{
  "best_metric": 1.3924267292022705,
  "best_model_checkpoint": "miner_id_24/checkpoint-400",
  "epoch": 0.2705444707473791,
  "eval_steps": 50,
  "global_step": 400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006763611768684477,
      "grad_norm": 0.11604132503271103,
      "learning_rate": 1e-05,
      "loss": 1.3722,
      "step": 1
    },
    {
      "epoch": 0.0006763611768684477,
      "eval_loss": 1.602122187614441,
      "eval_runtime": 58.3957,
      "eval_samples_per_second": 42.64,
      "eval_steps_per_second": 10.669,
      "step": 1
    },
    {
      "epoch": 0.0013527223537368955,
      "grad_norm": 0.1288396269083023,
      "learning_rate": 2e-05,
      "loss": 1.5532,
      "step": 2
    },
    {
      "epoch": 0.002029083530605343,
      "grad_norm": 0.1327093541622162,
      "learning_rate": 3e-05,
      "loss": 1.616,
      "step": 3
    },
    {
      "epoch": 0.002705444707473791,
      "grad_norm": 0.11062159389257431,
      "learning_rate": 4e-05,
      "loss": 1.467,
      "step": 4
    },
    {
      "epoch": 0.0033818058843422386,
      "grad_norm": 0.10954708606004715,
      "learning_rate": 5e-05,
      "loss": 1.4682,
      "step": 5
    },
    {
      "epoch": 0.004058167061210686,
      "grad_norm": 0.11840442568063736,
      "learning_rate": 6e-05,
      "loss": 1.5285,
      "step": 6
    },
    {
      "epoch": 0.0047345282380791345,
      "grad_norm": 0.11326661705970764,
      "learning_rate": 7e-05,
      "loss": 1.4841,
      "step": 7
    },
    {
      "epoch": 0.005410889414947582,
      "grad_norm": 0.10721534490585327,
      "learning_rate": 8e-05,
      "loss": 1.4474,
      "step": 8
    },
    {
      "epoch": 0.00608725059181603,
      "grad_norm": 0.10227765887975693,
      "learning_rate": 9e-05,
      "loss": 1.4565,
      "step": 9
    },
    {
      "epoch": 0.006763611768684477,
      "grad_norm": 0.09144428372383118,
      "learning_rate": 0.0001,
      "loss": 1.5008,
      "step": 10
    },
    {
      "epoch": 0.0074399729455529254,
      "grad_norm": 0.09755543619394302,
      "learning_rate": 9.99983777858264e-05,
      "loss": 1.6286,
      "step": 11
    },
    {
      "epoch": 0.008116334122421373,
      "grad_norm": 0.08014149963855743,
      "learning_rate": 9.999351124856874e-05,
      "loss": 1.5582,
      "step": 12
    },
    {
      "epoch": 0.00879269529928982,
      "grad_norm": 0.07479062676429749,
      "learning_rate": 9.998540070400966e-05,
      "loss": 1.5557,
      "step": 13
    },
    {
      "epoch": 0.009469056476158269,
      "grad_norm": 0.07556532323360443,
      "learning_rate": 9.997404667843075e-05,
      "loss": 1.5746,
      "step": 14
    },
    {
      "epoch": 0.010145417653026716,
      "grad_norm": 0.07913000881671906,
      "learning_rate": 9.995944990857849e-05,
      "loss": 1.4655,
      "step": 15
    },
    {
      "epoch": 0.010821778829895164,
      "grad_norm": 0.081476129591465,
      "learning_rate": 9.994161134161634e-05,
      "loss": 1.5327,
      "step": 16
    },
    {
      "epoch": 0.011498140006763611,
      "grad_norm": 0.07700520753860474,
      "learning_rate": 9.992053213506334e-05,
      "loss": 1.5462,
      "step": 17
    },
    {
      "epoch": 0.01217450118363206,
      "grad_norm": 0.08761648833751678,
      "learning_rate": 9.989621365671902e-05,
      "loss": 1.5837,
      "step": 18
    },
    {
      "epoch": 0.012850862360500507,
      "grad_norm": 0.10584254562854767,
      "learning_rate": 9.986865748457457e-05,
      "loss": 1.5232,
      "step": 19
    },
    {
      "epoch": 0.013527223537368955,
      "grad_norm": 0.08562006056308746,
      "learning_rate": 9.983786540671051e-05,
      "loss": 1.5188,
      "step": 20
    },
    {
      "epoch": 0.014203584714237404,
      "grad_norm": 0.08462590724229813,
      "learning_rate": 9.980383942118066e-05,
      "loss": 1.5743,
      "step": 21
    },
    {
      "epoch": 0.014879945891105851,
      "grad_norm": 0.07892792671918869,
      "learning_rate": 9.976658173588244e-05,
      "loss": 1.5606,
      "step": 22
    },
    {
      "epoch": 0.015556307067974298,
      "grad_norm": 0.06926483660936356,
      "learning_rate": 9.972609476841367e-05,
      "loss": 1.429,
      "step": 23
    },
    {
      "epoch": 0.016232668244842745,
      "grad_norm": 0.06770279258489609,
      "learning_rate": 9.968238114591566e-05,
      "loss": 1.564,
      "step": 24
    },
    {
      "epoch": 0.016909029421711193,
      "grad_norm": 0.06699114292860031,
      "learning_rate": 9.96354437049027e-05,
      "loss": 1.5362,
      "step": 25
    },
    {
      "epoch": 0.01758539059857964,
      "grad_norm": 0.0729045495390892,
      "learning_rate": 9.95852854910781e-05,
      "loss": 1.5552,
      "step": 26
    },
    {
      "epoch": 0.01826175177544809,
      "grad_norm": 0.07210293412208557,
      "learning_rate": 9.953190975913647e-05,
      "loss": 1.4896,
      "step": 27
    },
    {
      "epoch": 0.018938112952316538,
      "grad_norm": 0.06548850983381271,
      "learning_rate": 9.947531997255256e-05,
      "loss": 1.5583,
      "step": 28
    },
    {
      "epoch": 0.019614474129184985,
      "grad_norm": 0.06891623139381409,
      "learning_rate": 9.941551980335652e-05,
      "loss": 1.5435,
      "step": 29
    },
    {
      "epoch": 0.020290835306053433,
      "grad_norm": 0.06692694127559662,
      "learning_rate": 9.935251313189564e-05,
      "loss": 1.5053,
      "step": 30
    },
    {
      "epoch": 0.02096719648292188,
      "grad_norm": 0.07309979945421219,
      "learning_rate": 9.928630404658255e-05,
      "loss": 1.559,
      "step": 31
    },
    {
      "epoch": 0.021643557659790327,
      "grad_norm": 0.0623411126434803,
      "learning_rate": 9.921689684362989e-05,
      "loss": 1.4759,
      "step": 32
    },
    {
      "epoch": 0.022319918836658775,
      "grad_norm": 0.06196431815624237,
      "learning_rate": 9.914429602677162e-05,
      "loss": 1.436,
      "step": 33
    },
    {
      "epoch": 0.022996280013527222,
      "grad_norm": 0.06310637295246124,
      "learning_rate": 9.906850630697068e-05,
      "loss": 1.5004,
      "step": 34
    },
    {
      "epoch": 0.023672641190395673,
      "grad_norm": 0.06335233151912689,
      "learning_rate": 9.898953260211338e-05,
      "loss": 1.5498,
      "step": 35
    },
    {
      "epoch": 0.02434900236726412,
      "grad_norm": 0.06549704074859619,
      "learning_rate": 9.890738003669029e-05,
      "loss": 1.5757,
      "step": 36
    },
    {
      "epoch": 0.025025363544132567,
      "grad_norm": 0.06362990289926529,
      "learning_rate": 9.882205394146361e-05,
      "loss": 1.4404,
      "step": 37
    },
    {
      "epoch": 0.025701724721001015,
      "grad_norm": 0.065706267952919,
      "learning_rate": 9.87335598531214e-05,
      "loss": 1.5211,
      "step": 38
    },
    {
      "epoch": 0.026378085897869462,
      "grad_norm": 0.06316651403903961,
      "learning_rate": 9.864190351391822e-05,
      "loss": 1.4632,
      "step": 39
    },
    {
      "epoch": 0.02705444707473791,
      "grad_norm": 0.06697884202003479,
      "learning_rate": 9.85470908713026e-05,
      "loss": 1.5391,
      "step": 40
    },
    {
      "epoch": 0.027730808251606356,
      "grad_norm": 0.06642621010541916,
      "learning_rate": 9.844912807753104e-05,
      "loss": 1.4436,
      "step": 41
    },
    {
      "epoch": 0.028407169428474807,
      "grad_norm": 0.06622406840324402,
      "learning_rate": 9.834802148926882e-05,
      "loss": 1.5092,
      "step": 42
    },
    {
      "epoch": 0.029083530605343254,
      "grad_norm": 0.06932581961154938,
      "learning_rate": 9.824377766717759e-05,
      "loss": 1.4794,
      "step": 43
    },
    {
      "epoch": 0.029759891782211702,
      "grad_norm": 0.07035700976848602,
      "learning_rate": 9.813640337548954e-05,
      "loss": 1.3602,
      "step": 44
    },
    {
      "epoch": 0.03043625295908015,
      "grad_norm": 0.07323595881462097,
      "learning_rate": 9.802590558156862e-05,
      "loss": 1.5098,
      "step": 45
    },
    {
      "epoch": 0.031112614135948596,
      "grad_norm": 0.07331836223602295,
      "learning_rate": 9.791229145545831e-05,
      "loss": 1.4027,
      "step": 46
    },
    {
      "epoch": 0.03178897531281705,
      "grad_norm": 0.07488737255334854,
      "learning_rate": 9.779556836941645e-05,
      "loss": 1.4362,
      "step": 47
    },
    {
      "epoch": 0.03246533648968549,
      "grad_norm": 0.08538374304771423,
      "learning_rate": 9.767574389743682e-05,
      "loss": 1.4314,
      "step": 48
    },
    {
      "epoch": 0.03314169766655394,
      "grad_norm": 0.10251971334218979,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.4597,
      "step": 49
    },
    {
      "epoch": 0.033818058843422386,
      "grad_norm": 0.1345338076353073,
      "learning_rate": 9.742682209735727e-05,
      "loss": 1.4462,
      "step": 50
    },
    {
      "epoch": 0.033818058843422386,
