{
  "best_metric": 0.44540271162986755,
  "best_model_checkpoint": "miner_id_24/checkpoint-150",
  "epoch": 1.146131805157593,
  "eval_steps": 50,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0057306590257879654,
      "grad_norm": 1.0035889148712158,
      "learning_rate": 1e-05,
      "loss": 1.1815,
      "step": 1
    },
    {
      "epoch": 0.0057306590257879654,
      "eval_loss": 4.764197826385498,
      "eval_runtime": 31.1782,
      "eval_samples_per_second": 9.43,
      "eval_steps_per_second": 2.373,
      "step": 1
    },
    {
      "epoch": 0.011461318051575931,
      "grad_norm": 1.284098744392395,
      "learning_rate": 2e-05,
      "loss": 1.4512,
      "step": 2
    },
    {
      "epoch": 0.017191977077363897,
      "grad_norm": 1.4318172931671143,
      "learning_rate": 3e-05,
      "loss": 1.5606,
      "step": 3
    },
    {
      "epoch": 0.022922636103151862,
      "grad_norm": 1.313220500946045,
      "learning_rate": 4e-05,
      "loss": 1.3823,
      "step": 4
    },
    {
      "epoch": 0.02865329512893983,
      "grad_norm": 1.4467310905456543,
      "learning_rate": 5e-05,
      "loss": 1.4146,
      "step": 5
    },
    {
      "epoch": 0.034383954154727794,
      "grad_norm": 1.2124062776565552,
      "learning_rate": 6e-05,
      "loss": 0.9605,
      "step": 6
    },
    {
      "epoch": 0.04011461318051576,
      "grad_norm": 1.8278082609176636,
      "learning_rate": 7e-05,
      "loss": 1.0288,
      "step": 7
    },
    {
      "epoch": 0.045845272206303724,
      "grad_norm": 2.009721279144287,
      "learning_rate": 8e-05,
      "loss": 1.1359,
      "step": 8
    },
    {
      "epoch": 0.05157593123209169,
      "grad_norm": 2.1793124675750732,
      "learning_rate": 9e-05,
      "loss": 0.9358,
      "step": 9
    },
    {
      "epoch": 0.05730659025787966,
      "grad_norm": 19.282501220703125,
      "learning_rate": 0.0001,
      "loss": 2.2681,
      "step": 10
    },
    {
      "epoch": 0.06303724928366762,
      "grad_norm": 33.58476257324219,
      "learning_rate": 9.999316524962345e-05,
      "loss": 4.5587,
      "step": 11
    },
    {
      "epoch": 0.06876790830945559,
      "grad_norm": 15.062165260314941,
      "learning_rate": 9.997266286704631e-05,
      "loss": 1.4257,
      "step": 12
    },
    {
      "epoch": 0.07449856733524356,
      "grad_norm": 12.21485710144043,
      "learning_rate": 9.993849845741524e-05,
      "loss": 0.6596,
      "step": 13
    },
    {
      "epoch": 0.08022922636103152,
      "grad_norm": 6.862950325012207,
      "learning_rate": 9.989068136093873e-05,
      "loss": 0.4768,
      "step": 14
    },
    {
      "epoch": 0.08595988538681948,
      "grad_norm": 39.041934967041016,
      "learning_rate": 9.98292246503335e-05,
      "loss": 1.4254,
      "step": 15
    },
    {
      "epoch": 0.09169054441260745,
      "grad_norm": 14.904204368591309,
      "learning_rate": 9.975414512725057e-05,
      "loss": 0.5324,
      "step": 16
    },
    {
      "epoch": 0.09742120343839542,
      "grad_norm": 64.56082916259766,
      "learning_rate": 9.966546331768191e-05,
      "loss": 13.9855,
      "step": 17
    },
    {
      "epoch": 0.10315186246418338,
      "grad_norm": 65.52576446533203,
      "learning_rate": 9.956320346634876e-05,
      "loss": 11.8833,
      "step": 18
    },
    {
      "epoch": 0.10888252148997135,
      "grad_norm": 61.4244384765625,
      "learning_rate": 9.944739353007344e-05,
      "loss": 6.7465,
      "step": 19
    },
    {
      "epoch": 0.11461318051575932,
      "grad_norm": 27.389549255371094,
      "learning_rate": 9.931806517013612e-05,
      "loss": 1.5388,
      "step": 20
    },
    {
      "epoch": 0.12034383954154727,
      "grad_norm": 2.302321434020996,
      "learning_rate": 9.917525374361912e-05,
      "loss": 0.92,
      "step": 21
    },
    {
      "epoch": 0.12607449856733524,
      "grad_norm": 2.0892953872680664,
      "learning_rate": 9.901899829374047e-05,
      "loss": 0.7421,
      "step": 22
    },
    {
      "epoch": 0.1318051575931232,
      "grad_norm": 1.3867814540863037,
      "learning_rate": 9.884934153917997e-05,
      "loss": 0.6724,
      "step": 23
    },
    {
      "epoch": 0.13753581661891118,
      "grad_norm": 1.2110521793365479,
      "learning_rate": 9.86663298624003e-05,
      "loss": 0.5989,
      "step": 24
    },
    {
      "epoch": 0.14326647564469913,
      "grad_norm": 0.9389761686325073,
