{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9993522716757657,
  "eval_steps": 500,
  "global_step": 1350,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0007402609419820486,
      "grad_norm": 13.5625,
      "learning_rate": 1.4814814814814817e-07,
      "loss": 2.7636,
      "step": 1
    },
    {
      "epoch": 0.0037013047099102433,
      "grad_norm": 13.5,
      "learning_rate": 7.407407407407407e-07,
      "loss": 2.7044,
      "step": 5
    },
    {
      "epoch": 0.007402609419820487,
      "grad_norm": 12.0625,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 2.6933,
      "step": 10
    },
    {
      "epoch": 0.01110391412973073,
      "grad_norm": 9.75,
      "learning_rate": 2.222222222222222e-06,
      "loss": 2.6397,
      "step": 15
    },
    {
      "epoch": 0.014805218839640973,
      "grad_norm": 9.25,
      "learning_rate": 2.962962962962963e-06,
      "loss": 2.5469,
      "step": 20
    },
    {
      "epoch": 0.018506523549551217,
      "grad_norm": 4.3125,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 2.4799,
      "step": 25
    },
    {
      "epoch": 0.02220782825946146,
      "grad_norm": 3.4375,
      "learning_rate": 4.444444444444444e-06,
      "loss": 2.4496,
      "step": 30
    },
    {
      "epoch": 0.025909132969371702,
      "grad_norm": 3.171875,
      "learning_rate": 5.185185185185185e-06,
      "loss": 2.3571,
      "step": 35
    },
    {
      "epoch": 0.029610437679281947,
      "grad_norm": 2.8125,
      "learning_rate": 5.925925925925926e-06,
      "loss": 2.3253,
      "step": 40
    },
    {
      "epoch": 0.03331174238919219,
      "grad_norm": 2.75,
      "learning_rate": 6.666666666666667e-06,
      "loss": 2.3086,
      "step": 45
    },
    {
      "epoch": 0.037013047099102435,
      "grad_norm": 2.6875,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 2.2763,
      "step": 50
    },
    {
      "epoch": 0.04071435180901268,
      "grad_norm": 2.375,
      "learning_rate": 8.148148148148148e-06,
      "loss": 2.2683,
      "step": 55
    },
    {
      "epoch": 0.04441565651892292,
      "grad_norm": 2.515625,
      "learning_rate": 8.888888888888888e-06,
      "loss": 2.2694,
      "step": 60
    },
    {
      "epoch": 0.04811696122883316,
      "grad_norm": 2.25,
      "learning_rate": 9.62962962962963e-06,
      "loss": 2.2582,
      "step": 65
    },
    {
      "epoch": 0.051818265938743405,
      "grad_norm": 2.203125,
      "learning_rate": 1.037037037037037e-05,
      "loss": 2.2606,
      "step": 70
    },
    {
      "epoch": 0.05551957064865365,
      "grad_norm": 2.1875,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 2.2405,
      "step": 75
    },
    {
      "epoch": 0.05922087535856389,
      "grad_norm": 2.1875,
      "learning_rate": 1.1851851851851852e-05,
      "loss": 2.2496,
      "step": 80
    },
    {
      "epoch": 0.06292218006847414,
      "grad_norm": 2.1875,
      "learning_rate": 1.2592592592592593e-05,
      "loss": 2.2547,
      "step": 85
    },
    {
      "epoch": 0.06662348477838438,
      "grad_norm": 2.109375,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 2.2488,
      "step": 90
    },
    {
      "epoch": 0.07032478948829463,
      "grad_norm": 2.109375,
      "learning_rate": 1.4074074074074075e-05,
      "loss": 2.2479,
      "step": 95
    },
    {
      "epoch": 0.07402609419820487,
      "grad_norm": 2.03125,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 2.2405,
      "step": 100
    },
    {
      "epoch": 0.07772739890811511,
      "grad_norm": 2.109375,
      "learning_rate": 1.555555555555556e-05,
      "loss": 2.222,
      "step": 105
    },
    {
      "epoch": 0.08142870361802536,
      "grad_norm": 2.140625,
      "learning_rate": 1.6296296296296297e-05,
      "loss": 2.2254,
      "step": 110
    },
    {
      "epoch": 0.0851300083279356,
      "grad_norm": 2.0625,
      "learning_rate": 1.7037037037037038e-05,
      "loss": 2.2505,
      "step": 115
    },
    {
      "epoch": 0.08883131303784585,
      "grad_norm": 2.015625,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 2.2367,
      "step": 120
    },
    {
      "epoch": 0.09253261774775609,
      "grad_norm": 2.1875,
      "learning_rate": 1.851851851851852e-05,
      "loss": 2.2275,
      "step": 125
    },
    {
      "epoch": 0.09623392245766632,
      "grad_norm": 2.078125,
      "learning_rate": 1.925925925925926e-05,
      "loss": 2.2258,
      "step": 130
    },
    {
      "epoch": 0.09993522716757657,
      "grad_norm": 2.03125,
      "learning_rate": 2e-05,
      "loss": 2.2103,
      "step": 135
    },
    {
      "epoch": 0.10363653187748681,
      "grad_norm": 2.015625,
      "learning_rate": 1.9999164298554375e-05,
      "loss": 2.2442,
      "step": 140
    },
    {
      "epoch": 0.10733783658739705,
      "grad_norm": 2.109375,
      "learning_rate": 1.9996657333896875e-05,
      "loss": 2.2542,
      "step": 145
    },
    {
      "epoch": 0.1110391412973073,
      "grad_norm": 1.9765625,
      "learning_rate": 1.9992479525042305e-05,
      "loss": 2.2434,
      "step": 150
    },
    {
      "epoch": 0.11474044600721754,
      "grad_norm": 2.015625,
      "learning_rate": 1.9986631570270835e-05,
      "loss": 2.2195,
      "step": 155
    },
    {
      "epoch": 0.11844175071712779,
      "grad_norm": 2.03125,
      "learning_rate": 1.9979114447011323e-05,
      "loss": 2.2221,
      "step": 160
    },
    {
      "epoch": 0.12214305542703803,
      "grad_norm": 2.0,
      "learning_rate": 1.996992941167792e-05,
      "loss": 2.2276,
      "step": 165
    },
    {
      "epoch": 0.12584436013694827,
