{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999731255038968,
  "eval_steps": 500,
  "global_step": 930,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010749798441279225,
      "grad_norm": 22.762095682751983,
      "learning_rate": 1.0752688172043012e-07,
      "loss": 1.3377,
      "step": 1
    },
    {
      "epoch": 0.005374899220639613,
      "grad_norm": 20.69927583673527,
      "learning_rate": 5.376344086021506e-07,
      "loss": 1.3086,
      "step": 5
    },
    {
      "epoch": 0.010749798441279226,
      "grad_norm": 8.308333536609279,
      "learning_rate": 1.0752688172043011e-06,
      "loss": 1.1878,
      "step": 10
    },
    {
      "epoch": 0.01612469766191884,
      "grad_norm": 6.959676956693679,
      "learning_rate": 1.6129032258064516e-06,
      "loss": 1.0321,
      "step": 15
    },
    {
      "epoch": 0.021499596882558453,
      "grad_norm": 2.9820242928638794,
      "learning_rate": 2.1505376344086023e-06,
      "loss": 0.9101,
      "step": 20
    },
    {
      "epoch": 0.026874496103198066,
      "grad_norm": 2.5554546533203144,
      "learning_rate": 2.688172043010753e-06,
      "loss": 0.8617,
      "step": 25
    },
    {
      "epoch": 0.03224939532383768,
      "grad_norm": 2.286909995566289,
      "learning_rate": 3.225806451612903e-06,
      "loss": 0.8383,
      "step": 30
    },
    {
      "epoch": 0.03762429454447729,
      "grad_norm": 2.3398312761165907,
      "learning_rate": 3.763440860215054e-06,
      "loss": 0.8185,
      "step": 35
    },
    {
      "epoch": 0.042999193765116905,
      "grad_norm": 2.204837686008753,
      "learning_rate": 4.3010752688172045e-06,
      "loss": 0.8053,
      "step": 40
    },
    {
      "epoch": 0.04837409298575652,
      "grad_norm": 2.315534843885615,
      "learning_rate": 4.838709677419355e-06,
      "loss": 0.7857,
      "step": 45
    },
    {
      "epoch": 0.05374899220639613,
      "grad_norm": 2.5131629870127705,
      "learning_rate": 5.376344086021506e-06,
      "loss": 0.7691,
      "step": 50
    },
    {
      "epoch": 0.059123891427035745,
      "grad_norm": 2.4607920938058396,
      "learning_rate": 5.9139784946236566e-06,
      "loss": 0.758,
      "step": 55
    },
    {
      "epoch": 0.06449879064767536,
      "grad_norm": 2.367828432773105,
      "learning_rate": 6.451612903225806e-06,
      "loss": 0.7375,
      "step": 60
    },
    {
      "epoch": 0.06987368986831496,
      "grad_norm": 2.3306378375646406,
      "learning_rate": 6.989247311827958e-06,
      "loss": 0.7304,
      "step": 65
    },
    {
      "epoch": 0.07524858908895458,
      "grad_norm": 2.344704242935191,
      "learning_rate": 7.526881720430108e-06,
      "loss": 0.7131,
      "step": 70
    },
    {
      "epoch": 0.08062348830959419,
      "grad_norm": 2.3556743904747366,
      "learning_rate": 8.064516129032258e-06,
      "loss": 0.7109,
      "step": 75
    },
    {
      "epoch": 0.08599838753023381,
      "grad_norm": 2.2479332716993605,
      "learning_rate": 8.602150537634409e-06,
      "loss": 0.7022,
      "step": 80
    },
    {
      "epoch": 0.09137328675087342,
      "grad_norm": 2.3017400204623666,
      "learning_rate": 9.13978494623656e-06,
      "loss": 0.7046,
      "step": 85
    },
    {
      "epoch": 0.09674818597151304,
      "grad_norm": 2.3666760651980536,
      "learning_rate": 9.67741935483871e-06,
      "loss": 0.698,
      "step": 90
    },
    {
      "epoch": 0.10212308519215264,
      "grad_norm": 2.9278768472096433,
      "learning_rate": 9.999859120828162e-06,
      "loss": 0.6998,
      "step": 95
    },
    {
      "epoch": 0.10749798441279226,
      "grad_norm": 2.4524372124571165,
      "learning_rate": 9.998274321315453e-06,
      "loss": 0.6905,
      "step": 100
    },
    {
      "epoch": 0.11287288363343187,
      "grad_norm": 2.4347797178103394,
      "learning_rate": 9.994929183335237e-06,
      "loss": 0.679,
      "step": 105
    },
    {
      "epoch": 0.11824778285407149,
      "grad_norm": 2.3508265818164515,
      "learning_rate": 9.989824885009142e-06,
      "loss": 0.6843,
      "step": 110
    },
    {