      "learning_rate": 9.847001329696653e-05,
      "loss": 0.6053,
      "step": 25
    },
    {
      "epoch": 0.1489971346704871,
      "grad_norm": 1.0855313539505005,
      "learning_rate": 9.826044551386744e-05,
      "loss": 0.5802,
      "step": 26
    },
    {
      "epoch": 0.15472779369627507,
      "grad_norm": 1.1220835447311401,
      "learning_rate": 9.803768380684242e-05,
      "loss": 0.8231,
      "step": 27
    },
    {
      "epoch": 0.16045845272206305,
      "grad_norm": 1.0798989534378052,
      "learning_rate": 9.780178907671789e-05,
      "loss": 0.6332,
      "step": 28
    },
    {
      "epoch": 0.166189111747851,
      "grad_norm": 0.836507260799408,
      "learning_rate": 9.755282581475769e-05,
      "loss": 0.493,
      "step": 29
    },
    {
      "epoch": 0.17191977077363896,
      "grad_norm": 0.9351760149002075,
      "learning_rate": 9.729086208503174e-05,
      "loss": 0.4356,
      "step": 30
    },
    {
      "epoch": 0.17765042979942694,
      "grad_norm": 0.9819492101669312,
      "learning_rate": 9.701596950580806e-05,
      "loss": 0.4732,
      "step": 31
    },
    {
      "epoch": 0.1833810888252149,
      "grad_norm": 0.9373030662536621,
      "learning_rate": 9.672822322997305e-05,
      "loss": 0.534,
      "step": 32
    },
    {
      "epoch": 0.18911174785100288,
      "grad_norm": 1.21796452999115,
      "learning_rate": 9.642770192448536e-05,
      "loss": 0.5958,
      "step": 33
    },
    {
      "epoch": 0.19484240687679083,
      "grad_norm": 0.9990187883377075,
      "learning_rate": 9.611448774886924e-05,
      "loss": 0.5209,
      "step": 34
    },
    {
      "epoch": 0.20057306590257878,
      "grad_norm": 1.1247957944869995,
      "learning_rate": 9.578866633275288e-05,
      "loss": 0.6153,
      "step": 35
    },
    {
      "epoch": 0.20630372492836677,
      "grad_norm": 1.2250124216079712,
      "learning_rate": 9.545032675245813e-05,
      "loss": 0.624,
      "step": 36
    },
    {
      "epoch": 0.21203438395415472,
      "grad_norm": 1.1935553550720215,
      "learning_rate": 9.509956150664796e-05,
      "loss": 0.4949,
      "step": 37
    },
    {
      "epoch": 0.2177650429799427,
      "grad_norm": 0.9890780448913574,
      "learning_rate": 9.473646649103818e-05,
      "loss": 0.5411,
      "step": 38
    },
    {
      "epoch": 0.22349570200573066,
      "grad_norm": 1.2203408479690552,
      "learning_rate": 9.43611409721806e-05,
      "loss": 0.6145,
      "step": 39
    },
    {
      "epoch": 0.22922636103151864,
      "grad_norm": 0.9592453837394714,
      "learning_rate": 9.397368756032445e-05,
      "loss": 0.5206,
      "step": 40
    },
    {
      "epoch": 0.2349570200573066,
      "grad_norm": 1.0418148040771484,
      "learning_rate": 9.357421218136386e-05,
      "loss": 0.5794,
      "step": 41
    },
    {
      "epoch": 0.24068767908309455,
      "grad_norm": 1.1782774925231934,
      "learning_rate": 9.316282404787871e-05,
      "loss": 0.5522,
      "step": 42
    },
    {
      "epoch": 0.24641833810888253,
      "grad_norm": 1.3091051578521729,
      "learning_rate": 9.273963562927695e-05,
      "loss": 0.4101,
      "step": 43
    },
    {
      "epoch": 0.2521489971346705,
      "grad_norm": 1.1064131259918213,
      "learning_rate": 9.230476262104677e-05,
      "loss": 0.9251,
      "step": 44
    },
    {
      "epoch": 0.25787965616045844,
      "grad_norm": 0.7374259829521179,
      "learning_rate": 9.185832391312644e-05,
      "loss": 0.9514,
      "step": 45
    },
    {
      "epoch": 0.2636103151862464,
      "grad_norm": 0.6981614232063293,
      "learning_rate": 9.140044155740101e-05,
      "loss": 0.9857,
      "step": 46
    },
    {
      "epoch": 0.2693409742120344,
      "grad_norm": 0.7841881513595581,
      "learning_rate": 9.093124073433463e-05,
      "loss": 0.9988,
      "step": 47
    },
    {
      "epoch": 0.27507163323782235,
      "grad_norm": 0.7463127970695496,
      "learning_rate": 9.045084971874738e-05,
      "loss": 0.9756,
      "step": 48
    },
    {
      "epoch": 0.2808022922636103,
      "grad_norm": 0.7383707165718079,
      "learning_rate": 8.995939984474624e-05,
      "loss": 0.8701,
      "step": 49
    },
    {
      "epoch": 0.28653295128939826,
      "grad_norm": 0.8134977221488953,
      "learning_rate": 8.945702546981969e-05,
      "loss": 0.7029,
      "step": 50
    },
    {
      "epoch": 0.28653295128939826,
      "eval_loss": 0.9773313403129578,
