{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.03161089100564781,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00021073927337098542,
      "grad_norm": 4.289033889770508,
      "learning_rate": 4.998946303633145e-05,
      "loss": 3.8267,
      "step": 10
    },
    {
      "epoch": 0.00042147854674197085,
      "grad_norm": 4.060084342956543,
      "learning_rate": 4.997997976902976e-05,
      "loss": 3.6144,
      "step": 20
    },
    {
      "epoch": 0.0006322178201129563,
      "grad_norm": 4.173689842224121,
      "learning_rate": 4.996944280536121e-05,
      "loss": 3.4868,
      "step": 30
    },
    {
      "epoch": 0.0008429570934839417,
      "grad_norm": 5.22714376449585,
      "learning_rate": 4.995890584169266e-05,
      "loss": 3.4488,
      "step": 40
    },
    {
      "epoch": 0.001053696366854927,
      "grad_norm": 5.710427284240723,
      "learning_rate": 4.994836887802411e-05,
      "loss": 3.3072,
      "step": 50
    },
    {
      "epoch": 0.0012644356402259126,
      "grad_norm": 5.226071357727051,
      "learning_rate": 4.993783191435556e-05,
      "loss": 3.3811,
      "step": 60
    },
    {
      "epoch": 0.0014751749135968979,
      "grad_norm": 4.471126556396484,
      "learning_rate": 4.9927294950687014e-05,
      "loss": 3.3294,
      "step": 70
    },
    {
      "epoch": 0.0016859141869678834,
      "grad_norm": 3.511232376098633,
      "learning_rate": 4.991675798701846e-05,
      "loss": 3.274,
      "step": 80
    },
    {
      "epoch": 0.0018966534603388687,
      "grad_norm": 3.6000168323516846,
      "learning_rate": 4.990622102334991e-05,
      "loss": 3.2927,
      "step": 90
    },
    {
      "epoch": 0.002107392733709854,
      "grad_norm": 3.6504461765289307,
      "learning_rate": 4.9895684059681366e-05,
      "loss": 3.1517,
      "step": 100
    },
    {
      "epoch": 0.0023181320070808395,
      "grad_norm": 5.8762288093566895,
      "learning_rate": 4.9885147096012814e-05,
      "loss": 3.2165,
      "step": 110
    },
    {
      "epoch": 0.002528871280451825,
      "grad_norm": 3.862133264541626,
      "learning_rate": 4.987461013234427e-05,
      "loss": 3.2496,
      "step": 120
    },
    {
      "epoch": 0.0027396105538228105,
      "grad_norm": 3.1009256839752197,
      "learning_rate": 4.986407316867572e-05,
      "loss": 3.1269,
      "step": 130
    },
    {
      "epoch": 0.0029503498271937958,
      "grad_norm": 4.951446056365967,
      "learning_rate": 4.9853536205007165e-05,
      "loss": 3.142,
      "step": 140
    },
    {
      "epoch": 0.003161089100564781,
      "grad_norm": 4.305763244628906,
      "learning_rate": 4.9842999241338614e-05,
      "loss": 3.1518,
      "step": 150
    },
    {
      "epoch": 0.003371828373935767,
      "grad_norm": 4.36134672164917,
      "learning_rate": 4.983246227767007e-05,
      "loss": 3.1285,
      "step": 160
    },
    {
      "epoch": 0.003582567647306752,
      "grad_norm": 3.9053471088409424,
      "learning_rate": 4.9821925314001524e-05,
      "loss": 3.0738,
      "step": 170
    },
    {
      "epoch": 0.0037933069206777374,
      "grad_norm": 13.242171287536621,
      "learning_rate": 4.981138835033297e-05,
      "loss": 3.0425,
      "step": 180
    },
    {
      "epoch": 0.004004046194048723,
      "grad_norm": 3.8637614250183105,