      "grad_norm": 2.03125,
      "learning_rate": 1.9959077999460094e-05,
      "loss": 2.2112,
      "step": 170
    },
    {
      "epoch": 0.1295456648468585,
      "grad_norm": 2.046875,
      "learning_rate": 1.9946562024066018e-05,
      "loss": 2.2086,
      "step": 175
    },
    {
      "epoch": 0.13324696955676876,
      "grad_norm": 1.9765625,
      "learning_rate": 1.9932383577419432e-05,
      "loss": 2.2095,
      "step": 180
    },
    {
      "epoch": 0.136948274266679,
      "grad_norm": 1.9921875,
      "learning_rate": 1.991654502931001e-05,
      "loss": 2.2194,
      "step": 185
    },
    {
      "epoch": 0.14064957897658925,
      "grad_norm": 1.953125,
      "learning_rate": 1.9899049026997272e-05,
      "loss": 2.229,
      "step": 190
    },
    {
      "epoch": 0.14435088368649948,
      "grad_norm": 1.9921875,
      "learning_rate": 1.9879898494768093e-05,
      "loss": 2.2252,
      "step": 195
    },
    {
      "epoch": 0.14805218839640974,
      "grad_norm": 1.9140625,
      "learning_rate": 1.9859096633447965e-05,
      "loss": 2.2168,
      "step": 200
    },
    {
      "epoch": 0.15175349310631997,
      "grad_norm": 2.046875,
      "learning_rate": 1.9836646919866012e-05,
      "loss": 2.2246,
      "step": 205
    },
    {
      "epoch": 0.15545479781623023,
      "grad_norm": 1.9375,
      "learning_rate": 1.9812553106273848e-05,
      "loss": 2.2151,
      "step": 210
    },
    {
      "epoch": 0.15915610252614046,
      "grad_norm": 1.90625,
      "learning_rate": 1.9786819219718443e-05,
      "loss": 2.2176,
      "step": 215
    },
    {
      "epoch": 0.16285740723605072,
      "grad_norm": 1.9453125,
      "learning_rate": 1.9759449561369036e-05,
      "loss": 2.2279,
      "step": 220
    },
    {
      "epoch": 0.16655871194596095,
      "grad_norm": 2.015625,
      "learning_rate": 1.973044870579824e-05,
      "loss": 2.2113,
      "step": 225
    },
    {
      "epoch": 0.1702600166558712,
      "grad_norm": 1.875,
      "learning_rate": 1.9699821500217436e-05,
      "loss": 2.2418,
      "step": 230
    },
    {
      "epoch": 0.17396132136578144,
      "grad_norm": 1.8984375,
      "learning_rate": 1.9667573063666622e-05,
      "loss": 2.2206,
      "step": 235
    },
    {
      "epoch": 0.1776626260756917,
      "grad_norm": 1.9375,
      "learning_rate": 1.9633708786158803e-05,
      "loss": 2.2162,
      "step": 240
    },
    {
      "epoch": 0.18136393078560192,
      "grad_norm": 1.8828125,
      "learning_rate": 1.959823432777912e-05,
      "loss": 2.2015,
      "step": 245
    },
    {
      "epoch": 0.18506523549551218,
      "grad_norm": 1.90625,
      "learning_rate": 1.95611556177388e-05,
      "loss": 2.2095,
      "step": 250
    },
    {
      "epoch": 0.1887665402054224,
      "grad_norm": 1.828125,
      "learning_rate": 1.9522478853384154e-05,
      "loss": 2.2245,
      "step": 255
    },
    {
      "epoch": 0.19246784491533264,
      "grad_norm": 1.8984375,
      "learning_rate": 1.9482210499160767e-05,
      "loss": 2.2179,
      "step": 260
    },
    {
      "epoch": 0.1961691496252429,
      "grad_norm": 1.90625,
      "learning_rate": 1.9440357285533e-05,
      "loss": 2.2132,
      "step": 265
    },
    {
      "epoch": 0.19987045433515313,
      "grad_norm": 2.046875,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 2.2024,
      "step": 270
    },
    {
      "epoch": 0.2035717590450634,
      "grad_norm": 1.859375,
      "learning_rate": 1.93519245252219e-05,
      "loss": 2.1749,
      "step": 275
    },
    {
      "epoch": 0.20727306375497362,
      "grad_norm": 2.171875,
      "learning_rate": 1.9305359759215686e-05,
      "loss": 2.2137,
      "step": 280
    },
    {
      "epoch": 0.21097436846488388,
      "grad_norm": 1.921875,
      "learning_rate": 1.9257239692688907e-05,
      "loss": 2.2294,
      "step": 285
    },
    {
      "epoch": 0.2146756731747941,
      "grad_norm": 1.90625,
      "learning_rate": 1.9207572368443386e-05,
      "loss": 2.2023,
      "step": 290
    },
    {
      "epoch": 0.21837697788470437,
      "grad_norm": 1.890625,
      "learning_rate": 1.9156366087890062e-05,
      "loss": 2.2113,
      "step": 295
    },
    {
      "epoch": 0.2220782825946146,
      "grad_norm": 1.9375,
      "learning_rate": 1.9103629409661468e-05,
      "loss": 2.2245,
      "step": 300
    },
    {
      "epoch": 0.22577958730452485,
      "grad_norm": 1.90625,
      "learning_rate": 1.9049371148181253e-05,
      "loss": 2.2102,
      "step": 305
    },
    {
      "epoch": 0.22948089201443508,
      "grad_norm": 1.9609375,
      "learning_rate": 1.8993600372190933e-05,
      "loss": 2.2243,
      "step": 310
    },
    {
      "epoch": 0.23318219672434534,
      "grad_norm": 1.90625,
      "learning_rate": 1.8936326403234125e-05,
      "loss": 2.2009,
      "step": 315
    },
    {
      "epoch": 0.23688350143425557,
      "grad_norm": 1.828125,
      "learning_rate": 1.8877558814098564e-05,
      "loss": 2.2078,
      "step": 320
    },
    {
      "epoch": 0.24058480614416583,
      "grad_norm": 1.8515625,
      "learning_rate": 1.881730742721608e-05,
      "loss": 2.2031,
      "step": 325
    },
    {
      "epoch": 0.24428611085407606,
      "grad_norm": 1.8515625,
      "learning_rate": 1.8755582313020912e-05,
      "loss": 2.1597,
      "step": 330
    },
    {
      "epoch": 0.24798741556398632,
      "grad_norm": 1.875,
      "learning_rate": 1.8692393788266477e-05,
      "loss": 2.1922,
      "step": 335
    },
    {
      "epoch": 0.25168872027389655,
      "grad_norm": 1.8984375,
