{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2001067235859125,
  "eval_steps": 500,
  "global_step": 2250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008893632159373888,
      "grad_norm": 68.5,
      "learning_rate": 2.7e-06,
      "loss": 8.8776,
      "step": 10
    },
    {
      "epoch": 0.0017787264318747777,
      "grad_norm": 39.0,
      "learning_rate": 5.7000000000000005e-06,
      "loss": 7.711,
      "step": 20
    },
    {
      "epoch": 0.0026680896478121665,
      "grad_norm": 29.375,
      "learning_rate": 8.7e-06,
      "loss": 5.7038,
      "step": 30
    },
    {
      "epoch": 0.0035574528637495554,
      "grad_norm": 24.5,
      "learning_rate": 1.1700000000000001e-05,
      "loss": 4.746,
      "step": 40
    },
    {
      "epoch": 0.004446816079686944,
      "grad_norm": 22.625,
      "learning_rate": 1.47e-05,
      "loss": 4.1094,
      "step": 50
    },
    {
      "epoch": 0.005336179295624333,
      "grad_norm": 21.625,
      "learning_rate": 1.77e-05,
      "loss": 4.0746,
      "step": 60
    },
    {
      "epoch": 0.0062255425115617215,
      "grad_norm": 25.25,
      "learning_rate": 2.07e-05,
      "loss": 3.8396,
      "step": 70
    },
    {
      "epoch": 0.007114905727499111,
      "grad_norm": 20.875,
      "learning_rate": 2.37e-05,
      "loss": 3.8629,
      "step": 80
    },
    {
      "epoch": 0.0080042689434365,
      "grad_norm": 22.875,
      "learning_rate": 2.6700000000000002e-05,
      "loss": 3.9097,
      "step": 90
    },
    {
      "epoch": 0.008893632159373888,
      "grad_norm": 19.375,
      "learning_rate": 2.97e-05,
      "loss": 3.763,
      "step": 100
    },
    {
      "epoch": 0.009782995375311278,
      "grad_norm": 19.625,
      "learning_rate": 2.9874418604651165e-05,
      "loss": 4.1501,
      "step": 110
    },
    {
      "epoch": 0.010672358591248666,
      "grad_norm": 19.0,
      "learning_rate": 2.9734883720930235e-05,
      "loss": 3.9038,
      "step": 120
    },
    {
      "epoch": 0.011561721807186055,
      "grad_norm": 16.625,
      "learning_rate": 2.9595348837209305e-05,
      "loss": 4.0583,
      "step": 130
    },
    {
      "epoch": 0.012451085023123443,
      "grad_norm": 16.375,
      "learning_rate": 2.9455813953488376e-05,
      "loss": 3.5958,
      "step": 140
    },
    {
      "epoch": 0.013340448239060833,
      "grad_norm": 15.375,
      "learning_rate": 2.9316279069767443e-05,
      "loss": 3.9075,
      "step": 150
    },
    {
      "epoch": 0.014229811454998222,
      "grad_norm": 15.625,
      "learning_rate": 2.9176744186046513e-05,
      "loss": 3.9495,
      "step": 160
    },
    {
      "epoch": 0.01511917467093561,
      "grad_norm": 18.125,
      "learning_rate": 2.9037209302325583e-05,
      "loss": 3.3247,
      "step": 170
    },
    {
      "epoch": 0.016008537886873,
      "grad_norm": 18.75,
      "learning_rate": 2.889767441860465e-05,
      "loss": 3.6863,
      "step": 180
    },
    {
      "epoch": 0.01689790110281039,
      "grad_norm": 15.625,
      "learning_rate": 2.875813953488372e-05,
      "loss": 3.3097,
      "step": 190
    },
    {
      "epoch": 0.017787264318747775,
      "grad_norm": 14.3125,
      "learning_rate": 2.861860465116279e-05,
      "loss": 3.48,
      "step": 200
    },
    {
      "epoch": 0.018676627534685165,
      "grad_norm": 17.875,
      "learning_rate": 2.847906976744186e-05,
      "loss": 3.6216,
      "step": 210
    },
    {
      "epoch": 0.019565990750622556,
      "grad_norm": 15.4375,
      "learning_rate": 2.833953488372093e-05,
      "loss": 3.4751,
      "step": 220
    },
    {
      "epoch": 0.020455353966559942,
      "grad_norm": 14.75,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 3.9,
      "step": 230
    },
    {
      "epoch": 0.021344717182497332,
      "grad_norm": 14.8125,
      "learning_rate": 2.8060465116279068e-05,
      "loss": 3.7454,
      "step": 240
    },
    {
      "epoch": 0.02223408039843472,
      "grad_norm": 15.6875,
      "learning_rate": 2.7920930232558138e-05,
      "loss": 3.5053,
      "step": 250
    },
    {
      "epoch": 0.02312344361437211,
      "grad_norm": 14.5,
      "learning_rate": 2.778139534883721e-05,
      "loss": 3.3656,
      "step": 260
    },
    {
      "epoch": 0.0240128068303095,
      "grad_norm": 13.6875,
      "learning_rate": 2.764186046511628e-05,
      "loss": 3.769,
      "step": 270
    },
    {
      "epoch": 0.024902170046246886,
      "grad_norm": 15.0625,
      "learning_rate": 2.750232558139535e-05,
      "loss": 3.5721,
      "step": 280
    },
    {