      "eval_loss": 1.463585615158081,
      "eval_runtime": 58.5692,
      "eval_samples_per_second": 42.514,
      "eval_steps_per_second": 10.637,
      "step": 50
    },
    {
      "epoch": 0.034494420020290836,
      "grad_norm": 0.08610611408948898,
      "learning_rate": 9.729774092143627e-05,
      "loss": 1.243,
      "step": 51
    },
    {
      "epoch": 0.03517078119715928,
      "grad_norm": 0.09492066502571106,
      "learning_rate": 9.716559066288715e-05,
      "loss": 1.3992,
      "step": 52
    },
    {
      "epoch": 0.03584714237402773,
      "grad_norm": 0.08210044354200363,
      "learning_rate": 9.703037989675087e-05,
      "loss": 1.452,
      "step": 53
    },
    {
      "epoch": 0.03652350355089618,
      "grad_norm": 0.07704884558916092,
      "learning_rate": 9.689211739666023e-05,
      "loss": 1.321,
      "step": 54
    },
    {
      "epoch": 0.037199864727764625,
      "grad_norm": 0.07068493217229843,
      "learning_rate": 9.675081213427076e-05,
      "loss": 1.2485,
      "step": 55
    },
    {
      "epoch": 0.037876225904633076,
      "grad_norm": 0.07206089049577713,
      "learning_rate": 9.66064732786784e-05,
      "loss": 1.3866,
      "step": 56
    },
    {
      "epoch": 0.03855258708150152,
      "grad_norm": 0.06557848304510117,
      "learning_rate": 9.645911019582467e-05,
      "loss": 1.3046,
      "step": 57
    },
    {
      "epoch": 0.03922894825836997,
      "grad_norm": 0.0679512619972229,
      "learning_rate": 9.630873244788883e-05,
      "loss": 1.3663,
      "step": 58
    },
    {
      "epoch": 0.039905309435238415,
      "grad_norm": 0.06276343762874603,
      "learning_rate": 9.615534979266745e-05,
      "loss": 1.3611,
      "step": 59
    },
    {
      "epoch": 0.040581670612106865,
      "grad_norm": 0.06208965554833412,
      "learning_rate": 9.599897218294122e-05,
      "loss": 1.4392,
      "step": 60
    },
    {
      "epoch": 0.041258031788975316,
      "grad_norm": 0.05605030059814453,
      "learning_rate": 9.583960976582913e-05,
      "loss": 1.4283,
      "step": 61
    },
    {
      "epoch": 0.04193439296584376,
      "grad_norm": 0.0576203390955925,
      "learning_rate": 9.567727288213005e-05,
      "loss": 1.4846,
      "step": 62
    },
    {
      "epoch": 0.04261075414271221,
      "grad_norm": 0.056024856865406036,
      "learning_rate": 9.551197206565173e-05,
      "loss": 1.473,
      "step": 63
    },
    {
      "epoch": 0.043287115319580655,
      "grad_norm": 0.05627163499593735,
      "learning_rate": 9.534371804252728e-05,
      "loss": 1.4328,
      "step": 64
    },
    {
      "epoch": 0.043963476496449105,
      "grad_norm": 0.0586436428129673,
      "learning_rate": 9.517252173051911e-05,
      "loss": 1.493,
      "step": 65
    },
    {
      "epoch": 0.04463983767331755,
      "grad_norm": 0.058163754642009735,
      "learning_rate": 9.49983942383106e-05,
      "loss": 1.3742,
      "step": 66
    },
    {
      "epoch": 0.045316198850186,
      "grad_norm": 0.058487631380558014,
      "learning_rate": 9.482134686478519e-05,
      "loss": 1.4339,
      "step": 67
    },
    {
      "epoch": 0.045992560027054444,
      "grad_norm": 0.061149582266807556,
      "learning_rate": 9.464139109829321e-05,
      "loss": 1.3955,
      "step": 68
    },
    {
      "epoch": 0.046668921203922895,
      "grad_norm": 0.05813620612025261,
      "learning_rate": 9.445853861590647e-05,
      "loss": 1.3931,
      "step": 69
    },
    {
      "epoch": 0.047345282380791345,
      "grad_norm": 0.05958769470453262,
      "learning_rate": 9.42728012826605e-05,
      "loss": 1.4106,
      "step": 70
    },
    {
      "epoch": 0.04802164355765979,
      "grad_norm": 0.05971216782927513,
      "learning_rate": 9.408419115078471e-05,
      "loss": 1.4899,
      "step": 71
    },
    {
      "epoch": 0.04869800473452824,
      "grad_norm": 0.061666082590818405,
      "learning_rate": 9.389272045892024e-05,
      "loss": 1.4908,
      "step": 72
    },
    {
      "epoch": 0.049374365911396684,
      "grad_norm": 0.05941418558359146,
      "learning_rate": 9.36984016313259e-05,
      "loss": 1.4982,
      "step": 73
    },
    {
      "epoch": 0.050050727088265135,
      "grad_norm": 0.05943742021918297,
      "learning_rate": 9.350124727707197e-05,
      "loss": 1.4998,
      "step": 74
    },
    {
      "epoch": 0.05072708826513358,
      "grad_norm": 0.06307440251111984,
      "learning_rate": 9.330127018922194e-05,
      "loss": 1.387,
      "step": 75
    },
    {
      "epoch": 0.05140344944200203,
      "grad_norm": 0.05909983813762665,
      "learning_rate": 9.309848334400246e-05,
      "loss": 1.3985,
      "step": 76
    },
    {
      "epoch": 0.05207981061887048,
      "grad_norm": 0.061729796230793,
      "learning_rate": 9.289289989996133e-05,
      "loss": 1.4094,
      "step": 77
    },
    {
      "epoch": 0.052756171795738924,
      "grad_norm": 0.06409716606140137,
      "learning_rate": 9.268453319711363e-05,
      "loss": 1.4705,
      "step": 78
    },
    {
      "epoch": 0.053432532972607374,
      "grad_norm": 0.06077890843153,
      "learning_rate": 9.247339675607605e-05,
      "loss": 1.5073,
      "step": 79
    },
    {
      "epoch": 0.05410889414947582,
      "grad_norm": 0.06479129940271378,
      "learning_rate": 9.225950427718975e-05,
      "loss": 1.3732,
      "step": 80
    },
    {
      "epoch": 0.05478525532634427,
      "grad_norm": 0.06292920559644699,
      "learning_rate": 9.204286963963111e-05,
      "loss": 1.4566,
      "step": 81
    },
    {
      "epoch": 0.05546161650321271,
      "grad_norm": 0.06173981726169586,
      "learning_rate": 9.182350690051133e-05,
      "loss": 1.4232,
      "step": 82
    },
    {
      "epoch": 0.056137977680081164,
      "grad_norm": 0.06200195476412773,
      "learning_rate": 9.160143029396422e-05,
      "loss": 1.4592,
      "step": 83
    },
    {
      "epoch": 0.056814338856949614,
      "grad_norm": 0.0646771639585495,
      "learning_rate": 9.13766542302225e-05,
      "loss": 1.3876,
      "step": 84
    },
    {
      "epoch": 0.05749070003381806,
      "grad_norm": 0.06770598888397217,
      "learning_rate": 9.114919329468282e-05,
      "loss": 1.4498,
      "step": 85
    },
    {
      "epoch": 0.05816706121068651,
      "grad_norm": 0.06534601747989655,
      "learning_rate": 9.091906224695935e-05,
      "loss": 1.4767,
      "step": 86
    },
    {
      "epoch": 0.05884342238755495,
      "grad_norm": 0.06706451624631882,
      "learning_rate": 9.068627601992598e-05,
      "loss": 1.4541,
      "step": 87
    },
    {
      "epoch": 0.059519783564423404,
      "grad_norm": 0.06629147380590439,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.4449,
      "step": 88
    },
    {
      "epoch": 0.06019614474129185,
      "grad_norm": 0.07131273299455643,
      "learning_rate": 9.021279861989885e-05,
      "loss": 1.5306,
      "step": 89
    },
    {
      "epoch": 0.0608725059181603,
      "grad_norm": 0.07197068631649017,
      "learning_rate": 8.997213817017507e-05,
      "loss": 1.4657,
      "step": 90
    },
    {
      "epoch": 0.06154886709502874,
      "grad_norm": 0.07346373796463013,
      "learning_rate": 8.972888398568772e-05,
      "loss": 1.4437,
      "step": 91
    },
    {
      "epoch": 0.06222522827189719,
      "grad_norm": 0.0742584615945816,
      "learning_rate": 8.948305185085225e-05,
      "loss": 1.3633,
      "step": 92
    },
    {
      "epoch": 0.06290158944876564,
      "grad_norm": 0.07299409061670303,
      "learning_rate": 8.92346577173636e-05,
      "loss": 1.458,
      "step": 93
    },
    {
      "epoch": 0.0635779506256341,
      "grad_norm": 0.07521813362836838,
      "learning_rate": 8.898371770316111e-05,
      "loss": 1.5101,
      "step": 94
    },
    {
      "epoch": 0.06425431180250253,
      "grad_norm": 0.0773283913731575,
      "learning_rate": 8.873024809138272e-05,
      "loss": 1.4126,
      "step": 95
    },
    {
      "epoch": 0.06493067297937098,
      "grad_norm": 0.07289938628673553,
      "learning_rate": 8.847426532930831e-05,
      "loss": 1.4009,
      "step": 96
    },
    {
      "epoch": 0.06560703415623943,
      "grad_norm": 0.08150746673345566,
      "learning_rate": 8.821578602729242e-05,
      "loss": 1.4199,
      "step": 97
    },
    {
      "epoch": 0.06628339533310788,
      "grad_norm": 0.08864708989858627,
      "learning_rate": 8.795482695768658e-05,
      "loss": 1.3391,
      "step": 98
    },
    {
      "epoch": 0.06695975650997633,
      "grad_norm": 0.09325551986694336,
      "learning_rate": 8.769140505375085e-05,
      "loss": 1.4052,
      "step": 99
    },
    {
      "epoch": 0.06763611768684477,
      "grad_norm": 0.16029119491577148,
      "learning_rate": 8.742553740855506e-05,
      "loss": 1.5922,
      "step": 100
    },
    {
      "epoch": 0.06763611768684477,
      "eval_loss": 1.4291542768478394,
      "eval_runtime": 58.6048,
