231201172,
      "learning_rate": 5.0562471941716376e-08,
      "loss": 0.3154,
      "step": 28106
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9338258504867554,
      "learning_rate": 5.045039230015425e-08,
      "loss": 0.2861,
      "step": 28107
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.744452953338623,
      "learning_rate": 5.0338436703363246e-08,
      "loss": 0.2802,
      "step": 28108
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9367252588272095,
      "learning_rate": 5.0226605152736697e-08,
      "loss": 0.3833,
      "step": 28109
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.548297882080078,
      "learning_rate": 5.011489764967015e-08,
      "loss": 0.2946,
      "step": 28110
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.995832085609436,
      "learning_rate": 5.000331419555471e-08,
      "loss": 0.3454,
      "step": 28111
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7581523656845093,
      "learning_rate": 4.9891854791782604e-08,
      "loss": 0.248,
      "step": 28112
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.926316499710083,
      "learning_rate": 4.978051943974382e-08,
      "loss": 0.3533,
      "step": 28113
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7882611751556396,
      "learning_rate": 4.9669308140826154e-08,
      "loss": 0.2369,
      "step": 28114
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0586962699890137,
      "learning_rate": 4.955822089641516e-08,
      "loss": 0.2681,
      "step": 28115
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.139704942703247,
      "learning_rate": 4.944725770789749e-08,
      "loss": 0.3352,
      "step": 28116
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.611109495162964,
      "learning_rate": 4.9336418576655385e-08,
      "loss": 0.3421,
      "step": 28117
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8159229755401611,
      "learning_rate": 4.9225703504071075e-08,
      "loss": 0.2394,
      "step": 28118
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.896769404411316,
      "learning_rate": 4.9115112491524566e-08,
      "loss": 0.2558,
      "step": 28119
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8728547096252441,
      "learning_rate": 4.900464554039475e-08,
      "loss": 0.2992,
      "step": 28120
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8862990140914917,
      "learning_rate": 4.8894302652059414e-08,
      "loss": 0.2586,
      "step": 28121
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8280601501464844,
      "learning_rate": 4.878408382789301e-08,
      "loss": 0.2175,
      "step": 28122
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0394155979156494,
      "learning_rate": 4.867398906927112e-08,
      "loss": 0.358,
      "step": 28123
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8320090770721436,
      "learning_rate": 4.856401837756597e-08,
      "loss": 0.2644,
      "step": 28124
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.650197744369507,
      "learning_rate": 4.845417175414757e-08,
      "loss": 0.3125,
      "step": 28125
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.4571006298065186,
      "learning_rate": 4.834444920038706e-08,
      "loss": 0.2997,
      "step": 28126
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6682932376861572,
      "learning_rate": 4.8234850717651106e-08,
      "loss": 0.2585,
      "step": 28127
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.338163375854492,
      "learning_rate": 4.81253763073064e-08,
      "loss": 0.2672,
      "step": 28128
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6293625831604004,
      "learning_rate": 4.801602597071853e-08,
      "loss": 0.3072,
      "step": 28129
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0420374870300293,
      "learning_rate": 4.7906799709249716e-08,
      "loss": 0.228,
      "step": 28130
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.079082489013672,
      "learning_rate": 4.779769752426222e-08,
      "loss": 0.2704,
      "step": 28131
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8319073915481567,
      "learning_rate": 4.768871941711606e-08,
      "loss": 0.2227,
      "step": 28132
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.3682103157043457,
      "learning_rate": 4.757986538917014e-08,
      "loss": 0.3318,
      "step": 28133
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.969520926475525,
      "learning_rate": 4.747113544178228e-08,
      "loss": 0.3286,
      "step": 28134
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0506539344787598,
      "learning_rate": 4.736252957630583e-08,
      "loss": 0.3111,
      "step": 28135
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.9044413566589355,
      "learning_rate": 4.725404779409748e-08,
      "loss": 0.3786,
      "step": 28136
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7726918458938599,
      "learning_rate": 4.714569009650838e-08,
      "loss": 0.2563,
      "step": 28137
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.951067328453064,
      "learning_rate": 4.703745648488967e-08,
      "loss": 0.228,
      "step": 28138
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.819831371307373,
      "learning_rate": 4.692934696059026e-08,
      "loss": 0.3241,
      "step": 28139
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0664706230163574,
      "learning_rate": 4.6821361524959084e-08,
      "loss": 0.3452,
      "step": 28140
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0956430435180664,
      "learning_rate": 4.6713500179341733e-08,
      "loss": 0.2434,
      "step": 28141
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9102486371994019,
      "learning_rate": 4.6605762925082677e-08,
      "loss": 0.2264,
      "step": 28142
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0174436569213867,
      "learning_rate": 4.6498149763525294e-08,
      "loss": 0.3735,
      "step": 28143
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1192948818206787,
      "learning_rate": 4.6390660696010724e-08,
      "loss": 0.2392,
      "step": 28144
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0876832008361816,
      "learning_rate": 4.628329572388124e-08,
      "loss": 0.3072,
      "step": 28145
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.3301305770874023,
      "learning_rate": 4.617605484847243e-08,
      "loss": 0.336,
      "step": 28146
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1020188331604004,
      "learning_rate": 4.606893807112434e-08,
      "loss": 0.305,
      "step": 28147
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1247780323028564,
      "learning_rate": 4.596194539317034e-08,
      "loss": 0.2967,
      "step": 28148
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0667896270751953,
      "learning_rate": 4.585507681594381e-08,
      "loss": 0.2471,
      "step": 28149
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.718626618385315,
      "learning_rate": 4.574833234077924e-08,
      "loss": 0.325,
      "step": 28150
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8573195934295654,
      "learning_rate": 4.564171196900669e-08,
      "loss": 0.2995,
      "step": 28151
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9306761026382446,
      "learning_rate": 4.553521570195396e-08,
      "loss": 0.2255,
      "step": 28152
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.783541202545166,
      "learning_rate": 4.5428843540951116e-08,
      "loss": 0.2024,
      "step": 28153
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.2587549686431885,
      "learning_rate": 4.532259548732265e-08,
      "loss": 0.2925,
      "step": 28154
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.431182622909546,
      "learning_rate": 4.521647154239417e-08,
      "loss": 0.3028,
      "step": 28155
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.2386393547058105,
      "learning_rate": 4.511047170748795e-08,
      "loss": 0.3351,
      "step": 28156
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.472517490386963,
      "learning_rate": 4.500459598392515e-08,
      "loss": 0.3161,
      "step": 28157
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.214022159576416,
      "learning_rate": 4.489884437302805e-08,
      "loss": 0.2201,
      "step": 28158
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.896081805229187,
      "learning_rate": 4.479321687611227e-08,
      "loss": 0.3559,
      "step": 28159
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8573050498962402,
      "learning_rate": 4.4687713494496743e-08,
      "loss": 0.2519,
      "step": 28160
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8577851057052612,
      "learning_rate": 4.458233422949598e-08,
      "loss": 0.2702,
      "step": 28161
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.948871374130249,
      "learning_rate": 4.447707908242338e-08,
      "loss": 0.2622,
      "step": 28162
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.83856201171875,
      "learning_rate": 4.437194805459233e-08,
      "loss": 0.3159,
      "step": 28163
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.7608420848846436,
      "learning_rate": 4.426694114731289e-08,
      "loss": 0.3297,
      "step": 28164
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0636708736419678,
      "learning_rate": 4.416205836189402e-08,
      "loss": 0.3222,
      "step": 28165
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9910168647766113,
      "learning_rate": 4.4057299699643565e-08,
      "loss": 0.3702,
      "step": 28166
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6936841011047363,
      "learning_rate": 4.3952665161868247e-08,
      "loss": 0.2575,
      "step": 28167
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9130690097808838,
      "learning_rate": 4.384815474987147e-08,
      "loss": 0.3711,
      "step": 28168
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.623937964439392,
      "learning_rate": 4.374376846495554e-08,
      "loss": 0.3023,
      "step": 28169
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.868404746055603,
      "learning_rate": 4.363950630842384e-08,
      "loss": 0.2385,
      "step": 28170
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0095269680023193,
      "learning_rate": 4.3535368281576453e-08,
      "loss": 0.2667,
      "step": 28171
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.1945528984069824,
      "learning_rate": 4.3431354385709e-08,
      "loss": 0.3107,
      "step": 28172
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.22119402885437,
      "learning_rate": 4.332746462212045e-08,
      "loss": 0.2646,
      "step": 28173
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.0887374877929688,
      "learning_rate": 4.322369899210643e-08,
      "loss": 0.2962,
      "step": 28174
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.940202236175537,
      "learning_rate": 4.3120057496958134e-08,
      "loss": 0.2427,
      "step": 28175
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.2556400299072266,
      "learning_rate": 4.301654013797007e-08,
      "loss": 0.2632,
      "step": 28176
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9099483489990234,
      "learning_rate": 4.291314691643234e-08,
      "loss": 0.3161,
      "step": 28177
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.073124885559082,
      "learning_rate": 4.2809877833632775e-08,
      "loss": 0.2662,
      "step": 28178
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.8952723741531372,
      "learning_rate": 4.270673289086036e-08,
      "loss": 0.2709,
      "step": 28179
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9648691415786743,
      "learning_rate": 4.260371208940073e-08,
      "loss": 0.255,
      "step": 28180
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.001666784286499,
      "learning_rate": 4.250081543053841e-08,
      "loss": 0.2331,
      "step": 28181
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.633163332939148,
      "learning_rate": 4.2398042915554604e-08,
      "loss": 0.2391,
      "step": 28182
    },
    {
      "epoch": 2.91,
      "grad_norm": 2.2539968490600586,
      "learning_rate": 4.229539454573273e-08,
      "loss": 0.3003,
      "step": 28183
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.892076015472412,
      "learning_rate": 4.219287032235064e-08,
      "loss": 0.2745,
      "step": 28184
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9714324474334717,
      "learning_rate": 4.209047024668844e-08,
      "loss": 0.2602,
      "step": 28185
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8654793500900269,
      "learning_rate": 4.198819432002177e-08,
      "loss": 0.2921,
      "step": 28186
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8426419496536255,
      "learning_rate": 4.188604254362516e-08,
      "loss": 0.2604,
      "step": 28187
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.223919630050659,
      "learning_rate": 4.1784014918773154e-08,
      "loss": 0.4186,
      "step": 28188
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8552393913269043,
      "learning_rate": 4.168211144673806e-08,
      "loss": 0.2307,
      "step": 28189
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.986992359161377,
      "learning_rate": 4.158033212878998e-08,
      "loss": 0.3123,
      "step": 28190
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9130727052688599,
      "learning_rate": 4.1478676966196784e-08,
      "loss": 0.3264,
      "step": 28191
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7591685056686401,
      "learning_rate": 4.137714596022746e-08,
      "loss": 0.2885,
      "step": 28192
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1113758087158203,
      "learning_rate": 4.127573911214766e-08,
      "loss": 0.2891,
      "step": 28193
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7923046350479126,
      "learning_rate": 4.117445642321971e-08,
      "loss": 0.2941,
      "step": 28194
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.022315502166748,
      "learning_rate": 4.1073297894708156e-08,
      "loss": 0.3269,
      "step": 28195
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.6223671436309814,
      "learning_rate": 4.09722635278742e-08,
      "loss": 0.3683,
      "step": 28196
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7932769060134888,
      "learning_rate": 4.087135332397685e-08,
      "loss": 0.3162,
      "step": 28197
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.2337148189544678,
      "learning_rate": 4.077056728427398e-08,
      "loss": 0.2806,
      "step": 28198
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.6168503761291504,
      "learning_rate": 4.066990541002236e-08,
      "loss": 0.3314,
      "step": 28199
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.066798210144043,
      "learning_rate": 4.0569367702477655e-08,
      "loss": 0.3014,
      "step": 28200
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8801496028900146,
      "learning_rate": 4.0468954162892206e-08,
      "loss": 0.3287,
      "step": 28201
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1238362789154053,
      "learning_rate": 4.036866479251944e-08,
      "loss": 0.2678,
      "step": 28202
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9284429550170898,
      "learning_rate": 4.026849959260837e-08,
      "loss": 0.35,
      "step": 28203
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7701467275619507,
      "learning_rate": 4.016845856440798e-08,
      "loss": 0.2663,
      "step": 28204
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.030212640762329,
      "learning_rate": 4.0068541709165076e-08,
      "loss": 0.4178,
      "step": 28205
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.191568613052368,
      "learning_rate": 3.996874902812753e-08,
      "loss": 0.352,
      "step": 28206
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.7648489475250244,
      "learning_rate": 3.9869080522536574e-08,
      "loss": 0.3762,
      "step": 28207
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8457125425338745,
      "learning_rate": 3.976953619363677e-08,
      "loss": 0.3718,
      "step": 28208
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0569310188293457,
      "learning_rate": 3.967011604266824e-08,
      "loss": 0.3047,
      "step": 28209
    },
    {
      "epoch": 2.92,
      "grad_norm": 3.051457166671753,
      "learning_rate": 3.9570820070871094e-08,
      "loss": 0.2884,
      "step": 28210
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8127325773239136,
      "learning_rate": 3.9471648279483244e-08,
      "loss": 0.2597,
      "step": 28211
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.335167407989502,
      "learning_rate": 3.937260066974036e-08,
      "loss": 0.3401,
      "step": 28212
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0064024925231934,
      "learning_rate": 3.927367724287812e-08,
      "loss": 0.3145,
      "step": 28213
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9395743608474731,
      "learning_rate": 3.917487800012998e-08,
      "loss": 0.2717,
      "step": 28214
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.3331258296966553,
      "learning_rate": 3.907620294272718e-08,
      "loss": 0.3427,
      "step": 28215
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8952330350875854,
      "learning_rate": 3.897765207189985e-08,
      "loss": 0.2811,
      "step": 28216
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0760180950164795,
      "learning_rate": 3.8879225388877004e-08,
      "loss": 0.2625,
      "step": 28217
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.2159948348999023,
      "learning_rate": 3.878092289488544e-08,
      "loss": 0.2697,
      "step": 28218
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0948050022125244,