      "epoch": 0.025791533262184276,
      "grad_norm": 14.6875,
      "learning_rate": 2.736279069767442e-05,
      "loss": 3.6104,
      "step": 290
    },
    {
      "epoch": 0.026680896478121666,
      "grad_norm": 13.875,
      "learning_rate": 2.722325581395349e-05,
      "loss": 3.5904,
      "step": 300
    },
    {
      "epoch": 0.027570259694059053,
      "grad_norm": 13.8125,
      "learning_rate": 2.7083720930232556e-05,
      "loss": 3.5633,
      "step": 310
    },
    {
      "epoch": 0.028459622909996443,
      "grad_norm": 13.5625,
      "learning_rate": 2.6944186046511626e-05,
      "loss": 3.3366,
      "step": 320
    },
    {
      "epoch": 0.02934898612593383,
      "grad_norm": 14.3125,
      "learning_rate": 2.6804651162790697e-05,
      "loss": 3.6519,
      "step": 330
    },
    {
      "epoch": 0.03023834934187122,
      "grad_norm": 14.4375,
      "learning_rate": 2.6665116279069767e-05,
      "loss": 3.4482,
      "step": 340
    },
    {
      "epoch": 0.03112771255780861,
      "grad_norm": 13.9375,
      "learning_rate": 2.6525581395348837e-05,
      "loss": 3.3554,
      "step": 350
    },
    {
      "epoch": 0.032017075773746,
      "grad_norm": 14.5625,
      "learning_rate": 2.6386046511627907e-05,
      "loss": 3.4008,
      "step": 360
    },
    {
      "epoch": 0.03290643898968339,
      "grad_norm": 14.375,
      "learning_rate": 2.6246511627906978e-05,
      "loss": 3.3279,
      "step": 370
    },
    {
      "epoch": 0.03379580220562078,
      "grad_norm": 13.9375,
      "learning_rate": 2.6106976744186048e-05,
      "loss": 3.487,
      "step": 380
    },
    {
      "epoch": 0.03468516542155817,
      "grad_norm": 14.625,
      "learning_rate": 2.5967441860465115e-05,
      "loss": 3.2542,
      "step": 390
    },
    {
      "epoch": 0.03557452863749555,
      "grad_norm": 13.1875,
      "learning_rate": 2.5827906976744185e-05,
      "loss": 3.3328,
      "step": 400
    },
    {
      "epoch": 0.03646389185343294,
      "grad_norm": 13.25,
      "learning_rate": 2.5688372093023255e-05,
      "loss": 3.5205,
      "step": 410
    },
    {
      "epoch": 0.03735325506937033,
      "grad_norm": 11.9375,
      "learning_rate": 2.5548837209302325e-05,
      "loss": 3.4151,
      "step": 420
    },
    {
      "epoch": 0.03824261828530772,
      "grad_norm": 13.625,
      "learning_rate": 2.5409302325581396e-05,
      "loss": 3.1404,
      "step": 430
    },
    {
      "epoch": 0.03913198150124511,
      "grad_norm": 14.0625,
      "learning_rate": 2.5269767441860466e-05,
      "loss": 3.5277,
      "step": 440
    },
    {
      "epoch": 0.040021344717182494,
      "grad_norm": 13.1875,
      "learning_rate": 2.5130232558139536e-05,
      "loss": 3.1584,
      "step": 450
    },
    {
      "epoch": 0.040910707933119884,
      "grad_norm": 13.4375,
      "learning_rate": 2.4990697674418606e-05,
      "loss": 3.4669,
      "step": 460
    },
    {
      "epoch": 0.041800071149057275,
      "grad_norm": 14.5625,
      "learning_rate": 2.4851162790697673e-05,
      "loss": 3.2653,
      "step": 470
    },
    {
      "epoch": 0.042689434364994665,
      "grad_norm": 13.9375,
      "learning_rate": 2.4711627906976743e-05,
      "loss": 3.4219,
      "step": 480
    },
    {
      "epoch": 0.043578797580932055,
      "grad_norm": 12.875,
      "learning_rate": 2.4572093023255814e-05,
      "loss": 3.508,
      "step": 490
    },
    {
      "epoch": 0.04446816079686944,
      "grad_norm": 14.0625,
      "learning_rate": 2.4432558139534884e-05,
      "loss": 3.3343,
      "step": 500
    },
    {
      "epoch": 0.04535752401280683,
      "grad_norm": 13.75,
      "learning_rate": 2.4293023255813954e-05,
      "loss": 3.5054,
      "step": 510
    },
    {
      "epoch": 0.04624688722874422,
      "grad_norm": 15.25,
      "learning_rate": 2.4153488372093024e-05,
      "loss": 3.0298,
      "step": 520
    },
    {
      "epoch": 0.04713625044468161,
      "grad_norm": 13.75,
      "learning_rate": 2.4013953488372095e-05,
      "loss": 3.4306,
      "step": 530
    },
    {
      "epoch": 0.048025613660619,
      "grad_norm": 14.25,
      "learning_rate": 2.3874418604651165e-05,
      "loss": 3.1824,
      "step": 540
    },
    {
      "epoch": 0.04891497687655639,
      "grad_norm": 12.3125,
      "learning_rate": 2.373488372093023e-05,
      "loss": 3.211,
      "step": 550
    },
    {
      "epoch": 0.04980434009249377,
      "grad_norm": 13.75,
      "learning_rate": 2.3595348837209302e-05,
      "loss": 3.091,
      "step": 560
    },
    {
      "epoch": 0.05069370330843116,
      "grad_norm": 13.75,