      "learning_rate": 4.980085138666442e-05,
      "loss": 3.007,
      "step": 190
    },
    {
      "epoch": 0.004214785467419708,
      "grad_norm": 4.488512992858887,
      "learning_rate": 4.979031442299587e-05,
      "loss": 3.0578,
      "step": 200
    },
    {
      "epoch": 0.004425524740790694,
      "grad_norm": 5.628246307373047,
      "learning_rate": 4.9779777459327324e-05,
      "loss": 3.0215,
      "step": 210
    },
    {
      "epoch": 0.004636264014161679,
      "grad_norm": 3.368596315383911,
      "learning_rate": 4.976924049565878e-05,
      "loss": 3.096,
      "step": 220
    },
    {
      "epoch": 0.004847003287532664,
      "grad_norm": 7.204585075378418,
      "learning_rate": 4.975870353199023e-05,
      "loss": 2.9822,
      "step": 230
    },
    {
      "epoch": 0.00505774256090365,
      "grad_norm": 4.260004997253418,
      "learning_rate": 4.9748166568321675e-05,
      "loss": 3.0716,
      "step": 240
    },
    {
      "epoch": 0.005268481834274636,
      "grad_norm": 4.109992504119873,
      "learning_rate": 4.973762960465312e-05,
      "loss": 2.9789,
      "step": 250
    },
    {
      "epoch": 0.005479221107645621,
      "grad_norm": 10.826703071594238,
      "learning_rate": 4.972709264098457e-05,
      "loss": 2.9928,
      "step": 260
    },
    {
      "epoch": 0.005689960381016606,
      "grad_norm": 4.802614212036133,
      "learning_rate": 4.9716555677316027e-05,
      "loss": 3.0421,
      "step": 270
    },
    {
      "epoch": 0.0059006996543875916,
      "grad_norm": 4.688140392303467,
      "learning_rate": 4.970601871364748e-05,
      "loss": 2.8983,
      "step": 280
    },
    {
      "epoch": 0.006111438927758577,
      "grad_norm": 4.237213134765625,
      "learning_rate": 4.969548174997893e-05,
      "loss": 3.0775,
      "step": 290
    },
    {
      "epoch": 0.006322178201129562,
      "grad_norm": 6.8719000816345215,
      "learning_rate": 4.968494478631038e-05,
      "loss": 2.9356,
      "step": 300
    },
    {
      "epoch": 0.006532917474500548,
      "grad_norm": 6.765329360961914,
      "learning_rate": 4.9674407822641826e-05,
      "loss": 3.0783,
      "step": 310
    },
    {
      "epoch": 0.006743656747871534,
      "grad_norm": 4.296706199645996,
      "learning_rate": 4.9664924555340135e-05,
      "loss": 2.831,
      "step": 320
    },
    {
      "epoch": 0.006954396021242519,
      "grad_norm": 5.505979537963867,
      "learning_rate": 4.965438759167158e-05,
      "loss": 3.0016,
      "step": 330
    },
    {
      "epoch": 0.007165135294613504,
      "grad_norm": 15.19791316986084,
      "learning_rate": 4.964385062800304e-05,
      "loss": 2.9707,
      "step": 340
    },
    {
      "epoch": 0.0073758745679844894,
      "grad_norm": 3.5731112957000732,
      "learning_rate": 4.9633313664334487e-05,
      "loss": 3.0052,
      "step": 350
    },
    {
      "epoch": 0.007586613841355475,
      "grad_norm": 4.196691989898682,
      "learning_rate": 4.962277670066594e-05,
      "loss": 3.0018,
      "step": 360
    },
    {
      "epoch": 0.00779735311472646,
      "grad_norm": 3.7209060192108154,
      "learning_rate": 4.961223973699739e-05,
      "loss": 2.9956,
      "step": 370
    },
    {
      "epoch": 0.008008092388097446,