      "eval_runtime": 32.0113,
      "eval_samples_per_second": 9.184,
      "eval_steps_per_second": 2.312,
      "step": 50
    },
    {
      "epoch": 0.2922636103151863,
      "grad_norm": 1.1332144737243652,
      "learning_rate": 8.894386393810563e-05,
      "loss": 0.7462,
      "step": 51
    },
    {
      "epoch": 0.2979942693409742,
      "grad_norm": 1.1906083822250366,
      "learning_rate": 8.842005554284296e-05,
      "loss": 0.718,
      "step": 52
    },
    {
      "epoch": 0.3037249283667622,
      "grad_norm": 5.363713264465332,
      "learning_rate": 8.788574348801675e-05,
      "loss": 0.9616,
      "step": 53
    },
    {
      "epoch": 0.30945558739255014,
      "grad_norm": 15.01010799407959,
      "learning_rate": 8.73410738492077e-05,
      "loss": 1.598,
      "step": 54
    },
    {
      "epoch": 0.3151862464183381,
      "grad_norm": 4.782071113586426,
      "learning_rate": 8.678619553365659e-05,
      "loss": 0.6581,
      "step": 55
    },
    {
      "epoch": 0.3209169054441261,
      "grad_norm": 12.04438304901123,
      "learning_rate": 8.622126023955446e-05,
      "loss": 0.6126,
      "step": 56
    },
    {
      "epoch": 0.32664756446991405,
      "grad_norm": 7.532559394836426,
      "learning_rate": 8.564642241456986e-05,
      "loss": 0.3906,
      "step": 57
    },
    {
      "epoch": 0.332378223495702,
      "grad_norm": 2.730405330657959,
      "learning_rate": 8.506183921362443e-05,
      "loss": 0.3392,
      "step": 58
    },
    {
      "epoch": 0.33810888252148996,
      "grad_norm": 2.608391523361206,
      "learning_rate": 8.44676704559283e-05,
      "loss": 0.3518,
      "step": 59
    },
    {
      "epoch": 0.3438395415472779,
      "grad_norm": 17.62820816040039,
      "learning_rate": 8.386407858128706e-05,
      "loss": 0.6,
      "step": 60
    },
    {
      "epoch": 0.3495702005730659,
      "grad_norm": 3.376549482345581,
      "learning_rate": 8.32512286056924e-05,
      "loss": 0.8442,
      "step": 61
    },
    {
      "epoch": 0.3553008595988539,
      "grad_norm": 2.771148920059204,
      "learning_rate": 8.262928807620843e-05,
      "loss": 0.7439,
      "step": 62
    },
    {
      "epoch": 0.36103151862464183,
      "grad_norm": 1.198996901512146,
      "learning_rate": 8.199842702516583e-05,
      "loss": 0.4989,
      "step": 63
    },
    {
      "epoch": 0.3667621776504298,
      "grad_norm": 1.346926212310791,
      "learning_rate": 8.135881792367686e-05,
      "loss": 0.6355,
      "step": 64
    },
    {
      "epoch": 0.37249283667621774,
      "grad_norm": 1.0279195308685303,
      "learning_rate": 8.07106356344834e-05,
      "loss": 0.5252,
      "step": 65
    },
    {
      "epoch": 0.37822349570200575,
      "grad_norm": 1.1113489866256714,
      "learning_rate": 8.005405736415126e-05,
      "loss": 0.4989,
      "step": 66
    },
    {
      "epoch": 0.3839541547277937,
      "grad_norm": 1.2599353790283203,
      "learning_rate": 7.938926261462366e-05,
      "loss": 0.6129,
      "step": 67
    },
    {
      "epoch": 0.38968481375358166,
      "grad_norm": 1.3951442241668701,
      "learning_rate": 7.871643313414718e-05,
      "loss": 0.6764,
      "step": 68
    },
    {
      "epoch": 0.3954154727793696,
      "grad_norm": 0.8668970465660095,
      "learning_rate": 7.803575286758364e-05,
      "loss": 0.6259,
      "step": 69
    },
    {
      "epoch": 0.40114613180515757,
      "grad_norm": 0.9053806066513062,
      "learning_rate": 7.734740790612136e-05,
      "loss": 0.5455,
      "step": 70
    },
    {
      "epoch": 0.4068767908309456,
      "grad_norm": 1.2135175466537476,
      "learning_rate": 7.66515864363997e-05,
      "loss": 0.7378,
      "step": 71
    },
    {
      "epoch": 0.41260744985673353,
      "grad_norm": 0.8739562630653381,
      "learning_rate": 7.594847868906076e-05,
      "loss": 0.567,
      "step": 72
    },
    {
      "epoch": 0.4183381088825215,
      "grad_norm": 0.8985298871994019,
      "learning_rate": 7.52382768867422e-05,
      "loss": 0.5861,
      "step": 73
    },
    {
      "epoch": 0.42406876790830944,
      "grad_norm": 0.808602511882782,
      "learning_rate": 7.452117519152542e-05,
      "loss": 0.4248,
      "step": 74
    },
    {
      "epoch": 0.4297994269340974,
      "grad_norm": 0.9335209727287292,
      "learning_rate": 7.379736965185368e-05,
      "loss": 0.5191,
      "step": 75
    },
    {
      "epoch": 0.4355300859598854,