      "learning_rate": 3.8682744591150846e-08,
      "loss": 0.326,
      "step": 28219
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0062613487243652,
      "learning_rate": 3.858469047889779e-08,
      "loss": 0.3542,
      "step": 28220
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7245646715164185,
      "learning_rate": 3.8486760559347526e-08,
      "loss": 0.2886,
      "step": 28221
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.682721734046936,
      "learning_rate": 3.8388954833722406e-08,
      "loss": 0.2402,
      "step": 28222
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.102936029434204,
      "learning_rate": 3.829127330324034e-08,
      "loss": 0.3869,
      "step": 28223
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8496555089950562,
      "learning_rate": 3.819371596912036e-08,
      "loss": 0.255,
      "step": 28224
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6980195045471191,
      "learning_rate": 3.8096282832578154e-08,
      "loss": 0.2785,
      "step": 28225
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9384040832519531,
      "learning_rate": 3.7998973894829425e-08,
      "loss": 0.2817,
      "step": 28226
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7161785364151,
      "learning_rate": 3.7901789157086535e-08,
      "loss": 0.3141,
      "step": 28227
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.768093228340149,
      "learning_rate": 3.780472862056073e-08,
      "loss": 0.3052,
      "step": 28228
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.2390666007995605,
      "learning_rate": 3.770779228646215e-08,
      "loss": 0.2901,
      "step": 28229
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8513360023498535,
      "learning_rate": 3.761098015599984e-08,
      "loss": 0.3937,
      "step": 28230
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.2415225505828857,
      "learning_rate": 3.751429223038172e-08,
      "loss": 0.3413,
      "step": 28231
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7797064781188965,
      "learning_rate": 3.741772851081127e-08,
      "loss": 0.3225,
      "step": 28232
    },
    {
      "epoch": 2.92,
      "grad_norm": 4.017576694488525,
      "learning_rate": 3.732128899849419e-08,
      "loss": 0.3609,
      "step": 28233
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0837228298187256,
      "learning_rate": 3.722497369463063e-08,
      "loss": 0.2811,
      "step": 28234
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.96690034866333,
      "learning_rate": 3.7128782600424074e-08,
      "loss": 0.2892,
      "step": 28235
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0659940242767334,
      "learning_rate": 3.703271571707134e-08,
      "loss": 0.3201,
      "step": 28236
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8871434926986694,
      "learning_rate": 3.693677304577148e-08,
      "loss": 0.2086,
      "step": 28237
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.328530788421631,
      "learning_rate": 3.684095458772019e-08,
      "loss": 0.3188,
      "step": 28238
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.5952157974243164,
      "learning_rate": 3.674526034411319e-08,
      "loss": 0.2409,
      "step": 28239
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1543784141540527,
      "learning_rate": 3.664969031614174e-08,
      "loss": 0.401,
      "step": 28240
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5905247926712036,
      "learning_rate": 3.655424450499823e-08,
      "loss": 0.2445,
      "step": 28241
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7328810691833496,
      "learning_rate": 3.645892291187281e-08,
      "loss": 0.2506,
      "step": 28242
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.785494089126587,
      "learning_rate": 3.636372553795342e-08,
      "loss": 0.2515,
      "step": 28243
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.101346015930176,
      "learning_rate": 3.626865238442801e-08,
      "loss": 0.2485,
      "step": 28244
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.4145846366882324,
      "learning_rate": 3.6173703452480055e-08,
      "loss": 0.2949,
      "step": 28245
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.106497287750244,
      "learning_rate": 3.6078878743295296e-08,
      "loss": 0.2958,
      "step": 28246
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.429016351699829,
      "learning_rate": 3.598417825805389e-08,
      "loss": 0.287,
      "step": 28247
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9322360754013062,
      "learning_rate": 3.588960199793823e-08,
      "loss": 0.2216,
      "step": 28248
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7083418369293213,
      "learning_rate": 3.579514996412736e-08,
      "loss": 0.3112,
      "step": 28249
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5193582773208618,
      "learning_rate": 3.570082215779702e-08,
      "loss": 0.2874,
      "step": 28250
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.752326250076294,
      "learning_rate": 3.560661858012515e-08,
      "loss": 0.3209,
      "step": 28251
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.2459425926208496,
      "learning_rate": 3.551253923228526e-08,
      "loss": 0.3995,
      "step": 28252
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9140853881835938,
      "learning_rate": 3.541858411545085e-08,
      "loss": 0.3439,
      "step": 28253
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.2527670860290527,
      "learning_rate": 3.532475323079321e-08,
      "loss": 0.3247,
      "step": 28254
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9623271226882935,
      "learning_rate": 3.523104657948251e-08,
      "loss": 0.2879,
      "step": 28255
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.150965690612793,
      "learning_rate": 3.5137464162684484e-08,
      "loss": 0.3238,
      "step": 28256
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9338161945343018,
      "learning_rate": 3.504400598156932e-08,
      "loss": 0.3433,
      "step": 28257
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.07245135307312,
      "learning_rate": 3.4950672037301625e-08,
      "loss": 0.2884,
      "step": 28258
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.147421360015869,
      "learning_rate": 3.4857462331042705e-08,
      "loss": 0.4103,
      "step": 28259
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.3356211185455322,
      "learning_rate": 3.4764376863956064e-08,
      "loss": 0.3299,
      "step": 28260
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.4185054302215576,
      "learning_rate": 3.4671415637203e-08,
      "loss": 0.3129,
      "step": 28261
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.3171701431274414,
      "learning_rate": 3.457857865194147e-08,
      "loss": 0.3207,
      "step": 28262
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.4018962383270264,
      "learning_rate": 3.4485865909328345e-08,
      "loss": 0.2916,
      "step": 28263
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.4206409454345703,
      "learning_rate": 3.439327741052157e-08,
      "loss": 0.3447,
      "step": 28264
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.3259952068328857,
      "learning_rate": 3.4300813156673554e-08,
      "loss": 0.3566,
      "step": 28265
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9669564962387085,
      "learning_rate": 3.4208473148937826e-08,
      "loss": 0.3303,
      "step": 28266
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.847715139389038,
      "learning_rate": 3.41162573884668e-08,
      "loss": 0.2476,
      "step": 28267
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6794894933700562,
      "learning_rate": 3.402416587640733e-08,
      "loss": 0.2536,
      "step": 28268
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.115586042404175,
      "learning_rate": 3.3932198613909614e-08,
      "loss": 0.3058,
      "step": 28269
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9439150094985962,
      "learning_rate": 3.384035560212051e-08,
      "loss": 0.2893,
      "step": 28270
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8511918783187866,
      "learning_rate": 3.374863684218355e-08,
      "loss": 0.3098,
      "step": 28271
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0138351917266846,
      "learning_rate": 3.365704233524336e-08,
      "loss": 0.3808,
      "step": 28272
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.35080885887146,
      "learning_rate": 3.356557208244127e-08,
      "loss": 0.3364,
      "step": 28273
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.0597269535064697,
      "learning_rate": 3.347422608491857e-08,
      "loss": 0.2912,
      "step": 28274
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8846592903137207,
      "learning_rate": 3.338300434381325e-08,
      "loss": 0.3198,
      "step": 28275
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8709222078323364,
      "learning_rate": 3.329190686026218e-08,
      "loss": 0.263,
      "step": 28276
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.97565495967865,
      "learning_rate": 3.3200933635402224e-08,
      "loss": 0.3052,
      "step": 28277
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.507952928543091,
      "learning_rate": 3.3110084670368024e-08,
      "loss": 0.3278,
      "step": 28278
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.1746251583099365,
      "learning_rate": 3.30193599662898e-08,
      "loss": 0.22,
      "step": 28279
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6946558952331543,
      "learning_rate": 3.292875952430108e-08,
      "loss": 0.2732,
      "step": 28280
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.134223222732544,
      "learning_rate": 3.2838283345530964e-08,
      "loss": 0.299,
      "step": 28281
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.865714192390442,
      "learning_rate": 3.274793143110633e-08,
      "loss": 0.2261,
      "step": 28282
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0749289989471436,
      "learning_rate": 3.265770378215405e-08,
      "loss": 0.3219,
      "step": 28283
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.330855369567871,
      "learning_rate": 3.256760039979989e-08,
      "loss": 0.316,
      "step": 28284
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8531380891799927,
      "learning_rate": 3.24776212851663e-08,
      "loss": 0.3277,
      "step": 28285
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.689562201499939,
      "learning_rate": 3.2387766439375687e-08,
      "loss": 0.2982,
      "step": 28286
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.765090823173523,
      "learning_rate": 3.229803586354718e-08,
      "loss": 0.2878,
      "step": 28287
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9953718185424805,
      "learning_rate": 3.220842955880099e-08,
      "loss": 0.3814,
      "step": 28288
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.71720552444458,
      "learning_rate": 3.2118947526252886e-08,
      "loss": 0.2629,
      "step": 28289
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9095687866210938,
      "learning_rate": 3.2029589767019754e-08,
      "loss": 0.3149,
      "step": 28290
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0962555408477783,
      "learning_rate": 3.1940356282215147e-08,
      "loss": 0.2835,
      "step": 28291
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7294834852218628,
      "learning_rate": 3.185124707295151e-08,
      "loss": 0.2962,
      "step": 28292
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8715755939483643,
      "learning_rate": 3.176226214034017e-08,
      "loss": 0.2728,
      "step": 28293
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0299971103668213,
      "learning_rate": 3.1673401485489143e-08,
      "loss": 0.2866,
      "step": 28294
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1771726608276367,
      "learning_rate": 3.1584665109507527e-08,
      "loss": 0.3042,
      "step": 28295
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1033289432525635,
      "learning_rate": 3.149605301350223e-08,
      "loss": 0.2642,
      "step": 28296
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.084259271621704,
      "learning_rate": 3.140756519857679e-08,
      "loss": 0.2717,
      "step": 28297
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.423891305923462,
      "learning_rate": 3.13192016658348e-08,
      "loss": 0.3732,
      "step": 28298
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7569804191589355,
      "learning_rate": 3.123096241637757e-08,
      "loss": 0.2861,
      "step": 28299
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.958469271659851,
      "learning_rate": 3.114284745130536e-08,
      "loss": 0.3146,
      "step": 28300
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.6374030113220215,
      "learning_rate": 3.105485677171727e-08,
      "loss": 0.4321,
      "step": 28301
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.5482845306396484,
      "learning_rate": 3.09669903787102e-08,
      "loss": 0.2815,
      "step": 28302
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0074682235717773,
      "learning_rate": 3.0879248273378846e-08,
      "loss": 0.2569,
      "step": 28303
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8236838579177856,
      "learning_rate": 3.0791630456818987e-08,
      "loss": 0.2537,
      "step": 28304
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0164284706115723,
      "learning_rate": 3.070413693011976e-08,
      "loss": 0.3345,
      "step": 28305
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.011502504348755,
      "learning_rate": 3.061676769437583e-08,
      "loss": 0.2955,
      "step": 28306
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1276164054870605,
      "learning_rate": 3.0529522750673e-08,
      "loss": 0.2523,
      "step": 28307
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.988560438156128,
      "learning_rate": 3.044240210010152e-08,
      "loss": 0.2922,
      "step": 28308
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8626545667648315,
      "learning_rate": 3.035540574374607e-08,
      "loss": 0.2906,
      "step": 28309
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1586697101593018,
      "learning_rate": 3.0268533682691335e-08,
      "loss": 0.3136,
      "step": 28310
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8049808740615845,
      "learning_rate": 3.018178591802201e-08,
      "loss": 0.3171,
      "step": 28311
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8766340017318726,
      "learning_rate": 3.009516245081723e-08,
      "loss": 0.2851,
      "step": 28312
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.92770254611969,
      "learning_rate": 3.0008663282158344e-08,
      "loss": 0.3481,
      "step": 28313
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.284022331237793,
      "learning_rate": 2.99222884131245e-08,
      "loss": 0.2864,
      "step": 28314
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.836071252822876,
      "learning_rate": 2.9836037844790386e-08,
      "loss": 0.2634,
      "step": 28315
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8609552383422852,
      "learning_rate": 2.9749911578232927e-08,
      "loss": 0.3101,
      "step": 28316
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0939090251922607,
      "learning_rate": 2.96639096145257e-08,
      "loss": 0.3359,
      "step": 28317
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.117357015609741,
      "learning_rate": 2.9578031954741182e-08,
      "loss": 0.3235,
      "step": 28318
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.3355050086975098,
      "learning_rate": 2.949227859994852e-08,
      "loss": 0.3356,
      "step": 28319
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8027054071426392,
      "learning_rate": 2.940664955121908e-08,
      "loss": 0.2439,
      "step": 28320
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9530932903289795,
      "learning_rate": 2.9321144809618673e-08,
      "loss": 0.2781,
      "step": 28321
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5180763006210327,
      "learning_rate": 2.9235764376214225e-08,
      "loss": 0.2128,
      "step": 28322
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0828146934509277,
      "learning_rate": 2.915050825206933e-08,
      "loss": 0.2931,
      "step": 28323
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.6112916469573975,
      "learning_rate": 2.9065376438247584e-08,
      "loss": 0.3285,
      "step": 28324
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8206031322479248,
      "learning_rate": 2.8980368935810355e-08,
      "loss": 0.2269,
      "step": 28325
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.2244694232940674,
      "learning_rate": 2.8895485745817907e-08,
      "loss": 0.2671,
      "step": 28326
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9525021314620972,
      "learning_rate": 2.8810726869327176e-08,
      "loss": 0.331,
      "step": 28327
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9584335088729858,
      "learning_rate": 2.8726092307396202e-08,
      "loss": 0.2972,
      "step": 28328
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.821547269821167,
      "learning_rate": 2.8641582061078588e-08,
      "loss": 0.3354,
      "step": 28329
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9117693901062012,
      "learning_rate": 2.8557196131430154e-08,
      "loss": 0.3668,
      "step": 28330
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.977177381515503,
      "learning_rate": 2.847293451950117e-08,
      "loss": 0.3459,
      "step": 28331
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9688968658447266,