      "epoch": 0.1236226820747111,
      "grad_norm": 2.164423459020718,
      "learning_rate": 9.982963224016152e-06,
      "loss": 0.6784,
      "step": 115
    },
    {
      "epoch": 0.12899758129535072,
      "grad_norm": 2.4050223924123113,
      "learning_rate": 9.974346616959476e-06,
      "loss": 0.6693,
      "step": 120
    },
    {
      "epoch": 0.13437248051599032,
      "grad_norm": 2.1741064666149903,
      "learning_rate": 9.963978098515468e-06,
      "loss": 0.6667,
      "step": 125
    },
    {
      "epoch": 0.13974737973662993,
      "grad_norm": 2.151624445725951,
      "learning_rate": 9.951861320364822e-06,
      "loss": 0.6686,
      "step": 130
    },
    {
      "epoch": 0.14512227895726956,
      "grad_norm": 2.140470340411352,
      "learning_rate": 9.938000549906509e-06,
      "loss": 0.6626,
      "step": 135
    },
    {
      "epoch": 0.15049717817790917,
      "grad_norm": 2.2274245089556706,
      "learning_rate": 9.922400668754833e-06,
      "loss": 0.6517,
      "step": 140
    },
    {
      "epoch": 0.15587207739854878,
      "grad_norm": 1.95083198656833,
      "learning_rate": 9.905067171020185e-06,
      "loss": 0.6512,
      "step": 145
    },
    {
      "epoch": 0.16124697661918838,
      "grad_norm": 2.4043795722101136,
      "learning_rate": 9.88600616137407e-06,
      "loss": 0.6524,
      "step": 150
    },
    {
      "epoch": 0.16662187583982802,
      "grad_norm": 2.004729246454245,
      "learning_rate": 9.86522435289912e-06,
      "loss": 0.6495,
      "step": 155
    },
    {
      "epoch": 0.17199677506046762,
      "grad_norm": 2.0488055473575724,
      "learning_rate": 9.8427290647248e-06,
      "loss": 0.6417,
      "step": 160
    },
    {
      "epoch": 0.17737167428110723,
      "grad_norm": 2.026026305529713,
      "learning_rate": 9.818528219449705e-06,
      "loss": 0.6308,
      "step": 165
    },
    {
      "epoch": 0.18274657350174683,
      "grad_norm": 1.923497956617847,
      "learning_rate": 9.792630340351301e-06,
      "loss": 0.6464,
      "step": 170
    },
    {
      "epoch": 0.18812147272238647,
      "grad_norm": 1.8714980274335011,
      "learning_rate": 9.765044548384113e-06,
      "loss": 0.6406,
      "step": 175
    },
    {
      "epoch": 0.19349637194302607,
      "grad_norm": 2.130465692033948,
      "learning_rate": 9.735780558967434e-06,
      "loss": 0.6292,
      "step": 180
    },
    {
      "epoch": 0.19887127116366568,
      "grad_norm": 2.0472476860009086,
      "learning_rate": 9.70484867856365e-06,
      "loss": 0.6209,
      "step": 185
    },
    {
      "epoch": 0.2042461703843053,
      "grad_norm": 2.45138914999084,
      "learning_rate": 9.67225980104841e-06,
      "loss": 0.633,
      "step": 190
    },
    {
      "epoch": 0.20962106960494492,
      "grad_norm": 1.9725264804153964,
      "learning_rate": 9.638025403873939e-06,
      "loss": 0.625,
      "step": 195
    },
    {
      "epoch": 0.21499596882558453,
      "grad_norm": 1.956589833642945,
      "learning_rate": 9.602157544026785e-06,
      "loss": 0.6274,
      "step": 200
    },
    {
      "epoch": 0.22037086804622413,
      "grad_norm": 2.032060013563488,
      "learning_rate": 9.564668853781483e-06,
      "loss": 0.6143,
      "step": 205
    },
    {
      "epoch": 0.22574576726686374,
      "grad_norm": 1.9927039178935393,
      "learning_rate": 9.525572536251608e-06,
      "loss": 0.6131,
      "step": 210
    },
    {
      "epoch": 0.23112066648750335,
      "grad_norm": 1.946959208571873,
      "learning_rate": 9.484882360739772e-06,
      "loss": 0.6029,
      "step": 215
    },
    {
      "epoch": 0.23649556570814298,
      "grad_norm": 1.8806701194051492,
      "learning_rate": 9.442612657888237e-06,
      "loss": 0.6118,
      "step": 220
    },
    {
      "epoch": 0.24187046492878259,
      "grad_norm": 2.225666197462802,
      "learning_rate": 9.398778314631801e-06,
      "loss": 0.6028,
      "step": 225
    },
    {
      "epoch": 0.2472453641494222,
      "grad_norm": 2.001619119705015,
      "learning_rate": 9.353394768954791e-06,
      "loss": 0.609,
      "step": 230
    },
    {