      "grad_norm": 0.7914756536483765,
      "learning_rate": 7.30670581489344e-05,
      "loss": 0.4788,
      "step": 76
    },
    {
      "epoch": 0.44126074498567336,
      "grad_norm": 1.1445316076278687,
      "learning_rate": 7.233044034264034e-05,
      "loss": 0.5361,
      "step": 77
    },
    {
      "epoch": 0.4469914040114613,
      "grad_norm": 0.8748629093170166,
      "learning_rate": 7.158771761692464e-05,
      "loss": 0.499,
      "step": 78
    },
    {
      "epoch": 0.45272206303724927,
      "grad_norm": 1.1068371534347534,
      "learning_rate": 7.083909302476453e-05,
      "loss": 0.5626,
      "step": 79
    },
    {
      "epoch": 0.4584527220630373,
      "grad_norm": 1.013913869857788,
      "learning_rate": 7.008477123264848e-05,
      "loss": 0.5701,
      "step": 80
    },
    {
      "epoch": 0.46418338108882523,
      "grad_norm": 0.8930913209915161,
      "learning_rate": 6.932495846462261e-05,
      "loss": 0.4473,
      "step": 81
    },
    {
      "epoch": 0.4699140401146132,
      "grad_norm": 0.8025099039077759,
      "learning_rate": 6.855986244591104e-05,
      "loss": 0.4006,
      "step": 82
    },
    {
      "epoch": 0.47564469914040114,
      "grad_norm": 1.2134878635406494,
      "learning_rate": 6.778969234612584e-05,
      "loss": 0.5306,
      "step": 83
    },
    {
      "epoch": 0.4813753581661891,
      "grad_norm": 1.2197167873382568,
      "learning_rate": 6.701465872208216e-05,
      "loss": 0.6729,
      "step": 84
    },
    {
      "epoch": 0.4871060171919771,
      "grad_norm": 1.0436406135559082,
      "learning_rate": 6.623497346023418e-05,
      "loss": 0.5937,
      "step": 85
    },
    {
      "epoch": 0.49283667621776506,
      "grad_norm": 1.3491209745407104,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.5954,
      "step": 86
    },
    {
      "epoch": 0.498567335243553,
      "grad_norm": 0.75426185131073,
      "learning_rate": 6.466250186922325e-05,
      "loss": 0.7653,
      "step": 87
    },
    {
      "epoch": 0.504297994269341,
      "grad_norm": 0.7373204231262207,
      "learning_rate": 6.387014543809223e-05,
      "loss": 0.8416,
      "step": 88
    },
    {
      "epoch": 0.5100286532951289,
      "grad_norm": 0.683611273765564,
      "learning_rate": 6.307399704769099e-05,
      "loss": 0.7598,
      "step": 89
    },
    {
      "epoch": 0.5157593123209169,
      "grad_norm": 0.7615026831626892,
      "learning_rate": 6.227427435703997e-05,
      "loss": 0.8549,
      "step": 90
    },
    {
      "epoch": 0.5214899713467048,
      "grad_norm": 0.7129623293876648,
      "learning_rate": 6.147119600233758e-05,
      "loss": 0.6896,
      "step": 91
    },
    {
      "epoch": 0.5272206303724928,
      "grad_norm": 0.7031312584877014,
      "learning_rate": 6.066498153718735e-05,
      "loss": 0.7698,
      "step": 92
    },
    {
      "epoch": 0.5329512893982808,
      "grad_norm": 0.8572548031806946,
      "learning_rate": 5.985585137257401e-05,
      "loss": 0.5967,
      "step": 93
    },
    {
      "epoch": 0.5386819484240688,
      "grad_norm": 1.1666706800460815,
      "learning_rate": 5.90440267166055e-05,
      "loss": 0.5839,
      "step": 94
    },
    {
      "epoch": 0.5444126074498568,
      "grad_norm": 1.0940685272216797,
      "learning_rate": 5.8229729514036705e-05,
      "loss": 0.5916,
      "step": 95
    },
    {
      "epoch": 0.5501432664756447,
      "grad_norm": 1.2334367036819458,
      "learning_rate": 5.74131823855921e-05,
      "loss": 0.4533,
      "step": 96
    },
    {
      "epoch": 0.5558739255014327,
      "grad_norm": 5.839780807495117,
      "learning_rate": 5.6594608567103456e-05,
      "loss": 0.4335,
      "step": 97
    },
    {
      "epoch": 0.5616045845272206,
      "grad_norm": 1.2165459394454956,
      "learning_rate": 5.577423184847932e-05,
      "loss": 0.3048,
      "step": 98
    },
    {
      "epoch": 0.5673352435530086,
      "grad_norm": 4.578935623168945,
      "learning_rate": 5.495227651252315e-05,
      "loss": 0.3554,
      "step": 99
    },
    {
      "epoch": 0.5730659025787965,
      "grad_norm": 4.122792720794678,
      "learning_rate": 5.4128967273616625e-05,
      "loss": 0.4068,
      "step": 100
    },
    {
      "epoch": 0.5730659025787965,
      "eval_loss": 0.4951989948749542,
      "eval_runtime": 31.4583,
      "eval_samples_per_second": 9.346,
      "eval_steps_per_second": 2.352,
      "step": 100
    },
    {