      "learning_rate": 2.8388797226343024e-08,
      "loss": 0.3215,
      "step": 28332
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7785398960113525,
      "learning_rate": 2.830478425300487e-08,
      "loss": 0.2382,
      "step": 28333
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0957207679748535,
      "learning_rate": 2.8220895600532537e-08,
      "loss": 0.3095,
      "step": 28334
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9516301155090332,
      "learning_rate": 2.813713126997408e-08,
      "loss": 0.247,
      "step": 28335
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7347787618637085,
      "learning_rate": 2.805349126237311e-08,
      "loss": 0.2902,
      "step": 28336
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.069500207901001,
      "learning_rate": 2.7969975578771012e-08,
      "loss": 0.3063,
      "step": 28337
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0157387256622314,
      "learning_rate": 2.788658422021029e-08,
      "loss": 0.3665,
      "step": 28338
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0493481159210205,
      "learning_rate": 2.7803317187731217e-08,
      "loss": 0.3239,
      "step": 28339
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6116572618484497,
      "learning_rate": 2.7720174482370743e-08,
      "loss": 0.3295,
      "step": 28340
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.023815393447876,
      "learning_rate": 2.763715610516582e-08,
      "loss": 0.3351,
      "step": 28341
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7446963787078857,
      "learning_rate": 2.755426205715117e-08,
      "loss": 0.3306,
      "step": 28342
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7590968608856201,
      "learning_rate": 2.7471492339360418e-08,
      "loss": 0.3295,
      "step": 28343
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.952032446861267,
      "learning_rate": 2.7388846952826064e-08,
      "loss": 0.2799,
      "step": 28344
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.1174817085266113,
      "learning_rate": 2.73063258985784e-08,
      "loss": 0.3635,
      "step": 28345
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0754709243774414,
      "learning_rate": 2.722392917764549e-08,
      "loss": 0.3354,
      "step": 28346
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.2476043701171875,
      "learning_rate": 2.7141656791054295e-08,
      "loss": 0.3758,
      "step": 28347
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7716376781463623,
      "learning_rate": 2.7059508739831763e-08,
      "loss": 0.3184,
      "step": 28348
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9321646690368652,
      "learning_rate": 2.6977485025002636e-08,
      "loss": 0.3359,
      "step": 28349
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.079423189163208,
      "learning_rate": 2.689558564758721e-08,
      "loss": 0.3208,
      "step": 28350
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8629282712936401,
      "learning_rate": 2.6813810608607992e-08,
      "loss": 0.344,
      "step": 28351
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.31559681892395,
      "learning_rate": 2.6732159909084177e-08,
      "loss": 0.38,
      "step": 28352
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5561083555221558,
      "learning_rate": 2.6650633550033835e-08,
      "loss": 0.2357,
      "step": 28353
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.2285549640655518,
      "learning_rate": 2.6569231532473927e-08,
      "loss": 0.2984,
      "step": 28354
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9798686504364014,
      "learning_rate": 2.648795385741809e-08,
      "loss": 0.3831,
      "step": 28355
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8317450284957886,
      "learning_rate": 2.6406800525881072e-08,
      "loss": 0.3027,
      "step": 28356
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.3028762340545654,
      "learning_rate": 2.632577153887317e-08,
      "loss": 0.2892,
      "step": 28357
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9221711158752441,
      "learning_rate": 2.6244866897405796e-08,
      "loss": 0.3133,
      "step": 28358
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.78252375125885,
      "learning_rate": 2.6164086602487037e-08,
      "loss": 0.2776,
      "step": 28359
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7478394508361816,
      "learning_rate": 2.6083430655123866e-08,
      "loss": 0.3134,
      "step": 28360
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0049381256103516,
      "learning_rate": 2.600289905632325e-08,
      "loss": 0.2857,
      "step": 28361
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8155500888824463,
      "learning_rate": 2.5922491807087724e-08,
      "loss": 0.3117,
      "step": 28362
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6944671869277954,
      "learning_rate": 2.5842208908419818e-08,
      "loss": 0.3154,
      "step": 28363
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7728756666183472,
      "learning_rate": 2.5762050361319846e-08,
      "loss": 0.3603,
      "step": 28364
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9604523181915283,
      "learning_rate": 2.5682016166789224e-08,
      "loss": 0.2881,
      "step": 28365
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.514751434326172,
      "learning_rate": 2.560210632582494e-08,
      "loss": 0.3817,
      "step": 28366
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.9184046983718872,
      "learning_rate": 2.5522320839421742e-08,
      "loss": 0.3803,
      "step": 28367
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7285873889923096,
      "learning_rate": 2.5442659708576623e-08,
      "loss": 0.2615,
      "step": 28368
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.97221040725708,
      "learning_rate": 2.5363122934281003e-08,
      "loss": 0.4167,
      "step": 28369
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.4782204627990723,
      "learning_rate": 2.5283710517527426e-08,
      "loss": 0.3718,
      "step": 28370
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8056131601333618,
      "learning_rate": 2.520442245930621e-08,
      "loss": 0.2972,
      "step": 28371
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.658823013305664,
      "learning_rate": 2.5125258760604344e-08,
      "loss": 0.2683,
      "step": 28372
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.486631155014038,
      "learning_rate": 2.5046219422409923e-08,
      "loss": 0.2475,
      "step": 28373
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7760779857635498,
      "learning_rate": 2.4967304445708828e-08,
      "loss": 0.2764,
      "step": 28374
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.863844871520996,
      "learning_rate": 2.4888513831484717e-08,
      "loss": 0.323,
      "step": 28375
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.788352608680725,
      "learning_rate": 2.4809847580719025e-08,
      "loss": 0.326,
      "step": 28376
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.860409140586853,
      "learning_rate": 2.4731305694392084e-08,
      "loss": 0.2531,
      "step": 28377
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.563223123550415,
      "learning_rate": 2.4652888173485324e-08,
      "loss": 0.3246,
      "step": 28378
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9890753030776978,
      "learning_rate": 2.4574595018974633e-08,
      "loss": 0.3522,
      "step": 28379
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.882695198059082,
      "learning_rate": 2.449642623183701e-08,
      "loss": 0.311,
      "step": 28380
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.2523438930511475,
      "learning_rate": 2.441838181304612e-08,
      "loss": 0.3882,
      "step": 28381
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7177680730819702,
      "learning_rate": 2.4340461763575628e-08,
      "loss": 0.2574,
      "step": 28382
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.070965528488159,
      "learning_rate": 2.426266608439698e-08,
      "loss": 0.2997,
      "step": 28383
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.2358500957489014,
      "learning_rate": 2.4184994776479397e-08,
      "loss": 0.3194,
      "step": 28384
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7007113695144653,
      "learning_rate": 2.410744784079211e-08,
      "loss": 0.1931,
      "step": 28385
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.516078233718872,
      "learning_rate": 2.4030025278301005e-08,
      "loss": 0.2894,
      "step": 28386
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.118252992630005,
      "learning_rate": 2.3952727089973092e-08,
      "loss": 0.3513,
      "step": 28387
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1048319339752197,
      "learning_rate": 2.3875553276770935e-08,
      "loss": 0.2784,
      "step": 28388
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9504624605178833,
      "learning_rate": 2.3798503839655986e-08,
      "loss": 0.3162,
      "step": 28389
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.06476092338562,
      "learning_rate": 2.3721578779588584e-08,
      "loss": 0.3098,
      "step": 28390
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9307821989059448,
      "learning_rate": 2.3644778097530186e-08,
      "loss": 0.3302,
      "step": 28391
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.122429847717285,
      "learning_rate": 2.3568101794436693e-08,
      "loss": 0.3426,
      "step": 28392
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.480397939682007,
      "learning_rate": 2.3491549871264007e-08,
      "loss": 0.3154,
      "step": 28393
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.3224005699157715,
      "learning_rate": 2.3415122328965813e-08,
      "loss": 0.3922,
      "step": 28394
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9322943687438965,
      "learning_rate": 2.3338819168496896e-08,
      "loss": 0.2757,
      "step": 28395
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1056430339813232,
      "learning_rate": 2.326264039080761e-08,
      "loss": 0.2656,
      "step": 28396
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0508334636688232,
      "learning_rate": 2.31865859968472e-08,
      "loss": 0.3123,
      "step": 28397
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.2440454959869385,
      "learning_rate": 2.311065598756379e-08,
      "loss": 0.3689,
      "step": 28398
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8758171796798706,
      "learning_rate": 2.3034850363905513e-08,
      "loss": 0.2693,
      "step": 28399
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0423877239227295,
      "learning_rate": 2.295916912681606e-08,
      "loss": 0.2534,
      "step": 28400
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9702785015106201,
      "learning_rate": 2.288361227723912e-08,
      "loss": 0.3062,
      "step": 28401
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7812449932098389,
      "learning_rate": 2.280817981611727e-08,
      "loss": 0.2951,
      "step": 28402
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.851147174835205,
      "learning_rate": 2.2732871744390872e-08,
      "loss": 0.2577,
      "step": 28403
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6962559223175049,
      "learning_rate": 2.2657688062998062e-08,
      "loss": 0.2294,
      "step": 28404
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.729321002960205,
      "learning_rate": 2.258262877287698e-08,
      "loss": 0.2549,
      "step": 28405
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0099661350250244,
      "learning_rate": 2.2507693874963545e-08,
      "loss": 0.2871,
      "step": 28406
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.478515148162842,
      "learning_rate": 2.243288337019034e-08,
      "loss": 0.3788,
      "step": 28407
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8971024751663208,
      "learning_rate": 2.2358197259493285e-08,
      "loss": 0.2985,
      "step": 28408
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.986419439315796,
      "learning_rate": 2.228363554379942e-08,
      "loss": 0.2845,
      "step": 28409
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.577608346939087,
      "learning_rate": 2.220919822404244e-08,
      "loss": 0.3783,
      "step": 28410
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9240484237670898,
      "learning_rate": 2.2134885301147158e-08,
      "loss": 0.3003,
      "step": 28411
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7157361507415771,
      "learning_rate": 2.2060696776041723e-08,
      "loss": 0.3223,
      "step": 28412
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.44486927986145,
      "learning_rate": 2.198663264964984e-08,
      "loss": 0.3383,
      "step": 28413
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8601629734039307,
      "learning_rate": 2.1912692922897437e-08,
      "loss": 0.254,
      "step": 28414
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8212077617645264,
      "learning_rate": 2.1838877596702668e-08,
      "loss": 0.3456,
      "step": 28415
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1814558506011963,
      "learning_rate": 2.1765186671989234e-08,
      "loss": 0.3118,
      "step": 28416
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.216101884841919,
      "learning_rate": 2.1691620149674187e-08,
      "loss": 0.2899,
      "step": 28417
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0370349884033203,
      "learning_rate": 2.1618178030674563e-08,
      "loss": 0.3713,
      "step": 28418
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.887581467628479,
      "learning_rate": 2.15448603159063e-08,
      "loss": 0.2371,
      "step": 28419
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8572918176651,
      "learning_rate": 2.1471667006283113e-08,
      "loss": 0.2595,
      "step": 28420
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.323265790939331,
      "learning_rate": 2.1398598102718716e-08,
      "loss": 0.2864,
      "step": 28421
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8017685413360596,
      "learning_rate": 2.1325653606123486e-08,
      "loss": 0.2441,
      "step": 28422
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0387046337127686,
      "learning_rate": 2.1252833517406702e-08,
      "loss": 0.3162,
      "step": 28423
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9078435897827148,
      "learning_rate": 2.118013783747541e-08,
      "loss": 0.274,
      "step": 28424
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.020801305770874,
      "learning_rate": 2.1107566567237782e-08,
      "loss": 0.3708,
      "step": 28425
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1057164669036865,
      "learning_rate": 2.103511970759642e-08,
      "loss": 0.3137,
      "step": 28426
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9976122379302979,
      "learning_rate": 2.096279725945616e-08,
      "loss": 0.2569,
      "step": 28427
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9882473945617676,
      "learning_rate": 2.08905992237185e-08,
      "loss": 0.2894,
      "step": 28428
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9214056730270386,
      "learning_rate": 2.0818525601283834e-08,
      "loss": 0.2883,
      "step": 28429
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.7141687870025635,
      "learning_rate": 2.0746576393049222e-08,
      "loss": 0.4056,
      "step": 28430
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.3989217281341553,
      "learning_rate": 2.0674751599912833e-08,
      "loss": 0.316,
      "step": 28431
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9245258569717407,
      "learning_rate": 2.060305122276951e-08,
      "loss": 0.3077,
      "step": 28432
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0116608142852783,
      "learning_rate": 2.053147526251409e-08,
      "loss": 0.25,
      "step": 28433
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9866751432418823,
      "learning_rate": 2.046002372003808e-08,
      "loss": 0.3037,
      "step": 28434
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0111775398254395,
      "learning_rate": 2.0388696596232994e-08,
      "loss": 0.3098,
      "step": 28435
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0363502502441406,
      "learning_rate": 2.0317493891988117e-08,
      "loss": 0.2892,
      "step": 28436
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.3741419315338135,
      "learning_rate": 2.0246415608190516e-08,
      "loss": 0.287,
      "step": 28437
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0443665981292725,
      "learning_rate": 2.0175461745726154e-08,
      "loss": 0.2646,
      "step": 28438
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.2054593563079834,
      "learning_rate": 2.0104632305479877e-08,
      "loss": 0.3378,
      "step": 28439
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.096526861190796,
      "learning_rate": 2.003392728833542e-08,
      "loss": 0.3263,
      "step": 28440
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1129672527313232,
      "learning_rate": 1.9963346695174302e-08,
      "loss": 0.321,
      "step": 28441
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.028911590576172,
      "learning_rate": 1.989289052687471e-08,
      "loss": 0.272,
      "step": 28442
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.916186571121216,
      "learning_rate": 1.9822558784318158e-08,
      "loss": 0.315,
      "step": 28443
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.031829833984375,
      "learning_rate": 1.9752351468378396e-08,
      "loss": 0.2854,
      "step": 28444
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9430875778198242,
