{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 19.958677685950413,
  "eval_steps": 500,
  "global_step": 2415,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.08264462809917356,
      "grad_norm": 4.703310012817383,
      "learning_rate": 1.652892561983471e-05,
      "loss": 0.9413,
      "step": 10
    },
    {
      "epoch": 0.1652892561983471,
      "grad_norm": 2.9132559299468994,
      "learning_rate": 3.305785123966942e-05,
      "loss": 0.4897,
      "step": 20
    },
    {
      "epoch": 0.24793388429752067,
      "grad_norm": 1.5209449529647827,
      "learning_rate": 4.958677685950414e-05,
      "loss": 0.2702,
      "step": 30
    },
    {
      "epoch": 0.3305785123966942,
      "grad_norm": 0.6627697944641113,
      "learning_rate": 6.611570247933885e-05,
      "loss": 0.1735,
      "step": 40
    },
    {
      "epoch": 0.4132231404958678,
      "grad_norm": 1.2193719148635864,
      "learning_rate": 8.264462809917356e-05,
      "loss": 0.138,
      "step": 50
    },
    {
      "epoch": 0.49586776859504134,
      "grad_norm": 1.7384006977081299,
      "learning_rate": 9.917355371900827e-05,
      "loss": 0.1263,
      "step": 60
    },
    {
      "epoch": 0.5785123966942148,
      "grad_norm": 1.2912031412124634,
      "learning_rate": 0.00011570247933884298,
      "loss": 0.1205,
      "step": 70
    },
    {
      "epoch": 0.6611570247933884,
      "grad_norm": 0.9840922355651855,
      "learning_rate": 0.0001322314049586777,
      "loss": 0.1048,
      "step": 80
    },
    {
      "epoch": 0.743801652892562,
      "grad_norm": 0.4457073509693146,
      "learning_rate": 0.0001487603305785124,
      "loss": 0.0894,
      "step": 90
    },
    {
      "epoch": 0.8264462809917356,
      "grad_norm": 0.7870272994041443,
      "learning_rate": 0.00016528925619834712,
      "loss": 0.075,
      "step": 100
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.5919432640075684,
      "learning_rate": 0.00018181818181818183,
      "loss": 0.0839,
      "step": 110
    },
    {
      "epoch": 0.9917355371900827,
      "grad_norm": 0.5343506932258606,
      "learning_rate": 0.00019834710743801655,
      "loss": 0.0772,
      "step": 120
    },
    {
      "epoch": 1.0743801652892562,
      "grad_norm": 0.6658118963241577,
      "learning_rate": 0.00019999240439374418,
      "loss": 0.0697,
      "step": 130
    },
    {
      "epoch": 1.1570247933884297,
      "grad_norm": 0.8337640762329102,
      "learning_rate": 0.0001999661494584419,
      "loss": 0.067,
      "step": 140
    },
    {
      "epoch": 1.2396694214876034,
      "grad_norm": 0.4294738471508026,
      "learning_rate": 0.00019992114634399312,
      "loss": 0.07,
      "step": 150
    },
    {
      "epoch": 1.322314049586777,
      "grad_norm": 0.5833399891853333,
      "learning_rate": 0.00019985740349051935,
      "loss": 0.0684,
      "step": 160
    },
    {
      "epoch": 1.4049586776859504,
      "grad_norm": 0.5372803211212158,
      "learning_rate": 0.00019977493285269171,
      "loss": 0.0576,
      "step": 170
    },
    {
      "epoch": 1.487603305785124,
      "grad_norm": 0.4818974435329437,
      "learning_rate": 0.00019967374989748876,
      "loss": 0.0519,
      "step": 180
    },
    {
      "epoch": 1.5702479338842976,
      "grad_norm": 0.4126225709915161,
      "learning_rate": 0.0001995538736012958,
      "loss": 0.0542,
      "step": 190
    },
    {
      "epoch": 1.6528925619834711,
      "grad_norm": 0.4318453073501587,
      "learning_rate": 0.00019941532644634596,
      "loss": 0.0516,
      "step": 200
    },
    {
      "epoch": 1.7355371900826446,
      "grad_norm": 0.6815105676651001,
      "learning_rate": 0.0001992581344165037,
      "loss": 0.0559,
      "step": 210
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.5503532290458679,
      "learning_rate": 0.00019908232699239182,
      "loss": 0.0531,
      "step": 220
    },
    {
      "epoch": 1.9008264462809916,
      "grad_norm": 0.5689036846160889,
      "learning_rate": 0.0001988879371458622,
      "loss": 0.0528,
      "step": 230
    },
    {
      "epoch": 1.9834710743801653,
      "grad_norm": 0.5625288486480713,
      "learning_rate": 0.00019867500133381242,
      "loss": 0.0541,
      "step": 240
    },
    {
      "epoch": 2.0661157024793386,
      "grad_norm": 0.4508824646472931,
      "learning_rate": 0.00019844355949134824,
      "loss": 0.0472,
      "step": 250
    },
    {
      "epoch": 2.1487603305785123,
      "grad_norm": 0.5423370599746704,
      "learning_rate": 0.000198193655024294,
      "loss": 0.0455,
      "step": 260
    },
    {
      "epoch": 2.231404958677686,
      "grad_norm": 0.5978049039840698,
      "learning_rate": 0.00019792533480105203,
      "loss": 0.0453,
      "step": 270
    },
    {
      "epoch": 2.3140495867768593,
      "grad_norm": 0.301248162984848,
      "learning_rate": 0.0001976386491438129,
      "loss": 0.0401,
      "step": 280
    },
    {
      "epoch": 2.396694214876033,
      "grad_norm": 0.2878156006336212,
      "learning_rate": 0.00019733365181911748,
      "loss": 0.041,
      "step": 290
    },
    {
      "epoch": 2.479338842975207,
      "grad_norm": 0.3870035707950592,
      "learning_rate": 0.00019701040002777332,
      "loss": 0.0434,
      "step": 300
    },
    {