      "grad_norm": 4.67873477935791,
      "learning_rate": 4.960170277332884e-05,
      "loss": 3.0256,
      "step": 380
    },
    {
      "epoch": 0.008218831661468431,
      "grad_norm": 4.471567630767822,
      "learning_rate": 4.9591165809660286e-05,
      "loss": 2.9656,
      "step": 390
    },
    {
      "epoch": 0.008429570934839417,
      "grad_norm": 4.6043195724487305,
      "learning_rate": 4.958062884599174e-05,
      "loss": 2.9317,
      "step": 400
    },
    {
      "epoch": 0.008640310208210402,
      "grad_norm": 5.873717308044434,
      "learning_rate": 4.9570091882323196e-05,
      "loss": 3.0048,
      "step": 410
    },
    {
      "epoch": 0.008851049481581387,
      "grad_norm": 4.8152923583984375,
      "learning_rate": 4.9559554918654645e-05,
      "loss": 2.933,
      "step": 420
    },
    {
      "epoch": 0.009061788754952373,
      "grad_norm": 3.463914632797241,
      "learning_rate": 4.954901795498609e-05,
      "loss": 2.9188,
      "step": 430
    },
    {
      "epoch": 0.009272528028323358,
      "grad_norm": 5.305572986602783,
      "learning_rate": 4.953848099131754e-05,
      "loss": 3.1017,
      "step": 440
    },
    {
      "epoch": 0.009483267301694343,
      "grad_norm": 3.027397394180298,
      "learning_rate": 4.952794402764899e-05,
      "loss": 2.9651,
      "step": 450
    },
    {
      "epoch": 0.009694006575065328,
      "grad_norm": 3.619413375854492,
      "learning_rate": 4.9517407063980444e-05,
      "loss": 3.0447,
      "step": 460
    },
    {
      "epoch": 0.009904745848436314,
      "grad_norm": 3.6718661785125732,
      "learning_rate": 4.95068701003119e-05,
      "loss": 2.9183,
      "step": 470
    },
    {
      "epoch": 0.0101154851218073,
      "grad_norm": 3.527235984802246,
      "learning_rate": 4.949633313664335e-05,
      "loss": 2.9889,
      "step": 480
    },
    {
      "epoch": 0.010326224395178286,
      "grad_norm": 3.9344675540924072,
      "learning_rate": 4.9485796172974796e-05,
      "loss": 2.9665,
      "step": 490
    },
    {
      "epoch": 0.010536963668549271,
      "grad_norm": 6.459798336029053,
      "learning_rate": 4.9475259209306244e-05,
      "loss": 2.9079,
      "step": 500
    },
    {
      "epoch": 0.010747702941920257,
      "grad_norm": 3.535411834716797,
      "learning_rate": 4.94647222456377e-05,
      "loss": 3.0064,
      "step": 510
    },
    {
      "epoch": 0.010958442215291242,
      "grad_norm": 3.441504716873169,
      "learning_rate": 4.9454185281969154e-05,
      "loss": 2.8912,
      "step": 520
    },
    {
      "epoch": 0.011169181488662227,
      "grad_norm": 6.144808292388916,
      "learning_rate": 4.94436483183006e-05,
      "loss": 2.9239,
      "step": 530
    },
    {
      "epoch": 0.011379920762033213,
      "grad_norm": 3.7004387378692627,
      "learning_rate": 4.943311135463205e-05,
      "loss": 2.9526,
      "step": 540
    },
    {
      "epoch": 0.011590660035404198,
      "grad_norm": 6.244487762451172,
      "learning_rate": 4.94225743909635e-05,
      "loss": 2.8755,
      "step": 550
    },
    {
      "epoch": 0.011801399308775183,
      "grad_norm": 3.984456777572632,
      "learning_rate": 4.9412037427294954e-05,
      "loss": 2.835,
      "step": 560
    },
    {
      "epoch": 0.012012138582146168,
