{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 15630,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006397952655150352,
      "grad_norm": 40.859317264118296,
      "learning_rate": 4.264392324093817e-08,
      "loss": 2.4257,
      "step": 1
    },
    {
      "epoch": 0.0012795905310300703,
      "grad_norm": 42.50892178583353,
      "learning_rate": 8.528784648187634e-08,
      "loss": 2.5478,
      "step": 2
    },
    {
      "epoch": 0.0019193857965451055,
      "grad_norm": 39.02853470198785,
      "learning_rate": 1.279317697228145e-07,
      "loss": 2.47,
      "step": 3
    },
    {
      "epoch": 0.0025591810620601407,
      "grad_norm": 37.40191830646434,
      "learning_rate": 1.7057569296375268e-07,
      "loss": 2.5176,
      "step": 4
    },
    {
      "epoch": 0.003198976327575176,
      "grad_norm": 38.9959456862156,
      "learning_rate": 2.1321961620469084e-07,
      "loss": 2.5593,
      "step": 5
    },
    {
      "epoch": 0.003838771593090211,
      "grad_norm": 48.436746075458736,
      "learning_rate": 2.55863539445629e-07,
      "loss": 2.5419,
      "step": 6
    },
    {
      "epoch": 0.004478566858605247,
      "grad_norm": 37.33067509975441,
      "learning_rate": 2.9850746268656716e-07,
      "loss": 2.4747,
      "step": 7
    },
    {
      "epoch": 0.005118362124120281,
      "grad_norm": 56.328863102311296,
      "learning_rate": 3.4115138592750537e-07,
      "loss": 2.4182,
      "step": 8
    },
    {
      "epoch": 0.005758157389635317,
      "grad_norm": 40.89457775045127,
      "learning_rate": 3.837953091684436e-07,
      "loss": 2.4584,
      "step": 9
    },
    {
      "epoch": 0.006397952655150352,
      "grad_norm": 39.19032868893243,
      "learning_rate": 4.264392324093817e-07,
      "loss": 2.4932,
      "step": 10
    },
    {
      "epoch": 0.007037747920665387,
      "grad_norm": 34.710660478901495,
      "learning_rate": 4.690831556503199e-07,
      "loss": 2.5102,
      "step": 11
    },
    {
      "epoch": 0.007677543186180422,
      "grad_norm": 36.20874551128639,
      "learning_rate": 5.11727078891258e-07,
      "loss": 2.4694,
      "step": 12
    },
    {
      "epoch": 0.008317338451695458,
      "grad_norm": 39.13593207882229,
      "learning_rate": 5.543710021321962e-07,
      "loss": 2.4309,
      "step": 13
    },
    {
      "epoch": 0.008957133717210493,
      "grad_norm": 32.76590356061493,
      "learning_rate": 5.970149253731343e-07,
      "loss": 2.3756,
      "step": 14
    },
    {
      "epoch": 0.009596928982725527,
      "grad_norm": 62.770605636952695,
      "learning_rate": 6.396588486140725e-07,
      "loss": 2.32,
      "step": 15
    },
    {
      "epoch": 0.010236724248240563,
      "grad_norm": 25.472942299138555,
      "learning_rate": 6.823027718550107e-07,
      "loss": 2.2658,
      "step": 16
    },
    {
      "epoch": 0.010876519513755598,
      "grad_norm": 32.43314510722181,
      "learning_rate": 7.249466950959488e-07,
      "loss": 2.2928,
      "step": 17
    },
    {
      "epoch": 0.011516314779270634,
      "grad_norm": 39.80721678756784,
      "learning_rate": 7.675906183368872e-07,
      "loss": 2.2604,
      "step": 18
    },
    {
      "epoch": 0.01215611004478567,
      "grad_norm": 23.080609854294902,
      "learning_rate": 8.102345415778253e-07,
      "loss": 2.1888,
      "step": 19
    },
    {
      "epoch": 0.012795905310300703,
      "grad_norm": 24.25746931875127,
      "learning_rate": 8.528784648187634e-07,
      "loss": 2.1782,
      "step": 20
    },
    {
      "epoch": 0.013435700575815739,
      "grad_norm": 21.150700200587277,
      "learning_rate": 8.955223880597015e-07,
      "loss": 2.065,
      "step": 21
    },
    {
      "epoch": 0.014075495841330775,
      "grad_norm": 10.828815409590216,
      "learning_rate": 9.381663113006398e-07,
      "loss": 1.8762,
      "step": 22
    },
    {
      "epoch": 0.01471529110684581,
      "grad_norm": 12.54077014268372,
      "learning_rate": 9.808102345415779e-07,
      "loss": 1.9144,
      "step": 23
    },
    {
      "epoch": 0.015355086372360844,
      "grad_norm": 12.698839419599983,
      "learning_rate": 1.023454157782516e-06,
      "loss": 1.8904,
      "step": 24
    },
    {
      "epoch": 0.01599488163787588,
      "grad_norm": 12.849704027273102,
      "learning_rate": 1.066098081023454e-06,
      "loss": 1.9184,
      "step": 25
    },
    {
      "epoch": 0.016634676903390915,
      "grad_norm": 11.42476137157715,
      "learning_rate": 1.1087420042643924e-06,
      "loss": 1.9103,
      "step": 26
    },
    {
      "epoch": 0.01727447216890595,
      "grad_norm": 12.25597108342898,
      "learning_rate": 1.1513859275053305e-06,
      "loss": 1.8643,
      "step": 27
    },
    {
      "epoch": 0.017914267434420986,
      "grad_norm": 10.535412953496241,
      "learning_rate": 1.1940298507462686e-06,
      "loss": 1.816,
      "step": 28
    },
    {
      "epoch": 0.018554062699936022,
      "grad_norm": 8.850974856212073,
      "learning_rate": 1.236673773987207e-06,
      "loss": 1.7281,
      "step": 29
    },
    {
      "epoch": 0.019193857965451054,
      "grad_norm": 8.454591315067583,
      "learning_rate": 1.279317697228145e-06,
      "loss": 1.7593,
      "step": 30
    },
    {
      "epoch": 0.01983365323096609,
      "grad_norm": 6.983205380865425,
      "learning_rate": 1.3219616204690834e-06,
      "loss": 1.7848,
      "step": 31
    },
    {
      "epoch": 0.020473448496481125,
      "grad_norm": 7.043161488643078,
      "learning_rate": 1.3646055437100215e-06,
      "loss": 1.7635,
      "step": 32
    },
    {
      "epoch": 0.02111324376199616,
      "grad_norm": 6.942551734841916,
      "learning_rate": 1.4072494669509596e-06,
      "loss": 1.7234,
      "step": 33
    },
    {
      "epoch": 0.021753039027511197,
      "grad_norm": 7.297431836866321,
      "learning_rate": 1.4498933901918977e-06,
      "loss": 1.6397,
      "step": 34
    },
    {
      "epoch": 0.022392834293026232,
      "grad_norm": 5.976831446617759,
      "learning_rate": 1.4925373134328358e-06,
      "loss": 1.6435,
      "step": 35
    },
    {
      "epoch": 0.023032629558541268,
      "grad_norm": 6.331699742281076,
      "learning_rate": 1.5351812366737743e-06,
      "loss": 1.5886,
      "step": 36
    },
    {
      "epoch": 0.023672424824056303,
      "grad_norm": 5.25043106227297,
      "learning_rate": 1.5778251599147124e-06,
      "loss": 1.622,
      "step": 37
    },
    {
      "epoch": 0.02431222008957134,
      "grad_norm": 4.82676804474882,
      "learning_rate": 1.6204690831556505e-06,
      "loss": 1.6769,
      "step": 38
    },
    {
      "epoch": 0.02495201535508637,
      "grad_norm": 5.9340522050789595,
      "learning_rate": 1.6631130063965886e-06,
      "loss": 1.6507,
      "step": 39
    },
    {
      "epoch": 0.025591810620601407,
      "grad_norm": 5.508389528370756,
      "learning_rate": 1.7057569296375267e-06,
      "loss": 1.5854,
      "step": 40
    },
    {
      "epoch": 0.026231605886116442,
      "grad_norm": 5.804138787551976,
      "learning_rate": 1.7484008528784648e-06,
      "loss": 1.6414,
      "step": 41
    },
    {
      "epoch": 0.026871401151631478,
      "grad_norm": 6.204651305652198,
      "learning_rate": 1.791044776119403e-06,
      "loss": 1.4998,
      "step": 42
    },
    {
      "epoch": 0.027511196417146513,
      "grad_norm": 6.701367948460813,
      "learning_rate": 1.8336886993603415e-06,
      "loss": 1.5494,
      "step": 43
    },
    {
      "epoch": 0.02815099168266155,
      "grad_norm": 5.993505573681679,
      "learning_rate": 1.8763326226012796e-06,
      "loss": 1.5414,
      "step": 44
    },
    {
      "epoch": 0.028790786948176585,
      "grad_norm": 7.85182574128409,
      "learning_rate": 1.9189765458422177e-06,
      "loss": 1.5754,
      "step": 45
    },
    {
      "epoch": 0.02943058221369162,
      "grad_norm": 6.292543633815138,
      "learning_rate": 1.9616204690831558e-06,
      "loss": 1.613,
      "step": 46
    },
    {
      "epoch": 0.030070377479206652,
      "grad_norm": 5.752002464586006,
      "learning_rate": 2.004264392324094e-06,
      "loss": 1.4954,
      "step": 47
    },
    {
      "epoch": 0.030710172744721688,
      "grad_norm": 4.797802438809174,
      "learning_rate": 2.046908315565032e-06,
      "loss": 1.4421,
      "step": 48
    },
    {
      "epoch": 0.03134996801023673,
      "grad_norm": 4.136092895766244,
      "learning_rate": 2.08955223880597e-06,
      "loss": 1.4778,
      "step": 49
    },
    {
      "epoch": 0.03198976327575176,
      "grad_norm": 4.11280366546932,
      "learning_rate": 2.132196162046908e-06,
      "loss": 1.5176,
      "step": 50
    },
    {
      "epoch": 0.03262955854126679,
      "grad_norm": 4.064270956831667,
      "learning_rate": 2.1748400852878467e-06,
      "loss": 1.4789,
      "step": 51
    },
    {
      "epoch": 0.03326935380678183,
      "grad_norm": 4.323591851351226,
      "learning_rate": 2.217484008528785e-06,
      "loss": 1.5376,
      "step": 52
    },
    {
      "epoch": 0.03390914907229686,
      "grad_norm": 4.889006778815846,
      "learning_rate": 2.260127931769723e-06,
      "loss": 1.4974,
      "step": 53
    },
    {
      "epoch": 0.0345489443378119,
      "grad_norm": 4.8579810416853455,
      "learning_rate": 2.302771855010661e-06,
      "loss": 1.4455,
      "step": 54
    },
    {
      "epoch": 0.035188739603326934,
      "grad_norm": 5.538376018864462,
      "learning_rate": 2.345415778251599e-06,
      "loss": 1.4483,
      "step": 55
    },
    {
      "epoch": 0.03582853486884197,
      "grad_norm": 4.370635872248707,
      "learning_rate": 2.3880597014925373e-06,
      "loss": 1.4081,
      "step": 56
    },
    {
      "epoch": 0.036468330134357005,
      "grad_norm": 3.7045940067188066,
      "learning_rate": 2.4307036247334758e-06,
      "loss": 1.466,
      "step": 57
    },
    {
      "epoch": 0.037108125399872044,
      "grad_norm": 3.809719138775924,
      "learning_rate": 2.473347547974414e-06,
      "loss": 1.4399,
      "step": 58
    },
    {
      "epoch": 0.037747920665387076,
      "grad_norm": 3.7800457689296927,
      "learning_rate": 2.515991471215352e-06,
      "loss": 1.4572,
      "step": 59
    },
    {
      "epoch": 0.03838771593090211,
      "grad_norm": 3.224578358032259,
      "learning_rate": 2.55863539445629e-06,
      "loss": 1.3886,
      "step": 60
    },
    {
      "epoch": 0.03902751119641715,
      "grad_norm": 3.5660215215751734,
      "learning_rate": 2.601279317697228e-06,
      "loss": 1.3981,
      "step": 61
    },
    {
      "epoch": 0.03966730646193218,
      "grad_norm": 3.050144728271045,
      "learning_rate": 2.6439232409381667e-06,
      "loss": 1.4289,
      "step": 62
    },
    {
      "epoch": 0.04030710172744722,
      "grad_norm": 3.7368014444028934,
      "learning_rate": 2.686567164179105e-06,
      "loss": 1.4151,
      "step": 63
    },
    {
      "epoch": 0.04094689699296225,
      "grad_norm": 3.2081061681594303,
      "learning_rate": 2.729211087420043e-06,
      "loss": 1.41,
      "step": 64
    },
    {
      "epoch": 0.04158669225847729,
      "grad_norm": 3.2468913299074806,
      "learning_rate": 2.771855010660981e-06,
      "loss": 1.3994,
      "step": 65
    },
    {
      "epoch": 0.04222648752399232,
      "grad_norm": 3.5533385877808157,
      "learning_rate": 2.814498933901919e-06,
      "loss": 1.403,
      "step": 66
    },
    {
      "epoch": 0.04286628278950736,
      "grad_norm": 3.3796646301411277,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 1.4132,
      "step": 67
    },
    {
      "epoch": 0.04350607805502239,
      "grad_norm": 3.464484559559236,
      "learning_rate": 2.8997867803837954e-06,
      "loss": 1.3693,
      "step": 68
    },
    {
      "epoch": 0.044145873320537425,
      "grad_norm": 3.2466604419235012,
      "learning_rate": 2.9424307036247335e-06,
      "loss": 1.3122,
      "step": 69
    },
    {
      "epoch": 0.044785668586052464,
      "grad_norm": 3.018472556196756,
      "learning_rate": 2.9850746268656716e-06,
      "loss": 1.3696,
      "step": 70
    },
    {
      "epoch": 0.045425463851567496,
      "grad_norm": 3.106571825972309,
      "learning_rate": 3.0277185501066105e-06,
      "loss": 1.3904,
      "step": 71
    },
    {
      "epoch": 0.046065259117082535,
      "grad_norm": 3.0636524447931555,
      "learning_rate": 3.0703624733475486e-06,
      "loss": 1.3807,
      "step": 72
    },
    {
      "epoch": 0.04670505438259757,
      "grad_norm": 3.1851662277819175,
      "learning_rate": 3.1130063965884867e-06,
      "loss": 1.3652,
      "step": 73
    },
    {
      "epoch": 0.04734484964811261,
      "grad_norm": 3.587764967839034,
      "learning_rate": 3.155650319829425e-06,
      "loss": 1.3173,
      "step": 74
    },
    {
      "epoch": 0.04798464491362764,
      "grad_norm": 3.064134645611636,
      "learning_rate": 3.198294243070363e-06,
      "loss": 1.3368,
      "step": 75
    },
    {
      "epoch": 0.04862444017914268,
      "grad_norm": 3.0325331886877183,
      "learning_rate": 3.240938166311301e-06,
      "loss": 1.3423,
      "step": 76
    },
    {
      "epoch": 0.04926423544465771,
      "grad_norm": 3.035278137690572,
      "learning_rate": 3.283582089552239e-06,
      "loss": 1.3568,
      "step": 77
    },
    {
      "epoch": 0.04990403071017274,
      "grad_norm": 3.0601181586666617,
      "learning_rate": 3.3262260127931773e-06,
      "loss": 1.2949,
      "step": 78
    },
    {
      "epoch": 0.05054382597568778,
      "grad_norm": 3.360422884064213,
      "learning_rate": 3.3688699360341154e-06,
      "loss": 1.3132,
      "step": 79
    },
    {
      "epoch": 0.05118362124120281,
      "grad_norm": 3.4171545761919333,
      "learning_rate": 3.4115138592750535e-06,
      "loss": 1.3304,
      "step": 80
    },
    {
      "epoch": 0.05182341650671785,
      "grad_norm": 3.262104321269168,
      "learning_rate": 3.4541577825159916e-06,
      "loss": 1.4154,
      "step": 81
    },
    {
      "epoch": 0.052463211772232884,
      "grad_norm": 2.896267641966515,
      "learning_rate": 3.4968017057569297e-06,
      "loss": 1.2991,
      "step": 82
    },
    {
      "epoch": 0.053103007037747924,
      "grad_norm": 3.1668456353700356,
      "learning_rate": 3.5394456289978678e-06,
      "loss": 1.3494,
      "step": 83
    },
    {
      "epoch": 0.053742802303262956,
      "grad_norm": 3.104379779591339,
      "learning_rate": 3.582089552238806e-06,
      "loss": 1.3057,
      "step": 84
    },
    {
      "epoch": 0.05438259756877799,
      "grad_norm": 3.05531797388406,
      "learning_rate": 3.624733475479744e-06,
      "loss": 1.385,
      "step": 85
    },
    {
      "epoch": 0.05502239283429303,
      "grad_norm": 3.4149693850136646,
      "learning_rate": 3.667377398720683e-06,
      "loss": 1.2706,
      "step": 86
    },
    {
      "epoch": 0.05566218809980806,
      "grad_norm": 3.094867222297054,
      "learning_rate": 3.710021321961621e-06,
      "loss": 1.3472,
      "step": 87
    },
    {
      "epoch": 0.0563019833653231,
      "grad_norm": 2.8331399505321957,
      "learning_rate": 3.752665245202559e-06,
      "loss": 1.3188,
      "step": 88
    },
    {
      "epoch": 0.05694177863083813,
      "grad_norm": 2.6710922241492314,
      "learning_rate": 3.7953091684434973e-06,
      "loss": 1.3024,
      "step": 89
    },
    {
      "epoch": 0.05758157389635317,
      "grad_norm": 3.336683460034522,
      "learning_rate": 3.837953091684435e-06,
      "loss": 1.2542,
      "step": 90
    },
    {
      "epoch": 0.0582213691618682,
      "grad_norm": 3.1008114785080005,
      "learning_rate": 3.8805970149253735e-06,
      "loss": 1.3563,
      "step": 91
    },
    {
      "epoch": 0.05886116442738324,
      "grad_norm": 3.2382680283307304,
      "learning_rate": 3.9232409381663116e-06,
      "loss": 1.316,
      "step": 92
    },
    {
      "epoch": 0.05950095969289827,
      "grad_norm": 2.852421878801884,
      "learning_rate": 3.96588486140725e-06,
      "loss": 1.2991,
      "step": 93
    },
    {
      "epoch": 0.060140754958413305,
      "grad_norm": 3.1158963295644035,
      "learning_rate": 4.008528784648188e-06,
      "loss": 1.2586,
      "step": 94
    },
    {
      "epoch": 0.060780550223928344,
      "grad_norm": 2.988610914654297,
      "learning_rate": 4.051172707889126e-06,
      "loss": 1.2568,
      "step": 95
    },
    {
      "epoch": 0.061420345489443376,
      "grad_norm": 3.235006727821091,
      "learning_rate": 4.093816631130064e-06,
      "loss": 1.2169,
      "step": 96
    },
    {
      "epoch": 0.062060140754958415,
      "grad_norm": 2.9522660060444292,
      "learning_rate": 4.136460554371002e-06,
      "loss": 1.2727,
      "step": 97
    },
    {
      "epoch": 0.06269993602047345,
      "grad_norm": 3.1993383312680734,
      "learning_rate": 4.17910447761194e-06,
      "loss": 1.2218,
      "step": 98
    },
    {
      "epoch": 0.06333973128598848,
      "grad_norm": 3.1052773915577436,
      "learning_rate": 4.221748400852878e-06,
      "loss": 1.2564,
      "step": 99
    },
    {
      "epoch": 0.06397952655150352,
      "grad_norm": 3.0731839712066678,
      "learning_rate": 4.264392324093816e-06,
      "loss": 1.3506,
      "step": 100
    },
    {
      "epoch": 0.06461932181701856,
      "grad_norm": 2.8944583285644203,
      "learning_rate": 4.3070362473347545e-06,
      "loss": 1.2575,
      "step": 101
    },
    {
      "epoch": 0.06525911708253358,
      "grad_norm": 2.925775638033135,
      "learning_rate": 4.3496801705756935e-06,
      "loss": 1.2709,
      "step": 102
    },
    {
      "epoch": 0.06589891234804862,
      "grad_norm": 3.128038271739566,
      "learning_rate": 4.3923240938166316e-06,
      "loss": 1.2508,
      "step": 103
    },
    {
      "epoch": 0.06653870761356366,
      "grad_norm": 2.962059174175973,
      "learning_rate": 4.43496801705757e-06,
      "loss": 1.259,
      "step": 104
    },
    {
      "epoch": 0.0671785028790787,
      "grad_norm": 3.405781430969994,
      "learning_rate": 4.477611940298508e-06,
      "loss": 1.2242,
      "step": 105
    },
    {
      "epoch": 0.06781829814459372,
      "grad_norm": 2.841638000474648,
      "learning_rate": 4.520255863539446e-06,
      "loss": 1.304,
      "step": 106
    },
    {
      "epoch": 0.06845809341010876,
      "grad_norm": 2.9197587787542876,
      "learning_rate": 4.562899786780384e-06,
      "loss": 1.301,
      "step": 107
    },
    {
      "epoch": 0.0690978886756238,
      "grad_norm": 2.9206829216793606,
      "learning_rate": 4.605543710021322e-06,
      "loss": 1.2605,
      "step": 108
    },
    {
      "epoch": 0.06973768394113884,
      "grad_norm": 3.341174311112818,
      "learning_rate": 4.64818763326226e-06,
      "loss": 1.2401,
      "step": 109
    },
    {
      "epoch": 0.07037747920665387,
      "grad_norm": 3.354080028670265,
      "learning_rate": 4.690831556503198e-06,
      "loss": 1.2438,
      "step": 110
    },
    {
      "epoch": 0.0710172744721689,
      "grad_norm": 3.1976267760360724,
      "learning_rate": 4.733475479744136e-06,
      "loss": 1.298,
      "step": 111
    },
    {
      "epoch": 0.07165706973768395,
      "grad_norm": 3.185280482714834,
      "learning_rate": 4.7761194029850745e-06,
      "loss": 1.2151,
      "step": 112
    },
    {
      "epoch": 0.07229686500319897,
      "grad_norm": 4.568419447721809,
      "learning_rate": 4.8187633262260135e-06,
      "loss": 1.2389,
      "step": 113
    },
    {
      "epoch": 0.07293666026871401,
      "grad_norm": 2.903078279828431,
      "learning_rate": 4.8614072494669516e-06,
      "loss": 1.2214,
      "step": 114
    },
    {
      "epoch": 0.07357645553422905,
      "grad_norm": 3.1056687980579145,
      "learning_rate": 4.90405117270789e-06,
      "loss": 1.2546,
      "step": 115
    },
    {
      "epoch": 0.07421625079974409,
      "grad_norm": 3.313741190226211,
      "learning_rate": 4.946695095948828e-06,
      "loss": 1.21,
      "step": 116
    },
    {
      "epoch": 0.07485604606525911,
      "grad_norm": 3.345870889911724,
      "learning_rate": 4.989339019189766e-06,
      "loss": 1.2467,
      "step": 117
    },
    {
      "epoch": 0.07549584133077415,
      "grad_norm": 2.969987490120055,
      "learning_rate": 5.031982942430704e-06,
      "loss": 1.2192,
      "step": 118
    },
    {
      "epoch": 0.07613563659628919,
      "grad_norm": 3.583205000348064,
      "learning_rate": 5.074626865671642e-06,
      "loss": 1.2503,
      "step": 119
    },
    {
      "epoch": 0.07677543186180422,
      "grad_norm": 3.3609998452034677,
      "learning_rate": 5.11727078891258e-06,
      "loss": 1.26,
      "step": 120
    },
    {
      "epoch": 0.07741522712731926,
      "grad_norm": 3.393224163626569,
      "learning_rate": 5.159914712153518e-06,
      "loss": 1.2385,
      "step": 121
    },
    {
      "epoch": 0.0780550223928343,
      "grad_norm": 3.186851611894285,
      "learning_rate": 5.202558635394456e-06,
      "loss": 1.2552,
      "step": 122
    },
    {
      "epoch": 0.07869481765834933,
      "grad_norm": 3.1570602120835067,
      "learning_rate": 5.245202558635395e-06,
      "loss": 1.2206,
      "step": 123
    },
    {
      "epoch": 0.07933461292386436,
      "grad_norm": 2.900123662161653,
      "learning_rate": 5.2878464818763335e-06,
      "loss": 1.2339,
      "step": 124
    },
    {
      "epoch": 0.0799744081893794,
      "grad_norm": 3.4460707377503668,
      "learning_rate": 5.3304904051172716e-06,
      "loss": 1.2455,
      "step": 125
    },
    {
      "epoch": 0.08061420345489444,
      "grad_norm": 3.246991982630002,
      "learning_rate": 5.37313432835821e-06,
      "loss": 1.221,
      "step": 126
    },
    {
      "epoch": 0.08125399872040946,
      "grad_norm": 2.844650130149404,
      "learning_rate": 5.415778251599148e-06,
      "loss": 1.2803,
      "step": 127
    },
    {
      "epoch": 0.0818937939859245,
      "grad_norm": 3.2109226881974324,
      "learning_rate": 5.458422174840086e-06,
      "loss": 1.205,
      "step": 128
    },
    {
      "epoch": 0.08253358925143954,
      "grad_norm": 3.2852888862685843,
      "learning_rate": 5.501066098081024e-06,
      "loss": 1.1866,
      "step": 129
    },
    {
      "epoch": 0.08317338451695458,
      "grad_norm": 2.8390698341227028,
      "learning_rate": 5.543710021321962e-06,
      "loss": 1.2382,
      "step": 130
    },
    {
      "epoch": 0.0838131797824696,
      "grad_norm": 3.16012772907778,
      "learning_rate": 5.5863539445629e-06,
      "loss": 1.2394,
      "step": 131
    },
    {
      "epoch": 0.08445297504798464,
      "grad_norm": 2.959829160027203,
      "learning_rate": 5.628997867803838e-06,
      "loss": 1.2727,
      "step": 132
    },
    {
      "epoch": 0.08509277031349968,
      "grad_norm": 3.295227360767422,
      "learning_rate": 5.671641791044776e-06,
      "loss": 1.2299,
      "step": 133
    },
    {
      "epoch": 0.08573256557901472,
      "grad_norm": 3.2032214681157347,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 1.2411,
      "step": 134
    },
    {
      "epoch": 0.08637236084452975,
      "grad_norm": 3.101207106659726,
      "learning_rate": 5.756929637526653e-06,
      "loss": 1.148,
      "step": 135
    },
    {
      "epoch": 0.08701215611004479,
      "grad_norm": 2.942112378657255,
      "learning_rate": 5.799573560767591e-06,
      "loss": 1.2191,
      "step": 136
    },
    {
      "epoch": 0.08765195137555983,
      "grad_norm": 3.4551339187197723,
      "learning_rate": 5.842217484008529e-06,
      "loss": 1.1899,
      "step": 137
    },
    {
      "epoch": 0.08829174664107485,
      "grad_norm": 3.3816928384532052,
      "learning_rate": 5.884861407249467e-06,
      "loss": 1.2408,
      "step": 138
    },
    {
      "epoch": 0.08893154190658989,
      "grad_norm": 3.2395061660856412,
      "learning_rate": 5.927505330490405e-06,
      "loss": 1.2672,
      "step": 139
    },
    {
      "epoch": 0.08957133717210493,
      "grad_norm": 2.9743484576257333,
      "learning_rate": 5.970149253731343e-06,
      "loss": 1.2107,
      "step": 140
    },
    {
      "epoch": 0.09021113243761997,
      "grad_norm": 3.769278321969258,
      "learning_rate": 6.012793176972282e-06,
      "loss": 1.2634,
      "step": 141
    },
    {
      "epoch": 0.09085092770313499,
      "grad_norm": 3.6526708147930105,
      "learning_rate": 6.055437100213221e-06,
      "loss": 1.2166,
      "step": 142
    },
    {
      "epoch": 0.09149072296865003,
      "grad_norm": 3.226345674563243,
      "learning_rate": 6.098081023454159e-06,
      "loss": 1.251,
      "step": 143
    },
    {
      "epoch": 0.09213051823416507,
      "grad_norm": 3.3730519452212606,
      "learning_rate": 6.140724946695097e-06,
      "loss": 1.2602,
      "step": 144
    },
    {
      "epoch": 0.0927703134996801,
      "grad_norm": 3.1558003190628763,
      "learning_rate": 6.183368869936035e-06,
      "loss": 1.2123,
      "step": 145
    },
    {
      "epoch": 0.09341010876519514,
      "grad_norm": 3.2939092183219505,
      "learning_rate": 6.2260127931769735e-06,
      "loss": 1.1476,
      "step": 146
    },
    {
      "epoch": 0.09404990403071017,
      "grad_norm": 3.312805154145766,
      "learning_rate": 6.2686567164179116e-06,
      "loss": 1.1386,
      "step": 147
    },
    {
      "epoch": 0.09468969929622521,
      "grad_norm": 3.161731866738656,
      "learning_rate": 6.31130063965885e-06,
      "loss": 1.2594,
      "step": 148
    },
    {
      "epoch": 0.09532949456174024,
      "grad_norm": 3.219216329077004,
      "learning_rate": 6.353944562899788e-06,
      "loss": 1.1407,
      "step": 149
    },
    {
      "epoch": 0.09596928982725528,
      "grad_norm": 3.0786213568418472,
      "learning_rate": 6.396588486140726e-06,
      "loss": 1.2305,
      "step": 150
    },
    {
      "epoch": 0.09660908509277032,
      "grad_norm": 3.2722093049370526,
      "learning_rate": 6.439232409381664e-06,
      "loss": 1.2104,
      "step": 151
    },
    {
      "epoch": 0.09724888035828536,
      "grad_norm": 3.2305273403378574,
      "learning_rate": 6.481876332622602e-06,
      "loss": 1.2037,
      "step": 152
    },
    {
      "epoch": 0.09788867562380038,
      "grad_norm": 3.350538923041618,
      "learning_rate": 6.52452025586354e-06,
      "loss": 1.2253,
      "step": 153
    },
    {
      "epoch": 0.09852847088931542,
      "grad_norm": 3.5603576411716937,
      "learning_rate": 6.567164179104478e-06,
      "loss": 1.1287,
      "step": 154
    },
    {
      "epoch": 0.09916826615483046,
      "grad_norm": 3.318315943706194,
      "learning_rate": 6.609808102345416e-06,
      "loss": 1.2048,
      "step": 155
    },
    {
      "epoch": 0.09980806142034548,
      "grad_norm": 3.3625722787627033,
      "learning_rate": 6.6524520255863545e-06,
      "loss": 1.2377,
      "step": 156
    },
    {
      "epoch": 0.10044785668586052,
      "grad_norm": 3.293863687993725,
      "learning_rate": 6.695095948827293e-06,
      "loss": 1.2496,
      "step": 157
    },
    {
      "epoch": 0.10108765195137556,
      "grad_norm": 3.1007109468747953,
      "learning_rate": 6.737739872068231e-06,
      "loss": 1.1842,
      "step": 158
    },
    {
      "epoch": 0.1017274472168906,
      "grad_norm": 3.2241791868432252,
      "learning_rate": 6.780383795309169e-06,
      "loss": 1.2147,
      "step": 159
    },
    {
      "epoch": 0.10236724248240563,
      "grad_norm": 3.0874763766238855,
      "learning_rate": 6.823027718550107e-06,
      "loss": 1.2634,
      "step": 160
    },
    {
      "epoch": 0.10300703774792067,
      "grad_norm": 3.04226171945921,
      "learning_rate": 6.865671641791045e-06,
      "loss": 1.1764,
      "step": 161
    },
    {
      "epoch": 0.1036468330134357,
      "grad_norm": 3.5975597202814495,
      "learning_rate": 6.908315565031983e-06,
      "loss": 1.1246,
      "step": 162
    },
    {
      "epoch": 0.10428662827895073,
      "grad_norm": 3.410619517299722,
      "learning_rate": 6.950959488272921e-06,
      "loss": 1.2577,
      "step": 163
    },
    {
      "epoch": 0.10492642354446577,
      "grad_norm": 3.069978623167607,
      "learning_rate": 6.993603411513859e-06,
      "loss": 1.2518,
      "step": 164
    },
    {
      "epoch": 0.10556621880998081,
      "grad_norm": 3.3561516438782393,
      "learning_rate": 7.0362473347547975e-06,
      "loss": 1.1567,
      "step": 165
    },
    {
      "epoch": 0.10620601407549585,
      "grad_norm": 3.405445228117651,
      "learning_rate": 7.0788912579957356e-06,
      "loss": 1.1853,
      "step": 166
    },
    {
      "epoch": 0.10684580934101087,
      "grad_norm": 3.147405203641027,
      "learning_rate": 7.121535181236674e-06,
      "loss": 1.1615,
      "step": 167
    },
    {
      "epoch": 0.10748560460652591,
      "grad_norm": 2.912844634213437,
      "learning_rate": 7.164179104477612e-06,
      "loss": 1.2225,
      "step": 168
    },
    {
      "epoch": 0.10812539987204095,
      "grad_norm": 3.0213894832212107,
      "learning_rate": 7.20682302771855e-06,
      "loss": 1.241,
      "step": 169
    },
    {
      "epoch": 0.10876519513755598,
      "grad_norm": 3.5062031462607126,
      "learning_rate": 7.249466950959488e-06,
      "loss": 1.2366,
      "step": 170
    },
    {
      "epoch": 0.10940499040307101,
      "grad_norm": 3.2465176030505565,
      "learning_rate": 7.292110874200427e-06,
      "loss": 1.1889,
      "step": 171
    },
    {
      "epoch": 0.11004478566858605,
      "grad_norm": 2.7899602663161027,
      "learning_rate": 7.334754797441366e-06,
      "loss": 1.1848,
      "step": 172
    },
    {
      "epoch": 0.11068458093410109,
      "grad_norm": 3.2110729166433427,
      "learning_rate": 7.377398720682304e-06,
      "loss": 1.1523,
      "step": 173
    },
    {
      "epoch": 0.11132437619961612,
      "grad_norm": 3.6299403048166576,
      "learning_rate": 7.420042643923242e-06,
      "loss": 1.1899,
      "step": 174
    },
    {
      "epoch": 0.11196417146513116,
      "grad_norm": 3.096558144772406,
      "learning_rate": 7.46268656716418e-06,
      "loss": 1.2165,
      "step": 175
    },
    {
      "epoch": 0.1126039667306462,
      "grad_norm": 3.0559104508771155,
      "learning_rate": 7.505330490405118e-06,
      "loss": 1.1443,
      "step": 176
    },
    {
      "epoch": 0.11324376199616124,
      "grad_norm": 3.381432011041117,
      "learning_rate": 7.547974413646056e-06,
      "loss": 1.1972,
      "step": 177
    },
    {
      "epoch": 0.11388355726167626,
      "grad_norm": 3.6254391782041777,
      "learning_rate": 7.5906183368869945e-06,
      "loss": 1.1642,
      "step": 178
    },
    {
      "epoch": 0.1145233525271913,
      "grad_norm": 3.0624080823884707,
      "learning_rate": 7.633262260127933e-06,
      "loss": 1.1937,
      "step": 179
    },
    {
      "epoch": 0.11516314779270634,
      "grad_norm": 3.92948098649581,
      "learning_rate": 7.67590618336887e-06,
      "loss": 1.2678,
      "step": 180
    },
    {
      "epoch": 0.11580294305822136,
      "grad_norm": 3.644411116977709,
      "learning_rate": 7.718550106609809e-06,
      "loss": 1.1951,
      "step": 181
    },
    {
      "epoch": 0.1164427383237364,
      "grad_norm": 2.8784815861395043,
      "learning_rate": 7.761194029850747e-06,
      "loss": 1.1802,
      "step": 182
    },
    {
      "epoch": 0.11708253358925144,
      "grad_norm": 2.9341938877093727,
      "learning_rate": 7.803837953091685e-06,
      "loss": 1.0904,
      "step": 183
    },
    {
      "epoch": 0.11772232885476648,
      "grad_norm": 3.4437765358343224,
      "learning_rate": 7.846481876332623e-06,
      "loss": 1.1563,
      "step": 184
    },
    {
      "epoch": 0.1183621241202815,
      "grad_norm": 4.138256187122977,
      "learning_rate": 7.889125799573561e-06,
      "loss": 1.2283,
      "step": 185
    },
    {
      "epoch": 0.11900191938579655,
      "grad_norm": 3.7072589663885958,
      "learning_rate": 7.9317697228145e-06,
      "loss": 1.2191,
      "step": 186
    },
    {
      "epoch": 0.11964171465131158,
      "grad_norm": 3.2579551436689087,
      "learning_rate": 7.974413646055437e-06,
      "loss": 1.1861,
      "step": 187
    },
    {
      "epoch": 0.12028150991682661,
      "grad_norm": 3.5171017963941296,
      "learning_rate": 8.017057569296376e-06,
      "loss": 1.0956,
      "step": 188
    },
    {
      "epoch": 0.12092130518234165,
      "grad_norm": 4.042625693610425,
      "learning_rate": 8.059701492537314e-06,
      "loss": 1.1456,
      "step": 189
    },
    {
      "epoch": 0.12156110044785669,
      "grad_norm": 3.590222782821438,
      "learning_rate": 8.102345415778252e-06,
      "loss": 1.1465,
      "step": 190
    },
    {
      "epoch": 0.12220089571337173,
      "grad_norm": 3.3351900916761417,
      "learning_rate": 8.14498933901919e-06,
      "loss": 1.211,
      "step": 191
    },
    {
      "epoch": 0.12284069097888675,
      "grad_norm": 3.362576496039898,
      "learning_rate": 8.187633262260128e-06,
      "loss": 1.2046,
      "step": 192
    },
    {
      "epoch": 0.12348048624440179,
      "grad_norm": 3.4474720627928384,
      "learning_rate": 8.230277185501066e-06,
      "loss": 1.1713,
      "step": 193
    },
    {
      "epoch": 0.12412028150991683,
      "grad_norm": 3.4955330792918358,
      "learning_rate": 8.272921108742004e-06,
      "loss": 1.1771,
      "step": 194
    },
    {
      "epoch": 0.12476007677543186,
      "grad_norm": 3.1625555446754303,
      "learning_rate": 8.315565031982942e-06,
      "loss": 1.1554,
      "step": 195
    },
    {
      "epoch": 0.1253998720409469,
      "grad_norm": 3.4392689117766846,
      "learning_rate": 8.35820895522388e-06,
      "loss": 1.1908,
      "step": 196
    },
    {
      "epoch": 0.12603966730646193,
      "grad_norm": 4.025546441584292,
      "learning_rate": 8.400852878464819e-06,
      "loss": 1.2251,
      "step": 197
    },
    {
      "epoch": 0.12667946257197696,
      "grad_norm": 3.380909059651068,
      "learning_rate": 8.443496801705757e-06,
      "loss": 1.181,
      "step": 198
    },
    {
      "epoch": 0.127319257837492,
      "grad_norm": 3.2278758001592953,
      "learning_rate": 8.486140724946695e-06,
      "loss": 1.1649,
      "step": 199
    },
    {
      "epoch": 0.12795905310300704,
      "grad_norm": 3.2046563960859755,
      "learning_rate": 8.528784648187633e-06,
      "loss": 1.1796,
      "step": 200
    },
    {
      "epoch": 0.12859884836852206,
      "grad_norm": 3.2300485017804847,
      "learning_rate": 8.571428571428571e-06,
      "loss": 1.2446,
      "step": 201
    },
    {
      "epoch": 0.12923864363403711,
      "grad_norm": 3.118378246268439,
      "learning_rate": 8.614072494669509e-06,
      "loss": 1.1636,
      "step": 202
    },
    {
      "epoch": 0.12987843889955214,
      "grad_norm": 3.4156214239159866,
      "learning_rate": 8.656716417910447e-06,
      "loss": 1.1438,
      "step": 203
    },
    {
      "epoch": 0.13051823416506717,
      "grad_norm": 3.171059560804951,
      "learning_rate": 8.699360341151387e-06,
      "loss": 1.1898,
      "step": 204
    },
    {
      "epoch": 0.13115802943058222,
      "grad_norm": 3.4317295887842065,
      "learning_rate": 8.742004264392325e-06,
      "loss": 1.1837,
      "step": 205
    },
    {
      "epoch": 0.13179782469609724,
      "grad_norm": 3.029644188354263,
      "learning_rate": 8.784648187633263e-06,
      "loss": 1.1456,
      "step": 206
    },
    {
      "epoch": 0.1324376199616123,
      "grad_norm": 3.2225654641010997,
      "learning_rate": 8.827292110874201e-06,
      "loss": 1.1362,
      "step": 207
    },
    {
      "epoch": 0.13307741522712732,
      "grad_norm": 3.4156224699084277,
      "learning_rate": 8.86993603411514e-06,
      "loss": 1.1511,
      "step": 208
    },
    {
      "epoch": 0.13371721049264235,
      "grad_norm": 3.1621589528533374,
      "learning_rate": 8.912579957356077e-06,
      "loss": 1.1969,
      "step": 209
    },
    {
      "epoch": 0.1343570057581574,
      "grad_norm": 3.082553476585453,
      "learning_rate": 8.955223880597016e-06,
      "loss": 1.1333,
      "step": 210
    },
    {
      "epoch": 0.13499680102367242,
      "grad_norm": 3.2424073727663827,
      "learning_rate": 8.997867803837954e-06,
      "loss": 1.1808,
      "step": 211
    },
    {
      "epoch": 0.13563659628918745,
      "grad_norm": 3.599488876984696,
      "learning_rate": 9.040511727078892e-06,
      "loss": 1.1719,
      "step": 212
    },
    {
      "epoch": 0.1362763915547025,
      "grad_norm": 3.3333742239671835,
      "learning_rate": 9.08315565031983e-06,
      "loss": 1.1501,
      "step": 213
    },
    {
      "epoch": 0.13691618682021753,
      "grad_norm": 3.2695886068759923,
      "learning_rate": 9.125799573560768e-06,
      "loss": 1.1416,
      "step": 214
    },
    {
      "epoch": 0.13755598208573255,
      "grad_norm": 3.6145182945833825,
      "learning_rate": 9.168443496801706e-06,
      "loss": 1.171,
      "step": 215
    },
    {
      "epoch": 0.1381957773512476,
      "grad_norm": 3.918825188744275,
      "learning_rate": 9.211087420042644e-06,
      "loss": 1.155,
      "step": 216
    },
    {
      "epoch": 0.13883557261676263,
      "grad_norm": 3.725386101738647,
      "learning_rate": 9.253731343283582e-06,
      "loss": 1.1977,
      "step": 217
    },
    {
      "epoch": 0.13947536788227768,
      "grad_norm": 3.3013136248692265,
      "learning_rate": 9.29637526652452e-06,
      "loss": 1.1253,
      "step": 218
    },
    {
      "epoch": 0.1401151631477927,
      "grad_norm": 3.4511951614951917,
      "learning_rate": 9.339019189765458e-06,
      "loss": 1.1667,
      "step": 219
    },
    {
      "epoch": 0.14075495841330773,
      "grad_norm": 4.251245367744519,
      "learning_rate": 9.381663113006397e-06,
      "loss": 1.2073,
      "step": 220
    },
    {
      "epoch": 0.1413947536788228,
      "grad_norm": 3.3473803467412737,
      "learning_rate": 9.424307036247335e-06,
      "loss": 1.0963,
      "step": 221
    },
    {
      "epoch": 0.1420345489443378,
      "grad_norm": 3.238308285185854,
      "learning_rate": 9.466950959488273e-06,
      "loss": 1.161,
      "step": 222
    },
    {
      "epoch": 0.14267434420985284,
      "grad_norm": 3.6350143339135994,
      "learning_rate": 9.509594882729211e-06,
      "loss": 1.1361,
      "step": 223
    },
    {
      "epoch": 0.1433141394753679,
      "grad_norm": 3.347174323321505,
      "learning_rate": 9.552238805970149e-06,
      "loss": 1.143,
      "step": 224
    },
    {
      "epoch": 0.14395393474088292,
      "grad_norm": 3.3382717227492944,
      "learning_rate": 9.594882729211089e-06,
      "loss": 1.1695,
      "step": 225
    },
    {
      "epoch": 0.14459373000639794,
      "grad_norm": 3.202522362746085,
      "learning_rate": 9.637526652452027e-06,
      "loss": 1.1494,
      "step": 226
    },
    {
      "epoch": 0.145233525271913,
      "grad_norm": 3.385825706064351,
      "learning_rate": 9.680170575692965e-06,
      "loss": 1.1424,
      "step": 227
    },
    {
      "epoch": 0.14587332053742802,
      "grad_norm": 3.462150304079205,
      "learning_rate": 9.722814498933903e-06,
      "loss": 1.196,
      "step": 228
    },
    {
      "epoch": 0.14651311580294304,
      "grad_norm": 3.2038041175269707,
      "learning_rate": 9.765458422174841e-06,
      "loss": 1.1016,
      "step": 229
    },
    {
      "epoch": 0.1471529110684581,
      "grad_norm": 3.0438619568660483,
      "learning_rate": 9.80810234541578e-06,
      "loss": 1.1367,
      "step": 230
    },
    {
      "epoch": 0.14779270633397312,
      "grad_norm": 3.266916730485841,
      "learning_rate": 9.850746268656717e-06,
      "loss": 1.174,
      "step": 231
    },
    {
      "epoch": 0.14843250159948818,
      "grad_norm": 3.8170649686603264,
      "learning_rate": 9.893390191897656e-06,
      "loss": 1.1875,
      "step": 232
    },
    {
      "epoch": 0.1490722968650032,
      "grad_norm": 4.0888315387916645,
      "learning_rate": 9.936034115138594e-06,
      "loss": 1.1878,
      "step": 233
    },
    {
      "epoch": 0.14971209213051823,
      "grad_norm": 2.8710291993381896,
      "learning_rate": 9.978678038379532e-06,
      "loss": 1.2031,
      "step": 234
    },
    {
      "epoch": 0.15035188739603328,
      "grad_norm": 3.1159900078450553,
      "learning_rate": 1.002132196162047e-05,
      "loss": 1.1189,
      "step": 235
    },
    {
      "epoch": 0.1509916826615483,
      "grad_norm": 3.362766869183655,
      "learning_rate": 1.0063965884861408e-05,
      "loss": 1.1293,
      "step": 236
    },
    {
      "epoch": 0.15163147792706333,
      "grad_norm": 3.1178553247868903,
      "learning_rate": 1.0106609808102346e-05,
      "loss": 1.1116,
      "step": 237
    },
    {
      "epoch": 0.15227127319257838,
      "grad_norm": 3.401252114639348,
      "learning_rate": 1.0149253731343284e-05,
      "loss": 1.0962,
      "step": 238
    },
    {
      "epoch": 0.1529110684580934,
      "grad_norm": 3.2864113897260108,
      "learning_rate": 1.0191897654584222e-05,
      "loss": 1.2148,
      "step": 239
    },
    {
      "epoch": 0.15355086372360843,
      "grad_norm": 3.2275293356473314,
      "learning_rate": 1.023454157782516e-05,
      "loss": 1.1548,
      "step": 240
    },
    {
      "epoch": 0.15419065898912349,
      "grad_norm": 3.1188388328172136,
      "learning_rate": 1.0277185501066098e-05,
      "loss": 1.1485,
      "step": 241
    },
    {
      "epoch": 0.1548304542546385,
      "grad_norm": 3.2735091594124825,
      "learning_rate": 1.0319829424307037e-05,
      "loss": 1.1607,
      "step": 242
    },
    {
      "epoch": 0.15547024952015356,
      "grad_norm": 3.1736165629728292,
      "learning_rate": 1.0362473347547975e-05,
      "loss": 1.1909,
      "step": 243
    },
    {
      "epoch": 0.1561100447856686,
      "grad_norm": 3.8729744736541556,
      "learning_rate": 1.0405117270788913e-05,
      "loss": 1.0963,
      "step": 244
    },
    {
      "epoch": 0.15674984005118361,
      "grad_norm": 3.5258019497775916,
      "learning_rate": 1.0447761194029851e-05,
      "loss": 1.1327,
      "step": 245
    },
    {
      "epoch": 0.15738963531669867,
      "grad_norm": 3.064292684493223,
      "learning_rate": 1.049040511727079e-05,
      "loss": 1.1265,
      "step": 246
    },
    {
      "epoch": 0.1580294305822137,
      "grad_norm": 3.9782275681642405,
      "learning_rate": 1.0533049040511727e-05,
      "loss": 1.128,
      "step": 247
    },
    {
      "epoch": 0.15866922584772872,
      "grad_norm": 3.178941801923463,
      "learning_rate": 1.0575692963752667e-05,
      "loss": 1.2056,
      "step": 248
    },
    {
      "epoch": 0.15930902111324377,
      "grad_norm": 3.3048331727650333,
      "learning_rate": 1.0618336886993603e-05,
      "loss": 1.1213,
      "step": 249
    },
    {
      "epoch": 0.1599488163787588,
      "grad_norm": 3.448533678900044,
      "learning_rate": 1.0660980810234543e-05,
      "loss": 1.1243,
      "step": 250
    },
    {
      "epoch": 0.16058861164427382,
      "grad_norm": 2.991586230259323,
      "learning_rate": 1.0703624733475481e-05,
      "loss": 1.1076,
      "step": 251
    },
    {
      "epoch": 0.16122840690978887,
      "grad_norm": 3.312489689639886,
      "learning_rate": 1.074626865671642e-05,
      "loss": 1.1471,
      "step": 252
    },
    {
      "epoch": 0.1618682021753039,
      "grad_norm": 3.500908941141116,
      "learning_rate": 1.0788912579957357e-05,
      "loss": 1.1399,
      "step": 253
    },
    {
      "epoch": 0.16250799744081892,
      "grad_norm": 3.000634288231245,
      "learning_rate": 1.0831556503198296e-05,
      "loss": 1.1765,
      "step": 254
    },
    {
      "epoch": 0.16314779270633398,
      "grad_norm": 3.5036095821564133,
      "learning_rate": 1.0874200426439234e-05,
      "loss": 1.1546,
      "step": 255
    },
    {
      "epoch": 0.163787587971849,
      "grad_norm": 3.494461302628104,
      "learning_rate": 1.0916844349680172e-05,
      "loss": 1.0906,
      "step": 256
    },
    {
      "epoch": 0.16442738323736406,
      "grad_norm": 3.206948939564132,
      "learning_rate": 1.095948827292111e-05,
      "loss": 1.0982,
      "step": 257
    },
    {
      "epoch": 0.16506717850287908,
      "grad_norm": 3.8125710047003216,
      "learning_rate": 1.1002132196162048e-05,
      "loss": 1.1251,
      "step": 258
    },
    {
      "epoch": 0.1657069737683941,
      "grad_norm": 3.3406461485423264,
      "learning_rate": 1.1044776119402986e-05,
      "loss": 1.1714,
      "step": 259
    },
    {
      "epoch": 0.16634676903390916,
      "grad_norm": 3.3309233870557913,
      "learning_rate": 1.1087420042643924e-05,
      "loss": 1.1566,
      "step": 260
    },
    {
      "epoch": 0.16698656429942418,
      "grad_norm": 3.935968730822123,
      "learning_rate": 1.1130063965884862e-05,
      "loss": 1.1435,
      "step": 261
    },
    {
      "epoch": 0.1676263595649392,
      "grad_norm": 3.639687037770362,
      "learning_rate": 1.11727078891258e-05,
      "loss": 1.1455,
      "step": 262
    },
    {
      "epoch": 0.16826615483045426,
      "grad_norm": 3.7596559742325617,
      "learning_rate": 1.1215351812366738e-05,
      "loss": 1.1644,
      "step": 263
    },
    {
      "epoch": 0.1689059500959693,
      "grad_norm": 3.127163010908332,
      "learning_rate": 1.1257995735607677e-05,
      "loss": 1.1521,
      "step": 264
    },
    {
      "epoch": 0.1695457453614843,
      "grad_norm": 3.4765404021385895,
      "learning_rate": 1.1300639658848616e-05,
      "loss": 1.142,
      "step": 265
    },
    {
      "epoch": 0.17018554062699937,
      "grad_norm": 3.3672449405219202,
      "learning_rate": 1.1343283582089553e-05,
      "loss": 1.1312,
      "step": 266
    },
    {
      "epoch": 0.1708253358925144,
      "grad_norm": 3.764117514413339,
      "learning_rate": 1.1385927505330493e-05,
      "loss": 1.1382,
      "step": 267
    },
    {
      "epoch": 0.17146513115802944,
      "grad_norm": 3.400999513687139,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 1.1675,
      "step": 268
    },
    {
      "epoch": 0.17210492642354447,
      "grad_norm": 3.2940808950049263,
      "learning_rate": 1.1471215351812369e-05,
      "loss": 1.0715,
      "step": 269
    },
    {
      "epoch": 0.1727447216890595,
      "grad_norm": 4.004542796618932,
      "learning_rate": 1.1513859275053305e-05,
      "loss": 1.1828,
      "step": 270
    },
    {
      "epoch": 0.17338451695457455,
      "grad_norm": 3.725864088970507,
      "learning_rate": 1.1556503198294245e-05,
      "loss": 1.1394,
      "step": 271
    },
    {
      "epoch": 0.17402431222008957,
      "grad_norm": 3.497254394669451,
      "learning_rate": 1.1599147121535181e-05,
      "loss": 1.1643,
      "step": 272
    },
    {
      "epoch": 0.1746641074856046,
      "grad_norm": 3.3774715675711438,
      "learning_rate": 1.1641791044776121e-05,
      "loss": 1.1579,
      "step": 273
    },
    {
      "epoch": 0.17530390275111965,
      "grad_norm": 2.9231574071035484,
      "learning_rate": 1.1684434968017058e-05,
      "loss": 1.11,
      "step": 274
    },
    {
      "epoch": 0.17594369801663468,
      "grad_norm": 3.672759719739446,
      "learning_rate": 1.1727078891257997e-05,
      "loss": 1.1172,
      "step": 275
    },
    {
      "epoch": 0.1765834932821497,
      "grad_norm": 3.313100721393884,
      "learning_rate": 1.1769722814498934e-05,
      "loss": 1.191,
      "step": 276
    },
    {
      "epoch": 0.17722328854766475,
      "grad_norm": 3.3446342792613657,
      "learning_rate": 1.1812366737739874e-05,
      "loss": 1.135,
      "step": 277
    },
    {
      "epoch": 0.17786308381317978,
      "grad_norm": 3.5775866710604007,
      "learning_rate": 1.185501066098081e-05,
      "loss": 1.1409,
      "step": 278
    },
    {
      "epoch": 0.1785028790786948,
      "grad_norm": 3.693714452215452,
      "learning_rate": 1.189765458422175e-05,
      "loss": 1.1899,
      "step": 279
    },
    {
      "epoch": 0.17914267434420986,
      "grad_norm": 3.7764729284562186,
      "learning_rate": 1.1940298507462686e-05,
      "loss": 1.1453,
      "step": 280
    },
    {
      "epoch": 0.17978246960972488,
      "grad_norm": 3.8378546212637628,
      "learning_rate": 1.1982942430703626e-05,
      "loss": 1.1644,
      "step": 281
    },
    {
      "epoch": 0.18042226487523993,
      "grad_norm": 3.3025102656644187,
      "learning_rate": 1.2025586353944564e-05,
      "loss": 1.1647,
      "step": 282
    },
    {
      "epoch": 0.18106206014075496,
      "grad_norm": 3.423646558361597,
      "learning_rate": 1.2068230277185502e-05,
      "loss": 1.1064,
      "step": 283
    },
    {
      "epoch": 0.18170185540626999,
      "grad_norm": 4.1341908987257,
      "learning_rate": 1.2110874200426442e-05,
      "loss": 1.1283,
      "step": 284
    },
    {
      "epoch": 0.18234165067178504,
      "grad_norm": 3.2741158825598538,
      "learning_rate": 1.2153518123667378e-05,
      "loss": 1.1795,
      "step": 285
    },
    {
      "epoch": 0.18298144593730006,
      "grad_norm": 4.0812831411332136,
      "learning_rate": 1.2196162046908318e-05,
      "loss": 1.1006,
      "step": 286
    },
    {
      "epoch": 0.1836212412028151,
      "grad_norm": 3.9915973222805974,
      "learning_rate": 1.2238805970149255e-05,
      "loss": 1.1394,
      "step": 287
    },
    {
      "epoch": 0.18426103646833014,
      "grad_norm": 3.186458356609679,
      "learning_rate": 1.2281449893390195e-05,
      "loss": 1.0919,
      "step": 288
    },
    {
      "epoch": 0.18490083173384517,
      "grad_norm": 3.435493380667927,
      "learning_rate": 1.2324093816631131e-05,
      "loss": 1.1948,
      "step": 289
    },
    {
      "epoch": 0.1855406269993602,
      "grad_norm": 3.6390140009570753,
      "learning_rate": 1.236673773987207e-05,
      "loss": 1.1364,
      "step": 290
    },
    {
      "epoch": 0.18618042226487524,
      "grad_norm": 3.2192913907070273,
      "learning_rate": 1.2409381663113007e-05,
      "loss": 1.1021,
      "step": 291
    },
    {
      "epoch": 0.18682021753039027,
      "grad_norm": 3.387285414667607,
      "learning_rate": 1.2452025586353947e-05,
      "loss": 1.1236,
      "step": 292
    },
    {
      "epoch": 0.18746001279590532,
      "grad_norm": 3.718465225925112,
      "learning_rate": 1.2494669509594883e-05,
      "loss": 1.1297,
      "step": 293
    },
    {
      "epoch": 0.18809980806142035,
      "grad_norm": 3.678306254288396,
      "learning_rate": 1.2537313432835823e-05,
      "loss": 1.1778,
      "step": 294
    },
    {
      "epoch": 0.18873960332693537,
      "grad_norm": 2.853673663711992,
      "learning_rate": 1.257995735607676e-05,
      "loss": 1.2189,
      "step": 295
    },
    {
      "epoch": 0.18937939859245043,
      "grad_norm": 3.2510464819391935,
      "learning_rate": 1.26226012793177e-05,
      "loss": 1.1343,
      "step": 296
    },
    {
      "epoch": 0.19001919385796545,
      "grad_norm": 3.703613950913055,
      "learning_rate": 1.2665245202558636e-05,
      "loss": 1.1694,
      "step": 297
    },
    {
      "epoch": 0.19065898912348048,
      "grad_norm": 3.5236596281145185,
      "learning_rate": 1.2707889125799576e-05,
      "loss": 1.1046,
      "step": 298
    },
    {
      "epoch": 0.19129878438899553,
      "grad_norm": 2.9041668388286177,
      "learning_rate": 1.2750533049040512e-05,
      "loss": 1.1446,
      "step": 299
    },
    {
      "epoch": 0.19193857965451055,
      "grad_norm": 3.2734587135070763,
      "learning_rate": 1.2793176972281452e-05,
      "loss": 1.122,
      "step": 300
    },
    {
      "epoch": 0.19257837492002558,
      "grad_norm": 3.126202876230851,
      "learning_rate": 1.2835820895522388e-05,
      "loss": 1.1438,
      "step": 301
    },
    {
      "epoch": 0.19321817018554063,
      "grad_norm": 3.3714826161560545,
      "learning_rate": 1.2878464818763328e-05,
      "loss": 1.1541,
      "step": 302
    },
    {
      "epoch": 0.19385796545105566,
      "grad_norm": 3.03291078273618,
      "learning_rate": 1.2921108742004264e-05,
      "loss": 1.1218,
      "step": 303
    },
    {
      "epoch": 0.1944977607165707,
      "grad_norm": 4.125589073598504,
      "learning_rate": 1.2963752665245204e-05,
      "loss": 1.1229,
      "step": 304
    },
    {
      "epoch": 0.19513755598208574,
      "grad_norm": 3.5601867751942047,
      "learning_rate": 1.300639658848614e-05,
      "loss": 1.156,
      "step": 305
    },
    {
      "epoch": 0.19577735124760076,
      "grad_norm": 3.331342727462086,
      "learning_rate": 1.304904051172708e-05,
      "loss": 1.0942,
      "step": 306
    },
    {
      "epoch": 0.19641714651311581,
      "grad_norm": 3.4266222610487707,
      "learning_rate": 1.3091684434968017e-05,
      "loss": 1.1426,
      "step": 307
    },
    {
      "epoch": 0.19705694177863084,
      "grad_norm": 3.453686949902591,
      "learning_rate": 1.3134328358208957e-05,
      "loss": 1.1125,
      "step": 308
    },
    {
      "epoch": 0.19769673704414586,
      "grad_norm": 3.0922992636923525,
      "learning_rate": 1.3176972281449893e-05,
      "loss": 1.1774,
      "step": 309
    },
    {
      "epoch": 0.19833653230966092,
      "grad_norm": 3.303013967825049,
      "learning_rate": 1.3219616204690833e-05,
      "loss": 1.1345,
      "step": 310
    },
    {
      "epoch": 0.19897632757517594,
      "grad_norm": 3.243591580152601,
      "learning_rate": 1.326226012793177e-05,
      "loss": 1.0866,
      "step": 311
    },
    {
      "epoch": 0.19961612284069097,
      "grad_norm": 3.4531513948500336,
      "learning_rate": 1.3304904051172709e-05,
      "loss": 1.1116,
      "step": 312
    },
    {
      "epoch": 0.20025591810620602,
      "grad_norm": 3.1092694806743624,
      "learning_rate": 1.3347547974413649e-05,
      "loss": 1.1249,
      "step": 313
    },
    {
      "epoch": 0.20089571337172105,
      "grad_norm": 3.3750749649434377,
      "learning_rate": 1.3390191897654585e-05,
      "loss": 1.1169,
      "step": 314
    },
    {
      "epoch": 0.20153550863723607,
      "grad_norm": 3.5256558396350104,
      "learning_rate": 1.3432835820895525e-05,
      "loss": 1.0703,
      "step": 315
    },
    {
      "epoch": 0.20217530390275112,
      "grad_norm": 3.13509890184876,
      "learning_rate": 1.3475479744136461e-05,
      "loss": 1.1637,
      "step": 316
    },
    {
      "epoch": 0.20281509916826615,
      "grad_norm": 3.800454004266649,
      "learning_rate": 1.3518123667377401e-05,
      "loss": 1.1937,
      "step": 317
    },
    {
      "epoch": 0.2034548944337812,
      "grad_norm": 3.180713186709017,
      "learning_rate": 1.3560767590618338e-05,
      "loss": 1.1205,
      "step": 318
    },
    {
      "epoch": 0.20409468969929623,
      "grad_norm": 2.904628204114605,
      "learning_rate": 1.3603411513859277e-05,
      "loss": 1.1106,
      "step": 319
    },
    {
      "epoch": 0.20473448496481125,
      "grad_norm": 3.6857799247374836,
      "learning_rate": 1.3646055437100214e-05,
      "loss": 1.1082,
      "step": 320
    },
    {
      "epoch": 0.2053742802303263,
      "grad_norm": 3.3096080560630896,
      "learning_rate": 1.3688699360341154e-05,
      "loss": 1.1008,
      "step": 321
    },
    {
      "epoch": 0.20601407549584133,
      "grad_norm": 2.970199060649059,
      "learning_rate": 1.373134328358209e-05,
      "loss": 1.1533,
      "step": 322
    },
    {
      "epoch": 0.20665387076135636,
      "grad_norm": 3.3765469090763274,
      "learning_rate": 1.377398720682303e-05,
      "loss": 1.1536,
      "step": 323
    },
    {
      "epoch": 0.2072936660268714,
      "grad_norm": 3.1927873705597976,
      "learning_rate": 1.3816631130063966e-05,
      "loss": 1.1611,
      "step": 324
    },
    {
      "epoch": 0.20793346129238643,
      "grad_norm": 3.3576532589028636,
      "learning_rate": 1.3859275053304906e-05,
      "loss": 1.0898,
      "step": 325
    },
    {
      "epoch": 0.20857325655790146,
      "grad_norm": 3.26048513779939,
      "learning_rate": 1.3901918976545842e-05,
      "loss": 1.1715,
      "step": 326
    },
    {
      "epoch": 0.2092130518234165,
      "grad_norm": 3.1331824368149626,
      "learning_rate": 1.3944562899786782e-05,
      "loss": 1.1296,
      "step": 327
    },
    {
      "epoch": 0.20985284708893154,
      "grad_norm": 3.7052560725646826,
      "learning_rate": 1.3987206823027719e-05,
      "loss": 1.0821,
      "step": 328
    },
    {
      "epoch": 0.2104926423544466,
      "grad_norm": 3.2605684077242985,
      "learning_rate": 1.4029850746268658e-05,
      "loss": 1.0799,
      "step": 329
    },
    {
      "epoch": 0.21113243761996162,
      "grad_norm": 3.4493163463927043,
      "learning_rate": 1.4072494669509595e-05,
      "loss": 1.0932,
      "step": 330
    },
    {
      "epoch": 0.21177223288547664,
      "grad_norm": 3.3274098731634987,
      "learning_rate": 1.4115138592750535e-05,
      "loss": 1.1504,
      "step": 331
    },
    {
      "epoch": 0.2124120281509917,
      "grad_norm": 3.471279335996965,
      "learning_rate": 1.4157782515991471e-05,
      "loss": 1.1493,
      "step": 332
    },
    {
      "epoch": 0.21305182341650672,
      "grad_norm": 3.6975458153694403,
      "learning_rate": 1.4200426439232411e-05,
      "loss": 1.136,
      "step": 333
    },
    {
      "epoch": 0.21369161868202174,
      "grad_norm": 3.961727232202024,
      "learning_rate": 1.4243070362473347e-05,
      "loss": 1.123,
      "step": 334
    },
    {
      "epoch": 0.2143314139475368,
      "grad_norm": 3.402768094360313,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 1.1449,
      "step": 335
    },
    {
      "epoch": 0.21497120921305182,
      "grad_norm": 3.44087628108031,
      "learning_rate": 1.4328358208955224e-05,
      "loss": 1.0829,
      "step": 336
    },
    {
      "epoch": 0.21561100447856685,
      "grad_norm": 3.4395136204117938,
      "learning_rate": 1.4371002132196163e-05,
      "loss": 1.1312,
      "step": 337
    },
    {
      "epoch": 0.2162507997440819,
      "grad_norm": 3.037854414534309,
      "learning_rate": 1.44136460554371e-05,
      "loss": 1.1182,
      "step": 338
    },
    {
      "epoch": 0.21689059500959693,
      "grad_norm": 3.340596584154051,
      "learning_rate": 1.445628997867804e-05,
      "loss": 1.2137,
      "step": 339
    },
    {
      "epoch": 0.21753039027511195,
      "grad_norm": 3.195315779420266,
      "learning_rate": 1.4498933901918976e-05,
      "loss": 1.1559,
      "step": 340
    },
    {
      "epoch": 0.218170185540627,
      "grad_norm": 3.336461803883682,
      "learning_rate": 1.4541577825159916e-05,
      "loss": 1.1202,
      "step": 341
    },
    {
      "epoch": 0.21880998080614203,
      "grad_norm": 2.9625022833643864,
      "learning_rate": 1.4584221748400854e-05,
      "loss": 1.0949,
      "step": 342
    },
    {
      "epoch": 0.21944977607165708,
      "grad_norm": 3.209893911864078,
      "learning_rate": 1.4626865671641792e-05,
      "loss": 1.102,
      "step": 343
    },
    {
      "epoch": 0.2200895713371721,
      "grad_norm": 3.685136318076513,
      "learning_rate": 1.4669509594882732e-05,
      "loss": 1.1296,
      "step": 344
    },
    {
      "epoch": 0.22072936660268713,
      "grad_norm": 3.8793217285498027,
      "learning_rate": 1.4712153518123668e-05,
      "loss": 1.1113,
      "step": 345
    },
    {
      "epoch": 0.22136916186820219,
      "grad_norm": 3.517669177166419,
      "learning_rate": 1.4754797441364608e-05,
      "loss": 1.0691,
      "step": 346
    },
    {
      "epoch": 0.2220089571337172,
      "grad_norm": 3.32356455599649,
      "learning_rate": 1.4797441364605544e-05,
      "loss": 1.1338,
      "step": 347
    },
    {
      "epoch": 0.22264875239923224,
      "grad_norm": 3.5532630524976443,
      "learning_rate": 1.4840085287846484e-05,
      "loss": 1.1173,
      "step": 348
    },
    {
      "epoch": 0.2232885476647473,
      "grad_norm": 3.5617820522480312,
      "learning_rate": 1.488272921108742e-05,
      "loss": 1.1197,
      "step": 349
    },
    {
      "epoch": 0.22392834293026231,
      "grad_norm": 3.6146163697715172,
      "learning_rate": 1.492537313432836e-05,
      "loss": 1.1422,
      "step": 350
    },
    {
      "epoch": 0.22456813819577734,
      "grad_norm": 3.2758955686910762,
      "learning_rate": 1.4968017057569297e-05,
      "loss": 1.0373,
      "step": 351
    },
    {
      "epoch": 0.2252079334612924,
      "grad_norm": 3.4814751157816333,
      "learning_rate": 1.5010660980810237e-05,
      "loss": 1.1244,
      "step": 352
    },
    {
      "epoch": 0.22584772872680742,
      "grad_norm": 3.353785103563738,
      "learning_rate": 1.5053304904051173e-05,
      "loss": 1.1008,
      "step": 353
    },
    {
      "epoch": 0.22648752399232247,
      "grad_norm": 3.547789099444003,
      "learning_rate": 1.5095948827292113e-05,
      "loss": 1.1267,
      "step": 354
    },
    {
      "epoch": 0.2271273192578375,
      "grad_norm": 3.027491906841539,
      "learning_rate": 1.513859275053305e-05,
      "loss": 1.0774,
      "step": 355
    },
    {
      "epoch": 0.22776711452335252,
      "grad_norm": 3.3906688857051153,
      "learning_rate": 1.5181236673773989e-05,
      "loss": 1.1254,
      "step": 356
    },
    {
      "epoch": 0.22840690978886757,
      "grad_norm": 3.4684979882050153,
      "learning_rate": 1.5223880597014925e-05,
      "loss": 1.069,
      "step": 357
    },
    {
      "epoch": 0.2290467050543826,
      "grad_norm": 3.505767094972919,
      "learning_rate": 1.5266524520255865e-05,
      "loss": 1.1422,
      "step": 358
    },
    {
      "epoch": 0.22968650031989762,
      "grad_norm": 3.8212486955195724,
      "learning_rate": 1.5309168443496803e-05,
      "loss": 1.1519,
      "step": 359
    },
    {
      "epoch": 0.23032629558541268,
      "grad_norm": 3.59251181782098,
      "learning_rate": 1.535181236673774e-05,
      "loss": 1.1368,
      "step": 360
    },
    {
      "epoch": 0.2309660908509277,
      "grad_norm": 3.5880841295545376,
      "learning_rate": 1.539445628997868e-05,
      "loss": 1.1298,
      "step": 361
    },
    {
      "epoch": 0.23160588611644273,
      "grad_norm": 4.284348178910535,
      "learning_rate": 1.5437100213219618e-05,
      "loss": 1.0816,
      "step": 362
    },
    {
      "epoch": 0.23224568138195778,
      "grad_norm": 4.143500707066932,
      "learning_rate": 1.5479744136460556e-05,
      "loss": 1.1878,
      "step": 363
    },
    {
      "epoch": 0.2328854766474728,
      "grad_norm": 4.218800321189495,
      "learning_rate": 1.5522388059701494e-05,
      "loss": 1.1088,
      "step": 364
    },
    {
      "epoch": 0.23352527191298783,
      "grad_norm": 3.6744239701268064,
      "learning_rate": 1.5565031982942432e-05,
      "loss": 1.1113,
      "step": 365
    },
    {
      "epoch": 0.23416506717850288,
      "grad_norm": 3.9149520879386195,
      "learning_rate": 1.560767590618337e-05,
      "loss": 1.0613,
      "step": 366
    },
    {
      "epoch": 0.2348048624440179,
      "grad_norm": 3.4166968875829373,
      "learning_rate": 1.5650319829424308e-05,
      "loss": 1.1604,
      "step": 367
    },
    {
      "epoch": 0.23544465770953296,
      "grad_norm": 4.2231921495335945,
      "learning_rate": 1.5692963752665246e-05,
      "loss": 1.1608,
      "step": 368
    },
    {
      "epoch": 0.236084452975048,
      "grad_norm": 3.1842430962937875,
      "learning_rate": 1.5735607675906184e-05,
      "loss": 1.1608,
      "step": 369
    },
    {
      "epoch": 0.236724248240563,
      "grad_norm": 3.503999906686361,
      "learning_rate": 1.5778251599147122e-05,
      "loss": 1.1326,
      "step": 370
    },
    {
      "epoch": 0.23736404350607807,
      "grad_norm": 4.320438236191231,
      "learning_rate": 1.582089552238806e-05,
      "loss": 1.0758,
      "step": 371
    },
    {
      "epoch": 0.2380038387715931,
      "grad_norm": 3.476347990756033,
      "learning_rate": 1.5863539445629e-05,
      "loss": 1.0964,
      "step": 372
    },
    {
      "epoch": 0.23864363403710812,
      "grad_norm": 3.413477851987653,
      "learning_rate": 1.5906183368869937e-05,
      "loss": 1.0941,
      "step": 373
    },
    {
      "epoch": 0.23928342930262317,
      "grad_norm": 3.3770012162115353,
      "learning_rate": 1.5948827292110875e-05,
      "loss": 1.156,
      "step": 374
    },
    {
      "epoch": 0.2399232245681382,
      "grad_norm": 3.7120251393105264,
      "learning_rate": 1.5991471215351813e-05,
      "loss": 1.1524,
      "step": 375
    },
    {
      "epoch": 0.24056301983365322,
      "grad_norm": 3.314203841757412,
      "learning_rate": 1.603411513859275e-05,
      "loss": 1.0886,
      "step": 376
    },
    {
      "epoch": 0.24120281509916827,
      "grad_norm": 3.2009291857639197,
      "learning_rate": 1.6076759061833693e-05,
      "loss": 1.0912,
      "step": 377
    },
    {
      "epoch": 0.2418426103646833,
      "grad_norm": 3.4236438531676514,
      "learning_rate": 1.6119402985074627e-05,
      "loss": 1.0973,
      "step": 378
    },
    {
      "epoch": 0.24248240563019835,
      "grad_norm": 3.4637870415543395,
      "learning_rate": 1.616204690831557e-05,
      "loss": 1.0961,
      "step": 379
    },
    {
      "epoch": 0.24312220089571338,
      "grad_norm": 2.9287822663214382,
      "learning_rate": 1.6204690831556504e-05,
      "loss": 1.077,
      "step": 380
    },
    {
      "epoch": 0.2437619961612284,
      "grad_norm": 3.3010312658163117,
      "learning_rate": 1.6247334754797445e-05,
      "loss": 1.1235,
      "step": 381
    },
    {
      "epoch": 0.24440179142674345,
      "grad_norm": 3.16572276919956,
      "learning_rate": 1.628997867803838e-05,
      "loss": 1.1202,
      "step": 382
    },
    {
      "epoch": 0.24504158669225848,
      "grad_norm": 3.327234818498713,
      "learning_rate": 1.633262260127932e-05,
      "loss": 1.1034,
      "step": 383
    },
    {
      "epoch": 0.2456813819577735,
      "grad_norm": 3.460208586474368,
      "learning_rate": 1.6375266524520256e-05,
      "loss": 1.1135,
      "step": 384
    },
    {
      "epoch": 0.24632117722328856,
      "grad_norm": 3.630359566204365,
      "learning_rate": 1.6417910447761197e-05,
      "loss": 1.1097,
      "step": 385
    },
    {
      "epoch": 0.24696097248880358,
      "grad_norm": 3.5939905774575163,
      "learning_rate": 1.6460554371002132e-05,
      "loss": 1.1338,
      "step": 386
    },
    {
      "epoch": 0.2476007677543186,
      "grad_norm": 3.3980240613259887,
      "learning_rate": 1.6503198294243074e-05,
      "loss": 1.0903,
      "step": 387
    },
    {
      "epoch": 0.24824056301983366,
      "grad_norm": 3.4053687945501694,
      "learning_rate": 1.654584221748401e-05,
      "loss": 1.1348,
      "step": 388
    },
    {
      "epoch": 0.24888035828534869,
      "grad_norm": 3.4880895693819642,
      "learning_rate": 1.658848614072495e-05,
      "loss": 1.1368,
      "step": 389
    },
    {
      "epoch": 0.2495201535508637,
      "grad_norm": 3.0309296823440652,
      "learning_rate": 1.6631130063965885e-05,
      "loss": 1.1159,
      "step": 390
    },
    {
      "epoch": 0.25015994881637876,
      "grad_norm": 3.506745490016452,
      "learning_rate": 1.6673773987206826e-05,
      "loss": 1.1076,
      "step": 391
    },
    {
      "epoch": 0.2507997440818938,
      "grad_norm": 3.580614080558159,
      "learning_rate": 1.671641791044776e-05,
      "loss": 1.1273,
      "step": 392
    },
    {
      "epoch": 0.2514395393474088,
      "grad_norm": 3.428549052470366,
      "learning_rate": 1.6759061833688702e-05,
      "loss": 1.1667,
      "step": 393
    },
    {
      "epoch": 0.25207933461292387,
      "grad_norm": 3.417145522847714,
      "learning_rate": 1.6801705756929637e-05,
      "loss": 1.1988,
      "step": 394
    },
    {
      "epoch": 0.2527191298784389,
      "grad_norm": 4.041423206787811,
      "learning_rate": 1.684434968017058e-05,
      "loss": 1.087,
      "step": 395
    },
    {
      "epoch": 0.2533589251439539,
      "grad_norm": 3.738706038085192,
      "learning_rate": 1.6886993603411513e-05,
      "loss": 1.0899,
      "step": 396
    },
    {
      "epoch": 0.25399872040946897,
      "grad_norm": 3.2830649939784844,
      "learning_rate": 1.6929637526652455e-05,
      "loss": 1.0975,
      "step": 397
    },
    {
      "epoch": 0.254638515674984,
      "grad_norm": 3.6965568631559242,
      "learning_rate": 1.697228144989339e-05,
      "loss": 1.113,
      "step": 398
    },
    {
      "epoch": 0.255278310940499,
      "grad_norm": 3.4321909522785106,
      "learning_rate": 1.701492537313433e-05,
      "loss": 1.1766,
      "step": 399
    },
    {
      "epoch": 0.2559181062060141,
      "grad_norm": 3.1275188970534447,
      "learning_rate": 1.7057569296375266e-05,
      "loss": 1.1011,
      "step": 400
    },
    {
      "epoch": 0.2565579014715291,
      "grad_norm": 3.220900081994847,
      "learning_rate": 1.7100213219616207e-05,
      "loss": 1.1479,
      "step": 401
    },
    {
      "epoch": 0.2571976967370441,
      "grad_norm": 3.4091639927964326,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 1.1169,
      "step": 402
    },
    {
      "epoch": 0.2578374920025592,
      "grad_norm": 2.8004883716443367,
      "learning_rate": 1.7185501066098083e-05,
      "loss": 1.1588,
      "step": 403
    },
    {
      "epoch": 0.25847728726807423,
      "grad_norm": 3.2575359855868875,
      "learning_rate": 1.7228144989339018e-05,
      "loss": 1.0464,
      "step": 404
    },
    {
      "epoch": 0.2591170825335892,
      "grad_norm": 4.033348550582219,
      "learning_rate": 1.727078891257996e-05,
      "loss": 1.1709,
      "step": 405
    },
    {
      "epoch": 0.2597568777991043,
      "grad_norm": 3.5617732055899576,
      "learning_rate": 1.7313432835820894e-05,
      "loss": 1.1681,
      "step": 406
    },
    {
      "epoch": 0.26039667306461933,
      "grad_norm": 2.8644891123945966,
      "learning_rate": 1.7356076759061836e-05,
      "loss": 1.1178,
      "step": 407
    },
    {
      "epoch": 0.26103646833013433,
      "grad_norm": 3.34274346990883,
      "learning_rate": 1.7398720682302774e-05,
      "loss": 1.0489,
      "step": 408
    },
    {
      "epoch": 0.2616762635956494,
      "grad_norm": 3.5562848073481725,
      "learning_rate": 1.7441364605543712e-05,
      "loss": 1.1123,
      "step": 409
    },
    {
      "epoch": 0.26231605886116444,
      "grad_norm": 3.5643261127593573,
      "learning_rate": 1.748400852878465e-05,
      "loss": 1.088,
      "step": 410
    },
    {
      "epoch": 0.2629558541266795,
      "grad_norm": 3.1931922651184115,
      "learning_rate": 1.7526652452025588e-05,
      "loss": 1.1092,
      "step": 411
    },
    {
      "epoch": 0.2635956493921945,
      "grad_norm": 3.3152950165886916,
      "learning_rate": 1.7569296375266526e-05,
      "loss": 1.0507,
      "step": 412
    },
    {
      "epoch": 0.26423544465770954,
      "grad_norm": 3.3533072946041904,
      "learning_rate": 1.7611940298507464e-05,
      "loss": 1.1186,
      "step": 413
    },
    {
      "epoch": 0.2648752399232246,
      "grad_norm": 3.6345977889904555,
      "learning_rate": 1.7654584221748402e-05,
      "loss": 1.1379,
      "step": 414
    },
    {
      "epoch": 0.2655150351887396,
      "grad_norm": 3.128820218575811,
      "learning_rate": 1.769722814498934e-05,
      "loss": 1.134,
      "step": 415
    },
    {
      "epoch": 0.26615483045425464,
      "grad_norm": 2.937615065373637,
      "learning_rate": 1.773987206823028e-05,
      "loss": 1.0882,
      "step": 416
    },
    {
      "epoch": 0.2667946257197697,
      "grad_norm": 3.5986973604947257,
      "learning_rate": 1.7782515991471217e-05,
      "loss": 1.1235,
      "step": 417
    },
    {
      "epoch": 0.2674344209852847,
      "grad_norm": 3.4829394224826067,
      "learning_rate": 1.7825159914712155e-05,
      "loss": 1.0873,
      "step": 418
    },
    {
      "epoch": 0.26807421625079975,
      "grad_norm": 3.4790334703252546,
      "learning_rate": 1.7867803837953093e-05,
      "loss": 1.0858,
      "step": 419
    },
    {
      "epoch": 0.2687140115163148,
      "grad_norm": 3.225785347792026,
      "learning_rate": 1.791044776119403e-05,
      "loss": 1.0591,
      "step": 420
    },
    {
      "epoch": 0.2693538067818298,
      "grad_norm": 3.444037591639599,
      "learning_rate": 1.795309168443497e-05,
      "loss": 1.1147,
      "step": 421
    },
    {
      "epoch": 0.26999360204734485,
      "grad_norm": 3.7717758697160964,
      "learning_rate": 1.7995735607675907e-05,
      "loss": 1.1161,
      "step": 422
    },
    {
      "epoch": 0.2706333973128599,
      "grad_norm": 3.3592017940990053,
      "learning_rate": 1.8038379530916845e-05,
      "loss": 1.0376,
      "step": 423
    },
    {
      "epoch": 0.2712731925783749,
      "grad_norm": 4.1248989663068745,
      "learning_rate": 1.8081023454157784e-05,
      "loss": 1.1376,
      "step": 424
    },
    {
      "epoch": 0.27191298784388995,
      "grad_norm": 3.161996177495681,
      "learning_rate": 1.812366737739872e-05,
      "loss": 1.1135,
      "step": 425
    },
    {
      "epoch": 0.272552783109405,
      "grad_norm": 3.3812359169311725,
      "learning_rate": 1.816631130063966e-05,
      "loss": 1.1182,
      "step": 426
    },
    {
      "epoch": 0.27319257837492,
      "grad_norm": 4.369588351837968,
      "learning_rate": 1.8208955223880598e-05,
      "loss": 1.1102,
      "step": 427
    },
    {
      "epoch": 0.27383237364043506,
      "grad_norm": 3.861069921329403,
      "learning_rate": 1.8251599147121536e-05,
      "loss": 1.1654,
      "step": 428
    },
    {
      "epoch": 0.2744721689059501,
      "grad_norm": 3.879975779843111,
      "learning_rate": 1.8294243070362474e-05,
      "loss": 1.0886,
      "step": 429
    },
    {
      "epoch": 0.2751119641714651,
      "grad_norm": 3.4469036441849226,
      "learning_rate": 1.8336886993603412e-05,
      "loss": 1.0894,
      "step": 430
    },
    {
      "epoch": 0.27575175943698016,
      "grad_norm": 3.474280461845651,
      "learning_rate": 1.837953091684435e-05,
      "loss": 1.131,
      "step": 431
    },
    {
      "epoch": 0.2763915547024952,
      "grad_norm": 4.640802492172422,
      "learning_rate": 1.842217484008529e-05,
      "loss": 1.1229,
      "step": 432
    },
    {
      "epoch": 0.2770313499680102,
      "grad_norm": 4.0098521102348,
      "learning_rate": 1.8464818763326226e-05,
      "loss": 1.0819,
      "step": 433
    },
    {
      "epoch": 0.27767114523352526,
      "grad_norm": 3.9503731515881975,
      "learning_rate": 1.8507462686567165e-05,
      "loss": 1.1048,
      "step": 434
    },
    {
      "epoch": 0.2783109404990403,
      "grad_norm": 2.9167823998236506,
      "learning_rate": 1.8550106609808103e-05,
      "loss": 1.0801,
      "step": 435
    },
    {
      "epoch": 0.27895073576455537,
      "grad_norm": 3.895591321635814,
      "learning_rate": 1.859275053304904e-05,
      "loss": 1.0648,
      "step": 436
    },
    {
      "epoch": 0.27959053103007037,
      "grad_norm": 4.278236637137736,
      "learning_rate": 1.863539445628998e-05,
      "loss": 1.1122,
      "step": 437
    },
    {
      "epoch": 0.2802303262955854,
      "grad_norm": 3.199948932558296,
      "learning_rate": 1.8678038379530917e-05,
      "loss": 1.0892,
      "step": 438
    },
    {
      "epoch": 0.28087012156110047,
      "grad_norm": 3.6873527888576425,
      "learning_rate": 1.872068230277186e-05,
      "loss": 1.1305,
      "step": 439
    },
    {
      "epoch": 0.28150991682661547,
      "grad_norm": 3.754430016745001,
      "learning_rate": 1.8763326226012793e-05,
      "loss": 1.1403,
      "step": 440
    },
    {
      "epoch": 0.2821497120921305,
      "grad_norm": 3.9218863057714324,
      "learning_rate": 1.8805970149253735e-05,
      "loss": 1.1138,
      "step": 441
    },
    {
      "epoch": 0.2827895073576456,
      "grad_norm": 3.445027867658304,
      "learning_rate": 1.884861407249467e-05,
      "loss": 1.1042,
      "step": 442
    },
    {
      "epoch": 0.2834293026231606,
      "grad_norm": 3.283822297147565,
      "learning_rate": 1.889125799573561e-05,
      "loss": 1.1433,
      "step": 443
    },
    {
      "epoch": 0.2840690978886756,
      "grad_norm": 3.2065588400036917,
      "learning_rate": 1.8933901918976546e-05,
      "loss": 1.0591,
      "step": 444
    },
    {
      "epoch": 0.2847088931541907,
      "grad_norm": 3.3341790612076516,
      "learning_rate": 1.8976545842217487e-05,
      "loss": 1.0555,
      "step": 445
    },
    {
      "epoch": 0.2853486884197057,
      "grad_norm": 4.385538698244002,
      "learning_rate": 1.9019189765458422e-05,
      "loss": 1.0371,
      "step": 446
    },
    {
      "epoch": 0.28598848368522073,
      "grad_norm": 3.3295903569934375,
      "learning_rate": 1.9061833688699363e-05,
      "loss": 1.1217,
      "step": 447
    },
    {
      "epoch": 0.2866282789507358,
      "grad_norm": 3.2953200006914627,
      "learning_rate": 1.9104477611940298e-05,
      "loss": 1.0612,
      "step": 448
    },
    {
      "epoch": 0.2872680742162508,
      "grad_norm": 3.769827730177324,
      "learning_rate": 1.914712153518124e-05,
      "loss": 1.0977,
      "step": 449
    },
    {
      "epoch": 0.28790786948176583,
      "grad_norm": 3.7529200483995786,
      "learning_rate": 1.9189765458422178e-05,
      "loss": 1.1374,
      "step": 450
    },
    {
      "epoch": 0.2885476647472809,
      "grad_norm": 3.1273842202095326,
      "learning_rate": 1.9232409381663116e-05,
      "loss": 1.0462,
      "step": 451
    },
    {
      "epoch": 0.2891874600127959,
      "grad_norm": 3.306094062079741,
      "learning_rate": 1.9275053304904054e-05,
      "loss": 1.1009,
      "step": 452
    },
    {
      "epoch": 0.28982725527831094,
      "grad_norm": 3.073340483707346,
      "learning_rate": 1.9317697228144992e-05,
      "loss": 1.0892,
      "step": 453
    },
    {
      "epoch": 0.290467050543826,
      "grad_norm": 3.0844141873324564,
      "learning_rate": 1.936034115138593e-05,
      "loss": 1.0768,
      "step": 454
    },
    {
      "epoch": 0.291106845809341,
      "grad_norm": 3.268096885992606,
      "learning_rate": 1.9402985074626868e-05,
      "loss": 1.0596,
      "step": 455
    },
    {
      "epoch": 0.29174664107485604,
      "grad_norm": 3.239431985669449,
      "learning_rate": 1.9445628997867806e-05,
      "loss": 1.0283,
      "step": 456
    },
    {
      "epoch": 0.2923864363403711,
      "grad_norm": 3.0068988865807085,
      "learning_rate": 1.9488272921108744e-05,
      "loss": 1.1031,
      "step": 457
    },
    {
      "epoch": 0.2930262316058861,
      "grad_norm": 2.9194429742840544,
      "learning_rate": 1.9530916844349682e-05,
      "loss": 1.1206,
      "step": 458
    },
    {
      "epoch": 0.29366602687140114,
      "grad_norm": 3.4168618421886117,
      "learning_rate": 1.957356076759062e-05,
      "loss": 1.0946,
      "step": 459
    },
    {
      "epoch": 0.2943058221369162,
      "grad_norm": 3.2553279363702647,
      "learning_rate": 1.961620469083156e-05,
      "loss": 1.0978,
      "step": 460
    },
    {
      "epoch": 0.29494561740243125,
      "grad_norm": 2.9061472486938835,
      "learning_rate": 1.9658848614072497e-05,
      "loss": 1.0984,
      "step": 461
    },
    {
      "epoch": 0.29558541266794625,
      "grad_norm": 3.08441670155791,
      "learning_rate": 1.9701492537313435e-05,
      "loss": 1.0961,
      "step": 462
    },
    {
      "epoch": 0.2962252079334613,
      "grad_norm": 3.6371333727625106,
      "learning_rate": 1.9744136460554373e-05,
      "loss": 1.1231,
      "step": 463
    },
    {
      "epoch": 0.29686500319897635,
      "grad_norm": 3.439151172288207,
      "learning_rate": 1.978678038379531e-05,
      "loss": 1.0697,
      "step": 464
    },
    {
      "epoch": 0.29750479846449135,
      "grad_norm": 3.4482633484455683,
      "learning_rate": 1.982942430703625e-05,
      "loss": 1.1073,
      "step": 465
    },
    {
      "epoch": 0.2981445937300064,
      "grad_norm": 3.1045692798697555,
      "learning_rate": 1.9872068230277187e-05,
      "loss": 1.1463,
      "step": 466
    },
    {
      "epoch": 0.29878438899552145,
      "grad_norm": 2.9737076749506817,
      "learning_rate": 1.9914712153518125e-05,
      "loss": 1.1115,
      "step": 467
    },
    {
      "epoch": 0.29942418426103645,
      "grad_norm": 3.702654553888762,
      "learning_rate": 1.9957356076759064e-05,
      "loss": 1.0613,
      "step": 468
    },
    {
      "epoch": 0.3000639795265515,
      "grad_norm": 3.1202177793707078,
      "learning_rate": 2e-05,
      "loss": 1.1208,
      "step": 469
    },
    {
      "epoch": 0.30070377479206656,
      "grad_norm": 2.8816435262815907,
      "learning_rate": 1.9999999785308896e-05,
      "loss": 1.0851,
      "step": 470
    },
    {
      "epoch": 0.30134357005758156,
      "grad_norm": 3.5390428620748375,
      "learning_rate": 1.9999999141235587e-05,
      "loss": 1.0997,
      "step": 471
    },
    {
      "epoch": 0.3019833653230966,
      "grad_norm": 3.2167268044228097,
      "learning_rate": 1.9999998067780108e-05,
      "loss": 1.1277,
      "step": 472
    },
    {
      "epoch": 0.30262316058861166,
      "grad_norm": 3.646141591464044,
      "learning_rate": 1.9999996564942496e-05,
      "loss": 1.1474,
      "step": 473
    },
    {
      "epoch": 0.30326295585412666,
      "grad_norm": 3.662564808272209,
      "learning_rate": 1.999999463272282e-05,
      "loss": 1.0783,
      "step": 474
    },
    {
      "epoch": 0.3039027511196417,
      "grad_norm": 3.206781157087545,
      "learning_rate": 1.9999992271121165e-05,
      "loss": 1.0976,
      "step": 475
    },
    {
      "epoch": 0.30454254638515676,
      "grad_norm": 2.974518362826251,
      "learning_rate": 1.999998948013763e-05,
      "loss": 1.0252,
      "step": 476
    },
    {
      "epoch": 0.30518234165067176,
      "grad_norm": 3.7296505972520615,
      "learning_rate": 1.9999986259772333e-05,
      "loss": 1.0813,
      "step": 477
    },
    {
      "epoch": 0.3058221369161868,
      "grad_norm": 3.023844903271976,
      "learning_rate": 1.999998261002542e-05,
      "loss": 1.1106,
      "step": 478
    },
    {
      "epoch": 0.30646193218170187,
      "grad_norm": 3.0294456699135313,
      "learning_rate": 1.9999978530897037e-05,
      "loss": 1.0235,
      "step": 479
    },
    {
      "epoch": 0.30710172744721687,
      "grad_norm": 3.2526778850162086,
      "learning_rate": 1.9999974022387366e-05,
      "loss": 1.1466,
      "step": 480
    },
    {
      "epoch": 0.3077415227127319,
      "grad_norm": 3.6459262377936748,
      "learning_rate": 1.99999690844966e-05,
      "loss": 1.1059,
      "step": 481
    },
    {
      "epoch": 0.30838131797824697,
      "grad_norm": 3.4449103792810716,
      "learning_rate": 1.9999963717224952e-05,
      "loss": 1.1143,
      "step": 482
    },
    {
      "epoch": 0.30902111324376197,
      "grad_norm": 3.0195483765926627,
      "learning_rate": 1.9999957920572647e-05,
      "loss": 1.144,
      "step": 483
    },
    {
      "epoch": 0.309660908509277,
      "grad_norm": 3.646386742588589,
      "learning_rate": 1.999995169453994e-05,
      "loss": 1.0662,
      "step": 484
    },
    {
      "epoch": 0.3103007037747921,
      "grad_norm": 3.418551064221019,
      "learning_rate": 1.9999945039127092e-05,
      "loss": 1.0655,
      "step": 485
    },
    {
      "epoch": 0.31094049904030713,
      "grad_norm": 3.6277177438896664,
      "learning_rate": 1.9999937954334396e-05,
      "loss": 1.1139,
      "step": 486
    },
    {
      "epoch": 0.3115802943058221,
      "grad_norm": 3.124736613535684,
      "learning_rate": 1.9999930440162152e-05,
      "loss": 1.1246,
      "step": 487
    },
    {
      "epoch": 0.3122200895713372,
      "grad_norm": 3.469107928697995,
      "learning_rate": 1.9999922496610682e-05,
      "loss": 1.1255,
      "step": 488
    },
    {
      "epoch": 0.31285988483685223,
      "grad_norm": 4.227154268699066,
      "learning_rate": 1.9999914123680332e-05,
      "loss": 1.1257,
      "step": 489
    },
    {
      "epoch": 0.31349968010236723,
      "grad_norm": 3.8172130536964985,
      "learning_rate": 1.9999905321371454e-05,
      "loss": 1.0876,
      "step": 490
    },
    {
      "epoch": 0.3141394753678823,
      "grad_norm": 3.2542631268451836,
      "learning_rate": 1.999989608968443e-05,
      "loss": 1.1249,
      "step": 491
    },
    {
      "epoch": 0.31477927063339733,
      "grad_norm": 2.735841882355952,
      "learning_rate": 1.999988642861966e-05,
      "loss": 1.06,
      "step": 492
    },
    {
      "epoch": 0.31541906589891233,
      "grad_norm": 2.865675296321522,
      "learning_rate": 1.9999876338177554e-05,
      "loss": 1.0381,
      "step": 493
    },
    {
      "epoch": 0.3160588611644274,
      "grad_norm": 3.158056917359267,
      "learning_rate": 1.9999865818358547e-05,
      "loss": 1.0678,
      "step": 494
    },
    {
      "epoch": 0.31669865642994244,
      "grad_norm": 3.1346907267178836,
      "learning_rate": 1.9999854869163088e-05,
      "loss": 1.0028,
      "step": 495
    },
    {
      "epoch": 0.31733845169545744,
      "grad_norm": 3.0256958174733004,
      "learning_rate": 1.999984349059165e-05,
      "loss": 1.1063,
      "step": 496
    },
    {
      "epoch": 0.3179782469609725,
      "grad_norm": 3.2765704124867563,
      "learning_rate": 1.999983168264472e-05,
      "loss": 1.0841,
      "step": 497
    },
    {
      "epoch": 0.31861804222648754,
      "grad_norm": 2.7372166693492828,
      "learning_rate": 1.9999819445322805e-05,
      "loss": 1.0888,
      "step": 498
    },
    {
      "epoch": 0.31925783749200254,
      "grad_norm": 2.909691271498469,
      "learning_rate": 1.9999806778626435e-05,
      "loss": 1.0629,
      "step": 499
    },
    {
      "epoch": 0.3198976327575176,
      "grad_norm": 2.9996852749887584,
      "learning_rate": 1.999979368255615e-05,
      "loss": 1.0754,
      "step": 500
    },
    {
      "epoch": 0.32053742802303264,
      "grad_norm": 3.2921229579160696,
      "learning_rate": 1.9999780157112506e-05,
      "loss": 1.0988,
      "step": 501
    },
    {
      "epoch": 0.32117722328854764,
      "grad_norm": 2.9189441151418385,
      "learning_rate": 1.9999766202296098e-05,
      "loss": 1.0361,
      "step": 502
    },
    {
      "epoch": 0.3218170185540627,
      "grad_norm": 2.9130406133199904,
      "learning_rate": 1.9999751818107515e-05,
      "loss": 1.0436,
      "step": 503
    },
    {
      "epoch": 0.32245681381957775,
      "grad_norm": 3.8292409176661133,
      "learning_rate": 1.9999737004547377e-05,
      "loss": 1.0521,
      "step": 504
    },
    {
      "epoch": 0.32309660908509275,
      "grad_norm": 3.168002980495362,
      "learning_rate": 1.9999721761616317e-05,
      "loss": 1.0693,
      "step": 505
    },
    {
      "epoch": 0.3237364043506078,
      "grad_norm": 2.9400725159288204,
      "learning_rate": 1.9999706089314997e-05,
      "loss": 1.03,
      "step": 506
    },
    {
      "epoch": 0.32437619961612285,
      "grad_norm": 3.2873814904717236,
      "learning_rate": 1.9999689987644083e-05,
      "loss": 1.0786,
      "step": 507
    },
    {
      "epoch": 0.32501599488163785,
      "grad_norm": 3.6886766804443787,
      "learning_rate": 1.9999673456604268e-05,
      "loss": 1.075,
      "step": 508
    },
    {
      "epoch": 0.3256557901471529,
      "grad_norm": 3.463814812016267,
      "learning_rate": 1.999965649619627e-05,
      "loss": 1.054,
      "step": 509
    },
    {
      "epoch": 0.32629558541266795,
      "grad_norm": 2.4471713921832827,
      "learning_rate": 1.9999639106420804e-05,
      "loss": 1.1276,
      "step": 510
    },
    {
      "epoch": 0.326935380678183,
      "grad_norm": 3.0073909726184382,
      "learning_rate": 1.9999621287278622e-05,
      "loss": 1.0293,
      "step": 511
    },
    {
      "epoch": 0.327575175943698,
      "grad_norm": 3.2629372837389674,
      "learning_rate": 1.9999603038770493e-05,
      "loss": 0.9993,
      "step": 512
    },
    {
      "epoch": 0.32821497120921306,
      "grad_norm": 4.000542152271147,
      "learning_rate": 1.9999584360897195e-05,
      "loss": 1.1225,
      "step": 513
    },
    {
      "epoch": 0.3288547664747281,
      "grad_norm": 2.5736113983738758,
      "learning_rate": 1.9999565253659532e-05,
      "loss": 1.0428,
      "step": 514
    },
    {
      "epoch": 0.3294945617402431,
      "grad_norm": 2.8298050486090434,
      "learning_rate": 1.9999545717058327e-05,
      "loss": 1.0185,
      "step": 515
    },
    {
      "epoch": 0.33013435700575816,
      "grad_norm": 2.7348805685027484,
      "learning_rate": 1.9999525751094417e-05,
      "loss": 1.0638,
      "step": 516
    },
    {
      "epoch": 0.3307741522712732,
      "grad_norm": 3.6661602196529075,
      "learning_rate": 1.999950535576866e-05,
      "loss": 1.1264,
      "step": 517
    },
    {
      "epoch": 0.3314139475367882,
      "grad_norm": 2.9828233886774043,
      "learning_rate": 1.9999484531081926e-05,
      "loss": 1.0855,
      "step": 518
    },
    {
      "epoch": 0.33205374280230326,
      "grad_norm": 2.7569614893831207,
      "learning_rate": 1.9999463277035115e-05,
      "loss": 1.1262,
      "step": 519
    },
    {
      "epoch": 0.3326935380678183,
      "grad_norm": 2.795490651221398,
      "learning_rate": 1.9999441593629142e-05,
      "loss": 1.0821,
      "step": 520
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 3.2502644782957004,
      "learning_rate": 1.9999419480864933e-05,
      "loss": 1.0774,
      "step": 521
    },
    {
      "epoch": 0.33397312859884837,
      "grad_norm": 3.2759918658487175,
      "learning_rate": 1.999939693874344e-05,
      "loss": 1.1162,
      "step": 522
    },
    {
      "epoch": 0.3346129238643634,
      "grad_norm": 2.6044412257814766,
      "learning_rate": 1.9999373967265628e-05,
      "loss": 1.0801,
      "step": 523
    },
    {
      "epoch": 0.3352527191298784,
      "grad_norm": 3.1642359664101467,
      "learning_rate": 1.9999350566432487e-05,
      "loss": 1.1218,
      "step": 524
    },
    {
      "epoch": 0.33589251439539347,
      "grad_norm": 2.9157262224148055,
      "learning_rate": 1.999932673624502e-05,
      "loss": 1.0795,
      "step": 525
    },
    {
      "epoch": 0.3365323096609085,
      "grad_norm": 3.285932877073364,
      "learning_rate": 1.9999302476704246e-05,
      "loss": 1.0465,
      "step": 526
    },
    {
      "epoch": 0.3371721049264235,
      "grad_norm": 2.4007260170492812,
      "learning_rate": 1.9999277787811216e-05,
      "loss": 1.0669,
      "step": 527
    },
    {
      "epoch": 0.3378119001919386,
      "grad_norm": 2.778405941647659,
      "learning_rate": 1.9999252669566987e-05,
      "loss": 1.0326,
      "step": 528
    },
    {
      "epoch": 0.3384516954574536,
      "grad_norm": 2.9670837888074146,
      "learning_rate": 1.9999227121972633e-05,
      "loss": 1.0994,
      "step": 529
    },
    {
      "epoch": 0.3390914907229686,
      "grad_norm": 2.950465891681033,
      "learning_rate": 1.999920114502925e-05,
      "loss": 1.0694,
      "step": 530
    },
    {
      "epoch": 0.3397312859884837,
      "grad_norm": 2.830624887996146,
      "learning_rate": 1.9999174738737964e-05,
      "loss": 1.0991,
      "step": 531
    },
    {
      "epoch": 0.34037108125399873,
      "grad_norm": 3.0940806013168967,
      "learning_rate": 1.9999147903099896e-05,
      "loss": 1.1168,
      "step": 532
    },
    {
      "epoch": 0.34101087651951373,
      "grad_norm": 2.571665691712046,
      "learning_rate": 1.999912063811621e-05,
      "loss": 1.1155,
      "step": 533
    },
    {
      "epoch": 0.3416506717850288,
      "grad_norm": 2.8705474558134028,
      "learning_rate": 1.9999092943788065e-05,
      "loss": 1.0685,
      "step": 534
    },
    {
      "epoch": 0.34229046705054383,
      "grad_norm": 3.0472231362729767,
      "learning_rate": 1.9999064820116657e-05,
      "loss": 1.0269,
      "step": 535
    },
    {
      "epoch": 0.3429302623160589,
      "grad_norm": 2.9011514038217787,
      "learning_rate": 1.9999036267103195e-05,
      "loss": 1.0664,
      "step": 536
    },
    {
      "epoch": 0.3435700575815739,
      "grad_norm": 2.8408543480926887,
      "learning_rate": 1.99990072847489e-05,
      "loss": 1.0572,
      "step": 537
    },
    {
      "epoch": 0.34420985284708894,
      "grad_norm": 2.69142175052837,
      "learning_rate": 1.9998977873055022e-05,
      "loss": 1.0749,
      "step": 538
    },
    {
      "epoch": 0.344849648112604,
      "grad_norm": 2.922410893748301,
      "learning_rate": 1.9998948032022816e-05,
      "loss": 1.04,
      "step": 539
    },
    {
      "epoch": 0.345489443378119,
      "grad_norm": 2.737870771763212,
      "learning_rate": 1.9998917761653574e-05,
      "loss": 1.0535,
      "step": 540
    },
    {
      "epoch": 0.34612923864363404,
      "grad_norm": 3.7238968922527977,
      "learning_rate": 1.9998887061948585e-05,
      "loss": 1.0424,
      "step": 541
    },
    {
      "epoch": 0.3467690339091491,
      "grad_norm": 3.2052726976468544,
      "learning_rate": 1.999885593290918e-05,
      "loss": 1.0811,
      "step": 542
    },
    {
      "epoch": 0.3474088291746641,
      "grad_norm": 2.9555827997740254,
      "learning_rate": 1.999882437453668e-05,
      "loss": 1.0175,
      "step": 543
    },
    {
      "epoch": 0.34804862444017914,
      "grad_norm": 3.6274868819271835,
      "learning_rate": 1.999879238683245e-05,
      "loss": 1.0789,
      "step": 544
    },
    {
      "epoch": 0.3486884197056942,
      "grad_norm": 3.352041839282272,
      "learning_rate": 1.9998759969797864e-05,
      "loss": 1.0222,
      "step": 545
    },
    {
      "epoch": 0.3493282149712092,
      "grad_norm": 2.9874866073784334,
      "learning_rate": 1.9998727123434306e-05,
      "loss": 1.0673,
      "step": 546
    },
    {
      "epoch": 0.34996801023672425,
      "grad_norm": 3.105787137830318,
      "learning_rate": 1.9998693847743194e-05,
      "loss": 1.0017,
      "step": 547
    },
    {
      "epoch": 0.3506078055022393,
      "grad_norm": 3.3434433168957565,
      "learning_rate": 1.9998660142725956e-05,
      "loss": 1.0368,
      "step": 548
    },
    {
      "epoch": 0.3512476007677543,
      "grad_norm": 3.1536044537963512,
      "learning_rate": 1.999862600838404e-05,
      "loss": 1.0462,
      "step": 549
    },
    {
      "epoch": 0.35188739603326935,
      "grad_norm": 2.9356879163666605,
      "learning_rate": 1.9998591444718905e-05,
      "loss": 1.0595,
      "step": 550
    },
    {
      "epoch": 0.3525271912987844,
      "grad_norm": 2.9785591404319964,
      "learning_rate": 1.999855645173204e-05,
      "loss": 1.0979,
      "step": 551
    },
    {
      "epoch": 0.3531669865642994,
      "grad_norm": 3.1074233970417047,
      "learning_rate": 1.9998521029424948e-05,
      "loss": 1.1461,
      "step": 552
    },
    {
      "epoch": 0.35380678182981445,
      "grad_norm": 2.8914143522447717,
      "learning_rate": 1.999848517779915e-05,
      "loss": 1.1073,
      "step": 553
    },
    {
      "epoch": 0.3544465770953295,
      "grad_norm": 3.205634204874076,
      "learning_rate": 1.999844889685618e-05,
      "loss": 1.0495,
      "step": 554
    },
    {
      "epoch": 0.3550863723608445,
      "grad_norm": 2.899908564720441,
      "learning_rate": 1.9998412186597605e-05,
      "loss": 1.1138,
      "step": 555
    },
    {
      "epoch": 0.35572616762635956,
      "grad_norm": 3.011355032584398,
      "learning_rate": 1.9998375047024996e-05,
      "loss": 1.0644,
      "step": 556
    },
    {
      "epoch": 0.3563659628918746,
      "grad_norm": 2.904911493539732,
      "learning_rate": 1.9998337478139947e-05,
      "loss": 1.1052,
      "step": 557
    },
    {
      "epoch": 0.3570057581573896,
      "grad_norm": 2.902697597682554,
      "learning_rate": 1.9998299479944072e-05,
      "loss": 1.0967,
      "step": 558
    },
    {
      "epoch": 0.35764555342290466,
      "grad_norm": 2.9376138196479964,
      "learning_rate": 1.9998261052439005e-05,
      "loss": 1.0784,
      "step": 559
    },
    {
      "epoch": 0.3582853486884197,
      "grad_norm": 3.1899641893565445,
      "learning_rate": 1.9998222195626394e-05,
      "loss": 1.083,
      "step": 560
    },
    {
      "epoch": 0.35892514395393477,
      "grad_norm": 2.7838001466750066,
      "learning_rate": 1.9998182909507904e-05,
      "loss": 1.0418,
      "step": 561
    },
    {
      "epoch": 0.35956493921944976,
      "grad_norm": 2.73486865950769,
      "learning_rate": 1.9998143194085228e-05,
      "loss": 1.0792,
      "step": 562
    },
    {
      "epoch": 0.3602047344849648,
      "grad_norm": 3.2321933601981607,
      "learning_rate": 1.9998103049360068e-05,
      "loss": 1.1054,
      "step": 563
    },
    {
      "epoch": 0.36084452975047987,
      "grad_norm": 2.900968328643346,
      "learning_rate": 1.9998062475334148e-05,
      "loss": 1.0881,
      "step": 564
    },
    {
      "epoch": 0.36148432501599487,
      "grad_norm": 2.899654911818964,
      "learning_rate": 1.999802147200921e-05,
      "loss": 1.1064,
      "step": 565
    },
    {
      "epoch": 0.3621241202815099,
      "grad_norm": 3.226468299020007,
      "learning_rate": 1.9997980039387014e-05,
      "loss": 1.032,
      "step": 566
    },
    {
      "epoch": 0.362763915547025,
      "grad_norm": 3.1371049168147245,
      "learning_rate": 1.999793817746934e-05,
      "loss": 1.0954,
      "step": 567
    },
    {
      "epoch": 0.36340371081253997,
      "grad_norm": 3.0336542406731204,
      "learning_rate": 1.999789588625799e-05,
      "loss": 1.0502,
      "step": 568
    },
    {
      "epoch": 0.364043506078055,
      "grad_norm": 3.059076570429832,
      "learning_rate": 1.999785316575477e-05,
      "loss": 1.0499,
      "step": 569
    },
    {
      "epoch": 0.3646833013435701,
      "grad_norm": 3.2987152815123206,
      "learning_rate": 1.9997810015961524e-05,
      "loss": 1.0878,
      "step": 570
    },
    {
      "epoch": 0.3653230966090851,
      "grad_norm": 3.142419444403014,
      "learning_rate": 1.99977664368801e-05,
      "loss": 1.0314,
      "step": 571
    },
    {
      "epoch": 0.3659628918746001,
      "grad_norm": 3.265498630368467,
      "learning_rate": 1.999772242851237e-05,
      "loss": 1.0404,
      "step": 572
    },
    {
      "epoch": 0.3666026871401152,
      "grad_norm": 3.479714556091243,
      "learning_rate": 1.9997677990860223e-05,
      "loss": 1.0557,
      "step": 573
    },
    {
      "epoch": 0.3672424824056302,
      "grad_norm": 3.2402953238248826,
      "learning_rate": 1.9997633123925565e-05,
      "loss": 1.0827,
      "step": 574
    },
    {
      "epoch": 0.36788227767114523,
      "grad_norm": 3.2750265116200157,
      "learning_rate": 1.999758782771033e-05,
      "loss": 1.0848,
      "step": 575
    },
    {
      "epoch": 0.3685220729366603,
      "grad_norm": 3.7420925705549517,
      "learning_rate": 1.9997542102216454e-05,
      "loss": 1.0308,
      "step": 576
    },
    {
      "epoch": 0.3691618682021753,
      "grad_norm": 3.0857995617929093,
      "learning_rate": 1.9997495947445907e-05,
      "loss": 1.0431,
      "step": 577
    },
    {
      "epoch": 0.36980166346769033,
      "grad_norm": 3.247387104894026,
      "learning_rate": 1.999744936340067e-05,
      "loss": 1.0437,
      "step": 578
    },
    {
      "epoch": 0.3704414587332054,
      "grad_norm": 2.7460622116999427,
      "learning_rate": 1.9997402350082737e-05,
      "loss": 1.106,
      "step": 579
    },
    {
      "epoch": 0.3710812539987204,
      "grad_norm": 2.8044796368777427,
      "learning_rate": 1.9997354907494132e-05,
      "loss": 1.0765,
      "step": 580
    },
    {
      "epoch": 0.37172104926423544,
      "grad_norm": 3.202430542791296,
      "learning_rate": 1.9997307035636887e-05,
      "loss": 1.0705,
      "step": 581
    },
    {
      "epoch": 0.3723608445297505,
      "grad_norm": 3.4329379609278274,
      "learning_rate": 1.9997258734513068e-05,
      "loss": 1.0369,
      "step": 582
    },
    {
      "epoch": 0.3730006397952655,
      "grad_norm": 3.274268465509467,
      "learning_rate": 1.9997210004124742e-05,
      "loss": 1.057,
      "step": 583
    },
    {
      "epoch": 0.37364043506078054,
      "grad_norm": 3.0120223984549597,
      "learning_rate": 1.9997160844474e-05,
      "loss": 1.0968,
      "step": 584
    },
    {
      "epoch": 0.3742802303262956,
      "grad_norm": 3.504140907533389,
      "learning_rate": 1.9997111255562958e-05,
      "loss": 1.0548,
      "step": 585
    },
    {
      "epoch": 0.37492002559181065,
      "grad_norm": 3.421826939894412,
      "learning_rate": 1.999706123739374e-05,
      "loss": 1.0908,
      "step": 586
    },
    {
      "epoch": 0.37555982085732564,
      "grad_norm": 2.8995900097608485,
      "learning_rate": 1.9997010789968493e-05,
      "loss": 1.1008,
      "step": 587
    },
    {
      "epoch": 0.3761996161228407,
      "grad_norm": 3.0636423414702216,
      "learning_rate": 1.9996959913289392e-05,
      "loss": 1.0338,
      "step": 588
    },
    {
      "epoch": 0.37683941138835575,
      "grad_norm": 3.354681952816075,
      "learning_rate": 1.9996908607358612e-05,
      "loss": 1.0645,
      "step": 589
    },
    {
      "epoch": 0.37747920665387075,
      "grad_norm": 2.7575517018350215,
      "learning_rate": 1.9996856872178358e-05,
      "loss": 1.101,
      "step": 590
    },
    {
      "epoch": 0.3781190019193858,
      "grad_norm": 2.670543232284009,
      "learning_rate": 1.9996804707750853e-05,
      "loss": 1.1242,
      "step": 591
    },
    {
      "epoch": 0.37875879718490085,
      "grad_norm": 2.766263365019808,
      "learning_rate": 1.9996752114078343e-05,
      "loss": 1.0515,
      "step": 592
    },
    {
      "epoch": 0.37939859245041585,
      "grad_norm": 2.9036281017586525,
      "learning_rate": 1.9996699091163075e-05,
      "loss": 1.0387,
      "step": 593
    },
    {
      "epoch": 0.3800383877159309,
      "grad_norm": 2.6673009804505825,
      "learning_rate": 1.999664563900733e-05,
      "loss": 1.094,
      "step": 594
    },
    {
      "epoch": 0.38067818298144596,
      "grad_norm": 2.7218791340876463,
      "learning_rate": 1.9996591757613407e-05,
      "loss": 1.0639,
      "step": 595
    },
    {
      "epoch": 0.38131797824696095,
      "grad_norm": 2.553975935607351,
      "learning_rate": 1.999653744698361e-05,
      "loss": 1.0586,
      "step": 596
    },
    {
      "epoch": 0.381957773512476,
      "grad_norm": 2.744196277161132,
      "learning_rate": 1.9996482707120287e-05,
      "loss": 1.1128,
      "step": 597
    },
    {
      "epoch": 0.38259756877799106,
      "grad_norm": 2.887453997817572,
      "learning_rate": 1.9996427538025775e-05,
      "loss": 1.0716,
      "step": 598
    },
    {
      "epoch": 0.38323736404350606,
      "grad_norm": 2.788531848265296,
      "learning_rate": 1.9996371939702442e-05,
      "loss": 1.0572,
      "step": 599
    },
    {
      "epoch": 0.3838771593090211,
      "grad_norm": 2.477328503446145,
      "learning_rate": 1.999631591215269e-05,
      "loss": 1.0892,
      "step": 600
    },
    {
      "epoch": 0.38451695457453616,
      "grad_norm": 3.6500141136373614,
      "learning_rate": 1.9996259455378907e-05,
      "loss": 1.0486,
      "step": 601
    },
    {
      "epoch": 0.38515674984005116,
      "grad_norm": 2.9343440952646827,
      "learning_rate": 1.999620256938353e-05,
      "loss": 1.0055,
      "step": 602
    },
    {
      "epoch": 0.3857965451055662,
      "grad_norm": 2.624101261284732,
      "learning_rate": 1.999614525416899e-05,
      "loss": 1.052,
      "step": 603
    },
    {
      "epoch": 0.38643634037108127,
      "grad_norm": 2.8109122147799814,
      "learning_rate": 1.9996087509737757e-05,
      "loss": 1.0874,
      "step": 604
    },
    {
      "epoch": 0.38707613563659626,
      "grad_norm": 2.788609738775846,
      "learning_rate": 1.9996029336092308e-05,
      "loss": 1.0082,
      "step": 605
    },
    {
      "epoch": 0.3877159309021113,
      "grad_norm": 2.3113035831147917,
      "learning_rate": 1.9995970733235142e-05,
      "loss": 1.0943,
      "step": 606
    },
    {
      "epoch": 0.38835572616762637,
      "grad_norm": 2.954087631282599,
      "learning_rate": 1.9995911701168774e-05,
      "loss": 1.0674,
      "step": 607
    },
    {
      "epoch": 0.3889955214331414,
      "grad_norm": 3.094577724351017,
      "learning_rate": 1.999585223989574e-05,
      "loss": 1.0547,
      "step": 608
    },
    {
      "epoch": 0.3896353166986564,
      "grad_norm": 3.5336279472100913,
      "learning_rate": 1.999579234941859e-05,
      "loss": 1.0432,
      "step": 609
    },
    {
      "epoch": 0.3902751119641715,
      "grad_norm": 2.7700306487054376,
      "learning_rate": 1.99957320297399e-05,
      "loss": 1.084,
      "step": 610
    },
    {
      "epoch": 0.3909149072296865,
      "grad_norm": 2.5914800059215106,
      "learning_rate": 1.9995671280862257e-05,
      "loss": 1.1234,
      "step": 611
    },
    {
      "epoch": 0.3915547024952015,
      "grad_norm": 3.9977885378873257,
      "learning_rate": 1.9995610102788268e-05,
      "loss": 1.0388,
      "step": 612
    },
    {
      "epoch": 0.3921944977607166,
      "grad_norm": 3.213120502772563,
      "learning_rate": 1.9995548495520566e-05,
      "loss": 1.0866,
      "step": 613
    },
    {
      "epoch": 0.39283429302623163,
      "grad_norm": 2.9992114357916497,
      "learning_rate": 1.9995486459061788e-05,
      "loss": 1.0941,
      "step": 614
    },
    {
      "epoch": 0.3934740882917466,
      "grad_norm": 2.919657756951116,
      "learning_rate": 1.9995423993414604e-05,
      "loss": 1.0532,
      "step": 615
    },
    {
      "epoch": 0.3941138835572617,
      "grad_norm": 3.3472469872703328,
      "learning_rate": 1.9995361098581694e-05,
      "loss": 1.1198,
      "step": 616
    },
    {
      "epoch": 0.39475367882277673,
      "grad_norm": 2.69126766565489,
      "learning_rate": 1.999529777456576e-05,
      "loss": 1.0116,
      "step": 617
    },
    {
      "epoch": 0.39539347408829173,
      "grad_norm": 2.88701248603357,
      "learning_rate": 1.9995234021369517e-05,
      "loss": 1.0151,
      "step": 618
    },
    {
      "epoch": 0.3960332693538068,
      "grad_norm": 2.9967664806245833,
      "learning_rate": 1.9995169838995706e-05,
      "loss": 1.0611,
      "step": 619
    },
    {
      "epoch": 0.39667306461932184,
      "grad_norm": 2.9106601133265455,
      "learning_rate": 1.9995105227447084e-05,
      "loss": 1.0549,
      "step": 620
    },
    {
      "epoch": 0.39731285988483683,
      "grad_norm": 2.971037515479928,
      "learning_rate": 1.999504018672642e-05,
      "loss": 1.1149,
      "step": 621
    },
    {
      "epoch": 0.3979526551503519,
      "grad_norm": 2.824334426506958,
      "learning_rate": 1.9994974716836516e-05,
      "loss": 1.02,
      "step": 622
    },
    {
      "epoch": 0.39859245041586694,
      "grad_norm": 3.1898207489628527,
      "learning_rate": 1.9994908817780176e-05,
      "loss": 1.0176,
      "step": 623
    },
    {
      "epoch": 0.39923224568138194,
      "grad_norm": 2.896806682217505,
      "learning_rate": 1.9994842489560227e-05,
      "loss": 0.9982,
      "step": 624
    },
    {
      "epoch": 0.399872040946897,
      "grad_norm": 2.7169876839555425,
      "learning_rate": 1.999477573217952e-05,
      "loss": 1.0589,
      "step": 625
    },
    {
      "epoch": 0.40051183621241204,
      "grad_norm": 2.6997764181181956,
      "learning_rate": 1.9994708545640927e-05,
      "loss": 0.9967,
      "step": 626
    },
    {
      "epoch": 0.40115163147792704,
      "grad_norm": 3.0129705983951145,
      "learning_rate": 1.9994640929947328e-05,
      "loss": 1.0739,
      "step": 627
    },
    {
      "epoch": 0.4017914267434421,
      "grad_norm": 2.3546682283254845,
      "learning_rate": 1.9994572885101623e-05,
      "loss": 1.0058,
      "step": 628
    },
    {
      "epoch": 0.40243122200895715,
      "grad_norm": 2.7840302637280536,
      "learning_rate": 1.999450441110674e-05,
      "loss": 1.0885,
      "step": 629
    },
    {
      "epoch": 0.40307101727447214,
      "grad_norm": 3.5293849030698974,
      "learning_rate": 1.9994435507965616e-05,
      "loss": 1.0858,
      "step": 630
    },
    {
      "epoch": 0.4037108125399872,
      "grad_norm": 2.9729105281521706,
      "learning_rate": 1.999436617568121e-05,
      "loss": 0.9706,
      "step": 631
    },
    {
      "epoch": 0.40435060780550225,
      "grad_norm": 3.4554621275108452,
      "learning_rate": 1.9994296414256498e-05,
      "loss": 1.0532,
      "step": 632
    },
    {
      "epoch": 0.4049904030710173,
      "grad_norm": 3.0274572346626405,
      "learning_rate": 1.9994226223694476e-05,
      "loss": 1.1168,
      "step": 633
    },
    {
      "epoch": 0.4056301983365323,
      "grad_norm": 3.705146363171472,
      "learning_rate": 1.999415560399816e-05,
      "loss": 1.0512,
      "step": 634
    },
    {
      "epoch": 0.40626999360204735,
      "grad_norm": 3.214201222054201,
      "learning_rate": 1.9994084555170583e-05,
      "loss": 1.0296,
      "step": 635
    },
    {
      "epoch": 0.4069097888675624,
      "grad_norm": 2.9136813356200126,
      "learning_rate": 1.999401307721479e-05,
      "loss": 1.0775,
      "step": 636
    },
    {
      "epoch": 0.4075495841330774,
      "grad_norm": 3.2701891828700105,
      "learning_rate": 1.9993941170133856e-05,
      "loss": 1.0568,
      "step": 637
    },
    {
      "epoch": 0.40818937939859246,
      "grad_norm": 3.689611986653254,
      "learning_rate": 1.9993868833930863e-05,
      "loss": 1.0529,
      "step": 638
    },
    {
      "epoch": 0.4088291746641075,
      "grad_norm": 2.9936150338743848,
      "learning_rate": 1.999379606860892e-05,
      "loss": 1.0866,
      "step": 639
    },
    {
      "epoch": 0.4094689699296225,
      "grad_norm": 2.946534081517707,
      "learning_rate": 1.9993722874171153e-05,
      "loss": 1.0625,
      "step": 640
    },
    {
      "epoch": 0.41010876519513756,
      "grad_norm": 2.9486543440980704,
      "learning_rate": 1.99936492506207e-05,
      "loss": 1.0359,
      "step": 641
    },
    {
      "epoch": 0.4107485604606526,
      "grad_norm": 3.8508920019855686,
      "learning_rate": 1.9993575197960727e-05,
      "loss": 1.0783,
      "step": 642
    },
    {
      "epoch": 0.4113883557261676,
      "grad_norm": 3.105259950133739,
      "learning_rate": 1.9993500716194415e-05,
      "loss": 0.9675,
      "step": 643
    },
    {
      "epoch": 0.41202815099168266,
      "grad_norm": 3.0445955737654993,
      "learning_rate": 1.9993425805324956e-05,
      "loss": 1.1176,
      "step": 644
    },
    {
      "epoch": 0.4126679462571977,
      "grad_norm": 3.286504905355694,
      "learning_rate": 1.9993350465355572e-05,
      "loss": 1.0446,
      "step": 645
    },
    {
      "epoch": 0.4133077415227127,
      "grad_norm": 4.325127562414131,
      "learning_rate": 1.9993274696289498e-05,
      "loss": 1.0178,
      "step": 646
    },
    {
      "epoch": 0.41394753678822777,
      "grad_norm": 4.0497672647536085,
      "learning_rate": 1.999319849812998e-05,
      "loss": 0.9602,
      "step": 647
    },
    {
      "epoch": 0.4145873320537428,
      "grad_norm": 2.7107519636544004,
      "learning_rate": 1.99931218708803e-05,
      "loss": 1.0548,
      "step": 648
    },
    {
      "epoch": 0.4152271273192578,
      "grad_norm": 3.5765122422739384,
      "learning_rate": 1.999304481454374e-05,
      "loss": 1.0359,
      "step": 649
    },
    {
      "epoch": 0.41586692258477287,
      "grad_norm": 3.5041934641360966,
      "learning_rate": 1.9992967329123613e-05,
      "loss": 1.0565,
      "step": 650
    },
    {
      "epoch": 0.4165067178502879,
      "grad_norm": 3.0797651632554373,
      "learning_rate": 1.9992889414623244e-05,
      "loss": 1.0377,
      "step": 651
    },
    {
      "epoch": 0.4171465131158029,
      "grad_norm": 2.4096394525535283,
      "learning_rate": 1.999281107104598e-05,
      "loss": 1.0034,
      "step": 652
    },
    {
      "epoch": 0.417786308381318,
      "grad_norm": 3.325456608729587,
      "learning_rate": 1.9992732298395182e-05,
      "loss": 1.0587,
      "step": 653
    },
    {
      "epoch": 0.418426103646833,
      "grad_norm": 2.9493719303547357,
      "learning_rate": 1.9992653096674236e-05,
      "loss": 1.0183,
      "step": 654
    },
    {
      "epoch": 0.419065898912348,
      "grad_norm": 3.1000933822191614,
      "learning_rate": 1.9992573465886544e-05,
      "loss": 1.0273,
      "step": 655
    },
    {
      "epoch": 0.4197056941778631,
      "grad_norm": 2.6365825769198943,
      "learning_rate": 1.9992493406035522e-05,
      "loss": 1.0695,
      "step": 656
    },
    {
      "epoch": 0.42034548944337813,
      "grad_norm": 2.9704123527297135,
      "learning_rate": 1.999241291712461e-05,
      "loss": 1.031,
      "step": 657
    },
    {
      "epoch": 0.4209852847088932,
      "grad_norm": 2.7518458408470097,
      "learning_rate": 1.999233199915726e-05,
      "loss": 1.0931,
      "step": 658
    },
    {
      "epoch": 0.4216250799744082,
      "grad_norm": 3.0126301046703814,
      "learning_rate": 1.999225065213695e-05,
      "loss": 1.0121,
      "step": 659
    },
    {
      "epoch": 0.42226487523992323,
      "grad_norm": 2.723219411676696,
      "learning_rate": 1.9992168876067167e-05,
      "loss": 1.0475,
      "step": 660
    },
    {
      "epoch": 0.4229046705054383,
      "grad_norm": 3.103576229812221,
      "learning_rate": 1.9992086670951436e-05,
      "loss": 1.0046,
      "step": 661
    },
    {
      "epoch": 0.4235444657709533,
      "grad_norm": 2.609681195524442,
      "learning_rate": 1.999200403679327e-05,
      "loss": 1.0786,
      "step": 662
    },
    {
      "epoch": 0.42418426103646834,
      "grad_norm": 2.622943844079557,
      "learning_rate": 1.999192097359623e-05,
      "loss": 1.0695,
      "step": 663
    },
    {
      "epoch": 0.4248240563019834,
      "grad_norm": 2.7949127234569193,
      "learning_rate": 1.9991837481363876e-05,
      "loss": 1.0363,
      "step": 664
    },
    {
      "epoch": 0.4254638515674984,
      "grad_norm": 3.1713724994006474,
      "learning_rate": 1.9991753560099796e-05,
      "loss": 1.0221,
      "step": 665
    },
    {
      "epoch": 0.42610364683301344,
      "grad_norm": 2.853228510646763,
      "learning_rate": 1.999166920980759e-05,
      "loss": 1.0605,
      "step": 666
    },
    {
      "epoch": 0.4267434420985285,
      "grad_norm": 3.10216129621452,
      "learning_rate": 1.999158443049088e-05,
      "loss": 1.0302,
      "step": 667
    },
    {
      "epoch": 0.4273832373640435,
      "grad_norm": 3.027200944403017,
      "learning_rate": 1.9991499222153314e-05,
      "loss": 1.076,
      "step": 668
    },
    {
      "epoch": 0.42802303262955854,
      "grad_norm": 2.7409827028553715,
      "learning_rate": 1.9991413584798542e-05,
      "loss": 1.0722,
      "step": 669
    },
    {
      "epoch": 0.4286628278950736,
      "grad_norm": 3.0356997613755006,
      "learning_rate": 1.9991327518430242e-05,
      "loss": 1.0624,
      "step": 670
    },
    {
      "epoch": 0.4293026231605886,
      "grad_norm": 3.2729056384383917,
      "learning_rate": 1.999124102305211e-05,
      "loss": 1.0477,
      "step": 671
    },
    {
      "epoch": 0.42994241842610365,
      "grad_norm": 2.7642938468601654,
      "learning_rate": 1.999115409866787e-05,
      "loss": 1.1289,
      "step": 672
    },
    {
      "epoch": 0.4305822136916187,
      "grad_norm": 2.671847147936606,
      "learning_rate": 1.999106674528124e-05,
      "loss": 1.0402,
      "step": 673
    },
    {
      "epoch": 0.4312220089571337,
      "grad_norm": 3.423851876013969,
      "learning_rate": 1.9990978962895977e-05,
      "loss": 1.0078,
      "step": 674
    },
    {
      "epoch": 0.43186180422264875,
      "grad_norm": 3.462399531265576,
      "learning_rate": 1.999089075151585e-05,
      "loss": 1.0701,
      "step": 675
    },
    {
      "epoch": 0.4325015994881638,
      "grad_norm": 3.1132006984401097,
      "learning_rate": 1.999080211114465e-05,
      "loss": 1.0539,
      "step": 676
    },
    {
      "epoch": 0.4331413947536788,
      "grad_norm": 2.967100039906888,
      "learning_rate": 1.9990713041786175e-05,
      "loss": 1.049,
      "step": 677
    },
    {
      "epoch": 0.43378119001919385,
      "grad_norm": 3.1405805759359033,
      "learning_rate": 1.9990623543444254e-05,
      "loss": 1.0166,
      "step": 678
    },
    {
      "epoch": 0.4344209852847089,
      "grad_norm": 2.9446344623651073,
      "learning_rate": 1.999053361612273e-05,
      "loss": 1.0206,
      "step": 679
    },
    {
      "epoch": 0.4350607805502239,
      "grad_norm": 3.1064044637248136,
      "learning_rate": 1.9990443259825468e-05,
      "loss": 1.0096,
      "step": 680
    },
    {
      "epoch": 0.43570057581573896,
      "grad_norm": 2.9475536290749735,
      "learning_rate": 1.999035247455634e-05,
      "loss": 1.1144,
      "step": 681
    },
    {
      "epoch": 0.436340371081254,
      "grad_norm": 3.183240365908259,
      "learning_rate": 1.9990261260319253e-05,
      "loss": 1.0501,
      "step": 682
    },
    {
      "epoch": 0.43698016634676906,
      "grad_norm": 3.4545621113516662,
      "learning_rate": 1.9990169617118114e-05,
      "loss": 1.0567,
      "step": 683
    },
    {
      "epoch": 0.43761996161228406,
      "grad_norm": 2.5909609139655085,
      "learning_rate": 1.9990077544956865e-05,
      "loss": 1.0714,
      "step": 684
    },
    {
      "epoch": 0.4382597568777991,
      "grad_norm": 3.582201619915801,
      "learning_rate": 1.9989985043839456e-05,
      "loss": 1.0741,
      "step": 685
    },
    {
      "epoch": 0.43889955214331416,
      "grad_norm": 3.242901620046091,
      "learning_rate": 1.9989892113769862e-05,
      "loss": 0.9875,
      "step": 686
    },
    {
      "epoch": 0.43953934740882916,
      "grad_norm": 3.039257011236719,
      "learning_rate": 1.998979875475207e-05,
      "loss": 1.0633,
      "step": 687
    },
    {
      "epoch": 0.4401791426743442,
      "grad_norm": 3.001915365902321,
      "learning_rate": 1.998970496679009e-05,
      "loss": 1.0055,
      "step": 688
    },
    {
      "epoch": 0.44081893793985927,
      "grad_norm": 3.4402303318214233,
      "learning_rate": 1.9989610749887948e-05,
      "loss": 1.0718,
      "step": 689
    },
    {
      "epoch": 0.44145873320537427,
      "grad_norm": 3.480633862392414,
      "learning_rate": 1.9989516104049693e-05,
      "loss": 1.0696,
      "step": 690
    },
    {
      "epoch": 0.4420985284708893,
      "grad_norm": 2.8060313862639474,
      "learning_rate": 1.9989421029279384e-05,
      "loss": 1.0219,
      "step": 691
    },
    {
      "epoch": 0.44273832373640437,
      "grad_norm": 2.902787076108138,
      "learning_rate": 1.9989325525581108e-05,
      "loss": 1.0323,
      "step": 692
    },
    {
      "epoch": 0.44337811900191937,
      "grad_norm": 2.783367380601412,
      "learning_rate": 1.9989229592958965e-05,
      "loss": 1.0553,
      "step": 693
    },
    {
      "epoch": 0.4440179142674344,
      "grad_norm": 3.470032041722473,
      "learning_rate": 1.9989133231417067e-05,
      "loss": 1.0604,
      "step": 694
    },
    {
      "epoch": 0.4446577095329495,
      "grad_norm": 2.8800140349549763,
      "learning_rate": 1.9989036440959562e-05,
      "loss": 0.9989,
      "step": 695
    },
    {
      "epoch": 0.44529750479846447,
      "grad_norm": 2.9673369776152745,
      "learning_rate": 1.99889392215906e-05,
      "loss": 1.0572,
      "step": 696
    },
    {
      "epoch": 0.4459373000639795,
      "grad_norm": 2.5801370925003337,
      "learning_rate": 1.9988841573314355e-05,
      "loss": 1.0456,
      "step": 697
    },
    {
      "epoch": 0.4465770953294946,
      "grad_norm": 3.147900055789132,
      "learning_rate": 1.9988743496135025e-05,
      "loss": 1.0097,
      "step": 698
    },
    {
      "epoch": 0.4472168905950096,
      "grad_norm": 2.5680716940215067,
      "learning_rate": 1.9988644990056816e-05,
      "loss": 1.0328,
      "step": 699
    },
    {
      "epoch": 0.44785668586052463,
      "grad_norm": 2.604581334784409,
      "learning_rate": 1.9988546055083958e-05,
      "loss": 1.0206,
      "step": 700
    },
    {
      "epoch": 0.4484964811260397,
      "grad_norm": 2.9507084488215845,
      "learning_rate": 1.9988446691220704e-05,
      "loss": 1.0504,
      "step": 701
    },
    {
      "epoch": 0.4491362763915547,
      "grad_norm": 2.8314917036270564,
      "learning_rate": 1.9988346898471316e-05,
      "loss": 1.0757,
      "step": 702
    },
    {
      "epoch": 0.44977607165706973,
      "grad_norm": 2.9524933260609254,
      "learning_rate": 1.9988246676840076e-05,
      "loss": 1.0378,
      "step": 703
    },
    {
      "epoch": 0.4504158669225848,
      "grad_norm": 2.412869974658448,
      "learning_rate": 1.9988146026331295e-05,
      "loss": 1.0292,
      "step": 704
    },
    {
      "epoch": 0.4510556621880998,
      "grad_norm": 2.5302716951555584,
      "learning_rate": 1.9988044946949295e-05,
      "loss": 1.0588,
      "step": 705
    },
    {
      "epoch": 0.45169545745361483,
      "grad_norm": 3.25814229968259,
      "learning_rate": 1.9987943438698407e-05,
      "loss": 1.0062,
      "step": 706
    },
    {
      "epoch": 0.4523352527191299,
      "grad_norm": 3.1808760469099036,
      "learning_rate": 1.9987841501582995e-05,
      "loss": 1.0094,
      "step": 707
    },
    {
      "epoch": 0.45297504798464494,
      "grad_norm": 2.5208917712719443,
      "learning_rate": 1.998773913560744e-05,
      "loss": 1.0662,
      "step": 708
    },
    {
      "epoch": 0.45361484325015994,
      "grad_norm": 2.9821550941739243,
      "learning_rate": 1.998763634077613e-05,
      "loss": 1.0586,
      "step": 709
    },
    {
      "epoch": 0.454254638515675,
      "grad_norm": 2.6500207456031126,
      "learning_rate": 1.9987533117093484e-05,
      "loss": 0.9958,
      "step": 710
    },
    {
      "epoch": 0.45489443378119004,
      "grad_norm": 2.8179456550341233,
      "learning_rate": 1.998742946456393e-05,
      "loss": 1.0612,
      "step": 711
    },
    {
      "epoch": 0.45553422904670504,
      "grad_norm": 2.662155816096431,
      "learning_rate": 1.998732538319192e-05,
      "loss": 1.0236,
      "step": 712
    },
    {
      "epoch": 0.4561740243122201,
      "grad_norm": 3.2085639169542532,
      "learning_rate": 1.9987220872981926e-05,
      "loss": 0.9949,
      "step": 713
    },
    {
      "epoch": 0.45681381957773515,
      "grad_norm": 2.5950436698325343,
      "learning_rate": 1.9987115933938438e-05,
      "loss": 1.0759,
      "step": 714
    },
    {
      "epoch": 0.45745361484325014,
      "grad_norm": 2.7685552517107648,
      "learning_rate": 1.9987010566065953e-05,
      "loss": 1.0566,
      "step": 715
    },
    {
      "epoch": 0.4580934101087652,
      "grad_norm": 2.965229125472413,
      "learning_rate": 1.9986904769369e-05,
      "loss": 1.0758,
      "step": 716
    },
    {
      "epoch": 0.45873320537428025,
      "grad_norm": 2.845146031782788,
      "learning_rate": 1.998679854385212e-05,
      "loss": 0.9865,
      "step": 717
    },
    {
      "epoch": 0.45937300063979525,
      "grad_norm": 2.606398876514617,
      "learning_rate": 1.9986691889519878e-05,
      "loss": 0.9894,
      "step": 718
    },
    {
      "epoch": 0.4600127959053103,
      "grad_norm": 2.9540000578814443,
      "learning_rate": 1.9986584806376852e-05,
      "loss": 1.059,
      "step": 719
    },
    {
      "epoch": 0.46065259117082535,
      "grad_norm": 2.392916643620222,
      "learning_rate": 1.998647729442764e-05,
      "loss": 1.068,
      "step": 720
    },
    {
      "epoch": 0.46129238643634035,
      "grad_norm": 2.649045757102712,
      "learning_rate": 1.998636935367685e-05,
      "loss": 1.0972,
      "step": 721
    },
    {
      "epoch": 0.4619321817018554,
      "grad_norm": 2.8724071425847453,
      "learning_rate": 1.9986260984129132e-05,
      "loss": 1.0056,
      "step": 722
    },
    {
      "epoch": 0.46257197696737046,
      "grad_norm": 3.1989995920565457,
      "learning_rate": 1.998615218578913e-05,
      "loss": 1.0819,
      "step": 723
    },
    {
      "epoch": 0.46321177223288545,
      "grad_norm": 2.489410342726229,
      "learning_rate": 1.9986042958661515e-05,
      "loss": 1.0171,
      "step": 724
    },
    {
      "epoch": 0.4638515674984005,
      "grad_norm": 2.7445866129475127,
      "learning_rate": 1.9985933302750983e-05,
      "loss": 1.0108,
      "step": 725
    },
    {
      "epoch": 0.46449136276391556,
      "grad_norm": 3.1446061171592885,
      "learning_rate": 1.9985823218062236e-05,
      "loss": 1.0492,
      "step": 726
    },
    {
      "epoch": 0.46513115802943056,
      "grad_norm": 3.3504665270994756,
      "learning_rate": 1.99857127046e-05,
      "loss": 1.0146,
      "step": 727
    },
    {
      "epoch": 0.4657709532949456,
      "grad_norm": 2.87215982192055,
      "learning_rate": 1.998560176236903e-05,
      "loss": 1.038,
      "step": 728
    },
    {
      "epoch": 0.46641074856046066,
      "grad_norm": 3.104407239325188,
      "learning_rate": 1.9985490391374077e-05,
      "loss": 1.039,
      "step": 729
    },
    {
      "epoch": 0.46705054382597566,
      "grad_norm": 3.0449483610987076,
      "learning_rate": 1.9985378591619933e-05,
      "loss": 1.0552,
      "step": 730
    },
    {
      "epoch": 0.4676903390914907,
      "grad_norm": 4.256467022802163,
      "learning_rate": 1.9985266363111397e-05,
      "loss": 1.0778,
      "step": 731
    },
    {
      "epoch": 0.46833013435700577,
      "grad_norm": 2.7329334109727936,
      "learning_rate": 1.998515370585328e-05,
      "loss": 1.0394,
      "step": 732
    },
    {
      "epoch": 0.4689699296225208,
      "grad_norm": 2.778294900489492,
      "learning_rate": 1.9985040619850432e-05,
      "loss": 1.0624,
      "step": 733
    },
    {
      "epoch": 0.4696097248880358,
      "grad_norm": 3.283173736800804,
      "learning_rate": 1.9984927105107698e-05,
      "loss": 1.0038,
      "step": 734
    },
    {
      "epoch": 0.47024952015355087,
      "grad_norm": 3.2513362095043483,
      "learning_rate": 1.9984813161629957e-05,
      "loss": 1.0834,
      "step": 735
    },
    {
      "epoch": 0.4708893154190659,
      "grad_norm": 2.8258565082939064,
      "learning_rate": 1.99846987894221e-05,
      "loss": 0.9799,
      "step": 736
    },
    {
      "epoch": 0.4715291106845809,
      "grad_norm": 3.328161259780026,
      "learning_rate": 1.9984583988489037e-05,
      "loss": 1.0668,
      "step": 737
    },
    {
      "epoch": 0.472168905950096,
      "grad_norm": 2.9892010035044674,
      "learning_rate": 1.99844687588357e-05,
      "loss": 1.1046,
      "step": 738
    },
    {
      "epoch": 0.472808701215611,
      "grad_norm": 2.8195416189288505,
      "learning_rate": 1.9984353100467035e-05,
      "loss": 1.048,
      "step": 739
    },
    {
      "epoch": 0.473448496481126,
      "grad_norm": 2.4787646576033966,
      "learning_rate": 1.998423701338801e-05,
      "loss": 1.0389,
      "step": 740
    },
    {
      "epoch": 0.4740882917466411,
      "grad_norm": 2.7428198138478406,
      "learning_rate": 1.998412049760361e-05,
      "loss": 1.0431,
      "step": 741
    },
    {
      "epoch": 0.47472808701215613,
      "grad_norm": 3.0221335496508956,
      "learning_rate": 1.998400355311883e-05,
      "loss": 1.0292,
      "step": 742
    },
    {
      "epoch": 0.4753678822776711,
      "grad_norm": 3.36227395842786,
      "learning_rate": 1.9983886179938704e-05,
      "loss": 1.0136,
      "step": 743
    },
    {
      "epoch": 0.4760076775431862,
      "grad_norm": 2.6894807036004225,
      "learning_rate": 1.998376837806826e-05,
      "loss": 1.055,
      "step": 744
    },
    {
      "epoch": 0.47664747280870123,
      "grad_norm": 2.8735673893651876,
      "learning_rate": 1.9983650147512566e-05,
      "loss": 1.001,
      "step": 745
    },
    {
      "epoch": 0.47728726807421623,
      "grad_norm": 3.58093437407478,
      "learning_rate": 1.998353148827669e-05,
      "loss": 1.042,
      "step": 746
    },
    {
      "epoch": 0.4779270633397313,
      "grad_norm": 2.9287679088641907,
      "learning_rate": 1.9983412400365734e-05,
      "loss": 1.0618,
      "step": 747
    },
    {
      "epoch": 0.47856685860524634,
      "grad_norm": 2.873873905947819,
      "learning_rate": 1.998329288378481e-05,
      "loss": 1.0424,
      "step": 748
    },
    {
      "epoch": 0.47920665387076133,
      "grad_norm": 2.7733785976667615,
      "learning_rate": 1.9983172938539047e-05,
      "loss": 1.0855,
      "step": 749
    },
    {
      "epoch": 0.4798464491362764,
      "grad_norm": 2.6980787255270946,
      "learning_rate": 1.9983052564633593e-05,
      "loss": 1.089,
      "step": 750
    },
    {
      "epoch": 0.48048624440179144,
      "grad_norm": 3.1229232187456035,
      "learning_rate": 1.9982931762073624e-05,
      "loss": 1.0966,
      "step": 751
    },
    {
      "epoch": 0.48112603966730644,
      "grad_norm": 2.7234919533730966,
      "learning_rate": 1.9982810530864322e-05,
      "loss": 1.0276,
      "step": 752
    },
    {
      "epoch": 0.4817658349328215,
      "grad_norm": 2.8729878350804063,
      "learning_rate": 1.9982688871010896e-05,
      "loss": 1.0637,
      "step": 753
    },
    {
      "epoch": 0.48240563019833654,
      "grad_norm": 2.3397558272434797,
      "learning_rate": 1.9982566782518565e-05,
      "loss": 1.0566,
      "step": 754
    },
    {
      "epoch": 0.48304542546385154,
      "grad_norm": 2.408100467975977,
      "learning_rate": 1.9982444265392577e-05,
      "loss": 1.0612,
      "step": 755
    },
    {
      "epoch": 0.4836852207293666,
      "grad_norm": 2.585403426269804,
      "learning_rate": 1.998232131963819e-05,
      "loss": 1.0832,
      "step": 756
    },
    {
      "epoch": 0.48432501599488165,
      "grad_norm": 3.280635120498765,
      "learning_rate": 1.998219794526068e-05,
      "loss": 1.0486,
      "step": 757
    },
    {
      "epoch": 0.4849648112603967,
      "grad_norm": 2.7255067789636884,
      "learning_rate": 1.998207414226535e-05,
      "loss": 1.0182,
      "step": 758
    },
    {
      "epoch": 0.4856046065259117,
      "grad_norm": 3.4614699253176733,
      "learning_rate": 1.998194991065751e-05,
      "loss": 1.0238,
      "step": 759
    },
    {
      "epoch": 0.48624440179142675,
      "grad_norm": 3.819765608922573,
      "learning_rate": 1.9981825250442497e-05,
      "loss": 1.0416,
      "step": 760
    },
    {
      "epoch": 0.4868841970569418,
      "grad_norm": 2.236114532868026,
      "learning_rate": 1.9981700161625666e-05,
      "loss": 0.9786,
      "step": 761
    },
    {
      "epoch": 0.4875239923224568,
      "grad_norm": 2.5738243330741493,
      "learning_rate": 1.9981574644212386e-05,
      "loss": 1.0027,
      "step": 762
    },
    {
      "epoch": 0.48816378758797185,
      "grad_norm": 3.068880190366956,
      "learning_rate": 1.9981448698208045e-05,
      "loss": 1.0208,
      "step": 763
    },
    {
      "epoch": 0.4888035828534869,
      "grad_norm": 3.5948383656680423,
      "learning_rate": 1.9981322323618056e-05,
      "loss": 1.0293,
      "step": 764
    },
    {
      "epoch": 0.4894433781190019,
      "grad_norm": 2.9603248809642353,
      "learning_rate": 1.9981195520447838e-05,
      "loss": 1.0245,
      "step": 765
    },
    {
      "epoch": 0.49008317338451696,
      "grad_norm": 2.539100820687777,
      "learning_rate": 1.9981068288702842e-05,
      "loss": 1.0347,
      "step": 766
    },
    {
      "epoch": 0.490722968650032,
      "grad_norm": 2.301321596303269,
      "learning_rate": 1.9980940628388526e-05,
      "loss": 1.0386,
      "step": 767
    },
    {
      "epoch": 0.491362763915547,
      "grad_norm": 2.565396122548162,
      "learning_rate": 1.9980812539510378e-05,
      "loss": 1.0784,
      "step": 768
    },
    {
      "epoch": 0.49200255918106206,
      "grad_norm": 2.617116229275269,
      "learning_rate": 1.998068402207389e-05,
      "loss": 1.0243,
      "step": 769
    },
    {
      "epoch": 0.4926423544465771,
      "grad_norm": 2.792677234759974,
      "learning_rate": 1.9980555076084585e-05,
      "loss": 1.0143,
      "step": 770
    },
    {
      "epoch": 0.4932821497120921,
      "grad_norm": 2.898510475493918,
      "learning_rate": 1.9980425701548002e-05,
      "loss": 1.0127,
      "step": 771
    },
    {
      "epoch": 0.49392194497760716,
      "grad_norm": 2.6626569254492174,
      "learning_rate": 1.998029589846969e-05,
      "loss": 1.0423,
      "step": 772
    },
    {
      "epoch": 0.4945617402431222,
      "grad_norm": 2.8184249853693197,
      "learning_rate": 1.998016566685523e-05,
      "loss": 1.0797,
      "step": 773
    },
    {
      "epoch": 0.4952015355086372,
      "grad_norm": 2.992558248118656,
      "learning_rate": 1.9980035006710206e-05,
      "loss": 1.0258,
      "step": 774
    },
    {
      "epoch": 0.49584133077415227,
      "grad_norm": 2.7290951459677144,
      "learning_rate": 1.9979903918040234e-05,
      "loss": 1.0303,
      "step": 775
    },
    {
      "epoch": 0.4964811260396673,
      "grad_norm": 2.761133312091553,
      "learning_rate": 1.9979772400850937e-05,
      "loss": 1.0446,
      "step": 776
    },
    {
      "epoch": 0.4971209213051823,
      "grad_norm": 2.7810590612965203,
      "learning_rate": 1.997964045514797e-05,
      "loss": 1.0233,
      "step": 777
    },
    {
      "epoch": 0.49776071657069737,
      "grad_norm": 2.7574108022138546,
      "learning_rate": 1.9979508080936992e-05,
      "loss": 1.0465,
      "step": 778
    },
    {
      "epoch": 0.4984005118362124,
      "grad_norm": 2.878776591445976,
      "learning_rate": 1.9979375278223688e-05,
      "loss": 1.0712,
      "step": 779
    },
    {
      "epoch": 0.4990403071017274,
      "grad_norm": 2.9563980049929124,
      "learning_rate": 1.9979242047013762e-05,
      "loss": 0.9932,
      "step": 780
    },
    {
      "epoch": 0.4996801023672425,
      "grad_norm": 3.4702475232158796,
      "learning_rate": 1.997910838731294e-05,
      "loss": 1.0655,
      "step": 781
    },
    {
      "epoch": 0.5003198976327575,
      "grad_norm": 2.680401475165126,
      "learning_rate": 1.9978974299126946e-05,
      "loss": 0.988,
      "step": 782
    },
    {
      "epoch": 0.5009596928982726,
      "grad_norm": 2.4751312815824447,
      "learning_rate": 1.997883978246155e-05,
      "loss": 1.0437,
      "step": 783
    },
    {
      "epoch": 0.5015994881637876,
      "grad_norm": 2.618486986769146,
      "learning_rate": 1.9978704837322528e-05,
      "loss": 0.9601,
      "step": 784
    },
    {
      "epoch": 0.5022392834293026,
      "grad_norm": 2.5922967496278875,
      "learning_rate": 1.9978569463715672e-05,
      "loss": 1.0168,
      "step": 785
    },
    {
      "epoch": 0.5028790786948176,
      "grad_norm": 2.58196535514389,
      "learning_rate": 1.997843366164679e-05,
      "loss": 0.9927,
      "step": 786
    },
    {
      "epoch": 0.5035188739603327,
      "grad_norm": 2.5867019261946576,
      "learning_rate": 1.9978297431121718e-05,
      "loss": 1.0628,
      "step": 787
    },
    {
      "epoch": 0.5041586692258477,
      "grad_norm": 2.7128976927438546,
      "learning_rate": 1.9978160772146302e-05,
      "loss": 0.9978,
      "step": 788
    },
    {
      "epoch": 0.5047984644913628,
      "grad_norm": 3.1040592799977342,
      "learning_rate": 1.9978023684726415e-05,
      "loss": 1.0586,
      "step": 789
    },
    {
      "epoch": 0.5054382597568778,
      "grad_norm": 2.5868010427734567,
      "learning_rate": 1.997788616886794e-05,
      "loss": 0.9888,
      "step": 790
    },
    {
      "epoch": 0.5060780550223928,
      "grad_norm": 3.348014989752779,
      "learning_rate": 1.9977748224576784e-05,
      "loss": 0.9825,
      "step": 791
    },
    {
      "epoch": 0.5067178502879078,
      "grad_norm": 2.757530215723365,
      "learning_rate": 1.9977609851858865e-05,
      "loss": 1.0061,
      "step": 792
    },
    {
      "epoch": 0.5073576455534229,
      "grad_norm": 2.8845901887859964,
      "learning_rate": 1.9977471050720127e-05,
      "loss": 0.9641,
      "step": 793
    },
    {
      "epoch": 0.5079974408189379,
      "grad_norm": 2.607655423316454,
      "learning_rate": 1.9977331821166534e-05,
      "loss": 1.057,
      "step": 794
    },
    {
      "epoch": 0.508637236084453,
      "grad_norm": 2.8854996070483954,
      "learning_rate": 1.997719216320406e-05,
      "loss": 1.0217,
      "step": 795
    },
    {
      "epoch": 0.509277031349968,
      "grad_norm": 2.8142856023591416,
      "learning_rate": 1.99770520768387e-05,
      "loss": 1.0143,
      "step": 796
    },
    {
      "epoch": 0.5099168266154831,
      "grad_norm": 2.5070440016484725,
      "learning_rate": 1.9976911562076476e-05,
      "loss": 1.0747,
      "step": 797
    },
    {
      "epoch": 0.510556621880998,
      "grad_norm": 2.8104731553194897,
      "learning_rate": 1.997677061892341e-05,
      "loss": 1.0378,
      "step": 798
    },
    {
      "epoch": 0.5111964171465131,
      "grad_norm": 2.7410596928523527,
      "learning_rate": 1.9976629247385568e-05,
      "loss": 1.0338,
      "step": 799
    },
    {
      "epoch": 0.5118362124120281,
      "grad_norm": 2.877040753824934,
      "learning_rate": 1.9976487447469012e-05,
      "loss": 1.0212,
      "step": 800
    },
    {
      "epoch": 0.5124760076775432,
      "grad_norm": 2.9266802465988455,
      "learning_rate": 1.9976345219179828e-05,
      "loss": 0.9906,
      "step": 801
    },
    {
      "epoch": 0.5131158029430583,
      "grad_norm": 2.7458695870321423,
      "learning_rate": 1.997620256252413e-05,
      "loss": 0.9762,
      "step": 802
    },
    {
      "epoch": 0.5137555982085733,
      "grad_norm": 2.8573344685496167,
      "learning_rate": 1.9976059477508042e-05,
      "loss": 0.9609,
      "step": 803
    },
    {
      "epoch": 0.5143953934740882,
      "grad_norm": 2.896593033767702,
      "learning_rate": 1.9975915964137702e-05,
      "loss": 1.0276,
      "step": 804
    },
    {
      "epoch": 0.5150351887396033,
      "grad_norm": 3.0644436397884216,
      "learning_rate": 1.997577202241928e-05,
      "loss": 1.0326,
      "step": 805
    },
    {
      "epoch": 0.5156749840051184,
      "grad_norm": 2.3900848066627485,
      "learning_rate": 1.997562765235895e-05,
      "loss": 1.0155,
      "step": 806
    },
    {
      "epoch": 0.5163147792706334,
      "grad_norm": 2.5517193328900927,
      "learning_rate": 1.9975482853962913e-05,
      "loss": 1.0144,
      "step": 807
    },
    {
      "epoch": 0.5169545745361485,
      "grad_norm": 2.9179938256766436,
      "learning_rate": 1.997533762723739e-05,
      "loss": 0.9809,
      "step": 808
    },
    {
      "epoch": 0.5175943698016635,
      "grad_norm": 2.610683311110618,
      "learning_rate": 1.9975191972188615e-05,
      "loss": 1.0696,
      "step": 809
    },
    {
      "epoch": 0.5182341650671785,
      "grad_norm": 2.4772089301627984,
      "learning_rate": 1.9975045888822836e-05,
      "loss": 1.064,
      "step": 810
    },
    {
      "epoch": 0.5188739603326935,
      "grad_norm": 2.5047436767863753,
      "learning_rate": 1.9974899377146334e-05,
      "loss": 1.0383,
      "step": 811
    },
    {
      "epoch": 0.5195137555982086,
      "grad_norm": 2.8315288932983136,
      "learning_rate": 1.9974752437165396e-05,
      "loss": 1.0297,
      "step": 812
    },
    {
      "epoch": 0.5201535508637236,
      "grad_norm": 2.676588989889109,
      "learning_rate": 1.9974605068886334e-05,
      "loss": 1.0218,
      "step": 813
    },
    {
      "epoch": 0.5207933461292387,
      "grad_norm": 2.6878111982311896,
      "learning_rate": 1.9974457272315474e-05,
      "loss": 1.0448,
      "step": 814
    },
    {
      "epoch": 0.5214331413947537,
      "grad_norm": 2.66578614748554,
      "learning_rate": 1.997430904745916e-05,
      "loss": 1.0292,
      "step": 815
    },
    {
      "epoch": 0.5220729366602687,
      "grad_norm": 2.5757961261607796,
      "learning_rate": 1.9974160394323757e-05,
      "loss": 0.9872,
      "step": 816
    },
    {
      "epoch": 0.5227127319257837,
      "grad_norm": 2.5396868444304075,
      "learning_rate": 1.9974011312915654e-05,
      "loss": 1.0424,
      "step": 817
    },
    {
      "epoch": 0.5233525271912988,
      "grad_norm": 2.5353500078790026,
      "learning_rate": 1.9973861803241244e-05,
      "loss": 0.9882,
      "step": 818
    },
    {
      "epoch": 0.5239923224568138,
      "grad_norm": 2.277099794085869,
      "learning_rate": 1.997371186530695e-05,
      "loss": 1.0399,
      "step": 819
    },
    {
      "epoch": 0.5246321177223289,
      "grad_norm": 2.655087029508336,
      "learning_rate": 1.997356149911921e-05,
      "loss": 1.0552,
      "step": 820
    },
    {
      "epoch": 0.5252719129878439,
      "grad_norm": 2.530464276219458,
      "learning_rate": 1.9973410704684485e-05,
      "loss": 1.0302,
      "step": 821
    },
    {
      "epoch": 0.525911708253359,
      "grad_norm": 2.715165623984161,
      "learning_rate": 1.997325948200924e-05,
      "loss": 1.0251,
      "step": 822
    },
    {
      "epoch": 0.5265515035188739,
      "grad_norm": 2.799264061662147,
      "learning_rate": 1.997310783109998e-05,
      "loss": 0.9957,
      "step": 823
    },
    {
      "epoch": 0.527191298784389,
      "grad_norm": 2.813084758251407,
      "learning_rate": 1.9972955751963206e-05,
      "loss": 1.0276,
      "step": 824
    },
    {
      "epoch": 0.527831094049904,
      "grad_norm": 2.5738536034897836,
      "learning_rate": 1.9972803244605454e-05,
      "loss": 1.0249,
      "step": 825
    },
    {
      "epoch": 0.5284708893154191,
      "grad_norm": 2.7277687408097404,
      "learning_rate": 1.9972650309033272e-05,
      "loss": 1.0048,
      "step": 826
    },
    {
      "epoch": 0.5291106845809341,
      "grad_norm": 3.4732613789549656,
      "learning_rate": 1.9972496945253222e-05,
      "loss": 1.0277,
      "step": 827
    },
    {
      "epoch": 0.5297504798464492,
      "grad_norm": 2.911094663341328,
      "learning_rate": 1.9972343153271895e-05,
      "loss": 0.9951,
      "step": 828
    },
    {
      "epoch": 0.5303902751119641,
      "grad_norm": 2.7966063359536792,
      "learning_rate": 1.9972188933095895e-05,
      "loss": 0.9909,
      "step": 829
    },
    {
      "epoch": 0.5310300703774792,
      "grad_norm": 2.6867555482821417,
      "learning_rate": 1.997203428473184e-05,
      "loss": 0.98,
      "step": 830
    },
    {
      "epoch": 0.5316698656429942,
      "grad_norm": 2.85015028461485,
      "learning_rate": 1.997187920818637e-05,
      "loss": 1.0358,
      "step": 831
    },
    {
      "epoch": 0.5323096609085093,
      "grad_norm": 2.902896045909255,
      "learning_rate": 1.997172370346615e-05,
      "loss": 0.9895,
      "step": 832
    },
    {
      "epoch": 0.5329494561740243,
      "grad_norm": 2.8999742362664436,
      "learning_rate": 1.9971567770577846e-05,
      "loss": 1.0315,
      "step": 833
    },
    {
      "epoch": 0.5335892514395394,
      "grad_norm": 2.428905435295856,
      "learning_rate": 1.997141140952816e-05,
      "loss": 0.9848,
      "step": 834
    },
    {
      "epoch": 0.5342290467050543,
      "grad_norm": 2.444966907423109,
      "learning_rate": 1.997125462032381e-05,
      "loss": 0.9811,
      "step": 835
    },
    {
      "epoch": 0.5348688419705694,
      "grad_norm": 2.591853111710736,
      "learning_rate": 1.997109740297152e-05,
      "loss": 0.9805,
      "step": 836
    },
    {
      "epoch": 0.5355086372360844,
      "grad_norm": 2.6562564200791345,
      "learning_rate": 1.997093975747805e-05,
      "loss": 0.9871,
      "step": 837
    },
    {
      "epoch": 0.5361484325015995,
      "grad_norm": 2.9650917187879404,
      "learning_rate": 1.9970781683850163e-05,
      "loss": 1.0194,
      "step": 838
    },
    {
      "epoch": 0.5367882277671145,
      "grad_norm": 2.5626375623850755,
      "learning_rate": 1.9970623182094643e-05,
      "loss": 0.9998,
      "step": 839
    },
    {
      "epoch": 0.5374280230326296,
      "grad_norm": 2.4644425023957375,
      "learning_rate": 1.9970464252218303e-05,
      "loss": 1.0097,
      "step": 840
    },
    {
      "epoch": 0.5380678182981445,
      "grad_norm": 2.440085753535706,
      "learning_rate": 1.9970304894227965e-05,
      "loss": 1.0495,
      "step": 841
    },
    {
      "epoch": 0.5387076135636596,
      "grad_norm": 2.8021537625814634,
      "learning_rate": 1.9970145108130468e-05,
      "loss": 1.0305,
      "step": 842
    },
    {
      "epoch": 0.5393474088291746,
      "grad_norm": 2.4347102341598847,
      "learning_rate": 1.9969984893932673e-05,
      "loss": 1.0648,
      "step": 843
    },
    {
      "epoch": 0.5399872040946897,
      "grad_norm": 2.75659563181447,
      "learning_rate": 1.9969824251641466e-05,
      "loss": 1.0142,
      "step": 844
    },
    {
      "epoch": 0.5406269993602048,
      "grad_norm": 2.8678069987203365,
      "learning_rate": 1.9969663181263745e-05,
      "loss": 1.0198,
      "step": 845
    },
    {
      "epoch": 0.5412667946257198,
      "grad_norm": 2.642710107039389,
      "learning_rate": 1.9969501682806413e-05,
      "loss": 1.0151,
      "step": 846
    },
    {
      "epoch": 0.5419065898912349,
      "grad_norm": 2.7831095204880203,
      "learning_rate": 1.996933975627642e-05,
      "loss": 1.062,
      "step": 847
    },
    {
      "epoch": 0.5425463851567498,
      "grad_norm": 2.5638873309595387,
      "learning_rate": 1.9969177401680705e-05,
      "loss": 0.9437,
      "step": 848
    },
    {
      "epoch": 0.5431861804222649,
      "grad_norm": 2.8703471370795772,
      "learning_rate": 1.996901461902625e-05,
      "loss": 1.0073,
      "step": 849
    },
    {
      "epoch": 0.5438259756877799,
      "grad_norm": 3.106494281702087,
      "learning_rate": 1.9968851408320042e-05,
      "loss": 1.0495,
      "step": 850
    },
    {
      "epoch": 0.544465770953295,
      "grad_norm": 2.9165310744144888,
      "learning_rate": 1.9968687769569082e-05,
      "loss": 0.9915,
      "step": 851
    },
    {
      "epoch": 0.54510556621881,
      "grad_norm": 3.1752476684911923,
      "learning_rate": 1.996852370278041e-05,
      "loss": 1.0478,
      "step": 852
    },
    {
      "epoch": 0.5457453614843251,
      "grad_norm": 2.7004296303464113,
      "learning_rate": 1.9968359207961057e-05,
      "loss": 1.0481,
      "step": 853
    },
    {
      "epoch": 0.54638515674984,
      "grad_norm": 2.8464123764132596,
      "learning_rate": 1.9968194285118097e-05,
      "loss": 1.012,
      "step": 854
    },
    {
      "epoch": 0.5470249520153551,
      "grad_norm": 2.6424704857965273,
      "learning_rate": 1.9968028934258604e-05,
      "loss": 1.0437,
      "step": 855
    },
    {
      "epoch": 0.5476647472808701,
      "grad_norm": 2.8207095520272185,
      "learning_rate": 1.9967863155389676e-05,
      "loss": 1.0383,
      "step": 856
    },
    {
      "epoch": 0.5483045425463852,
      "grad_norm": 3.431460255524561,
      "learning_rate": 1.9967696948518442e-05,
      "loss": 1.0317,
      "step": 857
    },
    {
      "epoch": 0.5489443378119002,
      "grad_norm": 2.6646359609725305,
      "learning_rate": 1.996753031365203e-05,
      "loss": 1.0097,
      "step": 858
    },
    {
      "epoch": 0.5495841330774153,
      "grad_norm": 2.592993626519625,
      "learning_rate": 1.9967363250797596e-05,
      "loss": 1.0207,
      "step": 859
    },
    {
      "epoch": 0.5502239283429302,
      "grad_norm": 3.1182627002710674,
      "learning_rate": 1.9967195759962318e-05,
      "loss": 1.0186,
      "step": 860
    },
    {
      "epoch": 0.5508637236084453,
      "grad_norm": 2.712471948637913,
      "learning_rate": 1.996702784115338e-05,
      "loss": 0.9917,
      "step": 861
    },
    {
      "epoch": 0.5515035188739603,
      "grad_norm": 2.7543090106322885,
      "learning_rate": 1.9966859494377998e-05,
      "loss": 1.0538,
      "step": 862
    },
    {
      "epoch": 0.5521433141394754,
      "grad_norm": 2.7114268265544608,
      "learning_rate": 1.99666907196434e-05,
      "loss": 1.0157,
      "step": 863
    },
    {
      "epoch": 0.5527831094049904,
      "grad_norm": 2.849291943259465,
      "learning_rate": 1.996652151695683e-05,
      "loss": 1.0942,
      "step": 864
    },
    {
      "epoch": 0.5534229046705055,
      "grad_norm": 2.743970746997532,
      "learning_rate": 1.9966351886325557e-05,
      "loss": 0.9786,
      "step": 865
    },
    {
      "epoch": 0.5540626999360204,
      "grad_norm": 2.6573983667607224,
      "learning_rate": 1.996618182775686e-05,
      "loss": 1.0453,
      "step": 866
    },
    {
      "epoch": 0.5547024952015355,
      "grad_norm": 2.7183683591767704,
      "learning_rate": 1.9966011341258046e-05,
      "loss": 1.0039,
      "step": 867
    },
    {
      "epoch": 0.5553422904670505,
      "grad_norm": 3.0726512740623795,
      "learning_rate": 1.9965840426836433e-05,
      "loss": 1.0302,
      "step": 868
    },
    {
      "epoch": 0.5559820857325656,
      "grad_norm": 2.5510570971405073,
      "learning_rate": 1.9965669084499363e-05,
      "loss": 0.9841,
      "step": 869
    },
    {
      "epoch": 0.5566218809980806,
      "grad_norm": 2.448958251123029,
      "learning_rate": 1.9965497314254186e-05,
      "loss": 0.9999,
      "step": 870
    },
    {
      "epoch": 0.5572616762635957,
      "grad_norm": 2.544207755274459,
      "learning_rate": 1.9965325116108283e-05,
      "loss": 1.0295,
      "step": 871
    },
    {
      "epoch": 0.5579014715291107,
      "grad_norm": 2.4066491995597326,
      "learning_rate": 1.9965152490069043e-05,
      "loss": 1.0357,
      "step": 872
    },
    {
      "epoch": 0.5585412667946257,
      "grad_norm": 3.047153118928369,
      "learning_rate": 1.9964979436143887e-05,
      "loss": 1.0141,
      "step": 873
    },
    {
      "epoch": 0.5591810620601407,
      "grad_norm": 2.570445516422656,
      "learning_rate": 1.9964805954340237e-05,
      "loss": 1.0111,
      "step": 874
    },
    {
      "epoch": 0.5598208573256558,
      "grad_norm": 2.540627123290687,
      "learning_rate": 1.9964632044665548e-05,
      "loss": 1.0069,
      "step": 875
    },
    {
      "epoch": 0.5604606525911708,
      "grad_norm": 2.616975938824914,
      "learning_rate": 1.996445770712728e-05,
      "loss": 0.9794,
      "step": 876
    },
    {
      "epoch": 0.5611004478566859,
      "grad_norm": 3.39240014631203,
      "learning_rate": 1.996428294173293e-05,
      "loss": 1.0648,
      "step": 877
    },
    {
      "epoch": 0.5617402431222009,
      "grad_norm": 2.6414929320819596,
      "learning_rate": 1.9964107748489988e-05,
      "loss": 0.9729,
      "step": 878
    },
    {
      "epoch": 0.5623800383877159,
      "grad_norm": 2.7791471660038845,
      "learning_rate": 1.9963932127405988e-05,
      "loss": 1.0426,
      "step": 879
    },
    {
      "epoch": 0.5630198336532309,
      "grad_norm": 2.501676822809313,
      "learning_rate": 1.996375607848847e-05,
      "loss": 0.9972,
      "step": 880
    },
    {
      "epoch": 0.563659628918746,
      "grad_norm": 2.3746577178817163,
      "learning_rate": 1.9963579601744985e-05,
      "loss": 0.9894,
      "step": 881
    },
    {
      "epoch": 0.564299424184261,
      "grad_norm": 2.865213373738841,
      "learning_rate": 1.996340269718312e-05,
      "loss": 0.998,
      "step": 882
    },
    {
      "epoch": 0.5649392194497761,
      "grad_norm": 2.5237033923733914,
      "learning_rate": 1.9963225364810464e-05,
      "loss": 1.0305,
      "step": 883
    },
    {
      "epoch": 0.5655790147152912,
      "grad_norm": 2.651747656175551,
      "learning_rate": 1.9963047604634632e-05,
      "loss": 0.9758,
      "step": 884
    },
    {
      "epoch": 0.5662188099808061,
      "grad_norm": 2.863244277105121,
      "learning_rate": 1.9962869416663266e-05,
      "loss": 0.9731,
      "step": 885
    },
    {
      "epoch": 0.5668586052463211,
      "grad_norm": 2.3773774238007923,
      "learning_rate": 1.9962690800904e-05,
      "loss": 1.0177,
      "step": 886
    },
    {
      "epoch": 0.5674984005118362,
      "grad_norm": 2.6967604416714748,
      "learning_rate": 1.9962511757364522e-05,
      "loss": 0.9738,
      "step": 887
    },
    {
      "epoch": 0.5681381957773513,
      "grad_norm": 2.9778927177640764,
      "learning_rate": 1.9962332286052507e-05,
      "loss": 1.0175,
      "step": 888
    },
    {
      "epoch": 0.5687779910428663,
      "grad_norm": 2.5704792944797434,
      "learning_rate": 1.9962152386975668e-05,
      "loss": 0.9909,
      "step": 889
    },
    {
      "epoch": 0.5694177863083814,
      "grad_norm": 2.7171264490221683,
      "learning_rate": 1.996197206014172e-05,
      "loss": 1.0003,
      "step": 890
    },
    {
      "epoch": 0.5700575815738963,
      "grad_norm": 2.8211692179347794,
      "learning_rate": 1.996179130555842e-05,
      "loss": 0.9775,
      "step": 891
    },
    {
      "epoch": 0.5706973768394114,
      "grad_norm": 2.6801072699422592,
      "learning_rate": 1.996161012323352e-05,
      "loss": 1.0158,
      "step": 892
    },
    {
      "epoch": 0.5713371721049264,
      "grad_norm": 2.634131537319574,
      "learning_rate": 1.9961428513174798e-05,
      "loss": 1.0141,
      "step": 893
    },
    {
      "epoch": 0.5719769673704415,
      "grad_norm": 3.350110334929811,
      "learning_rate": 1.996124647539006e-05,
      "loss": 0.9984,
      "step": 894
    },
    {
      "epoch": 0.5726167626359565,
      "grad_norm": 2.642362053243956,
      "learning_rate": 1.9961064009887117e-05,
      "loss": 1.0026,
      "step": 895
    },
    {
      "epoch": 0.5732565579014716,
      "grad_norm": 2.7600233652245296,
      "learning_rate": 1.9960881116673803e-05,
      "loss": 0.9738,
      "step": 896
    },
    {
      "epoch": 0.5738963531669866,
      "grad_norm": 3.2495817608623185,
      "learning_rate": 1.9960697795757976e-05,
      "loss": 0.9688,
      "step": 897
    },
    {
      "epoch": 0.5745361484325016,
      "grad_norm": 2.6983890822180583,
      "learning_rate": 1.9960514047147502e-05,
      "loss": 0.9883,
      "step": 898
    },
    {
      "epoch": 0.5751759436980166,
      "grad_norm": 2.258019454803452,
      "learning_rate": 1.9960329870850272e-05,
      "loss": 1.0172,
      "step": 899
    },
    {
      "epoch": 0.5758157389635317,
      "grad_norm": 2.7606596094901987,
      "learning_rate": 1.99601452668742e-05,
      "loss": 1.0315,
      "step": 900
    },
    {
      "epoch": 0.5764555342290467,
      "grad_norm": 2.9937261825493633,
      "learning_rate": 1.99599602352272e-05,
      "loss": 0.9689,
      "step": 901
    },
    {
      "epoch": 0.5770953294945618,
      "grad_norm": 3.271675225261368,
      "learning_rate": 1.9959774775917234e-05,
      "loss": 0.9889,
      "step": 902
    },
    {
      "epoch": 0.5777351247600768,
      "grad_norm": 2.812078320220041,
      "learning_rate": 1.995958888895225e-05,
      "loss": 0.9712,
      "step": 903
    },
    {
      "epoch": 0.5783749200255918,
      "grad_norm": 2.937238093227635,
      "learning_rate": 1.995940257434024e-05,
      "loss": 0.9892,
      "step": 904
    },
    {
      "epoch": 0.5790147152911068,
      "grad_norm": 2.580799263062945,
      "learning_rate": 1.9959215832089197e-05,
      "loss": 1.0149,
      "step": 905
    },
    {
      "epoch": 0.5796545105566219,
      "grad_norm": 2.561031534984995,
      "learning_rate": 1.9959028662207146e-05,
      "loss": 0.9866,
      "step": 906
    },
    {
      "epoch": 0.5802943058221369,
      "grad_norm": 2.801316171256154,
      "learning_rate": 1.9958841064702114e-05,
      "loss": 1.0219,
      "step": 907
    },
    {
      "epoch": 0.580934101087652,
      "grad_norm": 2.906418547959419,
      "learning_rate": 1.995865303958217e-05,
      "loss": 0.9643,
      "step": 908
    },
    {
      "epoch": 0.581573896353167,
      "grad_norm": 2.4110630072085786,
      "learning_rate": 1.995846458685537e-05,
      "loss": 0.9808,
      "step": 909
    },
    {
      "epoch": 0.582213691618682,
      "grad_norm": 2.5515205273090915,
      "learning_rate": 1.9958275706529822e-05,
      "loss": 0.9707,
      "step": 910
    },
    {
      "epoch": 0.582853486884197,
      "grad_norm": 2.6862073891544855,
      "learning_rate": 1.9958086398613627e-05,
      "loss": 1.084,
      "step": 911
    },
    {
      "epoch": 0.5834932821497121,
      "grad_norm": 2.6739869014786066,
      "learning_rate": 1.9957896663114918e-05,
      "loss": 1.0121,
      "step": 912
    },
    {
      "epoch": 0.5841330774152271,
      "grad_norm": 2.6305694755677833,
      "learning_rate": 1.9957706500041838e-05,
      "loss": 1.0657,
      "step": 913
    },
    {
      "epoch": 0.5847728726807422,
      "grad_norm": 2.8440048083761074,
      "learning_rate": 1.9957515909402552e-05,
      "loss": 1.0409,
      "step": 914
    },
    {
      "epoch": 0.5854126679462572,
      "grad_norm": 2.505734800971472,
      "learning_rate": 1.9957324891205253e-05,
      "loss": 1.0138,
      "step": 915
    },
    {
      "epoch": 0.5860524632117722,
      "grad_norm": 2.523712523168351,
      "learning_rate": 1.9957133445458133e-05,
      "loss": 1.041,
      "step": 916
    },
    {
      "epoch": 0.5866922584772872,
      "grad_norm": 2.8563197278032946,
      "learning_rate": 1.995694157216941e-05,
      "loss": 0.9485,
      "step": 917
    },
    {
      "epoch": 0.5873320537428023,
      "grad_norm": 2.6133402544535307,
      "learning_rate": 1.9956749271347332e-05,
      "loss": 0.9919,
      "step": 918
    },
    {
      "epoch": 0.5879718490083173,
      "grad_norm": 3.0974081969082405,
      "learning_rate": 1.9956556543000155e-05,
      "loss": 1.06,
      "step": 919
    },
    {
      "epoch": 0.5886116442738324,
      "grad_norm": 2.7293982778961845,
      "learning_rate": 1.9956363387136147e-05,
      "loss": 0.9942,
      "step": 920
    },
    {
      "epoch": 0.5892514395393474,
      "grad_norm": 2.4955152716692335,
      "learning_rate": 1.9956169803763606e-05,
      "loss": 1.0067,
      "step": 921
    },
    {
      "epoch": 0.5898912348048625,
      "grad_norm": 2.8864321282979564,
      "learning_rate": 1.9955975792890847e-05,
      "loss": 1.0538,
      "step": 922
    },
    {
      "epoch": 0.5905310300703774,
      "grad_norm": 2.7849617188931304,
      "learning_rate": 1.9955781354526197e-05,
      "loss": 0.958,
      "step": 923
    },
    {
      "epoch": 0.5911708253358925,
      "grad_norm": 2.9652611313143815,
      "learning_rate": 1.9955586488678004e-05,
      "loss": 1.0322,
      "step": 924
    },
    {
      "epoch": 0.5918106206014075,
      "grad_norm": 2.7723852598338636,
      "learning_rate": 1.9955391195354638e-05,
      "loss": 0.9323,
      "step": 925
    },
    {
      "epoch": 0.5924504158669226,
      "grad_norm": 2.975648226689305,
      "learning_rate": 1.9955195474564488e-05,
      "loss": 1.0412,
      "step": 926
    },
    {
      "epoch": 0.5930902111324377,
      "grad_norm": 2.542689515339304,
      "learning_rate": 1.9954999326315945e-05,
      "loss": 0.9818,
      "step": 927
    },
    {
      "epoch": 0.5937300063979527,
      "grad_norm": 2.359076910596051,
      "learning_rate": 1.9954802750617443e-05,
      "loss": 0.9514,
      "step": 928
    },
    {
      "epoch": 0.5943698016634676,
      "grad_norm": 2.9509519787224785,
      "learning_rate": 1.9954605747477422e-05,
      "loss": 1.0385,
      "step": 929
    },
    {
      "epoch": 0.5950095969289827,
      "grad_norm": 2.2619028961290653,
      "learning_rate": 1.995440831690433e-05,
      "loss": 0.996,
      "step": 930
    },
    {
      "epoch": 0.5956493921944978,
      "grad_norm": 2.4490242861753964,
      "learning_rate": 1.995421045890666e-05,
      "loss": 0.9882,
      "step": 931
    },
    {
      "epoch": 0.5962891874600128,
      "grad_norm": 2.182409406798446,
      "learning_rate": 1.9954012173492895e-05,
      "loss": 1.068,
      "step": 932
    },
    {
      "epoch": 0.5969289827255279,
      "grad_norm": 3.3612080034826026,
      "learning_rate": 1.995381346067156e-05,
      "loss": 1.0204,
      "step": 933
    },
    {
      "epoch": 0.5975687779910429,
      "grad_norm": 2.7002793204578164,
      "learning_rate": 1.9953614320451178e-05,
      "loss": 1.0248,
      "step": 934
    },
    {
      "epoch": 0.5982085732565579,
      "grad_norm": 2.4662197576067646,
      "learning_rate": 1.9953414752840305e-05,
      "loss": 0.9492,
      "step": 935
    },
    {
      "epoch": 0.5988483685220729,
      "grad_norm": 2.6679912835905366,
      "learning_rate": 1.9953214757847505e-05,
      "loss": 0.972,
      "step": 936
    },
    {
      "epoch": 0.599488163787588,
      "grad_norm": 2.217285139174275,
      "learning_rate": 1.995301433548137e-05,
      "loss": 0.9579,
      "step": 937
    },
    {
      "epoch": 0.600127959053103,
      "grad_norm": 2.5436714951272688,
      "learning_rate": 1.9952813485750507e-05,
      "loss": 0.9879,
      "step": 938
    },
    {
      "epoch": 0.6007677543186181,
      "grad_norm": 2.964104019143105,
      "learning_rate": 1.9952612208663533e-05,
      "loss": 0.9644,
      "step": 939
    },
    {
      "epoch": 0.6014075495841331,
      "grad_norm": 2.6311080306939543,
      "learning_rate": 1.9952410504229098e-05,
      "loss": 0.9906,
      "step": 940
    },
    {
      "epoch": 0.6020473448496481,
      "grad_norm": 2.5731248986586537,
      "learning_rate": 1.9952208372455858e-05,
      "loss": 0.9217,
      "step": 941
    },
    {
      "epoch": 0.6026871401151631,
      "grad_norm": 2.276871913486961,
      "learning_rate": 1.9952005813352497e-05,
      "loss": 1.0096,
      "step": 942
    },
    {
      "epoch": 0.6033269353806782,
      "grad_norm": 3.0449277025840775,
      "learning_rate": 1.9951802826927705e-05,
      "loss": 0.9947,
      "step": 943
    },
    {
      "epoch": 0.6039667306461932,
      "grad_norm": 3.388170942342027,
      "learning_rate": 1.9951599413190208e-05,
      "loss": 1.0382,
      "step": 944
    },
    {
      "epoch": 0.6046065259117083,
      "grad_norm": 2.796600337388224,
      "learning_rate": 1.9951395572148733e-05,
      "loss": 0.9936,
      "step": 945
    },
    {
      "epoch": 0.6052463211772233,
      "grad_norm": 2.5891973007913482,
      "learning_rate": 1.995119130381203e-05,
      "loss": 1.0322,
      "step": 946
    },
    {
      "epoch": 0.6058861164427384,
      "grad_norm": 2.5231887793585694,
      "learning_rate": 1.995098660818888e-05,
      "loss": 1.0157,
      "step": 947
    },
    {
      "epoch": 0.6065259117082533,
      "grad_norm": 2.9349336436121436,
      "learning_rate": 1.9950781485288066e-05,
      "loss": 1.0346,
      "step": 948
    },
    {
      "epoch": 0.6071657069737684,
      "grad_norm": 2.7831338468481985,
      "learning_rate": 1.9950575935118394e-05,
      "loss": 0.9804,
      "step": 949
    },
    {
      "epoch": 0.6078055022392834,
      "grad_norm": 3.002844595125841,
      "learning_rate": 1.995036995768869e-05,
      "loss": 0.9843,
      "step": 950
    },
    {
      "epoch": 0.6084452975047985,
      "grad_norm": 2.4771914678722826,
      "learning_rate": 1.9950163553007807e-05,
      "loss": 1.0453,
      "step": 951
    },
    {
      "epoch": 0.6090850927703135,
      "grad_norm": 2.8170210381801843,
      "learning_rate": 1.9949956721084596e-05,
      "loss": 1.0178,
      "step": 952
    },
    {
      "epoch": 0.6097248880358286,
      "grad_norm": 2.482199760304433,
      "learning_rate": 1.9949749461927944e-05,
      "loss": 1.0174,
      "step": 953
    },
    {
      "epoch": 0.6103646833013435,
      "grad_norm": 2.7364749601814133,
      "learning_rate": 1.994954177554675e-05,
      "loss": 1.0304,
      "step": 954
    },
    {
      "epoch": 0.6110044785668586,
      "grad_norm": 2.711008795332919,
      "learning_rate": 1.994933366194993e-05,
      "loss": 0.9941,
      "step": 955
    },
    {
      "epoch": 0.6116442738323736,
      "grad_norm": 2.5890503359860495,
      "learning_rate": 1.994912512114642e-05,
      "loss": 1.0101,
      "step": 956
    },
    {
      "epoch": 0.6122840690978887,
      "grad_norm": 2.5464078910188745,
      "learning_rate": 1.9948916153145178e-05,
      "loss": 0.9427,
      "step": 957
    },
    {
      "epoch": 0.6129238643634037,
      "grad_norm": 2.4989358680683913,
      "learning_rate": 1.9948706757955172e-05,
      "loss": 0.9682,
      "step": 958
    },
    {
      "epoch": 0.6135636596289188,
      "grad_norm": 3.1509861831469723,
      "learning_rate": 1.9948496935585396e-05,
      "loss": 1.0892,
      "step": 959
    },
    {
      "epoch": 0.6142034548944337,
      "grad_norm": 2.5323546644635413,
      "learning_rate": 1.9948286686044857e-05,
      "loss": 0.9319,
      "step": 960
    },
    {
      "epoch": 0.6148432501599488,
      "grad_norm": 3.058749134050735,
      "learning_rate": 1.9948076009342588e-05,
      "loss": 0.9639,
      "step": 961
    },
    {
      "epoch": 0.6154830454254638,
      "grad_norm": 2.677216540337387,
      "learning_rate": 1.9947864905487627e-05,
      "loss": 1.0213,
      "step": 962
    },
    {
      "epoch": 0.6161228406909789,
      "grad_norm": 2.4318505224481215,
      "learning_rate": 1.9947653374489048e-05,
      "loss": 0.9652,
      "step": 963
    },
    {
      "epoch": 0.6167626359564939,
      "grad_norm": 2.4875645408902223,
      "learning_rate": 1.9947441416355925e-05,
      "loss": 1.0092,
      "step": 964
    },
    {
      "epoch": 0.617402431222009,
      "grad_norm": 2.7844843258262824,
      "learning_rate": 1.994722903109736e-05,
      "loss": 1.0628,
      "step": 965
    },
    {
      "epoch": 0.6180422264875239,
      "grad_norm": 3.273761953482096,
      "learning_rate": 1.9947016218722482e-05,
      "loss": 0.9748,
      "step": 966
    },
    {
      "epoch": 0.618682021753039,
      "grad_norm": 2.999538496399055,
      "learning_rate": 1.9946802979240417e-05,
      "loss": 1.0013,
      "step": 967
    },
    {
      "epoch": 0.619321817018554,
      "grad_norm": 2.4243947009563684,
      "learning_rate": 1.9946589312660328e-05,
      "loss": 1.1082,
      "step": 968
    },
    {
      "epoch": 0.6199616122840691,
      "grad_norm": 3.085077521082905,
      "learning_rate": 1.9946375218991388e-05,
      "loss": 0.992,
      "step": 969
    },
    {
      "epoch": 0.6206014075495841,
      "grad_norm": 3.1350875657664803,
      "learning_rate": 1.9946160698242787e-05,
      "loss": 0.9939,
      "step": 970
    },
    {
      "epoch": 0.6212412028150992,
      "grad_norm": 2.650332494710373,
      "learning_rate": 1.9945945750423742e-05,
      "loss": 0.978,
      "step": 971
    },
    {
      "epoch": 0.6218809980806143,
      "grad_norm": 2.526233210330557,
      "learning_rate": 1.9945730375543475e-05,
      "loss": 0.99,
      "step": 972
    },
    {
      "epoch": 0.6225207933461292,
      "grad_norm": 3.0625616595504632,
      "learning_rate": 1.994551457361124e-05,
      "loss": 0.9577,
      "step": 973
    },
    {
      "epoch": 0.6231605886116443,
      "grad_norm": 2.5180996086680456,
      "learning_rate": 1.9945298344636298e-05,
      "loss": 1.0537,
      "step": 974
    },
    {
      "epoch": 0.6238003838771593,
      "grad_norm": 2.602951412552846,
      "learning_rate": 1.994508168862794e-05,
      "loss": 1.0269,
      "step": 975
    },
    {
      "epoch": 0.6244401791426744,
      "grad_norm": 3.3700314789320274,
      "learning_rate": 1.9944864605595463e-05,
      "loss": 0.9854,
      "step": 976
    },
    {
      "epoch": 0.6250799744081894,
      "grad_norm": 2.60916375637917,
      "learning_rate": 1.9944647095548194e-05,
      "loss": 1.0137,
      "step": 977
    },
    {
      "epoch": 0.6257197696737045,
      "grad_norm": 2.600578293907656,
      "learning_rate": 1.9944429158495468e-05,
      "loss": 1.0027,
      "step": 978
    },
    {
      "epoch": 0.6263595649392194,
      "grad_norm": 3.1150148108663815,
      "learning_rate": 1.994421079444664e-05,
      "loss": 0.9933,
      "step": 979
    },
    {
      "epoch": 0.6269993602047345,
      "grad_norm": 3.054927154889795,
      "learning_rate": 1.9943992003411092e-05,
      "loss": 0.9746,
      "step": 980
    },
    {
      "epoch": 0.6276391554702495,
      "grad_norm": 2.206535302883484,
      "learning_rate": 1.9943772785398216e-05,
      "loss": 0.9695,
      "step": 981
    },
    {
      "epoch": 0.6282789507357646,
      "grad_norm": 2.8806555599424457,
      "learning_rate": 1.9943553140417427e-05,
      "loss": 1.0053,
      "step": 982
    },
    {
      "epoch": 0.6289187460012796,
      "grad_norm": 2.3954912131755286,
      "learning_rate": 1.994333306847815e-05,
      "loss": 1.0203,
      "step": 983
    },
    {
      "epoch": 0.6295585412667947,
      "grad_norm": 2.4518911968408914,
      "learning_rate": 1.994311256958984e-05,
      "loss": 0.9842,
      "step": 984
    },
    {
      "epoch": 0.6301983365323096,
      "grad_norm": 2.4202341416142685,
      "learning_rate": 1.994289164376197e-05,
      "loss": 1.031,
      "step": 985
    },
    {
      "epoch": 0.6308381317978247,
      "grad_norm": 3.453920717430681,
      "learning_rate": 1.9942670291004013e-05,
      "loss": 0.9971,
      "step": 986
    },
    {
      "epoch": 0.6314779270633397,
      "grad_norm": 2.979717899341948,
      "learning_rate": 1.9942448511325484e-05,
      "loss": 1.018,
      "step": 987
    },
    {
      "epoch": 0.6321177223288548,
      "grad_norm": 2.587278473153428,
      "learning_rate": 1.99422263047359e-05,
      "loss": 1.0261,
      "step": 988
    },
    {
      "epoch": 0.6327575175943698,
      "grad_norm": 2.4403979302132504,
      "learning_rate": 1.9942003671244806e-05,
      "loss": 1.0711,
      "step": 989
    },
    {
      "epoch": 0.6333973128598849,
      "grad_norm": 2.3827103964590033,
      "learning_rate": 1.994178061086176e-05,
      "loss": 1.0949,
      "step": 990
    },
    {
      "epoch": 0.6340371081253998,
      "grad_norm": 2.6860417994151096,
      "learning_rate": 1.994155712359634e-05,
      "loss": 1.0305,
      "step": 991
    },
    {
      "epoch": 0.6346769033909149,
      "grad_norm": 2.9282575262042903,
      "learning_rate": 1.9941333209458138e-05,
      "loss": 0.9823,
      "step": 992
    },
    {
      "epoch": 0.6353166986564299,
      "grad_norm": 2.5342062940108216,
      "learning_rate": 1.9941108868456775e-05,
      "loss": 1.0056,
      "step": 993
    },
    {
      "epoch": 0.635956493921945,
      "grad_norm": 2.717002896148259,
      "learning_rate": 1.994088410060188e-05,
      "loss": 0.9925,
      "step": 994
    },
    {
      "epoch": 0.63659628918746,
      "grad_norm": 2.710812897587225,
      "learning_rate": 1.9940658905903105e-05,
      "loss": 1.0154,
      "step": 995
    },
    {
      "epoch": 0.6372360844529751,
      "grad_norm": 3.13249285344138,
      "learning_rate": 1.994043328437012e-05,
      "loss": 1.0466,
      "step": 996
    },
    {
      "epoch": 0.6378758797184901,
      "grad_norm": 2.810630437543612,
      "learning_rate": 1.994020723601261e-05,
      "loss": 0.9942,
      "step": 997
    },
    {
      "epoch": 0.6385156749840051,
      "grad_norm": 2.6004017740770324,
      "learning_rate": 1.9939980760840288e-05,
      "loss": 0.9721,
      "step": 998
    },
    {
      "epoch": 0.6391554702495201,
      "grad_norm": 2.95580058619461,
      "learning_rate": 1.9939753858862874e-05,
      "loss": 0.9812,
      "step": 999
    },
    {
      "epoch": 0.6397952655150352,
      "grad_norm": 2.5248022805048254,
      "learning_rate": 1.993952653009011e-05,
      "loss": 1.0429,
      "step": 1000
    },
    {
      "epoch": 0.6404350607805502,
      "grad_norm": 2.256129004728206,
      "learning_rate": 1.9939298774531757e-05,
      "loss": 1.0314,
      "step": 1001
    },
    {
      "epoch": 0.6410748560460653,
      "grad_norm": 2.4022193342333678,
      "learning_rate": 1.9939070592197594e-05,
      "loss": 1.0429,
      "step": 1002
    },
    {
      "epoch": 0.6417146513115803,
      "grad_norm": 2.8896823533684346,
      "learning_rate": 1.993884198309742e-05,
      "loss": 1.0483,
      "step": 1003
    },
    {
      "epoch": 0.6423544465770953,
      "grad_norm": 2.6680790640856267,
      "learning_rate": 1.9938612947241053e-05,
      "loss": 0.9595,
      "step": 1004
    },
    {
      "epoch": 0.6429942418426103,
      "grad_norm": 2.7690449276064255,
      "learning_rate": 1.9938383484638326e-05,
      "loss": 0.9674,
      "step": 1005
    },
    {
      "epoch": 0.6436340371081254,
      "grad_norm": 2.2722734368914543,
      "learning_rate": 1.993815359529909e-05,
      "loss": 0.9697,
      "step": 1006
    },
    {
      "epoch": 0.6442738323736404,
      "grad_norm": 2.8848800546199165,
      "learning_rate": 1.9937923279233214e-05,
      "loss": 0.9962,
      "step": 1007
    },
    {
      "epoch": 0.6449136276391555,
      "grad_norm": 2.9894411580061027,
      "learning_rate": 1.9937692536450596e-05,
      "loss": 0.9675,
      "step": 1008
    },
    {
      "epoch": 0.6455534229046705,
      "grad_norm": 2.7520901695139246,
      "learning_rate": 1.9937461366961138e-05,
      "loss": 0.9888,
      "step": 1009
    },
    {
      "epoch": 0.6461932181701855,
      "grad_norm": 2.5997500652241885,
      "learning_rate": 1.9937229770774763e-05,
      "loss": 1.0006,
      "step": 1010
    },
    {
      "epoch": 0.6468330134357005,
      "grad_norm": 3.111652945066597,
      "learning_rate": 1.993699774790142e-05,
      "loss": 0.992,
      "step": 1011
    },
    {
      "epoch": 0.6474728087012156,
      "grad_norm": 2.640455831709262,
      "learning_rate": 1.9936765298351072e-05,
      "loss": 1.0659,
      "step": 1012
    },
    {
      "epoch": 0.6481126039667306,
      "grad_norm": 2.9197185007203013,
      "learning_rate": 1.9936532422133695e-05,
      "loss": 1.0068,
      "step": 1013
    },
    {
      "epoch": 0.6487523992322457,
      "grad_norm": 2.391687837202122,
      "learning_rate": 1.9936299119259293e-05,
      "loss": 0.987,
      "step": 1014
    },
    {
      "epoch": 0.6493921944977608,
      "grad_norm": 2.9151395938289553,
      "learning_rate": 1.9936065389737882e-05,
      "loss": 0.942,
      "step": 1015
    },
    {
      "epoch": 0.6500319897632757,
      "grad_norm": 3.11518387164241,
      "learning_rate": 1.9935831233579497e-05,
      "loss": 0.992,
      "step": 1016
    },
    {
      "epoch": 0.6506717850287908,
      "grad_norm": 2.4720448777349375,
      "learning_rate": 1.9935596650794196e-05,
      "loss": 1.0009,
      "step": 1017
    },
    {
      "epoch": 0.6513115802943058,
      "grad_norm": 2.466601682680757,
      "learning_rate": 1.9935361641392046e-05,
      "loss": 0.9864,
      "step": 1018
    },
    {
      "epoch": 0.6519513755598209,
      "grad_norm": 2.210073071682159,
      "learning_rate": 1.993512620538314e-05,
      "loss": 1.0011,
      "step": 1019
    },
    {
      "epoch": 0.6525911708253359,
      "grad_norm": 3.3835970573222984,
      "learning_rate": 1.9934890342777592e-05,
      "loss": 0.9919,
      "step": 1020
    },
    {
      "epoch": 0.653230966090851,
      "grad_norm": 2.6284677583014147,
      "learning_rate": 1.993465405358552e-05,
      "loss": 0.9808,
      "step": 1021
    },
    {
      "epoch": 0.653870761356366,
      "grad_norm": 2.349170943820921,
      "learning_rate": 1.993441733781708e-05,
      "loss": 1.0643,
      "step": 1022
    },
    {
      "epoch": 0.654510556621881,
      "grad_norm": 2.487199605001362,
      "learning_rate": 1.9934180195482433e-05,
      "loss": 1.0005,
      "step": 1023
    },
    {
      "epoch": 0.655150351887396,
      "grad_norm": 2.2360424595090476,
      "learning_rate": 1.993394262659176e-05,
      "loss": 0.981,
      "step": 1024
    },
    {
      "epoch": 0.6557901471529111,
      "grad_norm": 2.472094185498003,
      "learning_rate": 1.9933704631155257e-05,
      "loss": 0.9746,
      "step": 1025
    },
    {
      "epoch": 0.6564299424184261,
      "grad_norm": 3.227514568681453,
      "learning_rate": 1.993346620918315e-05,
      "loss": 0.9612,
      "step": 1026
    },
    {
      "epoch": 0.6570697376839412,
      "grad_norm": 2.471682515258181,
      "learning_rate": 1.993322736068567e-05,
      "loss": 0.954,
      "step": 1027
    },
    {
      "epoch": 0.6577095329494562,
      "grad_norm": 2.3967763963484194,
      "learning_rate": 1.993298808567308e-05,
      "loss": 0.925,
      "step": 1028
    },
    {
      "epoch": 0.6583493282149712,
      "grad_norm": 2.879221543302376,
      "learning_rate": 1.9932748384155657e-05,
      "loss": 0.9696,
      "step": 1029
    },
    {
      "epoch": 0.6589891234804862,
      "grad_norm": 2.3104645896639613,
      "learning_rate": 1.9932508256143683e-05,
      "loss": 1.0451,
      "step": 1030
    },
    {
      "epoch": 0.6596289187460013,
      "grad_norm": 2.513492697663256,
      "learning_rate": 1.993226770164747e-05,
      "loss": 0.9967,
      "step": 1031
    },
    {
      "epoch": 0.6602687140115163,
      "grad_norm": 2.8518779602294178,
      "learning_rate": 1.9932026720677352e-05,
      "loss": 0.9045,
      "step": 1032
    },
    {
      "epoch": 0.6609085092770314,
      "grad_norm": 2.613002575158052,
      "learning_rate": 1.9931785313243677e-05,
      "loss": 1.0195,
      "step": 1033
    },
    {
      "epoch": 0.6615483045425464,
      "grad_norm": 2.527371367496481,
      "learning_rate": 1.9931543479356804e-05,
      "loss": 0.999,
      "step": 1034
    },
    {
      "epoch": 0.6621880998080614,
      "grad_norm": 2.2357091109056553,
      "learning_rate": 1.9931301219027124e-05,
      "loss": 0.9411,
      "step": 1035
    },
    {
      "epoch": 0.6628278950735764,
      "grad_norm": 3.45236774318089,
      "learning_rate": 1.9931058532265033e-05,
      "loss": 1.0121,
      "step": 1036
    },
    {
      "epoch": 0.6634676903390915,
      "grad_norm": 3.2365135773951033,
      "learning_rate": 1.9930815419080955e-05,
      "loss": 0.9838,
      "step": 1037
    },
    {
      "epoch": 0.6641074856046065,
      "grad_norm": 3.3469317880730776,
      "learning_rate": 1.9930571879485333e-05,
      "loss": 1.0018,
      "step": 1038
    },
    {
      "epoch": 0.6647472808701216,
      "grad_norm": 2.90893569206141,
      "learning_rate": 1.9930327913488613e-05,
      "loss": 0.9532,
      "step": 1039
    },
    {
      "epoch": 0.6653870761356366,
      "grad_norm": 2.5867287137596433,
      "learning_rate": 1.993008352110128e-05,
      "loss": 0.9367,
      "step": 1040
    },
    {
      "epoch": 0.6660268714011516,
      "grad_norm": 2.5424783208334776,
      "learning_rate": 1.9929838702333825e-05,
      "loss": 1.0229,
      "step": 1041
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 2.684576655401913,
      "learning_rate": 1.992959345719676e-05,
      "loss": 0.9684,
      "step": 1042
    },
    {
      "epoch": 0.6673064619321817,
      "grad_norm": 2.841234149259597,
      "learning_rate": 1.9929347785700617e-05,
      "loss": 0.9745,
      "step": 1043
    },
    {
      "epoch": 0.6679462571976967,
      "grad_norm": 3.252807699019323,
      "learning_rate": 1.9929101687855942e-05,
      "loss": 0.9824,
      "step": 1044
    },
    {
      "epoch": 0.6685860524632118,
      "grad_norm": 3.1006968864225235,
      "learning_rate": 1.9928855163673303e-05,
      "loss": 1.0069,
      "step": 1045
    },
    {
      "epoch": 0.6692258477287268,
      "grad_norm": 2.6682743454755182,
      "learning_rate": 1.9928608213163284e-05,
      "loss": 0.9842,
      "step": 1046
    },
    {
      "epoch": 0.6698656429942419,
      "grad_norm": 2.190659571755242,
      "learning_rate": 1.992836083633649e-05,
      "loss": 0.9821,
      "step": 1047
    },
    {
      "epoch": 0.6705054382597568,
      "grad_norm": 3.8485246396826907,
      "learning_rate": 1.9928113033203545e-05,
      "loss": 1.0272,
      "step": 1048
    },
    {
      "epoch": 0.6711452335252719,
      "grad_norm": 3.0040621525348414,
      "learning_rate": 1.9927864803775084e-05,
      "loss": 0.9714,
      "step": 1049
    },
    {
      "epoch": 0.6717850287907869,
      "grad_norm": 2.483802133624241,
      "learning_rate": 1.9927616148061772e-05,
      "loss": 0.9813,
      "step": 1050
    },
    {
      "epoch": 0.672424824056302,
      "grad_norm": 2.2280544884946805,
      "learning_rate": 1.992736706607428e-05,
      "loss": 1.0,
      "step": 1051
    },
    {
      "epoch": 0.673064619321817,
      "grad_norm": 2.3449004425366775,
      "learning_rate": 1.9927117557823306e-05,
      "loss": 0.9794,
      "step": 1052
    },
    {
      "epoch": 0.6737044145873321,
      "grad_norm": 2.627684921408482,
      "learning_rate": 1.9926867623319562e-05,
      "loss": 1.004,
      "step": 1053
    },
    {
      "epoch": 0.674344209852847,
      "grad_norm": 2.7133400418871987,
      "learning_rate": 1.9926617262573783e-05,
      "loss": 1.0038,
      "step": 1054
    },
    {
      "epoch": 0.6749840051183621,
      "grad_norm": 2.8192885860639203,
      "learning_rate": 1.9926366475596718e-05,
      "loss": 0.9544,
      "step": 1055
    },
    {
      "epoch": 0.6756238003838771,
      "grad_norm": 2.6654800852090395,
      "learning_rate": 1.992611526239913e-05,
      "loss": 1.0101,
      "step": 1056
    },
    {
      "epoch": 0.6762635956493922,
      "grad_norm": 2.736326799549185,
      "learning_rate": 1.992586362299181e-05,
      "loss": 0.997,
      "step": 1057
    },
    {
      "epoch": 0.6769033909149073,
      "grad_norm": 2.758261391630249,
      "learning_rate": 1.9925611557385567e-05,
      "loss": 0.96,
      "step": 1058
    },
    {
      "epoch": 0.6775431861804223,
      "grad_norm": 2.6654733968622373,
      "learning_rate": 1.9925359065591218e-05,
      "loss": 0.9608,
      "step": 1059
    },
    {
      "epoch": 0.6781829814459372,
      "grad_norm": 2.4534332752923302,
      "learning_rate": 1.9925106147619608e-05,
      "loss": 0.9121,
      "step": 1060
    },
    {
      "epoch": 0.6788227767114523,
      "grad_norm": 2.3670685428935605,
      "learning_rate": 1.9924852803481595e-05,
      "loss": 0.994,
      "step": 1061
    },
    {
      "epoch": 0.6794625719769674,
      "grad_norm": 2.724645989029944,
      "learning_rate": 1.9924599033188055e-05,
      "loss": 0.9636,
      "step": 1062
    },
    {
      "epoch": 0.6801023672424824,
      "grad_norm": 2.6551130386942043,
      "learning_rate": 1.9924344836749888e-05,
      "loss": 1.0049,
      "step": 1063
    },
    {
      "epoch": 0.6807421625079975,
      "grad_norm": 2.9969296044831064,
      "learning_rate": 1.992409021417801e-05,
      "loss": 1.0013,
      "step": 1064
    },
    {
      "epoch": 0.6813819577735125,
      "grad_norm": 2.5130698304066863,
      "learning_rate": 1.9923835165483346e-05,
      "loss": 0.9988,
      "step": 1065
    },
    {
      "epoch": 0.6820217530390275,
      "grad_norm": 2.4554653572258944,
      "learning_rate": 1.9923579690676858e-05,
      "loss": 1.0209,
      "step": 1066
    },
    {
      "epoch": 0.6826615483045425,
      "grad_norm": 2.61571203669903,
      "learning_rate": 1.992332378976951e-05,
      "loss": 1.0169,
      "step": 1067
    },
    {
      "epoch": 0.6833013435700576,
      "grad_norm": 2.8517414490112367,
      "learning_rate": 1.9923067462772292e-05,
      "loss": 0.9878,
      "step": 1068
    },
    {
      "epoch": 0.6839411388355726,
      "grad_norm": 2.455734321850297,
      "learning_rate": 1.9922810709696207e-05,
      "loss": 1.0482,
      "step": 1069
    },
    {
      "epoch": 0.6845809341010877,
      "grad_norm": 2.829332070472135,
      "learning_rate": 1.9922553530552283e-05,
      "loss": 1.0311,
      "step": 1070
    },
    {
      "epoch": 0.6852207293666027,
      "grad_norm": 2.289131561073131,
      "learning_rate": 1.992229592535156e-05,
      "loss": 1.0288,
      "step": 1071
    },
    {
      "epoch": 0.6858605246321178,
      "grad_norm": 2.4356106651648544,
      "learning_rate": 1.99220378941051e-05,
      "loss": 0.9071,
      "step": 1072
    },
    {
      "epoch": 0.6865003198976327,
      "grad_norm": 2.4644478502667098,
      "learning_rate": 1.9921779436823983e-05,
      "loss": 1.0042,
      "step": 1073
    },
    {
      "epoch": 0.6871401151631478,
      "grad_norm": 2.7414972619417823,
      "learning_rate": 1.9921520553519308e-05,
      "loss": 0.959,
      "step": 1074
    },
    {
      "epoch": 0.6877799104286628,
      "grad_norm": 3.088378213678756,
      "learning_rate": 1.9921261244202186e-05,
      "loss": 0.9644,
      "step": 1075
    },
    {
      "epoch": 0.6884197056941779,
      "grad_norm": 2.0779145671847616,
      "learning_rate": 1.9921001508883758e-05,
      "loss": 0.961,
      "step": 1076
    },
    {
      "epoch": 0.6890595009596929,
      "grad_norm": 2.341441700953349,
      "learning_rate": 1.992074134757517e-05,
      "loss": 1.0129,
      "step": 1077
    },
    {
      "epoch": 0.689699296225208,
      "grad_norm": 2.387365786097861,
      "learning_rate": 1.99204807602876e-05,
      "loss": 0.9872,
      "step": 1078
    },
    {
      "epoch": 0.6903390914907229,
      "grad_norm": 2.592444300246129,
      "learning_rate": 1.992021974703223e-05,
      "loss": 0.9812,
      "step": 1079
    },
    {
      "epoch": 0.690978886756238,
      "grad_norm": 2.6391422687589645,
      "learning_rate": 1.9919958307820275e-05,
      "loss": 0.971,
      "step": 1080
    },
    {
      "epoch": 0.691618682021753,
      "grad_norm": 2.3631296961209056,
      "learning_rate": 1.9919696442662953e-05,
      "loss": 1.0016,
      "step": 1081
    },
    {
      "epoch": 0.6922584772872681,
      "grad_norm": 2.4022290917938665,
      "learning_rate": 1.9919434151571512e-05,
      "loss": 1.0123,
      "step": 1082
    },
    {
      "epoch": 0.6928982725527831,
      "grad_norm": 2.4946898211637425,
      "learning_rate": 1.9919171434557217e-05,
      "loss": 0.9519,
      "step": 1083
    },
    {
      "epoch": 0.6935380678182982,
      "grad_norm": 2.868635169897106,
      "learning_rate": 1.991890829163134e-05,
      "loss": 1.0141,
      "step": 1084
    },
    {
      "epoch": 0.6941778630838131,
      "grad_norm": 2.502171442622021,
      "learning_rate": 1.991864472280519e-05,
      "loss": 0.9639,
      "step": 1085
    },
    {
      "epoch": 0.6948176583493282,
      "grad_norm": 2.2956299047478024,
      "learning_rate": 1.9918380728090078e-05,
      "loss": 1.0448,
      "step": 1086
    },
    {
      "epoch": 0.6954574536148432,
      "grad_norm": 2.511855665739341,
      "learning_rate": 1.9918116307497336e-05,
      "loss": 0.9718,
      "step": 1087
    },
    {
      "epoch": 0.6960972488803583,
      "grad_norm": 2.5520137156292564,
      "learning_rate": 1.991785146103833e-05,
      "loss": 0.9699,
      "step": 1088
    },
    {
      "epoch": 0.6967370441458733,
      "grad_norm": 2.5751198001698445,
      "learning_rate": 1.991758618872442e-05,
      "loss": 1.0229,
      "step": 1089
    },
    {
      "epoch": 0.6973768394113884,
      "grad_norm": 2.5608434247649354,
      "learning_rate": 1.9917320490567003e-05,
      "loss": 0.9748,
      "step": 1090
    },
    {
      "epoch": 0.6980166346769033,
      "grad_norm": 2.577932979548477,
      "learning_rate": 1.991705436657749e-05,
      "loss": 1.0313,
      "step": 1091
    },
    {
      "epoch": 0.6986564299424184,
      "grad_norm": 2.2491903178016917,
      "learning_rate": 1.9916787816767294e-05,
      "loss": 0.997,
      "step": 1092
    },
    {
      "epoch": 0.6992962252079334,
      "grad_norm": 2.635433154331128,
      "learning_rate": 1.991652084114788e-05,
      "loss": 1.0502,
      "step": 1093
    },
    {
      "epoch": 0.6999360204734485,
      "grad_norm": 2.244754757715561,
      "learning_rate": 1.9916253439730693e-05,
      "loss": 0.9532,
      "step": 1094
    },
    {
      "epoch": 0.7005758157389635,
      "grad_norm": 2.4835505131692583,
      "learning_rate": 1.991598561252723e-05,
      "loss": 0.9832,
      "step": 1095
    },
    {
      "epoch": 0.7012156110044786,
      "grad_norm": 2.2593290766442773,
      "learning_rate": 1.9915717359548975e-05,
      "loss": 0.9868,
      "step": 1096
    },
    {
      "epoch": 0.7018554062699937,
      "grad_norm": 2.5535183965153267,
      "learning_rate": 1.991544868080746e-05,
      "loss": 1.0308,
      "step": 1097
    },
    {
      "epoch": 0.7024952015355086,
      "grad_norm": 2.7267020045923838,
      "learning_rate": 1.9915179576314222e-05,
      "loss": 1.0285,
      "step": 1098
    },
    {
      "epoch": 0.7031349968010236,
      "grad_norm": 2.1870177031529145,
      "learning_rate": 1.9914910046080803e-05,
      "loss": 1.0362,
      "step": 1099
    },
    {
      "epoch": 0.7037747920665387,
      "grad_norm": 2.8249090278881064,
      "learning_rate": 1.991464009011879e-05,
      "loss": 0.9886,
      "step": 1100
    },
    {
      "epoch": 0.7044145873320538,
      "grad_norm": 2.647743369821682,
      "learning_rate": 1.9914369708439762e-05,
      "loss": 1.0166,
      "step": 1101
    },
    {
      "epoch": 0.7050543825975688,
      "grad_norm": 2.137589387693219,
      "learning_rate": 1.9914098901055337e-05,
      "loss": 0.9776,
      "step": 1102
    },
    {
      "epoch": 0.7056941778630839,
      "grad_norm": 2.1777514083687413,
      "learning_rate": 1.9913827667977144e-05,
      "loss": 0.9839,
      "step": 1103
    },
    {
      "epoch": 0.7063339731285988,
      "grad_norm": 3.091741716348666,
      "learning_rate": 1.9913556009216825e-05,
      "loss": 1.0052,
      "step": 1104
    },
    {
      "epoch": 0.7069737683941139,
      "grad_norm": 2.842094458276322,
      "learning_rate": 1.9913283924786047e-05,
      "loss": 0.929,
      "step": 1105
    },
    {
      "epoch": 0.7076135636596289,
      "grad_norm": 3.099658332854651,
      "learning_rate": 1.991301141469649e-05,
      "loss": 0.9514,
      "step": 1106
    },
    {
      "epoch": 0.708253358925144,
      "grad_norm": 2.665029709714448,
      "learning_rate": 1.9912738478959856e-05,
      "loss": 0.9521,
      "step": 1107
    },
    {
      "epoch": 0.708893154190659,
      "grad_norm": 2.605927177829618,
      "learning_rate": 1.9912465117587865e-05,
      "loss": 0.9515,
      "step": 1108
    },
    {
      "epoch": 0.7095329494561741,
      "grad_norm": 2.5732327011568317,
      "learning_rate": 1.9912191330592258e-05,
      "loss": 0.9571,
      "step": 1109
    },
    {
      "epoch": 0.710172744721689,
      "grad_norm": 2.6631841943072017,
      "learning_rate": 1.9911917117984783e-05,
      "loss": 0.9875,
      "step": 1110
    },
    {
      "epoch": 0.7108125399872041,
      "grad_norm": 2.7803307980341794,
      "learning_rate": 1.991164247977722e-05,
      "loss": 0.9849,
      "step": 1111
    },
    {
      "epoch": 0.7114523352527191,
      "grad_norm": 2.758972636066955,
      "learning_rate": 1.9911367415981364e-05,
      "loss": 1.0278,
      "step": 1112
    },
    {
      "epoch": 0.7120921305182342,
      "grad_norm": 2.734395352854776,
      "learning_rate": 1.9911091926609018e-05,
      "loss": 1.0229,
      "step": 1113
    },
    {
      "epoch": 0.7127319257837492,
      "grad_norm": 2.6514254273304605,
      "learning_rate": 1.991081601167202e-05,
      "loss": 0.9584,
      "step": 1114
    },
    {
      "epoch": 0.7133717210492643,
      "grad_norm": 2.4958445460336343,
      "learning_rate": 1.9910539671182208e-05,
      "loss": 0.922,
      "step": 1115
    },
    {
      "epoch": 0.7140115163147792,
      "grad_norm": 2.2707320795266526,
      "learning_rate": 1.9910262905151453e-05,
      "loss": 1.0415,
      "step": 1116
    },
    {
      "epoch": 0.7146513115802943,
      "grad_norm": 2.3222311127210733,
      "learning_rate": 1.9909985713591634e-05,
      "loss": 0.9787,
      "step": 1117
    },
    {
      "epoch": 0.7152911068458093,
      "grad_norm": 2.518214933613484,
      "learning_rate": 1.9909708096514663e-05,
      "loss": 1.0445,
      "step": 1118
    },
    {
      "epoch": 0.7159309021113244,
      "grad_norm": 2.4045608602194757,
      "learning_rate": 1.9909430053932453e-05,
      "loss": 1.015,
      "step": 1119
    },
    {
      "epoch": 0.7165706973768394,
      "grad_norm": 2.3273283334197683,
      "learning_rate": 1.9909151585856943e-05,
      "loss": 1.0367,
      "step": 1120
    },
    {
      "epoch": 0.7172104926423545,
      "grad_norm": 3.0320637353156945,
      "learning_rate": 1.990887269230009e-05,
      "loss": 0.9804,
      "step": 1121
    },
    {
      "epoch": 0.7178502879078695,
      "grad_norm": 2.3036007862654246,
      "learning_rate": 1.9908593373273873e-05,
      "loss": 0.9613,
      "step": 1122
    },
    {
      "epoch": 0.7184900831733845,
      "grad_norm": 2.5916112138397143,
      "learning_rate": 1.990831362879028e-05,
      "loss": 0.924,
      "step": 1123
    },
    {
      "epoch": 0.7191298784388995,
      "grad_norm": 2.582479766357754,
      "learning_rate": 1.9908033458861326e-05,
      "loss": 0.9555,
      "step": 1124
    },
    {
      "epoch": 0.7197696737044146,
      "grad_norm": 2.4255938793907035,
      "learning_rate": 1.9907752863499042e-05,
      "loss": 0.9661,
      "step": 1125
    },
    {
      "epoch": 0.7204094689699296,
      "grad_norm": 2.7289202800913865,
      "learning_rate": 1.9907471842715474e-05,
      "loss": 0.9688,
      "step": 1126
    },
    {
      "epoch": 0.7210492642354447,
      "grad_norm": 2.2330797776845457,
      "learning_rate": 1.9907190396522686e-05,
      "loss": 0.9684,
      "step": 1127
    },
    {
      "epoch": 0.7216890595009597,
      "grad_norm": 2.325880811181374,
      "learning_rate": 1.9906908524932768e-05,
      "loss": 1.0541,
      "step": 1128
    },
    {
      "epoch": 0.7223288547664747,
      "grad_norm": 2.3254471726134116,
      "learning_rate": 1.990662622795782e-05,
      "loss": 0.9974,
      "step": 1129
    },
    {
      "epoch": 0.7229686500319897,
      "grad_norm": 2.37838742047191,
      "learning_rate": 1.990634350560997e-05,
      "loss": 0.9259,
      "step": 1130
    },
    {
      "epoch": 0.7236084452975048,
      "grad_norm": 2.5438736058301203,
      "learning_rate": 1.9906060357901344e-05,
      "loss": 0.959,
      "step": 1131
    },
    {
      "epoch": 0.7242482405630198,
      "grad_norm": 2.781468159116812,
      "learning_rate": 1.990577678484411e-05,
      "loss": 1.0602,
      "step": 1132
    },
    {
      "epoch": 0.7248880358285349,
      "grad_norm": 2.310911292043918,
      "learning_rate": 1.9905492786450443e-05,
      "loss": 0.964,
      "step": 1133
    },
    {
      "epoch": 0.72552783109405,
      "grad_norm": 2.4046300971510837,
      "learning_rate": 1.9905208362732538e-05,
      "loss": 0.9268,
      "step": 1134
    },
    {
      "epoch": 0.7261676263595649,
      "grad_norm": 2.5377210507904566,
      "learning_rate": 1.9904923513702602e-05,
      "loss": 1.0262,
      "step": 1135
    },
    {
      "epoch": 0.7268074216250799,
      "grad_norm": 2.506126262995948,
      "learning_rate": 1.9904638239372873e-05,
      "loss": 0.9957,
      "step": 1136
    },
    {
      "epoch": 0.727447216890595,
      "grad_norm": 2.974028437463711,
      "learning_rate": 1.9904352539755595e-05,
      "loss": 0.9862,
      "step": 1137
    },
    {
      "epoch": 0.72808701215611,
      "grad_norm": 2.341893077689245,
      "learning_rate": 1.9904066414863038e-05,
      "loss": 0.994,
      "step": 1138
    },
    {
      "epoch": 0.7287268074216251,
      "grad_norm": 2.507282260979315,
      "learning_rate": 1.990377986470749e-05,
      "loss": 0.9675,
      "step": 1139
    },
    {
      "epoch": 0.7293666026871402,
      "grad_norm": 3.186508981805413,
      "learning_rate": 1.9903492889301248e-05,
      "loss": 1.0021,
      "step": 1140
    },
    {
      "epoch": 0.7300063979526551,
      "grad_norm": 2.464989381584744,
      "learning_rate": 1.9903205488656638e-05,
      "loss": 0.967,
      "step": 1141
    },
    {
      "epoch": 0.7306461932181701,
      "grad_norm": 2.4395887900193842,
      "learning_rate": 1.9902917662786004e-05,
      "loss": 0.9863,
      "step": 1142
    },
    {
      "epoch": 0.7312859884836852,
      "grad_norm": 2.504140921024202,
      "learning_rate": 1.99026294117017e-05,
      "loss": 0.9561,
      "step": 1143
    },
    {
      "epoch": 0.7319257837492003,
      "grad_norm": 3.083455927364127,
      "learning_rate": 1.9902340735416102e-05,
      "loss": 0.9861,
      "step": 1144
    },
    {
      "epoch": 0.7325655790147153,
      "grad_norm": 2.45167718098798,
      "learning_rate": 1.9902051633941612e-05,
      "loss": 1.0097,
      "step": 1145
    },
    {
      "epoch": 0.7332053742802304,
      "grad_norm": 2.4156532782188753,
      "learning_rate": 1.9901762107290637e-05,
      "loss": 1.0279,
      "step": 1146
    },
    {
      "epoch": 0.7338451695457454,
      "grad_norm": 2.546541396493569,
      "learning_rate": 1.990147215547561e-05,
      "loss": 0.9764,
      "step": 1147
    },
    {
      "epoch": 0.7344849648112604,
      "grad_norm": 2.7167876259430064,
      "learning_rate": 1.9901181778508985e-05,
      "loss": 1.0066,
      "step": 1148
    },
    {
      "epoch": 0.7351247600767754,
      "grad_norm": 2.429381395867755,
      "learning_rate": 1.9900890976403224e-05,
      "loss": 0.9651,
      "step": 1149
    },
    {
      "epoch": 0.7357645553422905,
      "grad_norm": 2.5414275894838187,
      "learning_rate": 1.990059974917082e-05,
      "loss": 0.9564,
      "step": 1150
    },
    {
      "epoch": 0.7364043506078055,
      "grad_norm": 2.8154640023963258,
      "learning_rate": 1.9900308096824274e-05,
      "loss": 0.9736,
      "step": 1151
    },
    {
      "epoch": 0.7370441458733206,
      "grad_norm": 2.633270000575257,
      "learning_rate": 1.990001601937611e-05,
      "loss": 0.9546,
      "step": 1152
    },
    {
      "epoch": 0.7376839411388356,
      "grad_norm": 2.6438094191548354,
      "learning_rate": 1.9899723516838865e-05,
      "loss": 1.0095,
      "step": 1153
    },
    {
      "epoch": 0.7383237364043506,
      "grad_norm": 2.7375835594545004,
      "learning_rate": 1.9899430589225104e-05,
      "loss": 1.0206,
      "step": 1154
    },
    {
      "epoch": 0.7389635316698656,
      "grad_norm": 2.6472183735135664,
      "learning_rate": 1.9899137236547403e-05,
      "loss": 0.9736,
      "step": 1155
    },
    {
      "epoch": 0.7396033269353807,
      "grad_norm": 2.611902644605817,
      "learning_rate": 1.989884345881836e-05,
      "loss": 0.9611,
      "step": 1156
    },
    {
      "epoch": 0.7402431222008957,
      "grad_norm": 2.4940450842280812,
      "learning_rate": 1.989854925605059e-05,
      "loss": 0.9135,
      "step": 1157
    },
    {
      "epoch": 0.7408829174664108,
      "grad_norm": 2.3228927388944403,
      "learning_rate": 1.989825462825672e-05,
      "loss": 0.9456,
      "step": 1158
    },
    {
      "epoch": 0.7415227127319258,
      "grad_norm": 2.4784573697093673,
      "learning_rate": 1.9897959575449403e-05,
      "loss": 0.9822,
      "step": 1159
    },
    {
      "epoch": 0.7421625079974408,
      "grad_norm": 2.4338536748909823,
      "learning_rate": 1.989766409764131e-05,
      "loss": 0.9656,
      "step": 1160
    },
    {
      "epoch": 0.7428023032629558,
      "grad_norm": 2.5950963595761425,
      "learning_rate": 1.9897368194845126e-05,
      "loss": 1.028,
      "step": 1161
    },
    {
      "epoch": 0.7434420985284709,
      "grad_norm": 2.240871366246345,
      "learning_rate": 1.9897071867073557e-05,
      "loss": 1.0333,
      "step": 1162
    },
    {
      "epoch": 0.7440818937939859,
      "grad_norm": 2.536646981976395,
      "learning_rate": 1.989677511433933e-05,
      "loss": 0.9514,
      "step": 1163
    },
    {
      "epoch": 0.744721689059501,
      "grad_norm": 2.75169508229164,
      "learning_rate": 1.9896477936655185e-05,
      "loss": 0.9801,
      "step": 1164
    },
    {
      "epoch": 0.745361484325016,
      "grad_norm": 2.5911021303436335,
      "learning_rate": 1.989618033403388e-05,
      "loss": 0.9642,
      "step": 1165
    },
    {
      "epoch": 0.746001279590531,
      "grad_norm": 2.3600668473997892,
      "learning_rate": 1.9895882306488197e-05,
      "loss": 0.9455,
      "step": 1166
    },
    {
      "epoch": 0.746641074856046,
      "grad_norm": 2.4554144651243477,
      "learning_rate": 1.989558385403093e-05,
      "loss": 0.9525,
      "step": 1167
    },
    {
      "epoch": 0.7472808701215611,
      "grad_norm": 3.017471701752697,
      "learning_rate": 1.9895284976674896e-05,
      "loss": 0.9676,
      "step": 1168
    },
    {
      "epoch": 0.7479206653870761,
      "grad_norm": 2.1435869643949044,
      "learning_rate": 1.9894985674432926e-05,
      "loss": 0.9333,
      "step": 1169
    },
    {
      "epoch": 0.7485604606525912,
      "grad_norm": 2.273691471761247,
      "learning_rate": 1.9894685947317874e-05,
      "loss": 0.9907,
      "step": 1170
    },
    {
      "epoch": 0.7492002559181062,
      "grad_norm": 2.69007322378843,
      "learning_rate": 1.989438579534261e-05,
      "loss": 0.9847,
      "step": 1171
    },
    {
      "epoch": 0.7498400511836213,
      "grad_norm": 2.587453182546843,
      "learning_rate": 1.9894085218520018e-05,
      "loss": 0.9448,
      "step": 1172
    },
    {
      "epoch": 0.7504798464491362,
      "grad_norm": 2.7917187641194277,
      "learning_rate": 1.9893784216863012e-05,
      "loss": 0.9466,
      "step": 1173
    },
    {
      "epoch": 0.7511196417146513,
      "grad_norm": 2.544853600471547,
      "learning_rate": 1.9893482790384504e-05,
      "loss": 0.9765,
      "step": 1174
    },
    {
      "epoch": 0.7517594369801663,
      "grad_norm": 2.609695834159028,
      "learning_rate": 1.989318093909745e-05,
      "loss": 0.9977,
      "step": 1175
    },
    {
      "epoch": 0.7523992322456814,
      "grad_norm": 2.758029720620586,
      "learning_rate": 1.98928786630148e-05,
      "loss": 1.0118,
      "step": 1176
    },
    {
      "epoch": 0.7530390275111964,
      "grad_norm": 2.5354888479091824,
      "learning_rate": 1.9892575962149544e-05,
      "loss": 0.974,
      "step": 1177
    },
    {
      "epoch": 0.7536788227767115,
      "grad_norm": 2.582606198711522,
      "learning_rate": 1.989227283651467e-05,
      "loss": 0.9473,
      "step": 1178
    },
    {
      "epoch": 0.7543186180422264,
      "grad_norm": 2.3765296024889553,
      "learning_rate": 1.9891969286123197e-05,
      "loss": 1.0061,
      "step": 1179
    },
    {
      "epoch": 0.7549584133077415,
      "grad_norm": 2.1500968112952425,
      "learning_rate": 1.9891665310988162e-05,
      "loss": 1.0177,
      "step": 1180
    },
    {
      "epoch": 0.7555982085732565,
      "grad_norm": 2.556221056831829,
      "learning_rate": 1.9891360911122612e-05,
      "loss": 0.9484,
      "step": 1181
    },
    {
      "epoch": 0.7562380038387716,
      "grad_norm": 2.4059623451244465,
      "learning_rate": 1.989105608653962e-05,
      "loss": 0.967,
      "step": 1182
    },
    {
      "epoch": 0.7568777991042867,
      "grad_norm": 3.029744235402811,
      "learning_rate": 1.9890750837252277e-05,
      "loss": 0.9744,
      "step": 1183
    },
    {
      "epoch": 0.7575175943698017,
      "grad_norm": 3.23684295519874,
      "learning_rate": 1.9890445163273683e-05,
      "loss": 0.9809,
      "step": 1184
    },
    {
      "epoch": 0.7581573896353166,
      "grad_norm": 2.6077292603327615,
      "learning_rate": 1.9890139064616973e-05,
      "loss": 0.9812,
      "step": 1185
    },
    {
      "epoch": 0.7587971849008317,
      "grad_norm": 2.269664959131756,
      "learning_rate": 1.988983254129528e-05,
      "loss": 0.9219,
      "step": 1186
    },
    {
      "epoch": 0.7594369801663468,
      "grad_norm": 2.394119127190656,
      "learning_rate": 1.9889525593321773e-05,
      "loss": 0.9447,
      "step": 1187
    },
    {
      "epoch": 0.7600767754318618,
      "grad_norm": 2.5703841355750505,
      "learning_rate": 1.988921822070963e-05,
      "loss": 0.9476,
      "step": 1188
    },
    {
      "epoch": 0.7607165706973769,
      "grad_norm": 2.959962644372845,
      "learning_rate": 1.9888910423472044e-05,
      "loss": 1.0073,
      "step": 1189
    },
    {
      "epoch": 0.7613563659628919,
      "grad_norm": 3.040664262566946,
      "learning_rate": 1.988860220162224e-05,
      "loss": 0.9484,
      "step": 1190
    },
    {
      "epoch": 0.761996161228407,
      "grad_norm": 2.4993556600943028,
      "learning_rate": 1.9888293555173445e-05,
      "loss": 0.941,
      "step": 1191
    },
    {
      "epoch": 0.7626359564939219,
      "grad_norm": 2.0206905358641993,
      "learning_rate": 1.9887984484138916e-05,
      "loss": 0.91,
      "step": 1192
    },
    {
      "epoch": 0.763275751759437,
      "grad_norm": 2.7210000555631675,
      "learning_rate": 1.9887674988531924e-05,
      "loss": 1.0113,
      "step": 1193
    },
    {
      "epoch": 0.763915547024952,
      "grad_norm": 2.2995291138776177,
      "learning_rate": 1.9887365068365753e-05,
      "loss": 1.0111,
      "step": 1194
    },
    {
      "epoch": 0.7645553422904671,
      "grad_norm": 2.224902597556323,
      "learning_rate": 1.9887054723653717e-05,
      "loss": 1.0162,
      "step": 1195
    },
    {
      "epoch": 0.7651951375559821,
      "grad_norm": 2.5284594480863647,
      "learning_rate": 1.9886743954409137e-05,
      "loss": 1.016,
      "step": 1196
    },
    {
      "epoch": 0.7658349328214972,
      "grad_norm": 2.6172606030652408,
      "learning_rate": 1.988643276064536e-05,
      "loss": 0.9865,
      "step": 1197
    },
    {
      "epoch": 0.7664747280870121,
      "grad_norm": 2.4718423362197193,
      "learning_rate": 1.9886121142375745e-05,
      "loss": 0.9691,
      "step": 1198
    },
    {
      "epoch": 0.7671145233525272,
      "grad_norm": 2.420326204845833,
      "learning_rate": 1.9885809099613677e-05,
      "loss": 0.9365,
      "step": 1199
    },
    {
      "epoch": 0.7677543186180422,
      "grad_norm": 2.5060416923501134,
      "learning_rate": 1.988549663237255e-05,
      "loss": 0.9202,
      "step": 1200
    },
    {
      "epoch": 0.7683941138835573,
      "grad_norm": 2.2364652408139967,
      "learning_rate": 1.9885183740665783e-05,
      "loss": 0.9716,
      "step": 1201
    },
    {
      "epoch": 0.7690339091490723,
      "grad_norm": 2.8843635659341462,
      "learning_rate": 1.988487042450681e-05,
      "loss": 0.8801,
      "step": 1202
    },
    {
      "epoch": 0.7696737044145874,
      "grad_norm": 2.473775624763101,
      "learning_rate": 1.9884556683909088e-05,
      "loss": 0.9781,
      "step": 1203
    },
    {
      "epoch": 0.7703134996801023,
      "grad_norm": 2.7249211002536757,
      "learning_rate": 1.988424251888608e-05,
      "loss": 0.9556,
      "step": 1204
    },
    {
      "epoch": 0.7709532949456174,
      "grad_norm": 2.3627616168247276,
      "learning_rate": 1.9883927929451284e-05,
      "loss": 0.9226,
      "step": 1205
    },
    {
      "epoch": 0.7715930902111324,
      "grad_norm": 2.3795748665648047,
      "learning_rate": 1.9883612915618204e-05,
      "loss": 0.9895,
      "step": 1206
    },
    {
      "epoch": 0.7722328854766475,
      "grad_norm": 2.5004615063897653,
      "learning_rate": 1.988329747740037e-05,
      "loss": 0.9388,
      "step": 1207
    },
    {
      "epoch": 0.7728726807421625,
      "grad_norm": 2.6314000424024875,
      "learning_rate": 1.988298161481132e-05,
      "loss": 0.954,
      "step": 1208
    },
    {
      "epoch": 0.7735124760076776,
      "grad_norm": 2.5024586488077047,
      "learning_rate": 1.988266532786462e-05,
      "loss": 0.9393,
      "step": 1209
    },
    {
      "epoch": 0.7741522712731925,
      "grad_norm": 2.508211807999536,
      "learning_rate": 1.988234861657385e-05,
      "loss": 0.979,
      "step": 1210
    },
    {
      "epoch": 0.7747920665387076,
      "grad_norm": 2.0467696376554754,
      "learning_rate": 1.9882031480952614e-05,
      "loss": 0.9009,
      "step": 1211
    },
    {
      "epoch": 0.7754318618042226,
      "grad_norm": 2.265829851266147,
      "learning_rate": 1.988171392101452e-05,
      "loss": 0.9372,
      "step": 1212
    },
    {
      "epoch": 0.7760716570697377,
      "grad_norm": 2.5784166767621057,
      "learning_rate": 1.988139593677321e-05,
      "loss": 0.9355,
      "step": 1213
    },
    {
      "epoch": 0.7767114523352527,
      "grad_norm": 2.656904977032184,
      "learning_rate": 1.9881077528242335e-05,
      "loss": 0.983,
      "step": 1214
    },
    {
      "epoch": 0.7773512476007678,
      "grad_norm": 2.1139936050760024,
      "learning_rate": 1.988075869543557e-05,
      "loss": 0.9853,
      "step": 1215
    },
    {
      "epoch": 0.7779910428662828,
      "grad_norm": 2.2563411554277004,
      "learning_rate": 1.9880439438366605e-05,
      "loss": 0.9498,
      "step": 1216
    },
    {
      "epoch": 0.7786308381317978,
      "grad_norm": 2.543590470910679,
      "learning_rate": 1.9880119757049146e-05,
      "loss": 0.9966,
      "step": 1217
    },
    {
      "epoch": 0.7792706333973128,
      "grad_norm": 2.5279402509003543,
      "learning_rate": 1.9879799651496916e-05,
      "loss": 0.9811,
      "step": 1218
    },
    {
      "epoch": 0.7799104286628279,
      "grad_norm": 2.0308031728800384,
      "learning_rate": 1.9879479121723667e-05,
      "loss": 0.9712,
      "step": 1219
    },
    {
      "epoch": 0.780550223928343,
      "grad_norm": 2.5082396398682945,
      "learning_rate": 1.987915816774316e-05,
      "loss": 0.956,
      "step": 1220
    },
    {
      "epoch": 0.781190019193858,
      "grad_norm": 2.4719547934091852,
      "learning_rate": 1.987883678956917e-05,
      "loss": 0.9846,
      "step": 1221
    },
    {
      "epoch": 0.781829814459373,
      "grad_norm": 2.298147201392787,
      "learning_rate": 1.9878514987215505e-05,
      "loss": 0.9466,
      "step": 1222
    },
    {
      "epoch": 0.782469609724888,
      "grad_norm": 2.439953150763386,
      "learning_rate": 1.9878192760695977e-05,
      "loss": 0.9776,
      "step": 1223
    },
    {
      "epoch": 0.783109404990403,
      "grad_norm": 2.3749829281548718,
      "learning_rate": 1.9877870110024426e-05,
      "loss": 0.9641,
      "step": 1224
    },
    {
      "epoch": 0.7837492002559181,
      "grad_norm": 2.280499121456574,
      "learning_rate": 1.9877547035214703e-05,
      "loss": 0.9803,
      "step": 1225
    },
    {
      "epoch": 0.7843889955214332,
      "grad_norm": 2.216980510832572,
      "learning_rate": 1.9877223536280682e-05,
      "loss": 0.9245,
      "step": 1226
    },
    {
      "epoch": 0.7850287907869482,
      "grad_norm": 2.1862630743826266,
      "learning_rate": 1.9876899613236248e-05,
      "loss": 0.9338,
      "step": 1227
    },
    {
      "epoch": 0.7856685860524633,
      "grad_norm": 2.22061518519846,
      "learning_rate": 1.987657526609532e-05,
      "loss": 0.9473,
      "step": 1228
    },
    {
      "epoch": 0.7863083813179782,
      "grad_norm": 2.327811344935565,
      "learning_rate": 1.9876250494871815e-05,
      "loss": 1.0233,
      "step": 1229
    },
    {
      "epoch": 0.7869481765834933,
      "grad_norm": 2.3789817940934874,
      "learning_rate": 1.9875925299579686e-05,
      "loss": 0.9219,
      "step": 1230
    },
    {
      "epoch": 0.7875879718490083,
      "grad_norm": 2.529093087349271,
      "learning_rate": 1.987559968023289e-05,
      "loss": 0.9515,
      "step": 1231
    },
    {
      "epoch": 0.7882277671145234,
      "grad_norm": 2.5745388133243807,
      "learning_rate": 1.987527363684541e-05,
      "loss": 0.9704,
      "step": 1232
    },
    {
      "epoch": 0.7888675623800384,
      "grad_norm": 2.39727775173211,
      "learning_rate": 1.987494716943125e-05,
      "loss": 0.9716,
      "step": 1233
    },
    {
      "epoch": 0.7895073576455535,
      "grad_norm": 2.482504000275741,
      "learning_rate": 1.9874620278004422e-05,
      "loss": 0.985,
      "step": 1234
    },
    {
      "epoch": 0.7901471529110684,
      "grad_norm": 2.147160846649932,
      "learning_rate": 1.9874292962578967e-05,
      "loss": 0.9902,
      "step": 1235
    },
    {
      "epoch": 0.7907869481765835,
      "grad_norm": 2.31645043890959,
      "learning_rate": 1.9873965223168933e-05,
      "loss": 0.9532,
      "step": 1236
    },
    {
      "epoch": 0.7914267434420985,
      "grad_norm": 3.02920800360385,
      "learning_rate": 1.9873637059788403e-05,
      "loss": 0.9823,
      "step": 1237
    },
    {
      "epoch": 0.7920665387076136,
      "grad_norm": 2.7492960240041486,
      "learning_rate": 1.9873308472451458e-05,
      "loss": 1.0155,
      "step": 1238
    },
    {
      "epoch": 0.7927063339731286,
      "grad_norm": 2.2823356680067075,
      "learning_rate": 1.987297946117221e-05,
      "loss": 0.9637,
      "step": 1239
    },
    {
      "epoch": 0.7933461292386437,
      "grad_norm": 2.274432687925364,
      "learning_rate": 1.9872650025964786e-05,
      "loss": 0.9692,
      "step": 1240
    },
    {
      "epoch": 0.7939859245041587,
      "grad_norm": 2.449980333336272,
      "learning_rate": 1.9872320166843336e-05,
      "loss": 1.0234,
      "step": 1241
    },
    {
      "epoch": 0.7946257197696737,
      "grad_norm": 2.659588252477026,
      "learning_rate": 1.9871989883822016e-05,
      "loss": 1.0427,
      "step": 1242
    },
    {
      "epoch": 0.7952655150351887,
      "grad_norm": 2.6030729987012204,
      "learning_rate": 1.9871659176915014e-05,
      "loss": 1.0033,
      "step": 1243
    },
    {
      "epoch": 0.7959053103007038,
      "grad_norm": 2.3415293459915856,
      "learning_rate": 1.9871328046136523e-05,
      "loss": 0.9762,
      "step": 1244
    },
    {
      "epoch": 0.7965451055662188,
      "grad_norm": 2.4545221210513914,
      "learning_rate": 1.9870996491500767e-05,
      "loss": 0.9605,
      "step": 1245
    },
    {
      "epoch": 0.7971849008317339,
      "grad_norm": 2.5988696142997454,
      "learning_rate": 1.9870664513021985e-05,
      "loss": 0.9492,
      "step": 1246
    },
    {
      "epoch": 0.7978246960972489,
      "grad_norm": 2.4042999890452794,
      "learning_rate": 1.9870332110714422e-05,
      "loss": 0.9778,
      "step": 1247
    },
    {
      "epoch": 0.7984644913627639,
      "grad_norm": 2.2187349392815774,
      "learning_rate": 1.986999928459236e-05,
      "loss": 0.9859,
      "step": 1248
    },
    {
      "epoch": 0.7991042866282789,
      "grad_norm": 2.827998450027294,
      "learning_rate": 1.9869666034670083e-05,
      "loss": 0.9626,
      "step": 1249
    },
    {
      "epoch": 0.799744081893794,
      "grad_norm": 2.44646642312712,
      "learning_rate": 1.9869332360961905e-05,
      "loss": 1.0178,
      "step": 1250
    },
    {
      "epoch": 0.800383877159309,
      "grad_norm": 2.464551115974644,
      "learning_rate": 1.986899826348215e-05,
      "loss": 0.977,
      "step": 1251
    },
    {
      "epoch": 0.8010236724248241,
      "grad_norm": 2.431620876394442,
      "learning_rate": 1.9868663742245167e-05,
      "loss": 0.9441,
      "step": 1252
    },
    {
      "epoch": 0.8016634676903391,
      "grad_norm": 2.7507835863260532,
      "learning_rate": 1.9868328797265314e-05,
      "loss": 0.9788,
      "step": 1253
    },
    {
      "epoch": 0.8023032629558541,
      "grad_norm": 2.324811932125556,
      "learning_rate": 1.986799342855698e-05,
      "loss": 1.0145,
      "step": 1254
    },
    {
      "epoch": 0.8029430582213691,
      "grad_norm": 2.145433551358591,
      "learning_rate": 1.9867657636134563e-05,
      "loss": 0.9368,
      "step": 1255
    },
    {
      "epoch": 0.8035828534868842,
      "grad_norm": 2.4584618696711833,
      "learning_rate": 1.986732142001248e-05,
      "loss": 0.987,
      "step": 1256
    },
    {
      "epoch": 0.8042226487523992,
      "grad_norm": 2.5058165744299816,
      "learning_rate": 1.9866984780205166e-05,
      "loss": 0.9734,
      "step": 1257
    },
    {
      "epoch": 0.8048624440179143,
      "grad_norm": 2.608580703389801,
      "learning_rate": 1.9866647716727078e-05,
      "loss": 0.9937,
      "step": 1258
    },
    {
      "epoch": 0.8055022392834293,
      "grad_norm": 2.529498805877149,
      "learning_rate": 1.9866310229592692e-05,
      "loss": 0.9026,
      "step": 1259
    },
    {
      "epoch": 0.8061420345489443,
      "grad_norm": 2.1046609506378364,
      "learning_rate": 1.9865972318816492e-05,
      "loss": 0.9774,
      "step": 1260
    },
    {
      "epoch": 0.8067818298144593,
      "grad_norm": 2.4047168374246124,
      "learning_rate": 1.986563398441299e-05,
      "loss": 0.9195,
      "step": 1261
    },
    {
      "epoch": 0.8074216250799744,
      "grad_norm": 2.3055324398093817,
      "learning_rate": 1.986529522639672e-05,
      "loss": 0.9701,
      "step": 1262
    },
    {
      "epoch": 0.8080614203454894,
      "grad_norm": 2.6386193099199318,
      "learning_rate": 1.9864956044782215e-05,
      "loss": 0.9994,
      "step": 1263
    },
    {
      "epoch": 0.8087012156110045,
      "grad_norm": 2.3686766505775867,
      "learning_rate": 1.9864616439584048e-05,
      "loss": 0.9964,
      "step": 1264
    },
    {
      "epoch": 0.8093410108765196,
      "grad_norm": 2.428828941515484,
      "learning_rate": 1.98642764108168e-05,
      "loss": 0.9743,
      "step": 1265
    },
    {
      "epoch": 0.8099808061420346,
      "grad_norm": 2.504919715531219,
      "learning_rate": 1.986393595849507e-05,
      "loss": 0.9363,
      "step": 1266
    },
    {
      "epoch": 0.8106206014075495,
      "grad_norm": 2.3185107480086318,
      "learning_rate": 1.9863595082633475e-05,
      "loss": 0.9578,
      "step": 1267
    },
    {
      "epoch": 0.8112603966730646,
      "grad_norm": 2.7242432504879766,
      "learning_rate": 1.9863253783246655e-05,
      "loss": 0.9245,
      "step": 1268
    },
    {
      "epoch": 0.8119001919385797,
      "grad_norm": 2.376173689462891,
      "learning_rate": 1.9862912060349263e-05,
      "loss": 0.9893,
      "step": 1269
    },
    {
      "epoch": 0.8125399872040947,
      "grad_norm": 2.485857408564759,
      "learning_rate": 1.986256991395597e-05,
      "loss": 0.9906,
      "step": 1270
    },
    {
      "epoch": 0.8131797824696098,
      "grad_norm": 2.57171696172825,
      "learning_rate": 1.986222734408147e-05,
      "loss": 0.9632,
      "step": 1271
    },
    {
      "epoch": 0.8138195777351248,
      "grad_norm": 2.9582338427676738,
      "learning_rate": 1.986188435074047e-05,
      "loss": 0.91,
      "step": 1272
    },
    {
      "epoch": 0.8144593730006398,
      "grad_norm": 2.2773518563529147,
      "learning_rate": 1.98615409339477e-05,
      "loss": 0.9435,
      "step": 1273
    },
    {
      "epoch": 0.8150991682661548,
      "grad_norm": 2.2393028137185538,
      "learning_rate": 1.9861197093717908e-05,
      "loss": 0.9625,
      "step": 1274
    },
    {
      "epoch": 0.8157389635316699,
      "grad_norm": 2.817360251169623,
      "learning_rate": 1.986085283006585e-05,
      "loss": 1.0181,
      "step": 1275
    },
    {
      "epoch": 0.8163787587971849,
      "grad_norm": 2.686175465963356,
      "learning_rate": 1.9860508143006317e-05,
      "loss": 0.9502,
      "step": 1276
    },
    {
      "epoch": 0.8170185540627,
      "grad_norm": 2.220836478518023,
      "learning_rate": 1.98601630325541e-05,
      "loss": 0.8761,
      "step": 1277
    },
    {
      "epoch": 0.817658349328215,
      "grad_norm": 2.2245351599206113,
      "learning_rate": 1.9859817498724023e-05,
      "loss": 0.9871,
      "step": 1278
    },
    {
      "epoch": 0.81829814459373,
      "grad_norm": 1.9964496789065553,
      "learning_rate": 1.9859471541530924e-05,
      "loss": 0.9389,
      "step": 1279
    },
    {
      "epoch": 0.818937939859245,
      "grad_norm": 2.80202596454698,
      "learning_rate": 1.9859125160989653e-05,
      "loss": 0.9995,
      "step": 1280
    },
    {
      "epoch": 0.8195777351247601,
      "grad_norm": 2.792664071476345,
      "learning_rate": 1.9858778357115088e-05,
      "loss": 0.9865,
      "step": 1281
    },
    {
      "epoch": 0.8202175303902751,
      "grad_norm": 2.3277723829293184,
      "learning_rate": 1.985843112992212e-05,
      "loss": 0.9565,
      "step": 1282
    },
    {
      "epoch": 0.8208573256557902,
      "grad_norm": 2.1186312970168997,
      "learning_rate": 1.985808347942565e-05,
      "loss": 0.9247,
      "step": 1283
    },
    {
      "epoch": 0.8214971209213052,
      "grad_norm": 3.398878948502625,
      "learning_rate": 1.9857735405640616e-05,
      "loss": 0.9434,
      "step": 1284
    },
    {
      "epoch": 0.8221369161868202,
      "grad_norm": 2.5135256843437292,
      "learning_rate": 1.985738690858196e-05,
      "loss": 0.9287,
      "step": 1285
    },
    {
      "epoch": 0.8227767114523352,
      "grad_norm": 2.3548054129691276,
      "learning_rate": 1.985703798826464e-05,
      "loss": 0.9917,
      "step": 1286
    },
    {
      "epoch": 0.8234165067178503,
      "grad_norm": 3.0738354015191236,
      "learning_rate": 1.9856688644703647e-05,
      "loss": 0.9207,
      "step": 1287
    },
    {
      "epoch": 0.8240563019833653,
      "grad_norm": 2.5227581349712245,
      "learning_rate": 1.9856338877913975e-05,
      "loss": 0.9479,
      "step": 1288
    },
    {
      "epoch": 0.8246960972488804,
      "grad_norm": 2.3365924306264394,
      "learning_rate": 1.9855988687910647e-05,
      "loss": 1.0022,
      "step": 1289
    },
    {
      "epoch": 0.8253358925143954,
      "grad_norm": 2.3843358198867852,
      "learning_rate": 1.9855638074708693e-05,
      "loss": 0.9885,
      "step": 1290
    },
    {
      "epoch": 0.8259756877799105,
      "grad_norm": 2.783943827489314,
      "learning_rate": 1.985528703832318e-05,
      "loss": 0.9561,
      "step": 1291
    },
    {
      "epoch": 0.8266154830454254,
      "grad_norm": 3.390787059120649,
      "learning_rate": 1.9854935578769164e-05,
      "loss": 0.9115,
      "step": 1292
    },
    {
      "epoch": 0.8272552783109405,
      "grad_norm": 2.0929825476345716,
      "learning_rate": 1.985458369606175e-05,
      "loss": 0.9661,
      "step": 1293
    },
    {
      "epoch": 0.8278950735764555,
      "grad_norm": 2.661313767528074,
      "learning_rate": 1.9854231390216043e-05,
      "loss": 0.9256,
      "step": 1294
    },
    {
      "epoch": 0.8285348688419706,
      "grad_norm": 3.0356864237752217,
      "learning_rate": 1.9853878661247163e-05,
      "loss": 0.9861,
      "step": 1295
    },
    {
      "epoch": 0.8291746641074856,
      "grad_norm": 3.0425731778268,
      "learning_rate": 1.985352550917027e-05,
      "loss": 0.9719,
      "step": 1296
    },
    {
      "epoch": 0.8298144593730007,
      "grad_norm": 2.8161630563102644,
      "learning_rate": 1.9853171934000516e-05,
      "loss": 1.045,
      "step": 1297
    },
    {
      "epoch": 0.8304542546385156,
      "grad_norm": 2.157021403069524,
      "learning_rate": 1.9852817935753083e-05,
      "loss": 0.967,
      "step": 1298
    },
    {
      "epoch": 0.8310940499040307,
      "grad_norm": 2.498839401197811,
      "learning_rate": 1.985246351444318e-05,
      "loss": 0.9435,
      "step": 1299
    },
    {
      "epoch": 0.8317338451695457,
      "grad_norm": 2.75570358084422,
      "learning_rate": 1.9852108670086017e-05,
      "loss": 0.9626,
      "step": 1300
    },
    {
      "epoch": 0.8323736404350608,
      "grad_norm": 2.5537600847171857,
      "learning_rate": 1.9851753402696834e-05,
      "loss": 1.0087,
      "step": 1301
    },
    {
      "epoch": 0.8330134357005758,
      "grad_norm": 2.732522096053286,
      "learning_rate": 1.9851397712290886e-05,
      "loss": 0.9915,
      "step": 1302
    },
    {
      "epoch": 0.8336532309660909,
      "grad_norm": 2.3599368309112636,
      "learning_rate": 1.9851041598883443e-05,
      "loss": 0.8898,
      "step": 1303
    },
    {
      "epoch": 0.8342930262316058,
      "grad_norm": 2.65147829693628,
      "learning_rate": 1.98506850624898e-05,
      "loss": 1.0058,
      "step": 1304
    },
    {
      "epoch": 0.8349328214971209,
      "grad_norm": 2.6609498045180984,
      "learning_rate": 1.9850328103125257e-05,
      "loss": 0.9384,
      "step": 1305
    },
    {
      "epoch": 0.835572616762636,
      "grad_norm": 2.724259868046984,
      "learning_rate": 1.9849970720805152e-05,
      "loss": 0.9395,
      "step": 1306
    },
    {
      "epoch": 0.836212412028151,
      "grad_norm": 2.232941065931355,
      "learning_rate": 1.9849612915544828e-05,
      "loss": 1.0299,
      "step": 1307
    },
    {
      "epoch": 0.836852207293666,
      "grad_norm": 2.409039528149713,
      "learning_rate": 1.9849254687359645e-05,
      "loss": 0.9408,
      "step": 1308
    },
    {
      "epoch": 0.8374920025591811,
      "grad_norm": 3.3853202335879167,
      "learning_rate": 1.9848896036264985e-05,
      "loss": 0.9703,
      "step": 1309
    },
    {
      "epoch": 0.838131797824696,
      "grad_norm": 2.567988725632965,
      "learning_rate": 1.984853696227625e-05,
      "loss": 0.9284,
      "step": 1310
    },
    {
      "epoch": 0.8387715930902111,
      "grad_norm": 2.2085207582348105,
      "learning_rate": 1.9848177465408854e-05,
      "loss": 0.9256,
      "step": 1311
    },
    {
      "epoch": 0.8394113883557262,
      "grad_norm": 2.2594940156572982,
      "learning_rate": 1.9847817545678238e-05,
      "loss": 0.9496,
      "step": 1312
    },
    {
      "epoch": 0.8400511836212412,
      "grad_norm": 2.4246813056761796,
      "learning_rate": 1.9847457203099858e-05,
      "loss": 0.9612,
      "step": 1313
    },
    {
      "epoch": 0.8406909788867563,
      "grad_norm": 3.2922536766418413,
      "learning_rate": 1.984709643768918e-05,
      "loss": 1.0013,
      "step": 1314
    },
    {
      "epoch": 0.8413307741522713,
      "grad_norm": 2.369305363144554,
      "learning_rate": 1.9846735249461696e-05,
      "loss": 0.9424,
      "step": 1315
    },
    {
      "epoch": 0.8419705694177864,
      "grad_norm": 2.3331252806970726,
      "learning_rate": 1.984637363843292e-05,
      "loss": 0.9447,
      "step": 1316
    },
    {
      "epoch": 0.8426103646833013,
      "grad_norm": 2.2909261680473194,
      "learning_rate": 1.9846011604618375e-05,
      "loss": 0.8588,
      "step": 1317
    },
    {
      "epoch": 0.8432501599488164,
      "grad_norm": 2.469627290513508,
      "learning_rate": 1.9845649148033603e-05,
      "loss": 0.9649,
      "step": 1318
    },
    {
      "epoch": 0.8438899552143314,
      "grad_norm": 2.320814744770169,
      "learning_rate": 1.984528626869417e-05,
      "loss": 0.9545,
      "step": 1319
    },
    {
      "epoch": 0.8445297504798465,
      "grad_norm": 2.5411264548472783,
      "learning_rate": 1.9844922966615663e-05,
      "loss": 0.9273,
      "step": 1320
    },
    {
      "epoch": 0.8451695457453615,
      "grad_norm": 2.6299461348860698,
      "learning_rate": 1.9844559241813674e-05,
      "loss": 0.9681,
      "step": 1321
    },
    {
      "epoch": 0.8458093410108766,
      "grad_norm": 2.682163070989569,
      "learning_rate": 1.9844195094303826e-05,
      "loss": 0.9342,
      "step": 1322
    },
    {
      "epoch": 0.8464491362763915,
      "grad_norm": 2.318259415082762,
      "learning_rate": 1.9843830524101748e-05,
      "loss": 0.8945,
      "step": 1323
    },
    {
      "epoch": 0.8470889315419066,
      "grad_norm": 3.3179798041197075,
      "learning_rate": 1.9843465531223098e-05,
      "loss": 0.9511,
      "step": 1324
    },
    {
      "epoch": 0.8477287268074216,
      "grad_norm": 2.2696423931611176,
      "learning_rate": 1.984310011568355e-05,
      "loss": 0.9706,
      "step": 1325
    },
    {
      "epoch": 0.8483685220729367,
      "grad_norm": 2.347163231923836,
      "learning_rate": 1.9842734277498793e-05,
      "loss": 0.9142,
      "step": 1326
    },
    {
      "epoch": 0.8490083173384517,
      "grad_norm": 2.432074005870654,
      "learning_rate": 1.9842368016684534e-05,
      "loss": 0.9483,
      "step": 1327
    },
    {
      "epoch": 0.8496481126039668,
      "grad_norm": 2.720448553074431,
      "learning_rate": 1.98420013332565e-05,
      "loss": 0.9261,
      "step": 1328
    },
    {
      "epoch": 0.8502879078694817,
      "grad_norm": 2.3753307184471857,
      "learning_rate": 1.984163422723044e-05,
      "loss": 0.9853,
      "step": 1329
    },
    {
      "epoch": 0.8509277031349968,
      "grad_norm": 2.504461936962436,
      "learning_rate": 1.984126669862211e-05,
      "loss": 0.9148,
      "step": 1330
    },
    {
      "epoch": 0.8515674984005118,
      "grad_norm": 2.5168059842947317,
      "learning_rate": 1.9840898747447293e-05,
      "loss": 0.9249,
      "step": 1331
    },
    {
      "epoch": 0.8522072936660269,
      "grad_norm": 2.5354390692165545,
      "learning_rate": 1.984053037372179e-05,
      "loss": 0.9599,
      "step": 1332
    },
    {
      "epoch": 0.8528470889315419,
      "grad_norm": 2.127510255036001,
      "learning_rate": 1.9840161577461422e-05,
      "loss": 0.9619,
      "step": 1333
    },
    {
      "epoch": 0.853486884197057,
      "grad_norm": 2.676780903539025,
      "learning_rate": 1.9839792358682018e-05,
      "loss": 0.9029,
      "step": 1334
    },
    {
      "epoch": 0.8541266794625719,
      "grad_norm": 2.5479651850507024,
      "learning_rate": 1.983942271739943e-05,
      "loss": 0.9095,
      "step": 1335
    },
    {
      "epoch": 0.854766474728087,
      "grad_norm": 2.709565884132305,
      "learning_rate": 1.9839052653629535e-05,
      "loss": 0.9472,
      "step": 1336
    },
    {
      "epoch": 0.855406269993602,
      "grad_norm": 2.6539873058394354,
      "learning_rate": 1.9838682167388223e-05,
      "loss": 0.9369,
      "step": 1337
    },
    {
      "epoch": 0.8560460652591171,
      "grad_norm": 2.4005535009483787,
      "learning_rate": 1.98383112586914e-05,
      "loss": 0.935,
      "step": 1338
    },
    {
      "epoch": 0.8566858605246321,
      "grad_norm": 2.327741462597803,
      "learning_rate": 1.9837939927554995e-05,
      "loss": 0.9159,
      "step": 1339
    },
    {
      "epoch": 0.8573256557901472,
      "grad_norm": 2.8597564760973087,
      "learning_rate": 1.9837568173994947e-05,
      "loss": 0.9833,
      "step": 1340
    },
    {
      "epoch": 0.8579654510556622,
      "grad_norm": 2.5228854941434387,
      "learning_rate": 1.9837195998027222e-05,
      "loss": 0.9609,
      "step": 1341
    },
    {
      "epoch": 0.8586052463211772,
      "grad_norm": 2.3056774871865935,
      "learning_rate": 1.98368233996678e-05,
      "loss": 0.9362,
      "step": 1342
    },
    {
      "epoch": 0.8592450415866922,
      "grad_norm": 2.237550524055018,
      "learning_rate": 1.983645037893268e-05,
      "loss": 0.9162,
      "step": 1343
    },
    {
      "epoch": 0.8598848368522073,
      "grad_norm": 2.375952642741499,
      "learning_rate": 1.9836076935837876e-05,
      "loss": 0.9575,
      "step": 1344
    },
    {
      "epoch": 0.8605246321177223,
      "grad_norm": 2.4970126680893103,
      "learning_rate": 1.9835703070399428e-05,
      "loss": 0.9483,
      "step": 1345
    },
    {
      "epoch": 0.8611644273832374,
      "grad_norm": 2.26989904621074,
      "learning_rate": 1.9835328782633384e-05,
      "loss": 0.9647,
      "step": 1346
    },
    {
      "epoch": 0.8618042226487524,
      "grad_norm": 2.178162739102373,
      "learning_rate": 1.983495407255582e-05,
      "loss": 0.9146,
      "step": 1347
    },
    {
      "epoch": 0.8624440179142674,
      "grad_norm": 2.7768989107130757,
      "learning_rate": 1.9834578940182822e-05,
      "loss": 0.9796,
      "step": 1348
    },
    {
      "epoch": 0.8630838131797824,
      "grad_norm": 2.4031227710184138,
      "learning_rate": 1.98342033855305e-05,
      "loss": 0.9456,
      "step": 1349
    },
    {
      "epoch": 0.8637236084452975,
      "grad_norm": 2.4029075234358137,
      "learning_rate": 1.9833827408614976e-05,
      "loss": 1.0153,
      "step": 1350
    },
    {
      "epoch": 0.8643634037108126,
      "grad_norm": 3.2254085754334003,
      "learning_rate": 1.9833451009452398e-05,
      "loss": 0.9708,
      "step": 1351
    },
    {
      "epoch": 0.8650031989763276,
      "grad_norm": 2.532233302001729,
      "learning_rate": 1.9833074188058926e-05,
      "loss": 0.9755,
      "step": 1352
    },
    {
      "epoch": 0.8656429942418427,
      "grad_norm": 2.148763739341617,
      "learning_rate": 1.9832696944450738e-05,
      "loss": 0.9443,
      "step": 1353
    },
    {
      "epoch": 0.8662827895073576,
      "grad_norm": 2.3773360971635578,
      "learning_rate": 1.9832319278644036e-05,
      "loss": 0.9236,
      "step": 1354
    },
    {
      "epoch": 0.8669225847728727,
      "grad_norm": 2.8431294362089954,
      "learning_rate": 1.9831941190655035e-05,
      "loss": 0.8984,
      "step": 1355
    },
    {
      "epoch": 0.8675623800383877,
      "grad_norm": 2.6159884460283824,
      "learning_rate": 1.9831562680499966e-05,
      "loss": 0.9393,
      "step": 1356
    },
    {
      "epoch": 0.8682021753039028,
      "grad_norm": 2.494553652834235,
      "learning_rate": 1.9831183748195086e-05,
      "loss": 0.9593,
      "step": 1357
    },
    {
      "epoch": 0.8688419705694178,
      "grad_norm": 2.446091314167328,
      "learning_rate": 1.9830804393756666e-05,
      "loss": 0.9158,
      "step": 1358
    },
    {
      "epoch": 0.8694817658349329,
      "grad_norm": 2.4550521997750976,
      "learning_rate": 1.9830424617200994e-05,
      "loss": 0.9407,
      "step": 1359
    },
    {
      "epoch": 0.8701215611004478,
      "grad_norm": 2.2440946744272923,
      "learning_rate": 1.9830044418544372e-05,
      "loss": 0.9576,
      "step": 1360
    },
    {
      "epoch": 0.8707613563659629,
      "grad_norm": 2.3554585830887707,
      "learning_rate": 1.9829663797803128e-05,
      "loss": 0.9319,
      "step": 1361
    },
    {
      "epoch": 0.8714011516314779,
      "grad_norm": 2.396402881576814,
      "learning_rate": 1.982928275499361e-05,
      "loss": 0.9181,
      "step": 1362
    },
    {
      "epoch": 0.872040946896993,
      "grad_norm": 2.7035435422292426,
      "learning_rate": 1.9828901290132175e-05,
      "loss": 0.9333,
      "step": 1363
    },
    {
      "epoch": 0.872680742162508,
      "grad_norm": 2.4628796975970757,
      "learning_rate": 1.9828519403235202e-05,
      "loss": 0.94,
      "step": 1364
    },
    {
      "epoch": 0.8733205374280231,
      "grad_norm": 2.327284992289849,
      "learning_rate": 1.9828137094319093e-05,
      "loss": 0.9905,
      "step": 1365
    },
    {
      "epoch": 0.8739603326935381,
      "grad_norm": 2.3664122846443894,
      "learning_rate": 1.9827754363400253e-05,
      "loss": 0.9186,
      "step": 1366
    },
    {
      "epoch": 0.8746001279590531,
      "grad_norm": 2.4487724577809407,
      "learning_rate": 1.9827371210495128e-05,
      "loss": 0.9175,
      "step": 1367
    },
    {
      "epoch": 0.8752399232245681,
      "grad_norm": 2.253446038325371,
      "learning_rate": 1.9826987635620164e-05,
      "loss": 0.9509,
      "step": 1368
    },
    {
      "epoch": 0.8758797184900832,
      "grad_norm": 2.5143348908327647,
      "learning_rate": 1.982660363879183e-05,
      "loss": 0.9151,
      "step": 1369
    },
    {
      "epoch": 0.8765195137555982,
      "grad_norm": 1.8722563311706253,
      "learning_rate": 1.9826219220026615e-05,
      "loss": 0.9782,
      "step": 1370
    },
    {
      "epoch": 0.8771593090211133,
      "grad_norm": 2.311558944630403,
      "learning_rate": 1.9825834379341028e-05,
      "loss": 0.9327,
      "step": 1371
    },
    {
      "epoch": 0.8777991042866283,
      "grad_norm": 2.743064334291086,
      "learning_rate": 1.9825449116751595e-05,
      "loss": 1.0242,
      "step": 1372
    },
    {
      "epoch": 0.8784388995521433,
      "grad_norm": 2.3339594449467804,
      "learning_rate": 1.982506343227485e-05,
      "loss": 1.0011,
      "step": 1373
    },
    {
      "epoch": 0.8790786948176583,
      "grad_norm": 2.262413052327792,
      "learning_rate": 1.9824677325927362e-05,
      "loss": 0.9318,
      "step": 1374
    },
    {
      "epoch": 0.8797184900831734,
      "grad_norm": 2.1557252519746313,
      "learning_rate": 1.98242907977257e-05,
      "loss": 0.9749,
      "step": 1375
    },
    {
      "epoch": 0.8803582853486884,
      "grad_norm": 2.1146677170520256,
      "learning_rate": 1.9823903847686474e-05,
      "loss": 0.9315,
      "step": 1376
    },
    {
      "epoch": 0.8809980806142035,
      "grad_norm": 2.1304945333763183,
      "learning_rate": 1.982351647582629e-05,
      "loss": 0.9019,
      "step": 1377
    },
    {
      "epoch": 0.8816378758797185,
      "grad_norm": 2.4861983317406904,
      "learning_rate": 1.9823128682161785e-05,
      "loss": 0.9411,
      "step": 1378
    },
    {
      "epoch": 0.8822776711452335,
      "grad_norm": 2.3792986819291344,
      "learning_rate": 1.9822740466709606e-05,
      "loss": 0.9701,
      "step": 1379
    },
    {
      "epoch": 0.8829174664107485,
      "grad_norm": 2.0617470598758203,
      "learning_rate": 1.9822351829486424e-05,
      "loss": 0.9104,
      "step": 1380
    },
    {
      "epoch": 0.8835572616762636,
      "grad_norm": 2.1960596146449367,
      "learning_rate": 1.9821962770508928e-05,
      "loss": 0.9699,
      "step": 1381
    },
    {
      "epoch": 0.8841970569417786,
      "grad_norm": 2.137452660720591,
      "learning_rate": 1.9821573289793827e-05,
      "loss": 0.883,
      "step": 1382
    },
    {
      "epoch": 0.8848368522072937,
      "grad_norm": 2.4188287043686585,
      "learning_rate": 1.9821183387357832e-05,
      "loss": 0.9638,
      "step": 1383
    },
    {
      "epoch": 0.8854766474728087,
      "grad_norm": 2.88316765360612,
      "learning_rate": 1.9820793063217697e-05,
      "loss": 0.9489,
      "step": 1384
    },
    {
      "epoch": 0.8861164427383237,
      "grad_norm": 2.241193017048902,
      "learning_rate": 1.982040231739018e-05,
      "loss": 0.9585,
      "step": 1385
    },
    {
      "epoch": 0.8867562380038387,
      "grad_norm": 2.3996977691818255,
      "learning_rate": 1.9820011149892055e-05,
      "loss": 0.9552,
      "step": 1386
    },
    {
      "epoch": 0.8873960332693538,
      "grad_norm": 2.1103694617999675,
      "learning_rate": 1.981961956074012e-05,
      "loss": 0.9696,
      "step": 1387
    },
    {
      "epoch": 0.8880358285348688,
      "grad_norm": 2.442110431812703,
      "learning_rate": 1.9819227549951186e-05,
      "loss": 0.9315,
      "step": 1388
    },
    {
      "epoch": 0.8886756238003839,
      "grad_norm": 2.463476850501564,
      "learning_rate": 1.9818835117542088e-05,
      "loss": 0.9979,
      "step": 1389
    },
    {
      "epoch": 0.889315419065899,
      "grad_norm": 2.6351612780177045,
      "learning_rate": 1.9818442263529678e-05,
      "loss": 0.9743,
      "step": 1390
    },
    {
      "epoch": 0.889955214331414,
      "grad_norm": 2.4197212290212025,
      "learning_rate": 1.9818048987930824e-05,
      "loss": 0.9639,
      "step": 1391
    },
    {
      "epoch": 0.8905950095969289,
      "grad_norm": 2.458248609555108,
      "learning_rate": 1.981765529076241e-05,
      "loss": 0.929,
      "step": 1392
    },
    {
      "epoch": 0.891234804862444,
      "grad_norm": 2.557066997438783,
      "learning_rate": 1.981726117204134e-05,
      "loss": 0.9927,
      "step": 1393
    },
    {
      "epoch": 0.891874600127959,
      "grad_norm": 2.3124987683024343,
      "learning_rate": 1.9816866631784545e-05,
      "loss": 0.9333,
      "step": 1394
    },
    {
      "epoch": 0.8925143953934741,
      "grad_norm": 2.629036927103311,
      "learning_rate": 1.9816471670008956e-05,
      "loss": 0.9874,
      "step": 1395
    },
    {
      "epoch": 0.8931541906589892,
      "grad_norm": 2.5613358674809206,
      "learning_rate": 1.9816076286731535e-05,
      "loss": 0.9487,
      "step": 1396
    },
    {
      "epoch": 0.8937939859245042,
      "grad_norm": 2.687410593559699,
      "learning_rate": 1.9815680481969258e-05,
      "loss": 0.9496,
      "step": 1397
    },
    {
      "epoch": 0.8944337811900192,
      "grad_norm": 2.194584000321049,
      "learning_rate": 1.9815284255739128e-05,
      "loss": 0.9833,
      "step": 1398
    },
    {
      "epoch": 0.8950735764555342,
      "grad_norm": 2.2211272853707866,
      "learning_rate": 1.9814887608058146e-05,
      "loss": 0.9297,
      "step": 1399
    },
    {
      "epoch": 0.8957133717210493,
      "grad_norm": 2.737431694257807,
      "learning_rate": 1.981449053894335e-05,
      "loss": 0.9746,
      "step": 1400
    },
    {
      "epoch": 0.8963531669865643,
      "grad_norm": 2.4939726818279815,
      "learning_rate": 1.9814093048411793e-05,
      "loss": 0.9612,
      "step": 1401
    },
    {
      "epoch": 0.8969929622520794,
      "grad_norm": 2.5848463995137925,
      "learning_rate": 1.9813695136480534e-05,
      "loss": 0.9536,
      "step": 1402
    },
    {
      "epoch": 0.8976327575175944,
      "grad_norm": 2.273968942854673,
      "learning_rate": 1.9813296803166667e-05,
      "loss": 0.9281,
      "step": 1403
    },
    {
      "epoch": 0.8982725527831094,
      "grad_norm": 2.1581543772289327,
      "learning_rate": 1.981289804848729e-05,
      "loss": 0.9423,
      "step": 1404
    },
    {
      "epoch": 0.8989123480486244,
      "grad_norm": 2.470510843481598,
      "learning_rate": 1.9812498872459526e-05,
      "loss": 0.9486,
      "step": 1405
    },
    {
      "epoch": 0.8995521433141395,
      "grad_norm": 2.2549450841656733,
      "learning_rate": 1.9812099275100516e-05,
      "loss": 0.9283,
      "step": 1406
    },
    {
      "epoch": 0.9001919385796545,
      "grad_norm": 2.1441001756540623,
      "learning_rate": 1.9811699256427416e-05,
      "loss": 0.9129,
      "step": 1407
    },
    {
      "epoch": 0.9008317338451696,
      "grad_norm": 2.4089187256884346,
      "learning_rate": 1.9811298816457408e-05,
      "loss": 0.9609,
      "step": 1408
    },
    {
      "epoch": 0.9014715291106846,
      "grad_norm": 2.1231368782321796,
      "learning_rate": 1.9810897955207676e-05,
      "loss": 0.9457,
      "step": 1409
    },
    {
      "epoch": 0.9021113243761996,
      "grad_norm": 2.2766410459788515,
      "learning_rate": 1.9810496672695442e-05,
      "loss": 0.9413,
      "step": 1410
    },
    {
      "epoch": 0.9027511196417146,
      "grad_norm": 2.221750841175451,
      "learning_rate": 1.981009496893793e-05,
      "loss": 0.9101,
      "step": 1411
    },
    {
      "epoch": 0.9033909149072297,
      "grad_norm": 2.2641610302471498,
      "learning_rate": 1.9809692843952396e-05,
      "loss": 0.9029,
      "step": 1412
    },
    {
      "epoch": 0.9040307101727447,
      "grad_norm": 2.3684991540509914,
      "learning_rate": 1.9809290297756096e-05,
      "loss": 0.9443,
      "step": 1413
    },
    {
      "epoch": 0.9046705054382598,
      "grad_norm": 2.198452291020963,
      "learning_rate": 1.9808887330366323e-05,
      "loss": 0.958,
      "step": 1414
    },
    {
      "epoch": 0.9053103007037748,
      "grad_norm": 2.7743864521972266,
      "learning_rate": 1.980848394180037e-05,
      "loss": 0.969,
      "step": 1415
    },
    {
      "epoch": 0.9059500959692899,
      "grad_norm": 2.286137567550065,
      "learning_rate": 1.980808013207557e-05,
      "loss": 0.9529,
      "step": 1416
    },
    {
      "epoch": 0.9065898912348048,
      "grad_norm": 2.169381121901947,
      "learning_rate": 1.9807675901209258e-05,
      "loss": 0.9348,
      "step": 1417
    },
    {
      "epoch": 0.9072296865003199,
      "grad_norm": 2.321999801060504,
      "learning_rate": 1.9807271249218787e-05,
      "loss": 0.882,
      "step": 1418
    },
    {
      "epoch": 0.9078694817658349,
      "grad_norm": 2.4127500104449795,
      "learning_rate": 1.9806866176121537e-05,
      "loss": 0.9627,
      "step": 1419
    },
    {
      "epoch": 0.90850927703135,
      "grad_norm": 2.1600496163022447,
      "learning_rate": 1.9806460681934894e-05,
      "loss": 0.9024,
      "step": 1420
    },
    {
      "epoch": 0.909149072296865,
      "grad_norm": 2.326968782369151,
      "learning_rate": 1.9806054766676274e-05,
      "loss": 0.9583,
      "step": 1421
    },
    {
      "epoch": 0.9097888675623801,
      "grad_norm": 2.4781842847582505,
      "learning_rate": 1.980564843036311e-05,
      "loss": 0.9468,
      "step": 1422
    },
    {
      "epoch": 0.910428662827895,
      "grad_norm": 2.3410260378953804,
      "learning_rate": 1.9805241673012844e-05,
      "loss": 0.9695,
      "step": 1423
    },
    {
      "epoch": 0.9110684580934101,
      "grad_norm": 2.159352030664737,
      "learning_rate": 1.980483449464294e-05,
      "loss": 0.9635,
      "step": 1424
    },
    {
      "epoch": 0.9117082533589251,
      "grad_norm": 2.5695601542362896,
      "learning_rate": 1.9804426895270885e-05,
      "loss": 0.9139,
      "step": 1425
    },
    {
      "epoch": 0.9123480486244402,
      "grad_norm": 2.353194686637156,
      "learning_rate": 1.9804018874914183e-05,
      "loss": 1.0004,
      "step": 1426
    },
    {
      "epoch": 0.9129878438899552,
      "grad_norm": 2.5809646729882316,
      "learning_rate": 1.9803610433590347e-05,
      "loss": 0.9332,
      "step": 1427
    },
    {
      "epoch": 0.9136276391554703,
      "grad_norm": 2.4123833323215873,
      "learning_rate": 1.9803201571316923e-05,
      "loss": 0.9012,
      "step": 1428
    },
    {
      "epoch": 0.9142674344209852,
      "grad_norm": 2.5265947159403073,
      "learning_rate": 1.9802792288111458e-05,
      "loss": 0.9189,
      "step": 1429
    },
    {
      "epoch": 0.9149072296865003,
      "grad_norm": 2.1334321348017316,
      "learning_rate": 1.9802382583991527e-05,
      "loss": 0.965,
      "step": 1430
    },
    {
      "epoch": 0.9155470249520153,
      "grad_norm": 2.6662334676638033,
      "learning_rate": 1.980197245897473e-05,
      "loss": 0.9708,
      "step": 1431
    },
    {
      "epoch": 0.9161868202175304,
      "grad_norm": 2.409143257631634,
      "learning_rate": 1.980156191307867e-05,
      "loss": 0.9271,
      "step": 1432
    },
    {
      "epoch": 0.9168266154830454,
      "grad_norm": 2.9513689769400306,
      "learning_rate": 1.9801150946320976e-05,
      "loss": 0.9121,
      "step": 1433
    },
    {
      "epoch": 0.9174664107485605,
      "grad_norm": 2.844143402961616,
      "learning_rate": 1.98007395587193e-05,
      "loss": 0.9379,
      "step": 1434
    },
    {
      "epoch": 0.9181062060140754,
      "grad_norm": 2.4867851771859244,
      "learning_rate": 1.9800327750291296e-05,
      "loss": 0.919,
      "step": 1435
    },
    {
      "epoch": 0.9187460012795905,
      "grad_norm": 2.360835018060645,
      "learning_rate": 1.9799915521054656e-05,
      "loss": 0.9513,
      "step": 1436
    },
    {
      "epoch": 0.9193857965451055,
      "grad_norm": 2.2538064385749013,
      "learning_rate": 1.979950287102707e-05,
      "loss": 0.9782,
      "step": 1437
    },
    {
      "epoch": 0.9200255918106206,
      "grad_norm": 2.5591441839735354,
      "learning_rate": 1.979908980022627e-05,
      "loss": 0.9161,
      "step": 1438
    },
    {
      "epoch": 0.9206653870761357,
      "grad_norm": 2.508056685717502,
      "learning_rate": 1.9798676308669985e-05,
      "loss": 0.8563,
      "step": 1439
    },
    {
      "epoch": 0.9213051823416507,
      "grad_norm": 2.3891975182633236,
      "learning_rate": 1.9798262396375965e-05,
      "loss": 0.9427,
      "step": 1440
    },
    {
      "epoch": 0.9219449776071658,
      "grad_norm": 2.62710663414396,
      "learning_rate": 1.9797848063361988e-05,
      "loss": 0.9008,
      "step": 1441
    },
    {
      "epoch": 0.9225847728726807,
      "grad_norm": 2.814599909930273,
      "learning_rate": 1.9797433309645848e-05,
      "loss": 0.9981,
      "step": 1442
    },
    {
      "epoch": 0.9232245681381958,
      "grad_norm": 2.622369647692898,
      "learning_rate": 1.9797018135245344e-05,
      "loss": 0.9014,
      "step": 1443
    },
    {
      "epoch": 0.9238643634037108,
      "grad_norm": 2.608763119541987,
      "learning_rate": 1.9796602540178313e-05,
      "loss": 0.9781,
      "step": 1444
    },
    {
      "epoch": 0.9245041586692259,
      "grad_norm": 2.563752370550517,
      "learning_rate": 1.9796186524462598e-05,
      "loss": 0.9019,
      "step": 1445
    },
    {
      "epoch": 0.9251439539347409,
      "grad_norm": 2.407873483225868,
      "learning_rate": 1.9795770088116057e-05,
      "loss": 0.9585,
      "step": 1446
    },
    {
      "epoch": 0.925783749200256,
      "grad_norm": 2.4757645648793516,
      "learning_rate": 1.9795353231156574e-05,
      "loss": 0.9299,
      "step": 1447
    },
    {
      "epoch": 0.9264235444657709,
      "grad_norm": 2.418380020821488,
      "learning_rate": 1.9794935953602047e-05,
      "loss": 0.9657,
      "step": 1448
    },
    {
      "epoch": 0.927063339731286,
      "grad_norm": 2.228570202809564,
      "learning_rate": 1.9794518255470393e-05,
      "loss": 0.9599,
      "step": 1449
    },
    {
      "epoch": 0.927703134996801,
      "grad_norm": 2.194984604438835,
      "learning_rate": 1.9794100136779552e-05,
      "loss": 0.9414,
      "step": 1450
    },
    {
      "epoch": 0.9283429302623161,
      "grad_norm": 2.3909348238653787,
      "learning_rate": 1.9793681597547468e-05,
      "loss": 0.9641,
      "step": 1451
    },
    {
      "epoch": 0.9289827255278311,
      "grad_norm": 2.708653886192972,
      "learning_rate": 1.9793262637792125e-05,
      "loss": 0.9676,
      "step": 1452
    },
    {
      "epoch": 0.9296225207933462,
      "grad_norm": 3.131616539176903,
      "learning_rate": 1.9792843257531496e-05,
      "loss": 0.9306,
      "step": 1453
    },
    {
      "epoch": 0.9302623160588611,
      "grad_norm": 2.221269447636396,
      "learning_rate": 1.9792423456783605e-05,
      "loss": 1.0181,
      "step": 1454
    },
    {
      "epoch": 0.9309021113243762,
      "grad_norm": 2.2791700729301283,
      "learning_rate": 1.9792003235566467e-05,
      "loss": 0.9231,
      "step": 1455
    },
    {
      "epoch": 0.9315419065898912,
      "grad_norm": 2.2939837576274047,
      "learning_rate": 1.979158259389813e-05,
      "loss": 0.9517,
      "step": 1456
    },
    {
      "epoch": 0.9321817018554063,
      "grad_norm": 2.209374711506118,
      "learning_rate": 1.9791161531796654e-05,
      "loss": 0.8775,
      "step": 1457
    },
    {
      "epoch": 0.9328214971209213,
      "grad_norm": 2.2207418595166053,
      "learning_rate": 1.9790740049280122e-05,
      "loss": 0.9044,
      "step": 1458
    },
    {
      "epoch": 0.9334612923864364,
      "grad_norm": 2.425843325784377,
      "learning_rate": 1.9790318146366624e-05,
      "loss": 0.9334,
      "step": 1459
    },
    {
      "epoch": 0.9341010876519513,
      "grad_norm": 2.5086193144540228,
      "learning_rate": 1.978989582307428e-05,
      "loss": 0.9253,
      "step": 1460
    },
    {
      "epoch": 0.9347408829174664,
      "grad_norm": 2.3450085515007824,
      "learning_rate": 1.9789473079421228e-05,
      "loss": 0.8845,
      "step": 1461
    },
    {
      "epoch": 0.9353806781829814,
      "grad_norm": 2.695170392067216,
      "learning_rate": 1.9789049915425617e-05,
      "loss": 0.9637,
      "step": 1462
    },
    {
      "epoch": 0.9360204734484965,
      "grad_norm": 2.2152819986850103,
      "learning_rate": 1.978862633110561e-05,
      "loss": 0.9338,
      "step": 1463
    },
    {
      "epoch": 0.9366602687140115,
      "grad_norm": 2.5031388512259167,
      "learning_rate": 1.9788202326479404e-05,
      "loss": 0.9378,
      "step": 1464
    },
    {
      "epoch": 0.9373000639795266,
      "grad_norm": 2.5600925826142453,
      "learning_rate": 1.9787777901565204e-05,
      "loss": 0.9928,
      "step": 1465
    },
    {
      "epoch": 0.9379398592450416,
      "grad_norm": 2.2741933286876366,
      "learning_rate": 1.9787353056381228e-05,
      "loss": 0.9181,
      "step": 1466
    },
    {
      "epoch": 0.9385796545105566,
      "grad_norm": 2.1782846085788603,
      "learning_rate": 1.9786927790945728e-05,
      "loss": 0.9498,
      "step": 1467
    },
    {
      "epoch": 0.9392194497760716,
      "grad_norm": 2.42085342676774,
      "learning_rate": 1.9786502105276952e-05,
      "loss": 0.9229,
      "step": 1468
    },
    {
      "epoch": 0.9398592450415867,
      "grad_norm": 2.389171582952504,
      "learning_rate": 1.9786075999393187e-05,
      "loss": 0.9403,
      "step": 1469
    },
    {
      "epoch": 0.9404990403071017,
      "grad_norm": 2.5529763946978576,
      "learning_rate": 1.9785649473312727e-05,
      "loss": 0.9185,
      "step": 1470
    },
    {
      "epoch": 0.9411388355726168,
      "grad_norm": 2.0486917648493037,
      "learning_rate": 1.9785222527053887e-05,
      "loss": 0.8912,
      "step": 1471
    },
    {
      "epoch": 0.9417786308381318,
      "grad_norm": 2.574309209665542,
      "learning_rate": 1.9784795160634994e-05,
      "loss": 0.9441,
      "step": 1472
    },
    {
      "epoch": 0.9424184261036468,
      "grad_norm": 2.2194024889249033,
      "learning_rate": 1.9784367374074408e-05,
      "loss": 0.9085,
      "step": 1473
    },
    {
      "epoch": 0.9430582213691618,
      "grad_norm": 2.334552422970095,
      "learning_rate": 1.978393916739049e-05,
      "loss": 0.8939,
      "step": 1474
    },
    {
      "epoch": 0.9436980166346769,
      "grad_norm": 2.2066985722283663,
      "learning_rate": 1.9783510540601626e-05,
      "loss": 0.9142,
      "step": 1475
    },
    {
      "epoch": 0.944337811900192,
      "grad_norm": 2.1039070857505777,
      "learning_rate": 1.9783081493726227e-05,
      "loss": 0.9493,
      "step": 1476
    },
    {
      "epoch": 0.944977607165707,
      "grad_norm": 2.2455445851498124,
      "learning_rate": 1.9782652026782706e-05,
      "loss": 0.8896,
      "step": 1477
    },
    {
      "epoch": 0.945617402431222,
      "grad_norm": 2.2759619324290727,
      "learning_rate": 1.9782222139789512e-05,
      "loss": 0.8915,
      "step": 1478
    },
    {
      "epoch": 0.946257197696737,
      "grad_norm": 2.505110678551245,
      "learning_rate": 1.97817918327651e-05,
      "loss": 0.9538,
      "step": 1479
    },
    {
      "epoch": 0.946896992962252,
      "grad_norm": 2.7548093279753894,
      "learning_rate": 1.978136110572795e-05,
      "loss": 0.9337,
      "step": 1480
    },
    {
      "epoch": 0.9475367882277671,
      "grad_norm": 2.387076978066048,
      "learning_rate": 1.978092995869655e-05,
      "loss": 0.9515,
      "step": 1481
    },
    {
      "epoch": 0.9481765834932822,
      "grad_norm": 2.3069362607976798,
      "learning_rate": 1.978049839168942e-05,
      "loss": 0.9364,
      "step": 1482
    },
    {
      "epoch": 0.9488163787587972,
      "grad_norm": 2.2990809199917437,
      "learning_rate": 1.9780066404725086e-05,
      "loss": 0.9274,
      "step": 1483
    },
    {
      "epoch": 0.9494561740243123,
      "grad_norm": 2.4337240417372596,
      "learning_rate": 1.9779633997822098e-05,
      "loss": 0.8737,
      "step": 1484
    },
    {
      "epoch": 0.9500959692898272,
      "grad_norm": 1.9907577275610997,
      "learning_rate": 1.977920117099902e-05,
      "loss": 0.9755,
      "step": 1485
    },
    {
      "epoch": 0.9507357645553423,
      "grad_norm": 2.3694121981157763,
      "learning_rate": 1.977876792427444e-05,
      "loss": 0.933,
      "step": 1486
    },
    {
      "epoch": 0.9513755598208573,
      "grad_norm": 2.311138786943188,
      "learning_rate": 1.977833425766696e-05,
      "loss": 0.9126,
      "step": 1487
    },
    {
      "epoch": 0.9520153550863724,
      "grad_norm": 2.0797957029295984,
      "learning_rate": 1.9777900171195207e-05,
      "loss": 0.924,
      "step": 1488
    },
    {
      "epoch": 0.9526551503518874,
      "grad_norm": 2.3674660001719605,
      "learning_rate": 1.977746566487781e-05,
      "loss": 0.8906,
      "step": 1489
    },
    {
      "epoch": 0.9532949456174025,
      "grad_norm": 2.436061718921542,
      "learning_rate": 1.977703073873343e-05,
      "loss": 0.9835,
      "step": 1490
    },
    {
      "epoch": 0.9539347408829175,
      "grad_norm": 2.2143678089028946,
      "learning_rate": 1.9776595392780744e-05,
      "loss": 0.9112,
      "step": 1491
    },
    {
      "epoch": 0.9545745361484325,
      "grad_norm": 2.62950004508616,
      "learning_rate": 1.977615962703844e-05,
      "loss": 0.943,
      "step": 1492
    },
    {
      "epoch": 0.9552143314139475,
      "grad_norm": 2.190847470433349,
      "learning_rate": 1.9775723441525235e-05,
      "loss": 0.9063,
      "step": 1493
    },
    {
      "epoch": 0.9558541266794626,
      "grad_norm": 2.246285463110481,
      "learning_rate": 1.9775286836259856e-05,
      "loss": 0.9073,
      "step": 1494
    },
    {
      "epoch": 0.9564939219449776,
      "grad_norm": 2.4057295297330916,
      "learning_rate": 1.9774849811261045e-05,
      "loss": 0.8773,
      "step": 1495
    },
    {
      "epoch": 0.9571337172104927,
      "grad_norm": 2.273025432340165,
      "learning_rate": 1.9774412366547576e-05,
      "loss": 0.8672,
      "step": 1496
    },
    {
      "epoch": 0.9577735124760077,
      "grad_norm": 2.428964526269593,
      "learning_rate": 1.9773974502138224e-05,
      "loss": 0.9342,
      "step": 1497
    },
    {
      "epoch": 0.9584133077415227,
      "grad_norm": 2.5111266209396392,
      "learning_rate": 1.9773536218051795e-05,
      "loss": 0.8924,
      "step": 1498
    },
    {
      "epoch": 0.9590531030070377,
      "grad_norm": 3.064311877462918,
      "learning_rate": 1.9773097514307103e-05,
      "loss": 0.9226,
      "step": 1499
    },
    {
      "epoch": 0.9596928982725528,
      "grad_norm": 3.235064382500353,
      "learning_rate": 1.977265839092299e-05,
      "loss": 0.9608,
      "step": 1500
    },
    {
      "epoch": 0.9603326935380678,
      "grad_norm": 2.6078021732335666,
      "learning_rate": 1.9772218847918313e-05,
      "loss": 0.9067,
      "step": 1501
    },
    {
      "epoch": 0.9609724888035829,
      "grad_norm": 2.998288105995318,
      "learning_rate": 1.9771778885311937e-05,
      "loss": 0.8994,
      "step": 1502
    },
    {
      "epoch": 0.9616122840690979,
      "grad_norm": 3.2409958548318496,
      "learning_rate": 1.9771338503122767e-05,
      "loss": 0.9375,
      "step": 1503
    },
    {
      "epoch": 0.9622520793346129,
      "grad_norm": 2.3492686210472895,
      "learning_rate": 1.9770897701369697e-05,
      "loss": 0.8968,
      "step": 1504
    },
    {
      "epoch": 0.9628918746001279,
      "grad_norm": 2.432567309379934,
      "learning_rate": 1.9770456480071664e-05,
      "loss": 0.9551,
      "step": 1505
    },
    {
      "epoch": 0.963531669865643,
      "grad_norm": 2.443268446836983,
      "learning_rate": 1.9770014839247604e-05,
      "loss": 0.9701,
      "step": 1506
    },
    {
      "epoch": 0.964171465131158,
      "grad_norm": 2.9504634718139777,
      "learning_rate": 1.976957277891649e-05,
      "loss": 0.9686,
      "step": 1507
    },
    {
      "epoch": 0.9648112603966731,
      "grad_norm": 2.7067026139867805,
      "learning_rate": 1.97691302990973e-05,
      "loss": 0.9359,
      "step": 1508
    },
    {
      "epoch": 0.9654510556621881,
      "grad_norm": 2.6275223517342616,
      "learning_rate": 1.9768687399809035e-05,
      "loss": 0.9393,
      "step": 1509
    },
    {
      "epoch": 0.9660908509277031,
      "grad_norm": 2.3417755163603826,
      "learning_rate": 1.9768244081070708e-05,
      "loss": 0.8876,
      "step": 1510
    },
    {
      "epoch": 0.9667306461932181,
      "grad_norm": 2.7231864122221254,
      "learning_rate": 1.9767800342901357e-05,
      "loss": 0.9436,
      "step": 1511
    },
    {
      "epoch": 0.9673704414587332,
      "grad_norm": 2.4154349151083876,
      "learning_rate": 1.9767356185320034e-05,
      "loss": 0.9425,
      "step": 1512
    },
    {
      "epoch": 0.9680102367242482,
      "grad_norm": 2.1742286470217254,
      "learning_rate": 1.976691160834581e-05,
      "loss": 0.8985,
      "step": 1513
    },
    {
      "epoch": 0.9686500319897633,
      "grad_norm": 2.437784977231569,
      "learning_rate": 1.976646661199778e-05,
      "loss": 0.9374,
      "step": 1514
    },
    {
      "epoch": 0.9692898272552783,
      "grad_norm": 2.6321314693927835,
      "learning_rate": 1.976602119629504e-05,
      "loss": 0.8936,
      "step": 1515
    },
    {
      "epoch": 0.9699296225207934,
      "grad_norm": 2.2556387734679353,
      "learning_rate": 1.9765575361256727e-05,
      "loss": 0.8922,
      "step": 1516
    },
    {
      "epoch": 0.9705694177863083,
      "grad_norm": 2.7555058872410574,
      "learning_rate": 1.976512910690198e-05,
      "loss": 0.8965,
      "step": 1517
    },
    {
      "epoch": 0.9712092130518234,
      "grad_norm": 2.5337975761041824,
      "learning_rate": 1.9764682433249958e-05,
      "loss": 0.9124,
      "step": 1518
    },
    {
      "epoch": 0.9718490083173384,
      "grad_norm": 2.429107551464619,
      "learning_rate": 1.9764235340319842e-05,
      "loss": 0.975,
      "step": 1519
    },
    {
      "epoch": 0.9724888035828535,
      "grad_norm": 2.3453795426784048,
      "learning_rate": 1.976378782813083e-05,
      "loss": 0.9171,
      "step": 1520
    },
    {
      "epoch": 0.9731285988483686,
      "grad_norm": 2.752463402395549,
      "learning_rate": 1.9763339896702137e-05,
      "loss": 0.8649,
      "step": 1521
    },
    {
      "epoch": 0.9737683941138836,
      "grad_norm": 2.5156292197462515,
      "learning_rate": 1.9762891546053e-05,
      "loss": 0.9315,
      "step": 1522
    },
    {
      "epoch": 0.9744081893793985,
      "grad_norm": 2.7157565254674823,
      "learning_rate": 1.9762442776202664e-05,
      "loss": 0.8903,
      "step": 1523
    },
    {
      "epoch": 0.9750479846449136,
      "grad_norm": 2.6094259055128504,
      "learning_rate": 1.97619935871704e-05,
      "loss": 0.8865,
      "step": 1524
    },
    {
      "epoch": 0.9756877799104287,
      "grad_norm": 2.282136445943409,
      "learning_rate": 1.9761543978975498e-05,
      "loss": 0.9743,
      "step": 1525
    },
    {
      "epoch": 0.9763275751759437,
      "grad_norm": 2.4738799298173886,
      "learning_rate": 1.9761093951637265e-05,
      "loss": 0.9161,
      "step": 1526
    },
    {
      "epoch": 0.9769673704414588,
      "grad_norm": 2.655105672658034,
      "learning_rate": 1.9760643505175016e-05,
      "loss": 0.8642,
      "step": 1527
    },
    {
      "epoch": 0.9776071657069738,
      "grad_norm": 2.8932744617361297,
      "learning_rate": 1.9760192639608103e-05,
      "loss": 0.9177,
      "step": 1528
    },
    {
      "epoch": 0.9782469609724888,
      "grad_norm": 2.9206883371213452,
      "learning_rate": 1.9759741354955874e-05,
      "loss": 0.9547,
      "step": 1529
    },
    {
      "epoch": 0.9788867562380038,
      "grad_norm": 2.5312580420838255,
      "learning_rate": 1.9759289651237716e-05,
      "loss": 0.9285,
      "step": 1530
    },
    {
      "epoch": 0.9795265515035189,
      "grad_norm": 2.1677647706338035,
      "learning_rate": 1.975883752847302e-05,
      "loss": 0.8955,
      "step": 1531
    },
    {
      "epoch": 0.9801663467690339,
      "grad_norm": 2.449561835645445,
      "learning_rate": 1.9758384986681202e-05,
      "loss": 0.9043,
      "step": 1532
    },
    {
      "epoch": 0.980806142034549,
      "grad_norm": 2.800971437889875,
      "learning_rate": 1.9757932025881688e-05,
      "loss": 0.9459,
      "step": 1533
    },
    {
      "epoch": 0.981445937300064,
      "grad_norm": 2.5261385261279874,
      "learning_rate": 1.9757478646093935e-05,
      "loss": 0.9125,
      "step": 1534
    },
    {
      "epoch": 0.982085732565579,
      "grad_norm": 2.383364419117448,
      "learning_rate": 1.9757024847337404e-05,
      "loss": 0.8905,
      "step": 1535
    },
    {
      "epoch": 0.982725527831094,
      "grad_norm": 2.3851265722097956,
      "learning_rate": 1.975657062963158e-05,
      "loss": 0.918,
      "step": 1536
    },
    {
      "epoch": 0.9833653230966091,
      "grad_norm": 2.4801083224875486,
      "learning_rate": 1.9756115992995975e-05,
      "loss": 0.9035,
      "step": 1537
    },
    {
      "epoch": 0.9840051183621241,
      "grad_norm": 2.6861785816898798,
      "learning_rate": 1.9755660937450098e-05,
      "loss": 0.902,
      "step": 1538
    },
    {
      "epoch": 0.9846449136276392,
      "grad_norm": 2.2416508979424776,
      "learning_rate": 1.9755205463013495e-05,
      "loss": 0.8829,
      "step": 1539
    },
    {
      "epoch": 0.9852847088931542,
      "grad_norm": 2.2393339185922727,
      "learning_rate": 1.9754749569705727e-05,
      "loss": 0.9032,
      "step": 1540
    },
    {
      "epoch": 0.9859245041586693,
      "grad_norm": 2.449562993340201,
      "learning_rate": 1.975429325754636e-05,
      "loss": 0.9228,
      "step": 1541
    },
    {
      "epoch": 0.9865642994241842,
      "grad_norm": 2.412229668056761,
      "learning_rate": 1.9753836526554995e-05,
      "loss": 0.9217,
      "step": 1542
    },
    {
      "epoch": 0.9872040946896993,
      "grad_norm": 2.6211944245082575,
      "learning_rate": 1.975337937675124e-05,
      "loss": 0.9002,
      "step": 1543
    },
    {
      "epoch": 0.9878438899552143,
      "grad_norm": 2.281347215903675,
      "learning_rate": 1.9752921808154727e-05,
      "loss": 0.9575,
      "step": 1544
    },
    {
      "epoch": 0.9884836852207294,
      "grad_norm": 2.671941984104,
      "learning_rate": 1.9752463820785093e-05,
      "loss": 0.9435,
      "step": 1545
    },
    {
      "epoch": 0.9891234804862444,
      "grad_norm": 2.537768629734337,
      "learning_rate": 1.9752005414662015e-05,
      "loss": 0.9034,
      "step": 1546
    },
    {
      "epoch": 0.9897632757517595,
      "grad_norm": 2.442794406461084,
      "learning_rate": 1.9751546589805173e-05,
      "loss": 0.949,
      "step": 1547
    },
    {
      "epoch": 0.9904030710172744,
      "grad_norm": 2.5171491846794076,
      "learning_rate": 1.9751087346234267e-05,
      "loss": 0.9482,
      "step": 1548
    },
    {
      "epoch": 0.9910428662827895,
      "grad_norm": 2.2793303648220817,
      "learning_rate": 1.9750627683969016e-05,
      "loss": 0.9073,
      "step": 1549
    },
    {
      "epoch": 0.9916826615483045,
      "grad_norm": 2.3733947634587484,
      "learning_rate": 1.9750167603029155e-05,
      "loss": 0.9402,
      "step": 1550
    },
    {
      "epoch": 0.9923224568138196,
      "grad_norm": 2.6607515194807734,
      "learning_rate": 1.9749707103434447e-05,
      "loss": 0.8714,
      "step": 1551
    },
    {
      "epoch": 0.9929622520793346,
      "grad_norm": 2.161972660351689,
      "learning_rate": 1.9749246185204655e-05,
      "loss": 0.9278,
      "step": 1552
    },
    {
      "epoch": 0.9936020473448497,
      "grad_norm": 2.419355169425963,
      "learning_rate": 1.9748784848359573e-05,
      "loss": 0.9544,
      "step": 1553
    },
    {
      "epoch": 0.9942418426103646,
      "grad_norm": 2.363245502627612,
      "learning_rate": 1.9748323092919014e-05,
      "loss": 0.9017,
      "step": 1554
    },
    {
      "epoch": 0.9948816378758797,
      "grad_norm": 3.0102392970864758,
      "learning_rate": 1.97478609189028e-05,
      "loss": 0.915,
      "step": 1555
    },
    {
      "epoch": 0.9955214331413947,
      "grad_norm": 2.358058215346776,
      "learning_rate": 1.974739832633078e-05,
      "loss": 0.8516,
      "step": 1556
    },
    {
      "epoch": 0.9961612284069098,
      "grad_norm": 2.2945795951045094,
      "learning_rate": 1.9746935315222818e-05,
      "loss": 0.8919,
      "step": 1557
    },
    {
      "epoch": 0.9968010236724248,
      "grad_norm": 1.9839863982981452,
      "learning_rate": 1.9746471885598786e-05,
      "loss": 0.9149,
      "step": 1558
    },
    {
      "epoch": 0.9974408189379399,
      "grad_norm": 2.4174450978705964,
      "learning_rate": 1.974600803747859e-05,
      "loss": 0.9034,
      "step": 1559
    },
    {
      "epoch": 0.9980806142034548,
      "grad_norm": 2.6941125781114748,
      "learning_rate": 1.974554377088215e-05,
      "loss": 0.9409,
      "step": 1560
    },
    {
      "epoch": 0.9987204094689699,
      "grad_norm": 2.21176895774183,
      "learning_rate": 1.9745079085829395e-05,
      "loss": 0.8694,
      "step": 1561
    },
    {
      "epoch": 0.999360204734485,
      "grad_norm": 2.4844530334480197,
      "learning_rate": 1.974461398234028e-05,
      "loss": 0.9056,
      "step": 1562
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.672293874146878,
      "learning_rate": 1.974414846043477e-05,
      "loss": 0.9043,
      "step": 1563
    },
    {
      "epoch": 1.000639795265515,
      "grad_norm": 2.454602603975063,
      "learning_rate": 1.9743682520132863e-05,
      "loss": 0.8362,
      "step": 1564
    },
    {
      "epoch": 1.00127959053103,
      "grad_norm": 2.459343847842651,
      "learning_rate": 1.974321616145456e-05,
      "loss": 0.8072,
      "step": 1565
    },
    {
      "epoch": 1.0019193857965452,
      "grad_norm": 2.201341298937209,
      "learning_rate": 1.9742749384419886e-05,
      "loss": 0.8506,
      "step": 1566
    },
    {
      "epoch": 1.0025591810620602,
      "grad_norm": 2.610548664070499,
      "learning_rate": 1.9742282189048887e-05,
      "loss": 0.8325,
      "step": 1567
    },
    {
      "epoch": 1.0031989763275753,
      "grad_norm": 2.6065630541129674,
      "learning_rate": 1.9741814575361617e-05,
      "loss": 0.873,
      "step": 1568
    },
    {
      "epoch": 1.0038387715930903,
      "grad_norm": 2.5642068510475897,
      "learning_rate": 1.974134654337816e-05,
      "loss": 0.8515,
      "step": 1569
    },
    {
      "epoch": 1.0044785668586051,
      "grad_norm": 2.223933606382931,
      "learning_rate": 1.9740878093118614e-05,
      "loss": 0.7697,
      "step": 1570
    },
    {
      "epoch": 1.0051183621241202,
      "grad_norm": 2.096798732480728,
      "learning_rate": 1.9740409224603088e-05,
      "loss": 0.849,
      "step": 1571
    },
    {
      "epoch": 1.0057581573896353,
      "grad_norm": 2.365500071598967,
      "learning_rate": 1.973993993785172e-05,
      "loss": 0.8743,
      "step": 1572
    },
    {
      "epoch": 1.0063979526551503,
      "grad_norm": 2.684449584004415,
      "learning_rate": 1.973947023288465e-05,
      "loss": 0.8631,
      "step": 1573
    },
    {
      "epoch": 1.0070377479206654,
      "grad_norm": 2.2245334526339673,
      "learning_rate": 1.9739000109722058e-05,
      "loss": 0.8402,
      "step": 1574
    },
    {
      "epoch": 1.0076775431861804,
      "grad_norm": 2.7939231191989755,
      "learning_rate": 1.9738529568384122e-05,
      "loss": 0.8036,
      "step": 1575
    },
    {
      "epoch": 1.0083173384516955,
      "grad_norm": 2.4388273914260346,
      "learning_rate": 1.9738058608891056e-05,
      "loss": 0.8017,
      "step": 1576
    },
    {
      "epoch": 1.0089571337172105,
      "grad_norm": 2.1440797289765756,
      "learning_rate": 1.973758723126307e-05,
      "loss": 0.8136,
      "step": 1577
    },
    {
      "epoch": 1.0095969289827256,
      "grad_norm": 2.1539943251697586,
      "learning_rate": 1.9737115435520417e-05,
      "loss": 0.8439,
      "step": 1578
    },
    {
      "epoch": 1.0102367242482406,
      "grad_norm": 2.279571031940338,
      "learning_rate": 1.9736643221683343e-05,
      "loss": 0.7786,
      "step": 1579
    },
    {
      "epoch": 1.0108765195137557,
      "grad_norm": 2.7315663828908106,
      "learning_rate": 1.973617058977213e-05,
      "loss": 0.8231,
      "step": 1580
    },
    {
      "epoch": 1.0115163147792707,
      "grad_norm": 2.5972047917871595,
      "learning_rate": 1.973569753980707e-05,
      "loss": 0.8301,
      "step": 1581
    },
    {
      "epoch": 1.0121561100447856,
      "grad_norm": 2.4794150754896864,
      "learning_rate": 1.9735224071808478e-05,
      "loss": 0.8234,
      "step": 1582
    },
    {
      "epoch": 1.0127959053103006,
      "grad_norm": 2.515827226418928,
      "learning_rate": 1.973475018579668e-05,
      "loss": 0.7696,
      "step": 1583
    },
    {
      "epoch": 1.0134357005758157,
      "grad_norm": 2.5437924570807944,
      "learning_rate": 1.973427588179203e-05,
      "loss": 0.8077,
      "step": 1584
    },
    {
      "epoch": 1.0140754958413307,
      "grad_norm": 2.375503640479367,
      "learning_rate": 1.9733801159814883e-05,
      "loss": 0.7824,
      "step": 1585
    },
    {
      "epoch": 1.0147152911068458,
      "grad_norm": 2.2327157379820637,
      "learning_rate": 1.9733326019885634e-05,
      "loss": 0.7392,
      "step": 1586
    },
    {
      "epoch": 1.0153550863723608,
      "grad_norm": 2.4807134555998203,
      "learning_rate": 1.973285046202468e-05,
      "loss": 0.8265,
      "step": 1587
    },
    {
      "epoch": 1.0159948816378759,
      "grad_norm": 2.402638509730168,
      "learning_rate": 1.973237448625244e-05,
      "loss": 0.7964,
      "step": 1588
    },
    {
      "epoch": 1.016634676903391,
      "grad_norm": 2.3715724713648862,
      "learning_rate": 1.9731898092589352e-05,
      "loss": 0.8519,
      "step": 1589
    },
    {
      "epoch": 1.017274472168906,
      "grad_norm": 2.603814969433604,
      "learning_rate": 1.973142128105587e-05,
      "loss": 0.8541,
      "step": 1590
    },
    {
      "epoch": 1.017914267434421,
      "grad_norm": 2.3626469380879405,
      "learning_rate": 1.973094405167247e-05,
      "loss": 0.8204,
      "step": 1591
    },
    {
      "epoch": 1.018554062699936,
      "grad_norm": 2.333308954498033,
      "learning_rate": 1.9730466404459646e-05,
      "loss": 0.8884,
      "step": 1592
    },
    {
      "epoch": 1.0191938579654511,
      "grad_norm": 2.275247772055373,
      "learning_rate": 1.97299883394379e-05,
      "loss": 0.8366,
      "step": 1593
    },
    {
      "epoch": 1.0198336532309662,
      "grad_norm": 2.245536862693335,
      "learning_rate": 1.9729509856627763e-05,
      "loss": 0.8499,
      "step": 1594
    },
    {
      "epoch": 1.020473448496481,
      "grad_norm": 2.475855740633791,
      "learning_rate": 1.972903095604978e-05,
      "loss": 0.8348,
      "step": 1595
    },
    {
      "epoch": 1.021113243761996,
      "grad_norm": 2.029948317185615,
      "learning_rate": 1.9728551637724515e-05,
      "loss": 0.803,
      "step": 1596
    },
    {
      "epoch": 1.0217530390275111,
      "grad_norm": 2.823602153017728,
      "learning_rate": 1.9728071901672548e-05,
      "loss": 0.8395,
      "step": 1597
    },
    {
      "epoch": 1.0223928342930262,
      "grad_norm": 2.552540874662494,
      "learning_rate": 1.972759174791448e-05,
      "loss": 0.7856,
      "step": 1598
    },
    {
      "epoch": 1.0230326295585412,
      "grad_norm": 2.4725667754776466,
      "learning_rate": 1.9727111176470925e-05,
      "loss": 0.806,
      "step": 1599
    },
    {
      "epoch": 1.0236724248240563,
      "grad_norm": 2.1147829395218056,
      "learning_rate": 1.972663018736252e-05,
      "loss": 0.8431,
      "step": 1600
    },
    {
      "epoch": 1.0243122200895713,
      "grad_norm": 2.447021698976848,
      "learning_rate": 1.9726148780609916e-05,
      "loss": 0.7683,
      "step": 1601
    },
    {
      "epoch": 1.0249520153550864,
      "grad_norm": 2.2723763868447584,
      "learning_rate": 1.9725666956233787e-05,
      "loss": 0.8626,
      "step": 1602
    },
    {
      "epoch": 1.0255918106206015,
      "grad_norm": 2.7547539626341244,
      "learning_rate": 1.9725184714254815e-05,
      "loss": 0.7958,
      "step": 1603
    },
    {
      "epoch": 1.0262316058861165,
      "grad_norm": 2.1336840190385886,
      "learning_rate": 1.9724702054693717e-05,
      "loss": 0.8395,
      "step": 1604
    },
    {
      "epoch": 1.0268714011516316,
      "grad_norm": 2.5632873831358114,
      "learning_rate": 1.9724218977571207e-05,
      "loss": 0.8128,
      "step": 1605
    },
    {
      "epoch": 1.0275111964171466,
      "grad_norm": 2.2950675554147906,
      "learning_rate": 1.9723735482908037e-05,
      "loss": 0.7804,
      "step": 1606
    },
    {
      "epoch": 1.0281509916826614,
      "grad_norm": 2.1205654428738163,
      "learning_rate": 1.9723251570724956e-05,
      "loss": 0.8005,
      "step": 1607
    },
    {
      "epoch": 1.0287907869481765,
      "grad_norm": 2.3438693535370345,
      "learning_rate": 1.9722767241042754e-05,
      "loss": 0.8226,
      "step": 1608
    },
    {
      "epoch": 1.0294305822136915,
      "grad_norm": 2.9543955717182264,
      "learning_rate": 1.972228249388222e-05,
      "loss": 0.8361,
      "step": 1609
    },
    {
      "epoch": 1.0300703774792066,
      "grad_norm": 2.482513657598865,
      "learning_rate": 1.9721797329264173e-05,
      "loss": 0.8146,
      "step": 1610
    },
    {
      "epoch": 1.0307101727447217,
      "grad_norm": 2.2780440623886964,
      "learning_rate": 1.9721311747209437e-05,
      "loss": 0.7999,
      "step": 1611
    },
    {
      "epoch": 1.0313499680102367,
      "grad_norm": 2.0697576965634483,
      "learning_rate": 1.972082574773887e-05,
      "loss": 0.7662,
      "step": 1612
    },
    {
      "epoch": 1.0319897632757518,
      "grad_norm": 2.4890735722910238,
      "learning_rate": 1.9720339330873338e-05,
      "loss": 0.7622,
      "step": 1613
    },
    {
      "epoch": 1.0326295585412668,
      "grad_norm": 2.783872599948963,
      "learning_rate": 1.971985249663373e-05,
      "loss": 0.816,
      "step": 1614
    },
    {
      "epoch": 1.0332693538067819,
      "grad_norm": 2.2815272284238315,
      "learning_rate": 1.971936524504094e-05,
      "loss": 0.8108,
      "step": 1615
    },
    {
      "epoch": 1.033909149072297,
      "grad_norm": 2.588629082219566,
      "learning_rate": 1.97188775761159e-05,
      "loss": 0.817,
      "step": 1616
    },
    {
      "epoch": 1.034548944337812,
      "grad_norm": 2.245490707874264,
      "learning_rate": 1.9718389489879544e-05,
      "loss": 0.8393,
      "step": 1617
    },
    {
      "epoch": 1.035188739603327,
      "grad_norm": 2.637297875415707,
      "learning_rate": 1.971790098635283e-05,
      "loss": 0.8294,
      "step": 1618
    },
    {
      "epoch": 1.035828534868842,
      "grad_norm": 2.6950268543569953,
      "learning_rate": 1.9717412065556736e-05,
      "loss": 0.8259,
      "step": 1619
    },
    {
      "epoch": 1.036468330134357,
      "grad_norm": 2.5990367167245494,
      "learning_rate": 1.9716922727512253e-05,
      "loss": 0.8167,
      "step": 1620
    },
    {
      "epoch": 1.037108125399872,
      "grad_norm": 2.600810979876676,
      "learning_rate": 1.9716432972240393e-05,
      "loss": 0.784,
      "step": 1621
    },
    {
      "epoch": 1.037747920665387,
      "grad_norm": 2.6579000987963974,
      "learning_rate": 1.9715942799762183e-05,
      "loss": 0.8041,
      "step": 1622
    },
    {
      "epoch": 1.038387715930902,
      "grad_norm": 2.421259757896088,
      "learning_rate": 1.971545221009868e-05,
      "loss": 0.7925,
      "step": 1623
    },
    {
      "epoch": 1.0390275111964171,
      "grad_norm": 2.656524754697389,
      "learning_rate": 1.9714961203270936e-05,
      "loss": 0.8107,
      "step": 1624
    },
    {
      "epoch": 1.0396673064619322,
      "grad_norm": 2.9976687294426196,
      "learning_rate": 1.971446977930004e-05,
      "loss": 0.8387,
      "step": 1625
    },
    {
      "epoch": 1.0403071017274472,
      "grad_norm": 2.498416983495648,
      "learning_rate": 1.971397793820709e-05,
      "loss": 0.7894,
      "step": 1626
    },
    {
      "epoch": 1.0409468969929623,
      "grad_norm": 2.082495787406185,
      "learning_rate": 1.971348568001321e-05,
      "loss": 0.8437,
      "step": 1627
    },
    {
      "epoch": 1.0415866922584773,
      "grad_norm": 2.606269632867904,
      "learning_rate": 1.9712993004739532e-05,
      "loss": 0.8566,
      "step": 1628
    },
    {
      "epoch": 1.0422264875239924,
      "grad_norm": 2.607554625595292,
      "learning_rate": 1.9712499912407213e-05,
      "loss": 0.8097,
      "step": 1629
    },
    {
      "epoch": 1.0428662827895074,
      "grad_norm": 2.2997755145356025,
      "learning_rate": 1.9712006403037426e-05,
      "loss": 0.8807,
      "step": 1630
    },
    {
      "epoch": 1.0435060780550225,
      "grad_norm": 2.309825223227484,
      "learning_rate": 1.971151247665136e-05,
      "loss": 0.793,
      "step": 1631
    },
    {
      "epoch": 1.0441458733205373,
      "grad_norm": 2.1472045361888306,
      "learning_rate": 1.9711018133270227e-05,
      "loss": 0.7834,
      "step": 1632
    },
    {
      "epoch": 1.0447856685860524,
      "grad_norm": 2.4581968617880867,
      "learning_rate": 1.9710523372915244e-05,
      "loss": 0.8418,
      "step": 1633
    },
    {
      "epoch": 1.0454254638515674,
      "grad_norm": 2.227835349318698,
      "learning_rate": 1.9710028195607666e-05,
      "loss": 0.8175,
      "step": 1634
    },
    {
      "epoch": 1.0460652591170825,
      "grad_norm": 2.5217638182257875,
      "learning_rate": 1.970953260136875e-05,
      "loss": 0.8263,
      "step": 1635
    },
    {
      "epoch": 1.0467050543825975,
      "grad_norm": 2.4866277952377778,
      "learning_rate": 1.9709036590219773e-05,
      "loss": 0.865,
      "step": 1636
    },
    {
      "epoch": 1.0473448496481126,
      "grad_norm": 2.8137306811927805,
      "learning_rate": 1.9708540162182034e-05,
      "loss": 0.8438,
      "step": 1637
    },
    {
      "epoch": 1.0479846449136276,
      "grad_norm": 2.463604200706522,
      "learning_rate": 1.9708043317276854e-05,
      "loss": 0.8561,
      "step": 1638
    },
    {
      "epoch": 1.0486244401791427,
      "grad_norm": 2.154866141291066,
      "learning_rate": 1.9707546055525564e-05,
      "loss": 0.7985,
      "step": 1639
    },
    {
      "epoch": 1.0492642354446577,
      "grad_norm": 2.3569719166269314,
      "learning_rate": 1.9707048376949513e-05,
      "loss": 0.8055,
      "step": 1640
    },
    {
      "epoch": 1.0499040307101728,
      "grad_norm": 2.6620869612598086,
      "learning_rate": 1.9706550281570074e-05,
      "loss": 0.8465,
      "step": 1641
    },
    {
      "epoch": 1.0505438259756879,
      "grad_norm": 2.6262709465917107,
      "learning_rate": 1.970605176940863e-05,
      "loss": 0.8161,
      "step": 1642
    },
    {
      "epoch": 1.051183621241203,
      "grad_norm": 2.415265207144851,
      "learning_rate": 1.970555284048659e-05,
      "loss": 0.7995,
      "step": 1643
    },
    {
      "epoch": 1.051823416506718,
      "grad_norm": 2.1680835256533393,
      "learning_rate": 1.9705053494825376e-05,
      "loss": 0.7884,
      "step": 1644
    },
    {
      "epoch": 1.0524632117722328,
      "grad_norm": 2.1385197710052353,
      "learning_rate": 1.9704553732446428e-05,
      "loss": 0.8232,
      "step": 1645
    },
    {
      "epoch": 1.0531030070377478,
      "grad_norm": 2.5256904462706675,
      "learning_rate": 1.9704053553371207e-05,
      "loss": 0.872,
      "step": 1646
    },
    {
      "epoch": 1.053742802303263,
      "grad_norm": 2.421857611686675,
      "learning_rate": 1.9703552957621187e-05,
      "loss": 0.8271,
      "step": 1647
    },
    {
      "epoch": 1.054382597568778,
      "grad_norm": 2.072469842872589,
      "learning_rate": 1.9703051945217866e-05,
      "loss": 0.8588,
      "step": 1648
    },
    {
      "epoch": 1.055022392834293,
      "grad_norm": 2.348691595374194,
      "learning_rate": 1.9702550516182756e-05,
      "loss": 0.8349,
      "step": 1649
    },
    {
      "epoch": 1.055662188099808,
      "grad_norm": 2.3349354236119395,
      "learning_rate": 1.970204867053738e-05,
      "loss": 0.8498,
      "step": 1650
    },
    {
      "epoch": 1.056301983365323,
      "grad_norm": 2.0656981157352683,
      "learning_rate": 1.97015464083033e-05,
      "loss": 0.8106,
      "step": 1651
    },
    {
      "epoch": 1.0569417786308382,
      "grad_norm": 2.2395709889568334,
      "learning_rate": 1.970104372950207e-05,
      "loss": 0.8134,
      "step": 1652
    },
    {
      "epoch": 1.0575815738963532,
      "grad_norm": 2.3690686963906438,
      "learning_rate": 1.9700540634155283e-05,
      "loss": 0.8191,
      "step": 1653
    },
    {
      "epoch": 1.0582213691618683,
      "grad_norm": 2.58681221966543,
      "learning_rate": 1.9700037122284533e-05,
      "loss": 0.8647,
      "step": 1654
    },
    {
      "epoch": 1.0588611644273833,
      "grad_norm": 2.3386845557458558,
      "learning_rate": 1.969953319391145e-05,
      "loss": 0.7898,
      "step": 1655
    },
    {
      "epoch": 1.0595009596928984,
      "grad_norm": 2.442204593826714,
      "learning_rate": 1.969902884905766e-05,
      "loss": 0.8074,
      "step": 1656
    },
    {
      "epoch": 1.0601407549584132,
      "grad_norm": 2.0960855205578635,
      "learning_rate": 1.9698524087744828e-05,
      "loss": 0.7941,
      "step": 1657
    },
    {
      "epoch": 1.0607805502239283,
      "grad_norm": 2.209440057416185,
      "learning_rate": 1.9698018909994623e-05,
      "loss": 0.805,
      "step": 1658
    },
    {
      "epoch": 1.0614203454894433,
      "grad_norm": 2.4672922808247613,
      "learning_rate": 1.9697513315828734e-05,
      "loss": 0.8324,
      "step": 1659
    },
    {
      "epoch": 1.0620601407549584,
      "grad_norm": 2.265210100228491,
      "learning_rate": 1.969700730526888e-05,
      "loss": 0.7974,
      "step": 1660
    },
    {
      "epoch": 1.0626999360204734,
      "grad_norm": 2.125812601641627,
      "learning_rate": 1.9696500878336778e-05,
      "loss": 0.8553,
      "step": 1661
    },
    {
      "epoch": 1.0633397312859885,
      "grad_norm": 2.3325941458391406,
      "learning_rate": 1.9695994035054177e-05,
      "loss": 0.8514,
      "step": 1662
    },
    {
      "epoch": 1.0639795265515035,
      "grad_norm": 2.281223972140171,
      "learning_rate": 1.969548677544284e-05,
      "loss": 0.8178,
      "step": 1663
    },
    {
      "epoch": 1.0646193218170186,
      "grad_norm": 2.8147995274235242,
      "learning_rate": 1.9694979099524548e-05,
      "loss": 0.8528,
      "step": 1664
    },
    {
      "epoch": 1.0652591170825336,
      "grad_norm": 2.1881472087531564,
      "learning_rate": 1.9694471007321102e-05,
      "loss": 0.8266,
      "step": 1665
    },
    {
      "epoch": 1.0658989123480487,
      "grad_norm": 2.1841311833590114,
      "learning_rate": 1.9693962498854312e-05,
      "loss": 0.8271,
      "step": 1666
    },
    {
      "epoch": 1.0665387076135637,
      "grad_norm": 2.324071813648203,
      "learning_rate": 1.9693453574146018e-05,
      "loss": 0.7942,
      "step": 1667
    },
    {
      "epoch": 1.0671785028790788,
      "grad_norm": 2.2015463292912894,
      "learning_rate": 1.9692944233218074e-05,
      "loss": 0.8237,
      "step": 1668
    },
    {
      "epoch": 1.0678182981445938,
      "grad_norm": 2.485002721489981,
      "learning_rate": 1.9692434476092345e-05,
      "loss": 0.8717,
      "step": 1669
    },
    {
      "epoch": 1.0684580934101087,
      "grad_norm": 2.258594220609517,
      "learning_rate": 1.969192430279072e-05,
      "loss": 0.8455,
      "step": 1670
    },
    {
      "epoch": 1.0690978886756237,
      "grad_norm": 2.187329876537102,
      "learning_rate": 1.9691413713335107e-05,
      "loss": 0.8505,
      "step": 1671
    },
    {
      "epoch": 1.0697376839411388,
      "grad_norm": 2.2933779641288274,
      "learning_rate": 1.969090270774743e-05,
      "loss": 0.799,
      "step": 1672
    },
    {
      "epoch": 1.0703774792066538,
      "grad_norm": 2.2841585990082405,
      "learning_rate": 1.9690391286049625e-05,
      "loss": 0.8035,
      "step": 1673
    },
    {
      "epoch": 1.0710172744721689,
      "grad_norm": 2.2164706549072664,
      "learning_rate": 1.968987944826366e-05,
      "loss": 0.8504,
      "step": 1674
    },
    {
      "epoch": 1.071657069737684,
      "grad_norm": 2.1894214920753967,
      "learning_rate": 1.9689367194411505e-05,
      "loss": 0.8026,
      "step": 1675
    },
    {
      "epoch": 1.072296865003199,
      "grad_norm": 2.343857629511061,
      "learning_rate": 1.9688854524515162e-05,
      "loss": 0.8202,
      "step": 1676
    },
    {
      "epoch": 1.072936660268714,
      "grad_norm": 2.355865908786055,
      "learning_rate": 1.968834143859664e-05,
      "loss": 0.8307,
      "step": 1677
    },
    {
      "epoch": 1.073576455534229,
      "grad_norm": 2.268795139458161,
      "learning_rate": 1.968782793667797e-05,
      "loss": 0.7953,
      "step": 1678
    },
    {
      "epoch": 1.0742162507997441,
      "grad_norm": 2.455883129234087,
      "learning_rate": 1.9687314018781203e-05,
      "loss": 0.8511,
      "step": 1679
    },
    {
      "epoch": 1.0748560460652592,
      "grad_norm": 2.3853467054060755,
      "learning_rate": 1.9686799684928405e-05,
      "loss": 0.8481,
      "step": 1680
    },
    {
      "epoch": 1.0754958413307742,
      "grad_norm": 2.0197379920255947,
      "learning_rate": 1.968628493514166e-05,
      "loss": 0.8194,
      "step": 1681
    },
    {
      "epoch": 1.0761356365962893,
      "grad_norm": 2.36371295474788,
      "learning_rate": 1.9685769769443068e-05,
      "loss": 0.8059,
      "step": 1682
    },
    {
      "epoch": 1.0767754318618041,
      "grad_norm": 2.203550033685931,
      "learning_rate": 1.9685254187854754e-05,
      "loss": 0.8478,
      "step": 1683
    },
    {
      "epoch": 1.0774152271273192,
      "grad_norm": 2.4855708384212254,
      "learning_rate": 1.9684738190398853e-05,
      "loss": 0.8123,
      "step": 1684
    },
    {
      "epoch": 1.0780550223928342,
      "grad_norm": 2.203015402486109,
      "learning_rate": 1.9684221777097524e-05,
      "loss": 0.8354,
      "step": 1685
    },
    {
      "epoch": 1.0786948176583493,
      "grad_norm": 2.355084103158273,
      "learning_rate": 1.9683704947972935e-05,
      "loss": 0.8296,
      "step": 1686
    },
    {
      "epoch": 1.0793346129238643,
      "grad_norm": 2.1028131457183252,
      "learning_rate": 1.9683187703047283e-05,
      "loss": 0.8233,
      "step": 1687
    },
    {
      "epoch": 1.0799744081893794,
      "grad_norm": 2.1043297434667703,
      "learning_rate": 1.9682670042342776e-05,
      "loss": 0.8306,
      "step": 1688
    },
    {
      "epoch": 1.0806142034548945,
      "grad_norm": 2.3771256493029798,
      "learning_rate": 1.9682151965881643e-05,
      "loss": 0.8098,
      "step": 1689
    },
    {
      "epoch": 1.0812539987204095,
      "grad_norm": 2.3947521156592297,
      "learning_rate": 1.9681633473686125e-05,
      "loss": 0.8143,
      "step": 1690
    },
    {
      "epoch": 1.0818937939859246,
      "grad_norm": 2.0139182673064298,
      "learning_rate": 1.9681114565778488e-05,
      "loss": 0.8116,
      "step": 1691
    },
    {
      "epoch": 1.0825335892514396,
      "grad_norm": 2.3192416190022387,
      "learning_rate": 1.9680595242181016e-05,
      "loss": 0.8092,
      "step": 1692
    },
    {
      "epoch": 1.0831733845169547,
      "grad_norm": 2.099906460580957,
      "learning_rate": 1.9680075502916004e-05,
      "loss": 0.8524,
      "step": 1693
    },
    {
      "epoch": 1.0838131797824695,
      "grad_norm": 1.9335440574392917,
      "learning_rate": 1.9679555348005767e-05,
      "loss": 0.7898,
      "step": 1694
    },
    {
      "epoch": 1.0844529750479845,
      "grad_norm": 2.142476689530243,
      "learning_rate": 1.9679034777472645e-05,
      "loss": 0.8386,
      "step": 1695
    },
    {
      "epoch": 1.0850927703134996,
      "grad_norm": 2.0096199963498402,
      "learning_rate": 1.9678513791338986e-05,
      "loss": 0.8455,
      "step": 1696
    },
    {
      "epoch": 1.0857325655790147,
      "grad_norm": 2.62689895166026,
      "learning_rate": 1.967799238962716e-05,
      "loss": 0.8186,
      "step": 1697
    },
    {
      "epoch": 1.0863723608445297,
      "grad_norm": 2.578935327937364,
      "learning_rate": 1.967747057235956e-05,
      "loss": 0.7979,
      "step": 1698
    },
    {
      "epoch": 1.0870121561100448,
      "grad_norm": 2.3016261322198965,
      "learning_rate": 1.9676948339558587e-05,
      "loss": 0.838,
      "step": 1699
    },
    {
      "epoch": 1.0876519513755598,
      "grad_norm": 2.310749423736526,
      "learning_rate": 1.9676425691246666e-05,
      "loss": 0.8699,
      "step": 1700
    },
    {
      "epoch": 1.0882917466410749,
      "grad_norm": 2.4749487907214998,
      "learning_rate": 1.967590262744624e-05,
      "loss": 0.8294,
      "step": 1701
    },
    {
      "epoch": 1.08893154190659,
      "grad_norm": 2.2885051100627765,
      "learning_rate": 1.967537914817976e-05,
      "loss": 0.8089,
      "step": 1702
    },
    {
      "epoch": 1.089571337172105,
      "grad_norm": 2.4227662954154394,
      "learning_rate": 1.967485525346972e-05,
      "loss": 0.7602,
      "step": 1703
    },
    {
      "epoch": 1.09021113243762,
      "grad_norm": 2.3395101444751663,
      "learning_rate": 1.9674330943338602e-05,
      "loss": 0.8548,
      "step": 1704
    },
    {
      "epoch": 1.090850927703135,
      "grad_norm": 2.182497138597938,
      "learning_rate": 1.9673806217808923e-05,
      "loss": 0.8224,
      "step": 1705
    },
    {
      "epoch": 1.0914907229686501,
      "grad_norm": 2.3223588690797143,
      "learning_rate": 1.9673281076903214e-05,
      "loss": 0.8149,
      "step": 1706
    },
    {
      "epoch": 1.092130518234165,
      "grad_norm": 2.0257459618639606,
      "learning_rate": 1.9672755520644023e-05,
      "loss": 0.8242,
      "step": 1707
    },
    {
      "epoch": 1.09277031349968,
      "grad_norm": 2.1998030688219297,
      "learning_rate": 1.9672229549053913e-05,
      "loss": 0.8189,
      "step": 1708
    },
    {
      "epoch": 1.093410108765195,
      "grad_norm": 2.300440985636055,
      "learning_rate": 1.9671703162155474e-05,
      "loss": 0.7718,
      "step": 1709
    },
    {
      "epoch": 1.0940499040307101,
      "grad_norm": 2.5508292582584544,
      "learning_rate": 1.9671176359971306e-05,
      "loss": 0.8531,
      "step": 1710
    },
    {
      "epoch": 1.0946896992962252,
      "grad_norm": 2.2605065260416812,
      "learning_rate": 1.967064914252403e-05,
      "loss": 0.7935,
      "step": 1711
    },
    {
      "epoch": 1.0953294945617402,
      "grad_norm": 2.118709436854627,
      "learning_rate": 1.9670121509836284e-05,
      "loss": 0.7956,
      "step": 1712
    },
    {
      "epoch": 1.0959692898272553,
      "grad_norm": 2.3388431838765418,
      "learning_rate": 1.966959346193072e-05,
      "loss": 0.8433,
      "step": 1713
    },
    {
      "epoch": 1.0966090850927703,
      "grad_norm": 2.229264314993048,
      "learning_rate": 1.9669064998830012e-05,
      "loss": 0.8536,
      "step": 1714
    },
    {
      "epoch": 1.0972488803582854,
      "grad_norm": 2.4748960687465087,
      "learning_rate": 1.9668536120556853e-05,
      "loss": 0.8077,
      "step": 1715
    },
    {
      "epoch": 1.0978886756238004,
      "grad_norm": 2.5954392320819144,
      "learning_rate": 1.9668006827133953e-05,
      "loss": 0.7957,
      "step": 1716
    },
    {
      "epoch": 1.0985284708893155,
      "grad_norm": 2.477714028148731,
      "learning_rate": 1.9667477118584038e-05,
      "loss": 0.8246,
      "step": 1717
    },
    {
      "epoch": 1.0991682661548305,
      "grad_norm": 2.3020741944469894,
      "learning_rate": 1.9666946994929856e-05,
      "loss": 0.8245,
      "step": 1718
    },
    {
      "epoch": 1.0998080614203456,
      "grad_norm": 2.1394221575330534,
      "learning_rate": 1.966641645619416e-05,
      "loss": 0.8021,
      "step": 1719
    },
    {
      "epoch": 1.1004478566858604,
      "grad_norm": 2.526662138789805,
      "learning_rate": 1.966588550239974e-05,
      "loss": 0.8096,
      "step": 1720
    },
    {
      "epoch": 1.1010876519513755,
      "grad_norm": 2.6322162640336306,
      "learning_rate": 1.966535413356939e-05,
      "loss": 0.8522,
      "step": 1721
    },
    {
      "epoch": 1.1017274472168905,
      "grad_norm": 2.335823647958065,
      "learning_rate": 1.9664822349725927e-05,
      "loss": 0.8856,
      "step": 1722
    },
    {
      "epoch": 1.1023672424824056,
      "grad_norm": 2.4148234677085303,
      "learning_rate": 1.9664290150892187e-05,
      "loss": 0.7426,
      "step": 1723
    },
    {
      "epoch": 1.1030070377479206,
      "grad_norm": 2.451492143310843,
      "learning_rate": 1.9663757537091014e-05,
      "loss": 0.8209,
      "step": 1724
    },
    {
      "epoch": 1.1036468330134357,
      "grad_norm": 2.252290496161982,
      "learning_rate": 1.966322450834529e-05,
      "loss": 0.7898,
      "step": 1725
    },
    {
      "epoch": 1.1042866282789507,
      "grad_norm": 2.33867657351482,
      "learning_rate": 1.966269106467789e-05,
      "loss": 0.8759,
      "step": 1726
    },
    {
      "epoch": 1.1049264235444658,
      "grad_norm": 2.1708502085238495,
      "learning_rate": 1.9662157206111724e-05,
      "loss": 0.8018,
      "step": 1727
    },
    {
      "epoch": 1.1055662188099808,
      "grad_norm": 2.157568351115573,
      "learning_rate": 1.9661622932669718e-05,
      "loss": 0.8076,
      "step": 1728
    },
    {
      "epoch": 1.106206014075496,
      "grad_norm": 2.525720569835501,
      "learning_rate": 1.966108824437481e-05,
      "loss": 0.7752,
      "step": 1729
    },
    {
      "epoch": 1.106845809341011,
      "grad_norm": 2.49380431489408,
      "learning_rate": 1.9660553141249955e-05,
      "loss": 0.8221,
      "step": 1730
    },
    {
      "epoch": 1.107485604606526,
      "grad_norm": 2.3807230449224663,
      "learning_rate": 1.9660017623318134e-05,
      "loss": 0.8135,
      "step": 1731
    },
    {
      "epoch": 1.108125399872041,
      "grad_norm": 2.139604900946053,
      "learning_rate": 1.9659481690602342e-05,
      "loss": 0.8186,
      "step": 1732
    },
    {
      "epoch": 1.108765195137556,
      "grad_norm": 2.2074420281101195,
      "learning_rate": 1.965894534312559e-05,
      "loss": 0.8071,
      "step": 1733
    },
    {
      "epoch": 1.109404990403071,
      "grad_norm": 2.294041927915757,
      "learning_rate": 1.9658408580910903e-05,
      "loss": 0.8235,
      "step": 1734
    },
    {
      "epoch": 1.110044785668586,
      "grad_norm": 2.3856732563340035,
      "learning_rate": 1.9657871403981333e-05,
      "loss": 0.7411,
      "step": 1735
    },
    {
      "epoch": 1.110684580934101,
      "grad_norm": 2.1271025018502163,
      "learning_rate": 1.9657333812359948e-05,
      "loss": 0.8312,
      "step": 1736
    },
    {
      "epoch": 1.111324376199616,
      "grad_norm": 2.3615351663781903,
      "learning_rate": 1.965679580606983e-05,
      "loss": 0.8378,
      "step": 1737
    },
    {
      "epoch": 1.1119641714651312,
      "grad_norm": 2.129266112862535,
      "learning_rate": 1.9656257385134073e-05,
      "loss": 0.822,
      "step": 1738
    },
    {
      "epoch": 1.1126039667306462,
      "grad_norm": 2.1641680630465223,
      "learning_rate": 1.96557185495758e-05,
      "loss": 0.7906,
      "step": 1739
    },
    {
      "epoch": 1.1132437619961613,
      "grad_norm": 2.4987402563480643,
      "learning_rate": 1.9655179299418155e-05,
      "loss": 0.8123,
      "step": 1740
    },
    {
      "epoch": 1.1138835572616763,
      "grad_norm": 2.3662634250370322,
      "learning_rate": 1.965463963468428e-05,
      "loss": 0.7966,
      "step": 1741
    },
    {
      "epoch": 1.1145233525271914,
      "grad_norm": 2.038202633149386,
      "learning_rate": 1.9654099555397358e-05,
      "loss": 0.848,
      "step": 1742
    },
    {
      "epoch": 1.1151631477927064,
      "grad_norm": 2.13404116134029,
      "learning_rate": 1.9653559061580575e-05,
      "loss": 0.8158,
      "step": 1743
    },
    {
      "epoch": 1.1158029430582213,
      "grad_norm": 2.2575639326123893,
      "learning_rate": 1.9653018153257133e-05,
      "loss": 0.7813,
      "step": 1744
    },
    {
      "epoch": 1.1164427383237363,
      "grad_norm": 2.0976165437874474,
      "learning_rate": 1.9652476830450268e-05,
      "loss": 0.7852,
      "step": 1745
    },
    {
      "epoch": 1.1170825335892514,
      "grad_norm": 2.0503972718844974,
      "learning_rate": 1.9651935093183216e-05,
      "loss": 0.8416,
      "step": 1746
    },
    {
      "epoch": 1.1177223288547664,
      "grad_norm": 2.401837072348849,
      "learning_rate": 1.9651392941479236e-05,
      "loss": 0.7714,
      "step": 1747
    },
    {
      "epoch": 1.1183621241202815,
      "grad_norm": 2.0324020985829394,
      "learning_rate": 1.965085037536162e-05,
      "loss": 0.7967,
      "step": 1748
    },
    {
      "epoch": 1.1190019193857965,
      "grad_norm": 2.3726210916149593,
      "learning_rate": 1.965030739485365e-05,
      "loss": 0.796,
      "step": 1749
    },
    {
      "epoch": 1.1196417146513116,
      "grad_norm": 2.2991285500114813,
      "learning_rate": 1.964976399997865e-05,
      "loss": 0.8097,
      "step": 1750
    },
    {
      "epoch": 1.1202815099168266,
      "grad_norm": 2.3232782359354434,
      "learning_rate": 1.964922019075995e-05,
      "loss": 0.8608,
      "step": 1751
    },
    {
      "epoch": 1.1209213051823417,
      "grad_norm": 2.3844909729371686,
      "learning_rate": 1.9648675967220897e-05,
      "loss": 0.8516,
      "step": 1752
    },
    {
      "epoch": 1.1215611004478567,
      "grad_norm": 2.340328368461316,
      "learning_rate": 1.9648131329384866e-05,
      "loss": 0.7752,
      "step": 1753
    },
    {
      "epoch": 1.1222008957133718,
      "grad_norm": 2.7458655365602276,
      "learning_rate": 1.9647586277275236e-05,
      "loss": 0.7865,
      "step": 1754
    },
    {
      "epoch": 1.1228406909788868,
      "grad_norm": 2.290530877471875,
      "learning_rate": 1.964704081091541e-05,
      "loss": 0.8379,
      "step": 1755
    },
    {
      "epoch": 1.1234804862444019,
      "grad_norm": 1.9605837044659387,
      "learning_rate": 1.9646494930328815e-05,
      "loss": 0.7937,
      "step": 1756
    },
    {
      "epoch": 1.1241202815099167,
      "grad_norm": 2.6283793662258015,
      "learning_rate": 1.964594863553889e-05,
      "loss": 0.743,
      "step": 1757
    },
    {
      "epoch": 1.1247600767754318,
      "grad_norm": 2.574447533348572,
      "learning_rate": 1.9645401926569086e-05,
      "loss": 0.8588,
      "step": 1758
    },
    {
      "epoch": 1.1253998720409468,
      "grad_norm": 2.4484145042337686,
      "learning_rate": 1.964485480344288e-05,
      "loss": 0.8776,
      "step": 1759
    },
    {
      "epoch": 1.1260396673064619,
      "grad_norm": 2.202378838760484,
      "learning_rate": 1.964430726618377e-05,
      "loss": 0.7701,
      "step": 1760
    },
    {
      "epoch": 1.126679462571977,
      "grad_norm": 2.1740303297206856,
      "learning_rate": 1.964375931481526e-05,
      "loss": 0.7943,
      "step": 1761
    },
    {
      "epoch": 1.127319257837492,
      "grad_norm": 2.3864430154235468,
      "learning_rate": 1.964321094936088e-05,
      "loss": 0.7987,
      "step": 1762
    },
    {
      "epoch": 1.127959053103007,
      "grad_norm": 2.1635909370016737,
      "learning_rate": 1.9642662169844174e-05,
      "loss": 0.7335,
      "step": 1763
    },
    {
      "epoch": 1.128598848368522,
      "grad_norm": 2.222883918864306,
      "learning_rate": 1.9642112976288707e-05,
      "loss": 0.8248,
      "step": 1764
    },
    {
      "epoch": 1.1292386436340371,
      "grad_norm": 2.5267429086813653,
      "learning_rate": 1.964156336871806e-05,
      "loss": 0.8284,
      "step": 1765
    },
    {
      "epoch": 1.1298784388995522,
      "grad_norm": 2.385188626293987,
      "learning_rate": 1.9641013347155836e-05,
      "loss": 0.8929,
      "step": 1766
    },
    {
      "epoch": 1.1305182341650672,
      "grad_norm": 2.207196739064781,
      "learning_rate": 1.964046291162565e-05,
      "loss": 0.8096,
      "step": 1767
    },
    {
      "epoch": 1.1311580294305823,
      "grad_norm": 2.4947181926458617,
      "learning_rate": 1.9639912062151135e-05,
      "loss": 0.7939,
      "step": 1768
    },
    {
      "epoch": 1.1317978246960974,
      "grad_norm": 2.3065324521020756,
      "learning_rate": 1.9639360798755943e-05,
      "loss": 0.8112,
      "step": 1769
    },
    {
      "epoch": 1.1324376199616122,
      "grad_norm": 2.2129749690567926,
      "learning_rate": 1.9638809121463746e-05,
      "loss": 0.852,
      "step": 1770
    },
    {
      "epoch": 1.1330774152271272,
      "grad_norm": 2.405907874651535,
      "learning_rate": 1.963825703029823e-05,
      "loss": 0.8033,
      "step": 1771
    },
    {
      "epoch": 1.1337172104926423,
      "grad_norm": 2.5474591962229387,
      "learning_rate": 1.9637704525283102e-05,
      "loss": 0.8576,
      "step": 1772
    },
    {
      "epoch": 1.1343570057581573,
      "grad_norm": 2.696285377135883,
      "learning_rate": 1.963715160644209e-05,
      "loss": 0.7352,
      "step": 1773
    },
    {
      "epoch": 1.1349968010236724,
      "grad_norm": 2.7061221119306036,
      "learning_rate": 1.9636598273798927e-05,
      "loss": 0.818,
      "step": 1774
    },
    {
      "epoch": 1.1356365962891874,
      "grad_norm": 2.2168492777681483,
      "learning_rate": 1.9636044527377377e-05,
      "loss": 0.7781,
      "step": 1775
    },
    {
      "epoch": 1.1362763915547025,
      "grad_norm": 2.4091321582745064,
      "learning_rate": 1.963549036720122e-05,
      "loss": 0.7924,
      "step": 1776
    },
    {
      "epoch": 1.1369161868202176,
      "grad_norm": 2.6368334196085406,
      "learning_rate": 1.963493579329424e-05,
      "loss": 0.7735,
      "step": 1777
    },
    {
      "epoch": 1.1375559820857326,
      "grad_norm": 2.366331741424243,
      "learning_rate": 1.9634380805680258e-05,
      "loss": 0.8116,
      "step": 1778
    },
    {
      "epoch": 1.1381957773512477,
      "grad_norm": 2.621404739084732,
      "learning_rate": 1.9633825404383104e-05,
      "loss": 0.8068,
      "step": 1779
    },
    {
      "epoch": 1.1388355726167627,
      "grad_norm": 2.7383737637706522,
      "learning_rate": 1.9633269589426625e-05,
      "loss": 0.8204,
      "step": 1780
    },
    {
      "epoch": 1.1394753678822778,
      "grad_norm": 2.080802813091276,
      "learning_rate": 1.9632713360834685e-05,
      "loss": 0.7838,
      "step": 1781
    },
    {
      "epoch": 1.1401151631477928,
      "grad_norm": 2.3981851437315687,
      "learning_rate": 1.963215671863117e-05,
      "loss": 0.8471,
      "step": 1782
    },
    {
      "epoch": 1.1407549584133077,
      "grad_norm": 2.6175757550960657,
      "learning_rate": 1.963159966283998e-05,
      "loss": 0.8177,
      "step": 1783
    },
    {
      "epoch": 1.1413947536788227,
      "grad_norm": 2.476200963558999,
      "learning_rate": 1.963104219348503e-05,
      "loss": 0.8257,
      "step": 1784
    },
    {
      "epoch": 1.1420345489443378,
      "grad_norm": 2.8519182362357203,
      "learning_rate": 1.9630484310590265e-05,
      "loss": 0.8462,
      "step": 1785
    },
    {
      "epoch": 1.1426743442098528,
      "grad_norm": 2.6370955310401487,
      "learning_rate": 1.9629926014179634e-05,
      "loss": 0.7955,
      "step": 1786
    },
    {
      "epoch": 1.1433141394753679,
      "grad_norm": 2.621824508538184,
      "learning_rate": 1.962936730427711e-05,
      "loss": 0.7684,
      "step": 1787
    },
    {
      "epoch": 1.143953934740883,
      "grad_norm": 2.3234029571292463,
      "learning_rate": 1.962880818090668e-05,
      "loss": 0.8049,
      "step": 1788
    },
    {
      "epoch": 1.144593730006398,
      "grad_norm": 2.573153901478001,
      "learning_rate": 1.9628248644092358e-05,
      "loss": 0.8557,
      "step": 1789
    },
    {
      "epoch": 1.145233525271913,
      "grad_norm": 2.450078081505425,
      "learning_rate": 1.9627688693858168e-05,
      "loss": 0.782,
      "step": 1790
    },
    {
      "epoch": 1.145873320537428,
      "grad_norm": 2.344747341107107,
      "learning_rate": 1.962712833022815e-05,
      "loss": 0.8252,
      "step": 1791
    },
    {
      "epoch": 1.1465131158029431,
      "grad_norm": 2.264669891573534,
      "learning_rate": 1.9626567553226365e-05,
      "loss": 0.8076,
      "step": 1792
    },
    {
      "epoch": 1.1471529110684582,
      "grad_norm": 2.351316370221336,
      "learning_rate": 1.9626006362876898e-05,
      "loss": 0.8432,
      "step": 1793
    },
    {
      "epoch": 1.147792706333973,
      "grad_norm": 2.4348221842762303,
      "learning_rate": 1.9625444759203837e-05,
      "loss": 0.8366,
      "step": 1794
    },
    {
      "epoch": 1.1484325015994883,
      "grad_norm": 2.7124731352387297,
      "learning_rate": 1.9624882742231302e-05,
      "loss": 0.8224,
      "step": 1795
    },
    {
      "epoch": 1.1490722968650031,
      "grad_norm": 2.1648832146188544,
      "learning_rate": 1.9624320311983423e-05,
      "loss": 0.7841,
      "step": 1796
    },
    {
      "epoch": 1.1497120921305182,
      "grad_norm": 2.240961916310946,
      "learning_rate": 1.962375746848435e-05,
      "loss": 0.7893,
      "step": 1797
    },
    {
      "epoch": 1.1503518873960332,
      "grad_norm": 2.294343573669674,
      "learning_rate": 1.9623194211758253e-05,
      "loss": 0.8372,
      "step": 1798
    },
    {
      "epoch": 1.1509916826615483,
      "grad_norm": 2.544282151929317,
      "learning_rate": 1.9622630541829314e-05,
      "loss": 0.8647,
      "step": 1799
    },
    {
      "epoch": 1.1516314779270633,
      "grad_norm": 2.487624306589467,
      "learning_rate": 1.9622066458721735e-05,
      "loss": 0.8271,
      "step": 1800
    },
    {
      "epoch": 1.1522712731925784,
      "grad_norm": 2.0918757066733966,
      "learning_rate": 1.962150196245974e-05,
      "loss": 0.779,
      "step": 1801
    },
    {
      "epoch": 1.1529110684580934,
      "grad_norm": 2.615692874774432,
      "learning_rate": 1.9620937053067565e-05,
      "loss": 0.8395,
      "step": 1802
    },
    {
      "epoch": 1.1535508637236085,
      "grad_norm": 2.215192162124564,
      "learning_rate": 1.9620371730569468e-05,
      "loss": 0.8105,
      "step": 1803
    },
    {
      "epoch": 1.1541906589891235,
      "grad_norm": 2.431205114775517,
      "learning_rate": 1.9619805994989723e-05,
      "loss": 0.8134,
      "step": 1804
    },
    {
      "epoch": 1.1548304542546386,
      "grad_norm": 2.696505320274212,
      "learning_rate": 1.961923984635262e-05,
      "loss": 0.8262,
      "step": 1805
    },
    {
      "epoch": 1.1554702495201536,
      "grad_norm": 2.6615503980902724,
      "learning_rate": 1.9618673284682466e-05,
      "loss": 0.8601,
      "step": 1806
    },
    {
      "epoch": 1.1561100447856685,
      "grad_norm": 2.390359486104537,
      "learning_rate": 1.9618106310003597e-05,
      "loss": 0.8445,
      "step": 1807
    },
    {
      "epoch": 1.1567498400511835,
      "grad_norm": 2.2905210601358883,
      "learning_rate": 1.961753892234035e-05,
      "loss": 0.8388,
      "step": 1808
    },
    {
      "epoch": 1.1573896353166986,
      "grad_norm": 2.1476834792830157,
      "learning_rate": 1.961697112171709e-05,
      "loss": 0.7623,
      "step": 1809
    },
    {
      "epoch": 1.1580294305822136,
      "grad_norm": 2.4831177792054424,
      "learning_rate": 1.9616402908158195e-05,
      "loss": 0.7926,
      "step": 1810
    },
    {
      "epoch": 1.1586692258477287,
      "grad_norm": 2.4407696299194743,
      "learning_rate": 1.961583428168807e-05,
      "loss": 0.8071,
      "step": 1811
    },
    {
      "epoch": 1.1593090211132437,
      "grad_norm": 2.099875894264623,
      "learning_rate": 1.9615265242331125e-05,
      "loss": 0.8367,
      "step": 1812
    },
    {
      "epoch": 1.1599488163787588,
      "grad_norm": 2.0643987833411392,
      "learning_rate": 1.9614695790111795e-05,
      "loss": 0.8087,
      "step": 1813
    },
    {
      "epoch": 1.1605886116442738,
      "grad_norm": 2.044545087083899,
      "learning_rate": 1.9614125925054525e-05,
      "loss": 0.8356,
      "step": 1814
    },
    {
      "epoch": 1.161228406909789,
      "grad_norm": 2.16380872064768,
      "learning_rate": 1.96135556471838e-05,
      "loss": 0.8219,
      "step": 1815
    },
    {
      "epoch": 1.161868202175304,
      "grad_norm": 2.003740603445559,
      "learning_rate": 1.9612984956524088e-05,
      "loss": 0.7702,
      "step": 1816
    },
    {
      "epoch": 1.162507997440819,
      "grad_norm": 2.2325822791155154,
      "learning_rate": 1.9612413853099906e-05,
      "loss": 0.7901,
      "step": 1817
    },
    {
      "epoch": 1.163147792706334,
      "grad_norm": 2.3969378087100894,
      "learning_rate": 1.961184233693577e-05,
      "loss": 0.8785,
      "step": 1818
    },
    {
      "epoch": 1.1637875879718491,
      "grad_norm": 2.443369492132263,
      "learning_rate": 1.9611270408056226e-05,
      "loss": 0.8536,
      "step": 1819
    },
    {
      "epoch": 1.164427383237364,
      "grad_norm": 2.224308330685984,
      "learning_rate": 1.9610698066485825e-05,
      "loss": 0.756,
      "step": 1820
    },
    {
      "epoch": 1.165067178502879,
      "grad_norm": 2.2306569849748796,
      "learning_rate": 1.9610125312249147e-05,
      "loss": 0.8175,
      "step": 1821
    },
    {
      "epoch": 1.165706973768394,
      "grad_norm": 2.476803279681452,
      "learning_rate": 1.960955214537078e-05,
      "loss": 0.771,
      "step": 1822
    },
    {
      "epoch": 1.166346769033909,
      "grad_norm": 2.3805847353400718,
      "learning_rate": 1.960897856587534e-05,
      "loss": 0.7865,
      "step": 1823
    },
    {
      "epoch": 1.1669865642994242,
      "grad_norm": 2.047253715344869,
      "learning_rate": 1.9608404573787455e-05,
      "loss": 0.7825,
      "step": 1824
    },
    {
      "epoch": 1.1676263595649392,
      "grad_norm": 2.2969917431559277,
      "learning_rate": 1.9607830169131765e-05,
      "loss": 0.737,
      "step": 1825
    },
    {
      "epoch": 1.1682661548304543,
      "grad_norm": 2.700196416656631,
      "learning_rate": 1.9607255351932945e-05,
      "loss": 0.789,
      "step": 1826
    },
    {
      "epoch": 1.1689059500959693,
      "grad_norm": 2.4039475312338543,
      "learning_rate": 1.9606680122215665e-05,
      "loss": 0.7933,
      "step": 1827
    },
    {
      "epoch": 1.1695457453614844,
      "grad_norm": 3.0612711776169794,
      "learning_rate": 1.960610448000463e-05,
      "loss": 0.7718,
      "step": 1828
    },
    {
      "epoch": 1.1701855406269994,
      "grad_norm": 2.510698966897696,
      "learning_rate": 1.9605528425324557e-05,
      "loss": 0.8144,
      "step": 1829
    },
    {
      "epoch": 1.1708253358925145,
      "grad_norm": 2.140501238151346,
      "learning_rate": 1.9604951958200186e-05,
      "loss": 0.8646,
      "step": 1830
    },
    {
      "epoch": 1.1714651311580295,
      "grad_norm": 2.815176812708645,
      "learning_rate": 1.9604375078656255e-05,
      "loss": 0.8406,
      "step": 1831
    },
    {
      "epoch": 1.1721049264235446,
      "grad_norm": 2.9195792030087304,
      "learning_rate": 1.960379778671755e-05,
      "loss": 0.8431,
      "step": 1832
    },
    {
      "epoch": 1.1727447216890594,
      "grad_norm": 2.4645802657109157,
      "learning_rate": 1.9603220082408845e-05,
      "loss": 0.8188,
      "step": 1833
    },
    {
      "epoch": 1.1733845169545745,
      "grad_norm": 2.286018752267513,
      "learning_rate": 1.960264196575496e-05,
      "loss": 0.8297,
      "step": 1834
    },
    {
      "epoch": 1.1740243122200895,
      "grad_norm": 2.1889108172721223,
      "learning_rate": 1.9602063436780703e-05,
      "loss": 0.8127,
      "step": 1835
    },
    {
      "epoch": 1.1746641074856046,
      "grad_norm": 2.4696201929098414,
      "learning_rate": 1.9601484495510928e-05,
      "loss": 0.8063,
      "step": 1836
    },
    {
      "epoch": 1.1753039027511196,
      "grad_norm": 2.577395608762471,
      "learning_rate": 1.960090514197049e-05,
      "loss": 0.8049,
      "step": 1837
    },
    {
      "epoch": 1.1759436980166347,
      "grad_norm": 2.271994937162859,
      "learning_rate": 1.960032537618426e-05,
      "loss": 0.7596,
      "step": 1838
    },
    {
      "epoch": 1.1765834932821497,
      "grad_norm": 2.34619211920351,
      "learning_rate": 1.9599745198177135e-05,
      "loss": 0.796,
      "step": 1839
    },
    {
      "epoch": 1.1772232885476648,
      "grad_norm": 2.635054630882261,
      "learning_rate": 1.9599164607974034e-05,
      "loss": 0.8013,
      "step": 1840
    },
    {
      "epoch": 1.1778630838131798,
      "grad_norm": 2.488379140317775,
      "learning_rate": 1.9598583605599876e-05,
      "loss": 0.8178,
      "step": 1841
    },
    {
      "epoch": 1.1785028790786949,
      "grad_norm": 2.3898118754131934,
      "learning_rate": 1.959800219107961e-05,
      "loss": 0.8508,
      "step": 1842
    },
    {
      "epoch": 1.17914267434421,
      "grad_norm": 2.0954513876114427,
      "learning_rate": 1.9597420364438207e-05,
      "loss": 0.7993,
      "step": 1843
    },
    {
      "epoch": 1.1797824696097248,
      "grad_norm": 2.3925018608948663,
      "learning_rate": 1.9596838125700644e-05,
      "loss": 0.7713,
      "step": 1844
    },
    {
      "epoch": 1.18042226487524,
      "grad_norm": 2.3592889574217306,
      "learning_rate": 1.9596255474891926e-05,
      "loss": 0.8144,
      "step": 1845
    },
    {
      "epoch": 1.1810620601407549,
      "grad_norm": 2.239400308030216,
      "learning_rate": 1.9595672412037066e-05,
      "loss": 0.8474,
      "step": 1846
    },
    {
      "epoch": 1.18170185540627,
      "grad_norm": 2.603975541059594,
      "learning_rate": 1.95950889371611e-05,
      "loss": 0.7802,
      "step": 1847
    },
    {
      "epoch": 1.182341650671785,
      "grad_norm": 2.5392730543965323,
      "learning_rate": 1.959450505028909e-05,
      "loss": 0.8575,
      "step": 1848
    },
    {
      "epoch": 1.1829814459373,
      "grad_norm": 2.0671026997612247,
      "learning_rate": 1.9593920751446096e-05,
      "loss": 0.7607,
      "step": 1849
    },
    {
      "epoch": 1.183621241202815,
      "grad_norm": 2.680059499314906,
      "learning_rate": 1.959333604065721e-05,
      "loss": 0.8168,
      "step": 1850
    },
    {
      "epoch": 1.1842610364683301,
      "grad_norm": 2.42508909697098,
      "learning_rate": 1.9592750917947545e-05,
      "loss": 0.7632,
      "step": 1851
    },
    {
      "epoch": 1.1849008317338452,
      "grad_norm": 2.7580169032692057,
      "learning_rate": 1.9592165383342215e-05,
      "loss": 0.8276,
      "step": 1852
    },
    {
      "epoch": 1.1855406269993602,
      "grad_norm": 2.4302412481371234,
      "learning_rate": 1.9591579436866367e-05,
      "loss": 0.8571,
      "step": 1853
    },
    {
      "epoch": 1.1861804222648753,
      "grad_norm": 2.0644292675601386,
      "learning_rate": 1.9590993078545162e-05,
      "loss": 0.8106,
      "step": 1854
    },
    {
      "epoch": 1.1868202175303904,
      "grad_norm": 2.1338294766849364,
      "learning_rate": 1.9590406308403772e-05,
      "loss": 0.7658,
      "step": 1855
    },
    {
      "epoch": 1.1874600127959054,
      "grad_norm": 2.4311401040832004,
      "learning_rate": 1.95898191264674e-05,
      "loss": 0.8491,
      "step": 1856
    },
    {
      "epoch": 1.1880998080614202,
      "grad_norm": 2.322597037843722,
      "learning_rate": 1.958923153276125e-05,
      "loss": 0.8158,
      "step": 1857
    },
    {
      "epoch": 1.1887396033269353,
      "grad_norm": 2.248963529661031,
      "learning_rate": 1.9588643527310558e-05,
      "loss": 0.7756,
      "step": 1858
    },
    {
      "epoch": 1.1893793985924503,
      "grad_norm": 2.47780869428096,
      "learning_rate": 1.9588055110140568e-05,
      "loss": 0.7978,
      "step": 1859
    },
    {
      "epoch": 1.1900191938579654,
      "grad_norm": 2.193556405410384,
      "learning_rate": 1.9587466281276547e-05,
      "loss": 0.8449,
      "step": 1860
    },
    {
      "epoch": 1.1906589891234804,
      "grad_norm": 2.4698619959095613,
      "learning_rate": 1.958687704074378e-05,
      "loss": 0.8242,
      "step": 1861
    },
    {
      "epoch": 1.1912987843889955,
      "grad_norm": 2.352137466199482,
      "learning_rate": 1.958628738856757e-05,
      "loss": 0.7669,
      "step": 1862
    },
    {
      "epoch": 1.1919385796545106,
      "grad_norm": 2.4004826222128193,
      "learning_rate": 1.9585697324773228e-05,
      "loss": 0.851,
      "step": 1863
    },
    {
      "epoch": 1.1925783749200256,
      "grad_norm": 2.141826685567372,
      "learning_rate": 1.9585106849386096e-05,
      "loss": 0.837,
      "step": 1864
    },
    {
      "epoch": 1.1932181701855407,
      "grad_norm": 2.446686794899449,
      "learning_rate": 1.9584515962431523e-05,
      "loss": 0.8355,
      "step": 1865
    },
    {
      "epoch": 1.1938579654510557,
      "grad_norm": 2.257853187646201,
      "learning_rate": 1.9583924663934886e-05,
      "loss": 0.7977,
      "step": 1866
    },
    {
      "epoch": 1.1944977607165708,
      "grad_norm": 2.56495186797894,
      "learning_rate": 1.9583332953921573e-05,
      "loss": 0.8026,
      "step": 1867
    },
    {
      "epoch": 1.1951375559820858,
      "grad_norm": 2.466504255264755,
      "learning_rate": 1.9582740832416995e-05,
      "loss": 0.8537,
      "step": 1868
    },
    {
      "epoch": 1.1957773512476009,
      "grad_norm": 2.3138608958785634,
      "learning_rate": 1.9582148299446565e-05,
      "loss": 0.7746,
      "step": 1869
    },
    {
      "epoch": 1.1964171465131157,
      "grad_norm": 2.266792360126902,
      "learning_rate": 1.9581555355035733e-05,
      "loss": 0.811,
      "step": 1870
    },
    {
      "epoch": 1.1970569417786308,
      "grad_norm": 2.1846560865033826,
      "learning_rate": 1.9580961999209963e-05,
      "loss": 0.7892,
      "step": 1871
    },
    {
      "epoch": 1.1976967370441458,
      "grad_norm": 2.139139257130106,
      "learning_rate": 1.9580368231994724e-05,
      "loss": 0.7454,
      "step": 1872
    },
    {
      "epoch": 1.1983365323096609,
      "grad_norm": 2.2093599488273723,
      "learning_rate": 1.9579774053415518e-05,
      "loss": 0.795,
      "step": 1873
    },
    {
      "epoch": 1.198976327575176,
      "grad_norm": 2.433109918173029,
      "learning_rate": 1.957917946349785e-05,
      "loss": 0.7809,
      "step": 1874
    },
    {
      "epoch": 1.199616122840691,
      "grad_norm": 2.255242499677246,
      "learning_rate": 1.9578584462267264e-05,
      "loss": 0.8275,
      "step": 1875
    },
    {
      "epoch": 1.200255918106206,
      "grad_norm": 2.4296901829423043,
      "learning_rate": 1.9577989049749292e-05,
      "loss": 0.8122,
      "step": 1876
    },
    {
      "epoch": 1.200895713371721,
      "grad_norm": 2.488531642222606,
      "learning_rate": 1.9577393225969512e-05,
      "loss": 0.8205,
      "step": 1877
    },
    {
      "epoch": 1.2015355086372361,
      "grad_norm": 2.6304491441482685,
      "learning_rate": 1.9576796990953502e-05,
      "loss": 0.7721,
      "step": 1878
    },
    {
      "epoch": 1.2021753039027512,
      "grad_norm": 2.4081163760233792,
      "learning_rate": 1.957620034472687e-05,
      "loss": 0.8196,
      "step": 1879
    },
    {
      "epoch": 1.2028150991682662,
      "grad_norm": 2.181022911308375,
      "learning_rate": 1.9575603287315227e-05,
      "loss": 0.8034,
      "step": 1880
    },
    {
      "epoch": 1.2034548944337813,
      "grad_norm": 2.7015898783819603,
      "learning_rate": 1.957500581874421e-05,
      "loss": 0.801,
      "step": 1881
    },
    {
      "epoch": 1.2040946896992963,
      "grad_norm": 2.3750581543268927,
      "learning_rate": 1.9574407939039478e-05,
      "loss": 0.8363,
      "step": 1882
    },
    {
      "epoch": 1.2047344849648112,
      "grad_norm": 2.478083584739971,
      "learning_rate": 1.95738096482267e-05,
      "loss": 0.8137,
      "step": 1883
    },
    {
      "epoch": 1.2053742802303262,
      "grad_norm": 2.375306936840165,
      "learning_rate": 1.957321094633157e-05,
      "loss": 0.8368,
      "step": 1884
    },
    {
      "epoch": 1.2060140754958413,
      "grad_norm": 2.1046468586101623,
      "learning_rate": 1.9572611833379785e-05,
      "loss": 0.7519,
      "step": 1885
    },
    {
      "epoch": 1.2066538707613563,
      "grad_norm": 2.20301069013004,
      "learning_rate": 1.9572012309397078e-05,
      "loss": 0.7972,
      "step": 1886
    },
    {
      "epoch": 1.2072936660268714,
      "grad_norm": 2.3722748260628226,
      "learning_rate": 1.9571412374409193e-05,
      "loss": 0.8024,
      "step": 1887
    },
    {
      "epoch": 1.2079334612923864,
      "grad_norm": 2.44994509295316,
      "learning_rate": 1.9570812028441884e-05,
      "loss": 0.7791,
      "step": 1888
    },
    {
      "epoch": 1.2085732565579015,
      "grad_norm": 2.2361834341502833,
      "learning_rate": 1.9570211271520932e-05,
      "loss": 0.8225,
      "step": 1889
    },
    {
      "epoch": 1.2092130518234165,
      "grad_norm": 2.5548057842291656,
      "learning_rate": 1.9569610103672134e-05,
      "loss": 0.9099,
      "step": 1890
    },
    {
      "epoch": 1.2098528470889316,
      "grad_norm": 2.084415301818252,
      "learning_rate": 1.95690085249213e-05,
      "loss": 0.8337,
      "step": 1891
    },
    {
      "epoch": 1.2104926423544466,
      "grad_norm": 2.746954009208237,
      "learning_rate": 1.956840653529426e-05,
      "loss": 0.8143,
      "step": 1892
    },
    {
      "epoch": 1.2111324376199617,
      "grad_norm": 2.15615789758763,
      "learning_rate": 1.9567804134816864e-05,
      "loss": 0.7743,
      "step": 1893
    },
    {
      "epoch": 1.2117722328854765,
      "grad_norm": 2.376040026448349,
      "learning_rate": 1.9567201323514985e-05,
      "loss": 0.8245,
      "step": 1894
    },
    {
      "epoch": 1.2124120281509918,
      "grad_norm": 2.0856600751717354,
      "learning_rate": 1.9566598101414494e-05,
      "loss": 0.8212,
      "step": 1895
    },
    {
      "epoch": 1.2130518234165066,
      "grad_norm": 2.242233227167822,
      "learning_rate": 1.95659944685413e-05,
      "loss": 0.8297,
      "step": 1896
    },
    {
      "epoch": 1.2136916186820217,
      "grad_norm": 2.430046533671611,
      "learning_rate": 1.956539042492132e-05,
      "loss": 0.7797,
      "step": 1897
    },
    {
      "epoch": 1.2143314139475367,
      "grad_norm": 2.0267783475031944,
      "learning_rate": 1.9564785970580487e-05,
      "loss": 0.7899,
      "step": 1898
    },
    {
      "epoch": 1.2149712092130518,
      "grad_norm": 2.2279170434998674,
      "learning_rate": 1.956418110554476e-05,
      "loss": 0.7852,
      "step": 1899
    },
    {
      "epoch": 1.2156110044785668,
      "grad_norm": 2.0275088887164374,
      "learning_rate": 1.9563575829840115e-05,
      "loss": 0.7606,
      "step": 1900
    },
    {
      "epoch": 1.216250799744082,
      "grad_norm": 2.206530231681147,
      "learning_rate": 1.956297014349253e-05,
      "loss": 0.8004,
      "step": 1901
    },
    {
      "epoch": 1.216890595009597,
      "grad_norm": 2.307864780401253,
      "learning_rate": 1.956236404652802e-05,
      "loss": 0.7859,
      "step": 1902
    },
    {
      "epoch": 1.217530390275112,
      "grad_norm": 2.256648665482444,
      "learning_rate": 1.9561757538972613e-05,
      "loss": 0.765,
      "step": 1903
    },
    {
      "epoch": 1.218170185540627,
      "grad_norm": 2.160347423100246,
      "learning_rate": 1.956115062085234e-05,
      "loss": 0.7624,
      "step": 1904
    },
    {
      "epoch": 1.2188099808061421,
      "grad_norm": 2.201430755236383,
      "learning_rate": 1.956054329219327e-05,
      "loss": 0.8069,
      "step": 1905
    },
    {
      "epoch": 1.2194497760716572,
      "grad_norm": 2.1119426095750073,
      "learning_rate": 1.955993555302148e-05,
      "loss": 0.79,
      "step": 1906
    },
    {
      "epoch": 1.220089571337172,
      "grad_norm": 2.4903322592155868,
      "learning_rate": 1.955932740336306e-05,
      "loss": 0.8146,
      "step": 1907
    },
    {
      "epoch": 1.220729366602687,
      "grad_norm": 2.63954250907574,
      "learning_rate": 1.9558718843244127e-05,
      "loss": 0.7901,
      "step": 1908
    },
    {
      "epoch": 1.221369161868202,
      "grad_norm": 2.287297805657339,
      "learning_rate": 1.9558109872690808e-05,
      "loss": 0.8132,
      "step": 1909
    },
    {
      "epoch": 1.2220089571337172,
      "grad_norm": 2.1929907928874823,
      "learning_rate": 1.9557500491729257e-05,
      "loss": 0.8389,
      "step": 1910
    },
    {
      "epoch": 1.2226487523992322,
      "grad_norm": 2.4530356528577433,
      "learning_rate": 1.9556890700385637e-05,
      "loss": 0.7773,
      "step": 1911
    },
    {
      "epoch": 1.2232885476647473,
      "grad_norm": 2.57713631016997,
      "learning_rate": 1.9556280498686128e-05,
      "loss": 0.8308,
      "step": 1912
    },
    {
      "epoch": 1.2239283429302623,
      "grad_norm": 2.119898678041932,
      "learning_rate": 1.955566988665693e-05,
      "loss": 0.8378,
      "step": 1913
    },
    {
      "epoch": 1.2245681381957774,
      "grad_norm": 2.17166389590835,
      "learning_rate": 1.955505886432427e-05,
      "loss": 0.8372,
      "step": 1914
    },
    {
      "epoch": 1.2252079334612924,
      "grad_norm": 1.9803550161015853,
      "learning_rate": 1.9554447431714383e-05,
      "loss": 0.7622,
      "step": 1915
    },
    {
      "epoch": 1.2258477287268075,
      "grad_norm": 2.2829755476213154,
      "learning_rate": 1.9553835588853515e-05,
      "loss": 0.8264,
      "step": 1916
    },
    {
      "epoch": 1.2264875239923225,
      "grad_norm": 2.3354026812737727,
      "learning_rate": 1.955322333576794e-05,
      "loss": 0.7967,
      "step": 1917
    },
    {
      "epoch": 1.2271273192578376,
      "grad_norm": 2.4729933548056313,
      "learning_rate": 1.9552610672483954e-05,
      "loss": 0.8665,
      "step": 1918
    },
    {
      "epoch": 1.2277671145233526,
      "grad_norm": 2.057302732479572,
      "learning_rate": 1.9551997599027857e-05,
      "loss": 0.7307,
      "step": 1919
    },
    {
      "epoch": 1.2284069097888675,
      "grad_norm": 2.306679159157714,
      "learning_rate": 1.9551384115425973e-05,
      "loss": 0.8441,
      "step": 1920
    },
    {
      "epoch": 1.2290467050543825,
      "grad_norm": 2.2500622240729404,
      "learning_rate": 1.9550770221704644e-05,
      "loss": 0.7923,
      "step": 1921
    },
    {
      "epoch": 1.2296865003198976,
      "grad_norm": 2.217032803710492,
      "learning_rate": 1.9550155917890233e-05,
      "loss": 0.8106,
      "step": 1922
    },
    {
      "epoch": 1.2303262955854126,
      "grad_norm": 2.6061082307243404,
      "learning_rate": 1.9549541204009117e-05,
      "loss": 0.8483,
      "step": 1923
    },
    {
      "epoch": 1.2309660908509277,
      "grad_norm": 2.3958898455075883,
      "learning_rate": 1.954892608008769e-05,
      "loss": 0.7653,
      "step": 1924
    },
    {
      "epoch": 1.2316058861164427,
      "grad_norm": 2.191014023825517,
      "learning_rate": 1.954831054615236e-05,
      "loss": 0.796,
      "step": 1925
    },
    {
      "epoch": 1.2322456813819578,
      "grad_norm": 2.1431240428175853,
      "learning_rate": 1.9547694602229563e-05,
      "loss": 0.7833,
      "step": 1926
    },
    {
      "epoch": 1.2328854766474728,
      "grad_norm": 2.351721939590498,
      "learning_rate": 1.9547078248345738e-05,
      "loss": 0.777,
      "step": 1927
    },
    {
      "epoch": 1.2335252719129879,
      "grad_norm": 2.299229807739167,
      "learning_rate": 1.9546461484527362e-05,
      "loss": 0.7698,
      "step": 1928
    },
    {
      "epoch": 1.234165067178503,
      "grad_norm": 2.187655561927976,
      "learning_rate": 1.954584431080091e-05,
      "loss": 0.809,
      "step": 1929
    },
    {
      "epoch": 1.234804862444018,
      "grad_norm": 2.1434384482624504,
      "learning_rate": 1.9545226727192884e-05,
      "loss": 0.8139,
      "step": 1930
    },
    {
      "epoch": 1.235444657709533,
      "grad_norm": 2.0515006148499824,
      "learning_rate": 1.95446087337298e-05,
      "loss": 0.7505,
      "step": 1931
    },
    {
      "epoch": 1.236084452975048,
      "grad_norm": 2.4047434450024103,
      "learning_rate": 1.95439903304382e-05,
      "loss": 0.7872,
      "step": 1932
    },
    {
      "epoch": 1.236724248240563,
      "grad_norm": 1.9400459565585986,
      "learning_rate": 1.9543371517344628e-05,
      "loss": 0.7409,
      "step": 1933
    },
    {
      "epoch": 1.237364043506078,
      "grad_norm": 2.336249565162885,
      "learning_rate": 1.954275229447566e-05,
      "loss": 0.8224,
      "step": 1934
    },
    {
      "epoch": 1.238003838771593,
      "grad_norm": 2.1129481135621893,
      "learning_rate": 1.9542132661857885e-05,
      "loss": 0.828,
      "step": 1935
    },
    {
      "epoch": 1.238643634037108,
      "grad_norm": 2.0309951649662015,
      "learning_rate": 1.954151261951791e-05,
      "loss": 0.7394,
      "step": 1936
    },
    {
      "epoch": 1.2392834293026231,
      "grad_norm": 2.2318321762061326,
      "learning_rate": 1.9540892167482354e-05,
      "loss": 0.7668,
      "step": 1937
    },
    {
      "epoch": 1.2399232245681382,
      "grad_norm": 2.6322829427438483,
      "learning_rate": 1.954027130577786e-05,
      "loss": 0.8274,
      "step": 1938
    },
    {
      "epoch": 1.2405630198336532,
      "grad_norm": 2.353095105665104,
      "learning_rate": 1.9539650034431085e-05,
      "loss": 0.8116,
      "step": 1939
    },
    {
      "epoch": 1.2412028150991683,
      "grad_norm": 2.090128192427407,
      "learning_rate": 1.9539028353468713e-05,
      "loss": 0.785,
      "step": 1940
    },
    {
      "epoch": 1.2418426103646834,
      "grad_norm": 2.089789804276949,
      "learning_rate": 1.9538406262917427e-05,
      "loss": 0.7834,
      "step": 1941
    },
    {
      "epoch": 1.2424824056301984,
      "grad_norm": 2.1326116815922083,
      "learning_rate": 1.9537783762803945e-05,
      "loss": 0.8278,
      "step": 1942
    },
    {
      "epoch": 1.2431222008957135,
      "grad_norm": 1.985531477504481,
      "learning_rate": 1.9537160853154994e-05,
      "loss": 0.8083,
      "step": 1943
    },
    {
      "epoch": 1.2437619961612283,
      "grad_norm": 2.3880333857473994,
      "learning_rate": 1.9536537533997322e-05,
      "loss": 0.7911,
      "step": 1944
    },
    {
      "epoch": 1.2444017914267436,
      "grad_norm": 2.2336753193355876,
      "learning_rate": 1.9535913805357692e-05,
      "loss": 0.7711,
      "step": 1945
    },
    {
      "epoch": 1.2450415866922584,
      "grad_norm": 2.1259774799091526,
      "learning_rate": 1.9535289667262885e-05,
      "loss": 0.7472,
      "step": 1946
    },
    {
      "epoch": 1.2456813819577734,
      "grad_norm": 2.111948204114025,
      "learning_rate": 1.9534665119739707e-05,
      "loss": 0.7693,
      "step": 1947
    },
    {
      "epoch": 1.2463211772232885,
      "grad_norm": 2.387063363528489,
      "learning_rate": 1.953404016281496e-05,
      "loss": 0.8122,
      "step": 1948
    },
    {
      "epoch": 1.2469609724888036,
      "grad_norm": 2.6679452565966115,
      "learning_rate": 1.95334147965155e-05,
      "loss": 0.7889,
      "step": 1949
    },
    {
      "epoch": 1.2476007677543186,
      "grad_norm": 2.2142102666646646,
      "learning_rate": 1.9532789020868158e-05,
      "loss": 0.7711,
      "step": 1950
    },
    {
      "epoch": 1.2482405630198337,
      "grad_norm": 2.413408488450818,
      "learning_rate": 1.9532162835899816e-05,
      "loss": 0.8661,
      "step": 1951
    },
    {
      "epoch": 1.2488803582853487,
      "grad_norm": 2.12835097938973,
      "learning_rate": 1.9531536241637357e-05,
      "loss": 0.7674,
      "step": 1952
    },
    {
      "epoch": 1.2495201535508638,
      "grad_norm": 2.366306527992661,
      "learning_rate": 1.953090923810769e-05,
      "loss": 0.8263,
      "step": 1953
    },
    {
      "epoch": 1.2501599488163788,
      "grad_norm": 2.3341328912187183,
      "learning_rate": 1.9530281825337733e-05,
      "loss": 0.7507,
      "step": 1954
    },
    {
      "epoch": 1.2507997440818939,
      "grad_norm": 2.6841076051404897,
      "learning_rate": 1.952965400335443e-05,
      "loss": 0.8396,
      "step": 1955
    },
    {
      "epoch": 1.251439539347409,
      "grad_norm": 2.101430108189088,
      "learning_rate": 1.9529025772184732e-05,
      "loss": 0.8023,
      "step": 1956
    },
    {
      "epoch": 1.2520793346129238,
      "grad_norm": 2.3970532022592677,