      "learning_rate": 1.8627752414301087e-05,
      "loss": 2.1883,
      "step": 340
    },
    {
      "epoch": 0.2553900249838068,
      "grad_norm": 1.90625,
      "learning_rate": 1.8561668995302668e-05,
      "loss": 2.2097,
      "step": 345
    },
    {
      "epoch": 0.259091329693717,
      "grad_norm": 1.875,
      "learning_rate": 1.8494154576472976e-05,
      "loss": 2.2106,
      "step": 350
    },
    {
      "epoch": 0.26279263440362727,
      "grad_norm": 1.875,
      "learning_rate": 1.8425220442191496e-05,
      "loss": 2.2035,
      "step": 355
    },
    {
      "epoch": 0.2664939391135375,
      "grad_norm": 1.859375,
      "learning_rate": 1.8354878114129368e-05,
      "loss": 2.1937,
      "step": 360
    },
    {
      "epoch": 0.2701952438234478,
      "grad_norm": 1.8046875,
      "learning_rate": 1.8283139349323632e-05,
      "loss": 2.1828,
      "step": 365
    },
    {
      "epoch": 0.273896548533358,
      "grad_norm": 1.84375,
      "learning_rate": 1.8210016138212186e-05,
      "loss": 2.189,
      "step": 370
    },
    {
      "epoch": 0.27759785324326824,
      "grad_norm": 1.8203125,
      "learning_rate": 1.8135520702629677e-05,
      "loss": 2.2025,
      "step": 375
    },
    {
      "epoch": 0.2812991579531785,
      "grad_norm": 1.859375,
      "learning_rate": 1.8059665493764745e-05,
      "loss": 2.1967,
      "step": 380
    },
    {
      "epoch": 0.28500046266308876,
      "grad_norm": 1.796875,
      "learning_rate": 1.7982463190078928e-05,
      "loss": 2.1726,
      "step": 385
    },
    {
      "epoch": 0.28870176737299896,
      "grad_norm": 1.8203125,
      "learning_rate": 1.7903926695187595e-05,
      "loss": 2.1758,
      "step": 390
    },
    {
      "epoch": 0.2924030720829092,
      "grad_norm": 1.859375,
      "learning_rate": 1.78240691357032e-05,
      "loss": 2.186,
      "step": 395
    },
    {
      "epoch": 0.2961043767928195,
      "grad_norm": 1.828125,
      "learning_rate": 1.7742903859041324e-05,
      "loss": 2.1866,
      "step": 400
    },
    {
      "epoch": 0.29980568150272974,
      "grad_norm": 1.828125,
      "learning_rate": 1.766044443118978e-05,
      "loss": 2.1996,
      "step": 405
    },
    {
      "epoch": 0.30350698621263994,
      "grad_norm": 1.7890625,
      "learning_rate": 1.757670463444118e-05,
      "loss": 2.1657,
      "step": 410
    },
    {
      "epoch": 0.3072082909225502,
      "grad_norm": 1.8046875,
      "learning_rate": 1.749169846508936e-05,
      "loss": 2.1938,
      "step": 415
    },
    {
      "epoch": 0.31090959563246046,
      "grad_norm": 1.8359375,
      "learning_rate": 1.740544013109005e-05,
      "loss": 2.1802,
      "step": 420
    },
    {
      "epoch": 0.31461090034237066,
      "grad_norm": 1.8359375,
      "learning_rate": 1.7317944049686125e-05,
      "loss": 2.1961,
      "step": 425
    },
    {
      "epoch": 0.3183122050522809,
      "grad_norm": 1.8359375,
      "learning_rate": 1.722922484499793e-05,
      "loss": 2.1849,
      "step": 430
    },
    {
      "epoch": 0.3220135097621912,
      "grad_norm": 1.84375,
      "learning_rate": 1.7139297345578992e-05,
      "loss": 2.2075,
      "step": 435
    },
    {
      "epoch": 0.32571481447210143,
      "grad_norm": 1.8046875,
      "learning_rate": 1.7048176581937562e-05,
      "loss": 2.1531,
      "step": 440
    },
    {
      "epoch": 0.32941611918201164,
      "grad_norm": 1.8359375,
      "learning_rate": 1.6955877784024418e-05,
      "loss": 2.1763,
      "step": 445
    },
    {
      "epoch": 0.3331174238919219,
      "grad_norm": 1.8359375,
      "learning_rate": 1.686241637868734e-05,
      "loss": 2.2074,
      "step": 450
    },
    {
      "epoch": 0.33681872860183215,
      "grad_norm": 1.765625,
      "learning_rate": 1.676780798709262e-05,
      "loss": 2.1793,
      "step": 455
    },
    {
      "epoch": 0.3405200333117424,
      "grad_norm": 1.8203125,
      "learning_rate": 1.6672068422114195e-05,
      "loss": 2.1853,
      "step": 460
    },
    {
      "epoch": 0.3442213380216526,
      "grad_norm": 1.796875,
      "learning_rate": 1.657521368569064e-05,
      "loss": 2.1804,
      "step": 465
    },
    {
      "epoch": 0.34792264273156287,
      "grad_norm": 1.8515625,
      "learning_rate": 1.647725996615059e-05,
      "loss": 2.1836,
      "step": 470
    },
    {
      "epoch": 0.35162394744147313,
      "grad_norm": 1.9921875,
      "learning_rate": 1.637822363550706e-05,
      "loss": 2.1714,
      "step": 475
    },
    {
      "epoch": 0.3553252521513834,
      "grad_norm": 1.953125,
      "learning_rate": 1.627812124672099e-05,
      "loss": 2.2257,
      "step": 480
    },
    {
      "epoch": 0.3590265568612936,
      "grad_norm": 1.8515625,
      "learning_rate": 1.6176969530934573e-05,
      "loss": 2.1983,
      "step": 485
    },
    {
      "epoch": 0.36272786157120385,
      "grad_norm": 1.8125,
      "learning_rate": 1.6074785394674835e-05,
      "loss": 2.1925,
      "step": 490
    },
    {
      "epoch": 0.3664291662811141,
      "grad_norm": 1.8125,
      "learning_rate": 1.5971585917027864e-05,
      "loss": 2.1606,
      "step": 495
    },
    {
      "epoch": 0.37013047099102436,
      "grad_norm": 1.8125,
      "learning_rate": 1.586738834678418e-05,
      "loss": 2.1738,
      "step": 500
    },
    {
      "epoch": 0.37383177570093457,
      "grad_norm": 1.8203125,
      "learning_rate": 1.5762210099555804e-05,
      "loss": 2.17,
      "step": 505
    },
    {
      "epoch": 0.3775330804108448,
      "grad_norm": 1.859375,
      "learning_rate": 1.5656068754865388e-05,
      "loss": 2.1759,
      "step": 510