      "grad_norm": 4.6722941398620605,
      "learning_rate": 4.94015004636264e-05,
      "loss": 2.9199,
      "step": 570
    },
    {
      "epoch": 0.012222877855517154,
      "grad_norm": 4.149166584014893,
      "learning_rate": 4.939096349995786e-05,
      "loss": 3.0706,
      "step": 580
    },
    {
      "epoch": 0.012433617128888139,
      "grad_norm": 3.566357374191284,
      "learning_rate": 4.9380426536289305e-05,
      "loss": 2.9011,
      "step": 590
    },
    {
      "epoch": 0.012644356402259124,
      "grad_norm": 3.4722342491149902,
      "learning_rate": 4.9369889572620754e-05,
      "loss": 2.9345,
      "step": 600
    },
    {
      "epoch": 0.01285509567563011,
      "grad_norm": 3.4608707427978516,
      "learning_rate": 4.935935260895221e-05,
      "loss": 2.8419,
      "step": 610
    },
    {
      "epoch": 0.013065834949001097,
      "grad_norm": 3.4312050342559814,
      "learning_rate": 4.934881564528366e-05,
      "loss": 2.9087,
      "step": 620
    },
    {
      "epoch": 0.013276574222372082,
      "grad_norm": 6.524892330169678,
      "learning_rate": 4.9338278681615105e-05,
      "loss": 2.8818,
      "step": 630
    },
    {
      "epoch": 0.013487313495743067,
      "grad_norm": 5.508037567138672,
      "learning_rate": 4.932774171794656e-05,
      "loss": 2.7335,
      "step": 640
    },
    {
      "epoch": 0.013698052769114052,
      "grad_norm": 4.094667911529541,
      "learning_rate": 4.931720475427801e-05,
      "loss": 2.9197,
      "step": 650
    },
    {
      "epoch": 0.013908792042485038,
      "grad_norm": 3.6645474433898926,
      "learning_rate": 4.9306667790609463e-05,
      "loss": 2.9647,
      "step": 660
    },
    {
      "epoch": 0.014119531315856023,
      "grad_norm": 3.711179256439209,
      "learning_rate": 4.929613082694091e-05,
      "loss": 2.9863,
      "step": 670
    },
    {
      "epoch": 0.014330270589227008,
      "grad_norm": 3.904421329498291,
      "learning_rate": 4.928559386327236e-05,
      "loss": 2.8947,
      "step": 680
    },
    {
      "epoch": 0.014541009862597994,
      "grad_norm": 3.3760502338409424,
      "learning_rate": 4.927505689960381e-05,
      "loss": 2.7375,
      "step": 690
    },
    {
      "epoch": 0.014751749135968979,
      "grad_norm": 4.068516254425049,
      "learning_rate": 4.926451993593526e-05,
      "loss": 2.8502,
      "step": 700
    },
    {
      "epoch": 0.014962488409339964,
      "grad_norm": 4.1656999588012695,
      "learning_rate": 4.925398297226672e-05,
      "loss": 2.8688,
      "step": 710
    },
    {
      "epoch": 0.01517322768271095,
      "grad_norm": 3.5596961975097656,
      "learning_rate": 4.9243446008598166e-05,
      "loss": 2.9094,
      "step": 720
    },
    {
      "epoch": 0.015383966956081935,
      "grad_norm": 4.3002753257751465,
      "learning_rate": 4.9232909044929615e-05,
      "loss": 2.8964,
      "step": 730
    },
    {
      "epoch": 0.01559470622945292,
      "grad_norm": 5.381309509277344,
      "learning_rate": 4.922237208126106e-05,
      "loss": 2.8543,
      "step": 740
    },
    {
      "epoch": 0.015805445502823905,
      "grad_norm": 5.13466739654541,
      "learning_rate": 4.921183511759252e-05,
      "loss": 2.8754,
      "step": 750
    },
    {
      "epoch": 0.016016184776194892,