      "learning_rate": 2.3455813953488372e-05,
      "loss": 3.234,
      "step": 570
    },
    {
      "epoch": 0.05158306652436855,
      "grad_norm": 13.75,
      "learning_rate": 2.3316279069767442e-05,
      "loss": 3.1404,
      "step": 580
    },
    {
      "epoch": 0.05247242974030594,
      "grad_norm": 13.25,
      "learning_rate": 2.3176744186046513e-05,
      "loss": 3.2191,
      "step": 590
    },
    {
      "epoch": 0.05336179295624333,
      "grad_norm": 12.625,
      "learning_rate": 2.3037209302325583e-05,
      "loss": 3.0968,
      "step": 600
    },
    {
      "epoch": 0.054251156172180716,
      "grad_norm": 13.5625,
      "learning_rate": 2.2897674418604653e-05,
      "loss": 3.1108,
      "step": 610
    },
    {
      "epoch": 0.055140519388118106,
      "grad_norm": 12.75,
      "learning_rate": 2.2758139534883723e-05,
      "loss": 3.028,
      "step": 620
    },
    {
      "epoch": 0.056029882604055496,
      "grad_norm": 12.375,
      "learning_rate": 2.261860465116279e-05,
      "loss": 3.1469,
      "step": 630
    },
    {
      "epoch": 0.056919245819992886,
      "grad_norm": 14.8125,
      "learning_rate": 2.247906976744186e-05,
      "loss": 3.1258,
      "step": 640
    },
    {
      "epoch": 0.057808609035930276,
      "grad_norm": 12.5625,
      "learning_rate": 2.233953488372093e-05,
      "loss": 3.1147,
      "step": 650
    },
    {
      "epoch": 0.05869797225186766,
      "grad_norm": 13.1875,
      "learning_rate": 2.22e-05,
      "loss": 3.0447,
      "step": 660
    },
    {
      "epoch": 0.05958733546780505,
      "grad_norm": 15.3125,
      "learning_rate": 2.206046511627907e-05,
      "loss": 3.1177,
      "step": 670
    },
    {
      "epoch": 0.06047669868374244,
      "grad_norm": 13.6875,
      "learning_rate": 2.192093023255814e-05,
      "loss": 3.2542,
      "step": 680
    },
    {
      "epoch": 0.06136606189967983,
      "grad_norm": 13.1875,
      "learning_rate": 2.178139534883721e-05,
      "loss": 3.188,
      "step": 690
    },
    {
      "epoch": 0.06225542511561722,
      "grad_norm": 13.625,
      "learning_rate": 2.1641860465116282e-05,
      "loss": 3.2298,
      "step": 700
    },
    {
      "epoch": 0.06314478833155461,
      "grad_norm": 14.0625,
      "learning_rate": 2.150232558139535e-05,
      "loss": 3.1127,
      "step": 710
    },
    {
      "epoch": 0.064034151547492,
      "grad_norm": 15.1875,
      "learning_rate": 2.136279069767442e-05,
      "loss": 3.1468,
      "step": 720
    },
    {
      "epoch": 0.06492351476342939,
      "grad_norm": 12.625,
      "learning_rate": 2.122325581395349e-05,
      "loss": 3.2763,
      "step": 730
    },
    {
      "epoch": 0.06581287797936677,
      "grad_norm": 13.875,
      "learning_rate": 2.108372093023256e-05,
      "loss": 3.051,
      "step": 740
    },
    {
      "epoch": 0.06670224119530416,
      "grad_norm": 13.9375,
      "learning_rate": 2.094418604651163e-05,
      "loss": 3.1937,
      "step": 750
    },
    {
      "epoch": 0.06759160441124155,
      "grad_norm": 13.3125,
      "learning_rate": 2.08046511627907e-05,
      "loss": 3.0506,
      "step": 760
    },
    {
      "epoch": 0.06848096762717894,
      "grad_norm": 12.625,
      "learning_rate": 2.066511627906977e-05,
      "loss": 2.975,
      "step": 770
    },
    {
      "epoch": 0.06937033084311633,
      "grad_norm": 12.75,
      "learning_rate": 2.052558139534884e-05,
      "loss": 2.9174,
      "step": 780
    },
    {
      "epoch": 0.07025969405905372,
      "grad_norm": 12.875,
      "learning_rate": 2.0386046511627907e-05,
      "loss": 3.0532,
      "step": 790
    },
    {
      "epoch": 0.0711490572749911,
      "grad_norm": 12.4375,
      "learning_rate": 2.0246511627906977e-05,
      "loss": 3.2092,
      "step": 800
    },
    {
      "epoch": 0.0720384204909285,
      "grad_norm": 12.625,
      "learning_rate": 2.0106976744186048e-05,
      "loss": 2.9702,
      "step": 810
    },
    {
      "epoch": 0.07292778370686588,
      "grad_norm": 14.0625,
      "learning_rate": 1.9967441860465118e-05,
      "loss": 3.0595,
      "step": 820
    },
    {
      "epoch": 0.07381714692280328,
      "grad_norm": 14.5,
      "learning_rate": 1.9827906976744188e-05,
      "loss": 3.0102,
      "step": 830
    },
    {
      "epoch": 0.07470651013874066,
      "grad_norm": 13.25,
      "learning_rate": 1.968837209302326e-05,
      "loss": 3.0725,
      "step": 840
    },
    {
      "epoch": 0.07559587335467804,
      "grad_norm": 13.6875,
      "learning_rate": 1.954883720930233e-05,
      "loss": 3.0679,