      "eval_samples_per_second": 42.488,
      "eval_steps_per_second": 10.631,
      "step": 100
    },
    {
      "epoch": 0.06831247886371322,
      "grad_norm": 0.0940331295132637,
      "learning_rate": 8.715724127386972e-05,
      "loss": 1.4529,
      "step": 101
    },
    {
      "epoch": 0.06898884004058167,
      "grad_norm": 0.08114993572235107,
      "learning_rate": 8.688653405904652e-05,
      "loss": 1.4573,
      "step": 102
    },
    {
      "epoch": 0.06966520121745012,
      "grad_norm": 0.07999607920646667,
      "learning_rate": 8.661343332988869e-05,
      "loss": 1.4589,
      "step": 103
    },
    {
      "epoch": 0.07034156239431856,
      "grad_norm": 0.0692276656627655,
      "learning_rate": 8.633795680751116e-05,
      "loss": 1.3457,
      "step": 104
    },
    {
      "epoch": 0.07101792357118701,
      "grad_norm": 0.07577517628669739,
      "learning_rate": 8.606012236719073e-05,
      "loss": 1.4778,
      "step": 105
    },
    {
      "epoch": 0.07169428474805546,
      "grad_norm": 0.06558407098054886,
      "learning_rate": 8.577994803720606e-05,
      "loss": 1.3133,
      "step": 106
    },
    {
      "epoch": 0.07237064592492391,
      "grad_norm": 0.060306984931230545,
      "learning_rate": 8.549745199766792e-05,
      "loss": 1.2479,
      "step": 107
    },
    {
      "epoch": 0.07304700710179236,
      "grad_norm": 0.059541743248701096,
      "learning_rate": 8.521265257933948e-05,
      "loss": 1.2768,
      "step": 108
    },
    {
      "epoch": 0.0737233682786608,
      "grad_norm": 0.06409274786710739,
      "learning_rate": 8.492556826244687e-05,
      "loss": 1.3834,
      "step": 109
    },
    {
      "epoch": 0.07439972945552925,
      "grad_norm": 0.0610245019197464,
      "learning_rate": 8.463621767547998e-05,
      "loss": 1.4205,
      "step": 110
    },
    {
      "epoch": 0.0750760906323977,
      "grad_norm": 0.05981508269906044,
      "learning_rate": 8.434461959398376e-05,
      "loss": 1.4558,
      "step": 111
    },
    {
      "epoch": 0.07575245180926615,
      "grad_norm": 0.05959734693169594,
      "learning_rate": 8.405079293933986e-05,
      "loss": 1.3477,
      "step": 112
    },
    {
      "epoch": 0.07642881298613459,
      "grad_norm": 0.06075485795736313,
      "learning_rate": 8.375475677753881e-05,
      "loss": 1.4209,
      "step": 113
    },
    {
      "epoch": 0.07710517416300304,
      "grad_norm": 0.06361914426088333,
      "learning_rate": 8.345653031794292e-05,
      "loss": 1.4434,
      "step": 114
    },
    {
      "epoch": 0.07778153533987149,
      "grad_norm": 0.06411296874284744,
      "learning_rate": 8.315613291203976e-05,
      "loss": 1.4421,
      "step": 115
    },
    {
      "epoch": 0.07845789651673994,
      "grad_norm": 0.06421005725860596,
      "learning_rate": 8.285358405218655e-05,
      "loss": 1.4707,
      "step": 116
    },
    {
      "epoch": 0.07913425769360839,
      "grad_norm": 0.06386066973209381,
      "learning_rate": 8.25489033703452e-05,
      "loss": 1.4649,
      "step": 117
    },
    {
      "epoch": 0.07981061887047683,
      "grad_norm": 0.06911002844572067,
      "learning_rate": 8.224211063680853e-05,
      "loss": 1.4346,
      "step": 118
    },
    {
      "epoch": 0.08048698004734528,
      "grad_norm": 0.06442833691835403,
      "learning_rate": 8.19332257589174e-05,
      "loss": 1.4159,
      "step": 119
    },
    {
      "epoch": 0.08116334122421373,
      "grad_norm": 0.06367367506027222,
      "learning_rate": 8.162226877976887e-05,
      "loss": 1.4468,
      "step": 120
    },
    {
      "epoch": 0.08183970240108218,
      "grad_norm": 0.06462843716144562,
      "learning_rate": 8.130925987691569e-05,
      "loss": 1.4516,
      "step": 121
    },
    {
      "epoch": 0.08251606357795063,
      "grad_norm": 0.06435953825712204,
      "learning_rate": 8.099421936105702e-05,
      "loss": 1.4225,
      "step": 122
    },
    {
      "epoch": 0.08319242475481907,
      "grad_norm": 0.06690596044063568,
      "learning_rate": 8.067716767472045e-05,
      "loss": 1.4001,
      "step": 123
    },
    {
      "epoch": 0.08386878593168752,
      "grad_norm": 0.06788211315870285,
      "learning_rate": 8.035812539093557e-05,
      "loss": 1.4972,
      "step": 124
    },
    {
      "epoch": 0.08454514710855597,
      "grad_norm": 0.07012511044740677,
      "learning_rate": 8.003711321189895e-05,
      "loss": 1.5512,
      "step": 125
    },
    {
      "epoch": 0.08522150828542442,
      "grad_norm": 0.0646083876490593,
      "learning_rate": 7.971415196763088e-05,
      "loss": 1.3902,
      "step": 126
    },
    {
      "epoch": 0.08589786946229286,
      "grad_norm": 0.0692325159907341,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.4749,
      "step": 127
    },
    {
      "epoch": 0.08657423063916131,
      "grad_norm": 0.06617575138807297,
      "learning_rate": 7.906246623448183e-05,
      "loss": 1.445,
      "step": 128
    },
    {
      "epoch": 0.08725059181602976,
      "grad_norm": 0.06672117859125137,
      "learning_rate": 7.873378403255419e-05,
      "loss": 1.3787,
      "step": 129
    },
    {
      "epoch": 0.08792695299289821,
      "grad_norm": 0.06823927909135818,
      "learning_rate": 7.840323733655778e-05,
      "loss": 1.4454,
      "step": 130
    },
    {
      "epoch": 0.08860331416976666,
      "grad_norm": 0.06965020298957825,
      "learning_rate": 7.807084759519405e-05,
      "loss": 1.4482,
      "step": 131
    },
    {
      "epoch": 0.0892796753466351,
      "grad_norm": 0.06927543133497238,
      "learning_rate": 7.773663637675694e-05,
      "loss": 1.4027,
      "step": 132
    },
    {
      "epoch": 0.08995603652350355,
      "grad_norm": 0.07070069015026093,
      "learning_rate": 7.740062536773352e-05,
      "loss": 1.5726,
      "step": 133
    },
    {
      "epoch": 0.090632397700372,
      "grad_norm": 0.07122541964054108,
      "learning_rate": 7.706283637139658e-05,
      "loss": 1.4538,
      "step": 134
    },
    {
      "epoch": 0.09130875887724045,
      "grad_norm": 0.0677112340927124,
      "learning_rate": 7.672329130639005e-05,
      "loss": 1.4222,
      "step": 135
    },
    {
      "epoch": 0.09198512005410889,
      "grad_norm": 0.07415631413459778,
      "learning_rate": 7.638201220530665e-05,
      "loss": 1.4891,
      "step": 136
    },
    {
      "epoch": 0.09266148123097734,
      "grad_norm": 0.07131144404411316,
      "learning_rate": 7.603902121325813e-05,
      "loss": 1.393,
      "step": 137
    },
    {
      "epoch": 0.09333784240784579,
      "grad_norm": 0.07452413439750671,
      "learning_rate": 7.569434058643844e-05,
      "loss": 1.4818,
      "step": 138
    },
    {
      "epoch": 0.09401420358471424,
      "grad_norm": 0.0726298838853836,
      "learning_rate": 7.534799269067953e-05,
      "loss": 1.3627,
      "step": 139
    },
    {
      "epoch": 0.09469056476158269,
      "grad_norm": 0.07628554105758667,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.3693,
      "step": 140
    },
    {
      "epoch": 0.09536692593845113,
      "grad_norm": 0.07212232053279877,
      "learning_rate": 7.465038509514688e-05,
      "loss": 1.338,
      "step": 141
    },
    {
      "epoch": 0.09604328711531958,
      "grad_norm": 0.07615429162979126,
      "learning_rate": 7.42991706621303e-05,
      "loss": 1.3357,
      "step": 142
    },
    {
      "epoch": 0.09671964829218803,
      "grad_norm": 0.08060069382190704,
      "learning_rate": 7.394637949075154e-05,
      "loss": 1.448,
      "step": 143
    },
    {
      "epoch": 0.09739600946905648,
      "grad_norm": 0.07998302578926086,
      "learning_rate": 7.35920344731241e-05,
      "loss": 1.46,
      "step": 144
    },
    {
      "epoch": 0.09807237064592493,
      "grad_norm": 0.0826953873038292,
      "learning_rate": 7.323615860218843e-05,
      "loss": 1.3485,
      "step": 145
    },
    {
      "epoch": 0.09874873182279337,
      "grad_norm": 0.08640425652265549,
      "learning_rate": 7.287877497021978e-05,
      "loss": 1.4754,
      "step": 146
    },
    {
      "epoch": 0.09942509299966182,
      "grad_norm": 0.0944252610206604,
      "learning_rate": 7.251990676732984e-05,
      "loss": 1.4108,
      "step": 147
    },
    {
      "epoch": 0.10010145417653027,
      "grad_norm": 0.09773879498243332,
      "learning_rate": 7.215957727996207e-05,
      "loss": 1.3524,
      "step": 148
    },
    {
      "epoch": 0.10077781535339872,
      "grad_norm": 0.10976506024599075,
      "learning_rate": 7.179780988938051e-05,
      "loss": 1.3647,
      "step": 149
    },
    {
      "epoch": 0.10145417653026716,
      "grad_norm": 0.17261682450771332,
      "learning_rate": 7.143462807015271e-05,
      "loss": 1.3822,
      "step": 150
    },
    {
      "epoch": 0.10145417653026716,
      "eval_loss": 1.4144548177719116,
      "eval_runtime": 58.5001,
      "eval_samples_per_second": 42.564,