      "epoch": 0.2526202633700618,
      "grad_norm": 2.2733431261860777,
      "learning_rate": 9.30647800445397e-06,
      "loss": 0.601,
      "step": 235
    },
    {
      "epoch": 0.25799516259070143,
      "grad_norm": 2.230210940508007,
      "learning_rate": 9.258044544709276e-06,
      "loss": 0.5833,
      "step": 240
    },
    {
      "epoch": 0.26337006181134104,
      "grad_norm": 1.904244193327823,
      "learning_rate": 9.208111447464407e-06,
      "loss": 0.6101,
      "step": 245
    },
    {
      "epoch": 0.26874496103198064,
      "grad_norm": 1.88832266700029,
      "learning_rate": 9.156696298619266e-06,
      "loss": 0.5953,
      "step": 250
    },
    {
      "epoch": 0.27411986025262025,
      "grad_norm": 1.838793567250841,
      "learning_rate": 9.103817206036383e-06,
      "loss": 0.594,
      "step": 255
    },
    {
      "epoch": 0.27949475947325986,
      "grad_norm": 1.951838096647268,
      "learning_rate": 9.049492793163539e-06,
      "loss": 0.5783,
      "step": 260
    },
    {
      "epoch": 0.28486965869389946,
      "grad_norm": 2.1569848912019296,
      "learning_rate": 8.993742192474773e-06,
      "loss": 0.5897,
      "step": 265
    },
    {
      "epoch": 0.2902445579145391,
      "grad_norm": 2.0054996500986833,
      "learning_rate": 8.936585038732143e-06,
      "loss": 0.5866,
      "step": 270
    },
    {
      "epoch": 0.29561945713517873,
      "grad_norm": 2.112212926041267,
      "learning_rate": 8.878041462070556e-06,
      "loss": 0.5795,
      "step": 275
    },
    {
      "epoch": 0.30099435635581834,
      "grad_norm": 2.1036396849174492,
      "learning_rate": 8.818132080908178e-06,
      "loss": 0.5818,
      "step": 280
    },
    {
      "epoch": 0.30636925557645794,
      "grad_norm": 2.167096693852596,
      "learning_rate": 8.756877994684818e-06,
      "loss": 0.564,
      "step": 285
    },
    {
      "epoch": 0.31174415479709755,
      "grad_norm": 1.956775665235312,
      "learning_rate": 8.694300776430958e-06,
      "loss": 0.5683,
      "step": 290
    },
    {
      "epoch": 0.31711905401773716,
      "grad_norm": 1.97168436084639,
      "learning_rate": 8.630422465169947e-06,
      "loss": 0.5697,
      "step": 295
    },
    {
      "epoch": 0.32249395323837676,
      "grad_norm": 2.043558514889714,
      "learning_rate": 8.565265558156101e-06,
      "loss": 0.5635,
      "step": 300
    },
    {
      "epoch": 0.32786885245901637,
      "grad_norm": 1.9273574277695336,
      "learning_rate": 8.498853002951414e-06,
      "loss": 0.5728,
      "step": 305
    },
    {
      "epoch": 0.33324375167965603,
      "grad_norm": 1.9657306797632228,
      "learning_rate": 8.43120818934367e-06,
      "loss": 0.5619,
      "step": 310
    },
    {
      "epoch": 0.33861865090029564,
      "grad_norm": 1.9012625288122775,
      "learning_rate": 8.362354941108803e-06,
      "loss": 0.5702,
      "step": 315
    },
    {
      "epoch": 0.34399355012093524,
      "grad_norm": 2.123858299795287,
      "learning_rate": 8.292317507620438e-06,
      "loss": 0.554,
      "step": 320
    },
    {
      "epoch": 0.34936844934157485,
      "grad_norm": 1.9513877089408058,
      "learning_rate": 8.221120555309511e-06,
      "loss": 0.5498,
      "step": 325
    },
    {
      "epoch": 0.35474334856221446,
      "grad_norm": 1.914136234798412,
      "learning_rate": 8.148789158977012e-06,
      "loss": 0.5455,
      "step": 330
    },
    {
      "epoch": 0.36011824778285406,
      "grad_norm": 1.8672863149021612,
      "learning_rate": 8.075348792962924e-06,
      "loss": 0.5404,
      "step": 335
    },
    {
      "epoch": 0.36549314700349367,
      "grad_norm": 2.1793777540733177,
      "learning_rate": 8.000825322174424e-06,
      "loss": 0.5423,
      "step": 340
    },
    {
      "epoch": 0.3708680462241333,
      "grad_norm": 2.0291841027180455,
      "learning_rate": 7.925244992976538e-06,
      "loss": 0.5386,
      "step": 345
    },
    {
      "epoch": 0.37624294544477294,
      "grad_norm": 2.1016851887581662,
      "learning_rate": 7.848634423948468e-06,
      "loss": 0.5326,
      "step": 350