      "epoch": 2.56198347107438,
      "grad_norm": 0.5360531210899353,
      "learning_rate": 0.000196668954394127,
      "loss": 0.0496,
      "step": 310
    },
    {
      "epoch": 2.644628099173554,
      "grad_norm": 0.5107744336128235,
      "learning_rate": 0.0001963093789546944,
      "loss": 0.0418,
      "step": 320
    },
    {
      "epoch": 2.7272727272727275,
      "grad_norm": 0.46189603209495544,
      "learning_rate": 0.00019593174114615068,
      "loss": 0.0388,
      "step": 330
    },
    {
      "epoch": 2.809917355371901,
      "grad_norm": 0.3216039836406708,
      "learning_rate": 0.0001955361117926831,
      "loss": 0.0443,
      "step": 340
    },
    {
      "epoch": 2.8925619834710745,
      "grad_norm": 0.5578595399856567,
      "learning_rate": 0.00019512256509270822,
      "loss": 0.0443,
      "step": 350
    },
    {
      "epoch": 2.975206611570248,
      "grad_norm": 0.33241087198257446,
      "learning_rate": 0.0001946911786049562,
      "loss": 0.0446,
      "step": 360
    },
    {
      "epoch": 3.0578512396694215,
      "grad_norm": 0.3146415948867798,
      "learning_rate": 0.0001942420332339252,
      "loss": 0.0388,
      "step": 370
    },
    {
      "epoch": 3.1404958677685952,
      "grad_norm": 0.30948349833488464,
      "learning_rate": 0.00019377521321470805,
      "loss": 0.0349,
      "step": 380
    },
    {
      "epoch": 3.2231404958677685,
      "grad_norm": 0.3389824330806732,
      "learning_rate": 0.00019329080609719435,
      "loss": 0.0387,
      "step": 390
    },
    {
      "epoch": 3.3057851239669422,
      "grad_norm": 0.27903950214385986,
      "learning_rate": 0.00019278890272965096,
      "loss": 0.0365,
      "step": 400
    },
    {
      "epoch": 3.3884297520661155,
      "grad_norm": 0.3887731730937958,
      "learning_rate": 0.00019226959724168367,
      "loss": 0.0398,
      "step": 410
    },
    {
      "epoch": 3.4710743801652892,
      "grad_norm": 0.23304741084575653,
      "learning_rate": 0.0001917329870265839,
      "loss": 0.035,
      "step": 420
    },
    {
      "epoch": 3.553719008264463,
      "grad_norm": 0.4025675654411316,
      "learning_rate": 0.00019117917272306283,
      "loss": 0.0373,
      "step": 430
    },
    {
      "epoch": 3.6363636363636362,
      "grad_norm": 0.4401100277900696,
      "learning_rate": 0.00019060825819637716,
      "loss": 0.0374,
      "step": 440
    },
    {
      "epoch": 3.71900826446281,
      "grad_norm": 0.3880658447742462,
      "learning_rate": 0.00019002035051884973,
      "loss": 0.0396,
      "step": 450
    },
    {
      "epoch": 3.8016528925619832,
      "grad_norm": 0.3345029950141907,
      "learning_rate": 0.00018941555994978848,
      "loss": 0.0332,
      "step": 460
    },
    {
      "epoch": 3.884297520661157,
      "grad_norm": 0.297284334897995,
      "learning_rate": 0.00018879399991480792,
      "loss": 0.0321,
      "step": 470
    },
    {
      "epoch": 3.9669421487603307,
      "grad_norm": 0.27577996253967285,
      "learning_rate": 0.00018815578698455675,
      "loss": 0.0344,
      "step": 480
    },
    {
      "epoch": 4.049586776859504,
      "grad_norm": 0.2191019505262375,
      "learning_rate": 0.00018750104085285546,
      "loss": 0.0308,
      "step": 490
    },
    {
      "epoch": 4.132231404958677,
      "grad_norm": 0.2145395576953888,
      "learning_rate": 0.0001868298843142484,
      "loss": 0.0319,
      "step": 500
    },
    {
      "epoch": 4.214876033057851,
      "grad_norm": 0.2143179327249527,
      "learning_rate": 0.0001861424432409742,
      "loss": 0.031,
      "step": 510
    },
    {
      "epoch": 4.297520661157025,
      "grad_norm": 0.6055622100830078,
      "learning_rate": 0.0001854388465593591,
      "loss": 0.0355,
      "step": 520
    },
    {
      "epoch": 4.380165289256198,
      "grad_norm": 0.5569704174995422,
      "learning_rate": 0.00018471922622563728,
      "loss": 0.0319,
      "step": 530
    },
    {
      "epoch": 4.462809917355372,
      "grad_norm": 0.2579587399959564,
      "learning_rate": 0.0001839837172012033,
      "loss": 0.0331,
      "step": 540
    },
    {
      "epoch": 4.545454545454545,
      "grad_norm": 0.22480279207229614,
      "learning_rate": 0.00018323245742730054,
      "loss": 0.0306,
      "step": 550
    },
    {
      "epoch": 4.628099173553719,
      "grad_norm": 0.260975182056427,
      "learning_rate": 0.00018246558779915105,
      "loss": 0.0342,
      "step": 560
    },
    {
      "epoch": 4.710743801652892,
      "grad_norm": 0.35224688053131104,
      "learning_rate": 0.0001816832521395313,
      "loss": 0.0407,
      "step": 570
    },
    {
      "epoch": 4.793388429752066,
      "grad_norm": 0.265455961227417,
      "learning_rate": 0.000180885597171799,
      "loss": 0.0332,
      "step": 580
    },
    {
      "epoch": 4.87603305785124,
      "grad_norm": 0.2791174352169037,
      "learning_rate": 0.00018007277249237563,
      "loss": 0.0342,
      "step": 590
    },
    {
      "epoch": 4.958677685950414,
      "grad_norm": 0.39626914262771606,
      "learning_rate": 0.00017924493054269046,
      "loss": 0.0342,
      "step": 600
    },
    {
      "epoch": 5.041322314049586,
      "grad_norm": 0.3018089532852173,
      "learning_rate": 0.00017840222658059096,