      "eval_steps_per_second": 10.65,
      "step": 150
    },
    {
      "epoch": 0.10213053770713561,
      "grad_norm": 0.07265003770589828,
      "learning_rate": 7.107005538862646e-05,
      "loss": 1.2632,
      "step": 151
    },
    {
      "epoch": 0.10280689888400406,
      "grad_norm": 0.080800361931324,
      "learning_rate": 7.07041155014006e-05,
      "loss": 1.4302,
      "step": 152
    },
    {
      "epoch": 0.10348326006087251,
      "grad_norm": 0.07225494086742401,
      "learning_rate": 7.033683215379002e-05,
      "loss": 1.3344,
      "step": 153
    },
    {
      "epoch": 0.10415962123774096,
      "grad_norm": 0.0720028281211853,
      "learning_rate": 6.996822917828477e-05,
      "loss": 1.3406,
      "step": 154
    },
    {
      "epoch": 0.1048359824146094,
      "grad_norm": 0.06982496380805969,
      "learning_rate": 6.959833049300377e-05,
      "loss": 1.2951,
      "step": 155
    },
    {
      "epoch": 0.10551234359147785,
      "grad_norm": 0.0698671042919159,
      "learning_rate": 6.922716010014255e-05,
      "loss": 1.3759,
      "step": 156
    },
    {
      "epoch": 0.1061887047683463,
      "grad_norm": 0.06916831433773041,
      "learning_rate": 6.885474208441603e-05,
      "loss": 1.3699,
      "step": 157
    },
    {
      "epoch": 0.10686506594521475,
      "grad_norm": 0.06982008367776871,
      "learning_rate": 6.848110061149556e-05,
      "loss": 1.3838,
      "step": 158
    },
    {
      "epoch": 0.10754142712208319,
      "grad_norm": 0.06897356361150742,
      "learning_rate": 6.810625992644085e-05,
      "loss": 1.3406,
      "step": 159
    },
    {
      "epoch": 0.10821778829895164,
      "grad_norm": 0.06280040740966797,
      "learning_rate": 6.773024435212678e-05,
      "loss": 1.3497,
      "step": 160
    },
    {
      "epoch": 0.10889414947582009,
      "grad_norm": 0.06423673778772354,
      "learning_rate": 6.735307828766515e-05,
      "loss": 1.3508,
      "step": 161
    },
    {
      "epoch": 0.10957051065268854,
      "grad_norm": 0.06756166368722916,
      "learning_rate": 6.697478620682137e-05,
      "loss": 1.4382,
      "step": 162
    },
    {
      "epoch": 0.11024687182955699,
      "grad_norm": 0.06480084359645844,
      "learning_rate": 6.659539265642643e-05,
      "loss": 1.3933,
      "step": 163
    },
    {
      "epoch": 0.11092323300642543,
      "grad_norm": 0.06544915586709976,
      "learning_rate": 6.621492225478414e-05,
      "loss": 1.4351,
      "step": 164
    },
    {
      "epoch": 0.11159959418329388,
      "grad_norm": 0.06527773290872574,
      "learning_rate": 6.583339969007363e-05,
      "loss": 1.4121,
      "step": 165
    },
    {
      "epoch": 0.11227595536016233,
      "grad_norm": 0.06609731167554855,
      "learning_rate": 6.545084971874738e-05,
      "loss": 1.438,
      "step": 166
    },
    {
      "epoch": 0.11295231653703078,
      "grad_norm": 0.06664299219846725,
      "learning_rate": 6.506729716392481e-05,
      "loss": 1.5263,
      "step": 167
    },
    {
      "epoch": 0.11362867771389923,
      "grad_norm": 0.0701015368103981,
      "learning_rate": 6.468276691378155e-05,
      "loss": 1.4672,
      "step": 168
    },
    {
      "epoch": 0.11430503889076767,
      "grad_norm": 0.07150130718946457,
      "learning_rate": 6.429728391993446e-05,
      "loss": 1.4679,
      "step": 169
    },
    {
      "epoch": 0.11498140006763612,
      "grad_norm": 0.06744568049907684,
      "learning_rate": 6.391087319582264e-05,
      "loss": 1.4669,
      "step": 170
    },
    {
      "epoch": 0.11565776124450457,
      "grad_norm": 0.06979771703481674,
      "learning_rate": 6.35235598150842e-05,
      "loss": 1.4848,
      "step": 171
    },
    {
      "epoch": 0.11633412242137302,
      "grad_norm": 0.06854456663131714,
      "learning_rate": 6.313536890992935e-05,
      "loss": 1.4543,
      "step": 172
    },
    {
      "epoch": 0.11701048359824145,
      "grad_norm": 0.06629548966884613,
      "learning_rate": 6.274632566950967e-05,
      "loss": 1.4109,
      "step": 173
    },
    {
      "epoch": 0.1176868447751099,
      "grad_norm": 0.07091697305440903,
      "learning_rate": 6.235645533828349e-05,
      "loss": 1.3707,
      "step": 174
    },
    {
      "epoch": 0.11836320595197836,
      "grad_norm": 0.07082553952932358,
      "learning_rate": 6.19657832143779e-05,
      "loss": 1.3993,
      "step": 175
    },
    {
      "epoch": 0.11903956712884681,
      "grad_norm": 0.07125814259052277,
      "learning_rate": 6.157433464794716e-05,
      "loss": 1.4445,
      "step": 176
    },
    {
      "epoch": 0.11971592830571526,
      "grad_norm": 0.07349928468465805,
      "learning_rate": 6.118213503952779e-05,
      "loss": 1.4615,
      "step": 177
    },
    {
      "epoch": 0.1203922894825837,
      "grad_norm": 0.07393936812877655,
      "learning_rate": 6.078920983839031e-05,
      "loss": 1.4923,
      "step": 178
    },
    {
      "epoch": 0.12106865065945215,
      "grad_norm": 0.07127617299556732,
      "learning_rate": 6.0395584540887963e-05,
      "loss": 1.4849,
      "step": 179
    },
    {
      "epoch": 0.1217450118363206,
      "grad_norm": 0.0763372927904129,
      "learning_rate": 6.0001284688802226e-05,
      "loss": 1.5,
      "step": 180
    },
    {
      "epoch": 0.12242137301318905,
      "grad_norm": 0.07156749814748764,
      "learning_rate": 5.960633586768543e-05,
      "loss": 1.3908,
      "step": 181
    },
    {
      "epoch": 0.12309773419005748,
      "grad_norm": 0.07486128807067871,
      "learning_rate": 5.921076370520058e-05,
      "loss": 1.4367,
      "step": 182
    },
    {
      "epoch": 0.12377409536692593,
      "grad_norm": 0.0703478455543518,
      "learning_rate": 5.8814593869458455e-05,
      "loss": 1.3708,
      "step": 183
    },
    {
      "epoch": 0.12445045654379439,
      "grad_norm": 0.07301351428031921,
      "learning_rate": 5.841785206735192e-05,
      "loss": 1.4418,
      "step": 184
    },
    {
      "epoch": 0.12512681772066284,
      "grad_norm": 0.07492207735776901,
      "learning_rate": 5.8020564042888015e-05,
      "loss": 1.4297,
      "step": 185
    },
    {
      "epoch": 0.1258031788975313,
      "grad_norm": 0.07811646908521652,
      "learning_rate": 5.762275557551727e-05,
      "loss": 1.4119,
      "step": 186
    },
    {
      "epoch": 0.12647954007439974,
      "grad_norm": 0.07865872979164124,
      "learning_rate": 5.7224452478461064e-05,
      "loss": 1.4008,
      "step": 187
    },
    {
      "epoch": 0.1271559012512682,
      "grad_norm": 0.0770040899515152,
      "learning_rate": 5.682568059703659e-05,
      "loss": 1.4389,
      "step": 188
    },
    {
      "epoch": 0.1278322624281366,
      "grad_norm": 0.07645025849342346,
      "learning_rate": 5.642646580697973e-05,
      "loss": 1.3273,
      "step": 189
    },
    {
      "epoch": 0.12850862360500506,
      "grad_norm": 0.08479326963424683,
      "learning_rate": 5.602683401276615e-05,
      "loss": 1.4879,
      "step": 190
    },
    {
      "epoch": 0.1291849847818735,
      "grad_norm": 0.08738847821950912,
      "learning_rate": 5.562681114593028e-05,
      "loss": 1.4578,
      "step": 191
    },
    {
      "epoch": 0.12986134595874196,
      "grad_norm": 0.08168379962444305,
      "learning_rate": 5.522642316338268e-05,
      "loss": 1.375,
      "step": 192
    },
    {
      "epoch": 0.13053770713561041,
      "grad_norm": 0.08465580642223358,
      "learning_rate": 5.482569604572576e-05,
      "loss": 1.445,
      "step": 193
    },
    {
      "epoch": 0.13121406831247887,
      "grad_norm": 0.08332633227109909,
      "learning_rate": 5.442465579556793e-05,
      "loss": 1.3447,
      "step": 194
    },
    {
      "epoch": 0.13189042948934732,
      "grad_norm": 0.08931078761816025,
      "learning_rate": 5.402332843583631e-05,
      "loss": 1.4188,
      "step": 195
    },
    {
      "epoch": 0.13256679066621577,
      "grad_norm": 0.08860035240650177,
      "learning_rate": 5.3621740008088126e-05,
      "loss": 1.4052,
      "step": 196
    },
    {
      "epoch": 0.13324315184308422,
      "grad_norm": 0.09431397169828415,
      "learning_rate": 5.321991657082097e-05,
      "loss": 1.3552,
      "step": 197
    },
    {
      "epoch": 0.13391951301995267,
      "grad_norm": 0.09754069894552231,
      "learning_rate": 5.281788419778187e-05,
      "loss": 1.3225,
      "step": 198
    },
    {
      "epoch": 0.1345958741968211,
      "grad_norm": 0.10789535194635391,
      "learning_rate": 5.2415668976275355e-05,
      "loss": 1.3024,
      "step": 199
    },
    {
      "epoch": 0.13527223537368954,
      "grad_norm": 0.13092434406280518,
      "learning_rate": 5.201329700547076e-05,
      "loss": 1.3502,
      "step": 200
    },
    {
      "epoch": 0.13527223537368954,
      "eval_loss": 1.4045569896697998,
      "eval_runtime": 58.6172,
      "eval_samples_per_second": 42.479,
      "eval_steps_per_second": 10.628,
      "step": 200