    },
    {
      "epoch": 0.3812343851207551,
      "grad_norm": 1.8359375,
      "learning_rate": 1.554898205320797e-05,
      "loss": 2.2016,
      "step": 515
    },
    {
      "epoch": 0.3849356898306653,
      "grad_norm": 1.796875,
      "learning_rate": 1.5440967893085827e-05,
      "loss": 2.1711,
      "step": 520
    },
    {
      "epoch": 0.38863699454057554,
      "grad_norm": 1.84375,
      "learning_rate": 1.5332044328016916e-05,
      "loss": 2.1809,
      "step": 525
    },
    {
      "epoch": 0.3923382992504858,
      "grad_norm": 1.796875,
      "learning_rate": 1.5222229563517385e-05,
      "loss": 2.2018,
      "step": 530
    },
    {
      "epoch": 0.39603960396039606,
      "grad_norm": 1.765625,
      "learning_rate": 1.5111541954058733e-05,
      "loss": 2.1762,
      "step": 535
    },
    {
      "epoch": 0.39974090867030626,
      "grad_norm": 1.7578125,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 2.1955,
      "step": 540
    },
    {
      "epoch": 0.4034422133802165,
      "grad_norm": 1.7890625,
      "learning_rate": 1.4887622344495643e-05,
      "loss": 2.1855,
      "step": 545
    },
    {
      "epoch": 0.4071435180901268,
      "grad_norm": 1.8203125,
      "learning_rate": 1.4774427770379492e-05,
      "loss": 2.174,
      "step": 550
    },
    {
      "epoch": 0.41084482280003704,
      "grad_norm": 1.8125,
      "learning_rate": 1.4660435197025391e-05,
      "loss": 2.1727,
      "step": 555
    },
    {
      "epoch": 0.41454612750994724,
      "grad_norm": 1.828125,
      "learning_rate": 1.4545663677185007e-05,
      "loss": 2.1715,
      "step": 560
    },
    {
      "epoch": 0.4182474322198575,
      "grad_norm": 1.8203125,
      "learning_rate": 1.4430132393803353e-05,
      "loss": 2.1893,
      "step": 565
    },
    {
      "epoch": 0.42194873692976775,
      "grad_norm": 1.8046875,
      "learning_rate": 1.4313860656812537e-05,
      "loss": 2.1734,
      "step": 570
    },
    {
      "epoch": 0.425650041639678,
      "grad_norm": 1.8046875,
      "learning_rate": 1.4196867899904292e-05,
      "loss": 2.1759,
      "step": 575
    },
    {
      "epoch": 0.4293513463495882,
      "grad_norm": 1.78125,
      "learning_rate": 1.4079173677281836e-05,
      "loss": 2.1615,
      "step": 580
    },
    {
      "epoch": 0.4330526510594985,
      "grad_norm": 1.796875,
      "learning_rate": 1.396079766039157e-05,
      "loss": 2.1769,
      "step": 585
    },
    {
      "epoch": 0.43675395576940873,
      "grad_norm": 1.8125,
      "learning_rate": 1.3841759634635177e-05,
      "loss": 2.1867,
      "step": 590
    },
    {
      "epoch": 0.44045526047931893,
      "grad_norm": 1.8046875,
      "learning_rate": 1.3722079496062702e-05,
      "loss": 2.1836,
      "step": 595
    },
    {
      "epoch": 0.4441565651892292,
      "grad_norm": 1.765625,
      "learning_rate": 1.3601777248047105e-05,
      "loss": 2.1803,
      "step": 600
    },
    {
      "epoch": 0.44785786989913945,
      "grad_norm": 1.828125,
      "learning_rate": 1.3480872997940906e-05,
      "loss": 2.1667,
      "step": 605
    },
    {
      "epoch": 0.4515591746090497,
      "grad_norm": 1.8203125,
      "learning_rate": 1.3359386953715423e-05,
      "loss": 2.1644,
      "step": 610
    },
    {
      "epoch": 0.4552604793189599,
      "grad_norm": 1.8125,
      "learning_rate": 1.3237339420583213e-05,
      "loss": 2.1769,
      "step": 615
    },
    {
      "epoch": 0.45896178402887017,
      "grad_norm": 1.78125,
      "learning_rate": 1.3114750797604248e-05,
      "loss": 2.1611,
      "step": 620
    },
    {
      "epoch": 0.4626630887387804,
      "grad_norm": 1.796875,
      "learning_rate": 1.2991641574276419e-05,
      "loss": 2.1676,
      "step": 625
    },
    {
      "epoch": 0.4663643934486907,
      "grad_norm": 1.8671875,
      "learning_rate": 1.2868032327110904e-05,
      "loss": 2.2038,
      "step": 630
    },
    {
      "epoch": 0.4700656981586009,
      "grad_norm": 1.8203125,
      "learning_rate": 1.2743943716193017e-05,
      "loss": 2.2025,
      "step": 635
    },
    {
      "epoch": 0.47376700286851114,
      "grad_norm": 1.7890625,
      "learning_rate": 1.261939648172906e-05,
      "loss": 2.1784,
      "step": 640
    },
    {
      "epoch": 0.4774683075784214,
      "grad_norm": 1.796875,
      "learning_rate": 1.2494411440579814e-05,
      "loss": 2.1805,
      "step": 645
    },
    {
      "epoch": 0.48116961228833166,
      "grad_norm": 1.796875,
      "learning_rate": 1.2369009482781191e-05,
      "loss": 2.1945,
      "step": 650
    },
    {
      "epoch": 0.48487091699824186,
      "grad_norm": 1.8359375,
      "learning_rate": 1.2243211568052678e-05,
      "loss": 2.1775,
      "step": 655
    },
    {
      "epoch": 0.4885722217081521,
      "grad_norm": 1.796875,
      "learning_rate": 1.211703872229411e-05,
      "loss": 2.1832,
      "step": 660
    },
    {
      "epoch": 0.4922735264180624,
      "grad_norm": 1.84375,
      "learning_rate": 1.1990512034071407e-05,
      "loss": 2.1899,
      "step": 665
    },
    {
      "epoch": 0.49597483112797264,
      "grad_norm": 1.8203125,
      "learning_rate": 1.1863652651091824e-05,
      "loss": 2.1675,
      "step": 670
    },
    {
      "epoch": 0.49967613583788284,
      "grad_norm": 1.7734375,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 2.1881,
      "step": 675
    },
    {
      "epoch": 0.5033774405477931,
      "grad_norm": 1.7890625,
      "learning_rate": 1.1609020666180574e-05,
      "loss": 2.1825,
      "step": 680
    },
    {
      "epoch": 0.5070787452577034,