      "step": 850
    },
    {
      "epoch": 0.07648523657061544,
      "grad_norm": 13.1875,
      "learning_rate": 1.94093023255814e-05,
      "loss": 3.0764,
      "step": 860
    },
    {
      "epoch": 0.07737459978655283,
      "grad_norm": 14.5,
      "learning_rate": 1.9269767441860466e-05,
      "loss": 2.768,
      "step": 870
    },
    {
      "epoch": 0.07826396300249022,
      "grad_norm": 14.875,
      "learning_rate": 1.9130232558139536e-05,
      "loss": 3.2016,
      "step": 880
    },
    {
      "epoch": 0.0791533262184276,
      "grad_norm": 12.875,
      "learning_rate": 1.8990697674418606e-05,
      "loss": 3.0906,
      "step": 890
    },
    {
      "epoch": 0.08004268943436499,
      "grad_norm": 13.0,
      "learning_rate": 1.8851162790697673e-05,
      "loss": 3.1441,
      "step": 900
    },
    {
      "epoch": 0.08093205265030239,
      "grad_norm": 13.25,
      "learning_rate": 1.8711627906976743e-05,
      "loss": 2.9791,
      "step": 910
    },
    {
      "epoch": 0.08182141586623977,
      "grad_norm": 14.0625,
      "learning_rate": 1.8572093023255814e-05,
      "loss": 2.8393,
      "step": 920
    },
    {
      "epoch": 0.08271077908217717,
      "grad_norm": 13.1875,
      "learning_rate": 1.8432558139534884e-05,
      "loss": 2.9332,
      "step": 930
    },
    {
      "epoch": 0.08360014229811455,
      "grad_norm": 13.0625,
      "learning_rate": 1.8293023255813954e-05,
      "loss": 2.9787,
      "step": 940
    },
    {
      "epoch": 0.08448950551405193,
      "grad_norm": 15.375,
      "learning_rate": 1.815348837209302e-05,
      "loss": 2.8229,
      "step": 950
    },
    {
      "epoch": 0.08537886872998933,
      "grad_norm": 13.5625,
      "learning_rate": 1.801395348837209e-05,
      "loss": 2.8112,
      "step": 960
    },
    {
      "epoch": 0.08626823194592671,
      "grad_norm": 16.75,
      "learning_rate": 1.787441860465116e-05,
      "loss": 3.0131,
      "step": 970
    },
    {
      "epoch": 0.08715759516186411,
      "grad_norm": 13.9375,
      "learning_rate": 1.773488372093023e-05,
      "loss": 2.9496,
      "step": 980
    },
    {
      "epoch": 0.08804695837780149,
      "grad_norm": 15.0625,
      "learning_rate": 1.7595348837209302e-05,
      "loss": 3.0641,
      "step": 990
    },
    {
      "epoch": 0.08893632159373888,
      "grad_norm": 13.0,
      "learning_rate": 1.7455813953488372e-05,
      "loss": 2.8714,
      "step": 1000
    },
    {
      "epoch": 0.08982568480967627,
      "grad_norm": 13.8125,
      "learning_rate": 1.7316279069767442e-05,
      "loss": 2.7792,
      "step": 1010
    },
    {
      "epoch": 0.09071504802561366,
      "grad_norm": 13.875,
      "learning_rate": 1.7176744186046512e-05,
      "loss": 2.9472,
      "step": 1020
    },
    {
      "epoch": 0.09160441124155105,
      "grad_norm": 14.0,
      "learning_rate": 1.703720930232558e-05,
      "loss": 2.5662,
      "step": 1030
    },
    {
      "epoch": 0.09249377445748844,
      "grad_norm": 13.9375,
      "learning_rate": 1.689767441860465e-05,
      "loss": 2.82,
      "step": 1040
    },
    {
      "epoch": 0.09338313767342583,
      "grad_norm": 13.875,
      "learning_rate": 1.675813953488372e-05,
      "loss": 2.8245,
      "step": 1050
    },
    {
      "epoch": 0.09427250088936322,
      "grad_norm": 11.8125,
      "learning_rate": 1.661860465116279e-05,
      "loss": 2.8141,
      "step": 1060
    },
    {
      "epoch": 0.0951618641053006,
      "grad_norm": 13.1875,
      "learning_rate": 1.647906976744186e-05,
      "loss": 2.8975,
      "step": 1070
    },
    {
      "epoch": 0.096051227321238,
      "grad_norm": 14.625,
      "learning_rate": 1.633953488372093e-05,
      "loss": 2.7996,
      "step": 1080
    },
    {
      "epoch": 0.09694059053717538,
      "grad_norm": 12.625,
      "learning_rate": 1.62e-05,
      "loss": 2.7679,
      "step": 1090
    },
    {
      "epoch": 0.09782995375311278,
      "grad_norm": 14.0625,
      "learning_rate": 1.606046511627907e-05,
      "loss": 3.0057,
      "step": 1100
    },
    {
      "epoch": 0.09871931696905016,
      "grad_norm": 13.6875,
      "learning_rate": 1.5920930232558138e-05,
      "loss": 2.5742,
      "step": 1110
    },
    {
      "epoch": 0.09960868018498754,
      "grad_norm": 15.3125,
      "learning_rate": 1.5781395348837208e-05,
      "loss": 2.7903,
      "step": 1120
    },
    {
      "epoch": 0.10049804340092494,
      "grad_norm": 14.8125,
      "learning_rate": 1.564186046511628e-05,
      "loss": 2.5664,
      "step": 1130
    },
    {
      "epoch": 0.10138740661686232,