      "learning_rate": 1.952839713185562e-05,
      "loss": 0.709,
      "step": 1957
    },
    {
      "epoch": 1.252719129878439,
      "grad_norm": 2.5601065030086,
      "learning_rate": 1.9527768082394085e-05,
      "loss": 0.7542,
      "step": 1958
    },
    {
      "epoch": 1.2533589251439539,
      "grad_norm": 2.322598848813845,
      "learning_rate": 1.9527138623827137e-05,
      "loss": 0.8235,
      "step": 1959
    },
    {
      "epoch": 1.253998720409469,
      "grad_norm": 2.384074444458167,
      "learning_rate": 1.9526508756181805e-05,
      "loss": 0.8303,
      "step": 1960
    },
    {
      "epoch": 1.254638515674984,
      "grad_norm": 2.5210195696016298,
      "learning_rate": 1.9525878479485133e-05,
      "loss": 0.8156,
      "step": 1961
    },
    {
      "epoch": 1.255278310940499,
      "grad_norm": 2.438571769747396,
      "learning_rate": 1.9525247793764183e-05,
      "loss": 0.8378,
      "step": 1962
    },
    {
      "epoch": 1.255918106206014,
      "grad_norm": 2.418509468723335,
      "learning_rate": 1.9524616699046035e-05,
      "loss": 0.8149,
      "step": 1963
    },
    {
      "epoch": 1.2565579014715291,
      "grad_norm": 2.4132282693188127,
      "learning_rate": 1.9523985195357792e-05,
      "loss": 0.7931,
      "step": 1964
    },
    {
      "epoch": 1.2571976967370442,
      "grad_norm": 2.237885082449609,
      "learning_rate": 1.9523353282726565e-05,
      "loss": 0.8263,
      "step": 1965
    },
    {
      "epoch": 1.2578374920025592,
      "grad_norm": 2.2512315299895302,
      "learning_rate": 1.952272096117949e-05,
      "loss": 0.8034,
      "step": 1966
    },
    {
      "epoch": 1.2584772872680743,
      "grad_norm": 2.3267007936026793,
      "learning_rate": 1.9522088230743716e-05,
      "loss": 0.8187,
      "step": 1967
    },
    {
      "epoch": 1.2591170825335891,
      "grad_norm": 2.2580336174470967,
      "learning_rate": 1.9521455091446412e-05,
      "loss": 0.8233,
      "step": 1968
    },
    {
      "epoch": 1.2597568777991044,
      "grad_norm": 2.0936136482142307,
      "learning_rate": 1.9520821543314764e-05,
      "loss": 0.7882,
      "step": 1969
    },
    {
      "epoch": 1.2603966730646192,
      "grad_norm": 2.063952652647538,
      "learning_rate": 1.9520187586375974e-05,
      "loss": 0.7675,
      "step": 1970
    },
    {
      "epoch": 1.2610364683301343,
      "grad_norm": 2.244587817155308,
      "learning_rate": 1.9519553220657267e-05,
      "loss": 0.7939,
      "step": 1971
    },
    {
      "epoch": 1.2616762635956493,
      "grad_norm": 2.3568287785414865,
      "learning_rate": 1.9518918446185874e-05,
      "loss": 0.7454,
      "step": 1972
    },
    {
      "epoch": 1.2623160588611644,
      "grad_norm": 2.3168276521041222,
      "learning_rate": 1.951828326298906e-05,
      "loss": 0.8407,
      "step": 1973
    },
    {
      "epoch": 1.2629558541266794,
      "grad_norm": 2.3991726957279687,
      "learning_rate": 1.951764767109409e-05,
      "loss": 0.7935,
      "step": 1974
    },
    {
      "epoch": 1.2635956493921945,
      "grad_norm": 2.6759361358149185,
      "learning_rate": 1.9517011670528262e-05,
      "loss": 0.7756,
      "step": 1975
    },
    {
      "epoch": 1.2642354446577095,
      "grad_norm": 2.575806343224496,
      "learning_rate": 1.9516375261318883e-05,
      "loss": 0.7812,
      "step": 1976
    },
    {
      "epoch": 1.2648752399232246,
      "grad_norm": 2.314556281540089,
      "learning_rate": 1.951573844349328e-05,
      "loss": 0.8593,
      "step": 1977
    },
    {
      "epoch": 1.2655150351887396,
      "grad_norm": 2.111910100803403,
      "learning_rate": 1.9515101217078795e-05,
      "loss": 0.8087,
      "step": 1978
    },
    {
      "epoch": 1.2661548304542547,
      "grad_norm": 2.579230622043187,
      "learning_rate": 1.9514463582102786e-05,
      "loss": 0.7786,
      "step": 1979
    },
    {
      "epoch": 1.2667946257197698,
      "grad_norm": 2.3492148861812807,
      "learning_rate": 1.9513825538592636e-05,
      "loss": 0.7426,
      "step": 1980
    },
    {
      "epoch": 1.2674344209852846,
      "grad_norm": 2.249762519281704,
      "learning_rate": 1.9513187086575745e-05,
      "loss": 0.7562,
      "step": 1981
    },
    {
      "epoch": 1.2680742162507999,
      "grad_norm": 2.803206004758508,
      "learning_rate": 1.9512548226079522e-05,
      "loss": 0.7944,
      "step": 1982
    },
    {
      "epoch": 1.2687140115163147,
      "grad_norm": 2.263993945797272,
      "learning_rate": 1.9511908957131398e-05,
      "loss": 0.788,
      "step": 1983
    },
    {
      "epoch": 1.2693538067818297,
      "grad_norm": 2.2210890694491114,
      "learning_rate": 1.9511269279758824e-05,
      "loss": 0.8506,
      "step": 1984
    },
    {
      "epoch": 1.2699936020473448,
      "grad_norm": 2.3554624209566293,
      "learning_rate": 1.9510629193989266e-05,
      "loss": 0.7923,
      "step": 1985
    },
    {
      "epoch": 1.2706333973128598,
      "grad_norm": 2.554340611401795,
      "learning_rate": 1.9509988699850214e-05,
      "loss": 0.7747,
      "step": 1986
    },
    {
      "epoch": 1.271273192578375,
      "grad_norm": 2.3118175526855467,
      "learning_rate": 1.9509347797369157e-05,
      "loss": 0.8276,
      "step": 1987
    },
    {
      "epoch": 1.27191298784389,
      "grad_norm": 2.313972139043244,
      "learning_rate": 1.9508706486573626e-05,
      "loss": 0.8433,
      "step": 1988
    },
    {
      "epoch": 1.272552783109405,
      "grad_norm": 2.278496414190003,
      "learning_rate": 1.950806476749115e-05,
      "loss": 0.7945,
      "step": 1989
    },
    {
      "epoch": 1.27319257837492,
      "grad_norm": 2.286471336705756,
      "learning_rate": 1.950742264014929e-05,
      "loss": 0.7883,
      "step": 1990
    },
    {
      "epoch": 1.2738323736404351,
      "grad_norm": 2.507035675735442,
      "learning_rate": 1.9506780104575612e-05,
      "loss": 0.7992,
      "step": 1991
    },
    {
      "epoch": 1.2744721689059502,
      "grad_norm": 2.2549263951370015,
      "learning_rate": 1.9506137160797706e-05,
      "loss": 0.783,
      "step": 1992
    },
    {
      "epoch": 1.2751119641714652,
      "grad_norm": 2.069325017015306,
      "learning_rate": 1.9505493808843182e-05,
      "loss": 0.7819,
      "step": 1993
    },
    {
      "epoch": 1.27575175943698,
      "grad_norm": 2.68983087070434,
      "learning_rate": 1.950485004873966e-05,
      "loss": 0.7821,
      "step": 1994
    },
    {
      "epoch": 1.2763915547024953,
      "grad_norm": 2.4694662738010646,
      "learning_rate": 1.950420588051479e-05,
      "loss": 0.7995,
      "step": 1995
    },
    {
      "epoch": 1.2770313499680102,
      "grad_norm": 2.1814616133018423,
      "learning_rate": 1.9503561304196227e-05,
      "loss": 0.735,
      "step": 1996
    },
    {
      "epoch": 1.2776711452335252,
      "grad_norm": 2.3657433969365624,
      "learning_rate": 1.950291631981164e-05,
      "loss": 0.7729,
      "step": 1997
    },
    {
      "epoch": 1.2783109404990403,
      "grad_norm": 2.552330006515524,
      "learning_rate": 1.9502270927388736e-05,
      "loss": 0.8121,
      "step": 1998
    },
    {
      "epoch": 1.2789507357645553,
      "grad_norm": 2.499871970891502,
      "learning_rate": 1.950162512695522e-05,
      "loss": 0.8223,
      "step": 1999
    },
    {
      "epoch": 1.2795905310300704,
      "grad_norm": 2.4354980597901688,
      "learning_rate": 1.9500978918538823e-05,
      "loss": 0.7754,
      "step": 2000
    },
    {
      "epoch": 1.2802303262955854,
      "grad_norm": 2.3587625102174865,
      "learning_rate": 1.9500332302167292e-05,
      "loss": 0.7731,
      "step": 2001
    },
    {
      "epoch": 1.2808701215611005,
      "grad_norm": 2.2063367993280067,
      "learning_rate": 1.9499685277868394e-05,
      "loss": 0.7762,
      "step": 2002
    },
    {
      "epoch": 1.2815099168266155,
      "grad_norm": 2.2388005218941855,
      "learning_rate": 1.9499037845669907e-05,
      "loss": 0.8072,
      "step": 2003
    },
    {
      "epoch": 1.2821497120921306,
      "grad_norm": 2.105069436768517,
      "learning_rate": 1.9498390005599632e-05,
      "loss": 0.8126,
      "step": 2004
    },
    {
      "epoch": 1.2827895073576456,
      "grad_norm": 2.626262747845356,
      "learning_rate": 1.9497741757685388e-05,
      "loss": 0.7715,
      "step": 2005
    },
    {
      "epoch": 1.2834293026231607,
      "grad_norm": 2.652709943851835,
      "learning_rate": 1.9497093101955005e-05,
      "loss": 0.7523,
      "step": 2006
    },
    {
      "epoch": 1.2840690978886755,
      "grad_norm": 2.5424704102831224,
      "learning_rate": 1.9496444038436338e-05,
      "loss": 0.7744,
      "step": 2007
    },
    {
      "epoch": 1.2847088931541908,
      "grad_norm": 2.289710482541086,
      "learning_rate": 1.9495794567157264e-05,
      "loss": 0.7881,
      "step": 2008
    },
    {
      "epoch": 1.2853486884197056,
      "grad_norm": 2.194509612415007,
      "learning_rate": 1.9495144688145656e-05,
      "loss": 0.7425,
      "step": 2009
    },
    {
      "epoch": 1.2859884836852207,
      "grad_norm": 2.3465259134593466,
      "learning_rate": 1.9494494401429427e-05,
      "loss": 0.7695,
      "step": 2010
    },
    {
      "epoch": 1.2866282789507357,
      "grad_norm": 2.4622396950032823,
      "learning_rate": 1.94938437070365e-05,
      "loss": 0.7824,
      "step": 2011
    },
    {
      "epoch": 1.2872680742162508,
      "grad_norm": 1.8618633363195625,
      "learning_rate": 1.949319260499481e-05,
      "loss": 0.7432,
      "step": 2012
    },
    {
      "epoch": 1.2879078694817658,
      "grad_norm": 2.2212667429445827,
      "learning_rate": 1.9492541095332317e-05,
      "loss": 0.7615,
      "step": 2013
    },
    {
      "epoch": 1.2885476647472809,
      "grad_norm": 2.57979227451785,
      "learning_rate": 1.9491889178076997e-05,
      "loss": 0.8455,
      "step": 2014
    },
    {
      "epoch": 1.289187460012796,
      "grad_norm": 2.473545264973096,
      "learning_rate": 1.949123685325684e-05,
      "loss": 0.7839,
      "step": 2015
    },
    {
      "epoch": 1.289827255278311,
      "grad_norm": 2.330363516845394,
      "learning_rate": 1.9490584120899854e-05,
      "loss": 0.8147,
      "step": 2016
    },
    {
      "epoch": 1.290467050543826,
      "grad_norm": 2.235817996021775,
      "learning_rate": 1.948993098103407e-05,
      "loss": 0.7916,
      "step": 2017
    },
    {
      "epoch": 1.2911068458093409,
      "grad_norm": 1.9537079496409928,
      "learning_rate": 1.948927743368753e-05,
      "loss": 0.7873,
      "step": 2018
    },
    {
      "epoch": 1.2917466410748562,
      "grad_norm": 2.1445471441162334,
      "learning_rate": 1.94886234788883e-05,
      "loss": 0.7768,
      "step": 2019
    },
    {
      "epoch": 1.292386436340371,
      "grad_norm": 2.2608515193311285,
      "learning_rate": 1.9487969116664452e-05,
      "loss": 0.809,
      "step": 2020
    },
    {
      "epoch": 1.293026231605886,
      "grad_norm": 2.509488458667883,
      "learning_rate": 1.948731434704409e-05,
      "loss": 0.8248,
      "step": 2021
    },
    {
      "epoch": 1.293666026871401,
      "grad_norm": 2.113815724260608,
      "learning_rate": 1.9486659170055327e-05,
      "loss": 0.801,
      "step": 2022
    },
    {
      "epoch": 1.2943058221369161,
      "grad_norm": 2.449345512317645,
      "learning_rate": 1.9486003585726293e-05,
      "loss": 0.7589,
      "step": 2023
    },
    {
      "epoch": 1.2949456174024312,
      "grad_norm": 2.4211035061913813,
      "learning_rate": 1.9485347594085145e-05,
      "loss": 0.7463,
      "step": 2024
    },
    {
      "epoch": 1.2955854126679462,
      "grad_norm": 2.410784528295998,
      "learning_rate": 1.9484691195160035e-05,
      "loss": 0.8435,
      "step": 2025
    },
    {
      "epoch": 1.2962252079334613,
      "grad_norm": 2.2920845313612546,
      "learning_rate": 1.9484034388979167e-05,
      "loss": 0.8111,
      "step": 2026
    },
    {
      "epoch": 1.2968650031989764,
      "grad_norm": 2.417983754440061,
      "learning_rate": 1.9483377175570727e-05,
      "loss": 0.7822,
      "step": 2027
    },
    {
      "epoch": 1.2975047984644914,
      "grad_norm": 2.126701488603786,
      "learning_rate": 1.9482719554962944e-05,
      "loss": 0.7744,
      "step": 2028
    },
    {
      "epoch": 1.2981445937300065,
      "grad_norm": 2.4727063923448225,
      "learning_rate": 1.9482061527184048e-05,
      "loss": 0.7793,
      "step": 2029
    },
    {
      "epoch": 1.2987843889955215,
      "grad_norm": 2.5437206778054975,
      "learning_rate": 1.9481403092262303e-05,
      "loss": 0.7482,
      "step": 2030
    },
    {
      "epoch": 1.2994241842610363,
      "grad_norm": 2.5017237705890465,
      "learning_rate": 1.948074425022597e-05,
      "loss": 0.8029,
      "step": 2031
    },
    {
      "epoch": 1.3000639795265516,
      "grad_norm": 2.6402689907179018,
      "learning_rate": 1.9480085001103344e-05,
      "loss": 0.7621,
      "step": 2032
    },
    {
      "epoch": 1.3007037747920664,
      "grad_norm": 2.3117544372654906,
      "learning_rate": 1.9479425344922733e-05,
      "loss": 0.7953,
      "step": 2033
    },
    {
      "epoch": 1.3013435700575815,
      "grad_norm": 2.4382684189717243,
      "learning_rate": 1.947876528171246e-05,
      "loss": 0.7884,
      "step": 2034
    },
    {
      "epoch": 1.3019833653230966,
      "grad_norm": 2.7589026444046363,
      "learning_rate": 1.947810481150087e-05,
      "loss": 0.8479,
      "step": 2035
    },
    {
      "epoch": 1.3026231605886116,
      "grad_norm": 2.302513861570835,
      "learning_rate": 1.9477443934316316e-05,
      "loss": 0.8153,
      "step": 2036
    },
    {
      "epoch": 1.3032629558541267,
      "grad_norm": 2.2621559899747585,
      "learning_rate": 1.947678265018718e-05,
      "loss": 0.7608,
      "step": 2037
    },
    {
      "epoch": 1.3039027511196417,
      "grad_norm": 2.46064007553459,
      "learning_rate": 1.9476120959141857e-05,
      "loss": 0.7788,
      "step": 2038
    },
    {
      "epoch": 1.3045425463851568,
      "grad_norm": 2.282716861396147,
      "learning_rate": 1.9475458861208754e-05,
      "loss": 0.7596,
      "step": 2039
    },
    {
      "epoch": 1.3051823416506718,
      "grad_norm": 2.216711505648095,
      "learning_rate": 1.94747963564163e-05,
      "loss": 0.8036,
      "step": 2040
    },
    {
      "epoch": 1.3058221369161869,
      "grad_norm": 2.0550701054793667,
      "learning_rate": 1.947413344479295e-05,
      "loss": 0.7969,
      "step": 2041
    },
    {
      "epoch": 1.306461932181702,
      "grad_norm": 2.3359277708245085,
      "learning_rate": 1.9473470126367162e-05,
      "loss": 0.7805,
      "step": 2042
    },
    {
      "epoch": 1.307101727447217,
      "grad_norm": 2.435733286511535,
      "learning_rate": 1.947280640116742e-05,
      "loss": 0.8393,
      "step": 2043
    },
    {
      "epoch": 1.3077415227127318,
      "grad_norm": 2.151679403560261,
      "learning_rate": 1.9472142269222217e-05,
      "loss": 0.7735,
      "step": 2044
    },
    {
      "epoch": 1.308381317978247,
      "grad_norm": 2.102430601745267,
      "learning_rate": 1.947147773056008e-05,
      "loss": 0.8278,
      "step": 2045
    },
    {
      "epoch": 1.309021113243762,
      "grad_norm": 2.3930545769934475,
      "learning_rate": 1.9470812785209533e-05,
      "loss": 0.8178,
      "step": 2046
    },
    {
      "epoch": 1.309660908509277,
      "grad_norm": 2.0647698019957184,
      "learning_rate": 1.9470147433199135e-05,
      "loss": 0.8016,
      "step": 2047
    },
    {
      "epoch": 1.310300703774792,
      "grad_norm": 2.4871579073943764,
      "learning_rate": 1.9469481674557454e-05,
      "loss": 0.8021,
      "step": 2048
    },
    {
      "epoch": 1.310940499040307,
      "grad_norm": 2.198183371161303,
      "learning_rate": 1.946881550931307e-05,
      "loss": 0.7712,
      "step": 2049
    },
    {
      "epoch": 1.3115802943058221,
      "grad_norm": 2.056444750262997,
      "learning_rate": 1.9468148937494598e-05,
      "loss": 0.7682,
      "step": 2050
    },
    {
      "epoch": 1.3122200895713372,
      "grad_norm": 2.316547472080818,
      "learning_rate": 1.946748195913065e-05,
      "loss": 0.7527,
      "step": 2051
    },
    {
      "epoch": 1.3128598848368522,
      "grad_norm": 2.271653426320367,
      "learning_rate": 1.946681457424987e-05,
      "loss": 0.733,
      "step": 2052
    },
    {
      "epoch": 1.3134996801023673,
      "grad_norm": 2.2307773723415956,
      "learning_rate": 1.9466146782880907e-05,
      "loss": 0.7854,
      "step": 2053
    },
    {
      "epoch": 1.3141394753678823,
      "grad_norm": 2.258997355898933,
      "learning_rate": 1.946547858505244e-05,
      "loss": 0.7853,
      "step": 2054
    },
    {
      "epoch": 1.3147792706333974,
      "grad_norm": 2.4271213474957483,
      "learning_rate": 1.9464809980793166e-05,
      "loss": 0.739,
      "step": 2055
    },
    {
      "epoch": 1.3154190658989124,
      "grad_norm": 2.259595091505086,
      "learning_rate": 1.9464140970131786e-05,
      "loss": 0.7448,
      "step": 2056
    },
    {
      "epoch": 1.3160588611644273,
      "grad_norm": 2.629851468751884,
      "learning_rate": 1.946347155309703e-05,
      "loss": 0.769,
      "step": 2057
    },
    {
      "epoch": 1.3166986564299425,
      "grad_norm": 2.6051848501406076,
      "learning_rate": 1.9462801729717636e-05,
      "loss": 0.793,
      "step": 2058
    },
    {
      "epoch": 1.3173384516954574,
      "grad_norm": 2.4822966370004327,
      "learning_rate": 1.946213150002237e-05,
      "loss": 0.7712,
      "step": 2059
    },
    {
      "epoch": 1.3179782469609724,
      "grad_norm": 2.2053864011295485,
      "learning_rate": 1.946146086404001e-05,
      "loss": 0.7301,
      "step": 2060
    },
    {
      "epoch": 1.3186180422264875,
      "grad_norm": 2.0600906989846304,
      "learning_rate": 1.946078982179935e-05,
      "loss": 0.7115,
      "step": 2061
    },
    {
      "epoch": 1.3192578374920025,
      "grad_norm": 2.1983438162095474,
      "learning_rate": 1.9460118373329205e-05,
      "loss": 0.8165,
      "step": 2062
    },
    {
      "epoch": 1.3198976327575176,
      "grad_norm": 2.148194951110195,
      "learning_rate": 1.9459446518658404e-05,
      "loss": 0.7292,
      "step": 2063
    },
    {
      "epoch": 1.3205374280230326,
      "grad_norm": 2.054847409804944,
      "learning_rate": 1.94587742578158e-05,
      "loss": 0.7827,
      "step": 2064
    },
    {
      "epoch": 1.3211772232885477,
      "grad_norm": 2.258783534217873,
      "learning_rate": 1.9458101590830253e-05,
      "loss": 0.8146,
      "step": 2065
    },
    {
      "epoch": 1.3218170185540628,
      "grad_norm": 2.3215566075493035,
      "learning_rate": 1.945742851773065e-05,
      "loss": 0.7557,
      "step": 2066
    },
    {
      "epoch": 1.3224568138195778,
      "grad_norm": 2.1153030138009505,
      "learning_rate": 1.9456755038545884e-05,
      "loss": 0.7732,
      "step": 2067
    },
    {
      "epoch": 1.3230966090850926,
      "grad_norm": 2.2779914255888642,
      "learning_rate": 1.9456081153304886e-05,
      "loss": 0.7941,
      "step": 2068
    },
    {
      "epoch": 1.323736404350608,
      "grad_norm": 2.704725114899181,
      "learning_rate": 1.945540686203658e-05,
      "loss": 0.7692,
      "step": 2069
    },
    {
      "epoch": 1.3243761996161227,
      "grad_norm": 2.3243731886988908,
      "learning_rate": 1.9454732164769927e-05,
      "loss": 0.725,
      "step": 2070
    },
    {
      "epoch": 1.3250159948816378,
      "grad_norm": 2.2617775673241054,
      "learning_rate": 1.9454057061533887e-05,
      "loss": 0.7875,
      "step": 2071
    },
    {
      "epoch": 1.3256557901471528,
      "grad_norm": 2.2339089262664715,
      "learning_rate": 1.9453381552357463e-05,
      "loss": 0.7804,
      "step": 2072
    },
    {
      "epoch": 1.326295585412668,
      "grad_norm": 2.265442894277755,
      "learning_rate": 1.9452705637269647e-05,
      "loss": 0.7761,
      "step": 2073
    },
    {
      "epoch": 1.326935380678183,
      "grad_norm": 2.3071143404215864,
      "learning_rate": 1.9452029316299462e-05,
      "loss": 0.7626,
      "step": 2074
    },
    {
      "epoch": 1.327575175943698,
      "grad_norm": 2.291893351954054,
      "learning_rate": 1.9451352589475958e-05,
      "loss": 0.7501,
      "step": 2075
    },
    {
      "epoch": 1.328214971209213,
      "grad_norm": 2.2540201108717555,
      "learning_rate": 1.9450675456828184e-05,
      "loss": 0.7623,
      "step": 2076
    },
    {
      "epoch": 1.328854766474728,
      "grad_norm": 2.192319527976508,
      "learning_rate": 1.9449997918385217e-05,
      "loss": 0.8187,
      "step": 2077
    },
    {
      "epoch": 1.3294945617402432,
      "grad_norm": 2.421607560207123,
      "learning_rate": 1.9449319974176152e-05,
      "loss": 0.8014,
      "step": 2078
    },
    {
      "epoch": 1.3301343570057582,
      "grad_norm": 2.2371246765486377,
      "learning_rate": 1.9448641624230093e-05,
      "loss": 0.8056,
      "step": 2079
    },
    {
      "epoch": 1.3307741522712733,
      "grad_norm": 2.2146632236128467,
      "learning_rate": 1.9447962868576173e-05,
      "loss": 0.8089,
      "step": 2080
    },
    {
      "epoch": 1.331413947536788,
      "grad_norm": 2.202115462595019,
      "learning_rate": 1.9447283707243535e-05,
      "loss": 0.802,
      "step": 2081
    },
    {
      "epoch": 1.3320537428023034,
      "grad_norm": 2.2814842952511376,
      "learning_rate": 1.944660414026134e-05,
      "loss": 0.8178,
      "step": 2082
    },
    {
      "epoch": 1.3326935380678182,
      "grad_norm": 2.201015543231886,
      "learning_rate": 1.9445924167658763e-05,
      "loss": 0.7931,
      "step": 2083
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 2.0838437520867994,
      "learning_rate": 1.944524378946501e-05,
      "loss": 0.7952,
      "step": 2084
    },
    {
      "epoch": 1.3339731285988483,
      "grad_norm": 2.235382749346849,
      "learning_rate": 1.9444563005709286e-05,
      "loss": 0.7691,
      "step": 2085
    },
    {
      "epoch": 1.3346129238643634,
      "grad_norm": 2.6893500386552947,
      "learning_rate": 1.9443881816420828e-05,
      "loss": 0.8136,
      "step": 2086
    },
    {
      "epoch": 1.3352527191298784,
      "grad_norm": 2.079501825367117,
      "learning_rate": 1.944320022162889e-05,
      "loss": 0.7323,
      "step": 2087
    },
    {
      "epoch": 1.3358925143953935,
      "grad_norm": 2.1026441955085713,
      "learning_rate": 1.9442518221362725e-05,
      "loss": 0.7773,
      "step": 2088
    },
    {
      "epoch": 1.3365323096609085,
      "grad_norm": 2.311789148829814,
      "learning_rate": 1.944183581565163e-05,
      "loss": 0.7763,
      "step": 2089
    },
    {
      "epoch": 1.3371721049264236,
      "grad_norm": 2.0721956637083925,
      "learning_rate": 1.94411530045249e-05,
      "loss": 0.7455,
      "step": 2090
    },
    {
      "epoch": 1.3378119001919386,
      "grad_norm": 2.135522625249261,
      "learning_rate": 1.944046978801185e-05,
      "loss": 0.7488,
      "step": 2091
    },
    {
      "epoch": 1.3384516954574537,
      "grad_norm": 2.065430339717648,
      "learning_rate": 1.9439786166141824e-05,
      "loss": 0.7965,
      "step": 2092
    },
    {
      "epoch": 1.3390914907229687,
      "grad_norm": 2.31514039701495,
      "learning_rate": 1.943910213894417e-05,
      "loss": 0.7774,
      "step": 2093
    },
    {
      "epoch": 1.3397312859884836,
      "grad_norm": 2.2522104328255033,
      "learning_rate": 1.9438417706448267e-05,
      "loss": 0.7151,
      "step": 2094
    },
    {
      "epoch": 1.3403710812539988,
      "grad_norm": 2.322196998412051,
      "learning_rate": 1.943773286868349e-05,
      "loss": 0.7951,
      "step": 2095
    },
    {
      "epoch": 1.3410108765195137,
      "grad_norm": 2.176333881589983,
      "learning_rate": 1.9437047625679256e-05,
      "loss": 0.8064,
      "step": 2096
    },
    {
      "epoch": 1.3416506717850287,
      "grad_norm": 2.517529968982562,
      "learning_rate": 1.9436361977464985e-05,
      "loss": 0.785,
      "step": 2097
    },
    {
      "epoch": 1.3422904670505438,
      "grad_norm": 2.43468509345018,
      "learning_rate": 1.9435675924070115e-05,
      "loss": 0.7764,
      "step": 2098
    },
    {
      "epoch": 1.3429302623160588,
      "grad_norm": 2.2113556815245277,
      "learning_rate": 1.9434989465524103e-05,
      "loss": 0.7818,
      "step": 2099
    },
    {
      "epoch": 1.3435700575815739,
      "grad_norm": 2.202881111677277,
      "learning_rate": 1.943430260185643e-05,
      "loss": 0.7501,
      "step": 2100
    },
    {
      "epoch": 1.344209852847089,
      "grad_norm": 2.531203061765816,
      "learning_rate": 1.9433615333096588e-05,
      "loss": 0.7879,
      "step": 2101
    },
    {
      "epoch": 1.344849648112604,
      "grad_norm": 2.3614948034153733,
      "learning_rate": 1.943292765927408e-05,
      "loss": 0.7734,
      "step": 2102
    },
    {
      "epoch": 1.345489443378119,
      "grad_norm": 2.3456132910406713,
      "learning_rate": 1.9432239580418438e-05,
      "loss": 0.7641,
      "step": 2103
    },
    {
      "epoch": 1.346129238643634,
      "grad_norm": 2.2203685626743286,
      "learning_rate": 1.9431551096559213e-05,
      "loss": 0.7726,
      "step": 2104
    },
    {
      "epoch": 1.3467690339091491,
      "grad_norm": 2.2171057768815334,
      "learning_rate": 1.9430862207725957e-05,
      "loss": 0.7796,
      "step": 2105
    },
    {
      "epoch": 1.3474088291746642,
      "grad_norm": 2.3003885041220298,
      "learning_rate": 1.9430172913948254e-05,
      "loss": 0.8004,
      "step": 2106
    },
    {
      "epoch": 1.348048624440179,
      "grad_norm": 2.15543885038177,
      "learning_rate": 1.94294832152557e-05,
      "loss": 0.7301,
      "step": 2107
    },
    {
      "epoch": 1.3486884197056943,
      "grad_norm": 2.8391033915662574,
      "learning_rate": 1.9428793111677913e-05,
      "loss": 0.7882,
      "step": 2108
    },
    {
      "epoch": 1.3493282149712091,
      "grad_norm": 2.4409726298813617,
      "learning_rate": 1.9428102603244522e-05,
      "loss": 0.771,
      "step": 2109
    },
    {
      "epoch": 1.3499680102367242,
      "grad_norm": 2.2652584432737872,
      "learning_rate": 1.9427411689985175e-05,
      "loss": 0.7589,
      "step": 2110
    },
    {
      "epoch": 1.3506078055022392,
      "grad_norm": 2.3713696629240517,
      "learning_rate": 1.942672037192954e-05,
      "loss": 0.7834,
      "step": 2111
    },
    {
      "epoch": 1.3512476007677543,
      "grad_norm": 2.4427526498926895,
      "learning_rate": 1.9426028649107304e-05,
      "loss": 0.7558,
      "step": 2112
    },
    {
      "epoch": 1.3518873960332694,
      "grad_norm": 2.204517401645201,
      "learning_rate": 1.942533652154816e-05,
      "loss": 0.768,
      "step": 2113
    },
    {
      "epoch": 1.3525271912987844,
      "grad_norm": 2.2241790265030454,
      "learning_rate": 1.942464398928184e-05,
      "loss": 0.7965,
      "step": 2114
    },
    {
      "epoch": 1.3531669865642995,
      "grad_norm": 2.3273859479050154,
      "learning_rate": 1.9423951052338066e-05,
      "loss": 0.808,
      "step": 2115
    },
    {
      "epoch": 1.3538067818298145,
      "grad_norm": 2.0910213701996816,
      "learning_rate": 1.9423257710746596e-05,
      "loss": 0.828,
      "step": 2116
    },
    {
      "epoch": 1.3544465770953296,
      "grad_norm": 2.474271430836497,
      "learning_rate": 1.9422563964537208e-05,
      "loss": 0.8083,
      "step": 2117
    },
    {
      "epoch": 1.3550863723608444,
      "grad_norm": 2.049111339628477,
      "learning_rate": 1.9421869813739683e-05,
      "loss": 0.7669,
      "step": 2118
    },
    {
      "epoch": 1.3557261676263597,
      "grad_norm": 2.3103841552815014,
      "learning_rate": 1.9421175258383827e-05,
      "loss": 0.8222,
      "step": 2119
    },
    {
      "epoch": 1.3563659628918745,
      "grad_norm": 2.3697442114171183,
      "learning_rate": 1.9420480298499466e-05,
      "loss": 0.8013,
      "step": 2120
    },
    {
      "epoch": 1.3570057581573896,
      "grad_norm": 2.0449496071960493,
      "learning_rate": 1.9419784934116438e-05,
      "loss": 0.7638,
      "step": 2121
    },
    {
      "epoch": 1.3576455534229046,
      "grad_norm": 1.9610985794956577,
      "learning_rate": 1.9419089165264603e-05,
      "loss": 0.7508,
      "step": 2122
    },
    {
      "epoch": 1.3582853486884197,
      "grad_norm": 2.0087168563738413,
      "learning_rate": 1.9418392991973832e-05,
      "loss": 0.755,
      "step": 2123
    },
    {
      "epoch": 1.3589251439539347,
      "grad_norm": 2.1578790419559284,
      "learning_rate": 1.9417696414274024e-05,
      "loss": 0.7697,
      "step": 2124
    },
    {
      "epoch": 1.3595649392194498,
      "grad_norm": 2.253966575353397,
      "learning_rate": 1.941699943219508e-05,
      "loss": 0.759,
      "step": 2125
    },
    {
      "epoch": 1.3602047344849648,
      "grad_norm": 2.218501057278135,
      "learning_rate": 1.9416302045766936e-05,
      "loss": 0.7748,
      "step": 2126
    },
    {
      "epoch": 1.3608445297504799,
      "grad_norm": 2.1067660831194606,
      "learning_rate": 1.9415604255019528e-05,
      "loss": 0.7342,
      "step": 2127
    },
    {
      "epoch": 1.361484325015995,
      "grad_norm": 2.3006035585771865,
      "learning_rate": 1.9414906059982827e-05,
      "loss": 0.7239,
      "step": 2128
    },
    {
      "epoch": 1.36212412028151,
      "grad_norm": 2.148879861451695,
      "learning_rate": 1.9414207460686804e-05,
      "loss": 0.8395,
      "step": 2129
    },
    {
      "epoch": 1.362763915547025,
      "grad_norm": 2.3704149397615164,
      "learning_rate": 1.941350845716146e-05,
      "loss": 0.7753,
      "step": 2130
    },
    {
      "epoch": 1.3634037108125399,
      "grad_norm": 2.270229284801743,
      "learning_rate": 1.941280904943681e-05,
      "loss": 0.7573,
      "step": 2131
    },
    {
      "epoch": 1.3640435060780551,
      "grad_norm": 2.0250875651673725,
      "learning_rate": 1.9412109237542883e-05,
      "loss": 0.7701,
      "step": 2132
    },
    {
      "epoch": 1.36468330134357,
      "grad_norm": 2.2553559734273154,
      "learning_rate": 1.9411409021509727e-05,
      "loss": 0.7959,
      "step": 2133
    },
    {
      "epoch": 1.365323096609085,
      "grad_norm": 1.9560417477326912,
      "learning_rate": 1.9410708401367408e-05,
      "loss": 0.7558,
      "step": 2134
    },
    {
      "epoch": 1.3659628918746,
      "grad_norm": 2.180931243143994,
      "learning_rate": 1.9410007377146012e-05,
      "loss": 0.7302,
      "step": 2135
    },
    {
      "epoch": 1.3666026871401151,
      "grad_norm": 2.5317553888423023,
      "learning_rate": 1.940930594887564e-05,
      "loss": 0.7603,
      "step": 2136
    },
    {
      "epoch": 1.3672424824056302,
      "grad_norm": 2.209017022989144,
      "learning_rate": 1.9408604116586404e-05,
      "loss": 0.7388,
      "step": 2137
    },
    {
      "epoch": 1.3678822776711452,
      "grad_norm": 2.24411275227924,
      "learning_rate": 1.9407901880308447e-05,
      "loss": 0.8395,
      "step": 2138
    },
    {
      "epoch": 1.3685220729366603,
      "grad_norm": 2.2722504185709664,
      "learning_rate": 1.940719924007192e-05,
      "loss": 0.7863,
      "step": 2139
    },
    {
      "epoch": 1.3691618682021753,
      "grad_norm": 2.150813565496658,
      "learning_rate": 1.940649619590699e-05,
      "loss": 0.76,
      "step": 2140
    },
    {
      "epoch": 1.3698016634676904,
      "grad_norm": 2.0216188947089537,
      "learning_rate": 1.9405792747843845e-05,
      "loss": 0.7589,
      "step": 2141
    },
    {
      "epoch": 1.3704414587332054,
      "grad_norm": 2.1139161935867214,
      "learning_rate": 1.9405088895912693e-05,
      "loss": 0.7691,
      "step": 2142
    },
    {
      "epoch": 1.3710812539987205,
      "grad_norm": 1.9700807023695592,
      "learning_rate": 1.9404384640143754e-05,
      "loss": 0.8256,
      "step": 2143
    },
    {
      "epoch": 1.3717210492642353,
      "grad_norm": 2.754996093861431,
      "learning_rate": 1.9403679980567266e-05,
      "loss": 0.8091,
      "step": 2144
    },
    {
      "epoch": 1.3723608445297506,
      "grad_norm": 2.6999343430208835,
      "learning_rate": 1.940297491721349e-05,
      "loss": 0.8089,
      "step": 2145
    },
    {
      "epoch": 1.3730006397952654,
      "grad_norm": 2.080821200181188,
      "learning_rate": 1.9402269450112696e-05,
      "loss": 0.7693,
      "step": 2146
    },
    {
      "epoch": 1.3736404350607805,
      "grad_norm": 2.8084919941633864,
      "learning_rate": 1.9401563579295177e-05,
      "loss": 0.7504,
      "step": 2147
    },
    {
      "epoch": 1.3742802303262955,
      "grad_norm": 2.712426963140468,
      "learning_rate": 1.940085730479124e-05,
      "loss": 0.7875,
      "step": 2148
    },
    {
      "epoch": 1.3749200255918106,
      "grad_norm": 2.9991031307523786,
      "learning_rate": 1.9400150626631218e-05,
      "loss": 0.7769,
      "step": 2149
    },
    {
      "epoch": 1.3755598208573256,
      "grad_norm": 2.3156922428609956,
      "learning_rate": 1.9399443544845446e-05,
      "loss": 0.7864,
      "step": 2150
    },
    {
      "epoch": 1.3761996161228407,
      "grad_norm": 2.3171126592224716,
      "learning_rate": 1.939873605946429e-05,
      "loss": 0.7449,
      "step": 2151
    },
    {
      "epoch": 1.3768394113883557,
      "grad_norm": 3.6751005159634484,
      "learning_rate": 1.939802817051813e-05,
      "loss": 0.726,
      "step": 2152
    },
    {
      "epoch": 1.3774792066538708,
      "grad_norm": 2.4611035595703052,
      "learning_rate": 1.9397319878037352e-05,
      "loss": 0.7855,
      "step": 2153
    },
    {
      "epoch": 1.3781190019193859,
      "grad_norm": 2.2275571403630114,
      "learning_rate": 1.9396611182052378e-05,
      "loss": 0.76,
      "step": 2154
    },
    {
      "epoch": 1.378758797184901,
      "grad_norm": 2.358080791588997,
      "learning_rate": 1.9395902082593638e-05,
      "loss": 0.7722,
      "step": 2155
    },
    {
      "epoch": 1.379398592450416,
      "grad_norm": 2.4041845857127178,
      "learning_rate": 1.939519257969157e-05,
      "loss": 0.7753,
      "step": 2156
    },
    {
      "epoch": 1.3800383877159308,
      "grad_norm": 2.503856834267113,
      "learning_rate": 1.9394482673376647e-05,
      "loss": 0.7963,
      "step": 2157
    },
    {
      "epoch": 1.380678182981446,
      "grad_norm": 2.282889504919956,
      "learning_rate": 1.9393772363679353e-05,
      "loss": 0.7913,
      "step": 2158
    },
    {
      "epoch": 1.381317978246961,
      "grad_norm": 2.1400541011051644,
      "learning_rate": 1.9393061650630184e-05,
      "loss": 0.7617,
      "step": 2159
    },
    {
      "epoch": 1.381957773512476,
      "grad_norm": 2.382266841166675,
      "learning_rate": 1.9392350534259652e-05,
      "loss": 0.7791,
      "step": 2160
    },
    {
      "epoch": 1.382597568777991,
      "grad_norm": 2.641199248803219,
      "learning_rate": 1.9391639014598297e-05,
      "loss": 0.8125,
      "step": 2161
    },
    {
      "epoch": 1.383237364043506,
      "grad_norm": 2.312040929523328,
      "learning_rate": 1.939092709167667e-05,
      "loss": 0.7686,
      "step": 2162
    },
    {
      "epoch": 1.383877159309021,
      "grad_norm": 2.347322683560045,
      "learning_rate": 1.9390214765525343e-05,
      "loss": 0.7486,
      "step": 2163
    },
    {
      "epoch": 1.3845169545745362,
      "grad_norm": 2.1767085140038445,
      "learning_rate": 1.9389502036174893e-05,
      "loss": 0.769,
      "step": 2164
    },
    {
      "epoch": 1.3851567498400512,
      "grad_norm": 1.9695465597707458,
      "learning_rate": 1.9388788903655928e-05,
      "loss": 0.7939,
      "step": 2165
    },
    {
      "epoch": 1.3857965451055663,
      "grad_norm": 2.275865449490676,
      "learning_rate": 1.938807536799907e-05,
      "loss": 0.7323,
      "step": 2166
    },
    {
      "epoch": 1.3864363403710813,
      "grad_norm": 2.134322274705814,
      "learning_rate": 1.938736142923496e-05,
      "loss": 0.7758,
      "step": 2167
    },
    {
      "epoch": 1.3870761356365962,
      "grad_norm": 2.4412001181182617,
      "learning_rate": 1.9386647087394244e-05,
      "loss": 0.7822,
      "step": 2168
    },
    {
      "epoch": 1.3877159309021114,
      "grad_norm": 2.2479444567577787,
      "learning_rate": 1.9385932342507602e-05,
      "loss": 0.7772,
      "step": 2169
    },
    {
      "epoch": 1.3883557261676263,
      "grad_norm": 2.4335578692387285,
      "learning_rate": 1.938521719460572e-05,
      "loss": 0.7883,
      "step": 2170
    },
    {
      "epoch": 1.3889955214331415,
      "grad_norm": 2.0225885646561745,
      "learning_rate": 1.9384501643719304e-05,
      "loss": 0.7639,
      "step": 2171
    },
    {
      "epoch": 1.3896353166986564,
      "grad_norm": 2.083930859429548,
      "learning_rate": 1.9383785689879087e-05,
      "loss": 0.7705,
      "step": 2172
    },
    {
      "epoch": 1.3902751119641714,
      "grad_norm": 2.242368955810907,
      "learning_rate": 1.93830693331158e-05,
      "loss": 0.7387,
      "step": 2173
    },
    {
      "epoch": 1.3909149072296865,
      "grad_norm": 2.182321714428531,
      "learning_rate": 1.938235257346021e-05,
      "loss": 0.7966,
      "step": 2174
    },
    {
      "epoch": 1.3915547024952015,
      "grad_norm": 2.1258268646034812,
      "learning_rate": 1.938163541094309e-05,
      "loss": 0.7227,
      "step": 2175
    },
    {
      "epoch": 1.3921944977607166,
      "grad_norm": 2.136403574583713,
      "learning_rate": 1.9380917845595233e-05,
      "loss": 0.6821,
      "step": 2176
    },
    {
      "epoch": 1.3928342930262316,
      "grad_norm": 2.2387055713869985,
      "learning_rate": 1.9380199877447453e-05,
      "loss": 0.7433,
      "step": 2177
    },
    {
      "epoch": 1.3934740882917467,
      "grad_norm": 2.711506185734167,
      "learning_rate": 1.9379481506530572e-05,
      "loss": 0.7611,
      "step": 2178
    },
    {
      "epoch": 1.3941138835572617,
      "grad_norm": 2.4049853231695333,
      "learning_rate": 1.9378762732875445e-05,
      "loss": 0.7248,
      "step": 2179
    },
    {
      "epoch": 1.3947536788227768,
      "grad_norm": 2.464590211334854,
      "learning_rate": 1.937804355651293e-05,
      "loss": 0.7612,
      "step": 2180
    },
    {
      "epoch": 1.3953934740882916,
      "grad_norm": 2.3831963513264007,
      "learning_rate": 1.93773239774739e-05,
      "loss": 0.7449,
      "step": 2181
    },
    {
      "epoch": 1.396033269353807,
      "grad_norm": 2.6447722253867645,
      "learning_rate": 1.9376603995789268e-05,
      "loss": 0.776,
      "step": 2182
    },
    {
      "epoch": 1.3966730646193217,
      "grad_norm": 2.6005890836840653,
      "learning_rate": 1.9375883611489933e-05,
      "loss": 0.7682,
      "step": 2183
    },
    {
      "epoch": 1.3973128598848368,
      "grad_norm": 2.7833116655535575,
      "learning_rate": 1.937516282460684e-05,
      "loss": 0.8004,
      "step": 2184
    },
    {
      "epoch": 1.3979526551503518,
      "grad_norm": 2.4493422286503916,
      "learning_rate": 1.937444163517093e-05,
      "loss": 0.7907,
      "step": 2185
    },
    {
      "epoch": 1.3985924504158669,
      "grad_norm": 2.541719190535145,
      "learning_rate": 1.937372004321317e-05,
      "loss": 0.7435,
      "step": 2186
    },
    {
      "epoch": 1.399232245681382,
      "grad_norm": 2.7606649351921697,
      "learning_rate": 1.9372998048764548e-05,
      "loss": 0.7424,
      "step": 2187
    },
    {
      "epoch": 1.399872040946897,
      "grad_norm": 2.756056874804883,
      "learning_rate": 1.937227565185606e-05,
      "loss": 0.7392,
      "step": 2188
    },
    {
      "epoch": 1.400511836212412,
      "grad_norm": 2.668518110472334,
      "learning_rate": 1.937155285251873e-05,
      "loss": 0.8423,
      "step": 2189
    },
    {
      "epoch": 1.401151631477927,
      "grad_norm": 2.3528654753728824,
      "learning_rate": 1.9370829650783593e-05,
      "loss": 0.7789,
      "step": 2190
    },
    {
      "epoch": 1.4017914267434421,
      "grad_norm": 2.2056800833140717,
      "learning_rate": 1.93701060466817e-05,
      "loss": 0.7268,
      "step": 2191
    },
    {
      "epoch": 1.4024312220089572,
      "grad_norm": 2.5958823857245137,
      "learning_rate": 1.936938204024412e-05,
      "loss": 0.7744,
      "step": 2192
    },
    {
      "epoch": 1.4030710172744723,
      "grad_norm": 2.2486454854611715,
      "learning_rate": 1.9368657631501943e-05,
      "loss": 0.8056,
      "step": 2193
    },
    {
      "epoch": 1.403710812539987,
      "grad_norm": 2.5005105580665674,
      "learning_rate": 1.936793282048627e-05,
      "loss": 0.7478,
      "step": 2194
    },
    {
      "epoch": 1.4043506078055024,
      "grad_norm": 2.1524615114910137,
      "learning_rate": 1.936720760722823e-05,
      "loss": 0.7683,
      "step": 2195
    },
    {
      "epoch": 1.4049904030710172,
      "grad_norm": 2.1434711811496445,
      "learning_rate": 1.9366481991758955e-05,
      "loss": 0.8096,
      "step": 2196
    },
    {
      "epoch": 1.4056301983365322,
      "grad_norm": 2.2688751147680155,
      "learning_rate": 1.9365755974109604e-05,
      "loss": 0.7978,
      "step": 2197
    },
    {
      "epoch": 1.4062699936020473,
      "grad_norm": 2.2118902065534596,
      "learning_rate": 1.9365029554311355e-05,
      "loss": 0.7272,
      "step": 2198
    },
    {
      "epoch": 1.4069097888675623,
      "grad_norm": 2.1698446568802616,
      "learning_rate": 1.9364302732395395e-05,
      "loss": 0.7753,
      "step": 2199
    },
    {
      "epoch": 1.4075495841330774,
      "grad_norm": 2.6283503703443416,
      "learning_rate": 1.9363575508392933e-05,
      "loss": 0.7908,
      "step": 2200
    },
    {
      "epoch": 1.4081893793985925,
      "grad_norm": 2.3919038815097675,
      "learning_rate": 1.9362847882335196e-05,
      "loss": 0.787,
      "step": 2201
    },
    {
      "epoch": 1.4088291746641075,
      "grad_norm": 2.3902956949989322,
      "learning_rate": 1.9362119854253426e-05,
      "loss": 0.7649,
      "step": 2202
    },
    {
      "epoch": 1.4094689699296226,
      "grad_norm": 2.680328970790249,
      "learning_rate": 1.9361391424178883e-05,
      "loss": 0.7491,
      "step": 2203
    },
    {
      "epoch": 1.4101087651951376,
      "grad_norm": 2.4484887307514103,
      "learning_rate": 1.9360662592142846e-05,
      "loss": 0.8367,
      "step": 2204
    },
    {
      "epoch": 1.4107485604606527,
      "grad_norm": 2.6723994701864635,
      "learning_rate": 1.9359933358176607e-05,
      "loss": 0.7889,
      "step": 2205
    },
    {
      "epoch": 1.4113883557261677,
      "grad_norm": 2.4583948773951065,
      "learning_rate": 1.935920372231148e-05,
      "loss": 0.7766,
      "step": 2206
    },
    {
      "epoch": 1.4120281509916826,
      "grad_norm": 2.2868915869002557,
      "learning_rate": 1.9358473684578795e-05,
      "loss": 0.7342,
      "step": 2207
    },
    {
      "epoch": 1.4126679462571978,
      "grad_norm": 2.121200842722243,
      "learning_rate": 1.9357743245009898e-05,
      "loss": 0.794,
      "step": 2208
    },
    {
      "epoch": 1.4133077415227127,
      "grad_norm": 2.687391149146247,
      "learning_rate": 1.9357012403636147e-05,
      "loss": 0.7916,
      "step": 2209
    },
    {
      "epoch": 1.4139475367882277,
      "grad_norm": 2.693779323155532,
      "learning_rate": 1.9356281160488936e-05,
      "loss": 0.7367,
      "step": 2210
    },
    {
      "epoch": 1.4145873320537428,
      "grad_norm": 2.1784870447143265,
      "learning_rate": 1.935554951559965e-05,
      "loss": 0.7224,
      "step": 2211
    },
    {
      "epoch": 1.4152271273192578,
      "grad_norm": 2.2284351404554115,
      "learning_rate": 1.935481746899971e-05,
      "loss": 0.7478,
      "step": 2212
    },
    {
      "epoch": 1.4158669225847729,
      "grad_norm": 2.2588188577047204,
      "learning_rate": 1.935408502072055e-05,
      "loss": 0.722,
      "step": 2213
    },
    {
      "epoch": 1.416506717850288,
      "grad_norm": 2.1789935329488124,
      "learning_rate": 1.935335217079362e-05,
      "loss": 0.7988,
      "step": 2214
    },
    {
      "epoch": 1.417146513115803,
      "grad_norm": 2.277261060995996,
      "learning_rate": 1.9352618919250384e-05,
      "loss": 0.7496,
      "step": 2215
    },
    {
      "epoch": 1.417786308381318,
      "grad_norm": 2.145653488337721,
      "learning_rate": 1.9351885266122332e-05,
      "loss": 0.7539,
      "step": 2216
    },
    {
      "epoch": 1.418426103646833,
      "grad_norm": 2.1559035243702436,
      "learning_rate": 1.935115121144096e-05,
      "loss": 0.788,
      "step": 2217
    },
    {
      "epoch": 1.419065898912348,
      "grad_norm": 2.644345880402741,
      "learning_rate": 1.9350416755237792e-05,
      "loss": 0.8057,
      "step": 2218
    },
    {
      "epoch": 1.4197056941778632,
      "grad_norm": 2.4779013350418033,
      "learning_rate": 1.9349681897544358e-05,
      "loss": 0.7758,
      "step": 2219
    },
    {
      "epoch": 1.420345489443378,
      "grad_norm": 2.2450798266260286,
      "learning_rate": 1.9348946638392218e-05,
      "loss": 0.7765,
      "step": 2220
    },
    {
      "epoch": 1.4209852847088933,
      "grad_norm": 2.10135906316504,
      "learning_rate": 1.9348210977812936e-05,
      "loss": 0.7409,
      "step": 2221
    },
    {
      "epoch": 1.4216250799744081,
      "grad_norm": 2.8083522198365056,
      "learning_rate": 1.9347474915838107e-05,
      "loss": 0.7871,
      "step": 2222
    },
    {
      "epoch": 1.4222648752399232,
      "grad_norm": 2.6165382188331034,
      "learning_rate": 1.9346738452499332e-05,
      "loss": 0.7334,
      "step": 2223
    },
    {
      "epoch": 1.4229046705054382,
      "grad_norm": 2.4497737663876222,
      "learning_rate": 1.9346001587828235e-05,
      "loss": 0.6964,
      "step": 2224
    },
    {
      "epoch": 1.4235444657709533,
      "grad_norm": 2.7790391415691804,
      "learning_rate": 1.9345264321856454e-05,
      "loss": 0.8147,
      "step": 2225
    },
    {
      "epoch": 1.4241842610364683,
      "grad_norm": 2.562971867465885,
      "learning_rate": 1.9344526654615646e-05,
      "loss": 0.7692,
      "step": 2226
    },
    {
      "epoch": 1.4248240563019834,
      "grad_norm": 2.20891572529041,
      "learning_rate": 1.934378858613749e-05,
      "loss": 0.7304,
      "step": 2227
    },
    {
      "epoch": 1.4254638515674984,
      "grad_norm": 2.7921138319972387,
      "learning_rate": 1.934305011645367e-05,
      "loss": 0.7982,
      "step": 2228
    },
    {
      "epoch": 1.4261036468330135,
      "grad_norm": 2.096128234732554,
      "learning_rate": 1.9342311245595898e-05,
      "loss": 0.7585,
      "step": 2229
    },
    {
      "epoch": 1.4267434420985285,
      "grad_norm": 2.017881354904385,
      "learning_rate": 1.93415719735959e-05,
      "loss": 0.771,
      "step": 2230
    },
    {
      "epoch": 1.4273832373640434,
      "grad_norm": 2.25964432653882,
      "learning_rate": 1.9340832300485423e-05,
      "loss": 0.782,
      "step": 2231
    },
    {
      "epoch": 1.4280230326295587,
      "grad_norm": 2.096855726908655,
      "learning_rate": 1.934009222629622e-05,
      "loss": 0.7936,
      "step": 2232
    },
    {
      "epoch": 1.4286628278950735,
      "grad_norm": 2.5277247841168795,
      "learning_rate": 1.9339351751060068e-05,
      "loss": 0.7914,
      "step": 2233
    },
    {
      "epoch": 1.4293026231605885,
      "grad_norm": 2.489224916177137,
      "learning_rate": 1.933861087480877e-05,
      "loss": 0.7644,
      "step": 2234
    },
    {
      "epoch": 1.4299424184261036,
      "grad_norm": 2.125784567475569,
      "learning_rate": 1.9337869597574133e-05,
      "loss": 0.7368,
      "step": 2235
    },
    {
      "epoch": 1.4305822136916186,
      "grad_norm": 1.9541736020724572,
      "learning_rate": 1.9337127919387984e-05,
      "loss": 0.7293,
      "step": 2236
    },
    {
      "epoch": 1.4312220089571337,
      "grad_norm": 2.2395997599233164,
      "learning_rate": 1.933638584028217e-05,
      "loss": 0.7064,
      "step": 2237
    },
    {
      "epoch": 1.4318618042226487,
      "grad_norm": 2.4290865060723177,
      "learning_rate": 1.933564336028856e-05,
      "loss": 0.8008,
      "step": 2238
    },
    {
      "epoch": 1.4325015994881638,
      "grad_norm": 2.1892925631345994,
      "learning_rate": 1.9334900479439032e-05,
      "loss": 0.7546,
      "step": 2239
    },
    {
      "epoch": 1.4331413947536789,
      "grad_norm": 2.14755232331386,
      "learning_rate": 1.933415719776548e-05,
      "loss": 0.7924,
      "step": 2240
    },
    {
      "epoch": 1.433781190019194,
      "grad_norm": 2.277323458550757,
      "learning_rate": 1.933341351529982e-05,
      "loss": 0.7579,
      "step": 2241
    },
    {
      "epoch": 1.434420985284709,
      "grad_norm": 2.216314474364407,
      "learning_rate": 1.933266943207399e-05,
      "loss": 0.7295,
      "step": 2242
    },
    {
      "epoch": 1.435060780550224,
      "grad_norm": 2.1195835240712237,
      "learning_rate": 1.9331924948119935e-05,
      "loss": 0.7336,
      "step": 2243
    },
    {
      "epoch": 1.4357005758157388,
      "grad_norm": 2.30663097129268,
      "learning_rate": 1.9331180063469623e-05,
      "loss": 0.7636,
      "step": 2244
    },
    {
      "epoch": 1.4363403710812541,
      "grad_norm": 1.9522520514780792,
      "learning_rate": 1.9330434778155037e-05,
      "loss": 0.7588,
      "step": 2245
    },
    {
      "epoch": 1.436980166346769,
      "grad_norm": 2.0641068574115407,
      "learning_rate": 1.932968909220818e-05,
      "loss": 0.7898,
      "step": 2246
    },
    {
      "epoch": 1.437619961612284,
      "grad_norm": 2.3980340530885558,
      "learning_rate": 1.9328943005661072e-05,
      "loss": 0.7569,
      "step": 2247
    },
    {
      "epoch": 1.438259756877799,
      "grad_norm": 2.0567145419330046,
      "learning_rate": 1.9328196518545742e-05,
      "loss": 0.7123,
      "step": 2248
    },
    {
      "epoch": 1.438899552143314,
      "grad_norm": 2.3847020005610493,
      "learning_rate": 1.9327449630894254e-05,
      "loss": 0.7724,
      "step": 2249
    },
    {
      "epoch": 1.4395393474088292,
      "grad_norm": 2.046533518550811,
      "learning_rate": 1.9326702342738663e-05,
      "loss": 0.7024,
      "step": 2250
    },
    {
      "epoch": 1.4401791426743442,
      "grad_norm": 2.1356600975208475,
      "learning_rate": 1.9325954654111068e-05,
      "loss": 0.7825,
      "step": 2251
    },
    {
      "epoch": 1.4408189379398593,
      "grad_norm": 2.023211709840623,
      "learning_rate": 1.932520656504357e-05,
      "loss": 0.7692,
      "step": 2252
    },
    {
      "epoch": 1.4414587332053743,
      "grad_norm": 2.257431738533569,
      "learning_rate": 1.9324458075568288e-05,
      "loss": 0.7836,
      "step": 2253
    },
    {
      "epoch": 1.4420985284708894,
      "grad_norm": 2.1155075745690897,
      "learning_rate": 1.9323709185717365e-05,
      "loss": 0.7198,
      "step": 2254
    },
    {
      "epoch": 1.4427383237364044,
      "grad_norm": 2.1783277879517353,
      "learning_rate": 1.9322959895522957e-05,
      "loss": 0.7746,
      "step": 2255
    },
    {
      "epoch": 1.4433781190019195,
      "grad_norm": 2.4156897098665526,
      "learning_rate": 1.9322210205017234e-05,
      "loss": 0.7865,
      "step": 2256
    },
    {
      "epoch": 1.4440179142674343,
      "grad_norm": 2.09776349363517,
      "learning_rate": 1.9321460114232383e-05,
      "loss": 0.754,
      "step": 2257
    },
    {
      "epoch": 1.4446577095329496,
      "grad_norm": 2.3472206801717683,
      "learning_rate": 1.9320709623200623e-05,
      "loss": 0.8165,
      "step": 2258
    },
    {
      "epoch": 1.4452975047984644,
      "grad_norm": 2.2530333594528265,
      "learning_rate": 1.931995873195417e-05,
      "loss": 0.7574,
      "step": 2259
    },
    {
      "epoch": 1.4459373000639795,
      "grad_norm": 2.2161207876276348,
      "learning_rate": 1.9319207440525267e-05,
      "loss": 0.6998,
      "step": 2260
    },
    {
      "epoch": 1.4465770953294945,
      "grad_norm": 2.0549608087163564,
      "learning_rate": 1.9318455748946175e-05,
      "loss": 0.6726,
      "step": 2261
    },
    {
      "epoch": 1.4472168905950096,
      "grad_norm": 2.322626920128153,
      "learning_rate": 1.931770365724917e-05,
      "loss": 0.8239,
      "step": 2262
    },
    {
      "epoch": 1.4478566858605246,
      "grad_norm": 2.6226765172794297,
      "learning_rate": 1.9316951165466542e-05,
      "loss": 0.7873,
      "step": 2263
    },
    {
      "epoch": 1.4484964811260397,
      "grad_norm": 2.2122697199880466,
      "learning_rate": 1.9316198273630603e-05,
      "loss": 0.7373,
      "step": 2264
    },
    {
      "epoch": 1.4491362763915547,
      "grad_norm": 2.5369621226257992,
      "learning_rate": 1.931544498177369e-05,
      "loss": 0.7317,
      "step": 2265
    },
    {
      "epoch": 1.4497760716570698,
      "grad_norm": 2.078501123754023,
      "learning_rate": 1.9314691289928135e-05,
      "loss": 0.6926,
      "step": 2266
    },
    {
      "epoch": 1.4504158669225848,
      "grad_norm": 2.214060480791828,
      "learning_rate": 1.9313937198126305e-05,
      "loss": 0.7561,
      "step": 2267
    },
    {
      "epoch": 1.4510556621880997,
      "grad_norm": 2.5047731901633608,
      "learning_rate": 1.9313182706400583e-05,
      "loss": 0.7473,
      "step": 2268
    },
    {
      "epoch": 1.451695457453615,
      "grad_norm": 2.25520910914562,
      "learning_rate": 1.9312427814783363e-05,
      "loss": 0.7116,
      "step": 2269
    },
    {
      "epoch": 1.4523352527191298,
      "grad_norm": 2.016585583961043,
      "learning_rate": 1.931167252330705e-05,
      "loss": 0.7387,
      "step": 2270
    },
    {
      "epoch": 1.452975047984645,
      "grad_norm": 2.3060003435378333,
      "learning_rate": 1.931091683200409e-05,
      "loss": 0.7664,
      "step": 2271
    },
    {
      "epoch": 1.4536148432501599,
      "grad_norm": 2.12358189727134,
      "learning_rate": 1.9310160740906926e-05,
      "loss": 0.7652,
      "step": 2272
    },
    {
      "epoch": 1.454254638515675,
      "grad_norm": 2.157578726946466,
      "learning_rate": 1.930940425004802e-05,
      "loss": 0.7411,
      "step": 2273
    },
    {
      "epoch": 1.45489443378119,
      "grad_norm": 2.1189529014228197,
      "learning_rate": 1.9308647359459853e-05,
      "loss": 0.7241,
      "step": 2274
    },
    {
      "epoch": 1.455534229046705,
      "grad_norm": 2.046526527323013,
      "learning_rate": 1.930789006917493e-05,
      "loss": 0.7608,
      "step": 2275
    },
    {
      "epoch": 1.45617402431222,
      "grad_norm": 2.682637370841391,
      "learning_rate": 1.930713237922576e-05,
      "loss": 0.7197,
      "step": 2276
    },
    {
      "epoch": 1.4568138195777351,
      "grad_norm": 2.0085140487643236,
      "learning_rate": 1.9306374289644886e-05,
      "loss": 0.7319,
      "step": 2277
    },
    {
      "epoch": 1.4574536148432502,
      "grad_norm": 2.2349138717740065,
      "learning_rate": 1.9305615800464853e-05,
      "loss": 0.7692,
      "step": 2278
    },
    {
      "epoch": 1.4580934101087653,
      "grad_norm": 2.3412655664770154,
      "learning_rate": 1.930485691171823e-05,
      "loss": 0.7779,
      "step": 2279
    },
    {
      "epoch": 1.4587332053742803,
      "grad_norm": 2.323895891277818,
      "learning_rate": 1.9304097623437607e-05,
      "loss": 0.7548,
      "step": 2280
    },
    {
      "epoch": 1.4593730006397951,
      "grad_norm": 2.48568573853444,
      "learning_rate": 1.930333793565558e-05,
      "loss": 0.7999,
      "step": 2281
    },
    {
      "epoch": 1.4600127959053104,
      "grad_norm": 2.192979904357328,
      "learning_rate": 1.9302577848404772e-05,
      "loss": 0.7624,
      "step": 2282
    },
    {
      "epoch": 1.4606525911708252,
      "grad_norm": 2.0191367658025694,
      "learning_rate": 1.930181736171782e-05,
      "loss": 0.6892,
      "step": 2283
    },
    {
      "epoch": 1.4612923864363403,
      "grad_norm": 2.252551539671072,
      "learning_rate": 1.9301056475627374e-05,
      "loss": 0.7931,
      "step": 2284
    },
    {
      "epoch": 1.4619321817018553,
      "grad_norm": 2.203908964250321,
      "learning_rate": 1.9300295190166115e-05,
      "loss": 0.7671,
      "step": 2285
    },
    {
      "epoch": 1.4625719769673704,
      "grad_norm": 2.4828362458139828,
      "learning_rate": 1.9299533505366718e-05,
      "loss": 0.7976,
      "step": 2286
    },
    {
      "epoch": 1.4632117722328855,
      "grad_norm": 2.1772148280530024,
      "learning_rate": 1.9298771421261895e-05,
      "loss": 0.762,
      "step": 2287
    },
    {
      "epoch": 1.4638515674984005,
      "grad_norm": 2.131707814903448,
      "learning_rate": 1.9298008937884372e-05,
      "loss": 0.7596,
      "step": 2288
    },
    {
      "epoch": 1.4644913627639156,
      "grad_norm": 2.133608276909299,
      "learning_rate": 1.9297246055266882e-05,
      "loss": 0.6962,
      "step": 2289
    },
    {
      "epoch": 1.4651311580294306,
      "grad_norm": 2.169487092515522,
      "learning_rate": 1.9296482773442188e-05,
      "loss": 0.7371,
      "step": 2290
    },
    {
      "epoch": 1.4657709532949457,
      "grad_norm": 2.2017203585522114,
      "learning_rate": 1.929571909244306e-05,
      "loss": 0.7322,
      "step": 2291
    },
    {
      "epoch": 1.4664107485604607,
      "grad_norm": 2.483409776016947,
      "learning_rate": 1.9294955012302287e-05,
      "loss": 0.7234,
      "step": 2292
    },
    {
      "epoch": 1.4670505438259758,
      "grad_norm": 2.633866724807756,
      "learning_rate": 1.9294190533052682e-05,
      "loss": 0.8006,
      "step": 2293
    },
    {
      "epoch": 1.4676903390914906,
      "grad_norm": 2.6894838020278398,
      "learning_rate": 1.9293425654727072e-05,
      "loss": 0.7679,
      "step": 2294
    },
    {
      "epoch": 1.4683301343570059,
      "grad_norm": 2.252020442176766,
      "learning_rate": 1.929266037735829e-05,
      "loss": 0.7619,
      "step": 2295
    },
    {
      "epoch": 1.4689699296225207,
      "grad_norm": 2.351614896379744,
      "learning_rate": 1.9291894700979206e-05,
      "loss": 0.7545,
      "step": 2296
    },
    {
      "epoch": 1.4696097248880358,
      "grad_norm": 2.9035784990528133,
      "learning_rate": 1.929112862562269e-05,
      "loss": 0.7067,
      "step": 2297
    },
    {
      "epoch": 1.4702495201535508,
      "grad_norm": 2.628360262649446,
      "learning_rate": 1.9290362151321643e-05,
      "loss": 0.7922,
      "step": 2298
    },
    {
      "epoch": 1.4708893154190659,
      "grad_norm": 2.2695464285135385,
      "learning_rate": 1.928959527810897e-05,
      "loss": 0.7469,
      "step": 2299
    },
    {
      "epoch": 1.471529110684581,
      "grad_norm": 2.1471839757597335,
      "learning_rate": 1.9288828006017594e-05,
      "loss": 0.7207,
      "step": 2300
    },
    {
      "epoch": 1.472168905950096,
      "grad_norm": 2.2426343447629615,
      "learning_rate": 1.9288060335080474e-05,
      "loss": 0.8172,
      "step": 2301
    },
    {
      "epoch": 1.472808701215611,
      "grad_norm": 2.0405398998126123,
      "learning_rate": 1.9287292265330562e-05,
      "loss": 0.7359,
      "step": 2302
    },
    {
      "epoch": 1.473448496481126,
      "grad_norm": 2.2870030909266044,
      "learning_rate": 1.9286523796800846e-05,
      "loss": 0.7743,
      "step": 2303
    },
    {
      "epoch": 1.4740882917466411,
      "grad_norm": 2.3244642837237923,
      "learning_rate": 1.928575492952431e-05,
      "loss": 0.7249,
      "step": 2304
    },
    {
      "epoch": 1.4747280870121562,
      "grad_norm": 2.0315007224697967,
      "learning_rate": 1.928498566353398e-05,
      "loss": 0.7389,
      "step": 2305
    },
    {
      "epoch": 1.4753678822776712,
      "grad_norm": 2.303486272291626,
      "learning_rate": 1.928421599886288e-05,
      "loss": 0.8234,
      "step": 2306
    },
    {
      "epoch": 1.476007677543186,
      "grad_norm": 2.4523272989291214,
      "learning_rate": 1.9283445935544064e-05,
      "loss": 0.7254,
      "step": 2307
    },
    {
      "epoch": 1.4766474728087013,
      "grad_norm": 2.326164549636369,
      "learning_rate": 1.9282675473610594e-05,
      "loss": 0.7415,
      "step": 2308
    },
    {
      "epoch": 1.4772872680742162,
      "grad_norm": 2.0215740613282542,
      "learning_rate": 1.9281904613095546e-05,
      "loss": 0.753,
      "step": 2309
    },
    {
      "epoch": 1.4779270633397312,
      "grad_norm": 2.1322255860326185,
      "learning_rate": 1.9281133354032028e-05,
      "loss": 0.7249,
      "step": 2310
    },
    {
      "epoch": 1.4785668586052463,
      "grad_norm": 2.3334088554282237,
      "learning_rate": 1.9280361696453155e-05,
      "loss": 0.7283,
      "step": 2311
    },
    {
      "epoch": 1.4792066538707613,
      "grad_norm": 2.0845335731872083,
      "learning_rate": 1.9279589640392057e-05,
      "loss": 0.7453,
      "step": 2312
    },
    {
      "epoch": 1.4798464491362764,
      "grad_norm": 2.1434045449401884,
      "learning_rate": 1.9278817185881887e-05,
      "loss": 0.712,
      "step": 2313
    },
    {
      "epoch": 1.4804862444017914,
      "grad_norm": 2.1155488885805043,
      "learning_rate": 1.9278044332955815e-05,
      "loss": 0.7611,
      "step": 2314
    },
    {
      "epoch": 1.4811260396673065,
      "grad_norm": 2.299721900318412,
      "learning_rate": 1.9277271081647024e-05,
      "loss": 0.758,
      "step": 2315
    },
    {
      "epoch": 1.4817658349328215,
      "grad_norm": 2.1949935129182925,
      "learning_rate": 1.9276497431988713e-05,
      "loss": 0.7477,
      "step": 2316
    },
    {
      "epoch": 1.4824056301983366,
      "grad_norm": 2.090185369931863,
      "learning_rate": 1.9275723384014108e-05,
      "loss": 0.6983,
      "step": 2317
    },
    {
      "epoch": 1.4830454254638514,
      "grad_norm": 2.055891363251846,
      "learning_rate": 1.9274948937756436e-05,
      "loss": 0.7069,
      "step": 2318
    },
    {
      "epoch": 1.4836852207293667,
      "grad_norm": 2.2316089038513014,
      "learning_rate": 1.9274174093248957e-05,
      "loss": 0.73,
      "step": 2319
    },
    {
      "epoch": 1.4843250159948815,
      "grad_norm": 2.320616290725856,
      "learning_rate": 1.927339885052494e-05,
      "loss": 0.7383,
      "step": 2320
    },
    {
      "epoch": 1.4849648112603968,
      "grad_norm": 2.477515088469996,
      "learning_rate": 1.9272623209617675e-05,
      "loss": 0.8385,
      "step": 2321
    },
    {
      "epoch": 1.4856046065259116,
      "grad_norm": 1.830987985260361,
      "learning_rate": 1.927184717056046e-05,
      "loss": 0.7497,
      "step": 2322
    },
    {
      "epoch": 1.4862444017914267,
      "grad_norm": 2.131932120002586,
      "learning_rate": 1.9271070733386625e-05,
      "loss": 0.7268,
      "step": 2323
    },
    {
      "epoch": 1.4868841970569417,
      "grad_norm": 2.250264251822006,
      "learning_rate": 1.9270293898129503e-05,
      "loss": 0.7067,
      "step": 2324
    },
    {
      "epoch": 1.4875239923224568,
      "grad_norm": 2.092260204877089,
      "learning_rate": 1.926951666482245e-05,
      "loss": 0.7377,
      "step": 2325
    },
    {
      "epoch": 1.4881637875879719,
      "grad_norm": 2.260022622413138,
      "learning_rate": 1.9268739033498846e-05,
      "loss": 0.6933,
      "step": 2326
    },
    {
      "epoch": 1.488803582853487,
      "grad_norm": 2.1435314246676405,
      "learning_rate": 1.9267961004192067e-05,
      "loss": 0.6663,
      "step": 2327
    },
    {
      "epoch": 1.489443378119002,
      "grad_norm": 2.0900980423988393,
      "learning_rate": 1.9267182576935535e-05,
      "loss": 0.6938,
      "step": 2328
    },
    {
      "epoch": 1.490083173384517,
      "grad_norm": 2.265369893410068,
      "learning_rate": 1.9266403751762666e-05,
      "loss": 0.7434,
      "step": 2329
    },
    {
      "epoch": 1.490722968650032,
      "grad_norm": 2.0582314129849144,
      "learning_rate": 1.9265624528706906e-05,
      "loss": 0.7804,
      "step": 2330
    },
    {
      "epoch": 1.491362763915547,
      "grad_norm": 2.064287796244175,
      "learning_rate": 1.926484490780171e-05,
      "loss": 0.7469,
      "step": 2331
    },
    {
      "epoch": 1.4920025591810622,
      "grad_norm": 1.9811758158407478,
      "learning_rate": 1.9264064889080553e-05,
      "loss": 0.7085,
      "step": 2332
    },
    {
      "epoch": 1.492642354446577,
      "grad_norm": 2.080403875780074,
      "learning_rate": 1.9263284472576932e-05,
      "loss": 0.764,
      "step": 2333
    },
    {
      "epoch": 1.493282149712092,
      "grad_norm": 2.574272383672492,
      "learning_rate": 1.9262503658324353e-05,
      "loss": 0.7719,
      "step": 2334
    },
    {
      "epoch": 1.493921944977607,
      "grad_norm": 2.0853610870177186,
      "learning_rate": 1.926172244635634e-05,
      "loss": 0.7578,
      "step": 2335
    },
    {
      "epoch": 1.4945617402431222,
      "grad_norm": 2.0627619948223765,
      "learning_rate": 1.9260940836706443e-05,
      "loss": 0.702,
      "step": 2336
    },
    {
      "epoch": 1.4952015355086372,
      "grad_norm": 2.249546446406735,
      "learning_rate": 1.926015882940822e-05,
      "loss": 0.7448,
      "step": 2337
    },
    {
      "epoch": 1.4958413307741523,
      "grad_norm": 2.1271876500238442,
      "learning_rate": 1.9259376424495252e-05,
      "loss": 0.7994,
      "step": 2338
    },
    {
      "epoch": 1.4964811260396673,
      "grad_norm": 2.234346159825417,
      "learning_rate": 1.9258593622001128e-05,
      "loss": 0.732,
      "step": 2339
    },
    {
      "epoch": 1.4971209213051824,
      "grad_norm": 1.9049105794565837,
      "learning_rate": 1.9257810421959466e-05,
      "loss": 0.7383,
      "step": 2340
    },
    {
      "epoch": 1.4977607165706974,
      "grad_norm": 2.221256763629398,
      "learning_rate": 1.925702682440389e-05,
      "loss": 0.7429,
      "step": 2341
    },
    {
      "epoch": 1.4984005118362125,
      "grad_norm": 1.9959962592939517,
      "learning_rate": 1.9256242829368053e-05,
      "loss": 0.8503,
      "step": 2342
    },
    {
      "epoch": 1.4990403071017275,
      "grad_norm": 2.33442844207208,
      "learning_rate": 1.925545843688561e-05,
      "loss": 0.6857,
      "step": 2343
    },
    {
      "epoch": 1.4996801023672424,
      "grad_norm": 2.1842683741375173,
      "learning_rate": 1.925467364699025e-05,
      "loss": 0.7157,
      "step": 2344
    },
    {
      "epoch": 1.5003198976327576,
      "grad_norm": 2.279194261023527,
      "learning_rate": 1.9253888459715663e-05,
      "loss": 0.7814,
      "step": 2345
    },
    {
      "epoch": 1.5009596928982725,
      "grad_norm": 2.1648212277873182,
      "learning_rate": 1.9253102875095568e-05,
      "loss": 0.7771,
      "step": 2346
    },
    {
      "epoch": 1.5015994881637877,
      "grad_norm": 2.0313766583573276,
      "learning_rate": 1.92523168931637e-05,
      "loss": 0.6998,
      "step": 2347
    },
    {
      "epoch": 1.5022392834293026,
      "grad_norm": 2.0727893709211527,
      "learning_rate": 1.9251530513953794e-05,
      "loss": 0.7419,
      "step": 2348
    },
    {
      "epoch": 1.5028790786948176,
      "grad_norm": 1.993907248867296,
      "learning_rate": 1.9250743737499633e-05,
      "loss": 0.6873,
      "step": 2349
    },
    {
      "epoch": 1.5035188739603327,
      "grad_norm": 2.334420844372318,
      "learning_rate": 1.924995656383499e-05,
      "loss": 0.7894,
      "step": 2350
    },
    {
      "epoch": 1.5041586692258477,
      "grad_norm": 2.1943030647141595,
      "learning_rate": 1.9249168992993666e-05,
      "loss": 0.719,
      "step": 2351
    },
    {
      "epoch": 1.5047984644913628,
      "grad_norm": 2.3053231449423586,
      "learning_rate": 1.9248381025009476e-05,
      "loss": 0.697,
      "step": 2352
    },
    {
      "epoch": 1.5054382597568778,
      "grad_norm": 2.0397941308157925,
      "learning_rate": 1.924759265991626e-05,
      "loss": 0.7709,
      "step": 2353
    },
    {
      "epoch": 1.506078055022393,
      "grad_norm": 2.4011188919774167,
      "learning_rate": 1.9246803897747864e-05,
      "loss": 0.7847,
      "step": 2354
    },
    {
      "epoch": 1.5067178502879077,
      "grad_norm": 2.3021385450325833,
      "learning_rate": 1.9246014738538157e-05,
      "loss": 0.7298,
      "step": 2355
    },
    {
      "epoch": 1.507357645553423,
      "grad_norm": 2.5243197091870107,
      "learning_rate": 1.9245225182321025e-05,
      "loss": 0.7028,
      "step": 2356
    },
    {
      "epoch": 1.5079974408189378,
      "grad_norm": 2.6357139343179594,
      "learning_rate": 1.9244435229130368e-05,
      "loss": 0.7621,
      "step": 2357
    },
    {
      "epoch": 1.508637236084453,
      "grad_norm": 2.240635722010028,
      "learning_rate": 1.924364487900011e-05,
      "loss": 0.6612,
      "step": 2358
    },
    {
      "epoch": 1.509277031349968,
      "grad_norm": 2.364321374826784,
      "learning_rate": 1.9242854131964178e-05,
      "loss": 0.7573,
      "step": 2359
    },
    {
      "epoch": 1.5099168266154832,
      "grad_norm": 2.828219366039597,
      "learning_rate": 1.924206298805654e-05,
      "loss": 0.7369,
      "step": 2360
    },
    {
      "epoch": 1.510556621880998,
      "grad_norm": 2.227592681566136,
      "learning_rate": 1.9241271447311153e-05,
      "loss": 0.7462,
      "step": 2361
    },
    {
      "epoch": 1.511196417146513,
      "grad_norm": 2.640491577169284,
      "learning_rate": 1.9240479509762008e-05,
      "loss": 0.7683,
      "step": 2362
    },
    {
      "epoch": 1.5118362124120281,
      "grad_norm": 2.5694472866948006,
      "learning_rate": 1.923968717544311e-05,
      "loss": 0.7141,
      "step": 2363
    },
    {
      "epoch": 1.5124760076775432,
      "grad_norm": 2.555058064142759,
      "learning_rate": 1.9238894444388484e-05,
      "loss": 0.7694,
      "step": 2364
    },
    {
      "epoch": 1.5131158029430583,
      "grad_norm": 2.3118862255335086,
      "learning_rate": 1.923810131663216e-05,
      "loss": 0.8251,
      "step": 2365
    },
    {
      "epoch": 1.5137555982085733,
      "grad_norm": 2.136226222680582,
      "learning_rate": 1.9237307792208203e-05,
      "loss": 0.7561,
      "step": 2366
    },
    {
      "epoch": 1.5143953934740884,
      "grad_norm": 2.198255883893716,
      "learning_rate": 1.9236513871150682e-05,
      "loss": 0.7622,
      "step": 2367
    },
    {
      "epoch": 1.5150351887396032,
      "grad_norm": 2.4463921715553845,
      "learning_rate": 1.9235719553493684e-05,
      "loss": 0.6973,
      "step": 2368
    },
    {
      "epoch": 1.5156749840051185,
      "grad_norm": 2.214920321143559,
      "learning_rate": 1.923492483927132e-05,
      "loss": 0.7477,
      "step": 2369
    },
    {
      "epoch": 1.5163147792706333,
      "grad_norm": 2.3916378713563184,
      "learning_rate": 1.923412972851771e-05,
      "loss": 0.7608,
      "step": 2370
    },
    {
      "epoch": 1.5169545745361486,
      "grad_norm": 2.085312934595168,
      "learning_rate": 1.9233334221266995e-05,
      "loss": 0.7277,
      "step": 2371
    },
    {
      "epoch": 1.5175943698016634,
      "grad_norm": 2.4924486751269708,
      "learning_rate": 1.923253831755333e-05,
      "loss": 0.6809,
      "step": 2372
    },
    {
      "epoch": 1.5182341650671785,
      "grad_norm": 2.0635250831584404,
      "learning_rate": 1.9231742017410897e-05,
      "loss": 0.7841,
      "step": 2373
    },
    {
      "epoch": 1.5188739603326935,
      "grad_norm": 2.1817084820376063,
      "learning_rate": 1.9230945320873884e-05,
      "loss": 0.7021,
      "step": 2374
    },
    {
      "epoch": 1.5195137555982086,
      "grad_norm": 1.8666488705056596,
      "learning_rate": 1.92301482279765e-05,
      "loss": 0.7265,
      "step": 2375
    },
    {
      "epoch": 1.5201535508637236,
      "grad_norm": 2.0057492728813995,
      "learning_rate": 1.9229350738752967e-05,
      "loss": 0.7623,
      "step": 2376
    },
    {
      "epoch": 1.5207933461292387,
      "grad_norm": 2.082013072805825,
      "learning_rate": 1.9228552853237534e-05,
      "loss": 0.7061,
      "step": 2377
    },
    {
      "epoch": 1.5214331413947537,
      "grad_norm": 2.0252986680219456,
      "learning_rate": 1.9227754571464456e-05,
      "loss": 0.6667,
      "step": 2378
    },
    {
      "epoch": 1.5220729366602685,
      "grad_norm": 2.3106768342124417,
      "learning_rate": 1.9226955893468013e-05,
      "loss": 0.749,
      "step": 2379
    },
    {
      "epoch": 1.5227127319257838,
      "grad_norm": 2.05512257553192,
      "learning_rate": 1.9226156819282496e-05,
      "loss": 0.7689,
      "step": 2380
    },
    {
      "epoch": 1.5233525271912987,
      "grad_norm": 1.9912568757263993,
      "learning_rate": 1.922535734894222e-05,
      "loss": 0.7349,
      "step": 2381
    },
    {
      "epoch": 1.523992322456814,
      "grad_norm": 2.209039440882799,
      "learning_rate": 1.9224557482481505e-05,
      "loss": 0.7136,
      "step": 2382
    },
    {
      "epoch": 1.5246321177223288,
      "grad_norm": 1.9735846660522463,
      "learning_rate": 1.9223757219934703e-05,
      "loss": 0.7606,
      "step": 2383
    },
    {
      "epoch": 1.525271912987844,
      "grad_norm": 2.152409754999961,
      "learning_rate": 1.9222956561336176e-05,
      "loss": 0.7628,
      "step": 2384
    },
    {
      "epoch": 1.5259117082533589,
      "grad_norm": 2.024120001417869,
      "learning_rate": 1.92221555067203e-05,
      "loss": 0.7025,
      "step": 2385
    },
    {
      "epoch": 1.526551503518874,
      "grad_norm": 2.075406014709473,
      "learning_rate": 1.9221354056121472e-05,
      "loss": 0.7197,
      "step": 2386
    },
    {
      "epoch": 1.527191298784389,
      "grad_norm": 2.183881576347202,
      "learning_rate": 1.92205522095741e-05,
      "loss": 0.7253,
      "step": 2387
    },
    {
      "epoch": 1.527831094049904,
      "grad_norm": 2.26437162410687,
      "learning_rate": 1.9219749967112623e-05,
      "loss": 0.7389,
      "step": 2388
    },
    {
      "epoch": 1.528470889315419,
      "grad_norm": 2.125118490981066,
      "learning_rate": 1.921894732877148e-05,
      "loss": 0.7306,
      "step": 2389
    },
    {
      "epoch": 1.5291106845809341,
      "grad_norm": 2.4340931754547697,
      "learning_rate": 1.9218144294585142e-05,
      "loss": 0.7684,
      "step": 2390
    },
    {
      "epoch": 1.5297504798464492,
      "grad_norm": 2.418868673981315,
      "learning_rate": 1.921734086458808e-05,
      "loss": 0.7879,
      "step": 2391
    },
    {
      "epoch": 1.530390275111964,
      "grad_norm": 2.2195090165185376,
      "learning_rate": 1.9216537038814803e-05,
      "loss": 0.7213,
      "step": 2392
    },
    {
      "epoch": 1.5310300703774793,
      "grad_norm": 2.2504713984471,
      "learning_rate": 1.9215732817299818e-05,
      "loss": 0.69,
      "step": 2393
    },
    {
      "epoch": 1.5316698656429941,
      "grad_norm": 2.089360953155666,
      "learning_rate": 1.9214928200077657e-05,
      "loss": 0.7524,
      "step": 2394
    },
    {
      "epoch": 1.5323096609085094,
      "grad_norm": 2.022714251239008,
      "learning_rate": 1.9214123187182876e-05,
      "loss": 0.7311,
      "step": 2395
    },
    {
      "epoch": 1.5329494561740242,
      "grad_norm": 2.1973099862375896,
      "learning_rate": 1.921331777865003e-05,
      "loss": 0.7534,
      "step": 2396
    },
    {
      "epoch": 1.5335892514395395,
      "grad_norm": 2.185820373540746,
      "learning_rate": 1.9212511974513716e-05,
      "loss": 0.7388,
      "step": 2397
    },
    {
      "epoch": 1.5342290467050543,
      "grad_norm": 2.0324457601490242,
      "learning_rate": 1.921170577480852e-05,
      "loss": 0.6986,
      "step": 2398
    },
    {
      "epoch": 1.5348688419705694,
      "grad_norm": 2.1004571425226306,
      "learning_rate": 1.9210899179569064e-05,
      "loss": 0.751,
      "step": 2399
    },
    {
      "epoch": 1.5355086372360844,
      "grad_norm": 2.0616292784745647,
      "learning_rate": 1.9210092188829986e-05,
      "loss": 0.7478,
      "step": 2400
    },
    {
      "epoch": 1.5361484325015995,
      "grad_norm": 2.1518602981145194,
      "learning_rate": 1.920928480262593e-05,
      "loss": 0.7496,
      "step": 2401
    },
    {
      "epoch": 1.5367882277671145,
      "grad_norm": 2.388210845955321,
      "learning_rate": 1.9208477020991565e-05,
      "loss": 0.669,
      "step": 2402
    },
    {
      "epoch": 1.5374280230326296,
      "grad_norm": 2.255439318594395,
      "learning_rate": 1.9207668843961584e-05,
      "loss": 0.7495,
      "step": 2403
    },
    {
      "epoch": 1.5380678182981447,
      "grad_norm": 2.292010080764029,
      "learning_rate": 1.9206860271570677e-05,
      "loss": 0.7621,
      "step": 2404
    },
    {
      "epoch": 1.5387076135636595,
      "grad_norm": 2.2175683864158575,
      "learning_rate": 1.9206051303853566e-05,
      "loss": 0.7297,
      "step": 2405
    },
    {
      "epoch": 1.5393474088291748,
      "grad_norm": 2.0978737460476236,
      "learning_rate": 1.920524194084499e-05,
      "loss": 0.7347,
      "step": 2406
    },
    {
      "epoch": 1.5399872040946896,
      "grad_norm": 1.9578455162869652,
      "learning_rate": 1.9204432182579703e-05,
      "loss": 0.7297,
      "step": 2407
    },
    {
      "epoch": 1.5406269993602049,
      "grad_norm": 2.2130992831144747,
      "learning_rate": 1.920362202909247e-05,
      "loss": 0.7733,
      "step": 2408
    },
    {
      "epoch": 1.5412667946257197,
      "grad_norm": 2.307776100876724,
      "learning_rate": 1.9202811480418075e-05,
      "loss": 0.7217,
      "step": 2409
    },
    {
      "epoch": 1.541906589891235,
      "grad_norm": 2.0437517063328285,
      "learning_rate": 1.9202000536591327e-05,
      "loss": 0.7374,
      "step": 2410
    },
    {
      "epoch": 1.5425463851567498,
      "grad_norm": 2.0822092949442212,
      "learning_rate": 1.920118919764705e-05,
      "loss": 0.839,
      "step": 2411
    },
    {
      "epoch": 1.5431861804222649,
      "grad_norm": 2.145916864953087,
      "learning_rate": 1.9200377463620072e-05,
      "loss": 0.6825,
      "step": 2412
    },
    {
      "epoch": 1.54382597568778,
      "grad_norm": 2.2342669992946176,
      "learning_rate": 1.919956533454525e-05,
      "loss": 0.7501,
      "step": 2413
    },
    {
      "epoch": 1.544465770953295,
      "grad_norm": 2.4928049365919605,
      "learning_rate": 1.919875281045746e-05,
      "loss": 0.7325,
      "step": 2414
    },
    {
      "epoch": 1.54510556621881,
      "grad_norm": 2.368772653147749,
      "learning_rate": 1.919793989139159e-05,
      "loss": 0.7421,
      "step": 2415
    },
    {
      "epoch": 1.545745361484325,
      "grad_norm": 2.047886509237291,
      "learning_rate": 1.919712657738254e-05,
      "loss": 0.7052,
      "step": 2416
    },
    {
      "epoch": 1.5463851567498401,
      "grad_norm": 2.739153461872214,
      "learning_rate": 1.9196312868465236e-05,
      "loss": 0.7539,
      "step": 2417
    },
    {
      "epoch": 1.547024952015355,
      "grad_norm": 2.3850327183976208,
      "learning_rate": 1.919549876467462e-05,
      "loss": 0.7939,
      "step": 2418
    },
    {
      "epoch": 1.5476647472808702,
      "grad_norm": 2.3640160672752333,
      "learning_rate": 1.919468426604564e-05,
      "loss": 0.8022,
      "step": 2419
    },
    {
      "epoch": 1.548304542546385,
      "grad_norm": 2.163124688892263,
      "learning_rate": 1.919386937261327e-05,
      "loss": 0.7554,
      "step": 2420
    },
    {
      "epoch": 1.5489443378119003,
      "grad_norm": 2.389624239430598,
      "learning_rate": 1.9193054084412513e-05,
      "loss": 0.7396,
      "step": 2421
    },
    {
      "epoch": 1.5495841330774152,
      "grad_norm": 2.2887849797399102,
      "learning_rate": 1.919223840147836e-05,
      "loss": 0.7187,
      "step": 2422
    },
    {
      "epoch": 1.5502239283429302,
      "grad_norm": 2.0659010647003746,
      "learning_rate": 1.9191422323845847e-05,
      "loss": 0.7042,
      "step": 2423
    },
    {
      "epoch": 1.5508637236084453,
      "grad_norm": 2.766801909582163,
      "learning_rate": 1.9190605851550007e-05,
      "loss": 0.7337,
      "step": 2424
    },
    {
      "epoch": 1.5515035188739603,
      "grad_norm": 2.337013164050637,
      "learning_rate": 1.91897889846259e-05,
      "loss": 0.715,
      "step": 2425
    },
    {
      "epoch": 1.5521433141394754,
      "grad_norm": 2.014831395955237,
      "learning_rate": 1.9188971723108598e-05,
      "loss": 0.739,
      "step": 2426
    },
    {
      "epoch": 1.5527831094049904,
      "grad_norm": 2.1279712245556373,
      "learning_rate": 1.91881540670332e-05,
      "loss": 0.7437,
      "step": 2427
    },
    {
      "epoch": 1.5534229046705055,
      "grad_norm": 2.261228639500316,
      "learning_rate": 1.918733601643481e-05,
      "loss": 0.7961,
      "step": 2428
    },
    {
      "epoch": 1.5540626999360203,
      "grad_norm": 2.509812157343553,
      "learning_rate": 1.9186517571348554e-05,
      "loss": 0.716,
      "step": 2429
    },
    {
      "epoch": 1.5547024952015356,
      "grad_norm": 2.3935686486556604,
      "learning_rate": 1.9185698731809578e-05,
      "loss": 0.7383,
      "step": 2430
    },
    {
      "epoch": 1.5553422904670504,
      "grad_norm": 2.2817455333686847,
      "learning_rate": 1.9184879497853036e-05,
      "loss": 0.7046,
      "step": 2431
    },
    {
      "epoch": 1.5559820857325657,
      "grad_norm": 1.9918348741936112,
      "learning_rate": 1.918405986951411e-05,
      "loss": 0.6667,
      "step": 2432
    },
    {
      "epoch": 1.5566218809980805,
      "grad_norm": 2.300832827408224,
      "learning_rate": 1.9183239846827985e-05,
      "loss": 0.6978,
      "step": 2433
    },
    {
      "epoch": 1.5572616762635958,
      "grad_norm": 2.3928205610315056,
      "learning_rate": 1.918241942982988e-05,
      "loss": 0.7423,
      "step": 2434
    },
    {
      "epoch": 1.5579014715291106,
      "grad_norm": 2.643686725439739,
      "learning_rate": 1.918159861855502e-05,
      "loss": 0.6727,
      "step": 2435
    },
    {
      "epoch": 1.5585412667946257,
      "grad_norm": 2.5570853541665284,
      "learning_rate": 1.9180777413038647e-05,
      "loss": 0.7495,
      "step": 2436
    },
    {
      "epoch": 1.5591810620601407,
      "grad_norm": 2.193042619605655,
      "learning_rate": 1.917995581331602e-05,
      "loss": 0.7362,
      "step": 2437
    },
    {
      "epoch": 1.5598208573256558,
      "grad_norm": 2.674239052784458,
      "learning_rate": 1.9179133819422425e-05,
      "loss": 0.6864,
      "step": 2438
    },
    {
      "epoch": 1.5604606525911708,
      "grad_norm": 2.8299510255648817,
      "learning_rate": 1.9178311431393154e-05,
      "loss": 0.7432,
      "step": 2439
    },
    {
      "epoch": 1.561100447856686,
      "grad_norm": 2.2483211472045106,
      "learning_rate": 1.917748864926351e-05,
      "loss": 0.6719,
      "step": 2440
    },
    {
      "epoch": 1.561740243122201,
      "grad_norm": 2.270728846872936,
      "learning_rate": 1.9176665473068833e-05,
      "loss": 0.7232,
      "step": 2441
    },
    {
      "epoch": 1.5623800383877158,
      "grad_norm": 2.3854367635303304,
      "learning_rate": 1.9175841902844462e-05,
      "loss": 0.7363,
      "step": 2442
    },
    {
      "epoch": 1.563019833653231,
      "grad_norm": 2.1696560942746816,
      "learning_rate": 1.9175017938625766e-05,
      "loss": 0.7093,
      "step": 2443
    },
    {
      "epoch": 1.5636596289187459,
      "grad_norm": 2.818648648294859,
      "learning_rate": 1.917419358044812e-05,
      "loss": 0.7465,
      "step": 2444
    },
    {
      "epoch": 1.5642994241842612,
      "grad_norm": 2.452579131346442,
      "learning_rate": 1.917336882834692e-05,
      "loss": 0.7888,
      "step": 2445
    },
    {
      "epoch": 1.564939219449776,
      "grad_norm": 2.1028635612370152,
      "learning_rate": 1.917254368235758e-05,
      "loss": 0.6884,
      "step": 2446
    },
    {
      "epoch": 1.5655790147152913,
      "grad_norm": 2.315341866349145,
      "learning_rate": 1.917171814251553e-05,
      "loss": 0.7352,
      "step": 2447
    },
    {
      "epoch": 1.566218809980806,
      "grad_norm": 2.5788946436007016,
      "learning_rate": 1.9170892208856222e-05,
      "loss": 0.6918,
      "step": 2448
    },
    {
      "epoch": 1.5668586052463211,
      "grad_norm": 2.241285979747387,
      "learning_rate": 1.9170065881415115e-05,
      "loss": 0.7711,
      "step": 2449
    },
    {
      "epoch": 1.5674984005118362,
      "grad_norm": 2.2654366362083875,
      "learning_rate": 1.916923916022769e-05,
      "loss": 0.7017,
      "step": 2450
    },
    {
      "epoch": 1.5681381957773513,
      "grad_norm": 2.1522500239024565,
      "learning_rate": 1.9168412045329446e-05,
      "loss": 0.7292,
      "step": 2451
    },
    {
      "epoch": 1.5687779910428663,
      "grad_norm": 2.4305178670491774,
      "learning_rate": 1.91675845367559e-05,
      "loss": 0.7252,
      "step": 2452
    },
    {
      "epoch": 1.5694177863083814,
      "grad_norm": 2.0878968409468626,
      "learning_rate": 1.9166756634542583e-05,
      "loss": 0.6707,
      "step": 2453
    },
    {
      "epoch": 1.5700575815738964,
      "grad_norm": 2.623237815187598,
      "learning_rate": 1.916592833872504e-05,
      "loss": 0.7316,
      "step": 2454
    },
    {
      "epoch": 1.5706973768394112,
      "grad_norm": 2.1728561920352862,
      "learning_rate": 1.916509964933884e-05,
      "loss": 0.7375,
      "step": 2455
    },
    {
      "epoch": 1.5713371721049265,
      "grad_norm": 2.1418242597206993,
      "learning_rate": 1.916427056641956e-05,
      "loss": 0.7451,
      "step": 2456
    },
    {
      "epoch": 1.5719769673704413,
      "grad_norm": 2.237989806158043,
      "learning_rate": 1.916344109000281e-05,
      "loss": 0.7313,
      "step": 2457
    },
    {
      "epoch": 1.5726167626359566,
      "grad_norm": 2.254011085762031,
      "learning_rate": 1.91626112201242e-05,
      "loss": 0.748,
      "step": 2458
    },
    {
      "epoch": 1.5732565579014715,
      "grad_norm": 2.1065741385242,
      "learning_rate": 1.9161780956819363e-05,
      "loss": 0.693,
      "step": 2459
    },
    {
      "epoch": 1.5738963531669867,
      "grad_norm": 2.7545234330031185,
      "learning_rate": 1.9160950300123944e-05,
      "loss": 0.737,
      "step": 2460
    },
    {
      "epoch": 1.5745361484325016,
      "grad_norm": 2.4521897509665984,
      "learning_rate": 1.916011925007362e-05,
      "loss": 0.7761,
      "step": 2461
    },
    {
      "epoch": 1.5751759436980166,
      "grad_norm": 2.056574354015116,
      "learning_rate": 1.9159287806704068e-05,
      "loss": 0.6977,
      "step": 2462
    },
    {
      "epoch": 1.5758157389635317,
      "grad_norm": 2.218902164173838,
      "learning_rate": 1.9158455970050995e-05,
      "loss": 0.7764,
      "step": 2463
    },
    {
      "epoch": 1.5764555342290467,
      "grad_norm": 2.0348589157618178,
      "learning_rate": 1.9157623740150106e-05,
      "loss": 0.702,
      "step": 2464
    },
    {
      "epoch": 1.5770953294945618,
      "grad_norm": 2.3779073394005774,
      "learning_rate": 1.915679111703715e-05,
      "loss": 0.7667,
      "step": 2465
    },
    {
      "epoch": 1.5777351247600768,
      "grad_norm": 1.9774291819513115,
      "learning_rate": 1.915595810074787e-05,
      "loss": 0.7279,
      "step": 2466
    },
    {
      "epoch": 1.5783749200255919,
      "grad_norm": 2.4005746080069925,
      "learning_rate": 1.915512469131804e-05,
      "loss": 0.7908,
      "step": 2467
    },
    {
      "epoch": 1.5790147152911067,
      "grad_norm": 2.0340315052557134,
      "learning_rate": 1.9154290888783435e-05,
      "loss": 0.7715,
      "step": 2468
    },
    {
      "epoch": 1.579654510556622,
      "grad_norm": 1.996040297983495,
      "learning_rate": 1.9153456693179867e-05,
      "loss": 0.7841,
      "step": 2469
    },
    {
      "epoch": 1.5802943058221368,
      "grad_norm": 2.0514886250259075,
      "learning_rate": 1.915262210454315e-05,
      "loss": 0.7045,
      "step": 2470
    },
    {
      "epoch": 1.580934101087652,
      "grad_norm": 2.018701020619258,
      "learning_rate": 1.9151787122909122e-05,
      "loss": 0.6881,
      "step": 2471
    },
    {
      "epoch": 1.581573896353167,
      "grad_norm": 1.9587036363221149,
      "learning_rate": 1.915095174831364e-05,
      "loss": 0.6417,
      "step": 2472
    },
    {
      "epoch": 1.582213691618682,
      "grad_norm": 2.177117784071049,
      "learning_rate": 1.9150115980792557e-05,
      "loss": 0.7681,
      "step": 2473
    },
    {
      "epoch": 1.582853486884197,
      "grad_norm": 2.2318749240611657,
      "learning_rate": 1.914927982038178e-05,
      "loss": 0.7562,
      "step": 2474
    },
    {
      "epoch": 1.583493282149712,
      "grad_norm": 2.393085907631167,
      "learning_rate": 1.91484432671172e-05,
      "loss": 0.7101,
      "step": 2475
    },
    {
      "epoch": 1.5841330774152271,
      "grad_norm": 2.2977537785606,
      "learning_rate": 1.914760632103474e-05,
      "loss": 0.7153,
      "step": 2476
    },
    {
      "epoch": 1.5847728726807422,
      "grad_norm": 2.3774943777184614,
      "learning_rate": 1.9146768982170336e-05,
      "loss": 0.6877,
      "step": 2477
    },
    {
      "epoch": 1.5854126679462572,
      "grad_norm": 2.18489530437435,
      "learning_rate": 1.9145931250559943e-05,
      "loss": 0.6726,
      "step": 2478
    },
    {
      "epoch": 1.586052463211772,
      "grad_norm": 2.147618353304727,
      "learning_rate": 1.914509312623953e-05,
      "loss": 0.8029,
      "step": 2479
    },
    {
      "epoch": 1.5866922584772873,
      "grad_norm": 2.296465965598972,
      "learning_rate": 1.9144254609245086e-05,
      "loss": 0.7466,
      "step": 2480
    },
    {
      "epoch": 1.5873320537428022,
      "grad_norm": 2.156664381312445,
      "learning_rate": 1.9143415699612617e-05,
      "loss": 0.7127,
      "step": 2481
    },
    {
      "epoch": 1.5879718490083174,
      "grad_norm": 2.1583227153758964,
      "learning_rate": 1.9142576397378144e-05,
      "loss": 0.7071,
      "step": 2482
    },
    {
      "epoch": 1.5886116442738323,
      "grad_norm": 2.4579457299722565,
      "learning_rate": 1.9141736702577704e-05,
      "loss": 0.7374,
      "step": 2483
    },
    {
      "epoch": 1.5892514395393476,
      "grad_norm": 2.3167493071275964,
      "learning_rate": 1.914089661524735e-05,
      "loss": 0.7169,
      "step": 2484
    },
    {
      "epoch": 1.5898912348048624,
      "grad_norm": 2.04297423119437,
      "learning_rate": 1.9140056135423156e-05,
      "loss": 0.7041,
      "step": 2485
    },
    {
      "epoch": 1.5905310300703774,
      "grad_norm": 2.1692538527117846,
      "learning_rate": 1.913921526314121e-05,
      "loss": 0.7023,
      "step": 2486
    },
    {
      "epoch": 1.5911708253358925,
      "grad_norm": 2.1840769264025153,
      "learning_rate": 1.913837399843762e-05,
      "loss": 0.7443,
      "step": 2487
    },
    {
      "epoch": 1.5918106206014075,
      "grad_norm": 2.393092852021879,
      "learning_rate": 1.9137532341348503e-05,
      "loss": 0.7518,
      "step": 2488
    },
    {
      "epoch": 1.5924504158669226,
      "grad_norm": 2.6754728351402353,
      "learning_rate": 1.9136690291910005e-05,
      "loss": 0.7548,
      "step": 2489
    },
    {
      "epoch": 1.5930902111324377,
      "grad_norm": 2.491861696513052,
      "learning_rate": 1.9135847850158278e-05,
      "loss": 0.7065,
      "step": 2490
    },
    {
      "epoch": 1.5937300063979527,
      "grad_norm": 1.9110972240051884,
      "learning_rate": 1.9135005016129495e-05,
      "loss": 0.695,
      "step": 2491
    },
    {
      "epoch": 1.5943698016634675,
      "grad_norm": 2.1365251398609657,
      "learning_rate": 1.9134161789859847e-05,
      "loss": 0.7199,
      "step": 2492
    },
    {
      "epoch": 1.5950095969289828,
      "grad_norm": 2.790088734551431,
      "learning_rate": 1.9133318171385537e-05,
      "loss": 0.7902,
      "step": 2493
    },
    {
      "epoch": 1.5956493921944976,
      "grad_norm": 2.333903247852572,
      "learning_rate": 1.9132474160742797e-05,
      "loss": 0.729,
      "step": 2494
    },
    {
      "epoch": 1.596289187460013,
      "grad_norm": 2.1214300381730595,
      "learning_rate": 1.913162975796786e-05,
      "loss": 0.748,
      "step": 2495
    },
    {
      "epoch": 1.5969289827255277,
      "grad_norm": 2.1687291139239537,
      "learning_rate": 1.9130784963096986e-05,
      "loss": 0.6784,
      "step": 2496
    },
    {
      "epoch": 1.597568777991043,
      "grad_norm": 2.0381834798812783,
      "learning_rate": 1.9129939776166445e-05,
      "loss": 0.7533,
      "step": 2497
    },
    {
      "epoch": 1.5982085732565579,
      "grad_norm": 2.111403659680493,
      "learning_rate": 1.9129094197212533e-05,
      "loss": 0.739,
      "step": 2498
    },
    {
      "epoch": 1.598848368522073,
      "grad_norm": 2.52892055113424,
      "learning_rate": 1.9128248226271556e-05,
      "loss": 0.7701,
      "step": 2499
    },
    {
      "epoch": 1.599488163787588,
      "grad_norm": 2.315278751997447,
      "learning_rate": 1.9127401863379835e-05,
      "loss": 0.662,
      "step": 2500
    },
    {
      "epoch": 1.600127959053103,
      "grad_norm": 2.039892542437467,
      "learning_rate": 1.9126555108573718e-05,
      "loss": 0.7266,
      "step": 2501
    },
    {
      "epoch": 1.600767754318618,
      "grad_norm": 2.272345498498822,
      "learning_rate": 1.9125707961889556e-05,
      "loss": 0.7437,
      "step": 2502
    },
    {
      "epoch": 1.6014075495841331,
      "grad_norm": 2.219741742451431,
      "learning_rate": 1.912486042336373e-05,
      "loss": 0.7149,
      "step": 2503
    },
    {
      "epoch": 1.6020473448496482,
      "grad_norm": 2.4340391628716804,
      "learning_rate": 1.912401249303263e-05,
      "loss": 0.7343,
      "step": 2504
    },
    {
      "epoch": 1.602687140115163,
      "grad_norm": 2.423543276102326,
      "learning_rate": 1.9123164170932663e-05,
      "loss": 0.7497,
      "step": 2505
    },
    {
      "epoch": 1.6033269353806783,
      "grad_norm": 2.163020859057053,
      "learning_rate": 1.9122315457100255e-05,
      "loss": 0.7112,
      "step": 2506
    },
    {
      "epoch": 1.603966730646193,
      "grad_norm": 2.0546827678411628,
      "learning_rate": 1.9121466351571847e-05,
      "loss": 0.686,
      "step": 2507
    },
    {
      "epoch": 1.6046065259117084,
      "grad_norm": 2.183154423510172,
      "learning_rate": 1.91206168543839e-05,
      "loss": 0.7846,
      "step": 2508
    },
    {
      "epoch": 1.6052463211772232,
      "grad_norm": 2.048401645569352,
      "learning_rate": 1.9119766965572894e-05,
      "loss": 0.6675,
      "step": 2509
    },
    {
      "epoch": 1.6058861164427385,
      "grad_norm": 2.5283462934914347,
      "learning_rate": 1.9118916685175317e-05,
      "loss": 0.7136,
      "step": 2510
    },
    {
      "epoch": 1.6065259117082533,
      "grad_norm": 2.2874753735947517,
      "learning_rate": 1.9118066013227673e-05,
      "loss": 0.7385,
      "step": 2511
    },
    {
      "epoch": 1.6071657069737684,
      "grad_norm": 2.222405991123351,
      "learning_rate": 1.91172149497665e-05,
      "loss": 0.6916,
      "step": 2512
    },
    {
      "epoch": 1.6078055022392834,
      "grad_norm": 2.686430487273565,
      "learning_rate": 1.911636349482833e-05,
      "loss": 0.7404,
      "step": 2513
    },
    {
      "epoch": 1.6084452975047985,
      "grad_norm": 2.4211131155828256,
      "learning_rate": 1.9115511648449732e-05,
      "loss": 0.654,
      "step": 2514
    },
    {
      "epoch": 1.6090850927703135,
      "grad_norm": 2.0640185580029864,
      "learning_rate": 1.9114659410667275e-05,
      "loss": 0.7003,
      "step": 2515
    },
    {
      "epoch": 1.6097248880358286,
      "grad_norm": 2.0565537850500313,
      "learning_rate": 1.911380678151756e-05,
      "loss": 0.7571,
      "step": 2516
    },
    {
      "epoch": 1.6103646833013436,
      "grad_norm": 2.145997642347237,
      "learning_rate": 1.9112953761037195e-05,
      "loss": 0.6893,
      "step": 2517
    },
    {
      "epoch": 1.6110044785668585,
      "grad_norm": 2.3453371357483337,
      "learning_rate": 1.9112100349262807e-05,
      "loss": 0.7185,
      "step": 2518
    },
    {
      "epoch": 1.6116442738323737,
      "grad_norm": 2.298957682752517,
      "learning_rate": 1.9111246546231036e-05,
      "loss": 0.7382,
      "step": 2519
    },
    {
      "epoch": 1.6122840690978886,
      "grad_norm": 2.066830758600034,
      "learning_rate": 1.9110392351978545e-05,
      "loss": 0.7372,
      "step": 2520
    },
    {
      "epoch": 1.6129238643634038,
      "grad_norm": 2.3787395213223936,
      "learning_rate": 1.9109537766542017e-05,
      "loss": 0.7229,
      "step": 2521
    },
    {
      "epoch": 1.6135636596289187,
      "grad_norm": 2.3730623704464695,
      "learning_rate": 1.910868278995814e-05,
      "loss": 0.7382,
      "step": 2522
    },
    {
      "epoch": 1.6142034548944337,
      "grad_norm": 2.1397457558343986,
      "learning_rate": 1.9107827422263625e-05,
      "loss": 0.7728,
      "step": 2523
    },
    {
      "epoch": 1.6148432501599488,
      "grad_norm": 2.0809793348687657,
      "learning_rate": 1.9106971663495205e-05,
      "loss": 0.6958,
      "step": 2524
    },
    {
      "epoch": 1.6154830454254638,
      "grad_norm": 2.1562549104617417,
      "learning_rate": 1.910611551368962e-05,
      "loss": 0.6568,
      "step": 2525
    },
    {
      "epoch": 1.616122840690979,
      "grad_norm": 1.980854859606732,
      "learning_rate": 1.9105258972883632e-05,
      "loss": 0.7044,
      "step": 2526
    },
    {
      "epoch": 1.616762635956494,
      "grad_norm": 2.1858595212292213,
      "learning_rate": 1.9104402041114022e-05,
      "loss": 0.6769,
      "step": 2527
    },
    {
      "epoch": 1.617402431222009,
      "grad_norm": 2.2404974228911203,
      "learning_rate": 1.9103544718417585e-05,
      "loss": 0.7856,
      "step": 2528
    },
    {
      "epoch": 1.6180422264875238,
      "grad_norm": 2.094444075823284,
      "learning_rate": 1.9102687004831134e-05,
      "loss": 0.7076,
      "step": 2529
    },
    {
      "epoch": 1.618682021753039,
      "grad_norm": 2.477108537195321,
      "learning_rate": 1.910182890039149e-05,
      "loss": 0.7381,
      "step": 2530
    },
    {
      "epoch": 1.619321817018554,
      "grad_norm": 2.2416660366217083,
      "learning_rate": 1.9100970405135506e-05,
      "loss": 0.7426,
      "step": 2531
    },
    {
      "epoch": 1.6199616122840692,
      "grad_norm": 2.084427750008447,
      "learning_rate": 1.9100111519100042e-05,
      "loss": 0.7069,
      "step": 2532
    },
    {
      "epoch": 1.620601407549584,
      "grad_norm": 2.1222964352038334,
      "learning_rate": 1.9099252242321978e-05,
      "loss": 0.727,
      "step": 2533
    },
    {
      "epoch": 1.6212412028150993,
      "grad_norm": 2.137763734092989,
      "learning_rate": 1.909839257483821e-05,
      "loss": 0.7438,
      "step": 2534
    },
    {
      "epoch": 1.6218809980806141,
      "grad_norm": 2.3083621515208286,
      "learning_rate": 1.9097532516685645e-05,
      "loss": 0.7342,
      "step": 2535
    },
    {
      "epoch": 1.6225207933461292,
      "grad_norm": 2.2087773629095526,
      "learning_rate": 1.9096672067901224e-05,
      "loss": 0.6824,
      "step": 2536
    },
    {
      "epoch": 1.6231605886116443,
      "grad_norm": 2.173778351022402,
      "learning_rate": 1.9095811228521884e-05,
      "loss": 0.7175,
      "step": 2537
    },
    {
      "epoch": 1.6238003838771593,
      "grad_norm": 2.055900210562481,
      "learning_rate": 1.9094949998584587e-05,
      "loss": 0.7043,
      "step": 2538
    },
    {
      "epoch": 1.6244401791426744,
      "grad_norm": 2.418918944398108,
      "learning_rate": 1.9094088378126317e-05,
      "loss": 0.6714,
      "step": 2539
    },
    {
      "epoch": 1.6250799744081894,
      "grad_norm": 2.0811810997077878,
      "learning_rate": 1.909322636718407e-05,
      "loss": 0.6873,
      "step": 2540
    },
    {
      "epoch": 1.6257197696737045,
      "grad_norm": 2.052788358303775,
      "learning_rate": 1.9092363965794857e-05,
      "loss": 0.6942,
      "step": 2541
    },
    {
      "epoch": 1.6263595649392193,
      "grad_norm": 1.922074740335183,
      "learning_rate": 1.909150117399571e-05,
      "loss": 0.6329,
      "step": 2542
    },
    {
      "epoch": 1.6269993602047346,
      "grad_norm": 2.128094352012945,
      "learning_rate": 1.9090637991823676e-05,
      "loss": 0.7054,
      "step": 2543
    },
    {
      "epoch": 1.6276391554702494,
      "grad_norm": 2.121990575996939,
      "learning_rate": 1.9089774419315817e-05,
      "loss": 0.698,
      "step": 2544
    },
    {
      "epoch": 1.6282789507357647,
      "grad_norm": 2.186615951071917,
      "learning_rate": 1.9088910456509213e-05,
      "loss": 0.6932,
      "step": 2545
    },
    {
      "epoch": 1.6289187460012795,
      "grad_norm": 2.4261912598690585,
      "learning_rate": 1.908804610344096e-05,
      "loss": 0.701,
      "step": 2546
    },
    {
      "epoch": 1.6295585412667948,
      "grad_norm": 2.313748601323897,
      "learning_rate": 1.908718136014818e-05,
      "loss": 0.7596,
      "step": 2547
    },
    {
      "epoch": 1.6301983365323096,
      "grad_norm": 2.1181559187571732,
      "learning_rate": 1.9086316226667993e-05,
      "loss": 0.7296,
      "step": 2548
    },
    {
      "epoch": 1.6308381317978247,
      "grad_norm": 2.0230370229495644,
      "learning_rate": 1.908545070303755e-05,
      "loss": 0.7135,
      "step": 2549
    },
    {
      "epoch": 1.6314779270633397,
      "grad_norm": 2.3494800090926558,
      "learning_rate": 1.9084584789294018e-05,
      "loss": 0.7193,
      "step": 2550
    },
    {
      "epoch": 1.6321177223288548,
      "grad_norm": 1.9898029691828933,
      "learning_rate": 1.9083718485474576e-05,
      "loss": 0.7161,
      "step": 2551
    },
    {
      "epoch": 1.6327575175943698,
      "grad_norm": 2.2171874523709842,
      "learning_rate": 1.908285179161642e-05,
      "loss": 0.7146,
      "step": 2552
    },
    {
      "epoch": 1.6333973128598849,
      "grad_norm": 2.0425542334716416,
      "learning_rate": 1.908198470775676e-05,
      "loss": 0.6956,
      "step": 2553
    },
    {
      "epoch": 1.6340371081254,
      "grad_norm": 2.214227442308929,
      "learning_rate": 1.908111723393284e-05,
      "loss": 0.6709,
      "step": 2554
    },
    {
      "epoch": 1.6346769033909148,
      "grad_norm": 2.224330984992187,
      "learning_rate": 1.9080249370181897e-05,
      "loss": 0.7561,
      "step": 2555
    },
    {
      "epoch": 1.63531669865643,
      "grad_norm": 2.1556535442876994,
      "learning_rate": 1.9079381116541197e-05,
      "loss": 0.6604,
      "step": 2556
    },
    {
      "epoch": 1.6359564939219449,
      "grad_norm": 2.7064232118666665,
      "learning_rate": 1.9078512473048023e-05,
      "loss": 0.7273,
      "step": 2557
    },
    {
      "epoch": 1.6365962891874601,
      "grad_norm": 2.277590450402744,
      "learning_rate": 1.9077643439739675e-05,
      "loss": 0.7885,
      "step": 2558
    },
    {
      "epoch": 1.637236084452975,
      "grad_norm": 2.331561308441229,
      "learning_rate": 1.9076774016653463e-05,
      "loss": 0.7225,
      "step": 2559
    },
    {
      "epoch": 1.6378758797184902,
      "grad_norm": 2.217014338021757,
      "learning_rate": 1.9075904203826724e-05,
      "loss": 0.7145,
      "step": 2560
    },
    {
      "epoch": 1.638515674984005,
      "grad_norm": 2.7996254196221453,
      "learning_rate": 1.90750340012968e-05,
      "loss": 0.7021,
      "step": 2561
    },
    {
      "epoch": 1.6391554702495201,
      "grad_norm": 2.2151476857527532,
      "learning_rate": 1.907416340910106e-05,
      "loss": 0.7348,
      "step": 2562
    },
    {
      "epoch": 1.6397952655150352,
      "grad_norm": 2.207479479338689,
      "learning_rate": 1.9073292427276887e-05,
      "loss": 0.6669,
      "step": 2563
    },
    {
      "epoch": 1.6404350607805502,
      "grad_norm": 2.29120280996252,
      "learning_rate": 1.9072421055861675e-05,
      "loss": 0.6967,
      "step": 2564
    },
    {
      "epoch": 1.6410748560460653,
      "grad_norm": 2.0499960406537543,
      "learning_rate": 1.9071549294892843e-05,
      "loss": 0.7655,
      "step": 2565
    },
    {
      "epoch": 1.6417146513115803,
      "grad_norm": 2.275378590129151,
      "learning_rate": 1.9070677144407822e-05,
      "loss": 0.6909,
      "step": 2566
    },
    {
      "epoch": 1.6423544465770954,
      "grad_norm": 2.511276122614611,
      "learning_rate": 1.906980460444406e-05,
      "loss": 0.7675,
      "step": 2567
    },
    {
      "epoch": 1.6429942418426102,
      "grad_norm": 2.4079796494532926,
      "learning_rate": 1.906893167503902e-05,
      "loss": 0.7226,
      "step": 2568
    },
    {
      "epoch": 1.6436340371081255,
      "grad_norm": 2.3296464034134035,
      "learning_rate": 1.906805835623019e-05,
      "loss": 0.7503,
      "step": 2569
    },
    {
      "epoch": 1.6442738323736403,
      "grad_norm": 2.1513881209505588,
      "learning_rate": 1.906718464805506e-05,
      "loss": 0.6866,
      "step": 2570
    },
    {
      "epoch": 1.6449136276391556,
      "grad_norm": 2.2063262451999432,
      "learning_rate": 1.9066310550551153e-05,
      "loss": 0.6872,
      "step": 2571
    },
    {
      "epoch": 1.6455534229046704,
      "grad_norm": 2.084052625100513,
      "learning_rate": 1.9065436063755998e-05,
      "loss": 0.6961,
      "step": 2572
    },
    {
      "epoch": 1.6461932181701855,
      "grad_norm": 2.51717352935784,
      "learning_rate": 1.9064561187707143e-05,
      "loss": 0.7471,
      "step": 2573
    },
    {
      "epoch": 1.6468330134357005,
      "grad_norm": 2.2676206201308178,
      "learning_rate": 1.906368592244216e-05,
      "loss": 0.7338,
      "step": 2574
    },
    {
      "epoch": 1.6474728087012156,
      "grad_norm": 2.1075771013715854,
      "learning_rate": 1.9062810267998623e-05,
      "loss": 0.727,
      "step": 2575
    },
    {
      "epoch": 1.6481126039667306,
      "grad_norm": 2.0821456062465233,
      "learning_rate": 1.9061934224414136e-05,
      "loss": 0.6992,
      "step": 2576
    },
    {
      "epoch": 1.6487523992322457,
      "grad_norm": 1.963824127801277,
      "learning_rate": 1.9061057791726312e-05,
      "loss": 0.6603,
      "step": 2577
    },
    {
      "epoch": 1.6493921944977608,
      "grad_norm": 2.2825092123130464,
      "learning_rate": 1.9060180969972783e-05,
      "loss": 0.7697,
      "step": 2578
    },
    {
      "epoch": 1.6500319897632756,
      "grad_norm": 2.283164131749146,
      "learning_rate": 1.9059303759191203e-05,
      "loss": 0.6646,
      "step": 2579
    },
    {
      "epoch": 1.6506717850287909,
      "grad_norm": 2.1681692010241695,
      "learning_rate": 1.9058426159419236e-05,
      "loss": 0.7061,
      "step": 2580
    },
    {
      "epoch": 1.6513115802943057,
      "grad_norm": 2.5726237332462576,
      "learning_rate": 1.905754817069456e-05,
      "loss": 0.6805,
      "step": 2581
    },
    {
      "epoch": 1.651951375559821,
      "grad_norm": 2.0216229415946794,
      "learning_rate": 1.905666979305488e-05,
      "loss": 0.6663,
      "step": 2582
    },
    {
      "epoch": 1.6525911708253358,
      "grad_norm": 2.19983217627919,
      "learning_rate": 1.9055791026537908e-05,
      "loss": 0.6892,
      "step": 2583
    },
    {
      "epoch": 1.653230966090851,
      "grad_norm": 2.2399143313411756,
      "learning_rate": 1.9054911871181382e-05,
      "loss": 0.7296,
      "step": 2584
    },
    {
      "epoch": 1.653870761356366,
      "grad_norm": 2.2190512144549164,
      "learning_rate": 1.9054032327023044e-05,
      "loss": 0.7221,
      "step": 2585
    },
    {
      "epoch": 1.654510556621881,
      "grad_norm": 1.9982351646215852,
      "learning_rate": 1.9053152394100664e-05,
      "loss": 0.7569,
      "step": 2586
    },
    {
      "epoch": 1.655150351887396,
      "grad_norm": 2.015203320139472,
      "learning_rate": 1.9052272072452024e-05,
      "loss": 0.7035,
      "step": 2587
    },
    {
      "epoch": 1.655790147152911,
      "grad_norm": 2.1749738595337638,
      "learning_rate": 1.9051391362114927e-05,
      "loss": 0.6934,
      "step": 2588
    },
    {
      "epoch": 1.6564299424184261,
      "grad_norm": 2.1858161589491716,
      "learning_rate": 1.9050510263127184e-05,
      "loss": 0.7278,
      "step": 2589
    },
    {
      "epoch": 1.6570697376839412,
      "grad_norm": 1.9920912177717403,
      "learning_rate": 1.904962877552663e-05,
      "loss": 0.7022,
      "step": 2590
    },
    {
      "epoch": 1.6577095329494562,
      "grad_norm": 2.137377306284484,
      "learning_rate": 1.9048746899351115e-05,
      "loss": 0.7038,
      "step": 2591
    },
    {
      "epoch": 1.658349328214971,
      "grad_norm": 2.0631615649087354,
      "learning_rate": 1.9047864634638503e-05,
      "loss": 0.6835,
      "step": 2592
    },
    {
      "epoch": 1.6589891234804863,
      "grad_norm": 2.150072403163884,
      "learning_rate": 1.904698198142668e-05,
      "loss": 0.7019,
      "step": 2593
    },
    {
      "epoch": 1.6596289187460012,
      "grad_norm": 2.2234990677810478,
      "learning_rate": 1.9046098939753546e-05,
      "loss": 0.673,
      "step": 2594
    },
    {
      "epoch": 1.6602687140115164,
      "grad_norm": 2.060586206525205,
      "learning_rate": 1.904521550965701e-05,
      "loss": 0.7177,
      "step": 2595
    },
    {
      "epoch": 1.6609085092770313,
      "grad_norm": 2.4579445816374275,
      "learning_rate": 1.9044331691175014e-05,
      "loss": 0.7522,
      "step": 2596
    },
    {
      "epoch": 1.6615483045425465,
      "grad_norm": 2.3387816178021428,
      "learning_rate": 1.9043447484345504e-05,
      "loss": 0.7209,
      "step": 2597
    },
    {
      "epoch": 1.6621880998080614,
      "grad_norm": 2.1612925839567025,
      "learning_rate": 1.9042562889206446e-05,
      "loss": 0.7116,
      "step": 2598
    },
    {
      "epoch": 1.6628278950735764,
      "grad_norm": 2.0839914534560204,
      "learning_rate": 1.9041677905795822e-05,
      "loss": 0.6663,
      "step": 2599
    },
    {
      "epoch": 1.6634676903390915,
      "grad_norm": 2.229136002053383,
      "learning_rate": 1.9040792534151632e-05,
      "loss": 0.7101,
      "step": 2600
    },
    {
      "epoch": 1.6641074856046065,
      "grad_norm": 2.0298804420628778,
      "learning_rate": 1.9039906774311893e-05,
      "loss": 0.6943,
      "step": 2601
    },
    {
      "epoch": 1.6647472808701216,
      "grad_norm": 2.244550117129175,
      "learning_rate": 1.9039020626314637e-05,
      "loss": 0.7016,
      "step": 2602
    },
    {
      "epoch": 1.6653870761356366,
      "grad_norm": 2.1157036033666716,
      "learning_rate": 1.9038134090197916e-05,
      "loss": 0.6884,
      "step": 2603
    },
    {
      "epoch": 1.6660268714011517,
      "grad_norm": 2.055061070959906,
      "learning_rate": 1.9037247165999796e-05,
      "loss": 0.7052,
      "step": 2604
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 2.041661047248366,
      "learning_rate": 1.9036359853758356e-05,
      "loss": 0.7165,
      "step": 2605
    },
    {
      "epoch": 1.6673064619321818,
      "grad_norm": 2.047122078386663,
      "learning_rate": 1.90354721535117e-05,
      "loss": 0.725,
      "step": 2606
    },
    {
      "epoch": 1.6679462571976966,
      "grad_norm": 2.308681848063503,
      "learning_rate": 1.9034584065297945e-05,
      "loss": 0.6818,
      "step": 2607
    },
    {
      "epoch": 1.668586052463212,
      "grad_norm": 2.033875995582213,
      "learning_rate": 1.9033695589155217e-05,
      "loss": 0.7686,
      "step": 2608
    },
    {
      "epoch": 1.6692258477287267,
      "grad_norm": 2.0833760000755697,
      "learning_rate": 1.9032806725121672e-05,
      "loss": 0.6856,
      "step": 2609
    },
    {
      "epoch": 1.669865642994242,
      "grad_norm": 1.883251767338807,
      "learning_rate": 1.9031917473235474e-05,
      "loss": 0.7482,
      "step": 2610
    },
    {
      "epoch": 1.6705054382597568,
      "grad_norm": 2.2583724537064027,
      "learning_rate": 1.9031027833534808e-05,
      "loss": 0.7278,
      "step": 2611
    },
    {
      "epoch": 1.671145233525272,
      "grad_norm": 1.942142066863069,
      "learning_rate": 1.9030137806057868e-05,
      "loss": 0.7229,
      "step": 2612
    },
    {
      "epoch": 1.671785028790787,
      "grad_norm": 2.140911176258487,
      "learning_rate": 1.9029247390842878e-05,
      "loss": 0.7213,
      "step": 2613
    },
    {
      "epoch": 1.672424824056302,
      "grad_norm": 2.4037698558277367,
      "learning_rate": 1.9028356587928064e-05,
      "loss": 0.6904,
      "step": 2614
    },
    {
      "epoch": 1.673064619321817,
      "grad_norm": 2.555068647036384,
      "learning_rate": 1.902746539735168e-05,
      "loss": 0.7099,
      "step": 2615
    },
    {
      "epoch": 1.673704414587332,
      "grad_norm": 2.140102258768062,
      "learning_rate": 1.902657381915199e-05,
      "loss": 0.7091,
      "step": 2616
    },
    {
      "epoch": 1.6743442098528472,
      "grad_norm": 2.239017266675136,
      "learning_rate": 1.902568185336728e-05,
      "loss": 0.658,
      "step": 2617
    },
    {
      "epoch": 1.674984005118362,
      "grad_norm": 2.1374940994180305,
      "learning_rate": 1.9024789500035842e-05,
      "loss": 0.6519,
      "step": 2618
    },
    {
      "epoch": 1.6756238003838773,
      "grad_norm": 2.232032293428308,
      "learning_rate": 1.9023896759195998e-05,
      "loss": 0.7137,
      "step": 2619
    },
    {
      "epoch": 1.676263595649392,
      "grad_norm": 2.2595866442022667,
      "learning_rate": 1.9023003630886082e-05,
      "loss": 0.7801,
      "step": 2620
    },
    {
      "epoch": 1.6769033909149074,
      "grad_norm": 2.2208158150916306,
      "learning_rate": 1.9022110115144435e-05,
      "loss": 0.7334,
      "step": 2621
    },
    {
      "epoch": 1.6775431861804222,
      "grad_norm": 2.366664257063329,
      "learning_rate": 1.9021216212009433e-05,
      "loss": 0.6868,
      "step": 2622
    },
    {
      "epoch": 1.6781829814459372,
      "grad_norm": 2.450872146489939,
      "learning_rate": 1.9020321921519455e-05,
      "loss": 0.756,
      "step": 2623
    },
    {
      "epoch": 1.6788227767114523,
      "grad_norm": 2.036759487556971,
      "learning_rate": 1.9019427243712897e-05,
      "loss": 0.7087,
      "step": 2624
    },
    {
      "epoch": 1.6794625719769674,
      "grad_norm": 2.0882196238449264,
      "learning_rate": 1.901853217862818e-05,
      "loss": 0.72,
      "step": 2625
    },
    {
      "epoch": 1.6801023672424824,
      "grad_norm": 1.9436986412631376,
      "learning_rate": 1.9017636726303733e-05,
      "loss": 0.7346,
      "step": 2626
    },
    {
      "epoch": 1.6807421625079975,
      "grad_norm": 2.045417231593522,
      "learning_rate": 1.9016740886778006e-05,
      "loss": 0.6486,
      "step": 2627
    },
    {
      "epoch": 1.6813819577735125,
      "grad_norm": 2.1432706656399967,
      "learning_rate": 1.9015844660089464e-05,
      "loss": 0.6649,
      "step": 2628
    },
    {
      "epoch": 1.6820217530390273,
      "grad_norm": 2.279526035281614,
      "learning_rate": 1.9014948046276593e-05,
      "loss": 0.6577,
      "step": 2629
    },
    {
      "epoch": 1.6826615483045426,
      "grad_norm": 2.187156878546102,
      "learning_rate": 1.9014051045377887e-05,
      "loss": 0.737,
      "step": 2630
    },
    {
      "epoch": 1.6833013435700575,
      "grad_norm": 2.27995860402141,
      "learning_rate": 1.9013153657431863e-05,
      "loss": 0.7319,
      "step": 2631
    },
    {
      "epoch": 1.6839411388355727,
      "grad_norm": 2.4756495966941263,
      "learning_rate": 1.9012255882477056e-05,
      "loss": 0.7051,
      "step": 2632
    },
    {
      "epoch": 1.6845809341010876,
      "grad_norm": 2.1694969011919323,
      "learning_rate": 1.9011357720552014e-05,
      "loss": 0.742,
      "step": 2633
    },
    {
      "epoch": 1.6852207293666028,
      "grad_norm": 2.3029944183028634,
      "learning_rate": 1.9010459171695298e-05,
      "loss": 0.7054,
      "step": 2634
    },
    {
      "epoch": 1.6858605246321177,
      "grad_norm": 2.028823680608874,
      "learning_rate": 1.9009560235945495e-05,
      "loss": 0.6559,
      "step": 2635
    },
    {
      "epoch": 1.6865003198976327,
      "grad_norm": 2.0964852638372924,
      "learning_rate": 1.90086609133412e-05,
      "loss": 0.6614,
      "step": 2636
    },
    {
      "epoch": 1.6871401151631478,
      "grad_norm": 2.1264605645868886,
      "learning_rate": 1.9007761203921035e-05,
      "loss": 0.7319,
      "step": 2637
    },
    {
      "epoch": 1.6877799104286628,
      "grad_norm": 2.330071303193591,
      "learning_rate": 1.9006861107723623e-05,
      "loss": 0.7002,
      "step": 2638
    },
    {
      "epoch": 1.6884197056941779,
      "grad_norm": 2.269761582157185,
      "learning_rate": 1.900596062478762e-05,
      "loss": 0.6972,
      "step": 2639
    },
    {
      "epoch": 1.689059500959693,
      "grad_norm": 2.3039502764552626,
      "learning_rate": 1.9005059755151692e-05,
      "loss": 0.6398,
      "step": 2640
    },
    {
      "epoch": 1.689699296225208,
      "grad_norm": 2.0311593848781975,
      "learning_rate": 1.900415849885451e-05,
      "loss": 0.6388,
      "step": 2641
    },
    {
      "epoch": 1.6903390914907228,
      "grad_norm": 2.2946858141794957,
      "learning_rate": 1.9003256855934784e-05,
      "loss": 0.7298,
      "step": 2642
    },
    {
      "epoch": 1.690978886756238,
      "grad_norm": 2.1876856221332015,
      "learning_rate": 1.900235482643122e-05,
      "loss": 0.6657,
      "step": 2643
    },
    {
      "epoch": 1.691618682021753,
      "grad_norm": 2.0663989355955885,
      "learning_rate": 1.9001452410382553e-05,
      "loss": 0.6553,
      "step": 2644
    },
    {
      "epoch": 1.6922584772872682,
      "grad_norm": 1.986976128888432,
      "learning_rate": 1.9000549607827536e-05,
      "loss": 0.6944,
      "step": 2645
    },
    {
      "epoch": 1.692898272552783,
      "grad_norm": 2.0506559705758534,
      "learning_rate": 1.899964641880493e-05,
      "loss": 0.7213,
      "step": 2646
    },
    {
      "epoch": 1.6935380678182983,
      "grad_norm": 1.9383054339861407,
      "learning_rate": 1.8998742843353514e-05,
      "loss": 0.7054,
      "step": 2647
    },
    {
      "epoch": 1.6941778630838131,
      "grad_norm": 2.208952101310644,
      "learning_rate": 1.8997838881512084e-05,
      "loss": 0.6916,
      "step": 2648
    },
    {
      "epoch": 1.6948176583493282,
      "grad_norm": 2.019648883370489,
      "learning_rate": 1.8996934533319463e-05,
      "loss": 0.706,
      "step": 2649
    },
    {
      "epoch": 1.6954574536148432,
      "grad_norm": 1.957200679963791,
      "learning_rate": 1.8996029798814477e-05,
      "loss": 0.6611,
      "step": 2650
    },
    {
      "epoch": 1.6960972488803583,
      "grad_norm": 2.0499816589275013,
      "learning_rate": 1.8995124678035976e-05,
      "loss": 0.7,
      "step": 2651
    },
    {
      "epoch": 1.6967370441458733,
      "grad_norm": 2.259460895771713,
      "learning_rate": 1.8994219171022817e-05,
      "loss": 0.7522,
      "step": 2652
    },
    {
      "epoch": 1.6973768394113884,
      "grad_norm": 2.0452367622832783,
      "learning_rate": 1.899331327781389e-05,
      "loss": 0.71,
      "step": 2653
    },
    {
      "epoch": 1.6980166346769034,
      "grad_norm": 2.651033802299597,
      "learning_rate": 1.8992406998448087e-05,
      "loss": 0.6965,
      "step": 2654
    },
    {
      "epoch": 1.6986564299424183,
      "grad_norm": 2.1111860893955074,
      "learning_rate": 1.8991500332964324e-05,
      "loss": 0.7078,
      "step": 2655
    },
    {
      "epoch": 1.6992962252079336,
      "grad_norm": 1.9151802460811236,
      "learning_rate": 1.899059328140153e-05,
      "loss": 0.6629,
      "step": 2656
    },
    {
      "epoch": 1.6999360204734484,
      "grad_norm": 2.2146428508054696,
      "learning_rate": 1.8989685843798658e-05,
      "loss": 0.6707,
      "step": 2657
    },
    {
      "epoch": 1.7005758157389637,
      "grad_norm": 2.2833760788304054,
      "learning_rate": 1.8988778020194668e-05,
      "loss": 0.7321,
      "step": 2658
    },
    {
      "epoch": 1.7012156110044785,
      "grad_norm": 2.1447702386064558,
      "learning_rate": 1.8987869810628535e-05,
      "loss": 0.7158,
      "step": 2659
    },
    {
      "epoch": 1.7018554062699938,
      "grad_norm": 2.124540263443763,
      "learning_rate": 1.8986961215139263e-05,
      "loss": 0.6843,
      "step": 2660
    },
    {
      "epoch": 1.7024952015355086,
      "grad_norm": 2.012289016603621,
      "learning_rate": 1.898605223376586e-05,
      "loss": 0.6443,
      "step": 2661
    },
    {
      "epoch": 1.7031349968010236,
      "grad_norm": 2.316142572095157,
      "learning_rate": 1.898514286654736e-05,
      "loss": 0.6453,
      "step": 2662
    },
    {
      "epoch": 1.7037747920665387,
      "grad_norm": 2.6134458948378403,
      "learning_rate": 1.8984233113522812e-05,
      "loss": 0.6835,
      "step": 2663
    },
    {
      "epoch": 1.7044145873320538,
      "grad_norm": 2.4195848402550504,
      "learning_rate": 1.8983322974731274e-05,
      "loss": 0.6618,
      "step": 2664
    },
    {
      "epoch": 1.7050543825975688,
      "grad_norm": 2.217559169153059,
      "learning_rate": 1.8982412450211827e-05,
      "loss": 0.6287,
      "step": 2665
    },
    {
      "epoch": 1.7056941778630839,
      "grad_norm": 2.376337529472403,
      "learning_rate": 1.898150154000357e-05,
      "loss": 0.6671,
      "step": 2666
    },
    {
      "epoch": 1.706333973128599,
      "grad_norm": 2.256167841359425,
      "learning_rate": 1.898059024414561e-05,
      "loss": 0.6412,
      "step": 2667
    },
    {
      "epoch": 1.7069737683941137,
      "grad_norm": 2.177138505193986,
      "learning_rate": 1.897967856267708e-05,
      "loss": 0.7184,
      "step": 2668
    },
    {
      "epoch": 1.707613563659629,
      "grad_norm": 2.021410914191311,
      "learning_rate": 1.897876649563713e-05,
      "loss": 0.6924,
      "step": 2669
    },
    {
      "epoch": 1.7082533589251438,
      "grad_norm": 2.416056452844107,
      "learning_rate": 1.8977854043064914e-05,
      "loss": 0.6875,
      "step": 2670
    },
    {
      "epoch": 1.7088931541906591,
      "grad_norm": 2.2736528752490663,
      "learning_rate": 1.897694120499962e-05,
      "loss": 0.6857,
      "step": 2671
    },
    {
      "epoch": 1.709532949456174,
      "grad_norm": 2.1520123748277293,
      "learning_rate": 1.8976027981480437e-05,
      "loss": 0.6581,
      "step": 2672
    },
    {
      "epoch": 1.710172744721689,
      "grad_norm": 2.18360886181758,
      "learning_rate": 1.897511437254658e-05,
      "loss": 0.7299,
      "step": 2673
    },
    {
      "epoch": 1.710812539987204,
      "grad_norm": 2.107788526548479,
      "learning_rate": 1.8974200378237276e-05,
      "loss": 0.7086,
      "step": 2674
    },
    {
      "epoch": 1.7114523352527191,
      "grad_norm": 1.8995789680724633,
      "learning_rate": 1.8973285998591776e-05,
      "loss": 0.6943,
      "step": 2675
    },
    {
      "epoch": 1.7120921305182342,
      "grad_norm": 1.9063269724614063,
      "learning_rate": 1.8972371233649335e-05,
      "loss": 0.6547,
      "step": 2676
    },
    {
      "epoch": 1.7127319257837492,
      "grad_norm": 2.1969470280426413,
      "learning_rate": 1.897145608344923e-05,
      "loss": 0.7116,
      "step": 2677
    },
    {
      "epoch": 1.7133717210492643,
      "grad_norm": 1.9007324502223972,
      "learning_rate": 1.8970540548030764e-05,
      "loss": 0.6868,
      "step": 2678
    },
    {
      "epoch": 1.714011516314779,
      "grad_norm": 2.257043863497813,
      "learning_rate": 1.8969624627433246e-05,
      "loss": 0.6293,
      "step": 2679
    },
    {
      "epoch": 1.7146513115802944,
      "grad_norm": 1.9569842382607372,
      "learning_rate": 1.8968708321696e-05,
      "loss": 0.6414,
      "step": 2680
    },
    {
      "epoch": 1.7152911068458092,
      "grad_norm": 2.122936780566575,
      "learning_rate": 1.896779163085837e-05,
      "loss": 0.6663,
      "step": 2681
    },
    {
      "epoch": 1.7159309021113245,
      "grad_norm": 2.3245754431655117,
      "learning_rate": 1.8966874554959722e-05,
      "loss": 0.6842,
      "step": 2682
    },
    {
      "epoch": 1.7165706973768393,
      "grad_norm": 2.2021803310789,
      "learning_rate": 1.8965957094039434e-05,
      "loss": 0.7371,
      "step": 2683
    },
    {
      "epoch": 1.7172104926423546,
      "grad_norm": 2.3421823806292466,
      "learning_rate": 1.8965039248136895e-05,
      "loss": 0.6543,
      "step": 2684
    },
    {
      "epoch": 1.7178502879078694,
      "grad_norm": 2.247539332759749,
      "learning_rate": 1.8964121017291518e-05,
      "loss": 0.7052,
      "step": 2685
    },
    {
      "epoch": 1.7184900831733845,
      "grad_norm": 2.2400902222627868,
      "learning_rate": 1.8963202401542732e-05,
      "loss": 0.6522,
      "step": 2686
    },
    {
      "epoch": 1.7191298784388995,
      "grad_norm": 2.141248373645578,
      "learning_rate": 1.8962283400929977e-05,
      "loss": 0.6754,
      "step": 2687
    },
    {
      "epoch": 1.7197696737044146,
      "grad_norm": 2.282338221959114,
      "learning_rate": 1.8961364015492716e-05,
      "loss": 0.6923,
      "step": 2688
    },
    {
      "epoch": 1.7204094689699296,
      "grad_norm": 2.1927117914679006,
      "learning_rate": 1.8960444245270427e-05,
      "loss": 0.7146,
      "step": 2689
    },
    {
      "epoch": 1.7210492642354447,
      "grad_norm": 2.16592731500152,
      "learning_rate": 1.89595240903026e-05,
      "loss": 0.6524,
      "step": 2690
    },
    {
      "epoch": 1.7216890595009597,
      "grad_norm": 2.1746089761950214,
      "learning_rate": 1.8958603550628746e-05,
      "loss": 0.7332,
      "step": 2691
    },
    {
      "epoch": 1.7223288547664746,
      "grad_norm": 2.1624786956680078,
      "learning_rate": 1.8957682626288393e-05,
      "loss": 0.6543,
      "step": 2692
    },
    {
      "epoch": 1.7229686500319898,
      "grad_norm": 2.1780329605157602,
      "learning_rate": 1.8956761317321084e-05,
      "loss": 0.6061,
      "step": 2693
    },
    {
      "epoch": 1.7236084452975047,
      "grad_norm": 2.2003199854534636,
      "learning_rate": 1.8955839623766374e-05,
      "loss": 0.7143,
      "step": 2694
    },
    {
      "epoch": 1.72424824056302,
      "grad_norm": 2.1807029870394667,
      "learning_rate": 1.895491754566384e-05,
      "loss": 0.6806,
      "step": 2695
    },
    {
      "epoch": 1.7248880358285348,
      "grad_norm": 2.3626750074230456,
      "learning_rate": 1.895399508305308e-05,
      "loss": 0.7303,
      "step": 2696
    },
    {
      "epoch": 1.72552783109405,
      "grad_norm": 2.1167313165535964,
      "learning_rate": 1.89530722359737e-05,
      "loss": 0.6585,
      "step": 2697
    },
    {
      "epoch": 1.7261676263595649,
      "grad_norm": 2.160295757242435,
      "learning_rate": 1.895214900446532e-05,
      "loss": 0.6814,
      "step": 2698
    },
    {
      "epoch": 1.72680742162508,
      "grad_norm": 2.3875686976110235,
      "learning_rate": 1.8951225388567587e-05,
      "loss": 0.7284,
      "step": 2699
    },
    {
      "epoch": 1.727447216890595,
      "grad_norm": 2.108693929763817,
      "learning_rate": 1.8950301388320164e-05,
      "loss": 0.7279,
      "step": 2700
    },
    {
      "epoch": 1.72808701215611,
      "grad_norm": 2.2642664166818185,
      "learning_rate": 1.8949377003762716e-05,
      "loss": 0.7169,
      "step": 2701
    },
    {
      "epoch": 1.728726807421625,
      "grad_norm": 2.0513802707278517,
      "learning_rate": 1.894845223493494e-05,
      "loss": 0.6205,
      "step": 2702
    },
    {
      "epoch": 1.7293666026871402,
      "grad_norm": 2.2358106954732926,
      "learning_rate": 1.8947527081876544e-05,
      "loss": 0.6354,
      "step": 2703
    },
    {
      "epoch": 1.7300063979526552,
      "grad_norm": 2.1034359145385784,
      "learning_rate": 1.894660154462725e-05,
      "loss": 0.6781,
      "step": 2704
    },
    {
      "epoch": 1.73064619321817,
      "grad_norm": 2.3185261478357573,
      "learning_rate": 1.89456756232268e-05,
      "loss": 0.6702,
      "step": 2705
    },
    {
      "epoch": 1.7312859884836853,
      "grad_norm": 2.044212651434673,
      "learning_rate": 1.8944749317714956e-05,
      "loss": 0.6956,
      "step": 2706
    },
    {
      "epoch": 1.7319257837492001,
      "grad_norm": 2.2406732342802216,
      "learning_rate": 1.894382262813148e-05,
      "loss": 0.6901,
      "step": 2707
    },
    {
      "epoch": 1.7325655790147154,
      "grad_norm": 2.214023610269728,
      "learning_rate": 1.8942895554516178e-05,
      "loss": 0.6537,
      "step": 2708
    },
    {
      "epoch": 1.7332053742802302,
      "grad_norm": 2.060768414917095,
      "learning_rate": 1.8941968096908848e-05,
      "loss": 0.6705,
      "step": 2709
    },
    {
      "epoch": 1.7338451695457455,
      "grad_norm": 2.1764299665143967,
      "learning_rate": 1.8941040255349313e-05,
      "loss": 0.779,
      "step": 2710
    },
    {
      "epoch": 1.7344849648112604,
      "grad_norm": 2.0152017961692565,
      "learning_rate": 1.8940112029877415e-05,
      "loss": 0.6768,
      "step": 2711
    },
    {
      "epoch": 1.7351247600767754,
      "grad_norm": 2.300485016260876,
      "learning_rate": 1.8939183420533007e-05,
      "loss": 0.7036,
      "step": 2712
    },
    {
      "epoch": 1.7357645553422905,
      "grad_norm": 2.370335463592278,
      "learning_rate": 1.8938254427355964e-05,
      "loss": 0.7234,
      "step": 2713
    },
    {
      "epoch": 1.7364043506078055,
      "grad_norm": 2.187545128283072,
      "learning_rate": 1.8937325050386176e-05,
      "loss": 0.66,
      "step": 2714
    },
    {
      "epoch": 1.7370441458733206,
      "grad_norm": 2.2823829821501462,
      "learning_rate": 1.893639528966355e-05,
      "loss": 0.678,
      "step": 2715
    },
    {
      "epoch": 1.7376839411388356,
      "grad_norm": 2.394861101944455,
      "learning_rate": 1.893546514522801e-05,
      "loss": 0.6765,
      "step": 2716
    },
    {
      "epoch": 1.7383237364043507,
      "grad_norm": 2.3315373749441317,
      "learning_rate": 1.893453461711949e-05,
      "loss": 0.7216,
      "step": 2717
    },
    {
      "epoch": 1.7389635316698655,
      "grad_norm": 1.892333451420421,
      "learning_rate": 1.8933603705377945e-05,
      "loss": 0.6956,
      "step": 2718
    },
    {
      "epoch": 1.7396033269353808,
      "grad_norm": 2.4481043545947294,
      "learning_rate": 1.893267241004335e-05,
      "loss": 0.7182,
      "step": 2719
    },
    {
      "epoch": 1.7402431222008956,
      "grad_norm": 2.3468672751434903,
      "learning_rate": 1.893174073115569e-05,
      "loss": 0.6805,
      "step": 2720
    },
    {
      "epoch": 1.7408829174664109,
      "grad_norm": 2.5969287834411396,
      "learning_rate": 1.8930808668754972e-05,
      "loss": 0.677,
      "step": 2721
    },
    {
      "epoch": 1.7415227127319257,
      "grad_norm": 2.027440393083271,
      "learning_rate": 1.8929876222881217e-05,
      "loss": 0.6696,
      "step": 2722
    },
    {
      "epoch": 1.7421625079974408,
      "grad_norm": 2.3603760091887453,
      "learning_rate": 1.8928943393574464e-05,
      "loss": 0.6939,
      "step": 2723
    },
    {
      "epoch": 1.7428023032629558,
      "grad_norm": 2.234340132814777,
      "learning_rate": 1.892801018087476e-05,
      "loss": 0.7073,
      "step": 2724
    },
    {
      "epoch": 1.7434420985284709,
      "grad_norm": 2.421248445714457,
      "learning_rate": 1.8927076584822187e-05,
      "loss": 0.7343,
      "step": 2725
    },
    {
      "epoch": 1.744081893793986,
      "grad_norm": 2.1598394961427907,
      "learning_rate": 1.892614260545682e-05,
      "loss": 0.717,
      "step": 2726
    },
    {
      "epoch": 1.744721689059501,
      "grad_norm": 2.25304664636013,
      "learning_rate": 1.892520824281877e-05,
      "loss": 0.6885,
      "step": 2727
    },
    {
      "epoch": 1.745361484325016,
      "grad_norm": 2.38799932249825,
      "learning_rate": 1.8924273496948156e-05,
      "loss": 0.647,
      "step": 2728
    },
    {
      "epoch": 1.7460012795905309,
      "grad_norm": 2.1140195808497135,
      "learning_rate": 1.892333836788511e-05,
      "loss": 0.6681,
      "step": 2729
    },
    {
      "epoch": 1.7466410748560461,
      "grad_norm": 2.410618645973256,
      "learning_rate": 1.892240285566979e-05,
      "loss": 0.7394,
      "step": 2730
    },
    {
      "epoch": 1.747280870121561,
      "grad_norm": 2.167678095834506,
      "learning_rate": 1.8921466960342363e-05,
      "loss": 0.6828,
      "step": 2731
    },
    {
      "epoch": 1.7479206653870762,
      "grad_norm": 2.2074997906824683,
      "learning_rate": 1.8920530681943016e-05,
      "loss": 0.6758,
      "step": 2732
    },
    {
      "epoch": 1.748560460652591,
      "grad_norm": 2.132351200515226,
      "learning_rate": 1.8919594020511948e-05,
      "loss": 0.7412,
      "step": 2733
    },
    {
      "epoch": 1.7492002559181064,
      "grad_norm": 2.199651111670204,
      "learning_rate": 1.891865697608938e-05,
      "loss": 0.6924,
      "step": 2734
    },
    {
      "epoch": 1.7498400511836212,
      "grad_norm": 2.118610813935172,
      "learning_rate": 1.8917719548715545e-05,
      "loss": 0.6697,
      "step": 2735
    },
    {
      "epoch": 1.7504798464491362,
      "grad_norm": 2.268147374540476,
      "learning_rate": 1.8916781738430695e-05,
      "loss": 0.6597,
      "step": 2736
    },
    {
      "epoch": 1.7511196417146513,
      "grad_norm": 2.034600305505386,
      "learning_rate": 1.8915843545275102e-05,
      "loss": 0.6863,
      "step": 2737
    },
    {
      "epoch": 1.7517594369801663,
      "grad_norm": 2.073980337122526,
      "learning_rate": 1.891490496928905e-05,
      "loss": 0.7031,
      "step": 2738
    },
    {
      "epoch": 1.7523992322456814,
      "grad_norm": 2.1287459022288693,
      "learning_rate": 1.8913966010512833e-05,
      "loss": 0.6879,
      "step": 2739
    },
    {
      "epoch": 1.7530390275111964,
      "grad_norm": 2.0790685790270627,
      "learning_rate": 1.8913026668986772e-05,
      "loss": 0.688,
      "step": 2740
    },
    {
      "epoch": 1.7536788227767115,
      "grad_norm": 2.302172599379909,
      "learning_rate": 1.89120869447512e-05,
      "loss": 0.6657,
      "step": 2741
    },
    {
      "epoch": 1.7543186180422263,
      "grad_norm": 2.071057415447354,
      "learning_rate": 1.891114683784647e-05,
      "loss": 0.6855,
      "step": 2742
    },
    {
      "epoch": 1.7549584133077416,
      "grad_norm": 2.119791122856047,
      "learning_rate": 1.8910206348312945e-05,
      "loss": 0.7212,
      "step": 2743
    },
    {
      "epoch": 1.7555982085732564,
      "grad_norm": 2.242772486727632,
      "learning_rate": 1.8909265476191013e-05,
      "loss": 0.6961,
      "step": 2744
    },
    {
      "epoch": 1.7562380038387717,
      "grad_norm": 1.9453158014028509,
      "learning_rate": 1.8908324221521066e-05,
      "loss": 0.7088,
      "step": 2745
    },
    {
      "epoch": 1.7568777991042865,
      "grad_norm": 2.1482300159077434,
      "learning_rate": 1.8907382584343526e-05,
      "loss": 0.6913,
      "step": 2746
    },
    {
      "epoch": 1.7575175943698018,
      "grad_norm": 1.9990524126989992,
      "learning_rate": 1.890644056469882e-05,
      "loss": 0.6955,
      "step": 2747
    },
    {
      "epoch": 1.7581573896353166,
      "grad_norm": 2.121695896032684,
      "learning_rate": 1.8905498162627404e-05,
      "loss": 0.6401,
      "step": 2748
    },
    {
      "epoch": 1.7587971849008317,
      "grad_norm": 2.089873032345985,
      "learning_rate": 1.8904555378169733e-05,
      "loss": 0.7505,
      "step": 2749
    },
    {
      "epoch": 1.7594369801663468,
      "grad_norm": 2.110527763201646,
      "learning_rate": 1.8903612211366297e-05,
      "loss": 0.712,
      "step": 2750
    },
    {
      "epoch": 1.7600767754318618,
      "grad_norm": 2.531364490796449,
      "learning_rate": 1.890266866225759e-05,
      "loss": 0.6934,
      "step": 2751
    },
    {
      "epoch": 1.7607165706973769,
      "grad_norm": 2.0957752282575535,
      "learning_rate": 1.8901724730884128e-05,
      "loss": 0.6686,
      "step": 2752
    },
    {
      "epoch": 1.761356365962892,
      "grad_norm": 2.094828338249946,
      "learning_rate": 1.890078041728644e-05,
      "loss": 0.7277,
      "step": 2753
    },
    {
      "epoch": 1.761996161228407,
      "grad_norm": 2.0839661310931787,
      "learning_rate": 1.8899835721505074e-05,
      "loss": 0.7097,
      "step": 2754
    },
    {
      "epoch": 1.7626359564939218,
      "grad_norm": 2.1569285612828413,
      "learning_rate": 1.8898890643580596e-05,
      "loss": 0.6737,
      "step": 2755
    },
    {
      "epoch": 1.763275751759437,
      "grad_norm": 2.042780045972228,
      "learning_rate": 1.8897945183553585e-05,
      "loss": 0.6743,
      "step": 2756
    },
    {
      "epoch": 1.763915547024952,
      "grad_norm": 2.014514904394886,
      "learning_rate": 1.8896999341464632e-05,
      "loss": 0.6888,
      "step": 2757
    },
    {
      "epoch": 1.7645553422904672,
      "grad_norm": 1.9581313891277725,
      "learning_rate": 1.8896053117354357e-05,
      "loss": 0.6503,
      "step": 2758
    },
    {
      "epoch": 1.765195137555982,
      "grad_norm": 2.336083977230003,
      "learning_rate": 1.8895106511263383e-05,
      "loss": 0.6477,
      "step": 2759
    },
    {
      "epoch": 1.7658349328214973,
      "grad_norm": 1.9763280548715043,
      "learning_rate": 1.889415952323236e-05,
      "loss": 0.6418,
      "step": 2760
    },
    {
      "epoch": 1.7664747280870121,
      "grad_norm": 2.1636864910103246,
      "learning_rate": 1.889321215330195e-05,
      "loss": 0.6664,
      "step": 2761
    },
    {
      "epoch": 1.7671145233525272,
      "grad_norm": 2.2930924455360877,
      "learning_rate": 1.8892264401512825e-05,
      "loss": 0.7368,
      "step": 2762
    },
    {
      "epoch": 1.7677543186180422,
      "grad_norm": 2.4977649407356037,
      "learning_rate": 1.889131626790569e-05,
      "loss": 0.6689,
      "step": 2763
    },
    {
      "epoch": 1.7683941138835573,
      "grad_norm": 2.1072124797408573,
      "learning_rate": 1.889036775252125e-05,
      "loss": 0.6711,
      "step": 2764
    },
    {
      "epoch": 1.7690339091490723,
      "grad_norm": 2.1150980393176373,
      "learning_rate": 1.8889418855400232e-05,
      "loss": 0.644,
      "step": 2765
    },
    {
      "epoch": 1.7696737044145874,
      "grad_norm": 2.2840951878388696,
      "learning_rate": 1.888846957658338e-05,
      "loss": 0.6911,
      "step": 2766
    },
    {
      "epoch": 1.7703134996801024,
      "grad_norm": 2.064190496111294,
      "learning_rate": 1.888751991611146e-05,
      "loss": 0.6899,
      "step": 2767
    },
    {
      "epoch": 1.7709532949456173,
      "grad_norm": 2.087763337985181,
      "learning_rate": 1.888656987402524e-05,
      "loss": 0.6551,
      "step": 2768
    },
    {
      "epoch": 1.7715930902111325,
      "grad_norm": 2.135705483980106,
      "learning_rate": 1.888561945036552e-05,
      "loss": 0.6313,
      "step": 2769
    },
    {
      "epoch": 1.7722328854766474,
      "grad_norm": 2.1369461843654847,
      "learning_rate": 1.8884668645173105e-05,
      "loss": 0.639,
      "step": 2770
    },
    {
      "epoch": 1.7728726807421626,
      "grad_norm": 2.2157749267452065,
      "learning_rate": 1.8883717458488828e-05,
      "loss": 0.6924,
      "step": 2771
    },
    {
      "epoch": 1.7735124760076775,
      "grad_norm": 2.3356933219140443,
      "learning_rate": 1.888276589035352e-05,
      "loss": 0.6924,
      "step": 2772
    },
    {
      "epoch": 1.7741522712731925,
      "grad_norm": 2.1892333225783145,
      "learning_rate": 1.8881813940808048e-05,
      "loss": 0.6495,
      "step": 2773
    },
    {
      "epoch": 1.7747920665387076,
      "grad_norm": 1.981947568686018,
      "learning_rate": 1.8880861609893288e-05,
      "loss": 0.6742,
      "step": 2774
    },
    {
      "epoch": 1.7754318618042226,
      "grad_norm": 2.2318755759839273,
      "learning_rate": 1.8879908897650126e-05,
      "loss": 0.7007,
      "step": 2775
    },
    {
      "epoch": 1.7760716570697377,
      "grad_norm": 2.1112634092310407,
      "learning_rate": 1.887895580411947e-05,
      "loss": 0.65,
      "step": 2776
    },
    {
      "epoch": 1.7767114523352527,
      "grad_norm": 2.016495215656827,
      "learning_rate": 1.887800232934225e-05,
      "loss": 0.6439,
      "step": 2777
    },
    {
      "epoch": 1.7773512476007678,
      "grad_norm": 2.269186181573367,
      "learning_rate": 1.88770484733594e-05,
      "loss": 0.6388,
      "step": 2778
    },
    {
      "epoch": 1.7779910428662828,
      "grad_norm": 2.154696434133241,
      "learning_rate": 1.8876094236211882e-05,
      "loss": 0.7067,
      "step": 2779
    },
    {
      "epoch": 1.778630838131798,
      "grad_norm": 2.0281722065263907,
      "learning_rate": 1.8875139617940664e-05,
      "loss": 0.6365,
      "step": 2780
    },
    {
      "epoch": 1.7792706333973127,
      "grad_norm": 2.1395348040455184,
      "learning_rate": 1.887418461858674e-05,
      "loss": 0.7518,
      "step": 2781
    },
    {
      "epoch": 1.779910428662828,
      "grad_norm": 2.069262385262918,
      "learning_rate": 1.8873229238191116e-05,
      "loss": 0.6736,
      "step": 2782
    },
    {
      "epoch": 1.7805502239283428,
      "grad_norm": 2.0820693424529626,
      "learning_rate": 1.887227347679481e-05,
      "loss": 0.6691,
      "step": 2783
    },
    {
      "epoch": 1.781190019193858,
      "grad_norm": 2.263891239512507,
      "learning_rate": 1.8871317334438863e-05,
      "loss": 0.6907,
      "step": 2784
    },
    {
      "epoch": 1.781829814459373,
      "grad_norm": 2.068597397420223,
      "learning_rate": 1.8870360811164333e-05,
      "loss": 0.6544,
      "step": 2785
    },
    {
      "epoch": 1.782469609724888,
      "grad_norm": 2.007068223148545,
      "learning_rate": 1.886940390701229e-05,
      "loss": 0.6783,
      "step": 2786
    },
    {
      "epoch": 1.783109404990403,
      "grad_norm": 2.253053120196439,
      "learning_rate": 1.8868446622023818e-05,
      "loss": 0.6704,
      "step": 2787
    },
    {
      "epoch": 1.783749200255918,
      "grad_norm": 2.183648009544298,
      "learning_rate": 1.8867488956240023e-05,
      "loss": 0.74,
      "step": 2788
    },
    {
      "epoch": 1.7843889955214332,
      "grad_norm": 2.0614713626359022,
      "learning_rate": 1.8866530909702027e-05,
      "loss": 0.6797,
      "step": 2789
    },
    {
      "epoch": 1.7850287907869482,
      "grad_norm": 2.203496158989207,
      "learning_rate": 1.8865572482450967e-05,
      "loss": 0.6077,
      "step": 2790
    },
    {
      "epoch": 1.7856685860524633,
      "grad_norm": 2.129935326423506,
      "learning_rate": 1.8864613674527997e-05,
      "loss": 0.6647,
      "step": 2791
    },
    {
      "epoch": 1.786308381317978,
      "grad_norm": 2.380399704073007,
      "learning_rate": 1.8863654485974282e-05,
      "loss": 0.6327,
      "step": 2792
    },
    {
      "epoch": 1.7869481765834934,
      "grad_norm": 2.219782220164566,
      "learning_rate": 1.8862694916831012e-05,
      "loss": 0.6714,
      "step": 2793
    },
    {
      "epoch": 1.7875879718490082,
      "grad_norm": 2.0729180223604864,
      "learning_rate": 1.8861734967139386e-05,
      "loss": 0.6436,
      "step": 2794
    },
    {
      "epoch": 1.7882277671145235,
      "grad_norm": 2.362142858350717,
      "learning_rate": 1.8860774636940627e-05,
      "loss": 0.6768,
      "step": 2795
    },
    {
      "epoch": 1.7888675623800383,
      "grad_norm": 2.361367253320665,
      "learning_rate": 1.8859813926275963e-05,
      "loss": 0.7056,
      "step": 2796
    },
    {
      "epoch": 1.7895073576455536,
      "grad_norm": 2.078363497987319,
      "learning_rate": 1.885885283518665e-05,
      "loss": 0.6716,
      "step": 2797
    },
    {
      "epoch": 1.7901471529110684,
      "grad_norm": 2.340238475937407,
      "learning_rate": 1.8857891363713962e-05,
      "loss": 0.7226,
      "step": 2798
    },
    {
      "epoch": 1.7907869481765835,
      "grad_norm": 2.073615841828408,
      "learning_rate": 1.885692951189917e-05,
      "loss": 0.7225,
      "step": 2799
    },
    {
      "epoch": 1.7914267434420985,
      "grad_norm": 2.1759212817495355,
      "learning_rate": 1.8855967279783578e-05,
      "loss": 0.6497,
      "step": 2800
    },
    {
      "epoch": 1.7920665387076136,
      "grad_norm": 2.0154252671592,
      "learning_rate": 1.885500466740851e-05,
      "loss": 0.6756,
      "step": 2801
    },
    {
      "epoch": 1.7927063339731286,
      "grad_norm": 2.0888981276237777,
      "learning_rate": 1.8854041674815287e-05,
      "loss": 0.6788,
      "step": 2802
    },
    {
      "epoch": 1.7933461292386437,
      "grad_norm": 2.0077543669112776,
      "learning_rate": 1.885307830204527e-05,
      "loss": 0.682,
      "step": 2803
    },
    {
      "epoch": 1.7939859245041587,
      "grad_norm": 2.1116244885786872,
      "learning_rate": 1.885211454913982e-05,
      "loss": 0.6787,
      "step": 2804
    },
    {
      "epoch": 1.7946257197696736,
      "grad_norm": 2.0344915178158747,
      "learning_rate": 1.8851150416140314e-05,
      "loss": 0.5964,
      "step": 2805
    },
    {
      "epoch": 1.7952655150351888,
      "grad_norm": 2.5341635952255404,
      "learning_rate": 1.8850185903088155e-05,
      "loss": 0.6065,
      "step": 2806
    },
    {
      "epoch": 1.7959053103007037,
      "grad_norm": 2.174749372959101,
      "learning_rate": 1.8849221010024757e-05,
      "loss": 0.6695,
      "step": 2807
    },
    {
      "epoch": 1.796545105566219,
      "grad_norm": 2.060799537026917,
      "learning_rate": 1.884825573699155e-05,
      "loss": 0.6646,
      "step": 2808
    },
    {
      "epoch": 1.7971849008317338,
      "grad_norm": 2.1614050043531865,
      "learning_rate": 1.8847290084029982e-05,
      "loss": 0.6157,
      "step": 2809
    },
    {
      "epoch": 1.797824696097249,
      "grad_norm": 2.338695533042663,
      "learning_rate": 1.8846324051181514e-05,
      "loss": 0.7109,
      "step": 2810
    },
    {
      "epoch": 1.7984644913627639,
      "grad_norm": 2.138831073674361,
      "learning_rate": 1.884535763848763e-05,
      "loss": 0.682,
      "step": 2811
    },
    {
      "epoch": 1.799104286628279,
      "grad_norm": 2.0440916556673656,
      "learning_rate": 1.8844390845989824e-05,
      "loss": 0.6435,
      "step": 2812
    },
    {
      "epoch": 1.799744081893794,
      "grad_norm": 2.107317784418035,
      "learning_rate": 1.884342367372961e-05,
      "loss": 0.6394,
      "step": 2813
    },
    {
      "epoch": 1.800383877159309,
      "grad_norm": 2.20602380052985,
      "learning_rate": 1.884245612174851e-05,
      "loss": 0.7019,
      "step": 2814
    },
    {
      "epoch": 1.801023672424824,
      "grad_norm": 1.9773579314758138,
      "learning_rate": 1.8841488190088074e-05,
      "loss": 0.6924,
      "step": 2815
    },
    {
      "epoch": 1.8016634676903391,
      "grad_norm": 1.9367164107010717,
      "learning_rate": 1.8840519878789868e-05,
      "loss": 0.5786,
      "step": 2816
    },
    {
      "epoch": 1.8023032629558542,
      "grad_norm": 2.119794809621702,
      "learning_rate": 1.8839551187895456e-05,
      "loss": 0.6869,
      "step": 2817
    },
    {
      "epoch": 1.802943058221369,
      "grad_norm": 2.175628161779303,
      "learning_rate": 1.8838582117446448e-05,
      "loss": 0.6687,
      "step": 2818
    },
    {
      "epoch": 1.8035828534868843,
      "grad_norm": 2.0446474599157867,
      "learning_rate": 1.8837612667484443e-05,
      "loss": 0.715,
      "step": 2819
    },
    {
      "epoch": 1.8042226487523991,
      "grad_norm": 2.2041070898237574,
      "learning_rate": 1.883664283805107e-05,
      "loss": 0.6463,
      "step": 2820
    },
    {
      "epoch": 1.8048624440179144,
      "grad_norm": 1.9492276067861003,
      "learning_rate": 1.883567262918798e-05,
      "loss": 0.6813,
      "step": 2821
    },
    {
      "epoch": 1.8055022392834292,
      "grad_norm": 2.1006341506031494,
      "learning_rate": 1.8834702040936818e-05,
      "loss": 0.7327,
      "step": 2822
    },
    {
      "epoch": 1.8061420345489443,
      "grad_norm": 2.065615664686621,
      "learning_rate": 1.8833731073339264e-05,
      "loss": 0.6719,
      "step": 2823
    },
    {
      "epoch": 1.8067818298144593,
      "grad_norm": 2.129698483299012,
      "learning_rate": 1.8832759726437016e-05,
      "loss": 0.6363,
      "step": 2824
    },
    {
      "epoch": 1.8074216250799744,
      "grad_norm": 2.0209443017168836,
      "learning_rate": 1.8831788000271774e-05,
      "loss": 0.6564,
      "step": 2825
    },
    {
      "epoch": 1.8080614203454894,
      "grad_norm": 2.278885137756275,
      "learning_rate": 1.8830815894885268e-05,
      "loss": 0.6331,
      "step": 2826
    },
    {
      "epoch": 1.8087012156110045,
      "grad_norm": 2.0818056216351377,
      "learning_rate": 1.8829843410319238e-05,
      "loss": 0.7009,
      "step": 2827
    },
    {
      "epoch": 1.8093410108765196,
      "grad_norm": 2.0883617595020265,
      "learning_rate": 1.8828870546615436e-05,
      "loss": 0.6093,
      "step": 2828
    },
    {
      "epoch": 1.8099808061420346,
      "grad_norm": 2.243454546580723,
      "learning_rate": 1.8827897303815643e-05,
      "loss": 0.6616,
      "step": 2829
    },
    {
      "epoch": 1.8106206014075497,
      "grad_norm": 2.270376935894803,
      "learning_rate": 1.882692368196164e-05,
      "loss": 0.6319,
      "step": 2830
    },
    {
      "epoch": 1.8112603966730645,
      "grad_norm": 2.2919236932470977,
      "learning_rate": 1.8825949681095235e-05,
      "loss": 0.6451,
      "step": 2831
    },
    {
      "epoch": 1.8119001919385798,
      "grad_norm": 2.536961828024218,
      "learning_rate": 1.882497530125825e-05,
      "loss": 0.6675,
      "step": 2832
    },
    {
      "epoch": 1.8125399872040946,
      "grad_norm": 2.443430895486524,
      "learning_rate": 1.8824000542492525e-05,
      "loss": 0.7252,
      "step": 2833
    },
    {
      "epoch": 1.8131797824696099,
      "grad_norm": 2.233257977974462,
      "learning_rate": 1.8823025404839916e-05,
      "loss": 0.6758,
      "step": 2834
    },
    {
      "epoch": 1.8138195777351247,
      "grad_norm": 2.3232965048156142,
      "learning_rate": 1.882204988834229e-05,
      "loss": 0.6717,
      "step": 2835
    },
    {
      "epoch": 1.8144593730006398,
      "grad_norm": 1.9974870879052653,
      "learning_rate": 1.8821073993041534e-05,
      "loss": 0.6592,
      "step": 2836
    },
    {
      "epoch": 1.8150991682661548,
      "grad_norm": 2.610270093104226,
      "learning_rate": 1.882009771897955e-05,
      "loss": 0.6516,
      "step": 2837
    },
    {
      "epoch": 1.8157389635316699,
      "grad_norm": 2.1308623107950537,
      "learning_rate": 1.8819121066198262e-05,
      "loss": 0.6329,
      "step": 2838
    },
    {
      "epoch": 1.816378758797185,
      "grad_norm": 2.0297958886235103,
      "learning_rate": 1.8818144034739605e-05,
      "loss": 0.6344,
      "step": 2839
    },
    {
      "epoch": 1.8170185540627,
      "grad_norm": 2.286414630362289,
      "learning_rate": 1.8817166624645528e-05,
      "loss": 0.7265,
      "step": 2840
    },
    {
      "epoch": 1.817658349328215,
      "grad_norm": 1.913684597134493,
      "learning_rate": 1.8816188835958003e-05,
      "loss": 0.7239,
      "step": 2841
    },
    {
      "epoch": 1.8182981445937298,
      "grad_norm": 2.0483139460449187,
      "learning_rate": 1.881521066871901e-05,
      "loss": 0.6192,
      "step": 2842
    },
    {
      "epoch": 1.8189379398592451,
      "grad_norm": 2.054564098752808,
      "learning_rate": 1.881423212297055e-05,
      "loss": 0.6587,
      "step": 2843
    },
    {
      "epoch": 1.81957773512476,
      "grad_norm": 1.9628459613062912,
      "learning_rate": 1.8813253198754644e-05,
      "loss": 0.7059,
      "step": 2844
    },
    {
      "epoch": 1.8202175303902752,
      "grad_norm": 2.184627264017388,
      "learning_rate": 1.881227389611332e-05,
      "loss": 0.6171,
      "step": 2845
    },
    {
      "epoch": 1.82085732565579,
      "grad_norm": 2.2230777497604435,
      "learning_rate": 1.8811294215088636e-05,
      "loss": 0.6609,
      "step": 2846
    },
    {
      "epoch": 1.8214971209213053,
      "grad_norm": 2.1016952529713633,
      "learning_rate": 1.881031415572265e-05,
      "loss": 0.6945,
      "step": 2847
    },
    {
      "epoch": 1.8221369161868202,
      "grad_norm": 2.083370482573428,
      "learning_rate": 1.8809333718057446e-05,
      "loss": 0.6628,
      "step": 2848
    },
    {
      "epoch": 1.8227767114523352,
      "grad_norm": 2.215038084739677,
      "learning_rate": 1.8808352902135124e-05,
      "loss": 0.6818,
      "step": 2849
    },
    {
      "epoch": 1.8234165067178503,
      "grad_norm": 2.3449136054665196,
      "learning_rate": 1.8807371707997798e-05,
      "loss": 0.6827,
      "step": 2850
    },
    {
      "epoch": 1.8240563019833653,
      "grad_norm": 2.0411833663608387,
      "learning_rate": 1.8806390135687596e-05,
      "loss": 0.659,
      "step": 2851
    },
    {
      "epoch": 1.8246960972488804,
      "grad_norm": 2.571528313334715,
      "learning_rate": 1.8805408185246668e-05,
      "loss": 0.7119,
      "step": 2852
    },
    {
      "epoch": 1.8253358925143954,
      "grad_norm": 2.1883734734639164,
      "learning_rate": 1.8804425856717174e-05,
      "loss": 0.7116,
      "step": 2853
    },
    {
      "epoch": 1.8259756877799105,
      "grad_norm": 2.1117125841019484,
      "learning_rate": 1.8803443150141297e-05,
      "loss": 0.6237,
      "step": 2854
    },
    {
      "epoch": 1.8266154830454253,
      "grad_norm": 2.028125178587822,
      "learning_rate": 1.8802460065561233e-05,
      "loss": 0.6857,
      "step": 2855
    },
    {
      "epoch": 1.8272552783109406,
      "grad_norm": 2.1149311142465232,
      "learning_rate": 1.880147660301919e-05,
      "loss": 0.7253,
      "step": 2856
    },
    {
      "epoch": 1.8278950735764554,
      "grad_norm": 1.9620645461180461,
      "learning_rate": 1.8800492762557398e-05,
      "loss": 0.661,
      "step": 2857
    },
    {
      "epoch": 1.8285348688419707,
      "grad_norm": 2.146097052815693,
      "learning_rate": 1.8799508544218103e-05,
      "loss": 0.7289,
      "step": 2858
    },
    {
      "epoch": 1.8291746641074855,
      "grad_norm": 2.026803812573518,
      "learning_rate": 1.8798523948043563e-05,
      "loss": 0.6554,
      "step": 2859
    },
    {
      "epoch": 1.8298144593730008,
      "grad_norm": 2.136297877892902,
      "learning_rate": 1.8797538974076056e-05,
      "loss": 0.6111,
      "step": 2860
    },
    {
      "epoch": 1.8304542546385156,
      "grad_norm": 2.117544868605762,
      "learning_rate": 1.879655362235788e-05,
      "loss": 0.6441,
      "step": 2861
    },
    {
      "epoch": 1.8310940499040307,
      "grad_norm": 2.2239139348263812,
      "learning_rate": 1.879556789293133e-05,
      "loss": 0.6859,
      "step": 2862
    },
    {
      "epoch": 1.8317338451695457,
      "grad_norm": 2.2540561034597895,
      "learning_rate": 1.8794581785838746e-05,
      "loss": 0.6837,
      "step": 2863
    },
    {
      "epoch": 1.8323736404350608,
      "grad_norm": 2.293636007782129,
      "learning_rate": 1.8793595301122468e-05,
      "loss": 0.6652,
      "step": 2864
    },
    {
      "epoch": 1.8330134357005758,
      "grad_norm": 2.402518184006849,
      "learning_rate": 1.8792608438824845e-05,
      "loss": 0.6734,
      "step": 2865
    },
    {
      "epoch": 1.833653230966091,
      "grad_norm": 2.141014330688177,
      "learning_rate": 1.879162119898826e-05,
      "loss": 0.6953,
      "step": 2866
    },
    {
      "epoch": 1.834293026231606,
      "grad_norm": 2.2830837473638828,
      "learning_rate": 1.8790633581655095e-05,
      "loss": 0.681,
      "step": 2867
    },
    {
      "epoch": 1.8349328214971208,
      "grad_norm": 2.544048025663203,
      "learning_rate": 1.8789645586867764e-05,
      "loss": 0.6938,
      "step": 2868
    },
    {
      "epoch": 1.835572616762636,
      "grad_norm": 2.3887854893980895,
      "learning_rate": 1.8788657214668687e-05,
      "loss": 0.6506,
      "step": 2869
    },
    {
      "epoch": 1.8362124120281509,
      "grad_norm": 2.1642552224373133,
      "learning_rate": 1.8787668465100305e-05,
      "loss": 0.6675,
      "step": 2870
    },
    {
      "epoch": 1.8368522072936662,
      "grad_norm": 2.0647316693192552,
      "learning_rate": 1.878667933820507e-05,
      "loss": 0.5877,
      "step": 2871
    },
    {
      "epoch": 1.837492002559181,
      "grad_norm": 2.3376863801483476,
      "learning_rate": 1.8785689834025454e-05,
      "loss": 0.5991,
      "step": 2872
    },
    {
      "epoch": 1.838131797824696,
      "grad_norm": 2.3619543840568986,
      "learning_rate": 1.8784699952603943e-05,
      "loss": 0.6894,
      "step": 2873
    },
    {
      "epoch": 1.838771593090211,
      "grad_norm": 2.464412478380634,
      "learning_rate": 1.8783709693983044e-05,
      "loss": 0.6745,
      "step": 2874
    },
    {
      "epoch": 1.8394113883557262,
      "grad_norm": 2.271062143540166,
      "learning_rate": 1.878271905820528e-05,
      "loss": 0.6501,
      "step": 2875
    },
    {
      "epoch": 1.8400511836212412,
      "grad_norm": 2.0530131283786806,
      "learning_rate": 1.878172804531318e-05,
      "loss": 0.6263,
      "step": 2876
    },
    {
      "epoch": 1.8406909788867563,
      "grad_norm": 2.3036019713575047,
      "learning_rate": 1.8780736655349296e-05,
      "loss": 0.6407,
      "step": 2877
    },
    {
      "epoch": 1.8413307741522713,
      "grad_norm": 2.322440190387237,
      "learning_rate": 1.87797448883562e-05,
      "loss": 0.7063,
      "step": 2878
    },
    {
      "epoch": 1.8419705694177864,
      "grad_norm": 2.0582395937985862,
      "learning_rate": 1.8778752744376483e-05,
      "loss": 0.5971,
      "step": 2879
    },
    {
      "epoch": 1.8426103646833014,
      "grad_norm": 2.130931882830759,
      "learning_rate": 1.8777760223452735e-05,
      "loss": 0.6536,
      "step": 2880
    },
    {
      "epoch": 1.8432501599488162,
      "grad_norm": 2.2205308661947205,
      "learning_rate": 1.877676732562758e-05,
      "loss": 0.6638,
      "step": 2881
    },
    {
      "epoch": 1.8438899552143315,
      "grad_norm": 2.0195644300223137,
      "learning_rate": 1.8775774050943644e-05,
      "loss": 0.6684,
      "step": 2882
    },
    {
      "epoch": 1.8445297504798464,
      "grad_norm": 2.2006208323689638,
      "learning_rate": 1.877478039944358e-05,
      "loss": 0.6479,
      "step": 2883
    },
    {
      "epoch": 1.8451695457453616,
      "grad_norm": 2.320567637851761,
      "learning_rate": 1.8773786371170063e-05,
      "loss": 0.7129,
      "step": 2884
    },
    {
      "epoch": 1.8458093410108765,
      "grad_norm": 2.1565537411702493,
      "learning_rate": 1.877279196616576e-05,
      "loss": 0.673,
      "step": 2885
    },
    {
      "epoch": 1.8464491362763915,
      "grad_norm": 2.1956544161678577,
      "learning_rate": 1.8771797184473378e-05,
      "loss": 0.6946,
      "step": 2886
    },
    {
      "epoch": 1.8470889315419066,
      "grad_norm": 1.962900308178444,
      "learning_rate": 1.8770802026135627e-05,
      "loss": 0.6455,
      "step": 2887
    },
    {
      "epoch": 1.8477287268074216,
      "grad_norm": 1.9274870274161553,
      "learning_rate": 1.876980649119524e-05,
      "loss": 0.6551,
      "step": 2888
    },
    {
      "epoch": 1.8483685220729367,
      "grad_norm": 2.1341968987375055,
      "learning_rate": 1.8768810579694966e-05,
      "loss": 0.6229,
      "step": 2889
    },
    {
      "epoch": 1.8490083173384517,
      "grad_norm": 2.067511809739775,
      "learning_rate": 1.876781429167756e-05,
      "loss": 0.6132,
      "step": 2890
    },
    {
      "epoch": 1.8496481126039668,
      "grad_norm": 2.156622368636454,
      "learning_rate": 1.8766817627185804e-05,
      "loss": 0.6151,
      "step": 2891
    },
    {
      "epoch": 1.8502879078694816,
      "grad_norm": 2.2700510196834984,
      "learning_rate": 1.8765820586262498e-05,
      "loss": 0.6455,
      "step": 2892
    },
    {
      "epoch": 1.8509277031349969,
      "grad_norm": 2.023424895251842,
      "learning_rate": 1.8764823168950445e-05,
      "loss": 0.5853,
      "step": 2893
    },
    {
      "epoch": 1.8515674984005117,
      "grad_norm": 2.1921440206929983,
      "learning_rate": 1.876382537529248e-05,
      "loss": 0.6475,
      "step": 2894
    },
    {
      "epoch": 1.852207293666027,
      "grad_norm": 2.298105938650836,
      "learning_rate": 1.8762827205331442e-05,
      "loss": 0.712,
      "step": 2895
    },
    {
      "epoch": 1.8528470889315418,
      "grad_norm": 2.1600725235486324,
      "learning_rate": 1.876182865911019e-05,
      "loss": 0.5759,
      "step": 2896
    },
    {
      "epoch": 1.853486884197057,
      "grad_norm": 1.9678340327071158,
      "learning_rate": 1.8760829736671603e-05,
      "loss": 0.6218,
      "step": 2897
    },
    {
      "epoch": 1.854126679462572,
      "grad_norm": 1.939634828439285,
      "learning_rate": 1.875983043805857e-05,
      "loss": 0.6196,
      "step": 2898
    },
    {
      "epoch": 1.854766474728087,
      "grad_norm": 1.9448584446526076,
      "learning_rate": 1.8758830763314006e-05,
      "loss": 0.7022,
      "step": 2899
    },
    {
      "epoch": 1.855406269993602,
      "grad_norm": 2.071464684743019,
      "learning_rate": 1.8757830712480826e-05,
      "loss": 0.6386,
      "step": 2900
    },
    {
      "epoch": 1.856046065259117,
      "grad_norm": 2.144015368116956,
      "learning_rate": 1.875683028560197e-05,
      "loss": 0.6906,
      "step": 2901
    },
    {
      "epoch": 1.8566858605246321,
      "grad_norm": 2.085986024732545,
      "learning_rate": 1.87558294827204e-05,
      "loss": 0.6383,
      "step": 2902
    },
    {
      "epoch": 1.8573256557901472,
      "grad_norm": 2.197345486566377,
      "learning_rate": 1.8754828303879093e-05,
      "loss": 0.6787,
      "step": 2903
    },
    {
      "epoch": 1.8579654510556622,
      "grad_norm": 2.0701380168906196,
      "learning_rate": 1.875382674912103e-05,
      "loss": 0.6055,
      "step": 2904
    },
    {
      "epoch": 1.858605246321177,
      "grad_norm": 2.0020651321471967,
      "learning_rate": 1.8752824818489218e-05,
      "loss": 0.6961,
      "step": 2905
    },
    {
      "epoch": 1.8592450415866923,
      "grad_norm": 2.3351117867057023,
      "learning_rate": 1.875182251202668e-05,
      "loss": 0.6743,
      "step": 2906
    },
    {
      "epoch": 1.8598848368522072,
      "grad_norm": 2.0884909461968943,
      "learning_rate": 1.8750819829776445e-05,
      "loss": 0.6677,
      "step": 2907
    },
    {
      "epoch": 1.8605246321177225,
      "grad_norm": 2.162067237747071,
      "learning_rate": 1.874981677178158e-05,
      "loss": 0.6673,
      "step": 2908
    },
    {
      "epoch": 1.8611644273832373,
      "grad_norm": 2.1828946603925075,
      "learning_rate": 1.8748813338085142e-05,
      "loss": 0.6614,
      "step": 2909
    },
    {
      "epoch": 1.8618042226487526,
      "grad_norm": 1.9689256028824054,
      "learning_rate": 1.874780952873023e-05,
      "loss": 0.662,
      "step": 2910
    },
    {
      "epoch": 1.8624440179142674,
      "grad_norm": 1.9580966655546213,
      "learning_rate": 1.874680534375993e-05,
      "loss": 0.5701,
      "step": 2911
    },
    {
      "epoch": 1.8630838131797824,
      "grad_norm": 2.223746688154118,
      "learning_rate": 1.8745800783217373e-05,
      "loss": 0.6694,
      "step": 2912
    },
    {
      "epoch": 1.8637236084452975,
      "grad_norm": 2.1494936169255405,
      "learning_rate": 1.8744795847145684e-05,
      "loss": 0.6472,
      "step": 2913
    },
    {
      "epoch": 1.8643634037108126,
      "grad_norm": 2.1318609740216425,
      "learning_rate": 1.8743790535588017e-05,
      "loss": 0.6627,
      "step": 2914
    },
    {
      "epoch": 1.8650031989763276,
      "grad_norm": 2.356685423108424,
      "learning_rate": 1.8742784848587543e-05,
      "loss": 0.7047,
      "step": 2915
    },
    {
      "epoch": 1.8656429942418427,
      "grad_norm": 2.1761516909966865,
      "learning_rate": 1.8741778786187437e-05,
      "loss": 0.6877,
      "step": 2916
    },
    {
      "epoch": 1.8662827895073577,
      "grad_norm": 2.2067175728885133,
      "learning_rate": 1.87407723484309e-05,
      "loss": 0.5905,
      "step": 2917
    },
    {
      "epoch": 1.8669225847728725,
      "grad_norm": 2.498313191224839,
      "learning_rate": 1.8739765535361147e-05,
      "loss": 0.5738,
      "step": 2918
    },
    {
      "epoch": 1.8675623800383878,
      "grad_norm": 2.542396852192343,
      "learning_rate": 1.8738758347021412e-05,
      "loss": 0.6299,
      "step": 2919
    },
    {
      "epoch": 1.8682021753039026,
      "grad_norm": 2.1444051717323127,
      "learning_rate": 1.8737750783454934e-05,
      "loss": 0.7254,
      "step": 2920
    },
    {
      "epoch": 1.868841970569418,
      "grad_norm": 2.2589069718865304,
      "learning_rate": 1.8736742844704984e-05,
      "loss": 0.5731,
      "step": 2921
    },
    {
      "epoch": 1.8694817658349328,
      "grad_norm": 2.395776002185925,
      "learning_rate": 1.8735734530814837e-05,
      "loss": 0.6301,
      "step": 2922
    },
    {
      "epoch": 1.8701215611004478,
      "grad_norm": 2.6362725471676676,
      "learning_rate": 1.8734725841827788e-05,
      "loss": 0.5808,
      "step": 2923
    },
    {
      "epoch": 1.8707613563659629,
      "grad_norm": 2.4335820303521123,
      "learning_rate": 1.873371677778715e-05,
      "loss": 0.6838,
      "step": 2924
    },
    {
      "epoch": 1.871401151631478,
      "grad_norm": 2.1331490704758767,
      "learning_rate": 1.8732707338736248e-05,
      "loss": 0.6638,
      "step": 2925
    },
    {
      "epoch": 1.872040946896993,
      "grad_norm": 2.1150090194788915,
      "learning_rate": 1.873169752471843e-05,
      "loss": 0.6764,
      "step": 2926
    },
    {
      "epoch": 1.872680742162508,
      "grad_norm": 2.828613672568026,
      "learning_rate": 1.8730687335777048e-05,
      "loss": 0.6313,
      "step": 2927
    },
    {
      "epoch": 1.873320537428023,
      "grad_norm": 2.430235097938454,
      "learning_rate": 1.8729676771955486e-05,
      "loss": 0.658,
      "step": 2928
    },
    {
      "epoch": 1.8739603326935381,
      "grad_norm": 2.4744529497912993,
      "learning_rate": 1.8728665833297132e-05,
      "loss": 0.612,
      "step": 2929
    },
    {
      "epoch": 1.8746001279590532,
      "grad_norm": 2.3571077512157257,
      "learning_rate": 1.8727654519845395e-05,
      "loss": 0.6116,
      "step": 2930
    },
    {
      "epoch": 1.875239923224568,
      "grad_norm": 2.336644576124146,
      "learning_rate": 1.8726642831643696e-05,
      "loss": 0.6619,
      "step": 2931
    },
    {
      "epoch": 1.8758797184900833,
      "grad_norm": 2.2699454209038366,
      "learning_rate": 1.8725630768735483e-05,
      "loss": 0.6076,
      "step": 2932
    },
    {
      "epoch": 1.8765195137555981,
      "grad_norm": 2.3852872872386714,
      "learning_rate": 1.87246183311642e-05,
      "loss": 0.7097,
      "step": 2933
    },
    {
      "epoch": 1.8771593090211134,
      "grad_norm": 2.21017491434317,
      "learning_rate": 1.8723605518973328e-05,
      "loss": 0.6328,
      "step": 2934
    },
    {
      "epoch": 1.8777991042866282,
      "grad_norm": 2.0856849810733036,
      "learning_rate": 1.872259233220635e-05,
      "loss": 0.6258,
      "step": 2935
    },
    {
      "epoch": 1.8784388995521433,
      "grad_norm": 2.0319749063653485,
      "learning_rate": 1.8721578770906776e-05,
      "loss": 0.5848,
      "step": 2936
    },
    {
      "epoch": 1.8790786948176583,
      "grad_norm": 2.0755886245950794,
      "learning_rate": 1.8720564835118128e-05,
      "loss": 0.6402,
      "step": 2937
    },
    {
      "epoch": 1.8797184900831734,
      "grad_norm": 2.287828114172013,
      "learning_rate": 1.8719550524883935e-05,
      "loss": 0.6734,
      "step": 2938
    },
    {
      "epoch": 1.8803582853486884,
      "grad_norm": 2.111434722129777,
      "learning_rate": 1.8718535840247752e-05,
      "loss": 0.5645,
      "step": 2939
    },
    {
      "epoch": 1.8809980806142035,
      "grad_norm": 2.203025761901405,
      "learning_rate": 1.871752078125315e-05,
      "loss": 0.6729,
      "step": 2940
    },
    {
      "epoch": 1.8816378758797185,
      "grad_norm": 2.1625333469221495,
      "learning_rate": 1.8716505347943713e-05,
      "loss": 0.6241,
      "step": 2941
    },
    {
      "epoch": 1.8822776711452334,
      "grad_norm": 2.291537943817498,
      "learning_rate": 1.8715489540363043e-05,
      "loss": 0.6822,
      "step": 2942
    },
    {
      "epoch": 1.8829174664107486,
      "grad_norm": 2.1132165540312693,
      "learning_rate": 1.8714473358554753e-05,
      "loss": 0.647,
      "step": 2943
    },
    {
      "epoch": 1.8835572616762635,
      "grad_norm": 2.0427221962925084,
      "learning_rate": 1.8713456802562484e-05,
      "loss": 0.6997,
      "step": 2944
    },
    {
      "epoch": 1.8841970569417787,
      "grad_norm": 2.318219313393404,
      "learning_rate": 1.8712439872429876e-05,
      "loss": 0.694,
      "step": 2945
    },
    {
      "epoch": 1.8848368522072936,
      "grad_norm": 2.1585075253754504,
      "learning_rate": 1.87114225682006e-05,
      "loss": 0.6388,
      "step": 2946
    },
    {
      "epoch": 1.8854766474728089,
      "grad_norm": 1.9983622185234529,
      "learning_rate": 1.8710404889918334e-05,
      "loss": 0.5814,
      "step": 2947
    },
    {
      "epoch": 1.8861164427383237,
      "grad_norm": 2.0747055978570765,
      "learning_rate": 1.8709386837626778e-05,
      "loss": 0.6034,
      "step": 2948
    },
    {
      "epoch": 1.8867562380038387,
      "grad_norm": 2.06531699222886,
      "learning_rate": 1.8708368411369643e-05,
      "loss": 0.6524,
      "step": 2949
    },
    {
      "epoch": 1.8873960332693538,
      "grad_norm": 2.1771859655198282,
      "learning_rate": 1.870734961119066e-05,
      "loss": 0.655,
      "step": 2950
    },
    {
      "epoch": 1.8880358285348688,
      "grad_norm": 2.0251692180972665,
      "learning_rate": 1.870633043713357e-05,
      "loss": 0.5972,
      "step": 2951
    },
    {
      "epoch": 1.888675623800384,
      "grad_norm": 2.223004343875369,
      "learning_rate": 1.8705310889242143e-05,
      "loss": 0.6725,
      "step": 2952
    },
    {
      "epoch": 1.889315419065899,
      "grad_norm": 2.170263506189026,
      "learning_rate": 1.870429096756015e-05,
      "loss": 0.6402,
      "step": 2953
    },
    {
      "epoch": 1.889955214331414,
      "grad_norm": 2.1323455457179614,
      "learning_rate": 1.870327067213139e-05,
      "loss": 0.6976,
      "step": 2954
    },
    {
      "epoch": 1.8905950095969288,
      "grad_norm": 1.998202277673882,
      "learning_rate": 1.8702250002999664e-05,
      "loss": 0.6057,
      "step": 2955
    },
    {
      "epoch": 1.891234804862444,
      "grad_norm": 1.9055871121211276,
      "learning_rate": 1.8701228960208807e-05,
      "loss": 0.6002,
      "step": 2956
    },
    {
      "epoch": 1.891874600127959,
      "grad_norm": 2.125257676994862,
      "learning_rate": 1.8700207543802654e-05,
      "loss": 0.6496,
      "step": 2957
    },
    {
      "epoch": 1.8925143953934742,
      "grad_norm": 1.998481843321588,
      "learning_rate": 1.8699185753825067e-05,
      "loss": 0.605,
      "step": 2958
    },
    {
      "epoch": 1.893154190658989,
      "grad_norm": 2.267153906661925,
      "learning_rate": 1.8698163590319916e-05,
      "loss": 0.6771,
      "step": 2959
    },
    {
      "epoch": 1.8937939859245043,
      "grad_norm": 2.081850438869495,
      "learning_rate": 1.8697141053331095e-05,
      "loss": 0.6345,
      "step": 2960
    },
    {
      "epoch": 1.8944337811900192,
      "grad_norm": 1.9570025494850063,
      "learning_rate": 1.869611814290251e-05,
      "loss": 0.5646,
      "step": 2961
    },
    {
      "epoch": 1.8950735764555342,
      "grad_norm": 1.9878846233506289,
      "learning_rate": 1.869509485907808e-05,
      "loss": 0.6374,
      "step": 2962
    },
    {
      "epoch": 1.8957133717210493,
      "grad_norm": 2.106696994842439,
      "learning_rate": 1.8694071201901743e-05,
      "loss": 0.664,
      "step": 2963
    },
    {
      "epoch": 1.8963531669865643,
      "grad_norm": 1.9731596234610953,
      "learning_rate": 1.8693047171417456e-05,
      "loss": 0.6626,
      "step": 2964
    },
    {
      "epoch": 1.8969929622520794,
      "grad_norm": 2.083344837707872,
      "learning_rate": 1.8692022767669186e-05,
      "loss": 0.5898,
      "step": 2965
    },
    {
      "epoch": 1.8976327575175944,
      "grad_norm": 2.0727422823209016,
      "learning_rate": 1.869099799070092e-05,
      "loss": 0.6619,
      "step": 2966
    },
    {
      "epoch": 1.8982725527831095,
      "grad_norm": 2.16603485832285,
      "learning_rate": 1.868997284055666e-05,
      "loss": 0.6381,
      "step": 2967
    },
    {
      "epoch": 1.8989123480486243,
      "grad_norm": 2.0911258442021827,
      "learning_rate": 1.8688947317280428e-05,
      "loss": 0.5999,
      "step": 2968
    },
    {
      "epoch": 1.8995521433141396,
      "grad_norm": 2.0703208457278834,
      "learning_rate": 1.868792142091625e-05,
      "loss": 0.6452,
      "step": 2969
    },
    {
      "epoch": 1.9001919385796544,
      "grad_norm": 2.401220849175881,
      "learning_rate": 1.8686895151508186e-05,
      "loss": 0.6096,
      "step": 2970
    },
    {
      "epoch": 1.9008317338451697,
      "grad_norm": 2.0173742745568366,
      "learning_rate": 1.868586850910029e-05,
      "loss": 0.6325,
      "step": 2971
    },
    {
      "epoch": 1.9014715291106845,
      "grad_norm": 2.132260221003137,
      "learning_rate": 1.8684841493736657e-05,
      "loss": 0.6442,
      "step": 2972
    },
    {
      "epoch": 1.9021113243761996,
      "grad_norm": 1.986674881624077,
      "learning_rate": 1.8683814105461376e-05,
      "loss": 0.6654,
      "step": 2973
    },
    {
      "epoch": 1.9027511196417146,
      "grad_norm": 2.1394203744297005,
      "learning_rate": 1.8682786344318566e-05,
      "loss": 0.6165,
      "step": 2974
    },
    {
      "epoch": 1.9033909149072297,
      "grad_norm": 2.045911277458061,
      "learning_rate": 1.8681758210352355e-05,
      "loss": 0.6098,
      "step": 2975
    },
    {
      "epoch": 1.9040307101727447,
      "grad_norm": 2.1220594412197125,
      "learning_rate": 1.868072970360689e-05,
      "loss": 0.6538,
      "step": 2976
    },
    {
      "epoch": 1.9046705054382598,
      "grad_norm": 2.1277889617077994,
      "learning_rate": 1.8679700824126334e-05,
      "loss": 0.619,
      "step": 2977
    },
    {
      "epoch": 1.9053103007037748,
      "grad_norm": 2.076917923480786,
      "learning_rate": 1.8678671571954866e-05,
      "loss": 0.6715,
      "step": 2978
    },
    {
      "epoch": 1.9059500959692899,
      "grad_norm": 2.0913047084737224,
      "learning_rate": 1.8677641947136678e-05,
      "loss": 0.6768,
      "step": 2979
    },
    {
      "epoch": 1.906589891234805,
      "grad_norm": 2.1693675852416656,
      "learning_rate": 1.867661194971598e-05,
      "loss": 0.6173,
      "step": 2980
    },
    {
      "epoch": 1.9072296865003198,
      "grad_norm": 1.8342561867368454,
      "learning_rate": 1.8675581579737e-05,
      "loss": 0.6174,
      "step": 2981
    },
    {
      "epoch": 1.907869481765835,
      "grad_norm": 2.0082020413707027,
      "learning_rate": 1.867455083724398e-05,
      "loss": 0.631,
      "step": 2982
    },
    {
      "epoch": 1.9085092770313499,
      "grad_norm": 2.39283334305872,
      "learning_rate": 1.8673519722281178e-05,
      "loss": 0.672,
      "step": 2983
    },
    {
      "epoch": 1.9091490722968651,
      "grad_norm": 1.9750348379515792,
      "learning_rate": 1.8672488234892867e-05,
      "loss": 0.6187,
      "step": 2984
    },
    {
      "epoch": 1.90978886756238,
      "grad_norm": 2.26873346171672,
      "learning_rate": 1.867145637512334e-05,
      "loss": 0.6843,
      "step": 2985
    },
    {
      "epoch": 1.910428662827895,
      "grad_norm": 2.161951423718307,
      "learning_rate": 1.86704241430169e-05,
      "loss": 0.652,
      "step": 2986
    },
    {
      "epoch": 1.91106845809341,
      "grad_norm": 2.1551887875064817,
      "learning_rate": 1.866939153861787e-05,
      "loss": 0.6728,
      "step": 2987
    },
    {
      "epoch": 1.9117082533589251,
      "grad_norm": 2.0956918851587267,
      "learning_rate": 1.8668358561970593e-05,
      "loss": 0.5858,
      "step": 2988
    },
    {
      "epoch": 1.9123480486244402,
      "grad_norm": 2.3280294744520984,
      "learning_rate": 1.8667325213119417e-05,
      "loss": 0.6944,
      "step": 2989
    },
    {
      "epoch": 1.9129878438899552,
      "grad_norm": 2.076332339480808,
      "learning_rate": 1.8666291492108714e-05,
      "loss": 0.6491,
      "step": 2990
    },
    {
      "epoch": 1.9136276391554703,
      "grad_norm": 2.135187634385911,
      "learning_rate": 1.866525739898287e-05,
      "loss": 0.633,
      "step": 2991
    },
    {
      "epoch": 1.9142674344209851,
      "grad_norm": 2.486189081200541,
      "learning_rate": 1.8664222933786285e-05,
      "loss": 0.6053,
      "step": 2992
    },
    {
      "epoch": 1.9149072296865004,
      "grad_norm": 2.014953381536383,
      "learning_rate": 1.8663188096563384e-05,
      "loss": 0.6284,
      "step": 2993
    },
    {
      "epoch": 1.9155470249520152,
      "grad_norm": 1.8663929030026474,
      "learning_rate": 1.8662152887358594e-05,
      "loss": 0.6067,
      "step": 2994
    },
    {
      "epoch": 1.9161868202175305,
      "grad_norm": 1.886161961214352,
      "learning_rate": 1.8661117306216373e-05,
      "loss": 0.5804,
      "step": 2995
    },
    {
      "epoch": 1.9168266154830453,
      "grad_norm": 2.9601318469653957,
      "learning_rate": 1.8660081353181178e-05,
      "loss": 0.6822,
      "step": 2996
    },
    {
      "epoch": 1.9174664107485606,
      "grad_norm": 2.130795161347464,
      "learning_rate": 1.8659045028297492e-05,
      "loss": 0.6382,
      "step": 2997
    },
    {
      "epoch": 1.9181062060140754,
      "grad_norm": 2.2085028191281864,
      "learning_rate": 1.865800833160982e-05,
      "loss": 0.6472,
      "step": 2998
    },
    {
      "epoch": 1.9187460012795905,
      "grad_norm": 2.200939969066634,
      "learning_rate": 1.865697126316267e-05,
      "loss": 0.6337,
      "step": 2999
    },
    {
      "epoch": 1.9193857965451055,
      "grad_norm": 2.1190599662057275,
      "learning_rate": 1.8655933823000572e-05,
      "loss": 0.6089,
      "step": 3000
    },
    {
      "epoch": 1.9200255918106206,
      "grad_norm": 2.066541500135304,
      "learning_rate": 1.8654896011168074e-05,
      "loss": 0.6077,
      "step": 3001
    },
    {
      "epoch": 1.9206653870761357,
      "grad_norm": 2.454381997162432,
      "learning_rate": 1.865385782770974e-05,
      "loss": 0.6458,
      "step": 3002
    },
    {
      "epoch": 1.9213051823416507,
      "grad_norm": 2.238734307960251,
      "learning_rate": 1.8652819272670145e-05,
      "loss": 0.6409,
      "step": 3003
    },
    {
      "epoch": 1.9219449776071658,
      "grad_norm": 2.0826532627433845,
      "learning_rate": 1.8651780346093876e-05,
      "loss": 0.6117,
      "step": 3004
    },
    {
      "epoch": 1.9225847728726806,
      "grad_norm": 2.0837900054043814,
      "learning_rate": 1.8650741048025555e-05,
      "loss": 0.634,
      "step": 3005
    },
    {
      "epoch": 1.9232245681381959,
      "grad_norm": 2.2723561803244907,
      "learning_rate": 1.8649701378509803e-05,
      "loss": 0.6574,
      "step": 3006
    },
    {
      "epoch": 1.9238643634037107,
      "grad_norm": 2.349033142456975,
      "learning_rate": 1.864866133759126e-05,
      "loss": 0.6305,
      "step": 3007
    },
    {
      "epoch": 1.924504158669226,
      "grad_norm": 2.425427785665372,
      "learning_rate": 1.864762092531458e-05,
      "loss": 0.6388,
      "step": 3008
    },
    {
      "epoch": 1.9251439539347408,
      "grad_norm": 2.1463923016594504,
      "learning_rate": 1.8646580141724442e-05,
      "loss": 0.6336,
      "step": 3009
    },
    {
      "epoch": 1.925783749200256,
      "grad_norm": 2.136936428886688,
      "learning_rate": 1.8645538986865534e-05,
      "loss": 0.6187,
      "step": 3010
    },
    {
      "epoch": 1.926423544465771,
      "grad_norm": 1.9880347299921932,
      "learning_rate": 1.864449746078256e-05,
      "loss": 0.6064,
      "step": 3011
    },
    {
      "epoch": 1.927063339731286,
      "grad_norm": 2.533615596969976,
      "learning_rate": 1.8643455563520244e-05,
      "loss": 0.7026,
      "step": 3012
    },
    {
      "epoch": 1.927703134996801,
      "grad_norm": 2.3400199987322976,
      "learning_rate": 1.8642413295123326e-05,
      "loss": 0.6558,
      "step": 3013
    },
    {
      "epoch": 1.928342930262316,
      "grad_norm": 2.2643564662258893,
      "learning_rate": 1.8641370655636548e-05,
      "loss": 0.6052,
      "step": 3014
    },
    {
      "epoch": 1.9289827255278311,
      "grad_norm": 2.226639106358743,
      "learning_rate": 1.864032764510469e-05,
      "loss": 0.6265,
      "step": 3015
    },
    {
      "epoch": 1.9296225207933462,
      "grad_norm": 2.0875721661237545,
      "learning_rate": 1.863928426357253e-05,
      "loss": 0.6182,
      "step": 3016
    },
    {
      "epoch": 1.9302623160588612,
      "grad_norm": 2.145097389459872,
      "learning_rate": 1.863824051108487e-05,
      "loss": 0.674,
      "step": 3017
    },
    {
      "epoch": 1.930902111324376,
      "grad_norm": 2.3102039303552893,
      "learning_rate": 1.8637196387686534e-05,
      "loss": 0.6085,
      "step": 3018
    },
    {
      "epoch": 1.9315419065898913,
      "grad_norm": 2.2561165658537896,
      "learning_rate": 1.863615189342235e-05,
      "loss": 0.5987,
      "step": 3019
    },
    {
      "epoch": 1.9321817018554062,
      "grad_norm": 2.6013503413047236,
      "learning_rate": 1.863510702833716e-05,
      "loss": 0.6244,
      "step": 3020
    },
    {
      "epoch": 1.9328214971209214,
      "grad_norm": 2.0835712926882803,
      "learning_rate": 1.8634061792475838e-05,
      "loss": 0.6033,
      "step": 3021
    },
    {
      "epoch": 1.9334612923864363,
      "grad_norm": 2.117504485566291,
      "learning_rate": 1.863301618588326e-05,
      "loss": 0.6762,
      "step": 3022
    },
    {
      "epoch": 1.9341010876519513,
      "grad_norm": 2.0363810003085914,
      "learning_rate": 1.8631970208604325e-05,
      "loss": 0.653,
      "step": 3023
    },
    {
      "epoch": 1.9347408829174664,
      "grad_norm": 2.068349767762942,
      "learning_rate": 1.8630923860683948e-05,
      "loss": 0.6271,
      "step": 3024
    },
    {
      "epoch": 1.9353806781829814,
      "grad_norm": 2.312840822557149,
      "learning_rate": 1.862987714216705e-05,
      "loss": 0.6252,
      "step": 3025
    },
    {
      "epoch": 1.9360204734484965,
      "grad_norm": 2.009324662748502,
      "learning_rate": 1.8628830053098577e-05,
      "loss": 0.6265,
      "step": 3026
    },
    {
      "epoch": 1.9366602687140115,
      "grad_norm": 2.267279757305446,
      "learning_rate": 1.8627782593523493e-05,
      "loss": 0.6613,
      "step": 3027
    },
    {
      "epoch": 1.9373000639795266,
      "grad_norm": 2.06337547321514,
      "learning_rate": 1.8626734763486773e-05,
      "loss": 0.6056,
      "step": 3028
    },
    {
      "epoch": 1.9379398592450416,
      "grad_norm": 2.099433933353757,
      "learning_rate": 1.8625686563033407e-05,
      "loss": 0.665,
      "step": 3029
    },
    {
      "epoch": 1.9385796545105567,
      "grad_norm": 2.241098038520009,
      "learning_rate": 1.8624637992208407e-05,
      "loss": 0.6725,
      "step": 3030
    },
    {
      "epoch": 1.9392194497760715,
      "grad_norm": 2.0650951485684,
      "learning_rate": 1.862358905105679e-05,
      "loss": 0.5989,
      "step": 3031
    },
    {
      "epoch": 1.9398592450415868,
      "grad_norm": 2.0892470752470294,
      "learning_rate": 1.8622539739623596e-05,
      "loss": 0.6429,
      "step": 3032
    },
    {
      "epoch": 1.9404990403071016,
      "grad_norm": 2.3701541317759554,
      "learning_rate": 1.8621490057953893e-05,
      "loss": 0.6494,
      "step": 3033
    },
    {
      "epoch": 1.941138835572617,
      "grad_norm": 2.2409924523476414,
      "learning_rate": 1.8620440006092737e-05,
      "loss": 0.623,
      "step": 3034
    },
    {
      "epoch": 1.9417786308381317,
      "grad_norm": 2.2232893151975457,
      "learning_rate": 1.8619389584085223e-05,
      "loss": 0.6217,
      "step": 3035
    },
    {
      "epoch": 1.9424184261036468,
      "grad_norm": 2.0687038643047173,
      "learning_rate": 1.8618338791976454e-05,
      "loss": 0.6513,
      "step": 3036
    },
    {
      "epoch": 1.9430582213691618,
      "grad_norm": 2.227286829829822,
      "learning_rate": 1.8617287629811548e-05,
      "loss": 0.6344,
      "step": 3037
    },
    {
      "epoch": 1.943698016634677,
      "grad_norm": 2.0479408839774678,
      "learning_rate": 1.8616236097635636e-05,
      "loss": 0.6002,
      "step": 3038
    },
    {
      "epoch": 1.944337811900192,
      "grad_norm": 2.11177617503627,
      "learning_rate": 1.861518419549388e-05,
      "loss": 0.5666,
      "step": 3039
    },
    {
      "epoch": 1.944977607165707,
      "grad_norm": 2.1725622064530192,
      "learning_rate": 1.8614131923431436e-05,
      "loss": 0.6735,
      "step": 3040
    },
    {
      "epoch": 1.945617402431222,
      "grad_norm": 2.1734715458265885,
      "learning_rate": 1.861307928149349e-05,
      "loss": 0.6296,
      "step": 3041
    },
    {
      "epoch": 1.9462571976967369,
      "grad_norm": 2.1609102497854535,
      "learning_rate": 1.8612026269725243e-05,
      "loss": 0.7177,
      "step": 3042
    },
    {
      "epoch": 1.9468969929622522,
      "grad_norm": 2.1309776757564682,
      "learning_rate": 1.8610972888171906e-05,
      "loss": 0.6418,
      "step": 3043
    },
    {
      "epoch": 1.947536788227767,
      "grad_norm": 2.198656773444115,
      "learning_rate": 1.860991913687871e-05,
      "loss": 0.6455,
      "step": 3044
    },
    {
      "epoch": 1.9481765834932823,
      "grad_norm": 2.2333925916719064,
      "learning_rate": 1.8608865015890906e-05,
      "loss": 0.6147,
      "step": 3045
    },
    {
      "epoch": 1.948816378758797,
      "grad_norm": 2.2700164904793017,
      "learning_rate": 1.8607810525253746e-05,
      "loss": 0.6139,
      "step": 3046
    },
    {
      "epoch": 1.9494561740243124,
      "grad_norm": 2.3749570023234376,
      "learning_rate": 1.860675566501252e-05,
      "loss": 0.6848,
      "step": 3047
    },
    {
      "epoch": 1.9500959692898272,
      "grad_norm": 2.1554072452833015,
      "learning_rate": 1.860570043521251e-05,
      "loss": 0.6643,
      "step": 3048
    },
    {
      "epoch": 1.9507357645553423,
      "grad_norm": 1.9545279532273638,
      "learning_rate": 1.860464483589904e-05,
      "loss": 0.6812,
      "step": 3049
    },
    {
      "epoch": 1.9513755598208573,
      "grad_norm": 2.1330780934091815,
      "learning_rate": 1.8603588867117423e-05,
      "loss": 0.6045,
      "step": 3050
    },
    {
      "epoch": 1.9520153550863724,
      "grad_norm": 1.9985601611244095,
      "learning_rate": 1.8602532528913005e-05,
      "loss": 0.6379,
      "step": 3051
    },
    {
      "epoch": 1.9526551503518874,
      "grad_norm": 2.0249482791987403,
      "learning_rate": 1.860147582133114e-05,
      "loss": 0.6058,
      "step": 3052
    },
    {
      "epoch": 1.9532949456174025,
      "grad_norm": 2.0620362463580997,
      "learning_rate": 1.8600418744417208e-05,
      "loss": 0.6292,
      "step": 3053
    },
    {
      "epoch": 1.9539347408829175,
      "grad_norm": 2.3066487416158044,
      "learning_rate": 1.859936129821659e-05,
      "loss": 0.6358,
      "step": 3054
    },
    {
      "epoch": 1.9545745361484324,
      "grad_norm": 2.386368783982632,
      "learning_rate": 1.85983034827747e-05,
      "loss": 0.6286,
      "step": 3055
    },
    {
      "epoch": 1.9552143314139476,
      "grad_norm": 1.9551843562751008,
      "learning_rate": 1.859724529813695e-05,
      "loss": 0.6123,
      "step": 3056
    },
    {
      "epoch": 1.9558541266794625,
      "grad_norm": 2.00313715068162,
      "learning_rate": 1.8596186744348785e-05,
      "loss": 0.5838,
      "step": 3057
    },
    {
      "epoch": 1.9564939219449777,
      "grad_norm": 2.401538192295629,
      "learning_rate": 1.8595127821455644e-05,
      "loss": 0.643,
      "step": 3058
    },
    {
      "epoch": 1.9571337172104926,
      "grad_norm": 2.2526169941753236,
      "learning_rate": 1.859406852950301e-05,
      "loss": 0.6637,
      "step": 3059
    },
    {
      "epoch": 1.9577735124760078,
      "grad_norm": 2.736859821537199,
      "learning_rate": 1.859300886853636e-05,
      "loss": 0.6434,
      "step": 3060
    },
    {
      "epoch": 1.9584133077415227,
      "grad_norm": 2.285406694468725,
      "learning_rate": 1.8591948838601194e-05,
      "loss": 0.6266,
      "step": 3061
    },
    {
      "epoch": 1.9590531030070377,
      "grad_norm": 1.9635541263066554,
      "learning_rate": 1.8590888439743032e-05,
      "loss": 0.6264,
      "step": 3062
    },
    {
      "epoch": 1.9596928982725528,
      "grad_norm": 2.1410310375237147,
      "learning_rate": 1.8589827672007402e-05,
      "loss": 0.639,
      "step": 3063
    },
    {
      "epoch": 1.9603326935380678,
      "grad_norm": 1.9868208262318119,
      "learning_rate": 1.8588766535439847e-05,
      "loss": 0.6009,
      "step": 3064
    },
    {
      "epoch": 1.9609724888035829,
      "grad_norm": 2.192999090048026,
      "learning_rate": 1.8587705030085938e-05,
      "loss": 0.6576,
      "step": 3065
    },
    {
      "epoch": 1.961612284069098,
      "grad_norm": 2.589322329986278,
      "learning_rate": 1.858664315599125e-05,
      "loss": 0.682,
      "step": 3066
    },
    {
      "epoch": 1.962252079334613,
      "grad_norm": 1.9582636222418128,
      "learning_rate": 1.858558091320138e-05,
      "loss": 0.6209,
      "step": 3067
    },
    {
      "epoch": 1.9628918746001278,
      "grad_norm": 2.080205867815064,
      "learning_rate": 1.8584518301761942e-05,
      "loss": 0.6025,
      "step": 3068
    },
    {
      "epoch": 1.963531669865643,
      "grad_norm": 2.0704182040920984,
      "learning_rate": 1.8583455321718556e-05,
      "loss": 0.6591,
      "step": 3069
    },
    {
      "epoch": 1.964171465131158,
      "grad_norm": 1.941942696623456,
      "learning_rate": 1.8582391973116867e-05,
      "loss": 0.6066,
      "step": 3070
    },
    {
      "epoch": 1.9648112603966732,
      "grad_norm": 2.175785589541706,
      "learning_rate": 1.8581328256002536e-05,
      "loss": 0.6304,
      "step": 3071
    },
    {
      "epoch": 1.965451055662188,
      "grad_norm": 2.336543452727772,
      "learning_rate": 1.858026417042123e-05,
      "loss": 0.6712,
      "step": 3072
    },
    {
      "epoch": 1.966090850927703,
      "grad_norm": 2.3813474994378203,
      "learning_rate": 1.8579199716418645e-05,
      "loss": 0.6544,
      "step": 3073
    },
    {
      "epoch": 1.9667306461932181,
      "grad_norm": 2.1955409929898617,
      "learning_rate": 1.8578134894040488e-05,
      "loss": 0.6202,
      "step": 3074
    },
    {
      "epoch": 1.9673704414587332,
      "grad_norm": 2.174582182667151,
      "learning_rate": 1.8577069703332477e-05,
      "loss": 0.6154,
      "step": 3075
    },
    {
      "epoch": 1.9680102367242482,
      "grad_norm": 2.097354225094591,
      "learning_rate": 1.857600414434035e-05,
      "loss": 0.6702,
      "step": 3076
    },
    {
      "epoch": 1.9686500319897633,
      "grad_norm": 2.232343751734811,
      "learning_rate": 1.857493821710986e-05,
      "loss": 0.5524,
      "step": 3077
    },
    {
      "epoch": 1.9692898272552783,
      "grad_norm": 2.2802615527128633,
      "learning_rate": 1.8573871921686775e-05,
      "loss": 0.6018,
      "step": 3078
    },
    {
      "epoch": 1.9699296225207934,
      "grad_norm": 2.474513331626896,
      "learning_rate": 1.8572805258116885e-05,
      "loss": 0.6171,
      "step": 3079
    },
    {
      "epoch": 1.9705694177863085,
      "grad_norm": 2.1809728523495737,
      "learning_rate": 1.8571738226445982e-05,
      "loss": 0.5864,
      "step": 3080
    },
    {
      "epoch": 1.9712092130518233,
      "grad_norm": 1.9522163452269783,
      "learning_rate": 1.857067082671989e-05,
      "loss": 0.6136,
      "step": 3081
    },
    {
      "epoch": 1.9718490083173386,
      "grad_norm": 1.993959528036216,
      "learning_rate": 1.856960305898444e-05,
      "loss": 0.6277,
      "step": 3082
    },
    {
      "epoch": 1.9724888035828534,
      "grad_norm": 2.163065796805472,
      "learning_rate": 1.856853492328548e-05,
      "loss": 0.6206,
      "step": 3083
    },
    {
      "epoch": 1.9731285988483687,
      "grad_norm": 1.8139468090928754,
      "learning_rate": 1.8567466419668872e-05,
      "loss": 0.6141,
      "step": 3084
    },
    {
      "epoch": 1.9737683941138835,
      "grad_norm": 2.131451355210186,
      "learning_rate": 1.8566397548180493e-05,
      "loss": 0.6494,
      "step": 3085
    },
    {
      "epoch": 1.9744081893793985,
      "grad_norm": 2.1221009559268507,
      "learning_rate": 1.8565328308866248e-05,
      "loss": 0.5716,
      "step": 3086
    },
    {
      "epoch": 1.9750479846449136,
      "grad_norm": 1.9736365423797655,
      "learning_rate": 1.8564258701772034e-05,
      "loss": 0.5998,
      "step": 3087
    },
    {
      "epoch": 1.9756877799104287,
      "grad_norm": 2.1303944913332526,
      "learning_rate": 1.856318872694379e-05,
      "loss": 0.624,
      "step": 3088
    },
    {
      "epoch": 1.9763275751759437,
      "grad_norm": 1.983856125162394,
      "learning_rate": 1.8562118384427457e-05,
      "loss": 0.606,
      "step": 3089
    },
    {
      "epoch": 1.9769673704414588,
      "grad_norm": 2.0207523326811363,
      "learning_rate": 1.856104767426899e-05,
      "loss": 0.6131,
      "step": 3090
    },
    {
      "epoch": 1.9776071657069738,
      "grad_norm": 2.130064896125763,
      "learning_rate": 1.855997659651436e-05,
      "loss": 0.5989,
      "step": 3091
    },
    {
      "epoch": 1.9782469609724886,
      "grad_norm": 2.3427052202052674,
      "learning_rate": 1.8558905151209567e-05,
      "loss": 0.6672,
      "step": 3092
    },
    {
      "epoch": 1.978886756238004,
      "grad_norm": 2.2033205348344285,
      "learning_rate": 1.855783333840061e-05,
      "loss": 0.5691,
      "step": 3093
    },
    {
      "epoch": 1.9795265515035187,
      "grad_norm": 2.3324947065973354,
      "learning_rate": 1.8556761158133514e-05,
      "loss": 0.6183,
      "step": 3094
    },
    {
      "epoch": 1.980166346769034,
      "grad_norm": 2.2108478845682047,
      "learning_rate": 1.8555688610454314e-05,
      "loss": 0.6661,
      "step": 3095
    },
    {
      "epoch": 1.9808061420345489,
      "grad_norm": 2.3929733909368105,
      "learning_rate": 1.8554615695409067e-05,
      "loss": 0.6114,
      "step": 3096
    },
    {
      "epoch": 1.9814459373000641,
      "grad_norm": 2.0201276297909687,
      "learning_rate": 1.8553542413043836e-05,
      "loss": 0.6517,
      "step": 3097
    },
    {
      "epoch": 1.982085732565579,
      "grad_norm": 2.037212313797247,
      "learning_rate": 1.855246876340471e-05,
      "loss": 0.6532,
      "step": 3098
    },
    {
      "epoch": 1.982725527831094,
      "grad_norm": 2.076249753209353,
      "learning_rate": 1.855139474653779e-05,
      "loss": 0.6378,
      "step": 3099
    },
    {
      "epoch": 1.983365323096609,
      "grad_norm": 2.0974047167167176,
      "learning_rate": 1.8550320362489192e-05,
      "loss": 0.6164,
      "step": 3100
    },
    {
      "epoch": 1.9840051183621241,
      "grad_norm": 1.8962033651058705,
      "learning_rate": 1.8549245611305045e-05,
      "loss": 0.5841,
      "step": 3101
    },
    {
      "epoch": 1.9846449136276392,
      "grad_norm": 2.2535717149429657,
      "learning_rate": 1.8548170493031503e-05,
      "loss": 0.6388,
      "step": 3102
    },
    {
      "epoch": 1.9852847088931542,
      "grad_norm": 2.0371201146976756,
      "learning_rate": 1.8547095007714722e-05,
      "loss": 0.5431,
      "step": 3103
    },
    {
      "epoch": 1.9859245041586693,
      "grad_norm": 2.076747910699748,
      "learning_rate": 1.854601915540089e-05,
      "loss": 0.6099,
      "step": 3104
    },
    {
      "epoch": 1.986564299424184,
      "grad_norm": 1.9824555917232665,
      "learning_rate": 1.8544942936136193e-05,
      "loss": 0.5746,
      "step": 3105
    },
    {
      "epoch": 1.9872040946896994,
      "grad_norm": 2.0019009620475314,
      "learning_rate": 1.8543866349966852e-05,
      "loss": 0.6231,
      "step": 3106
    },
    {
      "epoch": 1.9878438899552142,
      "grad_norm": 2.0549619025270247,
      "learning_rate": 1.854278939693908e-05,
      "loss": 0.6273,
      "step": 3107
    },
    {
      "epoch": 1.9884836852207295,
      "grad_norm": 2.0777519285648682,
      "learning_rate": 1.8541712077099133e-05,
      "loss": 0.6006,
      "step": 3108
    },
    {
      "epoch": 1.9891234804862443,
      "grad_norm": 1.9622419807812275,
      "learning_rate": 1.8540634390493268e-05,
      "loss": 0.6115,
      "step": 3109
    },
    {
      "epoch": 1.9897632757517596,
      "grad_norm": 2.0028850537774323,
      "learning_rate": 1.8539556337167747e-05,
      "loss": 0.6336,
      "step": 3110
    },
    {
      "epoch": 1.9904030710172744,
      "grad_norm": 1.9552804130061776,
      "learning_rate": 1.8538477917168874e-05,
      "loss": 0.5794,
      "step": 3111
    },
    {
      "epoch": 1.9910428662827895,
      "grad_norm": 2.117339372406761,
      "learning_rate": 1.8537399130542946e-05,
      "loss": 0.5299,
      "step": 3112
    },
    {
      "epoch": 1.9916826615483045,
      "grad_norm": 2.07667653181443,
      "learning_rate": 1.8536319977336284e-05,
      "loss": 0.5738,
      "step": 3113
    },
    {
      "epoch": 1.9923224568138196,
      "grad_norm": 1.9939382382276731,
      "learning_rate": 1.8535240457595226e-05,
      "loss": 0.6511,
      "step": 3114
    },
    {
      "epoch": 1.9929622520793346,
      "grad_norm": 2.350045069513722,
      "learning_rate": 1.8534160571366132e-05,
      "loss": 0.6174,
      "step": 3115
    },
    {
      "epoch": 1.9936020473448497,
      "grad_norm": 2.3881817365575793,
      "learning_rate": 1.853308031869536e-05,
      "loss": 0.5945,
      "step": 3116
    },
    {
      "epoch": 1.9942418426103647,
      "grad_norm": 2.233917852832862,
      "learning_rate": 1.8531999699629298e-05,
      "loss": 0.626,
      "step": 3117
    },
    {
      "epoch": 1.9948816378758796,
      "grad_norm": 2.1447626631501318,
      "learning_rate": 1.8530918714214347e-05,
      "loss": 0.6593,
      "step": 3118
    },
    {
      "epoch": 1.9955214331413949,
      "grad_norm": 2.224266149639766,
      "learning_rate": 1.852983736249692e-05,
      "loss": 0.6516,
      "step": 3119
    },
    {
      "epoch": 1.9961612284069097,
      "grad_norm": 2.2414224331293227,
      "learning_rate": 1.8528755644523454e-05,
      "loss": 0.5754,
      "step": 3120
    },
    {
      "epoch": 1.996801023672425,
      "grad_norm": 2.117417790256333,
      "learning_rate": 1.852767356034039e-05,
      "loss": 0.6388,
      "step": 3121
    },
    {
      "epoch": 1.9974408189379398,
      "grad_norm": 2.3826489623607237,
      "learning_rate": 1.8526591109994193e-05,
      "loss": 0.6253,
      "step": 3122
    },
    {
      "epoch": 1.9980806142034548,
      "grad_norm": 2.2145966149643908,
      "learning_rate": 1.852550829353134e-05,
      "loss": 0.6128,
      "step": 3123
    },
    {
      "epoch": 1.99872040946897,
      "grad_norm": 2.1865540759801974,
      "learning_rate": 1.8524425110998328e-05,
      "loss": 0.5754,
      "step": 3124
    },
    {
      "epoch": 1.999360204734485,
      "grad_norm": 2.4174851069814407,
      "learning_rate": 1.8523341562441662e-05,
      "loss": 0.6303,
      "step": 3125
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.1564664160426763,
      "learning_rate": 1.8522257647907874e-05,
      "loss": 0.5745,
      "step": 3126
    },
    {
      "epoch": 2.000639795265515,
      "grad_norm": 2.084059545778687,
      "learning_rate": 1.8521173367443507e-05,
      "loss": 0.3999,
      "step": 3127
    },
    {
      "epoch": 2.00127959053103,
      "grad_norm": 1.7187662130790593,
      "learning_rate": 1.8520088721095106e-05,
      "loss": 0.425,
      "step": 3128
    },
    {
      "epoch": 2.001919385796545,
      "grad_norm": 2.0911970676912452,
      "learning_rate": 1.8519003708909255e-05,
      "loss": 0.4203,
      "step": 3129
    },
    {
      "epoch": 2.00255918106206,
      "grad_norm": 1.9132945411866957,
      "learning_rate": 1.851791833093254e-05,
      "loss": 0.4638,
      "step": 3130
    },
    {
      "epoch": 2.003198976327575,
      "grad_norm": 2.0039978489056285,
      "learning_rate": 1.851683258721156e-05,
      "loss": 0.4089,
      "step": 3131
    },
    {
      "epoch": 2.0038387715930903,
      "grad_norm": 1.9945634898558888,
      "learning_rate": 1.8515746477792944e-05,
      "loss": 0.4315,
      "step": 3132
    },
    {
      "epoch": 2.004478566858605,
      "grad_norm": 2.302720189388838,
      "learning_rate": 1.851466000272332e-05,
      "loss": 0.428,
      "step": 3133
    },
    {
      "epoch": 2.0051183621241204,
      "grad_norm": 1.975561509980411,
      "learning_rate": 1.8513573162049344e-05,
      "loss": 0.4155,
      "step": 3134
    },
    {
      "epoch": 2.0057581573896353,
      "grad_norm": 2.0818806132699863,
      "learning_rate": 1.8512485955817675e-05,
      "loss": 0.4893,
      "step": 3135
    },
    {
      "epoch": 2.0063979526551505,
      "grad_norm": 1.8668038555495534,
      "learning_rate": 1.851139838407501e-05,
      "loss": 0.422,
      "step": 3136
    },
    {
      "epoch": 2.0070377479206654,
      "grad_norm": 1.7439722324946,
      "learning_rate": 1.8510310446868032e-05,
      "loss": 0.4134,
      "step": 3137
    },
    {
      "epoch": 2.0076775431861806,
      "grad_norm": 1.8613737255501501,
      "learning_rate": 1.8509222144243466e-05,
      "loss": 0.4221,
      "step": 3138
    },
    {
      "epoch": 2.0083173384516955,
      "grad_norm": 1.9165363038659013,
      "learning_rate": 1.8508133476248036e-05,
      "loss": 0.4512,
      "step": 3139
    },
    {
      "epoch": 2.0089571337172103,
      "grad_norm": 1.873242051675235,
      "learning_rate": 1.8507044442928488e-05,
      "loss": 0.4464,
      "step": 3140
    },
    {
      "epoch": 2.0095969289827256,
      "grad_norm": 2.0248390047835443,
      "learning_rate": 1.8505955044331586e-05,
      "loss": 0.4175,
      "step": 3141
    },
    {
      "epoch": 2.0102367242482404,
      "grad_norm": 1.973883871908841,
      "learning_rate": 1.850486528050411e-05,
      "loss": 0.4313,
      "step": 3142
    },
    {
      "epoch": 2.0108765195137557,
      "grad_norm": 1.9527265587217062,
      "learning_rate": 1.850377515149284e-05,
      "loss": 0.4078,
      "step": 3143
    },
    {
      "epoch": 2.0115163147792705,
      "grad_norm": 1.924492421823363,
      "learning_rate": 1.85026846573446e-05,
      "loss": 0.422,
      "step": 3144
    },
    {
      "epoch": 2.012156110044786,
      "grad_norm": 1.9558795196266385,
      "learning_rate": 1.85015937981062e-05,
      "loss": 0.4374,
      "step": 3145
    },
    {
      "epoch": 2.0127959053103006,
      "grad_norm": 1.8494276613566805,
      "learning_rate": 1.8500502573824485e-05,
      "loss": 0.3838,
      "step": 3146
    },
    {
      "epoch": 2.013435700575816,
      "grad_norm": 2.091666567096285,
      "learning_rate": 1.8499410984546312e-05,
      "loss": 0.4477,
      "step": 3147
    },
    {
      "epoch": 2.0140754958413307,
      "grad_norm": 1.948785958992389,
      "learning_rate": 1.8498319030318552e-05,
      "loss": 0.4401,
      "step": 3148
    },
    {
      "epoch": 2.014715291106846,
      "grad_norm": 2.023159550150914,
      "learning_rate": 1.849722671118809e-05,
      "loss": 0.4494,
      "step": 3149
    },
    {
      "epoch": 2.015355086372361,
      "grad_norm": 1.9446750634263308,
      "learning_rate": 1.849613402720183e-05,
      "loss": 0.449,
      "step": 3150
    },
    {
      "epoch": 2.0159948816378757,
      "grad_norm": 1.7143725231592952,
      "learning_rate": 1.849504097840668e-05,
      "loss": 0.4187,
      "step": 3151
    },
    {
      "epoch": 2.016634676903391,
      "grad_norm": 1.8495522801463113,
      "learning_rate": 1.849394756484959e-05,
      "loss": 0.4279,
      "step": 3152
    },
    {
      "epoch": 2.0172744721689058,
      "grad_norm": 1.7875794778007565,
      "learning_rate": 1.84928537865775e-05,
      "loss": 0.4132,
      "step": 3153
    },
    {
      "epoch": 2.017914267434421,
      "grad_norm": 1.8108235372225423,
      "learning_rate": 1.8491759643637378e-05,
      "loss": 0.4092,
      "step": 3154
    },
    {
      "epoch": 2.018554062699936,
      "grad_norm": 1.9845018368967788,
      "learning_rate": 1.84906651360762e-05,
      "loss": 0.4251,
      "step": 3155
    },
    {
      "epoch": 2.019193857965451,
      "grad_norm": 1.9972595325593803,
      "learning_rate": 1.8489570263940963e-05,
      "loss": 0.4472,
      "step": 3156
    },
    {
      "epoch": 2.019833653230966,
      "grad_norm": 1.9475677966940472,
      "learning_rate": 1.8488475027278682e-05,
      "loss": 0.4206,
      "step": 3157
    },
    {
      "epoch": 2.0204734484964813,
      "grad_norm": 1.9372343680426976,
      "learning_rate": 1.848737942613638e-05,
      "loss": 0.4122,
      "step": 3158
    },
    {
      "epoch": 2.021113243761996,
      "grad_norm": 1.9672551176427477,
      "learning_rate": 1.848628346056111e-05,
      "loss": 0.4573,
      "step": 3159
    },
    {
      "epoch": 2.0217530390275114,
      "grad_norm": 1.8537549940497655,
      "learning_rate": 1.848518713059992e-05,
      "loss": 0.4126,
      "step": 3160
    },
    {
      "epoch": 2.022392834293026,
      "grad_norm": 1.928174403098956,
      "learning_rate": 1.848409043629989e-05,
      "loss": 0.4551,
      "step": 3161
    },
    {
      "epoch": 2.0230326295585415,
      "grad_norm": 1.928661384964577,
      "learning_rate": 1.848299337770811e-05,
      "loss": 0.4018,
      "step": 3162
    },
    {
      "epoch": 2.0236724248240563,
      "grad_norm": 2.0158648485277944,
      "learning_rate": 1.848189595487168e-05,
      "loss": 0.415,
      "step": 3163
    },
    {
      "epoch": 2.024312220089571,
      "grad_norm": 1.8224790669979314,
      "learning_rate": 1.848079816783773e-05,
      "loss": 0.4146,
      "step": 3164
    },
    {
      "epoch": 2.0249520153550864,
      "grad_norm": 1.987079205578158,
      "learning_rate": 1.847970001665339e-05,
      "loss": 0.4247,
      "step": 3165
    },
    {
      "epoch": 2.0255918106206012,
      "grad_norm": 1.9265975994698827,
      "learning_rate": 1.8478601501365817e-05,
      "loss": 0.4075,
      "step": 3166
    },
    {
      "epoch": 2.0262316058861165,
      "grad_norm": 1.8220842108468138,
      "learning_rate": 1.8477502622022178e-05,
      "loss": 0.4772,
      "step": 3167
    },
    {
      "epoch": 2.0268714011516313,
      "grad_norm": 2.410666307612683,
      "learning_rate": 1.8476403378669653e-05,
      "loss": 0.4324,
      "step": 3168
    },
    {
      "epoch": 2.0275111964171466,
      "grad_norm": 2.062826333402659,
      "learning_rate": 1.847530377135545e-05,
      "loss": 0.4033,
      "step": 3169
    },
    {
      "epoch": 2.0281509916826614,
      "grad_norm": 1.9521365716792138,
      "learning_rate": 1.8474203800126778e-05,
      "loss": 0.4139,
      "step": 3170
    },
    {
      "epoch": 2.0287907869481767,
      "grad_norm": 1.8719369989027914,
      "learning_rate": 1.847310346503087e-05,
      "loss": 0.4493,
      "step": 3171
    },
    {
      "epoch": 2.0294305822136915,
      "grad_norm": 1.7074980133038855,
      "learning_rate": 1.847200276611497e-05,
      "loss": 0.4173,
      "step": 3172
    },
    {
      "epoch": 2.030070377479207,
      "grad_norm": 1.9514061827707505,
      "learning_rate": 1.8470901703426343e-05,
      "loss": 0.4114,
      "step": 3173
    },
    {
      "epoch": 2.0307101727447217,
      "grad_norm": 1.8587934970932751,
      "learning_rate": 1.8469800277012263e-05,
      "loss": 0.4042,
      "step": 3174
    },
    {
      "epoch": 2.031349968010237,
      "grad_norm": 1.9756883325721832,
      "learning_rate": 1.846869848692003e-05,
      "loss": 0.4425,
      "step": 3175
    },
    {
      "epoch": 2.0319897632757518,
      "grad_norm": 1.8539167861328782,
      "learning_rate": 1.8467596333196945e-05,
      "loss": 0.4234,
      "step": 3176
    },
    {
      "epoch": 2.0326295585412666,
      "grad_norm": 1.8413735603418653,
      "learning_rate": 1.846649381589034e-05,
      "loss": 0.448,
      "step": 3177
    },
    {
      "epoch": 2.033269353806782,
      "grad_norm": 1.9995400621655735,
      "learning_rate": 1.8465390935047548e-05,
      "loss": 0.4626,
      "step": 3178
    },
    {
      "epoch": 2.0339091490722967,
      "grad_norm": 2.0400573722404025,
      "learning_rate": 1.8464287690715927e-05,
      "loss": 0.4821,
      "step": 3179
    },
    {
      "epoch": 2.034548944337812,
      "grad_norm": 2.106002514706204,
      "learning_rate": 1.8463184082942855e-05,
      "loss": 0.4397,
      "step": 3180
    },
    {
      "epoch": 2.035188739603327,
      "grad_norm": 1.9377492417359665,
      "learning_rate": 1.8462080111775707e-05,
      "loss": 0.4369,
      "step": 3181
    },
    {
      "epoch": 2.035828534868842,
      "grad_norm": 1.9458238957749483,
      "learning_rate": 1.8460975777261895e-05,
      "loss": 0.4444,
      "step": 3182
    },
    {
      "epoch": 2.036468330134357,
      "grad_norm": 1.957025237601507,
      "learning_rate": 1.845987107944883e-05,
      "loss": 0.4487,
      "step": 3183
    },
    {
      "epoch": 2.037108125399872,
      "grad_norm": 1.8484071016567107,
      "learning_rate": 1.8458766018383956e-05,
      "loss": 0.4051,
      "step": 3184
    },
    {
      "epoch": 2.037747920665387,
      "grad_norm": 2.0344380965009887,
      "learning_rate": 1.845766059411471e-05,
      "loss": 0.4259,
      "step": 3185
    },
    {
      "epoch": 2.0383877159309023,
      "grad_norm": 1.951575703578991,
      "learning_rate": 1.8456554806688568e-05,
      "loss": 0.4226,
      "step": 3186
    },
    {
      "epoch": 2.039027511196417,
      "grad_norm": 1.9468656246631184,
      "learning_rate": 1.8455448656153003e-05,
      "loss": 0.4493,
      "step": 3187
    },
    {
      "epoch": 2.0396673064619324,
      "grad_norm": 2.2359189556787045,
      "learning_rate": 1.845434214255552e-05,
      "loss": 0.4097,
      "step": 3188
    },
    {
      "epoch": 2.0403071017274472,
      "grad_norm": 1.9250777232243266,
      "learning_rate": 1.8453235265943616e-05,
      "loss": 0.4221,
      "step": 3189
    },
    {
      "epoch": 2.040946896992962,
      "grad_norm": 2.0299662791747806,
      "learning_rate": 1.845212802636483e-05,
      "loss": 0.4484,
      "step": 3190
    },
    {
      "epoch": 2.0415866922584773,
      "grad_norm": 1.9278365471933778,
      "learning_rate": 1.8451020423866703e-05,
      "loss": 0.3986,
      "step": 3191
    },
    {
      "epoch": 2.042226487523992,
      "grad_norm": 1.918593712295247,
      "learning_rate": 1.844991245849679e-05,
      "loss": 0.3994,
      "step": 3192
    },
    {
      "epoch": 2.0428662827895074,
      "grad_norm": 1.693676459326993,
      "learning_rate": 1.844880413030267e-05,
      "loss": 0.4015,
      "step": 3193
    },
    {
      "epoch": 2.0435060780550223,
      "grad_norm": 1.9765082378973853,
      "learning_rate": 1.8447695439331926e-05,
      "loss": 0.4368,
      "step": 3194
    },
    {
      "epoch": 2.0441458733205375,
      "grad_norm": 1.977322609738211,
      "learning_rate": 1.8446586385632172e-05,
      "loss": 0.4276,
      "step": 3195
    },
    {
      "epoch": 2.0447856685860524,
      "grad_norm": 2.040277940912987,
      "learning_rate": 1.844547696925102e-05,
      "loss": 0.442,
      "step": 3196
    },
    {
      "epoch": 2.0454254638515676,
      "grad_norm": 2.0385479478895,
      "learning_rate": 1.844436719023611e-05,
      "loss": 0.4739,
      "step": 3197
    },
    {
      "epoch": 2.0460652591170825,
      "grad_norm": 1.8774701885792902,
      "learning_rate": 1.8443257048635096e-05,
      "loss": 0.3869,
      "step": 3198
    },
    {
      "epoch": 2.0467050543825978,
      "grad_norm": 2.0492860654198926,
      "learning_rate": 1.8442146544495645e-05,
      "loss": 0.4372,
      "step": 3199
    },
    {
      "epoch": 2.0473448496481126,
      "grad_norm": 1.9760554377738957,
      "learning_rate": 1.844103567786544e-05,
      "loss": 0.4406,
      "step": 3200
    },
    {
      "epoch": 2.047984644913628,
      "grad_norm": 2.1791915502570776,
      "learning_rate": 1.8439924448792172e-05,
      "loss": 0.4302,
      "step": 3201
    },
    {
      "epoch": 2.0486244401791427,
      "grad_norm": 2.329888915316488,
      "learning_rate": 1.8438812857323564e-05,
      "loss": 0.4185,
      "step": 3202
    },
    {
      "epoch": 2.0492642354446575,
      "grad_norm": 2.0962639364549607,
      "learning_rate": 1.8437700903507346e-05,
      "loss": 0.4254,
      "step": 3203
    },
    {
      "epoch": 2.049904030710173,
      "grad_norm": 1.8213552169982439,
      "learning_rate": 1.843658858739126e-05,
      "loss": 0.403,
      "step": 3204
    },
    {
      "epoch": 2.0505438259756876,
      "grad_norm": 2.1465868860387904,
      "learning_rate": 1.843547590902307e-05,
      "loss": 0.4217,
      "step": 3205
    },
    {
      "epoch": 2.051183621241203,
      "grad_norm": 1.9326241991556135,
      "learning_rate": 1.8434362868450545e-05,
      "loss": 0.4996,
      "step": 3206
    },
    {
      "epoch": 2.0518234165067177,
      "grad_norm": 1.8655295346363663,
      "learning_rate": 1.843324946572148e-05,
      "loss": 0.4089,
      "step": 3207
    },
    {
      "epoch": 2.052463211772233,
      "grad_norm": 2.075437143335758,
      "learning_rate": 1.8432135700883692e-05,
      "loss": 0.4555,
      "step": 3208
    },
    {
      "epoch": 2.053103007037748,
      "grad_norm": 1.9800359863502646,
      "learning_rate": 1.8431021573984993e-05,
      "loss": 0.4221,
      "step": 3209
    },
    {
      "epoch": 2.053742802303263,
      "grad_norm": 2.2777303080148945,
      "learning_rate": 1.8429907085073225e-05,
      "loss": 0.4355,
      "step": 3210
    },
    {
      "epoch": 2.054382597568778,
      "grad_norm": 1.970868122943132,
      "learning_rate": 1.842879223419624e-05,
      "loss": 0.4278,
      "step": 3211
    },
    {
      "epoch": 2.055022392834293,
      "grad_norm": 1.8935526408562997,
      "learning_rate": 1.8427677021401914e-05,
      "loss": 0.4784,
      "step": 3212
    },
    {
      "epoch": 2.055662188099808,
      "grad_norm": 1.949167486742651,
      "learning_rate": 1.842656144673813e-05,
      "loss": 0.4247,
      "step": 3213
    },
    {
      "epoch": 2.056301983365323,
      "grad_norm": 1.9671022600320922,
      "learning_rate": 1.842544551025278e-05,
      "loss": 0.4279,
      "step": 3214
    },
    {
      "epoch": 2.056941778630838,
      "grad_norm": 2.126256159638697,
      "learning_rate": 1.8424329211993795e-05,
      "loss": 0.4204,
      "step": 3215
    },
    {
      "epoch": 2.057581573896353,
      "grad_norm": 2.1260533077259143,
      "learning_rate": 1.8423212552009094e-05,
      "loss": 0.4734,
      "step": 3216
    },
    {
      "epoch": 2.0582213691618683,
      "grad_norm": 1.9228473461748878,
      "learning_rate": 1.8422095530346635e-05,
      "loss": 0.4345,
      "step": 3217
    },
    {
      "epoch": 2.058861164427383,
      "grad_norm": 1.7932421490596548,
      "learning_rate": 1.842097814705437e-05,
      "loss": 0.403,
      "step": 3218
    },
    {
      "epoch": 2.0595009596928984,
      "grad_norm": 1.9465249632283361,
      "learning_rate": 1.8419860402180286e-05,
      "loss": 0.3873,
      "step": 3219
    },
    {
      "epoch": 2.060140754958413,
      "grad_norm": 1.9728760905584108,
      "learning_rate": 1.8418742295772377e-05,
      "loss": 0.4428,
      "step": 3220
    },
    {
      "epoch": 2.0607805502239285,
      "grad_norm": 1.9582296182933852,
      "learning_rate": 1.8417623827878647e-05,
      "loss": 0.3964,
      "step": 3221
    },
    {
      "epoch": 2.0614203454894433,
      "grad_norm": 1.9490344588688582,
      "learning_rate": 1.841650499854712e-05,
      "loss": 0.4187,
      "step": 3222
    },
    {
      "epoch": 2.0620601407549586,
      "grad_norm": 2.0861873522408114,
      "learning_rate": 1.8415385807825844e-05,
      "loss": 0.4697,
      "step": 3223
    },
    {
      "epoch": 2.0626999360204734,
      "grad_norm": 1.9142725320070881,
      "learning_rate": 1.841426625576287e-05,
      "loss": 0.3858,
      "step": 3224
    },
    {
      "epoch": 2.0633397312859887,
      "grad_norm": 1.8259687775211293,
      "learning_rate": 1.8413146342406273e-05,
      "loss": 0.411,
      "step": 3225
    },
    {
      "epoch": 2.0639795265515035,
      "grad_norm": 1.970237071240535,
      "learning_rate": 1.8412026067804137e-05,
      "loss": 0.4526,
      "step": 3226
    },
    {
      "epoch": 2.0646193218170183,
      "grad_norm": 1.9434295499978977,
      "learning_rate": 1.8410905432004564e-05,
      "loss": 0.4221,
      "step": 3227
    },
    {
      "epoch": 2.0652591170825336,
      "grad_norm": 1.9243691308047428,
      "learning_rate": 1.8409784435055672e-05,
      "loss": 0.4416,
      "step": 3228
    },
    {
      "epoch": 2.0658989123480485,
      "grad_norm": 1.947851752546957,
      "learning_rate": 1.84086630770056e-05,
      "loss": 0.4692,
      "step": 3229
    },
    {
      "epoch": 2.0665387076135637,
      "grad_norm": 1.7444205230981165,
      "learning_rate": 1.840754135790249e-05,
      "loss": 0.4756,
      "step": 3230
    },
    {
      "epoch": 2.0671785028790786,
      "grad_norm": 1.7853111202490537,
      "learning_rate": 1.8406419277794514e-05,
      "loss": 0.4512,
      "step": 3231
    },
    {
      "epoch": 2.067818298144594,
      "grad_norm": 1.8868726752043374,
      "learning_rate": 1.8405296836729846e-05,
      "loss": 0.4014,
      "step": 3232
    },
    {
      "epoch": 2.0684580934101087,
      "grad_norm": 1.9928027138929596,
      "learning_rate": 1.8404174034756684e-05,
      "loss": 0.4056,
      "step": 3233
    },
    {
      "epoch": 2.069097888675624,
      "grad_norm": 1.8981129081088788,
      "learning_rate": 1.8403050871923237e-05,
      "loss": 0.4225,
      "step": 3234
    },
    {
      "epoch": 2.0697376839411388,
      "grad_norm": 1.789284922698973,
      "learning_rate": 1.8401927348277737e-05,
      "loss": 0.425,
      "step": 3235
    },
    {
      "epoch": 2.070377479206654,
      "grad_norm": 1.8260995975945606,
      "learning_rate": 1.840080346386842e-05,
      "loss": 0.4313,
      "step": 3236
    },
    {
      "epoch": 2.071017274472169,
      "grad_norm": 1.9838147501752392,
      "learning_rate": 1.8399679218743543e-05,
      "loss": 0.4118,
      "step": 3237
    },
    {
      "epoch": 2.071657069737684,
      "grad_norm": 2.1553534866102404,
      "learning_rate": 1.839855461295139e-05,
      "loss": 0.4527,
      "step": 3238
    },
    {
      "epoch": 2.072296865003199,
      "grad_norm": 1.8534174829602124,
      "learning_rate": 1.8397429646540235e-05,
      "loss": 0.4466,
      "step": 3239
    },
    {
      "epoch": 2.072936660268714,
      "grad_norm": 1.953391947102055,
      "learning_rate": 1.8396304319558395e-05,
      "loss": 0.4187,
      "step": 3240
    },
    {
      "epoch": 2.073576455534229,
      "grad_norm": 1.945137420276321,
      "learning_rate": 1.839517863205418e-05,
      "loss": 0.3936,
      "step": 3241
    },
    {
      "epoch": 2.074216250799744,
      "grad_norm": 1.9413160166476033,
      "learning_rate": 1.8394052584075925e-05,
      "loss": 0.4643,
      "step": 3242
    },
    {
      "epoch": 2.074856046065259,
      "grad_norm": 1.8543137327536598,
      "learning_rate": 1.8392926175671986e-05,
      "loss": 0.4589,
      "step": 3243
    },
    {
      "epoch": 2.075495841330774,
      "grad_norm": 1.7565474241200842,
      "learning_rate": 1.8391799406890728e-05,
      "loss": 0.3624,
      "step": 3244
    },
    {
      "epoch": 2.0761356365962893,
      "grad_norm": 1.827802776569207,
      "learning_rate": 1.839067227778053e-05,
      "loss": 0.4074,
      "step": 3245
    },
    {
      "epoch": 2.076775431861804,
      "grad_norm": 1.7806953123931053,
      "learning_rate": 1.8389544788389795e-05,
      "loss": 0.4048,
      "step": 3246
    },
    {
      "epoch": 2.0774152271273194,
      "grad_norm": 1.9986808479175306,
      "learning_rate": 1.8388416938766926e-05,
      "loss": 0.4389,
      "step": 3247
    },
    {
      "epoch": 2.0780550223928342,
      "grad_norm": 1.915220781375992,
      "learning_rate": 1.838728872896036e-05,
      "loss": 0.4358,
      "step": 3248
    },
    {
      "epoch": 2.0786948176583495,
      "grad_norm": 1.9101337370083702,
      "learning_rate": 1.8386160159018527e-05,
      "loss": 0.4384,
      "step": 3249
    },
    {
      "epoch": 2.0793346129238643,
      "grad_norm": 1.9270199362849831,
      "learning_rate": 1.83850312289899e-05,
      "loss": 0.446,
      "step": 3250
    },
    {
      "epoch": 2.079974408189379,
      "grad_norm": 1.9595289773950673,
      "learning_rate": 1.838390193892295e-05,
      "loss": 0.4354,
      "step": 3251
    },
    {
      "epoch": 2.0806142034548945,
      "grad_norm": 2.00912826492078,
      "learning_rate": 1.8382772288866163e-05,
      "loss": 0.4464,
      "step": 3252
    },
    {
      "epoch": 2.0812539987204093,
      "grad_norm": 1.959185716250538,
      "learning_rate": 1.8381642278868042e-05,
      "loss": 0.4365,
      "step": 3253
    },
    {
      "epoch": 2.0818937939859246,
      "grad_norm": 1.97578238876535,
      "learning_rate": 1.8380511908977115e-05,
      "loss": 0.4409,
      "step": 3254
    },
    {
      "epoch": 2.0825335892514394,
      "grad_norm": 1.865946583838932,
      "learning_rate": 1.8379381179241916e-05,
      "loss": 0.4285,
      "step": 3255
    },
    {
      "epoch": 2.0831733845169547,
      "grad_norm": 1.8788576452901244,
      "learning_rate": 1.837825008971099e-05,
      "loss": 0.3907,
      "step": 3256
    },
    {
      "epoch": 2.0838131797824695,
      "grad_norm": 2.0005197778165305,
      "learning_rate": 1.8377118640432914e-05,
      "loss": 0.4192,
      "step": 3257
    },
    {
      "epoch": 2.0844529750479848,
      "grad_norm": 1.9496373946946617,
      "learning_rate": 1.8375986831456262e-05,
      "loss": 0.4245,
      "step": 3258
    },
    {
      "epoch": 2.0850927703134996,
      "grad_norm": 1.8986650658687299,
      "learning_rate": 1.8374854662829634e-05,
      "loss": 0.4014,
      "step": 3259
    },
    {
      "epoch": 2.085732565579015,
      "grad_norm": 2.0626404205839983,
      "learning_rate": 1.837372213460165e-05,
      "loss": 0.4249,
      "step": 3260
    },
    {
      "epoch": 2.0863723608445297,
      "grad_norm": 2.055140402344442,
      "learning_rate": 1.837258924682093e-05,
      "loss": 0.4405,
      "step": 3261
    },
    {
      "epoch": 2.087012156110045,
      "grad_norm": 1.9504413323882877,
      "learning_rate": 1.8371455999536122e-05,
      "loss": 0.4445,
      "step": 3262
    },
    {
      "epoch": 2.08765195137556,
      "grad_norm": 1.722100035724172,
      "learning_rate": 1.8370322392795882e-05,
      "loss": 0.4056,
      "step": 3263
    },
    {
      "epoch": 2.0882917466410746,
      "grad_norm": 1.7845310940839785,
      "learning_rate": 1.836918842664889e-05,
      "loss": 0.4147,
      "step": 3264
    },
    {
      "epoch": 2.08893154190659,
      "grad_norm": 2.270992231644015,
      "learning_rate": 1.8368054101143834e-05,
      "loss": 0.3821,
      "step": 3265
    },
    {
      "epoch": 2.0895713371721047,
      "grad_norm": 2.0523293565056386,
      "learning_rate": 1.8366919416329427e-05,
      "loss": 0.416,
      "step": 3266
    },
    {
      "epoch": 2.09021113243762,
      "grad_norm": 2.198521989908841,
      "learning_rate": 1.8365784372254378e-05,
      "loss": 0.4715,
      "step": 3267
    },
    {
      "epoch": 2.090850927703135,
      "grad_norm": 1.8015597388253355,
      "learning_rate": 1.8364648968967428e-05,
      "loss": 0.3814,
      "step": 3268
    },
    {
      "epoch": 2.09149072296865,
      "grad_norm": 2.0484758315860567,
      "learning_rate": 1.8363513206517332e-05,
      "loss": 0.4476,
      "step": 3269
    },
    {
      "epoch": 2.092130518234165,
      "grad_norm": 2.061740419808637,
      "learning_rate": 1.8362377084952857e-05,
      "loss": 0.4402,
      "step": 3270
    },
    {
      "epoch": 2.0927703134996802,
      "grad_norm": 1.8061684081658842,
      "learning_rate": 1.8361240604322784e-05,
      "loss": 0.3989,
      "step": 3271
    },
    {
      "epoch": 2.093410108765195,
      "grad_norm": 1.867960456873935,
      "learning_rate": 1.8360103764675916e-05,
      "loss": 0.4558,
      "step": 3272
    },
    {
      "epoch": 2.0940499040307103,
      "grad_norm": 1.773884245720134,
      "learning_rate": 1.8358966566061062e-05,
      "loss": 0.4007,
      "step": 3273
    },
    {
      "epoch": 2.094689699296225,
      "grad_norm": 1.9751977243443044,
      "learning_rate": 1.8357829008527055e-05,
      "loss": 0.4611,
      "step": 3274
    },
    {
      "epoch": 2.0953294945617404,
      "grad_norm": 1.812840673047891,
      "learning_rate": 1.8356691092122736e-05,
      "loss": 0.4536,
      "step": 3275
    },
    {
      "epoch": 2.0959692898272553,
      "grad_norm": 1.8777292162283041,
      "learning_rate": 1.8355552816896968e-05,
      "loss": 0.3808,
      "step": 3276
    },
    {
      "epoch": 2.09660908509277,
      "grad_norm": 2.0592970976906257,
      "learning_rate": 1.8354414182898625e-05,
      "loss": 0.3964,
      "step": 3277
    },
    {
      "epoch": 2.0972488803582854,
      "grad_norm": 1.9760246301045437,
      "learning_rate": 1.8353275190176597e-05,
      "loss": 0.4301,
      "step": 3278
    },
    {
      "epoch": 2.0978886756238,
      "grad_norm": 1.9915157313018432,
      "learning_rate": 1.8352135838779793e-05,
      "loss": 0.4045,
      "step": 3279
    },
    {
      "epoch": 2.0985284708893155,
      "grad_norm": 1.8629013887559842,
      "learning_rate": 1.8350996128757136e-05,
      "loss": 0.4736,
      "step": 3280
    },
    {
      "epoch": 2.0991682661548303,
      "grad_norm": 2.0677516905578006,
      "learning_rate": 1.8349856060157555e-05,
      "loss": 0.4626,
      "step": 3281
    },
    {
      "epoch": 2.0998080614203456,
      "grad_norm": 1.8325440417038705,
      "learning_rate": 1.834871563303001e-05,
      "loss": 0.3626,
      "step": 3282
    },
    {
      "epoch": 2.1004478566858604,
      "grad_norm": 1.9125509171387816,
      "learning_rate": 1.834757484742347e-05,
      "loss": 0.4277,
      "step": 3283
    },
    {
      "epoch": 2.1010876519513757,
      "grad_norm": 1.9796868994814476,
      "learning_rate": 1.834643370338691e-05,
      "loss": 0.4229,
      "step": 3284
    },
    {
      "epoch": 2.1017274472168905,
      "grad_norm": 1.8865869713343844,
      "learning_rate": 1.8345292200969334e-05,
      "loss": 0.399,
      "step": 3285
    },
    {
      "epoch": 2.102367242482406,
      "grad_norm": 1.8857583373473612,
      "learning_rate": 1.834415034021976e-05,
      "loss": 0.4596,
      "step": 3286
    },
    {
      "epoch": 2.1030070377479206,
      "grad_norm": 1.8004608614489461,
      "learning_rate": 1.8343008121187215e-05,
      "loss": 0.4263,
      "step": 3287
    },
    {
      "epoch": 2.103646833013436,
      "grad_norm": 1.9825251188367659,
      "learning_rate": 1.834186554392074e-05,
      "loss": 0.4644,
      "step": 3288
    },
    {
      "epoch": 2.1042866282789507,
      "grad_norm": 2.0205043961002698,
      "learning_rate": 1.8340722608469396e-05,
      "loss": 0.4205,
      "step": 3289
    },
    {
      "epoch": 2.1049264235444656,
      "grad_norm": 1.9120970478723414,
      "learning_rate": 1.833957931488226e-05,
      "loss": 0.4188,
      "step": 3290
    },
    {
      "epoch": 2.105566218809981,
      "grad_norm": 1.8954044267963233,
      "learning_rate": 1.8338435663208427e-05,
      "loss": 0.3838,
      "step": 3291
    },
    {
      "epoch": 2.1062060140754957,
      "grad_norm": 2.0762856534221603,
      "learning_rate": 1.8337291653496995e-05,
      "loss": 0.4733,
      "step": 3292
    },
    {
      "epoch": 2.106845809341011,
      "grad_norm": 1.9388227802621294,
      "learning_rate": 1.8336147285797096e-05,
      "loss": 0.4067,
      "step": 3293
    },
    {
      "epoch": 2.107485604606526,
      "grad_norm": 1.8290093051255139,
      "learning_rate": 1.8335002560157855e-05,
      "loss": 0.4258,
      "step": 3294
    },
    {
      "epoch": 2.108125399872041,
      "grad_norm": 1.7601531521233043,
      "learning_rate": 1.8333857476628434e-05,
      "loss": 0.4419,
      "step": 3295
    },
    {
      "epoch": 2.108765195137556,
      "grad_norm": 2.0692494995989024,
      "learning_rate": 1.8332712035257998e-05,
      "loss": 0.4342,
      "step": 3296
    },
    {
      "epoch": 2.109404990403071,
      "grad_norm": 1.8774067561259946,
      "learning_rate": 1.833156623609573e-05,
      "loss": 0.472,
      "step": 3297
    },
    {
      "epoch": 2.110044785668586,
      "grad_norm": 1.7686492765573572,
      "learning_rate": 1.8330420079190826e-05,
      "loss": 0.4079,
      "step": 3298
    },
    {
      "epoch": 2.1106845809341013,
      "grad_norm": 2.199239035261099,
      "learning_rate": 1.8329273564592505e-05,
      "loss": 0.4417,
      "step": 3299
    },
    {
      "epoch": 2.111324376199616,
      "grad_norm": 2.184358499039298,
      "learning_rate": 1.8328126692349992e-05,
      "loss": 0.4769,
      "step": 3300
    },
    {
      "epoch": 2.1119641714651314,
      "grad_norm": 1.7941051650493258,
      "learning_rate": 1.8326979462512534e-05,
      "loss": 0.3973,
      "step": 3301
    },
    {
      "epoch": 2.112603966730646,
      "grad_norm": 1.7999505014295578,
      "learning_rate": 1.832583187512939e-05,
      "loss": 0.4471,
      "step": 3302
    },
    {
      "epoch": 2.113243761996161,
      "grad_norm": 1.8232775487772315,
      "learning_rate": 1.832468393024984e-05,
      "loss": 0.4461,
      "step": 3303
    },
    {
      "epoch": 2.1138835572616763,
      "grad_norm": 1.9815446989418088,
      "learning_rate": 1.8323535627923167e-05,
      "loss": 0.425,
      "step": 3304
    },
    {
      "epoch": 2.114523352527191,
      "grad_norm": 1.968260054578301,
      "learning_rate": 1.832238696819868e-05,
      "loss": 0.42,
      "step": 3305
    },
    {
      "epoch": 2.1151631477927064,
      "grad_norm": 1.9049813841370276,
      "learning_rate": 1.83212379511257e-05,
      "loss": 0.4107,
      "step": 3306
    },
    {
      "epoch": 2.1158029430582213,
      "grad_norm": 1.7895909554006246,
      "learning_rate": 1.8320088576753568e-05,
      "loss": 0.381,
      "step": 3307
    },
    {
      "epoch": 2.1164427383237365,
      "grad_norm": 1.8817847990100394,
      "learning_rate": 1.8318938845131632e-05,
      "loss": 0.4185,
      "step": 3308
    },
    {
      "epoch": 2.1170825335892514,
      "grad_norm": 2.0009752220407853,
      "learning_rate": 1.8317788756309257e-05,
      "loss": 0.4572,
      "step": 3309
    },
    {
      "epoch": 2.1177223288547666,
      "grad_norm": 1.939039428033452,
      "learning_rate": 1.8316638310335834e-05,
      "loss": 0.438,
      "step": 3310
    },
    {
      "epoch": 2.1183621241202815,
      "grad_norm": 2.05436071639524,
      "learning_rate": 1.8315487507260753e-05,
      "loss": 0.3912,
      "step": 3311
    },
    {
      "epoch": 2.1190019193857967,
      "grad_norm": 1.8473415538211824,
      "learning_rate": 1.831433634713343e-05,
      "loss": 0.4429,
      "step": 3312
    },
    {
      "epoch": 2.1196417146513116,
      "grad_norm": 2.139888905560329,
      "learning_rate": 1.8313184830003298e-05,
      "loss": 0.455,
      "step": 3313
    },
    {
      "epoch": 2.1202815099168264,
      "grad_norm": 1.986982703288567,
      "learning_rate": 1.8312032955919793e-05,
      "loss": 0.4465,
      "step": 3314
    },
    {
      "epoch": 2.1209213051823417,
      "grad_norm": 1.9530078537371793,
      "learning_rate": 1.8310880724932382e-05,
      "loss": 0.4182,
      "step": 3315
    },
    {
      "epoch": 2.1215611004478565,
      "grad_norm": 2.0362713789359286,
      "learning_rate": 1.8309728137090536e-05,
      "loss": 0.4389,
      "step": 3316
    },
    {
      "epoch": 2.122200895713372,
      "grad_norm": 2.13534261288139,
      "learning_rate": 1.8308575192443742e-05,
      "loss": 0.4354,
      "step": 3317
    },
    {
      "epoch": 2.1228406909788866,
      "grad_norm": 2.114586556697625,
      "learning_rate": 1.8307421891041514e-05,
      "loss": 0.4694,
      "step": 3318
    },
    {
      "epoch": 2.123480486244402,
      "grad_norm": 2.0117215970970896,
      "learning_rate": 1.8306268232933363e-05,
      "loss": 0.4253,
      "step": 3319
    },
    {
      "epoch": 2.1241202815099167,
      "grad_norm": 1.8967570600626316,
      "learning_rate": 1.8305114218168833e-05,
      "loss": 0.4472,
      "step": 3320
    },
    {
      "epoch": 2.124760076775432,
      "grad_norm": 1.8829147095095173,
      "learning_rate": 1.8303959846797473e-05,
      "loss": 0.4424,
      "step": 3321
    },
    {
      "epoch": 2.125399872040947,
      "grad_norm": 1.8537549334011252,
      "learning_rate": 1.8302805118868847e-05,
      "loss": 0.4398,
      "step": 3322
    },
    {
      "epoch": 2.126039667306462,
      "grad_norm": 1.8972062601794657,
      "learning_rate": 1.8301650034432538e-05,
      "loss": 0.4149,
      "step": 3323
    },
    {
      "epoch": 2.126679462571977,
      "grad_norm": 2.0127057362050325,
      "learning_rate": 1.8300494593538147e-05,
      "loss": 0.413,
      "step": 3324
    },
    {
      "epoch": 2.127319257837492,
      "grad_norm": 2.0347770238275693,
      "learning_rate": 1.829933879623528e-05,
      "loss": 0.4827,
      "step": 3325
    },
    {
      "epoch": 2.127959053103007,
      "grad_norm": 1.9010445160630345,
      "learning_rate": 1.829818264257357e-05,
      "loss": 0.3628,
      "step": 3326
    },
    {
      "epoch": 2.128598848368522,
      "grad_norm": 1.9946225922705607,
      "learning_rate": 1.829702613260266e-05,
      "loss": 0.4267,
      "step": 3327
    },
    {
      "epoch": 2.129238643634037,
      "grad_norm": 1.9443789288831745,
      "learning_rate": 1.8295869266372208e-05,
      "loss": 0.4405,
      "step": 3328
    },
    {
      "epoch": 2.129878438899552,
      "grad_norm": 2.0170019270867563,
      "learning_rate": 1.8294712043931885e-05,
      "loss": 0.4368,
      "step": 3329
    },
    {
      "epoch": 2.1305182341650672,
      "grad_norm": 2.132693467248236,
      "learning_rate": 1.829355446533138e-05,
      "loss": 0.4335,
      "step": 3330
    },
    {
      "epoch": 2.131158029430582,
      "grad_norm": 1.9280219564466223,
      "learning_rate": 1.8292396530620402e-05,
      "loss": 0.4507,
      "step": 3331
    },
    {
      "epoch": 2.1317978246960974,
      "grad_norm": 1.8668140850848844,
      "learning_rate": 1.8291238239848668e-05,
      "loss": 0.3962,
      "step": 3332
    },
    {
      "epoch": 2.132437619961612,
      "grad_norm": 1.9580191153301132,
      "learning_rate": 1.829007959306591e-05,
      "loss": 0.398,
      "step": 3333
    },
    {
      "epoch": 2.1330774152271275,
      "grad_norm": 1.807058254663371,
      "learning_rate": 1.8288920590321884e-05,
      "loss": 0.4026,
      "step": 3334
    },
    {
      "epoch": 2.1337172104926423,
      "grad_norm": 2.104876649647512,
      "learning_rate": 1.828776123166635e-05,
      "loss": 0.4373,
      "step": 3335
    },
    {
      "epoch": 2.1343570057581576,
      "grad_norm": 2.0288959717684336,
      "learning_rate": 1.8286601517149096e-05,
      "loss": 0.405,
      "step": 3336
    },
    {
      "epoch": 2.1349968010236724,
      "grad_norm": 2.0750059386702873,
      "learning_rate": 1.828544144681991e-05,
      "loss": 0.3979,
      "step": 3337
    },
    {
      "epoch": 2.1356365962891877,
      "grad_norm": 1.9134636886577496,
      "learning_rate": 1.828428102072861e-05,
      "loss": 0.4255,
      "step": 3338
    },
    {
      "epoch": 2.1362763915547025,
      "grad_norm": 2.0015325416903598,
      "learning_rate": 1.828312023892502e-05,
      "loss": 0.4781,
      "step": 3339
    },
    {
      "epoch": 2.1369161868202173,
      "grad_norm": 2.137068407025569,
      "learning_rate": 1.8281959101458975e-05,
      "loss": 0.4061,
      "step": 3340
    },
    {
      "epoch": 2.1375559820857326,
      "grad_norm": 2.0421812413808844,
      "learning_rate": 1.828079760838034e-05,
      "loss": 0.4448,
      "step": 3341
    },
    {
      "epoch": 2.1381957773512474,
      "grad_norm": 1.8908074032676594,
      "learning_rate": 1.8279635759738995e-05,
      "loss": 0.4576,
      "step": 3342
    },
    {
      "epoch": 2.1388355726167627,
      "grad_norm": 1.9973195994441117,
      "learning_rate": 1.827847355558481e-05,
      "loss": 0.4398,
      "step": 3343
    },
    {
      "epoch": 2.1394753678822775,
      "grad_norm": 1.8300809476563706,
      "learning_rate": 1.8277310995967702e-05,
      "loss": 0.4206,
      "step": 3344
    },
    {
      "epoch": 2.140115163147793,
      "grad_norm": 2.027811967511443,
      "learning_rate": 1.827614808093758e-05,
      "loss": 0.4504,
      "step": 3345
    },
    {
      "epoch": 2.1407549584133077,
      "grad_norm": 1.7863686481242105,
      "learning_rate": 1.827498481054438e-05,
      "loss": 0.4614,
      "step": 3346
    },
    {
      "epoch": 2.141394753678823,
      "grad_norm": 1.8625785982180487,
      "learning_rate": 1.8273821184838058e-05,
      "loss": 0.38,
      "step": 3347
    },
    {
      "epoch": 2.1420345489443378,
      "grad_norm": 1.847212732709507,
      "learning_rate": 1.827265720386857e-05,
      "loss": 0.4065,
      "step": 3348
    },
    {
      "epoch": 2.142674344209853,
      "grad_norm": 1.97912232116686,
      "learning_rate": 1.82714928676859e-05,
      "loss": 0.4474,
      "step": 3349
    },
    {
      "epoch": 2.143314139475368,
      "grad_norm": 1.9116811796247069,
      "learning_rate": 1.8270328176340032e-05,
      "loss": 0.43,
      "step": 3350
    },
    {
      "epoch": 2.1439539347408827,
      "grad_norm": 1.9604050540459712,
      "learning_rate": 1.8269163129880993e-05,
      "loss": 0.4359,
      "step": 3351
    },
    {
      "epoch": 2.144593730006398,
      "grad_norm": 1.9236940425068014,
      "learning_rate": 1.8267997728358793e-05,
      "loss": 0.4341,
      "step": 3352
    },
    {
      "epoch": 2.145233525271913,
      "grad_norm": 1.9239228677989393,
      "learning_rate": 1.826683197182348e-05,
      "loss": 0.4113,
      "step": 3353
    },
    {
      "epoch": 2.145873320537428,
      "grad_norm": 1.9919487161986027,
      "learning_rate": 1.8265665860325108e-05,
      "loss": 0.4312,
      "step": 3354
    },
    {
      "epoch": 2.146513115802943,
      "grad_norm": 1.9336790706440752,
      "learning_rate": 1.8264499393913748e-05,
      "loss": 0.4667,
      "step": 3355
    },
    {
      "epoch": 2.147152911068458,
      "grad_norm": 2.069947929289291,
      "learning_rate": 1.8263332572639486e-05,
      "loss": 0.4253,
      "step": 3356
    },
    {
      "epoch": 2.147792706333973,
      "grad_norm": 1.892118018837534,
      "learning_rate": 1.826216539655242e-05,
      "loss": 0.4371,
      "step": 3357
    },
    {
      "epoch": 2.1484325015994883,
      "grad_norm": 1.860198074063458,
      "learning_rate": 1.826099786570267e-05,
      "loss": 0.4476,
      "step": 3358
    },
    {
      "epoch": 2.149072296865003,
      "grad_norm": 1.8355379763225563,
      "learning_rate": 1.8259829980140367e-05,
      "loss": 0.4133,
      "step": 3359
    },
    {
      "epoch": 2.1497120921305184,
      "grad_norm": 1.8679695826979177,
      "learning_rate": 1.8258661739915658e-05,
      "loss": 0.463,
      "step": 3360
    },
    {
      "epoch": 2.1503518873960332,
      "grad_norm": 1.8540602239317368,
      "learning_rate": 1.8257493145078706e-05,
      "loss": 0.3882,
      "step": 3361
    },
    {
      "epoch": 2.1509916826615485,
      "grad_norm": 1.8997426408538571,
      "learning_rate": 1.8256324195679687e-05,
      "loss": 0.3872,
      "step": 3362
    },
    {
      "epoch": 2.1516314779270633,
      "grad_norm": 1.9297012930354556,
      "learning_rate": 1.8255154891768794e-05,
      "loss": 0.4378,
      "step": 3363
    },
    {
      "epoch": 2.1522712731925786,
      "grad_norm": 1.8185480372645337,
      "learning_rate": 1.8253985233396233e-05,
      "loss": 0.4456,
      "step": 3364
    },
    {
      "epoch": 2.1529110684580934,
      "grad_norm": 1.935047214628324,
      "learning_rate": 1.825281522061223e-05,
      "loss": 0.3733,
      "step": 3365
    },
    {
      "epoch": 2.1535508637236083,
      "grad_norm": 1.8101013017779528,
      "learning_rate": 1.825164485346702e-05,
      "loss": 0.3847,
      "step": 3366
    },
    {
      "epoch": 2.1541906589891235,
      "grad_norm": 1.8208180748160185,
      "learning_rate": 1.825047413201086e-05,
      "loss": 0.4119,
      "step": 3367
    },
    {
      "epoch": 2.1548304542546384,
      "grad_norm": 1.838443353319937,
      "learning_rate": 1.824930305629402e-05,
      "loss": 0.3865,
      "step": 3368
    },
    {
      "epoch": 2.1554702495201536,
      "grad_norm": 1.977376562916549,
      "learning_rate": 1.8248131626366777e-05,
      "loss": 0.4511,
      "step": 3369
    },
    {
      "epoch": 2.1561100447856685,
      "grad_norm": 1.9434366504407579,
      "learning_rate": 1.8246959842279437e-05,
      "loss": 0.4391,
      "step": 3370
    },
    {
      "epoch": 2.1567498400511838,
      "grad_norm": 1.8134389378040998,
      "learning_rate": 1.824578770408231e-05,
      "loss": 0.4168,
      "step": 3371
    },
    {
      "epoch": 2.1573896353166986,
      "grad_norm": 1.751682007887646,
      "learning_rate": 1.8244615211825728e-05,
      "loss": 0.4169,
      "step": 3372
    },
    {
      "epoch": 2.158029430582214,
      "grad_norm": 1.9323132072815916,
      "learning_rate": 1.8243442365560034e-05,
      "loss": 0.4475,
      "step": 3373
    },
    {
      "epoch": 2.1586692258477287,
      "grad_norm": 1.8672699989717225,
      "learning_rate": 1.8242269165335588e-05,
      "loss": 0.4166,
      "step": 3374
    },
    {
      "epoch": 2.159309021113244,
      "grad_norm": 1.8560454074644106,
      "learning_rate": 1.824109561120277e-05,
      "loss": 0.4251,
      "step": 3375
    },
    {
      "epoch": 2.159948816378759,
      "grad_norm": 2.015985875620693,
      "learning_rate": 1.823992170321196e-05,
      "loss": 0.4572,
      "step": 3376
    },
    {
      "epoch": 2.1605886116442736,
      "grad_norm": 1.7804623489224212,
      "learning_rate": 1.8238747441413576e-05,
      "loss": 0.4045,
      "step": 3377
    },
    {
      "epoch": 2.161228406909789,
      "grad_norm": 1.9340888254483863,
      "learning_rate": 1.823757282585803e-05,
      "loss": 0.4032,
      "step": 3378
    },
    {
      "epoch": 2.1618682021753037,
      "grad_norm": 1.9305885131016778,
      "learning_rate": 1.823639785659576e-05,
      "loss": 0.4117,
      "step": 3379
    },
    {
      "epoch": 2.162507997440819,
      "grad_norm": 2.0212996674157457,
      "learning_rate": 1.8235222533677216e-05,
      "loss": 0.4139,
      "step": 3380
    },
    {
      "epoch": 2.163147792706334,
      "grad_norm": 1.95775599918358,
      "learning_rate": 1.8234046857152868e-05,
      "loss": 0.394,
      "step": 3381
    },
    {
      "epoch": 2.163787587971849,
      "grad_norm": 1.932471372631528,
      "learning_rate": 1.8232870827073196e-05,
      "loss": 0.4103,
      "step": 3382
    },
    {
      "epoch": 2.164427383237364,
      "grad_norm": 1.9445741040934335,
      "learning_rate": 1.823169444348869e-05,
      "loss": 0.3974,
      "step": 3383
    },
    {
      "epoch": 2.165067178502879,
      "grad_norm": 2.0506180606319995,
      "learning_rate": 1.8230517706449873e-05,
      "loss": 0.4242,
      "step": 3384
    },
    {
      "epoch": 2.165706973768394,
      "grad_norm": 1.9496987640597032,
      "learning_rate": 1.8229340616007263e-05,
      "loss": 0.4211,
      "step": 3385
    },
    {
      "epoch": 2.1663467690339093,
      "grad_norm": 2.007674844730726,
      "learning_rate": 1.8228163172211412e-05,
      "loss": 0.473,
      "step": 3386
    },
    {
      "epoch": 2.166986564299424,
      "grad_norm": 1.9545637974140306,
      "learning_rate": 1.8226985375112866e-05,
      "loss": 0.4278,
      "step": 3387
    },
    {
      "epoch": 2.167626359564939,
      "grad_norm": 1.916071403832295,
      "learning_rate": 1.8225807224762205e-05,
      "loss": 0.4258,
      "step": 3388
    },
    {
      "epoch": 2.1682661548304543,
      "grad_norm": 1.8444624918005568,
      "learning_rate": 1.8224628721210013e-05,
      "loss": 0.3845,
      "step": 3389
    },
    {
      "epoch": 2.168905950095969,
      "grad_norm": 1.895055756305579,
      "learning_rate": 1.8223449864506895e-05,
      "loss": 0.4115,
      "step": 3390
    },
    {
      "epoch": 2.1695457453614844,
      "grad_norm": 2.003050557614813,
      "learning_rate": 1.8222270654703466e-05,
      "loss": 0.4216,
      "step": 3391
    },
    {
      "epoch": 2.170185540626999,
      "grad_norm": 2.0016391864467664,
      "learning_rate": 1.8221091091850365e-05,
      "loss": 0.4338,
      "step": 3392
    },
    {
      "epoch": 2.1708253358925145,
      "grad_norm": 1.9470423887957582,
      "learning_rate": 1.8219911175998235e-05,
      "loss": 0.3966,
      "step": 3393
    },
    {
      "epoch": 2.1714651311580293,
      "grad_norm": 1.937451392609162,
      "learning_rate": 1.8218730907197743e-05,
      "loss": 0.514,
      "step": 3394
    },
    {
      "epoch": 2.1721049264235446,
      "grad_norm": 1.825060232131321,
      "learning_rate": 1.821755028549956e-05,
      "loss": 0.3706,
      "step": 3395
    },
    {
      "epoch": 2.1727447216890594,
      "grad_norm": 1.9506165652009881,
      "learning_rate": 1.821636931095439e-05,
      "loss": 0.4282,
      "step": 3396
    },
    {
      "epoch": 2.1733845169545747,
      "grad_norm": 1.9716254024284396,
      "learning_rate": 1.821518798361294e-05,
      "loss": 0.4255,
      "step": 3397
    },
    {
      "epoch": 2.1740243122200895,
      "grad_norm": 2.1158800069366377,
      "learning_rate": 1.8214006303525927e-05,
      "loss": 0.4786,
      "step": 3398
    },
    {
      "epoch": 2.174664107485605,
      "grad_norm": 1.9464315701968695,
      "learning_rate": 1.8212824270744095e-05,
      "loss": 0.4269,
      "step": 3399
    },
    {
      "epoch": 2.1753039027511196,
      "grad_norm": 1.9179402346363643,
      "learning_rate": 1.8211641885318202e-05,
      "loss": 0.4479,
      "step": 3400
    },
    {
      "epoch": 2.175943698016635,
      "grad_norm": 1.8235750478893449,
      "learning_rate": 1.821045914729901e-05,
      "loss": 0.3866,
      "step": 3401
    },
    {
      "epoch": 2.1765834932821497,
      "grad_norm": 2.027618232734173,
      "learning_rate": 1.820927605673731e-05,
      "loss": 0.426,
      "step": 3402
    },
    {
      "epoch": 2.1772232885476646,
      "grad_norm": 1.9223872957130623,
      "learning_rate": 1.82080926136839e-05,
      "loss": 0.3916,
      "step": 3403
    },
    {
      "epoch": 2.17786308381318,
      "grad_norm": 1.9627366994366893,
      "learning_rate": 1.8206908818189594e-05,
      "loss": 0.4161,
      "step": 3404
    },
    {
      "epoch": 2.1785028790786947,
      "grad_norm": 2.03361384299559,
      "learning_rate": 1.8205724670305218e-05,
      "loss": 0.4264,
      "step": 3405
    },
    {
      "epoch": 2.17914267434421,
      "grad_norm": 1.9453596020719908,
      "learning_rate": 1.8204540170081625e-05,
      "loss": 0.3885,
      "step": 3406
    },
    {
      "epoch": 2.1797824696097248,
      "grad_norm": 2.118837776020821,
      "learning_rate": 1.820335531756967e-05,
      "loss": 0.4775,
      "step": 3407
    },
    {
      "epoch": 2.18042226487524,
      "grad_norm": 2.7471610639334383,
      "learning_rate": 1.8202170112820233e-05,
      "loss": 0.4427,
      "step": 3408
    },
    {
      "epoch": 2.181062060140755,
      "grad_norm": 1.994642394090322,
      "learning_rate": 1.82009845558842e-05,
      "loss": 0.4401,
      "step": 3409
    },
    {
      "epoch": 2.18170185540627,
      "grad_norm": 1.8657991247040258,
      "learning_rate": 1.819979864681248e-05,
      "loss": 0.4465,
      "step": 3410
    },
    {
      "epoch": 2.182341650671785,
      "grad_norm": 1.9075575034219945,
      "learning_rate": 1.8198612385655993e-05,
      "loss": 0.4878,
      "step": 3411
    },
    {
      "epoch": 2.1829814459373003,
      "grad_norm": 1.8510792297410201,
      "learning_rate": 1.819742577246567e-05,
      "loss": 0.413,
      "step": 3412
    },
    {
      "epoch": 2.183621241202815,
      "grad_norm": 1.863173984032734,
      "learning_rate": 1.819623880729247e-05,
      "loss": 0.408,
      "step": 3413
    },
    {
      "epoch": 2.18426103646833,
      "grad_norm": 1.8563303199713972,
      "learning_rate": 1.8195051490187356e-05,
      "loss": 0.4173,
      "step": 3414
    },
    {
      "epoch": 2.184900831733845,
      "grad_norm": 1.9919861109521695,
      "learning_rate": 1.8193863821201306e-05,
      "loss": 0.458,
      "step": 3415
    },
    {
      "epoch": 2.18554062699936,
      "grad_norm": 1.8883627269014562,
      "learning_rate": 1.8192675800385323e-05,
      "loss": 0.4003,
      "step": 3416
    },
    {
      "epoch": 2.1861804222648753,
      "grad_norm": 1.9851025615143265,
      "learning_rate": 1.8191487427790413e-05,
      "loss": 0.419,
      "step": 3417
    },
    {
      "epoch": 2.18682021753039,
      "grad_norm": 2.007428475507898,
      "learning_rate": 1.8190298703467604e-05,
      "loss": 0.4387,
      "step": 3418
    },
    {
      "epoch": 2.1874600127959054,
      "grad_norm": 1.8582213642655743,
      "learning_rate": 1.818910962746794e-05,
      "loss": 0.4388,
      "step": 3419
    },
    {
      "epoch": 2.1880998080614202,
      "grad_norm": 1.8883473381743079,
      "learning_rate": 1.8187920199842473e-05,
      "loss": 0.4057,
      "step": 3420
    },
    {
      "epoch": 2.1887396033269355,
      "grad_norm": 2.114441977766257,
      "learning_rate": 1.818673042064228e-05,
      "loss": 0.4131,
      "step": 3421
    },
    {
      "epoch": 2.1893793985924503,
      "grad_norm": 1.997960298339548,
      "learning_rate": 1.8185540289918443e-05,
      "loss": 0.4656,
      "step": 3422
    },
    {
      "epoch": 2.1900191938579656,
      "grad_norm": 1.9425612433146173,
      "learning_rate": 1.818434980772207e-05,
      "loss": 0.4172,
      "step": 3423
    },
    {
      "epoch": 2.1906589891234804,
      "grad_norm": 1.9920013214059282,
      "learning_rate": 1.8183158974104275e-05,
      "loss": 0.4565,
      "step": 3424
    },
    {
      "epoch": 2.1912987843889957,
      "grad_norm": 1.8619632730193998,
      "learning_rate": 1.8181967789116188e-05,
      "loss": 0.4067,
      "step": 3425
    },
    {
      "epoch": 2.1919385796545106,
      "grad_norm": 1.9430436358682477,
      "learning_rate": 1.8180776252808958e-05,
      "loss": 0.4176,
      "step": 3426
    },
    {
      "epoch": 2.1925783749200254,
      "grad_norm": 2.1292540433996696,
      "learning_rate": 1.817958436523375e-05,
      "loss": 0.3943,
      "step": 3427
    },
    {
      "epoch": 2.1932181701855407,
      "grad_norm": 1.798569517385255,
      "learning_rate": 1.817839212644174e-05,
      "loss": 0.405,
      "step": 3428
    },
    {
      "epoch": 2.1938579654510555,
      "grad_norm": 1.7747839867695854,
      "learning_rate": 1.8177199536484123e-05,
      "loss": 0.4118,
      "step": 3429
    },
    {
      "epoch": 2.1944977607165708,
      "grad_norm": 2.1353494480671333,
      "learning_rate": 1.81760065954121e-05,
      "loss": 0.447,
      "step": 3430
    },
    {
      "epoch": 2.1951375559820856,
      "grad_norm": 1.9211064767416175,
      "learning_rate": 1.81748133032769e-05,
      "loss": 0.4304,
      "step": 3431
    },
    {
      "epoch": 2.195777351247601,
      "grad_norm": 1.860639889117535,
      "learning_rate": 1.8173619660129754e-05,
      "loss": 0.4422,
      "step": 3432
    },
    {
      "epoch": 2.1964171465131157,
      "grad_norm": 1.8690417257525034,
      "learning_rate": 1.8172425666021926e-05,
      "loss": 0.4405,
      "step": 3433
    },
    {
      "epoch": 2.197056941778631,
      "grad_norm": 1.9464881826509668,
      "learning_rate": 1.8171231321004675e-05,
      "loss": 0.4476,
      "step": 3434
    },
    {
      "epoch": 2.197696737044146,
      "grad_norm": 2.0053714537547433,
      "learning_rate": 1.8170036625129284e-05,
      "loss": 0.4321,
      "step": 3435
    },
    {
      "epoch": 2.198336532309661,
      "grad_norm": 1.8677571359337324,
      "learning_rate": 1.816884157844706e-05,
      "loss": 0.3801,
      "step": 3436
    },
    {
      "epoch": 2.198976327575176,
      "grad_norm": 2.0370871949229667,
      "learning_rate": 1.8167646181009302e-05,
      "loss": 0.4394,
      "step": 3437
    },
    {
      "epoch": 2.199616122840691,
      "grad_norm": 2.000028545213205,
      "learning_rate": 1.816645043286735e-05,
      "loss": 0.4402,
      "step": 3438
    },
    {
      "epoch": 2.200255918106206,
      "grad_norm": 2.1372857319325522,
      "learning_rate": 1.8165254334072543e-05,
      "loss": 0.4473,
      "step": 3439
    },
    {
      "epoch": 2.200895713371721,
      "grad_norm": 2.009005876576016,
      "learning_rate": 1.816405788467624e-05,
      "loss": 0.4263,
      "step": 3440
    },
    {
      "epoch": 2.201535508637236,
      "grad_norm": 1.8885886670336667,
      "learning_rate": 1.8162861084729816e-05,
      "loss": 0.3957,
      "step": 3441
    },
    {
      "epoch": 2.202175303902751,
      "grad_norm": 1.9530596464874712,
      "learning_rate": 1.8161663934284654e-05,
      "loss": 0.4246,
      "step": 3442
    },
    {
      "epoch": 2.2028150991682662,
      "grad_norm": 1.8335171180522079,
      "learning_rate": 1.8160466433392163e-05,
      "loss": 0.4255,
      "step": 3443
    },
    {
      "epoch": 2.203454894433781,
      "grad_norm": 1.7709282715094607,
      "learning_rate": 1.815926858210376e-05,
      "loss": 0.4007,
      "step": 3444
    },
    {
      "epoch": 2.2040946896992963,
      "grad_norm": 1.889916894128235,
      "learning_rate": 1.815807038047088e-05,
      "loss": 0.4,
      "step": 3445
    },
    {
      "epoch": 2.204734484964811,
      "grad_norm": 1.9749396929906402,
      "learning_rate": 1.8156871828544965e-05,
      "loss": 0.4086,
      "step": 3446
    },
    {
      "epoch": 2.2053742802303264,
      "grad_norm": 1.956517259201126,
      "learning_rate": 1.8155672926377488e-05,
      "loss": 0.3936,
      "step": 3447
    },
    {
      "epoch": 2.2060140754958413,
      "grad_norm": 1.9579913939436846,
      "learning_rate": 1.8154473674019922e-05,
      "loss": 0.4651,
      "step": 3448
    },
    {
      "epoch": 2.2066538707613566,
      "grad_norm": 1.9088549741025975,
      "learning_rate": 1.815327407152376e-05,
      "loss": 0.4237,
      "step": 3449
    },
    {
      "epoch": 2.2072936660268714,
      "grad_norm": 2.1859563310625147,
      "learning_rate": 1.8152074118940516e-05,
      "loss": 0.4408,
      "step": 3450
    },
    {
      "epoch": 2.207933461292386,
      "grad_norm": 2.019371034542143,
      "learning_rate": 1.815087381632171e-05,
      "loss": 0.4558,
      "step": 3451
    },
    {
      "epoch": 2.2085732565579015,
      "grad_norm": 2.010423315396098,
      "learning_rate": 1.814967316371888e-05,
      "loss": 0.4282,
      "step": 3452
    },
    {
      "epoch": 2.2092130518234163,
      "grad_norm": 1.9837652421099183,
      "learning_rate": 1.8148472161183584e-05,
      "loss": 0.4121,
      "step": 3453
    },
    {
      "epoch": 2.2098528470889316,
      "grad_norm": 1.9220935006184785,
      "learning_rate": 1.8147270808767386e-05,
      "loss": 0.4298,
      "step": 3454
    },
    {
      "epoch": 2.2104926423544464,
      "grad_norm": 2.1997094079473674,
      "learning_rate": 1.8146069106521875e-05,
      "loss": 0.4499,
      "step": 3455
    },
    {
      "epoch": 2.2111324376199617,
      "grad_norm": 1.8228433256082357,
      "learning_rate": 1.8144867054498645e-05,
      "loss": 0.3721,
      "step": 3456
    },
    {
      "epoch": 2.2117722328854765,
      "grad_norm": 1.860781524190097,
      "learning_rate": 1.8143664652749314e-05,
      "loss": 0.3661,
      "step": 3457
    },
    {
      "epoch": 2.212412028150992,
      "grad_norm": 1.8801491884092831,
      "learning_rate": 1.8142461901325507e-05,
      "loss": 0.4446,
      "step": 3458
    },
    {
      "epoch": 2.2130518234165066,
      "grad_norm": 1.9690923108756837,
      "learning_rate": 1.814125880027887e-05,
      "loss": 0.4261,
      "step": 3459
    },
    {
      "epoch": 2.213691618682022,
      "grad_norm": 1.9056005209459324,
      "learning_rate": 1.8140055349661065e-05,
      "loss": 0.4575,
      "step": 3460
    },
    {
      "epoch": 2.2143314139475367,
      "grad_norm": 1.7944581795535854,
      "learning_rate": 1.8138851549523764e-05,
      "loss": 0.4101,
      "step": 3461
    },
    {
      "epoch": 2.214971209213052,
      "grad_norm": 1.8598128126942317,
      "learning_rate": 1.8137647399918653e-05,
      "loss": 0.4254,
      "step": 3462
    },
    {
      "epoch": 2.215611004478567,
      "grad_norm": 1.838585083621075,
      "learning_rate": 1.813644290089744e-05,
      "loss": 0.4217,
      "step": 3463
    },
    {
      "epoch": 2.216250799744082,
      "grad_norm": 1.9221206608542576,
      "learning_rate": 1.813523805251184e-05,
      "loss": 0.4577,
      "step": 3464
    },
    {
      "epoch": 2.216890595009597,
      "grad_norm": 1.7722062963507557,
      "learning_rate": 1.8134032854813592e-05,
      "loss": 0.3974,
      "step": 3465
    },
    {
      "epoch": 2.217530390275112,
      "grad_norm": 1.866769010587804,
      "learning_rate": 1.813282730785444e-05,
      "loss": 0.4288,
      "step": 3466
    },
    {
      "epoch": 2.218170185540627,
      "grad_norm": 1.8403066344992431,
      "learning_rate": 1.8131621411686153e-05,
      "loss": 0.4117,
      "step": 3467
    },
    {
      "epoch": 2.218809980806142,
      "grad_norm": 1.8748605063721764,
      "learning_rate": 1.8130415166360506e-05,
      "loss": 0.3791,
      "step": 3468
    },
    {
      "epoch": 2.219449776071657,
      "grad_norm": 1.9943444794984042,
      "learning_rate": 1.8129208571929295e-05,
      "loss": 0.3834,
      "step": 3469
    },
    {
      "epoch": 2.220089571337172,
      "grad_norm": 2.1955245018129035,
      "learning_rate": 1.812800162844433e-05,
      "loss": 0.4226,
      "step": 3470
    },
    {
      "epoch": 2.2207293666026873,
      "grad_norm": 1.9557233606794384,
      "learning_rate": 1.8126794335957432e-05,
      "loss": 0.4008,
      "step": 3471
    },
    {
      "epoch": 2.221369161868202,
      "grad_norm": 1.7892675497552601,
      "learning_rate": 1.8125586694520443e-05,
      "loss": 0.3529,
      "step": 3472
    },
    {
      "epoch": 2.2220089571337174,
      "grad_norm": 2.0223921229020183,
      "learning_rate": 1.8124378704185213e-05,
      "loss": 0.4431,
      "step": 3473
    },
    {
      "epoch": 2.222648752399232,
      "grad_norm": 1.7837781621371533,
      "learning_rate": 1.8123170365003618e-05,
      "loss": 0.3757,
      "step": 3474
    },
    {
      "epoch": 2.2232885476647475,
      "grad_norm": 2.035757457215904,
      "learning_rate": 1.8121961677027533e-05,
      "loss": 0.4314,
      "step": 3475
    },
    {
      "epoch": 2.2239283429302623,
      "grad_norm": 2.0891620479249684,
      "learning_rate": 1.8120752640308864e-05,
      "loss": 0.4276,
      "step": 3476
    },
    {
      "epoch": 2.224568138195777,
      "grad_norm": 2.0628914832812297,
      "learning_rate": 1.811954325489952e-05,
      "loss": 0.4022,
      "step": 3477
    },
    {
      "epoch": 2.2252079334612924,
      "grad_norm": 1.9473032196850804,
      "learning_rate": 1.8118333520851438e-05,
      "loss": 0.4176,
      "step": 3478
    },
    {
      "epoch": 2.2258477287268073,
      "grad_norm": 1.9069511540810669,
      "learning_rate": 1.8117123438216547e-05,
      "loss": 0.4222,
      "step": 3479
    },
    {
      "epoch": 2.2264875239923225,
      "grad_norm": 1.9272235755448535,
      "learning_rate": 1.8115913007046822e-05,
      "loss": 0.3879,
      "step": 3480
    },
    {
      "epoch": 2.2271273192578374,
      "grad_norm": 2.0166315440075526,
      "learning_rate": 1.811470222739423e-05,
      "loss": 0.4839,
      "step": 3481
    },
    {
      "epoch": 2.2277671145233526,
      "grad_norm": 2.051288797366992,
      "learning_rate": 1.8113491099310755e-05,
      "loss": 0.436,
      "step": 3482
    },
    {
      "epoch": 2.2284069097888675,
      "grad_norm": 1.9967937401234088,
      "learning_rate": 1.8112279622848407e-05,
      "loss": 0.412,
      "step": 3483
    },
    {
      "epoch": 2.2290467050543827,
      "grad_norm": 1.7595000522131865,
      "learning_rate": 1.81110677980592e-05,
      "loss": 0.4329,
      "step": 3484
    },
    {
      "epoch": 2.2296865003198976,
      "grad_norm": 1.844393291108033,
      "learning_rate": 1.8109855624995174e-05,
      "loss": 0.4957,
      "step": 3485
    },
    {
      "epoch": 2.230326295585413,
      "grad_norm": 1.9395996318966526,
      "learning_rate": 1.8108643103708372e-05,
      "loss": 0.4037,
      "step": 3486
    },
    {
      "epoch": 2.2309660908509277,
      "grad_norm": 2.217045235153912,
      "learning_rate": 1.810743023425086e-05,
      "loss": 0.4597,
      "step": 3487
    },
    {
      "epoch": 2.2316058861164425,
      "grad_norm": 2.1028857062528767,
      "learning_rate": 1.8106217016674717e-05,
      "loss": 0.4571,
      "step": 3488
    },
    {
      "epoch": 2.232245681381958,
      "grad_norm": 1.763425397807173,
      "learning_rate": 1.810500345103203e-05,
      "loss": 0.3837,
      "step": 3489
    },
    {
      "epoch": 2.2328854766474726,
      "grad_norm": 1.8461408557902603,
      "learning_rate": 1.8103789537374918e-05,
      "loss": 0.4168,
      "step": 3490
    },
    {
      "epoch": 2.233525271912988,
      "grad_norm": 2.0421111770026883,
      "learning_rate": 1.8102575275755496e-05,
      "loss": 0.3994,
      "step": 3491
    },
    {
      "epoch": 2.2341650671785027,
      "grad_norm": 1.961598691079886,
      "learning_rate": 1.8101360666225907e-05,
      "loss": 0.4611,
      "step": 3492
    },
    {
      "epoch": 2.234804862444018,
      "grad_norm": 2.0706590587458535,
      "learning_rate": 1.8100145708838298e-05,
      "loss": 0.434,
      "step": 3493
    },
    {
      "epoch": 2.235444657709533,
      "grad_norm": 1.9390019318263017,
      "learning_rate": 1.8098930403644846e-05,
      "loss": 0.411,
      "step": 3494
    },
    {
      "epoch": 2.236084452975048,
      "grad_norm": 1.8731750802369855,
      "learning_rate": 1.8097714750697725e-05,
      "loss": 0.3783,
      "step": 3495
    },
    {
      "epoch": 2.236724248240563,
      "grad_norm": 1.944534829240339,
      "learning_rate": 1.809649875004914e-05,
      "loss": 0.4313,
      "step": 3496
    },
    {
      "epoch": 2.237364043506078,
      "grad_norm": 1.8769590673060172,
      "learning_rate": 1.8095282401751305e-05,
      "loss": 0.4497,
      "step": 3497
    },
    {
      "epoch": 2.238003838771593,
      "grad_norm": 2.0789577906831256,
      "learning_rate": 1.809406570585644e-05,
      "loss": 0.3934,
      "step": 3498
    },
    {
      "epoch": 2.2386436340371083,
      "grad_norm": 1.9249103072528226,
      "learning_rate": 1.809284866241679e-05,
      "loss": 0.4781,
      "step": 3499
    },
    {
      "epoch": 2.239283429302623,
      "grad_norm": 2.082932088076661,
      "learning_rate": 1.8091631271484617e-05,
      "loss": 0.432,
      "step": 3500
    },
    {
      "epoch": 2.2399232245681384,
      "grad_norm": 1.8923509844401596,
      "learning_rate": 1.809041353311219e-05,
      "loss": 0.4379,
      "step": 3501
    },
    {
      "epoch": 2.2405630198336532,
      "grad_norm": 1.7379964679542657,
      "learning_rate": 1.80891954473518e-05,
      "loss": 0.395,
      "step": 3502
    },
    {
      "epoch": 2.241202815099168,
      "grad_norm": 1.9150590053713872,
      "learning_rate": 1.8087977014255745e-05,
      "loss": 0.4514,
      "step": 3503
    },
    {
      "epoch": 2.2418426103646834,
      "grad_norm": 1.8041639069528812,
      "learning_rate": 1.8086758233876343e-05,
      "loss": 0.4184,
      "step": 3504
    },
    {
      "epoch": 2.242482405630198,
      "grad_norm": 1.8551334325574835,
      "learning_rate": 1.808553910626593e-05,
      "loss": 0.438,
      "step": 3505
    },
    {
      "epoch": 2.2431222008957135,
      "grad_norm": 1.960147983806571,
      "learning_rate": 1.808431963147685e-05,
      "loss": 0.4482,
      "step": 3506
    },
    {
      "epoch": 2.2437619961612283,
      "grad_norm": 2.217504806403352,
      "learning_rate": 1.8083099809561467e-05,
      "loss": 0.4678,
      "step": 3507
    },
    {
      "epoch": 2.2444017914267436,
      "grad_norm": 1.7480105229754912,
      "learning_rate": 1.8081879640572156e-05,
      "loss": 0.4097,
      "step": 3508
    },
    {
      "epoch": 2.2450415866922584,
      "grad_norm": 1.837162268876998,
      "learning_rate": 1.808065912456131e-05,
      "loss": 0.4052,
      "step": 3509
    },
    {
      "epoch": 2.2456813819577737,
      "grad_norm": 1.9678146868720556,
      "learning_rate": 1.8079438261581334e-05,
      "loss": 0.4099,
      "step": 3510
    },
    {
      "epoch": 2.2463211772232885,
      "grad_norm": 1.9854002686740218,
      "learning_rate": 1.8078217051684652e-05,
      "loss": 0.4543,
      "step": 3511
    },
    {
      "epoch": 2.2469609724888038,
      "grad_norm": 1.9360531383675121,
      "learning_rate": 1.8076995494923702e-05,
      "loss": 0.4228,
      "step": 3512
    },
    {
      "epoch": 2.2476007677543186,
      "grad_norm": 1.93743224900211,
      "learning_rate": 1.8075773591350933e-05,
      "loss": 0.4613,
      "step": 3513
    },
    {
      "epoch": 2.2482405630198334,
      "grad_norm": 1.9845820622299293,
      "learning_rate": 1.807455134101881e-05,
      "loss": 0.4036,
      "step": 3514
    },
    {
      "epoch": 2.2488803582853487,
      "grad_norm": 1.9831748392219757,
      "learning_rate": 1.8073328743979814e-05,
      "loss": 0.4048,
      "step": 3515
    },
    {
      "epoch": 2.2495201535508635,
      "grad_norm": 1.918295082919614,
      "learning_rate": 1.8072105800286445e-05,
      "loss": 0.4105,
      "step": 3516
    },
    {
      "epoch": 2.250159948816379,
      "grad_norm": 1.8922542314857818,
      "learning_rate": 1.8070882509991213e-05,
      "loss": 0.4134,
      "step": 3517
    },
    {
      "epoch": 2.2507997440818936,
      "grad_norm": 1.9193055940074468,
      "learning_rate": 1.8069658873146644e-05,
      "loss": 0.4326,
      "step": 3518
    },
    {
      "epoch": 2.251439539347409,
      "grad_norm": 2.061308210081607,
      "learning_rate": 1.8068434889805276e-05,
      "loss": 0.4388,
      "step": 3519
    },
    {
      "epoch": 2.2520793346129238,
      "grad_norm": 1.9782281662506886,
      "learning_rate": 1.8067210560019664e-05,
      "loss": 0.4215,
      "step": 3520
    },
    {
      "epoch": 2.252719129878439,
      "grad_norm": 1.8471458979501358,
      "learning_rate": 1.8065985883842385e-05,
      "loss": 0.4042,
      "step": 3521
    },
    {
      "epoch": 2.253358925143954,
      "grad_norm": 1.8353859877360137,
      "learning_rate": 1.806476086132602e-05,
      "loss": 0.3951,
      "step": 3522
    },
    {
      "epoch": 2.253998720409469,
      "grad_norm": 2.2106574331552786,
      "learning_rate": 1.8063535492523167e-05,
      "loss": 0.505,
      "step": 3523
    },
    {
      "epoch": 2.254638515674984,
      "grad_norm": 1.9420015667008197,
      "learning_rate": 1.8062309777486446e-05,
      "loss": 0.3784,
      "step": 3524
    },
    {
      "epoch": 2.255278310940499,
      "grad_norm": 1.9043542722851936,
      "learning_rate": 1.8061083716268484e-05,
      "loss": 0.4306,
      "step": 3525
    },
    {
      "epoch": 2.255918106206014,
      "grad_norm": 1.8217796695901805,
      "learning_rate": 1.805985730892193e-05,
      "loss": 0.4237,
      "step": 3526
    },
    {
      "epoch": 2.2565579014715293,
      "grad_norm": 1.9824919825797203,
      "learning_rate": 1.8058630555499435e-05,
      "loss": 0.3876,
      "step": 3527
    },
    {
      "epoch": 2.257197696737044,
      "grad_norm": 2.1061635846948312,
      "learning_rate": 1.8057403456053685e-05,
      "loss": 0.4621,
      "step": 3528
    },
    {
      "epoch": 2.257837492002559,
      "grad_norm": 2.1875020380817647,
      "learning_rate": 1.805617601063736e-05,
      "loss": 0.46,
      "step": 3529
    },
    {
      "epoch": 2.2584772872680743,
      "grad_norm": 1.8364821024029538,
      "learning_rate": 1.805494821930317e-05,
      "loss": 0.3745,
      "step": 3530
    },
    {
      "epoch": 2.259117082533589,
      "grad_norm": 1.7914903877731605,
      "learning_rate": 1.805372008210383e-05,
      "loss": 0.3965,
      "step": 3531
    },
    {
      "epoch": 2.2597568777991044,
      "grad_norm": 1.852040780378072,
      "learning_rate": 1.8052491599092078e-05,
      "loss": 0.4497,
      "step": 3532
    },
    {
      "epoch": 2.260396673064619,
      "grad_norm": 1.9510320533167136,
      "learning_rate": 1.805126277032066e-05,
      "loss": 0.4438,
      "step": 3533
    },
    {
      "epoch": 2.2610364683301345,
      "grad_norm": 1.9015679874773699,
      "learning_rate": 1.8050033595842337e-05,
      "loss": 0.4324,
      "step": 3534
    },
    {
      "epoch": 2.2616762635956493,
      "grad_norm": 2.02459802521425,
      "learning_rate": 1.8048804075709895e-05,
      "loss": 0.4277,
      "step": 3535
    },
    {
      "epoch": 2.2623160588611646,
      "grad_norm": 1.9428923886085214,
      "learning_rate": 1.8047574209976126e-05,
      "loss": 0.405,
      "step": 3536
    },
    {
      "epoch": 2.2629558541266794,
      "grad_norm": 1.948768069535332,
      "learning_rate": 1.8046343998693832e-05,
      "loss": 0.3943,
      "step": 3537
    },
    {
      "epoch": 2.2635956493921947,
      "grad_norm": 1.9812986597771882,
      "learning_rate": 1.8045113441915842e-05,
      "loss": 0.4001,
      "step": 3538
    },
    {
      "epoch": 2.2642354446577095,
      "grad_norm": 2.092833569635129,
      "learning_rate": 1.8043882539694994e-05,
      "loss": 0.4271,
      "step": 3539
    },
    {
      "epoch": 2.2648752399232244,
      "grad_norm": 1.8287193119151899,
      "learning_rate": 1.8042651292084135e-05,
      "loss": 0.4169,
      "step": 3540
    },
    {
      "epoch": 2.2655150351887396,
      "grad_norm": 2.0199182721524123,
      "learning_rate": 1.8041419699136134e-05,
      "loss": 0.4531,
      "step": 3541
    },
    {
      "epoch": 2.2661548304542545,
      "grad_norm": 1.913686018413337,
      "learning_rate": 1.804018776090388e-05,
      "loss": 0.3947,
      "step": 3542
    },
    {
      "epoch": 2.2667946257197698,
      "grad_norm": 1.8532705288914102,
      "learning_rate": 1.8038955477440266e-05,
      "loss": 0.3957,
      "step": 3543
    },
    {
      "epoch": 2.2674344209852846,
      "grad_norm": 1.8895964376523398,
      "learning_rate": 1.8037722848798203e-05,
      "loss": 0.3752,
      "step": 3544
    },
    {
      "epoch": 2.2680742162508,
      "grad_norm": 2.0171977052549273,
      "learning_rate": 1.803648987503062e-05,
      "loss": 0.4147,
      "step": 3545
    },
    {
      "epoch": 2.2687140115163147,
      "grad_norm": 2.2001466720165204,
      "learning_rate": 1.8035256556190456e-05,
      "loss": 0.4043,
      "step": 3546
    },
    {
      "epoch": 2.26935380678183,
      "grad_norm": 1.9684559457956627,
      "learning_rate": 1.803402289233067e-05,
      "loss": 0.417,
      "step": 3547
    },
    {
      "epoch": 2.269993602047345,
      "grad_norm": 1.7841234995053414,
      "learning_rate": 1.8032788883504233e-05,
      "loss": 0.4239,
      "step": 3548
    },
    {
      "epoch": 2.27063339731286,
      "grad_norm": 2.033667340052782,
      "learning_rate": 1.803155452976413e-05,
      "loss": 0.4328,
      "step": 3549
    },
    {
      "epoch": 2.271273192578375,
      "grad_norm": 1.841620879518432,
      "learning_rate": 1.8030319831163363e-05,
      "loss": 0.4319,
      "step": 3550
    },
    {
      "epoch": 2.2719129878438897,
      "grad_norm": 2.107647650711563,
      "learning_rate": 1.8029084787754945e-05,
      "loss": 0.3927,
      "step": 3551
    },
    {
      "epoch": 2.272552783109405,
      "grad_norm": 1.9003745321974603,
      "learning_rate": 1.802784939959191e-05,
      "loss": 0.4015,
      "step": 3552
    },
    {
      "epoch": 2.27319257837492,
      "grad_norm": 2.0679323986764824,
      "learning_rate": 1.80266136667273e-05,
      "loss": 0.448,
      "step": 3553
    },
    {
      "epoch": 2.273832373640435,
      "grad_norm": 1.8101177720597936,
      "learning_rate": 1.8025377589214183e-05,
      "loss": 0.3553,
      "step": 3554
    },
    {
      "epoch": 2.27447216890595,
      "grad_norm": 1.9190261358940768,
      "learning_rate": 1.8024141167105626e-05,
      "loss": 0.4072,
      "step": 3555
    },
    {
      "epoch": 2.275111964171465,
      "grad_norm": 1.749221350384138,
      "learning_rate": 1.802290440045472e-05,
      "loss": 0.4067,
      "step": 3556
    },
    {
      "epoch": 2.27575175943698,
      "grad_norm": 1.8513075206486531,
      "learning_rate": 1.802166728931457e-05,
      "loss": 0.3909,
      "step": 3557
    },
    {
      "epoch": 2.2763915547024953,
      "grad_norm": 2.0626216299113507,
      "learning_rate": 1.8020429833738298e-05,
      "loss": 0.3966,
      "step": 3558
    },
    {
      "epoch": 2.27703134996801,
      "grad_norm": 2.0035481368356756,
      "learning_rate": 1.8019192033779034e-05,
      "loss": 0.4437,
      "step": 3559
    },
    {
      "epoch": 2.2776711452335254,
      "grad_norm": 2.118829193258327,
      "learning_rate": 1.801795388948993e-05,
      "loss": 0.4172,
      "step": 3560
    },
    {
      "epoch": 2.2783109404990403,
      "grad_norm": 1.9594062212248127,
      "learning_rate": 1.801671540092415e-05,
      "loss": 0.4097,
      "step": 3561
    },
    {
      "epoch": 2.2789507357645555,
      "grad_norm": 1.9067477323980657,
      "learning_rate": 1.8015476568134872e-05,
      "loss": 0.4011,
      "step": 3562
    },
    {
      "epoch": 2.2795905310300704,
      "grad_norm": 2.0748426747795197,
      "learning_rate": 1.8014237391175284e-05,
      "loss": 0.4669,
      "step": 3563
    },
    {
      "epoch": 2.2802303262955856,
      "grad_norm": 1.9139599886328909,
      "learning_rate": 1.8012997870098603e-05,
      "loss": 0.4008,
      "step": 3564
    },
    {
      "epoch": 2.2808701215611005,
      "grad_norm": 1.8741478229145223,
      "learning_rate": 1.8011758004958047e-05,
      "loss": 0.434,
      "step": 3565
    },
    {
      "epoch": 2.2815099168266153,
      "grad_norm": 1.9956981359007766,
      "learning_rate": 1.801051779580685e-05,
      "loss": 0.4412,
      "step": 3566
    },
    {
      "epoch": 2.2821497120921306,
      "grad_norm": 1.8361863300853956,
      "learning_rate": 1.8009277242698274e-05,
      "loss": 0.3701,
      "step": 3567
    },
    {
      "epoch": 2.2827895073576454,
      "grad_norm": 1.8709826042017605,
      "learning_rate": 1.800803634568558e-05,
      "loss": 0.3942,
      "step": 3568
    },
    {
      "epoch": 2.2834293026231607,
      "grad_norm": 1.7276951323772223,
      "learning_rate": 1.8006795104822046e-05,
      "loss": 0.3957,
      "step": 3569
    },
    {
      "epoch": 2.2840690978886755,
      "grad_norm": 1.7766422914979971,
      "learning_rate": 1.800555352016098e-05,
      "loss": 0.4132,
      "step": 3570
    },
    {
      "epoch": 2.284708893154191,
      "grad_norm": 1.8479292343582796,
      "learning_rate": 1.8004311591755682e-05,
      "loss": 0.4217,
      "step": 3571
    },
    {
      "epoch": 2.2853486884197056,
      "grad_norm": 2.2778270465175496,
      "learning_rate": 1.800306931965948e-05,
      "loss": 0.4266,
      "step": 3572
    },
    {
      "epoch": 2.285988483685221,
      "grad_norm": 1.7991464472913514,
      "learning_rate": 1.8001826703925723e-05,
      "loss": 0.398,
      "step": 3573
    },
    {
      "epoch": 2.2866282789507357,
      "grad_norm": 1.9262031826870494,
      "learning_rate": 1.800058374460776e-05,
      "loss": 0.3953,
      "step": 3574
    },
    {
      "epoch": 2.287268074216251,
      "grad_norm": 1.7582540723400566,
      "learning_rate": 1.7999340441758963e-05,
      "loss": 0.3563,
      "step": 3575
    },
    {
      "epoch": 2.287907869481766,
      "grad_norm": 2.0165931761044966,
      "learning_rate": 1.799809679543272e-05,
      "loss": 0.4436,
      "step": 3576
    },
    {
      "epoch": 2.2885476647472807,
      "grad_norm": 2.270982573122261,
      "learning_rate": 1.7996852805682424e-05,
      "loss": 0.4525,
      "step": 3577
    },
    {
      "epoch": 2.289187460012796,
      "grad_norm": 1.8031889118719155,
      "learning_rate": 1.7995608472561493e-05,
      "loss": 0.3712,
      "step": 3578
    },
    {
      "epoch": 2.2898272552783108,
      "grad_norm": 1.9340649559566885,
      "learning_rate": 1.799436379612336e-05,
      "loss": 0.4119,
      "step": 3579
    },
    {
      "epoch": 2.290467050543826,
      "grad_norm": 1.8782513235040483,
      "learning_rate": 1.7993118776421467e-05,
      "loss": 0.4133,
      "step": 3580
    },
    {
      "epoch": 2.291106845809341,
      "grad_norm": 1.919909546393126,
      "learning_rate": 1.7991873413509266e-05,
      "loss": 0.4814,
      "step": 3581
    },
    {
      "epoch": 2.291746641074856,
      "grad_norm": 1.8547245455098869,
      "learning_rate": 1.7990627707440245e-05,
      "loss": 0.3906,
      "step": 3582
    },
    {
      "epoch": 2.292386436340371,
      "grad_norm": 1.8043814635215913,
      "learning_rate": 1.7989381658267883e-05,
      "loss": 0.3828,
      "step": 3583
    },
    {
      "epoch": 2.2930262316058863,
      "grad_norm": 2.126489261362778,
      "learning_rate": 1.798813526604568e-05,
      "loss": 0.4174,
      "step": 3584
    },
    {
      "epoch": 2.293666026871401,
      "grad_norm": 1.9291996131902192,
      "learning_rate": 1.7986888530827162e-05,
      "loss": 0.4663,
      "step": 3585
    },
    {
      "epoch": 2.2943058221369164,
      "grad_norm": 1.9201990613312305,
      "learning_rate": 1.7985641452665857e-05,
      "loss": 0.4558,
      "step": 3586
    },
    {
      "epoch": 2.294945617402431,
      "grad_norm": 1.9167099147698776,
      "learning_rate": 1.7984394031615315e-05,
      "loss": 0.4197,
      "step": 3587
    },
    {
      "epoch": 2.295585412667946,
      "grad_norm": 1.9096633009494342,
      "learning_rate": 1.7983146267729093e-05,
      "loss": 0.4129,
      "step": 3588
    },
    {
      "epoch": 2.2962252079334613,
      "grad_norm": 2.013013002702702,
      "learning_rate": 1.7981898161060776e-05,
      "loss": 0.4309,
      "step": 3589
    },
    {
      "epoch": 2.2968650031989766,
      "grad_norm": 2.105907464293832,
      "learning_rate": 1.7980649711663946e-05,
      "loss": 0.4234,
      "step": 3590
    },
    {
      "epoch": 2.2975047984644914,
      "grad_norm": 2.040857465098857,
      "learning_rate": 1.7979400919592218e-05,
      "loss": 0.4604,
      "step": 3591
    },
    {
      "epoch": 2.2981445937300062,
      "grad_norm": 1.8606098706217606,
      "learning_rate": 1.7978151784899207e-05,
      "loss": 0.3889,
      "step": 3592
    },
    {
      "epoch": 2.2987843889955215,
      "grad_norm": 2.1197586409514586,
      "learning_rate": 1.797690230763855e-05,
      "loss": 0.4714,
      "step": 3593
    },
    {
      "epoch": 2.2994241842610363,
      "grad_norm": 1.887867589302588,
      "learning_rate": 1.7975652487863896e-05,
      "loss": 0.4001,
      "step": 3594
    },
    {
      "epoch": 2.3000639795265516,
      "grad_norm": 1.6015780287194077,
      "learning_rate": 1.7974402325628915e-05,
      "loss": 0.4,
      "step": 3595
    },
    {
      "epoch": 2.3007037747920664,
      "grad_norm": 1.90335491471748,
      "learning_rate": 1.7973151820987282e-05,
      "loss": 0.4079,
      "step": 3596
    },
    {
      "epoch": 2.3013435700575817,
      "grad_norm": 1.8046561467938222,
      "learning_rate": 1.7971900973992693e-05,
      "loss": 0.397,
      "step": 3597
    },
    {
      "epoch": 2.3019833653230966,
      "grad_norm": 1.6884067572782628,
      "learning_rate": 1.797064978469886e-05,
      "loss": 0.3723,
      "step": 3598
    },
    {
      "epoch": 2.302623160588612,
      "grad_norm": 1.8929143726775466,
      "learning_rate": 1.79693982531595e-05,
      "loss": 0.4085,
      "step": 3599
    },
    {
      "epoch": 2.3032629558541267,
      "grad_norm": 1.956768676878961,
      "learning_rate": 1.7968146379428355e-05,
      "loss": 0.3841,
      "step": 3600
    },
    {
      "epoch": 2.303902751119642,
      "grad_norm": 1.9602501285219878,
      "learning_rate": 1.796689416355918e-05,
      "loss": 0.412,
      "step": 3601
    },
    {
      "epoch": 2.3045425463851568,
      "grad_norm": 1.9170614707784575,
      "learning_rate": 1.7965641605605743e-05,
      "loss": 0.4027,
      "step": 3602
    },
    {
      "epoch": 2.3051823416506716,
      "grad_norm": 1.9554085779519939,
      "learning_rate": 1.7964388705621824e-05,
      "loss": 0.4175,
      "step": 3603
    },
    {
      "epoch": 2.305822136916187,
      "grad_norm": 2.0279395068276274,
      "learning_rate": 1.7963135463661224e-05,
      "loss": 0.4021,
      "step": 3604
    },
    {
      "epoch": 2.3064619321817017,
      "grad_norm": 1.872558629852705,
      "learning_rate": 1.7961881879777746e-05,
      "loss": 0.4177,
      "step": 3605
    },
    {
      "epoch": 2.307101727447217,
      "grad_norm": 1.883361753524025,
      "learning_rate": 1.796062795402523e-05,
      "loss": 0.4303,
      "step": 3606
    },
    {
      "epoch": 2.307741522712732,
      "grad_norm": 2.0316150137820395,
      "learning_rate": 1.795937368645751e-05,
      "loss": 0.4786,
      "step": 3607
    },
    {
      "epoch": 2.308381317978247,
      "grad_norm": 1.994132866141619,
      "learning_rate": 1.7958119077128437e-05,
      "loss": 0.4119,
      "step": 3608
    },
    {
      "epoch": 2.309021113243762,
      "grad_norm": 1.7409277123316,
      "learning_rate": 1.7956864126091892e-05,
      "loss": 0.3888,
      "step": 3609
    },
    {
      "epoch": 2.309660908509277,
      "grad_norm": 1.9013723606492876,
      "learning_rate": 1.7955608833401754e-05,
      "loss": 0.46,
      "step": 3610
    },
    {
      "epoch": 2.310300703774792,
      "grad_norm": 1.9851196753982692,
      "learning_rate": 1.7954353199111923e-05,
      "loss": 0.4235,
      "step": 3611
    },
    {
      "epoch": 2.3109404990403073,
      "grad_norm": 1.7924061429278708,
      "learning_rate": 1.7953097223276317e-05,
      "loss": 0.4258,
      "step": 3612
    },
    {
      "epoch": 2.311580294305822,
      "grad_norm": 1.825240757244637,
      "learning_rate": 1.7951840905948865e-05,
      "loss": 0.4119,
      "step": 3613
    },
    {
      "epoch": 2.312220089571337,
      "grad_norm": 1.69331490006604,
      "learning_rate": 1.795058424718351e-05,
      "loss": 0.3575,
      "step": 3614
    },
    {
      "epoch": 2.3128598848368522,
      "grad_norm": 1.8085184834957406,
      "learning_rate": 1.7949327247034206e-05,
      "loss": 0.4293,
      "step": 3615
    },
    {
      "epoch": 2.313499680102367,
      "grad_norm": 1.9613087981262507,
      "learning_rate": 1.794806990555494e-05,
      "loss": 0.4193,
      "step": 3616
    },
    {
      "epoch": 2.3141394753678823,
      "grad_norm": 1.8557079090155588,
      "learning_rate": 1.7946812222799682e-05,
      "loss": 0.3859,
      "step": 3617
    },
    {
      "epoch": 2.314779270633397,
      "grad_norm": 1.9506576323480733,
      "learning_rate": 1.7945554198822445e-05,
      "loss": 0.4333,
      "step": 3618
    },
    {
      "epoch": 2.3154190658989124,
      "grad_norm": 1.934771938952554,
      "learning_rate": 1.7944295833677247e-05,
      "loss": 0.4044,
      "step": 3619
    },
    {
      "epoch": 2.3160588611644273,
      "grad_norm": 1.7944542799892018,
      "learning_rate": 1.7943037127418115e-05,
      "loss": 0.4163,
      "step": 3620
    },
    {
      "epoch": 2.3166986564299425,
      "grad_norm": 1.8051009348832094,
      "learning_rate": 1.79417780800991e-05,
      "loss": 0.4317,
      "step": 3621
    },
    {
      "epoch": 2.3173384516954574,
      "grad_norm": 1.7733915548452297,
      "learning_rate": 1.794051869177426e-05,
      "loss": 0.4146,
      "step": 3622
    },
    {
      "epoch": 2.3179782469609727,
      "grad_norm": 1.9062597192362285,
      "learning_rate": 1.7939258962497675e-05,
      "loss": 0.4154,
      "step": 3623
    },
    {
      "epoch": 2.3186180422264875,
      "grad_norm": 1.8124883996383756,
      "learning_rate": 1.793799889232343e-05,
      "loss": 0.3536,
      "step": 3624
    },
    {
      "epoch": 2.3192578374920023,
      "grad_norm": 1.9644457071231183,
      "learning_rate": 1.7936738481305636e-05,
      "loss": 0.4343,
      "step": 3625
    },
    {
      "epoch": 2.3198976327575176,
      "grad_norm": 1.8649886363106658,
      "learning_rate": 1.793547772949841e-05,
      "loss": 0.3836,
      "step": 3626
    },
    {
      "epoch": 2.320537428023033,
      "grad_norm": 1.973751483745757,
      "learning_rate": 1.7934216636955883e-05,
      "loss": 0.4033,
      "step": 3627
    },
    {
      "epoch": 2.3211772232885477,
      "grad_norm": 1.8465989460030434,
      "learning_rate": 1.7932955203732213e-05,
      "loss": 0.3973,
      "step": 3628
    },
    {
      "epoch": 2.3218170185540625,
      "grad_norm": 2.1439587136562372,
      "learning_rate": 1.793169342988155e-05,
      "loss": 0.4402,
      "step": 3629
    },
    {
      "epoch": 2.322456813819578,
      "grad_norm": 1.856535060081622,
      "learning_rate": 1.793043131545809e-05,
      "loss": 0.4268,
      "step": 3630
    },
    {
      "epoch": 2.3230966090850926,
      "grad_norm": 1.7466599159610632,
      "learning_rate": 1.792916886051601e-05,
      "loss": 0.4249,
      "step": 3631
    },
    {
      "epoch": 2.323736404350608,
      "grad_norm": 1.8142177708681237,
      "learning_rate": 1.7927906065109527e-05,
      "loss": 0.3707,
      "step": 3632
    },
    {
      "epoch": 2.3243761996161227,
      "grad_norm": 1.7364816115033572,
      "learning_rate": 1.792664292929286e-05,
      "loss": 0.3985,
      "step": 3633
    },
    {
      "epoch": 2.325015994881638,
      "grad_norm": 1.9102871629623392,
      "learning_rate": 1.7925379453120243e-05,
      "loss": 0.4447,
      "step": 3634
    },
    {
      "epoch": 2.325655790147153,
      "grad_norm": 2.0318901235938682,
      "learning_rate": 1.7924115636645933e-05,
      "loss": 0.4068,
      "step": 3635
    },
    {
      "epoch": 2.326295585412668,
      "grad_norm": 2.0406644658503357,
      "learning_rate": 1.7922851479924196e-05,
      "loss": 0.4304,
      "step": 3636
    },
    {
      "epoch": 2.326935380678183,
      "grad_norm": 1.837777533691941,
      "learning_rate": 1.7921586983009304e-05,
      "loss": 0.4303,
      "step": 3637
    },
    {
      "epoch": 2.3275751759436982,
      "grad_norm": 1.8851174201540624,
      "learning_rate": 1.7920322145955563e-05,
      "loss": 0.3956,
      "step": 3638
    },
    {
      "epoch": 2.328214971209213,
      "grad_norm": 1.9779942080226958,
      "learning_rate": 1.7919056968817277e-05,
      "loss": 0.4149,
      "step": 3639
    },
    {
      "epoch": 2.328854766474728,
      "grad_norm": 1.7547960927851523,
      "learning_rate": 1.791779145164877e-05,
      "loss": 0.3752,
      "step": 3640
    },
    {
      "epoch": 2.329494561740243,
      "grad_norm": 1.9453800779891366,
      "learning_rate": 1.7916525594504388e-05,
      "loss": 0.4254,
      "step": 3641
    },
    {
      "epoch": 2.330134357005758,
      "grad_norm": 1.8505358244479195,
      "learning_rate": 1.7915259397438473e-05,
      "loss": 0.431,
      "step": 3642
    },
    {
      "epoch": 2.3307741522712733,
      "grad_norm": 1.9422009544654333,
      "learning_rate": 1.7913992860505402e-05,
      "loss": 0.4314,
      "step": 3643
    },
    {
      "epoch": 2.331413947536788,
      "grad_norm": 1.8772143277768905,
      "learning_rate": 1.7912725983759556e-05,
      "loss": 0.4632,
      "step": 3644
    },
    {
      "epoch": 2.3320537428023034,
      "grad_norm": 1.9633252079800239,
      "learning_rate": 1.7911458767255333e-05,
      "loss": 0.4436,
      "step": 3645
    },
    {
      "epoch": 2.332693538067818,
      "grad_norm": 1.7989882675077356,
      "learning_rate": 1.791019121104714e-05,
      "loss": 0.414,
      "step": 3646
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 1.8908432226639758,
      "learning_rate": 1.7908923315189413e-05,
      "loss": 0.4033,
      "step": 3647
    },
    {
      "epoch": 2.3339731285988483,
      "grad_norm": 1.9025298206748755,
      "learning_rate": 1.7907655079736583e-05,
      "loss": 0.4305,
      "step": 3648
    },
    {
      "epoch": 2.3346129238643636,
      "grad_norm": 1.9224330861670853,
      "learning_rate": 1.7906386504743115e-05,
      "loss": 0.4295,
      "step": 3649
    },
    {
      "epoch": 2.3352527191298784,
      "grad_norm": 1.8169368179976386,
      "learning_rate": 1.790511759026347e-05,
      "loss": 0.4185,
      "step": 3650
    },
    {
      "epoch": 2.3358925143953932,
      "grad_norm": 1.9849478000420824,
      "learning_rate": 1.790384833635214e-05,
      "loss": 0.4508,
      "step": 3651
    },
    {
      "epoch": 2.3365323096609085,
      "grad_norm": 1.9019582173475926,
      "learning_rate": 1.790257874306362e-05,
      "loss": 0.4234,
      "step": 3652
    },
    {
      "epoch": 2.3371721049264234,
      "grad_norm": 2.0291510340303947,
      "learning_rate": 1.7901308810452433e-05,
      "loss": 0.3793,
      "step": 3653
    },
    {
      "epoch": 2.3378119001919386,
      "grad_norm": 1.845838986574629,
      "learning_rate": 1.7900038538573094e-05,
      "loss": 0.4068,
      "step": 3654
    },
    {
      "epoch": 2.3384516954574535,
      "grad_norm": 1.959823343912062,
      "learning_rate": 1.7898767927480157e-05,
      "loss": 0.4053,
      "step": 3655
    },
    {
      "epoch": 2.3390914907229687,
      "grad_norm": 2.0021146064002084,
      "learning_rate": 1.7897496977228174e-05,
      "loss": 0.4412,
      "step": 3656
    },
    {
      "epoch": 2.3397312859884836,
      "grad_norm": 1.8961538715604906,
      "learning_rate": 1.789622568787172e-05,
      "loss": 0.3679,
      "step": 3657
    },
    {
      "epoch": 2.340371081253999,
      "grad_norm": 1.990120538777861,
      "learning_rate": 1.789495405946538e-05,
      "loss": 0.4474,
      "step": 3658
    },
    {
      "epoch": 2.3410108765195137,
      "grad_norm": 2.076664293994804,
      "learning_rate": 1.789368209206376e-05,
      "loss": 0.4756,
      "step": 3659
    },
    {
      "epoch": 2.341650671785029,
      "grad_norm": 1.8837180696310678,
      "learning_rate": 1.789240978572147e-05,
      "loss": 0.4404,
      "step": 3660
    },
    {
      "epoch": 2.342290467050544,
      "grad_norm": 1.806444841753323,
      "learning_rate": 1.7891137140493144e-05,
      "loss": 0.3442,
      "step": 3661
    },
    {
      "epoch": 2.342930262316059,
      "grad_norm": 1.6490746157503022,
      "learning_rate": 1.7889864156433427e-05,
      "loss": 0.4014,
      "step": 3662
    },
    {
      "epoch": 2.343570057581574,
      "grad_norm": 1.841640560330009,
      "learning_rate": 1.7888590833596978e-05,
      "loss": 0.4397,
      "step": 3663
    },
    {
      "epoch": 2.344209852847089,
      "grad_norm": 1.8505566800436988,
      "learning_rate": 1.788731717203847e-05,
      "loss": 0.3885,
      "step": 3664
    },
    {
      "epoch": 2.344849648112604,
      "grad_norm": 1.947235031383629,
      "learning_rate": 1.7886043171812595e-05,
      "loss": 0.3994,
      "step": 3665
    },
    {
      "epoch": 2.345489443378119,
      "grad_norm": 1.8794042122976455,
      "learning_rate": 1.7884768832974055e-05,
      "loss": 0.3617,
      "step": 3666
    },
    {
      "epoch": 2.346129238643634,
      "grad_norm": 1.7865348779255033,
      "learning_rate": 1.7883494155577566e-05,
      "loss": 0.3796,
      "step": 3667
    },
    {
      "epoch": 2.346769033909149,
      "grad_norm": 1.8246674588087448,
      "learning_rate": 1.7882219139677862e-05,
      "loss": 0.43,
      "step": 3668
    },
    {
      "epoch": 2.347408829174664,
      "grad_norm": 2.0105918603119584,
      "learning_rate": 1.7880943785329688e-05,
      "loss": 0.4366,
      "step": 3669
    },
    {
      "epoch": 2.348048624440179,
      "grad_norm": 1.9527591458875315,
      "learning_rate": 1.7879668092587806e-05,
      "loss": 0.444,
      "step": 3670
    },
    {
      "epoch": 2.3486884197056943,
      "grad_norm": 1.8620080678352438,
      "learning_rate": 1.7878392061507e-05,
      "loss": 0.4323,
      "step": 3671
    },
    {
      "epoch": 2.349328214971209,
      "grad_norm": 1.8381742183838248,
      "learning_rate": 1.7877115692142046e-05,
      "loss": 0.4057,
      "step": 3672
    },
    {
      "epoch": 2.3499680102367244,
      "grad_norm": 1.7643864178662674,
      "learning_rate": 1.7875838984547757e-05,
      "loss": 0.4042,
      "step": 3673
    },
    {
      "epoch": 2.3506078055022392,
      "grad_norm": 1.8718926319391302,
      "learning_rate": 1.7874561938778955e-05,
      "loss": 0.45,
      "step": 3674
    },
    {
      "epoch": 2.3512476007677545,
      "grad_norm": 1.9383938931316491,
      "learning_rate": 1.787328455489047e-05,
      "loss": 0.4326,
      "step": 3675
    },
    {
      "epoch": 2.3518873960332694,
      "grad_norm": 1.9769193379787093,
      "learning_rate": 1.787200683293715e-05,
      "loss": 0.4335,
      "step": 3676
    },
    {
      "epoch": 2.352527191298784,
      "grad_norm": 1.9755495261263392,
      "learning_rate": 1.7870728772973863e-05,
      "loss": 0.4598,
      "step": 3677
    },
    {
      "epoch": 2.3531669865642995,
      "grad_norm": 1.8118409624274427,
      "learning_rate": 1.786945037505548e-05,
      "loss": 0.4168,
      "step": 3678
    },
    {
      "epoch": 2.3538067818298143,
      "grad_norm": 2.0103816563465546,
      "learning_rate": 1.78681716392369e-05,
      "loss": 0.4003,
      "step": 3679
    },
    {
      "epoch": 2.3544465770953296,
      "grad_norm": 1.8116177853479323,
      "learning_rate": 1.7866892565573025e-05,
      "loss": 0.3961,
      "step": 3680
    },
    {
      "epoch": 2.3550863723608444,
      "grad_norm": 1.9353617005456094,
      "learning_rate": 1.7865613154118776e-05,
      "loss": 0.4335,
      "step": 3681
    },
    {
      "epoch": 2.3557261676263597,
      "grad_norm": 1.9039513352945199,
      "learning_rate": 1.7864333404929092e-05,
      "loss": 0.43,
      "step": 3682
    },
    {
      "epoch": 2.3563659628918745,
      "grad_norm": 1.8504980434560707,
      "learning_rate": 1.786305331805892e-05,
      "loss": 0.4226,
      "step": 3683
    },
    {
      "epoch": 2.3570057581573898,
      "grad_norm": 1.9296054046200874,
      "learning_rate": 1.7861772893563224e-05,
      "loss": 0.4629,
      "step": 3684
    },
    {
      "epoch": 2.3576455534229046,
      "grad_norm": 1.9410713714130243,
      "learning_rate": 1.7860492131496987e-05,
      "loss": 0.4205,
      "step": 3685
    },
    {
      "epoch": 2.35828534868842,
      "grad_norm": 1.806224556748724,
      "learning_rate": 1.7859211031915202e-05,
      "loss": 0.4019,
      "step": 3686
    },
    {
      "epoch": 2.3589251439539347,
      "grad_norm": 1.7478821730356422,
      "learning_rate": 1.7857929594872875e-05,
      "loss": 0.3803,
      "step": 3687
    },
    {
      "epoch": 2.3595649392194495,
      "grad_norm": 1.9483592466487292,
      "learning_rate": 1.785664782042503e-05,
      "loss": 0.4339,
      "step": 3688
    },
    {
      "epoch": 2.360204734484965,
      "grad_norm": 1.9799447743340042,
      "learning_rate": 1.78553657086267e-05,
      "loss": 0.4676,
      "step": 3689
    },
    {
      "epoch": 2.36084452975048,
      "grad_norm": 1.8293840676959048,
      "learning_rate": 1.785408325953294e-05,
      "loss": 0.3799,
      "step": 3690
    },
    {
      "epoch": 2.361484325015995,
      "grad_norm": 1.8278838683446186,
      "learning_rate": 1.785280047319882e-05,
      "loss": 0.4088,
      "step": 3691
    },
    {
      "epoch": 2.3621241202815098,
      "grad_norm": 1.8068601073728898,
      "learning_rate": 1.7851517349679416e-05,
      "loss": 0.4081,
      "step": 3692
    },
    {
      "epoch": 2.362763915547025,
      "grad_norm": 1.8334796345541455,
      "learning_rate": 1.785023388902982e-05,
      "loss": 0.4059,
      "step": 3693
    },
    {
      "epoch": 2.36340371081254,
      "grad_norm": 1.7698253224931055,
      "learning_rate": 1.7848950091305152e-05,
      "loss": 0.373,
      "step": 3694
    },
    {
      "epoch": 2.364043506078055,
      "grad_norm": 1.8720243441938866,
      "learning_rate": 1.7847665956560524e-05,
      "loss": 0.3782,
      "step": 3695
    },
    {
      "epoch": 2.36468330134357,
      "grad_norm": 2.007383265340488,
      "learning_rate": 1.7846381484851082e-05,
      "loss": 0.4155,
      "step": 3696
    },
    {
      "epoch": 2.3653230966090852,
      "grad_norm": 1.8450111185130236,
      "learning_rate": 1.7845096676231974e-05,
      "loss": 0.382,
      "step": 3697
    },
    {
      "epoch": 2.3659628918746,
      "grad_norm": 1.8006794178008116,
      "learning_rate": 1.7843811530758373e-05,
      "loss": 0.383,
      "step": 3698
    },
    {
      "epoch": 2.3666026871401153,
      "grad_norm": 1.8440581071420201,
      "learning_rate": 1.7842526048485455e-05,
      "loss": 0.4513,
      "step": 3699
    },
    {
      "epoch": 2.36724248240563,
      "grad_norm": 2.0140423784419643,
      "learning_rate": 1.7841240229468422e-05,
      "loss": 0.4297,
      "step": 3700
    },
    {
      "epoch": 2.3678822776711455,
      "grad_norm": 1.8736967543650134,
      "learning_rate": 1.783995407376248e-05,
      "loss": 0.4242,
      "step": 3701
    },
    {
      "epoch": 2.3685220729366603,
      "grad_norm": 1.9988589783508828,
      "learning_rate": 1.783866758142286e-05,
      "loss": 0.3779,
      "step": 3702
    },
    {
      "epoch": 2.369161868202175,
      "grad_norm": 1.7974582345589285,
      "learning_rate": 1.7837380752504794e-05,
      "loss": 0.4054,
      "step": 3703
    },
    {
      "epoch": 2.3698016634676904,
      "grad_norm": 1.7183593531622408,
      "learning_rate": 1.7836093587063544e-05,
      "loss": 0.406,
      "step": 3704
    },
    {
      "epoch": 2.370441458733205,
      "grad_norm": 1.8907382805189645,
      "learning_rate": 1.783480608515437e-05,
      "loss": 0.3963,
      "step": 3705
    },
    {
      "epoch": 2.3710812539987205,
      "grad_norm": 1.831376260036698,
      "learning_rate": 1.783351824683256e-05,
      "loss": 0.4108,
      "step": 3706
    },
    {
      "epoch": 2.3717210492642353,
      "grad_norm": 1.8426653195376144,
      "learning_rate": 1.7832230072153415e-05,
      "loss": 0.42,
      "step": 3707
    },
    {
      "epoch": 2.3723608445297506,
      "grad_norm": 2.0571092142408376,
      "learning_rate": 1.783094156117224e-05,
      "loss": 0.4152,
      "step": 3708
    },
    {
      "epoch": 2.3730006397952654,
      "grad_norm": 1.9705537224916632,
      "learning_rate": 1.7829652713944368e-05,
      "loss": 0.4321,
      "step": 3709
    },
    {
      "epoch": 2.3736404350607807,
      "grad_norm": 1.711104916382243,
      "learning_rate": 1.7828363530525133e-05,
      "loss": 0.3691,
      "step": 3710
    },
    {
      "epoch": 2.3742802303262955,
      "grad_norm": 1.8130769624631322,
      "learning_rate": 1.7827074010969893e-05,
      "loss": 0.45,
      "step": 3711
    },
    {
      "epoch": 2.374920025591811,
      "grad_norm": 1.8134636476118426,
      "learning_rate": 1.782578415533402e-05,
      "loss": 0.3969,
      "step": 3712
    },
    {
      "epoch": 2.3755598208573256,
      "grad_norm": 1.9677420612975456,
      "learning_rate": 1.78244939636729e-05,
      "loss": 0.4027,
      "step": 3713
    },
    {
      "epoch": 2.3761996161228405,
      "grad_norm": 1.8962462994803935,
      "learning_rate": 1.7823203436041923e-05,
      "loss": 0.3947,
      "step": 3714
    },
    {
      "epoch": 2.3768394113883557,
      "grad_norm": 1.8708780157252909,
      "learning_rate": 1.7821912572496507e-05,
      "loss": 0.4402,
      "step": 3715
    },
    {
      "epoch": 2.3774792066538706,
      "grad_norm": 1.9444848185042676,
      "learning_rate": 1.782062137309208e-05,
      "loss": 0.427,
      "step": 3716
    },
    {
      "epoch": 2.378119001919386,
      "grad_norm": 1.8281759317216957,
      "learning_rate": 1.7819329837884087e-05,
      "loss": 0.4118,
      "step": 3717
    },
    {
      "epoch": 2.3787587971849007,
      "grad_norm": 2.008477874607617,
      "learning_rate": 1.7818037966927973e-05,
      "loss": 0.4475,
      "step": 3718
    },
    {
      "epoch": 2.379398592450416,
      "grad_norm": 1.8661197963123912,
      "learning_rate": 1.781674576027922e-05,
      "loss": 0.4619,
      "step": 3719
    },
    {
      "epoch": 2.380038387715931,
      "grad_norm": 2.0301578074961593,
      "learning_rate": 1.7815453217993308e-05,
      "loss": 0.4628,
      "step": 3720
    },
    {
      "epoch": 2.380678182981446,
      "grad_norm": 1.8656928962628954,
      "learning_rate": 1.7814160340125738e-05,
      "loss": 0.4388,
      "step": 3721
    },
    {
      "epoch": 2.381317978246961,
      "grad_norm": 1.7391446948759204,
      "learning_rate": 1.7812867126732026e-05,
      "loss": 0.429,
      "step": 3722
    },
    {
      "epoch": 2.381957773512476,
      "grad_norm": 1.7436336607680518,
      "learning_rate": 1.7811573577867693e-05,
      "loss": 0.4161,
      "step": 3723
    },
    {
      "epoch": 2.382597568777991,
      "grad_norm": 1.8831795622353453,
      "learning_rate": 1.7810279693588288e-05,
      "loss": 0.4118,
      "step": 3724
    },
    {
      "epoch": 2.383237364043506,
      "grad_norm": 2.054833004629259,
      "learning_rate": 1.7808985473949368e-05,
      "loss": 0.4034,
      "step": 3725
    },
    {
      "epoch": 2.383877159309021,
      "grad_norm": 1.8857518500773067,
      "learning_rate": 1.78076909190065e-05,
      "loss": 0.4073,
      "step": 3726
    },
    {
      "epoch": 2.3845169545745364,
      "grad_norm": 1.8373523228704058,
      "learning_rate": 1.7806396028815273e-05,
      "loss": 0.406,
      "step": 3727
    },
    {
      "epoch": 2.385156749840051,
      "grad_norm": 1.9948367847124695,
      "learning_rate": 1.780510080343129e-05,
      "loss": 0.404,
      "step": 3728
    },
    {
      "epoch": 2.385796545105566,
      "grad_norm": 1.7927438702819638,
      "learning_rate": 1.780380524291016e-05,
      "loss": 0.3781,
      "step": 3729
    },
    {
      "epoch": 2.3864363403710813,
      "grad_norm": 2.0044469609309754,
      "learning_rate": 1.7802509347307516e-05,
      "loss": 0.4062,
      "step": 3730
    },
    {
      "epoch": 2.387076135636596,
      "grad_norm": 2.041914454795127,
      "learning_rate": 1.7801213116679e-05,
      "loss": 0.4072,
      "step": 3731
    },
    {
      "epoch": 2.3877159309021114,
      "grad_norm": 1.9314415036855805,
      "learning_rate": 1.7799916551080268e-05,
      "loss": 0.4123,
      "step": 3732
    },
    {
      "epoch": 2.3883557261676263,
      "grad_norm": 1.693088911911756,
      "learning_rate": 1.7798619650567e-05,
      "loss": 0.3856,
      "step": 3733
    },
    {
      "epoch": 2.3889955214331415,
      "grad_norm": 1.9281649472331208,
      "learning_rate": 1.779732241519487e-05,
      "loss": 0.4104,
      "step": 3734
    },
    {
      "epoch": 2.3896353166986564,
      "grad_norm": 1.8451782836204185,
      "learning_rate": 1.779602484501959e-05,
      "loss": 0.362,
      "step": 3735
    },
    {
      "epoch": 2.3902751119641716,
      "grad_norm": 1.706953499606646,
      "learning_rate": 1.779472694009687e-05,
      "loss": 0.3652,
      "step": 3736
    },
    {
      "epoch": 2.3909149072296865,
      "grad_norm": 1.8113470119202228,
      "learning_rate": 1.779342870048244e-05,
      "loss": 0.3384,
      "step": 3737
    },
    {
      "epoch": 2.3915547024952017,
      "grad_norm": 1.8994498204808627,
      "learning_rate": 1.7792130126232047e-05,
      "loss": 0.3695,
      "step": 3738
    },
    {
      "epoch": 2.3921944977607166,
      "grad_norm": 1.859612382923212,
      "learning_rate": 1.779083121740145e-05,
      "loss": 0.3609,
      "step": 3739
    },
    {
      "epoch": 2.3928342930262314,
      "grad_norm": 1.9117063036739042,
      "learning_rate": 1.7789531974046413e-05,
      "loss": 0.4495,
      "step": 3740
    },
    {
      "epoch": 2.3934740882917467,
      "grad_norm": 1.7781480559831462,
      "learning_rate": 1.7788232396222732e-05,
      "loss": 0.3922,
      "step": 3741
    },
    {
      "epoch": 2.3941138835572615,
      "grad_norm": 1.9321223448656057,
      "learning_rate": 1.7786932483986205e-05,
      "loss": 0.4094,
      "step": 3742
    },
    {
      "epoch": 2.394753678822777,
      "grad_norm": 1.8259657517308578,
      "learning_rate": 1.778563223739265e-05,
      "loss": 0.4182,
      "step": 3743
    },
    {
      "epoch": 2.3953934740882916,
      "grad_norm": 1.790324552495318,
      "learning_rate": 1.7784331656497897e-05,
      "loss": 0.3986,
      "step": 3744
    },
    {
      "epoch": 2.396033269353807,
      "grad_norm": 1.9184203104020345,
      "learning_rate": 1.7783030741357792e-05,
      "loss": 0.4087,
      "step": 3745
    },
    {
      "epoch": 2.3966730646193217,
      "grad_norm": 1.9665303551119933,
      "learning_rate": 1.7781729492028187e-05,
      "loss": 0.4331,
      "step": 3746
    },
    {
      "epoch": 2.397312859884837,
      "grad_norm": 2.0289223492397626,
      "learning_rate": 1.7780427908564965e-05,
      "loss": 0.4481,
      "step": 3747
    },
    {
      "epoch": 2.397952655150352,
      "grad_norm": 1.7826701147903716,
      "learning_rate": 1.7779125991024007e-05,
      "loss": 0.3718,
      "step": 3748
    },
    {
      "epoch": 2.398592450415867,
      "grad_norm": 1.8418703408564427,
      "learning_rate": 1.7777823739461214e-05,
      "loss": 0.3942,
      "step": 3749
    },
    {
      "epoch": 2.399232245681382,
      "grad_norm": 1.885156978251458,
      "learning_rate": 1.777652115393251e-05,
      "loss": 0.4025,
      "step": 3750
    },
    {
      "epoch": 2.3998720409468968,
      "grad_norm": 1.907438912121078,
      "learning_rate": 1.777521823449382e-05,
      "loss": 0.4229,
      "step": 3751
    },
    {
      "epoch": 2.400511836212412,
      "grad_norm": 2.228347285323207,
      "learning_rate": 1.7773914981201087e-05,
      "loss": 0.4074,
      "step": 3752
    },
    {
      "epoch": 2.401151631477927,
      "grad_norm": 2.002649028238653,
      "learning_rate": 1.7772611394110275e-05,
      "loss": 0.4128,
      "step": 3753
    },
    {
      "epoch": 2.401791426743442,
      "grad_norm": 2.0427028907487403,
      "learning_rate": 1.7771307473277358e-05,
      "loss": 0.4229,
      "step": 3754
    },
    {
      "epoch": 2.402431222008957,
      "grad_norm": 1.7502513718698078,
      "learning_rate": 1.777000321875832e-05,
      "loss": 0.4035,
      "step": 3755
    },
    {
      "epoch": 2.4030710172744723,
      "grad_norm": 2.011228230300865,
      "learning_rate": 1.7768698630609168e-05,
      "loss": 0.3769,
      "step": 3756
    },
    {
      "epoch": 2.403710812539987,
      "grad_norm": 1.9912055545283192,
      "learning_rate": 1.7767393708885915e-05,
      "loss": 0.4049,
      "step": 3757
    },
    {
      "epoch": 2.4043506078055024,
      "grad_norm": 1.913553552501801,
      "learning_rate": 1.776608845364459e-05,
      "loss": 0.4355,
      "step": 3758
    },
    {
      "epoch": 2.404990403071017,
      "grad_norm": 1.7516219302197613,
      "learning_rate": 1.7764782864941247e-05,
      "loss": 0.3503,
      "step": 3759
    },
    {
      "epoch": 2.4056301983365325,
      "grad_norm": 1.8761034352139074,
      "learning_rate": 1.7763476942831938e-05,
      "loss": 0.4281,
      "step": 3760
    },
    {
      "epoch": 2.4062699936020473,
      "grad_norm": 1.6982835514758272,
      "learning_rate": 1.7762170687372738e-05,
      "loss": 0.3977,
      "step": 3761
    },
    {
      "epoch": 2.4069097888675626,
      "grad_norm": 1.8331941907887446,
      "learning_rate": 1.7760864098619737e-05,
      "loss": 0.4594,
      "step": 3762
    },
    {
      "epoch": 2.4075495841330774,
      "grad_norm": 1.8034565999451282,
      "learning_rate": 1.7759557176629037e-05,
      "loss": 0.3735,
      "step": 3763
    },
    {
      "epoch": 2.4081893793985927,
      "grad_norm": 1.8292646424015098,
      "learning_rate": 1.7758249921456755e-05,
      "loss": 0.3658,
      "step": 3764
    },
    {
      "epoch": 2.4088291746641075,
      "grad_norm": 1.871980185663682,
      "learning_rate": 1.7756942333159025e-05,
      "loss": 0.3909,
      "step": 3765
    },
    {
      "epoch": 2.4094689699296223,
      "grad_norm": 1.957365181178155,
      "learning_rate": 1.7755634411791985e-05,
      "loss": 0.3911,
      "step": 3766
    },
    {
      "epoch": 2.4101087651951376,
      "grad_norm": 1.8004851236955641,
      "learning_rate": 1.77543261574118e-05,
      "loss": 0.392,
      "step": 3767
    },
    {
      "epoch": 2.4107485604606524,
      "grad_norm": 1.825867289746648,
      "learning_rate": 1.775301757007465e-05,
      "loss": 0.4196,
      "step": 3768
    },
    {
      "epoch": 2.4113883557261677,
      "grad_norm": 1.994259445282941,
      "learning_rate": 1.7751708649836715e-05,
      "loss": 0.4224,
      "step": 3769
    },
    {
      "epoch": 2.4120281509916826,
      "grad_norm": 1.9134028267827097,
      "learning_rate": 1.7750399396754196e-05,
      "loss": 0.4073,
      "step": 3770
    },
    {
      "epoch": 2.412667946257198,
      "grad_norm": 2.132573623097441,
      "learning_rate": 1.7749089810883315e-05,
      "loss": 0.4242,
      "step": 3771
    },
    {
      "epoch": 2.4133077415227127,
      "grad_norm": 1.7473873441191599,
      "learning_rate": 1.77477798922803e-05,
      "loss": 0.3877,
      "step": 3772
    },
    {
      "epoch": 2.413947536788228,
      "grad_norm": 1.7296990805466343,
      "learning_rate": 1.7746469641001404e-05,
      "loss": 0.4135,
      "step": 3773
    },
    {
      "epoch": 2.4145873320537428,
      "grad_norm": 1.7403049180791632,
      "learning_rate": 1.7745159057102877e-05,
      "loss": 0.3815,
      "step": 3774
    },
    {
      "epoch": 2.415227127319258,
      "grad_norm": 1.712397177673351,
      "learning_rate": 1.7743848140641004e-05,
      "loss": 0.3663,
      "step": 3775
    },
    {
      "epoch": 2.415866922584773,
      "grad_norm": 1.9397175136664444,
      "learning_rate": 1.774253689167206e-05,
      "loss": 0.4051,
      "step": 3776
    },
    {
      "epoch": 2.4165067178502877,
      "grad_norm": 1.9924833028926918,
      "learning_rate": 1.7741225310252357e-05,
      "loss": 0.3963,
      "step": 3777
    },
    {
      "epoch": 2.417146513115803,
      "grad_norm": 1.82492893524662,
      "learning_rate": 1.773991339643821e-05,
      "loss": 0.4084,
      "step": 3778
    },
    {
      "epoch": 2.417786308381318,
      "grad_norm": 1.8544730261573648,
      "learning_rate": 1.773860115028595e-05,
      "loss": 0.3999,
      "step": 3779
    },
    {
      "epoch": 2.418426103646833,
      "grad_norm": 1.829024198624696,
      "learning_rate": 1.7737288571851923e-05,
      "loss": 0.4009,
      "step": 3780
    },
    {
      "epoch": 2.419065898912348,
      "grad_norm": 2.103283371481879,
      "learning_rate": 1.7735975661192485e-05,
      "loss": 0.4411,
      "step": 3781
    },
    {
      "epoch": 2.419705694177863,
      "grad_norm": 1.7735701007233255,
      "learning_rate": 1.773466241836402e-05,
      "loss": 0.3799,
      "step": 3782
    },
    {
      "epoch": 2.420345489443378,
      "grad_norm": 1.722027727012972,
      "learning_rate": 1.7733348843422903e-05,
      "loss": 0.3777,
      "step": 3783
    },
    {
      "epoch": 2.4209852847088933,
      "grad_norm": 1.7454219510062503,
      "learning_rate": 1.7732034936425547e-05,
      "loss": 0.4087,
      "step": 3784
    },
    {
      "epoch": 2.421625079974408,
      "grad_norm": 1.7991250120293383,
      "learning_rate": 1.7730720697428366e-05,
      "loss": 0.402,
      "step": 3785
    },
    {
      "epoch": 2.4222648752399234,
      "grad_norm": 1.8009203857109648,
      "learning_rate": 1.7729406126487786e-05,
      "loss": 0.4185,
      "step": 3786
    },
    {
      "epoch": 2.4229046705054382,
      "grad_norm": 1.7953829042980118,
      "learning_rate": 1.772809122366026e-05,
      "loss": 0.3779,
      "step": 3787
    },
    {
      "epoch": 2.423544465770953,
      "grad_norm": 1.8249149370917808,
      "learning_rate": 1.772677598900224e-05,
      "loss": 0.4035,
      "step": 3788
    },
    {
      "epoch": 2.4241842610364683,
      "grad_norm": 1.8887736959613641,
      "learning_rate": 1.7725460422570208e-05,
      "loss": 0.3812,
      "step": 3789
    },
    {
      "epoch": 2.4248240563019836,
      "grad_norm": 1.9383391751918202,
      "learning_rate": 1.7724144524420648e-05,
      "loss": 0.4006,
      "step": 3790
    },
    {
      "epoch": 2.4254638515674984,
      "grad_norm": 1.92287736130757,
      "learning_rate": 1.772282829461006e-05,
      "loss": 0.4314,
      "step": 3791
    },
    {
      "epoch": 2.4261036468330133,
      "grad_norm": 1.8826916547249601,
      "learning_rate": 1.7721511733194965e-05,
      "loss": 0.414,
      "step": 3792
    },
    {
      "epoch": 2.4267434420985285,
      "grad_norm": 1.9500319385827571,
      "learning_rate": 1.772019484023189e-05,
      "loss": 0.4153,
      "step": 3793
    },
    {
      "epoch": 2.4273832373640434,
      "grad_norm": 1.8485315643233178,
      "learning_rate": 1.7718877615777382e-05,
      "loss": 0.3955,
      "step": 3794
    },
    {
      "epoch": 2.4280230326295587,
      "grad_norm": 1.7431562691097005,
      "learning_rate": 1.7717560059888003e-05,
      "loss": 0.4083,
      "step": 3795
    },
    {
      "epoch": 2.4286628278950735,
      "grad_norm": 1.8828690363825125,
      "learning_rate": 1.771624217262032e-05,
      "loss": 0.4073,
      "step": 3796
    },
    {
      "epoch": 2.4293026231605888,
      "grad_norm": 1.849546481537548,
      "learning_rate": 1.7714923954030923e-05,
      "loss": 0.4227,
      "step": 3797
    },
    {
      "epoch": 2.4299424184261036,
      "grad_norm": 1.7942751645236539,
      "learning_rate": 1.7713605404176417e-05,
      "loss": 0.4186,
      "step": 3798
    },
    {
      "epoch": 2.430582213691619,
      "grad_norm": 1.930994274109048,
      "learning_rate": 1.7712286523113418e-05,
      "loss": 0.3792,
      "step": 3799
    },
    {
      "epoch": 2.4312220089571337,
      "grad_norm": 1.7100862943879709,
      "learning_rate": 1.771096731089855e-05,
      "loss": 0.3766,
      "step": 3800
    },
    {
      "epoch": 2.431861804222649,
      "grad_norm": 1.7484314671141565,
      "learning_rate": 1.770964776758847e-05,
      "loss": 0.4087,
      "step": 3801
    },
    {
      "epoch": 2.432501599488164,
      "grad_norm": 1.7410152527795493,
      "learning_rate": 1.7708327893239823e-05,
      "loss": 0.3843,
      "step": 3802
    },
    {
      "epoch": 2.4331413947536786,
      "grad_norm": 1.9150322732882137,
      "learning_rate": 1.770700768790929e-05,
      "loss": 0.3952,
      "step": 3803
    },
    {
      "epoch": 2.433781190019194,
      "grad_norm": 2.254899777567644,
      "learning_rate": 1.7705687151653558e-05,
      "loss": 0.4456,
      "step": 3804
    },
    {
      "epoch": 2.4344209852847087,
      "grad_norm": 1.89731884407171,
      "learning_rate": 1.7704366284529327e-05,
      "loss": 0.4292,
      "step": 3805
    },
    {
      "epoch": 2.435060780550224,
      "grad_norm": 1.946793011928812,
      "learning_rate": 1.7703045086593313e-05,
      "loss": 0.4394,
      "step": 3806
    },
    {
      "epoch": 2.435700575815739,
      "grad_norm": 1.9725399491220876,
      "learning_rate": 1.7701723557902242e-05,
      "loss": 0.4392,
      "step": 3807
    },
    {
      "epoch": 2.436340371081254,
      "grad_norm": 1.7843331045942759,
      "learning_rate": 1.7700401698512862e-05,
      "loss": 0.3747,
      "step": 3808
    },
    {
      "epoch": 2.436980166346769,
      "grad_norm": 1.6805671134381732,
      "learning_rate": 1.7699079508481936e-05,
      "loss": 0.3549,
      "step": 3809
    },
    {
      "epoch": 2.4376199616122842,
      "grad_norm": 1.922448181707184,
      "learning_rate": 1.7697756987866227e-05,
      "loss": 0.4095,
      "step": 3810
    },
    {
      "epoch": 2.438259756877799,
      "grad_norm": 1.840880104747533,
      "learning_rate": 1.7696434136722527e-05,
      "loss": 0.4159,
      "step": 3811
    },
    {
      "epoch": 2.4388995521433143,
      "grad_norm": 1.8111072229533738,
      "learning_rate": 1.769511095510764e-05,
      "loss": 0.3828,
      "step": 3812
    },
    {
      "epoch": 2.439539347408829,
      "grad_norm": 1.7836695791287729,
      "learning_rate": 1.7693787443078372e-05,
      "loss": 0.4037,
      "step": 3813
    },
    {
      "epoch": 2.440179142674344,
      "grad_norm": 1.7675019489344934,
      "learning_rate": 1.7692463600691557e-05,
      "loss": 0.3961,
      "step": 3814
    },
    {
      "epoch": 2.4408189379398593,
      "grad_norm": 1.8475455877304257,
      "learning_rate": 1.7691139428004045e-05,
      "loss": 0.4072,
      "step": 3815
    },
    {
      "epoch": 2.441458733205374,
      "grad_norm": 1.9090070201034124,
      "learning_rate": 1.768981492507268e-05,
      "loss": 0.423,
      "step": 3816
    },
    {
      "epoch": 2.4420985284708894,
      "grad_norm": 2.024861415138972,
      "learning_rate": 1.7688490091954345e-05,
      "loss": 0.4092,
      "step": 3817
    },
    {
      "epoch": 2.442738323736404,
      "grad_norm": 1.8996288644805694,
      "learning_rate": 1.7687164928705925e-05,
      "loss": 0.4147,
      "step": 3818
    },
    {
      "epoch": 2.4433781190019195,
      "grad_norm": 1.9830645082374334,
      "learning_rate": 1.7685839435384316e-05,
      "loss": 0.3813,
      "step": 3819
    },
    {
      "epoch": 2.4440179142674343,
      "grad_norm": 2.0116301139929287,
      "learning_rate": 1.768451361204643e-05,
      "loss": 0.4206,
      "step": 3820
    },
    {
      "epoch": 2.4446577095329496,
      "grad_norm": 1.9295553780405768,
      "learning_rate": 1.76831874587492e-05,
      "loss": 0.3785,
      "step": 3821
    },
    {
      "epoch": 2.4452975047984644,
      "grad_norm": 1.8093974359601186,
      "learning_rate": 1.768186097554957e-05,
      "loss": 0.4256,
      "step": 3822
    },
    {
      "epoch": 2.4459373000639797,
      "grad_norm": 1.8344893254976211,
      "learning_rate": 1.7680534162504495e-05,
      "loss": 0.4282,
      "step": 3823
    },
    {
      "epoch": 2.4465770953294945,
      "grad_norm": 1.8921528146036948,
      "learning_rate": 1.7679207019670945e-05,
      "loss": 0.3948,
      "step": 3824
    },
    {
      "epoch": 2.4472168905950094,
      "grad_norm": 1.7874379805271552,
      "learning_rate": 1.7677879547105904e-05,
      "loss": 0.3775,
      "step": 3825
    },
    {
      "epoch": 2.4478566858605246,
      "grad_norm": 1.77666161024984,
      "learning_rate": 1.7676551744866376e-05,
      "loss": 0.4248,
      "step": 3826
    },
    {
      "epoch": 2.44849648112604,
      "grad_norm": 1.9642551099766286,
      "learning_rate": 1.7675223613009368e-05,
      "loss": 0.4392,
      "step": 3827
    },
    {
      "epoch": 2.4491362763915547,
      "grad_norm": 1.8860531449640656,
      "learning_rate": 1.7673895151591914e-05,
      "loss": 0.3958,
      "step": 3828
    },
    {
      "epoch": 2.4497760716570696,
      "grad_norm": 1.9297619578567022,
      "learning_rate": 1.767256636067105e-05,
      "loss": 0.3791,
      "step": 3829
    },
    {
      "epoch": 2.450415866922585,
      "grad_norm": 1.9362142824787767,
      "learning_rate": 1.767123724030384e-05,
      "loss": 0.383,
      "step": 3830
    },
    {
      "epoch": 2.4510556621880997,
      "grad_norm": 1.9119917326886309,
      "learning_rate": 1.7669907790547346e-05,
      "loss": 0.4603,
      "step": 3831
    },
    {
      "epoch": 2.451695457453615,
      "grad_norm": 1.753177406323519,
      "learning_rate": 1.7668578011458657e-05,
      "loss": 0.3717,
      "step": 3832
    },
    {
      "epoch": 2.4523352527191298,
      "grad_norm": 1.954660681171379,
      "learning_rate": 1.7667247903094868e-05,
      "loss": 0.4668,
      "step": 3833
    },
    {
      "epoch": 2.452975047984645,
      "grad_norm": 1.7724363870817836,
      "learning_rate": 1.7665917465513092e-05,
      "loss": 0.3812,
      "step": 3834
    },
    {
      "epoch": 2.45361484325016,
      "grad_norm": 1.7533529961699015,
      "learning_rate": 1.7664586698770457e-05,
      "loss": 0.3704,
      "step": 3835
    },
    {
      "epoch": 2.454254638515675,
      "grad_norm": 1.94089256062134,
      "learning_rate": 1.7663255602924104e-05,
      "loss": 0.3746,
      "step": 3836
    },
    {
      "epoch": 2.45489443378119,
      "grad_norm": 1.8739747546143737,
      "learning_rate": 1.7661924178031187e-05,
      "loss": 0.4235,
      "step": 3837
    },
    {
      "epoch": 2.4555342290467053,
      "grad_norm": 1.9413446899350002,
      "learning_rate": 1.766059242414888e-05,
      "loss": 0.4329,
      "step": 3838
    },
    {
      "epoch": 2.45617402431222,
      "grad_norm": 1.7379214558602694,
      "learning_rate": 1.765926034133435e-05,
      "loss": 0.382,
      "step": 3839
    },
    {
      "epoch": 2.456813819577735,
      "grad_norm": 1.8287884140992505,
      "learning_rate": 1.7657927929644816e-05,
      "loss": 0.3648,
      "step": 3840
    },
    {
      "epoch": 2.45745361484325,
      "grad_norm": 1.9523222621788832,
      "learning_rate": 1.7656595189137476e-05,
      "loss": 0.3771,
      "step": 3841
    },
    {
      "epoch": 2.458093410108765,
      "grad_norm": 1.8133083476036922,
      "learning_rate": 1.7655262119869558e-05,
      "loss": 0.4272,
      "step": 3842
    },
    {
      "epoch": 2.4587332053742803,
      "grad_norm": 1.7112212499155106,
      "learning_rate": 1.76539287218983e-05,
      "loss": 0.3785,
      "step": 3843
    },
    {
      "epoch": 2.459373000639795,
      "grad_norm": 1.8690374519558148,
      "learning_rate": 1.7652594995280963e-05,
      "loss": 0.4265,
      "step": 3844
    },
    {
      "epoch": 2.4600127959053104,
      "grad_norm": 1.7689395667405556,
      "learning_rate": 1.7651260940074806e-05,
      "loss": 0.3872,
      "step": 3845
    },
    {
      "epoch": 2.4606525911708252,
      "grad_norm": 1.9080906535707807,
      "learning_rate": 1.7649926556337118e-05,
      "loss": 0.4354,
      "step": 3846
    },
    {
      "epoch": 2.4612923864363405,
      "grad_norm": 1.845108006516494,
      "learning_rate": 1.764859184412519e-05,
      "loss": 0.4119,
      "step": 3847
    },
    {
      "epoch": 2.4619321817018553,
      "grad_norm": 1.8824595156714317,
      "learning_rate": 1.7647256803496334e-05,
      "loss": 0.3694,
      "step": 3848
    },
    {
      "epoch": 2.4625719769673706,
      "grad_norm": 1.781445383987531,
      "learning_rate": 1.7645921434507872e-05,
      "loss": 0.4006,
      "step": 3849
    },
    {
      "epoch": 2.4632117722328855,
      "grad_norm": 1.8750165854702028,
      "learning_rate": 1.7644585737217148e-05,
      "loss": 0.4422,
      "step": 3850
    },
    {
      "epoch": 2.4638515674984003,
      "grad_norm": 1.798184613029389,
      "learning_rate": 1.7643249711681512e-05,
      "loss": 0.3505,
      "step": 3851
    },
    {
      "epoch": 2.4644913627639156,
      "grad_norm": 1.7764989139195988,
      "learning_rate": 1.7641913357958326e-05,
      "loss": 0.3314,
      "step": 3852
    },
    {
      "epoch": 2.4651311580294304,
      "grad_norm": 1.9538528246338858,
      "learning_rate": 1.7640576676104974e-05,
      "loss": 0.4338,
      "step": 3853
    },
    {
      "epoch": 2.4657709532949457,
      "grad_norm": 1.94450604514006,
      "learning_rate": 1.7639239666178855e-05,
      "loss": 0.3763,
      "step": 3854
    },
    {
      "epoch": 2.4664107485604605,
      "grad_norm": 1.825446041886476,
      "learning_rate": 1.7637902328237374e-05,
      "loss": 0.3947,
      "step": 3855
    },
    {
      "epoch": 2.4670505438259758,
      "grad_norm": 1.7934322149639164,
      "learning_rate": 1.763656466233795e-05,
      "loss": 0.404,
      "step": 3856
    },
    {
      "epoch": 2.4676903390914906,
      "grad_norm": 1.808780891500607,
      "learning_rate": 1.7635226668538024e-05,
      "loss": 0.3972,
      "step": 3857
    },
    {
      "epoch": 2.468330134357006,
      "grad_norm": 1.9156089098504818,
      "learning_rate": 1.763388834689505e-05,
      "loss": 0.414,
      "step": 3858
    },
    {
      "epoch": 2.4689699296225207,
      "grad_norm": 1.8818733717586469,
      "learning_rate": 1.7632549697466487e-05,
      "loss": 0.4263,
      "step": 3859
    },
    {
      "epoch": 2.469609724888036,
      "grad_norm": 2.030925520718161,
      "learning_rate": 1.763121072030982e-05,
      "loss": 0.3977,
      "step": 3860
    },
    {
      "epoch": 2.470249520153551,
      "grad_norm": 1.7829459756167527,
      "learning_rate": 1.762987141548254e-05,
      "loss": 0.4089,
      "step": 3861
    },
    {
      "epoch": 2.470889315419066,
      "grad_norm": 1.782645645842724,
      "learning_rate": 1.762853178304215e-05,
      "loss": 0.3955,
      "step": 3862
    },
    {
      "epoch": 2.471529110684581,
      "grad_norm": 1.947654784448355,
      "learning_rate": 1.762719182304618e-05,
      "loss": 0.3721,
      "step": 3863
    },
    {
      "epoch": 2.472168905950096,
      "grad_norm": 1.8292710869017197,
      "learning_rate": 1.7625851535552156e-05,
      "loss": 0.404,
      "step": 3864
    },
    {
      "epoch": 2.472808701215611,
      "grad_norm": 1.7828439657075839,
      "learning_rate": 1.7624510920617638e-05,
      "loss": 0.3367,
      "step": 3865
    },
    {
      "epoch": 2.473448496481126,
      "grad_norm": 1.7353817242005074,
      "learning_rate": 1.7623169978300177e-05,
      "loss": 0.3744,
      "step": 3866
    },
    {
      "epoch": 2.474088291746641,
      "grad_norm": 1.8978352433236554,
      "learning_rate": 1.7621828708657364e-05,
      "loss": 0.3993,
      "step": 3867
    },
    {
      "epoch": 2.474728087012156,
      "grad_norm": 1.9904202470216934,
      "learning_rate": 1.762048711174678e-05,
      "loss": 0.4441,
      "step": 3868
    },
    {
      "epoch": 2.4753678822776712,
      "grad_norm": 1.8859577423939988,
      "learning_rate": 1.7619145187626036e-05,
      "loss": 0.4444,
      "step": 3869
    },
    {
      "epoch": 2.476007677543186,
      "grad_norm": 1.743665652647882,
      "learning_rate": 1.7617802936352753e-05,
      "loss": 0.3707,
      "step": 3870
    },
    {
      "epoch": 2.4766474728087013,
      "grad_norm": 1.7371229326076243,
      "learning_rate": 1.761646035798456e-05,
      "loss": 0.3838,
      "step": 3871
    },
    {
      "epoch": 2.477287268074216,
      "grad_norm": 1.8653408214125349,
      "learning_rate": 1.7615117452579113e-05,
      "loss": 0.4188,
      "step": 3872
    },
    {
      "epoch": 2.4779270633397315,
      "grad_norm": 1.7218100362086715,
      "learning_rate": 1.7613774220194068e-05,
      "loss": 0.3733,
      "step": 3873
    },
    {
      "epoch": 2.4785668586052463,
      "grad_norm": 1.7185533105791093,
      "learning_rate": 1.7612430660887096e-05,
      "loss": 0.3722,
      "step": 3874
    },
    {
      "epoch": 2.4792066538707616,
      "grad_norm": 2.0093727891851247,
      "learning_rate": 1.7611086774715896e-05,
      "loss": 0.4373,
      "step": 3875
    },
    {
      "epoch": 2.4798464491362764,
      "grad_norm": 1.723277956271912,
      "learning_rate": 1.760974256173817e-05,
      "loss": 0.37,
      "step": 3876
    },
    {
      "epoch": 2.480486244401791,
      "grad_norm": 1.9295403510592004,
      "learning_rate": 1.7608398022011635e-05,
      "loss": 0.3826,
      "step": 3877
    },
    {
      "epoch": 2.4811260396673065,
      "grad_norm": 1.7428893288617064,
      "learning_rate": 1.7607053155594024e-05,
      "loss": 0.3608,
      "step": 3878
    },
    {
      "epoch": 2.4817658349328213,
      "grad_norm": 1.809352738769886,
      "learning_rate": 1.760570796254308e-05,
      "loss": 0.4395,
      "step": 3879
    },
    {
      "epoch": 2.4824056301983366,
      "grad_norm": 1.6848428883487925,
      "learning_rate": 1.7604362442916567e-05,
      "loss": 0.3802,
      "step": 3880
    },
    {
      "epoch": 2.4830454254638514,
      "grad_norm": 1.7659901421273532,
      "learning_rate": 1.7603016596772256e-05,
      "loss": 0.3953,
      "step": 3881
    },
    {
      "epoch": 2.4836852207293667,
      "grad_norm": 1.7525137453028081,
      "learning_rate": 1.7601670424167943e-05,
      "loss": 0.3739,
      "step": 3882
    },
    {
      "epoch": 2.4843250159948815,
      "grad_norm": 1.7849889108408634,
      "learning_rate": 1.7600323925161418e-05,
      "loss": 0.3491,
      "step": 3883
    },
    {
      "epoch": 2.484964811260397,
      "grad_norm": 1.8509329927050848,
      "learning_rate": 1.7598977099810506e-05,
      "loss": 0.4212,
      "step": 3884
    },
    {
      "epoch": 2.4856046065259116,
      "grad_norm": 1.8778777919962024,
      "learning_rate": 1.7597629948173033e-05,
      "loss": 0.3445,
      "step": 3885
    },
    {
      "epoch": 2.486244401791427,
      "grad_norm": 1.8575374927336399,
      "learning_rate": 1.759628247030685e-05,
      "loss": 0.409,
      "step": 3886
    },
    {
      "epoch": 2.4868841970569417,
      "grad_norm": 1.7343072860557212,
      "learning_rate": 1.7594934666269804e-05,
      "loss": 0.4315,
      "step": 3887
    },
    {
      "epoch": 2.4875239923224566,
      "grad_norm": 1.7860066593817594,
      "learning_rate": 1.759358653611978e-05,
      "loss": 0.4164,
      "step": 3888
    },
    {
      "epoch": 2.488163787587972,
      "grad_norm": 1.7444883712351238,
      "learning_rate": 1.7592238079914653e-05,
      "loss": 0.3891,
      "step": 3889
    },
    {
      "epoch": 2.488803582853487,
      "grad_norm": 1.7111622907849917,
      "learning_rate": 1.759088929771233e-05,
      "loss": 0.3866,
      "step": 3890
    },
    {
      "epoch": 2.489443378119002,
      "grad_norm": 2.1127394411662226,
      "learning_rate": 1.758954018957072e-05,
      "loss": 0.4437,
      "step": 3891
    },
    {
      "epoch": 2.490083173384517,
      "grad_norm": 1.7216953846564005,
      "learning_rate": 1.7588190755547763e-05,
      "loss": 0.3534,
      "step": 3892
    },
    {
      "epoch": 2.490722968650032,
      "grad_norm": 1.79052163398285,
      "learning_rate": 1.7586840995701387e-05,
      "loss": 0.4066,
      "step": 3893
    },
    {
      "epoch": 2.491362763915547,
      "grad_norm": 2.0305914850907723,
      "learning_rate": 1.758549091008956e-05,
      "loss": 0.4204,
      "step": 3894
    },
    {
      "epoch": 2.492002559181062,
      "grad_norm": 1.8713836240627577,
      "learning_rate": 1.758414049877024e-05,
      "loss": 0.4145,
      "step": 3895
    },
    {
      "epoch": 2.492642354446577,
      "grad_norm": 1.9202823019705344,
      "learning_rate": 1.7582789761801423e-05,
      "loss": 0.4087,
      "step": 3896
    },
    {
      "epoch": 2.4932821497120923,
      "grad_norm": 1.744729536660147,
      "learning_rate": 1.7581438699241102e-05,
      "loss": 0.4122,
      "step": 3897
    },
    {
      "epoch": 2.493921944977607,
      "grad_norm": 1.791537409590069,
      "learning_rate": 1.758008731114729e-05,
      "loss": 0.4512,
      "step": 3898
    },
    {
      "epoch": 2.4945617402431224,
      "grad_norm": 1.811481966540182,
      "learning_rate": 1.7578735597578008e-05,
      "loss": 0.3996,
      "step": 3899
    },
    {
      "epoch": 2.495201535508637,
      "grad_norm": 1.8892566229574805,
      "learning_rate": 1.7577383558591306e-05,
      "loss": 0.3931,
      "step": 3900
    },
    {
      "epoch": 2.4958413307741525,
      "grad_norm": 1.7411624571190607,
      "learning_rate": 1.7576031194245235e-05,
      "loss": 0.3871,
      "step": 3901
    },
    {
      "epoch": 2.4964811260396673,
      "grad_norm": 1.6884917939153268,
      "learning_rate": 1.7574678504597855e-05,
      "loss": 0.3553,
      "step": 3902
    },
    {
      "epoch": 2.497120921305182,
      "grad_norm": 1.7388336662472874,
      "learning_rate": 1.757332548970726e-05,
      "loss": 0.3281,
      "step": 3903
    },
    {
      "epoch": 2.4977607165706974,
      "grad_norm": 1.7127024143424308,
      "learning_rate": 1.7571972149631532e-05,
      "loss": 0.3725,
      "step": 3904
    },
    {
      "epoch": 2.4984005118362123,
      "grad_norm": 1.8089469660021689,
      "learning_rate": 1.7570618484428798e-05,
      "loss": 0.4004,
      "step": 3905
    },
    {
      "epoch": 2.4990403071017275,
      "grad_norm": 1.670944488250076,
      "learning_rate": 1.756926449415717e-05,
      "loss": 0.3551,
      "step": 3906
    },
    {
      "epoch": 2.4996801023672424,
      "grad_norm": 1.897708653564842,
      "learning_rate": 1.756791017887479e-05,
      "loss": 0.3863,
      "step": 3907
    },
    {
      "epoch": 2.5003198976327576,
      "grad_norm": 1.6506023146438347,
      "learning_rate": 1.756655553863981e-05,
      "loss": 0.3641,
      "step": 3908
    },
    {
      "epoch": 2.5009596928982725,
      "grad_norm": 2.0324936352098466,
      "learning_rate": 1.7565200573510392e-05,
      "loss": 0.4034,
      "step": 3909
    },
    {
      "epoch": 2.5015994881637877,
      "grad_norm": 1.8267718929019598,
      "learning_rate": 1.756384528354472e-05,
      "loss": 0.4073,
      "step": 3910
    },
    {
      "epoch": 2.5022392834293026,
      "grad_norm": 1.6957973615011064,
      "learning_rate": 1.756248966880099e-05,
      "loss": 0.3627,
      "step": 3911
    },
    {
      "epoch": 2.502879078694818,
      "grad_norm": 1.761879312104395,
