{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.996845425867508,
  "eval_steps": 500,
  "global_step": 1108,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.018026137899954935,
      "grad_norm": 18.020910263061523,
      "learning_rate": 4.9548736462093865e-05,
      "loss": 0.6809,
      "step": 10
    },
    {
      "epoch": 0.03605227579990987,
      "grad_norm": 30.794422149658203,
      "learning_rate": 4.909747292418773e-05,
      "loss": 0.572,
      "step": 20
    },
    {
      "epoch": 0.054078413699864804,
      "grad_norm": 24.659021377563477,
      "learning_rate": 4.864620938628159e-05,
      "loss": 0.554,
      "step": 30
    },
    {
      "epoch": 0.07210455159981974,
      "grad_norm": 17.13719367980957,
      "learning_rate": 4.819494584837546e-05,
      "loss": 0.5261,
      "step": 40
    },
    {
      "epoch": 0.09013068949977468,
      "grad_norm": 50.614078521728516,
      "learning_rate": 4.7743682310469314e-05,
      "loss": 0.5696,
      "step": 50
    },
    {
      "epoch": 0.10815682739972961,
      "grad_norm": 16.31667137145996,
      "learning_rate": 4.7292418772563177e-05,
      "loss": 0.4876,
      "step": 60
    },
    {
      "epoch": 0.12618296529968454,
      "grad_norm": 41.160762786865234,
      "learning_rate": 4.684115523465704e-05,
      "loss": 0.4647,
      "step": 70
    },
    {
      "epoch": 0.14420910319963948,
      "grad_norm": 20.278989791870117,
      "learning_rate": 4.63898916967509e-05,
      "loss": 0.4525,
      "step": 80
    },
    {
      "epoch": 0.16223524109959442,
      "grad_norm": 21.479433059692383,
      "learning_rate": 4.5938628158844764e-05,
      "loss": 0.4511,
      "step": 90
    },
    {
      "epoch": 0.18026137899954936,
      "grad_norm": 20.85803985595703,
      "learning_rate": 4.548736462093863e-05,
      "loss": 0.4828,
      "step": 100
    },
    {
      "epoch": 0.19828751689950427,
      "grad_norm": 19.689502716064453,
      "learning_rate": 4.5036101083032495e-05,
      "loss": 0.4785,
      "step": 110
    },
    {
      "epoch": 0.21631365479945922,
      "grad_norm": 23.28278350830078,
      "learning_rate": 4.458483754512636e-05,
      "loss": 0.4593,
      "step": 120
    },
    {
      "epoch": 0.23433979269941416,
      "grad_norm": 55.461483001708984,
      "learning_rate": 4.413357400722022e-05,
      "loss": 0.4265,
      "step": 130
    },
    {
      "epoch": 0.25236593059936907,
      "grad_norm": 21.591472625732422,
      "learning_rate": 4.368231046931408e-05,
      "loss": 0.4611,
      "step": 140
    },
    {
      "epoch": 0.270392068499324,
      "grad_norm": 31.00871467590332,
      "learning_rate": 4.3231046931407945e-05,
      "loss": 0.4489,
      "step": 150
    },
    {
      "epoch": 0.28841820639927895,
      "grad_norm": 21.315185546875,
      "learning_rate": 4.277978339350181e-05,
      "loss": 0.4755,
      "step": 160
    },
    {
      "epoch": 0.3064443442992339,
      "grad_norm": 25.012041091918945,
      "learning_rate": 4.232851985559567e-05,
      "loss": 0.4703,
      "step": 170
    },
    {
      "epoch": 0.32447048219918884,
      "grad_norm": 18.25890350341797,
      "learning_rate": 4.187725631768953e-05,
      "loss": 0.4456,
      "step": 180
    },
    {
      "epoch": 0.3424966200991438,
      "grad_norm": 19.33783721923828,
      "learning_rate": 4.1425992779783394e-05,
      "loss": 0.4764,
      "step": 190
    },
    {
      "epoch": 0.3605227579990987,
      "grad_norm": 21.992473602294922,
      "learning_rate": 4.0974729241877256e-05,
      "loss": 0.4204,
      "step": 200
    },
    {
      "epoch": 0.3785488958990536,
      "grad_norm": 20.253664016723633,
      "learning_rate": 4.052346570397112e-05,
      "loss": 0.4471,
      "step": 210
    },
    {
      "epoch": 0.39657503379900855,
      "grad_norm": 17.41029167175293,
      "learning_rate": 4.007220216606498e-05,
      "loss": 0.4387,
      "step": 220
