{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 1473,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00679001867255135,
      "grad_norm": 0.18874701857566833,
      "learning_rate": 6.081081081081082e-06,
      "loss": 0.0775,
      "step": 10
    },
    {
      "epoch": 0.0135800373451027,
      "grad_norm": 0.1703234165906906,
      "learning_rate": 1.2837837837837838e-05,
      "loss": 0.0726,
      "step": 20
    },
    {
      "epoch": 0.02037005601765405,
      "grad_norm": 0.17572833597660065,
      "learning_rate": 1.9594594594594595e-05,
      "loss": 0.0848,
      "step": 30
    },
    {
      "epoch": 0.0271600746902054,
      "grad_norm": 0.12470663338899612,
      "learning_rate": 2.635135135135135e-05,
      "loss": 0.0587,
      "step": 40
    },
    {
      "epoch": 0.03395009336275675,
      "grad_norm": 0.35773468017578125,
      "learning_rate": 3.310810810810811e-05,
      "loss": 0.0648,
      "step": 50
    },
    {
      "epoch": 0.0407401120353081,
      "grad_norm": 0.28681740164756775,
      "learning_rate": 3.986486486486487e-05,
      "loss": 0.0687,
      "step": 60
    },
    {
      "epoch": 0.04753013070785945,
      "grad_norm": 0.2551625967025757,
      "learning_rate": 4.662162162162162e-05,
      "loss": 0.0569,
      "step": 70
    },
    {
      "epoch": 0.0543201493804108,
      "grad_norm": 0.2852691411972046,
      "learning_rate": 5.337837837837838e-05,
      "loss": 0.0479,
      "step": 80
    },
    {
      "epoch": 0.06111016805296215,
      "grad_norm": 0.6955500245094299,
      "learning_rate": 6.013513513513514e-05,
      "loss": 0.0519,
      "step": 90
    },
    {
      "epoch": 0.0679001867255135,
      "grad_norm": 0.4106616973876953,
      "learning_rate": 6.68918918918919e-05,
      "loss": 0.044,
      "step": 100
    },
    {
      "epoch": 0.07469020539806484,
      "grad_norm": 0.40094736218452454,
      "learning_rate": 7.364864864864865e-05,
      "loss": 0.0395,
      "step": 110
    },
    {
      "epoch": 0.0814802240706162,
      "grad_norm": 0.5327359437942505,
      "learning_rate": 8.040540540540541e-05,
      "loss": 0.0272,
      "step": 120
    },
    {
      "epoch": 0.08827024274316754,
      "grad_norm": 0.6539778113365173,
      "learning_rate": 8.716216216216216e-05,
      "loss": 0.041,
      "step": 130
    },
    {
      "epoch": 0.0950602614157189,
      "grad_norm": 0.35497355461120605,
      "learning_rate": 9.391891891891892e-05,
      "loss": 0.0269,
      "step": 140
    },
    {
      "epoch": 0.10185028008827024,
      "grad_norm": 0.4729565978050232,
      "learning_rate": 9.999985945746134e-05,
      "loss": 0.0297,
      "step": 150
    },
    {
      "epoch": 0.1086402987608216,
      "grad_norm": 0.49648669362068176,
      "learning_rate": 9.998299530880787e-05,
      "loss": 0.0263,
      "step": 160
    },
    {
      "epoch": 0.11543031743337294,
      "grad_norm": 0.3857593536376953,
      "learning_rate": 9.993803351513094e-05,
      "loss": 0.0373,
      "step": 170
    },
    {
      "epoch": 0.1222203361059243,
      "grad_norm": 0.17934544384479523,
      "learning_rate": 9.986499935143679e-05,
      "loss": 0.0239,
      "step": 180
    },
    {
      "epoch": 0.12901035477847564,
      "grad_norm": 0.4041518270969391,
      "learning_rate": 9.976393387344834e-05,
      "loss": 0.0267,
      "step": 190
    },
    {
      "epoch": 0.135800373451027,
      "grad_norm": 0.3943810760974884,
      "learning_rate": 9.963489389452596e-05,
      "loss": 0.0241,
      "step": 200
    },
    {
      "epoch": 0.14259039212357835,
      "grad_norm": 0.09221380203962326,
      "learning_rate": 9.947795195373016e-05,