      "learning_rate": 1.75611337293374e-05,
      "loss": 0.3908,
      "step": 3912
    },
    {
      "epoch": 2.5035188739603327,
      "grad_norm": 1.7525121857459565,
      "learning_rate": 1.7559777465212183e-05,
      "loss": 0.3529,
      "step": 3913
    },
    {
      "epoch": 2.5041586692258475,
      "grad_norm": 1.862657064699888,
      "learning_rate": 1.7558420876483567e-05,
      "loss": 0.4004,
      "step": 3914
    },
    {
      "epoch": 2.504798464491363,
      "grad_norm": 1.762936060744748,
      "learning_rate": 1.7557063963209805e-05,
      "loss": 0.351,
      "step": 3915
    },
    {
      "epoch": 2.505438259756878,
      "grad_norm": 1.8201077153131187,
      "learning_rate": 1.755570672544916e-05,
      "loss": 0.4319,
      "step": 3916
    },
    {
      "epoch": 2.506078055022393,
      "grad_norm": 1.7603039823461115,
      "learning_rate": 1.7554349163259908e-05,
      "loss": 0.4168,
      "step": 3917
    },
    {
      "epoch": 2.5067178502879077,
      "grad_norm": 1.7425279892716001,
      "learning_rate": 1.7552991276700343e-05,
      "loss": 0.3833,
      "step": 3918
    },
    {
      "epoch": 2.507357645553423,
      "grad_norm": 1.8789863941858578,
      "learning_rate": 1.7551633065828765e-05,
      "loss": 0.3836,
      "step": 3919
    },
    {
      "epoch": 2.507997440818938,
      "grad_norm": 1.6842038832146176,
      "learning_rate": 1.7550274530703495e-05,
      "loss": 0.4058,
      "step": 3920
    },
    {
      "epoch": 2.508637236084453,
      "grad_norm": 1.8009105998833643,
      "learning_rate": 1.7548915671382873e-05,
      "loss": 0.3804,
      "step": 3921
    },
    {
      "epoch": 2.509277031349968,
      "grad_norm": 1.71584340819593,
      "learning_rate": 1.7547556487925236e-05,
      "loss": 0.3702,
      "step": 3922
    },
    {
      "epoch": 2.509916826615483,
      "grad_norm": 1.868447914140366,
      "learning_rate": 1.754619698038895e-05,
      "loss": 0.4245,
      "step": 3923
    },
    {
      "epoch": 2.510556621880998,
      "grad_norm": 1.9026674917553805,
      "learning_rate": 1.7544837148832393e-05,
      "loss": 0.4437,
      "step": 3924
    },
    {
      "epoch": 2.511196417146513,
      "grad_norm": 1.8106273502567642,
      "learning_rate": 1.7543476993313948e-05,
      "loss": 0.3881,
      "step": 3925
    },
    {
      "epoch": 2.511836212412028,
      "grad_norm": 1.7415172001108028,
      "learning_rate": 1.754211651389202e-05,
      "loss": 0.3942,
      "step": 3926
    },
    {
      "epoch": 2.5124760076775434,
      "grad_norm": 1.726320617633459,
      "learning_rate": 1.7540755710625023e-05,
      "loss": 0.3678,
      "step": 3927
    },
    {
      "epoch": 2.5131158029430583,
      "grad_norm": 1.8424051250211066,
      "learning_rate": 1.7539394583571393e-05,
      "loss": 0.4207,
      "step": 3928
    },
    {
      "epoch": 2.513755598208573,
      "grad_norm": 1.7380095905317712,
      "learning_rate": 1.753803313278957e-05,
      "loss": 0.3998,
      "step": 3929
    },
    {
      "epoch": 2.5143953934740884,
      "grad_norm": 1.8346779271457825,
      "learning_rate": 1.7536671358338012e-05,
      "loss": 0.3991,
      "step": 3930
    },
    {
      "epoch": 2.515035188739603,
      "grad_norm": 1.7898126832348045,
      "learning_rate": 1.7535309260275192e-05,
      "loss": 0.3666,
      "step": 3931
    },
    {
      "epoch": 2.5156749840051185,
      "grad_norm": 1.7533983390683092,
      "learning_rate": 1.75339468386596e-05,
      "loss": 0.4113,
      "step": 3932
    },
    {
      "epoch": 2.5163147792706333,
      "grad_norm": 2.2237356962155403,
      "learning_rate": 1.753258409354973e-05,
      "loss": 0.4017,
      "step": 3933
    },
    {
      "epoch": 2.5169545745361486,
      "grad_norm": 1.8405695558194268,
      "learning_rate": 1.75312210250041e-05,
      "loss": 0.395,
      "step": 3934
    },
    {
      "epoch": 2.5175943698016634,
      "grad_norm": 1.6185341384261762,
      "learning_rate": 1.7529857633081233e-05,
      "loss": 0.3927,
      "step": 3935
    },
    {
      "epoch": 2.5182341650671782,
      "grad_norm": 1.7459333410901519,
      "learning_rate": 1.7528493917839678e-05,
      "loss": 0.3635,
      "step": 3936
    },
    {
      "epoch": 2.5188739603326935,
      "grad_norm": 1.769668379498624,
      "learning_rate": 1.7527129879337984e-05,
      "loss": 0.389,
      "step": 3937
    },
    {
      "epoch": 2.519513755598209,
      "grad_norm": 1.9049433162769165,
      "learning_rate": 1.7525765517634724e-05,
      "loss": 0.4289,
      "step": 3938
    },
    {
      "epoch": 2.5201535508637236,
      "grad_norm": 1.9814084889221122,
      "learning_rate": 1.752440083278848e-05,
      "loss": 0.3848,
      "step": 3939
    },
    {
      "epoch": 2.5207933461292384,
      "grad_norm": 1.823964658514506,
      "learning_rate": 1.7523035824857845e-05,
      "loss": 0.3742,
      "step": 3940
    },
    {
      "epoch": 2.5214331413947537,
      "grad_norm": 1.8264170514868014,
      "learning_rate": 1.752167049390144e-05,
      "loss": 0.4168,
      "step": 3941
    },
    {
      "epoch": 2.5220729366602685,
      "grad_norm": 1.8364056052019457,
      "learning_rate": 1.752030483997788e-05,
      "loss": 0.405,
      "step": 3942
    },
    {
      "epoch": 2.522712731925784,
      "grad_norm": 1.8895400595075418,
      "learning_rate": 1.7518938863145808e-05,
      "loss": 0.3993,
      "step": 3943
    },
    {
      "epoch": 2.5233525271912987,
      "grad_norm": 1.692797797841324,
      "learning_rate": 1.7517572563463877e-05,
      "loss": 0.3546,
      "step": 3944
    },
    {
      "epoch": 2.523992322456814,
      "grad_norm": 1.9452832643723248,
      "learning_rate": 1.7516205940990755e-05,
      "loss": 0.4164,
      "step": 3945
    },
    {
      "epoch": 2.5246321177223288,
      "grad_norm": 1.7375527374380912,
      "learning_rate": 1.751483899578512e-05,
      "loss": 0.3895,
      "step": 3946
    },
    {
      "epoch": 2.525271912987844,
      "grad_norm": 1.8214431830038276,
      "learning_rate": 1.7513471727905666e-05,
      "loss": 0.3661,
      "step": 3947
    },
    {
      "epoch": 2.525911708253359,
      "grad_norm": 1.7409144346460335,
      "learning_rate": 1.7512104137411098e-05,
      "loss": 0.3706,
      "step": 3948
    },
    {
      "epoch": 2.526551503518874,
      "grad_norm": 1.8519838334553471,
      "learning_rate": 1.7510736224360146e-05,
      "loss": 0.3932,
      "step": 3949
    },
    {
      "epoch": 2.527191298784389,
      "grad_norm": 1.9252697980715727,
      "learning_rate": 1.750936798881154e-05,
      "loss": 0.3858,
      "step": 3950
    },
    {
      "epoch": 2.527831094049904,
      "grad_norm": 1.80948290859236,
      "learning_rate": 1.7507999430824026e-05,
      "loss": 0.4135,
      "step": 3951
    },
    {
      "epoch": 2.528470889315419,
      "grad_norm": 1.809371861094632,
      "learning_rate": 1.750663055045638e-05,
      "loss": 0.3909,
      "step": 3952
    },
    {
      "epoch": 2.5291106845809344,
      "grad_norm": 1.9019076400122414,
      "learning_rate": 1.7505261347767364e-05,
      "loss": 0.4531,
      "step": 3953
    },
    {
      "epoch": 2.529750479846449,
      "grad_norm": 1.8821138680202991,
      "learning_rate": 1.7503891822815783e-05,
      "loss": 0.3628,
      "step": 3954
    },
    {
      "epoch": 2.530390275111964,
      "grad_norm": 1.733516079677016,
      "learning_rate": 1.7502521975660428e-05,
      "loss": 0.3736,
      "step": 3955
    },
    {
      "epoch": 2.5310300703774793,
      "grad_norm": 1.7791584877111701,
      "learning_rate": 1.750115180636013e-05,
      "loss": 0.3978,
      "step": 3956
    },
    {
      "epoch": 2.531669865642994,
      "grad_norm": 1.8873273036330245,
      "learning_rate": 1.7499781314973717e-05,
      "loss": 0.3932,
      "step": 3957
    },
    {
      "epoch": 2.5323096609085094,
      "grad_norm": 1.7760063428687731,
      "learning_rate": 1.7498410501560035e-05,
      "loss": 0.4016,
      "step": 3958
    },
    {
      "epoch": 2.5329494561740242,
      "grad_norm": 1.7262499373288875,
      "learning_rate": 1.7497039366177942e-05,
      "loss": 0.4113,
      "step": 3959
    },
    {
      "epoch": 2.5335892514395395,
      "grad_norm": 1.7809619248059294,
      "learning_rate": 1.7495667908886318e-05,
      "loss": 0.3811,
      "step": 3960
    },
    {
      "epoch": 2.5342290467050543,
      "grad_norm": 1.931198594258376,
      "learning_rate": 1.749429612974405e-05,
      "loss": 0.375,
      "step": 3961
    },
    {
      "epoch": 2.534868841970569,
      "grad_norm": 1.6480004254124379,
      "learning_rate": 1.7492924028810036e-05,
      "loss": 0.3476,
      "step": 3962
    },
    {
      "epoch": 2.5355086372360844,
      "grad_norm": 1.7678955969701415,
      "learning_rate": 1.749155160614319e-05,
      "loss": 0.3928,
      "step": 3963
    },
    {
      "epoch": 2.5361484325015997,
      "grad_norm": 1.7662715815489862,
      "learning_rate": 1.7490178861802444e-05,
      "loss": 0.387,
      "step": 3964
    },
    {
      "epoch": 2.5367882277671145,
      "grad_norm": 1.7542559516743748,
      "learning_rate": 1.7488805795846744e-05,
      "loss": 0.3721,
      "step": 3965
    },
    {
      "epoch": 2.5374280230326294,
      "grad_norm": 1.8029805609085359,
      "learning_rate": 1.7487432408335047e-05,
      "loss": 0.3929,
      "step": 3966
    },
    {
      "epoch": 2.5380678182981447,
      "grad_norm": 1.7440170003741426,
      "learning_rate": 1.748605869932632e-05,
      "loss": 0.387,
      "step": 3967
    },
    {
      "epoch": 2.5387076135636595,
      "grad_norm": 1.7862339375475582,
      "learning_rate": 1.7484684668879545e-05,
      "loss": 0.4276,
      "step": 3968
    },
    {
      "epoch": 2.5393474088291748,
      "grad_norm": 1.824246729797484,
      "learning_rate": 1.748331031705373e-05,
      "loss": 0.3678,
      "step": 3969
    },
    {
      "epoch": 2.5399872040946896,
      "grad_norm": 1.7976812096197918,
      "learning_rate": 1.7481935643907875e-05,
      "loss": 0.4222,
      "step": 3970
    },
    {
      "epoch": 2.540626999360205,
      "grad_norm": 1.8410549020022184,
      "learning_rate": 1.748056064950102e-05,
      "loss": 0.4146,
      "step": 3971
    },
    {
      "epoch": 2.5412667946257197,
      "grad_norm": 1.883665434726397,
      "learning_rate": 1.7479185333892195e-05,
      "loss": 0.4261,
      "step": 3972
    },
    {
      "epoch": 2.541906589891235,
      "grad_norm": 1.870072872048158,
      "learning_rate": 1.7477809697140454e-05,
      "loss": 0.4125,
      "step": 3973
    },
    {
      "epoch": 2.54254638515675,
      "grad_norm": 1.7155327997273926,
      "learning_rate": 1.747643373930487e-05,
      "loss": 0.3709,
      "step": 3974
    },
    {
      "epoch": 2.543186180422265,
      "grad_norm": 1.8899943892963187,
      "learning_rate": 1.747505746044452e-05,
      "loss": 0.4003,
      "step": 3975
    },
    {
      "epoch": 2.54382597568778,
      "grad_norm": 1.7619570637880135,
      "learning_rate": 1.74736808606185e-05,
      "loss": 0.4147,
      "step": 3976
    },
    {
      "epoch": 2.5444657709532947,
      "grad_norm": 1.7993957629178257,
      "learning_rate": 1.747230393988592e-05,
      "loss": 0.3575,
      "step": 3977
    },
    {
      "epoch": 2.54510556621881,
      "grad_norm": 1.9253760277956224,
      "learning_rate": 1.7470926698305895e-05,
      "loss": 0.4272,
      "step": 3978
    },
    {
      "epoch": 2.5457453614843253,
      "grad_norm": 1.6882056282693507,
      "learning_rate": 1.746954913593757e-05,
      "loss": 0.3631,
      "step": 3979
    },
    {
      "epoch": 2.54638515674984,
      "grad_norm": 1.689552162862371,
      "learning_rate": 1.7468171252840094e-05,
      "loss": 0.3736,
      "step": 3980
    },
    {
      "epoch": 2.547024952015355,
      "grad_norm": 1.7535906809631943,
      "learning_rate": 1.746679304907263e-05,
      "loss": 0.4178,
      "step": 3981
    },
    {
      "epoch": 2.5476647472808702,
      "grad_norm": 1.6639569457975272,
      "learning_rate": 1.746541452469435e-05,
      "loss": 0.3904,
      "step": 3982
    },
    {
      "epoch": 2.548304542546385,
      "grad_norm": 1.6737837747374855,
      "learning_rate": 1.7464035679764455e-05,
      "loss": 0.3487,
      "step": 3983
    },
    {
      "epoch": 2.5489443378119003,
      "grad_norm": 1.762685459598082,
      "learning_rate": 1.7462656514342144e-05,
      "loss": 0.4114,
      "step": 3984
    },
    {
      "epoch": 2.549584133077415,
      "grad_norm": 1.797805432279225,
      "learning_rate": 1.746127702848664e-05,
      "loss": 0.3693,
      "step": 3985
    },
    {
      "epoch": 2.5502239283429304,
      "grad_norm": 1.6668187578074498,
      "learning_rate": 1.745989722225717e-05,
      "loss": 0.3556,
      "step": 3986
    },
    {
      "epoch": 2.5508637236084453,
      "grad_norm": 1.888905858728682,
      "learning_rate": 1.7458517095712984e-05,
      "loss": 0.3855,
      "step": 3987
    },
    {
      "epoch": 2.55150351887396,
      "grad_norm": 1.668776571195504,
      "learning_rate": 1.745713664891334e-05,
      "loss": 0.3265,
      "step": 3988
    },
    {
      "epoch": 2.5521433141394754,
      "grad_norm": 1.7743660937888495,
      "learning_rate": 1.7455755881917515e-05,
      "loss": 0.4005,
      "step": 3989
    },
    {
      "epoch": 2.5527831094049906,
      "grad_norm": 1.7790752115024595,
      "learning_rate": 1.7454374794784796e-05,
      "loss": 0.3793,
      "step": 3990
    },
    {
      "epoch": 2.5534229046705055,
      "grad_norm": 2.14641604604741,
      "learning_rate": 1.7452993387574482e-05,
      "loss": 0.3875,
      "step": 3991
    },
    {
      "epoch": 2.5540626999360203,
      "grad_norm": 1.8416244409640048,
      "learning_rate": 1.745161166034589e-05,
      "loss": 0.3668,
      "step": 3992
    },
    {
      "epoch": 2.5547024952015356,
      "grad_norm": 1.7898796416311449,
      "learning_rate": 1.745022961315835e-05,
      "loss": 0.4045,
      "step": 3993
    },
    {
      "epoch": 2.5553422904670504,
      "grad_norm": 1.7819102029569216,
      "learning_rate": 1.7448847246071203e-05,
      "loss": 0.388,
      "step": 3994
    },
    {
      "epoch": 2.5559820857325657,
      "grad_norm": 1.8040973938990879,
      "learning_rate": 1.7447464559143806e-05,
      "loss": 0.3972,
      "step": 3995
    },
    {
      "epoch": 2.5566218809980805,
      "grad_norm": 2.0513574226301037,
      "learning_rate": 1.744608155243553e-05,
      "loss": 0.387,
      "step": 3996
    },
    {
      "epoch": 2.557261676263596,
      "grad_norm": 2.0538777709067,
      "learning_rate": 1.744469822600575e-05,
      "loss": 0.4148,
      "step": 3997
    },
    {
      "epoch": 2.5579014715291106,
      "grad_norm": 1.9857208976552887,
      "learning_rate": 1.7443314579913878e-05,
      "loss": 0.3806,
      "step": 3998
    },
    {
      "epoch": 2.5585412667946255,
      "grad_norm": 2.053690965191097,
      "learning_rate": 1.744193061421932e-05,
      "loss": 0.4145,
      "step": 3999
    },
    {
      "epoch": 2.5591810620601407,
      "grad_norm": 1.9428802780724939,
      "learning_rate": 1.7440546328981497e-05,
      "loss": 0.4492,
      "step": 4000
    },
    {
      "epoch": 2.559820857325656,
      "grad_norm": 1.786681804962933,
      "learning_rate": 1.743916172425985e-05,
      "loss": 0.3626,
      "step": 4001
    },
    {
      "epoch": 2.560460652591171,
      "grad_norm": 1.9074462800456025,
      "learning_rate": 1.7437776800113833e-05,
      "loss": 0.4015,
      "step": 4002
    },
    {
      "epoch": 2.5611004478566857,
      "grad_norm": 1.9234199192519263,
      "learning_rate": 1.743639155660291e-05,
      "loss": 0.3812,
      "step": 4003
    },
    {
      "epoch": 2.561740243122201,
      "grad_norm": 1.7455078435168239,
      "learning_rate": 1.743500599378656e-05,
      "loss": 0.3628,
      "step": 4004
    },
    {
      "epoch": 2.5623800383877158,
      "grad_norm": 1.7704654814898837,
      "learning_rate": 1.7433620111724283e-05,
      "loss": 0.3698,
      "step": 4005
    },
    {
      "epoch": 2.563019833653231,
      "grad_norm": 1.844600618558509,
      "learning_rate": 1.743223391047558e-05,
      "loss": 0.3933,
      "step": 4006
    },
    {
      "epoch": 2.563659628918746,
      "grad_norm": 1.806637596326798,
      "learning_rate": 1.743084739009997e-05,
      "loss": 0.4171,
      "step": 4007
    },
    {
      "epoch": 2.564299424184261,
      "grad_norm": 1.7339509833343507,
      "learning_rate": 1.7429460550656997e-05,
      "loss": 0.3291,
      "step": 4008
    },
    {
      "epoch": 2.564939219449776,
      "grad_norm": 1.8956033056030184,
      "learning_rate": 1.7428073392206202e-05,
      "loss": 0.3891,
      "step": 4009
    },
    {
      "epoch": 2.5655790147152913,
      "grad_norm": 1.8978335413254221,
      "learning_rate": 1.742668591480715e-05,
      "loss": 0.3586,
      "step": 4010
    },
    {
      "epoch": 2.566218809980806,
      "grad_norm": 1.697193217996073,
      "learning_rate": 1.7425298118519413e-05,
      "loss": 0.4145,
      "step": 4011
    },
    {
      "epoch": 2.5668586052463214,
      "grad_norm": 1.745257982258745,
      "learning_rate": 1.7423910003402587e-05,
      "loss": 0.3762,
      "step": 4012
    },
    {
      "epoch": 2.567498400511836,
      "grad_norm": 1.7311585913726728,
      "learning_rate": 1.742252156951627e-05,
      "loss": 0.375,
      "step": 4013
    },
    {
      "epoch": 2.568138195777351,
      "grad_norm": 1.7365251678442684,
      "learning_rate": 1.7421132816920078e-05,
      "loss": 0.3782,
      "step": 4014
    },
    {
      "epoch": 2.5687779910428663,
      "grad_norm": 1.691839093382294,
      "learning_rate": 1.7419743745673647e-05,
      "loss": 0.3725,
      "step": 4015
    },
    {
      "epoch": 2.5694177863083816,
      "grad_norm": 1.7309834846263317,
      "learning_rate": 1.7418354355836617e-05,
      "loss": 0.3883,
      "step": 4016
    },
    {
      "epoch": 2.5700575815738964,
      "grad_norm": 1.7900522158738617,
      "learning_rate": 1.741696464746865e-05,
      "loss": 0.4134,
      "step": 4017
    },
    {
      "epoch": 2.5706973768394112,
      "grad_norm": 1.9767790494681892,
      "learning_rate": 1.741557462062941e-05,
      "loss": 0.4131,
      "step": 4018
    },
    {
      "epoch": 2.5713371721049265,
      "grad_norm": 1.804450096073915,
      "learning_rate": 1.741418427537859e-05,
      "loss": 0.4037,
      "step": 4019
    },
    {
      "epoch": 2.5719769673704413,
      "grad_norm": 1.8265950808415308,
      "learning_rate": 1.7412793611775883e-05,
      "loss": 0.3537,
      "step": 4020
    },
    {
      "epoch": 2.5726167626359566,
      "grad_norm": 1.6691209285965205,
      "learning_rate": 1.7411402629881007e-05,
      "loss": 0.3953,
      "step": 4021
    },
    {
      "epoch": 2.5732565579014715,
      "grad_norm": 1.7296891152088865,
      "learning_rate": 1.741001132975369e-05,
      "loss": 0.3788,
      "step": 4022
    },
    {
      "epoch": 2.5738963531669867,
      "grad_norm": 1.833203594611914,
      "learning_rate": 1.7408619711453662e-05,
      "loss": 0.3795,
      "step": 4023
    },
    {
      "epoch": 2.5745361484325016,
      "grad_norm": 2.000094446082064,
      "learning_rate": 1.7407227775040682e-05,
      "loss": 0.4446,
      "step": 4024
    },
    {
      "epoch": 2.5751759436980164,
      "grad_norm": 1.966084977982086,
      "learning_rate": 1.7405835520574524e-05,
      "loss": 0.4087,
      "step": 4025
    },
    {
      "epoch": 2.5758157389635317,
      "grad_norm": 1.9145367287126314,
      "learning_rate": 1.7404442948114956e-05,
      "loss": 0.3831,
      "step": 4026
    },
    {
      "epoch": 2.576455534229047,
      "grad_norm": 1.871077622657157,
      "learning_rate": 1.740305005772178e-05,
      "loss": 0.4351,
      "step": 4027
    },
    {
      "epoch": 2.5770953294945618,
      "grad_norm": 1.6855132070862004,
      "learning_rate": 1.740165684945481e-05,
      "loss": 0.403,
      "step": 4028
    },
    {
      "epoch": 2.5777351247600766,
      "grad_norm": 1.7067964076980782,
      "learning_rate": 1.7400263323373858e-05,
      "loss": 0.4064,
      "step": 4029
    },
    {
      "epoch": 2.578374920025592,
      "grad_norm": 1.7382136896492337,
      "learning_rate": 1.739886947953876e-05,
      "loss": 0.367,
      "step": 4030
    },
    {
      "epoch": 2.5790147152911067,
      "grad_norm": 1.820307078029869,
      "learning_rate": 1.739747531800937e-05,
      "loss": 0.3622,
      "step": 4031
    },
    {
      "epoch": 2.579654510556622,
      "grad_norm": 1.733555681981825,
      "learning_rate": 1.739608083884555e-05,
      "loss": 0.3888,
      "step": 4032
    },
    {
      "epoch": 2.580294305822137,
      "grad_norm": 1.8796406986012646,
      "learning_rate": 1.7394686042107173e-05,
      "loss": 0.3869,
      "step": 4033
    },
    {
      "epoch": 2.580934101087652,
      "grad_norm": 1.8566335463601624,
      "learning_rate": 1.7393290927854135e-05,
      "loss": 0.3969,
      "step": 4034
    },
    {
      "epoch": 2.581573896353167,
      "grad_norm": 1.804468992464907,
      "learning_rate": 1.7391895496146332e-05,
      "loss": 0.3664,
      "step": 4035
    },
    {
      "epoch": 2.5822136916186818,
      "grad_norm": 1.9554480092596411,
      "learning_rate": 1.739049974704369e-05,
      "loss": 0.3992,
      "step": 4036
    },
    {
      "epoch": 2.582853486884197,
      "grad_norm": 1.838237333712872,
      "learning_rate": 1.7389103680606132e-05,
      "loss": 0.385,
      "step": 4037
    },
    {
      "epoch": 2.5834932821497123,
      "grad_norm": 1.9024554163649305,
      "learning_rate": 1.738770729689361e-05,
      "loss": 0.4018,
      "step": 4038
    },
    {
      "epoch": 2.584133077415227,
      "grad_norm": 1.733415220032622,
      "learning_rate": 1.738631059596607e-05,
      "loss": 0.3839,
      "step": 4039
    },
    {
      "epoch": 2.584772872680742,
      "grad_norm": 1.6728038628648576,
      "learning_rate": 1.73849135778835e-05,
      "loss": 0.3757,
      "step": 4040
    },
    {
      "epoch": 2.5854126679462572,
      "grad_norm": 1.7815764046831206,
      "learning_rate": 1.7383516242705874e-05,
      "loss": 0.3833,
      "step": 4041
    },
    {
      "epoch": 2.586052463211772,
      "grad_norm": 1.7055566369393662,
      "learning_rate": 1.73821185904932e-05,
      "loss": 0.4216,
      "step": 4042
    },
    {
      "epoch": 2.5866922584772873,
      "grad_norm": 1.6518873010260593,
      "learning_rate": 1.7380720621305483e-05,
      "loss": 0.3266,
      "step": 4043
    },
    {
      "epoch": 2.587332053742802,
      "grad_norm": 1.8026495733172447,
      "learning_rate": 1.737932233520275e-05,
      "loss": 0.3908,
      "step": 4044
    },
    {
      "epoch": 2.5879718490083174,
      "grad_norm": 2.115753692813469,
      "learning_rate": 1.7377923732245043e-05,
      "loss": 0.4117,
      "step": 4045
    },
    {
      "epoch": 2.5886116442738323,
      "grad_norm": 1.7981664493654463,
      "learning_rate": 1.7376524812492415e-05,
      "loss": 0.3652,
      "step": 4046
    },
    {
      "epoch": 2.5892514395393476,
      "grad_norm": 1.9615092600873258,
      "learning_rate": 1.7375125576004933e-05,
      "loss": 0.3998,
      "step": 4047
    },
    {
      "epoch": 2.5898912348048624,
      "grad_norm": 2.012624370915542,
      "learning_rate": 1.737372602284268e-05,
      "loss": 0.4093,
      "step": 4048
    },
    {
      "epoch": 2.5905310300703777,
      "grad_norm": 2.0090571855549277,
      "learning_rate": 1.737232615306575e-05,
      "loss": 0.4474,
      "step": 4049
    },
    {
      "epoch": 2.5911708253358925,
      "grad_norm": 1.7850006916656984,
      "learning_rate": 1.7370925966734246e-05,
      "loss": 0.391,
      "step": 4050
    },
    {
      "epoch": 2.5918106206014073,
      "grad_norm": 1.7613686159758783,
      "learning_rate": 1.7369525463908292e-05,
      "loss": 0.3913,
      "step": 4051
    },
    {
      "epoch": 2.5924504158669226,
      "grad_norm": 1.842620019177169,
      "learning_rate": 1.7368124644648027e-05,
      "loss": 0.351,
      "step": 4052
    },
    {
      "epoch": 2.593090211132438,
      "grad_norm": 1.742863822972533,
      "learning_rate": 1.7366723509013598e-05,
      "loss": 0.3852,
      "step": 4053
    },
    {
      "epoch": 2.5937300063979527,
      "grad_norm": 1.7997795331778828,
      "learning_rate": 1.736532205706516e-05,
      "loss": 0.4295,
      "step": 4054
    },
    {
      "epoch": 2.5943698016634675,
      "grad_norm": 1.711612856246511,
      "learning_rate": 1.73639202888629e-05,
      "loss": 0.4227,
      "step": 4055
    },
    {
      "epoch": 2.595009596928983,
      "grad_norm": 1.8145686665693466,
      "learning_rate": 1.7362518204467e-05,
      "loss": 0.3994,
      "step": 4056
    },
    {
      "epoch": 2.5956493921944976,
      "grad_norm": 1.7286375984441469,
      "learning_rate": 1.7361115803937666e-05,
      "loss": 0.3812,
      "step": 4057
    },
    {
      "epoch": 2.596289187460013,
      "grad_norm": 1.9257583605454494,
      "learning_rate": 1.7359713087335112e-05,
      "loss": 0.3478,
      "step": 4058
    },
    {
      "epoch": 2.5969289827255277,
      "grad_norm": 1.810036138070011,
      "learning_rate": 1.7358310054719572e-05,
      "loss": 0.4329,
      "step": 4059
    },
    {
      "epoch": 2.597568777991043,
      "grad_norm": 2.002005879775642,
      "learning_rate": 1.7356906706151286e-05,
      "loss": 0.4127,
      "step": 4060
    },
    {
      "epoch": 2.598208573256558,
      "grad_norm": 1.8461517589372376,
      "learning_rate": 1.7355503041690513e-05,
      "loss": 0.4207,
      "step": 4061
    },
    {
      "epoch": 2.5988483685220727,
      "grad_norm": 1.705243151057416,
      "learning_rate": 1.7354099061397523e-05,
      "loss": 0.3473,
      "step": 4062
    },
    {
      "epoch": 2.599488163787588,
      "grad_norm": 1.815209812435502,
      "learning_rate": 1.7352694765332606e-05,
      "loss": 0.3831,
      "step": 4063
    },
    {
      "epoch": 2.6001279590531032,
      "grad_norm": 1.8611970190890406,
      "learning_rate": 1.7351290153556052e-05,
      "loss": 0.3865,
      "step": 4064
    },
    {
      "epoch": 2.600767754318618,
      "grad_norm": 1.960446418311431,
      "learning_rate": 1.7349885226128175e-05,
      "loss": 0.3987,
      "step": 4065
    },
    {
      "epoch": 2.601407549584133,
      "grad_norm": 1.9191678057366066,
      "learning_rate": 1.7348479983109303e-05,
      "loss": 0.3985,
      "step": 4066
    },
    {
      "epoch": 2.602047344849648,
      "grad_norm": 1.8254855390405464,
      "learning_rate": 1.7347074424559772e-05,
      "loss": 0.4381,
      "step": 4067
    },
    {
      "epoch": 2.602687140115163,
      "grad_norm": 1.861326715042625,
      "learning_rate": 1.734566855053993e-05,
      "loss": 0.3953,
      "step": 4068
    },
    {
      "epoch": 2.6033269353806783,
      "grad_norm": 1.9773486273314809,
      "learning_rate": 1.7344262361110154e-05,
      "loss": 0.4279,
      "step": 4069
    },
    {
      "epoch": 2.603966730646193,
      "grad_norm": 1.7992462720564821,
      "learning_rate": 1.7342855856330814e-05,
      "loss": 0.369,
      "step": 4070
    },
    {
      "epoch": 2.6046065259117084,
      "grad_norm": 1.9339541585190598,
      "learning_rate": 1.7341449036262304e-05,
      "loss": 0.4207,
      "step": 4071
    },
    {
      "epoch": 2.605246321177223,
      "grad_norm": 1.555179414806961,
      "learning_rate": 1.7340041900965035e-05,
      "loss": 0.3741,
      "step": 4072
    },
    {
      "epoch": 2.6058861164427385,
      "grad_norm": 1.9467209195553663,
      "learning_rate": 1.733863445049942e-05,
      "loss": 0.3806,
      "step": 4073
    },
    {
      "epoch": 2.6065259117082533,
      "grad_norm": 1.924464584222551,
      "learning_rate": 1.73372266849259e-05,
      "loss": 0.3886,
      "step": 4074
    },
    {
      "epoch": 2.6071657069737686,
      "grad_norm": 1.894784092598507,
      "learning_rate": 1.7335818604304916e-05,
      "loss": 0.4155,
      "step": 4075
    },
    {
      "epoch": 2.6078055022392834,
      "grad_norm": 1.8485790357905165,
      "learning_rate": 1.733441020869693e-05,
      "loss": 0.367,
      "step": 4076
    },
    {
      "epoch": 2.6084452975047983,
      "grad_norm": 1.8074478792865643,
      "learning_rate": 1.7333001498162415e-05,
      "loss": 0.4222,
      "step": 4077
    },
    {
      "epoch": 2.6090850927703135,
      "grad_norm": 1.7985040758262758,
      "learning_rate": 1.7331592472761865e-05,
      "loss": 0.3969,
      "step": 4078
    },
    {
      "epoch": 2.609724888035829,
      "grad_norm": 1.8745178796590454,
      "learning_rate": 1.733018313255577e-05,
      "loss": 0.4501,
      "step": 4079
    },
    {
      "epoch": 2.6103646833013436,
      "grad_norm": 1.8018018903241675,
      "learning_rate": 1.7328773477604656e-05,
      "loss": 0.3521,
      "step": 4080
    },
    {
      "epoch": 2.6110044785668585,
      "grad_norm": 1.759203931304201,
      "learning_rate": 1.732736350796904e-05,
      "loss": 0.3669,
      "step": 4081
    },
    {
      "epoch": 2.6116442738323737,
      "grad_norm": 1.8525602841853768,
      "learning_rate": 1.7325953223709477e-05,
      "loss": 0.4281,
      "step": 4082
    },
    {
      "epoch": 2.6122840690978886,
      "grad_norm": 1.9465751462543623,
      "learning_rate": 1.732454262488651e-05,
      "loss": 0.4558,
      "step": 4083
    },
    {
      "epoch": 2.612923864363404,
      "grad_norm": 1.802018193041992,
      "learning_rate": 1.7323131711560712e-05,
      "loss": 0.4113,
      "step": 4084
    },
    {
      "epoch": 2.6135636596289187,
      "grad_norm": 1.7861009750565247,
      "learning_rate": 1.7321720483792665e-05,
      "loss": 0.3617,
      "step": 4085
    },
    {
      "epoch": 2.614203454894434,
      "grad_norm": 1.926751686963373,
      "learning_rate": 1.7320308941642966e-05,
      "loss": 0.4316,
      "step": 4086
    },
    {
      "epoch": 2.614843250159949,
      "grad_norm": 1.8710400424481122,
      "learning_rate": 1.7318897085172222e-05,
      "loss": 0.4085,
      "step": 4087
    },
    {
      "epoch": 2.6154830454254636,
      "grad_norm": 1.7598958526265265,
      "learning_rate": 1.7317484914441057e-05,
      "loss": 0.4041,
      "step": 4088
    },
    {
      "epoch": 2.616122840690979,
      "grad_norm": 1.8915659259752748,
      "learning_rate": 1.7316072429510104e-05,
      "loss": 0.399,
      "step": 4089
    },
    {
      "epoch": 2.616762635956494,
      "grad_norm": 1.7501751515892752,
      "learning_rate": 1.731465963044002e-05,
      "loss": 0.3707,
      "step": 4090
    },
    {
      "epoch": 2.617402431222009,
      "grad_norm": 1.6726625865019966,
      "learning_rate": 1.731324651729146e-05,
      "loss": 0.4026,
      "step": 4091
    },
    {
      "epoch": 2.618042226487524,
      "grad_norm": 1.8030566365102898,
      "learning_rate": 1.7311833090125105e-05,
      "loss": 0.4451,
      "step": 4092
    },
    {
      "epoch": 2.618682021753039,
      "grad_norm": 1.80866987966382,
      "learning_rate": 1.7310419349001645e-05,
      "loss": 0.365,
      "step": 4093
    },
    {
      "epoch": 2.619321817018554,
      "grad_norm": 1.8813267332377728,
      "learning_rate": 1.730900529398178e-05,
      "loss": 0.3784,
      "step": 4094
    },
    {
      "epoch": 2.619961612284069,
      "grad_norm": 1.7701711747104383,
      "learning_rate": 1.7307590925126227e-05,
      "loss": 0.342,
      "step": 4095
    },
    {
      "epoch": 2.620601407549584,
      "grad_norm": 1.6102468843615563,
      "learning_rate": 1.7306176242495722e-05,
      "loss": 0.3683,
      "step": 4096
    },
    {
      "epoch": 2.6212412028150993,
      "grad_norm": 1.9232439145874722,
      "learning_rate": 1.7304761246151007e-05,
      "loss": 0.3607,
      "step": 4097
    },
    {
      "epoch": 2.621880998080614,
      "grad_norm": 1.814233274366818,
      "learning_rate": 1.730334593615284e-05,
      "loss": 0.4143,
      "step": 4098
    },
    {
      "epoch": 2.622520793346129,
      "grad_norm": 1.8002646340190185,
      "learning_rate": 1.7301930312561984e-05,
      "loss": 0.4177,
      "step": 4099
    },
    {
      "epoch": 2.6231605886116443,
      "grad_norm": 1.7989219518161248,
      "learning_rate": 1.7300514375439232e-05,
      "loss": 0.4202,
      "step": 4100
    },
    {
      "epoch": 2.6238003838771595,
      "grad_norm": 1.9398720198837291,
      "learning_rate": 1.7299098124845382e-05,
      "loss": 0.4321,
      "step": 4101
    },
    {
      "epoch": 2.6244401791426744,
      "grad_norm": 1.6705348366947332,
      "learning_rate": 1.729768156084124e-05,
      "loss": 0.3849,
      "step": 4102
    },
    {
      "epoch": 2.625079974408189,
      "grad_norm": 1.8043708130808922,
      "learning_rate": 1.7296264683487633e-05,
      "loss": 0.3631,
      "step": 4103
    },
    {
      "epoch": 2.6257197696737045,
      "grad_norm": 1.7703276260289647,
      "learning_rate": 1.72948474928454e-05,
      "loss": 0.3485,
      "step": 4104
    },
    {
      "epoch": 2.6263595649392193,
      "grad_norm": 1.9506773458423692,
      "learning_rate": 1.7293429988975393e-05,
      "loss": 0.3904,
      "step": 4105
    },
    {
      "epoch": 2.6269993602047346,
      "grad_norm": 1.9323145268550939,
      "learning_rate": 1.729201217193848e-05,
      "loss": 0.3966,
      "step": 4106
    },
    {
      "epoch": 2.6276391554702494,
      "grad_norm": 1.6775929548527997,
      "learning_rate": 1.7290594041795527e-05,
      "loss": 0.3611,
      "step": 4107
    },
    {
      "epoch": 2.6282789507357647,
      "grad_norm": 1.8639866776821123,
      "learning_rate": 1.7289175598607442e-05,
      "loss": 0.3906,
      "step": 4108
    },
    {
      "epoch": 2.6289187460012795,
      "grad_norm": 1.8387767671076305,
      "learning_rate": 1.728775684243512e-05,
      "loss": 0.3656,
      "step": 4109
    },
    {
      "epoch": 2.629558541266795,
      "grad_norm": 1.702665338327588,
      "learning_rate": 1.7286337773339484e-05,
      "loss": 0.3736,
      "step": 4110
    },
    {
      "epoch": 2.6301983365323096,
      "grad_norm": 1.841179204806438,
      "learning_rate": 1.7284918391381464e-05,
      "loss": 0.392,
      "step": 4111
    },
    {
      "epoch": 2.630838131797825,
      "grad_norm": 1.7339155569076,
      "learning_rate": 1.728349869662201e-05,
      "loss": 0.3657,
      "step": 4112
    },
    {
      "epoch": 2.6314779270633397,
      "grad_norm": 1.770636723827289,
      "learning_rate": 1.7282078689122074e-05,
      "loss": 0.4092,
      "step": 4113
    },
    {
      "epoch": 2.6321177223288545,
      "grad_norm": 1.82793840102438,
      "learning_rate": 1.7280658368942636e-05,
      "loss": 0.3945,
      "step": 4114
    },
    {
      "epoch": 2.63275751759437,
      "grad_norm": 1.8392596828648218,
      "learning_rate": 1.727923773614468e-05,
      "loss": 0.315,
      "step": 4115
    },
    {
      "epoch": 2.633397312859885,
      "grad_norm": 1.9812748757641534,
      "learning_rate": 1.72778167907892e-05,
      "loss": 0.3407,
      "step": 4116
    },
    {
      "epoch": 2.6340371081254,
      "grad_norm": 1.7232747812903881,
      "learning_rate": 1.727639553293722e-05,
      "loss": 0.3864,
      "step": 4117
    },
    {
      "epoch": 2.6346769033909148,
      "grad_norm": 2.005640901352231,
      "learning_rate": 1.7274973962649756e-05,
      "loss": 0.4104,
      "step": 4118
    },
    {
      "epoch": 2.63531669865643,
      "grad_norm": 1.8161423064307478,
      "learning_rate": 1.727355207998785e-05,
      "loss": 0.3651,
      "step": 4119
    },
    {
      "epoch": 2.635956493921945,
      "grad_norm": 1.9904450646591647,
      "learning_rate": 1.727212988501256e-05,
      "loss": 0.4206,
      "step": 4120
    },
    {
      "epoch": 2.63659628918746,
      "grad_norm": 1.8905510101519765,
      "learning_rate": 1.7270707377784944e-05,
      "loss": 0.4192,
      "step": 4121
    },
    {
      "epoch": 2.637236084452975,
      "grad_norm": 1.8226734033203402,
      "learning_rate": 1.7269284558366088e-05,
      "loss": 0.3665,
      "step": 4122
    },
    {
      "epoch": 2.6378758797184902,
      "grad_norm": 1.8279691929195698,
      "learning_rate": 1.7267861426817084e-05,
      "loss": 0.441,
      "step": 4123
    },
    {
      "epoch": 2.638515674984005,
      "grad_norm": 1.784030704777507,
      "learning_rate": 1.7266437983199042e-05,
      "loss": 0.3623,
      "step": 4124
    },
    {
      "epoch": 2.63915547024952,
      "grad_norm": 1.773151678006417,
      "learning_rate": 1.7265014227573074e-05,
      "loss": 0.3902,
      "step": 4125
    },
    {
      "epoch": 2.639795265515035,
      "grad_norm": 1.8118355983187133,
      "learning_rate": 1.7263590160000324e-05,
      "loss": 0.3552,
      "step": 4126
    },
    {
      "epoch": 2.6404350607805505,
      "grad_norm": 1.7736372248367176,
      "learning_rate": 1.726216578054193e-05,
      "loss": 0.4166,
      "step": 4127
    },
    {
      "epoch": 2.6410748560460653,
      "grad_norm": 1.9168600162725669,
      "learning_rate": 1.7260741089259056e-05,
      "loss": 0.3883,
      "step": 4128
    },
    {
      "epoch": 2.64171465131158,
      "grad_norm": 1.94585895306698,
      "learning_rate": 1.725931608621288e-05,
      "loss": 0.4007,
      "step": 4129
    },
    {
      "epoch": 2.6423544465770954,
      "grad_norm": 1.8695111253952696,
      "learning_rate": 1.7257890771464576e-05,
      "loss": 0.3956,
      "step": 4130
    },
    {
      "epoch": 2.6429942418426102,
      "grad_norm": 1.8122385376503092,
      "learning_rate": 1.7256465145075362e-05,
      "loss": 0.4444,
      "step": 4131
    },
    {
      "epoch": 2.6436340371081255,
      "grad_norm": 1.8704523245062872,
      "learning_rate": 1.7255039207106437e-05,
      "loss": 0.364,
      "step": 4132
    },
    {
      "epoch": 2.6442738323736403,
      "grad_norm": 1.7689764074255594,
      "learning_rate": 1.7253612957619034e-05,
      "loss": 0.3768,
      "step": 4133
    },
    {
      "epoch": 2.6449136276391556,
      "grad_norm": 1.7749045071412826,
      "learning_rate": 1.7252186396674395e-05,
      "loss": 0.4033,
      "step": 4134
    },
    {
      "epoch": 2.6455534229046704,
      "grad_norm": 1.747716209699103,
      "learning_rate": 1.7250759524333776e-05,
      "loss": 0.381,
      "step": 4135
    },
    {
      "epoch": 2.6461932181701853,
      "grad_norm": 1.7603395684861756,
      "learning_rate": 1.724933234065844e-05,
      "loss": 0.3512,
      "step": 4136
    },
    {
      "epoch": 2.6468330134357005,
      "grad_norm": 1.7478796578879332,
      "learning_rate": 1.7247904845709664e-05,
      "loss": 0.4231,
      "step": 4137
    },
    {
      "epoch": 2.647472808701216,
      "grad_norm": 1.8352032655572508,
      "learning_rate": 1.724647703954875e-05,
      "loss": 0.3559,
      "step": 4138
    },
    {
      "epoch": 2.6481126039667306,
      "grad_norm": 1.7763703404869264,
      "learning_rate": 1.7245048922237004e-05,
      "loss": 0.3969,
      "step": 4139
    },
    {
      "epoch": 2.6487523992322455,
      "grad_norm": 1.8318182627108288,
      "learning_rate": 1.7243620493835743e-05,
      "loss": 0.3494,
      "step": 4140
    },
    {
      "epoch": 2.6493921944977608,
      "grad_norm": 1.7562920979359382,
      "learning_rate": 1.7242191754406304e-05,
      "loss": 0.3499,
      "step": 4141
    },
    {
      "epoch": 2.6500319897632756,
      "grad_norm": 1.9173822357271368,
      "learning_rate": 1.7240762704010034e-05,
      "loss": 0.3705,
      "step": 4142
    },
    {
      "epoch": 2.650671785028791,
      "grad_norm": 1.8220291921447584,
      "learning_rate": 1.7239333342708297e-05,
      "loss": 0.4101,
      "step": 4143
    },
    {
      "epoch": 2.6513115802943057,
      "grad_norm": 1.8139124296326745,
      "learning_rate": 1.723790367056246e-05,
      "loss": 0.4046,
      "step": 4144
    },
    {
      "epoch": 2.651951375559821,
      "grad_norm": 1.7442199615791956,
      "learning_rate": 1.7236473687633916e-05,
      "loss": 0.3807,
      "step": 4145
    },
    {
      "epoch": 2.652591170825336,
      "grad_norm": 1.8364351248469952,
      "learning_rate": 1.723504339398406e-05,
      "loss": 0.3787,
      "step": 4146
    },
    {
      "epoch": 2.653230966090851,
      "grad_norm": 1.7472008687288674,
      "learning_rate": 1.7233612789674315e-05,
      "loss": 0.3733,
      "step": 4147
    },
    {
      "epoch": 2.653870761356366,
      "grad_norm": 1.8257635219420998,
      "learning_rate": 1.7232181874766106e-05,
      "loss": 0.3717,
      "step": 4148
    },
    {
      "epoch": 2.654510556621881,
      "grad_norm": 1.6978836489298452,
      "learning_rate": 1.7230750649320872e-05,
      "loss": 0.3714,
      "step": 4149
    },
    {
      "epoch": 2.655150351887396,
      "grad_norm": 1.5865309461707588,
      "learning_rate": 1.7229319113400066e-05,
      "loss": 0.3665,
      "step": 4150
    },
    {
      "epoch": 2.655790147152911,
      "grad_norm": 1.6829816096999182,
      "learning_rate": 1.7227887267065158e-05,
      "loss": 0.3587,
      "step": 4151
    },
    {
      "epoch": 2.656429942418426,
      "grad_norm": 1.7522749606800587,
      "learning_rate": 1.7226455110377628e-05,
      "loss": 0.3298,
      "step": 4152
    },
    {
      "epoch": 2.6570697376839414,
      "grad_norm": 1.7586430557932382,
      "learning_rate": 1.722502264339897e-05,
      "loss": 0.4044,
      "step": 4153
    },
    {
      "epoch": 2.657709532949456,
      "grad_norm": 1.9480387183259806,
      "learning_rate": 1.7223589866190693e-05,
      "loss": 0.3987,
      "step": 4154
    },
    {
      "epoch": 2.658349328214971,
      "grad_norm": 1.9333804018790168,
      "learning_rate": 1.7222156778814315e-05,
      "loss": 0.4364,
      "step": 4155
    },
    {
      "epoch": 2.6589891234804863,
      "grad_norm": 1.969846359606416,
      "learning_rate": 1.7220723381331376e-05,
      "loss": 0.4576,
      "step": 4156
    },
    {
      "epoch": 2.659628918746001,
      "grad_norm": 1.935402946006263,
      "learning_rate": 1.7219289673803413e-05,
      "loss": 0.4079,
      "step": 4157
    },
    {
      "epoch": 2.6602687140115164,
      "grad_norm": 1.7763713333573046,
      "learning_rate": 1.7217855656292e-05,
      "loss": 0.3523,
      "step": 4158
    },
    {
      "epoch": 2.6609085092770313,
      "grad_norm": 1.7829115888824134,
      "learning_rate": 1.7216421328858703e-05,
      "loss": 0.3643,
      "step": 4159
    },
    {
      "epoch": 2.6615483045425465,
      "grad_norm": 1.7546199228859856,
      "learning_rate": 1.721498669156511e-05,
      "loss": 0.3253,
      "step": 4160
    },
    {
      "epoch": 2.6621880998080614,
      "grad_norm": 1.6718202141158722,
      "learning_rate": 1.7213551744472822e-05,
      "loss": 0.3466,
      "step": 4161
    },
    {
      "epoch": 2.662827895073576,
      "grad_norm": 1.7047079227949105,
      "learning_rate": 1.7212116487643455e-05,
      "loss": 0.4414,
      "step": 4162
    },
    {
      "epoch": 2.6634676903390915,
      "grad_norm": 1.8297720796543298,
      "learning_rate": 1.7210680921138636e-05,
      "loss": 0.376,
      "step": 4163
    },
    {
      "epoch": 2.6641074856046068,
      "grad_norm": 1.66482422989058,
      "learning_rate": 1.7209245045020005e-05,
      "loss": 0.3476,
      "step": 4164
    },
    {
      "epoch": 2.6647472808701216,
      "grad_norm": 1.719396830329992,
      "learning_rate": 1.7207808859349216e-05,
      "loss": 0.3064,
      "step": 4165
    },
    {
      "epoch": 2.6653870761356364,
      "grad_norm": 1.7872142760971053,
      "learning_rate": 1.7206372364187937e-05,
      "loss": 0.344,
      "step": 4166
    },
    {
      "epoch": 2.6660268714011517,
      "grad_norm": 1.8070181407383783,
      "learning_rate": 1.7204935559597847e-05,
      "loss": 0.417,
      "step": 4167
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 1.7120476653151888,
      "learning_rate": 1.720349844564064e-05,
      "loss": 0.3553,
      "step": 4168
    },
    {
      "epoch": 2.667306461932182,
      "grad_norm": 1.665908886011083,
      "learning_rate": 1.7202061022378023e-05,
      "loss": 0.3676,
      "step": 4169
    },
    {
      "epoch": 2.6679462571976966,
      "grad_norm": 1.7936914476816834,
      "learning_rate": 1.720062328987172e-05,
      "loss": 0.3566,
      "step": 4170
    },
    {
      "epoch": 2.668586052463212,
      "grad_norm": 1.8835381291456432,
      "learning_rate": 1.719918524818346e-05,
      "loss": 0.3529,
      "step": 4171
    },
    {
      "epoch": 2.6692258477287267,
      "grad_norm": 1.9015183709415002,
      "learning_rate": 1.719774689737499e-05,
      "loss": 0.3923,
      "step": 4172
    },
    {
      "epoch": 2.669865642994242,
      "grad_norm": 1.7508450828286781,
      "learning_rate": 1.7196308237508078e-05,
      "loss": 0.3884,
      "step": 4173
    },
    {
      "epoch": 2.670505438259757,
      "grad_norm": 1.7345833339891312,
      "learning_rate": 1.7194869268644484e-05,
      "loss": 0.3632,
      "step": 4174
    },
    {
      "epoch": 2.671145233525272,
      "grad_norm": 1.7725237730222305,
      "learning_rate": 1.7193429990846006e-05,
      "loss": 0.403,
      "step": 4175
    },
    {
      "epoch": 2.671785028790787,
      "grad_norm": 1.7793800612492483,
      "learning_rate": 1.7191990404174442e-05,
      "loss": 0.3958,
      "step": 4176
    },
    {
      "epoch": 2.6724248240563018,
      "grad_norm": 1.7150407297576913,
      "learning_rate": 1.71905505086916e-05,
      "loss": 0.3954,
      "step": 4177
    },
    {
      "epoch": 2.673064619321817,
      "grad_norm": 1.7947053284971388,
      "learning_rate": 1.7189110304459313e-05,
      "loss": 0.4683,
      "step": 4178
    },
    {
      "epoch": 2.6737044145873323,
      "grad_norm": 1.8382227209710411,
      "learning_rate": 1.7187669791539418e-05,
      "loss": 0.4455,
      "step": 4179
    },
    {
      "epoch": 2.674344209852847,
      "grad_norm": 1.8514102644174162,
      "learning_rate": 1.7186228969993767e-05,
      "loss": 0.3738,
      "step": 4180
    },
    {
      "epoch": 2.674984005118362,
      "grad_norm": 1.70584611884999,
      "learning_rate": 1.7184787839884226e-05,
      "loss": 0.4027,
      "step": 4181
    },
    {
      "epoch": 2.6756238003838773,
      "grad_norm": 1.7716566742749396,
      "learning_rate": 1.7183346401272678e-05,
      "loss": 0.4086,
      "step": 4182
    },
    {
      "epoch": 2.676263595649392,
      "grad_norm": 1.7379068533549835,
      "learning_rate": 1.7181904654221015e-05,
      "loss": 0.3439,
      "step": 4183
    },
    {
      "epoch": 2.6769033909149074,
      "grad_norm": 1.7557549967343928,
      "learning_rate": 1.7180462598791137e-05,
      "loss": 0.3674,
      "step": 4184
    },
    {
      "epoch": 2.677543186180422,
      "grad_norm": 1.7832340174731147,
      "learning_rate": 1.7179020235044972e-05,
      "loss": 0.3655,
      "step": 4185
    },
    {
      "epoch": 2.6781829814459375,
      "grad_norm": 1.5842151741790043,
      "learning_rate": 1.7177577563044447e-05,
      "loss": 0.3132,
      "step": 4186
    },
    {
      "epoch": 2.6788227767114523,
      "grad_norm": 1.7259900008343292,
      "learning_rate": 1.717613458285151e-05,
      "loss": 0.3917,
      "step": 4187
    },
    {
      "epoch": 2.679462571976967,
      "grad_norm": 1.8570704582921185,
      "learning_rate": 1.717469129452812e-05,
      "loss": 0.3977,
      "step": 4188
    },
    {
      "epoch": 2.6801023672424824,
      "grad_norm": 1.7661843562645327,
      "learning_rate": 1.7173247698136246e-05,
      "loss": 0.3866,
      "step": 4189
    },
    {
      "epoch": 2.6807421625079977,
      "grad_norm": 1.835540730353934,
      "learning_rate": 1.717180379373788e-05,
      "loss": 0.4025,
      "step": 4190
    },
    {
      "epoch": 2.6813819577735125,
      "grad_norm": 1.7873414980437818,
      "learning_rate": 1.7170359581395014e-05,
      "loss": 0.3385,
      "step": 4191
    },
    {
      "epoch": 2.6820217530390273,
      "grad_norm": 1.7856366572585411,
      "learning_rate": 1.7168915061169662e-05,
      "loss": 0.3896,
      "step": 4192
    },
    {
      "epoch": 2.6826615483045426,
      "grad_norm": 1.7885240896273893,
      "learning_rate": 1.716747023312385e-05,
      "loss": 0.4245,
      "step": 4193
    },
    {
      "epoch": 2.6833013435700575,
      "grad_norm": 1.6236971524807078,
      "learning_rate": 1.7166025097319616e-05,
      "loss": 0.356,
      "step": 4194
    },
    {
      "epoch": 2.6839411388355727,
      "grad_norm": 1.8765348196387,
      "learning_rate": 1.7164579653819014e-05,
      "loss": 0.4008,
      "step": 4195
    },
    {
      "epoch": 2.6845809341010876,
      "grad_norm": 1.8458947126624259,
      "learning_rate": 1.7163133902684105e-05,
      "loss": 0.3657,
      "step": 4196
    },
    {
      "epoch": 2.685220729366603,
      "grad_norm": 1.809553568810735,
      "learning_rate": 1.7161687843976967e-05,
      "loss": 0.3716,
      "step": 4197
    },
    {
      "epoch": 2.6858605246321177,
      "grad_norm": 1.9193498484744111,
      "learning_rate": 1.71602414777597e-05,
      "loss": 0.3656,
      "step": 4198
    },
    {
      "epoch": 2.6865003198976325,
      "grad_norm": 1.769002915747781,
      "learning_rate": 1.7158794804094393e-05,
      "loss": 0.3599,
      "step": 4199
    },
    {
      "epoch": 2.6871401151631478,
      "grad_norm": 1.736731255667386,
      "learning_rate": 1.7157347823043173e-05,
      "loss": 0.4203,
      "step": 4200
    },
    {
      "epoch": 2.687779910428663,
      "grad_norm": 1.6731030122683819,
      "learning_rate": 1.7155900534668172e-05,
      "loss": 0.378,
      "step": 4201
    },
    {
      "epoch": 2.688419705694178,
      "grad_norm": 1.6927182192599621,
      "learning_rate": 1.715445293903153e-05,
      "loss": 0.3592,
      "step": 4202
    },
    {
      "epoch": 2.6890595009596927,
      "grad_norm": 1.9345358611515113,
      "learning_rate": 1.7153005036195402e-05,
      "loss": 0.4082,
      "step": 4203
    },
    {
      "epoch": 2.689699296225208,
      "grad_norm": 1.7712221948256368,
      "learning_rate": 1.7151556826221968e-05,
      "loss": 0.3846,
      "step": 4204
    },
    {
      "epoch": 2.690339091490723,
      "grad_norm": 1.7107122842236364,
      "learning_rate": 1.7150108309173404e-05,
      "loss": 0.3779,
      "step": 4205
    },
    {
      "epoch": 2.690978886756238,
      "grad_norm": 1.7194523644172077,
      "learning_rate": 1.7148659485111905e-05,
      "loss": 0.3466,
      "step": 4206
    },
    {
      "epoch": 2.691618682021753,
      "grad_norm": 1.7296740490563824,
      "learning_rate": 1.7147210354099682e-05,
      "loss": 0.3341,
      "step": 4207
    },
    {
      "epoch": 2.692258477287268,
      "grad_norm": 1.774090964715173,
      "learning_rate": 1.714576091619897e-05,
      "loss": 0.3627,
      "step": 4208
    },
    {
      "epoch": 2.692898272552783,
      "grad_norm": 1.831288806955999,
      "learning_rate": 1.714431117147199e-05,
      "loss": 0.3559,
      "step": 4209
    },
    {
      "epoch": 2.6935380678182983,
      "grad_norm": 1.7463674022141333,
      "learning_rate": 1.7142861119980995e-05,
      "loss": 0.4114,
      "step": 4210
    },
    {
      "epoch": 2.694177863083813,
      "grad_norm": 1.7716681578419173,
      "learning_rate": 1.714141076178825e-05,
      "loss": 0.3399,
      "step": 4211
    },
    {
      "epoch": 2.6948176583493284,
      "grad_norm": 1.828215795222698,
      "learning_rate": 1.7139960096956035e-05,
      "loss": 0.3804,
      "step": 4212
    },
    {
      "epoch": 2.6954574536148432,
      "grad_norm": 1.8353654554831011,
      "learning_rate": 1.713850912554663e-05,
      "loss": 0.4216,
      "step": 4213
    },
    {
      "epoch": 2.696097248880358,
      "grad_norm": 1.7446284229267146,
      "learning_rate": 1.7137057847622345e-05,
      "loss": 0.3269,
      "step": 4214
    },
    {
      "epoch": 2.6967370441458733,
      "grad_norm": 1.7651741937796377,
      "learning_rate": 1.713560626324549e-05,
      "loss": 0.3823,
      "step": 4215
    },
    {
      "epoch": 2.6973768394113886,
      "grad_norm": 1.7264546885313934,
      "learning_rate": 1.7134154372478395e-05,
      "loss": 0.4161,
      "step": 4216
    },
    {
      "epoch": 2.6980166346769034,
      "grad_norm": 1.7904020623674264,
      "learning_rate": 1.7132702175383405e-05,
      "loss": 0.3853,
      "step": 4217
    },
    {
      "epoch": 2.6986564299424183,
      "grad_norm": 1.7607482858808037,
      "learning_rate": 1.7131249672022866e-05,
      "loss": 0.3982,
      "step": 4218
    },
    {
      "epoch": 2.6992962252079336,
      "grad_norm": 1.7773275265185458,
      "learning_rate": 1.7129796862459158e-05,
      "loss": 0.3545,
      "step": 4219
    },
    {
      "epoch": 2.6999360204734484,
      "grad_norm": 1.8117025589809468,
      "learning_rate": 1.712834374675465e-05,
      "loss": 0.3764,
      "step": 4220
    },
    {
      "epoch": 2.7005758157389637,
      "grad_norm": 1.7164413027858032,
      "learning_rate": 1.7126890324971745e-05,
      "loss": 0.3517,
      "step": 4221
    },
    {
      "epoch": 2.7012156110044785,
      "grad_norm": 1.7775509173532187,
      "learning_rate": 1.7125436597172846e-05,
      "loss": 0.3575,
      "step": 4222
    },
    {
      "epoch": 2.7018554062699938,
      "grad_norm": 1.9674700817641275,
      "learning_rate": 1.7123982563420375e-05,
      "loss": 0.3963,
      "step": 4223
    },
    {
      "epoch": 2.7024952015355086,
      "grad_norm": 1.809674925343007,
      "learning_rate": 1.7122528223776768e-05,
      "loss": 0.3699,
      "step": 4224
    },
    {
      "epoch": 2.7031349968010234,
      "grad_norm": 1.8663052577433195,
      "learning_rate": 1.7121073578304464e-05,
      "loss": 0.3762,
      "step": 4225
    },
    {
      "epoch": 2.7037747920665387,
      "grad_norm": 1.6153105760958468,
      "learning_rate": 1.711961862706593e-05,
      "loss": 0.3938,
      "step": 4226
    },
    {
      "epoch": 2.704414587332054,
      "grad_norm": 1.6554706903866212,
      "learning_rate": 1.7118163370123637e-05,
      "loss": 0.3588,
      "step": 4227
    },
    {
      "epoch": 2.705054382597569,
      "grad_norm": 1.8646219866813818,
      "learning_rate": 1.711670780754007e-05,
      "loss": 0.4031,
      "step": 4228
    },
    {
      "epoch": 2.7056941778630836,
      "grad_norm": 1.8298264690258168,
      "learning_rate": 1.7115251939377734e-05,
      "loss": 0.4252,
      "step": 4229
    },
    {
      "epoch": 2.706333973128599,
      "grad_norm": 1.9270308916919932,
      "learning_rate": 1.7113795765699134e-05,
      "loss": 0.4026,
      "step": 4230
    },
    {
      "epoch": 2.7069737683941137,
      "grad_norm": 1.7120989414896741,
      "learning_rate": 1.7112339286566796e-05,
      "loss": 0.3505,
      "step": 4231
    },
    {
      "epoch": 2.707613563659629,
      "grad_norm": 1.8285021875643763,
      "learning_rate": 1.7110882502043265e-05,
      "loss": 0.3538,
      "step": 4232
    },
    {
      "epoch": 2.708253358925144,
      "grad_norm": 1.7549597955518932,
      "learning_rate": 1.7109425412191085e-05,
      "loss": 0.3635,
      "step": 4233
    },
    {
      "epoch": 2.708893154190659,
      "grad_norm": 1.8353217105812776,
      "learning_rate": 1.7107968017072825e-05,
      "loss": 0.404,
      "step": 4234
    },
    {
      "epoch": 2.709532949456174,
      "grad_norm": 1.7165150497584767,
      "learning_rate": 1.7106510316751068e-05,
      "loss": 0.4311,
      "step": 4235
    },
    {
      "epoch": 2.710172744721689,
      "grad_norm": 1.5829186661829064,
      "learning_rate": 1.710505231128839e-05,
      "loss": 0.3654,
      "step": 4236
    },
    {
      "epoch": 2.710812539987204,
      "grad_norm": 1.7376965096414259,
      "learning_rate": 1.710359400074741e-05,
      "loss": 0.3438,
      "step": 4237
    },
    {
      "epoch": 2.7114523352527193,
      "grad_norm": 1.9233707513510652,
      "learning_rate": 1.710213538519074e-05,
      "loss": 0.3726,
      "step": 4238
    },
    {
      "epoch": 2.712092130518234,
      "grad_norm": 1.9506586337750156,
      "learning_rate": 1.710067646468101e-05,
      "loss": 0.4073,
      "step": 4239
    },
    {
      "epoch": 2.712731925783749,
      "grad_norm": 1.721504360019392,
      "learning_rate": 1.7099217239280865e-05,
      "loss": 0.358,
      "step": 4240
    },
    {
      "epoch": 2.7133717210492643,
      "grad_norm": 1.835200193652629,
      "learning_rate": 1.7097757709052957e-05,
      "loss": 0.3469,
      "step": 4241
    },
    {
      "epoch": 2.714011516314779,
      "grad_norm": 1.8268131957633507,
      "learning_rate": 1.7096297874059962e-05,
      "loss": 0.3947,
      "step": 4242
    },
    {
      "epoch": 2.7146513115802944,
      "grad_norm": 1.7163493414092905,
      "learning_rate": 1.7094837734364557e-05,
      "loss": 0.3516,
      "step": 4243
    },
    {
      "epoch": 2.715291106845809,
      "grad_norm": 1.7784246102127537,
      "learning_rate": 1.7093377290029443e-05,
      "loss": 0.4012,
      "step": 4244
    },
    {
      "epoch": 2.7159309021113245,
      "grad_norm": 1.7385456682436053,
      "learning_rate": 1.7091916541117324e-05,
      "loss": 0.4103,
      "step": 4245
    },
    {
      "epoch": 2.7165706973768393,
      "grad_norm": 1.8053692381467816,
      "learning_rate": 1.7090455487690927e-05,
      "loss": 0.3551,
      "step": 4246
    },
    {
      "epoch": 2.7172104926423546,
      "grad_norm": 1.9472714147983223,
      "learning_rate": 1.7088994129812985e-05,
      "loss": 0.4427,
      "step": 4247
    },
    {
      "epoch": 2.7178502879078694,
      "grad_norm": 1.7341158941730135,
      "learning_rate": 1.708753246754624e-05,
      "loss": 0.3842,
      "step": 4248
    },
    {
      "epoch": 2.7184900831733847,
      "grad_norm": 1.7710248366507657,
      "learning_rate": 1.708607050095346e-05,
      "loss": 0.361,
      "step": 4249
    },
    {
      "epoch": 2.7191298784388995,
      "grad_norm": 1.8043352116668248,
      "learning_rate": 1.7084608230097423e-05,
      "loss": 0.412,
      "step": 4250
    },
    {
      "epoch": 2.7197696737044144,
      "grad_norm": 1.706850234450833,
      "learning_rate": 1.7083145655040904e-05,
      "loss": 0.3836,
      "step": 4251
    },
    {
      "epoch": 2.7204094689699296,
      "grad_norm": 1.974162886860278,
      "learning_rate": 1.7081682775846716e-05,
      "loss": 0.4211,
      "step": 4252
    },
    {
      "epoch": 2.721049264235445,
      "grad_norm": 1.766727233841734,
      "learning_rate": 1.708021959257766e-05,
      "loss": 0.3734,
      "step": 4253
    },
    {
      "epoch": 2.7216890595009597,
      "grad_norm": 1.697533966165444,
      "learning_rate": 1.707875610529658e-05,
      "loss": 0.3714,
      "step": 4254
    },
    {
      "epoch": 2.7223288547664746,
      "grad_norm": 1.8332796181658422,
      "learning_rate": 1.70772923140663e-05,
      "loss": 0.3659,
      "step": 4255
    },
    {
      "epoch": 2.72296865003199,
      "grad_norm": 1.7395745626879542,
      "learning_rate": 1.7075828218949675e-05,
      "loss": 0.3963,
      "step": 4256
    },
    {
      "epoch": 2.7236084452975047,
      "grad_norm": 1.8036464771770377,
      "learning_rate": 1.7074363820009577e-05,
      "loss": 0.4316,
      "step": 4257
    },
    {
      "epoch": 2.72424824056302,
      "grad_norm": 1.8419754026911839,
      "learning_rate": 1.7072899117308877e-05,
      "loss": 0.389,
      "step": 4258
    },
    {
      "epoch": 2.724888035828535,
      "grad_norm": 1.7263458399740723,
      "learning_rate": 1.7071434110910474e-05,
      "loss": 0.4186,
      "step": 4259
    },
    {
      "epoch": 2.72552783109405,
      "grad_norm": 1.7386892880610374,
      "learning_rate": 1.7069968800877265e-05,
      "loss": 0.3773,
      "step": 4260
    },
    {
      "epoch": 2.726167626359565,
      "grad_norm": 1.7929889552298663,
      "learning_rate": 1.7068503187272177e-05,
      "loss": 0.3992,
      "step": 4261
    },
    {
      "epoch": 2.7268074216250797,
      "grad_norm": 1.7341956491762023,
      "learning_rate": 1.7067037270158135e-05,
      "loss": 0.3827,
      "step": 4262
    },
    {
      "epoch": 2.727447216890595,
      "grad_norm": 1.873403646083764,
      "learning_rate": 1.706557104959808e-05,
      "loss": 0.4773,
      "step": 4263
    },
    {
      "epoch": 2.7280870121561103,
      "grad_norm": 1.7036558677359503,
      "learning_rate": 1.7064104525654982e-05,
      "loss": 0.3441,
      "step": 4264
    },
    {
      "epoch": 2.728726807421625,
      "grad_norm": 1.7106933028885087,
      "learning_rate": 1.7062637698391793e-05,
      "loss": 0.3985,
      "step": 4265
    },
    {
      "epoch": 2.72936660268714,
      "grad_norm": 1.769947585705654,
      "learning_rate": 1.7061170567871504e-05,
      "loss": 0.3541,
      "step": 4266
    },
    {
      "epoch": 2.730006397952655,
      "grad_norm": 1.7129898493564153,
      "learning_rate": 1.705970313415712e-05,
      "loss": 0.3286,
      "step": 4267
    },
    {
      "epoch": 2.73064619321817,
      "grad_norm": 1.8554877020243175,
      "learning_rate": 1.7058235397311637e-05,
      "loss": 0.4133,
      "step": 4268
    },
    {
      "epoch": 2.7312859884836853,
      "grad_norm": 1.7669024353243523,
      "learning_rate": 1.705676735739808e-05,
      "loss": 0.3898,
      "step": 4269
    },
    {
      "epoch": 2.7319257837492,
      "grad_norm": 1.830007346862392,
      "learning_rate": 1.7055299014479486e-05,
      "loss": 0.3621,
      "step": 4270
    },
    {
      "epoch": 2.7325655790147154,
      "grad_norm": 1.857715317302628,
      "learning_rate": 1.7053830368618904e-05,
      "loss": 0.409,
      "step": 4271
    },
    {
      "epoch": 2.7332053742802302,
      "grad_norm": 1.7574856491779138,
      "learning_rate": 1.7052361419879395e-05,
      "loss": 0.3753,
      "step": 4272
    },
    {
      "epoch": 2.7338451695457455,
      "grad_norm": 1.813791042663809,
      "learning_rate": 1.7050892168324028e-05,
      "loss": 0.443,
      "step": 4273
    },
    {
      "epoch": 2.7344849648112604,
      "grad_norm": 1.8547875713953415,
      "learning_rate": 1.7049422614015898e-05,
      "loss": 0.3818,
      "step": 4274
    },
    {
      "epoch": 2.7351247600767756,
      "grad_norm": 1.706158035632516,
      "learning_rate": 1.7047952757018097e-05,
      "loss": 0.407,
      "step": 4275
    },
    {
      "epoch": 2.7357645553422905,
      "grad_norm": 1.8492228660502112,
      "learning_rate": 1.7046482597393745e-05,
      "loss": 0.3748,
      "step": 4276
    },
    {
      "epoch": 2.7364043506078053,
      "grad_norm": 1.869048003418955,
      "learning_rate": 1.7045012135205964e-05,
      "loss": 0.3339,
      "step": 4277
    },
    {
      "epoch": 2.7370441458733206,
      "grad_norm": 1.6344924372002445,
      "learning_rate": 1.7043541370517896e-05,
      "loss": 0.4529,
      "step": 4278
    },
    {
      "epoch": 2.737683941138836,
      "grad_norm": 1.904609718329307,
      "learning_rate": 1.704207030339269e-05,
      "loss": 0.4029,
      "step": 4279
    },
    {
      "epoch": 2.7383237364043507,
      "grad_norm": 1.7175177765344118,
      "learning_rate": 1.704059893389351e-05,
      "loss": 0.3631,
      "step": 4280
    },
    {
      "epoch": 2.7389635316698655,
      "grad_norm": 1.6506109008863883,
      "learning_rate": 1.703912726208354e-05,
      "loss": 0.4007,
      "step": 4281
    },
    {
      "epoch": 2.739603326935381,
      "grad_norm": 1.7201489903849958,
      "learning_rate": 1.7037655288025962e-05,
      "loss": 0.3843,
      "step": 4282
    },
    {
      "epoch": 2.7402431222008956,
      "grad_norm": 1.767449633619511,
      "learning_rate": 1.7036183011783985e-05,
      "loss": 0.3525,
      "step": 4283
    },
    {
      "epoch": 2.740882917466411,
      "grad_norm": 1.8156092850987344,
      "learning_rate": 1.703471043342083e-05,
      "loss": 0.3441,
      "step": 4284
    },
    {
      "epoch": 2.7415227127319257,
      "grad_norm": 1.8178668747890563,
      "learning_rate": 1.703323755299972e-05,
      "loss": 0.3411,
      "step": 4285
    },
    {
      "epoch": 2.742162507997441,
      "grad_norm": 1.884349159920045,
      "learning_rate": 1.70317643705839e-05,
      "loss": 0.4153,
      "step": 4286
    },
    {
      "epoch": 2.742802303262956,
      "grad_norm": 1.7381898608622095,
      "learning_rate": 1.703029088623663e-05,
      "loss": 0.3824,
      "step": 4287
    },
    {
      "epoch": 2.7434420985284707,
      "grad_norm": 1.6970237904800463,
      "learning_rate": 1.7028817100021173e-05,
      "loss": 0.3486,
      "step": 4288
    },
    {
      "epoch": 2.744081893793986,
      "grad_norm": 1.8063775417589434,
      "learning_rate": 1.7027343012000813e-05,
      "loss": 0.3638,
      "step": 4289
    },
    {
      "epoch": 2.744721689059501,
      "grad_norm": 1.877608767719871,
      "learning_rate": 1.7025868622238844e-05,
      "loss": 0.3726,
      "step": 4290
    },
    {
      "epoch": 2.745361484325016,
      "grad_norm": 2.170919910575917,
      "learning_rate": 1.7024393930798577e-05,
      "loss": 0.4158,
      "step": 4291
    },
    {
      "epoch": 2.746001279590531,
      "grad_norm": 1.807877736214834,
      "learning_rate": 1.7022918937743327e-05,
      "loss": 0.4062,
      "step": 4292
    },
    {
      "epoch": 2.746641074856046,
      "grad_norm": 1.7264216713782112,
      "learning_rate": 1.7021443643136434e-05,
      "loss": 0.3559,
      "step": 4293
    },
    {
      "epoch": 2.747280870121561,
      "grad_norm": 1.704460211135688,
      "learning_rate": 1.7019968047041237e-05,
      "loss": 0.3852,
      "step": 4294
    },
    {
      "epoch": 2.7479206653870762,
      "grad_norm": 1.718053623794642,
      "learning_rate": 1.7018492149521103e-05,
      "loss": 0.3955,
      "step": 4295
    },
    {
      "epoch": 2.748560460652591,
      "grad_norm": 1.822313893446373,
      "learning_rate": 1.70170159506394e-05,
      "loss": 0.4126,
      "step": 4296
    },
    {
      "epoch": 2.7492002559181064,
      "grad_norm": 1.7086642549144675,
      "learning_rate": 1.7015539450459514e-05,
      "loss": 0.379,
      "step": 4297
    },
    {
      "epoch": 2.749840051183621,
      "grad_norm": 1.8147412705252748,
      "learning_rate": 1.7014062649044843e-05,
      "loss": 0.4006,
      "step": 4298
    },
    {
      "epoch": 2.750479846449136,
      "grad_norm": 1.8083257683234017,
      "learning_rate": 1.70125855464588e-05,
      "loss": 0.3831,
      "step": 4299
    },
    {
      "epoch": 2.7511196417146513,
      "grad_norm": 1.8682841446294642,
      "learning_rate": 1.701110814276481e-05,
      "loss": 0.4291,
      "step": 4300
    },
    {
      "epoch": 2.7517594369801666,
      "grad_norm": 1.6898758646780345,
      "learning_rate": 1.7009630438026303e-05,
      "loss": 0.3384,
      "step": 4301
    },
    {
      "epoch": 2.7523992322456814,
      "grad_norm": 1.7936982017143532,
      "learning_rate": 1.700815243230674e-05,
      "loss": 0.4223,
      "step": 4302
    },
    {
      "epoch": 2.7530390275111962,
      "grad_norm": 1.8199778256897348,
      "learning_rate": 1.7006674125669572e-05,
      "loss": 0.3852,
      "step": 4303
    },
    {
      "epoch": 2.7536788227767115,
      "grad_norm": 1.8184273356300502,
      "learning_rate": 1.7005195518178285e-05,
      "loss": 0.3722,
      "step": 4304
    },
    {
      "epoch": 2.7543186180422263,
      "grad_norm": 1.7805223188940327,
      "learning_rate": 1.700371660989636e-05,
      "loss": 0.3862,
      "step": 4305
    },
    {
      "epoch": 2.7549584133077416,
      "grad_norm": 1.6283950672819378,
      "learning_rate": 1.700223740088731e-05,
      "loss": 0.4032,
      "step": 4306
    },
    {
      "epoch": 2.7555982085732564,
      "grad_norm": 1.781035276475869,
      "learning_rate": 1.7000757891214637e-05,
      "loss": 0.4207,
      "step": 4307
    },
    {
      "epoch": 2.7562380038387717,
      "grad_norm": 1.7127041813696615,
      "learning_rate": 1.6999278080941874e-05,
      "loss": 0.3792,
      "step": 4308
    },
    {
      "epoch": 2.7568777991042865,
      "grad_norm": 1.6651704505032465,
      "learning_rate": 1.6997797970132557e-05,
      "loss": 0.3637,
      "step": 4309
    },
    {
      "epoch": 2.757517594369802,
      "grad_norm": 1.6668068863638255,
      "learning_rate": 1.699631755885025e-05,
      "loss": 0.4113,
      "step": 4310
    },
    {
      "epoch": 2.7581573896353166,
      "grad_norm": 1.8867486487566447,
      "learning_rate": 1.6994836847158512e-05,
      "loss": 0.422,
      "step": 4311
    },
    {
      "epoch": 2.758797184900832,
      "grad_norm": 1.7941898598488837,
      "learning_rate": 1.699335583512092e-05,
      "loss": 0.3823,
      "step": 4312
    },
    {
      "epoch": 2.7594369801663468,
      "grad_norm": 1.9327600051891602,
      "learning_rate": 1.6991874522801066e-05,
      "loss": 0.3905,
      "step": 4313
    },
    {
      "epoch": 2.7600767754318616,
      "grad_norm": 1.826522674585377,
      "learning_rate": 1.699039291026256e-05,
      "loss": 0.3899,
      "step": 4314
    },
    {
      "epoch": 2.760716570697377,
      "grad_norm": 1.6859512677159572,
      "learning_rate": 1.6988910997569018e-05,
      "loss": 0.3879,
      "step": 4315
    },
    {
      "epoch": 2.761356365962892,
      "grad_norm": 1.7512436940812803,
      "learning_rate": 1.6987428784784072e-05,
      "loss": 0.3777,
      "step": 4316
    },
    {
      "epoch": 2.761996161228407,
      "grad_norm": 1.659330021630049,
      "learning_rate": 1.698594627197136e-05,
      "loss": 0.3466,
      "step": 4317
    },
    {
      "epoch": 2.762635956493922,
      "grad_norm": 1.6258388262651555,
      "learning_rate": 1.6984463459194543e-05,
      "loss": 0.3608,
      "step": 4318
    },
    {
      "epoch": 2.763275751759437,
      "grad_norm": 1.5417894121763251,
      "learning_rate": 1.698298034651729e-05,
      "loss": 0.3117,
      "step": 4319
    },
    {
      "epoch": 2.763915547024952,
      "grad_norm": 1.6876899175204128,
      "learning_rate": 1.6981496934003284e-05,
      "loss": 0.3856,
      "step": 4320
    },
    {
      "epoch": 2.764555342290467,
      "grad_norm": 1.7556208729529585,
      "learning_rate": 1.6980013221716216e-05,
      "loss": 0.4561,
      "step": 4321
    },
    {
      "epoch": 2.765195137555982,
      "grad_norm": 1.7597745330934762,
      "learning_rate": 1.6978529209719797e-05,
      "loss": 0.3576,
      "step": 4322
    },
    {
      "epoch": 2.7658349328214973,
      "grad_norm": 1.6039700054529589,
      "learning_rate": 1.6977044898077752e-05,
      "loss": 0.3151,
      "step": 4323
    },
    {
      "epoch": 2.766474728087012,
      "grad_norm": 1.783808422923834,
      "learning_rate": 1.6975560286853807e-05,
      "loss": 0.3549,
      "step": 4324
    },
    {
      "epoch": 2.767114523352527,
      "grad_norm": 1.8090721217580015,
      "learning_rate": 1.697407537611171e-05,
      "loss": 0.414,
      "step": 4325
    },
    {
      "epoch": 2.767754318618042,
      "grad_norm": 1.6772723130710971,
      "learning_rate": 1.6972590165915228e-05,
      "loss": 0.3259,
      "step": 4326
    },
    {
      "epoch": 2.7683941138835575,
      "grad_norm": 1.8908458372559924,
      "learning_rate": 1.6971104656328123e-05,
      "loss": 0.3629,
      "step": 4327
    },
    {
      "epoch": 2.7690339091490723,
      "grad_norm": 1.7847998983747488,
      "learning_rate": 1.6969618847414183e-05,
      "loss": 0.3727,
      "step": 4328
    },
    {
      "epoch": 2.769673704414587,
      "grad_norm": 1.6730852868066424,
      "learning_rate": 1.6968132739237212e-05,
      "loss": 0.4193,
      "step": 4329
    },
    {
      "epoch": 2.7703134996801024,
      "grad_norm": 1.7475706904785167,
      "learning_rate": 1.6966646331861014e-05,
      "loss": 0.3354,
      "step": 4330
    },
    {
      "epoch": 2.7709532949456173,
      "grad_norm": 1.697060155139373,
      "learning_rate": 1.6965159625349418e-05,
      "loss": 0.3938,
      "step": 4331
    },
    {
      "epoch": 2.7715930902111325,
      "grad_norm": 1.5808038274845817,
      "learning_rate": 1.6963672619766254e-05,
      "loss": 0.3566,
      "step": 4332
    },
    {
      "epoch": 2.7722328854766474,
      "grad_norm": 1.76655098922167,
      "learning_rate": 1.6962185315175377e-05,
      "loss": 0.3707,
      "step": 4333
    },
    {
      "epoch": 2.7728726807421626,
      "grad_norm": 1.781114157436305,
      "learning_rate": 1.6960697711640647e-05,
      "loss": 0.3388,
      "step": 4334
    },
    {
      "epoch": 2.7735124760076775,
      "grad_norm": 1.7508553375739693,
      "learning_rate": 1.695920980922594e-05,
      "loss": 0.354,
      "step": 4335
    },
    {
      "epoch": 2.7741522712731923,
      "grad_norm": 1.6603586396486372,
      "learning_rate": 1.695772160799514e-05,
      "loss": 0.3682,
      "step": 4336
    },
    {
      "epoch": 2.7747920665387076,
      "grad_norm": 1.685686024941204,
      "learning_rate": 1.6956233108012156e-05,
      "loss": 0.3777,
      "step": 4337
    },
    {
      "epoch": 2.775431861804223,
      "grad_norm": 1.5598364035052075,
      "learning_rate": 1.6954744309340894e-05,
      "loss": 0.3154,
      "step": 4338
    },
    {
      "epoch": 2.7760716570697377,
      "grad_norm": 1.7720073095409334,
      "learning_rate": 1.6953255212045282e-05,
      "loss": 0.3487,
      "step": 4339
    },
    {
      "epoch": 2.7767114523352525,
      "grad_norm": 1.8224439386508307,
      "learning_rate": 1.6951765816189262e-05,
      "loss": 0.3447,
      "step": 4340
    },
    {
      "epoch": 2.777351247600768,
      "grad_norm": 1.9000520045579772,
      "learning_rate": 1.6950276121836786e-05,
      "loss": 0.4022,
      "step": 4341
    },
    {
      "epoch": 2.777991042866283,
      "grad_norm": 1.8915282342124704,
      "learning_rate": 1.694878612905181e-05,
      "loss": 0.4106,
      "step": 4342
    },
    {
      "epoch": 2.778630838131798,
      "grad_norm": 1.783898017224617,
      "learning_rate": 1.6947295837898323e-05,
      "loss": 0.3905,
      "step": 4343
    },
    {
      "epoch": 2.7792706333973127,
      "grad_norm": 1.828993686758776,
      "learning_rate": 1.6945805248440313e-05,
      "loss": 0.4355,
      "step": 4344
    },
    {
      "epoch": 2.779910428662828,
      "grad_norm": 1.8466846748349277,
      "learning_rate": 1.694431436074178e-05,
      "loss": 0.3841,
      "step": 4345
    },
    {
      "epoch": 2.780550223928343,
      "grad_norm": 1.8600994230576386,
      "learning_rate": 1.694282317486674e-05,
      "loss": 0.3675,
      "step": 4346
    },
    {
      "epoch": 2.781190019193858,
      "grad_norm": 1.7049429407452843,
      "learning_rate": 1.6941331690879223e-05,
      "loss": 0.3675,
      "step": 4347
    },
    {
      "epoch": 2.781829814459373,
      "grad_norm": 1.6791903514963886,
      "learning_rate": 1.693983990884327e-05,
      "loss": 0.4203,
      "step": 4348
    },
    {
      "epoch": 2.782469609724888,
      "grad_norm": 1.7384241948625776,
      "learning_rate": 1.693834782882294e-05,
      "loss": 0.3864,
      "step": 4349
    },
    {
      "epoch": 2.783109404990403,
      "grad_norm": 1.713129926668253,
      "learning_rate": 1.6936855450882293e-05,
      "loss": 0.3663,
      "step": 4350
    },
    {
      "epoch": 2.783749200255918,
      "grad_norm": 1.8583459864772212,
      "learning_rate": 1.6935362775085416e-05,
      "loss": 0.3581,
      "step": 4351
    },
    {
      "epoch": 2.784388995521433,
      "grad_norm": 1.8839110942831805,
      "learning_rate": 1.6933869801496398e-05,
      "loss": 0.3729,
      "step": 4352
    },
    {
      "epoch": 2.7850287907869484,
      "grad_norm": 1.5583452320802913,
      "learning_rate": 1.693237653017934e-05,
      "loss": 0.359,
      "step": 4353
    },
    {
      "epoch": 2.7856685860524633,
      "grad_norm": 1.815159109296719,
      "learning_rate": 1.693088296119837e-05,
      "loss": 0.3425,
      "step": 4354
    },
    {
      "epoch": 2.786308381317978,
      "grad_norm": 1.7210653579572857,
      "learning_rate": 1.6929389094617613e-05,
      "loss": 0.3497,
      "step": 4355
    },
    {
      "epoch": 2.7869481765834934,
      "grad_norm": 1.9043483717247318,
      "learning_rate": 1.6927894930501215e-05,
      "loss": 0.4517,
      "step": 4356
    },
    {
      "epoch": 2.787587971849008,
      "grad_norm": 1.6418030288996563,
      "learning_rate": 1.6926400468913334e-05,
      "loss": 0.3726,
      "step": 4357
    },
    {
      "epoch": 2.7882277671145235,
      "grad_norm": 1.678493861712606,
      "learning_rate": 1.692490570991814e-05,
      "loss": 0.3522,
      "step": 4358
    },
    {
      "epoch": 2.7888675623800383,
      "grad_norm": 1.7880625688412002,
      "learning_rate": 1.692341065357981e-05,
      "loss": 0.3898,
      "step": 4359
    },
    {
      "epoch": 2.7895073576455536,
      "grad_norm": 1.6262236940644956,
      "learning_rate": 1.6921915299962538e-05,
      "loss": 0.359,
      "step": 4360
    },
    {
      "epoch": 2.7901471529110684,
      "grad_norm": 1.92811385066931,
      "learning_rate": 1.6920419649130543e-05,
      "loss": 0.4092,
      "step": 4361
    },
    {
      "epoch": 2.7907869481765832,
      "grad_norm": 1.7833816130313427,
      "learning_rate": 1.6918923701148037e-05,
      "loss": 0.3732,
      "step": 4362
    },
    {
      "epoch": 2.7914267434420985,
      "grad_norm": 1.7904604916632985,
      "learning_rate": 1.6917427456079255e-05,
      "loss": 0.3535,
      "step": 4363
    },
    {
      "epoch": 2.792066538707614,
      "grad_norm": 1.8116274785381017,
      "learning_rate": 1.6915930913988442e-05,
      "loss": 0.469,
      "step": 4364
    },
    {
      "epoch": 2.7927063339731286,
      "grad_norm": 1.7765017847288482,
      "learning_rate": 1.691443407493986e-05,
      "loss": 0.372,
      "step": 4365
    },
    {
      "epoch": 2.7933461292386434,
      "grad_norm": 1.7647523512597947,
      "learning_rate": 1.6912936938997778e-05,
      "loss": 0.3721,
      "step": 4366
    },
    {
      "epoch": 2.7939859245041587,
      "grad_norm": 1.7253590702924018,
      "learning_rate": 1.691143950622648e-05,
      "loss": 0.3588,
      "step": 4367
    },
    {
      "epoch": 2.7946257197696736,
      "grad_norm": 1.6815406661607648,
      "learning_rate": 1.6909941776690264e-05,
      "loss": 0.4246,
      "step": 4368
    },
    {
      "epoch": 2.795265515035189,
      "grad_norm": 1.7671289757135158,
      "learning_rate": 1.690844375045344e-05,
      "loss": 0.3765,
      "step": 4369
    },
    {
      "epoch": 2.7959053103007037,
      "grad_norm": 1.8078237544343028,
      "learning_rate": 1.690694542758033e-05,
      "loss": 0.3616,
      "step": 4370
    },
    {
      "epoch": 2.796545105566219,
      "grad_norm": 1.8121334489583938,
      "learning_rate": 1.690544680813527e-05,
      "loss": 0.3652,
      "step": 4371
    },
    {
      "epoch": 2.7971849008317338,
      "grad_norm": 1.6040312031835278,
      "learning_rate": 1.690394789218261e-05,
      "loss": 0.3396,
      "step": 4372
    },
    {
      "epoch": 2.797824696097249,
      "grad_norm": 1.742311971361102,
      "learning_rate": 1.690244867978671e-05,
      "loss": 0.3427,
      "step": 4373
    },
    {
      "epoch": 2.798464491362764,
      "grad_norm": 1.8334651353110059,
      "learning_rate": 1.6900949171011937e-05,
      "loss": 0.3836,
      "step": 4374
    },
    {
      "epoch": 2.799104286628279,
      "grad_norm": 1.6686541559492665,
      "learning_rate": 1.6899449365922685e-05,
      "loss": 0.3285,
      "step": 4375
    },
    {
      "epoch": 2.799744081893794,
      "grad_norm": 1.719555601486259,
      "learning_rate": 1.6897949264583352e-05,
      "loss": 0.3792,
      "step": 4376
    },
    {
      "epoch": 2.800383877159309,
      "grad_norm": 1.7397851057922056,
      "learning_rate": 1.6896448867058345e-05,
      "loss": 0.4053,
      "step": 4377
    },
    {
      "epoch": 2.801023672424824,
      "grad_norm": 1.6680008066191618,
      "learning_rate": 1.6894948173412094e-05,
      "loss": 0.3521,
      "step": 4378
    },
    {
      "epoch": 2.8016634676903394,
      "grad_norm": 1.6569821623630485,
      "learning_rate": 1.6893447183709034e-05,
      "loss": 0.3471,
      "step": 4379
    },
    {
      "epoch": 2.802303262955854,
      "grad_norm": 1.7155189977439433,
      "learning_rate": 1.689194589801361e-05,
      "loss": 0.3814,
      "step": 4380
    },
    {
      "epoch": 2.802943058221369,
      "grad_norm": 1.741040600656813,
      "learning_rate": 1.6890444316390296e-05,
      "loss": 0.3777,
      "step": 4381
    },
    {
      "epoch": 2.8035828534868843,
      "grad_norm": 1.6835268046587337,
      "learning_rate": 1.6888942438903554e-05,
      "loss": 0.3614,
      "step": 4382
    },
    {
      "epoch": 2.804222648752399,
      "grad_norm": 1.6075618751059089,
      "learning_rate": 1.6887440265617884e-05,
      "loss": 0.3471,
      "step": 4383
    },
    {
      "epoch": 2.8048624440179144,
      "grad_norm": 1.7860455698080082,
      "learning_rate": 1.6885937796597775e-05,
      "loss": 0.3508,
      "step": 4384
    },
    {
      "epoch": 2.8055022392834292,
      "grad_norm": 1.818086990037608,
      "learning_rate": 1.6884435031907748e-05,
      "loss": 0.4082,
      "step": 4385
    },
    {
      "epoch": 2.8061420345489445,
      "grad_norm": 1.716019845112065,
      "learning_rate": 1.6882931971612326e-05,
      "loss": 0.3646,
      "step": 4386
    },
    {
      "epoch": 2.8067818298144593,
      "grad_norm": 1.6919508016002005,
      "learning_rate": 1.688142861577605e-05,
      "loss": 0.3875,
      "step": 4387
    },
    {
      "epoch": 2.807421625079974,
      "grad_norm": 1.745313271386049,
      "learning_rate": 1.6879924964463474e-05,
      "loss": 0.3408,
      "step": 4388
    },
    {
      "epoch": 2.8080614203454894,
      "grad_norm": 1.7450386129978737,
      "learning_rate": 1.6878421017739152e-05,
      "loss": 0.391,
      "step": 4389
    },
    {
      "epoch": 2.8087012156110047,
      "grad_norm": 1.8266779483508473,
      "learning_rate": 1.6876916775667672e-05,
      "loss": 0.3953,
      "step": 4390
    },
    {
      "epoch": 2.8093410108765196,
      "grad_norm": 1.6269361146747654,
      "learning_rate": 1.6875412238313615e-05,
      "loss": 0.3794,
      "step": 4391
    },
    {
      "epoch": 2.8099808061420344,
      "grad_norm": 1.7121828176471465,
      "learning_rate": 1.6873907405741593e-05,
      "loss": 0.3767,
      "step": 4392
    },
    {
      "epoch": 2.8106206014075497,
      "grad_norm": 1.8531402180658176,
      "learning_rate": 1.687240227801621e-05,
      "loss": 0.4153,
      "step": 4393
    },
    {
      "epoch": 2.8112603966730645,
      "grad_norm": 1.717159882148825,
      "learning_rate": 1.68708968552021e-05,
      "loss": 0.3606,
      "step": 4394
    },
    {
      "epoch": 2.8119001919385798,
      "grad_norm": 1.6411582060382663,
      "learning_rate": 1.68693911373639e-05,
      "loss": 0.3729,
      "step": 4395
    },
    {
      "epoch": 2.8125399872040946,
      "grad_norm": 1.7610993114027138,
      "learning_rate": 1.6867885124566265e-05,
      "loss": 0.3341,
      "step": 4396
    },
    {
      "epoch": 2.81317978246961,
      "grad_norm": 1.6826803746119858,
      "learning_rate": 1.6866378816873856e-05,
      "loss": 0.3691,
      "step": 4397
    },
    {
      "epoch": 2.8138195777351247,
      "grad_norm": 1.677139266667452,
      "learning_rate": 1.6864872214351362e-05,
      "loss": 0.3861,
      "step": 4398
    },
    {
      "epoch": 2.8144593730006395,
      "grad_norm": 1.6702430825827665,
      "learning_rate": 1.686336531706346e-05,
      "loss": 0.3691,
      "step": 4399
    },
    {
      "epoch": 2.815099168266155,
      "grad_norm": 1.7326150912865528,
      "learning_rate": 1.6861858125074864e-05,
      "loss": 0.3868,
      "step": 4400
    },
    {
      "epoch": 2.81573896353167,
      "grad_norm": 1.71643721363992,
      "learning_rate": 1.686035063845029e-05,
      "loss": 0.3759,
      "step": 4401
    },
    {
      "epoch": 2.816378758797185,
      "grad_norm": 1.6855653601341825,
      "learning_rate": 1.6858842857254457e-05,
      "loss": 0.3708,
      "step": 4402
    },
    {
      "epoch": 2.8170185540626997,
      "grad_norm": 1.6287678491250754,
      "learning_rate": 1.6857334781552115e-05,
      "loss": 0.3481,
      "step": 4403
    },
    {
      "epoch": 2.817658349328215,
      "grad_norm": 1.7845226928610847,
      "learning_rate": 1.6855826411408016e-05,
      "loss": 0.3731,
      "step": 4404
    },
    {
      "epoch": 2.81829814459373,
      "grad_norm": 1.6875665261657558,
      "learning_rate": 1.6854317746886923e-05,
      "loss": 0.316,
      "step": 4405
    },
    {
      "epoch": 2.818937939859245,
      "grad_norm": 1.714149151152457,
      "learning_rate": 1.6852808788053624e-05,
      "loss": 0.3872,
      "step": 4406
    },
    {
      "epoch": 2.81957773512476,
      "grad_norm": 1.8538124419590256,
      "learning_rate": 1.6851299534972904e-05,
      "loss": 0.3732,
      "step": 4407
    },
    {
      "epoch": 2.8202175303902752,
      "grad_norm": 1.77014375394254,
      "learning_rate": 1.6849789987709567e-05,
      "loss": 0.3859,
      "step": 4408
    },
    {
      "epoch": 2.82085732565579,
      "grad_norm": 1.7827507130132478,
      "learning_rate": 1.6848280146328436e-05,
      "loss": 0.3334,
      "step": 4409
    },
    {
      "epoch": 2.8214971209213053,
      "grad_norm": 1.6293160569778067,
      "learning_rate": 1.6846770010894335e-05,
      "loss": 0.3553,
      "step": 4410
    },
    {
      "epoch": 2.82213691618682,
      "grad_norm": 1.7137114885577582,
      "learning_rate": 1.6845259581472113e-05,
      "loss": 0.4301,
      "step": 4411
    },
    {
      "epoch": 2.8227767114523354,
      "grad_norm": 1.7942561360929592,
      "learning_rate": 1.684374885812662e-05,
      "loss": 0.3266,
      "step": 4412
    },
    {
      "epoch": 2.8234165067178503,
      "grad_norm": 1.613192040534359,
      "learning_rate": 1.6842237840922723e-05,
      "loss": 0.3279,
      "step": 4413
    },
    {
      "epoch": 2.824056301983365,
      "grad_norm": 1.7400832196111495,
      "learning_rate": 1.6840726529925303e-05,
      "loss": 0.3173,
      "step": 4414
    },
    {
      "epoch": 2.8246960972488804,
      "grad_norm": 1.7958092393441052,
      "learning_rate": 1.683921492519926e-05,
      "loss": 0.3302,
      "step": 4415
    },
    {
      "epoch": 2.8253358925143957,
      "grad_norm": 1.6658741940925397,
      "learning_rate": 1.6837703026809492e-05,
      "loss": 0.3777,
      "step": 4416
    },
    {
      "epoch": 2.8259756877799105,
      "grad_norm": 1.7311995771994244,
      "learning_rate": 1.6836190834820917e-05,
      "loss": 0.345,
      "step": 4417
    },
    {
      "epoch": 2.8266154830454253,
      "grad_norm": 1.7365808606067645,
      "learning_rate": 1.683467834929847e-05,
      "loss": 0.3825,
      "step": 4418
    },
    {
      "epoch": 2.8272552783109406,
      "grad_norm": 1.83186983521721,
      "learning_rate": 1.683316557030709e-05,
      "loss": 0.3667,
      "step": 4419
    },
    {
      "epoch": 2.8278950735764554,
      "grad_norm": 1.735827432744381,
      "learning_rate": 1.6831652497911742e-05,
      "loss": 0.403,
      "step": 4420
    },
    {
      "epoch": 2.8285348688419707,
      "grad_norm": 1.8548480761306028,
      "learning_rate": 1.6830139132177384e-05,
      "loss": 0.3648,
      "step": 4421
    },
    {
      "epoch": 2.8291746641074855,
      "grad_norm": 1.660859195680577,
      "learning_rate": 1.6828625473169e-05,
      "loss": 0.3573,
      "step": 4422
    },
    {
      "epoch": 2.829814459373001,
      "grad_norm": 1.630361647273811,
      "learning_rate": 1.6827111520951592e-05,
      "loss": 0.3237,
      "step": 4423
    },
    {
      "epoch": 2.8304542546385156,
      "grad_norm": 1.5825952024193681,
      "learning_rate": 1.6825597275590154e-05,
      "loss": 0.3936,
      "step": 4424
    },
    {
      "epoch": 2.8310940499040305,
      "grad_norm": 1.6576405079743664,
      "learning_rate": 1.682408273714971e-05,
      "loss": 0.3243,
      "step": 4425
    },
    {
      "epoch": 2.8317338451695457,
      "grad_norm": 1.66726008988646,
      "learning_rate": 1.6822567905695295e-05,
      "loss": 0.3354,
      "step": 4426
    },
    {
      "epoch": 2.832373640435061,
      "grad_norm": 1.614153534741518,
      "learning_rate": 1.6821052781291948e-05,
      "loss": 0.3567,
      "step": 4427
    },
    {
      "epoch": 2.833013435700576,
      "grad_norm": 1.7680854869621698,
      "learning_rate": 1.6819537364004733e-05,
      "loss": 0.4216,
      "step": 4428
    },
    {
      "epoch": 2.8336532309660907,
      "grad_norm": 1.7759777685045417,
      "learning_rate": 1.681802165389871e-05,
      "loss": 0.4289,
      "step": 4429
    },
    {
      "epoch": 2.834293026231606,
      "grad_norm": 1.7565202608487507,
      "learning_rate": 1.6816505651038967e-05,
      "loss": 0.3617,
      "step": 4430
    },
    {
      "epoch": 2.834932821497121,
      "grad_norm": 1.7721530716293474,
      "learning_rate": 1.68149893554906e-05,
      "loss": 0.3857,
      "step": 4431
    },
    {
      "epoch": 2.835572616762636,
      "grad_norm": 1.6322082001944103,
      "learning_rate": 1.6813472767318708e-05,
      "loss": 0.4127,
      "step": 4432
    },
    {
      "epoch": 2.836212412028151,
      "grad_norm": 1.6398043013799826,
      "learning_rate": 1.681195588658842e-05,
      "loss": 0.3621,
      "step": 4433
    },
    {
      "epoch": 2.836852207293666,
      "grad_norm": 1.6690305160984527,
      "learning_rate": 1.681043871336486e-05,
      "loss": 0.3944,
      "step": 4434
    },
    {
      "epoch": 2.837492002559181,
      "grad_norm": 1.751506420898826,
      "learning_rate": 1.6808921247713178e-05,
      "loss": 0.3429,
      "step": 4435
    },
    {
      "epoch": 2.838131797824696,
      "grad_norm": 1.7746366911846598,
      "learning_rate": 1.680740348969853e-05,
      "loss": 0.349,
      "step": 4436
    },
    {
      "epoch": 2.838771593090211,
      "grad_norm": 1.5551913398804966,
      "learning_rate": 1.6805885439386083e-05,
      "loss": 0.3608,
      "step": 4437
    },
    {
      "epoch": 2.8394113883557264,
      "grad_norm": 1.6715248669898755,
      "learning_rate": 1.6804367096841027e-05,
      "loss": 0.3237,
      "step": 4438
    },
    {
      "epoch": 2.840051183621241,
      "grad_norm": 1.726811847932644,
      "learning_rate": 1.680284846212855e-05,
      "loss": 0.3267,
      "step": 4439
    },
    {
      "epoch": 2.840690978886756,
      "grad_norm": 1.696375742495331,
      "learning_rate": 1.680132953531386e-05,
      "loss": 0.3406,
      "step": 4440
    },
    {
      "epoch": 2.8413307741522713,
      "grad_norm": 1.678416517231044,
      "learning_rate": 1.679981031646218e-05,
      "loss": 0.3293,
      "step": 4441
    },
    {
      "epoch": 2.8419705694177866,
      "grad_norm": 1.7550224977111522,
      "learning_rate": 1.679829080563874e-05,
      "loss": 0.3662,
      "step": 4442
    },
    {
      "epoch": 2.8426103646833014,
      "grad_norm": 1.6946214795828891,
      "learning_rate": 1.6796771002908788e-05,
      "loss": 0.3148,
      "step": 4443
    },
    {
      "epoch": 2.8432501599488162,
      "grad_norm": 1.7143890640707549,
      "learning_rate": 1.6795250908337575e-05,
      "loss": 0.3582,
      "step": 4444
    },
    {
      "epoch": 2.8438899552143315,
      "grad_norm": 1.6967782476447968,
      "learning_rate": 1.679373052199038e-05,
      "loss": 0.3747,
      "step": 4445
    },
    {
      "epoch": 2.8445297504798464,
      "grad_norm": 1.8052460855077506,
      "learning_rate": 1.6792209843932478e-05,
      "loss": 0.3765,
      "step": 4446
    },
    {
      "epoch": 2.8451695457453616,
      "grad_norm": 1.8860318772164597,
      "learning_rate": 1.679068887422917e-05,
      "loss": 0.3855,
      "step": 4447
    },
    {
      "epoch": 2.8458093410108765,
      "grad_norm": 1.5807758021856642,
      "learning_rate": 1.678916761294576e-05,
      "loss": 0.3555,
      "step": 4448
    },
    {
      "epoch": 2.8464491362763917,
      "grad_norm": 1.8397370015248307,
      "learning_rate": 1.6787646060147573e-05,
      "loss": 0.3698,
      "step": 4449
    },
    {
      "epoch": 2.8470889315419066,
      "grad_norm": 1.802411556773651,
      "learning_rate": 1.6786124215899934e-05,
      "loss": 0.3647,
      "step": 4450
    },
    {
      "epoch": 2.8477287268074214,
      "grad_norm": 1.6814720544970914,
      "learning_rate": 1.6784602080268197e-05,
      "loss": 0.3241,
      "step": 4451
    },
    {
      "epoch": 2.8483685220729367,
      "grad_norm": 1.608239870793842,
      "learning_rate": 1.6783079653317713e-05,
      "loss": 0.3544,
      "step": 4452
    },
    {
      "epoch": 2.849008317338452,
      "grad_norm": 1.6518593218985456,
      "learning_rate": 1.6781556935113855e-05,
      "loss": 0.3452,
      "step": 4453
    },
    {
      "epoch": 2.8496481126039668,
      "grad_norm": 1.6737287379935961,
      "learning_rate": 1.6780033925722005e-05,
      "loss": 0.324,
      "step": 4454
    },
    {
      "epoch": 2.8502879078694816,
      "grad_norm": 2.0221163562383464,
      "learning_rate": 1.6778510625207564e-05,
      "loss": 0.4075,
      "step": 4455
    },
    {
      "epoch": 2.850927703134997,
      "grad_norm": 1.7779843830595665,
      "learning_rate": 1.677698703363593e-05,
      "loss": 0.3805,
      "step": 4456
    },
    {
      "epoch": 2.8515674984005117,
      "grad_norm": 1.6896169860461439,
      "learning_rate": 1.6775463151072525e-05,
      "loss": 0.3585,
      "step": 4457
    },
    {
      "epoch": 2.852207293666027,
      "grad_norm": 1.7073014339932244,
      "learning_rate": 1.677393897758279e-05,
      "loss": 0.3632,
      "step": 4458
    },
    {
      "epoch": 2.852847088931542,
      "grad_norm": 1.6724915109727299,
      "learning_rate": 1.6772414513232164e-05,
      "loss": 0.3365,
      "step": 4459
    },
    {
      "epoch": 2.853486884197057,
      "grad_norm": 1.6090841647968201,
      "learning_rate": 1.6770889758086108e-05,
      "loss": 0.311,
      "step": 4460
    },
    {
      "epoch": 2.854126679462572,
      "grad_norm": 1.6465211225243075,
      "learning_rate": 1.6769364712210087e-05,
      "loss": 0.3415,
      "step": 4461
    },
    {
      "epoch": 2.8547664747280868,
      "grad_norm": 1.8489065229428998,
      "learning_rate": 1.6767839375669587e-05,
      "loss": 0.4148,
      "step": 4462
    },
    {
      "epoch": 2.855406269993602,
      "grad_norm": 1.716187014762682,
      "learning_rate": 1.6766313748530105e-05,
      "loss": 0.3884,
      "step": 4463
    },
    {
      "epoch": 2.8560460652591173,
      "grad_norm": 1.686958694248548,
      "learning_rate": 1.676478783085715e-05,
      "loss": 0.3256,
      "step": 4464
    },
    {
      "epoch": 2.856685860524632,
      "grad_norm": 1.7956243335062212,
      "learning_rate": 1.6763261622716233e-05,
      "loss": 0.3835,
      "step": 4465
    },
    {
      "epoch": 2.857325655790147,
      "grad_norm": 1.7627753751336117,
      "learning_rate": 1.6761735124172894e-05,
      "loss": 0.3507,
      "step": 4466
    },
    {
      "epoch": 2.8579654510556622,
      "grad_norm": 1.6678174238898784,
      "learning_rate": 1.676020833529268e-05,
      "loss": 0.3585,
      "step": 4467
    },
    {
      "epoch": 2.858605246321177,
      "grad_norm": 1.7721212681887917,
      "learning_rate": 1.6758681256141144e-05,
      "loss": 0.3503,
      "step": 4468
    },
    {
      "epoch": 2.8592450415866923,
      "grad_norm": 1.6289947390078605,
      "learning_rate": 1.675715388678386e-05,
      "loss": 0.3045,
      "step": 4469
    },
    {
      "epoch": 2.859884836852207,
      "grad_norm": 1.7077690839280875,
      "learning_rate": 1.6755626227286404e-05,
      "loss": 0.3744,
      "step": 4470
    },
    {
      "epoch": 2.8605246321177225,
      "grad_norm": 1.7852253788877657,
      "learning_rate": 1.6754098277714376e-05,
      "loss": 0.3458,
      "step": 4471
    },
    {
      "epoch": 2.8611644273832373,
      "grad_norm": 1.7662057957427766,
      "learning_rate": 1.6752570038133383e-05,
      "loss": 0.3668,
      "step": 4472
    },
    {
      "epoch": 2.8618042226487526,
      "grad_norm": 1.7939713528083403,
      "learning_rate": 1.6751041508609046e-05,
      "loss": 0.3753,
      "step": 4473
    },
    {
      "epoch": 2.8624440179142674,
      "grad_norm": 1.7178453224567491,
      "learning_rate": 1.6749512689206995e-05,
      "loss": 0.3721,
      "step": 4474
    },
    {
      "epoch": 2.8630838131797827,
      "grad_norm": 1.71176352943434,
      "learning_rate": 1.6747983579992878e-05,
      "loss": 0.3687,
      "step": 4475
    },
    {
      "epoch": 2.8637236084452975,
      "grad_norm": 1.5846379685342928,
      "learning_rate": 1.6746454181032344e-05,
      "loss": 0.385,
      "step": 4476
    },
    {
      "epoch": 2.8643634037108123,
      "grad_norm": 1.755686532371933,
      "learning_rate": 1.6744924492391072e-05,
      "loss": 0.3871,
      "step": 4477
    },
    {
      "epoch": 2.8650031989763276,
      "grad_norm": 1.8007474134352,
      "learning_rate": 1.674339451413474e-05,
      "loss": 0.3882,
      "step": 4478
    },
    {
      "epoch": 2.865642994241843,
      "grad_norm": 1.7980744184591662,
      "learning_rate": 1.6741864246329044e-05,
      "loss": 0.3801,
      "step": 4479
    },
    {
      "epoch": 2.8662827895073577,
      "grad_norm": 1.7091700200698474,
      "learning_rate": 1.674033368903969e-05,
      "loss": 0.3683,
      "step": 4480
    },
    {
      "epoch": 2.8669225847728725,
      "grad_norm": 1.6453596586813295,
      "learning_rate": 1.6738802842332397e-05,
      "loss": 0.3589,
      "step": 4481
    },
    {
      "epoch": 2.867562380038388,
      "grad_norm": 1.631568321405082,
      "learning_rate": 1.6737271706272897e-05,
      "loss": 0.4091,
      "step": 4482
    },
    {
      "epoch": 2.8682021753039026,
      "grad_norm": 1.6277980544018589,
      "learning_rate": 1.6735740280926936e-05,
      "loss": 0.3058,
      "step": 4483
    },
    {
      "epoch": 2.868841970569418,
      "grad_norm": 1.7570891930988948,
      "learning_rate": 1.6734208566360267e-05,
      "loss": 0.3721,
      "step": 4484
    },
    {
      "epoch": 2.8694817658349328,
      "grad_norm": 1.6721876728531058,
      "learning_rate": 1.6732676562638665e-05,
      "loss": 0.3276,
      "step": 4485
    },
    {
      "epoch": 2.870121561100448,
      "grad_norm": 1.7530240844181986,
      "learning_rate": 1.6731144269827902e-05,
      "loss": 0.3287,
      "step": 4486
    },
    {
      "epoch": 2.870761356365963,
      "grad_norm": 1.6707281314923381,
      "learning_rate": 1.672961168799378e-05,
      "loss": 0.4041,
      "step": 4487
    },
    {
      "epoch": 2.8714011516314777,
      "grad_norm": 1.7169151926667119,
      "learning_rate": 1.672807881720211e-05,
      "loss": 0.3188,
      "step": 4488
    },
    {
      "epoch": 2.872040946896993,
      "grad_norm": 1.7227259181511207,
      "learning_rate": 1.6726545657518696e-05,
      "loss": 0.4116,
      "step": 4489
    },
    {
      "epoch": 2.8726807421625082,
      "grad_norm": 1.6877710121303946,
      "learning_rate": 1.672501220900938e-05,
      "loss": 0.334,
      "step": 4490
    },
    {
      "epoch": 2.873320537428023,
      "grad_norm": 1.641608105861725,
      "learning_rate": 1.672347847174e-05,
      "loss": 0.3536,
      "step": 4491
    },
    {
      "epoch": 2.873960332693538,
      "grad_norm": 1.7685752629321292,
      "learning_rate": 1.672194444577642e-05,
      "loss": 0.3751,
      "step": 4492
    },
    {
      "epoch": 2.874600127959053,
      "grad_norm": 1.6275662373392927,
      "learning_rate": 1.6720410131184496e-05,
      "loss": 0.38,
      "step": 4493
    },
    {
      "epoch": 2.875239923224568,
      "grad_norm": 1.722939086728504,
      "learning_rate": 1.671887552803012e-05,
      "loss": 0.3499,
      "step": 4494
    },
    {
      "epoch": 2.8758797184900833,
      "grad_norm": 1.744047724418062,
      "learning_rate": 1.671734063637918e-05,
      "loss": 0.3245,
      "step": 4495
    },
    {
      "epoch": 2.876519513755598,
      "grad_norm": 1.781128054102234,
      "learning_rate": 1.6715805456297584e-05,
      "loss": 0.3854,
      "step": 4496
    },
    {
      "epoch": 2.8771593090211134,
      "grad_norm": 1.609262916959257,
      "learning_rate": 1.6714269987851245e-05,
      "loss": 0.3772,
      "step": 4497
    },
    {
      "epoch": 2.877799104286628,
      "grad_norm": 1.5266197559174182,
      "learning_rate": 1.67127342311061e-05,
      "loss": 0.3094,
      "step": 4498
    },
    {
      "epoch": 2.878438899552143,
      "grad_norm": 1.7177660930650935,
      "learning_rate": 1.6711198186128085e-05,
      "loss": 0.4117,
      "step": 4499
    },
    {
      "epoch": 2.8790786948176583,
      "grad_norm": 1.6061535999939798,
      "learning_rate": 1.6709661852983157e-05,
      "loss": 0.3913,
      "step": 4500
    },
    {
      "epoch": 2.8797184900831736,
      "grad_norm": 1.8876282268948448,
      "learning_rate": 1.670812523173729e-05,
      "loss": 0.3746,
      "step": 4501
    },
    {
      "epoch": 2.8803582853486884,
      "grad_norm": 1.6604889484063232,
      "learning_rate": 1.6706588322456455e-05,
      "loss": 0.3648,
      "step": 4502
    },
    {
      "epoch": 2.8809980806142033,
      "grad_norm": 1.791601554950426,
      "learning_rate": 1.6705051125206648e-05,
      "loss": 0.3889,
      "step": 4503
    },
    {
      "epoch": 2.8816378758797185,
      "grad_norm": 1.7238412631184292,
      "learning_rate": 1.6703513640053873e-05,
      "loss": 0.3606,
      "step": 4504
    },
    {
      "epoch": 2.8822776711452334,
      "grad_norm": 1.7149365167044075,
      "learning_rate": 1.6701975867064147e-05,
      "loss": 0.3534,
      "step": 4505
    },
    {
      "epoch": 2.8829174664107486,
      "grad_norm": 1.7358163933668715,
      "learning_rate": 1.6700437806303498e-05,
      "loss": 0.3292,
      "step": 4506
    },
    {
      "epoch": 2.8835572616762635,
      "grad_norm": 1.856527681449576,
      "learning_rate": 1.669889945783797e-05,
      "loss": 0.437,
      "step": 4507
    },
    {
      "epoch": 2.8841970569417787,
      "grad_norm": 1.656888120708167,
      "learning_rate": 1.6697360821733616e-05,
      "loss": 0.335,
      "step": 4508
    },
    {
      "epoch": 2.8848368522072936,
      "grad_norm": 1.7466804038475252,
      "learning_rate": 1.6695821898056502e-05,
      "loss": 0.3279,
      "step": 4509
    },
    {
      "epoch": 2.885476647472809,
      "grad_norm": 1.6642025261240976,
      "learning_rate": 1.6694282686872706e-05,
      "loss": 0.3281,
      "step": 4510
    },
    {
      "epoch": 2.8861164427383237,
      "grad_norm": 1.8502138138449986,
      "learning_rate": 1.6692743188248317e-05,
      "loss": 0.4356,
      "step": 4511
    },
    {
      "epoch": 2.886756238003839,
      "grad_norm": 1.474840205756029,
      "learning_rate": 1.6691203402249443e-05,
      "loss": 0.3269,
      "step": 4512
    },
    {
      "epoch": 2.887396033269354,
      "grad_norm": 1.7211085453728499,
      "learning_rate": 1.6689663328942202e-05,
      "loss": 0.3856,
      "step": 4513
    },
    {
      "epoch": 2.8880358285348686,
      "grad_norm": 1.6582724434707092,
      "learning_rate": 1.668812296839271e-05,
      "loss": 0.365,
      "step": 4514
    },
    {
      "epoch": 2.888675623800384,
      "grad_norm": 1.6729347351816863,
      "learning_rate": 1.6686582320667122e-05,
      "loss": 0.3847,
      "step": 4515
    },
    {
      "epoch": 2.889315419065899,
      "grad_norm": 1.5825350157544724,
      "learning_rate": 1.668504138583158e-05,
      "loss": 0.3542,
      "step": 4516
    },
    {
      "epoch": 2.889955214331414,
      "grad_norm": 1.6801710602692386,
      "learning_rate": 1.6683500163952253e-05,
      "loss": 0.3909,
      "step": 4517
    },
    {
      "epoch": 2.890595009596929,
      "grad_norm": 1.6647662548959694,
      "learning_rate": 1.6681958655095318e-05,
      "loss": 0.3491,
      "step": 4518
    },
    {
      "epoch": 2.891234804862444,
      "grad_norm": 1.6995297841360681,
      "learning_rate": 1.6680416859326964e-05,
      "loss": 0.3071,
      "step": 4519
    },
    {
      "epoch": 2.891874600127959,
      "grad_norm": 1.7693683187047866,
      "learning_rate": 1.6678874776713393e-05,
      "loss": 0.3392,
      "step": 4520
    },
    {
      "epoch": 2.892514395393474,
      "grad_norm": 1.7413187044143021,
      "learning_rate": 1.6677332407320822e-05,
      "loss": 0.4112,
      "step": 4521
    },
    {
      "epoch": 2.893154190658989,
      "grad_norm": 1.6897655920831736,
      "learning_rate": 1.667578975121548e-05,
      "loss": 0.3129,
      "step": 4522
    },
    {
      "epoch": 2.8937939859245043,
      "grad_norm": 1.9151746694707454,
      "learning_rate": 1.6674246808463593e-05,
      "loss": 0.3904,
      "step": 4523
    },
    {
      "epoch": 2.894433781190019,
      "grad_norm": 1.8299418254522066,
      "learning_rate": 1.667270357913142e-05,
      "loss": 0.326,
      "step": 4524
    },
    {
      "epoch": 2.895073576455534,
      "grad_norm": 1.7198077021001401,
      "learning_rate": 1.667116006328523e-05,
      "loss": 0.349,
      "step": 4525
    },
    {
      "epoch": 2.8957133717210493,
      "grad_norm": 1.7030329115280116,
      "learning_rate": 1.6669616260991296e-05,
      "loss": 0.3236,
      "step": 4526
    },
    {
      "epoch": 2.8963531669865645,
      "grad_norm": 1.631386939355666,
      "learning_rate": 1.66680721723159e-05,
      "loss": 0.3471,
      "step": 4527
    },
    {
      "epoch": 2.8969929622520794,
      "grad_norm": 1.6463122849289238,
      "learning_rate": 1.6666527797325348e-05,
      "loss": 0.3395,
      "step": 4528
    },
    {
      "epoch": 2.897632757517594,
      "grad_norm": 1.7716428914729248,
      "learning_rate": 1.666498313608595e-05,
      "loss": 0.3769,
      "step": 4529
    },
    {
      "epoch": 2.8982725527831095,
      "grad_norm": 1.7439341168853746,
      "learning_rate": 1.666343818866403e-05,
      "loss": 0.3832,
      "step": 4530
    },
    {
      "epoch": 2.8989123480486243,
      "grad_norm": 1.7760412390887068,
      "learning_rate": 1.6661892955125934e-05,
      "loss": 0.3223,
      "step": 4531
    },
    {
      "epoch": 2.8995521433141396,
      "grad_norm": 1.6521955843607048,
      "learning_rate": 1.6660347435538e-05,
      "loss": 0.3955,
      "step": 4532
    },
    {
      "epoch": 2.9001919385796544,
      "grad_norm": 1.6391961664975176,
      "learning_rate": 1.66588016299666e-05,
      "loss": 0.3586,
      "step": 4533
    },
    {
      "epoch": 2.9008317338451697,
      "grad_norm": 1.6289443876147676,
      "learning_rate": 1.6657255538478097e-05,
      "loss": 0.3393,
      "step": 4534
    },
    {
      "epoch": 2.9014715291106845,
      "grad_norm": 1.671904634458458,
      "learning_rate": 1.6655709161138886e-05,
      "loss": 0.4122,
      "step": 4535
    },
    {
      "epoch": 2.9021113243761993,
      "grad_norm": 1.7706758098616742,
      "learning_rate": 1.6654162498015366e-05,
      "loss": 0.3596,
      "step": 4536
    },
    {
      "epoch": 2.9027511196417146,
      "grad_norm": 1.6888066531440094,
      "learning_rate": 1.665261554917394e-05,
      "loss": 0.3304,
      "step": 4537
    },
    {
      "epoch": 2.90339091490723,
      "grad_norm": 1.7193677169415271,
      "learning_rate": 1.6651068314681037e-05,
      "loss": 0.3344,
      "step": 4538
    },
    {
      "epoch": 2.9040307101727447,
      "grad_norm": 1.6527935162565646,
      "learning_rate": 1.6649520794603097e-05,
      "loss": 0.3502,
      "step": 4539
    },
    {
      "epoch": 2.9046705054382596,
      "grad_norm": 1.7615480825984529,
      "learning_rate": 1.664797298900656e-05,
      "loss": 0.3398,
      "step": 4540
    },
    {
      "epoch": 2.905310300703775,
      "grad_norm": 1.666196443236934,
      "learning_rate": 1.6646424897957885e-05,
      "loss": 0.3213,
      "step": 4541
    },
    {
      "epoch": 2.90595009596929,
      "grad_norm": 1.7346512573911042,
      "learning_rate": 1.664487652152355e-05,
      "loss": 0.3325,
      "step": 4542
    },
    {
      "epoch": 2.906589891234805,
      "grad_norm": 1.8047078723772916,
      "learning_rate": 1.664332785977004e-05,
      "loss": 0.357,
      "step": 4543
    },
    {
      "epoch": 2.9072296865003198,
      "grad_norm": 1.650750268084151,
      "learning_rate": 1.664177891276385e-05,
      "loss": 0.3641,
      "step": 4544
    },
    {
      "epoch": 2.907869481765835,
      "grad_norm": 1.6048082521350608,
      "learning_rate": 1.6640229680571484e-05,
      "loss": 0.3641,
      "step": 4545
    },
    {
      "epoch": 2.90850927703135,
      "grad_norm": 1.584206721227444,
      "learning_rate": 1.6638680163259467e-05,
      "loss": 0.3034,
      "step": 4546
    },
    {
      "epoch": 2.909149072296865,
      "grad_norm": 1.6800405082824428,
      "learning_rate": 1.6637130360894338e-05,
      "loss": 0.3353,
      "step": 4547
    },
    {
      "epoch": 2.90978886756238,
      "grad_norm": 1.5997344797776947,
      "learning_rate": 1.6635580273542635e-05,
      "loss": 0.377,
      "step": 4548
    },
    {
      "epoch": 2.9104286628278953,
      "grad_norm": 1.8076118342493395,
      "learning_rate": 1.663402990127092e-05,
      "loss": 0.3773,
      "step": 4549
    },
    {
      "epoch": 2.91106845809341,
      "grad_norm": 1.6482346817820988,
      "learning_rate": 1.6632479244145764e-05,
      "loss": 0.3656,
      "step": 4550
    },
    {
      "epoch": 2.911708253358925,
      "grad_norm": 1.6326770283317884,
      "learning_rate": 1.6630928302233743e-05,
      "loss": 0.3127,
      "step": 4551
    },
    {
      "epoch": 2.91234804862444,
      "grad_norm": 1.6636970688430905,
      "learning_rate": 1.662937707560146e-05,
      "loss": 0.3093,
      "step": 4552
    },
    {
      "epoch": 2.9129878438899555,
      "grad_norm": 1.627508834417717,
      "learning_rate": 1.6627825564315517e-05,
      "loss": 0.3014,
      "step": 4553
    },
    {
      "epoch": 2.9136276391554703,
      "grad_norm": 1.945916311854438,
      "learning_rate": 1.662627376844253e-05,
      "loss": 0.4077,
      "step": 4554
    },
    {
      "epoch": 2.914267434420985,
      "grad_norm": 1.675037105694097,
      "learning_rate": 1.662472168804914e-05,
      "loss": 0.3286,
      "step": 4555
    },
    {
      "epoch": 2.9149072296865004,
      "grad_norm": 1.6679506001710678,
      "learning_rate": 1.662316932320198e-05,
      "loss": 0.339,
      "step": 4556
    },
    {
      "epoch": 2.9155470249520152,
      "grad_norm": 1.6746613742656202,
      "learning_rate": 1.6621616673967718e-05,
      "loss": 0.3491,
      "step": 4557
    },
    {
      "epoch": 2.9161868202175305,
      "grad_norm": 1.5884566669543378,
      "learning_rate": 1.662006374041301e-05,
      "loss": 0.3379,
      "step": 4558
    },
    {
      "epoch": 2.9168266154830453,
      "grad_norm": 1.7476052528823969,
      "learning_rate": 1.6618510522604544e-05,
      "loss": 0.4184,
      "step": 4559
    },
    {
      "epoch": 2.9174664107485606,
      "grad_norm": 1.6687910238041435,
      "learning_rate": 1.661695702060901e-05,
      "loss": 0.3129,
      "step": 4560
    },
    {
      "epoch": 2.9181062060140754,
      "grad_norm": 1.7074202926862792,
      "learning_rate": 1.6615403234493104e-05,
      "loss": 0.3653,
      "step": 4561
    },
    {
      "epoch": 2.9187460012795903,
      "grad_norm": 1.7219930207863838,
      "learning_rate": 1.6613849164323562e-05,
      "loss": 0.3747,
      "step": 4562
    },
    {
      "epoch": 2.9193857965451055,
      "grad_norm": 1.6772126958206024,
      "learning_rate": 1.66122948101671e-05,
      "loss": 0.3336,
      "step": 4563
    },
    {
      "epoch": 2.920025591810621,
      "grad_norm": 1.7827804756307575,
      "learning_rate": 1.6610740172090453e-05,
      "loss": 0.3822,
      "step": 4564
    },
    {
      "epoch": 2.9206653870761357,
      "grad_norm": 1.6001640509639246,
      "learning_rate": 1.660918525016039e-05,
      "loss": 0.3303,
      "step": 4565
    },
    {
      "epoch": 2.9213051823416505,
      "grad_norm": 1.5727941419321678,
      "learning_rate": 1.6607630044443665e-05,
      "loss": 0.298,
      "step": 4566
    },
    {
      "epoch": 2.9219449776071658,
      "grad_norm": 1.6685371500552821,
      "learning_rate": 1.6606074555007065e-05,
      "loss": 0.3413,
      "step": 4567
    },
    {
      "epoch": 2.9225847728726806,
      "grad_norm": 1.8964392483268744,
      "learning_rate": 1.660451878191737e-05,
      "loss": 0.3814,
      "step": 4568
    },
    {
      "epoch": 2.923224568138196,
      "grad_norm": 1.6714451270505455,
      "learning_rate": 1.6602962725241393e-05,
      "loss": 0.3442,
      "step": 4569
    },
    {
      "epoch": 2.9238643634037107,
      "grad_norm": 1.5574321945000544,
      "learning_rate": 1.6601406385045938e-05,
      "loss": 0.3368,
      "step": 4570
    },
    {
      "epoch": 2.924504158669226,
      "grad_norm": 1.6162341045021578,
      "learning_rate": 1.6599849761397836e-05,
      "loss": 0.3049,
      "step": 4571
    },
    {
      "epoch": 2.925143953934741,
      "grad_norm": 1.8484717540924434,
      "learning_rate": 1.6598292854363924e-05,
      "loss": 0.3757,
      "step": 4572
    },
    {
      "epoch": 2.925783749200256,
      "grad_norm": 1.6001415002462047,
      "learning_rate": 1.6596735664011054e-05,
      "loss": 0.3222,
      "step": 4573
    },
    {
      "epoch": 2.926423544465771,
      "grad_norm": 1.5987211583978007,
      "learning_rate": 1.6595178190406095e-05,
      "loss": 0.3423,
      "step": 4574
    },
    {
      "epoch": 2.927063339731286,
      "grad_norm": 1.7114251785363983,
      "learning_rate": 1.659362043361591e-05,
      "loss": 0.3524,
      "step": 4575
    },
    {
      "epoch": 2.927703134996801,
      "grad_norm": 1.8028389080879887,
      "learning_rate": 1.6592062393707395e-05,
      "loss": 0.3862,
      "step": 4576
    },
    {
      "epoch": 2.928342930262316,
      "grad_norm": 1.7336671183061458,
      "learning_rate": 1.659050407074745e-05,
      "loss": 0.3712,
      "step": 4577
    },
    {
      "epoch": 2.928982725527831,
      "grad_norm": 1.8351721085358548,
      "learning_rate": 1.658894546480298e-05,
      "loss": 0.3458,
      "step": 4578
    },
    {
      "epoch": 2.9296225207933464,
      "grad_norm": 1.7231796591387252,
      "learning_rate": 1.6587386575940915e-05,
      "loss": 0.375,
      "step": 4579
    },
    {
      "epoch": 2.9302623160588612,
      "grad_norm": 1.6506447714803103,
      "learning_rate": 1.658582740422819e-05,
      "loss": 0.3202,
      "step": 4580
    },
    {
      "epoch": 2.930902111324376,
      "grad_norm": 1.7665029528251248,
      "learning_rate": 1.6584267949731744e-05,
      "loss": 0.3225,
      "step": 4581
    },
    {
      "epoch": 2.9315419065898913,
      "grad_norm": 1.7942815740533828,
      "learning_rate": 1.6582708212518555e-05,
      "loss": 0.3069,
      "step": 4582
    },
    {
      "epoch": 2.932181701855406,
      "grad_norm": 1.6663326062845352,
      "learning_rate": 1.6581148192655575e-05,
      "loss": 0.3548,
      "step": 4583
    },
    {
      "epoch": 2.9328214971209214,
      "grad_norm": 1.65490370653639,
      "learning_rate": 1.6579587890209805e-05,
      "loss": 0.4019,
      "step": 4584
    },
    {
      "epoch": 2.9334612923864363,
      "grad_norm": 1.6785199291218527,
      "learning_rate": 1.6578027305248235e-05,
      "loss": 0.3969,
      "step": 4585
    },
    {
      "epoch": 2.9341010876519515,
      "grad_norm": 1.5522922598319355,
      "learning_rate": 1.657646643783787e-05,
      "loss": 0.3648,
      "step": 4586
    },
    {
      "epoch": 2.9347408829174664,
      "grad_norm": 1.64649116799236,
      "learning_rate": 1.657490528804574e-05,
      "loss": 0.3565,
      "step": 4587
    },
    {
      "epoch": 2.935380678182981,
      "grad_norm": 1.7198267106271943,
      "learning_rate": 1.657334385593887e-05,
      "loss": 0.3866,
      "step": 4588
    },
    {
      "epoch": 2.9360204734484965,
      "grad_norm": 1.5921274849459863,
      "learning_rate": 1.657178214158431e-05,
      "loss": 0.3341,
      "step": 4589
    },
    {
      "epoch": 2.9366602687140118,
      "grad_norm": 1.69022443540149,
      "learning_rate": 1.657022014504911e-05,
      "loss": 0.3557,
      "step": 4590
    },
    {
      "epoch": 2.9373000639795266,
      "grad_norm": 1.6754336913949737,
      "learning_rate": 1.6568657866400347e-05,
      "loss": 0.3569,
      "step": 4591
    },
    {
      "epoch": 2.9379398592450414,
      "grad_norm": 1.7677914579038356,
      "learning_rate": 1.65670953057051e-05,
      "loss": 0.3578,
      "step": 4592
    },
    {
      "epoch": 2.9385796545105567,
      "grad_norm": 1.8501713500699648,
      "learning_rate": 1.6565532463030466e-05,
      "loss": 0.3308,
      "step": 4593
    },
    {
      "epoch": 2.9392194497760715,
      "grad_norm": 1.80255985925026,
      "learning_rate": 1.6563969338443545e-05,
      "loss": 0.3752,
      "step": 4594
    },
    {
      "epoch": 2.939859245041587,
      "grad_norm": 1.7516666415226694,
      "learning_rate": 1.6562405932011453e-05,
      "loss": 0.338,
      "step": 4595
    },
    {
      "epoch": 2.9404990403071016,
      "grad_norm": 1.5973948161320444,
      "learning_rate": 1.6560842243801328e-05,
      "loss": 0.3432,
      "step": 4596
    },
    {
      "epoch": 2.941138835572617,
      "grad_norm": 1.727023281319214,
      "learning_rate": 1.655927827388031e-05,
      "loss": 0.3754,
      "step": 4597
    },
    {
      "epoch": 2.9417786308381317,
      "grad_norm": 1.7734029909284328,
      "learning_rate": 1.6557714022315544e-05,
      "loss": 0.3441,
      "step": 4598
    },
    {
      "epoch": 2.9424184261036466,
      "grad_norm": 1.753987655745499,
      "learning_rate": 1.655614948917421e-05,
      "loss": 0.3371,
      "step": 4599
    },
    {
      "epoch": 2.943058221369162,
      "grad_norm": 1.638637926976153,
      "learning_rate": 1.6554584674523475e-05,
      "loss": 0.3191,
      "step": 4600
    },
    {
      "epoch": 2.943698016634677,
      "grad_norm": 1.6699042065575953,
      "learning_rate": 1.6553019578430537e-05,
      "loss": 0.3794,
      "step": 4601
    },
    {
      "epoch": 2.944337811900192,
      "grad_norm": 1.6603902769955163,
      "learning_rate": 1.655145420096259e-05,
      "loss": 0.3505,
      "step": 4602
    },
    {
      "epoch": 2.944977607165707,
      "grad_norm": 1.6012491914511369,
      "learning_rate": 1.654988854218686e-05,
      "loss": 0.3336,
      "step": 4603
    },
    {
      "epoch": 2.945617402431222,
      "grad_norm": 1.7323267567977263,
      "learning_rate": 1.6548322602170565e-05,
      "loss": 0.4078,
      "step": 4604
    },
    {
      "epoch": 2.946257197696737,
      "grad_norm": 1.7422688670187794,
      "learning_rate": 1.6546756380980947e-05,
      "loss": 0.3347,
      "step": 4605
    },
    {
      "epoch": 2.946896992962252,
      "grad_norm": 1.6312802252830092,
      "learning_rate": 1.6545189878685253e-05,
      "loss": 0.3899,
      "step": 4606
    },
    {
      "epoch": 2.947536788227767,
      "grad_norm": 1.7230569156710225,
      "learning_rate": 1.6543623095350748e-05,
      "loss": 0.3247,
      "step": 4607
    },
    {
      "epoch": 2.9481765834932823,
      "grad_norm": 1.6701692400884853,
      "learning_rate": 1.654205603104471e-05,
      "loss": 0.3477,
      "step": 4608
    },
    {
      "epoch": 2.948816378758797,
      "grad_norm": 1.711777700837235,
      "learning_rate": 1.6540488685834425e-05,
      "loss": 0.3604,
      "step": 4609
    },
    {
      "epoch": 2.9494561740243124,
      "grad_norm": 1.7321922219984507,
      "learning_rate": 1.6538921059787186e-05,
      "loss": 0.4311,
      "step": 4610
    },
    {
      "epoch": 2.950095969289827,
      "grad_norm": 1.6007978782193009,
      "learning_rate": 1.653735315297031e-05,
      "loss": 0.3243,
      "step": 4611
    },
    {
      "epoch": 2.9507357645553425,
      "grad_norm": 1.5785720319210073,
      "learning_rate": 1.6535784965451123e-05,
      "loss": 0.3237,
      "step": 4612
    },
    {
      "epoch": 2.9513755598208573,
      "grad_norm": 1.67849899018217,
      "learning_rate": 1.6534216497296952e-05,
      "loss": 0.3911,
      "step": 4613
    },
    {
      "epoch": 2.952015355086372,
      "grad_norm": 1.6599455559299894,
      "learning_rate": 1.653264774857515e-05,
      "loss": 0.3249,
      "step": 4614
    },
    {
      "epoch": 2.9526551503518874,
      "grad_norm": 1.803714889410236,
      "learning_rate": 1.653107871935307e-05,
      "loss": 0.3799,
      "step": 4615
    },
    {
      "epoch": 2.9532949456174027,
      "grad_norm": 1.691227484853483,
      "learning_rate": 1.6529509409698093e-05,
      "loss": 0.3356,
      "step": 4616
    },
    {
      "epoch": 2.9539347408829175,
      "grad_norm": 1.853471015481459,
      "learning_rate": 1.6527939819677593e-05,
      "loss": 0.3703,
      "step": 4617
    },
    {
      "epoch": 2.9545745361484324,
      "grad_norm": 1.5800678331071303,
      "learning_rate": 1.6526369949358973e-05,
      "loss": 0.36,
      "step": 4618
    },
    {
      "epoch": 2.9552143314139476,
      "grad_norm": 1.572651868324834,
      "learning_rate": 1.6524799798809636e-05,
      "loss": 0.2874,
      "step": 4619
    },
    {
      "epoch": 2.9558541266794625,
      "grad_norm": 1.6638722722515928,
      "learning_rate": 1.6523229368097e-05,
      "loss": 0.3582,
      "step": 4620
    },
    {
      "epoch": 2.9564939219449777,
      "grad_norm": 1.5586807964019405,
      "learning_rate": 1.65216586572885e-05,
      "loss": 0.3454,
      "step": 4621
    },
    {
      "epoch": 2.9571337172104926,
      "grad_norm": 1.8733531569217399,
      "learning_rate": 1.652008766645158e-05,
      "loss": 0.3696,
      "step": 4622
    },
    {
      "epoch": 2.957773512476008,
      "grad_norm": 1.6903327176894078,
      "learning_rate": 1.651851639565369e-05,
      "loss": 0.3347,
      "step": 4623
    },
    {
      "epoch": 2.9584133077415227,
      "grad_norm": 1.828062772423982,
      "learning_rate": 1.6516944844962305e-05,
      "loss": 0.3693,
      "step": 4624
    },
    {
      "epoch": 2.9590531030070375,
      "grad_norm": 1.8267056451322612,
      "learning_rate": 1.65153730144449e-05,
      "loss": 0.4155,
      "step": 4625
    },
    {
      "epoch": 2.9596928982725528,
      "grad_norm": 1.600716210867755,
      "learning_rate": 1.651380090416897e-05,
      "loss": 0.3175,
      "step": 4626
    },
    {
      "epoch": 2.960332693538068,
      "grad_norm": 1.502786843697447,
      "learning_rate": 1.6512228514202014e-05,
      "loss": 0.3278,
      "step": 4627
    },
    {
      "epoch": 2.960972488803583,
      "grad_norm": 1.629878995698347,
      "learning_rate": 1.6510655844611546e-05,
      "loss": 0.3297,
      "step": 4628
    },
    {
      "epoch": 2.9616122840690977,
      "grad_norm": 1.6355942280971605,
      "learning_rate": 1.6509082895465104e-05,
      "loss": 0.334,
      "step": 4629
    },
    {
      "epoch": 2.962252079334613,
      "grad_norm": 1.6860552838175944,
      "learning_rate": 1.650750966683022e-05,
      "loss": 0.3445,
      "step": 4630
    },
    {
      "epoch": 2.962891874600128,
      "grad_norm": 1.6059347562281918,
      "learning_rate": 1.6505936158774445e-05,
      "loss": 0.2906,
      "step": 4631
    },
    {
      "epoch": 2.963531669865643,
      "grad_norm": 1.6214492650760446,
      "learning_rate": 1.6504362371365346e-05,
      "loss": 0.375,
      "step": 4632
    },
    {
      "epoch": 2.964171465131158,
      "grad_norm": 1.748654994425386,
      "learning_rate": 1.6502788304670495e-05,
      "loss": 0.3684,
      "step": 4633
    },
    {
      "epoch": 2.964811260396673,
      "grad_norm": 1.76469058126288,
      "learning_rate": 1.6501213958757483e-05,
      "loss": 0.3676,
      "step": 4634
    },
    {
      "epoch": 2.965451055662188,
      "grad_norm": 1.6178734467360993,
      "learning_rate": 1.649963933369391e-05,
      "loss": 0.3421,
      "step": 4635
    },
    {
      "epoch": 2.966090850927703,
      "grad_norm": 1.678442602553293,
      "learning_rate": 1.649806442954738e-05,
      "loss": 0.3182,
      "step": 4636
    },
    {
      "epoch": 2.966730646193218,
      "grad_norm": 1.6931421577061079,
      "learning_rate": 1.6496489246385534e-05,
      "loss": 0.3385,
      "step": 4637
    },
    {
      "epoch": 2.9673704414587334,
      "grad_norm": 1.621849201706208,
      "learning_rate": 1.6494913784275984e-05,
      "loss": 0.3254,
      "step": 4638
    },
    {
      "epoch": 2.9680102367242482,
      "grad_norm": 1.7263936585696982,
      "learning_rate": 1.6493338043286396e-05,
      "loss": 0.3465,
      "step": 4639
    },
    {
      "epoch": 2.968650031989763,
      "grad_norm": 1.7065965236192755,
      "learning_rate": 1.6491762023484422e-05,
      "loss": 0.3794,
      "step": 4640
    },
    {
      "epoch": 2.9692898272552783,
      "grad_norm": 1.6412759108469113,
      "learning_rate": 1.6490185724937738e-05,
      "loss": 0.3369,
      "step": 4641
    },
    {
      "epoch": 2.9699296225207936,
      "grad_norm": 1.68245021979716,
      "learning_rate": 1.648860914771402e-05,
      "loss": 0.3946,
      "step": 4642
    },
    {
      "epoch": 2.9705694177863085,
      "grad_norm": 1.4943065965833733,
      "learning_rate": 1.6487032291880968e-05,
      "loss": 0.2944,
      "step": 4643
    },
    {
      "epoch": 2.9712092130518233,
      "grad_norm": 1.6902933116132122,
      "learning_rate": 1.648545515750629e-05,
      "loss": 0.3049,
      "step": 4644
    },
    {
      "epoch": 2.9718490083173386,
      "grad_norm": 1.601708957213823,
      "learning_rate": 1.6483877744657703e-05,
      "loss": 0.3528,
      "step": 4645
    },
    {
      "epoch": 2.9724888035828534,
      "grad_norm": 1.6552351681364286,
      "learning_rate": 1.648230005340294e-05,
      "loss": 0.337,
      "step": 4646
    },
    {
      "epoch": 2.9731285988483687,
      "grad_norm": 1.5986561236249837,
      "learning_rate": 1.6480722083809743e-05,
      "loss": 0.3321,
      "step": 4647
    },
    {
      "epoch": 2.9737683941138835,
      "grad_norm": 1.6939129635842065,
      "learning_rate": 1.647914383594587e-05,
      "loss": 0.3435,
      "step": 4648
    },
    {
      "epoch": 2.9744081893793988,
      "grad_norm": 1.7662155060310665,
      "learning_rate": 1.6477565309879086e-05,
      "loss": 0.3337,
      "step": 4649
    },
    {
      "epoch": 2.9750479846449136,
      "grad_norm": 1.598305296726894,
      "learning_rate": 1.647598650567717e-05,
      "loss": 0.3138,
      "step": 4650
    },
    {
      "epoch": 2.9756877799104284,
      "grad_norm": 1.5557234939652143,
      "learning_rate": 1.6474407423407913e-05,
      "loss": 0.3267,
      "step": 4651
    },
    {
      "epoch": 2.9763275751759437,
      "grad_norm": 1.7724128031409858,
      "learning_rate": 1.6472828063139116e-05,
      "loss": 0.3868,
      "step": 4652
    },
    {
      "epoch": 2.976967370441459,
      "grad_norm": 1.6655086454967925,
      "learning_rate": 1.64712484249386e-05,
      "loss": 0.372,
      "step": 4653
    },
    {
      "epoch": 2.977607165706974,
      "grad_norm": 1.7055292656719925,
      "learning_rate": 1.6469668508874188e-05,
      "loss": 0.3857,
      "step": 4654
    },
    {
      "epoch": 2.9782469609724886,
      "grad_norm": 1.7080303078309989,
      "learning_rate": 1.6468088315013714e-05,
      "loss": 0.3667,
      "step": 4655
    },
    {
      "epoch": 2.978886756238004,
      "grad_norm": 1.7505691039549562,
      "learning_rate": 1.6466507843425036e-05,
      "loss": 0.3508,
      "step": 4656
    },
    {
      "epoch": 2.9795265515035187,
      "grad_norm": 1.6683502370498362,
      "learning_rate": 1.6464927094176016e-05,
      "loss": 0.3932,
      "step": 4657
    },
    {
      "epoch": 2.980166346769034,
      "grad_norm": 1.7340612667738373,
      "learning_rate": 1.6463346067334526e-05,
      "loss": 0.3427,
      "step": 4658
    },
    {
      "epoch": 2.980806142034549,
      "grad_norm": 1.6351563432023357,
      "learning_rate": 1.6461764762968455e-05,
      "loss": 0.322,
      "step": 4659
    },
    {
      "epoch": 2.981445937300064,
      "grad_norm": 1.6356092507391045,
      "learning_rate": 1.6460183181145696e-05,
      "loss": 0.3241,
      "step": 4660
    },
    {
      "epoch": 2.982085732565579,
      "grad_norm": 1.6255267833779574,
      "learning_rate": 1.645860132193417e-05,
      "loss": 0.3489,
      "step": 4661
    },
    {
      "epoch": 2.982725527831094,
      "grad_norm": 1.8103766112214095,
      "learning_rate": 1.6457019185401784e-05,
      "loss": 0.3138,
      "step": 4662
    },
    {
      "epoch": 2.983365323096609,
      "grad_norm": 1.7593223263610602,
      "learning_rate": 1.645543677161648e-05,
      "loss": 0.3865,
      "step": 4663
    },
    {
      "epoch": 2.9840051183621243,
      "grad_norm": 1.710909472982877,
      "learning_rate": 1.6453854080646213e-05,
      "loss": 0.348,
      "step": 4664
    },
    {
      "epoch": 2.984644913627639,
      "grad_norm": 1.5709324756716294,
      "learning_rate": 1.6452271112558924e-05,
      "loss": 0.335,
      "step": 4665
    },
    {
      "epoch": 2.985284708893154,
      "grad_norm": 1.7052920917711223,
      "learning_rate": 1.6450687867422595e-05,
      "loss": 0.3548,
      "step": 4666
    },
    {
      "epoch": 2.9859245041586693,
      "grad_norm": 1.621621796411796,
      "learning_rate": 1.6449104345305204e-05,
      "loss": 0.3185,
      "step": 4667
    },
    {
      "epoch": 2.986564299424184,
      "grad_norm": 1.5220621781175048,
      "learning_rate": 1.6447520546274742e-05,
      "loss": 0.3221,
      "step": 4668
    },
    {
      "epoch": 2.9872040946896994,
      "grad_norm": 1.672154451497806,
      "learning_rate": 1.644593647039922e-05,
      "loss": 0.3582,
      "step": 4669
    },
    {
      "epoch": 2.987843889955214,
      "grad_norm": 1.681860381565255,
      "learning_rate": 1.644435211774665e-05,
      "loss": 0.3247,
      "step": 4670
    },
    {
      "epoch": 2.9884836852207295,
      "grad_norm": 1.710301013567992,
      "learning_rate": 1.6442767488385066e-05,
      "loss": 0.3442,
      "step": 4671
    },
    {
      "epoch": 2.9891234804862443,
      "grad_norm": 1.7158225089948813,
      "learning_rate": 1.6441182582382506e-05,
      "loss": 0.3479,
      "step": 4672
    },
    {
      "epoch": 2.9897632757517596,
      "grad_norm": 1.6478901889473303,
      "learning_rate": 1.643959739980702e-05,
      "loss": 0.3231,
      "step": 4673
    },
    {
      "epoch": 2.9904030710172744,
      "grad_norm": 1.9074234621288932,
      "learning_rate": 1.643801194072668e-05,
      "loss": 0.3323,
      "step": 4674
    },
    {
      "epoch": 2.9910428662827897,
      "grad_norm": 1.8125322381170534,
      "learning_rate": 1.6436426205209563e-05,
      "loss": 0.3398,
      "step": 4675
    },
    {
      "epoch": 2.9916826615483045,
      "grad_norm": 1.712913396661657,
      "learning_rate": 1.643484019332375e-05,
      "loss": 0.372,
      "step": 4676
    },
    {
      "epoch": 2.9923224568138194,
      "grad_norm": 1.8090063042689708,
      "learning_rate": 1.6433253905137346e-05,
      "loss": 0.3845,
      "step": 4677
    },
    {
      "epoch": 2.9929622520793346,
      "grad_norm": 1.6692968776514459,
      "learning_rate": 1.6431667340718465e-05,
      "loss": 0.3617,
      "step": 4678
    },
    {
      "epoch": 2.99360204734485,
      "grad_norm": 1.7652834524854923,
      "learning_rate": 1.6430080500135228e-05,
      "loss": 0.3608,
      "step": 4679
    },
    {
      "epoch": 2.9942418426103647,
      "grad_norm": 1.7993284759444517,
      "learning_rate": 1.6428493383455774e-05,
      "loss": 0.3816,
      "step": 4680
    },
    {
      "epoch": 2.9948816378758796,
      "grad_norm": 1.6386239983954773,
      "learning_rate": 1.642690599074825e-05,
      "loss": 0.3029,
      "step": 4681
    },
    {
      "epoch": 2.995521433141395,
      "grad_norm": 1.7489689541477667,
      "learning_rate": 1.642531832208081e-05,
      "loss": 0.3313,
      "step": 4682
    },
    {
      "epoch": 2.9961612284069097,
      "grad_norm": 1.644979504748869,
      "learning_rate": 1.6423730377521635e-05,
      "loss": 0.364,
      "step": 4683
    },
    {
      "epoch": 2.996801023672425,
      "grad_norm": 1.56090335902693,
      "learning_rate": 1.6422142157138904e-05,
      "loss": 0.3506,
      "step": 4684
    },
    {
      "epoch": 2.99744081893794,
      "grad_norm": 1.6932313909258006,
      "learning_rate": 1.6420553661000813e-05,
      "loss": 0.3461,
      "step": 4685
    },
    {
      "epoch": 2.998080614203455,
      "grad_norm": 1.617305750562555,
      "learning_rate": 1.6418964889175572e-05,
      "loss": 0.3124,
      "step": 4686
    },
    {
      "epoch": 2.99872040946897,
      "grad_norm": 1.6437818983218706,
      "learning_rate": 1.6417375841731393e-05,
      "loss": 0.3246,
      "step": 4687
    },
    {
      "epoch": 2.9993602047344847,
      "grad_norm": 1.8129448064294993,
      "learning_rate": 1.641578651873651e-05,
      "loss": 0.3358,
      "step": 4688
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5870529375238869,
      "learning_rate": 1.641419692025917e-05,
      "loss": 0.3198,
      "step": 4689
    },
    {
      "epoch": 3.000639795265515,
      "grad_norm": 1.2994255739853695,
      "learning_rate": 1.6412607046367622e-05,
      "loss": 0.2112,
      "step": 4690
    },
    {
      "epoch": 3.00127959053103,
      "grad_norm": 1.474921902638091,
      "learning_rate": 1.6411016897130138e-05,
      "loss": 0.2414,
      "step": 4691
    },
    {
      "epoch": 3.001919385796545,
      "grad_norm": 1.3181886746280274,
      "learning_rate": 1.640942647261499e-05,
      "loss": 0.2253,
      "step": 4692
    },
    {
      "epoch": 3.00255918106206,
      "grad_norm": 1.4917072050010047,
      "learning_rate": 1.640783577289047e-05,
      "loss": 0.2411,
      "step": 4693
    },
    {
      "epoch": 3.003198976327575,
      "grad_norm": 1.4631496516818867,
      "learning_rate": 1.6406244798024884e-05,
      "loss": 0.2476,
      "step": 4694
    },
    {
      "epoch": 3.0038387715930903,
      "grad_norm": 1.5556156632949993,
      "learning_rate": 1.640465354808654e-05,
      "loss": 0.2473,
      "step": 4695
    },
    {
      "epoch": 3.004478566858605,
      "grad_norm": 1.5956488702192162,
      "learning_rate": 1.6403062023143767e-05,
      "loss": 0.2545,
      "step": 4696
    },
    {
      "epoch": 3.0051183621241204,
      "grad_norm": 1.6224869652575036,
      "learning_rate": 1.6401470223264898e-05,
      "loss": 0.2357,
      "step": 4697
    },
    {
      "epoch": 3.0057581573896353,
      "grad_norm": 1.4644434292391537,
      "learning_rate": 1.6399878148518288e-05,
      "loss": 0.221,
      "step": 4698
    },
    {
      "epoch": 3.0063979526551505,
      "grad_norm": 1.4835943627291415,
      "learning_rate": 1.6398285798972292e-05,
      "loss": 0.2153,
      "step": 4699
    },
    {
      "epoch": 3.0070377479206654,
      "grad_norm": 1.5054365507061338,
      "learning_rate": 1.6396693174695286e-05,
      "loss": 0.2382,
      "step": 4700
    },
    {
      "epoch": 3.0076775431861806,
      "grad_norm": 1.5832972736761384,
      "learning_rate": 1.6395100275755657e-05,
      "loss": 0.2224,
      "step": 4701
    },
    {
      "epoch": 3.0083173384516955,
      "grad_norm": 1.45387448442891,
      "learning_rate": 1.6393507102221793e-05,
      "loss": 0.2228,
      "step": 4702
    },
    {
      "epoch": 3.0089571337172103,
      "grad_norm": 1.5316604572835444,
      "learning_rate": 1.639191365416211e-05,
      "loss": 0.2268,
      "step": 4703
    },
    {
      "epoch": 3.0095969289827256,
      "grad_norm": 1.4326010645708447,
      "learning_rate": 1.639031993164503e-05,
      "loss": 0.2185,
      "step": 4704
    },
    {
      "epoch": 3.0102367242482404,
      "grad_norm": 1.337783457213617,
      "learning_rate": 1.6388725934738973e-05,
      "loss": 0.2278,
      "step": 4705
    },
    {
      "epoch": 3.0108765195137557,
      "grad_norm": 1.5054819163837871,
      "learning_rate": 1.6387131663512393e-05,
      "loss": 0.2434,
      "step": 4706
    },
    {
      "epoch": 3.0115163147792705,
      "grad_norm": 1.5303923325583693,
      "learning_rate": 1.6385537118033735e-05,
      "loss": 0.2242,
      "step": 4707
    },
    {
      "epoch": 3.012156110044786,
      "grad_norm": 1.5684534392376017,
      "learning_rate": 1.6383942298371475e-05,
      "loss": 0.2534,
      "step": 4708
    },
    {
      "epoch": 3.0127959053103006,
      "grad_norm": 1.440045803633677,
      "learning_rate": 1.6382347204594092e-05,
      "loss": 0.2162,
      "step": 4709
    },
    {
      "epoch": 3.013435700575816,
      "grad_norm": 1.496987016964713,
      "learning_rate": 1.6380751836770073e-05,
      "loss": 0.2198,
      "step": 4710
    },
    {
      "epoch": 3.0140754958413307,
      "grad_norm": 1.56785188256669,
      "learning_rate": 1.6379156194967916e-05,
      "loss": 0.2603,
      "step": 4711
    },
    {
      "epoch": 3.014715291106846,
      "grad_norm": 1.4124429826996117,
      "learning_rate": 1.6377560279256145e-05,
      "loss": 0.2129,
      "step": 4712
    },
    {
      "epoch": 3.015355086372361,
      "grad_norm": 1.4984044588105248,
      "learning_rate": 1.6375964089703278e-05,
      "loss": 0.2268,
      "step": 4713
    },
    {
      "epoch": 3.0159948816378757,
      "grad_norm": 1.5897045896067838,
      "learning_rate": 1.6374367626377852e-05,
      "loss": 0.2228,
      "step": 4714
    },
    {
      "epoch": 3.016634676903391,
      "grad_norm": 1.4027890395761082,
      "learning_rate": 1.6372770889348424e-05,
      "loss": 0.2299,
      "step": 4715
    },
    {
      "epoch": 3.0172744721689058,
      "grad_norm": 1.4926818638162183,
      "learning_rate": 1.637117387868355e-05,
      "loss": 0.2494,
      "step": 4716
    },
    {
      "epoch": 3.017914267434421,
      "grad_norm": 1.3211895345258797,
      "learning_rate": 1.6369576594451803e-05,
      "loss": 0.1944,
      "step": 4717
    },
    {
      "epoch": 3.018554062699936,
      "grad_norm": 1.5052010867174705,
      "learning_rate": 1.6367979036721764e-05,
      "loss": 0.2225,
      "step": 4718
    },
    {
      "epoch": 3.019193857965451,
      "grad_norm": 1.5082790909506913,
      "learning_rate": 1.6366381205562033e-05,
      "loss": 0.2179,
      "step": 4719
    },
    {
      "epoch": 3.019833653230966,
      "grad_norm": 1.4924426351504103,
      "learning_rate": 1.6364783101041222e-05,
      "loss": 0.2462,
      "step": 4720
    },
    {
      "epoch": 3.0204734484964813,
      "grad_norm": 1.5465333930666263,
      "learning_rate": 1.6363184723227944e-05,
      "loss": 0.2434,
      "step": 4721
    },
    {
      "epoch": 3.021113243761996,
      "grad_norm": 1.2875008747979422,
      "learning_rate": 1.6361586072190834e-05,
      "loss": 0.1939,
      "step": 4722
    },
    {
      "epoch": 3.0217530390275114,
      "grad_norm": 1.4801174636830003,
      "learning_rate": 1.6359987147998533e-05,
      "loss": 0.2597,
      "step": 4723
    },
    {
      "epoch": 3.022392834293026,
      "grad_norm": 1.3791036241719632,
      "learning_rate": 1.63583879507197e-05,
      "loss": 0.2146,
      "step": 4724
    },
    {
      "epoch": 3.0230326295585415,
      "grad_norm": 1.463392730281571,
      "learning_rate": 1.6356788480423e-05,
      "loss": 0.2226,
      "step": 4725
    },
    {
      "epoch": 3.0236724248240563,
      "grad_norm": 1.4105348916162188,
      "learning_rate": 1.6355188737177107e-05,
      "loss": 0.1856,
      "step": 4726
    },
    {
      "epoch": 3.024312220089571,
      "grad_norm": 1.4724591023761997,
      "learning_rate": 1.6353588721050714e-05,
      "loss": 0.2014,
      "step": 4727
    },
    {
      "epoch": 3.0249520153550864,
      "grad_norm": 1.5234773097849892,
      "learning_rate": 1.6351988432112526e-05,
      "loss": 0.2198,
      "step": 4728
    },
    {
      "epoch": 3.0255918106206012,
      "grad_norm": 1.482652360423871,
      "learning_rate": 1.6350387870431256e-05,
      "loss": 0.2255,
      "step": 4729
    },
    {
      "epoch": 3.0262316058861165,
      "grad_norm": 1.566812360379413,
      "learning_rate": 1.634878703607562e-05,
      "loss": 0.2285,
      "step": 4730
    },
    {
      "epoch": 3.0268714011516313,
      "grad_norm": 1.4262164082430502,
      "learning_rate": 1.634718592911437e-05,
      "loss": 0.2054,
      "step": 4731
    },
    {
      "epoch": 3.0275111964171466,
      "grad_norm": 1.4512829326659293,
      "learning_rate": 1.6345584549616248e-05,
      "loss": 0.2297,
      "step": 4732
    },
    {
      "epoch": 3.0281509916826614,
      "grad_norm": 1.480560535580434,
      "learning_rate": 1.6343982897650005e-05,
      "loss": 0.2453,
      "step": 4733
    },
    {
      "epoch": 3.0287907869481767,
      "grad_norm": 1.5499687509950921,
      "learning_rate": 1.6342380973284428e-05,
      "loss": 0.232,
      "step": 4734
    },
    {
      "epoch": 3.0294305822136915,
      "grad_norm": 1.4133894117790258,
      "learning_rate": 1.6340778776588294e-05,
      "loss": 0.2539,
      "step": 4735
    },
    {
      "epoch": 3.030070377479207,
      "grad_norm": 1.3634494047260066,
      "learning_rate": 1.6339176307630395e-05,
      "loss": 0.1965,
      "step": 4736
    },
    {
      "epoch": 3.0307101727447217,
      "grad_norm": 1.3885898803509968,
      "learning_rate": 1.6337573566479542e-05,
      "loss": 0.2053,
      "step": 4737
    },
    {
      "epoch": 3.031349968010237,
      "grad_norm": 1.509738831355315,
      "learning_rate": 1.6335970553204555e-05,
      "loss": 0.2282,
      "step": 4738
    },
    {
      "epoch": 3.0319897632757518,
      "grad_norm": 1.338815102741742,
      "learning_rate": 1.6334367267874265e-05,
      "loss": 0.2083,
      "step": 4739
    },
    {
      "epoch": 3.0326295585412666,
      "grad_norm": 1.4665325683330386,
      "learning_rate": 1.633276371055751e-05,
      "loss": 0.2387,
      "step": 4740
    },
    {
      "epoch": 3.033269353806782,
      "grad_norm": 1.4203321669959974,
      "learning_rate": 1.633115988132315e-05,
      "loss": 0.2079,
      "step": 4741
    },
    {
      "epoch": 3.0339091490722967,
      "grad_norm": 1.531295797668415,
      "learning_rate": 1.6329555780240043e-05,
      "loss": 0.2337,
      "step": 4742
    },
    {
      "epoch": 3.034548944337812,
      "grad_norm": 1.3945908735091612,
      "learning_rate": 1.632795140737707e-05,
      "loss": 0.1801,
      "step": 4743
    },
    {
      "epoch": 3.035188739603327,
      "grad_norm": 1.5693747766485906,
      "learning_rate": 1.6326346762803126e-05,
      "loss": 0.2363,
      "step": 4744
    },
    {
      "epoch": 3.035828534868842,
      "grad_norm": 1.462190602705006,
      "learning_rate": 1.63247418465871e-05,
      "loss": 0.2108,
      "step": 4745
    },
    {
      "epoch": 3.036468330134357,
      "grad_norm": 1.4527269771104367,
      "learning_rate": 1.632313665879791e-05,
      "loss": 0.234,
      "step": 4746
    },
    {
      "epoch": 3.037108125399872,
      "grad_norm": 1.4938791370113567,
      "learning_rate": 1.6321531199504483e-05,
      "loss": 0.2364,
      "step": 4747
    },
    {
      "epoch": 3.037747920665387,
      "grad_norm": 1.42100850300341,
      "learning_rate": 1.631992546877575e-05,
      "loss": 0.2219,
      "step": 4748
    },
    {
      "epoch": 3.0383877159309023,
      "grad_norm": 1.4137911480352632,
      "learning_rate": 1.6318319466680657e-05,
      "loss": 0.2243,
      "step": 4749
    },
    {
      "epoch": 3.039027511196417,
      "grad_norm": 1.4278379071305423,
      "learning_rate": 1.631671319328817e-05,
      "loss": 0.2135,
      "step": 4750
    },
    {
      "epoch": 3.0396673064619324,
      "grad_norm": 1.5258378519129605,
      "learning_rate": 1.6315106648667252e-05,
      "loss": 0.2315,
      "step": 4751
    },
    {
      "epoch": 3.0403071017274472,
      "grad_norm": 1.5400255791053317,
      "learning_rate": 1.631349983288689e-05,
      "loss": 0.2579,
      "step": 4752
    },
    {
      "epoch": 3.040946896992962,
      "grad_norm": 1.4853516769546937,
      "learning_rate": 1.6311892746016076e-05,
      "loss": 0.2415,
      "step": 4753
    },
    {
      "epoch": 3.0415866922584773,
      "grad_norm": 1.4434441261725384,
      "learning_rate": 1.6310285388123813e-05,
      "loss": 0.2232,
      "step": 4754
    },
    {
      "epoch": 3.042226487523992,
      "grad_norm": 1.4187922081852657,
      "learning_rate": 1.6308677759279123e-05,
      "loss": 0.2318,
      "step": 4755
    },
    {
      "epoch": 3.0428662827895074,
      "grad_norm": 1.6063103110754007,
      "learning_rate": 1.6307069859551034e-05,
      "loss": 0.252,
      "step": 4756
    },
    {
      "epoch": 3.0435060780550223,
      "grad_norm": 1.2764789243347696,
      "learning_rate": 1.6305461689008578e-05,
      "loss": 0.195,
      "step": 4757
    },
    {
      "epoch": 3.0441458733205375,
      "grad_norm": 1.353422658022811,
      "learning_rate": 1.630385324772082e-05,
      "loss": 0.2167,
      "step": 4758
    },
    {
      "epoch": 3.0447856685860524,
      "grad_norm": 1.4900838686942568,
      "learning_rate": 1.6302244535756814e-05,
      "loss": 0.2268,
      "step": 4759
    },
    {
      "epoch": 3.0454254638515676,
      "grad_norm": 1.4273277275815448,
      "learning_rate": 1.630063555318564e-05,
      "loss": 0.2152,
      "step": 4760
    },
    {
      "epoch": 3.0460652591170825,
      "grad_norm": 1.318779383296346,
      "learning_rate": 1.6299026300076382e-05,
      "loss": 0.2109,
      "step": 4761
    },
    {
      "epoch": 3.0467050543825978,
      "grad_norm": 1.515958845233502,
      "learning_rate": 1.6297416776498143e-05,
      "loss": 0.2358,
      "step": 4762
    },
    {
      "epoch": 3.0473448496481126,
      "grad_norm": 1.6267990866909798,
      "learning_rate": 1.6295806982520028e-05,
      "loss": 0.2575,
      "step": 4763
    },
    {
      "epoch": 3.047984644913628,
      "grad_norm": 1.5054007781788625,
      "learning_rate": 1.629419691821116e-05,
      "loss": 0.2206,
      "step": 4764
    },
    {
      "epoch": 3.0486244401791427,
      "grad_norm": 1.4634374647541857,
      "learning_rate": 1.6292586583640672e-05,
      "loss": 0.2117,
      "step": 4765
    },
    {
      "epoch": 3.0492642354446575,
      "grad_norm": 1.4317344861110866,
      "learning_rate": 1.629097597887771e-05,
      "loss": 0.2171,
      "step": 4766
    },
    {
      "epoch": 3.049904030710173,
      "grad_norm": 1.573906537364484,
      "learning_rate": 1.628936510399143e-05,
      "loss": 0.2392,
      "step": 4767
    },
    {
      "epoch": 3.0505438259756876,
      "grad_norm": 1.4149202696169518,
      "learning_rate": 1.6287753959051003e-05,
      "loss": 0.2271,
      "step": 4768
    },
    {
      "epoch": 3.051183621241203,
      "grad_norm": 1.444528354099821,
      "learning_rate": 1.62861425441256e-05,
      "loss": 0.2159,
      "step": 4769
    },
    {
      "epoch": 3.0518234165067177,
      "grad_norm": 1.4737126788710309,
      "learning_rate": 1.6284530859284427e-05,
      "loss": 0.2268,
      "step": 4770
    },
    {
      "epoch": 3.052463211772233,
      "grad_norm": 1.478398117826011,
      "learning_rate": 1.6282918904596675e-05,
      "loss": 0.2092,
      "step": 4771
    },
    {
      "epoch": 3.053103007037748,
      "grad_norm": 1.4744514963253856,
      "learning_rate": 1.6281306680131555e-05,
      "loss": 0.237,
      "step": 4772
    },
    {
      "epoch": 3.053742802303263,
      "grad_norm": 1.5507301686384898,
      "learning_rate": 1.6279694185958307e-05,
      "loss": 0.2348,
      "step": 4773
    },
    {
      "epoch": 3.054382597568778,
      "grad_norm": 1.379489414349326,
      "learning_rate": 1.627808142214616e-05,
      "loss": 0.2002,
      "step": 4774
    },
    {
      "epoch": 3.055022392834293,
      "grad_norm": 1.4347752773236357,
      "learning_rate": 1.6276468388764363e-05,
      "loss": 0.235,
      "step": 4775
    },
    {
      "epoch": 3.055662188099808,
      "grad_norm": 1.340283104707622,
      "learning_rate": 1.627485508588218e-05,
      "loss": 0.21,
      "step": 4776
    },
    {
      "epoch": 3.056301983365323,
      "grad_norm": 1.355773058475645,
      "learning_rate": 1.6273241513568884e-05,
      "loss": 0.2069,
      "step": 4777
    },
    {
      "epoch": 3.056941778630838,
      "grad_norm": 1.5167353108980917,
      "learning_rate": 1.6271627671893753e-05,
      "loss": 0.226,
      "step": 4778
    },
    {
      "epoch": 3.057581573896353,
      "grad_norm": 1.4778142770681224,
      "learning_rate": 1.6270013560926086e-05,
      "loss": 0.2319,
      "step": 4779
    },
    {
      "epoch": 3.0582213691618683,
      "grad_norm": 1.6451593126890336,
      "learning_rate": 1.626839918073519e-05,
      "loss": 0.2447,
      "step": 4780
    },
    {
      "epoch": 3.058861164427383,
      "grad_norm": 1.7074343742089721,
      "learning_rate": 1.6266784531390386e-05,
      "loss": 0.2658,
      "step": 4781
    },
    {
      "epoch": 3.0595009596928984,
      "grad_norm": 1.6260417056658913,
      "learning_rate": 1.6265169612961e-05,
      "loss": 0.2667,
      "step": 4782
    },
    {
      "epoch": 3.060140754958413,
      "grad_norm": 1.3715038318579813,
      "learning_rate": 1.6263554425516378e-05,
      "loss": 0.1921,
      "step": 4783
    },
    {
      "epoch": 3.0607805502239285,
      "grad_norm": 1.4339843222456223,
      "learning_rate": 1.6261938969125867e-05,
      "loss": 0.2151,
      "step": 4784
    },
    {
      "epoch": 3.0614203454894433,
      "grad_norm": 1.4812207218309636,
      "learning_rate": 1.626032324385884e-05,
      "loss": 0.2025,
      "step": 4785
    },
    {
      "epoch": 3.0620601407549586,
      "grad_norm": 1.4620267746133204,
      "learning_rate": 1.6258707249784666e-05,
      "loss": 0.2272,
      "step": 4786
    },
    {
      "epoch": 3.0626999360204734,
      "grad_norm": 1.4757571687282467,
      "learning_rate": 1.6257090986972738e-05,
      "loss": 0.2085,
      "step": 4787
    },
    {
      "epoch": 3.0633397312859887,
      "grad_norm": 1.4146895049535988,
      "learning_rate": 1.6255474455492455e-05,
      "loss": 0.2044,
      "step": 4788
    },
    {
      "epoch": 3.0639795265515035,
      "grad_norm": 1.4980480526445856,
      "learning_rate": 1.6253857655413226e-05,
      "loss": 0.2486,
      "step": 4789
    },
    {
      "epoch": 3.0646193218170183,
      "grad_norm": 1.5411648494219012,
      "learning_rate": 1.625224058680447e-05,
      "loss": 0.2245,
      "step": 4790
    },
    {
      "epoch": 3.0652591170825336,
      "grad_norm": 1.54587250284217,
      "learning_rate": 1.625062324973563e-05,
      "loss": 0.2316,
      "step": 4791
    },
    {
      "epoch": 3.0658989123480485,
      "grad_norm": 1.4015398041953813,
      "learning_rate": 1.6249005644276143e-05,
      "loss": 0.2389,
      "step": 4792
    },
    {
      "epoch": 3.0665387076135637,
      "grad_norm": 1.5303845424978946,
      "learning_rate": 1.6247387770495474e-05,
      "loss": 0.2153,
      "step": 4793
    },
    {
      "epoch": 3.0671785028790786,
      "grad_norm": 1.4481892253081994,
      "learning_rate": 1.624576962846308e-05,
      "loss": 0.2355,
      "step": 4794
    },
    {
      "epoch": 3.067818298144594,
      "grad_norm": 1.4550283839830942,
      "learning_rate": 1.6244151218248453e-05,
      "loss": 0.2122,
      "step": 4795
    },
    {
      "epoch": 3.0684580934101087,
      "grad_norm": 1.525852179832917,
      "learning_rate": 1.6242532539921084e-05,
      "loss": 0.2403,
      "step": 4796
    },
    {
      "epoch": 3.069097888675624,
      "grad_norm": 1.5115428850509578,
      "learning_rate": 1.624091359355047e-05,
      "loss": 0.223,
      "step": 4797
    },
    {
      "epoch": 3.0697376839411388,
      "grad_norm": 1.580406455629556,
      "learning_rate": 1.6239294379206123e-05,
      "loss": 0.2143,
      "step": 4798
    },
    {
      "epoch": 3.070377479206654,
      "grad_norm": 1.572482875713178,
      "learning_rate": 1.623767489695758e-05,
      "loss": 0.2346,
      "step": 4799
    },
    {
      "epoch": 3.071017274472169,
      "grad_norm": 1.4733497219380534,
      "learning_rate": 1.6236055146874367e-05,
      "loss": 0.2361,
      "step": 4800
    },
    {
      "epoch": 3.071657069737684,
      "grad_norm": 1.55609555837754,
      "learning_rate": 1.6234435129026046e-05,
      "loss": 0.2198,
      "step": 4801
    },
    {
      "epoch": 3.072296865003199,
      "grad_norm": 1.6352151087826514,
      "learning_rate": 1.6232814843482165e-05,
      "loss": 0.2566,
      "step": 4802
    },
    {
      "epoch": 3.072936660268714,
      "grad_norm": 1.5676544009610978,
      "learning_rate": 1.6231194290312304e-05,
      "loss": 0.2347,
      "step": 4803
    },
    {
      "epoch": 3.073576455534229,
      "grad_norm": 1.520366075654576,
      "learning_rate": 1.6229573469586047e-05,
      "loss": 0.2576,
      "step": 4804
    },
    {
      "epoch": 3.074216250799744,
      "grad_norm": 1.4994514572786788,
      "learning_rate": 1.6227952381372983e-05,
      "loss": 0.2555,
      "step": 4805
    },
    {
      "epoch": 3.074856046065259,
      "grad_norm": 1.4253764798427722,
      "learning_rate": 1.6226331025742723e-05,
      "loss": 0.2127,
      "step": 4806
    },
    {
      "epoch": 3.075495841330774,
      "grad_norm": 1.3487307609297932,
      "learning_rate": 1.6224709402764887e-05,
      "loss": 0.2157,
      "step": 4807
    },
    {
      "epoch": 3.0761356365962893,
      "grad_norm": 1.3924697895260465,
      "learning_rate": 1.62230875125091e-05,
      "loss": 0.2356,
      "step": 4808
    },
    {
      "epoch": 3.076775431861804,
      "grad_norm": 1.4561707427584436,
      "learning_rate": 1.6221465355045008e-05,
      "loss": 0.2245,
      "step": 4809
    },
    {
      "epoch": 3.0774152271273194,
      "grad_norm": 1.4664184843346413,
      "learning_rate": 1.621984293044226e-05,
      "loss": 0.2497,
      "step": 4810
    },
    {
      "epoch": 3.0780550223928342,
      "grad_norm": 1.3960481260474464,
      "learning_rate": 1.621822023877052e-05,
      "loss": 0.2089,
      "step": 4811
    },
    {
      "epoch": 3.0786948176583495,
      "grad_norm": 1.4271792086894366,
      "learning_rate": 1.621659728009946e-05,
      "loss": 0.2154,
      "step": 4812
    },
    {
      "epoch": 3.0793346129238643,
      "grad_norm": 1.4087186313291518,
      "learning_rate": 1.6214974054498776e-05,
      "loss": 0.2281,
      "step": 4813
    },
    {
      "epoch": 3.079974408189379,
      "grad_norm": 1.5483169860777646,
      "learning_rate": 1.621335056203816e-05,
      "loss": 0.2415,
      "step": 4814
    },
    {
      "epoch": 3.0806142034548945,
      "grad_norm": 1.5188176649027618,
      "learning_rate": 1.6211726802787328e-05,
      "loss": 0.2287,
      "step": 4815
    },
    {
      "epoch": 3.0812539987204093,
      "grad_norm": 1.4710621987325436,
      "learning_rate": 1.621010277681599e-05,
      "loss": 0.211,
      "step": 4816
    },
    {
      "epoch": 3.0818937939859246,
      "grad_norm": 1.5208352142503,
      "learning_rate": 1.620847848419389e-05,
      "loss": 0.2471,
      "step": 4817
    },
    {
      "epoch": 3.0825335892514394,
      "grad_norm": 1.4956004035340889,
      "learning_rate": 1.620685392499077e-05,
      "loss": 0.2337,
      "step": 4818
    },
    {
      "epoch": 3.0831733845169547,
      "grad_norm": 1.4751511670843973,
      "learning_rate": 1.620522909927638e-05,
      "loss": 0.2248,
      "step": 4819
    },
    {
      "epoch": 3.0838131797824695,
      "grad_norm": 1.4851239714544968,
      "learning_rate": 1.6203604007120487e-05,
      "loss": 0.2156,
      "step": 4820
    },
    {
      "epoch": 3.0844529750479848,
      "grad_norm": 1.4273508093769405,
      "learning_rate": 1.620197864859288e-05,
      "loss": 0.2272,
      "step": 4821
    },
    {
      "epoch": 3.0850927703134996,
      "grad_norm": 1.4094556647022705,
      "learning_rate": 1.620035302376334e-05,
      "loss": 0.2311,
      "step": 4822
    },
    {
      "epoch": 3.085732565579015,
      "grad_norm": 1.4760529346953244,
      "learning_rate": 1.619872713270167e-05,
      "loss": 0.2222,
      "step": 4823
    },
    {
      "epoch": 3.0863723608445297,
      "grad_norm": 1.4619153175282504,
      "learning_rate": 1.6197100975477685e-05,
      "loss": 0.2389,
      "step": 4824
    },
    {
      "epoch": 3.087012156110045,
      "grad_norm": 1.3279703172901685,
      "learning_rate": 1.619547455216121e-05,
      "loss": 0.1869,
      "step": 4825
    },
    {
      "epoch": 3.08765195137556,
      "grad_norm": 1.508661855054528,
      "learning_rate": 1.619384786282207e-05,
      "loss": 0.2638,
      "step": 4826
    },
    {
      "epoch": 3.0882917466410746,
      "grad_norm": 1.435785276638828,
      "learning_rate": 1.619222090753013e-05,
      "loss": 0.2098,
      "step": 4827
    },
    {
      "epoch": 3.08893154190659,
      "grad_norm": 1.5231677545094673,
      "learning_rate": 1.6190593686355237e-05,
      "loss": 0.2223,
      "step": 4828
    },
    {
      "epoch": 3.0895713371721047,
      "grad_norm": 1.4413436454763675,
      "learning_rate": 1.618896619936726e-05,
      "loss": 0.2041,
      "step": 4829
    },
    {
      "epoch": 3.09021113243762,
      "grad_norm": 1.5601373436374448,
      "learning_rate": 1.6187338446636086e-05,
      "loss": 0.2287,
      "step": 4830
    },
    {
      "epoch": 3.090850927703135,
      "grad_norm": 1.5924186293003073,
      "learning_rate": 1.6185710428231605e-05,
      "loss": 0.2459,
      "step": 4831
    },
    {
      "epoch": 3.09149072296865,
      "grad_norm": 1.3696893066222033,
      "learning_rate": 1.618408214422372e-05,
      "loss": 0.2203,
      "step": 4832
    },
    {
      "epoch": 3.092130518234165,
      "grad_norm": 1.4121135471938753,
      "learning_rate": 1.618245359468235e-05,
      "loss": 0.2038,
      "step": 4833
    },
    {
      "epoch": 3.0927703134996802,
      "grad_norm": 1.6096310534020237,
      "learning_rate": 1.618082477967742e-05,
      "loss": 0.2638,
      "step": 4834
    },
    {
      "epoch": 3.093410108765195,
      "grad_norm": 1.5773337222279282,
      "learning_rate": 1.6179195699278874e-05,
      "loss": 0.2496,
      "step": 4835
    },
    {
      "epoch": 3.0940499040307103,
      "grad_norm": 1.4872169030142248,
      "learning_rate": 1.617756635355665e-05,
      "loss": 0.2117,
      "step": 4836
    },
    {
      "epoch": 3.094689699296225,
      "grad_norm": 1.4296767791716982,
      "learning_rate": 1.617593674258072e-05,
      "loss": 0.2225,
      "step": 4837
    },
    {
      "epoch": 3.0953294945617404,
      "grad_norm": 1.6281063373782318,
      "learning_rate": 1.617430686642105e-05,
      "loss": 0.2545,
      "step": 4838
    },
    {
      "epoch": 3.0959692898272553,
      "grad_norm": 1.4194835542378237,
      "learning_rate": 1.6172676725147626e-05,
      "loss": 0.2228,
      "step": 4839
    },
    {
      "epoch": 3.09660908509277,
      "grad_norm": 1.4082644403069384,
      "learning_rate": 1.6171046318830446e-05,
      "loss": 0.2131,
      "step": 4840
    },
    {
      "epoch": 3.0972488803582854,
      "grad_norm": 1.454885823200044,
      "learning_rate": 1.6169415647539516e-05,
      "loss": 0.1979,
      "step": 4841
    },
    {
      "epoch": 3.0978886756238,
      "grad_norm": 1.529782601170936,
      "learning_rate": 1.6167784711344846e-05,
      "loss": 0.2445,
      "step": 4842
    },
    {
      "epoch": 3.0985284708893155,
      "grad_norm": 1.3864120761705512,
      "learning_rate": 1.616615351031648e-05,
      "loss": 0.2156,
      "step": 4843
    },
    {
      "epoch": 3.0991682661548303,
      "grad_norm": 1.5968742902317636,
      "learning_rate": 1.6164522044524444e-05,
      "loss": 0.2334,
      "step": 4844
    },
    {
      "epoch": 3.0998080614203456,
      "grad_norm": 1.6113784916328586,
      "learning_rate": 1.61628903140388e-05,
      "loss": 0.2729,
      "step": 4845
    },
    {
      "epoch": 3.1004478566858604,
      "grad_norm": 1.427065468016416,
      "learning_rate": 1.6161258318929608e-05,
      "loss": 0.2163,
      "step": 4846
    },
    {
      "epoch": 3.1010876519513757,
      "grad_norm": 1.518456093513846,
      "learning_rate": 1.6159626059266944e-05,
      "loss": 0.2457,
      "step": 4847
    },
    {
      "epoch": 3.1017274472168905,
      "grad_norm": 1.5655981541605666,
      "learning_rate": 1.6157993535120895e-05,
      "loss": 0.2313,
      "step": 4848
    },
    {
      "epoch": 3.102367242482406,
      "grad_norm": 1.390558691629576,
      "learning_rate": 1.615636074656156e-05,
      "loss": 0.2148,
      "step": 4849
    },
    {
      "epoch": 3.1030070377479206,
      "grad_norm": 1.4947994709536616,
      "learning_rate": 1.6154727693659045e-05,
      "loss": 0.243,
      "step": 4850
    },
    {
      "epoch": 3.103646833013436,
      "grad_norm": 1.4775019051501186,
      "learning_rate": 1.6153094376483468e-05,
      "loss": 0.23,
      "step": 4851
    },
    {
      "epoch": 3.1042866282789507,
      "grad_norm": 1.4239532239810253,
      "learning_rate": 1.6151460795104966e-05,
      "loss": 0.2053,
      "step": 4852
    },
    {
      "epoch": 3.1049264235444656,
      "grad_norm": 1.5780379746755469,
      "learning_rate": 1.614982694959368e-05,
      "loss": 0.2502,
      "step": 4853
    },
    {
      "epoch": 3.105566218809981,
      "grad_norm": 1.577542097966343,
      "learning_rate": 1.6148192840019766e-05,
      "loss": 0.2465,
      "step": 4854
    },
    {
      "epoch": 3.1062060140754957,
      "grad_norm": 1.5451262541760205,
      "learning_rate": 1.6146558466453384e-05,
      "loss": 0.2381,
      "step": 4855
    },
    {
      "epoch": 3.106845809341011,
      "grad_norm": 1.5874236584014303,
      "learning_rate": 1.614492382896472e-05,
      "loss": 0.2541,
      "step": 4856
    },
    {
      "epoch": 3.107485604606526,
      "grad_norm": 1.49786562416305,
      "learning_rate": 1.6143288927623954e-05,
      "loss": 0.2325,
      "step": 4857
    },
    {
      "epoch": 3.108125399872041,
      "grad_norm": 1.4877499014386286,
      "learning_rate": 1.614165376250129e-05,
      "loss": 0.2241,
      "step": 4858
    },
    {
      "epoch": 3.108765195137556,
      "grad_norm": 1.4641657177873055,
      "learning_rate": 1.614001833366694e-05,
      "loss": 0.2082,
      "step": 4859
    },
    {
      "epoch": 3.109404990403071,
      "grad_norm": 1.3110884439588595,
      "learning_rate": 1.613838264119113e-05,
      "loss": 0.2003,
      "step": 4860
    },
    {
      "epoch": 3.110044785668586,
      "grad_norm": 1.6068498707360666,
      "learning_rate": 1.613674668514408e-05,
      "loss": 0.248,
      "step": 4861
    },
    {
      "epoch": 3.1106845809341013,
      "grad_norm": 1.5018353679037335,
      "learning_rate": 1.6135110465596045e-05,
      "loss": 0.2386,
      "step": 4862
    },
    {
      "epoch": 3.111324376199616,
      "grad_norm": 1.6574819765633642,
      "learning_rate": 1.6133473982617284e-05,
      "loss": 0.2359,
      "step": 4863
    },
    {
      "epoch": 3.1119641714651314,
      "grad_norm": 1.4701042375084028,
      "learning_rate": 1.613183723627806e-05,
      "loss": 0.2283,
      "step": 4864
    },
    {
      "epoch": 3.112603966730646,
      "grad_norm": 1.5778692673535788,
      "learning_rate": 1.6130200226648653e-05,
      "loss": 0.2568,
      "step": 4865
    },
    {
      "epoch": 3.113243761996161,
      "grad_norm": 1.6234403680060028,
      "learning_rate": 1.6128562953799352e-05,
      "loss": 0.2195,
      "step": 4866
    },
    {
      "epoch": 3.1138835572616763,
      "grad_norm": 1.494313768645523,
      "learning_rate": 1.612692541780046e-05,
      "loss": 0.2079,
      "step": 4867
    },
    {
      "epoch": 3.114523352527191,
      "grad_norm": 1.3055027492857803,
      "learning_rate": 1.6125287618722288e-05,
      "loss": 0.2045,
      "step": 4868
    },
    {
      "epoch": 3.1151631477927064,
      "grad_norm": 1.4895680608579336,
      "learning_rate": 1.6123649556635165e-05,
      "loss": 0.2355,
      "step": 4869
    },
    {
      "epoch": 3.1158029430582213,
      "grad_norm": 1.4038712192815412,
      "learning_rate": 1.612201123160942e-05,
      "loss": 0.2156,
      "step": 4870
    },
    {
      "epoch": 3.1164427383237365,
      "grad_norm": 1.4364245718619035,
      "learning_rate": 1.6120372643715403e-05,
      "loss": 0.2208,
      "step": 4871
    },
    {
      "epoch": 3.1170825335892514,
      "grad_norm": 1.2924931126432375,
      "learning_rate": 1.6118733793023473e-05,
      "loss": 0.2034,
      "step": 4872
    },
    {
      "epoch": 3.1177223288547666,
      "grad_norm": 1.5120177315121195,
      "learning_rate": 1.6117094679604e-05,
      "loss": 0.2349,
      "step": 4873
    },
    {
      "epoch": 3.1183621241202815,
      "grad_norm": 1.4617499423895297,
      "learning_rate": 1.611545530352736e-05,
      "loss": 0.2319,
      "step": 4874
    },
    {
      "epoch": 3.1190019193857967,
      "grad_norm": 1.4909866767613968,
      "learning_rate": 1.611381566486395e-05,
      "loss": 0.2577,
      "step": 4875
    },
    {
      "epoch": 3.1196417146513116,
      "grad_norm": 1.4743441914184972,
      "learning_rate": 1.611217576368417e-05,
      "loss": 0.2365,
      "step": 4876
    },
    {
      "epoch": 3.1202815099168264,
      "grad_norm": 1.5517580296846092,
      "learning_rate": 1.611053560005843e-05,
      "loss": 0.2345,
      "step": 4877
    },
    {
      "epoch": 3.1209213051823417,
      "grad_norm": 1.4859499028740997,
      "learning_rate": 1.610889517405717e-05,
      "loss": 0.2575,
      "step": 4878
    },
    {
      "epoch": 3.1215611004478565,
      "grad_norm": 1.4581761096851498,
      "learning_rate": 1.6107254485750815e-05,
      "loss": 0.2271,
      "step": 4879
    },
    {
      "epoch": 3.122200895713372,
      "grad_norm": 1.5378011149326116,
      "learning_rate": 1.6105613535209815e-05,
      "loss": 0.2619,
      "step": 4880
    },
    {
      "epoch": 3.1228406909788866,
      "grad_norm": 1.3727706365424572,
      "learning_rate": 1.6103972322504634e-05,
      "loss": 0.202,
      "step": 4881
    },
    {
      "epoch": 3.123480486244402,
      "grad_norm": 1.5131873785138785,
      "learning_rate": 1.6102330847705737e-05,
      "loss": 0.2386,
      "step": 4882
    },
    {
      "epoch": 3.1241202815099167,
      "grad_norm": 1.538748412185953,
      "learning_rate": 1.610068911088361e-05,
      "loss": 0.245,
      "step": 4883
    },
    {
      "epoch": 3.124760076775432,
      "grad_norm": 1.5148008503292743,
      "learning_rate": 1.6099047112108742e-05,
      "loss": 0.2313,
      "step": 4884
    },
    {
      "epoch": 3.125399872040947,
      "grad_norm": 1.4643177542662713,
      "learning_rate": 1.6097404851451644e-05,
      "loss": 0.2228,
      "step": 4885
    },
    {
      "epoch": 3.126039667306462,
      "grad_norm": 1.4711978357455142,
      "learning_rate": 1.6095762328982828e-05,
      "loss": 0.2123,
      "step": 4886
    },
    {
      "epoch": 3.126679462571977,
      "grad_norm": 1.4899881866069233,
      "learning_rate": 1.6094119544772815e-05,
      "loss": 0.2556,
      "step": 4887
    },
    {
      "epoch": 3.127319257837492,
      "grad_norm": 1.4309462251671294,
      "learning_rate": 1.6092476498892157e-05,
      "loss": 0.2198,
      "step": 4888
    },
    {
      "epoch": 3.127959053103007,
      "grad_norm": 1.6127807544952182,
      "learning_rate": 1.609083319141139e-05,
      "loss": 0.2548,
      "step": 4889
    },
    {
      "epoch": 3.128598848368522,
      "grad_norm": 1.4806028643879958,
      "learning_rate": 1.6089189622401085e-05,
      "loss": 0.2308,
      "step": 4890
    },
    {
      "epoch": 3.129238643634037,
      "grad_norm": 1.4903449145784895,
      "learning_rate": 1.6087545791931805e-05,
      "loss": 0.2157,
      "step": 4891
    },
    {
      "epoch": 3.129878438899552,
      "grad_norm": 1.625845313305863,
      "learning_rate": 1.608590170007414e-05,
      "loss": 0.2447,
      "step": 4892
    },
    {
      "epoch": 3.1305182341650672,
      "grad_norm": 1.5116243349730567,
      "learning_rate": 1.6084257346898685e-05,
      "loss": 0.2123,
      "step": 4893
    },
    {
      "epoch": 3.131158029430582,
      "grad_norm": 1.3897884369320732,
      "learning_rate": 1.608261273247604e-05,
      "loss": 0.1998,
      "step": 4894
    },
    {
      "epoch": 3.1317978246960974,
      "grad_norm": 1.449325542740252,
      "learning_rate": 1.6080967856876822e-05,
      "loss": 0.2239,
      "step": 4895
    },
    {
      "epoch": 3.132437619961612,
      "grad_norm": 1.4203465309362362,
      "learning_rate": 1.6079322720171665e-05,
      "loss": 0.2135,
      "step": 4896
    },
    {
      "epoch": 3.1330774152271275,
      "grad_norm": 1.616544210987175,
      "learning_rate": 1.6077677322431203e-05,
      "loss": 0.251,
      "step": 4897
    },
    {
      "epoch": 3.1337172104926423,
      "grad_norm": 1.5493690227688381,
      "learning_rate": 1.6076031663726088e-05,
      "loss": 0.2459,
      "step": 4898
    },
    {
      "epoch": 3.1343570057581576,
      "grad_norm": 1.3042901173902468,
      "learning_rate": 1.607438574412698e-05,
      "loss": 0.1918,
      "step": 4899
    },
    {
      "epoch": 3.1349968010236724,
      "grad_norm": 1.3660508372695706,
      "learning_rate": 1.607273956370456e-05,
      "loss": 0.2047,
      "step": 4900
    },
    {
      "epoch": 3.1356365962891877,
      "grad_norm": 1.6566837148830476,
      "learning_rate": 1.6071093122529503e-05,
      "loss": 0.2588,
      "step": 4901
    },
    {
      "epoch": 3.1362763915547025,
      "grad_norm": 1.5107655812677008,
      "learning_rate": 1.60694464206725e-05,
      "loss": 0.2357,
      "step": 4902
    },
    {
      "epoch": 3.1369161868202173,
      "grad_norm": 1.5789246276538604,
      "learning_rate": 1.6067799458204275e-05,
      "loss": 0.2342,
      "step": 4903
    },
    {
      "epoch": 3.1375559820857326,
      "grad_norm": 1.3856269801194103,
      "learning_rate": 1.6066152235195527e-05,
      "loss": 0.1892,
      "step": 4904
    },
    {
      "epoch": 3.1381957773512474,
      "grad_norm": 1.4723024091062555,
      "learning_rate": 1.6064504751716997e-05,
      "loss": 0.2041,
      "step": 4905
    },
    {
      "epoch": 3.1388355726167627,
      "grad_norm": 1.4791039324621822,
      "learning_rate": 1.606285700783942e-05,
      "loss": 0.2216,
      "step": 4906
    },
    {
      "epoch": 3.1394753678822775,
      "grad_norm": 1.4877351510494823,
      "learning_rate": 1.6061209003633544e-05,
      "loss": 0.2305,
      "step": 4907
    },
    {
      "epoch": 3.140115163147793,
      "grad_norm": 1.4286889640972003,
      "learning_rate": 1.605956073917014e-05,
      "loss": 0.2345,
      "step": 4908
    },
    {
      "epoch": 3.1407549584133077,
      "grad_norm": 1.5409124522335702,
      "learning_rate": 1.6057912214519978e-05,
      "loss": 0.2327,
      "step": 4909
    },
    {
      "epoch": 3.141394753678823,
      "grad_norm": 1.4979351578958877,
      "learning_rate": 1.605626342975384e-05,
      "loss": 0.2395,
      "step": 4910
    },
    {
      "epoch": 3.1420345489443378,
      "grad_norm": 1.4697550476933032,
      "learning_rate": 1.605461438494252e-05,
      "loss": 0.2223,
      "step": 4911
    },
    {
      "epoch": 3.142674344209853,
      "grad_norm": 1.4766167154613825,
      "learning_rate": 1.6052965080156836e-05,
      "loss": 0.2332,
      "step": 4912
    },
    {
      "epoch": 3.143314139475368,
      "grad_norm": 1.4135228655063399,
      "learning_rate": 1.6051315515467597e-05,
      "loss": 0.2189,
      "step": 4913
    },
    {
      "epoch": 3.1439539347408827,
      "grad_norm": 1.4246316239359764,
      "learning_rate": 1.604966569094563e-05,
      "loss": 0.211,
      "step": 4914
    },
    {
      "epoch": 3.144593730006398,
      "grad_norm": 1.5571897676339652,
      "learning_rate": 1.6048015606661782e-05,
      "loss": 0.231,
      "step": 4915
    },
    {
      "epoch": 3.145233525271913,
      "grad_norm": 1.3613866054726016,
      "learning_rate": 1.6046365262686903e-05,
      "loss": 0.2084,
      "step": 4916
    },
    {
      "epoch": 3.145873320537428,
      "grad_norm": 1.5325106296389208,
      "learning_rate": 1.6044714659091855e-05,
      "loss": 0.2337,
      "step": 4917
    },
    {
      "epoch": 3.146513115802943,
      "grad_norm": 1.5066223243429089,
      "learning_rate": 1.604306379594751e-05,
      "loss": 0.2419,
      "step": 4918
    },
    {
      "epoch": 3.147152911068458,
      "grad_norm": 1.478363077562671,
      "learning_rate": 1.6041412673324757e-05,
      "loss": 0.2342,
      "step": 4919
    },
    {
      "epoch": 3.147792706333973,
      "grad_norm": 1.4801612446095331,
      "learning_rate": 1.6039761291294493e-05,
      "loss": 0.24,
      "step": 4920
    },
    {
      "epoch": 3.1484325015994883,
      "grad_norm": 1.4235203425984726,
      "learning_rate": 1.603810964992762e-05,
      "loss": 0.2158,
      "step": 4921
    },
    {
      "epoch": 3.149072296865003,
      "grad_norm": 1.4832325273045601,
      "learning_rate": 1.6036457749295063e-05,
      "loss": 0.2305,
      "step": 4922
    },
    {
      "epoch": 3.1497120921305184,
      "grad_norm": 1.5478038948706923,
      "learning_rate": 1.6034805589467743e-05,
      "loss": 0.2208,
      "step": 4923
    },
    {
      "epoch": 3.1503518873960332,
      "grad_norm": 1.5942135143848994,
      "learning_rate": 1.6033153170516608e-05,
      "loss": 0.2574,
      "step": 4924
    },
    {
      "epoch": 3.1509916826615485,
      "grad_norm": 1.4913786648545047,
      "learning_rate": 1.603150049251261e-05,
      "loss": 0.2262,
      "step": 4925
    },
    {
      "epoch": 3.1516314779270633,
      "grad_norm": 1.4956563210312863,
      "learning_rate": 1.602984755552671e-05,
      "loss": 0.2264,
      "step": 4926
    },
    {
      "epoch": 3.1522712731925786,
      "grad_norm": 1.599611105074739,
      "learning_rate": 1.602819435962988e-05,
      "loss": 0.248,
      "step": 4927
    },
    {
      "epoch": 3.1529110684580934,
      "grad_norm": 1.5197170120407144,
      "learning_rate": 1.602654090489311e-05,
      "loss": 0.2483,
      "step": 4928
    },
    {
      "epoch": 3.1535508637236083,
      "grad_norm": 1.5233486124774387,
      "learning_rate": 1.602488719138739e-05,
      "loss": 0.2244,
      "step": 4929
    },
    {
      "epoch": 3.1541906589891235,
      "grad_norm": 1.4404174559669254,
      "learning_rate": 1.6023233219183737e-05,
      "loss": 0.2402,
      "step": 4930
    },
    {
      "epoch": 3.1548304542546384,
      "grad_norm": 1.5424902656281276,
      "learning_rate": 1.602157898835316e-05,
      "loss": 0.2114,
      "step": 4931
    },
    {
      "epoch": 3.1554702495201536,
      "grad_norm": 1.59415752235417,
      "learning_rate": 1.6019924498966694e-05,
      "loss": 0.244,
      "step": 4932
    },
    {
      "epoch": 3.1561100447856685,
      "grad_norm": 1.5301098617379605,
      "learning_rate": 1.6018269751095377e-05,
      "loss": 0.209,
      "step": 4933
    },
    {
      "epoch": 3.1567498400511838,
      "grad_norm": 1.6161954656865676,
      "learning_rate": 1.6016614744810267e-05,
      "loss": 0.2417,
      "step": 4934
    },
    {
      "epoch": 3.1573896353166986,
      "grad_norm": 1.4330586841236788,
      "learning_rate": 1.6014959480182418e-05,
      "loss": 0.2052,
      "step": 4935
    },
    {
      "epoch": 3.158029430582214,
      "grad_norm": 1.4535960936274028,
      "learning_rate": 1.601330395728291e-05,
      "loss": 0.2352,
      "step": 4936
    },
    {
      "epoch": 3.1586692258477287,
      "grad_norm": 1.417256928744187,
      "learning_rate": 1.6011648176182828e-05,
      "loss": 0.2128,
      "step": 4937
    },
    {
      "epoch": 3.159309021113244,
      "grad_norm": 1.4537792401621712,
      "learning_rate": 1.6009992136953266e-05,
      "loss": 0.2254,
      "step": 4938
    },
    {
      "epoch": 3.159948816378759,
      "grad_norm": 1.3992463569416027,
      "learning_rate": 1.6008335839665332e-05,
      "loss": 0.2083,
      "step": 4939
    },
    {
      "epoch": 3.1605886116442736,
      "grad_norm": 1.552549412502794,
      "learning_rate": 1.6006679284390146e-05,
      "loss": 0.2278,
      "step": 4940
    },
    {
      "epoch": 3.161228406909789,
      "grad_norm": 1.4495302748243575,
      "learning_rate": 1.600502247119884e-05,
      "loss": 0.2222,
      "step": 4941
    },
    {
      "epoch": 3.1618682021753037,
      "grad_norm": 1.4967905898001486,
      "learning_rate": 1.6003365400162546e-05,
      "loss": 0.231,
      "step": 4942
    },
    {
      "epoch": 3.162507997440819,
      "grad_norm": 1.4339904666373204,
      "learning_rate": 1.600170807135242e-05,
      "loss": 0.1963,
      "step": 4943
    },
    {
      "epoch": 3.163147792706334,
      "grad_norm": 1.5917636524276704,
      "learning_rate": 1.6000050484839632e-05,
      "loss": 0.2559,
      "step": 4944
    },
    {
      "epoch": 3.163787587971849,
      "grad_norm": 1.509515004518665,
      "learning_rate": 1.5998392640695346e-05,
      "loss": 0.2463,
      "step": 4945
    },
    {
      "epoch": 3.164427383237364,
      "grad_norm": 1.3996185622028152,
      "learning_rate": 1.599673453899075e-05,
      "loss": 0.2054,
      "step": 4946
    },
    {
      "epoch": 3.165067178502879,
      "grad_norm": 1.429577745770418,
      "learning_rate": 1.599507617979704e-05,
      "loss": 0.2199,
      "step": 4947
    },
    {
      "epoch": 3.165706973768394,
      "grad_norm": 1.4693797168798468,
      "learning_rate": 1.5993417563185425e-05,
      "loss": 0.2295,
      "step": 4948
    },
    {
      "epoch": 3.1663467690339093,
      "grad_norm": 1.5492368200343252,
      "learning_rate": 1.5991758689227118e-05,
      "loss": 0.2482,
      "step": 4949
    },
    {
      "epoch": 3.166986564299424,
      "grad_norm": 1.5706659506783829,
      "learning_rate": 1.5990099557993354e-05,
      "loss": 0.229,
      "step": 4950
    },
    {
      "epoch": 3.167626359564939,
      "grad_norm": 1.5249737452370873,
      "learning_rate": 1.598844016955537e-05,
      "loss": 0.2478,
      "step": 4951
    },
    {
      "epoch": 3.1682661548304543,
      "grad_norm": 1.3996544065856336,
      "learning_rate": 1.5986780523984416e-05,
      "loss": 0.2327,
      "step": 4952
    },
    {
      "epoch": 3.168905950095969,
      "grad_norm": 1.5898087189748273,
      "learning_rate": 1.598512062135176e-05,
      "loss": 0.2387,
      "step": 4953
    },
    {
      "epoch": 3.1695457453614844,
      "grad_norm": 1.5148316464750522,
      "learning_rate": 1.598346046172867e-05,
      "loss": 0.2389,
      "step": 4954
    },
    {
      "epoch": 3.170185540626999,
      "grad_norm": 1.3334323137779334,
      "learning_rate": 1.598180004518643e-05,
      "loss": 0.2094,
      "step": 4955
    },
    {
      "epoch": 3.1708253358925145,
      "grad_norm": 1.5672673628694573,
      "learning_rate": 1.5980139371796333e-05,
      "loss": 0.2382,
      "step": 4956
    },
    {
      "epoch": 3.1714651311580293,
      "grad_norm": 1.5801269039094739,
      "learning_rate": 1.5978478441629697e-05,
      "loss": 0.2627,
      "step": 4957
    },
    {
      "epoch": 3.1721049264235446,
      "grad_norm": 1.3847679656094778,
      "learning_rate": 1.597681725475783e-05,
      "loss": 0.2088,
      "step": 4958
    },
    {
      "epoch": 3.1727447216890594,
      "grad_norm": 1.5480700054586491,
      "learning_rate": 1.5975155811252056e-05,
      "loss": 0.2451,
      "step": 4959
    },
    {
      "epoch": 3.1733845169545747,
      "grad_norm": 1.6549501744207882,
      "learning_rate": 1.5973494111183726e-05,
      "loss": 0.2747,
      "step": 4960
    },
    {
      "epoch": 3.1740243122200895,
      "grad_norm": 1.4065136938992209,
      "learning_rate": 1.5971832154624187e-05,
      "loss": 0.1947,
      "step": 4961
    },
    {
      "epoch": 3.174664107485605,
      "grad_norm": 1.5196147456226758,
      "learning_rate": 1.5970169941644793e-05,
      "loss": 0.231,
      "step": 4962
    },
    {
      "epoch": 3.1753039027511196,
      "grad_norm": 1.4707783563410077,
      "learning_rate": 1.5968507472316922e-05,
      "loss": 0.2203,
      "step": 4963
    },
    {
      "epoch": 3.175943698016635,
      "grad_norm": 1.5385979152150613,
      "learning_rate": 1.596684474671196e-05,
      "loss": 0.2286,
      "step": 4964
    },
    {
      "epoch": 3.1765834932821497,
      "grad_norm": 1.4227091232839923,
      "learning_rate": 1.5965181764901298e-05,
      "loss": 0.2069,
      "step": 4965
    },
    {
      "epoch": 3.1772232885476646,
      "grad_norm": 1.4347745652674553,
      "learning_rate": 1.596351852695634e-05,
      "loss": 0.2243,
      "step": 4966
    },
    {
      "epoch": 3.17786308381318,
      "grad_norm": 1.6000346535436554,
      "learning_rate": 1.5961855032948508e-05,
      "loss": 0.2348,
      "step": 4967
    },
    {
      "epoch": 3.1785028790786947,
      "grad_norm": 1.5961296055141807,
      "learning_rate": 1.5960191282949224e-05,
      "loss": 0.2691,
      "step": 4968
    },
    {
      "epoch": 3.17914267434421,
      "grad_norm": 1.4734734109996563,
      "learning_rate": 1.595852727702993e-05,
      "loss": 0.2179,
      "step": 4969
    },
    {
      "epoch": 3.1797824696097248,
      "grad_norm": 1.4488359259528847,
      "learning_rate": 1.5956863015262075e-05,
      "loss": 0.2164,
      "step": 4970
    },
    {
      "epoch": 3.18042226487524,
      "grad_norm": 1.5365302685813276,
      "learning_rate": 1.5955198497717116e-05,
      "loss": 0.2316,
      "step": 4971
    },
    {
      "epoch": 3.181062060140755,
      "grad_norm": 1.5296386971554985,
      "learning_rate": 1.5953533724466527e-05,
      "loss": 0.2364,
      "step": 4972
    },
    {
      "epoch": 3.18170185540627,
      "grad_norm": 1.557326686398437,
      "learning_rate": 1.5951868695581794e-05,
      "loss": 0.2457,
      "step": 4973
    },
    {
      "epoch": 3.182341650671785,
      "grad_norm": 1.4646757855211587,
      "learning_rate": 1.59502034111344e-05,
      "loss": 0.233,
      "step": 4974
    },
    {
      "epoch": 3.1829814459373003,
      "grad_norm": 1.4822637445679716,
      "learning_rate": 1.5948537871195863e-05,
      "loss": 0.2234,
      "step": 4975
    },
    {
      "epoch": 3.183621241202815,
      "grad_norm": 1.636961279017631,
      "learning_rate": 1.594687207583769e-05,
      "loss": 0.2389,
      "step": 4976
    },
    {
      "epoch": 3.18426103646833,
      "grad_norm": 1.5696370663457395,
      "learning_rate": 1.5945206025131404e-05,
      "loss": 0.2273,
      "step": 4977
    },
    {
      "epoch": 3.184900831733845,
      "grad_norm": 1.5375712375341164,
      "learning_rate": 1.5943539719148554e-05,
      "loss": 0.2262,
      "step": 4978
    },
    {
      "epoch": 3.18554062699936,
      "grad_norm": 1.4658244223832044,
      "learning_rate": 1.5941873157960677e-05,
      "loss": 0.2459,
      "step": 4979
    },
    {
      "epoch": 3.1861804222648753,
      "grad_norm": 1.5807377168315357,
      "learning_rate": 1.5940206341639338e-05,
      "loss": 0.2538,
      "step": 4980
    },
    {
      "epoch": 3.18682021753039,
      "grad_norm": 1.4030948687172335,
      "learning_rate": 1.5938539270256104e-05,
      "loss": 0.211,
      "step": 4981
    },
    {
      "epoch": 3.1874600127959054,
      "grad_norm": 1.5818531844216752,
      "learning_rate": 1.593687194388256e-05,
      "loss": 0.2367,
      "step": 4982
    },
    {
      "epoch": 3.1880998080614202,
      "grad_norm": 1.5099294457140726,
      "learning_rate": 1.5935204362590292e-05,
      "loss": 0.2315,
      "step": 4983
    },
    {
      "epoch": 3.1887396033269355,
      "grad_norm": 1.5146352413701585,
      "learning_rate": 1.5933536526450907e-05,
      "loss": 0.2074,
      "step": 4984
    },
    {
      "epoch": 3.1893793985924503,
      "grad_norm": 1.4763089434381766,
      "learning_rate": 1.593186843553602e-05,
      "loss": 0.232,
      "step": 4985
    },
    {
      "epoch": 3.1900191938579656,
      "grad_norm": 1.680231284520114,
      "learning_rate": 1.5930200089917252e-05,
      "loss": 0.2677,
      "step": 4986
    },
    {
      "epoch": 3.1906589891234804,
      "grad_norm": 1.4511276302375613,
      "learning_rate": 1.5928531489666247e-05,
      "loss": 0.2181,
      "step": 4987
    },
    {
      "epoch": 3.1912987843889957,
      "grad_norm": 1.5363875591762859,
      "learning_rate": 1.5926862634854644e-05,
      "loss": 0.2559,
      "step": 4988
    },
    {
      "epoch": 3.1919385796545106,
      "grad_norm": 1.48171906884283,
      "learning_rate": 1.59251935255541e-05,
      "loss": 0.2357,
      "step": 4989
    },
    {
      "epoch": 3.1925783749200254,
      "grad_norm": 1.3967701678402866,
      "learning_rate": 1.5923524161836287e-05,
      "loss": 0.2043,
      "step": 4990
    },
    {
      "epoch": 3.1932181701855407,
      "grad_norm": 1.5349381187758926,
      "learning_rate": 1.592185454377288e-05,
      "loss": 0.2358,
      "step": 4991
    },
    {
      "epoch": 3.1938579654510555,
      "grad_norm": 1.5249624688454229,
      "learning_rate": 1.592018467143558e-05,
      "loss": 0.224,
      "step": 4992
    },
    {
      "epoch": 3.1944977607165708,
      "grad_norm": 1.4327653430242469,
      "learning_rate": 1.5918514544896077e-05,
      "loss": 0.213,
      "step": 4993
    },
    {
      "epoch": 3.1951375559820856,
      "grad_norm": 1.4353430821747755,
      "learning_rate": 1.591684416422609e-05,
      "loss": 0.2054,
      "step": 4994
    },
    {
      "epoch": 3.195777351247601,
      "grad_norm": 1.5835652971521585,
      "learning_rate": 1.5915173529497337e-05,
      "loss": 0.2323,
      "step": 4995
    },
    {
      "epoch": 3.1964171465131157,
      "grad_norm": 1.5388996504740933,
      "learning_rate": 1.5913502640781554e-05,
      "loss": 0.2253,
      "step": 4996
    },
    {
      "epoch": 3.197056941778631,
      "grad_norm": 1.549345140429784,
      "learning_rate": 1.5911831498150492e-05,
      "loss": 0.2208,
      "step": 4997
    },
    {
      "epoch": 3.197696737044146,
      "grad_norm": 1.630535180186189,
      "learning_rate": 1.5910160101675896e-05,
      "loss": 0.245,
      "step": 4998
    },
    {
      "epoch": 3.198336532309661,
      "grad_norm": 1.6717102781700584,
      "learning_rate": 1.5908488451429546e-05,
      "loss": 0.2363,
      "step": 4999
    },
    {
      "epoch": 3.198976327575176,
      "grad_norm": 1.4914442475286167,
      "learning_rate": 1.5906816547483207e-05,
      "loss": 0.254,
      "step": 5000
    },
    {
      "epoch": 3.199616122840691,
      "grad_norm": 1.624943965396309,
      "learning_rate": 1.590514438990867e-05,
      "loss": 0.2268,
      "step": 5001
    },
    {
      "epoch": 3.200255918106206,
      "grad_norm": 1.5048395251754183,
      "learning_rate": 1.5903471978777743e-05,
      "loss": 0.2248,
      "step": 5002
    },
    {
      "epoch": 3.200895713371721,
      "grad_norm": 1.578790001683975,
      "learning_rate": 1.590179931416223e-05,
      "loss": 0.2289,
      "step": 5003
    },
    {
      "epoch": 3.201535508637236,
      "grad_norm": 1.4571678809666013,
      "learning_rate": 1.5900126396133953e-05,
      "loss": 0.2271,
      "step": 5004
    },
    {
      "epoch": 3.202175303902751,
      "grad_norm": 1.4991285963237162,
      "learning_rate": 1.589845322476474e-05,
      "loss": 0.2372,
      "step": 5005
    },
    {
      "epoch": 3.2028150991682662,
      "grad_norm": 1.4483210049547512,
      "learning_rate": 1.5896779800126444e-05,
      "loss": 0.2112,
      "step": 5006
    },
    {
      "epoch": 3.203454894433781,
      "grad_norm": 1.4502469590935447,
      "learning_rate": 1.589510612229091e-05,
      "loss": 0.2377,
      "step": 5007
    },
    {
      "epoch": 3.2040946896992963,
      "grad_norm": 1.4500640384453225,
      "learning_rate": 1.589343219133e-05,
      "loss": 0.2218,
      "step": 5008
    },
    {
      "epoch": 3.204734484964811,
      "grad_norm": 1.5520412394897931,
      "learning_rate": 1.5891758007315604e-05,
      "loss": 0.2181,
      "step": 5009
    },
    {
      "epoch": 3.2053742802303264,
      "grad_norm": 1.5788712072415274,
      "learning_rate": 1.5890083570319598e-05,
      "loss": 0.2774,
      "step": 5010
    },
    {
      "epoch": 3.2060140754958413,
      "grad_norm": 1.5072355204280043,
      "learning_rate": 1.588840888041388e-05,
      "loss": 0.2484,
      "step": 5011
    },
    {
      "epoch": 3.2066538707613566,
      "grad_norm": 1.391264842857285,
      "learning_rate": 1.588673393767036e-05,
      "loss": 0.2228,
      "step": 5012
    },
    {
      "epoch": 3.2072936660268714,
      "grad_norm": 1.599832893596739,
      "learning_rate": 1.5885058742160954e-05,
      "loss": 0.236,
      "step": 5013
    },
    {
      "epoch": 3.207933461292386,
      "grad_norm": 1.559481540946246,
      "learning_rate": 1.58833832939576e-05,
      "loss": 0.2522,
      "step": 5014
    },
    {
      "epoch": 3.2085732565579015,
      "grad_norm": 1.3133543369396512,
      "learning_rate": 1.5881707593132226e-05,
      "loss": 0.2051,
      "step": 5015
    },
    {
      "epoch": 3.2092130518234163,
      "grad_norm": 1.5235556232256755,
      "learning_rate": 1.588003163975679e-05,
      "loss": 0.2132,
      "step": 5016
    },
    {
      "epoch": 3.2098528470889316,
      "grad_norm": 1.649040752773713,
      "learning_rate": 1.5878355433903263e-05,
      "loss": 0.2561,
      "step": 5017
    },
    {
      "epoch": 3.2104926423544464,
      "grad_norm": 1.3795821381193958,
      "learning_rate": 1.5876678975643604e-05,
      "loss": 0.2134,
      "step": 5018
    },
    {
      "epoch": 3.2111324376199617,
      "grad_norm": 1.52502481634819,
      "learning_rate": 1.5875002265049808e-05,
      "loss": 0.2446,
      "step": 5019
    },
    {
      "epoch": 3.2117722328854765,
      "grad_norm": 1.4791448256768556,
      "learning_rate": 1.5873325302193862e-05,
      "loss": 0.2024,
      "step": 5020
    },
    {
      "epoch": 3.212412028150992,
      "grad_norm": 1.4326003116732542,
      "learning_rate": 1.5871648087147774e-05,
      "loss": 0.2283,
      "step": 5021
    },
    {
      "epoch": 3.2130518234165066,
      "grad_norm": 1.418828549907667,
      "learning_rate": 1.5869970619983567e-05,
      "loss": 0.2156,
      "step": 5022
    },
    {
      "epoch": 3.213691618682022,
      "grad_norm": 1.4877248252601971,
      "learning_rate": 1.586829290077326e-05,
      "loss": 0.2607,
      "step": 5023
    },
    {
      "epoch": 3.2143314139475367,
      "grad_norm": 1.5501160164489447,
      "learning_rate": 1.5866614929588898e-05,
      "loss": 0.2316,
      "step": 5024
    },
    {
      "epoch": 3.214971209213052,
      "grad_norm": 1.5248077267615334,
      "learning_rate": 1.5864936706502524e-05,
      "loss": 0.2421,
      "step": 5025
    },
    {
      "epoch": 3.215611004478567,
      "grad_norm": 1.6286681674141918,
      "learning_rate": 1.58632582315862e-05,
      "loss": 0.2374,
      "step": 5026
    },
    {
      "epoch": 3.216250799744082,
      "grad_norm": 1.502807209134612,
      "learning_rate": 1.5861579504912003e-05,
      "loss": 0.2296,
      "step": 5027
    },
    {
      "epoch": 3.216890595009597,
      "grad_norm": 1.685561706947756,
      "learning_rate": 1.5859900526552005e-05,
      "loss": 0.283,
      "step": 5028
    },
    {
      "epoch": 3.217530390275112,
      "grad_norm": 1.6287044143701688,
      "learning_rate": 1.58582212965783e-05,
      "loss": 0.2509,
      "step": 5029
    },
    {
      "epoch": 3.218170185540627,
      "grad_norm": 1.497197424943609,
      "learning_rate": 1.5856541815062995e-05,
      "loss": 0.2277,
      "step": 5030
    },
    {
      "epoch": 3.218809980806142,
      "grad_norm": 1.4822582196521839,
      "learning_rate": 1.5854862082078203e-05,
      "loss": 0.2395,
      "step": 5031
    },
    {
      "epoch": 3.219449776071657,
      "grad_norm": 1.5243489936814938,
      "learning_rate": 1.585318209769605e-05,
      "loss": 0.2525,
      "step": 5032
    },
    {
      "epoch": 3.220089571337172,
      "grad_norm": 1.448263566591502,
      "learning_rate": 1.5851501861988668e-05,
      "loss": 0.2174,
      "step": 5033
    },
    {
      "epoch": 3.2207293666026873,
      "grad_norm": 1.3851412859610064,
      "learning_rate": 1.5849821375028204e-05,
      "loss": 0.2066,
      "step": 5034
    },
    {
      "epoch": 3.221369161868202,
      "grad_norm": 1.5644927127989456,
      "learning_rate": 1.584814063688682e-05,
      "loss": 0.2596,
      "step": 5035
    },
    {
      "epoch": 3.2220089571337174,
      "grad_norm": 1.6142216145530979,
      "learning_rate": 1.5846459647636675e-05,
      "loss": 0.2405,
      "step": 5036
    },
    {
      "epoch": 3.222648752399232,
      "grad_norm": 1.5051069895525535,
      "learning_rate": 1.5844778407349956e-05,
      "loss": 0.2327,
      "step": 5037
    },
    {
      "epoch": 3.2232885476647475,
      "grad_norm": 1.5801588311212198,
      "learning_rate": 1.5843096916098848e-05,
      "loss": 0.2668,
      "step": 5038
    },
    {
      "epoch": 3.2239283429302623,
      "grad_norm": 1.4586838488510525,
      "learning_rate": 1.5841415173955554e-05,
      "loss": 0.2179,
      "step": 5039
    },
    {
      "epoch": 3.224568138195777,
      "grad_norm": 1.5629585244425452,
      "learning_rate": 1.5839733180992283e-05,
      "loss": 0.248,
      "step": 5040
    },
    {
      "epoch": 3.2252079334612924,
      "grad_norm": 1.4598036676863133,
      "learning_rate": 1.5838050937281255e-05,
      "loss": 0.2131,
      "step": 5041
    },
    {
      "epoch": 3.2258477287268073,
      "grad_norm": 1.4824347587029514,
      "learning_rate": 1.5836368442894706e-05,
      "loss": 0.231,
      "step": 5042
    },
    {
      "epoch": 3.2264875239923225,
      "grad_norm": 1.4110529058547783,
      "learning_rate": 1.583468569790488e-05,
      "loss": 0.2181,
      "step": 5043
    },
    {
      "epoch": 3.2271273192578374,
      "grad_norm": 1.496338798929377,
      "learning_rate": 1.5833002702384028e-05,
      "loss": 0.2223,
      "step": 5044
    },
    {
      "epoch": 3.2277671145233526,
      "grad_norm": 1.6805094353933132,
      "learning_rate": 1.5831319456404416e-05,
      "loss": 0.2662,
      "step": 5045
    },
    {
      "epoch": 3.2284069097888675,
      "grad_norm": 1.5595906870773955,
      "learning_rate": 1.5829635960038317e-05,
      "loss": 0.2423,
      "step": 5046
    },
    {
      "epoch": 3.2290467050543827,
      "grad_norm": 1.6879401281965705,
      "learning_rate": 1.5827952213358028e-05,
      "loss": 0.266,
      "step": 5047
    },
    {
      "epoch": 3.2296865003198976,
      "grad_norm": 1.604502546798409,
      "learning_rate": 1.582626821643583e-05,
      "loss": 0.2184,
      "step": 5048
    },
    {
      "epoch": 3.230326295585413,
      "grad_norm": 1.4585966709403142,
      "learning_rate": 1.5824583969344042e-05,
      "loss": 0.2346,
      "step": 5049
    },
    {
      "epoch": 3.2309660908509277,
      "grad_norm": 1.539780307281294,
      "learning_rate": 1.582289947215498e-05,
      "loss": 0.2187,
      "step": 5050
    },
    {
      "epoch": 3.2316058861164425,
      "grad_norm": 1.5778487823562717,
      "learning_rate": 1.5821214724940973e-05,
      "loss": 0.2312,
      "step": 5051
    },
    {
      "epoch": 3.232245681381958,
      "grad_norm": 1.3209479793074328,
      "learning_rate": 1.581952972777436e-05,
      "loss": 0.2195,
      "step": 5052
    },
    {
      "epoch": 3.2328854766474726,
      "grad_norm": 1.4724976762887945,
      "learning_rate": 1.581784448072749e-05,
      "loss": 0.2502,
      "step": 5053
    },
    {
      "epoch": 3.233525271912988,
      "grad_norm": 1.3913929943969896,
      "learning_rate": 1.581615898387273e-05,
      "loss": 0.2137,
      "step": 5054
    },
    {
      "epoch": 3.2341650671785027,
      "grad_norm": 1.4916866589080426,
      "learning_rate": 1.5814473237282452e-05,
      "loss": 0.2322,
      "step": 5055
    },
    {
      "epoch": 3.234804862444018,
      "grad_norm": 1.520990874151698,
      "learning_rate": 1.581278724102903e-05,
      "loss": 0.2607,
      "step": 5056
    },
    {
      "epoch": 3.235444657709533,
      "grad_norm": 1.6007114459658294,
      "learning_rate": 1.5811100995184866e-05,
      "loss": 0.2475,
      "step": 5057
    },
    {
      "epoch": 3.236084452975048,
      "grad_norm": 1.343521128867221,
      "learning_rate": 1.5809414499822365e-05,
      "loss": 0.2142,
      "step": 5058
    },
    {
      "epoch": 3.236724248240563,
      "grad_norm": 1.4499521351019071,
      "learning_rate": 1.580772775501394e-05,
      "loss": 0.2161,
      "step": 5059
    },
    {
      "epoch": 3.237364043506078,
      "grad_norm": 1.424661049663447,
      "learning_rate": 1.5806040760832014e-05,
      "loss": 0.2101,
      "step": 5060
    },
    {
      "epoch": 3.238003838771593,
      "grad_norm": 1.5571850671560854,
      "learning_rate": 1.5804353517349024e-05,
      "loss": 0.2458,
      "step": 5061
    },
    {
      "epoch": 3.2386436340371083,
      "grad_norm": 1.5407910955615232,
      "learning_rate": 1.5802666024637424e-05,
      "loss": 0.2291,
      "step": 5062
    },
    {
      "epoch": 3.239283429302623,
      "grad_norm": 1.495999063418005,
      "learning_rate": 1.5800978282769667e-05,
      "loss": 0.2177,
      "step": 5063
    },
    {
      "epoch": 3.2399232245681384,
      "grad_norm": 1.4249510915460017,
      "learning_rate": 1.5799290291818218e-05,
      "loss": 0.2294,
      "step": 5064
    },
    {
      "epoch": 3.2405630198336532,
      "grad_norm": 1.6046407688597544,
      "learning_rate": 1.5797602051855562e-05,
      "loss": 0.2252,
      "step": 5065
    },
    {
      "epoch": 3.241202815099168,
      "grad_norm": 1.597733346080674,
      "learning_rate": 1.5795913562954185e-05,
      "loss": 0.2643,
      "step": 5066
    },
    {
      "epoch": 3.2418426103646834,
      "grad_norm": 1.553034679764439,
      "learning_rate": 1.579422482518659e-05,
      "loss": 0.2342,
      "step": 5067
    },
    {
      "epoch": 3.242482405630198,
      "grad_norm": 1.548939558698884,
      "learning_rate": 1.579253583862529e-05,
      "loss": 0.2445,
      "step": 5068
    },
    {
      "epoch": 3.2431222008957135,
      "grad_norm": 1.6555149682381651,
      "learning_rate": 1.5790846603342805e-05,
      "loss": 0.251,
      "step": 5069
    },
    {
      "epoch": 3.2437619961612283,
      "grad_norm": 1.4806635681060059,
      "learning_rate": 1.5789157119411666e-05,
      "loss": 0.2404,
      "step": 5070
    },
    {
      "epoch": 3.2444017914267436,
      "grad_norm": 1.476318417145834,
      "learning_rate": 1.578746738690442e-05,
      "loss": 0.2119,
      "step": 5071
    },
    {
      "epoch": 3.2450415866922584,
      "grad_norm": 1.5358508959249797,
      "learning_rate": 1.5785777405893622e-05,
      "loss": 0.2517,
      "step": 5072
    },
    {
      "epoch": 3.2456813819577737,
      "grad_norm": 1.60409151249021,
      "learning_rate": 1.578408717645183e-05,
      "loss": 0.2581,
      "step": 5073
    },
    {
      "epoch": 3.2463211772232885,
      "grad_norm": 1.5744360738674843,
      "learning_rate": 1.5782396698651625e-05,
      "loss": 0.2612,
      "step": 5074
    },
    {
      "epoch": 3.2469609724888038,
      "grad_norm": 1.5738322711572046,
      "learning_rate": 1.578070597256559e-05,
      "loss": 0.225,
      "step": 5075
    },
    {
      "epoch": 3.2476007677543186,
      "grad_norm": 1.518418966335049,
      "learning_rate": 1.5779014998266327e-05,
      "loss": 0.233,
      "step": 5076
    },
    {
      "epoch": 3.2482405630198334,
      "grad_norm": 1.5005607971418937,
      "learning_rate": 1.577732377582644e-05,
      "loss": 0.2485,
      "step": 5077
    },
    {
      "epoch": 3.2488803582853487,
      "grad_norm": 1.6212653570897593,
      "learning_rate": 1.5775632305318548e-05,
      "loss": 0.2518,
      "step": 5078
    },
    {
      "epoch": 3.2495201535508635,
      "grad_norm": 1.5240287157782189,
      "learning_rate": 1.5773940586815275e-05,
      "loss": 0.2424,
      "step": 5079
    },
    {
      "epoch": 3.250159948816379,
      "grad_norm": 1.5310962740155902,
      "learning_rate": 1.5772248620389264e-05,
      "loss": 0.2187,
      "step": 5080
    },
    {
      "epoch": 3.2507997440818936,
      "grad_norm": 1.5836946698524954,
      "learning_rate": 1.577055640611317e-05,
      "loss": 0.2571,
      "step": 5081
    },
    {
      "epoch": 3.251439539347409,
      "grad_norm": 1.5813638935605407,
      "learning_rate": 1.5768863944059646e-05,
      "loss": 0.2388,
      "step": 5082
    },
    {
      "epoch": 3.2520793346129238,
      "grad_norm": 1.577381819357853,
      "learning_rate": 1.5767171234301363e-05,
      "loss": 0.2397,
      "step": 5083
    },
    {
      "epoch": 3.252719129878439,
      "grad_norm": 1.5111912476510247,
      "learning_rate": 1.576547827691101e-05,
      "loss": 0.2239,
      "step": 5084
    },
    {
      "epoch": 3.253358925143954,
      "grad_norm": 1.6275417878926672,
      "learning_rate": 1.576378507196128e-05,
      "loss": 0.2638,
      "step": 5085
    },
    {
      "epoch": 3.253998720409469,
      "grad_norm": 1.5148987525149975,
      "learning_rate": 1.5762091619524866e-05,
      "loss": 0.2442,
      "step": 5086
    },
    {
      "epoch": 3.254638515674984,
      "grad_norm": 1.4800755829480887,
      "learning_rate": 1.5760397919674487e-05,
      "loss": 0.2301,
      "step": 5087
    },
    {
      "epoch": 3.255278310940499,
      "grad_norm": 1.5261799686142614,
      "learning_rate": 1.5758703972482874e-05,
      "loss": 0.2064,
      "step": 5088
    },
    {
      "epoch": 3.255918106206014,
      "grad_norm": 1.548699350689226,
      "learning_rate": 1.575700977802275e-05,
      "loss": 0.2137,
      "step": 5089
    },
    {
      "epoch": 3.2565579014715293,
      "grad_norm": 1.6378465826590236,
      "learning_rate": 1.5755315336366872e-05,
      "loss": 0.2496,
      "step": 5090
    },
    {
      "epoch": 3.257197696737044,
      "grad_norm": 1.6669373782115926,
      "learning_rate": 1.575362064758799e-05,
      "loss": 0.2441,
      "step": 5091
    },
    {
      "epoch": 3.257837492002559,
      "grad_norm": 1.5921762328798261,
      "learning_rate": 1.575192571175887e-05,
      "loss": 0.2549,
      "step": 5092
    },
    {
      "epoch": 3.2584772872680743,
      "grad_norm": 1.4433428672885864,
      "learning_rate": 1.5750230528952295e-05,
      "loss": 0.2289,
      "step": 5093
    },
    {
      "epoch": 3.259117082533589,
      "grad_norm": 1.499597713320715,
      "learning_rate": 1.574853509924105e-05,
      "loss": 0.2288,
      "step": 5094
    },
    {
      "epoch": 3.2597568777991044,
      "grad_norm": 1.6169889038755327,
      "learning_rate": 1.574683942269793e-05,
      "loss": 0.2349,
      "step": 5095
    },
    {
      "epoch": 3.260396673064619,
      "grad_norm": 1.546155962136835,
      "learning_rate": 1.574514349939575e-05,
      "loss": 0.245,
      "step": 5096
    },
    {
      "epoch": 3.2610364683301345,
      "grad_norm": 1.6396579431604146,
      "learning_rate": 1.574344732940733e-05,
      "loss": 0.2484,
      "step": 5097
    },
    {
      "epoch": 3.2616762635956493,
      "grad_norm": 1.4797218419547458,
      "learning_rate": 1.5741750912805495e-05,
      "loss": 0.2687,
      "step": 5098
    },
    {
      "epoch": 3.2623160588611646,
      "grad_norm": 1.5166588437673518,
      "learning_rate": 1.574005424966309e-05,
      "loss": 0.2279,
      "step": 5099
    },
    {
      "epoch": 3.2629558541266794,
      "grad_norm": 1.6315818053499813,
      "learning_rate": 1.573835734005297e-05,
      "loss": 0.269,
      "step": 5100
    },
    {
      "epoch": 3.2635956493921947,
      "grad_norm": 1.5066640761122696,
      "learning_rate": 1.5736660184047993e-05,
      "loss": 0.241,
      "step": 5101
    },
    {
      "epoch": 3.2642354446577095,
      "grad_norm": 1.5772359457803693,
      "learning_rate": 1.573496278172103e-05,
      "loss": 0.2772,
      "step": 5102
    },
    {
      "epoch": 3.2648752399232244,
      "grad_norm": 1.5764558592663176,
      "learning_rate": 1.5733265133144965e-05,
      "loss": 0.2414,
      "step": 5103
    },
    {
      "epoch": 3.2655150351887396,
      "grad_norm": 1.4422182293347166,
      "learning_rate": 1.5731567238392697e-05,
      "loss": 0.2438,
      "step": 5104
    },
    {
      "epoch": 3.2661548304542545,
      "grad_norm": 1.4194088765935167,
      "learning_rate": 1.5729869097537126e-05,
      "loss": 0.2201,
      "step": 5105
    },
    {
      "epoch": 3.2667946257197698,
      "grad_norm": 1.4949913237282608,
      "learning_rate": 1.572817071065117e-05,
      "loss": 0.2254,
      "step": 5106
    },
    {
      "epoch": 3.2674344209852846,
      "grad_norm": 1.5934246561442504,
      "learning_rate": 1.572647207780775e-05,
      "loss": 0.2622,
      "step": 5107
    },
    {
      "epoch": 3.2680742162508,
      "grad_norm": 1.5523800080304095,
      "learning_rate": 1.572477319907981e-05,
      "loss": 0.2568,
      "step": 5108
    },
    {
      "epoch": 3.2687140115163147,
      "grad_norm": 1.525709492214153,
      "learning_rate": 1.5723074074540293e-05,
      "loss": 0.1967,
      "step": 5109
    },
    {
      "epoch": 3.26935380678183,
      "grad_norm": 1.6074888415905915,
      "learning_rate": 1.572137470426215e-05,
      "loss": 0.2344,
      "step": 5110
    },
    {
      "epoch": 3.269993602047345,
      "grad_norm": 1.5406717614154355,
      "learning_rate": 1.5719675088318356e-05,
      "loss": 0.2204,
      "step": 5111
    },
    {
      "epoch": 3.27063339731286,
      "grad_norm": 1.415629355959892,
      "learning_rate": 1.5717975226781893e-05,
      "loss": 0.2315,
      "step": 5112
    },
    {
      "epoch": 3.271273192578375,
      "grad_norm": 1.6724431846580334,
      "learning_rate": 1.571627511972574e-05,
      "loss": 0.2621,
      "step": 5113
    },
    {
      "epoch": 3.2719129878438897,
      "grad_norm": 1.510951059718763,
      "learning_rate": 1.5714574767222907e-05,
      "loss": 0.2414,
      "step": 5114
    },
    {
      "epoch": 3.272552783109405,
      "grad_norm": 1.5340921083531096,
      "learning_rate": 1.5712874169346398e-05,
      "loss": 0.235,
      "step": 5115
    },
    {
      "epoch": 3.27319257837492,
      "grad_norm": 1.4321701523259307,
      "learning_rate": 1.5711173326169233e-05,
      "loss": 0.227,
      "step": 5116
    },
    {
      "epoch": 3.273832373640435,
      "grad_norm": 1.3672110335445113,
      "learning_rate": 1.5709472237764445e-05,
      "loss": 0.2266,
      "step": 5117
    },
    {
      "epoch": 3.27447216890595,
      "grad_norm": 1.4681473317779614,
      "learning_rate": 1.5707770904205073e-05,
      "loss": 0.2215,
      "step": 5118
    },
    {
      "epoch": 3.275111964171465,
      "grad_norm": 1.4126537819985137,
      "learning_rate": 1.5706069325564175e-05,
      "loss": 0.2263,
      "step": 5119
    },
    {
      "epoch": 3.27575175943698,
      "grad_norm": 1.4462230917979741,
      "learning_rate": 1.570436750191481e-05,
      "loss": 0.2276,
      "step": 5120
    },
    {
      "epoch": 3.2763915547024953,
      "grad_norm": 1.528642383161223,
      "learning_rate": 1.570266543333005e-05,
      "loss": 0.2185,
      "step": 5121
    },
    {
      "epoch": 3.27703134996801,
      "grad_norm": 1.4020372528558096,
      "learning_rate": 1.5700963119882982e-05,
      "loss": 0.2043,
      "step": 5122
    },
    {
      "epoch": 3.2776711452335254,
      "grad_norm": 1.4986277393629002,
      "learning_rate": 1.5699260561646698e-05,
      "loss": 0.2112,
      "step": 5123
    },
    {
      "epoch": 3.2783109404990403,
      "grad_norm": 1.4700435723956302,
      "learning_rate": 1.5697557758694305e-05,
      "loss": 0.2108,
      "step": 5124
    },
    {
      "epoch": 3.2789507357645555,
      "grad_norm": 1.4530884970510392,
      "learning_rate": 1.5695854711098915e-05,
      "loss": 0.2105,
      "step": 5125
    },
    {
      "epoch": 3.2795905310300704,
      "grad_norm": 1.4846949857993013,
      "learning_rate": 1.569415141893366e-05,
      "loss": 0.2374,
      "step": 5126
    },
    {
      "epoch": 3.2802303262955856,
      "grad_norm": 1.6464096553137009,
      "learning_rate": 1.5692447882271667e-05,
      "loss": 0.2638,
      "step": 5127
    },
    {
      "epoch": 3.2808701215611005,
      "grad_norm": 1.4319423471957125,
      "learning_rate": 1.569074410118609e-05,
      "loss": 0.2151,
      "step": 5128
    },
    {
      "epoch": 3.2815099168266153,
      "grad_norm": 1.7632770769997432,
      "learning_rate": 1.5689040075750086e-05,
      "loss": 0.2392,
      "step": 5129
    },
    {
      "epoch": 3.2821497120921306,
      "grad_norm": 1.6517469767762225,
      "learning_rate": 1.5687335806036817e-05,
      "loss": 0.2527,
      "step": 5130
    },
    {
      "epoch": 3.2827895073576454,
      "grad_norm": 1.5155066984591206,
      "learning_rate": 1.568563129211947e-05,
      "loss": 0.2266,
      "step": 5131
    },
    {
      "epoch": 3.2834293026231607,
      "grad_norm": 1.5058324693125085,
      "learning_rate": 1.5683926534071226e-05,
      "loss": 0.2242,
      "step": 5132
    },
    {
      "epoch": 3.2840690978886755,
      "grad_norm": 1.444105845502471,
      "learning_rate": 1.5682221531965288e-05,
      "loss": 0.2184,
      "step": 5133
    },
    {
      "epoch": 3.284708893154191,
      "grad_norm": 1.5052851845408495,
      "learning_rate": 1.568051628587487e-05,
      "loss": 0.2578,
      "step": 5134
    },
    {
      "epoch": 3.2853486884197056,
      "grad_norm": 1.5921910773783652,
      "learning_rate": 1.567881079587318e-05,
      "loss": 0.2486,
      "step": 5135
    },
    {
      "epoch": 3.285988483685221,
      "grad_norm": 1.6114930987665947,
      "learning_rate": 1.5677105062033464e-05,
      "loss": 0.2224,
      "step": 5136
    },
    {
      "epoch": 3.2866282789507357,
      "grad_norm": 1.4649978221849587,
      "learning_rate": 1.5675399084428947e-05,
      "loss": 0.221,
      "step": 5137
    },
    {
      "epoch": 3.287268074216251,
      "grad_norm": 1.5589128223179298,
      "learning_rate": 1.5673692863132896e-05,
      "loss": 0.2349,
      "step": 5138
    },
    {
      "epoch": 3.287907869481766,
      "grad_norm": 1.5692736976793575,
      "learning_rate": 1.5671986398218564e-05,
      "loss": 0.2341,
      "step": 5139
    },
    {
      "epoch": 3.2885476647472807,
      "grad_norm": 1.6487978527064868,
      "learning_rate": 1.5670279689759228e-05,
      "loss": 0.2673,
      "step": 5140
    },
    {
      "epoch": 3.289187460012796,
      "grad_norm": 1.5262321593137778,
      "learning_rate": 1.5668572737828165e-05,
      "loss": 0.2189,
      "step": 5141
    },
    {
      "epoch": 3.2898272552783108,
      "grad_norm": 1.5958927760613193,
      "learning_rate": 1.566686554249867e-05,
      "loss": 0.2464,
      "step": 5142
    },
    {
      "epoch": 3.290467050543826,
      "grad_norm": 1.6158314968448233,
      "learning_rate": 1.566515810384406e-05,
      "loss": 0.2406,
      "step": 5143
    },
    {
      "epoch": 3.291106845809341,
      "grad_norm": 1.4500082935985095,
      "learning_rate": 1.5663450421937628e-05,
      "loss": 0.217,
      "step": 5144
    },
    {
      "epoch": 3.291746641074856,
      "grad_norm": 1.6168780282885693,
      "learning_rate": 1.566174249685271e-05,
      "loss": 0.2259,
      "step": 5145
    },
    {
      "epoch": 3.292386436340371,
      "grad_norm": 1.5470493461773205,
      "learning_rate": 1.5660034328662647e-05,
      "loss": 0.2314,
      "step": 5146
    },
    {
      "epoch": 3.2930262316058863,
      "grad_norm": 1.5390160276396803,
      "learning_rate": 1.5658325917440773e-05,
      "loss": 0.2334,
      "step": 5147
    },
    {
      "epoch": 3.293666026871401,
      "grad_norm": 1.5539051318653554,
      "learning_rate": 1.565661726326045e-05,
      "loss": 0.2332,
      "step": 5148
    },
    {
      "epoch": 3.2943058221369164,
      "grad_norm": 1.4942992376277704,
      "learning_rate": 1.5654908366195044e-05,
      "loss": 0.2453,
      "step": 5149
    },
    {
      "epoch": 3.294945617402431,
      "grad_norm": 1.466640101966074,
      "learning_rate": 1.5653199226317935e-05,
      "loss": 0.2135,
      "step": 5150
    },
    {
      "epoch": 3.295585412667946,
      "grad_norm": 1.4628422408789559,
      "learning_rate": 1.5651489843702506e-05,
      "loss": 0.2287,
      "step": 5151
    },
    {
      "epoch": 3.2962252079334613,
      "grad_norm": 1.5390607757509234,
      "learning_rate": 1.5649780218422156e-05,
      "loss": 0.2384,
      "step": 5152
    },
    {
      "epoch": 3.2968650031989766,
      "grad_norm": 1.5749745134808886,
      "learning_rate": 1.5648070350550292e-05,
      "loss": 0.2323,
      "step": 5153
    },
    {
      "epoch": 3.2975047984644914,
      "grad_norm": 1.6997348162879309,
      "learning_rate": 1.564636024016034e-05,
      "loss": 0.2513,
      "step": 5154
    },
    {
      "epoch": 3.2981445937300062,
      "grad_norm": 1.5050510619740016,
      "learning_rate": 1.5644649887325715e-05,
      "loss": 0.2076,
      "step": 5155
    },
    {
      "epoch": 3.2987843889955215,
      "grad_norm": 1.5342800031458486,
      "learning_rate": 1.5642939292119873e-05,
      "loss": 0.2133,
      "step": 5156
    },
    {
      "epoch": 3.2994241842610363,
      "grad_norm": 1.5017155995083933,
      "learning_rate": 1.5641228454616252e-05,
      "loss": 0.2212,
      "step": 5157
    },
    {
      "epoch": 3.3000639795265516,
      "grad_norm": 1.4782185536039216,
      "learning_rate": 1.5639517374888317e-05,
      "loss": 0.2477,
      "step": 5158
    },
    {
      "epoch": 3.3007037747920664,
      "grad_norm": 1.5841123201183387,
      "learning_rate": 1.5637806053009535e-05,
      "loss": 0.2429,
      "step": 5159
    },
    {
      "epoch": 3.3013435700575817,
      "grad_norm": 1.5240113134078255,
      "learning_rate": 1.563609448905339e-05,
      "loss": 0.2348,
      "step": 5160
    },
    {
      "epoch": 3.3019833653230966,
      "grad_norm": 1.5703316346138114,
      "learning_rate": 1.5634382683093377e-05,
      "loss": 0.2161,
      "step": 5161
    },
    {
      "epoch": 3.302623160588612,
      "grad_norm": 1.422711667217417,
      "learning_rate": 1.5632670635202992e-05,
      "loss": 0.2077,
      "step": 5162
    },
    {
      "epoch": 3.3032629558541267,
      "grad_norm": 1.7004667089333005,
      "learning_rate": 1.5630958345455752e-05,
      "loss": 0.2974,
      "step": 5163
    },
    {
      "epoch": 3.303902751119642,
      "grad_norm": 1.5718940720088055,
      "learning_rate": 1.5629245813925176e-05,
      "loss": 0.2531,
      "step": 5164
    },
    {
      "epoch": 3.3045425463851568,
      "grad_norm": 1.3588905957050603,
      "learning_rate": 1.5627533040684798e-05,
      "loss": 0.221,
      "step": 5165
    },
    {
      "epoch": 3.3051823416506716,
      "grad_norm": 1.4540822325381457,
      "learning_rate": 1.562582002580816e-05,
      "loss": 0.2546,
      "step": 5166
    },
    {
      "epoch": 3.305822136916187,
      "grad_norm": 1.6269872883307532,
      "learning_rate": 1.562410676936882e-05,
      "loss": 0.23,
      "step": 5167
    },
    {
      "epoch": 3.3064619321817017,
      "grad_norm": 1.5048838366081865,
      "learning_rate": 1.562239327144034e-05,
      "loss": 0.2182,
      "step": 5168
    },
    {
      "epoch": 3.307101727447217,
      "grad_norm": 1.5635880167226843,
      "learning_rate": 1.5620679532096296e-05,
      "loss": 0.2572,
      "step": 5169
    },
    {
      "epoch": 3.307741522712732,
      "grad_norm": 1.7448449831073516,
      "learning_rate": 1.5618965551410265e-05,
      "loss": 0.2742,
      "step": 5170
    },
    {
      "epoch": 3.308381317978247,
      "grad_norm": 1.4807440953210236,
      "learning_rate": 1.5617251329455857e-05,
      "loss": 0.2513,
      "step": 5171
    },
    {
      "epoch": 3.309021113243762,
      "grad_norm": 1.4573926870126175,
      "learning_rate": 1.5615536866306662e-05,
      "loss": 0.2244,
      "step": 5172
    },
    {
      "epoch": 3.309660908509277,
      "grad_norm": 1.5061013709165523,
      "learning_rate": 1.5613822162036308e-05,
      "loss": 0.2389,
      "step": 5173
    },
    {
      "epoch": 3.310300703774792,
      "grad_norm": 1.6208669341251882,
      "learning_rate": 1.5612107216718416e-05,
      "loss": 0.2556,
      "step": 5174
    },
    {
      "epoch": 3.3109404990403073,
      "grad_norm": 1.6117180432439346,
      "learning_rate": 1.561039203042662e-05,
      "loss": 0.2826,
      "step": 5175
    },
    {
      "epoch": 3.311580294305822,
      "grad_norm": 1.4546660425126152,
      "learning_rate": 1.5608676603234574e-05,
      "loss": 0.2197,
      "step": 5176
    },
    {
      "epoch": 3.312220089571337,
      "grad_norm": 1.470948666781644,
      "learning_rate": 1.560696093521593e-05,
      "loss": 0.2403,
      "step": 5177
    },
    {
      "epoch": 3.3128598848368522,
      "grad_norm": 1.4501510160671833,
      "learning_rate": 1.560524502644436e-05,
      "loss": 0.2275,
      "step": 5178
    },
    {
      "epoch": 3.313499680102367,
      "grad_norm": 1.6004196953844307,
      "learning_rate": 1.5603528876993533e-05,
      "loss": 0.2671,
      "step": 5179
    },
    {
      "epoch": 3.3141394753678823,
      "grad_norm": 1.4719014157293073,
      "learning_rate": 1.560181248693715e-05,
      "loss": 0.2291,
      "step": 5180
    },
    {
      "epoch": 3.314779270633397,
      "grad_norm": 1.4361579187123112,
      "learning_rate": 1.56000958563489e-05,
      "loss": 0.2326,
      "step": 5181
    },
    {
      "epoch": 3.3154190658989124,
      "grad_norm": 1.4910380123325184,
      "learning_rate": 1.5598378985302497e-05,
      "loss": 0.225,
      "step": 5182
    },
    {
      "epoch": 3.3160588611644273,
      "grad_norm": 1.3261294031067492,
      "learning_rate": 1.559666187387166e-05,
      "loss": 0.2035,
      "step": 5183
    },
    {
      "epoch": 3.3166986564299425,
      "grad_norm": 1.5410648598105672,
      "learning_rate": 1.559494452213012e-05,
      "loss": 0.2232,
      "step": 5184
    },
    {
      "epoch": 3.3173384516954574,
      "grad_norm": 1.5317834836880486,
      "learning_rate": 1.559322693015161e-05,
      "loss": 0.2467,
      "step": 5185
    },
    {
      "epoch": 3.3179782469609727,
      "grad_norm": 1.5515388388874765,
      "learning_rate": 1.5591509098009887e-05,
      "loss": 0.2579,
      "step": 5186
    },
    {
      "epoch": 3.3186180422264875,
      "grad_norm": 1.5010506569953785,
      "learning_rate": 1.5589791025778712e-05,
      "loss": 0.2324,
      "step": 5187
    },
    {
      "epoch": 3.3192578374920023,
      "grad_norm": 1.4488843954112667,
      "learning_rate": 1.558807271353185e-05,
      "loss": 0.2234,
      "step": 5188
    },
    {
      "epoch": 3.3198976327575176,
      "grad_norm": 1.481937617887409,
      "learning_rate": 1.558635416134309e-05,
      "loss": 0.22,
      "step": 5189
    },
    {
      "epoch": 3.320537428023033,
      "grad_norm": 1.5743677683747666,
      "learning_rate": 1.5584635369286218e-05,
      "loss": 0.2291,
      "step": 5190
    },
    {
      "epoch": 3.3211772232885477,
      "grad_norm": 1.569709629514696,
      "learning_rate": 1.5582916337435036e-05,
      "loss": 0.2458,
      "step": 5191
    },
    {
      "epoch": 3.3218170185540625,
      "grad_norm": 1.4809098613358305,
      "learning_rate": 1.5581197065863362e-05,
      "loss": 0.2255,
      "step": 5192
    },
    {
      "epoch": 3.322456813819578,
      "grad_norm": 1.466512337581624,
      "learning_rate": 1.557947755464501e-05,
      "loss": 0.2422,
      "step": 5193
    },
    {
      "epoch": 3.3230966090850926,
      "grad_norm": 1.5654548357011395,
      "learning_rate": 1.5577757803853817e-05,
      "loss": 0.2491,
      "step": 5194
    },
    {
      "epoch": 3.323736404350608,
      "grad_norm": 1.5110727270839817,
      "learning_rate": 1.5576037813563625e-05,
      "loss": 0.24,
      "step": 5195
    },
    {
      "epoch": 3.3243761996161227,
      "grad_norm": 1.4795766154911658,
      "learning_rate": 1.5574317583848295e-05,
      "loss": 0.2266,
      "step": 5196
    },
    {
      "epoch": 3.325015994881638,
      "grad_norm": 1.3749826167086212,
      "learning_rate": 1.5572597114781678e-05,
      "loss": 0.2168,
      "step": 5197
    },
    {
      "epoch": 3.325655790147153,
      "grad_norm": 1.5878186843519881,
      "learning_rate": 1.5570876406437655e-05,
      "loss": 0.2449,
      "step": 5198
    },
    {
      "epoch": 3.326295585412668,
      "grad_norm": 1.370697445962212,
      "learning_rate": 1.556915545889011e-05,
      "loss": 0.2028,
      "step": 5199
    },
    {
      "epoch": 3.326935380678183,
      "grad_norm": 1.4937407684957689,
      "learning_rate": 1.5567434272212932e-05,
      "loss": 0.2317,
      "step": 5200
    },
    {
      "epoch": 3.3275751759436982,
      "grad_norm": 1.5394753386148787,
      "learning_rate": 1.5565712846480037e-05,
      "loss": 0.2311,
      "step": 5201
    },
    {
      "epoch": 3.328214971209213,
      "grad_norm": 1.5469466352680812,
      "learning_rate": 1.556399118176533e-05,
      "loss": 0.2457,
      "step": 5202
    },
    {
      "epoch": 3.328854766474728,
      "grad_norm": 1.5754656807154828,
      "learning_rate": 1.556226927814274e-05,
      "loss": 0.2453,
      "step": 5203
    },
    {
      "epoch": 3.329494561740243,
      "grad_norm": 1.6212736520840976,
      "learning_rate": 1.55605471356862e-05,
      "loss": 0.2255,
      "step": 5204
    },
    {
      "epoch": 3.330134357005758,
      "grad_norm": 1.561390110364348,
      "learning_rate": 1.5558824754469654e-05,
      "loss": 0.237,
      "step": 5205
    },
    {
      "epoch": 3.3307741522712733,
      "grad_norm": 1.5143289832579467,
      "learning_rate": 1.5557102134567067e-05,
      "loss": 0.2012,
      "step": 5206
    },
    {
      "epoch": 3.331413947536788,
      "grad_norm": 1.6922278175447816,
      "learning_rate": 1.5555379276052397e-05,
      "loss": 0.2752,
      "step": 5207
    },
    {
      "epoch": 3.3320537428023034,
      "grad_norm": 1.6019017508003037,
      "learning_rate": 1.5553656178999627e-05,
      "loss": 0.2344,
      "step": 5208
    },
    {
      "epoch": 3.332693538067818,
      "grad_norm": 1.5615192160650873,
      "learning_rate": 1.5551932843482734e-05,
      "loss": 0.2414,
      "step": 5209
    },
    {
      "epoch": 3.3333333333333335,
      "grad_norm": 1.5442814839499397,
      "learning_rate": 1.5550209269575723e-05,
      "loss": 0.2231,
      "step": 5210
    },
    {
      "epoch": 3.3339731285988483,
      "grad_norm": 1.4313466529721892,
      "learning_rate": 1.55484854573526e-05,
      "loss": 0.2278,
      "step": 5211
    },
    {
      "epoch": 3.3346129238643636,
      "grad_norm": 1.6060483272575155,
      "learning_rate": 1.5546761406887382e-05,
      "loss": 0.2708,
      "step": 5212
    },
    {
      "epoch": 3.3352527191298784,
      "grad_norm": 1.5481156653059953,
      "learning_rate": 1.55450371182541e-05,
      "loss": 0.2435,
      "step": 5213
    },
    {
      "epoch": 3.3358925143953932,
      "grad_norm": 1.602296074231638,
      "learning_rate": 1.554331259152678e-05,
      "loss": 0.2553,
      "step": 5214
    },
    {
      "epoch": 3.3365323096609085,
      "grad_norm": 1.5649061478331439,
      "learning_rate": 1.554158782677948e-05,
      "loss": 0.2357,
      "step": 5215
    },
    {
      "epoch": 3.3371721049264234,
      "grad_norm": 1.6353575744442082,
      "learning_rate": 1.5539862824086262e-05,
      "loss": 0.2553,
      "step": 5216
    },
    {
      "epoch": 3.3378119001919386,
      "grad_norm": 1.54641693060766,
      "learning_rate": 1.553813758352118e-05,
      "loss": 0.2249,
      "step": 5217
    },
    {
      "epoch": 3.3384516954574535,
      "grad_norm": 1.3921834869885863,
      "learning_rate": 1.553641210515833e-05,
      "loss": 0.241,
      "step": 5218
    },
    {
      "epoch": 3.3390914907229687,
      "grad_norm": 1.4233359172363182,
      "learning_rate": 1.5534686389071788e-05,
      "loss": 0.197,
      "step": 5219
    },
    {
      "epoch": 3.3397312859884836,
      "grad_norm": 1.6638415116243928,
      "learning_rate": 1.553296043533566e-05,
      "loss": 0.2255,
      "step": 5220
    },
    {
      "epoch": 3.340371081253999,
      "grad_norm": 1.4238781677342798,
      "learning_rate": 1.553123424402405e-05,
      "loss": 0.2011,
      "step": 5221
    },
    {
      "epoch": 3.3410108765195137,
      "grad_norm": 1.4887122430714217,
      "learning_rate": 1.5529507815211085e-05,
      "loss": 0.2233,
      "step": 5222
    },
    {
      "epoch": 3.341650671785029,
      "grad_norm": 1.3874360889313324,
      "learning_rate": 1.5527781148970888e-05,
      "loss": 0.2248,
      "step": 5223
    },
    {
      "epoch": 3.342290467050544,
      "grad_norm": 1.4557813960285373,
      "learning_rate": 1.5526054245377606e-05,
      "loss": 0.2445,
      "step": 5224
    },
    {
      "epoch": 3.342930262316059,
      "grad_norm": 1.605598687879766,
      "learning_rate": 1.552432710450538e-05,
      "loss": 0.2302,
      "step": 5225
    },
    {
      "epoch": 3.343570057581574,
      "grad_norm": 1.526670232951678,
      "learning_rate": 1.5522599726428375e-05,
      "loss": 0.2383,
      "step": 5226
    },
    {
      "epoch": 3.344209852847089,
      "grad_norm": 1.595419843241311,
      "learning_rate": 1.5520872111220765e-05,
      "loss": 0.2472,
      "step": 5227
    },
    {
      "epoch": 3.344849648112604,
      "grad_norm": 1.6157788291662682,
      "learning_rate": 1.5519144258956728e-05,
      "loss": 0.2741,
      "step": 5228
    },
    {
      "epoch": 3.345489443378119,
      "grad_norm": 1.505908292862166,
      "learning_rate": 1.551741616971045e-05,
      "loss": 0.2167,
      "step": 5229
    },
    {
      "epoch": 3.346129238643634,
      "grad_norm": 1.512528304309214,
      "learning_rate": 1.5515687843556137e-05,
      "loss": 0.2265,
      "step": 5230
    },
    {
      "epoch": 3.346769033909149,
      "grad_norm": 1.540399410986239,
      "learning_rate": 1.5513959280568004e-05,
      "loss": 0.231,
      "step": 5231
    },
    {
      "epoch": 3.347408829174664,
      "grad_norm": 1.576372682060109,
      "learning_rate": 1.5512230480820264e-05,
      "loss": 0.2303,
      "step": 5232
    },
    {
      "epoch": 3.348048624440179,
      "grad_norm": 1.6590322733918794,
      "learning_rate": 1.5510501444387155e-05,
      "loss": 0.2488,
      "step": 5233
    },
    {
      "epoch": 3.3486884197056943,
      "grad_norm": 1.4636361256372548,
      "learning_rate": 1.5508772171342913e-05,
      "loss": 0.2067,
      "step": 5234
    },
    {
      "epoch": 3.349328214971209,
      "grad_norm": 1.5664480508066219,
      "learning_rate": 1.5507042661761794e-05,
      "loss": 0.2464,
      "step": 5235
    },
    {
      "epoch": 3.3499680102367244,
      "grad_norm": 1.532920034353949,
      "learning_rate": 1.550531291571806e-05,
      "loss": 0.225,
      "step": 5236
    },
    {
      "epoch": 3.3506078055022392,
      "grad_norm": 1.5524392184734044,
      "learning_rate": 1.5503582933285986e-05,
      "loss": 0.2247,
      "step": 5237
    },
    {
      "epoch": 3.3512476007677545,
      "grad_norm": 1.4345111667130526,
      "learning_rate": 1.5501852714539847e-05,
      "loss": 0.2344,
      "step": 5238
    },
    {
      "epoch": 3.3518873960332694,
      "grad_norm": 1.4546812945022816,
      "learning_rate": 1.5500122259553942e-05,
      "loss": 0.2381,
      "step": 5239
    },
    {
      "epoch": 3.352527191298784,
      "grad_norm": 1.2042554360371356,
      "learning_rate": 1.549839156840257e-05,
      "loss": 0.1756,
      "step": 5240
    },
    {
      "epoch": 3.3531669865642995,
      "grad_norm": 1.6624015503297422,
      "learning_rate": 1.5496660641160045e-05,
      "loss": 0.2669,
      "step": 5241
    },
    {
      "epoch": 3.3538067818298143,
      "grad_norm": 1.5386229511255347,
      "learning_rate": 1.549492947790069e-05,
      "loss": 0.2371,
      "step": 5242
    },
    {
      "epoch": 3.3544465770953296,
      "grad_norm": 1.5352408810015439,
      "learning_rate": 1.5493198078698838e-05,
      "loss": 0.2552,
      "step": 5243
    },
    {
      "epoch": 3.3550863723608444,
      "grad_norm": 1.535204362642921,
      "learning_rate": 1.549146644362883e-05,
      "loss": 0.2144,
      "step": 5244
    },
    {
      "epoch": 3.3557261676263597,
      "grad_norm": 1.4771701911532569,
      "learning_rate": 1.5489734572765025e-05,
      "loss": 0.2442,
      "step": 5245
    },
    {
      "epoch": 3.3563659628918745,
      "grad_norm": 1.5227645452146137,
      "learning_rate": 1.548800246618178e-05,
      "loss": 0.2426,
      "step": 5246
    },
    {
      "epoch": 3.3570057581573898,
      "grad_norm": 1.582360533726851,
      "learning_rate": 1.5486270123953472e-05,
      "loss": 0.2376,
      "step": 5247
    },
    {
      "epoch": 3.3576455534229046,
      "grad_norm": 1.569437049477695,
      "learning_rate": 1.5484537546154486e-05,
      "loss": 0.2274,
      "step": 5248
    },
    {
      "epoch": 3.35828534868842,
      "grad_norm": 1.4648802864433164,
      "learning_rate": 1.5482804732859212e-05,
      "loss": 0.2235,
      "step": 5249
    },
    {
      "epoch": 3.3589251439539347,
      "grad_norm": 1.4828273339438554,
      "learning_rate": 1.5481071684142058e-05,
      "loss": 0.2416,
      "step": 5250
    },
    {
      "epoch": 3.3595649392194495,
      "grad_norm": 1.4396196551508105,
      "learning_rate": 1.5479338400077435e-05,
      "loss": 0.2184,
      "step": 5251
    },
    {
      "epoch": 3.360204734484965,
      "grad_norm": 1.554941249882423,
      "learning_rate": 1.5477604880739766e-05,
      "loss": 0.2379,
      "step": 5252
    },
    {
      "epoch": 3.36084452975048,
      "grad_norm": 1.60326377452445,
      "learning_rate": 1.5475871126203494e-05,
      "loss": 0.2686,
      "step": 5253
    },
    {
      "epoch": 3.361484325015995,
      "grad_norm": 1.53368740355957,
      "learning_rate": 1.547413713654305e-05,
      "loss": 0.2411,
      "step": 5254
    },
    {
      "epoch": 3.3621241202815098,
      "grad_norm": 1.5415772831824137,
      "learning_rate": 1.54724029118329e-05,
      "loss": 0.2455,
      "step": 5255
    },
    {
      "epoch": 3.362763915547025,
      "grad_norm": 1.5670704786525356,
      "learning_rate": 1.54706684521475e-05,
      "loss": 0.2489,
      "step": 5256
    },
    {
      "epoch": 3.36340371081254,
      "grad_norm": 1.6457283669202616,
      "learning_rate": 1.5468933757561325e-05,
      "loss": 0.2464,
      "step": 5257
    },
    {
      "epoch": 3.364043506078055,
      "grad_norm": 1.4128804432707736,
      "learning_rate": 1.5467198828148867e-05,
      "loss": 0.2084,
      "step": 5258
    },
    {
      "epoch": 3.36468330134357,
      "grad_norm": 1.6016902570574594,
      "learning_rate": 1.5465463663984616e-05,
      "loss": 0.2512,
      "step": 5259
    },
    {
      "epoch": 3.3653230966090852,
      "grad_norm": 1.4317286489619776,
      "learning_rate": 1.546372826514308e-05,
      "loss": 0.2076,
      "step": 5260
    },
    {
      "epoch": 3.3659628918746,
      "grad_norm": 1.600899817693653,
      "learning_rate": 1.5461992631698768e-05,
      "loss": 0.244,
      "step": 5261
    },
    {
      "epoch": 3.3666026871401153,
      "grad_norm": 1.4180167151726628,
      "learning_rate": 1.546025676372621e-05,
      "loss": 0.2389,
      "step": 5262
    },
    {
      "epoch": 3.36724248240563,
      "grad_norm": 1.4945163990033279,
      "learning_rate": 1.545852066129994e-05,
      "loss": 0.2385,
      "step": 5263
    },
    {
      "epoch": 3.3678822776711455,
      "grad_norm": 1.4382529244705295,
      "learning_rate": 1.5456784324494504e-05,
      "loss": 0.2363,
      "step": 5264
    },
    {
      "epoch": 3.3685220729366603,
      "grad_norm": 1.6203972708943315,
      "learning_rate": 1.5455047753384455e-05,
      "loss": 0.2831,
      "step": 5265
    },
    {
      "epoch": 3.369161868202175,
      "grad_norm": 1.540361060678518,
      "learning_rate": 1.5453310948044358e-05,
      "loss": 0.246,
      "step": 5266
    },
    {
      "epoch": 3.3698016634676904,
      "grad_norm": 1.4656252664134162,
      "learning_rate": 1.545157390854879e-05,
      "loss": 0.2081,
      "step": 5267
    },
    {
      "epoch": 3.370441458733205,
      "grad_norm": 1.5625283981929314,
      "learning_rate": 1.5449836634972334e-05,
      "loss": 0.2491,
      "step": 5268
    },
    {
      "epoch": 3.3710812539987205,
      "grad_norm": 1.693607098906316,
      "learning_rate": 1.544809912738959e-05,
      "loss": 0.2587,
      "step": 5269
    },
    {
      "epoch": 3.3717210492642353,
      "grad_norm": 1.5656368040042523,
      "learning_rate": 1.5446361385875165e-05,
      "loss": 0.2426,
      "step": 5270
    },
    {
      "epoch": 3.3723608445297506,
      "grad_norm": 1.6035221796936932,
      "learning_rate": 1.5444623410503663e-05,
      "loss": 0.2473,
      "step": 5271
    },
    {
      "epoch": 3.3730006397952654,
      "grad_norm": 1.4687102803119925,
      "learning_rate": 1.544288520134972e-05,
      "loss": 0.2121,
      "step": 5272
    },
    {
      "epoch": 3.3736404350607807,
      "grad_norm": 1.6193641247389274,
      "learning_rate": 1.544114675848797e-05,
      "loss": 0.2292,
      "step": 5273
    },
    {
      "epoch": 3.3742802303262955,
      "grad_norm": 1.6093374362621902,
      "learning_rate": 1.5439408081993055e-05,
      "loss": 0.2154,
      "step": 5274
    },
    {
      "epoch": 3.374920025591811,
      "grad_norm": 1.448775187809153,
      "learning_rate": 1.5437669171939633e-05,
      "loss": 0.2378,
      "step": 5275
    },
    {
      "epoch": 3.3755598208573256,
      "grad_norm": 1.4953829159090004,
      "learning_rate": 1.543593002840237e-05,
      "loss": 0.2479,
      "step": 5276
    },
    {
      "epoch": 3.3761996161228405,
      "grad_norm": 1.5185945747270218,
      "learning_rate": 1.543419065145594e-05,
      "loss": 0.2323,
      "step": 5277
    },
    {
      "epoch": 3.3768394113883557,
      "grad_norm": 1.374185924143719,
      "learning_rate": 1.5432451041175035e-05,
      "loss": 0.2037,
      "step": 5278
    },
    {
      "epoch": 3.3774792066538706,
      "grad_norm": 1.6339556153894408,
      "learning_rate": 1.5430711197634343e-05,
      "loss": 0.2894,
      "step": 5279
    },
    {
      "epoch": 3.378119001919386,
      "grad_norm": 1.410287524648366,
      "learning_rate": 1.542897112090857e-05,
      "loss": 0.2313,
      "step": 5280
    },
    {
      "epoch": 3.3787587971849007,
      "grad_norm": 1.575892004959656,
      "learning_rate": 1.5427230811072437e-05,
      "loss": 0.2588,
      "step": 5281
    },
    {
      "epoch": 3.379398592450416,
      "grad_norm": 1.541751564205563,
      "learning_rate": 1.5425490268200666e-05,
      "loss": 0.2589,
      "step": 5282
    },
    {
      "epoch": 3.380038387715931,
      "grad_norm": 1.52352008000668,
      "learning_rate": 1.5423749492367997e-05,
      "loss": 0.2314,
      "step": 5283
    },
    {
      "epoch": 3.380678182981446,
      "grad_norm": 1.5029334903586742,
      "learning_rate": 1.542200848364917e-05,
      "loss": 0.2399,
      "step": 5284
    },
    {
      "epoch": 3.381317978246961,
      "grad_norm": 1.6648429353635181,
      "learning_rate": 1.5420267242118945e-05,
      "loss": 0.2574,
      "step": 5285
    },
    {
      "epoch": 3.381957773512476,
      "grad_norm": 1.6518623570285333,
      "learning_rate": 1.5418525767852087e-05,
      "loss": 0.2641,
      "step": 5286
    },
    {
      "epoch": 3.382597568777991,
      "grad_norm": 1.5698278263171945,
      "learning_rate": 1.541678406092337e-05,
      "loss": 0.2346,
      "step": 5287
    },
    {
      "epoch": 3.383237364043506,
      "grad_norm": 1.4189475441697204,
      "learning_rate": 1.5415042121407583e-05,
      "loss": 0.2108,
      "step": 5288
    },
    {
      "epoch": 3.383877159309021,
      "grad_norm": 1.3925663350330415,
      "learning_rate": 1.541329994937952e-05,
      "loss": 0.2235,
      "step": 5289
    },
    {
      "epoch": 3.3845169545745364,
      "grad_norm": 1.5112960135258686,
      "learning_rate": 1.5411557544913983e-05,
      "loss": 0.2026,
      "step": 5290
    },
    {
      "epoch": 3.385156749840051,
      "grad_norm": 1.502147241648329,
      "learning_rate": 1.5409814908085792e-05,
      "loss": 0.2477,
      "step": 5291
    },
    {
      "epoch": 3.385796545105566,
      "grad_norm": 1.4944704063335108,
      "learning_rate": 1.5408072038969776e-05,
      "loss": 0.2315,
      "step": 5292
    },
    {
      "epoch": 3.3864363403710813,
      "grad_norm": 1.49677758685156,
      "learning_rate": 1.5406328937640768e-05,
      "loss": 0.2299,
      "step": 5293
    },
    {
      "epoch": 3.387076135636596,
      "grad_norm": 1.5561694347925608,
      "learning_rate": 1.5404585604173606e-05,
      "loss": 0.2393,
      "step": 5294
    },
    {
      "epoch": 3.3877159309021114,
      "grad_norm": 1.5213532824253717,
      "learning_rate": 1.5402842038643157e-05,
      "loss": 0.2303,
      "step": 5295
    },
    {
      "epoch": 3.3883557261676263,
      "grad_norm": 1.5064450221853294,
      "learning_rate": 1.5401098241124278e-05,
      "loss": 0.2721,
      "step": 5296
    },
    {
      "epoch": 3.3889955214331415,
      "grad_norm": 1.4536675160585475,
      "learning_rate": 1.5399354211691854e-05,
      "loss": 0.2017,
      "step": 5297
    },
    {
      "epoch": 3.3896353166986564,
      "grad_norm": 1.5435600190000907,
      "learning_rate": 1.539760995042076e-05,
      "loss": 0.259,
      "step": 5298
    },
    {
      "epoch": 3.3902751119641716,
      "grad_norm": 1.5123536362910492,
      "learning_rate": 1.5395865457385896e-05,
      "loss": 0.2426,
      "step": 5299
    },
    {
      "epoch": 3.3909149072296865,
      "grad_norm": 1.4836723038008386,
      "learning_rate": 1.539412073266217e-05,
      "loss": 0.2359,
      "step": 5300
    },
    {
      "epoch": 3.3915547024952017,
      "grad_norm": 1.6331343541712398,
      "learning_rate": 1.5392375776324494e-05,
      "loss": 0.2362,
      "step": 5301
    },
    {
      "epoch": 3.3921944977607166,
      "grad_norm": 1.5293913931537593,
      "learning_rate": 1.53906305884478e-05,
      "loss": 0.2498,
      "step": 5302
    },
    {
      "epoch": 3.3928342930262314,
      "grad_norm": 1.4927414843950317,
      "learning_rate": 1.538888516910701e-05,
      "loss": 0.2239,
      "step": 5303
    },
    {
      "epoch": 3.3934740882917467,
      "grad_norm": 1.566209270243088,
      "learning_rate": 1.538713951837708e-05,
      "loss": 0.2225,
      "step": 5304
    },
    {
      "epoch": 3.3941138835572615,
      "grad_norm": 1.4913372773659712,
      "learning_rate": 1.538539363633296e-05,
      "loss": 0.2516,
      "step": 5305
    },
    {
      "epoch": 3.394753678822777,
      "grad_norm": 1.5582755266158963,
      "learning_rate": 1.538364752304962e-05,
      "loss": 0.2327,
      "step": 5306
    },
    {
      "epoch": 3.3953934740882916,
      "grad_norm": 1.5025160681473588,
      "learning_rate": 1.5381901178602032e-05,
      "loss": 0.2478,
      "step": 5307
    },
    {
      "epoch": 3.396033269353807,
      "grad_norm": 1.6111566545523115,
      "learning_rate": 1.5380154603065178e-05,
      "loss": 0.2351,
      "step": 5308
    },
    {
      "epoch": 3.3966730646193217,
      "grad_norm": 1.45373404858823,
      "learning_rate": 1.537840779651406e-05,
      "loss": 0.2081,
      "step": 5309
    },
    {
      "epoch": 3.397312859884837,
      "grad_norm": 1.4763976456301815,
      "learning_rate": 1.5376660759023678e-05,
      "loss": 0.2453,
      "step": 5310
    },
    {
      "epoch": 3.397952655150352,
      "grad_norm": 1.5216969271685312,
      "learning_rate": 1.5374913490669045e-05,
      "loss": 0.222,
      "step": 5311
    },
    {
      "epoch": 3.398592450415867,
      "grad_norm": 1.4890134610719903,
      "learning_rate": 1.537316599152519e-05,
      "loss": 0.2232,
      "step": 5312
    },
    {
      "epoch": 3.399232245681382,
      "grad_norm": 1.492811310825271,
      "learning_rate": 1.5371418261667145e-05,
      "loss": 0.2112,
      "step": 5313
    },
    {
      "epoch": 3.3998720409468968,
      "grad_norm": 1.492760179582933,
      "learning_rate": 1.536967030116995e-05,
      "loss": 0.2376,
      "step": 5314
    },
    {
      "epoch": 3.400511836212412,
      "grad_norm": 1.4939562321773543,
      "learning_rate": 1.536792211010867e-05,
      "loss": 0.2326,
      "step": 5315
    },
    {
      "epoch": 3.401151631477927,
      "grad_norm": 1.5476370247712294,
      "learning_rate": 1.5366173688558363e-05,
      "loss": 0.2382,
      "step": 5316
    },
    {
      "epoch": 3.401791426743442,
      "grad_norm": 1.5361831889847193,
      "learning_rate": 1.5364425036594103e-05,
      "loss": 0.2539,
      "step": 5317
    },
    {
      "epoch": 3.402431222008957,
      "grad_norm": 1.3973959463088073,
      "learning_rate": 1.5362676154290974e-05,
      "loss": 0.202,
      "step": 5318
    },
    {
      "epoch": 3.4030710172744723,
      "grad_norm": 1.490111870429855,
      "learning_rate": 1.5360927041724068e-05,
      "loss": 0.2471,
      "step": 5319
    },
    {
      "epoch": 3.403710812539987,
      "grad_norm": 1.4046061077275676,
      "learning_rate": 1.5359177698968495e-05,
      "loss": 0.2049,
      "step": 5320
    },
    {
      "epoch": 3.4043506078055024,
      "grad_norm": 1.4903623356785825,
      "learning_rate": 1.5357428126099363e-05,
      "loss": 0.2324,
      "step": 5321
    },
    {
      "epoch": 3.404990403071017,
      "grad_norm": 1.462201100062638,
      "learning_rate": 1.5355678323191797e-05,
      "loss": 0.2187,
      "step": 5322
    },
    {
      "epoch": 3.4056301983365325,
      "grad_norm": 1.5592351758112677,
      "learning_rate": 1.535392829032093e-05,
      "loss": 0.2388,
      "step": 5323
    },
    {
      "epoch": 3.4062699936020473,
      "grad_norm": 1.5023986605002424,
      "learning_rate": 1.5352178027561908e-05,
      "loss": 0.2191,
      "step": 5324
    },
    {
      "epoch": 3.4069097888675626,
      "grad_norm": 1.6149441141519423,
      "learning_rate": 1.5350427534989884e-05,
      "loss": 0.2322,
      "step": 5325
    },
    {
      "epoch": 3.4075495841330774,
      "grad_norm": 1.5509040452602592,
      "learning_rate": 1.5348676812680017e-05,
      "loss": 0.239,
      "step": 5326
    },
    {
      "epoch": 3.4081893793985927,
      "grad_norm": 1.411708933767092,
      "learning_rate": 1.5346925860707483e-05,
      "loss": 0.2351,
      "step": 5327
    },
    {
      "epoch": 3.4088291746641075,
      "grad_norm": 1.5346762777042169,
      "learning_rate": 1.5345174679147462e-05,
      "loss": 0.2448,
      "step": 5328
    },
    {
      "epoch": 3.4094689699296223,
      "grad_norm": 1.6316936352434033,
      "learning_rate": 1.534342326807515e-05,
      "loss": 0.2378,
      "step": 5329
    },
    {
      "epoch": 3.4101087651951376,
      "grad_norm": 1.4474359481924273,
      "learning_rate": 1.5341671627565748e-05,
      "loss": 0.2218,
      "step": 5330
    },
    {
      "epoch": 3.4107485604606524,
      "grad_norm": 1.4694610891380027,
      "learning_rate": 1.533991975769447e-05,
      "loss": 0.2444,
      "step": 5331
    },
    {
      "epoch": 3.4113883557261677,
      "grad_norm": 1.4313013507911834,
      "learning_rate": 1.5338167658536537e-05,
      "loss": 0.2266,
      "step": 5332
    },
    {
      "epoch": 3.4120281509916826,
      "grad_norm": 1.436767087365902,
      "learning_rate": 1.5336415330167177e-05,
      "loss": 0.214,
      "step": 5333
    },
    {
      "epoch": 3.412667946257198,
      "grad_norm": 1.4181765014170777,
      "learning_rate": 1.5334662772661635e-05,
      "loss": 0.2292,
      "step": 5334
    },
    {
      "epoch": 3.4133077415227127,
      "grad_norm": 1.508202719474824,
      "learning_rate": 1.5332909986095166e-05,
      "loss": 0.2402,
      "step": 5335
    },
    {
      "epoch": 3.413947536788228,
      "grad_norm": 1.669173059628539,
      "learning_rate": 1.533115697054303e-05,
      "loss": 0.2594,
      "step": 5336
    },
    {
      "epoch": 3.4145873320537428,
      "grad_norm": 1.5262223548460272,
      "learning_rate": 1.5329403726080496e-05,
      "loss": 0.234,
      "step": 5337
    },
    {
      "epoch": 3.415227127319258,
      "grad_norm": 1.4934658371980754,
      "learning_rate": 1.5327650252782843e-05,
      "loss": 0.2249,
      "step": 5338
    },
    {
      "epoch": 3.415866922584773,
      "grad_norm": 1.478636742027393,
      "learning_rate": 1.5325896550725366e-05,
      "loss": 0.2005,
      "step": 5339
    },
    {
      "epoch": 3.4165067178502877,
      "grad_norm": 1.6150630705931752,
      "learning_rate": 1.532414261998337e-05,
      "loss": 0.2632,
      "step": 5340
    },
    {
      "epoch": 3.417146513115803,
      "grad_norm": 1.5508722018485546,
      "learning_rate": 1.5322388460632154e-05,
      "loss": 0.241,
      "step": 5341
    },
    {
      "epoch": 3.417786308381318,
      "grad_norm": 1.6851268526734389,
      "learning_rate": 1.5320634072747052e-05,
      "loss": 0.2619,
      "step": 5342
    },
    {
      "epoch": 3.418426103646833,
      "grad_norm": 1.536574508874555,
      "learning_rate": 1.5318879456403384e-05,
      "loss": 0.2431,
      "step": 5343
    },
    {
      "epoch": 3.419065898912348,
      "grad_norm": 1.5749230897964563,
      "learning_rate": 1.5317124611676493e-05,
      "loss": 0.2854,
      "step": 5344
    },
    {
      "epoch": 3.419705694177863,
      "grad_norm": 1.3769783662658572,
      "learning_rate": 1.531536953864173e-05,
      "loss": 0.2231,
      "step": 5345
    },
    {
      "epoch": 3.420345489443378,
      "grad_norm": 1.3820375228569146,
      "learning_rate": 1.5313614237374458e-05,
      "loss": 0.1972,
      "step": 5346
    },
    {
      "epoch": 3.4209852847088933,
      "grad_norm": 1.5141369975115395,
      "learning_rate": 1.531185870795004e-05,
      "loss": 0.2453,
      "step": 5347
    },
    {
      "epoch": 3.421625079974408,
      "grad_norm": 1.536382309277205,
      "learning_rate": 1.5310102950443856e-05,
      "loss": 0.2411,
      "step": 5348
    },
    {
      "epoch": 3.4222648752399234,
      "grad_norm": 1.6442211814250656,
      "learning_rate": 1.53083469649313e-05,
      "loss": 0.2698,
      "step": 5349
    },
    {
      "epoch": 3.4229046705054382,
      "grad_norm": 1.3912443120346,
      "learning_rate": 1.5306590751487766e-05,
      "loss": 0.2186,
      "step": 5350
    },
    {
      "epoch": 3.423544465770953,
      "grad_norm": 1.4803824866516884,
      "learning_rate": 1.530483431018867e-05,
      "loss": 0.2225,
      "step": 5351
    },
    {
      "epoch": 3.4241842610364683,
      "grad_norm": 1.4180328329203737,
      "learning_rate": 1.530307764110942e-05,
      "loss": 0.2267,
      "step": 5352
    },
    {
      "epoch": 3.4248240563019836,
      "grad_norm": 1.5734725112996382,
      "learning_rate": 1.530132074432545e-05,
      "loss": 0.2509,
      "step": 5353
    },
    {
      "epoch": 3.4254638515674984,
      "grad_norm": 1.499525039857896,
      "learning_rate": 1.52995636199122e-05,
      "loss": 0.2255,
      "step": 5354
    },
    {
      "epoch": 3.4261036468330133,
      "grad_norm": 1.396839838394877,
      "learning_rate": 1.5297806267945115e-05,
      "loss": 0.2273,
      "step": 5355
    },
    {
      "epoch": 3.4267434420985285,
      "grad_norm": 1.3966539695540805,
      "learning_rate": 1.529604868849965e-05,
      "loss": 0.2007,
      "step": 5356
    },
    {
      "epoch": 3.4273832373640434,
      "grad_norm": 1.4711922276913834,
      "learning_rate": 1.5294290881651278e-05,
      "loss": 0.239,
      "step": 5357
    },
    {
      "epoch": 3.4280230326295587,
      "grad_norm": 1.6393586441573935,
      "learning_rate": 1.5292532847475472e-05,
      "loss": 0.2537,
      "step": 5358
    },
    {
      "epoch": 3.4286628278950735,
      "grad_norm": 1.6058202810136206,
      "learning_rate": 1.5290774586047722e-05,
      "loss": 0.2606,
      "step": 5359
    },
    {
      "epoch": 3.4293026231605888,
      "grad_norm": 1.4705668157551728,
      "learning_rate": 1.528901609744352e-05,
      "loss": 0.2115,
      "step": 5360
    },
    {
      "epoch": 3.4299424184261036,
      "grad_norm": 1.5767741088717082,
      "learning_rate": 1.5287257381738377e-05,
      "loss": 0.2336,
      "step": 5361
    },
    {
      "epoch": 3.430582213691619,
      "grad_norm": 1.5133607054934233,
      "learning_rate": 1.5285498439007807e-05,
      "loss": 0.2171,
      "step": 5362
    },
    {
      "epoch": 3.4312220089571337,
      "grad_norm": 1.564968961876221,
      "learning_rate": 1.5283739269327337e-05,
      "loss": 0.2346,
      "step": 5363
    },
    {
      "epoch": 3.431861804222649,
      "grad_norm": 1.6285837794801383,
      "learning_rate": 1.5281979872772503e-05,
      "loss": 0.2464,
      "step": 5364
    },
    {
      "epoch": 3.432501599488164,
      "grad_norm": 1.5931699771075118,
      "learning_rate": 1.5280220249418848e-05,
      "loss": 0.2717,
      "step": 5365
    },
    {
      "epoch": 3.4331413947536786,
      "grad_norm": 1.479106917986524,
      "learning_rate": 1.5278460399341923e-05,
      "loss": 0.2419,
      "step": 5366
    },
    {
      "epoch": 3.433781190019194,
      "grad_norm": 1.5454801740126407,
      "learning_rate": 1.5276700322617305e-05,
      "loss": 0.2008,
      "step": 5367
    },
    {
      "epoch": 3.4344209852847087,
      "grad_norm": 1.444858098909204,
      "learning_rate": 1.527494001932056e-05,
      "loss": 0.2147,
      "step": 5368
    },
    {
      "epoch": 3.435060780550224,
      "grad_norm": 1.525634295253168,
      "learning_rate": 1.5273179489527273e-05,
      "loss": 0.2275,
      "step": 5369
    },
    {
      "epoch": 3.435700575815739,
      "grad_norm": 1.4934475239280742,
      "learning_rate": 1.527141873331304e-05,
      "loss": 0.2369,
      "step": 5370
    },
    {
      "epoch": 3.436340371081254,
      "grad_norm": 1.5638119154562171,
      "learning_rate": 1.5269657750753463e-05,
      "loss": 0.2289,
      "step": 5371
    },
    {
      "epoch": 3.436980166346769,
      "grad_norm": 1.476127747526377,
      "learning_rate": 1.5267896541924157e-05,
      "loss": 0.237,
      "step": 5372
    },
    {
      "epoch": 3.4376199616122842,
      "grad_norm": 1.5477491391890976,
      "learning_rate": 1.526613510690074e-05,
      "loss": 0.2356,
      "step": 5373
    },
    {
      "epoch": 3.438259756877799,
      "grad_norm": 1.489355349492908,
      "learning_rate": 1.5264373445758853e-05,
      "loss": 0.22,
      "step": 5374
    },
    {
      "epoch": 3.4388995521433143,
      "grad_norm": 1.3882869422675252,
      "learning_rate": 1.5262611558574138e-05,
      "loss": 0.2304,
      "step": 5375
    },
    {
      "epoch": 3.439539347408829,
      "grad_norm": 1.5785401833693051,
      "learning_rate": 1.5260849445422242e-05,
      "loss": 0.2325,
      "step": 5376
    },
    {
      "epoch": 3.440179142674344,
      "grad_norm": 1.538741519758727,
      "learning_rate": 1.5259087106378826e-05,
      "loss": 0.2577,
      "step": 5377
    },
    {
      "epoch": 3.4408189379398593,
      "grad_norm": 1.5152008026947554,
      "learning_rate": 1.5257324541519567e-05,
      "loss": 0.2298,
      "step": 5378
    },
    {
      "epoch": 3.441458733205374,
      "grad_norm": 1.547418518670601,
      "learning_rate": 1.5255561750920144e-05,
      "loss": 0.2617,
      "step": 5379
    },
    {
      "epoch": 3.4420985284708894,
      "grad_norm": 1.6028318860660982,
      "learning_rate": 1.5253798734656248e-05,
      "loss": 0.2374,
      "step": 5380
    },
    {
      "epoch": 3.442738323736404,
      "grad_norm": 1.7018051056416559,
      "learning_rate": 1.525203549280358e-05,
      "loss": 0.2254,
      "step": 5381
    },
    {
      "epoch": 3.4433781190019195,
      "grad_norm": 1.5161867477948303,
      "learning_rate": 1.5250272025437852e-05,
      "loss": 0.2005,
      "step": 5382
    },
    {
      "epoch": 3.4440179142674343,
      "grad_norm": 1.4884071399200518,
      "learning_rate": 1.5248508332634782e-05,
      "loss": 0.2198,
      "step": 5383
    },
    {
      "epoch": 3.4446577095329496,
      "grad_norm": 1.435528407041549,
      "learning_rate": 1.5246744414470099e-05,
      "loss": 0.2093,
      "step": 5384
    },
    {
      "epoch": 3.4452975047984644,
      "grad_norm": 1.5117047805112656,
      "learning_rate": 1.5244980271019548e-05,
      "loss": 0.2063,
      "step": 5385
    },
    {
      "epoch": 3.4459373000639797,
      "grad_norm": 1.5383390812875937,
      "learning_rate": 1.524321590235887e-05,
      "loss": 0.225,
      "step": 5386
    },
    {
      "epoch": 3.4465770953294945,
      "grad_norm": 1.4589135976410545,
      "learning_rate": 1.524145130856383e-05,
      "loss": 0.2275,
      "step": 5387
    },
    {
      "epoch": 3.4472168905950094,
      "grad_norm": 1.4231948407692285,
      "learning_rate": 1.5239686489710192e-05,
      "loss": 0.2316,
      "step": 5388
    },
    {
      "epoch": 3.4478566858605246,
      "grad_norm": 1.54560253059137,
      "learning_rate": 1.5237921445873742e-05,
      "loss": 0.2524,
      "step": 5389
    },
    {
      "epoch": 3.44849648112604,
      "grad_norm": 1.546776587585119,
      "learning_rate": 1.5236156177130259e-05,
      "loss": 0.2247,
      "step": 5390
    },
    {
      "epoch": 3.4491362763915547,
      "grad_norm": 1.4460086629790618,
      "learning_rate": 1.5234390683555546e-05,
      "loss": 0.2614,
      "step": 5391
    },
    {
      "epoch": 3.4497760716570696,
      "grad_norm": 1.5258296319094515,
      "learning_rate": 1.5232624965225408e-05,
      "loss": 0.2439,
      "step": 5392
    },
    {
      "epoch": 3.450415866922585,
      "grad_norm": 1.6370111119569342,
      "learning_rate": 1.5230859022215661e-05,
      "loss": 0.2568,
      "step": 5393
    },
    {
      "epoch": 3.4510556621880997,
      "grad_norm": 1.4267494419641795,
      "learning_rate": 1.5229092854602135e-05,
      "loss": 0.2203,
      "step": 5394
    },
    {
      "epoch": 3.451695457453615,
      "grad_norm": 1.479196775136737,
      "learning_rate": 1.522732646246066e-05,
      "loss": 0.2345,
      "step": 5395
    },
    {
      "epoch": 3.4523352527191298,
      "grad_norm": 1.4610170696019915,
      "learning_rate": 1.5225559845867086e-05,
      "loss": 0.2225,
      "step": 5396
    },
    {
      "epoch": 3.452975047984645,
      "grad_norm": 1.6148840451530129,
      "learning_rate": 1.5223793004897269e-05,
      "loss": 0.2666,
      "step": 5397
    },
    {
      "epoch": 3.45361484325016,
      "grad_norm": 1.6117440615177785,
      "learning_rate": 1.5222025939627071e-05,
      "loss": 0.2527,
      "step": 5398
    },
    {
      "epoch": 3.454254638515675,
      "grad_norm": 1.545267887559899,
      "learning_rate": 1.5220258650132374e-05,
      "loss": 0.2223,
      "step": 5399
    },
    {
      "epoch": 3.45489443378119,
      "grad_norm": 1.4736705144847817,
      "learning_rate": 1.5218491136489052e-05,
      "loss": 0.2183,
      "step": 5400
    },
    {
      "epoch": 3.4555342290467053,
      "grad_norm": 1.587285258575007,
      "learning_rate": 1.5216723398773004e-05,
      "loss": 0.2506,
      "step": 5401
    },
    {
      "epoch": 3.45617402431222,
      "grad_norm": 1.3233183122817844,
      "learning_rate": 1.5214955437060135e-05,
      "loss": 0.1952,
      "step": 5402
    },
    {
      "epoch": 3.456813819577735,
      "grad_norm": 1.5621083916527883,
      "learning_rate": 1.5213187251426353e-05,
      "loss": 0.2265,
      "step": 5403
    },
    {
      "epoch": 3.45745361484325,
      "grad_norm": 1.5251664823428508,
      "learning_rate": 1.5211418841947586e-05,
      "loss": 0.2525,
      "step": 5404
    },
    {
      "epoch": 3.458093410108765,
      "grad_norm": 1.5863937625836209,
      "learning_rate": 1.5209650208699763e-05,
      "loss": 0.2498,
      "step": 5405
    },
    {
      "epoch": 3.4587332053742803,
      "grad_norm": 1.7016403021803181,
      "learning_rate": 1.520788135175883e-05,
      "loss": 0.2686,
      "step": 5406
    },
    {
      "epoch": 3.459373000639795,
      "grad_norm": 1.5286298717275164,
      "learning_rate": 1.5206112271200732e-05,
      "loss": 0.2303,
      "step": 5407
    },
    {
      "epoch": 3.4600127959053104,
      "grad_norm": 1.422223051981674,
      "learning_rate": 1.5204342967101436e-05,
      "loss": 0.2128,
      "step": 5408
    },
    {
      "epoch": 3.4606525911708252,
      "grad_norm": 1.5010075828700025,
      "learning_rate": 1.5202573439536911e-05,
      "loss": 0.223,
      "step": 5409
    },
    {
      "epoch": 3.4612923864363405,
      "grad_norm": 1.661010560350895,
      "learning_rate": 1.5200803688583139e-05,
      "loss": 0.2537,
      "step": 5410
    },
    {
      "epoch": 3.4619321817018553,
      "grad_norm": 1.4710803881196606,
      "learning_rate": 1.5199033714316103e-05,
      "loss": 0.2443,
      "step": 5411
    },
    {
      "epoch": 3.4625719769673706,
      "grad_norm": 1.56736865019446,
      "learning_rate": 1.5197263516811811e-05,
      "loss": 0.2389,
      "step": 5412
    },
    {
      "epoch": 3.4632117722328855,
      "grad_norm": 1.5293434787532851,
      "learning_rate": 1.519549309614627e-05,
      "loss": 0.2197,
      "step": 5413
    },
    {
      "epoch": 3.4638515674984003,
      "grad_norm": 1.5164066570201606,
      "learning_rate": 1.5193722452395496e-05,
      "loss": 0.2253,
      "step": 5414
    },
    {
      "epoch": 3.4644913627639156,
      "grad_norm": 1.6255879440320058,
      "learning_rate": 1.519195158563552e-05,
      "loss": 0.2704,
      "step": 5415
    },
    {
      "epoch": 3.4651311580294304,
      "grad_norm": 1.6323985120058075,
      "learning_rate": 1.5190180495942377e-05,
      "loss": 0.2551,
      "step": 5416
    },
    {
      "epoch": 3.4657709532949457,
      "grad_norm": 1.5839471295642298,
      "learning_rate": 1.518840918339212e-05,
      "loss": 0.2276,
      "step": 5417
    },
    {
      "epoch": 3.4664107485604605,
      "grad_norm": 1.435327203648858,
      "learning_rate": 1.5186637648060796e-05,
      "loss": 0.2044,
      "step": 5418
    },
    {
      "epoch": 3.4670505438259758,
      "grad_norm": 1.585967883075093,
      "learning_rate": 1.518486589002448e-05,
      "loss": 0.2462,
      "step": 5419
    },
    {
      "epoch": 3.4676903390914906,
      "grad_norm": 1.5402258832379285,
      "learning_rate": 1.5183093909359246e-05,
      "loss": 0.2449,
      "step": 5420
    },
    {
      "epoch": 3.468330134357006,
      "grad_norm": 1.4780876576542252,
      "learning_rate": 1.518132170614118e-05,
      "loss": 0.2153,
      "step": 5421
    },
    {
      "epoch": 3.4689699296225207,
      "grad_norm": 1.5223502529656396,
      "learning_rate": 1.5179549280446378e-05,
      "loss": 0.2435,
      "step": 5422
    },
    {
      "epoch": 3.469609724888036,
      "grad_norm": 1.6201442772189802,
      "learning_rate": 1.5177776632350942e-05,
      "loss": 0.2601,
      "step": 5423
    },
    {
      "epoch": 3.470249520153551,
      "grad_norm": 1.5969661668988855,
      "learning_rate": 1.5176003761930988e-05,
      "loss": 0.2528,
      "step": 5424
    },
    {
      "epoch": 3.470889315419066,
      "grad_norm": 1.600715647821204,
      "learning_rate": 1.5174230669262643e-05,
      "loss": 0.2368,
      "step": 5425
    },
    {
      "epoch": 3.471529110684581,
      "grad_norm": 1.5017046661343851,
      "learning_rate": 1.5172457354422033e-05,
      "loss": 0.242,
      "step": 5426
    },
    {
      "epoch": 3.472168905950096,
      "grad_norm": 1.4209644984036351,
      "learning_rate": 1.5170683817485306e-05,
      "loss": 0.2249,
      "step": 5427
    },
    {
      "epoch": 3.472808701215611,
      "grad_norm": 1.3970607731766809,
      "learning_rate": 1.5168910058528612e-05,
      "loss": 0.2083,
      "step": 5428
    },
    {
      "epoch": 3.473448496481126,
      "grad_norm": 2.3712750841543,
      "learning_rate": 1.516713607762812e-05,
      "loss": 0.2113,
      "step": 5429
    },
    {
      "epoch": 3.474088291746641,
      "grad_norm": 1.391391720616244,
      "learning_rate": 1.5165361874859996e-05,
      "loss": 0.2119,
      "step": 5430
    },
    {
      "epoch": 3.474728087012156,
      "grad_norm": 1.5234786496153045,
      "learning_rate": 1.5163587450300417e-05,
      "loss": 0.2422,
      "step": 5431
    },
    {
      "epoch": 3.4753678822776712,
      "grad_norm": 1.472058384368976,
      "learning_rate": 1.5161812804025582e-05,
      "loss": 0.2432,
      "step": 5432
    },
    {
      "epoch": 3.476007677543186,
      "grad_norm": 1.4595019282162203,
      "learning_rate": 1.5160037936111684e-05,
      "loss": 0.2179,
      "step": 5433
    },
    {
      "epoch": 3.4766474728087013,
      "grad_norm": 1.458859699074123,
      "learning_rate": 1.515826284663494e-05,
      "loss": 0.2263,
      "step": 5434
    },
    {
      "epoch": 3.477287268074216,
      "grad_norm": 1.4992390776477869,
      "learning_rate": 1.5156487535671563e-05,
      "loss": 0.2342,
      "step": 5435
    },
    {
      "epoch": 3.4779270633397315,
      "grad_norm": 1.6122555373147103,
      "learning_rate": 1.5154712003297784e-05,
      "loss": 0.2387,
      "step": 5436
    },
    {
      "epoch": 3.4785668586052463,
      "grad_norm": 1.480128976603674,
      "learning_rate": 1.5152936249589842e-05,
      "loss": 0.2234,
      "step": 5437
    },
    {
      "epoch": 3.4792066538707616,
      "grad_norm": 1.500503461571419,
      "learning_rate": 1.5151160274623982e-05,
      "loss": 0.2275,
      "step": 5438
    },
    {
      "epoch": 3.4798464491362764,
      "grad_norm": 1.4368472553720317,
      "learning_rate": 1.5149384078476463e-05,
      "loss": 0.2201,
      "step": 5439
    },
    {
      "epoch": 3.480486244401791,
      "grad_norm": 1.525979483409279,
      "learning_rate": 1.5147607661223554e-05,
      "loss": 0.2276,
      "step": 5440
    },
    {
      "epoch": 3.4811260396673065,
      "grad_norm": 1.5347799540815834,
      "learning_rate": 1.5145831022941526e-05,
      "loss": 0.2516,
      "step": 5441
    },
    {
      "epoch": 3.4817658349328213,
      "grad_norm": 1.5709249629216946,
      "learning_rate": 1.514405416370667e-05,
      "loss": 0.2337,
      "step": 5442
    },
    {
      "epoch": 3.4824056301983366,
      "grad_norm": 1.602483965363487,
      "learning_rate": 1.5142277083595278e-05,
      "loss": 0.2427,
      "step": 5443
    },
    {
      "epoch": 3.4830454254638514,
      "grad_norm": 1.517883826912121,
      "learning_rate": 1.5140499782683655e-05,
      "loss": 0.2414,
      "step": 5444
    },
    {
      "epoch": 3.4836852207293667,
      "grad_norm": 1.4896903733191718,
      "learning_rate": 1.5138722261048117e-05,
      "loss": 0.2433,
      "step": 5445
    },
    {
      "epoch": 3.4843250159948815,
      "grad_norm": 1.4511529900380689,
      "learning_rate": 1.5136944518764987e-05,
      "loss": 0.2451,
      "step": 5446
    },
    {
      "epoch": 3.484964811260397,
      "grad_norm": 1.504868148984125,
      "learning_rate": 1.5135166555910594e-05,
      "loss": 0.2252,
      "step": 5447
    },
    {
      "epoch": 3.4856046065259116,
      "grad_norm": 1.515318677592483,
      "learning_rate": 1.5133388372561286e-05,
      "loss": 0.2055,
      "step": 5448
    },
    {
      "epoch": 3.486244401791427,
      "grad_norm": 1.512534880565101,
      "learning_rate": 1.5131609968793412e-05,
      "loss": 0.2298,
      "step": 5449
    },
    {
      "epoch": 3.4868841970569417,
      "grad_norm": 1.4594672992874727,
      "learning_rate": 1.5129831344683333e-05,
      "loss": 0.2287,
      "step": 5450
    },
    {
      "epoch": 3.4875239923224566,
      "grad_norm": 1.7663326092471316,
      "learning_rate": 1.5128052500307426e-05,
      "loss": 0.273,
      "step": 5451
    },
    {
      "epoch": 3.488163787587972,
      "grad_norm": 1.6485075195728394,
      "learning_rate": 1.5126273435742063e-05,
      "loss": 0.2746,
      "step": 5452
    },
    {
      "epoch": 3.488803582853487,
      "grad_norm": 1.5196407832380536,
      "learning_rate": 1.5124494151063638e-05,
      "loss": 0.2359,
      "step": 5453
    },
    {
      "epoch": 3.489443378119002,
      "grad_norm": 1.4586262333766735,
      "learning_rate": 1.5122714646348552e-05,
      "loss": 0.2299,
      "step": 5454
    },
    {
      "epoch": 3.490083173384517,
      "grad_norm": 1.6037094159005592,
      "learning_rate": 1.5120934921673207e-05,
      "loss": 0.2186,
      "step": 5455
    },
    {
      "epoch": 3.490722968650032,
      "grad_norm": 1.5461405775228514,
      "learning_rate": 1.5119154977114028e-05,
      "loss": 0.2473,
      "step": 5456
    },
    {
      "epoch": 3.491362763915547,
      "grad_norm": 1.522029699257012,
      "learning_rate": 1.5117374812747443e-05,
      "loss": 0.2286,
      "step": 5457
    },
    {
      "epoch": 3.492002559181062,
      "grad_norm": 1.4634005195432955,
      "learning_rate": 1.5115594428649885e-05,
      "loss": 0.2264,
      "step": 5458
    },
    {
      "epoch": 3.492642354446577,
      "grad_norm": 1.4215457168119248,
      "learning_rate": 1.5113813824897801e-05,
      "loss": 0.2242,
      "step": 5459
    },
    {
      "epoch": 3.4932821497120923,
      "grad_norm": 1.4579368046402745,
      "learning_rate": 1.511203300156765e-05,
      "loss": 0.2039,
      "step": 5460
    },
    {
      "epoch": 3.493921944977607,
      "grad_norm": 1.5921282341643832,
      "learning_rate": 1.5110251958735896e-05,
      "loss": 0.2553,
      "step": 5461
    },
    {
      "epoch": 3.4945617402431224,
      "grad_norm": 1.4639652856964538,
      "learning_rate": 1.5108470696479011e-05,
      "loss": 0.2347,
      "step": 5462
    },
    {
      "epoch": 3.495201535508637,
      "grad_norm": 1.5093745939201972,
      "learning_rate": 1.5106689214873482e-05,
      "loss": 0.2446,
      "step": 5463
    },
    {
      "epoch": 3.4958413307741525,
      "grad_norm": 1.6041606488105586,
      "learning_rate": 1.5104907513995804e-05,
      "loss": 0.2596,
      "step": 5464
    },
    {
      "epoch": 3.4964811260396673,
      "grad_norm": 1.5952290270602574,
      "learning_rate": 1.5103125593922477e-05,
      "loss": 0.2526,
      "step": 5465
    },
    {
      "epoch": 3.497120921305182,
      "grad_norm": 1.3946763680909915,
      "learning_rate": 1.5101343454730015e-05,
      "loss": 0.2162,
      "step": 5466
    },
    {
      "epoch": 3.4977607165706974,
      "grad_norm": 1.436132057737899,
      "learning_rate": 1.5099561096494939e-05,
      "loss": 0.2048,
      "step": 5467
    },
    {
      "epoch": 3.4984005118362123,
      "grad_norm": 1.5457095697901497,
      "learning_rate": 1.5097778519293782e-05,
      "loss": 0.2119,
      "step": 5468
    },
    {
      "epoch": 3.4990403071017275,
      "grad_norm": 1.5759621134818984,
      "learning_rate": 1.5095995723203084e-05,
      "loss": 0.246,
      "step": 5469
    },
    {
      "epoch": 3.4996801023672424,
      "grad_norm": 1.5330942039470166,
      "learning_rate": 1.5094212708299392e-05,
      "loss": 0.2728,
      "step": 5470
    },
    {
      "epoch": 3.5003198976327576,
      "grad_norm": 1.5092300851022884,
      "learning_rate": 1.5092429474659272e-05,
      "loss": 0.2264,
      "step": 5471
    },
    {
      "epoch": 3.5009596928982725,
      "grad_norm": 1.6500761286961119,
      "learning_rate": 1.5090646022359286e-05,
      "loss": 0.2729,
      "step": 5472
    },
    {
      "epoch": 3.5015994881637877,
      "grad_norm": 1.4126849744782324,
      "learning_rate": 1.5088862351476015e-05,
      "loss": 0.2457,
      "step": 5473
    },
    {
      "epoch": 3.5022392834293026,
      "grad_norm": 1.3870614507535697,
      "learning_rate": 1.508707846208605e-05,
      "loss": 0.2272,
      "step": 5474
    },
    {
      "epoch": 3.502879078694818,
      "grad_norm": 1.4000734470163108,
      "learning_rate": 1.5085294354265982e-05,
      "loss": 0.2332,
      "step": 5475
    },
    {
      "epoch": 3.5035188739603327,
      "grad_norm": 1.5166547207185122,
      "learning_rate": 1.5083510028092425e-05,
      "loss": 0.2573,
      "step": 5476
    },
    {
      "epoch": 3.5041586692258475,
      "grad_norm": 1.5293740812376926,
      "learning_rate": 1.5081725483641988e-05,
      "loss": 0.2673,
      "step": 5477
    },
    {
      "epoch": 3.504798464491363,
      "grad_norm": 1.535872760741501,
      "learning_rate": 1.5079940720991296e-05,
      "loss": 0.2348,
      "step": 5478
    },
    {
      "epoch": 3.505438259756878,
      "grad_norm": 1.4024789878207555,
      "learning_rate": 1.5078155740216986e-05,
      "loss": 0.2235,
      "step": 5479
    },
    {
      "epoch": 3.506078055022393,
      "grad_norm": 1.4466818847198437,
      "learning_rate": 1.5076370541395704e-05,
      "loss": 0.2334,
      "step": 5480
    },
    {
      "epoch": 3.5067178502879077,
      "grad_norm": 1.6554335967694553,
      "learning_rate": 1.5074585124604102e-05,
      "loss": 0.2583,
      "step": 5481
    },
    {
      "epoch": 3.507357645553423,
      "grad_norm": 1.3813450185243559,
      "learning_rate": 1.5072799489918839e-05,
      "loss": 0.1905,
      "step": 5482
    },
    {
      "epoch": 3.507997440818938,
      "grad_norm": 1.4930760374838494,
      "learning_rate": 1.5071013637416592e-05,
      "loss": 0.2285,
      "step": 5483
    },
    {
      "epoch": 3.508637236084453,
      "grad_norm": 1.6033494408336009,
      "learning_rate": 1.506922756717404e-05,
      "loss": 0.2445,
      "step": 5484
    },
    {
      "epoch": 3.509277031349968,
      "grad_norm": 1.58227513954258,
      "learning_rate": 1.5067441279267872e-05,
      "loss": 0.253,
      "step": 5485
    },
    {
      "epoch": 3.509916826615483,
      "grad_norm": 1.5888767220353555,
      "learning_rate": 1.5065654773774792e-05,
      "loss": 0.2731,
      "step": 5486
    },
    {
      "epoch": 3.510556621880998,
      "grad_norm": 1.587421025172185,
      "learning_rate": 1.5063868050771505e-05,
      "loss": 0.2451,
      "step": 5487
    },
    {
      "epoch": 3.511196417146513,
      "grad_norm": 1.5783486608537178,
      "learning_rate": 1.506208111033473e-05,
      "loss": 0.2091,
      "step": 5488
    },
    {
      "epoch": 3.511836212412028,
      "grad_norm": 1.4416679072906535,
      "learning_rate": 1.50602939525412e-05,
      "loss": 0.2315,
      "step": 5489
    },
    {
      "epoch": 3.5124760076775434,
      "grad_norm": 1.49659004172779,
      "learning_rate": 1.5058506577467649e-05,
      "loss": 0.2419,
      "step": 5490
    },
    {
      "epoch": 3.5131158029430583,
      "grad_norm": 1.4145144386186996,
      "learning_rate": 1.5056718985190826e-05,
      "loss": 0.2453,
      "step": 5491
    },
    {
      "epoch": 3.513755598208573,
      "grad_norm": 1.413092682030215,
      "learning_rate": 1.5054931175787482e-05,
      "loss": 0.2221,
      "step": 5492
    },
    {
      "epoch": 3.5143953934740884,
      "grad_norm": 1.3715198304815466,
      "learning_rate": 1.5053143149334385e-05,
      "loss": 0.2027,
      "step": 5493
    },
    {
      "epoch": 3.515035188739603,
      "grad_norm": 1.4924508855280543,
      "learning_rate": 1.505135490590831e-05,
      "loss": 0.2545,
      "step": 5494
    },
    {
      "epoch": 3.5156749840051185,
      "grad_norm": 1.482714584428029,
      "learning_rate": 1.5049566445586042e-05,
      "loss": 0.2294,
      "step": 5495
    },
    {
      "epoch": 3.5163147792706333,
      "grad_norm": 1.4459951344077768,
      "learning_rate": 1.5047777768444374e-05,
      "loss": 0.2047,
      "step": 5496
    },
    {
      "epoch": 3.5169545745361486,
      "grad_norm": 1.5040468537314802,
      "learning_rate": 1.5045988874560107e-05,
      "loss": 0.2237,
      "step": 5497
    },
    {
      "epoch": 3.5175943698016634,
      "grad_norm": 1.51033935813243,
      "learning_rate": 1.5044199764010054e-05,
      "loss": 0.2444,
      "step": 5498
    },
    {
      "epoch": 3.5182341650671782,
      "grad_norm": 1.4989827219480345,
      "learning_rate": 1.5042410436871038e-05,
      "loss": 0.2207,
      "step": 5499
    },
    {
      "epoch": 3.5188739603326935,
      "grad_norm": 1.6023361911497465,
      "learning_rate": 1.5040620893219885e-05,
      "loss": 0.2003,
      "step": 5500
    },
    {
      "epoch": 3.519513755598209,
      "grad_norm": 1.5405657807935371,
      "learning_rate": 1.5038831133133437e-05,
      "loss": 0.2307,
      "step": 5501
    },
    {
      "epoch": 3.5201535508637236,
      "grad_norm": 1.472139637977917,
      "learning_rate": 1.5037041156688543e-05,
      "loss": 0.2075,
      "step": 5502
    },
    {
      "epoch": 3.5207933461292384,
      "grad_norm": 1.5457772397572052,
      "learning_rate": 1.5035250963962064e-05,
      "loss": 0.2247,
      "step": 5503
    },
    {
      "epoch": 3.5214331413947537,
      "grad_norm": 1.5428850127461324,
      "learning_rate": 1.5033460555030866e-05,
      "loss": 0.2375,
      "step": 5504
    },
    {
      "epoch": 3.5220729366602685,
      "grad_norm": 1.6864730523863718,
      "learning_rate": 1.5031669929971826e-05,
      "loss": 0.2412,
      "step": 5505
    },
    {
      "epoch": 3.522712731925784,
      "grad_norm": 1.5101451772861456,
      "learning_rate": 1.502987908886183e-05,
      "loss": 0.22,
      "step": 5506
    },
    {
      "epoch": 3.5233525271912987,
      "grad_norm": 1.4987770822905184,
      "learning_rate": 1.5028088031777767e-05,
      "loss": 0.2324,
      "step": 5507
    },
    {
      "epoch": 3.523992322456814,
      "grad_norm": 1.5608450291421743,
      "learning_rate": 1.5026296758796554e-05,
      "loss": 0.2683,
      "step": 5508
    },
    {
      "epoch": 3.5246321177223288,
      "grad_norm": 1.5265829292689315,
      "learning_rate": 1.50245052699951e-05,
      "loss": 0.2355,
      "step": 5509
    },
    {
      "epoch": 3.525271912987844,
      "grad_norm": 1.415191467342353,
      "learning_rate": 1.5022713565450324e-05,
      "loss": 0.1992,
      "step": 5510
    },
    {
      "epoch": 3.525911708253359,
      "grad_norm": 1.576869242079849,
      "learning_rate": 1.5020921645239167e-05,
      "loss": 0.2602,
      "step": 5511
    },
    {
      "epoch": 3.526551503518874,
      "grad_norm": 1.4995155439717431,
      "learning_rate": 1.5019129509438562e-05,
      "loss": 0.2178,
      "step": 5512
    },
    {
      "epoch": 3.527191298784389,
      "grad_norm": 1.5796095336001397,
      "learning_rate": 1.5017337158125465e-05,
      "loss": 0.2414,
      "step": 5513
    },
    {
      "epoch": 3.527831094049904,
      "grad_norm": 1.4156689952032429,
      "learning_rate": 1.5015544591376836e-05,
      "loss": 0.2074,
      "step": 5514
    },
    {
      "epoch": 3.528470889315419,
      "grad_norm": 1.5291118714124143,
      "learning_rate": 1.5013751809269646e-05,
      "loss": 0.23,
      "step": 5515
    },
    {
      "epoch": 3.5291106845809344,
      "grad_norm": 1.571072590407857,
      "learning_rate": 1.5011958811880871e-05,
      "loss": 0.2471,
      "step": 5516
    },
    {
      "epoch": 3.529750479846449,
      "grad_norm": 1.5247661890947528,
      "learning_rate": 1.5010165599287496e-05,
      "loss": 0.2396,
      "step": 5517
    },
    {
      "epoch": 3.530390275111964,
      "grad_norm": 1.5604383567025892,
      "learning_rate": 1.5008372171566525e-05,
      "loss": 0.2338,
      "step": 5518
    },
    {
      "epoch": 3.5310300703774793,
      "grad_norm": 1.4715897447543307,
      "learning_rate": 1.5006578528794965e-05,
      "loss": 0.2284,
      "step": 5519
    },
    {
      "epoch": 3.531669865642994,
      "grad_norm": 1.4819191137639984,
      "learning_rate": 1.500478467104983e-05,
      "loss": 0.2595,
      "step": 5520
    },
    {
      "epoch": 3.5323096609085094,
      "grad_norm": 1.434166554557028,
      "learning_rate": 1.5002990598408143e-05,
      "loss": 0.2247,
      "step": 5521
    },
    {
      "epoch": 3.5329494561740242,
      "grad_norm": 1.623295593857305,
      "learning_rate": 1.5001196310946935e-05,
      "loss": 0.2435,
      "step": 5522
    },
    {
      "epoch": 3.5335892514395395,
      "grad_norm": 1.4451352603717826,
      "learning_rate": 1.4999401808743255e-05,
      "loss": 0.2159,
      "step": 5523
    },
    {
      "epoch": 3.5342290467050543,
      "grad_norm": 1.3614577750089234,
      "learning_rate": 1.4997607091874158e-05,
      "loss": 0.2206,
      "step": 5524
    },
    {
      "epoch": 3.534868841970569,
      "grad_norm": 1.5317339606562261,
      "learning_rate": 1.4995812160416701e-05,
      "loss": 0.2424,
      "step": 5525
    },
    {
      "epoch": 3.5355086372360844,
      "grad_norm": 1.5936664942985956,
      "learning_rate": 1.4994017014447955e-05,
      "loss": 0.2399,
      "step": 5526
    },
    {
      "epoch": 3.5361484325015997,
      "grad_norm": 1.452455839172452,
      "learning_rate": 1.4992221654045006e-05,
      "loss": 0.2112,
      "step": 5527
    },
    {
      "epoch": 3.5367882277671145,
      "grad_norm": 1.4889135241772051,
      "learning_rate": 1.4990426079284934e-05,
      "loss": 0.2428,
      "step": 5528
    },
    {
      "epoch": 3.5374280230326294,
      "grad_norm": 1.556814754940809,
      "learning_rate": 1.4988630290244848e-05,
      "loss": 0.2187,
      "step": 5529
    },
    {
      "epoch": 3.5380678182981447,
      "grad_norm": 1.3540826736494942,
      "learning_rate": 1.498683428700185e-05,
      "loss": 0.2082,
      "step": 5530
    },
    {
      "epoch": 3.5387076135636595,
      "grad_norm": 1.5657932045448877,
      "learning_rate": 1.498503806963306e-05,
      "loss": 0.2217,
      "step": 5531
    },
    {
      "epoch": 3.5393474088291748,
      "grad_norm": 1.5873870551605822,
      "learning_rate": 1.49832416382156e-05,
      "loss": 0.2263,
      "step": 5532
    },
    {
      "epoch": 3.5399872040946896,
      "grad_norm": 1.5455881886998128,
      "learning_rate": 1.4981444992826608e-05,
      "loss": 0.2242,
      "step": 5533
    },
    {
      "epoch": 3.540626999360205,
      "grad_norm": 1.5988671227008833,
      "learning_rate": 1.4979648133543232e-05,
      "loss": 0.2453,
      "step": 5534
    },
    {
      "epoch": 3.5412667946257197,
      "grad_norm": 1.398420474100746,
      "learning_rate": 1.4977851060442623e-05,
      "loss": 0.2279,
      "step": 5535
    },
    {
      "epoch": 3.541906589891235,
      "grad_norm": 1.5015545101853693,
      "learning_rate": 1.4976053773601947e-05,
      "loss": 0.2337,
      "step": 5536
    },
    {
      "epoch": 3.54254638515675,
      "grad_norm": 1.4320310281788748,
      "learning_rate": 1.4974256273098369e-05,
      "loss": 0.2206,
      "step": 5537
    },
    {
      "epoch": 3.543186180422265,
      "grad_norm": 1.5445033245073247,
      "learning_rate": 1.4972458559009078e-05,
      "loss": 0.2522,
      "step": 5538
    },
    {
      "epoch": 3.54382597568778,
      "grad_norm": 1.4625880438382417,
      "learning_rate": 1.497066063141126e-05,
      "loss": 0.2324,
      "step": 5539
    },
    {
      "epoch": 3.5444657709532947,
      "grad_norm": 1.377182541507007,
      "learning_rate": 1.4968862490382114e-05,
      "loss": 0.2113,
      "step": 5540
    },
    {
      "epoch": 3.54510556621881,
      "grad_norm": 1.5324041596904823,
      "learning_rate": 1.4967064135998857e-05,
      "loss": 0.2295,
      "step": 5541
    },
    {
      "epoch": 3.5457453614843253,
      "grad_norm": 1.5812498473209282,
      "learning_rate": 1.49652655683387e-05,
      "loss": 0.2518,
      "step": 5542
    },
    {
      "epoch": 3.54638515674984,
      "grad_norm": 1.5189112164076977,
      "learning_rate": 1.4963466787478872e-05,
      "loss": 0.2325,
      "step": 5543
    },
    {
      "epoch": 3.547024952015355,
      "grad_norm": 1.5055032269177386,
      "learning_rate": 1.4961667793496608e-05,
      "loss": 0.2367,
      "step": 5544
    },
    {
      "epoch": 3.5476647472808702,
      "grad_norm": 1.4713349159768163,
      "learning_rate": 1.4959868586469154e-05,
      "loss": 0.2402,
      "step": 5545
    },
    {
      "epoch": 3.548304542546385,
      "grad_norm": 1.544907882708599,
      "learning_rate": 1.495806916647377e-05,
      "loss": 0.257,
      "step": 5546
    },
    {
      "epoch": 3.5489443378119003,
      "grad_norm": 1.495751916065043,
      "learning_rate": 1.4956269533587714e-05,
      "loss": 0.2224,
      "step": 5547
    },
    {
      "epoch": 3.549584133077415,
      "grad_norm": 1.3945411711390365,
      "learning_rate": 1.495446968788826e-05,
      "loss": 0.2146,
      "step": 5548
    },
    {
      "epoch": 3.5502239283429304,
      "grad_norm": 1.6991597950837654,
      "learning_rate": 1.4952669629452689e-05,
      "loss": 0.2751,
      "step": 5549
    },
    {
      "epoch": 3.5508637236084453,
      "grad_norm": 1.431071570537036,
      "learning_rate": 1.49508693583583e-05,
      "loss": 0.2227,
      "step": 5550
    },
    {
      "epoch": 3.55150351887396,
      "grad_norm": 1.5438479533095577,
      "learning_rate": 1.4949068874682384e-05,
      "loss": 0.2393,
      "step": 5551
    },
    {
      "epoch": 3.5521433141394754,
      "grad_norm": 1.431840271407325,
      "learning_rate": 1.4947268178502254e-05,
      "loss": 0.2163,
      "step": 5552
    },
    {
      "epoch": 3.5527831094049906,
      "grad_norm": 1.5147115561531665,
      "learning_rate": 1.494546726989523e-05,
      "loss": 0.2646,
      "step": 5553
    },
    {
      "epoch": 3.5534229046705055,
      "grad_norm": 1.4089613078255514,
      "learning_rate": 1.4943666148938636e-05,
      "loss": 0.2369,
      "step": 5554
    },
    {
      "epoch": 3.5540626999360203,
      "grad_norm": 1.4721316311776818,
      "learning_rate": 1.4941864815709814e-05,
      "loss": 0.2108,
      "step": 5555
    },
    {
      "epoch": 3.5547024952015356,
      "grad_norm": 1.56237942285916,
      "learning_rate": 1.4940063270286108e-05,
      "loss": 0.2383,
      "step": 5556
    },
    {
      "epoch": 3.5553422904670504,
      "grad_norm": 1.5442018672589517,
      "learning_rate": 1.4938261512744871e-05,
      "loss": 0.225,
      "step": 5557
    },
    {
      "epoch": 3.5559820857325657,
      "grad_norm": 1.5082931284718926,
      "learning_rate": 1.4936459543163473e-05,
      "loss": 0.2553,
      "step": 5558
    },
    {
      "epoch": 3.5566218809980805,
      "grad_norm": 1.595205609271742,
      "learning_rate": 1.493465736161928e-05,
      "loss": 0.2655,
      "step": 5559
    },
    {
      "epoch": 3.557261676263596,
      "grad_norm": 1.42086190191614,
      "learning_rate": 1.4932854968189679e-05,
      "loss": 0.2291,
      "step": 5560
    },
    {
      "epoch": 3.5579014715291106,
      "grad_norm": 1.4969188846250459,
      "learning_rate": 1.4931052362952061e-05,
      "loss": 0.2469,
      "step": 5561
    },
    {
      "epoch": 3.5585412667946255,
      "grad_norm": 1.6306345726987572,
      "learning_rate": 1.4929249545983824e-05,
      "loss": 0.2658,
      "step": 5562
    },
    {
      "epoch": 3.5591810620601407,
      "grad_norm": 1.4717189932695167,
      "learning_rate": 1.4927446517362382e-05,
      "loss": 0.241,
      "step": 5563
    },
    {
      "epoch": 3.559820857325656,
      "grad_norm": 1.5122006720419245,
      "learning_rate": 1.4925643277165151e-05,
      "loss": 0.2185,
      "step": 5564
    },
    {
      "epoch": 3.560460652591171,
      "grad_norm": 1.541532158482337,
      "learning_rate": 1.4923839825469561e-05,
      "loss": 0.2787,
      "step": 5565
    },
    {
      "epoch": 3.5611004478566857,
      "grad_norm": 1.5035996700545,
      "learning_rate": 1.4922036162353047e-05,
      "loss": 0.2305,
      "step": 5566
    },
    {
      "epoch": 3.561740243122201,
      "grad_norm": 1.5458840641043718,
      "learning_rate": 1.4920232287893055e-05,
      "loss": 0.211,
      "step": 5567
    },
    {
      "epoch": 3.5623800383877158,
      "grad_norm": 1.444776238529624,
      "learning_rate": 1.4918428202167041e-05,
      "loss": 0.2218,
      "step": 5568
    },
    {
      "epoch": 3.563019833653231,
      "grad_norm": 1.6081857021705617,
      "learning_rate": 1.491662390525247e-05,
      "loss": 0.2723,
      "step": 5569
    },
    {
      "epoch": 3.563659628918746,
      "grad_norm": 1.4989328160860316,
      "learning_rate": 1.4914819397226814e-05,
      "loss": 0.2327,
      "step": 5570
    },
    {
      "epoch": 3.564299424184261,
      "grad_norm": 1.5458479622015757,
      "learning_rate": 1.4913014678167557e-05,
      "loss": 0.2481,
      "step": 5571
    },
    {
      "epoch": 3.564939219449776,
      "grad_norm": 1.6020306246646348,
      "learning_rate": 1.491120974815219e-05,
      "loss": 0.2388,
      "step": 5572
    },
    {
      "epoch": 3.5655790147152913,
      "grad_norm": 1.688129603874725,
      "learning_rate": 1.490940460725821e-05,
      "loss": 0.2889,
      "step": 5573
    },
    {
      "epoch": 3.566218809980806,
      "grad_norm": 1.5056682594059398,
      "learning_rate": 1.490759925556313e-05,
      "loss": 0.2346,
      "step": 5574
    },
    {
      "epoch": 3.5668586052463214,
      "grad_norm": 1.4445186034701327,
      "learning_rate": 1.4905793693144469e-05,
      "loss": 0.2137,
      "step": 5575
    },
    {
      "epoch": 3.567498400511836,
      "grad_norm": 1.5915399168012567,
      "learning_rate": 1.4903987920079752e-05,
      "loss": 0.2322,
      "step": 5576
    },
    {
      "epoch": 3.568138195777351,
      "grad_norm": 1.4871480663997783,
      "learning_rate": 1.4902181936446518e-05,
      "loss": 0.2389,
      "step": 5577
    },
    {
      "epoch": 3.5687779910428663,
      "grad_norm": 1.44721415608641,
      "learning_rate": 1.4900375742322313e-05,
      "loss": 0.2202,
      "step": 5578
    },
    {
      "epoch": 3.5694177863083816,
      "grad_norm": 1.5173011626756137,
      "learning_rate": 1.489856933778469e-05,
      "loss": 0.2151,
      "step": 5579
    },
    {
      "epoch": 3.5700575815738964,
      "grad_norm": 1.41812278125269,
      "learning_rate": 1.4896762722911213e-05,
      "loss": 0.2359,
      "step": 5580
    },
    {
      "epoch": 3.5706973768394112,
      "grad_norm": 1.6796664833623118,
      "learning_rate": 1.4894955897779458e-05,
      "loss": 0.2407,
      "step": 5581
    },
    {
      "epoch": 3.5713371721049265,
      "grad_norm": 1.4531713523277332,
      "learning_rate": 1.4893148862467e-05,
      "loss": 0.2268,
      "step": 5582
    },
    {
      "epoch": 3.5719769673704413,
      "grad_norm": 1.4682914349238743,
      "learning_rate": 1.4891341617051435e-05,
      "loss": 0.2451,
      "step": 5583
    },
    {
      "epoch": 3.5726167626359566,
      "grad_norm": 1.5531621061554155,
      "learning_rate": 1.4889534161610364e-05,
      "loss": 0.2335,
      "step": 5584
    },
    {
      "epoch": 3.5732565579014715,
      "grad_norm": 1.4028298763156588,
      "learning_rate": 1.4887726496221393e-05,
      "loss": 0.2069,
      "step": 5585
    },
    {
      "epoch": 3.5738963531669867,
      "grad_norm": 1.592812861061851,
      "learning_rate": 1.488591862096214e-05,
      "loss": 0.2597,
      "step": 5586
    },
    {
      "epoch": 3.5745361484325016,
      "grad_norm": 1.547185342715031,
      "learning_rate": 1.4884110535910235e-05,
      "loss": 0.2361,
      "step": 5587
    },
    {
      "epoch": 3.5751759436980164,
      "grad_norm": 1.4688526033751783,
      "learning_rate": 1.4882302241143311e-05,
      "loss": 0.2201,
      "step": 5588
    },
    {
      "epoch": 3.5758157389635317,
      "grad_norm": 1.4726630861626653,
      "learning_rate": 1.4880493736739012e-05,
      "loss": 0.2263,
      "step": 5589
    },
    {
      "epoch": 3.576455534229047,
      "grad_norm": 1.515898711057944,
      "learning_rate": 1.4878685022774994e-05,
      "loss": 0.2202,
      "step": 5590
    },
    {
      "epoch": 3.5770953294945618,
      "grad_norm": 1.3055872025673059,
      "learning_rate": 1.487687609932892e-05,
      "loss": 0.196,
      "step": 5591
    },
    {
      "epoch": 3.5777351247600766,
      "grad_norm": 1.3816336387673773,
      "learning_rate": 1.4875066966478462e-05,
      "loss": 0.2077,
      "step": 5592
    },
    {
      "epoch": 3.578374920025592,
      "grad_norm": 1.4186498088059238,
      "learning_rate": 1.48732576243013e-05,
      "loss": 0.2332,
      "step": 5593
    },
    {
      "epoch": 3.5790147152911067,
      "grad_norm": 1.444259336092459,
      "learning_rate": 1.4871448072875126e-05,
      "loss": 0.2392,
      "step": 5594
    },
    {
      "epoch": 3.579654510556622,
      "grad_norm": 1.3739339899844671,
      "learning_rate": 1.4869638312277637e-05,
      "loss": 0.2122,
      "step": 5595
    },
    {
      "epoch": 3.580294305822137,
      "grad_norm": 1.4570252572516744,
      "learning_rate": 1.4867828342586541e-05,
      "loss": 0.2384,
      "step": 5596
    },
    {
      "epoch": 3.580934101087652,
      "grad_norm": 1.5760969023980642,
      "learning_rate": 1.4866018163879554e-05,
      "loss": 0.2487,
      "step": 5597
    },
    {
      "epoch": 3.581573896353167,
      "grad_norm": 1.6589482872557786,
      "learning_rate": 1.4864207776234405e-05,
      "loss": 0.2549,
      "step": 5598
    },
    {
      "epoch": 3.5822136916186818,
      "grad_norm": 1.5560126809401242,
      "learning_rate": 1.4862397179728825e-05,
      "loss": 0.2417,
      "step": 5599
    },
    {
      "epoch": 3.582853486884197,
      "grad_norm": 1.4835260201311466,
      "learning_rate": 1.4860586374440558e-05,
      "loss": 0.2396,
      "step": 5600
    },
    {
      "epoch": 3.5834932821497123,
      "grad_norm": 1.4229306739153684,
      "learning_rate": 1.4858775360447362e-05,
      "loss": 0.229,
      "step": 5601
    },
    {
      "epoch": 3.584133077415227,
      "grad_norm": 1.54842367871386,
      "learning_rate": 1.4856964137826994e-05,
      "loss": 0.2267,
      "step": 5602
    },
    {
      "epoch": 3.584772872680742,
      "grad_norm": 1.3138053392346107,
      "learning_rate": 1.4855152706657228e-05,
      "loss": 0.1918,
      "step": 5603
    },
    {
      "epoch": 3.5854126679462572,
      "grad_norm": 1.4423497783762431,
      "learning_rate": 1.4853341067015837e-05,
      "loss": 0.2118,
      "step": 5604
    },
    {
      "epoch": 3.586052463211772,
      "grad_norm": 1.4700655443966864,
      "learning_rate": 1.4851529218980613e-05,
      "loss": 0.2464,
      "step": 5605
    },
    {
      "epoch": 3.5866922584772873,
      "grad_norm": 1.573050894888458,
      "learning_rate": 1.484971716262936e-05,
      "loss": 0.2405,
      "step": 5606
    },
    {
      "epoch": 3.587332053742802,
      "grad_norm": 1.5617677703601145,
      "learning_rate": 1.4847904898039875e-05,
      "loss": 0.2515,
      "step": 5607
    },
    {
      "epoch": 3.5879718490083174,
      "grad_norm": 1.5175040213002209,
      "learning_rate": 1.4846092425289983e-05,
      "loss": 0.242,
      "step": 5608
    },
    {
      "epoch": 3.5886116442738323,
      "grad_norm": 1.4960384753757086,
      "learning_rate": 1.4844279744457495e-05,
      "loss": 0.241,
      "step": 5609
    },
    {
      "epoch": 3.5892514395393476,
      "grad_norm": 1.5232150867539855,
      "learning_rate": 1.4842466855620256e-05,
      "loss": 0.2317,
      "step": 5610
    },
    {
      "epoch": 3.5898912348048624,
      "grad_norm": 1.6030879026908351,
      "learning_rate": 1.4840653758856106e-05,
      "loss": 0.2163,
      "step": 5611
    },
    {
      "epoch": 3.5905310300703777,
      "grad_norm": 1.489208615641775,
      "learning_rate": 1.483884045424289e-05,
      "loss": 0.2317,
      "step": 5612
    },
    {
      "epoch": 3.5911708253358925,
      "grad_norm": 1.5538349395012983,
      "learning_rate": 1.4837026941858477e-05,
      "loss": 0.2414,
      "step": 5613
    },
    {
      "epoch": 3.5918106206014073,
      "grad_norm": 1.469532870735545,
      "learning_rate": 1.483521322178073e-05,
      "loss": 0.2379,
      "step": 5614
    },
    {
      "epoch": 3.5924504158669226,
      "grad_norm": 1.4527339746443966,
      "learning_rate": 1.4833399294087527e-05,
      "loss": 0.2215,
      "step": 5615
    },
    {
      "epoch": 3.593090211132438,
      "grad_norm": 1.4494033916552704,
      "learning_rate": 1.4831585158856756e-05,
      "loss": 0.2262,
      "step": 5616
    },
    {
      "epoch": 3.5937300063979527,
      "grad_norm": 1.5695497311762523,
      "learning_rate": 1.4829770816166317e-05,
      "loss": 0.2796,
      "step": 5617
    },
    {
      "epoch": 3.5943698016634675,
      "grad_norm": 1.4864971639769342,
      "learning_rate": 1.4827956266094108e-05,
      "loss": 0.269,
      "step": 5618
    },
    {
      "epoch": 3.595009596928983,
      "grad_norm": 1.477483252955438,
      "learning_rate": 1.4826141508718043e-05,
      "loss": 0.2402,
      "step": 5619
    },
    {
      "epoch": 3.5956493921944976,
      "grad_norm": 1.637834483739395,
      "learning_rate": 1.4824326544116051e-05,
      "loss": 0.2394,
      "step": 5620
    },
    {
      "epoch": 3.596289187460013,
      "grad_norm": 1.5585915829370756,
      "learning_rate": 1.4822511372366056e-05,
      "loss": 0.2333,
      "step": 5621
    },
    {
      "epoch": 3.5969289827255277,
      "grad_norm": 1.4224022188033,
      "learning_rate": 1.4820695993546001e-05,
      "loss": 0.2082,
      "step": 5622
    },
    {
      "epoch": 3.597568777991043,
      "grad_norm": 1.4410368569121537,
      "learning_rate": 1.4818880407733837e-05,
      "loss": 0.2294,
      "step": 5623
    },
    {
      "epoch": 3.598208573256558,
      "grad_norm": 1.4683339603619605,
      "learning_rate": 1.4817064615007522e-05,
      "loss": 0.2002,
      "step": 5624
    },
    {
      "epoch": 3.5988483685220727,
      "grad_norm": 1.4987526049498021,
      "learning_rate": 1.481524861544502e-05,
      "loss": 0.2473,
      "step": 5625
    },
    {
      "epoch": 3.599488163787588,
      "grad_norm": 1.515748361411373,
      "learning_rate": 1.4813432409124307e-05,
      "loss": 0.2225,
      "step": 5626
    },
    {
      "epoch": 3.6001279590531032,
      "grad_norm": 1.4689671289648267,
      "learning_rate": 1.4811615996123367e-05,
      "loss": 0.2309,
      "step": 5627
    },
    {
      "epoch": 3.600767754318618,
      "grad_norm": 1.4373204220762945,
      "learning_rate": 1.4809799376520198e-05,
      "loss": 0.2305,
      "step": 5628
    },
    {
      "epoch": 3.601407549584133,
      "grad_norm": 1.569070672793268,
      "learning_rate": 1.4807982550392799e-05,
      "loss": 0.2418,
      "step": 5629
    },
    {
      "epoch": 3.602047344849648,
      "grad_norm": 1.4250933175475835,
      "learning_rate": 1.4806165517819182e-05,
      "loss": 0.2403,
      "step": 5630
    },
    {
      "epoch": 3.602687140115163,
      "grad_norm": 1.3338111631069496,
      "learning_rate": 1.4804348278877368e-05,
      "loss": 0.1899,
      "step": 5631
    },
    {
      "epoch": 3.6033269353806783,
      "grad_norm": 1.4424374828650353,
      "learning_rate": 1.4802530833645383e-05,
      "loss": 0.225,
      "step": 5632
    },
    {
      "epoch": 3.603966730646193,
      "grad_norm": 1.5546248426268567,
      "learning_rate": 1.4800713182201269e-05,
      "loss": 0.238,
      "step": 5633
    },
    {
      "epoch": 3.6046065259117084,
      "grad_norm": 1.4259241234069342,
      "learning_rate": 1.4798895324623071e-05,
      "loss": 0.2481,
      "step": 5634
    },
    {
      "epoch": 3.605246321177223,
      "grad_norm": 1.4964224465664377,
      "learning_rate": 1.4797077260988845e-05,
      "loss": 0.2321,
      "step": 5635
    },
    {
      "epoch": 3.6058861164427385,
      "grad_norm": 1.5383937740912386,
      "learning_rate": 1.479525899137665e-05,
      "loss": 0.2336,
      "step": 5636
    },
    {
      "epoch": 3.6065259117082533,
      "grad_norm": 1.514280099237753,
      "learning_rate": 1.4793440515864568e-05,
      "loss": 0.2469,
      "step": 5637
    },
    {
      "epoch": 3.6071657069737686,
      "grad_norm": 1.4080343180212078,
      "learning_rate": 1.4791621834530678e-05,
      "loss": 0.2373,
      "step": 5638
    },
    {
      "epoch": 3.6078055022392834,
      "grad_norm": 1.5520376611025923,
      "learning_rate": 1.478980294745307e-05,
      "loss": 0.2624,
      "step": 5639
    },
    {
      "epoch": 3.6084452975047983,
      "grad_norm": 1.530831976380922,
      "learning_rate": 1.4787983854709842e-05,
      "loss": 0.2457,
      "step": 5640
    },
    {
      "epoch": 3.6090850927703135,
      "grad_norm": 1.5415893949155852,
      "learning_rate": 1.4786164556379103e-05,
      "loss": 0.2244,
      "step": 5641
    },
    {
      "epoch": 3.609724888035829,
      "grad_norm": 1.550711783328987,
      "learning_rate": 1.4784345052538975e-05,
      "loss": 0.2282,
      "step": 5642
    },
    {
      "epoch": 3.6103646833013436,
      "grad_norm": 1.5644837820596924,
      "learning_rate": 1.4782525343267579e-05,
      "loss": 0.2304,
      "step": 5643
    },
    {
      "epoch": 3.6110044785668585,
      "grad_norm": 1.4938339381873875,
      "learning_rate": 1.4780705428643053e-05,
      "loss": 0.223,
      "step": 5644
    },
    {
      "epoch": 3.6116442738323737,
      "grad_norm": 1.509807224077475,
      "learning_rate": 1.477888530874354e-05,
      "loss": 0.2458,
      "step": 5645
    },
    {
      "epoch": 3.6122840690978886,
      "grad_norm": 1.419631729177387,
      "learning_rate": 1.4777064983647194e-05,
      "loss": 0.2233,
      "step": 5646
    },
    {
      "epoch": 3.612923864363404,
      "grad_norm": 1.4816128096402876,
      "learning_rate": 1.4775244453432174e-05,
      "loss": 0.2415,
      "step": 5647
    },
    {
      "epoch": 3.6135636596289187,
      "grad_norm": 1.4537119397566958,
      "learning_rate": 1.4773423718176654e-05,
      "loss": 0.2483,
      "step": 5648
    },
    {
      "epoch": 3.614203454894434,
      "grad_norm": 1.424577393522732,
      "learning_rate": 1.4771602777958806e-05,
      "loss": 0.2255,
      "step": 5649
    },
    {
      "epoch": 3.614843250159949,
      "grad_norm": 1.4354306921377553,
      "learning_rate": 1.4769781632856828e-05,
      "loss": 0.224,
      "step": 5650
    },
    {
      "epoch": 3.6154830454254636,
      "grad_norm": 1.4610082728512452,
      "learning_rate": 1.4767960282948907e-05,
      "loss": 0.2405,
      "step": 5651
    },
    {
      "epoch": 3.616122840690979,
      "grad_norm": 1.5817686044443011,
      "learning_rate": 1.4766138728313255e-05,
      "loss": 0.2509,
      "step": 5652
    },
    {
      "epoch": 3.616762635956494,
      "grad_norm": 1.3895842136034469,
      "learning_rate": 1.4764316969028085e-05,
      "loss": 0.2064,
      "step": 5653
    },
    {
      "epoch": 3.617402431222009,
      "grad_norm": 1.4009468630658377,
      "learning_rate": 1.476249500517162e-05,
      "loss": 0.202,
      "step": 5654
    },
    {
      "epoch": 3.618042226487524,
      "grad_norm": 1.5559948322856758,
      "learning_rate": 1.4760672836822091e-05,
      "loss": 0.222,
      "step": 5655
    },
    {
      "epoch": 3.618682021753039,
      "grad_norm": 1.503812230832049,
      "learning_rate": 1.4758850464057737e-05,
      "loss": 0.2523,
      "step": 5656
    },
    {
      "epoch": 3.619321817018554,
      "grad_norm": 1.5082598690603233,
      "learning_rate": 1.475702788695681e-05,
      "loss": 0.2239,
      "step": 5657
    },
    {
      "epoch": 3.619961612284069,
      "grad_norm": 1.4226837801690306,
      "learning_rate": 1.4755205105597567e-05,
      "loss": 0.2122,
      "step": 5658
    },
    {
      "epoch": 3.620601407549584,
      "grad_norm": 1.4251972442835292,
      "learning_rate": 1.4753382120058278e-05,
      "loss": 0.2164,
      "step": 5659
    },
    {
      "epoch": 3.6212412028150993,
      "grad_norm": 1.532132243834398,
      "learning_rate": 1.4751558930417216e-05,
      "loss": 0.2679,
      "step": 5660
    },
    {
      "epoch": 3.621880998080614,
      "grad_norm": 1.5884703857500742,
      "learning_rate": 1.4749735536752662e-05,
      "loss": 0.2382,
      "step": 5661
    },
    {
      "epoch": 3.622520793346129,
      "grad_norm": 1.4302668333206712,
      "learning_rate": 1.4747911939142918e-05,
      "loss": 0.2083,
      "step": 5662
    },
    {
      "epoch": 3.6231605886116443,
      "grad_norm": 1.3916059000061347,
      "learning_rate": 1.4746088137666281e-05,
      "loss": 0.2089,
      "step": 5663
    },
    {
      "epoch": 3.6238003838771595,
      "grad_norm": 1.5338222354378603,
      "learning_rate": 1.474426413240106e-05,
      "loss": 0.2473,
      "step": 5664
    },
    {
      "epoch": 3.6244401791426744,
      "grad_norm": 1.4523902260102184,
      "learning_rate": 1.4742439923425578e-05,
      "loss": 0.2174,
      "step": 5665
    },
    {
      "epoch": 3.625079974408189,
      "grad_norm": 1.653402887988609,
      "learning_rate": 1.4740615510818159e-05,
      "loss": 0.2437,
      "step": 5666
    },
    {
      "epoch": 3.6257197696737045,
      "grad_norm": 1.5371578398380237,
      "learning_rate": 1.4738790894657142e-05,
      "loss": 0.2185,
      "step": 5667
    },
    {
      "epoch": 3.6263595649392193,
      "grad_norm": 1.3873263230453257,
      "learning_rate": 1.4736966075020878e-05,
      "loss": 0.1973,
      "step": 5668
    },
    {
      "epoch": 3.6269993602047346,
      "grad_norm": 1.6255658511943973,
      "learning_rate": 1.4735141051987719e-05,
      "loss": 0.2727,
      "step": 5669
    },
    {
      "epoch": 3.6276391554702494,
      "grad_norm": 1.5536412068661507,
      "learning_rate": 1.473331582563602e-05,
      "loss": 0.2656,
      "step": 5670
    },
    {
      "epoch": 3.6282789507357647,
      "grad_norm": 1.439974898404206,
      "learning_rate": 1.4731490396044162e-05,
      "loss": 0.2196,
      "step": 5671
    },
    {
      "epoch": 3.6289187460012795,
      "grad_norm": 1.400272795465504,
      "learning_rate": 1.4729664763290524e-05,
      "loss": 0.2173,
      "step": 5672
    },
    {
      "epoch": 3.629558541266795,
      "grad_norm": 1.4381545978314323,
      "learning_rate": 1.4727838927453493e-05,
      "loss": 0.2268,
      "step": 5673
    },
    {
      "epoch": 3.6301983365323096,
      "grad_norm": 1.472066153608141,
      "learning_rate": 1.4726012888611469e-05,
      "loss": 0.2095,
      "step": 5674
    },
    {
      "epoch": 3.630838131797825,
      "grad_norm": 1.4618495844319115,
      "learning_rate": 1.472418664684286e-05,
      "loss": 0.2185,
      "step": 5675
    },
    {
      "epoch": 3.6314779270633397,
      "grad_norm": 1.5658079710414463,
      "learning_rate": 1.4722360202226077e-05,
      "loss": 0.2372,
      "step": 5676
    },
    {
      "epoch": 3.6321177223288545,
      "grad_norm": 1.5183588769606573,
      "learning_rate": 1.4720533554839551e-05,
      "loss": 0.2308,
      "step": 5677
    },
    {
      "epoch": 3.63275751759437,
      "grad_norm": 1.5451225059545977,
      "learning_rate": 1.4718706704761709e-05,
      "loss": 0.2396,
      "step": 5678
    },
    {
      "epoch": 3.633397312859885,
      "grad_norm": 1.424191432138216,
      "learning_rate": 1.4716879652070995e-05,
      "loss": 0.226,
      "step": 5679
    },
    {
      "epoch": 3.6340371081254,
      "grad_norm": 1.4765293624090337,
      "learning_rate": 1.4715052396845862e-05,
      "loss": 0.2164,
      "step": 5680
    },
    {
      "epoch": 3.6346769033909148,
      "grad_norm": 1.4603946143380409,
      "learning_rate": 1.4713224939164762e-05,
      "loss": 0.2212,
      "step": 5681
    },
    {
      "epoch": 3.63531669865643,
      "grad_norm": 1.4551733604806931,
      "learning_rate": 1.471139727910617e-05,
      "loss": 0.2369,
      "step": 5682
    },
    {
      "epoch": 3.635956493921945,
      "grad_norm": 1.5108826911567523,
      "learning_rate": 1.4709569416748561e-05,
      "loss": 0.2337,
      "step": 5683
    },
    {
      "epoch": 3.63659628918746,
      "grad_norm": 1.5347053417343353,
      "learning_rate": 1.4707741352170418e-05,
      "loss": 0.2395,
      "step": 5684
    },
    {
      "epoch": 3.637236084452975,
      "grad_norm": 1.6065273574655343,
      "learning_rate": 1.4705913085450237e-05,
      "loss": 0.2532,
      "step": 5685
    },
    {
      "epoch": 3.6378758797184902,
      "grad_norm": 1.4002397410212257,
      "learning_rate": 1.4704084616666516e-05,
      "loss": 0.2064,
      "step": 5686
    },
    {
      "epoch": 3.638515674984005,
      "grad_norm": 1.3509952331294603,
      "learning_rate": 1.470225594589777e-05,
      "loss": 0.2191,
      "step": 5687
    },
    {
      "epoch": 3.63915547024952,
      "grad_norm": 1.4477985208531938,
      "learning_rate": 1.4700427073222518e-05,
      "loss": 0.211,
      "step": 5688
    },
    {
      "epoch": 3.639795265515035,
      "grad_norm": 1.573077669515796,
      "learning_rate": 1.4698597998719291e-05,
      "loss": 0.2199,
      "step": 5689
    },
    {
      "epoch": 3.6404350607805505,
      "grad_norm": 1.5046362081324074,
      "learning_rate": 1.4696768722466624e-05,
      "loss": 0.2403,
      "step": 5690
    },
    {
      "epoch": 3.6410748560460653,
      "grad_norm": 1.360091941346726,
      "learning_rate": 1.4694939244543061e-05,
      "loss": 0.189,
      "step": 5691
    },
    {
      "epoch": 3.64171465131158,
      "grad_norm": 1.5268320746452904,
      "learning_rate": 1.469310956502716e-05,
      "loss": 0.2306,
      "step": 5692
    },
    {
      "epoch": 3.6423544465770954,
      "grad_norm": 1.5266914157638605,
      "learning_rate": 1.4691279683997484e-05,
      "loss": 0.2105,
      "step": 5693
    },
    {
      "epoch": 3.6429942418426102,
      "grad_norm": 1.5066286068493477,
      "learning_rate": 1.4689449601532602e-05,
      "loss": 0.2208,
      "step": 5694
    },
    {
      "epoch": 3.6436340371081255,
      "grad_norm": 1.512863582118826,
      "learning_rate": 1.4687619317711096e-05,
      "loss": 0.23,
      "step": 5695
    },
    {
      "epoch": 3.6442738323736403,
      "grad_norm": 1.5074397272710978,
      "learning_rate": 1.4685788832611554e-05,
      "loss": 0.2415,
      "step": 5696
    },
    {
      "epoch": 3.6449136276391556,
      "grad_norm": 1.5005797117137711,
      "learning_rate": 1.4683958146312579e-05,
      "loss": 0.224,
      "step": 5697
    },
    {
      "epoch": 3.6455534229046704,
      "grad_norm": 1.5678455267652958,
      "learning_rate": 1.4682127258892766e-05,
      "loss": 0.2304,
      "step": 5698
    },
    {
      "epoch": 3.6461932181701853,
      "grad_norm": 1.4443905158210995,
      "learning_rate": 1.4680296170430744e-05,
      "loss": 0.2045,
      "step": 5699
    },
    {
      "epoch": 3.6468330134357005,
      "grad_norm": 1.5625618024265453,
      "learning_rate": 1.4678464881005128e-05,
      "loss": 0.2653,
      "step": 5700
    },
    {
      "epoch": 3.647472808701216,
      "grad_norm": 1.4928303311008513,
      "learning_rate": 1.4676633390694553e-05,
      "loss": 0.2357,
      "step": 5701
    },
    {
      "epoch": 3.6481126039667306,
      "grad_norm": 1.5001472629487913,
      "learning_rate": 1.4674801699577659e-05,
      "loss": 0.2155,
      "step": 5702
    },
    {
      "epoch": 3.6487523992322455,
      "grad_norm": 1.4320241997241963,
      "learning_rate": 1.4672969807733094e-05,
      "loss": 0.2202,
      "step": 5703
    },
    {
      "epoch": 3.6493921944977608,
      "grad_norm": 1.3814374220031673,
      "learning_rate": 1.4671137715239519e-05,
      "loss": 0.2265,
      "step": 5704
    },
    {
      "epoch": 3.6500319897632756,
      "grad_norm": 1.5252471706017687,
      "learning_rate": 1.4669305422175599e-05,
      "loss": 0.2434,
      "step": 5705
    },
    {
      "epoch": 3.650671785028791,
      "grad_norm": 1.5041966902127153,
      "learning_rate": 1.4667472928620011e-05,
      "loss": 0.2217,
      "step": 5706
    },
    {
      "epoch": 3.6513115802943057,
      "grad_norm": 1.5065845588480866,
      "learning_rate": 1.4665640234651439e-05,
      "loss": 0.2493,
      "step": 5707
    },
    {
      "epoch": 3.651951375559821,
      "grad_norm": 1.6487724018481793,
      "learning_rate": 1.4663807340348574e-05,
      "loss": 0.2615,
      "step": 5708
    },
    {
      "epoch": 3.652591170825336,
      "grad_norm": 1.4694665510962177,
      "learning_rate": 1.4661974245790118e-05,
      "loss": 0.2299,
      "step": 5709
    },
    {
      "epoch": 3.653230966090851,
      "grad_norm": 1.317087798824003,
      "learning_rate": 1.4660140951054781e-05,
      "loss": 0.1898,
      "step": 5710
    },
    {
      "epoch": 3.653870761356366,
      "grad_norm": 1.4625057136810016,
      "learning_rate": 1.465830745622128e-05,
      "loss": 0.2325,
      "step": 5711
    },
    {
      "epoch": 3.654510556621881,
      "grad_norm": 1.3989177537330308,
      "learning_rate": 1.4656473761368349e-05,
      "loss": 0.2135,
      "step": 5712
    },
    {
      "epoch": 3.655150351887396,
      "grad_norm": 1.4353099651376544,
      "learning_rate": 1.465463986657471e-05,
      "loss": 0.2106,
      "step": 5713
    },
    {
      "epoch": 3.655790147152911,
      "grad_norm": 1.4840654414766157,
      "learning_rate": 1.4652805771919121e-05,
      "loss": 0.2273,
      "step": 5714
    },
    {
      "epoch": 3.656429942418426,
      "grad_norm": 1.5260440717803518,
      "learning_rate": 1.465097147748033e-05,
      "loss": 0.2474,
      "step": 5715
    },
    {
      "epoch": 3.6570697376839414,
      "grad_norm": 1.4348444505245417,
      "learning_rate": 1.4649136983337091e-05,
      "loss": 0.217,
      "step": 5716
    },
    {
      "epoch": 3.657709532949456,
      "grad_norm": 1.5529048276736264,
      "learning_rate": 1.4647302289568184e-05,
      "loss": 0.2283,
      "step": 5717
    },
    {
      "epoch": 3.658349328214971,
      "grad_norm": 1.5658323501377371,
      "learning_rate": 1.4645467396252382e-05,
      "loss": 0.2426,
      "step": 5718
    },
    {
      "epoch": 3.6589891234804863,
      "grad_norm": 1.4322785519211434,
      "learning_rate": 1.4643632303468473e-05,
      "loss": 0.2227,
      "step": 5719
    },
    {
      "epoch": 3.659628918746001,
      "grad_norm": 1.6070281269041817,
      "learning_rate": 1.4641797011295255e-05,
      "loss": 0.2481,
      "step": 5720
    },
    {
      "epoch": 3.6602687140115164,
      "grad_norm": 1.330538491975617,
      "learning_rate": 1.463996151981153e-05,
      "loss": 0.1979,
      "step": 5721
    },
    {
      "epoch": 3.6609085092770313,
      "grad_norm": 1.5899763925544221,
      "learning_rate": 1.4638125829096113e-05,
      "loss": 0.2493,
      "step": 5722
    },
    {
      "epoch": 3.6615483045425465,
      "grad_norm": 1.4254092452889064,
      "learning_rate": 1.463628993922782e-05,
      "loss": 0.2251,
      "step": 5723
    },
    {
      "epoch": 3.6621880998080614,
      "grad_norm": 1.4705189321119383,
      "learning_rate": 1.4634453850285488e-05,
      "loss": 0.2228,
      "step": 5724
    },
    {
      "epoch": 3.662827895073576,
      "grad_norm": 1.346318241527392,
      "learning_rate": 1.4632617562347951e-05,
      "loss": 0.2189,
      "step": 5725
    },
    {
      "epoch": 3.6634676903390915,
      "grad_norm": 1.4073895748962255,
      "learning_rate": 1.4630781075494053e-05,
      "loss": 0.2075,
      "step": 5726
    },
    {
      "epoch": 3.6641074856046068,
      "grad_norm": 1.5188381482728133,
      "learning_rate": 1.4628944389802657e-05,
      "loss": 0.2373,
      "step": 5727
    },
    {
      "epoch": 3.6647472808701216,
      "grad_norm": 1.4869763165337335,
      "learning_rate": 1.462710750535262e-05,
      "loss": 0.228,
      "step": 5728
    },
    {
      "epoch": 3.6653870761356364,
      "grad_norm": 1.4586257990132088,
      "learning_rate": 1.4625270422222822e-05,
      "loss": 0.2334,
      "step": 5729
    },
    {
      "epoch": 3.6660268714011517,
      "grad_norm": 1.472408760723821,
      "learning_rate": 1.4623433140492135e-05,
      "loss": 0.2476,
      "step": 5730
    },
    {
      "epoch": 3.6666666666666665,
      "grad_norm": 1.5145607593913997,
      "learning_rate": 1.4621595660239455e-05,
      "loss": 0.2336,
      "step": 5731
    },
    {
      "epoch": 3.667306461932182,
      "grad_norm": 1.4040117349847743,
      "learning_rate": 1.4619757981543681e-05,
      "loss": 0.2031,
      "step": 5732
    },
    {
      "epoch": 3.6679462571976966,
      "grad_norm": 1.6565912969372834,
      "learning_rate": 1.4617920104483715e-05,
      "loss": 0.2667,
      "step": 5733
    },
    {
      "epoch": 3.668586052463212,
      "grad_norm": 1.3835013171323631,
      "learning_rate": 1.4616082029138473e-05,
      "loss": 0.2068,
      "step": 5734
    },
    {
      "epoch": 3.6692258477287267,
      "grad_norm": 1.4726715588557042,
      "learning_rate": 1.4614243755586883e-05,
      "loss": 0.2617,
      "step": 5735
    },
    {
      "epoch": 3.669865642994242,
      "grad_norm": 1.48282240955798,
      "learning_rate": 1.4612405283907873e-05,
      "loss": 0.2544,
      "step": 5736
    },
    {
      "epoch": 3.670505438259757,
      "grad_norm": 1.4932099250537865,
      "learning_rate": 1.4610566614180384e-05,
      "loss": 0.2505,
      "step": 5737
    },
    {
      "epoch": 3.671145233525272,
      "grad_norm": 1.8167799550585968,
      "learning_rate": 1.4608727746483369e-05,
      "loss": 0.2511,
      "step": 5738
    },
    {
      "epoch": 3.671785028790787,
      "grad_norm": 1.4832119327607396,
      "learning_rate": 1.4606888680895779e-05,
      "loss": 0.2112,
      "step": 5739
    },
    {
      "epoch": 3.6724248240563018,
      "grad_norm": 1.5091005235983599,
      "learning_rate": 1.4605049417496589e-05,
      "loss": 0.2305,
      "step": 5740
    },
    {
      "epoch": 3.673064619321817,
      "grad_norm": 1.5560896619706817,
      "learning_rate": 1.4603209956364764e-05,
      "loss": 0.2434,
      "step": 5741
    },
    {
      "epoch": 3.6737044145873323,
      "grad_norm": 1.509754173792451,
      "learning_rate": 1.4601370297579293e-05,
      "loss": 0.2614,
      "step": 5742
    },
    {
      "epoch": 3.674344209852847,
      "grad_norm": 1.486440066435346,
      "learning_rate": 1.4599530441219168e-05,
      "loss": 0.2208,
      "step": 5743
    },
    {
      "epoch": 3.674984005118362,
      "grad_norm": 1.4694535934306074,
      "learning_rate": 1.4597690387363388e-05,
      "loss": 0.2385,
      "step": 5744
    },
    {
      "epoch": 3.6756238003838773,
      "grad_norm": 1.4403016638101112,
      "learning_rate": 1.4595850136090961e-05,
      "loss": 0.2256,
      "step": 5745
    },
    {
      "epoch": 3.676263595649392,
      "grad_norm": 1.6326176944353978,
      "learning_rate": 1.4594009687480903e-05,
      "loss": 0.2471,
      "step": 5746
    },
    {
      "epoch": 3.6769033909149074,
      "grad_norm": 1.6945202005331652,
      "learning_rate": 1.4592169041612243e-05,
      "loss": 0.2289,
      "step": 5747
    },
    {
      "epoch": 3.677543186180422,
      "grad_norm": 1.681522326966433,
      "learning_rate": 1.4590328198564013e-05,
      "loss": 0.2557,
      "step": 5748
    },
    {
      "epoch": 3.6781829814459375,
      "grad_norm": 1.5323061777613938,
      "learning_rate": 1.4588487158415252e-05,
      "loss": 0.2372,
      "step": 5749
    },
    {
      "epoch": 3.6788227767114523,
      "grad_norm": 1.5441247644218774,
      "learning_rate": 1.4586645921245019e-05,
      "loss": 0.2403,
      "step": 5750
    },
    {
      "epoch": 3.679462571976967,
      "grad_norm": 1.52949483974214,
      "learning_rate": 1.4584804487132367e-05,
      "loss": 0.2475,
      "step": 5751
    },
    {
      "epoch": 3.6801023672424824,
      "grad_norm": 1.4713291917653244,
      "learning_rate": 1.4582962856156368e-05,
      "loss": 0.2027,
      "step": 5752
    },
    {
      "epoch": 3.6807421625079977,
      "grad_norm": 1.722244224355452,
      "learning_rate": 1.4581121028396098e-05,
      "loss": 0.2644,
      "step": 5753
    },
    {
      "epoch": 3.6813819577735125,
      "grad_norm": 1.4452190923839596,
      "learning_rate": 1.4579279003930635e-05,
      "loss": 0.2181,
      "step": 5754
    },
    {
      "epoch": 3.6820217530390273,
      "grad_norm": 1.444745338022213,
      "learning_rate": 1.457743678283908e-05,
      "loss": 0.1995,
      "step": 5755
    },
    {
      "epoch": 3.6826615483045426,
      "grad_norm": 1.4110991194508513,
      "learning_rate": 1.4575594365200532e-05,
      "loss": 0.2103,
      "step": 5756
    },
    {
      "epoch": 3.6833013435700575,
      "grad_norm": 1.4557935898409464,
      "learning_rate": 1.4573751751094103e-05,
      "loss": 0.2291,
      "step": 5757
    },
    {
      "epoch": 3.6839411388355727,
      "grad_norm": 1.4676848191777654,
      "learning_rate": 1.4571908940598907e-05,
      "loss": 0.2397,
      "step": 5758
    },
    {
      "epoch": 3.6845809341010876,
      "grad_norm": 1.5500711202087376,
      "learning_rate": 1.4570065933794077e-05,
      "loss": 0.2528,
      "step": 5759
    },
    {
      "epoch": 3.685220729366603,
      "grad_norm": 1.4063372360306294,
      "learning_rate": 1.4568222730758743e-05,
      "loss": 0.2415,
      "step": 5760
    },
    {
      "epoch": 3.6858605246321177,
      "grad_norm": 1.5028670715207457,
      "learning_rate": 1.4566379331572051e-05,
      "loss": 0.2316,
      "step": 5761
    },
    {
      "epoch": 3.6865003198976325,
      "grad_norm": 1.5048725198144817,
      "learning_rate": 1.4564535736313157e-05,
      "loss": 0.2391,
      "step": 5762
    },
    {
      "epoch": 3.6871401151631478,
      "grad_norm": 1.493478534481618,
      "learning_rate": 1.4562691945061214e-05,
      "loss": 0.2144,
      "step": 5763
    },
    {
      "epoch": 3.687779910428663,
      "grad_norm": 1.4455525564348326,
      "learning_rate": 1.4560847957895394e-05,
      "loss": 0.2114,
      "step": 5764
    },
    {
      "epoch": 3.688419705694178,
      "grad_norm": 1.5173869658570351,
      "learning_rate": 1.455900377489488e-05,
      "loss": 0.2079,
      "step": 5765
    },
    {
      "epoch": 3.6890595009596927,
      "grad_norm": 1.3337962462641157,
      "learning_rate": 1.455715939613885e-05,
      "loss": 0.1963,
      "step": 5766
    },
    {
      "epoch": 3.689699296225208,
      "grad_norm": 1.4218953966358945,
      "learning_rate": 1.4555314821706507e-05,
      "loss": 0.2196,
      "step": 5767
    },
    {
      "epoch": 3.690339091490723,
      "grad_norm": 1.3893213293155213,
      "learning_rate": 1.4553470051677048e-05,
      "loss": 0.209,
      "step": 5768
    },
    {
      "epoch": 3.690978886756238,
      "grad_norm": 1.5999437164372106,
      "learning_rate": 1.4551625086129681e-05,
      "loss": 0.2633,
      "step": 5769
    },
    {
      "epoch": 3.691618682021753,
      "grad_norm": 1.5555326378529222,
      "learning_rate": 1.4549779925143634e-05,
      "loss": 0.2412,
      "step": 5770
    },
    {
      "epoch": 3.692258477287268,
      "grad_norm": 1.597834449762471,
      "learning_rate": 1.4547934568798128e-05,
      "loss": 0.2472,
      "step": 5771
    },
    {
      "epoch": 3.692898272552783,
      "grad_norm": 1.4314513694734146,
      "learning_rate": 1.4546089017172402e-05,
      "loss": 0.2252,
      "step": 5772
    },
    {
      "epoch": 3.6935380678182983,
      "grad_norm": 1.676758053923561,
      "learning_rate": 1.45442432703457e-05,
      "loss": 0.246,
      "step": 5773
    },
    {
      "epoch": 3.694177863083813,
      "grad_norm": 1.3925700610952885,
      "learning_rate": 1.4542397328397276e-05,
      "loss": 0.2151,
      "step": 5774
    },
    {
      "epoch": 3.6948176583493284,
      "grad_norm": 1.5314433897693238,
      "learning_rate": 1.4540551191406392e-05,
      "loss": 0.2389,
      "step": 5775
    },
    {
      "epoch": 3.6954574536148432,
      "grad_norm": 1.3770803283424844,
      "learning_rate": 1.4538704859452315e-05,
      "loss": 0.2239,
      "step": 5776
    },
    {
      "epoch": 3.696097248880358,
      "grad_norm": 1.440420679775074,
      "learning_rate": 1.4536858332614327e-05,
      "loss": 0.2174,
      "step": 5777
    },
    {
      "epoch": 3.6967370441458733,
      "grad_norm": 1.4017599860335717,
      "learning_rate": 1.453501161097171e-05,
      "loss": 0.2035,
      "step": 5778
    },
    {
      "epoch": 3.6973768394113886,
      "grad_norm": 1.4563895915574216,
      "learning_rate": 1.4533164694603766e-05,
      "loss": 0.2017,
      "step": 5779
    },
    {
      "epoch": 3.6980166346769034,
      "grad_norm": 1.4327253525503103,
      "learning_rate": 1.4531317583589787e-05,
      "loss": 0.2112,
      "step": 5780
    },
    {
      "epoch": 3.6986564299424183,
      "grad_norm": 1.374794713997775,
      "learning_rate": 1.4529470278009097e-05,
      "loss": 0.2179,
      "step": 5781
    },
    {
      "epoch": 3.6992962252079336,
      "grad_norm": 1.538786144024654,
      "learning_rate": 1.4527622777941008e-05,
      "loss": 0.232,
      "step": 5782
    },
    {
      "epoch": 3.6999360204734484,
      "grad_norm": 1.5479901432498309,
      "learning_rate": 1.4525775083464852e-05,
      "loss": 0.2377,
      "step": 5783
    },
    {
      "epoch": 3.7005758157389637,
      "grad_norm": 1.5421591575789788,
      "learning_rate": 1.4523927194659968e-05,
      "loss": 0.2177,
      "step": 5784
    },
    {
      "epoch": 3.7012156110044785,
      "grad_norm": 1.4257330046750714,
      "learning_rate": 1.4522079111605693e-05,
      "loss": 0.2144,
      "step": 5785
    },
    {
      "epoch": 3.7018554062699938,
      "grad_norm": 1.437892567602319,
      "learning_rate": 1.4520230834381386e-05,
      "loss": 0.2177,
      "step": 5786
    },
    {
      "epoch": 3.7024952015355086,
      "grad_norm": 1.5391466471021031,
      "learning_rate": 1.451838236306641e-05,
      "loss": 0.2586,
      "step": 5787
    },
    {
      "epoch": 3.7031349968010234,
      "grad_norm": 1.469600593172427,
      "learning_rate": 1.451653369774013e-05,
      "loss": 0.2248,
      "step": 5788
    },
    {
      "epoch": 3.7037747920665387,
      "grad_norm": 1.5829313161844691,
      "learning_rate": 1.4514684838481929e-05,
      "loss": 0.2638,
      "step": 5789
    },
    {
      "epoch": 3.704414587332054,
      "grad_norm": 1.4241908283183375,
      "learning_rate": 1.4512835785371193e-05,
      "loss": 0.1856,
      "step": 5790
    },
    {
      "epoch": 3.705054382597569,
      "grad_norm": 1.4187859135913476,
      "learning_rate": 1.4510986538487312e-05,
      "loss": 0.2183,
      "step": 5791
    },
    {
      "epoch": 3.7056941778630836,
      "grad_norm": 1.4901580067987974,
      "learning_rate": 1.4509137097909698e-05,
      "loss": 0.2594,
      "step": 5792
    },
    {
      "epoch": 3.706333973128599,
      "grad_norm": 1.5122070217005088,
      "learning_rate": 1.4507287463717754e-05,
      "loss": 0.2241,
      "step": 5793
    },
    {
      "epoch": 3.7069737683941137,
      "grad_norm": 1.4571331534407546,
      "learning_rate": 1.4505437635990909e-05,
      "loss": 0.2157,
      "step": 5794
    },
    {
      "epoch": 3.707613563659629,
      "grad_norm": 1.5790986510769844,
      "learning_rate": 1.4503587614808582e-05,
      "loss": 0.2702,
      "step": 5795
    },
    {
      "epoch": 3.708253358925144,
      "grad_norm": 1.6702037107860592,
      "learning_rate": 1.4501737400250215e-05,
      "loss": 0.3087,
      "step": 5796
    },
    {
      "epoch": 3.708893154190659,
      "grad_norm": 1.5715860006584086,
      "learning_rate": 1.4499886992395253e-05,
      "loss": 0.2182,
      "step": 5797
    },
    {
      "epoch": 3.709532949456174,
      "grad_norm": 1.4714470783063098,
      "learning_rate": 1.449803639132315e-05,
      "loss": 0.2095,
      "step": 5798
    },
    {
      "epoch": 3.710172744721689,
      "grad_norm": 1.5090337965600467,
      "learning_rate": 1.4496185597113366e-05,
      "loss": 0.2697,
      "step": 5799
    },
    {
      "epoch": 3.710812539987204,
      "grad_norm": 1.55222183877453,
      "learning_rate": 1.4494334609845366e-05,
      "loss": 0.2502,
      "step": 5800
    },
    {
      "epoch": 3.7114523352527193,
      "grad_norm": 1.4175924498341024,
      "learning_rate": 1.4492483429598637e-05,
      "loss": 0.2257,
      "step": 5801
    },
    {
      "epoch": 3.712092130518234,
      "grad_norm": 1.3358817544321473,
      "learning_rate": 1.4490632056452658e-05,
      "loss": 0.1971,
      "step": 5802
    },
    {
      "epoch": 3.712731925783749,
      "grad_norm": 1.5463186047771054,
      "learning_rate": 1.4488780490486926e-05,
      "loss": 0.2042,
      "step": 5803
    },
    {
      "epoch": 3.7133717210492643,
      "grad_norm": 1.506785455838855,
      "learning_rate": 1.4486928731780947e-05,
      "loss": 0.2298,
      "step": 5804
    },
    {
      "epoch": 3.714011516314779,
      "grad_norm": 1.4575309891648418,
      "learning_rate": 1.4485076780414232e-05,
      "loss": 0.2121,
      "step": 5805
    },
    {
      "epoch": 3.7146513115802944,
      "grad_norm": 1.5958348004065732,
      "learning_rate": 1.4483224636466296e-05,
      "loss": 0.2685,
      "step": 5806
    },
    {
      "epoch": 3.715291106845809,
      "grad_norm": 1.5278745547608616,
      "learning_rate": 1.4481372300016668e-05,
      "loss": 0.2318,
      "step": 5807
    },
    {
      "epoch": 3.7159309021113245,
      "grad_norm": 1.7536241095297431,
      "learning_rate": 1.4479519771144885e-05,
      "loss": 0.2446,
      "step": 5808
    },
    {
      "epoch": 3.7165706973768393,
      "grad_norm": 1.5261565362820926,
      "learning_rate": 1.4477667049930495e-05,
      "loss": 0.2451,
      "step": 5809
    },
    {
      "epoch": 3.7172104926423546,
      "grad_norm": 1.3755199819907566,
      "learning_rate": 1.4475814136453043e-05,
      "loss": 0.2088,
      "step": 5810
    },
    {
      "epoch": 3.7178502879078694,
      "grad_norm": 1.5518842392000156,
      "learning_rate": 1.4473961030792095e-05,
      "loss": 0.2392,
      "step": 5811
    },
    {
      "epoch": 3.7184900831733847,
      "grad_norm": 1.4265129777356929,
      "learning_rate": 1.4472107733027221e-05,
      "loss": 0.2187,
      "step": 5812
    },
    {
      "epoch": 3.7191298784388995,
      "grad_norm": 1.4220083231804053,
      "learning_rate": 1.4470254243237993e-05,
      "loss": 0.2256,
      "step": 5813
    },
    {
      "epoch": 3.7197696737044144,
      "grad_norm": 1.5127637055988226,
      "learning_rate": 1.4468400561504004e-05,
      "loss": 0.25,
      "step": 5814
    },
    {
      "epoch": 3.7204094689699296,
      "grad_norm": 1.4347939355948192,
      "learning_rate": 1.4466546687904839e-05,
      "loss": 0.2283,
      "step": 5815
    },
    {
      "epoch": 3.721049264235445,
      "grad_norm": 1.3477240769621786,
      "learning_rate": 1.4464692622520107e-05,
      "loss": 0.2115,
      "step": 5816
    },
    {
      "epoch": 3.7216890595009597,
      "grad_norm": 1.5135768793278368,
      "learning_rate": 1.4462838365429414e-05,
      "loss": 0.2374,
      "step": 5817
    },
    {
      "epoch": 3.7223288547664746,
      "grad_norm": 1.541742950014774,
      "learning_rate": 1.4460983916712384e-05,
      "loss": 0.2436,
      "step": 5818
    },
    {
      "epoch": 3.72296865003199,
      "grad_norm": 1.615525925760475,
      "learning_rate": 1.4459129276448636e-05,
      "loss": 0.2756,
      "step": 5819
    },
    {
      "epoch": 3.7236084452975047,
      "grad_norm": 1.572934796199863,
      "learning_rate": 1.4457274444717812e-05,
      "loss": 0.2507,
      "step": 5820
    },
    {
      "epoch": 3.72424824056302,
      "grad_norm": 1.5584809906008479,
      "learning_rate": 1.445541942159955e-05,
      "loss": 0.2412,
      "step": 5821
    },
    {
      "epoch": 3.724888035828535,
      "grad_norm": 1.4813537884710337,
      "learning_rate": 1.4453564207173507e-05,
      "loss": 0.2322,
      "step": 5822
    },
    {
      "epoch": 3.72552783109405,
      "grad_norm": 1.4304286458737425,
      "learning_rate": 1.4451708801519335e-05,
      "loss": 0.2205,
      "step": 5823
    },
    {
      "epoch": 3.726167626359565,
      "grad_norm": 1.390378708963792,
      "learning_rate": 1.4449853204716711e-05,
      "loss": 0.2161,
      "step": 5824
    },
    {
      "epoch": 3.7268074216250797,
      "grad_norm": 1.5349009590223155,
      "learning_rate": 1.4447997416845301e-05,
      "loss": 0.2388,
      "step": 5825
    },
    {
      "epoch": 3.727447216890595,
      "grad_norm": 1.4312390462134759,
      "learning_rate": 1.4446141437984795e-05,
      "loss": 0.2145,
      "step": 5826
    },
    {
      "epoch": 3.7280870121561103,
      "grad_norm": 1.4522085761980827,
      "learning_rate": 1.4444285268214886e-05,
      "loss": 0.2302,
      "step": 5827
    },
    {
      "epoch": 3.728726807421625,
      "grad_norm": 1.5546823138163313,
      "learning_rate": 1.4442428907615275e-05,
      "loss": 0.2114,
      "step": 5828
    },
    {
      "epoch": 3.72936660268714,
      "grad_norm": 1.4961743060533255,
      "learning_rate": 1.4440572356265668e-05,
      "loss": 0.2184,
      "step": 5829
    },
    {
      "epoch": 3.730006397952655,
      "grad_norm": 1.4930331223857258,
      "learning_rate": 1.4438715614245781e-05,
      "loss": 0.2139,
      "step": 5830
    },
    {
      "epoch": 3.73064619321817,
      "grad_norm": 1.473671944943778,
      "learning_rate": 1.4436858681635343e-05,
      "loss": 0.2308,
      "step": 5831
    },
    {
      "epoch": 3.7312859884836853,
      "grad_norm": 1.5723910037455269,
      "learning_rate": 1.4435001558514089e-05,
      "loss": 0.2401,
      "step": 5832
    },
    {
      "epoch": 3.7319257837492,
      "grad_norm": 1.5853063158000043,
      "learning_rate": 1.4433144244961752e-05,
      "loss": 0.2299,
      "step": 5833
    },
    {
      "epoch": 3.7325655790147154,
      "grad_norm": 1.462661978231914,
      "learning_rate": 1.443128674105809e-05,
      "loss": 0.215,
      "step": 5834
    },
    {
      "epoch": 3.7332053742802302,
      "grad_norm": 1.6670729646738964,
      "learning_rate": 1.4429429046882857e-05,
      "loss": 0.2542,
      "step": 5835
    },
    {
      "epoch": 3.7338451695457455,
      "grad_norm": 1.61502187744441,
      "learning_rate": 1.4427571162515822e-05,
      "loss": 0.2506,
      "step": 5836
    },
    {
      "epoch": 3.7344849648112604,
      "grad_norm": 1.5250736064766646,
      "learning_rate": 1.4425713088036754e-05,
      "loss": 0.2348,
      "step": 5837
    },
    {
      "epoch": 3.7351247600767756,
      "grad_norm": 1.565075157013613,
      "learning_rate": 1.442385482352544e-05,
      "loss": 0.2596,
      "step": 5838
    },
    {
      "epoch": 3.7357645553422905,
      "grad_norm": 1.4268375792512138,
      "learning_rate": 1.442199636906167e-05,
      "loss": 0.2269,
      "step": 5839
    },
    {
      "epoch": 3.7364043506078053,
      "grad_norm": 1.440199434876275,
      "learning_rate": 1.4420137724725243e-05,
      "loss": 0.2269,
      "step": 5840
    },
    {
      "epoch": 3.7370441458733206,
      "grad_norm": 1.5471910273806073,
      "learning_rate": 1.4418278890595962e-05,
      "loss": 0.2405,
      "step": 5841
    },
    {
      "epoch": 3.737683941138836,
      "grad_norm": 1.5115946787917773,
      "learning_rate": 1.4416419866753648e-05,
      "loss": 0.248,
      "step": 5842
    },
    {
      "epoch": 3.7383237364043507,
      "grad_norm": 1.4459213952917662,
      "learning_rate": 1.441456065327812e-05,
      "loss": 0.2522,
      "step": 5843
    },
    {
      "epoch": 3.7389635316698655,
      "grad_norm": 1.5801709475273045,
      "learning_rate": 1.441270125024921e-05,
      "loss": 0.2733,
      "step": 5844
    },
    {
      "epoch": 3.739603326935381,
      "grad_norm": 4.099363323728811,
      "learning_rate": 1.4410841657746759e-05,
      "loss": 0.2309,
      "step": 5845
    },
    {
      "epoch": 3.7402431222008956,
      "grad_norm": 1.593694232937027,
      "learning_rate": 1.4408981875850613e-05,
      "loss": 0.2547,
      "step": 5846
    },
    {
      "epoch": 3.740882917466411,
      "grad_norm": 1.3399490605129225,
      "learning_rate": 1.4407121904640626e-05,
      "loss": 0.1866,
      "step": 5847
    },
    {
      "epoch": 3.7415227127319257,
      "grad_norm": 1.50945972801813,
      "learning_rate": 1.4405261744196665e-05,
      "loss": 0.2386,
      "step": 5848
    },
    {
      "epoch": 3.742162507997441,
      "grad_norm": 1.4261543603360416,
      "learning_rate": 1.4403401394598604e-05,
      "loss": 0.2133,
      "step": 5849
    },
    {
      "epoch": 3.742802303262956,
      "grad_norm": 1.3830605554974005,
      "learning_rate": 1.440154085592632e-05,
      "loss": 0.2139,
      "step": 5850
    },
    {
      "epoch": 3.7434420985284707,
      "grad_norm": 1.3684872836583044,
      "learning_rate": 1.4399680128259698e-05,
      "loss": 0.2201,
      "step": 5851
    },
    {
      "epoch": 3.744081893793986,
      "grad_norm": 1.4184241983949613,
      "learning_rate": 1.439781921167864e-05,
      "loss": 0.219,
      "step": 5852
    },
    {
      "epoch": 3.744721689059501,
      "grad_norm": 1.4712956560474348,
      "learning_rate": 1.4395958106263045e-05,
      "loss": 0.2064,
      "step": 5853
    },
    {
      "epoch": 3.745361484325016,
      "grad_norm": 1.399920090280922,
      "learning_rate": 1.4394096812092832e-05,
      "loss": 0.1914,
      "step": 5854
    },
    {
      "epoch": 3.746001279590531,
      "grad_norm": 1.541290857496778,
      "learning_rate": 1.4392235329247917e-05,
      "loss": 0.2626,
      "step": 5855
    },
    {
      "epoch": 3.746641074856046,
      "grad_norm": 1.650061905526006,
      "learning_rate": 1.4390373657808233e-05,
      "loss": 0.234,
      "step": 5856
    },
    {
      "epoch": 3.747280870121561,
      "grad_norm": 1.624481969765583,
      "learning_rate": 1.4388511797853707e-05,
      "loss": 0.2299,
      "step": 5857
    },
    {
      "epoch": 3.7479206653870762,
      "grad_norm": 1.641341725626894,
      "learning_rate": 1.4386649749464299e-05,
      "loss": 0.2396,
      "step": 5858
    },
    {
      "epoch": 3.748560460652591,
      "grad_norm": 1.5461889934600799,
      "learning_rate": 1.438478751271995e-05,
      "loss": 0.2558,
      "step": 5859
    },
    {
      "epoch": 3.7492002559181064,
      "grad_norm": 1.353030290240435,
      "learning_rate": 1.4382925087700625e-05,
      "loss": 0.2003,
      "step": 5860
    },
    {
      "epoch": 3.749840051183621,
      "grad_norm": 1.4660387165642357,
      "learning_rate": 1.4381062474486296e-05,
      "loss": 0.194,
      "step": 5861
    },
    {
      "epoch": 3.750479846449136,
      "grad_norm": 1.5538411815952557,
      "learning_rate": 1.4379199673156934e-05,
      "loss": 0.2153,
      "step": 5862
    },
    {
      "epoch": 3.7511196417146513,
      "grad_norm": 1.4510540361193263,
      "learning_rate": 1.437733668379253e-05,
      "loss": 0.2374,
      "step": 5863
    },
    {
      "epoch": 3.7517594369801666,
      "grad_norm": 1.581226318850098,
      "learning_rate": 1.4375473506473077e-05,
      "loss": 0.22,
      "step": 5864
    },
    {
      "epoch": 3.7523992322456814,
      "grad_norm": 1.4914047996126052,
      "learning_rate": 1.4373610141278573e-05,
      "loss": 0.2408,
      "step": 5865
    },
    {
      "epoch": 3.7530390275111962,
      "grad_norm": 1.4837765971487782,
      "learning_rate": 1.4371746588289031e-05,
      "loss": 0.2079,