    },
    {
      "epoch": 0.4146011716989635,
      "grad_norm": 18.28990364074707,
      "learning_rate": 3.962093862815885e-05,
      "loss": 0.4283,
      "step": 230
    },
    {
      "epoch": 0.43262730959891843,
      "grad_norm": 48.463008880615234,
      "learning_rate": 3.916967509025271e-05,
      "loss": 0.4447,
      "step": 240
    },
    {
      "epoch": 0.45065344749887337,
      "grad_norm": 15.58568000793457,
      "learning_rate": 3.8718411552346575e-05,
      "loss": 0.4257,
      "step": 250
    },
    {
      "epoch": 0.4686795853988283,
      "grad_norm": 31.015850067138672,
      "learning_rate": 3.826714801444044e-05,
      "loss": 0.4463,
      "step": 260
    },
    {
      "epoch": 0.48670572329878325,
      "grad_norm": 18.325271606445312,
      "learning_rate": 3.78158844765343e-05,
      "loss": 0.4207,
      "step": 270
    },
    {
      "epoch": 0.5047318611987381,
      "grad_norm": 20.241615295410156,
      "learning_rate": 3.7364620938628155e-05,
      "loss": 0.4084,
      "step": 280
    },
    {
      "epoch": 0.5227579990986931,
      "grad_norm": 19.043285369873047,
      "learning_rate": 3.6913357400722025e-05,
      "loss": 0.4345,
      "step": 290
    },
    {
      "epoch": 0.540784136998648,
      "grad_norm": 19.568265914916992,
      "learning_rate": 3.646209386281589e-05,
      "loss": 0.4393,
      "step": 300
    },
    {
      "epoch": 0.558810274898603,
      "grad_norm": 26.981136322021484,
      "learning_rate": 3.601083032490975e-05,
      "loss": 0.4532,
      "step": 310
    },
    {
      "epoch": 0.5768364127985579,
      "grad_norm": 17.03878402709961,
      "learning_rate": 3.555956678700361e-05,
      "loss": 0.4208,
      "step": 320
    },
    {
      "epoch": 0.5948625506985128,
      "grad_norm": 16.008604049682617,
      "learning_rate": 3.5108303249097474e-05,
      "loss": 0.4396,
      "step": 330
    },
    {
      "epoch": 0.6128886885984678,
      "grad_norm": 20.344444274902344,
      "learning_rate": 3.4657039711191336e-05,
      "loss": 0.4318,
      "step": 340
    },
    {
      "epoch": 0.6309148264984227,
      "grad_norm": 18.641759872436523,
      "learning_rate": 3.42057761732852e-05,
      "loss": 0.4481,
      "step": 350
    },
    {
      "epoch": 0.6489409643983777,
      "grad_norm": 25.64444351196289,
      "learning_rate": 3.375451263537907e-05,
      "loss": 0.4053,
      "step": 360
    },
    {
      "epoch": 0.6669671022983326,
      "grad_norm": 23.601179122924805,
      "learning_rate": 3.330324909747293e-05,
      "loss": 0.4463,
      "step": 370
    },
    {
      "epoch": 0.6849932401982876,
      "grad_norm": 29.48953628540039,
      "learning_rate": 3.2851985559566786e-05,
      "loss": 0.4086,
      "step": 380
    },
    {
      "epoch": 0.7030193780982424,
      "grad_norm": 15.503927230834961,
      "learning_rate": 3.240072202166065e-05,
      "loss": 0.4402,
      "step": 390
    },
    {
      "epoch": 0.7210455159981974,
      "grad_norm": 29.575830459594727,
      "learning_rate": 3.194945848375451e-05,
      "loss": 0.43,
      "step": 400
    },
    {
      "epoch": 0.7390716538981523,
      "grad_norm": 26.154592514038086,
      "learning_rate": 3.149819494584837e-05,
      "loss": 0.432,
      "step": 410
    },
    {
      "epoch": 0.7570977917981072,
      "grad_norm": 25.37771987915039,
      "learning_rate": 3.104693140794224e-05,
      "loss": 0.4277,
      "step": 420
    },
    {
      "epoch": 0.7751239296980622,
      "grad_norm": 31.5076904296875,
      "learning_rate": 3.0595667870036104e-05,
      "loss": 0.4131,
      "step": 430
    },
    {
      "epoch": 0.7931500675980171,
      "grad_norm": 16.534198760986328,
      "learning_rate": 3.0144404332129967e-05,
      "loss": 0.433,
      "step": 440
    },
    {
      "epoch": 0.8111762054979721,
      "grad_norm": 19.064289093017578,
      "learning_rate": 2.969314079422383e-05,
      "loss": 0.3964,