    },
    {
      "epoch": 0.135948596550558,
      "grad_norm": 0.07242485135793686,
      "learning_rate": 5.161079439470866e-05,
      "loss": 1.2782,
      "step": 201
    },
    {
      "epoch": 0.13662495772742644,
      "grad_norm": 0.07775568962097168,
      "learning_rate": 5.1208187261806615e-05,
      "loss": 1.3799,
      "step": 202
    },
    {
      "epoch": 0.1373013189042949,
      "grad_norm": 0.07579399645328522,
      "learning_rate": 5.080550173136457e-05,
      "loss": 1.3196,
      "step": 203
    },
    {
      "epoch": 0.13797768008116335,
      "grad_norm": 0.07350926846265793,
      "learning_rate": 5.0402763933069496e-05,
      "loss": 1.3123,
      "step": 204
    },
    {
      "epoch": 0.1386540412580318,
      "grad_norm": 0.07234349101781845,
      "learning_rate": 5e-05,
      "loss": 1.3502,
      "step": 205
    },
    {
      "epoch": 0.13933040243490025,
      "grad_norm": 0.07092832773923874,
      "learning_rate": 4.9597236066930516e-05,
      "loss": 1.3384,
      "step": 206
    },
    {
      "epoch": 0.1400067636117687,
      "grad_norm": 0.06819634884595871,
      "learning_rate": 4.919449826863544e-05,
      "loss": 1.3531,
      "step": 207
    },
    {
      "epoch": 0.14068312478863712,
      "grad_norm": 0.06536248326301575,
      "learning_rate": 4.87918127381934e-05,
      "loss": 1.218,
      "step": 208
    },
    {
      "epoch": 0.14135948596550557,
      "grad_norm": 0.07125400751829147,
      "learning_rate": 4.8389205605291365e-05,
      "loss": 1.3146,
      "step": 209
    },
    {
      "epoch": 0.14203584714237402,
      "grad_norm": 0.06933542340993881,
      "learning_rate": 4.798670299452926e-05,
      "loss": 1.3522,
      "step": 210
    },
    {
      "epoch": 0.14271220831924247,
      "grad_norm": 0.06874727457761765,
      "learning_rate": 4.758433102372466e-05,
      "loss": 1.3805,
      "step": 211
    },
    {
      "epoch": 0.14338856949611092,
      "grad_norm": 0.07114125788211823,
      "learning_rate": 4.7182115802218126e-05,
      "loss": 1.3567,
      "step": 212
    },
    {
      "epoch": 0.14406493067297937,
      "grad_norm": 0.07075918465852737,
      "learning_rate": 4.678008342917903e-05,
      "loss": 1.4714,
      "step": 213
    },
    {
      "epoch": 0.14474129184984783,
      "grad_norm": 0.07062017172574997,
      "learning_rate": 4.6378259991911886e-05,
      "loss": 1.3805,
      "step": 214
    },
    {
      "epoch": 0.14541765302671628,
      "grad_norm": 0.0695713609457016,
      "learning_rate": 4.597667156416371e-05,
      "loss": 1.4541,
      "step": 215
    },
    {
      "epoch": 0.14609401420358473,
      "grad_norm": 0.06924286484718323,
      "learning_rate": 4.5575344204432084e-05,
      "loss": 1.4296,
      "step": 216
    },
    {
      "epoch": 0.14677037538045315,
      "grad_norm": 0.07131223380565643,
      "learning_rate": 4.5174303954274244e-05,
      "loss": 1.5286,
      "step": 217
    },
    {
      "epoch": 0.1474467365573216,
      "grad_norm": 0.06701776385307312,
      "learning_rate": 4.477357683661734e-05,
      "loss": 1.3561,
      "step": 218
    },
    {
      "epoch": 0.14812309773419005,
      "grad_norm": 0.07408411055803299,
      "learning_rate": 4.437318885406973e-05,
      "loss": 1.4165,
      "step": 219
    },
    {
      "epoch": 0.1487994589110585,
      "grad_norm": 0.07193459570407867,
      "learning_rate": 4.397316598723385e-05,
      "loss": 1.3685,
      "step": 220
    },
    {
      "epoch": 0.14947582008792695,
      "grad_norm": 0.07403208315372467,
      "learning_rate": 4.3573534193020274e-05,
      "loss": 1.4735,
      "step": 221
    },
    {
      "epoch": 0.1501521812647954,
      "grad_norm": 0.07124771922826767,
      "learning_rate": 4.317431940296343e-05,
      "loss": 1.3149,
      "step": 222
    },
    {
      "epoch": 0.15082854244166385,
      "grad_norm": 0.07504601776599884,
      "learning_rate": 4.277554752153895e-05,
      "loss": 1.4289,
      "step": 223
    },
    {
      "epoch": 0.1515049036185323,
      "grad_norm": 0.07184187322854996,
      "learning_rate": 4.237724442448273e-05,
      "loss": 1.39,
      "step": 224
    },
    {
      "epoch": 0.15218126479540076,
      "grad_norm": 0.07542978227138519,
      "learning_rate": 4.197943595711198e-05,
      "loss": 1.4706,
      "step": 225
    },
    {
      "epoch": 0.15285762597226918,
      "grad_norm": 0.0755477175116539,
      "learning_rate": 4.1582147932648074e-05,
      "loss": 1.4443,
      "step": 226
    },
    {
      "epoch": 0.15353398714913763,
      "grad_norm": 0.07688594609498978,
      "learning_rate": 4.118540613054156e-05,
      "loss": 1.418,
      "step": 227
    },
    {
      "epoch": 0.15421034832600608,
      "grad_norm": 0.07337923347949982,
      "learning_rate": 4.078923629479943e-05,
      "loss": 1.4865,
      "step": 228
    },
    {
      "epoch": 0.15488670950287453,
      "grad_norm": 0.08003071695566177,
      "learning_rate": 4.039366413231458e-05,
      "loss": 1.4069,
      "step": 229
    },
    {
      "epoch": 0.15556307067974298,
      "grad_norm": 0.07447373121976852,
      "learning_rate": 3.9998715311197785e-05,
      "loss": 1.3647,
      "step": 230
    },
    {
      "epoch": 0.15623943185661143,
      "grad_norm": 0.082440085709095,
      "learning_rate": 3.960441545911204e-05,
      "loss": 1.4943,
      "step": 231
    },
    {
      "epoch": 0.15691579303347988,
      "grad_norm": 0.07682335376739502,
      "learning_rate": 3.92107901616097e-05,
      "loss": 1.4059,
      "step": 232
    },
    {
      "epoch": 0.15759215421034833,
      "grad_norm": 0.07876202464103699,
      "learning_rate": 3.8817864960472236e-05,
      "loss": 1.4894,
      "step": 233
    },
    {
      "epoch": 0.15826851538721678,
      "grad_norm": 0.08159554749727249,
      "learning_rate": 3.842566535205286e-05,
      "loss": 1.4321,
      "step": 234
    },
    {
      "epoch": 0.1589448765640852,
      "grad_norm": 0.0825141966342926,
      "learning_rate": 3.803421678562213e-05,
      "loss": 1.4251,
      "step": 235
    },
    {
      "epoch": 0.15962123774095366,
      "grad_norm": 0.07901477813720703,
      "learning_rate": 3.764354466171652e-05,
      "loss": 1.3717,
      "step": 236
    },
    {
      "epoch": 0.1602975989178221,
      "grad_norm": 0.07721833884716034,
      "learning_rate": 3.725367433049033e-05,
      "loss": 1.4613,
      "step": 237
    },
    {
      "epoch": 0.16097396009469056,
      "grad_norm": 0.0815616324543953,
      "learning_rate": 3.6864631090070655e-05,
      "loss": 1.3866,
      "step": 238
    },
    {
      "epoch": 0.161650321271559,
      "grad_norm": 0.08109608292579651,
      "learning_rate": 3.6476440184915815e-05,
      "loss": 1.3556,
      "step": 239
    },
    {
      "epoch": 0.16232668244842746,
      "grad_norm": 0.08702927082777023,
      "learning_rate": 3.608912680417737e-05,
      "loss": 1.3502,
      "step": 240
    },
    {
      "epoch": 0.1630030436252959,
      "grad_norm": 0.08374108374118805,
      "learning_rate": 3.570271608006555e-05,
      "loss": 1.3121,
      "step": 241
    },
    {
      "epoch": 0.16367940480216436,
      "grad_norm": 0.08754076808691025,
      "learning_rate": 3.531723308621847e-05,
      "loss": 1.4123,
      "step": 242
    },
    {
      "epoch": 0.16435576597903281,
      "grad_norm": 0.08081318438053131,
      "learning_rate": 3.493270283607522e-05,
      "loss": 1.2896,
      "step": 243
    },
    {
      "epoch": 0.16503212715590126,
      "grad_norm": 0.0907411053776741,
      "learning_rate": 3.4549150281252636e-05,
      "loss": 1.3749,
      "step": 244
    },
    {
      "epoch": 0.1657084883327697,
      "grad_norm": 0.08883036673069,
      "learning_rate": 3.4166600309926387e-05,
      "loss": 1.3238,
      "step": 245
    },
    {
      "epoch": 0.16638484950963814,
      "grad_norm": 0.101780466735363,
      "learning_rate": 3.3785077745215873e-05,
      "loss": 1.4447,
      "step": 246
    },
    {
      "epoch": 0.1670612106865066,
      "grad_norm": 0.09471210092306137,
      "learning_rate": 3.340460734357359e-05,
      "loss": 1.3208,
      "step": 247
    },
    {
      "epoch": 0.16773757186337504,
      "grad_norm": 0.1030389666557312,
      "learning_rate": 3.3025213793178646e-05,
      "loss": 1.3621,
      "step": 248
    },
    {
      "epoch": 0.1684139330402435,
      "grad_norm": 0.11591000854969025,
      "learning_rate": 3.264692171233485e-05,
      "loss": 1.3342,
      "step": 249
    },
    {
      "epoch": 0.16909029421711194,
      "grad_norm": 0.17307457327842712,
      "learning_rate": 3.226975564787322e-05,
      "loss": 1.4818,
      "step": 250
    },
    {
      "epoch": 0.16909029421711194,
      "eval_loss": 1.3978846073150635,
      "eval_runtime": 58.5594,
      "eval_samples_per_second": 42.521,
      "eval_steps_per_second": 10.639,
      "step": 250
    },
    {
      "epoch": 0.1697666553939804,