      "learning_rate": 1.9682268579932496e-08,
      "loss": 0.2954,
      "step": 28445
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8603551387786865,
      "learning_rate": 1.9612310119853094e-08,
      "loss": 0.2668,
      "step": 28446
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7948071956634521,
      "learning_rate": 1.9542476089013938e-08,
      "loss": 0.2667,
      "step": 28447
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.2630727291107178,
      "learning_rate": 1.9472766488284335e-08,
      "loss": 0.3273,
      "step": 28448
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7721439599990845,
      "learning_rate": 1.9403181318533582e-08,
      "loss": 0.3315,
      "step": 28449
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.744438886642456,
      "learning_rate": 1.933372058062988e-08,
      "loss": 0.2639,
      "step": 28450
    },
    {
      "epoch": 2.94,
      "grad_norm": 3.1859495639801025,
      "learning_rate": 1.9264384275438087e-08,
      "loss": 0.2386,
      "step": 28451
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.947047233581543,
      "learning_rate": 1.919517240382418e-08,
      "loss": 0.2746,
      "step": 28452
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.039003849029541,
      "learning_rate": 1.9126084966649695e-08,
      "loss": 0.4048,
      "step": 28453
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.132157325744629,
      "learning_rate": 1.905712196477727e-08,
      "loss": 0.3376,
      "step": 28454
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9378219842910767,
      "learning_rate": 1.898828339906511e-08,
      "loss": 0.3629,
      "step": 28455
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9134782552719116,
      "learning_rate": 1.8919569270372528e-08,
      "loss": 0.3292,
      "step": 28456
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.254340648651123,
      "learning_rate": 1.8850979579555505e-08,
      "loss": 0.2626,
      "step": 28457
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.182926893234253,
      "learning_rate": 1.8782514327470025e-08,
      "loss": 0.2484,
      "step": 28458
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8573421239852905,
      "learning_rate": 1.8714173514968735e-08,
      "loss": 0.3369,
      "step": 28459
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.887039065361023,
      "learning_rate": 1.8645957142905403e-08,
      "loss": 0.296,
      "step": 28460
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7118397951126099,
      "learning_rate": 1.857786521212823e-08,
      "loss": 0.3342,
      "step": 28461
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8317983150482178,
      "learning_rate": 1.8509897723487658e-08,
      "loss": 0.282,
      "step": 28462
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5356525182724,
      "learning_rate": 1.8442054677829667e-08,
      "loss": 0.2528,
      "step": 28463
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0141854286193848,
      "learning_rate": 1.8374336076002474e-08,
      "loss": 0.3678,
      "step": 28464
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1643757820129395,
      "learning_rate": 1.8306741918847626e-08,
      "loss": 0.3232,
      "step": 28465
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6954517364501953,
      "learning_rate": 1.823927220721e-08,
      "loss": 0.2927,
      "step": 28466
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.1881518363952637,
      "learning_rate": 1.817192694193004e-08,
      "loss": 0.3679,
      "step": 28467
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9381983280181885,
      "learning_rate": 1.8104706123845962e-08,
      "loss": 0.3035,
      "step": 28468
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.087770700454712,
      "learning_rate": 1.80376097537982e-08,
      "loss": 0.3774,
      "step": 28469
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.726955771446228,
      "learning_rate": 1.797063783262165e-08,
      "loss": 0.1971,
      "step": 28470
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8417258262634277,
      "learning_rate": 1.7903790361152308e-08,
      "loss": 0.3009,
      "step": 28471
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6309311389923096,
      "learning_rate": 1.7837067340222835e-08,
      "loss": 0.3247,
      "step": 28472
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0597481727600098,
      "learning_rate": 1.7770468770664794e-08,
      "loss": 0.3874,
      "step": 28473
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.027498245239258,
      "learning_rate": 1.770399465330974e-08,
      "loss": 0.3199,
      "step": 28474
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.606122612953186,
      "learning_rate": 1.7637644988985903e-08,
      "loss": 0.294,
      "step": 28475
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9832555055618286,
      "learning_rate": 1.7571419778519285e-08,
      "loss": 0.3355,
      "step": 28476
    },
    {
      "epoch": 2.95,
      "grad_norm": 3.140932083129883,
      "learning_rate": 1.7505319022737e-08,
      "loss": 0.366,
      "step": 28477
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1598732471466064,
      "learning_rate": 1.7439342722462837e-08,
      "loss": 0.225,
      "step": 28478
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2067017555236816,
      "learning_rate": 1.737349087851947e-08,
      "loss": 0.3137,
      "step": 28479
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.072965383529663,
      "learning_rate": 1.7307763491727358e-08,
      "loss": 0.3481,
      "step": 28480
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.0821919441223145,
      "learning_rate": 1.724216056290584e-08,
      "loss": 0.3378,
      "step": 28481
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.056638717651367,
      "learning_rate": 1.7176682092873153e-08,
      "loss": 0.2724,
      "step": 28482
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8945156335830688,
      "learning_rate": 1.711132808244642e-08,
      "loss": 0.3408,
      "step": 28483
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1375651359558105,
      "learning_rate": 1.7046098532439435e-08,
      "loss": 0.3376,
      "step": 28484
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6527782678604126,
      "learning_rate": 1.6980993443665994e-08,
      "loss": 0.2848,
      "step": 28485
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1536731719970703,
      "learning_rate": 1.6916012816936554e-08,
      "loss": 0.246,
      "step": 28486
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.006403684616089,
      "learning_rate": 1.6851156653062693e-08,
      "loss": 0.3321,
      "step": 28487
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3250539302825928,
      "learning_rate": 1.6786424952852653e-08,
      "loss": 0.3112,
      "step": 28488
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3190481662750244,
      "learning_rate": 1.6721817717113565e-08,
      "loss": 0.26,
      "step": 28489
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.4711620807647705,
      "learning_rate": 1.665733494664923e-08,
      "loss": 0.3156,
      "step": 28490
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.040066957473755,
      "learning_rate": 1.659297664226678e-08,
      "loss": 0.3082,
      "step": 28491
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.824766993522644,
      "learning_rate": 1.6528742804765575e-08,
      "loss": 0.2976,
      "step": 28492
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.731366515159607,
      "learning_rate": 1.6464633434948306e-08,
      "loss": 0.3373,
      "step": 28493
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9807381629943848,
      "learning_rate": 1.6400648533614338e-08,
      "loss": 0.2799,
      "step": 28494
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.932407259941101,
      "learning_rate": 1.6336788101559698e-08,
      "loss": 0.3628,
      "step": 28495
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8835387229919434,
      "learning_rate": 1.6273052139581523e-08,
      "loss": 0.2741,
      "step": 28496
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.947575330734253,
      "learning_rate": 1.620944064847474e-08,
      "loss": 0.328,
      "step": 28497
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.953181266784668,
      "learning_rate": 1.614595362903204e-08,
      "loss": 0.2639,
      "step": 28498
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.871975064277649,
      "learning_rate": 1.608259108204502e-08,
      "loss": 0.3436,
      "step": 28499
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6618564128875732,
      "learning_rate": 1.601935300830415e-08,
      "loss": 0.317,
      "step": 28500
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8750277757644653,
      "learning_rate": 1.5956239408596585e-08,
      "loss": 0.3558,
      "step": 28501
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3986294269561768,
      "learning_rate": 1.589325028371058e-08,
      "loss": 0.2781,
      "step": 28502
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.329709529876709,
      "learning_rate": 1.5830385634429958e-08,
      "loss": 0.2432,
      "step": 28503
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.126645565032959,
      "learning_rate": 1.5767645461540747e-08,
      "loss": 0.3316,
      "step": 28504
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.830558180809021,
      "learning_rate": 1.5705029765822333e-08,
      "loss": 0.3142,
      "step": 28505
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.067650556564331,
      "learning_rate": 1.5642538548057417e-08,
      "loss": 0.3156,
      "step": 28506
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6694879531860352,
      "learning_rate": 1.558017180902427e-08,
      "loss": 0.306,
      "step": 28507
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8241561651229858,
      "learning_rate": 1.5517929549501154e-08,
      "loss": 0.3002,
      "step": 28508
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.35243821144104,
      "learning_rate": 1.5455811770263008e-08,
      "loss": 0.2963,
      "step": 28509
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2469794750213623,
      "learning_rate": 1.5393818472084764e-08,
      "loss": 0.332,
      "step": 28510
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9871697425842285,
      "learning_rate": 1.5331949655740252e-08,
      "loss": 0.3467,
      "step": 28511
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2998828887939453,
      "learning_rate": 1.5270205321998853e-08,
      "loss": 0.3707,
      "step": 28512
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.045947313308716,
      "learning_rate": 1.5208585471632174e-08,
      "loss": 0.2974,
      "step": 28513
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.462179183959961,
      "learning_rate": 1.514709010540738e-08,
      "loss": 0.3787,
      "step": 28514
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1626853942871094,
      "learning_rate": 1.5085719224091632e-08,
      "loss": 0.2916,
      "step": 28515
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.497119665145874,
      "learning_rate": 1.5024472828449876e-08,
      "loss": 0.3074,
      "step": 28516
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.028810501098633,
      "learning_rate": 1.4963350919245946e-08,
      "loss": 0.3547,
      "step": 28517
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.187514543533325,
      "learning_rate": 1.490235349724145e-08,
      "loss": 0.395,
      "step": 28518
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9666175842285156,
      "learning_rate": 1.4841480563196898e-08,
      "loss": 0.3375,
      "step": 28519
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.883030652999878,
      "learning_rate": 1.4780732117871676e-08,
      "loss": 0.2442,
      "step": 28520
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1061527729034424,
      "learning_rate": 1.4720108162022962e-08,
      "loss": 0.3495,
      "step": 28521
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.066516160964966,
      "learning_rate": 1.4659608696405703e-08,
      "loss": 0.3422,
      "step": 28522
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2276225090026855,
      "learning_rate": 1.4599233721775962e-08,
      "loss": 0.3215,
      "step": 28523
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2839789390563965,
      "learning_rate": 1.4538983238885362e-08,
      "loss": 0.3656,
      "step": 28524
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2163994312286377,
      "learning_rate": 1.4478857248484413e-08,
      "loss": 0.3006,
      "step": 28525
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9230737686157227,
      "learning_rate": 1.4418855751323624e-08,
      "loss": 0.2635,
      "step": 28526
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.12876558303833,
      "learning_rate": 1.4358978748151286e-08,
      "loss": 0.2632,
      "step": 28527
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8986531496047974,
      "learning_rate": 1.4299226239713471e-08,
      "loss": 0.2581,
      "step": 28528
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9879173040390015,
      "learning_rate": 1.4239598226755135e-08,
      "loss": 0.2545,
      "step": 28529
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7848522663116455,
      "learning_rate": 1.418009471002013e-08,
      "loss": 0.3125,
      "step": 28530
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.74827241897583,
      "learning_rate": 1.4120715690250087e-08,
      "loss": 0.2584,
      "step": 28531
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.194873332977295,
      "learning_rate": 1.406146116818441e-08,
      "loss": 0.317,
      "step": 28532
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8951606750488281,
      "learning_rate": 1.4002331144562509e-08,
      "loss": 0.2483,
      "step": 28533
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.129448652267456,
      "learning_rate": 1.3943325620122683e-08,
      "loss": 0.3816,
      "step": 28534
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.229513645172119,
      "learning_rate": 1.3884444595597679e-08,
      "loss": 0.3162,
      "step": 28535
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9900587797164917,
      "learning_rate": 1.3825688071724686e-08,
      "loss": 0.329,
      "step": 28536
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.623884439468384,
      "learning_rate": 1.3767056049235338e-08,
      "loss": 0.3112,
      "step": 28537
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1028995513916016,
      "learning_rate": 1.3708548528860167e-08,
      "loss": 0.3153,
      "step": 28538
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8027558326721191,
      "learning_rate": 1.3650165511328584e-08,
      "loss": 0.254,
      "step": 28539
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3390414714813232,
      "learning_rate": 1.3591906997367788e-08,
      "loss": 0.3768,
      "step": 28540
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9506336450576782,
      "learning_rate": 1.3533772987706084e-08,
      "loss": 0.2891,
      "step": 28541
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.001411199569702,
      "learning_rate": 1.3475763483066228e-08,
      "loss": 0.2753,
      "step": 28542
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.851924180984497,
      "learning_rate": 1.3417878484173197e-08,
      "loss": 0.3357,
      "step": 28543
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3183186054229736,
      "learning_rate": 1.3360117991747523e-08,
      "loss": 0.2825,
      "step": 28544
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.6989452838897705,
      "learning_rate": 1.3302482006509743e-08,
      "loss": 0.3973,
      "step": 28545
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.866719126701355,
      "learning_rate": 1.324497052917817e-08,
      "loss": 0.2575,
      "step": 28546
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.992888331413269,
      "learning_rate": 1.3187583560470008e-08,
      "loss": 0.3412,
      "step": 28547
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6940348148345947,
      "learning_rate": 1.313032110110024e-08,
      "loss": 0.2785,
      "step": 28548
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7654951810836792,
      "learning_rate": 1.3073183151782742e-08,
      "loss": 0.2505,
      "step": 28549
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.5432050228118896,
      "learning_rate": 1.3016169713231386e-08,
      "loss": 0.2765,
      "step": 28550
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.180724859237671,
      "learning_rate": 1.2959280786155603e-08,
      "loss": 0.2391,
      "step": 28551
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.1656837463378906,
      "learning_rate": 1.2902516371264829e-08,
      "loss": 0.2729,
      "step": 28552
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8632172346115112,
      "learning_rate": 1.2845876469267382e-08,
      "loss": 0.2911,
      "step": 28553
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5805282592773438,
      "learning_rate": 1.2789361080868256e-08,
      "loss": 0.2657,
      "step": 28554
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.518801689147949,
      "learning_rate": 1.2732970206773553e-08,
      "loss": 0.2768,
      "step": 28555
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2578327655792236,
      "learning_rate": 1.2676703847684934e-08,
      "loss": 0.2857,
      "step": 28556
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8939402103424072,
      "learning_rate": 1.262056200430406e-08,
      "loss": 0.2842,
      "step": 28557
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7553868293762207,