      "grad_norm": 13.8125,
      "learning_rate": 1.550232558139535e-05,
      "loss": 2.9392,
      "step": 1140
    },
    {
      "epoch": 0.10227676983279972,
      "grad_norm": 13.9375,
      "learning_rate": 1.536279069767442e-05,
      "loss": 2.7105,
      "step": 1150
    },
    {
      "epoch": 0.1031661330487371,
      "grad_norm": 14.5,
      "learning_rate": 1.5223255813953489e-05,
      "loss": 2.9472,
      "step": 1160
    },
    {
      "epoch": 0.10405549626467449,
      "grad_norm": 13.4375,
      "learning_rate": 1.508372093023256e-05,
      "loss": 2.7698,
      "step": 1170
    },
    {
      "epoch": 0.10494485948061189,
      "grad_norm": 13.125,
      "learning_rate": 1.4944186046511628e-05,
      "loss": 2.8998,
      "step": 1180
    },
    {
      "epoch": 0.10583422269654927,
      "grad_norm": 14.125,
      "learning_rate": 1.4804651162790698e-05,
      "loss": 2.7952,
      "step": 1190
    },
    {
      "epoch": 0.10672358591248667,
      "grad_norm": 14.875,
      "learning_rate": 1.4665116279069768e-05,
      "loss": 2.7689,
      "step": 1200
    },
    {
      "epoch": 0.10761294912842405,
      "grad_norm": 13.375,
      "learning_rate": 1.4525581395348837e-05,
      "loss": 3.0368,
      "step": 1210
    },
    {
      "epoch": 0.10850231234436143,
      "grad_norm": 14.5,
      "learning_rate": 1.4386046511627907e-05,
      "loss": 2.8113,
      "step": 1220
    },
    {
      "epoch": 0.10939167556029883,
      "grad_norm": 13.25,
      "learning_rate": 1.4246511627906977e-05,
      "loss": 2.5883,
      "step": 1230
    },
    {
      "epoch": 0.11028103877623621,
      "grad_norm": 14.25,
      "learning_rate": 1.4106976744186048e-05,
      "loss": 2.9207,
      "step": 1240
    },
    {
      "epoch": 0.11117040199217361,
      "grad_norm": 12.9375,
      "learning_rate": 1.3967441860465116e-05,
      "loss": 2.8662,
      "step": 1250
    },
    {
      "epoch": 0.11205976520811099,
      "grad_norm": 14.25,
      "learning_rate": 1.3827906976744186e-05,
      "loss": 2.8439,
      "step": 1260
    },
    {
      "epoch": 0.11294912842404838,
      "grad_norm": 13.1875,
      "learning_rate": 1.3688372093023257e-05,
      "loss": 2.8322,
      "step": 1270
    },
    {
      "epoch": 0.11383849163998577,
      "grad_norm": 13.6875,
      "learning_rate": 1.3548837209302327e-05,
      "loss": 2.8627,
      "step": 1280
    },
    {
      "epoch": 0.11472785485592316,
      "grad_norm": 12.8125,
      "learning_rate": 1.3409302325581395e-05,
      "loss": 2.733,
      "step": 1290
    },
    {
      "epoch": 0.11561721807186055,
      "grad_norm": 12.75,
      "learning_rate": 1.3269767441860466e-05,
      "loss": 2.7608,
      "step": 1300
    },
    {
      "epoch": 0.11650658128779794,
      "grad_norm": 14.4375,
      "learning_rate": 1.3130232558139536e-05,
      "loss": 2.7633,
      "step": 1310
    },
    {
      "epoch": 0.11739594450373532,
      "grad_norm": 13.875,
      "learning_rate": 1.2990697674418606e-05,
      "loss": 2.682,
      "step": 1320
    },
    {
      "epoch": 0.11828530771967272,
      "grad_norm": 14.4375,
      "learning_rate": 1.2851162790697675e-05,
      "loss": 2.7205,
      "step": 1330
    },
    {
      "epoch": 0.1191746709356101,
      "grad_norm": 12.6875,
      "learning_rate": 1.2711627906976745e-05,
      "loss": 2.7408,
      "step": 1340
    },
    {
      "epoch": 0.1200640341515475,
      "grad_norm": 13.5,
      "learning_rate": 1.2572093023255815e-05,
      "loss": 3.0955,
      "step": 1350
    },
    {
      "epoch": 0.12095339736748488,
      "grad_norm": 13.8125,
      "learning_rate": 1.2432558139534885e-05,
      "loss": 2.8928,
      "step": 1360
    },
    {
      "epoch": 0.12184276058342226,
      "grad_norm": 12.5625,
      "learning_rate": 1.2293023255813954e-05,
      "loss": 2.7429,
      "step": 1370
    },
    {
      "epoch": 0.12273212379935966,
      "grad_norm": 14.3125,
      "learning_rate": 1.2153488372093024e-05,
      "loss": 2.6795,
      "step": 1380
    },
    {
      "epoch": 0.12362148701529704,
      "grad_norm": 13.125,
      "learning_rate": 1.2013953488372094e-05,
      "loss": 2.6802,
      "step": 1390
    },
    {
      "epoch": 0.12451085023123444,
      "grad_norm": 13.8125,
      "learning_rate": 1.1874418604651165e-05,
      "loss": 2.6818,
      "step": 1400
    },
    {
      "epoch": 0.12540021344717184,
      "grad_norm": 13.875,
      "learning_rate": 1.1734883720930233e-05,
      "loss": 2.5968,
      "step": 1410
    },
    {
      "epoch": 0.12628957666310922,
      "grad_norm": 13.75,