      "step": 5866
    },
    {
      "epoch": 3.7536788227767115,
      "grad_norm": 1.5027584020617297,
      "learning_rate": 1.4369882847584464e-05,
      "loss": 0.2387,
      "step": 5867
    },
    {
      "epoch": 3.7543186180422263,
      "grad_norm": 1.431058278502066,
      "learning_rate": 1.4368018919244904e-05,
      "loss": 0.2324,
      "step": 5868
    },
    {
      "epoch": 3.7549584133077416,
      "grad_norm": 1.5393359855584678,
      "learning_rate": 1.4366154803350381e-05,
      "loss": 0.2073,
      "step": 5869
    },
    {
      "epoch": 3.7555982085732564,
      "grad_norm": 1.551495144266616,
      "learning_rate": 1.4364290499980935e-05,
      "loss": 0.235,
      "step": 5870
    },
    {
      "epoch": 3.7562380038387717,
      "grad_norm": 1.4394592228971916,
      "learning_rate": 1.436242600921662e-05,
      "loss": 0.2108,
      "step": 5871
    },
    {
      "epoch": 3.7568777991042865,
      "grad_norm": 1.4914784064072935,
      "learning_rate": 1.4360561331137493e-05,
      "loss": 0.255,
      "step": 5872
    },
    {
      "epoch": 3.757517594369802,
      "grad_norm": 1.4067951263727665,
      "learning_rate": 1.435869646582362e-05,
      "loss": 0.2363,
      "step": 5873
    },
    {
      "epoch": 3.7581573896353166,
      "grad_norm": 1.45945217215376,
      "learning_rate": 1.4356831413355074e-05,
      "loss": 0.246,
      "step": 5874
    },
    {
      "epoch": 3.758797184900832,
      "grad_norm": 1.4168905270576428,
      "learning_rate": 1.4354966173811938e-05,
      "loss": 0.2307,
      "step": 5875
    },
    {
      "epoch": 3.7594369801663468,
      "grad_norm": 1.3889093243165254,
      "learning_rate": 1.43531007472743e-05,
      "loss": 0.2057,
      "step": 5876
    },
    {
      "epoch": 3.7600767754318616,
      "grad_norm": 1.4288544250996567,
      "learning_rate": 1.4351235133822259e-05,
      "loss": 0.2285,
      "step": 5877
    },
    {
      "epoch": 3.760716570697377,
      "grad_norm": 1.4602636157219648,
      "learning_rate": 1.434936933353592e-05,
      "loss": 0.2286,
      "step": 5878
    },
    {
      "epoch": 3.761356365962892,
      "grad_norm": 1.58474654974749,
      "learning_rate": 1.4347503346495405e-05,
      "loss": 0.2544,
      "step": 5879
    },
    {
      "epoch": 3.761996161228407,
      "grad_norm": 1.439798415983933,
      "learning_rate": 1.4345637172780826e-05,
      "loss": 0.2232,
      "step": 5880
    },
    {
      "epoch": 3.762635956493922,
      "grad_norm": 1.6692596715041992,
      "learning_rate": 1.4343770812472317e-05,
      "loss": 0.259,
      "step": 5881
    },
    {
      "epoch": 3.763275751759437,
      "grad_norm": 1.3390417532441408,
      "learning_rate": 1.4341904265650015e-05,
      "loss": 0.2135,
      "step": 5882
    },
    {
      "epoch": 3.763915547024952,
      "grad_norm": 1.4583328916461515,
      "learning_rate": 1.434003753239407e-05,
      "loss": 0.2031,
      "step": 5883
    },
    {
      "epoch": 3.764555342290467,
      "grad_norm": 1.4987643938369815,
      "learning_rate": 1.4338170612784633e-05,
      "loss": 0.1881,
      "step": 5884
    },
    {
      "epoch": 3.765195137555982,
      "grad_norm": 1.486457985141735,
      "learning_rate": 1.4336303506901863e-05,
      "loss": 0.2267,
      "step": 5885
    },
    {
      "epoch": 3.7658349328214973,
      "grad_norm": 1.3543146214183859,
      "learning_rate": 1.4334436214825939e-05,
      "loss": 0.1935,
      "step": 5886
    },
    {
      "epoch": 3.766474728087012,
      "grad_norm": 1.5103278957588604,
      "learning_rate": 1.4332568736637032e-05,
      "loss": 0.2199,
      "step": 5887
    },
    {
      "epoch": 3.767114523352527,
      "grad_norm": 1.4615105932889025,
      "learning_rate": 1.4330701072415331e-05,
      "loss": 0.2236,
      "step": 5888
    },
    {
      "epoch": 3.767754318618042,
      "grad_norm": 1.4278774250264783,
      "learning_rate": 1.4328833222241031e-05,
      "loss": 0.2195,
      "step": 5889
    },
    {
      "epoch": 3.7683941138835575,
      "grad_norm": 1.6058064415684703,
      "learning_rate": 1.4326965186194333e-05,
      "loss": 0.2748,
      "step": 5890
    },
    {
      "epoch": 3.7690339091490723,
      "grad_norm": 1.404607676306898,
      "learning_rate": 1.4325096964355445e-05,
      "loss": 0.2385,
      "step": 5891
    },
    {
      "epoch": 3.769673704414587,
      "grad_norm": 1.3893847547275746,
      "learning_rate": 1.4323228556804587e-05,
      "loss": 0.226,
      "step": 5892
    },
    {
      "epoch": 3.7703134996801024,
      "grad_norm": 1.5211192563009557,
      "learning_rate": 1.4321359963621985e-05,
      "loss": 0.2392,
      "step": 5893
    },
    {
      "epoch": 3.7709532949456173,
      "grad_norm": 1.3948620966438698,
      "learning_rate": 1.4319491184887874e-05,
      "loss": 0.1989,
      "step": 5894
    },
    {
      "epoch": 3.7715930902111325,
      "grad_norm": 1.6342104847541428,
      "learning_rate": 1.4317622220682498e-05,
      "loss": 0.2775,
      "step": 5895
    },
    {
      "epoch": 3.7722328854766474,
      "grad_norm": 1.4560503414681865,
      "learning_rate": 1.4315753071086103e-05,
      "loss": 0.2176,
      "step": 5896
    },
    {
      "epoch": 3.7728726807421626,
      "grad_norm": 1.4301208578256035,
      "learning_rate": 1.4313883736178946e-05,
      "loss": 0.2479,
      "step": 5897
    },
    {
      "epoch": 3.7735124760076775,
      "grad_norm": 1.450558455320564,
      "learning_rate": 1.4312014216041297e-05,
      "loss": 0.2427,
      "step": 5898
    },
    {
      "epoch": 3.7741522712731923,
      "grad_norm": 1.5419072865375283,
      "learning_rate": 1.4310144510753428e-05,
      "loss": 0.2551,
      "step": 5899
    },
    {
      "epoch": 3.7747920665387076,
      "grad_norm": 1.4516761133924716,
      "learning_rate": 1.4308274620395617e-05,
      "loss": 0.186,
      "step": 5900
    },
    {
      "epoch": 3.775431861804223,
      "grad_norm": 1.481765315470158,
      "learning_rate": 1.4306404545048163e-05,
      "loss": 0.2179,
      "step": 5901
    },
    {
      "epoch": 3.7760716570697377,
      "grad_norm": 1.7247892218248986,
      "learning_rate": 1.4304534284791354e-05,
      "loss": 0.2959,
      "step": 5902
    },
    {
      "epoch": 3.7767114523352525,
      "grad_norm": 1.4824131356655745,
      "learning_rate": 1.4302663839705504e-05,
      "loss": 0.245,
      "step": 5903
    },
    {
      "epoch": 3.777351247600768,
      "grad_norm": 1.5509925413653571,
      "learning_rate": 1.4300793209870922e-05,
      "loss": 0.247,
      "step": 5904
    },
    {
      "epoch": 3.777991042866283,
      "grad_norm": 1.4583429399423835,
      "learning_rate": 1.4298922395367927e-05,
      "loss": 0.2251,
      "step": 5905
    },
    {
      "epoch": 3.778630838131798,
      "grad_norm": 1.624477704407426,
      "learning_rate": 1.4297051396276855e-05,
      "loss": 0.2591,
      "step": 5906
    },
    {
      "epoch": 3.7792706333973127,
      "grad_norm": 1.4912332885355035,
      "learning_rate": 1.4295180212678037e-05,
      "loss": 0.2404,
      "step": 5907
    },
    {
      "epoch": 3.779910428662828,
      "grad_norm": 1.4742531263752114,
      "learning_rate": 1.429330884465182e-05,
      "loss": 0.2537,
      "step": 5908
    },
    {
      "epoch": 3.780550223928343,
      "grad_norm": 1.4206918625518496,
      "learning_rate": 1.4291437292278562e-05,
      "loss": 0.2256,
      "step": 5909
    },
    {
      "epoch": 3.781190019193858,
      "grad_norm": 1.4515326561000967,
      "learning_rate": 1.428956555563862e-05,
      "loss": 0.2471,
      "step": 5910
    },
    {
      "epoch": 3.781829814459373,
      "grad_norm": 1.5022690045380978,
      "learning_rate": 1.4287693634812366e-05,
      "loss": 0.2496,
      "step": 5911
    },
    {
      "epoch": 3.782469609724888,
      "grad_norm": 1.54942132433893,
      "learning_rate": 1.428582152988017e-05,
      "loss": 0.2151,
      "step": 5912
    },
    {
      "epoch": 3.783109404990403,
      "grad_norm": 1.5752098748088574,
      "learning_rate": 1.4283949240922423e-05,
      "loss": 0.2568,
      "step": 5913
    },
    {
      "epoch": 3.783749200255918,
      "grad_norm": 1.4154715633876938,
      "learning_rate": 1.4282076768019514e-05,
      "loss": 0.2083,
      "step": 5914
    },
    {
      "epoch": 3.784388995521433,
      "grad_norm": 1.5043700637589554,
      "learning_rate": 1.428020411125185e-05,
      "loss": 0.2289,
      "step": 5915
    },
    {
      "epoch": 3.7850287907869484,
      "grad_norm": 1.3601312523484657,
      "learning_rate": 1.4278331270699832e-05,
      "loss": 0.181,
      "step": 5916
    },
    {
      "epoch": 3.7856685860524633,
      "grad_norm": 1.5675934232431703,
      "learning_rate": 1.427645824644388e-05,
      "loss": 0.2381,
      "step": 5917
    },
    {
      "epoch": 3.786308381317978,
      "grad_norm": 1.5002142849114388,
      "learning_rate": 1.4274585038564422e-05,
      "loss": 0.2273,
      "step": 5918
    },
    {
      "epoch": 3.7869481765834934,
      "grad_norm": 1.5087832306646,
      "learning_rate": 1.4272711647141884e-05,
      "loss": 0.2326,
      "step": 5919
    },
    {
      "epoch": 3.787587971849008,
      "grad_norm": 1.504761675527817,
      "learning_rate": 1.4270838072256708e-05,
      "loss": 0.2268,
      "step": 5920
    },
    {
      "epoch": 3.7882277671145235,
      "grad_norm": 1.4836980950669774,
      "learning_rate": 1.4268964313989342e-05,
      "loss": 0.251,
      "step": 5921
    },
    {
      "epoch": 3.7888675623800383,
      "grad_norm": 1.5264484055239913,
      "learning_rate": 1.426709037242024e-05,
      "loss": 0.2269,
      "step": 5922
    },
    {
      "epoch": 3.7895073576455536,
      "grad_norm": 1.5089841847213838,
      "learning_rate": 1.4265216247629871e-05,
      "loss": 0.2328,
      "step": 5923
    },
    {
      "epoch": 3.7901471529110684,
      "grad_norm": 1.3650943708833259,
      "learning_rate": 1.4263341939698706e-05,
      "loss": 0.195,
      "step": 5924
    },
    {
      "epoch": 3.7907869481765832,
      "grad_norm": 1.5130084009695075,
      "learning_rate": 1.4261467448707219e-05,
      "loss": 0.2299,
      "step": 5925
    },
    {
      "epoch": 3.7914267434420985,
      "grad_norm": 1.6014300965236612,
      "learning_rate": 1.4259592774735901e-05,
      "loss": 0.2594,
      "step": 5926
    },
    {
      "epoch": 3.792066538707614,
      "grad_norm": 1.4854516595729075,
      "learning_rate": 1.4257717917865246e-05,
      "loss": 0.2359,
      "step": 5927
    },
    {
      "epoch": 3.7927063339731286,
      "grad_norm": 1.579563447586488,
      "learning_rate": 1.4255842878175753e-05,
      "loss": 0.2459,
      "step": 5928
    },
    {
      "epoch": 3.7933461292386434,
      "grad_norm": 1.3406041678944973,
      "learning_rate": 1.4253967655747945e-05,
      "loss": 0.2033,
      "step": 5929
    },
    {
      "epoch": 3.7939859245041587,
      "grad_norm": 1.4192384496856774,
      "learning_rate": 1.4252092250662328e-05,
      "loss": 0.2138,
      "step": 5930
    },
    {
      "epoch": 3.7946257197696736,
      "grad_norm": 1.6279126789594982,
      "learning_rate": 1.4250216662999437e-05,
      "loss": 0.2604,
      "step": 5931
    },
    {
      "epoch": 3.795265515035189,
      "grad_norm": 1.5854477609141788,
      "learning_rate": 1.4248340892839801e-05,
      "loss": 0.2235,
      "step": 5932
    },
    {
      "epoch": 3.7959053103007037,
      "grad_norm": 1.5996888096985724,
      "learning_rate": 1.4246464940263964e-05,
      "loss": 0.244,
      "step": 5933
    },
    {
      "epoch": 3.796545105566219,
      "grad_norm": 1.577535680797323,
      "learning_rate": 1.424458880535248e-05,
      "loss": 0.2283,
      "step": 5934
    },
    {
      "epoch": 3.7971849008317338,
      "grad_norm": 1.516999233134332,
      "learning_rate": 1.4242712488185899e-05,
      "loss": 0.2407,
      "step": 5935
    },
    {
      "epoch": 3.797824696097249,
      "grad_norm": 1.4530462745131316,
      "learning_rate": 1.4240835988844794e-05,
      "loss": 0.223,
      "step": 5936
    },
    {
      "epoch": 3.798464491362764,
      "grad_norm": 1.5626496683344595,
      "learning_rate": 1.4238959307409732e-05,
      "loss": 0.2487,
      "step": 5937
    },
    {
      "epoch": 3.799104286628279,
      "grad_norm": 1.483649885286992,
      "learning_rate": 1.4237082443961301e-05,
      "loss": 0.2351,
      "step": 5938
    },
    {
      "epoch": 3.799744081893794,
      "grad_norm": 1.6054143311195381,
      "learning_rate": 1.4235205398580085e-05,
      "loss": 0.2491,
      "step": 5939
    },
    {
      "epoch": 3.800383877159309,
      "grad_norm": 1.4288387966471976,
      "learning_rate": 1.4233328171346686e-05,
      "loss": 0.2263,
      "step": 5940
    },
    {
      "epoch": 3.801023672424824,
      "grad_norm": 1.495658256800239,
      "learning_rate": 1.4231450762341703e-05,
      "loss": 0.2276,
      "step": 5941
    },
    {
      "epoch": 3.8016634676903394,
      "grad_norm": 1.394329823683921,
      "learning_rate": 1.4229573171645751e-05,
      "loss": 0.2112,
      "step": 5942
    },
    {
      "epoch": 3.802303262955854,
      "grad_norm": 1.525065200257798,
      "learning_rate": 1.4227695399339454e-05,
      "loss": 0.2487,
      "step": 5943
    },
    {
      "epoch": 3.802943058221369,
      "grad_norm": 1.354004418718374,
      "learning_rate": 1.4225817445503433e-05,
      "loss": 0.2135,
      "step": 5944
    },
    {
      "epoch": 3.8035828534868843,
      "grad_norm": 1.513581619112431,
      "learning_rate": 1.4223939310218329e-05,
      "loss": 0.2333,
      "step": 5945
    },
    {
      "epoch": 3.804222648752399,
      "grad_norm": 1.4302130991074602,
      "learning_rate": 1.4222060993564788e-05,
      "loss": 0.2113,
      "step": 5946
    },
    {
      "epoch": 3.8048624440179144,
      "grad_norm": 1.5028513858242196,
      "learning_rate": 1.4220182495623454e-05,
      "loss": 0.2245,
      "step": 5947
    },
    {
      "epoch": 3.8055022392834292,
      "grad_norm": 1.486240246073738,
      "learning_rate": 1.4218303816474994e-05,
      "loss": 0.2247,
      "step": 5948
    },
    {
      "epoch": 3.8061420345489445,
      "grad_norm": 1.4274522487061323,
      "learning_rate": 1.4216424956200072e-05,
      "loss": 0.2316,
      "step": 5949
    },
    {
      "epoch": 3.8067818298144593,
      "grad_norm": 1.4592678354791682,
      "learning_rate": 1.421454591487936e-05,
      "loss": 0.2338,
      "step": 5950
    },
    {
      "epoch": 3.807421625079974,
      "grad_norm": 1.602893295485456,
      "learning_rate": 1.4212666692593546e-05,
      "loss": 0.2218,
      "step": 5951
    },
    {
      "epoch": 3.8080614203454894,
      "grad_norm": 1.4035095366250172,
      "learning_rate": 1.4210787289423315e-05,
      "loss": 0.2386,
      "step": 5952
    },
    {
      "epoch": 3.8087012156110047,
      "grad_norm": 1.4566137151240217,
      "learning_rate": 1.4208907705449374e-05,
      "loss": 0.2063,
      "step": 5953
    },
    {
      "epoch": 3.8093410108765196,
      "grad_norm": 1.3910190990216549,
      "learning_rate": 1.4207027940752418e-05,
      "loss": 0.2006,
      "step": 5954
    },
    {
      "epoch": 3.8099808061420344,
      "grad_norm": 1.3916756052591155,
      "learning_rate": 1.420514799541317e-05,
      "loss": 0.2236,
      "step": 5955
    },
    {
      "epoch": 3.8106206014075497,
      "grad_norm": 1.4645615547464872,
      "learning_rate": 1.4203267869512343e-05,
      "loss": 0.2286,
      "step": 5956
    },
    {
      "epoch": 3.8112603966730645,
      "grad_norm": 1.4047118988343694,
      "learning_rate": 1.4201387563130673e-05,
      "loss": 0.2042,
      "step": 5957
    },
    {
      "epoch": 3.8119001919385798,
      "grad_norm": 1.4707548089796896,
      "learning_rate": 1.4199507076348895e-05,
      "loss": 0.229,
      "step": 5958
    },
    {
      "epoch": 3.8125399872040946,
      "grad_norm": 1.4846440903998024,
      "learning_rate": 1.4197626409247752e-05,
      "loss": 0.2298,
      "step": 5959
    },
    {
      "epoch": 3.81317978246961,
      "grad_norm": 1.5971972345900085,
      "learning_rate": 1.4195745561907996e-05,
      "loss": 0.2788,
      "step": 5960
    },
    {
      "epoch": 3.8138195777351247,
      "grad_norm": 1.6364925959677044,
      "learning_rate": 1.4193864534410396e-05,
      "loss": 0.2587,
      "step": 5961
    },
    {
      "epoch": 3.8144593730006395,
      "grad_norm": 1.4653966391972995,
      "learning_rate": 1.4191983326835707e-05,
      "loss": 0.2261,
      "step": 5962
    },
    {
      "epoch": 3.815099168266155,
      "grad_norm": 1.3879507899134809,
      "learning_rate": 1.4190101939264716e-05,
      "loss": 0.2104,
      "step": 5963
    },
    {
      "epoch": 3.81573896353167,
      "grad_norm": 1.4554052376490278,
      "learning_rate": 1.4188220371778199e-05,
      "loss": 0.2442,
      "step": 5964
    },
    {
      "epoch": 3.816378758797185,
      "grad_norm": 1.5606490892597542,
      "learning_rate": 1.4186338624456948e-05,
      "loss": 0.263,
      "step": 5965
    },
    {
      "epoch": 3.8170185540626997,
      "grad_norm": 1.5117488717121401,
      "learning_rate": 1.4184456697381764e-05,
      "loss": 0.2299,
      "step": 5966
    },
    {
      "epoch": 3.817658349328215,
      "grad_norm": 1.5043480891270584,
      "learning_rate": 1.4182574590633453e-05,
      "loss": 0.2112,
      "step": 5967
    },
    {
      "epoch": 3.81829814459373,
      "grad_norm": 1.4758810136035763,
      "learning_rate": 1.4180692304292828e-05,
      "loss": 0.2386,
      "step": 5968
    },
    {
      "epoch": 3.818937939859245,
      "grad_norm": 1.365868692224636,
      "learning_rate": 1.4178809838440715e-05,
      "loss": 0.2104,
      "step": 5969
    },
    {
      "epoch": 3.81957773512476,
      "grad_norm": 1.2946648307337805,
      "learning_rate": 1.417692719315794e-05,
      "loss": 0.206,
      "step": 5970
    },
    {
      "epoch": 3.8202175303902752,
      "grad_norm": 1.3907668167969973,
      "learning_rate": 1.4175044368525344e-05,
      "loss": 0.2163,
      "step": 5971
    },
    {
      "epoch": 3.82085732565579,
      "grad_norm": 1.4191721190920812,
      "learning_rate": 1.4173161364623767e-05,
      "loss": 0.2223,
      "step": 5972
    },
    {
      "epoch": 3.8214971209213053,
      "grad_norm": 1.4958168363930961,
      "learning_rate": 1.4171278181534066e-05,
      "loss": 0.2313,
      "step": 5973
    },
    {
      "epoch": 3.82213691618682,
      "grad_norm": 1.4515560843029525,
      "learning_rate": 1.4169394819337096e-05,
      "loss": 0.2277,
      "step": 5974
    },
    {
      "epoch": 3.8227767114523354,
      "grad_norm": 1.4095911585334215,
      "learning_rate": 1.4167511278113732e-05,
      "loss": 0.216,
      "step": 5975
    },
    {
      "epoch": 3.8234165067178503,
      "grad_norm": 1.443479233916379,
      "learning_rate": 1.4165627557944847e-05,
      "loss": 0.2333,
      "step": 5976
    },
    {
      "epoch": 3.824056301983365,
      "grad_norm": 1.4306051741547559,
      "learning_rate": 1.4163743658911325e-05,
      "loss": 0.2239,
      "step": 5977
    },
    {
      "epoch": 3.8246960972488804,
      "grad_norm": 1.44079127301795,
      "learning_rate": 1.4161859581094057e-05,
      "loss": 0.2114,
      "step": 5978
    },
    {
      "epoch": 3.8253358925143957,
      "grad_norm": 1.4859875108256393,
      "learning_rate": 1.4159975324573942e-05,
      "loss": 0.2191,
      "step": 5979
    },
    {
      "epoch": 3.8259756877799105,
      "grad_norm": 1.5528125775001131,
      "learning_rate": 1.4158090889431888e-05,
      "loss": 0.2329,
      "step": 5980
    },
    {
      "epoch": 3.8266154830454253,
      "grad_norm": 1.4526983620808773,
      "learning_rate": 1.4156206275748808e-05,
      "loss": 0.2191,
      "step": 5981
    },
    {
      "epoch": 3.8272552783109406,
      "grad_norm": 1.279537152140492,
      "learning_rate": 1.415432148360562e-05,
      "loss": 0.1934,
      "step": 5982
    },
    {
      "epoch": 3.8278950735764554,
      "grad_norm": 1.4977867591346896,
      "learning_rate": 1.4152436513083259e-05,
      "loss": 0.2237,
      "step": 5983
    },
    {
      "epoch": 3.8285348688419707,
      "grad_norm": 1.3419975592590385,
      "learning_rate": 1.415055136426266e-05,
      "loss": 0.2157,
      "step": 5984
    },
    {
      "epoch": 3.8291746641074855,
      "grad_norm": 1.4131257045172394,
      "learning_rate": 1.4148666037224771e-05,
      "loss": 0.2239,
      "step": 5985
    },
    {
      "epoch": 3.829814459373001,
      "grad_norm": 1.5863265267727689,
      "learning_rate": 1.4146780532050544e-05,
      "loss": 0.2501,
      "step": 5986
    },
    {
      "epoch": 3.8304542546385156,
      "grad_norm": 1.5217837802548608,
      "learning_rate": 1.4144894848820933e-05,
      "loss": 0.2481,
      "step": 5987
    },
    {
      "epoch": 3.8310940499040305,
      "grad_norm": 1.3133809887603127,
      "learning_rate": 1.4143008987616913e-05,
      "loss": 0.2175,
      "step": 5988
    },
    {
      "epoch": 3.8317338451695457,
      "grad_norm": 1.4774017943395152,
      "learning_rate": 1.4141122948519454e-05,
      "loss": 0.2263,
      "step": 5989
    },
    {
      "epoch": 3.832373640435061,
      "grad_norm": 1.506004555983328,
      "learning_rate": 1.4139236731609545e-05,
      "loss": 0.2371,
      "step": 5990
    },
    {
      "epoch": 3.833013435700576,
      "grad_norm": 1.5748103544720542,
      "learning_rate": 1.4137350336968175e-05,
      "loss": 0.2516,
      "step": 5991
    },
    {
      "epoch": 3.8336532309660907,
      "grad_norm": 1.4282584885839895,
      "learning_rate": 1.4135463764676338e-05,
      "loss": 0.2275,
      "step": 5992
    },
    {
      "epoch": 3.834293026231606,
      "grad_norm": 1.421764896144006,
      "learning_rate": 1.4133577014815045e-05,
      "loss": 0.2343,
      "step": 5993
    },
    {
      "epoch": 3.834932821497121,
      "grad_norm": 1.3656031324698616,
      "learning_rate": 1.4131690087465308e-05,
      "loss": 0.2186,
      "step": 5994
    },
    {
      "epoch": 3.835572616762636,
      "grad_norm": 1.2883996139323264,
      "learning_rate": 1.4129802982708147e-05,
      "loss": 0.2108,
      "step": 5995
    },
    {
      "epoch": 3.836212412028151,
      "grad_norm": 1.5711008144952756,
      "learning_rate": 1.4127915700624595e-05,
      "loss": 0.2569,
      "step": 5996
    },
    {
      "epoch": 3.836852207293666,
      "grad_norm": 1.4975467491949206,
      "learning_rate": 1.4126028241295683e-05,
      "loss": 0.2297,
      "step": 5997
    },
    {
      "epoch": 3.837492002559181,
      "grad_norm": 1.306713770184282,
      "learning_rate": 1.4124140604802458e-05,
      "loss": 0.1834,
      "step": 5998
    },
    {
      "epoch": 3.838131797824696,
      "grad_norm": 1.5639429368784679,
      "learning_rate": 1.4122252791225974e-05,
      "loss": 0.2345,
      "step": 5999
    },
    {
      "epoch": 3.838771593090211,
      "grad_norm": 1.6128208352922666,
      "learning_rate": 1.4120364800647288e-05,
      "loss": 0.2746,
      "step": 6000
    },
    {
      "epoch": 3.8394113883557264,
      "grad_norm": 1.5270985206466323,
      "learning_rate": 1.4118476633147467e-05,
      "loss": 0.2328,
      "step": 6001
    },
    {
      "epoch": 3.840051183621241,
      "grad_norm": 1.519830451849325,
      "learning_rate": 1.4116588288807584e-05,
      "loss": 0.2238,
      "step": 6002
    },
    {
      "epoch": 3.840690978886756,
      "grad_norm": 1.4773737227059485,
      "learning_rate": 1.4114699767708725e-05,
      "loss": 0.2144,
      "step": 6003
    },
    {
      "epoch": 3.8413307741522713,
      "grad_norm": 1.6115222120898218,
      "learning_rate": 1.4112811069931973e-05,
      "loss": 0.2573,
      "step": 6004
    },
    {
      "epoch": 3.8419705694177866,
      "grad_norm": 1.5861319564017207,
      "learning_rate": 1.4110922195558433e-05,
      "loss": 0.2311,
      "step": 6005
    },
    {
      "epoch": 3.8426103646833014,
      "grad_norm": 1.5009926326879197,
      "learning_rate": 1.4109033144669206e-05,
      "loss": 0.2195,
      "step": 6006
    },
    {
      "epoch": 3.8432501599488162,
      "grad_norm": 1.4722302104995684,
      "learning_rate": 1.4107143917345405e-05,
      "loss": 0.2284,
      "step": 6007
    },
    {
      "epoch": 3.8438899552143315,
      "grad_norm": 1.4439433262450045,
      "learning_rate": 1.410525451366815e-05,
      "loss": 0.222,
      "step": 6008
    },
    {
      "epoch": 3.8445297504798464,
      "grad_norm": 1.4673763794979255,
      "learning_rate": 1.4103364933718571e-05,
      "loss": 0.1908,
      "step": 6009
    },
    {
      "epoch": 3.8451695457453616,
      "grad_norm": 1.5111990323749847,
      "learning_rate": 1.4101475177577801e-05,
      "loss": 0.219,
      "step": 6010
    },
    {
      "epoch": 3.8458093410108765,
      "grad_norm": 1.5228264283409902,
      "learning_rate": 1.409958524532698e-05,
      "loss": 0.2506,
      "step": 6011
    },
    {
      "epoch": 3.8464491362763917,
      "grad_norm": 1.5532989854222177,
      "learning_rate": 1.4097695137047263e-05,
      "loss": 0.2215,
      "step": 6012
    },
    {
      "epoch": 3.8470889315419066,
      "grad_norm": 1.390512451005287,
      "learning_rate": 1.4095804852819806e-05,
      "loss": 0.2143,
      "step": 6013
    },
    {
      "epoch": 3.8477287268074214,
      "grad_norm": 1.4597340637131526,
      "learning_rate": 1.4093914392725771e-05,
      "loss": 0.2304,
      "step": 6014
    },
    {
      "epoch": 3.8483685220729367,
      "grad_norm": 1.4774383735079102,
      "learning_rate": 1.4092023756846339e-05,
      "loss": 0.257,
      "step": 6015
    },
    {
      "epoch": 3.849008317338452,
      "grad_norm": 1.4610431537573922,
      "learning_rate": 1.4090132945262687e-05,
      "loss": 0.2161,
      "step": 6016
    },
    {
      "epoch": 3.8496481126039668,
      "grad_norm": 1.522576555904611,
      "learning_rate": 1.4088241958055997e-05,
      "loss": 0.2353,
      "step": 6017
    },
    {
      "epoch": 3.8502879078694816,
      "grad_norm": 1.3990317046808047,
      "learning_rate": 1.4086350795307474e-05,
      "loss": 0.209,
      "step": 6018
    },
    {
      "epoch": 3.850927703134997,
      "grad_norm": 1.5297007827727631,
      "learning_rate": 1.4084459457098314e-05,
      "loss": 0.238,
      "step": 6019
    },
    {
      "epoch": 3.8515674984005117,
      "grad_norm": 1.4987426924142224,
      "learning_rate": 1.4082567943509733e-05,
      "loss": 0.2176,
      "step": 6020
    },
    {
      "epoch": 3.852207293666027,
      "grad_norm": 1.4986482343924228,
      "learning_rate": 1.4080676254622945e-05,
      "loss": 0.2328,
      "step": 6021
    },
    {
      "epoch": 3.852847088931542,
      "grad_norm": 1.535887506590649,
      "learning_rate": 1.4078784390519178e-05,
      "loss": 0.2436,
      "step": 6022
    },
    {
      "epoch": 3.853486884197057,
      "grad_norm": 1.5900108347172066,
      "learning_rate": 1.4076892351279665e-05,
      "loss": 0.2328,
      "step": 6023
    },
    {
      "epoch": 3.854126679462572,
      "grad_norm": 1.6530184855027774,
      "learning_rate": 1.4075000136985648e-05,
      "loss": 0.2431,
      "step": 6024
    },
    {
      "epoch": 3.8547664747280868,
      "grad_norm": 1.4450504645588964,
      "learning_rate": 1.4073107747718373e-05,
      "loss": 0.2243,
      "step": 6025
    },
    {
      "epoch": 3.855406269993602,
      "grad_norm": 1.485330924435305,
      "learning_rate": 1.4071215183559095e-05,
      "loss": 0.2538,
      "step": 6026
    },
    {
      "epoch": 3.8560460652591173,
      "grad_norm": 1.5731982205022095,
      "learning_rate": 1.4069322444589082e-05,
      "loss": 0.2518,
      "step": 6027
    },
    {
      "epoch": 3.856685860524632,
      "grad_norm": 1.5899935295262204,
      "learning_rate": 1.40674295308896e-05,
      "loss": 0.2358,
      "step": 6028
    },
    {
      "epoch": 3.857325655790147,
      "grad_norm": 1.4709063537026275,
      "learning_rate": 1.406553644254193e-05,
      "loss": 0.2393,
      "step": 6029
    },
    {
      "epoch": 3.8579654510556622,
      "grad_norm": 1.4432662246345498,
      "learning_rate": 1.4063643179627358e-05,
      "loss": 0.2286,
      "step": 6030
    },
    {
      "epoch": 3.858605246321177,
      "grad_norm": 1.3100535754665101,
      "learning_rate": 1.4061749742227177e-05,
      "loss": 0.2019,
      "step": 6031
    },
    {
      "epoch": 3.8592450415866923,
      "grad_norm": 1.4893169831384352,
      "learning_rate": 1.4059856130422684e-05,
      "loss": 0.2511,
      "step": 6032
    },
    {
      "epoch": 3.859884836852207,
      "grad_norm": 1.475728110106718,
      "learning_rate": 1.4057962344295196e-05,
      "loss": 0.2236,
      "step": 6033
    },
    {
      "epoch": 3.8605246321177225,
      "grad_norm": 1.4951458124209678,
      "learning_rate": 1.4056068383926018e-05,
      "loss": 0.2122,
      "step": 6034
    },
    {
      "epoch": 3.8611644273832373,
      "grad_norm": 1.5062232289228954,
      "learning_rate": 1.4054174249396482e-05,
      "loss": 0.1933,
      "step": 6035
    },
    {
      "epoch": 3.8618042226487526,
      "grad_norm": 1.5066776658518106,
      "learning_rate": 1.4052279940787914e-05,
      "loss": 0.2485,
      "step": 6036
    },
    {
      "epoch": 3.8624440179142674,
      "grad_norm": 1.4240019026773765,
      "learning_rate": 1.4050385458181652e-05,
      "loss": 0.2214,
      "step": 6037
    },
    {
      "epoch": 3.8630838131797827,
      "grad_norm": 1.5277805177336352,
      "learning_rate": 1.4048490801659048e-05,
      "loss": 0.2207,
      "step": 6038
    },
    {
      "epoch": 3.8637236084452975,
      "grad_norm": 1.4945038927550969,
      "learning_rate": 1.404659597130145e-05,
      "loss": 0.2304,
      "step": 6039
    },
    {
      "epoch": 3.8643634037108123,
      "grad_norm": 1.5168483886367112,
      "learning_rate": 1.4044700967190218e-05,
      "loss": 0.2432,
      "step": 6040
    },
    {
      "epoch": 3.8650031989763276,
      "grad_norm": 1.537008242160798,
      "learning_rate": 1.4042805789406722e-05,
      "loss": 0.2248,
      "step": 6041
    },
    {
      "epoch": 3.865642994241843,
      "grad_norm": 1.4246889563484966,
      "learning_rate": 1.4040910438032333e-05,
      "loss": 0.1996,
      "step": 6042
    },
    {
      "epoch": 3.8662827895073577,
      "grad_norm": 1.5441834086640045,
      "learning_rate": 1.4039014913148443e-05,
      "loss": 0.2006,
      "step": 6043
    },
    {
      "epoch": 3.8669225847728725,
      "grad_norm": 1.531558631296265,
      "learning_rate": 1.4037119214836436e-05,
      "loss": 0.2148,
      "step": 6044
    },
    {
      "epoch": 3.867562380038388,
      "grad_norm": 1.502770636950017,
      "learning_rate": 1.4035223343177713e-05,
      "loss": 0.2318,
      "step": 6045
    },
    {
      "epoch": 3.8682021753039026,
      "grad_norm": 1.326798540541587,
      "learning_rate": 1.4033327298253676e-05,
      "loss": 0.2221,
      "step": 6046
    },
    {
      "epoch": 3.868841970569418,
      "grad_norm": 1.4633738785628148,
      "learning_rate": 1.4031431080145738e-05,
      "loss": 0.2403,
      "step": 6047
    },
    {
      "epoch": 3.8694817658349328,
      "grad_norm": 1.5216805221247423,
      "learning_rate": 1.4029534688935325e-05,
      "loss": 0.2577,
      "step": 6048
    },
    {
      "epoch": 3.870121561100448,
      "grad_norm": 1.4522230617260055,
      "learning_rate": 1.4027638124703855e-05,
      "loss": 0.1909,
      "step": 6049
    },
    {
      "epoch": 3.870761356365963,
      "grad_norm": 1.4974972219606428,
      "learning_rate": 1.4025741387532775e-05,
      "loss": 0.2343,
      "step": 6050
    },
    {
      "epoch": 3.8714011516314777,
      "grad_norm": 1.5592361592943693,
      "learning_rate": 1.4023844477503514e-05,
      "loss": 0.2425,
      "step": 6051
    },
    {
      "epoch": 3.872040946896993,
      "grad_norm": 1.413017096702922,
      "learning_rate": 1.4021947394697533e-05,
      "loss": 0.2139,
      "step": 6052
    },
    {
      "epoch": 3.8726807421625082,
      "grad_norm": 1.4839009490834334,
      "learning_rate": 1.4020050139196287e-05,
      "loss": 0.2311,
      "step": 6053
    },
    {
      "epoch": 3.873320537428023,
      "grad_norm": 1.5943076875513227,
      "learning_rate": 1.4018152711081236e-05,
      "loss": 0.2654,
      "step": 6054
    },
    {
      "epoch": 3.873960332693538,
      "grad_norm": 1.475700202758805,
      "learning_rate": 1.4016255110433857e-05,
      "loss": 0.2453,
      "step": 6055
    },
    {
      "epoch": 3.874600127959053,
      "grad_norm": 1.466084099239846,
      "learning_rate": 1.4014357337335626e-05,
      "loss": 0.2261,
      "step": 6056
    },
    {
      "epoch": 3.875239923224568,
      "grad_norm": 1.4317481597196744,
      "learning_rate": 1.4012459391868032e-05,
      "loss": 0.2293,
      "step": 6057
    },
    {
      "epoch": 3.8758797184900833,
      "grad_norm": 1.4333100000666112,
      "learning_rate": 1.4010561274112573e-05,
      "loss": 0.2286,
      "step": 6058
    },
    {
      "epoch": 3.876519513755598,
      "grad_norm": 1.3984614171046237,
      "learning_rate": 1.4008662984150742e-05,
      "loss": 0.223,
      "step": 6059
    },
    {
      "epoch": 3.8771593090211134,
      "grad_norm": 1.3570758802705258,
      "learning_rate": 1.4006764522064059e-05,
      "loss": 0.1923,
      "step": 6060
    },
    {
      "epoch": 3.877799104286628,
      "grad_norm": 1.455546689852338,
      "learning_rate": 1.4004865887934032e-05,
      "loss": 0.2234,
      "step": 6061
    },
    {
      "epoch": 3.878438899552143,
      "grad_norm": 1.4166031350600978,
      "learning_rate": 1.4002967081842184e-05,
      "loss": 0.215,
      "step": 6062
    },
    {
      "epoch": 3.8790786948176583,
      "grad_norm": 1.5324938410680788,
      "learning_rate": 1.4001068103870058e-05,
      "loss": 0.2428,
      "step": 6063
    },
    {
      "epoch": 3.8797184900831736,
      "grad_norm": 1.3769841352070122,
      "learning_rate": 1.3999168954099178e-05,
      "loss": 0.2034,
      "step": 6064
    },
    {
      "epoch": 3.8803582853486884,
      "grad_norm": 1.4092002509226995,
      "learning_rate": 1.3997269632611101e-05,
      "loss": 0.2362,
      "step": 6065
    },
    {
      "epoch": 3.8809980806142033,
      "grad_norm": 1.3541188212978994,
      "learning_rate": 1.3995370139487374e-05,
      "loss": 0.207,
      "step": 6066
    },
    {
      "epoch": 3.8816378758797185,
      "grad_norm": 1.4782968517213777,
      "learning_rate": 1.3993470474809562e-05,
      "loss": 0.24,
      "step": 6067
    },
    {
      "epoch": 3.8822776711452334,
      "grad_norm": 1.5426012342498332,
      "learning_rate": 1.3991570638659232e-05,
      "loss": 0.2524,
      "step": 6068
    },
    {
      "epoch": 3.8829174664107486,
      "grad_norm": 1.5952107791804664,
      "learning_rate": 1.398967063111796e-05,
      "loss": 0.2403,
      "step": 6069
    },
    {
      "epoch": 3.8835572616762635,
      "grad_norm": 1.5169083705750017,
      "learning_rate": 1.3987770452267328e-05,
      "loss": 0.2309,
      "step": 6070
    },
    {
      "epoch": 3.8841970569417787,
      "grad_norm": 1.4466624803527335,
      "learning_rate": 1.3985870102188925e-05,
      "loss": 0.193,
      "step": 6071
    },
    {
      "epoch": 3.8848368522072936,
      "grad_norm": 1.3519217724490564,
      "learning_rate": 1.398396958096435e-05,
      "loss": 0.2145,
      "step": 6072
    },
    {
      "epoch": 3.885476647472809,
      "grad_norm": 1.5645463483283286,
      "learning_rate": 1.3982068888675208e-05,
      "loss": 0.2381,
      "step": 6073
    },
    {
      "epoch": 3.8861164427383237,
      "grad_norm": 1.603896457122866,
      "learning_rate": 1.398016802540311e-05,
      "loss": 0.2552,
      "step": 6074
    },
    {
      "epoch": 3.886756238003839,
      "grad_norm": 1.586753139672477,
      "learning_rate": 1.3978266991229683e-05,
      "loss": 0.2553,
      "step": 6075
    },
    {
      "epoch": 3.887396033269354,
      "grad_norm": 1.5747443901796736,
      "learning_rate": 1.3976365786236542e-05,
      "loss": 0.2201,
      "step": 6076
    },
    {
      "epoch": 3.8880358285348686,
      "grad_norm": 1.4599178207927923,
      "learning_rate": 1.397446441050533e-05,
      "loss": 0.2162,
      "step": 6077
    },
    {
      "epoch": 3.888675623800384,
      "grad_norm": 1.4529411594293506,
      "learning_rate": 1.397256286411769e-05,
      "loss": 0.2276,
      "step": 6078
    },
    {
      "epoch": 3.889315419065899,
      "grad_norm": 1.5011843543688703,
      "learning_rate": 1.3970661147155263e-05,
      "loss": 0.2229,
      "step": 6079
    },
    {
      "epoch": 3.889955214331414,
      "grad_norm": 1.5172150213527147,
      "learning_rate": 1.3968759259699709e-05,
      "loss": 0.2117,
      "step": 6080
    },
    {
      "epoch": 3.890595009596929,
      "grad_norm": 1.4829782132973899,
      "learning_rate": 1.3966857201832694e-05,
      "loss": 0.2064,
      "step": 6081
    },
    {
      "epoch": 3.891234804862444,
      "grad_norm": 1.4028041205571213,
      "learning_rate": 1.3964954973635884e-05,
      "loss": 0.2264,
      "step": 6082
    },
    {
      "epoch": 3.891874600127959,
      "grad_norm": 1.6068782200443439,
      "learning_rate": 1.3963052575190963e-05,
      "loss": 0.2761,
      "step": 6083
    },
    {
      "epoch": 3.892514395393474,
      "grad_norm": 1.3553868167014647,
      "learning_rate": 1.3961150006579617e-05,
      "loss": 0.1913,
      "step": 6084
    },
    {
      "epoch": 3.893154190658989,
      "grad_norm": 1.4835712989742824,
      "learning_rate": 1.3959247267883533e-05,
      "loss": 0.2479,
      "step": 6085
    },
    {
      "epoch": 3.8937939859245043,
      "grad_norm": 1.4719247264510793,
      "learning_rate": 1.3957344359184414e-05,
      "loss": 0.2445,
      "step": 6086
    },
    {
      "epoch": 3.894433781190019,
      "grad_norm": 1.5248772061462395,
      "learning_rate": 1.395544128056397e-05,
      "loss": 0.2393,
      "step": 6087
    },
    {
      "epoch": 3.895073576455534,
      "grad_norm": 1.4426014130978861,
      "learning_rate": 1.395353803210391e-05,
      "loss": 0.2145,
      "step": 6088
    },
    {
      "epoch": 3.8957133717210493,
      "grad_norm": 1.5268370473347506,
      "learning_rate": 1.395163461388596e-05,
      "loss": 0.2076,
      "step": 6089
    },
    {
      "epoch": 3.8963531669865645,
      "grad_norm": 1.474906773228786,
      "learning_rate": 1.3949731025991853e-05,
      "loss": 0.2395,
      "step": 6090
    },
    {
      "epoch": 3.8969929622520794,
      "grad_norm": 1.4518278978845383,
      "learning_rate": 1.3947827268503323e-05,
      "loss": 0.1963,
      "step": 6091
    },
    {
      "epoch": 3.897632757517594,
      "grad_norm": 1.402786952107276,
      "learning_rate": 1.3945923341502108e-05,
      "loss": 0.1911,
      "step": 6092
    },
    {
      "epoch": 3.8982725527831095,
      "grad_norm": 1.567608380416922,
      "learning_rate": 1.3944019245069963e-05,
      "loss": 0.2165,
      "step": 6093
    },
    {
      "epoch": 3.8989123480486243,
      "grad_norm": 1.5216327727973535,
      "learning_rate": 1.394211497928865e-05,
      "loss": 0.2297,
      "step": 6094
    },
    {
      "epoch": 3.8995521433141396,
      "grad_norm": 1.499203839697814,
      "learning_rate": 1.3940210544239934e-05,
      "loss": 0.2245,
      "step": 6095
    },
    {
      "epoch": 3.9001919385796544,
      "grad_norm": 1.5100746327563317,
      "learning_rate": 1.3938305940005583e-05,
      "loss": 0.2421,
      "step": 6096
    },
    {
      "epoch": 3.9008317338451697,
      "grad_norm": 1.5690364875297362,
      "learning_rate": 1.3936401166667382e-05,
      "loss": 0.2153,
      "step": 6097
    },
    {
      "epoch": 3.9014715291106845,
      "grad_norm": 1.539795695097083,
      "learning_rate": 1.3934496224307117e-05,
      "loss": 0.2691,
      "step": 6098
    },
    {
      "epoch": 3.9021113243761993,
      "grad_norm": 1.4285603690086834,
      "learning_rate": 1.3932591113006584e-05,
      "loss": 0.1977,
      "step": 6099
    },
    {
      "epoch": 3.9027511196417146,
      "grad_norm": 1.4351779279183978,
      "learning_rate": 1.3930685832847584e-05,
      "loss": 0.2034,
      "step": 6100
    },
    {
      "epoch": 3.90339091490723,
      "grad_norm": 1.7948461090890664,
      "learning_rate": 1.3928780383911927e-05,
      "loss": 0.2795,
      "step": 6101
    },
    {
      "epoch": 3.9040307101727447,
      "grad_norm": 1.4511342777936016,
      "learning_rate": 1.3926874766281428e-05,
      "loss": 0.2082,
      "step": 6102
    },
    {
      "epoch": 3.9046705054382596,
      "grad_norm": 1.3324782524051688,
      "learning_rate": 1.392496898003791e-05,
      "loss": 0.1856,
      "step": 6103
    },
    {
      "epoch": 3.905310300703775,
      "grad_norm": 1.3404693184314302,
      "learning_rate": 1.3923063025263206e-05,
      "loss": 0.1955,
      "step": 6104
    },
    {
      "epoch": 3.90595009596929,
      "grad_norm": 1.3874636920601278,
      "learning_rate": 1.392115690203916e-05,
      "loss": 0.2124,
      "step": 6105
    },
    {
      "epoch": 3.906589891234805,
      "grad_norm": 1.4565674828199258,
      "learning_rate": 1.3919250610447608e-05,
      "loss": 0.217,
      "step": 6106
    },
    {
      "epoch": 3.9072296865003198,
      "grad_norm": 1.5234623578160307,
      "learning_rate": 1.3917344150570406e-05,
      "loss": 0.2124,
      "step": 6107
    },
    {
      "epoch": 3.907869481765835,
      "grad_norm": 1.465761936116586,
      "learning_rate": 1.3915437522489414e-05,
      "loss": 0.2256,
      "step": 6108
    },
    {
      "epoch": 3.90850927703135,
      "grad_norm": 1.373651888147878,
      "learning_rate": 1.3913530726286502e-05,
      "loss": 0.2002,
      "step": 6109
    },
    {
      "epoch": 3.909149072296865,
      "grad_norm": 1.3721852417724836,
      "learning_rate": 1.391162376204354e-05,
      "loss": 0.2045,
      "step": 6110
    },
    {
      "epoch": 3.90978886756238,
      "grad_norm": 1.4303646218064283,
      "learning_rate": 1.3909716629842414e-05,
      "loss": 0.2025,
      "step": 6111
    },
    {
      "epoch": 3.9104286628278953,
      "grad_norm": 1.3970812638814265,
      "learning_rate": 1.390780932976501e-05,
      "loss": 0.2034,
      "step": 6112
    },
    {
      "epoch": 3.91106845809341,
      "grad_norm": 1.3554856670480822,
      "learning_rate": 1.3905901861893223e-05,
      "loss": 0.2189,
      "step": 6113
    },
    {
      "epoch": 3.911708253358925,
      "grad_norm": 1.4162525481480297,
      "learning_rate": 1.3903994226308961e-05,
      "loss": 0.2284,
      "step": 6114
    },
    {
      "epoch": 3.91234804862444,
      "grad_norm": 1.4291055175847256,
      "learning_rate": 1.3902086423094131e-05,
      "loss": 0.2293,
      "step": 6115
    },
    {
      "epoch": 3.9129878438899555,
      "grad_norm": 1.367040933751415,
      "learning_rate": 1.390017845233065e-05,
      "loss": 0.2221,
      "step": 6116
    },
    {
      "epoch": 3.9136276391554703,
      "grad_norm": 1.4887100959875568,
      "learning_rate": 1.3898270314100445e-05,
      "loss": 0.242,
      "step": 6117
    },
    {
      "epoch": 3.914267434420985,
      "grad_norm": 1.4131246150418768,
      "learning_rate": 1.3896362008485446e-05,
      "loss": 0.221,
      "step": 6118
    },
    {
      "epoch": 3.9149072296865004,
      "grad_norm": 1.4863237768475197,
      "learning_rate": 1.3894453535567593e-05,
      "loss": 0.2325,
      "step": 6119
    },
    {
      "epoch": 3.9155470249520152,
      "grad_norm": 1.3301791925009148,
      "learning_rate": 1.3892544895428837e-05,
      "loss": 0.2028,
      "step": 6120
    },
    {
      "epoch": 3.9161868202175305,
      "grad_norm": 1.4892841073728362,
      "learning_rate": 1.3890636088151126e-05,
      "loss": 0.2248,
      "step": 6121
    },
    {
      "epoch": 3.9168266154830453,
      "grad_norm": 1.5229427578356147,
      "learning_rate": 1.3888727113816422e-05,
      "loss": 0.2311,
      "step": 6122
    },
    {
      "epoch": 3.9174664107485606,
      "grad_norm": 1.484149069628505,
      "learning_rate": 1.3886817972506693e-05,
      "loss": 0.2266,
      "step": 6123
    },
    {
      "epoch": 3.9181062060140754,
      "grad_norm": 1.3458633659420898,
      "learning_rate": 1.388490866430391e-05,
      "loss": 0.2136,
      "step": 6124
    },
    {
      "epoch": 3.9187460012795903,
      "grad_norm": 1.4675772139999415,
      "learning_rate": 1.3882999189290066e-05,
      "loss": 0.2187,
      "step": 6125
    },
    {
      "epoch": 3.9193857965451055,
      "grad_norm": 1.4963710200072362,
      "learning_rate": 1.388108954754714e-05,
      "loss": 0.2322,
      "step": 6126
    },
    {
      "epoch": 3.920025591810621,
      "grad_norm": 1.4153378772947611,
      "learning_rate": 1.3879179739157133e-05,
      "loss": 0.2124,
      "step": 6127
    },
    {
      "epoch": 3.9206653870761357,
      "grad_norm": 1.337936502050901,
      "learning_rate": 1.387726976420205e-05,
      "loss": 0.2038,
      "step": 6128
    },
    {
      "epoch": 3.9213051823416505,
      "grad_norm": 1.512558218460647,
      "learning_rate": 1.3875359622763902e-05,
      "loss": 0.2398,
      "step": 6129
    },
    {
      "epoch": 3.9219449776071658,
      "grad_norm": 1.4861874140953886,
      "learning_rate": 1.3873449314924706e-05,
      "loss": 0.2325,
      "step": 6130
    },
    {
      "epoch": 3.9225847728726806,
      "grad_norm": 1.2693461799501398,
      "learning_rate": 1.387153884076648e-05,
      "loss": 0.2116,
      "step": 6131
    },
    {
      "epoch": 3.923224568138196,
      "grad_norm": 1.3191325926837572,
      "learning_rate": 1.386962820037127e-05,
      "loss": 0.1991,
      "step": 6132
    },
    {
      "epoch": 3.9238643634037107,
      "grad_norm": 1.5058301195691735,
      "learning_rate": 1.3867717393821105e-05,
      "loss": 0.2308,
      "step": 6133
    },
    {
      "epoch": 3.924504158669226,
      "grad_norm": 1.5314201262740668,
      "learning_rate": 1.3865806421198036e-05,
      "loss": 0.2391,
      "step": 6134
    },
    {
      "epoch": 3.925143953934741,
      "grad_norm": 1.2987679909213006,
      "learning_rate": 1.3863895282584117e-05,
      "loss": 0.1978,
      "step": 6135
    },
    {
      "epoch": 3.925783749200256,
      "grad_norm": 1.491399833461101,
      "learning_rate": 1.3861983978061408e-05,
      "loss": 0.2251,
      "step": 6136
    },
    {
      "epoch": 3.926423544465771,
      "grad_norm": 1.5793186252911167,
      "learning_rate": 1.3860072507711979e-05,
      "loss": 0.2507,
      "step": 6137
    },
    {
      "epoch": 3.927063339731286,
      "grad_norm": 1.5984455941055347,
      "learning_rate": 1.3858160871617898e-05,
      "loss": 0.2564,
      "step": 6138
    },
    {
      "epoch": 3.927703134996801,
      "grad_norm": 1.482348881675251,
      "learning_rate": 1.3856249069861253e-05,
      "loss": 0.2261,
      "step": 6139
    },
    {
      "epoch": 3.928342930262316,
      "grad_norm": 1.5408926013344721,
      "learning_rate": 1.3854337102524138e-05,
      "loss": 0.2275,
      "step": 6140
    },
    {
      "epoch": 3.928982725527831,
      "grad_norm": 1.5450703483378347,
      "learning_rate": 1.3852424969688638e-05,
      "loss": 0.2229,
      "step": 6141
    },
    {
      "epoch": 3.9296225207933464,
      "grad_norm": 1.4697963902289377,
      "learning_rate": 1.3850512671436866e-05,
      "loss": 0.2371,
      "step": 6142
    },
    {
      "epoch": 3.9302623160588612,
      "grad_norm": 1.5408401302489694,
      "learning_rate": 1.3848600207850928e-05,
      "loss": 0.2072,
      "step": 6143
    },
    {
      "epoch": 3.930902111324376,
      "grad_norm": 1.4104092890856128,
      "learning_rate": 1.3846687579012946e-05,
      "loss": 0.2123,
      "step": 6144
    },
    {
      "epoch": 3.9315419065898913,
      "grad_norm": 1.576153857902473,
      "learning_rate": 1.384477478500504e-05,
      "loss": 0.2411,
      "step": 6145
    },
    {
      "epoch": 3.932181701855406,
      "grad_norm": 1.4419852172548353,
      "learning_rate": 1.3842861825909344e-05,
      "loss": 0.2183,
      "step": 6146
    },
    {
      "epoch": 3.9328214971209214,
      "grad_norm": 1.5612571364410741,
      "learning_rate": 1.3840948701807998e-05,
      "loss": 0.2135,
      "step": 6147
    },
    {
      "epoch": 3.9334612923864363,
      "grad_norm": 1.4428473246541607,
      "learning_rate": 1.3839035412783143e-05,
      "loss": 0.2366,
      "step": 6148
    },
    {
      "epoch": 3.9341010876519515,
      "grad_norm": 1.4457082441110753,
      "learning_rate": 1.3837121958916939e-05,
      "loss": 0.2242,
      "step": 6149
    },
    {
      "epoch": 3.9347408829174664,
      "grad_norm": 1.4071385267730598,
      "learning_rate": 1.3835208340291546e-05,
      "loss": 0.2046,
      "step": 6150
    },
    {
      "epoch": 3.935380678182981,
      "grad_norm": 1.540419795788152,
      "learning_rate": 1.3833294556989129e-05,
      "loss": 0.2388,
      "step": 6151
    },
    {
      "epoch": 3.9360204734484965,
      "grad_norm": 1.4680041935222643,
      "learning_rate": 1.3831380609091863e-05,
      "loss": 0.2367,
      "step": 6152
    },
    {
      "epoch": 3.9366602687140118,
      "grad_norm": 1.3305857723052632,
      "learning_rate": 1.3829466496681927e-05,
      "loss": 0.1961,
      "step": 6153
    },
    {
      "epoch": 3.9373000639795266,
      "grad_norm": 1.4841969073332832,
      "learning_rate": 1.3827552219841511e-05,
      "loss": 0.2377,
      "step": 6154
    },
    {
      "epoch": 3.9379398592450414,
      "grad_norm": 1.5433827399999063,
      "learning_rate": 1.3825637778652814e-05,
      "loss": 0.2568,
      "step": 6155
    },
    {
      "epoch": 3.9385796545105567,
      "grad_norm": 1.370956149529037,
      "learning_rate": 1.3823723173198033e-05,
      "loss": 0.2148,
      "step": 6156
    },
    {
      "epoch": 3.9392194497760715,
      "grad_norm": 1.3787820189184636,
      "learning_rate": 1.3821808403559382e-05,
      "loss": 0.225,
      "step": 6157
    },
    {
      "epoch": 3.939859245041587,
      "grad_norm": 1.4838827861932182,
      "learning_rate": 1.3819893469819081e-05,
      "loss": 0.2548,
      "step": 6158
    },
    {
      "epoch": 3.9404990403071016,
      "grad_norm": 1.3044750630934676,
      "learning_rate": 1.3817978372059344e-05,
      "loss": 0.211,
      "step": 6159
    },
    {
      "epoch": 3.941138835572617,
      "grad_norm": 1.474390528045609,
      "learning_rate": 1.381606311036241e-05,
      "loss": 0.2404,
      "step": 6160
    },
    {
      "epoch": 3.9417786308381317,
      "grad_norm": 1.475639628705748,
      "learning_rate": 1.3814147684810512e-05,
      "loss": 0.2303,
      "step": 6161
    },
    {
      "epoch": 3.9424184261036466,
      "grad_norm": 1.5699223882027875,
      "learning_rate": 1.38122320954859e-05,
      "loss": 0.248,
      "step": 6162
    },
    {
      "epoch": 3.943058221369162,
      "grad_norm": 1.3639754068362053,
      "learning_rate": 1.3810316342470819e-05,
      "loss": 0.1937,
      "step": 6163
    },
    {
      "epoch": 3.943698016634677,
      "grad_norm": 1.388511911221635,
      "learning_rate": 1.3808400425847535e-05,
      "loss": 0.2062,
      "step": 6164
    },
    {
      "epoch": 3.944337811900192,
      "grad_norm": 1.4382006552449387,
      "learning_rate": 1.3806484345698312e-05,
      "loss": 0.2198,
      "step": 6165
    },
    {
      "epoch": 3.944977607165707,
      "grad_norm": 1.3768101400147144,
      "learning_rate": 1.3804568102105425e-05,
      "loss": 0.2194,
      "step": 6166
    },
    {
      "epoch": 3.945617402431222,
      "grad_norm": 1.4128775073811568,
      "learning_rate": 1.3802651695151148e-05,
      "loss": 0.2277,
      "step": 6167
    },
    {
      "epoch": 3.946257197696737,
      "grad_norm": 1.5792231956985379,
      "learning_rate": 1.380073512491777e-05,
      "loss": 0.2424,
      "step": 6168
    },
    {
      "epoch": 3.946896992962252,
      "grad_norm": 1.5059198056679146,
      "learning_rate": 1.3798818391487591e-05,
      "loss": 0.2344,
      "step": 6169
    },
    {
      "epoch": 3.947536788227767,
      "grad_norm": 1.4767745370726646,
      "learning_rate": 1.3796901494942904e-05,
      "loss": 0.2345,
      "step": 6170
    },
    {
      "epoch": 3.9481765834932823,
      "grad_norm": 1.4478515155550629,
      "learning_rate": 1.3794984435366021e-05,
      "loss": 0.2164,
      "step": 6171
    },
    {
      "epoch": 3.948816378758797,
      "grad_norm": 1.4970096554445953,
      "learning_rate": 1.3793067212839259e-05,
      "loss": 0.2306,
      "step": 6172
    },
    {
      "epoch": 3.9494561740243124,
      "grad_norm": 1.463718394484502,
      "learning_rate": 1.379114982744494e-05,
      "loss": 0.2136,
      "step": 6173
    },
    {
      "epoch": 3.950095969289827,
      "grad_norm": 1.3145641519492866,
      "learning_rate": 1.378923227926539e-05,
      "loss": 0.2186,
      "step": 6174
    },
    {
      "epoch": 3.9507357645553425,
      "grad_norm": 1.4852755357946632,
      "learning_rate": 1.3787314568382946e-05,
      "loss": 0.218,
      "step": 6175
    },
    {
      "epoch": 3.9513755598208573,
      "grad_norm": 1.4430677102594873,
      "learning_rate": 1.3785396694879949e-05,
      "loss": 0.2088,
      "step": 6176
    },
    {
      "epoch": 3.952015355086372,
      "grad_norm": 1.270617436831412,
      "learning_rate": 1.3783478658838753e-05,
      "loss": 0.1857,
      "step": 6177
    },
    {
      "epoch": 3.9526551503518874,
      "grad_norm": 1.5884637496912428,
      "learning_rate": 1.3781560460341715e-05,
      "loss": 0.2327,
      "step": 6178
    },
    {
      "epoch": 3.9532949456174027,
      "grad_norm": 1.4878945914049597,
      "learning_rate": 1.3779642099471198e-05,
      "loss": 0.2304,
      "step": 6179
    },
    {
      "epoch": 3.9539347408829175,
      "grad_norm": 1.669201132808165,
      "learning_rate": 1.3777723576309572e-05,
      "loss": 0.2557,
      "step": 6180
    },
    {
      "epoch": 3.9545745361484324,
      "grad_norm": 1.3355211603965333,
      "learning_rate": 1.3775804890939216e-05,
      "loss": 0.1984,
      "step": 6181
    },
    {
      "epoch": 3.9552143314139476,
      "grad_norm": 1.3455504308835853,
      "learning_rate": 1.3773886043442514e-05,
      "loss": 0.2119,
      "step": 6182
    },
    {
      "epoch": 3.9558541266794625,
      "grad_norm": 1.4358727265330173,
      "learning_rate": 1.3771967033901858e-05,
      "loss": 0.2289,
      "step": 6183
    },
    {
      "epoch": 3.9564939219449777,
      "grad_norm": 1.3834080628790228,
      "learning_rate": 1.3770047862399648e-05,
      "loss": 0.2214,
      "step": 6184
    },
    {
      "epoch": 3.9571337172104926,
      "grad_norm": 1.3442769273727848,
      "learning_rate": 1.3768128529018289e-05,
      "loss": 0.2219,
      "step": 6185
    },
    {
      "epoch": 3.957773512476008,
      "grad_norm": 1.332555977874873,
      "learning_rate": 1.3766209033840193e-05,
      "loss": 0.1977,
      "step": 6186
    },
    {
      "epoch": 3.9584133077415227,
      "grad_norm": 1.413870191741213,
      "learning_rate": 1.3764289376947784e-05,
      "loss": 0.2489,
      "step": 6187
    },
    {
      "epoch": 3.9590531030070375,
      "grad_norm": 1.3241190507297842,
      "learning_rate": 1.3762369558423483e-05,
      "loss": 0.1933,
      "step": 6188
    },
    {
      "epoch": 3.9596928982725528,
      "grad_norm": 1.3954898209327529,
      "learning_rate": 1.3760449578349728e-05,
      "loss": 0.1932,
      "step": 6189
    },
    {
      "epoch": 3.960332693538068,
      "grad_norm": 1.5456796231652674,
      "learning_rate": 1.3758529436808955e-05,
      "loss": 0.2385,
      "step": 6190
    },
    {
      "epoch": 3.960972488803583,
      "grad_norm": 1.345680236924927,
      "learning_rate": 1.3756609133883611e-05,
      "loss": 0.2065,
      "step": 6191
    },
    {
      "epoch": 3.9616122840690977,
      "grad_norm": 1.4413709536065464,
      "learning_rate": 1.375468866965616e-05,
      "loss": 0.2161,
      "step": 6192
    },
    {
      "epoch": 3.962252079334613,
      "grad_norm": 1.4903663370167441,
      "learning_rate": 1.375276804420905e-05,
      "loss": 0.2328,
      "step": 6193
    },
    {
      "epoch": 3.962891874600128,
      "grad_norm": 1.6708908381347065,
      "learning_rate": 1.3750847257624762e-05,
      "loss": 0.2593,
      "step": 6194
    },
    {
      "epoch": 3.963531669865643,
      "grad_norm": 1.4694715581269993,
      "learning_rate": 1.374892630998576e-05,
      "loss": 0.2221,
      "step": 6195
    },
    {
      "epoch": 3.964171465131158,
      "grad_norm": 1.5322505840570302,
      "learning_rate": 1.3747005201374535e-05,
      "loss": 0.2574,
      "step": 6196
    },
    {
      "epoch": 3.964811260396673,
      "grad_norm": 1.4372580139216289,
      "learning_rate": 1.3745083931873572e-05,
      "loss": 0.2118,
      "step": 6197
    },
    {
      "epoch": 3.965451055662188,
      "grad_norm": 1.4791833494219777,
      "learning_rate": 1.3743162501565364e-05,
      "loss": 0.2369,
      "step": 6198
    },
    {
      "epoch": 3.966090850927703,
      "grad_norm": 1.2760155396244188,
      "learning_rate": 1.3741240910532416e-05,
      "loss": 0.2032,
      "step": 6199
    },
    {
      "epoch": 3.966730646193218,
      "grad_norm": 1.4818686967141148,
      "learning_rate": 1.373931915885724e-05,
      "loss": 0.2313,
      "step": 6200
    },
    {
      "epoch": 3.9673704414587334,
      "grad_norm": 1.3199104247633686,
      "learning_rate": 1.373739724662235e-05,
      "loss": 0.2122,
      "step": 6201
    },
    {
      "epoch": 3.9680102367242482,
      "grad_norm": 1.3169294257368058,
      "learning_rate": 1.3735475173910273e-05,
      "loss": 0.1992,
      "step": 6202
    },
    {
      "epoch": 3.968650031989763,
      "grad_norm": 1.3727919462507168,
      "learning_rate": 1.3733552940803539e-05,
      "loss": 0.2101,
      "step": 6203
    },
    {
      "epoch": 3.9692898272552783,
      "grad_norm": 1.5063156780097375,
      "learning_rate": 1.373163054738468e-05,
      "loss": 0.2403,
      "step": 6204
    },
    {
      "epoch": 3.9699296225207936,
      "grad_norm": 1.418327593166001,
      "learning_rate": 1.3729707993736242e-05,
      "loss": 0.2062,
      "step": 6205
    },
    {
      "epoch": 3.9705694177863085,
      "grad_norm": 1.5011720472612897,
      "learning_rate": 1.3727785279940777e-05,
      "loss": 0.2281,
      "step": 6206
    },
    {
      "epoch": 3.9712092130518233,
      "grad_norm": 1.3959562775944787,
      "learning_rate": 1.3725862406080847e-05,
      "loss": 0.2321,
      "step": 6207
    },
    {
      "epoch": 3.9718490083173386,
      "grad_norm": 1.6240625732833005,
      "learning_rate": 1.372393937223901e-05,
      "loss": 0.2617,
      "step": 6208
    },
    {
      "epoch": 3.9724888035828534,
      "grad_norm": 1.42225584217831,
      "learning_rate": 1.3722016178497844e-05,
      "loss": 0.1984,
      "step": 6209
    },
    {
      "epoch": 3.9731285988483687,
      "grad_norm": 1.490981461518347,
      "learning_rate": 1.3720092824939921e-05,
      "loss": 0.2144,
      "step": 6210
    },
    {
      "epoch": 3.9737683941138835,
      "grad_norm": 1.5735318600957287,
      "learning_rate": 1.3718169311647832e-05,
      "loss": 0.2401,
      "step": 6211
    },
    {
      "epoch": 3.9744081893793988,
      "grad_norm": 1.4168082775139255,
      "learning_rate": 1.3716245638704166e-05,
      "loss": 0.2308,
      "step": 6212
    },
    {
      "epoch": 3.9750479846449136,
      "grad_norm": 1.5017493320424462,
      "learning_rate": 1.3714321806191522e-05,
      "loss": 0.2364,
      "step": 6213
    },
    {
      "epoch": 3.9756877799104284,
      "grad_norm": 1.4457148006290523,
      "learning_rate": 1.3712397814192509e-05,
      "loss": 0.2203,
      "step": 6214
    },
    {
      "epoch": 3.9763275751759437,
      "grad_norm": 1.4807650787983764,
      "learning_rate": 1.3710473662789734e-05,
      "loss": 0.2606,
      "step": 6215
    },
    {
      "epoch": 3.976967370441459,
      "grad_norm": 1.3672019991287145,
      "learning_rate": 1.370854935206582e-05,
      "loss": 0.2048,
      "step": 6216
    },
    {
      "epoch": 3.977607165706974,
      "grad_norm": 1.4529589103675908,
      "learning_rate": 1.3706624882103399e-05,
      "loss": 0.2136,
      "step": 6217
    },
    {
      "epoch": 3.9782469609724886,
      "grad_norm": 1.3519469073913304,
      "learning_rate": 1.3704700252985097e-05,
      "loss": 0.2174,
      "step": 6218
    },
    {
      "epoch": 3.978886756238004,
      "grad_norm": 1.424371123992096,
      "learning_rate": 1.3702775464793558e-05,
      "loss": 0.2338,
      "step": 6219
    },
    {
      "epoch": 3.9795265515035187,
      "grad_norm": 1.5452804133746532,
      "learning_rate": 1.3700850517611423e-05,
      "loss": 0.2357,
      "step": 6220
    },
    {
      "epoch": 3.980166346769034,
      "grad_norm": 1.5622064504920612,
      "learning_rate": 1.369892541152135e-05,
      "loss": 0.2429,
      "step": 6221
    },
    {
      "epoch": 3.980806142034549,
      "grad_norm": 1.5966751513444664,
      "learning_rate": 1.3697000146606004e-05,
      "loss": 0.2443,
      "step": 6222
    },
    {
      "epoch": 3.981445937300064,
      "grad_norm": 1.4730509593362917,
      "learning_rate": 1.3695074722948046e-05,
      "loss": 0.2094,
      "step": 6223
    },
    {
      "epoch": 3.982085732565579,
      "grad_norm": 1.521004550973606,
      "learning_rate": 1.3693149140630153e-05,
      "loss": 0.2399,
      "step": 6224
    },
    {
      "epoch": 3.982725527831094,
      "grad_norm": 1.4159440390381752,
      "learning_rate": 1.3691223399735004e-05,
      "loss": 0.2268,
      "step": 6225
    },
    {
      "epoch": 3.983365323096609,
      "grad_norm": 1.4298759972529105,
      "learning_rate": 1.3689297500345289e-05,
      "loss": 0.2241,
      "step": 6226
    },
    {
      "epoch": 3.9840051183621243,
      "grad_norm": 1.4897000717392865,
      "learning_rate": 1.3687371442543704e-05,
      "loss": 0.2038,
      "step": 6227
    },
    {
      "epoch": 3.984644913627639,
      "grad_norm": 1.2644494472588268,
      "learning_rate": 1.3685445226412946e-05,
      "loss": 0.2098,
      "step": 6228
    },
    {
      "epoch": 3.985284708893154,
      "grad_norm": 1.346432944692962,
      "learning_rate": 1.3683518852035728e-05,
      "loss": 0.1949,
      "step": 6229
    },
    {
      "epoch": 3.9859245041586693,
      "grad_norm": 1.3452522026671574,
      "learning_rate": 1.368159231949476e-05,
      "loss": 0.1991,
      "step": 6230
    },
    {
      "epoch": 3.986564299424184,
      "grad_norm": 1.606831110927474,
      "learning_rate": 1.3679665628872769e-05,
      "loss": 0.235,
      "step": 6231
    },
    {
      "epoch": 3.9872040946896994,
      "grad_norm": 1.4827132358888564,
      "learning_rate": 1.3677738780252484e-05,
      "loss": 0.2487,
      "step": 6232
    },
    {
      "epoch": 3.987843889955214,
      "grad_norm": 1.522137426136074,
      "learning_rate": 1.3675811773716634e-05,
      "loss": 0.2309,
      "step": 6233
    },
    {
      "epoch": 3.9884836852207295,
      "grad_norm": 1.4089281528569737,
      "learning_rate": 1.3673884609347968e-05,
      "loss": 0.2265,
      "step": 6234
    },
    {
      "epoch": 3.9891234804862443,
      "grad_norm": 1.3501418229798337,
      "learning_rate": 1.367195728722923e-05,
      "loss": 0.2257,
      "step": 6235
    },
    {
      "epoch": 3.9897632757517596,
      "grad_norm": 1.4465108741905484,
      "learning_rate": 1.3670029807443177e-05,
      "loss": 0.241,
      "step": 6236
    },
    {
      "epoch": 3.9904030710172744,
      "grad_norm": 1.4461859144905858,
      "learning_rate": 1.3668102170072575e-05,
      "loss": 0.2537,
      "step": 6237
    },
    {
      "epoch": 3.9910428662827897,
      "grad_norm": 1.3858039925906431,
      "learning_rate": 1.3666174375200188e-05,
      "loss": 0.2092,
      "step": 6238
    },
    {
      "epoch": 3.9916826615483045,
      "grad_norm": 1.4329021911861322,
      "learning_rate": 1.3664246422908798e-05,
      "loss": 0.2241,
      "step": 6239
    },
    {
      "epoch": 3.9923224568138194,
      "grad_norm": 1.498389195078725,
      "learning_rate": 1.3662318313281184e-05,
      "loss": 0.2145,
      "step": 6240
    },
    {
      "epoch": 3.9929622520793346,
      "grad_norm": 1.3247598048175986,
      "learning_rate": 1.3660390046400136e-05,
      "loss": 0.2114,
      "step": 6241
    },
    {
      "epoch": 3.99360204734485,
      "grad_norm": 1.5338409428261186,
      "learning_rate": 1.365846162234845e-05,
      "loss": 0.2386,
      "step": 6242
    },
    {
      "epoch": 3.9942418426103647,
      "grad_norm": 1.5267597274218248,
      "learning_rate": 1.3656533041208933e-05,
      "loss": 0.2122,
      "step": 6243
    },
    {
      "epoch": 3.9948816378758796,
      "grad_norm": 1.478717864181993,
      "learning_rate": 1.365460430306439e-05,
      "loss": 0.21,
      "step": 6244
    },
    {
      "epoch": 3.995521433141395,
      "grad_norm": 1.385041476161103,
      "learning_rate": 1.3652675407997635e-05,
      "loss": 0.2123,
      "step": 6245
    },
    {
      "epoch": 3.9961612284069097,
      "grad_norm": 1.5407449940706879,
      "learning_rate": 1.3650746356091503e-05,
      "loss": 0.2273,
      "step": 6246
    },
    {
      "epoch": 3.996801023672425,
      "grad_norm": 1.6004415280633857,
      "learning_rate": 1.364881714742881e-05,
      "loss": 0.2425,
      "step": 6247
    },
    {
      "epoch": 3.99744081893794,
      "grad_norm": 1.5638922458372377,
      "learning_rate": 1.3646887782092404e-05,
      "loss": 0.2416,
      "step": 6248
    },
    {
      "epoch": 3.998080614203455,
      "grad_norm": 1.3872542464707578,
      "learning_rate": 1.3644958260165124e-05,
      "loss": 0.2227,
      "step": 6249
    },
    {
      "epoch": 3.99872040946897,
      "grad_norm": 1.4083016032916182,
      "learning_rate": 1.3643028581729815e-05,
      "loss": 0.2477,
      "step": 6250
    },
    {
      "epoch": 3.9993602047344847,
      "grad_norm": 1.2853884333536667,
      "learning_rate": 1.3641098746869346e-05,
      "loss": 0.2083,
      "step": 6251
    },
    {
      "epoch": 4.0,
      "grad_norm": 1.3086468508141198,
      "learning_rate": 1.3639168755666567e-05,
      "loss": 0.1751,
      "step": 6252
    },
    {
      "epoch": 4.000639795265515,
      "grad_norm": 1.182708756815443,
      "learning_rate": 1.363723860820436e-05,
      "loss": 0.1543,
      "step": 6253
    },
    {
      "epoch": 4.00127959053103,
      "grad_norm": 1.1084829596252612,
      "learning_rate": 1.3635308304565595e-05,
      "loss": 0.1449,
      "step": 6254
    },
    {
      "epoch": 4.001919385796545,
      "grad_norm": 1.1667336426545716,
      "learning_rate": 1.3633377844833158e-05,
      "loss": 0.1514,
      "step": 6255
    },
    {
      "epoch": 4.00255918106206,
      "grad_norm": 1.2506754423534099,
      "learning_rate": 1.3631447229089942e-05,
      "loss": 0.1477,
      "step": 6256
    },
    {
      "epoch": 4.0031989763275755,
      "grad_norm": 1.118046052921286,
      "learning_rate": 1.3629516457418843e-05,
      "loss": 0.1408,
      "step": 6257
    },
    {
      "epoch": 4.00383877159309,
      "grad_norm": 1.2766107639812816,
      "learning_rate": 1.3627585529902759e-05,
      "loss": 0.1368,
      "step": 6258
    },
    {
      "epoch": 4.004478566858605,
      "grad_norm": 1.3270423717738833,
      "learning_rate": 1.3625654446624608e-05,
      "loss": 0.1406,
      "step": 6259
    },
    {
      "epoch": 4.00511836212412,
      "grad_norm": 1.2617192878171897,
      "learning_rate": 1.3623723207667302e-05,
      "loss": 0.1406,
      "step": 6260
    },
    {
      "epoch": 4.005758157389636,
      "grad_norm": 1.287812379766795,
      "learning_rate": 1.3621791813113771e-05,
      "loss": 0.1328,
      "step": 6261
    },
    {
      "epoch": 4.00639795265515,
      "grad_norm": 1.3144319107272446,
      "learning_rate": 1.361986026304694e-05,
      "loss": 0.1436,
      "step": 6262
    },
    {
      "epoch": 4.007037747920665,
      "grad_norm": 1.468905390370415,
      "learning_rate": 1.3617928557549751e-05,
      "loss": 0.1408,
      "step": 6263
    },
    {
      "epoch": 4.007677543186181,
      "grad_norm": 1.241866282405594,
      "learning_rate": 1.3615996696705145e-05,
      "loss": 0.1383,
      "step": 6264
    },
    {
      "epoch": 4.008317338451695,
      "grad_norm": 1.296373232007353,
      "learning_rate": 1.3614064680596073e-05,
      "loss": 0.134,
      "step": 6265
    },
    {
      "epoch": 4.00895713371721,
      "grad_norm": 1.3350121540214543,
      "learning_rate": 1.361213250930549e-05,
      "loss": 0.1439,
      "step": 6266
    },
    {
      "epoch": 4.009596928982726,
      "grad_norm": 1.2622723795472677,
      "learning_rate": 1.3610200182916364e-05,
      "loss": 0.1305,
      "step": 6267
    },
    {
      "epoch": 4.010236724248241,
      "grad_norm": 1.1896512494627765,
      "learning_rate": 1.3608267701511663e-05,
      "loss": 0.1327,
      "step": 6268
    },
    {
      "epoch": 4.010876519513755,
      "grad_norm": 1.1794910458420094,
      "learning_rate": 1.360633506517437e-05,
      "loss": 0.1461,
      "step": 6269
    },
    {
      "epoch": 4.0115163147792705,
      "grad_norm": 1.1930776368673481,
      "learning_rate": 1.3604402273987459e-05,
      "loss": 0.139,
      "step": 6270
    },
    {
      "epoch": 4.012156110044786,
      "grad_norm": 1.2782691444310328,
      "learning_rate": 1.360246932803393e-05,
      "loss": 0.1402,
      "step": 6271
    },
    {
      "epoch": 4.012795905310301,
      "grad_norm": 1.1893691514166673,
      "learning_rate": 1.3600536227396774e-05,
      "loss": 0.1349,
      "step": 6272
    },
    {
      "epoch": 4.0134357005758154,
      "grad_norm": 1.2811017159228226,
      "learning_rate": 1.3598602972158998e-05,
      "loss": 0.1486,
      "step": 6273
    },
    {
      "epoch": 4.014075495841331,
      "grad_norm": 1.381383898404408,
      "learning_rate": 1.3596669562403616e-05,
      "loss": 0.145,
      "step": 6274
    },
    {
      "epoch": 4.014715291106846,
      "grad_norm": 1.1841872208956017,
      "learning_rate": 1.3594735998213634e-05,
      "loss": 0.1326,
      "step": 6275
    },
    {
      "epoch": 4.015355086372361,
      "grad_norm": 1.2477293442548898,
      "learning_rate": 1.3592802279672087e-05,
      "loss": 0.1338,
      "step": 6276
    },
    {
      "epoch": 4.015994881637876,
      "grad_norm": 1.241678141747811,
      "learning_rate": 1.3590868406862e-05,
      "loss": 0.1457,
      "step": 6277
    },
    {
      "epoch": 4.016634676903391,
      "grad_norm": 1.3153744582497464,
      "learning_rate": 1.358893437986641e-05,
      "loss": 0.1341,
      "step": 6278
    },
    {
      "epoch": 4.017274472168906,
      "grad_norm": 1.3220141890199344,
      "learning_rate": 1.3587000198768365e-05,
      "loss": 0.1426,
      "step": 6279
    },
    {
      "epoch": 4.017914267434421,
      "grad_norm": 1.3331193524684062,
      "learning_rate": 1.3585065863650908e-05,
      "loss": 0.1503,
      "step": 6280
    },
    {
      "epoch": 4.018554062699936,
      "grad_norm": 1.2594073966387038,
      "learning_rate": 1.3583131374597102e-05,
      "loss": 0.1411,
      "step": 6281
    },
    {
      "epoch": 4.019193857965451,
      "grad_norm": 1.2953729422047722,
      "learning_rate": 1.358119673169001e-05,
      "loss": 0.1447,
      "step": 6282
    },
    {
      "epoch": 4.019833653230966,
      "grad_norm": 1.2189918612538797,
      "learning_rate": 1.3579261935012698e-05,
      "loss": 0.145,
      "step": 6283
    },
    {
      "epoch": 4.020473448496481,
      "grad_norm": 1.3007582320950222,
      "learning_rate": 1.357732698464825e-05,
      "loss": 0.1387,
      "step": 6284
    },
    {
      "epoch": 4.021113243761996,
      "grad_norm": 1.2805838744646127,
      "learning_rate": 1.357539188067974e-05,
      "loss": 0.1375,
      "step": 6285
    },
    {
      "epoch": 4.021753039027511,
      "grad_norm": 1.328783509224516,
      "learning_rate": 1.3573456623190267e-05,
      "loss": 0.1324,
      "step": 6286
    },
    {
      "epoch": 4.022392834293027,
      "grad_norm": 1.2308012050031476,
      "learning_rate": 1.357152121226292e-05,
      "loss": 0.1441,
      "step": 6287
    },
    {
      "epoch": 4.023032629558541,
      "grad_norm": 1.3972817781159808,
      "learning_rate": 1.3569585647980808e-05,
      "loss": 0.1366,
      "step": 6288
    },
    {
      "epoch": 4.023672424824056,
      "grad_norm": 1.2091207464040103,
      "learning_rate": 1.3567649930427038e-05,
      "loss": 0.1376,
      "step": 6289
    },
    {
      "epoch": 4.024312220089572,
      "grad_norm": 1.2066660281420254,
      "learning_rate": 1.3565714059684724e-05,
      "loss": 0.1362,
      "step": 6290
    },
    {
      "epoch": 4.024952015355086,
      "grad_norm": 1.2090552440865774,
      "learning_rate": 1.3563778035836994e-05,
      "loss": 0.1422,
      "step": 6291
    },
    {
      "epoch": 4.025591810620601,
      "grad_norm": 1.2403750662034212,
      "learning_rate": 1.356184185896697e-05,
      "loss": 0.1332,
      "step": 6292
    },
    {
      "epoch": 4.0262316058861165,
      "grad_norm": 1.3148457527950548,
      "learning_rate": 1.3559905529157797e-05,
      "loss": 0.1365,
      "step": 6293
    },
    {
      "epoch": 4.026871401151632,
      "grad_norm": 1.4395940091765682,
      "learning_rate": 1.3557969046492613e-05,
      "loss": 0.1403,
      "step": 6294
    },
    {
      "epoch": 4.027511196417146,
      "grad_norm": 1.2503903664654439,
      "learning_rate": 1.3556032411054565e-05,
      "loss": 0.1277,
      "step": 6295
    },
    {
      "epoch": 4.028150991682661,
      "grad_norm": 1.1678269827532701,
      "learning_rate": 1.3554095622926812e-05,
      "loss": 0.1279,
      "step": 6296
    },
    {
      "epoch": 4.028790786948177,
      "grad_norm": 1.2539300553650237,
      "learning_rate": 1.3552158682192515e-05,
      "loss": 0.1396,
      "step": 6297
    },
    {
      "epoch": 4.029430582213692,
      "grad_norm": 1.218338535295174,
      "learning_rate": 1.3550221588934843e-05,
      "loss": 0.1413,
      "step": 6298
    },
    {
      "epoch": 4.030070377479206,
      "grad_norm": 1.1778091745465833,
      "learning_rate": 1.3548284343236973e-05,
      "loss": 0.1301,
      "step": 6299
    },
    {
      "epoch": 4.030710172744722,
      "grad_norm": 1.1797885384339761,
      "learning_rate": 1.3546346945182082e-05,
      "loss": 0.1357,
      "step": 6300
    },
    {
      "epoch": 4.031349968010237,
      "grad_norm": 1.2324099764755285,
      "learning_rate": 1.3544409394853363e-05,
      "loss": 0.1458,
      "step": 6301
    },
    {
      "epoch": 4.031989763275751,
      "grad_norm": 1.3287738416097605,
      "learning_rate": 1.3542471692334011e-05,
      "loss": 0.1499,
      "step": 6302
    },
    {
      "epoch": 4.032629558541267,
      "grad_norm": 1.2106152622367738,
      "learning_rate": 1.3540533837707223e-05,
      "loss": 0.1385,
      "step": 6303
    },
    {
      "epoch": 4.033269353806782,
      "grad_norm": 1.2111965927154336,
      "learning_rate": 1.3538595831056216e-05,
      "loss": 0.1381,
      "step": 6304
    },
    {
      "epoch": 4.033909149072297,
      "grad_norm": 1.2674297746247796,
      "learning_rate": 1.3536657672464193e-05,
      "loss": 0.1279,
      "step": 6305
    },
    {
      "epoch": 4.0345489443378115,
      "grad_norm": 1.1863319458912893,
      "learning_rate": 1.3534719362014383e-05,
      "loss": 0.1355,
      "step": 6306
    },
    {
      "epoch": 4.035188739603327,
      "grad_norm": 1.2234245239368902,
      "learning_rate": 1.353278089979001e-05,
      "loss": 0.1332,
      "step": 6307
    },
    {
      "epoch": 4.035828534868842,
      "grad_norm": 1.369436407253894,
      "learning_rate": 1.3530842285874313e-05,
      "loss": 0.1493,
      "step": 6308
    },
    {
      "epoch": 4.036468330134357,
      "grad_norm": 1.1727924309236508,
      "learning_rate": 1.3528903520350527e-05,
      "loss": 0.1363,
      "step": 6309
    },
    {
      "epoch": 4.037108125399872,
      "grad_norm": 1.2827208086680317,
      "learning_rate": 1.35269646033019e-05,
      "loss": 0.1509,
      "step": 6310
    },
    {
      "epoch": 4.037747920665387,
      "grad_norm": 1.257901333458684,
      "learning_rate": 1.3525025534811689e-05,
      "loss": 0.1479,
      "step": 6311
    },
    {
      "epoch": 4.038387715930902,
      "grad_norm": 1.2108266954072955,
      "learning_rate": 1.3523086314963148e-05,
      "loss": 0.1387,
      "step": 6312
    },
    {
      "epoch": 4.039027511196418,
      "grad_norm": 1.2810501841176873,
      "learning_rate": 1.352114694383955e-05,
      "loss": 0.1485,
      "step": 6313
    },
    {
      "epoch": 4.039667306461932,
      "grad_norm": 1.2667941860136214,
      "learning_rate": 1.3519207421524168e-05,
      "loss": 0.1367,
      "step": 6314
    },
    {
      "epoch": 4.040307101727447,
      "grad_norm": 1.2056779689391468,
      "learning_rate": 1.3517267748100276e-05,
      "loss": 0.1436,
      "step": 6315
    },
    {
      "epoch": 4.0409468969929625,
      "grad_norm": 1.3313018589767724,
      "learning_rate": 1.3515327923651167e-05,
      "loss": 0.1491,
      "step": 6316
    },
    {
      "epoch": 4.041586692258477,
      "grad_norm": 1.2197578343041688,
      "learning_rate": 1.351338794826013e-05,
      "loss": 0.1386,
      "step": 6317
    },
    {
      "epoch": 4.042226487523992,
      "grad_norm": 1.2891909961028543,
      "learning_rate": 1.3511447822010462e-05,
      "loss": 0.1467,
      "step": 6318
    },
    {
      "epoch": 4.042866282789507,
      "grad_norm": 1.2828248207972477,
      "learning_rate": 1.3509507544985475e-05,
      "loss": 0.1352,
      "step": 6319
    },
    {
      "epoch": 4.043506078055023,
      "grad_norm": 1.283089251840726,
      "learning_rate": 1.3507567117268471e-05,
      "loss": 0.1465,
      "step": 6320
    },
    {
      "epoch": 4.044145873320537,
      "grad_norm": 1.1789353928768758,
      "learning_rate": 1.350562653894278e-05,
      "loss": 0.1315,
      "step": 6321
    },
    {
      "epoch": 4.044785668586052,
      "grad_norm": 1.3222982759612354,
      "learning_rate": 1.3503685810091716e-05,
      "loss": 0.1433,
      "step": 6322
    },
    {
      "epoch": 4.045425463851568,
      "grad_norm": 1.1758136488344035,
      "learning_rate": 1.3501744930798619e-05,
      "loss": 0.1338,
      "step": 6323
    },
    {
      "epoch": 4.046065259117083,
      "grad_norm": 1.2098478973757103,
      "learning_rate": 1.3499803901146827e-05,
      "loss": 0.1303,
      "step": 6324
    },
    {
      "epoch": 4.046705054382597,
      "grad_norm": 1.1566392907031517,
      "learning_rate": 1.3497862721219677e-05,
      "loss": 0.1276,
      "step": 6325
    },
    {
      "epoch": 4.047344849648113,
      "grad_norm": 1.1716652357687583,
      "learning_rate": 1.3495921391100524e-05,
      "loss": 0.1321,
      "step": 6326
    },
    {
      "epoch": 4.047984644913628,
      "grad_norm": 1.2540659033909336,
      "learning_rate": 1.3493979910872725e-05,
      "loss": 0.1421,
      "step": 6327
    },
    {
      "epoch": 4.048624440179142,
      "grad_norm": 1.2343256959408033,
      "learning_rate": 1.3492038280619647e-05,
      "loss": 0.1392,
      "step": 6328
    },
    {
      "epoch": 4.0492642354446575,
      "grad_norm": 1.1686375568395786,
      "learning_rate": 1.3490096500424653e-05,
      "loss": 0.1436,
      "step": 6329
    },
    {
      "epoch": 4.049904030710173,
      "grad_norm": 1.2740130712217974,
      "learning_rate": 1.3488154570371124e-05,
      "loss": 0.1491,
      "step": 6330
    },
    {
      "epoch": 4.050543825975688,
      "grad_norm": 1.2421951504495286,
      "learning_rate": 1.3486212490542446e-05,
      "loss": 0.1433,
      "step": 6331
    },
    {
      "epoch": 4.0511836212412025,
      "grad_norm": 1.2789041835018917,
      "learning_rate": 1.3484270261022005e-05,
      "loss": 0.1485,
      "step": 6332
    },
    {
      "epoch": 4.051823416506718,
      "grad_norm": 1.1461478500171645,
      "learning_rate": 1.3482327881893197e-05,
      "loss": 0.1345,
      "step": 6333
    },
    {
      "epoch": 4.052463211772233,
      "grad_norm": 1.3020229809584705,
      "learning_rate": 1.348038535323942e-05,
      "loss": 0.1358,
      "step": 6334
    },
    {
      "epoch": 4.053103007037748,
      "grad_norm": 1.1806241346958233,
      "learning_rate": 1.347844267514409e-05,
      "loss": 0.1288,
      "step": 6335
    },
    {
      "epoch": 4.053742802303263,
      "grad_norm": 1.2378596192612432,
      "learning_rate": 1.347649984769062e-05,
      "loss": 0.1439,
      "step": 6336
    },
    {
      "epoch": 4.054382597568778,
      "grad_norm": 1.2942763030028324,
      "learning_rate": 1.3474556870962427e-05,
      "loss": 0.1522,
      "step": 6337
    },
    {
      "epoch": 4.055022392834293,
      "grad_norm": 1.1492745179867783,
      "learning_rate": 1.3472613745042948e-05,
      "loss": 0.1292,
      "step": 6338
    },
    {
      "epoch": 4.0556621880998085,
      "grad_norm": 1.11437100133588,
      "learning_rate": 1.347067047001561e-05,
      "loss": 0.1333,
      "step": 6339
    },
    {
      "epoch": 4.056301983365323,
      "grad_norm": 1.196856768169684,
      "learning_rate": 1.3468727045963856e-05,
      "loss": 0.134,
      "step": 6340
    },
    {
      "epoch": 4.056941778630838,
      "grad_norm": 1.3207362126775108,
      "learning_rate": 1.3466783472971132e-05,
      "loss": 0.1413,
      "step": 6341
    },
    {
      "epoch": 4.057581573896353,
      "grad_norm": 1.1636992452355173,
      "learning_rate": 1.3464839751120893e-05,
      "loss": 0.1278,
      "step": 6342
    },
    {
      "epoch": 4.058221369161868,
      "grad_norm": 1.3792074941690393,
      "learning_rate": 1.34628958804966e-05,
      "loss": 0.1611,
      "step": 6343
    },
    {
      "epoch": 4.058861164427383,
      "grad_norm": 1.1709084203765012,
      "learning_rate": 1.3460951861181714e-05,
      "loss": 0.1351,
      "step": 6344
    },
    {
      "epoch": 4.059500959692898,
      "grad_norm": 1.3037706863287506,
      "learning_rate": 1.3459007693259713e-05,
      "loss": 0.1385,
      "step": 6345
    },
    {
      "epoch": 4.060140754958414,
      "grad_norm": 1.1767659552442509,
      "learning_rate": 1.345706337681408e-05,
      "loss": 0.1266,
      "step": 6346
    },
    {
      "epoch": 4.060780550223928,
      "grad_norm": 1.2526178653839881,
      "learning_rate": 1.345511891192829e-05,
      "loss": 0.1463,
      "step": 6347
    },
    {
      "epoch": 4.061420345489443,
      "grad_norm": 1.3818169004353231,
      "learning_rate": 1.3453174298685843e-05,
      "loss": 0.1521,
      "step": 6348
    },
    {
      "epoch": 4.062060140754959,
      "grad_norm": 1.3869254573068885,
      "learning_rate": 1.3451229537170232e-05,
      "loss": 0.15,
      "step": 6349
    },
    {
      "epoch": 4.062699936020474,
      "grad_norm": 1.1909996180068452,
      "learning_rate": 1.3449284627464963e-05,
      "loss": 0.1352,
      "step": 6350
    },
    {
      "epoch": 4.063339731285988,
      "grad_norm": 1.0714281676064903,
      "learning_rate": 1.3447339569653551e-05,
      "loss": 0.1253,
      "step": 6351
    },
    {
      "epoch": 4.0639795265515035,
      "grad_norm": 1.1907146258595573,
      "learning_rate": 1.3445394363819508e-05,
      "loss": 0.1241,
      "step": 6352
    },
    {
      "epoch": 4.064619321817019,
      "grad_norm": 1.2771147645814513,
      "learning_rate": 1.3443449010046361e-05,
      "loss": 0.1374,
      "step": 6353
    },
    {
      "epoch": 4.065259117082533,
      "grad_norm": 1.2348214692016344,
      "learning_rate": 1.3441503508417638e-05,
      "loss": 0.1358,
      "step": 6354
    },
    {
      "epoch": 4.0658989123480485,
      "grad_norm": 1.2159204222557858,
      "learning_rate": 1.3439557859016876e-05,
      "loss": 0.1365,
      "step": 6355
    },
    {
      "epoch": 4.066538707613564,
      "grad_norm": 1.267788912499374,
      "learning_rate": 1.343761206192762e-05,
      "loss": 0.1333,
      "step": 6356
    },
    {
      "epoch": 4.067178502879079,
      "grad_norm": 1.2277226108586325,
      "learning_rate": 1.3435666117233417e-05,
      "loss": 0.134,
      "step": 6357
    },
    {
      "epoch": 4.067818298144593,
      "grad_norm": 1.3432435469849795,
      "learning_rate": 1.3433720025017822e-05,
      "loss": 0.1398,
      "step": 6358
    },
    {
      "epoch": 4.068458093410109,
      "grad_norm": 1.2640515822445662,
      "learning_rate": 1.3431773785364395e-05,
      "loss": 0.1482,
      "step": 6359
    },
    {
      "epoch": 4.069097888675624,
      "grad_norm": 1.130558988880538,
      "learning_rate": 1.3429827398356707e-05,
      "loss": 0.1237,
      "step": 6360
    },
    {
      "epoch": 4.069737683941139,
      "grad_norm": 1.2306847340815021,
      "learning_rate": 1.3427880864078336e-05,
      "loss": 0.1341,
      "step": 6361
    },
    {
      "epoch": 4.070377479206654,
      "grad_norm": 1.2633156223159239,
      "learning_rate": 1.3425934182612854e-05,
      "loss": 0.1299,
      "step": 6362
    },
    {
      "epoch": 4.071017274472169,
      "grad_norm": 1.2518392527356295,
      "learning_rate": 1.3423987354043854e-05,
      "loss": 0.1376,
      "step": 6363
    },
    {
      "epoch": 4.071657069737684,
      "grad_norm": 1.244545866408988,
      "learning_rate": 1.3422040378454928e-05,
      "loss": 0.1395,
      "step": 6364
    },
    {
      "epoch": 4.0722968650031985,
      "grad_norm": 1.2736673614673621,
      "learning_rate": 1.3420093255929673e-05,
      "loss": 0.1349,
      "step": 6365
    },
    {
      "epoch": 4.072936660268714,
      "grad_norm": 1.183536202974554,
      "learning_rate": 1.34181459865517e-05,
      "loss": 0.1394,
      "step": 6366
    },
    {
      "epoch": 4.073576455534229,
      "grad_norm": 1.1661988705343957,
      "learning_rate": 1.3416198570404618e-05,
      "loss": 0.1448,
      "step": 6367
    },
    {
      "epoch": 4.074216250799744,
      "grad_norm": 1.1515853819068742,
      "learning_rate": 1.3414251007572048e-05,
      "loss": 0.1422,
      "step": 6368
    },
    {
      "epoch": 4.074856046065259,
      "grad_norm": 1.2574716708105114,
      "learning_rate": 1.3412303298137612e-05,
      "loss": 0.1362,
      "step": 6369
    },
    {
      "epoch": 4.075495841330774,
      "grad_norm": 1.1336360189482477,
      "learning_rate": 1.341035544218494e-05,
      "loss": 0.1254,
      "step": 6370
    },
    {
      "epoch": 4.076135636596289,
      "grad_norm": 1.192534515677174,
      "learning_rate": 1.3408407439797678e-05,
      "loss": 0.1492,
      "step": 6371
    },
    {
      "epoch": 4.076775431861805,
      "grad_norm": 1.2027889800712566,
      "learning_rate": 1.3406459291059458e-05,
      "loss": 0.1337,
      "step": 6372
    },
    {
      "epoch": 4.077415227127319,
      "grad_norm": 1.1813344145173965,
      "learning_rate": 1.3404510996053936e-05,
      "loss": 0.1313,
      "step": 6373
    },
    {
      "epoch": 4.078055022392834,
      "grad_norm": 1.283065200947973,
      "learning_rate": 1.3402562554864769e-05,
      "loss": 0.1475,
      "step": 6374
    },
    {
      "epoch": 4.0786948176583495,
      "grad_norm": 1.276777082017552,
      "learning_rate": 1.3400613967575618e-05,
      "loss": 0.1366,
      "step": 6375
    },
    {
      "epoch": 4.079334612923865,
      "grad_norm": 1.359385946682142,
      "learning_rate": 1.3398665234270154e-05,
      "loss": 0.1536,
      "step": 6376
    },
    {
      "epoch": 4.079974408189379,
      "grad_norm": 1.1400030970634032,
      "learning_rate": 1.3396716355032051e-05,
      "loss": 0.1214,
      "step": 6377
    },
    {
      "epoch": 4.0806142034548945,
      "grad_norm": 1.3730345745308075,
      "learning_rate": 1.3394767329944989e-05,
      "loss": 0.1444,
      "step": 6378
    },
    {
      "epoch": 4.08125399872041,
      "grad_norm": 1.2336371874987055,
      "learning_rate": 1.3392818159092658e-05,
      "loss": 0.1318,
      "step": 6379
    },
    {
      "epoch": 4.081893793985924,
      "grad_norm": 1.2100348131029486,
      "learning_rate": 1.3390868842558746e-05,
      "loss": 0.1298,
      "step": 6380
    },
    {
      "epoch": 4.082533589251439,
      "grad_norm": 1.2062809315954082,
      "learning_rate": 1.3388919380426964e-05,
      "loss": 0.1329,
      "step": 6381
    },
    {
      "epoch": 4.083173384516955,
      "grad_norm": 1.2455428387241778,
      "learning_rate": 1.3386969772781007e-05,
      "loss": 0.1372,
      "step": 6382
    },
    {
      "epoch": 4.08381317978247,
      "grad_norm": 1.2541079347488844,
      "learning_rate": 1.3385020019704595e-05,
      "loss": 0.1321,
      "step": 6383
    },
    {
      "epoch": 4.084452975047984,
      "grad_norm": 1.1437086567833468,
      "learning_rate": 1.3383070121281447e-05,
      "loss": 0.1169,
      "step": 6384
    },
    {
      "epoch": 4.0850927703135,
      "grad_norm": 1.0628010173474949,
      "learning_rate": 1.3381120077595282e-05,
      "loss": 0.1185,
      "step": 6385
    },
    {
      "epoch": 4.085732565579015,
      "grad_norm": 1.2568096442616108,
      "learning_rate": 1.3379169888729839e-05,
      "loss": 0.1424,
      "step": 6386
    },
    {
      "epoch": 4.08637236084453,
      "grad_norm": 1.135335751992726,
      "learning_rate": 1.337721955476885e-05,
      "loss": 0.1214,
      "step": 6387
    },
    {
      "epoch": 4.0870121561100445,
      "grad_norm": 1.266530148370118,
      "learning_rate": 1.3375269075796063e-05,
      "loss": 0.1356,
      "step": 6388
    },
    {
      "epoch": 4.08765195137556,
      "grad_norm": 1.132687010210815,
      "learning_rate": 1.337331845189522e-05,
      "loss": 0.1299,
      "step": 6389
    },
    {
      "epoch": 4.088291746641075,
      "grad_norm": 1.2860799402887864,
      "learning_rate": 1.3371367683150087e-05,
      "loss": 0.1423,
      "step": 6390
    },
    {
      "epoch": 4.0889315419065895,
      "grad_norm": 1.3506217313038966,
      "learning_rate": 1.3369416769644426e-05,
      "loss": 0.1478,
      "step": 6391
    },
    {
      "epoch": 4.089571337172105,
      "grad_norm": 1.1957334882014778,
      "learning_rate": 1.3367465711462e-05,
      "loss": 0.1291,
      "step": 6392
    },
    {
      "epoch": 4.09021113243762,
      "grad_norm": 1.2302654914907913,
      "learning_rate": 1.336551450868659e-05,
      "loss": 0.1325,
      "step": 6393
    },
    {
      "epoch": 4.090850927703135,
      "grad_norm": 1.1673950337365206,
      "learning_rate": 1.336356316140197e-05,
      "loss": 0.1428,
      "step": 6394
    },
    {
      "epoch": 4.09149072296865,
      "grad_norm": 1.1913533298498964,
      "learning_rate": 1.336161166969193e-05,
      "loss": 0.1416,
      "step": 6395
    },
    {
      "epoch": 4.092130518234165,
      "grad_norm": 1.0495495969579811,
      "learning_rate": 1.3359660033640267e-05,
      "loss": 0.1153,
      "step": 6396
    },
    {
      "epoch": 4.09277031349968,
      "grad_norm": 1.0420800152022553,
      "learning_rate": 1.3357708253330776e-05,
      "loss": 0.1225,
      "step": 6397
    },
    {
      "epoch": 4.0934101087651955,
      "grad_norm": 1.3196250634312288,
      "learning_rate": 1.3355756328847268e-05,
      "loss": 0.1472,
      "step": 6398
    },
    {
      "epoch": 4.09404990403071,
      "grad_norm": 1.275982928878984,
      "learning_rate": 1.3353804260273555e-05,
      "loss": 0.1288,
      "step": 6399
    },
    {
      "epoch": 4.094689699296225,
      "grad_norm": 1.2665903822320348,
      "learning_rate": 1.335185204769345e-05,
      "loss": 0.1429,
      "step": 6400
    },
    {
      "epoch": 4.0953294945617404,
      "grad_norm": 1.2370617754796092,
      "learning_rate": 1.334989969119078e-05,
      "loss": 0.1424,
      "step": 6401
    },
    {
      "epoch": 4.095969289827256,
      "grad_norm": 1.2861403344582834,
      "learning_rate": 1.3347947190849376e-05,
      "loss": 0.1437,
      "step": 6402
    },
    {
      "epoch": 4.09660908509277,
      "grad_norm": 1.1884682346550028,
      "learning_rate": 1.3345994546753078e-05,
      "loss": 0.1253,
      "step": 6403
    },
    {
      "epoch": 4.097248880358285,
      "grad_norm": 1.1485267183261632,
      "learning_rate": 1.3344041758985724e-05,
      "loss": 0.1226,
      "step": 6404
    },
    {
      "epoch": 4.097888675623801,
      "grad_norm": 1.20195825876628,
      "learning_rate": 1.3342088827631166e-05,
      "loss": 0.1236,
      "step": 6405
    },
    {
      "epoch": 4.098528470889315,
      "grad_norm": 1.2505838364661515,
      "learning_rate": 1.3340135752773258e-05,
      "loss": 0.1454,
      "step": 6406
    },
    {
      "epoch": 4.09916826615483,
      "grad_norm": 1.3605570955440103,
      "learning_rate": 1.3338182534495863e-05,
      "loss": 0.1517,
      "step": 6407
    },
    {
      "epoch": 4.099808061420346,
      "grad_norm": 1.2741245534964087,
      "learning_rate": 1.333622917288285e-05,
      "loss": 0.1498,
      "step": 6408
    },
    {
      "epoch": 4.100447856685861,
      "grad_norm": 1.2292473490748346,
      "learning_rate": 1.3334275668018089e-05,
      "loss": 0.1306,
      "step": 6409
    },
    {
      "epoch": 4.101087651951375,
      "grad_norm": 1.2755154760982679,
      "learning_rate": 1.3332322019985462e-05,
      "loss": 0.1426,
      "step": 6410
    },
    {
      "epoch": 4.1017274472168905,
      "grad_norm": 1.1461570237134693,
      "learning_rate": 1.3330368228868856e-05,
      "loss": 0.1209,
      "step": 6411
    },
    {
      "epoch": 4.102367242482406,
      "grad_norm": 1.1732766586669563,
      "learning_rate": 1.332841429475216e-05,
      "loss": 0.1345,
      "step": 6412
    },
    {
      "epoch": 4.103007037747921,
      "grad_norm": 1.2889483505247772,
      "learning_rate": 1.332646021771928e-05,
      "loss": 0.1379,
      "step": 6413
    },
    {
      "epoch": 4.1036468330134355,
      "grad_norm": 1.3079032093882328,
      "learning_rate": 1.3324505997854112e-05,
      "loss": 0.1367,
      "step": 6414
    },
    {
      "epoch": 4.104286628278951,
      "grad_norm": 1.209987766798407,
      "learning_rate": 1.3322551635240571e-05,
      "loss": 0.1367,
      "step": 6415
    },
    {
      "epoch": 4.104926423544466,
      "grad_norm": 1.2823774919424886,
      "learning_rate": 1.3320597129962573e-05,
      "loss": 0.1435,
      "step": 6416
    },
    {
      "epoch": 4.10556621880998,
      "grad_norm": 1.1623566830442156,
      "learning_rate": 1.3318642482104038e-05,
      "loss": 0.1366,
      "step": 6417
    },
    {
      "epoch": 4.106206014075496,
      "grad_norm": 1.2023304178416003,
      "learning_rate": 1.3316687691748904e-05,
      "loss": 0.1299,
      "step": 6418
    },
    {
      "epoch": 4.106845809341011,
      "grad_norm": 1.0600763366559338,
      "learning_rate": 1.3314732758981096e-05,
      "loss": 0.121,
      "step": 6419
    },
    {
      "epoch": 4.107485604606526,
      "grad_norm": 1.307832201690089,
      "learning_rate": 1.3312777683884564e-05,
      "loss": 0.13,
      "step": 6420
    },
    {
      "epoch": 4.108125399872041,
      "grad_norm": 1.317415359958145,
      "learning_rate": 1.3310822466543247e-05,
      "loss": 0.1334,
      "step": 6421
    },
    {
      "epoch": 4.108765195137556,
      "grad_norm": 1.2115367613705064,
      "learning_rate": 1.3308867107041106e-05,
      "loss": 0.1358,
      "step": 6422
    },
    {
      "epoch": 4.109404990403071,
      "grad_norm": 1.2288430406859923,
      "learning_rate": 1.3306911605462096e-05,
      "loss": 0.145,
      "step": 6423
    },
    {
      "epoch": 4.110044785668586,
      "grad_norm": 1.1921677256448031,
      "learning_rate": 1.3304955961890185e-05,
      "loss": 0.142,
      "step": 6424
    },
    {
      "epoch": 4.110684580934101,
      "grad_norm": 1.2152138408335575,
      "learning_rate": 1.3303000176409344e-05,
      "loss": 0.1391,
      "step": 6425
    },
    {
      "epoch": 4.111324376199616,
      "grad_norm": 1.1145780100079248,
      "learning_rate": 1.3301044249103549e-05,
      "loss": 0.1227,
      "step": 6426
    },
    {
      "epoch": 4.111964171465131,
      "grad_norm": 1.2992767875394933,
      "learning_rate": 1.3299088180056785e-05,
      "loss": 0.1419,
      "step": 6427
    },
    {
      "epoch": 4.112603966730646,
      "grad_norm": 1.2508760445679226,
      "learning_rate": 1.3297131969353048e-05,
      "loss": 0.1469,
      "step": 6428
    },
    {
      "epoch": 4.113243761996161,
      "grad_norm": 1.1465587448742194,
      "learning_rate": 1.3295175617076327e-05,
      "loss": 0.13,
      "step": 6429
    },
    {
      "epoch": 4.113883557261676,
      "grad_norm": 1.323339634145322,
      "learning_rate": 1.3293219123310628e-05,
      "loss": 0.1383,
      "step": 6430
    },
    {
      "epoch": 4.114523352527192,
      "grad_norm": 1.2393942402112832,
      "learning_rate": 1.3291262488139955e-05,
      "loss": 0.1338,
      "step": 6431
    },
    {
      "epoch": 4.115163147792706,
      "grad_norm": 1.1724248149805423,
      "learning_rate": 1.3289305711648326e-05,
      "loss": 0.1271,
      "step": 6432
    },
    {
      "epoch": 4.115802943058221,
      "grad_norm": 1.1656263774338613,
      "learning_rate": 1.3287348793919764e-05,
      "loss": 0.1249,
      "step": 6433
    },
    {
      "epoch": 4.1164427383237365,
      "grad_norm": 1.2457801486282314,
      "learning_rate": 1.3285391735038289e-05,
      "loss": 0.1435,
      "step": 6434
    },
    {
      "epoch": 4.117082533589252,
      "grad_norm": 1.2825812991221355,
      "learning_rate": 1.328343453508794e-05,
      "loss": 0.1435,
      "step": 6435
    },
    {
      "epoch": 4.117722328854766,
      "grad_norm": 1.2800370667182788,
      "learning_rate": 1.328147719415275e-05,
      "loss": 0.142,
      "step": 6436
    },
    {
      "epoch": 4.1183621241202815,
      "grad_norm": 1.1912206870149136,
      "learning_rate": 1.327951971231677e-05,
      "loss": 0.1384,
      "step": 6437
    },
    {
      "epoch": 4.119001919385797,
      "grad_norm": 1.2662324275860168,
      "learning_rate": 1.3277562089664047e-05,
      "loss": 0.1393,
      "step": 6438
    },
    {
      "epoch": 4.119641714651312,
      "grad_norm": 1.127562861638717,
      "learning_rate": 1.3275604326278634e-05,
      "loss": 0.1223,
      "step": 6439
    },
    {
      "epoch": 4.120281509916826,
      "grad_norm": 1.15318081367395,
      "learning_rate": 1.3273646422244601e-05,
      "loss": 0.1362,
      "step": 6440
    },
    {
      "epoch": 4.120921305182342,
      "grad_norm": 1.0767445449244486,
      "learning_rate": 1.3271688377646013e-05,
      "loss": 0.1232,
      "step": 6441
    },
    {
      "epoch": 4.121561100447857,
      "grad_norm": 1.2978246890758194,
      "learning_rate": 1.3269730192566947e-05,
      "loss": 0.1421,
      "step": 6442
    },
    {
      "epoch": 4.122200895713371,
      "grad_norm": 1.1678663371353926,
      "learning_rate": 1.3267771867091483e-05,
      "loss": 0.1234,
      "step": 6443
    },
    {
      "epoch": 4.122840690978887,
      "grad_norm": 1.1604782545150036,
      "learning_rate": 1.3265813401303709e-05,
      "loss": 0.1281,
      "step": 6444
    },
    {
      "epoch": 4.123480486244402,
      "grad_norm": 1.2824233117808819,
      "learning_rate": 1.3263854795287716e-05,
      "loss": 0.1411,
      "step": 6445
    },
    {
      "epoch": 4.124120281509917,
      "grad_norm": 1.2213509495705295,
      "learning_rate": 1.3261896049127603e-05,
      "loss": 0.1359,
      "step": 6446
    },
    {
      "epoch": 4.1247600767754315,
      "grad_norm": 1.2440001583834066,
      "learning_rate": 1.3259937162907476e-05,
      "loss": 0.139,
      "step": 6447
    },
    {
      "epoch": 4.125399872040947,
      "grad_norm": 1.216729403792445,
      "learning_rate": 1.3257978136711447e-05,
      "loss": 0.1278,
      "step": 6448
    },
    {
      "epoch": 4.126039667306462,
      "grad_norm": 1.3789953914709479,
      "learning_rate": 1.3256018970623632e-05,
      "loss": 0.1478,
      "step": 6449
    },
    {
      "epoch": 4.126679462571977,
      "grad_norm": 1.2969746383580198,
      "learning_rate": 1.3254059664728155e-05,
      "loss": 0.1455,
      "step": 6450
    },
    {
      "epoch": 4.127319257837492,
      "grad_norm": 1.1228580528205623,
      "learning_rate": 1.3252100219109143e-05,
      "loss": 0.1341,
      "step": 6451
    },
    {
      "epoch": 4.127959053103007,
      "grad_norm": 1.3742987612455813,
      "learning_rate": 1.3250140633850736e-05,
      "loss": 0.1475,
      "step": 6452
    },
    {
      "epoch": 4.128598848368522,
      "grad_norm": 1.1761051109295662,
      "learning_rate": 1.324818090903707e-05,
      "loss": 0.1396,
      "step": 6453
    },
    {
      "epoch": 4.129238643634037,
      "grad_norm": 1.2234522252102147,
      "learning_rate": 1.3246221044752292e-05,
      "loss": 0.1445,
      "step": 6454
    },
    {
      "epoch": 4.129878438899552,
      "grad_norm": 1.3706014383187957,
      "learning_rate": 1.3244261041080559e-05,
      "loss": 0.1488,
      "step": 6455
    },
    {
      "epoch": 4.130518234165067,
      "grad_norm": 1.296339302658757,
      "learning_rate": 1.3242300898106026e-05,
      "loss": 0.141,
      "step": 6456
    },
    {
      "epoch": 4.1311580294305825,
      "grad_norm": 1.2829438281096481,
      "learning_rate": 1.324034061591286e-05,
      "loss": 0.1473,
      "step": 6457
    },
    {
      "epoch": 4.131797824696097,
      "grad_norm": 1.1480780823015535,
      "learning_rate": 1.3238380194585237e-05,
      "loss": 0.1358,
      "step": 6458
    },
    {
      "epoch": 4.132437619961612,
      "grad_norm": 1.1755996264222823,
      "learning_rate": 1.3236419634207325e-05,
      "loss": 0.1304,
      "step": 6459
    },
    {
      "epoch": 4.1330774152271275,
      "grad_norm": 1.2922563173092538,
      "learning_rate": 1.3234458934863312e-05,
      "loss": 0.134,
      "step": 6460
    },
    {
      "epoch": 4.133717210492643,
      "grad_norm": 1.355910324297651,
      "learning_rate": 1.3232498096637384e-05,
      "loss": 0.1592,
      "step": 6461
    },
    {
      "epoch": 4.134357005758157,
      "grad_norm": 1.1773892088494275,
      "learning_rate": 1.3230537119613738e-05,
      "loss": 0.128,
      "step": 6462
    },
    {
      "epoch": 4.134996801023672,
      "grad_norm": 1.2748810147754377,
      "learning_rate": 1.3228576003876578e-05,
      "loss": 0.1498,
      "step": 6463
    },
    {
      "epoch": 4.135636596289188,
      "grad_norm": 1.2206440159573906,
      "learning_rate": 1.3226614749510105e-05,
      "loss": 0.1358,
      "step": 6464
    },
    {
      "epoch": 4.136276391554702,
      "grad_norm": 1.313422653491305,
      "learning_rate": 1.3224653356598535e-05,
      "loss": 0.1367,
      "step": 6465
    },
    {
      "epoch": 4.136916186820217,
      "grad_norm": 1.1548909753435237,
      "learning_rate": 1.3222691825226086e-05,
      "loss": 0.1335,
      "step": 6466
    },
    {
      "epoch": 4.137555982085733,
      "grad_norm": 1.2067284318435965,
      "learning_rate": 1.3220730155476984e-05,
      "loss": 0.1306,
      "step": 6467
    },
    {
      "epoch": 4.138195777351248,
      "grad_norm": 1.2547169113138512,
      "learning_rate": 1.3218768347435458e-05,
      "loss": 0.1484,
      "step": 6468
    },
    {
      "epoch": 4.138835572616762,
      "grad_norm": 1.3013481076625462,
      "learning_rate": 1.3216806401185744e-05,
      "loss": 0.1373,
      "step": 6469
    },
    {
      "epoch": 4.1394753678822775,
      "grad_norm": 1.2560203032173063,
      "learning_rate": 1.3214844316812086e-05,
      "loss": 0.1453,
      "step": 6470
    },
    {
      "epoch": 4.140115163147793,
      "grad_norm": 1.2238771549974987,
      "learning_rate": 1.3212882094398731e-05,
      "loss": 0.1421,
      "step": 6471
    },
    {
      "epoch": 4.140754958413308,
      "grad_norm": 1.2058931661616203,
      "learning_rate": 1.3210919734029932e-05,
      "loss": 0.1346,
      "step": 6472
    },
    {
      "epoch": 4.1413947536788225,
      "grad_norm": 1.2487708854303174,
      "learning_rate": 1.3208957235789957e-05,
      "loss": 0.1368,
      "step": 6473
    },
    {
      "epoch": 4.142034548944338,
      "grad_norm": 1.265598474607933,
      "learning_rate": 1.3206994599763064e-05,
      "loss": 0.1352,
      "step": 6474
    },
    {
      "epoch": 4.142674344209853,
      "grad_norm": 1.2785288092400215,
      "learning_rate": 1.3205031826033528e-05,
      "loss": 0.146,
      "step": 6475
    },
    {
      "epoch": 4.143314139475368,
      "grad_norm": 1.3085001515274943,
      "learning_rate": 1.3203068914685624e-05,
      "loss": 0.1394,
      "step": 6476
    },
    {
      "epoch": 4.143953934740883,
      "grad_norm": 1.336671368842908,
      "learning_rate": 1.3201105865803641e-05,
      "loss": 0.1512,
      "step": 6477
    },
    {
      "epoch": 4.144593730006398,
      "grad_norm": 1.233995092623335,
      "learning_rate": 1.3199142679471865e-05,
      "loss": 0.1483,
      "step": 6478
    },
    {
      "epoch": 4.145233525271913,
      "grad_norm": 1.1561886095015046,
      "learning_rate": 1.3197179355774594e-05,
      "loss": 0.1293,
      "step": 6479
    },
    {
      "epoch": 4.145873320537428,
      "grad_norm": 1.079743117480665,
      "learning_rate": 1.319521589479613e-05,
      "loss": 0.1328,
      "step": 6480
    },
    {
      "epoch": 4.146513115802943,
      "grad_norm": 1.2708554091927209,
      "learning_rate": 1.3193252296620775e-05,
      "loss": 0.1357,
      "step": 6481
    },
    {
      "epoch": 4.147152911068458,
      "grad_norm": 1.3970631090314216,
      "learning_rate": 1.3191288561332853e-05,
      "loss": 0.1641,
      "step": 6482
    },
    {
      "epoch": 4.1477927063339735,
      "grad_norm": 1.3226802664193427,
      "learning_rate": 1.3189324689016676e-05,
      "loss": 0.1459,
      "step": 6483
    },
    {
      "epoch": 4.148432501599488,
      "grad_norm": 1.3165572571816193,
      "learning_rate": 1.3187360679756567e-05,
      "loss": 0.1496,
      "step": 6484
    },
    {
      "epoch": 4.149072296865003,
      "grad_norm": 1.2253690681313096,
      "learning_rate": 1.3185396533636862e-05,
      "loss": 0.1381,
      "step": 6485
    },
    {
      "epoch": 4.149712092130518,
      "grad_norm": 1.1949801943803502,
      "learning_rate": 1.3183432250741894e-05,
      "loss": 0.1366,
      "step": 6486
    },
    {
      "epoch": 4.150351887396034,
      "grad_norm": 1.2394050660133615,
      "learning_rate": 1.3181467831156011e-05,
      "loss": 0.1385,
      "step": 6487
    },
    {
      "epoch": 4.150991682661548,
      "grad_norm": 1.2940283368882937,
      "learning_rate": 1.3179503274963557e-05,
      "loss": 0.147,
      "step": 6488
    },
    {
      "epoch": 4.151631477927063,
      "grad_norm": 1.3000041449154258,
      "learning_rate": 1.317753858224889e-05,
      "loss": 0.1471,
      "step": 6489
    },
    {
      "epoch": 4.152271273192579,
      "grad_norm": 1.2403778347384142,
      "learning_rate": 1.3175573753096367e-05,
      "loss": 0.1309,
      "step": 6490
    },
    {
      "epoch": 4.152911068458093,
      "grad_norm": 1.1695220117355303,
      "learning_rate": 1.3173608787590358e-05,
      "loss": 0.1353,
      "step": 6491
    },
    {
      "epoch": 4.153550863723608,
      "grad_norm": 1.3238962702386379,
      "learning_rate": 1.3171643685815232e-05,
      "loss": 0.1362,
      "step": 6492
    },
    {
      "epoch": 4.1541906589891235,
      "grad_norm": 1.170436510761068,
      "learning_rate": 1.3169678447855366e-05,
      "loss": 0.1281,
      "step": 6493
    },
    {
      "epoch": 4.154830454254639,
      "grad_norm": 1.10735823968886,
      "learning_rate": 1.3167713073795145e-05,
      "loss": 0.1253,
      "step": 6494
    },
    {
      "epoch": 4.155470249520153,
      "grad_norm": 1.22172820979919,
      "learning_rate": 1.3165747563718965e-05,
      "loss": 0.1371,
      "step": 6495
    },
    {
      "epoch": 4.1561100447856685,
      "grad_norm": 1.4197818675301002,
      "learning_rate": 1.3163781917711212e-05,
      "loss": 0.1483,
      "step": 6496
    },
    {
      "epoch": 4.156749840051184,
      "grad_norm": 1.3011498499216392,
      "learning_rate": 1.3161816135856294e-05,
      "loss": 0.1515,
      "step": 6497
    },
    {
      "epoch": 4.157389635316699,
      "grad_norm": 1.2444813936218817,
      "learning_rate": 1.3159850218238614e-05,
      "loss": 0.1394,
      "step": 6498
    },
    {
      "epoch": 4.158029430582213,
      "grad_norm": 1.3101000408491517,
      "learning_rate": 1.3157884164942585e-05,
      "loss": 0.1365,
      "step": 6499
    },
    {
      "epoch": 4.158669225847729,
      "grad_norm": 1.2675778914366247,
      "learning_rate": 1.315591797605263e-05,
      "loss": 0.1422,
      "step": 6500
    },
    {
      "epoch": 4.159309021113244,
      "grad_norm": 1.3758004309511094,
      "learning_rate": 1.3153951651653167e-05,
      "loss": 0.1429,
      "step": 6501
    },
    {
      "epoch": 4.159948816378758,
      "grad_norm": 1.2612918163237836,
      "learning_rate": 1.3151985191828635e-05,
      "loss": 0.1471,
      "step": 6502
    },
    {
      "epoch": 4.160588611644274,
      "grad_norm": 1.2518374724485224,
      "learning_rate": 1.3150018596663463e-05,
      "loss": 0.1408,
      "step": 6503
    },
    {
      "epoch": 4.161228406909789,
      "grad_norm": 1.3637696475088317,
      "learning_rate": 1.3148051866242099e-05,
      "loss": 0.1497,
      "step": 6504
    },
    {
      "epoch": 4.161868202175304,
      "grad_norm": 1.2936952828445756,
      "learning_rate": 1.3146085000648987e-05,
      "loss": 0.1439,
      "step": 6505
    },
    {
      "epoch": 4.162507997440819,
      "grad_norm": 1.3079529115878328,
      "learning_rate": 1.314411799996858e-05,
      "loss": 0.1536,
      "step": 6506
    },
    {
      "epoch": 4.163147792706334,
      "grad_norm": 1.298720777715506,
      "learning_rate": 1.314215086428534e-05,
      "loss": 0.1485,
      "step": 6507
    },
    {
      "epoch": 4.163787587971849,
      "grad_norm": 1.2518015857169345,
      "learning_rate": 1.3140183593683731e-05,
      "loss": 0.1468,
      "step": 6508
    },
    {
      "epoch": 4.164427383237364,
      "grad_norm": 1.3417466765734316,
      "learning_rate": 1.3138216188248224e-05,
      "loss": 0.1554,
      "step": 6509
    },
    {
      "epoch": 4.165067178502879,
      "grad_norm": 1.2695200533957451,
      "learning_rate": 1.3136248648063296e-05,
      "loss": 0.1456,
      "step": 6510
    },
    {
      "epoch": 4.165706973768394,
      "grad_norm": 1.139545197246476,
      "learning_rate": 1.3134280973213432e-05,
      "loss": 0.14,
      "step": 6511
    },
    {
      "epoch": 4.166346769033909,
      "grad_norm": 1.0368694860447114,
      "learning_rate": 1.3132313163783118e-05,
      "loss": 0.1236,
      "step": 6512
    },
    {
      "epoch": 4.166986564299425,
      "grad_norm": 1.2166200305688422,
      "learning_rate": 1.3130345219856846e-05,
      "loss": 0.1384,
      "step": 6513
    },
    {
      "epoch": 4.167626359564939,
      "grad_norm": 1.1987093011545926,
      "learning_rate": 1.3128377141519119e-05,
      "loss": 0.1368,
      "step": 6514
    },
    {
      "epoch": 4.168266154830454,
      "grad_norm": 1.237490498492935,
      "learning_rate": 1.3126408928854447e-05,
      "loss": 0.1368,
      "step": 6515
    },
    {
      "epoch": 4.1689059500959695,
      "grad_norm": 1.2803703176620405,
      "learning_rate": 1.3124440581947332e-05,
      "loss": 0.1258,
      "step": 6516
    },
    {
      "epoch": 4.169545745361484,
      "grad_norm": 1.3191465371891724,
      "learning_rate": 1.31224721008823e-05,
      "loss": 0.1413,
      "step": 6517
    },
    {
      "epoch": 4.170185540626999,
      "grad_norm": 1.2631007320658443,
      "learning_rate": 1.3120503485743867e-05,
      "loss": 0.1268,
      "step": 6518
    },
    {
      "epoch": 4.1708253358925145,
      "grad_norm": 1.1701053850342378,
      "learning_rate": 1.3118534736616568e-05,
      "loss": 0.1225,
      "step": 6519
    },
    {
      "epoch": 4.17146513115803,
      "grad_norm": 1.2392605957587304,
      "learning_rate": 1.3116565853584937e-05,
      "loss": 0.1265,
      "step": 6520
    },
    {
      "epoch": 4.172104926423544,
      "grad_norm": 1.293237613808912,
      "learning_rate": 1.3114596836733507e-05,
      "loss": 0.1595,
      "step": 6521
    },
    {
      "epoch": 4.172744721689059,
      "grad_norm": 1.250881166002565,
      "learning_rate": 1.3112627686146833e-05,
      "loss": 0.1351,
      "step": 6522
    },
    {
      "epoch": 4.173384516954575,
      "grad_norm": 1.3081457311849232,
      "learning_rate": 1.3110658401909459e-05,
      "loss": 0.1448,
      "step": 6523
    },
    {
      "epoch": 4.17402431222009,
      "grad_norm": 1.2484104476167852,
      "learning_rate": 1.310868898410595e-05,
      "loss": 0.1462,
      "step": 6524
    },
    {
      "epoch": 4.174664107485604,
      "grad_norm": 1.1198269039039048,
      "learning_rate": 1.3106719432820864e-05,
      "loss": 0.1448,
      "step": 6525
    },
    {
      "epoch": 4.17530390275112,
      "grad_norm": 1.0930863577656353,
      "learning_rate": 1.3104749748138775e-05,
      "loss": 0.1352,
      "step": 6526
    },
    {
      "epoch": 4.175943698016635,
      "grad_norm": 1.1491381201930568,
      "learning_rate": 1.3102779930144252e-05,
      "loss": 0.1371,
      "step": 6527
    },
    {
      "epoch": 4.176583493282149,
      "grad_norm": 1.2209673015246114,
      "learning_rate": 1.3100809978921882e-05,
      "loss": 0.138,
      "step": 6528
    },
    {
      "epoch": 4.177223288547665,
      "grad_norm": 1.276464888135971,
      "learning_rate": 1.3098839894556243e-05,
      "loss": 0.1473,
      "step": 6529
    },
    {
      "epoch": 4.17786308381318,
      "grad_norm": 1.283718079813373,
      "learning_rate": 1.3096869677131933e-05,
      "loss": 0.1362,
      "step": 6530
    },
    {
      "epoch": 4.178502879078695,
      "grad_norm": 1.2290606181542971,
      "learning_rate": 1.3094899326733548e-05,
      "loss": 0.1325,
      "step": 6531
    },
    {
      "epoch": 4.1791426743442095,
      "grad_norm": 1.271235875629337,
      "learning_rate": 1.3092928843445691e-05,
      "loss": 0.1353,
      "step": 6532
    },
    {
      "epoch": 4.179782469609725,
      "grad_norm": 1.2526768038518956,
      "learning_rate": 1.309095822735297e-05,
      "loss": 0.1329,
      "step": 6533
    },
    {
      "epoch": 4.18042226487524,
      "grad_norm": 1.333376792162786,
      "learning_rate": 1.3088987478540006e-05,
      "loss": 0.1444,
      "step": 6534
    },
    {
      "epoch": 4.181062060140755,
      "grad_norm": 1.2853838744803292,
      "learning_rate": 1.3087016597091411e-05,
      "loss": 0.1431,
      "step": 6535
    },
    {
      "epoch": 4.18170185540627,
      "grad_norm": 1.2046742684922058,
      "learning_rate": 1.3085045583091813e-05,
      "loss": 0.1268,
      "step": 6536
    },
    {
      "epoch": 4.182341650671785,
      "grad_norm": 1.2696582961651395,
      "learning_rate": 1.3083074436625849e-05,
      "loss": 0.1419,
      "step": 6537
    },
    {
      "epoch": 4.1829814459373,
      "grad_norm": 1.0841214161779238,
      "learning_rate": 1.308110315777815e-05,
      "loss": 0.128,
      "step": 6538
    },
    {
      "epoch": 4.183621241202815,
      "grad_norm": 1.2503322519993953,
      "learning_rate": 1.3079131746633363e-05,
      "loss": 0.1366,
      "step": 6539
    },
    {
      "epoch": 4.18426103646833,
      "grad_norm": 1.2549142616375968,
      "learning_rate": 1.3077160203276138e-05,
      "loss": 0.1426,
      "step": 6540
    },
    {
      "epoch": 4.184900831733845,
      "grad_norm": 1.311062734520131,
      "learning_rate": 1.3075188527791126e-05,
      "loss": 0.1423,
      "step": 6541
    },
    {
      "epoch": 4.1855406269993605,
      "grad_norm": 1.1598427814815269,
      "learning_rate": 1.3073216720262989e-05,
      "loss": 0.1262,
      "step": 6542
    },
    {
      "epoch": 4.186180422264875,
      "grad_norm": 1.2966930445809575,
      "learning_rate": 1.3071244780776395e-05,
      "loss": 0.1532,
      "step": 6543
    },
    {
      "epoch": 4.18682021753039,
      "grad_norm": 1.2654019047293943,
      "learning_rate": 1.3069272709416011e-05,
      "loss": 0.1346,
      "step": 6544
    },
    {
      "epoch": 4.187460012795905,
      "grad_norm": 1.2309160696893604,
      "learning_rate": 1.3067300506266518e-05,
      "loss": 0.1366,
      "step": 6545
    },
    {
      "epoch": 4.188099808061421,
      "grad_norm": 1.3058539805842697,
      "learning_rate": 1.3065328171412597e-05,
      "loss": 0.1435,
      "step": 6546
    },
    {
      "epoch": 4.188739603326935,
      "grad_norm": 1.2448058215277413,
      "learning_rate": 1.3063355704938938e-05,
      "loss": 0.1408,
      "step": 6547
    },
    {
      "epoch": 4.18937939859245,
      "grad_norm": 1.227666552204223,
      "learning_rate": 1.3061383106930232e-05,
      "loss": 0.139,
      "step": 6548
    },
    {
      "epoch": 4.190019193857966,
      "grad_norm": 1.3622699876463635,
      "learning_rate": 1.3059410377471185e-05,
      "loss": 0.15,
      "step": 6549
    },
    {
      "epoch": 4.190658989123481,
      "grad_norm": 1.1770384979984008,
      "learning_rate": 1.3057437516646497e-05,
      "loss": 0.1309,
      "step": 6550
    },
    {
      "epoch": 4.191298784388995,
      "grad_norm": 1.2459902353687315,
      "learning_rate": 1.3055464524540877e-05,
      "loss": 0.1376,
      "step": 6551
    },
    {
      "epoch": 4.1919385796545106,
      "grad_norm": 1.2015715696936051,
      "learning_rate": 1.3053491401239051e-05,
      "loss": 0.1452,
      "step": 6552
    },
    {
      "epoch": 4.192578374920026,
      "grad_norm": 1.1824360280584667,
      "learning_rate": 1.3051518146825732e-05,
      "loss": 0.139,
      "step": 6553
    },
    {
      "epoch": 4.19321817018554,
      "grad_norm": 1.2534545549672955,
      "learning_rate": 1.3049544761385653e-05,
      "loss": 0.1419,
      "step": 6554
    },
    {
      "epoch": 4.1938579654510555,
      "grad_norm": 1.3127126036529375,
      "learning_rate": 1.3047571245003549e-05,
      "loss": 0.1437,
      "step": 6555
    },
    {
      "epoch": 4.194497760716571,
      "grad_norm": 1.234707172259688,
      "learning_rate": 1.3045597597764154e-05,
      "loss": 0.1383,
      "step": 6556
    },
    {
      "epoch": 4.195137555982086,
      "grad_norm": 1.2347969566238293,
      "learning_rate": 1.3043623819752218e-05,
      "loss": 0.1412,
      "step": 6557
    },
    {
      "epoch": 4.1957773512476,
      "grad_norm": 1.2207067773457116,
      "learning_rate": 1.304164991105249e-05,
      "loss": 0.1385,
      "step": 6558
    },
    {
      "epoch": 4.196417146513116,
      "grad_norm": 1.2530812217657448,
      "learning_rate": 1.3039675871749724e-05,
      "loss": 0.1503,
      "step": 6559
    },
    {
      "epoch": 4.197056941778631,
      "grad_norm": 1.1913833278412587,
      "learning_rate": 1.3037701701928682e-05,
      "loss": 0.1372,
      "step": 6560
    },
    {
      "epoch": 4.197696737044146,
      "grad_norm": 1.3018759707534322,
      "learning_rate": 1.3035727401674132e-05,
      "loss": 0.1495,
      "step": 6561
    },
    {
      "epoch": 4.198336532309661,
      "grad_norm": 1.1876027001086458,
      "learning_rate": 1.3033752971070852e-05,
      "loss": 0.1378,
      "step": 6562
    },
    {
      "epoch": 4.198976327575176,
      "grad_norm": 1.297180304047282,
      "learning_rate": 1.3031778410203612e-05,
      "loss": 0.1442,
      "step": 6563
    },
    {
      "epoch": 4.199616122840691,
      "grad_norm": 1.2864554476646293,
      "learning_rate": 1.3029803719157201e-05,
      "loss": 0.1404,
      "step": 6564
    },
    {
      "epoch": 4.2002559181062065,
      "grad_norm": 1.2337648190360058,
      "learning_rate": 1.302782889801641e-05,
      "loss": 0.1427,
      "step": 6565
    },
    {
      "epoch": 4.200895713371721,
      "grad_norm": 1.2615686901596987,
      "learning_rate": 1.3025853946866029e-05,
      "loss": 0.1499,
      "step": 6566
    },
    {
      "epoch": 4.201535508637236,
      "grad_norm": 1.246897557763951,
      "learning_rate": 1.3023878865790867e-05,
      "loss": 0.1414,
      "step": 6567
    },
    {
      "epoch": 4.202175303902751,
      "grad_norm": 1.288083485897449,
      "learning_rate": 1.302190365487572e-05,
      "loss": 0.1399,
      "step": 6568
    },
    {
      "epoch": 4.202815099168266,
      "grad_norm": 1.2226237610251622,
      "learning_rate": 1.3019928314205409e-05,
      "loss": 0.1474,
      "step": 6569
    },
    {
      "epoch": 4.203454894433781,
      "grad_norm": 1.2593275756727058,
      "learning_rate": 1.3017952843864748e-05,
      "loss": 0.1404,
      "step": 6570
    },
    {
      "epoch": 4.204094689699296,
      "grad_norm": 1.409837574798889,
      "learning_rate": 1.3015977243938558e-05,
      "loss": 0.1443,
      "step": 6571
    },
    {
      "epoch": 4.204734484964812,
      "grad_norm": 1.2513969507508196,
      "learning_rate": 1.3014001514511672e-05,
      "loss": 0.1308,
      "step": 6572
    },
    {
      "epoch": 4.205374280230326,
      "grad_norm": 1.2171148948709989,
      "learning_rate": 1.3012025655668923e-05,
      "loss": 0.1372,
      "step": 6573
    },
    {
      "epoch": 4.206014075495841,
      "grad_norm": 1.188244825069588,
      "learning_rate": 1.301004966749515e-05,
      "loss": 0.1264,
      "step": 6574
    },
    {
      "epoch": 4.2066538707613566,
      "grad_norm": 1.2393476744126237,
      "learning_rate": 1.3008073550075198e-05,
      "loss": 0.1386,
      "step": 6575
    },
    {
      "epoch": 4.207293666026872,
      "grad_norm": 1.291668512651603,
      "learning_rate": 1.3006097303493917e-05,
      "loss": 0.1392,
      "step": 6576
    },
    {
      "epoch": 4.207933461292386,
      "grad_norm": 1.0904225538407522,
      "learning_rate": 1.3004120927836168e-05,
      "loss": 0.1212,
      "step": 6577
    },
    {
      "epoch": 4.2085732565579015,
      "grad_norm": 1.2491572216785212,
      "learning_rate": 1.300214442318681e-05,
      "loss": 0.1372,
      "step": 6578
    },
    {
      "epoch": 4.209213051823417,
      "grad_norm": 1.35167396264769,
      "learning_rate": 1.300016778963071e-05,
      "loss": 0.1348,
      "step": 6579
    },
    {
      "epoch": 4.209852847088931,
      "grad_norm": 1.1888628206825715,
      "learning_rate": 1.2998191027252742e-05,
      "loss": 0.1349,
      "step": 6580
    },
    {
      "epoch": 4.210492642354446,
      "grad_norm": 1.1810173726354583,
      "learning_rate": 1.2996214136137786e-05,
      "loss": 0.1273,
      "step": 6581
    },
    {
      "epoch": 4.211132437619962,
      "grad_norm": 1.1995130647204375,
      "learning_rate": 1.2994237116370726e-05,
      "loss": 0.1263,
      "step": 6582
    },
    {
      "epoch": 4.211772232885477,
      "grad_norm": 1.3846125789474262,
      "learning_rate": 1.2992259968036444e-05,
      "loss": 0.1542,
      "step": 6583
    },
    {
      "epoch": 4.212412028150991,
      "grad_norm": 1.223779287767398,
      "learning_rate": 1.2990282691219848e-05,
      "loss": 0.1336,
      "step": 6584
    },
    {
      "epoch": 4.213051823416507,
      "grad_norm": 1.2090593461760313,
      "learning_rate": 1.298830528600583e-05,
      "loss": 0.1398,
      "step": 6585
    },
    {
      "epoch": 4.213691618682022,
      "grad_norm": 1.3776548362751786,
      "learning_rate": 1.2986327752479298e-05,
      "loss": 0.1549,
      "step": 6586
    },
    {
      "epoch": 4.214331413947537,
      "grad_norm": 1.3105770008982454,
      "learning_rate": 1.2984350090725166e-05,
      "loss": 0.1425,
      "step": 6587
    },
    {
      "epoch": 4.214971209213052,
      "grad_norm": 1.4464960327340137,
      "learning_rate": 1.2982372300828352e-05,
      "loss": 0.1483,
      "step": 6588
    },
    {
      "epoch": 4.215611004478567,
      "grad_norm": 1.3837961338835463,
      "learning_rate": 1.2980394382873775e-05,
      "loss": 0.1408,
      "step": 6589
    },
    {
      "epoch": 4.216250799744082,
      "grad_norm": 1.2901066290954672,
      "learning_rate": 1.2978416336946364e-05,
      "loss": 0.1501,
      "step": 6590
    },
    {
      "epoch": 4.2168905950095965,
      "grad_norm": 1.2742795767872908,
      "learning_rate": 1.2976438163131053e-05,
      "loss": 0.1345,
      "step": 6591
    },
    {
      "epoch": 4.217530390275112,
      "grad_norm": 1.2230270674555066,
      "learning_rate": 1.2974459861512784e-05,
      "loss": 0.1425,
      "step": 6592
    },
    {
      "epoch": 4.218170185540627,
      "grad_norm": 1.259008436038498,
      "learning_rate": 1.2972481432176497e-05,
      "loss": 0.1372,
      "step": 6593
    },
    {
      "epoch": 4.218809980806142,
      "grad_norm": 1.2876572663173813,
      "learning_rate": 1.297050287520715e-05,
      "loss": 0.1476,
      "step": 6594
    },
    {
      "epoch": 4.219449776071657,
      "grad_norm": 1.299093676281208,
      "learning_rate": 1.296852419068969e-05,
      "loss": 0.1395,
      "step": 6595
    },
    {
      "epoch": 4.220089571337172,
      "grad_norm": 1.2122651229227277,
      "learning_rate": 1.2966545378709081e-05,
      "loss": 0.1371,
      "step": 6596
    },
    {
      "epoch": 4.220729366602687,
      "grad_norm": 1.3523400900417635,
      "learning_rate": 1.2964566439350292e-05,
      "loss": 0.1505,
      "step": 6597
    },
    {
      "epoch": 4.2213691618682025,
      "grad_norm": 1.238383447615782,
      "learning_rate": 1.2962587372698293e-05,
      "loss": 0.1363,
      "step": 6598
    },
    {
      "epoch": 4.222008957133717,
      "grad_norm": 1.2382433036299978,
      "learning_rate": 1.2960608178838064e-05,
      "loss": 0.1355,
      "step": 6599
    },
    {
      "epoch": 4.222648752399232,
      "grad_norm": 1.2339263048175853,
      "learning_rate": 1.2958628857854584e-05,
      "loss": 0.1457,
      "step": 6600
    },
    {
      "epoch": 4.2232885476647475,
      "grad_norm": 1.2512702408586773,
      "learning_rate": 1.2956649409832842e-05,
      "loss": 0.1456,
      "step": 6601
    },
    {
      "epoch": 4.223928342930263,
      "grad_norm": 1.353092779682745,
      "learning_rate": 1.2954669834857838e-05,
      "loss": 0.1387,
      "step": 6602
    },
    {
      "epoch": 4.224568138195777,
      "grad_norm": 1.2202496885708798,
      "learning_rate": 1.2952690133014567e-05,
      "loss": 0.1404,
      "step": 6603
    },
    {
      "epoch": 4.225207933461292,
      "grad_norm": 1.188494523479143,
      "learning_rate": 1.2950710304388034e-05,
      "loss": 0.1339,
      "step": 6604
    },
    {
      "epoch": 4.225847728726808,
      "grad_norm": 1.3837764130438432,
      "learning_rate": 1.2948730349063247e-05,
      "loss": 0.1309,
      "step": 6605
    },
    {
      "epoch": 4.226487523992322,
      "grad_norm": 1.277713180159131,
      "learning_rate": 1.2946750267125226e-05,
      "loss": 0.1473,
      "step": 6606
    },
    {
      "epoch": 4.227127319257837,
      "grad_norm": 1.3034773801050916,
      "learning_rate": 1.2944770058658991e-05,
      "loss": 0.1419,
      "step": 6607
    },
    {
      "epoch": 4.227767114523353,
      "grad_norm": 1.2970905939654551,
      "learning_rate": 1.2942789723749564e-05,
      "loss": 0.1479,
      "step": 6608
    },
    {
      "epoch": 4.228406909788868,
      "grad_norm": 1.2362997618176168,
      "learning_rate": 1.2940809262481988e-05,
      "loss": 0.1337,
      "step": 6609
    },
    {
      "epoch": 4.229046705054382,
      "grad_norm": 1.2340937090626631,
      "learning_rate": 1.2938828674941291e-05,
      "loss": 0.1386,
      "step": 6610
    },
    {
      "epoch": 4.229686500319898,
      "grad_norm": 1.3918190180649839,
      "learning_rate": 1.2936847961212516e-05,
      "loss": 0.1574,
      "step": 6611
    },
    {
      "epoch": 4.230326295585413,
      "grad_norm": 1.3081712694708272,
      "learning_rate": 1.2934867121380714e-05,
      "loss": 0.1449,
      "step": 6612
    },
    {
      "epoch": 4.230966090850928,
      "grad_norm": 1.4059964979271353,
      "learning_rate": 1.2932886155530939e-05,
      "loss": 0.1544,
      "step": 6613
    },
    {
      "epoch": 4.2316058861164425,
      "grad_norm": 1.3039944549830629,
      "learning_rate": 1.293090506374825e-05,
      "loss": 0.1446,
      "step": 6614
    },
    {
      "epoch": 4.232245681381958,
      "grad_norm": 1.202641062096954,
      "learning_rate": 1.2928923846117712e-05,
      "loss": 0.1327,
      "step": 6615
    },
    {
      "epoch": 4.232885476647473,
      "grad_norm": 1.1474995165446324,
      "learning_rate": 1.2926942502724391e-05,
      "loss": 0.1333,
      "step": 6616
    },
    {
      "epoch": 4.233525271912987,
      "grad_norm": 1.2532915651541277,
      "learning_rate": 1.2924961033653369e-05,
      "loss": 0.135,
      "step": 6617
    },
    {
      "epoch": 4.234165067178503,
      "grad_norm": 1.2003254045202878,
      "learning_rate": 1.2922979438989725e-05,
      "loss": 0.1337,
      "step": 6618
    },
    {
      "epoch": 4.234804862444018,
      "grad_norm": 1.3160493266181392,
      "learning_rate": 1.292099771881854e-05,
      "loss": 0.1437,
      "step": 6619
    },
    {
      "epoch": 4.235444657709533,
      "grad_norm": 1.234689939130522,
      "learning_rate": 1.2919015873224909e-05,
      "loss": 0.1376,
      "step": 6620
    },
    {
      "epoch": 4.236084452975048,
      "grad_norm": 1.3284127466987896,
      "learning_rate": 1.291703390229393e-05,
      "loss": 0.1423,
      "step": 6621
    },
    {
      "epoch": 4.236724248240563,
      "grad_norm": 1.19300675616378,
      "learning_rate": 1.2915051806110706e-05,
      "loss": 0.1326,
      "step": 6622
    },
    {
      "epoch": 4.237364043506078,
      "grad_norm": 1.1560040281185868,
      "learning_rate": 1.2913069584760339e-05,
      "loss": 0.1273,
      "step": 6623
    },
    {
      "epoch": 4.2380038387715935,
      "grad_norm": 1.2193442449842689,
      "learning_rate": 1.2911087238327948e-05,
      "loss": 0.1368,
      "step": 6624
    },
    {
      "epoch": 4.238643634037108,
      "grad_norm": 1.359959751185359,
      "learning_rate": 1.290910476689865e-05,
      "loss": 0.1497,
      "step": 6625
    },
    {
      "epoch": 4.239283429302623,
      "grad_norm": 1.335703361952032,
      "learning_rate": 1.2907122170557567e-05,
      "loss": 0.1502,
      "step": 6626
    },
    {
      "epoch": 4.239923224568138,
      "grad_norm": 1.2317146383781132,
      "learning_rate": 1.2905139449389831e-05,
      "loss": 0.1484,
      "step": 6627
    },
    {
      "epoch": 4.240563019833653,
      "grad_norm": 1.274494276140197,
      "learning_rate": 1.2903156603480573e-05,
      "loss": 0.1427,
      "step": 6628
    },
    {
      "epoch": 4.241202815099168,
      "grad_norm": 1.2641677685717299,
      "learning_rate": 1.2901173632914937e-05,
      "loss": 0.1462,
      "step": 6629
    },
    {
      "epoch": 4.241842610364683,
      "grad_norm": 1.3284392951170327,
      "learning_rate": 1.2899190537778063e-05,
      "loss": 0.145,
      "step": 6630
    },
    {
      "epoch": 4.242482405630199,
      "grad_norm": 1.1524971660613466,
      "learning_rate": 1.2897207318155106e-05,
      "loss": 0.1269,
      "step": 6631
    },
    {
      "epoch": 4.243122200895713,
      "grad_norm": 1.2816655453146708,
      "learning_rate": 1.2895223974131221e-05,
      "loss": 0.147,
      "step": 6632
    },
    {
      "epoch": 4.243761996161228,
      "grad_norm": 1.2550504996648406,
      "learning_rate": 1.2893240505791569e-05,
      "loss": 0.1252,
      "step": 6633
    },
    {
      "epoch": 4.244401791426744,
      "grad_norm": 1.2409228467639646,
      "learning_rate": 1.2891256913221317e-05,
      "loss": 0.1444,
      "step": 6634
    },
    {
      "epoch": 4.245041586692259,
      "grad_norm": 1.2323902731712686,
      "learning_rate": 1.2889273196505633e-05,
      "loss": 0.1437,
      "step": 6635
    },
    {
      "epoch": 4.245681381957773,
      "grad_norm": 1.2868263013948018,
      "learning_rate": 1.2887289355729702e-05,
      "loss": 0.1504,
      "step": 6636
    },
    {
      "epoch": 4.2463211772232885,
      "grad_norm": 1.3015578670020773,
      "learning_rate": 1.2885305390978698e-05,
      "loss": 0.1477,
      "step": 6637
    },
    {
      "epoch": 4.246960972488804,
      "grad_norm": 1.2304955960510022,
      "learning_rate": 1.2883321302337814e-05,
      "loss": 0.1369,
      "step": 6638
    },
    {
      "epoch": 4.247600767754319,
      "grad_norm": 1.3530623333588199,
      "learning_rate": 1.2881337089892245e-05,
      "loss": 0.1459,
      "step": 6639
    },
    {
      "epoch": 4.248240563019833,
      "grad_norm": 1.2908508239356369,
      "learning_rate": 1.2879352753727185e-05,
      "loss": 0.1329,
      "step": 6640
    },
    {
      "epoch": 4.248880358285349,
      "grad_norm": 1.2171111083668131,
      "learning_rate": 1.2877368293927841e-05,
      "loss": 0.1335,
      "step": 6641
    },
    {
      "epoch": 4.249520153550864,
      "grad_norm": 1.3057822801400392,
      "learning_rate": 1.287538371057942e-05,
      "loss": 0.1474,
      "step": 6642
    },
    {
      "epoch": 4.250159948816378,
      "grad_norm": 1.2320732768586131,
      "learning_rate": 1.2873399003767135e-05,
      "loss": 0.1315,
      "step": 6643
    },
    {
      "epoch": 4.250799744081894,
      "grad_norm": 1.2342499097254411,
      "learning_rate": 1.2871414173576211e-05,
      "loss": 0.141,
      "step": 6644
    },
    {
      "epoch": 4.251439539347409,
      "grad_norm": 1.177645259710656,
      "learning_rate": 1.2869429220091867e-05,
      "loss": 0.1282,
      "step": 6645
    },
    {
      "epoch": 4.252079334612924,
      "grad_norm": 1.1030746325476815,
      "learning_rate": 1.2867444143399341e-05,
      "loss": 0.1277,
      "step": 6646
    },
    {
      "epoch": 4.252719129878439,
      "grad_norm": 1.1678655646765719,
      "learning_rate": 1.2865458943583862e-05,
      "loss": 0.1226,
      "step": 6647
    },
    {
      "epoch": 4.253358925143954,
      "grad_norm": 1.1582738699690847,
      "learning_rate": 1.2863473620730673e-05,
      "loss": 0.1278,
      "step": 6648
    },
    {
      "epoch": 4.253998720409469,
      "grad_norm": 1.3393498921930793,
      "learning_rate": 1.2861488174925024e-05,
      "loss": 0.1461,
      "step": 6649
    },
    {
      "epoch": 4.254638515674984,
      "grad_norm": 1.2118269415905063,
      "learning_rate": 1.2859502606252158e-05,
      "loss": 0.1445,
      "step": 6650
    },
    {
      "epoch": 4.255278310940499,
      "grad_norm": 1.2288745871413314,
      "learning_rate": 1.2857516914797339e-05,
      "loss": 0.1383,
      "step": 6651
    },
    {
      "epoch": 4.255918106206014,
      "grad_norm": 1.2852721815937764,
      "learning_rate": 1.2855531100645826e-05,
      "loss": 0.1392,
      "step": 6652
    },
    {
      "epoch": 4.256557901471529,
      "grad_norm": 1.3415942439351844,
      "learning_rate": 1.2853545163882886e-05,
      "loss": 0.1483,
      "step": 6653
    },
    {
      "epoch": 4.257197696737044,
      "grad_norm": 1.319489111448922,
      "learning_rate": 1.2851559104593795e-05,
      "loss": 0.1487,
      "step": 6654
    },
    {
      "epoch": 4.257837492002559,
      "grad_norm": 1.2634403706305155,
      "learning_rate": 1.284957292286383e-05,
      "loss": 0.1585,
      "step": 6655
    },
    {
      "epoch": 4.258477287268074,
      "grad_norm": 1.2539988048024266,
      "learning_rate": 1.2847586618778272e-05,
      "loss": 0.1452,
      "step": 6656
    },
    {
      "epoch": 4.25911708253359,
      "grad_norm": 1.1772533976543063,
      "learning_rate": 1.2845600192422407e-05,
      "loss": 0.1375,
      "step": 6657
    },
    {
      "epoch": 4.259756877799104,
      "grad_norm": 1.3418125944164154,
      "learning_rate": 1.2843613643881532e-05,
      "loss": 0.1515,
      "step": 6658
    },
    {
      "epoch": 4.260396673064619,
      "grad_norm": 1.4269118703504353,
      "learning_rate": 1.2841626973240948e-05,
      "loss": 0.152,
      "step": 6659
    },
    {
      "epoch": 4.2610364683301345,
      "grad_norm": 1.3596079263237206,
      "learning_rate": 1.2839640180585955e-05,
      "loss": 0.1465,
      "step": 6660
    },
    {
      "epoch": 4.26167626359565,
      "grad_norm": 1.3875926501939893,
      "learning_rate": 1.2837653266001863e-05,
      "loss": 0.1431,
      "step": 6661
    },
    {
      "epoch": 4.262316058861164,
      "grad_norm": 1.2191733754964857,
      "learning_rate": 1.2835666229573991e-05,
      "loss": 0.1298,
      "step": 6662
    },
    {
      "epoch": 4.262955854126679,
      "grad_norm": 1.3425855086364091,
      "learning_rate": 1.2833679071387652e-05,
      "loss": 0.14,
      "step": 6663
    },
    {
      "epoch": 4.263595649392195,
      "grad_norm": 1.2651172705434366,
      "learning_rate": 1.2831691791528177e-05,
      "loss": 0.1473,
      "step": 6664
    },
    {
      "epoch": 4.264235444657709,
      "grad_norm": 1.299656244896035,
      "learning_rate": 1.2829704390080893e-05,
      "loss": 0.1437,
      "step": 6665
    },
    {
      "epoch": 4.264875239923224,
      "grad_norm": 1.275978947087283,
      "learning_rate": 1.2827716867131135e-05,
      "loss": 0.1439,
      "step": 6666
    },
    {
      "epoch": 4.26551503518874,
      "grad_norm": 1.188743462690417,
      "learning_rate": 1.2825729222764245e-05,
      "loss": 0.1313,
      "step": 6667
    },
    {
      "epoch": 4.266154830454255,
      "grad_norm": 1.2251980455560576,
      "learning_rate": 1.2823741457065568e-05,
      "loss": 0.1244,
      "step": 6668
    },
    {
      "epoch": 4.266794625719769,
      "grad_norm": 1.4286502810329684,
      "learning_rate": 1.2821753570120458e-05,
      "loss": 0.1565,
      "step": 6669
    },
    {
      "epoch": 4.267434420985285,
      "grad_norm": 1.1722344839546408,
      "learning_rate": 1.281976556201427e-05,
      "loss": 0.1302,
      "step": 6670
    },
    {
      "epoch": 4.2680742162508,
      "grad_norm": 1.3756125246245892,
      "learning_rate": 1.2817777432832364e-05,
      "loss": 0.1591,
      "step": 6671
    },
    {
      "epoch": 4.268714011516315,
      "grad_norm": 1.2771577393875602,
      "learning_rate": 1.2815789182660106e-05,
      "loss": 0.1514,
      "step": 6672
    },
    {
      "epoch": 4.2693538067818295,
      "grad_norm": 1.20743890545672,
      "learning_rate": 1.2813800811582868e-05,
      "loss": 0.1376,
      "step": 6673
    },
    {
      "epoch": 4.269993602047345,
      "grad_norm": 1.2996637454400453,
      "learning_rate": 1.2811812319686033e-05,
      "loss": 0.1467,
      "step": 6674
    },
    {
      "epoch": 4.27063339731286,
      "grad_norm": 1.2305198321625421,
      "learning_rate": 1.2809823707054974e-05,
      "loss": 0.138,
      "step": 6675
    },
    {
      "epoch": 4.271273192578375,
      "grad_norm": 1.1521142607278876,
      "learning_rate": 1.2807834973775086e-05,
      "loss": 0.1308,
      "step": 6676
    },
    {
      "epoch": 4.27191298784389,
      "grad_norm": 1.1940880955054087,
      "learning_rate": 1.2805846119931757e-05,
      "loss": 0.1321,
      "step": 6677
    },
    {
      "epoch": 4.272552783109405,
      "grad_norm": 1.305432840619432,
      "learning_rate": 1.280385714561039e-05,
      "loss": 0.1553,
      "step": 6678
    },
    {
      "epoch": 4.27319257837492,
      "grad_norm": 1.1120687186944986,
      "learning_rate": 1.2801868050896383e-05,
      "loss": 0.1208,
      "step": 6679
    },
    {
      "epoch": 4.273832373640435,
      "grad_norm": 1.2920193494703345,
      "learning_rate": 1.2799878835875146e-05,
      "loss": 0.1362,
      "step": 6680
    },
    {
      "epoch": 4.27447216890595,
      "grad_norm": 1.2988288174199492,
      "learning_rate": 1.2797889500632093e-05,
      "loss": 0.1564,
      "step": 6681
    },
    {
      "epoch": 4.275111964171465,
      "grad_norm": 1.2011802484585072,
      "learning_rate": 1.2795900045252641e-05,
      "loss": 0.146,
      "step": 6682
    },
    {
      "epoch": 4.2757517594369805,
      "grad_norm": 1.2852437477010434,
      "learning_rate": 1.2793910469822213e-05,
      "loss": 0.1464,
      "step": 6683
    },
    {
      "epoch": 4.276391554702495,
      "grad_norm": 1.2955876496356296,
      "learning_rate": 1.2791920774426244e-05,
      "loss": 0.1485,
      "step": 6684
    },
    {
      "epoch": 4.27703134996801,
      "grad_norm": 1.2619349479686028,
      "learning_rate": 1.2789930959150161e-05,
      "loss": 0.1357,
      "step": 6685
    },
    {
      "epoch": 4.277671145233525,
      "grad_norm": 1.4584593796399297,
      "learning_rate": 1.2787941024079407e-05,
      "loss": 0.1614,
      "step": 6686
    },
    {
      "epoch": 4.278310940499041,
      "grad_norm": 1.1382132200817845,
      "learning_rate": 1.2785950969299421e-05,
      "loss": 0.1324,
      "step": 6687
    },
    {
      "epoch": 4.278950735764555,
      "grad_norm": 1.2539285460904037,
      "learning_rate": 1.2783960794895658e-05,
      "loss": 0.1477,
      "step": 6688
    },
    {
      "epoch": 4.27959053103007,
      "grad_norm": 1.1323119306856986,
      "learning_rate": 1.2781970500953571e-05,
      "loss": 0.1235,
      "step": 6689
    },
    {
      "epoch": 4.280230326295586,
      "grad_norm": 1.185088941246974,
      "learning_rate": 1.2779980087558619e-05,
      "loss": 0.1341,
      "step": 6690
    },
    {
      "epoch": 4.280870121561101,
      "grad_norm": 1.2210729899640094,
      "learning_rate": 1.2777989554796269e-05,
      "loss": 0.129,
      "step": 6691
    },
    {
      "epoch": 4.281509916826615,
      "grad_norm": 1.3632296400278714,
      "learning_rate": 1.2775998902751986e-05,
      "loss": 0.1459,
      "step": 6692
    },
    {
      "epoch": 4.282149712092131,
      "grad_norm": 1.3636221111523181,
      "learning_rate": 1.2774008131511254e-05,
      "loss": 0.1511,
      "step": 6693
    },
    {
      "epoch": 4.282789507357646,
      "grad_norm": 1.2749382875293247,
      "learning_rate": 1.2772017241159542e-05,
      "loss": 0.1444,
      "step": 6694
    },
    {
      "epoch": 4.28342930262316,
      "grad_norm": 1.3451668956627458,
      "learning_rate": 1.2770026231782342e-05,
      "loss": 0.1457,
      "step": 6695
    },
    {
      "epoch": 4.2840690978886755,
      "grad_norm": 1.1486369495986226,
      "learning_rate": 1.2768035103465143e-05,
      "loss": 0.1305,
      "step": 6696
    },
    {
      "epoch": 4.284708893154191,
      "grad_norm": 1.2702221148739141,
      "learning_rate": 1.276604385629344e-05,
      "loss": 0.1381,
      "step": 6697
    },
    {
      "epoch": 4.285348688419706,
      "grad_norm": 1.270004033483836,
      "learning_rate": 1.2764052490352734e-05,
      "loss": 0.1428,
      "step": 6698
    },
    {
      "epoch": 4.2859884836852205,
      "grad_norm": 1.352610820543988,
      "learning_rate": 1.2762061005728534e-05,
      "loss": 0.1419,
      "step": 6699
    },
    {
      "epoch": 4.286628278950736,
      "grad_norm": 1.3988971644277544,
      "learning_rate": 1.2760069402506348e-05,
      "loss": 0.1509,
      "step": 6700
    },
    {
      "epoch": 4.287268074216251,
      "grad_norm": 1.2034708772560183,
      "learning_rate": 1.275807768077169e-05,
      "loss": 0.1313,
      "step": 6701
    },
    {
      "epoch": 4.287907869481765,
      "grad_norm": 1.3960445025461095,
      "learning_rate": 1.2756085840610081e-05,
      "loss": 0.1427,
      "step": 6702
    },
    {
      "epoch": 4.288547664747281,
      "grad_norm": 1.1580276297637762,
      "learning_rate": 1.2754093882107048e-05,
      "loss": 0.1397,
      "step": 6703
    },
    {
      "epoch": 4.289187460012796,
      "grad_norm": 1.1661076431705983,
      "learning_rate": 1.2752101805348127e-05,
      "loss": 0.1247,
      "step": 6704
    },
    {
      "epoch": 4.289827255278311,
      "grad_norm": 1.2895948078071666,
      "learning_rate": 1.2750109610418847e-05,
      "loss": 0.1358,
      "step": 6705
    },
    {
      "epoch": 4.290467050543826,
      "grad_norm": 1.2429373784598745,
      "learning_rate": 1.2748117297404754e-05,
      "loss": 0.1355,
      "step": 6706
    },
    {
      "epoch": 4.291106845809341,
      "grad_norm": 1.2792774438097068,
      "learning_rate": 1.2746124866391391e-05,
      "loss": 0.1413,
      "step": 6707
    },
    {
      "epoch": 4.291746641074856,
      "grad_norm": 1.398899182352771,
      "learning_rate": 1.2744132317464313e-05,
      "loss": 0.1532,
      "step": 6708
    },
    {
      "epoch": 4.292386436340371,
      "grad_norm": 1.3615369331775642,
      "learning_rate": 1.2742139650709073e-05,
      "loss": 0.1448,
      "step": 6709
    },
    {
      "epoch": 4.293026231605886,
      "grad_norm": 1.3600795059579351,
      "learning_rate": 1.2740146866211236e-05,
      "loss": 0.1415,
      "step": 6710
    },
    {
      "epoch": 4.293666026871401,
      "grad_norm": 1.2599358886378553,
      "learning_rate": 1.2738153964056367e-05,
      "loss": 0.1413,
      "step": 6711
    },
    {
      "epoch": 4.294305822136916,
      "grad_norm": 1.345542182556654,
      "learning_rate": 1.2736160944330034e-05,
      "loss": 0.1477,
      "step": 6712
    },
    {
      "epoch": 4.294945617402432,
      "grad_norm": 1.3276146861458633,
      "learning_rate": 1.2734167807117821e-05,
      "loss": 0.158,
      "step": 6713
    },
    {
      "epoch": 4.295585412667946,
      "grad_norm": 1.2398092774323612,
      "learning_rate": 1.2732174552505306e-05,
      "loss": 0.1358,
      "step": 6714
    },
    {
      "epoch": 4.296225207933461,
      "grad_norm": 1.2202638059100783,
      "learning_rate": 1.2730181180578077e-05,
      "loss": 0.1401,
      "step": 6715
    },
    {
      "epoch": 4.296865003198977,
      "grad_norm": 1.2455559088848847,
      "learning_rate": 1.2728187691421723e-05,
      "loss": 0.1337,
      "step": 6716
    },
    {
      "epoch": 4.297504798464491,
      "grad_norm": 1.3907069216150199,
      "learning_rate": 1.272619408512184e-05,
      "loss": 0.1441,
      "step": 6717
    },
    {
      "epoch": 4.298144593730006,
      "grad_norm": 1.3570891150986144,
      "learning_rate": 1.2724200361764036e-05,
      "loss": 0.1571,
      "step": 6718
    },
    {
      "epoch": 4.2987843889955215,
      "grad_norm": 1.300277759725778,
      "learning_rate": 1.2722206521433912e-05,
      "loss": 0.1399,
      "step": 6719
    },
    {
      "epoch": 4.299424184261037,
      "grad_norm": 1.2580868612627776,
      "learning_rate": 1.2720212564217083e-05,
      "loss": 0.1439,
      "step": 6720
    },
    {
      "epoch": 4.300063979526551,
      "grad_norm": 1.230158596809827,
      "learning_rate": 1.2718218490199165e-05,
      "loss": 0.131,
      "step": 6721
    },
    {
      "epoch": 4.3007037747920664,
      "grad_norm": 1.2651675265041695,
      "learning_rate": 1.2716224299465782e-05,
      "loss": 0.1417,
      "step": 6722
    },
    {
      "epoch": 4.301343570057582,
      "grad_norm": 1.138292365354985,
      "learning_rate": 1.2714229992102563e-05,
      "loss": 0.131,
      "step": 6723
    },
    {
      "epoch": 4.301983365323097,
      "grad_norm": 1.4379584879918035,
      "learning_rate": 1.2712235568195128e-05,
      "loss": 0.1478,
      "step": 6724
    },
    {
      "epoch": 4.302623160588611,
      "grad_norm": 1.3304967398919452,
      "learning_rate": 1.2710241027829126e-05,
      "loss": 0.1404,
      "step": 6725
    },
    {
      "epoch": 4.303262955854127,
      "grad_norm": 1.303321146761973,
      "learning_rate": 1.2708246371090198e-05,
      "loss": 0.1426,
      "step": 6726
    },
    {
      "epoch": 4.303902751119642,
      "grad_norm": 1.3236424459120526,
      "learning_rate": 1.2706251598063985e-05,
      "loss": 0.1438,
      "step": 6727
    },
    {
      "epoch": 4.304542546385157,
      "grad_norm": 1.214041650955164,
      "learning_rate": 1.2704256708836145e-05,
      "loss": 0.1447,
      "step": 6728
    },
    {
      "epoch": 4.305182341650672,
      "grad_norm": 1.2801815130960719,
      "learning_rate": 1.2702261703492329e-05,
      "loss": 0.1352,
      "step": 6729
    },
    {
      "epoch": 4.305822136916187,
      "grad_norm": 1.373550457570286,
      "learning_rate": 1.2700266582118205e-05,
      "loss": 0.1466,
      "step": 6730
    },
    {
      "epoch": 4.306461932181702,
      "grad_norm": 1.2355161875225182,
      "learning_rate": 1.269827134479944e-05,
      "loss": 0.1449,
      "step": 6731
    },
    {
      "epoch": 4.3071017274472165,
      "grad_norm": 1.1900160237041153,
      "learning_rate": 1.2696275991621699e-05,
      "loss": 0.1415,
      "step": 6732
    },
    {
      "epoch": 4.307741522712732,
      "grad_norm": 1.2534700537697372,
      "learning_rate": 1.2694280522670665e-05,
      "loss": 0.143,
      "step": 6733
    },
    {
      "epoch": 4.308381317978247,
      "grad_norm": 1.2076049058159606,
      "learning_rate": 1.2692284938032016e-05,
      "loss": 0.137,
      "step": 6734
    },
    {
      "epoch": 4.309021113243762,
      "grad_norm": 1.1369476087467054,
      "learning_rate": 1.2690289237791441e-05,
      "loss": 0.1253,
      "step": 6735
    },
    {
      "epoch": 4.309660908509277,
      "grad_norm": 1.244451951052904,
      "learning_rate": 1.2688293422034635e-05,
      "loss": 0.1323,
      "step": 6736
    },
    {
      "epoch": 4.310300703774792,
      "grad_norm": 1.2707302792388864,
      "learning_rate": 1.268629749084729e-05,
      "loss": 0.1407,
      "step": 6737
    },
    {
      "epoch": 4.310940499040307,
      "grad_norm": 1.343750486624378,
      "learning_rate": 1.2684301444315113e-05,
      "loss": 0.1546,
      "step": 6738
    },
    {
      "epoch": 4.311580294305822,
      "grad_norm": 1.4092790190117797,
      "learning_rate": 1.2682305282523802e-05,
      "loss": 0.1534,
      "step": 6739
    },
    {
      "epoch": 4.312220089571337,
      "grad_norm": 1.2636311531981457,
      "learning_rate": 1.2680309005559073e-05,
      "loss": 0.1356,
      "step": 6740
    },
    {
      "epoch": 4.312859884836852,
      "grad_norm": 1.2869952336554156,
      "learning_rate": 1.2678312613506644e-05,
      "loss": 0.1477,
      "step": 6741
    },
    {
      "epoch": 4.3134996801023675,
      "grad_norm": 1.2022061049664843,
      "learning_rate": 1.2676316106452236e-05,
      "loss": 0.1295,
      "step": 6742
    },
    {
      "epoch": 4.314139475367882,
      "grad_norm": 1.2308093373058908,
      "learning_rate": 1.2674319484481577e-05,
      "loss": 0.1421,
      "step": 6743
    },
    {
      "epoch": 4.314779270633397,
      "grad_norm": 1.2402286723416418,
      "learning_rate": 1.2672322747680395e-05,
      "loss": 0.1307,
      "step": 6744
    },
    {
      "epoch": 4.315419065898912,
      "grad_norm": 1.3633765225502648,
      "learning_rate": 1.267032589613443e-05,
      "loss": 0.1446,
      "step": 6745
    },
    {
      "epoch": 4.316058861164428,
      "grad_norm": 1.301871712347017,
      "learning_rate": 1.266832892992942e-05,
      "loss": 0.1494,
      "step": 6746
    },
    {
      "epoch": 4.316698656429942,
      "grad_norm": 1.2938003237353535,
      "learning_rate": 1.266633184915111e-05,
      "loss": 0.1535,
      "step": 6747
    },
    {
      "epoch": 4.317338451695457,
      "grad_norm": 1.3215474838848877,
      "learning_rate": 1.2664334653885257e-05,
      "loss": 0.1585,
      "step": 6748
    },
    {
      "epoch": 4.317978246960973,
      "grad_norm": 1.3501573900830572,
      "learning_rate": 1.266233734421761e-05,
      "loss": 0.1351,
      "step": 6749
    },
    {
      "epoch": 4.318618042226488,
      "grad_norm": 1.2826628613387763,
      "learning_rate": 1.2660339920233936e-05,
      "loss": 0.1474,
      "step": 6750
    },
    {
      "epoch": 4.319257837492002,
      "grad_norm": 1.1677204123416542,
      "learning_rate": 1.2658342382019996e-05,
      "loss": 0.1366,
      "step": 6751
    },
    {
      "epoch": 4.319897632757518,
      "grad_norm": 1.3281392887234116,
      "learning_rate": 1.2656344729661567e-05,
      "loss": 0.1447,
      "step": 6752
    },
    {
      "epoch": 4.320537428023033,
      "grad_norm": 1.3343465300665434,
      "learning_rate": 1.2654346963244417e-05,
      "loss": 0.1436,
      "step": 6753
    },
    {
      "epoch": 4.321177223288547,
      "grad_norm": 1.270308298365699,
      "learning_rate": 1.2652349082854332e-05,
      "loss": 0.1433,
      "step": 6754
    },
    {
      "epoch": 4.3218170185540625,
      "grad_norm": 1.2574232394846003,
      "learning_rate": 1.2650351088577093e-05,
      "loss": 0.1402,
      "step": 6755
    },
    {
      "epoch": 4.322456813819578,
      "grad_norm": 1.2867558098126008,
      "learning_rate": 1.2648352980498495e-05,
      "loss": 0.1374,
      "step": 6756
    },
    {
      "epoch": 4.323096609085093,
      "grad_norm": 1.147170546152236,
      "learning_rate": 1.2646354758704328e-05,
      "loss": 0.1282,
      "step": 6757
    },
    {
      "epoch": 4.3237364043506075,
      "grad_norm": 1.1694987903195833,
      "learning_rate": 1.2644356423280399e-05,
      "loss": 0.1358,
      "step": 6758
    },
    {
      "epoch": 4.324376199616123,
      "grad_norm": 1.2129398679317123,
      "learning_rate": 1.2642357974312507e-05,
      "loss": 0.1352,
      "step": 6759
    },
    {
      "epoch": 4.325015994881638,
      "grad_norm": 1.2731763375588718,
      "learning_rate": 1.2640359411886467e-05,
      "loss": 0.135,
      "step": 6760
    },
    {
      "epoch": 4.325655790147153,
      "grad_norm": 1.2052126268802092,
      "learning_rate": 1.2638360736088089e-05,
      "loss": 0.1301,
      "step": 6761
    },
    {
      "epoch": 4.326295585412668,
      "grad_norm": 1.2589909392687355,
      "learning_rate": 1.2636361947003192e-05,
      "loss": 0.1267,
      "step": 6762
    },
    {
      "epoch": 4.326935380678183,
      "grad_norm": 1.2818248702525326,
      "learning_rate": 1.2634363044717606e-05,
      "loss": 0.1345,
      "step": 6763
    },
    {
      "epoch": 4.327575175943698,
      "grad_norm": 1.252173846058181,
      "learning_rate": 1.2632364029317155e-05,
      "loss": 0.1429,
      "step": 6764
    },
    {
      "epoch": 4.3282149712092135,
      "grad_norm": 1.2172454513288329,
      "learning_rate": 1.2630364900887673e-05,
      "loss": 0.1413,
      "step": 6765
    },
    {
      "epoch": 4.328854766474728,
      "grad_norm": 1.2376341081842048,
      "learning_rate": 1.2628365659515005e-05,
      "loss": 0.1334,
      "step": 6766
    },
    {
      "epoch": 4.329494561740243,
      "grad_norm": 1.2887722182602241,
      "learning_rate": 1.2626366305284991e-05,
      "loss": 0.1496,
      "step": 6767
    },
    {
      "epoch": 4.330134357005758,
      "grad_norm": 1.261127412396774,
      "learning_rate": 1.2624366838283479e-05,
      "loss": 0.1365,
      "step": 6768
    },
    {
      "epoch": 4.330774152271273,
      "grad_norm": 1.4196273433030246,
      "learning_rate": 1.2622367258596322e-05,
      "loss": 0.1477,
      "step": 6769
    },
    {
      "epoch": 4.331413947536788,
      "grad_norm": 1.2415381372748904,
      "learning_rate": 1.2620367566309381e-05,
      "loss": 0.1314,
      "step": 6770
    },
    {
      "epoch": 4.332053742802303,
      "grad_norm": 1.1715121800799835,
      "learning_rate": 1.261836776150852e-05,
      "loss": 0.1309,
      "step": 6771
    },
    {
      "epoch": 4.332693538067819,
      "grad_norm": 1.148605890417872,
      "learning_rate": 1.26163678442796e-05,
      "loss": 0.1286,
      "step": 6772
    },
    {
      "epoch": 4.333333333333333,
      "grad_norm": 1.3289349051314117,
      "learning_rate": 1.2614367814708503e-05,
      "loss": 0.1338,
      "step": 6773
    },
    {
      "epoch": 4.333973128598848,
      "grad_norm": 1.2531629633252686,
      "learning_rate": 1.26123676728811e-05,
      "loss": 0.1435,
      "step": 6774
    },
    {
      "epoch": 4.334612923864364,
      "grad_norm": 1.1873959370168357,
      "learning_rate": 1.2610367418883278e-05,
      "loss": 0.1269,
      "step": 6775
    },
    {
      "epoch": 4.335252719129878,
      "grad_norm": 1.144957957853231,
      "learning_rate": 1.2608367052800924e-05,
      "loss": 0.1327,
      "step": 6776
    },
    {
      "epoch": 4.335892514395393,
      "grad_norm": 1.3605286657129962,
      "learning_rate": 1.2606366574719926e-05,
      "loss": 0.1451,
      "step": 6777
    },
    {
      "epoch": 4.3365323096609085,
      "grad_norm": 1.2249004533916152,
      "learning_rate": 1.2604365984726186e-05,
      "loss": 0.1407,
      "step": 6778
    },
    {
      "epoch": 4.337172104926424,
      "grad_norm": 1.3162992710086763,
      "learning_rate": 1.26023652829056e-05,
      "loss": 0.1477,
      "step": 6779
    },
    {
      "epoch": 4.337811900191938,
      "grad_norm": 1.317544101557365,
      "learning_rate": 1.2600364469344077e-05,
      "loss": 0.1419,
      "step": 6780
    },
    {
      "epoch": 4.3384516954574535,
      "grad_norm": 1.2665333238636238,
      "learning_rate": 1.2598363544127535e-05,
      "loss": 0.1428,
      "step": 6781
    },
    {
      "epoch": 4.339091490722969,
      "grad_norm": 1.3751967422169569,
      "learning_rate": 1.2596362507341881e-05,
      "loss": 0.1444,
      "step": 6782
    },
    {
      "epoch": 4.339731285988484,
      "grad_norm": 1.3926759314858619,
      "learning_rate": 1.2594361359073042e-05,
      "loss": 0.144,
      "step": 6783
    },
    {
      "epoch": 4.340371081253998,
      "grad_norm": 1.3147953879309455,
      "learning_rate": 1.259236009940694e-05,
      "loss": 0.1475,
      "step": 6784
    },
    {
      "epoch": 4.341010876519514,
      "grad_norm": 1.3023015879275186,
      "learning_rate": 1.2590358728429504e-05,
      "loss": 0.1399,
      "step": 6785
    },
    {
      "epoch": 4.341650671785029,
      "grad_norm": 1.173815921700054,
      "learning_rate": 1.2588357246226676e-05,
      "loss": 0.1387,
      "step": 6786
    },
    {
      "epoch": 4.342290467050544,
      "grad_norm": 1.3129854657920372,
      "learning_rate": 1.2586355652884387e-05,
      "loss": 0.1413,
      "step": 6787
    },
    {
      "epoch": 4.342930262316059,
      "grad_norm": 1.2773045554646498,
      "learning_rate": 1.2584353948488594e-05,
      "loss": 0.1409,
      "step": 6788
    },
    {
      "epoch": 4.343570057581574,
      "grad_norm": 1.192025501936864,
      "learning_rate": 1.2582352133125235e-05,
      "loss": 0.1334,
      "step": 6789
    },
    {
      "epoch": 4.344209852847089,
      "grad_norm": 1.2742068714391674,
      "learning_rate": 1.258035020688027e-05,
      "loss": 0.1396,
      "step": 6790
    },
    {
      "epoch": 4.3448496481126035,
      "grad_norm": 1.339769841294734,
      "learning_rate": 1.2578348169839658e-05,
      "loss": 0.1515,
      "step": 6791
    },
    {
      "epoch": 4.345489443378119,
      "grad_norm": 1.3142463287264121,
      "learning_rate": 1.257634602208936e-05,
      "loss": 0.1464,
      "step": 6792
    },
    {
      "epoch": 4.346129238643634,
      "grad_norm": 1.2992029156151117,
      "learning_rate": 1.2574343763715351e-05,
      "loss": 0.1437,
      "step": 6793
    },
    {
      "epoch": 4.346769033909149,
      "grad_norm": 1.256047980991541,
      "learning_rate": 1.2572341394803597e-05,
      "loss": 0.1391,
      "step": 6794
    },
    {
      "epoch": 4.347408829174664,
      "grad_norm": 1.256757992314281,
      "learning_rate": 1.2570338915440081e-05,
      "loss": 0.1392,
      "step": 6795
    },
    {
      "epoch": 4.348048624440179,
      "grad_norm": 1.1586566481550045,
      "learning_rate": 1.2568336325710782e-05,
      "loss": 0.1411,
      "step": 6796
    },
    {
      "epoch": 4.348688419705694,
      "grad_norm": 1.3704858548161178,
      "learning_rate": 1.2566333625701692e-05,
      "loss": 0.157,
      "step": 6797
    },
    {
      "epoch": 4.34932821497121,
      "grad_norm": 1.3380014924143562,
      "learning_rate": 1.2564330815498801e-05,
      "loss": 0.1484,
      "step": 6798
    },
    {
      "epoch": 4.349968010236724,
      "grad_norm": 1.312055864160971,
      "learning_rate": 1.2562327895188104e-05,
      "loss": 0.14,
      "step": 6799
    },
    {
      "epoch": 4.350607805502239,
      "grad_norm": 1.2159378546999735,
      "learning_rate": 1.256032486485561e-05,
      "loss": 0.1339,
      "step": 6800
    },
    {
      "epoch": 4.3512476007677545,
      "grad_norm": 1.2014621576893296,
      "learning_rate": 1.2558321724587315e-05,
      "loss": 0.1278,
      "step": 6801
    },
    {
      "epoch": 4.35188739603327,
      "grad_norm": 1.426417159992691,
      "learning_rate": 1.2556318474469239e-05,
      "loss": 0.1568,
      "step": 6802
    },
    {
      "epoch": 4.352527191298784,
      "grad_norm": 1.2523769600499217,
      "learning_rate": 1.2554315114587395e-05,
      "loss": 0.1343,
      "step": 6803
    },
    {
      "epoch": 4.3531669865642995,
      "grad_norm": 1.3267208625836104,
      "learning_rate": 1.2552311645027802e-05,
      "loss": 0.1469,
      "step": 6804
    },
    {
      "epoch": 4.353806781829815,
      "grad_norm": 1.2767828007086814,
      "learning_rate": 1.255030806587649e-05,
      "loss": 0.142,
      "step": 6805
    },
    {
      "epoch": 4.354446577095329,
      "grad_norm": 1.1871352149306806,
      "learning_rate": 1.2548304377219486e-05,
      "loss": 0.1319,
      "step": 6806
    },
    {
      "epoch": 4.355086372360844,
      "grad_norm": 1.363706992651471,
      "learning_rate": 1.2546300579142823e-05,
      "loss": 0.1523,
      "step": 6807
    },
    {
      "epoch": 4.35572616762636,
      "grad_norm": 1.2925577908349113,
      "learning_rate": 1.2544296671732545e-05,
      "loss": 0.1397,
      "step": 6808
    },
    {
      "epoch": 4.356365962891875,
      "grad_norm": 1.3809637458000097,
      "learning_rate": 1.254229265507469e-05,
      "loss": 0.1484,
      "step": 6809
    },
    {
      "epoch": 4.357005758157389,
      "grad_norm": 1.1855246913876158,
      "learning_rate": 1.2540288529255316e-05,
      "loss": 0.1354,
      "step": 6810
    },
    {
      "epoch": 4.357645553422905,
      "grad_norm": 1.2657649746106012,
      "learning_rate": 1.2538284294360467e-05,
      "loss": 0.1451,
      "step": 6811
    },
    {
      "epoch": 4.35828534868842,
      "grad_norm": 1.2557817144052486,
      "learning_rate": 1.2536279950476209e-05,
      "loss": 0.1348,
      "step": 6812
    },
    {
      "epoch": 4.358925143953935,
      "grad_norm": 1.172168075983113,
      "learning_rate": 1.2534275497688602e-05,
      "loss": 0.121,
      "step": 6813
    },
    {
      "epoch": 4.3595649392194495,
      "grad_norm": 1.2602259539901104,
      "learning_rate": 1.2532270936083708e-05,
      "loss": 0.1369,
      "step": 6814
    },
    {
      "epoch": 4.360204734484965,
      "grad_norm": 1.1730286634039229,
      "learning_rate": 1.253026626574761e-05,
      "loss": 0.1324,
      "step": 6815
    },
    {
      "epoch": 4.36084452975048,
      "grad_norm": 1.250056166919793,
      "learning_rate": 1.2528261486766375e-05,
      "loss": 0.1421,
      "step": 6816
    },
    {
      "epoch": 4.3614843250159945,
      "grad_norm": 1.3535769454160322,
      "learning_rate": 1.2526256599226091e-05,
      "loss": 0.1437,
      "step": 6817
    },
    {
      "epoch": 4.36212412028151,
      "grad_norm": 1.270941268990759,
      "learning_rate": 1.2524251603212845e-05,
      "loss": 0.1351,
      "step": 6818
    },
    {
      "epoch": 4.362763915547025,
      "grad_norm": 1.3224469285084546,
      "learning_rate": 1.2522246498812723e-05,
      "loss": 0.1511,
      "step": 6819
    },
    {
      "epoch": 4.36340371081254,
      "grad_norm": 1.211087473579363,
      "learning_rate": 1.2520241286111822e-05,
      "loss": 0.1334,
      "step": 6820
    },
    {
      "epoch": 4.364043506078055,
      "grad_norm": 1.2606084916386073,
      "learning_rate": 1.2518235965196248e-05,
      "loss": 0.1273,
      "step": 6821
    },
    {
      "epoch": 4.36468330134357,
      "grad_norm": 1.2876863464349972,
      "learning_rate": 1.2516230536152097e-05,
      "loss": 0.1489,
      "step": 6822
    },
    {
      "epoch": 4.365323096609085,
      "grad_norm": 1.4156512982476932,
      "learning_rate": 1.2514224999065485e-05,
      "loss": 0.1656,
      "step": 6823
    },
    {
      "epoch": 4.3659628918746005,
      "grad_norm": 1.2714231843662542,
      "learning_rate": 1.2512219354022522e-05,
      "loss": 0.1434,
      "step": 6824
    },
    {
      "epoch": 4.366602687140115,
      "grad_norm": 1.265597841482503,
      "learning_rate": 1.2510213601109331e-05,
      "loss": 0.1505,
      "step": 6825
    },
    {
      "epoch": 4.36724248240563,
      "grad_norm": 1.1995382910706327,
      "learning_rate": 1.250820774041203e-05,
      "loss": 0.1318,
      "step": 6826
    },
    {
      "epoch": 4.3678822776711455,
      "grad_norm": 1.3317141422351961,
      "learning_rate": 1.2506201772016755e-05,
      "loss": 0.1494,
      "step": 6827
    },
    {
      "epoch": 4.36852207293666,
      "grad_norm": 1.3518134786211333,
      "learning_rate": 1.2504195696009633e-05,
      "loss": 0.1559,
      "step": 6828
    },
    {
      "epoch": 4.369161868202175,
      "grad_norm": 1.2895188984371762,
      "learning_rate": 1.2502189512476799e-05,
      "loss": 0.1541,
      "step": 6829
    },
    {
      "epoch": 4.36980166346769,
      "grad_norm": 1.176665520094814,
      "learning_rate": 1.25001832215044e-05,
      "loss": 0.1322,
      "step": 6830
    },
    {
      "epoch": 4.370441458733206,
      "grad_norm": 1.4247562792122554,
      "learning_rate": 1.2498176823178581e-05,
      "loss": 0.1533,
      "step": 6831
    },
    {
      "epoch": 4.37108125399872,
      "grad_norm": 1.219813639466111,
      "learning_rate": 1.2496170317585493e-05,
      "loss": 0.128,
      "step": 6832
    },
    {
      "epoch": 4.371721049264235,
      "grad_norm": 1.2143352121559452,
      "learning_rate": 1.2494163704811293e-05,
      "loss": 0.1264,
      "step": 6833
    },
    {
      "epoch": 4.372360844529751,
      "grad_norm": 1.3822057178307317,
      "learning_rate": 1.2492156984942139e-05,
      "loss": 0.144,
      "step": 6834
    },
    {
      "epoch": 4.373000639795266,
      "grad_norm": 1.2199646192780988,
      "learning_rate": 1.2490150158064198e-05,
      "loss": 0.1373,
      "step": 6835
    },
    {
      "epoch": 4.37364043506078,
      "grad_norm": 1.4030158268602622,
      "learning_rate": 1.248814322426364e-05,
      "loss": 0.1691,
      "step": 6836
    },
    {
      "epoch": 4.3742802303262955,
      "grad_norm": 1.4653321565264614,
      "learning_rate": 1.2486136183626635e-05,
      "loss": 0.1528,
      "step": 6837
    },
    {
      "epoch": 4.374920025591811,
      "grad_norm": 1.304757922862486,
      "learning_rate": 1.248412903623937e-05,
      "loss": 0.145,
      "step": 6838
    },
    {
      "epoch": 4.375559820857326,
      "grad_norm": 1.2119248879036837,
      "learning_rate": 1.2482121782188018e-05,
      "loss": 0.1451,
      "step": 6839
    },
    {
      "epoch": 4.3761996161228405,
      "grad_norm": 1.2547869190393766,
      "learning_rate": 1.2480114421558775e-05,
      "loss": 0.138,
      "step": 6840
    },
    {
      "epoch": 4.376839411388356,
      "grad_norm": 1.3356712851668635,
      "learning_rate": 1.2478106954437828e-05,
      "loss": 0.151,
      "step": 6841
    },
    {
      "epoch": 4.377479206653871,
      "grad_norm": 1.4072351801251815,
      "learning_rate": 1.247609938091138e-05,
      "loss": 0.1422,
      "step": 6842
    },
    {
      "epoch": 4.378119001919385,
      "grad_norm": 1.2166908196813555,
      "learning_rate": 1.2474091701065626e-05,
      "loss": 0.141,
      "step": 6843
    },
    {
      "epoch": 4.378758797184901,
      "grad_norm": 1.1918735093255415,
      "learning_rate": 1.2472083914986776e-05,
      "loss": 0.1317,
      "step": 6844
    },
    {
      "epoch": 4.379398592450416,
      "grad_norm": 1.2019478111754398,
      "learning_rate": 1.2470076022761041e-05,
      "loss": 0.1407,
      "step": 6845
    },
    {
      "epoch": 4.380038387715931,
      "grad_norm": 1.415719412097287,
      "learning_rate": 1.2468068024474635e-05,
      "loss": 0.1575,
      "step": 6846
    },
    {
      "epoch": 4.380678182981446,
      "grad_norm": 1.4595285801370548,
      "learning_rate": 1.2466059920213777e-05,
      "loss": 0.1651,
      "step": 6847
    },
    {
      "epoch": 4.381317978246961,
      "grad_norm": 1.4264715214745827,
      "learning_rate": 1.2464051710064695e-05,
      "loss": 0.1463,
      "step": 6848
    },
    {
      "epoch": 4.381957773512476,
      "grad_norm": 1.473329333078312,
      "learning_rate": 1.2462043394113615e-05,
      "loss": 0.1502,
      "step": 6849
    },
    {
      "epoch": 4.3825975687779914,
      "grad_norm": 1.2275863609954776,
      "learning_rate": 1.2460034972446772e-05,
      "loss": 0.1267,
      "step": 6850
    },
    {
      "epoch": 4.383237364043506,
      "grad_norm": 1.2201736647687857,
      "learning_rate": 1.2458026445150404e-05,
      "loss": 0.1448,
      "step": 6851
    },
    {
      "epoch": 4.383877159309021,
      "grad_norm": 1.3336018440086495,
      "learning_rate": 1.245601781231075e-05,
      "loss": 0.1434,
      "step": 6852
    },
    {
      "epoch": 4.384516954574536,
      "grad_norm": 1.3201098949398589,
      "learning_rate": 1.2454009074014063e-05,
      "loss": 0.1372,
      "step": 6853
    },
    {
      "epoch": 4.385156749840051,
      "grad_norm": 1.22752060030486,
      "learning_rate": 1.245200023034659e-05,
      "loss": 0.1314,
      "step": 6854
    },
    {
      "epoch": 4.385796545105566,
      "grad_norm": 1.2432685927334695,
      "learning_rate": 1.244999128139459e-05,
      "loss": 0.1349,
      "step": 6855
    },
    {
      "epoch": 4.386436340371081,
      "grad_norm": 1.2686505863779607,
      "learning_rate": 1.2447982227244323e-05,
      "loss": 0.1457,
      "step": 6856
    },
    {
      "epoch": 4.387076135636597,
      "grad_norm": 1.2571753390959335,
      "learning_rate": 1.2445973067982051e-05,
      "loss": 0.1413,
      "step": 6857
    },
    {
      "epoch": 4.387715930902111,
      "grad_norm": 1.0807601644976597,
      "learning_rate": 1.2443963803694051e-05,
      "loss": 0.129,
      "step": 6858
    },
    {
      "epoch": 4.388355726167626,
      "grad_norm": 1.1988290533363588,
      "learning_rate": 1.2441954434466588e-05,
      "loss": 0.1319,
      "step": 6859
    },
    {
      "epoch": 4.3889955214331415,
      "grad_norm": 1.2543356108378967,
      "learning_rate": 1.2439944960385949e-05,
      "loss": 0.1331,
      "step": 6860
    },
    {
      "epoch": 4.389635316698657,
      "grad_norm": 1.3241855369191118,
      "learning_rate": 1.2437935381538411e-05,
      "loss": 0.1495,
      "step": 6861
    },
    {
      "epoch": 4.390275111964171,
      "grad_norm": 1.2636404536780197,
      "learning_rate": 1.2435925698010266e-05,
      "loss": 0.1427,
      "step": 6862
    },
    {
      "epoch": 4.3909149072296865,
      "grad_norm": 1.3567066384102069,
      "learning_rate": 1.2433915909887804e-05,
      "loss": 0.1533,
      "step": 6863
    },
    {
      "epoch": 4.391554702495202,
      "grad_norm": 1.1531708273262744,
      "learning_rate": 1.243190601725732e-05,
      "loss": 0.128,
      "step": 6864
    },
    {
      "epoch": 4.392194497760716,
      "grad_norm": 1.2863431523671067,
      "learning_rate": 1.2429896020205124e-05,
      "loss": 0.1384,
      "step": 6865
    },
    {
      "epoch": 4.392834293026231,
      "grad_norm": 1.262338895392917,
      "learning_rate": 1.2427885918817511e-05,
      "loss": 0.1352,
      "step": 6866
    },
    {
      "epoch": 4.393474088291747,
      "grad_norm": 1.1863005955764767,
      "learning_rate": 1.2425875713180795e-05,
      "loss": 0.1336,
      "step": 6867
    },
    {
      "epoch": 4.394113883557262,
      "grad_norm": 1.2622453927778032,
      "learning_rate": 1.2423865403381295e-05,
      "loss": 0.1507,
      "step": 6868
    },
    {
      "epoch": 4.394753678822776,
      "grad_norm": 1.198348877695643,
      "learning_rate": 1.2421854989505322e-05,
      "loss": 0.1348,
      "step": 6869
    },
    {
      "epoch": 4.395393474088292,
      "grad_norm": 1.3372096159248028,
      "learning_rate": 1.2419844471639207e-05,
      "loss": 0.1447,
      "step": 6870
    },
    {
      "epoch": 4.396033269353807,
      "grad_norm": 1.30777745665728,
      "learning_rate": 1.2417833849869271e-05,
      "loss": 0.1397,
      "step": 6871
    },
    {
      "epoch": 4.396673064619322,
      "grad_norm": 1.248716513512587,
      "learning_rate": 1.2415823124281855e-05,
      "loss": 0.1331,
      "step": 6872
    },
    {
      "epoch": 4.3973128598848366,
      "grad_norm": 1.2299722365720949,
      "learning_rate": 1.241381229496329e-05,
      "loss": 0.1372,
      "step": 6873
    },
    {
      "epoch": 4.397952655150352,
      "grad_norm": 1.3160606402788584,
      "learning_rate": 1.2411801361999918e-05,
      "loss": 0.1443,
      "step": 6874
    },
    {
      "epoch": 4.398592450415867,
      "grad_norm": 1.316612038088884,
      "learning_rate": 1.2409790325478085e-05,
      "loss": 0.1419,
      "step": 6875
    },
    {
      "epoch": 4.399232245681382,
      "grad_norm": 1.2065215939784995,
      "learning_rate": 1.2407779185484143e-05,
      "loss": 0.1346,
      "step": 6876
    },
    {
      "epoch": 4.399872040946897,
      "grad_norm": 1.303722968758287,
      "learning_rate": 1.2405767942104447e-05,
      "loss": 0.1451,
      "step": 6877
    },
    {
      "epoch": 4.400511836212412,
      "grad_norm": 1.3286424229045954,
      "learning_rate": 1.2403756595425354e-05,
      "loss": 0.1461,
      "step": 6878
    },
    {
      "epoch": 4.401151631477927,
      "grad_norm": 1.2490547566841088,
      "learning_rate": 1.2401745145533228e-05,
      "loss": 0.1281,
      "step": 6879
    },
    {
      "epoch": 4.401791426743442,
      "grad_norm": 1.2371615322661385,
      "learning_rate": 1.239973359251444e-05,
      "loss": 0.1256,
      "step": 6880
    },
    {
      "epoch": 4.402431222008957,
      "grad_norm": 1.2776472029063792,
      "learning_rate": 1.239772193645536e-05,
      "loss": 0.1428,
      "step": 6881
    },
    {
      "epoch": 4.403071017274472,
      "grad_norm": 1.3481277098930011,
      "learning_rate": 1.2395710177442365e-05,
      "loss": 0.1406,
      "step": 6882
    },
    {
      "epoch": 4.4037108125399875,
      "grad_norm": 1.2414535884743214,
      "learning_rate": 1.2393698315561836e-05,
      "loss": 0.1407,
      "step": 6883
    },
    {
      "epoch": 4.404350607805502,
      "grad_norm": 1.2673784933330665,
      "learning_rate": 1.2391686350900158e-05,
      "loss": 0.1359,
      "step": 6884
    },
    {
      "epoch": 4.404990403071017,
      "grad_norm": 1.2522671330478534,
      "learning_rate": 1.2389674283543725e-05,
      "loss": 0.1346,
      "step": 6885
    },
    {
      "epoch": 4.4056301983365325,
      "grad_norm": 1.3175217643892596,
      "learning_rate": 1.2387662113578927e-05,
      "loss": 0.1382,
      "step": 6886
    },
    {
      "epoch": 4.406269993602048,
      "grad_norm": 1.3888951940096137,
      "learning_rate": 1.238564984109217e-05,
      "loss": 0.1583,
      "step": 6887
    },
    {
      "epoch": 4.406909788867562,
      "grad_norm": 1.2882200257890317,
      "learning_rate": 1.2383637466169848e-05,
      "loss": 0.128,
      "step": 6888
    },
    {
      "epoch": 4.407549584133077,
      "grad_norm": 1.2872546725922505,
      "learning_rate": 1.2381624988898373e-05,
      "loss": 0.1458,
      "step": 6889
    },
    {
      "epoch": 4.408189379398593,
      "grad_norm": 1.2276362827104323,
      "learning_rate": 1.2379612409364161e-05,
      "loss": 0.1383,
      "step": 6890
    },
    {
      "epoch": 4.408829174664108,
      "grad_norm": 1.3069634873761375,
      "learning_rate": 1.2377599727653622e-05,
      "loss": 0.1434,
      "step": 6891
    },
    {
      "epoch": 4.409468969929622,
      "grad_norm": 1.2467377112198916,
      "learning_rate": 1.2375586943853181e-05,
      "loss": 0.1297,
      "step": 6892
    },
    {
      "epoch": 4.410108765195138,
      "grad_norm": 1.3789978441431816,
      "learning_rate": 1.237357405804926e-05,
      "loss": 0.1576,
      "step": 6893
    },
    {
      "epoch": 4.410748560460653,
      "grad_norm": 1.4131954514413956,
      "learning_rate": 1.2371561070328295e-05,
      "loss": 0.1502,
      "step": 6894
    },
    {
      "epoch": 4.411388355726167,
      "grad_norm": 1.2879032171420863,
      "learning_rate": 1.2369547980776715e-05,
      "loss": 0.1474,
      "step": 6895
    },
    {
      "epoch": 4.4120281509916826,
      "grad_norm": 1.3494477493924908,
      "learning_rate": 1.2367534789480961e-05,
      "loss": 0.1497,
      "step": 6896
    },
    {
      "epoch": 4.412667946257198,
      "grad_norm": 1.3268969824747878,
      "learning_rate": 1.2365521496527476e-05,
      "loss": 0.1451,
      "step": 6897
    },
    {
      "epoch": 4.413307741522713,
      "grad_norm": 1.397078237520989,
      "learning_rate": 1.2363508102002702e-05,
      "loss": 0.1497,
      "step": 6898
    },
    {
      "epoch": 4.4139475367882275,
      "grad_norm": 1.2843185779754576,
      "learning_rate": 1.2361494605993094e-05,
      "loss": 0.1398,
      "step": 6899
    },
    {
      "epoch": 4.414587332053743,
      "grad_norm": 1.3754485089341237,
      "learning_rate": 1.235948100858511e-05,
      "loss": 0.1467,
      "step": 6900
    },
    {
      "epoch": 4.415227127319258,
      "grad_norm": 1.3361142692199628,
      "learning_rate": 1.235746730986521e-05,
      "loss": 0.145,
      "step": 6901
    },
    {
      "epoch": 4.415866922584772,
      "grad_norm": 1.4503904919175645,
      "learning_rate": 1.2355453509919857e-05,
      "loss": 0.1507,
      "step": 6902
    },
    {
      "epoch": 4.416506717850288,
      "grad_norm": 1.2878464823636095,
      "learning_rate": 1.235343960883552e-05,
      "loss": 0.147,
      "step": 6903
    },
    {
      "epoch": 4.417146513115803,
      "grad_norm": 1.1992065214513228,
      "learning_rate": 1.2351425606698673e-05,
      "loss": 0.1327,
      "step": 6904
    },
    {
      "epoch": 4.417786308381318,
      "grad_norm": 1.214825455044547,
      "learning_rate": 1.2349411503595793e-05,
      "loss": 0.1398,
      "step": 6905
    },
    {
      "epoch": 4.418426103646833,
      "grad_norm": 1.1985204732626042,
      "learning_rate": 1.234739729961336e-05,
      "loss": 0.1342,
      "step": 6906
    },
    {
      "epoch": 4.419065898912348,
      "grad_norm": 1.2080097814344564,
      "learning_rate": 1.2345382994837865e-05,
      "loss": 0.1295,
      "step": 6907
    },
    {
      "epoch": 4.419705694177863,
      "grad_norm": 1.381628224504263,
      "learning_rate": 1.2343368589355796e-05,
      "loss": 0.1514,
      "step": 6908
    },
    {
      "epoch": 4.4203454894433785,
      "grad_norm": 1.3548644748666878,
      "learning_rate": 1.2341354083253648e-05,
      "loss": 0.1531,
      "step": 6909
    },
    {
      "epoch": 4.420985284708893,
      "grad_norm": 1.2551186835160668,
      "learning_rate": 1.233933947661792e-05,
      "loss": 0.1356,
      "step": 6910
    },
    {
      "epoch": 4.421625079974408,
      "grad_norm": 1.3353880194255334,
      "learning_rate": 1.2337324769535116e-05,
      "loss": 0.1466,
      "step": 6911
    },
    {
      "epoch": 4.422264875239923,
      "grad_norm": 1.1459241251918613,
      "learning_rate": 1.2335309962091748e-05,
      "loss": 0.1341,
      "step": 6912
    },
    {
      "epoch": 4.422904670505439,
      "grad_norm": 1.1898218871179322,
      "learning_rate": 1.2333295054374322e-05,
      "loss": 0.1276,
      "step": 6913
    },
    {
      "epoch": 4.423544465770953,
      "grad_norm": 1.2446062016875221,
      "learning_rate": 1.2331280046469354e-05,
      "loss": 0.141,
      "step": 6914
    },
    {
      "epoch": 4.424184261036468,
      "grad_norm": 1.255723414528174,
      "learning_rate": 1.2329264938463371e-05,
      "loss": 0.1569,
      "step": 6915
    },
    {
      "epoch": 4.424824056301984,
      "grad_norm": 1.3371977835090032,
      "learning_rate": 1.2327249730442894e-05,
      "loss": 0.1434,
      "step": 6916
    },
    {
      "epoch": 4.425463851567498,
      "grad_norm": 1.3227641680503088,
      "learning_rate": 1.2325234422494454e-05,
      "loss": 0.1441,
      "step": 6917
    },
    {
      "epoch": 4.426103646833013,
      "grad_norm": 1.3228834714756752,
      "learning_rate": 1.2323219014704586e-05,
      "loss": 0.1459,
      "step": 6918
    },
    {
      "epoch": 4.4267434420985285,
      "grad_norm": 1.3105203862742074,
      "learning_rate": 1.2321203507159822e-05,
      "loss": 0.1403,
      "step": 6919
    },
    {
      "epoch": 4.427383237364044,
      "grad_norm": 1.2599315754248936,
      "learning_rate": 1.2319187899946713e-05,
      "loss": 0.1403,
      "step": 6920
    },
    {
      "epoch": 4.428023032629558,
      "grad_norm": 1.3163599665860735,
      "learning_rate": 1.2317172193151796e-05,
      "loss": 0.1491,
      "step": 6921
    },
    {
      "epoch": 4.4286628278950735,
      "grad_norm": 1.3780830079735045,
      "learning_rate": 1.2315156386861633e-05,
      "loss": 0.1389,
      "step": 6922
    },
    {
      "epoch": 4.429302623160589,
      "grad_norm": 1.3484019462463122,
      "learning_rate": 1.2313140481162768e-05,
      "loss": 0.1501,
      "step": 6923
    },
    {
      "epoch": 4.429942418426104,
      "grad_norm": 1.3867639607065108,
      "learning_rate": 1.2311124476141767e-05,
      "loss": 0.1548,
      "step": 6924
    },
    {
      "epoch": 4.430582213691618,
      "grad_norm": 1.2923703176014476,
      "learning_rate": 1.2309108371885193e-05,
      "loss": 0.1424,
      "step": 6925
    },
    {
      "epoch": 4.431222008957134,
      "grad_norm": 1.128229225498296,
      "learning_rate": 1.2307092168479613e-05,
      "loss": 0.1324,
      "step": 6926
    },
    {
      "epoch": 4.431861804222649,
      "grad_norm": 1.3019534091427964,
      "learning_rate": 1.23050758660116e-05,
      "loss": 0.1451,
      "step": 6927
    },
    {
      "epoch": 4.432501599488164,
      "grad_norm": 1.2017969859220035,
      "learning_rate": 1.2303059464567728e-05,
      "loss": 0.1306,
      "step": 6928
    },
    {
      "epoch": 4.433141394753679,
      "grad_norm": 1.1747468542853998,
      "learning_rate": 1.2301042964234578e-05,
      "loss": 0.1302,
      "step": 6929
    },
    {
      "epoch": 4.433781190019194,
      "grad_norm": 1.190523134440021,
      "learning_rate": 1.229902636509874e-05,
      "loss": 0.1459,
      "step": 6930
    },
    {
      "epoch": 4.434420985284709,
      "grad_norm": 1.3833623378338458,
      "learning_rate": 1.2297009667246798e-05,
      "loss": 0.1514,
      "step": 6931
    },
    {
      "epoch": 4.435060780550224,
      "grad_norm": 1.2524000481802418,
      "learning_rate": 1.2294992870765348e-05,
      "loss": 0.1536,
      "step": 6932
    },
    {
      "epoch": 4.435700575815739,
      "grad_norm": 1.3398316815369011,
      "learning_rate": 1.2292975975740987e-05,
      "loss": 0.1451,
      "step": 6933
    },
    {
      "epoch": 4.436340371081254,
      "grad_norm": 1.296782814412944,
      "learning_rate": 1.2290958982260312e-05,
      "loss": 0.142,
      "step": 6934
    },
    {
      "epoch": 4.436980166346769,
      "grad_norm": 1.2505345863881487,
      "learning_rate": 1.2288941890409937e-05,
      "loss": 0.1369,
      "step": 6935
    },
    {
      "epoch": 4.437619961612284,
      "grad_norm": 1.4030801661577403,
      "learning_rate": 1.2286924700276468e-05,
      "loss": 0.1455,
      "step": 6936
    },
    {
      "epoch": 4.438259756877799,
      "grad_norm": 1.3203863196068377,
      "learning_rate": 1.2284907411946523e-05,
      "loss": 0.147,
      "step": 6937
    },
    {
      "epoch": 4.438899552143314,
      "grad_norm": 1.3479481098372936,
      "learning_rate": 1.2282890025506715e-05,
      "loss": 0.1434,
      "step": 6938
    },
    {
      "epoch": 4.439539347408829,
      "grad_norm": 1.331411555587003,
      "learning_rate": 1.2280872541043669e-05,
      "loss": 0.1484,
      "step": 6939
    },
    {
      "epoch": 4.440179142674344,
      "grad_norm": 1.4136329124997797,
      "learning_rate": 1.2278854958644018e-05,
      "loss": 0.1608,
      "step": 6940
    },
    {
      "epoch": 4.440818937939859,
      "grad_norm": 1.2512062501803178,
      "learning_rate": 1.2276837278394387e-05,
      "loss": 0.1383,
      "step": 6941
    },
    {
      "epoch": 4.4414587332053745,
      "grad_norm": 1.2523689332823609,
      "learning_rate": 1.2274819500381414e-05,
      "loss": 0.1393,
      "step": 6942
    },
    {
      "epoch": 4.442098528470889,
      "grad_norm": 1.4100335756775144,
      "learning_rate": 1.2272801624691737e-05,
      "loss": 0.1502,
      "step": 6943
    },
    {
      "epoch": 4.442738323736404,
      "grad_norm": 1.2414387250645154,
      "learning_rate": 1.2270783651412002e-05,
      "loss": 0.14,
      "step": 6944
    },
    {
      "epoch": 4.4433781190019195,
      "grad_norm": 1.2482371385647542,
      "learning_rate": 1.2268765580628857e-05,
      "loss": 0.1399,
      "step": 6945
    },
    {
      "epoch": 4.444017914267435,
      "grad_norm": 1.2720796567007109,
      "learning_rate": 1.226674741242895e-05,
      "loss": 0.1465,
      "step": 6946
    },
    {
      "epoch": 4.444657709532949,
      "grad_norm": 1.103195346061251,
      "learning_rate": 1.2264729146898946e-05,
      "loss": 0.1296,
      "step": 6947
    },
    {
      "epoch": 4.445297504798464,
      "grad_norm": 1.2330278197784537,
      "learning_rate": 1.22627107841255e-05,
      "loss": 0.1433,
      "step": 6948
    },
    {
      "epoch": 4.44593730006398,
      "grad_norm": 1.1787146730098899,
      "learning_rate": 1.2260692324195277e-05,
      "loss": 0.1441,
      "step": 6949
    },
    {
      "epoch": 4.446577095329495,
      "grad_norm": 1.3795581192035329,
      "learning_rate": 1.2258673767194948e-05,
      "loss": 0.1452,
      "step": 6950
    },
    {
      "epoch": 4.447216890595009,
      "grad_norm": 1.2232223294949196,
      "learning_rate": 1.2256655113211182e-05,
      "loss": 0.1499,
      "step": 6951
    },
    {
      "epoch": 4.447856685860525,
      "grad_norm": 1.1676370872933821,
      "learning_rate": 1.2254636362330665e-05,
      "loss": 0.1337,
      "step": 6952
    },
    {
      "epoch": 4.44849648112604,
      "grad_norm": 1.1661111519601552,
      "learning_rate": 1.2252617514640068e-05,
      "loss": 0.1251,
      "step": 6953
    },
    {
      "epoch": 4.449136276391554,
      "grad_norm": 1.318735562917152,
      "learning_rate": 1.2250598570226087e-05,
      "loss": 0.1487,
      "step": 6954
    },
    {
      "epoch": 4.44977607165707,
      "grad_norm": 1.3779838539501035,
      "learning_rate": 1.2248579529175403e-05,
      "loss": 0.1588,
      "step": 6955
    },
    {
      "epoch": 4.450415866922585,
      "grad_norm": 1.3932176912815761,
      "learning_rate": 1.2246560391574718e-05,
      "loss": 0.1624,
      "step": 6956
    },
    {
      "epoch": 4.4510556621881,
      "grad_norm": 1.2795196670259292,
      "learning_rate": 1.2244541157510723e-05,
      "loss": 0.1367,
      "step": 6957
    },
    {
      "epoch": 4.4516954574536145,
      "grad_norm": 1.3386304998306406,
      "learning_rate": 1.2242521827070125e-05,
      "loss": 0.1598,
      "step": 6958
    },
    {
      "epoch": 4.45233525271913,
      "grad_norm": 1.320168176299263,
      "learning_rate": 1.2240502400339627e-05,
      "loss": 0.1416,
      "step": 6959
    },
    {
      "epoch": 4.452975047984645,
      "grad_norm": 1.2508685752236801,
      "learning_rate": 1.2238482877405942e-05,
      "loss": 0.1333,
      "step": 6960
    },
    {
      "epoch": 4.45361484325016,
      "grad_norm": 1.246897940363626,
      "learning_rate": 1.2236463258355783e-05,
      "loss": 0.1442,
      "step": 6961
    },
    {
      "epoch": 4.454254638515675,
      "grad_norm": 1.1790831284860783,
      "learning_rate": 1.2234443543275874e-05,
      "loss": 0.1227,
      "step": 6962
    },
    {
      "epoch": 4.45489443378119,
      "grad_norm": 1.1047808482369157,
      "learning_rate": 1.2232423732252932e-05,
      "loss": 0.1312,
      "step": 6963
    },
    {
      "epoch": 4.455534229046705,
      "grad_norm": 1.2844986982786233,
      "learning_rate": 1.2230403825373687e-05,
      "loss": 0.1405,
      "step": 6964
    },
    {
      "epoch": 4.4561740243122205,
      "grad_norm": 1.3419150712479533,
      "learning_rate": 1.2228383822724867e-05,
      "loss": 0.1531,
      "step": 6965
    },
    {
      "epoch": 4.456813819577735,
      "grad_norm": 1.2760777977769198,
      "learning_rate": 1.222636372439321e-05,
      "loss": 0.143,
      "step": 6966
    },
    {
      "epoch": 4.45745361484325,
      "grad_norm": 1.2809304035832856,
      "learning_rate": 1.2224343530465457e-05,
      "loss": 0.1472,
      "step": 6967
    },
    {
      "epoch": 4.4580934101087655,
      "grad_norm": 1.358756261620577,
      "learning_rate": 1.2222323241028346e-05,
      "loss": 0.1438,
      "step": 6968
    },
    {
      "epoch": 4.45873320537428,
      "grad_norm": 1.321592849417866,
      "learning_rate": 1.2220302856168634e-05,
      "loss": 0.147,
      "step": 6969
    },
    {
      "epoch": 4.459373000639795,
      "grad_norm": 1.2940391410532652,
      "learning_rate": 1.221828237597306e-05,
      "loss": 0.1474,
      "step": 6970
    },
    {
      "epoch": 4.46001279590531,
      "grad_norm": 1.3017900399034483,
      "learning_rate": 1.2216261800528392e-05,
      "loss": 0.1502,
      "step": 6971
    },
    {
      "epoch": 4.460652591170826,
      "grad_norm": 1.2493480862565374,
      "learning_rate": 1.2214241129921384e-05,
      "loss": 0.136,
      "step": 6972
    },
    {
      "epoch": 4.46129238643634,
      "grad_norm": 1.292674625089318,
      "learning_rate": 1.2212220364238798e-05,
      "loss": 0.1395,
      "step": 6973
    },
    {
      "epoch": 4.461932181701855,
      "grad_norm": 1.210262074879732,
      "learning_rate": 1.2210199503567409e-05,
      "loss": 0.1439,
      "step": 6974
    },
    {
      "epoch": 4.462571976967371,
      "grad_norm": 1.28719412456827,
      "learning_rate": 1.220817854799398e-05,
      "loss": 0.1471,
      "step": 6975
    },
    {
      "epoch": 4.463211772232885,
      "grad_norm": 1.342456972399331,
      "learning_rate": 1.2206157497605296e-05,
      "loss": 0.1493,
      "step": 6976
    },
    {
      "epoch": 4.4638515674984,
      "grad_norm": 1.2616570182672038,
      "learning_rate": 1.2204136352488132e-05,
      "loss": 0.1336,
      "step": 6977
    },
    {
      "epoch": 4.464491362763916,
      "grad_norm": 1.3700564782339617,
      "learning_rate": 1.2202115112729276e-05,
      "loss": 0.1513,
      "step": 6978
    },
    {
      "epoch": 4.465131158029431,
      "grad_norm": 1.1942011405422308,
      "learning_rate": 1.2200093778415512e-05,
      "loss": 0.1406,
      "step": 6979
    },
    {
      "epoch": 4.465770953294945,
      "grad_norm": 1.1984202619620745,
      "learning_rate": 1.2198072349633635e-05,
      "loss": 0.1329,
      "step": 6980
    },
    {
      "epoch": 4.4664107485604605,
      "grad_norm": 1.1805584570295995,
      "learning_rate": 1.219605082647044e-05,
      "loss": 0.1369,
      "step": 6981
    },
    {
      "epoch": 4.467050543825976,
      "grad_norm": 1.2973465569344311,
      "learning_rate": 1.2194029209012732e-05,
      "loss": 0.1374,
      "step": 6982
    },
    {
      "epoch": 4.467690339091491,
      "grad_norm": 1.1604192837368568,
      "learning_rate": 1.2192007497347311e-05,
      "loss": 0.1317,
      "step": 6983
    },
    {
      "epoch": 4.468330134357005,
      "grad_norm": 1.2524193760272133,
      "learning_rate": 1.2189985691560987e-05,
      "loss": 0.1512,
      "step": 6984
    },
    {
      "epoch": 4.468969929622521,
      "grad_norm": 1.2598786502466646,
      "learning_rate": 1.2187963791740574e-05,
      "loss": 0.1409,
      "step": 6985
    },
    {
      "epoch": 4.469609724888036,
      "grad_norm": 1.2720441279139443,
      "learning_rate": 1.2185941797972889e-05,
      "loss": 0.1448,
      "step": 6986
    },
    {
      "epoch": 4.470249520153551,
      "grad_norm": 1.2938781283861807,
      "learning_rate": 1.2183919710344751e-05,
      "loss": 0.1521,
      "step": 6987
    },
    {
      "epoch": 4.470889315419066,
      "grad_norm": 1.2545912154690668,
      "learning_rate": 1.2181897528942983e-05,
      "loss": 0.147,
      "step": 6988
    },
    {
      "epoch": 4.471529110684581,
      "grad_norm": 1.207356556036573,
      "learning_rate": 1.2179875253854419e-05,
      "loss": 0.1269,
      "step": 6989
    },
    {
      "epoch": 4.472168905950096,
      "grad_norm": 1.2631886175461478,
      "learning_rate": 1.2177852885165889e-05,
      "loss": 0.1497,
      "step": 6990
    },
    {
      "epoch": 4.472808701215611,
      "grad_norm": 1.2703061376919436,
      "learning_rate": 1.2175830422964228e-05,
      "loss": 0.1348,
      "step": 6991
    },
    {
      "epoch": 4.473448496481126,
      "grad_norm": 1.1164108962729644,
      "learning_rate": 1.217380786733628e-05,
      "loss": 0.1233,
      "step": 6992
    },
    {
      "epoch": 4.474088291746641,
      "grad_norm": 1.2892137975383118,
      "learning_rate": 1.2171785218368892e-05,
      "loss": 0.14,
      "step": 6993
    },
    {
      "epoch": 4.474728087012156,
      "grad_norm": 1.1404446861383137,
      "learning_rate": 1.216976247614891e-05,
      "loss": 0.1236,
      "step": 6994
    },
    {
      "epoch": 4.475367882277671,
      "grad_norm": 1.2750775050491472,
      "learning_rate": 1.2167739640763183e-05,
      "loss": 0.1373,
      "step": 6995
    },
    {
      "epoch": 4.476007677543186,
      "grad_norm": 1.2771526018690829,
      "learning_rate": 1.2165716712298572e-05,
      "loss": 0.1491,
      "step": 6996
    },
    {
      "epoch": 4.476647472808701,
      "grad_norm": 1.2419432910346624,
      "learning_rate": 1.2163693690841943e-05,
      "loss": 0.1328,
      "step": 6997
    },
    {
      "epoch": 4.477287268074217,
      "grad_norm": 1.4026958971475452,
      "learning_rate": 1.2161670576480148e-05,
      "loss": 0.15,
      "step": 6998
    },
    {
      "epoch": 4.477927063339731,
      "grad_norm": 1.2247945549130266,
      "learning_rate": 1.2159647369300071e-05,
      "loss": 0.1355,
      "step": 6999
    },
    {
      "epoch": 4.478566858605246,
      "grad_norm": 1.1993020912948695,
      "learning_rate": 1.2157624069388574e-05,
      "loss": 0.1339,
      "step": 7000
    },
    {
      "epoch": 4.479206653870762,
      "grad_norm": 1.2612780521166367,
      "learning_rate": 1.215560067683254e-05,
      "loss": 0.1451,
      "step": 7001
    },
    {
      "epoch": 4.479846449136277,
      "grad_norm": 1.356255206800079,
      "learning_rate": 1.2153577191718849e-05,
      "loss": 0.1611,
      "step": 7002
    },
    {
      "epoch": 4.480486244401791,
      "grad_norm": 1.1288099706118353,
      "learning_rate": 1.215155361413438e-05,
      "loss": 0.1257,
      "step": 7003
    },
    {
      "epoch": 4.4811260396673065,
      "grad_norm": 1.2782844864706888,
      "learning_rate": 1.2149529944166027e-05,
      "loss": 0.1397,
      "step": 7004
    },
    {
      "epoch": 4.481765834932822,
      "grad_norm": 1.2622769944013374,
      "learning_rate": 1.2147506181900681e-05,
      "loss": 0.148,
      "step": 7005
    },
    {
      "epoch": 4.482405630198336,
      "grad_norm": 1.2021244082729416,
      "learning_rate": 1.214548232742524e-05,
      "loss": 0.1455,
      "step": 7006
    },
    {
      "epoch": 4.483045425463851,
      "grad_norm": 1.2261423642972273,
      "learning_rate": 1.2143458380826608e-05,
      "loss": 0.135,
      "step": 7007
    },
    {
      "epoch": 4.483685220729367,
      "grad_norm": 1.1601159557512906,
      "learning_rate": 1.2141434342191683e-05,
      "loss": 0.1368,
      "step": 7008
    },
    {
      "epoch": 4.484325015994882,
      "grad_norm": 1.1760765682938994,
      "learning_rate": 1.2139410211607379e-05,
      "loss": 0.1314,
      "step": 7009
    },
    {
      "epoch": 4.484964811260396,
      "grad_norm": 1.2041281046730192,
      "learning_rate": 1.2137385989160602e-05,
      "loss": 0.1435,
      "step": 7010
    },
    {
      "epoch": 4.485604606525912,
      "grad_norm": 1.302344679701319,
      "learning_rate": 1.2135361674938276e-05,
      "loss": 0.1535,
      "step": 7011
    },
    {
      "epoch": 4.486244401791427,
      "grad_norm": 1.2255851235141793,
      "learning_rate": 1.2133337269027316e-05,
      "loss": 0.1306,
      "step": 7012
    },
    {
      "epoch": 4.486884197056942,
      "grad_norm": 1.3324934389736172,
      "learning_rate": 1.2131312771514649e-05,
      "loss": 0.1452,
      "step": 7013
    },
    {
      "epoch": 4.487523992322457,
      "grad_norm": 1.2152896056520888,
      "learning_rate": 1.2129288182487207e-05,
      "loss": 0.1312,
      "step": 7014
    },
    {
      "epoch": 4.488163787587972,
      "grad_norm": 1.3578261385861157,
      "learning_rate": 1.2127263502031912e-05,
      "loss": 0.1322,
      "step": 7015
    },
    {
      "epoch": 4.488803582853487,
      "grad_norm": 1.4545184623535572,
      "learning_rate": 1.212523873023571e-05,
      "loss": 0.1501,
      "step": 7016
    },
    {
      "epoch": 4.4894433781190015,
      "grad_norm": 1.347256438130226,
      "learning_rate": 1.212321386718554e-05,
      "loss": 0.1388,
      "step": 7017
    },
    {
      "epoch": 4.490083173384517,
      "grad_norm": 1.3661103605739295,
      "learning_rate": 1.2121188912968339e-05,
      "loss": 0.1525,
      "step": 7018
    },
    {
      "epoch": 4.490722968650032,
      "grad_norm": 1.2529749532671284,
      "learning_rate": 1.211916386767106e-05,
      "loss": 0.1389,
      "step": 7019
    },
    {
      "epoch": 4.491362763915547,
      "grad_norm": 1.3699018110263799,
      "learning_rate": 1.2117138731380653e-05,
      "loss": 0.1518,
      "step": 7020
    },
    {
      "epoch": 4.492002559181062,
      "grad_norm": 1.269823513170829,
      "learning_rate": 1.2115113504184077e-05,
      "loss": 0.1441,
      "step": 7021
    },
    {
      "epoch": 4.492642354446577,
      "grad_norm": 1.265559537781224,
      "learning_rate": 1.2113088186168289e-05,
      "loss": 0.1352,
      "step": 7022
    },
    {
      "epoch": 4.493282149712092,
      "grad_norm": 1.3012197100884293,
      "learning_rate": 1.2111062777420255e-05,
      "loss": 0.1508,
      "step": 7023
    },
    {
      "epoch": 4.4939219449776076,
      "grad_norm": 1.2641254443331682,
      "learning_rate": 1.210903727802694e-05,
      "loss": 0.1361,
      "step": 7024
    },
    {
      "epoch": 4.494561740243122,
      "grad_norm": 1.214390296356658,
      "learning_rate": 1.2107011688075313e-05,
      "loss": 0.1346,
      "step": 7025
    },
    {
      "epoch": 4.495201535508637,
      "grad_norm": 1.2978795459769283,
      "learning_rate": 1.2104986007652355e-05,
      "loss": 0.1382,
      "step": 7026
    },
    {
      "epoch": 4.4958413307741525,
      "grad_norm": 1.2573065244072634,
      "learning_rate": 1.2102960236845043e-05,
      "loss": 0.1342,
      "step": 7027
    },
    {
      "epoch": 4.496481126039667,
      "grad_norm": 1.2051598042675942,
      "learning_rate": 1.2100934375740357e-05,
      "loss": 0.1263,
      "step": 7028
    },
    {
      "epoch": 4.497120921305182,
      "grad_norm": 1.313770553987178,
      "learning_rate": 1.2098908424425288e-05,
      "loss": 0.1302,
      "step": 7029
    },
    {
      "epoch": 4.497760716570697,
      "grad_norm": 1.371420931878497,
      "learning_rate": 1.2096882382986823e-05,
      "loss": 0.1362,
      "step": 7030
    },
    {
      "epoch": 4.498400511836213,
      "grad_norm": 1.3692240443491066,
      "learning_rate": 1.209485625151196e-05,
      "loss": 0.1455,
      "step": 7031
    },
    {
      "epoch": 4.499040307101727,
      "grad_norm": 1.2906994136366103,
      "learning_rate": 1.2092830030087697e-05,
      "loss": 0.1316,
      "step": 7032
    },
    {
      "epoch": 4.499680102367242,
      "grad_norm": 1.2985824297649093,
      "learning_rate": 1.2090803718801032e-05,
      "loss": 0.1478,
      "step": 7033
    },
    {
      "epoch": 4.500319897632758,
      "grad_norm": 1.2154903817261205,
      "learning_rate": 1.2088777317738978e-05,
      "loss": 0.14,
      "step": 7034
    },
    {
      "epoch": 4.500959692898273,
      "grad_norm": 1.2142845269716316,
      "learning_rate": 1.208675082698854e-05,
      "loss": 0.1479,
      "step": 7035
    },
    {
      "epoch": 4.501599488163787,
      "grad_norm": 1.3091865262000029,
      "learning_rate": 1.2084724246636735e-05,
      "loss": 0.1387,
      "step": 7036
    },
    {
      "epoch": 4.502239283429303,
      "grad_norm": 1.2220924144484095,
      "learning_rate": 1.2082697576770577e-05,
      "loss": 0.1322,
      "step": 7037
    },
    {
      "epoch": 4.502879078694818,
      "grad_norm": 1.1833197328458807,
      "learning_rate": 1.2080670817477092e-05,
      "loss": 0.1286,
      "step": 7038
    },
    {
      "epoch": 4.503518873960333,
      "grad_norm": 1.4813082832667854,
      "learning_rate": 1.2078643968843303e-05,
      "loss": 0.1667,
      "step": 7039
    },
    {
      "epoch": 4.5041586692258475,
      "grad_norm": 1.3247673518261627,
      "learning_rate": 1.2076617030956237e-05,
      "loss": 0.1399,
      "step": 7040
    },
    {
      "epoch": 4.504798464491363,
      "grad_norm": 1.2162528580616325,
      "learning_rate": 1.2074590003902934e-05,
      "loss": 0.1408,
      "step": 7041
    },
    {
      "epoch": 4.505438259756878,
      "grad_norm": 1.3776498121381924,
      "learning_rate": 1.207256288777042e-05,
      "loss": 0.152,
      "step": 7042
    },
    {
      "epoch": 4.5060780550223924,
      "grad_norm": 1.2868179944687184,
      "learning_rate": 1.2070535682645748e-05,
      "loss": 0.1423,
      "step": 7043
    },
    {
      "epoch": 4.506717850287908,
      "grad_norm": 1.2639305832159184,
      "learning_rate": 1.2068508388615957e-05,
      "loss": 0.1451,
      "step": 7044
    },
    {
      "epoch": 4.507357645553423,
      "grad_norm": 1.2211233905512953,
      "learning_rate": 1.2066481005768091e-05,
      "loss": 0.1378,
      "step": 7045
    },
    {
      "epoch": 4.507997440818938,
      "grad_norm": 1.2457945182425736,
      "learning_rate": 1.2064453534189212e-05,
      "loss": 0.136,
      "step": 7046
    },
    {
      "epoch": 4.508637236084453,
      "grad_norm": 1.1766858341671087,
      "learning_rate": 1.2062425973966367e-05,
      "loss": 0.1338,
      "step": 7047
    },
    {
      "epoch": 4.509277031349968,
      "grad_norm": 1.2021971669366913,
      "learning_rate": 1.206039832518662e-05,
      "loss": 0.1369,
      "step": 7048
    },
    {
      "epoch": 4.509916826615483,
      "grad_norm": 1.409144469273563,
      "learning_rate": 1.2058370587937037e-05,
      "loss": 0.1511,
      "step": 7049
    },
    {
      "epoch": 4.510556621880998,
      "grad_norm": 1.2622607321344492,
      "learning_rate": 1.2056342762304679e-05,
      "loss": 0.1332,
      "step": 7050
    },
    {
      "epoch": 4.511196417146513,
      "grad_norm": 1.3361100659801455,
      "learning_rate": 1.2054314848376622e-05,
      "loss": 0.1525,
      "step": 7051
    },
    {
      "epoch": 4.511836212412028,
      "grad_norm": 1.2532725946608931,
      "learning_rate": 1.2052286846239938e-05,
      "loss": 0.1433,
      "step": 7052
    },
    {
      "epoch": 4.512476007677543,
      "grad_norm": 1.2340281781080196,
      "learning_rate": 1.205025875598171e-05,
      "loss": 0.131,
      "step": 7053
    },
    {
      "epoch": 4.513115802943059,
      "grad_norm": 1.2508087428437875,
      "learning_rate": 1.204823057768902e-05,
      "loss": 0.1368,
      "step": 7054
    },
    {
      "epoch": 4.513755598208573,
      "grad_norm": 1.2737000783222783,
      "learning_rate": 1.2046202311448947e-05,
      "loss": 0.1405,
      "step": 7055
    },
    {
      "epoch": 4.514395393474088,
      "grad_norm": 1.277133116210459,
      "learning_rate": 1.204417395734859e-05,
      "loss": 0.1477,
      "step": 7056
    },
    {
      "epoch": 4.515035188739604,
      "grad_norm": 1.1814392255877517,
      "learning_rate": 1.2042145515475038e-05,
      "loss": 0.1229,
      "step": 7057
    },
    {
      "epoch": 4.515674984005118,
      "grad_norm": 1.1952712800924445,
      "learning_rate": 1.204011698591539e-05,
      "loss": 0.1391,
      "step": 7058
    },
    {
      "epoch": 4.516314779270633,
      "grad_norm": 1.3480422281550555,
      "learning_rate": 1.203808836875675e-05,
      "loss": 0.1477,
      "step": 7059
    },
    {
      "epoch": 4.516954574536149,
      "grad_norm": 1.3394810055800555,
      "learning_rate": 1.2036059664086222e-05,
      "loss": 0.1348,
      "step": 7060
    },
    {
      "epoch": 4.517594369801664,
      "grad_norm": 1.2986680937452695,
      "learning_rate": 1.2034030871990911e-05,
      "loss": 0.1505,
      "step": 7061
    },
    {
      "epoch": 4.518234165067178,
      "grad_norm": 1.2628672550956193,
      "learning_rate": 1.2032001992557936e-05,
      "loss": 0.1258,
      "step": 7062
    },
    {
      "epoch": 4.5188739603326935,
      "grad_norm": 1.3247581886445734,
      "learning_rate": 1.2029973025874406e-05,
      "loss": 0.1537,
      "step": 7063
    },
    {
      "epoch": 4.519513755598209,
      "grad_norm": 1.370129105368203,
      "learning_rate": 1.2027943972027448e-05,
      "loss": 0.1558,
      "step": 7064
    },
    {
      "epoch": 4.520153550863723,
      "grad_norm": 1.3603565837359393,
      "learning_rate": 1.2025914831104179e-05,
      "loss": 0.1588,
      "step": 7065
    },
    {
      "epoch": 4.520793346129238,
      "grad_norm": 1.3552147029267874,
      "learning_rate": 1.2023885603191736e-05,
      "loss": 0.1396,
      "step": 7066
    },
    {
      "epoch": 4.521433141394754,
      "grad_norm": 1.3901139654061438,
      "learning_rate": 1.2021856288377243e-05,
      "loss": 0.1513,
      "step": 7067
    },
    {
      "epoch": 4.522072936660269,
      "grad_norm": 1.281581095081437,
      "learning_rate": 1.2019826886747841e-05,
      "loss": 0.1453,
      "step": 7068
    },
    {
      "epoch": 4.522712731925783,
      "grad_norm": 1.2623685212193176,
      "learning_rate": 1.2017797398390663e-05,
      "loss": 0.124,
      "step": 7069
    },
    {
      "epoch": 4.523352527191299,
      "grad_norm": 1.233149831005503,
      "learning_rate": 1.2015767823392851e-05,
      "loss": 0.131,
      "step": 7070
    },
    {
      "epoch": 4.523992322456814,
      "grad_norm": 1.1659855994988448,
      "learning_rate": 1.2013738161841559e-05,
      "loss": 0.1289,
      "step": 7071
    },
    {
      "epoch": 4.524632117722329,
      "grad_norm": 1.20783201463436,
      "learning_rate": 1.201170841382393e-05,
      "loss": 0.1283,
      "step": 7072
    },
    {
      "epoch": 4.525271912987844,
      "grad_norm": 1.2745083465233735,
      "learning_rate": 1.200967857942712e-05,
      "loss": 0.1458,
      "step": 7073
    },
    {
      "epoch": 4.525911708253359,
      "grad_norm": 1.2098382143100108,
      "learning_rate": 1.2007648658738288e-05,
      "loss": 0.1379,
      "step": 7074
    },
    {
      "epoch": 4.526551503518874,
      "grad_norm": 1.1636745878338899,
      "learning_rate": 1.2005618651844593e-05,
      "loss": 0.1419,
      "step": 7075
    },
    {
      "epoch": 4.527191298784389,
      "grad_norm": 1.273200486775004,
      "learning_rate": 1.20035885588332e-05,
      "loss": 0.1534,
      "step": 7076
    },
    {
      "epoch": 4.527831094049904,
      "grad_norm": 1.2726493892247281,
      "learning_rate": 1.2001558379791283e-05,
      "loss": 0.1356,
      "step": 7077
    },
    {
      "epoch": 4.528470889315419,
      "grad_norm": 1.4040850766757171,
      "learning_rate": 1.1999528114806005e-05,
      "loss": 0.1656,
      "step": 7078
    },
    {
      "epoch": 4.529110684580934,
      "grad_norm": 1.2534568941738087,
      "learning_rate": 1.199749776396455e-05,
      "loss": 0.1458,
      "step": 7079
    },
    {
      "epoch": 4.529750479846449,
      "grad_norm": 1.2839582141729324,
      "learning_rate": 1.199546732735409e-05,
      "loss": 0.1358,
      "step": 7080
    },
    {
      "epoch": 4.530390275111964,
      "grad_norm": 1.3757011070871312,
      "learning_rate": 1.1993436805061815e-05,
      "loss": 0.1543,
      "step": 7081
    },
    {
      "epoch": 4.531030070377479,
      "grad_norm": 1.2186889891563106,
      "learning_rate": 1.1991406197174911e-05,
      "loss": 0.1365,
      "step": 7082
    },
    {
      "epoch": 4.531669865642995,
      "grad_norm": 1.2547848344640153,
      "learning_rate": 1.1989375503780568e-05,
      "loss": 0.1372,
      "step": 7083
    },
    {
      "epoch": 4.532309660908509,
      "grad_norm": 1.159619400134232,
      "learning_rate": 1.1987344724965977e-05,
      "loss": 0.1306,
      "step": 7084
    },
    {
      "epoch": 4.532949456174024,
      "grad_norm": 1.3168915870997504,
      "learning_rate": 1.1985313860818336e-05,
      "loss": 0.1418,
      "step": 7085
    },
    {
      "epoch": 4.5335892514395395,
      "grad_norm": 1.2158562254391216,
      "learning_rate": 1.1983282911424856e-05,
      "loss": 0.1406,
      "step": 7086
    },
    {
      "epoch": 4.534229046705054,
      "grad_norm": 1.3310682314436022,
      "learning_rate": 1.198125187687273e-05,
      "loss": 0.1486,
      "step": 7087
    },
    {
      "epoch": 4.534868841970569,
      "grad_norm": 1.215119943594531,
      "learning_rate": 1.1979220757249174e-05,
      "loss": 0.1419,
      "step": 7088
    },
    {
      "epoch": 4.535508637236084,
      "grad_norm": 1.3560042221236175,
      "learning_rate": 1.1977189552641404e-05,
      "loss": 0.151,
      "step": 7089
    },
    {
      "epoch": 4.5361484325016,
      "grad_norm": 1.2117585076832615,
      "learning_rate": 1.1975158263136626e-05,
      "loss": 0.1404,
      "step": 7090
    },
    {
      "epoch": 4.536788227767115,
      "grad_norm": 1.2453260035937814,
      "learning_rate": 1.1973126888822072e-05,
      "loss": 0.1405,
      "step": 7091
    },
    {
      "epoch": 4.537428023032629,
      "grad_norm": 1.3144649442836434,
      "learning_rate": 1.1971095429784954e-05,
      "loss": 0.1485,
      "step": 7092
    },
    {
      "epoch": 4.538067818298145,
      "grad_norm": 1.2522330674342228,
      "learning_rate": 1.1969063886112507e-05,
      "loss": 0.1464,
      "step": 7093
    },
    {
      "epoch": 4.53870761356366,
      "grad_norm": 1.2573216300109726,
      "learning_rate": 1.1967032257891957e-05,
      "loss": 0.1383,
      "step": 7094
    },
    {
      "epoch": 4.539347408829174,
      "grad_norm": 1.360899348507786,
      "learning_rate": 1.1965000545210544e-05,
      "loss": 0.1643,
      "step": 7095
    },
    {
      "epoch": 4.53998720409469,
      "grad_norm": 1.310778512863593,
      "learning_rate": 1.1962968748155501e-05,
      "loss": 0.144,
      "step": 7096
    },
    {
      "epoch": 4.540626999360205,
      "grad_norm": 1.1761404923334076,
      "learning_rate": 1.1960936866814073e-05,
      "loss": 0.1346,
      "step": 7097
    },
    {
      "epoch": 4.54126679462572,
      "grad_norm": 1.2235062789600837,
      "learning_rate": 1.1958904901273503e-05,
      "loss": 0.1313,
      "step": 7098
    },
    {
      "epoch": 4.5419065898912345,
      "grad_norm": 1.1523270745542449,
      "learning_rate": 1.1956872851621043e-05,
      "loss": 0.1305,
      "step": 7099
    },
    {
      "epoch": 4.54254638515675,
      "grad_norm": 1.2569394087446988,
      "learning_rate": 1.195484071794394e-05,
      "loss": 0.1375,
      "step": 7100
    },
    {
      "epoch": 4.543186180422265,
      "grad_norm": 1.2961423320869325,
      "learning_rate": 1.1952808500329459e-05,
      "loss": 0.1454,
      "step": 7101
    },
    {
      "epoch": 4.5438259756877795,
      "grad_norm": 1.248154671596564,
      "learning_rate": 1.195077619886485e-05,
      "loss": 0.132,
      "step": 7102
    },
    {
      "epoch": 4.544465770953295,
      "grad_norm": 1.4244740376712268,
      "learning_rate": 1.1948743813637383e-05,
      "loss": 0.1578,
      "step": 7103
    },
    {
      "epoch": 4.54510556621881,
      "grad_norm": 1.2805721594082555,
      "learning_rate": 1.1946711344734326e-05,
      "loss": 0.1385,
      "step": 7104
    },
    {
      "epoch": 4.545745361484325,
      "grad_norm": 1.5148978689107429,
      "learning_rate": 1.194467879224294e-05,
      "loss": 0.16,
      "step": 7105
    },
    {
      "epoch": 4.54638515674984,
      "grad_norm": 1.169602879380975,
      "learning_rate": 1.1942646156250513e-05,
      "loss": 0.1415,
      "step": 7106
    },
    {
      "epoch": 4.547024952015355,
      "grad_norm": 1.2419461474459088,
      "learning_rate": 1.1940613436844315e-05,
      "loss": 0.1423,
      "step": 7107
    },
    {
      "epoch": 4.54766474728087,
      "grad_norm": 1.2872568630738692,
      "learning_rate": 1.1938580634111627e-05,
      "loss": 0.1505,
      "step": 7108
    },
    {
      "epoch": 4.5483045425463855,
      "grad_norm": 1.384776113039226,
      "learning_rate": 1.1936547748139733e-05,
      "loss": 0.1433,
      "step": 7109
    },
    {
      "epoch": 4.5489443378119,
      "grad_norm": 1.3384237465243645,
      "learning_rate": 1.1934514779015923e-05,
      "loss": 0.1422,
      "step": 7110
    },
    {
      "epoch": 4.549584133077415,
      "grad_norm": 1.294174654313507,
      "learning_rate": 1.1932481726827492e-05,
      "loss": 0.148,
      "step": 7111
    },
    {
      "epoch": 4.55022392834293,
      "grad_norm": 1.3248777548011217,
      "learning_rate": 1.1930448591661733e-05,
      "loss": 0.156,
      "step": 7112
    },
    {
      "epoch": 4.550863723608446,
      "grad_norm": 1.2723484416817985,
      "learning_rate": 1.1928415373605943e-05,
      "loss": 0.1368,
      "step": 7113
    },
    {
      "epoch": 4.55150351887396,
      "grad_norm": 1.378160693396387,
      "learning_rate": 1.1926382072747434e-05,
      "loss": 0.1446,
      "step": 7114
    },
    {
      "epoch": 4.552143314139475,
      "grad_norm": 1.4505452391587441,
      "learning_rate": 1.1924348689173497e-05,
      "loss": 0.1591,
      "step": 7115
    },
    {
      "epoch": 4.552783109404991,
      "grad_norm": 1.2662554185878196,
      "learning_rate": 1.1922315222971458e-05,
      "loss": 0.1375,
      "step": 7116
    },
    {
      "epoch": 4.553422904670505,
      "grad_norm": 1.3581688385542892,
      "learning_rate": 1.1920281674228617e-05,
      "loss": 0.1539,
      "step": 7117
    },
    {
      "epoch": 4.55406269993602,
      "grad_norm": 1.2773261057987924,
      "learning_rate": 1.19182480430323e-05,
      "loss": 0.1441,
      "step": 7118
    },
    {
      "epoch": 4.554702495201536,
      "grad_norm": 1.255650925276046,
      "learning_rate": 1.1916214329469822e-05,
      "loss": 0.1397,
      "step": 7119
    },
    {
      "epoch": 4.555342290467051,
      "grad_norm": 1.1376041762190514,
      "learning_rate": 1.191418053362851e-05,
      "loss": 0.133,
      "step": 7120
    },
    {
      "epoch": 4.555982085732565,
      "grad_norm": 1.230152079949002,
      "learning_rate": 1.1912146655595693e-05,
      "loss": 0.1323,
      "step": 7121
    },
    {
      "epoch": 4.5566218809980805,
      "grad_norm": 1.3828152253642534,
      "learning_rate": 1.19101126954587e-05,
      "loss": 0.1451,
      "step": 7122
    },
    {
      "epoch": 4.557261676263596,
      "grad_norm": 1.2607070872285322,
      "learning_rate": 1.1908078653304864e-05,
      "loss": 0.1457,
      "step": 7123
    },
    {
      "epoch": 4.557901471529111,
      "grad_norm": 1.3259316154756648,
      "learning_rate": 1.1906044529221525e-05,
      "loss": 0.1466,
      "step": 7124
    },
    {
      "epoch": 4.5585412667946255,
      "grad_norm": 1.2768385898522117,
      "learning_rate": 1.1904010323296022e-05,
      "loss": 0.1398,
      "step": 7125
    },
    {
      "epoch": 4.559181062060141,
      "grad_norm": 1.2588595100542457,
      "learning_rate": 1.1901976035615708e-05,
      "loss": 0.1406,
      "step": 7126
    },
    {
      "epoch": 4.559820857325656,
      "grad_norm": 1.2446670787073035,
      "learning_rate": 1.1899941666267922e-05,
      "loss": 0.1385,
      "step": 7127
    },
    {
      "epoch": 4.560460652591171,
      "grad_norm": 1.2253064761382602,
      "learning_rate": 1.1897907215340023e-05,
      "loss": 0.136,
      "step": 7128
    },
    {
      "epoch": 4.561100447856686,
      "grad_norm": 1.226858798318005,
      "learning_rate": 1.1895872682919364e-05,
      "loss": 0.1268,
      "step": 7129
    },
    {
      "epoch": 4.561740243122201,
      "grad_norm": 1.33007887248024,
      "learning_rate": 1.1893838069093304e-05,
      "loss": 0.1472,
      "step": 7130
    },
    {
      "epoch": 4.562380038387716,
      "grad_norm": 1.2965950347082806,
      "learning_rate": 1.1891803373949208e-05,
      "loss": 0.1502,
      "step": 7131
    },
    {
      "epoch": 4.563019833653231,
      "grad_norm": 1.348756195373535,
      "learning_rate": 1.1889768597574438e-05,
      "loss": 0.1563,
      "step": 7132
    },
    {
      "epoch": 4.563659628918746,
      "grad_norm": 1.1427086553367234,
      "learning_rate": 1.1887733740056369e-05,
      "loss": 0.1319,
      "step": 7133
    },
    {
      "epoch": 4.564299424184261,
      "grad_norm": 1.2360837947688836,
      "learning_rate": 1.1885698801482366e-05,
      "loss": 0.1457,
      "step": 7134
    },
    {
      "epoch": 4.564939219449776,
      "grad_norm": 1.2957524354177359,
      "learning_rate": 1.1883663781939814e-05,
      "loss": 0.1397,
      "step": 7135
    },
    {
      "epoch": 4.565579014715291,
      "grad_norm": 1.4525340422139763,
      "learning_rate": 1.188162868151609e-05,
      "loss": 0.1647,
      "step": 7136
    },
    {
      "epoch": 4.566218809980806,
      "grad_norm": 1.266768509819252,
      "learning_rate": 1.1879593500298579e-05,
      "loss": 0.1401,
      "step": 7137
    },
    {
      "epoch": 4.566858605246321,
      "grad_norm": 1.18047228222102,
      "learning_rate": 1.1877558238374666e-05,
      "loss": 0.1317,
      "step": 7138
    },
    {
      "epoch": 4.567498400511836,
      "grad_norm": 1.1299283379616833,
      "learning_rate": 1.1875522895831742e-05,
      "loss": 0.1252,
      "step": 7139
    },
    {
      "epoch": 4.568138195777351,
      "grad_norm": 1.35720841371251,
      "learning_rate": 1.1873487472757198e-05,
      "loss": 0.1529,
      "step": 7140
    },
    {
      "epoch": 4.568777991042866,
      "grad_norm": 1.3635295993044116,
      "learning_rate": 1.1871451969238437e-05,
      "loss": 0.1513,
      "step": 7141
    },
    {
      "epoch": 4.569417786308382,
      "grad_norm": 1.1859223079162653,
      "learning_rate": 1.1869416385362858e-05,
      "loss": 0.1255,
      "step": 7142
    },
    {
      "epoch": 4.570057581573896,
      "grad_norm": 1.2208563344573622,
      "learning_rate": 1.1867380721217864e-05,
      "loss": 0.1405,
      "step": 7143
    },
    {
      "epoch": 4.570697376839411,
      "grad_norm": 1.3636078107925356,
      "learning_rate": 1.1865344976890865e-05,
      "loss": 0.1451,
      "step": 7144
    },
    {
      "epoch": 4.5713371721049265,
      "grad_norm": 1.3183621191644224,
      "learning_rate": 1.186330915246927e-05,
      "loss": 0.1413,
      "step": 7145
    },
    {
      "epoch": 4.571976967370442,
      "grad_norm": 1.2608085114731529,
      "learning_rate": 1.1861273248040494e-05,
      "loss": 0.1307,
      "step": 7146
    },
    {
      "epoch": 4.572616762635956,
      "grad_norm": 1.2143820903537417,
      "learning_rate": 1.1859237263691956e-05,
      "loss": 0.1318,
      "step": 7147
    },
    {
      "epoch": 4.5732565579014715,
      "grad_norm": 1.2314260745951389,
      "learning_rate": 1.1857201199511077e-05,
      "loss": 0.1415,
      "step": 7148
    },
    {
      "epoch": 4.573896353166987,
      "grad_norm": 1.2513886957465281,
      "learning_rate": 1.1855165055585282e-05,
      "loss": 0.1349,
      "step": 7149
    },
    {
      "epoch": 4.574536148432502,
      "grad_norm": 1.2777454554062329,
      "learning_rate": 1.1853128832001999e-05,
      "loss": 0.1504,
      "step": 7150
    },
    {
      "epoch": 4.575175943698016,
      "grad_norm": 1.3600703640055876,
      "learning_rate": 1.1851092528848662e-05,
      "loss": 0.1565,
      "step": 7151
    },
    {
      "epoch": 4.575815738963532,
      "grad_norm": 1.2880396456192282,
      "learning_rate": 1.1849056146212707e-05,
      "loss": 0.1541,
      "step": 7152
    },
    {
      "epoch": 4.576455534229047,
      "grad_norm": 1.247573698302063,
      "learning_rate": 1.1847019684181568e-05,
      "loss": 0.1413,
      "step": 7153
    },
    {
      "epoch": 4.577095329494561,
      "grad_norm": 1.4732280073429127,
      "learning_rate": 1.1844983142842687e-05,
      "loss": 0.1599,
      "step": 7154
    },
    {
      "epoch": 4.577735124760077,
      "grad_norm": 1.2740111881342584,
      "learning_rate": 1.1842946522283514e-05,
      "loss": 0.1389,
      "step": 7155
    },
    {
      "epoch": 4.578374920025592,
      "grad_norm": 1.1407516007131944,
      "learning_rate": 1.1840909822591495e-05,
      "loss": 0.1287,
      "step": 7156
    },
    {
      "epoch": 4.579014715291107,
      "grad_norm": 1.3248153460254053,
      "learning_rate": 1.1838873043854083e-05,
      "loss": 0.143,
      "step": 7157
    },
    {
      "epoch": 4.5796545105566215,
      "grad_norm": 1.1657782378587096,
      "learning_rate": 1.1836836186158736e-05,
      "loss": 0.1303,
      "step": 7158
    },
    {
      "epoch": 4.580294305822137,
      "grad_norm": 1.2365075307692979,
      "learning_rate": 1.1834799249592908e-05,
      "loss": 0.1448,
      "step": 7159
    },
    {
      "epoch": 4.580934101087652,
      "grad_norm": 1.2871800492305683,
      "learning_rate": 1.1832762234244064e-05,
      "loss": 0.1352,
      "step": 7160
    },
    {
      "epoch": 4.581573896353167,
      "grad_norm": 1.2098834636743583,
      "learning_rate": 1.1830725140199672e-05,
      "loss": 0.1272,
      "step": 7161
    },
    {
      "epoch": 4.582213691618682,
      "grad_norm": 1.1825542962116173,
      "learning_rate": 1.1828687967547196e-05,
      "loss": 0.1318,
      "step": 7162
    },
    {
      "epoch": 4.582853486884197,
      "grad_norm": 1.2514648466138039,
      "learning_rate": 1.1826650716374115e-05,
      "loss": 0.1419,
      "step": 7163
    },
    {
      "epoch": 4.583493282149712,
      "grad_norm": 1.3379869806869134,
      "learning_rate": 1.1824613386767898e-05,
      "loss": 0.1408,
      "step": 7164
    },
    {
      "epoch": 4.584133077415228,
      "grad_norm": 1.2383508824092453,
      "learning_rate": 1.1822575978816028e-05,
      "loss": 0.1391,
      "step": 7165
    },
    {
      "epoch": 4.584772872680742,
      "grad_norm": 1.2995499950097706,
      "learning_rate": 1.182053849260599e-05,
      "loss": 0.1388,
      "step": 7166
    },
    {
      "epoch": 4.585412667946257,
      "grad_norm": 1.2816773629479241,
      "learning_rate": 1.1818500928225269e-05,
      "loss": 0.1393,
      "step": 7167
    },
    {
      "epoch": 4.5860524632117725,
      "grad_norm": 1.2581887081686687,
      "learning_rate": 1.1816463285761352e-05,
      "loss": 0.1287,
      "step": 7168
    },
    {
      "epoch": 4.586692258477287,
      "grad_norm": 1.3233702247144703,
      "learning_rate": 1.181442556530173e-05,
      "loss": 0.1444,
      "step": 7169
    },
    {
      "epoch": 4.587332053742802,
      "grad_norm": 1.309367043394958,
      "learning_rate": 1.1812387766933903e-05,
      "loss": 0.1351,
      "step": 7170
    },
    {
      "epoch": 4.5879718490083174,
      "grad_norm": 1.3305354523712414,
      "learning_rate": 1.1810349890745372e-05,
      "loss": 0.1339,
      "step": 7171
    },
    {
      "epoch": 4.588611644273833,
      "grad_norm": 1.3068030268478188,
      "learning_rate": 1.1808311936823634e-05,
      "loss": 0.1503,
      "step": 7172
    },
    {
      "epoch": 4.589251439539347,
      "grad_norm": 1.3307888561663026,
      "learning_rate": 1.1806273905256198e-05,
      "loss": 0.1404,
      "step": 7173
    },
    {
      "epoch": 4.589891234804862,
      "grad_norm": 1.2538921938532523,
      "learning_rate": 1.1804235796130577e-05,
      "loss": 0.1428,
      "step": 7174
    },
    {
      "epoch": 4.590531030070378,
      "grad_norm": 1.1763968372978604,
      "learning_rate": 1.1802197609534276e-05,
      "loss": 0.1368,
      "step": 7175
    },
    {
      "epoch": 4.591170825335892,
      "grad_norm": 1.3050253202941604,
      "learning_rate": 1.1800159345554818e-05,
      "loss": 0.1455,
      "step": 7176
    },
    {
      "epoch": 4.591810620601407,
      "grad_norm": 1.6827031909187349,
      "learning_rate": 1.179812100427972e-05,
      "loss": 0.1471,
      "step": 7177
    },
    {
      "epoch": 4.592450415866923,
      "grad_norm": 1.238838271294913,
      "learning_rate": 1.1796082585796505e-05,
      "loss": 0.127,
      "step": 7178
    },
    {
      "epoch": 4.593090211132438,
      "grad_norm": 1.1798966503802093,
      "learning_rate": 1.1794044090192696e-05,
      "loss": 0.1314,
      "step": 7179
    },
    {
      "epoch": 4.593730006397953,
      "grad_norm": 1.1561754705671468,
      "learning_rate": 1.1792005517555827e-05,
      "loss": 0.1342,
      "step": 7180
    },
    {
      "epoch": 4.5943698016634675,
      "grad_norm": 1.4081271330659917,
      "learning_rate": 1.1789966867973429e-05,
      "loss": 0.1458,
      "step": 7181
    },
    {
      "epoch": 4.595009596928983,
      "grad_norm": 1.277838431737153,
      "learning_rate": 1.1787928141533038e-05,
      "loss": 0.141,
      "step": 7182
    },
    {
      "epoch": 4.595649392194498,
      "grad_norm": 1.2399062006700177,
      "learning_rate": 1.1785889338322195e-05,
      "loss": 0.139,
      "step": 7183
    },
    {
      "epoch": 4.5962891874600125,
      "grad_norm": 1.221201588239478,
      "learning_rate": 1.1783850458428436e-05,
      "loss": 0.1323,
      "step": 7184
    },
    {
      "epoch": 4.596928982725528,
      "grad_norm": 1.2512219222770646,
      "learning_rate": 1.1781811501939313e-05,
      "loss": 0.1441,
      "step": 7185
    },
    {
      "epoch": 4.597568777991043,
      "grad_norm": 1.342198651129115,
      "learning_rate": 1.1779772468942377e-05,
      "loss": 0.1461,
      "step": 7186
    },
    {
      "epoch": 4.598208573256558,
      "grad_norm": 1.3416137007897528,
      "learning_rate": 1.1777733359525172e-05,
      "loss": 0.1552,
      "step": 7187
    },
    {
      "epoch": 4.598848368522073,
      "grad_norm": 1.305264272357973,
      "learning_rate": 1.1775694173775265e-05,
      "loss": 0.1389,
      "step": 7188
    },
    {
      "epoch": 4.599488163787588,
      "grad_norm": 1.409249706355143,
      "learning_rate": 1.1773654911780206e-05,
      "loss": 0.1517,
      "step": 7189
    },
    {
      "epoch": 4.600127959053103,
      "grad_norm": 1.2376712054381351,
      "learning_rate": 1.177161557362756e-05,
      "loss": 0.1432,
      "step": 7190
    },
    {
      "epoch": 4.600767754318618,
      "grad_norm": 1.450854284811107,
      "learning_rate": 1.1769576159404891e-05,
      "loss": 0.147,
      "step": 7191
    },
    {
      "epoch": 4.601407549584133,
      "grad_norm": 1.3541923471508386,
      "learning_rate": 1.1767536669199771e-05,
      "loss": 0.1488,
      "step": 7192
    },
    {
      "epoch": 4.602047344849648,
      "grad_norm": 1.1620539297378125,
      "learning_rate": 1.1765497103099772e-05,
      "loss": 0.1273,
      "step": 7193
    },
    {
      "epoch": 4.6026871401151634,
      "grad_norm": 1.3310523958410398,
      "learning_rate": 1.1763457461192464e-05,
      "loss": 0.1371,
      "step": 7194
    },
    {
      "epoch": 4.603326935380678,
      "grad_norm": 1.2635079449396855,
      "learning_rate": 1.1761417743565434e-05,
      "loss": 0.1335,
      "step": 7195
    },
    {
      "epoch": 4.603966730646193,
      "grad_norm": 1.1606300810512058,
      "learning_rate": 1.1759377950306257e-05,
      "loss": 0.1243,
      "step": 7196
    },
    {
      "epoch": 4.604606525911708,
      "grad_norm": 1.3193227233626725,
      "learning_rate": 1.1757338081502522e-05,
      "loss": 0.1417,
      "step": 7197
    },
    {
      "epoch": 4.605246321177224,
      "grad_norm": 1.30295632349406,
      "learning_rate": 1.1755298137241815e-05,
      "loss": 0.1421,
      "step": 7198
    },
    {
      "epoch": 4.605886116442738,
      "grad_norm": 1.1783640666514716,
      "learning_rate": 1.1753258117611727e-05,
      "loss": 0.1333,
      "step": 7199
    },
    {
      "epoch": 4.606525911708253,
      "grad_norm": 1.3104892556750498,
      "learning_rate": 1.1751218022699857e-05,
      "loss": 0.1363,
      "step": 7200
    },
    {
      "epoch": 4.607165706973769,
      "grad_norm": 1.1814158767777319,
      "learning_rate": 1.1749177852593798e-05,
      "loss": 0.141,
      "step": 7201
    },
    {
      "epoch": 4.607805502239284,
      "grad_norm": 1.2539332811977753,
      "learning_rate": 1.1747137607381154e-05,
      "loss": 0.1425,
      "step": 7202
    },
    {
      "epoch": 4.608445297504798,
      "grad_norm": 1.3206605523372066,
      "learning_rate": 1.174509728714953e-05,
      "loss": 0.1469,
      "step": 7203
    },
    {
      "epoch": 4.6090850927703135,
      "grad_norm": 1.2624149628918138,
      "learning_rate": 1.1743056891986533e-05,
      "loss": 0.1334,
      "step": 7204
    },
    {
      "epoch": 4.609724888035829,
      "grad_norm": 1.3168289977179766,
      "learning_rate": 1.1741016421979773e-05,
      "loss": 0.1498,
      "step": 7205
    },
    {
      "epoch": 4.610364683301343,
      "grad_norm": 1.3341050826272929,
      "learning_rate": 1.1738975877216865e-05,
      "loss": 0.1331,
      "step": 7206
    },
    {
      "epoch": 4.6110044785668585,
      "grad_norm": 1.2367872651145568,
      "learning_rate": 1.1736935257785425e-05,
      "loss": 0.1452,
      "step": 7207
    },
    {
      "epoch": 4.611644273832374,
      "grad_norm": 1.3035874527637408,
      "learning_rate": 1.1734894563773077e-05,
      "loss": 0.1528,
      "step": 7208
    },
    {
      "epoch": 4.612284069097889,
      "grad_norm": 1.1898794838343014,
      "learning_rate": 1.173285379526744e-05,
      "loss": 0.1384,
      "step": 7209
    },
    {
      "epoch": 4.612923864363403,
      "grad_norm": 1.268334400374927,
      "learning_rate": 1.1730812952356145e-05,
      "loss": 0.1357,
      "step": 7210
    },
    {
      "epoch": 4.613563659628919,
      "grad_norm": 1.298548560913332,
      "learning_rate": 1.1728772035126821e-05,
      "loss": 0.152,
      "step": 7211
    },
    {
      "epoch": 4.614203454894434,
      "grad_norm": 1.2828025732295105,
      "learning_rate": 1.1726731043667102e-05,
      "loss": 0.1398,
      "step": 7212
    },
    {
      "epoch": 4.614843250159948,
      "grad_norm": 1.2950965929473754,
      "learning_rate": 1.1724689978064622e-05,
      "loss": 0.1326,
      "step": 7213
    },
    {
      "epoch": 4.615483045425464,
      "grad_norm": 1.1617360076701844,
      "learning_rate": 1.1722648838407023e-05,
      "loss": 0.1359,
      "step": 7214
    },
    {
      "epoch": 4.616122840690979,
      "grad_norm": 1.2586993459863804,
      "learning_rate": 1.1720607624781948e-05,
      "loss": 0.1373,
      "step": 7215
    },
    {
      "epoch": 4.616762635956494,
      "grad_norm": 1.2015787748603488,
      "learning_rate": 1.171856633727704e-05,
      "loss": 0.1311,
      "step": 7216
    },
    {
      "epoch": 4.617402431222009,
      "grad_norm": 1.3088506879616548,
      "learning_rate": 1.171652497597995e-05,
      "loss": 0.141,
      "step": 7217
    },
    {
      "epoch": 4.618042226487524,
      "grad_norm": 1.2231251285353886,
      "learning_rate": 1.1714483540978333e-05,
      "loss": 0.133,
      "step": 7218
    },
    {
      "epoch": 4.618682021753039,
      "grad_norm": 1.2905776994833744,
      "learning_rate": 1.1712442032359845e-05,
      "loss": 0.1406,
      "step": 7219
    },
    {
      "epoch": 4.619321817018554,
      "grad_norm": 1.216666107990809,
      "learning_rate": 1.171040045021214e-05,
      "loss": 0.1438,
      "step": 7220
    },
    {
      "epoch": 4.619961612284069,
      "grad_norm": 1.2916233784129425,
      "learning_rate": 1.1708358794622883e-05,
      "loss": 0.1453,
      "step": 7221
    },
    {
      "epoch": 4.620601407549584,
      "grad_norm": 1.2395444016527104,
      "learning_rate": 1.1706317065679736e-05,
      "loss": 0.1327,
      "step": 7222
    },
    {
      "epoch": 4.621241202815099,
      "grad_norm": 1.2201327535292978,
      "learning_rate": 1.170427526347037e-05,
      "loss": 0.1453,
      "step": 7223
    },
    {
      "epoch": 4.621880998080615,
      "grad_norm": 1.283438317234243,
      "learning_rate": 1.1702233388082456e-05,
      "loss": 0.1465,
      "step": 7224
    },
    {
      "epoch": 4.622520793346129,
      "grad_norm": 1.2677478321245153,
      "learning_rate": 1.170019143960367e-05,
      "loss": 0.1348,
      "step": 7225
    },
    {
      "epoch": 4.623160588611644,
      "grad_norm": 1.43640137564427,
      "learning_rate": 1.1698149418121685e-05,
      "loss": 0.1622,
      "step": 7226
    },
    {
      "epoch": 4.6238003838771595,
      "grad_norm": 1.2735732629330083,
      "learning_rate": 1.1696107323724189e-05,
      "loss": 0.141,
      "step": 7227
    },
    {
      "epoch": 4.624440179142674,
      "grad_norm": 1.4167025447105255,
      "learning_rate": 1.1694065156498859e-05,
      "loss": 0.1557,
      "step": 7228
    },
    {
      "epoch": 4.625079974408189,
      "grad_norm": 1.398101263485292,
      "learning_rate": 1.1692022916533386e-05,
      "loss": 0.1469,
      "step": 7229
    },
    {
      "epoch": 4.6257197696737045,
      "grad_norm": 1.302143237954436,
      "learning_rate": 1.1689980603915459e-05,
      "loss": 0.1421,
      "step": 7230
    },
    {
      "epoch": 4.62635956493922,
      "grad_norm": 1.362673088110331,
      "learning_rate": 1.1687938218732768e-05,
      "loss": 0.1436,
      "step": 7231
    },
    {
      "epoch": 4.626999360204734,
      "grad_norm": 1.3636817752911639,
      "learning_rate": 1.1685895761073014e-05,
      "loss": 0.1545,
      "step": 7232
    },
    {
      "epoch": 4.627639155470249,
      "grad_norm": 1.2121388675275981,
      "learning_rate": 1.1683853231023896e-05,
      "loss": 0.1343,
      "step": 7233
    },
    {
      "epoch": 4.628278950735765,
      "grad_norm": 1.1775300720656066,
      "learning_rate": 1.1681810628673116e-05,
      "loss": 0.1317,
      "step": 7234
    },
    {
      "epoch": 4.62891874600128,
      "grad_norm": 1.2625052914465218,
      "learning_rate": 1.1679767954108379e-05,
      "loss": 0.1403,
      "step": 7235
    },
    {
      "epoch": 4.629558541266794,
      "grad_norm": 1.2344842273515149,
      "learning_rate": 1.1677725207417391e-05,
      "loss": 0.1405,
      "step": 7236
    },
    {
      "epoch": 4.63019833653231,
      "grad_norm": 1.3735698719229452,
      "learning_rate": 1.167568238868787e-05,
      "loss": 0.145,
      "step": 7237
    },
    {
      "epoch": 4.630838131797825,
      "grad_norm": 1.2301419664808786,
      "learning_rate": 1.1673639498007527e-05,
      "loss": 0.14,
      "step": 7238
    },
    {
      "epoch": 4.63147792706334,
      "grad_norm": 1.218492277841668,
      "learning_rate": 1.1671596535464083e-05,
      "loss": 0.14,
      "step": 7239
    },
    {
      "epoch": 4.6321177223288545,
      "grad_norm": 1.4220421121321432,
      "learning_rate": 1.1669553501145257e-05,
      "loss": 0.1609,
      "step": 7240
    },
    {
      "epoch": 4.63275751759437,
      "grad_norm": 1.359031668744068,
      "learning_rate": 1.1667510395138773e-05,
      "loss": 0.1398,
      "step": 7241
    },
    {
      "epoch": 4.633397312859885,
      "grad_norm": 1.355043278164365,
      "learning_rate": 1.1665467217532357e-05,
      "loss": 0.1501,
      "step": 7242
    },
    {
      "epoch": 4.6340371081253995,
      "grad_norm": 1.3173931492100335,
      "learning_rate": 1.1663423968413743e-05,
      "loss": 0.1484,
      "step": 7243
    },
    {
      "epoch": 4.634676903390915,
      "grad_norm": 1.2526790466061146,
      "learning_rate": 1.1661380647870661e-05,
      "loss": 0.1382,
      "step": 7244
    },
    {
      "epoch": 4.63531669865643,
      "grad_norm": 1.3623279540956503,
      "learning_rate": 1.1659337255990853e-05,
      "loss": 0.141,
      "step": 7245
    },
    {
      "epoch": 4.635956493921945,
      "grad_norm": 1.2670793412743944,
      "learning_rate": 1.1657293792862051e-05,
      "loss": 0.1409,
      "step": 7246
    },
    {
      "epoch": 4.63659628918746,
      "grad_norm": 1.2793627280854847,
      "learning_rate": 1.1655250258572005e-05,
      "loss": 0.1437,
      "step": 7247
    },
    {
      "epoch": 4.637236084452975,
      "grad_norm": 1.3730430610022757,
      "learning_rate": 1.1653206653208455e-05,
      "loss": 0.1569,
      "step": 7248
    },
    {
      "epoch": 4.63787587971849,
      "grad_norm": 1.3696326401300871,
      "learning_rate": 1.1651162976859155e-05,
      "loss": 0.1582,
      "step": 7249
    },
    {
      "epoch": 4.638515674984005,
      "grad_norm": 1.4188413762941818,
      "learning_rate": 1.1649119229611854e-05,
      "loss": 0.1464,
      "step": 7250
    },
    {
      "epoch": 4.63915547024952,
      "grad_norm": 1.4348413676766734,
      "learning_rate": 1.1647075411554302e-05,
      "loss": 0.1621,
      "step": 7251
    },
    {
      "epoch": 4.639795265515035,
      "grad_norm": 1.3611796750648995,
      "learning_rate": 1.1645031522774264e-05,
      "loss": 0.1477,
      "step": 7252
    },
    {
      "epoch": 4.6404350607805505,
      "grad_norm": 1.2364214865261454,
      "learning_rate": 1.16429875633595e-05,
      "loss": 0.1424,
      "step": 7253
    },
    {
      "epoch": 4.641074856046066,
      "grad_norm": 1.292655324916386,
      "learning_rate": 1.164094353339777e-05,
      "loss": 0.1463,
      "step": 7254
    },
    {
      "epoch": 4.64171465131158,
      "grad_norm": 1.2521797935740644,
      "learning_rate": 1.1638899432976846e-05,
      "loss": 0.1383,
      "step": 7255
    },
    {
      "epoch": 4.642354446577095,
      "grad_norm": 1.1609698858267141,
      "learning_rate": 1.1636855262184499e-05,
      "loss": 0.1302,
      "step": 7256
    },
    {
      "epoch": 4.642994241842611,
      "grad_norm": 1.1237245143579826,
      "learning_rate": 1.1634811021108494e-05,
      "loss": 0.1205,
      "step": 7257
    },
    {
      "epoch": 4.643634037108125,
      "grad_norm": 1.3412041750799453,
      "learning_rate": 1.1632766709836614e-05,
      "loss": 0.1468,
      "step": 7258
    },
    {
      "epoch": 4.64427383237364,
      "grad_norm": 1.1911120086225655,
      "learning_rate": 1.1630722328456635e-05,
      "loss": 0.1343,
      "step": 7259
    },
    {
      "epoch": 4.644913627639156,
      "grad_norm": 1.2252386921655596,
      "learning_rate": 1.162867787705634e-05,
      "loss": 0.1346,
      "step": 7260
    },
    {
      "epoch": 4.645553422904671,
      "grad_norm": 1.3564336647033015,
      "learning_rate": 1.1626633355723514e-05,
      "loss": 0.1516,
      "step": 7261
    },
    {
      "epoch": 4.646193218170185,
      "grad_norm": 1.4372805837831995,
      "learning_rate": 1.1624588764545944e-05,
      "loss": 0.158,
      "step": 7262
    },
    {
      "epoch": 4.6468330134357005,
      "grad_norm": 1.3697995989193876,
      "learning_rate": 1.1622544103611424e-05,
      "loss": 0.1549,
      "step": 7263
    },
    {
      "epoch": 4.647472808701216,
      "grad_norm": 1.2927098124864058,
      "learning_rate": 1.1620499373007747e-05,
      "loss": 0.1413,
      "step": 7264
    },
    {
      "epoch": 4.64811260396673,
      "grad_norm": 1.362042391909508,
      "learning_rate": 1.161845457282271e-05,
      "loss": 0.134,
      "step": 7265
    },
    {
      "epoch": 4.6487523992322455,
      "grad_norm": 1.3541199930901395,
      "learning_rate": 1.1616409703144109e-05,
      "loss": 0.1463,
      "step": 7266
    },
    {
      "epoch": 4.649392194497761,
      "grad_norm": 1.2611903762314565,
      "learning_rate": 1.1614364764059756e-05,
      "loss": 0.1414,
      "step": 7267
    },
    {
      "epoch": 4.650031989763276,
      "grad_norm": 1.3837561242177703,
      "learning_rate": 1.1612319755657446e-05,
      "loss": 0.1516,
      "step": 7268
    },
    {
      "epoch": 4.65067178502879,
      "grad_norm": 1.3356141076315327,
      "learning_rate": 1.1610274678024997e-05,
      "loss": 0.1482,
      "step": 7269
    },
    {
      "epoch": 4.651311580294306,
      "grad_norm": 1.3015696478050651,
      "learning_rate": 1.160822953125022e-05,
      "loss": 0.1333,
      "step": 7270
    },
    {
      "epoch": 4.651951375559821,
      "grad_norm": 1.1607354483064605,
      "learning_rate": 1.1606184315420924e-05,
      "loss": 0.1225,
      "step": 7271
    },
    {
      "epoch": 4.652591170825336,
      "grad_norm": 1.2132194078008554,
      "learning_rate": 1.1604139030624935e-05,
      "loss": 0.132,
      "step": 7272
    },
    {
      "epoch": 4.653230966090851,
      "grad_norm": 1.3193909631987402,
      "learning_rate": 1.1602093676950064e-05,
      "loss": 0.1587,
      "step": 7273
    },
    {
      "epoch": 4.653870761356366,
      "grad_norm": 1.2951610792776842,
      "learning_rate": 1.1600048254484143e-05,
      "loss": 0.1488,
      "step": 7274
    },
    {
      "epoch": 4.654510556621881,
      "grad_norm": 1.2660455875114436,
      "learning_rate": 1.1598002763314998e-05,
      "loss": 0.151,
      "step": 7275
    },
    {
      "epoch": 4.6551503518873965,
      "grad_norm": 1.0917276877310036,
      "learning_rate": 1.1595957203530453e-05,
      "loss": 0.1187,
      "step": 7276
    },
    {
      "epoch": 4.655790147152911,
      "grad_norm": 1.2606305762284056,
      "learning_rate": 1.1593911575218346e-05,
      "loss": 0.1417,
      "step": 7277
    },
    {
      "epoch": 4.656429942418426,
      "grad_norm": 1.5064429427476844,
      "learning_rate": 1.1591865878466514e-05,
      "loss": 0.1672,
      "step": 7278
    },
    {
      "epoch": 4.657069737683941,
      "grad_norm": 1.1723966434434976,
      "learning_rate": 1.158982011336279e-05,
      "loss": 0.1388,
      "step": 7279
    },
    {
      "epoch": 4.657709532949456,
      "grad_norm": 1.4064607874242665,
      "learning_rate": 1.1587774279995021e-05,
      "loss": 0.1335,
      "step": 7280
    },
    {
      "epoch": 4.658349328214971,
      "grad_norm": 1.3360512540096605,
      "learning_rate": 1.1585728378451046e-05,
      "loss": 0.1317,
      "step": 7281
    },
    {
      "epoch": 4.658989123480486,
      "grad_norm": 1.2507941034214718,
      "learning_rate": 1.1583682408818717e-05,
      "loss": 0.1281,
      "step": 7282
    },
    {
      "epoch": 4.659628918746002,
      "grad_norm": 1.4077994283691924,
      "learning_rate": 1.158163637118588e-05,
      "loss": 0.1596,
      "step": 7283
    },
    {
      "epoch": 4.660268714011516,
      "grad_norm": 1.3461719163110066,
      "learning_rate": 1.1579590265640392e-05,
      "loss": 0.1429,
      "step": 7284
    },
    {
      "epoch": 4.660908509277031,
      "grad_norm": 1.466672861946991,
      "learning_rate": 1.157754409227011e-05,
      "loss": 0.1616,
      "step": 7285
    },
    {
      "epoch": 4.6615483045425465,
      "grad_norm": 1.3511150465222808,
      "learning_rate": 1.1575497851162892e-05,
      "loss": 0.1568,
      "step": 7286
    },
    {
      "epoch": 4.662188099808061,
      "grad_norm": 1.1927107435595798,
      "learning_rate": 1.1573451542406597e-05,
      "loss": 0.1382,
      "step": 7287
    },
    {
      "epoch": 4.662827895073576,
      "grad_norm": 1.1730254693586313,
      "learning_rate": 1.157140516608909e-05,
      "loss": 0.1315,
      "step": 7288
    },
    {
      "epoch": 4.6634676903390915,
      "grad_norm": 1.2219838144101538,
      "learning_rate": 1.156935872229824e-05,
      "loss": 0.1471,
      "step": 7289
    },
    {
      "epoch": 4.664107485604607,
      "grad_norm": 1.2733862871620205,
      "learning_rate": 1.156731221112192e-05,
      "loss": 0.1271,
      "step": 7290
    },
    {
      "epoch": 4.664747280870122,
      "grad_norm": 1.214444228282716,
      "learning_rate": 1.1565265632648002e-05,
      "loss": 0.131,
      "step": 7291
    },
    {
      "epoch": 4.665387076135636,
      "grad_norm": 1.4077080577597216,
      "learning_rate": 1.1563218986964361e-05,
      "loss": 0.1437,
      "step": 7292
    },
    {
      "epoch": 4.666026871401152,
      "grad_norm": 1.279975231247181,
      "learning_rate": 1.1561172274158877e-05,
      "loss": 0.1509,
      "step": 7293
    },
    {
      "epoch": 4.666666666666667,
      "grad_norm": 1.315160141108038,
      "learning_rate": 1.1559125494319433e-05,
      "loss": 0.1389,
      "step": 7294
    },
    {
      "epoch": 4.667306461932181,
      "grad_norm": 1.4093917263093154,
      "learning_rate": 1.1557078647533916e-05,
      "loss": 0.1525,
      "step": 7295
    },
    {
      "epoch": 4.667946257197697,
      "grad_norm": 1.2962757805894067,
      "learning_rate": 1.155503173389021e-05,
      "loss": 0.1364,
      "step": 7296
    },
    {
      "epoch": 4.668586052463212,
      "grad_norm": 1.3092535456744827,
      "learning_rate": 1.1552984753476209e-05,
      "loss": 0.142,
      "step": 7297
    },
    {
      "epoch": 4.669225847728727,
      "grad_norm": 1.3006422269715427,
      "learning_rate": 1.1550937706379804e-05,
      "loss": 0.1392,
      "step": 7298
    },
    {
      "epoch": 4.669865642994242,
      "grad_norm": 1.388431898110892,
      "learning_rate": 1.154889059268889e-05,
      "loss": 0.1528,
      "step": 7299
    },
    {
      "epoch": 4.670505438259757,
      "grad_norm": 1.3138348007319587,
      "learning_rate": 1.1546843412491373e-05,
      "loss": 0.1309,
      "step": 7300
    },
    {
      "epoch": 4.671145233525272,
      "grad_norm": 1.3242479495673567,
      "learning_rate": 1.1544796165875152e-05,
      "loss": 0.1458,
      "step": 7301
    },
    {
      "epoch": 4.6717850287907865,
      "grad_norm": 1.262943503141905,
      "learning_rate": 1.1542748852928129e-05,
      "loss": 0.1324,
      "step": 7302
    },
    {
      "epoch": 4.672424824056302,
      "grad_norm": 1.248172990006481,
      "learning_rate": 1.1540701473738216e-05,
      "loss": 0.1395,
      "step": 7303
    },
    {
      "epoch": 4.673064619321817,
      "grad_norm": 1.1362839530438638,
      "learning_rate": 1.1538654028393321e-05,
      "loss": 0.1234,
      "step": 7304
    },
    {
      "epoch": 4.673704414587332,
      "grad_norm": 1.345355630470497,
      "learning_rate": 1.153660651698136e-05,
      "loss": 0.1467,
      "step": 7305
    },
    {
      "epoch": 4.674344209852847,
      "grad_norm": 1.3243973393163555,
      "learning_rate": 1.1534558939590248e-05,
      "loss": 0.1473,
      "step": 7306
    },
    {
      "epoch": 4.674984005118362,
      "grad_norm": 1.4400512289831973,
      "learning_rate": 1.1532511296307907e-05,
      "loss": 0.1585,
      "step": 7307
    },
    {
      "epoch": 4.675623800383877,
      "grad_norm": 1.3765559305673736,
      "learning_rate": 1.1530463587222253e-05,
      "loss": 0.1557,
      "step": 7308
    },
    {
      "epoch": 4.6762635956493925,
      "grad_norm": 1.3817343621127636,
      "learning_rate": 1.1528415812421222e-05,
      "loss": 0.1525,
      "step": 7309
    },
    {
      "epoch": 4.676903390914907,
      "grad_norm": 1.3452811236251274,
      "learning_rate": 1.152636797199273e-05,
      "loss": 0.1594,
      "step": 7310
    },
    {
      "epoch": 4.677543186180422,
      "grad_norm": 1.2793971962786337,
      "learning_rate": 1.1524320066024713e-05,
      "loss": 0.131,
      "step": 7311
    },
    {
      "epoch": 4.6781829814459375,
      "grad_norm": 1.3878720604415178,
      "learning_rate": 1.1522272094605103e-05,
      "loss": 0.1538,
      "step": 7312
    },
    {
      "epoch": 4.678822776711453,
      "grad_norm": 1.1916653262183132,
      "learning_rate": 1.1520224057821836e-05,
      "loss": 0.1333,
      "step": 7313
    },
    {
      "epoch": 4.679462571976967,
      "grad_norm": 1.1891632897980196,
      "learning_rate": 1.1518175955762854e-05,
      "loss": 0.1266,
      "step": 7314
    },
    {
      "epoch": 4.680102367242482,
      "grad_norm": 1.2117689586432936,
      "learning_rate": 1.1516127788516098e-05,
      "loss": 0.1327,
      "step": 7315
    },
    {
      "epoch": 4.680742162507998,
      "grad_norm": 1.0899925265210924,
      "learning_rate": 1.1514079556169514e-05,
      "loss": 0.1176,
      "step": 7316
    },
    {
      "epoch": 4.681381957773512,
      "grad_norm": 1.2479464998722005,
      "learning_rate": 1.1512031258811045e-05,
      "loss": 0.1346,
      "step": 7317
    },
    {
      "epoch": 4.682021753039027,
      "grad_norm": 1.2768157707694288,
      "learning_rate": 1.150998289652864e-05,
      "loss": 0.1425,
      "step": 7318
    },
    {
      "epoch": 4.682661548304543,
      "grad_norm": 1.3135676717347093,
      "learning_rate": 1.1507934469410259e-05,
      "loss": 0.149,
      "step": 7319
    },
    {
      "epoch": 4.683301343570058,
      "grad_norm": 1.3344771138230656,
      "learning_rate": 1.1505885977543853e-05,
      "loss": 0.1307,
      "step": 7320
    },
    {
      "epoch": 4.683941138835572,
      "grad_norm": 1.3508459551876666,
      "learning_rate": 1.150383742101738e-05,
      "loss": 0.1384,
      "step": 7321
    },
    {
      "epoch": 4.684580934101088,
      "grad_norm": 1.328380327393258,
      "learning_rate": 1.1501788799918807e-05,
      "loss": 0.147,
      "step": 7322
    },
    {
      "epoch": 4.685220729366603,
      "grad_norm": 1.2359160912984368,
      "learning_rate": 1.1499740114336092e-05,
      "loss": 0.1384,
      "step": 7323
    },
    {
      "epoch": 4.685860524632118,
      "grad_norm": 1.2701540857672058,
      "learning_rate": 1.1497691364357206e-05,
      "loss": 0.1405,
      "step": 7324
    },
    {
      "epoch": 4.6865003198976325,
      "grad_norm": 1.4023924312463878,
      "learning_rate": 1.1495642550070118e-05,
      "loss": 0.1625,
      "step": 7325
    },
    {
      "epoch": 4.687140115163148,
      "grad_norm": 1.3153429910221188,
      "learning_rate": 1.1493593671562797e-05,
      "loss": 0.1452,
      "step": 7326
    },
    {
      "epoch": 4.687779910428663,
      "grad_norm": 1.4221874870302187,
      "learning_rate": 1.1491544728923222e-05,
      "loss": 0.1537,
      "step": 7327
    },
    {
      "epoch": 4.688419705694178,
      "grad_norm": 1.2736808859713196,
      "learning_rate": 1.148949572223937e-05,
      "loss": 0.1469,
      "step": 7328
    },
    {
      "epoch": 4.689059500959693,
      "grad_norm": 1.2174512926766445,
      "learning_rate": 1.148744665159922e-05,
      "loss": 0.1353,
      "step": 7329
    },
    {
      "epoch": 4.689699296225208,
      "grad_norm": 1.227119307872872,
      "learning_rate": 1.1485397517090757e-05,
      "loss": 0.1384,
      "step": 7330
    },
    {
      "epoch": 4.690339091490723,
      "grad_norm": 1.0881225991060357,
      "learning_rate": 1.1483348318801968e-05,
      "loss": 0.1255,
      "step": 7331
    },
    {
      "epoch": 4.690978886756238,
      "grad_norm": 1.1963566763174358,
      "learning_rate": 1.148129905682084e-05,
      "loss": 0.1342,
      "step": 7332
    },
    {
      "epoch": 4.691618682021753,
      "grad_norm": 1.326058304178709,
      "learning_rate": 1.1479249731235365e-05,
      "loss": 0.1471,
      "step": 7333
    },
    {
      "epoch": 4.692258477287268,
      "grad_norm": 1.188485505853196,
      "learning_rate": 1.1477200342133537e-05,
      "loss": 0.1379,
      "step": 7334
    },
    {
      "epoch": 4.6928982725527835,
      "grad_norm": 1.1809706809387674,
      "learning_rate": 1.1475150889603358e-05,
      "loss": 0.1366,
      "step": 7335
    },
    {
      "epoch": 4.693538067818298,
      "grad_norm": 1.1874532994921136,
      "learning_rate": 1.1473101373732818e-05,
      "loss": 0.1367,
      "step": 7336
    },
    {
      "epoch": 4.694177863083813,
      "grad_norm": 1.2917517434096097,
      "learning_rate": 1.1471051794609931e-05,
      "loss": 0.1386,
      "step": 7337
    },
    {
      "epoch": 4.694817658349328,
      "grad_norm": 1.3224176249297335,
      "learning_rate": 1.1469002152322692e-05,
      "loss": 0.1467,
      "step": 7338
    },
    {
      "epoch": 4.695457453614843,
      "grad_norm": 1.1902566239073584,
      "learning_rate": 1.146695244695912e-05,
      "loss": 0.1258,
      "step": 7339
    },
    {
      "epoch": 4.696097248880358,
      "grad_norm": 1.4080755985244273,
      "learning_rate": 1.1464902678607215e-05,
      "loss": 0.1537,
      "step": 7340
    },
    {
      "epoch": 4.696737044145873,
      "grad_norm": 1.25015309877153,
      "learning_rate": 1.1462852847354995e-05,
      "loss": 0.1333,
      "step": 7341
    },
    {
      "epoch": 4.697376839411389,
      "grad_norm": 1.3274690118306662,
      "learning_rate": 1.1460802953290475e-05,
      "loss": 0.1289,
      "step": 7342
    },
    {
      "epoch": 4.698016634676903,
      "grad_norm": 1.069020422152222,
      "learning_rate": 1.1458752996501677e-05,
      "loss": 0.1247,
      "step": 7343
    },
    {
      "epoch": 4.698656429942418,
      "grad_norm": 1.3281596253113557,
      "learning_rate": 1.1456702977076618e-05,
      "loss": 0.1456,
      "step": 7344
    },
    {
      "epoch": 4.6992962252079336,
      "grad_norm": 1.3801752612803722,
      "learning_rate": 1.1454652895103327e-05,
      "loss": 0.1512,
      "step": 7345
    },
    {
      "epoch": 4.699936020473449,
      "grad_norm": 1.2614184914277184,
      "learning_rate": 1.1452602750669827e-05,
      "loss": 0.1442,
      "step": 7346
    },
    {
      "epoch": 4.700575815738963,
      "grad_norm": 1.2782994623735813,
      "learning_rate": 1.145055254386415e-05,
      "loss": 0.1408,
      "step": 7347
    },
    {
      "epoch": 4.7012156110044785,
      "grad_norm": 1.3383264587131842,
      "learning_rate": 1.1448502274774325e-05,
      "loss": 0.1449,
      "step": 7348
    },
    {
      "epoch": 4.701855406269994,
      "grad_norm": 1.2251385080642063,
      "learning_rate": 1.1446451943488393e-05,
      "loss": 0.1438,
      "step": 7349
    },
    {
      "epoch": 4.702495201535509,
      "grad_norm": 1.2938910990332946,
      "learning_rate": 1.1444401550094381e-05,
      "loss": 0.1427,
      "step": 7350
    },
    {
      "epoch": 4.703134996801023,
      "grad_norm": 1.3137391826927816,
      "learning_rate": 1.144235109468034e-05,
      "loss": 0.1424,
      "step": 7351
    },
    {
      "epoch": 4.703774792066539,
      "grad_norm": 1.2671695730994785,
      "learning_rate": 1.1440300577334309e-05,
      "loss": 0.1507,
      "step": 7352
    },
    {
      "epoch": 4.704414587332054,
      "grad_norm": 1.327750881822798,
      "learning_rate": 1.143824999814433e-05,
      "loss": 0.1461,
      "step": 7353
    },
    {
      "epoch": 4.705054382597568,
      "grad_norm": 1.184487862852702,
      "learning_rate": 1.143619935719846e-05,
      "loss": 0.1398,
      "step": 7354
    },
    {
      "epoch": 4.705694177863084,
      "grad_norm": 1.1013029837275083,
      "learning_rate": 1.1434148654584742e-05,
      "loss": 0.1225,
      "step": 7355
    },
    {
      "epoch": 4.706333973128599,
      "grad_norm": 1.2982668659091925,
      "learning_rate": 1.1432097890391231e-05,
      "loss": 0.1499,
      "step": 7356
    },
    {
      "epoch": 4.706973768394114,
      "grad_norm": 1.247216426671493,
      "learning_rate": 1.1430047064705984e-05,
      "loss": 0.1364,
      "step": 7357
    },
    {
      "epoch": 4.707613563659629,
      "grad_norm": 1.2477025945294977,
      "learning_rate": 1.142799617761706e-05,
      "loss": 0.1426,
      "step": 7358
    },
    {
      "epoch": 4.708253358925144,
      "grad_norm": 1.227937351411846,
      "learning_rate": 1.1425945229212523e-05,
      "loss": 0.13,
      "step": 7359
    },
    {
      "epoch": 4.708893154190659,
      "grad_norm": 1.3301278139019872,
      "learning_rate": 1.1423894219580431e-05,
      "loss": 0.139,
      "step": 7360
    },
    {
      "epoch": 4.709532949456174,
      "grad_norm": 1.4090645111178988,
      "learning_rate": 1.1421843148808856e-05,
      "loss": 0.1548,
      "step": 7361
    },
    {
      "epoch": 4.710172744721689,
      "grad_norm": 1.292734881443428,
      "learning_rate": 1.1419792016985867e-05,
      "loss": 0.1512,
      "step": 7362
    },
    {
      "epoch": 4.710812539987204,
      "grad_norm": 1.256058800213838,
      "learning_rate": 1.141774082419953e-05,
      "loss": 0.1339,
      "step": 7363
    },
    {
      "epoch": 4.711452335252719,
      "grad_norm": 1.241613921961653,
      "learning_rate": 1.1415689570537927e-05,
      "loss": 0.1266,
      "step": 7364
    },
    {
      "epoch": 4.712092130518235,
      "grad_norm": 1.30386765974784,
      "learning_rate": 1.141363825608913e-05,
      "loss": 0.1453,
      "step": 7365
    },
    {
      "epoch": 4.712731925783749,
      "grad_norm": 1.2459457634575493,
      "learning_rate": 1.141158688094122e-05,
      "loss": 0.1483,
      "step": 7366
    },
    {
      "epoch": 4.713371721049264,
      "grad_norm": 1.3462991839886063,
      "learning_rate": 1.1409535445182284e-05,
      "loss": 0.1464,
      "step": 7367
    },
    {
      "epoch": 4.7140115163147795,
      "grad_norm": 1.274834204465409,
      "learning_rate": 1.1407483948900401e-05,
      "loss": 0.1479,
      "step": 7368
    },
    {
      "epoch": 4.714651311580294,
      "grad_norm": 1.2384619569240876,
      "learning_rate": 1.1405432392183665e-05,
      "loss": 0.1426,
      "step": 7369
    },
    {
      "epoch": 4.715291106845809,
      "grad_norm": 1.174141259299779,
      "learning_rate": 1.1403380775120159e-05,
      "loss": 0.1375,
      "step": 7370
    },
    {
      "epoch": 4.7159309021113245,
      "grad_norm": 1.2426857025977125,
      "learning_rate": 1.1401329097797977e-05,
      "loss": 0.127,
      "step": 7371
    },
    {
      "epoch": 4.71657069737684,
      "grad_norm": 1.4588376472596294,
      "learning_rate": 1.139927736030522e-05,
      "loss": 0.1584,
      "step": 7372
    },
    {
      "epoch": 4.717210492642354,
      "grad_norm": 1.1541114739218317,
      "learning_rate": 1.1397225562729979e-05,
      "loss": 0.1247,
      "step": 7373
    },
    {
      "epoch": 4.717850287907869,
      "grad_norm": 1.3534002220057748,
      "learning_rate": 1.139517370516036e-05,
      "loss": 0.1472,
      "step": 7374
    },
    {
      "epoch": 4.718490083173385,
      "grad_norm": 1.3715404797823183,
      "learning_rate": 1.1393121787684463e-05,
      "loss": 0.1488,
      "step": 7375
    },
    {
      "epoch": 4.719129878438899,
      "grad_norm": 1.1305160335667983,
      "learning_rate": 1.1391069810390395e-05,
      "loss": 0.1284,
      "step": 7376
    },
    {
      "epoch": 4.719769673704414,
      "grad_norm": 1.3747853359325204,
      "learning_rate": 1.1389017773366267e-05,
      "loss": 0.1626,
      "step": 7377
    },
    {
      "epoch": 4.72040946896993,
      "grad_norm": 1.2655335095997733,
      "learning_rate": 1.1386965676700182e-05,
      "loss": 0.1422,
      "step": 7378
    },
    {
      "epoch": 4.721049264235445,
      "grad_norm": 1.400008709767201,
      "learning_rate": 1.1384913520480261e-05,
      "loss": 0.1493,
      "step": 7379
    },
    {
      "epoch": 4.72168905950096,
      "grad_norm": 1.2913543578196869,
      "learning_rate": 1.1382861304794615e-05,
      "loss": 0.1446,
      "step": 7380
    },
    {
      "epoch": 4.722328854766475,
      "grad_norm": 1.1824117458103462,
      "learning_rate": 1.1380809029731364e-05,
      "loss": 0.1396,
      "step": 7381
    },
    {
      "epoch": 4.72296865003199,
      "grad_norm": 1.304927235395577,
      "learning_rate": 1.137875669537863e-05,
      "loss": 0.1573,
      "step": 7382
    },
    {
      "epoch": 4.723608445297505,
      "grad_norm": 1.3815546769194902,
      "learning_rate": 1.1376704301824534e-05,
      "loss": 0.1347,
      "step": 7383
    },
    {
      "epoch": 4.7242482405630195,
      "grad_norm": 1.250398243045889,
      "learning_rate": 1.137465184915721e-05,
      "loss": 0.1457,
      "step": 7384
    },
    {
      "epoch": 4.724888035828535,
      "grad_norm": 1.2903620242976324,
      "learning_rate": 1.1372599337464778e-05,
      "loss": 0.1431,
      "step": 7385
    },
    {
      "epoch": 4.72552783109405,
      "grad_norm": 1.3171574416911251,
      "learning_rate": 1.1370546766835371e-05,
      "loss": 0.136,
      "step": 7386
    },
    {
      "epoch": 4.726167626359565,
      "grad_norm": 1.3679492994370817,
      "learning_rate": 1.1368494137357126e-05,
      "loss": 0.1427,
      "step": 7387
    },
    {
      "epoch": 4.72680742162508,
      "grad_norm": 1.3683493560492936,
      "learning_rate": 1.1366441449118176e-05,
      "loss": 0.1405,
      "step": 7388
    },
    {
      "epoch": 4.727447216890595,
      "grad_norm": 1.2424313698103737,
      "learning_rate": 1.136438870220666e-05,
      "loss": 0.1399,
      "step": 7389
    },
    {
      "epoch": 4.72808701215611,
      "grad_norm": 1.2355567999273263,
      "learning_rate": 1.1362335896710722e-05,
      "loss": 0.1424,
      "step": 7390
    },
    {
      "epoch": 4.728726807421625,
      "grad_norm": 1.2240873929160918,
      "learning_rate": 1.1360283032718501e-05,
      "loss": 0.1394,
      "step": 7391
    },
    {
      "epoch": 4.72936660268714,
      "grad_norm": 1.2756343823960492,
      "learning_rate": 1.1358230110318152e-05,
      "loss": 0.1316,
      "step": 7392
    },
    {
      "epoch": 4.730006397952655,
      "grad_norm": 1.1755773350318879,
      "learning_rate": 1.1356177129597811e-05,
      "loss": 0.1318,
      "step": 7393
    },
    {
      "epoch": 4.7306461932181705,
      "grad_norm": 1.286094955955093,
      "learning_rate": 1.1354124090645642e-05,
      "loss": 0.1511,
      "step": 7394
    },
    {
      "epoch": 4.731285988483685,
      "grad_norm": 1.3633295585272132,
      "learning_rate": 1.1352070993549792e-05,
      "loss": 0.1502,
      "step": 7395
    },
    {
      "epoch": 4.7319257837492,
      "grad_norm": 1.1886544537494161,
      "learning_rate": 1.1350017838398415e-05,
      "loss": 0.13,
      "step": 7396
    },
    {
      "epoch": 4.732565579014715,
      "grad_norm": 1.2025613407856073,
      "learning_rate": 1.1347964625279678e-05,
      "loss": 0.1481,
      "step": 7397
    },
    {
      "epoch": 4.733205374280231,
      "grad_norm": 1.1997586925935964,
      "learning_rate": 1.1345911354281736e-05,
      "loss": 0.1377,
      "step": 7398
    },
    {
      "epoch": 4.733845169545745,
      "grad_norm": 1.132650893872564,
      "learning_rate": 1.1343858025492756e-05,
      "loss": 0.1318,
      "step": 7399
    },
    {
      "epoch": 4.73448496481126,
      "grad_norm": 1.2152333721048971,
      "learning_rate": 1.1341804639000901e-05,
      "loss": 0.133,
      "step": 7400
    },
    {
      "epoch": 4.735124760076776,
      "grad_norm": 1.1803565972283174,
      "learning_rate": 1.1339751194894341e-05,
      "loss": 0.1308,
      "step": 7401
    },
    {
      "epoch": 4.735764555342291,
      "grad_norm": 1.2227566257906488,
      "learning_rate": 1.1337697693261251e-05,
      "loss": 0.1389,
      "step": 7402
    },
    {
      "epoch": 4.736404350607805,
      "grad_norm": 1.3844382060753104,
      "learning_rate": 1.1335644134189801e-05,
      "loss": 0.1499,
      "step": 7403
    },
    {
      "epoch": 4.737044145873321,
      "grad_norm": 1.2635715130149783,
      "learning_rate": 1.1333590517768168e-05,
      "loss": 0.129,
      "step": 7404
    },
    {
      "epoch": 4.737683941138836,
      "grad_norm": 1.0903004075893081,
      "learning_rate": 1.1331536844084527e-05,
      "loss": 0.1322,
      "step": 7405
    },
    {
      "epoch": 4.73832373640435,
      "grad_norm": 1.214862085360307,
      "learning_rate": 1.1329483113227067e-05,
      "loss": 0.1321,
      "step": 7406
    },
    {
      "epoch": 4.7389635316698655,
      "grad_norm": 1.235077333106939,
      "learning_rate": 1.1327429325283966e-05,
      "loss": 0.1391,
      "step": 7407
    },
    {
      "epoch": 4.739603326935381,
      "grad_norm": 1.2226202968093087,
      "learning_rate": 1.1325375480343407e-05,
      "loss": 0.1305,
      "step": 7408
    },
    {
      "epoch": 4.740243122200896,
      "grad_norm": 1.2762641000237331,
      "learning_rate": 1.1323321578493589e-05,
      "loss": 0.1446,
      "step": 7409
    },
    {
      "epoch": 4.74088291746641,
      "grad_norm": 1.4088994342227792,
      "learning_rate": 1.132126761982269e-05,
      "loss": 0.1559,
      "step": 7410
    },
    {
      "epoch": 4.741522712731926,
      "grad_norm": 1.3282291985976042,
      "learning_rate": 1.1319213604418913e-05,
      "loss": 0.137,
      "step": 7411
    },
    {
      "epoch": 4.742162507997441,
      "grad_norm": 1.3668422622138696,
      "learning_rate": 1.1317159532370452e-05,
      "loss": 0.1512,
      "step": 7412
    },
    {
      "epoch": 4.742802303262955,
      "grad_norm": 1.228482735857885,
      "learning_rate": 1.13151054037655e-05,
      "loss": 0.1489,
      "step": 7413
    },
    {
      "epoch": 4.743442098528471,
      "grad_norm": 1.2583808373783278,
      "learning_rate": 1.1313051218692266e-05,
      "loss": 0.1452,
      "step": 7414
    },
    {
      "epoch": 4.744081893793986,
      "grad_norm": 1.3533885993255235,
      "learning_rate": 1.1310996977238946e-05,
      "loss": 0.1483,
      "step": 7415
    },
    {
      "epoch": 4.744721689059501,
      "grad_norm": 1.2853419754228774,
      "learning_rate": 1.1308942679493746e-05,
      "loss": 0.1369,
      "step": 7416
    },
    {
      "epoch": 4.7453614843250165,
      "grad_norm": 1.297127538743022,
      "learning_rate": 1.1306888325544878e-05,
      "loss": 0.1377,
      "step": 7417
    },
    {
      "epoch": 4.746001279590531,
      "grad_norm": 1.3413733793318763,
      "learning_rate": 1.1304833915480549e-05,
      "loss": 0.1462,
      "step": 7418
    },
    {
      "epoch": 4.746641074856046,
      "grad_norm": 1.448661339061422,
      "learning_rate": 1.1302779449388974e-05,
      "loss": 0.1617,
      "step": 7419
    },
    {
      "epoch": 4.747280870121561,
      "grad_norm": 1.2311416837205533,
      "learning_rate": 1.1300724927358366e-05,
      "loss": 0.1383,
      "step": 7420
    },
    {
      "epoch": 4.747920665387076,
      "grad_norm": 1.2215388916536414,
      "learning_rate": 1.1298670349476943e-05,
      "loss": 0.1388,
      "step": 7421
    },
    {
      "epoch": 4.748560460652591,
      "grad_norm": 1.3547924884420206,
      "learning_rate": 1.1296615715832928e-05,
      "loss": 0.1486,
      "step": 7422
    },
    {
      "epoch": 4.749200255918106,
      "grad_norm": 1.1986657623809895,
      "learning_rate": 1.1294561026514539e-05,
      "loss": 0.1319,
      "step": 7423
    },
    {
      "epoch": 4.749840051183622,
      "grad_norm": 1.158634117098422,
      "learning_rate": 1.1292506281610004e-05,
      "loss": 0.141,
      "step": 7424
    },
    {
      "epoch": 4.750479846449136,
      "grad_norm": 1.2307544705417233,
      "learning_rate": 1.1290451481207543e-05,
      "loss": 0.1413,
      "step": 7425
    },
    {
      "epoch": 4.751119641714651,
      "grad_norm": 1.3230198617628737,
      "learning_rate": 1.1288396625395397e-05,
      "loss": 0.1515,
      "step": 7426
    },
    {
      "epoch": 4.751759436980167,
      "grad_norm": 1.2484799557882762,
      "learning_rate": 1.1286341714261786e-05,
      "loss": 0.1274,
      "step": 7427
    },
    {
      "epoch": 4.752399232245681,
      "grad_norm": 1.3321804109976243,
      "learning_rate": 1.1284286747894952e-05,
      "loss": 0.1444,
      "step": 7428
    },
    {
      "epoch": 4.753039027511196,
      "grad_norm": 1.2538984919751006,
      "learning_rate": 1.1282231726383133e-05,
      "loss": 0.1246,
      "step": 7429
    },
    {
      "epoch": 4.7536788227767115,
      "grad_norm": 1.2787961856295695,
      "learning_rate": 1.1280176649814565e-05,
      "loss": 0.137,
      "step": 7430
    },
    {
      "epoch": 4.754318618042227,
      "grad_norm": 1.2980294282761429,
      "learning_rate": 1.1278121518277487e-05,
      "loss": 0.1444,
      "step": 7431
    },
    {
      "epoch": 4.754958413307741,
      "grad_norm": 1.412315366838914,
      "learning_rate": 1.1276066331860142e-05,
      "loss": 0.1604,
      "step": 7432
    },
    {
      "epoch": 4.755598208573256,
      "grad_norm": 1.3077977363419866,
      "learning_rate": 1.1274011090650778e-05,
      "loss": 0.1447,
      "step": 7433
    },
    {
      "epoch": 4.756238003838772,
      "grad_norm": 1.2227918382943932,
      "learning_rate": 1.127195579473765e-05,
      "loss": 0.1433,
      "step": 7434
    },
    {
      "epoch": 4.756877799104287,
      "grad_norm": 1.1469904800236619,
      "learning_rate": 1.1269900444208996e-05,
      "loss": 0.1304,
      "step": 7435
    },
    {
      "epoch": 4.757517594369801,
      "grad_norm": 1.2721709433702433,
      "learning_rate": 1.1267845039153082e-05,
      "loss": 0.1424,
      "step": 7436
    },
    {
      "epoch": 4.758157389635317,
      "grad_norm": 1.2106723912546427,
      "learning_rate": 1.1265789579658156e-05,
      "loss": 0.1324,
      "step": 7437
    },
    {
      "epoch": 4.758797184900832,
      "grad_norm": 1.1187384427344127,
      "learning_rate": 1.1263734065812474e-05,
      "loss": 0.1243,
      "step": 7438
    },
    {
      "epoch": 4.759436980166347,
      "grad_norm": 1.193259817903358,
      "learning_rate": 1.1261678497704302e-05,
      "loss": 0.137,
      "step": 7439
    },
    {
      "epoch": 4.760076775431862,
      "grad_norm": 1.1692201352736387,
      "learning_rate": 1.1259622875421898e-05,
      "loss": 0.1246,
      "step": 7440
    },
    {
      "epoch": 4.760716570697377,
      "grad_norm": 1.2312068767242426,
      "learning_rate": 1.1257567199053529e-05,
      "loss": 0.1394,
      "step": 7441
    },
    {
      "epoch": 4.761356365962892,
      "grad_norm": 1.3133849493880805,
      "learning_rate": 1.1255511468687461e-05,
      "loss": 0.1429,
      "step": 7442
    },
    {
      "epoch": 4.7619961612284065,
      "grad_norm": 1.4024384513456876,
      "learning_rate": 1.1253455684411964e-05,
      "loss": 0.1427,
      "step": 7443
    },
    {
      "epoch": 4.762635956493922,
      "grad_norm": 1.2391638328557126,
      "learning_rate": 1.1251399846315311e-05,
      "loss": 0.1421,
      "step": 7444
    },
    {
      "epoch": 4.763275751759437,
      "grad_norm": 1.429647812748148,
      "learning_rate": 1.1249343954485777e-05,
      "loss": 0.1599,
      "step": 7445
    },
    {
      "epoch": 4.763915547024952,
      "grad_norm": 1.4041367204940562,
      "learning_rate": 1.1247288009011631e-05,
      "loss": 0.1479,
      "step": 7446
    },
    {
      "epoch": 4.764555342290467,
      "grad_norm": 1.317840436925923,
      "learning_rate": 1.124523200998116e-05,
      "loss": 0.1371,
      "step": 7447
    },
    {
      "epoch": 4.765195137555982,
      "grad_norm": 1.2237308029219178,
      "learning_rate": 1.1243175957482638e-05,
      "loss": 0.1405,
      "step": 7448
    },
    {
      "epoch": 4.765834932821497,
      "grad_norm": 1.1056928674539053,
      "learning_rate": 1.1241119851604355e-05,
      "loss": 0.1266,
      "step": 7449
    },
    {
      "epoch": 4.766474728087012,
      "grad_norm": 1.188260562276421,
      "learning_rate": 1.123906369243459e-05,
      "loss": 0.1355,
      "step": 7450
    },
    {
      "epoch": 4.767114523352527,
      "grad_norm": 1.2911473615985296,
      "learning_rate": 1.1237007480061639e-05,
      "loss": 0.1513,
      "step": 7451
    },
    {
      "epoch": 4.767754318618042,
      "grad_norm": 1.2922846337994367,
      "learning_rate": 1.1234951214573785e-05,
      "loss": 0.1459,
      "step": 7452
    },
    {
      "epoch": 4.7683941138835575,
      "grad_norm": 1.2551242506144915,
      "learning_rate": 1.1232894896059322e-05,
      "loss": 0.1388,
      "step": 7453
    },
    {
      "epoch": 4.769033909149073,
      "grad_norm": 1.390522681835493,
      "learning_rate": 1.1230838524606547e-05,
      "loss": 0.1601,
      "step": 7454
    },
    {
      "epoch": 4.769673704414587,
      "grad_norm": 1.1938728378885581,
      "learning_rate": 1.1228782100303752e-05,
      "loss": 0.1343,
      "step": 7455
    },
    {
      "epoch": 4.770313499680102,
      "grad_norm": 1.1881252918463958,
      "learning_rate": 1.1226725623239244e-05,
      "loss": 0.136,
      "step": 7456
    },
    {
      "epoch": 4.770953294945618,
      "grad_norm": 1.4029032336522425,
      "learning_rate": 1.1224669093501315e-05,
      "loss": 0.1632,
      "step": 7457
    },
    {
      "epoch": 4.771593090211132,
      "grad_norm": 1.3722468952370326,
      "learning_rate": 1.1222612511178274e-05,
      "loss": 0.1514,
      "step": 7458
    },
    {
      "epoch": 4.772232885476647,
      "grad_norm": 1.2379725809619004,
      "learning_rate": 1.122055587635843e-05,
      "loss": 0.1306,
      "step": 7459
    },
    {
      "epoch": 4.772872680742163,
      "grad_norm": 1.1381005986819304,
      "learning_rate": 1.1218499189130087e-05,
      "loss": 0.1295,
      "step": 7460
    },
    {
      "epoch": 4.773512476007678,
      "grad_norm": 1.3407227158406276,
      "learning_rate": 1.1216442449581558e-05,
      "loss": 0.1456,
      "step": 7461
    },
    {
      "epoch": 4.774152271273192,
      "grad_norm": 1.2615331284662816,
      "learning_rate": 1.121438565780115e-05,
      "loss": 0.1452,
      "step": 7462
    },
    {
      "epoch": 4.774792066538708,
      "grad_norm": 1.2080031536390357,
      "learning_rate": 1.1212328813877185e-05,
      "loss": 0.1366,
      "step": 7463
    },
    {
      "epoch": 4.775431861804223,
      "grad_norm": 1.3197957268454994,
      "learning_rate": 1.1210271917897977e-05,
      "loss": 0.15,
      "step": 7464
    },
    {
      "epoch": 4.776071657069737,
      "grad_norm": 1.4372503152660963,
      "learning_rate": 1.1208214969951843e-05,
      "loss": 0.1473,
      "step": 7465
    },
    {
      "epoch": 4.7767114523352525,
      "grad_norm": 1.3156615460916476,
      "learning_rate": 1.120615797012711e-05,
      "loss": 0.1441,
      "step": 7466
    },
    {
      "epoch": 4.777351247600768,
      "grad_norm": 1.2300454482368575,
      "learning_rate": 1.12041009185121e-05,
      "loss": 0.1332,
      "step": 7467
    },
    {
      "epoch": 4.777991042866283,
      "grad_norm": 1.1930978439482323,
      "learning_rate": 1.1202043815195137e-05,
      "loss": 0.1357,
      "step": 7468
    },
    {
      "epoch": 4.7786308381317975,
      "grad_norm": 1.3649624722414697,
      "learning_rate": 1.1199986660264552e-05,
      "loss": 0.1502,
      "step": 7469
    },
    {
      "epoch": 4.779270633397313,
      "grad_norm": 1.2104266994746837,
      "learning_rate": 1.1197929453808672e-05,
      "loss": 0.1339,
      "step": 7470
    },
    {
      "epoch": 4.779910428662828,
      "grad_norm": 1.2478887553964868,
      "learning_rate": 1.1195872195915836e-05,
      "loss": 0.1338,
      "step": 7471
    },
    {
      "epoch": 4.780550223928343,
      "grad_norm": 1.2904534273983375,
      "learning_rate": 1.119381488667437e-05,
      "loss": 0.1384,
      "step": 7472
    },
    {
      "epoch": 4.781190019193858,
      "grad_norm": 1.1500565151047304,
      "learning_rate": 1.1191757526172619e-05,
      "loss": 0.1232,
      "step": 7473
    },
    {
      "epoch": 4.781829814459373,
      "grad_norm": 1.3392389374300757,
      "learning_rate": 1.1189700114498923e-05,
      "loss": 0.1373,
      "step": 7474
    },
    {
      "epoch": 4.782469609724888,
      "grad_norm": 1.2183987846353597,
      "learning_rate": 1.1187642651741616e-05,
      "loss": 0.1244,
      "step": 7475
    },
    {
      "epoch": 4.7831094049904035,
      "grad_norm": 1.3219479707987425,
      "learning_rate": 1.118558513798905e-05,
      "loss": 0.1427,
      "step": 7476
    },
    {
      "epoch": 4.783749200255918,
      "grad_norm": 1.3350071666456096,
      "learning_rate": 1.1183527573329564e-05,
      "loss": 0.136,
      "step": 7477
    },
    {
      "epoch": 4.784388995521433,
      "grad_norm": 1.306021659363478,
      "learning_rate": 1.118146995785151e-05,
      "loss": 0.1369,
      "step": 7478
    },
    {
      "epoch": 4.785028790786948,
      "grad_norm": 1.167020456765086,
      "learning_rate": 1.1179412291643237e-05,
      "loss": 0.1265,
      "step": 7479
    },
    {
      "epoch": 4.785668586052463,
      "grad_norm": 1.3781968240718188,
      "learning_rate": 1.11773545747931e-05,
      "loss": 0.1515,
      "step": 7480
    },
    {
      "epoch": 4.786308381317978,
      "grad_norm": 1.2418496501329788,
      "learning_rate": 1.1175296807389451e-05,
      "loss": 0.1329,
      "step": 7481
    },
    {
      "epoch": 4.786948176583493,
      "grad_norm": 1.5028822478567225,
      "learning_rate": 1.1173238989520649e-05,
      "loss": 0.1308,
      "step": 7482
    },
    {
      "epoch": 4.787587971849009,
      "grad_norm": 1.3854787895447387,
      "learning_rate": 1.117118112127505e-05,
      "loss": 0.1596,
      "step": 7483
    },
    {
      "epoch": 4.788227767114523,
      "grad_norm": 1.1982778010610458,
      "learning_rate": 1.116912320274102e-05,
      "loss": 0.1324,
      "step": 7484
    },
    {
      "epoch": 4.788867562380038,
      "grad_norm": 1.4761748040186606,
      "learning_rate": 1.1167065234006915e-05,
      "loss": 0.151,
      "step": 7485
    },
    {
      "epoch": 4.789507357645554,
      "grad_norm": 1.1905560675202633,
      "learning_rate": 1.1165007215161108e-05,
      "loss": 0.139,
      "step": 7486
    },
    {
      "epoch": 4.790147152911068,
      "grad_norm": 1.4117389366677082,
      "learning_rate": 1.1162949146291962e-05,
      "loss": 0.1561,
      "step": 7487
    },
    {
      "epoch": 4.790786948176583,
      "grad_norm": 1.405097789482129,
      "learning_rate": 1.1160891027487846e-05,
      "loss": 0.1676,
      "step": 7488
    },
    {
      "epoch": 4.7914267434420985,
      "grad_norm": 1.4546348165509368,
      "learning_rate": 1.1158832858837139e-05,
      "loss": 0.1629,
      "step": 7489
    },
    {
      "epoch": 4.792066538707614,
      "grad_norm": 1.266261601781482,
      "learning_rate": 1.115677464042821e-05,
      "loss": 0.1332,
      "step": 7490
    },
    {
      "epoch": 4.792706333973129,
      "grad_norm": 1.3162891727124102,
      "learning_rate": 1.1154716372349435e-05,
      "loss": 0.1523,
      "step": 7491
    },
    {
      "epoch": 4.7933461292386434,
      "grad_norm": 1.154955722458843,
      "learning_rate": 1.1152658054689191e-05,
      "loss": 0.1277,
      "step": 7492
    },
    {
      "epoch": 4.793985924504159,
      "grad_norm": 1.2762401326142434,
      "learning_rate": 1.115059968753586e-05,
      "loss": 0.1368,
      "step": 7493
    },
    {
      "epoch": 4.794625719769674,
      "grad_norm": 1.2948527907724576,
      "learning_rate": 1.114854127097783e-05,
      "loss": 0.1503,
      "step": 7494
    },
    {
      "epoch": 4.795265515035188,
      "grad_norm": 1.3004860116172525,
      "learning_rate": 1.1146482805103476e-05,
      "loss": 0.1407,
      "step": 7495
    },
    {
      "epoch": 4.795905310300704,
      "grad_norm": 1.2357519988525003,
      "learning_rate": 1.1144424290001195e-05,
      "loss": 0.1329,
      "step": 7496
    },
    {
      "epoch": 4.796545105566219,
      "grad_norm": 1.1229608850324897,
      "learning_rate": 1.1142365725759367e-05,
      "loss": 0.13,
      "step": 7497
    },
    {
      "epoch": 4.797184900831734,
      "grad_norm": 1.3832552997125591,
      "learning_rate": 1.1140307112466387e-05,
      "loss": 0.151,
      "step": 7498
    },
    {
      "epoch": 4.797824696097249,
      "grad_norm": 1.2130902808475723,
      "learning_rate": 1.1138248450210648e-05,
      "loss": 0.1398,
      "step": 7499
    },
    {
      "epoch": 4.798464491362764,
      "grad_norm": 1.194101339055733,
      "learning_rate": 1.1136189739080545e-05,
      "loss": 0.133,
      "step": 7500
    },
    {
      "epoch": 4.799104286628279,
      "grad_norm": 1.1913894014723159,
      "learning_rate": 1.1134130979164479e-05,
      "loss": 0.1247,
      "step": 7501
    },
    {
      "epoch": 4.7997440818937935,
      "grad_norm": 1.207938305866352,
      "learning_rate": 1.1132072170550843e-05,
      "loss": 0.1328,
      "step": 7502
    },
    {
      "epoch": 4.800383877159309,
      "grad_norm": 1.222960910985384,
      "learning_rate": 1.1130013313328044e-05,
      "loss": 0.1414,
      "step": 7503
    },
    {
      "epoch": 4.801023672424824,
      "grad_norm": 1.2851020202809273,
      "learning_rate": 1.112795440758448e-05,
      "loss": 0.1385,
      "step": 7504
    },
    {
      "epoch": 4.801663467690339,
      "grad_norm": 1.1609867923012631,
      "learning_rate": 1.1125895453408564e-05,
      "loss": 0.1362,
      "step": 7505
    },
    {
      "epoch": 4.802303262955854,
      "grad_norm": 1.2360403350533797,
      "learning_rate": 1.11238364508887e-05,
      "loss": 0.1286,
      "step": 7506
    },
    {
      "epoch": 4.802943058221369,
      "grad_norm": 1.2011682205071041,
      "learning_rate": 1.1121777400113293e-05,
      "loss": 0.1375,
      "step": 7507
    },
    {
      "epoch": 4.803582853486884,
      "grad_norm": 1.2761630913515767,
      "learning_rate": 1.1119718301170766e-05,
      "loss": 0.1445,
      "step": 7508
    },
    {
      "epoch": 4.8042226487524,
      "grad_norm": 1.2373788025347972,
      "learning_rate": 1.1117659154149523e-05,
      "loss": 0.1378,
      "step": 7509
    },
    {
      "epoch": 4.804862444017914,
      "grad_norm": 1.0725580303907538,
      "learning_rate": 1.1115599959137985e-05,
      "loss": 0.1158,
      "step": 7510
    },
    {
      "epoch": 4.805502239283429,
      "grad_norm": 1.3576737177718228,
      "learning_rate": 1.1113540716224569e-05,
      "loss": 0.1464,
      "step": 7511
    },
    {
      "epoch": 4.8061420345489445,
      "grad_norm": 1.2993585044165208,
      "learning_rate": 1.1111481425497697e-05,
      "loss": 0.1398,
      "step": 7512
    },
    {
      "epoch": 4.80678182981446,
      "grad_norm": 1.3558383965483298,
      "learning_rate": 1.1109422087045789e-05,
      "loss": 0.1446,
      "step": 7513
    },
    {
      "epoch": 4.807421625079974,
      "grad_norm": 1.237639906846719,
      "learning_rate": 1.110736270095727e-05,
      "loss": 0.1361,
      "step": 7514
    },
    {
      "epoch": 4.8080614203454894,
      "grad_norm": 1.4604856487503926,
      "learning_rate": 1.1105303267320564e-05,
      "loss": 0.1634,
      "step": 7515
    },
    {
      "epoch": 4.808701215611005,
      "grad_norm": 1.2400956610523985,
      "learning_rate": 1.1103243786224105e-05,
      "loss": 0.1292,
      "step": 7516
    },
    {
      "epoch": 4.809341010876519,
      "grad_norm": 1.2380717924232192,
      "learning_rate": 1.110118425775632e-05,
      "loss": 0.1403,
      "step": 7517
    },
    {
      "epoch": 4.809980806142034,
      "grad_norm": 1.2118352379266448,
      "learning_rate": 1.1099124682005643e-05,
      "loss": 0.1328,
      "step": 7518
    },
    {
      "epoch": 4.81062060140755,
      "grad_norm": 1.202614589849649,
      "learning_rate": 1.1097065059060504e-05,
      "loss": 0.1311,
      "step": 7519
    },
    {
      "epoch": 4.811260396673065,
      "grad_norm": 1.408823935575597,
      "learning_rate": 1.1095005389009346e-05,
      "loss": 0.1555,
      "step": 7520
    },
    {
      "epoch": 4.811900191938579,
      "grad_norm": 1.3508963255578181,
      "learning_rate": 1.1092945671940603e-05,
      "loss": 0.1478,
      "step": 7521
    },
    {
      "epoch": 4.812539987204095,
      "grad_norm": 1.725635763410135,
      "learning_rate": 1.1090885907942717e-05,
      "loss": 0.1385,
      "step": 7522
    },
    {
      "epoch": 4.81317978246961,
      "grad_norm": 1.1784133069230798,
      "learning_rate": 1.108882609710413e-05,
      "loss": 0.133,
      "step": 7523
    },
    {
      "epoch": 4.813819577735125,
      "grad_norm": 1.453101383937968,
      "learning_rate": 1.1086766239513286e-05,
      "loss": 0.1328,
      "step": 7524
    },
    {
      "epoch": 4.8144593730006395,
      "grad_norm": 1.793810553994565,
      "learning_rate": 1.1084706335258634e-05,
      "loss": 0.136,
      "step": 7525
    },
    {
      "epoch": 4.815099168266155,
      "grad_norm": 1.2552827132726356,
      "learning_rate": 1.1082646384428626e-05,
      "loss": 0.1319,
      "step": 7526
    },
    {
      "epoch": 4.81573896353167,
      "grad_norm": 1.614279944586329,
      "learning_rate": 1.1080586387111704e-05,
      "loss": 0.1473,
      "step": 7527
    },
    {
      "epoch": 4.816378758797185,
      "grad_norm": 1.5909840596937392,
      "learning_rate": 1.1078526343396327e-05,
      "loss": 0.1377,
      "step": 7528
    },
    {
      "epoch": 4.8170185540627,
      "grad_norm": 1.87928724429927,
      "learning_rate": 1.1076466253370946e-05,
      "loss": 0.1545,
      "step": 7529
    },
    {
      "epoch": 4.817658349328215,
      "grad_norm": 1.6363344849215986,
      "learning_rate": 1.1074406117124018e-05,
      "loss": 0.1529,
      "step": 7530
    },
    {
      "epoch": 4.81829814459373,
      "grad_norm": 1.1969905613098,
      "learning_rate": 1.1072345934744002e-05,
      "loss": 0.1297,
      "step": 7531
    },
    {
      "epoch": 4.818937939859245,
      "grad_norm": 1.283469315178284,
      "learning_rate": 1.1070285706319359e-05,
      "loss": 0.1486,
      "step": 7532
    },
    {
      "epoch": 4.81957773512476,
      "grad_norm": 1.2869297275086449,
      "learning_rate": 1.1068225431938555e-05,
      "loss": 0.1406,
      "step": 7533
    },
    {
      "epoch": 4.820217530390275,
      "grad_norm": 1.4714047336237295,
      "learning_rate": 1.1066165111690047e-05,
      "loss": 0.1706,
      "step": 7534
    },
    {
      "epoch": 4.8208573256557905,
      "grad_norm": 1.2859240699115528,
      "learning_rate": 1.106410474566231e-05,
      "loss": 0.1399,
      "step": 7535
    },
    {
      "epoch": 4.821497120921305,
      "grad_norm": 1.1357850627305146,
      "learning_rate": 1.1062044333943807e-05,
      "loss": 0.1275,
      "step": 7536
    },
    {
      "epoch": 4.82213691618682,
      "grad_norm": 1.4447033664441187,
      "learning_rate": 1.1059983876623006e-05,
      "loss": 0.1576,
      "step": 7537
    },
    {
      "epoch": 4.822776711452335,
      "grad_norm": 1.3159167895334547,
      "learning_rate": 1.1057923373788387e-05,
      "loss": 0.1521,
      "step": 7538
    },
    {
      "epoch": 4.82341650671785,
      "grad_norm": 1.344943497344735,
      "learning_rate": 1.1055862825528419e-05,
      "loss": 0.1598,
      "step": 7539
    },
    {
      "epoch": 4.824056301983365,
      "grad_norm": 1.198130114457623,
      "learning_rate": 1.1053802231931576e-05,
      "loss": 0.1386,
      "step": 7540
    },
    {
      "epoch": 4.82469609724888,
      "grad_norm": 1.2884775971078444,
      "learning_rate": 1.1051741593086344e-05,
      "loss": 0.1419,
      "step": 7541
    },
    {
      "epoch": 4.825335892514396,
      "grad_norm": 1.2100023941540092,
      "learning_rate": 1.10496809090812e-05,
      "loss": 0.1333,
      "step": 7542
    },
    {
      "epoch": 4.825975687779911,
      "grad_norm": 1.2027651200143459,
      "learning_rate": 1.1047620180004624e-05,
      "loss": 0.1258,
      "step": 7543
    },
    {
      "epoch": 4.826615483045425,
      "grad_norm": 1.2267692223800228,
      "learning_rate": 1.1045559405945096e-05,
      "loss": 0.1333,
      "step": 7544
    },
    {
      "epoch": 4.827255278310941,
      "grad_norm": 1.2034611038589067,
      "learning_rate": 1.104349858699111e-05,
      "loss": 0.1241,
      "step": 7545
    },
    {
      "epoch": 4.827895073576456,
      "grad_norm": 1.325705475192894,
      "learning_rate": 1.1041437723231154e-05,
      "loss": 0.1395,
      "step": 7546
    },
    {
      "epoch": 4.82853486884197,
      "grad_norm": 1.3253600948411277,
      "learning_rate": 1.103937681475371e-05,
      "loss": 0.1455,
      "step": 7547
    },
    {
      "epoch": 4.8291746641074855,
      "grad_norm": 1.2651633636835413,
      "learning_rate": 1.103731586164728e-05,
      "loss": 0.136,
      "step": 7548
    },
    {
      "epoch": 4.829814459373001,
      "grad_norm": 1.287121052346073,
      "learning_rate": 1.1035254864000347e-05,
      "loss": 0.1506,
      "step": 7549
    },
    {
      "epoch": 4.830454254638516,
      "grad_norm": 1.2080865813867587,
      "learning_rate": 1.1033193821901414e-05,
      "loss": 0.1315,
      "step": 7550
    },
    {
      "epoch": 4.8310940499040305,
      "grad_norm": 1.2511778232564201,
      "learning_rate": 1.1031132735438978e-05,
      "loss": 0.1421,
      "step": 7551
    },
    {
      "epoch": 4.831733845169546,
      "grad_norm": 1.2699002081201636,
      "learning_rate": 1.1029071604701534e-05,
      "loss": 0.1534,
      "step": 7552
    },
    {
      "epoch": 4.832373640435061,
      "grad_norm": 1.149108192738389,
      "learning_rate": 1.1027010429777587e-05,
      "loss": 0.1364,
      "step": 7553
    },
    {
      "epoch": 4.833013435700575,
      "grad_norm": 1.238366990511465,
      "learning_rate": 1.1024949210755637e-05,
      "loss": 0.1331,
      "step": 7554
    },
    {
      "epoch": 4.833653230966091,
      "grad_norm": 1.2545799121549681,
      "learning_rate": 1.1022887947724192e-05,
      "loss": 0.1353,
      "step": 7555
    },
    {
      "epoch": 4.834293026231606,
      "grad_norm": 1.1850315410802363,
      "learning_rate": 1.1020826640771758e-05,
      "loss": 0.1295,
      "step": 7556
    },
    {
      "epoch": 4.834932821497121,
      "grad_norm": 1.250421390381327,
      "learning_rate": 1.1018765289986845e-05,
      "loss": 0.138,
      "step": 7557
    },
    {
      "epoch": 4.835572616762636,
      "grad_norm": 1.2510548031255955,
      "learning_rate": 1.1016703895457965e-05,
      "loss": 0.1317,
      "step": 7558
    },
    {
      "epoch": 4.836212412028151,
      "grad_norm": 1.3954225707011014,
      "learning_rate": 1.1014642457273626e-05,
      "loss": 0.1411,
      "step": 7559
    },
    {
      "epoch": 4.836852207293666,
      "grad_norm": 1.3699389350658888,
      "learning_rate": 1.1012580975522342e-05,
      "loss": 0.1523,
      "step": 7560
    },
    {
      "epoch": 4.837492002559181,
      "grad_norm": 1.2702971937671124,
      "learning_rate": 1.1010519450292636e-05,
      "loss": 0.1402,
      "step": 7561
    },
    {
      "epoch": 4.838131797824696,
      "grad_norm": 1.258233402133702,
      "learning_rate": 1.100845788167302e-05,
      "loss": 0.1338,
      "step": 7562
    },
    {
      "epoch": 4.838771593090211,
      "grad_norm": 1.3659875214931954,
      "learning_rate": 1.1006396269752017e-05,
      "loss": 0.1579,
      "step": 7563
    },
    {
      "epoch": 4.839411388355726,
      "grad_norm": 1.2593242507933182,
      "learning_rate": 1.1004334614618149e-05,
      "loss": 0.1396,
      "step": 7564
    },
    {
      "epoch": 4.840051183621242,
      "grad_norm": 1.3517652192596525,
      "learning_rate": 1.1002272916359939e-05,
      "loss": 0.1453,
      "step": 7565
    },
    {
      "epoch": 4.840690978886756,
      "grad_norm": 1.331580863790458,
      "learning_rate": 1.1000211175065915e-05,
      "loss": 0.1496,
      "step": 7566
    },
    {
      "epoch": 4.841330774152271,
      "grad_norm": 1.2577838489308035,
      "learning_rate": 1.09981493908246e-05,
      "loss": 0.1382,
      "step": 7567
    },
    {
      "epoch": 4.841970569417787,
      "grad_norm": 1.2591880951617882,
      "learning_rate": 1.0996087563724527e-05,
      "loss": 0.1311,
      "step": 7568
    },
    {
      "epoch": 4.842610364683301,
      "grad_norm": 1.3042151191235591,
      "learning_rate": 1.0994025693854222e-05,
      "loss": 0.1409,
      "step": 7569
    },
    {
      "epoch": 4.843250159948816,
      "grad_norm": 1.2272589639065958,
      "learning_rate": 1.0991963781302225e-05,
      "loss": 0.143,
      "step": 7570
    },
    {
      "epoch": 4.8438899552143315,
      "grad_norm": 1.2358011800221511,
      "learning_rate": 1.0989901826157067e-05,
      "loss": 0.1366,
      "step": 7571
    },
    {
      "epoch": 4.844529750479847,
      "grad_norm": 1.2699377894394197,
      "learning_rate": 1.098783982850729e-05,
      "loss": 0.1399,
      "step": 7572
    },
    {
      "epoch": 4.845169545745361,
      "grad_norm": 1.2205231353453128,
      "learning_rate": 1.0985777788441423e-05,
      "loss": 0.1354,
      "step": 7573
    },
    {
      "epoch": 4.8458093410108765,
      "grad_norm": 1.3195259571882993,
      "learning_rate": 1.0983715706048012e-05,
      "loss": 0.1331,
      "step": 7574
    },
    {
      "epoch": 4.846449136276392,
      "grad_norm": 1.268007862611401,
      "learning_rate": 1.0981653581415599e-05,
      "loss": 0.1343,
      "step": 7575
    },
    {
      "epoch": 4.847088931541906,
      "grad_norm": 1.3157151284001214,
      "learning_rate": 1.0979591414632729e-05,
      "loss": 0.1459,
      "step": 7576
    },
    {
      "epoch": 4.847728726807421,
      "grad_norm": 1.1643324941539086,
      "learning_rate": 1.0977529205787943e-05,
      "loss": 0.1225,
      "step": 7577
    },
    {
      "epoch": 4.848368522072937,
      "grad_norm": 1.3489736135747428,
      "learning_rate": 1.0975466954969794e-05,
      "loss": 0.1479,
      "step": 7578
    },
    {
      "epoch": 4.849008317338452,
      "grad_norm": 1.3032789077473936,
      "learning_rate": 1.0973404662266826e-05,
      "loss": 0.1411,
      "step": 7579
    },
    {
      "epoch": 4.849648112603967,
      "grad_norm": 1.3257841691468928,
      "learning_rate": 1.0971342327767598e-05,
      "loss": 0.1389,
      "step": 7580
    },
    {
      "epoch": 4.850287907869482,
      "grad_norm": 1.1368996683249493,
      "learning_rate": 1.0969279951560658e-05,
      "loss": 0.1322,
      "step": 7581
    },
    {
      "epoch": 4.850927703134997,
      "grad_norm": 1.2117564567648669,
      "learning_rate": 1.0967217533734558e-05,
      "loss": 0.1353,
      "step": 7582
    },
    {
      "epoch": 4.851567498400512,
      "grad_norm": 1.273470135557477,
      "learning_rate": 1.096515507437786e-05,
      "loss": 0.1396,
      "step": 7583
    },
    {
      "epoch": 4.8522072936660265,
      "grad_norm": 1.2083385464098795,
      "learning_rate": 1.0963092573579117e-05,
      "loss": 0.134,
      "step": 7584
    },
    {
      "epoch": 4.852847088931542,
      "grad_norm": 1.2552186421907092,
      "learning_rate": 1.0961030031426895e-05,
      "loss": 0.1386,
      "step": 7585
    },
    {
      "epoch": 4.853486884197057,
      "grad_norm": 1.3210693525711443,
      "learning_rate": 1.0958967448009752e-05,
      "loss": 0.134,
      "step": 7586
    },
    {
      "epoch": 4.854126679462572,
      "grad_norm": 1.1918574624783669,
      "learning_rate": 1.0956904823416255e-05,
      "loss": 0.1245,
      "step": 7587
    },
    {
      "epoch": 4.854766474728087,
      "grad_norm": 1.2970567007936775,
      "learning_rate": 1.0954842157734964e-05,
      "loss": 0.1458,
      "step": 7588
    },
    {
      "epoch": 4.855406269993602,
      "grad_norm": 1.2870041117431923,
      "learning_rate": 1.0952779451054449e-05,
      "loss": 0.1357,
      "step": 7589
    },
    {
      "epoch": 4.856046065259117,
      "grad_norm": 1.3992572638581307,
      "learning_rate": 1.0950716703463282e-05,
      "loss": 0.1562,
      "step": 7590
    },
    {
      "epoch": 4.856685860524632,
      "grad_norm": 1.379073336988822,
      "learning_rate": 1.0948653915050029e-05,
      "loss": 0.1544,
      "step": 7591
    },
    {
      "epoch": 4.857325655790147,
      "grad_norm": 1.2112066399033226,
      "learning_rate": 1.0946591085903262e-05,
      "loss": 0.131,
      "step": 7592
    },
    {
      "epoch": 4.857965451055662,
      "grad_norm": 1.2826523886799144,
      "learning_rate": 1.0944528216111563e-05,
      "loss": 0.1411,
      "step": 7593
    },
    {
      "epoch": 4.8586052463211775,
      "grad_norm": 1.213340550893876,
      "learning_rate": 1.0942465305763498e-05,
      "loss": 0.1324,
      "step": 7594
    },
    {
      "epoch": 4.859245041586692,
      "grad_norm": 1.3161775657651076,
      "learning_rate": 1.0940402354947652e-05,
      "loss": 0.1418,
      "step": 7595
    },
    {
      "epoch": 4.859884836852207,
      "grad_norm": 1.1398758505250852,
      "learning_rate": 1.0938339363752604e-05,
      "loss": 0.1348,
      "step": 7596
    },
    {
      "epoch": 4.8605246321177225,
      "grad_norm": 1.2685674666032243,
      "learning_rate": 1.0936276332266927e-05,
      "loss": 0.1325,
      "step": 7597
    },
    {
      "epoch": 4.861164427383238,
      "grad_norm": 1.2493073809198634,
      "learning_rate": 1.0934213260579214e-05,
      "loss": 0.14,
      "step": 7598
    },
    {
      "epoch": 4.861804222648752,
      "grad_norm": 1.2964756311044252,
      "learning_rate": 1.0932150148778042e-05,
      "loss": 0.1371,
      "step": 7599
    },
    {
      "epoch": 4.862444017914267,
      "grad_norm": 1.1991038941665275,
      "learning_rate": 1.0930086996952004e-05,
      "loss": 0.1376,
      "step": 7600
    },
    {
      "epoch": 4.863083813179783,
      "grad_norm": 1.2638424511041502,
      "learning_rate": 1.0928023805189683e-05,
      "loss": 0.1381,
      "step": 7601
    },
    {
      "epoch": 4.863723608445298,
      "grad_norm": 1.2423772524459233,
      "learning_rate": 1.0925960573579672e-05,
      "loss": 0.1328,
      "step": 7602
    },
    {
      "epoch": 4.864363403710812,
      "grad_norm": 1.1935651995901126,
      "learning_rate": 1.0923897302210562e-05,
      "loss": 0.126,
      "step": 7603
    },
    {
      "epoch": 4.865003198976328,
      "grad_norm": 1.263103191821304,
      "learning_rate": 1.0921833991170943e-05,
      "loss": 0.1338,
      "step": 7604
    },
    {
      "epoch": 4.865642994241843,
      "grad_norm": 1.3329109768148157,
      "learning_rate": 1.0919770640549415e-05,
      "loss": 0.1416,
      "step": 7605
    },
    {
      "epoch": 4.866282789507357,
      "grad_norm": 1.2781000767776123,
      "learning_rate": 1.0917707250434569e-05,
      "loss": 0.1521,
      "step": 7606
    },
    {
      "epoch": 4.8669225847728725,
      "grad_norm": 1.259681919768377,
      "learning_rate": 1.0915643820915005e-05,
      "loss": 0.1414,
      "step": 7607
    },
    {
      "epoch": 4.867562380038388,
      "grad_norm": 1.3020368870184198,
      "learning_rate": 1.0913580352079328e-05,
      "loss": 0.1374,
      "step": 7608
    },
    {
      "epoch": 4.868202175303903,
      "grad_norm": 1.244925066258494,
      "learning_rate": 1.0911516844016132e-05,
      "loss": 0.1406,
      "step": 7609
    },
    {
      "epoch": 4.8688419705694175,
      "grad_norm": 1.297813609338612,
      "learning_rate": 1.0909453296814027e-05,
      "loss": 0.1476,
      "step": 7610
    },
    {
      "epoch": 4.869481765834933,
      "grad_norm": 1.3030385620436251,
      "learning_rate": 1.0907389710561616e-05,
      "loss": 0.1322,
      "step": 7611
    },
    {
      "epoch": 4.870121561100448,
      "grad_norm": 1.2287912522826436,
      "learning_rate": 1.09053260853475e-05,
      "loss": 0.1336,
      "step": 7612
    },
    {
      "epoch": 4.870761356365962,
      "grad_norm": 1.2333784755857156,
      "learning_rate": 1.0903262421260299e-05,
      "loss": 0.1449,
      "step": 7613
    },
    {
      "epoch": 4.871401151631478,
      "grad_norm": 1.2884030765530539,
      "learning_rate": 1.090119871838861e-05,
      "loss": 0.1469,
      "step": 7614
    },
    {
      "epoch": 4.872040946896993,
      "grad_norm": 1.241721635251224,
      "learning_rate": 1.0899134976821058e-05,
      "loss": 0.1273,
      "step": 7615
    },
    {
      "epoch": 4.872680742162508,
      "grad_norm": 1.1621537279669643,
      "learning_rate": 1.0897071196646242e-05,
      "loss": 0.13,
      "step": 7616
    },
    {
      "epoch": 4.8733205374280235,
      "grad_norm": 1.1804201455039702,
      "learning_rate": 1.0895007377952792e-05,
      "loss": 0.1291,
      "step": 7617
    },
    {
      "epoch": 4.873960332693538,
      "grad_norm": 1.3375397193538798,
      "learning_rate": 1.0892943520829314e-05,
      "loss": 0.1485,
      "step": 7618
    },
    {
      "epoch": 4.874600127959053,
      "grad_norm": 1.2435925693317174,
      "learning_rate": 1.0890879625364429e-05,
      "loss": 0.129,
      "step": 7619
    },
    {
      "epoch": 4.8752399232245685,
      "grad_norm": 1.226804136978485,
      "learning_rate": 1.0888815691646756e-05,
      "loss": 0.1257,
      "step": 7620
    },
    {
      "epoch": 4.875879718490083,
      "grad_norm": 1.3255202833946553,
      "learning_rate": 1.088675171976492e-05,
      "loss": 0.1397,
      "step": 7621
    },
    {
      "epoch": 4.876519513755598,
      "grad_norm": 1.2519548311592656,
      "learning_rate": 1.0884687709807542e-05,
      "loss": 0.1353,
      "step": 7622
    },
    {
      "epoch": 4.877159309021113,
      "grad_norm": 1.099125401787497,
      "learning_rate": 1.088262366186325e-05,
      "loss": 0.1259,
      "step": 7623
    },
    {
      "epoch": 4.877799104286629,
      "grad_norm": 1.188335961148562,
      "learning_rate": 1.0880559576020665e-05,
      "loss": 0.1285,
      "step": 7624
    },
    {
      "epoch": 4.878438899552143,
      "grad_norm": 1.1226508412244884,
      "learning_rate": 1.087849545236842e-05,
      "loss": 0.1186,
      "step": 7625
    },
    {
      "epoch": 4.879078694817658,
      "grad_norm": 1.2719415612405045,
      "learning_rate": 1.0876431290995145e-05,
      "loss": 0.1407,
      "step": 7626
    },
    {
      "epoch": 4.879718490083174,
      "grad_norm": 1.290233229339943,
      "learning_rate": 1.0874367091989464e-05,
      "loss": 0.135,
      "step": 7627
    },
    {
      "epoch": 4.880358285348688,
      "grad_norm": 1.303613818081571,
      "learning_rate": 1.087230285544002e-05,
      "loss": 0.143,
      "step": 7628
    },
    {
      "epoch": 4.880998080614203,
      "grad_norm": 1.3500708515285957,
      "learning_rate": 1.087023858143544e-05,
      "loss": 0.1501,
      "step": 7629
    },
    {
      "epoch": 4.8816378758797185,
      "grad_norm": 1.166978301173851,
      "learning_rate": 1.0868174270064364e-05,
      "loss": 0.1273,
      "step": 7630
    },
    {
      "epoch": 4.882277671145234,
      "grad_norm": 1.204479228553452,
      "learning_rate": 1.086610992141543e-05,
      "loss": 0.1353,
      "step": 7631
    },
    {
      "epoch": 4.882917466410748,
      "grad_norm": 1.2359542822542242,
      "learning_rate": 1.0864045535577277e-05,
      "loss": 0.1377,
      "step": 7632
    },
    {
      "epoch": 4.8835572616762635,
      "grad_norm": 1.2156119491034247,
      "learning_rate": 1.0861981112638546e-05,
      "loss": 0.1274,
      "step": 7633
    },
    {
      "epoch": 4.884197056941779,
      "grad_norm": 1.3010168183143715,
      "learning_rate": 1.0859916652687878e-05,
      "loss": 0.1528,
      "step": 7634
    },
    {
      "epoch": 4.884836852207294,
      "grad_norm": 1.2692304027495978,
      "learning_rate": 1.0857852155813922e-05,
      "loss": 0.1413,
      "step": 7635
    },
    {
      "epoch": 4.885476647472808,
      "grad_norm": 1.214324822908753,
      "learning_rate": 1.0855787622105316e-05,
      "loss": 0.1413,
      "step": 7636
    },
    {
      "epoch": 4.886116442738324,
      "grad_norm": 1.1399049819850626,
      "learning_rate": 1.0853723051650713e-05,
      "loss": 0.1264,
      "step": 7637
    },
    {
      "epoch": 4.886756238003839,
      "grad_norm": 1.2498299363607326,
      "learning_rate": 1.0851658444538762e-05,
      "loss": 0.1481,
      "step": 7638
    },
    {
      "epoch": 4.887396033269354,
      "grad_norm": 1.233493033136847,
      "learning_rate": 1.084959380085811e-05,
      "loss": 0.1419,
      "step": 7639
    },
    {
      "epoch": 4.888035828534869,
      "grad_norm": 1.2955896492779702,
      "learning_rate": 1.0847529120697416e-05,
      "loss": 0.1422,
      "step": 7640
    },
    {
      "epoch": 4.888675623800384,
      "grad_norm": 1.2342399550987444,
      "learning_rate": 1.0845464404145324e-05,
      "loss": 0.1379,
      "step": 7641
    },
    {
      "epoch": 4.889315419065899,
      "grad_norm": 1.367713621517474,
      "learning_rate": 1.0843399651290498e-05,
      "loss": 0.1484,
      "step": 7642
    },
    {
      "epoch": 4.889955214331414,
      "grad_norm": 1.218539700121989,
      "learning_rate": 1.084133486222159e-05,
      "loss": 0.1256,
      "step": 7643
    },
    {
      "epoch": 4.890595009596929,
      "grad_norm": 1.2345491834736284,
      "learning_rate": 1.0839270037027259e-05,
      "loss": 0.1396,
      "step": 7644
    },
    {
      "epoch": 4.891234804862444,
      "grad_norm": 1.22054324368704,
      "learning_rate": 1.0837205175796168e-05,
      "loss": 0.1421,
      "step": 7645
    },
    {
      "epoch": 4.891874600127959,
      "grad_norm": 1.2699003418395094,
      "learning_rate": 1.0835140278616973e-05,
      "loss": 0.1375,
      "step": 7646
    },
    {
      "epoch": 4.892514395393474,
      "grad_norm": 1.2098472752382246,
      "learning_rate": 1.0833075345578343e-05,
      "loss": 0.1233,
      "step": 7647
    },
    {
      "epoch": 4.893154190658989,
      "grad_norm": 1.2203718269543118,
      "learning_rate": 1.0831010376768938e-05,
      "loss": 0.1202,
      "step": 7648
    },
    {
      "epoch": 4.893793985924504,
      "grad_norm": 1.2616662142142836,
      "learning_rate": 1.0828945372277427e-05,
      "loss": 0.1454,
      "step": 7649
    },
    {
      "epoch": 4.894433781190019,
      "grad_norm": 1.1651541023218255,
      "learning_rate": 1.0826880332192475e-05,
      "loss": 0.1338,
      "step": 7650
    },
    {
      "epoch": 4.895073576455534,
      "grad_norm": 1.385946855864732,
      "learning_rate": 1.0824815256602749e-05,
      "loss": 0.1434,
      "step": 7651
    },
    {
      "epoch": 4.895713371721049,
      "grad_norm": 1.2973397210740896,
      "learning_rate": 1.0822750145596928e-05,
      "loss": 0.1475,
      "step": 7652
    },
    {
      "epoch": 4.8963531669865645,
      "grad_norm": 1.2091478878682729,
      "learning_rate": 1.0820684999263678e-05,
      "loss": 0.1247,
      "step": 7653
    },
    {
      "epoch": 4.89699296225208,
      "grad_norm": 1.2917935669602583,
      "learning_rate": 1.0818619817691674e-05,
      "loss": 0.1464,
      "step": 7654
    },
    {
      "epoch": 4.897632757517594,
      "grad_norm": 1.180831619680591,
      "learning_rate": 1.0816554600969591e-05,
      "loss": 0.1372,
      "step": 7655
    },
    {
      "epoch": 4.8982725527831095,
      "grad_norm": 1.13052941601326,
      "learning_rate": 1.0814489349186106e-05,
      "loss": 0.126,
      "step": 7656
    },
    {
      "epoch": 4.898912348048625,
      "grad_norm": 1.3139004481313346,
      "learning_rate": 1.08124240624299e-05,
      "loss": 0.1467,
      "step": 7657
    },
    {
      "epoch": 4.899552143314139,
      "grad_norm": 1.2507261956781301,
      "learning_rate": 1.0810358740789646e-05,
      "loss": 0.1444,
      "step": 7658
    },
    {
      "epoch": 4.900191938579654,
      "grad_norm": 1.2528140764875955,
      "learning_rate": 1.080829338435403e-05,
      "loss": 0.1462,
      "step": 7659
    },
    {
      "epoch": 4.90083173384517,
      "grad_norm": 1.293672866644761,
      "learning_rate": 1.0806227993211734e-05,
      "loss": 0.1451,
      "step": 7660
    },
    {
      "epoch": 4.901471529110685,
      "grad_norm": 1.2378847601786422,
      "learning_rate": 1.0804162567451442e-05,
      "loss": 0.1391,
      "step": 7661
    },
    {
      "epoch": 4.902111324376199,
      "grad_norm": 1.175444463754358,
      "learning_rate": 1.0802097107161843e-05,
      "loss": 0.1242,
      "step": 7662
    },
    {
      "epoch": 4.902751119641715,
      "grad_norm": 1.200467040625829,
      "learning_rate": 1.0800031612431616e-05,
      "loss": 0.1277,
      "step": 7663
    },
    {
      "epoch": 4.90339091490723,
      "grad_norm": 1.2480462394561747,
      "learning_rate": 1.0797966083349456e-05,
      "loss": 0.1294,
      "step": 7664
    },
    {
      "epoch": 4.904030710172744,
      "grad_norm": 1.191523947808476,
      "learning_rate": 1.0795900520004054e-05,
      "loss": 0.1205,
      "step": 7665
    },
    {
      "epoch": 4.9046705054382596,
      "grad_norm": 1.295440888366616,
      "learning_rate": 1.0793834922484099e-05,
      "loss": 0.1383,
      "step": 7666
    },
    {
      "epoch": 4.905310300703775,
      "grad_norm": 1.2691191461396036,
      "learning_rate": 1.0791769290878284e-05,
      "loss": 0.1343,
      "step": 7667
    },
    {
      "epoch": 4.90595009596929,
      "grad_norm": 1.1830713288070782,
      "learning_rate": 1.0789703625275303e-05,
      "loss": 0.1206,
      "step": 7668
    },
    {
      "epoch": 4.9065898912348045,
      "grad_norm": 1.2407572944929177,
      "learning_rate": 1.0787637925763854e-05,
      "loss": 0.1331,
      "step": 7669
    },
    {
      "epoch": 4.90722968650032,
      "grad_norm": 1.2023661778962869,
      "learning_rate": 1.0785572192432634e-05,
      "loss": 0.1298,
      "step": 7670
    },
    {
      "epoch": 4.907869481765835,
      "grad_norm": 1.1393541751668752,
      "learning_rate": 1.0783506425370343e-05,
      "loss": 0.1288,
      "step": 7671
    },
    {
      "epoch": 4.90850927703135,
      "grad_norm": 1.2086483882210142,
      "learning_rate": 1.0781440624665678e-05,
      "loss": 0.1278,
      "step": 7672
    },
    {
      "epoch": 4.909149072296865,
      "grad_norm": 1.1886924963320589,
      "learning_rate": 1.0779374790407342e-05,
      "loss": 0.1341,
      "step": 7673
    },
    {
      "epoch": 4.90978886756238,
      "grad_norm": 1.3370864221978995,
      "learning_rate": 1.0777308922684038e-05,
      "loss": 0.1299,
      "step": 7674
    },
    {
      "epoch": 4.910428662827895,
      "grad_norm": 1.2179459323459016,
      "learning_rate": 1.0775243021584474e-05,
      "loss": 0.1318,
      "step": 7675
    },
    {
      "epoch": 4.9110684580934105,
      "grad_norm": 1.2039122236892215,
      "learning_rate": 1.0773177087197354e-05,
      "loss": 0.1377,
      "step": 7676
    },
    {
      "epoch": 4.911708253358925,
      "grad_norm": 1.3875803894326963,
      "learning_rate": 1.0771111119611384e-05,
      "loss": 0.1462,
      "step": 7677
    },
    {
      "epoch": 4.91234804862444,
      "grad_norm": 1.352188250686912,
      "learning_rate": 1.0769045118915275e-05,
      "loss": 0.1491,
      "step": 7678
    },
    {
      "epoch": 4.9129878438899555,
      "grad_norm": 1.3180590780367047,
      "learning_rate": 1.0766979085197735e-05,
      "loss": 0.1467,
      "step": 7679
    },
    {
      "epoch": 4.91362763915547,
      "grad_norm": 1.3392407077323831,
      "learning_rate": 1.0764913018547479e-05,
      "loss": 0.1357,
      "step": 7680
    },
    {
      "epoch": 4.914267434420985,
      "grad_norm": 1.21676612701304,
      "learning_rate": 1.0762846919053218e-05,
      "loss": 0.1475,
      "step": 7681
    },
    {
      "epoch": 4.9149072296865,
      "grad_norm": 1.200051294942864,
      "learning_rate": 1.0760780786803669e-05,
      "loss": 0.1312,
      "step": 7682
    },
    {
      "epoch": 4.915547024952016,
      "grad_norm": 1.2560811779686443,
      "learning_rate": 1.0758714621887542e-05,
      "loss": 0.143,
      "step": 7683
    },
    {
      "epoch": 4.91618682021753,
      "grad_norm": 1.2165148912422952,
      "learning_rate": 1.075664842439356e-05,
      "loss": 0.1387,
      "step": 7684
    },
    {
      "epoch": 4.916826615483045,
      "grad_norm": 1.2376929367677283,
      "learning_rate": 1.0754582194410444e-05,
      "loss": 0.1476,
      "step": 7685
    },
    {
      "epoch": 4.917466410748561,
      "grad_norm": 1.1972360002538647,
      "learning_rate": 1.075251593202691e-05,
      "loss": 0.1319,
      "step": 7686
    },
    {
      "epoch": 4.918106206014075,
      "grad_norm": 1.4133549297073822,
      "learning_rate": 1.075044963733168e-05,
      "loss": 0.1532,
      "step": 7687
    },
    {
      "epoch": 4.91874600127959,
      "grad_norm": 1.1041109130888278,
      "learning_rate": 1.0748383310413474e-05,
      "loss": 0.1294,
      "step": 7688
    },
    {
      "epoch": 4.9193857965451055,
      "grad_norm": 1.1305168208515781,
      "learning_rate": 1.0746316951361024e-05,
      "loss": 0.1303,
      "step": 7689
    },
    {
      "epoch": 4.920025591810621,
      "grad_norm": 1.104845097657789,
      "learning_rate": 1.0744250560263053e-05,
      "loss": 0.1296,
      "step": 7690
    },
    {
      "epoch": 4.920665387076136,
      "grad_norm": 1.1364961865314267,
      "learning_rate": 1.0742184137208282e-05,
      "loss": 0.1352,
      "step": 7691
    },
    {
      "epoch": 4.9213051823416505,
      "grad_norm": 1.3103806052814004,
      "learning_rate": 1.0740117682285446e-05,
      "loss": 0.1453,
      "step": 7692
    },
    {
      "epoch": 4.921944977607166,
      "grad_norm": 1.2590078134013432,
      "learning_rate": 1.0738051195583276e-05,
      "loss": 0.1375,
      "step": 7693
    },
    {
      "epoch": 4.922584772872681,
      "grad_norm": 1.2543268242899959,
      "learning_rate": 1.0735984677190498e-05,
      "loss": 0.1228,
      "step": 7694
    },
    {
      "epoch": 4.923224568138195,
      "grad_norm": 1.2411146584230224,
      "learning_rate": 1.0733918127195851e-05,
      "loss": 0.1353,
      "step": 7695
    },
    {
      "epoch": 4.923864363403711,
      "grad_norm": 1.133714104275744,
      "learning_rate": 1.073185154568806e-05,
      "loss": 0.1165,
      "step": 7696
    },
    {
      "epoch": 4.924504158669226,
      "grad_norm": 1.276291150193402,
      "learning_rate": 1.072978493275587e-05,
      "loss": 0.1402,
      "step": 7697
    },
    {
      "epoch": 4.925143953934741,
      "grad_norm": 1.338641709574382,
      "learning_rate": 1.0727718288488007e-05,
      "loss": 0.1338,
      "step": 7698
    },
    {
      "epoch": 4.925783749200256,
      "grad_norm": 1.424412913022771,
      "learning_rate": 1.0725651612973218e-05,
      "loss": 0.163,
      "step": 7699
    },
    {
      "epoch": 4.926423544465771,
      "grad_norm": 1.2997969420372528,
      "learning_rate": 1.0723584906300242e-05,
      "loss": 0.1417,
      "step": 7700
    },
    {
      "epoch": 4.927063339731286,
      "grad_norm": 1.251857155183291,
      "learning_rate": 1.0721518168557817e-05,
      "loss": 0.139,
      "step": 7701
    },
    {
      "epoch": 4.927703134996801,
      "grad_norm": 1.2913005908136785,
      "learning_rate": 1.0719451399834683e-05,
      "loss": 0.1415,
      "step": 7702
    },
    {
      "epoch": 4.928342930262316,
      "grad_norm": 1.3560195796072994,
      "learning_rate": 1.0717384600219587e-05,
      "loss": 0.1424,
      "step": 7703
    },
    {
      "epoch": 4.928982725527831,
      "grad_norm": 1.2497561079713047,
      "learning_rate": 1.0715317769801273e-05,
      "loss": 0.1377,
      "step": 7704
    },
    {
      "epoch": 4.929622520793346,
      "grad_norm": 1.240867927454618,
      "learning_rate": 1.0713250908668485e-05,
      "loss": 0.1472,
      "step": 7705
    },
    {
      "epoch": 4.930262316058861,
      "grad_norm": 1.3591415810627592,
      "learning_rate": 1.0711184016909972e-05,
      "loss": 0.1624,
      "step": 7706
    },
    {
      "epoch": 4.930902111324376,
      "grad_norm": 1.3579155495237227,
      "learning_rate": 1.0709117094614483e-05,
      "loss": 0.1461,
      "step": 7707
    },
    {
      "epoch": 4.931541906589891,
      "grad_norm": 1.258588812049571,
      "learning_rate": 1.070705014187077e-05,
      "loss": 0.1443,
      "step": 7708
    },
    {
      "epoch": 4.932181701855407,
      "grad_norm": 1.1997133906454447,
      "learning_rate": 1.0704983158767576e-05,
      "loss": 0.1371,
      "step": 7709
    },
    {
      "epoch": 4.932821497120921,
      "grad_norm": 1.1792376491978556,
      "learning_rate": 1.0702916145393662e-05,
      "loss": 0.1203,
      "step": 7710
    },
    {
      "epoch": 4.933461292386436,
      "grad_norm": 1.3094375927206539,
      "learning_rate": 1.070084910183778e-05,
      "loss": 0.1397,
      "step": 7711
    },
    {
      "epoch": 4.9341010876519515,
      "grad_norm": 1.241014024736284,
      "learning_rate": 1.0698782028188683e-05,
      "loss": 0.1331,
      "step": 7712
    },
    {
      "epoch": 4.934740882917467,
      "grad_norm": 1.3051231375661019,
      "learning_rate": 1.0696714924535129e-05,
      "loss": 0.1354,
      "step": 7713
    },
    {
      "epoch": 4.935380678182981,
      "grad_norm": 1.2159302062852653,
      "learning_rate": 1.0694647790965876e-05,
      "loss": 0.1305,
      "step": 7714
    },
    {
      "epoch": 4.9360204734484965,
      "grad_norm": 1.3595975105531464,
      "learning_rate": 1.0692580627569682e-05,
      "loss": 0.1522,
      "step": 7715
    },
    {
      "epoch": 4.936660268714012,
      "grad_norm": 1.3040804235853904,
      "learning_rate": 1.0690513434435309e-05,
      "loss": 0.1379,
      "step": 7716
    },
    {
      "epoch": 4.937300063979526,
      "grad_norm": 1.2272212434229233,
      "learning_rate": 1.0688446211651517e-05,
      "loss": 0.1391,
      "step": 7717
    },
    {
      "epoch": 4.937939859245041,
      "grad_norm": 1.2480219855553245,
      "learning_rate": 1.068637895930707e-05,
      "loss": 0.1277,
      "step": 7718
    },
    {
      "epoch": 4.938579654510557,
      "grad_norm": 1.1995274552563262,
      "learning_rate": 1.0684311677490726e-05,
      "loss": 0.1354,
      "step": 7719
    },
    {
      "epoch": 4.939219449776072,
      "grad_norm": 1.2981359101679106,
      "learning_rate": 1.0682244366291263e-05,
      "loss": 0.139,
      "step": 7720
    },
    {
      "epoch": 4.939859245041586,
      "grad_norm": 1.2914183044735332,
      "learning_rate": 1.0680177025797436e-05,
      "loss": 0.1492,
      "step": 7721
    },
    {
      "epoch": 4.940499040307102,
      "grad_norm": 1.0717888908290791,
      "learning_rate": 1.0678109656098021e-05,
      "loss": 0.1198,
      "step": 7722
    },
    {
      "epoch": 4.941138835572617,
      "grad_norm": 1.163208962564316,
      "learning_rate": 1.0676042257281783e-05,
      "loss": 0.1235,
      "step": 7723
    },
    {
      "epoch": 4.941778630838132,
      "grad_norm": 1.152856548023267,
      "learning_rate": 1.067397482943749e-05,
      "loss": 0.125,
      "step": 7724
    },
    {
      "epoch": 4.942418426103647,
      "grad_norm": 1.2134083831839264,
      "learning_rate": 1.067190737265392e-05,
      "loss": 0.1327,
      "step": 7725
    },
    {
      "epoch": 4.943058221369162,
      "grad_norm": 1.3281393786665303,
      "learning_rate": 1.0669839887019839e-05,
      "loss": 0.1457,
      "step": 7726
    },
    {
      "epoch": 4.943698016634677,
      "grad_norm": 1.1556265185611803,
      "learning_rate": 1.0667772372624029e-05,
      "loss": 0.1238,
      "step": 7727
    },
    {
      "epoch": 4.944337811900192,
      "grad_norm": 1.3113692987537888,
      "learning_rate": 1.0665704829555257e-05,
      "loss": 0.1493,
      "step": 7728
    },
    {
      "epoch": 4.944977607165707,
      "grad_norm": 1.3106139335880405,
      "learning_rate": 1.0663637257902307e-05,
      "loss": 0.1412,
      "step": 7729
    },
    {
      "epoch": 4.945617402431222,
      "grad_norm": 1.2768577576237037,
      "learning_rate": 1.0661569657753952e-05,
      "loss": 0.1414,
      "step": 7730
    },
    {
      "epoch": 4.946257197696737,
      "grad_norm": 1.2982529503246683,
      "learning_rate": 1.0659502029198973e-05,
      "loss": 0.1409,
      "step": 7731
    },
    {
      "epoch": 4.946896992962252,
      "grad_norm": 1.2359107827031426,
      "learning_rate": 1.0657434372326155e-05,
      "loss": 0.1373,
      "step": 7732
    },
    {
      "epoch": 4.947536788227767,
      "grad_norm": 1.29494552604399,
      "learning_rate": 1.0655366687224266e-05,
      "loss": 0.1388,
      "step": 7733
    },
    {
      "epoch": 4.948176583493282,
      "grad_norm": 1.2308290952716072,
      "learning_rate": 1.06532989739821e-05,
      "loss": 0.1327,
      "step": 7734
    },
    {
      "epoch": 4.9488163787587975,
      "grad_norm": 1.3554580989517993,
      "learning_rate": 1.0651231232688442e-05,
      "loss": 0.1391,
      "step": 7735
    },
    {
      "epoch": 4.949456174024312,
      "grad_norm": 1.3253328766359225,
      "learning_rate": 1.0649163463432068e-05,
      "loss": 0.1435,
      "step": 7736
    },
    {
      "epoch": 4.950095969289827,
      "grad_norm": 1.3506070087536746,
      "learning_rate": 1.0647095666301774e-05,
      "loss": 0.1332,
      "step": 7737
    },
    {
      "epoch": 4.9507357645553425,
      "grad_norm": 1.2321419365301567,
      "learning_rate": 1.0645027841386344e-05,
      "loss": 0.1356,
      "step": 7738
    },
    {
      "epoch": 4.951375559820857,
      "grad_norm": 1.198913990798165,
      "learning_rate": 1.0642959988774564e-05,
      "loss": 0.1419,
      "step": 7739
    },
    {
      "epoch": 4.952015355086372,
      "grad_norm": 1.2604926784560824,
      "learning_rate": 1.0640892108555226e-05,
      "loss": 0.1481,
      "step": 7740
    },
    {
      "epoch": 4.952655150351887,
      "grad_norm": 1.1039460803181844,
      "learning_rate": 1.0638824200817117e-05,
      "loss": 0.1158,
      "step": 7741
    },
    {
      "epoch": 4.953294945617403,
      "grad_norm": 1.377278936763423,
      "learning_rate": 1.0636756265649037e-05,
      "loss": 0.1498,
      "step": 7742
    },
    {
      "epoch": 4.953934740882918,
      "grad_norm": 1.0480999957165105,
      "learning_rate": 1.0634688303139775e-05,
      "loss": 0.123,
      "step": 7743
    },
    {
      "epoch": 4.954574536148432,
      "grad_norm": 1.2196847013535075,
      "learning_rate": 1.0632620313378128e-05,
      "loss": 0.138,
      "step": 7744
    },
    {
      "epoch": 4.955214331413948,
      "grad_norm": 1.1746156563866093,
      "learning_rate": 1.0630552296452887e-05,
      "loss": 0.1276,
      "step": 7745
    },
    {
      "epoch": 4.955854126679463,
      "grad_norm": 1.0699398146996124,
      "learning_rate": 1.0628484252452855e-05,
      "loss": 0.121,
      "step": 7746
    },
    {
      "epoch": 4.956493921944977,
      "grad_norm": 1.401157274907964,
      "learning_rate": 1.0626416181466825e-05,
      "loss": 0.1459,
      "step": 7747
    },
    {
      "epoch": 4.957133717210493,
      "grad_norm": 1.2541488712469904,
      "learning_rate": 1.06243480835836e-05,
      "loss": 0.1347,
      "step": 7748
    },
    {
      "epoch": 4.957773512476008,
      "grad_norm": 1.1960330599882794,
      "learning_rate": 1.0622279958891978e-05,
      "loss": 0.1344,
      "step": 7749
    },
    {
      "epoch": 4.958413307741523,
      "grad_norm": 1.332681068376442,
      "learning_rate": 1.062021180748076e-05,
      "loss": 0.1422,
      "step": 7750
    },
    {
      "epoch": 4.9590531030070375,
      "grad_norm": 1.1984445597083049,
      "learning_rate": 1.061814362943875e-05,
      "loss": 0.1395,
      "step": 7751
    },
    {
      "epoch": 4.959692898272553,
      "grad_norm": 1.3165305363107949,
      "learning_rate": 1.0616075424854757e-05,
      "loss": 0.1491,
      "step": 7752
    },
    {
      "epoch": 4.960332693538068,
      "grad_norm": 1.1532852709059096,
      "learning_rate": 1.0614007193817579e-05,
      "loss": 0.1279,
      "step": 7753
    },
    {
      "epoch": 4.960972488803582,
      "grad_norm": 1.2296030337040824,
      "learning_rate": 1.0611938936416024e-05,
      "loss": 0.1338,
      "step": 7754
    },
    {
      "epoch": 4.961612284069098,
      "grad_norm": 1.31692930460072,
      "learning_rate": 1.0609870652738898e-05,
      "loss": 0.1393,
      "step": 7755
    },
    {
      "epoch": 4.962252079334613,
      "grad_norm": 1.3270053089090141,
      "learning_rate": 1.0607802342875011e-05,
      "loss": 0.1438,
      "step": 7756
    },
    {
      "epoch": 4.962891874600128,
      "grad_norm": 1.3549172580951139,
      "learning_rate": 1.0605734006913175e-05,
      "loss": 0.1573,
      "step": 7757
    },
    {
      "epoch": 4.963531669865643,
      "grad_norm": 1.2576386938440058,
      "learning_rate": 1.0603665644942196e-05,
      "loss": 0.1473,
      "step": 7758
    },
    {
      "epoch": 4.964171465131158,
      "grad_norm": 1.1856528466409455,
      "learning_rate": 1.0601597257050891e-05,
      "loss": 0.1286,
      "step": 7759
    },
    {
      "epoch": 4.964811260396673,
      "grad_norm": 1.2621820022710022,
      "learning_rate": 1.0599528843328066e-05,
      "loss": 0.1305,
      "step": 7760
    },
    {
      "epoch": 4.9654510556621885,
      "grad_norm": 1.3369463759792768,
      "learning_rate": 1.0597460403862543e-05,
      "loss": 0.1464,
      "step": 7761
    },
    {
      "epoch": 4.966090850927703,
      "grad_norm": 1.3123161717896525,
      "learning_rate": 1.0595391938743131e-05,
      "loss": 0.1423,
      "step": 7762
    },
    {
      "epoch": 4.966730646193218,
      "grad_norm": 1.2859160478345266,
      "learning_rate": 1.0593323448058647e-05,
      "loss": 0.1409,
      "step": 7763
    },
    {
      "epoch": 4.967370441458733,
      "grad_norm": 1.2720503097243823,
      "learning_rate": 1.0591254931897912e-05,
      "loss": 0.1395,
      "step": 7764
    },
    {
      "epoch": 4.968010236724249,
      "grad_norm": 1.303306104145913,
      "learning_rate": 1.058918639034974e-05,
      "loss": 0.1339,
      "step": 7765
    },
    {
      "epoch": 4.968650031989763,
      "grad_norm": 1.2248927292491492,
      "learning_rate": 1.0587117823502952e-05,
      "loss": 0.1403,
      "step": 7766
    },
    {
      "epoch": 4.969289827255278,
      "grad_norm": 1.2432409866968372,
      "learning_rate": 1.0585049231446371e-05,
      "loss": 0.1503,
      "step": 7767
    },
    {
      "epoch": 4.969929622520794,
      "grad_norm": 1.2622911950568725,
      "learning_rate": 1.0582980614268819e-05,
      "loss": 0.1371,
      "step": 7768
    },
    {
      "epoch": 4.970569417786308,
      "grad_norm": 1.257769445693851,
      "learning_rate": 1.0580911972059114e-05,
      "loss": 0.1363,
      "step": 7769
    },
    {
      "epoch": 4.971209213051823,
      "grad_norm": 1.3005248341827553,
      "learning_rate": 1.0578843304906081e-05,
      "loss": 0.143,
      "step": 7770
    },
    {
      "epoch": 4.971849008317339,
      "grad_norm": 1.087083388027065,
      "learning_rate": 1.0576774612898545e-05,
      "loss": 0.1237,
      "step": 7771
    },
    {
      "epoch": 4.972488803582854,
      "grad_norm": 1.3002536932642652,
      "learning_rate": 1.0574705896125336e-05,
      "loss": 0.1394,
      "step": 7772
    },
    {
      "epoch": 4.973128598848368,
      "grad_norm": 1.1549094816487135,
      "learning_rate": 1.0572637154675276e-05,
      "loss": 0.1292,
      "step": 7773
    },
    {
      "epoch": 4.9737683941138835,
      "grad_norm": 1.2521785584634368,
      "learning_rate": 1.05705683886372e-05,
      "loss": 0.1288,
      "step": 7774
    },
    {
      "epoch": 4.974408189379399,
      "grad_norm": 1.2275350843556345,
      "learning_rate": 1.0568499598099926e-05,
      "loss": 0.1362,
      "step": 7775
    },
    {
      "epoch": 4.975047984644913,
      "grad_norm": 1.1164254096005173,
      "learning_rate": 1.0566430783152295e-05,
      "loss": 0.1279,
      "step": 7776
    },
    {
      "epoch": 4.975687779910428,
      "grad_norm": 1.2656678849918723,
      "learning_rate": 1.0564361943883133e-05,
      "loss": 0.143,
      "step": 7777
    },
    {
      "epoch": 4.976327575175944,
      "grad_norm": 1.240319512996512,
      "learning_rate": 1.0562293080381272e-05,
      "loss": 0.1369,
      "step": 7778
    },
    {
      "epoch": 4.976967370441459,
      "grad_norm": 1.2209292355008818,
      "learning_rate": 1.0560224192735547e-05,
      "loss": 0.1317,
      "step": 7779
    },
    {
      "epoch": 4.977607165706974,
      "grad_norm": 1.2465594906615285,
      "learning_rate": 1.055815528103479e-05,
      "loss": 0.1411,
      "step": 7780
    },
    {
      "epoch": 4.978246960972489,
      "grad_norm": 1.1400505699405659,
      "learning_rate": 1.0556086345367837e-05,
      "loss": 0.1211,
      "step": 7781
    },
    {
      "epoch": 4.978886756238004,
      "grad_norm": 1.199303795660861,
      "learning_rate": 1.055401738582353e-05,
      "loss": 0.1297,
      "step": 7782
    },
    {
      "epoch": 4.979526551503519,
      "grad_norm": 1.3464981839654564,
      "learning_rate": 1.0551948402490701e-05,
      "loss": 0.1412,
      "step": 7783
    },
    {
      "epoch": 4.980166346769034,
      "grad_norm": 1.2533267542646183,
      "learning_rate": 1.0549879395458188e-05,
      "loss": 0.1334,
      "step": 7784
    },
    {
      "epoch": 4.980806142034549,
      "grad_norm": 1.2367218582911863,
      "learning_rate": 1.0547810364814832e-05,
      "loss": 0.1285,
      "step": 7785
    },
    {
      "epoch": 4.981445937300064,
      "grad_norm": 1.24621020913255,
      "learning_rate": 1.0545741310649472e-05,
      "loss": 0.1375,
      "step": 7786
    },
    {
      "epoch": 4.982085732565579,
      "grad_norm": 1.310401375641122,
      "learning_rate": 1.0543672233050955e-05,
      "loss": 0.1398,
      "step": 7787
    },
    {
      "epoch": 4.982725527831094,
      "grad_norm": 1.2843517083540916,
      "learning_rate": 1.0541603132108116e-05,
      "loss": 0.1337,
      "step": 7788
    },
    {
      "epoch": 4.983365323096609,
      "grad_norm": 1.3786850789657579,
      "learning_rate": 1.0539534007909805e-05,
      "loss": 0.1435,
      "step": 7789
    },
    {
      "epoch": 4.984005118362124,
      "grad_norm": 1.1202921463201938,
      "learning_rate": 1.0537464860544861e-05,
      "loss": 0.1294,
      "step": 7790
    },
    {
      "epoch": 4.984644913627639,
      "grad_norm": 1.2467759653192254,
      "learning_rate": 1.0535395690102132e-05,
      "loss": 0.134,
      "step": 7791
    },
    {
      "epoch": 4.985284708893154,
      "grad_norm": 1.178914854169931,
      "learning_rate": 1.0533326496670466e-05,
      "loss": 0.138,
      "step": 7792
    },
    {
      "epoch": 4.985924504158669,
      "grad_norm": 1.2186327691002623,
      "learning_rate": 1.0531257280338708e-05,
      "loss": 0.1294,
      "step": 7793
    },
    {
      "epoch": 4.9865642994241846,
      "grad_norm": 0.9401491233844751,
      "learning_rate": 1.0529188041195708e-05,
      "loss": 0.1147,
      "step": 7794
    },
    {
      "epoch": 4.987204094689699,
      "grad_norm": 1.237192341839228,
      "learning_rate": 1.0527118779330312e-05,
      "loss": 0.1374,
      "step": 7795
    },
    {
      "epoch": 4.987843889955214,
      "grad_norm": 1.2095514935757585,
      "learning_rate": 1.0525049494831375e-05,
      "loss": 0.1253,
      "step": 7796
    },
    {
      "epoch": 4.9884836852207295,
      "grad_norm": 1.1291363891308193,
      "learning_rate": 1.0522980187787748e-05,
      "loss": 0.1287,
      "step": 7797
    },
    {
      "epoch": 4.989123480486245,
      "grad_norm": 1.2379927701800215,
      "learning_rate": 1.0520910858288282e-05,
      "loss": 0.1384,
      "step": 7798
    },
    {
      "epoch": 4.989763275751759,
      "grad_norm": 1.3087234835768362,
      "learning_rate": 1.0518841506421831e-05,
      "loss": 0.134,
      "step": 7799
    },
    {
      "epoch": 4.990403071017274,
      "grad_norm": 1.2678610719981587,
      "learning_rate": 1.0516772132277247e-05,
      "loss": 0.1333,
      "step": 7800
    },
    {
      "epoch": 4.99104286628279,
      "grad_norm": 1.3799994834155802,
      "learning_rate": 1.0514702735943385e-05,
      "loss": 0.1502,
      "step": 7801
    },
    {
      "epoch": 4.991682661548305,
      "grad_norm": 1.040553558942462,
      "learning_rate": 1.0512633317509106e-05,
      "loss": 0.1131,
      "step": 7802
    },
    {
      "epoch": 4.992322456813819,
      "grad_norm": 1.240297435913589,
      "learning_rate": 1.0510563877063263e-05,
      "loss": 0.1375,
      "step": 7803
    },
    {
      "epoch": 4.992962252079335,
      "grad_norm": 1.2631316783585684,
      "learning_rate": 1.0508494414694717e-05,
      "loss": 0.1374,
      "step": 7804
    },
    {
      "epoch": 4.99360204734485,
      "grad_norm": 1.261055216973503,
      "learning_rate": 1.0506424930492326e-05,
      "loss": 0.141,
      "step": 7805
    },
    {
      "epoch": 4.994241842610364,
      "grad_norm": 1.0842368021619284,
      "learning_rate": 1.0504355424544945e-05,
      "loss": 0.1247,
      "step": 7806
    },
    {
      "epoch": 4.99488163787588,
      "grad_norm": 1.0867996874425394,
      "learning_rate": 1.0502285896941441e-05,
      "loss": 0.1145,
      "step": 7807
    },
    {
      "epoch": 4.995521433141395,
      "grad_norm": 1.2958645834277651,
      "learning_rate": 1.0500216347770675e-05,
      "loss": 0.1504,
      "step": 7808
    },
    {
      "epoch": 4.99616122840691,
      "grad_norm": 1.169165466349309,
      "learning_rate": 1.049814677712151e-05,
      "loss": 0.1305,
      "step": 7809
    },
    {
      "epoch": 4.9968010236724245,
      "grad_norm": 1.228834459276936,
      "learning_rate": 1.0496077185082808e-05,
      "loss": 0.1408,
      "step": 7810
    },
    {
      "epoch": 4.99744081893794,
      "grad_norm": 1.2638180305988822,
      "learning_rate": 1.049400757174343e-05,
      "loss": 0.1418,
      "step": 7811
    },
    {
      "epoch": 4.998080614203455,
      "grad_norm": 1.100859087939892,
      "learning_rate": 1.0491937937192252e-05,
      "loss": 0.1243,
      "step": 7812
    },
    {
      "epoch": 4.9987204094689695,
      "grad_norm": 1.2225168860937399,
      "learning_rate": 1.0489868281518129e-05,
      "loss": 0.1359,
      "step": 7813
    },
    {
      "epoch": 4.999360204734485,
      "grad_norm": 1.1907471241125387,
      "learning_rate": 1.0487798604809938e-05,
      "loss": 0.1231,
      "step": 7814
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.1634774432327195,
      "learning_rate": 1.0485728907156539e-05,
      "loss": 0.1315,
      "step": 7815
    },
    {
      "epoch": 5.000639795265515,
      "grad_norm": 0.8171956650531078,
      "learning_rate": 1.0483659188646806e-05,
      "loss": 0.0897,
      "step": 7816
    },
    {
      "epoch": 5.00127959053103,
      "grad_norm": 0.8779715179715015,
      "learning_rate": 1.0481589449369606e-05,
      "loss": 0.0921,
      "step": 7817
    },
    {
      "epoch": 5.001919385796545,
      "grad_norm": 0.8243746954127156,
      "learning_rate": 1.0479519689413813e-05,
      "loss": 0.0912,
      "step": 7818
    },
    {
      "epoch": 5.00255918106206,
      "grad_norm": 1.0056868098613698,
      "learning_rate": 1.04774499088683e-05,
      "loss": 0.096,
      "step": 7819
    },
    {
      "epoch": 5.0031989763275755,
      "grad_norm": 0.7967825262983965,
      "learning_rate": 1.0475380107821936e-05,
      "loss": 0.0844,
      "step": 7820
    },
    {
      "epoch": 5.00383877159309,
      "grad_norm": 0.9597288651065097,
      "learning_rate": 1.0473310286363595e-05,
      "loss": 0.0909,
      "step": 7821
    },
    {
      "epoch": 5.004478566858605,
      "grad_norm": 1.0650924596092843,
      "learning_rate": 1.0471240444582151e-05,
      "loss": 0.0941,
      "step": 7822
    },
    {
      "epoch": 5.00511836212412,