      "learning_rate": 1.256454467733148e-08,
      "loss": 0.2451,
      "step": 28558
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.0254435539245605,
      "learning_rate": 1.2508651867465527e-08,
      "loss": 0.2897,
      "step": 28559
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.014025926589966,
      "learning_rate": 1.2452883575403419e-08,
      "loss": 0.2525,
      "step": 28560
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.3708624839782715,
      "learning_rate": 1.2397239801839045e-08,
      "loss": 0.3393,
      "step": 28561
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.99901282787323,
      "learning_rate": 1.2341720547467406e-08,
      "loss": 0.3243,
      "step": 28562
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2464771270751953,
      "learning_rate": 1.228632581298017e-08,
      "loss": 0.3223,
      "step": 28563
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.0587515830993652,
      "learning_rate": 1.2231055599067898e-08,
      "loss": 0.2771,
      "step": 28564
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6222480535507202,
      "learning_rate": 1.2175909906420035e-08,
      "loss": 0.2708,
      "step": 28565
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.2423441410064697,
      "learning_rate": 1.2120888735723812e-08,
      "loss": 0.2629,
      "step": 28566
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8779211044311523,
      "learning_rate": 1.2065992087666455e-08,
      "loss": 0.3593,
      "step": 28567
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7798115015029907,
      "learning_rate": 1.2011219962929643e-08,
      "loss": 0.287,
      "step": 28568
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.0271387100219727,
      "learning_rate": 1.1956572362198382e-08,
      "loss": 0.3016,
      "step": 28569
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9042603969573975,
      "learning_rate": 1.1902049286153239e-08,
      "loss": 0.3558,
      "step": 28570
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7511311769485474,
      "learning_rate": 1.184765073547367e-08,
      "loss": 0.3062,
      "step": 28571
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8033537864685059,
      "learning_rate": 1.1793376710839133e-08,
      "loss": 0.2607,
      "step": 28572
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7858654260635376,
      "learning_rate": 1.1739227212925752e-08,
      "loss": 0.2593,
      "step": 28573
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.753077745437622,
      "learning_rate": 1.1685202242407434e-08,
      "loss": 0.397,
      "step": 28574
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.570065498352051,
      "learning_rate": 1.1631301799959194e-08,
      "loss": 0.287,
      "step": 28575
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.788330078125,
      "learning_rate": 1.1577525886252715e-08,
      "loss": 0.212,
      "step": 28576
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.2667946815490723,
      "learning_rate": 1.1523874501957466e-08,
      "loss": 0.3209,
      "step": 28577
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1353518962860107,
      "learning_rate": 1.1470347647744017e-08,
      "loss": 0.3188,
      "step": 28578
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.449206590652466,
      "learning_rate": 1.1416945324278505e-08,
      "loss": 0.2974,
      "step": 28579
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1408932209014893,
      "learning_rate": 1.1363667532227062e-08,
      "loss": 0.3814,
      "step": 28580
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9592533111572266,
      "learning_rate": 1.1310514272253603e-08,
      "loss": 0.2709,
      "step": 28581
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8889013528823853,
      "learning_rate": 1.125748554502093e-08,
      "loss": 0.3251,
      "step": 28582
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.695949912071228,
      "learning_rate": 1.1204581351190735e-08,
      "loss": 0.283,
      "step": 28583
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.807800531387329,
      "learning_rate": 1.1151801691422492e-08,
      "loss": 0.2147,
      "step": 28584
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0057051181793213,
      "learning_rate": 1.1099146566373453e-08,
      "loss": 0.301,
      "step": 28585
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.988429069519043,
      "learning_rate": 1.104661597670087e-08,
      "loss": 0.3284,
      "step": 28586
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.461984157562256,
      "learning_rate": 1.0994209923058664e-08,
      "loss": 0.2814,
      "step": 28587
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.096707344055176,
      "learning_rate": 1.0941928406100755e-08,
      "loss": 0.3114,
      "step": 28588
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.2447826862335205,
      "learning_rate": 1.0889771426479956e-08,
      "loss": 0.3446,
      "step": 28589
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0133931636810303,
      "learning_rate": 1.0837738984845747e-08,
      "loss": 0.3177,
      "step": 28590
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9617326259613037,
      "learning_rate": 1.0785831081845388e-08,
      "loss": 0.3802,
      "step": 28591
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8667768239974976,
      "learning_rate": 1.073404771812836e-08,
      "loss": 0.323,
      "step": 28592
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.34683895111084,
      "learning_rate": 1.0682388894339702e-08,
      "loss": 0.3545,
      "step": 28593
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.4385266304016113,
      "learning_rate": 1.0630854611122232e-08,
      "loss": 0.3149,
      "step": 28594
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.150912284851074,
      "learning_rate": 1.0579444869119881e-08,
      "loss": 0.3402,
      "step": 28595
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8639752864837646,
      "learning_rate": 1.0528159668973247e-08,
      "loss": 0.2835,
      "step": 28596
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1978559494018555,
      "learning_rate": 1.047699901132071e-08,
      "loss": 0.2508,
      "step": 28597
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7624129056930542,
      "learning_rate": 1.0425962896800646e-08,
      "loss": 0.3357,
      "step": 28598
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9330132007598877,
      "learning_rate": 1.0375051326050323e-08,
      "loss": 0.3034,
      "step": 28599
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0241219997406006,
      "learning_rate": 1.032426429970368e-08,
      "loss": 0.2817,
      "step": 28600
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.468106269836426,
      "learning_rate": 1.027360181839354e-08,
      "loss": 0.2879,
      "step": 28601
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.672692060470581,
      "learning_rate": 1.0223063882751627e-08,
      "loss": 0.2348,
      "step": 28602
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.474684238433838,
      "learning_rate": 1.0172650493408542e-08,
      "loss": 0.3754,
      "step": 28603
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8498095273971558,
      "learning_rate": 1.0122361650992673e-08,
      "loss": 0.2914,
      "step": 28604
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8333219289779663,
      "learning_rate": 1.0072197356131298e-08,
      "loss": 0.3142,
      "step": 28605
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7129777669906616,
      "learning_rate": 1.0022157609449467e-08,
      "loss": 0.2889,
      "step": 28606
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9179174900054932,
      "learning_rate": 9.972242411570021e-09,
      "loss": 0.2818,
      "step": 28607
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.986432433128357,
      "learning_rate": 9.922451763116903e-09,
      "loss": 0.3107,
      "step": 28608
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8977177143096924,
      "learning_rate": 9.872785664709616e-09,
      "loss": 0.323,
      "step": 28609
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8013651371002197,
      "learning_rate": 9.823244116968778e-09,
      "loss": 0.2503,
      "step": 28610
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.5790607929229736,
      "learning_rate": 9.77382712051056e-09,
      "loss": 0.3544,
      "step": 28611
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.7898781299591064,
      "learning_rate": 9.724534675952247e-09,
      "loss": 0.3809,
      "step": 28612
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.687767505645752,
      "learning_rate": 9.675366783907792e-09,
      "loss": 0.3339,
      "step": 28613
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.623279094696045,
      "learning_rate": 9.626323444990038e-09,
      "loss": 0.2667,
      "step": 28614
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.63882315158844,
      "learning_rate": 9.577404659810719e-09,
      "loss": 0.2879,
      "step": 28615
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1885063648223877,
      "learning_rate": 9.528610428980457e-09,
      "loss": 0.2789,
      "step": 28616
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.172945737838745,
      "learning_rate": 9.479940753105433e-09,
      "loss": 0.2553,
      "step": 28617
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.863379955291748,
      "learning_rate": 9.43139563279516e-09,
      "loss": 0.2554,
      "step": 28618
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8833516836166382,
      "learning_rate": 9.38297506865249e-09,
      "loss": 0.2821,
      "step": 28619
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.2559807300567627,
      "learning_rate": 9.334679061282492e-09,
      "loss": 0.2582,
      "step": 28620
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1325461864471436,
      "learning_rate": 9.28650761128691e-09,
      "loss": 0.2838,
      "step": 28621
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.022036075592041,
      "learning_rate": 9.238460719266373e-09,
      "loss": 0.31,
      "step": 28622
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.292830228805542,
      "learning_rate": 9.190538385819292e-09,
      "loss": 0.2865,
      "step": 28623
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9206160306930542,
      "learning_rate": 9.142740611544076e-09,
      "loss": 0.3155,
      "step": 28624
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.957933783531189,
      "learning_rate": 9.095067397036916e-09,
      "loss": 0.3095,
      "step": 28625
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9156287908554077,
      "learning_rate": 9.047518742890671e-09,
      "loss": 0.2734,
      "step": 28626
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9588019847869873,
      "learning_rate": 9.000094649698199e-09,
      "loss": 0.3023,
      "step": 28627
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.588592767715454,
      "learning_rate": 8.95279511805347e-09,
      "loss": 0.3459,
      "step": 28628
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7590960264205933,
      "learning_rate": 8.905620148542682e-09,
      "loss": 0.2777,
      "step": 28629
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1292996406555176,
      "learning_rate": 8.858569741756473e-09,
      "loss": 0.4048,
      "step": 28630
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0526926517486572,
      "learning_rate": 8.811643898279931e-09,
      "loss": 0.2829,
      "step": 28631
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0615086555480957,
      "learning_rate": 8.764842618699254e-09,
      "loss": 0.2712,
      "step": 28632
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.3415071964263916,
      "learning_rate": 8.718165903596198e-09,
      "loss": 0.3762,
      "step": 28633
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.13299822807312,
      "learning_rate": 8.671613753555853e-09,
      "loss": 0.362,
      "step": 28634
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9474917650222778,
      "learning_rate": 8.625186169154421e-09,
      "loss": 0.2919,
      "step": 28635
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9830182790756226,
      "learning_rate": 8.57888315097477e-09,
      "loss": 0.3369,
      "step": 28636
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.2706077098846436,
      "learning_rate": 8.532704699590888e-09,
      "loss": 0.319,
      "step": 28637
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.7659096717834473,
      "learning_rate": 8.486650815581199e-09,
      "loss": 0.2358,
      "step": 28638
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8267911672592163,
      "learning_rate": 8.440721499517467e-09,
      "loss": 0.2875,
      "step": 28639
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0662841796875,
      "learning_rate": 8.39491675197368e-09,
      "loss": 0.3688,
      "step": 28640
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9953974485397339,
      "learning_rate": 8.349236573521601e-09,
      "loss": 0.2537,
      "step": 28641
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.3937556743621826,
      "learning_rate": 8.303680964728556e-09,
      "loss": 0.3569,
      "step": 28642
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.556222438812256,
      "learning_rate": 8.258249926165196e-09,
      "loss": 0.3701,
      "step": 28643
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9227107763290405,
      "learning_rate": 8.212943458395517e-09,
      "loss": 0.347,
      "step": 28644
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0998754501342773,
      "learning_rate": 8.167761561985731e-09,
      "loss": 0.2879,
      "step": 28645
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.91488516330719,
      "learning_rate": 8.122704237497614e-09,
      "loss": 0.3124,
      "step": 28646
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.0054798126220703,
      "learning_rate": 8.077771485495156e-09,
      "loss": 0.3387,
      "step": 28647
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1523828506469727,
      "learning_rate": 8.032963306537911e-09,
      "loss": 0.3424,
      "step": 28648
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9279096126556396,
      "learning_rate": 7.98827970118321e-09,
      "loss": 0.2709,
      "step": 28649
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.9512999057769775,
      "learning_rate": 7.943720669989497e-09,
      "loss": 0.2875,
      "step": 28650
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7860678434371948,
      "learning_rate": 7.899286213510771e-09,
      "loss": 0.2226,
      "step": 28651
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.151972770690918,
      "learning_rate": 7.854976332303255e-09,
      "loss": 0.3134,
      "step": 28652
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8546764850616455,
      "learning_rate": 7.81079102691762e-09,
      "loss": 0.3272,
      "step": 28653
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.283987522125244,
      "learning_rate": 7.766730297905644e-09,
      "loss": 0.2645,
      "step": 28654
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.701882243156433,
      "learning_rate": 7.722794145815782e-09,
      "loss": 0.2783,
      "step": 28655
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8688833713531494,
      "learning_rate": 7.67898257119648e-09,
      "loss": 0.2391,
      "step": 28656
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.5333685874938965,
      "learning_rate": 7.63529557459397e-09,
      "loss": 0.3407,
      "step": 28657
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.6580934524536133,
      "learning_rate": 7.591733156552261e-09,
      "loss": 0.365,
      "step": 28658
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.007000207901001,
      "learning_rate": 7.548295317615361e-09,
      "loss": 0.3201,
      "step": 28659
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7264069318771362,
      "learning_rate": 7.50498205832395e-09,
      "loss": 0.2875,
      "step": 28660
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5971075296401978,
      "learning_rate": 7.461793379219816e-09,
      "loss": 0.2387,
      "step": 28661
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8109469413757324,
      "learning_rate": 7.418729280838088e-09,
      "loss": 0.3804,
      "step": 28662
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.6740148067474365,
      "learning_rate": 7.375789763719443e-09,
      "loss": 0.3034,
      "step": 28663
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.054095506668091,
      "learning_rate": 7.332974828396788e-09,
      "loss": 0.2467,
      "step": 28664
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.953003168106079,
      "learning_rate": 7.290284475404141e-09,
      "loss": 0.2809,
      "step": 28665
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.439509630203247,
      "learning_rate": 7.247718705274409e-09,
      "loss": 0.2964,
      "step": 28666
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.908926010131836,
      "learning_rate": 7.205277518538279e-09,
      "loss": 0.3325,
      "step": 28667
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.4222347736358643,
      "learning_rate": 7.1629609157242155e-09,
      "loss": 0.3489,
      "step": 28668
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8260254859924316,
      "learning_rate": 7.1207688973595755e-09,
      "loss": 0.2879,
      "step": 28669
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.6147851943969727,
      "learning_rate": 7.078701463971716e-09,
      "loss": 0.3005,
      "step": 28670
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2014787197113037,