      "epoch": 0.5787965616045845,
      "grad_norm": 4.824379920959473,
      "learning_rate": 5.330452921628497e-05,
      "loss": 0.3382,
      "step": 101
    },
    {
      "epoch": 0.5845272206303725,
      "grad_norm": 1.6230909824371338,
      "learning_rate": 5.247918773366112e-05,
      "loss": 0.3311,
      "step": 102
    },
    {
      "epoch": 0.5902578796561605,
      "grad_norm": 2.886110544204712,
      "learning_rate": 5.165316846586541e-05,
      "loss": 0.3368,
      "step": 103
    },
    {
      "epoch": 0.5959885386819485,
      "grad_norm": 1.2518469095230103,
      "learning_rate": 5.0826697238317935e-05,
      "loss": 0.5906,
      "step": 104
    },
    {
      "epoch": 0.6017191977077364,
      "grad_norm": 1.0160014629364014,
      "learning_rate": 5e-05,
      "loss": 0.4486,
      "step": 105
    },
    {
      "epoch": 0.6074498567335244,
      "grad_norm": 0.7256372570991516,
      "learning_rate": 4.917330276168208e-05,
      "loss": 0.3544,
      "step": 106
    },
    {
      "epoch": 0.6131805157593123,
      "grad_norm": 0.9115315079689026,
      "learning_rate": 4.834683153413459e-05,
      "loss": 0.5429,
      "step": 107
    },
    {
      "epoch": 0.6189111747851003,
      "grad_norm": 0.9004347324371338,
      "learning_rate": 4.7520812266338885e-05,
      "loss": 0.6374,
      "step": 108
    },
    {
      "epoch": 0.6246418338108882,
      "grad_norm": 0.8389296531677246,
      "learning_rate": 4.669547078371504e-05,
      "loss": 0.4257,
      "step": 109
    },
    {
      "epoch": 0.6303724928366762,
      "grad_norm": 1.0888113975524902,
      "learning_rate": 4.5871032726383386e-05,
      "loss": 0.4651,
      "step": 110
    },
    {
      "epoch": 0.6361031518624641,
      "grad_norm": 0.6187241673469543,
      "learning_rate": 4.504772348747687e-05,
      "loss": 0.4474,
      "step": 111
    },
    {
      "epoch": 0.6418338108882522,
      "grad_norm": 0.7013351917266846,
      "learning_rate": 4.4225768151520694e-05,
      "loss": 0.4362,
      "step": 112
    },
    {
      "epoch": 0.6475644699140402,
      "grad_norm": 1.002426266670227,
      "learning_rate": 4.3405391432896555e-05,
      "loss": 0.5148,
      "step": 113
    },
    {
      "epoch": 0.6532951289398281,
      "grad_norm": 1.0781811475753784,
      "learning_rate": 4.2586817614407895e-05,
      "loss": 0.3857,
      "step": 114
    },
    {
      "epoch": 0.6590257879656161,
      "grad_norm": 0.8566449880599976,
      "learning_rate": 4.17702704859633e-05,
      "loss": 0.4931,
      "step": 115
    },
    {
      "epoch": 0.664756446991404,
      "grad_norm": 0.8623971939086914,
      "learning_rate": 4.095597328339452e-05,
      "loss": 0.5999,
      "step": 116
    },
    {
      "epoch": 0.670487106017192,
      "grad_norm": 0.7777952551841736,
      "learning_rate": 4.0144148627425993e-05,
      "loss": 0.4104,
      "step": 117
    },
    {
      "epoch": 0.6762177650429799,
      "grad_norm": 0.8600062131881714,
      "learning_rate": 3.933501846281267e-05,
      "loss": 0.476,
      "step": 118
    },
    {
      "epoch": 0.6819484240687679,
      "grad_norm": 0.9463000297546387,
      "learning_rate": 3.852880399766243e-05,
      "loss": 0.57,
      "step": 119
    },
    {
      "epoch": 0.6876790830945558,
      "grad_norm": 0.824280858039856,
      "learning_rate": 3.772572564296005e-05,
      "loss": 0.4579,
      "step": 120
    },
    {
      "epoch": 0.6934097421203438,
      "grad_norm": 0.9153361320495605,
      "learning_rate": 3.6926002952309016e-05,
      "loss": 0.6332,
      "step": 121
    },
    {
      "epoch": 0.6991404011461319,
      "grad_norm": 0.9660727977752686,
      "learning_rate": 3.612985456190778e-05,
      "loss": 0.4948,
      "step": 122
    },
    {
      "epoch": 0.7048710601719198,
      "grad_norm": 0.9707295298576355,
      "learning_rate": 3.533749813077677e-05,
      "loss": 0.544,
      "step": 123
    },
    {
      "epoch": 0.7106017191977078,
      "grad_norm": 0.9052512645721436,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 0.5014,
      "step": 124
    },
    {
      "epoch": 0.7163323782234957,
      "grad_norm": 1.1523258686065674,
      "learning_rate": 3.3765026539765834e-05,
      "loss": 0.5987,
      "step": 125
    },
    {
      "epoch": 0.7220630372492837,
      "grad_norm": 1.40993070602417,
      "learning_rate": 3.298534127791785e-05,
      "loss": 0.5708,
      "step": 126
    },