      "grad_norm": 0.9645450267289696,
      "learning_rate": 1.0469170582566482e-05,
      "loss": 0.09,
      "step": 7823
    },
    {
      "epoch": 5.005758157389636,
      "grad_norm": 0.9837390848689836,
      "learning_rate": 1.0467100700405464e-05,
      "loss": 0.0858,
      "step": 7824
    },
    {
      "epoch": 5.00639795265515,
      "grad_norm": 0.9552701443409547,
      "learning_rate": 1.046503079818797e-05,
      "loss": 0.0831,
      "step": 7825
    },
    {
      "epoch": 5.007037747920665,
      "grad_norm": 1.0674519678643908,
      "learning_rate": 1.0462960876002884e-05,
      "loss": 0.0899,
      "step": 7826
    },
    {
      "epoch": 5.007677543186181,
      "grad_norm": 1.021767253161758,
      "learning_rate": 1.046089093393908e-05,
      "loss": 0.0916,
      "step": 7827
    },
    {
      "epoch": 5.008317338451695,
      "grad_norm": 1.1081912665716072,
      "learning_rate": 1.0458820972085438e-05,
      "loss": 0.0999,
      "step": 7828
    },
    {
      "epoch": 5.00895713371721,
      "grad_norm": 1.0586935479691253,
      "learning_rate": 1.0456750990530841e-05,
      "loss": 0.0869,
      "step": 7829
    },
    {
      "epoch": 5.009596928982726,
      "grad_norm": 1.1220255838387858,
      "learning_rate": 1.0454680989364168e-05,
      "loss": 0.0941,
      "step": 7830
    },
    {
      "epoch": 5.010236724248241,
      "grad_norm": 0.8826573433931292,
      "learning_rate": 1.0452610968674303e-05,
      "loss": 0.0857,
      "step": 7831
    },
    {
      "epoch": 5.010876519513755,
      "grad_norm": 1.0079731181602463,
      "learning_rate": 1.0450540928550126e-05,
      "loss": 0.0849,
      "step": 7832
    },
    {
      "epoch": 5.0115163147792705,
      "grad_norm": 0.9835375864716557,
      "learning_rate": 1.0448470869080525e-05,
      "loss": 0.0887,
      "step": 7833
    },
    {
      "epoch": 5.012156110044786,
      "grad_norm": 1.0305461134293057,
      "learning_rate": 1.0446400790354384e-05,
      "loss": 0.09,
      "step": 7834
    },
    {
      "epoch": 5.012795905310301,
      "grad_norm": 0.9606349143728874,
      "learning_rate": 1.0444330692460584e-05,
      "loss": 0.0793,
      "step": 7835
    },
    {
      "epoch": 5.0134357005758154,
      "grad_norm": 1.0226462419715527,
      "learning_rate": 1.0442260575488019e-05,
      "loss": 0.0804,
      "step": 7836
    },
    {
      "epoch": 5.014075495841331,
      "grad_norm": 1.0389382115831383,
      "learning_rate": 1.0440190439525565e-05,
      "loss": 0.0899,
      "step": 7837
    },
    {
      "epoch": 5.014715291106846,
      "grad_norm": 1.0324790664418808,
      "learning_rate": 1.0438120284662121e-05,
      "loss": 0.0764,
      "step": 7838
    },
    {
      "epoch": 5.015355086372361,
      "grad_norm": 1.0333364520132826,
      "learning_rate": 1.0436050110986571e-05,
      "loss": 0.0863,
      "step": 7839
    },
    {
      "epoch": 5.015994881637876,
      "grad_norm": 1.0993325742284354,
      "learning_rate": 1.0433979918587802e-05,
      "loss": 0.0895,
      "step": 7840
    },
    {
      "epoch": 5.016634676903391,
      "grad_norm": 1.0676674167953313,
      "learning_rate": 1.0431909707554712e-05,
      "loss": 0.0839,
      "step": 7841
    },
    {
      "epoch": 5.017274472168906,
      "grad_norm": 1.0281088700629015,
      "learning_rate": 1.0429839477976184e-05,
      "loss": 0.0895,
      "step": 7842
    },
    {
      "epoch": 5.017914267434421,
      "grad_norm": 1.024467386948784,
      "learning_rate": 1.0427769229941116e-05,
      "loss": 0.0882,
      "step": 7843
    },
    {
      "epoch": 5.018554062699936,
      "grad_norm": 1.0385554351550388,
      "learning_rate": 1.0425698963538398e-05,
      "loss": 0.0892,
      "step": 7844
    },
    {
      "epoch": 5.019193857965451,
      "grad_norm": 0.9437137409139962,
      "learning_rate": 1.042362867885692e-05,
      "loss": 0.082,
      "step": 7845
    },
    {
      "epoch": 5.019833653230966,
      "grad_norm": 1.0166642050382728,
      "learning_rate": 1.0421558375985586e-05,
      "loss": 0.0916,
      "step": 7846
    },
    {
      "epoch": 5.020473448496481,
      "grad_norm": 0.9198084965073994,
      "learning_rate": 1.0419488055013278e-05,
      "loss": 0.0806,
      "step": 7847
    },
    {
      "epoch": 5.021113243761996,
      "grad_norm": 1.025700666095979,
      "learning_rate": 1.0417417716028903e-05,
      "loss": 0.0893,
      "step": 7848
    },
    {
      "epoch": 5.021753039027511,
      "grad_norm": 0.9412566851275124,
      "learning_rate": 1.0415347359121355e-05,
      "loss": 0.085,
      "step": 7849
    },
    {
      "epoch": 5.022392834293027,
      "grad_norm": 0.9826337838744051,
      "learning_rate": 1.041327698437953e-05,
      "loss": 0.0815,
      "step": 7850
    },
    {
      "epoch": 5.023032629558541,
      "grad_norm": 1.0507172714675286,
      "learning_rate": 1.0411206591892327e-05,
      "loss": 0.0912,
      "step": 7851
    },
    {
      "epoch": 5.023672424824056,
      "grad_norm": 1.0052886028437529,
      "learning_rate": 1.040913618174864e-05,
      "loss": 0.0795,
      "step": 7852
    },
    {
      "epoch": 5.024312220089572,
      "grad_norm": 0.9487728020656375,
      "learning_rate": 1.0407065754037376e-05,
      "loss": 0.0835,
      "step": 7853
    },
    {
      "epoch": 5.024952015355086,
      "grad_norm": 0.8677228800098475,
      "learning_rate": 1.0404995308847434e-05,
      "loss": 0.0809,
      "step": 7854
    },
    {
      "epoch": 5.025591810620601,
      "grad_norm": 0.9968448283965965,
      "learning_rate": 1.0402924846267713e-05,
      "loss": 0.0894,
      "step": 7855
    },
    {
      "epoch": 5.0262316058861165,
      "grad_norm": 1.1006404947492332,
      "learning_rate": 1.0400854366387115e-05,
      "loss": 0.0871,
      "step": 7856
    },
    {
      "epoch": 5.026871401151632,
      "grad_norm": 1.026953825819055,
      "learning_rate": 1.0398783869294541e-05,
      "loss": 0.0857,
      "step": 7857
    },
    {
      "epoch": 5.027511196417146,
      "grad_norm": 0.9569345560166628,
      "learning_rate": 1.0396713355078902e-05,
      "loss": 0.0782,
      "step": 7858
    },
    {
      "epoch": 5.028150991682661,
      "grad_norm": 0.9359485556164147,
      "learning_rate": 1.0394642823829098e-05,
      "loss": 0.0844,
      "step": 7859
    },
    {
      "epoch": 5.028790786948177,
      "grad_norm": 1.089734023899305,
      "learning_rate": 1.039257227563403e-05,
      "loss": 0.0892,
      "step": 7860
    },
    {
      "epoch": 5.029430582213692,
      "grad_norm": 0.8969923248721756,
      "learning_rate": 1.0390501710582608e-05,
      "loss": 0.0798,
      "step": 7861
    },
    {
      "epoch": 5.030070377479206,
      "grad_norm": 1.0711801094453623,
      "learning_rate": 1.0388431128763735e-05,
      "loss": 0.0865,
      "step": 7862
    },
    {
      "epoch": 5.030710172744722,
      "grad_norm": 1.0182192428656238,
      "learning_rate": 1.038636053026632e-05,
      "loss": 0.0838,
      "step": 7863
    },
    {
      "epoch": 5.031349968010237,
      "grad_norm": 1.1101365966770893,
      "learning_rate": 1.0384289915179273e-05,
      "loss": 0.0833,
      "step": 7864
    },
    {
      "epoch": 5.031989763275751,
      "grad_norm": 1.0580989669363048,
      "learning_rate": 1.0382219283591502e-05,
      "loss": 0.0797,
      "step": 7865
    },
    {
      "epoch": 5.032629558541267,
      "grad_norm": 1.0902425187397364,
      "learning_rate": 1.0380148635591916e-05,
      "loss": 0.087,
      "step": 7866
    },
    {
      "epoch": 5.033269353806782,
      "grad_norm": 1.0531112357609864,
      "learning_rate": 1.0378077971269418e-05,
      "loss": 0.0896,
      "step": 7867
    },
    {
      "epoch": 5.033909149072297,
      "grad_norm": 1.0914848986081316,
      "learning_rate": 1.0376007290712928e-05,
      "loss": 0.0839,
      "step": 7868
    },
    {
      "epoch": 5.0345489443378115,
      "grad_norm": 1.1192179484269795,
      "learning_rate": 1.0373936594011354e-05,
      "loss": 0.0878,
      "step": 7869
    },
    {
      "epoch": 5.035188739603327,
      "grad_norm": 1.081025408376401,
      "learning_rate": 1.0371865881253606e-05,
      "loss": 0.0921,
      "step": 7870
    },
    {
      "epoch": 5.035828534868842,
      "grad_norm": 0.9833826366697342,
      "learning_rate": 1.03697951525286e-05,
      "loss": 0.0876,
      "step": 7871
    },
    {
      "epoch": 5.036468330134357,
      "grad_norm": 1.0288328263230795,
      "learning_rate": 1.0367724407925248e-05,
      "loss": 0.0822,
      "step": 7872
    },
    {
      "epoch": 5.037108125399872,
      "grad_norm": 0.9748180112065032,
      "learning_rate": 1.0365653647532466e-05,
      "loss": 0.0869,
      "step": 7873
    },
    {
      "epoch": 5.037747920665387,
      "grad_norm": 1.037075608929546,
      "learning_rate": 1.0363582871439164e-05,
      "loss": 0.0848,
      "step": 7874
    },
    {
      "epoch": 5.038387715930902,
      "grad_norm": 1.1491580011550049,
      "learning_rate": 1.0361512079734259e-05,
      "loss": 0.0923,
      "step": 7875
    },
    {
      "epoch": 5.039027511196418,
      "grad_norm": 0.9893851946303032,
      "learning_rate": 1.0359441272506672e-05,
      "loss": 0.0796,
      "step": 7876
    },
    {
      "epoch": 5.039667306461932,
      "grad_norm": 1.1324317979934495,
      "learning_rate": 1.0357370449845313e-05,
      "loss": 0.0902,
      "step": 7877
    },
    {
      "epoch": 5.040307101727447,
      "grad_norm": 1.0471042877930625,
      "learning_rate": 1.0355299611839104e-05,
      "loss": 0.0852,
      "step": 7878
    },
    {
      "epoch": 5.0409468969929625,
      "grad_norm": 0.9553594514522539,
      "learning_rate": 1.0353228758576964e-05,
      "loss": 0.0882,
      "step": 7879
    },
    {
      "epoch": 5.041586692258477,
      "grad_norm": 1.0517178185995215,
      "learning_rate": 1.0351157890147808e-05,
      "loss": 0.0845,
      "step": 7880
    },
    {
      "epoch": 5.042226487523992,
      "grad_norm": 1.0521578089072743,
      "learning_rate": 1.0349087006640557e-05,
      "loss": 0.0875,
      "step": 7881
    },
    {
      "epoch": 5.042866282789507,
      "grad_norm": 0.9306512130261853,
      "learning_rate": 1.0347016108144128e-05,
      "loss": 0.0847,
      "step": 7882
    },
    {
      "epoch": 5.043506078055023,
      "grad_norm": 1.0877644896490473,
      "learning_rate": 1.0344945194747445e-05,
      "loss": 0.0927,
      "step": 7883
    },
    {
      "epoch": 5.044145873320537,
      "grad_norm": 1.0055632190893606,
      "learning_rate": 1.0342874266539431e-05,
      "loss": 0.0885,
      "step": 7884
    },
    {
      "epoch": 5.044785668586052,
      "grad_norm": 0.9675874614789257,
      "learning_rate": 1.0340803323609006e-05,
      "loss": 0.0844,
      "step": 7885
    },
    {
      "epoch": 5.045425463851568,
      "grad_norm": 1.027072261120058,
      "learning_rate": 1.0338732366045092e-05,
      "loss": 0.0888,
      "step": 7886
    },
    {
      "epoch": 5.046065259117083,
      "grad_norm": 1.0490186639564456,
      "learning_rate": 1.0336661393936611e-05,
      "loss": 0.084,
      "step": 7887
    },
    {
      "epoch": 5.046705054382597,
      "grad_norm": 1.087812599509547,
      "learning_rate": 1.0334590407372493e-05,
      "loss": 0.0825,
      "step": 7888
    },
    {
      "epoch": 5.047344849648113,
      "grad_norm": 1.0352278260097925,
      "learning_rate": 1.0332519406441657e-05,
      "loss": 0.0913,
      "step": 7889
    },
    {
      "epoch": 5.047984644913628,
      "grad_norm": 1.0921809982282957,
      "learning_rate": 1.0330448391233026e-05,
      "loss": 0.0882,
      "step": 7890
    },
    {
      "epoch": 5.048624440179142,
      "grad_norm": 0.927197767330293,
      "learning_rate": 1.032837736183553e-05,
      "loss": 0.0878,
      "step": 7891
    },
    {
      "epoch": 5.0492642354446575,
      "grad_norm": 1.0136489213183368,
      "learning_rate": 1.0326306318338095e-05,
      "loss": 0.0926,
      "step": 7892
    },
    {
      "epoch": 5.049904030710173,
      "grad_norm": 0.9811402497520189,
      "learning_rate": 1.0324235260829647e-05,
      "loss": 0.0841,
      "step": 7893
    },
    {
      "epoch": 5.050543825975688,
      "grad_norm": 1.0478650089004495,
      "learning_rate": 1.0322164189399114e-05,
      "loss": 0.0866,
      "step": 7894
    },
    {
      "epoch": 5.0511836212412025,
      "grad_norm": 0.9855316137182355,
      "learning_rate": 1.0320093104135425e-05,
      "loss": 0.0763,
      "step": 7895
    },
    {
      "epoch": 5.051823416506718,
      "grad_norm": 1.0123707060635891,
      "learning_rate": 1.0318022005127506e-05,
      "loss": 0.0775,
      "step": 7896
    },
    {
      "epoch": 5.052463211772233,
      "grad_norm": 1.0435404180524044,
      "learning_rate": 1.0315950892464285e-05,
      "loss": 0.0925,
      "step": 7897
    },
    {
      "epoch": 5.053103007037748,
      "grad_norm": 1.11805973984683,
      "learning_rate": 1.03138797662347e-05,
      "loss": 0.0916,
      "step": 7898
    },
    {
      "epoch": 5.053742802303263,
      "grad_norm": 1.1221822858964723,
      "learning_rate": 1.0311808626527673e-05,
      "loss": 0.0889,
      "step": 7899
    },
    {
      "epoch": 5.054382597568778,
      "grad_norm": 1.1654281104940327,
      "learning_rate": 1.0309737473432137e-05,
      "loss": 0.0973,
      "step": 7900
    },
    {
      "epoch": 5.055022392834293,
      "grad_norm": 1.028193609207784,
      "learning_rate": 1.0307666307037027e-05,
      "loss": 0.0872,
      "step": 7901
    },
    {
      "epoch": 5.0556621880998085,
      "grad_norm": 1.0655050146313787,
      "learning_rate": 1.0305595127431272e-05,
      "loss": 0.0963,
      "step": 7902
    },
    {
      "epoch": 5.056301983365323,
      "grad_norm": 0.9508417784275714,
      "learning_rate": 1.0303523934703807e-05,
      "loss": 0.087,
      "step": 7903
    },
    {
      "epoch": 5.056941778630838,
      "grad_norm": 1.0177429929468973,
      "learning_rate": 1.0301452728943566e-05,
      "loss": 0.0795,
      "step": 7904
    },
    {
      "epoch": 5.057581573896353,
      "grad_norm": 0.9792351750625504,
      "learning_rate": 1.029938151023948e-05,
      "loss": 0.0851,
      "step": 7905
    },
    {
      "epoch": 5.058221369161868,
      "grad_norm": 1.0368518923206664,
      "learning_rate": 1.0297310278680482e-05,
      "loss": 0.0828,
      "step": 7906
    },
    {
      "epoch": 5.058861164427383,
      "grad_norm": 1.06249989264819,
      "learning_rate": 1.0295239034355512e-05,
      "loss": 0.0876,
      "step": 7907
    },
    {
      "epoch": 5.059500959692898,
      "grad_norm": 1.0500825152167779,
      "learning_rate": 1.0293167777353501e-05,
      "loss": 0.0844,
      "step": 7908
    },
    {
      "epoch": 5.060140754958414,
      "grad_norm": 1.2327246051330436,
      "learning_rate": 1.0291096507763386e-05,
      "loss": 0.0895,
      "step": 7909
    },
    {
      "epoch": 5.060780550223928,
      "grad_norm": 0.8943613759552319,
      "learning_rate": 1.028902522567411e-05,
      "loss": 0.0842,
      "step": 7910
    },
    {
      "epoch": 5.061420345489443,
      "grad_norm": 0.977251735894535,
      "learning_rate": 1.0286953931174601e-05,
      "loss": 0.0851,
      "step": 7911
    },
    {
      "epoch": 5.062060140754959,
      "grad_norm": 1.049548898478562,
      "learning_rate": 1.02848826243538e-05,
      "loss": 0.0856,
      "step": 7912
    },
    {
      "epoch": 5.062699936020474,
      "grad_norm": 1.0466319692388075,
      "learning_rate": 1.028281130530065e-05,
      "loss": 0.0924,
      "step": 7913
    },
    {
      "epoch": 5.063339731285988,
      "grad_norm": 1.0910346401111735,
      "learning_rate": 1.0280739974104081e-05,
      "loss": 0.086,
      "step": 7914
    },
    {
      "epoch": 5.0639795265515035,
      "grad_norm": 1.0375901990731213,
      "learning_rate": 1.0278668630853039e-05,
      "loss": 0.0943,
      "step": 7915
    },
    {
      "epoch": 5.064619321817019,
      "grad_norm": 1.0097830168034936,
      "learning_rate": 1.0276597275636462e-05,
      "loss": 0.0877,
      "step": 7916
    },
    {
      "epoch": 5.065259117082533,
      "grad_norm": 1.0311706423158853,
      "learning_rate": 1.0274525908543289e-05,
      "loss": 0.0897,
      "step": 7917
    },
    {
      "epoch": 5.0658989123480485,
      "grad_norm": 0.9152415647051061,
      "learning_rate": 1.0272454529662463e-05,
      "loss": 0.0764,
      "step": 7918
    },
    {
      "epoch": 5.066538707613564,
      "grad_norm": 1.0008555585007541,
      "learning_rate": 1.0270383139082924e-05,
      "loss": 0.0833,
      "step": 7919
    },
    {
      "epoch": 5.067178502879079,
      "grad_norm": 0.9423891095547023,
      "learning_rate": 1.0268311736893612e-05,
      "loss": 0.0863,
      "step": 7920
    },
    {
      "epoch": 5.067818298144593,
      "grad_norm": 1.0982609852472753,
      "learning_rate": 1.0266240323183474e-05,
      "loss": 0.0907,
      "step": 7921
    },
    {
      "epoch": 5.068458093410109,
      "grad_norm": 0.9958470512674926,
      "learning_rate": 1.0264168898041449e-05,
      "loss": 0.0828,
      "step": 7922
    },
    {
      "epoch": 5.069097888675624,
      "grad_norm": 1.0173554304039862,
      "learning_rate": 1.0262097461556482e-05,
      "loss": 0.087,
      "step": 7923
    },
    {
      "epoch": 5.069737683941139,
      "grad_norm": 1.0382742732996906,
      "learning_rate": 1.0260026013817514e-05,
      "loss": 0.0849,
      "step": 7924
    },
    {
      "epoch": 5.070377479206654,
      "grad_norm": 1.0628236802318631,
      "learning_rate": 1.0257954554913496e-05,
      "loss": 0.0875,
      "step": 7925
    },
    {
      "epoch": 5.071017274472169,
      "grad_norm": 1.0029834950645795,
      "learning_rate": 1.0255883084933364e-05,
      "loss": 0.0892,
      "step": 7926
    },
    {
      "epoch": 5.071657069737684,
      "grad_norm": 1.1031033414792195,
      "learning_rate": 1.0253811603966069e-05,
      "loss": 0.0915,
      "step": 7927
    },
    {
      "epoch": 5.0722968650031985,
      "grad_norm": 0.9810395633946577,
      "learning_rate": 1.0251740112100557e-05,
      "loss": 0.0883,
      "step": 7928
    },
    {
      "epoch": 5.072936660268714,
      "grad_norm": 1.0430936470325145,
      "learning_rate": 1.0249668609425769e-05,
      "loss": 0.0902,
      "step": 7929
    },
    {
      "epoch": 5.073576455534229,
      "grad_norm": 0.990334214075758,
      "learning_rate": 1.0247597096030655e-05,
      "loss": 0.0834,
      "step": 7930
    },
    {
      "epoch": 5.074216250799744,
      "grad_norm": 1.050096373961879,
      "learning_rate": 1.0245525572004165e-05,
      "loss": 0.0901,
      "step": 7931
    },
    {
      "epoch": 5.074856046065259,
      "grad_norm": 1.0635238060840455,
      "learning_rate": 1.0243454037435241e-05,
      "loss": 0.0896,
      "step": 7932
    },
    {
      "epoch": 5.075495841330774,
      "grad_norm": 1.1777433597245002,
      "learning_rate": 1.0241382492412837e-05,
      "loss": 0.0912,
      "step": 7933
    },
    {
      "epoch": 5.076135636596289,
      "grad_norm": 1.003983936198598,
      "learning_rate": 1.0239310937025898e-05,
      "loss": 0.0838,
      "step": 7934
    },
    {
      "epoch": 5.076775431861805,
      "grad_norm": 1.1198325406591652,
      "learning_rate": 1.023723937136337e-05,
      "loss": 0.0924,
      "step": 7935
    },
    {
      "epoch": 5.077415227127319,
      "grad_norm": 1.121883468253292,
      "learning_rate": 1.0235167795514207e-05,
      "loss": 0.0835,
      "step": 7936
    },
    {
      "epoch": 5.078055022392834,
      "grad_norm": 0.9272283607191993,
      "learning_rate": 1.0233096209567357e-05,
      "loss": 0.0841,
      "step": 7937
    },
    {
      "epoch": 5.0786948176583495,
      "grad_norm": 1.0090426645597652,
      "learning_rate": 1.0231024613611771e-05,
      "loss": 0.0875,
      "step": 7938
    },
    {
      "epoch": 5.079334612923865,
      "grad_norm": 0.8918326962541365,
      "learning_rate": 1.0228953007736397e-05,
      "loss": 0.0804,
      "step": 7939
    },
    {
      "epoch": 5.079974408189379,
      "grad_norm": 0.9632918065681111,
      "learning_rate": 1.022688139203019e-05,
      "loss": 0.0814,
      "step": 7940
    },
    {
      "epoch": 5.0806142034548945,
      "grad_norm": 1.0550294479864815,
      "learning_rate": 1.0224809766582102e-05,
      "loss": 0.0867,
      "step": 7941
    },
    {
      "epoch": 5.08125399872041,
      "grad_norm": 0.970137829359286,
      "learning_rate": 1.0222738131481077e-05,
      "loss": 0.0834,
      "step": 7942
    },
    {
      "epoch": 5.081893793985924,
      "grad_norm": 0.9943522725383446,
      "learning_rate": 1.0220666486816078e-05,
      "loss": 0.0881,
      "step": 7943
    },
    {
      "epoch": 5.082533589251439,
      "grad_norm": 1.0593261736298754,
      "learning_rate": 1.021859483267605e-05,
      "loss": 0.0894,
      "step": 7944
    },
    {
      "epoch": 5.083173384516955,
      "grad_norm": 0.9291532076295705,
      "learning_rate": 1.0216523169149946e-05,
      "loss": 0.0822,
      "step": 7945
    },
    {
      "epoch": 5.08381317978247,
      "grad_norm": 1.0155885572399583,
      "learning_rate": 1.0214451496326727e-05,
      "loss": 0.079,
      "step": 7946
    },
    {
      "epoch": 5.084452975047984,
      "grad_norm": 0.9907022817709573,
      "learning_rate": 1.0212379814295339e-05,
      "loss": 0.0868,
      "step": 7947
    },
    {
      "epoch": 5.0850927703135,
      "grad_norm": 0.9321616352893206,
      "learning_rate": 1.0210308123144743e-05,
      "loss": 0.0849,
      "step": 7948
    },
    {
      "epoch": 5.085732565579015,
      "grad_norm": 1.0769741495250837,
      "learning_rate": 1.0208236422963888e-05,
      "loss": 0.0825,
      "step": 7949
    },
    {
      "epoch": 5.08637236084453,
      "grad_norm": 0.9666518018196998,
      "learning_rate": 1.0206164713841731e-05,
      "loss": 0.0906,
      "step": 7950
    },
    {
      "epoch": 5.0870121561100445,
      "grad_norm": 0.9232091882284278,
      "learning_rate": 1.020409299586723e-05,
      "loss": 0.0807,
      "step": 7951
    },
    {
      "epoch": 5.08765195137556,
      "grad_norm": 1.0170967509855111,
      "learning_rate": 1.0202021269129336e-05,
      "loss": 0.0877,
      "step": 7952
    },
    {
      "epoch": 5.088291746641075,
      "grad_norm": 0.9895491732425257,
      "learning_rate": 1.0199949533717011e-05,
      "loss": 0.0845,
      "step": 7953
    },
    {
      "epoch": 5.0889315419065895,
      "grad_norm": 1.0658362253439284,
      "learning_rate": 1.0197877789719206e-05,
      "loss": 0.096,
      "step": 7954
    },
    {
      "epoch": 5.089571337172105,
      "grad_norm": 0.9346562629780859,
      "learning_rate": 1.0195806037224882e-05,
      "loss": 0.0848,
      "step": 7955
    },
    {
      "epoch": 5.09021113243762,
      "grad_norm": 1.0334702658894477,
      "learning_rate": 1.0193734276322998e-05,
      "loss": 0.0889,
      "step": 7956
    },
    {
      "epoch": 5.090850927703135,
      "grad_norm": 1.0207718886860058,
      "learning_rate": 1.0191662507102505e-05,
      "loss": 0.0874,
      "step": 7957
    },
    {
      "epoch": 5.09149072296865,
      "grad_norm": 0.9221240896532382,
      "learning_rate": 1.0189590729652366e-05,
      "loss": 0.083,
      "step": 7958
    },
    {
      "epoch": 5.092130518234165,
      "grad_norm": 1.1502359272777805,
      "learning_rate": 1.0187518944061536e-05,
      "loss": 0.0891,
      "step": 7959
    },
    {
      "epoch": 5.09277031349968,
      "grad_norm": 1.0057177147216094,
      "learning_rate": 1.0185447150418976e-05,
      "loss": 0.0839,
      "step": 7960
    },
    {
      "epoch": 5.0934101087651955,
      "grad_norm": 0.879926814864042,
      "learning_rate": 1.018337534881365e-05,
      "loss": 0.0858,
      "step": 7961
    },
    {
      "epoch": 5.09404990403071,
      "grad_norm": 0.9496102300429254,
      "learning_rate": 1.0181303539334509e-05,
      "loss": 0.077,
      "step": 7962
    },
    {
      "epoch": 5.094689699296225,
      "grad_norm": 1.082463111117436,
      "learning_rate": 1.0179231722070518e-05,
      "loss": 0.0916,
      "step": 7963
    },
    {
      "epoch": 5.0953294945617404,
      "grad_norm": 1.037756679758704,
      "learning_rate": 1.0177159897110637e-05,
      "loss": 0.0901,
      "step": 7964
    },
    {
      "epoch": 5.095969289827256,
      "grad_norm": 0.9993561816216786,
      "learning_rate": 1.0175088064543822e-05,
      "loss": 0.089,
      "step": 7965
    },
    {
      "epoch": 5.09660908509277,
      "grad_norm": 1.0633280858045657,
      "learning_rate": 1.017301622445904e-05,
      "loss": 0.0931,
      "step": 7966
    },
    {
      "epoch": 5.097248880358285,
      "grad_norm": 0.9658821991292899,
      "learning_rate": 1.0170944376945247e-05,
      "loss": 0.0956,
      "step": 7967
    },
    {
      "epoch": 5.097888675623801,
      "grad_norm": 1.0080687618091628,
      "learning_rate": 1.0168872522091407e-05,
      "loss": 0.0877,
      "step": 7968
    },
    {
      "epoch": 5.098528470889315,
      "grad_norm": 0.9876668770391321,
      "learning_rate": 1.016680065998648e-05,
      "loss": 0.0865,
      "step": 7969
    },
    {
      "epoch": 5.09916826615483,
      "grad_norm": 1.0337956245802986,
      "learning_rate": 1.0164728790719432e-05,
      "loss": 0.0862,
      "step": 7970
    },
    {
      "epoch": 5.099808061420346,
      "grad_norm": 1.0638179921689808,
      "learning_rate": 1.0162656914379224e-05,
      "loss": 0.0887,
      "step": 7971
    },
    {
      "epoch": 5.100447856685861,
      "grad_norm": 1.0168853357428604,
      "learning_rate": 1.0160585031054813e-05,
      "loss": 0.0851,
      "step": 7972
    },
    {
      "epoch": 5.101087651951375,
      "grad_norm": 0.8713350187168916,
      "learning_rate": 1.0158513140835169e-05,
      "loss": 0.0875,
      "step": 7973
    },
    {
      "epoch": 5.1017274472168905,
      "grad_norm": 1.0087180315375965,
      "learning_rate": 1.0156441243809252e-05,
      "loss": 0.0791,
      "step": 7974
    },
    {
      "epoch": 5.102367242482406,
      "grad_norm": 0.9244151242310953,
      "learning_rate": 1.0154369340066026e-05,
      "loss": 0.0765,
      "step": 7975
    },
    {
      "epoch": 5.103007037747921,
      "grad_norm": 0.9743117338582364,
      "learning_rate": 1.0152297429694455e-05,
      "loss": 0.0863,
      "step": 7976
    },
    {
      "epoch": 5.1036468330134355,
      "grad_norm": 1.0492950870356934,
      "learning_rate": 1.0150225512783502e-05,
      "loss": 0.0847,
      "step": 7977
    },
    {
      "epoch": 5.104286628278951,
      "grad_norm": 1.063012093391282,
      "learning_rate": 1.0148153589422136e-05,
      "loss": 0.0942,
      "step": 7978
    },
    {
      "epoch": 5.104926423544466,
      "grad_norm": 1.0145798830686876,
      "learning_rate": 1.0146081659699318e-05,
      "loss": 0.0892,
      "step": 7979
    },
    {
      "epoch": 5.10556621880998,
      "grad_norm": 0.8798779829798925,
      "learning_rate": 1.0144009723704014e-05,
      "loss": 0.0827,
      "step": 7980
    },
    {
      "epoch": 5.106206014075496,
      "grad_norm": 1.0087987875940585,
      "learning_rate": 1.0141937781525186e-05,
      "loss": 0.0846,
      "step": 7981
    },
    {
      "epoch": 5.106845809341011,
      "grad_norm": 1.0399930296160684,
      "learning_rate": 1.0139865833251801e-05,
      "loss": 0.0853,
      "step": 7982
    },
    {
      "epoch": 5.107485604606526,
      "grad_norm": 0.9889335931713843,
      "learning_rate": 1.013779387897283e-05,
      "loss": 0.085,
      "step": 7983
    },
    {
      "epoch": 5.108125399872041,
      "grad_norm": 1.0673452999088586,
      "learning_rate": 1.013572191877723e-05,
      "loss": 0.0888,
      "step": 7984
    },
    {
      "epoch": 5.108765195137556,
      "grad_norm": 1.0315221722520587,
      "learning_rate": 1.0133649952753977e-05,
      "loss": 0.0928,
      "step": 7985
    },
    {
      "epoch": 5.109404990403071,
      "grad_norm": 1.05163809957222,
      "learning_rate": 1.0131577980992033e-05,
      "loss": 0.086,
      "step": 7986
    },
    {
      "epoch": 5.110044785668586,
      "grad_norm": 0.9998913052639045,
      "learning_rate": 1.012950600358036e-05,
      "loss": 0.0839,
      "step": 7987
    },
    {
      "epoch": 5.110684580934101,
      "grad_norm": 1.0715767277674102,
      "learning_rate": 1.0127434020607934e-05,
      "loss": 0.0852,
      "step": 7988
    },
    {
      "epoch": 5.111324376199616,
      "grad_norm": 0.9579507694957379,
      "learning_rate": 1.0125362032163715e-05,
      "loss": 0.0832,
      "step": 7989
    },
    {
      "epoch": 5.111964171465131,
      "grad_norm": 0.9642659518352188,
      "learning_rate": 1.0123290038336675e-05,
      "loss": 0.088,
      "step": 7990
    },
    {
      "epoch": 5.112603966730646,
      "grad_norm": 0.9215506072127987,
      "learning_rate": 1.0121218039215779e-05,
      "loss": 0.0854,
      "step": 7991
    },
    {
      "epoch": 5.113243761996161,
      "grad_norm": 0.9179736163743553,
      "learning_rate": 1.0119146034889995e-05,
      "loss": 0.0856,
      "step": 7992
    },
    {
      "epoch": 5.113883557261676,
      "grad_norm": 0.9328615429359262,
      "learning_rate": 1.0117074025448294e-05,
      "loss": 0.0775,
      "step": 7993
    },
    {
      "epoch": 5.114523352527192,
      "grad_norm": 1.0771634157156937,
      "learning_rate": 1.0115002010979644e-05,
      "loss": 0.0951,
      "step": 7994
    },
    {
      "epoch": 5.115163147792706,
      "grad_norm": 1.0223820330976106,
      "learning_rate": 1.0112929991573011e-05,
      "loss": 0.0819,
      "step": 7995
    },
    {
      "epoch": 5.115802943058221,
      "grad_norm": 1.003467332877868,
      "learning_rate": 1.0110857967317364e-05,
      "loss": 0.0805,
      "step": 7996
    },
    {
      "epoch": 5.1164427383237365,
      "grad_norm": 1.0368898377770395,
      "learning_rate": 1.0108785938301673e-05,
      "loss": 0.0838,
      "step": 7997
    },
    {
      "epoch": 5.117082533589252,
      "grad_norm": 1.0455928913452144,
      "learning_rate": 1.010671390461491e-05,
      "loss": 0.0917,
      "step": 7998
    },
    {
      "epoch": 5.117722328854766,
      "grad_norm": 1.0584699087190539,
      "learning_rate": 1.0104641866346037e-05,
      "loss": 0.0911,
      "step": 7999
    },
    {
      "epoch": 5.1183621241202815,
      "grad_norm": 0.9869746354010326,
      "learning_rate": 1.0102569823584032e-05,
      "loss": 0.0874,
      "step": 8000
    },
    {
      "epoch": 5.119001919385797,
      "grad_norm": 1.0721883267742718,
      "learning_rate": 1.0100497776417861e-05,
      "loss": 0.0885,
      "step": 8001
    },
    {
      "epoch": 5.119641714651312,
      "grad_norm": 0.9073927033022419,
      "learning_rate": 1.0098425724936492e-05,
      "loss": 0.0788,
      "step": 8002
    },
    {
      "epoch": 5.120281509916826,
      "grad_norm": 1.001332054533845,
      "learning_rate": 1.00963536692289e-05,
      "loss": 0.0793,
      "step": 8003
    },
    {
      "epoch": 5.120921305182342,
      "grad_norm": 0.8924091082567635,
      "learning_rate": 1.0094281609384051e-05,
      "loss": 0.0762,
      "step": 8004
    },
    {
      "epoch": 5.121561100447857,
      "grad_norm": 0.9562297230567642,
      "learning_rate": 1.0092209545490918e-05,
      "loss": 0.0805,
      "step": 8005
    },
    {
      "epoch": 5.122200895713371,
      "grad_norm": 0.9638936866580441,
      "learning_rate": 1.009013747763847e-05,
      "loss": 0.0771,
      "step": 8006
    },
    {
      "epoch": 5.122840690978887,
      "grad_norm": 1.0215101466405994,
      "learning_rate": 1.0088065405915681e-05,
      "loss": 0.0833,
      "step": 8007
    },
    {
      "epoch": 5.123480486244402,
      "grad_norm": 1.0294699637825526,
      "learning_rate": 1.008599333041152e-05,
      "loss": 0.0886,
      "step": 8008
    },
    {
      "epoch": 5.124120281509917,
      "grad_norm": 1.1044061647922552,
      "learning_rate": 1.0083921251214959e-05,
      "loss": 0.0852,
      "step": 8009
    },
    {
      "epoch": 5.1247600767754315,
      "grad_norm": 1.1606486478217515,
      "learning_rate": 1.008184916841497e-05,
      "loss": 0.087,
      "step": 8010
    },
    {
      "epoch": 5.125399872040947,
      "grad_norm": 1.1039986840902756,
      "learning_rate": 1.0079777082100518e-05,
      "loss": 0.0875,
      "step": 8011
    },
    {
      "epoch": 5.126039667306462,
      "grad_norm": 0.9649901138593657,
      "learning_rate": 1.0077704992360585e-05,
      "loss": 0.0843,
      "step": 8012
    },
    {
      "epoch": 5.126679462571977,
      "grad_norm": 1.1613715486241134,
      "learning_rate": 1.0075632899284136e-05,
      "loss": 0.0859,
      "step": 8013
    },
    {
      "epoch": 5.127319257837492,
      "grad_norm": 0.993226966299139,
      "learning_rate": 1.0073560802960143e-05,
      "loss": 0.0874,
      "step": 8014
    },
    {
      "epoch": 5.127959053103007,
      "grad_norm": 1.0188199718779039,
      "learning_rate": 1.0071488703477584e-05,
      "loss": 0.0854,
      "step": 8015
    },
    {
      "epoch": 5.128598848368522,
      "grad_norm": 0.9994668950510467,
      "learning_rate": 1.0069416600925423e-05,
      "loss": 0.0833,
      "step": 8016
    },
    {
      "epoch": 5.129238643634037,
      "grad_norm": 0.8800936730135207,
      "learning_rate": 1.006734449539264e-05,
      "loss": 0.0829,
      "step": 8017
    },
    {
      "epoch": 5.129878438899552,
      "grad_norm": 1.0223606565578065,
      "learning_rate": 1.0065272386968203e-05,
      "loss": 0.0859,
      "step": 8018
    },
    {
      "epoch": 5.130518234165067,
      "grad_norm": 1.0212870281036366,
      "learning_rate": 1.0063200275741083e-05,
      "loss": 0.0858,
      "step": 8019
    },
    {
      "epoch": 5.1311580294305825,
      "grad_norm": 1.0128707486864716,
      "learning_rate": 1.006112816180026e-05,
      "loss": 0.0893,
      "step": 8020
    },
    {
      "epoch": 5.131797824696097,
      "grad_norm": 0.9559748877153914,
      "learning_rate": 1.0059056045234698e-05,
      "loss": 0.0831,
      "step": 8021
    },
    {
      "epoch": 5.132437619961612,
      "grad_norm": 1.1428135618716264,
      "learning_rate": 1.0056983926133377e-05,
      "loss": 0.0977,
      "step": 8022
    },
    {
      "epoch": 5.1330774152271275,
      "grad_norm": 1.16322573369093,
      "learning_rate": 1.005491180458527e-05,
      "loss": 0.0905,
      "step": 8023
    },
    {
      "epoch": 5.133717210492643,
      "grad_norm": 0.9727029805857039,
      "learning_rate": 1.0052839680679345e-05,
      "loss": 0.0891,
      "step": 8024
    },
    {
      "epoch": 5.134357005758157,
      "grad_norm": 1.0162531902608043,
      "learning_rate": 1.0050767554504581e-05,
      "loss": 0.0866,
      "step": 8025
    },
    {
      "epoch": 5.134996801023672,
      "grad_norm": 1.1169176094030484,
      "learning_rate": 1.0048695426149944e-05,
      "loss": 0.0879,
      "step": 8026
    },
    {
      "epoch": 5.135636596289188,
      "grad_norm": 0.9462102680189866,
      "learning_rate": 1.0046623295704414e-05,
      "loss": 0.0852,
      "step": 8027
    },
    {
      "epoch": 5.136276391554702,
      "grad_norm": 1.0035741939174836,
      "learning_rate": 1.0044551163256964e-05,
      "loss": 0.0882,
      "step": 8028
    },
    {
      "epoch": 5.136916186820217,
      "grad_norm": 0.977145033872795,
      "learning_rate": 1.0042479028896563e-05,
      "loss": 0.0843,
      "step": 8029
    },
    {
      "epoch": 5.137555982085733,
      "grad_norm": 0.9595575285238667,
      "learning_rate": 1.0040406892712192e-05,
      "loss": 0.0819,
      "step": 8030
    },
    {
      "epoch": 5.138195777351248,
      "grad_norm": 1.1112444654471405,
      "learning_rate": 1.0038334754792821e-05,
      "loss": 0.0856,
      "step": 8031
    },
    {
      "epoch": 5.138835572616762,
      "grad_norm": 0.9434805120255036,
      "learning_rate": 1.003626261522742e-05,
      "loss": 0.0817,
      "step": 8032
    },
    {
      "epoch": 5.1394753678822775,
      "grad_norm": 0.9539115645213877,
      "learning_rate": 1.0034190474104968e-05,
      "loss": 0.0841,
      "step": 8033
    },
    {
      "epoch": 5.140115163147793,
      "grad_norm": 1.063215126783227,
      "learning_rate": 1.0032118331514438e-05,
      "loss": 0.0932,
      "step": 8034
    },
    {
      "epoch": 5.140754958413308,
      "grad_norm": 0.8955762761093878,
      "learning_rate": 1.0030046187544804e-05,
      "loss": 0.0797,
      "step": 8035
    },
    {
      "epoch": 5.1413947536788225,
      "grad_norm": 0.8985112005429761,
      "learning_rate": 1.002797404228504e-05,
      "loss": 0.0773,
      "step": 8036
    },
    {
      "epoch": 5.142034548944338,
      "grad_norm": 1.079710883015959,
      "learning_rate": 1.002590189582412e-05,
      "loss": 0.0854,
      "step": 8037
    },
    {
      "epoch": 5.142674344209853,
      "grad_norm": 0.9979648072165999,
      "learning_rate": 1.0023829748251021e-05,
      "loss": 0.0901,
      "step": 8038
    },
    {
      "epoch": 5.143314139475368,
      "grad_norm": 1.0020305861391992,
      "learning_rate": 1.0021757599654715e-05,
      "loss": 0.0832,
      "step": 8039
    },
    {
      "epoch": 5.143953934740883,
      "grad_norm": 1.1633307034737403,
      "learning_rate": 1.0019685450124176e-05,
      "loss": 0.0946,
      "step": 8040
    },
    {
      "epoch": 5.144593730006398,
      "grad_norm": 0.970228716208799,
      "learning_rate": 1.0017613299748376e-05,
      "loss": 0.0864,
      "step": 8041
    },
    {
      "epoch": 5.145233525271913,
      "grad_norm": 1.0326879567195637,
      "learning_rate": 1.0015541148616294e-05,
      "loss": 0.0937,
      "step": 8042
    },
    {
      "epoch": 5.145873320537428,
      "grad_norm": 0.9661872395187012,
      "learning_rate": 1.0013468996816905e-05,
      "loss": 0.0836,
      "step": 8043
    },
    {
      "epoch": 5.146513115802943,
      "grad_norm": 1.0636216224694128,
      "learning_rate": 1.0011396844439176e-05,
      "loss": 0.0891,
      "step": 8044
    },
    {
      "epoch": 5.147152911068458,
      "grad_norm": 1.031608499958614,
      "learning_rate": 1.0009324691572091e-05,
      "loss": 0.0899,
      "step": 8045
    },
    {
      "epoch": 5.1477927063339735,
      "grad_norm": 1.1271540976830654,
      "learning_rate": 1.0007252538304619e-05,
      "loss": 0.086,
      "step": 8046
    },
    {
      "epoch": 5.148432501599488,
      "grad_norm": 1.0035787619949217,
      "learning_rate": 1.0005180384725736e-05,
      "loss": 0.0885,
      "step": 8047
    },
    {
      "epoch": 5.149072296865003,
      "grad_norm": 1.1031771430469655,
      "learning_rate": 1.0003108230924418e-05,
      "loss": 0.0931,
      "step": 8048
    },
    {
      "epoch": 5.149712092130518,
      "grad_norm": 1.1526512226495158,
      "learning_rate": 1.0001036076989635e-05,
      "loss": 0.0941,
      "step": 8049
    },
    {
      "epoch": 5.150351887396034,
      "grad_norm": 0.9179899049516935,
      "learning_rate": 9.998963923010365e-06,
      "loss": 0.0798,
      "step": 8050
    },
    {
      "epoch": 5.150991682661548,
      "grad_norm": 0.9288669013802215,
      "learning_rate": 9.996891769075585e-06,
      "loss": 0.0899,
      "step": 8051
    },
    {
      "epoch": 5.151631477927063,
      "grad_norm": 0.9935201143867994,
      "learning_rate": 9.994819615274267e-06,
      "loss": 0.0863,
      "step": 8052
    },
    {
      "epoch": 5.152271273192579,
      "grad_norm": 1.1113454975317183,
      "learning_rate": 9.992747461695383e-06,
      "loss": 0.0895,
      "step": 8053
    },
    {
      "epoch": 5.152911068458093,
      "grad_norm": 1.1723983539661493,
      "learning_rate": 9.990675308427912e-06,
      "loss": 0.0934,
      "step": 8054
    },
    {
      "epoch": 5.153550863723608,
      "grad_norm": 1.0908296488177867,
      "learning_rate": 9.988603155560826e-06,
      "loss": 0.0862,
      "step": 8055
    },
    {
      "epoch": 5.1541906589891235,
      "grad_norm": 1.0790778655490447,
      "learning_rate": 9.986531003183102e-06,
      "loss": 0.0863,
      "step": 8056
    },
    {
      "epoch": 5.154830454254639,
      "grad_norm": 1.1287442504546152,
      "learning_rate": 9.984458851383707e-06,
      "loss": 0.0895,
      "step": 8057
    },
    {
      "epoch": 5.155470249520153,
      "grad_norm": 1.0149409727982137,
      "learning_rate": 9.982386700251625e-06,
      "loss": 0.0903,
      "step": 8058
    },
    {
      "epoch": 5.1561100447856685,
      "grad_norm": 1.0101146285541982,
      "learning_rate": 9.980314549875829e-06,
      "loss": 0.0837,
      "step": 8059
    },
    {
      "epoch": 5.156749840051184,
      "grad_norm": 1.0470540636571457,
      "learning_rate": 9.978242400345288e-06,
      "loss": 0.0952,
      "step": 8060
    },
    {
      "epoch": 5.157389635316699,
      "grad_norm": 0.9575307801693094,
      "learning_rate": 9.976170251748982e-06,
      "loss": 0.0804,
      "step": 8061
    },
    {
      "epoch": 5.158029430582213,
      "grad_norm": 1.013868709535358,
      "learning_rate": 9.974098104175882e-06,
      "loss": 0.0887,
      "step": 8062
    },
    {
      "epoch": 5.158669225847729,
      "grad_norm": 1.0757555389240878,
      "learning_rate": 9.97202595771496e-06,
      "loss": 0.0886,
      "step": 8063
    },
    {
      "epoch": 5.159309021113244,
      "grad_norm": 1.0231593099780687,
      "learning_rate": 9.969953812455198e-06,
      "loss": 0.0854,
      "step": 8064
    },
    {
      "epoch": 5.159948816378758,
      "grad_norm": 1.0651522072388475,
      "learning_rate": 9.967881668485566e-06,
      "loss": 0.0942,
      "step": 8065
    },
    {
      "epoch": 5.160588611644274,
      "grad_norm": 1.022116315042478,
      "learning_rate": 9.965809525895037e-06,
      "loss": 0.0906,
      "step": 8066
    },
    {
      "epoch": 5.161228406909789,
      "grad_norm": 1.1394208621343525,
      "learning_rate": 9.963737384772582e-06,
      "loss": 0.0937,
      "step": 8067
    },
    {
      "epoch": 5.161868202175304,
      "grad_norm": 1.0505441533157265,
      "learning_rate": 9.961665245207184e-06,
      "loss": 0.0833,
      "step": 8068
    },
    {
      "epoch": 5.162507997440819,
      "grad_norm": 1.0693728676390852,
      "learning_rate": 9.959593107287812e-06,
      "loss": 0.0936,
      "step": 8069
    },
    {
      "epoch": 5.163147792706334,
      "grad_norm": 1.0388226181007691,
      "learning_rate": 9.957520971103439e-06,
      "loss": 0.0837,
      "step": 8070
    },
    {
      "epoch": 5.163787587971849,
      "grad_norm": 0.8572864517811939,
      "learning_rate": 9.95544883674304e-06,
      "loss": 0.0767,
      "step": 8071
    },
    {
      "epoch": 5.164427383237364,
      "grad_norm": 1.0919217173912081,
      "learning_rate": 9.95337670429559e-06,
      "loss": 0.09,
      "step": 8072
    },
    {
      "epoch": 5.165067178502879,
      "grad_norm": 1.1240767161781384,
      "learning_rate": 9.951304573850056e-06,
      "loss": 0.0913,
      "step": 8073
    },
    {
      "epoch": 5.165706973768394,
      "grad_norm": 0.9661542407398334,
      "learning_rate": 9.949232445495422e-06,
      "loss": 0.0924,
      "step": 8074
    },
    {
      "epoch": 5.166346769033909,
      "grad_norm": 0.9534183286425193,
      "learning_rate": 9.947160319320656e-06,
      "loss": 0.0804,
      "step": 8075
    },
    {
      "epoch": 5.166986564299425,
      "grad_norm": 1.0226195127866051,
      "learning_rate": 9.945088195414734e-06,
      "loss": 0.0816,
      "step": 8076
    },
    {
      "epoch": 5.167626359564939,
      "grad_norm": 0.990505940129635,
      "learning_rate": 9.943016073866624e-06,
      "loss": 0.0865,
      "step": 8077
    },
    {
      "epoch": 5.168266154830454,
      "grad_norm": 1.1542446064981409,
      "learning_rate": 9.940943954765304e-06,
      "loss": 0.0948,
      "step": 8078
    },
    {
      "epoch": 5.1689059500959695,
      "grad_norm": 0.9306824674659482,
      "learning_rate": 9.938871838199745e-06,
      "loss": 0.0784,
      "step": 8079
    },
    {
      "epoch": 5.169545745361484,
      "grad_norm": 1.1691329948432128,
      "learning_rate": 9.936799724258917e-06,
      "loss": 0.0933,
      "step": 8080
    },
    {
      "epoch": 5.170185540626999,
      "grad_norm": 0.9479717997224026,
      "learning_rate": 9.9347276130318e-06,
      "loss": 0.0824,
      "step": 8081
    },
    {
      "epoch": 5.1708253358925145,
      "grad_norm": 1.0463580534227412,
      "learning_rate": 9.932655504607365e-06,
      "loss": 0.0916,
      "step": 8082
    },
    {
      "epoch": 5.17146513115803,
      "grad_norm": 1.0227656337193485,
      "learning_rate": 9.930583399074579e-06,
      "loss": 0.084,
      "step": 8083
    },
    {
      "epoch": 5.172104926423544,
      "grad_norm": 0.988528693067696,
      "learning_rate": 9.928511296522421e-06,
      "loss": 0.0854,
      "step": 8084
    },
    {
      "epoch": 5.172744721689059,
      "grad_norm": 0.897332786008926,
      "learning_rate": 9.92643919703986e-06,
      "loss": 0.0826,
      "step": 8085
    },
    {
      "epoch": 5.173384516954575,
      "grad_norm": 1.0427464249156666,
      "learning_rate": 9.92436710071587e-06,
      "loss": 0.0868,
      "step": 8086
    },
    {
      "epoch": 5.17402431222009,
      "grad_norm": 1.058224517222961,
      "learning_rate": 9.922295007639417e-06,
      "loss": 0.0905,
      "step": 8087
    },
    {
      "epoch": 5.174664107485604,
      "grad_norm": 0.9886735347584638,
      "learning_rate": 9.920222917899483e-06,
      "loss": 0.0911,
      "step": 8088
    },
    {
      "epoch": 5.17530390275112,
      "grad_norm": 0.9953282660045859,
      "learning_rate": 9.918150831585036e-06,
      "loss": 0.0824,
      "step": 8089
    },
    {
      "epoch": 5.175943698016635,
      "grad_norm": 0.9264904676308982,
      "learning_rate": 9.916078748785043e-06,
      "loss": 0.0801,
      "step": 8090
    },
    {
      "epoch": 5.176583493282149,
      "grad_norm": 1.0803361400717455,
      "learning_rate": 9.914006669588481e-06,
      "loss": 0.0903,
      "step": 8091
    },
    {
      "epoch": 5.177223288547665,
      "grad_norm": 0.9992722638165846,
      "learning_rate": 9.911934594084322e-06,
      "loss": 0.0892,
      "step": 8092
    },
    {
      "epoch": 5.17786308381318,
      "grad_norm": 0.9906957093163773,
      "learning_rate": 9.909862522361533e-06,
      "loss": 0.0872,
      "step": 8093
    },
    {
      "epoch": 5.178502879078695,
      "grad_norm": 0.9456955089652057,
      "learning_rate": 9.907790454509083e-06,
      "loss": 0.0831,
      "step": 8094
    },
    {
      "epoch": 5.1791426743442095,
      "grad_norm": 1.0476895878139303,
      "learning_rate": 9.905718390615952e-06,
      "loss": 0.092,
      "step": 8095
    },
    {
      "epoch": 5.179782469609725,
      "grad_norm": 1.0132077665992623,
      "learning_rate": 9.903646330771105e-06,
      "loss": 0.0933,
      "step": 8096
    },
    {
      "epoch": 5.18042226487524,
      "grad_norm": 1.0310989926278173,
      "learning_rate": 9.901574275063508e-06,
      "loss": 0.0919,
      "step": 8097
    },
    {
      "epoch": 5.181062060140755,
      "grad_norm": 0.9938286526380045,
      "learning_rate": 9.899502223582142e-06,
      "loss": 0.0856,
      "step": 8098
    },
    {
      "epoch": 5.18170185540627,
      "grad_norm": 0.9605018239647056,
      "learning_rate": 9.897430176415972e-06,
      "loss": 0.0892,
      "step": 8099
    },
    {
      "epoch": 5.182341650671785,
      "grad_norm": 1.134446228800222,
      "learning_rate": 9.895358133653964e-06,
      "loss": 0.0871,
      "step": 8100
    },
    {
      "epoch": 5.1829814459373,
      "grad_norm": 1.0139795757202468,
      "learning_rate": 9.893286095385096e-06,
      "loss": 0.0855,
      "step": 8101
    },
    {
      "epoch": 5.183621241202815,
      "grad_norm": 1.1372540901186132,
      "learning_rate": 9.891214061698332e-06,
      "loss": 0.0917,
      "step": 8102
    },
    {
      "epoch": 5.18426103646833,
      "grad_norm": 0.9633219412535584,
      "learning_rate": 9.88914203268264e-06,
      "loss": 0.0827,
      "step": 8103
    },
    {
      "epoch": 5.184900831733845,
      "grad_norm": 1.0316076342266134,
      "learning_rate": 9.887070008426992e-06,
      "loss": 0.0872,
      "step": 8104
    },
    {
      "epoch": 5.1855406269993605,
      "grad_norm": 1.0320162269606348,
      "learning_rate": 9.88499798902036e-06,
      "loss": 0.0839,
      "step": 8105
    },
    {
      "epoch": 5.186180422264875,
      "grad_norm": 0.9848676980922443,
      "learning_rate": 9.882925974551708e-06,
      "loss": 0.0819,
      "step": 8106
    },
    {
      "epoch": 5.18682021753039,
      "grad_norm": 1.1481051367747583,
      "learning_rate": 9.880853965110006e-06,
      "loss": 0.0949,
      "step": 8107
    },
    {
      "epoch": 5.187460012795905,
      "grad_norm": 1.0482402785173333,
      "learning_rate": 9.878781960784225e-06,
      "loss": 0.0933,
      "step": 8108
    },
    {
      "epoch": 5.188099808061421,
      "grad_norm": 1.0250225402025313,
      "learning_rate": 9.87670996166333e-06,
      "loss": 0.0847,
      "step": 8109
    },
    {
      "epoch": 5.188739603326935,
      "grad_norm": 0.873010321277344,
      "learning_rate": 9.874637967836286e-06,
      "loss": 0.0757,
      "step": 8110
    },
    {
      "epoch": 5.18937939859245,
      "grad_norm": 1.0811822390511123,
      "learning_rate": 9.87256597939207e-06,
      "loss": 0.0943,
      "step": 8111
    },
    {
      "epoch": 5.190019193857966,
      "grad_norm": 0.9946331109082861,
      "learning_rate": 9.870493996419643e-06,
      "loss": 0.085,
      "step": 8112
    },
    {
      "epoch": 5.190658989123481,
      "grad_norm": 0.9994203914172715,
      "learning_rate": 9.868422019007972e-06,
      "loss": 0.0831,
      "step": 8113
    },
    {
      "epoch": 5.191298784388995,
      "grad_norm": 1.0245457916891618,
      "learning_rate": 9.866350047246025e-06,
      "loss": 0.0837,
      "step": 8114
    },
    {
      "epoch": 5.1919385796545106,
      "grad_norm": 1.090352631079339,
      "learning_rate": 9.864278081222772e-06,
      "loss": 0.0956,
      "step": 8115
    },
    {
      "epoch": 5.192578374920026,
      "grad_norm": 1.0163821054123408,
      "learning_rate": 9.862206121027176e-06,
      "loss": 0.0784,
      "step": 8116
    },
    {
      "epoch": 5.19321817018554,
      "grad_norm": 1.0958393301618512,
      "learning_rate": 9.8601341667482e-06,
      "loss": 0.0908,
      "step": 8117
    },
    {
      "epoch": 5.1938579654510555,
      "grad_norm": 1.0438412506547188,
      "learning_rate": 9.858062218474818e-06,
      "loss": 0.0882,
      "step": 8118
    },
    {
      "epoch": 5.194497760716571,
      "grad_norm": 0.9210358427922134,
      "learning_rate": 9.85599027629599e-06,
      "loss": 0.0806,
      "step": 8119
    },
    {
      "epoch": 5.195137555982086,
      "grad_norm": 0.9882770256152316,
      "learning_rate": 9.853918340300684e-06,
      "loss": 0.0875,
      "step": 8120
    },
    {
      "epoch": 5.1957773512476,
      "grad_norm": 0.9564023309847175,
      "learning_rate": 9.851846410577866e-06,
      "loss": 0.079,
      "step": 8121
    },
    {
      "epoch": 5.196417146513116,
      "grad_norm": 1.0727153301082175,
      "learning_rate": 9.8497744872165e-06,
      "loss": 0.0888,
      "step": 8122
    },
    {
      "epoch": 5.197056941778631,
      "grad_norm": 1.0193118031087725,
      "learning_rate": 9.847702570305551e-06,
      "loss": 0.0774,
      "step": 8123
    },
    {
      "epoch": 5.197696737044146,
      "grad_norm": 1.0347436856335297,
      "learning_rate": 9.845630659933977e-06,
      "loss": 0.0881,
      "step": 8124
    },
    {
      "epoch": 5.198336532309661,
      "grad_norm": 0.9258941491790607,
      "learning_rate": 9.843558756190753e-06,
      "loss": 0.0842,
      "step": 8125
    },
    {
      "epoch": 5.198976327575176,
      "grad_norm": 1.0475205244395884,
      "learning_rate": 9.841486859164836e-06,
      "loss": 0.0821,
      "step": 8126
    },
    {
      "epoch": 5.199616122840691,
      "grad_norm": 1.0167393881328788,
      "learning_rate": 9.839414968945187e-06,
      "loss": 0.0903,
      "step": 8127
    },
    {
      "epoch": 5.2002559181062065,
      "grad_norm": 0.9669970034273734,
      "learning_rate": 9.83734308562078e-06,
      "loss": 0.0815,
      "step": 8128
    },
    {
      "epoch": 5.200895713371721,
      "grad_norm": 1.0464762314452256,
      "learning_rate": 9.83527120928057e-06,
      "loss": 0.0836,
      "step": 8129
    },
    {
      "epoch": 5.201535508637236,
      "grad_norm": 0.9006860101638013,
      "learning_rate": 9.833199340013521e-06,
      "loss": 0.0744,
      "step": 8130
    },
    {
      "epoch": 5.202175303902751,
      "grad_norm": 0.9981387812590311,
      "learning_rate": 9.831127477908595e-06,
      "loss": 0.0863,
      "step": 8131
    },
    {
      "epoch": 5.202815099168266,
      "grad_norm": 0.9424905606274168,
      "learning_rate": 9.829055623054757e-06,
      "loss": 0.0841,
      "step": 8132
    },
    {
      "epoch": 5.203454894433781,
      "grad_norm": 1.029009110390848,
      "learning_rate": 9.826983775540965e-06,
      "loss": 0.0938,
      "step": 8133
    },
    {
      "epoch": 5.204094689699296,
      "grad_norm": 1.089728521156135,
      "learning_rate": 9.824911935456178e-06,
      "loss": 0.0945,
      "step": 8134
    },
    {
      "epoch": 5.204734484964812,
      "grad_norm": 1.0151089906970514,
      "learning_rate": 9.822840102889366e-06,
      "loss": 0.0831,
      "step": 8135
    },
    {
      "epoch": 5.205374280230326,
      "grad_norm": 0.9072407723458347,
      "learning_rate": 9.820768277929484e-06,
      "loss": 0.0811,
      "step": 8136
    },
    {
      "epoch": 5.206014075495841,
      "grad_norm": 0.9245997261318225,
      "learning_rate": 9.818696460665493e-06,
      "loss": 0.0799,
      "step": 8137
    },
    {
      "epoch": 5.2066538707613566,
      "grad_norm": 0.9225195375157947,
      "learning_rate": 9.816624651186352e-06,
      "loss": 0.0871,
      "step": 8138
    },
    {
      "epoch": 5.207293666026872,
      "grad_norm": 0.95948684079286,
      "learning_rate": 9.814552849581027e-06,
      "loss": 0.0909,
      "step": 8139
    },
    {
      "epoch": 5.207933461292386,
      "grad_norm": 0.9852500035581467,
      "learning_rate": 9.812481055938463e-06,
      "loss": 0.085,
      "step": 8140
    },
    {
      "epoch": 5.2085732565579015,
      "grad_norm": 1.0708233716852724,
      "learning_rate": 9.810409270347638e-06,
      "loss": 0.0976,
      "step": 8141
    },
    {
      "epoch": 5.209213051823417,
      "grad_norm": 1.0094168714492706,
      "learning_rate": 9.8083374928975e-06,
      "loss": 0.0905,
      "step": 8142
    },
    {
      "epoch": 5.209852847088931,
      "grad_norm": 1.1089315327798062,
      "learning_rate": 9.806265723677009e-06,
      "loss": 0.0898,
      "step": 8143
    },
    {
      "epoch": 5.210492642354446,
      "grad_norm": 1.163999105038912,
      "learning_rate": 9.80419396277512e-06,
      "loss": 0.0972,
      "step": 8144
    },
    {
      "epoch": 5.211132437619962,
      "grad_norm": 1.1176215334359525,
      "learning_rate": 9.802122210280797e-06,
      "loss": 0.0946,
      "step": 8145
    },
    {
      "epoch": 5.211772232885477,
      "grad_norm": 0.8755976657395262,
      "learning_rate": 9.800050466282994e-06,
      "loss": 0.0844,
      "step": 8146
    },
    {
      "epoch": 5.212412028150991,
      "grad_norm": 1.0390891669818367,
      "learning_rate": 9.797978730870664e-06,
      "loss": 0.0862,
      "step": 8147
    },
    {
      "epoch": 5.213051823416507,
      "grad_norm": 1.2394984987892639,
      "learning_rate": 9.795907004132773e-06,
      "loss": 0.092,
      "step": 8148
    },
    {
      "epoch": 5.213691618682022,
      "grad_norm": 1.0990056871689826,
      "learning_rate": 9.793835286158272e-06,
      "loss": 0.0932,
      "step": 8149
    },
    {
      "epoch": 5.214331413947537,
      "grad_norm": 0.9724803401390592,
      "learning_rate": 9.791763577036114e-06,
      "loss": 0.0902,
      "step": 8150
    },
    {
      "epoch": 5.214971209213052,
      "grad_norm": 1.0476491522014015,
      "learning_rate": 9.78969187685526e-06,
      "loss": 0.0832,
      "step": 8151
    },
    {
      "epoch": 5.215611004478567,
      "grad_norm": 0.9814674273786927,
      "learning_rate": 9.787620185704664e-06,
      "loss": 0.0879,
      "step": 8152
    },
    {
      "epoch": 5.216250799744082,
      "grad_norm": 1.1056285431136679,
      "learning_rate": 9.785548503673278e-06,
      "loss": 0.0916,
      "step": 8153
    },
    {
      "epoch": 5.2168905950095965,
      "grad_norm": 0.9721717767979767,
      "learning_rate": 9.783476830850054e-06,
      "loss": 0.081,
      "step": 8154
    },
    {
      "epoch": 5.217530390275112,
      "grad_norm": 0.9872404606322692,
      "learning_rate": 9.781405167323954e-06,
      "loss": 0.0875,
      "step": 8155
    },
    {
      "epoch": 5.218170185540627,
      "grad_norm": 1.0657116430432891,
      "learning_rate": 9.779333513183927e-06,
      "loss": 0.0875,
      "step": 8156
    },
    {
      "epoch": 5.218809980806142,
      "grad_norm": 0.9958376866645005,
      "learning_rate": 9.777261868518923e-06,
      "loss": 0.091,
      "step": 8157
    },
    {
      "epoch": 5.219449776071657,
      "grad_norm": 0.9217416369249233,
      "learning_rate": 9.775190233417901e-06,
      "loss": 0.0933,
      "step": 8158
    },
    {
      "epoch": 5.220089571337172,
      "grad_norm": 1.0475676103131235,
      "learning_rate": 9.773118607969811e-06,
      "loss": 0.0883,
      "step": 8159
    },
    {
      "epoch": 5.220729366602687,
      "grad_norm": 1.1079794848723565,
      "learning_rate": 9.771046992263603e-06,
      "loss": 0.093,
      "step": 8160
    },
    {
      "epoch": 5.2213691618682025,
      "grad_norm": 1.0299191979399014,
      "learning_rate": 9.768975386388232e-06,
      "loss": 0.093,
      "step": 8161
    },
    {
      "epoch": 5.222008957133717,
      "grad_norm": 1.0285480825118498,
      "learning_rate": 9.766903790432645e-06,
      "loss": 0.0851,
      "step": 8162
    },
    {
      "epoch": 5.222648752399232,
      "grad_norm": 0.916648705000994,
      "learning_rate": 9.764832204485798e-06,
      "loss": 0.0837,
      "step": 8163
    },
    {
      "epoch": 5.2232885476647475,
      "grad_norm": 0.9861410635979301,
      "learning_rate": 9.76276062863663e-06,
      "loss": 0.0864,
      "step": 8164
    },
    {
      "epoch": 5.223928342930263,
      "grad_norm": 0.9236663753131523,
      "learning_rate": 9.760689062974104e-06,
      "loss": 0.0788,
      "step": 8165
    },
    {
      "epoch": 5.224568138195777,
      "grad_norm": 1.0778687131370386,
      "learning_rate": 9.758617507587165e-06,
      "loss": 0.0968,
      "step": 8166
    },
    {
      "epoch": 5.225207933461292,
      "grad_norm": 1.0353645488486702,
      "learning_rate": 9.756545962564759e-06,
      "loss": 0.0856,
      "step": 8167
    },
    {
      "epoch": 5.225847728726808,
      "grad_norm": 0.9192030877222748,
      "learning_rate": 9.754474427995836e-06,
      "loss": 0.0833,
      "step": 8168
    },
    {
      "epoch": 5.226487523992322,
      "grad_norm": 1.0152067315907098,
      "learning_rate": 9.752402903969347e-06,
      "loss": 0.0851,
      "step": 8169
    },
    {
      "epoch": 5.227127319257837,
      "grad_norm": 0.9819985494718897,
      "learning_rate": 9.750331390574236e-06,
      "loss": 0.0817,
      "step": 8170
    },
    {
      "epoch": 5.227767114523353,
      "grad_norm": 1.0238185826243074,
      "learning_rate": 9.748259887899447e-06,
      "loss": 0.0871,
      "step": 8171
    },
    {
      "epoch": 5.228406909788868,
      "grad_norm": 1.064223678490073,
      "learning_rate": 9.746188396033933e-06,
      "loss": 0.0902,
      "step": 8172
    },
    {
      "epoch": 5.229046705054382,
      "grad_norm": 0.948767357752723,
      "learning_rate": 9.74411691506664e-06,
      "loss": 0.0884,
      "step": 8173
    },
    {
      "epoch": 5.229686500319898,
      "grad_norm": 0.9525038933353291,
      "learning_rate": 9.742045445086507e-06,
      "loss": 0.0828,
      "step": 8174
    },
    {
      "epoch": 5.230326295585413,
      "grad_norm": 1.0366948681528132,
      "learning_rate": 9.739973986182488e-06,
      "loss": 0.0891,
      "step": 8175
    },
    {
      "epoch": 5.230966090850928,
      "grad_norm": 1.017102303558964,
      "learning_rate": 9.737902538443523e-06,
      "loss": 0.0913,
      "step": 8176
    },
    {
      "epoch": 5.2316058861164425,
      "grad_norm": 1.042964641909624,
      "learning_rate": 9.735831101958551e-06,
      "loss": 0.0888,
      "step": 8177
    },
    {
      "epoch": 5.232245681381958,
      "grad_norm": 1.0098154462990983,
      "learning_rate": 9.733759676816528e-06,
      "loss": 0.0833,
      "step": 8178
    },
    {
      "epoch": 5.232885476647473,
      "grad_norm": 0.9616747978563405,
      "learning_rate": 9.73168826310639e-06,
      "loss": 0.0877,
      "step": 8179
    },
    {
      "epoch": 5.233525271912987,
      "grad_norm": 0.9991825296368786,
      "learning_rate": 9.72961686091708e-06,
      "loss": 0.0885,
      "step": 8180
    },
    {
      "epoch": 5.234165067178503,
      "grad_norm": 1.0409550707192592,
      "learning_rate": 9.72754547033754e-06,
      "loss": 0.0851,
      "step": 8181
    },
    {
      "epoch": 5.234804862444018,
      "grad_norm": 1.0203795097574277,
      "learning_rate": 9.725474091456715e-06,
      "loss": 0.0907,
      "step": 8182
    },
    {
      "epoch": 5.235444657709533,
      "grad_norm": 0.9630616017493594,
      "learning_rate": 9.723402724363543e-06,
      "loss": 0.0824,
      "step": 8183
    },
    {
      "epoch": 5.236084452975048,
      "grad_norm": 1.059344311471976,
      "learning_rate": 9.721331369146963e-06,
      "loss": 0.0811,
      "step": 8184
    },
    {
      "epoch": 5.236724248240563,
      "grad_norm": 1.202354126368939,
      "learning_rate": 9.71926002589592e-06,
      "loss": 0.096,
      "step": 8185
    },
    {
      "epoch": 5.237364043506078,
      "grad_norm": 1.014548240741482,
      "learning_rate": 9.717188694699355e-06,
      "loss": 0.0816,
      "step": 8186
    },
    {
      "epoch": 5.2380038387715935,
      "grad_norm": 1.08852561820811,
      "learning_rate": 9.7151173756462e-06,
      "loss": 0.0946,
      "step": 8187
    },
    {
      "epoch": 5.238643634037108,
      "grad_norm": 1.0360884480903991,
      "learning_rate": 9.713046068825402e-06,
      "loss": 0.0867,
      "step": 8188
    },
    {
      "epoch": 5.239283429302623,
      "grad_norm": 1.0289741190662343,
      "learning_rate": 9.710974774325893e-06,
      "loss": 0.0891,
      "step": 8189
    },
    {
      "epoch": 5.239923224568138,
      "grad_norm": 0.9749472001284627,
      "learning_rate": 9.708903492236615e-06,
      "loss": 0.0861,
      "step": 8190
    },
    {
      "epoch": 5.240563019833653,
      "grad_norm": 1.0430987389009385,
      "learning_rate": 9.706832222646502e-06,
      "loss": 0.096,
      "step": 8191
    },
    {
      "epoch": 5.241202815099168,
      "grad_norm": 1.0973136196887803,
      "learning_rate": 9.704760965644493e-06,
      "loss": 0.0895,
      "step": 8192
    },
    {
      "epoch": 5.241842610364683,
      "grad_norm": 0.9655491491666465,
      "learning_rate": 9.702689721319522e-06,
      "loss": 0.0814,
      "step": 8193
    },
    {
      "epoch": 5.242482405630199,
      "grad_norm": 1.0465274015979873,
      "learning_rate": 9.700618489760522e-06,
      "loss": 0.0863,
      "step": 8194
    },
    {
      "epoch": 5.243122200895713,
      "grad_norm": 1.1248939769423734,
      "learning_rate": 9.698547271056436e-06,
      "loss": 0.0862,
      "step": 8195
    },
    {
      "epoch": 5.243761996161228,
      "grad_norm": 1.0117827005749438,
      "learning_rate": 9.696476065296195e-06,
      "loss": 0.0896,
      "step": 8196
    },
    {
      "epoch": 5.244401791426744,
      "grad_norm": 0.9959827500261105,
      "learning_rate": 9.694404872568728e-06,
      "loss": 0.0852,
      "step": 8197
    },
    {
      "epoch": 5.245041586692259,
      "grad_norm": 0.9273702241083809,
      "learning_rate": 9.692333692962976e-06,
      "loss": 0.0852,
      "step": 8198
    },
    {
      "epoch": 5.245681381957773,
      "grad_norm": 0.9568993467389298,
      "learning_rate": 9.690262526567866e-06,
      "loss": 0.0829,
      "step": 8199
    },
    {
      "epoch": 5.2463211772232885,
      "grad_norm": 1.004265363086918,
      "learning_rate": 9.688191373472332e-06,
      "loss": 0.0918,
      "step": 8200
    },
    {
      "epoch": 5.246960972488804,
      "grad_norm": 0.9807994820098602,
      "learning_rate": 9.686120233765303e-06,
      "loss": 0.0859,
      "step": 8201
    },
    {
      "epoch": 5.247600767754319,
      "grad_norm": 1.0256792480162087,
      "learning_rate": 9.684049107535716e-06,
      "loss": 0.0919,
      "step": 8202
    },
    {
      "epoch": 5.248240563019833,
      "grad_norm": 0.9920845766493267,
      "learning_rate": 9.6819779948725e-06,
      "loss": 0.091,
      "step": 8203
    },
    {
      "epoch": 5.248880358285349,
      "grad_norm": 0.9350170136928082,
      "learning_rate": 9.67990689586458e-06,
      "loss": 0.0813,
      "step": 8204
    },
    {
      "epoch": 5.249520153550864,
      "grad_norm": 1.0133772193176247,
      "learning_rate": 9.67783581060089e-06,
      "loss": 0.09,
      "step": 8205
    },
    {
      "epoch": 5.250159948816378,
      "grad_norm": 0.9348151531079529,
      "learning_rate": 9.675764739170357e-06,
      "loss": 0.0768,
      "step": 8206
    },
    {
      "epoch": 5.250799744081894,
      "grad_norm": 1.025132169187528,
      "learning_rate": 9.673693681661905e-06,
      "loss": 0.0957,
      "step": 8207
    },
    {
      "epoch": 5.251439539347409,
      "grad_norm": 1.0187205303806535,
      "learning_rate": 9.671622638164471e-06,
      "loss": 0.086,
      "step": 8208
    },
    {
      "epoch": 5.252079334612924,
      "grad_norm": 1.1253863055313356,
      "learning_rate": 9.669551608766978e-06,
      "loss": 0.0937,
      "step": 8209
    },
    {
      "epoch": 5.252719129878439,
      "grad_norm": 1.0023850570828206,
      "learning_rate": 9.66748059355835e-06,
      "loss": 0.0813,
      "step": 8210
    },
    {
      "epoch": 5.253358925143954,
      "grad_norm": 1.0195925792806975,
      "learning_rate": 9.66540959262751e-06,
      "loss": 0.0826,
      "step": 8211
    },
    {
      "epoch": 5.253998720409469,
      "grad_norm": 1.043609515034376,
      "learning_rate": 9.66333860606339e-06,
      "loss": 0.0862,
      "step": 8212
    },
    {
      "epoch": 5.254638515674984,
      "grad_norm": 0.9838255690588528,
      "learning_rate": 9.661267633954913e-06,
      "loss": 0.0813,
      "step": 8213
    },
    {
      "epoch": 5.255278310940499,
      "grad_norm": 0.9474815915364909,
      "learning_rate": 9.659196676390995e-06,
      "loss": 0.0844,
      "step": 8214
    },
    {
      "epoch": 5.255918106206014,
      "grad_norm": 1.0329878107512072,
      "learning_rate": 9.65712573346057e-06,
      "loss": 0.0856,
      "step": 8215
    },
    {
      "epoch": 5.256557901471529,
      "grad_norm": 1.0733158495345458,
      "learning_rate": 9.655054805252557e-06,
      "loss": 0.0855,
      "step": 8216
    },
    {
      "epoch": 5.257197696737044,
      "grad_norm": 1.0801112456617672,
      "learning_rate": 9.652983891855873e-06,
      "loss": 0.0884,
      "step": 8217
    },
    {
      "epoch": 5.257837492002559,
      "grad_norm": 0.928450808946394,
      "learning_rate": 9.650912993359446e-06,
      "loss": 0.0828,
      "step": 8218
    },
    {
      "epoch": 5.258477287268074,
      "grad_norm": 1.0436233275044435,
      "learning_rate": 9.648842109852196e-06,
      "loss": 0.088,
      "step": 8219
    },
    {
      "epoch": 5.25911708253359,
      "grad_norm": 1.0483092272681485,
      "learning_rate": 9.646771241423041e-06,
      "loss": 0.0909,
      "step": 8220
    },
    {
      "epoch": 5.259756877799104,
      "grad_norm": 1.033107936274786,
      "learning_rate": 9.644700388160898e-06,
      "loss": 0.082,
      "step": 8221
    },
    {
      "epoch": 5.260396673064619,
      "grad_norm": 1.0879647347484303,
      "learning_rate": 9.64262955015469e-06,
      "loss": 0.095,
      "step": 8222
    },
    {
      "epoch": 5.2610364683301345,
      "grad_norm": 1.0299300386557844,
      "learning_rate": 9.640558727493333e-06,
      "loss": 0.0881,
      "step": 8223
    },
    {
      "epoch": 5.26167626359565,
      "grad_norm": 0.9559537534662743,
      "learning_rate": 9.638487920265741e-06,
      "loss": 0.084,
      "step": 8224
    },
    {
      "epoch": 5.262316058861164,
      "grad_norm": 1.055691416783916,
      "learning_rate": 9.63641712856084e-06,
      "loss": 0.0842,
      "step": 8225
    },
    {
      "epoch": 5.262955854126679,
      "grad_norm": 1.1010246716237528,
      "learning_rate": 9.634346352467539e-06,
      "loss": 0.099,
      "step": 8226
    },
    {
      "epoch": 5.263595649392195,
      "grad_norm": 1.1204550925025543,
      "learning_rate": 9.632275592074754e-06,
      "loss": 0.0885,
      "step": 8227
    },
    {
      "epoch": 5.264235444657709,
      "grad_norm": 1.1120694631201598,
      "learning_rate": 9.630204847471401e-06,
      "loss": 0.0879,
      "step": 8228
    },
    {
      "epoch": 5.264875239923224,
      "grad_norm": 1.0420525027098442,
      "learning_rate": 9.628134118746397e-06,
      "loss": 0.0903,
      "step": 8229
    },
    {
      "epoch": 5.26551503518874,
      "grad_norm": 1.0772360250906725,
      "learning_rate": 9.626063405988651e-06,
      "loss": 0.0854,
      "step": 8230
    },
    {
      "epoch": 5.266154830454255,
      "grad_norm": 1.0522722758975651,
      "learning_rate": 9.623992709287073e-06,
      "loss": 0.0905,
      "step": 8231
    },
    {
      "epoch": 5.266794625719769,
      "grad_norm": 0.9683710193874244,
      "learning_rate": 9.621922028730584e-06,
      "loss": 0.0838,
      "step": 8232
    },
    {
      "epoch": 5.267434420985285,
      "grad_norm": 0.9862565623925772,
      "learning_rate": 9.61985136440809e-06,
      "loss": 0.0863,
      "step": 8233
    },
    {
      "epoch": 5.2680742162508,
      "grad_norm": 1.0060499724352683,
      "learning_rate": 9.6177807164085e-06,
      "loss": 0.096,
      "step": 8234
    },
    {
      "epoch": 5.268714011516315,
      "grad_norm": 1.1468480462372317,
      "learning_rate": 9.615710084820728e-06,
      "loss": 0.0886,
      "step": 8235
    },
    {
      "epoch": 5.2693538067818295,
      "grad_norm": 1.0911922325459005,
      "learning_rate": 9.613639469733684e-06,
      "loss": 0.094,
      "step": 8236
    },
    {
      "epoch": 5.269993602047345,
      "grad_norm": 1.016398568053289,
      "learning_rate": 9.611568871236267e-06,
      "loss": 0.0919,
      "step": 8237
    },
    {
      "epoch": 5.27063339731286,
      "grad_norm": 0.9977339883778297,
      "learning_rate": 9.609498289417397e-06,
      "loss": 0.0834,
      "step": 8238
    },
    {
      "epoch": 5.271273192578375,
      "grad_norm": 1.0252202486075537,
      "learning_rate": 9.607427724365974e-06,
      "loss": 0.0874,
      "step": 8239
    },
    {
      "epoch": 5.27191298784389,
      "grad_norm": 1.0733465325589837,
      "learning_rate": 9.605357176170907e-06,
      "loss": 0.0938,
      "step": 8240
    },
    {
      "epoch": 5.272552783109405,
      "grad_norm": 0.9002807834502647,
      "learning_rate": 9.6032866449211e-06,
      "loss": 0.0771,
      "step": 8241
    },
    {
      "epoch": 5.27319257837492,
      "grad_norm": 1.0684153080657237,
      "learning_rate": 9.60121613070546e-06,
      "loss": 0.0879,
      "step": 8242
    },
    {
      "epoch": 5.273832373640435,
      "grad_norm": 0.8814430494129433,
      "learning_rate": 9.59914563361289e-06,
      "loss": 0.0826,
      "step": 8243
    },
    {
      "epoch": 5.27447216890595,
      "grad_norm": 0.9854865301571911,
      "learning_rate": 9.597075153732289e-06,
      "loss": 0.079,
      "step": 8244
    },
    {
      "epoch": 5.275111964171465,
      "grad_norm": 1.0380819967652994,
      "learning_rate": 9.595004691152569e-06,
      "loss": 0.0954,
      "step": 8245
    },
    {
      "epoch": 5.2757517594369805,
      "grad_norm": 1.0576165449095065,
      "learning_rate": 9.592934245962626e-06,
      "loss": 0.0934,
      "step": 8246
    },
    {
      "epoch": 5.276391554702495,
      "grad_norm": 1.0034320747676373,
      "learning_rate": 9.590863818251364e-06,
      "loss": 0.0808,
      "step": 8247
    },
    {
      "epoch": 5.27703134996801,
      "grad_norm": 1.026884673852302,
      "learning_rate": 9.588793408107677e-06,
      "loss": 0.0839,
      "step": 8248
    },
    {
      "epoch": 5.277671145233525,
      "grad_norm": 1.0463720511585686,
      "learning_rate": 9.586723015620472e-06,
      "loss": 0.1012,
      "step": 8249
    },
    {
      "epoch": 5.278310940499041,
      "grad_norm": 1.006796706924782,
      "learning_rate": 9.584652640878649e-06,
      "loss": 0.0858,
      "step": 8250
    },
    {
      "epoch": 5.278950735764555,
      "grad_norm": 1.0567012487263525,
      "learning_rate": 9.582582283971098e-06,
      "loss": 0.0934,
      "step": 8251
    },
    {
      "epoch": 5.27959053103007,
      "grad_norm": 0.9674477664050748,
      "learning_rate": 9.580511944986723e-06,
      "loss": 0.0902,
      "step": 8252
    },
    {
      "epoch": 5.280230326295586,
      "grad_norm": 1.1693975533516825,
      "learning_rate": 9.57844162401442e-06,
      "loss": 0.1029,
      "step": 8253
    },
    {
      "epoch": 5.280870121561101,
      "grad_norm": 1.0168444922066975,
      "learning_rate": 9.57637132114308e-06,
      "loss": 0.0832,
      "step": 8254
    },
    {
      "epoch": 5.281509916826615,
      "grad_norm": 1.1270079580975272,
      "learning_rate": 9.574301036461606e-06,
      "loss": 0.0996,
      "step": 8255
    },
    {
      "epoch": 5.282149712092131,
      "grad_norm": 0.886042276325681,
      "learning_rate": 9.572230770058887e-06,
      "loss": 0.0796,
      "step": 8256
    },
    {
      "epoch": 5.282789507357646,
      "grad_norm": 1.0171321727661546,
      "learning_rate": 9.57016052202382e-06,
      "loss": 0.0848,
      "step": 8257
    },
    {
      "epoch": 5.28342930262316,
      "grad_norm": 1.006600751406872,
      "learning_rate": 9.568090292445291e-06,
      "loss": 0.0878,
      "step": 8258
    },
    {
      "epoch": 5.2840690978886755,
      "grad_norm": 1.015278153641355,
      "learning_rate": 9.5660200814122e-06,
      "loss": 0.088,
      "step": 8259
    },
    {
      "epoch": 5.284708893154191,
      "grad_norm": 0.9347894245777434,
      "learning_rate": 9.563949889013434e-06,
      "loss": 0.0764,
      "step": 8260
    },
    {
      "epoch": 5.285348688419706,
      "grad_norm": 1.0276243470052513,
      "learning_rate": 9.56187971533788e-06,
      "loss": 0.0761,
      "step": 8261
    },
    {
      "epoch": 5.2859884836852205,
      "grad_norm": 0.9388975015567229,
      "learning_rate": 9.559809560474436e-06,
      "loss": 0.0861,
      "step": 8262
    },
    {
      "epoch": 5.286628278950736,
      "grad_norm": 1.0643866120331316,
      "learning_rate": 9.557739424511986e-06,
      "loss": 0.0805,
      "step": 8263
    },
    {
      "epoch": 5.287268074216251,
      "grad_norm": 1.2099459708279647,
      "learning_rate": 9.555669307539417e-06,
      "loss": 0.0909,
      "step": 8264
    },
    {
      "epoch": 5.287907869481765,
      "grad_norm": 1.080241784860295,
      "learning_rate": 9.553599209645622e-06,
      "loss": 0.0908,
      "step": 8265
    },
    {
      "epoch": 5.288547664747281,
      "grad_norm": 0.9433431796366328,
      "learning_rate": 9.551529130919478e-06,
      "loss": 0.0777,
      "step": 8266
    },
    {
      "epoch": 5.289187460012796,
      "grad_norm": 1.0744589536609348,
      "learning_rate": 9.549459071449879e-06,
      "loss": 0.091,
      "step": 8267
    },
    {
      "epoch": 5.289827255278311,
      "grad_norm": 1.082815964487218,
      "learning_rate": 9.5473890313257e-06,
      "loss": 0.0927,
      "step": 8268
    },
    {
      "epoch": 5.290467050543826,
      "grad_norm": 1.0046200367358535,
      "learning_rate": 9.545319010635835e-06,
      "loss": 0.0796,
      "step": 8269
    },
    {
      "epoch": 5.291106845809341,
      "grad_norm": 1.076812947858891,
      "learning_rate": 9.543249009469164e-06,
      "loss": 0.0894,
      "step": 8270
    },
    {
      "epoch": 5.291746641074856,
      "grad_norm": 1.0866416519498512,
      "learning_rate": 9.541179027914566e-06,
      "loss": 0.0905,
      "step": 8271
    },
    {
      "epoch": 5.292386436340371,
      "grad_norm": 1.0861183579668219,
      "learning_rate": 9.539109066060926e-06,
      "loss": 0.0898,
      "step": 8272
    },
    {
      "epoch": 5.293026231605886,
      "grad_norm": 1.0528975403396754,
      "learning_rate": 9.537039123997121e-06,
      "loss": 0.0913,
      "step": 8273
    },
    {
      "epoch": 5.293666026871401,
      "grad_norm": 0.9643169157494366,
      "learning_rate": 9.534969201812032e-06,
      "loss": 0.082,
      "step": 8274
    },
    {
      "epoch": 5.294305822136916,
      "grad_norm": 0.9531406722952329,
      "learning_rate": 9.532899299594539e-06,
      "loss": 0.0847,
      "step": 8275
    },
    {
      "epoch": 5.294945617402432,
      "grad_norm": 1.0058543252008452,
      "learning_rate": 9.530829417433521e-06,
      "loss": 0.0863,
      "step": 8276
    },
    {
      "epoch": 5.295585412667946,
      "grad_norm": 1.1148566594657618,
      "learning_rate": 9.528759555417852e-06,
      "loss": 0.0974,
      "step": 8277
    },
    {
      "epoch": 5.296225207933461,
      "grad_norm": 1.0227485351593728,
      "learning_rate": 9.526689713636406e-06,
      "loss": 0.0873,
      "step": 8278
    },
    {
      "epoch": 5.296865003198977,
      "grad_norm": 1.0691317639616977,
      "learning_rate": 9.524619892178066e-06,
      "loss": 0.0898,
      "step": 8279
    },
    {
      "epoch": 5.297504798464491,
      "grad_norm": 0.9205922726822283,
      "learning_rate": 9.522550091131703e-06,
      "loss": 0.0808,
      "step": 8280
    },
    {
      "epoch": 5.298144593730006,
      "grad_norm": 0.8313311025231598,
      "learning_rate": 9.520480310586187e-06,
      "loss": 0.0836,
      "step": 8281
    },
    {
      "epoch": 5.2987843889955215,
      "grad_norm": 1.0021689434406997,
      "learning_rate": 9.518410550630396e-06,
      "loss": 0.0834,
      "step": 8282
    },
    {
      "epoch": 5.299424184261037,
      "grad_norm": 1.0006164620536777,
      "learning_rate": 9.516340811353197e-06,
      "loss": 0.084,
      "step": 8283
    },
    {
      "epoch": 5.300063979526551,
      "grad_norm": 0.9831129485722825,
      "learning_rate": 9.514271092843461e-06,
      "loss": 0.0901,
      "step": 8284
    },
    {
      "epoch": 5.3007037747920664,
      "grad_norm": 0.932554481333575,
      "learning_rate": 9.512201395190065e-06,
      "loss": 0.0829,
      "step": 8285
    },
    {
      "epoch": 5.301343570057582,
      "grad_norm": 1.0227940989214168,
      "learning_rate": 9.510131718481873e-06,
      "loss": 0.0812,
      "step": 8286
    },
    {
      "epoch": 5.301983365323097,
      "grad_norm": 1.1652566564734206,
      "learning_rate": 9.508062062807753e-06,
      "loss": 0.0997,
      "step": 8287
    },
    {
      "epoch": 5.302623160588611,
      "grad_norm": 0.9368920246007157,
      "learning_rate": 9.50599242825657e-06,
      "loss": 0.086,
      "step": 8288
    },
    {
      "epoch": 5.303262955854127,
      "grad_norm": 0.9381785154238513,
      "learning_rate": 9.503922814917197e-06,
      "loss": 0.0817,
      "step": 8289
    },
    {
      "epoch": 5.303902751119642,
      "grad_norm": 1.0508545847674242,
      "learning_rate": 9.501853222878495e-06,
      "loss": 0.0914,
      "step": 8290
    },
    {
      "epoch": 5.304542546385157,
      "grad_norm": 1.0652131035368038,
      "learning_rate": 9.499783652229325e-06,
      "loss": 0.0897,
      "step": 8291
    },
    {
      "epoch": 5.305182341650672,
      "grad_norm": 0.9892589278520484,
      "learning_rate": 9.49771410305856e-06,
      "loss": 0.0917,
      "step": 8292
    },
    {
      "epoch": 5.305822136916187,
      "grad_norm": 1.020445767789829,
      "learning_rate": 9.495644575455058e-06,
      "loss": 0.0944,
      "step": 8293
    },
    {
      "epoch": 5.306461932181702,
      "grad_norm": 1.1433076272070752,
      "learning_rate": 9.493575069507678e-06,
      "loss": 0.0899,
      "step": 8294
    },
    {
      "epoch": 5.3071017274472165,
      "grad_norm": 0.9206377439972036,
      "learning_rate": 9.491505585305287e-06,
      "loss": 0.0867,
      "step": 8295
    },
    {
      "epoch": 5.307741522712732,
      "grad_norm": 0.9546558098195603,
      "learning_rate": 9.48943612293674e-06,
      "loss": 0.0822,
      "step": 8296
    },
    {
      "epoch": 5.308381317978247,
      "grad_norm": 0.9645697664071715,
      "learning_rate": 9.487366682490899e-06,
      "loss": 0.0781,
      "step": 8297
    },
    {
      "epoch": 5.309021113243762,
      "grad_norm": 0.9980474946807131,
      "learning_rate": 9.485297264056615e-06,
      "loss": 0.0846,
      "step": 8298
    },
    {
      "epoch": 5.309660908509277,
      "grad_norm": 0.9849421196540377,
      "learning_rate": 9.483227867722757e-06,
      "loss": 0.0821,
      "step": 8299
    },
    {
      "epoch": 5.310300703774792,
      "grad_norm": 1.0102421827402306,
      "learning_rate": 9.481158493578174e-06,
      "loss": 0.0829,
      "step": 8300
    },
    {
      "epoch": 5.310940499040307,
      "grad_norm": 1.108128712041955,
      "learning_rate": 9.479089141711721e-06,
      "loss": 0.086,
      "step": 8301
    },
    {
      "epoch": 5.311580294305822,
      "grad_norm": 0.9708901722687001,
      "learning_rate": 9.477019812212254e-06,
      "loss": 0.076,
      "step": 8302
    },
    {
      "epoch": 5.312220089571337,
      "grad_norm": 1.157698093214824,
      "learning_rate": 9.474950505168628e-06,
      "loss": 0.0987,
      "step": 8303
    },
    {
      "epoch": 5.312859884836852,
      "grad_norm": 1.011188834234886,
      "learning_rate": 9.47288122066969e-06,
      "loss": 0.0841,
      "step": 8304
    },
    {
      "epoch": 5.3134996801023675,
      "grad_norm": 1.0526875053386264,
      "learning_rate": 9.470811958804295e-06,
      "loss": 0.0843,
      "step": 8305
    },
    {
      "epoch": 5.314139475367882,
      "grad_norm": 1.150061654943788,
      "learning_rate": 9.468742719661296e-06,
      "loss": 0.0926,
      "step": 8306
    },
    {
      "epoch": 5.314779270633397,
      "grad_norm": 1.035764552977847,
      "learning_rate": 9.466673503329539e-06,
      "loss": 0.0853,
      "step": 8307
    },
    {
      "epoch": 5.315419065898912,
      "grad_norm": 0.9714350211261986,
      "learning_rate": 9.464604309897868e-06,
      "loss": 0.08,
      "step": 8308
    },
    {
      "epoch": 5.316058861164428,
      "grad_norm": 1.0720245999285183,
      "learning_rate": 9.46253513945514e-06,
      "loss": 0.0902,
      "step": 8309
    },
    {
      "epoch": 5.316698656429942,
      "grad_norm": 0.9927568957638024,
      "learning_rate": 9.4604659920902e-06,
      "loss": 0.0888,
      "step": 8310
    },
    {
      "epoch": 5.317338451695457,
      "grad_norm": 1.198767911587105,
      "learning_rate": 9.458396867891884e-06,
      "loss": 0.1021,
      "step": 8311
    },
    {
      "epoch": 5.317978246960973,
      "grad_norm": 1.0134126439779763,
      "learning_rate": 9.456327766949047e-06,
      "loss": 0.0854,
      "step": 8312
    },
    {
      "epoch": 5.318618042226488,
      "grad_norm": 1.0057705407955777,
      "learning_rate": 9.454258689350529e-06,
      "loss": 0.0861,
      "step": 8313
    },
    {
      "epoch": 5.319257837492002,
      "grad_norm": 1.0936218492064076,
      "learning_rate": 9.452189635185168e-06,
      "loss": 0.0915,
      "step": 8314
    },
    {
      "epoch": 5.319897632757518,
      "grad_norm": 1.0243904860614228,
      "learning_rate": 9.450120604541813e-06,
      "loss": 0.0908,
      "step": 8315
    },
    {
      "epoch": 5.320537428023033,
      "grad_norm": 1.06624915771012,
      "learning_rate": 9.4480515975093e-06,
      "loss": 0.0937,
      "step": 8316
    },
    {
      "epoch": 5.321177223288547,
      "grad_norm": 1.106731793323065,
      "learning_rate": 9.445982614176471e-06,
      "loss": 0.091,
      "step": 8317
    },
    {
      "epoch": 5.3218170185540625,
      "grad_norm": 0.9948066493497324,
      "learning_rate": 9.443913654632161e-06,
      "loss": 0.0821,
      "step": 8318
    },
    {
      "epoch": 5.322456813819578,
      "grad_norm": 1.0442042612714133,
      "learning_rate": 9.441844718965213e-06,
      "loss": 0.1029,
      "step": 8319
    },
    {
      "epoch": 5.323096609085093,
      "grad_norm": 1.0025450198405863,
      "learning_rate": 9.439775807264458e-06,
      "loss": 0.0927,
      "step": 8320
    },
    {
      "epoch": 5.3237364043506075,
      "grad_norm": 1.1463373367051073,
      "learning_rate": 9.43770691961873e-06,
      "loss": 0.0981,
      "step": 8321
    },
    {
      "epoch": 5.324376199616123,
      "grad_norm": 0.955974986706086,
      "learning_rate": 9.43563805611687e-06,
      "loss": 0.0864,
      "step": 8322
    },
    {
      "epoch": 5.325015994881638,
      "grad_norm": 1.022912127256186,
      "learning_rate": 9.433569216847708e-06,
      "loss": 0.0874,
      "step": 8323
    },
    {
      "epoch": 5.325655790147153,
      "grad_norm": 0.9059070308796667,
      "learning_rate": 9.431500401900074e-06,
      "loss": 0.081,
      "step": 8324
    },
    {
      "epoch": 5.326295585412668,
      "grad_norm": 1.0375023863530537,
      "learning_rate": 9.429431611362804e-06,
      "loss": 0.0935,
      "step": 8325
    },
    {
      "epoch": 5.326935380678183,
      "grad_norm": 0.9983933849417131,
      "learning_rate": 9.427362845324726e-06,
      "loss": 0.0829,
      "step": 8326
    },
    {
      "epoch": 5.327575175943698,
      "grad_norm": 1.0042902534902571,
      "learning_rate": 9.425294103874669e-06,
      "loss": 0.0889,
      "step": 8327
    },
    {
      "epoch": 5.3282149712092135,
      "grad_norm": 0.9926712737390407,
      "learning_rate": 9.423225387101456e-06,
      "loss": 0.0883,
      "step": 8328
    },
    {
      "epoch": 5.328854766474728,
      "grad_norm": 1.1020502304332735,
      "learning_rate": 9.421156695093922e-06,
      "loss": 0.0923,
      "step": 8329
    },
    {
      "epoch": 5.329494561740243,
      "grad_norm": 0.9612963759429362,
      "learning_rate": 9.419088027940891e-06,
      "loss": 0.0739,
      "step": 8330
    },
    {
      "epoch": 5.330134357005758,
      "grad_norm": 1.0683901474920858,
      "learning_rate": 9.417019385731186e-06,
      "loss": 0.0888,
      "step": 8331
    },
    {
      "epoch": 5.330774152271273,
      "grad_norm": 1.0758419849428031,
      "learning_rate": 9.41495076855363e-06,
      "loss": 0.0867,
      "step": 8332
    },
    {
      "epoch": 5.331413947536788,
      "grad_norm": 1.0103458653461894,
      "learning_rate": 9.41288217649705e-06,
      "loss": 0.0859,
      "step": 8333
    },
    {
      "epoch": 5.332053742802303,
      "grad_norm": 1.0252534353219955,
      "learning_rate": 9.410813609650265e-06,
      "loss": 0.0915,
      "step": 8334
    },
    {
      "epoch": 5.332693538067819,
      "grad_norm": 1.0274349418095945,
      "learning_rate": 9.408745068102091e-06,
      "loss": 0.0884,
      "step": 8335
    },
    {
      "epoch": 5.333333333333333,
      "grad_norm": 1.141496237132421,
      "learning_rate": 9.406676551941355e-06,
      "loss": 0.0843,
      "step": 8336
    },
    {
      "epoch": 5.333973128598848,
      "grad_norm": 1.048659117640884,
      "learning_rate": 9.404608061256874e-06,
      "loss": 0.0943,
      "step": 8337
    },
    {
      "epoch": 5.334612923864364,
      "grad_norm": 1.0839609699086958,
      "learning_rate": 9.402539596137458e-06,
      "loss": 0.0906,
      "step": 8338
    },
    {
      "epoch": 5.335252719129878,
      "grad_norm": 1.0067144802331518,
      "learning_rate": 9.400471156671937e-06,
      "loss": 0.0876,
      "step": 8339
    },
    {
      "epoch": 5.335892514395393,
      "grad_norm": 0.9717929491263928,
      "learning_rate": 9.398402742949115e-06,
      "loss": 0.0806,
      "step": 8340
    },
    {
      "epoch": 5.3365323096609085,
      "grad_norm": 1.0449047291518492,
      "learning_rate": 9.396334355057805e-06,
      "loss": 0.0874,
      "step": 8341
    },
    {
      "epoch": 5.337172104926424,
      "grad_norm": 0.9347549486153958,
      "learning_rate": 9.394265993086827e-06,
      "loss": 0.087,
      "step": 8342
    },
    {
      "epoch": 5.337811900191938,
      "grad_norm": 0.9565018950084324,
      "learning_rate": 9.392197657124992e-06,
      "loss": 0.0846,
      "step": 8343
    },
    {
      "epoch": 5.3384516954574535,
      "grad_norm": 0.9705046436971955,
      "learning_rate": 9.390129347261107e-06,
      "loss": 0.088,
      "step": 8344
    },
    {
      "epoch": 5.339091490722969,
      "grad_norm": 1.0343540400757216,
      "learning_rate": 9.388061063583979e-06,
      "loss": 0.091,
      "step": 8345
    },
    {
      "epoch": 5.339731285988484,
      "grad_norm": 1.0216457826413476,
      "learning_rate": 9.385992806182425e-06,
      "loss": 0.0874,
      "step": 8346
    },
    {
      "epoch": 5.340371081253998,
      "grad_norm": 0.9445274256884826,
      "learning_rate": 9.383924575145247e-06,
      "loss": 0.0804,
      "step": 8347
    },
    {
      "epoch": 5.341010876519514,
      "grad_norm": 1.0865290538055017,
      "learning_rate": 9.38185637056125e-06,
      "loss": 0.0858,
      "step": 8348
    },
    {
      "epoch": 5.341650671785029,
      "grad_norm": 1.0305890689020438,
      "learning_rate": 9.379788192519242e-06,
      "loss": 0.0806,
      "step": 8349
    },
    {
      "epoch": 5.342290467050544,
      "grad_norm": 0.9862298744025103,
      "learning_rate": 9.377720041108027e-06,
      "loss": 0.09,
      "step": 8350
    },
    {
      "epoch": 5.342930262316059,
      "grad_norm": 1.0975885702301356,
      "learning_rate": 9.375651916416401e-06,
      "loss": 0.0842,
      "step": 8351
    },
    {
      "epoch": 5.343570057581574,
      "grad_norm": 0.9852966653928166,
      "learning_rate": 9.373583818533177e-06,
      "loss": 0.0858,
      "step": 8352
    },
    {
      "epoch": 5.344209852847089,
      "grad_norm": 1.041072646510002,
      "learning_rate": 9.371515747547148e-06,
      "loss": 0.0859,
      "step": 8353
    },
    {
      "epoch": 5.3448496481126035,
      "grad_norm": 0.9568485251665768,
      "learning_rate": 9.369447703547117e-06,
      "loss": 0.0886,
      "step": 8354
    },
    {
      "epoch": 5.345489443378119,
      "grad_norm": 0.9426834120095026,
      "learning_rate": 9.367379686621875e-06,
      "loss": 0.0839,
      "step": 8355
    },
    {
      "epoch": 5.346129238643634,
      "grad_norm": 0.9943958807046742,
      "learning_rate": 9.365311696860228e-06,
      "loss": 0.0856,
      "step": 8356
    },
    {
      "epoch": 5.346769033909149,
      "grad_norm": 1.071088817437423,
      "learning_rate": 9.363243734350968e-06,
      "loss": 0.0814,
      "step": 8357
    },
    {
      "epoch": 5.347408829174664,
      "grad_norm": 1.04253677885749,
      "learning_rate": 9.361175799182883e-06,
      "loss": 0.0889,
      "step": 8358
    },
    {
      "epoch": 5.348048624440179,
      "grad_norm": 1.0011031484272295,
      "learning_rate": 9.35910789144478e-06,
      "loss": 0.0843,
      "step": 8359
    },
    {
      "epoch": 5.348688419705694,
      "grad_norm": 0.9837330215092489,
      "learning_rate": 9.357040011225441e-06,
      "loss": 0.082,
      "step": 8360
    },
    {
      "epoch": 5.34932821497121,
      "grad_norm": 1.1062137591600891,
      "learning_rate": 9.35497215861366e-06,
      "loss": 0.0929,
      "step": 8361
    },
    {
      "epoch": 5.349968010236724,
      "grad_norm": 1.0943881137820295,
      "learning_rate": 9.352904333698229e-06,
      "loss": 0.0935,
      "step": 8362
    },
    {
      "epoch": 5.350607805502239,
      "grad_norm": 1.0287451480351026,
      "learning_rate": 9.350836536567934e-06,
      "loss": 0.093,
      "step": 8363
    },
    {
      "epoch": 5.3512476007677545,
      "grad_norm": 0.9193489728204699,
      "learning_rate": 9.348768767311563e-06,
      "loss": 0.081,
      "step": 8364
    },
    {
      "epoch": 5.35188739603327,
      "grad_norm": 1.0060947380887146,
      "learning_rate": 9.346701026017899e-06,
      "loss": 0.0855,
      "step": 8365
    },
    {
      "epoch": 5.352527191298784,
      "grad_norm": 1.0995069707402167,
      "learning_rate": 9.344633312775736e-06,
      "loss": 0.0923,
      "step": 8366
    },
    {
      "epoch": 5.3531669865642995,
      "grad_norm": 0.9357699495604903,
      "learning_rate": 9.342565627673852e-06,
      "loss": 0.0895,
      "step": 8367
    },
    {
      "epoch": 5.353806781829815,
      "grad_norm": 1.101141009648859,
      "learning_rate": 9.340497970801029e-06,
      "loss": 0.091,
      "step": 8368
    },
    {
      "epoch": 5.354446577095329,
      "grad_norm": 1.0633186871340627,
      "learning_rate": 9.338430342246052e-06,
      "loss": 0.0951,
      "step": 8369
    },
    {
      "epoch": 5.355086372360844,
      "grad_norm": 1.0615091314145337,
      "learning_rate": 9.336362742097698e-06,
      "loss": 0.0813,
      "step": 8370
    },
    {
      "epoch": 5.35572616762636,
      "grad_norm": 1.0334973876683755,
      "learning_rate": 9.334295170444743e-06,
      "loss": 0.0845,
      "step": 8371
    },
    {
      "epoch": 5.356365962891875,
      "grad_norm": 0.9026382510384491,
      "learning_rate": 9.332227627375973e-06,
      "loss": 0.0794,
      "step": 8372
    },
    {
      "epoch": 5.357005758157389,
      "grad_norm": 0.9143324146822454,
      "learning_rate": 9.330160112980163e-06,
      "loss": 0.0802,
      "step": 8373
    },
    {
      "epoch": 5.357645553422905,
      "grad_norm": 1.0657082846188701,
      "learning_rate": 9.328092627346085e-06,
      "loss": 0.094,
      "step": 8374
    },
    {
      "epoch": 5.35828534868842,
      "grad_norm": 1.0047710465630209,
      "learning_rate": 9.326025170562512e-06,
      "loss": 0.0857,
      "step": 8375
    },
    {
      "epoch": 5.358925143953935,
      "grad_norm": 1.0536708958877155,
      "learning_rate": 9.32395774271822e-06,
      "loss": 0.0861,
      "step": 8376
    },
    {
      "epoch": 5.3595649392194495,
      "grad_norm": 1.0528729750837114,
      "learning_rate": 9.321890343901982e-06,
      "loss": 0.082,
      "step": 8377
    },
    {
      "epoch": 5.360204734484965,
      "grad_norm": 1.0159835506956685,
      "learning_rate": 9.319822974202564e-06,
      "loss": 0.0835,
      "step": 8378
    },
    {
      "epoch": 5.36084452975048,
      "grad_norm": 1.30465346827098,
      "learning_rate": 9.31775563370874e-06,
      "loss": 0.1002,
      "step": 8379
    },
    {
      "epoch": 5.3614843250159945,
      "grad_norm": 0.9685795357672663,
      "learning_rate": 9.315688322509275e-06,
      "loss": 0.0816,
      "step": 8380
    },
    {
      "epoch": 5.36212412028151,
      "grad_norm": 1.0853482264981755,
      "learning_rate": 9.313621040692932e-06,
      "loss": 0.0947,
      "step": 8381
    },
    {
      "epoch": 5.362763915547025,
      "grad_norm": 1.0584401402640204,
      "learning_rate": 9.311553788348486e-06,
      "loss": 0.0884,
      "step": 8382
    },
    {
      "epoch": 5.36340371081254,
      "grad_norm": 0.9579219623440811,
      "learning_rate": 9.309486565564693e-06,
      "loss": 0.083,
      "step": 8383
    },
    {
      "epoch": 5.364043506078055,
      "grad_norm": 1.084678117536129,
      "learning_rate": 9.307419372430321e-06,
      "loss": 0.0851,
      "step": 8384
    },
    {
      "epoch": 5.36468330134357,
      "grad_norm": 0.9981676081980406,
      "learning_rate": 9.305352209034126e-06,
      "loss": 0.0861,
      "step": 8385
    },
    {
      "epoch": 5.365323096609085,
      "grad_norm": 1.076311030920995,
      "learning_rate": 9.303285075464874e-06,
      "loss": 0.0906,
      "step": 8386
    },
    {
      "epoch": 5.3659628918746005,
      "grad_norm": 1.080397598890348,
      "learning_rate": 9.301217971811322e-06,
      "loss": 0.0898,
      "step": 8387
    },
    {
      "epoch": 5.366602687140115,
      "grad_norm": 0.9717003099459665,
      "learning_rate": 9.299150898162221e-06,
      "loss": 0.0861,
      "step": 8388
    },
    {
      "epoch": 5.36724248240563,
      "grad_norm": 0.9406789505765145,
      "learning_rate": 9.29708385460634e-06,
      "loss": 0.0777,
      "step": 8389
    },
    {
      "epoch": 5.3678822776711455,
      "grad_norm": 1.0595423594496525,
      "learning_rate": 9.295016841232427e-06,
      "loss": 0.085,
      "step": 8390
    },
    {
      "epoch": 5.36852207293666,
      "grad_norm": 0.9387165566323066,
      "learning_rate": 9.292949858129234e-06,
      "loss": 0.0845,
      "step": 8391
    },
    {
      "epoch": 5.369161868202175,
      "grad_norm": 0.8604765254553433,
      "learning_rate": 9.29088290538552e-06,
      "loss": 0.0815,
      "step": 8392
    },
    {
      "epoch": 5.36980166346769,
      "grad_norm": 1.0478459241581917,
      "learning_rate": 9.288815983090031e-06,
      "loss": 0.0933,
      "step": 8393
    },
    {
      "epoch": 5.370441458733206,
      "grad_norm": 0.9650953115390688,
      "learning_rate": 9.28674909133152e-06,
      "loss": 0.0874,
      "step": 8394
    },
    {
      "epoch": 5.37108125399872,
      "grad_norm": 1.1600060895538908,
      "learning_rate": 9.284682230198729e-06,
      "loss": 0.0947,
      "step": 8395
    },
    {
      "epoch": 5.371721049264235,
      "grad_norm": 0.9713324974667795,
      "learning_rate": 9.282615399780414e-06,
      "loss": 0.0808,
      "step": 8396
    },
    {
      "epoch": 5.372360844529751,
      "grad_norm": 0.9935680931107789,
      "learning_rate": 9.28054860016532e-06,
      "loss": 0.0858,
      "step": 8397
    },
    {
      "epoch": 5.373000639795266,
      "grad_norm": 0.9409589920058143,
      "learning_rate": 9.278481831442187e-06,
      "loss": 0.075,
      "step": 8398
    },
    {
      "epoch": 5.37364043506078,
      "grad_norm": 0.9764805363008416,
      "learning_rate": 9.276415093699761e-06,
      "loss": 0.0802,
      "step": 8399
    },
    {
      "epoch": 5.3742802303262955,
      "grad_norm": 0.9435556028271979,
      "learning_rate": 9.274348387026785e-06,
      "loss": 0.0889,
      "step": 8400
    },
    {
      "epoch": 5.374920025591811,
      "grad_norm": 1.0679118426504788,
      "learning_rate": 9.272281711511993e-06,
      "loss": 0.0912,
      "step": 8401
    },
    {
      "epoch": 5.375559820857326,
      "grad_norm": 1.0982188835595206,
      "learning_rate": 9.270215067244136e-06,
      "loss": 0.0943,
      "step": 8402
    },
    {
      "epoch": 5.3761996161228405,
      "grad_norm": 1.0774821546211792,
      "learning_rate": 9.268148454311944e-06,
      "loss": 0.0871,
      "step": 8403
    },
    {
      "epoch": 5.376839411388356,
      "grad_norm": 1.048395799989279,
      "learning_rate": 9.266081872804156e-06,
      "loss": 0.0885,
      "step": 8404
    },
    {
      "epoch": 5.377479206653871,
      "grad_norm": 1.0047247764664102,
      "learning_rate": 9.264015322809502e-06,
      "loss": 0.0931,
      "step": 8405
    },
    {
      "epoch": 5.378119001919385,
      "grad_norm": 1.0312981851738465,
      "learning_rate": 9.261948804416726e-06,
      "loss": 0.0841,
      "step": 8406
    },
    {
      "epoch": 5.378758797184901,
      "grad_norm": 1.0960931585989848,
      "learning_rate": 9.259882317714555e-06,
      "loss": 0.0879,
      "step": 8407
    },
    {
      "epoch": 5.379398592450416,
      "grad_norm": 1.0570238359150572,
      "learning_rate": 9.25781586279172e-06,
      "loss": 0.0864,
      "step": 8408
    },
    {
      "epoch": 5.380038387715931,
      "grad_norm": 1.0196894976726973,
      "learning_rate": 9.255749439736952e-06,
      "loss": 0.0879,
      "step": 8409
    },
    {
      "epoch": 5.380678182981446,
      "grad_norm": 0.9551433819547118,
      "learning_rate": 9.253683048638979e-06,
      "loss": 0.0819,
      "step": 8410
    },
    {
      "epoch": 5.381317978246961,
      "grad_norm": 1.1340272712949802,
      "learning_rate": 9.25161668958653e-06,
      "loss": 0.0939,
      "step": 8411
    },
    {
      "epoch": 5.381957773512476,
      "grad_norm": 0.9847634574694348,
      "learning_rate": 9.249550362668322e-06,
      "loss": 0.0897,
      "step": 8412
    },
    {
      "epoch": 5.3825975687779914,
      "grad_norm": 1.019830316427981,
      "learning_rate": 9.247484067973093e-06,
      "loss": 0.0919,
      "step": 8413
    },
    {
      "epoch": 5.383237364043506,
      "grad_norm": 1.1159531565514607,
      "learning_rate": 9.24541780558956e-06,
      "loss": 0.0916,
      "step": 8414
    },
    {
      "epoch": 5.383877159309021,
      "grad_norm": 1.1036216292554066,
      "learning_rate": 9.243351575606441e-06,
      "loss": 0.089,
      "step": 8415
    },
    {
      "epoch": 5.384516954574536,
      "grad_norm": 0.9848293112971769,
      "learning_rate": 9.241285378112461e-06,
      "loss": 0.0774,
      "step": 8416
    },
    {
      "epoch": 5.385156749840051,
      "grad_norm": 0.8960768052062666,
      "learning_rate": 9.239219213196338e-06,
      "loss": 0.0766,
      "step": 8417
    },
    {
      "epoch": 5.385796545105566,
      "grad_norm": 1.1012299413682636,
      "learning_rate": 9.237153080946783e-06,
      "loss": 0.0825,
      "step": 8418
    },
    {
      "epoch": 5.386436340371081,
      "grad_norm": 1.0693345537189338,
      "learning_rate": 9.235086981452523e-06,
      "loss": 0.0893,
      "step": 8419
    },
    {
      "epoch": 5.387076135636597,
      "grad_norm": 1.0713025503018208,
      "learning_rate": 9.233020914802269e-06,
      "loss": 0.0901,
      "step": 8420
    },
    {
      "epoch": 5.387715930902111,
      "grad_norm": 1.0098727885396246,
      "learning_rate": 9.23095488108473e-06,
      "loss": 0.087,
      "step": 8421
    },
    {
      "epoch": 5.388355726167626,
      "grad_norm": 1.0615601868067668,
      "learning_rate": 9.22888888038862e-06,
      "loss": 0.0889,
      "step": 8422
    },
    {
      "epoch": 5.3889955214331415,
      "grad_norm": 1.0228052114701853,
      "learning_rate": 9.226822912802651e-06,
      "loss": 0.0847,
      "step": 8423
    },
    {
      "epoch": 5.389635316698657,
      "grad_norm": 1.161451577098814,
      "learning_rate": 9.22475697841553e-06,
      "loss": 0.0953,
      "step": 8424
    },
    {
      "epoch": 5.390275111964171,
      "grad_norm": 0.9515449236273518,
      "learning_rate": 9.222691077315962e-06,
      "loss": 0.0886,
      "step": 8425
    },
    {
      "epoch": 5.3909149072296865,
      "grad_norm": 1.0052338778140983,
      "learning_rate": 9.220625209592661e-06,
      "loss": 0.0895,
      "step": 8426
    },
    {
      "epoch": 5.391554702495202,
      "grad_norm": 0.967553846157717,
      "learning_rate": 9.218559375334327e-06,
      "loss": 0.087,
      "step": 8427
    },
    {
      "epoch": 5.392194497760716,
      "grad_norm": 0.9270142227601086,
      "learning_rate": 9.21649357462966e-06,
      "loss": 0.0731,
      "step": 8428
    },
    {
      "epoch": 5.392834293026231,
      "grad_norm": 0.9620777047203298,
      "learning_rate": 9.214427807567368e-06,
      "loss": 0.0812,
      "step": 8429
    },
    {
      "epoch": 5.393474088291747,
      "grad_norm": 1.004773928016514,
      "learning_rate": 9.212362074236149e-06,
      "loss": 0.0845,
      "step": 8430
    },
    {
      "epoch": 5.394113883557262,
      "grad_norm": 0.9939213111247358,
      "learning_rate": 9.210296374724702e-06,
      "loss": 0.0895,
      "step": 8431
    },
    {
      "epoch": 5.394753678822776,
      "grad_norm": 0.981043516603621,
      "learning_rate": 9.208230709121717e-06,
      "loss": 0.087,
      "step": 8432
    },
    {
      "epoch": 5.395393474088292,
      "grad_norm": 1.094114955104127,
      "learning_rate": 9.206165077515904e-06,
      "loss": 0.0894,
      "step": 8433
    },
    {
      "epoch": 5.396033269353807,
      "grad_norm": 1.002547641599214,
      "learning_rate": 9.20409947999595e-06,
      "loss": 0.093,
      "step": 8434
    },
    {
      "epoch": 5.396673064619322,
      "grad_norm": 0.9547750490683918,
      "learning_rate": 9.202033916650544e-06,
      "loss": 0.0791,
      "step": 8435
    },
    {
      "epoch": 5.3973128598848366,
      "grad_norm": 0.8669200529713703,
      "learning_rate": 9.199968387568386e-06,
      "loss": 0.0756,
      "step": 8436
    },
    {
      "epoch": 5.397952655150352,
      "grad_norm": 1.065526086025788,
      "learning_rate": 9.197902892838162e-06,
      "loss": 0.086,
      "step": 8437
    },
    {
      "epoch": 5.398592450415867,
      "grad_norm": 1.0534065731627154,
      "learning_rate": 9.19583743254856e-06,
      "loss": 0.0923,
      "step": 8438
    },
    {
      "epoch": 5.399232245681382,
      "grad_norm": 0.9873340573071859,
      "learning_rate": 9.193772006788268e-06,
      "loss": 0.0837,
      "step": 8439
    },
    {
      "epoch": 5.399872040946897,
      "grad_norm": 0.9698046264282733,
      "learning_rate": 9.191706615645974e-06,
      "loss": 0.0844,
      "step": 8440
    },
    {
      "epoch": 5.400511836212412,
      "grad_norm": 1.1326223314499524,
      "learning_rate": 9.18964125921036e-06,
      "loss": 0.0903,
      "step": 8441
    },
    {
      "epoch": 5.401151631477927,
      "grad_norm": 0.9292080817774013,
      "learning_rate": 9.187575937570103e-06,
      "loss": 0.077,
      "step": 8442
    },
    {
      "epoch": 5.401791426743442,
      "grad_norm": 1.041263684098862,
      "learning_rate": 9.185510650813895e-06,
      "loss": 0.086,
      "step": 8443
    },
    {
      "epoch": 5.402431222008957,
      "grad_norm": 0.9113162195052694,
      "learning_rate": 9.183445399030412e-06,
      "loss": 0.0843,
      "step": 8444
    },
    {
      "epoch": 5.403071017274472,
      "grad_norm": 1.1673398380393007,
      "learning_rate": 9.181380182308328e-06,
      "loss": 0.0883,
      "step": 8445
    },
    {
      "epoch": 5.4037108125399875,
      "grad_norm": 1.0444382395907281,
      "learning_rate": 9.179315000736325e-06,
      "loss": 0.0846,
      "step": 8446
    },
    {
      "epoch": 5.404350607805502,
      "grad_norm": 1.1485066097829524,
      "learning_rate": 9.177249854403076e-06,
      "loss": 0.0905,
      "step": 8447
    },
    {
      "epoch": 5.404990403071017,
      "grad_norm": 1.0425809616968154,
      "learning_rate": 9.17518474339725e-06,
      "loss": 0.097,
      "step": 8448
    },
    {
      "epoch": 5.4056301983365325,
      "grad_norm": 1.2836164013800044,
      "learning_rate": 9.173119667807529e-06,
      "loss": 0.0826,
      "step": 8449
    },
    {
      "epoch": 5.406269993602048,
      "grad_norm": 1.074230894322258,
      "learning_rate": 9.171054627722578e-06,
      "loss": 0.083,
      "step": 8450
    },
    {
      "epoch": 5.406909788867562,
      "grad_norm": 1.0069792511821927,
      "learning_rate": 9.168989623231065e-06,
      "loss": 0.0869,
      "step": 8451
    },
    {
      "epoch": 5.407549584133077,
      "grad_norm": 0.9748789099643397,
      "learning_rate": 9.16692465442166e-06,
      "loss": 0.0773,
      "step": 8452
    },
    {
      "epoch": 5.408189379398593,
      "grad_norm": 1.0019017276506976,
      "learning_rate": 9.16485972138303e-06,
      "loss": 0.0851,
      "step": 8453
    },
    {
      "epoch": 5.408829174664108,
      "grad_norm": 0.9641789177876199,
      "learning_rate": 9.162794824203837e-06,
      "loss": 0.0816,
      "step": 8454
    },
    {
      "epoch": 5.409468969929622,
      "grad_norm": 0.9056154611953021,
      "learning_rate": 9.160729962972741e-06,
      "loss": 0.0832,
      "step": 8455
    },
    {
      "epoch": 5.410108765195138,
      "grad_norm": 0.96473298057169,
      "learning_rate": 9.158665137778412e-06,
      "loss": 0.0859,
      "step": 8456
    },
    {
      "epoch": 5.410748560460653,
      "grad_norm": 1.0588000817099457,
      "learning_rate": 9.156600348709505e-06,
      "loss": 0.0964,
      "step": 8457
    },
    {
      "epoch": 5.411388355726167,
      "grad_norm": 0.8951408282369365,
      "learning_rate": 9.154535595854677e-06,
      "loss": 0.0783,
      "step": 8458
    },
    {
      "epoch": 5.4120281509916826,
      "grad_norm": 1.1728924089027493,
      "learning_rate": 9.152470879302589e-06,
      "loss": 0.0874,
      "step": 8459
    },
    {
      "epoch": 5.412667946257198,
      "grad_norm": 1.076483345156893,
      "learning_rate": 9.150406199141892e-06,
      "loss": 0.0861,
      "step": 8460
    },
    {
      "epoch": 5.413307741522713,
      "grad_norm": 0.9686490768323458,
      "learning_rate": 9.148341555461243e-06,
      "loss": 0.0898,
      "step": 8461
    },
    {
      "epoch": 5.4139475367882275,
      "grad_norm": 1.0475417292216513,
      "learning_rate": 9.146276948349289e-06,
      "loss": 0.0823,
      "step": 8462
    },
    {
      "epoch": 5.414587332053743,
      "grad_norm": 1.0350738046372394,
      "learning_rate": 9.144212377894689e-06,
      "loss": 0.0905,
      "step": 8463
    },
    {
      "epoch": 5.415227127319258,
      "grad_norm": 1.0628365062233665,
      "learning_rate": 9.142147844186085e-06,
      "loss": 0.087,
      "step": 8464
    },
    {
      "epoch": 5.415866922584772,
      "grad_norm": 1.1583901989155267,
      "learning_rate": 9.140083347312122e-06,
      "loss": 0.0951,
      "step": 8465
    },
    {
      "epoch": 5.416506717850288,
      "grad_norm": 0.9840339528376754,
      "learning_rate": 9.138018887361457e-06,
      "loss": 0.0907,
      "step": 8466
    },
    {
      "epoch": 5.417146513115803,
      "grad_norm": 0.976938617121144,
      "learning_rate": 9.135954464422725e-06,
      "loss": 0.0824,
      "step": 8467
    },
    {
      "epoch": 5.417786308381318,
      "grad_norm": 0.9995322000392382,
      "learning_rate": 9.133890078584571e-06,
      "loss": 0.0818,
      "step": 8468
    },
    {
      "epoch": 5.418426103646833,
      "grad_norm": 1.019841845357244,
      "learning_rate": 9.131825729935638e-06,
      "loss": 0.0839,
      "step": 8469
    },
    {
      "epoch": 5.419065898912348,
      "grad_norm": 1.1028493877064327,
      "learning_rate": 9.129761418564563e-06,
      "loss": 0.0893,
      "step": 8470
    },
    {
      "epoch": 5.419705694177863,
      "grad_norm": 0.994567334565205,
      "learning_rate": 9.127697144559987e-06,
      "loss": 0.089,
      "step": 8471
    },
    {
      "epoch": 5.4203454894433785,
      "grad_norm": 0.9076330310066946,
      "learning_rate": 9.125632908010536e-06,
      "loss": 0.086,
      "step": 8472
    },
    {
      "epoch": 5.420985284708893,
      "grad_norm": 0.984465479872619,
      "learning_rate": 9.12356870900486e-06,
      "loss": 0.0956,
      "step": 8473
    },
    {
      "epoch": 5.421625079974408,
      "grad_norm": 0.9811295116285806,
      "learning_rate": 9.121504547631583e-06,
      "loss": 0.0901,
      "step": 8474
    },
    {
      "epoch": 5.422264875239923,
      "grad_norm": 1.0448666456188616,
      "learning_rate": 9.119440423979337e-06,
      "loss": 0.0891,
      "step": 8475
    },
    {
      "epoch": 5.422904670505439,
      "grad_norm": 1.0857870870944952,
      "learning_rate": 9.117376338136755e-06,
      "loss": 0.0875,
      "step": 8476
    },
    {
      "epoch": 5.423544465770953,
      "grad_norm": 0.8783309302195319,
      "learning_rate": 9.115312290192461e-06,
      "loss": 0.0767,
      "step": 8477
    },
    {
      "epoch": 5.424184261036468,
      "grad_norm": 1.0800914877340773,
      "learning_rate": 9.113248280235081e-06,
      "loss": 0.0923,
      "step": 8478
    },
    {
      "epoch": 5.424824056301984,
      "grad_norm": 0.997404708804864,
      "learning_rate": 9.111184308353245e-06,
      "loss": 0.0874,
      "step": 8479
    },
    {
      "epoch": 5.425463851567498,
      "grad_norm": 1.1428969159991609,
      "learning_rate": 9.109120374635576e-06,
      "loss": 0.0895,
      "step": 8480
    },
    {
      "epoch": 5.426103646833013,
      "grad_norm": 1.061263912139614,
      "learning_rate": 9.107056479170691e-06,
      "loss": 0.0877,
      "step": 8481
    },
    {
      "epoch": 5.4267434420985285,
      "grad_norm": 1.0713989910011186,
      "learning_rate": 9.104992622047213e-06,
      "loss": 0.0849,
      "step": 8482
    },
    {
      "epoch": 5.427383237364044,
      "grad_norm": 1.0758986658331686,
      "learning_rate": 9.10292880335376e-06,
      "loss": 0.0827,
      "step": 8483
    },
    {
      "epoch": 5.428023032629558,
      "grad_norm": 0.947509051973523,
      "learning_rate": 9.100865023178949e-06,
      "loss": 0.0787,
      "step": 8484
    },
    {
      "epoch": 5.4286628278950735,
      "grad_norm": 1.0553795386206284,
      "learning_rate": 9.098801281611388e-06,
      "loss": 0.0825,
      "step": 8485
    },
    {
      "epoch": 5.429302623160589,
      "grad_norm": 0.9657889368905342,
      "learning_rate": 9.096737578739705e-06,
      "loss": 0.0826,
      "step": 8486
    },
    {
      "epoch": 5.429942418426104,
      "grad_norm": 1.1158011153976561,
      "learning_rate": 9.094673914652501e-06,
      "loss": 0.0954,
      "step": 8487
    },
    {
      "epoch": 5.430582213691618,
      "grad_norm": 1.1564113797550444,
      "learning_rate": 9.092610289438389e-06,
      "loss": 0.084,
      "step": 8488
    },
    {
      "epoch": 5.431222008957134,
      "grad_norm": 0.9687687908878975,
      "learning_rate": 9.090546703185976e-06,
      "loss": 0.091,
      "step": 8489
    },
    {
      "epoch": 5.431861804222649,
      "grad_norm": 1.0997445812393662,
      "learning_rate": 9.088483155983871e-06,
      "loss": 0.0926,
      "step": 8490
    },
    {
      "epoch": 5.432501599488164,
      "grad_norm": 1.0149131800555873,
      "learning_rate": 9.086419647920677e-06,
      "loss": 0.0865,
      "step": 8491
    },
    {
      "epoch": 5.433141394753679,
      "grad_norm": 0.9701054388415818,
      "learning_rate": 9.084356179084995e-06,
      "loss": 0.0813,
      "step": 8492
    },
    {
      "epoch": 5.433781190019194,
      "grad_norm": 1.0628224657158136,
      "learning_rate": 9.082292749565436e-06,
      "loss": 0.0874,
      "step": 8493
    },
    {
      "epoch": 5.434420985284709,
      "grad_norm": 0.9440780189077286,
      "learning_rate": 9.08022935945059e-06,
      "loss": 0.0799,
      "step": 8494
    },
    {
      "epoch": 5.435060780550224,
      "grad_norm": 0.9179951009365381,
      "learning_rate": 9.078166008829057e-06,
      "loss": 0.0893,
      "step": 8495
    },
    {
      "epoch": 5.435700575815739,
      "grad_norm": 1.056635840019559,
      "learning_rate": 9.07610269778944e-06,
      "loss": 0.0866,
      "step": 8496
    },
    {
      "epoch": 5.436340371081254,
      "grad_norm": 1.0660853526826595,
      "learning_rate": 9.07403942642033e-06,
      "loss": 0.08,
      "step": 8497
    },
    {
      "epoch": 5.436980166346769,
      "grad_norm": 0.9864210095493025,
      "learning_rate": 9.071976194810319e-06,
      "loss": 0.0889,
      "step": 8498
    },
    {
      "epoch": 5.437619961612284,
      "grad_norm": 0.9238733217925741,
      "learning_rate": 9.069913003047997e-06,
      "loss": 0.0808,
      "step": 8499
    },
    {
      "epoch": 5.438259756877799,
      "grad_norm": 1.0655220892770372,
      "learning_rate": 9.06784985122196e-06,
      "loss": 0.0921,
      "step": 8500
    },
    {
      "epoch": 5.438899552143314,
      "grad_norm": 0.8912826161653667,
      "learning_rate": 9.065786739420792e-06,
      "loss": 0.0815,
      "step": 8501
    },
    {
      "epoch": 5.439539347408829,
      "grad_norm": 0.9780561321041137,
      "learning_rate": 9.063723667733073e-06,
      "loss": 0.0856,
      "step": 8502
    },
    {
      "epoch": 5.440179142674344,
      "grad_norm": 1.0706764353908342,
      "learning_rate": 9.0616606362474e-06,
      "loss": 0.0991,
      "step": 8503
    },
    {
      "epoch": 5.440818937939859,
      "grad_norm": 0.9890391495049051,
      "learning_rate": 9.059597645052351e-06,
      "loss": 0.0844,
      "step": 8504
    },
    {
      "epoch": 5.4414587332053745,
      "grad_norm": 1.0180824694488047,
      "learning_rate": 9.057534694236503e-06,
      "loss": 0.0913,
      "step": 8505
    },
    {
      "epoch": 5.442098528470889,
      "grad_norm": 1.0567784107293763,
      "learning_rate": 9.05547178388844e-06,
      "loss": 0.0915,
      "step": 8506
    },
    {
      "epoch": 5.442738323736404,
      "grad_norm": 1.0948022237745525,
      "learning_rate": 9.053408914096739e-06,
      "loss": 0.0909,
      "step": 8507
    },
    {
      "epoch": 5.4433781190019195,
      "grad_norm": 1.020632527555052,
      "learning_rate": 9.051346084949976e-06,
      "loss": 0.0858,
      "step": 8508
    },
    {
      "epoch": 5.444017914267435,
      "grad_norm": 1.0929918865571033,
      "learning_rate": 9.049283296536721e-06,
      "loss": 0.0881,
      "step": 8509
    },
    {
      "epoch": 5.444657709532949,
      "grad_norm": 0.9637615198500908,
      "learning_rate": 9.047220548945553e-06,
      "loss": 0.088,
      "step": 8510
    },
    {
      "epoch": 5.445297504798464,
      "grad_norm": 1.0204690388671935,
      "learning_rate": 9.04515784226504e-06,
      "loss": 0.0897,
      "step": 8511
    },
    {
      "epoch": 5.44593730006398,
      "grad_norm": 1.0721906372916232,
      "learning_rate": 9.04309517658375e-06,
      "loss": 0.0889,
      "step": 8512
    },
    {
      "epoch": 5.446577095329495,
      "grad_norm": 0.9422618786507909,
      "learning_rate": 9.041032551990251e-06,
      "loss": 0.0802,
      "step": 8513
    },
    {
      "epoch": 5.447216890595009,
      "grad_norm": 0.9696615280320273,
      "learning_rate": 9.03896996857311e-06,
      "loss": 0.0888,
      "step": 8514
    },
    {
      "epoch": 5.447856685860525,
      "grad_norm": 0.9873589606615228,
      "learning_rate": 9.036907426420882e-06,
      "loss": 0.0924,
      "step": 8515
    },
    {
      "epoch": 5.44849648112604,
      "grad_norm": 1.116857407708841,
      "learning_rate": 9.034844925622142e-06,
      "loss": 0.0923,
      "step": 8516
    },
    {
      "epoch": 5.449136276391554,
      "grad_norm": 1.0057295195450373,
      "learning_rate": 9.032782466265446e-06,
      "loss": 0.0881,
      "step": 8517
    },
    {
      "epoch": 5.44977607165707,
      "grad_norm": 0.9866359265980126,
      "learning_rate": 9.030720048439347e-06,
      "loss": 0.0865,
      "step": 8518
    },
    {
      "epoch": 5.450415866922585,
      "grad_norm": 0.9812536419068812,
      "learning_rate": 9.028657672232404e-06,
      "loss": 0.0854,
      "step": 8519
    },
    {
      "epoch": 5.4510556621881,
      "grad_norm": 1.1134812962838045,
      "learning_rate": 9.026595337733175e-06,
      "loss": 0.0938,
      "step": 8520
    },
    {
      "epoch": 5.4516954574536145,
      "grad_norm": 1.1462523247816412,
      "learning_rate": 9.024533045030211e-06,
      "loss": 0.0923,
      "step": 8521
    },
    {
      "epoch": 5.45233525271913,
      "grad_norm": 1.1055439973692711,
      "learning_rate": 9.022470794212057e-06,
      "loss": 0.0956,
      "step": 8522
    },
    {
      "epoch": 5.452975047984645,
      "grad_norm": 1.1431302912809298,
      "learning_rate": 9.020408585367275e-06,
      "loss": 0.087,
      "step": 8523
    },
    {
      "epoch": 5.45361484325016,
      "grad_norm": 0.987852052000472,
      "learning_rate": 9.018346418584405e-06,
      "loss": 0.0921,
      "step": 8524
    },
    {
      "epoch": 5.454254638515675,
      "grad_norm": 1.025854980017721,
      "learning_rate": 9.016284293951988e-06,
      "loss": 0.0889,
      "step": 8525
    },
    {
      "epoch": 5.45489443378119,
      "grad_norm": 0.9060262754454327,
      "learning_rate": 9.014222211558579e-06,
      "loss": 0.0825,
      "step": 8526
    },
    {
      "epoch": 5.455534229046705,
      "grad_norm": 0.9078496205673359,
      "learning_rate": 9.012160171492713e-06,
      "loss": 0.0853,
      "step": 8527
    },
    {
      "epoch": 5.4561740243122205,
      "grad_norm": 1.0176935434057772,
      "learning_rate": 9.010098173842935e-06,
      "loss": 0.0827,
      "step": 8528
    },
    {
      "epoch": 5.456813819577735,
      "grad_norm": 1.1394988901284213,
      "learning_rate": 9.008036218697775e-06,
      "loss": 0.0968,
      "step": 8529
    },
    {
      "epoch": 5.45745361484325,
      "grad_norm": 0.8875819333510695,
      "learning_rate": 9.00597430614578e-06,
      "loss": 0.0836,
      "step": 8530
    },
    {
      "epoch": 5.4580934101087655,
      "grad_norm": 1.0332264428889484,
      "learning_rate": 9.00391243627548e-06,
      "loss": 0.0835,
      "step": 8531
    },
    {
      "epoch": 5.45873320537428,
      "grad_norm": 1.0197087312954745,
      "learning_rate": 9.001850609175402e-06,
      "loss": 0.087,
      "step": 8532
    },
    {
      "epoch": 5.459373000639795,
      "grad_norm": 1.0938001526058367,
      "learning_rate": 8.999788824934088e-06,
      "loss": 0.0905,
      "step": 8533
    },
    {
      "epoch": 5.46001279590531,
      "grad_norm": 0.9645002539936639,
      "learning_rate": 8.997727083640063e-06,
      "loss": 0.0801,
      "step": 8534
    },
    {
      "epoch": 5.460652591170826,
      "grad_norm": 0.8777473683201445,
      "learning_rate": 8.995665385381853e-06,
      "loss": 0.0876,
      "step": 8535
    },
    {
      "epoch": 5.46129238643634,
      "grad_norm": 0.9451736444448411,
      "learning_rate": 8.993603730247985e-06,
      "loss": 0.0943,
      "step": 8536
    },
    {
      "epoch": 5.461932181701855,
      "grad_norm": 0.9117051934694707,
      "learning_rate": 8.991542118326984e-06,
      "loss": 0.0862,
      "step": 8537
    },
    {
      "epoch": 5.462571976967371,
      "grad_norm": 1.0345327187148925,
      "learning_rate": 8.98948054970737e-06,
      "loss": 0.0856,
      "step": 8538
    },
    {
      "epoch": 5.463211772232885,
      "grad_norm": 1.0223804399217145,
      "learning_rate": 8.98741902447766e-06,
      "loss": 0.0893,
      "step": 8539
    },
    {
      "epoch": 5.4638515674984,
      "grad_norm": 1.0261366681480324,
      "learning_rate": 8.985357542726379e-06,
      "loss": 0.0894,
      "step": 8540
    },
    {
      "epoch": 5.464491362763916,
      "grad_norm": 1.0159644340392942,
      "learning_rate": 8.98329610454204e-06,
      "loss": 0.0885,
      "step": 8541
    },
    {
      "epoch": 5.465131158029431,
      "grad_norm": 0.9797369274670394,
      "learning_rate": 8.981234710013157e-06,
      "loss": 0.0796,
      "step": 8542
    },
    {
      "epoch": 5.465770953294945,
      "grad_norm": 1.0298336852782397,
      "learning_rate": 8.979173359228243e-06,
      "loss": 0.0903,
      "step": 8543
    },
    {
      "epoch": 5.4664107485604605,
      "grad_norm": 0.9095854922317405,
      "learning_rate": 8.977112052275811e-06,
      "loss": 0.0876,
      "step": 8544
    },
    {
      "epoch": 5.467050543825976,
      "grad_norm": 1.1062248748026318,
      "learning_rate": 8.975050789244365e-06,
      "loss": 0.0833,
      "step": 8545
    },
    {
      "epoch": 5.467690339091491,
      "grad_norm": 0.9894478375301216,
      "learning_rate": 8.972989570222416e-06,
      "loss": 0.0838,
      "step": 8546
    },
    {
      "epoch": 5.468330134357005,
      "grad_norm": 1.0817733485511656,
      "learning_rate": 8.970928395298471e-06,
      "loss": 0.0973,
      "step": 8547
    },
    {
      "epoch": 5.468969929622521,
      "grad_norm": 1.1096851298009933,
      "learning_rate": 8.968867264561027e-06,
      "loss": 0.0896,
      "step": 8548
    },
    {
      "epoch": 5.469609724888036,
      "grad_norm": 0.9614954711564653,
      "learning_rate": 8.966806178098587e-06,
      "loss": 0.0833,
      "step": 8549
    },
    {
      "epoch": 5.470249520153551,
      "grad_norm": 1.0327264182898956,
      "learning_rate": 8.964745135999656e-06,
      "loss": 0.0823,
      "step": 8550
    },
    {
      "epoch": 5.470889315419066,
      "grad_norm": 1.032947379737389,
      "learning_rate": 8.962684138352726e-06,
      "loss": 0.0869,
      "step": 8551
    },
    {
      "epoch": 5.471529110684581,
      "grad_norm": 1.0084412656348594,
      "learning_rate": 8.96062318524629e-06,
      "loss": 0.0886,
      "step": 8552
    },
    {
      "epoch": 5.472168905950096,
      "grad_norm": 1.0732872663138393,
      "learning_rate": 8.95856227676885e-06,
      "loss": 0.0946,
      "step": 8553
    },
    {
      "epoch": 5.472808701215611,
      "grad_norm": 1.0059187068010718,
      "learning_rate": 8.956501413008892e-06,
      "loss": 0.0847,
      "step": 8554
    },
    {
      "epoch": 5.473448496481126,
      "grad_norm": 0.9757356618263624,
      "learning_rate": 8.954440594054904e-06,
      "loss": 0.076,
      "step": 8555
    },
    {
      "epoch": 5.474088291746641,
      "grad_norm": 1.0383174921767584,
      "learning_rate": 8.952379819995381e-06,
      "loss": 0.0852,
      "step": 8556
    },
    {
      "epoch": 5.474728087012156,
      "grad_norm": 0.9936232350034836,
      "learning_rate": 8.950319090918804e-06,
      "loss": 0.0775,
      "step": 8557
    },
    {
      "epoch": 5.475367882277671,
      "grad_norm": 0.9583042409010611,
      "learning_rate": 8.94825840691366e-06,
      "loss": 0.0884,
      "step": 8558
    },
    {
      "epoch": 5.476007677543186,
      "grad_norm": 1.1299042965472346,
      "learning_rate": 8.946197768068424e-06,
      "loss": 0.0891,
      "step": 8559
    },
    {
      "epoch": 5.476647472808701,
      "grad_norm": 1.0280713839295417,
      "learning_rate": 8.944137174471585e-06,
      "loss": 0.0876,
      "step": 8560
    },
    {
      "epoch": 5.477287268074217,
      "grad_norm": 1.0824188666460464,
      "learning_rate": 8.942076626211617e-06,
      "loss": 0.0898,
      "step": 8561
    },
    {
      "epoch": 5.477927063339731,
      "grad_norm": 1.077092398893606,
      "learning_rate": 8.940016123376994e-06,
      "loss": 0.0864,
      "step": 8562
    },
    {
      "epoch": 5.478566858605246,
      "grad_norm": 0.9978893157534,
      "learning_rate": 8.937955666056198e-06,
      "loss": 0.0862,
      "step": 8563
    },
    {
      "epoch": 5.479206653870762,
      "grad_norm": 1.1071948617353402,
      "learning_rate": 8.935895254337694e-06,
      "loss": 0.092,
      "step": 8564
    },
    {
      "epoch": 5.479846449136277,
      "grad_norm": 1.058416148048843,
      "learning_rate": 8.933834888309956e-06,
      "loss": 0.0884,
      "step": 8565
    },
    {
      "epoch": 5.480486244401791,
      "grad_norm": 0.9797029747409508,
      "learning_rate": 8.931774568061449e-06,
      "loss": 0.0911,
      "step": 8566
    },
    {
      "epoch": 5.4811260396673065,
      "grad_norm": 0.9940131788920797,
      "learning_rate": 8.929714293680643e-06,
      "loss": 0.0871,
      "step": 8567
    },
    {
      "epoch": 5.481765834932822,
      "grad_norm": 0.9385077337800702,
      "learning_rate": 8.927654065256003e-06,
      "loss": 0.0813,
      "step": 8568
    },
    {
      "epoch": 5.482405630198336,
      "grad_norm": 1.0558618519276595,
      "learning_rate": 8.925593882875984e-06,
      "loss": 0.0894,
      "step": 8569
    },
    {
      "epoch": 5.483045425463851,
      "grad_norm": 1.0087241108266762,
      "learning_rate": 8.923533746629059e-06,
      "loss": 0.0889,
      "step": 8570
    },
    {
      "epoch": 5.483685220729367,
      "grad_norm": 1.0775374458306533,
      "learning_rate": 8.921473656603678e-06,
      "loss": 0.0908,
      "step": 8571
    },
    {
      "epoch": 5.484325015994882,
      "grad_norm": 1.1148045203762418,
      "learning_rate": 8.919413612888297e-06,
      "loss": 0.0929,
      "step": 8572
    },
    {
      "epoch": 5.484964811260396,
      "grad_norm": 1.0101171989631157,
      "learning_rate": 8.917353615571377e-06,
      "loss": 0.0873,
      "step": 8573
    },
    {
      "epoch": 5.485604606525912,
      "grad_norm": 0.9440911695523281,
      "learning_rate": 8.915293664741367e-06,
      "loss": 0.0819,
      "step": 8574
    },
    {
      "epoch": 5.486244401791427,
      "grad_norm": 1.0892141529441985,
      "learning_rate": 8.913233760486717e-06,
      "loss": 0.0836,
      "step": 8575
    },
    {
      "epoch": 5.486884197056942,
      "grad_norm": 1.0607106884586694,
      "learning_rate": 8.911173902895873e-06,
      "loss": 0.0888,
      "step": 8576
    },
    {
      "epoch": 5.487523992322457,
      "grad_norm": 0.936334263445555,
      "learning_rate": 8.909114092057286e-06,
      "loss": 0.0842,
      "step": 8577
    },
    {
      "epoch": 5.488163787587972,
      "grad_norm": 1.0081379015031655,
      "learning_rate": 8.907054328059402e-06,
      "loss": 0.0894,
      "step": 8578
    },
    {
      "epoch": 5.488803582853487,
      "grad_norm": 1.0147219538869,
      "learning_rate": 8.904994610990658e-06,
      "loss": 0.0827,
      "step": 8579
    },
    {
      "epoch": 5.4894433781190015,
      "grad_norm": 1.0270864510778948,
      "learning_rate": 8.9029349409395e-06,
      "loss": 0.0865,
      "step": 8580
    },
    {
      "epoch": 5.490083173384517,
      "grad_norm": 1.0871320180446638,
      "learning_rate": 8.900875317994362e-06,
      "loss": 0.089,
      "step": 8581
    },
    {
      "epoch": 5.490722968650032,
      "grad_norm": 0.9743654638172226,
      "learning_rate": 8.898815742243682e-06,
      "loss": 0.0807,
      "step": 8582
    },
    {
      "epoch": 5.491362763915547,
      "grad_norm": 1.0187626654390267,
      "learning_rate": 8.896756213775897e-06,
      "loss": 0.0783,
      "step": 8583
    },
    {
      "epoch": 5.492002559181062,
      "grad_norm": 1.0565038423960862,
      "learning_rate": 8.89469673267944e-06,
      "loss": 0.0892,
      "step": 8584
    },
    {
      "epoch": 5.492642354446577,
      "grad_norm": 1.1105884663926853,
      "learning_rate": 8.892637299042735e-06,
      "loss": 0.0899,
      "step": 8585
    },
    {
      "epoch": 5.493282149712092,
      "grad_norm": 1.0481314769256649,
      "learning_rate": 8.890577912954214e-06,
      "loss": 0.0867,
      "step": 8586
    },
    {
      "epoch": 5.4939219449776076,
      "grad_norm": 1.114963281848601,
      "learning_rate": 8.888518574502306e-06,
      "loss": 0.093,
      "step": 8587
    },
    {
      "epoch": 5.494561740243122,
      "grad_norm": 1.039146068579216,
      "learning_rate": 8.886459283775434e-06,
      "loss": 0.0859,
      "step": 8588
    },
    {
      "epoch": 5.495201535508637,
      "grad_norm": 0.9611099960537572,
      "learning_rate": 8.884400040862017e-06,
      "loss": 0.0883,
      "step": 8589
    },
    {
      "epoch": 5.4958413307741525,
      "grad_norm": 1.016384068654799,
      "learning_rate": 8.882340845850482e-06,
      "loss": 0.0872,
      "step": 8590
    },
    {
      "epoch": 5.496481126039667,
      "grad_norm": 1.1792062471734246,
      "learning_rate": 8.880281698829239e-06,
      "loss": 0.0921,
      "step": 8591
    },
    {
      "epoch": 5.497120921305182,
      "grad_norm": 1.017688646997759,
      "learning_rate": 8.878222599886707e-06,
      "loss": 0.09,
      "step": 8592
    },
    {
      "epoch": 5.497760716570697,
      "grad_norm": 1.1392146813335786,
      "learning_rate": 8.876163549111304e-06,
      "loss": 0.0996,
      "step": 8593
    },
    {
      "epoch": 5.498400511836213,
      "grad_norm": 1.0674856978660097,
      "learning_rate": 8.87410454659144e-06,
      "loss": 0.0974,
      "step": 8594
    },
    {
      "epoch": 5.499040307101727,
      "grad_norm": 0.9597348578530099,
      "learning_rate": 8.872045592415523e-06,
      "loss": 0.0847,
      "step": 8595
    },
    {
      "epoch": 5.499680102367242,
      "grad_norm": 0.9732459890279592,
      "learning_rate": 8.86998668667196e-06,
      "loss": 0.0841,
      "step": 8596
    },
    {
      "epoch": 5.500319897632758,
      "grad_norm": 0.9570744296303864,
      "learning_rate": 8.86792782944916e-06,
      "loss": 0.0911,
      "step": 8597
    },
    {
      "epoch": 5.500959692898273,
      "grad_norm": 1.0655098813460344,
      "learning_rate": 8.865869020835528e-06,
      "loss": 0.0927,
      "step": 8598
    },
    {
      "epoch": 5.501599488163787,
      "grad_norm": 0.9929215136378611,
      "learning_rate": 8.863810260919454e-06,
      "loss": 0.0801,
      "step": 8599
    },
    {
      "epoch": 5.502239283429303,
      "grad_norm": 1.1243022952417323,
      "learning_rate": 8.861751549789353e-06,
      "loss": 0.0924,
      "step": 8600
    },
    {
      "epoch": 5.502879078694818,
      "grad_norm": 1.1024651626491069,
      "learning_rate": 8.859692887533617e-06,
      "loss": 0.0857,
      "step": 8601
    },
    {
      "epoch": 5.503518873960333,
      "grad_norm": 1.0571640298892797,
      "learning_rate": 8.857634274240635e-06,
      "loss": 0.0846,
      "step": 8602
    },
    {
      "epoch": 5.5041586692258475,
      "grad_norm": 0.9767050198095328,
      "learning_rate": 8.855575709998809e-06,
      "loss": 0.0832,
      "step": 8603
    },
    {
      "epoch": 5.504798464491363,
      "grad_norm": 1.072575960004761,
      "learning_rate": 8.853517194896525e-06,
      "loss": 0.0882,
      "step": 8604
    },
    {
      "epoch": 5.505438259756878,
      "grad_norm": 1.0966350809735677,
      "learning_rate": 8.851458729022176e-06,
      "loss": 0.0918,
      "step": 8605
    },
    {
      "epoch": 5.5060780550223924,
      "grad_norm": 0.941457384021754,
      "learning_rate": 8.849400312464139e-06,
      "loss": 0.083,
      "step": 8606
    },
    {
      "epoch": 5.506717850287908,
      "grad_norm": 1.082760821134922,
      "learning_rate": 8.847341945310812e-06,
      "loss": 0.0887,
      "step": 8607
    },
    {
      "epoch": 5.507357645553423,
      "grad_norm": 0.9724154854416063,
      "learning_rate": 8.845283627650569e-06,
      "loss": 0.0798,
      "step": 8608
    },
    {
      "epoch": 5.507997440818938,
      "grad_norm": 1.0703535765704792,
      "learning_rate": 8.843225359571793e-06,
      "loss": 0.0905,
      "step": 8609
    },
    {
      "epoch": 5.508637236084453,
      "grad_norm": 1.041991956472516,
      "learning_rate": 8.841167141162864e-06,
      "loss": 0.0905,
      "step": 8610
    },
    {
      "epoch": 5.509277031349968,
      "grad_norm": 1.0654336033502798,
      "learning_rate": 8.839108972512156e-06,
      "loss": 0.0951,
      "step": 8611
    },
    {
      "epoch": 5.509916826615483,
      "grad_norm": 1.0266231045085146,
      "learning_rate": 8.83705085370804e-06,
      "loss": 0.0887,
      "step": 8612
    },
    {
      "epoch": 5.510556621880998,
      "grad_norm": 0.9617635136649997,
      "learning_rate": 8.834992784838895e-06,
      "loss": 0.0893,
      "step": 8613
    },
    {
      "epoch": 5.511196417146513,
      "grad_norm": 1.0050286734939304,
      "learning_rate": 8.832934765993088e-06,
      "loss": 0.0907,
      "step": 8614
    },
    {
      "epoch": 5.511836212412028,
      "grad_norm": 1.0009374200229564,
      "learning_rate": 8.830876797258987e-06,
      "loss": 0.0788,
      "step": 8615
    },
    {
      "epoch": 5.512476007677543,
      "grad_norm": 0.9471954749904559,
      "learning_rate": 8.828818878724951e-06,
      "loss": 0.0853,
      "step": 8616
    },
    {
      "epoch": 5.513115802943059,
      "grad_norm": 1.0636224278678816,
      "learning_rate": 8.826761010479353e-06,
      "loss": 0.0941,
      "step": 8617
    },
    {
      "epoch": 5.513755598208573,
      "grad_norm": 0.8769650945030805,
      "learning_rate": 8.824703192610552e-06,
      "loss": 0.0802,
      "step": 8618
    },
    {
      "epoch": 5.514395393474088,
      "grad_norm": 1.093844524879765,
      "learning_rate": 8.822645425206901e-06,
      "loss": 0.0957,
      "step": 8619
    },
    {
      "epoch": 5.515035188739604,
      "grad_norm": 1.1896763903161804,
      "learning_rate": 8.820587708356765e-06,
      "loss": 0.0936,
      "step": 8620
    },
    {
      "epoch": 5.515674984005118,
      "grad_norm": 1.0820907872938017,
      "learning_rate": 8.818530042148494e-06,
      "loss": 0.0908,
      "step": 8621
    },
    {
      "epoch": 5.516314779270633,
      "grad_norm": 0.9666760972486956,
      "learning_rate": 8.816472426670438e-06,
      "loss": 0.0869,
      "step": 8622
    },
    {
      "epoch": 5.516954574536149,
      "grad_norm": 0.9771650660428831,
      "learning_rate": 8.814414862010953e-06,
      "loss": 0.0874,
      "step": 8623
    },
    {
      "epoch": 5.517594369801664,
      "grad_norm": 1.1473767159340391,
      "learning_rate": 8.812357348258385e-06,
      "loss": 0.1015,
      "step": 8624
    },
    {
      "epoch": 5.518234165067178,
      "grad_norm": 0.9768414499151019,
      "learning_rate": 8.81029988550108e-06,
      "loss": 0.0844,
      "step": 8625
    },
    {
      "epoch": 5.5188739603326935,
      "grad_norm": 0.9637222415368187,
      "learning_rate": 8.808242473827383e-06,
      "loss": 0.0814,
      "step": 8626
    },
    {
      "epoch": 5.519513755598209,
      "grad_norm": 1.0037681188333305,
      "learning_rate": 8.806185113325632e-06,
      "loss": 0.0916,
      "step": 8627
    },
    {
      "epoch": 5.520153550863723,
      "grad_norm": 1.0176757471126103,
      "learning_rate": 8.80412780408417e-06,
      "loss": 0.089,
      "step": 8628
    },
    {
      "epoch": 5.520793346129238,
      "grad_norm": 0.9823397727948714,
      "learning_rate": 8.802070546191328e-06,
      "loss": 0.088,
      "step": 8629
    },
    {
      "epoch": 5.521433141394754,
      "grad_norm": 0.8670883680626167,
      "learning_rate": 8.800013339735451e-06,
      "loss": 0.0778,
      "step": 8630
    },
    {
      "epoch": 5.522072936660269,
      "grad_norm": 1.0137979406431317,
      "learning_rate": 8.797956184804867e-06,
      "loss": 0.0804,
      "step": 8631
    },
    {
      "epoch": 5.522712731925783,
      "grad_norm": 0.9744693681777185,
      "learning_rate": 8.795899081487902e-06,
      "loss": 0.0802,
      "step": 8632
    },
    {
      "epoch": 5.523352527191299,
      "grad_norm": 1.1396229603854318,
      "learning_rate": 8.793842029872892e-06,
      "loss": 0.0902,
      "step": 8633
    },
    {
      "epoch": 5.523992322456814,
      "grad_norm": 1.0266454497911819,
      "learning_rate": 8.79178503004816e-06,
      "loss": 0.0902,
      "step": 8634
    },
    {
      "epoch": 5.524632117722329,
      "grad_norm": 0.9755805374412381,
      "learning_rate": 8.78972808210203e-06,
      "loss": 0.0911,
      "step": 8635
    },
    {
      "epoch": 5.525271912987844,
      "grad_norm": 0.9952950802377447,
      "learning_rate": 8.787671186122816e-06,
      "loss": 0.0897,
      "step": 8636
    },
    {
      "epoch": 5.525911708253359,
      "grad_norm": 1.0169191324523204,
      "learning_rate": 8.785614342198852e-06,
      "loss": 0.0848,
      "step": 8637
    },
    {
      "epoch": 5.526551503518874,
      "grad_norm": 0.9661907531327099,
      "learning_rate": 8.783557550418447e-06,
      "loss": 0.0791,
      "step": 8638
    },
    {
      "epoch": 5.527191298784389,
      "grad_norm": 1.025977365211492,
      "learning_rate": 8.781500810869915e-06,
      "loss": 0.0895,
      "step": 8639
    },
    {
      "epoch": 5.527831094049904,
      "grad_norm": 0.8914627639187163,
      "learning_rate": 8.779444123641572e-06,
      "loss": 0.0796,
      "step": 8640
    },
    {
      "epoch": 5.528470889315419,
      "grad_norm": 1.1194148346766088,
      "learning_rate": 8.777387488821727e-06,
      "loss": 0.0879,
      "step": 8641
    },
    {
      "epoch": 5.529110684580934,
      "grad_norm": 0.9441575153012128,
      "learning_rate": 8.77533090649869e-06,
      "loss": 0.0822,
      "step": 8642
    },
    {
      "epoch": 5.529750479846449,
      "grad_norm": 1.0046144149210068,
      "learning_rate": 8.77327437676076e-06,
      "loss": 0.087,
      "step": 8643
    },
    {
      "epoch": 5.530390275111964,
      "grad_norm": 1.1315110694490302,
      "learning_rate": 8.77121789969625e-06,
      "loss": 0.0863,
      "step": 8644
    },
    {
      "epoch": 5.531030070377479,
      "grad_norm": 0.9465347041861787,
      "learning_rate": 8.769161475393458e-06,
      "loss": 0.0799,
      "step": 8645
    },
    {
      "epoch": 5.531669865642995,
      "grad_norm": 1.0494640207704176,
      "learning_rate": 8.767105103940678e-06,
      "loss": 0.0844,
      "step": 8646
    },
    {
      "epoch": 5.532309660908509,
      "grad_norm": 0.9574300385093993,
      "learning_rate": 8.765048785426216e-06,
      "loss": 0.0826,
      "step": 8647
    },
    {
      "epoch": 5.532949456174024,
      "grad_norm": 1.2012404408180748,
      "learning_rate": 8.762992519938363e-06,
      "loss": 0.0933,
      "step": 8648
    },
    {
      "epoch": 5.5335892514395395,
      "grad_norm": 1.0916236415417786,
      "learning_rate": 8.76093630756541e-06,
      "loss": 0.0866,
      "step": 8649
    },
    {
      "epoch": 5.534229046705054,
      "grad_norm": 1.0273931095731665,
      "learning_rate": 8.758880148395648e-06,
      "loss": 0.0799,
      "step": 8650
    },
    {
      "epoch": 5.534868841970569,
      "grad_norm": 0.9590008804894209,
      "learning_rate": 8.756824042517364e-06,
      "loss": 0.0918,
      "step": 8651
    },
    {
      "epoch": 5.535508637236084,
      "grad_norm": 0.9904145607448094,
      "learning_rate": 8.754767990018847e-06,
      "loss": 0.0884,
      "step": 8652
    },
    {
      "epoch": 5.5361484325016,
      "grad_norm": 1.0563637840247568,
      "learning_rate": 8.75271199098837e-06,
      "loss": 0.0911,
      "step": 8653
    },
    {
      "epoch": 5.536788227767115,
      "grad_norm": 1.1499269747726395,
      "learning_rate": 8.750656045514228e-06,
      "loss": 0.095,
      "step": 8654
    },
    {
      "epoch": 5.537428023032629,
      "grad_norm": 1.0318577711139192,
      "learning_rate": 8.74860015368469e-06,
      "loss": 0.0806,
      "step": 8655
    },
    {
      "epoch": 5.538067818298145,
      "grad_norm": 0.973472400433653,
      "learning_rate": 8.746544315588036e-06,
      "loss": 0.0814,
      "step": 8656
    },
    {
      "epoch": 5.53870761356366,
      "grad_norm": 1.0325372418963745,
      "learning_rate": 8.74448853131254e-06,
      "loss": 0.0896,
      "step": 8657
    },
    {
      "epoch": 5.539347408829174,
      "grad_norm": 1.0789046736912775,
      "learning_rate": 8.742432800946475e-06,
      "loss": 0.0884,
      "step": 8658
    },
    {
      "epoch": 5.53998720409469,
      "grad_norm": 1.0538307165685505,
      "learning_rate": 8.740377124578101e-06,
      "loss": 0.0905,
      "step": 8659
    },
    {
      "epoch": 5.540626999360205,
      "grad_norm": 1.0292410878386813,
      "learning_rate": 8.738321502295701e-06,
      "loss": 0.0917,
      "step": 8660
    },
    {
      "epoch": 5.54126679462572,
      "grad_norm": 1.1944189459522783,
      "learning_rate": 8.736265934187529e-06,
      "loss": 0.087,
      "step": 8661
    },
    {
      "epoch": 5.5419065898912345,
      "grad_norm": 1.1141067938286415,
      "learning_rate": 8.73421042034185e-06,
      "loss": 0.0913,
      "step": 8662
    },
    {
      "epoch": 5.54254638515675,
      "grad_norm": 1.004270429537511,
      "learning_rate": 8.732154960846921e-06,
      "loss": 0.0888,
      "step": 8663
    },
    {
      "epoch": 5.543186180422265,
      "grad_norm": 0.9843781838047698,
      "learning_rate": 8.730099555791006e-06,
      "loss": 0.0919,
      "step": 8664
    },
    {
      "epoch": 5.5438259756877795,
      "grad_norm": 0.9788223473277732,
      "learning_rate": 8.728044205262355e-06,
      "loss": 0.0952,
      "step": 8665
    },
    {
      "epoch": 5.544465770953295,
      "grad_norm": 0.9465144960032416,
      "learning_rate": 8.725988909349222e-06,
      "loss": 0.089,
      "step": 8666
    },
    {
      "epoch": 5.54510556621881,
      "grad_norm": 1.0400195946471864,
      "learning_rate": 8.723933668139861e-06,
      "loss": 0.088,
      "step": 8667
    },
    {
      "epoch": 5.545745361484325,
      "grad_norm": 0.9565483984829796,
      "learning_rate": 8.72187848172252e-06,
      "loss": 0.0835,
      "step": 8668
    },
    {
      "epoch": 5.54638515674984,
      "grad_norm": 1.0561604868736065,
      "learning_rate": 8.719823350185439e-06,
      "loss": 0.0824,
      "step": 8669
    },
    {
      "epoch": 5.547024952015355,
      "grad_norm": 0.9878382104573025,
      "learning_rate": 8.717768273616869e-06,
      "loss": 0.0856,
      "step": 8670
    },
    {
      "epoch": 5.54766474728087,
      "grad_norm": 1.085522010642822,
      "learning_rate": 8.71571325210505e-06,
      "loss": 0.0938,
      "step": 8671
    },
    {
      "epoch": 5.5483045425463855,
      "grad_norm": 1.0207213279994565,
      "learning_rate": 8.713658285738217e-06,
      "loss": 0.0886,
      "step": 8672
    },
    {
      "epoch": 5.5489443378119,
      "grad_norm": 1.0843464661635211,
      "learning_rate": 8.711603374604607e-06,
      "loss": 0.0957,
      "step": 8673
    },
    {
      "epoch": 5.549584133077415,
      "grad_norm": 1.063952011272016,
      "learning_rate": 8.709548518792459e-06,
      "loss": 0.0886,
      "step": 8674
    },
    {
      "epoch": 5.55022392834293,
      "grad_norm": 0.9444667809973946,
      "learning_rate": 8.707493718390003e-06,
      "loss": 0.082,
      "step": 8675
    },
    {
      "epoch": 5.550863723608446,
      "grad_norm": 1.0019550765382046,
      "learning_rate": 8.705438973485463e-06,
      "loss": 0.0876,
      "step": 8676
    },
    {
      "epoch": 5.55150351887396,
      "grad_norm": 1.0851171270264428,
      "learning_rate": 8.703384284167075e-06,
      "loss": 0.0895,
      "step": 8677
    },
    {
      "epoch": 5.552143314139475,
      "grad_norm": 1.0510258993739978,
      "learning_rate": 8.701329650523059e-06,
      "loss": 0.0929,
      "step": 8678
    },
    {
      "epoch": 5.552783109404991,
      "grad_norm": 1.0766350207690796,
      "learning_rate": 8.699275072641635e-06,
      "loss": 0.0901,
      "step": 8679
    },
    {
      "epoch": 5.553422904670505,
      "grad_norm": 1.013556333476096,
      "learning_rate": 8.697220550611028e-06,
      "loss": 0.0861,
      "step": 8680
    },
    {
      "epoch": 5.55406269993602,
      "grad_norm": 1.1180924903923877,
      "learning_rate": 8.695166084519454e-06,
      "loss": 0.0842,
      "step": 8681
    },
    {
      "epoch": 5.554702495201536,
      "grad_norm": 1.1391112174725226,
      "learning_rate": 8.693111674455126e-06,
      "loss": 0.0884,
      "step": 8682
    },
    {
      "epoch": 5.555342290467051,
      "grad_norm": 1.1447285114689982,
      "learning_rate": 8.691057320506254e-06,
      "loss": 0.0866,
      "step": 8683
    },
    {
      "epoch": 5.555982085732565,
      "grad_norm": 1.0404273626884237,
      "learning_rate": 8.689003022761059e-06,
      "loss": 0.0881,
      "step": 8684
    },
    {
      "epoch": 5.5566218809980805,
      "grad_norm": 1.0606280589337718,
      "learning_rate": 8.686948781307739e-06,
      "loss": 0.0828,
      "step": 8685
    },
    {
      "epoch": 5.557261676263596,
      "grad_norm": 1.0368645708060589,
      "learning_rate": 8.684894596234502e-06,
      "loss": 0.0925,
      "step": 8686
    },
    {
      "epoch": 5.557901471529111,
      "grad_norm": 1.0763830726933639,
      "learning_rate": 8.682840467629552e-06,
      "loss": 0.09,
      "step": 8687
    },
    {
      "epoch": 5.5585412667946255,
      "grad_norm": 1.089854937121145,
      "learning_rate": 8.68078639558109e-06,
      "loss": 0.0865,
      "step": 8688
    },
    {
      "epoch": 5.559181062060141,
      "grad_norm": 0.8842840449279045,
      "learning_rate": 8.678732380177309e-06,
      "loss": 0.0856,
      "step": 8689
    },
    {
      "epoch": 5.559820857325656,
      "grad_norm": 1.0264403032200067,
      "learning_rate": 8.676678421506415e-06,
      "loss": 0.0895,
      "step": 8690
    },
    {
      "epoch": 5.560460652591171,
      "grad_norm": 1.0154284820145825,
      "learning_rate": 8.674624519656594e-06,
      "loss": 0.0847,
      "step": 8691
    },
    {
      "epoch": 5.561100447856686,
      "grad_norm": 1.0553368167047743,
      "learning_rate": 8.67257067471604e-06,
      "loss": 0.0904,
      "step": 8692
    },
    {
      "epoch": 5.561740243122201,
      "grad_norm": 0.9231406937382057,
      "learning_rate": 8.670516886772936e-06,
      "loss": 0.0807,
      "step": 8693
    },
    {
      "epoch": 5.562380038387716,
      "grad_norm": 0.9935312821362621,
      "learning_rate": 8.668463155915474e-06,
      "loss": 0.0863,
      "step": 8694
    },
    {
      "epoch": 5.563019833653231,
      "grad_norm": 1.0284253075486036,
      "learning_rate": 8.666409482231837e-06,
      "loss": 0.0855,
      "step": 8695
    },
    {
      "epoch": 5.563659628918746,
      "grad_norm": 0.9527325788996411,
      "learning_rate": 8.6643558658102e-06,
      "loss": 0.0836,
      "step": 8696
    },
    {
      "epoch": 5.564299424184261,
      "grad_norm": 1.0302147573458393,
      "learning_rate": 8.66230230673875e-06,
      "loss": 0.0849,
      "step": 8697
    },
    {
      "epoch": 5.564939219449776,
      "grad_norm": 1.1250864173172834,
      "learning_rate": 8.66024880510566e-06,
      "loss": 0.094,
      "step": 8698
    },
    {
      "epoch": 5.565579014715291,
      "grad_norm": 0.9815550850830448,
      "learning_rate": 8.6581953609991e-06,
      "loss": 0.0779,
      "step": 8699
    },
    {
      "epoch": 5.566218809980806,
      "grad_norm": 0.9787260069556234,
      "learning_rate": 8.656141974507249e-06,
      "loss": 0.0859,
      "step": 8700
    },
    {
      "epoch": 5.566858605246321,
      "grad_norm": 0.9522429545800255,
      "learning_rate": 8.654088645718269e-06,
      "loss": 0.0843,
      "step": 8701
    },
    {
      "epoch": 5.567498400511836,
      "grad_norm": 1.155625756532909,
      "learning_rate": 8.652035374720327e-06,
      "loss": 0.092,
      "step": 8702
    },
    {
      "epoch": 5.568138195777351,
      "grad_norm": 0.9399647037194112,
      "learning_rate": 8.649982161601585e-06,
      "loss": 0.0848,
      "step": 8703
    },
    {
      "epoch": 5.568777991042866,
      "grad_norm": 1.0013742839536848,
      "learning_rate": 8.647929006450213e-06,
      "loss": 0.0878,
      "step": 8704
    },
    {
      "epoch": 5.569417786308382,
      "grad_norm": 1.0651332449669262,
      "learning_rate": 8.645875909354363e-06,
      "loss": 0.0892,
      "step": 8705
    },
    {
      "epoch": 5.570057581573896,
      "grad_norm": 1.098701769840622,
      "learning_rate": 8.643822870402188e-06,
      "loss": 0.0914,
      "step": 8706
    },
    {
      "epoch": 5.570697376839411,
      "grad_norm": 0.9934486690009166,
      "learning_rate": 8.641769889681852e-06,
      "loss": 0.088,
      "step": 8707
    },
    {
      "epoch": 5.5713371721049265,
      "grad_norm": 1.005268328976084,
      "learning_rate": 8.6397169672815e-06,
      "loss": 0.0816,
      "step": 8708
    },
    {
      "epoch": 5.571976967370442,
      "grad_norm": 1.1177540368217587,
      "learning_rate": 8.63766410328928e-06,
      "loss": 0.0934,
      "step": 8709
    },
    {
      "epoch": 5.572616762635956,
      "grad_norm": 1.2146954110514303,
      "learning_rate": 8.635611297793343e-06,
      "loss": 0.0915,
      "step": 8710
    },
    {
      "epoch": 5.5732565579014715,
      "grad_norm": 0.9843439205263623,
      "learning_rate": 8.63355855088183e-06,
      "loss": 0.0823,
      "step": 8711
    },
    {
      "epoch": 5.573896353166987,
      "grad_norm": 1.0630767904909817,
      "learning_rate": 8.631505862642879e-06,
      "loss": 0.0869,
      "step": 8712
    },
    {
      "epoch": 5.574536148432502,
      "grad_norm": 0.9706144852389464,
      "learning_rate": 8.62945323316463e-06,
      "loss": 0.081,
      "step": 8713
    },
    {
      "epoch": 5.575175943698016,
      "grad_norm": 0.9886121559533249,
      "learning_rate": 8.627400662535224e-06,
      "loss": 0.0828,
      "step": 8714
    },
    {
      "epoch": 5.575815738963532,
      "grad_norm": 0.9396506578070527,
      "learning_rate": 8.625348150842793e-06,
      "loss": 0.0854,
      "step": 8715
    },
    {
      "epoch": 5.576455534229047,
      "grad_norm": 1.0249985235969066,
      "learning_rate": 8.623295698175466e-06,
      "loss": 0.0861,
      "step": 8716
    },
    {
      "epoch": 5.577095329494561,
      "grad_norm": 0.9831870542117999,
      "learning_rate": 8.621243304621373e-06,
      "loss": 0.0853,
      "step": 8717
    },
    {
      "epoch": 5.577735124760077,
      "grad_norm": 0.9756799523255943,
      "learning_rate": 8.619190970268641e-06,
      "loss": 0.0817,
      "step": 8718
    },
    {
      "epoch": 5.578374920025592,
      "grad_norm": 1.0378235622306575,
      "learning_rate": 8.617138695205391e-06,
      "loss": 0.0899,
      "step": 8719
    },
    {
      "epoch": 5.579014715291107,
      "grad_norm": 1.0066257464113983,
      "learning_rate": 8.615086479519742e-06,
      "loss": 0.0872,
      "step": 8720
    },
    {
      "epoch": 5.5796545105566215,
      "grad_norm": 1.1419877881786071,
      "learning_rate": 8.613034323299821e-06,
      "loss": 0.0954,
      "step": 8721
    },
    {
      "epoch": 5.580294305822137,
      "grad_norm": 1.0370208049880891,
      "learning_rate": 8.61098222663374e-06,
      "loss": 0.084,
      "step": 8722
    },
    {
      "epoch": 5.580934101087652,
      "grad_norm": 1.0343897068519157,
      "learning_rate": 8.608930189609606e-06,
      "loss": 0.0883,
      "step": 8723
    },
    {
      "epoch": 5.581573896353167,
      "grad_norm": 1.1089794031925126,
      "learning_rate": 8.606878212315539e-06,
      "loss": 0.0858,
      "step": 8724
    },
    {
      "epoch": 5.582213691618682,
      "grad_norm": 0.9761550507933238,
      "learning_rate": 8.604826294839645e-06,
      "loss": 0.0851,
      "step": 8725
    },
    {
      "epoch": 5.582853486884197,
      "grad_norm": 1.076231456917918,
      "learning_rate": 8.602774437270021e-06,
      "loss": 0.0904,
      "step": 8726
    },
    {
      "epoch": 5.583493282149712,
      "grad_norm": 0.8613492078729845,
      "learning_rate": 8.600722639694783e-06,
      "loss": 0.0804,
      "step": 8727
    },
    {
      "epoch": 5.584133077415228,
      "grad_norm": 0.9154050074564952,
      "learning_rate": 8.598670902202025e-06,
      "loss": 0.0882,
      "step": 8728
    },
    {
      "epoch": 5.584772872680742,
      "grad_norm": 1.1342869201711188,
      "learning_rate": 8.596619224879846e-06,
      "loss": 0.0887,
      "step": 8729
    },
    {
      "epoch": 5.585412667946257,
      "grad_norm": 0.8661700217783564,
      "learning_rate": 8.594567607816339e-06,
      "loss": 0.0835,
      "step": 8730
    },
    {
      "epoch": 5.5860524632117725,
      "grad_norm": 1.0791471325391395,
      "learning_rate": 8.5925160510996e-06,
      "loss": 0.0907,
      "step": 8731
    },
    {
      "epoch": 5.586692258477287,
      "grad_norm": 0.9801860586450891,
      "learning_rate": 8.59046455481772e-06,
      "loss": 0.0867,
      "step": 8732
    },
    {
      "epoch": 5.587332053742802,
      "grad_norm": 1.0650102231460403,
      "learning_rate": 8.58841311905878e-06,
      "loss": 0.0901,
      "step": 8733
    },
    {
      "epoch": 5.5879718490083174,
      "grad_norm": 1.1373852144685863,
      "learning_rate": 8.586361743910873e-06,
      "loss": 0.097,
      "step": 8734
    },
    {
      "epoch": 5.588611644273833,
      "grad_norm": 1.0717327617544377,
      "learning_rate": 8.584310429462078e-06,
      "loss": 0.0852,
      "step": 8735
    },
    {
      "epoch": 5.589251439539347,
      "grad_norm": 1.083494094306059,
      "learning_rate": 8.582259175800472e-06,
      "loss": 0.0948,
      "step": 8736
    },
    {
      "epoch": 5.589891234804862,
      "grad_norm": 1.0707847163941453,
      "learning_rate": 8.580207983014136e-06,
      "loss": 0.0903,
      "step": 8737
    },
    {
      "epoch": 5.590531030070378,
      "grad_norm": 0.9717227223265443,
      "learning_rate": 8.578156851191145e-06,
      "loss": 0.0862,
      "step": 8738
    },
    {
      "epoch": 5.591170825335892,
      "grad_norm": 1.0756122072282417,
      "learning_rate": 8.576105780419572e-06,
      "loss": 0.0857,
      "step": 8739
    },
    {
      "epoch": 5.591810620601407,
      "grad_norm": 1.0643005924177193,
      "learning_rate": 8.57405477078748e-06,
      "loss": 0.0963,
      "step": 8740
    },
    {
      "epoch": 5.592450415866923,
      "grad_norm": 1.0810084389808072,
      "learning_rate": 8.572003822382942e-06,
      "loss": 0.0927,
      "step": 8741
    },
    {
      "epoch": 5.593090211132438,
      "grad_norm": 1.0312256923417686,
      "learning_rate": 8.569952935294019e-06,
      "loss": 0.0863,
      "step": 8742
    },
    {
      "epoch": 5.593730006397953,
      "grad_norm": 0.9118383599284923,
      "learning_rate": 8.56790210960877e-06,
      "loss": 0.081,
      "step": 8743
    },
    {
      "epoch": 5.5943698016634675,
      "grad_norm": 1.1003570426106932,
      "learning_rate": 8.56585134541526e-06,
      "loss": 0.0929,
      "step": 8744
    },
    {
      "epoch": 5.595009596928983,
      "grad_norm": 1.0066393692439568,
      "learning_rate": 8.563800642801543e-06,
      "loss": 0.0845,
      "step": 8745
    },
    {
      "epoch": 5.595649392194498,
      "grad_norm": 1.375246796458377,
      "learning_rate": 8.56175000185567e-06,
      "loss": 0.0949,
      "step": 8746
    },
    {
      "epoch": 5.5962891874600125,
      "grad_norm": 1.0550404649208744,
      "learning_rate": 8.559699422665694e-06,
      "loss": 0.0835,
      "step": 8747
    },
    {
      "epoch": 5.596928982725528,
      "grad_norm": 0.9435992954858671,
      "learning_rate": 8.557648905319663e-06,
      "loss": 0.0811,
      "step": 8748
    },
    {
      "epoch": 5.597568777991043,
      "grad_norm": 1.0300315027920073,
      "learning_rate": 8.555598449905622e-06,
      "loss": 0.0844,
      "step": 8749
    },
    {
      "epoch": 5.598208573256558,
      "grad_norm": 1.0270844613694878,
      "learning_rate": 8.55354805651161e-06,
      "loss": 0.0843,
      "step": 8750
    },
    {
      "epoch": 5.598848368522073,
      "grad_norm": 0.9937357266144347,
      "learning_rate": 8.551497725225677e-06,
      "loss": 0.0906,
      "step": 8751
    },
    {
      "epoch": 5.599488163787588,
      "grad_norm": 0.9509739212775803,
      "learning_rate": 8.549447456135855e-06,
      "loss": 0.0827,
      "step": 8752
    },
    {
      "epoch": 5.600127959053103,
      "grad_norm": 0.9104034042951763,
      "learning_rate": 8.547397249330174e-06,
      "loss": 0.0828,
      "step": 8753
    },
    {
      "epoch": 5.600767754318618,
      "grad_norm": 1.0798365832343442,
      "learning_rate": 8.545347104896677e-06,
      "loss": 0.092,
      "step": 8754
    },
    {
      "epoch": 5.601407549584133,
      "grad_norm": 0.9627267820066181,
      "learning_rate": 8.543297022923385e-06,
      "loss": 0.0837,
      "step": 8755
    },
    {
      "epoch": 5.602047344849648,
      "grad_norm": 1.0183147887992474,
      "learning_rate": 8.541247003498325e-06,
      "loss": 0.0807,
      "step": 8756
    },
    {
      "epoch": 5.6026871401151634,
      "grad_norm": 1.0836651612124029,
      "learning_rate": 8.539197046709526e-06,
      "loss": 0.0942,
      "step": 8757
    },
    {
      "epoch": 5.603326935380678,
      "grad_norm": 1.0646294180430314,
      "learning_rate": 8.537147152645009e-06,
      "loss": 0.0959,
      "step": 8758
    },
    {
      "epoch": 5.603966730646193,
      "grad_norm": 0.8704610023531917,
      "learning_rate": 8.535097321392792e-06,
      "loss": 0.0784,
      "step": 8759
    },
    {
      "epoch": 5.604606525911708,
      "grad_norm": 0.9415370409186634,
      "learning_rate": 8.533047553040886e-06,
      "loss": 0.0889,
      "step": 8760
    },
    {
      "epoch": 5.605246321177224,
      "grad_norm": 0.9927799917741204,
      "learning_rate": 8.530997847677311e-06,
      "loss": 0.0805,
      "step": 8761
    },
    {
      "epoch": 5.605886116442738,
      "grad_norm": 0.9367066534995209,
      "learning_rate": 8.528948205390076e-06,
      "loss": 0.0876,
      "step": 8762
    },
    {
      "epoch": 5.606525911708253,
      "grad_norm": 1.0336672125856112,
      "learning_rate": 8.526898626267182e-06,
      "loss": 0.0925,
      "step": 8763
    },
    {
      "epoch": 5.607165706973769,
      "grad_norm": 1.0568705390247624,
      "learning_rate": 8.524849110396647e-06,
      "loss": 0.0851,
      "step": 8764
    },
    {
      "epoch": 5.607805502239284,
      "grad_norm": 1.1744716542980573,
      "learning_rate": 8.522799657866466e-06,
      "loss": 0.0955,
      "step": 8765
    },
    {
      "epoch": 5.608445297504798,
      "grad_norm": 0.9340737388788414,
      "learning_rate": 8.520750268764636e-06,
      "loss": 0.0902,
      "step": 8766
    },
    {
      "epoch": 5.6090850927703135,
      "grad_norm": 1.0356868626411697,
      "learning_rate": 8.518700943179163e-06,
      "loss": 0.0898,
      "step": 8767
    },
    {
      "epoch": 5.609724888035829,
      "grad_norm": 1.0204718744364203,
      "learning_rate": 8.516651681198037e-06,
      "loss": 0.0866,
      "step": 8768
    },
    {
      "epoch": 5.610364683301343,
      "grad_norm": 0.9426205117796341,
      "learning_rate": 8.514602482909247e-06,
      "loss": 0.088,
      "step": 8769
    },
    {
      "epoch": 5.6110044785668585,
      "grad_norm": 0.942341902597029,
      "learning_rate": 8.512553348400782e-06,
      "loss": 0.0831,
      "step": 8770
    },
    {
      "epoch": 5.611644273832374,
      "grad_norm": 1.0055146224708367,
      "learning_rate": 8.510504277760633e-06,
      "loss": 0.0869,
      "step": 8771
    },
    {
      "epoch": 5.612284069097889,
      "grad_norm": 1.025510704974983,
      "learning_rate": 8.508455271076783e-06,
      "loss": 0.0877,
      "step": 8772
    },
    {
      "epoch": 5.612923864363403,
      "grad_norm": 1.09913459594915,
      "learning_rate": 8.506406328437203e-06,
      "loss": 0.0873,
      "step": 8773
    },
    {
      "epoch": 5.613563659628919,
      "grad_norm": 0.9281344371948391,
      "learning_rate": 8.504357449929885e-06,
      "loss": 0.0822,
      "step": 8774
    },
    {
      "epoch": 5.614203454894434,
      "grad_norm": 1.0104975443050577,
      "learning_rate": 8.502308635642795e-06,
      "loss": 0.0907,
      "step": 8775
    },
    {
      "epoch": 5.614843250159948,
      "grad_norm": 0.9823316980063372,
      "learning_rate": 8.500259885663908e-06,
      "loss": 0.0888,
      "step": 8776
    },
    {
      "epoch": 5.615483045425464,
      "grad_norm": 1.0510541897503907,
      "learning_rate": 8.498211200081194e-06,
      "loss": 0.0816,
      "step": 8777
    },
    {
      "epoch": 5.616122840690979,
      "grad_norm": 1.0512864505379877,
      "learning_rate": 8.496162578982621e-06,
      "loss": 0.0839,
      "step": 8778
    },
    {
      "epoch": 5.616762635956494,
      "grad_norm": 1.1125361196531616,
      "learning_rate": 8.494114022456152e-06,
      "loss": 0.088,
      "step": 8779
    },
    {
      "epoch": 5.617402431222009,
      "grad_norm": 1.0824718815019168,
      "learning_rate": 8.492065530589743e-06,
      "loss": 0.0882,
      "step": 8780
    },
    {
      "epoch": 5.618042226487524,
      "grad_norm": 0.9394481583658714,
      "learning_rate": 8.490017103471361e-06,
      "loss": 0.0905,
      "step": 8781
    },
    {
      "epoch": 5.618682021753039,
      "grad_norm": 1.0432395128235108,
      "learning_rate": 8.487968741188962e-06,
      "loss": 0.086,
      "step": 8782
    },
    {
      "epoch": 5.619321817018554,
      "grad_norm": 0.9443174580795628,
      "learning_rate": 8.48592044383049e-06,
      "loss": 0.0814,
      "step": 8783
    },
    {
      "epoch": 5.619961612284069,
      "grad_norm": 0.9839624939962888,
      "learning_rate": 8.483872211483905e-06,
      "loss": 0.0852,
      "step": 8784
    },
    {
      "epoch": 5.620601407549584,
      "grad_norm": 0.9679263259014804,
      "learning_rate": 8.481824044237147e-06,
      "loss": 0.0863,
      "step": 8785
    },
    {
      "epoch": 5.621241202815099,
      "grad_norm": 1.0598988868892474,
      "learning_rate": 8.479775942178164e-06,
      "loss": 0.0946,
      "step": 8786
    },
    {
      "epoch": 5.621880998080615,
      "grad_norm": 0.9643568141818459,
      "learning_rate": 8.477727905394899e-06,
      "loss": 0.0889,
      "step": 8787
    },
    {
      "epoch": 5.622520793346129,
      "grad_norm": 1.0315057679863369,
      "learning_rate": 8.475679933975292e-06,
      "loss": 0.0978,
      "step": 8788
    },
    {
      "epoch": 5.623160588611644,
      "grad_norm": 1.0320785370900603,
      "learning_rate": 8.473632028007275e-06,
      "loss": 0.0909,
      "step": 8789
    },
    {
      "epoch": 5.6238003838771595,
      "grad_norm": 1.0005867106395343,
      "learning_rate": 8.471584187578783e-06,
      "loss": 0.0925,
      "step": 8790
    },
    {
      "epoch": 5.624440179142674,
      "grad_norm": 1.0655846763913837,
      "learning_rate": 8.469536412777749e-06,
      "loss": 0.0918,
      "step": 8791
    },
    {
      "epoch": 5.625079974408189,
      "grad_norm": 1.0099820666121637,
      "learning_rate": 8.467488703692098e-06,
      "loss": 0.0821,
      "step": 8792
    },
    {
      "epoch": 5.6257197696737045,
      "grad_norm": 1.0166924556252248,
      "learning_rate": 8.465441060409752e-06,
      "loss": 0.084,
      "step": 8793
    },
    {
      "epoch": 5.62635956493922,
      "grad_norm": 0.8896932118263203,
      "learning_rate": 8.463393483018643e-06,
      "loss": 0.0804,
      "step": 8794
    },
    {
      "epoch": 5.626999360204734,
      "grad_norm": 1.003657866805717,
      "learning_rate": 8.461345971606682e-06,
      "loss": 0.0838,
      "step": 8795
    },
    {
      "epoch": 5.627639155470249,
      "grad_norm": 0.9945833875340123,
      "learning_rate": 8.459298526261789e-06,
      "loss": 0.0859,
      "step": 8796
    },
    {
      "epoch": 5.628278950735765,
      "grad_norm": 0.9934183010585363,
      "learning_rate": 8.457251147071873e-06,
      "loss": 0.0866,
      "step": 8797
    },
    {
      "epoch": 5.62891874600128,
      "grad_norm": 1.0440485874336818,
      "learning_rate": 8.455203834124851e-06,
      "loss": 0.0902,
      "step": 8798
    },
    {
      "epoch": 5.629558541266794,
      "grad_norm": 0.9811172766559384,
      "learning_rate": 8.453156587508629e-06,
      "loss": 0.0916,
      "step": 8799
    },
    {
      "epoch": 5.63019833653231,
      "grad_norm": 0.9458421273536439,
      "learning_rate": 8.451109407311111e-06,
      "loss": 0.0758,
      "step": 8800
    },
    {
      "epoch": 5.630838131797825,
      "grad_norm": 1.1820054524675834,
      "learning_rate": 8.4490622936202e-06,
      "loss": 0.1044,
      "step": 8801
    },
    {
      "epoch": 5.63147792706334,
      "grad_norm": 1.0691358753941698,
      "learning_rate": 8.447015246523796e-06,
      "loss": 0.0876,
      "step": 8802
    },
    {
      "epoch": 5.6321177223288545,
      "grad_norm": 1.1901937839915762,
      "learning_rate": 8.44496826610979e-06,
      "loss": 0.0941,
      "step": 8803
    },
    {
      "epoch": 5.63275751759437,
      "grad_norm": 1.0819058400441484,
      "learning_rate": 8.442921352466085e-06,
      "loss": 0.0876,
      "step": 8804
    },
    {
      "epoch": 5.633397312859885,
      "grad_norm": 1.1464731108801616,
      "learning_rate": 8.440874505680568e-06,
      "loss": 0.0913,
      "step": 8805
    },
    {
      "epoch": 5.6340371081253995,
      "grad_norm": 0.9672941612040329,
      "learning_rate": 8.438827725841125e-06,
      "loss": 0.0837,
      "step": 8806
    },
    {
      "epoch": 5.634676903390915,
      "grad_norm": 0.9525955190660917,
      "learning_rate": 8.436781013035642e-06,
      "loss": 0.08,
      "step": 8807
    },
    {
      "epoch": 5.63531669865643,
      "grad_norm": 1.0894995754081145,
      "learning_rate": 8.434734367352003e-06,
      "loss": 0.0868,
      "step": 8808
    },
    {
      "epoch": 5.635956493921945,
      "grad_norm": 0.9614864061347924,
      "learning_rate": 8.432687788878085e-06,
      "loss": 0.0872,
      "step": 8809
    },
    {
      "epoch": 5.63659628918746,
      "grad_norm": 0.9428060889661328,
      "learning_rate": 8.430641277701762e-06,
      "loss": 0.0862,
      "step": 8810
    },
    {
      "epoch": 5.637236084452975,
      "grad_norm": 0.9549637605186808,
      "learning_rate": 8.428594833910914e-06,
      "loss": 0.0838,
      "step": 8811
    },
    {
      "epoch": 5.63787587971849,
      "grad_norm": 1.061756443957204,
      "learning_rate": 8.426548457593408e-06,
      "loss": 0.0798,
      "step": 8812
    },
    {
      "epoch": 5.638515674984005,
      "grad_norm": 1.0145592018292358,
      "learning_rate": 8.424502148837112e-06,
      "loss": 0.0835,
      "step": 8813
    },
    {
      "epoch": 5.63915547024952,
      "grad_norm": 1.039605130497647,
      "learning_rate": 8.422455907729891e-06,
      "loss": 0.0902,
      "step": 8814
    },
    {
      "epoch": 5.639795265515035,
      "grad_norm": 1.0856567018876113,
      "learning_rate": 8.420409734359611e-06,
      "loss": 0.0872,
      "step": 8815
    },
    {
      "epoch": 5.6404350607805505,
      "grad_norm": 1.0828591450312244,
      "learning_rate": 8.418363628814124e-06,
      "loss": 0.0892,
      "step": 8816
    },
    {
      "epoch": 5.641074856046066,
      "grad_norm": 1.130191920686287,
      "learning_rate": 8.416317591181285e-06,
      "loss": 0.1011,
      "step": 8817
    },
    {
      "epoch": 5.64171465131158,
      "grad_norm": 1.0668499887819216,
      "learning_rate": 8.414271621548957e-06,
      "loss": 0.0913,
      "step": 8818
    },
    {
      "epoch": 5.642354446577095,
      "grad_norm": 1.0221638605210366,
      "learning_rate": 8.412225720004984e-06,
      "loss": 0.0931,
      "step": 8819
    },
    {
      "epoch": 5.642994241842611,
      "grad_norm": 0.9871511042777483,
      "learning_rate": 8.410179886637213e-06,
      "loss": 0.0865,
      "step": 8820
    },
    {
      "epoch": 5.643634037108125,
      "grad_norm": 0.98693915624854,
      "learning_rate": 8.408134121533491e-06,
      "loss": 0.0942,
      "step": 8821
    },
    {
      "epoch": 5.64427383237364,
      "grad_norm": 1.045351644189185,
      "learning_rate": 8.406088424781657e-06,
      "loss": 0.0955,
      "step": 8822
    },
    {
      "epoch": 5.644913627639156,
      "grad_norm": 0.9823640776215161,
      "learning_rate": 8.404042796469549e-06,
      "loss": 0.0818,
      "step": 8823
    },
    {
      "epoch": 5.645553422904671,
      "grad_norm": 1.1171154698357748,
      "learning_rate": 8.401997236685007e-06,
      "loss": 0.0885,
      "step": 8824
    },
    {
      "epoch": 5.646193218170185,
      "grad_norm": 0.9754075312941003,
      "learning_rate": 8.39995174551586e-06,
      "loss": 0.0863,
      "step": 8825
    },
    {
      "epoch": 5.6468330134357005,
      "grad_norm": 0.9660233486954256,
      "learning_rate": 8.397906323049941e-06,
      "loss": 0.0793,
      "step": 8826
    },
    {
      "epoch": 5.647472808701216,
      "grad_norm": 0.991283634643576,
      "learning_rate": 8.395860969375068e-06,
      "loss": 0.0906,
      "step": 8827
    },
    {
      "epoch": 5.64811260396673,
      "grad_norm": 1.0071347076641142,
      "learning_rate": 8.393815684579079e-06,
      "loss": 0.0788,
      "step": 8828
    },
    {
      "epoch": 5.6487523992322455,
      "grad_norm": 1.0352412981386412,
      "learning_rate": 8.391770468749784e-06,
      "loss": 0.0875,
      "step": 8829
    },
    {
      "epoch": 5.649392194497761,
      "grad_norm": 1.0216321241064337,
      "learning_rate": 8.389725321975005e-06,
      "loss": 0.0898,
      "step": 8830
    },
    {
      "epoch": 5.650031989763276,
      "grad_norm": 1.0087541665790494,
      "learning_rate": 8.387680244342555e-06,
      "loss": 0.087,
      "step": 8831
    },
    {
      "epoch": 5.65067178502879,
      "grad_norm": 0.9701850887911417,
      "learning_rate": 8.385635235940249e-06,
      "loss": 0.083,
      "step": 8832
    },
    {
      "epoch": 5.651311580294306,
      "grad_norm": 1.05437211252248,
      "learning_rate": 8.38359029685589e-06,
      "loss": 0.0916,
      "step": 8833
    },
    {
      "epoch": 5.651951375559821,
      "grad_norm": 1.0102457984674018,
      "learning_rate": 8.381545427177293e-06,
      "loss": 0.0875,
      "step": 8834
    },
    {
      "epoch": 5.652591170825336,
      "grad_norm": 1.0072003056254628,
      "learning_rate": 8.379500626992256e-06,
      "loss": 0.0852,
      "step": 8835
    },
    {
      "epoch": 5.653230966090851,
      "grad_norm": 1.015349959816428,
      "learning_rate": 8.377455896388579e-06,
      "loss": 0.0868,
      "step": 8836
    },
    {
      "epoch": 5.653870761356366,
      "grad_norm": 0.9954636049869604,
      "learning_rate": 8.375411235454058e-06,
      "loss": 0.0876,
      "step": 8837
    },
    {
      "epoch": 5.654510556621881,
      "grad_norm": 1.0443142800496494,
      "learning_rate": 8.373366644276491e-06,
      "loss": 0.0822,
      "step": 8838
    },
    {
      "epoch": 5.6551503518873965,
      "grad_norm": 0.9854665430370155,
      "learning_rate": 8.371322122943665e-06,
      "loss": 0.0897,
      "step": 8839
    },
    {
      "epoch": 5.655790147152911,
      "grad_norm": 1.046586446561738,
      "learning_rate": 8.369277671543367e-06,
      "loss": 0.0858,
      "step": 8840
    },
    {
      "epoch": 5.656429942418426,
      "grad_norm": 1.0797762337148649,
      "learning_rate": 8.36723329016339e-06,
      "loss": 0.0934,
      "step": 8841
    },
    {
      "epoch": 5.657069737683941,
      "grad_norm": 1.0405406416038385,
      "learning_rate": 8.36518897889151e-06,
      "loss": 0.0836,
      "step": 8842
    },
    {
      "epoch": 5.657709532949456,
      "grad_norm": 1.151269148912457,
      "learning_rate": 8.363144737815504e-06,
      "loss": 0.0899,
      "step": 8843
    },
    {
      "epoch": 5.658349328214971,
      "grad_norm": 0.9433331577388336,
      "learning_rate": 8.361100567023156e-06,
      "loss": 0.0833,
      "step": 8844
    },
    {
      "epoch": 5.658989123480486,
      "grad_norm": 0.9324121409369888,
      "learning_rate": 8.359056466602232e-06,
      "loss": 0.0874,
      "step": 8845
    },
    {
      "epoch": 5.659628918746002,
      "grad_norm": 1.0359493078758095,
      "learning_rate": 8.357012436640505e-06,
      "loss": 0.0867,
      "step": 8846
    },
    {
      "epoch": 5.660268714011516,
      "grad_norm": 1.0345203867952195,
      "learning_rate": 8.354968477225736e-06,
      "loss": 0.0894,
      "step": 8847
    },
    {
      "epoch": 5.660908509277031,
      "grad_norm": 0.8843832472465027,
      "learning_rate": 8.3529245884457e-06,
      "loss": 0.0823,
      "step": 8848
    },
    {
      "epoch": 5.6615483045425465,
      "grad_norm": 1.0155311700332517,
      "learning_rate": 8.350880770388151e-06,
      "loss": 0.0861,
      "step": 8849
    },
    {
      "epoch": 5.662188099808061,
      "grad_norm": 0.9370528725481446,
      "learning_rate": 8.348837023140847e-06,
      "loss": 0.0868,
      "step": 8850
    },
    {
      "epoch": 5.662827895073576,
      "grad_norm": 0.9553097907486948,
      "learning_rate": 8.346793346791547e-06,
      "loss": 0.0806,
      "step": 8851
    },
    {
      "epoch": 5.6634676903390915,
      "grad_norm": 0.9991128846349172,
      "learning_rate": 8.344749741427999e-06,
      "loss": 0.0793,
      "step": 8852
    },
    {
      "epoch": 5.664107485604607,
      "grad_norm": 1.075220279312573,
      "learning_rate": 8.342706207137949e-06,
      "loss": 0.0878,
      "step": 8853
    },
    {
      "epoch": 5.664747280870122,
      "grad_norm": 1.0055005101801633,
      "learning_rate": 8.340662744009149e-06,
      "loss": 0.0876,
      "step": 8854
    },
    {
      "epoch": 5.665387076135636,
      "grad_norm": 0.9815424985672847,
      "learning_rate": 8.33861935212934e-06,
      "loss": 0.0826,
      "step": 8855
    },
    {
      "epoch": 5.666026871401152,
      "grad_norm": 1.117454452854697,
      "learning_rate": 8.33657603158626e-06,
      "loss": 0.0912,
      "step": 8856
    },
    {
      "epoch": 5.666666666666667,
      "grad_norm": 0.9458861900519089,
      "learning_rate": 8.334532782467644e-06,
      "loss": 0.0796,
      "step": 8857
    },
    {
      "epoch": 5.667306461932181,
      "grad_norm": 0.9441645098253848,
      "learning_rate": 8.33248960486123e-06,
      "loss": 0.0891,
      "step": 8858
    },
    {
      "epoch": 5.667946257197697,
      "grad_norm": 0.944398988054084,
      "learning_rate": 8.33044649885475e-06,
      "loss": 0.085,
      "step": 8859
    },
    {
      "epoch": 5.668586052463212,
      "grad_norm": 1.0845019943665666,
      "learning_rate": 8.328403464535919e-06,
      "loss": 0.0984,
      "step": 8860
    },
    {
      "epoch": 5.669225847728727,
      "grad_norm": 1.0725798577698167,
      "learning_rate": 8.326360501992475e-06,
      "loss": 0.0883,
      "step": 8861
    },
    {
      "epoch": 5.669865642994242,
      "grad_norm": 1.0138971404769475,
      "learning_rate": 8.324317611312132e-06,
      "loss": 0.0864,
      "step": 8862
    },
    {
      "epoch": 5.670505438259757,
      "grad_norm": 0.9060412171612854,
      "learning_rate": 8.322274792582609e-06,
      "loss": 0.0719,
      "step": 8863
    },
    {
      "epoch": 5.671145233525272,
      "grad_norm": 0.9826859690083863,
      "learning_rate": 8.320232045891625e-06,
      "loss": 0.0845,
      "step": 8864
    },
    {
      "epoch": 5.6717850287907865,
      "grad_norm": 0.903787755771102,
      "learning_rate": 8.318189371326889e-06,
      "loss": 0.0796,
      "step": 8865
    },
    {
      "epoch": 5.672424824056302,
      "grad_norm": 0.999575529412447,
      "learning_rate": 8.316146768976107e-06,
      "loss": 0.0871,
      "step": 8866
    },
    {
      "epoch": 5.673064619321817,
      "grad_norm": 1.1293713390229152,
      "learning_rate": 8.314104238926988e-06,
      "loss": 0.0902,
      "step": 8867
    },
    {
      "epoch": 5.673704414587332,
      "grad_norm": 0.9600876752179123,
      "learning_rate": 8.312061781267235e-06,
      "loss": 0.0852,
      "step": 8868
    },
    {
      "epoch": 5.674344209852847,
      "grad_norm": 0.9379798782503201,
      "learning_rate": 8.310019396084548e-06,
      "loss": 0.0846,
      "step": 8869
    },
    {
      "epoch": 5.674984005118362,
      "grad_norm": 0.9603764067627347,
      "learning_rate": 8.307977083466616e-06,
      "loss": 0.0801,
      "step": 8870
    },
    {
      "epoch": 5.675623800383877,
      "grad_norm": 1.1093165163087109,
      "learning_rate": 8.305934843501143e-06,
      "loss": 0.089,
      "step": 8871
    },
    {
      "epoch": 5.6762635956493925,
      "grad_norm": 1.121524675492314,
      "learning_rate": 8.303892676275815e-06,
      "loss": 0.0923,
      "step": 8872
    },
    {
      "epoch": 5.676903390914907,
      "grad_norm": 1.0832177659513171,
      "learning_rate": 8.301850581878315e-06,
      "loss": 0.0935,
      "step": 8873
    },
    {
      "epoch": 5.677543186180422,
      "grad_norm": 1.0277760318573481,
      "learning_rate": 8.299808560396334e-06,
      "loss": 0.0896,
      "step": 8874
    },
    {
      "epoch": 5.6781829814459375,
      "grad_norm": 0.9903754743591395,
      "learning_rate": 8.297766611917547e-06,
      "loss": 0.0894,
      "step": 8875
    },
    {
      "epoch": 5.678822776711453,
      "grad_norm": 0.9298147814270239,
      "learning_rate": 8.295724736529635e-06,
      "loss": 0.0795,
      "step": 8876
    },
    {
      "epoch": 5.679462571976967,
      "grad_norm": 1.112356611033567,
      "learning_rate": 8.293682934320266e-06,
      "loss": 0.0886,
      "step": 8877
    },
    {
      "epoch": 5.680102367242482,
      "grad_norm": 0.9892971748230334,
      "learning_rate": 8.291641205377122e-06,
      "loss": 0.088,
      "step": 8878
    },
    {
      "epoch": 5.680742162507998,
      "grad_norm": 1.043301067167924,
      "learning_rate": 8.289599549787864e-06,
      "loss": 0.0846,
      "step": 8879
    },
    {
      "epoch": 5.681381957773512,
      "grad_norm": 0.9407205627317745,
      "learning_rate": 8.287557967640158e-06,
      "loss": 0.083,
      "step": 8880
    },
    {
      "epoch": 5.682021753039027,
      "grad_norm": 1.079629806938485,
      "learning_rate": 8.285516459021668e-06,
      "loss": 0.0859,
      "step": 8881
    },
    {
      "epoch": 5.682661548304543,
      "grad_norm": 1.1873267750535894,
      "learning_rate": 8.283475024020053e-06,
      "loss": 0.0894,
      "step": 8882
    },
    {
      "epoch": 5.683301343570058,
      "grad_norm": 0.8855353063115806,
      "learning_rate": 8.281433662722966e-06,
      "loss": 0.0816,
      "step": 8883
    },
    {
      "epoch": 5.683941138835572,
      "grad_norm": 0.9897731847298151,
      "learning_rate": 8.279392375218056e-06,
      "loss": 0.0808,
      "step": 8884
    },
    {
      "epoch": 5.684580934101088,
      "grad_norm": 1.0695195221855964,
      "learning_rate": 8.27735116159298e-06,
      "loss": 0.0969,
      "step": 8885
    },
    {
      "epoch": 5.685220729366603,
      "grad_norm": 1.0400388176766326,
      "learning_rate": 8.275310021935381e-06,
      "loss": 0.0924,
      "step": 8886
    },
    {
      "epoch": 5.685860524632118,
      "grad_norm": 1.032570941846042,
      "learning_rate": 8.2732689563329e-06,
      "loss": 0.0939,
      "step": 8887
    },
    {
      "epoch": 5.6865003198976325,
      "grad_norm": 1.0507265685302278,
      "learning_rate": 8.271227964873182e-06,
      "loss": 0.0877,
      "step": 8888
    },
    {
      "epoch": 5.687140115163148,
      "grad_norm": 1.0172696779409582,
      "learning_rate": 8.269187047643858e-06,
      "loss": 0.0843,
      "step": 8889
    },
    {
      "epoch": 5.687779910428663,
      "grad_norm": 1.0482382606806993,
      "learning_rate": 8.26714620473256e-06,
      "loss": 0.089,
      "step": 8890
    },
    {
      "epoch": 5.688419705694178,
      "grad_norm": 0.9766787765640507,
      "learning_rate": 8.265105436226925e-06,
      "loss": 0.0816,
      "step": 8891
    },
    {
      "epoch": 5.689059500959693,
      "grad_norm": 1.0542643078608858,
      "learning_rate": 8.263064742214577e-06,
      "loss": 0.0914,
      "step": 8892
    },
    {
      "epoch": 5.689699296225208,
      "grad_norm": 1.0036976011989434,
      "learning_rate": 8.26102412278314e-06,
      "loss": 0.0871,
      "step": 8893
    },
    {
      "epoch": 5.690339091490723,
      "grad_norm": 1.0960213391398206,
      "learning_rate": 8.258983578020228e-06,
      "loss": 0.0885,
      "step": 8894
    },
    {
      "epoch": 5.690978886756238,
      "grad_norm": 1.0405442022048097,
      "learning_rate": 8.256943108013468e-06,
      "loss": 0.0849,
      "step": 8895
    },
    {
      "epoch": 5.691618682021753,
      "grad_norm": 0.9660007522542355,
      "learning_rate": 8.254902712850472e-06,
      "loss": 0.0788,
      "step": 8896
    },
    {
      "epoch": 5.692258477287268,
      "grad_norm": 1.0717212148043063,
      "learning_rate": 8.252862392618848e-06,
      "loss": 0.0884,
      "step": 8897
    },
    {
      "epoch": 5.6928982725527835,
      "grad_norm": 0.9949093375991359,
      "learning_rate": 8.250822147406204e-06,
      "loss": 0.0839,
      "step": 8898
    },
    {
      "epoch": 5.693538067818298,
      "grad_norm": 0.9809042474460399,
      "learning_rate": 8.248781977300148e-06,
      "loss": 0.0813,
      "step": 8899
    },
    {
      "epoch": 5.694177863083813,
      "grad_norm": 1.030801303826371,
      "learning_rate": 8.246741882388273e-06,
      "loss": 0.0871,
      "step": 8900
    },
    {
      "epoch": 5.694817658349328,
      "grad_norm": 0.9471934664837216,
      "learning_rate": 8.244701862758187e-06,
      "loss": 0.0781,
      "step": 8901
    },
    {
      "epoch": 5.695457453614843,
      "grad_norm": 0.865473648910773,
      "learning_rate": 8.242661918497481e-06,
      "loss": 0.0789,
      "step": 8902
    },
    {
      "epoch": 5.696097248880358,
      "grad_norm": 0.8906620120969253,
      "learning_rate": 8.240622049693746e-06,
      "loss": 0.0814,
      "step": 8903
    },
    {
      "epoch": 5.696737044145873,
      "grad_norm": 1.1052182353098632,
      "learning_rate": 8.238582256434568e-06,
      "loss": 0.0845,
      "step": 8904
    },
    {
      "epoch": 5.697376839411389,
      "grad_norm": 1.06344696201439,
      "learning_rate": 8.236542538807539e-06,
      "loss": 0.0922,
      "step": 8905
    },
    {
      "epoch": 5.698016634676903,
      "grad_norm": 1.0631695346337353,
      "learning_rate": 8.234502896900234e-06,
      "loss": 0.0877,
      "step": 8906
    },
    {
      "epoch": 5.698656429942418,
      "grad_norm": 0.9906190415513031,
      "learning_rate": 8.23246333080023e-06,
      "loss": 0.0926,
      "step": 8907
    },
    {
      "epoch": 5.6992962252079336,
      "grad_norm": 1.1165739303352165,
      "learning_rate": 8.23042384059511e-06,
      "loss": 0.0966,
      "step": 8908
    },
    {
      "epoch": 5.699936020473449,
      "grad_norm": 1.1615543640179329,
      "learning_rate": 8.228384426372444e-06,
      "loss": 0.0984,
      "step": 8909
    },
    {
      "epoch": 5.700575815738963,
      "grad_norm": 1.0488643544332676,
      "learning_rate": 8.226345088219797e-06,
      "loss": 0.082,
      "step": 8910
    },
    {
      "epoch": 5.7012156110044785,
      "grad_norm": 0.9760929575294525,
      "learning_rate": 8.224305826224738e-06,
      "loss": 0.0886,
      "step": 8911
    },
    {
      "epoch": 5.701855406269994,
      "grad_norm": 1.0262159591966848,
      "learning_rate": 8.22226664047483e-06,
      "loss": 0.0903,
      "step": 8912
    },
    {
      "epoch": 5.702495201535509,
      "grad_norm": 0.9263301966100087,
      "learning_rate": 8.220227531057628e-06,
      "loss": 0.081,
      "step": 8913
    },
    {
      "epoch": 5.703134996801023,
      "grad_norm": 0.9707175529984007,
      "learning_rate": 8.218188498060686e-06,
      "loss": 0.0846,
      "step": 8914
    },
    {
      "epoch": 5.703774792066539,
      "grad_norm": 0.9755687654082285,
      "learning_rate": 8.216149541571565e-06,
      "loss": 0.0822,
      "step": 8915
    },
    {
      "epoch": 5.704414587332054,
      "grad_norm": 1.0230044463640477,
      "learning_rate": 8.21411066167781e-06,
      "loss": 0.0827,
      "step": 8916
    },
    {
      "epoch": 5.705054382597568,
      "grad_norm": 1.0175855592704892,
      "learning_rate": 8.212071858466963e-06,
      "loss": 0.087,
      "step": 8917
    },
    {
      "epoch": 5.705694177863084,
      "grad_norm": 1.1003040225471628,
      "learning_rate": 8.210033132026573e-06,
      "loss": 0.0961,
      "step": 8918
    },
    {
      "epoch": 5.706333973128599,
      "grad_norm": 0.9934226073280178,
      "learning_rate": 8.207994482444177e-06,
      "loss": 0.081,
      "step": 8919
    },
    {
      "epoch": 5.706973768394114,
      "grad_norm": 1.0476481640713693,
      "learning_rate": 8.205955909807304e-06,
      "loss": 0.0824,
      "step": 8920
    },
    {
      "epoch": 5.707613563659629,
      "grad_norm": 0.9961323761635557,
      "learning_rate": 8.203917414203498e-06,
      "loss": 0.0828,
      "step": 8921
    },
    {
      "epoch": 5.708253358925144,
      "grad_norm": 0.9600405677505588,
      "learning_rate": 8.201878995720284e-06,
      "loss": 0.0794,
      "step": 8922
    },
    {
      "epoch": 5.708893154190659,
      "grad_norm": 1.019310266820629,
      "learning_rate": 8.199840654445185e-06,
      "loss": 0.0915,
      "step": 8923
    },
    {
      "epoch": 5.709532949456174,
      "grad_norm": 1.1499346958096726,
      "learning_rate": 8.197802390465724e-06,
      "loss": 0.0924,
      "step": 8924
    },
    {
      "epoch": 5.710172744721689,
      "grad_norm": 0.9925693834216406,
      "learning_rate": 8.195764203869426e-06,
      "loss": 0.0806,
      "step": 8925
    },
    {
      "epoch": 5.710812539987204,
      "grad_norm": 1.1020201307700426,
      "learning_rate": 8.193726094743803e-06,
      "loss": 0.0892,
      "step": 8926
    },
    {
      "epoch": 5.711452335252719,
      "grad_norm": 1.0432953145407375,
      "learning_rate": 8.191688063176368e-06,
      "loss": 0.092,
      "step": 8927
    },
    {
      "epoch": 5.712092130518235,
      "grad_norm": 1.043356301788498,
      "learning_rate": 8.189650109254632e-06,
      "loss": 0.083,
      "step": 8928
    },
    {
      "epoch": 5.712731925783749,
      "grad_norm": 0.9909853709635276,
      "learning_rate": 8.1876122330661e-06,
      "loss": 0.0777,
      "step": 8929
    },
    {
      "epoch": 5.713371721049264,
      "grad_norm": 1.1068977246102567,
      "learning_rate": 8.18557443469827e-06,
      "loss": 0.0965,
      "step": 8930
    },
    {
      "epoch": 5.7140115163147795,
      "grad_norm": 1.0696048544988723,
      "learning_rate": 8.183536714238651e-06,
      "loss": 0.0833,
      "step": 8931
    },
    {
      "epoch": 5.714651311580294,
      "grad_norm": 1.0104198217052727,
      "learning_rate": 8.181499071774735e-06,
      "loss": 0.0875,
      "step": 8932
    },
    {
      "epoch": 5.715291106845809,
      "grad_norm": 0.9618550848632706,
      "learning_rate": 8.179461507394011e-06,
      "loss": 0.0811,
      "step": 8933
    },
    {
      "epoch": 5.7159309021113245,
      "grad_norm": 0.9294397620597716,
      "learning_rate": 8.177424021183972e-06,
      "loss": 0.0782,
      "step": 8934
    },
    {
      "epoch": 5.71657069737684,
      "grad_norm": 0.9844429193175953,
      "learning_rate": 8.175386613232104e-06,
      "loss": 0.0867,
      "step": 8935
    },
    {
      "epoch": 5.717210492642354,
      "grad_norm": 1.0555312366962382,
      "learning_rate": 8.173349283625892e-06,
      "loss": 0.0934,
      "step": 8936
    },
    {
      "epoch": 5.717850287907869,
      "grad_norm": 1.072012620352059,
      "learning_rate": 8.171312032452806e-06,
      "loss": 0.0948,
      "step": 8937
    },
    {
      "epoch": 5.718490083173385,
      "grad_norm": 1.0662039540731083,
      "learning_rate": 8.169274859800331e-06,
      "loss": 0.083,
      "step": 8938
    },
    {
      "epoch": 5.719129878438899,
      "grad_norm": 1.2495707193365542,
      "learning_rate": 8.167237765755938e-06,
      "loss": 0.0912,
      "step": 8939
    },
    {
      "epoch": 5.719769673704414,
      "grad_norm": 1.0060950747877346,
      "learning_rate": 8.165200750407094e-06,
      "loss": 0.0918,
      "step": 8940
    },
    {
      "epoch": 5.72040946896993,
      "grad_norm": 1.1470586736622013,
      "learning_rate": 8.163163813841269e-06,
      "loss": 0.0949,
      "step": 8941
    },
    {
      "epoch": 5.721049264235445,
      "grad_norm": 1.0728944994837997,
      "learning_rate": 8.161126956145919e-06,
      "loss": 0.0944,
      "step": 8942
    },
    {
      "epoch": 5.72168905950096,
      "grad_norm": 1.205514005122656,
      "learning_rate": 8.15909017740851e-06,
      "loss": 0.1131,
      "step": 8943
    },
    {
      "epoch": 5.722328854766475,
      "grad_norm": 0.9819265534955353,
      "learning_rate": 8.157053477716487e-06,
      "loss": 0.0861,
      "step": 8944
    },
    {
      "epoch": 5.72296865003199,
      "grad_norm": 1.0813796300432077,
      "learning_rate": 8.155016857157315e-06,
      "loss": 0.0879,
      "step": 8945
    },
    {
      "epoch": 5.723608445297505,
      "grad_norm": 0.9142706694587572,
      "learning_rate": 8.152980315818437e-06,
      "loss": 0.0883,
      "step": 8946
    },
    {
      "epoch": 5.7242482405630195,
      "grad_norm": 0.9922386212174134,
      "learning_rate": 8.150943853787296e-06,
      "loss": 0.0827,
      "step": 8947
    },
    {
      "epoch": 5.724888035828535,
      "grad_norm": 0.9426740435859717,
      "learning_rate": 8.14890747115134e-06,
      "loss": 0.0851,
      "step": 8948
    },
    {
      "epoch": 5.72552783109405,
      "grad_norm": 1.0519830921989697,
      "learning_rate": 8.146871167998003e-06,
      "loss": 0.0878,
      "step": 8949
    },
    {
      "epoch": 5.726167626359565,
      "grad_norm": 1.077132870104048,
      "learning_rate": 8.144834944414718e-06,
      "loss": 0.0919,
      "step": 8950
    },
    {
      "epoch": 5.72680742162508,
      "grad_norm": 1.0735410530618712,
      "learning_rate": 8.142798800488924e-06,
      "loss": 0.0894,
      "step": 8951
    },
    {
      "epoch": 5.727447216890595,
      "grad_norm": 1.0086314688435734,
      "learning_rate": 8.140762736308048e-06,
      "loss": 0.0836,
      "step": 8952
    },
    {
      "epoch": 5.72808701215611,
      "grad_norm": 1.0331622662153073,
      "learning_rate": 8.138726751959511e-06,
      "loss": 0.0886,
      "step": 8953
    },
    {
      "epoch": 5.728726807421625,
      "grad_norm": 0.9374356579785653,
      "learning_rate": 8.136690847530732e-06,
      "loss": 0.0784,
      "step": 8954
    },
    {
      "epoch": 5.72936660268714,
      "grad_norm": 0.9806952801797586,
      "learning_rate": 8.134655023109136e-06,
      "loss": 0.0809,
      "step": 8955
    },
    {
      "epoch": 5.730006397952655,
      "grad_norm": 1.0122277479262802,
      "learning_rate": 8.132619278782139e-06,
      "loss": 0.0831,
      "step": 8956
    },
    {
      "epoch": 5.7306461932181705,
      "grad_norm": 0.9928400291834785,
      "learning_rate": 8.130583614637144e-06,
      "loss": 0.0873,
      "step": 8957
    },
    {
      "epoch": 5.731285988483685,
      "grad_norm": 1.0769739971030063,
      "learning_rate": 8.128548030761565e-06,
      "loss": 0.0828,
      "step": 8958
    },
    {
      "epoch": 5.7319257837492,
      "grad_norm": 1.0528479253960734,
      "learning_rate": 8.126512527242804e-06,
      "loss": 0.0816,
      "step": 8959
    },
    {
      "epoch": 5.732565579014715,
      "grad_norm": 1.0561878216933653,
      "learning_rate": 8.124477104168265e-06,
      "loss": 0.0838,
      "step": 8960
    },
    {
      "epoch": 5.733205374280231,
      "grad_norm": 1.0093679032701732,
      "learning_rate": 8.122441761625337e-06,
      "loss": 0.0817,
      "step": 8961
    },
    {
      "epoch": 5.733845169545745,
      "grad_norm": 1.0447445308579935,
      "learning_rate": 8.120406499701423e-06,
      "loss": 0.0881,
      "step": 8962
    },
    {
      "epoch": 5.73448496481126,
      "grad_norm": 1.0217085855833843,
      "learning_rate": 8.118371318483911e-06,
      "loss": 0.0887,
      "step": 8963
    },
    {
      "epoch": 5.735124760076776,
      "grad_norm": 1.0347191672454072,
      "learning_rate": 8.116336218060186e-06,
      "loss": 0.0772,
      "step": 8964
    },
    {
      "epoch": 5.735764555342291,
      "grad_norm": 0.9827912530607298,
      "learning_rate": 8.114301198517635e-06,
      "loss": 0.0819,
      "step": 8965
    },
    {
      "epoch": 5.736404350607805,
      "grad_norm": 0.9108390435061992,
      "learning_rate": 8.112266259943638e-06,
      "loss": 0.0813,
      "step": 8966
    },
    {
      "epoch": 5.737044145873321,
      "grad_norm": 0.9960895551790565,
      "learning_rate": 8.110231402425562e-06,
      "loss": 0.0865,
      "step": 8967
    },
    {
      "epoch": 5.737683941138836,
      "grad_norm": 0.9498346649228419,
      "learning_rate": 8.108196626050795e-06,
      "loss": 0.0887,
      "step": 8968
    },
    {
      "epoch": 5.73832373640435,
      "grad_norm": 0.9826443039166824,
      "learning_rate": 8.106161930906697e-06,
      "loss": 0.0837,
      "step": 8969
    },
    {
      "epoch": 5.7389635316698655,
      "grad_norm": 1.0473075505079572,
      "learning_rate": 8.10412731708064e-06,
      "loss": 0.0878,
      "step": 8970
    },
    {
      "epoch": 5.739603326935381,
      "grad_norm": 1.023640491673056,
      "learning_rate": 8.102092784659978e-06,
      "loss": 0.086,
      "step": 8971
    },
    {
      "epoch": 5.740243122200896,
      "grad_norm": 1.14443372376883,
      "learning_rate": 8.10005833373208e-06,
      "loss": 0.089,
      "step": 8972
    },
    {
      "epoch": 5.74088291746641,
      "grad_norm": 0.9886351332098537,
      "learning_rate": 8.098023964384297e-06,
      "loss": 0.0883,
      "step": 8973
    },
    {
      "epoch": 5.741522712731926,
      "grad_norm": 1.0285558027555097,
      "learning_rate": 8.095989676703976e-06,
      "loss": 0.0813,
      "step": 8974
    },
    {
      "epoch": 5.742162507997441,
      "grad_norm": 1.0192223167745245,
      "learning_rate": 8.093955470778479e-06,
      "loss": 0.0835,
      "step": 8975
    },
    {
      "epoch": 5.742802303262955,
      "grad_norm": 1.0917179228632237,
      "learning_rate": 8.091921346695141e-06,
      "loss": 0.0882,
      "step": 8976
    },
    {
      "epoch": 5.743442098528471,
      "grad_norm": 1.0499479623693533,
      "learning_rate": 8.089887304541304e-06,
      "loss": 0.0977,
      "step": 8977
    },
    {
      "epoch": 5.744081893793986,
      "grad_norm": 1.001155005720094,
      "learning_rate": 8.08785334440431e-06,
      "loss": 0.0891,
      "step": 8978
    },
    {
      "epoch": 5.744721689059501,
      "grad_norm": 0.931310012186312,
      "learning_rate": 8.085819466371492e-06,
      "loss": 0.0866,
      "step": 8979
    },
    {
      "epoch": 5.7453614843250165,
      "grad_norm": 0.8914507734460824,
      "learning_rate": 8.083785670530183e-06,
      "loss": 0.0809,
      "step": 8980
    },
    {
      "epoch": 5.746001279590531,
      "grad_norm": 0.885395647280975,
      "learning_rate": 8.081751956967702e-06,
      "loss": 0.0753,
      "step": 8981
    },
    {
      "epoch": 5.746641074856046,
      "grad_norm": 1.0214913520876638,
      "learning_rate": 8.079718325771386e-06,
      "loss": 0.0838,
      "step": 8982
    },
    {
      "epoch": 5.747280870121561,
      "grad_norm": 1.143045291205945,
      "learning_rate": 8.077684777028547e-06,
      "loss": 0.0938,
      "step": 8983
    },
    {
      "epoch": 5.747920665387076,
      "grad_norm": 1.2270559440450006,
      "learning_rate": 8.075651310826503e-06,
      "loss": 0.1022,
      "step": 8984
    },
    {
      "epoch": 5.748560460652591,
      "grad_norm": 1.0329277145615883,
      "learning_rate": 8.073617927252571e-06,
      "loss": 0.0879,
      "step": 8985
    },
    {
      "epoch": 5.749200255918106,
      "grad_norm": 1.016120225518572,
      "learning_rate": 8.071584626394058e-06,
      "loss": 0.0741,
      "step": 8986
    },
    {
      "epoch": 5.749840051183622,
      "grad_norm": 1.170890318253563,
      "learning_rate": 8.069551408338269e-06,
      "loss": 0.0918,
      "step": 8987
    },
    {
      "epoch": 5.750479846449136,
      "grad_norm": 1.0695452204411815,
      "learning_rate": 8.06751827317251e-06,
      "loss": 0.0828,
      "step": 8988
    },
    {
      "epoch": 5.751119641714651,
      "grad_norm": 1.0790052430503816,
      "learning_rate": 8.065485220984079e-06,
      "loss": 0.0887,
      "step": 8989
    },
    {
      "epoch": 5.751759436980167,
      "grad_norm": 1.0831078417729585,
      "learning_rate": 8.063452251860273e-06,
      "loss": 0.0894,
      "step": 8990
    },
    {
      "epoch": 5.752399232245681,
      "grad_norm": 1.1518007898022793,
      "learning_rate": 8.061419365888376e-06,
      "loss": 0.0969,
      "step": 8991
    },
    {
      "epoch": 5.753039027511196,
      "grad_norm": 0.9757944747903763,
      "learning_rate": 8.059386563155689e-06,
      "loss": 0.0874,
      "step": 8992
    },
    {
      "epoch": 5.7536788227767115,
      "grad_norm": 0.9761949917074628,
      "learning_rate": 8.05735384374949e-06,
      "loss": 0.0897,
      "step": 8993
    },
    {
      "epoch": 5.754318618042227,
      "grad_norm": 0.980531918034324,
      "learning_rate": 8.055321207757058e-06,
      "loss": 0.0833,
      "step": 8994
    },
    {
      "epoch": 5.754958413307741,
      "grad_norm": 0.9332237890310386,
      "learning_rate": 8.05328865526568e-06,
      "loss": 0.0859,
      "step": 8995
    },
    {
      "epoch": 5.755598208573256,
      "grad_norm": 1.0984408862885309,
      "learning_rate": 8.05125618636262e-06,
      "loss": 0.0939,
      "step": 8996
    },
    {
      "epoch": 5.756238003838772,
      "grad_norm": 1.1003720049749819,
      "learning_rate": 8.04922380113515e-06,
      "loss": 0.0886,
      "step": 8997
    },
    {
      "epoch": 5.756877799104287,
      "grad_norm": 0.97361048502031,
      "learning_rate": 8.047191499670545e-06,
      "loss": 0.0885,
      "step": 8998
    },
    {
      "epoch": 5.757517594369801,
      "grad_norm": 0.9781360545663107,
      "learning_rate": 8.045159282056063e-06,
      "loss": 0.0889,
      "step": 8999
    },
    {
      "epoch": 5.758157389635317,
      "grad_norm": 0.9472628433075783,
      "learning_rate": 8.043127148378962e-06,
      "loss": 0.0903,
      "step": 9000
    },
    {
      "epoch": 5.758797184900832,
      "grad_norm": 0.9554343282708981,
      "learning_rate": 8.0410950987265e-06,
      "loss": 0.0858,
      "step": 9001
    },
    {
      "epoch": 5.759436980166347,
      "grad_norm": 1.0195620169017445,
      "learning_rate": 8.039063133185932e-06,
      "loss": 0.0913,
      "step": 9002
    },
    {
      "epoch": 5.760076775431862,
      "grad_norm": 1.0453922004249816,
      "learning_rate": 8.037031251844504e-06,
      "loss": 0.0897,
      "step": 9003
    },
    {
      "epoch": 5.760716570697377,
      "grad_norm": 1.0006980573489472,
      "learning_rate": 8.034999454789458e-06,
      "loss": 0.0866,
      "step": 9004
    },
    {
      "epoch": 5.761356365962892,
      "grad_norm": 0.9942998747206335,
      "learning_rate": 8.032967742108044e-06,
      "loss": 0.0914,
      "step": 9005
    },
    {
      "epoch": 5.7619961612284065,
      "grad_norm": 1.1268443194092543,
      "learning_rate": 8.030936113887498e-06,
      "loss": 0.0933,
      "step": 9006
    },
    {
      "epoch": 5.762635956493922,
      "grad_norm": 1.0308375934900516,
      "learning_rate": 8.028904570215048e-06,
      "loss": 0.0811,
      "step": 9007
    },
    {
      "epoch": 5.763275751759437,
      "grad_norm": 1.0383264324552015,
      "learning_rate": 8.026873111177933e-06,
      "loss": 0.0832,
      "step": 9008
    },
    {
      "epoch": 5.763915547024952,
      "grad_norm": 0.9437507266600753,
      "learning_rate": 8.024841736863375e-06,
      "loss": 0.0742,
      "step": 9009
    },
    {
      "epoch": 5.764555342290467,
      "grad_norm": 1.1005457751193541,
      "learning_rate": 8.022810447358602e-06,
      "loss": 0.0863,
      "step": 9010
    },
    {
      "epoch": 5.765195137555982,
      "grad_norm": 1.0742061807903829,
      "learning_rate": 8.020779242750825e-06,
      "loss": 0.0866,
      "step": 9011
    },
    {
      "epoch": 5.765834932821497,
      "grad_norm": 1.1063514453264367,
      "learning_rate": 8.018748123127272e-06,
      "loss": 0.0921,
      "step": 9012
    },
    {
      "epoch": 5.766474728087012,
      "grad_norm": 1.0091503020175752,
      "learning_rate": 8.016717088575149e-06,
      "loss": 0.0819,
      "step": 9013
    },
    {
      "epoch": 5.767114523352527,
      "grad_norm": 0.9296565295973337,
      "learning_rate": 8.014686139181662e-06,
      "loss": 0.0798,
      "step": 9014
    },
    {
      "epoch": 5.767754318618042,
      "grad_norm": 1.0214628225563365,
      "learning_rate": 8.012655275034026e-06,
      "loss": 0.0816,
      "step": 9015
    },
    {
      "epoch": 5.7683941138835575,
      "grad_norm": 1.0755745459474393,
      "learning_rate": 8.010624496219437e-06,
      "loss": 0.0938,
      "step": 9016
    },
    {
      "epoch": 5.769033909149073,
      "grad_norm": 1.0698267093521976,
      "learning_rate": 8.00859380282509e-06,
      "loss": 0.0882,
      "step": 9017
    },
    {
      "epoch": 5.769673704414587,
      "grad_norm": 1.038704518500869,
      "learning_rate": 8.006563194938186e-06,
      "loss": 0.0812,
      "step": 9018
    },
    {
      "epoch": 5.770313499680102,
      "grad_norm": 0.9503419775253353,
      "learning_rate": 8.004532672645913e-06,
      "loss": 0.0842,
      "step": 9019
    },
    {
      "epoch": 5.770953294945618,
      "grad_norm": 0.990828970730608,
      "learning_rate": 8.002502236035455e-06,
      "loss": 0.079,
      "step": 9020
    },
    {
      "epoch": 5.771593090211132,
      "grad_norm": 0.988526030070403,
      "learning_rate": 8.000471885193995e-06,
      "loss": 0.0945,
      "step": 9021
    },
    {
      "epoch": 5.772232885476647,
      "grad_norm": 1.0308043619755713,
      "learning_rate": 7.99844162020872e-06,
      "loss": 0.0846,
      "step": 9022
    },
    {
      "epoch": 5.772872680742163,
      "grad_norm": 1.0789255645496194,
      "learning_rate": 7.9964114411668e-06,
      "loss": 0.0912,
      "step": 9023
    },
    {
      "epoch": 5.773512476007678,
      "grad_norm": 1.1112502772119757,
      "learning_rate": 7.994381348155407e-06,
      "loss": 0.0943,
      "step": 9024
    },
    {
      "epoch": 5.774152271273192,
      "grad_norm": 1.0247728510465457,
      "learning_rate": 7.992351341261715e-06,
      "loss": 0.0872,
      "step": 9025
    },
    {
      "epoch": 5.774792066538708,
      "grad_norm": 1.0344948870291446,
      "learning_rate": 7.990321420572885e-06,
      "loss": 0.0902,
      "step": 9026
    },
    {
      "epoch": 5.775431861804223,
      "grad_norm": 0.994094220189481,
      "learning_rate": 7.988291586176071e-06,
      "loss": 0.0838,
      "step": 9027
    },
    {
      "epoch": 5.776071657069737,
      "grad_norm": 0.9944984453659463,
      "learning_rate": 7.986261838158443e-06,
      "loss": 0.0899,
      "step": 9028
    },
    {
      "epoch": 5.7767114523352525,
      "grad_norm": 0.8545614173013029,
      "learning_rate": 7.98423217660715e-06,
      "loss": 0.0825,
      "step": 9029
    },
    {
      "epoch": 5.777351247600768,
      "grad_norm": 1.0399410910571243,
      "learning_rate": 7.982202601609342e-06,
      "loss": 0.0905,
      "step": 9030
    },
    {
      "epoch": 5.777991042866283,
      "grad_norm": 1.1166725881818513,
      "learning_rate": 7.980173113252164e-06,
      "loss": 0.0957,
      "step": 9031
    },
    {
      "epoch": 5.7786308381317975,
      "grad_norm": 0.9815675867037671,
      "learning_rate": 7.978143711622759e-06,
      "loss": 0.0849,
      "step": 9032
    },
    {
      "epoch": 5.779270633397313,
      "grad_norm": 0.979758966373581,
      "learning_rate": 7.976114396808268e-06,
      "loss": 0.0813,
      "step": 9033
    },
    {
      "epoch": 5.779910428662828,
      "grad_norm": 1.1888979516847107,
      "learning_rate": 7.97408516889582e-06,
      "loss": 0.0941,
      "step": 9034
    },
    {
      "epoch": 5.780550223928343,
      "grad_norm": 1.0087989519971245,
      "learning_rate": 7.972056027972554e-06,
      "loss": 0.0867,
      "step": 9035
    },
    {
      "epoch": 5.781190019193858,
      "grad_norm": 0.9915964339973913,
      "learning_rate": 7.970026974125597e-06,
      "loss": 0.083,
      "step": 9036
    },
    {
      "epoch": 5.781829814459373,
      "grad_norm": 0.9488816050591649,
      "learning_rate": 7.96799800744207e-06,
      "loss": 0.0915,
      "step": 9037
    },
    {
      "epoch": 5.782469609724888,
      "grad_norm": 1.0052893851945641,
      "learning_rate": 7.96596912800909e-06,
      "loss": 0.0858,
      "step": 9038
    },
    {
      "epoch": 5.7831094049904035,
      "grad_norm": 1.0196036932776655,
      "learning_rate": 7.963940335913783e-06,
      "loss": 0.0858,
      "step": 9039
    },
    {
      "epoch": 5.783749200255918,
      "grad_norm": 0.973159357687972,
      "learning_rate": 7.961911631243253e-06,
      "loss": 0.077,
      "step": 9040
    },
    {
      "epoch": 5.784388995521433,
      "grad_norm": 0.936290209265184,
      "learning_rate": 7.959883014084608e-06,
      "loss": 0.0808,
      "step": 9041
    },
    {
      "epoch": 5.785028790786948,
      "grad_norm": 1.0304784066457298,
      "learning_rate": 7.957854484524963e-06,
      "loss": 0.0902,
      "step": 9042
    },
    {
      "epoch": 5.785668586052463,
      "grad_norm": 1.089250924468468,
      "learning_rate": 7.955826042651413e-06,
      "loss": 0.0913,
      "step": 9043
    },
    {
      "epoch": 5.786308381317978,
      "grad_norm": 1.084133526635982,
      "learning_rate": 7.953797688551053e-06,
      "loss": 0.0867,
      "step": 9044
    },
    {
      "epoch": 5.786948176583493,
      "grad_norm": 1.112563493668654,
      "learning_rate": 7.951769422310984e-06,
      "loss": 0.0875,
      "step": 9045
    },
    {
      "epoch": 5.787587971849009,
      "grad_norm": 0.9291193111508693,
      "learning_rate": 7.949741244018291e-06,
      "loss": 0.0845,
      "step": 9046
    },
    {
      "epoch": 5.788227767114523,
      "grad_norm": 0.9348017465899425,
      "learning_rate": 7.947713153760064e-06,
      "loss": 0.0732,
      "step": 9047
    },
    {
      "epoch": 5.788867562380038,
      "grad_norm": 0.9486032550821363,
      "learning_rate": 7.945685151623382e-06,
      "loss": 0.085,
      "step": 9048
    },
    {
      "epoch": 5.789507357645554,
      "grad_norm": 1.0023419110314045,
      "learning_rate": 7.943657237695325e-06,
      "loss": 0.0878,
      "step": 9049
    },
    {
      "epoch": 5.790147152911068,
      "grad_norm": 1.0105407584495667,
      "learning_rate": 7.941629412062968e-06,
      "loss": 0.0834,
      "step": 9050
    },
    {
      "epoch": 5.790786948176583,
      "grad_norm": 1.0619129403521907,
      "learning_rate": 7.93960167481338e-06,
      "loss": 0.0918,
      "step": 9051
    },
    {
      "epoch": 5.7914267434420985,
      "grad_norm": 0.906969935461116,
      "learning_rate": 7.937574026033636e-06,
      "loss": 0.0724,
      "step": 9052
    },
    {
      "epoch": 5.792066538707614,
      "grad_norm": 1.0512274462011288,
      "learning_rate": 7.935546465810793e-06,
      "loss": 0.0938,
      "step": 9053
    },
    {
      "epoch": 5.792706333973129,
      "grad_norm": 1.0586679407017037,
      "learning_rate": 7.93351899423191e-06,
      "loss": 0.0889,
      "step": 9054
    },
    {
      "epoch": 5.7933461292386434,
      "grad_norm": 0.9569676441156354,
      "learning_rate": 7.931491611384049e-06,
      "loss": 0.0817,
      "step": 9055
    },
    {
      "epoch": 5.793985924504159,
      "grad_norm": 1.0443784823762015,
      "learning_rate": 7.929464317354256e-06,
      "loss": 0.0899,
      "step": 9056
    },
    {
      "epoch": 5.794625719769674,
      "grad_norm": 0.9446957456523661,
      "learning_rate": 7.927437112229583e-06,
      "loss": 0.0821,
      "step": 9057
    },
    {
      "epoch": 5.795265515035188,
      "grad_norm": 1.0807377019958713,
      "learning_rate": 7.925409996097071e-06,
      "loss": 0.095,
      "step": 9058
    },
    {
      "epoch": 5.795905310300704,
      "grad_norm": 1.0909500894419057,
      "learning_rate": 7.923382969043766e-06,
      "loss": 0.0873,
      "step": 9059
    },
    {
      "epoch": 5.796545105566219,
      "grad_norm": 1.084488202702851,
      "learning_rate": 7.921356031156702e-06,
      "loss": 0.087,
      "step": 9060
    },
    {
      "epoch": 5.797184900831734,
      "grad_norm": 0.9990183972072237,
      "learning_rate": 7.919329182522913e-06,
      "loss": 0.0837,
      "step": 9061
    },
    {
      "epoch": 5.797824696097249,
      "grad_norm": 0.9959319758683277,
      "learning_rate": 7.917302423229428e-06,
      "loss": 0.0827,
      "step": 9062
    },
    {
      "epoch": 5.798464491362764,
      "grad_norm": 1.2455289579163016,
      "learning_rate": 7.91527575336327e-06,
      "loss": 0.0899,
      "step": 9063
    },
    {
      "epoch": 5.799104286628279,
      "grad_norm": 1.0118482689272803,
      "learning_rate": 7.913249173011461e-06,
      "loss": 0.0848,
      "step": 9064
    },
    {
      "epoch": 5.7997440818937935,
      "grad_norm": 1.007650761292732,
      "learning_rate": 7.911222682261024e-06,
      "loss": 0.0836,
      "step": 9065
    },
    {
      "epoch": 5.800383877159309,
      "grad_norm": 0.9590630178587439,
      "learning_rate": 7.90919628119897e-06,
      "loss": 0.0836,
      "step": 9066
    },
    {
      "epoch": 5.801023672424824,
      "grad_norm": 1.0902190139601424,
      "learning_rate": 7.907169969912308e-06,
      "loss": 0.0978,
      "step": 9067
    },
    {
      "epoch": 5.801663467690339,
      "grad_norm": 1.041456418458621,
      "learning_rate": 7.905143748488042e-06,
      "loss": 0.0898,
      "step": 9068
    },
    {
      "epoch": 5.802303262955854,
      "grad_norm": 1.0300548358021862,
      "learning_rate": 7.90311761701318e-06,
      "loss": 0.0882,
      "step": 9069
    },
    {
      "epoch": 5.802943058221369,
      "grad_norm": 0.9618146131378535,
      "learning_rate": 7.901091575574717e-06,
      "loss": 0.0911,
      "step": 9070
    },
    {
      "epoch": 5.803582853486884,
      "grad_norm": 1.1176161190294476,
      "learning_rate": 7.899065624259645e-06,
      "loss": 0.0957,
      "step": 9071
    },
    {
      "epoch": 5.8042226487524,
      "grad_norm": 1.0828329277743263,
      "learning_rate": 7.89703976315496e-06,
      "loss": 0.0883,
      "step": 9072
    },
    {
      "epoch": 5.804862444017914,
      "grad_norm": 1.0355566384785315,
      "learning_rate": 7.895013992347648e-06,
      "loss": 0.0889,
      "step": 9073
    },
    {
      "epoch": 5.805502239283429,
      "grad_norm": 0.9953579117309213,
      "learning_rate": 7.892988311924685e-06,
      "loss": 0.0831,
      "step": 9074
    },
    {
      "epoch": 5.8061420345489445,
      "grad_norm": 1.0632404749476463,
      "learning_rate": 7.890962721973061e-06,
      "loss": 0.0879,
      "step": 9075
    },
    {
      "epoch": 5.80678182981446,
      "grad_norm": 0.9294992818304694,
      "learning_rate": 7.888937222579746e-06,
      "loss": 0.0802,
      "step": 9076
    },
    {
      "epoch": 5.807421625079974,
      "grad_norm": 0.9721527923631496,
      "learning_rate": 7.886911813831713e-06,
      "loss": 0.0836,
      "step": 9077
    },
    {
      "epoch": 5.8080614203454894,
      "grad_norm": 0.9760123939075754,
      "learning_rate": 7.884886495815924e-06,
      "loss": 0.0871,
      "step": 9078
    },
    {
      "epoch": 5.808701215611005,
      "grad_norm": 0.9222887028295116,
      "learning_rate": 7.882861268619349e-06,
      "loss": 0.0807,
      "step": 9079
    },
    {
      "epoch": 5.809341010876519,
      "grad_norm": 0.9382830543901695,
      "learning_rate": 7.880836132328945e-06,
      "loss": 0.087,
      "step": 9080
    },
    {
      "epoch": 5.809980806142034,
      "grad_norm": 0.9374913386025034,
      "learning_rate": 7.878811087031663e-06,
      "loss": 0.0892,
      "step": 9081
    },
    {
      "epoch": 5.81062060140755,
      "grad_norm": 1.020814250018992,
      "learning_rate": 7.876786132814465e-06,
      "loss": 0.0806,
      "step": 9082
    },
    {
      "epoch": 5.811260396673065,
      "grad_norm": 1.042631895975429,
      "learning_rate": 7.874761269764291e-06,
      "loss": 0.0881,
      "step": 9083
    },
    {
      "epoch": 5.811900191938579,
      "grad_norm": 1.0430339890774791,
      "learning_rate": 7.872736497968088e-06,
      "loss": 0.0901,
      "step": 9084
    },
    {
      "epoch": 5.812539987204095,
      "grad_norm": 0.9380975822995175,
      "learning_rate": 7.870711817512798e-06,
      "loss": 0.0832,
      "step": 9085
    },
    {
      "epoch": 5.81317978246961,
      "grad_norm": 1.2108127392587542,
      "learning_rate": 7.868687228485353e-06,
      "loss": 0.0911,
      "step": 9086
    },
    {
      "epoch": 5.813819577735125,
      "grad_norm": 1.0282672228574585,
      "learning_rate": 7.866662730972687e-06,
      "loss": 0.0887,
      "step": 9087
    },
    {
      "epoch": 5.8144593730006395,
      "grad_norm": 1.0797967609749246,
      "learning_rate": 7.864638325061726e-06,
      "loss": 0.0865,
      "step": 9088
    },
    {
      "epoch": 5.815099168266155,
      "grad_norm": 0.945039138376266,
      "learning_rate": 7.8626140108394e-06,
      "loss": 0.083,
      "step": 9089
    },
    {
      "epoch": 5.81573896353167,
      "grad_norm": 0.9837153204569861,
      "learning_rate": 7.860589788392626e-06,
      "loss": 0.0832,
      "step": 9090
    },
    {
      "epoch": 5.816378758797185,
      "grad_norm": 1.0067354685344896,
      "learning_rate": 7.858565657808318e-06,
      "loss": 0.0902,
      "step": 9091
    },
    {
      "epoch": 5.8170185540627,
      "grad_norm": 1.0454654910903758,
      "learning_rate": 7.856541619173396e-06,
      "loss": 0.083,
      "step": 9092
    },
    {
      "epoch": 5.817658349328215,
      "grad_norm": 1.0380251578179693,
      "learning_rate": 7.854517672574762e-06,
      "loss": 0.0881,
      "step": 9093
    },
    {
      "epoch": 5.81829814459373,
      "grad_norm": 1.0566579954136832,
      "learning_rate": 7.852493818099319e-06,
      "loss": 0.0876,
      "step": 9094
    },
    {
      "epoch": 5.818937939859245,
      "grad_norm": 0.9242861220739692,
      "learning_rate": 7.850470055833976e-06,
      "loss": 0.0776,
      "step": 9095
    },
    {
      "epoch": 5.81957773512476,
      "grad_norm": 1.017126019563507,
      "learning_rate": 7.848446385865624e-06,
      "loss": 0.0882,
      "step": 9096
    },
    {
      "epoch": 5.820217530390275,
      "grad_norm": 1.0177627540738605,
      "learning_rate": 7.846422808281158e-06,
      "loss": 0.0836,
      "step": 9097
    },
    {
      "epoch": 5.8208573256557905,
      "grad_norm": 0.8997465112875803,
      "learning_rate": 7.844399323167462e-06,
      "loss": 0.0856,
      "step": 9098
    },
    {
      "epoch": 5.821497120921305,
      "grad_norm": 0.9347543598445786,
      "learning_rate": 7.842375930611428e-06,
      "loss": 0.0854,
      "step": 9099
    },
    {
      "epoch": 5.82213691618682,
      "grad_norm": 0.9099199712706516,
      "learning_rate": 7.840352630699934e-06,
      "loss": 0.0868,
      "step": 9100
    },
    {
      "epoch": 5.822776711452335,
      "grad_norm": 0.9164415859377459,
      "learning_rate": 7.83832942351985e-06,
      "loss": 0.0802,
      "step": 9101
    },
    {
      "epoch": 5.82341650671785,
      "grad_norm": 0.9533673919586744,
      "learning_rate": 7.836306309158062e-06,
      "loss": 0.0854,
      "step": 9102
    },
    {
      "epoch": 5.824056301983365,
      "grad_norm": 1.1915425583191084,
      "learning_rate": 7.83428328770143e-06,
      "loss": 0.0889,
      "step": 9103
    },
    {
      "epoch": 5.82469609724888,
      "grad_norm": 0.9700937749459863,
      "learning_rate": 7.832260359236817e-06,
      "loss": 0.0856,
      "step": 9104
    },
    {
      "epoch": 5.825335892514396,
      "grad_norm": 0.934384285600277,
      "learning_rate": 7.830237523851093e-06,
      "loss": 0.0857,
      "step": 9105
    },
    {
      "epoch": 5.825975687779911,
      "grad_norm": 1.0325877886305226,
      "learning_rate": 7.82821478163111e-06,
      "loss": 0.0848,
      "step": 9106
    },
    {
      "epoch": 5.826615483045425,
      "grad_norm": 1.0595925928307055,
      "learning_rate": 7.826192132663721e-06,
      "loss": 0.0935,
      "step": 9107
    },
    {
      "epoch": 5.827255278310941,
      "grad_norm": 0.9647461528603759,
      "learning_rate": 7.824169577035773e-06,
      "loss": 0.0836,
      "step": 9108
    },
    {
      "epoch": 5.827895073576456,
      "grad_norm": 0.9972521863699826,
      "learning_rate": 7.822147114834116e-06,
      "loss": 0.0837,
      "step": 9109
    },
    {
      "epoch": 5.82853486884197,
      "grad_norm": 0.8921072558878633,
      "learning_rate": 7.820124746145586e-06,
      "loss": 0.0804,
      "step": 9110
    },
    {
      "epoch": 5.8291746641074855,
      "grad_norm": 1.0486201519376246,
      "learning_rate": 7.818102471057017e-06,
      "loss": 0.0856,
      "step": 9111
    },
    {
      "epoch": 5.829814459373001,
      "grad_norm": 0.9797130687085,
      "learning_rate": 7.816080289655252e-06,
      "loss": 0.0901,
      "step": 9112
    },
    {
      "epoch": 5.830454254638516,
      "grad_norm": 0.9720178088400606,
      "learning_rate": 7.814058202027114e-06,
      "loss": 0.0851,
      "step": 9113
    },
    {
      "epoch": 5.8310940499040305,
      "grad_norm": 1.1056217878212453,
      "learning_rate": 7.81203620825943e-06,
      "loss": 0.0898,
      "step": 9114
    },
    {
      "epoch": 5.831733845169546,
      "grad_norm": 0.978568832900423,
      "learning_rate": 7.810014308439014e-06,
      "loss": 0.0835,
      "step": 9115
    },
    {
      "epoch": 5.832373640435061,
      "grad_norm": 0.9931757990780649,
      "learning_rate": 7.807992502652692e-06,
      "loss": 0.0886,
      "step": 9116
    },
    {
      "epoch": 5.833013435700575,
      "grad_norm": 1.100893203392811,
      "learning_rate": 7.805970790987273e-06,
      "loss": 0.0909,
      "step": 9117
    },
    {
      "epoch": 5.833653230966091,
      "grad_norm": 1.0732018920189397,
      "learning_rate": 7.80394917352956e-06,
      "loss": 0.0898,
      "step": 9118
    },
    {
      "epoch": 5.834293026231606,
      "grad_norm": 0.9492005340837287,
      "learning_rate": 7.801927650366367e-06,
      "loss": 0.0717,
      "step": 9119
    },
    {
      "epoch": 5.834932821497121,
      "grad_norm": 1.0360439755390307,
      "learning_rate": 7.799906221584493e-06,
      "loss": 0.0828,
      "step": 9120
    },
    {
      "epoch": 5.835572616762636,
      "grad_norm": 0.926390228783318,
      "learning_rate": 7.797884887270727e-06,
      "loss": 0.0783,
      "step": 9121
    },
    {
      "epoch": 5.836212412028151,
      "grad_norm": 1.0190706605627136,
      "learning_rate": 7.795863647511871e-06,
      "loss": 0.0906,
      "step": 9122
    },
    {
      "epoch": 5.836852207293666,
      "grad_norm": 1.0267286125660369,
      "learning_rate": 7.793842502394709e-06,
      "loss": 0.0858,
      "step": 9123
    },
    {
      "epoch": 5.837492002559181,
      "grad_norm": 0.8273637605647544,
      "learning_rate": 7.791821452006025e-06,
      "loss": 0.0778,
      "step": 9124
    },
    {
      "epoch": 5.838131797824696,
      "grad_norm": 1.0717107457183523,
      "learning_rate": 7.789800496432595e-06,
      "loss": 0.0865,
      "step": 9125
    },
    {
      "epoch": 5.838771593090211,
      "grad_norm": 0.9865697326081255,
      "learning_rate": 7.787779635761203e-06,
      "loss": 0.0854,
      "step": 9126
    },
    {
      "epoch": 5.839411388355726,
      "grad_norm": 0.9824192887165715,
      "learning_rate": 7.785758870078621e-06,
      "loss": 0.0805,
      "step": 9127
    },
    {
      "epoch": 5.840051183621242,
      "grad_norm": 1.0949594828030231,
      "learning_rate": 7.783738199471611e-06,
      "loss": 0.0879,
      "step": 9128
    },
    {
      "epoch": 5.840690978886756,
      "grad_norm": 0.9316153532727102,
      "learning_rate": 7.781717624026942e-06,
      "loss": 0.0803,
      "step": 9129
    },
    {
      "epoch": 5.841330774152271,
      "grad_norm": 1.068432929017773,
      "learning_rate": 7.779697143831373e-06,
      "loss": 0.0895,
      "step": 9130
    },
    {
      "epoch": 5.841970569417787,
      "grad_norm": 0.9913574266526807,
      "learning_rate": 7.777676758971654e-06,
      "loss": 0.0872,
      "step": 9131
    },
    {
      "epoch": 5.842610364683301,
      "grad_norm": 1.120024308289726,
      "learning_rate": 7.775656469534548e-06,
      "loss": 0.101,
      "step": 9132
    },
    {
      "epoch": 5.843250159948816,
      "grad_norm": 1.089142228744665,
      "learning_rate": 7.773636275606794e-06,
      "loss": 0.0877,
      "step": 9133
    },
    {
      "epoch": 5.8438899552143315,
      "grad_norm": 1.033343340695028,
      "learning_rate": 7.771616177275138e-06,
      "loss": 0.0873,
      "step": 9134
    },
    {
      "epoch": 5.844529750479847,
      "grad_norm": 1.1007064698084592,
      "learning_rate": 7.769596174626316e-06,
      "loss": 0.0923,
      "step": 9135
    },
    {
      "epoch": 5.845169545745361,
      "grad_norm": 0.9916919505528706,
      "learning_rate": 7.767576267747072e-06,
      "loss": 0.0799,
      "step": 9136
    },
    {
      "epoch": 5.8458093410108765,
      "grad_norm": 1.0367791187676214,
      "learning_rate": 7.76555645672413e-06,
      "loss": 0.0872,
      "step": 9137
    },
    {
      "epoch": 5.846449136276392,
      "grad_norm": 1.119307918999975,
      "learning_rate": 7.763536741644218e-06,
      "loss": 0.0896,
      "step": 9138
    },
    {
      "epoch": 5.847088931541906,
      "grad_norm": 1.0755157657722731,
      "learning_rate": 7.761517122594062e-06,
      "loss": 0.0936,
      "step": 9139
    },
    {
      "epoch": 5.847728726807421,
      "grad_norm": 1.0079249598433475,
      "learning_rate": 7.759497599660378e-06,
      "loss": 0.0849,
      "step": 9140
    },
    {
      "epoch": 5.848368522072937,
      "grad_norm": 1.105137041424428,
      "learning_rate": 7.757478172929877e-06,
      "loss": 0.0882,
      "step": 9141
    },
    {
      "epoch": 5.849008317338452,
      "grad_norm": 0.9448835781095447,
      "learning_rate": 7.75545884248928e-06,
      "loss": 0.0835,
      "step": 9142
    },
    {
      "epoch": 5.849648112603967,
      "grad_norm": 1.0826516061820455,
      "learning_rate": 7.753439608425287e-06,
      "loss": 0.0962,
      "step": 9143
    },
    {
      "epoch": 5.850287907869482,
      "grad_norm": 0.9951961494628575,
      "learning_rate": 7.7514204708246e-06,
      "loss": 0.0849,
      "step": 9144
    },
    {
      "epoch": 5.850927703134997,
      "grad_norm": 0.9397840698276682,
      "learning_rate": 7.749401429773916e-06,
      "loss": 0.0865,
      "step": 9145
    },
    {
      "epoch": 5.851567498400512,
      "grad_norm": 1.0222802169482135,
      "learning_rate": 7.747382485359934e-06,
      "loss": 0.0933,
      "step": 9146
    },
    {
      "epoch": 5.8522072936660265,
      "grad_norm": 1.096198879649337,
      "learning_rate": 7.745363637669341e-06,
      "loss": 0.0875,
      "step": 9147
    },
    {
      "epoch": 5.852847088931542,
      "grad_norm": 1.0172742007131508,
      "learning_rate": 7.743344886788818e-06,
      "loss": 0.0826,
      "step": 9148
    },
    {
      "epoch": 5.853486884197057,
      "grad_norm": 0.9631535836698034,
      "learning_rate": 7.741326232805055e-06,
      "loss": 0.0763,
      "step": 9149
    },
    {
      "epoch": 5.854126679462572,
      "grad_norm": 1.169745591326993,
      "learning_rate": 7.739307675804727e-06,
      "loss": 0.0958,
      "step": 9150
    },
    {
      "epoch": 5.854766474728087,
      "grad_norm": 0.9605269308854351,
      "learning_rate": 7.737289215874503e-06,
      "loss": 0.0787,
      "step": 9151
    },
    {
      "epoch": 5.855406269993602,
      "grad_norm": 1.0743718756697065,
      "learning_rate": 7.735270853101057e-06,
      "loss": 0.0872,
      "step": 9152
    },
    {
      "epoch": 5.856046065259117,
      "grad_norm": 0.948605728754003,
      "learning_rate": 7.733252587571053e-06,
      "loss": 0.0801,
      "step": 9153
    },
    {
      "epoch": 5.856685860524632,
      "grad_norm": 1.0798935018595759,
      "learning_rate": 7.73123441937115e-06,
      "loss": 0.0888,
      "step": 9154
    },
    {
      "epoch": 5.857325655790147,
      "grad_norm": 1.1657565021328895,
      "learning_rate": 7.729216348588e-06,
      "loss": 0.0912,
      "step": 9155
    },
    {
      "epoch": 5.857965451055662,
      "grad_norm": 0.9579214492411161,
      "learning_rate": 7.727198375308265e-06,
      "loss": 0.0834,
      "step": 9156
    },
    {
      "epoch": 5.8586052463211775,
      "grad_norm": 1.0330097439543044,
      "learning_rate": 7.725180499618591e-06,
      "loss": 0.0877,
      "step": 9157
    },
    {
      "epoch": 5.859245041586692,
      "grad_norm": 1.0351394952927533,
      "learning_rate": 7.723162721605616e-06,
      "loss": 0.0905,
      "step": 9158
    },
    {
      "epoch": 5.859884836852207,
      "grad_norm": 0.9270374327907238,
      "learning_rate": 7.721145041355985e-06,
      "loss": 0.086,
      "step": 9159
    },
    {
      "epoch": 5.8605246321177225,
      "grad_norm": 1.0953448264017187,
      "learning_rate": 7.719127458956333e-06,
      "loss": 0.0901,
      "step": 9160
    },
    {
      "epoch": 5.861164427383238,
      "grad_norm": 1.0169259147400431,
      "learning_rate": 7.717109974493287e-06,
      "loss": 0.0804,
      "step": 9161
    },
    {
      "epoch": 5.861804222648752,
      "grad_norm": 0.9966724563561047,
      "learning_rate": 7.71509258805348e-06,
      "loss": 0.0865,
      "step": 9162
    },
    {
      "epoch": 5.862444017914267,
      "grad_norm": 0.969597281835204,
      "learning_rate": 7.713075299723534e-06,
      "loss": 0.0861,
      "step": 9163
    },
    {
      "epoch": 5.863083813179783,
      "grad_norm": 1.0286166309244766,
      "learning_rate": 7.711058109590068e-06,
      "loss": 0.0846,
      "step": 9164
    },
    {
      "epoch": 5.863723608445298,
      "grad_norm": 0.9202649785934607,
      "learning_rate": 7.70904101773969e-06,
      "loss": 0.0865,
      "step": 9165
    },
    {
      "epoch": 5.864363403710812,
      "grad_norm": 1.1147127225501576,
      "learning_rate": 7.707024024259018e-06,
      "loss": 0.0842,
      "step": 9166
    },
    {
      "epoch": 5.865003198976328,
      "grad_norm": 0.8250649568190301,
      "learning_rate": 7.705007129234657e-06,
      "loss": 0.0801,
      "step": 9167
    },
    {
      "epoch": 5.865642994241843,
      "grad_norm": 1.1080917883557617,
      "learning_rate": 7.702990332753204e-06,
      "loss": 0.0927,
      "step": 9168
    },
    {
      "epoch": 5.866282789507357,
      "grad_norm": 1.1867255307583822,
      "learning_rate": 7.700973634901263e-06,
      "loss": 0.092,
      "step": 9169
    },
    {
      "epoch": 5.8669225847728725,
      "grad_norm": 1.0874433982143177,
      "learning_rate": 7.698957035765423e-06,
      "loss": 0.0933,
      "step": 9170
    },
    {
      "epoch": 5.867562380038388,
      "grad_norm": 0.9584348696053239,
      "learning_rate": 7.696940535432274e-06,
      "loss": 0.0771,
      "step": 9171
    },
    {
      "epoch": 5.868202175303903,
      "grad_norm": 0.9364218099698558,
      "learning_rate": 7.694924133988404e-06,
      "loss": 0.0854,
      "step": 9172
    },
    {
      "epoch": 5.8688419705694175,
      "grad_norm": 1.0650509975705023,
      "learning_rate": 7.69290783152039e-06,
      "loss": 0.0871,
      "step": 9173
    },
    {
      "epoch": 5.869481765834933,
      "grad_norm": 0.9824644747984417,
      "learning_rate": 7.69089162811481e-06,
      "loss": 0.0775,
      "step": 9174
    },
    {
      "epoch": 5.870121561100448,
      "grad_norm": 0.9745757286084132,
      "learning_rate": 7.688875523858235e-06,
      "loss": 0.0847,
      "step": 9175
    },
    {
      "epoch": 5.870761356365962,
      "grad_norm": 0.9470599915034512,
      "learning_rate": 7.686859518837235e-06,
      "loss": 0.0775,
      "step": 9176
    },
    {
      "epoch": 5.871401151631478,
      "grad_norm": 1.0822728814707878,
      "learning_rate": 7.684843613138372e-06,
      "loss": 0.1001,
      "step": 9177
    },
    {
      "epoch": 5.872040946896993,
      "grad_norm": 1.005906202772125,
      "learning_rate": 7.682827806848203e-06,
      "loss": 0.0848,
      "step": 9178
    },
    {
      "epoch": 5.872680742162508,
      "grad_norm": 0.9232751447700078,
      "learning_rate": 7.680812100053289e-06,
      "loss": 0.0887,
      "step": 9179
    },
    {
      "epoch": 5.8733205374280235,
      "grad_norm": 0.9618339264681282,
      "learning_rate": 7.67879649284018e-06,
      "loss": 0.0792,
      "step": 9180
    },
    {
      "epoch": 5.873960332693538,
      "grad_norm": 0.932930363363953,
      "learning_rate": 7.676780985295417e-06,
      "loss": 0.0814,
      "step": 9181
    },
    {
      "epoch": 5.874600127959053,
      "grad_norm": 1.052200664706974,
      "learning_rate": 7.674765577505548e-06,
      "loss": 0.0877,
      "step": 9182
    },
    {
      "epoch": 5.8752399232245685,
      "grad_norm": 1.1227081210848138,
      "learning_rate": 7.67275026955711e-06,
      "loss": 0.0933,
      "step": 9183
    },
    {
      "epoch": 5.875879718490083,
      "grad_norm": 0.9971130881315232,
      "learning_rate": 7.670735061536634e-06,
      "loss": 0.0863,
      "step": 9184
    },
    {
      "epoch": 5.876519513755598,
      "grad_norm": 0.9783728390941693,
      "learning_rate": 7.668719953530647e-06,
      "loss": 0.0842,
      "step": 9185
    },
    {
      "epoch": 5.877159309021113,
      "grad_norm": 1.041054378655578,
      "learning_rate": 7.666704945625683e-06,
      "loss": 0.0948,
      "step": 9186
    },
    {
      "epoch": 5.877799104286629,
      "grad_norm": 0.9595323287546195,
      "learning_rate": 7.664690037908257e-06,
      "loss": 0.0859,
      "step": 9187
    },
    {
      "epoch": 5.878438899552143,
      "grad_norm": 0.9169793799896544,
      "learning_rate": 7.662675230464885e-06,
      "loss": 0.0816,
      "step": 9188
    },
    {
      "epoch": 5.879078694817658,
      "grad_norm": 1.1217986095433663,
      "learning_rate": 7.660660523382083e-06,
      "loss": 0.0958,
      "step": 9189
    },
    {
      "epoch": 5.879718490083174,
      "grad_norm": 0.9243327142823903,
      "learning_rate": 7.658645916746357e-06,
      "loss": 0.076,
      "step": 9190
    },
    {
      "epoch": 5.880358285348688,
      "grad_norm": 0.9736754136359137,
      "learning_rate": 7.656631410644209e-06,
      "loss": 0.0867,
      "step": 9191
    },
    {
      "epoch": 5.880998080614203,
      "grad_norm": 1.0361521590651435,
      "learning_rate": 7.654617005162137e-06,
      "loss": 0.0878,
      "step": 9192
    },
    {
      "epoch": 5.8816378758797185,
      "grad_norm": 0.8824619589251769,
      "learning_rate": 7.652602700386644e-06,
      "loss": 0.0789,
      "step": 9193
    },
    {
      "epoch": 5.882277671145234,
      "grad_norm": 0.9643429827872129,
      "learning_rate": 7.650588496404212e-06,
      "loss": 0.0829,
      "step": 9194
    },
    {
      "epoch": 5.882917466410748,
      "grad_norm": 1.0465410351789541,
      "learning_rate": 7.648574393301328e-06,
      "loss": 0.0843,
      "step": 9195
    },
    {
      "epoch": 5.8835572616762635,
      "grad_norm": 1.0340739559058774,
      "learning_rate": 7.646560391164483e-06,
      "loss": 0.0896,
      "step": 9196
    },
    {
      "epoch": 5.884197056941779,
      "grad_norm": 0.978182753483474,
      "learning_rate": 7.644546490080146e-06,
      "loss": 0.0796,
      "step": 9197
    },
    {
      "epoch": 5.884836852207294,
      "grad_norm": 1.1129831468024518,
      "learning_rate": 7.642532690134792e-06,
      "loss": 0.0899,
      "step": 9198
    },
    {
      "epoch": 5.885476647472808,
      "grad_norm": 1.049693733775827,
      "learning_rate": 7.640518991414892e-06,
      "loss": 0.084,
      "step": 9199
    },
    {
      "epoch": 5.886116442738324,
      "grad_norm": 0.983955272006945,
      "learning_rate": 7.638505394006908e-06,
      "loss": 0.0819,
      "step": 9200
    },
    {
      "epoch": 5.886756238003839,
      "grad_norm": 1.0269531954440523,
      "learning_rate": 7.636491897997303e-06,
      "loss": 0.0845,
      "step": 9201
    },
    {
      "epoch": 5.887396033269354,
      "grad_norm": 0.9949875381585837,
      "learning_rate": 7.634478503472527e-06,
      "loss": 0.0826,
      "step": 9202
    },
    {
      "epoch": 5.888035828534869,
      "grad_norm": 0.9080942847902083,
      "learning_rate": 7.63246521051904e-06,
      "loss": 0.0763,
      "step": 9203
    },
    {
      "epoch": 5.888675623800384,
      "grad_norm": 0.9703327527641262,
      "learning_rate": 7.630452019223289e-06,
      "loss": 0.081,
      "step": 9204
    },
    {
      "epoch": 5.889315419065899,
      "grad_norm": 0.9998164166334002,
      "learning_rate": 7.628438929671706e-06,
      "loss": 0.0906,
      "step": 9205
    },
    {
      "epoch": 5.889955214331414,
      "grad_norm": 1.0594461551568737,
      "learning_rate": 7.6264259419507415e-06,
      "loss": 0.0821,
      "step": 9206
    },
    {
      "epoch": 5.890595009596929,
      "grad_norm": 1.0623707175512995,
      "learning_rate": 7.624413056146824e-06,
      "loss": 0.0931,
      "step": 9207
    },
    {
      "epoch": 5.891234804862444,
      "grad_norm": 0.8215726728309088,
      "learning_rate": 7.62240027234638e-06,
      "loss": 0.0736,
      "step": 9208
    },
    {
      "epoch": 5.891874600127959,
      "grad_norm": 1.0016749929227602,
      "learning_rate": 7.620387590635843e-06,
      "loss": 0.0934,
      "step": 9209
    },
    {
      "epoch": 5.892514395393474,
      "grad_norm": 0.9079518106154081,
      "learning_rate": 7.618375011101629e-06,
      "loss": 0.0786,
      "step": 9210
    },
    {
      "epoch": 5.893154190658989,
      "grad_norm": 0.9815207538796631,
      "learning_rate": 7.6163625338301575e-06,
      "loss": 0.0844,
      "step": 9211
    },
    {
      "epoch": 5.893793985924504,
      "grad_norm": 0.9125776591380029,
      "learning_rate": 7.614350158907835e-06,
      "loss": 0.0785,
      "step": 9212
    },
    {
      "epoch": 5.894433781190019,
      "grad_norm": 0.9551315356949341,
      "learning_rate": 7.612337886421075e-06,
      "loss": 0.0761,
      "step": 9213
    },
    {
      "epoch": 5.895073576455534,
      "grad_norm": 0.9841557857513763,
      "learning_rate": 7.610325716456279e-06,
      "loss": 0.0871,
      "step": 9214
    },
    {
      "epoch": 5.895713371721049,
      "grad_norm": 0.9969146131313205,
      "learning_rate": 7.608313649099842e-06,
      "loss": 0.0884,
      "step": 9215
    },
    {
      "epoch": 5.8963531669865645,
      "grad_norm": 1.1174572577755402,
      "learning_rate": 7.606301684438167e-06,
      "loss": 0.0894,
      "step": 9216
    },
    {
      "epoch": 5.89699296225208,
      "grad_norm": 1.0468167120208591,
      "learning_rate": 7.604289822557639e-06,
      "loss": 0.0885,
      "step": 9217
    },
    {
      "epoch": 5.897632757517594,
      "grad_norm": 1.079414628930873,
      "learning_rate": 7.602278063544643e-06,
      "loss": 0.0844,
      "step": 9218
    },
    {
      "epoch": 5.8982725527831095,
      "grad_norm": 1.0259822740174864,
      "learning_rate": 7.600266407485562e-06,
      "loss": 0.0842,
      "step": 9219
    },
    {
      "epoch": 5.898912348048625,
      "grad_norm": 0.94806562718201,
      "learning_rate": 7.598254854466774e-06,
      "loss": 0.0816,
      "step": 9220
    },
    {
      "epoch": 5.899552143314139,
      "grad_norm": 1.0243770398949512,
      "learning_rate": 7.5962434045746506e-06,
      "loss": 0.079,
      "step": 9221
    },
    {
      "epoch": 5.900191938579654,
      "grad_norm": 0.9868003247133371,
      "learning_rate": 7.594232057895554e-06,
      "loss": 0.0802,
      "step": 9222
    },
    {
      "epoch": 5.90083173384517,
      "grad_norm": 1.0775745892243307,
      "learning_rate": 7.5922208145158586e-06,
      "loss": 0.0955,
      "step": 9223
    },
    {
      "epoch": 5.901471529110685,
      "grad_norm": 0.9380302466634297,
      "learning_rate": 7.590209674521918e-06,
      "loss": 0.0831,
      "step": 9224
    },
    {
      "epoch": 5.902111324376199,
      "grad_norm": 1.1090409954296496,
      "learning_rate": 7.588198638000084e-06,
      "loss": 0.0888,
      "step": 9225
    },
    {
      "epoch": 5.902751119641715,
      "grad_norm": 0.9492424591522033,
      "learning_rate": 7.586187705036713e-06,
      "loss": 0.0827,
      "step": 9226
    },
    {
      "epoch": 5.90339091490723,
      "grad_norm": 1.0313979655291476,
      "learning_rate": 7.584176875718148e-06,
      "loss": 0.0933,
      "step": 9227
    },
    {
      "epoch": 5.904030710172744,
      "grad_norm": 0.961033241872888,
      "learning_rate": 7.58216615013073e-06,
      "loss": 0.0864,
      "step": 9228
    },
    {
      "epoch": 5.9046705054382596,
      "grad_norm": 1.053931299058054,
      "learning_rate": 7.580155528360797e-06,
      "loss": 0.0894,
      "step": 9229
    },
    {
      "epoch": 5.905310300703775,
      "grad_norm": 1.085012720580511,
      "learning_rate": 7.5781450104946816e-06,
      "loss": 0.0944,
      "step": 9230
    },
    {
      "epoch": 5.90595009596929,
      "grad_norm": 1.0342756047376422,
      "learning_rate": 7.576134596618712e-06,
      "loss": 0.0846,
      "step": 9231
    },
    {
      "epoch": 5.9065898912348045,
      "grad_norm": 1.0370210493478558,
      "learning_rate": 7.574124286819206e-06,
      "loss": 0.0941,
      "step": 9232
    },
    {
      "epoch": 5.90722968650032,
      "grad_norm": 0.9636366865034263,
      "learning_rate": 7.572114081182491e-06,
      "loss": 0.0862,
      "step": 9233
    },
    {
      "epoch": 5.907869481765835,
      "grad_norm": 1.0191495933255021,
      "learning_rate": 7.57010397979488e-06,
      "loss": 0.0864,
      "step": 9234
    },
    {
      "epoch": 5.90850927703135,
      "grad_norm": 1.0103082639488397,
      "learning_rate": 7.56809398274268e-06,
      "loss": 0.0856,
      "step": 9235
    },
    {
      "epoch": 5.909149072296865,
      "grad_norm": 1.0184645554485088,
      "learning_rate": 7.5660840901122005e-06,
      "loss": 0.0852,
      "step": 9236
    },
    {
      "epoch": 5.90978886756238,
      "grad_norm": 0.9249045925037139,
      "learning_rate": 7.5640743019897385e-06,
      "loss": 0.0801,
      "step": 9237
    },
    {
      "epoch": 5.910428662827895,
      "grad_norm": 0.949709954479771,
      "learning_rate": 7.562064618461591e-06,
      "loss": 0.0833,
      "step": 9238
    },
    {
      "epoch": 5.9110684580934105,
      "grad_norm": 0.9597255703323321,
      "learning_rate": 7.560055039614055e-06,
      "loss": 0.0864,
      "step": 9239
    },
    {
      "epoch": 5.911708253358925,
      "grad_norm": 0.9459062312529223,
      "learning_rate": 7.558045565533415e-06,
      "loss": 0.0857,
      "step": 9240
    },
    {
      "epoch": 5.91234804862444,
      "grad_norm": 1.005751847444791,
      "learning_rate": 7.556036196305955e-06,
      "loss": 0.0831,
      "step": 9241
    },
    {
      "epoch": 5.9129878438899555,
      "grad_norm": 0.9147711686215737,
      "learning_rate": 7.554026932017952e-06,
      "loss": 0.0807,
      "step": 9242
    },
    {
      "epoch": 5.91362763915547,
      "grad_norm": 1.0022640653476542,
      "learning_rate": 7.552017772755682e-06,
      "loss": 0.0905,
      "step": 9243
    },
    {
      "epoch": 5.914267434420985,
      "grad_norm": 0.9251679913975595,
      "learning_rate": 7.550008718605415e-06,
      "loss": 0.0809,
      "step": 9244
    },
    {
      "epoch": 5.9149072296865,
      "grad_norm": 1.0262230169249362,
      "learning_rate": 7.547999769653411e-06,
      "loss": 0.0936,
      "step": 9245
    },
    {
      "epoch": 5.915547024952016,
      "grad_norm": 0.9575161193115519,
      "learning_rate": 7.54599092598594e-06,
      "loss": 0.0874,
      "step": 9246
    },
    {
      "epoch": 5.91618682021753,
      "grad_norm": 0.9928920517628547,
      "learning_rate": 7.543982187689253e-06,
      "loss": 0.0827,
      "step": 9247
    },
    {
      "epoch": 5.916826615483045,
      "grad_norm": 1.0624802183873143,
      "learning_rate": 7.541973554849599e-06,
      "loss": 0.0881,
      "step": 9248
    },
    {
      "epoch": 5.917466410748561,
      "grad_norm": 0.9775612787433069,
      "learning_rate": 7.539965027553231e-06,
      "loss": 0.0843,
      "step": 9249
    },
    {
      "epoch": 5.918106206014075,
      "grad_norm": 1.006295194994921,
      "learning_rate": 7.537956605886388e-06,
      "loss": 0.0861,
      "step": 9250
    },
    {
      "epoch": 5.91874600127959,
      "grad_norm": 0.9559311979511121,
      "learning_rate": 7.5359482899353085e-06,
      "loss": 0.0959,
      "step": 9251
    },
    {
      "epoch": 5.9193857965451055,
      "grad_norm": 0.8694622859441282,
      "learning_rate": 7.533940079786224e-06,
      "loss": 0.0756,
      "step": 9252
    },
    {
      "epoch": 5.920025591810621,
      "grad_norm": 0.9124032126502155,
      "learning_rate": 7.531931975525368e-06,
      "loss": 0.0815,
      "step": 9253
    },
    {
      "epoch": 5.920665387076136,
      "grad_norm": 1.071294920223071,
      "learning_rate": 7.529923977238963e-06,
      "loss": 0.0854,
      "step": 9254
    },
    {
      "epoch": 5.9213051823416505,
      "grad_norm": 1.1734016355082937,
      "learning_rate": 7.527916085013225e-06,
      "loss": 0.0923,
      "step": 9255
    },
    {
      "epoch": 5.921944977607166,
      "grad_norm": 1.021242125081504,
      "learning_rate": 7.525908298934377e-06,
      "loss": 0.0856,
      "step": 9256
    },
    {
      "epoch": 5.922584772872681,
      "grad_norm": 1.0960313328986877,
      "learning_rate": 7.523900619088624e-06,
      "loss": 0.089,
      "step": 9257
    },
    {
      "epoch": 5.923224568138195,
      "grad_norm": 1.1437405944396886,
      "learning_rate": 7.521893045562173e-06,
      "loss": 0.088,
      "step": 9258
    },
    {
      "epoch": 5.923864363403711,
      "grad_norm": 1.0369569950771131,
      "learning_rate": 7.5198855784412285e-06,
      "loss": 0.0839,
      "step": 9259
    },
    {
      "epoch": 5.924504158669226,
      "grad_norm": 1.0432766385359107,
      "learning_rate": 7.517878217811985e-06,
      "loss": 0.0855,
      "step": 9260
    },
    {
      "epoch": 5.925143953934741,
      "grad_norm": 0.8856999967092407,
      "learning_rate": 7.5158709637606366e-06,
      "loss": 0.0793,
      "step": 9261
    },
    {
      "epoch": 5.925783749200256,
      "grad_norm": 0.9059259037804547,
      "learning_rate": 7.513863816373364e-06,
      "loss": 0.0931,
      "step": 9262
    },
    {
      "epoch": 5.926423544465771,
      "grad_norm": 1.0772981116991815,
      "learning_rate": 7.511856775736362e-06,
      "loss": 0.0868,
      "step": 9263
    },
    {
      "epoch": 5.927063339731286,
      "grad_norm": 0.9925821303851502,
      "learning_rate": 7.509849841935803e-06,
      "loss": 0.0861,
      "step": 9264
    },
    {
      "epoch": 5.927703134996801,
      "grad_norm": 0.9736853587124393,
      "learning_rate": 7.507843015057862e-06,
      "loss": 0.0831,
      "step": 9265
    },
    {
      "epoch": 5.928342930262316,
      "grad_norm": 1.0148907636660007,
      "learning_rate": 7.505836295188709e-06,
      "loss": 0.083,
      "step": 9266
    },
    {
      "epoch": 5.928982725527831,
      "grad_norm": 1.1297448595963795,
      "learning_rate": 7.50382968241451e-06,
      "loss": 0.0918,
      "step": 9267
    },
    {
      "epoch": 5.929622520793346,
      "grad_norm": 1.0705392758363521,
      "learning_rate": 7.501823176821423e-06,
      "loss": 0.0892,
      "step": 9268
    },
    {
      "epoch": 5.930262316058861,
      "grad_norm": 0.9955415688721716,
      "learning_rate": 7.499816778495601e-06,
      "loss": 0.0861,
      "step": 9269
    },
    {
      "epoch": 5.930902111324376,
      "grad_norm": 1.0176406526680433,
      "learning_rate": 7.497810487523204e-06,
      "loss": 0.0852,
      "step": 9270
    },
    {
      "epoch": 5.931541906589891,
      "grad_norm": 0.8036874571741875,
      "learning_rate": 7.4958043039903724e-06,
      "loss": 0.074,
      "step": 9271
    },
    {
      "epoch": 5.932181701855407,
      "grad_norm": 0.9960730737037531,
      "learning_rate": 7.493798227983249e-06,
      "loss": 0.0904,
      "step": 9272
    },
    {
      "epoch": 5.932821497120921,
      "grad_norm": 0.9401293488153716,
      "learning_rate": 7.491792259587972e-06,
      "loss": 0.0782,
      "step": 9273
    },
    {
      "epoch": 5.933461292386436,
      "grad_norm": 1.0754369905822512,
      "learning_rate": 7.489786398890675e-06,
      "loss": 0.0967,
      "step": 9274
    },
    {
      "epoch": 5.9341010876519515,
      "grad_norm": 0.8712134212204576,
      "learning_rate": 7.487780645977478e-06,
      "loss": 0.0869,
      "step": 9275
    },
    {
      "epoch": 5.934740882917467,
      "grad_norm": 0.9067971931156774,
      "learning_rate": 7.485775000934517e-06,
      "loss": 0.0832,
      "step": 9276
    },
    {
      "epoch": 5.935380678182981,
      "grad_norm": 1.0123268487127755,
      "learning_rate": 7.4837694638479055e-06,
      "loss": 0.0846,
      "step": 9277
    },
    {
      "epoch": 5.9360204734484965,
      "grad_norm": 0.9622598977568525,
      "learning_rate": 7.481764034803757e-06,
      "loss": 0.0894,
      "step": 9278
    },
    {
      "epoch": 5.936660268714012,
      "grad_norm": 1.0681486287734971,
      "learning_rate": 7.479758713888179e-06,
      "loss": 0.0902,
      "step": 9279
    },
    {
      "epoch": 5.937300063979526,
      "grad_norm": 1.0469253394190439,
      "learning_rate": 7.47775350118728e-06,
      "loss": 0.0876,
      "step": 9280
    },
    {
      "epoch": 5.937939859245041,
      "grad_norm": 0.9233067684933295,
      "learning_rate": 7.4757483967871605e-06,
      "loss": 0.0837,
      "step": 9281
    },
    {
      "epoch": 5.938579654510557,
      "grad_norm": 0.951597559623073,
      "learning_rate": 7.4737434007739094e-06,
      "loss": 0.0814,
      "step": 9282
    },
    {
      "epoch": 5.939219449776072,
      "grad_norm": 1.054682267837576,
      "learning_rate": 7.471738513233626e-06,
      "loss": 0.0923,
      "step": 9283
    },
    {
      "epoch": 5.939859245041586,
      "grad_norm": 0.9322315556338212,
      "learning_rate": 7.469733734252396e-06,
      "loss": 0.084,
      "step": 9284
    },
    {
      "epoch": 5.940499040307102,
      "grad_norm": 1.0796261992561795,
      "learning_rate": 7.467729063916291e-06,
      "loss": 0.0849,
      "step": 9285
    },
    {
      "epoch": 5.941138835572617,
      "grad_norm": 1.0072589575862165,
      "learning_rate": 7.465724502311402e-06,
      "loss": 0.0802,
      "step": 9286
    },
    {
      "epoch": 5.941778630838132,
      "grad_norm": 0.9478587246494674,
      "learning_rate": 7.463720049523795e-06,
      "loss": 0.0825,
      "step": 9287
    },
    {
      "epoch": 5.942418426103647,
      "grad_norm": 0.9755309650516685,
      "learning_rate": 7.461715705639536e-06,
      "loss": 0.0858,
      "step": 9288
    },
    {
      "epoch": 5.943058221369162,
      "grad_norm": 0.9406840966093701,
      "learning_rate": 7.459711470744687e-06,
      "loss": 0.0817,
      "step": 9289
    },
    {
      "epoch": 5.943698016634677,
      "grad_norm": 1.0625396626360115,
      "learning_rate": 7.457707344925312e-06,
      "loss": 0.0756,
      "step": 9290
    },
    {
      "epoch": 5.944337811900192,
      "grad_norm": 1.0259030664961253,
      "learning_rate": 7.455703328267461e-06,
      "loss": 0.0926,
      "step": 9291
    },
    {
      "epoch": 5.944977607165707,
      "grad_norm": 1.0254498219106332,
      "learning_rate": 7.453699420857178e-06,
      "loss": 0.0851,
      "step": 9292
    },
    {
      "epoch": 5.945617402431222,
      "grad_norm": 1.1044242534709314,
      "learning_rate": 7.451695622780517e-06,
      "loss": 0.0882,
      "step": 9293
    },
    {
      "epoch": 5.946257197696737,
      "grad_norm": 0.9798482511373161,
      "learning_rate": 7.449691934123513e-06,
      "loss": 0.0856,
      "step": 9294
    },
    {
      "epoch": 5.946896992962252,
      "grad_norm": 1.0064842544628037,
      "learning_rate": 7.447688354972198e-06,
      "loss": 0.0889,
      "step": 9295
    },
    {
      "epoch": 5.947536788227767,
      "grad_norm": 1.0992211423995755,
      "learning_rate": 7.445684885412607e-06,
      "loss": 0.0868,
      "step": 9296
    },
    {
      "epoch": 5.948176583493282,
      "grad_norm": 0.9800891636813813,
      "learning_rate": 7.443681525530765e-06,
      "loss": 0.0835,
      "step": 9297
    },
    {
      "epoch": 5.9488163787587975,
      "grad_norm": 1.0333177162423952,
      "learning_rate": 7.441678275412689e-06,
      "loss": 0.0835,
      "step": 9298
    },
    {
      "epoch": 5.949456174024312,
      "grad_norm": 0.8586419065182357,
      "learning_rate": 7.439675135144394e-06,
      "loss": 0.0778,
      "step": 9299
    },
    {
      "epoch": 5.950095969289827,
      "grad_norm": 0.9630750321902657,
      "learning_rate": 7.4376721048118984e-06,
      "loss": 0.0738,
      "step": 9300
    },
    {
      "epoch": 5.9507357645553425,
      "grad_norm": 0.882621553236987,
      "learning_rate": 7.4356691845012045e-06,
      "loss": 0.0843,
      "step": 9301
    },
    {
      "epoch": 5.951375559820857,
      "grad_norm": 0.9647828907933518,
      "learning_rate": 7.433666374298311e-06,
      "loss": 0.0777,
      "step": 9302
    },
    {
      "epoch": 5.952015355086372,
      "grad_norm": 0.9442131692126818,
      "learning_rate": 7.431663674289222e-06,
      "loss": 0.0846,
      "step": 9303
    },
    {
      "epoch": 5.952655150351887,
      "grad_norm": 0.9186030016434398,
      "learning_rate": 7.429661084559925e-06,
      "loss": 0.0789,
      "step": 9304
    },
    {
      "epoch": 5.953294945617403,
      "grad_norm": 0.9997901915259341,
      "learning_rate": 7.427658605196404e-06,
      "loss": 0.0887,
      "step": 9305
    },
    {
      "epoch": 5.953934740882918,
      "grad_norm": 0.9628918817089841,
      "learning_rate": 7.4256562362846505e-06,
      "loss": 0.081,
      "step": 9306
    },
    {
      "epoch": 5.954574536148432,
      "grad_norm": 1.0468378768179685,
      "learning_rate": 7.423653977910641e-06,
      "loss": 0.0886,
      "step": 9307
    },
    {
      "epoch": 5.955214331413948,
      "grad_norm": 1.144091841693801,
      "learning_rate": 7.4216518301603454e-06,
      "loss": 0.0939,
      "step": 9308
    },
    {
      "epoch": 5.955854126679463,
      "grad_norm": 0.9242375363594784,
      "learning_rate": 7.419649793119732e-06,
      "loss": 0.0746,
      "step": 9309
    },
    {
      "epoch": 5.956493921944977,
      "grad_norm": 0.9394895916328715,
      "learning_rate": 7.417647866874769e-06,
      "loss": 0.0822,
      "step": 9310
    },
    {
      "epoch": 5.957133717210493,
      "grad_norm": 0.9402356012862072,
      "learning_rate": 7.4156460515114115e-06,
      "loss": 0.0789,
      "step": 9311
    },
    {
      "epoch": 5.957773512476008,
      "grad_norm": 1.0902730360112316,
      "learning_rate": 7.413644347115612e-06,
      "loss": 0.091,
      "step": 9312
    },
    {
      "epoch": 5.958413307741523,
      "grad_norm": 0.9849273990917372,
      "learning_rate": 7.411642753773327e-06,
      "loss": 0.0887,
      "step": 9313
    },
    {
      "epoch": 5.9590531030070375,
      "grad_norm": 1.09031848107851,
      "learning_rate": 7.409641271570498e-06,
      "loss": 0.0859,
      "step": 9314
    },
    {
      "epoch": 5.959692898272553,
      "grad_norm": 1.0323907368510568,
      "learning_rate": 7.407639900593062e-06,
      "loss": 0.0928,
      "step": 9315
    },
    {
      "epoch": 5.960332693538068,
      "grad_norm": 1.0511584127423363,
      "learning_rate": 7.40563864092696e-06,
      "loss": 0.0903,
      "step": 9316
    },
    {
      "epoch": 5.960972488803582,
      "grad_norm": 1.1240112471388704,
      "learning_rate": 7.40363749265812e-06,
      "loss": 0.096,
      "step": 9317
    },
    {
      "epoch": 5.961612284069098,
      "grad_norm": 1.0303545819935838,
      "learning_rate": 7.401636455872469e-06,
      "loss": 0.0887,
      "step": 9318
    },
    {
      "epoch": 5.962252079334613,
      "grad_norm": 0.924660640916681,
      "learning_rate": 7.399635530655922e-06,
      "loss": 0.081,
      "step": 9319
    },
    {
      "epoch": 5.962891874600128,
      "grad_norm": 0.9073376438915698,
      "learning_rate": 7.397634717094403e-06,
      "loss": 0.0785,
      "step": 9320
    },
    {
      "epoch": 5.963531669865643,
      "grad_norm": 0.9362407306037782,
      "learning_rate": 7.3956340152738206e-06,
      "loss": 0.0836,
      "step": 9321
    },
    {
      "epoch": 5.964171465131158,
      "grad_norm": 1.0506236797010133,
      "learning_rate": 7.393633425280076e-06,
      "loss": 0.081,
      "step": 9322
    },
    {
      "epoch": 5.964811260396673,
      "grad_norm": 1.1647786633765767,
      "learning_rate": 7.391632947199079e-06,
      "loss": 0.0915,
      "step": 9323
    },
    {
      "epoch": 5.9654510556621885,
      "grad_norm": 0.9542582745292671,
      "learning_rate": 7.3896325811167235e-06,
      "loss": 0.081,
      "step": 9324
    },
    {
      "epoch": 5.966090850927703,
      "grad_norm": 0.9106521863647665,
      "learning_rate": 7.387632327118901e-06,
      "loss": 0.0809,
      "step": 9325
    },
    {
      "epoch": 5.966730646193218,
      "grad_norm": 0.9784916305994529,
      "learning_rate": 7.3856321852915e-06,
      "loss": 0.084,
      "step": 9326
    },
    {
      "epoch": 5.967370441458733,
      "grad_norm": 1.0301434776101677,
      "learning_rate": 7.383632155720403e-06,
      "loss": 0.0836,
      "step": 9327
    },
    {
      "epoch": 5.968010236724249,
      "grad_norm": 1.1118231173663964,
      "learning_rate": 7.381632238491487e-06,
      "loss": 0.0843,
      "step": 9328
    },
    {
      "epoch": 5.968650031989763,
      "grad_norm": 0.9066972855748539,
      "learning_rate": 7.37963243369062e-06,
      "loss": 0.0803,
      "step": 9329
    },
    {
      "epoch": 5.969289827255278,
      "grad_norm": 1.0046500387149706,
      "learning_rate": 7.377632741403679e-06,
      "loss": 0.0864,
      "step": 9330
    },
    {
      "epoch": 5.969929622520794,
      "grad_norm": 0.9819105775973775,
      "learning_rate": 7.375633161716524e-06,
      "loss": 0.077,
      "step": 9331
    },
    {
      "epoch": 5.970569417786308,
      "grad_norm": 0.9144693749255084,
      "learning_rate": 7.373633694715012e-06,
      "loss": 0.0817,
      "step": 9332
    },
    {
      "epoch": 5.971209213051823,
      "grad_norm": 1.061508438889832,
      "learning_rate": 7.371634340484998e-06,
      "loss": 0.0924,
      "step": 9333
    },
    {
      "epoch": 5.971849008317339,
      "grad_norm": 1.0306925563963434,
      "learning_rate": 7.3696350991123285e-06,
      "loss": 0.0816,
      "step": 9334
    },
    {
      "epoch": 5.972488803582854,
      "grad_norm": 0.9738679923981166,
      "learning_rate": 7.367635970682847e-06,
      "loss": 0.0876,
      "step": 9335
    },
    {
      "epoch": 5.973128598848368,
      "grad_norm": 0.9722519802244455,
      "learning_rate": 7.365636955282397e-06,
      "loss": 0.0881,
      "step": 9336
    },
    {
      "epoch": 5.9737683941138835,
      "grad_norm": 0.8996388194944559,
      "learning_rate": 7.363638052996811e-06,
      "loss": 0.0817,
      "step": 9337
    },
    {
      "epoch": 5.974408189379399,
      "grad_norm": 1.0289032197129182,
      "learning_rate": 7.361639263911916e-06,
      "loss": 0.0867,
      "step": 9338
    },
    {
      "epoch": 5.975047984644913,
      "grad_norm": 0.9856900263124926,
      "learning_rate": 7.359640588113537e-06,
      "loss": 0.0829,
      "step": 9339
    },
    {
      "epoch": 5.975687779910428,
      "grad_norm": 0.8330608758665512,
      "learning_rate": 7.357642025687495e-06,
      "loss": 0.0791,
      "step": 9340
    },
    {
      "epoch": 5.976327575175944,
      "grad_norm": 1.110466456222743,
      "learning_rate": 7.3556435767196045e-06,
      "loss": 0.0919,
      "step": 9341
    },
    {
      "epoch": 5.976967370441459,
      "grad_norm": 1.019235899854726,
      "learning_rate": 7.3536452412956716e-06,
      "loss": 0.083,
      "step": 9342
    },
    {
      "epoch": 5.977607165706974,
      "grad_norm": 1.0494859689847766,
      "learning_rate": 7.3516470195015075e-06,
      "loss": 0.0807,
      "step": 9343
    },
    {
      "epoch": 5.978246960972489,
      "grad_norm": 0.8341160584092359,
      "learning_rate": 7.349648911422911e-06,
      "loss": 0.0766,
      "step": 9344
    },
    {
      "epoch": 5.978886756238004,
      "grad_norm": 0.9599625577513736,
      "learning_rate": 7.34765091714567e-06,
      "loss": 0.0812,
      "step": 9345
    },
    {
      "epoch": 5.979526551503519,
      "grad_norm": 0.9825974246878485,
      "learning_rate": 7.345653036755585e-06,
      "loss": 0.0826,
      "step": 9346
    },
    {
      "epoch": 5.980166346769034,
      "grad_norm": 0.954447054568226,
      "learning_rate": 7.3436552703384365e-06,
      "loss": 0.0826,
      "step": 9347
    },
    {
      "epoch": 5.980806142034549,
      "grad_norm": 1.0138987564658457,
      "learning_rate": 7.3416576179800055e-06,
      "loss": 0.0876,
      "step": 9348
    },
    {
      "epoch": 5.981445937300064,
      "grad_norm": 1.0032535982090325,
      "learning_rate": 7.339660079766067e-06,
      "loss": 0.0874,
      "step": 9349
    },
    {
      "epoch": 5.982085732565579,
      "grad_norm": 1.020222560559342,
      "learning_rate": 7.3376626557823914e-06,
      "loss": 0.0851,
      "step": 9350
    },
    {
      "epoch": 5.982725527831094,
      "grad_norm": 0.9254712358130835,
      "learning_rate": 7.335665346114749e-06,
      "loss": 0.0812,
      "step": 9351
    },
    {
      "epoch": 5.983365323096609,
      "grad_norm": 1.085664244972199,
      "learning_rate": 7.333668150848889e-06,
      "loss": 0.0853,
      "step": 9352
    },
    {
      "epoch": 5.984005118362124,
      "grad_norm": 0.9961070853373456,
      "learning_rate": 7.331671070070583e-06,
      "loss": 0.0837,
      "step": 9353
    },
    {
      "epoch": 5.984644913627639,
      "grad_norm": 1.0755376772356606,
      "learning_rate": 7.329674103865574e-06,
      "loss": 0.0957,
      "step": 9354
    },
    {
      "epoch": 5.985284708893154,
      "grad_norm": 1.0697833128262182,
      "learning_rate": 7.327677252319607e-06,
      "loss": 0.0895,
      "step": 9355
    },
    {
      "epoch": 5.985924504158669,
      "grad_norm": 0.9315071957405818,
      "learning_rate": 7.325680515518425e-06,
      "loss": 0.0791,
      "step": 9356
    },
    {
      "epoch": 5.9865642994241846,
      "grad_norm": 1.0601101074084192,
      "learning_rate": 7.323683893547766e-06,
      "loss": 0.0903,
      "step": 9357
    },
    {
      "epoch": 5.987204094689699,
      "grad_norm": 1.020680794991849,
      "learning_rate": 7.321687386493361e-06,
      "loss": 0.0864,
      "step": 9358
    },
    {
      "epoch": 5.987843889955214,
      "grad_norm": 0.9183037423809676,
      "learning_rate": 7.319690994440929e-06,
      "loss": 0.0823,
      "step": 9359
    },
    {
      "epoch": 5.9884836852207295,
      "grad_norm": 1.1281699988000007,
      "learning_rate": 7.317694717476202e-06,
      "loss": 0.0885,
      "step": 9360
    },
    {
      "epoch": 5.989123480486245,
      "grad_norm": 0.9885722960139889,
      "learning_rate": 7.315698555684894e-06,
      "loss": 0.0882,
      "step": 9361
    },
    {
      "epoch": 5.989763275751759,
      "grad_norm": 1.0365242705404987,
      "learning_rate": 7.313702509152711e-06,
      "loss": 0.0984,
      "step": 9362
    },
    {
      "epoch": 5.990403071017274,
      "grad_norm": 1.025518081370114,
      "learning_rate": 7.311706577965367e-06,
      "loss": 0.0877,
      "step": 9363
    },
    {
      "epoch": 5.99104286628279,
      "grad_norm": 0.9348572081717371,
      "learning_rate": 7.309710762208562e-06,
      "loss": 0.084,
      "step": 9364
    },
    {
      "epoch": 5.991682661548305,
      "grad_norm": 1.007749844391999,
      "learning_rate": 7.307715061967988e-06,
      "loss": 0.0778,
      "step": 9365
    },
    {
      "epoch": 5.992322456813819,
      "grad_norm": 1.0338553827745933,
      "learning_rate": 7.305719477329338e-06,
      "loss": 0.0811,
      "step": 9366
    },
    {
      "epoch": 5.992962252079335,
      "grad_norm": 0.9586984915682669,
      "learning_rate": 7.303724008378304e-06,
      "loss": 0.0859,
      "step": 9367
    },
    {
      "epoch": 5.99360204734485,
      "grad_norm": 1.1205359663436518,
      "learning_rate": 7.301728655200567e-06,
      "loss": 0.0883,
      "step": 9368
    },
    {
      "epoch": 5.994241842610364,
      "grad_norm": 0.9131787723070495,
      "learning_rate": 7.299733417881796e-06,
      "loss": 0.0766,
      "step": 9369
    },
    {
      "epoch": 5.99488163787588,
      "grad_norm": 0.9954858550378964,
      "learning_rate": 7.297738296507673e-06,
      "loss": 0.0838,
      "step": 9370
    },
    {
      "epoch": 5.995521433141395,
      "grad_norm": 1.026473268419489,
      "learning_rate": 7.29574329116386e-06,
      "loss": 0.0837,
      "step": 9371
    },
    {
      "epoch": 5.99616122840691,
      "grad_norm": 1.0674462189921938,
      "learning_rate": 7.293748401936016e-06,
      "loss": 0.0909,
      "step": 9372
    },
    {
      "epoch": 5.9968010236724245,
      "grad_norm": 0.8688639520847274,
      "learning_rate": 7.291753628909806e-06,
      "loss": 0.0821,
      "step": 9373
    },
    {
      "epoch": 5.99744081893794,
      "grad_norm": 1.022305993419394,
      "learning_rate": 7.289758972170876e-06,
      "loss": 0.0893,
      "step": 9374
    },
    {
      "epoch": 5.998080614203455,
      "grad_norm": 1.0226615094644362,
      "learning_rate": 7.2877644318048754e-06,
      "loss": 0.0901,
      "step": 9375
    },
    {
      "epoch": 5.9987204094689695,
      "grad_norm": 1.0181066718672047,
      "learning_rate": 7.2857700078974415e-06,
      "loss": 0.0878,
      "step": 9376
    },
    {
      "epoch": 5.999360204734485,
      "grad_norm": 0.8991292716179171,
      "learning_rate": 7.2837757005342194e-06,
      "loss": 0.079,
      "step": 9377
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.9903154413263181,
      "learning_rate": 7.281781509800838e-06,
      "loss": 0.0777,
      "step": 9378
    },
    {
      "epoch": 6.000639795265515,
      "grad_norm": 0.7314119736021562,
      "learning_rate": 7.279787435782919e-06,
      "loss": 0.0582,
      "step": 9379
    },
    {
      "epoch": 6.00127959053103,
      "grad_norm": 0.7525238537940039,
      "learning_rate": 7.277793478566091e-06,
      "loss": 0.0572,
      "step": 9380
    },
    {
      "epoch": 6.001919385796545,
      "grad_norm": 0.73233855263517,
      "learning_rate": 7.2757996382359694e-06,
      "loss": 0.0653,
      "step": 9381
    },
    {
      "epoch": 6.00255918106206,
      "grad_norm": 0.7438519219342177,
      "learning_rate": 7.27380591487816e-06,
      "loss": 0.0545,
      "step": 9382
    },
    {
      "epoch": 6.0031989763275755,
      "grad_norm": 0.7298662770426099,
      "learning_rate": 7.271812308578281e-06,
      "loss": 0.056,
      "step": 9383
    },
    {
      "epoch": 6.00383877159309,
      "grad_norm": 0.6908064226585872,
      "learning_rate": 7.2698188194219275e-06,
      "loss": 0.051,
      "step": 9384
    },
    {
      "epoch": 6.004478566858605,
      "grad_norm": 0.8585903200778997,
      "learning_rate": 7.267825447494698e-06,
      "loss": 0.0562,
      "step": 9385
    },
    {
      "epoch": 6.00511836212412,
      "grad_norm": 0.8562176254672683,
      "learning_rate": 7.2658321928821804e-06,
      "loss": 0.054,
      "step": 9386
    },
    {
      "epoch": 6.005758157389636,
      "grad_norm": 0.7386545603106305,
      "learning_rate": 7.263839055669968e-06,
      "loss": 0.0567,
      "step": 9387
    },
    {
      "epoch": 6.00639795265515,
      "grad_norm": 0.7732160928745163,
      "learning_rate": 7.261846035943639e-06,
      "loss": 0.0496,
      "step": 9388
    },
    {
      "epoch": 6.007037747920665,
      "grad_norm": 0.774934657777729,
      "learning_rate": 7.259853133788765e-06,
      "loss": 0.0508,
      "step": 9389
    },
    {
      "epoch": 6.007677543186181,
      "grad_norm": 0.8410547299428415,
      "learning_rate": 7.257860349290928e-06,
      "loss": 0.0505,
      "step": 9390
    },
    {
      "epoch": 6.008317338451695,
      "grad_norm": 0.7660124817353184,
      "learning_rate": 7.255867682535691e-06,
      "loss": 0.0518,
      "step": 9391
    },
    {
      "epoch": 6.00895713371721,
      "grad_norm": 0.7729034117556096,
      "learning_rate": 7.25387513360861e-06,
      "loss": 0.0524,
      "step": 9392
    },
    {
      "epoch": 6.009596928982726,
      "grad_norm": 0.8767756889874782,
      "learning_rate": 7.25188270259525e-06,
      "loss": 0.0564,
      "step": 9393
    },
    {
      "epoch": 6.010236724248241,
      "grad_norm": 0.9269648503007866,
      "learning_rate": 7.2498903895811565e-06,
      "loss": 0.0562,
      "step": 9394
    },
    {
      "epoch": 6.010876519513755,
      "grad_norm": 0.7748793986766676,
      "learning_rate": 7.247898194651879e-06,
      "loss": 0.0528,
      "step": 9395
    },
    {
      "epoch": 6.0115163147792705,
      "grad_norm": 0.8013297563510944,
      "learning_rate": 7.245906117892951e-06,
      "loss": 0.0537,
      "step": 9396
    },
    {
      "epoch": 6.012156110044786,
      "grad_norm": 0.8303995866800556,
      "learning_rate": 7.243914159389922e-06,
      "loss": 0.0542,
      "step": 9397
    },
    {
      "epoch": 6.012795905310301,
      "grad_norm": 0.7910314217880084,
      "learning_rate": 7.2419223192283154e-06,
      "loss": 0.0541,
      "step": 9398
    },
    {
      "epoch": 6.0134357005758154,
      "grad_norm": 0.8834646202268914,
      "learning_rate": 7.2399305974936566e-06,
      "loss": 0.0508,
      "step": 9399
    },
    {
      "epoch": 6.014075495841331,
      "grad_norm": 0.8395328516085322,
      "learning_rate": 7.2379389942714685e-06,
      "loss": 0.0549,
      "step": 9400
    },
    {
      "epoch": 6.014715291106846,
      "grad_norm": 0.8249591666020915,
      "learning_rate": 7.235947509647268e-06,
      "loss": 0.0508,
      "step": 9401
    },
    {
      "epoch": 6.015355086372361,
      "grad_norm": 0.8717553205560656,
      "learning_rate": 7.23395614370656e-06,
      "loss": 0.0561,
      "step": 9402
    },
    {
      "epoch": 6.015994881637876,
      "grad_norm": 0.8680303774551819,
      "learning_rate": 7.231964896534859e-06,
      "loss": 0.0574,
      "step": 9403
    },
    {
      "epoch": 6.016634676903391,
      "grad_norm": 0.7614119980790877,
      "learning_rate": 7.229973768217662e-06,
      "loss": 0.0535,
      "step": 9404
    },
    {
      "epoch": 6.017274472168906,
      "grad_norm": 0.8380532986241379,
      "learning_rate": 7.227982758840461e-06,
      "loss": 0.0527,
      "step": 9405
    },
    {
      "epoch": 6.017914267434421,
      "grad_norm": 0.7649714828048919,
      "learning_rate": 7.225991868488749e-06,
      "loss": 0.0518,
      "step": 9406
    },
    {
      "epoch": 6.018554062699936,
      "grad_norm": 0.8138986776778663,
      "learning_rate": 7.224001097248017e-06,
      "loss": 0.0437,
      "step": 9407
    },
    {
      "epoch": 6.019193857965451,
      "grad_norm": 0.9224022733969506,
      "learning_rate": 7.222010445203735e-06,
      "loss": 0.0544,
      "step": 9408
    },
    {
      "epoch": 6.019833653230966,
      "grad_norm": 0.9525191080795665,
      "learning_rate": 7.220019912441382e-06,
      "loss": 0.0495,
      "step": 9409
    },
    {
      "epoch": 6.020473448496481,
      "grad_norm": 0.8639188164669429,
      "learning_rate": 7.218029499046431e-06,
      "loss": 0.0619,
      "step": 9410
    },
    {
      "epoch": 6.021113243761996,
      "grad_norm": 0.8751671272500173,
      "learning_rate": 7.216039205104346e-06,
      "loss": 0.0593,
      "step": 9411
    },
    {
      "epoch": 6.021753039027511,
      "grad_norm": 0.9913322979253203,
      "learning_rate": 7.214049030700579e-06,
      "loss": 0.0574,
      "step": 9412
    },
    {
      "epoch": 6.022392834293027,
      "grad_norm": 0.843800418052034,
      "learning_rate": 7.212058975920598e-06,
      "loss": 0.0455,
      "step": 9413
    },
    {
      "epoch": 6.023032629558541,
      "grad_norm": 0.777342485185414,
      "learning_rate": 7.210069040849842e-06,
      "loss": 0.0519,
      "step": 9414
    },
    {
      "epoch": 6.023672424824056,
      "grad_norm": 0.8528802888389051,
      "learning_rate": 7.20807922557376e-06,
      "loss": 0.0514,
      "step": 9415
    },
    {
      "epoch": 6.024312220089572,
      "grad_norm": 0.7354695542915833,
      "learning_rate": 7.2060895301777866e-06,
      "loss": 0.0487,
      "step": 9416
    },
    {
      "epoch": 6.024952015355086,
      "grad_norm": 0.7981537022425044,
      "learning_rate": 7.204099954747362e-06,
      "loss": 0.0501,
      "step": 9417
    },
    {
      "epoch": 6.025591810620601,
      "grad_norm": 0.7361361052979203,
      "learning_rate": 7.202110499367911e-06,
      "loss": 0.0466,
      "step": 9418
    },
    {
      "epoch": 6.0262316058861165,
      "grad_norm": 0.8440774296236886,
      "learning_rate": 7.200121164124856e-06,
      "loss": 0.0522,
      "step": 9419
    },
    {
      "epoch": 6.026871401151632,
      "grad_norm": 0.818521764965017,
      "learning_rate": 7.1981319491036196e-06,
      "loss": 0.0491,
      "step": 9420
    },
    {
      "epoch": 6.027511196417146,
      "grad_norm": 0.822149279886653,
      "learning_rate": 7.196142854389614e-06,
      "loss": 0.058,
      "step": 9421
    },
    {
      "epoch": 6.028150991682661,
      "grad_norm": 0.751027356206936,
      "learning_rate": 7.194153880068244e-06,
      "loss": 0.0523,
      "step": 9422
    },
    {
      "epoch": 6.028790786948177,
      "grad_norm": 0.8596775004094956,
      "learning_rate": 7.192165026224917e-06,
      "loss": 0.0545,
      "step": 9423
    },
    {
      "epoch": 6.029430582213692,
      "grad_norm": 0.7439441506280642,
      "learning_rate": 7.190176292945031e-06,
      "loss": 0.0488,
      "step": 9424
    },
    {
      "epoch": 6.030070377479206,
      "grad_norm": 0.7407679219752544,
      "learning_rate": 7.188187680313974e-06,
      "loss": 0.0524,
      "step": 9425
    },
    {
      "epoch": 6.030710172744722,
      "grad_norm": 0.8270448980095153,
      "learning_rate": 7.186199188417133e-06,
      "loss": 0.0528,
      "step": 9426
    },
    {
      "epoch": 6.031349968010237,
      "grad_norm": 0.7822853215016963,
      "learning_rate": 7.184210817339899e-06,
      "loss": 0.0501,
      "step": 9427
    },
    {
      "epoch": 6.031989763275751,
      "grad_norm": 0.8215330886491944,
      "learning_rate": 7.182222567167641e-06,
      "loss": 0.0532,
      "step": 9428
    },
    {
      "epoch": 6.032629558541267,
      "grad_norm": 0.8428721090798361,
      "learning_rate": 7.1802344379857335e-06,
      "loss": 0.0526,
      "step": 9429
    },
    {
      "epoch": 6.033269353806782,
      "grad_norm": 0.7931792066881757,
      "learning_rate": 7.178246429879545e-06,
      "loss": 0.0403,
      "step": 9430
    },
    {
      "epoch": 6.033909149072297,
      "grad_norm": 0.7451932318573634,
      "learning_rate": 7.176258542934436e-06,
      "loss": 0.0439,
      "step": 9431
    },
    {
      "epoch": 6.0345489443378115,
      "grad_norm": 0.9268848055714706,
      "learning_rate": 7.17427077723576e-06,
      "loss": 0.053,
      "step": 9432
    },
    {
      "epoch": 6.035188739603327,
      "grad_norm": 0.785663627244465,
      "learning_rate": 7.172283132868867e-06,
      "loss": 0.0499,
      "step": 9433
    },
    {
      "epoch": 6.035828534868842,
      "grad_norm": 0.9194006474447675,
      "learning_rate": 7.1702956099191115e-06,
      "loss": 0.0514,
      "step": 9434
    },
    {
      "epoch": 6.036468330134357,
      "grad_norm": 0.7744674786889206,
      "learning_rate": 7.168308208471827e-06,
      "loss": 0.0461,
      "step": 9435
    },
    {
      "epoch": 6.037108125399872,
      "grad_norm": 0.8429612239080287,
      "learning_rate": 7.166320928612349e-06,
      "loss": 0.0528,
      "step": 9436
    },
    {
      "epoch": 6.037747920665387,
      "grad_norm": 0.9194180064846268,
      "learning_rate": 7.164333770426013e-06,
      "loss": 0.0562,
      "step": 9437
    },
    {
      "epoch": 6.038387715930902,
      "grad_norm": 0.8601332560460407,
      "learning_rate": 7.16234673399814e-06,
      "loss": 0.0514,
      "step": 9438
    },
    {
      "epoch": 6.039027511196418,
      "grad_norm": 0.8221816865296178,
      "learning_rate": 7.160359819414047e-06,
      "loss": 0.055,
      "step": 9439
    },
    {
      "epoch": 6.039667306461932,
      "grad_norm": 0.8090289384343264,
      "learning_rate": 7.158373026759056e-06,
      "loss": 0.0541,
      "step": 9440
    },
    {
      "epoch": 6.040307101727447,
      "grad_norm": 0.7063790759894515,
      "learning_rate": 7.156386356118471e-06,
      "loss": 0.0451,
      "step": 9441
    },
    {
      "epoch": 6.0409468969929625,
      "grad_norm": 0.831293224761604,
      "learning_rate": 7.154399807577599e-06,
      "loss": 0.0504,
      "step": 9442
    },
    {
      "epoch": 6.041586692258477,
      "grad_norm": 0.8639468590903675,
      "learning_rate": 7.1524133812217325e-06,
      "loss": 0.0579,
      "step": 9443
    },
    {
      "epoch": 6.042226487523992,
      "grad_norm": 0.8870552970000873,
      "learning_rate": 7.1504270771361724e-06,
      "loss": 0.0551,
      "step": 9444
    },
    {
      "epoch": 6.042866282789507,
      "grad_norm": 0.7746559956923824,
      "learning_rate": 7.148440895406208e-06,
      "loss": 0.0415,
      "step": 9445
    },
    {
      "epoch": 6.043506078055023,
      "grad_norm": 0.8552088480189204,
      "learning_rate": 7.1464548361171135e-06,
      "loss": 0.0495,
      "step": 9446
    },
    {
      "epoch": 6.044145873320537,
      "grad_norm": 0.9332842997359739,
      "learning_rate": 7.144468899354177e-06,
      "loss": 0.0543,
      "step": 9447
    },
    {
      "epoch": 6.044785668586052,
      "grad_norm": 0.9050437053862843,
      "learning_rate": 7.142483085202665e-06,
      "loss": 0.0533,
      "step": 9448
    },
    {
      "epoch": 6.045425463851568,
      "grad_norm": 0.7476165435089006,
      "learning_rate": 7.140497393747843e-06,
      "loss": 0.0468,
      "step": 9449
    },
    {
      "epoch": 6.046065259117083,
      "grad_norm": 0.8589712749001888,
      "learning_rate": 7.13851182507498e-06,
      "loss": 0.0462,
      "step": 9450
    },
    {
      "epoch": 6.046705054382597,
      "grad_norm": 0.8181955965928467,
      "learning_rate": 7.136526379269329e-06,
      "loss": 0.0449,
      "step": 9451
    },
    {
      "epoch": 6.047344849648113,
      "grad_norm": 0.8836923583051935,
      "learning_rate": 7.134541056416142e-06,
      "loss": 0.0507,
      "step": 9452
    },
    {
      "epoch": 6.047984644913628,
      "grad_norm": 0.855958915486856,
      "learning_rate": 7.132555856600662e-06,
      "loss": 0.0539,
      "step": 9453
    },
    {
      "epoch": 6.048624440179142,
      "grad_norm": 0.9357660877708285,
      "learning_rate": 7.1305707799081345e-06,
      "loss": 0.0514,
      "step": 9454
    },
    {
      "epoch": 6.0492642354446575,
      "grad_norm": 0.882883619798223,
      "learning_rate": 7.128585826423794e-06,
      "loss": 0.0468,
      "step": 9455
    },
    {
      "epoch": 6.049904030710173,
      "grad_norm": 0.7974026867043928,
      "learning_rate": 7.126600996232867e-06,
      "loss": 0.0508,
      "step": 9456
    },
    {
      "epoch": 6.050543825975688,
      "grad_norm": 0.7656134321824669,
      "learning_rate": 7.124616289420584e-06,
      "loss": 0.0547,
      "step": 9457
    },
    {
      "epoch": 6.0511836212412025,
      "grad_norm": 0.8939895213002296,
      "learning_rate": 7.122631706072164e-06,
      "loss": 0.0553,
      "step": 9458
    },
    {
      "epoch": 6.051823416506718,
      "grad_norm": 0.8210176051161371,
      "learning_rate": 7.120647246272817e-06,
      "loss": 0.057,
      "step": 9459
    },
    {
      "epoch": 6.052463211772233,
      "grad_norm": 0.7925498115070124,
      "learning_rate": 7.118662910107758e-06,
      "loss": 0.0466,
      "step": 9460
    },
    {
      "epoch": 6.053103007037748,
      "grad_norm": 0.6621361891849726,
      "learning_rate": 7.116678697662189e-06,
      "loss": 0.0446,
      "step": 9461
    },
    {
      "epoch": 6.053742802303263,
      "grad_norm": 0.8386735088940276,
      "learning_rate": 7.114694609021306e-06,
      "loss": 0.0514,
      "step": 9462
    },
    {
      "epoch": 6.054382597568778,
      "grad_norm": 0.8002647618855254,
      "learning_rate": 7.112710644270301e-06,
      "loss": 0.0463,
      "step": 9463
    },
    {
      "epoch": 6.055022392834293,
      "grad_norm": 0.8069597819161904,
      "learning_rate": 7.1107268034943685e-06,
      "loss": 0.0463,
      "step": 9464
    },
    {
      "epoch": 6.0556621880998085,
      "grad_norm": 0.8554240736654805,
      "learning_rate": 7.108743086778688e-06,
      "loss": 0.0514,
      "step": 9465
    },
    {
      "epoch": 6.056301983365323,
      "grad_norm": 0.8783343316549286,
      "learning_rate": 7.106759494208433e-06,
      "loss": 0.0492,
      "step": 9466
    },
    {
      "epoch": 6.056941778630838,
      "grad_norm": 0.7952045289720391,
      "learning_rate": 7.104776025868781e-06,
      "loss": 0.057,
      "step": 9467
    },
    {
      "epoch": 6.057581573896353,
      "grad_norm": 0.8255215786473951,
      "learning_rate": 7.102792681844897e-06,
      "loss": 0.0477,
      "step": 9468
    },
    {
      "epoch": 6.058221369161868,
      "grad_norm": 0.7660497119815015,
      "learning_rate": 7.100809462221938e-06,
      "loss": 0.0518,
      "step": 9469
    },
    {
      "epoch": 6.058861164427383,
      "grad_norm": 0.792602719719813,
      "learning_rate": 7.098826367085065e-06,
      "loss": 0.0487,
      "step": 9470
    },
    {
      "epoch": 6.059500959692898,
      "grad_norm": 0.869732135671838,
      "learning_rate": 7.096843396519429e-06,
      "loss": 0.0516,
      "step": 9471
    },
    {
      "epoch": 6.060140754958414,
      "grad_norm": 0.8047718829426437,
      "learning_rate": 7.0948605506101745e-06,
      "loss": 0.0496,
      "step": 9472
    },
    {
      "epoch": 6.060780550223928,
      "grad_norm": 0.8444431397934477,
      "learning_rate": 7.092877829442433e-06,
      "loss": 0.0556,
      "step": 9473
    },
    {
      "epoch": 6.061420345489443,
      "grad_norm": 0.7712845438898247,
      "learning_rate": 7.090895233101351e-06,
      "loss": 0.0487,
      "step": 9474
    },
    {
      "epoch": 6.062060140754959,
      "grad_norm": 0.8667531763253025,
      "learning_rate": 7.088912761672055e-06,
      "loss": 0.0542,
      "step": 9475
    },
    {
      "epoch": 6.062699936020474,
      "grad_norm": 0.8830206759643792,
      "learning_rate": 7.086930415239663e-06,
      "loss": 0.0488,
      "step": 9476
    },
    {
      "epoch": 6.063339731285988,
      "grad_norm": 0.812672592161428,
      "learning_rate": 7.0849481938892986e-06,
      "loss": 0.0532,
      "step": 9477
    },
    {
      "epoch": 6.0639795265515035,
      "grad_norm": 0.8391011655505548,
      "learning_rate": 7.082966097706073e-06,
      "loss": 0.0517,
      "step": 9478
    },
    {
      "epoch": 6.064619321817019,
      "grad_norm": 0.9296156951725556,
      "learning_rate": 7.08098412677509e-06,
      "loss": 0.0512,
      "step": 9479
    },
    {
      "epoch": 6.065259117082533,
      "grad_norm": 0.9378027012476218,
      "learning_rate": 7.079002281181462e-06,
      "loss": 0.0527,
      "step": 9480
    },
    {
      "epoch": 6.0658989123480485,
      "grad_norm": 0.8120297062533691,
      "learning_rate": 7.077020561010279e-06,
      "loss": 0.0494,
      "step": 9481
    },
    {
      "epoch": 6.066538707613564,
      "grad_norm": 0.9100353468489062,
      "learning_rate": 7.075038966346632e-06,
      "loss": 0.0519,
      "step": 9482
    },
    {
      "epoch": 6.067178502879079,
      "grad_norm": 0.829044546868418,
      "learning_rate": 7.073057497275608e-06,
      "loss": 0.0513,
      "step": 9483
    },
    {
      "epoch": 6.067818298144593,
      "grad_norm": 0.79256857981101,
      "learning_rate": 7.071076153882292e-06,
      "loss": 0.0569,
      "step": 9484
    },
    {
      "epoch": 6.068458093410109,
      "grad_norm": 0.912498136253117,
      "learning_rate": 7.069094936251754e-06,
      "loss": 0.0485,
      "step": 9485
    },
    {
      "epoch": 6.069097888675624,
      "grad_norm": 0.8863337432839824,
      "learning_rate": 7.067113844469063e-06,
      "loss": 0.054,
      "step": 9486
    },
    {
      "epoch": 6.069737683941139,
      "grad_norm": 0.7272747859340537,
      "learning_rate": 7.065132878619288e-06,
      "loss": 0.0466,
      "step": 9487
    },
    {
      "epoch": 6.070377479206654,
      "grad_norm": 0.8230817534117295,
      "learning_rate": 7.0631520387874885e-06,
      "loss": 0.0487,
      "step": 9488
    },
    {
      "epoch": 6.071017274472169,
      "grad_norm": 0.807032770983593,
      "learning_rate": 7.061171325058713e-06,
      "loss": 0.0538,
      "step": 9489
    },
    {
      "epoch": 6.071657069737684,
      "grad_norm": 0.808132047348372,
      "learning_rate": 7.0591907375180156e-06,
      "loss": 0.0459,
      "step": 9490
    },
    {
      "epoch": 6.0722968650031985,
      "grad_norm": 0.6737425730689108,
      "learning_rate": 7.057210276250437e-06,
      "loss": 0.0441,
      "step": 9491
    },
    {
      "epoch": 6.072936660268714,
      "grad_norm": 0.8428713931420313,
      "learning_rate": 7.0552299413410145e-06,
      "loss": 0.0549,
      "step": 9492
    },
    {
      "epoch": 6.073576455534229,
      "grad_norm": 0.7488482135384865,
      "learning_rate": 7.053249732874775e-06,
      "loss": 0.047,
      "step": 9493
    },
    {
      "epoch": 6.074216250799744,
      "grad_norm": 0.7925337937050403,
      "learning_rate": 7.051269650936754e-06,
      "loss": 0.0484,
      "step": 9494
    },
    {
      "epoch": 6.074856046065259,
      "grad_norm": 0.8614016103066297,
      "learning_rate": 7.0492896956119694e-06,
      "loss": 0.055,
      "step": 9495
    },
    {
      "epoch": 6.075495841330774,
      "grad_norm": 0.8062992902655716,
      "learning_rate": 7.047309866985434e-06,
      "loss": 0.0499,
      "step": 9496
    },
    {
      "epoch": 6.076135636596289,
      "grad_norm": 0.8242825212472632,
      "learning_rate": 7.045330165142164e-06,
      "loss": 0.0501,
      "step": 9497
    },
    {
      "epoch": 6.076775431861805,
      "grad_norm": 0.7785600697878146,
      "learning_rate": 7.043350590167159e-06,
      "loss": 0.0494,
      "step": 9498
    },
    {
      "epoch": 6.077415227127319,
      "grad_norm": 0.7384155523063755,
      "learning_rate": 7.041371142145417e-06,
      "loss": 0.0437,
      "step": 9499
    },
    {
      "epoch": 6.078055022392834,
      "grad_norm": 0.8624094554809774,
      "learning_rate": 7.039391821161938e-06,
      "loss": 0.0481,
      "step": 9500
    },
    {
      "epoch": 6.0786948176583495,
      "grad_norm": 0.8137625798098221,
      "learning_rate": 7.03741262730171e-06,
      "loss": 0.0476,
      "step": 9501
    },
    {
      "epoch": 6.079334612923865,
      "grad_norm": 0.942928011388473,
      "learning_rate": 7.035433560649712e-06,
      "loss": 0.0551,
      "step": 9502
    },
    {
      "epoch": 6.079974408189379,
      "grad_norm": 0.8336464083926236,
      "learning_rate": 7.0334546212909195e-06,
      "loss": 0.0587,
      "step": 9503
    },
    {
      "epoch": 6.0806142034548945,
      "grad_norm": 0.8140855649315262,
      "learning_rate": 7.031475809310313e-06,
      "loss": 0.0511,
      "step": 9504
    },
    {
      "epoch": 6.08125399872041,
      "grad_norm": 0.6835733015831208,
      "learning_rate": 7.0294971247928545e-06,
      "loss": 0.0443,
      "step": 9505
    },
    {
      "epoch": 6.081893793985924,
      "grad_norm": 0.776799511493839,
      "learning_rate": 7.0275185678235025e-06,
      "loss": 0.0521,
      "step": 9506
    },
    {
      "epoch": 6.082533589251439,
      "grad_norm": 0.7047979909086374,
      "learning_rate": 7.025540138487219e-06,
      "loss": 0.0452,
      "step": 9507
    },
    {
      "epoch": 6.083173384516955,
      "grad_norm": 0.7794507188700557,
      "learning_rate": 7.02356183686895e-06,
      "loss": 0.0487,
      "step": 9508
    },
    {
      "epoch": 6.08381317978247,
      "grad_norm": 0.805030889643017,
      "learning_rate": 7.021583663053641e-06,
      "loss": 0.0524,
      "step": 9509
    },
    {
      "epoch": 6.084452975047984,
      "grad_norm": 0.7972146602497798,
      "learning_rate": 7.0196056171262274e-06,
      "loss": 0.0492,
      "step": 9510
    },
    {
      "epoch": 6.0850927703135,
      "grad_norm": 0.8961937648484779,
      "learning_rate": 7.0176276991716505e-06,
      "loss": 0.0507,
      "step": 9511
    },
    {
      "epoch": 6.085732565579015,
      "grad_norm": 0.8322438141860983,
      "learning_rate": 7.015649909274835e-06,
      "loss": 0.0564,
      "step": 9512
    },
    {
      "epoch": 6.08637236084453,
      "grad_norm": 0.8397941005910813,
      "learning_rate": 7.013672247520702e-06,
      "loss": 0.0491,
      "step": 9513
    },
    {
      "epoch": 6.0870121561100445,
      "grad_norm": 0.6511540920172876,
      "learning_rate": 7.011694713994173e-06,
      "loss": 0.0411,
      "step": 9514
    },
    {
      "epoch": 6.08765195137556,
      "grad_norm": 0.9144219363852971,
      "learning_rate": 7.009717308780158e-06,
      "loss": 0.0529,
      "step": 9515
    },
    {
      "epoch": 6.088291746641075,
      "grad_norm": 0.8935158308697413,
      "learning_rate": 7.007740031963556e-06,
      "loss": 0.052,
      "step": 9516
    },
    {
      "epoch": 6.0889315419065895,
      "grad_norm": 0.8318642655737463,
      "learning_rate": 7.005762883629279e-06,
      "loss": 0.0492,
      "step": 9517
    },
    {
      "epoch": 6.089571337172105,
      "grad_norm": 0.7707997652957349,
      "learning_rate": 7.003785863862218e-06,
      "loss": 0.0462,
      "step": 9518
    },
    {
      "epoch": 6.09021113243762,
      "grad_norm": 0.9097129189439095,
      "learning_rate": 7.001808972747261e-06,
      "loss": 0.0522,
      "step": 9519
    },
    {
      "epoch": 6.090850927703135,
      "grad_norm": 0.7526626376674075,
      "learning_rate": 6.999832210369293e-06,
      "loss": 0.0484,
      "step": 9520
    },
    {
      "epoch": 6.09149072296865,
      "grad_norm": 0.8297499155342691,
      "learning_rate": 6.997855576813195e-06,
      "loss": 0.0445,
      "step": 9521
    },
    {
      "epoch": 6.092130518234165,
      "grad_norm": 0.9367377600731517,
      "learning_rate": 6.995879072163837e-06,
      "loss": 0.053,
      "step": 9522
    },
    {
      "epoch": 6.09277031349968,
      "grad_norm": 0.8625700717347948,
      "learning_rate": 6.993902696506083e-06,
      "loss": 0.0556,
      "step": 9523
    },
    {
      "epoch": 6.0934101087651955,
      "grad_norm": 0.7405833767846915,
      "learning_rate": 6.991926449924806e-06,
      "loss": 0.0484,
      "step": 9524
    },
    {
      "epoch": 6.09404990403071,
      "grad_norm": 0.6876206460060831,
      "learning_rate": 6.989950332504856e-06,
      "loss": 0.046,
      "step": 9525
    },
    {
      "epoch": 6.094689699296225,
      "grad_norm": 0.8627297597422517,
      "learning_rate": 6.987974344331081e-06,
      "loss": 0.0519,
      "step": 9526
    },
    {
      "epoch": 6.0953294945617404,
      "grad_norm": 0.8763562119892674,
      "learning_rate": 6.985998485488332e-06,
      "loss": 0.0575,
      "step": 9527
    },
    {
      "epoch": 6.095969289827256,
      "grad_norm": 0.9150920791116399,
      "learning_rate": 6.984022756061446e-06,
      "loss": 0.0509,
      "step": 9528
    },
    {
      "epoch": 6.09660908509277,
      "grad_norm": 0.7185700408243024,
      "learning_rate": 6.982047156135259e-06,
      "loss": 0.0482,
      "step": 9529
    },
    {
      "epoch": 6.097248880358285,
      "grad_norm": 0.779455529678677,
      "learning_rate": 6.980071685794594e-06,
      "loss": 0.0474,
      "step": 9530
    },
    {
      "epoch": 6.097888675623801,
      "grad_norm": 0.7683185472107503,
      "learning_rate": 6.978096345124283e-06,
      "loss": 0.0487,
      "step": 9531
    },
    {
      "epoch": 6.098528470889315,
      "grad_norm": 0.9303130677140249,
      "learning_rate": 6.97612113420914e-06,
      "loss": 0.0503,
      "step": 9532
    },
    {
      "epoch": 6.09916826615483,
      "grad_norm": 0.8072017621217118,
      "learning_rate": 6.97414605313397e-06,
      "loss": 0.0512,
      "step": 9533
    },
    {
      "epoch": 6.099808061420346,
      "grad_norm": 0.7998417182543887,
      "learning_rate": 6.9721711019835925e-06,
      "loss": 0.0464,
      "step": 9534
    },
    {
      "epoch": 6.100447856685861,
      "grad_norm": 0.8464127925486246,
      "learning_rate": 6.9701962808428e-06,
      "loss": 0.0534,
      "step": 9535
    },
    {
      "epoch": 6.101087651951375,
      "grad_norm": 0.8697754589583363,
      "learning_rate": 6.968221589796389e-06,
      "loss": 0.0549,
      "step": 9536
    },
    {
      "epoch": 6.1017274472168905,
      "grad_norm": 0.816156591677778,
      "learning_rate": 6.966247028929151e-06,
      "loss": 0.0522,
      "step": 9537
    },
    {
      "epoch": 6.102367242482406,
      "grad_norm": 0.8773738817076423,
      "learning_rate": 6.96427259832587e-06,
      "loss": 0.046,
      "step": 9538
    },
    {
      "epoch": 6.103007037747921,
      "grad_norm": 0.9361828372213686,
      "learning_rate": 6.9622982980713225e-06,
      "loss": 0.0557,
      "step": 9539
    },
    {
      "epoch": 6.1036468330134355,
      "grad_norm": 0.8153054882063279,
      "learning_rate": 6.960324128250279e-06,
      "loss": 0.0498,
      "step": 9540
    },
    {
      "epoch": 6.104286628278951,
      "grad_norm": 0.8031297216840267,
      "learning_rate": 6.958350088947514e-06,
      "loss": 0.0479,
      "step": 9541
    },
    {
      "epoch": 6.104926423544466,
      "grad_norm": 0.7648551847466436,
      "learning_rate": 6.9563761802477855e-06,
      "loss": 0.0484,
      "step": 9542
    },
    {
      "epoch": 6.10556621880998,
      "grad_norm": 0.7482659589158155,
      "learning_rate": 6.954402402235847e-06,
      "loss": 0.0491,
      "step": 9543
    },
    {
      "epoch": 6.106206014075496,
      "grad_norm": 1.0210824298747903,
      "learning_rate": 6.952428754996454e-06,
      "loss": 0.051,
      "step": 9544
    },
    {
      "epoch": 6.106845809341011,
      "grad_norm": 0.7920531414677969,
      "learning_rate": 6.95045523861435e-06,
      "loss": 0.0456,
      "step": 9545
    },
    {
      "epoch": 6.107485604606526,
      "grad_norm": 0.7750568104070292,
      "learning_rate": 6.948481853174268e-06,
      "loss": 0.0448,
      "step": 9546
    },
    {
      "epoch": 6.108125399872041,
      "grad_norm": 0.8068366693921535,
      "learning_rate": 6.946508598760953e-06,
      "loss": 0.0567,
      "step": 9547
    },
    {
      "epoch": 6.108765195137556,
      "grad_norm": 0.870965730187807,
      "learning_rate": 6.944535475459124e-06,
      "loss": 0.053,
      "step": 9548
    },
    {
      "epoch": 6.109404990403071,
      "grad_norm": 0.78983517907255,
      "learning_rate": 6.942562483353509e-06,
      "loss": 0.0506,
      "step": 9549
    },
    {
      "epoch": 6.110044785668586,
      "grad_norm": 0.7811896858586186,
      "learning_rate": 6.9405896225288195e-06,
      "loss": 0.0475,
      "step": 9550
    },
    {
      "epoch": 6.110684580934101,
      "grad_norm": 0.8841715570259634,
      "learning_rate": 6.93861689306977e-06,
      "loss": 0.0573,
      "step": 9551
    },
    {
      "epoch": 6.111324376199616,
      "grad_norm": 0.7875277564020067,
      "learning_rate": 6.936644295061068e-06,
      "loss": 0.0487,
      "step": 9552
    },
    {
      "epoch": 6.111964171465131,
      "grad_norm": 0.7774739832694778,
      "learning_rate": 6.934671828587405e-06,
      "loss": 0.0516,
      "step": 9553
    },
    {
      "epoch": 6.112603966730646,
      "grad_norm": 0.8640500292576032,
      "learning_rate": 6.932699493733484e-06,
      "loss": 0.0621,
      "step": 9554
    },
    {
      "epoch": 6.113243761996161,
      "grad_norm": 0.758177116814216,
      "learning_rate": 6.930727290583992e-06,
      "loss": 0.047,
      "step": 9555
    },
    {
      "epoch": 6.113883557261676,
      "grad_norm": 0.8347578888149146,
      "learning_rate": 6.928755219223607e-06,
      "loss": 0.0432,
      "step": 9556
    },
    {
      "epoch": 6.114523352527192,
      "grad_norm": 0.8713514040812941,
      "learning_rate": 6.926783279737013e-06,
      "loss": 0.0457,
      "step": 9557
    },
    {
      "epoch": 6.115163147792706,
      "grad_norm": 0.8691809189952338,
      "learning_rate": 6.924811472208878e-06,
      "loss": 0.0472,
      "step": 9558
    },
    {
      "epoch": 6.115802943058221,
      "grad_norm": 0.881820102974921,
      "learning_rate": 6.922839796723867e-06,
      "loss": 0.0554,
      "step": 9559
    },
    {
      "epoch": 6.1164427383237365,
      "grad_norm": 0.864573779206445,
      "learning_rate": 6.920868253366637e-06,
      "loss": 0.0533,
      "step": 9560
    },
    {
      "epoch": 6.117082533589252,
      "grad_norm": 0.9818528690262934,
      "learning_rate": 6.9188968422218525e-06,
      "loss": 0.0505,
      "step": 9561
    },
    {
      "epoch": 6.117722328854766,
      "grad_norm": 0.7491257260832551,
      "learning_rate": 6.916925563374156e-06,
      "loss": 0.0487,
      "step": 9562
    },
    {
      "epoch": 6.1183621241202815,
      "grad_norm": 0.8679570975882251,
      "learning_rate": 6.9149544169081876e-06,
      "loss": 0.0527,
      "step": 9563
    },
    {
      "epoch": 6.119001919385797,
      "grad_norm": 0.7228860170272438,
      "learning_rate": 6.912983402908593e-06,
      "loss": 0.0484,
      "step": 9564
    },
    {
      "epoch": 6.119641714651312,
      "grad_norm": 0.8488918902457196,
      "learning_rate": 6.911012521459999e-06,
      "loss": 0.0524,
      "step": 9565
    },
    {
      "epoch": 6.120281509916826,
      "grad_norm": 0.9651220267873654,
      "learning_rate": 6.90904177264703e-06,
      "loss": 0.053,
      "step": 9566
    },
    {
      "epoch": 6.120921305182342,
      "grad_norm": 0.7879159330031842,
      "learning_rate": 6.907071156554311e-06,
      "loss": 0.0506,
      "step": 9567
    },
    {
      "epoch": 6.121561100447857,
      "grad_norm": 0.707013254054738,
      "learning_rate": 6.905100673266456e-06,
      "loss": 0.0482,
      "step": 9568
    },
    {
      "epoch": 6.122200895713371,
      "grad_norm": 0.9501039483982584,
      "learning_rate": 6.903130322868071e-06,
      "loss": 0.0598,
      "step": 9569
    },
    {
      "epoch": 6.122840690978887,
      "grad_norm": 0.8257929418660792,
      "learning_rate": 6.9011601054437586e-06,
      "loss": 0.05,
      "step": 9570
    },
    {
      "epoch": 6.123480486244402,
      "grad_norm": 0.7697626461189062,
      "learning_rate": 6.899190021078122e-06,
      "loss": 0.053,
      "step": 9571
    },
    {
      "epoch": 6.124120281509917,
      "grad_norm": 0.8961129546172834,
      "learning_rate": 6.89722006985575e-06,
      "loss": 0.0561,
      "step": 9572
    },
    {
      "epoch": 6.1247600767754315,
      "grad_norm": 0.7225299600122008,
      "learning_rate": 6.8952502518612274e-06,
      "loss": 0.0485,
      "step": 9573
    },
    {
      "epoch": 6.125399872040947,
      "grad_norm": 0.8703601624829663,
      "learning_rate": 6.893280567179137e-06,
      "loss": 0.0543,
      "step": 9574
    },
    {
      "epoch": 6.126039667306462,
      "grad_norm": 0.8589409568284375,
      "learning_rate": 6.891311015894054e-06,
      "loss": 0.0532,
      "step": 9575
    },
    {
      "epoch": 6.126679462571977,
      "grad_norm": 0.9505451081883673,
      "learning_rate": 6.88934159809054e-06,
      "loss": 0.0455,
      "step": 9576
    },
    {
      "epoch": 6.127319257837492,
      "grad_norm": 0.8403462294916887,
      "learning_rate": 6.887372313853169e-06,
      "loss": 0.0504,
      "step": 9577
    },
    {
      "epoch": 6.127959053103007,
      "grad_norm": 0.7962347043287451,
      "learning_rate": 6.885403163266495e-06,
      "loss": 0.0493,
      "step": 9578
    },
    {
      "epoch": 6.128598848368522,
      "grad_norm": 0.9001211745878567,
      "learning_rate": 6.883434146415069e-06,
      "loss": 0.0529,
      "step": 9579
    },
    {
      "epoch": 6.129238643634037,
      "grad_norm": 0.847218066627192,
      "learning_rate": 6.881465263383433e-06,
      "loss": 0.0602,
      "step": 9580
    },
    {
      "epoch": 6.129878438899552,
      "grad_norm": 0.7678079104858438,
      "learning_rate": 6.879496514256134e-06,
      "loss": 0.0519,
      "step": 9581
    },
    {
      "epoch": 6.130518234165067,
      "grad_norm": 0.6705358598735334,
      "learning_rate": 6.877527899117704e-06,
      "loss": 0.0453,
      "step": 9582
    },
    {
      "epoch": 6.1311580294305825,
      "grad_norm": 0.7674567788439117,
      "learning_rate": 6.875559418052668e-06,
      "loss": 0.0483,
      "step": 9583
    },
    {
      "epoch": 6.131797824696097,
      "grad_norm": 0.7853698065775413,
      "learning_rate": 6.873591071145556e-06,
      "loss": 0.0466,
      "step": 9584
    },
    {
      "epoch": 6.132437619961612,
      "grad_norm": 0.8254792958415218,
      "learning_rate": 6.871622858480882e-06,
      "loss": 0.0498,
      "step": 9585
    },
    {
      "epoch": 6.1330774152271275,
      "grad_norm": 0.7556742044689019,
      "learning_rate": 6.869654780143158e-06,
      "loss": 0.0485,
      "step": 9586
    },
    {
      "epoch": 6.133717210492643,
      "grad_norm": 0.8140403083484428,
      "learning_rate": 6.867686836216887e-06,
      "loss": 0.0466,
      "step": 9587
    },
    {
      "epoch": 6.134357005758157,
      "grad_norm": 0.8657498581420933,
      "learning_rate": 6.865719026786573e-06,
      "loss": 0.0455,
      "step": 9588
    },
    {
      "epoch": 6.134996801023672,
      "grad_norm": 0.9202430350770958,
      "learning_rate": 6.863751351936708e-06,
      "loss": 0.0541,
      "step": 9589
    },
    {
      "epoch": 6.135636596289188,
      "grad_norm": 0.8226654868224108,
      "learning_rate": 6.861783811751779e-06,
      "loss": 0.0476,
      "step": 9590
    },
    {
      "epoch": 6.136276391554702,
      "grad_norm": 0.8404815905377757,
      "learning_rate": 6.859816406316273e-06,
      "loss": 0.05,
      "step": 9591
    },
    {
      "epoch": 6.136916186820217,
      "grad_norm": 0.826166871509949,
      "learning_rate": 6.857849135714664e-06,
      "loss": 0.0493,
      "step": 9592
    },
    {
      "epoch": 6.137555982085733,
      "grad_norm": 0.7890288770614455,
      "learning_rate": 6.8558820000314205e-06,
      "loss": 0.0502,
      "step": 9593
    },
    {
      "epoch": 6.138195777351248,
      "grad_norm": 0.7312160195247616,
      "learning_rate": 6.853914999351014e-06,
      "loss": 0.0445,
      "step": 9594
    },
    {
      "epoch": 6.138835572616762,
      "grad_norm": 0.865327253825697,
      "learning_rate": 6.851948133757904e-06,
      "loss": 0.0533,
      "step": 9595
    },
    {
      "epoch": 6.1394753678822775,
      "grad_norm": 0.7503693150592562,
      "learning_rate": 6.849981403336539e-06,
      "loss": 0.042,
      "step": 9596
    },
    {
      "epoch": 6.140115163147793,
      "grad_norm": 0.9137985623104431,
      "learning_rate": 6.848014808171366e-06,
      "loss": 0.0494,
      "step": 9597
    },
    {
      "epoch": 6.140754958413308,
      "grad_norm": 0.8279341190075434,
      "learning_rate": 6.846048348346834e-06,
      "loss": 0.0461,
      "step": 9598
    },
    {
      "epoch": 6.1413947536788225,
      "grad_norm": 0.8180555667237044,
      "learning_rate": 6.844082023947376e-06,
      "loss": 0.0538,
      "step": 9599
    },
    {
      "epoch": 6.142034548944338,
      "grad_norm": 0.811780150600098,
      "learning_rate": 6.842115835057417e-06,
      "loss": 0.0456,
      "step": 9600
    },
    {
      "epoch": 6.142674344209853,
      "grad_norm": 0.8653610902928627,
      "learning_rate": 6.8401497817613895e-06,
      "loss": 0.0532,
      "step": 9601
    },
    {
      "epoch": 6.143314139475368,
      "grad_norm": 0.7811894819835421,
      "learning_rate": 6.83818386414371e-06,
      "loss": 0.0499,
      "step": 9602
    },
    {
      "epoch": 6.143953934740883,
      "grad_norm": 0.8903685348181873,
      "learning_rate": 6.83621808228879e-06,
      "loss": 0.0553,
      "step": 9603
    },
    {
      "epoch": 6.144593730006398,
      "grad_norm": 0.8546678109377559,
      "learning_rate": 6.834252436281039e-06,
      "loss": 0.0534,
      "step": 9604
    },
    {
      "epoch": 6.145233525271913,
      "grad_norm": 0.7265484581853292,
      "learning_rate": 6.832286926204857e-06,
      "loss": 0.0478,
      "step": 9605
    },
    {
      "epoch": 6.145873320537428,
      "grad_norm": 0.865129604066096,
      "learning_rate": 6.830321552144638e-06,
      "loss": 0.0536,
      "step": 9606
    },
    {
      "epoch": 6.146513115802943,
      "grad_norm": 0.8595066385186234,
      "learning_rate": 6.828356314184772e-06,
      "loss": 0.0473,
      "step": 9607
    },
    {
      "epoch": 6.147152911068458,
      "grad_norm": 0.8249764799631406,
      "learning_rate": 6.826391212409647e-06,
      "loss": 0.0513,
      "step": 9608
    },
    {
      "epoch": 6.1477927063339735,
      "grad_norm": 0.8947043914957471,
      "learning_rate": 6.824426246903637e-06,
      "loss": 0.0515,
      "step": 9609
    },
    {
      "epoch": 6.148432501599488,
      "grad_norm": 0.8651279753361286,
      "learning_rate": 6.822461417751114e-06,
      "loss": 0.0523,
      "step": 9610
    },
    {
      "epoch": 6.149072296865003,
      "grad_norm": 0.9014030674801135,
      "learning_rate": 6.820496725036446e-06,
      "loss": 0.0533,
      "step": 9611
    },
    {
      "epoch": 6.149712092130518,
      "grad_norm": 0.8596524879264179,
      "learning_rate": 6.8185321688439945e-06,
      "loss": 0.0514,
      "step": 9612
    },
    {
      "epoch": 6.150351887396034,
      "grad_norm": 0.8414626965173624,
      "learning_rate": 6.816567749258106e-06,
      "loss": 0.0523,
      "step": 9613
    },
    {
      "epoch": 6.150991682661548,
      "grad_norm": 0.8410416863368936,
      "learning_rate": 6.814603466363141e-06,
      "loss": 0.0554,
      "step": 9614
    },
    {
      "epoch": 6.151631477927063,
      "grad_norm": 0.7015726323765308,
      "learning_rate": 6.812639320243437e-06,
      "loss": 0.0484,
      "step": 9615
    },
    {
      "epoch": 6.152271273192579,
      "grad_norm": 0.9104749011729707,
      "learning_rate": 6.81067531098333e-06,
      "loss": 0.0546,
      "step": 9616
    },
    {
      "epoch": 6.152911068458093,
      "grad_norm": 0.8834665774919147,
      "learning_rate": 6.80871143866715e-06,
      "loss": 0.0511,
      "step": 9617
    },
    {
      "epoch": 6.153550863723608,
      "grad_norm": 0.8839251313670029,
      "learning_rate": 6.806747703379226e-06,
      "loss": 0.0496,
      "step": 9618
    },
    {
      "epoch": 6.1541906589891235,
      "grad_norm": 1.0169483173389193,
      "learning_rate": 6.804784105203876e-06,
      "loss": 0.056,
      "step": 9619
    },
    {
      "epoch": 6.154830454254639,
      "grad_norm": 0.8653770186524139,
      "learning_rate": 6.8028206442254076e-06,
      "loss": 0.0518,
      "step": 9620
    },
    {
      "epoch": 6.155470249520153,
      "grad_norm": 0.8751738138571441,
      "learning_rate": 6.800857320528137e-06,
      "loss": 0.0496,
      "step": 9621
    },
    {
      "epoch": 6.1561100447856685,
      "grad_norm": 0.8071672865358067,
      "learning_rate": 6.798894134196363e-06,
      "loss": 0.0458,
      "step": 9622
    },
    {
      "epoch": 6.156749840051184,
      "grad_norm": 0.841634042210744,
      "learning_rate": 6.796931085314376e-06,
      "loss": 0.0544,
      "step": 9623
    },
    {
      "epoch": 6.157389635316699,
      "grad_norm": 0.7940777740467528,
      "learning_rate": 6.794968173966475e-06,
      "loss": 0.0446,
      "step": 9624
    },
    {
      "epoch": 6.158029430582213,
      "grad_norm": 0.9942123256242515,
      "learning_rate": 6.79300540023694e-06,
      "loss": 0.0626,
      "step": 9625
    },
    {
      "epoch": 6.158669225847729,
      "grad_norm": 0.9441694608482725,
      "learning_rate": 6.791042764210048e-06,
      "loss": 0.0494,
      "step": 9626
    },
    {
      "epoch": 6.159309021113244,
      "grad_norm": 0.8335767393545384,
      "learning_rate": 6.7890802659700685e-06,
      "loss": 0.0556,
      "step": 9627
    },
    {
      "epoch": 6.159948816378758,
      "grad_norm": 0.8368899657593705,
      "learning_rate": 6.787117905601273e-06,
      "loss": 0.0499,
      "step": 9628
    },
    {
      "epoch": 6.160588611644274,
      "grad_norm": 0.7708845426703721,
      "learning_rate": 6.785155683187919e-06,
      "loss": 0.0472,
      "step": 9629
    },
    {
      "epoch": 6.161228406909789,
      "grad_norm": 0.8306223234639187,
      "learning_rate": 6.783193598814257e-06,
      "loss": 0.0519,
      "step": 9630
    },
    {
      "epoch": 6.161868202175304,
      "grad_norm": 0.8394982325582869,
      "learning_rate": 6.781231652564544e-06,
      "loss": 0.0464,
      "step": 9631
    },
    {
      "epoch": 6.162507997440819,
      "grad_norm": 0.7952066121574374,
      "learning_rate": 6.779269844523019e-06,
      "loss": 0.0483,
      "step": 9632
    },
    {
      "epoch": 6.163147792706334,
      "grad_norm": 0.8295201911361301,
      "learning_rate": 6.777308174773916e-06,
      "loss": 0.0549,
      "step": 9633
    },
    {
      "epoch": 6.163787587971849,
      "grad_norm": 0.7523158868620522,
      "learning_rate": 6.7753466434014666e-06,
      "loss": 0.0479,
      "step": 9634
    },
    {
      "epoch": 6.164427383237364,
      "grad_norm": 0.7222054925976732,
      "learning_rate": 6.773385250489899e-06,
      "loss": 0.0465,
      "step": 9635
    },
    {
      "epoch": 6.165067178502879,
      "grad_norm": 0.7615191588753675,
      "learning_rate": 6.771423996123427e-06,
      "loss": 0.0497,
      "step": 9636
    },
    {
      "epoch": 6.165706973768394,
      "grad_norm": 0.8189472835583556,
      "learning_rate": 6.769462880386262e-06,
      "loss": 0.0486,
      "step": 9637
    },
    {
      "epoch": 6.166346769033909,
      "grad_norm": 0.8035964206706683,
      "learning_rate": 6.767501903362619e-06,
      "loss": 0.0426,
      "step": 9638
    },
    {
      "epoch": 6.166986564299425,
      "grad_norm": 0.8339709337205807,
      "learning_rate": 6.7655410651366925e-06,
      "loss": 0.051,
      "step": 9639
    },
    {
      "epoch": 6.167626359564939,
      "grad_norm": 0.7482550637202433,
      "learning_rate": 6.763580365792678e-06,
      "loss": 0.0465,
      "step": 9640
    },
    {
      "epoch": 6.168266154830454,
      "grad_norm": 0.8921778394880812,
      "learning_rate": 6.761619805414768e-06,
      "loss": 0.0536,
      "step": 9641
    },
    {
      "epoch": 6.1689059500959695,
      "grad_norm": 0.7955241802503037,
      "learning_rate": 6.759659384087143e-06,
      "loss": 0.0522,
      "step": 9642
    },
    {
      "epoch": 6.169545745361484,
      "grad_norm": 0.8257659962620414,
      "learning_rate": 6.757699101893974e-06,
      "loss": 0.0516,
      "step": 9643
    },
    {
      "epoch": 6.170185540626999,
      "grad_norm": 0.941219948788986,
      "learning_rate": 6.755738958919444e-06,
      "loss": 0.0504,
      "step": 9644
    },
    {
      "epoch": 6.1708253358925145,
      "grad_norm": 0.9304306626095775,
      "learning_rate": 6.753778955247711e-06,
      "loss": 0.056,
      "step": 9645
    },
    {
      "epoch": 6.17146513115803,
      "grad_norm": 0.9231297015659289,
      "learning_rate": 6.751819090962934e-06,
      "loss": 0.0537,
      "step": 9646
    },
    {
      "epoch": 6.172104926423544,
      "grad_norm": 0.8191812193645897,
      "learning_rate": 6.749859366149267e-06,
      "loss": 0.0471,
      "step": 9647
    },
    {
      "epoch": 6.172744721689059,
      "grad_norm": 0.8905363913719844,
      "learning_rate": 6.747899780890858e-06,
      "loss": 0.0511,
      "step": 9648
    },
    {
      "epoch": 6.173384516954575,
      "grad_norm": 0.8936965621020133,
      "learning_rate": 6.74594033527185e-06,
      "loss": 0.0599,
      "step": 9649
    },
    {
      "epoch": 6.17402431222009,
      "grad_norm": 0.8474076821725791,
      "learning_rate": 6.74398102937637e-06,
      "loss": 0.0504,
      "step": 9650
    },
    {
      "epoch": 6.174664107485604,
      "grad_norm": 0.8101947154765451,
      "learning_rate": 6.742021863288555e-06,
      "loss": 0.0486,
      "step": 9651
    },
    {
      "epoch": 6.17530390275112,
      "grad_norm": 0.8737040308461602,
      "learning_rate": 6.740062837092527e-06,
      "loss": 0.0486,
      "step": 9652
    },
    {
      "epoch": 6.175943698016635,
      "grad_norm": 0.8742736960550169,
      "learning_rate": 6.738103950872397e-06,
      "loss": 0.0532,
      "step": 9653
    },
    {
      "epoch": 6.176583493282149,
      "grad_norm": 0.8575930342316083,
      "learning_rate": 6.736145204712287e-06,
      "loss": 0.052,
      "step": 9654
    },
    {
      "epoch": 6.177223288547665,
      "grad_norm": 0.8538191765503449,
      "learning_rate": 6.734186598696294e-06,
      "loss": 0.0545,
      "step": 9655
    },
    {
      "epoch": 6.17786308381318,
      "grad_norm": 0.8098493956158122,
      "learning_rate": 6.7322281329085204e-06,
      "loss": 0.0503,
      "step": 9656
    },
    {
      "epoch": 6.178502879078695,
      "grad_norm": 0.8058475988582309,
      "learning_rate": 6.730269807433054e-06,
      "loss": 0.0512,
      "step": 9657
    },
    {
      "epoch": 6.1791426743442095,
      "grad_norm": 0.8703959795936566,
      "learning_rate": 6.728311622353989e-06,
      "loss": 0.0518,
      "step": 9658
    },
    {
      "epoch": 6.179782469609725,
      "grad_norm": 0.7114189607482804,
      "learning_rate": 6.726353577755402e-06,
      "loss": 0.0481,
      "step": 9659
    },
    {
      "epoch": 6.18042226487524,
      "grad_norm": 0.8257796471651815,
      "learning_rate": 6.7243956737213665e-06,
      "loss": 0.0417,
      "step": 9660
    },
    {
      "epoch": 6.181062060140755,
      "grad_norm": 0.9526236709224978,
      "learning_rate": 6.722437910335957e-06,
      "loss": 0.0514,
      "step": 9661
    },
    {
      "epoch": 6.18170185540627,
      "grad_norm": 0.8123724311434162,
      "learning_rate": 6.720480287683233e-06,
      "loss": 0.0517,
      "step": 9662
    },
    {
      "epoch": 6.182341650671785,
      "grad_norm": 0.8151942246101279,
      "learning_rate": 6.718522805847253e-06,
      "loss": 0.0513,
      "step": 9663
    },
    {
      "epoch": 6.1829814459373,
      "grad_norm": 0.7493210826899108,
      "learning_rate": 6.716565464912063e-06,
      "loss": 0.052,
      "step": 9664
    },
    {
      "epoch": 6.183621241202815,
      "grad_norm": 0.8203614899022719,
      "learning_rate": 6.714608264961714e-06,
      "loss": 0.0509,
      "step": 9665
    },
    {
      "epoch": 6.18426103646833,
      "grad_norm": 0.8200918555292422,
      "learning_rate": 6.712651206080241e-06,
      "loss": 0.0489,
      "step": 9666
    },
    {
      "epoch": 6.184900831733845,
      "grad_norm": 0.8409145899032401,
      "learning_rate": 6.710694288351675e-06,
      "loss": 0.0463,
      "step": 9667
    },
    {
      "epoch": 6.1855406269993605,
      "grad_norm": 0.8164110915856291,
      "learning_rate": 6.708737511860048e-06,
      "loss": 0.0493,
      "step": 9668
    },
    {
      "epoch": 6.186180422264875,
      "grad_norm": 0.8344678856302598,
      "learning_rate": 6.706780876689377e-06,
      "loss": 0.0507,
      "step": 9669
    },
    {
      "epoch": 6.18682021753039,
      "grad_norm": 0.8108374495636151,
      "learning_rate": 6.704824382923676e-06,
      "loss": 0.043,
      "step": 9670
    },
    {
      "epoch": 6.187460012795905,
      "grad_norm": 0.7885876680312017,
      "learning_rate": 6.702868030646956e-06,
      "loss": 0.0497,
      "step": 9671
    },
    {
      "epoch": 6.188099808061421,
      "grad_norm": 0.8278175243575887,
      "learning_rate": 6.700911819943216e-06,
      "loss": 0.0523,
      "step": 9672
    },
    {
      "epoch": 6.188739603326935,
      "grad_norm": 0.8190613060057725,
      "learning_rate": 6.698955750896457e-06,
      "loss": 0.0521,
      "step": 9673
    },
    {
      "epoch": 6.18937939859245,
      "grad_norm": 0.9067452612635473,
      "learning_rate": 6.69699982359066e-06,
      "loss": 0.0586,
      "step": 9674
    },
    {
      "epoch": 6.190019193857966,
      "grad_norm": 0.794619766175695,
      "learning_rate": 6.695044038109819e-06,
      "loss": 0.0504,
      "step": 9675
    },
    {
      "epoch": 6.190658989123481,
      "grad_norm": 0.8238322072762769,
      "learning_rate": 6.693088394537908e-06,
      "loss": 0.0505,
      "step": 9676
    },
    {
      "epoch": 6.191298784388995,
      "grad_norm": 0.8048594769475153,
      "learning_rate": 6.691132892958897e-06,
      "loss": 0.0523,
      "step": 9677
    },
    {
      "epoch": 6.1919385796545106,
      "grad_norm": 0.7688606365400317,
      "learning_rate": 6.6891775334567566e-06,
      "loss": 0.05,
      "step": 9678
    },
    {
      "epoch": 6.192578374920026,
      "grad_norm": 0.7635435583409196,
      "learning_rate": 6.6872223161154414e-06,
      "loss": 0.0449,
      "step": 9679
    },
    {
      "epoch": 6.19321817018554,
      "grad_norm": 0.8513292922221909,
      "learning_rate": 6.685267241018904e-06,
      "loss": 0.0486,
      "step": 9680
    },
    {
      "epoch": 6.1938579654510555,
      "grad_norm": 0.9089781185083043,
      "learning_rate": 6.683312308251099e-06,
      "loss": 0.0541,
      "step": 9681
    },
    {
      "epoch": 6.194497760716571,
      "grad_norm": 0.8757297648496923,
      "learning_rate": 6.6813575178959635e-06,
      "loss": 0.0495,
      "step": 9682
    },
    {
      "epoch": 6.195137555982086,
      "grad_norm": 0.7109332572527652,
      "learning_rate": 6.6794028700374325e-06,
      "loss": 0.0444,
      "step": 9683
    },
    {
      "epoch": 6.1957773512476,
      "grad_norm": 0.7374122971014161,
      "learning_rate": 6.6774483647594314e-06,
      "loss": 0.0493,
      "step": 9684
    },
    {
      "epoch": 6.196417146513116,
      "grad_norm": 0.8504674788716049,
      "learning_rate": 6.675494002145891e-06,
      "loss": 0.0493,
      "step": 9685
    },
    {
      "epoch": 6.197056941778631,
      "grad_norm": 0.8508779196394514,
      "learning_rate": 6.673539782280725e-06,
      "loss": 0.0527,
      "step": 9686
    },
    {
      "epoch": 6.197696737044146,
      "grad_norm": 0.8904414756501823,
      "learning_rate": 6.671585705247841e-06,
      "loss": 0.0584,
      "step": 9687
    },
    {
      "epoch": 6.198336532309661,
      "grad_norm": 0.7476067529655782,
      "learning_rate": 6.6696317711311484e-06,
      "loss": 0.0544,
      "step": 9688
    },
    {
      "epoch": 6.198976327575176,
      "grad_norm": 0.8945291816741262,
      "learning_rate": 6.667677980014542e-06,
      "loss": 0.051,
      "step": 9689
    },
    {
      "epoch": 6.199616122840691,
      "grad_norm": 0.7692704394546657,
      "learning_rate": 6.6657243319819135e-06,
      "loss": 0.0479,
      "step": 9690
    },
    {
      "epoch": 6.2002559181062065,
      "grad_norm": 0.7206505201102126,
      "learning_rate": 6.663770827117153e-06,
      "loss": 0.0482,
      "step": 9691
    },
    {
      "epoch": 6.200895713371721,
      "grad_norm": 0.797027833697124,
      "learning_rate": 6.661817465504139e-06,
      "loss": 0.0501,
      "step": 9692
    },
    {
      "epoch": 6.201535508637236,
      "grad_norm": 0.932827374994183,
      "learning_rate": 6.6598642472267465e-06,
      "loss": 0.0574,
      "step": 9693
    },
    {
      "epoch": 6.202175303902751,
      "grad_norm": 0.818306580879808,
      "learning_rate": 6.657911172368837e-06,
      "loss": 0.0468,
      "step": 9694
    },
    {
      "epoch": 6.202815099168266,
      "grad_norm": 0.8640488417795474,
      "learning_rate": 6.65595824101428e-06,
      "loss": 0.0547,
      "step": 9695
    },
    {
      "epoch": 6.203454894433781,
      "grad_norm": 0.8514143886404778,
      "learning_rate": 6.654005453246927e-06,
      "loss": 0.0494,
      "step": 9696
    },
    {
      "epoch": 6.204094689699296,
      "grad_norm": 0.7569120255542664,
      "learning_rate": 6.652052809150624e-06,
      "loss": 0.0464,
      "step": 9697
    },
    {
      "epoch": 6.204734484964812,
      "grad_norm": 0.8468738911354675,
      "learning_rate": 6.650100308809222e-06,
      "loss": 0.0507,
      "step": 9698
    },
    {
      "epoch": 6.205374280230326,
      "grad_norm": 0.7991466413932059,
      "learning_rate": 6.648147952306554e-06,
      "loss": 0.0458,
      "step": 9699
    },
    {
      "epoch": 6.206014075495841,
      "grad_norm": 0.8125228093278378,
      "learning_rate": 6.646195739726448e-06,
      "loss": 0.0587,
      "step": 9700
    },
    {
      "epoch": 6.2066538707613566,
      "grad_norm": 0.8324026277117197,
      "learning_rate": 6.644243671152734e-06,
      "loss": 0.0477,
      "step": 9701
    },
    {
      "epoch": 6.207293666026872,
      "grad_norm": 0.8114067875494053,
      "learning_rate": 6.642291746669226e-06,
      "loss": 0.047,
      "step": 9702
    },
    {
      "epoch": 6.207933461292386,
      "grad_norm": 0.7642729757186221,
      "learning_rate": 6.640339966359739e-06,
      "loss": 0.0493,
      "step": 9703
    },
    {
      "epoch": 6.2085732565579015,
      "grad_norm": 0.9555239554365855,
      "learning_rate": 6.638388330308072e-06,
      "loss": 0.0489,
      "step": 9704
    },
    {
      "epoch": 6.209213051823417,
      "grad_norm": 0.937862527942406,
      "learning_rate": 6.636436838598035e-06,
      "loss": 0.0523,
      "step": 9705
    },
    {
      "epoch": 6.209852847088931,
      "grad_norm": 0.9699074591976116,
      "learning_rate": 6.634485491313416e-06,
      "loss": 0.0543,
      "step": 9706
    },
    {
      "epoch": 6.210492642354446,
      "grad_norm": 0.7975657289373976,
      "learning_rate": 6.632534288538002e-06,
      "loss": 0.0458,
      "step": 9707
    },
    {
      "epoch": 6.211132437619962,
      "grad_norm": 0.8231493496906077,
      "learning_rate": 6.630583230355576e-06,
      "loss": 0.0449,
      "step": 9708
    },
    {
      "epoch": 6.211772232885477,
      "grad_norm": 0.8069141992242479,
      "learning_rate": 6.628632316849915e-06,
      "loss": 0.0407,
      "step": 9709
    },
    {
      "epoch": 6.212412028150991,
      "grad_norm": 0.8557960128824221,
      "learning_rate": 6.626681548104779e-06,
      "loss": 0.0483,
      "step": 9710
    },
    {
      "epoch": 6.213051823416507,
      "grad_norm": 0.7089831610532107,
      "learning_rate": 6.624730924203941e-06,
      "loss": 0.0487,
      "step": 9711
    },
    {
      "epoch": 6.213691618682022,
      "grad_norm": 1.0160323603062749,
      "learning_rate": 6.622780445231154e-06,
      "loss": 0.0571,
      "step": 9712
    },
    {
      "epoch": 6.214331413947537,
      "grad_norm": 0.8003884187816367,
      "learning_rate": 6.620830111270167e-06,
      "loss": 0.046,
      "step": 9713
    },
    {
      "epoch": 6.214971209213052,
      "grad_norm": 0.7881701387010271,
      "learning_rate": 6.618879922404718e-06,
      "loss": 0.0477,
      "step": 9714
    },
    {
      "epoch": 6.215611004478567,
      "grad_norm": 0.8563166657814819,
      "learning_rate": 6.616929878718556e-06,
      "loss": 0.0559,
      "step": 9715
    },
    {
      "epoch": 6.216250799744082,
      "grad_norm": 0.7931555588805717,
      "learning_rate": 6.614979980295407e-06,
      "loss": 0.0513,
      "step": 9716
    },
    {
      "epoch": 6.2168905950095965,
      "grad_norm": 0.9061730941049569,
      "learning_rate": 6.613030227218994e-06,
      "loss": 0.0536,
      "step": 9717
    },
    {
      "epoch": 6.217530390275112,
      "grad_norm": 0.8835656717528041,
      "learning_rate": 6.61108061957304e-06,
      "loss": 0.06,
      "step": 9718
    },
    {
      "epoch": 6.218170185540627,
      "grad_norm": 0.8583578433980368,
      "learning_rate": 6.609131157441256e-06,
      "loss": 0.0528,
      "step": 9719
    },
    {
      "epoch": 6.218809980806142,
      "grad_norm": 1.0435777737002991,
      "learning_rate": 6.6071818409073445e-06,
      "loss": 0.0532,
      "step": 9720
    },
    {
      "epoch": 6.219449776071657,
      "grad_norm": 0.8743078368783751,
      "learning_rate": 6.605232670055013e-06,
      "loss": 0.0565,
      "step": 9721
    },
    {
      "epoch": 6.220089571337172,
      "grad_norm": 0.7735321832691698,
      "learning_rate": 6.6032836449679515e-06,
      "loss": 0.0491,
      "step": 9722
    },
    {
      "epoch": 6.220729366602687,
      "grad_norm": 0.7066526582046997,
      "learning_rate": 6.601334765729849e-06,
      "loss": 0.0501,
      "step": 9723
    },
    {
      "epoch": 6.2213691618682025,
      "grad_norm": 0.9286349117622577,
      "learning_rate": 6.599386032424383e-06,
      "loss": 0.0508,
      "step": 9724
    },
    {
      "epoch": 6.222008957133717,
      "grad_norm": 0.782079471743681,
      "learning_rate": 6.597437445135233e-06,
      "loss": 0.0581,
      "step": 9725
    },
    {
      "epoch": 6.222648752399232,
      "grad_norm": 0.8410493000799785,
      "learning_rate": 6.595489003946067e-06,
      "loss": 0.0557,
      "step": 9726
    },
    {
      "epoch": 6.2232885476647475,
      "grad_norm": 0.7440166501565649,
      "learning_rate": 6.593540708940545e-06,
      "loss": 0.0472,
      "step": 9727
    },
    {
      "epoch": 6.223928342930263,
      "grad_norm": 0.8849221706111029,
      "learning_rate": 6.591592560202326e-06,
      "loss": 0.053,
      "step": 9728
    },
    {
      "epoch": 6.224568138195777,
      "grad_norm": 0.8616931635713696,
      "learning_rate": 6.589644557815061e-06,
      "loss": 0.0529,
      "step": 9729
    },
    {
      "epoch": 6.225207933461292,
      "grad_norm": 0.8902689469495939,
      "learning_rate": 6.587696701862391e-06,
      "loss": 0.0499,
      "step": 9730
    },
    {
      "epoch": 6.225847728726808,
      "grad_norm": 0.831394573057733,
      "learning_rate": 6.585748992427956e-06,
      "loss": 0.0542,
      "step": 9731
    },
    {
      "epoch": 6.226487523992322,
      "grad_norm": 0.8289065739877128,
      "learning_rate": 6.583801429595384e-06,
      "loss": 0.048,
      "step": 9732
    },
    {
      "epoch": 6.227127319257837,
      "grad_norm": 0.8658494104187742,
      "learning_rate": 6.581854013448304e-06,
      "loss": 0.052,
      "step": 9733
    },
    {
      "epoch": 6.227767114523353,
      "grad_norm": 0.8192378982664451,
      "learning_rate": 6.579906744070328e-06,
      "loss": 0.0473,
      "step": 9734
    },
    {
      "epoch": 6.228406909788868,
      "grad_norm": 0.8928583729690961,
      "learning_rate": 6.577959621545076e-06,
      "loss": 0.0503,
      "step": 9735
    },
    {
      "epoch": 6.229046705054382,
      "grad_norm": 0.8941831319320023,
      "learning_rate": 6.576012645956149e-06,
      "loss": 0.0496,
      "step": 9736
    },
    {
      "epoch": 6.229686500319898,
      "grad_norm": 0.8564438914755568,
      "learning_rate": 6.5740658173871485e-06,
      "loss": 0.0543,
      "step": 9737
    },
    {
      "epoch": 6.230326295585413,
      "grad_norm": 0.8434348950924226,
      "learning_rate": 6.572119135921668e-06,
      "loss": 0.0488,
      "step": 9738
    },
    {
      "epoch": 6.230966090850928,
      "grad_norm": 0.9248550975646909,
      "learning_rate": 6.5701726016432955e-06,
      "loss": 0.0568,
      "step": 9739
    },
    {
      "epoch": 6.2316058861164425,
      "grad_norm": 0.7946209945693513,
      "learning_rate": 6.568226214635609e-06,
      "loss": 0.0523,
      "step": 9740
    },
    {
      "epoch": 6.232245681381958,
      "grad_norm": 0.7223365595363386,
      "learning_rate": 6.566279974982182e-06,
      "loss": 0.0478,
      "step": 9741
    },
    {
      "epoch": 6.232885476647473,
      "grad_norm": 0.7979775719532205,
      "learning_rate": 6.564333882766587e-06,
      "loss": 0.0519,
      "step": 9742
    },
    {
      "epoch": 6.233525271912987,
      "grad_norm": 0.835425881984282,
      "learning_rate": 6.562387938072383e-06,
      "loss": 0.0516,
      "step": 9743
    },
    {
      "epoch": 6.234165067178503,
      "grad_norm": 0.8850080307506216,
      "learning_rate": 6.560442140983123e-06,
      "loss": 0.0543,
      "step": 9744
    },
    {
      "epoch": 6.234804862444018,
      "grad_norm": 0.9300416767943065,
      "learning_rate": 6.558496491582362e-06,
      "loss": 0.0533,
      "step": 9745
    },
    {
      "epoch": 6.235444657709533,
      "grad_norm": 0.9322349669631839,
      "learning_rate": 6.556550989953641e-06,
      "loss": 0.0452,
      "step": 9746
    },
    {
      "epoch": 6.236084452975048,
      "grad_norm": 0.8091979435195878,
      "learning_rate": 6.554605636180493e-06,
      "loss": 0.0513,
      "step": 9747
    },
    {
      "epoch": 6.236724248240563,
      "grad_norm": 0.8372212857032424,
      "learning_rate": 6.5526604303464515e-06,
      "loss": 0.055,
      "step": 9748
    },
    {
      "epoch": 6.237364043506078,
      "grad_norm": 0.7998840194711406,
      "learning_rate": 6.5507153725350395e-06,
      "loss": 0.0544,
      "step": 9749
    },
    {
      "epoch": 6.2380038387715935,
      "grad_norm": 0.9242423729624955,
      "learning_rate": 6.5487704628297735e-06,
      "loss": 0.0579,
      "step": 9750
    },
    {
      "epoch": 6.238643634037108,
      "grad_norm": 0.9504172942646608,
      "learning_rate": 6.546825701314159e-06,
      "loss": 0.0527,
      "step": 9751
    },
    {
      "epoch": 6.239283429302623,
      "grad_norm": 0.7949114697582061,
      "learning_rate": 6.544881088071712e-06,
      "loss": 0.0511,
      "step": 9752
    },
    {
      "epoch": 6.239923224568138,
      "grad_norm": 0.9554319120054696,
      "learning_rate": 6.542936623185925e-06,
      "loss": 0.0554,
      "step": 9753
    },
    {
      "epoch": 6.240563019833653,
      "grad_norm": 0.9581505006257472,
      "learning_rate": 6.540992306740286e-06,
      "loss": 0.0515,
      "step": 9754
    },
    {
      "epoch": 6.241202815099168,
      "grad_norm": 0.925762933197951,
      "learning_rate": 6.539048138818287e-06,
      "loss": 0.0528,
      "step": 9755
    },
    {
      "epoch": 6.241842610364683,
      "grad_norm": 0.8531923780950993,
      "learning_rate": 6.537104119503406e-06,
      "loss": 0.0514,
      "step": 9756
    },
    {
      "epoch": 6.242482405630199,
      "grad_norm": 0.9056708973333736,
      "learning_rate": 6.535160248879108e-06,
      "loss": 0.0519,
      "step": 9757
    },
    {
      "epoch": 6.243122200895713,
      "grad_norm": 0.9051152896327558,
      "learning_rate": 6.533216527028869e-06,
      "loss": 0.0527,
      "step": 9758
    },
    {
      "epoch": 6.243761996161228,
      "grad_norm": 0.86305882122961,
      "learning_rate": 6.5312729540361475e-06,
      "loss": 0.0478,
      "step": 9759
    },
    {
      "epoch": 6.244401791426744,
      "grad_norm": 0.8746604102167966,
      "learning_rate": 6.529329529984394e-06,
      "loss": 0.0464,
      "step": 9760
    },
    {
      "epoch": 6.245041586692259,
      "grad_norm": 0.804988068759879,
      "learning_rate": 6.527386254957055e-06,
      "loss": 0.0509,
      "step": 9761
    },
    {
      "epoch": 6.245681381957773,
      "grad_norm": 0.8088818558267934,
      "learning_rate": 6.525443129037574e-06,
      "loss": 0.0419,
      "step": 9762
    },
    {
      "epoch": 6.2463211772232885,
      "grad_norm": 0.8984816760396388,
      "learning_rate": 6.5235001523093856e-06,
      "loss": 0.0503,
      "step": 9763
    },
    {
      "epoch": 6.246960972488804,
      "grad_norm": 0.8090579135758402,
      "learning_rate": 6.521557324855912e-06,
      "loss": 0.0523,
      "step": 9764
    },
    {
      "epoch": 6.247600767754319,
      "grad_norm": 0.7138256947928808,
      "learning_rate": 6.519614646760582e-06,
      "loss": 0.0471,
      "step": 9765
    },
    {
      "epoch": 6.248240563019833,
      "grad_norm": 0.8599401858011289,
      "learning_rate": 6.5176721181068085e-06,
      "loss": 0.0505,
      "step": 9766
    },
    {
      "epoch": 6.248880358285349,
      "grad_norm": 0.8020392884989408,
      "learning_rate": 6.515729738977998e-06,
      "loss": 0.0517,
      "step": 9767
    },
    {
      "epoch": 6.249520153550864,
      "grad_norm": 0.8807076122900194,
      "learning_rate": 6.513787509457557e-06,
      "loss": 0.0554,
      "step": 9768
    },
    {
      "epoch": 6.250159948816378,
      "grad_norm": 0.7482830386518087,
      "learning_rate": 6.511845429628878e-06,
      "loss": 0.0499,
      "step": 9769
    },
    {
      "epoch": 6.250799744081894,
      "grad_norm": 0.7522697698084359,
      "learning_rate": 6.509903499575352e-06,
      "loss": 0.0435,
      "step": 9770
    },
    {
      "epoch": 6.251439539347409,
      "grad_norm": 0.8146820029042667,
      "learning_rate": 6.507961719380356e-06,
      "loss": 0.0526,
      "step": 9771
    },
    {
      "epoch": 6.252079334612924,
      "grad_norm": 0.9136100356399508,
      "learning_rate": 6.506020089127277e-06,
      "loss": 0.0597,
      "step": 9772
    },
    {
      "epoch": 6.252719129878439,
      "grad_norm": 0.802499727307561,
      "learning_rate": 6.50407860889948e-06,
      "loss": 0.0524,
      "step": 9773
    },
    {
      "epoch": 6.253358925143954,
      "grad_norm": 0.8627803463486741,
      "learning_rate": 6.502137278780325e-06,
      "loss": 0.0545,
      "step": 9774
    },
    {
      "epoch": 6.253998720409469,
      "grad_norm": 0.8475058407603762,
      "learning_rate": 6.5001960988531765e-06,
      "loss": 0.0497,
      "step": 9775
    },
    {
      "epoch": 6.254638515674984,
      "grad_norm": 0.9911793495932578,
      "learning_rate": 6.4982550692013826e-06,
      "loss": 0.054,
      "step": 9776
    },
    {
      "epoch": 6.255278310940499,
      "grad_norm": 0.8258468222762021,
      "learning_rate": 6.496314189908284e-06,
      "loss": 0.0524,
      "step": 9777
    },
    {
      "epoch": 6.255918106206014,
      "grad_norm": 0.9704130784213728,
      "learning_rate": 6.494373461057224e-06,
      "loss": 0.0583,
      "step": 9778
    },
    {
      "epoch": 6.256557901471529,
      "grad_norm": 0.8696089635600228,
      "learning_rate": 6.492432882731532e-06,
      "loss": 0.0524,
      "step": 9779
    },
    {
      "epoch": 6.257197696737044,
      "grad_norm": 0.8769489580115257,
      "learning_rate": 6.490492455014532e-06,
      "loss": 0.0467,
      "step": 9780
    },
    {
      "epoch": 6.257837492002559,
      "grad_norm": 0.8426938569533543,
      "learning_rate": 6.488552177989537e-06,
      "loss": 0.0503,
      "step": 9781
    },
    {
      "epoch": 6.258477287268074,
      "grad_norm": 0.8673567171066547,
      "learning_rate": 6.486612051739872e-06,
      "loss": 0.052,
      "step": 9782
    },
    {
      "epoch": 6.25911708253359,
      "grad_norm": 0.8410592084622063,
      "learning_rate": 6.484672076348834e-06,
      "loss": 0.0467,
      "step": 9783
    },
    {
      "epoch": 6.259756877799104,
      "grad_norm": 0.7748451035007816,
      "learning_rate": 6.4827322518997235e-06,
      "loss": 0.0474,
      "step": 9784
    },
    {
      "epoch": 6.260396673064619,
      "grad_norm": 0.8135893450393716,
      "learning_rate": 6.480792578475834e-06,
      "loss": 0.0488,
      "step": 9785
    },
    {
      "epoch": 6.2610364683301345,
      "grad_norm": 0.8452356318549739,
      "learning_rate": 6.4788530561604525e-06,
      "loss": 0.0557,
      "step": 9786
    },
    {
      "epoch": 6.26167626359565,
      "grad_norm": 0.7933931289293626,
      "learning_rate": 6.476913685036852e-06,
      "loss": 0.0488,
      "step": 9787
    },
    {
      "epoch": 6.262316058861164,
      "grad_norm": 0.7928186372419588,
      "learning_rate": 6.474974465188315e-06,
      "loss": 0.0469,
      "step": 9788
    },
    {
      "epoch": 6.262955854126679,