      "learning_rate": 7.03675861608466e-09,
      "loss": 0.2717,
      "step": 28671
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9523630142211914,
      "learning_rate": 6.994940354220214e-09,
      "loss": 0.3019,
      "step": 28672
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.201253652572632,
      "learning_rate": 6.953246678901293e-09,
      "loss": 0.3581,
      "step": 28673
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8651823997497559,
      "learning_rate": 6.911677590647481e-09,
      "loss": 0.3258,
      "step": 28674
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8363144397735596,
      "learning_rate": 6.870233089976141e-09,
      "loss": 0.2956,
      "step": 28675
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.455138921737671,
      "learning_rate": 6.828913177403529e-09,
      "loss": 0.3618,
      "step": 28676
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5293585062026978,
      "learning_rate": 6.787717853447007e-09,
      "loss": 0.2167,
      "step": 28677
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.768297553062439,
      "learning_rate": 6.746647118617278e-09,
      "loss": 0.292,
      "step": 28678
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8515349626541138,
      "learning_rate": 6.705700973429485e-09,
      "loss": 0.2882,
      "step": 28679
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8454327583312988,
      "learning_rate": 6.664879418392112e-09,
      "loss": 0.3526,
      "step": 28680
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6706278324127197,
      "learning_rate": 6.624182454013639e-09,
      "loss": 0.3566,
      "step": 28681
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2476890087127686,
      "learning_rate": 6.5836100808036594e-09,
      "loss": 0.2745,
      "step": 28682
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.048316240310669,
      "learning_rate": 6.5431622992651045e-09,
      "loss": 0.2917,
      "step": 28683
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.160431385040283,
      "learning_rate": 6.502839109905346e-09,
      "loss": 0.3416,
      "step": 28684
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.726670265197754,
      "learning_rate": 6.4626405132250936e-09,
      "loss": 0.2968,
      "step": 28685
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6201837062835693,
      "learning_rate": 6.422566509726169e-09,
      "loss": 0.3088,
      "step": 28686
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7093535661697388,
      "learning_rate": 6.3826170999081725e-09,
      "loss": 0.2259,
      "step": 28687
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.3173046112060547,
      "learning_rate": 6.342792284268484e-09,
      "loss": 0.3145,
      "step": 28688
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.4912655353546143,
      "learning_rate": 6.303092063304484e-09,
      "loss": 0.3439,
      "step": 28689
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.147294044494629,
      "learning_rate": 6.26351643751133e-09,
      "loss": 0.3072,
      "step": 28690
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8890379667282104,
      "learning_rate": 6.2240654073808526e-09,
      "loss": 0.3544,
      "step": 28691
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.754584550857544,
      "learning_rate": 6.1847389734071e-09,
      "loss": 0.3391,
      "step": 28692
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.952064037322998,
      "learning_rate": 6.145537136078572e-09,
      "loss": 0.3048,
      "step": 28693
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.007673740386963,
      "learning_rate": 6.106459895884875e-09,
      "loss": 0.2989,
      "step": 28694
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.872609257698059,
      "learning_rate": 6.067507253312288e-09,
      "loss": 0.244,
      "step": 28695
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9541292190551758,
      "learning_rate": 6.028679208848198e-09,
      "loss": 0.2854,
      "step": 28696
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.203066349029541,
      "learning_rate": 5.989975762975553e-09,
      "loss": 0.3176,
      "step": 28697
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9316909313201904,
      "learning_rate": 5.951396916176188e-09,
      "loss": 0.3641,
      "step": 28698
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0732388496398926,
      "learning_rate": 5.912942668933053e-09,
      "loss": 0.3502,
      "step": 28699
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2735273838043213,
      "learning_rate": 5.874613021723541e-09,
      "loss": 0.323,
      "step": 28700
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0711216926574707,
      "learning_rate": 5.8364079750272695e-09,
      "loss": 0.3443,
      "step": 28701
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.858062744140625,
      "learning_rate": 5.798327529318304e-09,
      "loss": 0.2887,
      "step": 28702
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2420847415924072,
      "learning_rate": 5.76037168507404e-09,
      "loss": 0.3837,
      "step": 28703
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.896592617034912,
      "learning_rate": 5.722540442765212e-09,
      "loss": 0.2975,
      "step": 28704
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7870820760726929,
      "learning_rate": 5.684833802865886e-09,
      "loss": 0.2575,
      "step": 28705
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.219719409942627,
      "learning_rate": 5.647251765844574e-09,
      "loss": 0.4203,
      "step": 28706
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.43550181388855,
      "learning_rate": 5.6097943321697936e-09,
      "loss": 0.3135,
      "step": 28707
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2153921127319336,
      "learning_rate": 5.572461502310056e-09,
      "loss": 0.3633,
      "step": 28708
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.44948148727417,
      "learning_rate": 5.535253276728325e-09,
      "loss": 0.2978,
      "step": 28709
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.234222888946533,
      "learning_rate": 5.498169655890895e-09,
      "loss": 0.3046,
      "step": 28710
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.3100783824920654,
      "learning_rate": 5.4612106402585075e-09,
      "loss": 0.3426,
      "step": 28711
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.733678936958313,
      "learning_rate": 5.424376230293016e-09,
      "loss": 0.2883,
      "step": 28712
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8025531768798828,
      "learning_rate": 5.387666426452942e-09,
      "loss": 0.3191,
      "step": 28713
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.506377935409546,
      "learning_rate": 5.351081229195698e-09,
      "loss": 0.1718,
      "step": 28714
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.758514165878296,
      "learning_rate": 5.314620638978696e-09,
      "loss": 0.2661,
      "step": 28715
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.1884572505950928,
      "learning_rate": 5.2782846562549065e-09,
      "loss": 0.2224,
      "step": 28716
    },
    {
      "epoch": 2.97,
      "grad_norm": 3.101815700531006,
      "learning_rate": 5.24207328147841e-09,
      "loss": 0.3939,
      "step": 28717
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.232071876525879,
      "learning_rate": 5.205986515101069e-09,
      "loss": 0.3101,
      "step": 28718
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.211226463317871,
      "learning_rate": 5.170024357571413e-09,
      "loss": 0.3542,
      "step": 28719
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6022642850875854,
      "learning_rate": 5.134186809337971e-09,
      "loss": 0.2365,
      "step": 28720
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6913799047470093,
      "learning_rate": 5.098473870849274e-09,
      "loss": 0.3099,
      "step": 28721
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2235727310180664,
      "learning_rate": 5.0628855425483016e-09,
      "loss": 0.3323,
      "step": 28722
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.014212131500244,
      "learning_rate": 5.027421824880252e-09,
      "loss": 0.3283,
      "step": 28723
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8214545249938965,
      "learning_rate": 4.992082718286995e-09,
      "loss": 0.2795,
      "step": 28724
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8191425800323486,
      "learning_rate": 4.956868223208178e-09,
      "loss": 0.2121,
      "step": 28725
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2783923149108887,
      "learning_rate": 4.92177834008456e-09,
      "loss": 0.4054,
      "step": 28726
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.852821946144104,
      "learning_rate": 4.886813069352459e-09,
      "loss": 0.3714,
      "step": 28727
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.942625880241394,
      "learning_rate": 4.851972411447081e-09,
      "loss": 0.3064,
      "step": 28728
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.056046485900879,
      "learning_rate": 4.817256366804746e-09,
      "loss": 0.3899,
      "step": 28729
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9865937232971191,
      "learning_rate": 4.782664935857329e-09,
      "loss": 0.3333,
      "step": 28730
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0114777088165283,
      "learning_rate": 4.7481981190355964e-09,
      "loss": 0.3258,
      "step": 28731
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2525126934051514,
      "learning_rate": 4.7138559167692055e-09,
      "loss": 0.3647,
      "step": 28732
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7545254230499268,
      "learning_rate": 4.679638329486702e-09,
      "loss": 0.3214,
      "step": 28733
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9621418714523315,
      "learning_rate": 4.6455453576155216e-09,
      "loss": 0.3346,
      "step": 28734
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0164709091186523,
      "learning_rate": 4.61157700157977e-09,
      "loss": 0.2882,
      "step": 28735
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.1405863761901855,
      "learning_rate": 4.577733261802442e-09,
      "loss": 0.3159,
      "step": 28736
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7804195880889893,
      "learning_rate": 4.544014138706532e-09,
      "loss": 0.2615,
      "step": 28737
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8487721681594849,
      "learning_rate": 4.5104196327128145e-09,
      "loss": 0.2979,
      "step": 28738
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.243760347366333,
      "learning_rate": 4.476949744238734e-09,
      "loss": 0.3523,
      "step": 28739
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0589489936828613,
      "learning_rate": 4.443604473701735e-09,
      "loss": 0.3378,
      "step": 28740
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.150454044342041,
      "learning_rate": 4.410383821519259e-09,
      "loss": 0.3121,
      "step": 28741
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2200164794921875,
      "learning_rate": 4.377287788103201e-09,
      "loss": 0.2436,
      "step": 28742
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.43607759475708,
      "learning_rate": 4.344316373867674e-09,
      "loss": 0.351,
      "step": 28743
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.3570008277893066,
      "learning_rate": 4.311469579223459e-09,
      "loss": 0.3311,
      "step": 28744
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.356330394744873,
      "learning_rate": 4.27874740457912e-09,
      "loss": 0.3595,
      "step": 28745
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9583606719970703,
      "learning_rate": 4.246149850344327e-09,
      "loss": 0.2942,
      "step": 28746
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0242297649383545,
      "learning_rate": 4.213676916923204e-09,
      "loss": 0.2341,
      "step": 28747
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9250560998916626,
      "learning_rate": 4.1813286047232005e-09,
      "loss": 0.2709,
      "step": 28748
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8291432857513428,
      "learning_rate": 4.149104914146218e-09,
      "loss": 0.2645,
      "step": 28749
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.678486704826355,
      "learning_rate": 4.117005845594157e-09,
      "loss": 0.2222,
      "step": 28750
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2103006839752197,
      "learning_rate": 4.085031399466699e-09,
      "loss": 0.3934,
      "step": 28751
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.3539414405822754,
      "learning_rate": 4.0531815761635225e-09,
      "loss": 0.2894,
      "step": 28752
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.4021806716918945,
      "learning_rate": 4.021456376080979e-09,
      "loss": 0.3842,
      "step": 28753
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9100643396377563,
      "learning_rate": 3.989855799614306e-09,
      "loss": 0.3512,
      "step": 28754
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9441317319869995,
      "learning_rate": 3.958379847158744e-09,
      "loss": 0.2826,
      "step": 28755
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9049675464630127,
      "learning_rate": 3.927028519105091e-09,
      "loss": 0.3483,
      "step": 28756
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6339058876037598,
      "learning_rate": 3.895801815845257e-09,
      "loss": 0.2941,
      "step": 28757
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.415499448776245,
      "learning_rate": 3.864699737768929e-09,
      "loss": 0.3199,
      "step": 28758
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.414910316467285,
      "learning_rate": 3.8337222852635745e-09,
      "loss": 0.3856,
      "step": 28759
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8102360963821411,
      "learning_rate": 3.802869458714442e-09,
      "loss": 0.2351,
      "step": 28760
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.3187193870544434,
      "learning_rate": 3.7721412585067785e-09,
      "loss": 0.3704,
      "step": 28761
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.2252633571624756,
      "learning_rate": 3.741537685023611e-09,
      "loss": 0.3792,
      "step": 28762
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.938558578491211,
      "learning_rate": 3.7110587386468553e-09,
      "loss": 0.3089,
      "step": 28763
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0527191162109375,
      "learning_rate": 3.6807044197573194e-09,
      "loss": 0.2954,
      "step": 28764
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.168945550918579,
      "learning_rate": 3.650474728731368e-09,
      "loss": 0.3766,
      "step": 28765
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9752728939056396,
      "learning_rate": 3.620369665946477e-09,
      "loss": 0.2415,
      "step": 28766
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.660233974456787,
      "learning_rate": 3.5903892317801226e-09,
      "loss": 0.3268,
      "step": 28767
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8071107864379883,
      "learning_rate": 3.560533426603119e-09,
      "loss": 0.2445,
      "step": 28768
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8783063888549805,
      "learning_rate": 3.530802250788501e-09,
      "loss": 0.3085,
      "step": 28769
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.158013343811035,
      "learning_rate": 3.5011957047081935e-09,
      "loss": 0.3468,
      "step": 28770
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.241356372833252,
      "learning_rate": 3.4717137887307905e-09,
      "loss": 0.3859,
      "step": 28771
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9497073888778687,
      "learning_rate": 3.442356503223776e-09,
      "loss": 0.2908,
      "step": 28772
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0285768508911133,
      "learning_rate": 3.413123848552413e-09,
      "loss": 0.3181,
      "step": 28773
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9428156614303589,
      "learning_rate": 3.3840158250819653e-09,
      "loss": 0.32,
      "step": 28774
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8160260915756226,
      "learning_rate": 3.3550324331743656e-09,
      "loss": 0.3012,
      "step": 28775
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0892670154571533,
      "learning_rate": 3.3261736731926565e-09,
      "loss": 0.352,
      "step": 28776
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9146101474761963,
      "learning_rate": 3.29743954549544e-09,
      "loss": 0.2352,
      "step": 28777
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8514509201049805,
      "learning_rate": 3.2688300504413184e-09,
      "loss": 0.3218,
      "step": 28778
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9383817911148071,
      "learning_rate": 3.240345188385563e-09,
      "loss": 0.3074,
      "step": 28779
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.2869982719421387,
      "learning_rate": 3.2119849596856657e-09,
      "loss": 0.291,
      "step": 28780
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1853725910186768,
      "learning_rate": 3.183749364694677e-09,
      "loss": 0.2905,
      "step": 28781
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.32955002784729,
      "learning_rate": 3.1556384037623177e-09,
      "loss": 0.2983,
      "step": 28782
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.2245802879333496,
      "learning_rate": 3.1276520772427487e-09,
      "loss": 0.2833,
      "step": 28783
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6316139698028564,
      "learning_rate": 3.0997903854812496e-09,