      "loss": 0.0306,
      "step": 610
    },
    {
      "epoch": 5.12396694214876,
      "grad_norm": 0.4229370653629303,
      "learning_rate": 0.00017754481865122474,
      "loss": 0.0293,
      "step": 620
    },
    {
      "epoch": 5.206611570247934,
      "grad_norm": 0.35095998644828796,
      "learning_rate": 0.00017667286755739895,
      "loss": 0.0334,
      "step": 630
    },
    {
      "epoch": 5.289256198347108,
      "grad_norm": 0.3846903145313263,
      "learning_rate": 0.00017578653682942275,
      "loss": 0.0346,
      "step": 640
    },
    {
      "epoch": 5.371900826446281,
      "grad_norm": 0.3931705057621002,
      "learning_rate": 0.00017488599269443753,
      "loss": 0.0363,
      "step": 650
    },
    {
      "epoch": 5.454545454545454,
      "grad_norm": 0.39143308997154236,
      "learning_rate": 0.0001739714040452422,
      "loss": 0.0274,
      "step": 660
    },
    {
      "epoch": 5.537190082644628,
      "grad_norm": 0.26642906665802,
      "learning_rate": 0.0001730429424086179,
      "loss": 0.0294,
      "step": 670
    },
    {
      "epoch": 5.619834710743802,
      "grad_norm": 0.27915745973587036,
      "learning_rate": 0.00017210078191315904,
      "loss": 0.0275,
      "step": 680
    },
    {
      "epoch": 5.702479338842975,
      "grad_norm": 0.3474787771701813,
      "learning_rate": 0.0001711450992566163,
      "loss": 0.0249,
      "step": 690
    },
    {
      "epoch": 5.785123966942149,
      "grad_norm": 0.4248145818710327,
      "learning_rate": 0.00017017607367275777,
      "loss": 0.0266,
      "step": 700
    },
    {
      "epoch": 5.867768595041323,
      "grad_norm": 0.35364291071891785,
      "learning_rate": 0.00016919388689775464,
      "loss": 0.0295,
      "step": 710
    },
    {
      "epoch": 5.950413223140496,
      "grad_norm": 0.32174554467201233,
      "learning_rate": 0.00016819872313609716,
      "loss": 0.033,
      "step": 720
    },
    {
      "epoch": 6.033057851239669,
      "grad_norm": 0.3670883774757385,
      "learning_rate": 0.00016719076902604836,
      "loss": 0.0325,
      "step": 730
    },
    {
      "epoch": 6.115702479338843,
      "grad_norm": 0.2812848687171936,
      "learning_rate": 0.00016617021360464045,
      "loss": 0.0324,
      "step": 740
    },
    {
      "epoch": 6.198347107438017,
      "grad_norm": 0.2678919732570648,
      "learning_rate": 0.00016513724827222227,
      "loss": 0.0258,
      "step": 750
    },
    {
      "epoch": 6.2809917355371905,
      "grad_norm": 0.33413299918174744,
      "learning_rate": 0.00016409206675656264,
      "loss": 0.0295,
      "step": 760
    },
    {
      "epoch": 6.363636363636363,
      "grad_norm": 0.31874868273735046,
      "learning_rate": 0.00016303486507651807,
      "loss": 0.0267,
      "step": 770
    },
    {
      "epoch": 6.446280991735537,
      "grad_norm": 0.2576686441898346,
      "learning_rate": 0.00016196584150527,
      "loss": 0.033,
      "step": 780
    },
    {
      "epoch": 6.528925619834711,
      "grad_norm": 0.16514086723327637,
      "learning_rate": 0.0001608851965331397,
      "loss": 0.0305,
      "step": 790
    },
    {
      "epoch": 6.6115702479338845,
      "grad_norm": 0.2784286141395569,
      "learning_rate": 0.0001597931328299874,
      "loss": 0.0288,
      "step": 800
    },
    {
      "epoch": 6.694214876033058,
      "grad_norm": 0.21867933869361877,
      "learning_rate": 0.0001586898552072023,
      "loss": 0.0257,
      "step": 810
    },
    {
      "epoch": 6.776859504132231,
      "grad_norm": 0.13699640333652496,
      "learning_rate": 0.00015757557057929127,
      "loss": 0.0243,
      "step": 820
    },
    {
      "epoch": 6.859504132231405,
      "grad_norm": 0.2195553332567215,
      "learning_rate": 0.00015645048792507297,
      "loss": 0.0298,
      "step": 830
    },
    {
      "epoch": 6.9421487603305785,
      "grad_norm": 0.21387651562690735,
      "learning_rate": 0.00015531481824848504,
      "loss": 0.0285,
      "step": 840
    },
    {
      "epoch": 7.024793388429752,
      "grad_norm": 0.1597180813550949,
      "learning_rate": 0.00015416877453901096,
      "loss": 0.0259,
      "step": 850
    },
    {
      "epoch": 7.107438016528926,
      "grad_norm": 0.2567303776741028,
      "learning_rate": 0.00015301257173173523,
      "loss": 0.0258,
      "step": 860
    },
    {
      "epoch": 7.190082644628099,
      "grad_norm": 0.18878251314163208,
      "learning_rate": 0.00015184642666703308,
      "loss": 0.0293,
      "step": 870
    },
    {
      "epoch": 7.2727272727272725,
      "grad_norm": 0.24876275658607483,
      "learning_rate": 0.00015067055804990318,
      "loss": 0.0283,
      "step": 880
    },
    {
      "epoch": 7.355371900826446,
      "grad_norm": 0.29571253061294556,
      "learning_rate": 0.0001494851864089505,
      "loss": 0.0288,
      "step": 890
    },
    {
      "epoch": 7.43801652892562,
      "grad_norm": 0.2919931709766388,
      "learning_rate": 0.00014829053405502723,
      "loss": 0.0282,
      "step": 900
    },
    {
      "epoch": 7.520661157024794,
      "grad_norm": 0.25039032101631165,
      "learning_rate": 0.00014708682503953956,
      "loss": 0.0258,
      "step": 910
    },
    {
      "epoch": 7.6033057851239665,
      "grad_norm": 0.2013220340013504,