      "learning_rate": 1.1595348837209303e-05,
      "loss": 2.7026,
      "step": 1420
    },
    {
      "epoch": 0.1271789398790466,
      "grad_norm": 13.5625,
      "learning_rate": 1.1455813953488372e-05,
      "loss": 2.8092,
      "step": 1430
    },
    {
      "epoch": 0.128068303094984,
      "grad_norm": 13.9375,
      "learning_rate": 1.1316279069767442e-05,
      "loss": 2.6957,
      "step": 1440
    },
    {
      "epoch": 0.12895766631092137,
      "grad_norm": 14.0,
      "learning_rate": 1.117674418604651e-05,
      "loss": 2.8901,
      "step": 1450
    },
    {
      "epoch": 0.12984702952685878,
      "grad_norm": 14.3125,
      "learning_rate": 1.1037209302325581e-05,
      "loss": 2.6284,
      "step": 1460
    },
    {
      "epoch": 0.13073639274279616,
      "grad_norm": 13.875,
      "learning_rate": 1.0897674418604651e-05,
      "loss": 2.9493,
      "step": 1470
    },
    {
      "epoch": 0.13162575595873355,
      "grad_norm": 13.5625,
      "learning_rate": 1.0758139534883721e-05,
      "loss": 2.6665,
      "step": 1480
    },
    {
      "epoch": 0.13251511917467093,
      "grad_norm": 14.125,
      "learning_rate": 1.061860465116279e-05,
      "loss": 2.8124,
      "step": 1490
    },
    {
      "epoch": 0.13340448239060831,
      "grad_norm": 13.8125,
      "learning_rate": 1.047906976744186e-05,
      "loss": 2.8094,
      "step": 1500
    },
    {
      "epoch": 0.13429384560654573,
      "grad_norm": 14.3125,
      "learning_rate": 1.033953488372093e-05,
      "loss": 2.6246,
      "step": 1510
    },
    {
      "epoch": 0.1351832088224831,
      "grad_norm": 14.5625,
      "learning_rate": 1.02e-05,
      "loss": 2.8486,
      "step": 1520
    },
    {
      "epoch": 0.1360725720384205,
      "grad_norm": 13.625,
      "learning_rate": 1.0060465116279069e-05,
      "loss": 2.7712,
      "step": 1530
    },
    {
      "epoch": 0.13696193525435787,
      "grad_norm": 14.5625,
      "learning_rate": 9.92093023255814e-06,
      "loss": 2.7667,
      "step": 1540
    },
    {
      "epoch": 0.13785129847029526,
      "grad_norm": 14.375,
      "learning_rate": 9.78139534883721e-06,
      "loss": 2.6918,
      "step": 1550
    },
    {
      "epoch": 0.13874066168623267,
      "grad_norm": 14.1875,
      "learning_rate": 9.64186046511628e-06,
      "loss": 2.872,
      "step": 1560
    },
    {
      "epoch": 0.13963002490217005,
      "grad_norm": 14.375,
      "learning_rate": 9.502325581395348e-06,
      "loss": 2.9586,
      "step": 1570
    },
    {
      "epoch": 0.14051938811810744,
      "grad_norm": 13.8125,
      "learning_rate": 9.362790697674419e-06,
      "loss": 2.5803,
      "step": 1580
    },
    {
      "epoch": 0.14140875133404482,
      "grad_norm": 13.8125,
      "learning_rate": 9.223255813953489e-06,
      "loss": 2.72,
      "step": 1590
    },
    {
      "epoch": 0.1422981145499822,
      "grad_norm": 13.4375,
      "learning_rate": 9.083720930232559e-06,
      "loss": 2.7268,
      "step": 1600
    },
    {
      "epoch": 0.1431874777659196,
      "grad_norm": 14.0625,
      "learning_rate": 8.944186046511628e-06,
      "loss": 2.7176,
      "step": 1610
    },
    {
      "epoch": 0.144076840981857,
      "grad_norm": 15.125,
      "learning_rate": 8.804651162790698e-06,
      "loss": 2.6748,
      "step": 1620
    },
    {
      "epoch": 0.14496620419779438,
      "grad_norm": 15.25,
      "learning_rate": 8.665116279069768e-06,
      "loss": 2.7349,
      "step": 1630
    },
    {
      "epoch": 0.14585556741373176,
      "grad_norm": 15.0625,
      "learning_rate": 8.525581395348838e-06,
      "loss": 2.729,
      "step": 1640
    },
    {
      "epoch": 0.14674493062966915,
      "grad_norm": 14.5625,
      "learning_rate": 8.386046511627907e-06,
      "loss": 2.6542,
      "step": 1650
    },
    {
      "epoch": 0.14763429384560656,
      "grad_norm": 14.625,
      "learning_rate": 8.246511627906977e-06,
      "loss": 2.6102,
      "step": 1660
    },
    {
      "epoch": 0.14852365706154394,
      "grad_norm": 13.8125,
      "learning_rate": 8.106976744186047e-06,
      "loss": 2.7355,
      "step": 1670
    },
    {
      "epoch": 0.14941302027748132,
      "grad_norm": 14.1875,
      "learning_rate": 7.967441860465118e-06,
      "loss": 2.7689,
      "step": 1680
    },
    {
      "epoch": 0.1503023834934187,
      "grad_norm": 14.3125,
      "learning_rate": 7.827906976744186e-06,
      "loss": 2.7373,
      "step": 1690
    },
    {
      "epoch": 0.1511917467093561,
      "grad_norm": 15.25,
      "learning_rate": 7.688372093023256e-06,