      "loss": 0.3068,
      "step": 28784
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1802496910095215,
      "learning_rate": 3.0720533288275402e-09,
      "loss": 0.3733,
      "step": 28785
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.2828052043914795,
      "learning_rate": 3.0444409076269e-09,
      "loss": 0.3118,
      "step": 28786
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8169472217559814,
      "learning_rate": 3.016953122223498e-09,
      "loss": 0.3135,
      "step": 28787
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9239542484283447,
      "learning_rate": 2.9895899729603937e-09,
      "loss": 0.34,
      "step": 28788
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.762891173362732,
      "learning_rate": 2.962351460177315e-09,
      "loss": 0.3252,
      "step": 28789
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0329291820526123,
      "learning_rate": 2.9352375842151005e-09,
      "loss": 0.2519,
      "step": 28790
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.2178642749786377,
      "learning_rate": 2.9082483454112574e-09,
      "loss": 0.2606,
      "step": 28791
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.118556499481201,
      "learning_rate": 2.8813837441032943e-09,
      "loss": 0.2209,
      "step": 28792
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.061717987060547,
      "learning_rate": 2.854643780625388e-09,
      "loss": 0.3631,
      "step": 28793
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.350126028060913,
      "learning_rate": 2.828028455310605e-09,
      "loss": 0.3455,
      "step": 28794
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8797544240951538,
      "learning_rate": 2.801537768490903e-09,
      "loss": 0.3009,
      "step": 28795
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0482661724090576,
      "learning_rate": 2.7751717204971275e-09,
      "loss": 0.2773,
      "step": 28796
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.3576765060424805,
      "learning_rate": 2.7489303116579047e-09,
      "loss": 0.3277,
      "step": 28797
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.948528528213501,
      "learning_rate": 2.7228135422996404e-09,
      "loss": 0.2156,
      "step": 28798
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.3169214725494385,
      "learning_rate": 2.6968214127476298e-09,
      "loss": 0.3201,
      "step": 28799
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.367725372314453,
      "learning_rate": 2.670953923328279e-09,
      "loss": 0.3321,
      "step": 28800
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.5723443031311035,
      "learning_rate": 2.645211074361331e-09,
      "loss": 0.402,
      "step": 28801
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.174924850463867,
      "learning_rate": 2.619592866168752e-09,
      "loss": 0.3045,
      "step": 28802
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9411672353744507,
      "learning_rate": 2.594099299071395e-09,
      "loss": 0.2439,
      "step": 28803
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.313433885574341,
      "learning_rate": 2.568730373384565e-09,
      "loss": 0.3493,
      "step": 28804
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.089104652404785,
      "learning_rate": 2.543486089426894e-09,
      "loss": 0.3298,
      "step": 28805
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0315158367156982,
      "learning_rate": 2.518366447511467e-09,
      "loss": 0.3099,
      "step": 28806
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8208377361297607,
      "learning_rate": 2.493371447951365e-09,
      "loss": 0.2323,
      "step": 28807
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.808553695678711,
      "learning_rate": 2.4685010910596717e-09,
      "loss": 0.3053,
      "step": 28808
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.997349739074707,
      "learning_rate": 2.4437553771450295e-09,
      "loss": 0.3838,
      "step": 28809
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1080245971679688,
      "learning_rate": 2.41913430651608e-09,
      "loss": 0.2782,
      "step": 28810
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.16237211227417,
      "learning_rate": 2.3946378794814653e-09,
      "loss": 0.3071,
      "step": 28811
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.958261489868164,
      "learning_rate": 2.370266096344276e-09,
      "loss": 0.2864,
      "step": 28812
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9186843633651733,
      "learning_rate": 2.346018957410934e-09,
      "loss": 0.2761,
      "step": 28813
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.462930202484131,
      "learning_rate": 2.321896462981199e-09,
      "loss": 0.3546,
      "step": 28814
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1393654346466064,
      "learning_rate": 2.2978986133570524e-09,
      "loss": 0.2814,
      "step": 28815
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8904049396514893,
      "learning_rate": 2.274025408837144e-09,
      "loss": 0.319,
      "step": 28816
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8833622932434082,
      "learning_rate": 2.2502768497201235e-09,
      "loss": 0.3617,
      "step": 28817
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.022759199142456,
      "learning_rate": 2.2266529363002e-09,
      "loss": 0.2889,
      "step": 28818
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9799911975860596,
      "learning_rate": 2.2031536688749134e-09,
      "loss": 0.2726,
      "step": 28819
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1285524368286133,
      "learning_rate": 2.179779047734032e-09,
      "loss": 0.2754,
      "step": 28820
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.4175522327423096,
      "learning_rate": 2.1565290731706544e-09,
      "loss": 0.2777,
      "step": 28821
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7204276323318481,
      "learning_rate": 2.1334037454745495e-09,
      "loss": 0.3162,
      "step": 28822
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1074883937835693,
      "learning_rate": 2.1104030649343744e-09,
      "loss": 0.2871,
      "step": 28823
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.24395489692688,
      "learning_rate": 2.0875270318354566e-09,
      "loss": 0.2698,
      "step": 28824
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.634441614151001,
      "learning_rate": 2.0647756464642345e-09,
      "loss": 0.3339,
      "step": 28825
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.030702590942383,
      "learning_rate": 2.042148909104924e-09,
      "loss": 0.3473,
      "step": 28826
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8615089654922485,
      "learning_rate": 2.019646820037302e-09,
      "loss": 0.3257,
      "step": 28827
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8402986526489258,
      "learning_rate": 1.997269379544475e-09,
      "loss": 0.2464,
      "step": 28828
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6975345611572266,
      "learning_rate": 1.9750165879039996e-09,
      "loss": 0.2492,
      "step": 28829
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.2528064250946045,
      "learning_rate": 1.9528884453934305e-09,
      "loss": 0.3165,
      "step": 28830
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.023153305053711,
      "learning_rate": 1.930884952289214e-09,
      "loss": 0.3188,
      "step": 28831
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.8966903686523438,
      "learning_rate": 1.909006108864464e-09,
      "loss": 0.3715,
      "step": 28832
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1512937545776367,
      "learning_rate": 1.8872519153934067e-09,
      "loss": 0.2981,
      "step": 28833
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.3041324615478516,
      "learning_rate": 1.8656223721469357e-09,
      "loss": 0.3954,
      "step": 28834
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8611440658569336,
      "learning_rate": 1.8441174793937256e-09,
      "loss": 0.255,
      "step": 28835
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.2543091773986816,
      "learning_rate": 1.82273723740245e-09,
      "loss": 0.2728,
      "step": 28836
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.3245160579681396,
      "learning_rate": 1.8014816464406726e-09,
      "loss": 0.2896,
      "step": 28837
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8200550079345703,
      "learning_rate": 1.7803507067715164e-09,
      "loss": 0.3319,
      "step": 28838
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0538089275360107,
      "learning_rate": 1.7593444186603248e-09,
      "loss": 0.3026,
      "step": 28839
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.059213638305664,
      "learning_rate": 1.7384627823680001e-09,
      "loss": 0.3272,
      "step": 28840
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9869221448898315,
      "learning_rate": 1.7177057981543344e-09,
      "loss": 0.3443,
      "step": 28841
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7456269264221191,
      "learning_rate": 1.6970734662791199e-09,
      "loss": 0.2676,
      "step": 28842
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1529526710510254,
      "learning_rate": 1.6765657869988183e-09,
      "loss": 0.3227,
      "step": 28843
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0522968769073486,
      "learning_rate": 1.656182760569891e-09,
      "loss": 0.2905,
      "step": 28844
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.215022087097168,
      "learning_rate": 1.6359243872465792e-09,
      "loss": 0.2854,
      "step": 28845
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9586237668991089,
      "learning_rate": 1.6157906672809032e-09,
      "loss": 0.2805,
      "step": 28846
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.294891119003296,
      "learning_rate": 1.5957816009226635e-09,
      "loss": 0.3143,
      "step": 28847
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8070846796035767,
      "learning_rate": 1.5758971884238805e-09,
      "loss": 0.2794,
      "step": 28848
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0096733570098877,
      "learning_rate": 1.5561374300299136e-09,
      "loss": 0.3259,
      "step": 28849
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.2595622539520264,
      "learning_rate": 1.536502325989453e-09,
      "loss": 0.2857,
      "step": 28850
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.993273377418518,
      "learning_rate": 1.5169918765456372e-09,
      "loss": 0.3139,
      "step": 28851
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9341223239898682,
      "learning_rate": 1.4976060819427152e-09,
      "loss": 0.2807,
      "step": 28852
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.6036040782928467,
      "learning_rate": 1.4783449424216057e-09,
      "loss": 0.3215,
      "step": 28853
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0129234790802,
      "learning_rate": 1.459208458223227e-09,
      "loss": 0.3229,
      "step": 28854
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.844585657119751,
      "learning_rate": 1.4401966295851666e-09,
      "loss": 0.3319,
      "step": 28855
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.094114303588867,
      "learning_rate": 1.4213094567450124e-09,
      "loss": 0.2967,
      "step": 28856
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9752534627914429,
      "learning_rate": 1.402546939938132e-09,
      "loss": 0.3524,
      "step": 28857
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.094264507293701,
      "learning_rate": 1.383909079398782e-09,
      "loss": 0.2968,
      "step": 28858
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7233753204345703,
      "learning_rate": 1.3653958753601093e-09,
      "loss": 0.3877,
      "step": 28859
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.1773288249969482,
      "learning_rate": 1.34700732805082e-09,
      "loss": 0.3534,
      "step": 28860
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.0014913082122803,
      "learning_rate": 1.3287434377007302e-09,
      "loss": 0.2953,
      "step": 28861
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8082923889160156,
      "learning_rate": 1.310604204538546e-09,
      "loss": 0.3075,
      "step": 28862
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.951133131980896,
      "learning_rate": 1.2925896287907525e-09,
      "loss": 0.3285,
      "step": 28863
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1710422039031982,
      "learning_rate": 1.2746997106793946e-09,
      "loss": 0.3899,
      "step": 28864
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5212711095809937,
      "learning_rate": 1.2569344504309577e-09,
      "loss": 0.273,
      "step": 28865
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.941396713256836,
      "learning_rate": 1.2392938482641559e-09,
      "loss": 0.2846,
      "step": 28866
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1141273975372314,
      "learning_rate": 1.2217779043999234e-09,
      "loss": 0.3503,
      "step": 28867
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.2511632442474365,
      "learning_rate": 1.2043866190580843e-09,
      "loss": 0.3071,
      "step": 28868
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.3925998210906982,
      "learning_rate": 1.187119992452912e-09,
      "loss": 0.2878,
      "step": 28869
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.956369161605835,
      "learning_rate": 1.1699780248008996e-09,
      "loss": 0.2689,
      "step": 28870
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7829861640930176,
      "learning_rate": 1.15296071631521e-09,
      "loss": 0.2932,
      "step": 28871
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9936439990997314,
      "learning_rate": 1.1360680672090062e-09,
      "loss": 0.3624,
      "step": 28872
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9247275590896606,
      "learning_rate": 1.1193000776932307e-09,
      "loss": 0.3092,
      "step": 28873
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.067157745361328,
      "learning_rate": 1.1026567479754947e-09,
      "loss": 0.3758,
      "step": 28874
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.855088472366333,
      "learning_rate": 1.08613807826341e-09,
      "loss": 0.2801,
      "step": 28875
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.929064154624939,
      "learning_rate": 1.0697440687634785e-09,
      "loss": 0.2859,
      "step": 28876
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.841336488723755,
      "learning_rate": 1.053474719679981e-09,
      "loss": 0.3482,
      "step": 28877
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.528557777404785,
      "learning_rate": 1.0373300312160884e-09,
      "loss": 0.3531,
      "step": 28878
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.416130304336548,
      "learning_rate": 1.0213100035727507e-09,
      "loss": 0.3543,
      "step": 28879
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.035841226577759,
      "learning_rate": 1.0054146369498086e-09,
      "loss": 0.3106,
      "step": 28880
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.916771411895752,
      "learning_rate": 9.896439315448814e-10,
      "loss": 0.3129,
      "step": 28881
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.176384925842285,
      "learning_rate": 9.73997887555589e-10,
      "loss": 0.2547,
      "step": 28882
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7130882740020752,
      "learning_rate": 9.5847650517622e-10,
      "loss": 0.262,
      "step": 28883
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.808711051940918,
      "learning_rate": 9.430797846010642e-10,
      "loss": 0.3615,
      "step": 28884
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9012455940246582,
      "learning_rate": 9.278077260210794e-10,
      "loss": 0.3231,
      "step": 28885
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.701722502708435,
      "learning_rate": 9.126603296272241e-10,
      "loss": 0.2772,
      "step": 28886
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7287856340408325,
      "learning_rate": 8.976375956071259e-10,
      "loss": 0.3142,
      "step": 28887
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.69850492477417,
      "learning_rate": 8.827395241506331e-10,
      "loss": 0.2749,
      "step": 28888
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9016534090042114,
      "learning_rate": 8.679661154420426e-10,
      "loss": 0.2418,
      "step": 28889
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.214153528213501,
      "learning_rate": 8.533173696645413e-10,
      "loss": 0.2397,
      "step": 28890
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7809422016143799,
      "learning_rate": 8.387932870024262e-10,
      "loss": 0.3049,
      "step": 28891
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.078594207763672,
      "learning_rate": 8.243938676366637e-10,
      "loss": 0.342,
      "step": 28892
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.405625104904175,
      "learning_rate": 8.101191117448892e-10,
      "loss": 0.2979,
      "step": 28893
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.074415683746338,
      "learning_rate": 7.959690195080694e-10,
      "loss": 0.3421,
      "step": 28894
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7718180418014526,
      "learning_rate": 7.819435910993989e-10,
      "loss": 0.3256,
      "step": 28895
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.3878371715545654,
      "learning_rate": 7.680428266965135e-10,
      "loss": 0.3518,
      "step": 28896
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.190516233444214,
      "learning_rate": 7.542667264703874e-10,