      "learning_rate": 0.00014587428511242766,
      "loss": 0.0247,
      "step": 920
    },
    {
      "epoch": 7.68595041322314,
      "grad_norm": 0.3657705783843994,
      "learning_rate": 0.0001446531416798275,
      "loss": 0.0287,
      "step": 930
    },
    {
      "epoch": 7.768595041322314,
      "grad_norm": 0.36600229144096375,
      "learning_rate": 0.00014342362376142185,
      "loss": 0.0271,
      "step": 940
    },
    {
      "epoch": 7.851239669421488,
      "grad_norm": 0.20832963287830353,
      "learning_rate": 0.00014218596194748875,
      "loss": 0.0253,
      "step": 950
    },
    {
      "epoch": 7.933884297520661,
      "grad_norm": 0.15197999775409698,
      "learning_rate": 0.0001409403883556554,
      "loss": 0.0271,
      "step": 960
    },
    {
      "epoch": 8.016528925619834,
      "grad_norm": 0.16558091342449188,
      "learning_rate": 0.00013968713658736552,
      "loss": 0.0265,
      "step": 970
    },
    {
      "epoch": 8.099173553719009,
      "grad_norm": 0.1627083718776703,
      "learning_rate": 0.00013842644168406864,
      "loss": 0.0234,
      "step": 980
    },
    {
      "epoch": 8.181818181818182,
      "grad_norm": 0.33982279896736145,
      "learning_rate": 0.000137158540083139,
      "loss": 0.025,
      "step": 990
    },
    {
      "epoch": 8.264462809917354,
      "grad_norm": 0.415449857711792,
      "learning_rate": 0.000135883669573533,
      "loss": 0.0279,
      "step": 1000
    },
    {
      "epoch": 8.347107438016529,
      "grad_norm": 0.3493754267692566,
      "learning_rate": 0.00013460206925119297,
      "loss": 0.0242,
      "step": 1010
    },
    {
      "epoch": 8.429752066115702,
      "grad_norm": 0.1715659201145172,
      "learning_rate": 0.00013331397947420576,
      "loss": 0.031,
      "step": 1020
    },
    {
      "epoch": 8.512396694214877,
      "grad_norm": 0.27542078495025635,
      "learning_rate": 0.00013201964181772485,
      "loss": 0.0247,
      "step": 1030
    },
    {
      "epoch": 8.59504132231405,
      "grad_norm": 0.25437498092651367,
      "learning_rate": 0.000130719299028664,
      "loss": 0.0239,
      "step": 1040
    },
    {
      "epoch": 8.677685950413224,
      "grad_norm": 0.15187709033489227,
      "learning_rate": 0.00012941319498017116,
      "loss": 0.0219,
      "step": 1050
    },
    {
      "epoch": 8.760330578512397,
      "grad_norm": 0.2681483328342438,
      "learning_rate": 0.00012810157462589137,
      "loss": 0.0229,
      "step": 1060
    },
    {
      "epoch": 8.84297520661157,
      "grad_norm": 0.1943976879119873,
      "learning_rate": 0.00012678468395402662,
      "loss": 0.0224,
      "step": 1070
    },
    {
      "epoch": 8.925619834710744,
      "grad_norm": 0.25881528854370117,
      "learning_rate": 0.00012546276994120228,
      "loss": 0.026,
      "step": 1080
    },
    {
      "epoch": 9.008264462809917,
      "grad_norm": 0.21106192469596863,
      "learning_rate": 0.0001241360805061474,
      "loss": 0.0208,
      "step": 1090
    },
    {
      "epoch": 9.090909090909092,
      "grad_norm": 0.2531886398792267,
      "learning_rate": 0.00012280486446319895,
      "loss": 0.0257,
      "step": 1100
    },
    {
      "epoch": 9.173553719008265,
      "grad_norm": 0.22699125111103058,
      "learning_rate": 0.00012146937147563793,
      "loss": 0.0172,
      "step": 1110
    },
    {
      "epoch": 9.256198347107437,
      "grad_norm": 0.1889602094888687,
      "learning_rate": 0.00012012985200886602,
      "loss": 0.022,
      "step": 1120
    },
    {
      "epoch": 9.338842975206612,
      "grad_norm": 0.24796439707279205,
      "learning_rate": 0.00011878655728343204,
      "loss": 0.0217,
      "step": 1130
    },
    {
      "epoch": 9.421487603305785,
      "grad_norm": 0.2704314589500427,
      "learning_rate": 0.00011743973922791678,
      "loss": 0.0229,
      "step": 1140
    },
    {
      "epoch": 9.50413223140496,
      "grad_norm": 0.26962810754776,
      "learning_rate": 0.00011608965043168496,
      "loss": 0.0201,
      "step": 1150
    },
    {
      "epoch": 9.586776859504132,
      "grad_norm": 0.1735532283782959,
      "learning_rate": 0.0001147365440975132,
      "loss": 0.0215,
      "step": 1160
    },
    {
      "epoch": 9.669421487603305,
      "grad_norm": 0.15696899592876434,
      "learning_rate": 0.00011338067399410319,
      "loss": 0.0241,
      "step": 1170
    },
    {
      "epoch": 9.75206611570248,
      "grad_norm": 0.23969663679599762,
      "learning_rate": 0.00011202229440848844,
      "loss": 0.0238,
      "step": 1180
    },
    {
      "epoch": 9.834710743801653,
      "grad_norm": 0.23791910707950592,
      "learning_rate": 0.000110661660098344,
      "loss": 0.0207,
      "step": 1190
    },
    {
      "epoch": 9.917355371900827,
      "grad_norm": 0.24661310017108917,
      "learning_rate": 0.00010929902624420787,
      "loss": 0.025,
      "step": 1200
    },
    {
      "epoch": 10.0,
      "grad_norm": 0.3029937446117401,
      "learning_rate": 0.00010793464840162302,
      "loss": 0.0198,
      "step": 1210
    },
    {
      "epoch": 10.082644628099173,
      "grad_norm": 0.22985076904296875,
      "learning_rate": 0.00010656878245320933,
      "loss": 0.0186,
      "step": 1220
    },
    {
      "epoch": 10.165289256198347,