    },
    {
      "epoch": 0.38161784466541254,
      "grad_norm": 2.029810803364677,
      "learning_rate": 7.7710205965088e-06,
      "loss": 0.5403,
      "step": 355
    },
    {
      "epoch": 0.38699274388605215,
      "grad_norm": 1.8590989815971988,
      "learning_rate": 7.692430845412946e-06,
      "loss": 0.5333,
      "step": 360
    },
    {
      "epoch": 0.39236764310669175,
      "grad_norm": 1.9140526286879267,
      "learning_rate": 7.612892849126132e-06,
      "loss": 0.5252,
      "step": 365
    },
    {
      "epoch": 0.39774254232733136,
      "grad_norm": 1.9988915945152586,
      "learning_rate": 7.532434620075349e-06,
      "loss": 0.5242,
      "step": 370
    },
    {
      "epoch": 0.40311744154797097,
      "grad_norm": 1.9746262065023925,
      "learning_rate": 7.451084494783668e-06,
      "loss": 0.5085,
      "step": 375
    },
    {
      "epoch": 0.4084923407686106,
      "grad_norm": 1.9222623607017046,
      "learning_rate": 7.368871123890425e-06,
      "loss": 0.5247,
      "step": 380
    },
    {
      "epoch": 0.4138672399892502,
      "grad_norm": 2.184926573195526,
      "learning_rate": 7.285823462060776e-06,
      "loss": 0.5153,
      "step": 385
    },
    {
      "epoch": 0.41924213920988984,
      "grad_norm": 1.9008034480063025,
      "learning_rate": 7.201970757788172e-06,
      "loss": 0.5146,
      "step": 390
    },
    {
      "epoch": 0.42461703843052945,
      "grad_norm": 2.0387228861562776,
      "learning_rate": 7.117342543093358e-06,
      "loss": 0.506,
      "step": 395
    },
    {
      "epoch": 0.42999193765116905,
      "grad_norm": 2.165202028954103,
      "learning_rate": 7.031968623123503e-06,
      "loss": 0.5086,
      "step": 400
    },
    {
      "epoch": 0.43536683687180866,
      "grad_norm": 2.0059909450641435,
      "learning_rate": 6.945879065655164e-06,
      "loss": 0.5052,
      "step": 405
    },
    {
      "epoch": 0.44074173609244827,
      "grad_norm": 2.3690249701648725,
      "learning_rate": 6.859104190504725e-06,
      "loss": 0.5042,
      "step": 410
    },
    {
      "epoch": 0.4461166353130879,
      "grad_norm": 2.010697016188747,
      "learning_rate": 6.771674558850088e-06,
      "loss": 0.4958,
      "step": 415
    },
    {
      "epoch": 0.4514915345337275,
      "grad_norm": 1.9600695859304653,
      "learning_rate": 6.6836209624673575e-06,
      "loss": 0.5028,
      "step": 420
    },
    {
      "epoch": 0.4568664337543671,
      "grad_norm": 1.897791733525091,
      "learning_rate": 6.5949744128863026e-06,
      "loss": 0.4918,
      "step": 425
    },
    {
      "epoch": 0.4622413329750067,
      "grad_norm": 2.0272846103958755,
      "learning_rate": 6.5057661304684314e-06,
      "loss": 0.4863,
      "step": 430
    },
    {
      "epoch": 0.46761623219564635,
      "grad_norm": 1.888723644878263,
      "learning_rate": 6.41602753341152e-06,
      "loss": 0.4816,
      "step": 435
    },
    {
      "epoch": 0.47299113141628596,
      "grad_norm": 1.9192743092814748,
      "learning_rate": 6.32579022668446e-06,
      "loss": 0.4651,
      "step": 440
    },
    {
      "epoch": 0.47836603063692557,
      "grad_norm": 2.015499929746225,
      "learning_rate": 6.235085990896317e-06,
      "loss": 0.4843,
      "step": 445
    },
    {
      "epoch": 0.48374092985756517,
      "grad_norm": 1.8591635323006084,
      "learning_rate": 6.143946771103561e-06,
      "loss": 0.4792,
      "step": 450
    },
    {
      "epoch": 0.4891158290782048,
      "grad_norm": 1.9418548689130606,
      "learning_rate": 6.052404665559342e-06,
      "loss": 0.4808,
      "step": 455
    },
    {
      "epoch": 0.4944907282988444,
      "grad_norm": 1.944779410362686,
      "learning_rate": 5.960491914408846e-06,
      "loss": 0.478,
      "step": 460
    },
    {
      "epoch": 0.499865627519484,
      "grad_norm": 1.9179287908203055,
      "learning_rate": 5.8682408883346535e-06,
      "loss": 0.4741,
      "step": 465
    },
    {
      "epoch": 0.5052405267401237,
      "grad_norm": 1.939456137738699,
      "learning_rate": 5.775684077156133e-06,
      "loss": 0.4711,