      "grad_norm": 0.07893766462802887,
      "learning_rate": 3.189374007355917e-05,
      "loss": 1.3761,
      "step": 251
    },
    {
      "epoch": 0.17044301657084884,
      "grad_norm": 0.07456450909376144,
      "learning_rate": 3.151889938850445e-05,
      "loss": 1.3348,
      "step": 252
    },
    {
      "epoch": 0.1711193777477173,
      "grad_norm": 0.07382918149232864,
      "learning_rate": 3.114525791558398e-05,
      "loss": 1.3499,
      "step": 253
    },
    {
      "epoch": 0.17179573892458572,
      "grad_norm": 0.07121651619672775,
      "learning_rate": 3.0772839899857464e-05,
      "loss": 1.3402,
      "step": 254
    },
    {
      "epoch": 0.17247210010145417,
      "grad_norm": 0.07278232276439667,
      "learning_rate": 3.0401669506996256e-05,
      "loss": 1.3145,
      "step": 255
    },
    {
      "epoch": 0.17314846127832262,
      "grad_norm": 0.07363329082727432,
      "learning_rate": 3.003177082171523e-05,
      "loss": 1.3581,
      "step": 256
    },
    {
      "epoch": 0.17382482245519107,
      "grad_norm": 0.06870734691619873,
      "learning_rate": 2.9663167846209998e-05,
      "loss": 1.3378,
      "step": 257
    },
    {
      "epoch": 0.17450118363205952,
      "grad_norm": 0.06792139261960983,
      "learning_rate": 2.9295884498599414e-05,
      "loss": 1.3419,
      "step": 258
    },
    {
      "epoch": 0.17517754480892797,
      "grad_norm": 0.0735265463590622,
      "learning_rate": 2.8929944611373554e-05,
      "loss": 1.3432,
      "step": 259
    },
    {
      "epoch": 0.17585390598579642,
      "grad_norm": 0.07180746644735336,
      "learning_rate": 2.8565371929847284e-05,
      "loss": 1.3547,
      "step": 260
    },
    {
      "epoch": 0.17653026716266487,
      "grad_norm": 0.07569875568151474,
      "learning_rate": 2.8202190110619493e-05,
      "loss": 1.4529,
      "step": 261
    },
    {
      "epoch": 0.17720662833953332,
      "grad_norm": 0.07017538696527481,
      "learning_rate": 2.784042272003794e-05,
      "loss": 1.3556,
      "step": 262
    },
    {
      "epoch": 0.17788298951640175,
      "grad_norm": 0.07113169133663177,
      "learning_rate": 2.7480093232670158e-05,
      "loss": 1.4059,
      "step": 263
    },
    {
      "epoch": 0.1785593506932702,
      "grad_norm": 0.06814723461866379,
      "learning_rate": 2.712122502978024e-05,
      "loss": 1.3724,
      "step": 264
    },
    {
      "epoch": 0.17923571187013865,
      "grad_norm": 0.06991983950138092,
      "learning_rate": 2.6763841397811573e-05,
      "loss": 1.3626,
      "step": 265
    },
    {
      "epoch": 0.1799120730470071,
      "grad_norm": 0.07174625992774963,
      "learning_rate": 2.64079655268759e-05,
      "loss": 1.4458,
      "step": 266
    },
    {
      "epoch": 0.18058843422387555,
      "grad_norm": 0.07207131385803223,
      "learning_rate": 2.605362050924848e-05,
      "loss": 1.4582,
      "step": 267
    },
    {
      "epoch": 0.181264795400744,
      "grad_norm": 0.07133448123931885,
      "learning_rate": 2.57008293378697e-05,
      "loss": 1.4014,
      "step": 268
    },
    {
      "epoch": 0.18194115657761245,
      "grad_norm": 0.07459162175655365,
      "learning_rate": 2.534961490485313e-05,
      "loss": 1.5518,
      "step": 269
    },
    {
      "epoch": 0.1826175177544809,
      "grad_norm": 0.07072808593511581,
      "learning_rate": 2.500000000000001e-05,
      "loss": 1.4785,
      "step": 270
    },
    {
      "epoch": 0.18329387893134935,
      "grad_norm": 0.07045449316501617,
      "learning_rate": 2.4652007309320498e-05,
      "loss": 1.3619,
      "step": 271
    },
    {
      "epoch": 0.18397024010821778,
      "grad_norm": 0.07227825373411179,
      "learning_rate": 2.430565941356157e-05,
      "loss": 1.4089,
      "step": 272
    },
    {
      "epoch": 0.18464660128508623,
      "grad_norm": 0.0711086317896843,
      "learning_rate": 2.3960978786741877e-05,
      "loss": 1.4513,
      "step": 273
    },
    {
      "epoch": 0.18532296246195468,
      "grad_norm": 0.07261015474796295,
      "learning_rate": 2.361798779469336e-05,
      "loss": 1.4833,
      "step": 274
    },
    {
      "epoch": 0.18599932363882313,
      "grad_norm": 0.07312192767858505,
      "learning_rate": 2.3276708693609943e-05,
      "loss": 1.4728,
      "step": 275
    },
    {
      "epoch": 0.18667568481569158,
      "grad_norm": 0.07378038763999939,
      "learning_rate": 2.2937163628603435e-05,
      "loss": 1.4263,
      "step": 276
    },
    {
      "epoch": 0.18735204599256003,
      "grad_norm": 0.0779234915971756,
      "learning_rate": 2.259937463226651e-05,
      "loss": 1.3974,
      "step": 277
    },
    {
      "epoch": 0.18802840716942848,
      "grad_norm": 0.07699842005968094,
      "learning_rate": 2.2263363623243054e-05,
      "loss": 1.4611,
      "step": 278
    },
    {
      "epoch": 0.18870476834629693,
      "grad_norm": 0.07660985738039017,
      "learning_rate": 2.192915240480596e-05,
      "loss": 1.3248,
      "step": 279
    },
    {
      "epoch": 0.18938112952316538,
      "grad_norm": 0.0773986205458641,
      "learning_rate": 2.1596762663442218e-05,
      "loss": 1.4168,
      "step": 280
    },
    {
      "epoch": 0.1900574907000338,
      "grad_norm": 0.08041651546955109,
      "learning_rate": 2.1266215967445824e-05,
      "loss": 1.4745,
      "step": 281
    },
    {
      "epoch": 0.19073385187690226,
      "grad_norm": 0.07643985748291016,
      "learning_rate": 2.0937533765518187e-05,
      "loss": 1.428,
      "step": 282
    },
    {
      "epoch": 0.1914102130537707,
      "grad_norm": 0.07961355894804001,
      "learning_rate": 2.061073738537635e-05,
      "loss": 1.4152,
      "step": 283
    },
    {
      "epoch": 0.19208657423063916,
      "grad_norm": 0.0767185166478157,
      "learning_rate": 2.0285848032369137e-05,
      "loss": 1.3493,
      "step": 284
    },
    {
      "epoch": 0.1927629354075076,
      "grad_norm": 0.08084074407815933,
      "learning_rate": 1.996288678810105e-05,
      "loss": 1.4049,
      "step": 285
    },
    {
      "epoch": 0.19343929658437606,
      "grad_norm": 0.07845637947320938,
      "learning_rate": 1.9641874609064443e-05,
      "loss": 1.4237,
      "step": 286
    },
    {
      "epoch": 0.1941156577612445,
      "grad_norm": 0.08574547618627548,
      "learning_rate": 1.932283232527956e-05,
      "loss": 1.429,
      "step": 287
    },
    {
      "epoch": 0.19479201893811296,
      "grad_norm": 0.08614347875118256,
      "learning_rate": 1.9005780638942982e-05,
      "loss": 1.4629,
      "step": 288
    },
    {
      "epoch": 0.1954683801149814,
      "grad_norm": 0.08351199328899384,
      "learning_rate": 1.8690740123084316e-05,
      "loss": 1.3625,
      "step": 289
    },
    {
      "epoch": 0.19614474129184986,
      "grad_norm": 0.08555345237255096,
      "learning_rate": 1.837773122023114e-05,
      "loss": 1.3805,
      "step": 290
    },
    {
      "epoch": 0.19682110246871828,
      "grad_norm": 0.09179560840129852,
      "learning_rate": 1.8066774241082612e-05,
      "loss": 1.4212,
      "step": 291
    },
    {
      "epoch": 0.19749746364558673,
      "grad_norm": 0.08746284991502762,
      "learning_rate": 1.7757889363191483e-05,
      "loss": 1.3866,
      "step": 292
    },
    {
      "epoch": 0.19817382482245519,
      "grad_norm": 0.09664510190486908,
      "learning_rate": 1.745109662965481e-05,
      "loss": 1.4013,
      "step": 293
    },
    {
      "epoch": 0.19885018599932364,
      "grad_norm": 0.09385596960783005,
      "learning_rate": 1.714641594781347e-05,
      "loss": 1.4426,
      "step": 294
    },
    {
      "epoch": 0.1995265471761921,
      "grad_norm": 0.09728514403104782,
      "learning_rate": 1.684386708796025e-05,
      "loss": 1.5043,
      "step": 295
    },
    {
      "epoch": 0.20020290835306054,
      "grad_norm": 0.09565767645835876,
      "learning_rate": 1.6543469682057106e-05,
      "loss": 1.4168,
      "step": 296
    },
    {
      "epoch": 0.200879269529929,
      "grad_norm": 0.09778063744306564,
      "learning_rate": 1.62452432224612e-05,
      "loss": 1.4201,
      "step": 297
    },
    {
      "epoch": 0.20155563070679744,
      "grad_norm": 0.11353736370801926,
      "learning_rate": 1.5949207060660138e-05,
      "loss": 1.4805,
      "step": 298
    },
    {
      "epoch": 0.2022319918836659,
      "grad_norm": 0.1255834847688675,
      "learning_rate": 1.5655380406016235e-05,
      "loss": 1.3583,
      "step": 299
    },
    {
      "epoch": 0.2029083530605343,
      "grad_norm": 0.1988285779953003,
      "learning_rate": 1.536378232452003e-05,
      "loss": 1.3486,
      "step": 300
    },
    {
      "epoch": 0.2029083530605343,
      "eval_loss": 1.394815444946289,
      "eval_runtime": 58.5752,
      "eval_samples_per_second": 42.509,
      "eval_steps_per_second": 10.636,
      "step": 300
    },
    {
      "epoch": 0.20358471423740276,
      "grad_norm": 0.06902935355901718,