      "grad_norm": 1.78125,
      "learning_rate": 1.1481290623512491e-05,
      "loss": 2.1875,
      "step": 685
    },
    {
      "epoch": 0.5107800499676136,
      "grad_norm": 1.828125,
      "learning_rate": 1.1353312997501313e-05,
      "loss": 2.1693,
      "step": 690
    },
    {
      "epoch": 0.5144813546775239,
      "grad_norm": 1.78125,
      "learning_rate": 1.1225109178364456e-05,
      "loss": 2.169,
      "step": 695
    },
    {
      "epoch": 0.518182659387434,
      "grad_norm": 1.765625,
      "learning_rate": 1.1096700594125318e-05,
      "loss": 2.1714,
      "step": 700
    },
    {
      "epoch": 0.5218839640973443,
      "grad_norm": 1.8203125,
      "learning_rate": 1.0968108707031792e-05,
      "loss": 2.179,
      "step": 705
    },
    {
      "epoch": 0.5255852688072545,
      "grad_norm": 1.765625,
      "learning_rate": 1.0839355009969068e-05,
      "loss": 2.153,
      "step": 710
    },
    {
      "epoch": 0.5292865735171648,
      "grad_norm": 1.796875,
      "learning_rate": 1.0710461022867303e-05,
      "loss": 2.1683,
      "step": 715
    },
    {
      "epoch": 0.532987878227075,
      "grad_norm": 1.7578125,
      "learning_rate": 1.0581448289104759e-05,
      "loss": 2.1459,
      "step": 720
    },
    {
      "epoch": 0.5366891829369853,
      "grad_norm": 1.8515625,
      "learning_rate": 1.0452338371907065e-05,
      "loss": 2.1666,
      "step": 725
    },
    {
      "epoch": 0.5403904876468956,
      "grad_norm": 1.765625,
      "learning_rate": 1.0323152850743107e-05,
      "loss": 2.173,
      "step": 730
    },
    {
      "epoch": 0.5440917923568058,
      "grad_norm": 1.8046875,
      "learning_rate": 1.0193913317718245e-05,
      "loss": 2.1745,
      "step": 735
    },
    {
      "epoch": 0.547793097066716,
      "grad_norm": 1.7734375,
      "learning_rate": 1.0064641373965394e-05,
      "loss": 2.1602,
      "step": 740
    },
    {
      "epoch": 0.5514944017766262,
      "grad_norm": 1.7578125,
      "learning_rate": 9.935358626034607e-06,
      "loss": 2.1608,
      "step": 745
    },
    {
      "epoch": 0.5551957064865365,
      "grad_norm": 1.796875,
      "learning_rate": 9.806086682281759e-06,
      "loss": 2.1906,
      "step": 750
    },
    {
      "epoch": 0.5588970111964467,
      "grad_norm": 1.7734375,
      "learning_rate": 9.676847149256894e-06,
      "loss": 2.1502,
      "step": 755
    },
    {
      "epoch": 0.562598315906357,
      "grad_norm": 1.765625,
      "learning_rate": 9.547661628092938e-06,
      "loss": 2.1648,
      "step": 760
    },
    {
      "epoch": 0.5662996206162673,
      "grad_norm": 1.8515625,
      "learning_rate": 9.418551710895243e-06,
      "loss": 2.1866,
      "step": 765
    },
    {
      "epoch": 0.5700009253261775,
      "grad_norm": 1.7734375,
      "learning_rate": 9.289538977132702e-06,
      "loss": 2.1776,
      "step": 770
    },
    {
      "epoch": 0.5737022300360877,
      "grad_norm": 1.7734375,
      "learning_rate": 9.160644990030932e-06,
      "loss": 2.1665,
      "step": 775
    },
    {
      "epoch": 0.5774035347459979,
      "grad_norm": 1.7734375,
      "learning_rate": 9.03189129296821e-06,
      "loss": 2.1891,
      "step": 780
    },
    {
      "epoch": 0.5811048394559082,
      "grad_norm": 1.8203125,
      "learning_rate": 8.903299405874685e-06,
      "loss": 2.1894,
      "step": 785
    },
    {
      "epoch": 0.5848061441658184,
      "grad_norm": 1.7734375,
      "learning_rate": 8.774890821635548e-06,
      "loss": 2.167,
      "step": 790
    },
    {
      "epoch": 0.5885074488757287,
      "grad_norm": 1.8359375,
      "learning_rate": 8.646687002498692e-06,
      "loss": 2.1759,
      "step": 795
    },
    {
      "epoch": 0.592208753585639,
      "grad_norm": 1.8046875,
      "learning_rate": 8.518709376487515e-06,
      "loss": 2.155,
      "step": 800
    },
    {
      "epoch": 0.5959100582955492,
      "grad_norm": 1.796875,
      "learning_rate": 8.390979333819427e-06,
      "loss": 2.1734,
      "step": 805
    },
    {
      "epoch": 0.5996113630054595,
      "grad_norm": 1.7890625,
      "learning_rate": 8.263518223330698e-06,
      "loss": 2.1619,
      "step": 810
    },
    {
      "epoch": 0.6033126677153696,
      "grad_norm": 1.8125,
      "learning_rate": 8.13634734890818e-06,
      "loss": 2.1509,
      "step": 815
    },
    {
      "epoch": 0.6070139724252799,
      "grad_norm": 1.796875,
      "learning_rate": 8.009487965928597e-06,
      "loss": 2.1662,
      "step": 820
    },
    {
      "epoch": 0.6107152771351901,
      "grad_norm": 1.796875,
      "learning_rate": 7.882961277705897e-06,
      "loss": 2.1315,
      "step": 825
    },
    {
      "epoch": 0.6144165818451004,
      "grad_norm": 1.8046875,
      "learning_rate": 7.756788431947327e-06,
      "loss": 2.1737,
      "step": 830
    },
    {
      "epoch": 0.6181178865550107,
      "grad_norm": 1.8359375,
      "learning_rate": 7.630990517218809e-06,
      "loss": 2.1673,
      "step": 835
    },
    {
      "epoch": 0.6218191912649209,
      "grad_norm": 1.859375,
      "learning_rate": 7.505588559420188e-06,
      "loss": 2.156,
      "step": 840
    },
    {
      "epoch": 0.6255204959748312,
      "grad_norm": 1.796875,
      "learning_rate": 7.380603518270942e-06,
      "loss": 2.1576,
      "step": 845
    },
    {
      "epoch": 0.6292218006847413,
      "grad_norm": 1.7890625,
      "learning_rate": 7.256056283806987e-06,
      "loss": 2.1715,
      "step": 850
    },
    {
      "epoch": 0.6329231053946516,
      "grad_norm": 1.78125,
      "learning_rate": 7.131967672889101e-06,