    {
      "epoch": 0.7277936962750716,
      "grad_norm": 1.1037057638168335,
      "learning_rate": 3.221030765387417e-05,
      "loss": 0.4856,
      "step": 127
    },
    {
      "epoch": 0.7335243553008596,
      "grad_norm": 1.0423033237457275,
      "learning_rate": 3.144013755408895e-05,
      "loss": 0.4701,
      "step": 128
    },
    {
      "epoch": 0.7392550143266475,
      "grad_norm": 1.3933699131011963,
      "learning_rate": 3.0675041535377405e-05,
      "loss": 0.5281,
      "step": 129
    },
    {
      "epoch": 0.7449856733524355,
      "grad_norm": 0.6178534626960754,
      "learning_rate": 2.991522876735154e-05,
      "loss": 0.6392,
      "step": 130
    },
    {
      "epoch": 0.7507163323782235,
      "grad_norm": 0.6612771153450012,
      "learning_rate": 2.916090697523549e-05,
      "loss": 0.7174,
      "step": 131
    },
    {
      "epoch": 0.7564469914040115,
      "grad_norm": 0.6470524668693542,
      "learning_rate": 2.8412282383075363e-05,
      "loss": 0.6859,
      "step": 132
    },
    {
      "epoch": 0.7621776504297995,
      "grad_norm": 0.6631524562835693,
      "learning_rate": 2.766955965735968e-05,
      "loss": 0.7743,
      "step": 133
    },
    {
      "epoch": 0.7679083094555874,
      "grad_norm": 0.9311091899871826,
      "learning_rate": 2.693294185106562e-05,
      "loss": 0.7462,
      "step": 134
    },
    {
      "epoch": 0.7736389684813754,
      "grad_norm": 0.6406381130218506,
      "learning_rate": 2.6202630348146324e-05,
      "loss": 0.6433,
      "step": 135
    },
    {
      "epoch": 0.7793696275071633,
      "grad_norm": 1.0116075277328491,
      "learning_rate": 2.547882480847461e-05,
      "loss": 0.4622,
      "step": 136
    },
    {
      "epoch": 0.7851002865329513,
      "grad_norm": 1.2839752435684204,
      "learning_rate": 2.476172311325783e-05,
      "loss": 0.6618,
      "step": 137
    },
    {
      "epoch": 0.7908309455587392,
      "grad_norm": 1.2551158666610718,
      "learning_rate": 2.405152131093926e-05,
      "loss": 0.4768,
      "step": 138
    },
    {
      "epoch": 0.7965616045845272,
      "grad_norm": 7.313023090362549,
      "learning_rate": 2.3348413563600325e-05,
      "loss": 0.563,
      "step": 139
    },
    {
      "epoch": 0.8022922636103151,
      "grad_norm": 8.236926078796387,
      "learning_rate": 2.2652592093878666e-05,
      "loss": 0.5689,
      "step": 140
    },
    {
      "epoch": 0.8080229226361032,
      "grad_norm": 6.016308784484863,
      "learning_rate": 2.196424713241637e-05,
      "loss": 0.4673,
      "step": 141
    },
    {
      "epoch": 0.8137535816618912,
      "grad_norm": 1.0700188875198364,
      "learning_rate": 2.128356686585282e-05,
      "loss": 0.3485,
      "step": 142
    },
    {
      "epoch": 0.8194842406876791,
      "grad_norm": 4.893228530883789,
      "learning_rate": 2.061073738537635e-05,
      "loss": 0.414,
      "step": 143
    },
    {
      "epoch": 0.8252148997134671,
      "grad_norm": 3.841538190841675,
      "learning_rate": 1.9945942635848748e-05,
      "loss": 0.3559,
      "step": 144
    },
    {
      "epoch": 0.830945558739255,
      "grad_norm": 4.237437725067139,
      "learning_rate": 1.928936436551661e-05,
      "loss": 0.3312,
      "step": 145
    },
    {
      "epoch": 0.836676217765043,
      "grad_norm": 0.8320844173431396,
      "learning_rate": 1.8641182076323148e-05,
      "loss": 0.3532,
      "step": 146
    },
    {
      "epoch": 0.8424068767908309,
      "grad_norm": 0.8265153765678406,
      "learning_rate": 1.800157297483417e-05,
      "loss": 0.5201,
      "step": 147
    },
    {
      "epoch": 0.8481375358166189,
      "grad_norm": 1.0047736167907715,
      "learning_rate": 1.7370711923791567e-05,
      "loss": 0.4563,
      "step": 148
    },
    {
      "epoch": 0.8538681948424068,
      "grad_norm": 0.9188067317008972,
      "learning_rate": 1.6748771394307585e-05,
      "loss": 0.422,
      "step": 149
    },
    {
      "epoch": 0.8595988538681948,
      "grad_norm": 0.8439174294471741,
      "learning_rate": 1.6135921418712956e-05,
      "loss": 0.4918,
      "step": 150
    },
    {
      "epoch": 0.8595988538681948,
      "eval_loss": 0.44540271162986755,
      "eval_runtime": 31.7224,
      "eval_samples_per_second": 9.268,
      "eval_steps_per_second": 2.333,
      "step": 150
    },
    {
      "epoch": 0.8653295128939829,
      "grad_norm": 0.8964242935180664,