      "learning_rate": 1.5074431737553157e-05,
      "loss": 1.2703,
      "step": 301
    },
    {
      "epoch": 0.20426107541427121,
      "grad_norm": 0.07079865038394928,
      "learning_rate": 1.4787347420660541e-05,
      "loss": 1.2929,
      "step": 302
    },
    {
      "epoch": 0.20493743659113967,
      "grad_norm": 0.07172349840402603,
      "learning_rate": 1.4502548002332088e-05,
      "loss": 1.3659,
      "step": 303
    },
    {
      "epoch": 0.20561379776800812,
      "grad_norm": 0.06566619873046875,
      "learning_rate": 1.422005196279395e-05,
      "loss": 1.2871,
      "step": 304
    },
    {
      "epoch": 0.20629015894487657,
      "grad_norm": 0.06977768242359161,
      "learning_rate": 1.3939877632809278e-05,
      "loss": 1.3399,
      "step": 305
    },
    {
      "epoch": 0.20696652012174502,
      "grad_norm": 0.0717938095331192,
      "learning_rate": 1.3662043192488849e-05,
      "loss": 1.4212,
      "step": 306
    },
    {
      "epoch": 0.20764288129861347,
      "grad_norm": 0.06987138837575912,
      "learning_rate": 1.338656667011134e-05,
      "loss": 1.3077,
      "step": 307
    },
    {
      "epoch": 0.20831924247548192,
      "grad_norm": 0.07141832262277603,
      "learning_rate": 1.3113465940953495e-05,
      "loss": 1.3018,
      "step": 308
    },
    {
      "epoch": 0.20899560365235034,
      "grad_norm": 0.06794386357069016,
      "learning_rate": 1.2842758726130283e-05,
      "loss": 1.2765,
      "step": 309
    },
    {
      "epoch": 0.2096719648292188,
      "grad_norm": 0.07226680219173431,
      "learning_rate": 1.257446259144494e-05,
      "loss": 1.2678,
      "step": 310
    },
    {
      "epoch": 0.21034832600608724,
      "grad_norm": 0.06923844665288925,
      "learning_rate": 1.2308594946249163e-05,
      "loss": 1.3101,
      "step": 311
    },
    {
      "epoch": 0.2110246871829557,
      "grad_norm": 0.07099943608045578,
      "learning_rate": 1.204517304231343e-05,
      "loss": 1.3425,
      "step": 312
    },
    {
      "epoch": 0.21170104835982415,
      "grad_norm": 0.07071360945701599,
      "learning_rate": 1.178421397270758e-05,
      "loss": 1.3551,
      "step": 313
    },
    {
      "epoch": 0.2123774095366926,
      "grad_norm": 0.07278096675872803,
      "learning_rate": 1.1525734670691701e-05,
      "loss": 1.3774,
      "step": 314
    },
    {
      "epoch": 0.21305377071356105,
      "grad_norm": 0.07339213043451309,
      "learning_rate": 1.1269751908617277e-05,
      "loss": 1.4096,
      "step": 315
    },
    {
      "epoch": 0.2137301318904295,
      "grad_norm": 0.07134059816598892,
      "learning_rate": 1.1016282296838887e-05,
      "loss": 1.39,
      "step": 316
    },
    {
      "epoch": 0.21440649306729795,
      "grad_norm": 0.0759441927075386,
      "learning_rate": 1.0765342282636416e-05,
      "loss": 1.5035,
      "step": 317
    },
    {
      "epoch": 0.21508285424416637,
      "grad_norm": 0.0740458145737648,
      "learning_rate": 1.0516948149147754e-05,
      "loss": 1.4417,
      "step": 318
    },
    {
      "epoch": 0.21575921542103482,
      "grad_norm": 0.07227642089128494,
      "learning_rate": 1.0271116014312293e-05,
      "loss": 1.3583,
      "step": 319
    },
    {
      "epoch": 0.21643557659790327,
      "grad_norm": 0.07315854728221893,
      "learning_rate": 1.0027861829824952e-05,
      "loss": 1.4062,
      "step": 320
    },
    {
      "epoch": 0.21711193777477172,
      "grad_norm": 0.07590021938085556,
      "learning_rate": 9.787201380101157e-06,
      "loss": 1.3649,
      "step": 321
    },
    {
      "epoch": 0.21778829895164017,
      "grad_norm": 0.07322739064693451,
      "learning_rate": 9.549150281252633e-06,
      "loss": 1.4691,
      "step": 322
    },
    {
      "epoch": 0.21846466012850863,
      "grad_norm": 0.07656604051589966,
      "learning_rate": 9.313723980074018e-06,
      "loss": 1.5084,
      "step": 323
    },
    {
      "epoch": 0.21914102130537708,
      "grad_norm": 0.07473529130220413,
      "learning_rate": 9.080937753040646e-06,
      "loss": 1.4019,
      "step": 324
    },
    {
      "epoch": 0.21981738248224553,
      "grad_norm": 0.07312378287315369,
      "learning_rate": 8.850806705317183e-06,
      "loss": 1.3843,
      "step": 325
    },
    {
      "epoch": 0.22049374365911398,
      "grad_norm": 0.0723285973072052,
      "learning_rate": 8.623345769777514e-06,
      "loss": 1.4158,
      "step": 326
    },
    {
      "epoch": 0.22117010483598243,
      "grad_norm": 0.07800653576850891,
      "learning_rate": 8.398569706035792e-06,
      "loss": 1.4794,
      "step": 327
    },
    {
      "epoch": 0.22184646601285085,
      "grad_norm": 0.0782305896282196,
      "learning_rate": 8.176493099488663e-06,
      "loss": 1.368,
      "step": 328
    },
    {
      "epoch": 0.2225228271897193,
      "grad_norm": 0.07785584032535553,
      "learning_rate": 7.957130360368898e-06,
      "loss": 1.4589,
      "step": 329
    },
    {
      "epoch": 0.22319918836658775,
      "grad_norm": 0.07849667221307755,
      "learning_rate": 7.740495722810271e-06,
      "loss": 1.4204,
      "step": 330
    },
    {
      "epoch": 0.2238755495434562,
      "grad_norm": 0.07931456714868546,
      "learning_rate": 7.526603243923957e-06,
      "loss": 1.4728,
      "step": 331
    },
    {
      "epoch": 0.22455191072032465,
      "grad_norm": 0.07695943862199783,
      "learning_rate": 7.315466802886401e-06,
      "loss": 1.3767,
      "step": 332
    },
    {
      "epoch": 0.2252282718971931,
      "grad_norm": 0.07418181747198105,
      "learning_rate": 7.107100100038671e-06,
      "loss": 1.3109,
      "step": 333
    },
    {
      "epoch": 0.22590463307406156,
      "grad_norm": 0.08536362648010254,
      "learning_rate": 6.901516655997536e-06,
      "loss": 1.4548,
      "step": 334
    },
    {
      "epoch": 0.22658099425093,
      "grad_norm": 0.0794777050614357,
      "learning_rate": 6.698729810778065e-06,
      "loss": 1.4105,
      "step": 335
    },
    {
      "epoch": 0.22725735542779846,
      "grad_norm": 0.08513446897268295,
      "learning_rate": 6.498752722928042e-06,
      "loss": 1.4203,
      "step": 336
    },
    {
      "epoch": 0.22793371660466688,
      "grad_norm": 0.0815361961722374,
      "learning_rate": 6.301598368674105e-06,
      "loss": 1.4362,
      "step": 337
    },
    {
      "epoch": 0.22861007778153533,
      "grad_norm": 0.0861455649137497,
      "learning_rate": 6.107279541079769e-06,
      "loss": 1.4163,
      "step": 338
    },
    {
      "epoch": 0.22928643895840378,
      "grad_norm": 0.08055116981267929,
      "learning_rate": 5.915808849215304e-06,
      "loss": 1.3815,
      "step": 339
    },
    {
      "epoch": 0.22996280013527223,
      "grad_norm": 0.08424156159162521,
      "learning_rate": 5.727198717339511e-06,
      "loss": 1.3917,
      "step": 340
    },
    {
      "epoch": 0.23063916131214068,
      "grad_norm": 0.0870758518576622,
      "learning_rate": 5.54146138409355e-06,
      "loss": 1.4611,
      "step": 341
    },
    {
      "epoch": 0.23131552248900913,
      "grad_norm": 0.09209448844194412,
      "learning_rate": 5.358608901706802e-06,
      "loss": 1.4124,
      "step": 342
    },
    {
      "epoch": 0.23199188366587759,
      "grad_norm": 0.09123307466506958,
      "learning_rate": 5.178653135214812e-06,
      "loss": 1.3352,
      "step": 343
    },
    {
      "epoch": 0.23266824484274604,
      "grad_norm": 0.09511040896177292,
      "learning_rate": 5.001605761689398e-06,
      "loss": 1.4073,
      "step": 344
    },
    {
      "epoch": 0.2333446060196145,
      "grad_norm": 0.0982203483581543,
      "learning_rate": 4.827478269480895e-06,
      "loss": 1.4191,
      "step": 345
    },
    {
      "epoch": 0.2340209671964829,
      "grad_norm": 0.09550726413726807,
      "learning_rate": 4.65628195747273e-06,
      "loss": 1.4069,
      "step": 346
    },
    {
      "epoch": 0.23469732837335136,
      "grad_norm": 0.10381521284580231,
      "learning_rate": 4.488027934348271e-06,
      "loss": 1.3493,
      "step": 347
    },
    {
      "epoch": 0.2353736895502198,
      "grad_norm": 0.10688237845897675,
      "learning_rate": 4.322727117869951e-06,
      "loss": 1.4009,
      "step": 348
    },
    {
      "epoch": 0.23605005072708826,
      "grad_norm": 0.1287006139755249,
      "learning_rate": 4.16039023417088e-06,
      "loss": 1.3225,
      "step": 349
    },
    {
      "epoch": 0.2367264119039567,
      "grad_norm": 0.16477414965629578,
      "learning_rate": 4.001027817058789e-06,
      "loss": 1.1971,
      "step": 350
    },
    {
      "epoch": 0.2367264119039567,
      "eval_loss": 1.392931342124939,
      "eval_runtime": 58.606,
      "eval_samples_per_second": 42.487,
      "eval_steps_per_second": 10.63,
      "step": 350
    },
    {
      "epoch": 0.23740277308082516,
      "grad_norm": 0.07476108521223068,
      "learning_rate": 3.844650207332562e-06,