      "loss": 2.1872,
      "step": 855
    },
    {
      "epoch": 0.6366244101045618,
      "grad_norm": 1.828125,
      "learning_rate": 7.008358425723586e-06,
      "loss": 2.1596,
      "step": 860
    },
    {
      "epoch": 0.6403257148144721,
      "grad_norm": 1.765625,
      "learning_rate": 6.885249202395754e-06,
      "loss": 2.1546,
      "step": 865
    },
    {
      "epoch": 0.6440270195243823,
      "grad_norm": 1.796875,
      "learning_rate": 6.762660579416791e-06,
      "loss": 2.1499,
      "step": 870
    },
    {
      "epoch": 0.6477283242342926,
      "grad_norm": 1.8203125,
      "learning_rate": 6.640613046284581e-06,
      "loss": 2.1816,
      "step": 875
    },
    {
      "epoch": 0.6514296289442029,
      "grad_norm": 1.8359375,
      "learning_rate": 6.519127002059096e-06,
      "loss": 2.1675,
      "step": 880
    },
    {
      "epoch": 0.6551309336541131,
      "grad_norm": 1.796875,
      "learning_rate": 6.3982227519528986e-06,
      "loss": 2.1819,
      "step": 885
    },
    {
      "epoch": 0.6588322383640233,
      "grad_norm": 1.8359375,
      "learning_rate": 6.277920503937303e-06,
      "loss": 2.1628,
      "step": 890
    },
    {
      "epoch": 0.6625335430739335,
      "grad_norm": 1.8046875,
      "learning_rate": 6.158240365364823e-06,
      "loss": 2.1595,
      "step": 895
    },
    {
      "epoch": 0.6662348477838438,
      "grad_norm": 1.7890625,
      "learning_rate": 6.039202339608432e-06,
      "loss": 2.1604,
      "step": 900
    },
    {
      "epoch": 0.669936152493754,
      "grad_norm": 1.7890625,
      "learning_rate": 5.920826322718165e-06,
      "loss": 2.1498,
      "step": 905
    },
    {
      "epoch": 0.6736374572036643,
      "grad_norm": 1.8125,
      "learning_rate": 5.80313210009571e-06,
      "loss": 2.1612,
      "step": 910
    },
    {
      "epoch": 0.6773387619135746,
      "grad_norm": 1.7890625,
      "learning_rate": 5.686139343187468e-06,
      "loss": 2.1668,
      "step": 915
    },
    {
      "epoch": 0.6810400666234848,
      "grad_norm": 1.75,
      "learning_rate": 5.569867606196652e-06,
      "loss": 2.1952,
      "step": 920
    },
    {
      "epoch": 0.6847413713333951,
      "grad_norm": 1.7578125,
      "learning_rate": 5.454336322814995e-06,
      "loss": 2.1699,
      "step": 925
    },
    {
      "epoch": 0.6884426760433052,
      "grad_norm": 1.7578125,
      "learning_rate": 5.339564802974615e-06,
      "loss": 2.176,
      "step": 930
    },
    {
      "epoch": 0.6921439807532155,
      "grad_norm": 1.8515625,
      "learning_rate": 5.2255722296205104e-06,
      "loss": 2.1927,
      "step": 935
    },
    {
      "epoch": 0.6958452854631257,
      "grad_norm": 1.796875,
      "learning_rate": 5.112377655504359e-06,
      "loss": 2.1742,
      "step": 940
    },
    {
      "epoch": 0.699546590173036,
      "grad_norm": 1.78125,
      "learning_rate": 5.000000000000003e-06,
      "loss": 2.1835,
      "step": 945
    },
    {
      "epoch": 0.7032478948829463,
      "grad_norm": 1.84375,
      "learning_rate": 4.888458045941269e-06,
      "loss": 2.1855,
      "step": 950
    },
    {
      "epoch": 0.7069491995928565,
      "grad_norm": 1.8125,
      "learning_rate": 4.7777704364826175e-06,
      "loss": 2.1922,
      "step": 955
    },
    {
      "epoch": 0.7106505043027668,
      "grad_norm": 1.8046875,
      "learning_rate": 4.66795567198309e-06,
      "loss": 2.1563,
      "step": 960
    },
    {
      "epoch": 0.7143518090126769,
      "grad_norm": 1.78125,
      "learning_rate": 4.559032106914173e-06,
      "loss": 2.1522,
      "step": 965
    },
    {
      "epoch": 0.7180531137225872,
      "grad_norm": 1.8203125,
      "learning_rate": 4.4510179467920325e-06,
      "loss": 2.175,
      "step": 970
    },
    {
      "epoch": 0.7217544184324974,
      "grad_norm": 1.765625,
      "learning_rate": 4.343931245134616e-06,
      "loss": 2.1506,
      "step": 975
    },
    {
      "epoch": 0.7254557231424077,
      "grad_norm": 1.8046875,
      "learning_rate": 4.237789900444197e-06,
      "loss": 2.1555,
      "step": 980
    },
    {
      "epoch": 0.729157027852318,
      "grad_norm": 1.78125,
      "learning_rate": 4.132611653215822e-06,
      "loss": 2.1142,
      "step": 985
    },
    {
      "epoch": 0.7328583325622282,
      "grad_norm": 1.7890625,
      "learning_rate": 4.028414082972141e-06,
      "loss": 2.1617,
      "step": 990
    },
    {
      "epoch": 0.7365596372721385,
      "grad_norm": 1.8046875,
      "learning_rate": 3.925214605325164e-06,
      "loss": 2.1854,
      "step": 995
    },
    {
      "epoch": 0.7402609419820487,
      "grad_norm": 1.7734375,
      "learning_rate": 3.823030469065431e-06,
      "loss": 2.1767,
      "step": 1000
    },
    {
      "epoch": 0.7439622466919589,
      "grad_norm": 1.8125,
      "learning_rate": 3.7218787532790167e-06,
      "loss": 2.1667,
      "step": 1005
    },
    {
      "epoch": 0.7476635514018691,
      "grad_norm": 1.8046875,
      "learning_rate": 3.6217763644929393e-06,
      "loss": 2.1532,
      "step": 1010
    },
    {
      "epoch": 0.7513648561117794,
      "grad_norm": 1.8203125,
      "learning_rate": 3.522740033849411e-06,
      "loss": 2.1617,
      "step": 1015
    },
    {
      "epoch": 0.7550661608216896,
      "grad_norm": 1.8828125,
      "learning_rate": 3.424786314309365e-06,
      "loss": 2.1549,
      "step": 1020
    },
    {
      "epoch": 0.7587674655315999,
      "grad_norm": 1.7265625,
      "learning_rate": 3.3279315778858034e-06,
      "loss": 2.1501,
      "step": 1025
    },
    {
      "epoch": 0.7624687702415102,