      "learning_rate": 1.553232954407171e-05,
      "loss": 0.4643,
      "step": 151
    },
    {
      "epoch": 0.8710601719197708,
      "grad_norm": 0.824439287185669,
      "learning_rate": 1.4938160786375572e-05,
      "loss": 0.3892,
      "step": 152
    },
    {
      "epoch": 0.8767908309455588,
      "grad_norm": 0.7776252627372742,
      "learning_rate": 1.435357758543015e-05,
      "loss": 0.4145,
      "step": 153
    },
    {
      "epoch": 0.8825214899713467,
      "grad_norm": 0.9371297359466553,
      "learning_rate": 1.3778739760445552e-05,
      "loss": 0.5634,
      "step": 154
    },
    {
      "epoch": 0.8882521489971347,
      "grad_norm": 1.0299814939498901,
      "learning_rate": 1.3213804466343421e-05,
      "loss": 0.4847,
      "step": 155
    },
    {
      "epoch": 0.8939828080229226,
      "grad_norm": 1.4246488809585571,
      "learning_rate": 1.2658926150792322e-05,
      "loss": 0.6393,
      "step": 156
    },
    {
      "epoch": 0.8997134670487106,
      "grad_norm": 0.7245753407478333,
      "learning_rate": 1.2114256511983274e-05,
      "loss": 0.4044,
      "step": 157
    },
    {
      "epoch": 0.9054441260744985,
      "grad_norm": 0.7005698680877686,
      "learning_rate": 1.157994445715706e-05,
      "loss": 0.4756,
      "step": 158
    },
    {
      "epoch": 0.9111747851002865,
      "grad_norm": 0.8856043219566345,
      "learning_rate": 1.1056136061894384e-05,
      "loss": 0.5241,
      "step": 159
    },
    {
      "epoch": 0.9169054441260746,
      "grad_norm": 0.7303268909454346,
      "learning_rate": 1.0542974530180327e-05,
      "loss": 0.4489,
      "step": 160
    },
    {
      "epoch": 0.9226361031518625,
      "grad_norm": 0.7562816739082336,
      "learning_rate": 1.0040600155253765e-05,
      "loss": 0.3839,
      "step": 161
    },
    {
      "epoch": 0.9283667621776505,
      "grad_norm": 0.8168535828590393,
      "learning_rate": 9.549150281252633e-06,
      "loss": 0.4504,
      "step": 162
    },
    {
      "epoch": 0.9340974212034384,
      "grad_norm": 1.0767041444778442,
      "learning_rate": 9.068759265665384e-06,
      "loss": 0.6333,
      "step": 163
    },
    {
      "epoch": 0.9398280802292264,
      "grad_norm": 1.0518041849136353,
      "learning_rate": 8.599558442598998e-06,
      "loss": 0.3752,
      "step": 164
    },
    {
      "epoch": 0.9455587392550143,
      "grad_norm": 1.162166953086853,
      "learning_rate": 8.141676086873572e-06,
      "loss": 0.4307,
      "step": 165
    },
    {
      "epoch": 0.9512893982808023,
      "grad_norm": 1.3328009843826294,
      "learning_rate": 7.695237378953223e-06,
      "loss": 0.5075,
      "step": 166
    },
    {
      "epoch": 0.9570200573065902,
      "grad_norm": 0.9067739248275757,
      "learning_rate": 7.260364370723044e-06,
      "loss": 0.4621,
      "step": 167
    },
    {
      "epoch": 0.9627507163323782,
      "grad_norm": 0.9629002213478088,
      "learning_rate": 6.837175952121306e-06,
      "loss": 0.4683,
      "step": 168
    },
    {
      "epoch": 0.9684813753581661,
      "grad_norm": 0.9616539478302002,
      "learning_rate": 6.425787818636131e-06,
      "loss": 0.5045,
      "step": 169
    },
    {
      "epoch": 0.9742120343839542,
      "grad_norm": 1.5788521766662598,
      "learning_rate": 6.026312439675552e-06,
      "loss": 0.353,
      "step": 170
    },
    {
      "epoch": 0.9799426934097422,
      "grad_norm": 1.116830825805664,
      "learning_rate": 5.6388590278194096e-06,
      "loss": 0.5412,
      "step": 171
    },
    {
      "epoch": 0.9856733524355301,
      "grad_norm": 1.3788104057312012,
      "learning_rate": 5.263533508961827e-06,
      "loss": 0.3874,
      "step": 172
    },
    {
      "epoch": 0.9914040114613181,
      "grad_norm": 0.8184030652046204,
      "learning_rate": 4.900438493352055e-06,
      "loss": 0.591,
      "step": 173
    },
    {
      "epoch": 0.997134670487106,
      "grad_norm": 0.849399209022522,
      "learning_rate": 4.549673247541875e-06,
      "loss": 0.353,
      "step": 174
    },
    {
      "epoch": 1.002865329512894,
      "grad_norm": 2.4538636207580566,
      "learning_rate": 4.2113336672471245e-06,
      "loss": 0.7994,
      "step": 175
    },
    {
      "epoch": 1.008595988538682,
      "grad_norm": 0.49595823884010315,
      "learning_rate": 3.885512251130763e-06,
      "loss": 0.5555,
      "step": 176
    },
    {
      "epoch": 1.0143266475644699,