      "grad_norm": 0.7731334862721171,
      "learning_rate": 6.473035396698104e-06,
      "loss": 0.0456,
      "step": 9789
    },
    {
      "epoch": 6.263595649392195,
      "grad_norm": 0.9869249176662764,
      "learning_rate": 6.471096479649479e-06,
      "loss": 0.0555,
      "step": 9790
    },
    {
      "epoch": 6.264235444657709,
      "grad_norm": 0.8772482480277185,
      "learning_rate": 6.469157714125691e-06,
      "loss": 0.0462,
      "step": 9791
    },
    {
      "epoch": 6.264875239923224,
      "grad_norm": 0.8941016457536102,
      "learning_rate": 6.467219100209993e-06,
      "loss": 0.0501,
      "step": 9792
    },
    {
      "epoch": 6.26551503518874,
      "grad_norm": 0.8747999935540361,
      "learning_rate": 6.465280637985622e-06,
      "loss": 0.0584,
      "step": 9793
    },
    {
      "epoch": 6.266154830454255,
      "grad_norm": 0.8592342293885117,
      "learning_rate": 6.463342327535809e-06,
      "loss": 0.0533,
      "step": 9794
    },
    {
      "epoch": 6.266794625719769,
      "grad_norm": 0.9892668645180289,
      "learning_rate": 6.461404168943788e-06,
      "loss": 0.0531,
      "step": 9795
    },
    {
      "epoch": 6.267434420985285,
      "grad_norm": 0.8641153110682726,
      "learning_rate": 6.459466162292778e-06,
      "loss": 0.0519,
      "step": 9796
    },
    {
      "epoch": 6.2680742162508,
      "grad_norm": 0.792298176000867,
      "learning_rate": 6.457528307665991e-06,
      "loss": 0.0494,
      "step": 9797
    },
    {
      "epoch": 6.268714011516315,
      "grad_norm": 0.8727809505844778,
      "learning_rate": 6.4555906051466385e-06,
      "loss": 0.0537,
      "step": 9798
    },
    {
      "epoch": 6.2693538067818295,
      "grad_norm": 0.8457552462912347,
      "learning_rate": 6.453653054817921e-06,
      "loss": 0.0452,
      "step": 9799
    },
    {
      "epoch": 6.269993602047345,
      "grad_norm": 0.7575662119489694,
      "learning_rate": 6.451715656763033e-06,
      "loss": 0.0516,
      "step": 9800
    },
    {
      "epoch": 6.27063339731286,
      "grad_norm": 0.850605940196718,
      "learning_rate": 6.449778411065158e-06,
      "loss": 0.0558,
      "step": 9801
    },
    {
      "epoch": 6.271273192578375,
      "grad_norm": 0.8323874334438296,
      "learning_rate": 6.447841317807487e-06,
      "loss": 0.0461,
      "step": 9802
    },
    {
      "epoch": 6.27191298784389,
      "grad_norm": 0.8181072800801336,
      "learning_rate": 6.445904377073191e-06,
      "loss": 0.0503,
      "step": 9803
    },
    {
      "epoch": 6.272552783109405,
      "grad_norm": 0.9039021080905074,
      "learning_rate": 6.4439675889454355e-06,
      "loss": 0.0519,
      "step": 9804
    },
    {
      "epoch": 6.27319257837492,
      "grad_norm": 0.9084611494883195,
      "learning_rate": 6.44203095350739e-06,
      "loss": 0.0512,
      "step": 9805
    },
    {
      "epoch": 6.273832373640435,
      "grad_norm": 0.9094041820820417,
      "learning_rate": 6.440094470842205e-06,
      "loss": 0.0537,
      "step": 9806
    },
    {
      "epoch": 6.27447216890595,
      "grad_norm": 0.8232218428910124,
      "learning_rate": 6.438158141033029e-06,
      "loss": 0.0479,
      "step": 9807
    },
    {
      "epoch": 6.275111964171465,
      "grad_norm": 0.8156399208923157,
      "learning_rate": 6.43622196416301e-06,
      "loss": 0.0529,
      "step": 9808
    },
    {
      "epoch": 6.2757517594369805,
      "grad_norm": 0.9413639573573533,
      "learning_rate": 6.434285940315279e-06,
      "loss": 0.0574,
      "step": 9809
    },
    {
      "epoch": 6.276391554702495,
      "grad_norm": 0.9261154538510149,
      "learning_rate": 6.432350069572968e-06,
      "loss": 0.0512,
      "step": 9810
    },
    {
      "epoch": 6.27703134996801,
      "grad_norm": 0.9956628769604491,
      "learning_rate": 6.430414352019193e-06,
      "loss": 0.0525,
      "step": 9811
    },
    {
      "epoch": 6.277671145233525,
      "grad_norm": 0.7953888386705807,
      "learning_rate": 6.42847878773708e-06,
      "loss": 0.0529,
      "step": 9812
    },
    {
      "epoch": 6.278310940499041,
      "grad_norm": 0.7603487140075702,
      "learning_rate": 6.426543376809736e-06,
      "loss": 0.0516,
      "step": 9813
    },
    {
      "epoch": 6.278950735764555,
      "grad_norm": 0.8051931228235261,
      "learning_rate": 6.42460811932026e-06,
      "loss": 0.0513,
      "step": 9814
    },
    {
      "epoch": 6.27959053103007,
      "grad_norm": 1.0057193333243124,
      "learning_rate": 6.422673015351754e-06,
      "loss": 0.0539,
      "step": 9815
    },
    {
      "epoch": 6.280230326295586,
      "grad_norm": 0.8268923912613396,
      "learning_rate": 6.420738064987304e-06,
      "loss": 0.0519,
      "step": 9816
    },
    {
      "epoch": 6.280870121561101,
      "grad_norm": 0.8576794100773085,
      "learning_rate": 6.4188032683099944e-06,
      "loss": 0.0612,
      "step": 9817
    },
    {
      "epoch": 6.281509916826615,
      "grad_norm": 0.8947692812500114,
      "learning_rate": 6.416868625402899e-06,
      "loss": 0.0516,
      "step": 9818
    },
    {
      "epoch": 6.282149712092131,
      "grad_norm": 0.9547198363453553,
      "learning_rate": 6.414934136349095e-06,
      "loss": 0.0594,
      "step": 9819
    },
    {
      "epoch": 6.282789507357646,
      "grad_norm": 0.8338750289872612,
      "learning_rate": 6.4129998012316405e-06,
      "loss": 0.0482,
      "step": 9820
    },
    {
      "epoch": 6.28342930262316,
      "grad_norm": 0.8424004090500756,
      "learning_rate": 6.411065620133593e-06,
      "loss": 0.0513,
      "step": 9821
    },
    {
      "epoch": 6.2840690978886755,
      "grad_norm": 0.8721255692879009,
      "learning_rate": 6.409131593138005e-06,
      "loss": 0.0541,
      "step": 9822
    },
    {
      "epoch": 6.284708893154191,
      "grad_norm": 0.9231864104232652,
      "learning_rate": 6.407197720327919e-06,
      "loss": 0.0567,
      "step": 9823
    },
    {
      "epoch": 6.285348688419706,
      "grad_norm": 0.9567083616302295,
      "learning_rate": 6.405264001786366e-06,
      "loss": 0.0599,
      "step": 9824
    },
    {
      "epoch": 6.2859884836852205,
      "grad_norm": 0.8123588161660013,
      "learning_rate": 6.403330437596389e-06,
      "loss": 0.0521,
      "step": 9825
    },
    {
      "epoch": 6.286628278950736,
      "grad_norm": 0.8251117762296677,
      "learning_rate": 6.401397027841004e-06,
      "loss": 0.06,
      "step": 9826
    },
    {
      "epoch": 6.287268074216251,
      "grad_norm": 0.8321056541632116,
      "learning_rate": 6.3994637726032295e-06,
      "loss": 0.0505,
      "step": 9827
    },
    {
      "epoch": 6.287907869481765,
      "grad_norm": 0.7961337411993824,
      "learning_rate": 6.3975306719660726e-06,
      "loss": 0.0473,
      "step": 9828
    },
    {
      "epoch": 6.288547664747281,
      "grad_norm": 0.796251557062966,
      "learning_rate": 6.3955977260125435e-06,
      "loss": 0.0497,
      "step": 9829
    },
    {
      "epoch": 6.289187460012796,
      "grad_norm": 0.9065644437798669,
      "learning_rate": 6.393664934825636e-06,
      "loss": 0.0521,
      "step": 9830
    },
    {
      "epoch": 6.289827255278311,
      "grad_norm": 0.8898018416425441,
      "learning_rate": 6.391732298488337e-06,
      "loss": 0.0511,
      "step": 9831
    },
    {
      "epoch": 6.290467050543826,
      "grad_norm": 0.929879148385523,
      "learning_rate": 6.3897998170836384e-06,
      "loss": 0.0515,
      "step": 9832
    },
    {
      "epoch": 6.291106845809341,
      "grad_norm": 0.8731085177606243,
      "learning_rate": 6.387867490694513e-06,
      "loss": 0.0552,
      "step": 9833
    },
    {
      "epoch": 6.291746641074856,
      "grad_norm": 0.8115872296578889,
      "learning_rate": 6.385935319403929e-06,
      "loss": 0.0529,
      "step": 9834
    },
    {
      "epoch": 6.292386436340371,
      "grad_norm": 0.7619212251746305,
      "learning_rate": 6.384003303294856e-06,
      "loss": 0.049,
      "step": 9835
    },
    {
      "epoch": 6.293026231605886,
      "grad_norm": 0.7498649874080069,
      "learning_rate": 6.38207144245025e-06,
      "loss": 0.0441,
      "step": 9836
    },
    {
      "epoch": 6.293666026871401,
      "grad_norm": 0.8297081524328799,
      "learning_rate": 6.380139736953061e-06,
      "loss": 0.047,
      "step": 9837
    },
    {
      "epoch": 6.294305822136916,
      "grad_norm": 0.8782332420668413,
      "learning_rate": 6.37820818688623e-06,
      "loss": 0.0517,
      "step": 9838
    },
    {
      "epoch": 6.294945617402432,
      "grad_norm": 0.8758079813521527,
      "learning_rate": 6.376276792332698e-06,
      "loss": 0.0483,
      "step": 9839
    },
    {
      "epoch": 6.295585412667946,
      "grad_norm": 0.8101762924867907,
      "learning_rate": 6.374345553375396e-06,
      "loss": 0.0458,
      "step": 9840
    },
    {
      "epoch": 6.296225207933461,
      "grad_norm": 0.8995234978619341,
      "learning_rate": 6.3724144700972414e-06,
      "loss": 0.0509,
      "step": 9841
    },
    {
      "epoch": 6.296865003198977,
      "grad_norm": 0.7864790301200503,
      "learning_rate": 6.37048354258116e-06,
      "loss": 0.0491,
      "step": 9842
    },
    {
      "epoch": 6.297504798464491,
      "grad_norm": 0.7735093544843843,
      "learning_rate": 6.3685527709100605e-06,
      "loss": 0.052,
      "step": 9843
    },
    {
      "epoch": 6.298144593730006,
      "grad_norm": 0.9054795949571148,
      "learning_rate": 6.366622155166842e-06,
      "loss": 0.0502,
      "step": 9844
    },
    {
      "epoch": 6.2987843889955215,
      "grad_norm": 0.8081852594084542,
      "learning_rate": 6.364691695434406e-06,
      "loss": 0.0558,
      "step": 9845
    },
    {
      "epoch": 6.299424184261037,
      "grad_norm": 0.8703665589581006,
      "learning_rate": 6.362761391795644e-06,
      "loss": 0.0527,
      "step": 9846
    },
    {
      "epoch": 6.300063979526551,
      "grad_norm": 0.8512111242470716,
      "learning_rate": 6.360831244333437e-06,
      "loss": 0.051,
      "step": 9847
    },
    {
      "epoch": 6.3007037747920664,
      "grad_norm": 0.8725677213157839,
      "learning_rate": 6.358901253130658e-06,
      "loss": 0.0506,
      "step": 9848
    },
    {
      "epoch": 6.301343570057582,
      "grad_norm": 0.8970705403233351,
      "learning_rate": 6.3569714182701855e-06,
      "loss": 0.0524,
      "step": 9849
    },
    {
      "epoch": 6.301983365323097,
      "grad_norm": 0.897882794075581,
      "learning_rate": 6.3550417398348814e-06,
      "loss": 0.0495,
      "step": 9850
    },
    {
      "epoch": 6.302623160588611,
      "grad_norm": 0.880772254510672,
      "learning_rate": 6.353112217907599e-06,
      "loss": 0.0561,
      "step": 9851
    },
    {
      "epoch": 6.303262955854127,
      "grad_norm": 0.896868126558404,
      "learning_rate": 6.351182852571192e-06,
      "loss": 0.0507,
      "step": 9852
    },
    {
      "epoch": 6.303902751119642,
      "grad_norm": 0.8295842710410598,
      "learning_rate": 6.349253643908503e-06,
      "loss": 0.0471,
      "step": 9853
    },
    {
      "epoch": 6.304542546385157,
      "grad_norm": 0.8541910450540949,
      "learning_rate": 6.347324592002364e-06,
      "loss": 0.0497,
      "step": 9854
    },
    {
      "epoch": 6.305182341650672,
      "grad_norm": 0.7780617583149516,
      "learning_rate": 6.345395696935613e-06,
      "loss": 0.0485,
      "step": 9855
    },
    {
      "epoch": 6.305822136916187,
      "grad_norm": 0.7693749540421307,
      "learning_rate": 6.343466958791071e-06,
      "loss": 0.0497,
      "step": 9856
    },
    {
      "epoch": 6.306461932181702,
      "grad_norm": 0.7516841312135942,
      "learning_rate": 6.341538377651553e-06,
      "loss": 0.0447,
      "step": 9857
    },
    {
      "epoch": 6.3071017274472165,
      "grad_norm": 0.7894682526895499,
      "learning_rate": 6.3396099535998664e-06,
      "loss": 0.0519,
      "step": 9858
    },
    {
      "epoch": 6.307741522712732,
      "grad_norm": 0.9134303905816915,
      "learning_rate": 6.3376816867188194e-06,
      "loss": 0.0549,
      "step": 9859
    },
    {
      "epoch": 6.308381317978247,
      "grad_norm": 0.9343494861426938,
      "learning_rate": 6.335753577091206e-06,
      "loss": 0.0546,
      "step": 9860
    },
    {
      "epoch": 6.309021113243762,
      "grad_norm": 0.8960080939046401,
      "learning_rate": 6.333825624799812e-06,
      "loss": 0.0577,
      "step": 9861
    },
    {
      "epoch": 6.309660908509277,
      "grad_norm": 0.865766864455,
      "learning_rate": 6.331897829927428e-06,
      "loss": 0.053,
      "step": 9862
    },
    {
      "epoch": 6.310300703774792,
      "grad_norm": 0.9504955121318802,
      "learning_rate": 6.329970192556827e-06,
      "loss": 0.0495,
      "step": 9863
    },
    {
      "epoch": 6.310940499040307,
      "grad_norm": 0.8678713969311508,
      "learning_rate": 6.328042712770772e-06,
      "loss": 0.0564,
      "step": 9864
    },
    {
      "epoch": 6.311580294305822,
      "grad_norm": 0.8271519582485285,
      "learning_rate": 6.3261153906520345e-06,
      "loss": 0.0577,
      "step": 9865
    },
    {
      "epoch": 6.312220089571337,
      "grad_norm": 0.7758082780924599,
      "learning_rate": 6.324188226283369e-06,
      "loss": 0.0509,
      "step": 9866
    },
    {
      "epoch": 6.312859884836852,
      "grad_norm": 0.7534549523475632,
      "learning_rate": 6.322261219747521e-06,
      "loss": 0.0489,
      "step": 9867
    },
    {
      "epoch": 6.3134996801023675,
      "grad_norm": 0.9267148566978936,
      "learning_rate": 6.3203343711272325e-06,
      "loss": 0.0549,
      "step": 9868
    },
    {
      "epoch": 6.314139475367882,
      "grad_norm": 0.9073440323911878,
      "learning_rate": 6.318407680505242e-06,
      "loss": 0.0594,
      "step": 9869
    },
    {
      "epoch": 6.314779270633397,
      "grad_norm": 0.8277806911068945,
      "learning_rate": 6.3164811479642775e-06,
      "loss": 0.0509,
      "step": 9870
    },
    {
      "epoch": 6.315419065898912,
      "grad_norm": 0.9280878925840154,
      "learning_rate": 6.314554773587054e-06,
      "loss": 0.0585,
      "step": 9871
    },
    {
      "epoch": 6.316058861164428,
      "grad_norm": 0.7814559715447762,
      "learning_rate": 6.312628557456299e-06,
      "loss": 0.0507,
      "step": 9872
    },
    {
      "epoch": 6.316698656429942,
      "grad_norm": 0.7659811300985206,
      "learning_rate": 6.310702499654713e-06,
      "loss": 0.0494,
      "step": 9873
    },
    {
      "epoch": 6.317338451695457,
      "grad_norm": 0.7779040498408594,
      "learning_rate": 6.308776600264996e-06,
      "loss": 0.0514,
      "step": 9874
    },
    {
      "epoch": 6.317978246960973,
      "grad_norm": 0.8591984974337749,
      "learning_rate": 6.3068508593698504e-06,
      "loss": 0.0574,
      "step": 9875
    },
    {
      "epoch": 6.318618042226488,
      "grad_norm": 0.7976839291825809,
      "learning_rate": 6.3049252770519575e-06,
      "loss": 0.0529,
      "step": 9876
    },
    {
      "epoch": 6.319257837492002,
      "grad_norm": 0.7721934567477646,
      "learning_rate": 6.302999853394001e-06,
      "loss": 0.0491,
      "step": 9877
    },
    {
      "epoch": 6.319897632757518,
      "grad_norm": 0.8763257396785687,
      "learning_rate": 6.301074588478649e-06,
      "loss": 0.051,
      "step": 9878
    },
    {
      "epoch": 6.320537428023033,
      "grad_norm": 0.8798290007845068,
      "learning_rate": 6.29914948238858e-06,
      "loss": 0.0546,
      "step": 9879
    },
    {
      "epoch": 6.321177223288547,
      "grad_norm": 0.9081452216670675,
      "learning_rate": 6.2972245352064475e-06,
      "loss": 0.0532,
      "step": 9880
    },
    {
      "epoch": 6.3218170185540625,
      "grad_norm": 0.7284383990809816,
      "learning_rate": 6.295299747014905e-06,
      "loss": 0.0468,
      "step": 9881
    },
    {
      "epoch": 6.322456813819578,
      "grad_norm": 0.7929130527886751,
      "learning_rate": 6.293375117896604e-06,
      "loss": 0.0485,
      "step": 9882
    },
    {
      "epoch": 6.323096609085093,
      "grad_norm": 1.0018177155863925,
      "learning_rate": 6.291450647934181e-06,
      "loss": 0.0615,
      "step": 9883
    },
    {
      "epoch": 6.3237364043506075,
      "grad_norm": 0.8490960817315784,
      "learning_rate": 6.289526337210266e-06,
      "loss": 0.0509,
      "step": 9884
    },
    {
      "epoch": 6.324376199616123,
      "grad_norm": 0.8382465879709302,
      "learning_rate": 6.2876021858074955e-06,
      "loss": 0.0458,
      "step": 9885
    },
    {
      "epoch": 6.325015994881638,
      "grad_norm": 0.798219174113772,
      "learning_rate": 6.285678193808482e-06,
      "loss": 0.0492,
      "step": 9886
    },
    {
      "epoch": 6.325655790147153,
      "grad_norm": 0.875197185852495,
      "learning_rate": 6.28375436129584e-06,
      "loss": 0.0534,
      "step": 9887
    },
    {
      "epoch": 6.326295585412668,
      "grad_norm": 0.9149382233511026,
      "learning_rate": 6.281830688352172e-06,
      "loss": 0.0555,
      "step": 9888
    },
    {
      "epoch": 6.326935380678183,
      "grad_norm": 0.7765841766292961,
      "learning_rate": 6.279907175060083e-06,
      "loss": 0.046,
      "step": 9889
    },
    {
      "epoch": 6.327575175943698,
      "grad_norm": 0.7790225600938719,
      "learning_rate": 6.277983821502161e-06,
      "loss": 0.0478,
      "step": 9890
    },
    {
      "epoch": 6.3282149712092135,
      "grad_norm": 0.9680209697786418,
      "learning_rate": 6.27606062776099e-06,
      "loss": 0.0574,
      "step": 9891
    },
    {
      "epoch": 6.328854766474728,
      "grad_norm": 0.7955154327194858,
      "learning_rate": 6.274137593919156e-06,
      "loss": 0.0492,
      "step": 9892
    },
    {
      "epoch": 6.329494561740243,
      "grad_norm": 0.895000883602678,
      "learning_rate": 6.2722147200592244e-06,
      "loss": 0.0542,
      "step": 9893
    },
    {
      "epoch": 6.330134357005758,
      "grad_norm": 0.7590728234030061,
      "learning_rate": 6.270292006263758e-06,
      "loss": 0.049,
      "step": 9894
    },
    {
      "epoch": 6.330774152271273,
      "grad_norm": 0.7996736339988705,
      "learning_rate": 6.2683694526153226e-06,
      "loss": 0.0549,
      "step": 9895
    },
    {
      "epoch": 6.331413947536788,
      "grad_norm": 0.7592892048598052,
      "learning_rate": 6.2664470591964655e-06,
      "loss": 0.0502,
      "step": 9896
    },
    {
      "epoch": 6.332053742802303,
      "grad_norm": 0.7012576390736517,
      "learning_rate": 6.264524826089729e-06,
      "loss": 0.0442,
      "step": 9897
    },
    {
      "epoch": 6.332693538067819,
      "grad_norm": 0.8508696230372101,
      "learning_rate": 6.262602753377649e-06,
      "loss": 0.0518,
      "step": 9898
    },
    {
      "epoch": 6.333333333333333,
      "grad_norm": 0.8032039094538808,
      "learning_rate": 6.260680841142762e-06,
      "loss": 0.0454,
      "step": 9899
    },
    {
      "epoch": 6.333973128598848,
      "grad_norm": 0.7950429902291037,
      "learning_rate": 6.258759089467588e-06,
      "loss": 0.0459,
      "step": 9900
    },
    {
      "epoch": 6.334612923864364,
      "grad_norm": 0.7379398704563743,
      "learning_rate": 6.256837498434638e-06,
      "loss": 0.0518,
      "step": 9901
    },
    {
      "epoch": 6.335252719129878,
      "grad_norm": 0.8097110805363041,
      "learning_rate": 6.2549160681264335e-06,
      "loss": 0.048,
      "step": 9902
    },
    {
      "epoch": 6.335892514395393,
      "grad_norm": 0.8657527617979148,
      "learning_rate": 6.252994798625469e-06,
      "loss": 0.0534,
      "step": 9903
    },
    {
      "epoch": 6.3365323096609085,
      "grad_norm": 0.8925873712255274,
      "learning_rate": 6.251073690014242e-06,
      "loss": 0.0566,
      "step": 9904
    },
    {
      "epoch": 6.337172104926424,
      "grad_norm": 0.7893339721681871,
      "learning_rate": 6.249152742375242e-06,
      "loss": 0.0543,
      "step": 9905
    },
    {
      "epoch": 6.337811900191938,
      "grad_norm": 0.9229677208694371,
      "learning_rate": 6.247231955790951e-06,
      "loss": 0.0581,
      "step": 9906
    },
    {
      "epoch": 6.3384516954574535,
      "grad_norm": 0.7266130842762456,
      "learning_rate": 6.245311330343846e-06,
      "loss": 0.0454,
      "step": 9907
    },
    {
      "epoch": 6.339091490722969,
      "grad_norm": 0.7969798403067765,
      "learning_rate": 6.243390866116388e-06,
      "loss": 0.0466,
      "step": 9908
    },
    {
      "epoch": 6.339731285988484,
      "grad_norm": 0.8657755810699259,
      "learning_rate": 6.241470563191049e-06,
      "loss": 0.0558,
      "step": 9909
    },
    {
      "epoch": 6.340371081253998,
      "grad_norm": 0.7831710038799873,
      "learning_rate": 6.239550421650278e-06,
      "loss": 0.049,
      "step": 9910
    },
    {
      "epoch": 6.341010876519514,
      "grad_norm": 0.8033972829497537,
      "learning_rate": 6.2376304415765186e-06,
      "loss": 0.0449,
      "step": 9911
    },
    {
      "epoch": 6.341650671785029,
      "grad_norm": 0.8738666138789714,
      "learning_rate": 6.23571062305222e-06,
      "loss": 0.0567,
      "step": 9912
    },
    {
      "epoch": 6.342290467050544,
      "grad_norm": 0.724239657223808,
      "learning_rate": 6.23379096615981e-06,
      "loss": 0.0526,
      "step": 9913
    },
    {
      "epoch": 6.342930262316059,
      "grad_norm": 0.8502987573881439,
      "learning_rate": 6.231871470981716e-06,
      "loss": 0.0572,
      "step": 9914
    },
    {
      "epoch": 6.343570057581574,
      "grad_norm": 0.8085063545675294,
      "learning_rate": 6.229952137600353e-06,
      "loss": 0.0518,
      "step": 9915
    },
    {
      "epoch": 6.344209852847089,
      "grad_norm": 0.7862943951244707,
      "learning_rate": 6.2280329660981456e-06,
      "loss": 0.0503,
      "step": 9916
    },
    {
      "epoch": 6.3448496481126035,
      "grad_norm": 0.9306683398847968,
      "learning_rate": 6.22611395655749e-06,
      "loss": 0.0477,
      "step": 9917
    },
    {
      "epoch": 6.345489443378119,
      "grad_norm": 0.8391091063855696,
      "learning_rate": 6.224195109060788e-06,
      "loss": 0.0575,
      "step": 9918
    },
    {
      "epoch": 6.346129238643634,
      "grad_norm": 0.8798587844543075,
      "learning_rate": 6.222276423690432e-06,
      "loss": 0.0483,
      "step": 9919
    },
    {
      "epoch": 6.346769033909149,
      "grad_norm": 0.9821318310608825,
      "learning_rate": 6.220357900528806e-06,
      "loss": 0.0577,
      "step": 9920
    },
    {
      "epoch": 6.347408829174664,
      "grad_norm": 0.8063218308749668,
      "learning_rate": 6.218439539658285e-06,
      "loss": 0.0451,
      "step": 9921
    },
    {
      "epoch": 6.348048624440179,
      "grad_norm": 0.7516094512025655,
      "learning_rate": 6.216521341161248e-06,
      "loss": 0.0441,
      "step": 9922
    },
    {
      "epoch": 6.348688419705694,
      "grad_norm": 0.9332414689429281,
      "learning_rate": 6.214603305120055e-06,
      "loss": 0.0507,
      "step": 9923
    },
    {
      "epoch": 6.34932821497121,
      "grad_norm": 0.9096387338868968,
      "learning_rate": 6.21268543161706e-06,
      "loss": 0.0477,
      "step": 9924
    },
    {
      "epoch": 6.349968010236724,
      "grad_norm": 0.7950650311847225,
      "learning_rate": 6.210767720734613e-06,
      "loss": 0.0425,
      "step": 9925
    },
    {
      "epoch": 6.350607805502239,
      "grad_norm": 0.8541193998253572,
      "learning_rate": 6.208850172555063e-06,
      "loss": 0.0528,
      "step": 9926
    },
    {
      "epoch": 6.3512476007677545,
      "grad_norm": 1.0488287662071307,
      "learning_rate": 6.206932787160744e-06,
      "loss": 0.0606,
      "step": 9927
    },
    {
      "epoch": 6.35188739603327,
      "grad_norm": 0.8412157707401672,
      "learning_rate": 6.20501556463398e-06,
      "loss": 0.051,
      "step": 9928
    },
    {
      "epoch": 6.352527191298784,
      "grad_norm": 0.891027422314959,
      "learning_rate": 6.203098505057099e-06,
      "loss": 0.0503,
      "step": 9929
    },
    {
      "epoch": 6.3531669865642995,
      "grad_norm": 0.943279552232496,
      "learning_rate": 6.2011816085124145e-06,
      "loss": 0.0476,
      "step": 9930
    },
    {
      "epoch": 6.353806781829815,
      "grad_norm": 0.908274064317826,
      "learning_rate": 6.199264875082229e-06,
      "loss": 0.0598,
      "step": 9931
    },
    {
      "epoch": 6.354446577095329,
      "grad_norm": 0.95923638245346,
      "learning_rate": 6.197348304848855e-06,
      "loss": 0.0572,
      "step": 9932
    },
    {
      "epoch": 6.355086372360844,
      "grad_norm": 0.9026025325378219,
      "learning_rate": 6.1954318978945794e-06,
      "loss": 0.0481,
      "step": 9933
    },
    {
      "epoch": 6.35572616762636,
      "grad_norm": 0.8522113025772463,
      "learning_rate": 6.19351565430169e-06,
      "loss": 0.0536,
      "step": 9934
    },
    {
      "epoch": 6.356365962891875,
      "grad_norm": 0.9392561943060238,
      "learning_rate": 6.191599574152465e-06,
      "loss": 0.0541,
      "step": 9935
    },
    {
      "epoch": 6.357005758157389,
      "grad_norm": 0.7817076993949379,
      "learning_rate": 6.189683657529182e-06,
      "loss": 0.0547,
      "step": 9936
    },
    {
      "epoch": 6.357645553422905,
      "grad_norm": 0.8597878977867159,
      "learning_rate": 6.187767904514106e-06,
      "loss": 0.0513,
      "step": 9937
    },
    {
      "epoch": 6.35828534868842,
      "grad_norm": 0.8608644557521139,
      "learning_rate": 6.185852315189489e-06,
      "loss": 0.0522,
      "step": 9938
    },
    {
      "epoch": 6.358925143953935,
      "grad_norm": 0.8612747713218383,
      "learning_rate": 6.183936889637593e-06,
      "loss": 0.0585,
      "step": 9939
    },
    {
      "epoch": 6.3595649392194495,
      "grad_norm": 0.9601082610447486,
      "learning_rate": 6.18202162794066e-06,
      "loss": 0.0501,
      "step": 9940
    },
    {
      "epoch": 6.360204734484965,
      "grad_norm": 0.8951626062152795,
      "learning_rate": 6.180106530180923e-06,
      "loss": 0.0514,
      "step": 9941
    },
    {
      "epoch": 6.36084452975048,
      "grad_norm": 0.9066395747154082,
      "learning_rate": 6.178191596440619e-06,
      "loss": 0.0585,
      "step": 9942
    },
    {
      "epoch": 6.3614843250159945,
      "grad_norm": 0.8905039283659315,
      "learning_rate": 6.17627682680197e-06,
      "loss": 0.0533,
      "step": 9943
    },
    {
      "epoch": 6.36212412028151,
      "grad_norm": 0.7617441026127846,
      "learning_rate": 6.174362221347192e-06,
      "loss": 0.0506,
      "step": 9944
    },
    {
      "epoch": 6.362763915547025,
      "grad_norm": 0.8910496033746602,
      "learning_rate": 6.1724477801584905e-06,
      "loss": 0.0544,
      "step": 9945
    },
    {
      "epoch": 6.36340371081254,
      "grad_norm": 0.8454403420900378,
      "learning_rate": 6.170533503318077e-06,
      "loss": 0.0499,
      "step": 9946
    },
    {
      "epoch": 6.364043506078055,
      "grad_norm": 0.7277697721757524,
      "learning_rate": 6.168619390908141e-06,
      "loss": 0.0509,
      "step": 9947
    },
    {
      "epoch": 6.36468330134357,
      "grad_norm": 0.7615344476631855,
      "learning_rate": 6.166705443010873e-06,
      "loss": 0.0488,
      "step": 9948
    },
    {
      "epoch": 6.365323096609085,
      "grad_norm": 0.8147303662884018,
      "learning_rate": 6.164791659708456e-06,
      "loss": 0.05,
      "step": 9949
    },
    {
      "epoch": 6.3659628918746005,
      "grad_norm": 0.8168736284842997,
      "learning_rate": 6.162878041083063e-06,
      "loss": 0.0472,
      "step": 9950
    },
    {
      "epoch": 6.366602687140115,
      "grad_norm": 0.8606682939780901,
      "learning_rate": 6.160964587216857e-06,
      "loss": 0.0515,
      "step": 9951
    },
    {
      "epoch": 6.36724248240563,
      "grad_norm": 0.8395657618277707,
      "learning_rate": 6.159051298192006e-06,
      "loss": 0.0584,
      "step": 9952
    },
    {
      "epoch": 6.3678822776711455,
      "grad_norm": 0.7726428260914188,
      "learning_rate": 6.15713817409066e-06,
      "loss": 0.0425,
      "step": 9953
    },
    {
      "epoch": 6.36852207293666,
      "grad_norm": 0.87229102311083,
      "learning_rate": 6.155225214994964e-06,
      "loss": 0.0461,
      "step": 9954
    },
    {
      "epoch": 6.369161868202175,
      "grad_norm": 0.9212242716121216,
      "learning_rate": 6.1533124209870566e-06,
      "loss": 0.0548,
      "step": 9955
    },
    {
      "epoch": 6.36980166346769,
      "grad_norm": 0.9153107962032141,
      "learning_rate": 6.1513997921490754e-06,
      "loss": 0.0527,
      "step": 9956
    },
    {
      "epoch": 6.370441458733206,
      "grad_norm": 0.9210785768560846,
      "learning_rate": 6.149487328563139e-06,
      "loss": 0.0554,
      "step": 9957
    },
    {
      "epoch": 6.37108125399872,
      "grad_norm": 0.7801202775014925,
      "learning_rate": 6.147575030311362e-06,
      "loss": 0.052,
      "step": 9958
    },
    {
      "epoch": 6.371721049264235,
      "grad_norm": 0.7952531690927462,
      "learning_rate": 6.1456628974758655e-06,
      "loss": 0.048,
      "step": 9959
    },
    {
      "epoch": 6.372360844529751,
      "grad_norm": 0.8661929373876285,
      "learning_rate": 6.143750930138747e-06,
      "loss": 0.0477,
      "step": 9960
    },
    {
      "epoch": 6.373000639795266,
      "grad_norm": 0.7871448584600992,
      "learning_rate": 6.141839128382102e-06,
      "loss": 0.0458,
      "step": 9961
    },
    {
      "epoch": 6.37364043506078,
      "grad_norm": 0.7737154317526236,
      "learning_rate": 6.139927492288025e-06,
      "loss": 0.0551,
      "step": 9962
    },
    {
      "epoch": 6.3742802303262955,
      "grad_norm": 1.0008726256785738,
      "learning_rate": 6.138016021938593e-06,
      "loss": 0.0495,
      "step": 9963
    },
    {
      "epoch": 6.374920025591811,
      "grad_norm": 0.8772046204894085,
      "learning_rate": 6.136104717415885e-06,
      "loss": 0.0532,
      "step": 9964
    },
    {
      "epoch": 6.375559820857326,
      "grad_norm": 0.8400482570882841,
      "learning_rate": 6.134193578801965e-06,
      "loss": 0.05,
      "step": 9965
    },
    {
      "epoch": 6.3761996161228405,
      "grad_norm": 0.728274867266892,
      "learning_rate": 6.132282606178896e-06,
      "loss": 0.0498,
      "step": 9966
    },
    {
      "epoch": 6.376839411388356,
      "grad_norm": 0.7825202898630362,
      "learning_rate": 6.130371799628734e-06,
      "loss": 0.0513,
      "step": 9967
    },
    {
      "epoch": 6.377479206653871,
      "grad_norm": 0.833521525113017,
      "learning_rate": 6.128461159233519e-06,
      "loss": 0.0545,
      "step": 9968
    },
    {
      "epoch": 6.378119001919385,
      "grad_norm": 0.8333798025578514,
      "learning_rate": 6.126550685075298e-06,
      "loss": 0.055,
      "step": 9969
    },
    {
      "epoch": 6.378758797184901,
      "grad_norm": 0.9030067316996728,
      "learning_rate": 6.124640377236102e-06,
      "loss": 0.0551,
      "step": 9970
    },
    {
      "epoch": 6.379398592450416,
      "grad_norm": 0.8112118228530127,
      "learning_rate": 6.12273023579795e-06,
      "loss": 0.0553,
      "step": 9971
    },
    {
      "epoch": 6.380038387715931,
      "grad_norm": 0.8672323910530849,
      "learning_rate": 6.120820260842868e-06,
      "loss": 0.0561,
      "step": 9972
    },
    {
      "epoch": 6.380678182981446,
      "grad_norm": 0.7799972183155238,
      "learning_rate": 6.118910452452863e-06,
      "loss": 0.053,
      "step": 9973
    },
    {
      "epoch": 6.381317978246961,
      "grad_norm": 0.7836475013783641,
      "learning_rate": 6.11700081070994e-06,
      "loss": 0.0494,
      "step": 9974
    },
    {
      "epoch": 6.381957773512476,
      "grad_norm": 0.7737749235743266,
      "learning_rate": 6.1150913356960905e-06,
      "loss": 0.0463,
      "step": 9975
    },
    {
      "epoch": 6.3825975687779914,
      "grad_norm": 0.8420851311835371,
      "learning_rate": 6.1131820274933115e-06,
      "loss": 0.0583,
      "step": 9976
    },
    {
      "epoch": 6.383237364043506,
      "grad_norm": 0.8979233390741921,
      "learning_rate": 6.111272886183584e-06,
      "loss": 0.0528,
      "step": 9977
    },
    {
      "epoch": 6.383877159309021,
      "grad_norm": 0.8015276442022289,
      "learning_rate": 6.109363911848878e-06,
      "loss": 0.0507,
      "step": 9978
    },
    {
      "epoch": 6.384516954574536,
      "grad_norm": 0.8598150864575647,
      "learning_rate": 6.107455104571166e-06,
      "loss": 0.0552,
      "step": 9979
    },
    {
      "epoch": 6.385156749840051,
      "grad_norm": 0.7647122596777864,
      "learning_rate": 6.105546464432408e-06,
      "loss": 0.0529,
      "step": 9980
    },
    {
      "epoch": 6.385796545105566,
      "grad_norm": 0.8129743307654709,
      "learning_rate": 6.103637991514557e-06,
      "loss": 0.0507,
      "step": 9981
    },
    {
      "epoch": 6.386436340371081,
      "grad_norm": 0.9257425948942868,
      "learning_rate": 6.101729685899557e-06,
      "loss": 0.0509,
      "step": 9982
    },
    {
      "epoch": 6.387076135636597,
      "grad_norm": 0.8189661056387353,
      "learning_rate": 6.0998215476693525e-06,
      "loss": 0.0437,
      "step": 9983
    },
    {
      "epoch": 6.387715930902111,
      "grad_norm": 0.7526565427204015,
      "learning_rate": 6.097913576905873e-06,
      "loss": 0.0497,
      "step": 9984
    },
    {
      "epoch": 6.388355726167626,
      "grad_norm": 0.7744243505720148,
      "learning_rate": 6.096005773691042e-06,
      "loss": 0.042,
      "step": 9985
    },
    {
      "epoch": 6.3889955214331415,
      "grad_norm": 0.859637062555763,
      "learning_rate": 6.094098138106779e-06,
      "loss": 0.0515,
      "step": 9986
    },
    {
      "epoch": 6.389635316698657,
      "grad_norm": 0.903147921251402,
      "learning_rate": 6.092190670234994e-06,
      "loss": 0.0515,
      "step": 9987
    },
    {
      "epoch": 6.390275111964171,
      "grad_norm": 0.876572386815825,
      "learning_rate": 6.090283370157588e-06,
      "loss": 0.053,
      "step": 9988
    },
    {
      "epoch": 6.3909149072296865,
      "grad_norm": 0.8761528912209666,
      "learning_rate": 6.088376237956461e-06,
      "loss": 0.0534,
      "step": 9989
    },
    {
      "epoch": 6.391554702495202,
      "grad_norm": 0.7090999996891112,
      "learning_rate": 6.086469273713502e-06,
      "loss": 0.0467,
      "step": 9990
    },
    {
      "epoch": 6.392194497760716,
      "grad_norm": 0.9011455479079006,
      "learning_rate": 6.084562477510589e-06,
      "loss": 0.0467,
      "step": 9991
    },
    {
      "epoch": 6.392834293026231,
      "grad_norm": 0.9547575803821141,
      "learning_rate": 6.082655849429596e-06,
      "loss": 0.0474,
      "step": 9992
    },
    {
      "epoch": 6.393474088291747,
      "grad_norm": 0.9045855856180646,
      "learning_rate": 6.080749389552396e-06,
      "loss": 0.0509,
      "step": 9993
    },
    {
      "epoch": 6.394113883557262,
      "grad_norm": 1.2279358648623657,
      "learning_rate": 6.078843097960845e-06,
      "loss": 0.0618,
      "step": 9994
    },
    {
      "epoch": 6.394753678822776,
      "grad_norm": 0.7600340977959604,
      "learning_rate": 6.076936974736793e-06,
      "loss": 0.0501,
      "step": 9995
    },
    {
      "epoch": 6.395393474088292,
      "grad_norm": 0.8938644920800018,
      "learning_rate": 6.0750310199620925e-06,
      "loss": 0.0532,
      "step": 9996
    },
    {
      "epoch": 6.396033269353807,
      "grad_norm": 0.8612642029896168,
      "learning_rate": 6.073125233718578e-06,
      "loss": 0.0514,
      "step": 9997
    },
    {
      "epoch": 6.396673064619322,
      "grad_norm": 0.832684069236236,
      "learning_rate": 6.071219616088076e-06,
      "loss": 0.057,
      "step": 9998
    },
    {
      "epoch": 6.3973128598848366,
      "grad_norm": 0.8250984197147162,
      "learning_rate": 6.069314167152418e-06,
      "loss": 0.0543,
      "step": 9999
    },
    {
      "epoch": 6.397952655150352,
      "grad_norm": 0.9100395066665737,
      "learning_rate": 6.067408886993418e-06,
      "loss": 0.0517,
      "step": 10000
    },
    {
      "epoch": 6.398592450415867,
      "grad_norm": 0.8449594019866407,
      "learning_rate": 6.065503775692886e-06,
      "loss": 0.0574,
      "step": 10001
    },
    {
      "epoch": 6.399232245681382,
      "grad_norm": 0.8363716017998517,
      "learning_rate": 6.06359883333262e-06,
      "loss": 0.049,
      "step": 10002
    },
    {
      "epoch": 6.399872040946897,
      "grad_norm": 0.8419282402128663,
      "learning_rate": 6.0616940599944206e-06,
      "loss": 0.0532,
      "step": 10003
    },
    {
      "epoch": 6.400511836212412,
      "grad_norm": 0.7555724784789937,
      "learning_rate": 6.059789455760071e-06,
      "loss": 0.0443,
      "step": 10004
    },
    {
      "epoch": 6.401151631477927,
      "grad_norm": 0.8325685570696192,
      "learning_rate": 6.0578850207113505e-06,
      "loss": 0.0483,
      "step": 10005
    },
    {
      "epoch": 6.401791426743442,
      "grad_norm": 0.8343426112233244,
      "learning_rate": 6.0559807549300375e-06,
      "loss": 0.0512,
      "step": 10006
    },
    {
      "epoch": 6.402431222008957,
      "grad_norm": 0.8904377666433662,
      "learning_rate": 6.054076658497896e-06,
      "loss": 0.0564,
      "step": 10007
    },
    {
      "epoch": 6.403071017274472,
      "grad_norm": 0.7894846002541952,
      "learning_rate": 6.052172731496681e-06,
      "loss": 0.0432,
      "step": 10008
    },
    {
      "epoch": 6.4037108125399875,
      "grad_norm": 0.9711412710398114,
      "learning_rate": 6.050268974008149e-06,
      "loss": 0.0549,
      "step": 10009
    },
    {
      "epoch": 6.404350607805502,
      "grad_norm": 0.9371247767218939,
      "learning_rate": 6.048365386114041e-06,
      "loss": 0.0525,
      "step": 10010
    },
    {
      "epoch": 6.404990403071017,
      "grad_norm": 0.8533965195127249,
      "learning_rate": 6.046461967896093e-06,
      "loss": 0.0489,
      "step": 10011
    },
    {
      "epoch": 6.4056301983365325,
      "grad_norm": 0.8267945998786773,
      "learning_rate": 6.0445587194360335e-06,
      "loss": 0.0484,
      "step": 10012
    },
    {
      "epoch": 6.406269993602048,
      "grad_norm": 0.8463844184834978,
      "learning_rate": 6.042655640815589e-06,
      "loss": 0.05,
      "step": 10013
    },
    {
      "epoch": 6.406909788867562,
      "grad_norm": 0.8442536618499389,
      "learning_rate": 6.04075273211647e-06,
      "loss": 0.0525,
      "step": 10014
    },
    {
      "epoch": 6.407549584133077,
      "grad_norm": 0.7187361131077517,
      "learning_rate": 6.038849993420386e-06,
      "loss": 0.0468,
      "step": 10015
    },
    {
      "epoch": 6.408189379398593,
      "grad_norm": 0.8913908921088591,
      "learning_rate": 6.036947424809039e-06,
      "loss": 0.0524,
      "step": 10016
    },
    {
      "epoch": 6.408829174664108,
      "grad_norm": 1.0548974175581494,
      "learning_rate": 6.0350450263641194e-06,
      "loss": 0.0567,
      "step": 10017
    },
    {
      "epoch": 6.409468969929622,
      "grad_norm": 0.929202047194783,
      "learning_rate": 6.033142798167309e-06,
      "loss": 0.048,
      "step": 10018
    },
    {
      "epoch": 6.410108765195138,
      "grad_norm": 0.8132162163247405,
      "learning_rate": 6.0312407403002934e-06,
      "loss": 0.0517,
      "step": 10019
    },
    {
      "epoch": 6.410748560460653,
      "grad_norm": 0.8034553799365517,
      "learning_rate": 6.029338852844742e-06,
      "loss": 0.0494,
      "step": 10020
    },
    {
      "epoch": 6.411388355726167,
      "grad_norm": 0.8490483165545035,
      "learning_rate": 6.027437135882317e-06,
      "loss": 0.0536,
      "step": 10021
    },
    {
      "epoch": 6.4120281509916826,
      "grad_norm": 0.8541311991031021,
      "learning_rate": 6.0255355894946685e-06,
      "loss": 0.0566,
      "step": 10022
    },
    {
      "epoch": 6.412667946257198,
      "grad_norm": 0.7927473517153464,
      "learning_rate": 6.0236342137634585e-06,
      "loss": 0.0523,
      "step": 10023
    },
    {
      "epoch": 6.413307741522713,
      "grad_norm": 0.771862309751794,
      "learning_rate": 6.021733008770321e-06,
      "loss": 0.0456,
      "step": 10024
    },
    {
      "epoch": 6.4139475367882275,
      "grad_norm": 0.8014934610109842,
      "learning_rate": 6.019831974596889e-06,
      "loss": 0.0529,
      "step": 10025
    },
    {
      "epoch": 6.414587332053743,
      "grad_norm": 0.7986631631747637,
      "learning_rate": 6.017931111324795e-06,
      "loss": 0.0512,
      "step": 10026
    },
    {
      "epoch": 6.415227127319258,
      "grad_norm": 0.8286831246521296,
      "learning_rate": 6.016030419035654e-06,
      "loss": 0.0473,
      "step": 10027
    },
    {
      "epoch": 6.415866922584772,
      "grad_norm": 0.8450190547036008,
      "learning_rate": 6.0141298978110765e-06,
      "loss": 0.0518,
      "step": 10028
    },
    {
      "epoch": 6.416506717850288,
      "grad_norm": 0.8413087345715937,
      "learning_rate": 6.012229547732675e-06,
      "loss": 0.0523,
      "step": 10029
    },
    {
      "epoch": 6.417146513115803,
      "grad_norm": 0.8054613959738168,
      "learning_rate": 6.010329368882043e-06,
      "loss": 0.045,
      "step": 10030
    },
    {
      "epoch": 6.417786308381318,
      "grad_norm": 0.7962806658680612,
      "learning_rate": 6.008429361340771e-06,
      "loss": 0.0498,
      "step": 10031
    },
    {
      "epoch": 6.418426103646833,
      "grad_norm": 0.883499686980218,
      "learning_rate": 6.006529525190439e-06,
      "loss": 0.0536,
      "step": 10032
    },
    {
      "epoch": 6.419065898912348,
      "grad_norm": 0.7735749120778624,
      "learning_rate": 6.004629860512627e-06,
      "loss": 0.0485,
      "step": 10033
    },
    {
      "epoch": 6.419705694177863,
      "grad_norm": 0.7971678857688916,
      "learning_rate": 6.002730367388904e-06,
      "loss": 0.0534,
      "step": 10034
    },
    {
      "epoch": 6.4203454894433785,
      "grad_norm": 0.918564822170313,
      "learning_rate": 6.000831045900822e-06,
      "loss": 0.0543,
      "step": 10035
    },
    {
      "epoch": 6.420985284708893,
      "grad_norm": 0.8586810381725057,
      "learning_rate": 5.998931896129947e-06,
      "loss": 0.0551,
      "step": 10036
    },
    {
      "epoch": 6.421625079974408,
      "grad_norm": 0.9036984473486933,
      "learning_rate": 5.997032918157817e-06,
      "loss": 0.0528,
      "step": 10037
    },
    {
      "epoch": 6.422264875239923,
      "grad_norm": 0.8092767966212703,
      "learning_rate": 5.9951341120659725e-06,
      "loss": 0.0509,
      "step": 10038
    },
    {
      "epoch": 6.422904670505439,
      "grad_norm": 0.8839579647568322,
      "learning_rate": 5.9932354779359455e-06,
      "loss": 0.0485,
      "step": 10039
    },
    {
      "epoch": 6.423544465770953,
      "grad_norm": 0.925675554360472,
      "learning_rate": 5.9913370158492605e-06,
      "loss": 0.054,
      "step": 10040
    },
    {
      "epoch": 6.424184261036468,
      "grad_norm": 0.9593420566613066,
      "learning_rate": 5.989438725887433e-06,
      "loss": 0.0536,
      "step": 10041
    },
    {
      "epoch": 6.424824056301984,
      "grad_norm": 0.9908328873246206,
      "learning_rate": 5.9875406081319675e-06,
      "loss": 0.0542,
      "step": 10042
    },
    {
      "epoch": 6.425463851567498,
      "grad_norm": 0.891937274641799,
      "learning_rate": 5.985642662664377e-06,
      "loss": 0.0579,
      "step": 10043
    },
    {
      "epoch": 6.426103646833013,
      "grad_norm": 0.724474164544849,
      "learning_rate": 5.983744889566148e-06,
      "loss": 0.0474,
      "step": 10044
    },
    {
      "epoch": 6.4267434420985285,
      "grad_norm": 0.9695118627286872,
      "learning_rate": 5.981847288918767e-06,
      "loss": 0.0543,
      "step": 10045
    },
    {
      "epoch": 6.427383237364044,
      "grad_norm": 1.0238025132509736,
      "learning_rate": 5.9799498608037175e-06,
      "loss": 0.0552,
      "step": 10046
    },
    {
      "epoch": 6.428023032629558,
      "grad_norm": 0.9210074170810192,
      "learning_rate": 5.97805260530247e-06,
      "loss": 0.0538,
      "step": 10047
    },
    {
      "epoch": 6.4286628278950735,
      "grad_norm": 0.8550249228493323,
      "learning_rate": 5.976155522496485e-06,
      "loss": 0.0459,
      "step": 10048
    },
    {
      "epoch": 6.429302623160589,
      "grad_norm": 0.7816698717036318,
      "learning_rate": 5.97425861246723e-06,
      "loss": 0.0479,
      "step": 10049
    },
    {
      "epoch": 6.429942418426104,
      "grad_norm": 0.8415083389693737,
      "learning_rate": 5.972361875296147e-06,
      "loss": 0.0481,
      "step": 10050
    },
    {
      "epoch": 6.430582213691618,
      "grad_norm": 1.0299162496207674,
      "learning_rate": 5.97046531106468e-06,
      "loss": 0.0713,
      "step": 10051
    },
    {
      "epoch": 6.431222008957134,
      "grad_norm": 0.7265142934631529,
      "learning_rate": 5.968568919854262e-06,
      "loss": 0.0477,
      "step": 10052
    },
    {
      "epoch": 6.431861804222649,
      "grad_norm": 0.8308541934139171,
      "learning_rate": 5.966672701746326e-06,
      "loss": 0.048,
      "step": 10053
    },
    {
      "epoch": 6.432501599488164,
      "grad_norm": 0.7580490326345087,
      "learning_rate": 5.964776656822291e-06,
      "loss": 0.0495,
      "step": 10054
    },
    {
      "epoch": 6.433141394753679,
      "grad_norm": 0.8207141183013413,
      "learning_rate": 5.962880785163565e-06,
      "loss": 0.0521,
      "step": 10055
    },
    {
      "epoch": 6.433781190019194,
      "grad_norm": 0.7984872003232004,
      "learning_rate": 5.960985086851558e-06,
      "loss": 0.0539,
      "step": 10056
    },
    {
      "epoch": 6.434420985284709,
      "grad_norm": 0.8343575967664427,
      "learning_rate": 5.959089561967668e-06,
      "loss": 0.056,
      "step": 10057
    },
    {
      "epoch": 6.435060780550224,
      "grad_norm": 0.8127040991792784,
      "learning_rate": 5.957194210593283e-06,
      "loss": 0.0463,
      "step": 10058
    },
    {
      "epoch": 6.435700575815739,
      "grad_norm": 0.8067625737244638,
      "learning_rate": 5.955299032809784e-06,
      "loss": 0.0494,
      "step": 10059
    },
    {
      "epoch": 6.436340371081254,
      "grad_norm": 0.7738558503453202,
      "learning_rate": 5.9534040286985526e-06,
      "loss": 0.0529,
      "step": 10060
    },
    {
      "epoch": 6.436980166346769,
      "grad_norm": 0.8076574730923131,
      "learning_rate": 5.951509198340955e-06,
      "loss": 0.0539,
      "step": 10061
    },
    {
      "epoch": 6.437619961612284,
      "grad_norm": 0.8180623431588953,
      "learning_rate": 5.949614541818347e-06,
      "loss": 0.0486,
      "step": 10062
    },
    {
      "epoch": 6.438259756877799,
      "grad_norm": 0.8938590203055226,
      "learning_rate": 5.9477200592120884e-06,
      "loss": 0.0576,
      "step": 10063
    },
    {
      "epoch": 6.438899552143314,
      "grad_norm": 0.8311773977631294,
      "learning_rate": 5.945825750603522e-06,
      "loss": 0.0529,
      "step": 10064
    },
    {
      "epoch": 6.439539347408829,
      "grad_norm": 0.8383283880599305,
      "learning_rate": 5.943931616073983e-06,
      "loss": 0.0491,
      "step": 10065
    },
    {
      "epoch": 6.440179142674344,
      "grad_norm": 0.873984474376873,
      "learning_rate": 5.942037655704808e-06,
      "loss": 0.0551,
      "step": 10066
    },
    {
      "epoch": 6.440818937939859,
      "grad_norm": 0.7572274756514197,
      "learning_rate": 5.940143869577317e-06,
      "loss": 0.0502,
      "step": 10067
    },
    {
      "epoch": 6.4414587332053745,
      "grad_norm": 0.8402704483040285,
      "learning_rate": 5.9382502577728285e-06,
      "loss": 0.0518,
      "step": 10068
    },
    {
      "epoch": 6.442098528470889,
      "grad_norm": 0.8394362052906595,
      "learning_rate": 5.936356820372644e-06,
      "loss": 0.0529,
      "step": 10069
    },
    {
      "epoch": 6.442738323736404,
      "grad_norm": 0.8737961086811573,
      "learning_rate": 5.934463557458072e-06,
      "loss": 0.0571,
      "step": 10070
    },
    {
      "epoch": 6.4433781190019195,
      "grad_norm": 0.8487046894162332,
      "learning_rate": 5.932570469110403e-06,
      "loss": 0.0468,
      "step": 10071
    },
    {
      "epoch": 6.444017914267435,
      "grad_norm": 0.7782818436599065,
      "learning_rate": 5.93067755541092e-06,
      "loss": 0.0472,
      "step": 10072
    },
    {
      "epoch": 6.444657709532949,
      "grad_norm": 0.8139587926602936,
      "learning_rate": 5.928784816440907e-06,
      "loss": 0.0464,
      "step": 10073
    },
    {
      "epoch": 6.445297504798464,
      "grad_norm": 0.8292143340811939,
      "learning_rate": 5.926892252281631e-06,
      "loss": 0.0524,
      "step": 10074
    },
    {
      "epoch": 6.44593730006398,
      "grad_norm": 0.8430863203028159,
      "learning_rate": 5.9249998630143556e-06,
      "loss": 0.0477,
      "step": 10075
    },
    {
      "epoch": 6.446577095329495,
      "grad_norm": 0.8415047345013406,
      "learning_rate": 5.923107648720338e-06,
      "loss": 0.0439,
      "step": 10076
    },
    {
      "epoch": 6.447216890595009,
      "grad_norm": 0.8447906625842703,
      "learning_rate": 5.921215609480826e-06,
      "loss": 0.053,
      "step": 10077
    },
    {
      "epoch": 6.447856685860525,
      "grad_norm": 0.7215375118011238,
      "learning_rate": 5.91932374537706e-06,
      "loss": 0.046,
      "step": 10078
    },
    {
      "epoch": 6.44849648112604,
      "grad_norm": 0.8209688877706257,
      "learning_rate": 5.91743205649027e-06,
      "loss": 0.0493,
      "step": 10079
    },
    {
      "epoch": 6.449136276391554,
      "grad_norm": 0.7436399415515454,
      "learning_rate": 5.915540542901689e-06,
      "loss": 0.0474,
      "step": 10080
    },
    {
      "epoch": 6.44977607165707,
      "grad_norm": 0.8389996500553316,
      "learning_rate": 5.9136492046925305e-06,
      "loss": 0.046,
      "step": 10081
    },
    {
      "epoch": 6.450415866922585,
      "grad_norm": 0.8105735418248993,
      "learning_rate": 5.9117580419440034e-06,
      "loss": 0.0537,
      "step": 10082
    },
    {
      "epoch": 6.4510556621881,
      "grad_norm": 0.839898041316843,
      "learning_rate": 5.909867054737317e-06,
      "loss": 0.0508,
      "step": 10083
    },
    {
      "epoch": 6.4516954574536145,
      "grad_norm": 0.8251397882125785,
      "learning_rate": 5.907976243153664e-06,
      "loss": 0.0476,
      "step": 10084
    },
    {
      "epoch": 6.45233525271913,
      "grad_norm": 0.834897751351002,
      "learning_rate": 5.906085607274229e-06,
      "loss": 0.0525,
      "step": 10085
    },
    {
      "epoch": 6.452975047984645,
      "grad_norm": 0.8739554998489119,
      "learning_rate": 5.904195147180198e-06,
      "loss": 0.0543,
      "step": 10086
    },
    {
      "epoch": 6.45361484325016,
      "grad_norm": 0.9107474049854656,
      "learning_rate": 5.902304862952741e-06,
      "loss": 0.0563,
      "step": 10087
    },
    {
      "epoch": 6.454254638515675,
      "grad_norm": 0.804581902455562,
      "learning_rate": 5.900414754673025e-06,
      "loss": 0.0494,
      "step": 10088
    },
    {
      "epoch": 6.45489443378119,
      "grad_norm": 0.8150821583038551,
      "learning_rate": 5.898524822422203e-06,
      "loss": 0.0566,
      "step": 10089
    },
    {
      "epoch": 6.455534229046705,
      "grad_norm": 0.808066055344944,
      "learning_rate": 5.896635066281432e-06,
      "loss": 0.0509,
      "step": 10090
    },
    {
      "epoch": 6.4561740243122205,
      "grad_norm": 0.8823403905512227,
      "learning_rate": 5.894745486331852e-06,
      "loss": 0.0543,
      "step": 10091
    },
    {
      "epoch": 6.456813819577735,
      "grad_norm": 0.7688274653878544,
      "learning_rate": 5.892856082654597e-06,
      "loss": 0.0458,
      "step": 10092
    },
    {
      "epoch": 6.45745361484325,
      "grad_norm": 0.8498547153329981,
      "learning_rate": 5.8909668553307974e-06,
      "loss": 0.0539,
      "step": 10093
    },
    {
      "epoch": 6.4580934101087655,
      "grad_norm": 0.9280799610216297,
      "learning_rate": 5.889077804441572e-06,
      "loss": 0.0531,
      "step": 10094
    },
    {
      "epoch": 6.45873320537428,
      "grad_norm": 0.8525326801821911,
      "learning_rate": 5.887188930068027e-06,
      "loss": 0.0598,
      "step": 10095
    },
    {
      "epoch": 6.459373000639795,
      "grad_norm": 0.8691637269083067,
      "learning_rate": 5.88530023229128e-06,
      "loss": 0.049,
      "step": 10096
    },
    {
      "epoch": 6.46001279590531,
      "grad_norm": 0.9172472024369124,
      "learning_rate": 5.88341171119242e-06,
      "loss": 0.0525,
      "step": 10097
    },
    {
      "epoch": 6.460652591170826,
      "grad_norm": 0.7971477909728769,
      "learning_rate": 5.881523366852538e-06,
      "loss": 0.0519,
      "step": 10098
    },
    {
      "epoch": 6.46129238643634,
      "grad_norm": 0.8876071084820694,
      "learning_rate": 5.879635199352715e-06,
      "loss": 0.0568,
      "step": 10099
    },
    {
      "epoch": 6.461932181701855,
      "grad_norm": 0.7068288038784961,
      "learning_rate": 5.8777472087740295e-06,
      "loss": 0.0491,
      "step": 10100
    },
    {
      "epoch": 6.462571976967371,
      "grad_norm": 0.870410472234208,
      "learning_rate": 5.875859395197545e-06,
      "loss": 0.0481,
      "step": 10101
    },
    {
      "epoch": 6.463211772232885,
      "grad_norm": 0.8109743434021242,
      "learning_rate": 5.8739717587043175e-06,
      "loss": 0.0531,
      "step": 10102
    },
    {
      "epoch": 6.4638515674984,
      "grad_norm": 0.7686867456053295,
      "learning_rate": 5.8720842993754075e-06,
      "loss": 0.0479,
      "step": 10103
    },
    {
      "epoch": 6.464491362763916,
      "grad_norm": 0.8701938925556789,
      "learning_rate": 5.870197017291855e-06,
      "loss": 0.05,
      "step": 10104
    },
    {
      "epoch": 6.465131158029431,
      "grad_norm": 0.8039154287730377,
      "learning_rate": 5.868309912534694e-06,
      "loss": 0.0497,
      "step": 10105
    },
    {
      "epoch": 6.465770953294945,
      "grad_norm": 0.7964083257433336,
      "learning_rate": 5.866422985184958e-06,
      "loss": 0.0521,
      "step": 10106
    },
    {
      "epoch": 6.4664107485604605,
      "grad_norm": 0.808669436543782,
      "learning_rate": 5.8645362353236654e-06,
      "loss": 0.0487,
      "step": 10107
    },
    {
      "epoch": 6.467050543825976,
      "grad_norm": 0.8768026278257479,
      "learning_rate": 5.862649663031832e-06,
      "loss": 0.0487,
      "step": 10108
    },
    {
      "epoch": 6.467690339091491,
      "grad_norm": 0.8207516758554841,
      "learning_rate": 5.860763268390457e-06,
      "loss": 0.0464,
      "step": 10109
    },
    {
      "epoch": 6.468330134357005,
      "grad_norm": 0.8099898356814151,
      "learning_rate": 5.858877051480547e-06,
      "loss": 0.0486,
      "step": 10110
    },
    {
      "epoch": 6.468969929622521,
      "grad_norm": 0.9030090598818823,
      "learning_rate": 5.856991012383091e-06,
      "loss": 0.056,
      "step": 10111
    },
    {
      "epoch": 6.469609724888036,
      "grad_norm": 0.8103998763566401,
      "learning_rate": 5.855105151179068e-06,
      "loss": 0.0542,
      "step": 10112
    },
    {
      "epoch": 6.470249520153551,
      "grad_norm": 0.9397918533866374,
      "learning_rate": 5.853219467949459e-06,
      "loss": 0.0555,
      "step": 10113
    },
    {
      "epoch": 6.470889315419066,
      "grad_norm": 0.7176586356796488,
      "learning_rate": 5.851333962775231e-06,
      "loss": 0.049,
      "step": 10114
    },
    {
      "epoch": 6.471529110684581,
      "grad_norm": 0.8893806530180353,
      "learning_rate": 5.84944863573734e-06,
      "loss": 0.0442,
      "step": 10115
    },
    {
      "epoch": 6.472168905950096,
      "grad_norm": 0.9474460256158517,
      "learning_rate": 5.847563486916743e-06,
      "loss": 0.0599,
      "step": 10116
    },
    {
      "epoch": 6.472808701215611,
      "grad_norm": 0.8639830104142536,
      "learning_rate": 5.845678516394384e-06,
      "loss": 0.0527,
      "step": 10117
    },
    {
      "epoch": 6.473448496481126,
      "grad_norm": 0.7973058118101776,
      "learning_rate": 5.843793724251199e-06,
      "loss": 0.0469,
      "step": 10118
    },
    {
      "epoch": 6.474088291746641,
      "grad_norm": 0.8834463385172431,
      "learning_rate": 5.841909110568114e-06,
      "loss": 0.0544,
      "step": 10119
    },
    {
      "epoch": 6.474728087012156,
      "grad_norm": 0.7139408425294286,
      "learning_rate": 5.840024675426059e-06,
      "loss": 0.0471,
      "step": 10120
    },
    {
      "epoch": 6.475367882277671,
      "grad_norm": 0.8014857329281115,
      "learning_rate": 5.8381404189059466e-06,
      "loss": 0.0517,
      "step": 10121
    },
    {
      "epoch": 6.476007677543186,
      "grad_norm": 0.759147271355477,
      "learning_rate": 5.836256341088675e-06,
      "loss": 0.0487,
      "step": 10122
    },
    {
      "epoch": 6.476647472808701,
      "grad_norm": 0.7523476867500295,
      "learning_rate": 5.834372442055155e-06,
      "loss": 0.0513,
      "step": 10123
    },
    {
      "epoch": 6.477287268074217,
      "grad_norm": 0.8730765471128094,
      "learning_rate": 5.8324887218862714e-06,
      "loss": 0.0578,
      "step": 10124
    },
    {
      "epoch": 6.477927063339731,
      "grad_norm": 0.6985305162446147,
      "learning_rate": 5.8306051806629035e-06,
      "loss": 0.0457,
      "step": 10125
    },
    {
      "epoch": 6.478566858605246,
      "grad_norm": 0.7660320830135136,
      "learning_rate": 5.8287218184659375e-06,
      "loss": 0.0464,
      "step": 10126
    },
    {
      "epoch": 6.479206653870762,
      "grad_norm": 0.867014466717202,
      "learning_rate": 5.826838635376236e-06,
      "loss": 0.0541,
      "step": 10127
    },
    {
      "epoch": 6.479846449136277,
      "grad_norm": 0.8980394525305441,
      "learning_rate": 5.82495563147466e-06,
      "loss": 0.0591,
      "step": 10128
    },
    {
      "epoch": 6.480486244401791,
      "grad_norm": 0.7601277221874055,
      "learning_rate": 5.823072806842059e-06,
      "loss": 0.0514,
      "step": 10129
    },
    {
      "epoch": 6.4811260396673065,
      "grad_norm": 0.762030913593661,
      "learning_rate": 5.8211901615592845e-06,
      "loss": 0.0458,
      "step": 10130
    },
    {
      "epoch": 6.481765834932822,
      "grad_norm": 0.7697126725210565,
      "learning_rate": 5.8193076957071725e-06,
      "loss": 0.0445,
      "step": 10131
    },
    {
      "epoch": 6.482405630198336,
      "grad_norm": 0.7731224851474328,
      "learning_rate": 5.817425409366551e-06,
      "loss": 0.0482,
      "step": 10132
    },
    {
      "epoch": 6.483045425463851,
      "grad_norm": 0.7471025146406899,
      "learning_rate": 5.81554330261824e-06,
      "loss": 0.0461,
      "step": 10133
    },
    {
      "epoch": 6.483685220729367,
      "grad_norm": 0.8935654853894024,
      "learning_rate": 5.813661375543058e-06,
      "loss": 0.058,
      "step": 10134
    },
    {
      "epoch": 6.484325015994882,
      "grad_norm": 0.8364852544140302,
      "learning_rate": 5.8117796282218085e-06,
      "loss": 0.0542,
      "step": 10135
    },
    {
      "epoch": 6.484964811260396,
      "grad_norm": 0.8455712891235354,
      "learning_rate": 5.809898060735288e-06,
      "loss": 0.0528,
      "step": 10136
    },
    {
      "epoch": 6.485604606525912,
      "grad_norm": 0.9496993571952953,
      "learning_rate": 5.808016673164296e-06,
      "loss": 0.0488,
      "step": 10137
    },
    {
      "epoch": 6.486244401791427,
      "grad_norm": 0.8570661732732063,
      "learning_rate": 5.80613546558961e-06,
      "loss": 0.0537,
      "step": 10138
    },
    {
      "epoch": 6.486884197056942,
      "grad_norm": 0.8415874482829822,
      "learning_rate": 5.804254438092003e-06,
      "loss": 0.0528,
      "step": 10139
    },
    {
      "epoch": 6.487523992322457,
      "grad_norm": 0.7892211577499049,
      "learning_rate": 5.802373590752251e-06,
      "loss": 0.0456,
      "step": 10140
    },
    {
      "epoch": 6.488163787587972,
      "grad_norm": 0.8668009975447609,
      "learning_rate": 5.80049292365111e-06,
      "loss": 0.0503,
      "step": 10141
    },
    {
      "epoch": 6.488803582853487,
      "grad_norm": 0.8613701959726676,
      "learning_rate": 5.798612436869329e-06,
      "loss": 0.0539,
      "step": 10142
    },
    {
      "epoch": 6.4894433781190015,
      "grad_norm": 0.7995635215023562,
      "learning_rate": 5.796732130487659e-06,
      "loss": 0.0477,
      "step": 10143
    },
    {
      "epoch": 6.490083173384517,
      "grad_norm": 0.8676188732670466,
      "learning_rate": 5.794852004586835e-06,
      "loss": 0.0546,
      "step": 10144
    },
    {
      "epoch": 6.490722968650032,
      "grad_norm": 0.7695315092351034,
      "learning_rate": 5.792972059247586e-06,
      "loss": 0.0509,
      "step": 10145
    },
    {
      "epoch": 6.491362763915547,
      "grad_norm": 0.8356221080330976,
      "learning_rate": 5.791092294550628e-06,
      "loss": 0.0553,
      "step": 10146
    },
    {
      "epoch": 6.492002559181062,
      "grad_norm": 0.8604207148516644,
      "learning_rate": 5.789212710576684e-06,
      "loss": 0.0546,
      "step": 10147
    },
    {
      "epoch": 6.492642354446577,
      "grad_norm": 0.7531409173221668,
      "learning_rate": 5.787333307406457e-06,
      "loss": 0.0479,
      "step": 10148
    },
    {
      "epoch": 6.493282149712092,
      "grad_norm": 0.8935023572501689,
      "learning_rate": 5.785454085120643e-06,
      "loss": 0.045,
      "step": 10149
    },
    {
      "epoch": 6.4939219449776076,
      "grad_norm": 0.8891379674888487,
      "learning_rate": 5.783575043799934e-06,
      "loss": 0.0519,
      "step": 10150
    },
    {
      "epoch": 6.494561740243122,
      "grad_norm": 0.9537499609313512,
      "learning_rate": 5.781696183525011e-06,
      "loss": 0.0548,
      "step": 10151
    },
    {
      "epoch": 6.495201535508637,
      "grad_norm": 0.8605418468258829,
      "learning_rate": 5.779817504376547e-06,
      "loss": 0.0505,
      "step": 10152
    },
    {
      "epoch": 6.4958413307741525,
      "grad_norm": 0.8479099790006113,
      "learning_rate": 5.777939006435216e-06,
      "loss": 0.0568,
      "step": 10153
    },
    {
      "epoch": 6.496481126039667,
      "grad_norm": 0.7877848070160572,
      "learning_rate": 5.776060689781674e-06,
      "loss": 0.0475,
      "step": 10154
    },
    {
      "epoch": 6.497120921305182,
      "grad_norm": 0.801676357319478,
      "learning_rate": 5.774182554496573e-06,
      "loss": 0.0469,
      "step": 10155
    },
    {
      "epoch": 6.497760716570697,
      "grad_norm": 0.8497961361914066,
      "learning_rate": 5.77230460066055e-06,
      "loss": 0.0529,
      "step": 10156
    },
    {
      "epoch": 6.498400511836213,
      "grad_norm": 0.761181975835295,
      "learning_rate": 5.770426828354252e-06,
      "loss": 0.0443,
      "step": 10157
    },
    {
      "epoch": 6.499040307101727,
      "grad_norm": 0.8536207610022504,
      "learning_rate": 5.768549237658301e-06,
      "loss": 0.053,
      "step": 10158
    },
    {
      "epoch": 6.499680102367242,
      "grad_norm": 0.8649915844506009,
      "learning_rate": 5.766671828653317e-06,
      "loss": 0.0509,
      "step": 10159
    },
    {
      "epoch": 6.500319897632758,
      "grad_norm": 0.8510811195812272,
      "learning_rate": 5.764794601419916e-06,
      "loss": 0.0497,
      "step": 10160
    },
    {
      "epoch": 6.500959692898273,
      "grad_norm": 0.7630012310048717,
      "learning_rate": 5.762917556038702e-06,
      "loss": 0.047,
      "step": 10161
    },
    {
      "epoch": 6.501599488163787,
      "grad_norm": 0.8086463940061155,
      "learning_rate": 5.76104069259027e-06,
      "loss": 0.054,
      "step": 10162
    },
    {
      "epoch": 6.502239283429303,
      "grad_norm": 0.849653740273666,
      "learning_rate": 5.759164011155212e-06,
      "loss": 0.0558,
      "step": 10163
    },
    {
      "epoch": 6.502879078694818,
      "grad_norm": 0.743591964826896,
      "learning_rate": 5.7572875118141054e-06,
      "loss": 0.0482,
      "step": 10164
    },
    {
      "epoch": 6.503518873960333,
      "grad_norm": 0.7646483905468464,
      "learning_rate": 5.755411194647527e-06,
      "loss": 0.044,
      "step": 10165
    },
    {
      "epoch": 6.5041586692258475,
      "grad_norm": 0.745968197606875,
      "learning_rate": 5.753535059736037e-06,
      "loss": 0.0491,
      "step": 10166
    },
    {
      "epoch": 6.504798464491363,
      "grad_norm": 0.7832252319789266,
      "learning_rate": 5.751659107160201e-06,
      "loss": 0.0535,
      "step": 10167
    },
    {
      "epoch": 6.505438259756878,
      "grad_norm": 0.8327963188669285,
      "learning_rate": 5.749783337000567e-06,
      "loss": 0.0526,
      "step": 10168
    },
    {
      "epoch": 6.5060780550223924,
      "grad_norm": 0.7510554354035432,
      "learning_rate": 5.747907749337672e-06,
      "loss": 0.0433,
      "step": 10169
    },
    {
      "epoch": 6.506717850287908,
      "grad_norm": 0.8899302199626552,
      "learning_rate": 5.746032344252057e-06,
      "loss": 0.0531,
      "step": 10170
    },
    {
      "epoch": 6.507357645553423,
      "grad_norm": 0.7415174346899056,
      "learning_rate": 5.7441571218242475e-06,
      "loss": 0.0433,
      "step": 10171
    },
    {
      "epoch": 6.507997440818938,
      "grad_norm": 0.8032605128810892,
      "learning_rate": 5.742282082134755e-06,
      "loss": 0.052,
      "step": 10172
    },
    {
      "epoch": 6.508637236084453,
      "grad_norm": 0.8029155463078373,
      "learning_rate": 5.740407225264102e-06,
      "loss": 0.0472,
      "step": 10173
    },
    {
      "epoch": 6.509277031349968,
      "grad_norm": 0.839293492729461,
      "learning_rate": 5.738532551292784e-06,
      "loss": 0.0563,
      "step": 10174
    },
    {
      "epoch": 6.509916826615483,
      "grad_norm": 0.762896177906835,
      "learning_rate": 5.7366580603012985e-06,
      "loss": 0.0533,
      "step": 10175
    },
    {
      "epoch": 6.510556621880998,
      "grad_norm": 0.8131021659516642,
      "learning_rate": 5.734783752370128e-06,
      "loss": 0.0545,
      "step": 10176
    },
    {
      "epoch": 6.511196417146513,
      "grad_norm": 0.8194130484856991,
      "learning_rate": 5.732909627579759e-06,
      "loss": 0.0533,
      "step": 10177
    },
    {
      "epoch": 6.511836212412028,
      "grad_norm": 0.9050652500272348,
      "learning_rate": 5.731035686010661e-06,
      "loss": 0.054,
      "step": 10178
    },
    {
      "epoch": 6.512476007677543,
      "grad_norm": 0.835742438105351,
      "learning_rate": 5.729161927743297e-06,
      "loss": 0.0477,
      "step": 10179
    },
    {
      "epoch": 6.513115802943059,
      "grad_norm": 0.8506847973588275,
      "learning_rate": 5.72728835285812e-06,
      "loss": 0.0519,
      "step": 10180
    },
    {
      "epoch": 6.513755598208573,
      "grad_norm": 0.8365933870791278,
      "learning_rate": 5.725414961435584e-06,
      "loss": 0.0466,
      "step": 10181
    },
    {
      "epoch": 6.514395393474088,
      "grad_norm": 0.7658037420262412,
      "learning_rate": 5.723541753556119e-06,
      "loss": 0.0459,
      "step": 10182
    },
    {
      "epoch": 6.515035188739604,
      "grad_norm": 0.9308424808507382,
      "learning_rate": 5.721668729300169e-06,
      "loss": 0.0557,
      "step": 10183
    },
    {
      "epoch": 6.515674984005118,
      "grad_norm": 0.8293800064086336,
      "learning_rate": 5.719795888748154e-06,
      "loss": 0.0561,
      "step": 10184
    },
    {
      "epoch": 6.516314779270633,
      "grad_norm": 0.8952622629850828,
      "learning_rate": 5.717923231980489e-06,
      "loss": 0.0559,
      "step": 10185
    },
    {
      "epoch": 6.516954574536149,
      "grad_norm": 0.7967303934705255,
      "learning_rate": 5.716050759077578e-06,
      "loss": 0.044,
      "step": 10186
    },
    {
      "epoch": 6.517594369801664,
      "grad_norm": 0.7931696285534564,
      "learning_rate": 5.7141784701198335e-06,
      "loss": 0.049,
      "step": 10187
    },
    {
      "epoch": 6.518234165067178,
      "grad_norm": 0.7946775961721837,
      "learning_rate": 5.712306365187639e-06,
      "loss": 0.0482,
      "step": 10188
    },
    {
      "epoch": 6.5188739603326935,
      "grad_norm": 0.7631058125997967,
      "learning_rate": 5.710434444361379e-06,
      "loss": 0.0455,
      "step": 10189
    },
    {
      "epoch": 6.519513755598209,
      "grad_norm": 0.6843572100414957,
      "learning_rate": 5.708562707721437e-06,
      "loss": 0.0418,
      "step": 10190
    },
    {
      "epoch": 6.520153550863723,
      "grad_norm": 0.8370867652407561,
      "learning_rate": 5.706691155348178e-06,
      "loss": 0.0537,
      "step": 10191
    },
    {
      "epoch": 6.520793346129238,
      "grad_norm": 0.7803638208150612,
      "learning_rate": 5.7048197873219656e-06,
      "loss": 0.0497,
      "step": 10192
    },
    {
      "epoch": 6.521433141394754,
      "grad_norm": 0.9290573432755252,
      "learning_rate": 5.7029486037231504e-06,
      "loss": 0.0616,
      "step": 10193
    },
    {
      "epoch": 6.522072936660269,
      "grad_norm": 0.775121076040483,
      "learning_rate": 5.701077604632077e-06,
      "loss": 0.0483,
      "step": 10194
    },
    {
      "epoch": 6.522712731925783,
      "grad_norm": 0.7857961641280744,
      "learning_rate": 5.699206790129085e-06,
      "loss": 0.0484,
      "step": 10195
    },
    {
      "epoch": 6.523352527191299,
      "grad_norm": 0.7472202980280739,
      "learning_rate": 5.6973361602944985e-06,
      "loss": 0.0561,
      "step": 10196
    },
    {
      "epoch": 6.523992322456814,
      "grad_norm": 0.8392845467680856,
      "learning_rate": 5.695465715208648e-06,
      "loss": 0.0465,
      "step": 10197
    },
    {
      "epoch": 6.524632117722329,
      "grad_norm": 0.7614269949079142,
      "learning_rate": 5.693595454951841e-06,
      "loss": 0.0439,
      "step": 10198
    },
    {
      "epoch": 6.525271912987844,
      "grad_norm": 0.83811856547044,
      "learning_rate": 5.6917253796043825e-06,
      "loss": 0.0456,
      "step": 10199
    },
    {
      "epoch": 6.525911708253359,
      "grad_norm": 0.9133774545393093,
      "learning_rate": 5.6898554892465765e-06,
      "loss": 0.0502,
      "step": 10200
    },
    {
      "epoch": 6.526551503518874,
      "grad_norm": 0.866748954926135,
      "learning_rate": 5.687985783958707e-06,
      "loss": 0.0482,
      "step": 10201
    },
    {
      "epoch": 6.527191298784389,
      "grad_norm": 0.7850536966065679,
      "learning_rate": 5.686116263821055e-06,
      "loss": 0.0518,
      "step": 10202
    },
    {
      "epoch": 6.527831094049904,
      "grad_norm": 0.894883357388319,
      "learning_rate": 5.684246928913899e-06,
      "loss": 0.0518,
      "step": 10203
    },
    {
      "epoch": 6.528470889315419,
      "grad_norm": 0.8282719814021225,
      "learning_rate": 5.682377779317505e-06,
      "loss": 0.05,
      "step": 10204
    },
    {
      "epoch": 6.529110684580934,
      "grad_norm": 0.8432438997341537,
      "learning_rate": 5.680508815112127e-06,
      "loss": 0.0514,
      "step": 10205
    },
    {
      "epoch": 6.529750479846449,
      "grad_norm": 0.8443990959454745,
      "learning_rate": 5.678640036378014e-06,
      "loss": 0.046,
      "step": 10206
    },
    {
      "epoch": 6.530390275111964,
      "grad_norm": 0.7746800133474654,
      "learning_rate": 5.676771443195413e-06,
      "loss": 0.0448,
      "step": 10207
    },
    {
      "epoch": 6.531030070377479,
      "grad_norm": 0.8225693316281412,
      "learning_rate": 5.674903035644557e-06,
      "loss": 0.0512,
      "step": 10208
    },
    {
      "epoch": 6.531669865642995,
      "grad_norm": 0.8408065298760572,
      "learning_rate": 5.673034813805671e-06,
      "loss": 0.0561,
      "step": 10209
    },
    {
      "epoch": 6.532309660908509,
      "grad_norm": 0.7766685163090434,
      "learning_rate": 5.671166777758972e-06,
      "loss": 0.0468,
      "step": 10210
    },
    {
      "epoch": 6.532949456174024,
      "grad_norm": 0.8482366544223992,
      "learning_rate": 5.669298927584672e-06,
      "loss": 0.0448,
      "step": 10211
    },
    {
      "epoch": 6.5335892514395395,
      "grad_norm": 0.8619694704983774,
      "learning_rate": 5.667431263362973e-06,
      "loss": 0.0557,
      "step": 10212
    },
    {
      "epoch": 6.534229046705054,
      "grad_norm": 0.8407778600179675,
      "learning_rate": 5.665563785174064e-06,
      "loss": 0.0479,
      "step": 10213
    },
    {
      "epoch": 6.534868841970569,
      "grad_norm": 0.8527268212003546,
      "learning_rate": 5.6636964930981385e-06,
      "loss": 0.057,
      "step": 10214
    },
    {
      "epoch": 6.535508637236084,
      "grad_norm": 0.9463845504780015,
      "learning_rate": 5.661829387215373e-06,
      "loss": 0.0547,
      "step": 10215
    },
    {
      "epoch": 6.5361484325016,
      "grad_norm": 0.9897419458210778,
      "learning_rate": 5.659962467605933e-06,
      "loss": 0.0572,
      "step": 10216
    },
    {
      "epoch": 6.536788227767115,
      "grad_norm": 0.7338726960250564,
      "learning_rate": 5.658095734349987e-06,
      "loss": 0.0485,
      "step": 10217
    },
    {
      "epoch": 6.537428023032629,
      "grad_norm": 0.9256482257762562,
      "learning_rate": 5.656229187527687e-06,
      "loss": 0.0526,
      "step": 10218
    },
    {
      "epoch": 6.538067818298145,
      "grad_norm": 0.8828736303360645,
      "learning_rate": 5.654362827219175e-06,
      "loss": 0.0489,
      "step": 10219
    },
    {
      "epoch": 6.53870761356366,
      "grad_norm": 0.8030964603525599,
      "learning_rate": 5.652496653504598e-06,
      "loss": 0.051,
      "step": 10220
    },
    {
      "epoch": 6.539347408829174,
      "grad_norm": 0.870481739487129,
      "learning_rate": 5.650630666464079e-06,
      "loss": 0.0557,
      "step": 10221
    },
    {
      "epoch": 6.53998720409469,
      "grad_norm": 0.8543900717088159,
      "learning_rate": 5.648764866177744e-06,
      "loss": 0.0558,
      "step": 10222
    },
    {
      "epoch": 6.540626999360205,
      "grad_norm": 0.8310772381624072,
      "learning_rate": 5.6468992527257034e-06,
      "loss": 0.0464,
      "step": 10223
    },
    {
      "epoch": 6.54126679462572,
      "grad_norm": 0.8902125645208531,
      "learning_rate": 5.645033826188067e-06,
      "loss": 0.0544,
      "step": 10224
    },
    {
      "epoch": 6.5419065898912345,
      "grad_norm": 0.8674654823611049,
      "learning_rate": 5.6431685866449315e-06,
      "loss": 0.0532,
      "step": 10225
    },
    {
      "epoch": 6.54254638515675,
      "grad_norm": 0.7705133686238762,
      "learning_rate": 5.641303534176381e-06,
      "loss": 0.0492,
      "step": 10226
    },
    {
      "epoch": 6.543186180422265,
      "grad_norm": 0.7791747368214774,
      "learning_rate": 5.639438668862509e-06,
      "loss": 0.041,
      "step": 10227
    },
    {
      "epoch": 6.5438259756877795,
      "grad_norm": 0.7737987975829522,
      "learning_rate": 5.6375739907833826e-06,
      "loss": 0.0484,
      "step": 10228
    },
    {
      "epoch": 6.544465770953295,
      "grad_norm": 0.8357880023324535,
      "learning_rate": 5.635709500019065e-06,
      "loss": 0.0457,
      "step": 10229
    },
    {
      "epoch": 6.54510556621881,
      "grad_norm": 0.7808069081276545,
      "learning_rate": 5.633845196649624e-06,
      "loss": 0.0435,
      "step": 10230
    },
    {
      "epoch": 6.545745361484325,
      "grad_norm": 0.8486910891705944,
      "learning_rate": 5.6319810807551e-06,
      "loss": 0.0469,
      "step": 10231
    },
    {
      "epoch": 6.54638515674984,
      "grad_norm": 0.9289221948917622,
      "learning_rate": 5.630117152415539e-06,
      "loss": 0.0539,
      "step": 10232
    },
    {
      "epoch": 6.547024952015355,
      "grad_norm": 0.8022831300339734,
      "learning_rate": 5.628253411710972e-06,
      "loss": 0.047,
      "step": 10233
    },
    {
      "epoch": 6.54766474728087,
      "grad_norm": 0.7292380775172338,
      "learning_rate": 5.62638985872143e-06,
      "loss": 0.0447,
      "step": 10234
    },
    {
      "epoch": 6.5483045425463855,
      "grad_norm": 0.7980509313489317,
      "learning_rate": 5.624526493526927e-06,
      "loss": 0.0476,
      "step": 10235
    },
    {
      "epoch": 6.5489443378119,
      "grad_norm": 0.8745530614825919,
      "learning_rate": 5.62266331620747e-06,
      "loss": 0.054,
      "step": 10236
    },
    {
      "epoch": 6.549584133077415,
      "grad_norm": 0.8311544128523398,
      "learning_rate": 5.620800326843066e-06,
      "loss": 0.0491,
      "step": 10237
    },
    {
      "epoch": 6.55022392834293,
      "grad_norm": 0.9046976656103803,
      "learning_rate": 5.618937525513707e-06,
      "loss": 0.0486,
      "step": 10238
    },
    {
      "epoch": 6.550863723608446,
      "grad_norm": 0.8003389205013265,
      "learning_rate": 5.617074912299376e-06,
      "loss": 0.0502,
      "step": 10239
    },
    {
      "epoch": 6.55150351887396,
      "grad_norm": 0.8784915909183543,
      "learning_rate": 5.615212487280053e-06,
      "loss": 0.0503,
      "step": 10240
    },
    {
      "epoch": 6.552143314139475,
      "grad_norm": 0.9095420759774878,
      "learning_rate": 5.613350250535706e-06,
      "loss": 0.0515,
      "step": 10241
    },
    {
      "epoch": 6.552783109404991,
      "grad_norm": 0.8044438512524246,
      "learning_rate": 5.611488202146296e-06,
      "loss": 0.0498,
      "step": 10242
    },
    {
      "epoch": 6.553422904670505,
      "grad_norm": 1.037088926603698,
      "learning_rate": 5.609626342191772e-06,
      "loss": 0.0513,
      "step": 10243
    },
    {
      "epoch": 6.55406269993602,
      "grad_norm": 0.8128748757143877,
      "learning_rate": 5.607764670752085e-06,
      "loss": 0.0449,
      "step": 10244
    },
    {
      "epoch": 6.554702495201536,
      "grad_norm": 0.9700064700800991,
      "learning_rate": 5.605903187907172e-06,
      "loss": 0.0537,
      "step": 10245
    },
    {
      "epoch": 6.555342290467051,
      "grad_norm": 0.9345787189855987,
      "learning_rate": 5.604041893736955e-06,
      "loss": 0.0539,
      "step": 10246
    },
    {
      "epoch": 6.555982085732565,
      "grad_norm": 0.8451730238570764,
      "learning_rate": 5.602180788321363e-06,
      "loss": 0.0513,
      "step": 10247
    },
    {
      "epoch": 6.5566218809980805,
      "grad_norm": 0.7592905967238738,
      "learning_rate": 5.600319871740305e-06,
      "loss": 0.048,
      "step": 10248
    },
    {
      "epoch": 6.557261676263596,
      "grad_norm": 0.7909182865046998,
      "learning_rate": 5.598459144073682e-06,
      "loss": 0.0451,
      "step": 10249
    },
    {
      "epoch": 6.557901471529111,
      "grad_norm": 0.7766448018056037,
      "learning_rate": 5.596598605401398e-06,
      "loss": 0.0478,
      "step": 10250
    },
    {
      "epoch": 6.5585412667946255,
      "grad_norm": 0.8342988066256812,
      "learning_rate": 5.594738255803336e-06,
      "loss": 0.0524,
      "step": 10251
    },
    {
      "epoch": 6.559181062060141,
      "grad_norm": 0.7880005699363241,
      "learning_rate": 5.592878095359375e-06,
      "loss": 0.0486,
      "step": 10252
    },
    {
      "epoch": 6.559820857325656,
      "grad_norm": 0.8049990018816038,
      "learning_rate": 5.5910181241493925e-06,
      "loss": 0.0512,
      "step": 10253
    },
    {
      "epoch": 6.560460652591171,
      "grad_norm": 0.8379432495219032,
      "learning_rate": 5.589158342253247e-06,
      "loss": 0.0464,
      "step": 10254
    },
    {
      "epoch": 6.561100447856686,
      "grad_norm": 0.8464644826053278,
      "learning_rate": 5.587298749750796e-06,
      "loss": 0.0518,
      "step": 10255
    },
    {
      "epoch": 6.561740243122201,
      "grad_norm": 0.722125957614492,
      "learning_rate": 5.585439346721882e-06,
      "loss": 0.0508,
      "step": 10256
    },
    {
      "epoch": 6.562380038387716,
      "grad_norm": 0.8367186970676297,
      "learning_rate": 5.583580133246355e-06,
      "loss": 0.0465,
      "step": 10257
    },
    {
      "epoch": 6.563019833653231,
      "grad_norm": 0.8084562531281606,
      "learning_rate": 5.581721109404041e-06,
      "loss": 0.0441,
      "step": 10258
    },
    {
      "epoch": 6.563659628918746,
      "grad_norm": 0.7184132814215299,
      "learning_rate": 5.579862275274759e-06,
      "loss": 0.0529,
      "step": 10259
    },
    {
      "epoch": 6.564299424184261,
      "grad_norm": 0.8531705744395933,
      "learning_rate": 5.578003630938331e-06,
      "loss": 0.0495,
      "step": 10260
    },
    {
      "epoch": 6.564939219449776,
      "grad_norm": 0.8197407257027036,
      "learning_rate": 5.576145176474562e-06,
      "loss": 0.0492,
      "step": 10261
    },
    {
      "epoch": 6.565579014715291,
      "grad_norm": 0.720176396701422,
      "learning_rate": 5.574286911963249e-06,
      "loss": 0.0453,
      "step": 10262
    },
    {
      "epoch": 6.566218809980806,
      "grad_norm": 0.9165611054737327,
      "learning_rate": 5.572428837484181e-06,
      "loss": 0.0497,
      "step": 10263
    },
    {
      "epoch": 6.566858605246321,
      "grad_norm": 0.8493457904592623,
      "learning_rate": 5.570570953117145e-06,
      "loss": 0.0491,
      "step": 10264
    },
    {
      "epoch": 6.567498400511836,
      "grad_norm": 0.8479966719465567,
      "learning_rate": 5.5687132589419135e-06,
      "loss": 0.0523,
      "step": 10265
    },
    {
      "epoch": 6.568138195777351,
      "grad_norm": 0.9184387381998083,
      "learning_rate": 5.5668557550382475e-06,
      "loss": 0.0562,
      "step": 10266
    },
    {
      "epoch": 6.568777991042866,
      "grad_norm": 0.8125473101235998,
      "learning_rate": 5.564998441485914e-06,
      "loss": 0.0571,
      "step": 10267
    },
    {
      "epoch": 6.569417786308382,
      "grad_norm": 0.9242810711246574,
      "learning_rate": 5.563141318364657e-06,
      "loss": 0.053,
      "step": 10268
    },
    {
      "epoch": 6.570057581573896,
      "grad_norm": 0.8509697552438725,
      "learning_rate": 5.56128438575422e-06,
      "loss": 0.0497,
      "step": 10269
    },
    {
      "epoch": 6.570697376839411,
      "grad_norm": 0.8473694900826203,
      "learning_rate": 5.559427643734336e-06,
      "loss": 0.0502,
      "step": 10270
    },
    {
      "epoch": 6.5713371721049265,
      "grad_norm": 0.7697676318160378,
      "learning_rate": 5.5575710923847305e-06,
      "loss": 0.0466,
      "step": 10271
    },
    {
      "epoch": 6.571976967370442,
      "grad_norm": 0.8192277483130095,
      "learning_rate": 5.555714731785118e-06,
      "loss": 0.0577,
      "step": 10272
    },
    {
      "epoch": 6.572616762635956,
      "grad_norm": 0.89005382633621,
      "learning_rate": 5.5538585620152065e-06,
      "loss": 0.0536,
      "step": 10273
    },
    {
      "epoch": 6.5732565579014715,
      "grad_norm": 0.7825188031386753,
      "learning_rate": 5.552002583154703e-06,
      "loss": 0.0472,
      "step": 10274
    },
    {
      "epoch": 6.573896353166987,
      "grad_norm": 0.8143252283288379,
      "learning_rate": 5.550146795283295e-06,
      "loss": 0.05,
      "step": 10275
    },
    {
      "epoch": 6.574536148432502,
      "grad_norm": 0.7747188382900679,
      "learning_rate": 5.548291198480665e-06,
      "loss": 0.0438,
      "step": 10276
    },
    {
      "epoch": 6.575175943698016,
      "grad_norm": 0.7562826838974533,
      "learning_rate": 5.546435792826497e-06,
      "loss": 0.0435,
      "step": 10277
    },
    {
      "epoch": 6.575815738963532,
      "grad_norm": 0.8192147693783338,
      "learning_rate": 5.544580578400452e-06,
      "loss": 0.0551,
      "step": 10278
    },
    {
      "epoch": 6.576455534229047,
      "grad_norm": 0.7316436577565486,
      "learning_rate": 5.542725555282188e-06,
      "loss": 0.041,
      "step": 10279
    },
    {
      "epoch": 6.577095329494561,
      "grad_norm": 0.8992402273196035,
      "learning_rate": 5.540870723551364e-06,
      "loss": 0.0473,
      "step": 10280
    },
    {
      "epoch": 6.577735124760077,
      "grad_norm": 0.8329022601407683,
      "learning_rate": 5.539016083287618e-06,
      "loss": 0.0493,
      "step": 10281
    },
    {
      "epoch": 6.578374920025592,
      "grad_norm": 0.9287610006569699,
      "learning_rate": 5.537161634570587e-06,
      "loss": 0.0536,
      "step": 10282
    },
    {
      "epoch": 6.579014715291107,
      "grad_norm": 0.9311372461031262,
      "learning_rate": 5.535307377479897e-06,
      "loss": 0.0563,
      "step": 10283
    },
    {
      "epoch": 6.5796545105566215,
      "grad_norm": 0.7444264497648777,
      "learning_rate": 5.533453312095165e-06,
      "loss": 0.0513,
      "step": 10284
    },
    {
      "epoch": 6.580294305822137,
      "grad_norm": 0.8552273318858333,
      "learning_rate": 5.5315994384960026e-06,
      "loss": 0.0539,
      "step": 10285
    },
    {
      "epoch": 6.580934101087652,
      "grad_norm": 0.8001168529639139,
      "learning_rate": 5.529745756762008e-06,
      "loss": 0.0441,
      "step": 10286
    },
    {
      "epoch": 6.581573896353167,
      "grad_norm": 0.8137714321149578,
      "learning_rate": 5.527892266972783e-06,
      "loss": 0.0435,
      "step": 10287
    },
    {
      "epoch": 6.582213691618682,
      "grad_norm": 0.7972225002139196,
      "learning_rate": 5.526038969207907e-06,
      "loss": 0.0478,
      "step": 10288
    },
    {
      "epoch": 6.582853486884197,
      "grad_norm": 0.7626384032393425,
      "learning_rate": 5.524185863546963e-06,
      "loss": 0.049,
      "step": 10289
    },
    {
      "epoch": 6.583493282149712,
      "grad_norm": 0.851134746367999,
      "learning_rate": 5.522332950069509e-06,
      "loss": 0.047,
      "step": 10290
    },
    {
      "epoch": 6.584133077415228,
      "grad_norm": 0.7086041447599203,
      "learning_rate": 5.520480228855116e-06,
      "loss": 0.0433,
      "step": 10291
    },
    {
      "epoch": 6.584772872680742,
      "grad_norm": 0.7066735087392316,
      "learning_rate": 5.518627699983336e-06,
      "loss": 0.0457,
      "step": 10292
    },
    {
      "epoch": 6.585412667946257,
      "grad_norm": 0.7790994281860194,
      "learning_rate": 5.516775363533706e-06,
      "loss": 0.0492,
      "step": 10293
    },
    {
      "epoch": 6.5860524632117725,
      "grad_norm": 0.8023381549616362,
      "learning_rate": 5.514923219585771e-06,
      "loss": 0.0527,
      "step": 10294
    },
    {
      "epoch": 6.586692258477287,
      "grad_norm": 0.961786433069365,
      "learning_rate": 5.513071268219055e-06,
      "loss": 0.0537,
      "step": 10295
    },
    {
      "epoch": 6.587332053742802,
      "grad_norm": 0.7475814939371649,
      "learning_rate": 5.511219509513072e-06,
      "loss": 0.0538,
      "step": 10296
    },
    {
      "epoch": 6.5879718490083174,
      "grad_norm": 0.8239644052875297,
      "learning_rate": 5.509367943547344e-06,
      "loss": 0.0486,
      "step": 10297
    },
    {
      "epoch": 6.588611644273833,
      "grad_norm": 0.9090151228580955,
      "learning_rate": 5.507516570401366e-06,
      "loss": 0.055,
      "step": 10298
    },
    {
      "epoch": 6.589251439539347,
      "grad_norm": 0.7931034267472228,
      "learning_rate": 5.505665390154636e-06,
      "loss": 0.0546,
      "step": 10299
    },
    {
      "epoch": 6.589891234804862,
      "grad_norm": 0.7948416268601097,
      "learning_rate": 5.50381440288664e-06,
      "loss": 0.0476,
      "step": 10300
    },
    {
      "epoch": 6.590531030070378,
      "grad_norm": 0.8275689328091824,
      "learning_rate": 5.501963608676855e-06,
      "loss": 0.0448,
      "step": 10301
    },
    {
      "epoch": 6.591170825335892,
      "grad_norm": 0.8707426695722487,
      "learning_rate": 5.5001130076047505e-06,
      "loss": 0.0479,
      "step": 10302
    },
    {
      "epoch": 6.591810620601407,
      "grad_norm": 0.8857790062836495,
      "learning_rate": 5.498262599749787e-06,
      "loss": 0.0462,
      "step": 10303
    },
    {
      "epoch": 6.592450415866923,
      "grad_norm": 0.9133390657599201,
      "learning_rate": 5.496412385191421e-06,
      "loss": 0.0487,
      "step": 10304
    },
    {
      "epoch": 6.593090211132438,
      "grad_norm": 0.790778988184272,
      "learning_rate": 5.4945623640090964e-06,
      "loss": 0.0475,
      "step": 10305
    },
    {
      "epoch": 6.593730006397953,
      "grad_norm": 0.7590249478246073,
      "learning_rate": 5.4927125362822455e-06,
      "loss": 0.0433,
      "step": 10306
    },
    {
      "epoch": 6.5943698016634675,
      "grad_norm": 0.791892485785069,
      "learning_rate": 5.4908629020903044e-06,
      "loss": 0.0463,
      "step": 10307
    },
    {
      "epoch": 6.595009596928983,
      "grad_norm": 0.8180153784568713,
      "learning_rate": 5.48901346151269e-06,
      "loss": 0.0464,
      "step": 10308
    },
    {
      "epoch": 6.595649392194498,
      "grad_norm": 0.7628721382930506,
      "learning_rate": 5.487164214628813e-06,
      "loss": 0.0512,
      "step": 10309
    },
    {
      "epoch": 6.5962891874600125,
      "grad_norm": 0.7376295319024481,
      "learning_rate": 5.4853151615180725e-06,
      "loss": 0.044,
      "step": 10310
    },
    {
      "epoch": 6.596928982725528,
      "grad_norm": 0.8374525904765718,
      "learning_rate": 5.483466302259872e-06,
      "loss": 0.0513,
      "step": 10311
    },
    {
      "epoch": 6.597568777991043,
      "grad_norm": 0.9023659241308463,
      "learning_rate": 5.481617636933594e-06,
      "loss": 0.0513,
      "step": 10312
    },
    {
      "epoch": 6.598208573256558,
      "grad_norm": 0.9034954792813689,
      "learning_rate": 5.479769165618618e-06,
      "loss": 0.0516,
      "step": 10313
    },
    {
      "epoch": 6.598848368522073,
      "grad_norm": 0.7513352379921112,
      "learning_rate": 5.4779208883943125e-06,
      "loss": 0.0488,
      "step": 10314
    },
    {
      "epoch": 6.599488163787588,
      "grad_norm": 0.7390349089244744,
      "learning_rate": 5.476072805340039e-06,
      "loss": 0.0428,
      "step": 10315
    },
    {
      "epoch": 6.600127959053103,
      "grad_norm": 0.7414699014878231,
      "learning_rate": 5.474224916535149e-06,
      "loss": 0.0475,
      "step": 10316
    },
    {
      "epoch": 6.600767754318618,
      "grad_norm": 0.8218464744282843,
      "learning_rate": 5.472377222058993e-06,
      "loss": 0.0486,
      "step": 10317
    },
    {
      "epoch": 6.601407549584133,
      "grad_norm": 0.9385062278539802,
      "learning_rate": 5.470529721990907e-06,
      "loss": 0.06,
      "step": 10318
    },
    {
      "epoch": 6.602047344849648,
      "grad_norm": 0.8999439884445157,
      "learning_rate": 5.468682416410216e-06,
      "loss": 0.0467,
      "step": 10319
    },
    {
      "epoch": 6.6026871401151634,
      "grad_norm": 0.7380867573161887,
      "learning_rate": 5.466835305396239e-06,
      "loss": 0.0424,
      "step": 10320
    },
    {
      "epoch": 6.603326935380678,
      "grad_norm": 0.8438240255817037,
      "learning_rate": 5.464988389028292e-06,
      "loss": 0.0551,
      "step": 10321
    },
    {
      "epoch": 6.603966730646193,
      "grad_norm": 0.7539420915645945,
      "learning_rate": 5.4631416673856786e-06,
      "loss": 0.0468,
      "step": 10322
    },
    {
      "epoch": 6.604606525911708,
      "grad_norm": 0.7648315158840969,
      "learning_rate": 5.461295140547687e-06,
      "loss": 0.0503,
      "step": 10323
    },
    {
      "epoch": 6.605246321177224,
      "grad_norm": 0.7902090355705121,
      "learning_rate": 5.459448808593611e-06,
      "loss": 0.0408,
      "step": 10324
    },
    {
      "epoch": 6.605886116442738,
      "grad_norm": 0.8633984844378517,
      "learning_rate": 5.457602671602727e-06,
      "loss": 0.0557,
      "step": 10325
    },
    {
      "epoch": 6.606525911708253,
      "grad_norm": 0.8385796637349971,
      "learning_rate": 5.4557567296543e-06,
      "loss": 0.0573,
      "step": 10326
    },
    {
      "epoch": 6.607165706973769,
      "grad_norm": 0.7875092256633504,
      "learning_rate": 5.453910982827599e-06,
      "loss": 0.049,
      "step": 10327
    },
    {
      "epoch": 6.607805502239284,
      "grad_norm": 0.8954029156101979,
      "learning_rate": 5.452065431201874e-06,
      "loss": 0.0491,
      "step": 10328
    },
    {
      "epoch": 6.608445297504798,
      "grad_norm": 0.7686649762450383,
      "learning_rate": 5.4502200748563695e-06,
      "loss": 0.0544,
      "step": 10329
    },
    {
      "epoch": 6.6090850927703135,
      "grad_norm": 0.8715837912152875,
      "learning_rate": 5.44837491387032e-06,
      "loss": 0.0543,
      "step": 10330
    },
    {
      "epoch": 6.609724888035829,
      "grad_norm": 0.9065748743690986,
      "learning_rate": 5.446529948322956e-06,
      "loss": 0.0523,
      "step": 10331
    },
    {
      "epoch": 6.610364683301343,
      "grad_norm": 0.8093640960550189,
      "learning_rate": 5.444685178293496e-06,
      "loss": 0.0498,
      "step": 10332
    },
    {
      "epoch": 6.6110044785668585,
      "grad_norm": 0.8369960318483324,
      "learning_rate": 5.442840603861148e-06,
      "loss": 0.0465,
      "step": 10333
    },
    {
      "epoch": 6.611644273832374,
      "grad_norm": 0.8827720504594677,
      "learning_rate": 5.440996225105122e-06,
      "loss": 0.0512,
      "step": 10334
    },
    {
      "epoch": 6.612284069097889,
      "grad_norm": 0.671841959932314,
      "learning_rate": 5.4391520421046076e-06,
      "loss": 0.0465,
      "step": 10335
    },
    {
      "epoch": 6.612923864363403,
      "grad_norm": 0.757777885973128,
      "learning_rate": 5.437308054938787e-06,
      "loss": 0.0482,
      "step": 10336
    },
    {
      "epoch": 6.613563659628919,
      "grad_norm": 0.8669536849757571,
      "learning_rate": 5.435464263686847e-06,
      "loss": 0.0554,
      "step": 10337
    },
    {
      "epoch": 6.614203454894434,
      "grad_norm": 0.8231525580948199,
      "learning_rate": 5.43362066842795e-06,
      "loss": 0.0551,
      "step": 10338
    },
    {
      "epoch": 6.614843250159948,
      "grad_norm": 0.8328652417164559,
      "learning_rate": 5.431777269241261e-06,
      "loss": 0.0524,
      "step": 10339
    },
    {
      "epoch": 6.615483045425464,
      "grad_norm": 0.769285487740853,
      "learning_rate": 5.429934066205924e-06,
      "loss": 0.0508,
      "step": 10340
    },
    {
      "epoch": 6.616122840690979,
      "grad_norm": 0.9530674667256792,
      "learning_rate": 5.428091059401093e-06,
      "loss": 0.0532,
      "step": 10341
    },
    {
      "epoch": 6.616762635956494,
      "grad_norm": 0.9573590655514054,
      "learning_rate": 5.4262482489058995e-06,
      "loss": 0.05,
      "step": 10342
    },
    {
      "epoch": 6.617402431222009,
      "grad_norm": 0.7345673222362618,
      "learning_rate": 5.42440563479947e-06,
      "loss": 0.0492,
      "step": 10343
    },
    {
      "epoch": 6.618042226487524,
      "grad_norm": 0.8746587337340719,
      "learning_rate": 5.422563217160923e-06,
      "loss": 0.052,
      "step": 10344
    },
    {
      "epoch": 6.618682021753039,
      "grad_norm": 0.7917318321407946,
      "learning_rate": 5.4207209960693694e-06,
      "loss": 0.0483,
      "step": 10345
    },
    {
      "epoch": 6.619321817018554,
      "grad_norm": 0.7188871939882577,
      "learning_rate": 5.418878971603907e-06,
      "loss": 0.0435,
      "step": 10346
    },
    {
      "epoch": 6.619961612284069,
      "grad_norm": 0.8283604056224837,
      "learning_rate": 5.417037143843635e-06,
      "loss": 0.0528,
      "step": 10347
    },
    {
      "epoch": 6.620601407549584,
      "grad_norm": 0.922533611100723,
      "learning_rate": 5.4151955128676346e-06,
      "loss": 0.0539,
      "step": 10348
    },
    {
      "epoch": 6.621241202815099,
      "grad_norm": 0.748579482167675,
      "learning_rate": 5.413354078754985e-06,
      "loss": 0.0469,
      "step": 10349
    },
    {
      "epoch": 6.621880998080615,
      "grad_norm": 0.8622768465013781,
      "learning_rate": 5.4115128415847484e-06,
      "loss": 0.0454,
      "step": 10350
    },
    {
      "epoch": 6.622520793346129,
      "grad_norm": 0.9145319442105974,
      "learning_rate": 5.409671801435991e-06,
      "loss": 0.0515,
      "step": 10351
    },
    {
      "epoch": 6.623160588611644,
      "grad_norm": 0.7195933991208951,
      "learning_rate": 5.407830958387761e-06,
      "loss": 0.0503,
      "step": 10352
    },
    {
      "epoch": 6.6238003838771595,
      "grad_norm": 0.8646924986424835,
      "learning_rate": 5.405990312519098e-06,
      "loss": 0.0496,
      "step": 10353
    },
    {
      "epoch": 6.624440179142674,
      "grad_norm": 0.7516122325546052,
      "learning_rate": 5.4041498639090415e-06,
      "loss": 0.0452,
      "step": 10354
    },
    {
      "epoch": 6.625079974408189,
      "grad_norm": 0.8115080901606456,
      "learning_rate": 5.402309612636616e-06,
      "loss": 0.0499,
      "step": 10355
    },
    {
      "epoch": 6.6257197696737045,
      "grad_norm": 0.8867622324084231,
      "learning_rate": 5.400469558780832e-06,
      "loss": 0.0476,
      "step": 10356
    },
    {
      "epoch": 6.62635956493922,
      "grad_norm": 0.7764525273810028,
      "learning_rate": 5.398629702420707e-06,
      "loss": 0.0462,
      "step": 10357
    },
    {
      "epoch": 6.626999360204734,
      "grad_norm": 0.8529950636178362,
      "learning_rate": 5.396790043635238e-06,
      "loss": 0.051,
      "step": 10358
    },
    {
      "epoch": 6.627639155470249,
      "grad_norm": 0.8033618126100515,
      "learning_rate": 5.394950582503415e-06,
      "loss": 0.0503,
      "step": 10359
    },
    {
      "epoch": 6.628278950735765,
      "grad_norm": 0.7004339355186411,
      "learning_rate": 5.3931113191042225e-06,
      "loss": 0.0463,
      "step": 10360
    },
    {
      "epoch": 6.62891874600128,
      "grad_norm": 0.8552280371821913,
      "learning_rate": 5.3912722535166355e-06,
      "loss": 0.0544,
      "step": 10361
    },
    {
      "epoch": 6.629558541266794,
      "grad_norm": 0.8200169986768402,
      "learning_rate": 5.38943338581962e-06,
      "loss": 0.0529,
      "step": 10362
    },
    {
      "epoch": 6.63019833653231,
      "grad_norm": 0.8026938017296128,
      "learning_rate": 5.387594716092128e-06,
      "loss": 0.0475,
      "step": 10363
    },
    {
      "epoch": 6.630838131797825,
      "grad_norm": 0.785112676657086,
      "learning_rate": 5.385756244413119e-06,
      "loss": 0.0469,
      "step": 10364
    },
    {
      "epoch": 6.63147792706334,
      "grad_norm": 0.966782240395898,
      "learning_rate": 5.38391797086153e-06,
      "loss": 0.051,
      "step": 10365
    },
    {
      "epoch": 6.6321177223288545,
      "grad_norm": 0.8990261020061275,
      "learning_rate": 5.382079895516285e-06,
      "loss": 0.0552,
      "step": 10366
    },
    {
      "epoch": 6.63275751759437,
      "grad_norm": 0.8881687012850851,
      "learning_rate": 5.380242018456321e-06,
      "loss": 0.0564,
      "step": 10367
    },
    {
      "epoch": 6.633397312859885,
      "grad_norm": 0.8876639521416257,
      "learning_rate": 5.378404339760546e-06,
      "loss": 0.0503,
      "step": 10368
    },
    {
      "epoch": 6.6340371081253995,
      "grad_norm": 0.9321098484430381,
      "learning_rate": 5.376566859507868e-06,
      "loss": 0.051,
      "step": 10369
    },
    {
      "epoch": 6.634676903390915,
      "grad_norm": 0.7944625000071108,
      "learning_rate": 5.37472957777718e-06,
      "loss": 0.0549,
      "step": 10370
    },
    {
      "epoch": 6.63531669865643,
      "grad_norm": 0.7998917709517775,
      "learning_rate": 5.3728924946473804e-06,
      "loss": 0.0453,
      "step": 10371
    },
    {
      "epoch": 6.635956493921945,
      "grad_norm": 0.9696612274745292,
      "learning_rate": 5.371055610197347e-06,
      "loss": 0.0549,
      "step": 10372
    },
    {
      "epoch": 6.63659628918746,
      "grad_norm": 0.9642066009401137,
      "learning_rate": 5.36921892450595e-06,
      "loss": 0.0476,
      "step": 10373
    },
    {
      "epoch": 6.637236084452975,
      "grad_norm": 0.8405011620118334,
      "learning_rate": 5.3673824376520555e-06,
      "loss": 0.0482,
      "step": 10374
    },
    {
      "epoch": 6.63787587971849,
      "grad_norm": 0.9379360910546963,
      "learning_rate": 5.365546149714518e-06,
      "loss": 0.0533,
      "step": 10375
    },
    {
      "epoch": 6.638515674984005,
      "grad_norm": 0.7732135720074899,
      "learning_rate": 5.3637100607721845e-06,
      "loss": 0.0405,
      "step": 10376
    },
    {
      "epoch": 6.63915547024952,
      "grad_norm": 0.7195623877935773,
      "learning_rate": 5.3618741709038904e-06,
      "loss": 0.0422,
      "step": 10377
    },
    {
      "epoch": 6.639795265515035,
      "grad_norm": 0.8989258526359253,
      "learning_rate": 5.360038480188472e-06,
      "loss": 0.0524,
      "step": 10378
    },
    {
      "epoch": 6.6404350607805505,
      "grad_norm": 0.9198214474109357,
      "learning_rate": 5.358202988704748e-06,
      "loss": 0.0626,
      "step": 10379
    },
    {
      "epoch": 6.641074856046066,
      "grad_norm": 0.674585595128488,
      "learning_rate": 5.356367696531527e-06,
      "loss": 0.0362,
      "step": 10380
    },
    {
      "epoch": 6.64171465131158,
      "grad_norm": 0.7320525245091296,
      "learning_rate": 5.35453260374762e-06,
      "loss": 0.0438,
      "step": 10381
    },
    {
      "epoch": 6.642354446577095,
      "grad_norm": 0.7651774311933762,
      "learning_rate": 5.35269771043182e-06,
      "loss": 0.0424,
      "step": 10382
    },
    {
      "epoch": 6.642994241842611,
      "grad_norm": 0.894904294309028,
      "learning_rate": 5.3508630166629096e-06,
      "loss": 0.0531,
      "step": 10383
    },
    {
      "epoch": 6.643634037108125,
      "grad_norm": 0.8880315080012571,
      "learning_rate": 5.349028522519675e-06,
      "loss": 0.0537,
      "step": 10384
    },
    {
      "epoch": 6.64427383237364,
      "grad_norm": 0.9032314489615529,
      "learning_rate": 5.347194228080881e-06,
      "loss": 0.0485,
      "step": 10385
    },
    {
      "epoch": 6.644913627639156,
      "grad_norm": 0.8923345201172597,
      "learning_rate": 5.34536013342529e-06,
      "loss": 0.051,
      "step": 10386
    },
    {
      "epoch": 6.645553422904671,
      "grad_norm": 0.8460728072793766,
      "learning_rate": 5.343526238631653e-06,
      "loss": 0.0493,
      "step": 10387
    },
    {
      "epoch": 6.646193218170185,
      "grad_norm": 0.8551039518189449,
      "learning_rate": 5.341692543778718e-06,
      "loss": 0.0503,
      "step": 10388
    },
    {
      "epoch": 6.6468330134357005,
      "grad_norm": 0.8233152090837185,
      "learning_rate": 5.339859048945219e-06,
      "loss": 0.0504,
      "step": 10389
    },
    {
      "epoch": 6.647472808701216,
      "grad_norm": 0.8417843915923197,
      "learning_rate": 5.338025754209883e-06,
      "loss": 0.0501,
      "step": 10390
    },
    {
      "epoch": 6.64811260396673,
      "grad_norm": 0.9865385937635309,
      "learning_rate": 5.336192659651429e-06,
      "loss": 0.0554,
      "step": 10391
    },
    {
      "epoch": 6.6487523992322455,
      "grad_norm": 0.8007244012137931,
      "learning_rate": 5.334359765348566e-06,
      "loss": 0.0478,
      "step": 10392
    },
    {
      "epoch": 6.649392194497761,
      "grad_norm": 0.8753071332059786,
      "learning_rate": 5.332527071379991e-06,
      "loss": 0.0564,
      "step": 10393
    },
    {
      "epoch": 6.650031989763276,
      "grad_norm": 0.7352106202332884,
      "learning_rate": 5.330694577824403e-06,
      "loss": 0.0465,
      "step": 10394
    },
    {
      "epoch": 6.65067178502879,
      "grad_norm": 0.7573974673238671,
      "learning_rate": 5.3288622847604856e-06,
      "loss": 0.0443,
      "step": 10395
    },
    {
      "epoch": 6.651311580294306,
      "grad_norm": 0.814797651382131,
      "learning_rate": 5.327030192266911e-06,
      "loss": 0.0547,
      "step": 10396
    },
    {
      "epoch": 6.651951375559821,
      "grad_norm": 0.8044752466715498,
      "learning_rate": 5.3251983004223455e-06,
      "loss": 0.0509,
      "step": 10397
    },
    {
      "epoch": 6.652591170825336,
      "grad_norm": 0.9219091323695148,
      "learning_rate": 5.323366609305451e-06,
      "loss": 0.0506,
      "step": 10398
    },
    {
      "epoch": 6.653230966090851,
      "grad_norm": 0.8272710399119642,
      "learning_rate": 5.321535118994876e-06,
      "loss": 0.0462,
      "step": 10399
    },
    {
      "epoch": 6.653870761356366,
      "grad_norm": 0.8037158554175982,
      "learning_rate": 5.319703829569256e-06,
      "loss": 0.053,
      "step": 10400
    },
    {
      "epoch": 6.654510556621881,
      "grad_norm": 0.8741236911196406,
      "learning_rate": 5.317872741107233e-06,
      "loss": 0.0447,
      "step": 10401
    },
    {
      "epoch": 6.6551503518873965,
      "grad_norm": 0.8816227296969866,
      "learning_rate": 5.316041853687426e-06,
      "loss": 0.0544,
      "step": 10402
    },
    {
      "epoch": 6.655790147152911,
      "grad_norm": 0.8055110194868151,
      "learning_rate": 5.314211167388448e-06,
      "loss": 0.0505,
      "step": 10403
    },
    {
      "epoch": 6.656429942418426,
      "grad_norm": 0.8054291973889393,
      "learning_rate": 5.312380682288909e-06,
      "loss": 0.0493,
      "step": 10404
    },
    {
      "epoch": 6.657069737683941,
      "grad_norm": 0.96686024229896,
      "learning_rate": 5.310550398467403e-06,
      "loss": 0.0559,
      "step": 10405
    },
    {
      "epoch": 6.657709532949456,
      "grad_norm": 0.7845607988470105,
      "learning_rate": 5.3087203160025226e-06,
      "loss": 0.0519,
      "step": 10406
    },
    {
      "epoch": 6.658349328214971,
      "grad_norm": 0.8471511131478474,
      "learning_rate": 5.3068904349728425e-06,
      "loss": 0.0531,
      "step": 10407
    },
    {
      "epoch": 6.658989123480486,
      "grad_norm": 0.9039666784591042,
      "learning_rate": 5.305060755456941e-06,
      "loss": 0.051,
      "step": 10408
    },
    {
      "epoch": 6.659628918746002,
      "grad_norm": 0.8339809537521807,
      "learning_rate": 5.30323127753338e-06,
      "loss": 0.0559,
      "step": 10409
    },
    {
      "epoch": 6.660268714011516,
      "grad_norm": 0.8735379201087778,
      "learning_rate": 5.30140200128071e-06,
      "loss": 0.0562,
      "step": 10410
    },
    {
      "epoch": 6.660908509277031,
      "grad_norm": 0.8840327197376415,
      "learning_rate": 5.299572926777483e-06,
      "loss": 0.044,
      "step": 10411
    },
    {
      "epoch": 6.6615483045425465,
      "grad_norm": 0.8526535351667813,
      "learning_rate": 5.297744054102234e-06,
      "loss": 0.0502,
      "step": 10412
    },
    {
      "epoch": 6.662188099808061,
      "grad_norm": 0.7781040945046871,
      "learning_rate": 5.295915383333485e-06,
      "loss": 0.0517,
      "step": 10413
    },
    {
      "epoch": 6.662827895073576,
      "grad_norm": 0.887488141103646,
      "learning_rate": 5.294086914549767e-06,
      "loss": 0.05,
      "step": 10414
    },
    {
      "epoch": 6.6634676903390915,
      "grad_norm": 0.7956871094239835,
      "learning_rate": 5.292258647829585e-06,
      "loss": 0.0453,
      "step": 10415
    },
    {
      "epoch": 6.664107485604607,
      "grad_norm": 0.7993573848751597,
      "learning_rate": 5.2904305832514424e-06,
      "loss": 0.0511,
      "step": 10416
    },
    {
      "epoch": 6.664747280870122,
      "grad_norm": 0.8755313521474117,
      "learning_rate": 5.288602720893829e-06,
      "loss": 0.0533,
      "step": 10417
    },
    {
      "epoch": 6.665387076135636,
      "grad_norm": 0.8622948757003455,
      "learning_rate": 5.286775060835238e-06,
      "loss": 0.0537,
      "step": 10418
    },
    {
      "epoch": 6.666026871401152,
      "grad_norm": 0.7326321772465286,
      "learning_rate": 5.284947603154141e-06,
      "loss": 0.0449,
      "step": 10419
    },
    {
      "epoch": 6.666666666666667,
      "grad_norm": 0.8106347993204668,
      "learning_rate": 5.283120347929007e-06,
      "loss": 0.0479,
      "step": 10420
    },
    {
      "epoch": 6.667306461932181,
      "grad_norm": 0.9667484153492252,
      "learning_rate": 5.281293295238295e-06,
      "loss": 0.0593,
      "step": 10421
    },
    {
      "epoch": 6.667946257197697,
      "grad_norm": 0.9066576078238061,
      "learning_rate": 5.279466445160454e-06,
      "loss": 0.0513,
      "step": 10422
    },
    {
      "epoch": 6.668586052463212,
      "grad_norm": 0.8084952228611627,
      "learning_rate": 5.277639797773923e-06,
      "loss": 0.054,
      "step": 10423
    },
    {
      "epoch": 6.669225847728727,
      "grad_norm": 0.833220101159631,
      "learning_rate": 5.275813353157143e-06,
      "loss": 0.054,
      "step": 10424
    },
    {
      "epoch": 6.669865642994242,
      "grad_norm": 0.9366720098072959,
      "learning_rate": 5.273987111388534e-06,
      "loss": 0.055,
      "step": 10425
    },
    {
      "epoch": 6.670505438259757,
      "grad_norm": 0.7678339180428798,
      "learning_rate": 5.272161072546511e-06,
      "loss": 0.0477,
      "step": 10426
    },
    {
      "epoch": 6.671145233525272,
      "grad_norm": 0.8190530488691945,
      "learning_rate": 5.270335236709479e-06,
      "loss": 0.048,
      "step": 10427
    },
    {
      "epoch": 6.6717850287907865,
      "grad_norm": 0.7132677623205597,
      "learning_rate": 5.26850960395584e-06,
      "loss": 0.0468,
      "step": 10428
    },
    {
      "epoch": 6.672424824056302,
      "grad_norm": 0.8114930775130875,
      "learning_rate": 5.266684174363983e-06,
      "loss": 0.0492,
      "step": 10429
    },
    {
      "epoch": 6.673064619321817,
      "grad_norm": 0.9427619278882113,
      "learning_rate": 5.264858948012284e-06,
      "loss": 0.0573,
      "step": 10430
    },
    {
      "epoch": 6.673704414587332,
      "grad_norm": 0.8353305572097498,
      "learning_rate": 5.263033924979122e-06,
      "loss": 0.0495,
      "step": 10431
    },
    {
      "epoch": 6.674344209852847,
      "grad_norm": 0.8538146685500771,
      "learning_rate": 5.261209105342857e-06,
      "loss": 0.0549,
      "step": 10432
    },
    {
      "epoch": 6.674984005118362,
      "grad_norm": 0.8517190388232008,
      "learning_rate": 5.259384489181844e-06,
      "loss": 0.0456,
      "step": 10433
    },
    {
      "epoch": 6.675623800383877,
      "grad_norm": 0.7943513316953679,
      "learning_rate": 5.257560076574428e-06,
      "loss": 0.0496,
      "step": 10434
    },
    {
      "epoch": 6.6762635956493925,
      "grad_norm": 0.8154027313176505,
      "learning_rate": 5.255735867598946e-06,
      "loss": 0.0497,
      "step": 10435
    },
    {
      "epoch": 6.676903390914907,
      "grad_norm": 0.7450326298788338,
      "learning_rate": 5.253911862333726e-06,
      "loss": 0.0459,
      "step": 10436
    },
    {
      "epoch": 6.677543186180422,
      "grad_norm": 0.7323932845298313,
      "learning_rate": 5.252088060857083e-06,
      "loss": 0.0479,
      "step": 10437
    },
    {
      "epoch": 6.6781829814459375,
      "grad_norm": 0.9475984262545712,
      "learning_rate": 5.250264463247339e-06,
      "loss": 0.0526,
      "step": 10438
    },
    {
      "epoch": 6.678822776711453,
      "grad_norm": 0.9178771159984987,
      "learning_rate": 5.248441069582789e-06,
      "loss": 0.0548,
      "step": 10439
    },
    {
      "epoch": 6.679462571976967,
      "grad_norm": 0.845460641361275,
      "learning_rate": 5.246617879941724e-06,
      "loss": 0.0457,
      "step": 10440
    },
    {
      "epoch": 6.680102367242482,
      "grad_norm": 0.7948904483851973,
      "learning_rate": 5.244794894402434e-06,
      "loss": 0.049,
      "step": 10441
    },
    {
      "epoch": 6.680742162507998,
      "grad_norm": 0.8267000498826995,
      "learning_rate": 5.242972113043193e-06,
      "loss": 0.0488,
      "step": 10442
    },
    {
      "epoch": 6.681381957773512,
      "grad_norm": 0.7678732914908376,
      "learning_rate": 5.2411495359422635e-06,
      "loss": 0.046,
      "step": 10443
    },
    {
      "epoch": 6.682021753039027,
      "grad_norm": 0.8317318758309417,
      "learning_rate": 5.239327163177912e-06,
      "loss": 0.0542,
      "step": 10444
    },
    {
      "epoch": 6.682661548304543,
      "grad_norm": 0.8323943064217519,
      "learning_rate": 5.237504994828382e-06,
      "loss": 0.0545,
      "step": 10445
    },
    {
      "epoch": 6.683301343570058,
      "grad_norm": 0.735531482192476,
      "learning_rate": 5.235683030971916e-06,
      "loss": 0.0484,
      "step": 10446
    },
    {
      "epoch": 6.683941138835572,
      "grad_norm": 0.803775366094278,
      "learning_rate": 5.2338612716867445e-06,
      "loss": 0.0472,
      "step": 10447
    },
    {
      "epoch": 6.684580934101088,
      "grad_norm": 0.8063709840732415,
      "learning_rate": 5.232039717051093e-06,
      "loss": 0.0474,
      "step": 10448
    },
    {
      "epoch": 6.685220729366603,
      "grad_norm": 0.8125196713394746,
      "learning_rate": 5.230218367143174e-06,
      "loss": 0.0524,
      "step": 10449
    },
    {
      "epoch": 6.685860524632118,
      "grad_norm": 0.8769958874200564,
      "learning_rate": 5.228397222041195e-06,
      "loss": 0.0467,
      "step": 10450
    },
    {
      "epoch": 6.6865003198976325,
      "grad_norm": 0.800719033026391,
      "learning_rate": 5.22657628182335e-06,
      "loss": 0.048,
      "step": 10451
    },
    {
      "epoch": 6.687140115163148,
      "grad_norm": 0.8508268918707035,
      "learning_rate": 5.22475554656783e-06,
      "loss": 0.0534,
      "step": 10452
    },
    {
      "epoch": 6.687779910428663,
      "grad_norm": 0.7447708770328773,
      "learning_rate": 5.222935016352812e-06,
      "loss": 0.0445,
      "step": 10453
    },
    {
      "epoch": 6.688419705694178,
      "grad_norm": 0.8087679865240803,
      "learning_rate": 5.221114691256461e-06,
      "loss": 0.0475,
      "step": 10454
    },
    {
      "epoch": 6.689059500959693,
      "grad_norm": 0.8350423851638872,
      "learning_rate": 5.21929457135695e-06,
      "loss": 0.0518,
      "step": 10455
    },
    {
      "epoch": 6.689699296225208,
      "grad_norm": 0.6599274033341108,
      "learning_rate": 5.217474656732424e-06,
      "loss": 0.0416,
      "step": 10456
    },
    {
      "epoch": 6.690339091490723,
      "grad_norm": 0.8726095848175972,
      "learning_rate": 5.2156549474610265e-06,
      "loss": 0.0577,
      "step": 10457
    },
    {
      "epoch": 6.690978886756238,
      "grad_norm": 0.9694200198151953,
      "learning_rate": 5.213835443620898e-06,
      "loss": 0.0554,
      "step": 10458
    },
    {
      "epoch": 6.691618682021753,
      "grad_norm": 0.7197563529514401,
      "learning_rate": 5.212016145290162e-06,
      "loss": 0.0467,
      "step": 10459
    },
    {
      "epoch": 6.692258477287268,
      "grad_norm": 0.7145016373167513,
      "learning_rate": 5.210197052546931e-06,
      "loss": 0.0435,
      "step": 10460
    },
    {
      "epoch": 6.6928982725527835,
      "grad_norm": 0.8070908809851421,
      "learning_rate": 5.2083781654693234e-06,
      "loss": 0.0494,
      "step": 10461
    },
    {
      "epoch": 6.693538067818298,
      "grad_norm": 0.8920249574929512,
      "learning_rate": 5.2065594841354315e-06,
      "loss": 0.0546,
      "step": 10462
    },
    {
      "epoch": 6.694177863083813,
      "grad_norm": 0.9115832163728308,
      "learning_rate": 5.204741008623351e-06,
      "loss": 0.0501,
      "step": 10463
    },
    {
      "epoch": 6.694817658349328,
      "grad_norm": 0.7626137873235042,
      "learning_rate": 5.202922739011161e-06,
      "loss": 0.0484,
      "step": 10464
    },
    {
      "epoch": 6.695457453614843,
      "grad_norm": 0.7545413185674427,
      "learning_rate": 5.201104675376933e-06,
      "loss": 0.0429,
      "step": 10465
    },
    {
      "epoch": 6.696097248880358,
      "grad_norm": 0.8749706879407645,
      "learning_rate": 5.199286817798736e-06,
      "loss": 0.0502,
      "step": 10466
    },
    {
      "epoch": 6.696737044145873,
      "grad_norm": 0.9460964543117408,
      "learning_rate": 5.197469166354619e-06,
      "loss": 0.0579,
      "step": 10467
    },
    {
      "epoch": 6.697376839411389,
      "grad_norm": 0.7646691263368725,
      "learning_rate": 5.1956517211226366e-06,
      "loss": 0.0497,
      "step": 10468
    },
    {
      "epoch": 6.698016634676903,
      "grad_norm": 0.8747534672877306,
      "learning_rate": 5.193834482180822e-06,
      "loss": 0.0529,
      "step": 10469
    },
    {
      "epoch": 6.698656429942418,
      "grad_norm": 0.9162999898899898,
      "learning_rate": 5.192017449607203e-06,
      "loss": 0.055,
      "step": 10470
    },
    {
      "epoch": 6.6992962252079336,
      "grad_norm": 0.7867715334555255,
      "learning_rate": 5.1902006234798045e-06,
      "loss": 0.0478,
      "step": 10471
    },
    {
      "epoch": 6.699936020473449,
      "grad_norm": 0.7010243202495785,
      "learning_rate": 5.188384003876637e-06,
      "loss": 0.044,
      "step": 10472
    },
    {
      "epoch": 6.700575815738963,
      "grad_norm": 0.8250624502951593,
      "learning_rate": 5.1865675908757e-06,
      "loss": 0.0523,
      "step": 10473
    },
    {
      "epoch": 6.7012156110044785,
      "grad_norm": 0.8381076677386307,
      "learning_rate": 5.184751384554983e-06,
      "loss": 0.0459,
      "step": 10474
    },
    {
      "epoch": 6.701855406269994,
      "grad_norm": 0.8202771261064226,
      "learning_rate": 5.182935384992482e-06,
      "loss": 0.049,
      "step": 10475
    },
    {
      "epoch": 6.702495201535509,
      "grad_norm": 0.7403393593929237,
      "learning_rate": 5.181119592266165e-06,
      "loss": 0.0484,
      "step": 10476
    },
    {
      "epoch": 6.703134996801023,
      "grad_norm": 0.8366119647476431,
      "learning_rate": 5.179304006453998e-06,
      "loss": 0.051,
      "step": 10477
    },
    {
      "epoch": 6.703774792066539,
      "grad_norm": 0.8550008097459514,
      "learning_rate": 5.177488627633944e-06,
      "loss": 0.0447,
      "step": 10478
    },
    {
      "epoch": 6.704414587332054,
      "grad_norm": 0.8493480608110192,
      "learning_rate": 5.175673455883952e-06,
      "loss": 0.0504,
      "step": 10479
    },
    {
      "epoch": 6.705054382597568,
      "grad_norm": 0.7492846231306757,
      "learning_rate": 5.173858491281958e-06,
      "loss": 0.0464,
      "step": 10480
    },
    {
      "epoch": 6.705694177863084,
      "grad_norm": 0.7970716378706013,
      "learning_rate": 5.172043733905896e-06,
      "loss": 0.0487,
      "step": 10481
    },
    {
      "epoch": 6.706333973128599,
      "grad_norm": 0.7989712274013491,
      "learning_rate": 5.1702291838336885e-06,
      "loss": 0.0487,
      "step": 10482
    },
    {
      "epoch": 6.706973768394114,
      "grad_norm": 0.7906937025094025,
      "learning_rate": 5.168414841143248e-06,
      "loss": 0.0488,
      "step": 10483
    },
    {
      "epoch": 6.707613563659629,
      "grad_norm": 0.7916731654286466,
      "learning_rate": 5.166600705912475e-06,
      "loss": 0.0464,
      "step": 10484
    },
    {
      "epoch": 6.708253358925144,
      "grad_norm": 0.8553947310345231,
      "learning_rate": 5.1647867782192725e-06,
      "loss": 0.052,
      "step": 10485
    },
    {
      "epoch": 6.708893154190659,
      "grad_norm": 0.7838263002329023,
      "learning_rate": 5.162973058141527e-06,
      "loss": 0.0509,
      "step": 10486
    },
    {
      "epoch": 6.709532949456174,
      "grad_norm": 0.8228936713907874,
      "learning_rate": 5.161159545757109e-06,
      "loss": 0.0541,
      "step": 10487
    },
    {
      "epoch": 6.710172744721689,
      "grad_norm": 0.7863712524517129,
      "learning_rate": 5.159346241143897e-06,
      "loss": 0.0484,
      "step": 10488
    },
    {
      "epoch": 6.710812539987204,
      "grad_norm": 0.9009089674337454,
      "learning_rate": 5.157533144379746e-06,
      "loss": 0.052,
      "step": 10489
    },
    {
      "epoch": 6.711452335252719,
      "grad_norm": 0.8864450870199166,
      "learning_rate": 5.155720255542503e-06,
      "loss": 0.0506,
      "step": 10490
    },
    {
      "epoch": 6.712092130518235,
      "grad_norm": 0.9094429825930066,
      "learning_rate": 5.15390757471002e-06,
      "loss": 0.0521,
      "step": 10491
    },
    {
      "epoch": 6.712731925783749,
      "grad_norm": 0.8355259663350236,
      "learning_rate": 5.1520951019601255e-06,
      "loss": 0.0537,
      "step": 10492
    },
    {
      "epoch": 6.713371721049264,
      "grad_norm": 0.9583399933488335,
      "learning_rate": 5.150282837370643e-06,
      "loss": 0.0571,
      "step": 10493
    },
    {
      "epoch": 6.7140115163147795,
      "grad_norm": 0.9709664730881289,
      "learning_rate": 5.148470781019388e-06,
      "loss": 0.0534,
      "step": 10494
    },
    {
      "epoch": 6.714651311580294,
      "grad_norm": 0.793171833161329,
      "learning_rate": 5.1466589329841675e-06,
      "loss": 0.0469,
      "step": 10495
    },
    {
      "epoch": 6.715291106845809,
      "grad_norm": 0.9281285861981092,
      "learning_rate": 5.14484729334278e-06,
      "loss": 0.0535,
      "step": 10496
    },
    {
      "epoch": 6.7159309021113245,
      "grad_norm": 0.9730028953945334,
      "learning_rate": 5.143035862173008e-06,
      "loss": 0.0535,
      "step": 10497
    },
    {
      "epoch": 6.71657069737684,
      "grad_norm": 0.9005947529192853,
      "learning_rate": 5.141224639552641e-06,
      "loss": 0.0465,
      "step": 10498
    },
    {
      "epoch": 6.717210492642354,
      "grad_norm": 0.9298404731187724,
      "learning_rate": 5.139413625559444e-06,
      "loss": 0.0519,
      "step": 10499
    },
    {
      "epoch": 6.717850287907869,
      "grad_norm": 0.814670622774817,
      "learning_rate": 5.137602820271177e-06,
      "loss": 0.0474,
      "step": 10500
    },
    {
      "epoch": 6.718490083173385,
      "grad_norm": 0.796016335767336,
      "learning_rate": 5.135792223765599e-06,
      "loss": 0.0479,
      "step": 10501
    },
    {
      "epoch": 6.719129878438899,
      "grad_norm": 0.6793106837718115,
      "learning_rate": 5.13398183612045e-06,
      "loss": 0.0442,
      "step": 10502
    },
    {
      "epoch": 6.719769673704414,
      "grad_norm": 0.820284867360053,
      "learning_rate": 5.132171657413464e-06,
      "loss": 0.0493,
      "step": 10503
    },
    {
      "epoch": 6.72040946896993,
      "grad_norm": 0.7552337196473989,
      "learning_rate": 5.130361687722365e-06,
      "loss": 0.0459,
      "step": 10504
    },
    {
      "epoch": 6.721049264235445,
      "grad_norm": 0.7277700840839181,
      "learning_rate": 5.1285519271248765e-06,
      "loss": 0.0476,
      "step": 10505
    },
    {
      "epoch": 6.72168905950096,
      "grad_norm": 0.7843663343175966,
      "learning_rate": 5.126742375698701e-06,
      "loss": 0.0464,
      "step": 10506
    },
    {
      "epoch": 6.722328854766475,
      "grad_norm": 0.7829140488183102,
      "learning_rate": 5.124933033521537e-06,
      "loss": 0.0475,
      "step": 10507
    },
    {
      "epoch": 6.72296865003199,
      "grad_norm": 0.8190900307427328,
      "learning_rate": 5.12312390067108e-06,
      "loss": 0.0477,
      "step": 10508
    },
    {
      "epoch": 6.723608445297505,
      "grad_norm": 0.7853982529350151,
      "learning_rate": 5.121314977225007e-06,
      "loss": 0.0454,
      "step": 10509
    },
    {
      "epoch": 6.7242482405630195,
      "grad_norm": 0.8996223438537262,
      "learning_rate": 5.1195062632609906e-06,
      "loss": 0.0541,
      "step": 10510
    },
    {
      "epoch": 6.724888035828535,
      "grad_norm": 0.9904042763102121,
      "learning_rate": 5.117697758856693e-06,
      "loss": 0.0603,
      "step": 10511
    },
    {
      "epoch": 6.72552783109405,
      "grad_norm": 0.8309503175673266,
      "learning_rate": 5.11588946408977e-06,
      "loss": 0.0504,
      "step": 10512
    },
    {
      "epoch": 6.726167626359565,
      "grad_norm": 0.821668386813267,
      "learning_rate": 5.114081379037864e-06,
      "loss": 0.0497,
      "step": 10513
    },
    {
      "epoch": 6.72680742162508,
      "grad_norm": 0.7563442365395813,
      "learning_rate": 5.112273503778609e-06,
      "loss": 0.0428,
      "step": 10514
    },
    {
      "epoch": 6.727447216890595,
      "grad_norm": 0.8139314856701269,
      "learning_rate": 5.110465838389639e-06,
      "loss": 0.0482,
      "step": 10515
    },
    {
      "epoch": 6.72808701215611,
      "grad_norm": 0.8550808922471739,
      "learning_rate": 5.108658382948568e-06,
      "loss": 0.051,
      "step": 10516
    },
    {
      "epoch": 6.728726807421625,
      "grad_norm": 0.8271632007864509,
      "learning_rate": 5.106851137533001e-06,
      "loss": 0.0485,
      "step": 10517
    },
    {
      "epoch": 6.72936660268714,
      "grad_norm": 0.8382817093448498,
      "learning_rate": 5.105044102220546e-06,
      "loss": 0.0467,
      "step": 10518
    },
    {
      "epoch": 6.730006397952655,
      "grad_norm": 0.865213903011685,
      "learning_rate": 5.1032372770887895e-06,
      "loss": 0.0447,
      "step": 10519
    },
    {
      "epoch": 6.7306461932181705,
      "grad_norm": 0.7957262760822141,
      "learning_rate": 5.1014306622153095e-06,
      "loss": 0.0471,
      "step": 10520
    },
    {
      "epoch": 6.731285988483685,
      "grad_norm": 0.8747184848406128,
      "learning_rate": 5.099624257677688e-06,
      "loss": 0.0507,
      "step": 10521
    },
    {
      "epoch": 6.7319257837492,
      "grad_norm": 0.7917013151343056,
      "learning_rate": 5.097818063553482e-06,
      "loss": 0.0486,
      "step": 10522
    },
    {
      "epoch": 6.732565579014715,
      "grad_norm": 0.8146798201996244,
      "learning_rate": 5.096012079920249e-06,
      "loss": 0.0512,
      "step": 10523
    },
    {
      "epoch": 6.733205374280231,
      "grad_norm": 0.8191971874466863,
      "learning_rate": 5.094206306855535e-06,
      "loss": 0.0486,
      "step": 10524
    },
    {
      "epoch": 6.733845169545745,
      "grad_norm": 0.844229920150335,
      "learning_rate": 5.092400744436874e-06,
      "loss": 0.0521,
      "step": 10525
    },
    {
      "epoch": 6.73448496481126,
      "grad_norm": 0.7726862278711092,
      "learning_rate": 5.090595392741795e-06,
      "loss": 0.0451,
      "step": 10526
    },
    {
      "epoch": 6.735124760076776,
      "grad_norm": 0.8550239184358793,
      "learning_rate": 5.088790251847814e-06,
      "loss": 0.0495,
      "step": 10527
    },
    {
      "epoch": 6.735764555342291,
      "grad_norm": 0.800003410571151,
      "learning_rate": 5.086985321832446e-06,
      "loss": 0.0505,
      "step": 10528
    },
    {
      "epoch": 6.736404350607805,
      "grad_norm": 0.7316283888877122,
      "learning_rate": 5.08518060277319e-06,
      "loss": 0.046,
      "step": 10529
    },
    {
      "epoch": 6.737044145873321,
      "grad_norm": 0.7608947935799738,
      "learning_rate": 5.083376094747534e-06,
      "loss": 0.053,
      "step": 10530
    },
    {
      "epoch": 6.737683941138836,
      "grad_norm": 0.799207012246887,
      "learning_rate": 5.08157179783296e-06,
      "loss": 0.0517,
      "step": 10531
    },
    {
      "epoch": 6.73832373640435,
      "grad_norm": 0.8141162429983729,
      "learning_rate": 5.079767712106948e-06,
      "loss": 0.0561,
      "step": 10532
    },
    {
      "epoch": 6.7389635316698655,
      "grad_norm": 0.8617039110716638,
      "learning_rate": 5.077963837646957e-06,
      "loss": 0.0497,
      "step": 10533
    },
    {
      "epoch": 6.739603326935381,
      "grad_norm": 0.7553246046013926,
      "learning_rate": 5.07616017453044e-06,
      "loss": 0.0473,
      "step": 10534
    },
    {
      "epoch": 6.740243122200896,
      "grad_norm": 0.8492734150779802,
      "learning_rate": 5.07435672283485e-06,
      "loss": 0.051,
      "step": 10535
    },
    {
      "epoch": 6.74088291746641,
      "grad_norm": 0.8825141958614048,
      "learning_rate": 5.072553482637621e-06,
      "loss": 0.0496,
      "step": 10536
    },
    {
      "epoch": 6.741522712731926,
      "grad_norm": 0.7162828276340585,
      "learning_rate": 5.070750454016176e-06,
      "loss": 0.0435,
      "step": 10537
    },
    {
      "epoch": 6.742162507997441,
      "grad_norm": 0.8130100247151522,
      "learning_rate": 5.068947637047941e-06,
      "loss": 0.0486,
      "step": 10538
    },
    {
      "epoch": 6.742802303262955,
      "grad_norm": 0.8436389943263358,
      "learning_rate": 5.067145031810323e-06,
      "loss": 0.0571,
      "step": 10539
    },
    {
      "epoch": 6.743442098528471,
      "grad_norm": 0.77758652859161,
      "learning_rate": 5.065342638380722e-06,
      "loss": 0.0485,
      "step": 10540
    },
    {
      "epoch": 6.744081893793986,
      "grad_norm": 0.7943232673143457,
      "learning_rate": 5.063540456836531e-06,
      "loss": 0.0481,
      "step": 10541
    },
    {
      "epoch": 6.744721689059501,
      "grad_norm": 0.717511310145335,
      "learning_rate": 5.061738487255132e-06,
      "loss": 0.0448,
      "step": 10542
    },
    {
      "epoch": 6.7453614843250165,
      "grad_norm": 0.8647916076196204,
      "learning_rate": 5.0599367297138965e-06,
      "loss": 0.0546,
      "step": 10543
    },
    {
      "epoch": 6.746001279590531,
      "grad_norm": 0.7721066811047143,
      "learning_rate": 5.058135184290187e-06,
      "loss": 0.0465,
      "step": 10544
    },
    {
      "epoch": 6.746641074856046,
      "grad_norm": 0.9796819165980536,
      "learning_rate": 5.056333851061366e-06,
      "loss": 0.0519,
      "step": 10545
    },
    {
      "epoch": 6.747280870121561,
      "grad_norm": 0.8637763260213415,
      "learning_rate": 5.054532730104775e-06,
      "loss": 0.0509,
      "step": 10546
    },
    {
      "epoch": 6.747920665387076,
      "grad_norm": 0.8183664574006553,
      "learning_rate": 5.052731821497747e-06,
      "loss": 0.0512,
      "step": 10547
    },
    {
      "epoch": 6.748560460652591,
      "grad_norm": 0.7876997291830671,
      "learning_rate": 5.050931125317618e-06,
      "loss": 0.0469,
      "step": 10548
    },
    {
      "epoch": 6.749200255918106,
      "grad_norm": 0.8239721218682985,
      "learning_rate": 5.049130641641704e-06,
      "loss": 0.0493,
      "step": 10549
    },
    {
      "epoch": 6.749840051183622,
      "grad_norm": 0.84692975725848,
      "learning_rate": 5.047330370547312e-06,
      "loss": 0.053,
      "step": 10550
    },
    {
      "epoch": 6.750479846449136,
      "grad_norm": 0.8036767249500415,
      "learning_rate": 5.045530312111741e-06,
      "loss": 0.0521,
      "step": 10551
    },
    {
      "epoch": 6.751119641714651,
      "grad_norm": 0.7166681559970892,
      "learning_rate": 5.043730466412288e-06,
      "loss": 0.0449,
      "step": 10552
    },
    {
      "epoch": 6.751759436980167,
      "grad_norm": 0.7914503142007902,
      "learning_rate": 5.041930833526231e-06,
      "loss": 0.0527,
      "step": 10553
    },
    {
      "epoch": 6.752399232245681,
      "grad_norm": 0.8778003671220705,
      "learning_rate": 5.040131413530846e-06,
      "loss": 0.0488,
      "step": 10554
    },
    {
      "epoch": 6.753039027511196,
      "grad_norm": 0.7721399641178149,
      "learning_rate": 5.038332206503396e-06,
      "loss": 0.0455,
      "step": 10555
    },
    {
      "epoch": 6.7536788227767115,
      "grad_norm": 0.830205951464023,
      "learning_rate": 5.036533212521134e-06,
      "loss": 0.0482,
      "step": 10556
    },
    {
      "epoch": 6.754318618042227,
      "grad_norm": 0.709979971899556,
      "learning_rate": 5.0347344316613035e-06,
      "loss": 0.043,
      "step": 10557
    },
    {
      "epoch": 6.754958413307741,
      "grad_norm": 0.9407254329061461,
      "learning_rate": 5.032935864001146e-06,
      "loss": 0.0503,
      "step": 10558
    },
    {
      "epoch": 6.755598208573256,
      "grad_norm": 0.8642612990083116,
      "learning_rate": 5.031137509617887e-06,
      "loss": 0.0543,
      "step": 10559
    },
    {
      "epoch": 6.756238003838772,
      "grad_norm": 0.9581795041213153,
      "learning_rate": 5.029339368588746e-06,
      "loss": 0.0611,
      "step": 10560
    },
    {
      "epoch": 6.756877799104287,
      "grad_norm": 0.8784555766428271,
      "learning_rate": 5.027541440990926e-06,
      "loss": 0.0502,
      "step": 10561
    },
    {
      "epoch": 6.757517594369801,
      "grad_norm": 0.7820837615078046,
      "learning_rate": 5.025743726901634e-06,
      "loss": 0.0457,
      "step": 10562
    },
    {
      "epoch": 6.758157389635317,
      "grad_norm": 0.8564997661257654,
      "learning_rate": 5.023946226398059e-06,
      "loss": 0.0468,
      "step": 10563
    },
    {
      "epoch": 6.758797184900832,
      "grad_norm": 0.7272912661025202,
      "learning_rate": 5.022148939557376e-06,
      "loss": 0.0474,
      "step": 10564
    },
    {
      "epoch": 6.759436980166347,
      "grad_norm": 0.7738250445639299,
      "learning_rate": 5.020351866456768e-06,
      "loss": 0.0445,
      "step": 10565
    },
    {
      "epoch": 6.760076775431862,
      "grad_norm": 0.803724875417739,
      "learning_rate": 5.018555007173392e-06,
      "loss": 0.0495,
      "step": 10566
    },
    {
      "epoch": 6.760716570697377,
      "grad_norm": 0.8230185868664132,
      "learning_rate": 5.0167583617844e-06,
      "loss": 0.0387,
      "step": 10567
    },
    {
      "epoch": 6.761356365962892,
      "grad_norm": 0.9256525997463133,
      "learning_rate": 5.0149619303669415e-06,
      "loss": 0.0475,
      "step": 10568
    },
    {
      "epoch": 6.7619961612284065,
      "grad_norm": 0.8666697737379088,
      "learning_rate": 5.013165712998151e-06,
      "loss": 0.0504,
      "step": 10569
    },
    {
      "epoch": 6.762635956493922,
      "grad_norm": 0.7847516215379376,
      "learning_rate": 5.011369709755154e-06,
      "loss": 0.0461,
      "step": 10570
    },
    {
      "epoch": 6.763275751759437,
      "grad_norm": 0.7730616356827391,
      "learning_rate": 5.0095739207150675e-06,
      "loss": 0.0476,
      "step": 10571
    },
    {
      "epoch": 6.763915547024952,
      "grad_norm": 0.7307581563150959,
      "learning_rate": 5.0077783459549986e-06,
      "loss": 0.047,
      "step": 10572
    },
    {
      "epoch": 6.764555342290467,
      "grad_norm": 0.7896596555075825,
      "learning_rate": 5.005982985552049e-06,
      "loss": 0.045,
      "step": 10573
    },
    {
      "epoch": 6.765195137555982,
      "grad_norm": 0.7940544141911503,
      "learning_rate": 5.004187839583301e-06,
      "loss": 0.0536,
      "step": 10574
    },
    {
      "epoch": 6.765834932821497,
      "grad_norm": 0.7597110125582298,
      "learning_rate": 5.002392908125845e-06,
      "loss": 0.0458,
      "step": 10575
    },
    {
      "epoch": 6.766474728087012,
      "grad_norm": 0.7728555482566075,
      "learning_rate": 5.000598191256747e-06,
      "loss": 0.0502,
      "step": 10576
    },
    {
      "epoch": 6.767114523352527,
      "grad_norm": 0.8372522779312027,
      "learning_rate": 4.998803689053065e-06,
      "loss": 0.055,
      "step": 10577
    },
    {
      "epoch": 6.767754318618042,
      "grad_norm": 0.80868179878267,
      "learning_rate": 4.997009401591861e-06,
      "loss": 0.0488,
      "step": 10578
    },
    {
      "epoch": 6.7683941138835575,
      "grad_norm": 0.8662947827851506,
      "learning_rate": 4.995215328950173e-06,
      "loss": 0.0513,
      "step": 10579
    },
    {
      "epoch": 6.769033909149073,
      "grad_norm": 0.9215317313582289,
      "learning_rate": 4.993421471205037e-06,
      "loss": 0.0516,
      "step": 10580
    },
    {
      "epoch": 6.769673704414587,
      "grad_norm": 0.863524978020963,
      "learning_rate": 4.991627828433472e-06,
      "loss": 0.056,
      "step": 10581
    },
    {
      "epoch": 6.770313499680102,
      "grad_norm": 0.8229387111256921,
      "learning_rate": 4.989834400712503e-06,
      "loss": 0.0469,
      "step": 10582
    },
    {
      "epoch": 6.770953294945618,
      "grad_norm": 0.8844245173980056,
      "learning_rate": 4.988041188119133e-06,
      "loss": 0.0512,
      "step": 10583
    },
    {
      "epoch": 6.771593090211132,
      "grad_norm": 0.816842194594646,
      "learning_rate": 4.986248190730357e-06,
      "loss": 0.0433,
      "step": 10584
    },
    {
      "epoch": 6.772232885476647,
      "grad_norm": 0.797021842540153,
      "learning_rate": 4.984455408623167e-06,
      "loss": 0.0486,
      "step": 10585
    },
    {
      "epoch": 6.772872680742163,
      "grad_norm": 0.8491930240629261,
      "learning_rate": 4.9826628418745394e-06,
      "loss": 0.0481,
      "step": 10586
    },
    {
      "epoch": 6.773512476007678,
      "grad_norm": 0.8224037155019853,
      "learning_rate": 4.980870490561439e-06,
      "loss": 0.0476,
      "step": 10587
    },
    {
      "epoch": 6.774152271273192,
      "grad_norm": 0.9333101847962884,
      "learning_rate": 4.979078354760837e-06,
      "loss": 0.0488,
      "step": 10588
    },
    {
      "epoch": 6.774792066538708,
      "grad_norm": 0.8257445777500577,
      "learning_rate": 4.977286434549678e-06,
      "loss": 0.0486,
      "step": 10589
    },
    {
      "epoch": 6.775431861804223,
      "grad_norm": 0.8776745753267933,
      "learning_rate": 4.975494730004905e-06,
      "loss": 0.0557,
      "step": 10590
    },
    {
      "epoch": 6.776071657069737,
      "grad_norm": 0.8297972297878827,
      "learning_rate": 4.973703241203447e-06,
      "loss": 0.0485,
      "step": 10591
    },
    {
      "epoch": 6.7767114523352525,
      "grad_norm": 0.8485988238510629,
      "learning_rate": 4.9719119682222335e-06,
      "loss": 0.0505,
      "step": 10592
    },
    {
      "epoch": 6.777351247600768,
      "grad_norm": 0.7721797514319059,
      "learning_rate": 4.970120911138177e-06,
      "loss": 0.046,
      "step": 10593
    },
    {
      "epoch": 6.777991042866283,
      "grad_norm": 0.8655571253836454,
      "learning_rate": 4.968330070028175e-06,
      "loss": 0.0471,
      "step": 10594
    },
    {
      "epoch": 6.7786308381317975,
      "grad_norm": 0.7322873675072435,
      "learning_rate": 4.966539444969136e-06,
      "loss": 0.0505,
      "step": 10595
    },
    {
      "epoch": 6.779270633397313,
      "grad_norm": 0.7917834824828884,
      "learning_rate": 4.964749036037938e-06,
      "loss": 0.047,
      "step": 10596
    },
    {
      "epoch": 6.779910428662828,
      "grad_norm": 0.805754998938214,
      "learning_rate": 4.962958843311455e-06,
      "loss": 0.0448,
      "step": 10597
    },
    {
      "epoch": 6.780550223928343,
      "grad_norm": 0.7649043380846382,
      "learning_rate": 4.961168866866563e-06,
      "loss": 0.0497,
      "step": 10598
    },
    {
      "epoch": 6.781190019193858,
      "grad_norm": 0.8125294586968163,
      "learning_rate": 4.959379106780118e-06,
      "loss": 0.05,
      "step": 10599
    },
    {
      "epoch": 6.781829814459373,
      "grad_norm": 0.7473366077018027,
      "learning_rate": 4.957589563128966e-06,
      "loss": 0.0473,
      "step": 10600
    },
    {
      "epoch": 6.782469609724888,
      "grad_norm": 0.8388026619470333,
      "learning_rate": 4.95580023598995e-06,
      "loss": 0.0495,
      "step": 10601
    },
    {
      "epoch": 6.7831094049904035,
      "grad_norm": 0.7698116615636209,
      "learning_rate": 4.954011125439897e-06,
      "loss": 0.0499,
      "step": 10602
    },
    {
      "epoch": 6.783749200255918,
      "grad_norm": 0.7327342223980756,
      "learning_rate": 4.952222231555631e-06,
      "loss": 0.045,
      "step": 10603
    },
    {
      "epoch": 6.784388995521433,
      "grad_norm": 0.7954038346153863,
      "learning_rate": 4.950433554413959e-06,
      "loss": 0.0475,
      "step": 10604
    },
    {
      "epoch": 6.785028790786948,
      "grad_norm": 0.7807286215134276,
      "learning_rate": 4.948645094091692e-06,
      "loss": 0.0426,
      "step": 10605
    },
    {
      "epoch": 6.785668586052463,
      "grad_norm": 0.8173012189635964,
      "learning_rate": 4.946856850665619e-06,
      "loss": 0.0501,
      "step": 10606
    },
    {
      "epoch": 6.786308381317978,
      "grad_norm": 0.8288593702874242,
      "learning_rate": 4.945068824212523e-06,
      "loss": 0.051,
      "step": 10607
    },
    {
      "epoch": 6.786948176583493,
      "grad_norm": 0.7771494014021422,
      "learning_rate": 4.943281014809177e-06,
      "loss": 0.0538,
      "step": 10608
    },
    {
      "epoch": 6.787587971849009,
      "grad_norm": 0.8375720396489214,
      "learning_rate": 4.941493422532353e-06,
      "loss": 0.0476,
      "step": 10609
    },
    {
      "epoch": 6.788227767114523,
      "grad_norm": 0.8273519567769767,
      "learning_rate": 4.939706047458802e-06,
      "loss": 0.0549,
      "step": 10610
    },
    {
      "epoch": 6.788867562380038,
      "grad_norm": 0.9174269079709553,
      "learning_rate": 4.937918889665268e-06,
      "loss": 0.0555,
      "step": 10611
    },
    {
      "epoch": 6.789507357645554,
      "grad_norm": 0.8301513463348488,
      "learning_rate": 4.936131949228498e-06,
      "loss": 0.0441,
      "step": 10612
    },
    {
      "epoch": 6.790147152911068,
      "grad_norm": 0.7423058162743033,
      "learning_rate": 4.934345226225212e-06,
      "loss": 0.051,
      "step": 10613
    },
    {
      "epoch": 6.790786948176583,
      "grad_norm": 0.8946652951462238,
      "learning_rate": 4.932558720732131e-06,
      "loss": 0.0516,
      "step": 10614
    },
    {
      "epoch": 6.7914267434420985,
      "grad_norm": 0.724712727222273,
      "learning_rate": 4.930772432825966e-06,
      "loss": 0.0423,
      "step": 10615
    },
    {
      "epoch": 6.792066538707614,
      "grad_norm": 0.7597676322701268,
      "learning_rate": 4.9289863625834134e-06,
      "loss": 0.0433,
      "step": 10616
    },
    {
      "epoch": 6.792706333973129,
      "grad_norm": 0.765006844387284,
      "learning_rate": 4.927200510081166e-06,
      "loss": 0.0448,
      "step": 10617
    },
    {
      "epoch": 6.7933461292386434,
      "grad_norm": 0.8888708484094752,
      "learning_rate": 4.925414875395901e-06,
      "loss": 0.0451,
      "step": 10618
    },
    {
      "epoch": 6.793985924504159,
      "grad_norm": 0.8132918961411674,
      "learning_rate": 4.923629458604298e-06,
      "loss": 0.0464,
      "step": 10619
    },
    {
      "epoch": 6.794625719769674,
      "grad_norm": 0.7293217039361125,
      "learning_rate": 4.921844259783017e-06,
      "loss": 0.0444,
      "step": 10620
    },
    {
      "epoch": 6.795265515035188,
      "grad_norm": 0.9217288395529526,
      "learning_rate": 4.920059279008706e-06,
      "loss": 0.0598,
      "step": 10621
    },
    {
      "epoch": 6.795905310300704,
      "grad_norm": 0.8413471765759776,
      "learning_rate": 4.918274516358017e-06,
      "loss": 0.0513,
      "step": 10622
    },
    {
      "epoch": 6.796545105566219,
      "grad_norm": 0.8033557937466551,
      "learning_rate": 4.91648997190758e-06,
      "loss": 0.046,
      "step": 10623
    },
    {
      "epoch": 6.797184900831734,
      "grad_norm": 0.8456755360312173,
      "learning_rate": 4.914705645734017e-06,
      "loss": 0.0549,
      "step": 10624
    },
    {
      "epoch": 6.797824696097249,
      "grad_norm": 0.7700172118327255,
      "learning_rate": 4.912921537913952e-06,
      "loss": 0.0486,
      "step": 10625
    },
    {
      "epoch": 6.798464491362764,
      "grad_norm": 0.8191295600729457,
      "learning_rate": 4.911137648523986e-06,
      "loss": 0.047,
      "step": 10626
    },
    {
      "epoch": 6.799104286628279,
      "grad_norm": 0.8633408931129255,
      "learning_rate": 4.909353977640717e-06,
      "loss": 0.0492,
      "step": 10627
    },
    {
      "epoch": 6.7997440818937935,
      "grad_norm": 0.8720413098335918,
      "learning_rate": 4.907570525340729e-06,
      "loss": 0.0508,
      "step": 10628
    },
    {
      "epoch": 6.800383877159309,
      "grad_norm": 0.8054462496866315,
      "learning_rate": 4.9057872917006075e-06,
      "loss": 0.0504,
      "step": 10629
    },
    {
      "epoch": 6.801023672424824,
      "grad_norm": 0.8025676743405701,
      "learning_rate": 4.904004276796919e-06,
      "loss": 0.0485,
      "step": 10630
    },
    {
      "epoch": 6.801663467690339,
      "grad_norm": 0.7002250371102798,
      "learning_rate": 4.902221480706221e-06,
      "loss": 0.0439,
      "step": 10631
    },
    {
      "epoch": 6.802303262955854,
      "grad_norm": 0.7734265258832468,
      "learning_rate": 4.900438903505064e-06,
      "loss": 0.0483,
      "step": 10632
    },
    {
      "epoch": 6.802943058221369,
      "grad_norm": 0.7444047843424394,
      "learning_rate": 4.89865654526999e-06,
      "loss": 0.0466,
      "step": 10633
    },
    {
      "epoch": 6.803582853486884,
      "grad_norm": 0.8378909666638077,
      "learning_rate": 4.896874406077524e-06,
      "loss": 0.0515,
      "step": 10634
    },
    {
      "epoch": 6.8042226487524,
      "grad_norm": 0.9464384469782852,
      "learning_rate": 4.895092486004198e-06,
      "loss": 0.0516,
      "step": 10635
    },
    {
      "epoch": 6.804862444017914,
      "grad_norm": 0.9173769087515086,
      "learning_rate": 4.89331078512652e-06,
      "loss": 0.0464,
      "step": 10636
    },
    {
      "epoch": 6.805502239283429,
      "grad_norm": 0.7177993498888267,
      "learning_rate": 4.891529303520993e-06,
      "loss": 0.0452,
      "step": 10637
    },
    {
      "epoch": 6.8061420345489445,
      "grad_norm": 0.8419719640110513,
      "learning_rate": 4.889748041264107e-06,
      "loss": 0.0486,
      "step": 10638
    },
    {
      "epoch": 6.80678182981446,
      "grad_norm": 0.8273023495696692,
      "learning_rate": 4.887966998432352e-06,
      "loss": 0.0489,
      "step": 10639
    },
    {
      "epoch": 6.807421625079974,
      "grad_norm": 0.7247152039258895,
      "learning_rate": 4.886186175102201e-06,
      "loss": 0.0472,
      "step": 10640
    },
    {
      "epoch": 6.8080614203454894,
      "grad_norm": 0.7781429350245546,
      "learning_rate": 4.8844055713501145e-06,
      "loss": 0.0495,
      "step": 10641
    },
    {
      "epoch": 6.808701215611005,
      "grad_norm": 0.7062499882638836,
      "learning_rate": 4.882625187252559e-06,
      "loss": 0.0441,
      "step": 10642
    },
    {
      "epoch": 6.809341010876519,
      "grad_norm": 0.748587378758223,
      "learning_rate": 4.880845022885973e-06,
      "loss": 0.0494,
      "step": 10643
    },
    {
      "epoch": 6.809980806142034,
      "grad_norm": 0.7516065266459214,
      "learning_rate": 4.879065078326796e-06,
      "loss": 0.0479,
      "step": 10644
    },
    {
      "epoch": 6.81062060140755,
      "grad_norm": 0.8864164424502591,
      "learning_rate": 4.877285353651454e-06,
      "loss": 0.0567,
      "step": 10645
    },
    {
      "epoch": 6.811260396673065,
      "grad_norm": 0.7709822749286496,
      "learning_rate": 4.875505848936367e-06,
      "loss": 0.0469,
      "step": 10646
    },
    {
      "epoch": 6.811900191938579,
      "grad_norm": 0.8913240239966723,
      "learning_rate": 4.873726564257943e-06,
      "loss": 0.0511,
      "step": 10647
    },
    {
      "epoch": 6.812539987204095,
      "grad_norm": 0.8188406847584246,
      "learning_rate": 4.871947499692578e-06,
      "loss": 0.0537,
      "step": 10648
    },
    {
      "epoch": 6.81317978246961,
      "grad_norm": 0.7918218685165378,
      "learning_rate": 4.870168655316668e-06,
      "loss": 0.0485,
      "step": 10649
    },
    {
      "epoch": 6.813819577735125,
      "grad_norm": 0.7709823930157537,
      "learning_rate": 4.868390031206592e-06,
      "loss": 0.0484,
      "step": 10650
    },
    {
      "epoch": 6.8144593730006395,
      "grad_norm": 0.7681557921209438,
      "learning_rate": 4.866611627438716e-06,
      "loss": 0.0539,
      "step": 10651
    },
    {
      "epoch": 6.815099168266155,
      "grad_norm": 0.8812248555814186,
      "learning_rate": 4.864833444089407e-06,
      "loss": 0.0516,
      "step": 10652
    },
    {
      "epoch": 6.81573896353167,
      "grad_norm": 0.8458965338711925,
      "learning_rate": 4.863055481235018e-06,
      "loss": 0.0534,
      "step": 10653
    },
    {
      "epoch": 6.816378758797185,
      "grad_norm": 0.8431471415950939,
      "learning_rate": 4.861277738951883e-06,
      "loss": 0.0468,
      "step": 10654
    },
    {
      "epoch": 6.8170185540627,
      "grad_norm": 0.8602369600850622,
      "learning_rate": 4.859500217316344e-06,
      "loss": 0.0492,
      "step": 10655
    },
    {
      "epoch": 6.817658349328215,
      "grad_norm": 0.7925516923795114,
      "learning_rate": 4.857722916404723e-06,
      "loss": 0.0454,
      "step": 10656
    },
    {
      "epoch": 6.81829814459373,
      "grad_norm": 0.9093242039438768,
      "learning_rate": 4.855945836293332e-06,
      "loss": 0.0543,
      "step": 10657
    },
    {
      "epoch": 6.818937939859245,
      "grad_norm": 0.7505166931070664,
      "learning_rate": 4.854168977058472e-06,
      "loss": 0.0469,
      "step": 10658
    },
    {
      "epoch": 6.81957773512476,
      "grad_norm": 0.8290513443209485,
      "learning_rate": 4.8523923387764474e-06,
      "loss": 0.0451,
      "step": 10659
    },
    {
      "epoch": 6.820217530390275,
      "grad_norm": 0.7816920633532642,
      "learning_rate": 4.850615921523537e-06,
      "loss": 0.0447,
      "step": 10660
    },
    {
      "epoch": 6.8208573256557905,
      "grad_norm": 0.7424585935137956,
      "learning_rate": 4.848839725376021e-06,
      "loss": 0.0442,
      "step": 10661
    },
    {
      "epoch": 6.821497120921305,
      "grad_norm": 0.8914059607377971,
      "learning_rate": 4.847063750410163e-06,
      "loss": 0.0543,
      "step": 10662
    },
    {
      "epoch": 6.82213691618682,
      "grad_norm": 0.8506329370619847,
      "learning_rate": 4.84528799670222e-06,
      "loss": 0.0516,
      "step": 10663
    },
    {
      "epoch": 6.822776711452335,
      "grad_norm": 0.7566119333134295,
      "learning_rate": 4.843512464328439e-06,
      "loss": 0.0428,
      "step": 10664
    },
    {
      "epoch": 6.82341650671785,
      "grad_norm": 0.85870940853343,
      "learning_rate": 4.841737153365063e-06,
      "loss": 0.05,
      "step": 10665
    },
    {
      "epoch": 6.824056301983365,
      "grad_norm": 0.7732908254130547,
      "learning_rate": 4.8399620638883186e-06,
      "loss": 0.0497,
      "step": 10666
    },
    {
      "epoch": 6.82469609724888,
      "grad_norm": 0.8042745227899661,
      "learning_rate": 4.838187195974423e-06,
      "loss": 0.0513,
      "step": 10667
    },
    {
      "epoch": 6.825335892514396,
      "grad_norm": 0.71134771478773,
      "learning_rate": 4.836412549699584e-06,
      "loss": 0.0491,
      "step": 10668
    },
    {
      "epoch": 6.825975687779911,
      "grad_norm": 0.769802359930727,
      "learning_rate": 4.834638125140007e-06,
      "loss": 0.0505,
      "step": 10669
    },
    {
      "epoch": 6.826615483045425,
      "grad_norm": 0.8421386099392345,
      "learning_rate": 4.832863922371883e-06,
      "loss": 0.0504,
      "step": 10670
    },
    {
      "epoch": 6.827255278310941,
      "grad_norm": 0.8893911201675462,
      "learning_rate": 4.831089941471385e-06,
      "loss": 0.0503,
      "step": 10671
    },
    {
      "epoch": 6.827895073576456,
      "grad_norm": 0.8928705067179856,
      "learning_rate": 4.829316182514695e-06,
      "loss": 0.0489,
      "step": 10672
    },
    {
      "epoch": 6.82853486884197,
      "grad_norm": 0.8342470927318519,
      "learning_rate": 4.827542645577969e-06,
      "loss": 0.0511,
      "step": 10673
    },
    {
      "epoch": 6.8291746641074855,
      "grad_norm": 0.8590568217693002,
      "learning_rate": 4.8257693307373625e-06,
      "loss": 0.0526,
      "step": 10674
    },
    {
      "epoch": 6.829814459373001,
      "grad_norm": 0.8475079229973682,
      "learning_rate": 4.823996238069015e-06,
      "loss": 0.0473,
      "step": 10675
    },
    {
      "epoch": 6.830454254638516,
      "grad_norm": 0.8608177838404314,
      "learning_rate": 4.822223367649063e-06,
      "loss": 0.053,
      "step": 10676
    },
    {
      "epoch": 6.8310940499040305,
      "grad_norm": 0.8770070696091993,
      "learning_rate": 4.820450719553628e-06,
      "loss": 0.0501,
      "step": 10677
    },
    {
      "epoch": 6.831733845169546,
      "grad_norm": 0.8199624444465683,
      "learning_rate": 4.818678293858822e-06,
      "loss": 0.0507,
      "step": 10678
    },
    {
      "epoch": 6.832373640435061,
      "grad_norm": 0.8161324507682867,
      "learning_rate": 4.816906090640756e-06,
      "loss": 0.0539,
      "step": 10679
    },
    {
      "epoch": 6.833013435700575,
      "grad_norm": 0.8024641346981921,
      "learning_rate": 4.815134109975524e-06,
      "loss": 0.0475,
      "step": 10680
    },
    {
      "epoch": 6.833653230966091,
      "grad_norm": 0.7333815002658433,
      "learning_rate": 4.813362351939206e-06,
      "loss": 0.0449,
      "step": 10681
    },
    {
      "epoch": 6.834293026231606,
      "grad_norm": 0.8331657702521595,
      "learning_rate": 4.8115908166078854e-06,
      "loss": 0.0525,
      "step": 10682
    },
    {
      "epoch": 6.834932821497121,
      "grad_norm": 0.8967148948542222,
      "learning_rate": 4.809819504057626e-06,
      "loss": 0.0491,
      "step": 10683
    },
    {
      "epoch": 6.835572616762636,
      "grad_norm": 0.8009156629941996,
      "learning_rate": 4.808048414364484e-06,
      "loss": 0.0547,
      "step": 10684
    },
    {
      "epoch": 6.836212412028151,
      "grad_norm": 0.8784448681502828,
      "learning_rate": 4.8062775476045046e-06,
      "loss": 0.0458,
      "step": 10685
    },
    {
      "epoch": 6.836852207293666,
      "grad_norm": 0.8096571440980731,
      "learning_rate": 4.804506903853732e-06,
      "loss": 0.0483,
      "step": 10686
    },
    {
      "epoch": 6.837492002559181,
      "grad_norm": 0.7481901274276476,
      "learning_rate": 4.8027364831881905e-06,
      "loss": 0.044,
      "step": 10687
    },
    {
      "epoch": 6.838131797824696,
      "grad_norm": 0.846855969234893,
      "learning_rate": 4.800966285683896e-06,
      "loss": 0.0508,
      "step": 10688
    },
    {
      "epoch": 6.838771593090211,
      "grad_norm": 0.7390670533806422,
      "learning_rate": 4.7991963114168635e-06,
      "loss": 0.0455,
      "step": 10689
    },
    {
      "epoch": 6.839411388355726,
      "grad_norm": 0.776690685380885,
      "learning_rate": 4.79742656046309e-06,
      "loss": 0.0511,
      "step": 10690
    },
    {
      "epoch": 6.840051183621242,
      "grad_norm": 0.8248146293398925,
      "learning_rate": 4.795657032898566e-06,
      "loss": 0.0525,
      "step": 10691
    },
    {
      "epoch": 6.840690978886756,
      "grad_norm": 0.6723698018880149,
      "learning_rate": 4.793887728799272e-06,
      "loss": 0.0454,
      "step": 10692
    },
    {
      "epoch": 6.841330774152271,
      "grad_norm": 0.8366170313806909,
      "learning_rate": 4.792118648241177e-06,
      "loss": 0.0528,
      "step": 10693
    },
    {
      "epoch": 6.841970569417787,
      "grad_norm": 0.8156438074251554,
      "learning_rate": 4.790349791300241e-06,
      "loss": 0.0471,
      "step": 10694
    },
    {
      "epoch": 6.842610364683301,
      "grad_norm": 0.8119976821812536,
      "learning_rate": 4.788581158052418e-06,
      "loss": 0.0568,
      "step": 10695
    },
    {
      "epoch": 6.843250159948816,
      "grad_norm": 0.8191524634263386,
      "learning_rate": 4.78681274857365e-06,
      "loss": 0.0465,
      "step": 10696
    },
    {
      "epoch": 6.8438899552143315,
      "grad_norm": 0.800315705280224,
      "learning_rate": 4.7850445629398716e-06,
      "loss": 0.0459,
      "step": 10697
    },
    {
      "epoch": 6.844529750479847,
      "grad_norm": 0.8134964211036504,
      "learning_rate": 4.783276601226999e-06,
      "loss": 0.0539,
      "step": 10698
    },
    {
      "epoch": 6.845169545745361,
      "grad_norm": 0.7881949869665984,
      "learning_rate": 4.781508863510951e-06,
      "loss": 0.0452,
      "step": 10699
    },
    {
      "epoch": 6.8458093410108765,
      "grad_norm": 0.8645010407848127,
      "learning_rate": 4.77974134986763e-06,
      "loss": 0.0438,
      "step": 10700
    },
    {
      "epoch": 6.846449136276392,
      "grad_norm": 0.8649972572611129,
      "learning_rate": 4.777974060372926e-06,
      "loss": 0.0531,
      "step": 10701
    },
    {
      "epoch": 6.847088931541906,
      "grad_norm": 0.8091873905780539,
      "learning_rate": 4.776206995102731e-06,
      "loss": 0.0449,
      "step": 10702
    },
    {
      "epoch": 6.847728726807421,
      "grad_norm": 0.8919605814834457,
      "learning_rate": 4.774440154132916e-06,
      "loss": 0.0499,
      "step": 10703
    },
    {
      "epoch": 6.848368522072937,
      "grad_norm": 0.7737750029024391,
      "learning_rate": 4.772673537539343e-06,
      "loss": 0.0504,
      "step": 10704
    },
    {
      "epoch": 6.849008317338452,
      "grad_norm": 0.734662952023083,
      "learning_rate": 4.77090714539787e-06,
      "loss": 0.047,
      "step": 10705
    },
    {
      "epoch": 6.849648112603967,
      "grad_norm": 0.8693099244015714,
      "learning_rate": 4.769140977784344e-06,
      "loss": 0.0506,
      "step": 10706
    },
    {
      "epoch": 6.850287907869482,
      "grad_norm": 0.8082550453844615,
      "learning_rate": 4.767375034774598e-06,
      "loss": 0.0531,
      "step": 10707
    },
    {
      "epoch": 6.850927703134997,
      "grad_norm": 0.9018001573766982,
      "learning_rate": 4.765609316444456e-06,
      "loss": 0.0571,
      "step": 10708
    },
    {
      "epoch": 6.851567498400512,
      "grad_norm": 0.7432409036373469,
      "learning_rate": 4.7638438228697436e-06,
      "loss": 0.0465,
      "step": 10709
    },
    {
      "epoch": 6.8522072936660265,
      "grad_norm": 0.8454902246781111,
      "learning_rate": 4.7620785541262626e-06,
      "loss": 0.0498,
      "step": 10710
    },
    {
      "epoch": 6.852847088931542,
      "grad_norm": 0.9236328056650577,
      "learning_rate": 4.7603135102898065e-06,
      "loss": 0.0562,
      "step": 10711
    },
    {
      "epoch": 6.853486884197057,
      "grad_norm": 0.8591423743746639,
      "learning_rate": 4.758548691436172e-06,
      "loss": 0.052,
      "step": 10712
    },
    {
      "epoch": 6.854126679462572,
      "grad_norm": 0.7972950579115312,
      "learning_rate": 4.756784097641133e-06,
      "loss": 0.046,
      "step": 10713
    },
    {
      "epoch": 6.854766474728087,
      "grad_norm": 0.8277689262346449,
      "learning_rate": 4.755019728980458e-06,
      "loss": 0.0527,
      "step": 10714
    },
    {
      "epoch": 6.855406269993602,
      "grad_norm": 0.6981173200599176,
      "learning_rate": 4.753255585529901e-06,
      "loss": 0.0475,
      "step": 10715
    },
    {
      "epoch": 6.856046065259117,
      "grad_norm": 0.7095657890675618,
      "learning_rate": 4.75149166736522e-06,
      "loss": 0.0516,
      "step": 10716
    },
    {
      "epoch": 6.856685860524632,
      "grad_norm": 0.7994179748664964,
      "learning_rate": 4.749727974562151e-06,
      "loss": 0.0539,
      "step": 10717
    },
    {
      "epoch": 6.857325655790147,
      "grad_norm": 0.8688972358390116,
      "learning_rate": 4.747964507196419e-06,
      "loss": 0.048,
      "step": 10718
    },
    {
      "epoch": 6.857965451055662,
      "grad_norm": 0.866868798266758,
      "learning_rate": 4.7462012653437525e-06,
      "loss": 0.054,
      "step": 10719
    },
    {
      "epoch": 6.8586052463211775,
      "grad_norm": 0.7909403120860966,
      "learning_rate": 4.744438249079858e-06,
      "loss": 0.0509,
      "step": 10720
    },
    {
      "epoch": 6.859245041586692,
      "grad_norm": 0.6838742157135173,
      "learning_rate": 4.7426754584804354e-06,
      "loss": 0.0448,
      "step": 10721
    },
    {
      "epoch": 6.859884836852207,
      "grad_norm": 1.0366176937199374,
      "learning_rate": 4.740912893621178e-06,
      "loss": 0.0533,
      "step": 10722
    },
    {
      "epoch": 6.8605246321177225,
      "grad_norm": 0.7976915485447154,
      "learning_rate": 4.739150554577764e-06,
      "loss": 0.044,
      "step": 10723
    },
    {
      "epoch": 6.861164427383238,
      "grad_norm": 0.6837943320954188,
      "learning_rate": 4.7373884414258675e-06,
      "loss": 0.0431,
      "step": 10724
    },
    {
      "epoch": 6.861804222648752,
      "grad_norm": 0.6923650933461821,
      "learning_rate": 4.735626554241147e-06,
      "loss": 0.0464,
      "step": 10725
    },
    {
      "epoch": 6.862444017914267,
      "grad_norm": 0.9238398766477482,
      "learning_rate": 4.7338648930992605e-06,
      "loss": 0.0505,
      "step": 10726
    },
    {
      "epoch": 6.863083813179783,
      "grad_norm": 0.7993259030597999,
      "learning_rate": 4.732103458075849e-06,
      "loss": 0.047,
      "step": 10727
    },
    {
      "epoch": 6.863723608445298,
      "grad_norm": 0.8570221795511827,
      "learning_rate": 4.730342249246538e-06,
      "loss": 0.0562,
      "step": 10728
    },
    {
      "epoch": 6.864363403710812,
      "grad_norm": 0.7929994021584315,
      "learning_rate": 4.728581266686963e-06,
      "loss": 0.0446,
      "step": 10729
    },
    {
      "epoch": 6.865003198976328,
      "grad_norm": 0.8295096916934505,
      "learning_rate": 4.726820510472729e-06,
      "loss": 0.0432,
      "step": 10730
    },
    {
      "epoch": 6.865642994241843,
      "grad_norm": 0.7425316962179463,
      "learning_rate": 4.72505998067944e-06,
      "loss": 0.0512,
      "step": 10731
    },
    {
      "epoch": 6.866282789507357,
      "grad_norm": 0.762224014628206,
      "learning_rate": 4.723299677382696e-06,
      "loss": 0.0519,
      "step": 10732
    },
    {
      "epoch": 6.8669225847728725,
      "grad_norm": 0.9101602895675547,
      "learning_rate": 4.721539600658076e-06,
      "loss": 0.0477,
      "step": 10733
    },
    {
      "epoch": 6.867562380038388,
      "grad_norm": 0.8723358573917873,
      "learning_rate": 4.719779750581156e-06,
      "loss": 0.0481,
      "step": 10734
    },
    {
      "epoch": 6.868202175303903,
      "grad_norm": 1.0115064040631299,
      "learning_rate": 4.7180201272275014e-06,
      "loss": 0.0546,
      "step": 10735
    },
    {
      "epoch": 6.8688419705694175,
      "grad_norm": 0.7759716325011775,
      "learning_rate": 4.716260730672667e-06,
      "loss": 0.0519,
      "step": 10736
    },
    {
      "epoch": 6.869481765834933,
      "grad_norm": 0.7981154681149453,
      "learning_rate": 4.7145015609921975e-06,
      "loss": 0.0484,
      "step": 10737
    },
    {
      "epoch": 6.870121561100448,
      "grad_norm": 0.8589391756364007,
      "learning_rate": 4.712742618261624e-06,
      "loss": 0.046,
      "step": 10738
    },
    {
      "epoch": 6.870761356365962,
      "grad_norm": 0.9354440495955227,
      "learning_rate": 4.710983902556482e-06,
      "loss": 0.0559,
      "step": 10739
    },
    {
      "epoch": 6.871401151631478,
      "grad_norm": 0.8188546392275551,
      "learning_rate": 4.709225413952283e-06,
      "loss": 0.0505,
      "step": 10740
    },
    {
      "epoch": 6.872040946896993,
      "grad_norm": 0.746614812615665,
      "learning_rate": 4.707467152524528e-06,
      "loss": 0.0499,
      "step": 10741
    },
    {
      "epoch": 6.872680742162508,
      "grad_norm": 0.7900924698843993,
      "learning_rate": 4.705709118348725e-06,
      "loss": 0.0525,
      "step": 10742
    },
    {
      "epoch": 6.8733205374280235,
      "grad_norm": 0.8102728333478986,
      "learning_rate": 4.703951311500352e-06,
      "loss": 0.0523,
      "step": 10743
    },
    {
      "epoch": 6.873960332693538,
      "grad_norm": 0.8278660002768388,
      "learning_rate": 4.7021937320548905e-06,
      "loss": 0.0544,
      "step": 10744
    },
    {
      "epoch": 6.874600127959053,
      "grad_norm": 0.8576498180361646,
      "learning_rate": 4.700436380087802e-06,
      "loss": 0.0473,
      "step": 10745
    },
    {
      "epoch": 6.8752399232245685,
      "grad_norm": 0.9141130361629209,
      "learning_rate": 4.698679255674551e-06,
      "loss": 0.0476,
      "step": 10746
    },
    {
      "epoch": 6.875879718490083,
      "grad_norm": 0.8688208834669495,
      "learning_rate": 4.696922358890583e-06,
      "loss": 0.0553,
      "step": 10747
    },
    {
      "epoch": 6.876519513755598,
      "grad_norm": 0.7890757291276147,
      "learning_rate": 4.695165689811332e-06,
      "loss": 0.0519,
      "step": 10748
    },
    {
      "epoch": 6.877159309021113,
      "grad_norm": 0.7956487839787254,
      "learning_rate": 4.6934092485122325e-06,
      "loss": 0.0481,
      "step": 10749
    },
    {
      "epoch": 6.877799104286629,
      "grad_norm": 0.8376544767447782,
      "learning_rate": 4.691653035068705e-06,
      "loss": 0.0505,
      "step": 10750
    },
    {
      "epoch": 6.878438899552143,
      "grad_norm": 0.8086955948700371,
      "learning_rate": 4.689897049556146e-06,
      "loss": 0.0482,
      "step": 10751
    },
    {
      "epoch": 6.879078694817658,
      "grad_norm": 0.7870448799149365,
      "learning_rate": 4.688141292049965e-06,
      "loss": 0.0444,
      "step": 10752
    },
    {
      "epoch": 6.879718490083174,
      "grad_norm": 0.8781628363964686,
      "learning_rate": 4.686385762625547e-06,
      "loss": 0.0501,
      "step": 10753
    },
    {
      "epoch": 6.880358285348688,
      "grad_norm": 0.8477264955482499,
      "learning_rate": 4.684630461358274e-06,
      "loss": 0.0513,
      "step": 10754
    },
    {
      "epoch": 6.880998080614203,
      "grad_norm": 0.7421943470800699,
      "learning_rate": 4.682875388323509e-06,
      "loss": 0.0466,
      "step": 10755
    },
    {
      "epoch": 6.8816378758797185,
      "grad_norm": 0.8258492876658657,
      "learning_rate": 4.68112054359662e-06,
      "loss": 0.0488,
      "step": 10756
    },
    {
      "epoch": 6.882277671145234,
      "grad_norm": 0.7817451732864045,
      "learning_rate": 4.679365927252953e-06,
      "loss": 0.0501,
      "step": 10757
    },
    {
      "epoch": 6.882917466410748,
      "grad_norm": 0.8197538680022461,
      "learning_rate": 4.677611539367846e-06,
      "loss": 0.0449,
      "step": 10758
    },
    {
      "epoch": 6.8835572616762635,
      "grad_norm": 0.7527521673657855,
      "learning_rate": 4.6758573800166344e-06,
      "loss": 0.0537,
      "step": 10759
    },
    {
      "epoch": 6.884197056941779,
      "grad_norm": 0.8522983453554378,
      "learning_rate": 4.674103449274635e-06,
      "loss": 0.0558,
      "step": 10760
    },
    {
      "epoch": 6.884836852207294,
      "grad_norm": 0.7729678959036946,
      "learning_rate": 4.672349747217161e-06,
      "loss": 0.0484,
      "step": 10761
    },
    {
      "epoch": 6.885476647472808,
      "grad_norm": 0.822658272255188,
      "learning_rate": 4.670596273919508e-06,
      "loss": 0.0484,
      "step": 10762
    },
    {
      "epoch": 6.886116442738324,
      "grad_norm": 0.8805008821576351,
      "learning_rate": 4.668843029456973e-06,
      "loss": 0.0539,
      "step": 10763
    },
    {
      "epoch": 6.886756238003839,
      "grad_norm": 0.8410456437780693,
      "learning_rate": 4.667090013904836e-06,
      "loss": 0.0464,
      "step": 10764
    },
    {
      "epoch": 6.887396033269354,
      "grad_norm": 0.9497949569057097,
      "learning_rate": 4.665337227338368e-06,
      "loss": 0.0519,
      "step": 10765
    },
    {
      "epoch": 6.888035828534869,
      "grad_norm": 0.7943110905556412,
      "learning_rate": 4.663584669832828e-06,
      "loss": 0.0463,
      "step": 10766
    },
    {
      "epoch": 6.888675623800384,
      "grad_norm": 0.7561528843239274,
      "learning_rate": 4.661832341463471e-06,
      "loss": 0.0498,
      "step": 10767
    },
    {
      "epoch": 6.889315419065899,
      "grad_norm": 0.8032024976633176,
      "learning_rate": 4.660080242305534e-06,
      "loss": 0.0527,
      "step": 10768
    },
    {
      "epoch": 6.889955214331414,
      "grad_norm": 0.7648117990688316,
      "learning_rate": 4.658328372434254e-06,
      "loss": 0.0469,
      "step": 10769
    },
    {
      "epoch": 6.890595009596929,
      "grad_norm": 0.7935719855284631,
      "learning_rate": 4.656576731924853e-06,
      "loss": 0.0493,
      "step": 10770
    },
    {
      "epoch": 6.891234804862444,
      "grad_norm": 0.7703179707776264,
      "learning_rate": 4.654825320852543e-06,
      "loss": 0.0518,
      "step": 10771
    },
    {
      "epoch": 6.891874600127959,
      "grad_norm": 0.7899618879043138,
      "learning_rate": 4.65307413929252e-06,
      "loss": 0.0482,
      "step": 10772
    },
    {
      "epoch": 6.892514395393474,
      "grad_norm": 0.850943962026489,
      "learning_rate": 4.651323187319986e-06,
      "loss": 0.0517,
      "step": 10773
    },
    {
      "epoch": 6.893154190658989,
      "grad_norm": 0.7631787730980388,
      "learning_rate": 4.64957246501012e-06,
      "loss": 0.0442,
      "step": 10774
    },
    {
      "epoch": 6.893793985924504,
      "grad_norm": 0.7131721991302958,
      "learning_rate": 4.647821972438092e-06,
      "loss": 0.0431,
      "step": 10775
    },
    {
      "epoch": 6.894433781190019,
      "grad_norm": 0.929529353473411,
      "learning_rate": 4.64607170967907e-06,
      "loss": 0.0522,
      "step": 10776
    },
    {
      "epoch": 6.895073576455534,
      "grad_norm": 0.7842865207405659,
      "learning_rate": 4.644321676808206e-06,
      "loss": 0.0511,
      "step": 10777
    },
    {
      "epoch": 6.895713371721049,
      "grad_norm": 0.844631208059752,
      "learning_rate": 4.642571873900637e-06,
      "loss": 0.0533,
      "step": 10778
    },
    {
      "epoch": 6.8963531669865645,
      "grad_norm": 0.8552140066322067,
      "learning_rate": 4.6408223010315105e-06,
      "loss": 0.0487,
      "step": 10779
    },
    {
      "epoch": 6.89699296225208,
      "grad_norm": 0.8329585646803703,
      "learning_rate": 4.639072958275937e-06,
      "loss": 0.0486,
      "step": 10780
    },
    {
      "epoch": 6.897632757517594,
      "grad_norm": 0.7980716561233243,
      "learning_rate": 4.637323845709033e-06,
      "loss": 0.051,
      "step": 10781
    },
    {
      "epoch": 6.8982725527831095,
      "grad_norm": 0.7869560620128836,
      "learning_rate": 4.635574963405901e-06,
      "loss": 0.0533,
      "step": 10782
    },
    {
      "epoch": 6.898912348048625,
      "grad_norm": 0.7682461768554086,
      "learning_rate": 4.633826311441641e-06,
      "loss": 0.0469,
      "step": 10783
    },
    {
      "epoch": 6.899552143314139,
      "grad_norm": 0.8713241484224359,
      "learning_rate": 4.632077889891334e-06,
      "loss": 0.0486,
      "step": 10784
    },
    {
      "epoch": 6.900191938579654,
      "grad_norm": 0.7257700902987807,
      "learning_rate": 4.630329698830049e-06,
      "loss": 0.0426,
      "step": 10785
    },
    {
      "epoch": 6.90083173384517,
      "grad_norm": 0.7598723461772652,
      "learning_rate": 4.62858173833286e-06,
      "loss": 0.0444,
      "step": 10786
    },
    {
      "epoch": 6.901471529110685,
      "grad_norm": 0.8015583702849791,
      "learning_rate": 4.626834008474815e-06,
      "loss": 0.0507,
      "step": 10787
    },
    {
      "epoch": 6.902111324376199,
      "grad_norm": 0.9044647613390521,
      "learning_rate": 4.625086509330956e-06,
      "loss": 0.0519,
      "step": 10788
    },
    {
      "epoch": 6.902751119641715,
      "grad_norm": 0.8836495186553486,
      "learning_rate": 4.623339240976325e-06,
      "loss": 0.0611,
      "step": 10789
    },
    {
      "epoch": 6.90339091490723,
      "grad_norm": 0.6608925485298939,
      "learning_rate": 4.6215922034859425e-06,
      "loss": 0.0418,
      "step": 10790
    },
    {
      "epoch": 6.904030710172744,
      "grad_norm": 0.7913360394408202,
      "learning_rate": 4.619845396934823e-06,
      "loss": 0.0475,
      "step": 10791
    },
    {
      "epoch": 6.9046705054382596,
      "grad_norm": 0.7756908460549082,
      "learning_rate": 4.61809882139797e-06,
      "loss": 0.0476,
      "step": 10792
    },
    {
      "epoch": 6.905310300703775,
      "grad_norm": 0.7474659600630584,
      "learning_rate": 4.616352476950381e-06,
      "loss": 0.0438,
      "step": 10793
    },
    {
      "epoch": 6.90595009596929,
      "grad_norm": 0.6858350619737682,
      "learning_rate": 4.614606363667041e-06,
      "loss": 0.0471,
      "step": 10794
    },
    {
      "epoch": 6.9065898912348045,
      "grad_norm": 0.8119309708530801,
      "learning_rate": 4.612860481622924e-06,
      "loss": 0.0511,
      "step": 10795
    },
    {
      "epoch": 6.90722968650032,
      "grad_norm": 0.7987625979682935,
      "learning_rate": 4.611114830892994e-06,
      "loss": 0.0506,
      "step": 10796
    },
    {
      "epoch": 6.907869481765835,
      "grad_norm": 0.9895655130218474,
      "learning_rate": 4.609369411552208e-06,
      "loss": 0.0547,
      "step": 10797
    },
    {
      "epoch": 6.90850927703135,
      "grad_norm": 0.7943225327529452,
      "learning_rate": 4.607624223675506e-06,
      "loss": 0.0432,
      "step": 10798
    },
    {
      "epoch": 6.909149072296865,
      "grad_norm": 0.7265998488372439,
      "learning_rate": 4.605879267337832e-06,
      "loss": 0.0457,
      "step": 10799
    },
    {
      "epoch": 6.90978886756238,
      "grad_norm": 0.8716191819432584,
      "learning_rate": 4.604134542614106e-06,
      "loss": 0.0449,
      "step": 10800
    },
    {
      "epoch": 6.910428662827895,
      "grad_norm": 0.7535897176702747,
      "learning_rate": 4.602390049579245e-06,
      "loss": 0.0475,
      "step": 10801
    },
    {
      "epoch": 6.9110684580934105,
      "grad_norm": 0.8987308385009117,
      "learning_rate": 4.60064578830815e-06,
      "loss": 0.0517,
      "step": 10802
    },
    {
      "epoch": 6.911708253358925,
      "grad_norm": 0.8131614565876043,
      "learning_rate": 4.598901758875724e-06,
      "loss": 0.048,
      "step": 10803
    },
    {
      "epoch": 6.91234804862444,
      "grad_norm": 0.758700508308506,
      "learning_rate": 4.5971579613568474e-06,
      "loss": 0.0423,
      "step": 10804
    },
    {
      "epoch": 6.9129878438899555,
      "grad_norm": 0.7575554476130919,
      "learning_rate": 4.595414395826394e-06,
      "loss": 0.0487,
      "step": 10805
    },
    {
      "epoch": 6.91362763915547,
      "grad_norm": 0.7849419810784164,
      "learning_rate": 4.593671062359236e-06,
      "loss": 0.0513,
      "step": 10806
    },
    {
      "epoch": 6.914267434420985,
      "grad_norm": 0.7302084961160727,
      "learning_rate": 4.591927961030226e-06,
      "loss": 0.0436,
      "step": 10807
    },
    {
      "epoch": 6.9149072296865,
      "grad_norm": 0.8535726172292639,
      "learning_rate": 4.5901850919142085e-06,
      "loss": 0.0472,
      "step": 10808
    },
    {
      "epoch": 6.915547024952016,
      "grad_norm": 0.763878888785899,
      "learning_rate": 4.588442455086021e-06,
      "loss": 0.0465,
      "step": 10809
    },
    {
      "epoch": 6.91618682021753,
      "grad_norm": 0.9130856677943257,
      "learning_rate": 4.586700050620486e-06,
      "loss": 0.0497,
      "step": 10810
    },
    {
      "epoch": 6.916826615483045,
      "grad_norm": 0.8140415172733788,
      "learning_rate": 4.584957878592423e-06,
      "loss": 0.0482,
      "step": 10811
    },
    {
      "epoch": 6.917466410748561,
      "grad_norm": 0.8252962273093023,
      "learning_rate": 4.583215939076632e-06,
      "loss": 0.0443,
      "step": 10812
    },
    {
      "epoch": 6.918106206014075,
      "grad_norm": 0.7907975146694383,
      "learning_rate": 4.581474232147916e-06,
      "loss": 0.0496,
      "step": 10813
    },
    {
      "epoch": 6.91874600127959,
      "grad_norm": 0.772442482720504,
      "learning_rate": 4.579732757881059e-06,
      "loss": 0.0463,
      "step": 10814
    },
    {
      "epoch": 6.9193857965451055,
      "grad_norm": 0.8576670364442464,
      "learning_rate": 4.57799151635083e-06,
      "loss": 0.0515,
      "step": 10815
    },
    {
      "epoch": 6.920025591810621,
      "grad_norm": 0.9561255782211873,
      "learning_rate": 4.576250507632005e-06,
      "loss": 0.0518,
      "step": 10816
    },
    {
      "epoch": 6.920665387076136,
      "grad_norm": 0.8341378630216776,
      "learning_rate": 4.574509731799335e-06,
      "loss": 0.0484,
      "step": 10817
    },
    {
      "epoch": 6.9213051823416505,
      "grad_norm": 0.8268007555861212,
      "learning_rate": 4.572769188927563e-06,
      "loss": 0.0486,
      "step": 10818
    },
    {
      "epoch": 6.921944977607166,
      "grad_norm": 0.821972213406335,
      "learning_rate": 4.571028879091431e-06,
      "loss": 0.0496,
      "step": 10819
    },
    {
      "epoch": 6.922584772872681,
      "grad_norm": 0.7750900803259944,
      "learning_rate": 4.56928880236566e-06,
      "loss": 0.0451,
      "step": 10820
    },
    {
      "epoch": 6.923224568138195,
      "grad_norm": 0.8731118274815203,
      "learning_rate": 4.567548958824969e-06,
      "loss": 0.0555,
      "step": 10821
    },
    {
      "epoch": 6.923864363403711,
      "grad_norm": 0.7900200842348002,
      "learning_rate": 4.565809348544058e-06,
      "loss": 0.0437,
      "step": 10822
    },
    {
      "epoch": 6.924504158669226,
      "grad_norm": 0.9907983990016774,
      "learning_rate": 4.56406997159763e-06,
      "loss": 0.0553,
      "step": 10823
    },
    {
      "epoch": 6.925143953934741,
      "grad_norm": 0.7501627623401198,
      "learning_rate": 4.562330828060369e-06,
      "loss": 0.0489,
      "step": 10824
    },
    {
      "epoch": 6.925783749200256,
      "grad_norm": 0.8663998477772591,
      "learning_rate": 4.560591918006948e-06,
      "loss": 0.0488,
      "step": 10825
    },
    {
      "epoch": 6.926423544465771,
      "grad_norm": 0.8761146399348216,
      "learning_rate": 4.558853241512035e-06,
      "loss": 0.0537,
      "step": 10826
    },
    {
      "epoch": 6.927063339731286,
      "grad_norm": 0.8059810692514554,
      "learning_rate": 4.557114798650284e-06,
      "loss": 0.0449,
      "step": 10827
    },
    {
      "epoch": 6.927703134996801,
      "grad_norm": 0.817697854616692,
      "learning_rate": 4.5553765894963386e-06,
      "loss": 0.0473,
      "step": 10828
    },
    {
      "epoch": 6.928342930262316,
      "grad_norm": 0.8441310170794051,
      "learning_rate": 4.55363861412484e-06,
      "loss": 0.0499,
      "step": 10829
    },
    {
      "epoch": 6.928982725527831,
      "grad_norm": 0.8319344240157931,
      "learning_rate": 4.551900872610412e-06,
      "loss": 0.0497,
      "step": 10830
    },
    {
      "epoch": 6.929622520793346,
      "grad_norm": 0.8175499507383195,
      "learning_rate": 4.5501633650276675e-06,
      "loss": 0.0499,
      "step": 10831
    },
    {
      "epoch": 6.930262316058861,
      "grad_norm": 0.7749213533563427,
      "learning_rate": 4.548426091451211e-06,
      "loss": 0.0424,
      "step": 10832
    },
    {
      "epoch": 6.930902111324376,
      "grad_norm": 0.8392540244557193,
      "learning_rate": 4.546689051955645e-06,
      "loss": 0.0478,
      "step": 10833
    },
    {
      "epoch": 6.931541906589891,
      "grad_norm": 0.8089487958199219,
      "learning_rate": 4.544952246615549e-06,
      "loss": 0.0545,
      "step": 10834
    },
    {
      "epoch": 6.932181701855407,
      "grad_norm": 0.7656324862920731,
      "learning_rate": 4.543215675505497e-06,
      "loss": 0.0466,
      "step": 10835
    },
    {
      "epoch": 6.932821497120921,
      "grad_norm": 0.8363662684157591,
      "learning_rate": 4.54147933870006e-06,
      "loss": 0.0486,
      "step": 10836
    },
    {
      "epoch": 6.933461292386436,
      "grad_norm": 0.7922614190483535,
      "learning_rate": 4.53974323627379e-06,
      "loss": 0.0494,
      "step": 10837
    },
    {
      "epoch": 6.9341010876519515,
      "grad_norm": 0.9624260403605375,
      "learning_rate": 4.538007368301234e-06,
      "loss": 0.0592,
      "step": 10838
    },
    {
      "epoch": 6.934740882917467,
      "grad_norm": 0.8078759202820468,
      "learning_rate": 4.536271734856924e-06,
      "loss": 0.05,
      "step": 10839
    },
    {
      "epoch": 6.935380678182981,
      "grad_norm": 0.8857685142494266,
      "learning_rate": 4.534536336015387e-06,
      "loss": 0.049,
      "step": 10840
    },
    {
      "epoch": 6.9360204734484965,
      "grad_norm": 0.8858415133880493,
      "learning_rate": 4.532801171851138e-06,
      "loss": 0.047,
      "step": 10841
    },
    {
      "epoch": 6.936660268714012,
      "grad_norm": 0.8320164269942739,
      "learning_rate": 4.5310662424386765e-06,
      "loss": 0.0469,
      "step": 10842
    },
    {
      "epoch": 6.937300063979526,
      "grad_norm": 0.7297177674497333,
      "learning_rate": 4.529331547852507e-06,
      "loss": 0.0434,
      "step": 10843
    },
    {
      "epoch": 6.937939859245041,
      "grad_norm": 0.8700314204630117,
      "learning_rate": 4.527597088167109e-06,
      "loss": 0.0533,
      "step": 10844
    },
    {
      "epoch": 6.938579654510557,
      "grad_norm": 0.8413194666832896,
      "learning_rate": 4.525862863456952e-06,
      "loss": 0.0498,
      "step": 10845
    },
    {
      "epoch": 6.939219449776072,
      "grad_norm": 0.8106408532403758,
      "learning_rate": 4.5241288737965104e-06,
      "loss": 0.0458,
      "step": 10846
    },
    {
      "epoch": 6.939859245041586,
      "grad_norm": 0.797815123372782,
      "learning_rate": 4.5223951192602355e-06,
      "loss": 0.0497,
      "step": 10847
    },
    {
      "epoch": 6.940499040307102,
      "grad_norm": 0.8119180933192485,
      "learning_rate": 4.520661599922569e-06,
      "loss": 0.0455,
      "step": 10848
    },
    {
      "epoch": 6.941138835572617,
      "grad_norm": 0.8891206012223478,
      "learning_rate": 4.518928315857943e-06,
      "loss": 0.0542,
      "step": 10849
    },
    {
      "epoch": 6.941778630838132,
      "grad_norm": 0.7475301120514101,
      "learning_rate": 4.517195267140788e-06,
      "loss": 0.0479,
      "step": 10850
    },
    {
      "epoch": 6.942418426103647,
      "grad_norm": 0.8778824149521612,
      "learning_rate": 4.515462453845516e-06,
      "loss": 0.0514,
      "step": 10851
    },
    {
      "epoch": 6.943058221369162,
      "grad_norm": 0.789293525530128,
      "learning_rate": 4.513729876046527e-06,
      "loss": 0.0503,
      "step": 10852
    },
    {
      "epoch": 6.943698016634677,
      "grad_norm": 0.7369133866548588,
      "learning_rate": 4.511997533818221e-06,
      "loss": 0.044,
      "step": 10853
    },
    {
      "epoch": 6.944337811900192,
      "grad_norm": 0.776048863508292,
      "learning_rate": 4.510265427234978e-06,
      "loss": 0.0468,
      "step": 10854
    },
    {
      "epoch": 6.944977607165707,
      "grad_norm": 0.8906429399914644,
      "learning_rate": 4.508533556371172e-06,
      "loss": 0.0485,
      "step": 10855
    },
    {
      "epoch": 6.945617402431222,
      "grad_norm": 0.7528996526581426,
      "learning_rate": 4.506801921301166e-06,
      "loss": 0.0507,
      "step": 10856
    },
    {
      "epoch": 6.946257197696737,
      "grad_norm": 0.8438009136380762,
      "learning_rate": 4.505070522099315e-06,
      "loss": 0.0485,
      "step": 10857
    },
    {
      "epoch": 6.946896992962252,
      "grad_norm": 0.7623003829435612,
      "learning_rate": 4.503339358839961e-06,
      "loss": 0.0426,
      "step": 10858
    },
    {
      "epoch": 6.947536788227767,
      "grad_norm": 0.8108876675045289,
      "learning_rate": 4.501608431597433e-06,
      "loss": 0.0489,
      "step": 10859
    },
    {
      "epoch": 6.948176583493282,
      "grad_norm": 0.7470153074029572,
      "learning_rate": 4.499877740446061e-06,
      "loss": 0.0444,
      "step": 10860
    },
    {
      "epoch": 6.9488163787587975,
      "grad_norm": 0.7490200957161945,
      "learning_rate": 4.498147285460157e-06,
      "loss": 0.0536,
      "step": 10861
    },
    {
      "epoch": 6.949456174024312,
      "grad_norm": 0.8035623250144031,
      "learning_rate": 4.496417066714016e-06,
      "loss": 0.0434,
      "step": 10862
    },
    {
      "epoch": 6.950095969289827,
      "grad_norm": 0.9056273608877271,
      "learning_rate": 4.4946870842819405e-06,
      "loss": 0.0521,
      "step": 10863
    },
    {
      "epoch": 6.9507357645553425,
      "grad_norm": 0.8077423785081328,
      "learning_rate": 4.492957338238209e-06,
      "loss": 0.0485,
      "step": 10864
    },
    {
      "epoch": 6.951375559820857,
      "grad_norm": 0.7239243121143959,
      "learning_rate": 4.4912278286570875e-06,
      "loss": 0.0493,
      "step": 10865
    },
    {
      "epoch": 6.952015355086372,
      "grad_norm": 0.9565401591602841,
      "learning_rate": 4.4894985556128486e-06,
      "loss": 0.0582,
      "step": 10866
    },
    {
      "epoch": 6.952655150351887,
      "grad_norm": 0.7409704940619335,
      "learning_rate": 4.48776951917974e-06,
      "loss": 0.045,
      "step": 10867
    },
    {
      "epoch": 6.953294945617403,
      "grad_norm": 0.7990225003869971,
      "learning_rate": 4.486040719432001e-06,
      "loss": 0.0456,
      "step": 10868
    },
    {
      "epoch": 6.953934740882918,
      "grad_norm": 0.8163061749622851,
      "learning_rate": 4.4843121564438665e-06,
      "loss": 0.0489,
      "step": 10869
    },
    {
      "epoch": 6.954574536148432,
      "grad_norm": 0.707468088245876,
      "learning_rate": 4.482583830289555e-06,
      "loss": 0.0448,
      "step": 10870
    },
    {
      "epoch": 6.955214331413948,
      "grad_norm": 0.8037057575979439,
      "learning_rate": 4.480855741043281e-06,
      "loss": 0.0521,
      "step": 10871
    },
    {
      "epoch": 6.955854126679463,
      "grad_norm": 0.8330550563015291,
      "learning_rate": 4.4791278887792385e-06,
      "loss": 0.0499,
      "step": 10872
    },
    {
      "epoch": 6.956493921944977,
      "grad_norm": 0.7735652530989227,
      "learning_rate": 4.477400273571627e-06,
      "loss": 0.0411,
      "step": 10873
    },
    {
      "epoch": 6.957133717210493,
      "grad_norm": 0.8788883142202285,
      "learning_rate": 4.475672895494625e-06,
      "loss": 0.0523,
      "step": 10874
    },
    {
      "epoch": 6.957773512476008,
      "grad_norm": 0.75546907714598,
      "learning_rate": 4.4739457546223975e-06,
      "loss": 0.0476,
      "step": 10875
    },
    {
      "epoch": 6.958413307741523,
      "grad_norm": 0.8308841897341076,
      "learning_rate": 4.472218851029113e-06,
      "loss": 0.0454,
      "step": 10876
    },
    {
      "epoch": 6.9590531030070375,
      "grad_norm": 0.8552528616826063,
      "learning_rate": 4.470492184788918e-06,
      "loss": 0.0507,
      "step": 10877
    },
    {
      "epoch": 6.959692898272553,
      "grad_norm": 0.8886141088389792,
      "learning_rate": 4.468765755975952e-06,
      "loss": 0.0506,
      "step": 10878
    },
    {
      "epoch": 6.960332693538068,
      "grad_norm": 0.7583723896401459,
      "learning_rate": 4.467039564664342e-06,
      "loss": 0.0456,
      "step": 10879
    },
    {
      "epoch": 6.960972488803582,
      "grad_norm": 0.7859971951315367,
      "learning_rate": 4.4653136109282146e-06,
      "loss": 0.0486,
      "step": 10880
    },
    {
      "epoch": 6.961612284069098,
      "grad_norm": 0.8556777269073254,
      "learning_rate": 4.463587894841674e-06,
      "loss": 0.0503,
      "step": 10881
    },
    {
      "epoch": 6.962252079334613,
      "grad_norm": 0.8000302226944479,
      "learning_rate": 4.461862416478818e-06,
      "loss": 0.0494,
      "step": 10882
    },
    {
      "epoch": 6.962891874600128,
      "grad_norm": 0.7639620916203886,
      "learning_rate": 4.4601371759137424e-06,
      "loss": 0.0487,
      "step": 10883
    },
    {
      "epoch": 6.963531669865643,
      "grad_norm": 0.8550331062553984,
      "learning_rate": 4.458412173220521e-06,
      "loss": 0.0502,
      "step": 10884
    },
    {
      "epoch": 6.964171465131158,
      "grad_norm": 0.799744573018089,
      "learning_rate": 4.456687408473222e-06,
      "loss": 0.0521,
      "step": 10885
    },
    {
      "epoch": 6.964811260396673,
      "grad_norm": 0.7977268210662286,
      "learning_rate": 4.4549628817459066e-06,
      "loss": 0.0527,
      "step": 10886
    },
    {
      "epoch": 6.9654510556621885,
      "grad_norm": 0.776764955522324,
      "learning_rate": 4.453238593112621e-06,
      "loss": 0.0452,
      "step": 10887
    },
    {
      "epoch": 6.966090850927703,
      "grad_norm": 0.8110958412210716,
      "learning_rate": 4.4515145426474035e-06,
      "loss": 0.0452,
      "step": 10888
    },
    {
      "epoch": 6.966730646193218,
      "grad_norm": 0.7427433847328149,
      "learning_rate": 4.449790730424277e-06,
      "loss": 0.0451,
      "step": 10889
    },
    {
      "epoch": 6.967370441458733,
      "grad_norm": 0.7928411994584883,
      "learning_rate": 4.448067156517268e-06,
      "loss": 0.0462,
      "step": 10890
    },
    {
      "epoch": 6.968010236724249,
      "grad_norm": 0.798402162521014,
      "learning_rate": 4.446343821000378e-06,
      "loss": 0.0463,
      "step": 10891
    },
    {
      "epoch": 6.968650031989763,
      "grad_norm": 0.762477461943378,
      "learning_rate": 4.444620723947603e-06,
      "loss": 0.0471,
      "step": 10892
    },
    {
      "epoch": 6.969289827255278,
      "grad_norm": 0.7977578891202793,
      "learning_rate": 4.442897865432934e-06,
      "loss": 0.0497,
      "step": 10893
    },
    {
      "epoch": 6.969929622520794,
      "grad_norm": 0.8176982039765226,
      "learning_rate": 4.441175245530347e-06,
      "loss": 0.054,
      "step": 10894
    },
    {
      "epoch": 6.970569417786308,
      "grad_norm": 0.7517354065568175,
      "learning_rate": 4.4394528643138025e-06,
      "loss": 0.0512,
      "step": 10895
    },
    {
      "epoch": 6.971209213051823,
      "grad_norm": 0.7571658373666307,
      "learning_rate": 4.437730721857263e-06,
      "loss": 0.0429,
      "step": 10896
    },
    {
      "epoch": 6.971849008317339,
      "grad_norm": 0.7976853350855848,
      "learning_rate": 4.436008818234673e-06,
      "loss": 0.0503,
      "step": 10897
    },
    {
      "epoch": 6.972488803582854,
      "grad_norm": 0.8261256085623258,
      "learning_rate": 4.434287153519966e-06,
      "loss": 0.0521,
      "step": 10898
    },
    {
      "epoch": 6.973128598848368,
      "grad_norm": 0.6360106284377289,
      "learning_rate": 4.432565727787069e-06,
      "loss": 0.0395,
      "step": 10899
    },
    {
      "epoch": 6.9737683941138835,
      "grad_norm": 0.8590095222438845,
      "learning_rate": 4.4308445411098954e-06,
      "loss": 0.0514,
      "step": 10900
    },
    {
      "epoch": 6.974408189379399,
      "grad_norm": 0.8195204168574215,
      "learning_rate": 4.42912359356235e-06,
      "loss": 0.0474,
      "step": 10901
    },
    {
      "epoch": 6.975047984644913,
      "grad_norm": 0.7520944891005522,
      "learning_rate": 4.427402885218325e-06,
      "loss": 0.0471,
      "step": 10902
    },
    {
      "epoch": 6.975687779910428,
      "grad_norm": 0.7491490142829822,
      "learning_rate": 4.42568241615171e-06,
      "loss": 0.0483,
      "step": 10903
    },
    {
      "epoch": 6.976327575175944,
      "grad_norm": 0.853375488619133,
      "learning_rate": 4.4239621864363755e-06,
      "loss": 0.0516,
      "step": 10904
    },
    {
      "epoch": 6.976967370441459,
      "grad_norm": 0.8603760361616946,
      "learning_rate": 4.4222421961461834e-06,
      "loss": 0.0568,
      "step": 10905
    },
    {
      "epoch": 6.977607165706974,
      "grad_norm": 0.7595459761505127,
      "learning_rate": 4.420522445354992e-06,
      "loss": 0.0478,
      "step": 10906
    },
    {
      "epoch": 6.978246960972489,
      "grad_norm": 0.9233537792012778,
      "learning_rate": 4.418802934136642e-06,
      "loss": 0.0568,
      "step": 10907
    },
    {
      "epoch": 6.978886756238004,
      "grad_norm": 0.7930911237074106,
      "learning_rate": 4.417083662564966e-06,
      "loss": 0.0501,
      "step": 10908
    },
    {
      "epoch": 6.979526551503519,
      "grad_norm": 0.7220699693035505,
      "learning_rate": 4.4153646307137836e-06,
      "loss": 0.0451,
      "step": 10909
    },
    {
      "epoch": 6.980166346769034,
      "grad_norm": 0.8630535127474118,
      "learning_rate": 4.413645838656911e-06,
      "loss": 0.0486,
      "step": 10910
    },
    {
      "epoch": 6.980806142034549,
      "grad_norm": 0.9528679989761136,
      "learning_rate": 4.411927286468151e-06,
      "loss": 0.0474,
      "step": 10911
    },
    {
      "epoch": 6.981445937300064,
      "grad_norm": 0.8416581998352143,
      "learning_rate": 4.4102089742212885e-06,
      "loss": 0.0462,
      "step": 10912
    },
    {
      "epoch": 6.982085732565579,
      "grad_norm": 0.9208024789786421,
      "learning_rate": 4.408490901990112e-06,
      "loss": 0.0526,
      "step": 10913
    },
    {
      "epoch": 6.982725527831094,
      "grad_norm": 0.7708484827712901,
      "learning_rate": 4.406773069848391e-06,
      "loss": 0.0503,
      "step": 10914
    },
    {
      "epoch": 6.983365323096609,
      "grad_norm": 0.8467152749246051,
      "learning_rate": 4.405055477869884e-06,
      "loss": 0.0523,
      "step": 10915
    },
    {
      "epoch": 6.984005118362124,
      "grad_norm": 0.6314736605393657,
      "learning_rate": 4.403338126128343e-06,
      "loss": 0.0417,
      "step": 10916
    },
    {
      "epoch": 6.984644913627639,
      "grad_norm": 0.9695052842897866,
      "learning_rate": 4.401621014697506e-06,
      "loss": 0.049,
      "step": 10917
    },
    {
      "epoch": 6.985284708893154,
      "grad_norm": 0.8322368451086961,
      "learning_rate": 4.3999041436511035e-06,
      "loss": 0.0474,
      "step": 10918
    },
    {
      "epoch": 6.985924504158669,
      "grad_norm": 0.8313037051965726,
      "learning_rate": 4.398187513062853e-06,
      "loss": 0.0445,
      "step": 10919
    },
    {
      "epoch": 6.9865642994241846,
      "grad_norm": 0.8964376348835684,
      "learning_rate": 4.396471123006469e-06,
      "loss": 0.0502,
      "step": 10920
    },
    {
      "epoch": 6.987204094689699,
      "grad_norm": 0.7813218722902895,
      "learning_rate": 4.3947549735556476e-06,
      "loss": 0.0485,
      "step": 10921
    },
    {
      "epoch": 6.987843889955214,
      "grad_norm": 0.8414803941662119,
      "learning_rate": 4.393039064784072e-06,
      "loss": 0.0472,
      "step": 10922
    },
    {
      "epoch": 6.9884836852207295,
      "grad_norm": 0.9107670120672986,
      "learning_rate": 4.391323396765429e-06,
      "loss": 0.0525,
      "step": 10923
    },
    {
      "epoch": 6.989123480486245,
      "grad_norm": 0.7928660661328154,
      "learning_rate": 4.389607969573382e-06,
      "loss": 0.0449,
      "step": 10924
    },
    {
      "epoch": 6.989763275751759,
      "grad_norm": 0.8539306835647923,
      "learning_rate": 4.387892783281589e-06,
      "loss": 0.0567,
      "step": 10925
    },
    {
      "epoch": 6.990403071017274,
      "grad_norm": 0.8994563326423718,
      "learning_rate": 4.3861778379636935e-06,
      "loss": 0.0515,
      "step": 10926
    },
    {
      "epoch": 6.99104286628279,
      "grad_norm": 0.8303397455868116,
      "learning_rate": 4.384463133693339e-06,
      "loss": 0.049,
      "step": 10927
    },
    {
      "epoch": 6.991682661548305,
      "grad_norm": 0.7705104630934144,
      "learning_rate": 4.382748670544147e-06,
      "loss": 0.0456,
      "step": 10928
    },
    {
      "epoch": 6.992322456813819,
      "grad_norm": 0.7654288651428437,
      "learning_rate": 4.381034448589736e-06,
      "loss": 0.0453,
      "step": 10929
    },
    {
      "epoch": 6.992962252079335,
      "grad_norm": 0.8515712399622621,
      "learning_rate": 4.379320467903711e-06,
      "loss": 0.0459,
      "step": 10930
    },
    {
      "epoch": 6.99360204734485,
      "grad_norm": 0.7868870659096509,
      "learning_rate": 4.377606728559664e-06,
      "loss": 0.0457,
      "step": 10931
    },
    {
      "epoch": 6.994241842610364,
      "grad_norm": 0.871287559393763,
      "learning_rate": 4.375893230631182e-06,
      "loss": 0.049,
      "step": 10932
    },
    {
      "epoch": 6.99488163787588,
      "grad_norm": 0.7876730282919926,
      "learning_rate": 4.374179974191842e-06,
      "loss": 0.0472,
      "step": 10933
    },
    {
      "epoch": 6.995521433141395,
      "grad_norm": 0.8153438837284085,
      "learning_rate": 4.3724669593152065e-06,
      "loss": 0.0556,
      "step": 10934
    },
    {
      "epoch": 6.99616122840691,
      "grad_norm": 0.767717473354232,
      "learning_rate": 4.3707541860748305e-06,
      "loss": 0.0503,
      "step": 10935
    },
    {
      "epoch": 6.9968010236724245,
      "grad_norm": 0.7603229614381711,
      "learning_rate": 4.369041654544251e-06,
      "loss": 0.0481,
      "step": 10936
    },
    {
      "epoch": 6.99744081893794,
      "grad_norm": 0.7352478288285761,
      "learning_rate": 4.3673293647970105e-06,
      "loss": 0.0443,
      "step": 10937
    },
    {
      "epoch": 6.998080614203455,
      "grad_norm": 0.8228577530024638,
      "learning_rate": 4.3656173169066275e-06,
      "loss": 0.0453,
      "step": 10938
    },
    {
      "epoch": 6.9987204094689695,
      "grad_norm": 0.7809240465257489,
      "learning_rate": 4.36390551094661e-06,
      "loss": 0.0498,
      "step": 10939
    },
    {
      "epoch": 6.999360204734485,
      "grad_norm": 0.76905430717918,
      "learning_rate": 4.362193946990467e-06,
      "loss": 0.0503,
      "step": 10940
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.6759700815136456,
      "learning_rate": 4.360482625111688e-06,
      "loss": 0.0418,
      "step": 10941
    },
    {
      "epoch": 7.000639795265515,
      "grad_norm": 0.6180923891108273,
      "learning_rate": 4.35877154538375e-06,
      "loss": 0.0324,
      "step": 10942
    },
    {
      "epoch": 7.00127959053103,
      "grad_norm": 0.550960601418994,
      "learning_rate": 4.35706070788013e-06,
      "loss": 0.026,
      "step": 10943
    },
    {
      "epoch": 7.001919385796545,
      "grad_norm": 0.5598819350413567,
      "learning_rate": 4.355350112674285e-06,
      "loss": 0.0314,
      "step": 10944
    },
    {
      "epoch": 7.00255918106206,
      "grad_norm": 0.5056654952364772,
      "learning_rate": 4.3536397598396655e-06,
      "loss": 0.03,
      "step": 10945
    },
    {
      "epoch": 7.0031989763275755,
      "grad_norm": 0.5281293662362467,
      "learning_rate": 4.35192964944971e-06,
      "loss": 0.0302,
      "step": 10946
    },
    {
      "epoch": 7.00383877159309,
      "grad_norm": 0.5875953189466421,
      "learning_rate": 4.350219781577848e-06,
      "loss": 0.0234,
      "step": 10947
    },
    {
      "epoch": 7.004478566858605,
      "grad_norm": 0.6618444292348159,
      "learning_rate": 4.3485101562975e-06,
      "loss": 0.0268,
      "step": 10948
    },
    {
      "epoch": 7.00511836212412,
      "grad_norm": 0.5367583824341955,
      "learning_rate": 4.346800773682067e-06,
      "loss": 0.0267,
      "step": 10949
    },
    {
      "epoch": 7.005758157389636,
      "grad_norm": 0.592757166553704,
      "learning_rate": 4.345091633804957e-06,
      "loss": 0.0286,
      "step": 10950
    },
    {
      "epoch": 7.00639795265515,
      "grad_norm": 0.6114852879234086,
      "learning_rate": 4.343382736739553e-06,
      "loss": 0.0233,
      "step": 10951
    },
    {
      "epoch": 7.007037747920665,
      "grad_norm": 0.6401398811951036,
      "learning_rate": 4.341674082559228e-06,
      "loss": 0.0237,
      "step": 10952
    },
    {
      "epoch": 7.007677543186181,
      "grad_norm": 0.648757232288395,
      "learning_rate": 4.339965671337358e-06,
      "loss": 0.0303,
      "step": 10953
    },
    {
      "epoch": 7.008317338451695,
      "grad_norm": 0.5570487283917721,
      "learning_rate": 4.338257503147291e-06,
      "loss": 0.0181,
      "step": 10954
    },
    {
      "epoch": 7.00895713371721,
      "grad_norm": 0.526626942678054,
      "learning_rate": 4.336549578062377e-06,
      "loss": 0.0233,
      "step": 10955
    },
    {
      "epoch": 7.009596928982726,
      "grad_norm": 0.5305684984608455,
      "learning_rate": 4.334841896155945e-06,
      "loss": 0.0218,
      "step": 10956
    },
    {
      "epoch": 7.010236724248241,
      "grad_norm": 0.6826356748825128,
      "learning_rate": 4.333134457501329e-06,
      "loss": 0.0263,
      "step": 10957
    },
    {
      "epoch": 7.010876519513755,
      "grad_norm": 0.6960712118086532,
      "learning_rate": 4.331427262171838e-06,
      "loss": 0.0242,
      "step": 10958
    },
    {
      "epoch": 7.0115163147792705,
      "grad_norm": 0.7236594810152172,
      "learning_rate": 4.329720310240777e-06,
      "loss": 0.0296,
      "step": 10959
    },
    {
      "epoch": 7.012156110044786,
      "grad_norm": 0.5919788963570148,
      "learning_rate": 4.32801360178144e-06,
      "loss": 0.0204,
      "step": 10960
    },
    {
      "epoch": 7.012795905310301,
      "grad_norm": 0.6075909393838579,
      "learning_rate": 4.3263071368671085e-06,
      "loss": 0.0223,
      "step": 10961
    },
    {
      "epoch": 7.0134357005758154,
      "grad_norm": 0.6695756637449644,
      "learning_rate": 4.324600915571052e-06,
      "loss": 0.0247,
      "step": 10962
    },
    {
      "epoch": 7.014075495841331,
      "grad_norm": 0.6926535810437789,
      "learning_rate": 4.322894937966541e-06,
      "loss": 0.0254,
      "step": 10963
    },
    {
      "epoch": 7.014715291106846,
      "grad_norm": 0.7156816593254041,
      "learning_rate": 4.321189204126821e-06,
      "loss": 0.0253,
      "step": 10964
    },
    {
      "epoch": 7.015355086372361,
      "grad_norm": 0.6668908731381539,
      "learning_rate": 4.319483714125136e-06,
      "loss": 0.0288,
      "step": 10965
    },
    {
      "epoch": 7.015994881637876,
      "grad_norm": 0.49567236757081007,
      "learning_rate": 4.317778468034711e-06,
      "loss": 0.0203,
      "step": 10966
    },
    {
      "epoch": 7.016634676903391,
      "grad_norm": 0.6218167505713831,
      "learning_rate": 4.3160734659287765e-06,
      "loss": 0.0279,
      "step": 10967
    },
    {
      "epoch": 7.017274472168906,
      "grad_norm": 0.6402863478408132,
      "learning_rate": 4.314368707880534e-06,
      "loss": 0.0282,
      "step": 10968
    },
    {
      "epoch": 7.017914267434421,
      "grad_norm": 0.6039538737306871,
      "learning_rate": 4.312664193963182e-06,
      "loss": 0.0245,
      "step": 10969
    },
    {
      "epoch": 7.018554062699936,
      "grad_norm": 0.6081609477955583,
      "learning_rate": 4.310959924249917e-06,
      "loss": 0.0261,
      "step": 10970
    },
    {
      "epoch": 7.019193857965451,
      "grad_norm": 0.7448931741790641,
      "learning_rate": 4.3092558988139126e-06,
      "loss": 0.0245,
      "step": 10971
    },
    {
      "epoch": 7.019833653230966,
      "grad_norm": 0.7192408831025835,
      "learning_rate": 4.307552117728333e-06,
      "loss": 0.0292,
      "step": 10972
    },
    {
      "epoch": 7.020473448496481,
      "grad_norm": 0.5817050382062507,
      "learning_rate": 4.305848581066343e-06,
      "loss": 0.0234,
      "step": 10973
    },
    {
      "epoch": 7.021113243761996,
      "grad_norm": 0.5761947658938161,
      "learning_rate": 4.304145288901086e-06,
      "loss": 0.0233,
      "step": 10974
    },
    {
      "epoch": 7.021753039027511,
      "grad_norm": 0.6459735249882247,
      "learning_rate": 4.3024422413056975e-06,
      "loss": 0.0243,
      "step": 10975
    },
    {
      "epoch": 7.022392834293027,
      "grad_norm": 0.5808845936134474,
      "learning_rate": 4.300739438353305e-06,
      "loss": 0.0254,
      "step": 10976
    },
    {
      "epoch": 7.023032629558541,
      "grad_norm": 0.7002974509927798,
      "learning_rate": 4.2990368801170225e-06,
      "loss": 0.0251,
      "step": 10977
    },
    {
      "epoch": 7.023672424824056,
      "grad_norm": 0.5616001605703426,
      "learning_rate": 4.297334566669955e-06,
      "loss": 0.0219,
      "step": 10978
    },
    {
      "epoch": 7.024312220089572,
      "grad_norm": 0.6000667927812628,
      "learning_rate": 4.295632498085193e-06,
      "loss": 0.0202,
      "step": 10979
    },
    {
      "epoch": 7.024952015355086,
      "grad_norm": 0.6611534861708034,
      "learning_rate": 4.293930674435828e-06,
      "loss": 0.0227,
      "step": 10980
    },
    {
      "epoch": 7.025591810620601,
      "grad_norm": 0.5932067270354013,
      "learning_rate": 4.29222909579493e-06,
      "loss": 0.023,
      "step": 10981
    },
    {
      "epoch": 7.0262316058861165,
      "grad_norm": 0.6760033825905312,
      "learning_rate": 4.290527762235558e-06,
      "loss": 0.0317,
      "step": 10982
    },
    {
      "epoch": 7.026871401151632,
      "grad_norm": 0.6531340910455845,
      "learning_rate": 4.288826673830771e-06,
      "loss": 0.0263,
      "step": 10983
    },
    {
      "epoch": 7.027511196417146,
      "grad_norm": 0.7039180894230803,
      "learning_rate": 4.287125830653605e-06,
      "loss": 0.0267,
      "step": 10984
    },
    {
      "epoch": 7.028150991682661,
      "grad_norm": 0.6307101433793981,
      "learning_rate": 4.2854252327770965e-06,
      "loss": 0.0233,
      "step": 10985
    },
    {
      "epoch": 7.028790786948177,
      "grad_norm": 0.7044632808175111,
      "learning_rate": 4.283724880274257e-06,
      "loss": 0.0239,
      "step": 10986
    },
    {
      "epoch": 7.029430582213692,
      "grad_norm": 0.6565090435393623,
      "learning_rate": 4.282024773218107e-06,
      "loss": 0.024,
      "step": 10987
    },
    {
      "epoch": 7.030070377479206,
      "grad_norm": 0.6523959973606287,
      "learning_rate": 4.280324911681644e-06,
      "loss": 0.024,
      "step": 10988
    },
    {
      "epoch": 7.030710172744722,
      "grad_norm": 0.7422999890176962,
      "learning_rate": 4.278625295737853e-06,
      "loss": 0.0216,
      "step": 10989
    },
    {
      "epoch": 7.031349968010237,
      "grad_norm": 0.7084834007412485,
      "learning_rate": 4.276925925459713e-06,
      "loss": 0.0219,
      "step": 10990
    },
    {
      "epoch": 7.031989763275751,
      "grad_norm": 0.5353850073336419,
      "learning_rate": 4.2752268009201945e-06,
      "loss": 0.0226,
      "step": 10991
    },
    {
      "epoch": 7.032629558541267,
      "grad_norm": 0.5488320253051058,
      "learning_rate": 4.2735279221922495e-06,
      "loss": 0.0213,
      "step": 10992
    },
    {
      "epoch": 7.033269353806782,
      "grad_norm": 0.6823499123253156,
      "learning_rate": 4.271829289348834e-06,
      "loss": 0.0238,
      "step": 10993
    },
    {
      "epoch": 7.033909149072297,
      "grad_norm": 0.6556762928127441,
      "learning_rate": 4.270130902462878e-06,
      "loss": 0.0217,
      "step": 10994
    },
    {
      "epoch": 7.0345489443378115,
      "grad_norm": 0.6811722290379104,
      "learning_rate": 4.2684327616073085e-06,
      "loss": 0.0253,
      "step": 10995
    },
    {
      "epoch": 7.035188739603327,
      "grad_norm": 0.5796822843718218,
      "learning_rate": 4.266734866855037e-06,
      "loss": 0.0257,
      "step": 10996
    },
    {
      "epoch": 7.035828534868842,
      "grad_norm": 0.6607718644006634,
      "learning_rate": 4.2650372182789745e-06,
      "loss": 0.0246,
      "step": 10997
    },
    {
      "epoch": 7.036468330134357,
      "grad_norm": 0.6032932108335716,
      "learning_rate": 4.263339815952014e-06,
      "loss": 0.0214,
      "step": 10998
    },
    {
      "epoch": 7.037108125399872,
      "grad_norm": 0.6977630357558922,
      "learning_rate": 4.261642659947032e-06,
      "loss": 0.0215,
      "step": 10999
    },
    {
      "epoch": 7.037747920665387,
      "grad_norm": 0.5198689505795598,
      "learning_rate": 4.2599457503369105e-06,
      "loss": 0.0219,
      "step": 11000
    },
    {
      "epoch": 7.038387715930902,
      "grad_norm": 0.5868511177070703,
      "learning_rate": 4.258249087194507e-06,
      "loss": 0.0216,
      "step": 11001
    },
    {
      "epoch": 7.039027511196418,
      "grad_norm": 0.7802855751855073,
      "learning_rate": 4.256552670592674e-06,
      "loss": 0.0274,
      "step": 11002
    },
    {
      "epoch": 7.039667306461932,
      "grad_norm": 0.7671947731622161,
      "learning_rate": 4.2548565006042495e-06,
      "loss": 0.0254,
      "step": 11003
    },
    {
      "epoch": 7.040307101727447,
      "grad_norm": 0.5881879382456617,
      "learning_rate": 4.253160577302071e-06,
      "loss": 0.022,
      "step": 11004
    },
    {
      "epoch": 7.0409468969929625,
      "grad_norm": 0.5683373782312939,
      "learning_rate": 4.251464900758953e-06,
      "loss": 0.0209,
      "step": 11005
    },
    {
      "epoch": 7.041586692258477,
      "grad_norm": 0.6641333219772076,
      "learning_rate": 4.249769471047708e-06,
      "loss": 0.024,
      "step": 11006
    },
    {
      "epoch": 7.042226487523992,
      "grad_norm": 0.8502983642062598,
      "learning_rate": 4.248074288241133e-06,
      "loss": 0.0302,
      "step": 11007
    },
    {
      "epoch": 7.042866282789507,
      "grad_norm": 0.5829615700467333,
      "learning_rate": 4.246379352412016e-06,
      "loss": 0.0247,
      "step": 11008
    },
    {
      "epoch": 7.043506078055023,
      "grad_norm": 0.5564211037318165,
      "learning_rate": 4.244684663633129e-06,
      "loss": 0.0215,
      "step": 11009
    },
    {
      "epoch": 7.044145873320537,
      "grad_norm": 0.6181411062182243,
      "learning_rate": 4.242990221977251e-06,
      "loss": 0.0231,
      "step": 11010
    },
    {
      "epoch": 7.044785668586052,
      "grad_norm": 0.6835109535622265,
      "learning_rate": 4.241296027517131e-06,
      "loss": 0.0263,
      "step": 11011
    },
    {
      "epoch": 7.045425463851568,
      "grad_norm": 0.5860106469918929,
      "learning_rate": 4.2396020803255156e-06,
      "loss": 0.0217,
      "step": 11012
    },
    {
      "epoch": 7.046065259117083,
      "grad_norm": 0.5964705050994953,
      "learning_rate": 4.2379083804751355e-06,
      "loss": 0.023,
      "step": 11013
    },
    {
      "epoch": 7.046705054382597,
      "grad_norm": 0.6931027110148541,
      "learning_rate": 4.236214928038724e-06,
      "loss": 0.0245,
      "step": 11014
    },
    {
      "epoch": 7.047344849648113,
      "grad_norm": 0.6286320038010677,
      "learning_rate": 4.23452172308899e-06,
      "loss": 0.0221,
      "step": 11015
    },
    {
      "epoch": 7.047984644913628,
      "grad_norm": 0.6321190426721174,
      "learning_rate": 4.232828765698634e-06,
      "loss": 0.0231,
      "step": 11016
    },
    {
      "epoch": 7.048624440179142,
      "grad_norm": 0.6629758067447754,
      "learning_rate": 4.231136055940356e-06,
      "loss": 0.0269,
      "step": 11017
    },
    {
      "epoch": 7.0492642354446575,
      "grad_norm": 0.6616802912845887,
      "learning_rate": 4.229443593886833e-06,
      "loss": 0.0252,
      "step": 11018
    },
    {
      "epoch": 7.049904030710173,
      "grad_norm": 0.6901079925733673,
      "learning_rate": 4.227751379610737e-06,
      "loss": 0.0188,
      "step": 11019
    },
    {
      "epoch": 7.050543825975688,
      "grad_norm": 0.6463557109670706,
      "learning_rate": 4.226059413184729e-06,
      "loss": 0.0213,
      "step": 11020
    },
    {
      "epoch": 7.0511836212412025,
      "grad_norm": 0.7519231000902123,
      "learning_rate": 4.22436769468146e-06,
      "loss": 0.028,
      "step": 11021
    },
    {
      "epoch": 7.051823416506718,
      "grad_norm": 0.6543935597140841,
      "learning_rate": 4.222676224173565e-06,
      "loss": 0.025,
      "step": 11022
    },
    {
      "epoch": 7.052463211772233,
      "grad_norm": 0.7389063069316941,
      "learning_rate": 4.220985001733675e-06,
      "loss": 0.0255,
      "step": 11023
    },
    {
      "epoch": 7.053103007037748,
      "grad_norm": 0.6095439589872944,
      "learning_rate": 4.219294027434412e-06,
      "loss": 0.0231,
      "step": 11024
    },
    {
      "epoch": 7.053742802303263,
      "grad_norm": 0.7456642112040499,
      "learning_rate": 4.21760330134838e-06,
      "loss": 0.0233,
      "step": 11025
    },
    {
      "epoch": 7.054382597568778,
      "grad_norm": 0.6886827990497979,
      "learning_rate": 4.215912823548172e-06,
      "loss": 0.0256,
      "step": 11026
    },
    {
      "epoch": 7.055022392834293,
      "grad_norm": 0.7578396981341939,
      "learning_rate": 4.214222594106383e-06,
      "loss": 0.0276,
      "step": 11027
    },
    {
      "epoch": 7.0556621880998085,
      "grad_norm": 0.5773505121356426,
      "learning_rate": 4.212532613095582e-06,
      "loss": 0.0206,
      "step": 11028
    },
    {
      "epoch": 7.056301983365323,
      "grad_norm": 0.7745131638400413,
      "learning_rate": 4.210842880588334e-06,
      "loss": 0.0245,
      "step": 11029
    },
    {
      "epoch": 7.056941778630838,
      "grad_norm": 0.7940730442578723,
      "learning_rate": 4.209153396657197e-06,
      "loss": 0.0281,
      "step": 11030
    },
    {
      "epoch": 7.057581573896353,
      "grad_norm": 0.6780704949786708,
      "learning_rate": 4.207464161374713e-06,
      "loss": 0.0236,
      "step": 11031
    },
    {
      "epoch": 7.058221369161868,
      "grad_norm": 0.7264139877760736,
      "learning_rate": 4.205775174813412e-06,
      "loss": 0.0271,
      "step": 11032
    },
    {
      "epoch": 7.058861164427383,
      "grad_norm": 0.6835044008513358,
      "learning_rate": 4.204086437045816e-06,
      "loss": 0.0248,
      "step": 11033
    },
    {
      "epoch": 7.059500959692898,
      "grad_norm": 0.6243807070789668,
      "learning_rate": 4.202397948144441e-06,
      "loss": 0.0212,
      "step": 11034
    },
    {
      "epoch": 7.060140754958414,
      "grad_norm": 0.6003275650778115,
      "learning_rate": 4.200709708181785e-06,
      "loss": 0.0199,
      "step": 11035
    },
    {
      "epoch": 7.060780550223928,
      "grad_norm": 0.5517205960588789,
      "learning_rate": 4.199021717230339e-06,
      "loss": 0.0212,
      "step": 11036
    },
    {
      "epoch": 7.061420345489443,
      "grad_norm": 0.6934114126970268,
      "learning_rate": 4.19733397536258e-06,
      "loss": 0.0241,
      "step": 11037
    },
    {
      "epoch": 7.062060140754959,
      "grad_norm": 0.5630776440118852,
      "learning_rate": 4.195646482650978e-06,
      "loss": 0.0199,
      "step": 11038
    },
    {
      "epoch": 7.062699936020474,
      "grad_norm": 0.6029932389712144,
      "learning_rate": 4.193959239167989e-06,
      "loss": 0.0203,
      "step": 11039
    },
    {
      "epoch": 7.063339731285988,
      "grad_norm": 0.5993233177986753,
      "learning_rate": 4.1922722449860634e-06,
      "loss": 0.0223,
      "step": 11040
    },
    {
      "epoch": 7.0639795265515035,
      "grad_norm": 0.6109546339595339,
      "learning_rate": 4.190585500177638e-06,
      "loss": 0.0219,
      "step": 11041
    },
    {
      "epoch": 7.064619321817019,
      "grad_norm": 0.6962102542722011,
      "learning_rate": 4.188899004815135e-06,
      "loss": 0.0236,
      "step": 11042
    },
    {
      "epoch": 7.065259117082533,
      "grad_norm": 0.7241651872587879,
      "learning_rate": 4.187212758970971e-06,
      "loss": 0.0232,
      "step": 11043
    },
    {
      "epoch": 7.0658989123480485,
      "grad_norm": 0.6228188039003293,
      "learning_rate": 4.185526762717553e-06,
      "loss": 0.0217,
      "step": 11044
    },
    {
      "epoch": 7.066538707613564,
      "grad_norm": 0.6830620156170617,
      "learning_rate": 4.183841016127273e-06,
      "loss": 0.0262,
      "step": 11045
    },
    {
      "epoch": 7.067178502879079,
      "grad_norm": 0.6865591794872227,
      "learning_rate": 4.18215551927251e-06,
      "loss": 0.0251,
      "step": 11046
    },
    {
      "epoch": 7.067818298144593,
      "grad_norm": 0.6090124930237392,
      "learning_rate": 4.1804702722256426e-06,
      "loss": 0.0221,
      "step": 11047
    },
    {
      "epoch": 7.068458093410109,
      "grad_norm": 0.5866831071572428,
      "learning_rate": 4.178785275059031e-06,
      "loss": 0.0201,
      "step": 11048
    },
    {
      "epoch": 7.069097888675624,
      "grad_norm": 0.6568370468834261,
      "learning_rate": 4.177100527845024e-06,
      "loss": 0.0238,
      "step": 11049
    },
    {
      "epoch": 7.069737683941139,
      "grad_norm": 0.5891973767760194,
      "learning_rate": 4.175416030655961e-06,
      "loss": 0.0225,
      "step": 11050
    },
    {
      "epoch": 7.070377479206654,
      "grad_norm": 0.7261007072523817,
      "learning_rate": 4.173731783564174e-06,
      "loss": 0.0222,
      "step": 11051
    },
    {
      "epoch": 7.071017274472169,
      "grad_norm": 0.5481482599697892,
      "learning_rate": 4.172047786641979e-06,
      "loss": 0.0187,
      "step": 11052
    },
    {
      "epoch": 7.071657069737684,
      "grad_norm": 0.6770120086465218,
      "learning_rate": 4.170364039961683e-06,
      "loss": 0.0233,
      "step": 11053
    },
    {
      "epoch": 7.0722968650031985,
      "grad_norm": 0.6971598642944732,
      "learning_rate": 4.168680543595587e-06,
      "loss": 0.0242,
      "step": 11054
    },
    {
      "epoch": 7.072936660268714,
      "grad_norm": 0.6693466140878129,
      "learning_rate": 4.166997297615976e-06,
      "loss": 0.025,
      "step": 11055
    },
    {
      "epoch": 7.073576455534229,
      "grad_norm": 0.6258243729149873,
      "learning_rate": 4.165314302095121e-06,
      "loss": 0.0226,
      "step": 11056
    },
    {
      "epoch": 7.074216250799744,
      "grad_norm": 0.610236585098562,
      "learning_rate": 4.163631557105294e-06,
      "loss": 0.0211,
      "step": 11057
    },
    {
      "epoch": 7.074856046065259,
      "grad_norm": 0.7141282384482649,
      "learning_rate": 4.1619490627187475e-06,
      "loss": 0.0214,
      "step": 11058
    },
    {
      "epoch": 7.075495841330774,
      "grad_norm": 0.7183687194299193,
      "learning_rate": 4.160266819007718e-06,
      "loss": 0.0224,
      "step": 11059
    },
    {
      "epoch": 7.076135636596289,
      "grad_norm": 0.910723924301988,
      "learning_rate": 4.158584826044447e-06,
      "loss": 0.0278,
      "step": 11060
    },
    {
      "epoch": 7.076775431861805,
      "grad_norm": 0.7344808617905186,
      "learning_rate": 4.1569030839011535e-06,
      "loss": 0.025,
      "step": 11061
    },
    {
      "epoch": 7.077415227127319,
      "grad_norm": 0.6277292548072883,
      "learning_rate": 4.155221592650047e-06,
      "loss": 0.0207,
      "step": 11062
    },
    {
      "epoch": 7.078055022392834,
      "grad_norm": 0.7098736591665146,
      "learning_rate": 4.153540352363325e-06,
      "loss": 0.0266,
      "step": 11063
    },
    {
      "epoch": 7.0786948176583495,
      "grad_norm": 0.7496813544987824,
      "learning_rate": 4.151859363113182e-06,
      "loss": 0.026,
      "step": 11064
    },
    {
      "epoch": 7.079334612923865,
      "grad_norm": 0.8431342832826166,
      "learning_rate": 4.150178624971797e-06,
      "loss": 0.0295,
      "step": 11065
    },
    {
      "epoch": 7.079974408189379,
      "grad_norm": 0.6245486099035922,
      "learning_rate": 4.148498138011334e-06,
      "loss": 0.0221,
      "step": 11066
    },
    {
      "epoch": 7.0806142034548945,
      "grad_norm": 0.6706466304121582,
      "learning_rate": 4.146817902303953e-06,
      "loss": 0.0231,
      "step": 11067
    },
    {
      "epoch": 7.08125399872041,
      "grad_norm": 0.6396034043787032,
      "learning_rate": 4.1451379179218e-06,
      "loss": 0.0212,
      "step": 11068
    },
    {
      "epoch": 7.081893793985924,
      "grad_norm": 0.7387120358596054,
      "learning_rate": 4.143458184937006e-06,
      "loss": 0.0284,
      "step": 11069
    },
    {
      "epoch": 7.082533589251439,
      "grad_norm": 0.645578401377111,
      "learning_rate": 4.1417787034217015e-06,
      "loss": 0.0234,
      "step": 11070
    },
    {
      "epoch": 7.083173384516955,
      "grad_norm": 0.8062707061273132,
      "learning_rate": 4.1400994734480005e-06,
      "loss": 0.026,
      "step": 11071
    },
    {
      "epoch": 7.08381317978247,
      "grad_norm": 0.648475103118255,
      "learning_rate": 4.138420495088004e-06,
      "loss": 0.0193,
      "step": 11072
    },
    {
      "epoch": 7.084452975047984,
      "grad_norm": 0.6344656411206526,
      "learning_rate": 4.136741768413799e-06,
      "loss": 0.0246,
      "step": 11073
    },
    {
      "epoch": 7.0850927703135,
      "grad_norm": 0.7099837154947921,
      "learning_rate": 4.135063293497478e-06,
      "loss": 0.0207,
      "step": 11074
    },
    {
      "epoch": 7.085732565579015,
      "grad_norm": 0.7534174832437847,
      "learning_rate": 4.133385070411105e-06,
      "loss": 0.0277,
      "step": 11075
    },
    {
      "epoch": 7.08637236084453,
      "grad_norm": 0.6988571799280602,
      "learning_rate": 4.131707099226738e-06,
      "loss": 0.0262,
      "step": 11076
    },
    {
      "epoch": 7.0870121561100445,
      "grad_norm": 0.6664531664239387,
      "learning_rate": 4.130029380016434e-06,
      "loss": 0.0255,
      "step": 11077
    },
    {
      "epoch": 7.08765195137556,
      "grad_norm": 0.6761756103765677,
      "learning_rate": 4.1283519128522255e-06,
      "loss": 0.0279,
      "step": 11078
    },
    {
      "epoch": 7.088291746641075,
      "grad_norm": 0.8904053746717667,
      "learning_rate": 4.12667469780614e-06,
      "loss": 0.0229,
      "step": 11079
    },
    {
      "epoch": 7.0889315419065895,
      "grad_norm": 0.6200062221416123,
      "learning_rate": 4.124997734950197e-06,
      "loss": 0.0239,
      "step": 11080
    },
    {
      "epoch": 7.089571337172105,
      "grad_norm": 0.5744129430587193,
      "learning_rate": 4.123321024356399e-06,
      "loss": 0.0217,
      "step": 11081
    },
    {
      "epoch": 7.09021113243762,
      "grad_norm": 0.7026058027292205,
      "learning_rate": 4.1216445660967444e-06,
      "loss": 0.0243,
      "step": 11082
    },
    {
      "epoch": 7.090850927703135,
      "grad_norm": 0.6336171444668803,
      "learning_rate": 4.119968360243209e-06,
      "loss": 0.024,
      "step": 11083
    },
    {
      "epoch": 7.09149072296865,
      "grad_norm": 0.6584194022588845,
      "learning_rate": 4.1182924068677774e-06,
      "loss": 0.0207,
      "step": 11084
    },
    {
      "epoch": 7.092130518234165,
      "grad_norm": 0.7933983349422299,
      "learning_rate": 4.1166167060424075e-06,
      "loss": 0.0285,
      "step": 11085
    },
    {
      "epoch": 7.09277031349968,
      "grad_norm": 0.7044325311669835,
      "learning_rate": 4.1149412578390465e-06,
      "loss": 0.0266,
      "step": 11086
    },
    {
      "epoch": 7.0934101087651955,
      "grad_norm": 0.7018537975939358,
      "learning_rate": 4.113266062329644e-06,
      "loss": 0.0227,
      "step": 11087
    },
    {
      "epoch": 7.09404990403071,
      "grad_norm": 0.7087865983390466,
      "learning_rate": 4.111591119586123e-06,
      "loss": 0.026,
      "step": 11088
    },
    {
      "epoch": 7.094689699296225,
      "grad_norm": 0.6570387652739308,
      "learning_rate": 4.109916429680406e-06,
      "loss": 0.0243,
      "step": 11089
    },
    {
      "epoch": 7.0953294945617404,
      "grad_norm": 0.7160452087905949,
      "learning_rate": 4.1082419926843965e-06,
      "loss": 0.0254,
      "step": 11090
    },
    {
      "epoch": 7.095969289827256,
      "grad_norm": 0.6679825334105343,
      "learning_rate": 4.106567808669998e-06,
      "loss": 0.0228,
      "step": 11091
    },
    {
      "epoch": 7.09660908509277,
      "grad_norm": 0.6420525216487424,
      "learning_rate": 4.1048938777090945e-06,
      "loss": 0.0236,
      "step": 11092
    },
    {
      "epoch": 7.097248880358285,
      "grad_norm": 0.841405952471294,
      "learning_rate": 4.103220199873558e-06,
      "loss": 0.0225,
      "step": 11093
    },
    {
      "epoch": 7.097888675623801,
      "grad_norm": 0.8050755882128652,
      "learning_rate": 4.101546775235259e-06,
      "loss": 0.0222,
      "step": 11094
    },
    {
      "epoch": 7.098528470889315,
      "grad_norm": 0.7787184253162299,
      "learning_rate": 4.09987360386605e-06,
      "loss": 0.0256,
      "step": 11095
    },
    {
      "epoch": 7.09916826615483,
      "grad_norm": 0.6914319087489925,
      "learning_rate": 4.098200685837772e-06,
      "loss": 0.0265,
      "step": 11096
    },
    {
      "epoch": 7.099808061420346,
      "grad_norm": 0.7041064169153808,
      "learning_rate": 4.096528021222259e-06,
      "loss": 0.0248,
      "step": 11097
    },
    {
      "epoch": 7.100447856685861,
      "grad_norm": 0.7445176467344984,
      "learning_rate": 4.094855610091331e-06,
      "loss": 0.0256,
      "step": 11098
    },
    {
      "epoch": 7.101087651951375,
      "grad_norm": 0.6938219219881601,
      "learning_rate": 4.093183452516799e-06,
      "loss": 0.0278,
      "step": 11099
    },
    {
      "epoch": 7.1017274472168905,
      "grad_norm": 0.5958604566998423,
      "learning_rate": 4.0915115485704586e-06,
      "loss": 0.0204,
      "step": 11100
    },
    {
      "epoch": 7.102367242482406,
      "grad_norm": 0.5691076464126003,
      "learning_rate": 4.089839898324105e-06,
      "loss": 0.0234,
      "step": 11101
    },
    {
      "epoch": 7.103007037747921,
      "grad_norm": 0.7205475302532164,
      "learning_rate": 4.088168501849513e-06,
      "loss": 0.0238,
      "step": 11102
    },
    {
      "epoch": 7.1036468330134355,
      "grad_norm": 0.694979179423209,
      "learning_rate": 4.086497359218446e-06,
      "loss": 0.0242,
      "step": 11103
    },
    {
      "epoch": 7.104286628278951,
      "grad_norm": 0.7579731664170223,
      "learning_rate": 4.084826470502665e-06,
      "loss": 0.0229,
      "step": 11104
    },
    {
      "epoch": 7.104926423544466,
      "grad_norm": 0.6731351722088033,
      "learning_rate": 4.0831558357739155e-06,
      "loss": 0.0256,
      "step": 11105
    },
    {
      "epoch": 7.10556621880998,
      "grad_norm": 0.6780507780823137,
      "learning_rate": 4.081485455103924e-06,
      "loss": 0.0259,
      "step": 11106
    },
    {
      "epoch": 7.106206014075496,
      "grad_norm": 0.6408807547018311,
      "learning_rate": 4.079815328564423e-06,
      "loss": 0.0244,
      "step": 11107
    },
    {
      "epoch": 7.106845809341011,
      "grad_norm": 0.739650574013109,
      "learning_rate": 4.0781454562271206e-06,
      "loss": 0.0256,
      "step": 11108
    },
    {
      "epoch": 7.107485604606526,
      "grad_norm": 0.7167176912289223,
      "learning_rate": 4.076475838163717e-06,
      "loss": 0.0248,
      "step": 11109
    },
    {
      "epoch": 7.108125399872041,
      "grad_norm": 0.6415225155990353,
      "learning_rate": 4.074806474445905e-06,
      "loss": 0.0244,
      "step": 11110
    },
    {
      "epoch": 7.108765195137556,
      "grad_norm": 0.6611258866914348,
      "learning_rate": 4.0731373651453635e-06,
      "loss": 0.0217,
      "step": 11111
    },
    {
      "epoch": 7.109404990403071,
      "grad_norm": 0.6707949769186141,
      "learning_rate": 4.071468510333759e-06,
      "loss": 0.0212,
      "step": 11112
    },
    {
      "epoch": 7.110044785668586,
      "grad_norm": 0.6745921408610424,
      "learning_rate": 4.069799910082748e-06,
      "loss": 0.0207,
      "step": 11113
    },
    {
      "epoch": 7.110684580934101,
      "grad_norm": 0.7135324654671189,
      "learning_rate": 4.068131564463983e-06,
      "loss": 0.0235,
      "step": 11114
    },
    {
      "epoch": 7.111324376199616,
      "grad_norm": 0.5851134642699212,
      "learning_rate": 4.066463473549096e-06,
      "loss": 0.0227,
      "step": 11115
    },
    {
      "epoch": 7.111964171465131,
      "grad_norm": 0.846763293594921,
      "learning_rate": 4.06479563740971e-06,
      "loss": 0.0272,
      "step": 11116
    },
    {
      "epoch": 7.112603966730646,
      "grad_norm": 0.7245034799222108,
      "learning_rate": 4.063128056117445e-06,
      "loss": 0.0247,
      "step": 11117
    },
    {
      "epoch": 7.113243761996161,
      "grad_norm": 0.6271745416006668,
      "learning_rate": 4.0614607297439e-06,
      "loss": 0.0226,
      "step": 11118
    },
    {
      "epoch": 7.113883557261676,
      "grad_norm": 0.6711231289162105,
      "learning_rate": 4.059793658360667e-06,
      "loss": 0.0238,
      "step": 11119
    },
    {
      "epoch": 7.114523352527192,
      "grad_norm": 0.6616380116986418,
      "learning_rate": 4.058126842039325e-06,
      "loss": 0.0249,
      "step": 11120
    },
    {
      "epoch": 7.115163147792706,
      "grad_norm": 0.577962814247184,
      "learning_rate": 4.056460280851449e-06,
      "loss": 0.0205,
      "step": 11121
    },
    {
      "epoch": 7.115802943058221,
      "grad_norm": 0.6898732838247223,
      "learning_rate": 4.054793974868597e-06,
      "loss": 0.022,
      "step": 11122
    },
    {
      "epoch": 7.1164427383237365,
      "grad_norm": 0.8278541817150719,
      "learning_rate": 4.053127924162311e-06,
      "loss": 0.025,
      "step": 11123
    },
    {
      "epoch": 7.117082533589252,
      "grad_norm": 0.7944699632306023,
      "learning_rate": 4.0514621288041374e-06,
      "loss": 0.0285,
      "step": 11124
    },
    {
      "epoch": 7.117722328854766,
      "grad_norm": 0.6307489935482992,
      "learning_rate": 4.0497965888655996e-06,
      "loss": 0.02,
      "step": 11125
    },
    {
      "epoch": 7.1183621241202815,
      "grad_norm": 0.6670267445155386,
      "learning_rate": 4.04813130441821e-06,
      "loss": 0.0232,
      "step": 11126
    },
    {
      "epoch": 7.119001919385797,
      "grad_norm": 0.7590395700314095,
      "learning_rate": 4.046466275533476e-06,
      "loss": 0.0328,
      "step": 11127
    },
    {
      "epoch": 7.119641714651312,
      "grad_norm": 0.6728012500518362,
      "learning_rate": 4.044801502282889e-06,
      "loss": 0.0233,
      "step": 11128
    },
    {
      "epoch": 7.120281509916826,
      "grad_norm": 0.6916103118379057,
      "learning_rate": 4.043136984737932e-06,
      "loss": 0.0248,
      "step": 11129
    },
    {
      "epoch": 7.120921305182342,
      "grad_norm": 0.5781244372654106,
      "learning_rate": 4.041472722970072e-06,
      "loss": 0.0203,
      "step": 11130
    },
    {
      "epoch": 7.121561100447857,
      "grad_norm": 0.6303794752311679,
      "learning_rate": 4.03980871705078e-06,
      "loss": 0.0263,
      "step": 11131
    },
    {
      "epoch": 7.122200895713371,
      "grad_norm": 0.7014830461210548,
      "learning_rate": 4.038144967051498e-06,
      "loss": 0.0249,
      "step": 11132
    },
    {
      "epoch": 7.122840690978887,
      "grad_norm": 0.7206543739888678,
      "learning_rate": 4.0364814730436606e-06,
      "loss": 0.023,
      "step": 11133
    },
    {
      "epoch": 7.123480486244402,
      "grad_norm": 0.6958162340268962,
      "learning_rate": 4.034818235098706e-06,
      "loss": 0.0235,
      "step": 11134
    },
    {
      "epoch": 7.124120281509917,
      "grad_norm": 0.7111624829276536,
      "learning_rate": 4.033155253288044e-06,
      "loss": 0.0239,
      "step": 11135
    },
    {
      "epoch": 7.1247600767754315,
      "grad_norm": 0.6199552941503883,
      "learning_rate": 4.031492527683079e-06,
      "loss": 0.0208,
      "step": 11136
    },
    {
      "epoch": 7.125399872040947,
      "grad_norm": 0.6411764386535129,
      "learning_rate": 4.0298300583552095e-06,
      "loss": 0.0238,
      "step": 11137
    },
    {
      "epoch": 7.126039667306462,
      "grad_norm": 0.6791070664730231,
      "learning_rate": 4.0281678453758175e-06,
      "loss": 0.0219,
      "step": 11138
    },
    {
      "epoch": 7.126679462571977,
      "grad_norm": 0.8164324393914193,
      "learning_rate": 4.026505888816277e-06,
      "loss": 0.0244,
      "step": 11139
    },
    {
      "epoch": 7.127319257837492,
      "grad_norm": 0.6669433195220986,
      "learning_rate": 4.024844188747946e-06,
      "loss": 0.0223,
      "step": 11140
    },
    {
      "epoch": 7.127959053103007,
      "grad_norm": 0.8334357366899886,
      "learning_rate": 4.023182745242177e-06,
      "loss": 0.0245,
      "step": 11141
    },
    {
      "epoch": 7.128598848368522,
      "grad_norm": 0.6923043552802384,
      "learning_rate": 4.021521558370308e-06,
      "loss": 0.025,
      "step": 11142
    },
    {
      "epoch": 7.129238643634037,
      "grad_norm": 0.7617487309147373,
      "learning_rate": 4.019860628203666e-06,
      "loss": 0.0244,
      "step": 11143
    },
    {
      "epoch": 7.129878438899552,
      "grad_norm": 0.8326367117878244,
      "learning_rate": 4.018199954813574e-06,
      "loss": 0.024,
      "step": 11144
    },
    {
      "epoch": 7.130518234165067,
      "grad_norm": 0.5507528896880514,
      "learning_rate": 4.016539538271336e-06,
      "loss": 0.0204,
      "step": 11145
    },
    {
      "epoch": 7.1311580294305825,
      "grad_norm": 0.7094473237858474,
      "learning_rate": 4.014879378648242e-06,
      "loss": 0.0245,
      "step": 11146
    },
    {
      "epoch": 7.131797824696097,
      "grad_norm": 0.535563596846639,
      "learning_rate": 4.0132194760155844e-06,
      "loss": 0.0188,
      "step": 11147
    },
    {
      "epoch": 7.132437619961612,
      "grad_norm": 0.6822733828127644,
      "learning_rate": 4.011559830444632e-06,
      "loss": 0.0258,
      "step": 11148
    },
    {
      "epoch": 7.1330774152271275,
      "grad_norm": 0.6130910255459177,
      "learning_rate": 4.009900442006649e-06,
      "loss": 0.0234,
      "step": 11149
    },
    {
      "epoch": 7.133717210492643,
      "grad_norm": 0.6443354370890582,
      "learning_rate": 4.008241310772882e-06,
      "loss": 0.022,
      "step": 11150
    },
    {
      "epoch": 7.134357005758157,
      "grad_norm": 0.7435381497744218,
      "learning_rate": 4.006582436814578e-06,
      "loss": 0.0226,
      "step": 11151
    },
    {
      "epoch": 7.134996801023672,
      "grad_norm": 0.704028114035078,
      "learning_rate": 4.004923820202963e-06,
      "loss": 0.0227,
      "step": 11152
    },
    {
      "epoch": 7.135636596289188,
      "grad_norm": 0.7504422850136113,
      "learning_rate": 4.003265461009251e-06,
      "loss": 0.0291,
      "step": 11153
    },
    {
      "epoch": 7.136276391554702,
      "grad_norm": 0.7368842705523726,
      "learning_rate": 4.001607359304656e-06,
      "loss": 0.0232,
      "step": 11154
    },
    {
      "epoch": 7.136916186820217,
      "grad_norm": 0.6358034603243482,
      "learning_rate": 3.999949515160369e-06,
      "loss": 0.0215,
      "step": 11155
    },
    {
      "epoch": 7.137555982085733,
      "grad_norm": 0.6697056727489169,
      "learning_rate": 3.998291928647579e-06,
      "loss": 0.0213,
      "step": 11156
    },
    {
      "epoch": 7.138195777351248,
      "grad_norm": 0.6294489419008522,
      "learning_rate": 3.996634599837457e-06,
      "loss": 0.0247,
      "step": 11157
    },
    {
      "epoch": 7.138835572616762,
      "grad_norm": 0.6627792373596418,
      "learning_rate": 3.994977528801166e-06,
      "loss": 0.0227,
      "step": 11158
    },
    {
      "epoch": 7.1394753678822775,
      "grad_norm": 0.6730579953967364,
      "learning_rate": 3.993320715609857e-06,
      "loss": 0.0211,
      "step": 11159
    },
    {
      "epoch": 7.140115163147793,
      "grad_norm": 0.6018910883199617,
      "learning_rate": 3.9916641603346685e-06,
      "loss": 0.0218,
      "step": 11160
    },
    {
      "epoch": 7.140754958413308,
      "grad_norm": 0.6794607231081842,
      "learning_rate": 3.990007863046737e-06,
      "loss": 0.0221,
      "step": 11161
    },
    {
      "epoch": 7.1413947536788225,
      "grad_norm": 0.7001267940489436,
      "learning_rate": 3.988351823817176e-06,
      "loss": 0.0244,
      "step": 11162
    },
    {
      "epoch": 7.142034548944338,
      "grad_norm": 0.6804868752126972,
      "learning_rate": 3.986696042717091e-06,
      "loss": 0.0244,
      "step": 11163
    },
    {
      "epoch": 7.142674344209853,
      "grad_norm": 0.6739722935096157,
      "learning_rate": 3.985040519817584e-06,
      "loss": 0.0262,
      "step": 11164
    },
    {
      "epoch": 7.143314139475368,
      "grad_norm": 0.6695624580539069,
      "learning_rate": 3.983385255189738e-06,
      "loss": 0.0277,
      "step": 11165
    },
    {
      "epoch": 7.143953934740883,
      "grad_norm": 0.6066485670424935,
      "learning_rate": 3.981730248904625e-06,
      "loss": 0.0247,
      "step": 11166
    },
    {
      "epoch": 7.144593730006398,
      "grad_norm": 0.537523351344593,
      "learning_rate": 3.980075501033307e-06,
      "loss": 0.0199,
      "step": 11167
    },
    {
      "epoch": 7.145233525271913,
      "grad_norm": 0.6436522352745927,
      "learning_rate": 3.978421011646843e-06,
      "loss": 0.0207,
      "step": 11168
    },
    {
      "epoch": 7.145873320537428,
      "grad_norm": 0.7968662342215812,
      "learning_rate": 3.976766780816267e-06,
      "loss": 0.0271,
      "step": 11169
    },
    {
      "epoch": 7.146513115802943,
      "grad_norm": 0.6773442420545965,
      "learning_rate": 3.9751128086126125e-06,
      "loss": 0.0225,
      "step": 11170
    },
    {
      "epoch": 7.147152911068458,
      "grad_norm": 0.6796248052259694,
      "learning_rate": 3.973459095106895e-06,
      "loss": 0.0256,
      "step": 11171
    },
    {
      "epoch": 7.1477927063339735,
      "grad_norm": 0.6701913623780427,
      "learning_rate": 3.971805640370125e-06,
      "loss": 0.021,
      "step": 11172
    },
    {
      "epoch": 7.148432501599488,
      "grad_norm": 0.6814851980558471,
      "learning_rate": 3.970152444473293e-06,
      "loss": 0.0204,
      "step": 11173
    },
    {
      "epoch": 7.149072296865003,
      "grad_norm": 0.6813029591287119,
      "learning_rate": 3.968499507487393e-06,
      "loss": 0.0245,
      "step": 11174
    },
    {
      "epoch": 7.149712092130518,
      "grad_norm": 0.6307116420239702,
      "learning_rate": 3.966846829483395e-06,
      "loss": 0.0212,
      "step": 11175
    },
    {
      "epoch": 7.150351887396034,
      "grad_norm": 0.6626967410768763,
      "learning_rate": 3.9651944105322615e-06,
      "loss": 0.02,
      "step": 11176
    },
    {
      "epoch": 7.150991682661548,
      "grad_norm": 0.7217914218671472,
      "learning_rate": 3.9635422507049415e-06,
      "loss": 0.0223,
      "step": 11177
    },
    {
      "epoch": 7.151631477927063,
      "grad_norm": 0.7955585038095901,
      "learning_rate": 3.961890350072382e-06,
      "loss": 0.0302,
      "step": 11178
    },
    {
      "epoch": 7.152271273192579,
      "grad_norm": 0.7354400677972566,
      "learning_rate": 3.9602387087055116e-06,
      "loss": 0.0241,
      "step": 11179
    },
    {
      "epoch": 7.152911068458093,
      "grad_norm": 0.6694475219589618,
      "learning_rate": 3.9585873266752425e-06,
      "loss": 0.023,
      "step": 11180
    },
    {
      "epoch": 7.153550863723608,
      "grad_norm": 0.6613187653910417,
      "learning_rate": 3.956936204052491e-06,
      "loss": 0.0229,
      "step": 11181
    },
    {
      "epoch": 7.1541906589891235,
      "grad_norm": 0.5442341050811357,
      "learning_rate": 3.95528534090815e-06,
      "loss": 0.0176,
      "step": 11182
    },
    {
      "epoch": 7.154830454254639,
      "grad_norm": 0.6120603865745452,
      "learning_rate": 3.953634737313098e-06,
      "loss": 0.0212,
      "step": 11183
    },
    {
      "epoch": 7.155470249520153,
      "grad_norm": 0.6867008505364143,
      "learning_rate": 3.951984393338219e-06,
      "loss": 0.0235,
      "step": 11184
    },
    {
      "epoch": 7.1561100447856685,
      "grad_norm": 0.7442191538660091,
      "learning_rate": 3.950334309054372e-06,
      "loss": 0.0296,
      "step": 11185
    },
    {
      "epoch": 7.156749840051184,
      "grad_norm": 0.7951422673324637,
      "learning_rate": 3.948684484532408e-06,
      "loss": 0.0274,
      "step": 11186
    },
    {
      "epoch": 7.157389635316699,
      "grad_norm": 0.6969237320606528,
      "learning_rate": 3.947034919843168e-06,
      "loss": 0.0221,
      "step": 11187
    },
    {
      "epoch": 7.158029430582213,
      "grad_norm": 0.5952816148128349,
      "learning_rate": 3.94538561505748e-06,
      "loss": 0.0205,
      "step": 11188
    },
    {
      "epoch": 7.158669225847729,
      "grad_norm": 0.684399738392253,
      "learning_rate": 3.943736570246165e-06,
      "loss": 0.0272,
      "step": 11189
    },
    {
      "epoch": 7.159309021113244,
      "grad_norm": 0.6772109451233225,
      "learning_rate": 3.9420877854800246e-06,
      "loss": 0.0247,
      "step": 11190
    },
    {
      "epoch": 7.159948816378758,
      "grad_norm": 0.6346166244413475,
      "learning_rate": 3.94043926082986e-06,
      "loss": 0.0225,
      "step": 11191
    },
    {
      "epoch": 7.160588611644274,
      "grad_norm": 0.716871652942588,
      "learning_rate": 3.938790996366457e-06,
      "loss": 0.02,
      "step": 11192
    },
    {
      "epoch": 7.161228406909789,
      "grad_norm": 0.7748104233893605,
      "learning_rate": 3.937142992160582e-06,
      "loss": 0.0308,
      "step": 11193
    },
    {
      "epoch": 7.161868202175304,
      "grad_norm": 0.7505037759840274,
      "learning_rate": 3.935495248283006e-06,
      "loss": 0.0254,
      "step": 11194
    },
    {
      "epoch": 7.162507997440819,
      "grad_norm": 0.6005549136182649,
      "learning_rate": 3.933847764804475e-06,
      "loss": 0.0215,
      "step": 11195
    },
    {
      "epoch": 7.163147792706334,
      "grad_norm": 0.6289081473138196,
      "learning_rate": 3.932200541795731e-06,
      "loss": 0.0192,
      "step": 11196
    },
    {
      "epoch": 7.163787587971849,
      "grad_norm": 0.7096426570144585,
      "learning_rate": 3.930553579327499e-06,
      "loss": 0.024,
      "step": 11197
    },
    {
      "epoch": 7.164427383237364,
      "grad_norm": 0.6344550135078252,
      "learning_rate": 3.928906877470501e-06,
      "loss": 0.0199,
      "step": 11198
    },
    {
      "epoch": 7.165067178502879,
      "grad_norm": 0.7331850335558454,
      "learning_rate": 3.927260436295444e-06,
      "loss": 0.0233,
      "step": 11199
    },
    {
      "epoch": 7.165706973768394,
      "grad_norm": 0.7164774204834303,
      "learning_rate": 3.925614255873021e-06,
      "loss": 0.0258,
      "step": 11200
    },
    {
      "epoch": 7.166346769033909,
      "grad_norm": 0.5848926616251658,
      "learning_rate": 3.923968336273916e-06,
      "loss": 0.0247,
      "step": 11201
    },
    {
      "epoch": 7.166986564299425,
      "grad_norm": 0.7243008049314175,
      "learning_rate": 3.922322677568802e-06,
      "loss": 0.023,
      "step": 11202
    },
    {
      "epoch": 7.167626359564939,
      "grad_norm": 0.7392120725615314,
      "learning_rate": 3.920677279828336e-06,
      "loss": 0.0253,
      "step": 11203
    },
    {
      "epoch": 7.168266154830454,
      "grad_norm": 0.6689731305354949,
      "learning_rate": 3.919032143123179e-06,
      "loss": 0.0221,
      "step": 11204
    },
    {
      "epoch": 7.1689059500959695,
      "grad_norm": 0.7511941628851895,
      "learning_rate": 3.917387267523964e-06,
      "loss": 0.0281,
      "step": 11205
    },
    {
      "epoch": 7.169545745361484,
      "grad_norm": 0.6791898807300477,
      "learning_rate": 3.91574265310132e-06,
      "loss": 0.0239,
      "step": 11206
    },
    {
      "epoch": 7.170185540626999,
      "grad_norm": 0.7128609548468715,
      "learning_rate": 3.914098299925859e-06,
      "loss": 0.0247,
      "step": 11207
    },
    {
      "epoch": 7.1708253358925145,
      "grad_norm": 0.6714954946614005,
      "learning_rate": 3.912454208068196e-06,
      "loss": 0.0238,
      "step": 11208
    },
    {
      "epoch": 7.17146513115803,
      "grad_norm": 0.7369806144986595,
      "learning_rate": 3.910810377598919e-06,
      "loss": 0.0248,
      "step": 11209
    },
    {
      "epoch": 7.172104926423544,
      "grad_norm": 0.6781129675435449,
      "learning_rate": 3.90916680858861e-06,
      "loss": 0.0221,
      "step": 11210
    },
    {
      "epoch": 7.172744721689059,
      "grad_norm": 0.6923453284718555,
      "learning_rate": 3.907523501107846e-06,
      "loss": 0.0238,
      "step": 11211
    },
    {
      "epoch": 7.173384516954575,
      "grad_norm": 0.6866898891074823,
      "learning_rate": 3.905880455227185e-06,
      "loss": 0.0239,
      "step": 11212
    },
    {
      "epoch": 7.17402431222009,
      "grad_norm": 0.6706471134951287,
      "learning_rate": 3.904237671017175e-06,
      "loss": 0.0227,
      "step": 11213
    },
    {
      "epoch": 7.174664107485604,
      "grad_norm": 0.7330951175869547,
      "learning_rate": 3.9025951485483584e-06,
      "loss": 0.0219,
      "step": 11214
    },
    {
      "epoch": 7.17530390275112,
      "grad_norm": 0.6727887566818973,
      "learning_rate": 3.900952887891259e-06,
      "loss": 0.023,
      "step": 11215
    },
    {
      "epoch": 7.175943698016635,
      "grad_norm": 0.6866884111153139,
      "learning_rate": 3.8993108891163934e-06,
      "loss": 0.0251,
      "step": 11216
    },
    {
      "epoch": 7.176583493282149,
      "grad_norm": 0.6965480837230685,
      "learning_rate": 3.897669152294267e-06,
      "loss": 0.0239,
      "step": 11217
    },
    {
      "epoch": 7.177223288547665,
      "grad_norm": 0.6875297828272988,
      "learning_rate": 3.896027677495371e-06,
      "loss": 0.0255,
      "step": 11218
    },
    {
      "epoch": 7.17786308381318,
      "grad_norm": 0.7191699776097933,
      "learning_rate": 3.894386464790189e-06,
      "loss": 0.0254,
      "step": 11219
    },
    {
      "epoch": 7.178502879078695,
      "grad_norm": 0.7483559961224936,
      "learning_rate": 3.892745514249187e-06,
      "loss": 0.0269,
      "step": 11220
    },
    {
      "epoch": 7.1791426743442095,
      "grad_norm": 0.7092830774058118,
      "learning_rate": 3.891104825942832e-06,
      "loss": 0.023,
      "step": 11221
    },
    {
      "epoch": 7.179782469609725,
      "grad_norm": 0.625606117508176,
      "learning_rate": 3.889464399941569e-06,
      "loss": 0.0202,
      "step": 11222
    },
    {
      "epoch": 7.18042226487524,
      "grad_norm": 0.6827922569687599,
      "learning_rate": 3.887824236315833e-06,
      "loss": 0.0231,
      "step": 11223
    },
    {
      "epoch": 7.181062060140755,
      "grad_norm": 0.7089977402728262,
      "learning_rate": 3.886184335136053e-06,
      "loss": 0.0278,
      "step": 11224
    },
    {
      "epoch": 7.18170185540627,
      "grad_norm": 0.5700594842874035,
      "learning_rate": 3.884544696472643e-06,
      "loss": 0.0208,
      "step": 11225
    },
    {
      "epoch": 7.182341650671785,
      "grad_norm": 0.6356274155754332,
      "learning_rate": 3.882905320396005e-06,
      "loss": 0.0228,
      "step": 11226
    },
    {
      "epoch": 7.1829814459373,
      "grad_norm": 0.6990058118465823,
      "learning_rate": 3.881266206976528e-06,
      "loss": 0.0237,
      "step": 11227
    },
    {
      "epoch": 7.183621241202815,
      "grad_norm": 0.6173243759287754,
      "learning_rate": 3.879627356284599e-06,
      "loss": 0.0219,
      "step": 11228
    },
    {
      "epoch": 7.18426103646833,
      "grad_norm": 0.7687378730253221,
      "learning_rate": 3.877988768390584e-06,
      "loss": 0.0268,
      "step": 11229
    },
    {
      "epoch": 7.184900831733845,
      "grad_norm": 0.6185994421594873,
      "learning_rate": 3.8763504433648406e-06,
      "loss": 0.0219,
      "step": 11230
    },
    {
      "epoch": 7.1855406269993605,
      "grad_norm": 0.6689543462434304,
      "learning_rate": 3.874712381277716e-06,
      "loss": 0.0267,
      "step": 11231
    },
    {
      "epoch": 7.186180422264875,
      "grad_norm": 0.6807663835623222,
      "learning_rate": 3.873074582199546e-06,
      "loss": 0.0273,
      "step": 11232
    },
    {
      "epoch": 7.18682021753039,
      "grad_norm": 0.5673897687440699,
      "learning_rate": 3.871437046200654e-06,
      "loss": 0.0209,
      "step": 11233
    },
    {
      "epoch": 7.187460012795905,
      "grad_norm": 0.7815225956388264,
      "learning_rate": 3.869799773351351e-06,
      "loss": 0.0263,
      "step": 11234
    },
    {
      "epoch": 7.188099808061421,
      "grad_norm": 0.7025150346745954,
      "learning_rate": 3.868162763721943e-06,
      "loss": 0.028,
      "step": 11235
    },
    {
      "epoch": 7.188739603326935,
      "grad_norm": 0.754123459838148,
      "learning_rate": 3.86652601738272e-06,
      "loss": 0.0327,
      "step": 11236
    },
    {
      "epoch": 7.18937939859245,
      "grad_norm": 0.7698726562468464,
      "learning_rate": 3.864889534403955e-06,
      "loss": 0.0307,
      "step": 11237
    },
    {
      "epoch": 7.190019193857966,
      "grad_norm": 0.5879457624646327,
      "learning_rate": 3.863253314855922e-06,
      "loss": 0.0225,
      "step": 11238
    },
    {
      "epoch": 7.190658989123481,
      "grad_norm": 0.6098275914917326,
      "learning_rate": 3.861617358808877e-06,
      "loss": 0.0208,
      "step": 11239
    },
    {
      "epoch": 7.191298784388995,
      "grad_norm": 0.6664020050839745,
      "learning_rate": 3.859981666333059e-06,
      "loss": 0.0265,
      "step": 11240
    },
    {
      "epoch": 7.1919385796545106,
      "grad_norm": 0.58925767534336,
      "learning_rate": 3.85834623749871e-06,
      "loss": 0.0212,
      "step": 11241
    },
    {
      "epoch": 7.192578374920026,
      "grad_norm": 0.7331800456193936,
      "learning_rate": 3.856711072376048e-06,
      "loss": 0.0269,
      "step": 11242
    },
    {
      "epoch": 7.19321817018554,
      "grad_norm": 0.6311837490028687,
      "learning_rate": 3.855076171035285e-06,
      "loss": 0.0216,
      "step": 11243
    },
    {
      "epoch": 7.1938579654510555,
      "grad_norm": 0.70649124781121,
      "learning_rate": 3.853441533546616e-06,
      "loss": 0.0226,
      "step": 11244
    },
    {
      "epoch": 7.194497760716571,
      "grad_norm": 0.8379746849681977,
      "learning_rate": 3.851807159980238e-06,
      "loss": 0.0291,
      "step": 11245
    },
    {
      "epoch": 7.195137555982086,
      "grad_norm": 0.6296293534955069,
      "learning_rate": 3.850173050406323e-06,
      "loss": 0.0238,
      "step": 11246
    },
    {
      "epoch": 7.1957773512476,
      "grad_norm": 0.5928850313875894,
      "learning_rate": 3.848539204895038e-06,
      "loss": 0.0186,
      "step": 11247
    },
    {
      "epoch": 7.196417146513116,
      "grad_norm": 0.6736651237378098,
      "learning_rate": 3.846905623516536e-06,
      "loss": 0.0227,
      "step": 11248
    },
    {
      "epoch": 7.197056941778631,
      "grad_norm": 0.6580202233970749,
      "learning_rate": 3.845272306340962e-06,
      "loss": 0.0255,
      "step": 11249
    },
    {
      "epoch": 7.197696737044146,
      "grad_norm": 0.6700549351908547,
      "learning_rate": 3.843639253438443e-06,
      "loss": 0.0234,
      "step": 11250
    },
    {
      "epoch": 7.198336532309661,
      "grad_norm": 0.8531985620388834,
      "learning_rate": 3.842006464879107e-06,
      "loss": 0.0239,
      "step": 11251
    },
    {
      "epoch": 7.198976327575176,
      "grad_norm": 0.6397595184000252,
      "learning_rate": 3.8403739407330575e-06,
      "loss": 0.0238,
      "step": 11252
    },
    {
      "epoch": 7.199616122840691,
      "grad_norm": 0.6361336014862238,
      "learning_rate": 3.838741681070396e-06,
      "loss": 0.0225,
      "step": 11253
    },
    {
      "epoch": 7.2002559181062065,
      "grad_norm": 0.717521963283427,
      "learning_rate": 3.837109685961202e-06,
      "loss": 0.0229,
      "step": 11254
    },
    {
      "epoch": 7.200895713371721,
      "grad_norm": 0.8007423746082568,
      "learning_rate": 3.83547795547556e-06,
      "loss": 0.0278,
      "step": 11255
    },
    {
      "epoch": 7.201535508637236,
      "grad_norm": 0.6533023761823615,
      "learning_rate": 3.833846489683527e-06,
      "loss": 0.0244,
      "step": 11256
    },
    {
      "epoch": 7.202175303902751,
      "grad_norm": 0.7435920251583529,
      "learning_rate": 3.8322152886551545e-06,
      "loss": 0.0282,
      "step": 11257
    },
    {
      "epoch": 7.202815099168266,
      "grad_norm": 0.5878375286048109,
      "learning_rate": 3.830584352460489e-06,
      "loss": 0.0205,
      "step": 11258
    },
    {
      "epoch": 7.203454894433781,
      "grad_norm": 0.6407774214935545,
      "learning_rate": 3.8289536811695575e-06,
      "loss": 0.0206,
      "step": 11259
    },
    {
      "epoch": 7.204094689699296,
      "grad_norm": 0.6597693516151164,
      "learning_rate": 3.827323274852376e-06,
      "loss": 0.025,
      "step": 11260
    },
    {
      "epoch": 7.204734484964812,
      "grad_norm": 0.7459196356932601,
      "learning_rate": 3.825693133578954e-06,
      "loss": 0.0282,
      "step": 11261
    },
    {
      "epoch": 7.205374280230326,
      "grad_norm": 0.675424061315642,
      "learning_rate": 3.824063257419285e-06,
      "loss": 0.0239,
      "step": 11262
    },
    {
      "epoch": 7.206014075495841,
      "grad_norm": 0.7889970684074836,
      "learning_rate": 3.822433646443354e-06,
      "loss": 0.0261,
      "step": 11263
    },
    {
      "epoch": 7.2066538707613566,
      "grad_norm": 0.6602115961243948,
      "learning_rate": 3.820804300721129e-06,
      "loss": 0.0227,
      "step": 11264
    },
    {
      "epoch": 7.207293666026872,
      "grad_norm": 0.6348448544782905,
      "learning_rate": 3.81917522032258e-06,
      "loss": 0.0242,
      "step": 11265
    },
    {
      "epoch": 7.207933461292386,
      "grad_norm": 0.6277874928940899,
      "learning_rate": 3.817546405317652e-06,
      "loss": 0.0217,
      "step": 11266
    },
    {
      "epoch": 7.2085732565579015,
      "grad_norm": 0.6970756137306476,
      "learning_rate": 3.81591785577628e-06,
      "loss": 0.0266,
      "step": 11267
    },
    {
      "epoch": 7.209213051823417,
      "grad_norm": 0.7640196277720899,
      "learning_rate": 3.814289571768397e-06,
      "loss": 0.0253,
      "step": 11268
    },
    {
      "epoch": 7.209852847088931,
      "grad_norm": 0.7881957566015377,
      "learning_rate": 3.8126615533639177e-06,
      "loss": 0.0275,
      "step": 11269
    },
    {
      "epoch": 7.210492642354446,
      "grad_norm": 0.6489785286204562,
      "learning_rate": 3.8110338006327406e-06,
      "loss": 0.0243,
      "step": 11270
    },
    {
      "epoch": 7.211132437619962,
      "grad_norm": 0.6311011314997406,
      "learning_rate": 3.809406313644767e-06,
      "loss": 0.0214,
      "step": 11271
    },
    {
      "epoch": 7.211772232885477,
      "grad_norm": 0.6821912270288397,
      "learning_rate": 3.807779092469873e-06,
      "loss": 0.0253,
      "step": 11272
    },
    {
      "epoch": 7.212412028150991,
      "grad_norm": 0.7215603601723632,
      "learning_rate": 3.8061521371779307e-06,
      "loss": 0.0238,
      "step": 11273
    },
    {
      "epoch": 7.213051823416507,
      "grad_norm": 0.7008204828383197,
      "learning_rate": 3.8045254478387937e-06,
      "loss": 0.0214,
      "step": 11274
    },
    {
      "epoch": 7.213691618682022,
      "grad_norm": 0.7732392825088382,
      "learning_rate": 3.8028990245223164e-06,
      "loss": 0.0288,
      "step": 11275
    },
    {
      "epoch": 7.214331413947537,
      "grad_norm": 0.7095931984619878,
      "learning_rate": 3.8012728672983312e-06,
      "loss": 0.0263,
      "step": 11276
    },
    {
      "epoch": 7.214971209213052,
      "grad_norm": 0.7536663124104304,
      "learning_rate": 3.799646976236663e-06,
      "loss": 0.0234,
      "step": 11277
    },
    {
      "epoch": 7.215611004478567,
      "grad_norm": 0.7116797442679068,
      "learning_rate": 3.798021351407124e-06,
      "loss": 0.0219,
      "step": 11278
    },
    {
      "epoch": 7.216250799744082,
      "grad_norm": 0.6572726400956087,
      "learning_rate": 3.796395992879516e-06,
      "loss": 0.0216,
      "step": 11279
    },
    {
      "epoch": 7.2168905950095965,
      "grad_norm": 0.7356505080948165,
      "learning_rate": 3.794770900723623e-06,
      "loss": 0.0245,
      "step": 11280
    },
    {
      "epoch": 7.217530390275112,
      "grad_norm": 0.7545969048559378,
      "learning_rate": 3.7931460750092353e-06,
      "loss": 0.0238,
      "step": 11281
    },
    {
      "epoch": 7.218170185540627,
      "grad_norm": 0.6942631086439627,
      "learning_rate": 3.7915215158061126e-06,
      "loss": 0.0239,
      "step": 11282
    },
    {
      "epoch": 7.218809980806142,
      "grad_norm": 0.6382592911256868,
      "learning_rate": 3.7898972231840125e-06,
      "loss": 0.0217,
      "step": 11283
    },
    {
      "epoch": 7.219449776071657,
      "grad_norm": 0.6838295643280212,
      "learning_rate": 3.788273197212675e-06,
      "loss": 0.0202,
      "step": 11284
    },
    {
      "epoch": 7.220089571337172,
      "grad_norm": 0.6994440781722474,
      "learning_rate": 3.7866494379618414e-06,
      "loss": 0.0262,
      "step": 11285
    },
    {
      "epoch": 7.220729366602687,
      "grad_norm": 0.7712630830292088,
      "learning_rate": 3.785025945501226e-06,
      "loss": 0.0263,
      "step": 11286
    },
    {
      "epoch": 7.2213691618682025,
      "grad_norm": 0.7833078253107194,
      "learning_rate": 3.7834027199005395e-06,
      "loss": 0.028,
      "step": 11287
    },
    {
      "epoch": 7.222008957133717,
      "grad_norm": 0.6097410146374751,
      "learning_rate": 3.7817797612294838e-06,
      "loss": 0.0223,
      "step": 11288
    },
    {
      "epoch": 7.222648752399232,
      "grad_norm": 0.5666084091547298,
      "learning_rate": 3.780157069557744e-06,
      "loss": 0.0198,
      "step": 11289
    },
    {
      "epoch": 7.2232885476647475,
      "grad_norm": 0.665012231621521,
      "learning_rate": 3.7785346449549964e-06,
      "loss": 0.0201,
      "step": 11290
    },
    {
      "epoch": 7.223928342930263,
      "grad_norm": 0.7531368284665981,
      "learning_rate": 3.7769124874909025e-06,
      "loss": 0.026,
      "step": 11291
    },
    {
      "epoch": 7.224568138195777,
      "grad_norm": 0.7365895465675479,
      "learning_rate": 3.7752905972351175e-06,
      "loss": 0.0277,
      "step": 11292
    },
    {
      "epoch": 7.225207933461292,
      "grad_norm": 0.6672983204156957,
      "learning_rate": 3.7736689742572806e-06,
      "loss": 0.0225,
      "step": 11293
    },
    {
      "epoch": 7.225847728726808,
      "grad_norm": 0.6598605045802716,
      "learning_rate": 3.7720476186270193e-06,
      "loss": 0.023,
      "step": 11294
    },
    {
      "epoch": 7.226487523992322,
      "grad_norm": 0.6277564403020696,
      "learning_rate": 3.770426530413959e-06,
      "loss": 0.0207,
      "step": 11295
    },
    {
      "epoch": 7.227127319257837,
      "grad_norm": 0.6973730055619806,
      "learning_rate": 3.7688057096876994e-06,
      "loss": 0.0246,
      "step": 11296
    },
    {
      "epoch": 7.227767114523353,
      "grad_norm": 0.8230377686951369,
      "learning_rate": 3.767185156517836e-06,
      "loss": 0.0218,
      "step": 11297
    },
    {
      "epoch": 7.228406909788868,
      "grad_norm": 0.633119592278954,
      "learning_rate": 3.765564870973959e-06,
      "loss": 0.0216,
      "step": 11298
    },
    {
      "epoch": 7.229046705054382,
      "grad_norm": 0.7030133929801513,
      "learning_rate": 3.763944853125635e-06,
      "loss": 0.0266,
      "step": 11299
    },
    {
      "epoch": 7.229686500319898,
      "grad_norm": 0.7905420405010832,
      "learning_rate": 3.7623251030424233e-06,
      "loss": 0.0214,
      "step": 11300
    },
    {
      "epoch": 7.230326295585413,
      "grad_norm": 0.7032216711017883,
      "learning_rate": 3.7607056207938786e-06,
      "loss": 0.0249,
      "step": 11301
    },
    {
      "epoch": 7.230966090850928,
      "grad_norm": 0.7255752414255031,
      "learning_rate": 3.759086406449536e-06,
      "loss": 0.0274,
      "step": 11302
    },
    {
      "epoch": 7.2316058861164425,
      "grad_norm": 0.7570700583280855,
      "learning_rate": 3.757467460078922e-06,
      "loss": 0.026,
      "step": 11303
    },
    {
      "epoch": 7.232245681381958,
      "grad_norm": 0.6777200475679995,
      "learning_rate": 3.7558487817515464e-06,
      "loss": 0.0243,
      "step": 11304
    },
    {
      "epoch": 7.232885476647473,
      "grad_norm": 0.7307780899841698,
      "learning_rate": 3.75423037153692e-06,
      "loss": 0.0234,
      "step": 11305
    },
    {
      "epoch": 7.233525271912987,
      "grad_norm": 0.7654585000914641,
      "learning_rate": 3.752612229504531e-06,
      "loss": 0.0245,
      "step": 11306
    },
    {
      "epoch": 7.234165067178503,
      "grad_norm": 0.6944927688282941,
      "learning_rate": 3.7509943557238605e-06,
      "loss": 0.0204,
      "step": 11307
    },
    {
      "epoch": 7.234804862444018,
      "grad_norm": 0.6075163534909132,
      "learning_rate": 3.7493767502643753e-06,
      "loss": 0.0204,
      "step": 11308
    },
    {
      "epoch": 7.235444657709533,
      "grad_norm": 0.6827180133561752,
      "learning_rate": 3.747759413195534e-06,
      "loss": 0.0254,
      "step": 11309
    },
    {
      "epoch": 7.236084452975048,
      "grad_norm": 0.9477735071284381,
      "learning_rate": 3.7461423445867807e-06,
      "loss": 0.0295,
      "step": 11310
    },
    {
      "epoch": 7.236724248240563,
      "grad_norm": 0.6468487189843425,
      "learning_rate": 3.744525544507548e-06,
      "loss": 0.0259,
      "step": 11311
    },
    {
      "epoch": 7.237364043506078,
      "grad_norm": 0.689584878146903,
      "learning_rate": 3.742909013027264e-06,
      "loss": 0.0235,
      "step": 11312
    },
    {
      "epoch": 7.2380038387715935,
      "grad_norm": 0.723017896769308,
      "learning_rate": 3.7412927502153364e-06,
      "loss": 0.0226,
      "step": 11313
    },
    {
      "epoch": 7.238643634037108,
      "grad_norm": 0.6859875613393134,
      "learning_rate": 3.7396767561411605e-06,
      "loss": 0.026,
      "step": 11314
    },
    {
      "epoch": 7.239283429302623,
      "grad_norm": 0.6319584431554867,
      "learning_rate": 3.7380610308741326e-06,
      "loss": 0.0211,
      "step": 11315
    },
    {
      "epoch": 7.239923224568138,
      "grad_norm": 0.6782334277845795,
      "learning_rate": 3.736445574483626e-06,
      "loss": 0.0242,
      "step": 11316
    },
    {
      "epoch": 7.240563019833653,
      "grad_norm": 0.786266833039473,
      "learning_rate": 3.734830387038999e-06,
      "loss": 0.0274,
      "step": 11317
    },
    {
      "epoch": 7.241202815099168,
      "grad_norm": 0.6192699593737215,
      "learning_rate": 3.7332154686096154e-06,
      "loss": 0.0195,
      "step": 11318
    },
    {
      "epoch": 7.241842610364683,
      "grad_norm": 0.7286370935028975,
      "learning_rate": 3.7316008192648114e-06,
      "loss": 0.0235,
      "step": 11319
    },
    {
      "epoch": 7.242482405630199,
      "grad_norm": 0.7125227254911507,
      "learning_rate": 3.729986439073917e-06,
      "loss": 0.0236,
      "step": 11320
    },
    {
      "epoch": 7.243122200895713,
      "grad_norm": 0.5959748333275883,
      "learning_rate": 3.728372328106251e-06,
      "loss": 0.0237,
      "step": 11321
    },
    {
      "epoch": 7.243761996161228,
      "grad_norm": 0.7059682984790979,
      "learning_rate": 3.7267584864311224e-06,
      "loss": 0.0271,
      "step": 11322
    },
    {
      "epoch": 7.244401791426744,
      "grad_norm": 0.6226647745783448,
      "learning_rate": 3.725144914117824e-06,
      "loss": 0.0211,
      "step": 11323
    },
    {
      "epoch": 7.245041586692259,
      "grad_norm": 0.6539720979406946,
      "learning_rate": 3.723531611235638e-06,
      "loss": 0.0266,
      "step": 11324
    },
    {
      "epoch": 7.245681381957773,
      "grad_norm": 0.5895407118688586,
      "learning_rate": 3.7219185778538425e-06,
      "loss": 0.0215,
      "step": 11325
    },
    {
      "epoch": 7.2463211772232885,
      "grad_norm": 0.7381026964068798,
      "learning_rate": 3.720305814041696e-06,
      "loss": 0.0277,
      "step": 11326
    },
    {
      "epoch": 7.246960972488804,
      "grad_norm": 0.698215367561776,
      "learning_rate": 3.718693319868444e-06,
      "loss": 0.0256,
      "step": 11327
    },
    {
      "epoch": 7.247600767754319,
      "grad_norm": 0.6785691835201906,
      "learning_rate": 3.7170810954033297e-06,
      "loss": 0.022,
      "step": 11328
    },
    {
      "epoch": 7.248240563019833,
      "grad_norm": 0.7914903315996986,
      "learning_rate": 3.7154691407155774e-06,
      "loss": 0.0268,
      "step": 11329
    },
    {
      "epoch": 7.248880358285349,
      "grad_norm": 0.7160330474439937,
      "learning_rate": 3.713857455874401e-06,
      "loss": 0.0219,
      "step": 11330
    },
    {
      "epoch": 7.249520153550864,
      "grad_norm": 0.7328350879392598,
      "learning_rate": 3.712246040948999e-06,
      "loss": 0.0237,
      "step": 11331
    },
    {
      "epoch": 7.250159948816378,
      "grad_norm": 0.7662083298321973,
      "learning_rate": 3.7106348960085716e-06,
      "loss": 0.0234,
      "step": 11332
    },
    {
      "epoch": 7.250799744081894,
      "grad_norm": 0.652257459058318,
      "learning_rate": 3.7090240211222927e-06,
      "loss": 0.0223,
      "step": 11333
    },
    {
      "epoch": 7.251439539347409,
      "grad_norm": 0.7720963946335171,
      "learning_rate": 3.707413416359329e-06,
      "loss": 0.0274,
      "step": 11334
    },
    {
      "epoch": 7.252079334612924,
      "grad_norm": 0.7086451664161082,
      "learning_rate": 3.7058030817888424e-06,
      "loss": 0.0256,
      "step": 11335
    },
    {
      "epoch": 7.252719129878439,
      "grad_norm": 0.6761370669056617,
      "learning_rate": 3.7041930174799755e-06,
      "loss": 0.0242,
      "step": 11336
    },
    {
      "epoch": 7.253358925143954,
      "grad_norm": 0.5784234053358162,
      "learning_rate": 3.7025832235018612e-06,
      "loss": 0.018,
      "step": 11337
    },
    {
      "epoch": 7.253998720409469,
      "grad_norm": 0.6154285540648294,
      "learning_rate": 3.70097369992362e-06,
      "loss": 0.0232,
      "step": 11338
    },
    {
      "epoch": 7.254638515674984,
      "grad_norm": 0.6208876494543261,
      "learning_rate": 3.699364446814363e-06,
      "loss": 0.0228,
      "step": 11339
    },
    {
      "epoch": 7.255278310940499,
      "grad_norm": 0.5912233567339767,
      "learning_rate": 3.69775546424319e-06,
      "loss": 0.0191,
      "step": 11340
    },
    {
      "epoch": 7.255918106206014,
      "grad_norm": 0.6622815829119724,
      "learning_rate": 3.6961467522791827e-06,
      "loss": 0.0209,
      "step": 11341
    },
    {
      "epoch": 7.256557901471529,
      "grad_norm": 0.7157844843686261,
      "learning_rate": 3.694538310991422e-06,
      "loss": 0.0244,
      "step": 11342
    },
    {
      "epoch": 7.257197696737044,
      "grad_norm": 0.6970211109003549,
      "learning_rate": 3.6929301404489714e-06,
      "loss": 0.0258,
      "step": 11343
    },
    {
      "epoch": 7.257837492002559,
      "grad_norm": 0.760864947152318,
      "learning_rate": 3.6913222407208783e-06,
      "loss": 0.0242,
      "step": 11344
    },
    {
      "epoch": 7.258477287268074,
      "grad_norm": 0.6872877538214753,
      "learning_rate": 3.689714611876187e-06,
      "loss": 0.0247,
      "step": 11345
    },
    {
      "epoch": 7.25911708253359,
      "grad_norm": 0.60078224577362,
      "learning_rate": 3.688107253983927e-06,
      "loss": 0.0201,
      "step": 11346
    },
    {
      "epoch": 7.259756877799104,
      "grad_norm": 0.6313893210730385,
      "learning_rate": 3.6865001671131095e-06,
      "loss": 0.0231,
      "step": 11347
    },
    {
      "epoch": 7.260396673064619,
      "grad_norm": 0.7616029408036163,
      "learning_rate": 3.6848933513327478e-06,
      "loss": 0.0288,
      "step": 11348
    },
    {
      "epoch": 7.2610364683301345,
      "grad_norm": 0.6269643859143278,
      "learning_rate": 3.6832868067118323e-06,
      "loss": 0.0214,
      "step": 11349
    },
    {
      "epoch": 7.26167626359565,
      "grad_norm": 0.7604402218131571,
      "learning_rate": 3.681680533319344e-06,
      "loss": 0.0263,
      "step": 11350
    },
    {
      "epoch": 7.262316058861164,
      "grad_norm": 0.5830495418551429,
      "learning_rate": 3.6800745312242534e-06,
      "loss": 0.023,
      "step": 11351
    },
    {
      "epoch": 7.262955854126679,
      "grad_norm": 0.6020577070558006,
      "learning_rate": 3.6784688004955214e-06,
      "loss": 0.0212,
      "step": 11352
    },
    {
      "epoch": 7.263595649392195,
      "grad_norm": 0.6418860109388116,
      "learning_rate": 3.676863341202094e-06,
      "loss": 0.0223,
      "step": 11353
    },
    {
      "epoch": 7.264235444657709,
      "grad_norm": 0.5886912790192103,
      "learning_rate": 3.6752581534129026e-06,
      "loss": 0.0184,
      "step": 11354
    },
    {
      "epoch": 7.264875239923224,
      "grad_norm": 0.803752297591294,
      "learning_rate": 3.673653237196879e-06,
      "loss": 0.019,
      "step": 11355
    },
    {
      "epoch": 7.26551503518874,
      "grad_norm": 0.7140068487325693,
      "learning_rate": 3.672048592622931e-06,
      "loss": 0.0208,
      "step": 11356
    },
    {
      "epoch": 7.266154830454255,
      "grad_norm": 0.7062252573803961,
      "learning_rate": 3.6704442197599576e-06,
      "loss": 0.0242,
      "step": 11357
    },
    {
      "epoch": 7.266794625719769,
      "grad_norm": 0.742257476061499,
      "learning_rate": 3.668840118676853e-06,
      "loss": 0.0237,
      "step": 11358
    },
    {
      "epoch": 7.267434420985285,
      "grad_norm": 0.6873907705963008,
      "learning_rate": 3.6672362894424918e-06,
      "loss": 0.0214,
      "step": 11359
    },
    {
      "epoch": 7.2680742162508,
      "grad_norm": 0.6580860722506429,
      "learning_rate": 3.665632732125738e-06,
      "loss": 0.0197,
      "step": 11360
    },
    {
      "epoch": 7.268714011516315,
      "grad_norm": 0.6722726205390351,
      "learning_rate": 3.664029446795444e-06,
      "loss": 0.0242,
      "step": 11361
    },
    {
      "epoch": 7.2693538067818295,
      "grad_norm": 0.6288347405377392,
      "learning_rate": 3.6624264335204585e-06,
      "loss": 0.023,
      "step": 11362
    },
    {
      "epoch": 7.269993602047345,
      "grad_norm": 0.5741944804005005,
      "learning_rate": 3.6608236923696085e-06,
      "loss": 0.0233,
      "step": 11363
    },
    {
      "epoch": 7.27063339731286,
      "grad_norm": 0.6989949988673934,
      "learning_rate": 3.6592212234117085e-06,
      "loss": 0.025,
      "step": 11364
    },
    {
      "epoch": 7.271273192578375,
      "grad_norm": 0.6816665688011184,
      "learning_rate": 3.657619026715573e-06,
      "loss": 0.0231,
      "step": 11365
    },
    {
      "epoch": 7.27191298784389,
      "grad_norm": 0.6364308532752999,
      "learning_rate": 3.6560171023499947e-06,
      "loss": 0.0226,
      "step": 11366
    },
    {
      "epoch": 7.272552783109405,
      "grad_norm": 0.7344190104854361,
      "learning_rate": 3.6544154503837572e-06,
      "loss": 0.028,
      "step": 11367
    },
    {
      "epoch": 7.27319257837492,
      "grad_norm": 0.6052060662271866,
      "learning_rate": 3.652814070885632e-06,
      "loss": 0.0215,
      "step": 11368
    },
    {
      "epoch": 7.273832373640435,
      "grad_norm": 0.6334055744080916,
      "learning_rate": 3.6512129639243798e-06,
      "loss": 0.0234,
      "step": 11369
    },
    {
      "epoch": 7.27447216890595,
      "grad_norm": 0.7196242519479568,
      "learning_rate": 3.6496121295687505e-06,
      "loss": 0.0222,
      "step": 11370
    },
    {
      "epoch": 7.275111964171465,
      "grad_norm": 0.7646994073274792,
      "learning_rate": 3.648011567887475e-06,
      "loss": 0.0265,
      "step": 11371
    },
    {
      "epoch": 7.2757517594369805,
      "grad_norm": 0.7465866046791807,
      "learning_rate": 3.6464112789492867e-06,
      "loss": 0.0279,
      "step": 11372
    },
    {
      "epoch": 7.276391554702495,
      "grad_norm": 0.6726836004212516,
      "learning_rate": 3.644811262822897e-06,
      "loss": 0.0263,
      "step": 11373
    },
    {
      "epoch": 7.27703134996801,
      "grad_norm": 0.723945273378176,
      "learning_rate": 3.6432115195770023e-06,
      "loss": 0.0248,
      "step": 11374
    },
    {
      "epoch": 7.277671145233525,
      "grad_norm": 0.7008495908618102,
      "learning_rate": 3.641612049280302e-06,
      "loss": 0.0265,
      "step": 11375
    },
    {
      "epoch": 7.278310940499041,
      "grad_norm": 0.6232149903976612,
      "learning_rate": 3.640012852001468e-06,
      "loss": 0.0232,
      "step": 11376
    },
    {
      "epoch": 7.278950735764555,
      "grad_norm": 0.7622267445346629,
      "learning_rate": 3.638413927809166e-06,
      "loss": 0.024,
      "step": 11377
    },
    {
      "epoch": 7.27959053103007,
      "grad_norm": 0.7096600728824425,
      "learning_rate": 3.6368152767720565e-06,
      "loss": 0.0264,
      "step": 11378
    },
    {
      "epoch": 7.280230326295586,
      "grad_norm": 0.7381096105335653,
      "learning_rate": 3.63521689895878e-06,
      "loss": 0.0249,
      "step": 11379
    },
    {
      "epoch": 7.280870121561101,
      "grad_norm": 0.6913455065614993,
      "learning_rate": 3.6336187944379686e-06,
      "loss": 0.0258,
      "step": 11380
    },
    {
      "epoch": 7.281509916826615,
      "grad_norm": 0.7071286110058604,
      "learning_rate": 3.6320209632782398e-06,
      "loss": 0.0246,
      "step": 11381
    },
    {
      "epoch": 7.282149712092131,
      "grad_norm": 0.6095466133620667,
      "learning_rate": 3.6304234055482036e-06,
      "loss": 0.0191,
      "step": 11382
    },
    {
      "epoch": 7.282789507357646,
      "grad_norm": 0.5987874877208073,
      "learning_rate": 3.6288261213164567e-06,
      "loss": 0.0197,
      "step": 11383
    },
    {
      "epoch": 7.28342930262316,
      "grad_norm": 0.7298207366358618,
      "learning_rate": 3.6272291106515777e-06,
      "loss": 0.0274,
      "step": 11384
    },
    {
      "epoch": 7.2840690978886755,
      "grad_norm": 0.7155758368767043,
      "learning_rate": 3.6256323736221487e-06,
      "loss": 0.0284,
      "step": 11385
    },
    {
      "epoch": 7.284708893154191,
      "grad_norm": 0.758756654479808,
      "learning_rate": 3.624035910296727e-06,
      "loss": 0.0221,
      "step": 11386
    },
    {
      "epoch": 7.285348688419706,
      "grad_norm": 0.6462009432508009,
      "learning_rate": 3.622439720743858e-06,
      "loss": 0.0232,
      "step": 11387
    },
    {
      "epoch": 7.2859884836852205,
      "grad_norm": 0.7881429303361569,
      "learning_rate": 3.620843805032085e-06,
      "loss": 0.0258,
      "step": 11388
    },
    {
      "epoch": 7.286628278950736,
      "grad_norm": 0.6092621654182271,
      "learning_rate": 3.6192481632299312e-06,
      "loss": 0.0208,
      "step": 11389
    },
    {
      "epoch": 7.287268074216251,
      "grad_norm": 0.6658669714941143,
      "learning_rate": 3.6176527954059114e-06,
      "loss": 0.0266,
      "step": 11390
    },
    {
      "epoch": 7.287907869481765,
      "grad_norm": 0.6785931862512845,
      "learning_rate": 3.6160577016285236e-06,
      "loss": 0.0244,
      "step": 11391
    },
    {
      "epoch": 7.288547664747281,
      "grad_norm": 0.9292856426896688,
      "learning_rate": 3.614462881966266e-06,
      "loss": 0.0363,
      "step": 11392
    },
    {
      "epoch": 7.289187460012796,
      "grad_norm": 0.6262904616414744,
      "learning_rate": 3.612868336487613e-06,
      "loss": 0.021,
      "step": 11393
    },
    {
      "epoch": 7.289827255278311,
      "grad_norm": 0.5942822527967605,
      "learning_rate": 3.611274065261028e-06,
      "loss": 0.0191,
      "step": 11394
    },
    {
      "epoch": 7.290467050543826,
      "grad_norm": 0.6275315036110725,
      "learning_rate": 3.6096800683549738e-06,
      "loss": 0.0232,
      "step": 11395
    },
    {
      "epoch": 7.291106845809341,
      "grad_norm": 0.704363157055601,
      "learning_rate": 3.608086345837889e-06,
      "loss": 0.0229,
      "step": 11396
    },
    {
      "epoch": 7.291746641074856,
      "grad_norm": 0.7546865514507832,
      "learning_rate": 3.6064928977782078e-06,
      "loss": 0.0245,
      "step": 11397
    },
    {
      "epoch": 7.292386436340371,
      "grad_norm": 0.6205929278849817,
      "learning_rate": 3.6048997242443472e-06,
      "loss": 0.0251,
      "step": 11398
    },
    {
      "epoch": 7.293026231605886,
      "grad_norm": 0.7128074901115059,
      "learning_rate": 3.6033068253047167e-06,
      "loss": 0.0173,
      "step": 11399
    },
    {
      "epoch": 7.293666026871401,
      "grad_norm": 0.6543034511813401,
      "learning_rate": 3.601714201027712e-06,
      "loss": 0.0243,
      "step": 11400
    },
    {
      "epoch": 7.294305822136916,
      "grad_norm": 0.5799131972780435,
      "learning_rate": 3.600121851481715e-06,
      "loss": 0.0235,
      "step": 11401
    },
    {
      "epoch": 7.294945617402432,
      "grad_norm": 0.7837447215566478,
      "learning_rate": 3.5985297767351047e-06,
      "loss": 0.0273,
      "step": 11402
    },
    {
      "epoch": 7.295585412667946,
      "grad_norm": 0.8154280611977677,
      "learning_rate": 3.596937976856237e-06,
      "loss": 0.0246,
      "step": 11403
    },
    {
      "epoch": 7.296225207933461,
      "grad_norm": 0.7338849474479364,
      "learning_rate": 3.595346451913461e-06,
      "loss": 0.0243,
      "step": 11404
    },
    {
      "epoch": 7.296865003198977,
      "grad_norm": 0.7233867867194138,
      "learning_rate": 3.5937552019751177e-06,
      "loss": 0.0255,
      "step": 11405
    },
    {
      "epoch": 7.297504798464491,
      "grad_norm": 0.714188126495953,
      "learning_rate": 3.5921642271095303e-06,
      "loss": 0.0234,
      "step": 11406
    },
    {
      "epoch": 7.298144593730006,
      "grad_norm": 0.6751899146299374,
      "learning_rate": 3.5905735273850127e-06,
      "loss": 0.0273,
      "step": 11407
    },
    {
      "epoch": 7.2987843889955215,
      "grad_norm": 0.715685536455184,
      "learning_rate": 3.5889831028698628e-06,
      "loss": 0.0233,
      "step": 11408
    },
    {
      "epoch": 7.299424184261037,
      "grad_norm": 0.7093520800105225,
      "learning_rate": 3.587392953632377e-06,
      "loss": 0.0261,
      "step": 11409
    },
    {
      "epoch": 7.300063979526551,
      "grad_norm": 0.7553206913031083,
      "learning_rate": 3.585803079740832e-06,
      "loss": 0.0286,
      "step": 11410
    },
    {
      "epoch": 7.3007037747920664,
      "grad_norm": 0.6724776873153343,
      "learning_rate": 3.584213481263491e-06,
      "loss": 0.0231,
      "step": 11411
    },
    {
      "epoch": 7.301343570057582,
      "grad_norm": 0.7887509594684359,
      "learning_rate": 3.5826241582686115e-06,
      "loss": 0.0261,
      "step": 11412
    },
    {
      "epoch": 7.301983365323097,
      "grad_norm": 0.53107831837529,
      "learning_rate": 3.5810351108244347e-06,
      "loss": 0.0191,
      "step": 11413
    },
    {
      "epoch": 7.302623160588611,
      "grad_norm": 0.8771181462601229,
      "learning_rate": 3.5794463389991886e-06,
      "loss": 0.0251,
      "step": 11414
    },
    {
      "epoch": 7.303262955854127,
      "grad_norm": 0.7405265786446564,
      "learning_rate": 3.5778578428610978e-06,
      "loss": 0.0269,
      "step": 11415
    },
    {
      "epoch": 7.303902751119642,
      "grad_norm": 0.6387043127009403,
      "learning_rate": 3.5762696224783687e-06,
      "loss": 0.026,
      "step": 11416
    },
    {
      "epoch": 7.304542546385157,
      "grad_norm": 0.7014291095280981,
      "learning_rate": 3.574681677919194e-06,
      "loss": 0.027,
      "step": 11417
    },
    {
      "epoch": 7.305182341650672,
      "grad_norm": 0.7492761826729997,
      "learning_rate": 3.573094009251755e-06,
      "loss": 0.0236,
      "step": 11418
    },
    {
      "epoch": 7.305822136916187,
      "grad_norm": 0.6062923689037791,
      "learning_rate": 3.5715066165442304e-06,
      "loss": 0.022,
      "step": 11419
    },
    {
      "epoch": 7.306461932181702,
      "grad_norm": 0.7428612673167133,
      "learning_rate": 3.5699194998647758e-06,
      "loss": 0.0275,
      "step": 11420
    },
    {
      "epoch": 7.3071017274472165,
      "grad_norm": 0.6940229848566111,
      "learning_rate": 3.568332659281537e-06,
      "loss": 0.0239,
      "step": 11421
    },
    {
      "epoch": 7.307741522712732,
      "grad_norm": 0.6731587574528495,
      "learning_rate": 3.5667460948626554e-06,
      "loss": 0.0193,
      "step": 11422
    },
    {
      "epoch": 7.308381317978247,
      "grad_norm": 0.6777004483694858,
      "learning_rate": 3.5651598066762528e-06,
      "loss": 0.023,
      "step": 11423
    },
    {
      "epoch": 7.309021113243762,
      "grad_norm": 0.6550446916627888,
      "learning_rate": 3.563573794790438e-06,
      "loss": 0.0224,
      "step": 11424
    },
    {
      "epoch": 7.309660908509277,
      "grad_norm": 0.729033934671041,
      "learning_rate": 3.5619880592733193e-06,
      "loss": 0.0241,
      "step": 11425
    },
    {
      "epoch": 7.310300703774792,
      "grad_norm": 0.704315534474657,
      "learning_rate": 3.5604026001929803e-06,
      "loss": 0.0286,
      "step": 11426
    },
    {
      "epoch": 7.310940499040307,
      "grad_norm": 0.6442574463657473,
      "learning_rate": 3.558817417617498e-06,
      "loss": 0.0225,
      "step": 11427
    },
    {
      "epoch": 7.311580294305822,
      "grad_norm": 0.5895370229058987,
      "learning_rate": 3.5572325116149386e-06,
      "loss": 0.0223,
      "step": 11428
    },
    {
      "epoch": 7.312220089571337,
      "grad_norm": 0.7118634051838536,
      "learning_rate": 3.5556478822533536e-06,
      "loss": 0.0241,
      "step": 11429
    },
    {
      "epoch": 7.312859884836852,
      "grad_norm": 0.583461864757141,
      "learning_rate": 3.5540635296007854e-06,
      "loss": 0.0206,
      "step": 11430
    },
    {
      "epoch": 7.3134996801023675,
      "grad_norm": 0.7213709971548847,
      "learning_rate": 3.5524794537252594e-06,
      "loss": 0.0255,
      "step": 11431
    },
    {
      "epoch": 7.314139475367882,
      "grad_norm": 0.7557248951755183,
      "learning_rate": 3.5508956546947994e-06,
      "loss": 0.0256,
      "step": 11432
    },
    {
      "epoch": 7.314779270633397,
      "grad_norm": 0.6939959724504187,
      "learning_rate": 3.5493121325774084e-06,
      "loss": 0.026,
      "step": 11433
    },
    {
      "epoch": 7.315419065898912,
      "grad_norm": 0.7336964225562378,
      "learning_rate": 3.5477288874410765e-06,
      "loss": 0.0287,
      "step": 11434
    },
    {
      "epoch": 7.316058861164428,
      "grad_norm": 0.6759334845428024,
      "learning_rate": 3.5461459193537905e-06,
      "loss": 0.0249,
      "step": 11435
    },
    {
      "epoch": 7.316698656429942,
      "grad_norm": 0.6298437601005143,
      "learning_rate": 3.5445632283835195e-06,
      "loss": 0.022,
      "step": 11436
    },
    {
      "epoch": 7.317338451695457,
      "grad_norm": 0.6563388607690398,
      "learning_rate": 3.542980814598219e-06,
      "loss": 0.0218,
      "step": 11437
    },
    {
      "epoch": 7.317978246960973,
      "grad_norm": 0.546072561792758,
      "learning_rate": 3.5413986780658337e-06,
      "loss": 0.0173,
      "step": 11438
    },
    {
      "epoch": 7.318618042226488,
      "grad_norm": 0.6354355815794663,
      "learning_rate": 3.5398168188543037e-06,
      "loss": 0.0215,
      "step": 11439
    },
    {
      "epoch": 7.319257837492002,
      "grad_norm": 0.6564598935163859,
      "learning_rate": 3.5382352370315474e-06,
      "loss": 0.024,
      "step": 11440
    },
    {
      "epoch": 7.319897632757518,
      "grad_norm": 0.5679220283859372,
      "learning_rate": 3.536653932665476e-06,
      "loss": 0.0192,
      "step": 11441
    },
    {
      "epoch": 7.320537428023033,
      "grad_norm": 0.8630216767616176,
      "learning_rate": 3.535072905823986e-06,
      "loss": 0.0306,
      "step": 11442
    },
    {
      "epoch": 7.321177223288547,
      "grad_norm": 0.6868476109634839,
      "learning_rate": 3.5334921565749656e-06,
      "loss": 0.0235,
      "step": 11443
    },
    {
      "epoch": 7.3218170185540625,
      "grad_norm": 0.6243446915936935,
      "learning_rate": 3.5319116849862865e-06,
      "loss": 0.0225,
      "step": 11444
    },
    {
      "epoch": 7.322456813819578,
      "grad_norm": 0.6600563857869061,
      "learning_rate": 3.530331491125817e-06,
      "loss": 0.0216,
      "step": 11445
    },
    {
      "epoch": 7.323096609085093,
      "grad_norm": 0.6569202516106326,
      "learning_rate": 3.528751575061404e-06,
      "loss": 0.0205,
      "step": 11446
    },
    {
      "epoch": 7.3237364043506075,
      "grad_norm": 0.5342927936841719,
      "learning_rate": 3.527171936860887e-06,
      "loss": 0.0199,
      "step": 11447
    },
    {
      "epoch": 7.324376199616123,
      "grad_norm": 0.5992246814776443,
      "learning_rate": 3.5255925765920895e-06,
      "loss": 0.0207,
      "step": 11448
    },
    {
      "epoch": 7.325015994881638,
      "grad_norm": 0.6882526823954588,
      "learning_rate": 3.524013494322833e-06,
      "loss": 0.024,
      "step": 11449
    },
    {
      "epoch": 7.325655790147153,
      "grad_norm": 0.6245278276839757,
      "learning_rate": 3.522434690120917e-06,
      "loss": 0.0207,
      "step": 11450
    },
    {
      "epoch": 7.326295585412668,
      "grad_norm": 0.7396237583894023,
      "learning_rate": 3.5208561640541308e-06,
      "loss": 0.0245,
      "step": 11451
    },
    {
      "epoch": 7.326935380678183,
      "grad_norm": 0.7637570795460913,
      "learning_rate": 3.519277916190257e-06,
      "loss": 0.0223,
      "step": 11452
    },
    {
      "epoch": 7.327575175943698,
      "grad_norm": 0.6721888018880429,
      "learning_rate": 3.517699946597062e-06,
      "loss": 0.0237,
      "step": 11453
    },
    {
      "epoch": 7.3282149712092135,
      "grad_norm": 0.6494085878287373,
      "learning_rate": 3.516122255342297e-06,
      "loss": 0.0228,
      "step": 11454
    },
    {
      "epoch": 7.328854766474728,
      "grad_norm": 0.7010811655458653,
      "learning_rate": 3.514544842493711e-06,
      "loss": 0.0262,
      "step": 11455
    },
    {
      "epoch": 7.329494561740243,
      "grad_norm": 0.7141767225341797,
      "learning_rate": 3.512967708119033e-06,
      "loss": 0.0268,
      "step": 11456
    },
    {
      "epoch": 7.330134357005758,
      "grad_norm": 0.6030621255400722,
      "learning_rate": 3.511390852285982e-06,
      "loss": 0.0218,
      "step": 11457
    },
    {
      "epoch": 7.330774152271273,
      "grad_norm": 0.645406528854119,
      "learning_rate": 3.5098142750622654e-06,
      "loss": 0.0237,
      "step": 11458
    },
    {
      "epoch": 7.331413947536788,
      "grad_norm": 0.5714110321564625,
      "learning_rate": 3.5082379765155795e-06,
      "loss": 0.0235,
      "step": 11459
    },
    {
      "epoch": 7.332053742802303,
      "grad_norm": 0.6305643436437269,
      "learning_rate": 3.506661956713606e-06,
      "loss": 0.0252,
      "step": 11460
    },
    {
      "epoch": 7.332693538067819,
      "grad_norm": 0.6812310650316512,
      "learning_rate": 3.505086215724015e-06,
      "loss": 0.0216,
      "step": 11461
    },
    {
      "epoch": 7.333333333333333,
      "grad_norm": 0.6686132546829467,
      "learning_rate": 3.5035107536144707e-06,
      "loss": 0.0218,
      "step": 11462
    },
    {
      "epoch": 7.333973128598848,
      "grad_norm": 0.6713764004307283,
      "learning_rate": 3.501935570452619e-06,
      "loss": 0.0226,
      "step": 11463
    },
    {
      "epoch": 7.334612923864364,
      "grad_norm": 0.616813485396044,
      "learning_rate": 3.500360666306091e-06,
      "loss": 0.0236,
      "step": 11464
    },
    {
      "epoch": 7.335252719129878,
      "grad_norm": 0.617233047674344,
      "learning_rate": 3.498786041242517e-06,
      "loss": 0.0234,
      "step": 11465
    },
    {
      "epoch": 7.335892514395393,
      "grad_norm": 0.5657967622223533,
      "learning_rate": 3.4972116953295067e-06,
      "loss": 0.0221,
      "step": 11466
    },
    {
      "epoch": 7.3365323096609085,
      "grad_norm": 0.6549000587531026,
      "learning_rate": 3.495637628634657e-06,
      "loss": 0.021,
      "step": 11467
    },
    {
      "epoch": 7.337172104926424,
      "grad_norm": 0.6817483449585872,
      "learning_rate": 3.4940638412255557e-06,
      "loss": 0.0279,
      "step": 11468
    },
    {
      "epoch": 7.337811900191938,
      "grad_norm": 0.6388695457526139,
      "learning_rate": 3.4924903331697822e-06,
      "loss": 0.0228,
      "step": 11469
    },
    {
      "epoch": 7.3384516954574535,
      "grad_norm": 0.7600547426304846,
      "learning_rate": 3.4909171045348976e-06,
      "loss": 0.0289,
      "step": 11470
    },
    {
      "epoch": 7.339091490722969,
      "grad_norm": 0.6870607716781387,
      "learning_rate": 3.4893441553884544e-06,
      "loss": 0.0243,
      "step": 11471
    },
    {
      "epoch": 7.339731285988484,
      "grad_norm": 0.629998098127544,
      "learning_rate": 3.4877714857979915e-06,
      "loss": 0.0218,
      "step": 11472
    },
    {
      "epoch": 7.340371081253998,
      "grad_norm": 0.6662585864610568,
      "learning_rate": 3.4861990958310367e-06,
      "loss": 0.0217,
      "step": 11473
    },
    {
      "epoch": 7.341010876519514,
      "grad_norm": 0.6561197250686686,
      "learning_rate": 3.484626985555105e-06,
      "loss": 0.0259,
      "step": 11474
    },
    {
      "epoch": 7.341650671785029,
      "grad_norm": 0.7603090122246905,
      "learning_rate": 3.483055155037698e-06,
      "loss": 0.0276,
      "step": 11475
    },
    {
      "epoch": 7.342290467050544,
      "grad_norm": 0.6615220524995639,
      "learning_rate": 3.481483604346313e-06,
      "loss": 0.026,
      "step": 11476
    },
    {
      "epoch": 7.342930262316059,
      "grad_norm": 0.7448108954196055,
      "learning_rate": 3.479912333548425e-06,
      "loss": 0.0291,
      "step": 11477
    },
    {
      "epoch": 7.343570057581574,
      "grad_norm": 0.6789547796654094,
      "learning_rate": 3.4783413427115022e-06,
      "loss": 0.0195,
      "step": 11478
    },
    {
      "epoch": 7.344209852847089,
      "grad_norm": 0.7039821917511555,
      "learning_rate": 3.4767706319030025e-06,
      "loss": 0.0251,
      "step": 11479
    },
    {
      "epoch": 7.3448496481126035,
      "grad_norm": 0.6773088090218816,
      "learning_rate": 3.4752002011903685e-06,
      "loss": 0.0235,
      "step": 11480
    },
    {
      "epoch": 7.345489443378119,
      "grad_norm": 0.8877916032130881,
      "learning_rate": 3.4736300506410282e-06,
      "loss": 0.0274,
      "step": 11481
    },
    {
      "epoch": 7.346129238643634,
      "grad_norm": 0.7558603912716715,
      "learning_rate": 3.4720601803224076e-06,
      "loss": 0.0205,
      "step": 11482
    },
    {
      "epoch": 7.346769033909149,
      "grad_norm": 0.6564378581469156,
      "learning_rate": 3.4704905903019092e-06,
      "loss": 0.0247,
      "step": 11483
    },
    {
      "epoch": 7.347408829174664,
      "grad_norm": 0.7086390417932573,
      "learning_rate": 3.4689212806469308e-06,
      "loss": 0.0257,
      "step": 11484
    },
    {
      "epoch": 7.348048624440179,
      "grad_norm": 0.6509165675067224,
      "learning_rate": 3.4673522514248516e-06,
      "loss": 0.0219,
      "step": 11485
    },
    {
      "epoch": 7.348688419705694,
      "grad_norm": 0.6872602238029483,
      "learning_rate": 3.4657835027030496e-06,
      "loss": 0.025,
      "step": 11486
    },
    {
      "epoch": 7.34932821497121,
      "grad_norm": 0.6863417098689341,
      "learning_rate": 3.464215034548879e-06,
      "loss": 0.0219,
      "step": 11487
    },
    {
      "epoch": 7.349968010236724,
      "grad_norm": 0.7491846148173174,
      "learning_rate": 3.4626468470296892e-06,
      "loss": 0.0281,
      "step": 11488
    },
    {
      "epoch": 7.350607805502239,
      "grad_norm": 0.6551341857423877,
      "learning_rate": 3.4610789402128154e-06,
      "loss": 0.0252,
      "step": 11489
    },
    {
      "epoch": 7.3512476007677545,
      "grad_norm": 0.5829128339318246,
      "learning_rate": 3.45951131416558e-06,
      "loss": 0.0209,
      "step": 11490
    },
    {
      "epoch": 7.35188739603327,
      "grad_norm": 0.7286494956692215,
      "learning_rate": 3.4579439689552907e-06,
      "loss": 0.0267,
      "step": 11491
    },
    {
      "epoch": 7.352527191298784,
      "grad_norm": 0.6637404088413293,
      "learning_rate": 3.456376904649252e-06,
      "loss": 0.0221,
      "step": 11492
    },
    {
      "epoch": 7.3531669865642995,
      "grad_norm": 0.6804154329892358,
      "learning_rate": 3.4548101213147512e-06,
      "loss": 0.0262,
      "step": 11493
    },
    {
      "epoch": 7.353806781829815,
      "grad_norm": 0.5646786972907823,
      "learning_rate": 3.453243619019059e-06,
      "loss": 0.0212,
      "step": 11494
    },
    {
      "epoch": 7.354446577095329,
      "grad_norm": 0.7356012244627226,
      "learning_rate": 3.451677397829437e-06,
      "loss": 0.0186,
      "step": 11495
    },
    {
      "epoch": 7.355086372360844,
      "grad_norm": 0.5917465712214056,
      "learning_rate": 3.4501114578131425e-06,
      "loss": 0.0195,
      "step": 11496
    },
    {
      "epoch": 7.35572616762636,
      "grad_norm": 0.680412331232924,
      "learning_rate": 3.4485457990374104e-06,
      "loss": 0.0248,
      "step": 11497
    },
    {
      "epoch": 7.356365962891875,
      "grad_norm": 0.770244442430586,
      "learning_rate": 3.4469804215694647e-06,
      "loss": 0.0316,
      "step": 11498
    },
    {
      "epoch": 7.357005758157389,
      "grad_norm": 0.8427475783110019,
      "learning_rate": 3.4454153254765256e-06,
      "loss": 0.0256,
      "step": 11499
    },
    {
      "epoch": 7.357645553422905,
      "grad_norm": 0.6948271946579344,
      "learning_rate": 3.4438505108257936e-06,
      "loss": 0.0235,
      "step": 11500
    },
    {
      "epoch": 7.35828534868842,
      "grad_norm": 0.553540511327082,
      "learning_rate": 3.4422859776844576e-06,
      "loss": 0.0191,
      "step": 11501
    },
    {
      "epoch": 7.358925143953935,
      "grad_norm": 0.6449111678419155,
      "learning_rate": 3.4407217261196957e-06,
      "loss": 0.0224,
      "step": 11502
    },
    {
      "epoch": 7.3595649392194495,
      "grad_norm": 0.6741740363826552,
      "learning_rate": 3.4391577561986755e-06,
      "loss": 0.0263,
      "step": 11503
    },
    {
      "epoch": 7.360204734484965,
      "grad_norm": 0.6437418197044716,
      "learning_rate": 3.437594067988551e-06,
      "loss": 0.0217,
      "step": 11504
    },
    {
      "epoch": 7.36084452975048,
      "grad_norm": 0.7424881557433052,
      "learning_rate": 3.43603066155646e-06,
      "loss": 0.0264,
      "step": 11505
    },
    {
      "epoch": 7.3614843250159945,
      "grad_norm": 0.5838581723114765,
      "learning_rate": 3.434467536969539e-06,
      "loss": 0.0197,
      "step": 11506
    },
    {
      "epoch": 7.36212412028151,
      "grad_norm": 0.6491392263145288,
      "learning_rate": 3.432904694294903e-06,
      "loss": 0.0228,
      "step": 11507
    },
    {
      "epoch": 7.362763915547025,
      "grad_norm": 0.6037447031141242,
      "learning_rate": 3.431342133599653e-06,
      "loss": 0.0196,
      "step": 11508
    },
    {
      "epoch": 7.36340371081254,
      "grad_norm": 0.7283949851074959,
      "learning_rate": 3.429779854950892e-06,
      "loss": 0.0209,
      "step": 11509
    },
    {
      "epoch": 7.364043506078055,
      "grad_norm": 0.6437835681766058,
      "learning_rate": 3.4282178584156956e-06,
      "loss": 0.0253,
      "step": 11510
    },
    {
      "epoch": 7.36468330134357,
      "grad_norm": 0.680063978834284,
      "learning_rate": 3.426656144061131e-06,
      "loss": 0.024,
      "step": 11511
    },
    {
      "epoch": 7.365323096609085,
      "grad_norm": 0.7066394803166599,
      "learning_rate": 3.4250947119542623e-06,
      "loss": 0.0224,
      "step": 11512
    },
    {
      "epoch": 7.3659628918746005,
      "grad_norm": 0.7562256127118415,
      "learning_rate": 3.42353356216213e-06,
      "loss": 0.0256,
      "step": 11513
    },
    {
      "epoch": 7.366602687140115,
      "grad_norm": 0.657099206332194,
      "learning_rate": 3.4219726947517683e-06,
      "loss": 0.0226,
      "step": 11514
    },
    {
      "epoch": 7.36724248240563,
      "grad_norm": 0.8783236427658695,
      "learning_rate": 3.420412109790194e-06,
      "loss": 0.0266,
      "step": 11515
    },
    {
      "epoch": 7.3678822776711455,
      "grad_norm": 0.703301337720672,
      "learning_rate": 3.4188518073444245e-06,
      "loss": 0.0194,
      "step": 11516
    },
    {
      "epoch": 7.36852207293666,
      "grad_norm": 0.7203115378011773,
      "learning_rate": 3.417291787481449e-06,
      "loss": 0.0256,
      "step": 11517
    },
    {
      "epoch": 7.369161868202175,
      "grad_norm": 0.6497646769818675,
      "learning_rate": 3.4157320502682555e-06,
      "loss": 0.0243,
      "step": 11518
    },
    {
      "epoch": 7.36980166346769,
      "grad_norm": 0.8903830343067912,
      "learning_rate": 3.4141725957718165e-06,
      "loss": 0.027,
      "step": 11519
    },
    {
      "epoch": 7.370441458733206,
      "grad_norm": 0.5935669167553483,
      "learning_rate": 3.412613424059089e-06,
      "loss": 0.0236,
      "step": 11520
    },
    {
      "epoch": 7.37108125399872,
      "grad_norm": 0.646780773197703,
      "learning_rate": 3.4110545351970205e-06,
      "loss": 0.0248,
      "step": 11521
    },
    {
      "epoch": 7.371721049264235,
      "grad_norm": 0.7444783578630101,
      "learning_rate": 3.409495929252553e-06,
      "loss": 0.0225,
      "step": 11522
    },
    {
      "epoch": 7.372360844529751,
      "grad_norm": 0.5972217553532344,
      "learning_rate": 3.407937606292606e-06,
      "loss": 0.0186,
      "step": 11523
    },
    {
      "epoch": 7.373000639795266,
      "grad_norm": 0.6603902834372312,
      "learning_rate": 3.4063795663840925e-06,
      "loss": 0.0231,
      "step": 11524
    },
    {
      "epoch": 7.37364043506078,
      "grad_norm": 0.683996162349986,
      "learning_rate": 3.404821809593908e-06,
      "loss": 0.0259,
      "step": 11525
    },
    {
      "epoch": 7.3742802303262955,
      "grad_norm": 0.7207897242522158,
      "learning_rate": 3.403264335988946e-06,
      "loss": 0.0266,
      "step": 11526
    },
    {
      "epoch": 7.374920025591811,
      "grad_norm": 0.6279237537817841,
      "learning_rate": 3.4017071456360785e-06,
      "loss": 0.0213,
      "step": 11527
    },
    {
      "epoch": 7.375559820857326,
      "grad_norm": 0.7119209573998654,
      "learning_rate": 3.4001502386021657e-06,
      "loss": 0.0197,
      "step": 11528
    },
    {
      "epoch": 7.3761996161228405,
      "grad_norm": 0.6452503434923148,
      "learning_rate": 3.3985936149540656e-06,
      "loss": 0.0228,
      "step": 11529
    },
    {
      "epoch": 7.376839411388356,
      "grad_norm": 0.6446700159063837,
      "learning_rate": 3.397037274758611e-06,
      "loss": 0.0232,
      "step": 11530
    },
    {
      "epoch": 7.377479206653871,
      "grad_norm": 0.6291335047723917,
      "learning_rate": 3.3954812180826312e-06,
      "loss": 0.0204,
      "step": 11531
    },
    {
      "epoch": 7.378119001919385,
      "grad_norm": 0.6474623514496333,
      "learning_rate": 3.3939254449929392e-06,
      "loss": 0.019,
      "step": 11532
    },
    {
      "epoch": 7.378758797184901,
      "grad_norm": 0.7402414806252755,
      "learning_rate": 3.392369955556337e-06,
      "loss": 0.0219,
      "step": 11533
    },
    {
      "epoch": 7.379398592450416,
      "grad_norm": 0.6071962208223705,
      "learning_rate": 3.390814749839615e-06,
      "loss": 0.0217,
      "step": 11534
    },
    {
      "epoch": 7.380038387715931,
      "grad_norm": 0.7512517106865126,
      "learning_rate": 3.3892598279095477e-06,
      "loss": 0.0237,
      "step": 11535
    },
    {
      "epoch": 7.380678182981446,
      "grad_norm": 0.6909842632794598,
      "learning_rate": 3.3877051898329073e-06,
      "loss": 0.0229,
      "step": 11536
    },
    {
      "epoch": 7.381317978246961,
      "grad_norm": 0.6577789787486668,
      "learning_rate": 3.3861508356764426e-06,
      "loss": 0.022,
      "step": 11537
    },
    {
      "epoch": 7.381957773512476,
      "grad_norm": 0.6085437984633366,
      "learning_rate": 3.3845967655068935e-06,
      "loss": 0.0207,
      "step": 11538
    },
    {
      "epoch": 7.3825975687779914,
      "grad_norm": 0.7213917250954868,
      "learning_rate": 3.383042979390995e-06,
      "loss": 0.0236,
      "step": 11539
    },
    {
      "epoch": 7.383237364043506,
      "grad_norm": 0.5670147501299656,
      "learning_rate": 3.3814894773954598e-06,
      "loss": 0.021,
      "step": 11540
    },
    {
      "epoch": 7.383877159309021,
      "grad_norm": 0.6513294176834172,
      "learning_rate": 3.37993625958699e-06,
      "loss": 0.0218,
      "step": 11541
    },
    {
      "epoch": 7.384516954574536,
      "grad_norm": 0.664602529904669,
      "learning_rate": 3.3783833260322842e-06,
      "loss": 0.0236,
      "step": 11542
    },
    {
      "epoch": 7.385156749840051,
      "grad_norm": 0.7057034825678185,
      "learning_rate": 3.3768306767980184e-06,
      "loss": 0.0228,
      "step": 11543
    },
    {
      "epoch": 7.385796545105566,
      "grad_norm": 0.6865000889013069,
      "learning_rate": 3.3752783119508626e-06,
      "loss": 0.0225,
      "step": 11544
    },
    {
      "epoch": 7.386436340371081,
      "grad_norm": 0.6914577066227758,
      "learning_rate": 3.3737262315574683e-06,
      "loss": 0.0252,
      "step": 11545
    },
    {
      "epoch": 7.387076135636597,
      "grad_norm": 0.6995353387775556,
      "learning_rate": 3.3721744356844854e-06,
      "loss": 0.0252,
      "step": 11546
    },
    {
      "epoch": 7.387715930902111,
      "grad_norm": 0.760875009102138,
      "learning_rate": 3.3706229243985435e-06,
      "loss": 0.0245,
      "step": 11547
    },
    {
      "epoch": 7.388355726167626,
      "grad_norm": 0.7258639204545164,
      "learning_rate": 3.3690716977662584e-06,
      "loss": 0.0273,
      "step": 11548
    },
    {
      "epoch": 7.3889955214331415,
      "grad_norm": 0.5678170055252171,
      "learning_rate": 3.367520755854241e-06,
      "loss": 0.0187,
      "step": 11549
    },
    {
      "epoch": 7.389635316698657,
      "grad_norm": 0.6091971862379932,
      "learning_rate": 3.3659700987290834e-06,
      "loss": 0.021,
      "step": 11550
    },
    {
      "epoch": 7.390275111964171,
      "grad_norm": 0.6312212912515189,
      "learning_rate": 3.3644197264573687e-06,
      "loss": 0.0227,
      "step": 11551
    },
    {
      "epoch": 7.3909149072296865,
      "grad_norm": 0.7159471421422978,
      "learning_rate": 3.3628696391056637e-06,
      "loss": 0.0256,
      "step": 11552
    },
    {
      "epoch": 7.391554702495202,
      "grad_norm": 0.6341811409650585,
      "learning_rate": 3.3613198367405332e-06,
      "loss": 0.022,
      "step": 11553
    },
    {
      "epoch": 7.392194497760716,
      "grad_norm": 0.5867179303740339,
      "learning_rate": 3.359770319428519e-06,
      "loss": 0.0202,
      "step": 11554
    },
    {
      "epoch": 7.392834293026231,
      "grad_norm": 0.634131412947477,
      "learning_rate": 3.358221087236153e-06,
      "loss": 0.0241,
      "step": 11555
    },
    {
      "epoch": 7.393474088291747,
      "grad_norm": 0.7414036077017492,
      "learning_rate": 3.356672140229962e-06,
      "loss": 0.0243,
      "step": 11556
    },
    {
      "epoch": 7.394113883557262,
      "grad_norm": 0.7816827457575409,
      "learning_rate": 3.3551234784764506e-06,
      "loss": 0.0302,
      "step": 11557
    },
    {
      "epoch": 7.394753678822776,
      "grad_norm": 0.7013781707583451,
      "learning_rate": 3.353575102042115e-06,
      "loss": 0.0237,
      "step": 11558
    },
    {
      "epoch": 7.395393474088292,
      "grad_norm": 0.6076383090565068,
      "learning_rate": 3.352027010993443e-06,
      "loss": 0.02,
      "step": 11559
    },
    {
      "epoch": 7.396033269353807,
      "grad_norm": 0.5624593049646228,
      "learning_rate": 3.350479205396906e-06,
      "loss": 0.0199,
      "step": 11560
    },
    {
      "epoch": 7.396673064619322,
      "grad_norm": 0.6520275984567947,
      "learning_rate": 3.3489316853189637e-06,
      "loss": 0.0217,
      "step": 11561
    },
    {
      "epoch": 7.3973128598848366,
      "grad_norm": 0.6053438800905309,
      "learning_rate": 3.3473844508260634e-06,
      "loss": 0.0237,
      "step": 11562
    },
    {
      "epoch": 7.397952655150352,
      "grad_norm": 0.7161717939853698,
      "learning_rate": 3.3458375019846413e-06,
      "loss": 0.0245,
      "step": 11563
    },
    {
      "epoch": 7.398592450415867,
      "grad_norm": 0.7026655725305276,
      "learning_rate": 3.3442908388611186e-06,
      "loss": 0.0259,
      "step": 11564
    },
    {
      "epoch": 7.399232245681382,
      "grad_norm": 0.7793855949161568,
      "learning_rate": 3.3427444615219053e-06,
      "loss": 0.0241,
      "step": 11565
    },
    {
      "epoch": 7.399872040946897,
      "grad_norm": 0.6627302936400679,
      "learning_rate": 3.3411983700334062e-06,
      "loss": 0.0229,
      "step": 11566
    },
    {
      "epoch": 7.400511836212412,
      "grad_norm": 0.6390738650552797,
      "learning_rate": 3.3396525644620025e-06,
      "loss": 0.0249,
      "step": 11567
    },
    {
      "epoch": 7.401151631477927,
      "grad_norm": 0.6834207834606034,
      "learning_rate": 3.338107044874068e-06,
      "loss": 0.023,
      "step": 11568
    },
    {
      "epoch": 7.401791426743442,
      "grad_norm": 0.6524380670319987,
      "learning_rate": 3.336561811335969e-06,
      "loss": 0.0239,
      "step": 11569
    },
    {
      "epoch": 7.402431222008957,
      "grad_norm": 0.7125329791754427,
      "learning_rate": 3.3350168639140524e-06,
      "loss": 0.0245,
      "step": 11570
    },
    {
      "epoch": 7.403071017274472,
      "grad_norm": 0.7258587218629852,
      "learning_rate": 3.333472202674656e-06,
      "loss": 0.0263,
      "step": 11571
    },
    {
      "epoch": 7.4037108125399875,
      "grad_norm": 0.6169198441538868,
      "learning_rate": 3.3319278276841016e-06,
      "loss": 0.0216,
      "step": 11572
    },
    {
      "epoch": 7.404350607805502,
      "grad_norm": 0.6950187567391177,
      "learning_rate": 3.330383739008707e-06,
      "loss": 0.0226,
      "step": 11573
    },
    {
      "epoch": 7.404990403071017,
      "grad_norm": 0.7451412756863774,
      "learning_rate": 3.3288399367147707e-06,
      "loss": 0.0226,
      "step": 11574
    },
    {
      "epoch": 7.4056301983365325,
      "grad_norm": 0.806909338083968,
      "learning_rate": 3.3272964208685764e-06,
      "loss": 0.0256,
      "step": 11575
    },
    {
      "epoch": 7.406269993602048,
      "grad_norm": 0.6379429557281503,
      "learning_rate": 3.3257531915364083e-06,
      "loss": 0.0236,
      "step": 11576
    },
    {
      "epoch": 7.406909788867562,
      "grad_norm": 0.7331422110258581,
      "learning_rate": 3.324210248784525e-06,
      "loss": 0.024,
      "step": 11577
    },
    {
      "epoch": 7.407549584133077,
      "grad_norm": 0.6723988245377214,
      "learning_rate": 3.3226675926791795e-06,
      "loss": 0.0248,
      "step": 11578
    },
    {
      "epoch": 7.408189379398593,
      "grad_norm": 0.6515885919783689,
      "learning_rate": 3.3211252232866075e-06,
      "loss": 0.0269,
      "step": 11579
    },
    {
      "epoch": 7.408829174664108,
      "grad_norm": 0.5746079098642078,
      "learning_rate": 3.3195831406730394e-06,
      "loss": 0.0187,
      "step": 11580
    },
    {
      "epoch": 7.409468969929622,
      "grad_norm": 0.6900127345965426,
      "learning_rate": 3.3180413449046865e-06,
      "loss": 0.0247,
      "step": 11581
    },
    {
      "epoch": 7.410108765195138,
      "grad_norm": 0.7177630775582393,
      "learning_rate": 3.3164998360477495e-06,
      "loss": 0.0252,
      "step": 11582
    },
    {
      "epoch": 7.410748560460653,
      "grad_norm": 0.6709601469511968,
      "learning_rate": 3.3149586141684232e-06,
      "loss": 0.0255,
      "step": 11583
    },
    {
      "epoch": 7.411388355726167,
      "grad_norm": 0.6977401292697661,
      "learning_rate": 3.3134176793328833e-06,
      "loss": 0.0242,
      "step": 11584
    },
    {
      "epoch": 7.4120281509916826,
      "grad_norm": 0.6637405038350412,
      "learning_rate": 3.311877031607289e-06,
      "loss": 0.0251,
      "step": 11585
    },
    {
      "epoch": 7.412667946257198,
      "grad_norm": 0.7079648990403202,
      "learning_rate": 3.3103366710578024e-06,
      "loss": 0.0265,
      "step": 11586
    },
    {
      "epoch": 7.413307741522713,
      "grad_norm": 0.6478357758025529,
      "learning_rate": 3.308796597750558e-06,
      "loss": 0.0252,
      "step": 11587
    },
    {
      "epoch": 7.4139475367882275,
      "grad_norm": 0.6762273251140041,
      "learning_rate": 3.3072568117516822e-06,
      "loss": 0.022,
      "step": 11588
    },
    {
      "epoch": 7.414587332053743,
      "grad_norm": 0.6415718114390058,
      "learning_rate": 3.3057173131272967e-06,
      "loss": 0.0239,
      "step": 11589
    },
    {
      "epoch": 7.415227127319258,
      "grad_norm": 0.6934542452156602,
      "learning_rate": 3.3041781019435017e-06,
      "loss": 0.0235,
      "step": 11590
    },
    {
      "epoch": 7.415866922584772,
      "grad_norm": 0.8442267403587743,
      "learning_rate": 3.3026391782663867e-06,
      "loss": 0.0285,
      "step": 11591
    },
    {
      "epoch": 7.416506717850288,
      "grad_norm": 0.6589750875268906,
      "learning_rate": 3.3011005421620333e-06,
      "loss": 0.0227,
      "step": 11592
    },
    {
      "epoch": 7.417146513115803,
      "grad_norm": 0.7849939378316252,
      "learning_rate": 3.299562193696506e-06,
      "loss": 0.0248,
      "step": 11593
    },
    {
      "epoch": 7.417786308381318,
      "grad_norm": 0.6886093696334585,
      "learning_rate": 3.2980241329358586e-06,
      "loss": 0.025,
      "step": 11594
    },
    {
      "epoch": 7.418426103646833,
      "grad_norm": 0.722494174237689,
      "learning_rate": 3.2964863599461306e-06,
      "loss": 0.0236,
      "step": 11595
    },
    {
      "epoch": 7.419065898912348,
      "grad_norm": 0.8491577451959108,
      "learning_rate": 3.2949488747933557e-06,
      "loss": 0.0243,
      "step": 11596
    },
    {
      "epoch": 7.419705694177863,
      "grad_norm": 0.6247761310133806,
      "learning_rate": 3.2934116775435487e-06,
      "loss": 0.0239,
      "step": 11597
    },
    {
      "epoch": 7.4203454894433785,
      "grad_norm": 0.6412504408139517,
      "learning_rate": 3.2918747682627116e-06,
      "loss": 0.0236,
      "step": 11598
    },
    {
      "epoch": 7.420985284708893,
      "grad_norm": 0.7377094018044146,
      "learning_rate": 3.290338147016843e-06,
      "loss": 0.0259,
      "step": 11599
    },
    {
      "epoch": 7.421625079974408,
      "grad_norm": 0.7302513190934111,
      "learning_rate": 3.288801813871918e-06,
      "loss": 0.0265,
      "step": 11600
    },
    {
      "epoch": 7.422264875239923,
      "grad_norm": 0.598943659589163,
      "learning_rate": 3.2872657688939045e-06,
      "loss": 0.0192,
      "step": 11601
    },
    {
      "epoch": 7.422904670505439,
      "grad_norm": 0.699161723354199,
      "learning_rate": 3.2857300121487557e-06,
      "loss": 0.0247,
      "step": 11602
    },
    {
      "epoch": 7.423544465770953,
      "grad_norm": 0.5498170226015882,
      "learning_rate": 3.284194543702418e-06,
      "loss": 0.0193,
      "step": 11603
    },
    {
      "epoch": 7.424184261036468,
      "grad_norm": 0.6597549787955319,
      "learning_rate": 3.2826593636208214e-06,
      "loss": 0.0206,
      "step": 11604
    },
    {
      "epoch": 7.424824056301984,
      "grad_norm": 0.660745950960408,
      "learning_rate": 3.2811244719698787e-06,
      "loss": 0.0207,
      "step": 11605
    },
    {
      "epoch": 7.425463851567498,
      "grad_norm": 0.6171403459565978,
      "learning_rate": 3.2795898688155026e-06,
      "loss": 0.0238,
      "step": 11606
    },
    {
      "epoch": 7.426103646833013,
      "grad_norm": 0.6618111805260978,
      "learning_rate": 3.2780555542235836e-06,
      "loss": 0.0212,
      "step": 11607
    },
    {
      "epoch": 7.4267434420985285,
      "grad_norm": 0.6949389605499616,
      "learning_rate": 3.2765215282600006e-06,
      "loss": 0.0225,
      "step": 11608
    },
    {
      "epoch": 7.427383237364044,
      "grad_norm": 0.7851187342770167,
      "learning_rate": 3.2749877909906236e-06,
      "loss": 0.0275,
      "step": 11609
    },
    {
      "epoch": 7.428023032629558,
      "grad_norm": 0.7217472937178432,
      "learning_rate": 3.273454342481308e-06,
      "loss": 0.0263,
      "step": 11610
    },
    {
      "epoch": 7.4286628278950735,
      "grad_norm": 0.6638435470850589,
      "learning_rate": 3.271921182797897e-06,
      "loss": 0.0215,
      "step": 11611
    },
    {
      "epoch": 7.429302623160589,
      "grad_norm": 0.6508509521934932,
      "learning_rate": 3.2703883120062198e-06,
      "loss": 0.021,
      "step": 11612
    },
    {
      "epoch": 7.429942418426104,
      "grad_norm": 0.7373532134019278,
      "learning_rate": 3.268855730172099e-06,
      "loss": 0.0248,
      "step": 11613
    },
    {
      "epoch": 7.430582213691618,
      "grad_norm": 0.6415646670644195,
      "learning_rate": 3.2673234373613407e-06,
      "loss": 0.021,
      "step": 11614
    },
    {
      "epoch": 7.431222008957134,
      "grad_norm": 0.640346137524866,
      "learning_rate": 3.2657914336397334e-06,
      "loss": 0.0228,
      "step": 11615
    },
    {
      "epoch": 7.431861804222649,
      "grad_norm": 0.6372850006427868,
      "learning_rate": 3.2642597190730662e-06,
      "loss": 0.0211,
      "step": 11616
    },
    {
      "epoch": 7.432501599488164,
      "grad_norm": 0.7380211409393095,
      "learning_rate": 3.2627282937271055e-06,
      "loss": 0.0205,
      "step": 11617
    },
    {
      "epoch": 7.433141394753679,
      "grad_norm": 0.6570645184309685,
      "learning_rate": 3.261197157667604e-06,
      "loss": 0.0238,
      "step": 11618
    },
    {
      "epoch": 7.433781190019194,
      "grad_norm": 0.5976493447423987,
      "learning_rate": 3.259666310960311e-06,
      "loss": 0.0194,
      "step": 11619
    },
    {
      "epoch": 7.434420985284709,
      "grad_norm": 0.7098367002575372,
      "learning_rate": 3.258135753670957e-06,
      "loss": 0.0252,
      "step": 11620
    },
    {
      "epoch": 7.435060780550224,
      "grad_norm": 0.6610037085587694,
      "learning_rate": 3.2566054858652617e-06,
      "loss": 0.0221,
      "step": 11621
    },
    {
      "epoch": 7.435700575815739,
      "grad_norm": 0.6782150273592515,
      "learning_rate": 3.25507550760893e-06,
      "loss": 0.0241,
      "step": 11622
    },
    {
      "epoch": 7.436340371081254,
      "grad_norm": 0.7907848003714211,
      "learning_rate": 3.2535458189676583e-06,
      "loss": 0.0254,
      "step": 11623
    },
    {
      "epoch": 7.436980166346769,
      "grad_norm": 0.5755290337957092,
      "learning_rate": 3.2520164200071293e-06,
      "loss": 0.0197,
      "step": 11624
    },
    {
      "epoch": 7.437619961612284,
      "grad_norm": 0.6844600614974279,
      "learning_rate": 3.250487310793007e-06,
      "loss": 0.025,
      "step": 11625
    },
    {
      "epoch": 7.438259756877799,
      "grad_norm": 0.6680623458899864,
      "learning_rate": 3.2489584913909566e-06,
      "loss": 0.0206,
      "step": 11626
    },
    {
      "epoch": 7.438899552143314,
      "grad_norm": 0.6888907762735075,
      "learning_rate": 3.2474299618666195e-06,
      "loss": 0.0225,
      "step": 11627
    },
    {
      "epoch": 7.439539347408829,
      "grad_norm": 0.7340644147031004,
      "learning_rate": 3.245901722285627e-06,
      "loss": 0.0217,
      "step": 11628
    },
    {
      "epoch": 7.440179142674344,
      "grad_norm": 0.809342038866065,
      "learning_rate": 3.244373772713598e-06,
      "loss": 0.0244,
      "step": 11629
    },
    {
      "epoch": 7.440818937939859,
      "grad_norm": 0.6690994352602674,
      "learning_rate": 3.2428461132161437e-06,
      "loss": 0.0242,
      "step": 11630
    },
    {
      "epoch": 7.4414587332053745,
      "grad_norm": 0.7613408806905945,
      "learning_rate": 3.241318743858859e-06,
      "loss": 0.0257,
      "step": 11631
    },
    {
      "epoch": 7.442098528470889,
      "grad_norm": 0.722661518823592,
      "learning_rate": 3.23979166470732e-06,
      "loss": 0.028,
      "step": 11632
    },
    {
      "epoch": 7.442738323736404,
      "grad_norm": 0.6456456992482186,
      "learning_rate": 3.238264875827105e-06,
      "loss": 0.0196,
      "step": 11633
    },
    {
      "epoch": 7.4433781190019195,
      "grad_norm": 0.7328861959351215,
      "learning_rate": 3.2367383772837692e-06,
      "loss": 0.0276,
      "step": 11634
    },
    {
      "epoch": 7.444017914267435,
      "grad_norm": 0.725588249801463,
      "learning_rate": 3.2352121691428516e-06,
      "loss": 0.0256,
      "step": 11635
    },
    {
      "epoch": 7.444657709532949,
      "grad_norm": 0.6915963913710557,
      "learning_rate": 3.2336862514698943e-06,
      "loss": 0.0269,
      "step": 11636
    },
    {
      "epoch": 7.445297504798464,
      "grad_norm": 0.6176179244267299,
      "learning_rate": 3.2321606243304128e-06,
      "loss": 0.0221,
      "step": 11637
    },
    {
      "epoch": 7.44593730006398,
      "grad_norm": 0.6075221794259591,
      "learning_rate": 3.230635287789915e-06,
      "loss": 0.0197,
      "step": 11638
    },
    {
      "epoch": 7.446577095329495,
      "grad_norm": 0.6808958370915305,
      "learning_rate": 3.2291102419138964e-06,
      "loss": 0.0221,
      "step": 11639
    },
    {
      "epoch": 7.447216890595009,
      "grad_norm": 0.7595684554632709,
      "learning_rate": 3.227585486767839e-06,
      "loss": 0.0238,
      "step": 11640
    },
    {
      "epoch": 7.447856685860525,
      "grad_norm": 0.7457005014324554,
      "learning_rate": 3.226061022417214e-06,
      "loss": 0.0288,
      "step": 11641
    },
    {
      "epoch": 7.44849648112604,
      "grad_norm": 0.7316341298237974,
      "learning_rate": 3.2245368489274753e-06,
      "loss": 0.0269,
      "step": 11642
    },
    {
      "epoch": 7.449136276391554,
      "grad_norm": 0.5648806117759937,
      "learning_rate": 3.223012966364074e-06,
      "loss": 0.0204,
      "step": 11643
    },
    {
      "epoch": 7.44977607165707,
      "grad_norm": 0.6528959489070018,
      "learning_rate": 3.2214893747924423e-06,
      "loss": 0.0244,
      "step": 11644
    },
    {
      "epoch": 7.450415866922585,
      "grad_norm": 0.6368890186706959,
      "learning_rate": 3.219966074277995e-06,
      "loss": 0.0251,
      "step": 11645
    },
    {
      "epoch": 7.4510556621881,
      "grad_norm": 0.6749818465859713,
      "learning_rate": 3.2184430648861463e-06,
      "loss": 0.0246,
      "step": 11646
    },
    {
      "epoch": 7.4516954574536145,
      "grad_norm": 0.616300631166614,
      "learning_rate": 3.2169203466822897e-06,
      "loss": 0.0211,
      "step": 11647
    },
    {
      "epoch": 7.45233525271913,
      "grad_norm": 0.7201604085476826,
      "learning_rate": 3.2153979197318076e-06,
      "loss": 0.0194,
      "step": 11648
    },
    {
      "epoch": 7.452975047984645,
      "grad_norm": 0.6398996547402651,
      "learning_rate": 3.2138757841000656e-06,
      "loss": 0.0232,
      "step": 11649
    },
    {
      "epoch": 7.45361484325016,
      "grad_norm": 0.648160886074385,
      "learning_rate": 3.2123539398524296e-06,
      "loss": 0.0205,
      "step": 11650
    },
    {
      "epoch": 7.454254638515675,
      "grad_norm": 0.6177402599605666,
      "learning_rate": 3.210832387054241e-06,
      "loss": 0.0191,
      "step": 11651
    },
    {
      "epoch": 7.45489443378119,
      "grad_norm": 0.6462235031568647,
      "learning_rate": 3.209311125770833e-06,
      "loss": 0.0243,
      "step": 11652
    },
    {
      "epoch": 7.455534229046705,
      "grad_norm": 0.5837322077376537,
      "learning_rate": 3.207790156067525e-06,
      "loss": 0.0189,
      "step": 11653
    },
    {
      "epoch": 7.4561740243122205,
      "grad_norm": 0.6842771620500766,
      "learning_rate": 3.206269478009626e-06,
      "loss": 0.0202,
      "step": 11654
    },
    {
      "epoch": 7.456813819577735,
      "grad_norm": 0.6767770945918163,
      "learning_rate": 3.204749091662427e-06,
      "loss": 0.0252,
      "step": 11655
    },
    {
      "epoch": 7.45745361484325,
      "grad_norm": 0.6595849288096167,
      "learning_rate": 3.203228997091218e-06,
      "loss": 0.0255,
      "step": 11656
    },
    {
      "epoch": 7.4580934101087655,
      "grad_norm": 0.6420429283346548,
      "learning_rate": 3.2017091943612645e-06,
      "loss": 0.0207,
      "step": 11657
    },
    {
      "epoch": 7.45873320537428,
      "grad_norm": 0.6495866985561736,
      "learning_rate": 3.2001896835378255e-06,
      "loss": 0.0216,
      "step": 11658
    },
    {
      "epoch": 7.459373000639795,
      "grad_norm": 0.678721163979012,
      "learning_rate": 3.198670464686141e-06,
      "loss": 0.0225,
      "step": 11659
    },
    {
      "epoch": 7.46001279590531,
      "grad_norm": 0.6200740737909304,
      "learning_rate": 3.1971515378714537e-06,
      "loss": 0.0168,
      "step": 11660
    },
    {
      "epoch": 7.460652591170826,
      "grad_norm": 0.6410894311685389,
      "learning_rate": 3.195632903158976e-06,
      "loss": 0.0258,
      "step": 11661
    },
    {
      "epoch": 7.46129238643634,
      "grad_norm": 0.6664254547465651,
      "learning_rate": 3.1941145606139146e-06,
      "loss": 0.0213,
      "step": 11662
    },
    {
      "epoch": 7.461932181701855,
      "grad_norm": 0.7083504727474584,
      "learning_rate": 3.1925965103014713e-06,
      "loss": 0.0222,
      "step": 11663
    },
    {
      "epoch": 7.462571976967371,
      "grad_norm": 0.7576892574035697,
      "learning_rate": 3.191078752286824e-06,
      "loss": 0.027,
      "step": 11664
    },
    {
      "epoch": 7.463211772232885,
      "grad_norm": 0.7220546481567913,
      "learning_rate": 3.189561286635139e-06,
      "loss": 0.0253,
      "step": 11665
    },
    {
      "epoch": 7.4638515674984,
      "grad_norm": 0.7971396488925419,
      "learning_rate": 3.1880441134115814e-06,
      "loss": 0.0275,
      "step": 11666
    },
    {
      "epoch": 7.464491362763916,
      "grad_norm": 0.6522299958492257,
      "learning_rate": 3.1865272326812923e-06,
      "loss": 0.0229,
      "step": 11667
    },
    {
      "epoch": 7.465131158029431,
      "grad_norm": 0.8039947795184382,
      "learning_rate": 3.185010644509403e-06,
      "loss": 0.0282,
      "step": 11668
    },
    {
      "epoch": 7.465770953294945,
      "grad_norm": 0.8272047545394012,
      "learning_rate": 3.1834943489610346e-06,
      "loss": 0.024,
      "step": 11669
    },
    {
      "epoch": 7.4664107485604605,
      "grad_norm": 0.6750380229315468,
      "learning_rate": 3.1819783461012922e-06,
      "loss": 0.0232,
      "step": 11670
    },
    {
      "epoch": 7.467050543825976,
      "grad_norm": 0.7632379029668448,
      "learning_rate": 3.1804626359952717e-06,
      "loss": 0.0234,
      "step": 11671
    },
    {
      "epoch": 7.467690339091491,
      "grad_norm": 0.7469993469604409,
      "learning_rate": 3.1789472187080515e-06,
      "loss": 0.0251,
      "step": 11672
    },
    {
      "epoch": 7.468330134357005,
      "grad_norm": 0.7440437343149295,
      "learning_rate": 3.177432094304708e-06,
      "loss": 0.0237,
      "step": 11673
    },
    {
      "epoch": 7.468969929622521,
      "grad_norm": 0.6115364463961971,
      "learning_rate": 3.175917262850292e-06,
      "loss": 0.0224,
      "step": 11674
    },
    {
      "epoch": 7.469609724888036,
      "grad_norm": 0.7027619072338397,
      "learning_rate": 3.1744027244098485e-06,
      "loss": 0.0231,
      "step": 11675
    },
    {
      "epoch": 7.470249520153551,
      "grad_norm": 0.7002110089882474,
      "learning_rate": 3.1728884790484117e-06,
      "loss": 0.0216,
      "step": 11676
    },
    {
      "epoch": 7.470889315419066,
      "grad_norm": 0.8286742693687579,
      "learning_rate": 3.1713745268310003e-06,
      "loss": 0.0289,
      "step": 11677
    },
    {
      "epoch": 7.471529110684581,
      "grad_norm": 0.6849892498175727,
      "learning_rate": 3.1698608678226193e-06,
      "loss": 0.0243,
      "step": 11678
    },
    {
      "epoch": 7.472168905950096,
      "grad_norm": 0.7432641509114731,
      "learning_rate": 3.168347502088259e-06,
      "loss": 0.0246,
      "step": 11679
    },
    {
      "epoch": 7.472808701215611,
      "grad_norm": 0.8043127785296412,
      "learning_rate": 3.166834429692908e-06,
      "loss": 0.0262,
      "step": 11680
    },
    {
      "epoch": 7.473448496481126,
      "grad_norm": 0.7680753172724077,
      "learning_rate": 3.1653216507015317e-06,
      "loss": 0.0263,
      "step": 11681
    },
    {
      "epoch": 7.474088291746641,
      "grad_norm": 0.7370808695000691,
      "learning_rate": 3.1638091651790846e-06,
      "loss": 0.023,
      "step": 11682
    },
    {
      "epoch": 7.474728087012156,
      "grad_norm": 0.6923468320575195,
      "learning_rate": 3.162296973190513e-06,
      "loss": 0.0246,
      "step": 11683
    },
    {
      "epoch": 7.475367882277671,
      "grad_norm": 0.5632706293045896,
      "learning_rate": 3.1607850748007453e-06,
      "loss": 0.0189,
      "step": 11684
    },
    {
      "epoch": 7.476007677543186,
      "grad_norm": 0.6985299713551484,
      "learning_rate": 3.159273470074696e-06,
      "loss": 0.0225,
      "step": 11685
    },
    {
      "epoch": 7.476647472808701,
      "grad_norm": 0.6903522290684881,
      "learning_rate": 3.15776215907728e-06,
      "loss": 0.0249,
      "step": 11686
    },
    {
      "epoch": 7.477287268074217,
      "grad_norm": 0.6295148540484932,
      "learning_rate": 3.1562511418733853e-06,
      "loss": 0.0228,
      "step": 11687
    },
    {
      "epoch": 7.477927063339731,
      "grad_norm": 0.7331348338399607,
      "learning_rate": 3.1547404185278916e-06,
      "loss": 0.0239,
      "step": 11688
    },
    {
      "epoch": 7.478566858605246,
      "grad_norm": 0.8670108771863264,
      "learning_rate": 3.1532299891056652e-06,
      "loss": 0.0292,
      "step": 11689
    },
    {
      "epoch": 7.479206653870762,
      "grad_norm": 0.9302393960150422,
      "learning_rate": 3.1517198536715664e-06,
      "loss": 0.026,
      "step": 11690
    },
    {
      "epoch": 7.479846449136277,
      "grad_norm": 0.8083211778786598,
      "learning_rate": 3.1502100122904356e-06,
      "loss": 0.0262,
      "step": 11691
    },
    {
      "epoch": 7.480486244401791,
      "grad_norm": 0.667572113659502,
      "learning_rate": 3.148700465027098e-06,
      "loss": 0.0249,
      "step": 11692
    },
    {
      "epoch": 7.4811260396673065,
      "grad_norm": 0.6821684080816113,
      "learning_rate": 3.1471912119463777e-06,
      "loss": 0.0212,
      "step": 11693
    },
    {
      "epoch": 7.481765834932822,
      "grad_norm": 0.6799814934637429,
      "learning_rate": 3.145682253113078e-06,
      "loss": 0.0249,
      "step": 11694
    },
    {
      "epoch": 7.482405630198336,
      "grad_norm": 0.5903707135542126,
      "learning_rate": 3.1441735885919856e-06,
      "loss": 0.0193,
      "step": 11695
    },
    {
      "epoch": 7.483045425463851,
      "grad_norm": 0.7493703393717386,
      "learning_rate": 3.1426652184478867e-06,
      "loss": 0.0189,
      "step": 11696
    },
    {
      "epoch": 7.483685220729367,
      "grad_norm": 0.5436368851858052,
      "learning_rate": 3.1411571427455455e-06,
      "loss": 0.015,
      "step": 11697
    },
    {
      "epoch": 7.484325015994882,
      "grad_norm": 0.667089202516436,
      "learning_rate": 3.139649361549715e-06,
      "loss": 0.0214,
      "step": 11698
    },
    {
      "epoch": 7.484964811260396,
      "grad_norm": 0.7935827313011515,
      "learning_rate": 3.138141874925138e-06,
      "loss": 0.0279,
      "step": 11699
    },
    {
      "epoch": 7.485604606525912,
      "grad_norm": 0.8355171894975721,
      "learning_rate": 3.1366346829365423e-06,
      "loss": 0.0272,
      "step": 11700
    },
    {
      "epoch": 7.486244401791427,
      "grad_norm": 0.6450561446893582,
      "learning_rate": 3.135127785648644e-06,
      "loss": 0.0204,
      "step": 11701
    },
    {
      "epoch": 7.486884197056942,
      "grad_norm": 0.7081311792364711,
      "learning_rate": 3.133621183126144e-06,
      "loss": 0.0238,
      "step": 11702
    },
    {
      "epoch": 7.487523992322457,
      "grad_norm": 0.5908018067922876,
      "learning_rate": 3.1321148754337392e-06,
      "loss": 0.0196,
      "step": 11703
    },
    {
      "epoch": 7.488163787587972,
      "grad_norm": 0.6713494704751595,
      "learning_rate": 3.130608862636105e-06,
      "loss": 0.0239,
      "step": 11704
    },
    {
      "epoch": 7.488803582853487,
      "grad_norm": 0.7111685087298035,
      "learning_rate": 3.129103144797906e-06,
      "loss": 0.0269,
      "step": 11705
    },
    {
      "epoch": 7.4894433781190015,
      "grad_norm": 0.6468834939387561,
      "learning_rate": 3.127597721983793e-06,
      "loss": 0.0187,
      "step": 11706
    },
    {
      "epoch": 7.490083173384517,
      "grad_norm": 0.7123927954811587,
      "learning_rate": 3.126092594258412e-06,
      "loss": 0.0232,
      "step": 11707
    },
    {
      "epoch": 7.490722968650032,
      "grad_norm": 0.677496340484206,
      "learning_rate": 3.124587761686386e-06,
      "loss": 0.0215,
      "step": 11708
    },
    {
      "epoch": 7.491362763915547,
      "grad_norm": 0.7103026854348993,
      "learning_rate": 3.1230832243323295e-06,
      "loss": 0.024,
      "step": 11709
    },
    {
      "epoch": 7.492002559181062,
      "grad_norm": 0.7173563577687553,
      "learning_rate": 3.121578982260849e-06,
      "loss": 0.0267,
      "step": 11710
    },
    {
      "epoch": 7.492642354446577,
      "grad_norm": 0.7470048738483127,
      "learning_rate": 3.1200750355365307e-06,
      "loss": 0.0246,
      "step": 11711
    },
    {
      "epoch": 7.493282149712092,
      "grad_norm": 0.7579050126022193,
      "learning_rate": 3.1185713842239517e-06,
      "loss": 0.0242,
      "step": 11712
    },
    {
      "epoch": 7.4939219449776076,
      "grad_norm": 0.7272691881269919,
      "learning_rate": 3.1170680283876775e-06,
      "loss": 0.0232,
      "step": 11713
    },
    {
      "epoch": 7.494561740243122,
      "grad_norm": 0.7118855207366008,
      "learning_rate": 3.1155649680922574e-06,
      "loss": 0.0247,
      "step": 11714
    },
    {
      "epoch": 7.495201535508637,
      "grad_norm": 0.7200859410906707,
      "learning_rate": 3.1140622034022316e-06,
      "loss": 0.0208,
      "step": 11715
    },
    {
      "epoch": 7.4958413307741525,
      "grad_norm": 0.7338463433850584,
      "learning_rate": 3.1125597343821223e-06,
      "loss": 0.0252,
      "step": 11716
    },
    {
      "epoch": 7.496481126039667,
      "grad_norm": 0.7110592136024082,
      "learning_rate": 3.111057561096449e-06,
      "loss": 0.0256,
      "step": 11717
    },
    {
      "epoch": 7.497120921305182,
      "grad_norm": 0.6342596323129114,
      "learning_rate": 3.1095556836097105e-06,
      "loss": 0.0233,
      "step": 11718
    },
    {
      "epoch": 7.497760716570697,
      "grad_norm": 0.796865444958724,
      "learning_rate": 3.1080541019863895e-06,
      "loss": 0.0278,
      "step": 11719
    },
    {
      "epoch": 7.498400511836213,
      "grad_norm": 0.7244415934460432,
      "learning_rate": 3.1065528162909698e-06,
      "loss": 0.0217,
      "step": 11720
    },
    {
      "epoch": 7.499040307101727,
      "grad_norm": 0.6010011227838281,
      "learning_rate": 3.105051826587909e-06,
      "loss": 0.0203,
      "step": 11721
    },
    {
      "epoch": 7.499680102367242,
      "grad_norm": 0.6838976066746245,
      "learning_rate": 3.1035511329416546e-06,
      "loss": 0.0231,
      "step": 11722
    },
    {
      "epoch": 7.500319897632758,
      "grad_norm": 0.6306882216185342,
      "learning_rate": 3.1020507354166496e-06,
      "loss": 0.0208,
      "step": 11723
    },
    {
      "epoch": 7.500959692898273,
      "grad_norm": 0.7070717602455894,
      "learning_rate": 3.100550634077316e-06,
      "loss": 0.0257,
      "step": 11724
    },
    {
      "epoch": 7.501599488163787,
      "grad_norm": 0.7300446693305627,
      "learning_rate": 3.099050828988065e-06,
      "loss": 0.0239,
      "step": 11725
    },
    {
      "epoch": 7.502239283429303,
      "grad_norm": 0.7250555806408747,
      "learning_rate": 3.0975513202132924e-06,
      "loss": 0.0196,
      "step": 11726
    },
    {
      "epoch": 7.502879078694818,
      "grad_norm": 0.6858680687161088,
      "learning_rate": 3.09605210781739e-06,
      "loss": 0.0227,
      "step": 11727
    },
    {
      "epoch": 7.503518873960333,
      "grad_norm": 0.6974813667304276,
      "learning_rate": 3.0945531918647297e-06,
      "loss": 0.0243,
      "step": 11728
    },
    {
      "epoch": 7.5041586692258475,
      "grad_norm": 0.6220625851982879,
      "learning_rate": 3.0930545724196714e-06,
      "loss": 0.0221,
      "step": 11729
    },
    {
      "epoch": 7.504798464491363,
      "grad_norm": 0.8811734918594337,
      "learning_rate": 3.0915562495465634e-06,
      "loss": 0.0293,
      "step": 11730
    },
    {
      "epoch": 7.505438259756878,
      "grad_norm": 0.7196650367200449,
      "learning_rate": 3.090058223309741e-06,
      "loss": 0.0267,
      "step": 11731
    },
    {
      "epoch": 7.5060780550223924,
      "grad_norm": 0.6561970187487192,
      "learning_rate": 3.0885604937735225e-06,
      "loss": 0.0237,
      "step": 11732
    },
    {
      "epoch": 7.506717850287908,
      "grad_norm": 0.6728404991627294,
      "learning_rate": 3.0870630610022258e-06,
      "loss": 0.0199,
      "step": 11733
    },
    {
      "epoch": 7.507357645553423,
      "grad_norm": 0.6375922953048845,
      "learning_rate": 3.0855659250601445e-06,
      "loss": 0.0214,
      "step": 11734
    },
    {
      "epoch": 7.507997440818938,
      "grad_norm": 0.6796206782081651,
      "learning_rate": 3.0840690860115617e-06,
      "loss": 0.0217,
      "step": 11735
    },
    {
      "epoch": 7.508637236084453,
      "grad_norm": 0.5905228661717745,
      "learning_rate": 3.0825725439207464e-06,
      "loss": 0.0191,
      "step": 11736
    },
    {
      "epoch": 7.509277031349968,
      "grad_norm": 0.6848116930877398,
      "learning_rate": 3.081076298851965e-06,
      "loss": 0.0211,
      "step": 11737
    },
    {
      "epoch": 7.509916826615483,
      "grad_norm": 0.6271356426169701,
      "learning_rate": 3.07958035086946e-06,
      "loss": 0.0218,
      "step": 11738
    },
    {
      "epoch": 7.510556621880998,
      "grad_norm": 0.7248858887033045,
      "learning_rate": 3.0780847000374602e-06,
      "loss": 0.0259,
      "step": 11739
    },
    {
      "epoch": 7.511196417146513,
      "grad_norm": 0.7360323753763413,
      "learning_rate": 3.0765893464201936e-06,
      "loss": 0.0253,
      "step": 11740
    },
    {
      "epoch": 7.511836212412028,
      "grad_norm": 0.7249184279280019,
      "learning_rate": 3.0750942900818647e-06,
      "loss": 0.0222,
      "step": 11741
    },
    {
      "epoch": 7.512476007677543,
      "grad_norm": 0.6104549643911386,