      "loss": 0.0169,
      "step": 210
    },
    {
      "epoch": 0.14938041079612968,
      "grad_norm": 0.26015597581863403,
      "learning_rate": 9.92931962750442e-05,
      "loss": 0.0123,
      "step": 220
    },
    {
      "epoch": 0.15617042946868104,
      "grad_norm": 0.7605913281440735,
      "learning_rate": 9.908073071777954e-05,
      "loss": 0.017,
      "step": 230
    },
    {
      "epoch": 0.1629604481412324,
      "grad_norm": 0.3495510220527649,
      "learning_rate": 9.884067471819184e-05,
      "loss": 0.0137,
      "step": 240
    },
    {
      "epoch": 0.16975046681378375,
      "grad_norm": 0.2134857475757599,
      "learning_rate": 9.857316322234067e-05,
      "loss": 0.0143,
      "step": 250
    },
    {
      "epoch": 0.17654048548633508,
      "grad_norm": 0.1996830254793167,
      "learning_rate": 9.827834661023034e-05,
      "loss": 0.0177,
      "step": 260
    },
    {
      "epoch": 0.18333050415888644,
      "grad_norm": 0.44393885135650635,
      "learning_rate": 9.795639061127468e-05,
      "loss": 0.015,
      "step": 270
    },
    {
      "epoch": 0.1901205228314378,
      "grad_norm": 0.5129746198654175,
      "learning_rate": 9.760747621113325e-05,
      "loss": 0.0109,
      "step": 280
    },
    {
      "epoch": 0.19691054150398912,
      "grad_norm": 0.31817689538002014,
      "learning_rate": 9.723179954997125e-05,
      "loss": 0.0143,
      "step": 290
    },
    {
      "epoch": 0.20370056017654048,
      "grad_norm": 0.17889945209026337,
      "learning_rate": 9.682957181220062e-05,
      "loss": 0.0123,
      "step": 300
    },
    {
      "epoch": 0.21049057884909184,
      "grad_norm": 0.3338984251022339,
      "learning_rate": 9.640101910776381e-05,
      "loss": 0.0068,
      "step": 310
    },
    {
      "epoch": 0.2172805975216432,
      "grad_norm": 0.4307146966457367,
      "learning_rate": 9.594638234502753e-05,
      "loss": 0.0189,
      "step": 320
    },
    {
      "epoch": 0.22407061619419452,
      "grad_norm": 0.3624001443386078,
      "learning_rate": 9.546591709535751e-05,
      "loss": 0.0062,
      "step": 330
    },
    {
      "epoch": 0.23086063486674588,
      "grad_norm": 0.28696736693382263,
      "learning_rate": 9.495989344945056e-05,
      "loss": 0.0038,
      "step": 340
    },
    {
      "epoch": 0.23765065353929724,
      "grad_norm": 0.7615050673484802,
      "learning_rate": 9.442859586550468e-05,
      "loss": 0.0064,
      "step": 350
    },
    {
      "epoch": 0.2444406722118486,
      "grad_norm": 0.3822503983974457,
      "learning_rate": 9.387232300931255e-05,
      "loss": 0.0076,
      "step": 360
    },
    {
      "epoch": 0.2512306908843999,
      "grad_norm": 0.17375706136226654,
      "learning_rate": 9.32913875863684e-05,
      "loss": 0.0133,
      "step": 370
    },
    {
      "epoch": 0.2580207095569513,
      "grad_norm": 0.3693685531616211,
      "learning_rate": 9.268611616608236e-05,
      "loss": 0.0089,
      "step": 380
    },
    {
      "epoch": 0.26481072822950263,
      "grad_norm": 0.5162019729614258,
      "learning_rate": 9.205684899820147e-05,
      "loss": 0.0149,
      "step": 390
    },
    {
      "epoch": 0.271600746902054,
      "grad_norm": 0.07368151843547821,
      "learning_rate": 9.140393982154026e-05,
      "loss": 0.0107,
      "step": 400
    },
    {
      "epoch": 0.27839076557460535,
      "grad_norm": 0.3283478915691376,
      "learning_rate": 9.072775566512863e-05,
      "loss": 0.0148,
      "step": 410
    },
    {
      "epoch": 0.2851807842471567,
      "grad_norm": 0.041360560804605484,
      "learning_rate": 9.002867664188858e-05,