      "loss": 0.3948,
      "step": 28897
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.0812485218048096,
      "learning_rate": 7.406152905953257e-10,
      "loss": 0.2827,
      "step": 28898
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7023917436599731,
      "learning_rate": 7.270885192389721e-10,
      "loss": 0.2657,
      "step": 28899
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.051891326904297,
      "learning_rate": 7.136864125711906e-10,
      "loss": 0.2829,
      "step": 28900
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.142827272415161,
      "learning_rate": 7.004089707596251e-10,
      "loss": 0.3409,
      "step": 28901
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7862040996551514,
      "learning_rate": 6.872561939685885e-10,
      "loss": 0.3269,
      "step": 28902
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6908009052276611,
      "learning_rate": 6.742280823623937e-10,
      "loss": 0.2695,
      "step": 28903
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.376581907272339,
      "learning_rate": 6.613246361042435e-10,
      "loss": 0.3723,
      "step": 28904
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1210131645202637,
      "learning_rate": 6.485458553540103e-10,
      "loss": 0.3058,
      "step": 28905
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.893868088722229,
      "learning_rate": 6.358917402715658e-10,
      "loss": 0.3384,
      "step": 28906
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9348596334457397,
      "learning_rate": 6.233622910156723e-10,
      "loss": 0.2786,
      "step": 28907
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.077716588973999,
      "learning_rate": 6.109575077395402e-10,
      "loss": 0.295,
      "step": 28908
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.043124198913574,
      "learning_rate": 5.986773906008214e-10,
      "loss": 0.2786,
      "step": 28909
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.96877920627594,
      "learning_rate": 5.865219397516164e-10,
      "loss": 0.3397,
      "step": 28910
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.345773935317993,
      "learning_rate": 5.744911553440258e-10,
      "loss": 0.3141,
      "step": 28911
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.718548059463501,
      "learning_rate": 5.625850375257092e-10,
      "loss": 0.2931,
      "step": 28912
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.0556256771087646,
      "learning_rate": 5.508035864476569e-10,
      "loss": 0.2975,
      "step": 28913
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9760057926177979,
      "learning_rate": 5.391468022564184e-10,
      "loss": 0.2979,
      "step": 28914
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9546009302139282,
      "learning_rate": 5.276146850963226e-10,
      "loss": 0.3406,
      "step": 28915
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9601430892944336,
      "learning_rate": 5.162072351116987e-10,
      "loss": 0.3204,
      "step": 28916
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.801613211631775,
      "learning_rate": 5.049244524446551e-10,
      "loss": 0.3293,
      "step": 28917
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9241148233413696,
      "learning_rate": 4.937663372361901e-10,
      "loss": 0.2317,
      "step": 28918
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9684209823608398,
      "learning_rate": 4.827328896239713e-10,
      "loss": 0.2817,
      "step": 28919
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.455132484436035,
      "learning_rate": 4.718241097478871e-10,
      "loss": 0.3762,
      "step": 28920
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.510899782180786,
      "learning_rate": 4.6103999774227415e-10,
      "loss": 0.3421,
      "step": 28921
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7302155494689941,
      "learning_rate": 4.5038055374257984e-10,
      "loss": 0.2328,
      "step": 28922
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.973044991493225,
      "learning_rate": 4.3984577788092065e-10,
      "loss": 0.2645,
      "step": 28923
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.982157826423645,
      "learning_rate": 4.294356702883029e-10,
      "loss": 0.2713,
      "step": 28924
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.04097580909729,
      "learning_rate": 4.1915023109573295e-10,
      "loss": 0.3611,
      "step": 28925
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9501012563705444,
      "learning_rate": 4.0898946043088637e-10,
      "loss": 0.3636,
      "step": 28926
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.411789894104004,
      "learning_rate": 3.9895335842032866e-10,
      "loss": 0.2924,
      "step": 28927
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9052091836929321,
      "learning_rate": 3.890419251884048e-10,
      "loss": 0.3173,
      "step": 28928
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.6039488315582275,
      "learning_rate": 3.792551608605699e-10,
      "loss": 0.3906,
      "step": 28929
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.000084638595581,
      "learning_rate": 3.6959306555672814e-10,
      "loss": 0.316,
      "step": 28930
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.907240390777588,
      "learning_rate": 3.6005563939900403e-10,
      "loss": 0.3349,
      "step": 28931
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9461655616760254,
      "learning_rate": 3.506428825061914e-10,
      "loss": 0.2798,
      "step": 28932
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.100714921951294,
      "learning_rate": 3.413547949948637e-10,
      "loss": 0.2692,
      "step": 28933
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9549297094345093,
      "learning_rate": 3.321913769804841e-10,
      "loss": 0.3628,
      "step": 28934
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5441232919692993,
      "learning_rate": 3.231526285785158e-10,
      "loss": 0.3069,
      "step": 28935
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.997456431388855,
      "learning_rate": 3.1423854990109137e-10,
      "loss": 0.2663,
      "step": 28936
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9586764574050903,
      "learning_rate": 3.0544914105923306e-10,
      "loss": 0.2611,
      "step": 28937
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.390144109725952,
      "learning_rate": 2.9678440216174274e-10,
      "loss": 0.3179,
      "step": 28938
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8838605880737305,
      "learning_rate": 2.8824433331853254e-10,
      "loss": 0.333,
      "step": 28939
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.0738914012908936,
      "learning_rate": 2.798289346350735e-10,
      "loss": 0.3049,
      "step": 28940
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1273226737976074,
      "learning_rate": 2.715382062157268e-10,
      "loss": 0.2956,
      "step": 28941
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9661682844161987,
      "learning_rate": 2.6337214816485323e-10,
      "loss": 0.3551,
      "step": 28942
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9493184089660645,
      "learning_rate": 2.553307605834832e-10,
      "loss": 0.243,
      "step": 28943
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7541587352752686,
      "learning_rate": 2.4741404357153663e-10,
      "loss": 0.2787,
      "step": 28944
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9722756147384644,
      "learning_rate": 2.3962199722893374e-10,
      "loss": 0.3074,
      "step": 28945
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.4577925205230713,
      "learning_rate": 2.319546216522639e-10,
      "loss": 0.3178,
      "step": 28946
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.581118583679199,
      "learning_rate": 2.2441191693700626e-10,
      "loss": 0.3691,
      "step": 28947
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1707963943481445,
      "learning_rate": 2.1699388317752978e-10,
      "loss": 0.3544,
      "step": 28948
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8822072744369507,
      "learning_rate": 2.09700520465983e-10,
      "loss": 0.2886,
      "step": 28949
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.2186944484710693,
      "learning_rate": 2.0253182889340417e-10,
      "loss": 0.3255,
      "step": 28950
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6728392839431763,
      "learning_rate": 1.9548780854861117e-10,
      "loss": 0.2118,
      "step": 28951
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.038053035736084,
      "learning_rate": 1.885684595204218e-10,
      "loss": 0.2385,
      "step": 28952
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.2050344944000244,
      "learning_rate": 1.817737818954335e-10,
      "loss": 0.3221,
      "step": 28953
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.1701247692108154,
      "learning_rate": 1.7510377575691295e-10,
      "loss": 0.3417,
      "step": 28954
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9829379320144653,
      "learning_rate": 1.685584411881269e-10,
      "loss": 0.3422,
      "step": 28955
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.839501142501831,
      "learning_rate": 1.621377782712319e-10,
      "loss": 0.2793,
      "step": 28956
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.9536687135696411,
      "learning_rate": 1.5584178708727415e-10,
      "loss": 0.4133,
      "step": 28957
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.3696706295013428,
      "learning_rate": 1.4967046771285909e-10,
      "loss": 0.3149,
      "step": 28958
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.172330617904663,
      "learning_rate": 1.436238202257023e-10,
      "loss": 0.3245,
      "step": 28959
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.2678816318511963,
      "learning_rate": 1.377018447024092e-10,
      "loss": 0.3251,
      "step": 28960
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.249180793762207,
      "learning_rate": 1.3190454121403408e-10,
      "loss": 0.2791,
      "step": 28961
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8550670146942139,
      "learning_rate": 1.2623190983607203e-10,
      "loss": 0.308,
      "step": 28962
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8873056173324585,
      "learning_rate": 1.2068395063735694e-10,
      "loss": 0.3056,
      "step": 28963
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.17646861076355,
      "learning_rate": 1.1526066368672262e-10,
      "loss": 0.247,
      "step": 28964
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.3988940715789795,
      "learning_rate": 1.0996204905300289e-10,
      "loss": 0.2814,
      "step": 28965
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6715664863586426,
      "learning_rate": 1.0478810680170093e-10,
      "loss": 0.2477,
      "step": 28966
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.2108314037323,
      "learning_rate": 9.973883699720965e-11,
      "loss": 0.3811,
      "step": 28967
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.9435316324234009,
      "learning_rate": 9.481423970170156e-11,
      "loss": 0.2982,
      "step": 28968
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.118544101715088,
      "learning_rate": 9.001431497845936e-11,
      "loss": 0.2812,
      "step": 28969
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.3304903507232666,
      "learning_rate": 8.533906288632488e-11,
      "loss": 0.3822,
      "step": 28970
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.968979001045227,
      "learning_rate": 8.078848348413993e-11,
      "loss": 0.3347,
      "step": 28971
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.9468716382980347,
      "learning_rate": 7.636257682741566e-11,
      "loss": 0.2785,
      "step": 28972
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.1489386558532715,
      "learning_rate": 7.206134297166322e-11,
      "loss": 0.3132,
      "step": 28973
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.46958065032959,
      "learning_rate": 6.788478197128357e-11,
      "loss": 0.3094,
      "step": 28974
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.378878593444824,
      "learning_rate": 6.383289387734693e-11,
      "loss": 0.2442,
      "step": 28975
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.03186297416687,
      "learning_rate": 5.99056787420338e-11,
      "loss": 0.3149,
      "step": 28976
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.361887216567993,
      "learning_rate": 5.610313661197353e-11,
      "loss": 0.3246,
      "step": 28977
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.930734872817993,
      "learning_rate": 5.242526753712618e-11,
      "loss": 0.2662,
      "step": 28978
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7274881601333618,
      "learning_rate": 4.887207156079043e-11,
      "loss": 0.2511,
      "step": 28979
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7573628425598145,
      "learning_rate": 4.5443548729595666e-11,
      "loss": 0.3061,
      "step": 28980
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7325918674468994,
      "learning_rate": 4.213969908462012e-11,
      "loss": 0.202,
      "step": 28981
    },
    {
      "epoch": 3.0,
      "grad_norm": 3.853550672531128,
      "learning_rate": 3.896052266805228e-11,
      "loss": 0.3235,
      "step": 28982
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8212394714355469,
      "learning_rate": 3.5906019518749943e-11,
      "loss": 0.3437,
      "step": 28983
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.917029857635498,
      "learning_rate": 3.2976189675570926e-11,
      "loss": 0.2856,
      "step": 28984
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.0120627880096436,
      "learning_rate": 3.0171033174042355e-11,
      "loss": 0.3895,
      "step": 28985
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.705405592918396,
      "learning_rate": 2.7490550049691367e-11,
      "loss": 0.318,
      "step": 28986
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.0982139110565186,
      "learning_rate": 2.493474033582466e-11,
      "loss": 0.3077,
      "step": 28987
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.016167640686035,
      "learning_rate": 2.250360406574892e-11,
      "loss": 0.2443,
      "step": 28988
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.9557193517684937,
      "learning_rate": 2.019714126721972e-11,
      "loss": 0.2698,
      "step": 28989
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8901747465133667,
      "learning_rate": 1.8015351970213092e-11,
      "loss": 0.2781,
      "step": 28990
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7655307054519653,
      "learning_rate": 1.59582362024846e-11,
      "loss": 0.2996,
      "step": 28991
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.087355136871338,
      "learning_rate": 1.4025793989569381e-11,
      "loss": 0.2525,
      "step": 28992
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.796848177909851,
      "learning_rate": 1.2218025354782115e-11,
      "loss": 0.3053,
      "step": 28993
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.9571720361709595,
      "learning_rate": 1.0534930320327263e-11,
      "loss": 0.2462,
      "step": 28994
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7163604497909546,
      "learning_rate": 8.976508908409286e-12,
      "loss": 0.2941,
      "step": 28995
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.1664717197418213,
      "learning_rate": 7.542761137901978e-12,
      "loss": 0.3404,
      "step": 28996
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.2508738040924072,
      "learning_rate": 6.233687026568902e-12,
      "loss": 0.2946,
      "step": 28997
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8566919565200806,
      "learning_rate": 5.049286591063407e-12,
      "loss": 0.2825,
      "step": 28998
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5702537298202515,
      "learning_rate": 3.989559846928615e-12,
      "loss": 0.225,
      "step": 28999
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.735158085823059,
      "learning_rate": 3.054506805266755e-12,
      "loss": 0.2625,
      "step": 29000
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.334102153778076,
      "learning_rate": 2.244127479400504e-12,
      "loss": 0.3359,
      "step": 29001
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6875548362731934,
      "learning_rate": 1.5584218782116467e-12,
      "loss": 0.2893,
      "step": 29002
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8265315294265747,
      "learning_rate": 9.973900116921897e-13,
      "loss": 0.2544,
      "step": 29003
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.641510009765625,
      "learning_rate": 5.610318853932484e-13,
      "loss": 0.2556,
      "step": 29004
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.023818254470825,
      "learning_rate": 2.493475059761608e-13,
      "loss": 0.2769,
      "step": 29005
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8402198553085327,
      "learning_rate": 6.233687677159594e-14,
      "loss": 0.2478,
      "step": 29006
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5446624755859375,
      "learning_rate": 0.0,
      "loss": 0.2644,
      "step": 29007
    },
    {
      "epoch": 3.0,
      "step": 29007,
      "total_flos": 7.106036455115325e+17,
      "train_loss": 0.5416545932868169,
      "train_runtime": 18952.8185,
      "train_samples_per_second": 24.486,
      "train_steps_per_second": 1.53
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 29007,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 2400.0,
  "total_flos": 7.106036455115325e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}