      "learning_rate": 3.0735995310866686e-06,
      "loss": 0.0216,
      "step": 11742
    },
    {
      "epoch": 7.513115802943059,
      "grad_norm": 0.7692321660173727,
      "learning_rate": 3.072105069498786e-06,
      "loss": 0.0241,
      "step": 11743
    },
    {
      "epoch": 7.513755598208573,
      "grad_norm": 0.7562489071007564,
      "learning_rate": 3.070610905382391e-06,
      "loss": 0.0243,
      "step": 11744
    },
    {
      "epoch": 7.514395393474088,
      "grad_norm": 0.7117444904265373,
      "learning_rate": 3.0691170388016347e-06,
      "loss": 0.0229,
      "step": 11745
    },
    {
      "epoch": 7.515035188739604,
      "grad_norm": 0.6061993543084002,
      "learning_rate": 3.067623469820662e-06,
      "loss": 0.0225,
      "step": 11746
    },
    {
      "epoch": 7.515674984005118,
      "grad_norm": 0.7678196491516996,
      "learning_rate": 3.0661301985036083e-06,
      "loss": 0.0246,
      "step": 11747
    },
    {
      "epoch": 7.516314779270633,
      "grad_norm": 0.7243874944295726,
      "learning_rate": 3.0646372249145895e-06,
      "loss": 0.0215,
      "step": 11748
    },
    {
      "epoch": 7.516954574536149,
      "grad_norm": 0.7094999992330586,
      "learning_rate": 3.063144549117708e-06,
      "loss": 0.0249,
      "step": 11749
    },
    {
      "epoch": 7.517594369801664,
      "grad_norm": 0.770355661905073,
      "learning_rate": 3.061652171177063e-06,
      "loss": 0.0293,
      "step": 11750
    },
    {
      "epoch": 7.518234165067178,
      "grad_norm": 0.6137244229102828,
      "learning_rate": 3.060160091156732e-06,
      "loss": 0.0209,
      "step": 11751
    },
    {
      "epoch": 7.5188739603326935,
      "grad_norm": 0.618290600173726,
      "learning_rate": 3.058668309120777e-06,
      "loss": 0.0191,
      "step": 11752
    },
    {
      "epoch": 7.519513755598209,
      "grad_norm": 0.5711306250013418,
      "learning_rate": 3.0571768251332622e-06,
      "loss": 0.018,
      "step": 11753
    },
    {
      "epoch": 7.520153550863723,
      "grad_norm": 0.6673949846816765,
      "learning_rate": 3.0556856392582234e-06,
      "loss": 0.0233,
      "step": 11754
    },
    {
      "epoch": 7.520793346129238,
      "grad_norm": 0.5823831514402393,
      "learning_rate": 3.0541947515596893e-06,
      "loss": 0.0211,
      "step": 11755
    },
    {
      "epoch": 7.521433141394754,
      "grad_norm": 0.7238700090757264,
      "learning_rate": 3.052704162101675e-06,
      "loss": 0.0277,
      "step": 11756
    },
    {
      "epoch": 7.522072936660269,
      "grad_norm": 0.6983933264939562,
      "learning_rate": 3.051213870948189e-06,
      "loss": 0.0247,
      "step": 11757
    },
    {
      "epoch": 7.522712731925783,
      "grad_norm": 0.6373714522764539,
      "learning_rate": 3.0497238781632175e-06,
      "loss": 0.0206,
      "step": 11758
    },
    {
      "epoch": 7.523352527191299,
      "grad_norm": 0.6560875350794916,
      "learning_rate": 3.04823418381074e-06,
      "loss": 0.0214,
      "step": 11759
    },
    {
      "epoch": 7.523992322456814,
      "grad_norm": 0.7148981005587555,
      "learning_rate": 3.0467447879547196e-06,
      "loss": 0.0243,
      "step": 11760
    },
    {
      "epoch": 7.524632117722329,
      "grad_norm": 0.7725652926428024,
      "learning_rate": 3.0452556906591092e-06,
      "loss": 0.0252,
      "step": 11761
    },
    {
      "epoch": 7.525271912987844,
      "grad_norm": 0.7244380908825665,
      "learning_rate": 3.043766891987845e-06,
      "loss": 0.025,
      "step": 11762
    },
    {
      "epoch": 7.525911708253359,
      "grad_norm": 0.6359399973840201,
      "learning_rate": 3.0422783920048603e-06,
      "loss": 0.0233,
      "step": 11763
    },
    {
      "epoch": 7.526551503518874,
      "grad_norm": 0.6049193063034491,
      "learning_rate": 3.0407901907740636e-06,
      "loss": 0.0201,
      "step": 11764
    },
    {
      "epoch": 7.527191298784389,
      "grad_norm": 0.5862700038662904,
      "learning_rate": 3.0393022883593572e-06,
      "loss": 0.0201,
      "step": 11765
    },
    {
      "epoch": 7.527831094049904,
      "grad_norm": 0.7775422884988233,
      "learning_rate": 3.037814684824625e-06,
      "loss": 0.0237,
      "step": 11766
    },
    {
      "epoch": 7.528470889315419,
      "grad_norm": 0.7317649661753559,
      "learning_rate": 3.036327380233748e-06,
      "loss": 0.0245,
      "step": 11767
    },
    {
      "epoch": 7.529110684580934,
      "grad_norm": 0.7334305380309301,
      "learning_rate": 3.034840374650586e-06,
      "loss": 0.0244,
      "step": 11768
    },
    {
      "epoch": 7.529750479846449,
      "grad_norm": 0.6377531744563559,
      "learning_rate": 3.0333536681389864e-06,
      "loss": 0.0218,
      "step": 11769
    },
    {
      "epoch": 7.530390275111964,
      "grad_norm": 0.6595161616316063,
      "learning_rate": 3.0318672607627896e-06,
      "loss": 0.0216,
      "step": 11770
    },
    {
      "epoch": 7.531030070377479,
      "grad_norm": 0.6089660791327445,
      "learning_rate": 3.0303811525858174e-06,
      "loss": 0.0197,
      "step": 11771
    },
    {
      "epoch": 7.531669865642995,
      "grad_norm": 0.6873481282530046,
      "learning_rate": 3.0288953436718806e-06,
      "loss": 0.0232,
      "step": 11772
    },
    {
      "epoch": 7.532309660908509,
      "grad_norm": 0.6363828256519624,
      "learning_rate": 3.027409834084777e-06,
      "loss": 0.0244,
      "step": 11773
    },
    {
      "epoch": 7.532949456174024,
      "grad_norm": 0.696828520435397,
      "learning_rate": 3.025924623888292e-06,
      "loss": 0.021,
      "step": 11774
    },
    {
      "epoch": 7.5335892514395395,
      "grad_norm": 0.6634464358337367,
      "learning_rate": 3.0244397131461976e-06,
      "loss": 0.0217,
      "step": 11775
    },
    {
      "epoch": 7.534229046705054,
      "grad_norm": 0.6728034475589431,
      "learning_rate": 3.02295510192225e-06,
      "loss": 0.0223,
      "step": 11776
    },
    {
      "epoch": 7.534868841970569,
      "grad_norm": 0.5646524659203142,
      "learning_rate": 3.021470790280203e-06,
      "loss": 0.0161,
      "step": 11777
    },
    {
      "epoch": 7.535508637236084,
      "grad_norm": 0.6467650129643309,
      "learning_rate": 3.0199867782837865e-06,
      "loss": 0.0228,
      "step": 11778
    },
    {
      "epoch": 7.5361484325016,
      "grad_norm": 0.8370321184143481,
      "learning_rate": 3.018503065996717e-06,
      "loss": 0.0231,
      "step": 11779
    },
    {
      "epoch": 7.536788227767115,
      "grad_norm": 0.7939476780437312,
      "learning_rate": 3.017019653482711e-06,
      "loss": 0.0283,
      "step": 11780
    },
    {
      "epoch": 7.537428023032629,
      "grad_norm": 0.6989820135893715,
      "learning_rate": 3.0155365408054595e-06,
      "loss": 0.0237,
      "step": 11781
    },
    {
      "epoch": 7.538067818298145,
      "grad_norm": 0.7023481125148362,
      "learning_rate": 3.014053728028643e-06,
      "loss": 0.0217,
      "step": 11782
    },
    {
      "epoch": 7.53870761356366,
      "grad_norm": 0.7644658121825949,
      "learning_rate": 3.0125712152159303e-06,
      "loss": 0.0229,
      "step": 11783
    },
    {
      "epoch": 7.539347408829174,
      "grad_norm": 0.6410151686321574,
      "learning_rate": 3.0110890024309835e-06,
      "loss": 0.0218,
      "step": 11784
    },
    {
      "epoch": 7.53998720409469,
      "grad_norm": 0.6900884870079526,
      "learning_rate": 3.009607089737442e-06,
      "loss": 0.025,
      "step": 11785
    },
    {
      "epoch": 7.540626999360205,
      "grad_norm": 0.7085334750657368,
      "learning_rate": 3.0081254771989333e-06,
      "loss": 0.026,
      "step": 11786
    },
    {
      "epoch": 7.54126679462572,
      "grad_norm": 0.6955009839663253,
      "learning_rate": 3.0066441648790832e-06,
      "loss": 0.0246,
      "step": 11787
    },
    {
      "epoch": 7.5419065898912345,
      "grad_norm": 0.608231125846817,
      "learning_rate": 3.0051631528414927e-06,
      "loss": 0.0201,
      "step": 11788
    },
    {
      "epoch": 7.54254638515675,
      "grad_norm": 0.6650826964525798,
      "learning_rate": 3.0036824411497523e-06,
      "loss": 0.0188,
      "step": 11789
    },
    {
      "epoch": 7.543186180422265,
      "grad_norm": 0.7404059114216136,
      "learning_rate": 3.0022020298674438e-06,
      "loss": 0.0228,
      "step": 11790
    },
    {
      "epoch": 7.5438259756877795,
      "grad_norm": 0.7492395430723641,
      "learning_rate": 3.0007219190581305e-06,
      "loss": 0.0265,
      "step": 11791
    },
    {
      "epoch": 7.544465770953295,
      "grad_norm": 0.6858401660665587,
      "learning_rate": 2.9992421087853684e-06,
      "loss": 0.0245,
      "step": 11792
    },
    {
      "epoch": 7.54510556621881,
      "grad_norm": 0.6325602525663384,
      "learning_rate": 2.9977625991126934e-06,
      "loss": 0.0208,
      "step": 11793
    },
    {
      "epoch": 7.545745361484325,
      "grad_norm": 0.6562995978434714,
      "learning_rate": 2.9962833901036392e-06,
      "loss": 0.0204,
      "step": 11794
    },
    {
      "epoch": 7.54638515674984,
      "grad_norm": 0.5984695853152844,
      "learning_rate": 2.9948044818217183e-06,
      "loss": 0.0221,
      "step": 11795
    },
    {
      "epoch": 7.547024952015355,
      "grad_norm": 0.7111448116898249,
      "learning_rate": 2.9933258743304282e-06,
      "loss": 0.0279,
      "step": 11796
    },
    {
      "epoch": 7.54766474728087,
      "grad_norm": 0.6043886572210198,
      "learning_rate": 2.9918475676932636e-06,
      "loss": 0.0172,
      "step": 11797
    },
    {
      "epoch": 7.5483045425463855,
      "grad_norm": 0.7073929669452675,
      "learning_rate": 2.9903695619736985e-06,
      "loss": 0.0214,
      "step": 11798
    },
    {
      "epoch": 7.5489443378119,
      "grad_norm": 0.6536188805079984,
      "learning_rate": 2.988891857235192e-06,
      "loss": 0.0208,
      "step": 11799
    },
    {
      "epoch": 7.549584133077415,
      "grad_norm": 0.785266094714605,
      "learning_rate": 2.987414453541201e-06,
      "loss": 0.0268,
      "step": 11800
    },
    {
      "epoch": 7.55022392834293,
      "grad_norm": 0.570113240315728,
      "learning_rate": 2.9859373509551582e-06,
      "loss": 0.0209,
      "step": 11801
    },
    {
      "epoch": 7.550863723608446,
      "grad_norm": 0.6694027462798394,
      "learning_rate": 2.9844605495404887e-06,
      "loss": 0.0226,
      "step": 11802
    },
    {
      "epoch": 7.55150351887396,
      "grad_norm": 0.7556744206003938,
      "learning_rate": 2.9829840493606032e-06,
      "loss": 0.0248,
      "step": 11803
    },
    {
      "epoch": 7.552143314139475,
      "grad_norm": 0.6791793828149151,
      "learning_rate": 2.9815078504789006e-06,
      "loss": 0.0208,
      "step": 11804
    },
    {
      "epoch": 7.552783109404991,
      "grad_norm": 0.7280140290509849,
      "learning_rate": 2.9800319529587673e-06,
      "loss": 0.0238,
      "step": 11805
    },
    {
      "epoch": 7.553422904670505,
      "grad_norm": 0.686194629698262,
      "learning_rate": 2.9785563568635702e-06,
      "loss": 0.0209,
      "step": 11806
    },
    {
      "epoch": 7.55406269993602,
      "grad_norm": 0.6378536915413946,
      "learning_rate": 2.9770810622566747e-06,
      "loss": 0.0243,
      "step": 11807
    },
    {
      "epoch": 7.554702495201536,
      "grad_norm": 0.6505209551825009,
      "learning_rate": 2.9756060692014276e-06,
      "loss": 0.0208,
      "step": 11808
    },
    {
      "epoch": 7.555342290467051,
      "grad_norm": 0.7625238710113554,
      "learning_rate": 2.9741313777611568e-06,
      "loss": 0.0231,
      "step": 11809
    },
    {
      "epoch": 7.555982085732565,
      "grad_norm": 0.6573919006303937,
      "learning_rate": 2.9726569879991896e-06,
      "loss": 0.0247,
      "step": 11810
    },
    {
      "epoch": 7.5566218809980805,
      "grad_norm": 0.7660327445053526,
      "learning_rate": 2.9711828999788295e-06,
      "loss": 0.0297,
      "step": 11811
    },
    {
      "epoch": 7.557261676263596,
      "grad_norm": 0.5884032979083744,
      "learning_rate": 2.9697091137633737e-06,
      "loss": 0.0196,
      "step": 11812
    },
    {
      "epoch": 7.557901471529111,
      "grad_norm": 0.7121141606643929,
      "learning_rate": 2.9682356294160996e-06,
      "loss": 0.0208,
      "step": 11813
    },
    {
      "epoch": 7.5585412667946255,
      "grad_norm": 0.7343448816781354,
      "learning_rate": 2.966762447000282e-06,
      "loss": 0.0269,
      "step": 11814
    },
    {
      "epoch": 7.559181062060141,
      "grad_norm": 0.6559420773283136,
      "learning_rate": 2.9652895665791735e-06,
      "loss": 0.0225,
      "step": 11815
    },
    {
      "epoch": 7.559820857325656,
      "grad_norm": 0.806098691882307,
      "learning_rate": 2.963816988216014e-06,
      "loss": 0.0274,
      "step": 11816
    },
    {
      "epoch": 7.560460652591171,
      "grad_norm": 0.620206961846094,
      "learning_rate": 2.96234471197404e-06,
      "loss": 0.0211,
      "step": 11817
    },
    {
      "epoch": 7.561100447856686,
      "grad_norm": 0.6493175004091977,
      "learning_rate": 2.9608727379164647e-06,
      "loss": 0.0223,
      "step": 11818
    },
    {
      "epoch": 7.561740243122201,
      "grad_norm": 0.6141223553905975,
      "learning_rate": 2.959401066106493e-06,
      "loss": 0.0206,
      "step": 11819
    },
    {
      "epoch": 7.562380038387716,
      "grad_norm": 0.7613393537255628,
      "learning_rate": 2.9579296966073144e-06,
      "loss": 0.0249,
      "step": 11820
    },
    {
      "epoch": 7.563019833653231,
      "grad_norm": 0.7306254441163352,
      "learning_rate": 2.9564586294821084e-06,
      "loss": 0.026,
      "step": 11821
    },
    {
      "epoch": 7.563659628918746,
      "grad_norm": 0.7206638840337113,
      "learning_rate": 2.954987864794039e-06,
      "loss": 0.0255,
      "step": 11822
    },
    {
      "epoch": 7.564299424184261,
      "grad_norm": 0.5443197867580517,
      "learning_rate": 2.953517402606256e-06,
      "loss": 0.018,
      "step": 11823
    },
    {
      "epoch": 7.564939219449776,
      "grad_norm": 0.8292867384994324,
      "learning_rate": 2.952047242981904e-06,
      "loss": 0.019,
      "step": 11824
    },
    {
      "epoch": 7.565579014715291,
      "grad_norm": 0.7385131795685087,
      "learning_rate": 2.950577385984106e-06,
      "loss": 0.029,
      "step": 11825
    },
    {
      "epoch": 7.566218809980806,
      "grad_norm": 0.5945256877318221,
      "learning_rate": 2.9491078316759712e-06,
      "loss": 0.0189,
      "step": 11826
    },
    {
      "epoch": 7.566858605246321,
      "grad_norm": 0.7144566850717969,
      "learning_rate": 2.9476385801206075e-06,
      "loss": 0.0261,
      "step": 11827
    },
    {
      "epoch": 7.567498400511836,
      "grad_norm": 0.5566034455809962,
      "learning_rate": 2.9461696313810982e-06,
      "loss": 0.0181,
      "step": 11828
    },
    {
      "epoch": 7.568138195777351,
      "grad_norm": 0.6648751326501176,
      "learning_rate": 2.944700985520513e-06,
      "loss": 0.0199,
      "step": 11829
    },
    {
      "epoch": 7.568777991042866,
      "grad_norm": 0.6663236496846412,
      "learning_rate": 2.9432326426019208e-06,
      "loss": 0.0267,
      "step": 11830
    },
    {
      "epoch": 7.569417786308382,
      "grad_norm": 0.724006206674993,
      "learning_rate": 2.941764602688366e-06,
      "loss": 0.0272,
      "step": 11831
    },
    {
      "epoch": 7.570057581573896,
      "grad_norm": 0.5809334430394092,
      "learning_rate": 2.940296865842884e-06,
      "loss": 0.0191,
      "step": 11832
    },
    {
      "epoch": 7.570697376839411,
      "grad_norm": 0.5537242817108035,
      "learning_rate": 2.938829432128496e-06,
      "loss": 0.019,
      "step": 11833
    },
    {
      "epoch": 7.5713371721049265,
      "grad_norm": 0.583342113356987,
      "learning_rate": 2.9373623016082107e-06,
      "loss": 0.0201,
      "step": 11834
    },
    {
      "epoch": 7.571976967370442,
      "grad_norm": 0.7809835604281156,
      "learning_rate": 2.9358954743450253e-06,
      "loss": 0.0242,
      "step": 11835
    },
    {
      "epoch": 7.572616762635956,
      "grad_norm": 0.6157846385341433,
      "learning_rate": 2.9344289504019184e-06,
      "loss": 0.0193,
      "step": 11836
    },
    {
      "epoch": 7.5732565579014715,
      "grad_norm": 0.6288022769131542,
      "learning_rate": 2.9329627298418685e-06,
      "loss": 0.0212,
      "step": 11837
    },
    {
      "epoch": 7.573896353166987,
      "grad_norm": 0.7227589856505853,
      "learning_rate": 2.9314968127278263e-06,
      "loss": 0.0204,
      "step": 11838
    },
    {
      "epoch": 7.574536148432502,
      "grad_norm": 0.6503774852682792,
      "learning_rate": 2.930031199122734e-06,
      "loss": 0.022,
      "step": 11839
    },
    {
      "epoch": 7.575175943698016,
      "grad_norm": 0.6980364352259161,
      "learning_rate": 2.9285658890895285e-06,
      "loss": 0.0215,
      "step": 11840
    },
    {
      "epoch": 7.575815738963532,
      "grad_norm": 0.6882099953594792,
      "learning_rate": 2.927100882691125e-06,
      "loss": 0.0238,
      "step": 11841
    },
    {
      "epoch": 7.576455534229047,
      "grad_norm": 0.6148510917740314,
      "learning_rate": 2.9256361799904276e-06,
      "loss": 0.0186,
      "step": 11842
    },
    {
      "epoch": 7.577095329494561,
      "grad_norm": 0.6706521870078742,
      "learning_rate": 2.9241717810503255e-06,
      "loss": 0.0231,
      "step": 11843
    },
    {
      "epoch": 7.577735124760077,
      "grad_norm": 0.6290337531477352,
      "learning_rate": 2.9227076859337033e-06,
      "loss": 0.0217,
      "step": 11844
    },
    {
      "epoch": 7.578374920025592,
      "grad_norm": 0.7100239411916546,
      "learning_rate": 2.921243894703424e-06,
      "loss": 0.0207,
      "step": 11845
    },
    {
      "epoch": 7.579014715291107,
      "grad_norm": 0.5712182168099346,
      "learning_rate": 2.919780407422336e-06,
      "loss": 0.0208,
      "step": 11846
    },
    {
      "epoch": 7.5796545105566215,
      "grad_norm": 0.7508508263457933,
      "learning_rate": 2.9183172241532887e-06,
      "loss": 0.0223,
      "step": 11847
    },
    {
      "epoch": 7.580294305822137,
      "grad_norm": 0.6133745446414037,
      "learning_rate": 2.916854344959099e-06,
      "loss": 0.0217,
      "step": 11848
    },
    {
      "epoch": 7.580934101087652,
      "grad_norm": 0.7764344392693433,
      "learning_rate": 2.915391769902581e-06,
      "loss": 0.026,
      "step": 11849
    },
    {
      "epoch": 7.581573896353167,
      "grad_norm": 0.620808547293715,
      "learning_rate": 2.9139294990465416e-06,
      "loss": 0.0223,
      "step": 11850
    },
    {
      "epoch": 7.582213691618682,
      "grad_norm": 0.7010876368891563,
      "learning_rate": 2.9124675324537645e-06,
      "loss": 0.0229,
      "step": 11851
    },
    {
      "epoch": 7.582853486884197,
      "grad_norm": 0.6666976808364954,
      "learning_rate": 2.9110058701870224e-06,
      "loss": 0.0246,
      "step": 11852
    },
    {
      "epoch": 7.583493282149712,
      "grad_norm": 0.6673707368542195,
      "learning_rate": 2.909544512309076e-06,
      "loss": 0.0225,
      "step": 11853
    },
    {
      "epoch": 7.584133077415228,
      "grad_norm": 0.6514970653090345,
      "learning_rate": 2.908083458882678e-06,
      "loss": 0.0213,
      "step": 11854
    },
    {
      "epoch": 7.584772872680742,
      "grad_norm": 0.7221162952537473,
      "learning_rate": 2.906622709970561e-06,
      "loss": 0.0207,
      "step": 11855
    },
    {
      "epoch": 7.585412667946257,
      "grad_norm": 0.607022992166649,
      "learning_rate": 2.905162265635443e-06,
      "loss": 0.0223,
      "step": 11856
    },
    {
      "epoch": 7.5860524632117725,
      "grad_norm": 0.7791370598745876,
      "learning_rate": 2.90370212594004e-06,
      "loss": 0.0274,
      "step": 11857
    },
    {
      "epoch": 7.586692258477287,
      "grad_norm": 0.5979926211411667,
      "learning_rate": 2.902242290947045e-06,
      "loss": 0.0205,
      "step": 11858
    },
    {
      "epoch": 7.587332053742802,
      "grad_norm": 0.6861577818092236,
      "learning_rate": 2.9007827607191364e-06,
      "loss": 0.0218,
      "step": 11859
    },
    {
      "epoch": 7.5879718490083174,
      "grad_norm": 0.8106223124270001,
      "learning_rate": 2.8993235353189908e-06,
      "loss": 0.026,
      "step": 11860
    },
    {
      "epoch": 7.588611644273833,
      "grad_norm": 0.6397345563293528,
      "learning_rate": 2.8978646148092616e-06,
      "loss": 0.0224,
      "step": 11861
    },
    {
      "epoch": 7.589251439539347,
      "grad_norm": 0.7110546519307368,
      "learning_rate": 2.8964059992525918e-06,
      "loss": 0.0219,
      "step": 11862
    },
    {
      "epoch": 7.589891234804862,
      "grad_norm": 0.6451967479982265,
      "learning_rate": 2.8949476887116112e-06,
      "loss": 0.0219,
      "step": 11863
    },
    {
      "epoch": 7.590531030070378,
      "grad_norm": 0.6397777293121248,
      "learning_rate": 2.893489683248939e-06,
      "loss": 0.0231,
      "step": 11864
    },
    {
      "epoch": 7.591170825335892,
      "grad_norm": 0.8237016417913571,
      "learning_rate": 2.892031982927178e-06,
      "loss": 0.0284,
      "step": 11865
    },
    {
      "epoch": 7.591810620601407,
      "grad_norm": 0.6927984671945436,
      "learning_rate": 2.8905745878089166e-06,
      "loss": 0.025,
      "step": 11866
    },
    {
      "epoch": 7.592450415866923,
      "grad_norm": 0.7226009939035964,
      "learning_rate": 2.8891174979567392e-06,
      "loss": 0.0247,
      "step": 11867
    },
    {
      "epoch": 7.593090211132438,
      "grad_norm": 0.6477609899083854,
      "learning_rate": 2.887660713433207e-06,
      "loss": 0.0232,
      "step": 11868
    },
    {
      "epoch": 7.593730006397953,
      "grad_norm": 0.6547666145272276,
      "learning_rate": 2.886204234300871e-06,
      "loss": 0.02,
      "step": 11869
    },
    {
      "epoch": 7.5943698016634675,
      "grad_norm": 0.6507937296487113,
      "learning_rate": 2.884748060622269e-06,
      "loss": 0.0218,
      "step": 11870
    },
    {
      "epoch": 7.595009596928983,
      "grad_norm": 0.6560963046641108,
      "learning_rate": 2.88329219245993e-06,
      "loss": 0.0239,
      "step": 11871
    },
    {
      "epoch": 7.595649392194498,
      "grad_norm": 0.6860558695496602,
      "learning_rate": 2.8818366298763656e-06,
      "loss": 0.0214,
      "step": 11872
    },
    {
      "epoch": 7.5962891874600125,
      "grad_norm": 0.739837381589917,
      "learning_rate": 2.88038137293407e-06,
      "loss": 0.0228,
      "step": 11873
    },
    {
      "epoch": 7.596928982725528,
      "grad_norm": 0.657815058357702,
      "learning_rate": 2.8789264216955382e-06,
      "loss": 0.0233,
      "step": 11874
    },
    {
      "epoch": 7.597568777991043,
      "grad_norm": 0.8547987169701683,
      "learning_rate": 2.8774717762232373e-06,
      "loss": 0.0259,
      "step": 11875
    },
    {
      "epoch": 7.598208573256558,
      "grad_norm": 0.6168476584138207,
      "learning_rate": 2.876017436579628e-06,
      "loss": 0.017,
      "step": 11876
    },
    {
      "epoch": 7.598848368522073,
      "grad_norm": 0.7199343725405614,
      "learning_rate": 2.8745634028271586e-06,
      "loss": 0.0246,
      "step": 11877
    },
    {
      "epoch": 7.599488163787588,
      "grad_norm": 0.5830505383185449,
      "learning_rate": 2.8731096750282605e-06,
      "loss": 0.0175,
      "step": 11878
    },
    {
      "epoch": 7.600127959053103,
      "grad_norm": 0.7551269774637724,
      "learning_rate": 2.871656253245355e-06,
      "loss": 0.0243,
      "step": 11879
    },
    {
      "epoch": 7.600767754318618,
      "grad_norm": 0.76749220427589,
      "learning_rate": 2.870203137540847e-06,
      "loss": 0.0239,
      "step": 11880
    },
    {
      "epoch": 7.601407549584133,
      "grad_norm": 0.7096983509428649,
      "learning_rate": 2.8687503279771367e-06,
      "loss": 0.0276,
      "step": 11881
    },
    {
      "epoch": 7.602047344849648,
      "grad_norm": 0.6467852702355308,
      "learning_rate": 2.8672978246166017e-06,
      "loss": 0.0222,
      "step": 11882
    },
    {
      "epoch": 7.6026871401151634,
      "grad_norm": 0.6721878036853003,
      "learning_rate": 2.865845627521606e-06,
      "loss": 0.023,
      "step": 11883
    },
    {
      "epoch": 7.603326935380678,
      "grad_norm": 0.6671613781520254,
      "learning_rate": 2.8643937367545126e-06,
      "loss": 0.0222,
      "step": 11884
    },
    {
      "epoch": 7.603966730646193,
      "grad_norm": 0.6990666045157692,
      "learning_rate": 2.862942152377659e-06,
      "loss": 0.0215,
      "step": 11885
    },
    {
      "epoch": 7.604606525911708,
      "grad_norm": 0.6850994538025498,
      "learning_rate": 2.86149087445337e-06,
      "loss": 0.0243,
      "step": 11886
    },
    {
      "epoch": 7.605246321177224,
      "grad_norm": 0.6255680236832435,
      "learning_rate": 2.8600399030439675e-06,
      "loss": 0.0222,
      "step": 11887
    },
    {
      "epoch": 7.605886116442738,
      "grad_norm": 0.6052357716286868,
      "learning_rate": 2.8585892382117495e-06,
      "loss": 0.0216,
      "step": 11888
    },
    {
      "epoch": 7.606525911708253,
      "grad_norm": 0.596444068183964,
      "learning_rate": 2.857138880019008e-06,
      "loss": 0.0205,
      "step": 11889
    },
    {
      "epoch": 7.607165706973769,
      "grad_norm": 0.6222244741354295,
      "learning_rate": 2.8556888285280126e-06,
      "loss": 0.0227,
      "step": 11890
    },
    {
      "epoch": 7.607805502239284,
      "grad_norm": 0.6076334417436721,
      "learning_rate": 2.8542390838010337e-06,
      "loss": 0.0197,
      "step": 11891
    },
    {
      "epoch": 7.608445297504798,
      "grad_norm": 0.7386519897471763,
      "learning_rate": 2.852789645900317e-06,
      "loss": 0.0237,
      "step": 11892
    },
    {
      "epoch": 7.6090850927703135,
      "grad_norm": 0.693485965542081,
      "learning_rate": 2.851340514888098e-06,
      "loss": 0.021,
      "step": 11893
    },
    {
      "epoch": 7.609724888035829,
      "grad_norm": 0.6942478227033817,
      "learning_rate": 2.8498916908266027e-06,
      "loss": 0.0229,
      "step": 11894
    },
    {
      "epoch": 7.610364683301343,
      "grad_norm": 0.6846347557136756,
      "learning_rate": 2.8484431737780382e-06,
      "loss": 0.0226,
      "step": 11895
    },
    {
      "epoch": 7.6110044785668585,
      "grad_norm": 0.737540756319222,
      "learning_rate": 2.846994963804599e-06,
      "loss": 0.0235,
      "step": 11896
    },
    {
      "epoch": 7.611644273832374,
      "grad_norm": 0.6100861869117458,
      "learning_rate": 2.8455470609684743e-06,
      "loss": 0.0205,
      "step": 11897
    },
    {
      "epoch": 7.612284069097889,
      "grad_norm": 0.6651727590274747,
      "learning_rate": 2.8440994653318333e-06,
      "loss": 0.0224,
      "step": 11898
    },
    {
      "epoch": 7.612923864363403,
      "grad_norm": 0.6111003548215516,
      "learning_rate": 2.8426521769568318e-06,
      "loss": 0.0233,
      "step": 11899
    },
    {
      "epoch": 7.613563659628919,
      "grad_norm": 0.6948776089463097,
      "learning_rate": 2.84120519590561e-06,
      "loss": 0.0254,
      "step": 11900
    },
    {
      "epoch": 7.614203454894434,
      "grad_norm": 0.6665830329984158,
      "learning_rate": 2.8397585222403058e-06,
      "loss": 0.0249,
      "step": 11901
    },
    {
      "epoch": 7.614843250159948,
      "grad_norm": 0.7336116081379974,
      "learning_rate": 2.8383121560230343e-06,
      "loss": 0.0228,
      "step": 11902
    },
    {
      "epoch": 7.615483045425464,
      "grad_norm": 0.6399669782744652,
      "learning_rate": 2.8368660973158955e-06,
      "loss": 0.0221,
      "step": 11903
    },
    {
      "epoch": 7.616122840690979,
      "grad_norm": 0.6930711480598583,
      "learning_rate": 2.835420346180987e-06,
      "loss": 0.0222,
      "step": 11904
    },
    {
      "epoch": 7.616762635956494,
      "grad_norm": 0.6745560763106953,
      "learning_rate": 2.8339749026803844e-06,
      "loss": 0.0214,
      "step": 11905
    },
    {
      "epoch": 7.617402431222009,
      "grad_norm": 0.5569774704199256,
      "learning_rate": 2.832529766876152e-06,
      "loss": 0.0211,
      "step": 11906
    },
    {
      "epoch": 7.618042226487524,
      "grad_norm": 0.7972942769152562,
      "learning_rate": 2.831084938830342e-06,
      "loss": 0.0237,
      "step": 11907
    },
    {
      "epoch": 7.618682021753039,
      "grad_norm": 0.7036014610151613,
      "learning_rate": 2.829640418604992e-06,
      "loss": 0.0245,
      "step": 11908
    },
    {
      "epoch": 7.619321817018554,
      "grad_norm": 0.6585373096019382,
      "learning_rate": 2.828196206262127e-06,
      "loss": 0.0233,
      "step": 11909
    },
    {
      "epoch": 7.619961612284069,
      "grad_norm": 0.7282106219991242,
      "learning_rate": 2.8267523018637566e-06,
      "loss": 0.022,
      "step": 11910
    },
    {
      "epoch": 7.620601407549584,
      "grad_norm": 0.6108418759470878,
      "learning_rate": 2.8253087054718854e-06,
      "loss": 0.0208,
      "step": 11911
    },
    {
      "epoch": 7.621241202815099,
      "grad_norm": 0.6725761370345793,
      "learning_rate": 2.8238654171484946e-06,
      "loss": 0.0239,
      "step": 11912
    },
    {
      "epoch": 7.621880998080615,
      "grad_norm": 0.7033005424456368,
      "learning_rate": 2.822422436955555e-06,
      "loss": 0.0225,
      "step": 11913
    },
    {
      "epoch": 7.622520793346129,
      "grad_norm": 0.59071884313594,
      "learning_rate": 2.820979764955031e-06,
      "loss": 0.0194,
      "step": 11914
    },
    {
      "epoch": 7.623160588611644,
      "grad_norm": 0.6924842321486852,
      "learning_rate": 2.819537401208866e-06,
      "loss": 0.0238,
      "step": 11915
    },
    {
      "epoch": 7.6238003838771595,
      "grad_norm": 0.6767279045775423,
      "learning_rate": 2.8180953457789884e-06,
      "loss": 0.0223,
      "step": 11916
    },
    {
      "epoch": 7.624440179142674,
      "grad_norm": 0.7156825826982285,
      "learning_rate": 2.8166535987273236e-06,
      "loss": 0.0257,
      "step": 11917
    },
    {
      "epoch": 7.625079974408189,
      "grad_norm": 0.7341657772701341,
      "learning_rate": 2.8152121601157757e-06,
      "loss": 0.0245,
      "step": 11918
    },
    {
      "epoch": 7.6257197696737045,
      "grad_norm": 0.5909207776112485,
      "learning_rate": 2.813771030006237e-06,
      "loss": 0.0194,
      "step": 11919
    },
    {
      "epoch": 7.62635956493922,
      "grad_norm": 0.7815616845557405,
      "learning_rate": 2.812330208460583e-06,
      "loss": 0.0266,
      "step": 11920
    },
    {
      "epoch": 7.626999360204734,
      "grad_norm": 0.7247864645436923,
      "learning_rate": 2.8108896955406883e-06,
      "loss": 0.0253,
      "step": 11921
    },
    {
      "epoch": 7.627639155470249,
      "grad_norm": 0.7162708631618778,
      "learning_rate": 2.8094494913084003e-06,
      "loss": 0.0208,
      "step": 11922
    },
    {
      "epoch": 7.628278950735765,
      "grad_norm": 0.746153485091804,
      "learning_rate": 2.8080095958255617e-06,
      "loss": 0.0257,
      "step": 11923
    },
    {
      "epoch": 7.62891874600128,
      "grad_norm": 0.7605030930847599,
      "learning_rate": 2.806570009153996e-06,
      "loss": 0.0271,
      "step": 11924
    },
    {
      "epoch": 7.629558541266794,
      "grad_norm": 0.6615462996687095,
      "learning_rate": 2.8051307313555186e-06,
      "loss": 0.023,
      "step": 11925
    },
    {
      "epoch": 7.63019833653231,
      "grad_norm": 0.6492832716643068,
      "learning_rate": 2.8036917624919257e-06,
      "loss": 0.0214,
      "step": 11926
    },
    {
      "epoch": 7.630838131797825,
      "grad_norm": 0.6717528575962208,
      "learning_rate": 2.802253102625011e-06,
      "loss": 0.0234,
      "step": 11927
    },
    {
      "epoch": 7.63147792706334,
      "grad_norm": 0.6735723863740921,
      "learning_rate": 2.800814751816544e-06,
      "loss": 0.0218,
      "step": 11928
    },
    {
      "epoch": 7.6321177223288545,
      "grad_norm": 0.7625113575034863,
      "learning_rate": 2.7993767101282855e-06,
      "loss": 0.0233,
      "step": 11929
    },
    {
      "epoch": 7.63275751759437,
      "grad_norm": 0.7359244273944138,
      "learning_rate": 2.7979389776219777e-06,
      "loss": 0.0288,
      "step": 11930
    },
    {
      "epoch": 7.633397312859885,
      "grad_norm": 0.7927091169354139,
      "learning_rate": 2.796501554359363e-06,
      "loss": 0.0272,
      "step": 11931
    },
    {
      "epoch": 7.6340371081253995,
      "grad_norm": 0.6338214152819813,
      "learning_rate": 2.7950644404021576e-06,
      "loss": 0.0221,
      "step": 11932
    },
    {
      "epoch": 7.634676903390915,
      "grad_norm": 0.719689466251286,
      "learning_rate": 2.793627635812065e-06,
      "loss": 0.0262,
      "step": 11933
    },
    {
      "epoch": 7.63531669865643,
      "grad_norm": 0.6904387441674004,
      "learning_rate": 2.792191140650785e-06,
      "loss": 0.025,
      "step": 11934
    },
    {
      "epoch": 7.635956493921945,
      "grad_norm": 0.7388007084468816,
      "learning_rate": 2.790754954979997e-06,
      "loss": 0.0244,
      "step": 11935
    },
    {
      "epoch": 7.63659628918746,
      "grad_norm": 0.7128239983293457,
      "learning_rate": 2.7893190788613665e-06,
      "loss": 0.0233,
      "step": 11936
    },
    {
      "epoch": 7.637236084452975,
      "grad_norm": 0.6679310490385211,
      "learning_rate": 2.7878835123565474e-06,
      "loss": 0.0245,
      "step": 11937
    },
    {
      "epoch": 7.63787587971849,
      "grad_norm": 0.7349476066636306,
      "learning_rate": 2.7864482555271822e-06,
      "loss": 0.0242,
      "step": 11938
    },
    {
      "epoch": 7.638515674984005,
      "grad_norm": 0.6167721958050628,
      "learning_rate": 2.7850133084348964e-06,
      "loss": 0.0182,
      "step": 11939
    },
    {
      "epoch": 7.63915547024952,
      "grad_norm": 0.554158604183613,
      "learning_rate": 2.7835786711413005e-06,
      "loss": 0.0186,
      "step": 11940
    },
    {
      "epoch": 7.639795265515035,
      "grad_norm": 0.6907251981109792,
      "learning_rate": 2.782144343708003e-06,
      "loss": 0.0224,
      "step": 11941
    },
    {
      "epoch": 7.6404350607805505,
      "grad_norm": 0.6737246961532962,
      "learning_rate": 2.7807103261965886e-06,
      "loss": 0.0237,
      "step": 11942
    },
    {
      "epoch": 7.641074856046066,
      "grad_norm": 0.6821717663994209,
      "learning_rate": 2.779276618668627e-06,
      "loss": 0.0249,
      "step": 11943
    },
    {
      "epoch": 7.64171465131158,
      "grad_norm": 0.783922607999773,
      "learning_rate": 2.7778432211856864e-06,
      "loss": 0.0266,
      "step": 11944
    },
    {
      "epoch": 7.642354446577095,
      "grad_norm": 0.721065050739122,
      "learning_rate": 2.7764101338093108e-06,
      "loss": 0.0249,
      "step": 11945
    },
    {
      "epoch": 7.642994241842611,
      "grad_norm": 0.7398578098119561,
      "learning_rate": 2.774977356601033e-06,
      "loss": 0.0266,
      "step": 11946
    },
    {
      "epoch": 7.643634037108125,
      "grad_norm": 0.7468561856654358,
      "learning_rate": 2.773544889622374e-06,
      "loss": 0.027,
      "step": 11947
    },
    {
      "epoch": 7.64427383237364,
      "grad_norm": 0.6147935192541304,
      "learning_rate": 2.772112732934844e-06,
      "loss": 0.0203,
      "step": 11948
    },
    {
      "epoch": 7.644913627639156,
      "grad_norm": 0.6545934678778471,
      "learning_rate": 2.7706808865999356e-06,
      "loss": 0.0213,
      "step": 11949
    },
    {
      "epoch": 7.645553422904671,
      "grad_norm": 0.636393232508618,
      "learning_rate": 2.7692493506791283e-06,
      "loss": 0.0211,
      "step": 11950
    },
    {
      "epoch": 7.646193218170185,
      "grad_norm": 0.7034741618851232,
      "learning_rate": 2.767818125233894e-06,
      "loss": 0.0246,
      "step": 11951
    },
    {
      "epoch": 7.6468330134357005,
      "grad_norm": 0.6103698083764343,
      "learning_rate": 2.766387210325684e-06,
      "loss": 0.0198,
      "step": 11952
    },
    {
      "epoch": 7.647472808701216,
      "grad_norm": 0.6606485484649031,
      "learning_rate": 2.7649566060159396e-06,
      "loss": 0.022,
      "step": 11953
    },
    {
      "epoch": 7.64811260396673,
      "grad_norm": 0.7168667962013859,
      "learning_rate": 2.763526312366088e-06,
      "loss": 0.0233,
      "step": 11954
    },
    {
      "epoch": 7.6487523992322455,
      "grad_norm": 0.5294350148029602,
      "learning_rate": 2.7620963294375445e-06,
      "loss": 0.0179,
      "step": 11955
    },
    {
      "epoch": 7.649392194497761,
      "grad_norm": 0.6691284577186263,
      "learning_rate": 2.7606666572917094e-06,
      "loss": 0.0238,
      "step": 11956
    },
    {
      "epoch": 7.650031989763276,
      "grad_norm": 0.7172617292458064,
      "learning_rate": 2.7592372959899673e-06,
      "loss": 0.0258,
      "step": 11957
    },
    {
      "epoch": 7.65067178502879,
      "grad_norm": 0.6652680570125684,
      "learning_rate": 2.7578082455936983e-06,
      "loss": 0.0239,
      "step": 11958
    },
    {
      "epoch": 7.651311580294306,
      "grad_norm": 0.6048215605208658,
      "learning_rate": 2.756379506164261e-06,
      "loss": 0.0193,
      "step": 11959
    },
    {
      "epoch": 7.651951375559821,
      "grad_norm": 0.7602953519775101,
      "learning_rate": 2.7549510777629986e-06,
      "loss": 0.026,
      "step": 11960
    },
    {
      "epoch": 7.652591170825336,
      "grad_norm": 0.6547373541812921,
      "learning_rate": 2.7535229604512516e-06,
      "loss": 0.0238,
      "step": 11961
    },
    {
      "epoch": 7.653230966090851,
      "grad_norm": 0.7574968862758352,
      "learning_rate": 2.7520951542903386e-06,
      "loss": 0.0228,
      "step": 11962
    },
    {
      "epoch": 7.653870761356366,
      "grad_norm": 0.6785042604235725,
      "learning_rate": 2.750667659341564e-06,
      "loss": 0.0229,
      "step": 11963
    },
    {
      "epoch": 7.654510556621881,
      "grad_norm": 0.8314669354610337,
      "learning_rate": 2.7492404756662272e-06,
      "loss": 0.0234,
      "step": 11964
    },
    {
      "epoch": 7.6551503518873965,
      "grad_norm": 0.7259335427158129,
      "learning_rate": 2.7478136033256055e-06,
      "loss": 0.0246,
      "step": 11965
    },
    {
      "epoch": 7.655790147152911,
      "grad_norm": 0.6759597605200879,
      "learning_rate": 2.746387042380968e-06,
      "loss": 0.0231,
      "step": 11966
    },
    {
      "epoch": 7.656429942418426,
      "grad_norm": 0.8019243490201251,
      "learning_rate": 2.7449607928935673e-06,
      "loss": 0.0249,
      "step": 11967
    },
    {
      "epoch": 7.657069737683941,
      "grad_norm": 0.6531141049729912,
      "learning_rate": 2.7435348549246454e-06,
      "loss": 0.0243,
      "step": 11968
    },
    {
      "epoch": 7.657709532949456,
      "grad_norm": 0.7077858698721857,
      "learning_rate": 2.7421092285354277e-06,
      "loss": 0.023,
      "step": 11969
    },
    {
      "epoch": 7.658349328214971,
      "grad_norm": 0.6803033373295391,
      "learning_rate": 2.7406839137871255e-06,
      "loss": 0.0205,
      "step": 11970
    },
    {
      "epoch": 7.658989123480486,
      "grad_norm": 0.6812533580757684,
      "learning_rate": 2.7392589107409463e-06,
      "loss": 0.0252,
      "step": 11971
    },
    {
      "epoch": 7.659628918746002,
      "grad_norm": 0.6971519736028289,
      "learning_rate": 2.737834219458074e-06,
      "loss": 0.0224,
      "step": 11972
    },
    {
      "epoch": 7.660268714011516,
      "grad_norm": 0.6078974741595701,
      "learning_rate": 2.7364098399996784e-06,
      "loss": 0.0225,
      "step": 11973
    },
    {
      "epoch": 7.660908509277031,
      "grad_norm": 0.6756181685184506,
      "learning_rate": 2.7349857724269268e-06,
      "loss": 0.0215,
      "step": 11974
    },
    {
      "epoch": 7.6615483045425465,
      "grad_norm": 0.7029454252077578,
      "learning_rate": 2.7335620168009626e-06,
      "loss": 0.0234,
      "step": 11975
    },
    {
      "epoch": 7.662188099808061,
      "grad_norm": 0.6873402379664569,
      "learning_rate": 2.7321385731829186e-06,
      "loss": 0.0233,
      "step": 11976
    },
    {
      "epoch": 7.662827895073576,
      "grad_norm": 0.6584959100601492,
      "learning_rate": 2.730715441633913e-06,
      "loss": 0.0222,
      "step": 11977
    },
    {
      "epoch": 7.6634676903390915,
      "grad_norm": 0.7321361210317698,
      "learning_rate": 2.7292926222150583e-06,
      "loss": 0.0235,
      "step": 11978
    },
    {
      "epoch": 7.664107485604607,
      "grad_norm": 0.6675335243834611,
      "learning_rate": 2.7278701149874456e-06,
      "loss": 0.0211,
      "step": 11979
    },
    {
      "epoch": 7.664747280870122,
      "grad_norm": 0.6882338148528357,
      "learning_rate": 2.7264479200121507e-06,
      "loss": 0.025,
      "step": 11980
    },
    {
      "epoch": 7.665387076135636,
      "grad_norm": 0.6576851418384709,
      "learning_rate": 2.7250260373502456e-06,
      "loss": 0.0241,
      "step": 11981
    },
    {
      "epoch": 7.666026871401152,
      "grad_norm": 0.6665295474540994,
      "learning_rate": 2.7236044670627828e-06,
      "loss": 0.0221,
      "step": 11982
    },
    {
      "epoch": 7.666666666666667,
      "grad_norm": 0.7562178756315576,
      "learning_rate": 2.7221832092107993e-06,
      "loss": 0.0231,
      "step": 11983
    },
    {
      "epoch": 7.667306461932181,
      "grad_norm": 0.6630076684532101,
      "learning_rate": 2.720762263855323e-06,
      "loss": 0.0233,
      "step": 11984
    },
    {
      "epoch": 7.667946257197697,
      "grad_norm": 0.6460810581123018,
      "learning_rate": 2.719341631057366e-06,
      "loss": 0.021,
      "step": 11985
    },
    {
      "epoch": 7.668586052463212,
      "grad_norm": 0.7511606376993489,
      "learning_rate": 2.717921310877929e-06,
      "loss": 0.0237,
      "step": 11986
    },
    {
      "epoch": 7.669225847728727,
      "grad_norm": 0.6928961395851484,
      "learning_rate": 2.716501303377993e-06,
      "loss": 0.0214,
      "step": 11987
    },
    {
      "epoch": 7.669865642994242,
      "grad_norm": 0.6407798629941831,
      "learning_rate": 2.7150816086185384e-06,
      "loss": 0.0209,
      "step": 11988
    },
    {
      "epoch": 7.670505438259757,
      "grad_norm": 0.7454369343904855,
      "learning_rate": 2.7136622266605205e-06,
      "loss": 0.0211,
      "step": 11989
    },
    {
      "epoch": 7.671145233525272,
      "grad_norm": 0.6503367304903072,
      "learning_rate": 2.7122431575648823e-06,
      "loss": 0.0235,
      "step": 11990
    },
    {
      "epoch": 7.6717850287907865,
      "grad_norm": 0.6885603129456883,
      "learning_rate": 2.710824401392561e-06,
      "loss": 0.021,
      "step": 11991
    },
    {
      "epoch": 7.672424824056302,
      "grad_norm": 0.699312158832741,
      "learning_rate": 2.709405958204475e-06,
      "loss": 0.0227,
      "step": 11992
    },
    {
      "epoch": 7.673064619321817,
      "grad_norm": 0.7243451867939639,
      "learning_rate": 2.7079878280615247e-06,
      "loss": 0.0228,
      "step": 11993
    },
    {
      "epoch": 7.673704414587332,
      "grad_norm": 0.6794614530382438,
      "learning_rate": 2.7065700110246074e-06,
      "loss": 0.0253,
      "step": 11994
    },
    {
      "epoch": 7.674344209852847,
      "grad_norm": 0.6595317714970462,
      "learning_rate": 2.7051525071546014e-06,
      "loss": 0.0212,
      "step": 11995
    },
    {
      "epoch": 7.674984005118362,
      "grad_norm": 0.7218880543792726,
      "learning_rate": 2.7037353165123692e-06,
      "loss": 0.0202,
      "step": 11996
    },
    {
      "epoch": 7.675623800383877,
      "grad_norm": 0.6822817933308222,
      "learning_rate": 2.702318439158764e-06,
      "loss": 0.0218,
      "step": 11997
    },
    {
      "epoch": 7.6762635956493925,
      "grad_norm": 0.6575066418459755,
      "learning_rate": 2.7009018751546233e-06,
      "loss": 0.0192,
      "step": 11998
    },
    {
      "epoch": 7.676903390914907,
      "grad_norm": 0.6932054673407332,
      "learning_rate": 2.6994856245607728e-06,
      "loss": 0.0218,
      "step": 11999
    },
    {
      "epoch": 7.677543186180422,
      "grad_norm": 0.6212722466515281,
      "learning_rate": 2.6980696874380184e-06,
      "loss": 0.0192,
      "step": 12000
    },
    {
      "epoch": 7.6781829814459375,
      "grad_norm": 0.6951069481513887,
      "learning_rate": 2.6966540638471662e-06,
      "loss": 0.0213,
      "step": 12001
    },
    {
      "epoch": 7.678822776711453,
      "grad_norm": 0.7266463002217396,
      "learning_rate": 2.6952387538489967e-06,
      "loss": 0.0221,
      "step": 12002
    },
    {
      "epoch": 7.679462571976967,
      "grad_norm": 0.6855441922815532,
      "learning_rate": 2.693823757504278e-06,
      "loss": 0.0253,
      "step": 12003
    },
    {
      "epoch": 7.680102367242482,
      "grad_norm": 0.6965722134267222,
      "learning_rate": 2.692409074873774e-06,
      "loss": 0.0223,
      "step": 12004
    },
    {
      "epoch": 7.680742162507998,
      "grad_norm": 0.6999213771044767,
      "learning_rate": 2.690994706018225e-06,
      "loss": 0.0222,
      "step": 12005
    },
    {
      "epoch": 7.681381957773512,
      "grad_norm": 0.7859868474658959,
      "learning_rate": 2.6895806509983613e-06,
      "loss": 0.0218,
      "step": 12006
    },
    {
      "epoch": 7.682021753039027,
      "grad_norm": 0.6361179791842421,
      "learning_rate": 2.6881669098748974e-06,
      "loss": 0.0186,
      "step": 12007
    },
    {
      "epoch": 7.682661548304543,
      "grad_norm": 0.6195538540727151,
      "learning_rate": 2.686753482708543e-06,
      "loss": 0.0214,
      "step": 12008
    },
    {
      "epoch": 7.683301343570058,
      "grad_norm": 0.6932736345421953,
      "learning_rate": 2.6853403695599834e-06,
      "loss": 0.0228,
      "step": 12009
    },
    {
      "epoch": 7.683941138835572,
      "grad_norm": 0.6775950156211867,
      "learning_rate": 2.6839275704898947e-06,
      "loss": 0.0208,
      "step": 12010
    },
    {
      "epoch": 7.684580934101088,
      "grad_norm": 0.6186268352548634,
      "learning_rate": 2.6825150855589445e-06,
      "loss": 0.0188,
      "step": 12011
    },
    {
      "epoch": 7.685220729366603,
      "grad_norm": 0.7272540504833577,
      "learning_rate": 2.68110291482778e-06,
      "loss": 0.0234,
      "step": 12012
    },
    {
      "epoch": 7.685860524632118,
      "grad_norm": 0.6268476056223897,
      "learning_rate": 2.679691058357037e-06,
      "loss": 0.0203,
      "step": 12013
    },
    {
      "epoch": 7.6865003198976325,
      "grad_norm": 0.6510284725996709,
      "learning_rate": 2.6782795162073385e-06,
      "loss": 0.0213,
      "step": 12014
    },
    {
      "epoch": 7.687140115163148,
      "grad_norm": 0.7035645821493415,
      "learning_rate": 2.6768682884392928e-06,
      "loss": 0.023,
      "step": 12015
    },
    {
      "epoch": 7.687779910428663,
      "grad_norm": 0.6908507983353718,
      "learning_rate": 2.675457375113495e-06,
      "loss": 0.0244,
      "step": 12016
    },
    {
      "epoch": 7.688419705694178,
      "grad_norm": 0.5935052747495381,
      "learning_rate": 2.6740467762905264e-06,
      "loss": 0.0184,
      "step": 12017
    },
    {
      "epoch": 7.689059500959693,
      "grad_norm": 0.828458535107327,
      "learning_rate": 2.67263649203096e-06,
      "loss": 0.0249,
      "step": 12018
    },
    {
      "epoch": 7.689699296225208,
      "grad_norm": 0.7372706253893643,
      "learning_rate": 2.671226522395348e-06,
      "loss": 0.0239,
      "step": 12019
    },
    {
      "epoch": 7.690339091490723,
      "grad_norm": 0.6832192245253643,
      "learning_rate": 2.6698168674442304e-06,
      "loss": 0.0228,
      "step": 12020
    },
    {
      "epoch": 7.690978886756238,
      "grad_norm": 0.6192722380124858,
      "learning_rate": 2.6684075272381395e-06,
      "loss": 0.0192,
      "step": 12021
    },
    {
      "epoch": 7.691618682021753,
      "grad_norm": 0.7137029375226187,
      "learning_rate": 2.6669985018375864e-06,
      "loss": 0.024,
      "step": 12022
    },
    {
      "epoch": 7.692258477287268,
      "grad_norm": 0.686046940168572,
      "learning_rate": 2.665589791303075e-06,
      "loss": 0.0245,
      "step": 12023
    },
    {
      "epoch": 7.6928982725527835,
      "grad_norm": 0.6587275279085297,
      "learning_rate": 2.664181395695087e-06,
      "loss": 0.0194,
      "step": 12024
    },
    {
      "epoch": 7.693538067818298,
      "grad_norm": 0.7890308929543525,
      "learning_rate": 2.662773315074103e-06,
      "loss": 0.0226,
      "step": 12025
    },
    {
      "epoch": 7.694177863083813,
      "grad_norm": 0.806929317381531,
      "learning_rate": 2.661365549500582e-06,
      "loss": 0.0209,
      "step": 12026
    },
    {
      "epoch": 7.694817658349328,
      "grad_norm": 0.6809226590379219,
      "learning_rate": 2.6599580990349694e-06,
      "loss": 0.0225,
      "step": 12027
    },
    {
      "epoch": 7.695457453614843,
      "grad_norm": 0.660628598603376,
      "learning_rate": 2.6585509637376996e-06,
      "loss": 0.021,
      "step": 12028
    },
    {
      "epoch": 7.696097248880358,
      "grad_norm": 0.6929457350211892,
      "learning_rate": 2.6571441436691913e-06,
      "loss": 0.0227,
      "step": 12029
    },
    {
      "epoch": 7.696737044145873,
      "grad_norm": 0.7289539361930619,
      "learning_rate": 2.655737638889848e-06,
      "loss": 0.023,
      "step": 12030
    },
    {
      "epoch": 7.697376839411389,
      "grad_norm": 0.7001263607644718,
      "learning_rate": 2.6543314494600704e-06,
      "loss": 0.0241,
      "step": 12031
    },
    {
      "epoch": 7.698016634676903,
      "grad_norm": 0.7188967480904365,
      "learning_rate": 2.652925575440233e-06,
      "loss": 0.0235,
      "step": 12032
    },
    {
      "epoch": 7.698656429942418,
      "grad_norm": 0.6572565913807524,
      "learning_rate": 2.651520016890702e-06,
      "loss": 0.0205,
      "step": 12033
    },
    {
      "epoch": 7.6992962252079336,
      "grad_norm": 0.6916182068847401,
      "learning_rate": 2.650114773871826e-06,
      "loss": 0.018,
      "step": 12034
    },
    {
      "epoch": 7.699936020473449,
      "grad_norm": 0.7002645175453343,
      "learning_rate": 2.648709846443951e-06,
      "loss": 0.0229,
      "step": 12035
    },
    {
      "epoch": 7.700575815738963,
      "grad_norm": 0.8787799380389982,
      "learning_rate": 2.6473052346673978e-06,
      "loss": 0.0233,
      "step": 12036
    },
    {
      "epoch": 7.7012156110044785,
      "grad_norm": 0.8221167277399761,
      "learning_rate": 2.645900938602475e-06,
      "loss": 0.0293,
      "step": 12037
    },
    {
      "epoch": 7.701855406269994,
      "grad_norm": 0.7166246788705825,
      "learning_rate": 2.6444969583094872e-06,
      "loss": 0.0255,
      "step": 12038
    },
    {
      "epoch": 7.702495201535509,
      "grad_norm": 0.6477518318754286,
      "learning_rate": 2.6430932938487163e-06,
      "loss": 0.0207,
      "step": 12039
    },
    {
      "epoch": 7.703134996801023,
      "grad_norm": 0.7412867393483779,
      "learning_rate": 2.6416899452804288e-06,
      "loss": 0.0237,
      "step": 12040
    },
    {
      "epoch": 7.703774792066539,
      "grad_norm": 0.6675752844773292,
      "learning_rate": 2.640286912664889e-06,
      "loss": 0.0233,
      "step": 12041
    },
    {
      "epoch": 7.704414587332054,
      "grad_norm": 0.6109126265331721,
      "learning_rate": 2.6388841960623367e-06,
      "loss": 0.0225,
      "step": 12042
    },
    {
      "epoch": 7.705054382597568,
      "grad_norm": 0.7429344049347034,
      "learning_rate": 2.6374817955330034e-06,
      "loss": 0.023,
      "step": 12043
    },
    {
      "epoch": 7.705694177863084,
      "grad_norm": 0.7331071691656319,
      "learning_rate": 2.636079711137104e-06,
      "loss": 0.0214,
      "step": 12044
    },
    {
      "epoch": 7.706333973128599,
      "grad_norm": 0.7044420472638142,
      "learning_rate": 2.6346779429348425e-06,
      "loss": 0.0242,
      "step": 12045
    },
    {
      "epoch": 7.706973768394114,
      "grad_norm": 0.5822493168756822,
      "learning_rate": 2.6332764909864083e-06,
      "loss": 0.0195,
      "step": 12046
    },
    {
      "epoch": 7.707613563659629,
      "grad_norm": 0.7492719508545722,
      "learning_rate": 2.631875355351975e-06,
      "loss": 0.0234,
      "step": 12047
    },
    {
      "epoch": 7.708253358925144,
      "grad_norm": 0.7260420871589113,
      "learning_rate": 2.6304745360917095e-06,
      "loss": 0.0248,
      "step": 12048
    },
    {
      "epoch": 7.708893154190659,
      "grad_norm": 0.7192518524510488,
      "learning_rate": 2.6290740332657584e-06,
      "loss": 0.0243,
      "step": 12049
    },
    {
      "epoch": 7.709532949456174,
      "grad_norm": 0.6875903760711395,
      "learning_rate": 2.6276738469342533e-06,
      "loss": 0.0227,
      "step": 12050
    },
    {
      "epoch": 7.710172744721689,
      "grad_norm": 0.6878025271421401,
      "learning_rate": 2.626273977157321e-06,
      "loss": 0.02,
      "step": 12051
    },
    {
      "epoch": 7.710812539987204,
      "grad_norm": 0.6186758240500968,
      "learning_rate": 2.6248744239950686e-06,
      "loss": 0.0183,
      "step": 12052
    },
    {
      "epoch": 7.711452335252719,
      "grad_norm": 0.8470782265804156,
      "learning_rate": 2.623475187507588e-06,
      "loss": 0.0292,
      "step": 12053
    },
    {
      "epoch": 7.712092130518235,
      "grad_norm": 0.697737059154552,
      "learning_rate": 2.622076267754958e-06,
      "loss": 0.0256,
      "step": 12054
    },
    {
      "epoch": 7.712731925783749,
      "grad_norm": 0.6946235402919764,
      "learning_rate": 2.6206776647972533e-06,
      "loss": 0.0253,
      "step": 12055
    },
    {
      "epoch": 7.713371721049264,
      "grad_norm": 0.718618375720918,
      "learning_rate": 2.6192793786945213e-06,
      "loss": 0.0239,
      "step": 12056
    },
    {
      "epoch": 7.7140115163147795,
      "grad_norm": 0.7698412719143092,
      "learning_rate": 2.617881409506804e-06,
      "loss": 0.0266,
      "step": 12057
    },
    {
      "epoch": 7.714651311580294,
      "grad_norm": 0.7472332854329905,
      "learning_rate": 2.6164837572941272e-06,
      "loss": 0.0246,
      "step": 12058
    },
    {
      "epoch": 7.715291106845809,
      "grad_norm": 0.8093457633828028,
      "learning_rate": 2.615086422116504e-06,
      "loss": 0.024,
      "step": 12059
    },
    {
      "epoch": 7.7159309021113245,
      "grad_norm": 0.5362934639176614,
      "learning_rate": 2.613689404033929e-06,
      "loss": 0.0153,
      "step": 12060
    },
    {
      "epoch": 7.71657069737684,
      "grad_norm": 0.674557089142069,
      "learning_rate": 2.612292703106396e-06,
      "loss": 0.0227,
      "step": 12061
    },
    {
      "epoch": 7.717210492642354,
      "grad_norm": 0.6745843696338008,
      "learning_rate": 2.6108963193938718e-06,
      "loss": 0.0225,
      "step": 12062
    },
    {
      "epoch": 7.717850287907869,
      "grad_norm": 0.7614484616780112,
      "learning_rate": 2.609500252956316e-06,
      "loss": 0.0226,
      "step": 12063
    },
    {
      "epoch": 7.718490083173385,
      "grad_norm": 0.6817532713322406,
      "learning_rate": 2.608104503853669e-06,
      "loss": 0.0226,
      "step": 12064
    },
    {
      "epoch": 7.719129878438899,
      "grad_norm": 0.632783921996947,
      "learning_rate": 2.6067090721458687e-06,
      "loss": 0.0198,
      "step": 12065
    },
    {
      "epoch": 7.719769673704414,
      "grad_norm": 0.7619411302428429,
      "learning_rate": 2.60531395789283e-06,
      "loss": 0.0242,
      "step": 12066
    },
    {
      "epoch": 7.72040946896993,
      "grad_norm": 0.5489985564052178,
      "learning_rate": 2.6039191611544523e-06,
      "loss": 0.0189,
      "step": 12067
    },
    {
      "epoch": 7.721049264235445,
      "grad_norm": 0.6150140048869318,
      "learning_rate": 2.6025246819906314e-06,
      "loss": 0.0211,
      "step": 12068
    },
    {
      "epoch": 7.72168905950096,
      "grad_norm": 0.737548635928296,
      "learning_rate": 2.6011305204612426e-06,
      "loss": 0.0222,
      "step": 12069
    },
    {
      "epoch": 7.722328854766475,
      "grad_norm": 0.6518999506820181,
      "learning_rate": 2.5997366766261445e-06,
      "loss": 0.0192,
      "step": 12070
    },
    {
      "epoch": 7.72296865003199,
      "grad_norm": 0.7825028589398407,
      "learning_rate": 2.5983431505451917e-06,
      "loss": 0.024,
      "step": 12071
    },
    {
      "epoch": 7.723608445297505,
      "grad_norm": 0.6169371033059958,
      "learning_rate": 2.5969499422782183e-06,
      "loss": 0.0183,
      "step": 12072
    },
    {
      "epoch": 7.7242482405630195,
      "grad_norm": 0.610292193330289,
      "learning_rate": 2.595557051885046e-06,
      "loss": 0.0193,
      "step": 12073
    },
    {
      "epoch": 7.724888035828535,
      "grad_norm": 0.7864081896352415,
      "learning_rate": 2.5941644794254804e-06,
      "loss": 0.0255,
      "step": 12074
    },
    {
      "epoch": 7.72552783109405,
      "grad_norm": 0.7518579294362006,
      "learning_rate": 2.5927722249593203e-06,
      "loss": 0.0249,
      "step": 12075
    },
    {
      "epoch": 7.726167626359565,
      "grad_norm": 0.693326670472122,
      "learning_rate": 2.5913802885463425e-06,
      "loss": 0.0231,
      "step": 12076
    },
    {
      "epoch": 7.72680742162508,
      "grad_norm": 0.7111419164174079,
      "learning_rate": 2.5899886702463138e-06,
      "loss": 0.0228,
      "step": 12077
    },
    {
      "epoch": 7.727447216890595,
      "grad_norm": 0.7073386247875666,
      "learning_rate": 2.588597370118994e-06,
      "loss": 0.0226,
      "step": 12078
    },
    {
      "epoch": 7.72808701215611,
      "grad_norm": 0.6411709631411355,
      "learning_rate": 2.5872063882241193e-06,
      "loss": 0.0195,
      "step": 12079
    },
    {
      "epoch": 7.728726807421625,
      "grad_norm": 0.6870023724141714,
      "learning_rate": 2.5858157246214123e-06,
      "loss": 0.0218,
      "step": 12080
    },
    {
      "epoch": 7.72936660268714,
      "grad_norm": 0.7582101552542717,
      "learning_rate": 2.584425379370592e-06,
      "loss": 0.0206,
      "step": 12081
    },
    {
      "epoch": 7.730006397952655,
      "grad_norm": 0.6620864760505589,
      "learning_rate": 2.5830353525313556e-06,
      "loss": 0.0213,
      "step": 12082
    },
    {
      "epoch": 7.7306461932181705,
      "grad_norm": 0.6941443798971437,
      "learning_rate": 2.5816456441633863e-06,
      "loss": 0.0217,
      "step": 12083
    },
    {
      "epoch": 7.731285988483685,
      "grad_norm": 0.7572116749568625,
      "learning_rate": 2.580256254326354e-06,
      "loss": 0.0235,
      "step": 12084
    },
    {
      "epoch": 7.7319257837492,
      "grad_norm": 0.672944236583544,
      "learning_rate": 2.5788671830799227e-06,
      "loss": 0.0233,
      "step": 12085
    },
    {
      "epoch": 7.732565579014715,
      "grad_norm": 0.6935925570862347,
      "learning_rate": 2.577478430483734e-06,
      "loss": 0.0237,
      "step": 12086
    },
    {
      "epoch": 7.733205374280231,
      "grad_norm": 0.6299490637640242,
      "learning_rate": 2.5760899965974173e-06,
      "loss": 0.0218,
      "step": 12087
    },
    {
      "epoch": 7.733845169545745,
      "grad_norm": 0.5865669351537085,
      "learning_rate": 2.57470188148059e-06,
      "loss": 0.0185,
      "step": 12088
    },
    {
      "epoch": 7.73448496481126,
      "grad_norm": 0.6551411203790474,
      "learning_rate": 2.573314085192855e-06,
      "loss": 0.0217,
      "step": 12089
    },
    {
      "epoch": 7.735124760076776,
      "grad_norm": 0.7577663734232003,
      "learning_rate": 2.5719266077938e-06,
      "loss": 0.0256,
      "step": 12090
    },
    {
      "epoch": 7.735764555342291,
      "grad_norm": 0.6619367437646066,
      "learning_rate": 2.5705394493430056e-06,
      "loss": 0.0215,
      "step": 12091
    },
    {
      "epoch": 7.736404350607805,
      "grad_norm": 0.6452376311513616,
      "learning_rate": 2.5691526099000308e-06,
      "loss": 0.0223,
      "step": 12092
    },
    {
      "epoch": 7.737044145873321,
      "grad_norm": 0.6285444723394087,
      "learning_rate": 2.567766089524425e-06,
      "loss": 0.0216,
      "step": 12093
    },
    {
      "epoch": 7.737683941138836,
      "grad_norm": 0.6608398749512354,
      "learning_rate": 2.5663798882757185e-06,
      "loss": 0.021,
      "step": 12094
    },
    {
      "epoch": 7.73832373640435,
      "grad_norm": 0.774903673375478,
      "learning_rate": 2.5649940062134405e-06,
      "loss": 0.0302,
      "step": 12095
    },
    {
      "epoch": 7.7389635316698655,
      "grad_norm": 0.6442085258977823,
      "learning_rate": 2.563608443397093e-06,
      "loss": 0.0201,
      "step": 12096
    },
    {
      "epoch": 7.739603326935381,
      "grad_norm": 0.7141012236480735,
      "learning_rate": 2.5622231998861678e-06,
      "loss": 0.0232,
      "step": 12097
    },
    {
      "epoch": 7.740243122200896,
      "grad_norm": 0.6297132138238893,
      "learning_rate": 2.560838275740152e-06,
      "loss": 0.0203,
      "step": 12098
    },
    {
      "epoch": 7.74088291746641,
      "grad_norm": 0.6300459253903369,
      "learning_rate": 2.559453671018506e-06,
      "loss": 0.0206,
      "step": 12099
    },
    {
      "epoch": 7.741522712731926,
      "grad_norm": 0.66534571816013,
      "learning_rate": 2.558069385780684e-06,
      "loss": 0.0249,
      "step": 12100
    },
    {
      "epoch": 7.742162507997441,
      "grad_norm": 0.663361311031848,
      "learning_rate": 2.5566854200861213e-06,
      "loss": 0.0211,
      "step": 12101
    },
    {
      "epoch": 7.742802303262955,
      "grad_norm": 0.7039341626752906,
      "learning_rate": 2.555301773994249e-06,
      "loss": 0.0251,
      "step": 12102
    },
    {
      "epoch": 7.743442098528471,
      "grad_norm": 0.6847792076928481,
      "learning_rate": 2.553918447564475e-06,
      "loss": 0.0231,
      "step": 12103
    },
    {
      "epoch": 7.744081893793986,
      "grad_norm": 0.6317469437107489,
      "learning_rate": 2.5525354408561974e-06,
      "loss": 0.0214,
      "step": 12104
    },
    {
      "epoch": 7.744721689059501,
      "grad_norm": 0.6334023260423146,
      "learning_rate": 2.5511527539288007e-06,
      "loss": 0.0195,
      "step": 12105
    },
    {
      "epoch": 7.7453614843250165,
      "grad_norm": 0.532905598110573,
      "learning_rate": 2.5497703868416544e-06,
      "loss": 0.0151,
      "step": 12106
    },
    {
      "epoch": 7.746001279590531,
      "grad_norm": 0.6750798506185138,
      "learning_rate": 2.548388339654111e-06,
      "loss": 0.0258,
      "step": 12107
    },
    {
      "epoch": 7.746641074856046,
      "grad_norm": 0.6508758841808265,
      "learning_rate": 2.54700661242552e-06,
      "loss": 0.0236,
      "step": 12108
    },
    {
      "epoch": 7.747280870121561,
      "grad_norm": 0.6819930324196223,
      "learning_rate": 2.5456252052152074e-06,
      "loss": 0.0241,
      "step": 12109
    },
    {
      "epoch": 7.747920665387076,
      "grad_norm": 0.6254025507485845,
      "learning_rate": 2.5442441180824886e-06,
      "loss": 0.0219,
      "step": 12110
    },
    {
      "epoch": 7.748560460652591,
      "grad_norm": 0.7145548328258823,
      "learning_rate": 2.5428633510866608e-06,
      "loss": 0.0237,
      "step": 12111
    },
    {
      "epoch": 7.749200255918106,
      "grad_norm": 0.6982386314882947,
      "learning_rate": 2.541482904287019e-06,
      "loss": 0.0214,
      "step": 12112
    },
    {
      "epoch": 7.749840051183622,
      "grad_norm": 0.7410019029417915,
      "learning_rate": 2.5401027777428333e-06,
      "loss": 0.0264,
      "step": 12113
    },
    {
      "epoch": 7.750479846449136,
      "grad_norm": 0.7264069402413855,
      "learning_rate": 2.538722971513362e-06,
      "loss": 0.0262,
      "step": 12114
    },
    {
      "epoch": 7.751119641714651,
      "grad_norm": 0.7847670977474881,
      "learning_rate": 2.537343485657856e-06,
      "loss": 0.0265,
      "step": 12115
    },
    {
      "epoch": 7.751759436980167,
      "grad_norm": 0.6263034426281753,
      "learning_rate": 2.5359643202355456e-06,
      "loss": 0.0239,
      "step": 12116
    },
    {
      "epoch": 7.752399232245681,
      "grad_norm": 0.6411305602105206,
      "learning_rate": 2.53458547530565e-06,
      "loss": 0.0232,
      "step": 12117
    },
    {
      "epoch": 7.753039027511196,
      "grad_norm": 0.675204599670505,
      "learning_rate": 2.533206950927375e-06,
      "loss": 0.0248,
      "step": 12118
    },
    {
      "epoch": 7.7536788227767115,
      "grad_norm": 0.6214578082668737,
      "learning_rate": 2.53182874715991e-06,
      "loss": 0.0212,
      "step": 12119
    },
    {
      "epoch": 7.754318618042227,
      "grad_norm": 0.6450834077822887,
      "learning_rate": 2.5304508640624338e-06,
      "loss": 0.0217,
      "step": 12120
    },
    {
      "epoch": 7.754958413307741,
      "grad_norm": 0.693976477316699,
      "learning_rate": 2.529073301694107e-06,
      "loss": 0.0229,
      "step": 12121
    },
    {
      "epoch": 7.755598208573256,
      "grad_norm": 0.6967470543898542,
      "learning_rate": 2.5276960601140865e-06,
      "loss": 0.0238,
      "step": 12122
    },
    {
      "epoch": 7.756238003838772,
      "grad_norm": 0.7684084020126339,
      "learning_rate": 2.5263191393815057e-06,
      "loss": 0.0284,
      "step": 12123
    },
    {
      "epoch": 7.756877799104287,
      "grad_norm": 0.6450300081445196,
      "learning_rate": 2.5249425395554817e-06,
      "loss": 0.0246,
      "step": 12124
    },
    {
      "epoch": 7.757517594369801,
      "grad_norm": 0.5705627627654338,
      "learning_rate": 2.523566260695132e-06,
      "loss": 0.0218,
      "step": 12125
    },
    {
      "epoch": 7.758157389635317,
      "grad_norm": 0.6016660602505429,
      "learning_rate": 2.5221903028595474e-06,
      "loss": 0.0228,
      "step": 12126
    },
    {
      "epoch": 7.758797184900832,
      "grad_norm": 0.5910912953368732,
      "learning_rate": 2.520814666107806e-06,
      "loss": 0.0185,
      "step": 12127
    },
    {
      "epoch": 7.759436980166347,
      "grad_norm": 0.624623150973433,
      "learning_rate": 2.519439350498981e-06,
      "loss": 0.0232,
      "step": 12128
    },
    {
      "epoch": 7.760076775431862,
      "grad_norm": 0.6119105730520693,
      "learning_rate": 2.5180643560921245e-06,
      "loss": 0.0219,
      "step": 12129
    },
    {
      "epoch": 7.760716570697377,
      "grad_norm": 0.6465929179744997,
      "learning_rate": 2.516689682946275e-06,
      "loss": 0.0215,
      "step": 12130
    },
    {
      "epoch": 7.761356365962892,
      "grad_norm": 0.634317908121759,
      "learning_rate": 2.515315331120455e-06,
      "loss": 0.0256,
      "step": 12131
    },
    {
      "epoch": 7.7619961612284065,
      "grad_norm": 0.7206407189990773,
      "learning_rate": 2.5139413006736835e-06,
      "loss": 0.022,
      "step": 12132
    },
    {
      "epoch": 7.762635956493922,
      "grad_norm": 0.6086619986043117,
      "learning_rate": 2.512567591664956e-06,
      "loss": 0.0212,
      "step": 12133
    },
    {
      "epoch": 7.763275751759437,
      "grad_norm": 0.672717472344836,
      "learning_rate": 2.511194204153258e-06,
      "loss": 0.0259,
      "step": 12134
    },
    {
      "epoch": 7.763915547024952,
      "grad_norm": 0.6450418233540886,
      "learning_rate": 2.509821138197558e-06,
      "loss": 0.0222,
      "step": 12135
    },
    {
      "epoch": 7.764555342290467,
      "grad_norm": 0.6929600386145224,
      "learning_rate": 2.508448393856815e-06,
      "loss": 0.0197,
      "step": 12136
    },
    {
      "epoch": 7.765195137555982,
      "grad_norm": 0.6494804006702884,
      "learning_rate": 2.507075971189968e-06,
      "loss": 0.0225,
      "step": 12137
    },
    {
      "epoch": 7.765834932821497,
      "grad_norm": 0.6219537522084715,
      "learning_rate": 2.5057038702559533e-06,
      "loss": 0.0212,
      "step": 12138
    },
    {
      "epoch": 7.766474728087012,
      "grad_norm": 0.6453744295612286,
      "learning_rate": 2.5043320911136836e-06,
      "loss": 0.0221,
      "step": 12139
    },
    {
      "epoch": 7.767114523352527,
      "grad_norm": 0.7366407351675139,
      "learning_rate": 2.5029606338220593e-06,
      "loss": 0.024,
      "step": 12140
    },
    {
      "epoch": 7.767754318618042,
      "grad_norm": 0.7143263355538275,
      "learning_rate": 2.501589498439967e-06,
      "loss": 0.0217,
      "step": 12141
    },
    {
      "epoch": 7.7683941138835575,
      "grad_norm": 0.6286516175103792,
      "learning_rate": 2.5002186850262857e-06,
      "loss": 0.0201,
      "step": 12142
    },
    {
      "epoch": 7.769033909149073,
      "grad_norm": 0.7556940092152965,
      "learning_rate": 2.4988481936398723e-06,
      "loss": 0.0235,
      "step": 12143
    },
    {
      "epoch": 7.769673704414587,
      "grad_norm": 0.7963415559479141,
      "learning_rate": 2.4974780243395714e-06,
      "loss": 0.0215,
      "step": 12144
    },
    {
      "epoch": 7.770313499680102,
      "grad_norm": 0.6434584613756575,
      "learning_rate": 2.496108177184222e-06,
      "loss": 0.0227,
      "step": 12145
    },
    {
      "epoch": 7.770953294945618,
      "grad_norm": 0.6918263118013925,
      "learning_rate": 2.4947386522326376e-06,
      "loss": 0.0255,
      "step": 12146
    },
    {
      "epoch": 7.771593090211132,
      "grad_norm": 0.752492316627432,
      "learning_rate": 2.4933694495436255e-06,
      "loss": 0.0257,
      "step": 12147
    },
    {
      "epoch": 7.772232885476647,
      "grad_norm": 0.7076020790214474,
      "learning_rate": 2.4920005691759765e-06,
      "loss": 0.0244,
      "step": 12148
    },
    {
      "epoch": 7.772872680742163,
      "grad_norm": 0.7107089550181482,
      "learning_rate": 2.490632011188466e-06,
      "loss": 0.0257,
      "step": 12149
    },
    {
      "epoch": 7.773512476007678,
      "grad_norm": 0.6052768587356837,
      "learning_rate": 2.4892637756398597e-06,
      "loss": 0.0182,
      "step": 12150
    },
    {
      "epoch": 7.774152271273192,
      "grad_norm": 0.5792984258351146,
      "learning_rate": 2.487895862588904e-06,
      "loss": 0.0195,
      "step": 12151
    },
    {
      "epoch": 7.774792066538708,
      "grad_norm": 0.6394947181645723,
      "learning_rate": 2.4865282720943386e-06,
      "loss": 0.0229,
      "step": 12152
    },
    {
      "epoch": 7.775431861804223,
      "grad_norm": 0.7320464359550927,
      "learning_rate": 2.4851610042148846e-06,
      "loss": 0.0251,
      "step": 12153
    },
    {
      "epoch": 7.776071657069737,
      "grad_norm": 0.7323148606301684,
      "learning_rate": 2.4837940590092458e-06,
      "loss": 0.0225,
      "step": 12154
    },
    {
      "epoch": 7.7767114523352525,
      "grad_norm": 0.6464613023256067,
      "learning_rate": 2.4824274365361225e-06,
      "loss": 0.0209,
      "step": 12155
    },
    {
      "epoch": 7.777351247600768,
      "grad_norm": 0.7231031115635091,
      "learning_rate": 2.4810611368541924e-06,
      "loss": 0.0216,
      "step": 12156
    },
    {
      "epoch": 7.777991042866283,
      "grad_norm": 0.7153120989660595,
      "learning_rate": 2.47969516002212e-06,
      "loss": 0.0254,
      "step": 12157
    },
    {
      "epoch": 7.7786308381317975,
      "grad_norm": 0.7558997312493314,
      "learning_rate": 2.478329506098562e-06,
      "loss": 0.0279,
      "step": 12158
    },
    {
      "epoch": 7.779270633397313,
      "grad_norm": 0.6268347196875727,
      "learning_rate": 2.4769641751421545e-06,
      "loss": 0.0189,
      "step": 12159
    },
    {
      "epoch": 7.779910428662828,
      "grad_norm": 0.5585454663220496,
      "learning_rate": 2.475599167211524e-06,
      "loss": 0.0184,
      "step": 12160
    },
    {
      "epoch": 7.780550223928343,
      "grad_norm": 0.7064768279535881,
      "learning_rate": 2.4742344823652773e-06,
      "loss": 0.025,
      "step": 12161
    },
    {
      "epoch": 7.781190019193858,
      "grad_norm": 0.6664558677918383,
      "learning_rate": 2.4728701206620165e-06,
      "loss": 0.023,
      "step": 12162
    },
    {
      "epoch": 7.781829814459373,
      "grad_norm": 0.6125397990444557,
      "learning_rate": 2.471506082160324e-06,
      "loss": 0.0198,
      "step": 12163
    },
    {
      "epoch": 7.782469609724888,
      "grad_norm": 0.786234715363562,
      "learning_rate": 2.4701423669187686e-06,
      "loss": 0.0237,
      "step": 12164
    },
    {
      "epoch": 7.7831094049904035,
      "grad_norm": 0.6946532300431156,
      "learning_rate": 2.4687789749959046e-06,
      "loss": 0.0262,
      "step": 12165
    },
    {
      "epoch": 7.783749200255918,
      "grad_norm": 0.49946019055235863,
      "learning_rate": 2.4674159064502744e-06,
      "loss": 0.0147,
      "step": 12166
    },
    {
      "epoch": 7.784388995521433,
      "grad_norm": 0.5853704624663684,
      "learning_rate": 2.4660531613404024e-06,
      "loss": 0.0207,
      "step": 12167
    },
    {
      "epoch": 7.785028790786948,
      "grad_norm": 0.5629815203287112,
      "learning_rate": 2.4646907397248098e-06,
      "loss": 0.0179,
      "step": 12168
    },
    {
      "epoch": 7.785668586052463,
      "grad_norm": 0.6072643367000414,
      "learning_rate": 2.4633286416619916e-06,
      "loss": 0.0205,
      "step": 12169
    },
    {
      "epoch": 7.786308381317978,
      "grad_norm": 0.5931866171070034,
      "learning_rate": 2.461966867210435e-06,
      "loss": 0.0225,
      "step": 12170
    },
    {
      "epoch": 7.786948176583493,
      "grad_norm": 0.60170493056873,
      "learning_rate": 2.4606054164286096e-06,
      "loss": 0.0217,
      "step": 12171
    },
    {
      "epoch": 7.787587971849009,
      "grad_norm": 0.6827828456307933,
      "learning_rate": 2.459244289374979e-06,
      "loss": 0.0241,
      "step": 12172
    },
    {
      "epoch": 7.788227767114523,
      "grad_norm": 0.6372649378095658,
      "learning_rate": 2.457883486107984e-06,
      "loss": 0.0208,
      "step": 12173
    },
    {
      "epoch": 7.788867562380038,
      "grad_norm": 0.5875770177940897,
      "learning_rate": 2.4565230066860534e-06,
      "loss": 0.0192,
      "step": 12174
    },
    {
      "epoch": 7.789507357645554,
      "grad_norm": 0.680964344707888,
      "learning_rate": 2.4551628511676086e-06,
      "loss": 0.0196,
      "step": 12175
    },
    {
      "epoch": 7.790147152911068,
      "grad_norm": 0.614093624082983,
      "learning_rate": 2.4538030196110497e-06,
      "loss": 0.0203,
      "step": 12176
    },
    {
      "epoch": 7.790786948176583,
      "grad_norm": 0.6975307163345035,
      "learning_rate": 2.452443512074766e-06,
      "loss": 0.0234,
      "step": 12177
    },
    {
      "epoch": 7.7914267434420985,
      "grad_norm": 0.7016765232326564,
      "learning_rate": 2.4510843286171317e-06,
      "loss": 0.0233,
      "step": 12178
    },
    {
      "epoch": 7.792066538707614,
      "grad_norm": 0.6375943659527525,
      "learning_rate": 2.4497254692965077e-06,
      "loss": 0.021,
      "step": 12179
    },
    {
      "epoch": 7.792706333973129,
      "grad_norm": 0.5538552964541885,
      "learning_rate": 2.4483669341712403e-06,
      "loss": 0.0167,
      "step": 12180
    },
    {
      "epoch": 7.7933461292386434,
      "grad_norm": 0.8211894321894374,
      "learning_rate": 2.447008723299662e-06,
      "loss": 0.0282,
      "step": 12181
    },
    {
      "epoch": 7.793985924504159,
      "grad_norm": 0.6925490753522326,
      "learning_rate": 2.4456508367400956e-06,
      "loss": 0.0228,
      "step": 12182
    },
    {
      "epoch": 7.794625719769674,
      "grad_norm": 0.6725132911043037,
      "learning_rate": 2.444293274550844e-06,
      "loss": 0.0222,
      "step": 12183
    },
    {
      "epoch": 7.795265515035188,
      "grad_norm": 0.7511025759435602,
      "learning_rate": 2.4429360367901956e-06,
      "loss": 0.0239,
      "step": 12184
    },
    {
      "epoch": 7.795905310300704,
      "grad_norm": 0.6423958488196391,
      "learning_rate": 2.4415791235164345e-06,
      "loss": 0.0202,
      "step": 12185
    },
    {
      "epoch": 7.796545105566219,
      "grad_norm": 0.6472258180888208,
      "learning_rate": 2.4402225347878196e-06,
      "loss": 0.02,
      "step": 12186
    },
    {
      "epoch": 7.797184900831734,
      "grad_norm": 0.6350295704263583,
      "learning_rate": 2.438866270662602e-06,
      "loss": 0.0243,
      "step": 12187
    },
    {
      "epoch": 7.797824696097249,
      "grad_norm": 0.6121579000483914,
      "learning_rate": 2.437510331199013e-06,
      "loss": 0.023,
      "step": 12188
    },
    {
      "epoch": 7.798464491362764,
      "grad_norm": 0.6904517687556737,
      "learning_rate": 2.436154716455281e-06,
      "loss": 0.0219,
      "step": 12189
    },
    {
      "epoch": 7.799104286628279,
      "grad_norm": 0.6363818299309291,
      "learning_rate": 2.4347994264896102e-06,
      "loss": 0.0218,
      "step": 12190
    },
    {
      "epoch": 7.7997440818937935,
      "grad_norm": 0.5860643175795409,
      "learning_rate": 2.4334444613601915e-06,
      "loss": 0.0223,
      "step": 12191
    },
    {
      "epoch": 7.800383877159309,
      "grad_norm": 0.7618498209307789,
      "learning_rate": 2.432089821125212e-06,
      "loss": 0.0235,
      "step": 12192
    },
    {
      "epoch": 7.801023672424824,
      "grad_norm": 0.7043439622817594,
      "learning_rate": 2.430735505842832e-06,
      "loss": 0.0239,
      "step": 12193
    },
    {
      "epoch": 7.801663467690339,
      "grad_norm": 0.730467191568084,
      "learning_rate": 2.429381515571205e-06,
      "loss": 0.024,
      "step": 12194
    },
    {
      "epoch": 7.802303262955854,
      "grad_norm": 0.7216671354274572,
      "learning_rate": 2.4280278503684685e-06,
      "loss": 0.0212,
      "step": 12195
    },
    {
      "epoch": 7.802943058221369,
      "grad_norm": 0.6867936687848528,
      "learning_rate": 2.426674510292747e-06,
      "loss": 0.0236,
      "step": 12196
    },
    {
      "epoch": 7.803582853486884,
      "grad_norm": 0.6717699449964268,
      "learning_rate": 2.4253214954021497e-06,
      "loss": 0.0223,
      "step": 12197
    },
    {
      "epoch": 7.8042226487524,
      "grad_norm": 0.7353501748348269,
      "learning_rate": 2.4239688057547705e-06,
      "loss": 0.0215,
      "step": 12198
    },
    {
      "epoch": 7.804862444017914,
      "grad_norm": 0.6488934771250229,
      "learning_rate": 2.422616441408696e-06,
      "loss": 0.0215,
      "step": 12199
    },
    {
      "epoch": 7.805502239283429,
      "grad_norm": 0.6725129319701673,
      "learning_rate": 2.4212644024219933e-06,
      "loss": 0.0221,
      "step": 12200
    },
    {
      "epoch": 7.8061420345489445,
      "grad_norm": 0.6876878782071557,
      "learning_rate": 2.4199126888527125e-06,
      "loss": 0.0234,
      "step": 12201
    },
    {
      "epoch": 7.80678182981446,
      "grad_norm": 0.620077185233817,
      "learning_rate": 2.4185613007589003e-06,
      "loss": 0.0218,
      "step": 12202
    },
    {
      "epoch": 7.807421625079974,
      "grad_norm": 0.7126853355373332,
      "learning_rate": 2.4172102381985797e-06,
      "loss": 0.0258,
      "step": 12203
    },
    {
      "epoch": 7.8080614203454894,
      "grad_norm": 0.6799371192419872,
      "learning_rate": 2.4158595012297594e-06,
      "loss": 0.0187,
      "step": 12204
    },
    {
      "epoch": 7.808701215611005,
      "grad_norm": 0.596845292919448,
      "learning_rate": 2.414509089910443e-06,
      "loss": 0.0206,
      "step": 12205
    },
    {
      "epoch": 7.809341010876519,
      "grad_norm": 0.8311976739028942,
      "learning_rate": 2.4131590042986142e-06,
      "loss": 0.0288,
      "step": 12206
    },
    {
      "epoch": 7.809980806142034,
      "grad_norm": 0.7356973695168538,
      "learning_rate": 2.411809244452241e-06,
      "loss": 0.0207,
      "step": 12207
    },
    {
      "epoch": 7.81062060140755,
      "grad_norm": 0.8156938978247888,
      "learning_rate": 2.4104598104292807e-06,
      "loss": 0.0225,
      "step": 12208
    },
    {
      "epoch": 7.811260396673065,
      "grad_norm": 0.656089347708617,
      "learning_rate": 2.409110702287675e-06,
      "loss": 0.0222,
      "step": 12209
    },
    {
      "epoch": 7.811900191938579,
      "grad_norm": 0.7512071836756178,
      "learning_rate": 2.407761920085353e-06,
      "loss": 0.0234,
      "step": 12210
    },
    {
      "epoch": 7.812539987204095,
      "grad_norm": 0.7545598141189427,
      "learning_rate": 2.4064134638802262e-06,
      "loss": 0.0238,
      "step": 12211
    },
    {
      "epoch": 7.81317978246961,
      "grad_norm": 0.57245466589661,
      "learning_rate": 2.4050653337301987e-06,
      "loss": 0.0175,
      "step": 12212
    },
    {
      "epoch": 7.813819577735125,
      "grad_norm": 0.7127551279478249,
      "learning_rate": 2.403717529693157e-06,
      "loss": 0.023,
      "step": 12213
    },
    {
      "epoch": 7.8144593730006395,
      "grad_norm": 0.5830330154611049,
      "learning_rate": 2.4023700518269675e-06,
      "loss": 0.0187,
      "step": 12214
    },
    {
      "epoch": 7.815099168266155,
      "grad_norm": 0.7008208900624813,
      "learning_rate": 2.401022900189497e-06,
      "loss": 0.0251,
      "step": 12215
    },
    {
      "epoch": 7.81573896353167,
      "grad_norm": 0.7104506594041348,
      "learning_rate": 2.399676074838585e-06,
      "loss": 0.0226,
      "step": 12216
    },
    {
      "epoch": 7.816378758797185,
      "grad_norm": 0.6908230400328818,
      "learning_rate": 2.3983295758320623e-06,
      "loss": 0.0244,
      "step": 12217
    },
    {
      "epoch": 7.8170185540627,
      "grad_norm": 0.6128017807864656,
      "learning_rate": 2.3969834032277427e-06,
      "loss": 0.0206,
      "step": 12218
    },
    {
      "epoch": 7.817658349328215,
      "grad_norm": 0.7725188849672624,
      "learning_rate": 2.3956375570834346e-06,
      "loss": 0.0239,
      "step": 12219
    },
    {
      "epoch": 7.81829814459373,
      "grad_norm": 0.5850328290639356,
      "learning_rate": 2.394292037456922e-06,
      "loss": 0.0204,
      "step": 12220
    },
    {
      "epoch": 7.818937939859245,
      "grad_norm": 0.6960549348093125,
      "learning_rate": 2.3929468444059766e-06,
      "loss": 0.0212,
      "step": 12221
    },
    {
      "epoch": 7.81957773512476,
      "grad_norm": 0.638462418585801,
      "learning_rate": 2.391601977988366e-06,
      "loss": 0.0199,
      "step": 12222
    },
    {
      "epoch": 7.820217530390275,
      "grad_norm": 0.686485877589621,
      "learning_rate": 2.3902574382618304e-06,
      "loss": 0.0226,
      "step": 12223
    },
    {
      "epoch": 7.8208573256557905,
      "grad_norm": 0.8244883161938208,
      "learning_rate": 2.388913225284105e-06,
      "loss": 0.0278,
      "step": 12224
    },
    {
      "epoch": 7.821497120921305,
      "grad_norm": 0.5903576929538216,
      "learning_rate": 2.387569339112906e-06,
      "loss": 0.0197,
      "step": 12225
    },
    {
      "epoch": 7.82213691618682,
      "grad_norm": 0.6157156767163157,
      "learning_rate": 2.3862257798059384e-06,
      "loss": 0.0219,
      "step": 12226
    },
    {
      "epoch": 7.822776711452335,
      "grad_norm": 0.6712022024094506,
      "learning_rate": 2.3848825474208912e-06,
      "loss": 0.0226,
      "step": 12227
    },
    {
      "epoch": 7.82341650671785,
      "grad_norm": 0.867695674808315,
      "learning_rate": 2.383539642015439e-06,
      "loss": 0.0302,
      "step": 12228
    },
    {
      "epoch": 7.824056301983365,
      "grad_norm": 0.7121113255961969,
      "learning_rate": 2.382197063647249e-06,
      "loss": 0.0209,
      "step": 12229
    },
    {
      "epoch": 7.82469609724888,
      "grad_norm": 0.6275037596978045,
      "learning_rate": 2.3808548123739662e-06,
      "loss": 0.0222,
      "step": 12230
    },
    {
      "epoch": 7.825335892514396,
      "grad_norm": 0.7067869676425641,
      "learning_rate": 2.3795128882532205e-06,
      "loss": 0.0241,
      "step": 12231
    },
    {
      "epoch": 7.825975687779911,
      "grad_norm": 0.7270807866570541,
      "learning_rate": 2.3781712913426393e-06,
      "loss": 0.023,
      "step": 12232
    },
    {
      "epoch": 7.826615483045425,
      "grad_norm": 0.7054871602220649,
      "learning_rate": 2.376830021699824e-06,
      "loss": 0.0232,
      "step": 12233
    },
    {
      "epoch": 7.827255278310941,
      "grad_norm": 0.6805784346789071,
      "learning_rate": 2.375489079382365e-06,
      "loss": 0.0191,
      "step": 12234
    },
    {
      "epoch": 7.827895073576456,
      "grad_norm": 0.6938265216445622,
      "learning_rate": 2.374148464447844e-06,
      "loss": 0.0223,
      "step": 12235
    },
    {
      "epoch": 7.82853486884197,
      "grad_norm": 0.7947232661190042,
      "learning_rate": 2.3728081769538235e-06,
      "loss": 0.028,
      "step": 12236
    },
    {
      "epoch": 7.8291746641074855,
      "grad_norm": 0.68957893232724,
      "learning_rate": 2.371468216957852e-06,
      "loss": 0.0242,
      "step": 12237
    },
    {
      "epoch": 7.829814459373001,
      "grad_norm": 0.6591979931163908,
      "learning_rate": 2.370128584517465e-06,
      "loss": 0.0185,
      "step": 12238
    },
    {
      "epoch": 7.830454254638516,
      "grad_norm": 0.8853749657481427,
      "learning_rate": 2.3687892796901835e-06,
      "loss": 0.0212,
      "step": 12239
    },
    {
      "epoch": 7.8310940499040305,
      "grad_norm": 0.6126473758697196,
      "learning_rate": 2.367450302533516e-06,
      "loss": 0.02,
      "step": 12240
    },
    {
      "epoch": 7.831733845169546,
      "grad_norm": 0.5461962836889117,
      "learning_rate": 2.3661116531049524e-06,
      "loss": 0.0181,
      "step": 12241
    },
    {
      "epoch": 7.832373640435061,
      "grad_norm": 0.7519104833438176,
      "learning_rate": 2.364773331461977e-06,
      "loss": 0.0238,
      "step": 12242
    },
    {
      "epoch": 7.833013435700575,
      "grad_norm": 0.7073546316192821,
      "learning_rate": 2.3634353376620535e-06,
      "loss": 0.0231,
      "step": 12243
    },
    {
      "epoch": 7.833653230966091,
      "grad_norm": 0.6838057590553742,
      "learning_rate": 2.3620976717626286e-06,
      "loss": 0.0239,
      "step": 12244
    },
    {
      "epoch": 7.834293026231606,
      "grad_norm": 0.6664124737265568,
      "learning_rate": 2.360760333821146e-06,
      "loss": 0.0219,
      "step": 12245
    },
    {
      "epoch": 7.834932821497121,
      "grad_norm": 0.6261074785758781,
      "learning_rate": 2.359423323895026e-06,
      "loss": 0.0171,
      "step": 12246
    },
    {
      "epoch": 7.835572616762636,
      "grad_norm": 0.7578941474425354,
      "learning_rate": 2.358086642041677e-06,
      "loss": 0.0217,
      "step": 12247
    },
    {
      "epoch": 7.836212412028151,
      "grad_norm": 0.669632910257975,
      "learning_rate": 2.35675028831849e-06,
      "loss": 0.0232,
      "step": 12248
    },
    {
      "epoch": 7.836852207293666,
      "grad_norm": 0.6834092188351738,
      "learning_rate": 2.3554142627828535e-06,
      "loss": 0.0247,
      "step": 12249
    },
    {
      "epoch": 7.837492002559181,
      "grad_norm": 0.6960043631668924,
      "learning_rate": 2.3540785654921284e-06,
      "loss": 0.0196,
      "step": 12250
    },
    {
      "epoch": 7.838131797824696,
      "grad_norm": 0.6873423807488137,
      "learning_rate": 2.3527431965036673e-06,
      "loss": 0.0226,
      "step": 12251
    },
    {
      "epoch": 7.838771593090211,
      "grad_norm": 0.8013840438948545,
      "learning_rate": 2.3514081558748115e-06,
      "loss": 0.0261,
      "step": 12252
    },
    {
      "epoch": 7.839411388355726,
      "grad_norm": 0.8295804787880853,
      "learning_rate": 2.3500734436628847e-06,
      "loss": 0.0263,
      "step": 12253
    },
    {
      "epoch": 7.840051183621242,
      "grad_norm": 0.7421311831869545,
      "learning_rate": 2.3487390599251946e-06,
      "loss": 0.0239,
      "step": 12254
    },
    {
      "epoch": 7.840690978886756,
      "grad_norm": 0.6938697378666302,
      "learning_rate": 2.34740500471904e-06,
      "loss": 0.0246,
      "step": 12255
    },
    {
      "epoch": 7.841330774152271,
      "grad_norm": 0.7099737919354097,
      "learning_rate": 2.3460712781017016e-06,
      "loss": 0.0243,
      "step": 12256
    },
    {
      "epoch": 7.841970569417787,
      "grad_norm": 0.6875950624351862,
      "learning_rate": 2.3447378801304464e-06,
      "loss": 0.0241,
      "step": 12257
    },
    {
      "epoch": 7.842610364683301,
      "grad_norm": 0.6168711668736793,
      "learning_rate": 2.3434048108625275e-06,
      "loss": 0.0199,
      "step": 12258
    },
    {
      "epoch": 7.843250159948816,
      "grad_norm": 0.6325532810010096,
      "learning_rate": 2.342072070355187e-06,
      "loss": 0.0229,
      "step": 12259
    },
    {
      "epoch": 7.8438899552143315,
      "grad_norm": 0.8393730616705632,
      "learning_rate": 2.3407396586656505e-06,
      "loss": 0.0281,
      "step": 12260
    },
    {
      "epoch": 7.844529750479847,
      "grad_norm": 0.6533004784562876,
      "learning_rate": 2.339407575851126e-06,
      "loss": 0.0224,
      "step": 12261
    },
    {
      "epoch": 7.845169545745361,
      "grad_norm": 0.6598896823357597,
      "learning_rate": 2.338075821968815e-06,
      "loss": 0.0188,
      "step": 12262
    },
    {
      "epoch": 7.8458093410108765,
      "grad_norm": 0.580846423785414,
      "learning_rate": 2.3367443970758984e-06,
      "loss": 0.0196,
      "step": 12263
    },
    {
      "epoch": 7.846449136276392,
      "grad_norm": 0.6857400432671135,
      "learning_rate": 2.3354133012295468e-06,
      "loss": 0.0226,
      "step": 12264
    },
    {
      "epoch": 7.847088931541906,
      "grad_norm": 0.5054078455324428,
      "learning_rate": 2.33408253448691e-06,
      "loss": 0.0186,
      "step": 12265
    },
    {
      "epoch": 7.847728726807421,
      "grad_norm": 0.6802289162763299,
      "learning_rate": 2.3327520969051354e-06,
      "loss": 0.025,
      "step": 12266
    },
    {
      "epoch": 7.848368522072937,
      "grad_norm": 0.7300598940163419,
      "learning_rate": 2.3314219885413468e-06,
      "loss": 0.0242,
      "step": 12267
    },
    {
      "epoch": 7.849008317338452,
      "grad_norm": 0.6243014071418418,
      "learning_rate": 2.330092209452657e-06,
      "loss": 0.0226,
      "step": 12268
    },
    {
      "epoch": 7.849648112603967,
      "grad_norm": 0.8130485773423253,
      "learning_rate": 2.328762759696164e-06,
      "loss": 0.0262,
      "step": 12269
    },
    {
      "epoch": 7.850287907869482,
      "grad_norm": 0.603743598882866,
      "learning_rate": 2.327433639328952e-06,
      "loss": 0.0195,
      "step": 12270
    },
    {
      "epoch": 7.850927703134997,
      "grad_norm": 0.7331715474843364,
      "learning_rate": 2.3261048484080885e-06,
      "loss": 0.0226,
      "step": 12271
    },
    {
      "epoch": 7.851567498400512,
      "grad_norm": 0.692433919759694,
      "learning_rate": 2.3247763869906337e-06,
      "loss": 0.0197,
      "step": 12272
    },
    {
      "epoch": 7.8522072936660265,
      "grad_norm": 0.7548226200802937,
      "learning_rate": 2.323448255133629e-06,
      "loss": 0.0268,
      "step": 12273
    },
    {
      "epoch": 7.852847088931542,
      "grad_norm": 0.6284002629713998,
      "learning_rate": 2.3221204528941e-06,
      "loss": 0.0192,
      "step": 12274
    },
    {
      "epoch": 7.853486884197057,
      "grad_norm": 0.7020803213937238,
      "learning_rate": 2.3207929803290573e-06,
      "loss": 0.0251,
      "step": 12275
    },
    {
      "epoch": 7.854126679462572,
      "grad_norm": 0.6111684516110357,
      "learning_rate": 2.319465837495508e-06,
      "loss": 0.022,
      "step": 12276
    },
    {
      "epoch": 7.854766474728087,
      "grad_norm": 1.404609779666455,
      "learning_rate": 2.318139024450433e-06,
      "loss": 0.0213,
      "step": 12277
    },
    {
      "epoch": 7.855406269993602,
      "grad_norm": 0.6899105957968956,
      "learning_rate": 2.3168125412507993e-06,
      "loss": 0.021,
      "step": 12278
    },
    {
      "epoch": 7.856046065259117,
      "grad_norm": 0.671305144288533,
      "learning_rate": 2.3154863879535714e-06,
      "loss": 0.0216,
      "step": 12279
    },
    {
      "epoch": 7.856685860524632,
      "grad_norm": 0.8240152639494236,
      "learning_rate": 2.3141605646156882e-06,
      "loss": 0.0227,
      "step": 12280
    },
    {
      "epoch": 7.857325655790147,
      "grad_norm": 0.7177616141633802,
      "learning_rate": 2.3128350712940763e-06,
      "loss": 0.0221,
      "step": 12281
    },
    {
      "epoch": 7.857965451055662,
      "grad_norm": 0.6800533465107309,
      "learning_rate": 2.311509908045654e-06,
      "loss": 0.0209,
      "step": 12282
    },
    {
      "epoch": 7.8586052463211775,
      "grad_norm": 0.8167461538083598,
      "learning_rate": 2.310185074927319e-06,
      "loss": 0.0257,
      "step": 12283
    },
    {
      "epoch": 7.859245041586692,
      "grad_norm": 0.7013831462542893,
      "learning_rate": 2.3088605719959577e-06,
      "loss": 0.0248,
      "step": 12284
    },
    {
      "epoch": 7.859884836852207,
      "grad_norm": 0.6882492517875707,
      "learning_rate": 2.307536399308443e-06,
      "loss": 0.0194,
      "step": 12285
    },
    {
      "epoch": 7.8605246321177225,
      "grad_norm": 0.7334507342652414,
      "learning_rate": 2.306212556921631e-06,
      "loss": 0.0264,
      "step": 12286
    },
    {
      "epoch": 7.861164427383238,
      "grad_norm": 0.6700646096915053,
      "learning_rate": 2.3048890448923655e-06,
      "loss": 0.0199,
      "step": 12287
    },
    {
      "epoch": 7.861804222648752,
      "grad_norm": 0.618809879533919,
      "learning_rate": 2.303565863277474e-06,
      "loss": 0.0192,
      "step": 12288
    },
    {
      "epoch": 7.862444017914267,
      "grad_norm": 0.6609750147745183,
      "learning_rate": 2.302243012133775e-06,
      "loss": 0.021,
      "step": 12289
    },
    {
      "epoch": 7.863083813179783,
      "grad_norm": 0.7427584737797922,
      "learning_rate": 2.3009204915180684e-06,
      "loss": 0.027,
      "step": 12290
    },
    {
      "epoch": 7.863723608445298,
      "grad_norm": 0.6425611191159992,
      "learning_rate": 2.299598301487137e-06,
      "loss": 0.021,
      "step": 12291
    },
    {
      "epoch": 7.864363403710812,
      "grad_norm": 0.6916605833051123,
      "learning_rate": 2.29827644209776e-06,
      "loss": 0.0228,
      "step": 12292
    },
    {
      "epoch": 7.865003198976328,
      "grad_norm": 0.7176335644745937,
      "learning_rate": 2.2969549134066927e-06,
      "loss": 0.0211,
      "step": 12293
    },
    {
      "epoch": 7.865642994241843,
      "grad_norm": 0.6487229647253904,
      "learning_rate": 2.2956337154706777e-06,
      "loss": 0.0213,
      "step": 12294
    },
    {
      "epoch": 7.866282789507357,
      "grad_norm": 0.6800657901899974,
      "learning_rate": 2.2943128483464426e-06,
      "loss": 0.0229,
      "step": 12295
    },
    {
      "epoch": 7.8669225847728725,
      "grad_norm": 0.77804349550238,
      "learning_rate": 2.2929923120907104e-06,
      "loss": 0.0266,
      "step": 12296
    },
    {
      "epoch": 7.867562380038388,
      "grad_norm": 0.6906600632179467,
      "learning_rate": 2.291672106760179e-06,
      "loss": 0.0211,
      "step": 12297
    },
    {
      "epoch": 7.868202175303903,
      "grad_norm": 0.6325684569466853,
      "learning_rate": 2.2903522324115344e-06,
      "loss": 0.0203,
      "step": 12298
    },
    {
      "epoch": 7.8688419705694175,
      "grad_norm": 0.6355563543911172,
      "learning_rate": 2.2890326891014516e-06,
      "loss": 0.0218,
      "step": 12299
    },
    {
      "epoch": 7.869481765834933,
      "grad_norm": 0.6614717060892561,
      "learning_rate": 2.287713476886587e-06,
      "loss": 0.0216,
      "step": 12300
    },
    {
      "epoch": 7.870121561100448,
      "grad_norm": 0.6924773001959604,
      "learning_rate": 2.2863945958235843e-06,
      "loss": 0.0231,
      "step": 12301
    },
    {
      "epoch": 7.870761356365962,
      "grad_norm": 0.6435648859698008,
      "learning_rate": 2.285076045969079e-06,
      "loss": 0.0214,
      "step": 12302
    },
    {
      "epoch": 7.871401151631478,
      "grad_norm": 0.6361698484274039,
      "learning_rate": 2.2837578273796847e-06,
      "loss": 0.0193,
      "step": 12303
    },
    {
      "epoch": 7.872040946896993,
      "grad_norm": 0.6395010652577267,
      "learning_rate": 2.2824399401120036e-06,
      "loss": 0.0204,
      "step": 12304
    },
    {
      "epoch": 7.872680742162508,
      "grad_norm": 0.7688139097143861,
      "learning_rate": 2.2811223842226193e-06,
      "loss": 0.0267,
      "step": 12305
    },
    {
      "epoch": 7.8733205374280235,
      "grad_norm": 0.6779861389410332,
      "learning_rate": 2.2798051597681127e-06,
      "loss": 0.0235,
      "step": 12306
    },
    {
      "epoch": 7.873960332693538,
      "grad_norm": 0.7102524524449086,
      "learning_rate": 2.2784882668050378e-06,
      "loss": 0.0216,
      "step": 12307
    },
    {
      "epoch": 7.874600127959053,
      "grad_norm": 0.6610939638899511,
      "learning_rate": 2.2771717053899402e-06,
      "loss": 0.0191,
      "step": 12308
    },
    {
      "epoch": 7.8752399232245685,
      "grad_norm": 0.6706706669311446,
      "learning_rate": 2.2758554755793536e-06,
      "loss": 0.0229,
      "step": 12309
    },
    {
      "epoch": 7.875879718490083,
      "grad_norm": 0.6704814632306221,
      "learning_rate": 2.274539577429793e-06,
      "loss": 0.024,
      "step": 12310
    },
    {
      "epoch": 7.876519513755598,
      "grad_norm": 0.6209488860457509,
      "learning_rate": 2.2732240109977575e-06,
      "loss": 0.0207,
      "step": 12311
    },
    {
      "epoch": 7.877159309021113,
      "grad_norm": 0.6946575938972263,
      "learning_rate": 2.2719087763397405e-06,
      "loss": 0.0228,
      "step": 12312
    },
    {
      "epoch": 7.877799104286629,
      "grad_norm": 0.5908015362256193,
      "learning_rate": 2.2705938735122144e-06,
      "loss": 0.0207,
      "step": 12313
    },
    {
      "epoch": 7.878438899552143,
      "grad_norm": 0.6227778213633047,
      "learning_rate": 2.2692793025716377e-06,
      "loss": 0.0195,
      "step": 12314
    },
    {
      "epoch": 7.879078694817658,
      "grad_norm": 0.6442719643677457,
      "learning_rate": 2.2679650635744553e-06,
      "loss": 0.0209,
      "step": 12315
    },
    {
      "epoch": 7.879718490083174,
      "grad_norm": 0.7495377586773375,
      "learning_rate": 2.266651156577099e-06,
      "loss": 0.0254,
      "step": 12316
    },
    {
      "epoch": 7.880358285348688,
      "grad_norm": 0.6441398822802572,
      "learning_rate": 2.2653375816359845e-06,
      "loss": 0.0206,
      "step": 12317
    },
    {
      "epoch": 7.880998080614203,
      "grad_norm": 0.6901873081805351,
      "learning_rate": 2.2640243388075144e-06,
      "loss": 0.0203,
      "step": 12318
    },
    {
      "epoch": 7.8816378758797185,
      "grad_norm": 0.6616556074214582,
      "learning_rate": 2.26271142814808e-06,
      "loss": 0.0215,
      "step": 12319
    },
    {
      "epoch": 7.882277671145234,
      "grad_norm": 0.6278029113978424,
      "learning_rate": 2.2613988497140526e-06,
      "loss": 0.0208,
      "step": 12320
    },
    {
      "epoch": 7.882917466410748,
      "grad_norm": 0.6011511218718707,
      "learning_rate": 2.2600866035617908e-06,
      "loss": 0.0201,
      "step": 12321
    },
    {
      "epoch": 7.8835572616762635,
      "grad_norm": 0.6350848877521786,
      "learning_rate": 2.258774689747645e-06,
      "loss": 0.019,
      "step": 12322
    },
    {
      "epoch": 7.884197056941779,
      "grad_norm": 0.6182386094512218,
      "learning_rate": 2.257463108327942e-06,
      "loss": 0.0211,
      "step": 12323
    },
    {
      "epoch": 7.884836852207294,
      "grad_norm": 0.6545936000093096,
      "learning_rate": 2.2561518593590014e-06,
      "loss": 0.0212,
      "step": 12324
    },
    {
      "epoch": 7.885476647472808,
      "grad_norm": 0.6170600724414689,
      "learning_rate": 2.254840942897122e-06,
      "loss": 0.0226,
      "step": 12325
    },
    {
      "epoch": 7.886116442738324,
      "grad_norm": 0.6956349862071399,
      "learning_rate": 2.2535303589985967e-06,
      "loss": 0.0214,
      "step": 12326
    },
    {
      "epoch": 7.886756238003839,
      "grad_norm": 0.6109416558037813,
      "learning_rate": 2.2522201077196993e-06,
      "loss": 0.0206,
      "step": 12327
    },
    {
      "epoch": 7.887396033269354,
      "grad_norm": 0.6513177247752983,
      "learning_rate": 2.250910189116687e-06,
      "loss": 0.0226,
      "step": 12328
    },
    {
      "epoch": 7.888035828534869,
      "grad_norm": 0.6102533534005428,
      "learning_rate": 2.2496006032458075e-06,
      "loss": 0.0215,
      "step": 12329
    },
    {
      "epoch": 7.888675623800384,
      "grad_norm": 0.8492075843726956,
      "learning_rate": 2.248291350163291e-06,
      "loss": 0.0263,
      "step": 12330
    },
    {
      "epoch": 7.889315419065899,
      "grad_norm": 0.6674001404630857,
      "learning_rate": 2.2469824299253538e-06,
      "loss": 0.0197,
      "step": 12331
    },
    {
      "epoch": 7.889955214331414,
      "grad_norm": 0.6653494982416334,
      "learning_rate": 2.2456738425881976e-06,
      "loss": 0.0244,
      "step": 12332
    },
    {
      "epoch": 7.890595009596929,
      "grad_norm": 0.7856428165142019,
      "learning_rate": 2.2443655882080162e-06,
      "loss": 0.0265,
      "step": 12333
    },
    {
      "epoch": 7.891234804862444,
      "grad_norm": 0.6250007243952342,
      "learning_rate": 2.2430576668409785e-06,
      "loss": 0.0222,
      "step": 12334
    },
    {
      "epoch": 7.891874600127959,
      "grad_norm": 0.6748929738377827,
      "learning_rate": 2.2417500785432444e-06,
      "loss": 0.0232,
      "step": 12335
    },
    {
      "epoch": 7.892514395393474,
      "grad_norm": 0.6852940011563847,
      "learning_rate": 2.240442823370964e-06,
      "loss": 0.0227,
      "step": 12336
    },
    {
      "epoch": 7.893154190658989,
      "grad_norm": 0.7486837181597175,
      "learning_rate": 2.239135901380265e-06,
      "loss": 0.0245,
      "step": 12337
    },
    {
      "epoch": 7.893793985924504,
      "grad_norm": 0.5887752195047741,
      "learning_rate": 2.2378293126272622e-06,
      "loss": 0.0177,
      "step": 12338
    },
    {
      "epoch": 7.894433781190019,
      "grad_norm": 0.6861885593176543,
      "learning_rate": 2.2365230571680642e-06,
      "loss": 0.0223,
      "step": 12339
    },
    {
      "epoch": 7.895073576455534,
      "grad_norm": 0.6571645214220317,
      "learning_rate": 2.235217135058756e-06,
      "loss": 0.0222,
      "step": 12340
    },
    {
      "epoch": 7.895713371721049,
      "grad_norm": 0.7008157990539106,
      "learning_rate": 2.2339115463554104e-06,
      "loss": 0.0225,
      "step": 12341
    },
    {
      "epoch": 7.8963531669865645,
      "grad_norm": 0.7299380733848048,
      "learning_rate": 2.2326062911140865e-06,
      "loss": 0.0234,
      "step": 12342
    },
    {
      "epoch": 7.89699296225208,
      "grad_norm": 0.6975262219731435,
      "learning_rate": 2.2313013693908336e-06,
      "loss": 0.0203,
      "step": 12343
    },
    {
      "epoch": 7.897632757517594,
      "grad_norm": 0.5954212086385932,
      "learning_rate": 2.2299967812416813e-06,
      "loss": 0.0164,
      "step": 12344
    },
    {
      "epoch": 7.8982725527831095,
      "grad_norm": 0.6612755378948465,
      "learning_rate": 2.2286925267226446e-06,
      "loss": 0.0218,
      "step": 12345
    },
    {
      "epoch": 7.898912348048625,
      "grad_norm": 0.7694907767292404,
      "learning_rate": 2.2273886058897265e-06,
      "loss": 0.0267,
      "step": 12346
    },
    {
      "epoch": 7.899552143314139,
      "grad_norm": 0.7189435937843183,
      "learning_rate": 2.226085018798916e-06,
      "loss": 0.0235,
      "step": 12347
    },
    {
      "epoch": 7.900191938579654,
      "grad_norm": 0.7000004365026223,
      "learning_rate": 2.224781765506182e-06,
      "loss": 0.0231,
      "step": 12348
    },
    {
      "epoch": 7.90083173384517,
      "grad_norm": 0.6718974029658509,
      "learning_rate": 2.223478846067493e-06,
      "loss": 0.0235,
      "step": 12349
    },
    {
      "epoch": 7.901471529110685,
      "grad_norm": 0.6361386768758205,
      "learning_rate": 2.2221762605387866e-06,
      "loss": 0.0182,
      "step": 12350
    },
    {
      "epoch": 7.902111324376199,
      "grad_norm": 0.6289180054818321,
      "learning_rate": 2.2208740089759975e-06,
      "loss": 0.0208,
      "step": 12351
    },
    {
      "epoch": 7.902751119641715,
      "grad_norm": 0.690651250520683,
      "learning_rate": 2.219572091435037e-06,
      "loss": 0.0212,
      "step": 12352
    },
    {
      "epoch": 7.90339091490723,
      "grad_norm": 0.65870619007797,
      "learning_rate": 2.2182705079718135e-06,
      "loss": 0.0186,
      "step": 12353
    },
    {
      "epoch": 7.904030710172744,
      "grad_norm": 0.7057093542990347,
      "learning_rate": 2.216969258642212e-06,
      "loss": 0.0221,
      "step": 12354
    },
    {
      "epoch": 7.9046705054382596,
      "grad_norm": 0.6527568257826862,
      "learning_rate": 2.215668343502102e-06,
      "loss": 0.0216,
      "step": 12355
    },
    {
      "epoch": 7.905310300703775,
      "grad_norm": 0.6413579687123526,
      "learning_rate": 2.2143677626073502e-06,
      "loss": 0.0195,
      "step": 12356
    },
    {
      "epoch": 7.90595009596929,
      "grad_norm": 0.7621675152730454,
      "learning_rate": 2.213067516013796e-06,
      "loss": 0.0268,
      "step": 12357
    },
    {
      "epoch": 7.9065898912348045,
      "grad_norm": 0.8227082167851154,
      "learning_rate": 2.2117676037772705e-06,
      "loss": 0.02,
      "step": 12358
    },
    {
      "epoch": 7.90722968650032,
      "grad_norm": 0.5934120466517009,
      "learning_rate": 2.2104680259535916e-06,
      "loss": 0.0195,
      "step": 12359
    },
    {
      "epoch": 7.907869481765835,
      "grad_norm": 0.7102516209962534,
      "learning_rate": 2.2091687825985575e-06,
      "loss": 0.0221,
      "step": 12360
    },
    {
      "epoch": 7.90850927703135,
      "grad_norm": 0.6315338741409725,
      "learning_rate": 2.207869873767958e-06,
      "loss": 0.0188,
      "step": 12361
    },
    {
      "epoch": 7.909149072296865,
      "grad_norm": 0.7247468708868856,
      "learning_rate": 2.2065712995175616e-06,
      "loss": 0.0268,
      "step": 12362
    },
    {
      "epoch": 7.90978886756238,
      "grad_norm": 0.702874648677657,
      "learning_rate": 2.2052730599031337e-06,
      "loss": 0.0196,
      "step": 12363
    },
    {
      "epoch": 7.910428662827895,
      "grad_norm": 0.6186466738958128,
      "learning_rate": 2.2039751549804147e-06,
      "loss": 0.0206,
      "step": 12364
    },
    {
      "epoch": 7.9110684580934105,
      "grad_norm": 0.6311711976403993,
      "learning_rate": 2.2026775848051308e-06,
      "loss": 0.0213,
      "step": 12365
    },
    {
      "epoch": 7.911708253358925,
      "grad_norm": 0.714289985614982,
      "learning_rate": 2.2013803494330045e-06,
      "loss": 0.024,
      "step": 12366
    },
    {
      "epoch": 7.91234804862444,
      "grad_norm": 0.651462157803175,
      "learning_rate": 2.200083448919733e-06,
      "loss": 0.022,
      "step": 12367
    },
    {
      "epoch": 7.9129878438899555,
      "grad_norm": 0.695437630216662,
      "learning_rate": 2.1987868833210013e-06,
      "loss": 0.0232,
      "step": 12368
    },
    {
      "epoch": 7.91362763915547,
      "grad_norm": 0.6185252860547654,
      "learning_rate": 2.1974906526924855e-06,
      "loss": 0.0213,
      "step": 12369
    },
    {
      "epoch": 7.914267434420985,
      "grad_norm": 0.6858254853397484,
      "learning_rate": 2.196194757089841e-06,
      "loss": 0.0196,
      "step": 12370
    },
    {
      "epoch": 7.9149072296865,
      "grad_norm": 0.6855960150938222,
      "learning_rate": 2.1948991965687127e-06,
      "loss": 0.0204,
      "step": 12371
    },
    {
      "epoch": 7.915547024952016,
      "grad_norm": 0.6603754498627045,
      "learning_rate": 2.1936039711847258e-06,
      "loss": 0.0213,
      "step": 12372
    },
    {
      "epoch": 7.91618682021753,
      "grad_norm": 0.7170438517870972,
      "learning_rate": 2.192309080993501e-06,
      "loss": 0.0244,
      "step": 12373
    },
    {
      "epoch": 7.916826615483045,
      "grad_norm": 0.591848403589971,
      "learning_rate": 2.191014526050634e-06,
      "loss": 0.0182,
      "step": 12374
    },
    {
      "epoch": 7.917466410748561,
      "grad_norm": 0.6385477360185727,
      "learning_rate": 2.1897203064117132e-06,
      "loss": 0.02,
      "step": 12375
    },
    {
      "epoch": 7.918106206014075,
      "grad_norm": 0.6430575756319166,
      "learning_rate": 2.188426422132309e-06,
      "loss": 0.0193,
      "step": 12376
    },
    {
      "epoch": 7.91874600127959,
      "grad_norm": 0.6695969151173072,
      "learning_rate": 2.1871328732679787e-06,
      "loss": 0.0173,
      "step": 12377
    },
    {
      "epoch": 7.9193857965451055,
      "grad_norm": 0.586485506329755,
      "learning_rate": 2.185839659874263e-06,
      "loss": 0.02,
      "step": 12378
    },
    {
      "epoch": 7.920025591810621,
      "grad_norm": 0.6430191327711833,
      "learning_rate": 2.1845467820066925e-06,
      "loss": 0.0223,
      "step": 12379
    },
    {
      "epoch": 7.920665387076136,
      "grad_norm": 0.7154854110906177,
      "learning_rate": 2.1832542397207823e-06,
      "loss": 0.0221,
      "step": 12380
    },
    {
      "epoch": 7.9213051823416505,
      "grad_norm": 0.613743788841971,
      "learning_rate": 2.1819620330720294e-06,
      "loss": 0.0197,
      "step": 12381
    },
    {
      "epoch": 7.921944977607166,
      "grad_norm": 0.6379305074392896,
      "learning_rate": 2.1806701621159166e-06,
      "loss": 0.0191,
      "step": 12382
    },
    {
      "epoch": 7.922584772872681,
      "grad_norm": 0.7511763682290541,
      "learning_rate": 2.179378626907921e-06,
      "loss": 0.0229,
      "step": 12383
    },
    {
      "epoch": 7.923224568138195,
      "grad_norm": 0.7131839298617527,
      "learning_rate": 2.1780874275034948e-06,
      "loss": 0.0188,
      "step": 12384
    },
    {
      "epoch": 7.923864363403711,
      "grad_norm": 0.6631527382220663,
      "learning_rate": 2.176796563958078e-06,
      "loss": 0.0214,
      "step": 12385
    },
    {
      "epoch": 7.924504158669226,
      "grad_norm": 0.6797465672459614,
      "learning_rate": 2.175506036327103e-06,
      "loss": 0.0222,
      "step": 12386
    },
    {
      "epoch": 7.925143953934741,
      "grad_norm": 0.7674043339803596,
      "learning_rate": 2.17421584466598e-06,
      "loss": 0.0231,
      "step": 12387
    },
    {
      "epoch": 7.925783749200256,
      "grad_norm": 0.6356779772282942,
      "learning_rate": 2.172925989030107e-06,
      "loss": 0.0191,
      "step": 12388
    },
    {
      "epoch": 7.926423544465771,
      "grad_norm": 0.6540008006760587,
      "learning_rate": 2.1716364694748693e-06,
      "loss": 0.0187,
      "step": 12389
    },
    {
      "epoch": 7.927063339731286,
      "grad_norm": 0.6702181067269746,
      "learning_rate": 2.170347286055636e-06,
      "loss": 0.0221,
      "step": 12390
    },
    {
      "epoch": 7.927703134996801,
      "grad_norm": 0.7875585879033782,
      "learning_rate": 2.1690584388277624e-06,
      "loss": 0.0254,
      "step": 12391
    },
    {
      "epoch": 7.928342930262316,
      "grad_norm": 0.5706257454723397,
      "learning_rate": 2.167769927846587e-06,
      "loss": 0.0176,
      "step": 12392
    },
    {
      "epoch": 7.928982725527831,
      "grad_norm": 0.7107224678929346,
      "learning_rate": 2.1664817531674407e-06,
      "loss": 0.0227,
      "step": 12393
    },
    {
      "epoch": 7.929622520793346,
      "grad_norm": 0.6273317676152669,
      "learning_rate": 2.1651939148456334e-06,
      "loss": 0.0196,
      "step": 12394
    },
    {
      "epoch": 7.930262316058861,
      "grad_norm": 0.6734686282871819,
      "learning_rate": 2.1639064129364594e-06,
      "loss": 0.026,
      "step": 12395
    },
    {
      "epoch": 7.930902111324376,
      "grad_norm": 0.6818604777155427,
      "learning_rate": 2.162619247495208e-06,
      "loss": 0.0219,
      "step": 12396
    },
    {
      "epoch": 7.931541906589891,
      "grad_norm": 0.6389760850889952,
      "learning_rate": 2.1613324185771433e-06,
      "loss": 0.0225,
      "step": 12397
    },
    {
      "epoch": 7.932181701855407,
      "grad_norm": 0.7856785421935737,
      "learning_rate": 2.1600459262375182e-06,
      "loss": 0.0259,
      "step": 12398
    },
    {
      "epoch": 7.932821497120921,
      "grad_norm": 0.6874403677254819,
      "learning_rate": 2.1587597705315788e-06,
      "loss": 0.0208,
      "step": 12399
    },
    {
      "epoch": 7.933461292386436,
      "grad_norm": 0.6678053108880525,
      "learning_rate": 2.1574739515145447e-06,
      "loss": 0.0204,
      "step": 12400
    },
    {
      "epoch": 7.9341010876519515,
      "grad_norm": 0.6616703885187877,
      "learning_rate": 2.156188469241629e-06,
      "loss": 0.0227,
      "step": 12401
    },
    {
      "epoch": 7.934740882917467,
      "grad_norm": 0.6275157681795496,
      "learning_rate": 2.154903323768025e-06,
      "loss": 0.0235,
      "step": 12402
    },
    {
      "epoch": 7.935380678182981,
      "grad_norm": 0.6813908447574721,
      "learning_rate": 2.1536185151489198e-06,
      "loss": 0.0207,
      "step": 12403
    },
    {
      "epoch": 7.9360204734484965,
      "grad_norm": 0.6104840227927657,
      "learning_rate": 2.152334043439478e-06,
      "loss": 0.0214,
      "step": 12404
    },
    {
      "epoch": 7.936660268714012,
      "grad_norm": 0.5803645204586769,
      "learning_rate": 2.151049908694852e-06,
      "loss": 0.017,
      "step": 12405
    },
    {
      "epoch": 7.937300063979526,
      "grad_norm": 0.6734720803728302,
      "learning_rate": 2.1497661109701808e-06,
      "loss": 0.0214,
      "step": 12406
    },
    {
      "epoch": 7.937939859245041,
      "grad_norm": 0.7087630226479702,
      "learning_rate": 2.148482650320588e-06,
      "loss": 0.0236,
      "step": 12407
    },
    {
      "epoch": 7.938579654510557,
      "grad_norm": 0.646866275510068,
      "learning_rate": 2.1471995268011815e-06,
      "loss": 0.0195,
      "step": 12408
    },
    {
      "epoch": 7.939219449776072,
      "grad_norm": 0.6816262800318675,
      "learning_rate": 2.14591674046706e-06,
      "loss": 0.0252,
      "step": 12409
    },
    {
      "epoch": 7.939859245041586,
      "grad_norm": 0.6558473312987486,
      "learning_rate": 2.1446342913733033e-06,
      "loss": 0.0228,
      "step": 12410
    },
    {
      "epoch": 7.940499040307102,
      "grad_norm": 0.6766889361892682,
      "learning_rate": 2.1433521795749757e-06,
      "loss": 0.0219,
      "step": 12411
    },
    {
      "epoch": 7.941138835572617,
      "grad_norm": 0.6225007871155342,
      "learning_rate": 2.142070405127127e-06,
      "loss": 0.0197,
      "step": 12412
    },
    {
      "epoch": 7.941778630838132,
      "grad_norm": 0.731393663470705,
      "learning_rate": 2.1407889680848004e-06,
      "loss": 0.02,
      "step": 12413
    },
    {
      "epoch": 7.942418426103647,
      "grad_norm": 0.6717740830134492,
      "learning_rate": 2.1395078685030145e-06,
      "loss": 0.0246,
      "step": 12414
    },
    {
      "epoch": 7.943058221369162,
      "grad_norm": 0.6861680131339973,
      "learning_rate": 2.1382271064367755e-06,
      "loss": 0.0191,
      "step": 12415
    },
    {
      "epoch": 7.943698016634677,
      "grad_norm": 0.6955148820662221,
      "learning_rate": 2.136946681941081e-06,
      "loss": 0.0238,
      "step": 12416
    },
    {
      "epoch": 7.944337811900192,
      "grad_norm": 0.664015005338006,
      "learning_rate": 2.13566659507091e-06,
      "loss": 0.023,
      "step": 12417
    },
    {
      "epoch": 7.944977607165707,
      "grad_norm": 0.7193205829748108,
      "learning_rate": 2.1343868458812254e-06,
      "loss": 0.021,
      "step": 12418
    },
    {
      "epoch": 7.945617402431222,
      "grad_norm": 0.7387982488955225,
      "learning_rate": 2.1331074344269788e-06,
      "loss": 0.0218,
      "step": 12419
    },
    {
      "epoch": 7.946257197696737,
      "grad_norm": 0.6594819766958904,
      "learning_rate": 2.131828360763104e-06,
      "loss": 0.0206,
      "step": 12420
    },
    {
      "epoch": 7.946896992962252,
      "grad_norm": 0.6111760294801304,
      "learning_rate": 2.1305496249445233e-06,
      "loss": 0.0199,
      "step": 12421
    },
    {
      "epoch": 7.947536788227767,
      "grad_norm": 0.6715809278401905,
      "learning_rate": 2.1292712270261394e-06,
      "loss": 0.0224,
      "step": 12422
    },
    {
      "epoch": 7.948176583493282,
      "grad_norm": 0.6362259967907468,
      "learning_rate": 2.1279931670628516e-06,
      "loss": 0.0205,
      "step": 12423
    },
    {
      "epoch": 7.9488163787587975,
      "grad_norm": 0.8148336387375729,
      "learning_rate": 2.126715445109534e-06,
      "loss": 0.0261,
      "step": 12424
    },
    {
      "epoch": 7.949456174024312,
      "grad_norm": 0.7248233309492745,
      "learning_rate": 2.1254380612210467e-06,
      "loss": 0.0219,
      "step": 12425
    },
    {
      "epoch": 7.950095969289827,
      "grad_norm": 0.7099997100938892,
      "learning_rate": 2.124161015452244e-06,
      "loss": 0.0231,
      "step": 12426
    },
    {
      "epoch": 7.9507357645553425,
      "grad_norm": 0.6120864357793909,
      "learning_rate": 2.1228843078579575e-06,
      "loss": 0.0207,
      "step": 12427
    },
    {
      "epoch": 7.951375559820857,
      "grad_norm": 0.6112061735054319,
      "learning_rate": 2.1216079384930066e-06,
      "loss": 0.0192,
      "step": 12428
    },
    {
      "epoch": 7.952015355086372,
      "grad_norm": 0.7248624520940672,
      "learning_rate": 2.120331907412193e-06,
      "loss": 0.0229,
      "step": 12429
    },
    {
      "epoch": 7.952655150351887,
      "grad_norm": 0.74173443337597,
      "learning_rate": 2.1190562146703143e-06,
      "loss": 0.0256,
      "step": 12430
    },
    {
      "epoch": 7.953294945617403,
      "grad_norm": 0.628837792461432,
      "learning_rate": 2.1177808603221418e-06,
      "loss": 0.0228,
      "step": 12431
    },
    {
      "epoch": 7.953934740882918,
      "grad_norm": 0.755387644562201,
      "learning_rate": 2.1165058444224352e-06,
      "loss": 0.0279,
      "step": 12432
    },
    {
      "epoch": 7.954574536148432,
      "grad_norm": 0.7401825938009732,
      "learning_rate": 2.115231167025946e-06,
      "loss": 0.0224,
      "step": 12433
    },
    {
      "epoch": 7.955214331413948,
      "grad_norm": 0.6157180604433621,
      "learning_rate": 2.1139568281874057e-06,
      "loss": 0.0209,
      "step": 12434
    },
    {
      "epoch": 7.955854126679463,
      "grad_norm": 0.6834028943930064,
      "learning_rate": 2.112682827961531e-06,
      "loss": 0.0229,
      "step": 12435
    },
    {
      "epoch": 7.956493921944977,
      "grad_norm": 0.704178793011521,
      "learning_rate": 2.1114091664030244e-06,
      "loss": 0.0224,
      "step": 12436
    },
    {
      "epoch": 7.957133717210493,
      "grad_norm": 0.6077299576803966,
      "learning_rate": 2.1101358435665753e-06,
      "loss": 0.0182,
      "step": 12437
    },
    {
      "epoch": 7.957773512476008,
      "grad_norm": 0.6642987255310623,
      "learning_rate": 2.1088628595068593e-06,
      "loss": 0.0219,
      "step": 12438
    },
    {
      "epoch": 7.958413307741523,
      "grad_norm": 0.7056650515667431,
      "learning_rate": 2.107590214278531e-06,
      "loss": 0.0213,
      "step": 12439
    },
    {
      "epoch": 7.9590531030070375,
      "grad_norm": 0.6277573246300686,
      "learning_rate": 2.1063179079362428e-06,
      "loss": 0.02,
      "step": 12440
    },
    {
      "epoch": 7.959692898272553,
      "grad_norm": 0.6648633180409873,
      "learning_rate": 2.1050459405346213e-06,
      "loss": 0.0194,
      "step": 12441
    },
    {
      "epoch": 7.960332693538068,
      "grad_norm": 0.643232741573026,
      "learning_rate": 2.1037743121282815e-06,
      "loss": 0.0233,
      "step": 12442
    },
    {
      "epoch": 7.960972488803582,
      "grad_norm": 0.6922675546002637,
      "learning_rate": 2.102503022771828e-06,
      "loss": 0.0216,
      "step": 12443
    },
    {
      "epoch": 7.961612284069098,
      "grad_norm": 0.7040947334906847,
      "learning_rate": 2.1012320725198463e-06,
      "loss": 0.0232,
      "step": 12444
    },
    {
      "epoch": 7.962252079334613,
      "grad_norm": 0.7246076160725042,
      "learning_rate": 2.099961461426906e-06,
      "loss": 0.0212,
      "step": 12445
    },
    {
      "epoch": 7.962891874600128,
      "grad_norm": 0.6907434409977001,
      "learning_rate": 2.0986911895475704e-06,
      "loss": 0.0204,
      "step": 12446
    },
    {
      "epoch": 7.963531669865643,
      "grad_norm": 0.7721745426669681,
      "learning_rate": 2.0974212569363793e-06,
      "loss": 0.0226,
      "step": 12447
    },
    {
      "epoch": 7.964171465131158,
      "grad_norm": 0.6549637519986556,
      "learning_rate": 2.096151663647863e-06,
      "loss": 0.0201,
      "step": 12448
    },
    {
      "epoch": 7.964811260396673,
      "grad_norm": 0.713151336231653,
      "learning_rate": 2.0948824097365338e-06,
      "loss": 0.0211,
      "step": 12449
    },
    {
      "epoch": 7.9654510556621885,
      "grad_norm": 0.5554054658684888,
      "learning_rate": 2.093613495256891e-06,
      "loss": 0.0195,
      "step": 12450
    },
    {
      "epoch": 7.966090850927703,
      "grad_norm": 0.6387611023457815,
      "learning_rate": 2.0923449202634215e-06,
      "loss": 0.0178,
      "step": 12451
    },
    {
      "epoch": 7.966730646193218,
      "grad_norm": 0.7456674066047301,
      "learning_rate": 2.0910766848105914e-06,
      "loss": 0.0229,
      "step": 12452
    },
    {
      "epoch": 7.967370441458733,
      "grad_norm": 0.7136019116108818,
      "learning_rate": 2.0898087889528617e-06,
      "loss": 0.0235,
      "step": 12453
    },
    {
      "epoch": 7.968010236724249,
      "grad_norm": 0.6040535835987569,
      "learning_rate": 2.088541232744672e-06,
      "loss": 0.018,
      "step": 12454
    },
    {
      "epoch": 7.968650031989763,
      "grad_norm": 0.6122728066694112,
      "learning_rate": 2.087274016240445e-06,
      "loss": 0.0182,
      "step": 12455
    },
    {
      "epoch": 7.969289827255278,
      "grad_norm": 0.6453020513712752,
      "learning_rate": 2.0860071394945992e-06,
      "loss": 0.0215,
      "step": 12456
    },
    {
      "epoch": 7.969929622520794,
      "grad_norm": 0.7054124961591868,
      "learning_rate": 2.0847406025615293e-06,
      "loss": 0.0204,
      "step": 12457
    },
    {
      "epoch": 7.970569417786308,
      "grad_norm": 0.6255282124912044,
      "learning_rate": 2.083474405495618e-06,
      "loss": 0.0199,
      "step": 12458
    },
    {
      "epoch": 7.971209213051823,
      "grad_norm": 0.6773287067035906,
      "learning_rate": 2.08220854835123e-06,
      "loss": 0.0217,
      "step": 12459
    },
    {
      "epoch": 7.971849008317339,
      "grad_norm": 0.734169851921537,
      "learning_rate": 2.080943031182725e-06,
      "loss": 0.0254,
      "step": 12460
    },
    {
      "epoch": 7.972488803582854,
      "grad_norm": 0.7676063985616719,
      "learning_rate": 2.0796778540444397e-06,
      "loss": 0.0187,
      "step": 12461
    },
    {
      "epoch": 7.973128598848368,
      "grad_norm": 0.5826186689225937,
      "learning_rate": 2.0784130169906945e-06,
      "loss": 0.0186,
      "step": 12462
    },
    {
      "epoch": 7.9737683941138835,
      "grad_norm": 0.7066833345444112,
      "learning_rate": 2.0771485200758067e-06,
      "loss": 0.022,
      "step": 12463
    },
    {
      "epoch": 7.974408189379399,
      "grad_norm": 0.48937810125858106,
      "learning_rate": 2.0758843633540683e-06,
      "loss": 0.016,
      "step": 12464
    },
    {
      "epoch": 7.975047984644913,
      "grad_norm": 0.6522985637864399,
      "learning_rate": 2.0746205468797574e-06,
      "loss": 0.0222,
      "step": 12465
    },
    {
      "epoch": 7.975687779910428,
      "grad_norm": 0.6848186579751411,
      "learning_rate": 2.0733570707071436e-06,
      "loss": 0.0205,
      "step": 12466
    },
    {
      "epoch": 7.976327575175944,
      "grad_norm": 0.6448318002816372,
      "learning_rate": 2.0720939348904767e-06,
      "loss": 0.0212,
      "step": 12467
    },
    {
      "epoch": 7.976967370441459,
      "grad_norm": 0.6610504874033447,
      "learning_rate": 2.070831139483993e-06,
      "loss": 0.0254,
      "step": 12468
    },
    {
      "epoch": 7.977607165706974,
      "grad_norm": 0.6057837062970923,
      "learning_rate": 2.069568684541914e-06,
      "loss": 0.0213,
      "step": 12469
    },
    {
      "epoch": 7.978246960972489,
      "grad_norm": 0.7437823659299287,
      "learning_rate": 2.068306570118449e-06,
      "loss": 0.0277,
      "step": 12470
    },
    {
      "epoch": 7.978886756238004,
      "grad_norm": 0.7869706678131986,
      "learning_rate": 2.067044796267792e-06,
      "loss": 0.0224,
      "step": 12471
    },
    {
      "epoch": 7.979526551503519,
      "grad_norm": 0.5784161935464814,
      "learning_rate": 2.065783363044117e-06,
      "loss": 0.0183,
      "step": 12472
    },
    {
      "epoch": 7.980166346769034,
      "grad_norm": 0.7474044173142158,
      "learning_rate": 2.0645222705015934e-06,
      "loss": 0.0226,
      "step": 12473
    },
    {
      "epoch": 7.980806142034549,
      "grad_norm": 0.627969033016195,
      "learning_rate": 2.063261518694366e-06,
      "loss": 0.0177,
      "step": 12474
    },
    {
      "epoch": 7.981445937300064,
      "grad_norm": 0.6632331449938017,
      "learning_rate": 2.062001107676569e-06,
      "loss": 0.022,
      "step": 12475
    },
    {
      "epoch": 7.982085732565579,
      "grad_norm": 0.6223877056538604,
      "learning_rate": 2.0607410375023275e-06,
      "loss": 0.0186,
      "step": 12476
    },
    {
      "epoch": 7.982725527831094,
      "grad_norm": 0.6708956921515044,
      "learning_rate": 2.0594813082257414e-06,
      "loss": 0.0195,
      "step": 12477
    },
    {
      "epoch": 7.983365323096609,
      "grad_norm": 0.7658939100473802,
      "learning_rate": 2.0582219199009034e-06,
      "loss": 0.0196,
      "step": 12478
    },
    {
      "epoch": 7.984005118362124,
      "grad_norm": 0.810069047764647,
      "learning_rate": 2.056962872581889e-06,
      "loss": 0.0243,
      "step": 12479
    },
    {
      "epoch": 7.984644913627639,
      "grad_norm": 0.6627246173473889,
      "learning_rate": 2.0557041663227582e-06,
      "loss": 0.0194,
      "step": 12480
    },
    {
      "epoch": 7.985284708893154,
      "grad_norm": 0.7276182430615057,
      "learning_rate": 2.0544458011775603e-06,
      "loss": 0.0231,
      "step": 12481
    },
    {
      "epoch": 7.985924504158669,
      "grad_norm": 0.6880692990128103,
      "learning_rate": 2.0531877772003217e-06,
      "loss": 0.0191,
      "step": 12482
    },
    {
      "epoch": 7.9865642994241846,
      "grad_norm": 0.7393500763370989,
      "learning_rate": 2.051930094445067e-06,
      "loss": 0.0228,
      "step": 12483
    },
    {
      "epoch": 7.987204094689699,
      "grad_norm": 0.6566212276474015,
      "learning_rate": 2.0506727529657954e-06,
      "loss": 0.0218,
      "step": 12484
    },
    {
      "epoch": 7.987843889955214,
      "grad_norm": 0.6222574811307934,
      "learning_rate": 2.0494157528164925e-06,
      "loss": 0.0181,
      "step": 12485
    },
    {
      "epoch": 7.9884836852207295,
      "grad_norm": 0.6998071443362288,
      "learning_rate": 2.0481590940511364e-06,
      "loss": 0.0209,
      "step": 12486
    },
    {
      "epoch": 7.989123480486245,
      "grad_norm": 0.6711943679721019,
      "learning_rate": 2.0469027767236848e-06,
      "loss": 0.0226,
      "step": 12487
    },
    {
      "epoch": 7.989763275751759,
      "grad_norm": 0.683748352060417,
      "learning_rate": 2.045646800888079e-06,
      "loss": 0.0231,
      "step": 12488
    },
    {
      "epoch": 7.990403071017274,
      "grad_norm": 0.7366141612560977,
      "learning_rate": 2.0443911665982473e-06,
      "loss": 0.0223,
      "step": 12489
    },
    {
      "epoch": 7.99104286628279,
      "grad_norm": 0.677512578797958,
      "learning_rate": 2.0431358739081108e-06,
      "loss": 0.0212,
      "step": 12490
    },
    {
      "epoch": 7.991682661548305,
      "grad_norm": 0.6716991672971032,
      "learning_rate": 2.041880922871565e-06,
      "loss": 0.0205,
      "step": 12491
    },
    {
      "epoch": 7.992322456813819,
      "grad_norm": 0.7040834478627295,
      "learning_rate": 2.040626313542493e-06,
      "loss": 0.0239,
      "step": 12492
    },
    {
      "epoch": 7.992962252079335,
      "grad_norm": 0.760063858091788,
      "learning_rate": 2.0393720459747723e-06,
      "loss": 0.024,
      "step": 12493
    },
    {
      "epoch": 7.99360204734485,
      "grad_norm": 0.5667415844264875,
      "learning_rate": 2.0381181202222543e-06,
      "loss": 0.0187,
      "step": 12494
    },
    {
      "epoch": 7.994241842610364,
      "grad_norm": 0.6931735274468768,
      "learning_rate": 2.0368645363387806e-06,
      "loss": 0.0225,
      "step": 12495
    },
    {
      "epoch": 7.99488163787588,
      "grad_norm": 0.5846947684712608,
      "learning_rate": 2.035611294378179e-06,
      "loss": 0.0183,
      "step": 12496
    },
    {
      "epoch": 7.995521433141395,
      "grad_norm": 0.6957520479314939,
      "learning_rate": 2.0343583943942612e-06,
      "loss": 0.0227,
      "step": 12497
    },
    {
      "epoch": 7.99616122840691,
      "grad_norm": 0.6961883211602308,
      "learning_rate": 2.033105836440823e-06,
      "loss": 0.0203,
      "step": 12498
    },
    {
      "epoch": 7.9968010236724245,
      "grad_norm": 0.7163950786985035,
      "learning_rate": 2.0318536205716467e-06,
      "loss": 0.0234,
      "step": 12499
    },
    {
      "epoch": 7.99744081893794,
      "grad_norm": 0.6291474181486971,
      "learning_rate": 2.030601746840504e-06,
      "loss": 0.0182,
      "step": 12500
    },
    {
      "epoch": 7.998080614203455,
      "grad_norm": 0.6421199341430469,
      "learning_rate": 2.0293502153011456e-06,
      "loss": 0.0231,
      "step": 12501
    },
    {
      "epoch": 7.9987204094689695,
      "grad_norm": 0.6947761090372107,
      "learning_rate": 2.0280990260073075e-06,
      "loss": 0.0208,
      "step": 12502
    },
    {
      "epoch": 7.999360204734485,
      "grad_norm": 0.6550737817704017,
      "learning_rate": 2.0268481790127192e-06,
      "loss": 0.0219,
      "step": 12503
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.752645929107226,
      "learning_rate": 2.025597674371087e-06,
      "loss": 0.0204,
      "step": 12504
    },
    {
      "epoch": 8.000639795265515,
      "grad_norm": 0.32809351315208996,
      "learning_rate": 2.0243475121361057e-06,
      "loss": 0.0066,
      "step": 12505
    },
    {
      "epoch": 8.00127959053103,
      "grad_norm": 0.29611510060946244,
      "learning_rate": 2.023097692361451e-06,
      "loss": 0.0077,
      "step": 12506
    },
    {
      "epoch": 8.001919385796546,
      "grad_norm": 0.30099113197070654,
      "learning_rate": 2.0218482151007956e-06,
      "loss": 0.008,
      "step": 12507
    },
    {
      "epoch": 8.00255918106206,
      "grad_norm": 0.3315593136268088,
      "learning_rate": 2.0205990804077848e-06,
      "loss": 0.0068,
      "step": 12508
    },
    {
      "epoch": 8.003198976327575,
      "grad_norm": 0.33919996880315656,
      "learning_rate": 2.019350288336055e-06,
      "loss": 0.0068,
      "step": 12509
    },
    {
      "epoch": 8.00383877159309,
      "grad_norm": 0.39726785067957554,
      "learning_rate": 2.018101838939228e-06,
      "loss": 0.0099,
      "step": 12510
    },
    {
      "epoch": 8.004478566858605,
      "grad_norm": 0.4221258771533326,
      "learning_rate": 2.016853732270909e-06,
      "loss": 0.0094,
      "step": 12511
    },
    {
      "epoch": 8.00511836212412,
      "grad_norm": 0.3344772749827228,
      "learning_rate": 2.0156059683846875e-06,
      "loss": 0.0083,
      "step": 12512
    },
    {
      "epoch": 8.005758157389636,
      "grad_norm": 0.38251422579307814,
      "learning_rate": 2.0143585473341454e-06,
      "loss": 0.0093,
      "step": 12513
    },
    {
      "epoch": 8.006397952655151,
      "grad_norm": 0.3766830707362484,
      "learning_rate": 2.0131114691728416e-06,
      "loss": 0.007,
      "step": 12514
    },
    {
      "epoch": 8.007037747920666,
      "grad_norm": 0.4386496207305214,
      "learning_rate": 2.011864733954324e-06,
      "loss": 0.0101,
      "step": 12515
    },
    {
      "epoch": 8.00767754318618,
      "grad_norm": 0.3872341262376731,
      "learning_rate": 2.0106183417321213e-06,
      "loss": 0.0085,
      "step": 12516
    },
    {
      "epoch": 8.008317338451695,
      "grad_norm": 0.2973195751016853,
      "learning_rate": 2.009372292559758e-06,
      "loss": 0.0055,
      "step": 12517
    },
    {
      "epoch": 8.00895713371721,
      "grad_norm": 0.43289577252605865,
      "learning_rate": 2.0081265864907353e-06,
      "loss": 0.008,
      "step": 12518
    },
    {
      "epoch": 8.009596928982726,
      "grad_norm": 0.40802600732409716,
      "learning_rate": 2.0068812235785375e-06,
      "loss": 0.0077,
      "step": 12519
    },
    {
      "epoch": 8.01023672424824,
      "grad_norm": 0.3691498284523459,
      "learning_rate": 2.005636203876643e-06,
      "loss": 0.0078,
      "step": 12520
    },
    {
      "epoch": 8.010876519513756,
      "grad_norm": 0.47492758704031846,
      "learning_rate": 2.0043915274385095e-06,
      "loss": 0.0081,
      "step": 12521
    },
    {
      "epoch": 8.011516314779271,
      "grad_norm": 0.464659331504764,
      "learning_rate": 2.0031471943175783e-06,
      "loss": 0.0084,
      "step": 12522
    },
    {
      "epoch": 8.012156110044785,
      "grad_norm": 0.43632529491369026,
      "learning_rate": 2.001903204567284e-06,
      "loss": 0.009,
      "step": 12523
    },
    {
      "epoch": 8.0127959053103,
      "grad_norm": 0.391053820950473,
      "learning_rate": 2.0006595582410383e-06,
      "loss": 0.0062,
      "step": 12524
    },
    {
      "epoch": 8.013435700575815,
      "grad_norm": 0.4012987069871319,
      "learning_rate": 1.9994162553922415e-06,
      "loss": 0.0078,
      "step": 12525
    },
    {
      "epoch": 8.01407549584133,
      "grad_norm": 0.41294349773467465,
      "learning_rate": 1.9981732960742796e-06,
      "loss": 0.0077,
      "step": 12526
    },
    {
      "epoch": 8.014715291106846,
      "grad_norm": 0.3861572581497946,
      "learning_rate": 1.996930680340521e-06,
      "loss": 0.0067,
      "step": 12527
    },
    {
      "epoch": 8.015355086372361,
      "grad_norm": 0.43167090057968827,
      "learning_rate": 1.995688408244323e-06,
      "loss": 0.0068,
      "step": 12528
    },
    {
      "epoch": 8.015994881637877,
      "grad_norm": 0.4078414076034888,
      "learning_rate": 1.994446479839024e-06,
      "loss": 0.0063,
      "step": 12529
    },
    {
      "epoch": 8.01663467690339,
      "grad_norm": 0.3730676171910358,
      "learning_rate": 1.9932048951779547e-06,
      "loss": 0.006,
      "step": 12530
    },
    {
      "epoch": 8.017274472168905,
      "grad_norm": 0.45128503126876895,
      "learning_rate": 1.991963654314424e-06,
      "loss": 0.0073,
      "step": 12531
    },
    {
      "epoch": 8.01791426743442,
      "grad_norm": 0.400897723588924,
      "learning_rate": 1.9907227573017264e-06,
      "loss": 0.0068,
      "step": 12532
    },
    {
      "epoch": 8.018554062699936,
      "grad_norm": 0.39783285142416347,
      "learning_rate": 1.9894822041931485e-06,
      "loss": 0.0072,
      "step": 12533
    },
    {
      "epoch": 8.019193857965451,
      "grad_norm": 0.3823746971408917,
      "learning_rate": 1.988241995041955e-06,
      "loss": 0.007,
      "step": 12534
    },
    {
      "epoch": 8.019833653230966,
      "grad_norm": 0.43061308664835757,
      "learning_rate": 1.9870021299013995e-06,
      "loss": 0.0076,
      "step": 12535
    },
    {
      "epoch": 8.020473448496482,
      "grad_norm": 0.3146213915988917,
      "learning_rate": 1.985762608824715e-06,
      "loss": 0.0055,
      "step": 12536
    },
    {
      "epoch": 8.021113243761997,
      "grad_norm": 0.3739280509424224,
      "learning_rate": 1.984523431865131e-06,
      "loss": 0.0061,
      "step": 12537
    },
    {
      "epoch": 8.02175303902751,
      "grad_norm": 0.4359871158669589,
      "learning_rate": 1.983284599075852e-06,
      "loss": 0.0072,
      "step": 12538
    },
    {
      "epoch": 8.022392834293026,
      "grad_norm": 0.40433467562837727,
      "learning_rate": 1.982046110510072e-06,
      "loss": 0.0066,
      "step": 12539
    },
    {
      "epoch": 8.023032629558541,
      "grad_norm": 0.3673386103040803,
      "learning_rate": 1.980807966220969e-06,
      "loss": 0.0073,
      "step": 12540
    },
    {
      "epoch": 8.023672424824056,
      "grad_norm": 0.2992841710191939,
      "learning_rate": 1.9795701662617074e-06,
      "loss": 0.0057,
      "step": 12541
    },
    {
      "epoch": 8.024312220089572,
      "grad_norm": 0.35533527925719177,
      "learning_rate": 1.9783327106854322e-06,
      "loss": 0.0065,
      "step": 12542
    },
    {
      "epoch": 8.024952015355087,
      "grad_norm": 0.37122136889026625,
      "learning_rate": 1.9770955995452833e-06,
      "loss": 0.0064,
      "step": 12543
    },
    {
      "epoch": 8.025591810620602,
      "grad_norm": 0.41607884928216443,
      "learning_rate": 1.9758588328943784e-06,
      "loss": 0.008,
      "step": 12544
    },
    {
      "epoch": 8.026231605886116,
      "grad_norm": 0.493843903100876,
      "learning_rate": 1.9746224107858216e-06,
      "loss": 0.0096,
      "step": 12545
    },
    {
      "epoch": 8.026871401151631,
      "grad_norm": 0.320616301786242,
      "learning_rate": 1.9733863332726988e-06,
      "loss": 0.0062,
      "step": 12546
    },
    {
      "epoch": 8.027511196417146,
      "grad_norm": 0.4137350365329338,
      "learning_rate": 1.9721506004080927e-06,
      "loss": 0.0079,
      "step": 12547
    },
    {
      "epoch": 8.028150991682661,
      "grad_norm": 0.38449169348073053,
      "learning_rate": 1.9709152122450582e-06,
      "loss": 0.0064,
      "step": 12548
    },
    {
      "epoch": 8.028790786948177,
      "grad_norm": 0.4936336006125126,
      "learning_rate": 1.9696801688366394e-06,
      "loss": 0.0089,
      "step": 12549
    },
    {
      "epoch": 8.029430582213692,
      "grad_norm": 0.3831381812246577,
      "learning_rate": 1.9684454702358725e-06,
      "loss": 0.0066,
      "step": 12550
    },
    {
      "epoch": 8.030070377479207,
      "grad_norm": 0.33436966438021676,
      "learning_rate": 1.9672111164957697e-06,
      "loss": 0.0058,
      "step": 12551
    },
    {
      "epoch": 8.030710172744723,
      "grad_norm": 0.3675067651860286,
      "learning_rate": 1.96597710766933e-06,
      "loss": 0.0064,
      "step": 12552
    },
    {
      "epoch": 8.031349968010236,
      "grad_norm": 0.35725344789787417,
      "learning_rate": 1.9647434438095434e-06,
      "loss": 0.0062,
      "step": 12553
    },
    {
      "epoch": 8.031989763275751,
      "grad_norm": 0.38364090949431073,
      "learning_rate": 1.963510124969381e-06,
      "loss": 0.0064,
      "step": 12554
    },
    {
      "epoch": 8.032629558541267,
      "grad_norm": 0.30424987741146453,
      "learning_rate": 1.962277151201798e-06,
      "loss": 0.0047,
      "step": 12555
    },
    {
      "epoch": 8.033269353806782,
      "grad_norm": 0.36903695881998017,
      "learning_rate": 1.9610445225597353e-06,
      "loss": 0.0059,
      "step": 12556
    },
    {
      "epoch": 8.033909149072297,
      "grad_norm": 0.3973007690688407,
      "learning_rate": 1.9598122390961216e-06,
      "loss": 0.0067,
      "step": 12557
    },
    {
      "epoch": 8.034548944337812,
      "grad_norm": 0.4169132957927371,
      "learning_rate": 1.958580300863867e-06,
      "loss": 0.0071,
      "step": 12558
    },
    {
      "epoch": 8.035188739603328,
      "grad_norm": 0.4072487075249227,
      "learning_rate": 1.957348707915868e-06,
      "loss": 0.008,
      "step": 12559
    },
    {
      "epoch": 8.035828534868841,
      "grad_norm": 0.3504104481459631,
      "learning_rate": 1.9561174603050102e-06,
      "loss": 0.0057,
      "step": 12560
    },
    {
      "epoch": 8.036468330134356,
      "grad_norm": 0.3999467829487302,
      "learning_rate": 1.95488655808416e-06,
      "loss": 0.0057,
      "step": 12561
    },
    {
      "epoch": 8.037108125399872,
      "grad_norm": 0.45189683181765045,
      "learning_rate": 1.9536560013061677e-06,
      "loss": 0.0076,
      "step": 12562
    },
    {
      "epoch": 8.037747920665387,
      "grad_norm": 0.4653773631736632,
      "learning_rate": 1.9524257900238765e-06,
      "loss": 0.0066,
      "step": 12563
    },
    {
      "epoch": 8.038387715930902,
      "grad_norm": 0.40240437384865996,
      "learning_rate": 1.951195924290106e-06,
      "loss": 0.0048,
      "step": 12564
    },
    {
      "epoch": 8.039027511196418,
      "grad_norm": 0.5041925510239312,
      "learning_rate": 1.949966404157664e-06,
      "loss": 0.0068,
      "step": 12565
    },
    {
      "epoch": 8.039667306461933,
      "grad_norm": 0.49370381647104966,
      "learning_rate": 1.948737229679343e-06,
      "loss": 0.0081,
      "step": 12566
    },
    {
      "epoch": 8.040307101727446,
      "grad_norm": 0.3462423852696134,
      "learning_rate": 1.947508400907925e-06,
      "loss": 0.0056,
      "step": 12567
    },
    {
      "epoch": 8.040946896992962,
      "grad_norm": 0.42341502860868074,
      "learning_rate": 1.946279917896172e-06,
      "loss": 0.0079,
      "step": 12568
    },
    {
      "epoch": 8.041586692258477,
      "grad_norm": 0.4739353190949195,
      "learning_rate": 1.945051780696834e-06,
      "loss": 0.0067,
      "step": 12569
    },
    {
      "epoch": 8.042226487523992,
      "grad_norm": 0.5342240290242978,
      "learning_rate": 1.943823989362643e-06,
      "loss": 0.0097,
      "step": 12570
    },
    {
      "epoch": 8.042866282789507,
      "grad_norm": 0.4959886777612178,
      "learning_rate": 1.942596543946319e-06,
      "loss": 0.0074,
      "step": 12571
    },
    {
      "epoch": 8.043506078055023,
      "grad_norm": 0.37294585216190135,
      "learning_rate": 1.941369444500567e-06,
      "loss": 0.0072,
      "step": 12572
    },
    {
      "epoch": 8.044145873320538,
      "grad_norm": 0.3383137282823556,
      "learning_rate": 1.940142691078073e-06,
      "loss": 0.0048,
      "step": 12573
    },
    {
      "epoch": 8.044785668586053,
      "grad_norm": 0.4508861045132767,
      "learning_rate": 1.9389162837315168e-06,
      "loss": 0.0073,
      "step": 12574
    },
    {
      "epoch": 8.045425463851567,
      "grad_norm": 0.39791173506392596,
      "learning_rate": 1.9376902225135565e-06,
      "loss": 0.007,
      "step": 12575
    },
    {
      "epoch": 8.046065259117082,
      "grad_norm": 0.5111419427076979,
      "learning_rate": 1.936464507476833e-06,
      "loss": 0.0071,
      "step": 12576
    },
    {
      "epoch": 8.046705054382597,
      "grad_norm": 0.32976775939440295,
      "learning_rate": 1.935239138673982e-06,
      "loss": 0.005,
      "step": 12577
    },
    {
      "epoch": 8.047344849648113,
      "grad_norm": 0.3348842108628288,
      "learning_rate": 1.9340141161576176e-06,
      "loss": 0.0054,
      "step": 12578
    },
    {
      "epoch": 8.047984644913628,
      "grad_norm": 0.6201030029240955,
      "learning_rate": 1.932789439980335e-06,
      "loss": 0.0083,
      "step": 12579
    },
    {
      "epoch": 8.048624440179143,
      "grad_norm": 0.38223157985549044,
      "learning_rate": 1.9315651101947262e-06,
      "loss": 0.0069,
      "step": 12580
    },
    {
      "epoch": 8.049264235444658,
      "grad_norm": 0.3142263802643546,
      "learning_rate": 1.930341126853359e-06,
      "loss": 0.005,
      "step": 12581
    },
    {
      "epoch": 8.049904030710172,
      "grad_norm": 0.4381344544364267,
      "learning_rate": 1.9291174900087892e-06,
      "loss": 0.0075,
      "step": 12582
    },
    {
      "epoch": 8.050543825975687,
      "grad_norm": 0.40811561830337467,
      "learning_rate": 1.9278941997135546e-06,
      "loss": 0.0076,
      "step": 12583
    },
    {
      "epoch": 8.051183621241202,
      "grad_norm": 0.3597242818901956,
      "learning_rate": 1.9266712560201874e-06,
      "loss": 0.0056,
      "step": 12584
    },
    {
      "epoch": 8.051823416506718,
      "grad_norm": 0.3739830564405683,
      "learning_rate": 1.925448658981194e-06,
      "loss": 0.0057,
      "step": 12585
    },
    {
      "epoch": 8.052463211772233,
      "grad_norm": 0.33625528274908456,
      "learning_rate": 1.9242264086490714e-06,
      "loss": 0.0051,
      "step": 12586
    },
    {
      "epoch": 8.053103007037748,
      "grad_norm": 0.46056814634598103,
      "learning_rate": 1.9230045050763026e-06,
      "loss": 0.009,
      "step": 12587
    },
    {
      "epoch": 8.053742802303264,
      "grad_norm": 0.4426479322723882,
      "learning_rate": 1.9217829483153516e-06,
      "loss": 0.0068,
      "step": 12588
    },
    {
      "epoch": 8.054382597568779,
      "grad_norm": 0.48811848308140454,
      "learning_rate": 1.9205617384186683e-06,
      "loss": 0.0067,
      "step": 12589
    },
    {
      "epoch": 8.055022392834292,
      "grad_norm": 0.43953220311999264,
      "learning_rate": 1.9193408754386933e-06,
      "loss": 0.0058,
      "step": 12590
    },
    {
      "epoch": 8.055662188099808,
      "grad_norm": 0.3982110805169687,
      "learning_rate": 1.9181203594278475e-06,
      "loss": 0.0053,
      "step": 12591
    },
    {
      "epoch": 8.056301983365323,
      "grad_norm": 0.38357099267585926,
      "learning_rate": 1.9169001904385376e-06,
      "loss": 0.0065,
      "step": 12592
    },
    {
      "epoch": 8.056941778630838,
      "grad_norm": 0.3902528136769906,
      "learning_rate": 1.9156803685231518e-06,
      "loss": 0.0055,
      "step": 12593
    },
    {
      "epoch": 8.057581573896353,
      "grad_norm": 0.47006100732034006,
      "learning_rate": 1.9144608937340726e-06,
      "loss": 0.0056,
      "step": 12594
    },
    {
      "epoch": 8.058221369161869,
      "grad_norm": 0.3649854621090495,
      "learning_rate": 1.913241766123659e-06,
      "loss": 0.0056,
      "step": 12595
    },
    {
      "epoch": 8.058861164427384,
      "grad_norm": 0.40882998331763926,
      "learning_rate": 1.9120229857442573e-06,
      "loss": 0.0055,
      "step": 12596
    },
    {
      "epoch": 8.059500959692897,
      "grad_norm": 0.36525776228212214,
      "learning_rate": 1.910804552648202e-06,
      "loss": 0.0063,
      "step": 12597
    },
    {
      "epoch": 8.060140754958413,
      "grad_norm": 0.33994099025616314,
      "learning_rate": 1.909586466887812e-06,
      "loss": 0.0061,
      "step": 12598
    },
    {
      "epoch": 8.060780550223928,
      "grad_norm": 0.5628923824970086,
      "learning_rate": 1.908368728515385e-06,
      "loss": 0.0084,
      "step": 12599
    },
    {
      "epoch": 8.061420345489443,
      "grad_norm": 0.449637642275486,
      "learning_rate": 1.907151337583213e-06,
      "loss": 0.007,
      "step": 12600
    },
    {
      "epoch": 8.062060140754959,
      "grad_norm": 0.4111440401888317,
      "learning_rate": 1.9059342941435654e-06,
      "loss": 0.0054,
      "step": 12601
    },
    {
      "epoch": 8.062699936020474,
      "grad_norm": 0.4788138842873767,
      "learning_rate": 1.904717598248701e-06,
      "loss": 0.0064,
      "step": 12602
    },
    {
      "epoch": 8.06333973128599,
      "grad_norm": 0.4136491088979151,
      "learning_rate": 1.9035012499508598e-06,
      "loss": 0.0069,
      "step": 12603
    },
    {
      "epoch": 8.063979526551503,
      "grad_norm": 0.4508132643663942,
      "learning_rate": 1.902285249302276e-06,
      "loss": 0.0056,
      "step": 12604
    },
    {
      "epoch": 8.064619321817018,
      "grad_norm": 0.4251609088046973,
      "learning_rate": 1.9010695963551572e-06,
      "loss": 0.0067,
      "step": 12605
    },
    {
      "epoch": 8.065259117082533,
      "grad_norm": 0.3908566665071184,
      "learning_rate": 1.8998542911617023e-06,
      "loss": 0.0059,
      "step": 12606
    },
    {
      "epoch": 8.065898912348048,
      "grad_norm": 0.4279573939640756,
      "learning_rate": 1.898639333774096e-06,
      "loss": 0.0073,
      "step": 12607
    },
    {
      "epoch": 8.066538707613564,
      "grad_norm": 0.3832782571442945,
      "learning_rate": 1.8974247242445066e-06,
      "loss": 0.0065,
      "step": 12608
    },
    {
      "epoch": 8.067178502879079,
      "grad_norm": 0.6218355934154335,
      "learning_rate": 1.8962104626250832e-06,
      "loss": 0.0089,
      "step": 12609
    },
    {
      "epoch": 8.067818298144594,
      "grad_norm": 0.4916320854878031,
      "learning_rate": 1.89499654896797e-06,
      "loss": 0.0084,
      "step": 12610
    },
    {
      "epoch": 8.06845809341011,
      "grad_norm": 0.415191351181675,
      "learning_rate": 1.8937829833252875e-06,
      "loss": 0.0071,
      "step": 12611
    },
    {
      "epoch": 8.069097888675623,
      "grad_norm": 0.44612948192836416,
      "learning_rate": 1.8925697657491426e-06,
      "loss": 0.0076,
      "step": 12612
    },
    {
      "epoch": 8.069737683941138,
      "grad_norm": 0.53408281681231,
      "learning_rate": 1.8913568962916285e-06,
      "loss": 0.0092,
      "step": 12613
    },
    {
      "epoch": 8.070377479206654,
      "grad_norm": 0.3629212890609767,
      "learning_rate": 1.8901443750048277e-06,
      "loss": 0.006,
      "step": 12614
    },
    {
      "epoch": 8.071017274472169,
      "grad_norm": 0.40353010564451786,
      "learning_rate": 1.8889322019408008e-06,
      "loss": 0.0061,
      "step": 12615
    },
    {
      "epoch": 8.071657069737684,
      "grad_norm": 0.37654691042985433,
      "learning_rate": 1.887720377151596e-06,
      "loss": 0.0053,
      "step": 12616
    },
    {
      "epoch": 8.0722968650032,
      "grad_norm": 0.44324955572198266,
      "learning_rate": 1.8865089006892478e-06,
      "loss": 0.0059,
      "step": 12617
    },
    {
      "epoch": 8.072936660268715,
      "grad_norm": 0.38330847946234525,
      "learning_rate": 1.8852977726057752e-06,
      "loss": 0.0051,
      "step": 12618
    },
    {
      "epoch": 8.073576455534228,
      "grad_norm": 0.492662739473812,
      "learning_rate": 1.884086992953179e-06,
      "loss": 0.008,
      "step": 12619
    },
    {
      "epoch": 8.074216250799743,
      "grad_norm": 0.4560164993362394,
      "learning_rate": 1.8828765617834521e-06,
      "loss": 0.0075,
      "step": 12620
    },
    {
      "epoch": 8.074856046065259,
      "grad_norm": 0.36910250959753377,
      "learning_rate": 1.8816664791485672e-06,
      "loss": 0.0054,
      "step": 12621
    },
    {
      "epoch": 8.075495841330774,
      "grad_norm": 0.42922270010280694,
      "learning_rate": 1.8804567451004818e-06,
      "loss": 0.0068,
      "step": 12622
    },
    {
      "epoch": 8.07613563659629,
      "grad_norm": 0.48321887458256774,
      "learning_rate": 1.8792473596911375e-06,
      "loss": 0.0079,
      "step": 12623
    },
    {
      "epoch": 8.076775431861805,
      "grad_norm": 0.3568996117752809,
      "learning_rate": 1.8780383229724685e-06,
      "loss": 0.0041,
      "step": 12624
    },
    {
      "epoch": 8.07741522712732,
      "grad_norm": 0.37074227858951,
      "learning_rate": 1.8768296349963855e-06,
      "loss": 0.0056,
      "step": 12625
    },
    {
      "epoch": 8.078055022392835,
      "grad_norm": 0.40868626070485764,
      "learning_rate": 1.8756212958147868e-06,
      "loss": 0.0066,
      "step": 12626
    },
    {
      "epoch": 8.078694817658349,
      "grad_norm": 0.3810362508297691,
      "learning_rate": 1.8744133054795588e-06,
      "loss": 0.005,
      "step": 12627
    },
    {
      "epoch": 8.079334612923864,
      "grad_norm": 0.4027005589669834,
      "learning_rate": 1.8732056640425699e-06,
      "loss": 0.0071,
      "step": 12628
    },
    {
      "epoch": 8.07997440818938,
      "grad_norm": 0.46897134682429764,
      "learning_rate": 1.8719983715556733e-06,
      "loss": 0.0059,
      "step": 12629
    },
    {
      "epoch": 8.080614203454894,
      "grad_norm": 0.39239180369234916,
      "learning_rate": 1.8707914280707073e-06,
      "loss": 0.0061,
      "step": 12630
    },
    {
      "epoch": 8.08125399872041,
      "grad_norm": 0.42485234660470217,
      "learning_rate": 1.8695848336394974e-06,
      "loss": 0.0065,
      "step": 12631
    },
    {
      "epoch": 8.081893793985925,
      "grad_norm": 0.4289099799394216,
      "learning_rate": 1.8683785883138516e-06,
      "loss": 0.0071,
      "step": 12632
    },
    {
      "epoch": 8.08253358925144,
      "grad_norm": 0.4295401642673043,
      "learning_rate": 1.8671726921455613e-06,
      "loss": 0.0061,
      "step": 12633
    },
    {
      "epoch": 8.083173384516954,
      "grad_norm": 0.3879829953397521,
      "learning_rate": 1.8659671451864114e-06,
      "loss": 0.0066,
      "step": 12634
    },
    {
      "epoch": 8.083813179782469,
      "grad_norm": 0.4047851193462028,
      "learning_rate": 1.864761947488163e-06,
      "loss": 0.0061,
      "step": 12635
    },
    {
      "epoch": 8.084452975047984,
      "grad_norm": 0.3483250726150666,
      "learning_rate": 1.8635570991025631e-06,
      "loss": 0.0058,
      "step": 12636
    },
    {
      "epoch": 8.0850927703135,
      "grad_norm": 0.40235231740957134,
      "learning_rate": 1.8623526000813496e-06,
      "loss": 0.0048,
      "step": 12637
    },
    {
      "epoch": 8.085732565579015,
      "grad_norm": 0.43919342144798157,
      "learning_rate": 1.861148450476239e-06,
      "loss": 0.0078,
      "step": 12638
    },
    {
      "epoch": 8.08637236084453,
      "grad_norm": 0.3273688323283511,
      "learning_rate": 1.859944650338935e-06,
      "loss": 0.0053,
      "step": 12639
    },
    {
      "epoch": 8.087012156110045,
      "grad_norm": 0.46812758993832243,
      "learning_rate": 1.858741199721129e-06,
      "loss": 0.0074,
      "step": 12640
    },
    {
      "epoch": 8.08765195137556,
      "grad_norm": 0.35875522358170947,
      "learning_rate": 1.857538098674494e-06,
      "loss": 0.0053,
      "step": 12641
    },
    {
      "epoch": 8.088291746641074,
      "grad_norm": 0.50199089353103,
      "learning_rate": 1.8563353472506895e-06,
      "loss": 0.008,
      "step": 12642
    },
    {
      "epoch": 8.08893154190659,
      "grad_norm": 0.45009238815562824,
      "learning_rate": 1.8551329455013556e-06,
      "loss": 0.0075,
      "step": 12643
    },
    {
      "epoch": 8.089571337172105,
      "grad_norm": 0.38662717769691735,
      "learning_rate": 1.8539308934781265e-06,
      "loss": 0.0057,
      "step": 12644
    },
    {
      "epoch": 8.09021113243762,
      "grad_norm": 0.3182956949268309,
      "learning_rate": 1.8527291912326151e-06,
      "loss": 0.0049,
      "step": 12645
    },
    {
      "epoch": 8.090850927703135,
      "grad_norm": 0.4806514906697556,
      "learning_rate": 1.8515278388164182e-06,
      "loss": 0.0061,
      "step": 12646
    },
    {
      "epoch": 8.09149072296865,
      "grad_norm": 0.4386133655005112,
      "learning_rate": 1.8503268362811222e-06,
      "loss": 0.0073,
      "step": 12647
    },
    {
      "epoch": 8.092130518234166,
      "grad_norm": 0.5037563654433503,
      "learning_rate": 1.8491261836782936e-06,
      "loss": 0.0078,
      "step": 12648
    },
    {
      "epoch": 8.09277031349968,
      "grad_norm": 0.4362519893887088,
      "learning_rate": 1.847925881059488e-06,
      "loss": 0.0059,
      "step": 12649
    },
    {
      "epoch": 8.093410108765195,
      "grad_norm": 0.38686130191276635,
      "learning_rate": 1.8467259284762407e-06,
      "loss": 0.0057,
      "step": 12650
    },
    {
      "epoch": 8.09404990403071,
      "grad_norm": 0.44049277425830513,
      "learning_rate": 1.8455263259800815e-06,
      "loss": 0.0074,
      "step": 12651
    },
    {
      "epoch": 8.094689699296225,
      "grad_norm": 0.3538573712127394,
      "learning_rate": 1.844327073622516e-06,
      "loss": 0.006,
      "step": 12652
    },
    {
      "epoch": 8.09532949456174,
      "grad_norm": 0.3587819356568097,
      "learning_rate": 1.8431281714550354e-06,
      "loss": 0.0054,
      "step": 12653
    },
    {
      "epoch": 8.095969289827256,
      "grad_norm": 0.3720011260357874,
      "learning_rate": 1.8419296195291236e-06,
      "loss": 0.006,
      "step": 12654
    },
    {
      "epoch": 8.096609085092771,
      "grad_norm": 0.4726754756907371,
      "learning_rate": 1.8407314178962421e-06,
      "loss": 0.0061,
      "step": 12655
    },
    {
      "epoch": 8.097248880358284,
      "grad_norm": 0.3669488802248791,
      "learning_rate": 1.8395335666078373e-06,
      "loss": 0.0064,
      "step": 12656
    },
    {
      "epoch": 8.0978886756238,
      "grad_norm": 0.39055772403164857,
      "learning_rate": 1.838336065715347e-06,
      "loss": 0.0064,
      "step": 12657
    },
    {
      "epoch": 8.098528470889315,
      "grad_norm": 0.44725746541403294,
      "learning_rate": 1.8371389152701868e-06,
      "loss": 0.0075,
      "step": 12658
    },
    {
      "epoch": 8.09916826615483,
      "grad_norm": 0.4164002987437643,
      "learning_rate": 1.8359421153237623e-06,
      "loss": 0.0059,
      "step": 12659
    },
    {
      "epoch": 8.099808061420346,
      "grad_norm": 0.48197986487132893,
      "learning_rate": 1.8347456659274587e-06,
      "loss": 0.0081,
      "step": 12660
    },
    {
      "epoch": 8.10044785668586,
      "grad_norm": 0.37777803300210805,
      "learning_rate": 1.833549567132653e-06,
      "loss": 0.0061,
      "step": 12661
    },
    {
      "epoch": 8.101087651951376,
      "grad_norm": 0.32490278999015126,
      "learning_rate": 1.832353818990702e-06,
      "loss": 0.0044,
      "step": 12662
    },
    {
      "epoch": 8.101727447216891,
      "grad_norm": 0.4246890730334709,
      "learning_rate": 1.8311584215529454e-06,
      "loss": 0.0067,
      "step": 12663
    },
    {
      "epoch": 8.102367242482405,
      "grad_norm": 0.46429660656048694,
      "learning_rate": 1.8299633748707179e-06,
      "loss": 0.0062,
      "step": 12664
    },
    {
      "epoch": 8.10300703774792,
      "grad_norm": 0.43204623044838764,
      "learning_rate": 1.8287686789953296e-06,
      "loss": 0.006,
      "step": 12665
    },
    {
      "epoch": 8.103646833013435,
      "grad_norm": 0.35180608484604003,
      "learning_rate": 1.827574333978076e-06,
      "loss": 0.0056,
      "step": 12666
    },
    {
      "epoch": 8.10428662827895,
      "grad_norm": 0.3783962019030857,
      "learning_rate": 1.8263803398702462e-06,
      "loss": 0.0059,
      "step": 12667
    },
    {
      "epoch": 8.104926423544466,
      "grad_norm": 0.3911739235798467,
      "learning_rate": 1.8251866967231035e-06,
      "loss": 0.0067,
      "step": 12668
    },
    {
      "epoch": 8.105566218809981,
      "grad_norm": 0.3480961010596643,
      "learning_rate": 1.8239934045879038e-06,
      "loss": 0.0048,
      "step": 12669
    },
    {
      "epoch": 8.106206014075497,
      "grad_norm": 0.5006350727943748,
      "learning_rate": 1.8228004635158802e-06,
      "loss": 0.0077,
      "step": 12670
    },
    {
      "epoch": 8.10684580934101,
      "grad_norm": 0.41807246926398006,
      "learning_rate": 1.821607873558261e-06,
      "loss": 0.0067,
      "step": 12671
    },
    {
      "epoch": 8.107485604606525,
      "grad_norm": 0.41337268782435793,
      "learning_rate": 1.8204156347662505e-06,
      "loss": 0.0068,
      "step": 12672
    },
    {
      "epoch": 8.10812539987204,
      "grad_norm": 0.3657281336682914,
      "learning_rate": 1.8192237471910412e-06,
      "loss": 0.0055,
      "step": 12673
    },
    {
      "epoch": 8.108765195137556,
      "grad_norm": 0.470568125410643,
      "learning_rate": 1.8180322108838134e-06,
      "loss": 0.0059,
      "step": 12674
    },
    {
      "epoch": 8.109404990403071,
      "grad_norm": 0.40728358829748373,
      "learning_rate": 1.816841025895728e-06,
      "loss": 0.006,
      "step": 12675
    },
    {
      "epoch": 8.110044785668586,
      "grad_norm": 0.4482191351596994,
      "learning_rate": 1.8156501922779335e-06,
      "loss": 0.0066,
      "step": 12676
    },
    {
      "epoch": 8.110684580934102,
      "grad_norm": 0.4678612163372572,
      "learning_rate": 1.8144597100815597e-06,
      "loss": 0.0062,
      "step": 12677
    },
    {
      "epoch": 8.111324376199617,
      "grad_norm": 0.5162297098871823,
      "learning_rate": 1.8132695793577248e-06,
      "loss": 0.0089,
      "step": 12678
    },
    {
      "epoch": 8.11196417146513,
      "grad_norm": 0.4440611132936173,
      "learning_rate": 1.8120798001575312e-06,
      "loss": 0.0066,
      "step": 12679
    },
    {
      "epoch": 8.112603966730646,
      "grad_norm": 0.38902984968821774,
      "learning_rate": 1.8108903725320637e-06,
      "loss": 0.006,
      "step": 12680
    },
    {
      "epoch": 8.113243761996161,
      "grad_norm": 0.4567127696306382,
      "learning_rate": 1.809701296532399e-06,
      "loss": 0.0079,
      "step": 12681
    },
    {
      "epoch": 8.113883557261676,
      "grad_norm": 0.40100858140110485,
      "learning_rate": 1.8085125722095909e-06,
      "loss": 0.0055,
      "step": 12682
    },
    {
      "epoch": 8.114523352527192,
      "grad_norm": 0.4574503352127965,
      "learning_rate": 1.8073241996146785e-06,
      "loss": 0.0067,
      "step": 12683
    },
    {
      "epoch": 8.115163147792707,
      "grad_norm": 0.41640522411650616,
      "learning_rate": 1.8061361787986942e-06,
      "loss": 0.0055,
      "step": 12684
    },
    {
      "epoch": 8.115802943058222,
      "grad_norm": 0.39788043869326967,
      "learning_rate": 1.8049485098126474e-06,
      "loss": 0.0049,
      "step": 12685
    },
    {
      "epoch": 8.116442738323736,
      "grad_norm": 0.4126096295577201,
      "learning_rate": 1.8037611927075294e-06,
      "loss": 0.0063,
      "step": 12686
    },
    {
      "epoch": 8.117082533589251,
      "grad_norm": 0.4017772878131167,
      "learning_rate": 1.8025742275343295e-06,
      "loss": 0.0054,
      "step": 12687
    },
    {
      "epoch": 8.117722328854766,
      "grad_norm": 0.46398437479122895,
      "learning_rate": 1.8013876143440101e-06,
      "loss": 0.0069,
      "step": 12688
    },
    {
      "epoch": 8.118362124120281,
      "grad_norm": 0.4247833620753187,
      "learning_rate": 1.8002013531875228e-06,
      "loss": 0.0063,
      "step": 12689
    },
    {
      "epoch": 8.119001919385797,
      "grad_norm": 0.49450174834552546,
      "learning_rate": 1.7990154441158025e-06,
      "loss": 0.0063,
      "step": 12690
    },
    {
      "epoch": 8.119641714651312,
      "grad_norm": 0.4251028085744135,
      "learning_rate": 1.7978298871797705e-06,
      "loss": 0.0065,
      "step": 12691
    },
    {
      "epoch": 8.120281509916827,
      "grad_norm": 0.45131825580847196,
      "learning_rate": 1.796644682430333e-06,
      "loss": 0.0073,
      "step": 12692
    },
    {
      "epoch": 8.12092130518234,
      "grad_norm": 0.4350977890351921,
      "learning_rate": 1.7954598299183778e-06,
      "loss": 0.0069,
      "step": 12693
    },
    {
      "epoch": 8.121561100447856,
      "grad_norm": 0.4442982710344643,
      "learning_rate": 1.7942753296947846e-06,
      "loss": 0.0069,
      "step": 12694
    },
    {
      "epoch": 8.122200895713371,
      "grad_norm": 0.48203372693482005,
      "learning_rate": 1.7930911818104125e-06,
      "loss": 0.0069,
      "step": 12695
    },
    {
      "epoch": 8.122840690978887,
      "grad_norm": 0.3863954185465748,
      "learning_rate": 1.791907386316103e-06,
      "loss": 0.0047,
      "step": 12696
    },
    {
      "epoch": 8.123480486244402,
      "grad_norm": 0.5679299627121865,
      "learning_rate": 1.7907239432626922e-06,
      "loss": 0.0086,
      "step": 12697
    },
    {
      "epoch": 8.124120281509917,
      "grad_norm": 0.45542035659880054,
      "learning_rate": 1.7895408527009917e-06,
      "loss": 0.0067,
      "step": 12698
    },
    {
      "epoch": 8.124760076775432,
      "grad_norm": 0.3738937376687491,
      "learning_rate": 1.788358114681803e-06,
      "loss": 0.0053,
      "step": 12699
    },
    {
      "epoch": 8.125399872040948,
      "grad_norm": 0.47464315722312367,
      "learning_rate": 1.7871757292559056e-06,
      "loss": 0.0076,
      "step": 12700
    },
    {
      "epoch": 8.126039667306461,
      "grad_norm": 0.5776939580318874,
      "learning_rate": 1.7859936964740754e-06,
      "loss": 0.0097,
      "step": 12701
    },
    {
      "epoch": 8.126679462571976,
      "grad_norm": 0.4198319731150127,
      "learning_rate": 1.7848120163870653e-06,
      "loss": 0.007,
      "step": 12702
    },
    {
      "epoch": 8.127319257837492,
      "grad_norm": 0.4710500053364376,
      "learning_rate": 1.7836306890456102e-06,
      "loss": 0.0076,
      "step": 12703
    },
    {
      "epoch": 8.127959053103007,
      "grad_norm": 0.4403672175256589,
      "learning_rate": 1.7824497145004404e-06,
      "loss": 0.0065,
      "step": 12704
    },
    {
      "epoch": 8.128598848368522,
      "grad_norm": 0.4560557534422248,
      "learning_rate": 1.7812690928022614e-06,
      "loss": 0.0068,
      "step": 12705
    },
    {
      "epoch": 8.129238643634038,
      "grad_norm": 0.3747727950131596,
      "learning_rate": 1.7800888240017688e-06,
      "loss": 0.0054,
      "step": 12706
    },
    {
      "epoch": 8.129878438899553,
      "grad_norm": 0.42171203127059315,
      "learning_rate": 1.7789089081496391e-06,
      "loss": 0.0059,
      "step": 12707
    },
    {
      "epoch": 8.130518234165066,
      "grad_norm": 0.4384279491169966,
      "learning_rate": 1.777729345296536e-06,
      "loss": 0.0071,
      "step": 12708
    },
    {
      "epoch": 8.131158029430582,
      "grad_norm": 0.38039343812372933,
      "learning_rate": 1.7765501354931103e-06,
      "loss": 0.0064,
      "step": 12709
    },
    {
      "epoch": 8.131797824696097,
      "grad_norm": 0.33325639603541085,
      "learning_rate": 1.775371278789989e-06,
      "loss": 0.0042,
      "step": 12710
    },
    {
      "epoch": 8.132437619961612,
      "grad_norm": 0.3625873676549834,
      "learning_rate": 1.7741927752377986e-06,
      "loss": 0.0056,
      "step": 12711
    },
    {
      "epoch": 8.133077415227127,
      "grad_norm": 0.34092273720128957,
      "learning_rate": 1.7730146248871371e-06,
      "loss": 0.0045,
      "step": 12712
    },
    {
      "epoch": 8.133717210492643,
      "grad_norm": 0.44763729938802055,
      "learning_rate": 1.77183682778859e-06,
      "loss": 0.0066,
      "step": 12713
    },
    {
      "epoch": 8.134357005758158,
      "grad_norm": 0.5287330233442837,
      "learning_rate": 1.7706593839927366e-06,
      "loss": 0.0082,
      "step": 12714
    },
    {
      "epoch": 8.134996801023673,
      "grad_norm": 0.44725575092386405,
      "learning_rate": 1.7694822935501288e-06,
      "loss": 0.0076,
      "step": 12715
    },
    {
      "epoch": 8.135636596289187,
      "grad_norm": 0.40827982916639016,
      "learning_rate": 1.7683055565113094e-06,
      "loss": 0.006,
      "step": 12716
    },
    {
      "epoch": 8.136276391554702,
      "grad_norm": 0.39974929197169523,
      "learning_rate": 1.7671291729268081e-06,
      "loss": 0.0058,
      "step": 12717
    },
    {
      "epoch": 8.136916186820217,
      "grad_norm": 0.4439098921062927,
      "learning_rate": 1.7659531428471343e-06,
      "loss": 0.0057,
      "step": 12718
    },
    {
      "epoch": 8.137555982085733,
      "grad_norm": 0.4757208497320024,
      "learning_rate": 1.7647774663227867e-06,
      "loss": 0.0066,
      "step": 12719
    },
    {
      "epoch": 8.138195777351248,
      "grad_norm": 0.5021868314441311,
      "learning_rate": 1.763602143404244e-06,
      "loss": 0.0067,
      "step": 12720
    },
    {
      "epoch": 8.138835572616763,
      "grad_norm": 0.49697751021412406,
      "learning_rate": 1.7624271741419752e-06,
      "loss": 0.0086,
      "step": 12721
    },
    {
      "epoch": 8.139475367882278,
      "grad_norm": 0.44057385115291353,
      "learning_rate": 1.7612525585864293e-06,
      "loss": 0.0058,
      "step": 12722
    },
    {
      "epoch": 8.140115163147792,
      "grad_norm": 0.4888828777352772,
      "learning_rate": 1.7600782967880404e-06,
      "loss": 0.0082,
      "step": 12723
    },
    {
      "epoch": 8.140754958413307,
      "grad_norm": 0.4589719306953728,
      "learning_rate": 1.7589043887972347e-06,
      "loss": 0.0052,
      "step": 12724
    },
    {
      "epoch": 8.141394753678822,
      "grad_norm": 0.49225038354397305,
      "learning_rate": 1.7577308346644139e-06,
      "loss": 0.0067,
      "step": 12725
    },
    {
      "epoch": 8.142034548944338,
      "grad_norm": 0.40747644550630147,
      "learning_rate": 1.7565576344399693e-06,
      "loss": 0.0071,
      "step": 12726
    },
    {
      "epoch": 8.142674344209853,
      "grad_norm": 0.48457350801391663,
      "learning_rate": 1.7553847881742737e-06,
      "loss": 0.0057,
      "step": 12727
    },
    {
      "epoch": 8.143314139475368,
      "grad_norm": 0.3851025858333066,
      "learning_rate": 1.7542122959176921e-06,
      "loss": 0.0054,
      "step": 12728
    },
    {
      "epoch": 8.143953934740884,
      "grad_norm": 0.6037475541010817,
      "learning_rate": 1.753040157720567e-06,
      "loss": 0.0086,
      "step": 12729
    },
    {
      "epoch": 8.144593730006397,
      "grad_norm": 0.40528130118167177,
      "learning_rate": 1.7518683736332233e-06,
      "loss": 0.0054,
      "step": 12730
    },
    {
      "epoch": 8.145233525271912,
      "grad_norm": 0.45236804985982326,
      "learning_rate": 1.750696943705983e-06,
      "loss": 0.0055,
      "step": 12731
    },
    {
      "epoch": 8.145873320537428,
      "grad_norm": 0.4326449604490435,
      "learning_rate": 1.7495258679891414e-06,
      "loss": 0.0058,
      "step": 12732
    },
    {
      "epoch": 8.146513115802943,
      "grad_norm": 0.34282545653131463,
      "learning_rate": 1.7483551465329796e-06,
      "loss": 0.005,
      "step": 12733
    },
    {
      "epoch": 8.147152911068458,
      "grad_norm": 0.3986178404686815,
      "learning_rate": 1.7471847793877716e-06,
      "loss": 0.006,
      "step": 12734
    },
    {
      "epoch": 8.147792706333973,
      "grad_norm": 0.40852855789482545,
      "learning_rate": 1.7460147666037686e-06,
      "loss": 0.0069,
      "step": 12735
    },
    {
      "epoch": 8.148432501599489,
      "grad_norm": 0.46135641422942286,
      "learning_rate": 1.744845108231209e-06,
      "loss": 0.0072,
      "step": 12736
    },
    {
      "epoch": 8.149072296865004,
      "grad_norm": 0.4682731159865328,
      "learning_rate": 1.7436758043203161e-06,
      "loss": 0.0062,
      "step": 12737
    },
    {
      "epoch": 8.149712092130518,
      "grad_norm": 0.4296563945841116,
      "learning_rate": 1.7425068549212975e-06,
      "loss": 0.0065,
      "step": 12738
    },
    {
      "epoch": 8.150351887396033,
      "grad_norm": 0.4250845718367122,
      "learning_rate": 1.7413382600843455e-06,
      "loss": 0.007,
      "step": 12739
    },
    {
      "epoch": 8.150991682661548,
      "grad_norm": 0.41119278279831006,
      "learning_rate": 1.740170019859634e-06,
      "loss": 0.0071,
      "step": 12740
    },
    {
      "epoch": 8.151631477927063,
      "grad_norm": 0.47170751926965576,
      "learning_rate": 1.7390021342973318e-06,
      "loss": 0.0076,
      "step": 12741
    },
    {
      "epoch": 8.152271273192579,
      "grad_norm": 0.4218016114308907,
      "learning_rate": 1.7378346034475834e-06,
      "loss": 0.0054,
      "step": 12742
    },
    {
      "epoch": 8.152911068458094,
      "grad_norm": 0.42820696797900437,
      "learning_rate": 1.7366674273605166e-06,
      "loss": 0.0061,
      "step": 12743
    },
    {
      "epoch": 8.15355086372361,
      "grad_norm": 0.4910566734252451,
      "learning_rate": 1.7355006060862544e-06,
      "loss": 0.0062,
      "step": 12744
    },
    {
      "epoch": 8.154190658989123,
      "grad_norm": 0.4559728234405737,
      "learning_rate": 1.7343341396748935e-06,
      "loss": 0.0069,
      "step": 12745
    },
    {
      "epoch": 8.154830454254638,
      "grad_norm": 0.4730023887311357,
      "learning_rate": 1.7331680281765228e-06,
      "loss": 0.0071,
      "step": 12746
    },
    {
      "epoch": 8.155470249520153,
      "grad_norm": 0.3930320042464431,
      "learning_rate": 1.7320022716412077e-06,
      "loss": 0.005,
      "step": 12747
    },
    {
      "epoch": 8.156110044785668,
      "grad_norm": 0.4751079641209024,
      "learning_rate": 1.7308368701190104e-06,
      "loss": 0.0062,
      "step": 12748
    },
    {
      "epoch": 8.156749840051184,
      "grad_norm": 0.4234287398996055,
      "learning_rate": 1.7296718236599686e-06,
      "loss": 0.0059,
      "step": 12749
    },
    {
      "epoch": 8.157389635316699,
      "grad_norm": 0.45011197127668917,
      "learning_rate": 1.7285071323141068e-06,
      "loss": 0.0062,
      "step": 12750
    },
    {
      "epoch": 8.158029430582214,
      "grad_norm": 0.29837075576961924,
      "learning_rate": 1.7273427961314337e-06,
      "loss": 0.0038,
      "step": 12751
    },
    {
      "epoch": 8.15866922584773,
      "grad_norm": 0.4029696555148285,
      "learning_rate": 1.7261788151619464e-06,
      "loss": 0.0055,
      "step": 12752
    },
    {
      "epoch": 8.159309021113243,
      "grad_norm": 0.43834960991542893,
      "learning_rate": 1.72501518945562e-06,
      "loss": 0.0072,
      "step": 12753
    },
    {
      "epoch": 8.159948816378758,
      "grad_norm": 0.5538755775807073,
      "learning_rate": 1.7238519190624226e-06,
      "loss": 0.0078,
      "step": 12754
    },
    {
      "epoch": 8.160588611644274,
      "grad_norm": 0.511542081424264,
      "learning_rate": 1.722689004032303e-06,
      "loss": 0.0077,
      "step": 12755
    },
    {
      "epoch": 8.161228406909789,
      "grad_norm": 0.3881552856053649,
      "learning_rate": 1.7215264444151935e-06,
      "loss": 0.0066,
      "step": 12756
    },
    {
      "epoch": 8.161868202175304,
      "grad_norm": 0.425948249135335,
      "learning_rate": 1.7203642402610087e-06,
      "loss": 0.007,
      "step": 12757
    },
    {
      "epoch": 8.16250799744082,
      "grad_norm": 0.4635128714536437,
      "learning_rate": 1.7192023916196587e-06,
      "loss": 0.0075,
      "step": 12758
    },
    {
      "epoch": 8.163147792706335,
      "grad_norm": 0.39543013800186594,
      "learning_rate": 1.7180408985410269e-06,
      "loss": 0.006,
      "step": 12759
    },
    {
      "epoch": 8.163787587971848,
      "grad_norm": 0.5479202007632853,
      "learning_rate": 1.7168797610749844e-06,
      "loss": 0.0071,
      "step": 12760
    },
    {
      "epoch": 8.164427383237364,
      "grad_norm": 0.48870615689410035,
      "learning_rate": 1.7157189792713924e-06,
      "loss": 0.0065,
      "step": 12761
    },
    {
      "epoch": 8.165067178502879,
      "grad_norm": 0.45585691332020817,
      "learning_rate": 1.7145585531800912e-06,
      "loss": 0.0068,
      "step": 12762
    },
    {
      "epoch": 8.165706973768394,
      "grad_norm": 0.3913314122252132,
      "learning_rate": 1.7133984828509043e-06,
      "loss": 0.0054,
      "step": 12763
    },
    {
      "epoch": 8.16634676903391,
      "grad_norm": 0.32269621329762505,
      "learning_rate": 1.7122387683336483e-06,
      "loss": 0.0052,
      "step": 12764
    },
    {
      "epoch": 8.166986564299425,
      "grad_norm": 0.4001001117496467,
      "learning_rate": 1.711079409678117e-06,
      "loss": 0.0073,
      "step": 12765
    },
    {
      "epoch": 8.16762635956494,
      "grad_norm": 0.38190458042263287,
      "learning_rate": 1.7099204069340902e-06,
      "loss": 0.0051,
      "step": 12766
    },
    {
      "epoch": 8.168266154830455,
      "grad_norm": 0.4524685946037837,
      "learning_rate": 1.7087617601513352e-06,
      "loss": 0.0056,
      "step": 12767
    },
    {
      "epoch": 8.168905950095969,
      "grad_norm": 0.3791847929800885,
      "learning_rate": 1.7076034693796018e-06,
      "loss": 0.0059,
      "step": 12768
    },
    {
      "epoch": 8.169545745361484,
      "grad_norm": 0.35080047323073,
      "learning_rate": 1.7064455346686227e-06,
      "loss": 0.0046,
      "step": 12769
    },
    {
      "epoch": 8.170185540627,
      "grad_norm": 0.3565586118301213,
      "learning_rate": 1.705287956068118e-06,
      "loss": 0.0047,
      "step": 12770
    },
    {
      "epoch": 8.170825335892514,
      "grad_norm": 0.41728289962871334,
      "learning_rate": 1.7041307336277957e-06,
      "loss": 0.0054,
      "step": 12771
    },
    {
      "epoch": 8.17146513115803,
      "grad_norm": 0.38455426985400953,
      "learning_rate": 1.7029738673973417e-06,
      "loss": 0.005,
      "step": 12772
    },
    {
      "epoch": 8.172104926423545,
      "grad_norm": 0.43394943408511477,
      "learning_rate": 1.7018173574264295e-06,
      "loss": 0.0064,
      "step": 12773
    },
    {
      "epoch": 8.17274472168906,
      "grad_norm": 0.32754913604058855,
      "learning_rate": 1.7006612037647196e-06,
      "loss": 0.0048,
      "step": 12774
    },
    {
      "epoch": 8.173384516954574,
      "grad_norm": 0.4533110483145973,
      "learning_rate": 1.6995054064618554e-06,
      "loss": 0.0064,
      "step": 12775
    },
    {
      "epoch": 8.174024312220089,
      "grad_norm": 0.4466365878257805,
      "learning_rate": 1.698349965567463e-06,
      "loss": 0.0055,
      "step": 12776
    },
    {
      "epoch": 8.174664107485604,
      "grad_norm": 0.45385256546273967,
      "learning_rate": 1.6971948811311533e-06,
      "loss": 0.0061,
      "step": 12777
    },
    {
      "epoch": 8.17530390275112,
      "grad_norm": 0.3702966353861636,
      "learning_rate": 1.6960401532025283e-06,
      "loss": 0.0056,
      "step": 12778
    },
    {
      "epoch": 8.175943698016635,
      "grad_norm": 0.42818952156876705,
      "learning_rate": 1.694885781831167e-06,
      "loss": 0.0059,
      "step": 12779
    },
    {
      "epoch": 8.17658349328215,
      "grad_norm": 0.4018283348575647,
      "learning_rate": 1.6937317670666375e-06,
      "loss": 0.0051,
      "step": 12780
    },
    {
      "epoch": 8.177223288547665,
      "grad_norm": 0.47324454304744656,
      "learning_rate": 1.6925781089584903e-06,
      "loss": 0.0074,
      "step": 12781
    },
    {
      "epoch": 8.177863083813179,
      "grad_norm": 0.5022492232965252,
      "learning_rate": 1.69142480755626e-06,
      "loss": 0.0065,
      "step": 12782
    },
    {
      "epoch": 8.178502879078694,
      "grad_norm": 0.5282233453247589,
      "learning_rate": 1.6902718629094673e-06,
      "loss": 0.0077,
      "step": 12783
    },
    {
      "epoch": 8.17914267434421,
      "grad_norm": 0.49316002310908785,
      "learning_rate": 1.6891192750676212e-06,
      "loss": 0.0079,
      "step": 12784
    },
    {
      "epoch": 8.179782469609725,
      "grad_norm": 0.4513319615958747,
      "learning_rate": 1.6879670440802099e-06,
      "loss": 0.0058,
      "step": 12785
    },
    {
      "epoch": 8.18042226487524,
      "grad_norm": 0.390190374020194,
      "learning_rate": 1.6868151699967072e-06,
      "loss": 0.0054,
      "step": 12786
    },
    {
      "epoch": 8.181062060140755,
      "grad_norm": 0.4203013446221984,
      "learning_rate": 1.6856636528665705e-06,
      "loss": 0.0065,
      "step": 12787
    },
    {
      "epoch": 8.18170185540627,
      "grad_norm": 0.37400104929610567,
      "learning_rate": 1.6845124927392498e-06,
      "loss": 0.0052,
      "step": 12788
    },
    {
      "epoch": 8.182341650671786,
      "grad_norm": 0.37296541944439904,
      "learning_rate": 1.6833616896641692e-06,
      "loss": 0.0048,
      "step": 12789
    },
    {
      "epoch": 8.1829814459373,
      "grad_norm": 0.4486720427045273,
      "learning_rate": 1.6822112436907422e-06,
      "loss": 0.0061,
      "step": 12790
    },
    {
      "epoch": 8.183621241202815,
      "grad_norm": 0.44383931570722285,
      "learning_rate": 1.6810611548683697e-06,
      "loss": 0.0041,
      "step": 12791
    },
    {
      "epoch": 8.18426103646833,
      "grad_norm": 0.37378061119107486,
      "learning_rate": 1.6799114232464341e-06,
      "loss": 0.0067,
      "step": 12792
    },
    {
      "epoch": 8.184900831733845,
      "grad_norm": 0.37588924041509164,
      "learning_rate": 1.6787620488742984e-06,
      "loss": 0.0049,
      "step": 12793
    },
    {
      "epoch": 8.18554062699936,
      "grad_norm": 0.5016154527539278,
      "learning_rate": 1.6776130318013207e-06,
      "loss": 0.007,
      "step": 12794
    },
    {
      "epoch": 8.186180422264876,
      "grad_norm": 0.46048108130656484,
      "learning_rate": 1.6764643720768347e-06,
      "loss": 0.0071,
      "step": 12795
    },
    {
      "epoch": 8.186820217530391,
      "grad_norm": 0.47996938830681596,
      "learning_rate": 1.6753160697501625e-06,
      "loss": 0.0069,
      "step": 12796
    },
    {
      "epoch": 8.187460012795905,
      "grad_norm": 0.340567957121149,
      "learning_rate": 1.6741681248706099e-06,
      "loss": 0.0045,
      "step": 12797
    },
    {
      "epoch": 8.18809980806142,
      "grad_norm": 0.45567917428678895,
      "learning_rate": 1.6730205374874674e-06,
      "loss": 0.0073,
      "step": 12798
    },
    {
      "epoch": 8.188739603326935,
      "grad_norm": 0.2407120990635701,
      "learning_rate": 1.6718733076500104e-06,
      "loss": 0.0028,
      "step": 12799
    },
    {
      "epoch": 8.18937939859245,
      "grad_norm": 0.39680758660344834,
      "learning_rate": 1.670726435407497e-06,
      "loss": 0.0056,
      "step": 12800
    },
    {
      "epoch": 8.190019193857966,
      "grad_norm": 0.40679024194142194,
      "learning_rate": 1.6695799208091757e-06,
      "loss": 0.0067,
      "step": 12801
    },
    {
      "epoch": 8.190658989123481,
      "grad_norm": 0.43428404089543776,
      "learning_rate": 1.6684337639042748e-06,
      "loss": 0.0057,
      "step": 12802
    },
    {
      "epoch": 8.191298784388996,
      "grad_norm": 0.46164812505295005,
      "learning_rate": 1.667287964742006e-06,
      "loss": 0.0071,
      "step": 12803
    },
    {
      "epoch": 8.191938579654511,
      "grad_norm": 0.3446788072466963,
      "learning_rate": 1.6661425233715677e-06,
      "loss": 0.0053,
      "step": 12804
    },
    {
      "epoch": 8.192578374920025,
      "grad_norm": 0.3899762255221645,
      "learning_rate": 1.6649974398421475e-06,
      "loss": 0.0055,
      "step": 12805
    },
    {
      "epoch": 8.19321817018554,
      "grad_norm": 0.38229014676614154,
      "learning_rate": 1.6638527142029093e-06,
      "loss": 0.0062,
      "step": 12806
    },
    {
      "epoch": 8.193857965451055,
      "grad_norm": 0.37114709067811485,
      "learning_rate": 1.6627083465030048e-06,
      "loss": 0.0049,
      "step": 12807
    },
    {
      "epoch": 8.19449776071657,
      "grad_norm": 0.39408998763679165,
      "learning_rate": 1.6615643367915757e-06,
      "loss": 0.0058,
      "step": 12808
    },
    {
      "epoch": 8.195137555982086,
      "grad_norm": 0.41908100195747566,
      "learning_rate": 1.6604206851177407e-06,
      "loss": 0.0062,
      "step": 12809
    },
    {
      "epoch": 8.195777351247601,
      "grad_norm": 0.38777745913085676,
      "learning_rate": 1.659277391530606e-06,
      "loss": 0.0052,
      "step": 12810
    },
    {
      "epoch": 8.196417146513117,
      "grad_norm": 0.4498702882576427,
      "learning_rate": 1.6581344560792644e-06,
      "loss": 0.0073,
      "step": 12811
    },
    {
      "epoch": 8.19705694177863,
      "grad_norm": 0.4224919935292318,
      "learning_rate": 1.6569918788127893e-06,
      "loss": 0.0065,
      "step": 12812
    },
    {
      "epoch": 8.197696737044145,
      "grad_norm": 0.3886143099490104,
      "learning_rate": 1.655849659780242e-06,
      "loss": 0.0052,
      "step": 12813
    },
    {
      "epoch": 8.19833653230966,
      "grad_norm": 0.48085805118483477,
      "learning_rate": 1.6547077990306647e-06,
      "loss": 0.0057,
      "step": 12814
    },
    {
      "epoch": 8.198976327575176,
      "grad_norm": 0.4166400153819978,
      "learning_rate": 1.6535662966130927e-06,
      "loss": 0.006,
      "step": 12815
    },
    {
      "epoch": 8.199616122840691,
      "grad_norm": 0.553289004127435,
      "learning_rate": 1.6524251525765355e-06,
      "loss": 0.0074,
      "step": 12816
    },
    {
      "epoch": 8.200255918106206,
      "grad_norm": 0.38256344339366244,
      "learning_rate": 1.6512843669699919e-06,
      "loss": 0.0049,
      "step": 12817
    },
    {
      "epoch": 8.200895713371722,
      "grad_norm": 0.40092122646370465,
      "learning_rate": 1.6501439398424479e-06,
      "loss": 0.0059,
      "step": 12818
    },
    {
      "epoch": 8.201535508637235,
      "grad_norm": 0.41335879537938097,
      "learning_rate": 1.6490038712428701e-06,
      "loss": 0.0061,
      "step": 12819
    },
    {
      "epoch": 8.20217530390275,
      "grad_norm": 0.386635568774325,
      "learning_rate": 1.6478641612202085e-06,
      "loss": 0.0055,
      "step": 12820
    },
    {
      "epoch": 8.202815099168266,
      "grad_norm": 0.5904110188311044,
      "learning_rate": 1.6467248098234045e-06,
      "loss": 0.0054,
      "step": 12821
    },
    {
      "epoch": 8.203454894433781,
      "grad_norm": 0.49828171308619373,
      "learning_rate": 1.6455858171013783e-06,
      "loss": 0.0068,
      "step": 12822
    },
    {
      "epoch": 8.204094689699296,
      "grad_norm": 0.3665388964050344,
      "learning_rate": 1.644447183103035e-06,
      "loss": 0.0052,
      "step": 12823
    },
    {
      "epoch": 8.204734484964812,
      "grad_norm": 0.352615111539862,
      "learning_rate": 1.6433089078772646e-06,
      "loss": 0.0049,
      "step": 12824
    },
    {
      "epoch": 8.205374280230327,
      "grad_norm": 0.4423303304350615,
      "learning_rate": 1.6421709914729466e-06,
      "loss": 0.0063,
      "step": 12825
    },
    {
      "epoch": 8.206014075495842,
      "grad_norm": 0.4751884809747827,
      "learning_rate": 1.6410334339389389e-06,
      "loss": 0.0063,
      "step": 12826
    },
    {
      "epoch": 8.206653870761356,
      "grad_norm": 0.41473520597381086,
      "learning_rate": 1.6398962353240856e-06,
      "loss": 0.0051,
      "step": 12827
    },
    {
      "epoch": 8.207293666026871,
      "grad_norm": 0.4426140502222422,
      "learning_rate": 1.6387593956772175e-06,
      "loss": 0.0066,
      "step": 12828
    },
    {
      "epoch": 8.207933461292386,
      "grad_norm": 0.5661495434891896,
      "learning_rate": 1.6376229150471467e-06,
      "loss": 0.0056,
      "step": 12829
    },
    {
      "epoch": 8.208573256557901,
      "grad_norm": 0.36157297011028816,
      "learning_rate": 1.63648679348267e-06,
      "loss": 0.0048,
      "step": 12830
    },
    {
      "epoch": 8.209213051823417,
      "grad_norm": 0.39675100175700756,
      "learning_rate": 1.635351031032576e-06,
      "loss": 0.0055,
      "step": 12831
    },
    {
      "epoch": 8.209852847088932,
      "grad_norm": 0.45628895710605405,
      "learning_rate": 1.6342156277456279e-06,
      "loss": 0.0053,
      "step": 12832
    },
    {
      "epoch": 8.210492642354447,
      "grad_norm": 0.3376615341510218,
      "learning_rate": 1.63308058367058e-06,
      "loss": 0.0052,
      "step": 12833
    },
    {
      "epoch": 8.21113243761996,
      "grad_norm": 0.5021479041981801,
      "learning_rate": 1.6319458988561655e-06,
      "loss": 0.007,
      "step": 12834
    },
    {
      "epoch": 8.211772232885476,
      "grad_norm": 0.3814802377708425,
      "learning_rate": 1.630811573351111e-06,
      "loss": 0.0046,
      "step": 12835
    },
    {
      "epoch": 8.212412028150991,
      "grad_norm": 0.373238214887286,
      "learning_rate": 1.62967760720412e-06,
      "loss": 0.0054,
      "step": 12836
    },
    {
      "epoch": 8.213051823416507,
      "grad_norm": 0.3734669135422957,
      "learning_rate": 1.6285440004638809e-06,
      "loss": 0.0046,
      "step": 12837
    },
    {
      "epoch": 8.213691618682022,
      "grad_norm": 0.4309023685234012,
      "learning_rate": 1.6274107531790727e-06,
      "loss": 0.0057,
      "step": 12838
    },
    {
      "epoch": 8.214331413947537,
      "grad_norm": 0.4667567431043938,
      "learning_rate": 1.6262778653983525e-06,
      "loss": 0.0057,
      "step": 12839
    },
    {
      "epoch": 8.214971209213052,
      "grad_norm": 0.3683686649616474,
      "learning_rate": 1.6251453371703662e-06,
      "loss": 0.0052,
      "step": 12840
    },
    {
      "epoch": 8.215611004478568,
      "grad_norm": 0.40486598435068494,
      "learning_rate": 1.624013168543741e-06,
      "loss": 0.0053,
      "step": 12841
    },
    {
      "epoch": 8.216250799744081,
      "grad_norm": 0.3630133252427087,
      "learning_rate": 1.62288135956709e-06,
      "loss": 0.0042,
      "step": 12842
    },
    {
      "epoch": 8.216890595009597,
      "grad_norm": 0.44060653335283023,
      "learning_rate": 1.6217499102890122e-06,
      "loss": 0.0054,
      "step": 12843
    },
    {
      "epoch": 8.217530390275112,
      "grad_norm": 0.3563430060808915,
      "learning_rate": 1.6206188207580876e-06,
      "loss": 0.0048,
      "step": 12844
    },
    {
      "epoch": 8.218170185540627,
      "grad_norm": 0.3978855272692393,
      "learning_rate": 1.619488091022886e-06,
      "loss": 0.006,
      "step": 12845
    },
    {
      "epoch": 8.218809980806142,
      "grad_norm": 0.4220200528942508,
      "learning_rate": 1.6183577211319602e-06,
      "loss": 0.0057,
      "step": 12846
    },
    {
      "epoch": 8.219449776071658,
      "grad_norm": 0.3727025299750756,
      "learning_rate": 1.6172277111338397e-06,
      "loss": 0.0058,
      "step": 12847
    },
    {
      "epoch": 8.220089571337173,
      "grad_norm": 0.4271603582949197,
      "learning_rate": 1.6160980610770526e-06,
      "loss": 0.0042,
      "step": 12848
    },
    {
      "epoch": 8.220729366602686,
      "grad_norm": 0.4073779912830996,
      "learning_rate": 1.6149687710101014e-06,
      "loss": 0.0048,
      "step": 12849
    },
    {
      "epoch": 8.221369161868202,
      "grad_norm": 0.45943338383986676,
      "learning_rate": 1.6138398409814725e-06,
      "loss": 0.006,
      "step": 12850
    },
    {
      "epoch": 8.222008957133717,
      "grad_norm": 0.42565389225075656,
      "learning_rate": 1.6127112710396453e-06,
      "loss": 0.0055,
      "step": 12851
    },
    {
      "epoch": 8.222648752399232,
      "grad_norm": 0.4840741115571528,
      "learning_rate": 1.6115830612330764e-06,
      "loss": 0.0066,
      "step": 12852
    },
    {
      "epoch": 8.223288547664747,
      "grad_norm": 0.40889262120343844,
      "learning_rate": 1.610455211610209e-06,
      "loss": 0.0052,
      "step": 12853
    },
    {
      "epoch": 8.223928342930263,
      "grad_norm": 0.37306908462866367,
      "learning_rate": 1.6093277222194681e-06,
      "loss": 0.0056,
      "step": 12854
    },
    {
      "epoch": 8.224568138195778,
      "grad_norm": 0.4697595246463783,
      "learning_rate": 1.6082005931092726e-06,
      "loss": 0.0057,
      "step": 12855
    },
    {
      "epoch": 8.225207933461292,
      "grad_norm": 0.379841829312815,
      "learning_rate": 1.607073824328015e-06,
      "loss": 0.0044,
      "step": 12856
    },
    {
      "epoch": 8.225847728726807,
      "grad_norm": 0.39904708249759396,
      "learning_rate": 1.6059474159240773e-06,
      "loss": 0.0055,
      "step": 12857
    },
    {
      "epoch": 8.226487523992322,
      "grad_norm": 0.40292498891849193,
      "learning_rate": 1.6048213679458258e-06,
      "loss": 0.0058,
      "step": 12858
    },
    {
      "epoch": 8.227127319257837,
      "grad_norm": 0.5893939159679606,
      "learning_rate": 1.6036956804416114e-06,
      "loss": 0.0091,
      "step": 12859
    },
    {
      "epoch": 8.227767114523353,
      "grad_norm": 0.39399912965486955,
      "learning_rate": 1.602570353459766e-06,
      "loss": 0.0061,
      "step": 12860
    },
    {
      "epoch": 8.228406909788868,
      "grad_norm": 0.39044339405713235,
      "learning_rate": 1.6014453870486136e-06,
      "loss": 0.0049,
      "step": 12861
    },
    {
      "epoch": 8.229046705054383,
      "grad_norm": 0.42055392968934197,
      "learning_rate": 1.6003207812564569e-06,
      "loss": 0.0061,
      "step": 12862
    },
    {
      "epoch": 8.229686500319898,
      "grad_norm": 0.38137209782234754,
      "learning_rate": 1.5991965361315842e-06,
      "loss": 0.0053,
      "step": 12863
    },
    {
      "epoch": 8.230326295585412,
      "grad_norm": 0.39239252185105217,
      "learning_rate": 1.5980726517222656e-06,
      "loss": 0.0056,
      "step": 12864
    },
    {
      "epoch": 8.230966090850927,
      "grad_norm": 0.40509113152561793,
      "learning_rate": 1.5969491280767634e-06,
      "loss": 0.0059,
      "step": 12865
    },
    {
      "epoch": 8.231605886116443,
      "grad_norm": 0.5485673154462747,
      "learning_rate": 1.5958259652433184e-06,
      "loss": 0.0087,
      "step": 12866
    },
    {
      "epoch": 8.232245681381958,
      "grad_norm": 0.3639297522322495,
      "learning_rate": 1.594703163270155e-06,
      "loss": 0.0051,
      "step": 12867
    },
    {
      "epoch": 8.232885476647473,
      "grad_norm": 0.46858837078671695,
      "learning_rate": 1.5935807222054866e-06,
      "loss": 0.0066,
      "step": 12868
    },
    {
      "epoch": 8.233525271912988,
      "grad_norm": 0.4976856273484638,
      "learning_rate": 1.5924586420975097e-06,
      "loss": 0.0098,
      "step": 12869
    },
    {
      "epoch": 8.234165067178504,
      "grad_norm": 0.434278406627835,
      "learning_rate": 1.591336922994402e-06,
      "loss": 0.0063,
      "step": 12870
    },
    {
      "epoch": 8.234804862444017,
      "grad_norm": 0.40761334747545697,
      "learning_rate": 1.5902155649443284e-06,
      "loss": 0.0052,
      "step": 12871
    },
    {
      "epoch": 8.235444657709532,
      "grad_norm": 0.5497863659096314,
      "learning_rate": 1.5890945679954395e-06,
      "loss": 0.0085,
      "step": 12872
    },
    {
      "epoch": 8.236084452975048,
      "grad_norm": 0.41750482266699307,
      "learning_rate": 1.5879739321958677e-06,
      "loss": 0.0053,
      "step": 12873
    },
    {
      "epoch": 8.236724248240563,
      "grad_norm": 0.3945236822239659,
      "learning_rate": 1.5868536575937287e-06,
      "loss": 0.0053,
      "step": 12874
    },
    {
      "epoch": 8.237364043506078,
      "grad_norm": 0.5218357588502723,
      "learning_rate": 1.5857337442371312e-06,
      "loss": 0.0073,
      "step": 12875
    },
    {
      "epoch": 8.238003838771593,
      "grad_norm": 0.40441022306719004,
      "learning_rate": 1.5846141921741587e-06,
      "loss": 0.0053,
      "step": 12876
    },
    {
      "epoch": 8.238643634037109,
      "grad_norm": 0.4485853523316625,
      "learning_rate": 1.5834950014528794e-06,
      "loss": 0.0072,
      "step": 12877
    },
    {
      "epoch": 8.239283429302624,
      "grad_norm": 0.40577471803862275,
      "learning_rate": 1.582376172121357e-06,
      "loss": 0.006,
      "step": 12878
    },
    {
      "epoch": 8.239923224568138,
      "grad_norm": 0.4397787861540983,
      "learning_rate": 1.5812577042276267e-06,
      "loss": 0.0055,
      "step": 12879
    },
    {
      "epoch": 8.240563019833653,
      "grad_norm": 0.35214302757695126,
      "learning_rate": 1.5801395978197133e-06,
      "loss": 0.005,
      "step": 12880
    },
    {
      "epoch": 8.241202815099168,
      "grad_norm": 0.431129193958917,
      "learning_rate": 1.5790218529456292e-06,
      "loss": 0.0056,
      "step": 12881
    },
    {
      "epoch": 8.241842610364683,
      "grad_norm": 0.3402562855845315,
      "learning_rate": 1.5779044696533673e-06,
      "loss": 0.0047,
      "step": 12882
    },
    {
      "epoch": 8.242482405630199,
      "grad_norm": 0.35157361453028063,
      "learning_rate": 1.5767874479909061e-06,
      "loss": 0.005,
      "step": 12883
    },
    {
      "epoch": 8.243122200895714,
      "grad_norm": 0.47870060106417656,
      "learning_rate": 1.5756707880062061e-06,
      "loss": 0.0065,
      "step": 12884
    },
    {
      "epoch": 8.24376199616123,
      "grad_norm": 0.42836734969810564,
      "learning_rate": 1.5745544897472176e-06,
      "loss": 0.0067,
      "step": 12885
    },
    {
      "epoch": 8.244401791426743,
      "grad_norm": 0.3787249740325688,
      "learning_rate": 1.5734385532618734e-06,
      "loss": 0.0052,
      "step": 12886
    },
    {
      "epoch": 8.245041586692258,
      "grad_norm": 0.44654476963782497,
      "learning_rate": 1.5723229785980865e-06,
      "loss": 0.0055,
      "step": 12887
    },
    {
      "epoch": 8.245681381957773,
      "grad_norm": 0.4934409868769016,
      "learning_rate": 1.5712077658037606e-06,
      "loss": 0.0076,
      "step": 12888
    },
    {
      "epoch": 8.246321177223289,
      "grad_norm": 0.506823587439112,
      "learning_rate": 1.5700929149267785e-06,
      "loss": 0.0068,
      "step": 12889
    },
    {
      "epoch": 8.246960972488804,
      "grad_norm": 0.3307341050431741,
      "learning_rate": 1.568978426015012e-06,
      "loss": 0.0045,
      "step": 12890
    },
    {
      "epoch": 8.247600767754319,
      "grad_norm": 0.41795850580291527,
      "learning_rate": 1.5678642991163106e-06,
      "loss": 0.0051,
      "step": 12891
    },
    {
      "epoch": 8.248240563019834,
      "grad_norm": 0.3710163908551262,
      "learning_rate": 1.5667505342785206e-06,
      "loss": 0.0053,
      "step": 12892
    },
    {
      "epoch": 8.248880358285348,
      "grad_norm": 0.36676393773051824,
      "learning_rate": 1.5656371315494601e-06,
      "loss": 0.0049,
      "step": 12893
    },
    {
      "epoch": 8.249520153550863,
      "grad_norm": 0.5141808893414614,
      "learning_rate": 1.5645240909769343e-06,
      "loss": 0.008,
      "step": 12894
    },
    {
      "epoch": 8.250159948816378,
      "grad_norm": 0.5470707910655844,
      "learning_rate": 1.563411412608743e-06,
      "loss": 0.0083,
      "step": 12895
    },
    {
      "epoch": 8.250799744081894,
      "grad_norm": 0.4833982316056128,
      "learning_rate": 1.5622990964926566e-06,
      "loss": 0.0075,
      "step": 12896
    },
    {
      "epoch": 8.251439539347409,
      "grad_norm": 0.42874373515048,
      "learning_rate": 1.5611871426764358e-06,
      "loss": 0.0066,
      "step": 12897
    },
    {
      "epoch": 8.252079334612924,
      "grad_norm": 0.4218785477785462,
      "learning_rate": 1.560075551207829e-06,
      "loss": 0.0062,
      "step": 12898
    },
    {
      "epoch": 8.25271912987844,
      "grad_norm": 0.41336297638435193,
      "learning_rate": 1.5589643221345651e-06,
      "loss": 0.0064,
      "step": 12899
    },
    {
      "epoch": 8.253358925143955,
      "grad_norm": 0.532324863396059,
      "learning_rate": 1.557853455504358e-06,
      "loss": 0.0082,
      "step": 12900
    },
    {
      "epoch": 8.253998720409468,
      "grad_norm": 0.4457197349814687,
      "learning_rate": 1.5567429513649057e-06,
      "loss": 0.0064,
      "step": 12901
    },
    {
      "epoch": 8.254638515674984,
      "grad_norm": 0.3119485950093008,
      "learning_rate": 1.555632809763893e-06,
      "loss": 0.0041,
      "step": 12902
    },
    {
      "epoch": 8.255278310940499,
      "grad_norm": 0.4152991982653571,
      "learning_rate": 1.5545230307489844e-06,
      "loss": 0.0051,
      "step": 12903
    },
    {
      "epoch": 8.255918106206014,
      "grad_norm": 0.39268571579761263,
      "learning_rate": 1.5534136143678324e-06,
      "loss": 0.0059,
      "step": 12904
    },
    {
      "epoch": 8.25655790147153,
      "grad_norm": 0.32890257064963735,
      "learning_rate": 1.5523045606680765e-06,
      "loss": 0.0046,
      "step": 12905
    },
    {
      "epoch": 8.257197696737045,
      "grad_norm": 0.37031682031792645,
      "learning_rate": 1.5511958696973351e-06,
      "loss": 0.0052,
      "step": 12906
    },
    {
      "epoch": 8.25783749200256,
      "grad_norm": 0.45817423010837793,
      "learning_rate": 1.5500875415032113e-06,
      "loss": 0.0072,
      "step": 12907
    },
    {
      "epoch": 8.258477287268073,
      "grad_norm": 0.48715729413168996,
      "learning_rate": 1.5489795761332992e-06,
      "loss": 0.0063,
      "step": 12908
    },
    {
      "epoch": 8.259117082533589,
      "grad_norm": 0.49892822447698054,
      "learning_rate": 1.5478719736351721e-06,
      "loss": 0.0061,
      "step": 12909
    },
    {
      "epoch": 8.259756877799104,
      "grad_norm": 0.4009466889678047,
      "learning_rate": 1.5467647340563873e-06,
      "loss": 0.0057,
      "step": 12910
    },
    {
      "epoch": 8.26039667306462,
      "grad_norm": 0.3742150388734514,
      "learning_rate": 1.5456578574444848e-06,
      "loss": 0.005,
      "step": 12911
    },
    {
      "epoch": 8.261036468330134,
      "grad_norm": 0.40618814397514774,
      "learning_rate": 1.5445513438469972e-06,
      "loss": 0.0048,
      "step": 12912
    },
    {
      "epoch": 8.26167626359565,
      "grad_norm": 0.48079833373594855,
      "learning_rate": 1.5434451933114335e-06,
      "loss": 0.0071,
      "step": 12913
    },
    {
      "epoch": 8.262316058861165,
      "grad_norm": 0.3890945712487516,
      "learning_rate": 1.542339405885288e-06,
      "loss": 0.0054,
      "step": 12914
    },
    {
      "epoch": 8.26295585412668,
      "grad_norm": 0.4574339931688127,
      "learning_rate": 1.5412339816160459e-06,
      "loss": 0.0073,
      "step": 12915
    },
    {
      "epoch": 8.263595649392194,
      "grad_norm": 0.36674056907977126,
      "learning_rate": 1.5401289205511715e-06,
      "loss": 0.0053,
      "step": 12916
    },
    {
      "epoch": 8.264235444657709,
      "grad_norm": 0.40971653703077354,
      "learning_rate": 1.539024222738108e-06,
      "loss": 0.0055,
      "step": 12917
    },
    {
      "epoch": 8.264875239923224,
      "grad_norm": 0.4384766009679335,
      "learning_rate": 1.5379198882242963e-06,
      "loss": 0.0063,
      "step": 12918
    },
    {
      "epoch": 8.26551503518874,
      "grad_norm": 0.4331812340072143,
      "learning_rate": 1.5368159170571517e-06,
      "loss": 0.0056,
      "step": 12919
    },
    {
      "epoch": 8.266154830454255,
      "grad_norm": 0.4328327090232242,
      "learning_rate": 1.5357123092840754e-06,
      "loss": 0.0056,
      "step": 12920
    },
    {
      "epoch": 8.26679462571977,
      "grad_norm": 0.47587715415949916,
      "learning_rate": 1.5346090649524548e-06,
      "loss": 0.0063,
      "step": 12921
    },
    {
      "epoch": 8.267434420985285,
      "grad_norm": 0.45217259023418127,
      "learning_rate": 1.5335061841096643e-06,
      "loss": 0.0055,
      "step": 12922
    },
    {
      "epoch": 8.268074216250799,
      "grad_norm": 0.4118829436496806,
      "learning_rate": 1.5324036668030572e-06,
      "loss": 0.0059,
      "step": 12923
    },
    {
      "epoch": 8.268714011516314,
      "grad_norm": 0.504122790615225,
      "learning_rate": 1.5313015130799713e-06,
      "loss": 0.0068,
      "step": 12924
    },
    {
      "epoch": 8.26935380678183,
      "grad_norm": 0.515589180042899,
      "learning_rate": 1.5301997229877374e-06,
      "loss": 0.0071,
      "step": 12925
    },
    {
      "epoch": 8.269993602047345,
      "grad_norm": 0.4542773868698005,
      "learning_rate": 1.5290982965736601e-06,
      "loss": 0.006,
      "step": 12926
    },
    {
      "epoch": 8.27063339731286,
      "grad_norm": 0.4353950385567067,
      "learning_rate": 1.5279972338850303e-06,
      "loss": 0.0045,
      "step": 12927
    },
    {
      "epoch": 8.271273192578375,
      "grad_norm": 0.4539806931134819,
      "learning_rate": 1.5268965349691322e-06,
      "loss": 0.0058,
      "step": 12928
    },
    {
      "epoch": 8.27191298784389,
      "grad_norm": 0.45009501602443663,
      "learning_rate": 1.525796199873223e-06,
      "loss": 0.0066,
      "step": 12929
    },
    {
      "epoch": 8.272552783109404,
      "grad_norm": 0.37626718147250815,
      "learning_rate": 1.5246962286445521e-06,
      "loss": 0.0056,
      "step": 12930
    },
    {
      "epoch": 8.27319257837492,
      "grad_norm": 0.48897159768305903,
      "learning_rate": 1.5235966213303478e-06,
      "loss": 0.0074,
      "step": 12931
    },
    {
      "epoch": 8.273832373640435,
      "grad_norm": 0.4089917287745855,
      "learning_rate": 1.5224973779778262e-06,
      "loss": 0.0055,
      "step": 12932
    },
    {
      "epoch": 8.27447216890595,
      "grad_norm": 0.5785694979450474,
      "learning_rate": 1.521398498634188e-06,
      "loss": 0.0077,
      "step": 12933
    },
    {
      "epoch": 8.275111964171465,
      "grad_norm": 0.33295339698991505,
      "learning_rate": 1.5202999833466126e-06,
      "loss": 0.0043,
      "step": 12934
    },
    {
      "epoch": 8.27575175943698,
      "grad_norm": 0.4870221174739288,
      "learning_rate": 1.5192018321622736e-06,
      "loss": 0.0068,
      "step": 12935
    },
    {
      "epoch": 8.276391554702496,
      "grad_norm": 0.35215275078678365,
      "learning_rate": 1.5181040451283235e-06,
      "loss": 0.0051,
      "step": 12936
    },
    {
      "epoch": 8.277031349968011,
      "grad_norm": 0.37238535349044277,
      "learning_rate": 1.5170066222918934e-06,
      "loss": 0.0049,
      "step": 12937
    },
    {
      "epoch": 8.277671145233525,
      "grad_norm": 0.4744047613207955,
      "learning_rate": 1.515909563700112e-06,
      "loss": 0.0067,
      "step": 12938
    },
    {
      "epoch": 8.27831094049904,
      "grad_norm": 0.33291413116222807,
      "learning_rate": 1.514812869400082e-06,
      "loss": 0.004,
      "step": 12939
    },
    {
      "epoch": 8.278950735764555,
      "grad_norm": 0.5119781993216063,
      "learning_rate": 1.513716539438893e-06,
      "loss": 0.0067,
      "step": 12940
    },
    {
      "epoch": 8.27959053103007,
      "grad_norm": 0.4400274625916037,
      "learning_rate": 1.512620573863618e-06,
      "loss": 0.0051,
      "step": 12941
    },
    {
      "epoch": 8.280230326295586,
      "grad_norm": 0.5277621608096775,
      "learning_rate": 1.51152497272132e-06,
      "loss": 0.0043,
      "step": 12942
    },
    {
      "epoch": 8.280870121561101,
      "grad_norm": 0.4058201284422392,
      "learning_rate": 1.5104297360590404e-06,
      "loss": 0.0051,
      "step": 12943
    },
    {
      "epoch": 8.281509916826616,
      "grad_norm": 0.4310610881627397,
      "learning_rate": 1.5093348639238025e-06,
      "loss": 0.0066,
      "step": 12944
    },
    {
      "epoch": 8.28214971209213,
      "grad_norm": 0.4401130749450274,
      "learning_rate": 1.508240356362627e-06,
      "loss": 0.005,
      "step": 12945
    },
    {
      "epoch": 8.282789507357645,
      "grad_norm": 0.4030240109968164,
      "learning_rate": 1.5071462134225034e-06,
      "loss": 0.0049,
      "step": 12946
    },
    {
      "epoch": 8.28342930262316,
      "grad_norm": 0.38110726949977536,
      "learning_rate": 1.5060524351504102e-06,
      "loss": 0.0052,
      "step": 12947
    },
    {
      "epoch": 8.284069097888676,
      "grad_norm": 0.3804775964897852,
      "learning_rate": 1.5049590215933185e-06,
      "loss": 0.0057,
      "step": 12948
    },
    {
      "epoch": 8.28470889315419,
      "grad_norm": 0.4553714463368746,
      "learning_rate": 1.5038659727981752e-06,
      "loss": 0.0052,
      "step": 12949
    },
    {
      "epoch": 8.285348688419706,
      "grad_norm": 0.5017401952767522,
      "learning_rate": 1.502773288811914e-06,
      "loss": 0.0075,
      "step": 12950
    },
    {
      "epoch": 8.285988483685221,
      "grad_norm": 0.42548179315845663,
      "learning_rate": 1.5016809696814493e-06,
      "loss": 0.0059,
      "step": 12951
    },
    {
      "epoch": 8.286628278950737,
      "grad_norm": 0.3647836839112077,
      "learning_rate": 1.5005890154536894e-06,
      "loss": 0.0058,
      "step": 12952
    },
    {
      "epoch": 8.28726807421625,
      "grad_norm": 0.4366679157941187,
      "learning_rate": 1.499497426175518e-06,
      "loss": 0.0069,
      "step": 12953
    },
    {
      "epoch": 8.287907869481765,
      "grad_norm": 0.4259884948801391,
      "learning_rate": 1.4984062018938028e-06,
      "loss": 0.0051,
      "step": 12954
    },
    {
      "epoch": 8.28854766474728,
      "grad_norm": 0.4178623166946619,
      "learning_rate": 1.4973153426554042e-06,
      "loss": 0.0051,
      "step": 12955
    },
    {
      "epoch": 8.289187460012796,
      "grad_norm": 0.45660424899032526,
      "learning_rate": 1.4962248485071607e-06,
      "loss": 0.0072,
      "step": 12956
    },
    {
      "epoch": 8.289827255278311,
      "grad_norm": 0.4017197141699849,
      "learning_rate": 1.4951347194958933e-06,
      "loss": 0.0052,
      "step": 12957
    },
    {
      "epoch": 8.290467050543826,
      "grad_norm": 0.4747505680322071,
      "learning_rate": 1.4940449556684133e-06,
      "loss": 0.0063,
      "step": 12958
    },
    {
      "epoch": 8.291106845809342,
      "grad_norm": 0.36205493889987533,
      "learning_rate": 1.4929555570715116e-06,
      "loss": 0.0048,
      "step": 12959
    },
    {
      "epoch": 8.291746641074855,
      "grad_norm": 0.39705763490610035,
      "learning_rate": 1.4918665237519659e-06,
      "loss": 0.0063,
      "step": 12960
    },
    {
      "epoch": 8.29238643634037,
      "grad_norm": 0.4106123871137704,
      "learning_rate": 1.4907778557565379e-06,
      "loss": 0.0039,
      "step": 12961
    },
    {
      "epoch": 8.293026231605886,
      "grad_norm": 0.3968631774058702,
      "learning_rate": 1.4896895531319699e-06,
      "loss": 0.0053,
      "step": 12962
    },
    {
      "epoch": 8.293666026871401,
      "grad_norm": 0.42323331147320165,
      "learning_rate": 1.488601615924996e-06,
      "loss": 0.0061,
      "step": 12963
    },
    {
      "epoch": 8.294305822136916,
      "grad_norm": 0.36102372714610054,
      "learning_rate": 1.4875140441823243e-06,
      "loss": 0.0056,
      "step": 12964
    },
    {
      "epoch": 8.294945617402432,
      "grad_norm": 0.4956084242868606,
      "learning_rate": 1.4864268379506597e-06,
      "loss": 0.0064,
      "step": 12965
    },
    {
      "epoch": 8.295585412667947,
      "grad_norm": 0.54420747346082,
      "learning_rate": 1.4853399972766824e-06,
      "loss": 0.0065,
      "step": 12966
    },
    {
      "epoch": 8.29622520793346,
      "grad_norm": 0.5680980142364618,
      "learning_rate": 1.4842535222070598e-06,
      "loss": 0.0087,
      "step": 12967
    },
    {
      "epoch": 8.296865003198976,
      "grad_norm": 0.42704782873970026,
      "learning_rate": 1.4831674127884399e-06,
      "loss": 0.0084,
      "step": 12968
    },
    {
      "epoch": 8.297504798464491,
      "grad_norm": 0.4310794844313818,
      "learning_rate": 1.4820816690674622e-06,
      "loss": 0.0052,
      "step": 12969
    },
    {
      "epoch": 8.298144593730006,
      "grad_norm": 0.4919916509168484,
      "learning_rate": 1.4809962910907472e-06,
      "loss": 0.0068,
      "step": 12970
    },
    {
      "epoch": 8.298784388995522,
      "grad_norm": 0.4239374516284166,
      "learning_rate": 1.4799112789048942e-06,
      "loss": 0.0059,
      "step": 12971
    },
    {
      "epoch": 8.299424184261037,
      "grad_norm": 0.6627522455741004,
      "learning_rate": 1.478826632556497e-06,
      "loss": 0.0061,
      "step": 12972
    },
    {
      "epoch": 8.300063979526552,
      "grad_norm": 0.38385256017975794,
      "learning_rate": 1.4777423520921264e-06,
      "loss": 0.0057,
      "step": 12973
    },
    {
      "epoch": 8.300703774792067,
      "grad_norm": 0.36666969892935514,
      "learning_rate": 1.4766584375583382e-06,
      "loss": 0.0048,
      "step": 12974
    },
    {
      "epoch": 8.30134357005758,
      "grad_norm": 0.3407655981532601,
      "learning_rate": 1.4755748890016762e-06,
      "loss": 0.0046,
      "step": 12975
    },
    {
      "epoch": 8.301983365323096,
      "grad_norm": 0.3475264942075852,
      "learning_rate": 1.4744917064686648e-06,
      "loss": 0.0045,
      "step": 12976
    },
    {
      "epoch": 8.302623160588611,
      "grad_norm": 0.42876312023143587,
      "learning_rate": 1.473408890005813e-06,
      "loss": 0.0053,
      "step": 12977
    },
    {
      "epoch": 8.303262955854127,
      "grad_norm": 0.5345372595233563,
      "learning_rate": 1.4723264396596137e-06,
      "loss": 0.0062,
      "step": 12978
    },
    {
      "epoch": 8.303902751119642,
      "grad_norm": 0.4452744298960779,
      "learning_rate": 1.47124435547655e-06,
      "loss": 0.0059,
      "step": 12979
    },
    {
      "epoch": 8.304542546385157,
      "grad_norm": 0.4772762653949263,
      "learning_rate": 1.4701626375030819e-06,
      "loss": 0.0055,
      "step": 12980
    },
    {
      "epoch": 8.305182341650672,
      "grad_norm": 0.5104790642015493,
      "learning_rate": 1.469081285785655e-06,
      "loss": 0.0058,
      "step": 12981
    },
    {
      "epoch": 8.305822136916186,
      "grad_norm": 0.41447002432773017,
      "learning_rate": 1.4680003003707044e-06,
      "loss": 0.005,
      "step": 12982
    },
    {
      "epoch": 8.306461932181701,
      "grad_norm": 0.38795326472375075,
      "learning_rate": 1.4669196813046439e-06,
      "loss": 0.005,
      "step": 12983
    },
    {
      "epoch": 8.307101727447217,
      "grad_norm": 0.38112617417928357,
      "learning_rate": 1.4658394286338706e-06,
      "loss": 0.0048,
      "step": 12984
    },
    {
      "epoch": 8.307741522712732,
      "grad_norm": 0.45335969707827056,
      "learning_rate": 1.4647595424047735e-06,
      "loss": 0.0067,
      "step": 12985
    },
    {
      "epoch": 8.308381317978247,
      "grad_norm": 0.5314094872433688,
      "learning_rate": 1.4636800226637183e-06,
      "loss": 0.0072,
      "step": 12986
    },
    {
      "epoch": 8.309021113243762,
      "grad_norm": 0.5368263864859502,
      "learning_rate": 1.4626008694570582e-06,
      "loss": 0.0071,
      "step": 12987
    },
    {
      "epoch": 8.309660908509278,
      "grad_norm": 0.3894790292568693,
      "learning_rate": 1.4615220828311284e-06,
      "loss": 0.0058,
      "step": 12988
    },
    {
      "epoch": 8.310300703774793,
      "grad_norm": 0.44852187859259357,
      "learning_rate": 1.4604436628322527e-06,
      "loss": 0.0058,
      "step": 12989
    },
    {
      "epoch": 8.310940499040306,
      "grad_norm": 0.30207180992687593,
      "learning_rate": 1.4593656095067354e-06,
      "loss": 0.0036,
      "step": 12990
    },
    {
      "epoch": 8.311580294305822,
      "grad_norm": 0.41547217615003945,
      "learning_rate": 1.4582879229008673e-06,
      "loss": 0.0047,
      "step": 12991
    },
    {
      "epoch": 8.312220089571337,
      "grad_norm": 0.37088547998837323,
      "learning_rate": 1.4572106030609201e-06,
      "loss": 0.0054,
      "step": 12992
    },
    {
      "epoch": 8.312859884836852,
      "grad_norm": 0.5139001080876433,
      "learning_rate": 1.4561336500331535e-06,
      "loss": 0.0058,
      "step": 12993
    },
    {
      "epoch": 8.313499680102368,
      "grad_norm": 0.3746090979031293,
      "learning_rate": 1.4550570638638072e-06,
      "loss": 0.0053,
      "step": 12994
    },
    {
      "epoch": 8.314139475367883,
      "grad_norm": 0.3654614494417123,
      "learning_rate": 1.4539808445991133e-06,
      "loss": 0.0046,
      "step": 12995
    },
    {
      "epoch": 8.314779270633398,
      "grad_norm": 0.4177036211769618,
      "learning_rate": 1.45290499228528e-06,
      "loss": 0.0057,
      "step": 12996
    },
    {
      "epoch": 8.315419065898912,
      "grad_norm": 0.4245838512118672,
      "learning_rate": 1.4518295069685008e-06,
      "loss": 0.0053,
      "step": 12997
    },
    {
      "epoch": 8.316058861164427,
      "grad_norm": 0.43572208645622545,
      "learning_rate": 1.4507543886949548e-06,
      "loss": 0.0061,
      "step": 12998
    },
    {
      "epoch": 8.316698656429942,
      "grad_norm": 0.40067559344923226,
      "learning_rate": 1.4496796375108102e-06,
      "loss": 0.0046,
      "step": 12999
    },
    {
      "epoch": 8.317338451695457,
      "grad_norm": 0.3940042900268918,
      "learning_rate": 1.448605253462212e-06,
      "loss": 0.0054,
      "step": 13000
    },
    {
      "epoch": 8.317978246960973,
      "grad_norm": 0.3742493406743258,
      "learning_rate": 1.4475312365952898e-06,
      "loss": 0.0057,
      "step": 13001
    },
    {
      "epoch": 8.318618042226488,
      "grad_norm": 0.45871345636586586,
      "learning_rate": 1.4464575869561659e-06,
      "loss": 0.0064,
      "step": 13002
    },
    {
      "epoch": 8.319257837492003,
      "grad_norm": 0.35458920837609353,
      "learning_rate": 1.4453843045909365e-06,
      "loss": 0.0038,
      "step": 13003
    },
    {
      "epoch": 8.319897632757517,
      "grad_norm": 0.7300059441290231,
      "learning_rate": 1.4443113895456885e-06,
      "loss": 0.0079,
      "step": 13004
    },
    {
      "epoch": 8.320537428023032,
      "grad_norm": 0.3428806161831243,
      "learning_rate": 1.4432388418664888e-06,
      "loss": 0.0049,
      "step": 13005
    },
    {
      "epoch": 8.321177223288547,
      "grad_norm": 0.41758648821313493,
      "learning_rate": 1.4421666615993933e-06,
      "loss": 0.0062,
      "step": 13006
    },
    {
      "epoch": 8.321817018554063,
      "grad_norm": 0.4305990379714516,
      "learning_rate": 1.4410948487904374e-06,
      "loss": 0.0057,
      "step": 13007
    },
    {
      "epoch": 8.322456813819578,
      "grad_norm": 0.5001716984937985,
      "learning_rate": 1.440023403485642e-06,
      "loss": 0.0063,
      "step": 13008
    },
    {
      "epoch": 8.323096609085093,
      "grad_norm": 0.46009524713444566,
      "learning_rate": 1.4389523257310156e-06,
      "loss": 0.0064,
      "step": 13009
    },
    {
      "epoch": 8.323736404350608,
      "grad_norm": 0.3740710777887433,
      "learning_rate": 1.4378816155725483e-06,
      "loss": 0.0053,
      "step": 13010
    },
    {
      "epoch": 8.324376199616124,
      "grad_norm": 0.4247813513223,
      "learning_rate": 1.436811273056211e-06,
      "loss": 0.0065,
      "step": 13011
    },
    {
      "epoch": 8.325015994881637,
      "grad_norm": 0.4546657085072686,
      "learning_rate": 1.4357412982279672e-06,
      "loss": 0.0064,
      "step": 13012
    },
    {
      "epoch": 8.325655790147152,
      "grad_norm": 0.40219985646355183,
      "learning_rate": 1.4346716911337576e-06,
      "loss": 0.0054,
      "step": 13013
    },
    {
      "epoch": 8.326295585412668,
      "grad_norm": 0.4597046371384626,
      "learning_rate": 1.433602451819507e-06,
      "loss": 0.0059,
      "step": 13014
    },
    {
      "epoch": 8.326935380678183,
      "grad_norm": 0.4158275977319702,
      "learning_rate": 1.4325335803311313e-06,
      "loss": 0.0055,
      "step": 13015
    },
    {
      "epoch": 8.327575175943698,
      "grad_norm": 0.4434224145990274,
      "learning_rate": 1.4314650767145222e-06,
      "loss": 0.0064,
      "step": 13016
    },
    {
      "epoch": 8.328214971209214,
      "grad_norm": 0.39706674507301015,
      "learning_rate": 1.4303969410155616e-06,
      "loss": 0.0056,
      "step": 13017
    },
    {
      "epoch": 8.328854766474729,
      "grad_norm": 0.4005373768494253,
      "learning_rate": 1.4293291732801084e-06,
      "loss": 0.005,
      "step": 13018
    },
    {
      "epoch": 8.329494561740242,
      "grad_norm": 0.3759698788390848,
      "learning_rate": 1.4282617735540183e-06,
      "loss": 0.0046,
      "step": 13019
    },
    {
      "epoch": 8.330134357005758,
      "grad_norm": 0.41153811437966037,
      "learning_rate": 1.4271947418831188e-06,
      "loss": 0.0046,
      "step": 13020
    },
    {
      "epoch": 8.330774152271273,
      "grad_norm": 0.44763939261279606,
      "learning_rate": 1.4261280783132281e-06,
      "loss": 0.0053,
      "step": 13021
    },
    {
      "epoch": 8.331413947536788,
      "grad_norm": 0.4347474860126545,
      "learning_rate": 1.4250617828901448e-06,
      "loss": 0.0055,
      "step": 13022
    },
    {
      "epoch": 8.332053742802303,
      "grad_norm": 0.44023979736156477,
      "learning_rate": 1.4239958556596556e-06,
      "loss": 0.006,
      "step": 13023
    },
    {
      "epoch": 8.332693538067819,
      "grad_norm": 0.4570704146981267,
      "learning_rate": 1.4229302966675263e-06,
      "loss": 0.0053,
      "step": 13024
    },
    {
      "epoch": 8.333333333333334,
      "grad_norm": 0.3747629827346997,
      "learning_rate": 1.4218651059595156e-06,
      "loss": 0.0047,
      "step": 13025
    },
    {
      "epoch": 8.33397312859885,
      "grad_norm": 0.5169010764922225,
      "learning_rate": 1.4208002835813573e-06,
      "loss": 0.0046,
      "step": 13026
    },
    {
      "epoch": 8.334612923864363,
      "grad_norm": 0.38642791076878935,
      "learning_rate": 1.4197358295787734e-06,
      "loss": 0.0053,
      "step": 13027
    },
    {
      "epoch": 8.335252719129878,
      "grad_norm": 0.5596069857466639,
      "learning_rate": 1.4186717439974685e-06,
      "loss": 0.0065,
      "step": 13028
    },
    {
      "epoch": 8.335892514395393,
      "grad_norm": 0.42383999954265184,
      "learning_rate": 1.4176080268831349e-06,
      "loss": 0.0057,
      "step": 13029
    },
    {
      "epoch": 8.336532309660909,
      "grad_norm": 0.43098560110190226,
      "learning_rate": 1.4165446782814462e-06,
      "loss": 0.0058,
      "step": 13030
    },
    {
      "epoch": 8.337172104926424,
      "grad_norm": 0.5159546639717933,
      "learning_rate": 1.4154816982380581e-06,
      "loss": 0.0084,
      "step": 13031
    },
    {
      "epoch": 8.337811900191939,
      "grad_norm": 0.4505796473965692,
      "learning_rate": 1.4144190867986184e-06,
      "loss": 0.0064,
      "step": 13032
    },
    {
      "epoch": 8.338451695457454,
      "grad_norm": 0.48128608935027406,
      "learning_rate": 1.4133568440087497e-06,
      "loss": 0.0053,
      "step": 13033
    },
    {
      "epoch": 8.339091490722968,
      "grad_norm": 0.4096208937759436,
      "learning_rate": 1.4122949699140632e-06,
      "loss": 0.0067,
      "step": 13034
    },
    {
      "epoch": 8.339731285988483,
      "grad_norm": 0.4334444322643912,
      "learning_rate": 1.4112334645601556e-06,
      "loss": 0.0063,
      "step": 13035
    },
    {
      "epoch": 8.340371081253998,
      "grad_norm": 0.4138813199448937,
      "learning_rate": 1.4101723279926038e-06,
      "loss": 0.0058,
      "step": 13036
    },
    {
      "epoch": 8.341010876519514,
      "grad_norm": 0.3330064893580779,
      "learning_rate": 1.4091115602569716e-06,
      "loss": 0.0045,
      "step": 13037
    },
    {
      "epoch": 8.341650671785029,
      "grad_norm": 0.4333519461379826,
      "learning_rate": 1.4080511613988057e-06,
      "loss": 0.006,
      "step": 13038
    },
    {
      "epoch": 8.342290467050544,
      "grad_norm": 0.4709235538628137,
      "learning_rate": 1.4069911314636408e-06,
      "loss": 0.007,
      "step": 13039
    },
    {
      "epoch": 8.34293026231606,
      "grad_norm": 0.38987830621657793,
      "learning_rate": 1.4059314704969917e-06,
      "loss": 0.0056,
      "step": 13040
    },
    {
      "epoch": 8.343570057581573,
      "grad_norm": 0.4461903883379822,
      "learning_rate": 1.4048721785443552e-06,
      "loss": 0.0062,
      "step": 13041
    },
    {
      "epoch": 8.344209852847088,
      "grad_norm": 0.43283761667720155,
      "learning_rate": 1.4038132556512196e-06,
      "loss": 0.0053,
      "step": 13042
    },
    {
      "epoch": 8.344849648112604,
      "grad_norm": 0.38506136194423174,
      "learning_rate": 1.4027547018630517e-06,
      "loss": 0.0048,
      "step": 13043
    },
    {
      "epoch": 8.345489443378119,
      "grad_norm": 0.30049235976513466,
      "learning_rate": 1.4016965172253027e-06,
      "loss": 0.0035,
      "step": 13044
    },
    {
      "epoch": 8.346129238643634,
      "grad_norm": 0.40531939095214603,
      "learning_rate": 1.4006387017834088e-06,
      "loss": 0.0064,
      "step": 13045
    },
    {
      "epoch": 8.34676903390915,
      "grad_norm": 0.31661053173445974,
      "learning_rate": 1.3995812555827936e-06,
      "loss": 0.0045,
      "step": 13046
    },
    {
      "epoch": 8.347408829174665,
      "grad_norm": 0.5011163614301608,
      "learning_rate": 1.398524178668861e-06,
      "loss": 0.0057,
      "step": 13047
    },
    {
      "epoch": 8.34804862444018,
      "grad_norm": 0.49172157589820303,
      "learning_rate": 1.3974674710869962e-06,
      "loss": 0.0057,
      "step": 13048
    },
    {
      "epoch": 8.348688419705693,
      "grad_norm": 0.36957943551575895,
      "learning_rate": 1.3964111328825791e-06,
      "loss": 0.0052,
      "step": 13049
    },
    {
      "epoch": 8.349328214971209,
      "grad_norm": 0.4485237222320445,
      "learning_rate": 1.395355164100962e-06,
      "loss": 0.0053,
      "step": 13050
    },
    {
      "epoch": 8.349968010236724,
      "grad_norm": 0.4908376609914067,
      "learning_rate": 1.394299564787488e-06,
      "loss": 0.0059,
      "step": 13051
    },
    {
      "epoch": 8.35060780550224,
      "grad_norm": 0.41839780234141055,
      "learning_rate": 1.3932443349874824e-06,
      "loss": 0.0057,
      "step": 13052
    },
    {
      "epoch": 8.351247600767755,
      "grad_norm": 0.4520609146843116,
      "learning_rate": 1.392189474746255e-06,
      "loss": 0.0077,
      "step": 13053
    },
    {
      "epoch": 8.35188739603327,
      "grad_norm": 0.4314800753438864,
      "learning_rate": 1.3911349841090993e-06,
      "loss": 0.004,
      "step": 13054
    },
    {
      "epoch": 8.352527191298785,
      "grad_norm": 0.40706029976846747,
      "learning_rate": 1.3900808631212915e-06,
      "loss": 0.0047,
      "step": 13055
    },
    {
      "epoch": 8.353166986564299,
      "grad_norm": 0.4591703680914273,
      "learning_rate": 1.3890271118280973e-06,
      "loss": 0.0061,
      "step": 13056
    },
    {
      "epoch": 8.353806781829814,
      "grad_norm": 0.45967966013713735,
      "learning_rate": 1.387973730274762e-06,
      "loss": 0.0066,
      "step": 13057
    },
    {
      "epoch": 8.35444657709533,
      "grad_norm": 0.41208054874319594,
      "learning_rate": 1.3869207185065126e-06,
      "loss": 0.0043,
      "step": 13058
    },
    {
      "epoch": 8.355086372360844,
      "grad_norm": 0.41839904281463686,
      "learning_rate": 1.3858680765685672e-06,
      "loss": 0.0056,
      "step": 13059
    },
    {
      "epoch": 8.35572616762636,
      "grad_norm": 0.44705972169042846,
      "learning_rate": 1.3848158045061244e-06,
      "loss": 0.0053,
      "step": 13060
    },
    {
      "epoch": 8.356365962891875,
      "grad_norm": 0.4156861779065228,
      "learning_rate": 1.3837639023643634e-06,
      "loss": 0.0052,
      "step": 13061
    },
    {
      "epoch": 8.35700575815739,
      "grad_norm": 0.41044775053814186,
      "learning_rate": 1.3827123701884548e-06,
      "loss": 0.0056,
      "step": 13062
    },
    {
      "epoch": 8.357645553422905,
      "grad_norm": 0.44404735164204934,
      "learning_rate": 1.3816612080235491e-06,
      "loss": 0.0056,
      "step": 13063
    },
    {
      "epoch": 8.358285348688419,
      "grad_norm": 0.4234569721696355,
      "learning_rate": 1.3806104159147792e-06,
      "loss": 0.0065,
      "step": 13064
    },
    {
      "epoch": 8.358925143953934,
      "grad_norm": 0.5111895416856025,
      "learning_rate": 1.3795599939072669e-06,
      "loss": 0.0072,
      "step": 13065
    },
    {
      "epoch": 8.35956493921945,
      "grad_norm": 0.4131516575417765,
      "learning_rate": 1.3785099420461123e-06,
      "loss": 0.0054,
      "step": 13066
    },
    {
      "epoch": 8.360204734484965,
      "grad_norm": 0.3736585265111846,
      "learning_rate": 1.3774602603764053e-06,
      "loss": 0.005,
      "step": 13067
    },
    {
      "epoch": 8.36084452975048,
      "grad_norm": 0.4406672937262677,
      "learning_rate": 1.376410948943213e-06,
      "loss": 0.0052,
      "step": 13068
    },
    {
      "epoch": 8.361484325015995,
      "grad_norm": 0.4096145997639424,
      "learning_rate": 1.3753620077915975e-06,
      "loss": 0.0046,
      "step": 13069
    },
    {
      "epoch": 8.36212412028151,
      "grad_norm": 0.45981847756971395,
      "learning_rate": 1.3743134369665956e-06,
      "loss": 0.0068,
      "step": 13070
    },
    {
      "epoch": 8.362763915547024,
      "grad_norm": 0.5298907414014571,
      "learning_rate": 1.3732652365132282e-06,
      "loss": 0.0063,
      "step": 13071
    },
    {
      "epoch": 8.36340371081254,
      "grad_norm": 0.3775545208995684,
      "learning_rate": 1.3722174064765081e-06,
      "loss": 0.0054,
      "step": 13072
    },
    {
      "epoch": 8.364043506078055,
      "grad_norm": 0.3703669388452463,
      "learning_rate": 1.371169946901425e-06,
      "loss": 0.0051,
      "step": 13073
    },
    {
      "epoch": 8.36468330134357,
      "grad_norm": 0.5189580384957834,
      "learning_rate": 1.370122857832955e-06,
      "loss": 0.0078,
      "step": 13074
    },
    {
      "epoch": 8.365323096609085,
      "grad_norm": 0.4967962809834867,
      "learning_rate": 1.3690761393160556e-06,
      "loss": 0.0064,
      "step": 13075
    },
    {
      "epoch": 8.3659628918746,
      "grad_norm": 0.3639803817079406,
      "learning_rate": 1.3680297913956752e-06,
      "loss": 0.0046,
      "step": 13076
    },
    {
      "epoch": 8.366602687140116,
      "grad_norm": 0.5018169735478083,
      "learning_rate": 1.3669838141167412e-06,
      "loss": 0.0065,
      "step": 13077
    },
    {
      "epoch": 8.36724248240563,
      "grad_norm": 0.43256014975237134,
      "learning_rate": 1.3659382075241634e-06,
      "loss": 0.0064,
      "step": 13078
    },
    {
      "epoch": 8.367882277671145,
      "grad_norm": 0.49449718255621794,
      "learning_rate": 1.3648929716628423e-06,
      "loss": 0.0074,
      "step": 13079
    },
    {
      "epoch": 8.36852207293666,
      "grad_norm": 0.382319533148516,
      "learning_rate": 1.3638481065776555e-06,
      "loss": 0.0053,
      "step": 13080
    },
    {
      "epoch": 8.369161868202175,
      "grad_norm": 0.43400524847114136,
      "learning_rate": 1.3628036123134692e-06,
      "loss": 0.0054,
      "step": 13081
    },
    {
      "epoch": 8.36980166346769,
      "grad_norm": 0.5276753149913918,
      "learning_rate": 1.3617594889151308e-06,
      "loss": 0.0066,
      "step": 13082
    },
    {
      "epoch": 8.370441458733206,
      "grad_norm": 0.45129755815265177,
      "learning_rate": 1.3607157364274747e-06,
      "loss": 0.0059,
      "step": 13083
    },
    {
      "epoch": 8.371081253998721,
      "grad_norm": 0.3681211290810267,
      "learning_rate": 1.3596723548953161e-06,
      "loss": 0.0048,
      "step": 13084
    },
    {
      "epoch": 8.371721049264236,
      "grad_norm": 0.3562389081623214,
      "learning_rate": 1.3586293443634546e-06,
      "loss": 0.0039,
      "step": 13085
    },
    {
      "epoch": 8.37236084452975,
      "grad_norm": 0.4306456637541144,
      "learning_rate": 1.3575867048766789e-06,
      "loss": 0.0051,
      "step": 13086
    },
    {
      "epoch": 8.373000639795265,
      "grad_norm": 0.46453808153754395,
      "learning_rate": 1.3565444364797574e-06,
      "loss": 0.006,
      "step": 13087
    },
    {
      "epoch": 8.37364043506078,
      "grad_norm": 0.43507507619359986,
      "learning_rate": 1.3555025392174393e-06,
      "loss": 0.0059,
      "step": 13088
    },
    {
      "epoch": 8.374280230326296,
      "grad_norm": 0.3642930624592677,
      "learning_rate": 1.3544610131344672e-06,
      "loss": 0.0039,
      "step": 13089
    },
    {
      "epoch": 8.37492002559181,
      "grad_norm": 0.4501686341987547,
      "learning_rate": 1.3534198582755598e-06,
      "loss": 0.0063,
      "step": 13090
    },
    {
      "epoch": 8.375559820857326,
      "grad_norm": 0.40884362710373773,
      "learning_rate": 1.3523790746854215e-06,
      "loss": 0.0054,
      "step": 13091
    },
    {
      "epoch": 8.376199616122841,
      "grad_norm": 0.4003739180957497,
      "learning_rate": 1.3513386624087433e-06,
      "loss": 0.0048,
      "step": 13092
    },
    {
      "epoch": 8.376839411388357,
      "grad_norm": 0.463767344198273,
      "learning_rate": 1.3502986214901991e-06,
      "loss": 0.0053,
      "step": 13093
    },
    {
      "epoch": 8.37747920665387,
      "grad_norm": 0.5423963628392974,
      "learning_rate": 1.349258951974446e-06,
      "loss": 0.0065,
      "step": 13094
    },
    {
      "epoch": 8.378119001919385,
      "grad_norm": 0.5014978738469063,
      "learning_rate": 1.3482196539061244e-06,
      "loss": 0.0066,
      "step": 13095
    },
    {
      "epoch": 8.3787587971849,
      "grad_norm": 0.37878482740450886,
      "learning_rate": 1.34718072732986e-06,
      "loss": 0.0049,
      "step": 13096
    },
    {
      "epoch": 8.379398592450416,
      "grad_norm": 0.4072444684600854,
      "learning_rate": 1.3461421722902634e-06,
      "loss": 0.0055,
      "step": 13097
    },
    {
      "epoch": 8.380038387715931,
      "grad_norm": 0.5076455249598478,
      "learning_rate": 1.3451039888319262e-06,
      "loss": 0.0069,
      "step": 13098
    },
    {
      "epoch": 8.380678182981447,
      "grad_norm": 0.4022555322233996,
      "learning_rate": 1.3440661769994302e-06,
      "loss": 0.0047,
      "step": 13099
    },
    {
      "epoch": 8.381317978246962,
      "grad_norm": 0.43600911212778654,
      "learning_rate": 1.343028736837334e-06,
      "loss": 0.0074,
      "step": 13100
    },
    {
      "epoch": 8.381957773512475,
      "grad_norm": 0.35559472240951473,
      "learning_rate": 1.3419916683901824e-06,
      "loss": 0.0043,
      "step": 13101
    },
    {
      "epoch": 8.38259756877799,
      "grad_norm": 0.39624631563494894,
      "learning_rate": 1.3409549717025094e-06,
      "loss": 0.0055,
      "step": 13102
    },
    {
      "epoch": 8.383237364043506,
      "grad_norm": 0.4226026388686675,
      "learning_rate": 1.3399186468188263e-06,
      "loss": 0.0053,
      "step": 13103
    },
    {
      "epoch": 8.383877159309021,
      "grad_norm": 0.6232715356988563,
      "learning_rate": 1.338882693783632e-06,
      "loss": 0.0084,
      "step": 13104
    },
    {
      "epoch": 8.384516954574536,
      "grad_norm": 0.5841539652385404,
      "learning_rate": 1.337847112641406e-06,
      "loss": 0.0069,
      "step": 13105
    },
    {
      "epoch": 8.385156749840052,
      "grad_norm": 0.4807717479994068,
      "learning_rate": 1.3368119034366179e-06,
      "loss": 0.007,
      "step": 13106
    },
    {
      "epoch": 8.385796545105567,
      "grad_norm": 0.53075589804395,
      "learning_rate": 1.335777066213716e-06,
      "loss": 0.0083,
      "step": 13107
    },
    {
      "epoch": 8.38643634037108,
      "grad_norm": 0.47712473674695316,
      "learning_rate": 1.3347426010171316e-06,
      "loss": 0.0072,
      "step": 13108
    },
    {
      "epoch": 8.387076135636596,
      "grad_norm": 0.4200457588529897,
      "learning_rate": 1.3337085078912892e-06,
      "loss": 0.0058,
      "step": 13109
    },
    {
      "epoch": 8.387715930902111,
      "grad_norm": 0.4430493313895195,
      "learning_rate": 1.3326747868805867e-06,
      "loss": 0.0055,
      "step": 13110
    },
    {
      "epoch": 8.388355726167626,
      "grad_norm": 0.4016532091637223,
      "learning_rate": 1.3316414380294107e-06,
      "loss": 0.0047,
      "step": 13111
    },
    {
      "epoch": 8.388995521433142,
      "grad_norm": 0.4320361064099798,
      "learning_rate": 1.3306084613821312e-06,
      "loss": 0.0054,
      "step": 13112
    },
    {
      "epoch": 8.389635316698657,
      "grad_norm": 0.5007221613287957,
      "learning_rate": 1.3295758569831029e-06,
      "loss": 0.0066,
      "step": 13113
    },
    {
      "epoch": 8.390275111964172,
      "grad_norm": 0.5064972222685865,
      "learning_rate": 1.3285436248766648e-06,
      "loss": 0.0044,
      "step": 13114
    },
    {
      "epoch": 8.390914907229687,
      "grad_norm": 0.48770339764899373,
      "learning_rate": 1.3275117651071346e-06,
      "loss": 0.0056,
      "step": 13115
    },
    {
      "epoch": 8.3915547024952,
      "grad_norm": 0.4932580717184476,
      "learning_rate": 1.3264802777188246e-06,
      "loss": 0.0047,
      "step": 13116
    },
    {
      "epoch": 8.392194497760716,
      "grad_norm": 0.5720905100994602,
      "learning_rate": 1.325449162756023e-06,
      "loss": 0.0072,
      "step": 13117
    },
    {
      "epoch": 8.392834293026231,
      "grad_norm": 0.6082987968643597,
      "learning_rate": 1.3244184202630017e-06,
      "loss": 0.0081,
      "step": 13118
    },
    {
      "epoch": 8.393474088291747,
      "grad_norm": 0.5374301188041537,
      "learning_rate": 1.3233880502840213e-06,
      "loss": 0.0076,
      "step": 13119
    },
    {
      "epoch": 8.394113883557262,
      "grad_norm": 0.5432668417536827,
      "learning_rate": 1.3223580528633251e-06,
      "loss": 0.0074,
      "step": 13120
    },
    {
      "epoch": 8.394753678822777,
      "grad_norm": 0.6097184797981199,
      "learning_rate": 1.3213284280451366e-06,
      "loss": 0.0071,
      "step": 13121
    },
    {
      "epoch": 8.395393474088293,
      "grad_norm": 0.36472314260254707,
      "learning_rate": 1.3202991758736662e-06,
      "loss": 0.0043,
      "step": 13122
    },
    {
      "epoch": 8.396033269353806,
      "grad_norm": 0.36543891862779815,
      "learning_rate": 1.3192702963931104e-06,
      "loss": 0.0042,
      "step": 13123
    },
    {
      "epoch": 8.396673064619321,
      "grad_norm": 0.539616797857562,
      "learning_rate": 1.3182417896476473e-06,
      "loss": 0.0086,
      "step": 13124
    },
    {
      "epoch": 8.397312859884837,
      "grad_norm": 0.44814340892363747,
      "learning_rate": 1.3172136556814375e-06,
      "loss": 0.0062,
      "step": 13125
    },
    {
      "epoch": 8.397952655150352,
      "grad_norm": 0.448083546967688,
      "learning_rate": 1.3161858945386275e-06,
      "loss": 0.005,
      "step": 13126
    },
    {
      "epoch": 8.398592450415867,
      "grad_norm": 0.42559658919543003,
      "learning_rate": 1.3151585062633486e-06,
      "loss": 0.0051,
      "step": 13127
    },
    {
      "epoch": 8.399232245681382,
      "grad_norm": 0.38877375580915485,
      "learning_rate": 1.3141314908997115e-06,
      "loss": 0.0053,
      "step": 13128
    },
    {
      "epoch": 8.399872040946898,
      "grad_norm": 0.5348322296152074,
      "learning_rate": 1.3131048484918197e-06,
      "loss": 0.007,
      "step": 13129
    },
    {
      "epoch": 8.400511836212413,
      "grad_norm": 0.3830277131644772,
      "learning_rate": 1.3120785790837531e-06,
      "loss": 0.0047,
      "step": 13130
    },
    {
      "epoch": 8.401151631477926,
      "grad_norm": 0.40496435243060386,
      "learning_rate": 1.3110526827195768e-06,
      "loss": 0.0053,
      "step": 13131
    },
    {
      "epoch": 8.401791426743442,
      "grad_norm": 0.42647133770683965,
      "learning_rate": 1.3100271594433412e-06,
      "loss": 0.0064,
      "step": 13132
    },
    {
      "epoch": 8.402431222008957,
      "grad_norm": 0.3327799869807191,
      "learning_rate": 1.309002009299082e-06,
      "loss": 0.0035,
      "step": 13133
    },
    {
      "epoch": 8.403071017274472,
      "grad_norm": 0.4475870525974451,
      "learning_rate": 1.307977232330817e-06,
      "loss": 0.0054,
      "step": 13134
    },
    {
      "epoch": 8.403710812539988,
      "grad_norm": 0.5073933106664137,
      "learning_rate": 1.306952828582545e-06,
      "loss": 0.0061,
      "step": 13135
    },
    {
      "epoch": 8.404350607805503,
      "grad_norm": 0.4048648306841362,
      "learning_rate": 1.3059287980982583e-06,
      "loss": 0.0052,
      "step": 13136
    },
    {
      "epoch": 8.404990403071018,
      "grad_norm": 0.433360223215817,
      "learning_rate": 1.3049051409219227e-06,
      "loss": 0.0069,
      "step": 13137
    },
    {
      "epoch": 8.405630198336532,
      "grad_norm": 0.5227017347717805,
      "learning_rate": 1.3038818570974897e-06,
      "loss": 0.007,
      "step": 13138
    },
    {
      "epoch": 8.406269993602047,
      "grad_norm": 0.4193520083594454,
      "learning_rate": 1.3028589466689045e-06,
      "loss": 0.0047,
      "step": 13139
    },
    {
      "epoch": 8.406909788867562,
      "grad_norm": 0.4064856811063989,
      "learning_rate": 1.301836409680084e-06,
      "loss": 0.0051,
      "step": 13140
    },
    {
      "epoch": 8.407549584133077,
      "grad_norm": 0.4527599181752718,
      "learning_rate": 1.300814246174935e-06,
      "loss": 0.0046,
      "step": 13141
    },
    {
      "epoch": 8.408189379398593,
      "grad_norm": 0.3158799104925121,
      "learning_rate": 1.2997924561973485e-06,
      "loss": 0.0037,
      "step": 13142
    },
    {
      "epoch": 8.408829174664108,
      "grad_norm": 0.4656758109560124,
      "learning_rate": 1.2987710397911978e-06,
      "loss": 0.0062,
      "step": 13143
    },
    {
      "epoch": 8.409468969929623,
      "grad_norm": 0.3911091435523172,
      "learning_rate": 1.2977499970003393e-06,
      "loss": 0.0056,
      "step": 13144
    },
    {
      "epoch": 8.410108765195137,
      "grad_norm": 0.3509721163745536,
      "learning_rate": 1.2967293278686145e-06,
      "loss": 0.0048,
      "step": 13145
    },
    {
      "epoch": 8.410748560460652,
      "grad_norm": 0.3738845818307007,
      "learning_rate": 1.2957090324398514e-06,
      "loss": 0.0049,
      "step": 13146
    },
    {
      "epoch": 8.411388355726167,
      "grad_norm": 0.4627777489920291,
      "learning_rate": 1.2946891107578596e-06,
      "loss": 0.0056,
      "step": 13147
    },
    {
      "epoch": 8.412028150991683,
      "grad_norm": 0.403867594653585,
      "learning_rate": 1.2936695628664297e-06,
      "loss": 0.0055,
      "step": 13148
    },
    {
      "epoch": 8.412667946257198,
      "grad_norm": 0.4509892336606439,
      "learning_rate": 1.2926503888093433e-06,
      "loss": 0.0051,
      "step": 13149
    },
    {
      "epoch": 8.413307741522713,
      "grad_norm": 0.4938323178789607,
      "learning_rate": 1.2916315886303599e-06,
      "loss": 0.0067,
      "step": 13150
    },
    {
      "epoch": 8.413947536788228,
      "grad_norm": 0.47903297381287074,
      "learning_rate": 1.2906131623732255e-06,
      "loss": 0.0067,
      "step": 13151
    },
    {
      "epoch": 8.414587332053744,
      "grad_norm": 0.43930184494286295,
      "learning_rate": 1.2895951100816673e-06,
      "loss": 0.005,
      "step": 13152
    },
    {
      "epoch": 8.415227127319257,
      "grad_norm": 0.3182746283689503,
      "learning_rate": 1.2885774317994015e-06,
      "loss": 0.0038,
      "step": 13153
    },
    {
      "epoch": 8.415866922584772,
      "grad_norm": 0.4507872923077549,
      "learning_rate": 1.2875601275701254e-06,
      "loss": 0.0061,
      "step": 13154
    },
    {
      "epoch": 8.416506717850288,
      "grad_norm": 0.4051630442490314,
      "learning_rate": 1.2865431974375188e-06,
      "loss": 0.0051,
      "step": 13155
    },
    {
      "epoch": 8.417146513115803,
      "grad_norm": 0.46985280886070796,
      "learning_rate": 1.285526641445247e-06,
      "loss": 0.0061,
      "step": 13156
    },
    {
      "epoch": 8.417786308381318,
      "grad_norm": 0.4887505846525386,
      "learning_rate": 1.2845104596369606e-06,
      "loss": 0.0072,
      "step": 13157
    },
    {
      "epoch": 8.418426103646834,
      "grad_norm": 0.49264207201537774,
      "learning_rate": 1.2834946520562874e-06,
      "loss": 0.0061,
      "step": 13158
    },
    {
      "epoch": 8.419065898912349,
      "grad_norm": 0.4434665969062536,
      "learning_rate": 1.2824792187468516e-06,
      "loss": 0.0054,
      "step": 13159
    },
    {
      "epoch": 8.419705694177862,
      "grad_norm": 0.4551534515105831,
      "learning_rate": 1.2814641597522515e-06,
      "loss": 0.006,
      "step": 13160
    },
    {
      "epoch": 8.420345489443378,
      "grad_norm": 0.4378886731552207,
      "learning_rate": 1.28044947511607e-06,
      "loss": 0.0046,
      "step": 13161
    },
    {
      "epoch": 8.420985284708893,
      "grad_norm": 0.45160964227495015,
      "learning_rate": 1.279435164881876e-06,
      "loss": 0.0062,
      "step": 13162
    },
    {
      "epoch": 8.421625079974408,
      "grad_norm": 0.47390721922458884,
      "learning_rate": 1.2784212290932242e-06,
      "loss": 0.0053,
      "step": 13163
    },
    {
      "epoch": 8.422264875239923,
      "grad_norm": 0.5417354539141438,
      "learning_rate": 1.2774076677936508e-06,
      "loss": 0.007,
      "step": 13164
    },
    {
      "epoch": 8.422904670505439,
      "grad_norm": 0.4493288646587694,
      "learning_rate": 1.276394481026675e-06,
      "loss": 0.0063,
      "step": 13165
    },
    {
      "epoch": 8.423544465770954,
      "grad_norm": 0.4794824582425425,
      "learning_rate": 1.2753816688358033e-06,
      "loss": 0.0066,
      "step": 13166
    },
    {
      "epoch": 8.42418426103647,
      "grad_norm": 0.5315672250971134,
      "learning_rate": 1.2743692312645228e-06,
      "loss": 0.0071,
      "step": 13167
    },
    {
      "epoch": 8.424824056301983,
      "grad_norm": 0.4516140895727068,
      "learning_rate": 1.273357168356304e-06,
      "loss": 0.005,
      "step": 13168
    },
    {
      "epoch": 8.425463851567498,
      "grad_norm": 0.4550124564825827,
      "learning_rate": 1.2723454801546065e-06,
      "loss": 0.006,
      "step": 13169
    },
    {
      "epoch": 8.426103646833013,
      "grad_norm": 0.479139542631626,
      "learning_rate": 1.2713341667028689e-06,
      "loss": 0.0067,
      "step": 13170
    },
    {
      "epoch": 8.426743442098529,
      "grad_norm": 0.4489272080889013,
      "learning_rate": 1.2703232280445155e-06,
      "loss": 0.0057,
      "step": 13171
    },
    {
      "epoch": 8.427383237364044,
      "grad_norm": 0.463176628101727,
      "learning_rate": 1.2693126642229535e-06,
      "loss": 0.0056,
      "step": 13172
    },
    {
      "epoch": 8.428023032629559,
      "grad_norm": 0.5070442978034454,
      "learning_rate": 1.2683024752815743e-06,
      "loss": 0.0068,
      "step": 13173
    },
    {
      "epoch": 8.428662827895074,
      "grad_norm": 0.3431184985426145,
      "learning_rate": 1.267292661263756e-06,
      "loss": 0.0038,
      "step": 13174
    },
    {
      "epoch": 8.429302623160588,
      "grad_norm": 0.6131358232555598,
      "learning_rate": 1.266283222212853e-06,
      "loss": 0.0063,
      "step": 13175
    },
    {
      "epoch": 8.429942418426103,
      "grad_norm": 0.45556992657804607,
      "learning_rate": 1.265274158172215e-06,
      "loss": 0.0067,
      "step": 13176
    },
    {
      "epoch": 8.430582213691618,
      "grad_norm": 0.4248040546859423,
      "learning_rate": 1.2642654691851664e-06,
      "loss": 0.0051,
      "step": 13177
    },
    {
      "epoch": 8.431222008957134,
      "grad_norm": 0.3565011298784416,
      "learning_rate": 1.263257155295018e-06,
      "loss": 0.0044,
      "step": 13178
    },
    {
      "epoch": 8.431861804222649,
      "grad_norm": 0.40403029133583385,
      "learning_rate": 1.262249216545066e-06,
      "loss": 0.0053,
      "step": 13179
    },
    {
      "epoch": 8.432501599488164,
      "grad_norm": 0.4001141381316882,
      "learning_rate": 1.261241652978591e-06,
      "loss": 0.0056,
      "step": 13180
    },
    {
      "epoch": 8.43314139475368,
      "grad_norm": 0.4660010059509438,
      "learning_rate": 1.2602344646388542e-06,
      "loss": 0.0056,
      "step": 13181
    },
    {
      "epoch": 8.433781190019193,
      "grad_norm": 0.32623175998629617,
      "learning_rate": 1.2592276515691005e-06,
      "loss": 0.0044,
      "step": 13182
    },
    {
      "epoch": 8.434420985284708,
      "grad_norm": 0.40209734605709546,
      "learning_rate": 1.2582212138125638e-06,
      "loss": 0.0055,
      "step": 13183
    },
    {
      "epoch": 8.435060780550224,
      "grad_norm": 0.3985922220061304,
      "learning_rate": 1.2572151514124587e-06,
      "loss": 0.0054,
      "step": 13184
    },
    {
      "epoch": 8.435700575815739,
      "grad_norm": 0.6736507519892324,
      "learning_rate": 1.2562094644119826e-06,
      "loss": 0.0092,
      "step": 13185
    },
    {
      "epoch": 8.436340371081254,
      "grad_norm": 0.46327052083193915,
      "learning_rate": 1.255204152854319e-06,
      "loss": 0.0048,
      "step": 13186
    },
    {
      "epoch": 8.43698016634677,
      "grad_norm": 0.38147547381904107,
      "learning_rate": 1.2541992167826323e-06,
      "loss": 0.005,
      "step": 13187
    },
    {
      "epoch": 8.437619961612285,
      "grad_norm": 0.44906892468101245,
      "learning_rate": 1.253194656240072e-06,
      "loss": 0.0064,
      "step": 13188
    },
    {
      "epoch": 8.4382597568778,
      "grad_norm": 0.3996270463718132,
      "learning_rate": 1.2521904712697752e-06,
      "loss": 0.0052,
      "step": 13189
    },
    {
      "epoch": 8.438899552143313,
      "grad_norm": 0.44015237516642025,
      "learning_rate": 1.2511866619148593e-06,
      "loss": 0.0057,
      "step": 13190
    },
    {
      "epoch": 8.439539347408829,
      "grad_norm": 0.36591400260166396,
      "learning_rate": 1.2501832282184244e-06,
      "loss": 0.0044,
      "step": 13191
    },
    {
      "epoch": 8.440179142674344,
      "grad_norm": 0.3774556644005787,
      "learning_rate": 1.2491801702235552e-06,
      "loss": 0.0052,
      "step": 13192
    },
    {
      "epoch": 8.44081893793986,
      "grad_norm": 0.39666852599198515,
      "learning_rate": 1.2481774879733255e-06,
      "loss": 0.0055,
      "step": 13193
    },
    {
      "epoch": 8.441458733205375,
      "grad_norm": 0.4426553425152741,
      "learning_rate": 1.2471751815107858e-06,
      "loss": 0.005,
      "step": 13194
    },
    {
      "epoch": 8.44209852847089,
      "grad_norm": 0.47432532954974077,
      "learning_rate": 1.2461732508789714e-06,
      "loss": 0.0063,
      "step": 13195
    },
    {
      "epoch": 8.442738323736405,
      "grad_norm": 0.459823620631367,
      "learning_rate": 1.2451716961209082e-06,
      "loss": 0.0067,
      "step": 13196
    },
    {
      "epoch": 8.443378119001919,
      "grad_norm": 0.3804064756259917,
      "learning_rate": 1.2441705172795981e-06,
      "loss": 0.0053,
      "step": 13197
    },
    {
      "epoch": 8.444017914267434,
      "grad_norm": 0.5279693912395473,
      "learning_rate": 1.2431697143980314e-06,
      "loss": 0.0076,
      "step": 13198
    },
    {
      "epoch": 8.44465770953295,
      "grad_norm": 0.4751007417916736,
      "learning_rate": 1.2421692875191783e-06,
      "loss": 0.0067,
      "step": 13199
    },
    {
      "epoch": 8.445297504798464,
      "grad_norm": 0.36792028259995424,
      "learning_rate": 1.2411692366859972e-06,
      "loss": 0.0036,
      "step": 13200
    },
    {
      "epoch": 8.44593730006398,
      "grad_norm": 0.42139002678157333,
      "learning_rate": 1.24016956194143e-06,
      "loss": 0.0043,
      "step": 13201
    },
    {
      "epoch": 8.446577095329495,
      "grad_norm": 0.47796438930192553,
      "learning_rate": 1.2391702633283986e-06,
      "loss": 0.0066,
      "step": 13202
    },
    {
      "epoch": 8.44721689059501,
      "grad_norm": 0.2739072752375373,
      "learning_rate": 1.2381713408898122e-06,
      "loss": 0.0035,
      "step": 13203
    },
    {
      "epoch": 8.447856685860526,
      "grad_norm": 0.3533271738838019,
      "learning_rate": 1.2371727946685618e-06,
      "loss": 0.0043,
      "step": 13204
    },
    {
      "epoch": 8.448496481126039,
      "grad_norm": 0.4323315393156913,
      "learning_rate": 1.2361746247075224e-06,
      "loss": 0.0058,
      "step": 13205
    },
    {
      "epoch": 8.449136276391554,
      "grad_norm": 0.42456282608229834,
      "learning_rate": 1.2351768310495571e-06,
      "loss": 0.0054,
      "step": 13206
    },
    {
      "epoch": 8.44977607165707,
      "grad_norm": 0.31814393406903974,
      "learning_rate": 1.2341794137375062e-06,
      "loss": 0.0042,
      "step": 13207
    },
    {
      "epoch": 8.450415866922585,
      "grad_norm": 0.5330094606823798,
      "learning_rate": 1.2331823728141989e-06,
      "loss": 0.0059,
      "step": 13208
    },
    {
      "epoch": 8.4510556621881,
      "grad_norm": 0.48155009817651323,
      "learning_rate": 1.232185708322443e-06,
      "loss": 0.0062,
      "step": 13209
    },
    {
      "epoch": 8.451695457453615,
      "grad_norm": 0.49733392645429303,
      "learning_rate": 1.231189420305039e-06,
      "loss": 0.0056,
      "step": 13210
    },
    {
      "epoch": 8.45233525271913,
      "grad_norm": 0.45451279312112625,
      "learning_rate": 1.2301935088047622e-06,
      "loss": 0.0055,
      "step": 13211
    },
    {
      "epoch": 8.452975047984644,
      "grad_norm": 0.37466793154953815,
      "learning_rate": 1.2291979738643734e-06,
      "loss": 0.0042,
      "step": 13212
    },
    {
      "epoch": 8.45361484325016,
      "grad_norm": 0.4612320717166756,
      "learning_rate": 1.2282028155266246e-06,
      "loss": 0.0058,
      "step": 13213
    },
    {
      "epoch": 8.454254638515675,
      "grad_norm": 0.44199002470284726,
      "learning_rate": 1.2272080338342418e-06,
      "loss": 0.0063,
      "step": 13214
    },
    {
      "epoch": 8.45489443378119,
      "grad_norm": 0.44291993985112627,
      "learning_rate": 1.226213628829942e-06,
      "loss": 0.0057,
      "step": 13215
    },
    {
      "epoch": 8.455534229046705,
      "grad_norm": 0.48585184409751747,
      "learning_rate": 1.2252196005564198e-06,
      "loss": 0.0069,
      "step": 13216
    },
    {
      "epoch": 8.45617402431222,
      "grad_norm": 0.41936561834555597,
      "learning_rate": 1.2242259490563602e-06,
      "loss": 0.0055,
      "step": 13217
    },
    {
      "epoch": 8.456813819577736,
      "grad_norm": 0.4370784075446082,
      "learning_rate": 1.2232326743724276e-06,
      "loss": 0.0054,
      "step": 13218
    },
    {
      "epoch": 8.45745361484325,
      "grad_norm": 0.33833490044634956,
      "learning_rate": 1.2222397765472682e-06,
      "loss": 0.004,
      "step": 13219
    },
    {
      "epoch": 8.458093410108765,
      "grad_norm": 0.4586328054934998,
      "learning_rate": 1.2212472556235211e-06,
      "loss": 0.0053,
      "step": 13220
    },
    {
      "epoch": 8.45873320537428,
      "grad_norm": 0.5221842092068276,
      "learning_rate": 1.2202551116438e-06,
      "loss": 0.0051,
      "step": 13221
    },
    {
      "epoch": 8.459373000639795,
      "grad_norm": 0.4363119785935827,
      "learning_rate": 1.2192633446507052e-06,
      "loss": 0.005,
      "step": 13222
    },
    {
      "epoch": 8.46001279590531,
      "grad_norm": 0.4159482552915613,
      "learning_rate": 1.2182719546868249e-06,
      "loss": 0.0054,
      "step": 13223
    },
    {
      "epoch": 8.460652591170826,
      "grad_norm": 0.4764737976632694,
      "learning_rate": 1.217280941794724e-06,
      "loss": 0.005,
      "step": 13224
    },
    {
      "epoch": 8.461292386436341,
      "grad_norm": 0.5567971041752762,
      "learning_rate": 1.2162903060169552e-06,
      "loss": 0.0077,
      "step": 13225
    },
    {
      "epoch": 8.461932181701856,
      "grad_norm": 0.47700664724682257,
      "learning_rate": 1.2153000473960575e-06,
      "loss": 0.0068,
      "step": 13226
    },
    {
      "epoch": 8.46257197696737,
      "grad_norm": 0.3785939350704572,
      "learning_rate": 1.2143101659745494e-06,
      "loss": 0.0049,
      "step": 13227
    },
    {
      "epoch": 8.463211772232885,
      "grad_norm": 0.39001662357283345,
      "learning_rate": 1.2133206617949334e-06,
      "loss": 0.0038,
      "step": 13228
    },
    {
      "epoch": 8.4638515674984,
      "grad_norm": 0.434039395598794,
      "learning_rate": 1.2123315348996966e-06,
      "loss": 0.006,
      "step": 13229
    },
    {
      "epoch": 8.464491362763916,
      "grad_norm": 0.4324984930700336,
      "learning_rate": 1.2113427853313132e-06,
      "loss": 0.0056,
      "step": 13230
    },
    {
      "epoch": 8.46513115802943,
      "grad_norm": 0.4225500548936478,
      "learning_rate": 1.2103544131322375e-06,
      "loss": 0.0053,
      "step": 13231
    },
    {
      "epoch": 8.465770953294946,
      "grad_norm": 0.43435102788285923,
      "learning_rate": 1.2093664183449071e-06,
      "loss": 0.0056,
      "step": 13232
    },
    {
      "epoch": 8.466410748560461,
      "grad_norm": 0.4371738939007003,
      "learning_rate": 1.2083788010117458e-06,
      "loss": 0.005,
      "step": 13233
    },
    {
      "epoch": 8.467050543825975,
      "grad_norm": 0.4154826743987583,
      "learning_rate": 1.2073915611751586e-06,
      "loss": 0.0042,
      "step": 13234
    },
    {
      "epoch": 8.46769033909149,
      "grad_norm": 0.3423258513996876,
      "learning_rate": 1.2064046988775358e-06,
      "loss": 0.0037,
      "step": 13235
    },
    {
      "epoch": 8.468330134357005,
      "grad_norm": 0.41539696963697836,
      "learning_rate": 1.2054182141612547e-06,
      "loss": 0.0049,
      "step": 13236
    },
    {
      "epoch": 8.46896992962252,
      "grad_norm": 0.3796825429195963,
      "learning_rate": 1.2044321070686715e-06,
      "loss": 0.0047,
      "step": 13237
    },
    {
      "epoch": 8.469609724888036,
      "grad_norm": 0.4861468414447173,
      "learning_rate": 1.2034463776421268e-06,
      "loss": 0.0067,
      "step": 13238
    },
    {
      "epoch": 8.470249520153551,
      "grad_norm": 0.46271412357390446,
      "learning_rate": 1.2024610259239445e-06,
      "loss": 0.0049,
      "step": 13239
    },
    {
      "epoch": 8.470889315419067,
      "grad_norm": 0.5277856771069621,
      "learning_rate": 1.2014760519564394e-06,
      "loss": 0.007,
      "step": 13240
    },
    {
      "epoch": 8.471529110684582,
      "grad_norm": 0.4123911472338296,
      "learning_rate": 1.2004914557819004e-06,
      "loss": 0.0058,
      "step": 13241
    },
    {
      "epoch": 8.472168905950095,
      "grad_norm": 0.45365984788769653,
      "learning_rate": 1.199507237442603e-06,
      "loss": 0.0064,
      "step": 13242
    },
    {
      "epoch": 8.47280870121561,
      "grad_norm": 0.4671187917157396,
      "learning_rate": 1.1985233969808118e-06,
      "loss": 0.0052,
      "step": 13243
    },
    {
      "epoch": 8.473448496481126,
      "grad_norm": 0.3893461846744969,
      "learning_rate": 1.1975399344387707e-06,
      "loss": 0.0046,
      "step": 13244
    },
    {
      "epoch": 8.474088291746641,
      "grad_norm": 0.4329168964835562,
      "learning_rate": 1.1965568498587055e-06,
      "loss": 0.0061,
      "step": 13245
    },
    {
      "epoch": 8.474728087012156,
      "grad_norm": 0.39269663767981716,
      "learning_rate": 1.1955741432828283e-06,
      "loss": 0.0051,
      "step": 13246
    },
    {
      "epoch": 8.475367882277672,
      "grad_norm": 0.3732887972097558,
      "learning_rate": 1.1945918147533364e-06,
      "loss": 0.0046,
      "step": 13247
    },
    {
      "epoch": 8.476007677543187,
      "grad_norm": 0.479363564851608,
      "learning_rate": 1.1936098643124083e-06,
      "loss": 0.0058,
      "step": 13248
    },
    {
      "epoch": 8.4766474728087,
      "grad_norm": 0.3466057417811353,
      "learning_rate": 1.1926282920022058e-06,
      "loss": 0.0046,
      "step": 13249
    },
    {
      "epoch": 8.477287268074216,
      "grad_norm": 0.42571949629635697,
      "learning_rate": 1.1916470978648787e-06,
      "loss": 0.0066,
      "step": 13250
    },
    {
      "epoch": 8.477927063339731,
      "grad_norm": 0.438747236965553,
      "learning_rate": 1.1906662819425563e-06,
      "loss": 0.0056,
      "step": 13251
    },
    {
      "epoch": 8.478566858605246,
      "grad_norm": 0.4905160666285632,
      "learning_rate": 1.1896858442773517e-06,
      "loss": 0.0052,
      "step": 13252
    },
    {
      "epoch": 8.479206653870762,
      "grad_norm": 0.5151336906464616,
      "learning_rate": 1.188705784911366e-06,
      "loss": 0.0057,
      "step": 13253
    },
    {
      "epoch": 8.479846449136277,
      "grad_norm": 0.5019960155565589,
      "learning_rate": 1.1877261038866794e-06,
      "loss": 0.0065,
      "step": 13254
    },
    {
      "epoch": 8.480486244401792,
      "grad_norm": 0.4264785682780601,
      "learning_rate": 1.1867468012453575e-06,
      "loss": 0.0049,
      "step": 13255
    },
    {
      "epoch": 8.481126039667306,
      "grad_norm": 0.4015864508775749,
      "learning_rate": 1.1857678770294511e-06,
      "loss": 0.0044,
      "step": 13256
    },
    {
      "epoch": 8.48176583493282,
      "grad_norm": 0.3725641837044689,
      "learning_rate": 1.1847893312809932e-06,
      "loss": 0.0051,
      "step": 13257
    },
    {
      "epoch": 8.482405630198336,
      "grad_norm": 0.47928728510570906,
      "learning_rate": 1.1838111640420014e-06,
      "loss": 0.006,
      "step": 13258
    },
    {
      "epoch": 8.483045425463851,
      "grad_norm": 0.5584473213624843,
      "learning_rate": 1.1828333753544718e-06,
      "loss": 0.0077,
      "step": 13259
    },
    {
      "epoch": 8.483685220729367,
      "grad_norm": 0.4961408399202067,
      "learning_rate": 1.1818559652603956e-06,
      "loss": 0.0057,
      "step": 13260
    },
    {
      "epoch": 8.484325015994882,
      "grad_norm": 0.4035670634399079,
      "learning_rate": 1.180878933801738e-06,
      "loss": 0.0054,
      "step": 13261
    },
    {
      "epoch": 8.484964811260397,
      "grad_norm": 0.41109851526701785,
      "learning_rate": 1.1799022810204497e-06,
      "loss": 0.0046,
      "step": 13262
    },
    {
      "epoch": 8.485604606525913,
      "grad_norm": 0.4454015977345431,
      "learning_rate": 1.1789260069584696e-06,
      "loss": 0.006,
      "step": 13263
    },
    {
      "epoch": 8.486244401791426,
      "grad_norm": 0.4711282562158258,
      "learning_rate": 1.177950111657714e-06,
      "loss": 0.0069,
      "step": 13264
    },
    {
      "epoch": 8.486884197056941,
      "grad_norm": 0.3897214330354719,
      "learning_rate": 1.1769745951600852e-06,
      "loss": 0.005,
      "step": 13265
    },
    {
      "epoch": 8.487523992322457,
      "grad_norm": 0.33890706527468634,
      "learning_rate": 1.175999457507475e-06,
      "loss": 0.0045,
      "step": 13266
    },
    {
      "epoch": 8.488163787587972,
      "grad_norm": 0.4738887758033389,
      "learning_rate": 1.1750246987417512e-06,
      "loss": 0.006,
      "step": 13267
    },
    {
      "epoch": 8.488803582853487,
      "grad_norm": 0.40322031360227306,
      "learning_rate": 1.1740503189047691e-06,
      "loss": 0.006,
      "step": 13268
    },
    {
      "epoch": 8.489443378119002,
      "grad_norm": 0.48860671145926304,
      "learning_rate": 1.173076318038363e-06,
      "loss": 0.0067,
      "step": 13269
    },
    {
      "epoch": 8.490083173384518,
      "grad_norm": 0.39462342748770873,
      "learning_rate": 1.1721026961843608e-06,
      "loss": 0.0051,
      "step": 13270
    },
    {
      "epoch": 8.490722968650031,
      "grad_norm": 0.48745845466546245,
      "learning_rate": 1.1711294533845651e-06,
      "loss": 0.0065,
      "step": 13271
    },
    {
      "epoch": 8.491362763915546,
      "grad_norm": 0.4620848924912874,
      "learning_rate": 1.1701565896807631e-06,
      "loss": 0.0061,
      "step": 13272
    },
    {
      "epoch": 8.492002559181062,
      "grad_norm": 0.5634504399079173,
      "learning_rate": 1.1691841051147313e-06,
      "loss": 0.0078,
      "step": 13273
    },
    {
      "epoch": 8.492642354446577,
      "grad_norm": 0.46155150649362736,
      "learning_rate": 1.1682119997282259e-06,
      "loss": 0.0062,
      "step": 13274
    },
    {
      "epoch": 8.493282149712092,
      "grad_norm": 0.4308960817849752,
      "learning_rate": 1.1672402735629873e-06,
      "loss": 0.005,
      "step": 13275
    },
    {
      "epoch": 8.493921944977608,
      "grad_norm": 0.4190437583683335,
      "learning_rate": 1.1662689266607385e-06,
      "loss": 0.0055,
      "step": 13276
    },
    {
      "epoch": 8.494561740243123,
      "grad_norm": 0.46035194319659517,
      "learning_rate": 1.1652979590631884e-06,
      "loss": 0.0057,
      "step": 13277
    },
    {
      "epoch": 8.495201535508638,
      "grad_norm": 0.3674843332616302,
      "learning_rate": 1.1643273708120274e-06,
      "loss": 0.0049,
      "step": 13278
    },
    {
      "epoch": 8.495841330774152,
      "grad_norm": 0.4371871449510139,
      "learning_rate": 1.1633571619489303e-06,
      "loss": 0.0059,
      "step": 13279
    },
    {
      "epoch": 8.496481126039667,
      "grad_norm": 0.42585194541130567,
      "learning_rate": 1.1623873325155598e-06,
      "loss": 0.005,
      "step": 13280
    },
    {
      "epoch": 8.497120921305182,
      "grad_norm": 0.4028968550706599,
      "learning_rate": 1.161417882553556e-06,
      "loss": 0.0063,
      "step": 13281
    },
    {
      "epoch": 8.497760716570697,
      "grad_norm": 0.4977253012253542,
      "learning_rate": 1.1604488121045431e-06,
      "loss": 0.0071,
      "step": 13282
    },
    {
      "epoch": 8.498400511836213,
      "grad_norm": 0.509479129082234,
      "learning_rate": 1.1594801212101369e-06,
      "loss": 0.0066,
      "step": 13283
    },
    {
      "epoch": 8.499040307101728,
      "grad_norm": 0.4598446992813008,
      "learning_rate": 1.1585118099119275e-06,
      "loss": 0.0055,
      "step": 13284
    },
    {
      "epoch": 8.499680102367243,
      "grad_norm": 0.36114334948681415,
      "learning_rate": 1.1575438782514937e-06,
      "loss": 0.0042,
      "step": 13285
    },
    {
      "epoch": 8.500319897632757,
      "grad_norm": 0.3951233644623775,
      "learning_rate": 1.1565763262703932e-06,
      "loss": 0.0049,
      "step": 13286
    },
    {
      "epoch": 8.500959692898272,
      "grad_norm": 0.5018741966569967,
      "learning_rate": 1.1556091540101777e-06,
      "loss": 0.0076,
      "step": 13287
    },
    {
      "epoch": 8.501599488163787,
      "grad_norm": 0.3546235941193951,
      "learning_rate": 1.1546423615123715e-06,
      "loss": 0.0044,
      "step": 13288
    },
    {
      "epoch": 8.502239283429303,
      "grad_norm": 0.3826665302150067,
      "learning_rate": 1.1536759488184846e-06,
      "loss": 0.0041,
      "step": 13289
    },
    {
      "epoch": 8.502879078694818,
      "grad_norm": 0.46139077723813243,
      "learning_rate": 1.1527099159700194e-06,
      "loss": 0.0056,
      "step": 13290
    },
    {
      "epoch": 8.503518873960333,
      "grad_norm": 0.38295911323496096,
      "learning_rate": 1.151744263008452e-06,
      "loss": 0.0047,
      "step": 13291
    },
    {
      "epoch": 8.504158669225848,
      "grad_norm": 0.5116846420936427,
      "learning_rate": 1.1507789899752453e-06,
      "loss": 0.0077,
      "step": 13292
    },
    {
      "epoch": 8.504798464491362,
      "grad_norm": 0.35308813429308095,
      "learning_rate": 1.1498140969118476e-06,
      "loss": 0.0052,
      "step": 13293
    },
    {
      "epoch": 8.505438259756877,
      "grad_norm": 0.46990722862140905,
      "learning_rate": 1.1488495838596903e-06,
      "loss": 0.0059,
      "step": 13294
    },
    {
      "epoch": 8.506078055022392,
      "grad_norm": 0.41446649865654644,
      "learning_rate": 1.1478854508601855e-06,
      "loss": 0.0048,
      "step": 13295
    },
    {
      "epoch": 8.506717850287908,
      "grad_norm": 0.3647768259509093,
      "learning_rate": 1.1469216979547315e-06,
      "loss": 0.0037,
      "step": 13296
    },
    {
      "epoch": 8.507357645553423,
      "grad_norm": 0.44275626434750204,
      "learning_rate": 1.145958325184713e-06,
      "loss": 0.0054,
      "step": 13297
    },
    {
      "epoch": 8.507997440818938,
      "grad_norm": 0.364553326957004,
      "learning_rate": 1.1449953325914942e-06,
      "loss": 0.0046,
      "step": 13298
    },
    {
      "epoch": 8.508637236084454,
      "grad_norm": 0.3717540598711817,
      "learning_rate": 1.1440327202164225e-06,
      "loss": 0.0045,
      "step": 13299
    },
    {
      "epoch": 8.509277031349969,
      "grad_norm": 0.4696221996255852,
      "learning_rate": 1.1430704881008336e-06,
      "loss": 0.0068,
      "step": 13300
    },
    {
      "epoch": 8.509916826615482,
      "grad_norm": 0.4883336086345865,
      "learning_rate": 1.1421086362860422e-06,
      "loss": 0.0058,
      "step": 13301
    },
    {
      "epoch": 8.510556621880998,
      "grad_norm": 0.4475157475238674,
      "learning_rate": 1.1411471648133477e-06,
      "loss": 0.0061,
      "step": 13302
    },
    {
      "epoch": 8.511196417146513,
      "grad_norm": 0.4812730873382848,
      "learning_rate": 1.1401860737240377e-06,
      "loss": 0.0071,
      "step": 13303
    },
    {
      "epoch": 8.511836212412028,
      "grad_norm": 0.3533678580649376,
      "learning_rate": 1.1392253630593764e-06,
      "loss": 0.005,
      "step": 13304
    },
    {
      "epoch": 8.512476007677543,
      "grad_norm": 0.5173451570248394,
      "learning_rate": 1.138265032860616e-06,
      "loss": 0.0066,
      "step": 13305
    },
    {
      "epoch": 8.513115802943059,
      "grad_norm": 0.5516539505237678,
      "learning_rate": 1.1373050831689913e-06,
      "loss": 0.0074,
      "step": 13306
    },
    {
      "epoch": 8.513755598208574,
      "grad_norm": 0.42641388976451544,
      "learning_rate": 1.1363455140257218e-06,
      "loss": 0.0059,
      "step": 13307
    },
    {
      "epoch": 8.514395393474087,
      "grad_norm": 0.6199208208573651,
      "learning_rate": 1.1353863254720076e-06,
      "loss": 0.0056,
      "step": 13308
    },
    {
      "epoch": 8.515035188739603,
      "grad_norm": 0.4174403372486057,
      "learning_rate": 1.1344275175490337e-06,
      "loss": 0.0053,
      "step": 13309
    },
    {
      "epoch": 8.515674984005118,
      "grad_norm": 0.4861220993805294,
      "learning_rate": 1.133469090297975e-06,
      "loss": 0.007,
      "step": 13310
    },
    {
      "epoch": 8.516314779270633,
      "grad_norm": 0.399395369397305,
      "learning_rate": 1.1325110437599797e-06,
      "loss": 0.0046,
      "step": 13311
    },
    {
      "epoch": 8.516954574536149,
      "grad_norm": 0.4031830165855053,
      "learning_rate": 1.1315533779761846e-06,
      "loss": 0.0053,
      "step": 13312
    },
    {
      "epoch": 8.517594369801664,
      "grad_norm": 0.3541860005242424,
      "learning_rate": 1.1305960929877135e-06,
      "loss": 0.0042,
      "step": 13313
    },
    {
      "epoch": 8.51823416506718,
      "grad_norm": 0.43752883043976176,
      "learning_rate": 1.1296391888356694e-06,
      "loss": 0.0048,
      "step": 13314
    },
    {
      "epoch": 8.518873960332694,
      "grad_norm": 0.5605079385716331,
      "learning_rate": 1.1286826655611384e-06,
      "loss": 0.0071,
      "step": 13315
    },
    {
      "epoch": 8.519513755598208,
      "grad_norm": 0.40994474754226057,
      "learning_rate": 1.1277265232051914e-06,
      "loss": 0.005,
      "step": 13316
    },
    {
      "epoch": 8.520153550863723,
      "grad_norm": 0.3702934309267107,
      "learning_rate": 1.1267707618088875e-06,
      "loss": 0.006,
      "step": 13317
    },
    {
      "epoch": 8.520793346129238,
      "grad_norm": 0.4033023089959052,
      "learning_rate": 1.1258153814132622e-06,
      "loss": 0.0065,
      "step": 13318
    },
    {
      "epoch": 8.521433141394754,
      "grad_norm": 0.43831836098734844,
      "learning_rate": 1.1248603820593362e-06,
      "loss": 0.0057,
      "step": 13319
    },
    {
      "epoch": 8.522072936660269,
      "grad_norm": 0.3136629036072645,
      "learning_rate": 1.12390576378812e-06,
      "loss": 0.0038,
      "step": 13320
    },
    {
      "epoch": 8.522712731925784,
      "grad_norm": 0.592119534820234,
      "learning_rate": 1.122951526640601e-06,
      "loss": 0.0085,
      "step": 13321
    },
    {
      "epoch": 8.5233525271913,
      "grad_norm": 0.39982851212924725,
      "learning_rate": 1.1219976706577529e-06,
      "loss": 0.0044,
      "step": 13322
    },
    {
      "epoch": 8.523992322456813,
      "grad_norm": 0.5376346554869261,
      "learning_rate": 1.1210441958805306e-06,
      "loss": 0.0057,
      "step": 13323
    },
    {
      "epoch": 8.524632117722328,
      "grad_norm": 0.40287535497362603,
      "learning_rate": 1.1200911023498773e-06,
      "loss": 0.0053,
      "step": 13324
    },
    {
      "epoch": 8.525271912987844,
      "grad_norm": 0.37310897646130003,
      "learning_rate": 1.1191383901067166e-06,
      "loss": 0.0048,
      "step": 13325
    },
    {
      "epoch": 8.525911708253359,
      "grad_norm": 0.4217043610717656,
      "learning_rate": 1.1181860591919524e-06,
      "loss": 0.0061,
      "step": 13326
    },
    {
      "epoch": 8.526551503518874,
      "grad_norm": 0.43468659869903126,
      "learning_rate": 1.1172341096464812e-06,
      "loss": 0.0049,
      "step": 13327
    },
    {
      "epoch": 8.52719129878439,
      "grad_norm": 0.46472740187906614,
      "learning_rate": 1.1162825415111766e-06,
      "loss": 0.0061,
      "step": 13328
    },
    {
      "epoch": 8.527831094049905,
      "grad_norm": 0.47043568567457833,
      "learning_rate": 1.1153313548268951e-06,
      "loss": 0.0063,
      "step": 13329
    },
    {
      "epoch": 8.528470889315418,
      "grad_norm": 0.3982057106262636,
      "learning_rate": 1.1143805496344818e-06,
      "loss": 0.0051,
      "step": 13330
    },
    {
      "epoch": 8.529110684580933,
      "grad_norm": 0.3765905513088053,
      "learning_rate": 1.1134301259747627e-06,
      "loss": 0.0055,
      "step": 13331
    },
    {
      "epoch": 8.529750479846449,
      "grad_norm": 0.44464203479332126,
      "learning_rate": 1.112480083888543e-06,
      "loss": 0.0053,
      "step": 13332
    },
    {
      "epoch": 8.530390275111964,
      "grad_norm": 0.543306920357483,
      "learning_rate": 1.1115304234166201e-06,
      "loss": 0.0075,
      "step": 13333
    },
    {
      "epoch": 8.53103007037748,
      "grad_norm": 0.3502108865528481,
      "learning_rate": 1.1105811445997705e-06,
      "loss": 0.0037,
      "step": 13334
    },
    {
      "epoch": 8.531669865642995,
      "grad_norm": 0.4429491316504422,
      "learning_rate": 1.1096322474787535e-06,
      "loss": 0.0058,
      "step": 13335
    },
    {
      "epoch": 8.53230966090851,
      "grad_norm": 0.4432439077027579,
      "learning_rate": 1.1086837320943134e-06,
      "loss": 0.0047,
      "step": 13336
    },
    {
      "epoch": 8.532949456174025,
      "grad_norm": 0.4847278685805212,
      "learning_rate": 1.1077355984871763e-06,
      "loss": 0.0063,
      "step": 13337
    },
    {
      "epoch": 8.533589251439539,
      "grad_norm": 0.41254976549921746,
      "learning_rate": 1.1067878466980552e-06,
      "loss": 0.0057,
      "step": 13338
    },
    {
      "epoch": 8.534229046705054,
      "grad_norm": 0.40784537411357247,
      "learning_rate": 1.1058404767676422e-06,
      "loss": 0.0061,
      "step": 13339
    },
    {
      "epoch": 8.53486884197057,
      "grad_norm": 0.35039171057780993,
      "learning_rate": 1.1048934887366191e-06,
      "loss": 0.0049,
      "step": 13340
    },
    {
      "epoch": 8.535508637236084,
      "grad_norm": 0.4115065238232297,
      "learning_rate": 1.1039468826456468e-06,
      "loss": 0.0051,
      "step": 13341
    },
    {
      "epoch": 8.5361484325016,
      "grad_norm": 0.4433513083573397,
      "learning_rate": 1.1030006585353692e-06,
      "loss": 0.0052,
      "step": 13342
    },
    {
      "epoch": 8.536788227767115,
      "grad_norm": 0.4191994266194077,
      "learning_rate": 1.1020548164464183e-06,
      "loss": 0.0065,
      "step": 13343
    },
    {
      "epoch": 8.53742802303263,
      "grad_norm": 0.3439597785033764,
      "learning_rate": 1.1011093564194054e-06,
      "loss": 0.0039,
      "step": 13344
    },
    {
      "epoch": 8.538067818298144,
      "grad_norm": 0.4868922283521357,
      "learning_rate": 1.1001642784949263e-06,
      "loss": 0.0065,
      "step": 13345
    },
    {
      "epoch": 8.538707613563659,
      "grad_norm": 0.4508094585831318,
      "learning_rate": 1.09921958271356e-06,
      "loss": 0.006,
      "step": 13346
    },
    {
      "epoch": 8.539347408829174,
      "grad_norm": 0.38135756155085876,
      "learning_rate": 1.0982752691158727e-06,
      "loss": 0.005,
      "step": 13347
    },
    {
      "epoch": 8.53998720409469,
      "grad_norm": 0.43381698214364045,
      "learning_rate": 1.0973313377424112e-06,
      "loss": 0.0055,
      "step": 13348
    },
    {
      "epoch": 8.540626999360205,
      "grad_norm": 0.358766629739091,
      "learning_rate": 1.0963877886337038e-06,
      "loss": 0.0049,
      "step": 13349
    },
    {
      "epoch": 8.54126679462572,
      "grad_norm": 0.49961962424752476,
      "learning_rate": 1.0954446218302684e-06,
      "loss": 0.0065,
      "step": 13350
    },
    {
      "epoch": 8.541906589891235,
      "grad_norm": 0.43016420033185165,
      "learning_rate": 1.0945018373726001e-06,
      "loss": 0.0056,
      "step": 13351
    },
    {
      "epoch": 8.54254638515675,
      "grad_norm": 0.5699577021154297,
      "learning_rate": 1.0935594353011813e-06,
      "loss": 0.0066,
      "step": 13352
    },
    {
      "epoch": 8.543186180422264,
      "grad_norm": 0.5354889241145857,
      "learning_rate": 1.0926174156564773e-06,
      "loss": 0.0074,
      "step": 13353
    },
    {
      "epoch": 8.54382597568778,
      "grad_norm": 0.4496528689894343,
      "learning_rate": 1.0916757784789367e-06,
      "loss": 0.0062,
      "step": 13354
    },
    {
      "epoch": 8.544465770953295,
      "grad_norm": 0.4759547360362423,
      "learning_rate": 1.0907345238089916e-06,
      "loss": 0.0062,
      "step": 13355
    },
    {
      "epoch": 8.54510556621881,
      "grad_norm": 0.4585394646773197,
      "learning_rate": 1.0897936516870556e-06,
      "loss": 0.0066,
      "step": 13356
    },
    {
      "epoch": 8.545745361484325,
      "grad_norm": 0.42765677384485246,
      "learning_rate": 1.0888531621535326e-06,
      "loss": 0.0053,
      "step": 13357
    },
    {
      "epoch": 8.54638515674984,
      "grad_norm": 0.42976361199460456,
      "learning_rate": 1.0879130552488016e-06,
      "loss": 0.0053,
      "step": 13358
    },
    {
      "epoch": 8.547024952015356,
      "grad_norm": 0.5328592054943151,
      "learning_rate": 1.0869733310132303e-06,
      "loss": 0.0066,
      "step": 13359
    },
    {
      "epoch": 8.54766474728087,
      "grad_norm": 0.40195060060368204,
      "learning_rate": 1.0860339894871697e-06,
      "loss": 0.0047,
      "step": 13360
    },
    {
      "epoch": 8.548304542546385,
      "grad_norm": 0.34285643575614627,
      "learning_rate": 1.0850950307109543e-06,
      "loss": 0.0039,
      "step": 13361
    },
    {
      "epoch": 8.5489443378119,
      "grad_norm": 0.34872923412850687,
      "learning_rate": 1.0841564547248994e-06,
      "loss": 0.0041,
      "step": 13362
    },
    {
      "epoch": 8.549584133077415,
      "grad_norm": 0.49738743186284334,
      "learning_rate": 1.0832182615693033e-06,
      "loss": 0.0054,
      "step": 13363
    },
    {
      "epoch": 8.55022392834293,
      "grad_norm": 0.5533256675073607,
      "learning_rate": 1.0822804512844564e-06,
      "loss": 0.0062,
      "step": 13364
    },
    {
      "epoch": 8.550863723608446,
      "grad_norm": 0.413530865458762,
      "learning_rate": 1.0813430239106238e-06,
      "loss": 0.0052,
      "step": 13365
    },
    {
      "epoch": 8.551503518873961,
      "grad_norm": 0.5051686484125425,
      "learning_rate": 1.0804059794880562e-06,
      "loss": 0.0043,
      "step": 13366
    },
    {
      "epoch": 8.552143314139474,
      "grad_norm": 0.5189887550012249,
      "learning_rate": 1.0794693180569882e-06,
      "loss": 0.0062,
      "step": 13367
    },
    {
      "epoch": 8.55278310940499,
      "grad_norm": 0.39151576406710015,
      "learning_rate": 1.0785330396576398e-06,
      "loss": 0.0049,
      "step": 13368
    },
    {
      "epoch": 8.553422904670505,
      "grad_norm": 0.4004631073501692,
      "learning_rate": 1.0775971443302114e-06,
      "loss": 0.0047,
      "step": 13369
    },
    {
      "epoch": 8.55406269993602,
      "grad_norm": 0.5002474429106061,
      "learning_rate": 1.0766616321148915e-06,
      "loss": 0.0071,
      "step": 13370
    },
    {
      "epoch": 8.554702495201536,
      "grad_norm": 0.3410941353131591,
      "learning_rate": 1.0757265030518482e-06,
      "loss": 0.0042,
      "step": 13371
    },
    {
      "epoch": 8.55534229046705,
      "grad_norm": 0.4106140437695633,
      "learning_rate": 1.0747917571812328e-06,
      "loss": 0.0051,
      "step": 13372
    },
    {
      "epoch": 8.555982085732566,
      "grad_norm": 0.4148851180320382,
      "learning_rate": 1.0738573945431807e-06,
      "loss": 0.0054,
      "step": 13373
    },
    {
      "epoch": 8.556621880998081,
      "grad_norm": 0.46735342639756794,
      "learning_rate": 1.0729234151778168e-06,
      "loss": 0.0065,
      "step": 13374
    },
    {
      "epoch": 8.557261676263595,
      "grad_norm": 0.4181402156135958,
      "learning_rate": 1.071989819125241e-06,
      "loss": 0.0053,
      "step": 13375
    },
    {
      "epoch": 8.55790147152911,
      "grad_norm": 0.4391828060602117,
      "learning_rate": 1.0710566064255378e-06,
      "loss": 0.006,
      "step": 13376
    },
    {
      "epoch": 8.558541266794625,
      "grad_norm": 0.5006986107477494,
      "learning_rate": 1.0701237771187833e-06,
      "loss": 0.0065,
      "step": 13377
    },
    {
      "epoch": 8.55918106206014,
      "grad_norm": 0.44612561523336614,
      "learning_rate": 1.0691913312450287e-06,
      "loss": 0.0064,
      "step": 13378
    },
    {
      "epoch": 8.559820857325656,
      "grad_norm": 0.36964673184738217,
      "learning_rate": 1.0682592688443094e-06,
      "loss": 0.0039,
      "step": 13379
    },
    {
      "epoch": 8.560460652591171,
      "grad_norm": 0.5139097684149426,
      "learning_rate": 1.0673275899566515e-06,
      "loss": 0.0067,
      "step": 13380
    },
    {
      "epoch": 8.561100447856687,
      "grad_norm": 0.3457502203679387,
      "learning_rate": 1.0663962946220563e-06,
      "loss": 0.0042,
      "step": 13381
    },
    {
      "epoch": 8.561740243122202,
      "grad_norm": 0.47386382092288315,
      "learning_rate": 1.0654653828805128e-06,
      "loss": 0.0051,
      "step": 13382
    },
    {
      "epoch": 8.562380038387715,
      "grad_norm": 0.51898269104847,
      "learning_rate": 1.0645348547719925e-06,
      "loss": 0.0064,
      "step": 13383
    },
    {
      "epoch": 8.56301983365323,
      "grad_norm": 0.39836693400008677,
      "learning_rate": 1.06360471033645e-06,
      "loss": 0.0057,
      "step": 13384
    },
    {
      "epoch": 8.563659628918746,
      "grad_norm": 0.42194228333915984,
      "learning_rate": 1.0626749496138245e-06,
      "loss": 0.0042,
      "step": 13385
    },
    {
      "epoch": 8.564299424184261,
      "grad_norm": 0.5196332904200008,
      "learning_rate": 1.0617455726440373e-06,
      "loss": 0.0082,
      "step": 13386
    },
    {
      "epoch": 8.564939219449776,
      "grad_norm": 0.4641870803126981,
      "learning_rate": 1.0608165794669968e-06,
      "loss": 0.0072,
      "step": 13387
    },
    {
      "epoch": 8.565579014715292,
      "grad_norm": 0.4309673487538754,
      "learning_rate": 1.0598879701225906e-06,
      "loss": 0.0067,
      "step": 13388
    },
    {
      "epoch": 8.566218809980807,
      "grad_norm": 0.44070568785360253,
      "learning_rate": 1.0589597446506905e-06,
      "loss": 0.0065,
      "step": 13389
    },
    {
      "epoch": 8.56685860524632,
      "grad_norm": 0.4907964993965354,
      "learning_rate": 1.0580319030911556e-06,
      "loss": 0.007,
      "step": 13390
    },
    {
      "epoch": 8.567498400511836,
      "grad_norm": 0.5175455973155361,
      "learning_rate": 1.057104445483824e-06,
      "loss": 0.0068,
      "step": 13391
    },
    {
      "epoch": 8.568138195777351,
      "grad_norm": 0.4495372694326497,
      "learning_rate": 1.0561773718685198e-06,
      "loss": 0.0047,
      "step": 13392
    },
    {
      "epoch": 8.568777991042866,
      "grad_norm": 0.3942649196090423,
      "learning_rate": 1.0552506822850472e-06,
      "loss": 0.0048,
      "step": 13393
    },
    {
      "epoch": 8.569417786308382,
      "grad_norm": 0.5074111753531516,
      "learning_rate": 1.0543243767732004e-06,
      "loss": 0.0053,
      "step": 13394
    },
    {
      "epoch": 8.570057581573897,
      "grad_norm": 0.3895055913288915,
      "learning_rate": 1.0533984553727527e-06,
      "loss": 0.0048,
      "step": 13395
    },
    {
      "epoch": 8.570697376839412,
      "grad_norm": 0.38112610286843995,
      "learning_rate": 1.0524729181234594e-06,
      "loss": 0.0048,
      "step": 13396
    },
    {
      "epoch": 8.571337172104926,
      "grad_norm": 0.4599513346499114,
      "learning_rate": 1.0515477650650642e-06,
      "loss": 0.0059,
      "step": 13397
    },
    {
      "epoch": 8.571976967370441,
      "grad_norm": 0.374663808637149,
      "learning_rate": 1.0506229962372882e-06,
      "loss": 0.0048,
      "step": 13398
    },
    {
      "epoch": 8.572616762635956,
      "grad_norm": 0.42886194549632217,
      "learning_rate": 1.0496986116798403e-06,
      "loss": 0.0057,
      "step": 13399
    },
    {
      "epoch": 8.573256557901471,
      "grad_norm": 0.41712929732673343,
      "learning_rate": 1.0487746114324139e-06,
      "loss": 0.0047,
      "step": 13400
    },
    {
      "epoch": 8.573896353166987,
      "grad_norm": 0.5504727287880442,
      "learning_rate": 1.0478509955346827e-06,
      "loss": 0.0068,
      "step": 13401
    },
    {
      "epoch": 8.574536148432502,
      "grad_norm": 0.4429925582847526,
      "learning_rate": 1.0469277640263055e-06,
      "loss": 0.0055,
      "step": 13402
    },
    {
      "epoch": 8.575175943698017,
      "grad_norm": 0.5914418230369852,
      "learning_rate": 1.046004916946921e-06,
      "loss": 0.008,
      "step": 13403
    },
    {
      "epoch": 8.57581573896353,
      "grad_norm": 0.4717819446069525,
      "learning_rate": 1.0450824543361603e-06,
      "loss": 0.0056,
      "step": 13404
    },
    {
      "epoch": 8.576455534229046,
      "grad_norm": 0.39737772778712294,
      "learning_rate": 1.0441603762336294e-06,
      "loss": 0.0046,
      "step": 13405
    },
    {
      "epoch": 8.577095329494561,
      "grad_norm": 0.47832837828707436,
      "learning_rate": 1.0432386826789187e-06,
      "loss": 0.0055,
      "step": 13406
    },
    {
      "epoch": 8.577735124760077,
      "grad_norm": 0.3720741234884826,
      "learning_rate": 1.0423173737116076e-06,
      "loss": 0.0044,
      "step": 13407
    },
    {
      "epoch": 8.578374920025592,
      "grad_norm": 0.45185026759886965,
      "learning_rate": 1.0413964493712548e-06,
      "loss": 0.0059,
      "step": 13408
    },
    {
      "epoch": 8.579014715291107,
      "grad_norm": 0.5036015469242807,
      "learning_rate": 1.0404759096974003e-06,
      "loss": 0.0061,
      "step": 13409
    },
    {
      "epoch": 8.579654510556622,
      "grad_norm": 0.37608945003701666,
      "learning_rate": 1.0395557547295742e-06,
      "loss": 0.0051,
      "step": 13410
    },
    {
      "epoch": 8.580294305822138,
      "grad_norm": 0.4620101145126748,
      "learning_rate": 1.0386359845072846e-06,
      "loss": 0.0054,
      "step": 13411
    },
    {
      "epoch": 8.580934101087651,
      "grad_norm": 0.3970170325690429,
      "learning_rate": 1.0377165990700244e-06,
      "loss": 0.005,
      "step": 13412
    },
    {
      "epoch": 8.581573896353166,
      "grad_norm": 0.35506427669093216,
      "learning_rate": 1.0367975984572709e-06,
      "loss": 0.0042,
      "step": 13413
    },
    {
      "epoch": 8.582213691618682,
      "grad_norm": 0.3928826724337926,
      "learning_rate": 1.0358789827084847e-06,
      "loss": 0.0044,
      "step": 13414
    },
    {
      "epoch": 8.582853486884197,
      "grad_norm": 0.4153162498909074,
      "learning_rate": 1.0349607518631089e-06,
      "loss": 0.0046,
      "step": 13415
    },
    {
      "epoch": 8.583493282149712,
      "grad_norm": 0.452973467626943,
      "learning_rate": 1.0340429059605683e-06,
      "loss": 0.0055,
      "step": 13416
    },
    {
      "epoch": 8.584133077415228,
      "grad_norm": 0.4954229576936018,
      "learning_rate": 1.0331254450402783e-06,
      "loss": 0.006,
      "step": 13417
    },
    {
      "epoch": 8.584772872680743,
      "grad_norm": 0.46690306230726175,
      "learning_rate": 1.0322083691416318e-06,
      "loss": 0.0064,
      "step": 13418
    },
    {
      "epoch": 8.585412667946258,
      "grad_norm": 0.447976889872746,
      "learning_rate": 1.0312916783040027e-06,
      "loss": 0.0051,
      "step": 13419
    },
    {
      "epoch": 8.586052463211772,
      "grad_norm": 0.4077549862572863,
      "learning_rate": 1.0303753725667565e-06,
      "loss": 0.0044,
      "step": 13420
    },
    {
      "epoch": 8.586692258477287,
      "grad_norm": 0.5016806328932375,
      "learning_rate": 1.029459451969237e-06,
      "loss": 0.0073,
      "step": 13421
    },
    {
      "epoch": 8.587332053742802,
      "grad_norm": 0.5091364614593187,
      "learning_rate": 1.0285439165507705e-06,
      "loss": 0.0049,
      "step": 13422
    },
    {
      "epoch": 8.587971849008317,
      "grad_norm": 0.38827544024407623,
      "learning_rate": 1.027628766350668e-06,
      "loss": 0.0046,
      "step": 13423
    },
    {
      "epoch": 8.588611644273833,
      "grad_norm": 0.4315635027693341,
      "learning_rate": 1.0267140014082266e-06,
      "loss": 0.0055,
      "step": 13424
    },
    {
      "epoch": 8.589251439539348,
      "grad_norm": 0.5133391082261993,
      "learning_rate": 1.0257996217627241e-06,
      "loss": 0.0068,
      "step": 13425
    },
    {
      "epoch": 8.589891234804863,
      "grad_norm": 0.5415039364770708,
      "learning_rate": 1.0248856274534224e-06,
      "loss": 0.0085,
      "step": 13426
    },
    {
      "epoch": 8.590531030070377,
      "grad_norm": 0.513713916764342,
      "learning_rate": 1.0239720185195668e-06,
      "loss": 0.0061,
      "step": 13427
    },
    {
      "epoch": 8.591170825335892,
      "grad_norm": 0.4103460093871945,
      "learning_rate": 1.0230587950003845e-06,
      "loss": 0.0049,
      "step": 13428
    },
    {
      "epoch": 8.591810620601407,
      "grad_norm": 0.5487857278881274,
      "learning_rate": 1.0221459569350867e-06,
      "loss": 0.0064,
      "step": 13429
    },
    {
      "epoch": 8.592450415866923,
      "grad_norm": 0.3938004045043837,
      "learning_rate": 1.021233504362874e-06,
      "loss": 0.005,
      "step": 13430
    },
    {
      "epoch": 8.593090211132438,
      "grad_norm": 0.5031368612238772,
      "learning_rate": 1.020321437322922e-06,
      "loss": 0.0054,
      "step": 13431
    },
    {
      "epoch": 8.593730006397953,
      "grad_norm": 0.39479415959954495,
      "learning_rate": 1.0194097558543948e-06,
      "loss": 0.0048,
      "step": 13432
    },
    {
      "epoch": 8.594369801663468,
      "grad_norm": 0.47476156434230143,
      "learning_rate": 1.0184984599964342e-06,
      "loss": 0.005,
      "step": 13433
    },
    {
      "epoch": 8.595009596928982,
      "grad_norm": 0.41258956377247513,
      "learning_rate": 1.017587549788176e-06,
      "loss": 0.005,
      "step": 13434
    },
    {
      "epoch": 8.595649392194497,
      "grad_norm": 0.4182895086548728,
      "learning_rate": 1.0166770252687296e-06,
      "loss": 0.0054,
      "step": 13435
    },
    {
      "epoch": 8.596289187460012,
      "grad_norm": 0.5543987044522823,
      "learning_rate": 1.0157668864771896e-06,
      "loss": 0.0055,
      "step": 13436
    },
    {
      "epoch": 8.596928982725528,
      "grad_norm": 0.49526104009210636,
      "learning_rate": 1.0148571334526392e-06,
      "loss": 0.0057,
      "step": 13437
    },
    {
      "epoch": 8.597568777991043,
      "grad_norm": 0.5054929600066823,
      "learning_rate": 1.0139477662341413e-06,
      "loss": 0.0059,
      "step": 13438
    },
    {
      "epoch": 8.598208573256558,
      "grad_norm": 0.4285907905875033,
      "learning_rate": 1.0130387848607404e-06,
      "loss": 0.0057,
      "step": 13439
    },
    {
      "epoch": 8.598848368522074,
      "grad_norm": 0.4566036104460776,
      "learning_rate": 1.0121301893714662e-06,
      "loss": 0.0062,
      "step": 13440
    },
    {
      "epoch": 8.599488163787587,
      "grad_norm": 0.33231598052553246,
      "learning_rate": 1.0112219798053347e-06,
      "loss": 0.0038,
      "step": 13441
    },
    {
      "epoch": 8.600127959053102,
      "grad_norm": 0.4035568782323153,
      "learning_rate": 1.0103141562013418e-06,
      "loss": 0.0053,
      "step": 13442
    },
    {
      "epoch": 8.600767754318618,
      "grad_norm": 0.4677725588507669,
      "learning_rate": 1.009406718598468e-06,
      "loss": 0.0054,
      "step": 13443
    },
    {
      "epoch": 8.601407549584133,
      "grad_norm": 0.4322640059892681,
      "learning_rate": 1.0084996670356772e-06,
      "loss": 0.0059,
      "step": 13444
    },
    {
      "epoch": 8.602047344849648,
      "grad_norm": 0.4271439271272157,
      "learning_rate": 1.0075930015519153e-06,
      "loss": 0.0051,
      "step": 13445
    },
    {
      "epoch": 8.602687140115163,
      "grad_norm": 0.4526144866334064,
      "learning_rate": 1.0066867221861109e-06,
      "loss": 0.0062,
      "step": 13446
    },
    {
      "epoch": 8.603326935380679,
      "grad_norm": 0.41443170905068966,
      "learning_rate": 1.0057808289771843e-06,
      "loss": 0.0054,
      "step": 13447
    },
    {
      "epoch": 8.603966730646194,
      "grad_norm": 0.4116131101604089,
      "learning_rate": 1.0048753219640284e-06,
      "loss": 0.0058,
      "step": 13448
    },
    {
      "epoch": 8.604606525911707,
      "grad_norm": 0.44345995607119876,
      "learning_rate": 1.0039702011855257e-06,
      "loss": 0.0064,
      "step": 13449
    },
    {
      "epoch": 8.605246321177223,
      "grad_norm": 0.35885038427626664,
      "learning_rate": 1.0030654666805373e-06,
      "loss": 0.0048,
      "step": 13450
    },
    {
      "epoch": 8.605886116442738,
      "grad_norm": 0.4953140817047235,
      "learning_rate": 1.0021611184879165e-06,
      "loss": 0.0069,
      "step": 13451
    },
    {
      "epoch": 8.606525911708253,
      "grad_norm": 0.35521226646189424,
      "learning_rate": 1.0012571566464901e-06,
      "loss": 0.0046,
      "step": 13452
    },
    {
      "epoch": 8.607165706973769,
      "grad_norm": 0.39999411402087076,
      "learning_rate": 1.0003535811950726e-06,
      "loss": 0.0058,
      "step": 13453
    },
    {
      "epoch": 8.607805502239284,
      "grad_norm": 0.4045831118493177,
      "learning_rate": 9.994503921724652e-07,
      "loss": 0.0046,
      "step": 13454
    },
    {
      "epoch": 8.6084452975048,
      "grad_norm": 0.40702307056476883,
      "learning_rate": 9.985475896174468e-07,
      "loss": 0.0047,
      "step": 13455
    },
    {
      "epoch": 8.609085092770314,
      "grad_norm": 0.41687790252209134,
      "learning_rate": 9.97645173568783e-07,
      "loss": 0.0057,
      "step": 13456
    },
    {
      "epoch": 8.609724888035828,
      "grad_norm": 0.48993462994965203,
      "learning_rate": 9.967431440652208e-07,
      "loss": 0.0061,
      "step": 13457
    },
    {
      "epoch": 8.610364683301343,
      "grad_norm": 0.40976917799260004,
      "learning_rate": 9.958415011454935e-07,
      "loss": 0.0048,
      "step": 13458
    },
    {
      "epoch": 8.611004478566858,
      "grad_norm": 0.3977143656815109,
      "learning_rate": 9.949402448483137e-07,
      "loss": 0.0046,
      "step": 13459
    },
    {
      "epoch": 8.611644273832374,
      "grad_norm": 0.4985807053271524,
      "learning_rate": 9.9403937521238e-07,
      "loss": 0.007,
      "step": 13460
    },
    {
      "epoch": 8.612284069097889,
      "grad_norm": 0.46408138899126966,
      "learning_rate": 9.931388922763774e-07,
      "loss": 0.0063,
      "step": 13461
    },
    {
      "epoch": 8.612923864363404,
      "grad_norm": 0.5521004211595109,
      "learning_rate": 9.922387960789682e-07,
      "loss": 0.0072,
      "step": 13462
    },
    {
      "epoch": 8.61356365962892,
      "grad_norm": 0.4261395836521237,
      "learning_rate": 9.91339086658799e-07,
      "loss": 0.0053,
      "step": 13463
    },
    {
      "epoch": 8.614203454894433,
      "grad_norm": 0.4420382598621068,
      "learning_rate": 9.904397640545071e-07,
      "loss": 0.0058,
      "step": 13464
    },
    {
      "epoch": 8.614843250159948,
      "grad_norm": 0.481586464922376,
      "learning_rate": 9.895408283047047e-07,
      "loss": 0.006,
      "step": 13465
    },
    {
      "epoch": 8.615483045425464,
      "grad_norm": 0.5677335056526233,
      "learning_rate": 9.886422794479888e-07,
      "loss": 0.007,
      "step": 13466
    },
    {
      "epoch": 8.616122840690979,
      "grad_norm": 0.4058112507594972,
      "learning_rate": 9.877441175229462e-07,
      "loss": 0.0045,
      "step": 13467
    },
    {
      "epoch": 8.616762635956494,
      "grad_norm": 0.5108247008984035,
      "learning_rate": 9.868463425681384e-07,
      "loss": 0.0065,
      "step": 13468
    },
    {
      "epoch": 8.61740243122201,
      "grad_norm": 0.44355579099972936,
      "learning_rate": 9.859489546221157e-07,
      "loss": 0.0064,
      "step": 13469
    },
    {
      "epoch": 8.618042226487525,
      "grad_norm": 0.4420383660631518,
      "learning_rate": 9.850519537234093e-07,
      "loss": 0.0055,
      "step": 13470
    },
    {
      "epoch": 8.618682021753038,
      "grad_norm": 0.4414102231717232,
      "learning_rate": 9.841553399105364e-07,
      "loss": 0.0061,
      "step": 13471
    },
    {
      "epoch": 8.619321817018553,
      "grad_norm": 0.40414481808899494,
      "learning_rate": 9.832591132219959e-07,
      "loss": 0.0059,
      "step": 13472
    },
    {
      "epoch": 8.619961612284069,
      "grad_norm": 0.4914007854219369,
      "learning_rate": 9.823632736962695e-07,
      "loss": 0.0071,
      "step": 13473
    },
    {
      "epoch": 8.620601407549584,
      "grad_norm": 0.35899127362518873,
      "learning_rate": 9.814678213718222e-07,
      "loss": 0.0047,
      "step": 13474
    },
    {
      "epoch": 8.6212412028151,
      "grad_norm": 0.5663369219078281,
      "learning_rate": 9.805727562871048e-07,
      "loss": 0.0071,
      "step": 13475
    },
    {
      "epoch": 8.621880998080615,
      "grad_norm": 0.4844246249326885,
      "learning_rate": 9.79678078480547e-07,
      "loss": 0.0063,
      "step": 13476
    },
    {
      "epoch": 8.62252079334613,
      "grad_norm": 0.36903047139756684,
      "learning_rate": 9.787837879905682e-07,
      "loss": 0.0049,
      "step": 13477
    },
    {
      "epoch": 8.623160588611643,
      "grad_norm": 0.3646112561312084,
      "learning_rate": 9.778898848555663e-07,
      "loss": 0.0048,
      "step": 13478
    },
    {
      "epoch": 8.623800383877159,
      "grad_norm": 0.48004266684442576,
      "learning_rate": 9.769963691139227e-07,
      "loss": 0.0058,
      "step": 13479
    },
    {
      "epoch": 8.624440179142674,
      "grad_norm": 0.3958276457448452,
      "learning_rate": 9.761032408040038e-07,
      "loss": 0.0051,
      "step": 13480
    },
    {
      "epoch": 8.62507997440819,
      "grad_norm": 0.3659637944613869,
      "learning_rate": 9.75210499964161e-07,
      "loss": 0.0053,
      "step": 13481
    },
    {
      "epoch": 8.625719769673704,
      "grad_norm": 0.4116905736813109,
      "learning_rate": 9.743181466327244e-07,
      "loss": 0.0063,
      "step": 13482
    },
    {
      "epoch": 8.62635956493922,
      "grad_norm": 0.40015498581567877,
      "learning_rate": 9.734261808480106e-07,
      "loss": 0.0045,
      "step": 13483
    },
    {
      "epoch": 8.626999360204735,
      "grad_norm": 0.3877341718610854,
      "learning_rate": 9.725346026483207e-07,
      "loss": 0.0048,
      "step": 13484
    },
    {
      "epoch": 8.62763915547025,
      "grad_norm": 0.542929421214227,
      "learning_rate": 9.716434120719365e-07,
      "loss": 0.0085,
      "step": 13485
    },
    {
      "epoch": 8.628278950735764,
      "grad_norm": 0.44267064004276774,
      "learning_rate": 9.70752609157124e-07,
      "loss": 0.0063,
      "step": 13486
    },
    {
      "epoch": 8.628918746001279,
      "grad_norm": 0.45980050270872747,
      "learning_rate": 9.698621939421327e-07,
      "loss": 0.006,
      "step": 13487
    },
    {
      "epoch": 8.629558541266794,
      "grad_norm": 0.4674600482571774,
      "learning_rate": 9.689721664651951e-07,
      "loss": 0.0067,
      "step": 13488
    },
    {
      "epoch": 8.63019833653231,
      "grad_norm": 0.4055308295010369,
      "learning_rate": 9.680825267645288e-07,
      "loss": 0.0064,
      "step": 13489
    },
    {
      "epoch": 8.630838131797825,
      "grad_norm": 0.5111777556109482,
      "learning_rate": 9.671932748783297e-07,
      "loss": 0.0061,
      "step": 13490
    },
    {
      "epoch": 8.63147792706334,
      "grad_norm": 0.5589444859894962,
      "learning_rate": 9.66304410844785e-07,
      "loss": 0.0073,
      "step": 13491
    },
    {
      "epoch": 8.632117722328855,
      "grad_norm": 0.3817802664190503,
      "learning_rate": 9.654159347020598e-07,
      "loss": 0.0049,
      "step": 13492
    },
    {
      "epoch": 8.63275751759437,
      "grad_norm": 0.45164376458380734,
      "learning_rate": 9.645278464883001e-07,
      "loss": 0.0055,
      "step": 13493
    },
    {
      "epoch": 8.633397312859884,
      "grad_norm": 0.4571091407247332,
      "learning_rate": 9.636401462416444e-07,
      "loss": 0.0068,
      "step": 13494
    },
    {
      "epoch": 8.6340371081254,
      "grad_norm": 0.4934124173400625,
      "learning_rate": 9.627528340002057e-07,
      "loss": 0.0063,
      "step": 13495
    },
    {
      "epoch": 8.634676903390915,
      "grad_norm": 0.3600035210955749,
      "learning_rate": 9.618659098020832e-07,
      "loss": 0.0042,
      "step": 13496
    },
    {
      "epoch": 8.63531669865643,
      "grad_norm": 0.36543296634442757,
      "learning_rate": 9.609793736853634e-07,
      "loss": 0.0046,
      "step": 13497
    },
    {
      "epoch": 8.635956493921945,
      "grad_norm": 0.5464396791578586,
      "learning_rate": 9.60093225688109e-07,
      "loss": 0.0066,
      "step": 13498
    },
    {
      "epoch": 8.63659628918746,
      "grad_norm": 0.38862161874769247,
      "learning_rate": 9.592074658483708e-07,
      "loss": 0.0046,
      "step": 13499
    },
    {
      "epoch": 8.637236084452976,
      "grad_norm": 0.5093508510141714,
      "learning_rate": 9.583220942041804e-07,
      "loss": 0.0049,
      "step": 13500
    },
    {
      "epoch": 8.63787587971849,
      "grad_norm": 0.46052978104642084,
      "learning_rate": 9.574371107935564e-07,
      "loss": 0.0056,
      "step": 13501
    },
    {
      "epoch": 8.638515674984005,
      "grad_norm": 0.38170005797502066,
      "learning_rate": 9.565525156544986e-07,
      "loss": 0.0046,
      "step": 13502
    },
    {
      "epoch": 8.63915547024952,
      "grad_norm": 0.41661613682583315,
      "learning_rate": 9.556683088249875e-07,
      "loss": 0.0045,
      "step": 13503
    },
    {
      "epoch": 8.639795265515035,
      "grad_norm": 0.49235721755122547,
      "learning_rate": 9.547844903429914e-07,
      "loss": 0.0063,
      "step": 13504
    },
    {
      "epoch": 8.64043506078055,
      "grad_norm": 0.4204015837233706,
      "learning_rate": 9.539010602464594e-07,
      "loss": 0.0053,
      "step": 13505
    },
    {
      "epoch": 8.641074856046066,
      "grad_norm": 0.43962643395625617,
      "learning_rate": 9.530180185733217e-07,
      "loss": 0.0057,
      "step": 13506
    },
    {
      "epoch": 8.641714651311581,
      "grad_norm": 0.34584146792702536,
      "learning_rate": 9.521353653614984e-07,
      "loss": 0.0034,
      "step": 13507
    },
    {
      "epoch": 8.642354446577095,
      "grad_norm": 0.39694835161879305,
      "learning_rate": 9.512531006488878e-07,
      "loss": 0.0042,
      "step": 13508
    },
    {
      "epoch": 8.64299424184261,
      "grad_norm": 0.38790606852939685,
      "learning_rate": 9.503712244733732e-07,
      "loss": 0.0051,
      "step": 13509
    },
    {
      "epoch": 8.643634037108125,
      "grad_norm": 0.503305591102613,
      "learning_rate": 9.494897368728184e-07,
      "loss": 0.0064,
      "step": 13510
    },
    {
      "epoch": 8.64427383237364,
      "grad_norm": 0.3501758888233701,
      "learning_rate": 9.486086378850756e-07,
      "loss": 0.0046,
      "step": 13511
    },
    {
      "epoch": 8.644913627639156,
      "grad_norm": 0.401650241955172,
      "learning_rate": 9.477279275479767e-07,
      "loss": 0.0041,
      "step": 13512
    },
    {
      "epoch": 8.64555342290467,
      "grad_norm": 0.5531459754124571,
      "learning_rate": 9.46847605899337e-07,
      "loss": 0.0063,
      "step": 13513
    },
    {
      "epoch": 8.646193218170186,
      "grad_norm": 0.4785569331585113,
      "learning_rate": 9.459676729769584e-07,
      "loss": 0.0037,
      "step": 13514
    },
    {
      "epoch": 8.6468330134357,
      "grad_norm": 0.5194621449390032,
      "learning_rate": 9.45088128818622e-07,
      "loss": 0.0058,
      "step": 13515
    },
    {
      "epoch": 8.647472808701215,
      "grad_norm": 0.36861005064885455,
      "learning_rate": 9.442089734620929e-07,
      "loss": 0.0038,
      "step": 13516
    },
    {
      "epoch": 8.64811260396673,
      "grad_norm": 0.40041624517891644,
      "learning_rate": 9.433302069451222e-07,
      "loss": 0.0052,
      "step": 13517
    },
    {
      "epoch": 8.648752399232245,
      "grad_norm": 0.46770477745383054,
      "learning_rate": 9.42451829305443e-07,
      "loss": 0.0055,
      "step": 13518
    },
    {
      "epoch": 8.64939219449776,
      "grad_norm": 0.43433767588119904,
      "learning_rate": 9.415738405807684e-07,
      "loss": 0.0054,
      "step": 13519
    },
    {
      "epoch": 8.650031989763276,
      "grad_norm": 0.352212375281801,
      "learning_rate": 9.406962408087983e-07,
      "loss": 0.0041,
      "step": 13520
    },
    {
      "epoch": 8.650671785028791,
      "grad_norm": 0.47866987805115646,
      "learning_rate": 9.398190300272181e-07,
      "loss": 0.0049,
      "step": 13521
    },
    {
      "epoch": 8.651311580294307,
      "grad_norm": 0.480751609036714,
      "learning_rate": 9.38942208273692e-07,
      "loss": 0.0063,
      "step": 13522
    },
    {
      "epoch": 8.65195137555982,
      "grad_norm": 0.38384176017507315,
      "learning_rate": 9.380657755858669e-07,
      "loss": 0.0038,
      "step": 13523
    },
    {
      "epoch": 8.652591170825335,
      "grad_norm": 0.6097029550863796,
      "learning_rate": 9.3718973200138e-07,
      "loss": 0.0093,
      "step": 13524
    },
    {
      "epoch": 8.65323096609085,
      "grad_norm": 0.5325283983158665,
      "learning_rate": 9.363140775578439e-07,
      "loss": 0.0071,
      "step": 13525
    },
    {
      "epoch": 8.653870761356366,
      "grad_norm": 0.3892862917699612,
      "learning_rate": 9.354388122928582e-07,
      "loss": 0.0047,
      "step": 13526
    },
    {
      "epoch": 8.654510556621881,
      "grad_norm": 0.609244245737372,
      "learning_rate": 9.34563936244004e-07,
      "loss": 0.0064,
      "step": 13527
    },
    {
      "epoch": 8.655150351887396,
      "grad_norm": 0.32725050722682725,
      "learning_rate": 9.336894494488502e-07,
      "loss": 0.0035,
      "step": 13528
    },
    {
      "epoch": 8.655790147152912,
      "grad_norm": 0.46100734442302843,
      "learning_rate": 9.328153519449435e-07,
      "loss": 0.004,
      "step": 13529
    },
    {
      "epoch": 8.656429942418427,
      "grad_norm": 0.35694153170629933,
      "learning_rate": 9.319416437698137e-07,
      "loss": 0.0042,
      "step": 13530
    },
    {
      "epoch": 8.65706973768394,
      "grad_norm": 0.4065451054880711,
      "learning_rate": 9.31068324960982e-07,
      "loss": 0.0046,
      "step": 13531
    },
    {
      "epoch": 8.657709532949456,
      "grad_norm": 0.5176278704835998,
      "learning_rate": 9.301953955559428e-07,
      "loss": 0.0064,
      "step": 13532
    },
    {
      "epoch": 8.658349328214971,
      "grad_norm": 0.46258376756871405,
      "learning_rate": 9.293228555921807e-07,
      "loss": 0.0063,
      "step": 13533
    },
    {
      "epoch": 8.658989123480486,
      "grad_norm": 0.4686100995706043,
      "learning_rate": 9.28450705107159e-07,
      "loss": 0.0067,
      "step": 13534
    },
    {
      "epoch": 8.659628918746002,
      "grad_norm": 0.47091260877733576,
      "learning_rate": 9.275789441383265e-07,
      "loss": 0.0043,
      "step": 13535
    },
    {
      "epoch": 8.660268714011517,
      "grad_norm": 0.41990943203945275,
      "learning_rate": 9.267075727231156e-07,
      "loss": 0.005,
      "step": 13536
    },
    {
      "epoch": 8.660908509277032,
      "grad_norm": 0.41784731772974465,
      "learning_rate": 9.258365908989409e-07,
      "loss": 0.0056,
      "step": 13537
    },
    {
      "epoch": 8.661548304542546,
      "grad_norm": 0.38995128086755687,
      "learning_rate": 9.249659987032022e-07,
      "loss": 0.0055,
      "step": 13538
    },
    {
      "epoch": 8.662188099808061,
      "grad_norm": 0.4511079812448764,
      "learning_rate": 9.240957961732799e-07,
      "loss": 0.0065,
      "step": 13539
    },
    {
      "epoch": 8.662827895073576,
      "grad_norm": 0.4136162703076829,
      "learning_rate": 9.232259833465385e-07,
      "loss": 0.0051,
      "step": 13540
    },
    {
      "epoch": 8.663467690339091,
      "grad_norm": 0.40898741894373886,
      "learning_rate": 9.22356560260328e-07,
      "loss": 0.0056,
      "step": 13541
    },
    {
      "epoch": 8.664107485604607,
      "grad_norm": 0.4249425013125344,
      "learning_rate": 9.214875269519785e-07,
      "loss": 0.0049,
      "step": 13542
    },
    {
      "epoch": 8.664747280870122,
      "grad_norm": 0.47523712001972457,
      "learning_rate": 9.206188834588048e-07,
      "loss": 0.0057,
      "step": 13543
    },
    {
      "epoch": 8.665387076135637,
      "grad_norm": 0.40875580602020006,
      "learning_rate": 9.197506298181058e-07,
      "loss": 0.0059,
      "step": 13544
    },
    {
      "epoch": 8.66602687140115,
      "grad_norm": 0.428119102715801,
      "learning_rate": 9.188827660671629e-07,
      "loss": 0.0059,
      "step": 13545
    },
    {
      "epoch": 8.666666666666666,
      "grad_norm": 0.46821065933499123,
      "learning_rate": 9.180152922432395e-07,
      "loss": 0.0068,
      "step": 13546
    },
    {
      "epoch": 8.667306461932181,
      "grad_norm": 0.3782250905882895,
      "learning_rate": 9.171482083835847e-07,
      "loss": 0.0045,
      "step": 13547
    },
    {
      "epoch": 8.667946257197697,
      "grad_norm": 0.3686459985276722,
      "learning_rate": 9.162815145254278e-07,
      "loss": 0.0042,
      "step": 13548
    },
    {
      "epoch": 8.668586052463212,
      "grad_norm": 0.4502122770689791,
      "learning_rate": 9.154152107059843e-07,
      "loss": 0.0068,
      "step": 13549
    },
    {
      "epoch": 8.669225847728727,
      "grad_norm": 0.43136993744156005,
      "learning_rate": 9.145492969624503e-07,
      "loss": 0.004,
      "step": 13550
    },
    {
      "epoch": 8.669865642994242,
      "grad_norm": 0.413349776410738,
      "learning_rate": 9.136837733320092e-07,
      "loss": 0.0053,
      "step": 13551
    },
    {
      "epoch": 8.670505438259756,
      "grad_norm": 0.44115431618206996,
      "learning_rate": 9.128186398518235e-07,
      "loss": 0.0065,
      "step": 13552
    },
    {
      "epoch": 8.671145233525271,
      "grad_norm": 0.4986732587479403,
      "learning_rate": 9.11953896559038e-07,
      "loss": 0.0056,
      "step": 13553
    },
    {
      "epoch": 8.671785028790786,
      "grad_norm": 0.529823319032126,
      "learning_rate": 9.110895434907885e-07,
      "loss": 0.0073,
      "step": 13554
    },
    {
      "epoch": 8.672424824056302,
      "grad_norm": 0.4262355478271545,
      "learning_rate": 9.102255806841853e-07,
      "loss": 0.0042,
      "step": 13555
    },
    {
      "epoch": 8.673064619321817,
      "grad_norm": 0.5014395555955534,
      "learning_rate": 9.093620081763266e-07,
      "loss": 0.0063,
      "step": 13556
    },
    {
      "epoch": 8.673704414587332,
      "grad_norm": 0.403130495109541,
      "learning_rate": 9.084988260042904e-07,
      "loss": 0.005,
      "step": 13557
    },
    {
      "epoch": 8.674344209852848,
      "grad_norm": 0.47566778414332744,
      "learning_rate": 9.076360342051449e-07,
      "loss": 0.0057,
      "step": 13558
    },
    {
      "epoch": 8.674984005118363,
      "grad_norm": 0.41537591790152734,
      "learning_rate": 9.067736328159327e-07,
      "loss": 0.0052,
      "step": 13559
    },
    {
      "epoch": 8.675623800383876,
      "grad_norm": 0.3716377314168491,
      "learning_rate": 9.059116218736841e-07,
      "loss": 0.0044,
      "step": 13560
    },
    {
      "epoch": 8.676263595649392,
      "grad_norm": 0.48665197248574804,
      "learning_rate": 9.05050001415415e-07,
      "loss": 0.0078,
      "step": 13561
    },
    {
      "epoch": 8.676903390914907,
      "grad_norm": 0.37377039230596976,
      "learning_rate": 9.041887714781195e-07,
      "loss": 0.0044,
      "step": 13562
    },
    {
      "epoch": 8.677543186180422,
      "grad_norm": 0.4198714644241347,
      "learning_rate": 9.033279320987786e-07,
      "loss": 0.0058,
      "step": 13563
    },
    {
      "epoch": 8.678182981445937,
      "grad_norm": 0.47682314351609517,
      "learning_rate": 9.024674833143543e-07,
      "loss": 0.0064,
      "step": 13564
    },
    {
      "epoch": 8.678822776711453,
      "grad_norm": 0.4020574860143794,
      "learning_rate": 9.016074251617935e-07,
      "loss": 0.0049,
      "step": 13565
    },
    {
      "epoch": 8.679462571976968,
      "grad_norm": 0.46262800474590016,
      "learning_rate": 9.007477576780244e-07,
      "loss": 0.0058,
      "step": 13566
    },
    {
      "epoch": 8.680102367242483,
      "grad_norm": 0.4759928500955227,
      "learning_rate": 8.998884808999586e-07,
      "loss": 0.0061,
      "step": 13567
    },
    {
      "epoch": 8.680742162507997,
      "grad_norm": 0.4542810927133547,
      "learning_rate": 8.990295948644956e-07,
      "loss": 0.0068,
      "step": 13568
    },
    {
      "epoch": 8.681381957773512,
      "grad_norm": 0.40315776460389063,
      "learning_rate": 8.981710996085124e-07,
      "loss": 0.0055,
      "step": 13569
    },
    {
      "epoch": 8.682021753039027,
      "grad_norm": 0.389356551498713,
      "learning_rate": 8.973129951688697e-07,
      "loss": 0.0049,
      "step": 13570
    },
    {
      "epoch": 8.682661548304543,
      "grad_norm": 0.38045256500861596,
      "learning_rate": 8.964552815824168e-07,
      "loss": 0.0043,
      "step": 13571
    },
    {
      "epoch": 8.683301343570058,
      "grad_norm": 0.4699939450326696,
      "learning_rate": 8.955979588859786e-07,
      "loss": 0.0049,
      "step": 13572
    },
    {
      "epoch": 8.683941138835573,
      "grad_norm": 0.6358451919786396,
      "learning_rate": 8.947410271163681e-07,
      "loss": 0.007,
      "step": 13573
    },
    {
      "epoch": 8.684580934101088,
      "grad_norm": 0.5791645637998291,
      "learning_rate": 8.938844863103824e-07,
      "loss": 0.007,
      "step": 13574
    },
    {
      "epoch": 8.685220729366602,
      "grad_norm": 0.4079981223543104,
      "learning_rate": 8.930283365047976e-07,
      "loss": 0.0051,
      "step": 13575
    },
    {
      "epoch": 8.685860524632117,
      "grad_norm": 0.8446846097504761,
      "learning_rate": 8.921725777363765e-07,
      "loss": 0.0047,
      "step": 13576
    },
    {
      "epoch": 8.686500319897632,
      "grad_norm": 0.5378396091465141,
      "learning_rate": 8.913172100418643e-07,
      "loss": 0.0069,
      "step": 13577
    },
    {
      "epoch": 8.687140115163148,
      "grad_norm": 0.5584661539173299,
      "learning_rate": 8.904622334579871e-07,
      "loss": 0.0078,
      "step": 13578
    },
    {
      "epoch": 8.687779910428663,
      "grad_norm": 0.40049260344586235,
      "learning_rate": 8.896076480214566e-07,
      "loss": 0.0062,
      "step": 13579
    },
    {
      "epoch": 8.688419705694178,
      "grad_norm": 0.44544525346740255,
      "learning_rate": 8.887534537689668e-07,
      "loss": 0.0054,
      "step": 13580
    },
    {
      "epoch": 8.689059500959694,
      "grad_norm": 0.3758740235410257,
      "learning_rate": 8.878996507371973e-07,
      "loss": 0.0052,
      "step": 13581
    },
    {
      "epoch": 8.689699296225207,
      "grad_norm": 0.37419412803993307,
      "learning_rate": 8.870462389628076e-07,
      "loss": 0.0054,
      "step": 13582
    },
    {
      "epoch": 8.690339091490722,
      "grad_norm": 0.44439131217871997,
      "learning_rate": 8.861932184824395e-07,
      "loss": 0.0059,
      "step": 13583
    },
    {
      "epoch": 8.690978886756238,
      "grad_norm": 0.47251424177458085,
      "learning_rate": 8.853405893327249e-07,
      "loss": 0.0067,
      "step": 13584
    },
    {
      "epoch": 8.691618682021753,
      "grad_norm": 0.45078451158524707,
      "learning_rate": 8.844883515502711e-07,
      "loss": 0.0052,
      "step": 13585
    },
    {
      "epoch": 8.692258477287268,
      "grad_norm": 0.6278787019713187,
      "learning_rate": 8.836365051716723e-07,
      "loss": 0.0068,
      "step": 13586
    },
    {
      "epoch": 8.692898272552783,
      "grad_norm": 0.40059571723236165,
      "learning_rate": 8.827850502335034e-07,
      "loss": 0.0052,
      "step": 13587
    },
    {
      "epoch": 8.693538067818299,
      "grad_norm": 0.43696278294223295,
      "learning_rate": 8.819339867723287e-07,
      "loss": 0.0056,
      "step": 13588
    },
    {
      "epoch": 8.694177863083812,
      "grad_norm": 0.4879187314283488,
      "learning_rate": 8.810833148246877e-07,
      "loss": 0.0057,
      "step": 13589
    },
    {
      "epoch": 8.694817658349328,
      "grad_norm": 0.46299575428076845,
      "learning_rate": 8.802330344271071e-07,
      "loss": 0.005,
      "step": 13590
    },
    {
      "epoch": 8.695457453614843,
      "grad_norm": 0.45368837623326563,
      "learning_rate": 8.793831456160984e-07,
      "loss": 0.006,
      "step": 13591
    },
    {
      "epoch": 8.696097248880358,
      "grad_norm": 0.36886700617758894,
      "learning_rate": 8.785336484281537e-07,
      "loss": 0.004,
      "step": 13592
    },
    {
      "epoch": 8.696737044145873,
      "grad_norm": 0.4702127224341265,
      "learning_rate": 8.776845428997471e-07,
      "loss": 0.0057,
      "step": 13593
    },
    {
      "epoch": 8.697376839411389,
      "grad_norm": 0.4244650664353884,
      "learning_rate": 8.768358290673407e-07,
      "loss": 0.0055,
      "step": 13594
    },
    {
      "epoch": 8.698016634676904,
      "grad_norm": 0.5256464973096759,
      "learning_rate": 8.759875069673729e-07,
      "loss": 0.0066,
      "step": 13595
    },
    {
      "epoch": 8.69865642994242,
      "grad_norm": 0.3753946428684555,
      "learning_rate": 8.751395766362725e-07,
      "loss": 0.005,
      "step": 13596
    },
    {
      "epoch": 8.699296225207933,
      "grad_norm": 0.5118939678452534,
      "learning_rate": 8.742920381104447e-07,
      "loss": 0.0071,
      "step": 13597
    },
    {
      "epoch": 8.699936020473448,
      "grad_norm": 0.4580945961053416,
      "learning_rate": 8.73444891426285e-07,
      "loss": 0.0055,
      "step": 13598
    },
    {
      "epoch": 8.700575815738963,
      "grad_norm": 0.344491799724246,
      "learning_rate": 8.725981366201675e-07,
      "loss": 0.0041,
      "step": 13599
    },
    {
      "epoch": 8.701215611004478,
      "grad_norm": 0.4596519413301914,
      "learning_rate": 8.717517737284475e-07,
      "loss": 0.005,
      "step": 13600
    },
    {
      "epoch": 8.701855406269994,
      "grad_norm": 0.48732274070742204,
      "learning_rate": 8.709058027874695e-07,
      "loss": 0.0061,
      "step": 13601
    },
    {
      "epoch": 8.702495201535509,
      "grad_norm": 0.4035544120046555,
      "learning_rate": 8.700602238335576e-07,
      "loss": 0.0056,
      "step": 13602
    },
    {
      "epoch": 8.703134996801024,
      "grad_norm": 0.4288851734327541,
      "learning_rate": 8.692150369030195e-07,
      "loss": 0.0055,
      "step": 13603
    },
    {
      "epoch": 8.70377479206654,
      "grad_norm": 0.5225682842330629,
      "learning_rate": 8.683702420321427e-07,
      "loss": 0.006,
      "step": 13604
    },
    {
      "epoch": 8.704414587332053,
      "grad_norm": 0.4466159424530086,
      "learning_rate": 8.67525839257205e-07,
      "loss": 0.0065,
      "step": 13605
    },
    {
      "epoch": 8.705054382597568,
      "grad_norm": 0.42158163569906154,
      "learning_rate": 8.666818286144629e-07,
      "loss": 0.0062,
      "step": 13606
    },
    {
      "epoch": 8.705694177863084,
      "grad_norm": 0.46620540950074124,
      "learning_rate": 8.658382101401564e-07,
      "loss": 0.0061,
      "step": 13607
    },
    {
      "epoch": 8.706333973128599,
      "grad_norm": 0.3528563225974131,
      "learning_rate": 8.649949838705086e-07,
      "loss": 0.0044,
      "step": 13608
    },
    {
      "epoch": 8.706973768394114,
      "grad_norm": 0.43747567251971325,
      "learning_rate": 8.641521498417261e-07,
      "loss": 0.0058,
      "step": 13609
    },
    {
      "epoch": 8.70761356365963,
      "grad_norm": 0.43866283419071,
      "learning_rate": 8.633097080899977e-07,
      "loss": 0.0045,
      "step": 13610
    },
    {
      "epoch": 8.708253358925145,
      "grad_norm": 0.4250395869077879,
      "learning_rate": 8.624676586514991e-07,
      "loss": 0.0054,
      "step": 13611
    },
    {
      "epoch": 8.708893154190658,
      "grad_norm": 0.4901406442742908,
      "learning_rate": 8.616260015623845e-07,
      "loss": 0.0044,
      "step": 13612
    },
    {
      "epoch": 8.709532949456174,
      "grad_norm": 0.5163635935293487,
      "learning_rate": 8.607847368587929e-07,
      "loss": 0.0068,
      "step": 13613
    },
    {
      "epoch": 8.710172744721689,
      "grad_norm": 0.4898546450941461,
      "learning_rate": 8.599438645768465e-07,
      "loss": 0.0048,
      "step": 13614
    },
    {
      "epoch": 8.710812539987204,
      "grad_norm": 0.34435523192129514,
      "learning_rate": 8.591033847526531e-07,
      "loss": 0.0038,
      "step": 13615
    },
    {
      "epoch": 8.71145233525272,
      "grad_norm": 0.38802113056627136,
      "learning_rate": 8.582632974222993e-07,
      "loss": 0.0058,
      "step": 13616
    },
    {
      "epoch": 8.712092130518235,
      "grad_norm": 0.3324793819790707,
      "learning_rate": 8.574236026218574e-07,
      "loss": 0.0035,
      "step": 13617
    },
    {
      "epoch": 8.71273192578375,
      "grad_norm": 0.4165033590628269,
      "learning_rate": 8.565843003873831e-07,
      "loss": 0.0052,
      "step": 13618
    },
    {
      "epoch": 8.713371721049263,
      "grad_norm": 0.43046678465082516,
      "learning_rate": 8.55745390754914e-07,
      "loss": 0.0057,
      "step": 13619
    },
    {
      "epoch": 8.714011516314779,
      "grad_norm": 0.4507036645469782,
      "learning_rate": 8.549068737604704e-07,
      "loss": 0.0047,
      "step": 13620
    },
    {
      "epoch": 8.714651311580294,
      "grad_norm": 0.3961481570637186,
      "learning_rate": 8.540687494400591e-07,
      "loss": 0.004,
      "step": 13621
    },
    {
      "epoch": 8.71529110684581,
      "grad_norm": 0.5419024163051837,
      "learning_rate": 8.532310178296666e-07,
      "loss": 0.0066,
      "step": 13622
    },
    {
      "epoch": 8.715930902111324,
      "grad_norm": 0.4639523101033465,
      "learning_rate": 8.523936789652643e-07,
      "loss": 0.0068,
      "step": 13623
    },
    {
      "epoch": 8.71657069737684,
      "grad_norm": 0.479685867595414,
      "learning_rate": 8.515567328828033e-07,
      "loss": 0.0058,
      "step": 13624
    },
    {
      "epoch": 8.717210492642355,
      "grad_norm": 0.4260704577086642,
      "learning_rate": 8.50720179618223e-07,
      "loss": 0.0047,
      "step": 13625
    },
    {
      "epoch": 8.71785028790787,
      "grad_norm": 0.4944770253970087,
      "learning_rate": 8.498840192074431e-07,
      "loss": 0.0068,
      "step": 13626
    },
    {
      "epoch": 8.718490083173384,
      "grad_norm": 0.6740545111453384,
      "learning_rate": 8.490482516863651e-07,
      "loss": 0.0062,
      "step": 13627
    },
    {
      "epoch": 8.719129878438899,
      "grad_norm": 0.4768106472542135,
      "learning_rate": 8.482128770908782e-07,
      "loss": 0.006,
      "step": 13628
    },
    {
      "epoch": 8.719769673704414,
      "grad_norm": 0.4545146367380864,
      "learning_rate": 8.473778954568512e-07,
      "loss": 0.0055,
      "step": 13629
    },
    {
      "epoch": 8.72040946896993,
      "grad_norm": 0.3804994090254284,
      "learning_rate": 8.465433068201334e-07,
      "loss": 0.0047,
      "step": 13630
    },
    {
      "epoch": 8.721049264235445,
      "grad_norm": 0.3397109573665856,
      "learning_rate": 8.45709111216566e-07,
      "loss": 0.0041,
      "step": 13631
    },
    {
      "epoch": 8.72168905950096,
      "grad_norm": 0.5045707522851396,
      "learning_rate": 8.44875308681965e-07,
      "loss": 0.0056,
      "step": 13632
    },
    {
      "epoch": 8.722328854766475,
      "grad_norm": 0.44826775573975775,
      "learning_rate": 8.440418992521316e-07,
      "loss": 0.0054,
      "step": 13633
    },
    {
      "epoch": 8.722968650031989,
      "grad_norm": 0.44694003595546733,
      "learning_rate": 8.432088829628515e-07,
      "loss": 0.0049,
      "step": 13634
    },
    {
      "epoch": 8.723608445297504,
      "grad_norm": 0.40004062504876403,
      "learning_rate": 8.423762598498942e-07,
      "loss": 0.0053,
      "step": 13635
    },
    {
      "epoch": 8.72424824056302,
      "grad_norm": 0.39445812015010867,
      "learning_rate": 8.415440299490096e-07,
      "loss": 0.0055,
      "step": 13636
    },
    {
      "epoch": 8.724888035828535,
      "grad_norm": 0.4359141015359039,
      "learning_rate": 8.407121932959339e-07,
      "loss": 0.0045,
      "step": 13637
    },
    {
      "epoch": 8.72552783109405,
      "grad_norm": 0.4467734394657568,
      "learning_rate": 8.398807499263828e-07,
      "loss": 0.005,
      "step": 13638
    },
    {
      "epoch": 8.726167626359565,
      "grad_norm": 0.4778525823839304,
      "learning_rate": 8.390496998760578e-07,
      "loss": 0.0067,
      "step": 13639
    },
    {
      "epoch": 8.72680742162508,
      "grad_norm": 0.4360182562885361,
      "learning_rate": 8.382190431806414e-07,
      "loss": 0.0056,
      "step": 13640
    },
    {
      "epoch": 8.727447216890596,
      "grad_norm": 0.4613801376113197,
      "learning_rate": 8.373887798758029e-07,
      "loss": 0.0078,
      "step": 13641
    },
    {
      "epoch": 8.72808701215611,
      "grad_norm": 0.42958230852488133,
      "learning_rate": 8.365589099971915e-07,
      "loss": 0.0053,
      "step": 13642
    },
    {
      "epoch": 8.728726807421625,
      "grad_norm": 0.3465314857727834,
      "learning_rate": 8.3572943358044e-07,
      "loss": 0.0037,
      "step": 13643
    },
    {
      "epoch": 8.72936660268714,
      "grad_norm": 0.5439286427545711,
      "learning_rate": 8.349003506611619e-07,
      "loss": 0.0049,
      "step": 13644
    },
    {
      "epoch": 8.730006397952655,
      "grad_norm": 0.369326464460959,
      "learning_rate": 8.340716612749622e-07,
      "loss": 0.0041,
      "step": 13645
    },
    {
      "epoch": 8.73064619321817,
      "grad_norm": 0.6849664070276088,
      "learning_rate": 8.332433654574201e-07,
      "loss": 0.0074,
      "step": 13646
    },
    {
      "epoch": 8.731285988483686,
      "grad_norm": 0.4500379504418876,
      "learning_rate": 8.324154632440995e-07,
      "loss": 0.0048,
      "step": 13647
    },
    {
      "epoch": 8.731925783749201,
      "grad_norm": 0.43919980064207553,
      "learning_rate": 8.315879546705529e-07,
      "loss": 0.0062,
      "step": 13648
    },
    {
      "epoch": 8.732565579014715,
      "grad_norm": 0.48324845394190635,
      "learning_rate": 8.307608397723111e-07,
      "loss": 0.0054,
      "step": 13649
    },
    {
      "epoch": 8.73320537428023,
      "grad_norm": 0.4492422938641875,
      "learning_rate": 8.299341185848852e-07,
      "loss": 0.0043,
      "step": 13650
    },
    {
      "epoch": 8.733845169545745,
      "grad_norm": 0.4558820011211914,
      "learning_rate": 8.291077911437784e-07,
      "loss": 0.0056,
      "step": 13651
    },
    {
      "epoch": 8.73448496481126,
      "grad_norm": 0.5511910109120745,
      "learning_rate": 8.282818574844687e-07,
      "loss": 0.0073,
      "step": 13652
    },
    {
      "epoch": 8.735124760076776,
      "grad_norm": 0.4467657415751572,
      "learning_rate": 8.274563176424211e-07,
      "loss": 0.0054,
      "step": 13653
    },
    {
      "epoch": 8.735764555342291,
      "grad_norm": 0.4756028211304533,
      "learning_rate": 8.266311716530828e-07,
      "loss": 0.0064,
      "step": 13654
    },
    {
      "epoch": 8.736404350607806,
      "grad_norm": 0.4575913675501839,
      "learning_rate": 8.258064195518834e-07,
      "loss": 0.0051,
      "step": 13655
    },
    {
      "epoch": 8.73704414587332,
      "grad_norm": 0.36165214499222204,
      "learning_rate": 8.249820613742365e-07,
      "loss": 0.0043,
      "step": 13656
    },
    {
      "epoch": 8.737683941138835,
      "grad_norm": 0.4664866339541804,
      "learning_rate": 8.241580971555374e-07,
      "loss": 0.0067,
      "step": 13657
    },
    {
      "epoch": 8.73832373640435,
      "grad_norm": 0.3965696655768527,
      "learning_rate": 8.233345269311687e-07,
      "loss": 0.0046,
      "step": 13658
    },
    {
      "epoch": 8.738963531669866,
      "grad_norm": 0.5033825615733044,
      "learning_rate": 8.225113507364913e-07,
      "loss": 0.0059,
      "step": 13659
    },
    {
      "epoch": 8.73960332693538,
      "grad_norm": 0.48914269705939095,
      "learning_rate": 8.216885686068499e-07,
      "loss": 0.0058,
      "step": 13660
    },
    {
      "epoch": 8.740243122200896,
      "grad_norm": 0.4989530285968258,
      "learning_rate": 8.208661805775752e-07,
      "loss": 0.0064,
      "step": 13661
    },
    {
      "epoch": 8.740882917466411,
      "grad_norm": 0.3762844976694946,
      "learning_rate": 8.200441866839792e-07,
      "loss": 0.0044,
      "step": 13662
    },
    {
      "epoch": 8.741522712731927,
      "grad_norm": 0.34936950913636655,
      "learning_rate": 8.192225869613557e-07,
      "loss": 0.0043,
      "step": 13663
    },
    {
      "epoch": 8.74216250799744,
      "grad_norm": 0.41689919084442706,
      "learning_rate": 8.184013814449809e-07,
      "loss": 0.0049,
      "step": 13664
    },
    {
      "epoch": 8.742802303262955,
      "grad_norm": 0.436881547086602,
      "learning_rate": 8.1758057017012e-07,
      "loss": 0.0053,
      "step": 13665
    },
    {
      "epoch": 8.74344209852847,
      "grad_norm": 0.3305386517170733,
      "learning_rate": 8.167601531720149e-07,
      "loss": 0.0034,
      "step": 13666
    },
    {
      "epoch": 8.744081893793986,
      "grad_norm": 0.37280333711621827,
      "learning_rate": 8.159401304858939e-07,
      "loss": 0.0045,
      "step": 13667
    },
    {
      "epoch": 8.744721689059501,
      "grad_norm": 0.5124805723213327,
      "learning_rate": 8.151205021469655e-07,
      "loss": 0.0062,
      "step": 13668
    },
    {
      "epoch": 8.745361484325016,
      "grad_norm": 0.44062826126664423,
      "learning_rate": 8.143012681904238e-07,
      "loss": 0.0049,
      "step": 13669
    },
    {
      "epoch": 8.746001279590532,
      "grad_norm": 0.37286316018306276,
      "learning_rate": 8.134824286514464e-07,
      "loss": 0.0049,
      "step": 13670
    },
    {
      "epoch": 8.746641074856045,
      "grad_norm": 0.4604208398712177,
      "learning_rate": 8.126639835651906e-07,
      "loss": 0.006,
      "step": 13671
    },
    {
      "epoch": 8.74728087012156,
      "grad_norm": 0.4898718211069133,
      "learning_rate": 8.118459329668016e-07,
      "loss": 0.0058,
      "step": 13672
    },
    {
      "epoch": 8.747920665387076,
      "grad_norm": 0.46292598872115004,
      "learning_rate": 8.110282768914035e-07,
      "loss": 0.0055,
      "step": 13673
    },
    {
      "epoch": 8.748560460652591,
      "grad_norm": 0.48920289594123967,
      "learning_rate": 8.102110153741039e-07,
      "loss": 0.0051,
      "step": 13674
    },
    {
      "epoch": 8.749200255918106,
      "grad_norm": 0.39655066895938196,
      "learning_rate": 8.093941484499979e-07,
      "loss": 0.0048,
      "step": 13675
    },
    {
      "epoch": 8.749840051183622,
      "grad_norm": 0.4273480792235554,
      "learning_rate": 8.085776761541574e-07,
      "loss": 0.0049,
      "step": 13676
    },
    {
      "epoch": 8.750479846449137,
      "grad_norm": 0.4182124009194838,
      "learning_rate": 8.077615985216403e-07,
      "loss": 0.0054,
      "step": 13677
    },
    {
      "epoch": 8.751119641714652,
      "grad_norm": 0.403033967736268,
      "learning_rate": 8.069459155874904e-07,
      "loss": 0.0049,
      "step": 13678
    },
    {
      "epoch": 8.751759436980166,
      "grad_norm": 0.4834223648362714,
      "learning_rate": 8.061306273867287e-07,
      "loss": 0.0066,
      "step": 13679
    },
    {
      "epoch": 8.752399232245681,
      "grad_norm": 0.41594094667608283,
      "learning_rate": 8.053157339543638e-07,
      "loss": 0.0061,
      "step": 13680
    },
    {
      "epoch": 8.753039027511196,
      "grad_norm": 0.46338431923304485,
      "learning_rate": 8.045012353253834e-07,
      "loss": 0.006,
      "step": 13681
    },
    {
      "epoch": 8.753678822776711,
      "grad_norm": 0.5383484623031412,
      "learning_rate": 8.036871315347638e-07,
      "loss": 0.0057,
      "step": 13682
    },
    {
      "epoch": 8.754318618042227,
      "grad_norm": 0.45023528699870774,
      "learning_rate": 8.028734226174606e-07,
      "loss": 0.0057,
      "step": 13683
    },
    {
      "epoch": 8.754958413307742,
      "grad_norm": 0.4878289727807897,
      "learning_rate": 8.020601086084124e-07,
      "loss": 0.0055,
      "step": 13684
    },
    {
      "epoch": 8.755598208573257,
      "grad_norm": 0.46230939622259587,
      "learning_rate": 8.0124718954254e-07,
      "loss": 0.0066,
      "step": 13685
    },
    {
      "epoch": 8.75623800383877,
      "grad_norm": 0.45990191711624256,
      "learning_rate": 8.004346654547512e-07,
      "loss": 0.0064,
      "step": 13686
    },
    {
      "epoch": 8.756877799104286,
      "grad_norm": 0.34018419878562794,
      "learning_rate": 7.996225363799303e-07,
      "loss": 0.0038,
      "step": 13687
    },
    {
      "epoch": 8.757517594369801,
      "grad_norm": 0.43111134944361296,
      "learning_rate": 7.98810802352954e-07,