      "grad_norm": 0.5643259882926941,
      "learning_rate": 3.5722980755146517e-06,
      "loss": 0.6956,
      "step": 177
    },
    {
      "epoch": 1.0200573065902578,
      "grad_norm": 0.8207282423973083,
      "learning_rate": 3.271776770026963e-06,
      "loss": 0.5886,
      "step": 178
    },
    {
      "epoch": 1.0257879656160458,
      "grad_norm": 0.6101379990577698,
      "learning_rate": 2.9840304941919415e-06,
      "loss": 0.7186,
      "step": 179
    },
    {
      "epoch": 1.0315186246418337,
      "grad_norm": 0.6199820041656494,
      "learning_rate": 2.7091379149682685e-06,
      "loss": 0.6548,
      "step": 180
    },
    {
      "epoch": 1.0372492836676217,
      "grad_norm": 0.6589621901512146,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 0.608,
      "step": 181
    },
    {
      "epoch": 1.0429799426934097,
      "grad_norm": 0.8309704661369324,
      "learning_rate": 2.1982109232821178e-06,
      "loss": 0.3727,
      "step": 182
    },
    {
      "epoch": 1.0487106017191976,
      "grad_norm": 0.7659175992012024,
      "learning_rate": 1.962316193157593e-06,
      "loss": 0.352,
      "step": 183
    },
    {
      "epoch": 1.0544412607449858,
      "grad_norm": 0.9337642788887024,
      "learning_rate": 1.7395544861325718e-06,
      "loss": 0.473,
      "step": 184
    },
    {
      "epoch": 1.0601719197707737,
      "grad_norm": 1.664469599723816,
      "learning_rate": 1.5299867030334814e-06,
      "loss": 0.1855,
      "step": 185
    },
    {
      "epoch": 1.0659025787965617,
      "grad_norm": 0.8137052059173584,
      "learning_rate": 1.333670137599713e-06,
      "loss": 0.2174,
      "step": 186
    },
    {
      "epoch": 1.0716332378223496,
      "grad_norm": 6.241861343383789,
      "learning_rate": 1.1506584608200367e-06,
      "loss": 0.4824,
      "step": 187
    },
    {
      "epoch": 1.0773638968481376,
      "grad_norm": 1.8748527765274048,
      "learning_rate": 9.810017062595322e-07,
      "loss": 0.2882,
      "step": 188
    },
    {
      "epoch": 1.0830945558739256,
      "grad_norm": 4.546536445617676,
      "learning_rate": 8.247462563808817e-07,
      "loss": 0.4016,
      "step": 189
    },
    {
      "epoch": 1.0888252148997135,
      "grad_norm": 3.190882444381714,
      "learning_rate": 6.819348298638839e-07,
      "loss": 0.3463,
      "step": 190
    },
    {
      "epoch": 1.0945558739255015,
      "grad_norm": 0.5799930691719055,
      "learning_rate": 5.526064699265753e-07,
      "loss": 0.2081,
      "step": 191
    },
    {
      "epoch": 1.1002865329512894,
      "grad_norm": 0.46898940205574036,
      "learning_rate": 4.367965336512403e-07,
      "loss": 0.1307,
      "step": 192
    },
    {
      "epoch": 1.1060171919770774,
      "grad_norm": 1.3466490507125854,
      "learning_rate": 3.3453668231809286e-07,
      "loss": 0.4705,
      "step": 193
    },
    {
      "epoch": 1.1117478510028653,
      "grad_norm": 0.7763919830322266,
      "learning_rate": 2.458548727494292e-07,
      "loss": 0.4452,
      "step": 194
    },
    {
      "epoch": 1.1174785100286533,
      "grad_norm": 0.6729264259338379,
      "learning_rate": 1.7077534966650766e-07,
      "loss": 0.2995,
      "step": 195
    },
    {
      "epoch": 1.1232091690544412,
      "grad_norm": 0.6211886405944824,
      "learning_rate": 1.0931863906127327e-07,
      "loss": 0.2793,
      "step": 196
    },
    {
      "epoch": 1.1289398280802292,
      "grad_norm": 0.6557769775390625,
      "learning_rate": 6.150154258476315e-08,
      "loss": 0.2858,
      "step": 197
    },
    {
      "epoch": 1.1346704871060171,
      "grad_norm": 0.7445409297943115,
      "learning_rate": 2.7337132953697554e-08,
      "loss": 0.3551,
      "step": 198
    },
    {
      "epoch": 1.140401146131805,
      "grad_norm": 0.6006184220314026,
      "learning_rate": 6.834750376549792e-09,
      "loss": 0.3054,
      "step": 199
    },
    {
      "epoch": 1.146131805157593,
      "grad_norm": 0.6899487972259521,
      "learning_rate": 0.0,
      "loss": 0.4284,
      "step": 200
    },
    {
      "epoch": 1.146131805157593,
      "eval_loss": 0.45417553186416626,
      "eval_runtime": 31.4756,
      "eval_samples_per_second": 9.341,
      "eval_steps_per_second": 2.351,
      "step": 200
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 1
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.341690643827917e+17,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}