      "grad_norm": 0.23791086673736572,
      "learning_rate": 0.00010520168456067381,
      "loss": 0.0218,
      "step": 1230
    },
    {
      "epoch": 10.24793388429752,
      "grad_norm": 0.23052431643009186,
      "learning_rate": 0.0001038336111167688,
      "loss": 0.0207,
      "step": 1240
    },
    {
      "epoch": 10.330578512396695,
      "grad_norm": 0.18408218026161194,
      "learning_rate": 0.00010246481869720665,
      "loss": 0.0189,
      "step": 1250
    },
    {
      "epoch": 10.413223140495868,
      "grad_norm": 0.1350499838590622,
      "learning_rate": 0.00010109556401254032,
      "loss": 0.0214,
      "step": 1260
    },
    {
      "epoch": 10.49586776859504,
      "grad_norm": 0.21356913447380066,
      "learning_rate": 9.972610386001824e-05,
      "loss": 0.0204,
      "step": 1270
    },
    {
      "epoch": 10.578512396694215,
      "grad_norm": 0.19436530768871307,
      "learning_rate": 9.83566950754235e-05,
      "loss": 0.0234,
      "step": 1280
    },
    {
      "epoch": 10.661157024793388,
      "grad_norm": 0.22124770283699036,
      "learning_rate": 9.698759448490528e-05,
      "loss": 0.0204,
      "step": 1290
    },
    {
      "epoch": 10.743801652892563,
      "grad_norm": 0.23618550598621368,
      "learning_rate": 9.561905885681252e-05,
      "loss": 0.02,
      "step": 1300
    },
    {
      "epoch": 10.826446280991735,
      "grad_norm": 0.22095705568790436,
      "learning_rate": 9.425134485353804e-05,
      "loss": 0.0191,
      "step": 1310
    },
    {
      "epoch": 10.909090909090908,
      "grad_norm": 0.1535860002040863,
      "learning_rate": 9.288470898338283e-05,
      "loss": 0.0205,
      "step": 1320
    },
    {
      "epoch": 10.991735537190083,
      "grad_norm": 0.23291295766830444,
      "learning_rate": 9.151940755244912e-05,
      "loss": 0.0228,
      "step": 1330
    },
    {
      "epoch": 11.074380165289256,
      "grad_norm": 0.22103387117385864,
      "learning_rate": 9.01556966165713e-05,
      "loss": 0.021,
      "step": 1340
    },
    {
      "epoch": 11.15702479338843,
      "grad_norm": 0.2173365205526352,
      "learning_rate": 8.879383193329409e-05,
      "loss": 0.0224,
      "step": 1350
    },
    {
      "epoch": 11.239669421487603,
      "grad_norm": 0.18998365104198456,
      "learning_rate": 8.743406891390607e-05,
      "loss": 0.0191,
      "step": 1360
    },
    {
      "epoch": 11.322314049586776,
      "grad_norm": 0.22178007662296295,
      "learning_rate": 8.607666257553881e-05,
      "loss": 0.0182,
      "step": 1370
    },
    {
      "epoch": 11.40495867768595,
      "grad_norm": 0.17217198014259338,
      "learning_rate": 8.472186749333941e-05,
      "loss": 0.017,
      "step": 1380
    },
    {
      "epoch": 11.487603305785123,
      "grad_norm": 0.2341386377811432,
      "learning_rate": 8.336993775272624e-05,
      "loss": 0.017,
      "step": 1390
    },
    {
      "epoch": 11.570247933884298,
      "grad_norm": 0.2643473744392395,
      "learning_rate": 8.202112690173643e-05,
      "loss": 0.0182,
      "step": 1400
    },
    {
      "epoch": 11.652892561983471,
      "grad_norm": 0.20025375485420227,
      "learning_rate": 8.067568790347394e-05,
      "loss": 0.0184,
      "step": 1410
    },
    {
      "epoch": 11.735537190082646,
      "grad_norm": 0.3023671507835388,
      "learning_rate": 7.933387308866793e-05,
      "loss": 0.0196,
      "step": 1420
    },
    {
      "epoch": 11.818181818181818,
      "grad_norm": 0.3398241400718689,
      "learning_rate": 7.799593410834882e-05,
      "loss": 0.0219,
      "step": 1430
    },
    {
      "epoch": 11.900826446280991,
      "grad_norm": 0.26556509733200073,
      "learning_rate": 7.666212188665278e-05,
      "loss": 0.0167,
      "step": 1440
    },
    {
      "epoch": 11.983471074380166,
      "grad_norm": 0.16627757251262665,
      "learning_rate": 7.533268657376175e-05,
      "loss": 0.0173,
      "step": 1450
    },
    {
      "epoch": 12.066115702479339,
      "grad_norm": 0.1822843700647354,
      "learning_rate": 7.400787749898919e-05,
      "loss": 0.0211,
      "step": 1460
    },
    {
      "epoch": 12.148760330578513,
      "grad_norm": 0.20193220674991608,
      "learning_rate": 7.268794312401934e-05,
      "loss": 0.0158,
      "step": 1470
    },
    {
      "epoch": 12.231404958677686,
      "grad_norm": 0.16035108268260956,
      "learning_rate": 7.137313099630965e-05,
      "loss": 0.0181,
      "step": 1480
    },
    {
      "epoch": 12.314049586776859,
      "grad_norm": 0.14358443021774292,
      "learning_rate": 7.006368770266421e-05,
      "loss": 0.0176,
      "step": 1490
    },
    {
      "epoch": 12.396694214876034,
      "grad_norm": 0.2652245759963989,
      "learning_rate": 6.875985882298767e-05,
      "loss": 0.0151,
      "step": 1500
    },
    {
      "epoch": 12.479338842975206,
      "grad_norm": 0.16228072345256805,
      "learning_rate": 6.746188888422788e-05,
      "loss": 0.0189,
      "step": 1510
    },
    {
      "epoch": 12.561983471074381,
      "grad_norm": 0.2025652527809143,
      "learning_rate": 6.617002131451595e-05,
      "loss": 0.0175,
      "step": 1520
    },
    {
      "epoch": 12.644628099173554,
      "grad_norm": 0.19407929480075836,
      "learning_rate": 6.488449839751248e-05,