      "grad_norm": 4.221550464630127,
      "learning_rate": 4.9201298153923966e-05,
      "loss": 2.9192,
      "step": 760
    },
    {
      "epoch": 0.016226924049565876,
      "grad_norm": 4.988082408905029,
      "learning_rate": 4.919076119025542e-05,
      "loss": 2.8325,
      "step": 770
    },
    {
      "epoch": 0.016437663322936863,
      "grad_norm": 6.187538146972656,
      "learning_rate": 4.918022422658687e-05,
      "loss": 2.8819,
      "step": 780
    },
    {
      "epoch": 0.016648402596307846,
      "grad_norm": 3.920051336288452,
      "learning_rate": 4.916968726291832e-05,
      "loss": 2.8123,
      "step": 790
    },
    {
      "epoch": 0.016859141869678834,
      "grad_norm": 3.4514334201812744,
      "learning_rate": 4.915915029924977e-05,
      "loss": 2.8523,
      "step": 800
    },
    {
      "epoch": 0.017069881143049817,
      "grad_norm": 5.007158279418945,
      "learning_rate": 4.914861333558122e-05,
      "loss": 2.8607,
      "step": 810
    },
    {
      "epoch": 0.017280620416420804,
      "grad_norm": 4.900107383728027,
      "learning_rate": 4.913807637191267e-05,
      "loss": 2.7612,
      "step": 820
    },
    {
      "epoch": 0.01749135968979179,
      "grad_norm": 7.45019006729126,
      "learning_rate": 4.9127539408244124e-05,
      "loss": 2.913,
      "step": 830
    },
    {
      "epoch": 0.017702098963162775,
      "grad_norm": 5.6856770515441895,
      "learning_rate": 4.911700244457557e-05,
      "loss": 2.8877,
      "step": 840
    },
    {
      "epoch": 0.01791283823653376,
      "grad_norm": 4.228877067565918,
      "learning_rate": 4.910646548090703e-05,
      "loss": 2.7294,
      "step": 850
    },
    {
      "epoch": 0.018123577509904745,
      "grad_norm": 3.372474193572998,
      "learning_rate": 4.9095928517238476e-05,
      "loss": 2.7951,
      "step": 860
    },
    {
      "epoch": 0.018334316783275732,
      "grad_norm": 3.3396689891815186,
      "learning_rate": 4.9085391553569924e-05,
      "loss": 2.8488,
      "step": 870
    },
    {
      "epoch": 0.018545056056646716,
      "grad_norm": 3.912947177886963,
      "learning_rate": 4.907485458990138e-05,
      "loss": 2.7816,
      "step": 880
    },
    {
      "epoch": 0.018755795330017703,
      "grad_norm": 3.7946770191192627,
      "learning_rate": 4.906431762623283e-05,
      "loss": 2.8841,
      "step": 890
    },
    {
      "epoch": 0.018966534603388686,
      "grad_norm": 4.324241638183594,
      "learning_rate": 4.905378066256428e-05,
      "loss": 2.8666,
      "step": 900
    },
    {
      "epoch": 0.019177273876759673,
      "grad_norm": 3.172816038131714,
      "learning_rate": 4.904324369889573e-05,
      "loss": 2.7794,
      "step": 910
    },
    {
      "epoch": 0.019388013150130657,
      "grad_norm": 3.8265621662139893,
      "learning_rate": 4.903270673522718e-05,
      "loss": 2.8162,
      "step": 920
    },
    {
      "epoch": 0.019598752423501644,
      "grad_norm": 3.4845385551452637,
      "learning_rate": 4.902216977155863e-05,
      "loss": 2.7385,
      "step": 930
    },
    {
      "epoch": 0.019809491696872628,
      "grad_norm": 4.434839248657227,
      "learning_rate": 4.901163280789008e-05,
      "loss": 2.6926,
      "step": 940
    },
    {
      "epoch": 0.020020230970243615,