      "step": 470
    },
    {
      "epoch": 0.5106154259607633,
      "grad_norm": 2.030961825347864,
      "learning_rate": 5.682854078386882e-06,
      "loss": 0.4684,
      "step": 475
    },
    {
      "epoch": 0.5159903251814029,
      "grad_norm": 2.017192647267279,
      "learning_rate": 5.5897835857542315e-06,
      "loss": 0.4716,
      "step": 480
    },
    {
      "epoch": 0.5213652244020425,
      "grad_norm": 1.8797704382098077,
      "learning_rate": 5.496505377684858e-06,
      "loss": 0.46,
      "step": 485
    },
    {
      "epoch": 0.5267401236226821,
      "grad_norm": 1.937682332181844,
      "learning_rate": 5.4030523057605865e-06,
      "loss": 0.4561,
      "step": 490
    },
    {
      "epoch": 0.5321150228433217,
      "grad_norm": 2.030922067258147,
      "learning_rate": 5.30945728314841e-06,
      "loss": 0.4558,
      "step": 495
    },
    {
      "epoch": 0.5374899220639613,
      "grad_norm": 2.0947701090151347,
      "learning_rate": 5.215753273008828e-06,
      "loss": 0.4483,
      "step": 500
    },
    {
      "epoch": 0.5428648212846009,
      "grad_norm": 1.9527895814298004,
      "learning_rate": 5.1219732768865744e-06,
      "loss": 0.4546,
      "step": 505
    },
    {
      "epoch": 0.5482397205052405,
      "grad_norm": 1.925264597102223,
      "learning_rate": 5.0281503230878304e-06,
      "loss": 0.454,
      "step": 510
    },
    {
      "epoch": 0.5536146197258801,
      "grad_norm": 1.870811300114845,
      "learning_rate": 4.934317455048005e-06,
      "loss": 0.4519,
      "step": 515
    },
    {
      "epoch": 0.5589895189465197,
      "grad_norm": 1.8517150460463325,
      "learning_rate": 4.840507719694202e-06,
      "loss": 0.4465,
      "step": 520
    },
    {
      "epoch": 0.5643644181671593,
      "grad_norm": 1.9901783088282807,
      "learning_rate": 4.746754155806437e-06,
      "loss": 0.4426,
      "step": 525
    },
    {
      "epoch": 0.5697393173877989,
      "grad_norm": 1.92526563478314,
      "learning_rate": 4.6530897823817425e-06,
      "loss": 0.447,
      "step": 530
    },
    {
      "epoch": 0.5751142166084386,
      "grad_norm": 2.0060885985999297,
      "learning_rate": 4.559547587005227e-06,
      "loss": 0.4324,
      "step": 535
    },
    {
      "epoch": 0.5804891158290783,
      "grad_norm": 1.9893587686847451,
      "learning_rate": 4.466160514232206e-06,
      "loss": 0.4333,
      "step": 540
    },
    {
      "epoch": 0.5858640150497179,
      "grad_norm": 1.9557768830171276,
      "learning_rate": 4.3729614539854815e-06,
      "loss": 0.4317,
      "step": 545
    },
    {
      "epoch": 0.5912389142703575,
      "grad_norm": 1.934922488824026,
      "learning_rate": 4.279983229971863e-06,
      "loss": 0.4385,
      "step": 550
    },
    {
      "epoch": 0.5966138134909971,
      "grad_norm": 1.876883342350168,
      "learning_rate": 4.187258588122019e-06,
      "loss": 0.4308,
      "step": 555
    },
    {
      "epoch": 0.6019887127116367,
      "grad_norm": 1.9145406252528392,
      "learning_rate": 4.094820185057701e-06,
      "loss": 0.429,
      "step": 560
    },
    {
      "epoch": 0.6073636119322763,
      "grad_norm": 1.9668078483751361,
      "learning_rate": 4.002700576590441e-06,
      "loss": 0.4355,
      "step": 565
    },
    {
      "epoch": 0.6127385111529159,
      "grad_norm": 1.961317833983311,
      "learning_rate": 3.910932206255742e-06,
      "loss": 0.4307,
      "step": 570
    },
    {
      "epoch": 0.6181134103735555,
      "grad_norm": 1.9266836648594068,
      "learning_rate": 3.819547393886816e-06,
      "loss": 0.4228,
      "step": 575
    },
    {
      "epoch": 0.6234883095941951,
      "grad_norm": 1.9178449532415156,
      "learning_rate": 3.7285783242318773e-06,
      "loss": 0.4208,
      "step": 580
    },
    {
      "epoch": 0.6288632088148347,
      "grad_norm": 1.9258945419319073,
      "learning_rate": 3.6380570356190346e-06,
      "loss": 0.4198,
      "step": 585
    },
    {
      "epoch": 0.6342381080354743,
      "grad_norm": 1.900131272139124,
      "learning_rate": 3.548015408672723e-06,
      "loss": 0.4166,