      "loss": 2.7359,
      "step": 1700
    },
    {
      "epoch": 0.1520811099252935,
      "grad_norm": 15.0,
      "learning_rate": 7.548837209302326e-06,
      "loss": 2.6088,
      "step": 1710
    },
    {
      "epoch": 0.15297047314123088,
      "grad_norm": 14.4375,
      "learning_rate": 7.409302325581395e-06,
      "loss": 2.7727,
      "step": 1720
    },
    {
      "epoch": 0.15385983635716827,
      "grad_norm": 13.1875,
      "learning_rate": 7.269767441860465e-06,
      "loss": 2.8457,
      "step": 1730
    },
    {
      "epoch": 0.15474919957310565,
      "grad_norm": 14.0625,
      "learning_rate": 7.130232558139535e-06,
      "loss": 2.6671,
      "step": 1740
    },
    {
      "epoch": 0.15563856278904303,
      "grad_norm": 14.0,
      "learning_rate": 6.990697674418605e-06,
      "loss": 2.593,
      "step": 1750
    },
    {
      "epoch": 0.15652792600498044,
      "grad_norm": 13.6875,
      "learning_rate": 6.851162790697674e-06,
      "loss": 2.6174,
      "step": 1760
    },
    {
      "epoch": 0.15741728922091783,
      "grad_norm": 14.375,
      "learning_rate": 6.711627906976745e-06,
      "loss": 2.6143,
      "step": 1770
    },
    {
      "epoch": 0.1583066524368552,
      "grad_norm": 12.875,
      "learning_rate": 6.572093023255814e-06,
      "loss": 2.7958,
      "step": 1780
    },
    {
      "epoch": 0.1591960156527926,
      "grad_norm": 13.875,
      "learning_rate": 6.432558139534884e-06,
      "loss": 2.6567,
      "step": 1790
    },
    {
      "epoch": 0.16008537886872998,
      "grad_norm": 14.625,
      "learning_rate": 6.293023255813954e-06,
      "loss": 2.5427,
      "step": 1800
    },
    {
      "epoch": 0.1609747420846674,
      "grad_norm": 14.1875,
      "learning_rate": 6.153488372093024e-06,
      "loss": 2.6013,
      "step": 1810
    },
    {
      "epoch": 0.16186410530060477,
      "grad_norm": 14.8125,
      "learning_rate": 6.013953488372093e-06,
      "loss": 2.6624,
      "step": 1820
    },
    {
      "epoch": 0.16275346851654215,
      "grad_norm": 14.625,
      "learning_rate": 5.8744186046511635e-06,
      "loss": 2.6861,
      "step": 1830
    },
    {
      "epoch": 0.16364283173247954,
      "grad_norm": 12.875,
      "learning_rate": 5.734883720930233e-06,
      "loss": 2.7337,
      "step": 1840
    },
    {
      "epoch": 0.16453219494841692,
      "grad_norm": 13.25,
      "learning_rate": 5.595348837209303e-06,
      "loss": 2.532,
      "step": 1850
    },
    {
      "epoch": 0.16542155816435433,
      "grad_norm": 14.0625,
      "learning_rate": 5.4558139534883726e-06,
      "loss": 2.5045,
      "step": 1860
    },
    {
      "epoch": 0.16631092138029172,
      "grad_norm": 14.5625,
      "learning_rate": 5.316279069767443e-06,
      "loss": 2.8413,
      "step": 1870
    },
    {
      "epoch": 0.1672002845962291,
      "grad_norm": 14.8125,
      "learning_rate": 5.176744186046511e-06,
      "loss": 2.6274,
      "step": 1880
    },
    {
      "epoch": 0.16808964781216648,
      "grad_norm": 13.5625,
      "learning_rate": 5.0372093023255816e-06,
      "loss": 2.5649,
      "step": 1890
    },
    {
      "epoch": 0.16897901102810386,
      "grad_norm": 15.0625,
      "learning_rate": 4.897674418604651e-06,
      "loss": 2.7143,
      "step": 1900
    },
    {
      "epoch": 0.16986837424404128,
      "grad_norm": 15.125,
      "learning_rate": 4.758139534883721e-06,
      "loss": 2.642,
      "step": 1910
    },
    {
      "epoch": 0.17075773745997866,
      "grad_norm": 12.875,
      "learning_rate": 4.618604651162791e-06,
      "loss": 2.8059,
      "step": 1920
    },
    {
      "epoch": 0.17164710067591604,
      "grad_norm": 13.8125,
      "learning_rate": 4.479069767441861e-06,
      "loss": 2.6971,
      "step": 1930
    },
    {
      "epoch": 0.17253646389185343,
      "grad_norm": 13.6875,
      "learning_rate": 4.33953488372093e-06,
      "loss": 2.6398,
      "step": 1940
    },
    {
      "epoch": 0.1734258271077908,
      "grad_norm": 16.375,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 2.9306,
      "step": 1950
    },
    {
      "epoch": 0.17431519032372822,
      "grad_norm": 13.8125,
      "learning_rate": 4.06046511627907e-06,
      "loss": 2.7713,
      "step": 1960
    },
    {
      "epoch": 0.1752045535396656,
      "grad_norm": 13.125,
      "learning_rate": 3.92093023255814e-06,
      "loss": 2.7845,
      "step": 1970
    },
    {
      "epoch": 0.17609391675560299,
      "grad_norm": 13.5,
      "learning_rate": 3.7813953488372095e-06,
      "loss": 2.5285,
      "step": 1980
    },
    {