      "loss": 0.0195,
      "step": 1530
    },
    {
      "epoch": 12.727272727272727,
      "grad_norm": 0.19878722727298737,
      "learning_rate": 6.360556122696847e-05,
      "loss": 0.0179,
      "step": 1540
    },
    {
      "epoch": 12.809917355371901,
      "grad_norm": 0.14901714026927948,
      "learning_rate": 6.233344966150931e-05,
      "loss": 0.0174,
      "step": 1550
    },
    {
      "epoch": 12.892561983471074,
      "grad_norm": 0.19965632259845734,
      "learning_rate": 6.106840227965031e-05,
      "loss": 0.0202,
      "step": 1560
    },
    {
      "epoch": 12.975206611570249,
      "grad_norm": 0.19168788194656372,
      "learning_rate": 5.98106563350526e-05,
      "loss": 0.0184,
      "step": 1570
    },
    {
      "epoch": 13.057851239669422,
      "grad_norm": 0.21455532312393188,
      "learning_rate": 5.8560447712027176e-05,
      "loss": 0.0185,
      "step": 1580
    },
    {
      "epoch": 13.140495867768594,
      "grad_norm": 0.14643830060958862,
      "learning_rate": 5.7318010881296e-05,
      "loss": 0.0156,
      "step": 1590
    },
    {
      "epoch": 13.223140495867769,
      "grad_norm": 0.15653955936431885,
      "learning_rate": 5.6083578856018116e-05,
      "loss": 0.0172,
      "step": 1600
    },
    {
      "epoch": 13.305785123966942,
      "grad_norm": 0.19662214815616608,
      "learning_rate": 5.485738314808905e-05,
      "loss": 0.0176,
      "step": 1610
    },
    {
      "epoch": 13.388429752066116,
      "grad_norm": 0.17977331578731537,
      "learning_rate": 5.363965372472195e-05,
      "loss": 0.0149,
      "step": 1620
    },
    {
      "epoch": 13.47107438016529,
      "grad_norm": 0.16984397172927856,
      "learning_rate": 5.2430618965318356e-05,
      "loss": 0.0155,
      "step": 1630
    },
    {
      "epoch": 13.553719008264462,
      "grad_norm": 0.23604920506477356,
      "learning_rate": 5.123050561863657e-05,
      "loss": 0.0172,
      "step": 1640
    },
    {
      "epoch": 13.636363636363637,
      "grad_norm": 0.20416028797626495,
      "learning_rate": 5.0039538760266034e-05,
      "loss": 0.015,
      "step": 1650
    },
    {
      "epoch": 13.71900826446281,
      "grad_norm": 0.17371866106987,
      "learning_rate": 4.885794175041567e-05,
      "loss": 0.0145,
      "step": 1660
    },
    {
      "epoch": 13.801652892561984,
      "grad_norm": 0.18007588386535645,
      "learning_rate": 4.7685936192023415e-05,
      "loss": 0.0159,
      "step": 1670
    },
    {
      "epoch": 13.884297520661157,
      "grad_norm": 0.1721491813659668,
      "learning_rate": 4.652374188919584e-05,
      "loss": 0.0167,
      "step": 1680
    },
    {
      "epoch": 13.96694214876033,
      "grad_norm": 0.24759605526924133,
      "learning_rate": 4.537157680598486e-05,
      "loss": 0.0191,
      "step": 1690
    },
    {
      "epoch": 14.049586776859504,
      "grad_norm": 0.147015780210495,
      "learning_rate": 4.422965702550971e-05,
      "loss": 0.018,
      "step": 1700
    },
    {
      "epoch": 14.132231404958677,
      "grad_norm": 0.2071155458688736,
      "learning_rate": 4.309819670943139e-05,
      "loss": 0.0174,
      "step": 1710
    },
    {
      "epoch": 14.214876033057852,
      "grad_norm": 0.19617584347724915,
      "learning_rate": 4.197740805778779e-05,
      "loss": 0.0147,
      "step": 1720
    },
    {
      "epoch": 14.297520661157025,
      "grad_norm": 0.20283041894435883,
      "learning_rate": 4.086750126919655e-05,
      "loss": 0.0149,
      "step": 1730
    },
    {
      "epoch": 14.380165289256198,
      "grad_norm": 0.23828528821468353,
      "learning_rate": 3.976868450143329e-05,
      "loss": 0.0166,
      "step": 1740
    },
    {
      "epoch": 14.462809917355372,
      "grad_norm": 0.2454332709312439,
      "learning_rate": 3.868116383239259e-05,
      "loss": 0.016,
      "step": 1750
    },
    {
      "epoch": 14.545454545454545,
      "grad_norm": 0.21714092791080475,
      "learning_rate": 3.7605143221439066e-05,
      "loss": 0.0154,
      "step": 1760
    },
    {
      "epoch": 14.62809917355372,
      "grad_norm": 0.20886965095996857,
      "learning_rate": 3.6540824471155856e-05,
      "loss": 0.0176,
      "step": 1770
    },
    {
      "epoch": 14.710743801652892,
      "grad_norm": 0.20230558514595032,
      "learning_rate": 3.548840718949724e-05,
      "loss": 0.0164,
      "step": 1780
    },
    {
      "epoch": 14.793388429752067,
      "grad_norm": 0.18747635185718536,
      "learning_rate": 3.4448088752353434e-05,
      "loss": 0.0159,
      "step": 1790
    },
    {
      "epoch": 14.87603305785124,
      "grad_norm": 0.15160124003887177,
      "learning_rate": 3.3420064266533415e-05,
      "loss": 0.0153,
      "step": 1800
    },
    {
      "epoch": 14.958677685950413,
      "grad_norm": 0.21707528829574585,
      "learning_rate": 3.240452653317367e-05,
      "loss": 0.0163,
      "step": 1810
    },
    {
      "epoch": 15.041322314049587,
      "grad_norm": 0.2630583345890045,
      "learning_rate": 3.140166601157928e-05,
      "loss": 0.0153,
      "step": 1820
    },
    {
      "epoch": 15.12396694214876,
      "grad_norm": 0.2211904525756836,
      "learning_rate": 3.0411670783504053e-05,
      "loss": 0.0159,
      "step": 1830
    },
    {