      "step": 450
    },
    {
      "epoch": 0.829202343397927,
      "grad_norm": 18.9260196685791,
      "learning_rate": 2.924187725631769e-05,
      "loss": 0.396,
      "step": 460
    },
    {
      "epoch": 0.847228481297882,
      "grad_norm": 27.17948341369629,
      "learning_rate": 2.879061371841155e-05,
      "loss": 0.4079,
      "step": 470
    },
    {
      "epoch": 0.8652546191978369,
      "grad_norm": 21.348743438720703,
      "learning_rate": 2.8339350180505413e-05,
      "loss": 0.4545,
      "step": 480
    },
    {
      "epoch": 0.8832807570977917,
      "grad_norm": 15.41019344329834,
      "learning_rate": 2.7888086642599282e-05,
      "loss": 0.4024,
      "step": 490
    },
    {
      "epoch": 0.9013068949977467,
      "grad_norm": 26.81648826599121,
      "learning_rate": 2.7436823104693144e-05,
      "loss": 0.3888,
      "step": 500
    },
    {
      "epoch": 0.9193330328977016,
      "grad_norm": 27.93416976928711,
      "learning_rate": 2.6985559566787007e-05,
      "loss": 0.4086,
      "step": 510
    },
    {
      "epoch": 0.9373591707976566,
      "grad_norm": 14.353543281555176,
      "learning_rate": 2.6534296028880866e-05,
      "loss": 0.416,
      "step": 520
    },
    {
      "epoch": 0.9553853086976115,
      "grad_norm": 15.367998123168945,
      "learning_rate": 2.6083032490974728e-05,
      "loss": 0.4184,
      "step": 530
    },
    {
      "epoch": 0.9734114465975665,
      "grad_norm": 22.00202178955078,
      "learning_rate": 2.563176895306859e-05,
      "loss": 0.4294,
      "step": 540
    },
    {
      "epoch": 0.9914375844975214,
      "grad_norm": 15.376652717590332,
      "learning_rate": 2.518050541516246e-05,
      "loss": 0.4068,
      "step": 550
    },
    {
      "epoch": 1.0,
      "eval_f1": 0.8437814956661963,
      "eval_fn": 317,
      "eval_fp": 458,
      "eval_loss": 0.3667502701282501,
      "eval_precision": 0.8204625637005096,
      "eval_recall": 0.8684647302904565,
      "eval_runtime": 14.8642,
      "eval_samples_per_second": 314.379,
      "eval_steps_per_second": 39.356,
      "eval_tn": 1805,
      "eval_tp": 2093,
      "step": 555
    },
    {
      "epoch": 1.0090130689499774,
      "grad_norm": 33.98088455200195,
      "learning_rate": 2.472924187725632e-05,
      "loss": 0.3578,
      "step": 560
    },
    {
      "epoch": 1.0270392068499323,
      "grad_norm": 24.8944091796875,
      "learning_rate": 2.427797833935018e-05,
      "loss": 0.3203,
      "step": 570
    },
    {
      "epoch": 1.0450653447498874,
      "grad_norm": 17.574607849121094,
      "learning_rate": 2.3826714801444043e-05,
      "loss": 0.335,
      "step": 580
    },
    {
      "epoch": 1.0630914826498423,
      "grad_norm": 26.903356552124023,
      "learning_rate": 2.3375451263537906e-05,
      "loss": 0.289,
      "step": 590
    },
    {
      "epoch": 1.0811176205497972,
      "grad_norm": 31.987625122070312,
      "learning_rate": 2.292418772563177e-05,
      "loss": 0.3511,
      "step": 600
    },
    {
      "epoch": 1.099143758449752,
      "grad_norm": 29.54005241394043,
      "learning_rate": 2.2472924187725634e-05,
      "loss": 0.3126,
      "step": 610
    },
    {
      "epoch": 1.117169896349707,
      "grad_norm": 22.088634490966797,
      "learning_rate": 2.2021660649819496e-05,
      "loss": 0.3004,
      "step": 620
    },
    {
      "epoch": 1.135196034249662,
      "grad_norm": 30.805749893188477,
      "learning_rate": 2.157039711191336e-05,
      "loss": 0.3323,
      "step": 630
    },
    {
      "epoch": 1.153222172149617,
      "grad_norm": 16.607519149780273,
      "learning_rate": 2.111913357400722e-05,
      "loss": 0.3345,
      "step": 640
    },
    {
      "epoch": 1.1712483100495719,
      "grad_norm": 25.248517990112305,
      "learning_rate": 2.0667870036101083e-05,
      "loss": 0.3001,
      "step": 650
    },
    {
      "epoch": 1.1892744479495267,