      "loss": 0.0051,
      "step": 420
    },
    {
      "epoch": 0.291970802919708,
      "grad_norm": 0.2192244976758957,
      "learning_rate": 8.930709573495602e-05,
      "loss": 0.0031,
      "step": 430
    },
    {
      "epoch": 0.29876082159225936,
      "grad_norm": 0.36345595121383667,
      "learning_rate": 8.856341857676758e-05,
      "loss": 0.0077,
      "step": 440
    },
    {
      "epoch": 0.3055508402648107,
      "grad_norm": 0.22609934210777283,
      "learning_rate": 8.779806322103682e-05,
      "loss": 0.0089,
      "step": 450
    },
    {
      "epoch": 0.3123408589373621,
      "grad_norm": 0.5008813738822937,
      "learning_rate": 8.701145990774775e-05,
      "loss": 0.0063,
      "step": 460
    },
    {
      "epoch": 0.31913087760991343,
      "grad_norm": 0.030453462153673172,
      "learning_rate": 8.620405082129798e-05,
      "loss": 0.0077,
      "step": 470
    },
    {
      "epoch": 0.3259208962824648,
      "grad_norm": 0.21866290271282196,
      "learning_rate": 8.537628984192736e-05,
      "loss": 0.0025,
      "step": 480
    },
    {
      "epoch": 0.33271091495501615,
      "grad_norm": 0.5812458395957947,
      "learning_rate": 8.452864229057191e-05,
      "loss": 0.0038,
      "step": 490
    },
    {
      "epoch": 0.3395009336275675,
      "grad_norm": 0.07420260459184647,
      "learning_rate": 8.366158466728644e-05,
      "loss": 0.0033,
      "step": 500
    },
    {
      "epoch": 0.3462909523001188,
      "grad_norm": 0.008573359809815884,
      "learning_rate": 8.277560438338278e-05,
      "loss": 0.0012,
      "step": 510
    },
    {
      "epoch": 0.35308097097267016,
      "grad_norm": 0.012327054515480995,
      "learning_rate": 8.18711994874345e-05,
      "loss": 0.0049,
      "step": 520
    },
    {
      "epoch": 0.3598709896452215,
      "grad_norm": 0.1665765792131424,
      "learning_rate": 8.094887838530174e-05,
      "loss": 0.0041,
      "step": 530
    },
    {
      "epoch": 0.3666610083177729,
      "grad_norm": 0.02857097238302231,
      "learning_rate": 8.000915955433396e-05,
      "loss": 0.0072,
      "step": 540
    },
    {
      "epoch": 0.37345102699032423,
      "grad_norm": 0.2231156975030899,
      "learning_rate": 7.905257125191094e-05,
      "loss": 0.0079,
      "step": 550
    },
    {
      "epoch": 0.3802410456628756,
      "grad_norm": 0.06368722766637802,
      "learning_rate": 7.807965121848597e-05,
      "loss": 0.005,
      "step": 560
    },
    {
      "epoch": 0.38703106433542694,
      "grad_norm": 0.01118999719619751,
      "learning_rate": 7.70909463752983e-05,
      "loss": 0.0019,
      "step": 570
    },
    {
      "epoch": 0.39382108300797825,
      "grad_norm": 0.21869057416915894,
      "learning_rate": 7.608701251692457e-05,
      "loss": 0.0061,
      "step": 580
    },
    {
      "epoch": 0.4006111016805296,
      "grad_norm": 0.5839071869850159,
      "learning_rate": 7.506841399884217e-05,
      "loss": 0.0057,
      "step": 590
    },
    {
      "epoch": 0.40740112035308096,
      "grad_norm": 0.009670675732195377,
      "learning_rate": 7.403572342018021e-05,
      "loss": 0.0066,
      "step": 600
    },
    {
      "epoch": 0.4141911390256323,
      "grad_norm": 0.02666589990258217,
      "learning_rate": 7.29895213018362e-05,
      "loss": 0.0037,
      "step": 610
    },
    {
      "epoch": 0.4209811576981837,
      "grad_norm": 0.012181079015135765,
      "learning_rate": 7.193039576013976e-05,
      "loss": 0.0027,
      "step": 620
    },
    {
      "epoch": 0.42777117637073503,
      "grad_norm": 0.031734827905893326,
      "learning_rate": 7.085894217624645e-05,
      "loss": 0.0009,