      "grad_norm": 1.7734375,
      "learning_rate": 3.2321920129073815e-06,
      "loss": 2.164,
      "step": 1030
    },
    {
      "epoch": 0.7661700749514204,
      "grad_norm": 1.765625,
      "learning_rate": 3.1375836213126653e-06,
      "loss": 2.1473,
      "step": 1035
    },
    {
      "epoch": 0.7698713796613306,
      "grad_norm": 1.7890625,
      "learning_rate": 3.04412221597558e-06,
      "loss": 2.1597,
      "step": 1040
    },
    {
      "epoch": 0.7735726843712408,
      "grad_norm": 1.8359375,
      "learning_rate": 2.9518234180624393e-06,
      "loss": 2.165,
      "step": 1045
    },
    {
      "epoch": 0.7772739890811511,
      "grad_norm": 1.7890625,
      "learning_rate": 2.8607026544210115e-06,
      "loss": 2.1634,
      "step": 1050
    },
    {
      "epoch": 0.7809752937910613,
      "grad_norm": 1.7421875,
      "learning_rate": 2.770775155002071e-06,
      "loss": 2.1559,
      "step": 1055
    },
    {
      "epoch": 0.7846765985009716,
      "grad_norm": 1.7734375,
      "learning_rate": 2.6820559503138797e-06,
      "loss": 2.1638,
      "step": 1060
    },
    {
      "epoch": 0.7883779032108819,
      "grad_norm": 1.796875,
      "learning_rate": 2.594559868909956e-06,
      "loss": 2.1604,
      "step": 1065
    },
    {
      "epoch": 0.7920792079207921,
      "grad_norm": 1.7578125,
      "learning_rate": 2.50830153491064e-06,
      "loss": 2.1602,
      "step": 1070
    },
    {
      "epoch": 0.7957805126307024,
      "grad_norm": 1.765625,
      "learning_rate": 2.423295365558821e-06,
      "loss": 2.1588,
      "step": 1075
    },
    {
      "epoch": 0.7994818173406125,
      "grad_norm": 1.8125,
      "learning_rate": 2.339555568810221e-06,
      "loss": 2.1525,
      "step": 1080
    },
    {
      "epoch": 0.8031831220505228,
      "grad_norm": 1.7578125,
      "learning_rate": 2.2570961409586756e-06,
      "loss": 2.1622,
      "step": 1085
    },
    {
      "epoch": 0.806884426760433,
      "grad_norm": 1.7890625,
      "learning_rate": 2.1759308642968024e-06,
      "loss": 2.1544,
      "step": 1090
    },
    {
      "epoch": 0.8105857314703433,
      "grad_norm": 1.8046875,
      "learning_rate": 2.0960733048124082e-06,
      "loss": 2.158,
      "step": 1095
    },
    {
      "epoch": 0.8142870361802536,
      "grad_norm": 1.828125,
      "learning_rate": 2.01753680992107e-06,
      "loss": 2.1827,
      "step": 1100
    },
    {
      "epoch": 0.8179883408901638,
      "grad_norm": 1.7890625,
      "learning_rate": 1.9403345062352574e-06,
      "loss": 2.1579,
      "step": 1105
    },
    {
      "epoch": 0.8216896456000741,
      "grad_norm": 1.7890625,
      "learning_rate": 1.8644792973703252e-06,
      "loss": 2.1601,
      "step": 1110
    },
    {
      "epoch": 0.8253909503099842,
      "grad_norm": 1.8515625,
      "learning_rate": 1.7899838617878163e-06,
      "loss": 2.168,
      "step": 1115
    },
    {
      "epoch": 0.8290922550198945,
      "grad_norm": 1.8359375,
      "learning_rate": 1.7168606506763696e-06,
      "loss": 2.1706,
      "step": 1120
    },
    {
      "epoch": 0.8327935597298047,
      "grad_norm": 1.796875,
      "learning_rate": 1.6451218858706374e-06,
      "loss": 2.175,
      "step": 1125
    },
    {
      "epoch": 0.836494864439715,
      "grad_norm": 1.765625,
      "learning_rate": 1.5747795578085046e-06,
      "loss": 2.1803,
      "step": 1130
    },
    {
      "epoch": 0.8401961691496252,
      "grad_norm": 1.7578125,
      "learning_rate": 1.505845423527027e-06,
      "loss": 2.1543,
      "step": 1135
    },
    {
      "epoch": 0.8438974738595355,
      "grad_norm": 1.7890625,
      "learning_rate": 1.4383310046973365e-06,
      "loss": 2.1556,
      "step": 1140
    },
    {
      "epoch": 0.8475987785694458,
      "grad_norm": 1.828125,
      "learning_rate": 1.372247585698916e-06,
      "loss": 2.1772,
      "step": 1145
    },
    {
      "epoch": 0.851300083279356,
      "grad_norm": 1.7734375,
      "learning_rate": 1.307606211733522e-06,
      "loss": 2.1542,
      "step": 1150
    },
    {
      "epoch": 0.8550013879892662,
      "grad_norm": 1.8125,
      "learning_rate": 1.2444176869790925e-06,
      "loss": 2.1635,
      "step": 1155
    },
    {
      "epoch": 0.8587026926991764,
      "grad_norm": 1.8046875,
      "learning_rate": 1.18269257278392e-06,
      "loss": 2.144,
      "step": 1160
    },
    {
      "epoch": 0.8624039974090867,
      "grad_norm": 1.765625,
      "learning_rate": 1.1224411859014417e-06,
      "loss": 2.153,
      "step": 1165
    },
    {
      "epoch": 0.866105302118997,
      "grad_norm": 1.84375,
      "learning_rate": 1.0636735967658785e-06,
      "loss": 2.1537,
      "step": 1170
    },
    {
      "epoch": 0.8698066068289072,
      "grad_norm": 1.7890625,
      "learning_rate": 1.0063996278090704e-06,
      "loss": 2.1913,
      "step": 1175
    },
    {
      "epoch": 0.8735079115388175,
      "grad_norm": 1.8203125,
      "learning_rate": 9.506288518187468e-07,
      "loss": 2.1783,
      "step": 1180
    },
    {
      "epoch": 0.8772092162487277,
      "grad_norm": 1.75,
      "learning_rate": 8.963705903385344e-07,
      "loss": 2.1502,
      "step": 1185
    },
    {
      "epoch": 0.8809105209586379,
      "grad_norm": 1.7890625,
      "learning_rate": 8.436339121099413e-07,
      "loss": 2.1642,
      "step": 1190
    },
    {
      "epoch": 0.8846118256685481,
      "grad_norm": 1.7734375,
      "learning_rate": 7.924276315566171e-07,
      "loss": 2.1625,
      "step": 1195
    },
    {
      "epoch": 0.8883131303784584,
      "grad_norm": 1.8046875,