      "step": 590
    },
    {
      "epoch": 0.6396130072561139,
      "grad_norm": 1.9436783208199349,
      "learning_rate": 3.4584851550857007e-06,
      "loss": 0.4097,
      "step": 595
    },
    {
      "epoch": 0.6449879064767535,
      "grad_norm": 1.9775742266969982,
      "learning_rate": 3.3694978064505258e-06,
      "loss": 0.4129,
      "step": 600
    },
    {
      "epoch": 0.6503628056973931,
      "grad_norm": 1.8269664591538408,
      "learning_rate": 3.2810847031544703e-06,
      "loss": 0.4088,
      "step": 605
    },
    {
      "epoch": 0.6557377049180327,
      "grad_norm": 1.886926257627791,
      "learning_rate": 3.193276983341773e-06,
      "loss": 0.4047,
      "step": 610
    },
    {
      "epoch": 0.6611126041386725,
      "grad_norm": 1.9544062693179776,
      "learning_rate": 3.10610557194712e-06,
      "loss": 0.4062,
      "step": 615
    },
    {
      "epoch": 0.6664875033593121,
      "grad_norm": 1.8781772152552532,
      "learning_rate": 3.019601169804216e-06,
      "loss": 0.4104,
      "step": 620
    },
    {
      "epoch": 0.6718624025799517,
      "grad_norm": 1.8152357686298919,
      "learning_rate": 2.9337942428332787e-06,
      "loss": 0.4045,
      "step": 625
    },
    {
      "epoch": 0.6772373018005913,
      "grad_norm": 1.8231399150815806,
      "learning_rate": 2.848715011311271e-06,
      "loss": 0.392,
      "step": 630
    },
    {
      "epoch": 0.6826122010212309,
      "grad_norm": 1.8660672229253348,
      "learning_rate": 2.764393439228643e-06,
      "loss": 0.3943,
      "step": 635
    },
    {
      "epoch": 0.6879871002418705,
      "grad_norm": 1.8610053663644113,
      "learning_rate": 2.6808592237363364e-06,
      "loss": 0.4008,
      "step": 640
    },
    {
      "epoch": 0.6933619994625101,
      "grad_norm": 1.8382570488101446,
      "learning_rate": 2.5981417846867753e-06,
      "loss": 0.4016,
      "step": 645
    },
    {
      "epoch": 0.6987368986831497,
      "grad_norm": 2.1963817133336487,
      "learning_rate": 2.5162702542724924e-06,
      "loss": 0.3897,
      "step": 650
    },
    {
      "epoch": 0.7041117979037893,
      "grad_norm": 1.9176668039213747,
      "learning_rate": 2.4352734667661073e-06,
      "loss": 0.3828,
      "step": 655
    },
    {
      "epoch": 0.7094866971244289,
      "grad_norm": 1.7727520223769333,
      "learning_rate": 2.3551799483651894e-06,
      "loss": 0.3918,
      "step": 660
    },
    {
      "epoch": 0.7148615963450685,
      "grad_norm": 1.8508452992048683,
      "learning_rate": 2.2760179071456356e-06,
      "loss": 0.3923,
      "step": 665
    },
    {
      "epoch": 0.7202364955657081,
      "grad_norm": 1.8898208151955234,
      "learning_rate": 2.1978152231271077e-06,
      "loss": 0.3889,
      "step": 670
    },
    {
      "epoch": 0.7256113947863477,
      "grad_norm": 1.8442479576678996,
      "learning_rate": 2.120599438453968e-06,
      "loss": 0.3803,
      "step": 675
    },
    {
      "epoch": 0.7309862940069873,
      "grad_norm": 1.8287369121112131,
      "learning_rate": 2.044397747695247e-06,
      "loss": 0.3803,
      "step": 680
    },
    {
      "epoch": 0.7363611932276269,
      "grad_norm": 1.8799793582063802,
      "learning_rate": 1.969236988267005e-06,
      "loss": 0.3761,
      "step": 685
    },
    {
      "epoch": 0.7417360924482665,
      "grad_norm": 1.7804441999040213,
      "learning_rate": 1.8951436309804766e-06,
      "loss": 0.3803,
      "step": 690
    },
    {
      "epoch": 0.7471109916689062,
      "grad_norm": 1.8057045901843964,
      "learning_rate": 1.8221437707193424e-06,
      "loss": 0.3791,
      "step": 695
    },
    {
      "epoch": 0.7524858908895459,
      "grad_norm": 1.8066778942795885,
      "learning_rate": 1.7502631172493878e-06,
      "loss": 0.3787,
      "step": 700
    },
    {
      "epoch": 0.7578607901101855,
      "grad_norm": 1.8302691938217297,
      "learning_rate": 1.6795269861638041e-06,
      "loss": 0.3881,
      "step": 705
    },
    {
      "epoch": 0.7632356893308251,
      "grad_norm": 1.8234451924525408,
      "learning_rate": 1.6099602899673083e-06,