      "grad_norm": 29.825464248657227,
      "learning_rate": 2.0216606498194946e-05,
      "loss": 0.3146,
      "step": 660
    },
    {
      "epoch": 1.2073005858494819,
      "grad_norm": 30.085296630859375,
      "learning_rate": 1.976534296028881e-05,
      "loss": 0.3335,
      "step": 670
    },
    {
      "epoch": 1.2253267237494367,
      "grad_norm": 23.349525451660156,
      "learning_rate": 1.9314079422382674e-05,
      "loss": 0.27,
      "step": 680
    },
    {
      "epoch": 1.2433528616493916,
      "grad_norm": 34.48153305053711,
      "learning_rate": 1.8862815884476533e-05,
      "loss": 0.3333,
      "step": 690
    },
    {
      "epoch": 1.2613789995493465,
      "grad_norm": 49.01852798461914,
      "learning_rate": 1.84115523465704e-05,
      "loss": 0.275,
      "step": 700
    },
    {
      "epoch": 1.2794051374493014,
      "grad_norm": 24.860200881958008,
      "learning_rate": 1.796028880866426e-05,
      "loss": 0.311,
      "step": 710
    },
    {
      "epoch": 1.2974312753492563,
      "grad_norm": 30.17133331298828,
      "learning_rate": 1.7509025270758123e-05,
      "loss": 0.2988,
      "step": 720
    },
    {
      "epoch": 1.3154574132492114,
      "grad_norm": 47.71062469482422,
      "learning_rate": 1.7057761732851986e-05,
      "loss": 0.2696,
      "step": 730
    },
    {
      "epoch": 1.3334835511491663,
      "grad_norm": 40.605262756347656,
      "learning_rate": 1.6606498194945848e-05,
      "loss": 0.2966,
      "step": 740
    },
    {
      "epoch": 1.3515096890491212,
      "grad_norm": 23.67529296875,
      "learning_rate": 1.615523465703971e-05,
      "loss": 0.2967,
      "step": 750
    },
    {
      "epoch": 1.3695358269490763,
      "grad_norm": 25.661808013916016,
      "learning_rate": 1.5703971119133576e-05,
      "loss": 0.2856,
      "step": 760
    },
    {
      "epoch": 1.3875619648490312,
      "grad_norm": 21.13004493713379,
      "learning_rate": 1.5252707581227438e-05,
      "loss": 0.3415,
      "step": 770
    },
    {
      "epoch": 1.405588102748986,
      "grad_norm": 24.510709762573242,
      "learning_rate": 1.4801444043321299e-05,
      "loss": 0.3044,
      "step": 780
    },
    {
      "epoch": 1.423614240648941,
      "grad_norm": 22.614412307739258,
      "learning_rate": 1.4350180505415165e-05,
      "loss": 0.3042,
      "step": 790
    },
    {
      "epoch": 1.4416403785488958,
      "grad_norm": 26.001209259033203,
      "learning_rate": 1.3898916967509026e-05,
      "loss": 0.2906,
      "step": 800
    },
    {
      "epoch": 1.4596665164488507,
      "grad_norm": 26.30026626586914,
      "learning_rate": 1.3447653429602888e-05,
      "loss": 0.2603,
      "step": 810
    },
    {
      "epoch": 1.4776926543488058,
      "grad_norm": 26.163490295410156,
      "learning_rate": 1.299638989169675e-05,
      "loss": 0.2768,
      "step": 820
    },
    {
      "epoch": 1.4957187922487607,
      "grad_norm": 48.625335693359375,
      "learning_rate": 1.2545126353790614e-05,
      "loss": 0.3203,
      "step": 830
    },
    {
      "epoch": 1.5137449301487156,
      "grad_norm": 19.079364776611328,
      "learning_rate": 1.2093862815884477e-05,
      "loss": 0.3069,
      "step": 840
    },
    {
      "epoch": 1.5317710680486707,
      "grad_norm": 21.295671463012695,
      "learning_rate": 1.164259927797834e-05,
      "loss": 0.2658,
      "step": 850
    },
    {
      "epoch": 1.5497972059486256,
      "grad_norm": 26.641357421875,
      "learning_rate": 1.1191335740072201e-05,
      "loss": 0.281,
      "step": 860
    },
    {
      "epoch": 1.5678233438485805,
      "grad_norm": 33.70462417602539,
      "learning_rate": 1.0740072202166066e-05,
      "loss": 0.2754,
      "step": 870
    },
    {
      "epoch": 1.5858494817485354,
      "grad_norm": 30.827627182006836,
      "learning_rate": 1.028880866425993e-05,
      "loss": 0.2913,
      "step": 880
    },
    {