      "grad_norm": 6.55767822265625,
      "learning_rate": 4.900109584422154e-05,
      "loss": 2.6728,
      "step": 950
    },
    {
      "epoch": 0.0202309702436146,
      "grad_norm": 3.1376519203186035,
      "learning_rate": 4.8990558880552985e-05,
      "loss": 2.8854,
      "step": 960
    },
    {
      "epoch": 0.020441709516985585,
      "grad_norm": 3.569209337234497,
      "learning_rate": 4.8980021916884433e-05,
      "loss": 2.7862,
      "step": 970
    },
    {
      "epoch": 0.020652448790356572,
      "grad_norm": 16.01164436340332,
      "learning_rate": 4.896948495321588e-05,
      "loss": 2.8844,
      "step": 980
    },
    {
      "epoch": 0.020863188063727556,
      "grad_norm": 3.2311861515045166,
      "learning_rate": 4.895894798954733e-05,
      "loss": 2.8773,
      "step": 990
    },
    {
      "epoch": 0.021073927337098543,
      "grad_norm": 4.047968864440918,
      "learning_rate": 4.8948411025878785e-05,
      "loss": 2.6955,
      "step": 1000
    },
    {
      "epoch": 0.021284666610469526,
      "grad_norm": 3.3213765621185303,
      "learning_rate": 4.893787406221024e-05,
      "loss": 2.8308,
      "step": 1010
    },
    {
      "epoch": 0.021495405883840513,
      "grad_norm": 3.658327102661133,
      "learning_rate": 4.892733709854169e-05,
      "loss": 2.8409,
      "step": 1020
    },
    {
      "epoch": 0.021706145157211497,
      "grad_norm": 3.6058104038238525,
      "learning_rate": 4.8916800134873136e-05,
      "loss": 2.8073,
      "step": 1030
    },
    {
      "epoch": 0.021916884430582484,
      "grad_norm": 3.622807741165161,
      "learning_rate": 4.8906263171204585e-05,
      "loss": 2.7389,
      "step": 1040
    },
    {
      "epoch": 0.022127623703953467,
      "grad_norm": 3.6087045669555664,
      "learning_rate": 4.889572620753604e-05,
      "loss": 2.7506,
      "step": 1050
    },
    {
      "epoch": 0.022338362977324455,
      "grad_norm": 3.1047306060791016,
      "learning_rate": 4.888518924386749e-05,
      "loss": 2.77,
      "step": 1060
    },
    {
      "epoch": 0.022549102250695438,
      "grad_norm": 3.4957761764526367,
      "learning_rate": 4.887465228019894e-05,
      "loss": 2.8526,
      "step": 1070
    },
    {
      "epoch": 0.022759841524066425,
      "grad_norm": 5.486833095550537,
      "learning_rate": 4.886411531653039e-05,
      "loss": 2.8144,
      "step": 1080
    },
    {
      "epoch": 0.022970580797437412,
      "grad_norm": 6.019243240356445,
      "learning_rate": 4.885357835286184e-05,
      "loss": 2.8095,
      "step": 1090
    },
    {
      "epoch": 0.023181320070808396,
      "grad_norm": 5.631433486938477,
      "learning_rate": 4.8843041389193294e-05,
      "loss": 2.6904,
      "step": 1100
    },
    {
      "epoch": 0.023392059344179383,
      "grad_norm": 3.162370204925537,
      "learning_rate": 4.883250442552474e-05,
      "loss": 2.6805,
      "step": 1110
    },
    {
      "epoch": 0.023602798617550366,
      "grad_norm": 3.7855355739593506,
      "learning_rate": 4.882196746185619e-05,
      "loss": 2.7785,
      "step": 1120
    },
    {
      "epoch": 0.023813537890921353,
      "grad_norm": 3.0686984062194824,
      "learning_rate": 4.8811430498187646e-05,
      "loss": 2.7163,
      "step": 1130
    },
    {
      "epoch": 0.024024277164292337,