      "loss": 0.3755,
      "step": 710
    },
    {
      "epoch": 0.7686105885514647,
      "grad_norm": 1.767257969497289,
      "learning_rate": 1.5415875293022181e-06,
      "loss": 0.3767,
      "step": 715
    },
    {
      "epoch": 0.7739854877721043,
      "grad_norm": 1.8275002961230755,
      "learning_rate": 1.4744327843196043e-06,
      "loss": 0.3676,
      "step": 720
    },
    {
      "epoch": 0.7793603869927439,
      "grad_norm": 1.8120986388263132,
      "learning_rate": 1.4085197061985022e-06,
      "loss": 0.378,
      "step": 725
    },
    {
      "epoch": 0.7847352862133835,
      "grad_norm": 1.7559930973845725,
      "learning_rate": 1.3438715088162403e-06,
      "loss": 0.3676,
      "step": 730
    },
    {
      "epoch": 0.7901101854340231,
      "grad_norm": 1.8425030002430842,
      "learning_rate": 1.280510960572745e-06,
      "loss": 0.3721,
      "step": 735
    },
    {
      "epoch": 0.7954850846546627,
      "grad_norm": 1.8012025523517194,
      "learning_rate": 1.2184603763717684e-06,
      "loss": 0.3624,
      "step": 740
    },
    {
      "epoch": 0.8008599838753023,
      "grad_norm": 1.861454253307067,
      "learning_rate": 1.1577416097618138e-06,
      "loss": 0.3628,
      "step": 745
    },
    {
      "epoch": 0.8062348830959419,
      "grad_norm": 1.7687205113359847,
      "learning_rate": 1.0983760452395415e-06,
      "loss": 0.3624,
      "step": 750
    },
    {
      "epoch": 0.8116097823165815,
      "grad_norm": 1.9487313717243606,
      "learning_rate": 1.040384590718399e-06,
      "loss": 0.3554,
      "step": 755
    },
    {
      "epoch": 0.8169846815372211,
      "grad_norm": 1.6883835685983015,
      "learning_rate": 9.837876701650606e-07,
      "loss": 0.3552,
      "step": 760
    },
    {
      "epoch": 0.8223595807578608,
      "grad_norm": 1.8609027176051518,
      "learning_rate": 9.286052164063369e-07,
      "loss": 0.3658,
      "step": 765
    },
    {
      "epoch": 0.8277344799785004,
      "grad_norm": 1.8177990709248102,
      "learning_rate": 8.748566641090433e-07,
      "loss": 0.3649,
      "step": 770
    },
    {
      "epoch": 0.83310937919914,
      "grad_norm": 1.8719226675515828,
      "learning_rate": 8.225609429353187e-07,
      "loss": 0.3679,
      "step": 775
    },
    {
      "epoch": 0.8384842784197797,
      "grad_norm": 1.740827963426437,
      "learning_rate": 7.717364708758024e-07,
      "loss": 0.3587,
      "step": 780
    },
    {
      "epoch": 0.8438591776404193,
      "grad_norm": 1.7094014840076446,
      "learning_rate": 7.224011477630166e-07,
      "loss": 0.3615,
      "step": 785
    },
    {
      "epoch": 0.8492340768610589,
      "grad_norm": 1.790943519825671,
      "learning_rate": 6.745723489672412e-07,
      "loss": 0.3401,
      "step": 790
    },
    {
      "epoch": 0.8546089760816985,
      "grad_norm": 1.852602320274836,
      "learning_rate": 6.282669192770896e-07,
      "loss": 0.3615,
      "step": 795
    },
    {
      "epoch": 0.8599838753023381,
      "grad_norm": 1.7448118489231759,
      "learning_rate": 5.83501166966956e-07,
      "loss": 0.356,
      "step": 800
    },
    {
      "epoch": 0.8653587745229777,
      "grad_norm": 1.7138058444443134,
      "learning_rate": 5.402908580534233e-07,
      "loss": 0.3507,
      "step": 805
    },
    {
      "epoch": 0.8707336737436173,
      "grad_norm": 1.786529360611276,
      "learning_rate": 4.986512107426283e-07,
      "loss": 0.3545,
      "step": 810
    },
    {
      "epoch": 0.8761085729642569,
      "grad_norm": 1.756925743913887,
      "learning_rate": 4.5859689007058896e-07,
      "loss": 0.3561,
      "step": 815
    },
    {
      "epoch": 0.8814834721848965,
      "grad_norm": 1.7845712546592873,
      "learning_rate": 4.2014200273832406e-07,
      "loss": 0.3484,
      "step": 820
    },
    {
      "epoch": 0.8868583714055361,
      "grad_norm": 1.8932772047523514,
      "learning_rate": 3.8330009214363197e-07,
      "loss": 0.3568,
      "step": 825
    },
    {
      "epoch": 0.8922332706261757,
      "grad_norm": 1.6852471718751032,
      "learning_rate": 3.4808413361125004e-07,