      "epoch": 1.6038756196484902,
      "grad_norm": 19.456621170043945,
      "learning_rate": 9.83754512635379e-06,
      "loss": 0.3002,
      "step": 890
    },
    {
      "epoch": 1.6219017575484451,
      "grad_norm": 22.82554817199707,
      "learning_rate": 9.386281588447654e-06,
      "loss": 0.3197,
      "step": 900
    },
    {
      "epoch": 1.6399278954484002,
      "grad_norm": 22.25665283203125,
      "learning_rate": 8.935018050541517e-06,
      "loss": 0.3069,
      "step": 910
    },
    {
      "epoch": 1.6579540333483551,
      "grad_norm": 30.897809982299805,
      "learning_rate": 8.483754512635379e-06,
      "loss": 0.2965,
      "step": 920
    },
    {
      "epoch": 1.67598017124831,
      "grad_norm": 21.626707077026367,
      "learning_rate": 8.032490974729243e-06,
      "loss": 0.287,
      "step": 930
    },
    {
      "epoch": 1.694006309148265,
      "grad_norm": 32.338802337646484,
      "learning_rate": 7.581227436823105e-06,
      "loss": 0.2863,
      "step": 940
    },
    {
      "epoch": 1.71203244704822,
      "grad_norm": 21.633394241333008,
      "learning_rate": 7.129963898916968e-06,
      "loss": 0.2822,
      "step": 950
    },
    {
      "epoch": 1.7300585849481749,
      "grad_norm": 18.33571434020996,
      "learning_rate": 6.678700361010831e-06,
      "loss": 0.2676,
      "step": 960
    },
    {
      "epoch": 1.7480847228481298,
      "grad_norm": 29.97617530822754,
      "learning_rate": 6.227436823104693e-06,
      "loss": 0.2652,
      "step": 970
    },
    {
      "epoch": 1.7661108607480847,
      "grad_norm": 37.65591812133789,
      "learning_rate": 5.776173285198557e-06,
      "loss": 0.2956,
      "step": 980
    },
    {
      "epoch": 1.7841369986480395,
      "grad_norm": 33.256507873535156,
      "learning_rate": 5.324909747292419e-06,
      "loss": 0.3358,
      "step": 990
    },
    {
      "epoch": 1.8021631365479944,
      "grad_norm": 24.32988739013672,
      "learning_rate": 4.873646209386281e-06,
      "loss": 0.3075,
      "step": 1000
    },
    {
      "epoch": 1.8201892744479495,
      "grad_norm": 40.130863189697266,
      "learning_rate": 4.422382671480145e-06,
      "loss": 0.2682,
      "step": 1010
    },
    {
      "epoch": 1.8382154123479044,
      "grad_norm": 20.9071044921875,
      "learning_rate": 3.971119133574008e-06,
      "loss": 0.2849,
      "step": 1020
    },
    {
      "epoch": 1.8562415502478595,
      "grad_norm": 39.87733840942383,
      "learning_rate": 3.51985559566787e-06,
      "loss": 0.2967,
      "step": 1030
    },
    {
      "epoch": 1.8742676881478144,
      "grad_norm": 27.898658752441406,
      "learning_rate": 3.068592057761733e-06,
      "loss": 0.2846,
      "step": 1040
    },
    {
      "epoch": 1.8922938260477693,
      "grad_norm": 34.07933807373047,
      "learning_rate": 2.6173285198555957e-06,
      "loss": 0.2228,
      "step": 1050
    },
    {
      "epoch": 1.9103199639477242,
      "grad_norm": 20.768266677856445,
      "learning_rate": 2.1660649819494585e-06,
      "loss": 0.269,
      "step": 1060
    },
    {
      "epoch": 1.928346101847679,
      "grad_norm": 29.845983505249023,
      "learning_rate": 1.7148014440433215e-06,
      "loss": 0.2628,
      "step": 1070
    },
    {
      "epoch": 1.946372239747634,
      "grad_norm": 31.16872787475586,
      "learning_rate": 1.263537906137184e-06,
      "loss": 0.2821,
      "step": 1080
    },
    {
      "epoch": 1.9643983776475888,
      "grad_norm": 24.65738868713379,
      "learning_rate": 8.122743682310469e-07,
      "loss": 0.2961,
      "step": 1090
    },
    {
      "epoch": 1.982424515547544,
      "grad_norm": 33.531166076660156,
      "learning_rate": 3.610108303249098e-07,
      "loss": 0.2663,
      "step": 1100
    }
  ],
  "logging_steps": 10,
  "max_steps": 1108,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9326760690401280.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}