      "epoch": 0.17698327997154037,
      "grad_norm": 14.75,
      "learning_rate": 3.6418604651162793e-06,
      "loss": 2.7907,
      "step": 1990
    },
    {
      "epoch": 0.17787264318747775,
      "grad_norm": 14.25,
      "learning_rate": 3.502325581395349e-06,
      "loss": 2.6383,
      "step": 2000
    },
    {
      "epoch": 0.17876200640341516,
      "grad_norm": 13.4375,
      "learning_rate": 3.3627906976744185e-06,
      "loss": 2.62,
      "step": 2010
    },
    {
      "epoch": 0.17965136961935255,
      "grad_norm": 16.0,
      "learning_rate": 3.2232558139534883e-06,
      "loss": 2.6536,
      "step": 2020
    },
    {
      "epoch": 0.18054073283528993,
      "grad_norm": 13.3125,
      "learning_rate": 3.083720930232558e-06,
      "loss": 2.6178,
      "step": 2030
    },
    {
      "epoch": 0.1814300960512273,
      "grad_norm": 15.0,
      "learning_rate": 2.944186046511628e-06,
      "loss": 2.7065,
      "step": 2040
    },
    {
      "epoch": 0.1823194592671647,
      "grad_norm": 12.375,
      "learning_rate": 2.8046511627906977e-06,
      "loss": 2.6898,
      "step": 2050
    },
    {
      "epoch": 0.1832088224831021,
      "grad_norm": 14.0625,
      "learning_rate": 2.6651162790697675e-06,
      "loss": 2.5896,
      "step": 2060
    },
    {
      "epoch": 0.1840981856990395,
      "grad_norm": 15.4375,
      "learning_rate": 2.5255813953488374e-06,
      "loss": 2.6833,
      "step": 2070
    },
    {
      "epoch": 0.18498754891497687,
      "grad_norm": 15.5,
      "learning_rate": 2.386046511627907e-06,
      "loss": 2.7408,
      "step": 2080
    },
    {
      "epoch": 0.18587691213091426,
      "grad_norm": 13.375,
      "learning_rate": 2.246511627906977e-06,
      "loss": 2.7091,
      "step": 2090
    },
    {
      "epoch": 0.18676627534685167,
      "grad_norm": 14.1875,
      "learning_rate": 2.1069767441860464e-06,
      "loss": 2.4644,
      "step": 2100
    },
    {
      "epoch": 0.18765563856278905,
      "grad_norm": 14.625,
      "learning_rate": 1.967441860465116e-06,
      "loss": 2.7544,
      "step": 2110
    },
    {
      "epoch": 0.18854500177872643,
      "grad_norm": 14.625,
      "learning_rate": 1.8279069767441862e-06,
      "loss": 2.7613,
      "step": 2120
    },
    {
      "epoch": 0.18943436499466382,
      "grad_norm": 14.0625,
      "learning_rate": 1.6883720930232558e-06,
      "loss": 2.5281,
      "step": 2130
    },
    {
      "epoch": 0.1903237282106012,
      "grad_norm": 14.1875,
      "learning_rate": 1.5488372093023256e-06,
      "loss": 2.56,
      "step": 2140
    },
    {
      "epoch": 0.1912130914265386,
      "grad_norm": 14.0,
      "learning_rate": 1.4093023255813954e-06,
      "loss": 2.6563,
      "step": 2150
    },
    {
      "epoch": 0.192102454642476,
      "grad_norm": 14.0,
      "learning_rate": 1.2697674418604653e-06,
      "loss": 2.6377,
      "step": 2160
    },
    {
      "epoch": 0.19299181785841338,
      "grad_norm": 14.4375,
      "learning_rate": 1.1302325581395349e-06,
      "loss": 2.694,
      "step": 2170
    },
    {
      "epoch": 0.19388118107435076,
      "grad_norm": 14.5,
      "learning_rate": 9.906976744186047e-07,
      "loss": 2.6686,
      "step": 2180
    },
    {
      "epoch": 0.19477054429028814,
      "grad_norm": 14.0625,
      "learning_rate": 8.511627906976745e-07,
      "loss": 2.7232,
      "step": 2190
    },
    {
      "epoch": 0.19565990750622556,
      "grad_norm": 14.5625,
      "learning_rate": 7.116279069767442e-07,
      "loss": 2.7544,
      "step": 2200
    },
    {
      "epoch": 0.19654927072216294,
      "grad_norm": 13.5,
      "learning_rate": 5.72093023255814e-07,
      "loss": 2.6411,
      "step": 2210
    },
    {
      "epoch": 0.19743863393810032,
      "grad_norm": 13.1875,
      "learning_rate": 4.325581395348837e-07,
      "loss": 2.5296,
      "step": 2220
    },
    {
      "epoch": 0.1983279971540377,
      "grad_norm": 15.25,
      "learning_rate": 2.9302325581395347e-07,
      "loss": 2.5208,
      "step": 2230
    },
    {
      "epoch": 0.1992173603699751,
      "grad_norm": 13.125,
      "learning_rate": 1.5348837209302325e-07,
      "loss": 2.6465,
      "step": 2240
    },
    {
      "epoch": 0.2001067235859125,
      "grad_norm": 14.75,
      "learning_rate": 1.3953488372093025e-08,
      "loss": 2.7407,
      "step": 2250
    }
  ],
  "logging_steps": 10,
  "max_steps": 2250,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "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": 1.3699129752e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}