      "grad_norm": 4.290556907653809,
      "learning_rate": 4.8800893534519094e-05,
      "loss": 2.8549,
      "step": 1140
    },
    {
      "epoch": 0.024235016437663324,
      "grad_norm": 4.224363327026367,
      "learning_rate": 4.879035657085055e-05,
      "loss": 2.7858,
      "step": 1150
    },
    {
      "epoch": 0.024445755711034307,
      "grad_norm": 4.114129066467285,
      "learning_rate": 4.8779819607182e-05,
      "loss": 2.7969,
      "step": 1160
    },
    {
      "epoch": 0.024656494984405294,
      "grad_norm": 4.066317081451416,
      "learning_rate": 4.8769282643513446e-05,
      "loss": 2.701,
      "step": 1170
    },
    {
      "epoch": 0.024867234257776278,
      "grad_norm": 3.0110621452331543,
      "learning_rate": 4.8758745679844894e-05,
      "loss": 2.9004,
      "step": 1180
    },
    {
      "epoch": 0.025077973531147265,
      "grad_norm": 4.144210338592529,
      "learning_rate": 4.874820871617635e-05,
      "loss": 2.8651,
      "step": 1190
    },
    {
      "epoch": 0.02528871280451825,
      "grad_norm": 3.226668119430542,
      "learning_rate": 4.8737671752507804e-05,
      "loss": 2.8332,
      "step": 1200
    },
    {
      "epoch": 0.025499452077889236,
      "grad_norm": 3.7419745922088623,
      "learning_rate": 4.872713478883925e-05,
      "loss": 2.8502,
      "step": 1210
    },
    {
      "epoch": 0.02571019135126022,
      "grad_norm": 6.740761756896973,
      "learning_rate": 4.87165978251707e-05,
      "loss": 2.7892,
      "step": 1220
    },
    {
      "epoch": 0.025920930624631206,
      "grad_norm": 4.075252056121826,
      "learning_rate": 4.870606086150215e-05,
      "loss": 2.7037,
      "step": 1230
    },
    {
      "epoch": 0.026131669898002193,
      "grad_norm": 3.7171854972839355,
      "learning_rate": 4.8695523897833604e-05,
      "loss": 2.7885,
      "step": 1240
    },
    {
      "epoch": 0.026342409171373177,
      "grad_norm": 3.2372097969055176,
      "learning_rate": 4.868498693416506e-05,
      "loss": 2.7827,
      "step": 1250
    },
    {
      "epoch": 0.026553148444744164,
      "grad_norm": 5.390050888061523,
      "learning_rate": 4.867444997049651e-05,
      "loss": 2.7929,
      "step": 1260
    },
    {
      "epoch": 0.026763887718115147,
      "grad_norm": 6.445965766906738,
      "learning_rate": 4.8663913006827955e-05,
      "loss": 2.7683,
      "step": 1270
    },
    {
      "epoch": 0.026974626991486134,
      "grad_norm": 4.038309574127197,
      "learning_rate": 4.8653376043159403e-05,
      "loss": 2.7644,
      "step": 1280
    },
    {
      "epoch": 0.027185366264857118,
      "grad_norm": 3.422018527984619,
      "learning_rate": 4.864283907949085e-05,
      "loss": 2.759,
      "step": 1290
    },
    {
      "epoch": 0.027396105538228105,
      "grad_norm": 3.4551455974578857,
      "learning_rate": 4.863230211582231e-05,
      "loss": 2.6919,
      "step": 1300
    },
    {
      "epoch": 0.02760684481159909,
      "grad_norm": 4.99464225769043,
      "learning_rate": 4.862176515215376e-05,
      "loss": 2.8362,
      "step": 1310
    },
    {
      "epoch": 0.027817584084970075,
      "grad_norm": 3.759676933288574,
      "learning_rate": 4.861122818848521e-05,
      "loss": 2.7712,
      "step": 1320
    },
    {