      "learning_rate": 7.427603073110967e-07,
      "loss": 2.1741,
      "step": 1200
    },
    {
      "epoch": 0.8920144350883686,
      "grad_norm": 1.796875,
      "learning_rate": 6.946402407843156e-07,
      "loss": 2.1467,
      "step": 1205
    },
    {
      "epoch": 0.8957157397982789,
      "grad_norm": 1.7578125,
      "learning_rate": 6.480754747781037e-07,
      "loss": 2.1597,
      "step": 1210
    },
    {
      "epoch": 0.8994170445081892,
      "grad_norm": 1.8046875,
      "learning_rate": 6.030737921409169e-07,
      "loss": 2.1653,
      "step": 1215
    },
    {
      "epoch": 0.9031183492180994,
      "grad_norm": 1.796875,
      "learning_rate": 5.596427144670002e-07,
      "loss": 2.173,
      "step": 1220
    },
    {
      "epoch": 0.9068196539280097,
      "grad_norm": 1.8125,
      "learning_rate": 5.177895008392353e-07,
      "loss": 2.1844,
      "step": 1225
    },
    {
      "epoch": 0.9105209586379198,
      "grad_norm": 1.7109375,
      "learning_rate": 4.775211466158469e-07,
      "loss": 2.1385,
      "step": 1230
    },
    {
      "epoch": 0.9142222633478301,
      "grad_norm": 1.7734375,
      "learning_rate": 4.388443822612043e-07,
      "loss": 2.1521,
      "step": 1235
    },
    {
      "epoch": 0.9179235680577403,
      "grad_norm": 1.7890625,
      "learning_rate": 4.017656722208807e-07,
      "loss": 2.1692,
      "step": 1240
    },
    {
      "epoch": 0.9216248727676506,
      "grad_norm": 1.7421875,
      "learning_rate": 3.662912138411967e-07,
      "loss": 2.1767,
      "step": 1245
    },
    {
      "epoch": 0.9253261774775609,
      "grad_norm": 1.796875,
      "learning_rate": 3.3242693633337986e-07,
      "loss": 2.1517,
      "step": 1250
    },
    {
      "epoch": 0.9290274821874711,
      "grad_norm": 1.796875,
      "learning_rate": 3.001784997825652e-07,
      "loss": 2.1849,
      "step": 1255
    },
    {
      "epoch": 0.9327287868973814,
      "grad_norm": 1.796875,
      "learning_rate": 2.6955129420176193e-07,
      "loss": 2.1564,
      "step": 1260
    },
    {
      "epoch": 0.9364300916072915,
      "grad_norm": 1.8125,
      "learning_rate": 2.405504386309643e-07,
      "loss": 2.1831,
      "step": 1265
    },
    {
      "epoch": 0.9401313963172018,
      "grad_norm": 1.7734375,
      "learning_rate": 2.1318078028155886e-07,
      "loss": 2.1436,
      "step": 1270
    },
    {
      "epoch": 0.943832701027112,
      "grad_norm": 1.796875,
      "learning_rate": 1.874468937261531e-07,
      "loss": 2.159,
      "step": 1275
    },
    {
      "epoch": 0.9475340057370223,
      "grad_norm": 1.78125,
      "learning_rate": 1.6335308013398888e-07,
      "loss": 2.1677,
      "step": 1280
    },
    {
      "epoch": 0.9512353104469325,
      "grad_norm": 1.7890625,
      "learning_rate": 1.409033665520354e-07,
      "loss": 2.1872,
      "step": 1285
    },
    {
      "epoch": 0.9549366151568428,
      "grad_norm": 1.78125,
      "learning_rate": 1.201015052319099e-07,
      "loss": 2.1408,
      "step": 1290
    },
    {
      "epoch": 0.9586379198667531,
      "grad_norm": 1.75,
      "learning_rate": 1.0095097300273026e-07,
      "loss": 2.1421,
      "step": 1295
    },
    {
      "epoch": 0.9623392245766633,
      "grad_norm": 1.7890625,
      "learning_rate": 8.345497068998897e-08,
      "loss": 2.1545,
      "step": 1300
    },
    {
      "epoch": 0.9660405292865735,
      "grad_norm": 1.7734375,
      "learning_rate": 6.761642258056977e-08,
      "loss": 2.1459,
      "step": 1305
    },
    {
      "epoch": 0.9697418339964837,
      "grad_norm": 1.7421875,
      "learning_rate": 5.3437975933985366e-08,
      "loss": 2.1508,
      "step": 1310
    },
    {
      "epoch": 0.973443138706394,
      "grad_norm": 1.78125,
      "learning_rate": 4.0922000539906914e-08,
      "loss": 2.1533,
      "step": 1315
    },
    {
      "epoch": 0.9771444434163042,
      "grad_norm": 1.78125,
      "learning_rate": 3.0070588322079765e-08,
      "loss": 2.1527,
      "step": 1320
    },
    {
      "epoch": 0.9808457481262145,
      "grad_norm": 1.921875,
      "learning_rate": 2.088555298867978e-08,
      "loss": 2.1634,
      "step": 1325
    },
    {
      "epoch": 0.9845470528361248,
      "grad_norm": 1.75,
      "learning_rate": 1.3368429729168075e-08,
      "loss": 2.1714,
      "step": 1330
    },
    {
      "epoch": 0.988248357546035,
      "grad_norm": 1.75,
      "learning_rate": 7.520474957699586e-09,
      "loss": 2.1449,
      "step": 1335
    },
    {
      "epoch": 0.9919496622559453,
      "grad_norm": 1.7578125,
      "learning_rate": 3.3426661031255024e-09,
      "loss": 2.1756,
      "step": 1340
    },
    {
      "epoch": 0.9956509669658554,
      "grad_norm": 1.765625,
      "learning_rate": 8.357014456272794e-10,
      "loss": 2.1534,
      "step": 1345
    },
    {
      "epoch": 0.9993522716757657,
      "grad_norm": 1.796875,
      "learning_rate": 0.0,
      "loss": 2.1688,
      "step": 1350
    },
    {
      "epoch": 0.9993522716757657,
      "eval_loss": 2.16412353515625,
      "eval_runtime": 221.9604,
      "eval_samples_per_second": 5.379,
      "eval_steps_per_second": 2.69,
      "step": 1350
    },
    {
      "epoch": 0.9993522716757657,
      "step": 1350,
      "total_flos": 5.373491620442276e+17,
      "train_loss": 2.19371005181913,
      "train_runtime": 17533.5186,
      "train_samples_per_second": 1.233,
      "train_steps_per_second": 0.077
    }
  ],
  "logging_steps": 5,
  "max_steps": 1350,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.373491620442276e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}