      "epoch": 15.206611570247935,
      "grad_norm": 0.27094215154647827,
      "learning_rate": 2.94347265178769e-05,
      "loss": 0.0143,
      "step": 1840
    },
    {
      "epoch": 15.289256198347108,
      "grad_norm": 0.2170286327600479,
      "learning_rate": 2.847101643598028e-05,
      "loss": 0.0154,
      "step": 1850
    },
    {
      "epoch": 15.37190082644628,
      "grad_norm": 0.15928302705287933,
      "learning_rate": 2.7520721277088024e-05,
      "loss": 0.0162,
      "step": 1860
    },
    {
      "epoch": 15.454545454545455,
      "grad_norm": 0.20816448330879211,
      "learning_rate": 2.6584019264568495e-05,
      "loss": 0.0127,
      "step": 1870
    },
    {
      "epoch": 15.537190082644628,
      "grad_norm": 0.14063189923763275,
      "learning_rate": 2.5661086072459707e-05,
      "loss": 0.0168,
      "step": 1880
    },
    {
      "epoch": 15.619834710743802,
      "grad_norm": 0.2043224722146988,
      "learning_rate": 2.4752094792522372e-05,
      "loss": 0.0152,
      "step": 1890
    },
    {
      "epoch": 15.702479338842975,
      "grad_norm": 0.17056132853031158,
      "learning_rate": 2.385721590177754e-05,
      "loss": 0.0154,
      "step": 1900
    },
    {
      "epoch": 15.785123966942148,
      "grad_norm": 0.20709876716136932,
      "learning_rate": 2.2976617230534202e-05,
      "loss": 0.0132,
      "step": 1910
    },
    {
      "epoch": 15.867768595041323,
      "grad_norm": 0.14429618418216705,
      "learning_rate": 2.211046393091374e-05,
      "loss": 0.0165,
      "step": 1920
    },
    {
      "epoch": 15.950413223140496,
      "grad_norm": 0.15559087693691254,
      "learning_rate": 2.1258918445876263e-05,
      "loss": 0.0147,
      "step": 1930
    },
    {
      "epoch": 16.03305785123967,
      "grad_norm": 0.17711517214775085,
      "learning_rate": 2.0422140478755248e-05,
      "loss": 0.016,
      "step": 1940
    },
    {
      "epoch": 16.115702479338843,
      "grad_norm": 0.14507338404655457,
      "learning_rate": 1.9600286963305957e-05,
      "loss": 0.0126,
      "step": 1950
    },
    {
      "epoch": 16.198347107438018,
      "grad_norm": 0.12779353559017181,
      "learning_rate": 1.8793512034273242e-05,
      "loss": 0.0133,
      "step": 1960
    },
    {
      "epoch": 16.28099173553719,
      "grad_norm": 0.1657179296016693,
      "learning_rate": 1.8001966998484253e-05,
      "loss": 0.0134,
      "step": 1970
    },
    {
      "epoch": 16.363636363636363,
      "grad_norm": 0.17538750171661377,
      "learning_rate": 1.7225800306471696e-05,
      "loss": 0.0135,
      "step": 1980
    },
    {
      "epoch": 16.446280991735538,
      "grad_norm": 0.18412277102470398,
      "learning_rate": 1.6465157524632436e-05,
      "loss": 0.0138,
      "step": 1990
    },
    {
      "epoch": 16.52892561983471,
      "grad_norm": 0.152567520737648,
      "learning_rate": 1.5720181307927373e-05,
      "loss": 0.012,
      "step": 2000
    },
    {
      "epoch": 16.611570247933884,
      "grad_norm": 0.2471626251935959,
      "learning_rate": 1.4991011373127084e-05,
      "loss": 0.0149,
      "step": 2010
    },
    {
      "epoch": 16.694214876033058,
      "grad_norm": 0.22137291729450226,
      "learning_rate": 1.4277784472608635e-05,
      "loss": 0.0126,
      "step": 2020
    },
    {
      "epoch": 16.776859504132233,
      "grad_norm": 0.16841760277748108,
      "learning_rate": 1.3580634368708366e-05,
      "loss": 0.0151,
      "step": 2030
    },
    {
      "epoch": 16.859504132231404,
      "grad_norm": 0.17957067489624023,
      "learning_rate": 1.2899691808635283e-05,
      "loss": 0.015,
      "step": 2040
    },
    {
      "epoch": 16.94214876033058,
      "grad_norm": 0.17319756746292114,
      "learning_rate": 1.2235084499950178e-05,
      "loss": 0.0128,
      "step": 2050
    },
    {
      "epoch": 17.024793388429753,
      "grad_norm": 0.14869403839111328,
      "learning_rate": 1.1586937086614525e-05,
      "loss": 0.0133,
      "step": 2060
    },
    {
      "epoch": 17.107438016528924,
      "grad_norm": 0.15348997712135315,
      "learning_rate": 1.0955371125614223e-05,
      "loss": 0.0137,
      "step": 2070
    },
    {
      "epoch": 17.1900826446281,
      "grad_norm": 0.2330683022737503,
      "learning_rate": 1.0340505064162076e-05,
      "loss": 0.0143,
      "step": 2080
    },
    {
      "epoch": 17.272727272727273,
      "grad_norm": 0.17106673121452332,
      "learning_rate": 9.742454217483632e-06,
      "loss": 0.014,
      "step": 2090
    },
    {
      "epoch": 17.355371900826448,
      "grad_norm": 0.15164901316165924,
      "learning_rate": 9.161330747190321e-06,
      "loss": 0.0151,
      "step": 2100
    },
    {
      "epoch": 17.43801652892562,
      "grad_norm": 0.18551285564899445,
      "learning_rate": 8.59724364024409e-06,
      "loss": 0.011,
      "step": 2110
    },
    {
      "epoch": 17.520661157024794,
      "grad_norm": 0.13972045481204987,
      "learning_rate": 8.050298688517444e-06,
      "loss": 0.0104,
      "step": 2120
    },
    {
      "epoch": 17.60330578512397,
      "grad_norm": 0.17032235860824585,
      "learning_rate": 7.520598468952578e-06,
      "loss": 0.0136,
      "step": 2130
    },
    {
      "epoch": 17.68595041322314,
      "grad_norm": 0.14614145457744598,