      "loss": 0.3483,
      "step": 830
    },
    {
      "epoch": 0.8976081698468154,
      "grad_norm": 1.774916611408541,
      "learning_rate": 3.1450652982307815e-07,
      "loss": 0.3549,
      "step": 835
    },
    {
      "epoch": 0.902983069067455,
      "grad_norm": 1.9063855186007104,
      "learning_rate": 2.8257910645009935e-07,
      "loss": 0.3551,
      "step": 840
    },
    {
      "epoch": 0.9083579682880946,
      "grad_norm": 1.7048794107309078,
      "learning_rate": 2.523131079874963e-07,
      "loss": 0.3542,
      "step": 845
    },
    {
      "epoch": 0.9137328675087342,
      "grad_norm": 1.8134483816091473,
      "learning_rate": 2.2371919379446495e-07,
      "loss": 0.3546,
      "step": 850
    },
    {
      "epoch": 0.9191077667293738,
      "grad_norm": 1.847919769203762,
      "learning_rate": 1.9680743434010385e-07,
      "loss": 0.3468,
      "step": 855
    },
    {
      "epoch": 0.9244826659500134,
      "grad_norm": 1.7597705426288421,
      "learning_rate": 1.7158730765669817e-07,
      "loss": 0.3598,
      "step": 860
    },
    {
      "epoch": 0.9298575651706531,
      "grad_norm": 1.8107276346763232,
      "learning_rate": 1.480676960016636e-07,
      "loss": 0.3596,
      "step": 865
    },
    {
      "epoch": 0.9352324643912927,
      "grad_norm": 1.726387338435163,
      "learning_rate": 1.2625688272930925e-07,
      "loss": 0.3555,
      "step": 870
    },
    {
      "epoch": 0.9406073636119323,
      "grad_norm": 1.657583954677237,
      "learning_rate": 1.0616254937352966e-07,
      "loss": 0.3478,
      "step": 875
    },
    {
      "epoch": 0.9459822628325719,
      "grad_norm": 1.7186030940673933,
      "learning_rate": 8.779177294245044e-08,
      "loss": 0.3476,
      "step": 880
    },
    {
      "epoch": 0.9513571620532115,
      "grad_norm": 1.803294093442226,
      "learning_rate": 7.115102342598101e-08,
      "loss": 0.3473,
      "step": 885
    },
    {
      "epoch": 0.9567320612738511,
      "grad_norm": 1.9161389425178095,
      "learning_rate": 5.6246161517158336e-08,
      "loss": 0.346,
      "step": 890
    },
    {
      "epoch": 0.9621069604944907,
      "grad_norm": 1.7244304109116553,
      "learning_rate": 4.308243654806643e-08,
      "loss": 0.3522,
      "step": 895
    },
    {
      "epoch": 0.9674818597151303,
      "grad_norm": 1.798195655862142,
      "learning_rate": 3.166448464108629e-08,
      "loss": 0.348,
      "step": 900
    },
    {
      "epoch": 0.97285675893577,
      "grad_norm": 1.7093455607124064,
      "learning_rate": 2.1996327076096446e-08,
      "loss": 0.3497,
      "step": 905
    },
    {
      "epoch": 0.9782316581564096,
      "grad_norm": 1.79086386462793,
      "learning_rate": 1.4081368874226398e-08,
      "loss": 0.3483,
      "step": 910
    },
    {
      "epoch": 0.9836065573770492,
      "grad_norm": 1.7408308676379578,
      "learning_rate": 7.922397598642551e-09,
      "loss": 0.3451,
      "step": 915
    },
    {
      "epoch": 0.9889814565976888,
      "grad_norm": 1.7334185566757478,
      "learning_rate": 3.5215823727974274e-09,
      "loss": 0.3496,
      "step": 920
    },
    {
      "epoch": 0.9943563558183284,
      "grad_norm": 1.7114218675449298,
      "learning_rate": 8.804731164901991e-10,
      "loss": 0.345,
      "step": 925
    },
    {
      "epoch": 0.999731255038968,
      "grad_norm": 1.7912428537040286,
      "learning_rate": 0.0,
      "loss": 0.3503,
      "step": 930
    },
    {
      "epoch": 0.999731255038968,
      "eval_loss": 0.3442555069923401,
      "eval_runtime": 95.8162,
      "eval_samples_per_second": 3.152,
      "eval_steps_per_second": 0.793,
      "step": 930
    },
    {
      "epoch": 0.999731255038968,
      "step": 930,
      "total_flos": 194670734868480.0,
      "train_loss": 0.508077065149943,
      "train_runtime": 21031.9081,
      "train_samples_per_second": 1.415,
      "train_steps_per_second": 0.044
    }
  ],
  "logging_steps": 5,
  "max_steps": 930,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 194670734868480.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}