      "loss": 1.2627,
      "step": 351
    },
    {
      "epoch": 0.23807913425769361,
      "grad_norm": 0.07067526876926422,
      "learning_rate": 3.691267552111183e-06,
      "loss": 1.4219,
      "step": 352
    },
    {
      "epoch": 0.23875549543456207,
      "grad_norm": 0.06555365771055222,
      "learning_rate": 3.54088980417534e-06,
      "loss": 1.2736,
      "step": 353
    },
    {
      "epoch": 0.23943185661143052,
      "grad_norm": 0.07065010815858841,
      "learning_rate": 3.393526721321616e-06,
      "loss": 1.3441,
      "step": 354
    },
    {
      "epoch": 0.24010821778829894,
      "grad_norm": 0.066973976790905,
      "learning_rate": 3.249187865729264e-06,
      "loss": 1.3049,
      "step": 355
    },
    {
      "epoch": 0.2407845789651674,
      "grad_norm": 0.06522707641124725,
      "learning_rate": 3.1078826033397843e-06,
      "loss": 1.3121,
      "step": 356
    },
    {
      "epoch": 0.24146094014203584,
      "grad_norm": 0.06993697583675385,
      "learning_rate": 2.9696201032491434e-06,
      "loss": 1.3958,
      "step": 357
    },
    {
      "epoch": 0.2421373013189043,
      "grad_norm": 0.06835866719484329,
      "learning_rate": 2.8344093371128424e-06,
      "loss": 1.2915,
      "step": 358
    },
    {
      "epoch": 0.24281366249577274,
      "grad_norm": 0.06519079953432083,
      "learning_rate": 2.70225907856374e-06,
      "loss": 1.296,
      "step": 359
    },
    {
      "epoch": 0.2434900236726412,
      "grad_norm": 0.06798028200864792,
      "learning_rate": 2.573177902642726e-06,
      "loss": 1.3299,
      "step": 360
    },
    {
      "epoch": 0.24416638484950964,
      "grad_norm": 0.06898199021816254,
      "learning_rate": 2.4471741852423237e-06,
      "loss": 1.3754,
      "step": 361
    },
    {
      "epoch": 0.2448427460263781,
      "grad_norm": 0.07092057168483734,
      "learning_rate": 2.324256102563188e-06,
      "loss": 1.3248,
      "step": 362
    },
    {
      "epoch": 0.24551910720324654,
      "grad_norm": 0.07142449170351028,
      "learning_rate": 2.204431630583548e-06,
      "loss": 1.4185,
      "step": 363
    },
    {
      "epoch": 0.24619546838011497,
      "grad_norm": 0.06955256313085556,
      "learning_rate": 2.087708544541689e-06,
      "loss": 1.3,
      "step": 364
    },
    {
      "epoch": 0.24687182955698342,
      "grad_norm": 0.07163543999195099,
      "learning_rate": 1.974094418431388e-06,
      "loss": 1.4182,
      "step": 365
    },
    {
      "epoch": 0.24754819073385187,
      "grad_norm": 0.06957574933767319,
      "learning_rate": 1.8635966245104664e-06,
      "loss": 1.3436,
      "step": 366
    },
    {
      "epoch": 0.24822455191072032,
      "grad_norm": 0.07044114917516708,
      "learning_rate": 1.7562223328224325e-06,
      "loss": 1.3497,
      "step": 367
    },
    {
      "epoch": 0.24890091308758877,
      "grad_norm": 0.07231567054986954,
      "learning_rate": 1.6519785107311891e-06,
      "loss": 1.4305,
      "step": 368
    },
    {
      "epoch": 0.24957727426445722,
      "grad_norm": 0.071175716817379,
      "learning_rate": 1.5508719224689717e-06,
      "loss": 1.3585,
      "step": 369
    },
    {
      "epoch": 0.2502536354413257,
      "grad_norm": 0.07513930648565292,
      "learning_rate": 1.4529091286973995e-06,
      "loss": 1.4127,
      "step": 370
    },
    {
      "epoch": 0.2509299966181941,
      "grad_norm": 0.07834319770336151,
      "learning_rate": 1.358096486081778e-06,
      "loss": 1.4904,
      "step": 371
    },
    {
      "epoch": 0.2516063577950626,
      "grad_norm": 0.0735422819852829,
      "learning_rate": 1.2664401468786114e-06,
      "loss": 1.4379,
      "step": 372
    },
    {
      "epoch": 0.252282718971931,
      "grad_norm": 0.07638537883758545,
      "learning_rate": 1.1779460585363944e-06,
      "loss": 1.4957,
      "step": 373
    },
    {
      "epoch": 0.2529590801487995,
      "grad_norm": 0.07571271061897278,
      "learning_rate": 1.0926199633097157e-06,
      "loss": 1.4089,
      "step": 374
    },
    {
      "epoch": 0.2536354413256679,
      "grad_norm": 0.07592733204364777,
      "learning_rate": 1.0104673978866164e-06,
      "loss": 1.3655,
      "step": 375
    },
    {
      "epoch": 0.2543118025025364,
      "grad_norm": 0.07473273575305939,
      "learning_rate": 9.314936930293283e-07,
      "loss": 1.3653,
      "step": 376
    },
    {
      "epoch": 0.2549881636794048,
      "grad_norm": 0.07662557065486908,
      "learning_rate": 8.557039732283944e-07,
      "loss": 1.3438,
      "step": 377
    },
    {
      "epoch": 0.2556645248562732,
      "grad_norm": 0.07830655574798584,
      "learning_rate": 7.83103156370113e-07,
      "loss": 1.4263,
      "step": 378
    },
    {
      "epoch": 0.2563408860331417,
      "grad_norm": 0.07488763332366943,
      "learning_rate": 7.136959534174592e-07,
      "loss": 1.3887,
      "step": 379
    },
    {
      "epoch": 0.2570172472100101,
      "grad_norm": 0.07890254259109497,
      "learning_rate": 6.474868681043578e-07,
      "loss": 1.4591,
      "step": 380
    },
    {
      "epoch": 0.2576936083868786,
      "grad_norm": 0.08007548749446869,
      "learning_rate": 5.844801966434832e-07,
      "loss": 1.5055,
      "step": 381
    },
    {
      "epoch": 0.258369969563747,
      "grad_norm": 0.07806207239627838,
      "learning_rate": 5.246800274474439e-07,
      "loss": 1.421,
      "step": 382
    },
    {
      "epoch": 0.2590463307406155,
      "grad_norm": 0.0787520781159401,
      "learning_rate": 4.680902408635335e-07,
      "loss": 1.4013,
      "step": 383
    },
    {
      "epoch": 0.2597226919174839,
      "grad_norm": 0.07986165583133698,
      "learning_rate": 4.1471450892189846e-07,
      "loss": 1.433,
      "step": 384
    },
    {
      "epoch": 0.2603990530943524,
      "grad_norm": 0.0808686912059784,
      "learning_rate": 3.6455629509730136e-07,
      "loss": 1.4251,
      "step": 385
    },
    {
      "epoch": 0.26107541427122083,
      "grad_norm": 0.08336784690618515,
      "learning_rate": 3.1761885408435054e-07,
      "loss": 1.4741,
      "step": 386
    },
    {
      "epoch": 0.26175177544808925,
      "grad_norm": 0.08735602349042892,
      "learning_rate": 2.7390523158633554e-07,
      "loss": 1.4814,
      "step": 387
    },
    {
      "epoch": 0.26242813662495773,
      "grad_norm": 0.07778208702802658,
      "learning_rate": 2.334182641175686e-07,
      "loss": 1.3246,
      "step": 388
    },
    {
      "epoch": 0.26310449780182615,
      "grad_norm": 0.08358029276132584,
      "learning_rate": 1.9616057881935436e-07,
      "loss": 1.5294,
      "step": 389
    },
    {
      "epoch": 0.26378085897869463,
      "grad_norm": 0.08617830276489258,
      "learning_rate": 1.6213459328950352e-07,
      "loss": 1.4288,
      "step": 390
    },
    {
      "epoch": 0.26445722015556306,
      "grad_norm": 0.08489940315485,
      "learning_rate": 1.3134251542544774e-07,
      "loss": 1.387,
      "step": 391
    },
    {
      "epoch": 0.26513358133243153,
      "grad_norm": 0.0889124944806099,
      "learning_rate": 1.0378634328099269e-07,
      "loss": 1.4393,
      "step": 392
    },
    {
      "epoch": 0.26580994250929996,
      "grad_norm": 0.09041323512792587,
      "learning_rate": 7.946786493666647e-08,
      "loss": 1.4255,
      "step": 393
    },
    {
      "epoch": 0.26648630368616844,
      "grad_norm": 0.09535028040409088,
      "learning_rate": 5.838865838366792e-08,
      "loss": 1.3967,
      "step": 394
    },
    {
      "epoch": 0.26716266486303686,
      "grad_norm": 0.09417252242565155,
      "learning_rate": 4.055009142152067e-08,
      "loss": 1.3455,
      "step": 395
    },
    {
      "epoch": 0.26783902603990534,
      "grad_norm": 0.09836997091770172,
      "learning_rate": 2.595332156925534e-08,
      "loss": 1.4622,
      "step": 396
    },
    {
      "epoch": 0.26851538721677376,
      "grad_norm": 0.10254459083080292,
      "learning_rate": 1.4599295990352924e-08,
      "loss": 1.418,
      "step": 397
    },
    {
      "epoch": 0.2691917483936422,
      "grad_norm": 0.11125387996435165,
      "learning_rate": 6.488751431266149e-09,
      "loss": 1.4415,
      "step": 398
    },
    {
      "epoch": 0.26986810957051066,
      "grad_norm": 0.1215406209230423,
      "learning_rate": 1.622214173602199e-09,
      "loss": 1.3851,
      "step": 399
    },
    {
      "epoch": 0.2705444707473791,
      "grad_norm": 0.15271972119808197,
      "learning_rate": 0.0,
      "loss": 1.3691,
      "step": 400
    },
    {
      "epoch": 0.2705444707473791,
      "eval_loss": 1.3924267292022705,
      "eval_runtime": 58.679,
      "eval_samples_per_second": 42.434,
      "eval_steps_per_second": 10.617,
      "step": 400
    }
  ],
  "logging_steps": 1,
  "max_steps": 400,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.285041640112128e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}