      "learning_rate": 7.0082423243236995e-06,
      "loss": 0.0134,
      "step": 2140
    },
    {
      "epoch": 17.768595041322314,
      "grad_norm": 0.19540473818778992,
      "learning_rate": 6.513326344605608e-06,
      "loss": 0.013,
      "step": 2150
    },
    {
      "epoch": 17.85123966942149,
      "grad_norm": 0.16889305412769318,
      "learning_rate": 6.035943348952611e-06,
      "loss": 0.0128,
      "step": 2160
    },
    {
      "epoch": 17.93388429752066,
      "grad_norm": 0.13315258920192719,
      "learning_rate": 5.576182868290669e-06,
      "loss": 0.0102,
      "step": 2170
    },
    {
      "epoch": 18.016528925619834,
      "grad_norm": 0.197654590010643,
      "learning_rate": 5.134131128526243e-06,
      "loss": 0.012,
      "step": 2180
    },
    {
      "epoch": 18.09917355371901,
      "grad_norm": 0.16169218719005585,
      "learning_rate": 4.709871034375146e-06,
      "loss": 0.0103,
      "step": 2190
    },
    {
      "epoch": 18.181818181818183,
      "grad_norm": 0.1659018099308014,
      "learning_rate": 4.303482153814064e-06,
      "loss": 0.0125,
      "step": 2200
    },
    {
      "epoch": 18.264462809917354,
      "grad_norm": 0.3515506982803345,
      "learning_rate": 3.915040703157991e-06,
      "loss": 0.0141,
      "step": 2210
    },
    {
      "epoch": 18.34710743801653,
      "grad_norm": 0.1042797714471817,
      "learning_rate": 3.5446195327662133e-06,
      "loss": 0.0128,
      "step": 2220
    },
    {
      "epoch": 18.429752066115704,
      "grad_norm": 0.19790691137313843,
      "learning_rate": 3.1922881133795825e-06,
      "loss": 0.015,
      "step": 2230
    },
    {
      "epoch": 18.512396694214875,
      "grad_norm": 0.21495257318019867,
      "learning_rate": 2.8581125230915605e-06,
      "loss": 0.011,
      "step": 2240
    },
    {
      "epoch": 18.59504132231405,
      "grad_norm": 0.35978537797927856,
      "learning_rate": 2.5421554349556733e-06,
      "loss": 0.0119,
      "step": 2250
    },
    {
      "epoch": 18.677685950413224,
      "grad_norm": 0.1598854809999466,
      "learning_rate": 2.2444761052313856e-06,
      "loss": 0.0136,
      "step": 2260
    },
    {
      "epoch": 18.760330578512395,
      "grad_norm": 0.1521085500717163,
      "learning_rate": 1.965130362270917e-06,
      "loss": 0.0115,
      "step": 2270
    },
    {
      "epoch": 18.84297520661157,
      "grad_norm": 0.185346782207489,
      "learning_rate": 1.704170596048893e-06,
      "loss": 0.0134,
      "step": 2280
    },
    {
      "epoch": 18.925619834710744,
      "grad_norm": 0.23003143072128296,
      "learning_rate": 1.4616457483368263e-06,
      "loss": 0.0102,
      "step": 2290
    },
    {
      "epoch": 19.00826446280992,
      "grad_norm": 0.23142951726913452,
      "learning_rate": 1.2376013035243606e-06,
      "loss": 0.0116,
      "step": 2300
    },
    {
      "epoch": 19.09090909090909,
      "grad_norm": 0.12257048487663269,
      "learning_rate": 1.0320792800888602e-06,
      "loss": 0.0132,
      "step": 2310
    },
    {
      "epoch": 19.173553719008265,
      "grad_norm": 0.11083350330591202,
      "learning_rate": 8.451182227150467e-07,
      "loss": 0.0128,
      "step": 2320
    },
    {
      "epoch": 19.25619834710744,
      "grad_norm": 0.17381551861763,
      "learning_rate": 6.767531950661377e-07,
      "loss": 0.0137,
      "step": 2330
    },
    {
      "epoch": 19.33884297520661,
      "grad_norm": 0.20176014304161072,
      "learning_rate": 5.270157732077951e-07,
      "loss": 0.0151,
      "step": 2340
    },
    {
      "epoch": 19.421487603305785,
      "grad_norm": 0.17078755795955658,
      "learning_rate": 3.959340396862188e-07,
      "loss": 0.0125,
      "step": 2350
    },
    {
      "epoch": 19.50413223140496,
      "grad_norm": 0.23273628950119019,
      "learning_rate": 2.8353257826136957e-07,
      "loss": 0.0148,
      "step": 2360
    },
    {
      "epoch": 19.586776859504134,
      "grad_norm": 0.11116746068000793,
      "learning_rate": 1.8983246929642485e-07,
      "loss": 0.0119,
      "step": 2370
    },
    {
      "epoch": 19.669421487603305,
      "grad_norm": 0.16333059966564178,
      "learning_rate": 1.1485128580422944e-07,
      "loss": 0.0108,
      "step": 2380
    },
    {
      "epoch": 19.75206611570248,
      "grad_norm": 0.12590940296649933,
      "learning_rate": 5.860309015156551e-08,
      "loss": 0.0109,
      "step": 2390
    },
    {
      "epoch": 19.834710743801654,
      "grad_norm": 0.12979190051555634,
      "learning_rate": 2.1098431421862076e-08,
      "loss": 0.0131,
      "step": 2400
    },
    {
      "epoch": 19.917355371900825,
      "grad_norm": 0.18272912502288818,
      "learning_rate": 2.3443434366887673e-09,
      "loss": 0.0123,
      "step": 2410
    },
    {
      "epoch": 19.958677685950413,
      "step": 2415,
      "total_flos": 3.496526197912458e+17,
      "train_loss": 0.034255825163599866,
      "train_runtime": 2744.0812,
      "train_samples_per_second": 56.325,
      "train_steps_per_second": 0.88
    }
  ],
  "logging_steps": 10,
  "max_steps": 2415,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 10000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.496526197912458e+17,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}