      "epoch": 0.02802832335834106,
      "grad_norm": 4.082064151763916,
      "learning_rate": 4.860069122481666e-05,
      "loss": 2.8224,
      "step": 1330
    },
    {
      "epoch": 0.028239062631712046,
      "grad_norm": 3.157449960708618,
      "learning_rate": 4.8590154261148106e-05,
      "loss": 2.7863,
      "step": 1340
    },
    {
      "epoch": 0.02844980190508303,
      "grad_norm": 5.148242950439453,
      "learning_rate": 4.857961729747956e-05,
      "loss": 2.7383,
      "step": 1350
    },
    {
      "epoch": 0.028660541178454017,
      "grad_norm": 3.794858694076538,
      "learning_rate": 4.856908033381101e-05,
      "loss": 2.7118,
      "step": 1360
    },
    {
      "epoch": 0.028871280451825004,
      "grad_norm": 3.0035576820373535,
      "learning_rate": 4.8558543370142465e-05,
      "loss": 2.7787,
      "step": 1370
    },
    {
      "epoch": 0.029082019725195987,
      "grad_norm": 4.196252822875977,
      "learning_rate": 4.854800640647391e-05,
      "loss": 2.7501,
      "step": 1380
    },
    {
      "epoch": 0.029292758998566974,
      "grad_norm": 7.164214611053467,
      "learning_rate": 4.853746944280536e-05,
      "loss": 2.7885,
      "step": 1390
    },
    {
      "epoch": 0.029503498271937958,
      "grad_norm": 3.252230167388916,
      "learning_rate": 4.8526932479136816e-05,
      "loss": 2.7875,
      "step": 1400
    },
    {
      "epoch": 0.029714237545308945,
      "grad_norm": 7.037599563598633,
      "learning_rate": 4.8516395515468265e-05,
      "loss": 2.7787,
      "step": 1410
    },
    {
      "epoch": 0.02992497681867993,
      "grad_norm": 3.5410661697387695,
      "learning_rate": 4.850585855179971e-05,
      "loss": 2.8644,
      "step": 1420
    },
    {
      "epoch": 0.030135716092050915,
      "grad_norm": 3.7410356998443604,
      "learning_rate": 4.849532158813117e-05,
      "loss": 2.8146,
      "step": 1430
    },
    {
      "epoch": 0.0303464553654219,
      "grad_norm": 7.049551963806152,
      "learning_rate": 4.8484784624462616e-05,
      "loss": 2.8427,
      "step": 1440
    },
    {
      "epoch": 0.030557194638792886,
      "grad_norm": 3.5179500579833984,
      "learning_rate": 4.847424766079407e-05,
      "loss": 2.7396,
      "step": 1450
    },
    {
      "epoch": 0.03076793391216387,
      "grad_norm": 3.117222785949707,
      "learning_rate": 4.846371069712552e-05,
      "loss": 2.6701,
      "step": 1460
    },
    {
      "epoch": 0.030978673185534857,
      "grad_norm": 2.9434943199157715,
      "learning_rate": 4.845317373345697e-05,
      "loss": 2.7827,
      "step": 1470
    },
    {
      "epoch": 0.03118941245890584,
      "grad_norm": 7.40554666519165,
      "learning_rate": 4.8442636769788416e-05,
      "loss": 2.6663,
      "step": 1480
    },
    {
      "epoch": 0.031400151732276824,
      "grad_norm": 3.6940155029296875,
      "learning_rate": 4.843209980611987e-05,
      "loss": 2.7505,
      "step": 1490
    },
    {
      "epoch": 0.03161089100564781,
      "grad_norm": 3.142557382583618,
      "learning_rate": 4.8421562842451326e-05,
      "loss": 2.7381,
      "step": 1500
    }
  ],
  "logging_steps": 10,
  "max_steps": 47452,
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3376102996231680.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}