      "loss": 0.0053,
      "step": 13688
    },
    {
      "epoch": 8.758157389635317,
      "grad_norm": 0.3778808891933915,
      "learning_rate": 7.979994634086741e-07,
      "loss": 0.0044,
      "step": 13689
    },
    {
      "epoch": 8.758797184900832,
      "grad_norm": 0.38313863948640986,
      "learning_rate": 7.971885195819284e-07,
      "loss": 0.0049,
      "step": 13690
    },
    {
      "epoch": 8.759436980166347,
      "grad_norm": 0.38201982456548145,
      "learning_rate": 7.963779709075347e-07,
      "loss": 0.0044,
      "step": 13691
    },
    {
      "epoch": 8.760076775431862,
      "grad_norm": 0.5020651545606274,
      "learning_rate": 7.955678174203008e-07,
      "loss": 0.0065,
      "step": 13692
    },
    {
      "epoch": 8.760716570697376,
      "grad_norm": 0.32571889617358296,
      "learning_rate": 7.947580591550119e-07,
      "loss": 0.0039,
      "step": 13693
    },
    {
      "epoch": 8.761356365962891,
      "grad_norm": 0.3827356759976299,
      "learning_rate": 7.939486961464349e-07,
      "loss": 0.0043,
      "step": 13694
    },
    {
      "epoch": 8.761996161228407,
      "grad_norm": 0.5017255639589213,
      "learning_rate": 7.931397284293263e-07,
      "loss": 0.0063,
      "step": 13695
    },
    {
      "epoch": 8.762635956493922,
      "grad_norm": 0.3194161449673768,
      "learning_rate": 7.923311560384206e-07,
      "loss": 0.0036,
      "step": 13696
    },
    {
      "epoch": 8.763275751759437,
      "grad_norm": 0.4698127953542094,
      "learning_rate": 7.915229790084356e-07,
      "loss": 0.0066,
      "step": 13697
    },
    {
      "epoch": 8.763915547024952,
      "grad_norm": 0.4218739885648816,
      "learning_rate": 7.907151973740734e-07,
      "loss": 0.0049,
      "step": 13698
    },
    {
      "epoch": 8.764555342290468,
      "grad_norm": 0.3981458332116828,
      "learning_rate": 7.899078111700176e-07,
      "loss": 0.0057,
      "step": 13699
    },
    {
      "epoch": 8.765195137555983,
      "grad_norm": 0.4257481650254901,
      "learning_rate": 7.891008204309381e-07,
      "loss": 0.0056,
      "step": 13700
    },
    {
      "epoch": 8.765834932821496,
      "grad_norm": 0.4264642477876266,
      "learning_rate": 7.882942251914827e-07,
      "loss": 0.0045,
      "step": 13701
    },
    {
      "epoch": 8.766474728087012,
      "grad_norm": 0.5102126139791189,
      "learning_rate": 7.874880254862882e-07,
      "loss": 0.0061,
      "step": 13702
    },
    {
      "epoch": 8.767114523352527,
      "grad_norm": 0.5305126290168055,
      "learning_rate": 7.866822213499703e-07,
      "loss": 0.007,
      "step": 13703
    },
    {
      "epoch": 8.767754318618042,
      "grad_norm": 0.42624443744392854,
      "learning_rate": 7.858768128171268e-07,
      "loss": 0.0048,
      "step": 13704
    },
    {
      "epoch": 8.768394113883557,
      "grad_norm": 0.43509411406830606,
      "learning_rate": 7.850717999223434e-07,
      "loss": 0.0051,
      "step": 13705
    },
    {
      "epoch": 8.769033909149073,
      "grad_norm": 0.43724917673921354,
      "learning_rate": 7.842671827001857e-07,
      "loss": 0.0055,
      "step": 13706
    },
    {
      "epoch": 8.769673704414588,
      "grad_norm": 0.355105232998505,
      "learning_rate": 7.834629611851996e-07,
      "loss": 0.0045,
      "step": 13707
    },
    {
      "epoch": 8.770313499680102,
      "grad_norm": 0.41293759314727985,
      "learning_rate": 7.826591354119206e-07,
      "loss": 0.0055,
      "step": 13708
    },
    {
      "epoch": 8.770953294945617,
      "grad_norm": 0.4682149900577752,
      "learning_rate": 7.818557054148624e-07,
      "loss": 0.0071,
      "step": 13709
    },
    {
      "epoch": 8.771593090211132,
      "grad_norm": 0.46066904939016734,
      "learning_rate": 7.810526712285216e-07,
      "loss": 0.0063,
      "step": 13710
    },
    {
      "epoch": 8.772232885476647,
      "grad_norm": 0.43783678163499185,
      "learning_rate": 7.802500328873786e-07,
      "loss": 0.006,
      "step": 13711
    },
    {
      "epoch": 8.772872680742163,
      "grad_norm": 0.5434709909257042,
      "learning_rate": 7.794477904259002e-07,
      "loss": 0.008,
      "step": 13712
    },
    {
      "epoch": 8.773512476007678,
      "grad_norm": 0.41654106631786,
      "learning_rate": 7.786459438785321e-07,
      "loss": 0.0057,
      "step": 13713
    },
    {
      "epoch": 8.774152271273193,
      "grad_norm": 0.4705996178663126,
      "learning_rate": 7.778444932797025e-07,
      "loss": 0.006,
      "step": 13714
    },
    {
      "epoch": 8.774792066538708,
      "grad_norm": 0.43005419190336674,
      "learning_rate": 7.770434386638259e-07,
      "loss": 0.0056,
      "step": 13715
    },
    {
      "epoch": 8.775431861804222,
      "grad_norm": 0.46941093097940484,
      "learning_rate": 7.762427800652983e-07,
      "loss": 0.0051,
      "step": 13716
    },
    {
      "epoch": 8.776071657069737,
      "grad_norm": 0.358321998551092,
      "learning_rate": 7.754425175184954e-07,
      "loss": 0.0048,
      "step": 13717
    },
    {
      "epoch": 8.776711452335253,
      "grad_norm": 0.4202627996198747,
      "learning_rate": 7.746426510577843e-07,
      "loss": 0.0055,
      "step": 13718
    },
    {
      "epoch": 8.777351247600768,
      "grad_norm": 0.41478638954306013,
      "learning_rate": 7.738431807175062e-07,
      "loss": 0.0056,
      "step": 13719
    },
    {
      "epoch": 8.777991042866283,
      "grad_norm": 0.37112036823030625,
      "learning_rate": 7.730441065319904e-07,
      "loss": 0.0044,
      "step": 13720
    },
    {
      "epoch": 8.778630838131798,
      "grad_norm": 0.44136501269670353,
      "learning_rate": 7.722454285355452e-07,
      "loss": 0.0059,
      "step": 13721
    },
    {
      "epoch": 8.779270633397314,
      "grad_norm": 0.4289803107581612,
      "learning_rate": 7.714471467624684e-07,
      "loss": 0.0054,
      "step": 13722
    },
    {
      "epoch": 8.779910428662827,
      "grad_norm": 0.38706820132470465,
      "learning_rate": 7.70649261247034e-07,
      "loss": 0.0044,
      "step": 13723
    },
    {
      "epoch": 8.780550223928342,
      "grad_norm": 0.34313584303724154,
      "learning_rate": 7.698517720235021e-07,
      "loss": 0.0037,
      "step": 13724
    },
    {
      "epoch": 8.781190019193858,
      "grad_norm": 0.4399423051030314,
      "learning_rate": 7.690546791261166e-07,
      "loss": 0.0056,
      "step": 13725
    },
    {
      "epoch": 8.781829814459373,
      "grad_norm": 0.42725363482801854,
      "learning_rate": 7.682579825891034e-07,
      "loss": 0.0058,
      "step": 13726
    },
    {
      "epoch": 8.782469609724888,
      "grad_norm": 0.4190136463715427,
      "learning_rate": 7.674616824466697e-07,
      "loss": 0.0046,
      "step": 13727
    },
    {
      "epoch": 8.783109404990403,
      "grad_norm": 0.4331964471022406,
      "learning_rate": 7.666657787330089e-07,
      "loss": 0.0051,
      "step": 13728
    },
    {
      "epoch": 8.783749200255919,
      "grad_norm": 0.41373603401030556,
      "learning_rate": 7.658702714822941e-07,
      "loss": 0.005,
      "step": 13729
    },
    {
      "epoch": 8.784388995521432,
      "grad_norm": 0.5111752060935268,
      "learning_rate": 7.650751607286833e-07,
      "loss": 0.0064,
      "step": 13730
    },
    {
      "epoch": 8.785028790786948,
      "grad_norm": 0.4219014097280208,
      "learning_rate": 7.642804465063158e-07,
      "loss": 0.0054,
      "step": 13731
    },
    {
      "epoch": 8.785668586052463,
      "grad_norm": 0.502970512837658,
      "learning_rate": 7.634861288493189e-07,
      "loss": 0.007,
      "step": 13732
    },
    {
      "epoch": 8.786308381317978,
      "grad_norm": 0.5030539737277209,
      "learning_rate": 7.626922077917975e-07,
      "loss": 0.0063,
      "step": 13733
    },
    {
      "epoch": 8.786948176583493,
      "grad_norm": 0.46599623129316836,
      "learning_rate": 7.618986833678377e-07,
      "loss": 0.0057,
      "step": 13734
    },
    {
      "epoch": 8.787587971849009,
      "grad_norm": 0.5240419600974596,
      "learning_rate": 7.611055556115188e-07,
      "loss": 0.0064,
      "step": 13735
    },
    {
      "epoch": 8.788227767114524,
      "grad_norm": 0.39449471143450243,
      "learning_rate": 7.603128245568913e-07,
      "loss": 0.0052,
      "step": 13736
    },
    {
      "epoch": 8.78886756238004,
      "grad_norm": 0.4226450553428835,
      "learning_rate": 7.595204902379938e-07,
      "loss": 0.0047,
      "step": 13737
    },
    {
      "epoch": 8.789507357645553,
      "grad_norm": 0.5247061553022024,
      "learning_rate": 7.587285526888499e-07,
      "loss": 0.006,
      "step": 13738
    },
    {
      "epoch": 8.790147152911068,
      "grad_norm": 0.5058178824156189,
      "learning_rate": 7.579370119434637e-07,
      "loss": 0.0061,
      "step": 13739
    },
    {
      "epoch": 8.790786948176583,
      "grad_norm": 0.3549822620173089,
      "learning_rate": 7.571458680358212e-07,
      "loss": 0.0038,
      "step": 13740
    },
    {
      "epoch": 8.791426743442099,
      "grad_norm": 0.5151079552220477,
      "learning_rate": 7.56355120999892e-07,
      "loss": 0.0042,
      "step": 13741
    },
    {
      "epoch": 8.792066538707614,
      "grad_norm": 0.4253970127254065,
      "learning_rate": 7.555647708696334e-07,
      "loss": 0.0044,
      "step": 13742
    },
    {
      "epoch": 8.792706333973129,
      "grad_norm": 0.4090252025030053,
      "learning_rate": 7.547748176789782e-07,
      "loss": 0.0056,
      "step": 13743
    },
    {
      "epoch": 8.793346129238644,
      "grad_norm": 0.4786229735384796,
      "learning_rate": 7.539852614618459e-07,
      "loss": 0.0052,
      "step": 13744
    },
    {
      "epoch": 8.79398592450416,
      "grad_norm": 0.44039467103920665,
      "learning_rate": 7.531961022521395e-07,
      "loss": 0.0068,
      "step": 13745
    },
    {
      "epoch": 8.794625719769673,
      "grad_norm": 0.3997260746086678,
      "learning_rate": 7.52407340083744e-07,
      "loss": 0.0047,
      "step": 13746
    },
    {
      "epoch": 8.795265515035188,
      "grad_norm": 0.4275223328244261,
      "learning_rate": 7.516189749905267e-07,
      "loss": 0.0053,
      "step": 13747
    },
    {
      "epoch": 8.795905310300704,
      "grad_norm": 0.39367685674207203,
      "learning_rate": 7.508310070063374e-07,
      "loss": 0.0052,
      "step": 13748
    },
    {
      "epoch": 8.796545105566219,
      "grad_norm": 0.39655823469038015,
      "learning_rate": 7.500434361650133e-07,
      "loss": 0.0041,
      "step": 13749
    },
    {
      "epoch": 8.797184900831734,
      "grad_norm": 0.4772298076055901,
      "learning_rate": 7.492562625003697e-07,
      "loss": 0.0054,
      "step": 13750
    },
    {
      "epoch": 8.79782469609725,
      "grad_norm": 0.3746036373634638,
      "learning_rate": 7.484694860462038e-07,
      "loss": 0.0044,
      "step": 13751
    },
    {
      "epoch": 8.798464491362765,
      "grad_norm": 0.33376128994105336,
      "learning_rate": 7.476831068363044e-07,
      "loss": 0.0042,
      "step": 13752
    },
    {
      "epoch": 8.799104286628278,
      "grad_norm": 0.40591200363731006,
      "learning_rate": 7.468971249044321e-07,
      "loss": 0.0051,
      "step": 13753
    },
    {
      "epoch": 8.799744081893794,
      "grad_norm": 0.5675072978460217,
      "learning_rate": 7.461115402843367e-07,
      "loss": 0.0069,
      "step": 13754
    },
    {
      "epoch": 8.800383877159309,
      "grad_norm": 0.33288106600674766,
      "learning_rate": 7.453263530097521e-07,
      "loss": 0.004,
      "step": 13755
    },
    {
      "epoch": 8.801023672424824,
      "grad_norm": 0.344425566630337,
      "learning_rate": 7.445415631143905e-07,
      "loss": 0.004,
      "step": 13756
    },
    {
      "epoch": 8.80166346769034,
      "grad_norm": 0.4189248375948763,
      "learning_rate": 7.437571706319501e-07,
      "loss": 0.005,
      "step": 13757
    },
    {
      "epoch": 8.802303262955855,
      "grad_norm": 0.5180783908454852,
      "learning_rate": 7.429731755961123e-07,
      "loss": 0.006,
      "step": 13758
    },
    {
      "epoch": 8.80294305822137,
      "grad_norm": 0.4616793404389281,
      "learning_rate": 7.421895780405386e-07,
      "loss": 0.0056,
      "step": 13759
    },
    {
      "epoch": 8.803582853486883,
      "grad_norm": 0.4292075578090236,
      "learning_rate": 7.414063779988756e-07,
      "loss": 0.0051,
      "step": 13760
    },
    {
      "epoch": 8.804222648752399,
      "grad_norm": 0.4049914193407783,
      "learning_rate": 7.406235755047519e-07,
      "loss": 0.004,
      "step": 13761
    },
    {
      "epoch": 8.804862444017914,
      "grad_norm": 0.46509841113590067,
      "learning_rate": 7.398411705917818e-07,
      "loss": 0.0057,
      "step": 13762
    },
    {
      "epoch": 8.80550223928343,
      "grad_norm": 0.43497839319184195,
      "learning_rate": 7.390591632935595e-07,
      "loss": 0.0048,
      "step": 13763
    },
    {
      "epoch": 8.806142034548945,
      "grad_norm": 0.5355503891904377,
      "learning_rate": 7.382775536436615e-07,
      "loss": 0.0073,
      "step": 13764
    },
    {
      "epoch": 8.80678182981446,
      "grad_norm": 0.5092735917251825,
      "learning_rate": 7.374963416756509e-07,
      "loss": 0.0064,
      "step": 13765
    },
    {
      "epoch": 8.807421625079975,
      "grad_norm": 0.40145170515922485,
      "learning_rate": 7.36715527423072e-07,
      "loss": 0.0044,
      "step": 13766
    },
    {
      "epoch": 8.808061420345489,
      "grad_norm": 0.47690286881479893,
      "learning_rate": 7.35935110919449e-07,
      "loss": 0.0058,
      "step": 13767
    },
    {
      "epoch": 8.808701215611004,
      "grad_norm": 0.44846742627916214,
      "learning_rate": 7.351550921982919e-07,
      "loss": 0.005,
      "step": 13768
    },
    {
      "epoch": 8.809341010876519,
      "grad_norm": 0.64596487858899,
      "learning_rate": 7.343754712930951e-07,
      "loss": 0.0085,
      "step": 13769
    },
    {
      "epoch": 8.809980806142034,
      "grad_norm": 0.4787086183230883,
      "learning_rate": 7.335962482373338e-07,
      "loss": 0.005,
      "step": 13770
    },
    {
      "epoch": 8.81062060140755,
      "grad_norm": 0.40071557535513197,
      "learning_rate": 7.328174230644647e-07,
      "loss": 0.0056,
      "step": 13771
    },
    {
      "epoch": 8.811260396673065,
      "grad_norm": 0.42598347426034366,
      "learning_rate": 7.32038995807931e-07,
      "loss": 0.005,
      "step": 13772
    },
    {
      "epoch": 8.81190019193858,
      "grad_norm": 0.3998130113508627,
      "learning_rate": 7.312609665011572e-07,
      "loss": 0.0043,
      "step": 13773
    },
    {
      "epoch": 8.812539987204095,
      "grad_norm": 0.4068265516530607,
      "learning_rate": 7.304833351775498e-07,
      "loss": 0.0056,
      "step": 13774
    },
    {
      "epoch": 8.813179782469609,
      "grad_norm": 0.43482329293581345,
      "learning_rate": 7.297061018704987e-07,
      "loss": 0.0054,
      "step": 13775
    },
    {
      "epoch": 8.813819577735124,
      "grad_norm": 0.4851643766582371,
      "learning_rate": 7.289292666133774e-07,
      "loss": 0.0057,
      "step": 13776
    },
    {
      "epoch": 8.81445937300064,
      "grad_norm": 0.36205363313057504,
      "learning_rate": 7.281528294395402e-07,
      "loss": 0.0042,
      "step": 13777
    },
    {
      "epoch": 8.815099168266155,
      "grad_norm": 0.5256531941814099,
      "learning_rate": 7.27376790382327e-07,
      "loss": 0.0055,
      "step": 13778
    },
    {
      "epoch": 8.81573896353167,
      "grad_norm": 0.35789797752821956,
      "learning_rate": 7.266011494750602e-07,
      "loss": 0.0035,
      "step": 13779
    },
    {
      "epoch": 8.816378758797185,
      "grad_norm": 0.48506650437540794,
      "learning_rate": 7.258259067510443e-07,
      "loss": 0.0058,
      "step": 13780
    },
    {
      "epoch": 8.8170185540627,
      "grad_norm": 0.28411094047020813,
      "learning_rate": 7.250510622435647e-07,
      "loss": 0.0038,
      "step": 13781
    },
    {
      "epoch": 8.817658349328216,
      "grad_norm": 0.4453150582807869,
      "learning_rate": 7.24276615985896e-07,
      "loss": 0.0044,
      "step": 13782
    },
    {
      "epoch": 8.81829814459373,
      "grad_norm": 0.45752847440313726,
      "learning_rate": 7.235025680112883e-07,
      "loss": 0.0056,
      "step": 13783
    },
    {
      "epoch": 8.818937939859245,
      "grad_norm": 0.41481614696115326,
      "learning_rate": 7.227289183529773e-07,
      "loss": 0.005,
      "step": 13784
    },
    {
      "epoch": 8.81957773512476,
      "grad_norm": 0.43616266104723816,
      "learning_rate": 7.219556670441852e-07,
      "loss": 0.0036,
      "step": 13785
    },
    {
      "epoch": 8.820217530390275,
      "grad_norm": 0.38349483194335254,
      "learning_rate": 7.211828141181132e-07,
      "loss": 0.0044,
      "step": 13786
    },
    {
      "epoch": 8.82085732565579,
      "grad_norm": 0.4222655789592428,
      "learning_rate": 7.204103596079437e-07,
      "loss": 0.0048,
      "step": 13787
    },
    {
      "epoch": 8.821497120921306,
      "grad_norm": 0.4344436628118916,
      "learning_rate": 7.19638303546848e-07,
      "loss": 0.0049,
      "step": 13788
    },
    {
      "epoch": 8.822136916186821,
      "grad_norm": 0.4595784916532007,
      "learning_rate": 7.188666459679738e-07,
      "loss": 0.0064,
      "step": 13789
    },
    {
      "epoch": 8.822776711452335,
      "grad_norm": 0.3471361734134549,
      "learning_rate": 7.180953869044572e-07,
      "loss": 0.0045,
      "step": 13790
    },
    {
      "epoch": 8.82341650671785,
      "grad_norm": 0.5271857033314725,
      "learning_rate": 7.173245263894114e-07,
      "loss": 0.0069,
      "step": 13791
    },
    {
      "epoch": 8.824056301983365,
      "grad_norm": 0.6164240897348614,
      "learning_rate": 7.165540644559387e-07,
      "loss": 0.007,
      "step": 13792
    },
    {
      "epoch": 8.82469609724888,
      "grad_norm": 0.4992743983740257,
      "learning_rate": 7.157840011371208e-07,
      "loss": 0.0049,
      "step": 13793
    },
    {
      "epoch": 8.825335892514396,
      "grad_norm": 0.44073572395496674,
      "learning_rate": 7.150143364660212e-07,
      "loss": 0.0052,
      "step": 13794
    },
    {
      "epoch": 8.825975687779911,
      "grad_norm": 0.4422236270185082,
      "learning_rate": 7.142450704756909e-07,
      "loss": 0.0063,
      "step": 13795
    },
    {
      "epoch": 8.826615483045426,
      "grad_norm": 0.4218424947600563,
      "learning_rate": 7.134762031991594e-07,
      "loss": 0.0051,
      "step": 13796
    },
    {
      "epoch": 8.82725527831094,
      "grad_norm": 0.47209906591061174,
      "learning_rate": 7.1270773466944e-07,
      "loss": 0.0061,
      "step": 13797
    },
    {
      "epoch": 8.827895073576455,
      "grad_norm": 0.40653108248579845,
      "learning_rate": 7.119396649195276e-07,
      "loss": 0.0049,
      "step": 13798
    },
    {
      "epoch": 8.82853486884197,
      "grad_norm": 0.43118120125807535,
      "learning_rate": 7.111719939824058e-07,
      "loss": 0.0051,
      "step": 13799
    },
    {
      "epoch": 8.829174664107486,
      "grad_norm": 0.45461982819263536,
      "learning_rate": 7.10404721891036e-07,
      "loss": 0.0043,
      "step": 13800
    },
    {
      "epoch": 8.829814459373,
      "grad_norm": 0.35072503895355533,
      "learning_rate": 7.096378486783595e-07,
      "loss": 0.0041,
      "step": 13801
    },
    {
      "epoch": 8.830454254638516,
      "grad_norm": 0.4685712759807752,
      "learning_rate": 7.088713743773101e-07,
      "loss": 0.005,
      "step": 13802
    },
    {
      "epoch": 8.831094049904031,
      "grad_norm": 0.3936978270111812,
      "learning_rate": 7.081052990207959e-07,
      "loss": 0.0045,
      "step": 13803
    },
    {
      "epoch": 8.831733845169545,
      "grad_norm": 0.42426676312519257,
      "learning_rate": 7.073396226417106e-07,
      "loss": 0.0058,
      "step": 13804
    },
    {
      "epoch": 8.83237364043506,
      "grad_norm": 0.38694703623476695,
      "learning_rate": 7.065743452729324e-07,
      "loss": 0.0048,
      "step": 13805
    },
    {
      "epoch": 8.833013435700575,
      "grad_norm": 0.43675554973875663,
      "learning_rate": 7.058094669473192e-07,
      "loss": 0.0055,
      "step": 13806
    },
    {
      "epoch": 8.83365323096609,
      "grad_norm": 0.4254698072698087,
      "learning_rate": 7.05044987697715e-07,
      "loss": 0.0058,
      "step": 13807
    },
    {
      "epoch": 8.834293026231606,
      "grad_norm": 0.4029524803188116,
      "learning_rate": 7.042809075569435e-07,
      "loss": 0.0045,
      "step": 13808
    },
    {
      "epoch": 8.834932821497121,
      "grad_norm": 0.4701287263322202,
      "learning_rate": 7.03517226557815e-07,
      "loss": 0.0054,
      "step": 13809
    },
    {
      "epoch": 8.835572616762636,
      "grad_norm": 0.48856374752861437,
      "learning_rate": 7.0275394473312e-07,
      "loss": 0.0054,
      "step": 13810
    },
    {
      "epoch": 8.836212412028152,
      "grad_norm": 0.4746583823231882,
      "learning_rate": 7.0199106211563e-07,
      "loss": 0.0053,
      "step": 13811
    },
    {
      "epoch": 8.836852207293665,
      "grad_norm": 0.3513862152647439,
      "learning_rate": 7.012285787381056e-07,
      "loss": 0.0043,
      "step": 13812
    },
    {
      "epoch": 8.83749200255918,
      "grad_norm": 0.46880626982691576,
      "learning_rate": 7.00466494633285e-07,
      "loss": 0.0067,
      "step": 13813
    },
    {
      "epoch": 8.838131797824696,
      "grad_norm": 0.5396627794797487,
      "learning_rate": 6.997048098338888e-07,
      "loss": 0.0066,
      "step": 13814
    },
    {
      "epoch": 8.838771593090211,
      "grad_norm": 0.4040768836550669,
      "learning_rate": 6.98943524372625e-07,
      "loss": 0.0055,
      "step": 13815
    },
    {
      "epoch": 8.839411388355726,
      "grad_norm": 0.45025466809133735,
      "learning_rate": 6.981826382821821e-07,
      "loss": 0.0056,
      "step": 13816
    },
    {
      "epoch": 8.840051183621242,
      "grad_norm": 0.4056960503828836,
      "learning_rate": 6.974221515952295e-07,
      "loss": 0.005,
      "step": 13817
    },
    {
      "epoch": 8.840690978886757,
      "grad_norm": 0.47254238259576525,
      "learning_rate": 6.966620643444222e-07,
      "loss": 0.0059,
      "step": 13818
    },
    {
      "epoch": 8.841330774152272,
      "grad_norm": 0.3960368137713664,
      "learning_rate": 6.959023765623962e-07,
      "loss": 0.0044,
      "step": 13819
    },
    {
      "epoch": 8.841970569417786,
      "grad_norm": 0.543405073815851,
      "learning_rate": 6.951430882817712e-07,
      "loss": 0.0064,
      "step": 13820
    },
    {
      "epoch": 8.842610364683301,
      "grad_norm": 0.42528427682272507,
      "learning_rate": 6.943841995351486e-07,
      "loss": 0.0058,
      "step": 13821
    },
    {
      "epoch": 8.843250159948816,
      "grad_norm": 0.46239029950554783,
      "learning_rate": 6.93625710355117e-07,
      "loss": 0.0061,
      "step": 13822
    },
    {
      "epoch": 8.843889955214332,
      "grad_norm": 0.3687574895390154,
      "learning_rate": 6.928676207742425e-07,
      "loss": 0.0035,
      "step": 13823
    },
    {
      "epoch": 8.844529750479847,
      "grad_norm": 0.3633426249089123,
      "learning_rate": 6.921099308250756e-07,
      "loss": 0.0052,
      "step": 13824
    },
    {
      "epoch": 8.845169545745362,
      "grad_norm": 0.3679897045726106,
      "learning_rate": 6.913526405401494e-07,
      "loss": 0.0041,
      "step": 13825
    },
    {
      "epoch": 8.845809341010877,
      "grad_norm": 0.4324441744236299,
      "learning_rate": 6.905957499519844e-07,
      "loss": 0.0046,
      "step": 13826
    },
    {
      "epoch": 8.84644913627639,
      "grad_norm": 0.4361219170098835,
      "learning_rate": 6.898392590930769e-07,
      "loss": 0.0057,
      "step": 13827
    },
    {
      "epoch": 8.847088931541906,
      "grad_norm": 0.4476677944139078,
      "learning_rate": 6.890831679959086e-07,
      "loss": 0.0045,
      "step": 13828
    },
    {
      "epoch": 8.847728726807421,
      "grad_norm": 0.44187814000318615,
      "learning_rate": 6.883274766929482e-07,
      "loss": 0.0054,
      "step": 13829
    },
    {
      "epoch": 8.848368522072937,
      "grad_norm": 0.4603783736125089,
      "learning_rate": 6.875721852166417e-07,
      "loss": 0.0051,
      "step": 13830
    },
    {
      "epoch": 8.849008317338452,
      "grad_norm": 0.4011754422408708,
      "learning_rate": 6.868172935994188e-07,
      "loss": 0.0059,
      "step": 13831
    },
    {
      "epoch": 8.849648112603967,
      "grad_norm": 0.339780500707748,
      "learning_rate": 6.860628018736948e-07,
      "loss": 0.0034,
      "step": 13832
    },
    {
      "epoch": 8.850287907869482,
      "grad_norm": 0.4552922430994997,
      "learning_rate": 6.853087100718669e-07,
      "loss": 0.0049,
      "step": 13833
    },
    {
      "epoch": 8.850927703134996,
      "grad_norm": 0.512409918996805,
      "learning_rate": 6.845550182263138e-07,
      "loss": 0.0066,
      "step": 13834
    },
    {
      "epoch": 8.851567498400511,
      "grad_norm": 0.4217814137622757,
      "learning_rate": 6.838017263693964e-07,
      "loss": 0.0056,
      "step": 13835
    },
    {
      "epoch": 8.852207293666027,
      "grad_norm": 0.3914573130415289,
      "learning_rate": 6.830488345334607e-07,
      "loss": 0.0051,
      "step": 13836
    },
    {
      "epoch": 8.852847088931542,
      "grad_norm": 0.5600561410285649,
      "learning_rate": 6.822963427508356e-07,
      "loss": 0.0074,
      "step": 13837
    },
    {
      "epoch": 8.853486884197057,
      "grad_norm": 0.5060490459094225,
      "learning_rate": 6.815442510538273e-07,
      "loss": 0.006,
      "step": 13838
    },
    {
      "epoch": 8.854126679462572,
      "grad_norm": 0.5182602138703447,
      "learning_rate": 6.807925594747355e-07,
      "loss": 0.0054,
      "step": 13839
    },
    {
      "epoch": 8.854766474728088,
      "grad_norm": 0.5018505085354503,
      "learning_rate": 6.800412680458335e-07,
      "loss": 0.0067,
      "step": 13840
    },
    {
      "epoch": 8.855406269993601,
      "grad_norm": 0.4324586808301658,
      "learning_rate": 6.792903767993786e-07,
      "loss": 0.0047,
      "step": 13841
    },
    {
      "epoch": 8.856046065259116,
      "grad_norm": 0.47849680101033043,
      "learning_rate": 6.785398857676162e-07,
      "loss": 0.0051,
      "step": 13842
    },
    {
      "epoch": 8.856685860524632,
      "grad_norm": 0.41006723909238163,
      "learning_rate": 6.777897949827694e-07,
      "loss": 0.0047,
      "step": 13843
    },
    {
      "epoch": 8.857325655790147,
      "grad_norm": 0.5000896835848953,
      "learning_rate": 6.77040104477047e-07,
      "loss": 0.0059,
      "step": 13844
    },
    {
      "epoch": 8.857965451055662,
      "grad_norm": 0.3541188895285364,
      "learning_rate": 6.762908142826352e-07,
      "loss": 0.0038,
      "step": 13845
    },
    {
      "epoch": 8.858605246321178,
      "grad_norm": 0.3856534320917173,
      "learning_rate": 6.75541924431713e-07,
      "loss": 0.0052,
      "step": 13846
    },
    {
      "epoch": 8.859245041586693,
      "grad_norm": 0.38884332319367415,
      "learning_rate": 6.747934349564322e-07,
      "loss": 0.0049,
      "step": 13847
    },
    {
      "epoch": 8.859884836852208,
      "grad_norm": 0.40572290150641066,
      "learning_rate": 6.74045345888934e-07,
      "loss": 0.0053,
      "step": 13848
    },
    {
      "epoch": 8.860524632117722,
      "grad_norm": 0.3204219828208601,
      "learning_rate": 6.732976572613393e-07,
      "loss": 0.004,
      "step": 13849
    },
    {
      "epoch": 8.861164427383237,
      "grad_norm": 0.3181247050286971,
      "learning_rate": 6.725503691057523e-07,
      "loss": 0.004,
      "step": 13850
    },
    {
      "epoch": 8.861804222648752,
      "grad_norm": 0.43089998986565714,
      "learning_rate": 6.718034814542585e-07,
      "loss": 0.0055,
      "step": 13851
    },
    {
      "epoch": 8.862444017914267,
      "grad_norm": 0.5275896808352578,
      "learning_rate": 6.7105699433893e-07,
      "loss": 0.006,
      "step": 13852
    },
    {
      "epoch": 8.863083813179783,
      "grad_norm": 0.3744933718072571,
      "learning_rate": 6.703109077918202e-07,
      "loss": 0.0043,
      "step": 13853
    },
    {
      "epoch": 8.863723608445298,
      "grad_norm": 0.36346008971952903,
      "learning_rate": 6.695652218449644e-07,
      "loss": 0.0046,
      "step": 13854
    },
    {
      "epoch": 8.864363403710813,
      "grad_norm": 0.3782249840188488,
      "learning_rate": 6.688199365303782e-07,
      "loss": 0.005,
      "step": 13855
    },
    {
      "epoch": 8.865003198976328,
      "grad_norm": 0.4534901513612238,
      "learning_rate": 6.680750518800672e-07,
      "loss": 0.0061,
      "step": 13856
    },
    {
      "epoch": 8.865642994241842,
      "grad_norm": 0.45527836036276015,
      "learning_rate": 6.673305679260123e-07,
      "loss": 0.0058,
      "step": 13857
    },
    {
      "epoch": 8.866282789507357,
      "grad_norm": 0.46288813616651986,
      "learning_rate": 6.665864847001802e-07,
      "loss": 0.005,
      "step": 13858
    },
    {
      "epoch": 8.866922584772873,
      "grad_norm": 0.3467338923940496,
      "learning_rate": 6.658428022345232e-07,
      "loss": 0.0037,
      "step": 13859
    },
    {
      "epoch": 8.867562380038388,
      "grad_norm": 0.42240832312970644,
      "learning_rate": 6.650995205609723e-07,
      "loss": 0.0051,
      "step": 13860
    },
    {
      "epoch": 8.868202175303903,
      "grad_norm": 0.45392777849000515,
      "learning_rate": 6.643566397114398e-07,
      "loss": 0.0065,
      "step": 13861
    },
    {
      "epoch": 8.868841970569418,
      "grad_norm": 0.43692139929608426,
      "learning_rate": 6.636141597178292e-07,
      "loss": 0.0065,
      "step": 13862
    },
    {
      "epoch": 8.869481765834934,
      "grad_norm": 0.4846427727303518,
      "learning_rate": 6.628720806120181e-07,
      "loss": 0.0068,
      "step": 13863
    },
    {
      "epoch": 8.870121561100447,
      "grad_norm": 0.4299220957017988,
      "learning_rate": 6.621304024258701e-07,
      "loss": 0.0049,
      "step": 13864
    },
    {
      "epoch": 8.870761356365962,
      "grad_norm": 0.4246325331635346,
      "learning_rate": 6.613891251912318e-07,
      "loss": 0.0046,
      "step": 13865
    },
    {
      "epoch": 8.871401151631478,
      "grad_norm": 0.3687631266593537,
      "learning_rate": 6.606482489399335e-07,
      "loss": 0.0043,
      "step": 13866
    },
    {
      "epoch": 8.872040946896993,
      "grad_norm": 0.49109282075716437,
      "learning_rate": 6.599077737037851e-07,
      "loss": 0.0052,
      "step": 13867
    },
    {
      "epoch": 8.872680742162508,
      "grad_norm": 0.34313644973867924,
      "learning_rate": 6.591676995145801e-07,
      "loss": 0.0034,
      "step": 13868
    },
    {
      "epoch": 8.873320537428024,
      "grad_norm": 0.49184854128350264,
      "learning_rate": 6.584280264040999e-07,
      "loss": 0.0066,
      "step": 13869
    },
    {
      "epoch": 8.873960332693539,
      "grad_norm": 0.4174963640265576,
      "learning_rate": 6.576887544041033e-07,
      "loss": 0.0049,
      "step": 13870
    },
    {
      "epoch": 8.874600127959052,
      "grad_norm": 0.44453007770941516,
      "learning_rate": 6.569498835463317e-07,
      "loss": 0.0055,
      "step": 13871
    },
    {
      "epoch": 8.875239923224568,
      "grad_norm": 0.45924774221778825,
      "learning_rate": 6.562114138625131e-07,
      "loss": 0.0063,
      "step": 13872
    },
    {
      "epoch": 8.875879718490083,
      "grad_norm": 0.45200755433815876,
      "learning_rate": 6.554733453843542e-07,
      "loss": 0.0053,
      "step": 13873
    },
    {
      "epoch": 8.876519513755598,
      "grad_norm": 0.3592661067675065,
      "learning_rate": 6.547356781435488e-07,
      "loss": 0.0045,
      "step": 13874
    },
    {
      "epoch": 8.877159309021113,
      "grad_norm": 0.39630018237383,
      "learning_rate": 6.539984121717668e-07,
      "loss": 0.0052,
      "step": 13875
    },
    {
      "epoch": 8.877799104286629,
      "grad_norm": 0.48687744035698055,
      "learning_rate": 6.532615475006699e-07,
      "loss": 0.0065,
      "step": 13876
    },
    {
      "epoch": 8.878438899552144,
      "grad_norm": 0.39335979020765655,
      "learning_rate": 6.525250841618946e-07,
      "loss": 0.0043,
      "step": 13877
    },
    {
      "epoch": 8.879078694817657,
      "grad_norm": 0.3667004748581699,
      "learning_rate": 6.51789022187066e-07,
      "loss": 0.0038,
      "step": 13878
    },
    {
      "epoch": 8.879718490083173,
      "grad_norm": 0.392624021028618,
      "learning_rate": 6.510533616077863e-07,
      "loss": 0.005,
      "step": 13879
    },
    {
      "epoch": 8.880358285348688,
      "grad_norm": 0.3674017826039141,
      "learning_rate": 6.503181024556448e-07,
      "loss": 0.0041,
      "step": 13880
    },
    {
      "epoch": 8.880998080614203,
      "grad_norm": 0.35740572531941595,
      "learning_rate": 6.495832447622108e-07,
      "loss": 0.0044,
      "step": 13881
    },
    {
      "epoch": 8.881637875879719,
      "grad_norm": 0.3573204898649544,
      "learning_rate": 6.488487885590411e-07,
      "loss": 0.0041,
      "step": 13882
    },
    {
      "epoch": 8.882277671145234,
      "grad_norm": 0.45154502392202633,
      "learning_rate": 6.481147338776705e-07,
      "loss": 0.0057,
      "step": 13883
    },
    {
      "epoch": 8.882917466410749,
      "grad_norm": 0.4589477776884184,
      "learning_rate": 6.473810807496161e-07,
      "loss": 0.0058,
      "step": 13884
    },
    {
      "epoch": 8.883557261676264,
      "grad_norm": 0.705606188024245,
      "learning_rate": 6.466478292063805e-07,
      "loss": 0.0082,
      "step": 13885
    },
    {
      "epoch": 8.884197056941778,
      "grad_norm": 0.4440887676760569,
      "learning_rate": 6.459149792794506e-07,
      "loss": 0.006,
      "step": 13886
    },
    {
      "epoch": 8.884836852207293,
      "grad_norm": 0.4369291927778015,
      "learning_rate": 6.451825310002913e-07,
      "loss": 0.0059,
      "step": 13887
    },
    {
      "epoch": 8.885476647472808,
      "grad_norm": 0.42315298391754047,
      "learning_rate": 6.444504844003518e-07,
      "loss": 0.0049,
      "step": 13888
    },
    {
      "epoch": 8.886116442738324,
      "grad_norm": 0.4785114118141049,
      "learning_rate": 6.437188395110672e-07,
      "loss": 0.0062,
      "step": 13889
    },
    {
      "epoch": 8.886756238003839,
      "grad_norm": 0.4926484389213625,
      "learning_rate": 6.429875963638532e-07,
      "loss": 0.0054,
      "step": 13890
    },
    {
      "epoch": 8.887396033269354,
      "grad_norm": 0.4149053835730093,
      "learning_rate": 6.422567549901037e-07,
      "loss": 0.0051,
      "step": 13891
    },
    {
      "epoch": 8.88803582853487,
      "grad_norm": 0.38813557799115034,
      "learning_rate": 6.41526315421206e-07,
      "loss": 0.004,
      "step": 13892
    },
    {
      "epoch": 8.888675623800385,
      "grad_norm": 0.4029095269763956,
      "learning_rate": 6.407962776885212e-07,
      "loss": 0.0056,
      "step": 13893
    },
    {
      "epoch": 8.889315419065898,
      "grad_norm": 0.5008780198453523,
      "learning_rate": 6.400666418233947e-07,
      "loss": 0.0068,
      "step": 13894
    },
    {
      "epoch": 8.889955214331414,
      "grad_norm": 0.3368640740594059,
      "learning_rate": 6.393374078571568e-07,
      "loss": 0.005,
      "step": 13895
    },
    {
      "epoch": 8.890595009596929,
      "grad_norm": 0.37033245431749756,
      "learning_rate": 6.386085758211191e-07,
      "loss": 0.0045,
      "step": 13896
    },
    {
      "epoch": 8.891234804862444,
      "grad_norm": 0.4136547913380819,
      "learning_rate": 6.378801457465767e-07,
      "loss": 0.0048,
      "step": 13897
    },
    {
      "epoch": 8.89187460012796,
      "grad_norm": 0.469093165982912,
      "learning_rate": 6.371521176648054e-07,
      "loss": 0.0053,
      "step": 13898
    },
    {
      "epoch": 8.892514395393475,
      "grad_norm": 0.47084524628183383,
      "learning_rate": 6.364244916070684e-07,
      "loss": 0.0064,
      "step": 13899
    },
    {
      "epoch": 8.89315419065899,
      "grad_norm": 0.40261138837641786,
      "learning_rate": 6.35697267604607e-07,
      "loss": 0.0053,
      "step": 13900
    },
    {
      "epoch": 8.893793985924503,
      "grad_norm": 0.531299625343349,
      "learning_rate": 6.349704456886463e-07,
      "loss": 0.0064,
      "step": 13901
    },
    {
      "epoch": 8.894433781190019,
      "grad_norm": 0.3297462867961143,
      "learning_rate": 6.342440258903959e-07,
      "loss": 0.0037,
      "step": 13902
    },
    {
      "epoch": 8.895073576455534,
      "grad_norm": 0.4154148197943964,
      "learning_rate": 6.335180082410475e-07,
      "loss": 0.0047,
      "step": 13903
    },
    {
      "epoch": 8.89571337172105,
      "grad_norm": 0.3471439220378131,
      "learning_rate": 6.32792392771775e-07,
      "loss": 0.0043,
      "step": 13904
    },
    {
      "epoch": 8.896353166986565,
      "grad_norm": 0.40203361153559225,
      "learning_rate": 6.320671795137312e-07,
      "loss": 0.0047,
      "step": 13905
    },
    {
      "epoch": 8.89699296225208,
      "grad_norm": 0.3112328422396422,
      "learning_rate": 6.313423684980602e-07,
      "loss": 0.0032,
      "step": 13906
    },
    {
      "epoch": 8.897632757517595,
      "grad_norm": 0.39643391033853975,
      "learning_rate": 6.306179597558825e-07,
      "loss": 0.0043,
      "step": 13907
    },
    {
      "epoch": 8.898272552783109,
      "grad_norm": 0.45668832372326135,
      "learning_rate": 6.298939533183035e-07,
      "loss": 0.0057,
      "step": 13908
    },
    {
      "epoch": 8.898912348048624,
      "grad_norm": 0.36323026604683056,
      "learning_rate": 6.291703492164091e-07,
      "loss": 0.0043,
      "step": 13909
    },
    {
      "epoch": 8.89955214331414,
      "grad_norm": 0.2666407065554783,
      "learning_rate": 6.284471474812704e-07,
      "loss": 0.0026,
      "step": 13910
    },
    {
      "epoch": 8.900191938579654,
      "grad_norm": 0.48124695074965745,
      "learning_rate": 6.27724348143941e-07,
      "loss": 0.0045,
      "step": 13911
    },
    {
      "epoch": 8.90083173384517,
      "grad_norm": 0.4296309517884549,
      "learning_rate": 6.270019512354552e-07,
      "loss": 0.0051,
      "step": 13912
    },
    {
      "epoch": 8.901471529110685,
      "grad_norm": 0.4273614816619048,
      "learning_rate": 6.262799567868327e-07,
      "loss": 0.0042,
      "step": 13913
    },
    {
      "epoch": 8.9021113243762,
      "grad_norm": 0.3669164290987377,
      "learning_rate": 6.255583648290753e-07,
      "loss": 0.0039,
      "step": 13914
    },
    {
      "epoch": 8.902751119641714,
      "grad_norm": 0.40500357442174356,
      "learning_rate": 6.248371753931637e-07,
      "loss": 0.0053,
      "step": 13915
    },
    {
      "epoch": 8.903390914907229,
      "grad_norm": 0.5035786191876893,
      "learning_rate": 6.241163885100676e-07,
      "loss": 0.0056,
      "step": 13916
    },
    {
      "epoch": 8.904030710172744,
      "grad_norm": 0.5505541240041358,
      "learning_rate": 6.233960042107368e-07,
      "loss": 0.0073,
      "step": 13917
    },
    {
      "epoch": 8.90467050543826,
      "grad_norm": 0.3216033565810026,
      "learning_rate": 6.226760225260986e-07,
      "loss": 0.0034,
      "step": 13918
    },
    {
      "epoch": 8.905310300703775,
      "grad_norm": 0.47743916335093817,
      "learning_rate": 6.219564434870739e-07,
      "loss": 0.0061,
      "step": 13919
    },
    {
      "epoch": 8.90595009596929,
      "grad_norm": 0.4669121476820101,
      "learning_rate": 6.21237267124557e-07,
      "loss": 0.0062,
      "step": 13920
    },
    {
      "epoch": 8.906589891234805,
      "grad_norm": 0.5275169748384597,
      "learning_rate": 6.205184934694286e-07,
      "loss": 0.0062,
      "step": 13921
    },
    {
      "epoch": 8.90722968650032,
      "grad_norm": 0.49095680195342156,
      "learning_rate": 6.198001225525497e-07,
      "loss": 0.0058,
      "step": 13922
    },
    {
      "epoch": 8.907869481765834,
      "grad_norm": 0.4553537932637871,
      "learning_rate": 6.190821544047687e-07,
      "loss": 0.0054,
      "step": 13923
    },
    {
      "epoch": 8.90850927703135,
      "grad_norm": 0.5554126844682723,
      "learning_rate": 6.183645890569123e-07,
      "loss": 0.0066,
      "step": 13924
    },
    {
      "epoch": 8.909149072296865,
      "grad_norm": 0.510539713527056,
      "learning_rate": 6.176474265397925e-07,
      "loss": 0.0052,
      "step": 13925
    },
    {
      "epoch": 8.90978886756238,
      "grad_norm": 0.4012993953485451,
      "learning_rate": 6.169306668842012e-07,
      "loss": 0.0062,
      "step": 13926
    },
    {
      "epoch": 8.910428662827895,
      "grad_norm": 0.4418336129031859,
      "learning_rate": 6.162143101209173e-07,
      "loss": 0.0048,
      "step": 13927
    },
    {
      "epoch": 8.91106845809341,
      "grad_norm": 0.4327116037562413,
      "learning_rate": 6.15498356280696e-07,
      "loss": 0.0055,
      "step": 13928
    },
    {
      "epoch": 8.911708253358926,
      "grad_norm": 0.4762055267434935,
      "learning_rate": 6.147828053942828e-07,
      "loss": 0.0069,
      "step": 13929
    },
    {
      "epoch": 8.912348048624441,
      "grad_norm": 0.3161785500546031,
      "learning_rate": 6.14067657492402e-07,
      "loss": 0.0036,
      "step": 13930
    },
    {
      "epoch": 8.912987843889955,
      "grad_norm": 0.3865616150475753,
      "learning_rate": 6.133529126057591e-07,
      "loss": 0.0062,
      "step": 13931
    },
    {
      "epoch": 8.91362763915547,
      "grad_norm": 0.33288584310945735,
      "learning_rate": 6.126385707650428e-07,
      "loss": 0.0036,
      "step": 13932
    },
    {
      "epoch": 8.914267434420985,
      "grad_norm": 0.40378728968183697,
      "learning_rate": 6.119246320009297e-07,
      "loss": 0.0046,
      "step": 13933
    },
    {
      "epoch": 8.9149072296865,
      "grad_norm": 0.432685658538069,
      "learning_rate": 6.112110963440721e-07,
      "loss": 0.0054,
      "step": 13934
    },
    {
      "epoch": 8.915547024952016,
      "grad_norm": 0.3632681841378661,
      "learning_rate": 6.104979638251074e-07,
      "loss": 0.0045,
      "step": 13935
    },
    {
      "epoch": 8.916186820217531,
      "grad_norm": 0.5942425028424068,
      "learning_rate": 6.097852344746591e-07,
      "loss": 0.0073,
      "step": 13936
    },
    {
      "epoch": 8.916826615483046,
      "grad_norm": 0.37746784925158666,
      "learning_rate": 6.090729083233293e-07,
      "loss": 0.0045,
      "step": 13937
    },
    {
      "epoch": 8.91746641074856,
      "grad_norm": 0.40378695385026087,
      "learning_rate": 6.083609854017036e-07,
      "loss": 0.006,
      "step": 13938
    },
    {
      "epoch": 8.918106206014075,
      "grad_norm": 0.3931816563320282,
      "learning_rate": 6.076494657403498e-07,
      "loss": 0.005,
      "step": 13939
    },
    {
      "epoch": 8.91874600127959,
      "grad_norm": 0.41724268563471273,
      "learning_rate": 6.069383493698211e-07,
      "loss": 0.0045,
      "step": 13940
    },
    {
      "epoch": 8.919385796545106,
      "grad_norm": 0.42192778587695245,
      "learning_rate": 6.06227636320651e-07,
      "loss": 0.0044,
      "step": 13941
    },
    {
      "epoch": 8.92002559181062,
      "grad_norm": 0.42800257453773916,
      "learning_rate": 6.055173266233538e-07,
      "loss": 0.0044,
      "step": 13942
    },
    {
      "epoch": 8.920665387076136,
      "grad_norm": 0.4401263357853731,
      "learning_rate": 6.048074203084331e-07,
      "loss": 0.0054,
      "step": 13943
    },
    {
      "epoch": 8.921305182341651,
      "grad_norm": 0.3067666307632997,
      "learning_rate": 6.040979174063677e-07,
      "loss": 0.0034,
      "step": 13944
    },
    {
      "epoch": 8.921944977607165,
      "grad_norm": 0.37144841711074145,
      "learning_rate": 6.033888179476233e-07,
      "loss": 0.004,
      "step": 13945
    },
    {
      "epoch": 8.92258477287268,
      "grad_norm": 0.45285374478193396,
      "learning_rate": 6.026801219626499e-07,
      "loss": 0.0057,
      "step": 13946
    },
    {
      "epoch": 8.923224568138195,
      "grad_norm": 0.5111066034475421,
      "learning_rate": 6.019718294818744e-07,
      "loss": 0.0067,
      "step": 13947
    },
    {
      "epoch": 8.92386436340371,
      "grad_norm": 0.4199720259596497,
      "learning_rate": 6.012639405357101e-07,
      "loss": 0.0045,
      "step": 13948
    },
    {
      "epoch": 8.924504158669226,
      "grad_norm": 0.459615283243916,
      "learning_rate": 6.00556455154554e-07,
      "loss": 0.0053,
      "step": 13949
    },
    {
      "epoch": 8.925143953934741,
      "grad_norm": 0.4318321669252702,
      "learning_rate": 5.998493733687838e-07,
      "loss": 0.0048,
      "step": 13950
    },
    {
      "epoch": 8.925783749200257,
      "grad_norm": 0.37299113779049076,
      "learning_rate": 5.991426952087598e-07,
      "loss": 0.0049,
      "step": 13951
    },
    {
      "epoch": 8.92642354446577,
      "grad_norm": 0.4502186941401046,
      "learning_rate": 5.984364207048232e-07,
      "loss": 0.0051,
      "step": 13952
    },
    {
      "epoch": 8.927063339731285,
      "grad_norm": 0.5240463126706505,
      "learning_rate": 5.977305498873054e-07,
      "loss": 0.0071,
      "step": 13953
    },
    {
      "epoch": 8.9277031349968,
      "grad_norm": 0.42050892313229105,
      "learning_rate": 5.97025082786512e-07,
      "loss": 0.0046,
      "step": 13954
    },
    {
      "epoch": 8.928342930262316,
      "grad_norm": 0.38086704642162333,
      "learning_rate": 5.963200194327357e-07,
      "loss": 0.0044,
      "step": 13955
    },
    {
      "epoch": 8.928982725527831,
      "grad_norm": 0.46952682246100247,
      "learning_rate": 5.956153598562486e-07,
      "loss": 0.0062,
      "step": 13956
    },
    {
      "epoch": 8.929622520793346,
      "grad_norm": 0.5138321508071058,
      "learning_rate": 5.949111040873101e-07,
      "loss": 0.0065,
      "step": 13957
    },
    {
      "epoch": 8.930262316058862,
      "grad_norm": 0.40124457771699273,
      "learning_rate": 5.942072521561559e-07,
      "loss": 0.0045,
      "step": 13958
    },
    {
      "epoch": 8.930902111324377,
      "grad_norm": 0.412494922392538,
      "learning_rate": 5.935038040930119e-07,
      "loss": 0.0055,
      "step": 13959
    },
    {
      "epoch": 8.93154190658989,
      "grad_norm": 0.42424075594061367,
      "learning_rate": 5.928007599280827e-07,
      "loss": 0.0046,
      "step": 13960
    },
    {
      "epoch": 8.932181701855406,
      "grad_norm": 0.38801382124992906,
      "learning_rate": 5.920981196915543e-07,
      "loss": 0.0041,
      "step": 13961
    },
    {
      "epoch": 8.932821497120921,
      "grad_norm": 0.3992368727545239,
      "learning_rate": 5.913958834135958e-07,
      "loss": 0.0044,
      "step": 13962
    },
    {
      "epoch": 8.933461292386436,
      "grad_norm": 0.4523543318499393,
      "learning_rate": 5.906940511243631e-07,
      "loss": 0.0057,
      "step": 13963
    },
    {
      "epoch": 8.934101087651952,
      "grad_norm": 0.42277262877739213,
      "learning_rate": 5.899926228539898e-07,
      "loss": 0.005,
      "step": 13964
    },
    {
      "epoch": 8.934740882917467,
      "grad_norm": 0.33449148903499143,
      "learning_rate": 5.892915986325931e-07,
      "loss": 0.0034,
      "step": 13965
    },
    {
      "epoch": 8.935380678182982,
      "grad_norm": 0.507766507873428,
      "learning_rate": 5.885909784902766e-07,
      "loss": 0.0058,
      "step": 13966
    },
    {
      "epoch": 8.936020473448497,
      "grad_norm": 0.47987292329929215,
      "learning_rate": 5.878907624571207e-07,
      "loss": 0.0061,
      "step": 13967
    },
    {
      "epoch": 8.93666026871401,
      "grad_norm": 0.4606897820707521,
      "learning_rate": 5.871909505631934e-07,
      "loss": 0.0051,
      "step": 13968
    },
    {
      "epoch": 8.937300063979526,
      "grad_norm": 0.5601653442162228,
      "learning_rate": 5.864915428385432e-07,
      "loss": 0.0061,
      "step": 13969
    },
    {
      "epoch": 8.937939859245041,
      "grad_norm": 0.4523900247829778,
      "learning_rate": 5.857925393131991e-07,
      "loss": 0.0063,
      "step": 13970
    },
    {
      "epoch": 8.938579654510557,
      "grad_norm": 0.4526554503499803,
      "learning_rate": 5.850939400171784e-07,
      "loss": 0.0053,
      "step": 13971
    },
    {
      "epoch": 8.939219449776072,
      "grad_norm": 0.43569666655141664,
      "learning_rate": 5.843957449804738e-07,
      "loss": 0.0052,
      "step": 13972
    },
    {
      "epoch": 8.939859245041587,
      "grad_norm": 0.4842843745389259,
      "learning_rate": 5.836979542330679e-07,
      "loss": 0.0055,
      "step": 13973
    },
    {
      "epoch": 8.940499040307103,
      "grad_norm": 0.29135980562255714,
      "learning_rate": 5.830005678049222e-07,
      "loss": 0.0028,
      "step": 13974
    },
    {
      "epoch": 8.941138835572616,
      "grad_norm": 0.47133928391974805,
      "learning_rate": 5.823035857259796e-07,
      "loss": 0.0073,
      "step": 13975
    },
    {
      "epoch": 8.941778630838131,
      "grad_norm": 0.5068836502339918,
      "learning_rate": 5.816070080261693e-07,
      "loss": 0.0065,
      "step": 13976
    },
    {
      "epoch": 8.942418426103647,
      "grad_norm": 0.42035901618465255,
      "learning_rate": 5.809108347353998e-07,
      "loss": 0.0055,
      "step": 13977
    },
    {
      "epoch": 8.943058221369162,
      "grad_norm": 0.36080598645124395,
      "learning_rate": 5.802150658835626e-07,
      "loss": 0.0049,
      "step": 13978
    },
    {
      "epoch": 8.943698016634677,
      "grad_norm": 0.49873360724778965,
      "learning_rate": 5.79519701500535e-07,
      "loss": 0.0057,
      "step": 13979
    },
    {
      "epoch": 8.944337811900192,
      "grad_norm": 0.5794169778552682,
      "learning_rate": 5.78824741616174e-07,
      "loss": 0.0049,
      "step": 13980
    },
    {
      "epoch": 8.944977607165708,
      "grad_norm": 0.6559853265513402,
      "learning_rate": 5.781301862603194e-07,
      "loss": 0.0074,
      "step": 13981
    },
    {
      "epoch": 8.945617402431221,
      "grad_norm": 0.3591712742413758,
      "learning_rate": 5.774360354627928e-07,
      "loss": 0.0039,
      "step": 13982
    },
    {
      "epoch": 8.946257197696736,
      "grad_norm": 0.3809924718280135,
      "learning_rate": 5.767422892534036e-07,
      "loss": 0.0043,
      "step": 13983
    },
    {
      "epoch": 8.946896992962252,
      "grad_norm": 0.4154211087413583,
      "learning_rate": 5.760489476619391e-07,
      "loss": 0.0049,
      "step": 13984
    },
    {
      "epoch": 8.947536788227767,
      "grad_norm": 0.3924963546323766,
      "learning_rate": 5.753560107181655e-07,
      "loss": 0.0051,
      "step": 13985
    },
    {
      "epoch": 8.948176583493282,
      "grad_norm": 0.44814895799640375,
      "learning_rate": 5.746634784518412e-07,
      "loss": 0.0064,
      "step": 13986
    },
    {
      "epoch": 8.948816378758798,
      "grad_norm": 0.6325997009546688,
      "learning_rate": 5.739713508927003e-07,
      "loss": 0.006,
      "step": 13987
    },
    {
      "epoch": 8.949456174024313,
      "grad_norm": 0.4263737157119238,
      "learning_rate": 5.732796280704634e-07,
      "loss": 0.0048,
      "step": 13988
    },
    {
      "epoch": 8.950095969289826,
      "grad_norm": 0.4206923414373466,
      "learning_rate": 5.725883100148277e-07,
      "loss": 0.0049,
      "step": 13989
    },
    {
      "epoch": 8.950735764555342,
      "grad_norm": 0.48351719090105305,
      "learning_rate": 5.71897396755482e-07,
      "loss": 0.0057,
      "step": 13990
    },
    {
      "epoch": 8.951375559820857,
      "grad_norm": 0.4358598394982186,
      "learning_rate": 5.7120688832209e-07,
      "loss": 0.0044,
      "step": 13991
    },
    {
      "epoch": 8.952015355086372,
      "grad_norm": 0.42967377952910346,
      "learning_rate": 5.705167847443016e-07,
      "loss": 0.005,
      "step": 13992
    },
    {
      "epoch": 8.952655150351887,
      "grad_norm": 0.45460324130099206,
      "learning_rate": 5.698270860517486e-07,
      "loss": 0.0046,
      "step": 13993
    },
    {
      "epoch": 8.953294945617403,
      "grad_norm": 0.4653069299844644,
      "learning_rate": 5.691377922740471e-07,
      "loss": 0.0063,
      "step": 13994
    },
    {
      "epoch": 8.953934740882918,
      "grad_norm": 0.45079000134276903,
      "learning_rate": 5.684489034407903e-07,
      "loss": 0.0045,
      "step": 13995
    },
    {
      "epoch": 8.954574536148433,
      "grad_norm": 0.501200155005459,
      "learning_rate": 5.677604195815612e-07,
      "loss": 0.0067,
      "step": 13996
    },
    {
      "epoch": 8.955214331413947,
      "grad_norm": 0.39570885985761767,
      "learning_rate": 5.670723407259215e-07,
      "loss": 0.0048,
      "step": 13997
    },
    {
      "epoch": 8.955854126679462,
      "grad_norm": 0.4186971160415396,
      "learning_rate": 5.663846669034156e-07,
      "loss": 0.0048,
      "step": 13998
    },
    {
      "epoch": 8.956493921944977,
      "grad_norm": 0.37645745092477817,
      "learning_rate": 5.65697398143571e-07,
      "loss": 0.0047,
      "step": 13999
    },
    {
      "epoch": 8.957133717210493,
      "grad_norm": 0.4761485225289767,
      "learning_rate": 5.650105344758982e-07,
      "loss": 0.0048,
      "step": 14000
    },
    {
      "epoch": 8.957773512476008,
      "grad_norm": 0.5046049217761246,
      "learning_rate": 5.643240759298896e-07,
      "loss": 0.0077,
      "step": 14001
    },
    {
      "epoch": 8.958413307741523,
      "grad_norm": 0.5070840344163818,
      "learning_rate": 5.636380225350179e-07,
      "loss": 0.006,
      "step": 14002
    },
    {
      "epoch": 8.959053103007038,
      "grad_norm": 0.5111348021779422,
      "learning_rate": 5.629523743207454e-07,
      "loss": 0.0073,
      "step": 14003
    },
    {
      "epoch": 8.959692898272554,
      "grad_norm": 0.35088104103303985,
      "learning_rate": 5.622671313165117e-07,
      "loss": 0.0044,
      "step": 14004
    },
    {
      "epoch": 8.960332693538067,
      "grad_norm": 0.3244389003507644,
      "learning_rate": 5.615822935517368e-07,
      "loss": 0.0034,
      "step": 14005
    },
    {
      "epoch": 8.960972488803582,
      "grad_norm": 0.39997031938965244,
      "learning_rate": 5.608978610558291e-07,
      "loss": 0.0054,
      "step": 14006
    },
    {
      "epoch": 8.961612284069098,
      "grad_norm": 0.3231122759853712,
      "learning_rate": 5.602138338581775e-07,
      "loss": 0.0034,
      "step": 14007
    },
    {
      "epoch": 8.962252079334613,
      "grad_norm": 0.5463655142140147,
      "learning_rate": 5.595302119881518e-07,
      "loss": 0.0053,
      "step": 14008
    },
    {
      "epoch": 8.962891874600128,
      "grad_norm": 0.43098896510872986,
      "learning_rate": 5.588469954751041e-07,
      "loss": 0.005,
      "step": 14009
    },
    {
      "epoch": 8.963531669865644,
      "grad_norm": 0.35941002946418926,
      "learning_rate": 5.581641843483731e-07,
      "loss": 0.0048,
      "step": 14010
    },
    {
      "epoch": 8.964171465131159,
      "grad_norm": 0.4095490921560911,
      "learning_rate": 5.574817786372754e-07,
      "loss": 0.0045,
      "step": 14011
    },
    {
      "epoch": 8.964811260396672,
      "grad_norm": 0.4141856964135463,
      "learning_rate": 5.567997783711132e-07,
      "loss": 0.0043,
      "step": 14012
    },
    {
      "epoch": 8.965451055662188,
      "grad_norm": 0.4136527440574189,
      "learning_rate": 5.561181835791729e-07,
      "loss": 0.005,
      "step": 14013
    },
    {
      "epoch": 8.966090850927703,
      "grad_norm": 0.4764400951811388,
      "learning_rate": 5.554369942907167e-07,
      "loss": 0.0048,
      "step": 14014
    },
    {
      "epoch": 8.966730646193218,
      "grad_norm": 0.40828166839705404,
      "learning_rate": 5.547562105349935e-07,
      "loss": 0.0047,
      "step": 14015
    },
    {
      "epoch": 8.967370441458733,
      "grad_norm": 0.39957815254259,
      "learning_rate": 5.540758323412388e-07,
      "loss": 0.0049,
      "step": 14016
    },
    {
      "epoch": 8.968010236724249,
      "grad_norm": 0.4894625335588888,
      "learning_rate": 5.533958597386646e-07,
      "loss": 0.0057,
      "step": 14017
    },
    {
      "epoch": 8.968650031989764,
      "grad_norm": 0.6643728297437674,
      "learning_rate": 5.527162927564689e-07,
      "loss": 0.0061,
      "step": 14018
    },
    {
      "epoch": 8.969289827255277,
      "grad_norm": 0.37821291296395904,
      "learning_rate": 5.520371314238282e-07,
      "loss": 0.0041,
      "step": 14019
    },
    {
      "epoch": 8.969929622520793,
      "grad_norm": 0.3076504865396771,
      "learning_rate": 5.51358375769907e-07,
      "loss": 0.0034,
      "step": 14020
    },
    {
      "epoch": 8.970569417786308,
      "grad_norm": 0.4264936173909115,
      "learning_rate": 5.50680025823851e-07,
      "loss": 0.0044,
      "step": 14021
    },
    {
      "epoch": 8.971209213051823,
      "grad_norm": 0.4822072278073843,
      "learning_rate": 5.500020816147833e-07,
      "loss": 0.0062,
      "step": 14022
    },
    {
      "epoch": 8.971849008317339,
      "grad_norm": 0.3554745137987027,
      "learning_rate": 5.493245431718164e-07,
      "loss": 0.0042,
      "step": 14023
    },
    {
      "epoch": 8.972488803582854,
      "grad_norm": 0.31924063981932416,
      "learning_rate": 5.486474105240436e-07,
      "loss": 0.004,
      "step": 14024
    },
    {
      "epoch": 8.973128598848369,
      "grad_norm": 0.4225132055793702,
      "learning_rate": 5.479706837005361e-07,
      "loss": 0.0052,
      "step": 14025
    },
    {
      "epoch": 8.973768394113884,
      "grad_norm": 0.45383086846018983,
      "learning_rate": 5.472943627303562e-07,
      "loss": 0.006,
      "step": 14026
    },
    {
      "epoch": 8.974408189379398,
      "grad_norm": 0.4767296086052536,
      "learning_rate": 5.466184476425396e-07,
      "loss": 0.0059,
      "step": 14027
    },
    {
      "epoch": 8.975047984644913,
      "grad_norm": 0.32719945378039067,
      "learning_rate": 5.459429384661119e-07,
      "loss": 0.004,
      "step": 14028
    },
    {
      "epoch": 8.975687779910428,
      "grad_norm": 0.42743420623124806,
      "learning_rate": 5.452678352300767e-07,
      "loss": 0.0061,
      "step": 14029
    },
    {
      "epoch": 8.976327575175944,
      "grad_norm": 0.4085362497744778,
      "learning_rate": 5.445931379634218e-07,
      "loss": 0.0055,
      "step": 14030
    },
    {
      "epoch": 8.976967370441459,
      "grad_norm": 0.3728840493020343,
      "learning_rate": 5.439188466951173e-07,
      "loss": 0.0041,
      "step": 14031
    },
    {
      "epoch": 8.977607165706974,
      "grad_norm": 0.3213856829468427,
      "learning_rate": 5.432449614541157e-07,
      "loss": 0.0032,
      "step": 14032
    },
    {
      "epoch": 8.97824696097249,
      "grad_norm": 0.4235668992677781,
      "learning_rate": 5.425714822693539e-07,
      "loss": 0.0054,
      "step": 14033
    },
    {
      "epoch": 8.978886756238003,
      "grad_norm": 0.37497473659041525,
      "learning_rate": 5.418984091697499e-07,
      "loss": 0.004,
      "step": 14034
    },
    {
      "epoch": 8.979526551503518,
      "grad_norm": 0.4407264838453872,
      "learning_rate": 5.412257421842027e-07,
      "loss": 0.0055,
      "step": 14035
    },
    {
      "epoch": 8.980166346769034,
      "grad_norm": 0.41503197242357964,
      "learning_rate": 5.405534813415969e-07,
      "loss": 0.0056,
      "step": 14036
    },
    {
      "epoch": 8.980806142034549,
      "grad_norm": 0.513420942748452,
      "learning_rate": 5.398816266707973e-07,
      "loss": 0.0056,
      "step": 14037
    },
    {
      "epoch": 8.981445937300064,
      "grad_norm": 0.4231207140032295,
      "learning_rate": 5.39210178200652e-07,
      "loss": 0.005,
      "step": 14038
    },
    {
      "epoch": 8.98208573256558,
      "grad_norm": 0.4600222426897581,
      "learning_rate": 5.385391359599923e-07,
      "loss": 0.0049,
      "step": 14039
    },
    {
      "epoch": 8.982725527831095,
      "grad_norm": 0.4988722313337832,
      "learning_rate": 5.378684999776318e-07,
      "loss": 0.0048,
      "step": 14040
    },
    {
      "epoch": 8.98336532309661,
      "grad_norm": 0.39827472895356897,
      "learning_rate": 5.371982702823664e-07,
      "loss": 0.0054,
      "step": 14041
    },
    {
      "epoch": 8.984005118362123,
      "grad_norm": 0.48098503264941833,
      "learning_rate": 5.365284469029741e-07,
      "loss": 0.0064,
      "step": 14042
    },
    {
      "epoch": 8.984644913627639,
      "grad_norm": 0.33934754196374284,
      "learning_rate": 5.358590298682165e-07,
      "loss": 0.0036,
      "step": 14043
    },
    {
      "epoch": 8.985284708893154,
      "grad_norm": 0.41231858407314337,
      "learning_rate": 5.35190019206836e-07,
      "loss": 0.0047,
      "step": 14044
    },
    {
      "epoch": 8.98592450415867,
      "grad_norm": 0.3406636030163703,
      "learning_rate": 5.345214149475586e-07,
      "loss": 0.0044,
      "step": 14045
    },
    {
      "epoch": 8.986564299424185,
      "grad_norm": 0.4720857262139089,
      "learning_rate": 5.338532171190946e-07,
      "loss": 0.0054,
      "step": 14046
    },
    {
      "epoch": 8.9872040946897,
      "grad_norm": 0.3904824880561148,
      "learning_rate": 5.331854257501357e-07,
      "loss": 0.0042,
      "step": 14047
    },
    {
      "epoch": 8.987843889955215,
      "grad_norm": 0.440497388206338,
      "learning_rate": 5.325180408693531e-07,
      "loss": 0.0049,
      "step": 14048
    },
    {
      "epoch": 8.988483685220729,
      "grad_norm": 0.4284757769964367,
      "learning_rate": 5.318510625054041e-07,
      "loss": 0.0053,
      "step": 14049
    },
    {
      "epoch": 8.989123480486244,
      "grad_norm": 0.4608777005577883,
      "learning_rate": 5.311844906869291e-07,
      "loss": 0.0056,
      "step": 14050
    },
    {
      "epoch": 8.98976327575176,
      "grad_norm": 0.5449732899974185,
      "learning_rate": 5.305183254425484e-07,
      "loss": 0.0075,
      "step": 14051
    },
    {
      "epoch": 8.990403071017274,
      "grad_norm": 0.36551612803672123,
      "learning_rate": 5.298525668008648e-07,
      "loss": 0.0048,
      "step": 14052
    },
    {
      "epoch": 8.99104286628279,
      "grad_norm": 0.4167851189083351,
      "learning_rate": 5.291872147904664e-07,
      "loss": 0.0048,
      "step": 14053
    },
    {
      "epoch": 8.991682661548305,
      "grad_norm": 0.318370534026376,
      "learning_rate": 5.285222694399228e-07,
      "loss": 0.0034,
      "step": 14054
    },
    {
      "epoch": 8.99232245681382,
      "grad_norm": 0.4298341400858817,
      "learning_rate": 5.278577307777821e-07,
      "loss": 0.0049,
      "step": 14055
    },
    {
      "epoch": 8.992962252079334,
      "grad_norm": 0.49050942035109957,
      "learning_rate": 5.271935988325828e-07,
      "loss": 0.0059,
      "step": 14056
    },
    {
      "epoch": 8.993602047344849,
      "grad_norm": 0.48333212002780185,
      "learning_rate": 5.265298736328395e-07,
      "loss": 0.0068,
      "step": 14057
    },
    {
      "epoch": 8.994241842610364,
      "grad_norm": 0.4918817401673311,
      "learning_rate": 5.258665552070508e-07,
      "loss": 0.0062,
      "step": 14058
    },
    {
      "epoch": 8.99488163787588,
      "grad_norm": 0.43421592345639204,
      "learning_rate": 5.252036435836993e-07,
      "loss": 0.0043,
      "step": 14059
    },
    {
      "epoch": 8.995521433141395,
      "grad_norm": 0.49264183454198657,
      "learning_rate": 5.245411387912491e-07,
      "loss": 0.0049,
      "step": 14060
    },
    {
      "epoch": 8.99616122840691,
      "grad_norm": 0.33429591963617156,
      "learning_rate": 5.238790408581474e-07,
      "loss": 0.0037,
      "step": 14061
    },
    {
      "epoch": 8.996801023672425,
      "grad_norm": 0.47719244553242396,
      "learning_rate": 5.232173498128212e-07,
      "loss": 0.0053,
      "step": 14062
    },
    {
      "epoch": 8.99744081893794,
      "grad_norm": 0.3802897939387877,
      "learning_rate": 5.225560656836859e-07,
      "loss": 0.0046,
      "step": 14063
    },
    {
      "epoch": 8.998080614203454,
      "grad_norm": 0.4396434509311198,
      "learning_rate": 5.218951884991331e-07,
      "loss": 0.0051,
      "step": 14064
    },
    {
      "epoch": 8.99872040946897,
      "grad_norm": 0.6054497341781506,
      "learning_rate": 5.21234718287541e-07,
      "loss": 0.0044,
      "step": 14065
    },
    {
      "epoch": 8.999360204734485,
      "grad_norm": 0.41943785566415903,
      "learning_rate": 5.205746550772684e-07,
      "loss": 0.0046,
      "step": 14066
    },
    {
      "epoch": 9.0,
      "grad_norm": 0.3552155678943211,
      "learning_rate": 5.199149988966578e-07,
      "loss": 0.004,
      "step": 14067
    },
    {
      "epoch": 9.000639795265515,
      "grad_norm": 0.1594783115448008,
      "learning_rate": 5.192557497740336e-07,
      "loss": 0.0013,
      "step": 14068
    },
    {
      "epoch": 9.00127959053103,
      "grad_norm": 0.11706245272497158,
      "learning_rate": 5.185969077377007e-07,
      "loss": 0.0014,
      "step": 14069
    },
    {
      "epoch": 9.001919385796546,
      "grad_norm": 0.14365397373191902,
      "learning_rate": 5.179384728159531e-07,
      "loss": 0.0016,
      "step": 14070
    },
    {
      "epoch": 9.00255918106206,
      "grad_norm": 0.09286939862928183,
      "learning_rate": 5.172804450370594e-07,
      "loss": 0.0011,
      "step": 14071
    },
    {
      "epoch": 9.003198976327575,
      "grad_norm": 0.19157763545164178,
      "learning_rate": 5.166228244292748e-07,
      "loss": 0.0016,
      "step": 14072
    },
    {
      "epoch": 9.00383877159309,
      "grad_norm": 0.10480759309826818,
      "learning_rate": 5.159656110208367e-07,
      "loss": 0.0012,
      "step": 14073
    },
    {
      "epoch": 9.004478566858605,
      "grad_norm": 0.12270075177389057,
      "learning_rate": 5.153088048399646e-07,
      "loss": 0.0013,
      "step": 14074
    },
    {
      "epoch": 9.00511836212412,
      "grad_norm": 0.13138011045014794,
      "learning_rate": 5.146524059148605e-07,
      "loss": 0.0012,
      "step": 14075
    },
    {
      "epoch": 9.005758157389636,
      "grad_norm": 0.13202604444988736,
      "learning_rate": 5.139964142737064e-07,
      "loss": 0.0013,
      "step": 14076
    },
    {
      "epoch": 9.006397952655151,
      "grad_norm": 0.09589774568972101,
      "learning_rate": 5.133408299446752e-07,
      "loss": 0.0011,
      "step": 14077
    },
    {
      "epoch": 9.007037747920666,
      "grad_norm": 0.13121321637024025,
      "learning_rate": 5.126856529559121e-07,
      "loss": 0.0015,
      "step": 14078
    },
    {
      "epoch": 9.00767754318618,
      "grad_norm": 0.13770988289632988,
      "learning_rate": 5.120308833355492e-07,
      "loss": 0.0019,
      "step": 14079
    },
    {
      "epoch": 9.008317338451695,
      "grad_norm": 0.09985441748485817,
      "learning_rate": 5.113765211117039e-07,
      "loss": 0.0012,
      "step": 14080
    },
    {
      "epoch": 9.00895713371721,
      "grad_norm": 0.13638832869418016,
      "learning_rate": 5.107225663124726e-07,
      "loss": 0.0014,
      "step": 14081
    },
    {
      "epoch": 9.009596928982726,
      "grad_norm": 0.13074259399195148,
      "learning_rate": 5.100690189659319e-07,
      "loss": 0.0014,
      "step": 14082
    },
    {
      "epoch": 9.01023672424824,
      "grad_norm": 0.12497644856261005,
      "learning_rate": 5.09415879100148e-07,
      "loss": 0.0013,
      "step": 14083
    },
    {
      "epoch": 9.010876519513756,
      "grad_norm": 0.15577747231879538,
      "learning_rate": 5.08763146743163e-07,
      "loss": 0.0013,
      "step": 14084
    },
    {
      "epoch": 9.011516314779271,
      "grad_norm": 0.1331610580211973,
      "learning_rate": 5.081108219230058e-07,
      "loss": 0.0016,
      "step": 14085
    },
    {
      "epoch": 9.012156110044785,
      "grad_norm": 0.11939767929648297,
      "learning_rate": 5.074589046676837e-07,
      "loss": 0.0013,
      "step": 14086
    },
    {
      "epoch": 9.0127959053103,
      "grad_norm": 0.16998775912246566,
      "learning_rate": 5.06807395005191e-07,
      "loss": 0.0011,
      "step": 14087
    },
    {
      "epoch": 9.013435700575815,
      "grad_norm": 0.14941748633480054,
      "learning_rate": 5.061562929635022e-07,
      "loss": 0.0017,
      "step": 14088
    },
    {
      "epoch": 9.01407549584133,
      "grad_norm": 0.12649645835710063,
      "learning_rate": 5.055055985705748e-07,
      "loss": 0.0015,
      "step": 14089
    },
    {
      "epoch": 9.014715291106846,
      "grad_norm": 0.11357120589469631,
      "learning_rate": 5.048553118543465e-07,
      "loss": 0.0012,
      "step": 14090
    },
    {
      "epoch": 9.015355086372361,
      "grad_norm": 0.10743297733981393,
      "learning_rate": 5.042054328427404e-07,
      "loss": 0.001,
      "step": 14091
    },
    {
      "epoch": 9.015994881637877,
      "grad_norm": 0.15594065287012435,
      "learning_rate": 5.035559615636609e-07,
      "loss": 0.0012,
      "step": 14092
    },
    {
      "epoch": 9.01663467690339,
      "grad_norm": 0.11307295309441254,
      "learning_rate": 5.029068980449958e-07,
      "loss": 0.0011,
      "step": 14093
    },
    {
      "epoch": 9.017274472168905,
      "grad_norm": 0.10495600375683377,
      "learning_rate": 5.022582423146149e-07,
      "loss": 0.001,
      "step": 14094
    },
    {
      "epoch": 9.01791426743442,
      "grad_norm": 0.08916923438089065,
      "learning_rate": 5.016099944003705e-07,
      "loss": 0.0009,
      "step": 14095
    },
    {
      "epoch": 9.018554062699936,
      "grad_norm": 0.10586637315270213,
      "learning_rate": 5.009621543300947e-07,
      "loss": 0.0013,
      "step": 14096
    },
    {
      "epoch": 9.019193857965451,
      "grad_norm": 0.08989291531476941,
      "learning_rate": 5.003147221316085e-07,
      "loss": 0.0009,
      "step": 14097
    },
    {
      "epoch": 9.019833653230966,
      "grad_norm": 0.1667603309969024,
      "learning_rate": 4.996676978327087e-07,
      "loss": 0.0016,
      "step": 14098
    },
    {
      "epoch": 9.020473448496482,
      "grad_norm": 0.11685721517415205,
      "learning_rate": 4.990210814611773e-07,
      "loss": 0.0013,
      "step": 14099
    },
    {
      "epoch": 9.021113243761997,
      "grad_norm": 0.08833056404997991,
      "learning_rate": 4.983748730447802e-07,
      "loss": 0.0009,
      "step": 14100
    },
    {
      "epoch": 9.02175303902751,
      "grad_norm": 0.10903118899109811,
      "learning_rate": 4.977290726112649e-07,
      "loss": 0.0008,
      "step": 14101
    },
    {
      "epoch": 9.022392834293026,
      "grad_norm": 0.13818639822564072,
      "learning_rate": 4.970836801883594e-07,
      "loss": 0.001,
      "step": 14102
    },
    {
      "epoch": 9.023032629558541,
      "grad_norm": 0.15008801725520052,
      "learning_rate": 4.96438695803777e-07,
      "loss": 0.0011,
      "step": 14103
    },
    {
      "epoch": 9.023672424824056,
      "grad_norm": 0.12148418151686101,
      "learning_rate": 4.957941194852112e-07,
      "loss": 0.0014,
      "step": 14104
    },
    {
      "epoch": 9.024312220089572,
      "grad_norm": 0.21210352268122415,
      "learning_rate": 4.951499512603386e-07,
      "loss": 0.0017,
      "step": 14105
    },
    {
      "epoch": 9.024952015355087,
      "grad_norm": 0.21160148505312001,
      "learning_rate": 4.945061911568183e-07,
      "loss": 0.0016,
      "step": 14106
    },
    {
      "epoch": 9.025591810620602,
      "grad_norm": 0.16281582334594089,
      "learning_rate": 4.938628392022948e-07,
      "loss": 0.0012,
      "step": 14107
    },
    {
      "epoch": 9.026231605886116,
      "grad_norm": 0.12912800482498704,
      "learning_rate": 4.932198954243916e-07,
      "loss": 0.001,
      "step": 14108
    },
    {
      "epoch": 9.026871401151631,
      "grad_norm": 0.11821492709271021,
      "learning_rate": 4.925773598507122e-07,
      "loss": 0.0012,
      "step": 14109
    },
    {
      "epoch": 9.027511196417146,
      "grad_norm": 0.13911228573887408,
      "learning_rate": 4.9193523250885e-07,
      "loss": 0.0012,
      "step": 14110
    },
    {
      "epoch": 9.028150991682661,
      "grad_norm": 0.09096437938226083,
      "learning_rate": 4.912935134263763e-07,
      "loss": 0.001,
      "step": 14111
    },
    {
      "epoch": 9.028790786948177,
      "grad_norm": 0.14420803461860313,
      "learning_rate": 4.906522026308425e-07,
      "loss": 0.0013,
      "step": 14112
    },
    {
      "epoch": 9.029430582213692,
      "grad_norm": 0.11415079792974114,
      "learning_rate": 4.900113001497885e-07,
      "loss": 0.0011,
      "step": 14113
    },
    {
      "epoch": 9.030070377479207,
      "grad_norm": 0.1004394269407646,
      "learning_rate": 4.893708060107328e-07,
      "loss": 0.001,
      "step": 14114
    },
    {
      "epoch": 9.030710172744723,
      "grad_norm": 0.12132803591049202,
      "learning_rate": 4.887307202411761e-07,
      "loss": 0.0012,
      "step": 14115
    },
    {
      "epoch": 9.031349968010236,
      "grad_norm": 0.15471307561181388,
      "learning_rate": 4.880910428686026e-07,
      "loss": 0.0011,
      "step": 14116
    },
    {
      "epoch": 9.031989763275751,
      "grad_norm": 0.11962540212010206,
      "learning_rate": 4.874517739204798e-07,
      "loss": 0.0011,
      "step": 14117
    },
    {
      "epoch": 9.032629558541267,
      "grad_norm": 0.09075010292781441,
      "learning_rate": 4.86812913424256e-07,
      "loss": 0.0009,
      "step": 14118
    },
    {
      "epoch": 9.033269353806782,
      "grad_norm": 0.11174345617973803,
      "learning_rate": 4.861744614073638e-07,
      "loss": 0.0011,
      "step": 14119
    },
    {
      "epoch": 9.033909149072297,
      "grad_norm": 0.1270708186545829,
      "learning_rate": 4.855364178972166e-07,
      "loss": 0.0015,
      "step": 14120
    },
    {
      "epoch": 9.034548944337812,
      "grad_norm": 0.12682758699399338,
      "learning_rate": 4.848987829212104e-07,
      "loss": 0.001,
      "step": 14121
    },
    {
      "epoch": 9.035188739603328,
      "grad_norm": 0.14147895072341476,
      "learning_rate": 4.842615565067222e-07,
      "loss": 0.001,
      "step": 14122
    },
    {
      "epoch": 9.035828534868841,
      "grad_norm": 0.14263868448407996,
      "learning_rate": 4.836247386811177e-07,
      "loss": 0.0013,
      "step": 14123
    },
    {
      "epoch": 9.036468330134356,
      "grad_norm": 0.1445658349919938,
      "learning_rate": 4.829883294717396e-07,
      "loss": 0.0013,
      "step": 14124
    },
    {
      "epoch": 9.037108125399872,
      "grad_norm": 0.11732951267212414,
      "learning_rate": 4.823523289059118e-07,
      "loss": 0.0012,
      "step": 14125
    },
    {
      "epoch": 9.037747920665387,
      "grad_norm": 0.14019697680702897,
      "learning_rate": 4.817167370109433e-07,
      "loss": 0.0013,
      "step": 14126
    },
    {
      "epoch": 9.038387715930902,
      "grad_norm": 0.10853951533058301,
      "learning_rate": 4.810815538141278e-07,
      "loss": 0.0011,
      "step": 14127
    },
    {
      "epoch": 9.039027511196418,
      "grad_norm": 0.13627722831870467,
      "learning_rate": 4.804467793427381e-07,
      "loss": 0.0012,
      "step": 14128
    },
    {
      "epoch": 9.039667306461933,
      "grad_norm": 0.18087098370560464,
      "learning_rate": 4.798124136240279e-07,
      "loss": 0.0012,
      "step": 14129
    },
    {
      "epoch": 9.040307101727446,
      "grad_norm": 0.1331604930351011,
      "learning_rate": 4.791784566852386e-07,
      "loss": 0.0012,
      "step": 14130
    },
    {
      "epoch": 9.040946896992962,
      "grad_norm": 0.1300691098521577,
      "learning_rate": 4.785449085535909e-07,
      "loss": 0.001,
      "step": 14131
    },
    {
      "epoch": 9.041586692258477,
      "grad_norm": 0.09733537332791624,
      "learning_rate": 4.779117692562873e-07,
      "loss": 0.001,
      "step": 14132
    },
    {
      "epoch": 9.042226487523992,
      "grad_norm": 0.13370089253810324,
      "learning_rate": 4.772790388205128e-07,
      "loss": 0.0013,
      "step": 14133
    },
    {
      "epoch": 9.042866282789507,
      "grad_norm": 0.14560619163069463,
      "learning_rate": 4.7664671727343793e-07,
      "loss": 0.0011,
      "step": 14134
    },
    {
      "epoch": 9.043506078055023,
      "grad_norm": 0.10746565873818839,
      "learning_rate": 4.7601480464221196e-07,
      "loss": 0.001,
      "step": 14135
    },
    {
      "epoch": 9.044145873320538,
      "grad_norm": 0.09878341618710766,
      "learning_rate": 4.753833009539666e-07,
      "loss": 0.001,
      "step": 14136
    },
    {
      "epoch": 9.044785668586053,
      "grad_norm": 0.12116785081809238,
      "learning_rate": 4.7475220623582117e-07,
      "loss": 0.001,
      "step": 14137
    },
    {
      "epoch": 9.045425463851567,
      "grad_norm": 0.13168162629649643,
      "learning_rate": 4.741215205148719e-07,
      "loss": 0.0012,
      "step": 14138
    },
    {
      "epoch": 9.046065259117082,
      "grad_norm": 0.11504799993582192,
      "learning_rate": 4.734912438181982e-07,
      "loss": 0.0009,
      "step": 14139
    },
    {
      "epoch": 9.046705054382597,
      "grad_norm": 0.16406209160421625,
      "learning_rate": 4.728613761728651e-07,
      "loss": 0.0014,
      "step": 14140
    },
    {
      "epoch": 9.047344849648113,
      "grad_norm": 0.07398549382763915,
      "learning_rate": 4.7223191760591756e-07,
      "loss": 0.0007,
      "step": 14141
    },
    {
      "epoch": 9.047984644913628,
      "grad_norm": 0.12339687725541225,
      "learning_rate": 4.7160286814438184e-07,
      "loss": 0.001,
      "step": 14142
    },
    {
      "epoch": 9.048624440179143,
      "grad_norm": 0.13885438627414043,
      "learning_rate": 4.709742278152685e-07,
      "loss": 0.0012,
      "step": 14143
    },
    {
      "epoch": 9.049264235444658,
      "grad_norm": 0.1121713670511689,
      "learning_rate": 4.703459966455726e-07,
      "loss": 0.001,
      "step": 14144
    },
    {
      "epoch": 9.049904030710172,
      "grad_norm": 0.14757918040237197,
      "learning_rate": 4.697181746622681e-07,
      "loss": 0.0013,
      "step": 14145
    },
    {
      "epoch": 9.050543825975687,
      "grad_norm": 0.17580718953554578,
      "learning_rate": 4.690907618923102e-07,
      "loss": 0.0013,
      "step": 14146
    },
    {
      "epoch": 9.051183621241202,
      "grad_norm": 0.33411431277868187,
      "learning_rate": 4.6846375836264167e-07,
      "loss": 0.0018,
      "step": 14147
    },
    {
      "epoch": 9.051823416506718,
      "grad_norm": 0.11630236034528375,
      "learning_rate": 4.6783716410018443e-07,
      "loss": 0.0012,
      "step": 14148
    },
    {
      "epoch": 9.052463211772233,
      "grad_norm": 0.1760942864474372,
      "learning_rate": 4.6721097913184244e-07,
      "loss": 0.0013,
      "step": 14149
    },
    {
      "epoch": 9.053103007037748,
      "grad_norm": 0.10267884885085565,
      "learning_rate": 4.665852034845042e-07,
      "loss": 0.0009,
      "step": 14150
    },
    {
      "epoch": 9.053742802303264,
      "grad_norm": 0.14787161273539953,
      "learning_rate": 4.6595983718503934e-07,
      "loss": 0.0012,
      "step": 14151
    },
    {
      "epoch": 9.054382597568779,
      "grad_norm": 0.20488292669056066,
      "learning_rate": 4.653348802602986e-07,
      "loss": 0.0011,
      "step": 14152
    },
    {
      "epoch": 9.055022392834292,
      "grad_norm": 0.0947431535844544,
      "learning_rate": 4.64710332737115e-07,
      "loss": 0.0009,
      "step": 14153
    },
    {
      "epoch": 9.055662188099808,
      "grad_norm": 0.20034210008920156,
      "learning_rate": 4.640861946423103e-07,
      "loss": 0.0012,
      "step": 14154
    },
    {
      "epoch": 9.056301983365323,
      "grad_norm": 0.19336279998435535,
      "learning_rate": 4.6346246600268096e-07,
      "loss": 0.0014,
      "step": 14155
    },
    {
      "epoch": 9.056941778630838,
      "grad_norm": 0.1070794916363582,
      "learning_rate": 4.6283914684500775e-07,
      "loss": 0.001,
      "step": 14156
    },
    {
      "epoch": 9.057581573896353,
      "grad_norm": 0.07775397967613455,
      "learning_rate": 4.6221623719605814e-07,
      "loss": 0.0008,
      "step": 14157
    },
    {
      "epoch": 9.058221369161869,
      "grad_norm": 0.15494302838006963,
      "learning_rate": 4.615937370825763e-07,
      "loss": 0.0012,
      "step": 14158
    },
    {
      "epoch": 9.058861164427384,
      "grad_norm": 0.07851082787726905,
      "learning_rate": 4.609716465312908e-07,
      "loss": 0.0008,
      "step": 14159
    },
    {
      "epoch": 9.059500959692897,
      "grad_norm": 0.1280328715252289,
      "learning_rate": 4.6034996556891476e-07,
      "loss": 0.001,
      "step": 14160
    },
    {
      "epoch": 9.060140754958413,
      "grad_norm": 0.08864871834497995,
      "learning_rate": 4.597286942221424e-07,
      "loss": 0.0009,
      "step": 14161
    },
    {
      "epoch": 9.060780550223928,
      "grad_norm": 0.15571063835140975,
      "learning_rate": 4.59107832517649e-07,
      "loss": 0.0013,
      "step": 14162
    },
    {
      "epoch": 9.061420345489443,
      "grad_norm": 0.08118502078001814,
      "learning_rate": 4.584873804820933e-07,
      "loss": 0.0008,
      "step": 14163
    },
    {
      "epoch": 9.062060140754959,
      "grad_norm": 0.11575571813052911,
      "learning_rate": 4.5786733814211613e-07,
      "loss": 0.0012,
      "step": 14164
    },
    {
      "epoch": 9.062699936020474,
      "grad_norm": 0.08329098469564246,
      "learning_rate": 4.5724770552434186e-07,
      "loss": 0.0008,
      "step": 14165
    },
    {
      "epoch": 9.06333973128599,
      "grad_norm": 0.10569539044537632,
      "learning_rate": 4.566284826553735e-07,
      "loss": 0.0009,
      "step": 14166
    },
    {
      "epoch": 9.063979526551503,
      "grad_norm": 0.11057308482864785,
      "learning_rate": 4.560096695618044e-07,
      "loss": 0.001,
      "step": 14167
    },
    {
      "epoch": 9.064619321817018,
      "grad_norm": 0.11425376994875557,
      "learning_rate": 4.55391266270202e-07,
      "loss": 0.001,
      "step": 14168
    },
    {
      "epoch": 9.065259117082533,
      "grad_norm": 0.11774161905891771,
      "learning_rate": 4.547732728071186e-07,
      "loss": 0.0011,
      "step": 14169
    },
    {
      "epoch": 9.065898912348048,
      "grad_norm": 0.10399120899445666,
      "learning_rate": 4.5415568919909283e-07,
      "loss": 0.0008,
      "step": 14170
    },
    {
      "epoch": 9.066538707613564,
      "grad_norm": 0.06359899446143726,
      "learning_rate": 4.535385154726413e-07,
      "loss": 0.0007,
      "step": 14171
    },
    {
      "epoch": 9.067178502879079,
      "grad_norm": 0.145844005144512,
      "learning_rate": 4.5292175165426276e-07,
      "loss": 0.0012,
      "step": 14172
    },
    {
      "epoch": 9.067818298144594,
      "grad_norm": 0.18836095489512025,
      "learning_rate": 4.5230539777044167e-07,
      "loss": 0.0016,
      "step": 14173
    },
    {
      "epoch": 9.06845809341011,
      "grad_norm": 0.13434497491837977,
      "learning_rate": 4.516894538476424e-07,
      "loss": 0.0013,
      "step": 14174
    },
    {
      "epoch": 9.069097888675623,
      "grad_norm": 0.13916571149884924,
      "learning_rate": 4.5107391991231373e-07,
      "loss": 0.0011,
      "step": 14175
    },
    {
      "epoch": 9.069737683941138,
      "grad_norm": 0.12449997544861487,
      "learning_rate": 4.504587959908835e-07,
      "loss": 0.001,
      "step": 14176
    },
    {
      "epoch": 9.070377479206654,
      "grad_norm": 0.1025390037325357,
      "learning_rate": 4.4984408210976715e-07,
      "loss": 0.0009,
      "step": 14177
    },
    {
      "epoch": 9.071017274472169,
      "grad_norm": 0.08371641706267174,
      "learning_rate": 4.4922977829535694e-07,
      "loss": 0.0008,
      "step": 14178
    },
    {
      "epoch": 9.071657069737684,
      "grad_norm": 0.09915520701971851,
      "learning_rate": 4.4861588457403074e-07,
      "loss": 0.0009,
      "step": 14179
    },
    {
      "epoch": 9.0722968650032,
      "grad_norm": 0.12633569598155261,
      "learning_rate": 4.480024009721473e-07,
      "loss": 0.001,
      "step": 14180
    },
    {
      "epoch": 9.072936660268715,
      "grad_norm": 0.1303923378112897,
      "learning_rate": 4.4738932751604903e-07,
      "loss": 0.001,
      "step": 14181
    },
    {
      "epoch": 9.073576455534228,
      "grad_norm": 0.15876149048227184,
      "learning_rate": 4.4677666423206035e-07,
      "loss": 0.0016,
      "step": 14182
    },
    {
      "epoch": 9.074216250799743,
      "grad_norm": 0.160546740665002,
      "learning_rate": 4.461644111464869e-07,
      "loss": 0.0012,
      "step": 14183
    },
    {
      "epoch": 9.074856046065259,
      "grad_norm": 0.1443740974688424,
      "learning_rate": 4.455525682856199e-07,
      "loss": 0.0011,
      "step": 14184
    },
    {
      "epoch": 9.075495841330774,
      "grad_norm": 0.11508751226710356,
      "learning_rate": 4.449411356757294e-07,
      "loss": 0.001,
      "step": 14185
    },
    {
      "epoch": 9.07613563659629,
      "grad_norm": 0.19049280856224649,
      "learning_rate": 4.443301133430689e-07,
      "loss": 0.0012,
      "step": 14186
    },
    {
      "epoch": 9.076775431861805,
      "grad_norm": 0.18393031370989457,
      "learning_rate": 4.437195013138751e-07,
      "loss": 0.0016,
      "step": 14187
    },
    {
      "epoch": 9.07741522712732,
      "grad_norm": 0.10365351192393496,
      "learning_rate": 4.4310929961436713e-07,
      "loss": 0.0009,
      "step": 14188
    },
    {
      "epoch": 9.078055022392835,
      "grad_norm": 0.10918595502817047,
      "learning_rate": 4.4249950827074504e-07,
      "loss": 0.0009,
      "step": 14189
    },
    {
      "epoch": 9.078694817658349,
      "grad_norm": 0.21060863458331047,
      "learning_rate": 4.4189012730919247e-07,
      "loss": 0.0012,
      "step": 14190
    },
    {
      "epoch": 9.079334612923864,
      "grad_norm": 0.12139940759660528,
      "learning_rate": 4.412811567558761e-07,
      "loss": 0.0009,
      "step": 14191
    },
    {
      "epoch": 9.07997440818938,
      "grad_norm": 0.13321546541220994,
      "learning_rate": 4.40672596636943e-07,
      "loss": 0.001,
      "step": 14192
    },
    {
      "epoch": 9.080614203454894,
      "grad_norm": 0.1541935836394607,
      "learning_rate": 4.4006444697852424e-07,
      "loss": 0.0013,
      "step": 14193
    },
    {
      "epoch": 9.08125399872041,
      "grad_norm": 0.23838217605821044,
      "learning_rate": 4.3945670780673245e-07,
      "loss": 0.0015,
      "step": 14194
    },
    {
      "epoch": 9.081893793985925,
      "grad_norm": 0.1610954167693601,
      "learning_rate": 4.3884937914766226e-07,
      "loss": 0.0011,
      "step": 14195
    },
    {
      "epoch": 9.08253358925144,
      "grad_norm": 0.17001157622773566,
      "learning_rate": 4.382424610273905e-07,
      "loss": 0.0012,
      "step": 14196
    },
    {
      "epoch": 9.083173384516954,
      "grad_norm": 0.12428799267868003,
      "learning_rate": 4.3763595347197875e-07,
      "loss": 0.0009,
      "step": 14197
    },
    {
      "epoch": 9.083813179782469,
      "grad_norm": 0.15237835711122769,
      "learning_rate": 4.370298565074704e-07,
      "loss": 0.0009,
      "step": 14198
    },
    {
      "epoch": 9.084452975047984,
      "grad_norm": 0.1405092201084759,
      "learning_rate": 4.36424170159887e-07,
      "loss": 0.0014,
      "step": 14199
    },
    {
      "epoch": 9.0850927703135,
      "grad_norm": 0.09578904251856493,
      "learning_rate": 4.358188944552388e-07,
      "loss": 0.0009,
      "step": 14200
    },
    {
      "epoch": 9.085732565579015,
      "grad_norm": 0.2696581927594159,
      "learning_rate": 4.3521402941951397e-07,
      "loss": 0.0015,
      "step": 14201
    },
    {
      "epoch": 9.08637236084453,
      "grad_norm": 0.13088073748170043,
      "learning_rate": 4.3460957507868496e-07,
      "loss": 0.0011,
      "step": 14202
    },
    {
      "epoch": 9.087012156110045,
      "grad_norm": 0.11066507021907299,
      "learning_rate": 4.340055314587033e-07,
      "loss": 0.0011,
      "step": 14203
    },
    {
      "epoch": 9.08765195137556,
      "grad_norm": 0.09622258747883536,
      "learning_rate": 4.334018985855082e-07,
      "loss": 0.0009,
      "step": 14204
    },
    {
      "epoch": 9.088291746641074,
      "grad_norm": 0.1249731604946111,
      "learning_rate": 4.327986764850189e-07,
      "loss": 0.0009,
      "step": 14205
    },
    {
      "epoch": 9.08893154190659,
      "grad_norm": 0.11959219165828275,
      "learning_rate": 4.321958651831337e-07,
      "loss": 0.0009,
      "step": 14206
    },
    {
      "epoch": 9.089571337172105,
      "grad_norm": 0.16035657817890098,
      "learning_rate": 4.315934647057396e-07,
      "loss": 0.0012,
      "step": 14207
    },
    {
      "epoch": 9.09021113243762,
      "grad_norm": 0.11866459739252477,
      "learning_rate": 4.309914750787025e-07,
      "loss": 0.001,
      "step": 14208
    },
    {
      "epoch": 9.090850927703135,
      "grad_norm": 0.14711064169070578,
      "learning_rate": 4.3038989632786853e-07,
      "loss": 0.0009,
      "step": 14209
    },
    {
      "epoch": 9.09149072296865,
      "grad_norm": 0.1587709635014603,
      "learning_rate": 4.2978872847906916e-07,
      "loss": 0.0011,
      "step": 14210
    },
    {
      "epoch": 9.092130518234166,
      "grad_norm": 0.1084620996492171,
      "learning_rate": 4.291879715581182e-07,
      "loss": 0.001,
      "step": 14211
    },
    {
      "epoch": 9.09277031349968,
      "grad_norm": 0.08527571482306111,
      "learning_rate": 4.285876255908106e-07,
      "loss": 0.0008,
      "step": 14212
    },
    {
      "epoch": 9.093410108765195,
      "grad_norm": 0.11636713149680411,
      "learning_rate": 4.2798769060292233e-07,
      "loss": 0.0011,
      "step": 14213
    },
    {
      "epoch": 9.09404990403071,
      "grad_norm": 0.20459723729211274,
      "learning_rate": 4.273881666202162e-07,
      "loss": 0.0018,
      "step": 14214
    },
    {
      "epoch": 9.094689699296225,
      "grad_norm": 0.06682856395893845,
      "learning_rate": 4.267890536684349e-07,
      "loss": 0.0007,
      "step": 14215
    },
    {
      "epoch": 9.09532949456174,
      "grad_norm": 0.22367766204039166,
      "learning_rate": 4.2619035177330014e-07,
      "loss": 0.0015,
      "step": 14216
    },
    {
      "epoch": 9.095969289827256,
      "grad_norm": 0.220784904611136,
      "learning_rate": 4.255920609605235e-07,
      "loss": 0.0012,
      "step": 14217
    },
    {
      "epoch": 9.096609085092771,
      "grad_norm": 0.14150734345770788,
      "learning_rate": 4.2499418125579116e-07,
      "loss": 0.0012,
      "step": 14218
    },
    {
      "epoch": 9.097248880358284,
      "grad_norm": 0.12829680403885418,
      "learning_rate": 4.24396712684777e-07,
      "loss": 0.001,
      "step": 14219
    },
    {
      "epoch": 9.0978886756238,
      "grad_norm": 0.13007780170150787,
      "learning_rate": 4.237996552731327e-07,
      "loss": 0.0012,
      "step": 14220
    },
    {
      "epoch": 9.098528470889315,
      "grad_norm": 0.13203113212387552,
      "learning_rate": 4.232030090464978e-07,
      "loss": 0.001,
      "step": 14221
    },
    {
      "epoch": 9.09916826615483,
      "grad_norm": 0.12663134922066452,
      "learning_rate": 4.2260677403048953e-07,
      "loss": 0.0009,
      "step": 14222
    },
    {
      "epoch": 9.099808061420346,
      "grad_norm": 0.1421364964464744,
      "learning_rate": 4.220109502507097e-07,
      "loss": 0.0011,
      "step": 14223
    },
    {
      "epoch": 9.10044785668586,
      "grad_norm": 0.15865843534264906,
      "learning_rate": 4.2141553773274113e-07,
      "loss": 0.0011,
      "step": 14224
    },
    {
      "epoch": 9.101087651951376,
      "grad_norm": 0.14865876720716112,
      "learning_rate": 4.2082053650215117e-07,
      "loss": 0.0009,
      "step": 14225
    },
    {
      "epoch": 9.101727447216891,
      "grad_norm": 0.06737878197731292,
      "learning_rate": 4.20225946584486e-07,
      "loss": 0.0007,
      "step": 14226
    },
    {
      "epoch": 9.102367242482405,
      "grad_norm": 0.10206663876245155,
      "learning_rate": 4.196317680052786e-07,
      "loss": 0.0009,
      "step": 14227
    },
    {
      "epoch": 9.10300703774792,
      "grad_norm": 0.11766347827093893,
      "learning_rate": 4.1903800079004073e-07,
      "loss": 0.001,
      "step": 14228
    },
    {
      "epoch": 9.103646833013435,
      "grad_norm": 0.14838618493636024,
      "learning_rate": 4.1844464496426876e-07,
      "loss": 0.0011,
      "step": 14229
    },
    {
      "epoch": 9.10428662827895,
      "grad_norm": 0.14709866759916784,
      "learning_rate": 4.178517005534377e-07,
      "loss": 0.0012,
      "step": 14230
    },
    {
      "epoch": 9.104926423544466,
      "grad_norm": 0.11877857524150974,
      "learning_rate": 4.172591675830107e-07,
      "loss": 0.001,
      "step": 14231
    },
    {
      "epoch": 9.105566218809981,
      "grad_norm": 0.2317476009185684,
      "learning_rate": 4.1666704607842833e-07,
      "loss": 0.0011,
      "step": 14232
    },
    {
      "epoch": 9.106206014075497,
      "grad_norm": 0.09401708375630059,
      "learning_rate": 4.1607533606511484e-07,
      "loss": 0.0009,
      "step": 14233
    },
    {
      "epoch": 9.10684580934101,
      "grad_norm": 0.10930272213872058,
      "learning_rate": 4.154840375684788e-07,
      "loss": 0.0009,
      "step": 14234
    },
    {
      "epoch": 9.107485604606525,
      "grad_norm": 0.14267424057077902,
      "learning_rate": 4.1489315061390864e-07,
      "loss": 0.0013,
      "step": 14235
    },
    {
      "epoch": 9.10812539987204,
      "grad_norm": 0.08462015270353752,
      "learning_rate": 4.1430267522677424e-07,
      "loss": 0.0007,
      "step": 14236
    },
    {
      "epoch": 9.108765195137556,
      "grad_norm": 0.09532028106392834,
      "learning_rate": 4.1371261143243304e-07,
      "loss": 0.0009,
      "step": 14237
    },
    {
      "epoch": 9.109404990403071,
      "grad_norm": 0.10686079862219937,
      "learning_rate": 4.1312295925622024e-07,
      "loss": 0.001,
      "step": 14238
    },
    {
      "epoch": 9.110044785668586,
      "grad_norm": 0.09594399184313247,
      "learning_rate": 4.125337187234535e-07,
      "loss": 0.0008,
      "step": 14239
    },
    {
      "epoch": 9.110684580934102,
      "grad_norm": 0.2472565040878195,
      "learning_rate": 4.119448898594336e-07,
      "loss": 0.0012,
      "step": 14240
    },
    {
      "epoch": 9.111324376199617,
      "grad_norm": 0.201235225174458,
      "learning_rate": 4.113564726894448e-07,
      "loss": 0.0014,
      "step": 14241
    },
    {
      "epoch": 9.11196417146513,
      "grad_norm": 0.12828943886070443,
      "learning_rate": 4.107684672387524e-07,
      "loss": 0.001,
      "step": 14242
    },
    {
      "epoch": 9.112603966730646,
      "grad_norm": 0.07288607754548328,
      "learning_rate": 4.1018087353260185e-07,
      "loss": 0.0007,
      "step": 14243
    },
    {
      "epoch": 9.113243761996161,
      "grad_norm": 0.10267645215185352,
      "learning_rate": 4.095936915962284e-07,
      "loss": 0.0009,
      "step": 14244
    },
    {
      "epoch": 9.113883557261676,
      "grad_norm": 0.15188843811820305,
      "learning_rate": 4.090069214548409e-07,
      "loss": 0.0012,
      "step": 14245
    },
    {
      "epoch": 9.114523352527192,
      "grad_norm": 0.24729945896180178,
      "learning_rate": 4.0842056313363356e-07,
      "loss": 0.0017,
      "step": 14246
    },
    {
      "epoch": 9.115163147792707,
      "grad_norm": 0.13108601951263468,
      "learning_rate": 4.0783461665778623e-07,
      "loss": 0.0009,
      "step": 14247
    },
    {
      "epoch": 9.115802943058222,
      "grad_norm": 0.17008800932327886,
      "learning_rate": 4.072490820524588e-07,
      "loss": 0.0012,
      "step": 14248
    },
    {
      "epoch": 9.116442738323736,
      "grad_norm": 0.1675074523982011,
      "learning_rate": 4.0666395934279014e-07,
      "loss": 0.0012,
      "step": 14249
    },
    {
      "epoch": 9.117082533589251,
      "grad_norm": 0.10095258737427913,
      "learning_rate": 4.0607924855390557e-07,
      "loss": 0.0008,
      "step": 14250
    },
    {
      "epoch": 9.117722328854766,
      "grad_norm": 0.09477154450923475,
      "learning_rate": 4.0549494971091175e-07,
      "loss": 0.0008,
      "step": 14251
    },
    {
      "epoch": 9.118362124120281,
      "grad_norm": 0.13484904816885676,
      "learning_rate": 4.049110628388986e-07,
      "loss": 0.0012,
      "step": 14252
    },
    {
      "epoch": 9.119001919385797,
      "grad_norm": 0.14357473262559167,
      "learning_rate": 4.04327587962936e-07,
      "loss": 0.0012,
      "step": 14253
    },
    {
      "epoch": 9.119641714651312,
      "grad_norm": 0.18640595430437235,
      "learning_rate": 4.0374452510807737e-07,
      "loss": 0.0012,
      "step": 14254
    },
    {
      "epoch": 9.120281509916827,
      "grad_norm": 0.09275512036736487,
      "learning_rate": 4.0316187429935816e-07,
      "loss": 0.0009,
      "step": 14255
    },
    {
      "epoch": 9.12092130518234,
      "grad_norm": 0.13884512340112679,
      "learning_rate": 4.02579635561795e-07,
      "loss": 0.0011,
      "step": 14256
    },
    {
      "epoch": 9.121561100447856,
      "grad_norm": 0.08223743537835289,
      "learning_rate": 4.019978089203924e-07,
      "loss": 0.0008,
      "step": 14257
    },
    {
      "epoch": 9.122200895713371,
      "grad_norm": 0.10699629516844522,
      "learning_rate": 4.0141639440012926e-07,
      "loss": 0.0009,
      "step": 14258
    },
    {
      "epoch": 9.122840690978887,
      "grad_norm": 0.0856537370554797,
      "learning_rate": 4.008353920259711e-07,
      "loss": 0.0008,
      "step": 14259
    },
    {
      "epoch": 9.123480486244402,
      "grad_norm": 0.09818406642241262,
      "learning_rate": 4.002548018228658e-07,
      "loss": 0.0008,
      "step": 14260
    },
    {
      "epoch": 9.124120281509917,
      "grad_norm": 0.1586248647411751,
      "learning_rate": 3.996746238157434e-07,
      "loss": 0.0009,
      "step": 14261
    },
    {
      "epoch": 9.124760076775432,
      "grad_norm": 0.14364807403475227,
      "learning_rate": 3.9909485802951397e-07,
      "loss": 0.0008,
      "step": 14262
    },
    {
      "epoch": 9.125399872040948,
      "grad_norm": 0.1255790378610287,
      "learning_rate": 3.9851550448907316e-07,
      "loss": 0.0012,
      "step": 14263
    },
    {
      "epoch": 9.126039667306461,
      "grad_norm": 0.09363674570936607,
      "learning_rate": 3.979365632192966e-07,
      "loss": 0.0008,
      "step": 14264
    },
    {
      "epoch": 9.126679462571976,
      "grad_norm": 0.11246464674500099,
      "learning_rate": 3.973580342450445e-07,
      "loss": 0.001,
      "step": 14265
    },
    {
      "epoch": 9.127319257837492,
      "grad_norm": 0.12561594098641676,
      "learning_rate": 3.967799175911546e-07,
      "loss": 0.0011,
      "step": 14266
    },
    {
      "epoch": 9.127959053103007,
      "grad_norm": 0.14117431182447168,
      "learning_rate": 3.9620221328245277e-07,
      "loss": 0.0011,
      "step": 14267
    },
    {
      "epoch": 9.128598848368522,
      "grad_norm": 0.11128465579437322,
      "learning_rate": 3.9562492134374577e-07,
      "loss": 0.0012,
      "step": 14268
    },
    {
      "epoch": 9.129238643634038,
      "grad_norm": 0.15404123732810246,
      "learning_rate": 3.9504804179981813e-07,
      "loss": 0.0011,
      "step": 14269
    },
    {
      "epoch": 9.129878438899553,
      "grad_norm": 0.13593277457879305,
      "learning_rate": 3.944715746754435e-07,
      "loss": 0.0008,
      "step": 14270
    },
    {
      "epoch": 9.130518234165066,
      "grad_norm": 0.08307059529140795,
      "learning_rate": 3.938955199953709e-07,
      "loss": 0.0007,
      "step": 14271
    },
    {
      "epoch": 9.131158029430582,
      "grad_norm": 0.15996438039355992,
      "learning_rate": 3.933198777843372e-07,
      "loss": 0.0011,
      "step": 14272
    },
    {
      "epoch": 9.131797824696097,
      "grad_norm": 0.12798250876074127,
      "learning_rate": 3.9274464806705826e-07,
      "loss": 0.0009,
      "step": 14273
    },
    {
      "epoch": 9.132437619961612,
      "grad_norm": 0.0958104074730341,
      "learning_rate": 3.9216983086823536e-07,
      "loss": 0.0008,
      "step": 14274
    },
    {
      "epoch": 9.133077415227127,
      "grad_norm": 0.1357838939876697,
      "learning_rate": 3.9159542621254875e-07,
      "loss": 0.0011,
      "step": 14275
    },
    {
      "epoch": 9.133717210492643,
      "grad_norm": 0.19579969187614005,
      "learning_rate": 3.91021434124661e-07,
      "loss": 0.0011,
      "step": 14276
    },
    {
      "epoch": 9.134357005758158,
      "grad_norm": 0.102458469522811,
      "learning_rate": 3.904478546292212e-07,
      "loss": 0.0009,
      "step": 14277
    },
    {
      "epoch": 9.134996801023673,
      "grad_norm": 0.10979870071160201,
      "learning_rate": 3.8987468775085637e-07,
      "loss": 0.0009,
      "step": 14278
    },
    {
      "epoch": 9.135636596289187,
      "grad_norm": 0.12436988670873168,
      "learning_rate": 3.893019335141779e-07,
      "loss": 0.0009,
      "step": 14279
    },
    {
      "epoch": 9.136276391554702,
      "grad_norm": 0.11908945393257162,
      "learning_rate": 3.8872959194377614e-07,
      "loss": 0.001,
      "step": 14280
    },
    {
      "epoch": 9.136916186820217,
      "grad_norm": 0.09930607534952053,
      "learning_rate": 3.881576630642303e-07,
      "loss": 0.0009,
      "step": 14281
    },
    {
      "epoch": 9.137555982085733,
      "grad_norm": 0.10044094719132417,
      "learning_rate": 3.875861469000963e-07,
      "loss": 0.0008,
      "step": 14282
    },
    {
      "epoch": 9.138195777351248,
      "grad_norm": 0.08911078717416583,
      "learning_rate": 3.8701504347591457e-07,
      "loss": 0.0008,
      "step": 14283
    },
    {
      "epoch": 9.138835572616763,
      "grad_norm": 0.1431698938233496,
      "learning_rate": 3.864443528162054e-07,
      "loss": 0.0011,
      "step": 14284
    },
    {
      "epoch": 9.139475367882278,
      "grad_norm": 0.16167013233700075,
      "learning_rate": 3.858740749454759e-07,
      "loss": 0.0013,
      "step": 14285
    },
    {
      "epoch": 9.140115163147792,
      "grad_norm": 0.11889204376603792,
      "learning_rate": 3.853042098882087e-07,
      "loss": 0.0009,
      "step": 14286
    },
    {
      "epoch": 9.140754958413307,
      "grad_norm": 0.17006844021423545,
      "learning_rate": 3.847347576688776e-07,
      "loss": 0.0012,
      "step": 14287
    },
    {
      "epoch": 9.141394753678822,
      "grad_norm": 0.16681947402612546,
      "learning_rate": 3.8416571831193186e-07,
      "loss": 0.0012,
      "step": 14288
    },
    {
      "epoch": 9.142034548944338,
      "grad_norm": 0.14758252563969934,
      "learning_rate": 3.835970918418053e-07,
      "loss": 0.0012,
      "step": 14289
    },
    {
      "epoch": 9.142674344209853,
      "grad_norm": 0.12312676641266664,
      "learning_rate": 3.8302887828291167e-07,
      "loss": 0.001,
      "step": 14290
    },
    {
      "epoch": 9.143314139475368,
      "grad_norm": 0.15683901911908046,
      "learning_rate": 3.8246107765965157e-07,
      "loss": 0.0013,
      "step": 14291
    },
    {
      "epoch": 9.143953934740884,
      "grad_norm": 0.17478306124330037,
      "learning_rate": 3.8189368999640544e-07,
      "loss": 0.0014,
      "step": 14292
    },
    {
      "epoch": 9.144593730006397,
      "grad_norm": 0.11382758206142589,
      "learning_rate": 3.813267153175326e-07,
      "loss": 0.001,
      "step": 14293
    },
    {
      "epoch": 9.145233525271912,
      "grad_norm": 0.13329971591308223,
      "learning_rate": 3.807601536473826e-07,
      "loss": 0.0011,
      "step": 14294
    },
    {
      "epoch": 9.145873320537428,
      "grad_norm": 0.10139824344715173,
      "learning_rate": 3.8019400501027926e-07,
      "loss": 0.0008,
      "step": 14295
    },
    {
      "epoch": 9.146513115802943,
      "grad_norm": 0.11763442447920067,
      "learning_rate": 3.7962826943053423e-07,
      "loss": 0.0009,
      "step": 14296
    },
    {
      "epoch": 9.147152911068458,
      "grad_norm": 0.1598008863315075,
      "learning_rate": 3.7906294693243584e-07,
      "loss": 0.0013,
      "step": 14297
    },
    {
      "epoch": 9.147792706333973,
      "grad_norm": 0.1634903466552253,
      "learning_rate": 3.7849803754026136e-07,
      "loss": 0.0014,
      "step": 14298
    },
    {
      "epoch": 9.148432501599489,
      "grad_norm": 0.1372223912679645,
      "learning_rate": 3.77933541278267e-07,
      "loss": 0.0011,
      "step": 14299
    },
    {
      "epoch": 9.149072296865004,
      "grad_norm": 0.1428804963021998,
      "learning_rate": 3.7736945817068883e-07,
      "loss": 0.0011,
      "step": 14300
    },
    {
      "epoch": 9.149712092130518,
      "grad_norm": 0.09281014282347298,
      "learning_rate": 3.7680578824174973e-07,
      "loss": 0.0008,
      "step": 14301
    },
    {
      "epoch": 9.150351887396033,
      "grad_norm": 0.09372140263909667,
      "learning_rate": 3.762425315156504e-07,
      "loss": 0.0008,
      "step": 14302
    },
    {
      "epoch": 9.150991682661548,
      "grad_norm": 0.25229845782231225,
      "learning_rate": 3.7567968801657693e-07,
      "loss": 0.0015,
      "step": 14303
    },
    {
      "epoch": 9.151631477927063,
      "grad_norm": 0.14947296885747055,
      "learning_rate": 3.75117257768699e-07,
      "loss": 0.0012,
      "step": 14304
    },
    {
      "epoch": 9.152271273192579,
      "grad_norm": 0.09129768422752442,
      "learning_rate": 3.745552407961639e-07,
      "loss": 0.0009,
      "step": 14305
    },
    {
      "epoch": 9.152911068458094,
      "grad_norm": 0.1150694201937823,
      "learning_rate": 3.7399363712310564e-07,
      "loss": 0.0009,
      "step": 14306
    },
    {
      "epoch": 9.15355086372361,
      "grad_norm": 0.09163828159824125,
      "learning_rate": 3.73432446773635e-07,
      "loss": 0.0008,
      "step": 14307
    },
    {
      "epoch": 9.154190658989123,
      "grad_norm": 0.1391754958751605,
      "learning_rate": 3.7287166977185263e-07,
      "loss": 0.0009,
      "step": 14308
    },
    {
      "epoch": 9.154830454254638,
      "grad_norm": 0.27988964580389736,
      "learning_rate": 3.723113061418349e-07,
      "loss": 0.0011,
      "step": 14309
    },
    {
      "epoch": 9.155470249520153,
      "grad_norm": 0.19827809308214572,
      "learning_rate": 3.7175135590764246e-07,
      "loss": 0.0012,
      "step": 14310
    },
    {
      "epoch": 9.156110044785668,
      "grad_norm": 0.09332646074347002,
      "learning_rate": 3.711918190933206e-07,
      "loss": 0.0008,
      "step": 14311
    },
    {
      "epoch": 9.156749840051184,
      "grad_norm": 0.08149517749848227,
      "learning_rate": 3.706326957228934e-07,
      "loss": 0.0008,
      "step": 14312
    },
    {
      "epoch": 9.157389635316699,
      "grad_norm": 0.12187789827924425,
      "learning_rate": 3.7007398582036833e-07,
      "loss": 0.0009,
      "step": 14313
    },
    {
      "epoch": 9.158029430582214,
      "grad_norm": 0.1376269815355712,
      "learning_rate": 3.695156894097374e-07,
      "loss": 0.0009,
      "step": 14314
    },
    {
      "epoch": 9.15866922584773,
      "grad_norm": 0.11378131013059033,
      "learning_rate": 3.689578065149713e-07,
      "loss": 0.001,
      "step": 14315
    },
    {
      "epoch": 9.159309021113243,
      "grad_norm": 0.112050297074594,
      "learning_rate": 3.684003371600242e-07,
      "loss": 0.0011,
      "step": 14316
    },
    {
      "epoch": 9.159948816378758,
      "grad_norm": 0.18033005263226523,
      "learning_rate": 3.6784328136883153e-07,
      "loss": 0.0009,
      "step": 14317
    },
    {
      "epoch": 9.160588611644274,
      "grad_norm": 0.174564971771807,
      "learning_rate": 3.6728663916531625e-07,
      "loss": 0.0012,
      "step": 14318
    },
    {
      "epoch": 9.161228406909789,
      "grad_norm": 0.1454935750987848,
      "learning_rate": 3.667304105733771e-07,
      "loss": 0.0012,
      "step": 14319
    },
    {
      "epoch": 9.161868202175304,
      "grad_norm": 0.14152122674898096,
      "learning_rate": 3.6617459561689605e-07,
      "loss": 0.001,
      "step": 14320
    },
    {
      "epoch": 9.16250799744082,
      "grad_norm": 0.11276414198204932,
      "learning_rate": 3.656191943197429e-07,
      "loss": 0.0009,
      "step": 14321
    },
    {
      "epoch": 9.163147792706335,
      "grad_norm": 0.0810776916575295,
      "learning_rate": 3.65064206705763e-07,
      "loss": 0.0008,
      "step": 14322
    },
    {
      "epoch": 9.163787587971848,
      "grad_norm": 0.09933070123097877,
      "learning_rate": 3.6450963279878517e-07,
      "loss": 0.0008,
      "step": 14323
    },
    {
      "epoch": 9.164427383237364,
      "grad_norm": 0.06759451159327495,
      "learning_rate": 3.639554726226258e-07,
      "loss": 0.0007,
      "step": 14324
    },
    {
      "epoch": 9.165067178502879,
      "grad_norm": 0.12582532829865453,
      "learning_rate": 3.6340172620107583e-07,
      "loss": 0.0011,
      "step": 14325
    },
    {
      "epoch": 9.165706973768394,
      "grad_norm": 0.1952446628775912,
      "learning_rate": 3.628483935579141e-07,
      "loss": 0.0014,
      "step": 14326
    },
    {
      "epoch": 9.16634676903391,
      "grad_norm": 0.12056329467556104,
      "learning_rate": 3.622954747168983e-07,
      "loss": 0.0009,
      "step": 14327
    },
    {
      "epoch": 9.166986564299425,
      "grad_norm": 0.08616327569114879,
      "learning_rate": 3.617429697017716e-07,
      "loss": 0.0008,
      "step": 14328
    },
    {
      "epoch": 9.16762635956494,
      "grad_norm": 0.13477295790040197,
      "learning_rate": 3.611908785362572e-07,
      "loss": 0.0008,
      "step": 14329
    },
    {
      "epoch": 9.168266154830455,
      "grad_norm": 0.12273872951575264,
      "learning_rate": 3.606392012440596e-07,
      "loss": 0.001,
      "step": 14330
    },
    {
      "epoch": 9.168905950095969,
      "grad_norm": 0.08449348417169317,
      "learning_rate": 3.6008793784886754e-07,
      "loss": 0.0008,
      "step": 14331
    },
    {
      "epoch": 9.169545745361484,
      "grad_norm": 0.11123969941462475,
      "learning_rate": 3.5953708837435206e-07,
      "loss": 0.0009,
      "step": 14332
    },
    {
      "epoch": 9.170185540627,
      "grad_norm": 0.12247734398819234,
      "learning_rate": 3.5898665284416324e-07,
      "loss": 0.0008,
      "step": 14333
    },
    {
      "epoch": 9.170825335892514,
      "grad_norm": 0.17989055575401755,
      "learning_rate": 3.584366312819387e-07,
      "loss": 0.0013,
      "step": 14334
    },
    {
      "epoch": 9.17146513115803,
      "grad_norm": 0.21207530720284537,
      "learning_rate": 3.5788702371129416e-07,
      "loss": 0.0013,
      "step": 14335
    },
    {
      "epoch": 9.172104926423545,
      "grad_norm": 0.0976421750005534,
      "learning_rate": 3.5733783015582946e-07,
      "loss": 0.0009,
      "step": 14336
    },
    {
      "epoch": 9.17274472168906,
      "grad_norm": 0.1401302554203876,
      "learning_rate": 3.5678905063912363e-07,
      "loss": 0.0012,
      "step": 14337
    },
    {
      "epoch": 9.173384516954574,
      "grad_norm": 0.13028119456851556,
      "learning_rate": 3.562406851847433e-07,
      "loss": 0.001,
      "step": 14338
    },
    {
      "epoch": 9.174024312220089,
      "grad_norm": 0.0923452031330927,
      "learning_rate": 3.55692733816233e-07,
      "loss": 0.0008,
      "step": 14339
    },
    {
      "epoch": 9.174664107485604,
      "grad_norm": 0.22660010338742145,
      "learning_rate": 3.5514519655711955e-07,
      "loss": 0.0017,
      "step": 14340
    },
    {
      "epoch": 9.17530390275112,
      "grad_norm": 0.12668294993652512,
      "learning_rate": 3.545980734309162e-07,
      "loss": 0.001,
      "step": 14341
    },
    {
      "epoch": 9.175943698016635,
      "grad_norm": 0.20516362678125946,
      "learning_rate": 3.540513644611132e-07,
      "loss": 0.001,
      "step": 14342
    },
    {
      "epoch": 9.17658349328215,
      "grad_norm": 0.11091798035827727,
      "learning_rate": 3.5350506967118615e-07,
      "loss": 0.0008,
      "step": 14343
    },
    {
      "epoch": 9.177223288547665,
      "grad_norm": 0.09586093454766666,
      "learning_rate": 3.5295918908459183e-07,
      "loss": 0.0009,
      "step": 14344
    },
    {
      "epoch": 9.177863083813179,
      "grad_norm": 0.10917112036267365,
      "learning_rate": 3.5241372272476815e-07,
      "loss": 0.001,
      "step": 14345
    },
    {
      "epoch": 9.178502879078694,
      "grad_norm": 0.09628051845200837,
      "learning_rate": 3.5186867061513865e-07,
      "loss": 0.0009,
      "step": 14346
    },
    {
      "epoch": 9.17914267434421,
      "grad_norm": 0.09913863866790194,
      "learning_rate": 3.513240327791034e-07,
      "loss": 0.0009,
      "step": 14347
    },
    {
      "epoch": 9.179782469609725,
      "grad_norm": 0.10563672263556274,
      "learning_rate": 3.507798092400527e-07,
      "loss": 0.0009,
      "step": 14348
    },
    {
      "epoch": 9.18042226487524,
      "grad_norm": 0.23624299097719156,
      "learning_rate": 3.5023600002135226e-07,
      "loss": 0.0011,
      "step": 14349
    },
    {
      "epoch": 9.181062060140755,
      "grad_norm": 0.10189710275351077,
      "learning_rate": 3.4969260514635006e-07,
      "loss": 0.0008,
      "step": 14350
    },
    {
      "epoch": 9.18170185540627,
      "grad_norm": 0.27470833476363044,
      "learning_rate": 3.49149624638383e-07,
      "loss": 0.0015,
      "step": 14351
    },
    {
      "epoch": 9.182341650671786,
      "grad_norm": 0.15743246482755088,
      "learning_rate": 3.486070585207635e-07,
      "loss": 0.0011,
      "step": 14352
    },
    {
      "epoch": 9.1829814459373,
      "grad_norm": 0.06678563391995894,
      "learning_rate": 3.480649068167874e-07,
      "loss": 0.0006,
      "step": 14353
    },
    {
      "epoch": 9.183621241202815,
      "grad_norm": 0.11034424547335066,
      "learning_rate": 3.4752316954973607e-07,
      "loss": 0.0009,
      "step": 14354
    },
    {
      "epoch": 9.18426103646833,
      "grad_norm": 0.21918615407496075,
      "learning_rate": 3.4698184674286873e-07,
      "loss": 0.0017,
      "step": 14355
    },
    {
      "epoch": 9.184900831733845,
      "grad_norm": 0.11079085432977565,
      "learning_rate": 3.4644093841943005e-07,
      "loss": 0.0009,
      "step": 14356
    },
    {
      "epoch": 9.18554062699936,
      "grad_norm": 0.11936636360679816,
      "learning_rate": 3.4590044460264374e-07,
      "loss": 0.0008,
      "step": 14357
    },
    {
      "epoch": 9.186180422264876,
      "grad_norm": 0.10853738196163143,
      "learning_rate": 3.4536036531572006e-07,
      "loss": 0.0008,
      "step": 14358
    },
    {
      "epoch": 9.186820217530391,
      "grad_norm": 0.09670548960815933,
      "learning_rate": 3.4482070058184825e-07,
      "loss": 0.0008,
      "step": 14359
    },
    {
      "epoch": 9.187460012795905,
      "grad_norm": 0.10301347950867362,
      "learning_rate": 3.442814504241998e-07,
      "loss": 0.0009,
      "step": 14360
    },
    {
      "epoch": 9.18809980806142,
      "grad_norm": 0.17808504299478325,
      "learning_rate": 3.4374261486592955e-07,
      "loss": 0.0014,
      "step": 14361
    },
    {
      "epoch": 9.188739603326935,
      "grad_norm": 0.19312475639986865,
      "learning_rate": 3.4320419393017555e-07,
      "loss": 0.0009,
      "step": 14362
    },
    {
      "epoch": 9.18937939859245,
      "grad_norm": 0.09080165482263963,
      "learning_rate": 3.426661876400528e-07,
      "loss": 0.0007,
      "step": 14363
    },
    {
      "epoch": 9.190019193857966,
      "grad_norm": 0.23649952458441242,
      "learning_rate": 3.421285960186671e-07,
      "loss": 0.0011,
      "step": 14364
    },
    {
      "epoch": 9.190658989123481,
      "grad_norm": 0.13341850803885433,
      "learning_rate": 3.4159141908909896e-07,
      "loss": 0.0009,
      "step": 14365
    },
    {
      "epoch": 9.191298784388996,
      "grad_norm": 0.13653942426439866,
      "learning_rate": 3.410546568744144e-07,
      "loss": 0.0009,
      "step": 14366
    },
    {
      "epoch": 9.191938579654511,
      "grad_norm": 0.11491542415374945,
      "learning_rate": 3.405183093976594e-07,
      "loss": 0.0009,
      "step": 14367
    },
    {
      "epoch": 9.192578374920025,
      "grad_norm": 0.1418218970705065,
      "learning_rate": 3.399823766818666e-07,
      "loss": 0.001,
      "step": 14368
    },
    {
      "epoch": 9.19321817018554,
      "grad_norm": 0.13234747126184368,
      "learning_rate": 3.394468587500477e-07,
      "loss": 0.0009,
      "step": 14369
    },
    {
      "epoch": 9.193857965451055,
      "grad_norm": 0.13658023015010878,
      "learning_rate": 3.389117556251942e-07,
      "loss": 0.0011,
      "step": 14370
    },
    {
      "epoch": 9.19449776071657,
      "grad_norm": 0.1525643700471913,
      "learning_rate": 3.383770673302844e-07,
      "loss": 0.001,
      "step": 14371
    },
    {
      "epoch": 9.195137555982086,
      "grad_norm": 0.1215335686224602,
      "learning_rate": 3.3784279388827777e-07,
      "loss": 0.0009,
      "step": 14372
    },
    {
      "epoch": 9.195777351247601,
      "grad_norm": 0.1661784285793464,
      "learning_rate": 3.373089353221137e-07,
      "loss": 0.0011,
      "step": 14373
    },
    {
      "epoch": 9.196417146513117,
      "grad_norm": 0.06375483567964266,
      "learning_rate": 3.36775491654715e-07,
      "loss": 0.0006,
      "step": 14374
    },
    {
      "epoch": 9.19705694177863,
      "grad_norm": 0.14567319903365833,
      "learning_rate": 3.3624246290898667e-07,
      "loss": 0.0009,
      "step": 14375
    },
    {
      "epoch": 9.197696737044145,
      "grad_norm": 0.13250979142563515,
      "learning_rate": 3.35709849107817e-07,
      "loss": 0.0008,
      "step": 14376
    },
    {
      "epoch": 9.19833653230966,
      "grad_norm": 0.10661354308896391,
      "learning_rate": 3.3517765027407446e-07,
      "loss": 0.0008,
      "step": 14377
    },
    {
      "epoch": 9.198976327575176,
      "grad_norm": 0.13907073074188145,
      "learning_rate": 3.346458664306118e-07,
      "loss": 0.001,
      "step": 14378
    },
    {
      "epoch": 9.199616122840691,
      "grad_norm": 0.13196304305409984,
      "learning_rate": 3.3411449760026194e-07,
      "loss": 0.0011,
      "step": 14379
    },
    {
      "epoch": 9.200255918106206,
      "grad_norm": 0.10003427545805886,
      "learning_rate": 3.3358354380584103e-07,
      "loss": 0.0007,
      "step": 14380
    },
    {
      "epoch": 9.200895713371722,
      "grad_norm": 0.10901467166453697,
      "learning_rate": 3.330530050701486e-07,
      "loss": 0.0008,
      "step": 14381
    },
    {
      "epoch": 9.201535508637235,
      "grad_norm": 0.08947278103115791,
      "learning_rate": 3.325228814159631e-07,
      "loss": 0.0009,
      "step": 14382
    },
    {
      "epoch": 9.20217530390275,
      "grad_norm": 0.16510768059405034,
      "learning_rate": 3.319931728660486e-07,
      "loss": 0.0011,
      "step": 14383
    },
    {
      "epoch": 9.202815099168266,
      "grad_norm": 0.1189498808075877,
      "learning_rate": 3.3146387944314796e-07,
      "loss": 0.0011,
      "step": 14384
    },
    {
      "epoch": 9.203454894433781,
      "grad_norm": 0.11992964658831343,
      "learning_rate": 3.3093500116998966e-07,
      "loss": 0.0009,
      "step": 14385
    },
    {
      "epoch": 9.204094689699296,
      "grad_norm": 0.18535021438373975,
      "learning_rate": 3.3040653806928337e-07,
      "loss": 0.0018,
      "step": 14386
    },
    {
      "epoch": 9.204734484964812,
      "grad_norm": 0.1341090123817236,
      "learning_rate": 3.298784901637186e-07,
      "loss": 0.001,
      "step": 14387
    },
    {
      "epoch": 9.205374280230327,
      "grad_norm": 0.12490441623012127,
      "learning_rate": 3.2935085747597074e-07,
      "loss": 0.001,
      "step": 14388
    },
    {
      "epoch": 9.206014075495842,
      "grad_norm": 0.18456507293851776,
      "learning_rate": 3.288236400286937e-07,
      "loss": 0.0012,
      "step": 14389
    },
    {
      "epoch": 9.206653870761356,
      "grad_norm": 0.15002359896438755,
      "learning_rate": 3.2829683784452616e-07,
      "loss": 0.001,
      "step": 14390
    },
    {
      "epoch": 9.207293666026871,
      "grad_norm": 0.1005573463772663,
      "learning_rate": 3.277704509460877e-07,
      "loss": 0.0007,
      "step": 14391
    },
    {
      "epoch": 9.207933461292386,
      "grad_norm": 0.09700448119117644,
      "learning_rate": 3.2724447935598037e-07,
      "loss": 0.0009,
      "step": 14392
    },
    {
      "epoch": 9.208573256557901,
      "grad_norm": 0.08075051231115664,
      "learning_rate": 3.267189230967893e-07,
      "loss": 0.0007,
      "step": 14393
    },
    {
      "epoch": 9.209213051823417,
      "grad_norm": 0.16152149556973536,
      "learning_rate": 3.2619378219107766e-07,
      "loss": 0.001,
      "step": 14394
    },
    {
      "epoch": 9.209852847088932,
      "grad_norm": 0.12145028488807774,
      "learning_rate": 3.256690566613996e-07,
      "loss": 0.0009,
      "step": 14395
    },
    {
      "epoch": 9.210492642354447,
      "grad_norm": 0.277137258700625,
      "learning_rate": 3.251447465302815e-07,
      "loss": 0.001,
      "step": 14396
    },
    {
      "epoch": 9.21113243761996,
      "grad_norm": 0.07522525040821641,
      "learning_rate": 3.246208518202376e-07,
      "loss": 0.0006,
      "step": 14397
    },
    {
      "epoch": 9.211772232885476,
      "grad_norm": 0.10388259479210574,
      "learning_rate": 3.240973725537633e-07,
      "loss": 0.0008,
      "step": 14398
    },
    {
      "epoch": 9.212412028150991,
      "grad_norm": 0.1245626846037962,
      "learning_rate": 3.2357430875333606e-07,
      "loss": 0.0008,
      "step": 14399
    },
    {
      "epoch": 9.213051823416507,
      "grad_norm": 0.09291884923170794,
      "learning_rate": 3.2305166044141466e-07,
      "loss": 0.0008,
      "step": 14400
    },
    {
      "epoch": 9.213691618682022,
      "grad_norm": 0.12870686341218915,
      "learning_rate": 3.2252942764044114e-07,
      "loss": 0.001,
      "step": 14401
    },
    {
      "epoch": 9.214331413947537,
      "grad_norm": 0.0820174537151038,
      "learning_rate": 3.220076103728398e-07,
      "loss": 0.0008,
      "step": 14402
    },
    {
      "epoch": 9.214971209213052,
      "grad_norm": 0.16068184841191682,
      "learning_rate": 3.21486208661016e-07,
      "loss": 0.0012,
      "step": 14403
    },
    {
      "epoch": 9.215611004478568,
      "grad_norm": 0.11751054712330336,
      "learning_rate": 3.209652225273563e-07,
      "loss": 0.0009,
      "step": 14404
    },
    {
      "epoch": 9.216250799744081,
      "grad_norm": 0.09564051073264329,
      "learning_rate": 3.204446519942339e-07,
      "loss": 0.0008,
      "step": 14405
    },
    {
      "epoch": 9.216890595009597,
      "grad_norm": 0.17302161291508522,
      "learning_rate": 3.1992449708399875e-07,
      "loss": 0.0009,
      "step": 14406
    },
    {
      "epoch": 9.217530390275112,
      "grad_norm": 0.17406227442816635,
      "learning_rate": 3.1940475781898516e-07,
      "loss": 0.0011,
      "step": 14407
    },
    {
      "epoch": 9.218170185540627,
      "grad_norm": 0.1213576448550891,
      "learning_rate": 3.1888543422151195e-07,
      "loss": 0.0009,
      "step": 14408
    },
    {
      "epoch": 9.218809980806142,
      "grad_norm": 0.08144643081252374,
      "learning_rate": 3.1836652631387686e-07,
      "loss": 0.0008,
      "step": 14409
    },
    {
      "epoch": 9.219449776071658,
      "grad_norm": 0.14208098088283996,
      "learning_rate": 3.1784803411835985e-07,
      "loss": 0.001,
      "step": 14410
    },
    {
      "epoch": 9.220089571337173,
      "grad_norm": 0.14577136509601257,
      "learning_rate": 3.173299576572253e-07,
      "loss": 0.0012,
      "step": 14411
    },
    {
      "epoch": 9.220729366602686,
      "grad_norm": 0.13987532218908813,
      "learning_rate": 3.1681229695271985e-07,
      "loss": 0.0009,
      "step": 14412
    },
    {
      "epoch": 9.221369161868202,
      "grad_norm": 0.08146911683782737,
      "learning_rate": 3.16295052027068e-07,
      "loss": 0.0007,
      "step": 14413
    },
    {
      "epoch": 9.222008957133717,
      "grad_norm": 0.163726525344793,
      "learning_rate": 3.157782229024797e-07,
      "loss": 0.001,
      "step": 14414
    },
    {
      "epoch": 9.222648752399232,
      "grad_norm": 0.15480002159686893,
      "learning_rate": 3.152618096011495e-07,
      "loss": 0.0012,
      "step": 14415
    },
    {
      "epoch": 9.223288547664747,
      "grad_norm": 0.12539753416838387,
      "learning_rate": 3.1474581214524845e-07,
      "loss": 0.001,
      "step": 14416
    },
    {
      "epoch": 9.223928342930263,
      "grad_norm": 0.2159333605964942,
      "learning_rate": 3.1423023055693334e-07,
      "loss": 0.0017,
      "step": 14417
    },
    {
      "epoch": 9.224568138195778,
      "grad_norm": 0.10028026596605043,
      "learning_rate": 3.13715064858342e-07,
      "loss": 0.0008,
      "step": 14418
    },
    {
      "epoch": 9.225207933461292,
      "grad_norm": 0.10958394871774385,
      "learning_rate": 3.132003150715968e-07,
      "loss": 0.0008,
      "step": 14419
    },
    {
      "epoch": 9.225847728726807,
      "grad_norm": 0.11557808658276154,
      "learning_rate": 3.1268598121879767e-07,
      "loss": 0.0008,
      "step": 14420
    },
    {
      "epoch": 9.226487523992322,
      "grad_norm": 0.25244624716165526,
      "learning_rate": 3.1217206332203043e-07,
      "loss": 0.0011,
      "step": 14421
    },
    {
      "epoch": 9.227127319257837,
      "grad_norm": 0.18253304149252472,
      "learning_rate": 3.116585614033618e-07,
      "loss": 0.0007,
      "step": 14422
    },
    {
      "epoch": 9.227767114523353,
      "grad_norm": 0.14530227575850577,
      "learning_rate": 3.111454754848398e-07,
      "loss": 0.001,
      "step": 14423
    },
    {
      "epoch": 9.228406909788868,
      "grad_norm": 0.13886595830838885,
      "learning_rate": 3.106328055884955e-07,
      "loss": 0.0011,
      "step": 14424
    },
    {
      "epoch": 9.229046705054383,
      "grad_norm": 0.09537995107534265,
      "learning_rate": 3.101205517363426e-07,
      "loss": 0.0008,
      "step": 14425
    },
    {
      "epoch": 9.229686500319898,
      "grad_norm": 0.21180513019660752,
      "learning_rate": 3.096087139503778e-07,
      "loss": 0.0015,
      "step": 14426
    },
    {
      "epoch": 9.230326295585412,
      "grad_norm": 0.11821332501904425,
      "learning_rate": 3.090972922525748e-07,
      "loss": 0.0009,
      "step": 14427
    },
    {
      "epoch": 9.230966090850927,
      "grad_norm": 0.12789858328255246,
      "learning_rate": 3.085862866648959e-07,
      "loss": 0.001,
      "step": 14428
    },
    {
      "epoch": 9.231605886116443,
      "grad_norm": 0.09835911903221736,
      "learning_rate": 3.080756972092824e-07,
      "loss": 0.0009,
      "step": 14429
    },
    {
      "epoch": 9.232245681381958,
      "grad_norm": 0.11115871331155643,
      "learning_rate": 3.07565523907658e-07,
      "loss": 0.0009,
      "step": 14430
    },
    {
      "epoch": 9.232885476647473,
      "grad_norm": 0.1375855998963092,
      "learning_rate": 3.070557667819285e-07,
      "loss": 0.0008,
      "step": 14431
    },
    {
      "epoch": 9.233525271912988,
      "grad_norm": 0.19258263343295928,
      "learning_rate": 3.0654642585398185e-07,
      "loss": 0.0014,
      "step": 14432
    },
    {
      "epoch": 9.234165067178504,
      "grad_norm": 0.15231248849347734,
      "learning_rate": 3.0603750114568844e-07,
      "loss": 0.0012,
      "step": 14433
    },
    {
      "epoch": 9.234804862444017,
      "grad_norm": 0.1182502216528024,
      "learning_rate": 3.0552899267890067e-07,
      "loss": 0.0009,
      "step": 14434
    },
    {
      "epoch": 9.235444657709532,
      "grad_norm": 0.10258443241285625,
      "learning_rate": 3.050209004754534e-07,
      "loss": 0.0008,
      "step": 14435
    },
    {
      "epoch": 9.236084452975048,
      "grad_norm": 0.10411724444375957,
      "learning_rate": 3.0451322455716246e-07,
      "loss": 0.0008,
      "step": 14436
    },
    {
      "epoch": 9.236724248240563,
      "grad_norm": 0.14238494164238277,
      "learning_rate": 3.040059649458249e-07,
      "loss": 0.0012,
      "step": 14437
    },
    {
      "epoch": 9.237364043506078,
      "grad_norm": 0.1229452124826967,
      "learning_rate": 3.034991216632244e-07,
      "loss": 0.0008,
      "step": 14438
    },
    {
      "epoch": 9.238003838771593,
      "grad_norm": 0.11029668027467558,
      "learning_rate": 3.029926947311235e-07,
      "loss": 0.0007,
      "step": 14439
    },
    {
      "epoch": 9.238643634037109,
      "grad_norm": 0.1414678432120471,
      "learning_rate": 3.024866841712659e-07,
      "loss": 0.0011,
      "step": 14440
    },
    {
      "epoch": 9.239283429302624,
      "grad_norm": 0.14681253722570312,
      "learning_rate": 3.019810900053799e-07,
      "loss": 0.0009,
      "step": 14441
    },
    {
      "epoch": 9.239923224568138,
      "grad_norm": 0.143725898432111,
      "learning_rate": 3.014759122551747e-07,
      "loss": 0.0011,
      "step": 14442
    },
    {
      "epoch": 9.240563019833653,
      "grad_norm": 0.18505549977149569,
      "learning_rate": 3.0097115094234185e-07,
      "loss": 0.0014,
      "step": 14443
    },
    {
      "epoch": 9.241202815099168,
      "grad_norm": 0.09029112316967058,
      "learning_rate": 3.0046680608855295e-07,
      "loss": 0.0007,
      "step": 14444
    },
    {
      "epoch": 9.241842610364683,
      "grad_norm": 0.13397319718472628,
      "learning_rate": 2.999628777154662e-07,
      "loss": 0.0009,
      "step": 14445
    },
    {
      "epoch": 9.242482405630199,
      "grad_norm": 0.125354306806055,
      "learning_rate": 2.994593658447187e-07,
      "loss": 0.0009,
      "step": 14446
    },
    {
      "epoch": 9.243122200895714,
      "grad_norm": 0.10920901044679346,
      "learning_rate": 2.9895627049792874e-07,
      "loss": 0.001,
      "step": 14447
    },
    {
      "epoch": 9.24376199616123,
      "grad_norm": 0.11616004145204618,
      "learning_rate": 2.984535916967013e-07,
      "loss": 0.0008,
      "step": 14448
    },
    {
      "epoch": 9.244401791426743,
      "grad_norm": 0.12435046867085443,
      "learning_rate": 2.9795132946261906e-07,
      "loss": 0.0009,
      "step": 14449
    },
    {
      "epoch": 9.245041586692258,
      "grad_norm": 0.08612610898805635,
      "learning_rate": 2.9744948381724813e-07,
      "loss": 0.0007,
      "step": 14450
    },
    {
      "epoch": 9.245681381957773,
      "grad_norm": 0.12177313608135382,
      "learning_rate": 2.969480547821357e-07,
      "loss": 0.0009,
      "step": 14451
    },
    {
      "epoch": 9.246321177223289,
      "grad_norm": 0.13630939167241363,
      "learning_rate": 2.9644704237881463e-07,
      "loss": 0.0011,
      "step": 14452
    },
    {
      "epoch": 9.246960972488804,
      "grad_norm": 0.10467333853163999,
      "learning_rate": 2.9594644662879536e-07,
      "loss": 0.0009,
      "step": 14453
    },
    {
      "epoch": 9.247600767754319,
      "grad_norm": 0.18794372090656,
      "learning_rate": 2.9544626755357296e-07,
      "loss": 0.0013,
      "step": 14454
    },
    {
      "epoch": 9.248240563019834,
      "grad_norm": 0.20140096423948325,
      "learning_rate": 2.9494650517462585e-07,
      "loss": 0.0018,
      "step": 14455
    },
    {
      "epoch": 9.248880358285348,
      "grad_norm": 0.2761968545390428,
      "learning_rate": 2.944471595134124e-07,
      "loss": 0.0016,
      "step": 14456
    },
    {
      "epoch": 9.249520153550863,
      "grad_norm": 0.17244820590137813,
      "learning_rate": 2.9394823059137103e-07,
      "loss": 0.0013,
      "step": 14457
    },
    {
      "epoch": 9.250159948816378,
      "grad_norm": 0.16269304813512722,
      "learning_rate": 2.93449718429929e-07,
      "loss": 0.0012,
      "step": 14458
    },
    {
      "epoch": 9.250799744081894,
      "grad_norm": 0.12260400860390883,
      "learning_rate": 2.929516230504892e-07,
      "loss": 0.0009,
      "step": 14459
    },
    {
      "epoch": 9.251439539347409,
      "grad_norm": 0.11341653455181373,
      "learning_rate": 2.924539444744379e-07,
      "loss": 0.0007,
      "step": 14460
    },
    {
      "epoch": 9.252079334612924,
      "grad_norm": 0.07687365916006601,
      "learning_rate": 2.919566827231457e-07,
      "loss": 0.0006,
      "step": 14461
    },
    {
      "epoch": 9.25271912987844,
      "grad_norm": 0.11525821500045119,
      "learning_rate": 2.914598378179656e-07,
      "loss": 0.001,
      "step": 14462
    },
    {
      "epoch": 9.253358925143955,
      "grad_norm": 0.10178162967933152,
      "learning_rate": 2.909634097802294e-07,
      "loss": 0.0007,
      "step": 14463
    },
    {
      "epoch": 9.253998720409468,
      "grad_norm": 0.10991719258546273,
      "learning_rate": 2.904673986312545e-07,
      "loss": 0.0009,
      "step": 14464
    },
    {
      "epoch": 9.254638515674984,
      "grad_norm": 0.1200405986781304,
      "learning_rate": 2.8997180439233605e-07,
      "loss": 0.001,
      "step": 14465
    },
    {
      "epoch": 9.255278310940499,
      "grad_norm": 0.10547624472999165,
      "learning_rate": 2.89476627084756e-07,
      "loss": 0.0009,
      "step": 14466
    },
    {
      "epoch": 9.255918106206014,
      "grad_norm": 0.09007784491670526,
      "learning_rate": 2.8898186672977503e-07,
      "loss": 0.0006,
      "step": 14467
    },
    {
      "epoch": 9.25655790147153,
      "grad_norm": 0.15397109600990985,
      "learning_rate": 2.884875233486395e-07,
      "loss": 0.0011,
      "step": 14468
    },
    {
      "epoch": 9.257197696737045,
      "grad_norm": 0.17735058600747172,
      "learning_rate": 2.879935969625747e-07,
      "loss": 0.0011,
      "step": 14469
    },
    {
      "epoch": 9.25783749200256,
      "grad_norm": 0.13683718792091698,
      "learning_rate": 2.875000875927869e-07,
      "loss": 0.0011,
      "step": 14470
    },
    {
      "epoch": 9.258477287268073,
      "grad_norm": 0.10060984537161972,
      "learning_rate": 2.8700699526046817e-07,
      "loss": 0.0008,
      "step": 14471
    },
    {
      "epoch": 9.259117082533589,
      "grad_norm": 0.15906825222115772,
      "learning_rate": 2.865143199867926e-07,
      "loss": 0.0011,
      "step": 14472
    },
    {
      "epoch": 9.259756877799104,
      "grad_norm": 0.10918555115760366,
      "learning_rate": 2.860220617929121e-07,
      "loss": 0.0008,
      "step": 14473
    },
    {
      "epoch": 9.26039667306462,
      "grad_norm": 0.10096023837410645,
      "learning_rate": 2.855302206999633e-07,
      "loss": 0.0007,
      "step": 14474
    },
    {
      "epoch": 9.261036468330134,
      "grad_norm": 0.10344991493653483,
      "learning_rate": 2.85038796729068e-07,
      "loss": 0.0008,
      "step": 14475
    },
    {
      "epoch": 9.26167626359565,
      "grad_norm": 0.15779580235091883,
      "learning_rate": 2.845477899013249e-07,
      "loss": 0.001,
      "step": 14476
    },
    {
      "epoch": 9.262316058861165,
      "grad_norm": 0.1332060657716768,
      "learning_rate": 2.840572002378161e-07,
      "loss": 0.001,
      "step": 14477
    },
    {
      "epoch": 9.26295585412668,
      "grad_norm": 0.08952608931398569,
      "learning_rate": 2.835670277596081e-07,
      "loss": 0.0008,
      "step": 14478
    },
    {
      "epoch": 9.263595649392194,
      "grad_norm": 0.11891650384259922,
      "learning_rate": 2.8307727248774954e-07,
      "loss": 0.0008,
      "step": 14479
    },
    {
      "epoch": 9.264235444657709,
      "grad_norm": 0.1656974356299246,
      "learning_rate": 2.825879344432658e-07,
      "loss": 0.0014,
      "step": 14480
    },
    {
      "epoch": 9.264875239923224,
      "grad_norm": 0.14539601235119917,
      "learning_rate": 2.8209901364717127e-07,
      "loss": 0.0009,
      "step": 14481
    },
    {
      "epoch": 9.26551503518874,
      "grad_norm": 0.09687540368549492,
      "learning_rate": 2.8161051012045803e-07,
      "loss": 0.0009,
      "step": 14482
    },
    {
      "epoch": 9.266154830454255,
      "grad_norm": 0.16787027006706534,
      "learning_rate": 2.811224238841026e-07,
      "loss": 0.0013,
      "step": 14483
    },
    {
      "epoch": 9.26679462571977,
      "grad_norm": 0.13352255905100352,
      "learning_rate": 2.8063475495906045e-07,
      "loss": 0.0009,
      "step": 14484
    },
    {
      "epoch": 9.267434420985285,
      "grad_norm": 0.07539373981202284,
      "learning_rate": 2.8014750336627264e-07,
      "loss": 0.0007,
      "step": 14485
    },
    {
      "epoch": 9.268074216250799,
      "grad_norm": 0.14938861408052664,
      "learning_rate": 2.796606691266612e-07,
      "loss": 0.0011,
      "step": 14486
    },
    {
      "epoch": 9.268714011516314,
      "grad_norm": 0.09973478565534047,
      "learning_rate": 2.791742522611285e-07,
      "loss": 0.0007,
      "step": 14487
    },
    {
      "epoch": 9.26935380678183,
      "grad_norm": 0.2599101522669424,
      "learning_rate": 2.786882527905632e-07,
      "loss": 0.0009,
      "step": 14488
    },
    {
      "epoch": 9.269993602047345,
      "grad_norm": 0.14886727896495644,
      "learning_rate": 2.7820267073583096e-07,
      "loss": 0.0009,
      "step": 14489
    },
    {
      "epoch": 9.27063339731286,
      "grad_norm": 0.25023860486085325,
      "learning_rate": 2.777175061177817e-07,
      "loss": 0.0014,
      "step": 14490
    },
    {
      "epoch": 9.271273192578375,
      "grad_norm": 0.10925744424107028,
      "learning_rate": 2.7723275895724766e-07,
      "loss": 0.0008,
      "step": 14491
    },
    {
      "epoch": 9.27191298784389,
      "grad_norm": 0.13586015891879397,
      "learning_rate": 2.767484292750444e-07,
      "loss": 0.0008,
      "step": 14492
    },
    {
      "epoch": 9.272552783109404,
      "grad_norm": 0.13056408119282017,
      "learning_rate": 2.7626451709196643e-07,
      "loss": 0.0009,
      "step": 14493
    },
    {
      "epoch": 9.27319257837492,
      "grad_norm": 0.11805739418839652,
      "learning_rate": 2.757810224287938e-07,
      "loss": 0.0009,
      "step": 14494
    },
    {
      "epoch": 9.273832373640435,
      "grad_norm": 0.12533984088218691,
      "learning_rate": 2.752979453062865e-07,
      "loss": 0.001,
      "step": 14495
    },
    {
      "epoch": 9.27447216890595,
      "grad_norm": 0.3096289736670887,
      "learning_rate": 2.748152857451858e-07,
      "loss": 0.0012,
      "step": 14496
    },
    {
      "epoch": 9.275111964171465,
      "grad_norm": 0.21002826677065015,
      "learning_rate": 2.743330437662162e-07,
      "loss": 0.0012,
      "step": 14497
    },
    {
      "epoch": 9.27575175943698,
      "grad_norm": 0.17711114889636223,
      "learning_rate": 2.738512193900855e-07,
      "loss": 0.0013,
      "step": 14498
    },
    {
      "epoch": 9.276391554702496,
      "grad_norm": 0.09221709108309648,
      "learning_rate": 2.7336981263748284e-07,
      "loss": 0.0007,
      "step": 14499
    },
    {
      "epoch": 9.277031349968011,
      "grad_norm": 0.15279142967314854,
      "learning_rate": 2.728888235290783e-07,
      "loss": 0.0009,
      "step": 14500
    },
    {
      "epoch": 9.277671145233525,
      "grad_norm": 0.08767596085155636,
      "learning_rate": 2.72408252085522e-07,
      "loss": 0.0008,
      "step": 14501
    },
    {
      "epoch": 9.27831094049904,
      "grad_norm": 0.11093540269243454,
      "learning_rate": 2.7192809832745305e-07,
      "loss": 0.0009,
      "step": 14502
    },
    {
      "epoch": 9.278950735764555,
      "grad_norm": 0.15135768355422263,
      "learning_rate": 2.714483622754871e-07,
      "loss": 0.0012,
      "step": 14503
    },
    {
      "epoch": 9.27959053103007,
      "grad_norm": 0.557350041894307,
      "learning_rate": 2.7096904395022217e-07,
      "loss": 0.0012,
      "step": 14504
    },
    {
      "epoch": 9.280230326295586,
      "grad_norm": 0.11425132084318519,
      "learning_rate": 2.7049014337223956e-07,
      "loss": 0.001,
      "step": 14505
    },
    {
      "epoch": 9.280870121561101,
      "grad_norm": 0.15516610795410635,
      "learning_rate": 2.7001166056210393e-07,
      "loss": 0.0011,
      "step": 14506
    },
    {
      "epoch": 9.281509916826616,
      "grad_norm": 0.09502574531458854,
      "learning_rate": 2.6953359554035776e-07,
      "loss": 0.0007,
      "step": 14507
    },
    {
      "epoch": 9.28214971209213,
      "grad_norm": 0.11445966084395064,
      "learning_rate": 2.6905594832753014e-07,
      "loss": 0.0008,
      "step": 14508
    },
    {
      "epoch": 9.282789507357645,
      "grad_norm": 0.15538646824381527,
      "learning_rate": 2.6857871894413135e-07,
      "loss": 0.0012,
      "step": 14509
    },
    {
      "epoch": 9.28342930262316,
      "grad_norm": 0.14619057110351738,
      "learning_rate": 2.681019074106506e-07,
      "loss": 0.001,
      "step": 14510
    },
    {
      "epoch": 9.284069097888676,
      "grad_norm": 0.12137058573380934,
      "learning_rate": 2.6762551374756253e-07,
      "loss": 0.001,
      "step": 14511
    },
    {
      "epoch": 9.28470889315419,
      "grad_norm": 0.07219018838923584,
      "learning_rate": 2.6714953797532193e-07,
      "loss": 0.0006,
      "step": 14512
    },
    {
      "epoch": 9.285348688419706,
      "grad_norm": 0.15507231269517238,
      "learning_rate": 2.6667398011436695e-07,
      "loss": 0.0012,
      "step": 14513
    },
    {
      "epoch": 9.285988483685221,
      "grad_norm": 0.1554138375273473,
      "learning_rate": 2.661988401851168e-07,
      "loss": 0.0011,
      "step": 14514
    },
    {
      "epoch": 9.286628278950737,
      "grad_norm": 0.12445766125337916,
      "learning_rate": 2.6572411820797394e-07,
      "loss": 0.0009,
      "step": 14515
    },
    {
      "epoch": 9.28726807421625,
      "grad_norm": 0.14752565881326116,
      "learning_rate": 2.652498142033211e-07,
      "loss": 0.0009,
      "step": 14516
    },
    {
      "epoch": 9.287907869481765,
      "grad_norm": 0.11177900315410257,
      "learning_rate": 2.6477592819152407e-07,
      "loss": 0.0008,
      "step": 14517
    },
    {
      "epoch": 9.28854766474728,
      "grad_norm": 0.1510246626328962,
      "learning_rate": 2.6430246019293116e-07,
      "loss": 0.0012,
      "step": 14518
    },
    {
      "epoch": 9.289187460012796,
      "grad_norm": 0.11771807249991428,
      "learning_rate": 2.6382941022787376e-07,
      "loss": 0.0009,
      "step": 14519
    },
    {
      "epoch": 9.289827255278311,
      "grad_norm": 0.25614147244967334,
      "learning_rate": 2.6335677831666015e-07,
      "loss": 0.0014,
      "step": 14520
    },
    {
      "epoch": 9.290467050543826,
      "grad_norm": 0.12342685761606763,
      "learning_rate": 2.628845644795863e-07,
      "loss": 0.0009,
      "step": 14521
    },
    {
      "epoch": 9.291106845809342,
      "grad_norm": 0.14613882441830467,
      "learning_rate": 2.624127687369293e-07,
      "loss": 0.0011,
      "step": 14522
    },
    {
      "epoch": 9.291746641074855,
      "grad_norm": 0.1519510129416873,
      "learning_rate": 2.619413911089463e-07,
      "loss": 0.001,
      "step": 14523
    },
    {
      "epoch": 9.29238643634037,
      "grad_norm": 0.16465465231390464,
      "learning_rate": 2.614704316158778e-07,
      "loss": 0.0011,
      "step": 14524
    },
    {
      "epoch": 9.293026231605886,
      "grad_norm": 0.15813981652349254,
      "learning_rate": 2.6099989027794427e-07,
      "loss": 0.0009,
      "step": 14525
    },
    {
      "epoch": 9.293666026871401,
      "grad_norm": 0.1771011454078896,
      "learning_rate": 2.605297671153517e-07,
      "loss": 0.0013,
      "step": 14526
    },
    {
      "epoch": 9.294305822136916,
      "grad_norm": 0.144452192876888,
      "learning_rate": 2.6006006214828516e-07,
      "loss": 0.001,
      "step": 14527
    },
    {
      "epoch": 9.294945617402432,
      "grad_norm": 0.17975442326149815,
      "learning_rate": 2.5959077539691404e-07,
      "loss": 0.0015,
      "step": 14528
    },
    {
      "epoch": 9.295585412667947,
      "grad_norm": 0.14834662778569455,
      "learning_rate": 2.5912190688138885e-07,
      "loss": 0.0011,
      "step": 14529
    },
    {
      "epoch": 9.29622520793346,
      "grad_norm": 0.09182043692889244,
      "learning_rate": 2.586534566218402e-07,
      "loss": 0.0007,
      "step": 14530
    },
    {
      "epoch": 9.296865003198976,
      "grad_norm": 0.14979377069050429,
      "learning_rate": 2.5818542463838304e-07,
      "loss": 0.001,
      "step": 14531
    },
    {
      "epoch": 9.297504798464491,
      "grad_norm": 0.2857467504596076,
      "learning_rate": 2.5771781095111583e-07,
      "loss": 0.0013,
      "step": 14532
    },
    {
      "epoch": 9.298144593730006,
      "grad_norm": 0.12138757146499803,
      "learning_rate": 2.572506155801158e-07,
      "loss": 0.001,
      "step": 14533
    },
    {
      "epoch": 9.298784388995522,
      "grad_norm": 0.13533127106378434,
      "learning_rate": 2.567838385454413e-07,
      "loss": 0.0009,
      "step": 14534
    },
    {
      "epoch": 9.299424184261037,
      "grad_norm": 0.2734962109808441,
      "learning_rate": 2.563174798671386e-07,
      "loss": 0.0014,
      "step": 14535
    },
    {
      "epoch": 9.300063979526552,
      "grad_norm": 0.1280438021069097,
      "learning_rate": 2.5585153956523057e-07,
      "loss": 0.0009,
      "step": 14536
    },
    {
      "epoch": 9.300703774792067,
      "grad_norm": 0.15542496788138804,
      "learning_rate": 2.553860176597245e-07,
      "loss": 0.0012,
      "step": 14537
    },
    {
      "epoch": 9.30134357005758,
      "grad_norm": 0.14006029876105874,
      "learning_rate": 2.549209141706066e-07,
      "loss": 0.0009,
      "step": 14538
    },
    {
      "epoch": 9.301983365323096,
      "grad_norm": 0.15391089040145853,
      "learning_rate": 2.5445622911785097e-07,
      "loss": 0.0011,
      "step": 14539
    },
    {
      "epoch": 9.302623160588611,
      "grad_norm": 0.12463548561925554,
      "learning_rate": 2.539919625214082e-07,
      "loss": 0.0009,
      "step": 14540
    },
    {
      "epoch": 9.303262955854127,
      "grad_norm": 0.10340418151070077,
      "learning_rate": 2.5352811440121474e-07,
      "loss": 0.0008,
      "step": 14541
    },
    {
      "epoch": 9.303902751119642,
      "grad_norm": 0.121208352024799,
      "learning_rate": 2.530646847771856e-07,
      "loss": 0.001,
      "step": 14542
    },
    {
      "epoch": 9.304542546385157,
      "grad_norm": 0.15619801681646886,
      "learning_rate": 2.526016736692205e-07,
      "loss": 0.0011,
      "step": 14543
    },
    {
      "epoch": 9.305182341650672,
      "grad_norm": 0.10991571357225745,
      "learning_rate": 2.521390810972002e-07,
      "loss": 0.001,
      "step": 14544
    },
    {
      "epoch": 9.305822136916186,
      "grad_norm": 0.13600466755599666,
      "learning_rate": 2.5167690708098767e-07,
      "loss": 0.0011,
      "step": 14545
    },
    {
      "epoch": 9.306461932181701,
      "grad_norm": 0.08237968589364483,
      "learning_rate": 2.512151516404282e-07,
      "loss": 0.0007,
      "step": 14546
    },
    {
      "epoch": 9.307101727447217,
      "grad_norm": 0.135496112673045,
      "learning_rate": 2.507538147953481e-07,
      "loss": 0.0011,
      "step": 14547
    },
    {
      "epoch": 9.307741522712732,
      "grad_norm": 0.12086494752960517,
      "learning_rate": 2.50292896565556e-07,
      "loss": 0.0009,
      "step": 14548
    },
    {
      "epoch": 9.308381317978247,
      "grad_norm": 0.24946540844965404,
      "learning_rate": 2.4983239697084493e-07,
      "loss": 0.0013,
      "step": 14549
    },
    {
      "epoch": 9.309021113243762,
      "grad_norm": 0.11233646246335086,
      "learning_rate": 2.493723160309858e-07,
      "loss": 0.0008,
      "step": 14550
    },
    {
      "epoch": 9.309660908509278,
      "grad_norm": 0.10780221162751728,
      "learning_rate": 2.4891265376573384e-07,
      "loss": 0.0009,
      "step": 14551
    },
    {
      "epoch": 9.310300703774793,
      "grad_norm": 0.10493704056793515,
      "learning_rate": 2.484534101948277e-07,
      "loss": 0.0009,
      "step": 14552
    },
    {
      "epoch": 9.310940499040306,
      "grad_norm": 0.09935345188317111,
      "learning_rate": 2.4799458533798504e-07,
      "loss": 0.0007,
      "step": 14553
    },
    {
      "epoch": 9.311580294305822,
      "grad_norm": 0.20347582860119867,
      "learning_rate": 2.475361792149078e-07,
      "loss": 0.0014,
      "step": 14554
    },
    {
      "epoch": 9.312220089571337,
      "grad_norm": 0.1243548064992199,
      "learning_rate": 2.47078191845278e-07,
      "loss": 0.0009,
      "step": 14555
    },
    {
      "epoch": 9.312859884836852,
      "grad_norm": 0.2705660005264615,
      "learning_rate": 2.4662062324876225e-07,
      "loss": 0.0012,
      "step": 14556
    },
    {
      "epoch": 9.313499680102368,
      "grad_norm": 0.11119127711008575,
      "learning_rate": 2.461634734450069e-07,
      "loss": 0.0008,
      "step": 14557
    },
    {
      "epoch": 9.314139475367883,
      "grad_norm": 0.10690525067490163,
      "learning_rate": 2.457067424536397e-07,
      "loss": 0.0009,
      "step": 14558
    },
    {
      "epoch": 9.314779270633398,
      "grad_norm": 0.1132554086613765,
      "learning_rate": 2.452504302942749e-07,
      "loss": 0.0009,
      "step": 14559
    },
    {
      "epoch": 9.315419065898912,
      "grad_norm": 0.10888944302462268,
      "learning_rate": 2.4479453698650455e-07,
      "loss": 0.0009,
      "step": 14560
    },
    {
      "epoch": 9.316058861164427,
      "grad_norm": 0.1789060684947106,
      "learning_rate": 2.443390625499031e-07,
      "loss": 0.0011,
      "step": 14561
    },
    {
      "epoch": 9.316698656429942,
      "grad_norm": 0.12043901357833346,
      "learning_rate": 2.4388400700402824e-07,
      "loss": 0.0008,
      "step": 14562
    },
    {
      "epoch": 9.317338451695457,
      "grad_norm": 0.11282761503013918,
      "learning_rate": 2.4342937036841983e-07,
      "loss": 0.001,
      "step": 14563
    },
    {
      "epoch": 9.317978246960973,
      "grad_norm": 0.10251993724261002,
      "learning_rate": 2.4297515266259784e-07,
      "loss": 0.0008,
      "step": 14564
    },
    {
      "epoch": 9.318618042226488,
      "grad_norm": 0.13038217415567022,
      "learning_rate": 2.4252135390606666e-07,
      "loss": 0.0009,
      "step": 14565
    },
    {
      "epoch": 9.319257837492003,
      "grad_norm": 0.13121128756028969,
      "learning_rate": 2.420679741183118e-07,
      "loss": 0.0008,
      "step": 14566
    },
    {
      "epoch": 9.319897632757517,
      "grad_norm": 0.08089664700022176,
      "learning_rate": 2.416150133188e-07,
      "loss": 0.0006,
      "step": 14567
    },
    {
      "epoch": 9.320537428023032,
      "grad_norm": 0.11040472032113749,
      "learning_rate": 2.4116247152698e-07,
      "loss": 0.0008,
      "step": 14568
    },
    {
      "epoch": 9.321177223288547,
      "grad_norm": 0.08474355542935667,
      "learning_rate": 2.4071034876228527e-07,
      "loss": 0.0007,
      "step": 14569
    },
    {
      "epoch": 9.321817018554063,
      "grad_norm": 0.17818347294213166,
      "learning_rate": 2.402586450441269e-07,
      "loss": 0.0013,
      "step": 14570
    },
    {
      "epoch": 9.322456813819578,
      "grad_norm": 0.15454912119119377,
      "learning_rate": 2.398073603919004e-07,
      "loss": 0.0011,
      "step": 14571
    },
    {
      "epoch": 9.323096609085093,
      "grad_norm": 0.10505364314554397,
      "learning_rate": 2.393564948249849e-07,
      "loss": 0.0008,
      "step": 14572
    },
    {
      "epoch": 9.323736404350608,
      "grad_norm": 0.15349369237144958,
      "learning_rate": 2.3890604836273924e-07,
      "loss": 0.0012,
      "step": 14573
    },
    {
      "epoch": 9.324376199616124,
      "grad_norm": 0.10283793946282473,
      "learning_rate": 2.384560210245024e-07,
      "loss": 0.0009,
      "step": 14574
    },
    {
      "epoch": 9.325015994881637,
      "grad_norm": 0.16576130803390937,
      "learning_rate": 2.380064128296e-07,
      "loss": 0.0011,
      "step": 14575
    },
    {
      "epoch": 9.325655790147152,
      "grad_norm": 0.2185089075318969,
      "learning_rate": 2.3755722379733782e-07,
      "loss": 0.0014,
      "step": 14576
    },
    {
      "epoch": 9.326295585412668,
      "grad_norm": 0.1356827274230054,
      "learning_rate": 2.3710845394700254e-07,
      "loss": 0.001,
      "step": 14577
    },
    {
      "epoch": 9.326935380678183,
      "grad_norm": 0.1419803910681923,
      "learning_rate": 2.3666010329786216e-07,
      "loss": 0.0011,
      "step": 14578
    },
    {
      "epoch": 9.327575175943698,
      "grad_norm": 0.11874363598038196,
      "learning_rate": 2.362121718691701e-07,
      "loss": 0.0008,
      "step": 14579
    },
    {
      "epoch": 9.328214971209214,
      "grad_norm": 0.08665856671702798,
      "learning_rate": 2.3576465968015883e-07,
      "loss": 0.0007,
      "step": 14580
    },
    {
      "epoch": 9.328854766474729,
      "grad_norm": 0.13120490388197376,
      "learning_rate": 2.3531756675004403e-07,
      "loss": 0.001,
      "step": 14581
    },
    {
      "epoch": 9.329494561740242,
      "grad_norm": 0.12221520720648327,
      "learning_rate": 2.348708930980226e-07,
      "loss": 0.0009,
      "step": 14582
    },
    {
      "epoch": 9.330134357005758,
      "grad_norm": 0.21860401803200843,
      "learning_rate": 2.3442463874327471e-07,
      "loss": 0.0014,
      "step": 14583
    },
    {
      "epoch": 9.330774152271273,
      "grad_norm": 0.12700844618993581,
      "learning_rate": 2.3397880370496062e-07,
      "loss": 0.0009,
      "step": 14584
    },
    {
      "epoch": 9.331413947536788,
      "grad_norm": 0.14684975228216696,
      "learning_rate": 2.3353338800222503e-07,
      "loss": 0.0007,
      "step": 14585
    },
    {
      "epoch": 9.332053742802303,
      "grad_norm": 0.17809921764174866,
      "learning_rate": 2.3308839165419263e-07,
      "loss": 0.0013,
      "step": 14586
    },
    {
      "epoch": 9.332693538067819,
      "grad_norm": 0.07206493025128886,
      "learning_rate": 2.3264381467997033e-07,
      "loss": 0.0007,
      "step": 14587
    },
    {
      "epoch": 9.333333333333334,
      "grad_norm": 0.15900861671112296,
      "learning_rate": 2.3219965709864623e-07,
      "loss": 0.0011,
      "step": 14588
    },
    {
      "epoch": 9.33397312859885,
      "grad_norm": 0.17350252527380983,
      "learning_rate": 2.3175591892929506e-07,
      "loss": 0.001,
      "step": 14589
    },
    {
      "epoch": 9.334612923864363,
      "grad_norm": 0.11763784588983392,
      "learning_rate": 2.3131260019096824e-07,
      "loss": 0.001,
      "step": 14590
    },
    {
      "epoch": 9.335252719129878,
      "grad_norm": 0.12563918424755105,
      "learning_rate": 2.3086970090270056e-07,
      "loss": 0.001,
      "step": 14591
    },
    {
      "epoch": 9.335892514395393,
      "grad_norm": 0.10029777261827925,
      "learning_rate": 2.3042722108351013e-07,
      "loss": 0.0007,
      "step": 14592
    },
    {
      "epoch": 9.336532309660909,
      "grad_norm": 0.24298638410964668,
      "learning_rate": 2.299851607523973e-07,
      "loss": 0.0011,
      "step": 14593
    },
    {
      "epoch": 9.337172104926424,
      "grad_norm": 0.27622965894033896,
      "learning_rate": 2.2954351992834023e-07,
      "loss": 0.0019,
      "step": 14594
    },
    {
      "epoch": 9.337811900191939,
      "grad_norm": 0.15048933584864255,
      "learning_rate": 2.2910229863030597e-07,
      "loss": 0.001,
      "step": 14595
    },
    {
      "epoch": 9.338451695457454,
      "grad_norm": 0.1964590891746309,
      "learning_rate": 2.2866149687723715e-07,
      "loss": 0.0015,
      "step": 14596
    },
    {
      "epoch": 9.339091490722968,
      "grad_norm": 0.17969614089674854,
      "learning_rate": 2.2822111468806197e-07,
      "loss": 0.0012,
      "step": 14597
    },
    {
      "epoch": 9.339731285988483,
      "grad_norm": 0.17203283164738503,
      "learning_rate": 2.2778115208168861e-07,
      "loss": 0.0011,
      "step": 14598
    },
    {
      "epoch": 9.340371081253998,
      "grad_norm": 0.1884664496172365,
      "learning_rate": 2.2734160907700976e-07,
      "loss": 0.0013,
      "step": 14599
    },
    {
      "epoch": 9.341010876519514,
      "grad_norm": 0.14166335019770546,
      "learning_rate": 2.2690248569289809e-07,
      "loss": 0.0009,
      "step": 14600
    },
    {
      "epoch": 9.341650671785029,
      "grad_norm": 0.09120221643929056,
      "learning_rate": 2.264637819482096e-07,
      "loss": 0.0008,
      "step": 14601
    },
    {
      "epoch": 9.342290467050544,
      "grad_norm": 0.09450342277639476,
      "learning_rate": 2.260254978617793e-07,
      "loss": 0.0007,
      "step": 14602
    },
    {
      "epoch": 9.34293026231606,
      "grad_norm": 0.1437218798435191,
      "learning_rate": 2.2558763345242874e-07,
      "loss": 0.0011,
      "step": 14603
    },
    {
      "epoch": 9.343570057581573,
      "grad_norm": 0.21214910722177294,
      "learning_rate": 2.2515018873895622e-07,
      "loss": 0.0015,
      "step": 14604
    },
    {
      "epoch": 9.344209852847088,
      "grad_norm": 0.17863790239511107,
      "learning_rate": 2.2471316374014672e-07,
      "loss": 0.0012,
      "step": 14605
    },
    {
      "epoch": 9.344849648112604,
      "grad_norm": 0.09417011515136262,
      "learning_rate": 2.2427655847476636e-07,
      "loss": 0.0008,
      "step": 14606
    },
    {
      "epoch": 9.345489443378119,
      "grad_norm": 0.11063685146536865,
      "learning_rate": 2.2384037296156014e-07,
      "loss": 0.0008,
      "step": 14607
    },
    {
      "epoch": 9.346129238643634,
      "grad_norm": 0.14381101744579985,
      "learning_rate": 2.234046072192575e-07,
      "loss": 0.001,
      "step": 14608
    },
    {
      "epoch": 9.34676903390915,
      "grad_norm": 0.3535364390475069,
      "learning_rate": 2.2296926126657015e-07,
      "loss": 0.0017,
      "step": 14609
    },
    {
      "epoch": 9.347408829174665,
      "grad_norm": 0.30682390083942856,
      "learning_rate": 2.225343351221909e-07,
      "loss": 0.001,
      "step": 14610
    },
    {
      "epoch": 9.34804862444018,
      "grad_norm": 0.1052526758722408,
      "learning_rate": 2.2209982880479374e-07,
      "loss": 0.0008,
      "step": 14611
    },
    {
      "epoch": 9.348688419705693,
      "grad_norm": 0.1306668602400201,
      "learning_rate": 2.2166574233303706e-07,
      "loss": 0.001,
      "step": 14612
    },
    {
      "epoch": 9.349328214971209,
      "grad_norm": 0.12935263892384666,
      "learning_rate": 2.212320757255604e-07,
      "loss": 0.0008,
      "step": 14613
    },
    {
      "epoch": 9.349968010236724,
      "grad_norm": 0.17137520437120576,
      "learning_rate": 2.2079882900098214e-07,
      "loss": 0.001,
      "step": 14614
    },
    {
      "epoch": 9.35060780550224,
      "grad_norm": 0.14499896344169533,
      "learning_rate": 2.203660021779064e-07,
      "loss": 0.0011,
      "step": 14615
    },
    {
      "epoch": 9.351247600767755,
      "grad_norm": 0.06180386334121307,
      "learning_rate": 2.199335952749171e-07,
      "loss": 0.0006,
      "step": 14616
    },
    {
      "epoch": 9.35188739603327,
      "grad_norm": 0.14180956616691268,
      "learning_rate": 2.1950160831058277e-07,
      "loss": 0.001,
      "step": 14617
    },
    {
      "epoch": 9.352527191298785,
      "grad_norm": 0.1203927368908974,
      "learning_rate": 2.1907004130345078e-07,
      "loss": 0.0009,
      "step": 14618
    },
    {
      "epoch": 9.353166986564299,
      "grad_norm": 0.18150486654405948,
      "learning_rate": 2.1863889427205297e-07,
      "loss": 0.0009,
      "step": 14619
    },
    {
      "epoch": 9.353806781829814,
      "grad_norm": 0.12030500708065887,
      "learning_rate": 2.1820816723490012e-07,
      "loss": 0.0009,
      "step": 14620
    },
    {
      "epoch": 9.35444657709533,
      "grad_norm": 0.30946742463547544,
      "learning_rate": 2.1777786021048854e-07,
      "loss": 0.0012,
      "step": 14621
    },
    {
      "epoch": 9.355086372360844,
      "grad_norm": 0.10977732071720651,
      "learning_rate": 2.173479732172945e-07,
      "loss": 0.0009,
      "step": 14622
    },
    {
      "epoch": 9.35572616762636,
      "grad_norm": 0.13264805647328332,
      "learning_rate": 2.1691850627377774e-07,
      "loss": 0.0009,
      "step": 14623
    },
    {
      "epoch": 9.356365962891875,
      "grad_norm": 0.1412568751150925,
      "learning_rate": 2.1648945939837685e-07,
      "loss": 0.0008,
      "step": 14624
    },
    {
      "epoch": 9.35700575815739,
      "grad_norm": 0.13633597062341804,
      "learning_rate": 2.160608326095137e-07,
      "loss": 0.0008,
      "step": 14625
    },
    {
      "epoch": 9.357645553422905,
      "grad_norm": 0.16099395582818954,
      "learning_rate": 2.1563262592559475e-07,
      "loss": 0.0012,
      "step": 14626
    },
    {
      "epoch": 9.358285348688419,
      "grad_norm": 0.09371988900559002,
      "learning_rate": 2.1520483936500746e-07,
      "loss": 0.0007,
      "step": 14627
    },
    {
      "epoch": 9.358925143953934,
      "grad_norm": 0.16491905805361434,
      "learning_rate": 2.1477747294611607e-07,
      "loss": 0.001,
      "step": 14628
    },
    {
      "epoch": 9.35956493921945,
      "grad_norm": 0.09815706751580958,
      "learning_rate": 2.1435052668727474e-07,
      "loss": 0.0008,
      "step": 14629
    },
    {
      "epoch": 9.360204734484965,
      "grad_norm": 0.18322437734100216,
      "learning_rate": 2.139240006068144e-07,
      "loss": 0.0011,
      "step": 14630
    },
    {
      "epoch": 9.36084452975048,
      "grad_norm": 0.11820647818602137,
      "learning_rate": 2.134978947230504e-07,
      "loss": 0.001,
      "step": 14631
    },
    {
      "epoch": 9.361484325015995,
      "grad_norm": 0.3435501475700914,
      "learning_rate": 2.1307220905427695e-07,
      "loss": 0.0019,
      "step": 14632
    },
    {
      "epoch": 9.36212412028151,
      "grad_norm": 0.1253693167308227,
      "learning_rate": 2.126469436187728e-07,
      "loss": 0.0008,
      "step": 14633
    },
    {
      "epoch": 9.362763915547024,
      "grad_norm": 0.17477690545905028,
      "learning_rate": 2.1222209843480003e-07,
      "loss": 0.0011,
      "step": 14634
    },
    {
      "epoch": 9.36340371081254,
      "grad_norm": 0.14811513448053976,
      "learning_rate": 2.1179767352059734e-07,
      "loss": 0.001,
      "step": 14635
    },
    {
      "epoch": 9.364043506078055,
      "grad_norm": 0.11712280947585549,
      "learning_rate": 2.1137366889439126e-07,
      "loss": 0.0009,
      "step": 14636
    },
    {
      "epoch": 9.36468330134357,
      "grad_norm": 0.12111549288110578,
      "learning_rate": 2.1095008457438837e-07,
      "loss": 0.0011,
      "step": 14637
    },
    {
      "epoch": 9.365323096609085,
      "grad_norm": 0.40457177471695865,
      "learning_rate": 2.105269205787741e-07,
      "loss": 0.0012,
      "step": 14638
    },
    {
      "epoch": 9.3659628918746,
      "grad_norm": 0.09281937514826093,
      "learning_rate": 2.101041769257206e-07,
      "loss": 0.0007,
      "step": 14639
    },
    {
      "epoch": 9.366602687140116,
      "grad_norm": 0.13065722166913493,
      "learning_rate": 2.096818536333789e-07,
      "loss": 0.001,
      "step": 14640
    },
    {
      "epoch": 9.36724248240563,
      "grad_norm": 0.10706026794870405,
      "learning_rate": 2.0925995071988114e-07,
      "loss": 0.0009,
      "step": 14641
    },
    {
      "epoch": 9.367882277671145,
      "grad_norm": 0.17509400526601432,
      "learning_rate": 2.0883846820334618e-07,
      "loss": 0.0008,
      "step": 14642
    },
    {
      "epoch": 9.36852207293666,
      "grad_norm": 0.23865510805463533,
      "learning_rate": 2.0841740610187066e-07,
      "loss": 0.001,
      "step": 14643
    },
    {
      "epoch": 9.369161868202175,
      "grad_norm": 0.13805338176682136,
      "learning_rate": 2.0799676443353346e-07,
      "loss": 0.0009,
      "step": 14644
    },
    {
      "epoch": 9.36980166346769,
      "grad_norm": 0.08732441786472472,
      "learning_rate": 2.0757654321639676e-07,
      "loss": 0.0007,
      "step": 14645
    },
    {
      "epoch": 9.370441458733206,
      "grad_norm": 0.08580620329206132,
      "learning_rate": 2.0715674246850393e-07,
      "loss": 0.0008,
      "step": 14646
    },
    {
      "epoch": 9.371081253998721,
      "grad_norm": 0.11178158007263368,
      "learning_rate": 2.0673736220787943e-07,
      "loss": 0.0008,
      "step": 14647
    },
    {
      "epoch": 9.371721049264236,
      "grad_norm": 0.13729121090149024,
      "learning_rate": 2.063184024525322e-07,
      "loss": 0.0009,
      "step": 14648
    },
    {
      "epoch": 9.37236084452975,
      "grad_norm": 0.1688117996722735,
      "learning_rate": 2.058998632204512e-07,
      "loss": 0.0017,
      "step": 14649
    },
    {
      "epoch": 9.373000639795265,
      "grad_norm": 0.10593709093308551,
      "learning_rate": 2.0548174452960868e-07,
      "loss": 0.0009,
      "step": 14650
    },
    {
      "epoch": 9.37364043506078,
      "grad_norm": 0.11420033444834696,
      "learning_rate": 2.0506404639795474e-07,
      "loss": 0.0009,
      "step": 14651
    },
    {
      "epoch": 9.374280230326296,
      "grad_norm": 0.08972562334052414,
      "learning_rate": 2.0464676884342837e-07,
      "loss": 0.0008,
      "step": 14652
    },
    {
      "epoch": 9.37492002559181,
      "grad_norm": 0.12648101919290364,
      "learning_rate": 2.0422991188394526e-07,
      "loss": 0.0009,
      "step": 14653
    },
    {
      "epoch": 9.375559820857326,
      "grad_norm": 0.09339085070298678,
      "learning_rate": 2.0381347553740437e-07,
      "loss": 0.0008,
      "step": 14654
    },
    {
      "epoch": 9.376199616122841,
      "grad_norm": 0.10916779048554746,
      "learning_rate": 2.033974598216859e-07,
      "loss": 0.0008,
      "step": 14655
    },
    {
      "epoch": 9.376839411388357,
      "grad_norm": 0.11987455269373726,
      "learning_rate": 2.0298186475465553e-07,
      "loss": 0.0008,
      "step": 14656
    },
    {
      "epoch": 9.37747920665387,
      "grad_norm": 0.11817605016254717,
      "learning_rate": 2.0256669035415454e-07,
      "loss": 0.001,
      "step": 14657
    },
    {
      "epoch": 9.378119001919385,
      "grad_norm": 0.1210223323236355,
      "learning_rate": 2.0215193663801203e-07,
      "loss": 0.0009,
      "step": 14658
    },
    {
      "epoch": 9.3787587971849,
      "grad_norm": 0.08417436084984235,
      "learning_rate": 2.0173760362403704e-07,
      "loss": 0.0007,
      "step": 14659
    },
    {
      "epoch": 9.379398592450416,
      "grad_norm": 0.2183453029072348,
      "learning_rate": 2.0132369133001873e-07,
      "loss": 0.0011,
      "step": 14660
    },
    {
      "epoch": 9.380038387715931,
      "grad_norm": 0.1334207247063911,
      "learning_rate": 2.0091019977373172e-07,
      "loss": 0.0009,
      "step": 14661
    },
    {
      "epoch": 9.380678182981447,
      "grad_norm": 0.05682563238056977,
      "learning_rate": 2.004971289729285e-07,
      "loss": 0.0005,
      "step": 14662
    },
    {
      "epoch": 9.381317978246962,
      "grad_norm": 0.20174392192781376,
      "learning_rate": 2.000844789453471e-07,
      "loss": 0.0009,
      "step": 14663
    },
    {
      "epoch": 9.381957773512475,
      "grad_norm": 0.10690397948966121,
      "learning_rate": 1.9967224970870558e-07,
      "loss": 0.0007,
      "step": 14664
    },
    {
      "epoch": 9.38259756877799,
      "grad_norm": 0.12895443859650874,
      "learning_rate": 1.992604412807031e-07,
      "loss": 0.0009,
      "step": 14665
    },
    {
      "epoch": 9.383237364043506,
      "grad_norm": 0.13167725883951847,
      "learning_rate": 1.988490536790244e-07,
      "loss": 0.001,
      "step": 14666
    },
    {
      "epoch": 9.383877159309021,
      "grad_norm": 0.07077218757528986,
      "learning_rate": 1.9843808692133204e-07,
      "loss": 0.0006,
      "step": 14667
    },
    {
      "epoch": 9.384516954574536,
      "grad_norm": 0.09847355007256375,
      "learning_rate": 1.980275410252719e-07,
      "loss": 0.0008,
      "step": 14668
    },
    {
      "epoch": 9.385156749840052,
      "grad_norm": 0.10460572149854404,
      "learning_rate": 1.9761741600847318e-07,
      "loss": 0.0009,
      "step": 14669
    },
    {
      "epoch": 9.385796545105567,
      "grad_norm": 0.07998930226865207,
      "learning_rate": 1.9720771188854626e-07,
      "loss": 0.0005,
      "step": 14670
    },
    {
      "epoch": 9.38643634037108,
      "grad_norm": 0.10745540601204143,
      "learning_rate": 1.9679842868308152e-07,
      "loss": 0.0008,
      "step": 14671
    },
    {
      "epoch": 9.387076135636596,
      "grad_norm": 0.0809027397903271,
      "learning_rate": 1.963895664096538e-07,
      "loss": 0.0007,
      "step": 14672
    },
    {
      "epoch": 9.387715930902111,
      "grad_norm": 0.09699740974405957,
      "learning_rate": 1.95981125085819e-07,
      "loss": 0.0008,
      "step": 14673
    },
    {
      "epoch": 9.388355726167626,
      "grad_norm": 0.08775527170469936,
      "learning_rate": 1.9557310472911538e-07,
      "loss": 0.0008,
      "step": 14674
    },
    {
      "epoch": 9.388995521433142,
      "grad_norm": 0.13872774765050647,
      "learning_rate": 1.9516550535706115e-07,
      "loss": 0.001,
      "step": 14675
    },
    {
      "epoch": 9.389635316698657,
      "grad_norm": 0.10046451231560033,
      "learning_rate": 1.9475832698716e-07,
      "loss": 0.0008,
      "step": 14676
    },
    {
      "epoch": 9.390275111964172,
      "grad_norm": 0.08317435734045173,
      "learning_rate": 1.943515696368925e-07,
      "loss": 0.0006,
      "step": 14677
    },
    {
      "epoch": 9.390914907229687,
      "grad_norm": 0.1657943225240117,
      "learning_rate": 1.9394523332372573e-07,
      "loss": 0.001,
      "step": 14678
    },
    {
      "epoch": 9.3915547024952,
      "grad_norm": 0.16306713546710333,
      "learning_rate": 1.9353931806510794e-07,
      "loss": 0.0009,
      "step": 14679
    },
    {
      "epoch": 9.392194497760716,
      "grad_norm": 0.11867558291007702,
      "learning_rate": 1.9313382387846635e-07,
      "loss": 0.0009,
      "step": 14680
    },
    {
      "epoch": 9.392834293026231,
      "grad_norm": 0.16229198373402762,
      "learning_rate": 1.9272875078121367e-07,
      "loss": 0.001,
      "step": 14681
    },
    {
      "epoch": 9.393474088291747,
      "grad_norm": 0.13976546839014486,
      "learning_rate": 1.9232409879074375e-07,
      "loss": 0.0012,
      "step": 14682
    },
    {
      "epoch": 9.394113883557262,
      "grad_norm": 0.1765417202944935,
      "learning_rate": 1.919198679244294e-07,
      "loss": 0.001,
      "step": 14683
    },
    {
      "epoch": 9.394753678822777,
      "grad_norm": 0.1654410667239934,
      "learning_rate": 1.9151605819962894e-07,
      "loss": 0.001,
      "step": 14684
    },
    {
      "epoch": 9.395393474088293,
      "grad_norm": 0.15154257645513614,
      "learning_rate": 1.911126696336807e-07,
      "loss": 0.0009,
      "step": 14685
    },
    {
      "epoch": 9.396033269353806,
      "grad_norm": 0.13118926239190887,
      "learning_rate": 1.907097022439064e-07,
      "loss": 0.0009,
      "step": 14686
    },
    {
      "epoch": 9.396673064619321,
      "grad_norm": 0.10451903387853599,
      "learning_rate": 1.9030715604760775e-07,
      "loss": 0.0008,
      "step": 14687
    },
    {
      "epoch": 9.397312859884837,
      "grad_norm": 0.14051623015352196,
      "learning_rate": 1.8990503106206982e-07,
      "loss": 0.0012,
      "step": 14688
    },
    {
      "epoch": 9.397952655150352,
      "grad_norm": 0.12964082628404122,
      "learning_rate": 1.895033273045588e-07,
      "loss": 0.0008,
      "step": 14689
    },
    {
      "epoch": 9.398592450415867,
      "grad_norm": 0.1418892926069493,
      "learning_rate": 1.891020447923242e-07,
      "loss": 0.0009,
      "step": 14690
    },
    {
      "epoch": 9.399232245681382,
      "grad_norm": 0.15941582563623796,
      "learning_rate": 1.8870118354259448e-07,
      "loss": 0.0011,
      "step": 14691
    },
    {
      "epoch": 9.399872040946898,
      "grad_norm": 0.14417399921135027,
      "learning_rate": 1.8830074357258367e-07,
      "loss": 0.001,
      "step": 14692
    },
    {
      "epoch": 9.400511836212413,
      "grad_norm": 0.15651645407247977,
      "learning_rate": 1.8790072489948574e-07,
      "loss": 0.0009,
      "step": 14693
    },
    {
      "epoch": 9.401151631477926,
      "grad_norm": 0.12789967208708655,
      "learning_rate": 1.8750112754047588e-07,
      "loss": 0.001,
      "step": 14694
    },
    {
      "epoch": 9.401791426743442,
      "grad_norm": 0.15849677382420155,
      "learning_rate": 1.871019515127115e-07,
      "loss": 0.0009,
      "step": 14695
    },
    {
      "epoch": 9.402431222008957,
      "grad_norm": 0.21912272092929597,
      "learning_rate": 1.8670319683333438e-07,
      "loss": 0.0011,
      "step": 14696
    },
    {
      "epoch": 9.403071017274472,
      "grad_norm": 0.13466364188520993,
      "learning_rate": 1.863048635194664e-07,
      "loss": 0.0012,
      "step": 14697
    },
    {
      "epoch": 9.403710812539988,
      "grad_norm": 0.19017620991265227,
      "learning_rate": 1.8590695158820838e-07,
      "loss": 0.0014,
      "step": 14698
    },
    {
      "epoch": 9.404350607805503,
      "grad_norm": 0.12321485896660246,
      "learning_rate": 1.855094610566499e-07,
      "loss": 0.0007,
      "step": 14699
    },
    {
      "epoch": 9.404990403071018,
      "grad_norm": 0.17556916711168544,
      "learning_rate": 1.8511239194185515e-07,
      "loss": 0.0011,
      "step": 14700
    },
    {
      "epoch": 9.405630198336532,
      "grad_norm": 0.11260331741000892,
      "learning_rate": 1.84715744260876e-07,
      "loss": 0.0009,
      "step": 14701
    },
    {
      "epoch": 9.406269993602047,
      "grad_norm": 0.1183013969801682,
      "learning_rate": 1.843195180307411e-07,
      "loss": 0.0009,
      "step": 14702
    },
    {
      "epoch": 9.406909788867562,
      "grad_norm": 0.20070667437657488,
      "learning_rate": 1.8392371326846682e-07,
      "loss": 0.0014,
      "step": 14703
    },
    {
      "epoch": 9.407549584133077,
      "grad_norm": 0.1103588212105573,
      "learning_rate": 1.835283299910473e-07,
      "loss": 0.0009,
      "step": 14704
    },
    {
      "epoch": 9.408189379398593,
      "grad_norm": 0.10368047016747767,
      "learning_rate": 1.831333682154579e-07,
      "loss": 0.001,
      "step": 14705
    },
    {
      "epoch": 9.408829174664108,
      "grad_norm": 0.18164811419747406,
      "learning_rate": 1.827388279586595e-07,
      "loss": 0.0009,
      "step": 14706
    },
    {
      "epoch": 9.409468969929623,
      "grad_norm": 0.25010547196901983,
      "learning_rate": 1.8234470923759185e-07,
      "loss": 0.0013,
      "step": 14707
    },
    {
      "epoch": 9.410108765195137,
      "grad_norm": 0.08739191383123836,
      "learning_rate": 1.8195101206917808e-07,
      "loss": 0.0008,
      "step": 14708
    },
    {
      "epoch": 9.410748560460652,
      "grad_norm": 0.10146862281145418,
      "learning_rate": 1.815577364703236e-07,
      "loss": 0.0008,
      "step": 14709
    },
    {
      "epoch": 9.411388355726167,
      "grad_norm": 0.10926520095919696,
      "learning_rate": 1.8116488245791376e-07,
      "loss": 0.0008,
      "step": 14710
    },
    {
      "epoch": 9.412028150991683,
      "grad_norm": 0.09406651936147122,
      "learning_rate": 1.8077245004881728e-07,
      "loss": 0.0008,
      "step": 14711
    },
    {
      "epoch": 9.412667946257198,
      "grad_norm": 0.12462621837913566,
      "learning_rate": 1.8038043925988512e-07,
      "loss": 0.0009,
      "step": 14712
    },
    {
      "epoch": 9.413307741522713,
      "grad_norm": 0.1291516197300304,
      "learning_rate": 1.7998885010794942e-07,
      "loss": 0.0009,
      "step": 14713
    },
    {
      "epoch": 9.413947536788228,
      "grad_norm": 0.18195724532848992,
      "learning_rate": 1.7959768260982336e-07,
      "loss": 0.0009,
      "step": 14714
    },
    {
      "epoch": 9.414587332053744,
      "grad_norm": 0.13665328752012001,
      "learning_rate": 1.792069367823035e-07,
      "loss": 0.001,
      "step": 14715
    },
    {
      "epoch": 9.415227127319257,
      "grad_norm": 0.11840228696018477,
      "learning_rate": 1.7881661264216866e-07,
      "loss": 0.0007,
      "step": 14716
    },
    {
      "epoch": 9.415866922584772,
      "grad_norm": 0.08179676875835408,
      "learning_rate": 1.7842671020617763e-07,
      "loss": 0.0007,
      "step": 14717
    },
    {
      "epoch": 9.416506717850288,
      "grad_norm": 0.08731363095507019,
      "learning_rate": 1.7803722949107149e-07,
      "loss": 0.0008,
      "step": 14718
    },
    {
      "epoch": 9.417146513115803,
      "grad_norm": 0.23042848949911357,
      "learning_rate": 1.7764817051357687e-07,
      "loss": 0.0011,
      "step": 14719
    },
    {
      "epoch": 9.417786308381318,
      "grad_norm": 0.1995587112006397,
      "learning_rate": 1.7725953329039592e-07,
      "loss": 0.0013,
      "step": 14720
    },
    {
      "epoch": 9.418426103646834,
      "grad_norm": 0.11121553091084539,
      "learning_rate": 1.7687131783821754e-07,
      "loss": 0.0008,
      "step": 14721
    },
    {
      "epoch": 9.419065898912349,
      "grad_norm": 0.14028929520643785,
      "learning_rate": 1.7648352417371174e-07,
      "loss": 0.0009,
      "step": 14722
    },
    {
      "epoch": 9.419705694177862,
      "grad_norm": 0.0892575540426823,
      "learning_rate": 1.7609615231352738e-07,
      "loss": 0.0008,
      "step": 14723
    },
    {
      "epoch": 9.420345489443378,
      "grad_norm": 0.1968917186322585,
      "learning_rate": 1.757092022743001e-07,
      "loss": 0.0012,
      "step": 14724
    },
    {
      "epoch": 9.420985284708893,
      "grad_norm": 0.11162375835203303,
      "learning_rate": 1.7532267407264215e-07,
      "loss": 0.0009,
      "step": 14725
    },
    {
      "epoch": 9.421625079974408,
      "grad_norm": 0.12772438968473965,
      "learning_rate": 1.7493656772515244e-07,
      "loss": 0.0008,
      "step": 14726
    },
    {
      "epoch": 9.422264875239923,
      "grad_norm": 0.10666959536896455,
      "learning_rate": 1.7455088324840886e-07,
      "loss": 0.0006,
      "step": 14727
    },
    {
      "epoch": 9.422904670505439,
      "grad_norm": 0.09941662792291821,
      "learning_rate": 1.7416562065897146e-07,
      "loss": 0.0008,
      "step": 14728
    },
    {
      "epoch": 9.423544465770954,
      "grad_norm": 0.17060959781308904,
      "learning_rate": 1.7378077997338483e-07,
      "loss": 0.0014,
      "step": 14729
    },
    {
      "epoch": 9.42418426103647,
      "grad_norm": 0.21468330216249218,
      "learning_rate": 1.733963612081724e-07,
      "loss": 0.001,
      "step": 14730
    },
    {
      "epoch": 9.424824056301983,
      "grad_norm": 0.16899050224771736,
      "learning_rate": 1.730123643798398e-07,
      "loss": 0.001,
      "step": 14731
    },
    {
      "epoch": 9.425463851567498,
      "grad_norm": 0.09733178428890471,
      "learning_rate": 1.7262878950487394e-07,
      "loss": 0.0008,
      "step": 14732
    },
    {
      "epoch": 9.426103646833013,
      "grad_norm": 0.11245271493121707,
      "learning_rate": 1.7224563659974713e-07,
      "loss": 0.0009,
      "step": 14733
    },
    {
      "epoch": 9.426743442098529,
      "grad_norm": 0.16790306519527098,
      "learning_rate": 1.718629056809118e-07,
      "loss": 0.001,
      "step": 14734
    },
    {
      "epoch": 9.427383237364044,
      "grad_norm": 0.09963710371678029,
      "learning_rate": 1.7148059676479922e-07,
      "loss": 0.0009,
      "step": 14735
    },
    {
      "epoch": 9.428023032629559,
      "grad_norm": 0.12040227827135136,
      "learning_rate": 1.7109870986782627e-07,
      "loss": 0.0009,
      "step": 14736
    },
    {
      "epoch": 9.428662827895074,
      "grad_norm": 0.09208042221622732,
      "learning_rate": 1.7071724500639096e-07,
      "loss": 0.0008,
      "step": 14737
    },
    {
      "epoch": 9.429302623160588,
      "grad_norm": 0.10340079458038291,
      "learning_rate": 1.703362021968713e-07,
      "loss": 0.0008,
      "step": 14738
    },
    {
      "epoch": 9.429942418426103,
      "grad_norm": 0.08636214375711061,
      "learning_rate": 1.6995558145562975e-07,
      "loss": 0.0007,
      "step": 14739
    },
    {
      "epoch": 9.430582213691618,
      "grad_norm": 0.1332031563925271,
      "learning_rate": 1.6957538279900988e-07,
      "loss": 0.001,
      "step": 14740
    },
    {
      "epoch": 9.431222008957134,
      "grad_norm": 0.13409210536400187,
      "learning_rate": 1.691956062433353e-07,
      "loss": 0.001,
      "step": 14741
    },
    {
      "epoch": 9.431861804222649,
      "grad_norm": 0.19207628947102126,
      "learning_rate": 1.6881625180491413e-07,
      "loss": 0.0011,
      "step": 14742
    },
    {
      "epoch": 9.432501599488164,
      "grad_norm": 0.3666688628457752,
      "learning_rate": 1.684373195000344e-07,
      "loss": 0.0014,
      "step": 14743
    },
    {
      "epoch": 9.43314139475368,
      "grad_norm": 0.12910347749143508,
      "learning_rate": 1.6805880934496754e-07,
      "loss": 0.0009,
      "step": 14744
    },
    {
      "epoch": 9.433781190019193,
      "grad_norm": 0.08867339284383428,
      "learning_rate": 1.6768072135596504e-07,
      "loss": 0.0007,
      "step": 14745
    },
    {
      "epoch": 9.434420985284708,
      "grad_norm": 0.11597973945855874,
      "learning_rate": 1.6730305554926385e-07,
      "loss": 0.0008,
      "step": 14746
    },
    {
      "epoch": 9.435060780550224,
      "grad_norm": 0.16578856997317126,
      "learning_rate": 1.669258119410766e-07,
      "loss": 0.0013,
      "step": 14747
    },
    {
      "epoch": 9.435700575815739,
      "grad_norm": 0.20530227919436647,
      "learning_rate": 1.6654899054760366e-07,
      "loss": 0.001,
      "step": 14748
    },
    {
      "epoch": 9.436340371081254,
      "grad_norm": 0.11295864238107275,
      "learning_rate": 1.661725913850254e-07,
      "loss": 0.0008,
      "step": 14749
    },
    {
      "epoch": 9.43698016634677,
      "grad_norm": 0.11716146100175338,
      "learning_rate": 1.6579661446950334e-07,
      "loss": 0.0008,
      "step": 14750
    },
    {
      "epoch": 9.437619961612285,
      "grad_norm": 0.1795959312951831,
      "learning_rate": 1.6542105981718016e-07,
      "loss": 0.0011,
      "step": 14751
    },
    {
      "epoch": 9.4382597568778,
      "grad_norm": 0.1556648369108809,
      "learning_rate": 1.6504592744418292e-07,
      "loss": 0.0012,
      "step": 14752
    },
    {
      "epoch": 9.438899552143313,
      "grad_norm": 0.13044810036384052,
      "learning_rate": 1.6467121736661762e-07,
      "loss": 0.0008,
      "step": 14753
    },
    {
      "epoch": 9.439539347408829,
      "grad_norm": 0.10629389313308918,
      "learning_rate": 1.6429692960057586e-07,
      "loss": 0.0008,
      "step": 14754
    },
    {
      "epoch": 9.440179142674344,
      "grad_norm": 0.1826947525107814,
      "learning_rate": 1.639230641621259e-07,
      "loss": 0.0009,
      "step": 14755
    },
    {
      "epoch": 9.44081893793986,
      "grad_norm": 0.11953619779432387,
      "learning_rate": 1.6354962106732375e-07,
      "loss": 0.0008,
      "step": 14756
    },
    {
      "epoch": 9.441458733205375,
      "grad_norm": 0.10860624324083434,
      "learning_rate": 1.631766003322033e-07,
      "loss": 0.0009,
      "step": 14757
    },
    {
      "epoch": 9.44209852847089,
      "grad_norm": 0.1639909760797282,
      "learning_rate": 1.6280400197277945e-07,
      "loss": 0.0014,
      "step": 14758
    },
    {
      "epoch": 9.442738323736405,
      "grad_norm": 0.14434937373880774,
      "learning_rate": 1.6243182600505503e-07,
      "loss": 0.001,
      "step": 14759
    },
    {
      "epoch": 9.443378119001919,
      "grad_norm": 0.12060227531414842,
      "learning_rate": 1.6206007244500721e-07,
      "loss": 0.0009,
      "step": 14760
    },
    {
      "epoch": 9.444017914267434,
      "grad_norm": 0.11413579591888119,
      "learning_rate": 1.616887413085999e-07,
      "loss": 0.0008,
      "step": 14761
    },
    {
      "epoch": 9.44465770953295,
      "grad_norm": 0.11265713985020769,
      "learning_rate": 1.6131783261177591e-07,
      "loss": 0.0009,
      "step": 14762
    },
    {
      "epoch": 9.445297504798464,
      "grad_norm": 0.1464068786566266,
      "learning_rate": 1.609473463704647e-07,
      "loss": 0.001,
      "step": 14763
    },
    {
      "epoch": 9.44593730006398,
      "grad_norm": 0.21560666150529825,
      "learning_rate": 1.6057728260057025e-07,
      "loss": 0.0013,
      "step": 14764
    },
    {
      "epoch": 9.446577095329495,
      "grad_norm": 0.5648090668077429,
      "learning_rate": 1.6020764131798538e-07,
      "loss": 0.0013,
      "step": 14765
    },
    {
      "epoch": 9.44721689059501,
      "grad_norm": 0.14456571784351363,
      "learning_rate": 1.5983842253858074e-07,
      "loss": 0.0012,
      "step": 14766
    },
    {
      "epoch": 9.447856685860526,
      "grad_norm": 0.1217844732531275,
      "learning_rate": 1.5946962627820917e-07,
      "loss": 0.001,
      "step": 14767
    },
    {
      "epoch": 9.448496481126039,
      "grad_norm": 0.2017135042444951,
      "learning_rate": 1.5910125255270803e-07,
      "loss": 0.0011,
      "step": 14768
    },
    {
      "epoch": 9.449136276391554,
      "grad_norm": 0.16014000931873434,
      "learning_rate": 1.5873330137789245e-07,
      "loss": 0.0011,
      "step": 14769
    },
    {
      "epoch": 9.44977607165707,
      "grad_norm": 0.13039665503572315,
      "learning_rate": 1.583657727695631e-07,
      "loss": 0.0007,
      "step": 14770
    },
    {
      "epoch": 9.450415866922585,
      "grad_norm": 0.19685853888394894,
      "learning_rate": 1.5799866674350074e-07,
      "loss": 0.001,
      "step": 14771
    },
    {
      "epoch": 9.4510556621881,
      "grad_norm": 0.1315625184737033,
      "learning_rate": 1.5763198331546713e-07,
      "loss": 0.001,
      "step": 14772
    },
    {
      "epoch": 9.451695457453615,
      "grad_norm": 0.10623877252747449,
      "learning_rate": 1.5726572250120864e-07,
      "loss": 0.0008,
      "step": 14773
    },
    {
      "epoch": 9.45233525271913,
      "grad_norm": 0.12457995258920125,
      "learning_rate": 1.5689988431645155e-07,
      "loss": 0.0007,
      "step": 14774
    },
    {
      "epoch": 9.452975047984644,
      "grad_norm": 0.11104632044874284,
      "learning_rate": 1.5653446877690216e-07,
      "loss": 0.0008,
      "step": 14775
    },
    {
      "epoch": 9.45361484325016,
      "grad_norm": 0.18366013240814902,
      "learning_rate": 1.561694758982546e-07,
      "loss": 0.0009,
      "step": 14776
    },
    {
      "epoch": 9.454254638515675,
      "grad_norm": 0.08757466243762307,
      "learning_rate": 1.5580490569617745e-07,
      "loss": 0.0007,
      "step": 14777
    },
    {
      "epoch": 9.45489443378119,
      "grad_norm": 0.24159293893539052,
      "learning_rate": 1.554407581863271e-07,
      "loss": 0.0012,
      "step": 14778
    },
    {
      "epoch": 9.455534229046705,
      "grad_norm": 0.21235466595081615,
      "learning_rate": 1.5507703338433767e-07,
      "loss": 0.001,
      "step": 14779
    },
    {
      "epoch": 9.45617402431222,
      "grad_norm": 0.11263473133597406,
      "learning_rate": 1.547137313058289e-07,
      "loss": 0.0008,
      "step": 14780
    },
    {
      "epoch": 9.456813819577736,
      "grad_norm": 0.2679419937560905,
      "learning_rate": 1.5435085196639833e-07,
      "loss": 0.0014,
      "step": 14781
    },
    {
      "epoch": 9.45745361484325,
      "grad_norm": 0.10471277602960302,
      "learning_rate": 1.5398839538162902e-07,
      "loss": 0.001,
      "step": 14782
    },
    {
      "epoch": 9.458093410108765,
      "grad_norm": 0.07567865154712781,
      "learning_rate": 1.5362636156708188e-07,
      "loss": 0.0007,
      "step": 14783
    },
    {
      "epoch": 9.45873320537428,
      "grad_norm": 0.0703519428367898,
      "learning_rate": 1.5326475053830448e-07,
      "loss": 0.0006,
      "step": 14784
    },
    {
      "epoch": 9.459373000639795,
      "grad_norm": 0.07612550419518548,
      "learning_rate": 1.5290356231082216e-07,
      "loss": 0.0007,
      "step": 14785
    },
    {
      "epoch": 9.46001279590531,
      "grad_norm": 0.0823515475138406,
      "learning_rate": 1.5254279690014474e-07,
      "loss": 0.0007,
      "step": 14786
    },
    {
      "epoch": 9.460652591170826,
      "grad_norm": 0.15820243181422997,
      "learning_rate": 1.5218245432176205e-07,
      "loss": 0.001,
      "step": 14787
    },
    {
      "epoch": 9.461292386436341,
      "grad_norm": 0.06612887469991117,
      "learning_rate": 1.5182253459114726e-07,
      "loss": 0.0006,
      "step": 14788
    },
    {
      "epoch": 9.461932181701856,
      "grad_norm": 0.0753626867072825,
      "learning_rate": 1.5146303772375248e-07,
      "loss": 0.0007,
      "step": 14789
    },
    {
      "epoch": 9.46257197696737,
      "grad_norm": 0.10277756835666489,
      "learning_rate": 1.5110396373501758e-07,
      "loss": 0.0008,
      "step": 14790
    },
    {
      "epoch": 9.463211772232885,
      "grad_norm": 0.1083178110311075,
      "learning_rate": 1.5074531264035798e-07,
      "loss": 0.0009,
      "step": 14791
    },
    {
      "epoch": 9.4638515674984,
      "grad_norm": 0.21799788333014802,
      "learning_rate": 1.503870844551747e-07,
      "loss": 0.0008,
      "step": 14792
    },
    {
      "epoch": 9.464491362763916,
      "grad_norm": 0.12910204662309765,
      "learning_rate": 1.5002927919484765e-07,
      "loss": 0.0009,
      "step": 14793
    },
    {
      "epoch": 9.46513115802943,
      "grad_norm": 0.08303199995611886,
      "learning_rate": 1.4967189687474348e-07,
      "loss": 0.0007,
      "step": 14794
    },
    {
      "epoch": 9.465770953294946,
      "grad_norm": 0.10997058483466425,
      "learning_rate": 1.493149375102043e-07,
      "loss": 0.0009,
      "step": 14795
    },
    {
      "epoch": 9.466410748560461,
      "grad_norm": 0.10983095371455595,
      "learning_rate": 1.4895840111656013e-07,
      "loss": 0.0008,
      "step": 14796
    },
    {
      "epoch": 9.467050543825975,
      "grad_norm": 0.05359831885009148,
      "learning_rate": 1.4860228770911757e-07,
      "loss": 0.0006,
      "step": 14797
    },
    {
      "epoch": 9.46769033909149,
      "grad_norm": 0.11650739396239902,
      "learning_rate": 1.4824659730316883e-07,
      "loss": 0.0008,
      "step": 14798
    },
    {
      "epoch": 9.468330134357005,
      "grad_norm": 0.17526816824225075,
      "learning_rate": 1.4789132991398504e-07,
      "loss": 0.0008,
      "step": 14799
    },
    {
      "epoch": 9.46896992962252,
      "grad_norm": 0.10017537537989324,
      "learning_rate": 1.4753648555682288e-07,
      "loss": 0.0008,
      "step": 14800
    },
    {
      "epoch": 9.469609724888036,
      "grad_norm": 0.1688240043999374,
      "learning_rate": 1.4718206424691793e-07,
      "loss": 0.0012,
      "step": 14801
    },
    {
      "epoch": 9.470249520153551,
      "grad_norm": 0.1126485374821095,
      "learning_rate": 1.4682806599948807e-07,
      "loss": 0.0008,
      "step": 14802
    },
    {
      "epoch": 9.470889315419067,
      "grad_norm": 0.09361993160126723,
      "learning_rate": 1.464744908297333e-07,
      "loss": 0.0007,
      "step": 14803
    },
    {
      "epoch": 9.471529110684582,
      "grad_norm": 0.17187401689201037,
      "learning_rate": 1.4612133875283708e-07,
      "loss": 0.0013,
      "step": 14804
    },
    {
      "epoch": 9.472168905950095,
      "grad_norm": 0.16703585236311372,
      "learning_rate": 1.4576860978396057e-07,
      "loss": 0.0012,
      "step": 14805
    },
    {
      "epoch": 9.47280870121561,
      "grad_norm": 0.10384648256333535,
      "learning_rate": 1.4541630393825167e-07,
      "loss": 0.0008,
      "step": 14806
    },
    {
      "epoch": 9.473448496481126,
      "grad_norm": 0.16239890691670725,
      "learning_rate": 1.450644212308361e-07,
      "loss": 0.0009,
      "step": 14807
    },
    {
      "epoch": 9.474088291746641,
      "grad_norm": 0.11106139327766315,
      "learning_rate": 1.4471296167682392e-07,
      "loss": 0.0009,
      "step": 14808
    },
    {
      "epoch": 9.474728087012156,
      "grad_norm": 0.17628327239754354,
      "learning_rate": 1.4436192529130532e-07,
      "loss": 0.0011,
      "step": 14809
    },
    {
      "epoch": 9.475367882277672,
      "grad_norm": 0.1259894235197116,
      "learning_rate": 1.4401131208935493e-07,
      "loss": 0.0009,
      "step": 14810
    },
    {
      "epoch": 9.476007677543187,
      "grad_norm": 0.2008708377504474,
      "learning_rate": 1.4366112208602623e-07,
      "loss": 0.0012,
      "step": 14811
    },
    {
      "epoch": 9.4766474728087,
      "grad_norm": 0.2041774919013365,
      "learning_rate": 1.433113552963561e-07,
      "loss": 0.0009,
      "step": 14812
    },
    {
      "epoch": 9.477287268074216,
      "grad_norm": 0.13457906186652097,
      "learning_rate": 1.429620117353614e-07,
      "loss": 0.0011,
      "step": 14813
    },
    {
      "epoch": 9.477927063339731,
      "grad_norm": 0.14185534365522987,
      "learning_rate": 1.4261309141804458e-07,
      "loss": 0.0009,
      "step": 14814
    },
    {
      "epoch": 9.478566858605246,
      "grad_norm": 0.2010882460963694,
      "learning_rate": 1.4226459435938588e-07,
      "loss": 0.0013,
      "step": 14815
    },
    {
      "epoch": 9.479206653870762,
      "grad_norm": 0.08957970386182162,
      "learning_rate": 1.4191652057435112e-07,
      "loss": 0.0008,
      "step": 14816
    },
    {
      "epoch": 9.479846449136277,
      "grad_norm": 0.10458561291616152,
      "learning_rate": 1.4156887007788388e-07,
      "loss": 0.0009,
      "step": 14817
    },
    {
      "epoch": 9.480486244401792,
      "grad_norm": 0.12385119032191551,
      "learning_rate": 1.4122164288491336e-07,
      "loss": 0.001,
      "step": 14818
    },
    {
      "epoch": 9.481126039667306,
      "grad_norm": 0.17617223877057586,
      "learning_rate": 1.408748390103476e-07,
      "loss": 0.001,
      "step": 14819
    },
    {
      "epoch": 9.48176583493282,
      "grad_norm": 0.07632356783365353,
      "learning_rate": 1.4052845846907803e-07,
      "loss": 0.0006,
      "step": 14820
    },
    {
      "epoch": 9.482405630198336,
      "grad_norm": 0.11145278455483584,
      "learning_rate": 1.401825012759783e-07,
      "loss": 0.0008,
      "step": 14821
    },
    {
      "epoch": 9.483045425463851,
      "grad_norm": 0.11257308797709706,
      "learning_rate": 1.3983696744590215e-07,
      "loss": 0.0011,
      "step": 14822
    },
    {
      "epoch": 9.483685220729367,
      "grad_norm": 0.14245922236000294,
      "learning_rate": 1.394918569936865e-07,
      "loss": 0.001,
      "step": 14823
    },
    {
      "epoch": 9.484325015994882,
      "grad_norm": 0.19435479021694305,
      "learning_rate": 1.3914716993415068e-07,
      "loss": 0.001,
      "step": 14824
    },
    {
      "epoch": 9.484964811260397,
      "grad_norm": 0.13252892191795548,
      "learning_rate": 1.3880290628209504e-07,
      "loss": 0.001,
      "step": 14825
    },
    {
      "epoch": 9.485604606525913,
      "grad_norm": 0.12387834467129631,
      "learning_rate": 1.3845906605229998e-07,
      "loss": 0.0009,
      "step": 14826
    },
    {
      "epoch": 9.486244401791426,
      "grad_norm": 0.10648315028828577,
      "learning_rate": 1.3811564925953035e-07,
      "loss": 0.0008,
      "step": 14827
    },
    {
      "epoch": 9.486884197056941,
      "grad_norm": 0.13289184246015995,
      "learning_rate": 1.3777265591853218e-07,
      "loss": 0.0011,
      "step": 14828
    },
    {
      "epoch": 9.487523992322457,
      "grad_norm": 0.28372696922845725,
      "learning_rate": 1.3743008604403142e-07,
      "loss": 0.0017,
      "step": 14829
    },
    {
      "epoch": 9.488163787587972,
      "grad_norm": 0.11794818727773197,
      "learning_rate": 1.3708793965073963e-07,
      "loss": 0.0008,
      "step": 14830
    },
    {
      "epoch": 9.488803582853487,
      "grad_norm": 0.1810192029785874,
      "learning_rate": 1.367462167533473e-07,
      "loss": 0.0013,
      "step": 14831
    },
    {
      "epoch": 9.489443378119002,
      "grad_norm": 0.11219751896971783,
      "learning_rate": 1.3640491736652606e-07,
      "loss": 0.0008,
      "step": 14832
    },
    {
      "epoch": 9.490083173384518,
      "grad_norm": 0.15631291651571655,
      "learning_rate": 1.3606404150493191e-07,
      "loss": 0.001,
      "step": 14833
    },
    {
      "epoch": 9.490722968650031,
      "grad_norm": 0.13436980148467945,
      "learning_rate": 1.3572358918320205e-07,
      "loss": 0.0008,
      "step": 14834
    },
    {
      "epoch": 9.491362763915546,
      "grad_norm": 0.1608125585395249,
      "learning_rate": 1.3538356041595258e-07,
      "loss": 0.0009,
      "step": 14835
    },
    {
      "epoch": 9.492002559181062,
      "grad_norm": 0.11007172769467738,
      "learning_rate": 1.3504395521778625e-07,
      "loss": 0.0008,
      "step": 14836
    },
    {
      "epoch": 9.492642354446577,
      "grad_norm": 0.161255489499563,
      "learning_rate": 1.3470477360328472e-07,
      "loss": 0.0008,
      "step": 14837
    },
    {
      "epoch": 9.493282149712092,
      "grad_norm": 0.12600188325499254,
      "learning_rate": 1.343660155870108e-07,
      "loss": 0.001,
      "step": 14838
    },
    {
      "epoch": 9.493921944977608,
      "grad_norm": 0.1310916829724581,
      "learning_rate": 1.340276811835095e-07,
      "loss": 0.0009,
      "step": 14839
    },
    {
      "epoch": 9.494561740243123,
      "grad_norm": 0.12893339329500472,
      "learning_rate": 1.3368977040731034e-07,
      "loss": 0.001,
      "step": 14840
    },
    {
      "epoch": 9.495201535508638,
      "grad_norm": 0.19017586475908965,
      "learning_rate": 1.333522832729217e-07,
      "loss": 0.0009,
      "step": 14841
    },
    {
      "epoch": 9.495841330774152,
      "grad_norm": 0.10867333210027355,
      "learning_rate": 1.3301521979483423e-07,
      "loss": 0.0008,
      "step": 14842
    },
    {
      "epoch": 9.496481126039667,
      "grad_norm": 0.11811988807475675,
      "learning_rate": 1.3267857998752186e-07,
      "loss": 0.001,
      "step": 14843
    },
    {
      "epoch": 9.497120921305182,
      "grad_norm": 0.22316432944735282,
      "learning_rate": 1.3234236386543752e-07,
      "loss": 0.0013,
      "step": 14844
    },
    {
      "epoch": 9.497760716570697,
      "grad_norm": 0.12876838217811465,
      "learning_rate": 1.3200657144301965e-07,
      "loss": 0.0011,
      "step": 14845
    },
    {
      "epoch": 9.498400511836213,
      "grad_norm": 0.15511903133087424,
      "learning_rate": 1.316712027346856e-07,
      "loss": 0.0009,
      "step": 14846
    },
    {
      "epoch": 9.499040307101728,
      "grad_norm": 0.16447655039514358,
      "learning_rate": 1.3133625775483604e-07,
      "loss": 0.0011,
      "step": 14847
    },
    {
      "epoch": 9.499680102367243,
      "grad_norm": 0.08713685655707147,
      "learning_rate": 1.3100173651785174e-07,
      "loss": 0.0007,
      "step": 14848
    },
    {
      "epoch": 9.500319897632757,
      "grad_norm": 0.12380233715490098,
      "learning_rate": 1.3066763903809788e-07,
      "loss": 0.0009,
      "step": 14849
    },
    {
      "epoch": 9.500959692898272,
      "grad_norm": 0.16366337142323284,
      "learning_rate": 1.303339653299196e-07,
      "loss": 0.0013,
      "step": 14850
    },
    {
      "epoch": 9.501599488163787,
      "grad_norm": 0.09203898982242116,
      "learning_rate": 1.3000071540764326e-07,
      "loss": 0.0007,
      "step": 14851
    },
    {
      "epoch": 9.502239283429303,
      "grad_norm": 0.14281018533725945,
      "learning_rate": 1.296678892855796e-07,
      "loss": 0.0009,
      "step": 14852
    },
    {
      "epoch": 9.502879078694818,
      "grad_norm": 0.0734878455185831,
      "learning_rate": 1.2933548697801834e-07,
      "loss": 0.0007,
      "step": 14853
    },
    {
      "epoch": 9.503518873960333,
      "grad_norm": 0.10835658381448715,
      "learning_rate": 1.290035084992336e-07,
      "loss": 0.0008,
      "step": 14854
    },
    {
      "epoch": 9.504158669225848,
      "grad_norm": 0.10736400527123408,
      "learning_rate": 1.2867195386347843e-07,
      "loss": 0.0008,
      "step": 14855
    },
    {
      "epoch": 9.504798464491362,
      "grad_norm": 0.14382102919630402,
      "learning_rate": 1.283408230849903e-07,
      "loss": 0.0011,
      "step": 14856
    },
    {
      "epoch": 9.505438259756877,
      "grad_norm": 0.11297545410655159,
      "learning_rate": 1.2801011617798676e-07,
      "loss": 0.0009,
      "step": 14857
    },
    {
      "epoch": 9.506078055022392,
      "grad_norm": 0.1311348378319785,
      "learning_rate": 1.2767983315666755e-07,
      "loss": 0.001,
      "step": 14858
    },
    {
      "epoch": 9.506717850287908,
      "grad_norm": 0.1635349386272084,
      "learning_rate": 1.2734997403521465e-07,
      "loss": 0.0011,
      "step": 14859
    },
    {
      "epoch": 9.507357645553423,
      "grad_norm": 0.07964924765369008,
      "learning_rate": 1.270205388277923e-07,
      "loss": 0.0006,
      "step": 14860
    },
    {
      "epoch": 9.507997440818938,
      "grad_norm": 0.16163010744071632,
      "learning_rate": 1.2669152754854474e-07,
      "loss": 0.001,
      "step": 14861
    },
    {
      "epoch": 9.508637236084454,
      "grad_norm": 0.10463548562974678,
      "learning_rate": 1.2636294021159955e-07,
      "loss": 0.0009,
      "step": 14862
    },
    {
      "epoch": 9.509277031349969,
      "grad_norm": 0.12691035587746516,
      "learning_rate": 1.2603477683106657e-07,
      "loss": 0.0008,
      "step": 14863
    },
    {
      "epoch": 9.509916826615482,
      "grad_norm": 0.141248322652214,
      "learning_rate": 1.2570703742103562e-07,
      "loss": 0.001,
      "step": 14864
    },
    {
      "epoch": 9.510556621880998,
      "grad_norm": 0.09001042351723443,
      "learning_rate": 1.2537972199557992e-07,
      "loss": 0.0008,
      "step": 14865
    },
    {
      "epoch": 9.511196417146513,
      "grad_norm": 0.1314541604523205,
      "learning_rate": 1.2505283056875263e-07,
      "loss": 0.0009,
      "step": 14866
    },
    {
      "epoch": 9.511836212412028,
      "grad_norm": 0.1809340356865274,
      "learning_rate": 1.2472636315459142e-07,
      "loss": 0.0011,
      "step": 14867
    },
    {
      "epoch": 9.512476007677543,
      "grad_norm": 0.13370574294629411,
      "learning_rate": 1.2440031976711287e-07,
      "loss": 0.0009,
      "step": 14868
    },
    {
      "epoch": 9.513115802943059,
      "grad_norm": 0.1090328449560812,
      "learning_rate": 1.240747004203169e-07,
      "loss": 0.0008,
      "step": 14869
    },
    {
      "epoch": 9.513755598208574,
      "grad_norm": 0.07010401836743527,
      "learning_rate": 1.237495051281856e-07,
      "loss": 0.0006,
      "step": 14870
    },
    {
      "epoch": 9.514395393474087,
      "grad_norm": 0.1400153626345532,
      "learning_rate": 1.2342473390468235e-07,
      "loss": 0.0009,
      "step": 14871
    },
    {
      "epoch": 9.515035188739603,
      "grad_norm": 0.13473572059584285,
      "learning_rate": 1.2310038676375257e-07,
      "loss": 0.0008,
      "step": 14872
    },
    {
      "epoch": 9.515674984005118,
      "grad_norm": 0.17339608245567573,
      "learning_rate": 1.2277646371932184e-07,
      "loss": 0.0011,
      "step": 14873
    },
    {
      "epoch": 9.516314779270633,
      "grad_norm": 0.12677570830229756,
      "learning_rate": 1.2245296478530012e-07,
      "loss": 0.0008,
      "step": 14874
    },
    {
      "epoch": 9.516954574536149,
      "grad_norm": 0.13085939766056415,
      "learning_rate": 1.221298899755774e-07,
      "loss": 0.001,
      "step": 14875
    },
    {
      "epoch": 9.517594369801664,
      "grad_norm": 0.11795605865023687,
      "learning_rate": 1.2180723930402373e-07,
      "loss": 0.0008,
      "step": 14876
    },
    {
      "epoch": 9.51823416506718,
      "grad_norm": 0.0712316252889191,
      "learning_rate": 1.2148501278449687e-07,
      "loss": 0.0006,
      "step": 14877
    },
    {
      "epoch": 9.518873960332694,
      "grad_norm": 0.10581980363640917,
      "learning_rate": 1.2116321043083135e-07,
      "loss": 0.0008,
      "step": 14878
    },
    {
      "epoch": 9.519513755598208,
      "grad_norm": 0.16425918140439494,
      "learning_rate": 1.2084183225684386e-07,
      "loss": 0.001,
      "step": 14879
    },
    {
      "epoch": 9.520153550863723,
      "grad_norm": 0.20600832452554155,
      "learning_rate": 1.2052087827633453e-07,
      "loss": 0.001,
      "step": 14880
    },
    {
      "epoch": 9.520793346129238,
      "grad_norm": 0.19417121002923862,
      "learning_rate": 1.202003485030845e-07,
      "loss": 0.0014,
      "step": 14881
    },
    {
      "epoch": 9.521433141394754,
      "grad_norm": 0.18925823740618616,
      "learning_rate": 1.198802429508572e-07,
      "loss": 0.0013,
      "step": 14882
    },
    {
      "epoch": 9.522072936660269,
      "grad_norm": 0.16072410897888229,
      "learning_rate": 1.1956056163339614e-07,
      "loss": 0.0011,
      "step": 14883
    },
    {
      "epoch": 9.522712731925784,
      "grad_norm": 0.206282504058688,
      "learning_rate": 1.1924130456442916e-07,
      "loss": 0.0015,
      "step": 14884
    },
    {
      "epoch": 9.5233525271913,
      "grad_norm": 0.174330822783774,
      "learning_rate": 1.1892247175766425e-07,
      "loss": 0.0013,
      "step": 14885
    },
    {
      "epoch": 9.523992322456813,
      "grad_norm": 0.22370043517085098,
      "learning_rate": 1.1860406322679041e-07,
      "loss": 0.0014,
      "step": 14886
    },
    {
      "epoch": 9.524632117722328,
      "grad_norm": 0.12313146007285394,
      "learning_rate": 1.1828607898548117e-07,
      "loss": 0.001,
      "step": 14887
    },
    {
      "epoch": 9.525271912987844,
      "grad_norm": 0.08225605997436024,
      "learning_rate": 1.1796851904739004e-07,
      "loss": 0.0008,
      "step": 14888
    },
    {
      "epoch": 9.525911708253359,
      "grad_norm": 0.11147059110991826,
      "learning_rate": 1.1765138342615057e-07,
      "loss": 0.0008,
      "step": 14889
    },
    {
      "epoch": 9.526551503518874,
      "grad_norm": 0.1421325138570956,
      "learning_rate": 1.1733467213538186e-07,
      "loss": 0.001,
      "step": 14890
    },
    {
      "epoch": 9.52719129878439,
      "grad_norm": 0.14337229022889753,
      "learning_rate": 1.1701838518868302e-07,
      "loss": 0.0009,
      "step": 14891
    },
    {
      "epoch": 9.527831094049905,
      "grad_norm": 0.20032418132069482,
      "learning_rate": 1.1670252259963321e-07,
      "loss": 0.001,
      "step": 14892
    },
    {
      "epoch": 9.528470889315418,
      "grad_norm": 0.12970123345994972,
      "learning_rate": 1.1638708438179713e-07,
      "loss": 0.0008,
      "step": 14893
    },
    {
      "epoch": 9.529110684580933,
      "grad_norm": 0.18434612150020951,
      "learning_rate": 1.1607207054871727e-07,
      "loss": 0.0011,
      "step": 14894
    },
    {
      "epoch": 9.529750479846449,
      "grad_norm": 0.10508874839289425,
      "learning_rate": 1.1575748111392172e-07,
      "loss": 0.0008,
      "step": 14895
    },
    {
      "epoch": 9.530390275111964,
      "grad_norm": 0.10564538939535942,
      "learning_rate": 1.1544331609091519e-07,
      "loss": 0.0008,
      "step": 14896
    },
    {
      "epoch": 9.53103007037748,
      "grad_norm": 0.1862857195469138,
      "learning_rate": 1.1512957549319137e-07,
      "loss": 0.0014,
      "step": 14897
    },
    {
      "epoch": 9.531669865642995,
      "grad_norm": 0.09326738505548515,
      "learning_rate": 1.1481625933421836e-07,
      "loss": 0.0007,
      "step": 14898
    },
    {
      "epoch": 9.53230966090851,
      "grad_norm": 0.20421941169989266,
      "learning_rate": 1.1450336762745096e-07,
      "loss": 0.0014,
      "step": 14899
    },
    {
      "epoch": 9.532949456174025,
      "grad_norm": 0.10840530280881598,
      "learning_rate": 1.1419090038632507e-07,
      "loss": 0.0006,
      "step": 14900
    },
    {
      "epoch": 9.533589251439539,
      "grad_norm": 0.15885609668189918,
      "learning_rate": 1.1387885762425555e-07,
      "loss": 0.001,
      "step": 14901
    },
    {
      "epoch": 9.534229046705054,
      "grad_norm": 0.19397267072969765,
      "learning_rate": 1.1356723935464275e-07,
      "loss": 0.0012,
      "step": 14902
    },
    {
      "epoch": 9.53486884197057,
      "grad_norm": 0.13899555270650668,
      "learning_rate": 1.1325604559086489e-07,
      "loss": 0.0011,
      "step": 14903
    },
    {
      "epoch": 9.535508637236084,
      "grad_norm": 0.14045778684263163,
      "learning_rate": 1.1294527634628682e-07,
      "loss": 0.0009,
      "step": 14904
    },
    {
      "epoch": 9.5361484325016,
      "grad_norm": 0.13327391709380737,
      "learning_rate": 1.126349316342501e-07,
      "loss": 0.0007,
      "step": 14905
    },
    {
      "epoch": 9.536788227767115,
      "grad_norm": 0.1426595126671142,
      "learning_rate": 1.1232501146807961e-07,
      "loss": 0.001,
      "step": 14906
    },
    {
      "epoch": 9.53742802303263,
      "grad_norm": 0.14334585342134779,
      "learning_rate": 1.1201551586108584e-07,
      "loss": 0.0009,
      "step": 14907
    },
    {
      "epoch": 9.538067818298144,
      "grad_norm": 0.08803346264789659,
      "learning_rate": 1.1170644482655701e-07,
      "loss": 0.0007,
      "step": 14908
    },
    {
      "epoch": 9.538707613563659,
      "grad_norm": 0.10106418997875728,
      "learning_rate": 1.1139779837776254e-07,
      "loss": 0.0009,
      "step": 14909
    },
    {
      "epoch": 9.539347408829174,
      "grad_norm": 0.16179753896927848,
      "learning_rate": 1.1108957652795627e-07,
      "loss": 0.0012,
      "step": 14910
    },
    {
      "epoch": 9.53998720409469,
      "grad_norm": 0.14861678140867352,
      "learning_rate": 1.1078177929037315e-07,
      "loss": 0.0011,
      "step": 14911
    },
    {
      "epoch": 9.540626999360205,
      "grad_norm": 0.23466772939912625,
      "learning_rate": 1.1047440667822818e-07,
      "loss": 0.0015,
      "step": 14912
    },
    {
      "epoch": 9.54126679462572,
      "grad_norm": 0.09408584281548649,
      "learning_rate": 1.1016745870471968e-07,
      "loss": 0.0008,
      "step": 14913
    },
    {
      "epoch": 9.541906589891235,
      "grad_norm": 0.15228759764104863,
      "learning_rate": 1.0986093538302933e-07,
      "loss": 0.0012,
      "step": 14914
    },
    {
      "epoch": 9.54254638515675,
      "grad_norm": 0.1861877350921196,
      "learning_rate": 1.0955483672631662e-07,
      "loss": 0.0012,
      "step": 14915
    },
    {
      "epoch": 9.543186180422264,
      "grad_norm": 0.11931533823644357,
      "learning_rate": 1.0924916274772546e-07,
      "loss": 0.0008,
      "step": 14916
    },
    {
      "epoch": 9.54382597568778,
      "grad_norm": 0.1652167559519369,
      "learning_rate": 1.0894391346038091e-07,
      "loss": 0.0011,
      "step": 14917
    },
    {
      "epoch": 9.544465770953295,
      "grad_norm": 0.10696364583363097,
      "learning_rate": 1.0863908887738916e-07,
      "loss": 0.001,
      "step": 14918
    },
    {
      "epoch": 9.54510556621881,
      "grad_norm": 0.1370589060347654,
      "learning_rate": 1.0833468901183974e-07,
      "loss": 0.0009,
      "step": 14919
    },
    {
      "epoch": 9.545745361484325,
      "grad_norm": 0.15649051004650888,
      "learning_rate": 1.0803071387680442e-07,
      "loss": 0.0011,
      "step": 14920
    },
    {
      "epoch": 9.54638515674984,
      "grad_norm": 0.18305429909915374,
      "learning_rate": 1.0772716348533274e-07,
      "loss": 0.0009,
      "step": 14921
    },
    {
      "epoch": 9.547024952015356,
      "grad_norm": 0.14610603941773262,
      "learning_rate": 1.0742403785045874e-07,
      "loss": 0.0009,
      "step": 14922
    },
    {
      "epoch": 9.54766474728087,
      "grad_norm": 0.092092156991815,
      "learning_rate": 1.0712133698519978e-07,
      "loss": 0.0007,
      "step": 14923
    },
    {
      "epoch": 9.548304542546385,
      "grad_norm": 0.07669767828907757,
      "learning_rate": 1.0681906090255323e-07,
      "loss": 0.0007,
      "step": 14924
    },
    {
      "epoch": 9.5489443378119,
      "grad_norm": 0.44834224214645446,
      "learning_rate": 1.0651720961549761e-07,
      "loss": 0.0011,
      "step": 14925
    },
    {
      "epoch": 9.549584133077415,
      "grad_norm": 0.11822306005765389,
      "learning_rate": 1.0621578313699254e-07,
      "loss": 0.0009,
      "step": 14926
    },
    {
      "epoch": 9.55022392834293,
      "grad_norm": 0.0963498022566908,
      "learning_rate": 1.0591478147998324e-07,
      "loss": 0.0007,
      "step": 14927
    },
    {
      "epoch": 9.550863723608446,
      "grad_norm": 0.127063084563014,
      "learning_rate": 1.056142046573927e-07,
      "loss": 0.0008,
      "step": 14928
    },
    {
      "epoch": 9.551503518873961,
      "grad_norm": 0.09512904493769367,
      "learning_rate": 1.0531405268212613e-07,
      "loss": 0.0008,
      "step": 14929
    },
    {
      "epoch": 9.552143314139474,
      "grad_norm": 0.16500312707033923,
      "learning_rate": 1.0501432556707436e-07,
      "loss": 0.0011,
      "step": 14930
    },
    {
      "epoch": 9.55278310940499,
      "grad_norm": 0.10791723159438833,
      "learning_rate": 1.0471502332510597e-07,
      "loss": 0.0009,
      "step": 14931
    },
    {
      "epoch": 9.553422904670505,
      "grad_norm": 0.14635449460372904,
      "learning_rate": 1.044161459690718e-07,
      "loss": 0.001,
      "step": 14932
    },
    {
      "epoch": 9.55406269993602,
      "grad_norm": 0.14000514265858072,
      "learning_rate": 1.0411769351180489e-07,
      "loss": 0.0006,
      "step": 14933
    },
    {
      "epoch": 9.554702495201536,
      "grad_norm": 0.13634598466347073,
      "learning_rate": 1.0381966596612169e-07,
      "loss": 0.0008,
      "step": 14934
    },
    {
      "epoch": 9.55534229046705,
      "grad_norm": 0.09712979743188449,
      "learning_rate": 1.0352206334481751e-07,
      "loss": 0.0007,
      "step": 14935
    },
    {
      "epoch": 9.555982085732566,
      "grad_norm": 0.11337960346149786,
      "learning_rate": 1.0322488566067101e-07,
      "loss": 0.0009,
      "step": 14936
    },
    {
      "epoch": 9.556621880998081,
      "grad_norm": 0.06627553531350322,
      "learning_rate": 1.029281329264431e-07,
      "loss": 0.0007,
      "step": 14937
    },
    {
      "epoch": 9.557261676263595,
      "grad_norm": 0.19686577619902637,
      "learning_rate": 1.026318051548758e-07,
      "loss": 0.0014,
      "step": 14938
    },
    {
      "epoch": 9.55790147152911,
      "grad_norm": 0.13357645338133894,
      "learning_rate": 1.0233590235869229e-07,
      "loss": 0.001,
      "step": 14939
    },
    {
      "epoch": 9.558541266794625,
      "grad_norm": 0.12151486978790317,
      "learning_rate": 1.0204042455060014e-07,
      "loss": 0.001,
      "step": 14940
    },
    {
      "epoch": 9.55918106206014,
      "grad_norm": 0.11473281110814783,
      "learning_rate": 1.0174537174328369e-07,
      "loss": 0.0008,
      "step": 14941
    },
    {
      "epoch": 9.559820857325656,
      "grad_norm": 0.19933024890388096,
      "learning_rate": 1.0145074394941389e-07,
      "loss": 0.0009,
      "step": 14942
    },
    {
      "epoch": 9.560460652591171,
      "grad_norm": 0.16212491162249779,
      "learning_rate": 1.0115654118164064e-07,
      "loss": 0.0012,
      "step": 14943
    },
    {
      "epoch": 9.561100447856687,
      "grad_norm": 0.11275277670863995,
      "learning_rate": 1.0086276345259716e-07,
      "loss": 0.0008,
      "step": 14944
    },
    {
      "epoch": 9.561740243122202,
      "grad_norm": 0.11462290254566836,
      "learning_rate": 1.0056941077489668e-07,
      "loss": 0.0009,
      "step": 14945
    },
    {
      "epoch": 9.562380038387715,
      "grad_norm": 0.09777548513855397,
      "learning_rate": 1.0027648316113692e-07,
      "loss": 0.0008,
      "step": 14946
    },
    {
      "epoch": 9.56301983365323,
      "grad_norm": 0.20503414895953348,
      "learning_rate": 9.998398062389448e-08,
      "loss": 0.0011,
      "step": 14947
    },
    {
      "epoch": 9.563659628918746,
      "grad_norm": 0.14770062730509842,
      "learning_rate": 9.96919031757293e-08,
      "loss": 0.0011,
      "step": 14948
    },
    {
      "epoch": 9.564299424184261,
      "grad_norm": 0.18033612909837451,
      "learning_rate": 9.940025082918137e-08,
      "loss": 0.0009,
      "step": 14949
    },
    {
      "epoch": 9.564939219449776,
      "grad_norm": 0.1326206315994504,
      "learning_rate": 9.910902359677621e-08,
      "loss": 0.0011,
      "step": 14950
    },
    {
      "epoch": 9.565579014715292,
      "grad_norm": 0.10014542425207337,
      "learning_rate": 9.881822149101716e-08,
      "loss": 0.0008,
      "step": 14951
    },
    {
      "epoch": 9.566218809980807,
      "grad_norm": 0.18704355140529624,
      "learning_rate": 9.85278445243898e-08,
      "loss": 0.001,
      "step": 14952
    },
    {
      "epoch": 9.56685860524632,
      "grad_norm": 0.16610731791124916,
      "learning_rate": 9.823789270936413e-08,
      "loss": 0.0011,
      "step": 14953
    },
    {
      "epoch": 9.567498400511836,
      "grad_norm": 0.10284786890997388,
      "learning_rate": 9.79483660583891e-08,
      "loss": 0.0008,
      "step": 14954
    },
    {
      "epoch": 9.568138195777351,
      "grad_norm": 0.0796321455428631,
      "learning_rate": 9.765926458389808e-08,
      "loss": 0.0007,
      "step": 14955
    },
    {
      "epoch": 9.568777991042866,
      "grad_norm": 0.1701187840798069,
      "learning_rate": 9.737058829830226e-08,
      "loss": 0.0012,
      "step": 14956
    },
    {
      "epoch": 9.569417786308382,
      "grad_norm": 0.06967587407788199,
      "learning_rate": 9.708233721399729e-08,
      "loss": 0.0006,
      "step": 14957
    },
    {
      "epoch": 9.570057581573897,
      "grad_norm": 0.10389874933113197,
      "learning_rate": 9.679451134336216e-08,
      "loss": 0.0008,
      "step": 14958
    },
    {
      "epoch": 9.570697376839412,
      "grad_norm": 0.20714214588173882,
      "learning_rate": 9.650711069875363e-08,
      "loss": 0.0012,
      "step": 14959
    },
    {
      "epoch": 9.571337172104926,
      "grad_norm": 0.18362390699710374,
      "learning_rate": 9.622013529251295e-08,
      "loss": 0.0011,
      "step": 14960
    },
    {
      "epoch": 9.571976967370441,
      "grad_norm": 0.10021745456885699,
      "learning_rate": 9.59335851369625e-08,
      "loss": 0.0009,
      "step": 14961
    },
    {
      "epoch": 9.572616762635956,
      "grad_norm": 0.10683500957966195,
      "learning_rate": 9.564746024440574e-08,
      "loss": 0.0008,
      "step": 14962
    },
    {
      "epoch": 9.573256557901471,
      "grad_norm": 0.14347345668986725,
      "learning_rate": 9.536176062712843e-08,
      "loss": 0.0011,
      "step": 14963
    },
    {
      "epoch": 9.573896353166987,
      "grad_norm": 0.1142079219874599,
      "learning_rate": 9.507648629739852e-08,
      "loss": 0.0008,
      "step": 14964
    },
    {
      "epoch": 9.574536148432502,
      "grad_norm": 0.15004451525095586,
      "learning_rate": 9.479163726746509e-08,
      "loss": 0.0012,
      "step": 14965
    },
    {
      "epoch": 9.575175943698017,
      "grad_norm": 0.16640180590481496,
      "learning_rate": 9.450721354955727e-08,
      "loss": 0.0011,
      "step": 14966
    },
    {
      "epoch": 9.57581573896353,
      "grad_norm": 0.14884488520539416,
      "learning_rate": 9.422321515589084e-08,
      "loss": 0.0011,
      "step": 14967
    },
    {
      "epoch": 9.576455534229046,
      "grad_norm": 0.1362146836294164,
      "learning_rate": 9.393964209865713e-08,
      "loss": 0.0011,
      "step": 14968
    },
    {
      "epoch": 9.577095329494561,
      "grad_norm": 0.2758948115158492,
      "learning_rate": 9.365649439003421e-08,
      "loss": 0.0012,
      "step": 14969
    },
    {
      "epoch": 9.577735124760077,
      "grad_norm": 0.13226739144539823,
      "learning_rate": 9.337377204218012e-08,
      "loss": 0.0008,
      "step": 14970
    },
    {
      "epoch": 9.578374920025592,
      "grad_norm": 0.19679912552075413,
      "learning_rate": 9.309147506723293e-08,
      "loss": 0.0012,
      "step": 14971
    },
    {
      "epoch": 9.579014715291107,
      "grad_norm": 0.11485483510042264,
      "learning_rate": 9.280960347731516e-08,
      "loss": 0.0009,
      "step": 14972
    },
    {
      "epoch": 9.579654510556622,
      "grad_norm": 0.20463765554739347,
      "learning_rate": 9.252815728452935e-08,
      "loss": 0.0011,
      "step": 14973
    },
    {
      "epoch": 9.580294305822138,
      "grad_norm": 0.07504560230565072,
      "learning_rate": 9.22471365009603e-08,
      "loss": 0.0007,
      "step": 14974
    },
    {
      "epoch": 9.580934101087651,
      "grad_norm": 0.12956230198109747,
      "learning_rate": 9.196654113867498e-08,
      "loss": 0.0008,
      "step": 14975
    },
    {
      "epoch": 9.581573896353166,
      "grad_norm": 0.12274870574580944,
      "learning_rate": 9.168637120972157e-08,
      "loss": 0.0008,
      "step": 14976
    },
    {
      "epoch": 9.582213691618682,
      "grad_norm": 0.19780375580371506,
      "learning_rate": 9.140662672612932e-08,
      "loss": 0.0013,
      "step": 14977
    },
    {
      "epoch": 9.582853486884197,
      "grad_norm": 0.13704697322286483,
      "learning_rate": 9.112730769991085e-08,
      "loss": 0.0008,
      "step": 14978
    },
    {
      "epoch": 9.583493282149712,
      "grad_norm": 0.1481772134617004,
      "learning_rate": 9.084841414305878e-08,
      "loss": 0.0008,
      "step": 14979
    },
    {
      "epoch": 9.584133077415228,
      "grad_norm": 0.09053795307325392,
      "learning_rate": 9.05699460675491e-08,
      "loss": 0.0007,
      "step": 14980
    },
    {
      "epoch": 9.584772872680743,
      "grad_norm": 0.1327766624686144,
      "learning_rate": 9.029190348533779e-08,
      "loss": 0.001,
      "step": 14981
    },
    {
      "epoch": 9.585412667946258,
      "grad_norm": 0.15827537767452776,
      "learning_rate": 9.00142864083653e-08,
      "loss": 0.0013,
      "step": 14982
    },
    {
      "epoch": 9.586052463211772,
      "grad_norm": 0.08713427531974613,
      "learning_rate": 8.973709484854987e-08,
      "loss": 0.0006,
      "step": 14983
    },
    {
      "epoch": 9.586692258477287,
      "grad_norm": 0.08300269878227841,
      "learning_rate": 8.946032881779532e-08,
      "loss": 0.0007,
      "step": 14984
    },
    {
      "epoch": 9.587332053742802,
      "grad_norm": 0.12877048867564228,
      "learning_rate": 8.918398832798436e-08,
      "loss": 0.001,
      "step": 14985
    },
    {
      "epoch": 9.587971849008317,
      "grad_norm": 0.10432597712658477,
      "learning_rate": 8.890807339098307e-08,
      "loss": 0.0008,
      "step": 14986
    },
    {
      "epoch": 9.588611644273833,
      "grad_norm": 0.13558628404595544,
      "learning_rate": 8.863258401863861e-08,
      "loss": 0.0009,
      "step": 14987
    },
    {
      "epoch": 9.589251439539348,
      "grad_norm": 0.10337309757995963,
      "learning_rate": 8.835752022278043e-08,
      "loss": 0.0007,
      "step": 14988
    },
    {
      "epoch": 9.589891234804863,
      "grad_norm": 0.2798750239242195,
      "learning_rate": 8.808288201521797e-08,
      "loss": 0.0018,
      "step": 14989
    },
    {
      "epoch": 9.590531030070377,
      "grad_norm": 0.14361216036170069,
      "learning_rate": 8.780866940774512e-08,
      "loss": 0.001,
      "step": 14990
    },
    {
      "epoch": 9.591170825335892,
      "grad_norm": 0.08258384142886122,
      "learning_rate": 8.753488241213581e-08,
      "loss": 0.0007,
      "step": 14991
    },
    {
      "epoch": 9.591810620601407,
      "grad_norm": 0.11765836274856674,
      "learning_rate": 8.726152104014619e-08,
      "loss": 0.001,
      "step": 14992
    },
    {
      "epoch": 9.592450415866923,
      "grad_norm": 0.1458494345634921,
      "learning_rate": 8.69885853035124e-08,
      "loss": 0.0011,
      "step": 14993
    },
    {
      "epoch": 9.593090211132438,
      "grad_norm": 0.09547704848277937,
      "learning_rate": 8.67160752139562e-08,
      "loss": 0.0008,
      "step": 14994
    },
    {
      "epoch": 9.593730006397953,
      "grad_norm": 0.1413266180095518,
      "learning_rate": 8.6443990783176e-08,
      "loss": 0.0009,
      "step": 14995
    },
    {
      "epoch": 9.594369801663468,
      "grad_norm": 0.10599700807341116,
      "learning_rate": 8.61723320228569e-08,
      "loss": 0.0008,
      "step": 14996
    },
    {
      "epoch": 9.595009596928982,
      "grad_norm": 0.1160823283603376,
      "learning_rate": 8.590109894466181e-08,
      "loss": 0.0009,
      "step": 14997
    },
    {
      "epoch": 9.595649392194497,
      "grad_norm": 0.17051130851018428,
      "learning_rate": 8.563029156023805e-08,
      "loss": 0.0011,
      "step": 14998
    },
    {
      "epoch": 9.596289187460012,
      "grad_norm": 0.10293995879040285,
      "learning_rate": 8.535990988121301e-08,
      "loss": 0.0009,
      "step": 14999
    },
    {
      "epoch": 9.596928982725528,
      "grad_norm": 0.1672402176461381,
      "learning_rate": 8.508995391919739e-08,
      "loss": 0.0011,
      "step": 15000
    },
    {
      "epoch": 9.597568777991043,
      "grad_norm": 0.08098388772310124,
      "learning_rate": 8.482042368578081e-08,
      "loss": 0.0006,
      "step": 15001
    },
    {
      "epoch": 9.598208573256558,
      "grad_norm": 0.11052166448483382,
      "learning_rate": 8.455131919253846e-08,
      "loss": 0.0009,
      "step": 15002
    },
    {
      "epoch": 9.598848368522074,
      "grad_norm": 0.15476137093804185,
      "learning_rate": 8.42826404510233e-08,
      "loss": 0.001,
      "step": 15003
    },
    {
      "epoch": 9.599488163787587,
      "grad_norm": 0.11089772260649763,
      "learning_rate": 8.40143874727739e-08,
      "loss": 0.0008,
      "step": 15004
    },
    {
      "epoch": 9.600127959053102,
      "grad_norm": 0.1380078726810572,
      "learning_rate": 8.374656026930771e-08,
      "loss": 0.001,
      "step": 15005
    },
    {
      "epoch": 9.600767754318618,
      "grad_norm": 0.09580672507882765,
      "learning_rate": 8.347915885212332e-08,
      "loss": 0.0007,
      "step": 15006
    },
    {
      "epoch": 9.601407549584133,
      "grad_norm": 0.18521554692198297,
      "learning_rate": 8.321218323270485e-08,
      "loss": 0.0012,
      "step": 15007
    },
    {
      "epoch": 9.602047344849648,
      "grad_norm": 0.17729220357779102,
      "learning_rate": 8.294563342251427e-08,
      "loss": 0.0011,
      "step": 15008
    },
    {
      "epoch": 9.602687140115163,
      "grad_norm": 0.08393247483617454,
      "learning_rate": 8.267950943299685e-08,
      "loss": 0.0007,
      "step": 15009
    },
    {
      "epoch": 9.603326935380679,
      "grad_norm": 0.24154287931451432,
      "learning_rate": 8.241381127558012e-08,
      "loss": 0.0012,
      "step": 15010
    },
    {
      "epoch": 9.603966730646194,
      "grad_norm": 0.12104271283328542,
      "learning_rate": 8.214853896167163e-08,
      "loss": 0.0009,
      "step": 15011
    },
    {
      "epoch": 9.604606525911707,
      "grad_norm": 0.15672361638518956,
      "learning_rate": 8.188369250266337e-08,
      "loss": 0.0009,
      "step": 15012
    },
    {
      "epoch": 9.605246321177223,
      "grad_norm": 0.1340261531372345,
      "learning_rate": 8.161927190992513e-08,
      "loss": 0.0007,
      "step": 15013
    },
    {
      "epoch": 9.605886116442738,
      "grad_norm": 0.10557175091455312,
      "learning_rate": 8.13552771948134e-08,
      "loss": 0.0008,
      "step": 15014
    },
    {
      "epoch": 9.606525911708253,
      "grad_norm": 0.12926674155376927,
      "learning_rate": 8.109170836866132e-08,
      "loss": 0.0009,
      "step": 15015
    },
    {
      "epoch": 9.607165706973769,
      "grad_norm": 0.17937842883624258,
      "learning_rate": 8.08285654427865e-08,
      "loss": 0.001,
      "step": 15016
    },
    {
      "epoch": 9.607805502239284,
      "grad_norm": 0.0994608973549316,
      "learning_rate": 8.056584842848881e-08,
      "loss": 0.0008,
      "step": 15017
    },
    {
      "epoch": 9.6084452975048,
      "grad_norm": 0.1120892328387728,
      "learning_rate": 8.03035573370492e-08,
      "loss": 0.0007,
      "step": 15018
    },
    {
      "epoch": 9.609085092770314,
      "grad_norm": 0.1364467837368868,
      "learning_rate": 8.004169217972757e-08,
      "loss": 0.0009,
      "step": 15019
    },
    {
      "epoch": 9.609724888035828,
      "grad_norm": 0.2890094113918229,
      "learning_rate": 7.978025296776937e-08,
      "loss": 0.0014,
      "step": 15020
    },
    {
      "epoch": 9.610364683301343,
      "grad_norm": 0.15070247538546366,
      "learning_rate": 7.951923971240117e-08,
      "loss": 0.0009,
      "step": 15021
    },
    {
      "epoch": 9.611004478566858,
      "grad_norm": 0.15067716872113926,
      "learning_rate": 7.925865242482955e-08,
      "loss": 0.0009,
      "step": 15022
    },
    {
      "epoch": 9.611644273832374,
      "grad_norm": 0.1271543035986673,
      "learning_rate": 7.899849111624447e-08,
      "loss": 0.0009,
      "step": 15023
    },
    {
      "epoch": 9.612284069097889,
      "grad_norm": 0.13240937000968106,
      "learning_rate": 7.873875579781587e-08,
      "loss": 0.0009,
      "step": 15024
    },
    {
      "epoch": 9.612923864363404,
      "grad_norm": 0.15940308283310922,
      "learning_rate": 7.847944648069594e-08,
      "loss": 0.0008,
      "step": 15025
    },
    {
      "epoch": 9.61356365962892,
      "grad_norm": 0.14431389441517647,
      "learning_rate": 7.822056317601911e-08,
      "loss": 0.0009,
      "step": 15026
    },
    {
      "epoch": 9.614203454894433,
      "grad_norm": 0.14870593373770474,
      "learning_rate": 7.796210589490205e-08,
      "loss": 0.001,
      "step": 15027
    },
    {
      "epoch": 9.614843250159948,
      "grad_norm": 0.09415880815877545,
      "learning_rate": 7.770407464844254e-08,
      "loss": 0.0007,
      "step": 15028
    },
    {
      "epoch": 9.615483045425464,
      "grad_norm": 0.12019525295237829,
      "learning_rate": 7.744646944772061e-08,
      "loss": 0.0007,
      "step": 15029
    },
    {
      "epoch": 9.616122840690979,
      "grad_norm": 0.1322837140596051,
      "learning_rate": 7.718929030379518e-08,
      "loss": 0.001,
      "step": 15030
    },
    {
      "epoch": 9.616762635956494,
      "grad_norm": 0.079462659544323,
      "learning_rate": 7.693253722770966e-08,
      "loss": 0.0007,
      "step": 15031
    },
    {
      "epoch": 9.61740243122201,
      "grad_norm": 0.11842667021868336,
      "learning_rate": 7.667621023049077e-08,
      "loss": 0.0008,
      "step": 15032
    },
    {
      "epoch": 9.618042226487525,
      "grad_norm": 0.15828288435194088,
      "learning_rate": 7.642030932314193e-08,
      "loss": 0.0011,
      "step": 15033
    },
    {
      "epoch": 9.618682021753038,
      "grad_norm": 0.1847571722075445,
      "learning_rate": 7.616483451665324e-08,
      "loss": 0.0013,
      "step": 15034
    },
    {
      "epoch": 9.619321817018553,
      "grad_norm": 0.17466143419233546,
      "learning_rate": 7.590978582199371e-08,
      "loss": 0.0015,
      "step": 15035
    },
    {
      "epoch": 9.619961612284069,
      "grad_norm": 0.14576191900589033,
      "learning_rate": 7.565516325011346e-08,
      "loss": 0.001,
      "step": 15036
    },
    {
      "epoch": 9.620601407549584,
      "grad_norm": 0.08291480682849015,
      "learning_rate": 7.540096681194708e-08,
      "loss": 0.0008,
      "step": 15037
    },
    {
      "epoch": 9.6212412028151,
      "grad_norm": 0.11888271007899948,
      "learning_rate": 7.514719651840918e-08,
      "loss": 0.0009,
      "step": 15038
    },
    {
      "epoch": 9.621880998080615,
      "grad_norm": 0.1385000797142773,
      "learning_rate": 7.489385238039437e-08,
      "loss": 0.001,
      "step": 15039
    },
    {
      "epoch": 9.62252079334613,
      "grad_norm": 0.21217746973229284,
      "learning_rate": 7.464093440878284e-08,
      "loss": 0.0011,
      "step": 15040
    },
    {
      "epoch": 9.623160588611643,
      "grad_norm": 0.2143236551225506,
      "learning_rate": 7.438844261443367e-08,
      "loss": 0.0013,
      "step": 15041
    },
    {
      "epoch": 9.623800383877159,
      "grad_norm": 0.2821611261547207,
      "learning_rate": 7.413637700818822e-08,
      "loss": 0.0011,
      "step": 15042
    },
    {
      "epoch": 9.624440179142674,
      "grad_norm": 0.13110991123238047,
      "learning_rate": 7.388473760087001e-08,
      "loss": 0.001,
      "step": 15043
    },
    {
      "epoch": 9.62507997440819,
      "grad_norm": 0.1176125013908614,
      "learning_rate": 7.363352440328486e-08,
      "loss": 0.0009,
      "step": 15044
    },
    {
      "epoch": 9.625719769673704,
      "grad_norm": 0.1885497501770076,
      "learning_rate": 7.338273742621749e-08,
      "loss": 0.001,
      "step": 15045
    },
    {
      "epoch": 9.62635956493922,
      "grad_norm": 0.0825891891216751,
      "learning_rate": 7.313237668043816e-08,
      "loss": 0.0007,
      "step": 15046
    },
    {
      "epoch": 9.626999360204735,
      "grad_norm": 0.16871677850266822,
      "learning_rate": 7.288244217669493e-08,
      "loss": 0.0011,
      "step": 15047
    },
    {
      "epoch": 9.62763915547025,
      "grad_norm": 0.12045498591221325,
      "learning_rate": 7.263293392572146e-08,
      "loss": 0.0008,
      "step": 15048
    },
    {
      "epoch": 9.628278950735764,
      "grad_norm": 0.0931374148704819,
      "learning_rate": 7.238385193823028e-08,
      "loss": 0.0008,
      "step": 15049
    },
    {
      "epoch": 9.628918746001279,
      "grad_norm": 0.1499402072090999,
      "learning_rate": 7.213519622491726e-08,
      "loss": 0.001,
      "step": 15050
    },
    {
      "epoch": 9.629558541266794,
      "grad_norm": 0.14538270007810025,
      "learning_rate": 7.188696679645724e-08,
      "loss": 0.0008,
      "step": 15051
    },
    {
      "epoch": 9.63019833653231,
      "grad_norm": 0.12028624078130223,
      "learning_rate": 7.163916366351165e-08,
      "loss": 0.0009,
      "step": 15052
    },
    {
      "epoch": 9.630838131797825,
      "grad_norm": 0.30375202472863905,
      "learning_rate": 7.139178683671865e-08,
      "loss": 0.001,
      "step": 15053
    },
    {
      "epoch": 9.63147792706334,
      "grad_norm": 0.11384811591970775,
      "learning_rate": 7.114483632669977e-08,
      "loss": 0.0009,
      "step": 15054
    },
    {
      "epoch": 9.632117722328855,
      "grad_norm": 0.13010657403528206,
      "learning_rate": 7.089831214406095e-08,
      "loss": 0.0008,
      "step": 15055
    },
    {
      "epoch": 9.63275751759437,
      "grad_norm": 0.1359934764725575,
      "learning_rate": 7.065221429938485e-08,
      "loss": 0.001,
      "step": 15056
    },
    {
      "epoch": 9.633397312859884,
      "grad_norm": 0.14118251344981453,
      "learning_rate": 7.040654280324077e-08,
      "loss": 0.0009,
      "step": 15057
    },
    {
      "epoch": 9.6340371081254,
      "grad_norm": 0.07339701339396773,
      "learning_rate": 7.016129766617586e-08,
      "loss": 0.0006,
      "step": 15058
    },
    {
      "epoch": 9.634676903390915,
      "grad_norm": 0.13549733120287188,
      "learning_rate": 6.991647889872055e-08,
      "loss": 0.0011,
      "step": 15059
    },
    {
      "epoch": 9.63531669865643,
      "grad_norm": 0.11795331158278415,
      "learning_rate": 6.967208651138758e-08,
      "loss": 0.0008,
      "step": 15060
    },
    {
      "epoch": 9.635956493921945,
      "grad_norm": 0.14600221537499833,
      "learning_rate": 6.942812051467073e-08,
      "loss": 0.001,
      "step": 15061
    },
    {
      "epoch": 9.63659628918746,
      "grad_norm": 0.128398886166326,
      "learning_rate": 6.918458091904612e-08,
      "loss": 0.0009,
      "step": 15062
    },
    {
      "epoch": 9.637236084452976,
      "grad_norm": 0.0877654866762155,
      "learning_rate": 6.894146773496868e-08,
      "loss": 0.0008,
      "step": 15063
    },
    {
      "epoch": 9.63787587971849,
      "grad_norm": 0.14810782706926592,
      "learning_rate": 6.869878097287897e-08,
      "loss": 0.0011,
      "step": 15064
    },
    {
      "epoch": 9.638515674984005,
      "grad_norm": 0.0748490413171626,
      "learning_rate": 6.845652064319752e-08,
      "loss": 0.0006,
      "step": 15065
    },
    {
      "epoch": 9.63915547024952,
      "grad_norm": 0.11987885077542795,
      "learning_rate": 6.821468675632715e-08,
      "loss": 0.0009,
      "step": 15066
    },
    {
      "epoch": 9.639795265515035,
      "grad_norm": 0.11964943427909078,
      "learning_rate": 6.797327932264952e-08,
      "loss": 0.0009,
      "step": 15067
    },
    {
      "epoch": 9.64043506078055,
      "grad_norm": 0.17249872630069302,
      "learning_rate": 6.773229835253192e-08,
      "loss": 0.001,
      "step": 15068
    },
    {
      "epoch": 9.641074856046066,
      "grad_norm": 0.20709626699816291,
      "learning_rate": 6.749174385632162e-08,
      "loss": 0.0011,
      "step": 15069
    },
    {
      "epoch": 9.641714651311581,
      "grad_norm": 0.11964114455135684,
      "learning_rate": 6.725161584434591e-08,
      "loss": 0.0008,
      "step": 15070
    },
    {
      "epoch": 9.642354446577095,
      "grad_norm": 0.11515416026281791,
      "learning_rate": 6.701191432691878e-08,
      "loss": 0.001,
      "step": 15071
    },
    {
      "epoch": 9.64299424184261,
      "grad_norm": 0.2218270165176908,
      "learning_rate": 6.677263931432976e-08,
      "loss": 0.0014,
      "step": 15072
    },
    {
      "epoch": 9.643634037108125,
      "grad_norm": 0.16743702437295807,
      "learning_rate": 6.653379081685284e-08,
      "loss": 0.001,
      "step": 15073
    },
    {
      "epoch": 9.64427383237364,
      "grad_norm": 0.2152120187530932,
      "learning_rate": 6.629536884474652e-08,
      "loss": 0.0013,
      "step": 15074
    },
    {
      "epoch": 9.644913627639156,
      "grad_norm": 0.12844119815024194,
      "learning_rate": 6.605737340824481e-08,
      "loss": 0.0009,
      "step": 15075
    },
    {
      "epoch": 9.64555342290467,
      "grad_norm": 0.15599181465033482,
      "learning_rate": 6.581980451756842e-08,
      "loss": 0.001,
      "step": 15076
    },
    {
      "epoch": 9.646193218170186,
      "grad_norm": 0.18184818333398053,
      "learning_rate": 6.55826621829192e-08,
      "loss": 0.0011,
      "step": 15077
    },
    {
      "epoch": 9.6468330134357,
      "grad_norm": 0.1271414741382751,
      "learning_rate": 6.534594641447789e-08,
      "loss": 0.001,
      "step": 15078
    },
    {
      "epoch": 9.647472808701215,
      "grad_norm": 0.16782936509170512,
      "learning_rate": 6.51096572224097e-08,
      "loss": 0.001,
      "step": 15079
    },
    {
      "epoch": 9.64811260396673,
      "grad_norm": 0.3695517253995549,
      "learning_rate": 6.487379461685871e-08,
      "loss": 0.0016,
      "step": 15080
    },
    {
      "epoch": 9.648752399232245,
      "grad_norm": 0.1509499042873079,
      "learning_rate": 6.463835860795575e-08,
      "loss": 0.001,
      "step": 15081
    },
    {
      "epoch": 9.64939219449776,
      "grad_norm": 0.13186735000520647,
      "learning_rate": 6.440334920580715e-08,
      "loss": 0.001,
      "step": 15082
    },
    {
      "epoch": 9.650031989763276,
      "grad_norm": 0.17934078839836015,
      "learning_rate": 6.416876642050373e-08,
      "loss": 0.0009,
      "step": 15083
    },
    {
      "epoch": 9.650671785028791,
      "grad_norm": 0.1592325209843799,
      "learning_rate": 6.393461026212078e-08,
      "loss": 0.0011,
      "step": 15084
    },
    {
      "epoch": 9.651311580294307,
      "grad_norm": 0.1256943371602615,
      "learning_rate": 6.370088074070912e-08,
      "loss": 0.0008,
      "step": 15085
    },
    {
      "epoch": 9.65195137555982,
      "grad_norm": 0.12954006013272834,
      "learning_rate": 6.346757786630742e-08,
      "loss": 0.001,
      "step": 15086
    },
    {
      "epoch": 9.652591170825335,
      "grad_norm": 0.12903352346115546,
      "learning_rate": 6.323470164893208e-08,
      "loss": 0.0009,
      "step": 15087
    },
    {
      "epoch": 9.65323096609085,
      "grad_norm": 0.07474041320316763,
      "learning_rate": 6.30022520985818e-08,
      "loss": 0.0006,
      "step": 15088
    },
    {
      "epoch": 9.653870761356366,
      "grad_norm": 0.08778083983080333,
      "learning_rate": 6.277022922523856e-08,
      "loss": 0.0007,
      "step": 15089
    },
    {
      "epoch": 9.654510556621881,
      "grad_norm": 0.1910589858980673,
      "learning_rate": 6.253863303886443e-08,
      "loss": 0.0011,
      "step": 15090
    },
    {
      "epoch": 9.655150351887396,
      "grad_norm": 0.053855062486496956,
      "learning_rate": 6.230746354940476e-08,
      "loss": 0.0006,
      "step": 15091
    },
    {
      "epoch": 9.655790147152912,
      "grad_norm": 0.2600152646631757,
      "learning_rate": 6.207672076678495e-08,
      "loss": 0.0014,
      "step": 15092
    },
    {
      "epoch": 9.656429942418427,
      "grad_norm": 0.24047626873692804,
      "learning_rate": 6.184640470091152e-08,
      "loss": 0.0014,
      "step": 15093
    },
    {
      "epoch": 9.65706973768394,
      "grad_norm": 0.11745133308645801,
      "learning_rate": 6.161651536167546e-08,
      "loss": 0.0009,
      "step": 15094
    },
    {
      "epoch": 9.657709532949456,
      "grad_norm": 0.166517961295707,
      "learning_rate": 6.138705275894774e-08,
      "loss": 0.0014,
      "step": 15095
    },
    {
      "epoch": 9.658349328214971,
      "grad_norm": 0.11844658983747602,
      "learning_rate": 6.11580169025805e-08,
      "loss": 0.0009,
      "step": 15096
    },
    {
      "epoch": 9.658989123480486,
      "grad_norm": 0.09888154459738824,
      "learning_rate": 6.092940780240697e-08,
      "loss": 0.0007,
      "step": 15097
    },
    {
      "epoch": 9.659628918746002,
      "grad_norm": 0.15481479656943703,
      "learning_rate": 6.070122546824597e-08,
      "loss": 0.0011,
      "step": 15098
    },
    {
      "epoch": 9.660268714011517,
      "grad_norm": 0.11535824081399605,
      "learning_rate": 6.0473469909893e-08,
      "loss": 0.0008,
      "step": 15099
    },
    {
      "epoch": 9.660908509277032,
      "grad_norm": 0.1496199747266684,
      "learning_rate": 6.02461411371269e-08,
      "loss": 0.001,
      "step": 15100
    },
    {
      "epoch": 9.661548304542546,
      "grad_norm": 0.156411080138729,
      "learning_rate": 6.001923915971208e-08,
      "loss": 0.0012,
      "step": 15101
    },
    {
      "epoch": 9.662188099808061,
      "grad_norm": 0.0864791912858373,
      "learning_rate": 5.979276398738854e-08,
      "loss": 0.0007,
      "step": 15102
    },
    {
      "epoch": 9.662827895073576,
      "grad_norm": 0.10297551883413515,
      "learning_rate": 5.95667156298807e-08,
      "loss": 0.0008,
      "step": 15103
    },
    {
      "epoch": 9.663467690339091,
      "grad_norm": 0.09079595301236712,
      "learning_rate": 5.9341094096896366e-08,
      "loss": 0.0007,
      "step": 15104
    },
    {
      "epoch": 9.664107485604607,
      "grad_norm": 0.14227450357423777,
      "learning_rate": 5.9115899398122236e-08,
      "loss": 0.001,
      "step": 15105
    },
    {
      "epoch": 9.664747280870122,
      "grad_norm": 0.16664748706921134,
      "learning_rate": 5.889113154322834e-08,
      "loss": 0.001,
      "step": 15106
    },
    {
      "epoch": 9.665387076135637,
      "grad_norm": 0.11187893503062332,
      "learning_rate": 5.866679054186364e-08,
      "loss": 0.0008,
      "step": 15107
    },
    {
      "epoch": 9.66602687140115,
      "grad_norm": 0.06181173534540797,
      "learning_rate": 5.844287640366486e-08,
      "loss": 0.0006,
      "step": 15108
    },
    {
      "epoch": 9.666666666666666,
      "grad_norm": 0.18276028217494744,
      "learning_rate": 5.821938913824321e-08,
      "loss": 0.0009,
      "step": 15109
    },
    {
      "epoch": 9.667306461932181,
      "grad_norm": 0.1728231601975618,
      "learning_rate": 5.7996328755195455e-08,
      "loss": 0.001,
      "step": 15110
    },
    {
      "epoch": 9.667946257197697,
      "grad_norm": 0.1644918259891956,
      "learning_rate": 5.77736952641017e-08,
      "loss": 0.0011,
      "step": 15111
    },
    {
      "epoch": 9.668586052463212,
      "grad_norm": 0.22298421860321518,
      "learning_rate": 5.755148867451876e-08,
      "loss": 0.0018,
      "step": 15112
    },
    {
      "epoch": 9.669225847728727,
      "grad_norm": 0.08897459020616189,
      "learning_rate": 5.7329708995987884e-08,
      "loss": 0.0008,
      "step": 15113
    },
    {
      "epoch": 9.669865642994242,
      "grad_norm": 0.10527182798253899,
      "learning_rate": 5.710835623803257e-08,
      "loss": 0.0008,
      "step": 15114
    },
    {
      "epoch": 9.670505438259756,
      "grad_norm": 0.09077478085721487,
      "learning_rate": 5.688743041015854e-08,
      "loss": 0.0007,
      "step": 15115
    },
    {
      "epoch": 9.671145233525271,
      "grad_norm": 0.14378308807913434,
      "learning_rate": 5.666693152185043e-08,
      "loss": 0.001,
      "step": 15116
    },
    {
      "epoch": 9.671785028790786,
      "grad_norm": 0.14966472733343336,
      "learning_rate": 5.644685958257623e-08,
      "loss": 0.0011,
      "step": 15117
    },
    {
      "epoch": 9.672424824056302,
      "grad_norm": 0.12689999919268954,
      "learning_rate": 5.622721460178504e-08,
      "loss": 0.0009,
      "step": 15118
    },
    {
      "epoch": 9.673064619321817,
      "grad_norm": 0.1641884978434786,
      "learning_rate": 5.600799658890932e-08,
      "loss": 0.0012,
      "step": 15119
    },
    {
      "epoch": 9.673704414587332,
      "grad_norm": 0.09245377550127738,
      "learning_rate": 5.5789205553361536e-08,
      "loss": 0.0007,
      "step": 15120
    },
    {
      "epoch": 9.674344209852848,
      "grad_norm": 0.1132345890093602,
      "learning_rate": 5.557084150453529e-08,
      "loss": 0.001,
      "step": 15121
    },
    {
      "epoch": 9.674984005118363,
      "grad_norm": 0.17573736349046218,
      "learning_rate": 5.535290445180752e-08,
      "loss": 0.0011,
      "step": 15122
    },
    {
      "epoch": 9.675623800383876,
      "grad_norm": 0.0850611651464922,
      "learning_rate": 5.51353944045363e-08,
      "loss": 0.0007,
      "step": 15123
    },
    {
      "epoch": 9.676263595649392,
      "grad_norm": 0.14349421158775105,
      "learning_rate": 5.491831137206083e-08,
      "loss": 0.001,
      "step": 15124
    },
    {
      "epoch": 9.676903390914907,
      "grad_norm": 0.22077911337286463,
      "learning_rate": 5.4701655363701424e-08,
      "loss": 0.0012,
      "step": 15125
    },
    {
      "epoch": 9.677543186180422,
      "grad_norm": 0.12872405803439088,
      "learning_rate": 5.448542638876175e-08,
      "loss": 0.0008,
      "step": 15126
    },
    {
      "epoch": 9.678182981445937,
      "grad_norm": 0.20660200314950075,
      "learning_rate": 5.426962445652661e-08,
      "loss": 0.0012,
      "step": 15127
    },
    {
      "epoch": 9.678822776711453,
      "grad_norm": 0.08463690728827825,
      "learning_rate": 5.405424957626193e-08,
      "loss": 0.0007,
      "step": 15128
    },
    {
      "epoch": 9.679462571976968,
      "grad_norm": 0.0973502882840343,
      "learning_rate": 5.383930175721475e-08,
      "loss": 0.0008,
      "step": 15129
    },
    {
      "epoch": 9.680102367242483,
      "grad_norm": 0.0987211101136341,
      "learning_rate": 5.3624781008614343e-08,
      "loss": 0.0007,
      "step": 15130
    },
    {
      "epoch": 9.680742162507997,
      "grad_norm": 0.13931621121921228,
      "learning_rate": 5.341068733967447e-08,
      "loss": 0.0011,
      "step": 15131
    },
    {
      "epoch": 9.681381957773512,
      "grad_norm": 0.15756671317910348,
      "learning_rate": 5.319702075958444e-08,
      "loss": 0.0009,
      "step": 15132
    },
    {
      "epoch": 9.682021753039027,
      "grad_norm": 0.15322792420703465,
      "learning_rate": 5.298378127752135e-08,
      "loss": 0.001,
      "step": 15133
    },
    {
      "epoch": 9.682661548304543,
      "grad_norm": 0.09194704078404013,
      "learning_rate": 5.277096890264011e-08,
      "loss": 0.0007,
      "step": 15134
    },
    {
      "epoch": 9.683301343570058,
      "grad_norm": 0.1340702006224883,
      "learning_rate": 5.255858364407784e-08,
      "loss": 0.001,
      "step": 15135
    },
    {
      "epoch": 9.683941138835573,
      "grad_norm": 0.22574809105508395,
      "learning_rate": 5.234662551095615e-08,
      "loss": 0.0009,
      "step": 15136
    },
    {
      "epoch": 9.684580934101088,
      "grad_norm": 0.11567657544152196,
      "learning_rate": 5.213509451237331e-08,
      "loss": 0.0009,
      "step": 15137
    },
    {
      "epoch": 9.685220729366602,
      "grad_norm": 0.20969819924385327,
      "learning_rate": 5.1923990657415377e-08,
      "loss": 0.0014,
      "step": 15138
    },
    {
      "epoch": 9.685860524632117,
      "grad_norm": 0.15566480105353334,
      "learning_rate": 5.171331395514401e-08,
      "loss": 0.001,
      "step": 15139
    },
    {
      "epoch": 9.686500319897632,
      "grad_norm": 0.08258366551035427,
      "learning_rate": 5.15030644146064e-08,
      "loss": 0.0007,
      "step": 15140
    },
    {
      "epoch": 9.687140115163148,
      "grad_norm": 0.13656796253496062,
      "learning_rate": 5.129324204483088e-08,
      "loss": 0.0008,
      "step": 15141
    },
    {
      "epoch": 9.687779910428663,
      "grad_norm": 0.12008873981544281,
      "learning_rate": 5.1083846854825816e-08,
      "loss": 0.0008,
      "step": 15142
    },
    {
      "epoch": 9.688419705694178,
      "grad_norm": 0.10078847859905188,
      "learning_rate": 5.0874878853581774e-08,
      "loss": 0.0008,
      "step": 15143
    },
    {
      "epoch": 9.689059500959694,
      "grad_norm": 0.13558970119483918,
      "learning_rate": 5.0666338050073816e-08,
      "loss": 0.0008,
      "step": 15144
    },
    {
      "epoch": 9.689699296225207,
      "grad_norm": 0.1660553263976175,
      "learning_rate": 5.0458224453253656e-08,
      "loss": 0.0011,
      "step": 15145
    },
    {
      "epoch": 9.690339091490722,
      "grad_norm": 0.1320920626794633,
      "learning_rate": 5.02505380720586e-08,
      "loss": 0.0009,
      "step": 15146
    },
    {
      "epoch": 9.690978886756238,
      "grad_norm": 0.0742143522279387,
      "learning_rate": 5.004327891540706e-08,
      "loss": 0.0007,
      "step": 15147
    },
    {
      "epoch": 9.691618682021753,
      "grad_norm": 0.09872010467771077,
      "learning_rate": 4.983644699219636e-08,
      "loss": 0.0008,
      "step": 15148
    },
    {
      "epoch": 9.692258477287268,
      "grad_norm": 0.078534891838433,
      "learning_rate": 4.963004231130941e-08,
      "loss": 0.0006,
      "step": 15149
    },
    {
      "epoch": 9.692898272552783,
      "grad_norm": 0.1361297466328345,
      "learning_rate": 4.9424064881608005e-08,
      "loss": 0.0008,
      "step": 15150
    },
    {
      "epoch": 9.693538067818299,
      "grad_norm": 0.09413334873927294,
      "learning_rate": 4.921851471193617e-08,
      "loss": 0.0007,
      "step": 15151
    },
    {
      "epoch": 9.694177863083812,
      "grad_norm": 0.08154708374170434,
      "learning_rate": 4.901339181112019e-08,
      "loss": 0.0007,
      "step": 15152
    },
    {
      "epoch": 9.694817658349328,
      "grad_norm": 0.19532830001441398,
      "learning_rate": 4.880869618796746e-08,
      "loss": 0.0012,
      "step": 15153
    },
    {
      "epoch": 9.695457453614843,
      "grad_norm": 0.12448711987602558,
      "learning_rate": 4.860442785126873e-08,
      "loss": 0.001,
      "step": 15154
    },
    {
      "epoch": 9.696097248880358,
      "grad_norm": 0.26602415804353063,
      "learning_rate": 4.840058680979365e-08,
      "loss": 0.0012,
      "step": 15155
    },
    {
      "epoch": 9.696737044145873,
      "grad_norm": 0.12049537173877647,
      "learning_rate": 4.81971730722941e-08,
      "loss": 0.0007,
      "step": 15156
    },
    {
      "epoch": 9.697376839411389,
      "grad_norm": 0.10190461222535316,
      "learning_rate": 4.799418664750532e-08,
      "loss": 0.0008,
      "step": 15157
    },
    {
      "epoch": 9.698016634676904,
      "grad_norm": 0.21077886332880333,
      "learning_rate": 4.7791627544142574e-08,
      "loss": 0.0011,
      "step": 15158
    },
    {
      "epoch": 9.69865642994242,
      "grad_norm": 0.09936514921407472,
      "learning_rate": 4.758949577090444e-08,
      "loss": 0.0009,
      "step": 15159
    },
    {
      "epoch": 9.699296225207933,
      "grad_norm": 0.1907539282458532,
      "learning_rate": 4.738779133646842e-08,
      "loss": 0.0011,
      "step": 15160
    },
    {
      "epoch": 9.699936020473448,
      "grad_norm": 0.19089127033583797,
      "learning_rate": 4.7186514249496494e-08,
      "loss": 0.0015,
      "step": 15161
    },
    {
      "epoch": 9.700575815738963,
      "grad_norm": 0.10074653997966,
      "learning_rate": 4.6985664518631736e-08,
      "loss": 0.0008,
      "step": 15162
    },
    {
      "epoch": 9.701215611004478,
      "grad_norm": 0.08630848357762014,
      "learning_rate": 4.678524215249725e-08,
      "loss": 0.0008,
      "step": 15163
    },
    {
      "epoch": 9.701855406269994,
      "grad_norm": 0.15154924241041498,
      "learning_rate": 4.658524715969837e-08,
      "loss": 0.0009,
      "step": 15164
    },
    {
      "epoch": 9.702495201535509,
      "grad_norm": 0.16211763098671944,
      "learning_rate": 4.638567954882378e-08,
      "loss": 0.0011,
      "step": 15165
    },
    {
      "epoch": 9.703134996801024,
      "grad_norm": 0.12689592417460724,
      "learning_rate": 4.618653932844219e-08,
      "loss": 0.0007,
      "step": 15166
    },
    {
      "epoch": 9.70377479206654,
      "grad_norm": 0.17230234917040158,
      "learning_rate": 4.598782650710343e-08,
      "loss": 0.0012,
      "step": 15167
    },
    {
      "epoch": 9.704414587332053,
      "grad_norm": 0.14183276417193816,
      "learning_rate": 4.578954109334177e-08,
      "loss": 0.001,
      "step": 15168
    },
    {
      "epoch": 9.705054382597568,
      "grad_norm": 0.09167542229124095,
      "learning_rate": 4.5591683095669305e-08,
      "loss": 0.0007,
      "step": 15169
    },
    {
      "epoch": 9.705694177863084,
      "grad_norm": 0.18028101913522554,
      "learning_rate": 4.539425252258256e-08,
      "loss": 0.0012,
      "step": 15170
    },
    {
      "epoch": 9.706333973128599,
      "grad_norm": 0.12714587587206289,
      "learning_rate": 4.51972493825592e-08,
      "loss": 0.0009,
      "step": 15171
    },
    {
      "epoch": 9.706973768394114,
      "grad_norm": 0.11123784036498616,
      "learning_rate": 4.500067368405692e-08,
      "loss": 0.0008,
      "step": 15172
    },
    {
      "epoch": 9.70761356365963,
      "grad_norm": 0.11569405027607317,
      "learning_rate": 4.480452543551783e-08,
      "loss": 0.0009,
      "step": 15173
    },
    {
      "epoch": 9.708253358925145,
      "grad_norm": 0.11567258993367488,
      "learning_rate": 4.460880464536299e-08,
      "loss": 0.0009,
      "step": 15174
    },
    {
      "epoch": 9.708893154190658,
      "grad_norm": 0.1327128236285377,
      "learning_rate": 4.4413511321996785e-08,
      "loss": 0.0009,
      "step": 15175
    },
    {
      "epoch": 9.709532949456174,
      "grad_norm": 0.15193317292768005,
      "learning_rate": 4.4218645473804724e-08,
      "loss": 0.0011,
      "step": 15176
    },
    {
      "epoch": 9.710172744721689,
      "grad_norm": 0.1340963915314597,
      "learning_rate": 4.402420710915456e-08,
      "loss": 0.0008,
      "step": 15177
    },
    {
      "epoch": 9.710812539987204,
      "grad_norm": 0.13645404701134423,
      "learning_rate": 4.3830196236394063e-08,
      "loss": 0.0009,
      "step": 15178
    },
    {
      "epoch": 9.71145233525272,
      "grad_norm": 0.11645649708771856,
      "learning_rate": 4.3636612863854344e-08,
      "loss": 0.0008,
      "step": 15179
    },
    {
      "epoch": 9.712092130518235,
      "grad_norm": 0.11952226919309251,
      "learning_rate": 4.3443456999847644e-08,
      "loss": 0.0009,
      "step": 15180
    },
    {
      "epoch": 9.71273192578375,
      "grad_norm": 0.19384296909391827,
      "learning_rate": 4.325072865266844e-08,
      "loss": 0.001,
      "step": 15181
    },
    {
      "epoch": 9.713371721049263,
      "grad_norm": 0.37534302285289733,
      "learning_rate": 4.3058427830590113e-08,
      "loss": 0.0011,
      "step": 15182
    },
    {
      "epoch": 9.714011516314779,
      "grad_norm": 0.11332599503206096,
      "learning_rate": 4.2866554541871606e-08,
      "loss": 0.0009,
      "step": 15183
    },
    {
      "epoch": 9.714651311580294,
      "grad_norm": 0.1341843928612155,
      "learning_rate": 4.267510879474968e-08,
      "loss": 0.0009,
      "step": 15184
    },
    {
      "epoch": 9.71529110684581,
      "grad_norm": 0.14282061501150384,
      "learning_rate": 4.2484090597446623e-08,
      "loss": 0.0009,
      "step": 15185
    },
    {
      "epoch": 9.715930902111324,
      "grad_norm": 0.16623253129761623,
      "learning_rate": 4.2293499958163674e-08,
      "loss": 0.0011,
      "step": 15186
    },
    {
      "epoch": 9.71657069737684,
      "grad_norm": 0.12602254491213216,
      "learning_rate": 4.210333688508539e-08,
      "loss": 0.0009,
      "step": 15187
    },
    {
      "epoch": 9.717210492642355,
      "grad_norm": 0.1273172654942111,
      "learning_rate": 4.191360138637523e-08,
      "loss": 0.0009,
      "step": 15188
    },
    {
      "epoch": 9.71785028790787,
      "grad_norm": 0.1118897939203142,
      "learning_rate": 4.172429347018114e-08,
      "loss": 0.0009,
      "step": 15189
    },
    {
      "epoch": 9.718490083173384,
      "grad_norm": 0.20279665070318798,
      "learning_rate": 4.153541314463105e-08,
      "loss": 0.0008,
      "step": 15190
    },
    {
      "epoch": 9.719129878438899,
      "grad_norm": 0.126495519977393,
      "learning_rate": 4.1346960417835146e-08,
      "loss": 0.0009,
      "step": 15191
    },
    {
      "epoch": 9.719769673704414,
      "grad_norm": 0.06568576108804128,
      "learning_rate": 4.115893529788695e-08,
      "loss": 0.0006,
      "step": 15192
    },
    {
      "epoch": 9.72040946896993,
      "grad_norm": 0.18168696318333868,
      "learning_rate": 4.097133779285778e-08,
      "loss": 0.001,
      "step": 15193
    },
    {
      "epoch": 9.721049264235445,
      "grad_norm": 0.09995582502533133,
      "learning_rate": 4.078416791080342e-08,
      "loss": 0.0008,
      "step": 15194
    },
    {
      "epoch": 9.72168905950096,
      "grad_norm": 0.14408328256254269,
      "learning_rate": 4.059742565976188e-08,
      "loss": 0.0009,
      "step": 15195
    },
    {
      "epoch": 9.722328854766475,
      "grad_norm": 0.112894181383743,
      "learning_rate": 4.041111104775008e-08,
      "loss": 0.0007,
      "step": 15196
    },
    {
      "epoch": 9.722968650031989,
      "grad_norm": 0.12825869537341975,
      "learning_rate": 4.02252240827683e-08,
      "loss": 0.001,
      "step": 15197
    },
    {
      "epoch": 9.723608445297504,
      "grad_norm": 0.10705279415539079,
      "learning_rate": 4.003976477279792e-08,
      "loss": 0.0009,
      "step": 15198
    },
    {
      "epoch": 9.72424824056302,
      "grad_norm": 0.07215384134148704,
      "learning_rate": 3.985473312580368e-08,
      "loss": 0.0006,
      "step": 15199
    },
    {
      "epoch": 9.724888035828535,
      "grad_norm": 0.12521357685427079,
      "learning_rate": 3.9670129149728123e-08,
      "loss": 0.0009,
      "step": 15200
    },
    {
      "epoch": 9.72552783109405,
      "grad_norm": 0.18925282328311374,
      "learning_rate": 3.948595285250045e-08,
      "loss": 0.001,
      "step": 15201
    },
    {
      "epoch": 9.726167626359565,
      "grad_norm": 0.10970015438764101,
      "learning_rate": 3.930220424202658e-08,
      "loss": 0.0007,
      "step": 15202
    },
    {
      "epoch": 9.72680742162508,
      "grad_norm": 0.1361864530273169,
      "learning_rate": 3.9118883326197953e-08,
      "loss": 0.001,
      "step": 15203
    },
    {
      "epoch": 9.727447216890596,
      "grad_norm": 0.10318800038887316,
      "learning_rate": 3.8935990112884957e-08,
      "loss": 0.0007,
      "step": 15204
    },
    {
      "epoch": 9.72808701215611,
      "grad_norm": 0.15787113908021394,
      "learning_rate": 3.8753524609942415e-08,
      "loss": 0.001,
      "step": 15205
    },
    {
      "epoch": 9.728726807421625,
      "grad_norm": 0.07865391358952391,
      "learning_rate": 3.8571486825202954e-08,
      "loss": 0.0007,
      "step": 15206
    },
    {
      "epoch": 9.72936660268714,
      "grad_norm": 0.1479450992501016,
      "learning_rate": 3.838987676648365e-08,
      "loss": 0.001,
      "step": 15207
    },
    {
      "epoch": 9.730006397952655,
      "grad_norm": 0.14476688695665213,
      "learning_rate": 3.8208694441582704e-08,
      "loss": 0.001,
      "step": 15208
    },
    {
      "epoch": 9.73064619321817,
      "grad_norm": 0.09573163172186012,
      "learning_rate": 3.802793985828057e-08,
      "loss": 0.0008,
      "step": 15209
    },
    {
      "epoch": 9.731285988483686,
      "grad_norm": 0.1880687861577502,
      "learning_rate": 3.7847613024336595e-08,
      "loss": 0.0012,
      "step": 15210
    },
    {
      "epoch": 9.731925783749201,
      "grad_norm": 0.11732267266145469,
      "learning_rate": 3.7667713947494577e-08,
      "loss": 0.0008,
      "step": 15211
    },
    {
      "epoch": 9.732565579014715,
      "grad_norm": 0.195633050820399,
      "learning_rate": 3.748824263548057e-08,
      "loss": 0.0014,
      "step": 15212
    },
    {
      "epoch": 9.73320537428023,
      "grad_norm": 0.1181597620619954,
      "learning_rate": 3.7309199095999505e-08,
      "loss": 0.0008,
      "step": 15213
    },
    {
      "epoch": 9.733845169545745,
      "grad_norm": 0.2239885594326139,
      "learning_rate": 3.7130583336738576e-08,
      "loss": 0.0012,
      "step": 15214
    },
    {
      "epoch": 9.73448496481126,
      "grad_norm": 0.108586415217814,
      "learning_rate": 3.695239536536721e-08,
      "loss": 0.0009,
      "step": 15215
    },
    {
      "epoch": 9.735124760076776,
      "grad_norm": 0.08328243992363973,
      "learning_rate": 3.6774635189538163e-08,
      "loss": 0.0008,
      "step": 15216
    },
    {
      "epoch": 9.735764555342291,
      "grad_norm": 0.17901506744467638,
      "learning_rate": 3.659730281688312e-08,
      "loss": 0.0014,
      "step": 15217
    },
    {
      "epoch": 9.736404350607806,
      "grad_norm": 0.12191674320093947,
      "learning_rate": 3.642039825501598e-08,
      "loss": 0.0009,
      "step": 15218
    },
    {
      "epoch": 9.73704414587332,
      "grad_norm": 0.12045668143747497,
      "learning_rate": 3.6243921511532886e-08,
      "loss": 0.0009,
      "step": 15219
    },
    {
      "epoch": 9.737683941138835,
      "grad_norm": 0.11238918772690054,
      "learning_rate": 3.606787259401223e-08,
      "loss": 0.0008,
      "step": 15220
    },
    {
      "epoch": 9.73832373640435,
      "grad_norm": 0.15690006608574117,
      "learning_rate": 3.589225151001241e-08,
      "loss": 0.001,
      "step": 15221
    },
    {
      "epoch": 9.738963531669866,
      "grad_norm": 0.09424024460385451,
      "learning_rate": 3.5717058267074054e-08,
      "loss": 0.0007,
      "step": 15222
    },
    {
      "epoch": 9.73960332693538,
      "grad_norm": 0.06813689535891376,
      "learning_rate": 3.5542292872720044e-08,
      "loss": 0.0006,
      "step": 15223
    },
    {
      "epoch": 9.740243122200896,
      "grad_norm": 0.15445943147776645,
      "learning_rate": 3.536795533445547e-08,
      "loss": 0.0008,
      "step": 15224
    },
    {
      "epoch": 9.740882917466411,
      "grad_norm": 0.15321765731778467,
      "learning_rate": 3.519404565976437e-08,
      "loss": 0.001,
      "step": 15225
    },
    {
      "epoch": 9.741522712731927,
      "grad_norm": 0.11190074239402922,
      "learning_rate": 3.502056385611519e-08,
      "loss": 0.0008,
      "step": 15226
    },
    {
      "epoch": 9.74216250799744,
      "grad_norm": 0.202432712237613,
      "learning_rate": 3.4847509930957535e-08,
      "loss": 0.0008,
      "step": 15227
    },
    {
      "epoch": 9.742802303262955,
      "grad_norm": 0.11795579226571953,
      "learning_rate": 3.4674883891719914e-08,
      "loss": 0.0008,
      "step": 15228
    },
    {
      "epoch": 9.74344209852847,
      "grad_norm": 0.13519540484165857,
      "learning_rate": 3.450268574581639e-08,
      "loss": 0.001,
      "step": 15229
    },
    {
      "epoch": 9.744081893793986,
      "grad_norm": 0.11297823407804893,
      "learning_rate": 3.433091550064105e-08,
      "loss": 0.0009,
      "step": 15230
    },
    {
      "epoch": 9.744721689059501,
      "grad_norm": 0.13953415704436656,
      "learning_rate": 3.4159573163567994e-08,
      "loss": 0.001,
      "step": 15231
    },
    {
      "epoch": 9.745361484325016,
      "grad_norm": 0.13950005670800178,
      "learning_rate": 3.398865874195467e-08,
      "loss": 0.0009,
      "step": 15232
    },
    {
      "epoch": 9.746001279590532,
      "grad_norm": 0.14828950234332997,
      "learning_rate": 3.381817224314077e-08,
      "loss": 0.001,
      "step": 15233
    },
    {
      "epoch": 9.746641074856045,
      "grad_norm": 0.08170621686855942,
      "learning_rate": 3.364811367444487e-08,
      "loss": 0.0007,
      "step": 15234
    },
    {
      "epoch": 9.74728087012156,
      "grad_norm": 0.1772126855555272,
      "learning_rate": 3.347848304317114e-08,
      "loss": 0.0012,
      "step": 15235
    },
    {
      "epoch": 9.747920665387076,
      "grad_norm": 0.1290558152427802,
      "learning_rate": 3.330928035660264e-08,
      "loss": 0.0008,
      "step": 15236
    },
    {
      "epoch": 9.748560460652591,
      "grad_norm": 0.08512273196060476,
      "learning_rate": 3.314050562200355e-08,
      "loss": 0.0007,
      "step": 15237
    },
    {
      "epoch": 9.749200255918106,
      "grad_norm": 0.13455177396931295,
      "learning_rate": 3.297215884662142e-08,
      "loss": 0.0008,
      "step": 15238
    },
    {
      "epoch": 9.749840051183622,
      "grad_norm": 0.08239084258874771,
      "learning_rate": 3.280424003768601e-08,
      "loss": 0.0007,
      "step": 15239
    },
    {
      "epoch": 9.750479846449137,
      "grad_norm": 0.10222663464978501,
      "learning_rate": 3.2636749202404894e-08,
      "loss": 0.0008,
      "step": 15240
    },
    {
      "epoch": 9.751119641714652,
      "grad_norm": 0.1341099734629669,
      "learning_rate": 3.2469686347971205e-08,
      "loss": 0.0008,
      "step": 15241
    },
    {
      "epoch": 9.751759436980166,
      "grad_norm": 0.12374275431669078,
      "learning_rate": 3.230305148155921e-08,
      "loss": 0.0008,
      "step": 15242
    },
    {
      "epoch": 9.752399232245681,
      "grad_norm": 0.09571377409300909,
      "learning_rate": 3.213684461032318e-08,
      "loss": 0.0007,
      "step": 15243
    },
    {
      "epoch": 9.753039027511196,
      "grad_norm": 0.12334122359520516,
      "learning_rate": 3.197106574139852e-08,
      "loss": 0.001,
      "step": 15244
    },
    {
      "epoch": 9.753678822776711,
      "grad_norm": 0.11835393493891794,
      "learning_rate": 3.180571488190509e-08,
      "loss": 0.0008,
      "step": 15245
    },
    {
      "epoch": 9.754318618042227,
      "grad_norm": 0.09993729925832877,
      "learning_rate": 3.164079203894277e-08,
      "loss": 0.0008,
      "step": 15246
    },
    {
      "epoch": 9.754958413307742,
      "grad_norm": 0.09725449577621091,
      "learning_rate": 3.147629721959145e-08,
      "loss": 0.0008,
      "step": 15247
    },
    {
      "epoch": 9.755598208573257,
      "grad_norm": 0.19662620763215496,
      "learning_rate": 3.131223043091658e-08,
      "loss": 0.0012,
      "step": 15248
    },
    {
      "epoch": 9.75623800383877,
      "grad_norm": 0.15020187988555503,
      "learning_rate": 3.114859167996032e-08,
      "loss": 0.001,
      "step": 15249
    },
    {
      "epoch": 9.756877799104286,
      "grad_norm": 0.1532527741176727,
      "learning_rate": 3.0985380973751476e-08,
      "loss": 0.0011,
      "step": 15250
    },
    {
      "epoch": 9.757517594369801,
      "grad_norm": 0.1847883237481503,
      "learning_rate": 3.082259831929557e-08,
      "loss": 0.001,
      "step": 15251
    },
    {
      "epoch": 9.758157389635317,
      "grad_norm": 0.12049161710938215,
      "learning_rate": 3.066024372358478e-08,
      "loss": 0.0008,
      "step": 15252
    },
    {
      "epoch": 9.758797184900832,
      "grad_norm": 0.13519276996028642,
      "learning_rate": 3.0498317193587975e-08,
      "loss": 0.001,
      "step": 15253
    },
    {
      "epoch": 9.759436980166347,
      "grad_norm": 0.14371955995335048,
      "learning_rate": 3.03368187362596e-08,
      "loss": 0.001,
      "step": 15254
    },
    {
      "epoch": 9.760076775431862,
      "grad_norm": 0.15826912039726154,
      "learning_rate": 3.0175748358534095e-08,
      "loss": 0.001,
      "step": 15255
    },
    {
      "epoch": 9.760716570697376,
      "grad_norm": 0.10453478984505007,
      "learning_rate": 3.001510606732594e-08,
      "loss": 0.0007,
      "step": 15256
    },
    {
      "epoch": 9.761356365962891,
      "grad_norm": 0.18949396898019885,
      "learning_rate": 2.985489186953405e-08,
      "loss": 0.0011,
      "step": 15257
    },
    {
      "epoch": 9.761996161228407,
      "grad_norm": 0.0787003764468962,
      "learning_rate": 2.969510577203849e-08,
      "loss": 0.0007,
      "step": 15258
    },
    {
      "epoch": 9.762635956493922,
      "grad_norm": 0.060877107965660544,
      "learning_rate": 2.9535747781698208e-08,
      "loss": 0.0006,
      "step": 15259
    },
    {
      "epoch": 9.763275751759437,
      "grad_norm": 0.11181026851754824,
      "learning_rate": 2.9376817905357734e-08,
      "loss": 0.0008,
      "step": 15260
    },
    {
      "epoch": 9.763915547024952,
      "grad_norm": 0.08677077520643398,
      "learning_rate": 2.9218316149840498e-08,
      "loss": 0.0007,
      "step": 15261
    },
    {
      "epoch": 9.764555342290468,
      "grad_norm": 0.12218064860079,
      "learning_rate": 2.9060242521951054e-08,
      "loss": 0.0009,
      "step": 15262
    },
    {
      "epoch": 9.765195137555983,
      "grad_norm": 0.3051591993732467,
      "learning_rate": 2.8902597028478417e-08,
      "loss": 0.0019,
      "step": 15263
    },
    {
      "epoch": 9.765834932821496,
      "grad_norm": 0.22567540501781785,
      "learning_rate": 2.8745379676191622e-08,
      "loss": 0.0011,
      "step": 15264
    },
    {
      "epoch": 9.766474728087012,
      "grad_norm": 0.14151581179110437,
      "learning_rate": 2.8588590471839707e-08,
      "loss": 0.0008,
      "step": 15265
    },
    {
      "epoch": 9.767114523352527,
      "grad_norm": 0.36375471626319883,
      "learning_rate": 2.8432229422156178e-08,
      "loss": 0.001,
      "step": 15266
    },
    {
      "epoch": 9.767754318618042,
      "grad_norm": 0.0674070742964711,
      "learning_rate": 2.8276296533854552e-08,
      "loss": 0.0006,
      "step": 15267
    },
    {
      "epoch": 9.768394113883557,
      "grad_norm": 0.15794518109431271,
      "learning_rate": 2.81207918136317e-08,
      "loss": 0.001,
      "step": 15268
    },
    {
      "epoch": 9.769033909149073,
      "grad_norm": 0.17195926719254737,
      "learning_rate": 2.7965715268162275e-08,
      "loss": 0.0011,
      "step": 15269
    },
    {
      "epoch": 9.769673704414588,
      "grad_norm": 0.2010770468220199,
      "learning_rate": 2.7811066904106508e-08,
      "loss": 0.0008,
      "step": 15270
    },
    {
      "epoch": 9.770313499680102,
      "grad_norm": 0.1258192923152595,
      "learning_rate": 2.7656846728103537e-08,
      "loss": 0.0009,
      "step": 15271
    },
    {
      "epoch": 9.770953294945617,
      "grad_norm": 0.11547713421748597,
      "learning_rate": 2.7503054746778057e-08,
      "loss": 0.001,
      "step": 15272
    },
    {
      "epoch": 9.771593090211132,
      "grad_norm": 0.09328978550773202,
      "learning_rate": 2.7349690966730347e-08,
      "loss": 0.0007,
      "step": 15273
    },
    {
      "epoch": 9.772232885476647,
      "grad_norm": 0.11221528844281413,
      "learning_rate": 2.719675539454625e-08,
      "loss": 0.0007,
      "step": 15274
    },
    {
      "epoch": 9.772872680742163,
      "grad_norm": 0.16672818194576852,
      "learning_rate": 2.7044248036794953e-08,
      "loss": 0.001,
      "step": 15275
    },
    {
      "epoch": 9.773512476007678,
      "grad_norm": 0.1508411765551476,
      "learning_rate": 2.689216890002233e-08,
      "loss": 0.001,
      "step": 15276
    },
    {
      "epoch": 9.774152271273193,
      "grad_norm": 0.10717824751146017,
      "learning_rate": 2.6740517990758718e-08,
      "loss": 0.0008,
      "step": 15277
    },
    {
      "epoch": 9.774792066538708,
      "grad_norm": 0.16827416868510026,
      "learning_rate": 2.6589295315516685e-08,
      "loss": 0.0009,
      "step": 15278
    },
    {
      "epoch": 9.775431861804222,
      "grad_norm": 0.10520385513720974,
      "learning_rate": 2.643850088078992e-08,
      "loss": 0.0007,
      "step": 15279
    },
    {
      "epoch": 9.776071657069737,
      "grad_norm": 0.134442301268197,
      "learning_rate": 2.6288134693051026e-08,
      "loss": 0.0009,
      "step": 15280
    },
    {
      "epoch": 9.776711452335253,
      "grad_norm": 0.11798945580441869,
      "learning_rate": 2.6138196758758173e-08,
      "loss": 0.0008,
      "step": 15281
    },
    {
      "epoch": 9.777351247600768,
      "grad_norm": 0.08430938205601247,
      "learning_rate": 2.598868708434843e-08,
      "loss": 0.0006,
      "step": 15282
    },
    {
      "epoch": 9.777991042866283,
      "grad_norm": 0.21629870506158796,
      "learning_rate": 2.583960567624333e-08,
      "loss": 0.0012,
      "step": 15283
    },
    {
      "epoch": 9.778630838131798,
      "grad_norm": 0.09144415883012356,
      "learning_rate": 2.5690952540842196e-08,
      "loss": 0.0008,
      "step": 15284
    },
    {
      "epoch": 9.779270633397314,
      "grad_norm": 0.16279674854535303,
      "learning_rate": 2.554272768452881e-08,
      "loss": 0.001,
      "step": 15285
    },
    {
      "epoch": 9.779910428662827,
      "grad_norm": 0.05238617380686359,
      "learning_rate": 2.5394931113666975e-08,
      "loss": 0.0005,
      "step": 15286
    },
    {
      "epoch": 9.780550223928342,
      "grad_norm": 0.19971112174202263,
      "learning_rate": 2.5247562834603832e-08,
      "loss": 0.0015,
      "step": 15287
    },
    {
      "epoch": 9.781190019193858,
      "grad_norm": 0.16155201634094293,
      "learning_rate": 2.5100622853666545e-08,
      "loss": 0.0013,
      "step": 15288
    },
    {
      "epoch": 9.781829814459373,
      "grad_norm": 0.13096402937553894,
      "learning_rate": 2.495411117716451e-08,
      "loss": 0.0009,
      "step": 15289
    },
    {
      "epoch": 9.782469609724888,
      "grad_norm": 0.1099805021042403,
      "learning_rate": 2.4808027811388247e-08,
      "loss": 0.0008,
      "step": 15290
    },
    {
      "epoch": 9.783109404990403,
      "grad_norm": 0.1033023748613802,
      "learning_rate": 2.466237276261163e-08,
      "loss": 0.0006,
      "step": 15291
    },
    {
      "epoch": 9.783749200255919,
      "grad_norm": 0.2196377282262589,
      "learning_rate": 2.4517146037087438e-08,
      "loss": 0.0014,
      "step": 15292
    },
    {
      "epoch": 9.784388995521432,
      "grad_norm": 0.11084069408664533,
      "learning_rate": 2.4372347641052897e-08,
      "loss": 0.0008,
      "step": 15293
    },
    {
      "epoch": 9.785028790786948,
      "grad_norm": 0.09510058127958695,
      "learning_rate": 2.422797758072304e-08,
      "loss": 0.0007,
      "step": 15294
    },
    {
      "epoch": 9.785668586052463,
      "grad_norm": 0.12216896673003093,
      "learning_rate": 2.4084035862299572e-08,
      "loss": 0.0009,
      "step": 15295
    },
    {
      "epoch": 9.786308381317978,
      "grad_norm": 0.10872331787556018,
      "learning_rate": 2.3940522491960884e-08,
      "loss": 0.0007,
      "step": 15296
    },
    {
      "epoch": 9.786948176583493,
      "grad_norm": 0.2237252810951601,
      "learning_rate": 2.3797437475870933e-08,
      "loss": 0.001,
      "step": 15297
    },
    {
      "epoch": 9.787587971849009,
      "grad_norm": 0.13406168727762843,
      "learning_rate": 2.3654780820171476e-08,
      "loss": 0.0009,
      "step": 15298
    },
    {
      "epoch": 9.788227767114524,
      "grad_norm": 0.1696080057640516,
      "learning_rate": 2.351255253099094e-08,
      "loss": 0.001,
      "step": 15299
    },
    {
      "epoch": 9.78886756238004,
      "grad_norm": 0.08791687481922211,
      "learning_rate": 2.3370752614433334e-08,
      "loss": 0.0007,
      "step": 15300
    },
    {
      "epoch": 9.789507357645553,
      "grad_norm": 0.2204929597058454,
      "learning_rate": 2.322938107658823e-08,
      "loss": 0.0011,
      "step": 15301
    },
    {
      "epoch": 9.790147152911068,
      "grad_norm": 0.0927466036488297,
      "learning_rate": 2.308843792352744e-08,
      "loss": 0.0008,
      "step": 15302
    },
    {
      "epoch": 9.790786948176583,
      "grad_norm": 0.1790629865741802,
      "learning_rate": 2.294792316130168e-08,
      "loss": 0.0009,
      "step": 15303
    },
    {
      "epoch": 9.791426743442099,
      "grad_norm": 0.10784189255111905,
      "learning_rate": 2.2807836795942784e-08,
      "loss": 0.0007,
      "step": 15304
    },
    {
      "epoch": 9.792066538707614,
      "grad_norm": 0.1343884003142166,
      "learning_rate": 2.266817883346817e-08,
      "loss": 0.0008,
      "step": 15305
    },
    {
      "epoch": 9.792706333973129,
      "grad_norm": 0.15393498938793893,
      "learning_rate": 2.2528949279874145e-08,
      "loss": 0.0013,
      "step": 15306
    },
    {
      "epoch": 9.793346129238644,
      "grad_norm": 0.20999505597007279,
      "learning_rate": 2.2390148141138157e-08,
      "loss": 0.0014,
      "step": 15307
    },
    {
      "epoch": 9.79398592450416,
      "grad_norm": 0.1430514533213237,
      "learning_rate": 2.225177542321988e-08,
      "loss": 0.0009,
      "step": 15308
    },
    {
      "epoch": 9.794625719769673,
      "grad_norm": 0.11786562933614779,
      "learning_rate": 2.2113831132062334e-08,
      "loss": 0.0009,
      "step": 15309
    },
    {
      "epoch": 9.795265515035188,
      "grad_norm": 0.40761782816727843,
      "learning_rate": 2.1976315273586347e-08,
      "loss": 0.0009,
      "step": 15310
    },
    {
      "epoch": 9.795905310300704,
      "grad_norm": 0.10989254909875822,
      "learning_rate": 2.1839227853698297e-08,
      "loss": 0.0009,
      "step": 15311
    },
    {
      "epoch": 9.796545105566219,
      "grad_norm": 0.16572810306218025,
      "learning_rate": 2.1702568878284592e-08,
      "loss": 0.0008,
      "step": 15312
    },
    {
      "epoch": 9.797184900831734,
      "grad_norm": 0.21709196941226666,
      "learning_rate": 2.1566338353212757e-08,
      "loss": 0.0012,
      "step": 15313
    },
    {
      "epoch": 9.79782469609725,
      "grad_norm": 0.1078447837842535,
      "learning_rate": 2.143053628433145e-08,
      "loss": 0.0009,
      "step": 15314
    },
    {
      "epoch": 9.798464491362765,
      "grad_norm": 0.23569019363113086,
      "learning_rate": 2.129516267747267e-08,
      "loss": 0.0014,
      "step": 15315
    },
    {
      "epoch": 9.799104286628278,
      "grad_norm": 0.117006618613359,
      "learning_rate": 2.1160217538448434e-08,
      "loss": 0.0009,
      "step": 15316
    },
    {
      "epoch": 9.799744081893794,
      "grad_norm": 0.08900848164609268,
      "learning_rate": 2.102570087305411e-08,
      "loss": 0.0007,
      "step": 15317
    },
    {
      "epoch": 9.800383877159309,
      "grad_norm": 0.2016847029533532,
      "learning_rate": 2.0891612687063967e-08,
      "loss": 0.0012,
      "step": 15318
    },
    {
      "epoch": 9.801023672424824,
      "grad_norm": 0.22081155893921206,
      "learning_rate": 2.075795298623784e-08,
      "loss": 0.001,
      "step": 15319
    },
    {
      "epoch": 9.80166346769034,
      "grad_norm": 0.26899753672687005,
      "learning_rate": 2.0624721776312252e-08,
      "loss": 0.0009,
      "step": 15320
    },
    {
      "epoch": 9.802303262955855,
      "grad_norm": 0.12909082447577827,
      "learning_rate": 2.04919190630104e-08,
      "loss": 0.0009,
      "step": 15321
    },
    {
      "epoch": 9.80294305822137,
      "grad_norm": 0.12296819382270467,
      "learning_rate": 2.0359544852032177e-08,
      "loss": 0.0009,
      "step": 15322
    },
    {
      "epoch": 9.803582853486883,
      "grad_norm": 0.12933241711057877,
      "learning_rate": 2.0227599149064136e-08,
      "loss": 0.0009,
      "step": 15323
    },
    {
      "epoch": 9.804222648752399,
      "grad_norm": 0.08118004527459986,
      "learning_rate": 2.009608195976842e-08,
      "loss": 0.0007,
      "step": 15324
    },
    {
      "epoch": 9.804862444017914,
      "grad_norm": 0.16412797627980222,
      "learning_rate": 1.9964993289796065e-08,
      "loss": 0.001,
      "step": 15325
    },
    {
      "epoch": 9.80550223928343,
      "grad_norm": 0.0941211660578149,
      "learning_rate": 1.9834333144772566e-08,
      "loss": 0.0007,
      "step": 15326
    },
    {
      "epoch": 9.806142034548945,
      "grad_norm": 0.22615057822177673,
      "learning_rate": 1.9704101530310104e-08,
      "loss": 0.0011,
      "step": 15327
    },
    {
      "epoch": 9.80678182981446,
      "grad_norm": 0.17644359697975648,
      "learning_rate": 1.957429845199865e-08,
      "loss": 0.001,
      "step": 15328
    },
    {
      "epoch": 9.807421625079975,
      "grad_norm": 0.13233692109982043,
      "learning_rate": 1.944492391541486e-08,
      "loss": 0.0009,
      "step": 15329
    },
    {
      "epoch": 9.808061420345489,
      "grad_norm": 0.08931637132704937,
      "learning_rate": 1.9315977926110952e-08,
      "loss": 0.0007,
      "step": 15330
    },
    {
      "epoch": 9.808701215611004,
      "grad_norm": 0.09416585735530222,
      "learning_rate": 1.918746048962472e-08,
      "loss": 0.0008,
      "step": 15331
    },
    {
      "epoch": 9.809341010876519,
      "grad_norm": 0.1046997258800229,
      "learning_rate": 1.9059371611475084e-08,
      "loss": 0.0007,
      "step": 15332
    },
    {
      "epoch": 9.809980806142034,
      "grad_norm": 0.14267364155286313,
      "learning_rate": 1.8931711297160983e-08,
      "loss": 0.0011,
      "step": 15333
    },
    {
      "epoch": 9.81062060140755,
      "grad_norm": 0.14050403821924345,
      "learning_rate": 1.8804479552163578e-08,
      "loss": 0.0008,
      "step": 15334
    },
    {
      "epoch": 9.811260396673065,
      "grad_norm": 0.14590273823370034,
      "learning_rate": 1.8677676381947395e-08,
      "loss": 0.001,
      "step": 15335
    },
    {
      "epoch": 9.81190019193858,
      "grad_norm": 0.14711994666837339,
      "learning_rate": 1.8551301791955856e-08,
      "loss": 0.0012,
      "step": 15336
    },
    {
      "epoch": 9.812539987204095,
      "grad_norm": 0.25015469633433873,
      "learning_rate": 1.842535578761573e-08,
      "loss": 0.0015,
      "step": 15337
    },
    {
      "epoch": 9.813179782469609,
      "grad_norm": 0.22397252316945734,
      "learning_rate": 1.829983837433602e-08,
      "loss": 0.0016,
      "step": 15338
    },
    {
      "epoch": 9.813819577735124,
      "grad_norm": 0.17005784387468453,
      "learning_rate": 1.8174749557504643e-08,
      "loss": 0.0011,
      "step": 15339
    },
    {
      "epoch": 9.81445937300064,
      "grad_norm": 0.12466636207721946,
      "learning_rate": 1.805008934249286e-08,
      "loss": 0.0008,
      "step": 15340
    },
    {
      "epoch": 9.815099168266155,
      "grad_norm": 0.15160776704787501,
      "learning_rate": 1.7925857734654163e-08,
      "loss": 0.0008,
      "step": 15341
    },
    {
      "epoch": 9.81573896353167,
      "grad_norm": 0.22652552787334493,
      "learning_rate": 1.780205473932206e-08,
      "loss": 0.0011,
      "step": 15342
    },
    {
      "epoch": 9.816378758797185,
      "grad_norm": 0.12082668412398617,
      "learning_rate": 1.7678680361813415e-08,
      "loss": 0.0009,
      "step": 15343
    },
    {
      "epoch": 9.8170185540627,
      "grad_norm": 0.0958800418234719,
      "learning_rate": 1.7555734607423992e-08,
      "loss": 0.0009,
      "step": 15344
    },
    {
      "epoch": 9.817658349328216,
      "grad_norm": 0.20139939947015248,
      "learning_rate": 1.7433217481435115e-08,
      "loss": 0.0013,
      "step": 15345
    },
    {
      "epoch": 9.81829814459373,
      "grad_norm": 0.07627915020374561,
      "learning_rate": 1.731112898910592e-08,
      "loss": 0.0007,
      "step": 15346
    },
    {
      "epoch": 9.818937939859245,
      "grad_norm": 0.13061089449662155,
      "learning_rate": 1.718946913567776e-08,
      "loss": 0.0009,
      "step": 15347
    },
    {
      "epoch": 9.81957773512476,
      "grad_norm": 0.12234183064157132,
      "learning_rate": 1.7068237926377575e-08,
      "loss": 0.0008,
      "step": 15348
    },
    {
      "epoch": 9.820217530390275,
      "grad_norm": 0.16893597074782443,
      "learning_rate": 1.694743536640786e-08,
      "loss": 0.0011,
      "step": 15349
    },
    {
      "epoch": 9.82085732565579,
      "grad_norm": 0.1246571647374111,
      "learning_rate": 1.6827061460956695e-08,
      "loss": 0.0011,
      "step": 15350
    },
    {
      "epoch": 9.821497120921306,
      "grad_norm": 0.13418509981224966,
      "learning_rate": 1.6707116215194384e-08,
      "loss": 0.001,
      "step": 15351
    },
    {
      "epoch": 9.822136916186821,
      "grad_norm": 0.14962586256160038,
      "learning_rate": 1.6587599634267926e-08,
      "loss": 0.001,
      "step": 15352
    },
    {
      "epoch": 9.822776711452335,
      "grad_norm": 0.09803676443002536,
      "learning_rate": 1.6468511723310987e-08,
      "loss": 0.0008,
      "step": 15353
    },
    {
      "epoch": 9.82341650671785,
      "grad_norm": 0.1655965051251695,
      "learning_rate": 1.6349852487437256e-08,
      "loss": 0.001,
      "step": 15354
    },
    {
      "epoch": 9.824056301983365,
      "grad_norm": 0.14861079892461215,
      "learning_rate": 1.6231621931741547e-08,
      "loss": 0.0009,
      "step": 15355
    },
    {
      "epoch": 9.82469609724888,
      "grad_norm": 0.19198855477473933,
      "learning_rate": 1.61138200612998e-08,
      "loss": 0.0014,
      "step": 15356
    },
    {
      "epoch": 9.825335892514396,
      "grad_norm": 0.11076908026673098,
      "learning_rate": 1.5996446881171303e-08,
      "loss": 0.0007,
      "step": 15357
    },
    {
      "epoch": 9.825975687779911,
      "grad_norm": 0.19703836536441677,
      "learning_rate": 1.587950239639424e-08,
      "loss": 0.0014,
      "step": 15358
    },
    {
      "epoch": 9.826615483045426,
      "grad_norm": 0.18692506101534692,
      "learning_rate": 1.576298661199238e-08,
      "loss": 0.0013,
      "step": 15359
    },
    {
      "epoch": 9.82725527831094,
      "grad_norm": 0.19048941783923776,
      "learning_rate": 1.5646899532966165e-08,
      "loss": 0.0012,
      "step": 15360
    },
    {
      "epoch": 9.827895073576455,
      "grad_norm": 0.12663139868628642,
      "learning_rate": 1.5531241164301604e-08,
      "loss": 0.0008,
      "step": 15361
    },
    {
      "epoch": 9.82853486884197,
      "grad_norm": 0.17273003334772835,
      "learning_rate": 1.5416011510964725e-08,
      "loss": 0.0011,
      "step": 15362
    },
    {
      "epoch": 9.829174664107486,
      "grad_norm": 0.1783182012682112,
      "learning_rate": 1.530121057790268e-08,
      "loss": 0.0011,
      "step": 15363
    },
    {
      "epoch": 9.829814459373,
      "grad_norm": 0.1665760948438022,
      "learning_rate": 1.5186838370044865e-08,
      "loss": 0.001,
      "step": 15364
    },
    {
      "epoch": 9.830454254638516,
      "grad_norm": 0.1391320086854076,
      "learning_rate": 1.5072894892304014e-08,
      "loss": 0.001,
      "step": 15365
    },
    {
      "epoch": 9.831094049904031,
      "grad_norm": 0.10665402367446902,
      "learning_rate": 1.4959380149569548e-08,
      "loss": 0.0009,
      "step": 15366
    },
    {
      "epoch": 9.831733845169545,
      "grad_norm": 0.1879648508853368,
      "learning_rate": 1.4846294146718677e-08,
      "loss": 0.0012,
      "step": 15367
    },
    {
      "epoch": 9.83237364043506,
      "grad_norm": 0.14277561721602847,
      "learning_rate": 1.4733636888605296e-08,
      "loss": 0.0009,
      "step": 15368
    },
    {
      "epoch": 9.833013435700575,
      "grad_norm": 0.12617315589191083,
      "learning_rate": 1.4621408380066648e-08,
      "loss": 0.001,
      "step": 15369
    },
    {
      "epoch": 9.83365323096609,
      "grad_norm": 0.0882958853709554,
      "learning_rate": 1.450960862592332e-08,
      "loss": 0.0007,
      "step": 15370
    },
    {
      "epoch": 9.834293026231606,
      "grad_norm": 0.25618122293793216,
      "learning_rate": 1.43982376309737e-08,
      "loss": 0.0014,
      "step": 15371
    },
    {
      "epoch": 9.834932821497121,
      "grad_norm": 0.15229202335160835,
      "learning_rate": 1.4287295400000623e-08,
      "loss": 0.0009,
      "step": 15372
    },
    {
      "epoch": 9.835572616762636,
      "grad_norm": 0.12274157323513789,
      "learning_rate": 1.4176781937768058e-08,
      "loss": 0.0008,
      "step": 15373
    },
    {
      "epoch": 9.836212412028152,
      "grad_norm": 0.10508118237738519,
      "learning_rate": 1.40666972490211e-08,
      "loss": 0.0008,
      "step": 15374
    },
    {
      "epoch": 9.836852207293665,
      "grad_norm": 0.2221381765656579,
      "learning_rate": 1.3957041338487077e-08,
      "loss": 0.0014,
      "step": 15375
    },
    {
      "epoch": 9.83749200255918,
      "grad_norm": 0.08525963841593963,
      "learning_rate": 1.3847814210873333e-08,
      "loss": 0.0007,
      "step": 15376
    },
    {
      "epoch": 9.838131797824696,
      "grad_norm": 0.13996458705977183,
      "learning_rate": 1.3739015870870565e-08,
      "loss": 0.0011,
      "step": 15377
    },
    {
      "epoch": 9.838771593090211,
      "grad_norm": 0.10433752054453674,
      "learning_rate": 1.3630646323149476e-08,
      "loss": 0.0009,
      "step": 15378
    },
    {
      "epoch": 9.839411388355726,
      "grad_norm": 0.09957455933571033,
      "learning_rate": 1.3522705572365236e-08,
      "loss": 0.0009,
      "step": 15379
    },
    {
      "epoch": 9.840051183621242,
      "grad_norm": 0.19235926546761942,
      "learning_rate": 1.3415193623150802e-08,
      "loss": 0.001,
      "step": 15380
    },
    {
      "epoch": 9.840690978886757,
      "grad_norm": 0.11653072925347699,
      "learning_rate": 1.3308110480123592e-08,
      "loss": 0.0008,
      "step": 15381
    },
    {
      "epoch": 9.841330774152272,
      "grad_norm": 0.11526097431366039,
      "learning_rate": 1.3201456147881042e-08,
      "loss": 0.0009,
      "step": 15382
    },
    {
      "epoch": 9.841970569417786,
      "grad_norm": 0.22455536573202792,
      "learning_rate": 1.309523063100282e-08,
      "loss": 0.0011,
      "step": 15383
    },
    {
      "epoch": 9.842610364683301,
      "grad_norm": 0.21868230326228738,
      "learning_rate": 1.2989433934049723e-08,
      "loss": 0.0008,
      "step": 15384
    },
    {
      "epoch": 9.843250159948816,
      "grad_norm": 0.12680363528193975,
      "learning_rate": 1.2884066061565893e-08,
      "loss": 0.0008,
      "step": 15385
    },
    {
      "epoch": 9.843889955214332,
      "grad_norm": 0.13723344700678208,
      "learning_rate": 1.2779127018073267e-08,
      "loss": 0.001,
      "step": 15386
    },
    {
      "epoch": 9.844529750479847,
      "grad_norm": 0.11257914488679308,
      "learning_rate": 1.2674616808080464e-08,
      "loss": 0.0008,
      "step": 15387
    },
    {
      "epoch": 9.845169545745362,
      "grad_norm": 0.09880096095234632,
      "learning_rate": 1.2570535436071674e-08,
      "loss": 0.0008,
      "step": 15388
    },
    {
      "epoch": 9.845809341010877,
      "grad_norm": 0.13426862694274166,
      "learning_rate": 1.2466882906518874e-08,
      "loss": 0.0009,
      "step": 15389
    },
    {
      "epoch": 9.84644913627639,
      "grad_norm": 0.15691477339929508,
      "learning_rate": 1.2363659223871838e-08,
      "loss": 0.0011,
      "step": 15390
    },
    {
      "epoch": 9.847088931541906,
      "grad_norm": 0.15127848228244994,
      "learning_rate": 1.2260864392562577e-08,
      "loss": 0.0008,
      "step": 15391
    },
    {
      "epoch": 9.847728726807421,
      "grad_norm": 0.11482286002384473,
      "learning_rate": 1.2158498417005338e-08,
      "loss": 0.0008,
      "step": 15392
    },
    {
      "epoch": 9.848368522072937,
      "grad_norm": 0.08035135683556625,
      "learning_rate": 1.2056561301594383e-08,
      "loss": 0.0007,
      "step": 15393
    },
    {
      "epoch": 9.849008317338452,
      "grad_norm": 0.12718717440433197,
      "learning_rate": 1.1955053050707322e-08,
      "loss": 0.001,
      "step": 15394
    },
    {
      "epoch": 9.849648112603967,
      "grad_norm": 0.12432115925951642,
      "learning_rate": 1.1853973668704e-08,
      "loss": 0.0007,
      "step": 15395
    },
    {
      "epoch": 9.850287907869482,
      "grad_norm": 0.14111351451144075,
      "learning_rate": 1.1753323159923169e-08,
      "loss": 0.0011,
      "step": 15396
    },
    {
      "epoch": 9.850927703134996,
      "grad_norm": 0.13886638231241982,
      "learning_rate": 1.1653101528686928e-08,
      "loss": 0.001,
      "step": 15397
    },
    {
      "epoch": 9.851567498400511,
      "grad_norm": 0.10349122196299589,
      "learning_rate": 1.15533087792985e-08,
      "loss": 0.0007,
      "step": 15398
    },
    {
      "epoch": 9.852207293666027,
      "grad_norm": 0.12940386737983162,
      "learning_rate": 1.1453944916042237e-08,
      "loss": 0.0009,
      "step": 15399
    },
    {
      "epoch": 9.852847088931542,
      "grad_norm": 0.18114344823162146,
      "learning_rate": 1.1355009943185835e-08,
      "loss": 0.0011,
      "step": 15400
    },
    {
      "epoch": 9.853486884197057,
      "grad_norm": 0.14778771303314095,
      "learning_rate": 1.125650386497701e-08,
      "loss": 0.001,
      "step": 15401
    },
    {
      "epoch": 9.854126679462572,
      "grad_norm": 0.13814464346363187,
      "learning_rate": 1.1158426685645707e-08,
      "loss": 0.0009,
      "step": 15402
    },
    {
      "epoch": 9.854766474728088,
      "grad_norm": 0.1949743991990142,
      "learning_rate": 1.1060778409401895e-08,
      "loss": 0.0012,
      "step": 15403
    },
    {
      "epoch": 9.855406269993601,
      "grad_norm": 0.140748599049785,
      "learning_rate": 1.0963559040439996e-08,
      "loss": 0.0008,
      "step": 15404
    },
    {
      "epoch": 9.856046065259116,
      "grad_norm": 0.13286863091907045,
      "learning_rate": 1.086676858293334e-08,
      "loss": 0.0009,
      "step": 15405
    },
    {
      "epoch": 9.856685860524632,
      "grad_norm": 0.1202327811309614,
      "learning_rate": 1.07704070410386e-08,
      "loss": 0.0008,
      "step": 15406
    },
    {
      "epoch": 9.857325655790147,
      "grad_norm": 0.1682609822813978,
      "learning_rate": 1.0674474418893577e-08,
      "loss": 0.001,
      "step": 15407
    },
    {
      "epoch": 9.857965451055662,
      "grad_norm": 0.2250612773429429,
      "learning_rate": 1.057897072061609e-08,
      "loss": 0.0014,
      "step": 15408
    },
    {
      "epoch": 9.858605246321178,
      "grad_norm": 0.10880289967319137,
      "learning_rate": 1.048389595030841e-08,
      "loss": 0.0008,
      "step": 15409
    },
    {
      "epoch": 9.859245041586693,
      "grad_norm": 0.18807655839082613,
      "learning_rate": 1.0389250112051719e-08,
      "loss": 0.0009,
      "step": 15410
    },
    {
      "epoch": 9.859884836852208,
      "grad_norm": 0.1057895195132854,
      "learning_rate": 1.0295033209911653e-08,
      "loss": 0.0007,
      "step": 15411
    },
    {
      "epoch": 9.860524632117722,
      "grad_norm": 0.16030436100397544,
      "learning_rate": 1.0201245247931646e-08,
      "loss": 0.0013,
      "step": 15412
    },
    {
      "epoch": 9.861164427383237,
      "grad_norm": 0.2396575855630622,
      "learning_rate": 1.0107886230139585e-08,
      "loss": 0.0014,
      "step": 15413
    },
    {
      "epoch": 9.861804222648752,
      "grad_norm": 0.34354847915937825,
      "learning_rate": 1.0014956160544487e-08,
      "loss": 0.0019,
      "step": 15414
    },
    {
      "epoch": 9.862444017914267,
      "grad_norm": 0.13244891300294656,
      "learning_rate": 9.922455043136492e-09,
      "loss": 0.0009,
      "step": 15415
    },
    {
      "epoch": 9.863083813179783,
      "grad_norm": 0.14343416540524154,
      "learning_rate": 9.830382881886868e-09,
      "loss": 0.001,
      "step": 15416
    },
    {
      "epoch": 9.863723608445298,
      "grad_norm": 0.12329551163669812,
      "learning_rate": 9.738739680750232e-09,
      "loss": 0.001,
      "step": 15417
    },
    {
      "epoch": 9.864363403710813,
      "grad_norm": 0.11368721838412116,
      "learning_rate": 9.647525443660099e-09,
      "loss": 0.0008,
      "step": 15418
    },
    {
      "epoch": 9.865003198976328,
      "grad_norm": 0.18469892322797296,
      "learning_rate": 9.556740174533341e-09,
      "loss": 0.001,
      "step": 15419
    },
    {
      "epoch": 9.865642994241842,
      "grad_norm": 0.19371197719424904,
      "learning_rate": 9.466383877270169e-09,
      "loss": 0.0016,
      "step": 15420
    },
    {
      "epoch": 9.866282789507357,
      "grad_norm": 0.10116102540442846,
      "learning_rate": 9.376456555747482e-09,
      "loss": 0.0008,
      "step": 15421
    },
    {
      "epoch": 9.866922584772873,
      "grad_norm": 0.12995081099198097,
      "learning_rate": 9.286958213827746e-09,
      "loss": 0.0008,
      "step": 15422
    },
    {
      "epoch": 9.867562380038388,
      "grad_norm": 0.12104171951017656,
      "learning_rate": 9.197888855354553e-09,
      "loss": 0.001,
      "step": 15423
    },
    {
      "epoch": 9.868202175303903,
      "grad_norm": 0.10901698297225049,
      "learning_rate": 9.109248484151511e-09,
      "loss": 0.0009,
      "step": 15424
    },
    {
      "epoch": 9.868841970569418,
      "grad_norm": 0.1661820253670065,
      "learning_rate": 9.021037104024466e-09,
      "loss": 0.0011,
      "step": 15425
    },
    {
      "epoch": 9.869481765834934,
      "grad_norm": 0.11067055977433163,
      "learning_rate": 8.933254718761497e-09,
      "loss": 0.0008,
      "step": 15426
    },
    {
      "epoch": 9.870121561100447,
      "grad_norm": 0.10268664062357634,
      "learning_rate": 8.845901332132923e-09,
      "loss": 0.0008,
      "step": 15427
    },
    {
      "epoch": 9.870761356365962,
      "grad_norm": 0.12438826731817487,
      "learning_rate": 8.758976947887966e-09,
      "loss": 0.001,
      "step": 15428
    },
    {
      "epoch": 9.871401151631478,
      "grad_norm": 0.11297858975418691,
      "learning_rate": 8.672481569759194e-09,
      "loss": 0.0008,
      "step": 15429
    },
    {
      "epoch": 9.872040946896993,
      "grad_norm": 0.3487940576128272,
      "learning_rate": 8.586415201460308e-09,
      "loss": 0.0009,
      "step": 15430
    },
    {
      "epoch": 9.872680742162508,
      "grad_norm": 0.10164781883235507,
      "learning_rate": 8.500777846688347e-09,
      "loss": 0.0008,
      "step": 15431
    },
    {
      "epoch": 9.873320537428024,
      "grad_norm": 0.17391340883242337,
      "learning_rate": 8.415569509119258e-09,
      "loss": 0.0012,
      "step": 15432
    },
    {
      "epoch": 9.873960332693539,
      "grad_norm": 0.1475423042036329,
      "learning_rate": 8.330790192412341e-09,
      "loss": 0.0009,
      "step": 15433
    },
    {
      "epoch": 9.874600127959052,
      "grad_norm": 0.08719120878843756,
      "learning_rate": 8.246439900206904e-09,
      "loss": 0.0007,
      "step": 15434
    },
    {
      "epoch": 9.875239923224568,
      "grad_norm": 0.13031722514620112,
      "learning_rate": 8.162518636125605e-09,
      "loss": 0.0009,
      "step": 15435
    },
    {
      "epoch": 9.875879718490083,
      "grad_norm": 0.08100455379939597,
      "learning_rate": 8.079026403771117e-09,
      "loss": 0.0007,
      "step": 15436
    },
    {
      "epoch": 9.876519513755598,
      "grad_norm": 0.0680821819268255,
      "learning_rate": 7.995963206729462e-09,
      "loss": 0.0007,
      "step": 15437
    },
    {
      "epoch": 9.877159309021113,
      "grad_norm": 0.10867973021714457,
      "learning_rate": 7.913329048566676e-09,
      "loss": 0.0008,
      "step": 15438
    },
    {
      "epoch": 9.877799104286629,
      "grad_norm": 0.13564400440447863,
      "learning_rate": 7.83112393283103e-09,
      "loss": 0.001,
      "step": 15439
    },
    {
      "epoch": 9.878438899552144,
      "grad_norm": 0.1425603511428601,
      "learning_rate": 7.749347863053035e-09,
      "loss": 0.0011,
      "step": 15440
    },
    {
      "epoch": 9.879078694817657,
      "grad_norm": 0.12525172814394406,
      "learning_rate": 7.668000842742108e-09,
      "loss": 0.0009,
      "step": 15441
    },
    {
      "epoch": 9.879718490083173,
      "grad_norm": 0.08548739964158215,
      "learning_rate": 7.587082875393225e-09,
      "loss": 0.0007,
      "step": 15442
    },
    {
      "epoch": 9.880358285348688,
      "grad_norm": 0.12357803288237493,
      "learning_rate": 7.506593964479169e-09,
      "loss": 0.001,
      "step": 15443
    },
    {
      "epoch": 9.880998080614203,
      "grad_norm": 0.08700240531094443,
      "learning_rate": 7.4265341134571736e-09,
      "loss": 0.0006,
      "step": 15444
    },
    {
      "epoch": 9.881637875879719,
      "grad_norm": 0.08120941344541233,
      "learning_rate": 7.3469033257633774e-09,
      "loss": 0.0007,
      "step": 15445
    },
    {
      "epoch": 9.882277671145234,
      "grad_norm": 0.11007361233488393,
      "learning_rate": 7.267701604818378e-09,
      "loss": 0.0008,
      "step": 15446
    },
    {
      "epoch": 9.882917466410749,
      "grad_norm": 0.09160150360271595,
      "learning_rate": 7.188928954022789e-09,
      "loss": 0.0008,
      "step": 15447
    },
    {
      "epoch": 9.883557261676264,
      "grad_norm": 0.11256280188222943,
      "learning_rate": 7.11058537675946e-09,
      "loss": 0.0008,
      "step": 15448
    },
    {
      "epoch": 9.884197056941778,
      "grad_norm": 0.17092372102888506,
      "learning_rate": 7.032670876390146e-09,
      "loss": 0.001,
      "step": 15449
    },
    {
      "epoch": 9.884836852207293,
      "grad_norm": 0.12500931811744692,
      "learning_rate": 6.9551854562632805e-09,
      "loss": 0.0008,
      "step": 15450
    },
    {
      "epoch": 9.885476647472808,
      "grad_norm": 0.21357233483982185,
      "learning_rate": 6.87812911970398e-09,
      "loss": 0.0017,
      "step": 15451
    },
    {
      "epoch": 9.886116442738324,
      "grad_norm": 0.16129071361586705,
      "learning_rate": 6.80150187002071e-09,
      "loss": 0.0011,
      "step": 15452
    },
    {
      "epoch": 9.886756238003839,
      "grad_norm": 0.09503052219517645,
      "learning_rate": 6.725303710505282e-09,
      "loss": 0.0007,
      "step": 15453
    },
    {
      "epoch": 9.887396033269354,
      "grad_norm": 0.15407249326165343,
      "learning_rate": 6.649534644428412e-09,
      "loss": 0.0011,
      "step": 15454
    },
    {
      "epoch": 9.88803582853487,
      "grad_norm": 0.1634997260445164,
      "learning_rate": 6.574194675044165e-09,
      "loss": 0.001,
      "step": 15455
    },
    {
      "epoch": 9.888675623800385,
      "grad_norm": 0.1323519995300905,
      "learning_rate": 6.49928380558662e-09,
      "loss": 0.0008,
      "step": 15456
    },
    {
      "epoch": 9.889315419065898,
      "grad_norm": 0.1135862866539266,
      "learning_rate": 6.424802039272093e-09,
      "loss": 0.0009,
      "step": 15457
    },
    {
      "epoch": 9.889955214331414,
      "grad_norm": 0.15242208382874944,
      "learning_rate": 6.350749379300247e-09,
      "loss": 0.001,
      "step": 15458
    },
    {
      "epoch": 9.890595009596929,
      "grad_norm": 0.1503711626296458,
      "learning_rate": 6.277125828850761e-09,
      "loss": 0.001,
      "step": 15459
    },
    {
      "epoch": 9.891234804862444,
      "grad_norm": 0.23179554855776113,
      "learning_rate": 6.203931391082219e-09,
      "loss": 0.0015,
      "step": 15460
    },
    {
      "epoch": 9.89187460012796,
      "grad_norm": 0.13643299253451954,
      "learning_rate": 6.131166069139882e-09,
      "loss": 0.0008,
      "step": 15461
    },
    {
      "epoch": 9.892514395393475,
      "grad_norm": 0.13588512192980798,
      "learning_rate": 6.05882986614792e-09,
      "loss": 0.0009,
      "step": 15462
    },
    {
      "epoch": 9.89315419065899,
      "grad_norm": 0.10487720102183046,
      "learning_rate": 5.986922785211624e-09,
      "loss": 0.0008,
      "step": 15463
    },
    {
      "epoch": 9.893793985924503,
      "grad_norm": 0.1328542817570363,
      "learning_rate": 5.915444829419636e-09,
      "loss": 0.0009,
      "step": 15464
    },
    {
      "epoch": 9.894433781190019,
      "grad_norm": 0.11687630220559632,
      "learning_rate": 5.844396001839503e-09,
      "loss": 0.0009,
      "step": 15465
    },
    {
      "epoch": 9.895073576455534,
      "grad_norm": 0.180690156002016,
      "learning_rate": 5.773776305523226e-09,
      "loss": 0.0011,
      "step": 15466
    },
    {
      "epoch": 9.89571337172105,
      "grad_norm": 0.11128751681100524,
      "learning_rate": 5.703585743502826e-09,
      "loss": 0.0009,
      "step": 15467
    },
    {
      "epoch": 9.896353166986565,
      "grad_norm": 0.1705211392653577,
      "learning_rate": 5.633824318791448e-09,
      "loss": 0.001,
      "step": 15468
    },
    {
      "epoch": 9.89699296225208,
      "grad_norm": 0.13062631824367388,
      "learning_rate": 5.564492034385582e-09,
      "loss": 0.0008,
      "step": 15469
    },
    {
      "epoch": 9.897632757517595,
      "grad_norm": 0.11130130573091822,
      "learning_rate": 5.495588893260628e-09,
      "loss": 0.0008,
      "step": 15470
    },
    {
      "epoch": 9.898272552783109,
      "grad_norm": 0.15646194489541249,
      "learning_rate": 5.42711489837755e-09,
      "loss": 0.0011,
      "step": 15471
    },
    {
      "epoch": 9.898912348048624,
      "grad_norm": 0.17131658251049703,
      "learning_rate": 5.359070052673998e-09,
      "loss": 0.0011,
      "step": 15472
    },
    {
      "epoch": 9.89955214331414,
      "grad_norm": 0.12122771259415084,
      "learning_rate": 5.2914543590743e-09,
      "loss": 0.0009,
      "step": 15473
    },
    {
      "epoch": 9.900191938579654,
      "grad_norm": 0.16799786712903206,
      "learning_rate": 5.224267820479467e-09,
      "loss": 0.0011,
      "step": 15474
    },
    {
      "epoch": 9.90083173384517,
      "grad_norm": 0.13636203166557773,
      "learning_rate": 5.157510439774971e-09,
      "loss": 0.0008,
      "step": 15475
    },
    {
      "epoch": 9.901471529110685,
      "grad_norm": 0.1539451352504816,
      "learning_rate": 5.091182219827406e-09,
      "loss": 0.0012,
      "step": 15476
    },
    {
      "epoch": 9.9021113243762,
      "grad_norm": 0.16328309169636562,
      "learning_rate": 5.025283163485606e-09,
      "loss": 0.0013,
      "step": 15477
    },
    {
      "epoch": 9.902751119641714,
      "grad_norm": 0.10006978880648137,
      "learning_rate": 4.959813273578418e-09,
      "loss": 0.0008,
      "step": 15478
    },
    {
      "epoch": 9.903390914907229,
      "grad_norm": 0.10955578455342144,
      "learning_rate": 4.894772552916926e-09,
      "loss": 0.0007,
      "step": 15479
    },
    {
      "epoch": 9.904030710172744,
      "grad_norm": 0.11013644583438426,
      "learning_rate": 4.830161004294454e-09,
      "loss": 0.0007,
      "step": 15480
    },
    {
      "epoch": 9.90467050543826,
      "grad_norm": 0.14866890236393018,
      "learning_rate": 4.765978630484336e-09,
      "loss": 0.0008,
      "step": 15481
    },
    {
      "epoch": 9.905310300703775,
      "grad_norm": 0.12094873373033889,
      "learning_rate": 4.702225434242147e-09,
      "loss": 0.0008,
      "step": 15482
    },
    {
      "epoch": 9.90595009596929,
      "grad_norm": 0.14745013820116118,
      "learning_rate": 4.6389014183079176e-09,
      "loss": 0.001,
      "step": 15483
    },
    {
      "epoch": 9.906589891234805,
      "grad_norm": 0.17566014577912278,
      "learning_rate": 4.5760065853972524e-09,
      "loss": 0.0011,
      "step": 15484
    },
    {
      "epoch": 9.90722968650032,
      "grad_norm": 0.14702825653732468,
      "learning_rate": 4.513540938213546e-09,
      "loss": 0.0009,
      "step": 15485
    },
    {
      "epoch": 9.907869481765834,
      "grad_norm": 0.10168140406350423,
      "learning_rate": 4.451504479436875e-09,
      "loss": 0.0007,
      "step": 15486
    },
    {
      "epoch": 9.90850927703135,
      "grad_norm": 0.11926105675371977,
      "learning_rate": 4.3898972117328856e-09,
      "loss": 0.0008,
      "step": 15487
    },
    {
      "epoch": 9.909149072296865,
      "grad_norm": 0.08121123440529417,
      "learning_rate": 4.328719137745019e-09,
      "loss": 0.0007,
      "step": 15488
    },
    {
      "epoch": 9.90978886756238,
      "grad_norm": 0.10512985782425989,
      "learning_rate": 4.267970260101173e-09,
      "loss": 0.0008,
      "step": 15489
    },
    {
      "epoch": 9.910428662827895,
      "grad_norm": 0.20405680169559354,
      "learning_rate": 4.207650581410372e-09,
      "loss": 0.0013,
      "step": 15490
    },
    {
      "epoch": 9.91106845809341,
      "grad_norm": 0.16956209889867538,
      "learning_rate": 4.147760104261656e-09,
      "loss": 0.0011,
      "step": 15491
    },
    {
      "epoch": 9.911708253358926,
      "grad_norm": 0.10816312082926911,
      "learning_rate": 4.088298831226301e-09,
      "loss": 0.0008,
      "step": 15492
    },
    {
      "epoch": 9.912348048624441,
      "grad_norm": 0.10173345946258201,
      "learning_rate": 4.02926676485782e-09,
      "loss": 0.0009,
      "step": 15493
    },
    {
      "epoch": 9.912987843889955,
      "grad_norm": 0.17612935161901427,
      "learning_rate": 3.9706639076919625e-09,
      "loss": 0.001,
      "step": 15494
    },
    {
      "epoch": 9.91362763915547,
      "grad_norm": 0.14080974555600384,
      "learning_rate": 3.9124902622444945e-09,
      "loss": 0.001,
      "step": 15495
    },
    {
      "epoch": 9.914267434420985,
      "grad_norm": 0.19932392087170306,
      "learning_rate": 3.854745831012308e-09,
      "loss": 0.0012,
      "step": 15496
    },
    {
      "epoch": 9.9149072296865,
      "grad_norm": 0.16341879714845192,
      "learning_rate": 3.7974306164756395e-09,
      "loss": 0.0009,
      "step": 15497
    },
    {
      "epoch": 9.915547024952016,
      "grad_norm": 0.09875976289238227,
      "learning_rate": 3.740544621095854e-09,
      "loss": 0.0007,
      "step": 15498
    },
    {
      "epoch": 9.916186820217531,
      "grad_norm": 0.09204216218510623,
      "learning_rate": 3.6840878473143327e-09,
      "loss": 0.0008,
      "step": 15499
    },
    {
      "epoch": 9.916826615483046,
      "grad_norm": 0.15600459779436032,
      "learning_rate": 3.6280602975569122e-09,
      "loss": 0.0009,
      "step": 15500
    },
    {
      "epoch": 9.91746641074856,
      "grad_norm": 0.14272520903758776,
      "learning_rate": 3.572461974228336e-09,
      "loss": 0.0009,
      "step": 15501
    },
    {
      "epoch": 9.918106206014075,
      "grad_norm": 0.135362126346429,
      "learning_rate": 3.517292879715584e-09,
      "loss": 0.001,
      "step": 15502
    },
    {
      "epoch": 9.91874600127959,
      "grad_norm": 0.1390325755789485,
      "learning_rate": 3.4625530163878706e-09,
      "loss": 0.0011,
      "step": 15503
    },
    {
      "epoch": 9.919385796545106,
      "grad_norm": 0.11438512257715615,
      "learning_rate": 3.4082423865955396e-09,
      "loss": 0.0007,
      "step": 15504
    },
    {
      "epoch": 9.92002559181062,
      "grad_norm": 0.08963420496354063,
      "learning_rate": 3.3543609926711685e-09,
      "loss": 0.0006,
      "step": 15505
    },
    {
      "epoch": 9.920665387076136,
      "grad_norm": 0.11257766973925251,
      "learning_rate": 3.300908836927352e-09,
      "loss": 0.0008,
      "step": 15506
    },
    {
      "epoch": 9.921305182341651,
      "grad_norm": 0.17655965041834912,
      "learning_rate": 3.2478859216611425e-09,
      "loss": 0.0012,
      "step": 15507
    },
    {
      "epoch": 9.921944977607165,
      "grad_norm": 0.09540985086643707,
      "learning_rate": 3.1952922491462755e-09,
      "loss": 0.0006,
      "step": 15508
    },
    {
      "epoch": 9.92258477287268,
      "grad_norm": 0.0847791437789807,
      "learning_rate": 3.143127821643166e-09,
      "loss": 0.0006,
      "step": 15509
    },
    {
      "epoch": 9.923224568138195,
      "grad_norm": 0.11577961886312362,
      "learning_rate": 3.091392641391133e-09,
      "loss": 0.001,
      "step": 15510
    },
    {
      "epoch": 9.92386436340371,
      "grad_norm": 0.1283813135535754,
      "learning_rate": 3.0400867106117336e-09,
      "loss": 0.001,
      "step": 15511
    },
    {
      "epoch": 9.924504158669226,
      "grad_norm": 0.13252325974840629,
      "learning_rate": 2.98921003150765e-09,
      "loss": 0.0009,
      "step": 15512
    },
    {
      "epoch": 9.925143953934741,
      "grad_norm": 0.11546032663191502,
      "learning_rate": 2.9387626062638007e-09,
      "loss": 0.0009,
      "step": 15513
    },
    {
      "epoch": 9.925783749200257,
      "grad_norm": 0.15343963900498203,
      "learning_rate": 2.8887444370462315e-09,
      "loss": 0.001,
      "step": 15514
    },
    {
      "epoch": 9.92642354446577,
      "grad_norm": 0.1434432596558574,
      "learning_rate": 2.839155526001003e-09,
      "loss": 0.001,
      "step": 15515
    },
    {
      "epoch": 9.927063339731285,
      "grad_norm": 0.34542127562975783,
      "learning_rate": 2.7899958752608535e-09,
      "loss": 0.0013,
      "step": 15516
    },
    {
      "epoch": 9.9277031349968,
      "grad_norm": 0.10070615870601278,
      "learning_rate": 2.7412654869329867e-09,
      "loss": 0.0008,
      "step": 15517
    },
    {
      "epoch": 9.928342930262316,
      "grad_norm": 0.1413029172593261,
      "learning_rate": 2.6929643631123938e-09,
      "loss": 0.0011,
      "step": 15518
    },
    {
      "epoch": 9.928982725527831,
      "grad_norm": 0.08037696769008454,
      "learning_rate": 2.6450925058718603e-09,
      "loss": 0.0007,
      "step": 15519
    },
    {
      "epoch": 9.929622520793346,
      "grad_norm": 0.09626424033979411,
      "learning_rate": 2.5976499172675198e-09,
      "loss": 0.0007,
      "step": 15520
    },
    {
      "epoch": 9.930262316058862,
      "grad_norm": 0.17788371294491476,
      "learning_rate": 2.550636599335521e-09,
      "loss": 0.0011,
      "step": 15521
    },
    {
      "epoch": 9.930902111324377,
      "grad_norm": 0.18633957852489996,
      "learning_rate": 2.50405255409536e-09,
      "loss": 0.0015,
      "step": 15522
    },
    {
      "epoch": 9.93154190658989,
      "grad_norm": 0.16253986631537296,
      "learning_rate": 2.457897783547658e-09,
      "loss": 0.0011,
      "step": 15523
    },
    {
      "epoch": 9.932181701855406,
      "grad_norm": 0.12686956709919175,
      "learning_rate": 2.4121722896730536e-09,
      "loss": 0.0011,
      "step": 15524
    },
    {
      "epoch": 9.932821497120921,
      "grad_norm": 0.10301416047894041,
      "learning_rate": 2.366876074435531e-09,
      "loss": 0.0008,
      "step": 15525
    },
    {
      "epoch": 9.933461292386436,
      "grad_norm": 0.08914574229961744,
      "learning_rate": 2.3220091397790913e-09,
      "loss": 0.0007,
      "step": 15526
    },
    {
      "epoch": 9.934101087651952,
      "grad_norm": 0.16431521661737447,
      "learning_rate": 2.2775714876321908e-09,
      "loss": 0.0013,
      "step": 15527
    },
    {
      "epoch": 9.934740882917467,
      "grad_norm": 0.13604922795985344,
      "learning_rate": 2.2335631199010834e-09,
      "loss": 0.0009,
      "step": 15528
    },
    {
      "epoch": 9.935380678182982,
      "grad_norm": 0.12034700104146584,
      "learning_rate": 2.189984038476478e-09,
      "loss": 0.0009,
      "step": 15529
    },
    {
      "epoch": 9.936020473448497,
      "grad_norm": 0.0944723391293663,
      "learning_rate": 2.1468342452291012e-09,
      "loss": 0.0007,
      "step": 15530
    },
    {
      "epoch": 9.93666026871401,
      "grad_norm": 0.10358130989191022,
      "learning_rate": 2.1041137420119152e-09,
      "loss": 0.0008,
      "step": 15531
    },
    {
      "epoch": 9.937300063979526,
      "grad_norm": 0.10669428723100842,
      "learning_rate": 2.061822530659008e-09,
      "loss": 0.0008,
      "step": 15532
    },
    {
      "epoch": 9.937939859245041,
      "grad_norm": 0.10855889937284663,
      "learning_rate": 2.0199606129867045e-09,
      "loss": 0.0008,
      "step": 15533
    },
    {
      "epoch": 9.938579654510557,
      "grad_norm": 0.06635109202757634,
      "learning_rate": 1.978527990792456e-09,
      "loss": 0.0006,
      "step": 15534
    },
    {
      "epoch": 9.939219449776072,
      "grad_norm": 0.07538908703849317,
      "learning_rate": 1.9375246658548396e-09,
      "loss": 0.0006,
      "step": 15535
    },
    {
      "epoch": 9.939859245041587,
      "grad_norm": 0.1293710725206695,
      "learning_rate": 1.8969506399346694e-09,
      "loss": 0.0008,
      "step": 15536
    },
    {
      "epoch": 9.940499040307103,
      "grad_norm": 0.2860845284097169,
      "learning_rate": 1.8568059147738848e-09,
      "loss": 0.001,
      "step": 15537
    },
    {
      "epoch": 9.941138835572616,
      "grad_norm": 0.10092646023617155,
      "learning_rate": 1.8170904920977727e-09,
      "loss": 0.0008,
      "step": 15538
    },
    {
      "epoch": 9.941778630838131,
      "grad_norm": 0.12725062468831153,
      "learning_rate": 1.7778043736094152e-09,
      "loss": 0.0009,
      "step": 15539
    },
    {
      "epoch": 9.942418426103647,
      "grad_norm": 0.11940363528298785,
      "learning_rate": 1.7389475609963512e-09,
      "loss": 0.0009,
      "step": 15540
    },
    {
      "epoch": 9.943058221369162,
      "grad_norm": 0.13211831045151906,
      "learning_rate": 1.7005200559283564e-09,
      "loss": 0.0008,
      "step": 15541
    },
    {
      "epoch": 9.943698016634677,
      "grad_norm": 0.11423105391090374,
      "learning_rate": 1.6625218600541114e-09,
      "loss": 0.0007,
      "step": 15542
    },
    {
      "epoch": 9.944337811900192,
      "grad_norm": 0.09386081197010807,
      "learning_rate": 1.6249529750056447e-09,
      "loss": 0.0007,
      "step": 15543
    },
    {
      "epoch": 9.944977607165708,
      "grad_norm": 0.1523708196615129,
      "learning_rate": 1.5878134023949999e-09,
      "loss": 0.0008,
      "step": 15544
    },
    {
      "epoch": 9.945617402431221,
      "grad_norm": 0.1855069505205077,
      "learning_rate": 1.5511031438186775e-09,
      "loss": 0.0012,
      "step": 15545
    },
    {
      "epoch": 9.946257197696736,
      "grad_norm": 0.14202077877584615,
      "learning_rate": 1.5148222008520842e-09,
      "loss": 0.001,
      "step": 15546
    },
    {
      "epoch": 9.946896992962252,
      "grad_norm": 0.17528099140790082,
      "learning_rate": 1.4789705750528627e-09,
      "loss": 0.0009,
      "step": 15547
    },
    {
      "epoch": 9.947536788227767,
      "grad_norm": 0.14502294811773142,
      "learning_rate": 1.4435482679608925e-09,
      "loss": 0.001,
      "step": 15548
    },
    {
      "epoch": 9.948176583493282,
      "grad_norm": 0.10020575122543451,
      "learning_rate": 1.4085552810971792e-09,
      "loss": 0.0008,
      "step": 15549
    },
    {
      "epoch": 9.948816378758798,
      "grad_norm": 0.14232655411886536,
      "learning_rate": 1.3739916159627441e-09,
      "loss": 0.001,
      "step": 15550
    },
    {
      "epoch": 9.949456174024313,
      "grad_norm": 0.1740279216782998,
      "learning_rate": 1.3398572740441762e-09,
      "loss": 0.0011,
      "step": 15551
    },
    {
      "epoch": 9.950095969289826,
      "grad_norm": 0.10384778663963674,
      "learning_rate": 1.3061522568058592e-09,
      "loss": 0.0008,
      "step": 15552
    },
    {
      "epoch": 9.950735764555342,
      "grad_norm": 0.1477930690369477,
      "learning_rate": 1.2728765656944141e-09,
      "loss": 0.001,
      "step": 15553
    },
    {
      "epoch": 9.951375559820857,
      "grad_norm": 0.07776594289542464,
      "learning_rate": 1.240030202139808e-09,
      "loss": 0.0008,
      "step": 15554
    },
    {
      "epoch": 9.952015355086372,
      "grad_norm": 0.24635176357592758,
      "learning_rate": 1.2076131675520241e-09,
      "loss": 0.0011,
      "step": 15555
    },
    {
      "epoch": 9.952655150351887,
      "grad_norm": 0.1609977657594626,
      "learning_rate": 1.175625463322172e-09,
      "loss": 0.0011,
      "step": 15556
    },
    {
      "epoch": 9.953294945617403,
      "grad_norm": 0.13370627509678715,
      "learning_rate": 1.1440670908247076e-09,
      "loss": 0.001,
      "step": 15557
    },
    {
      "epoch": 9.953934740882918,
      "grad_norm": 0.1107489687852359,
      "learning_rate": 1.112938051414103e-09,
      "loss": 0.0009,
      "step": 15558
    },
    {
      "epoch": 9.954574536148433,
      "grad_norm": 0.1347926316731202,
      "learning_rate": 1.0822383464281771e-09,
      "loss": 0.0008,
      "step": 15559
    },
    {
      "epoch": 9.955214331413947,
      "grad_norm": 0.08029798122924048,
      "learning_rate": 1.0519679771836545e-09,
      "loss": 0.0007,
      "step": 15560
    },
    {
      "epoch": 9.955854126679462,
      "grad_norm": 0.13008566929120594,
      "learning_rate": 1.022126944980606e-09,
      "loss": 0.0008,
      "step": 15561
    },
    {
      "epoch": 9.956493921944977,
      "grad_norm": 0.19356066598883556,
      "learning_rate": 9.927152511013393e-10,
      "loss": 0.0011,
      "step": 15562
    },
    {
      "epoch": 9.957133717210493,
      "grad_norm": 0.15415887930067335,
      "learning_rate": 9.637328968081783e-10,
      "loss": 0.0012,
      "step": 15563
    },
    {
      "epoch": 9.957773512476008,
      "grad_norm": 0.15319746255795483,
      "learning_rate": 9.351798833445724e-10,
      "loss": 0.0009,
      "step": 15564
    },
    {
      "epoch": 9.958413307741523,
      "grad_norm": 0.09747556933331128,
      "learning_rate": 9.070562119373182e-10,
      "loss": 0.0008,
      "step": 15565
    },
    {
      "epoch": 9.959053103007038,
      "grad_norm": 0.10652500265884972,
      "learning_rate": 8.793618837954487e-10,
      "loss": 0.0008,
      "step": 15566
    },
    {
      "epoch": 9.959692898272554,
      "grad_norm": 0.2592861344792776,
      "learning_rate": 8.520969001057922e-10,
      "loss": 0.0009,
      "step": 15567
    },
    {
      "epoch": 9.960332693538067,
      "grad_norm": 0.0922447796839637,
      "learning_rate": 8.252612620396339e-10,
      "loss": 0.0007,
      "step": 15568
    },
    {
      "epoch": 9.960972488803582,
      "grad_norm": 0.09513269591756687,
      "learning_rate": 7.988549707493853e-10,
      "loss": 0.0008,
      "step": 15569
    },
    {
      "epoch": 9.961612284069098,
      "grad_norm": 0.14384924293816034,
      "learning_rate": 7.728780273696945e-10,
      "loss": 0.0008,
      "step": 15570
    },
    {
      "epoch": 9.962252079334613,
      "grad_norm": 0.15706694743632882,
      "learning_rate": 7.473304330152253e-10,
      "loss": 0.001,
      "step": 15571
    },
    {
      "epoch": 9.962891874600128,
      "grad_norm": 0.16540547885707638,
      "learning_rate": 7.222121887839883e-10,
      "loss": 0.0012,
      "step": 15572
    },
    {
      "epoch": 9.963531669865644,
      "grad_norm": 0.21000367938065542,
      "learning_rate": 6.975232957528999e-10,
      "loss": 0.0014,
      "step": 15573
    },
    {
      "epoch": 9.964171465131159,
      "grad_norm": 0.14741123250778634,
      "learning_rate": 6.732637549822229e-10,
      "loss": 0.001,
      "step": 15574
    },
    {
      "epoch": 9.964811260396672,
      "grad_norm": 0.12395272304117823,
      "learning_rate": 6.49433567514457e-10,
      "loss": 0.0009,
      "step": 15575
    },
    {
      "epoch": 9.965451055662188,
      "grad_norm": 0.11637632213593244,
      "learning_rate": 6.260327343732275e-10,
      "loss": 0.0007,
      "step": 15576
    },
    {
      "epoch": 9.966090850927703,
      "grad_norm": 0.11687642707809254,
      "learning_rate": 6.030612565621763e-10,
      "loss": 0.0007,
      "step": 15577
    },
    {
      "epoch": 9.966730646193218,
      "grad_norm": 0.12316607385249728,
      "learning_rate": 5.805191350682915e-10,
      "loss": 0.0009,
      "step": 15578
    },
    {
      "epoch": 9.967370441458733,
      "grad_norm": 0.19831892191304937,
      "learning_rate": 5.584063708585774e-10,
      "loss": 0.0014,
      "step": 15579
    },
    {
      "epoch": 9.968010236724249,
      "grad_norm": 0.10650853546899061,
      "learning_rate": 5.367229648844951e-10,
      "loss": 0.0008,
      "step": 15580
    },
    {
      "epoch": 9.968650031989764,
      "grad_norm": 0.12058593594495791,
      "learning_rate": 5.154689180753014e-10,
      "loss": 0.0008,
      "step": 15581
    },
    {
      "epoch": 9.969289827255277,
      "grad_norm": 0.14881950217569156,
      "learning_rate": 4.946442313435995e-10,
      "loss": 0.0008,
      "step": 15582
    },
    {
      "epoch": 9.969929622520793,
      "grad_norm": 0.07519806414363943,
      "learning_rate": 4.742489055853394e-10,
      "loss": 0.0006,
      "step": 15583
    },
    {
      "epoch": 9.970569417786308,
      "grad_norm": 0.1501858047825114,
      "learning_rate": 4.542829416742667e-10,
      "loss": 0.001,
      "step": 15584
    },
    {
      "epoch": 9.971209213051823,
      "grad_norm": 0.1441040765086884,
      "learning_rate": 4.3474634046858366e-10,
      "loss": 0.0009,
      "step": 15585
    },
    {
      "epoch": 9.971849008317339,
      "grad_norm": 0.11696558198766997,
      "learning_rate": 4.156391028065088e-10,
      "loss": 0.0009,
      "step": 15586
    },
    {
      "epoch": 9.972488803582854,
      "grad_norm": 0.18170294181689023,
      "learning_rate": 3.9696122950960705e-10,
      "loss": 0.0015,
      "step": 15587
    },
    {
      "epoch": 9.973128598848369,
      "grad_norm": 0.10648888132364745,
      "learning_rate": 3.787127213794595e-10,
      "loss": 0.0007,
      "step": 15588
    },
    {
      "epoch": 9.973768394113884,
      "grad_norm": 0.12198431600423991,
      "learning_rate": 3.608935791987733e-10,
      "loss": 0.0007,
      "step": 15589
    },
    {
      "epoch": 9.974408189379398,
      "grad_norm": 0.1401118833559223,
      "learning_rate": 3.435038037336025e-10,
      "loss": 0.0009,
      "step": 15590
    },
    {
      "epoch": 9.975047984644913,
      "grad_norm": 0.17770334735164547,
      "learning_rate": 3.26543395731127e-10,
      "loss": 0.0012,
      "step": 15591
    },
    {
      "epoch": 9.975687779910428,
      "grad_norm": 0.19959518774817533,
      "learning_rate": 3.1001235591854307e-10,
      "loss": 0.0014,
      "step": 15592
    },
    {
      "epoch": 9.976327575175944,
      "grad_norm": 0.08132979618083393,
      "learning_rate": 2.9391068500528307e-10,
      "loss": 0.0006,
      "step": 15593
    },
    {
      "epoch": 9.976967370441459,
      "grad_norm": 0.19722683748360836,
      "learning_rate": 2.7823838368412626e-10,
      "loss": 0.0011,
      "step": 15594
    },
    {
      "epoch": 9.977607165706974,
      "grad_norm": 0.2228561922546228,
      "learning_rate": 2.629954526267575e-10,
      "loss": 0.0013,
      "step": 15595
    },
    {
      "epoch": 9.97824696097249,
      "grad_norm": 0.10863164940749422,
      "learning_rate": 2.4818189248820846e-10,
      "loss": 0.0008,
      "step": 15596
    },
    {
      "epoch": 9.978886756238003,
      "grad_norm": 0.23794514844707265,
      "learning_rate": 2.3379770390463686e-10,
      "loss": 0.001,
      "step": 15597
    },
    {
      "epoch": 9.979526551503518,
      "grad_norm": 0.2023766560977767,
      "learning_rate": 2.1984288749332672e-10,
      "loss": 0.0009,
      "step": 15598
    },
    {
      "epoch": 9.980166346769034,
      "grad_norm": 0.20781461456026543,
      "learning_rate": 2.0631744385379848e-10,
      "loss": 0.001,
      "step": 15599
    },
    {
      "epoch": 9.980806142034549,
      "grad_norm": 0.10112570905407356,
      "learning_rate": 1.9322137356780902e-10,
      "loss": 0.0008,
      "step": 15600
    },
    {
      "epoch": 9.981445937300064,
      "grad_norm": 0.13186668354030204,
      "learning_rate": 1.8055467719602094e-10,
      "loss": 0.001,
      "step": 15601
    },
    {
      "epoch": 9.98208573256558,
      "grad_norm": 0.12814762450237477,
      "learning_rate": 1.6831735528244353e-10,
      "loss": 0.0009,
      "step": 15602
    },
    {
      "epoch": 9.982725527831095,
      "grad_norm": 0.11918524549667422,
      "learning_rate": 1.565094083533225e-10,
      "loss": 0.0009,
      "step": 15603
    },
    {
      "epoch": 9.98336532309661,
      "grad_norm": 0.11905158684402262,
      "learning_rate": 1.451308369149196e-10,
      "loss": 0.0009,
      "step": 15604
    },
    {
      "epoch": 9.984005118362123,
      "grad_norm": 0.1197089994351157,
      "learning_rate": 1.341816414568431e-10,
      "loss": 0.0009,
      "step": 15605
    },
    {
      "epoch": 9.984644913627639,
      "grad_norm": 0.15132320112332542,
      "learning_rate": 1.2366182244871738e-10,
      "loss": 0.001,
      "step": 15606
    },
    {
      "epoch": 9.985284708893154,
      "grad_norm": 0.18185966616067184,
      "learning_rate": 1.13571380341293e-10,
      "loss": 0.001,
      "step": 15607
    },
    {
      "epoch": 9.98592450415867,
      "grad_norm": 0.12001644774459208,
      "learning_rate": 1.0391031556977737e-10,
      "loss": 0.0008,
      "step": 15608
    },
    {
      "epoch": 9.986564299424185,
      "grad_norm": 0.18170553394455,
      "learning_rate": 9.467862854717347e-11,
      "loss": 0.0018,
      "step": 15609
    },
    {
      "epoch": 9.9872040946897,
      "grad_norm": 0.07753590751504916,
      "learning_rate": 8.587631967094112e-11,
      "loss": 0.0007,
      "step": 15610
    },
    {
      "epoch": 9.987843889955215,
      "grad_norm": 0.11534037606983613,
      "learning_rate": 7.750338931855616e-11,
      "loss": 0.0007,
      "step": 15611
    },
    {
      "epoch": 9.988483685220729,
      "grad_norm": 0.15286649665323054,
      "learning_rate": 6.955983784973085e-11,
      "loss": 0.0011,
      "step": 15612
    },
    {
      "epoch": 9.989123480486244,
      "grad_norm": 0.1872715557462907,
      "learning_rate": 6.204566560530368e-11,
      "loss": 0.001,
      "step": 15613
    },
    {
      "epoch": 9.98976327575176,
      "grad_norm": 0.1921251764986967,
      "learning_rate": 5.49608729083495e-11,
      "loss": 0.0012,
      "step": 15614
    },
    {
      "epoch": 9.990403071017274,
      "grad_norm": 0.13791906358945866,
      "learning_rate": 4.830546006306947e-11,
      "loss": 0.001,
      "step": 15615
    },
    {
      "epoch": 9.99104286628279,
      "grad_norm": 0.19029673155449806,
      "learning_rate": 4.207942735479087e-11,
      "loss": 0.0008,
      "step": 15616
    },
    {
      "epoch": 9.991682661548305,
      "grad_norm": 0.10818470463544945,
      "learning_rate": 3.628277505107747e-11,
      "loss": 0.0009,
      "step": 15617
    },
    {
      "epoch": 9.99232245681382,
      "grad_norm": 0.16053096657215243,
      "learning_rate": 3.091550340061922e-11,
      "loss": 0.0009,
      "step": 15618
    },
    {
      "epoch": 9.992962252079334,
      "grad_norm": 0.2235529567175166,
      "learning_rate": 2.597761263434251e-11,
      "loss": 0.0009,
      "step": 15619
    },
    {
      "epoch": 9.993602047344849,
      "grad_norm": 0.18771753761025142,
      "learning_rate": 2.146910296429994e-11,
      "loss": 0.0013,
      "step": 15620
    },
    {
      "epoch": 9.994241842610364,
      "grad_norm": 0.257994555978058,
      "learning_rate": 1.7389974582560088e-11,
      "loss": 0.0013,
      "step": 15621
    },
    {
      "epoch": 9.99488163787588,
      "grad_norm": 0.21292153514284978,
      "learning_rate": 1.3740227666758643e-11,
      "loss": 0.0012,
      "step": 15622
    },
    {
      "epoch": 9.995521433141395,
      "grad_norm": 0.17050682116904498,
      "learning_rate": 1.0519862372326827e-11,
      "loss": 0.0011,
      "step": 15623
    },
    {
      "epoch": 9.99616122840691,
      "grad_norm": 0.12385434257523004,
      "learning_rate": 7.728878836932297e-12,
      "loss": 0.0009,
      "step": 15624
    },
    {
      "epoch": 9.996801023672425,
      "grad_norm": 0.09001091698497901,
      "learning_rate": 5.367277181589359e-12,
      "loss": 0.0008,
      "step": 15625
    },
    {
      "epoch": 9.99744081893794,
      "grad_norm": 0.08740432408436134,
      "learning_rate": 3.435057506218087e-12,
      "loss": 0.0008,
      "step": 15626
    },
    {
      "epoch": 9.998080614203454,
      "grad_norm": 0.10736607004065903,
      "learning_rate": 1.932219896305654e-12,
      "loss": 0.0008,
      "step": 15627
    },
    {
      "epoch": 9.99872040946897,
      "grad_norm": 0.17314923434424706,
      "learning_rate": 8.587644140245488e-13,
      "loss": 0.001,
      "step": 15628
    },
    {
      "epoch": 9.999360204734485,
      "grad_norm": 0.16073018211525064,
      "learning_rate": 2.1469110600413902e-13,
      "loss": 0.001,
      "step": 15629
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.08114614608436933,
      "learning_rate": 0.0,
      "loss": 0.0006,
      "step": 15630
    },
    {
      "epoch": 10.0,
      "step": 15630,
      "total_flos": 1163529710051328.0,
      "train_loss": 0.0029745390078046025,
      "train_runtime": 17453.2799,
      "train_samples_per_second": 114.592,
      "train_steps_per_second": 0.896
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 15630,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 50000.0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1163529710051328.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}