      "step": 630
    },
    {
      "epoch": 0.4345611950432864,
      "grad_norm": 0.003946160431951284,
      "learning_rate": 6.977576286144784e-05,
      "loss": 0.001,
      "step": 640
    },
    {
      "epoch": 0.44135121371583774,
      "grad_norm": 0.002033177064731717,
      "learning_rate": 6.868146671858567e-05,
      "loss": 0.0021,
      "step": 650
    },
    {
      "epoch": 0.44814123238838904,
      "grad_norm": 0.02190972864627838,
      "learning_rate": 6.757666889976085e-05,
      "loss": 0.0002,
      "step": 660
    },
    {
      "epoch": 0.4549312510609404,
      "grad_norm": 0.008110946044325829,
      "learning_rate": 6.646199046052921e-05,
      "loss": 0.0027,
      "step": 670
    },
    {
      "epoch": 0.46172126973349176,
      "grad_norm": 0.03216719999909401,
      "learning_rate": 6.533805801077899e-05,
      "loss": 0.0011,
      "step": 680
    },
    {
      "epoch": 0.4685112884060431,
      "grad_norm": 0.004164108540862799,
      "learning_rate": 6.420550336248558e-05,
      "loss": 0.002,
      "step": 690
    },
    {
      "epoch": 0.47530130707859447,
      "grad_norm": 0.003367300610989332,
      "learning_rate": 6.306496317454227e-05,
      "loss": 0.0002,
      "step": 700
    },
    {
      "epoch": 0.4820913257511458,
      "grad_norm": 0.004391076508909464,
      "learning_rate": 6.191707859486622e-05,
      "loss": 0.0071,
      "step": 710
    },
    {
      "epoch": 0.4888813444236972,
      "grad_norm": 0.007109352853149176,
      "learning_rate": 6.076249489998097e-05,
      "loss": 0.0002,
      "step": 720
    },
    {
      "epoch": 0.49567136309624854,
      "grad_norm": 0.004730923566967249,
      "learning_rate": 5.9601861132278e-05,
      "loss": 0.0019,
      "step": 730
    },
    {
      "epoch": 0.5024613817687998,
      "grad_norm": 0.5947654843330383,
      "learning_rate": 5.8435829735161416e-05,
      "loss": 0.0049,
      "step": 740
    },
    {
      "epoch": 0.5092514004413512,
      "grad_norm": 0.009204763919115067,
      "learning_rate": 5.7265056186280754e-05,
      "loss": 0.0013,
      "step": 750
    },
    {
      "epoch": 0.5160414191139026,
      "grad_norm": 0.14082202315330505,
      "learning_rate": 5.6090198629058134e-05,
      "loss": 0.0012,
      "step": 760
    },
    {
      "epoch": 0.5228314377864539,
      "grad_norm": 0.01261549536138773,
      "learning_rate": 5.491191750271677e-05,
      "loss": 0.0096,
      "step": 770
    },
    {
      "epoch": 0.5296214564590053,
      "grad_norm": 0.016321472823619843,
      "learning_rate": 5.373087517101899e-05,
      "loss": 0.0004,
      "step": 780
    },
    {
      "epoch": 0.5364114751315566,
      "grad_norm": 0.00708204647526145,
      "learning_rate": 5.254773554992242e-05,
      "loss": 0.0008,
      "step": 790
    },
    {
      "epoch": 0.543201493804108,
      "grad_norm": 0.4712439775466919,
      "learning_rate": 5.136316373436343e-05,
      "loss": 0.0006,
      "step": 800
    },
    {
      "epoch": 0.5499915124766593,
      "grad_norm": 0.006097007542848587,
      "learning_rate": 5.0177825624378114e-05,
      "loss": 0.0002,
      "step": 810
    },
    {
      "epoch": 0.5567815311492107,
      "grad_norm": 0.32815349102020264,
      "learning_rate": 4.8992387550770373e-05,
      "loss": 0.0054,
      "step": 820
    },
    {
      "epoch": 0.563571549821762,
      "grad_norm": 0.026494784280657768,
      "learning_rate": 4.780751590053813e-05,
      "loss": 0.0038,
      "step": 830
    },
    {
      "epoch": 0.5703615684943134,
      "grad_norm": 0.1990426480770111,
      "learning_rate": 4.662387674226771e-05,
      "loss": 0.0006,
      "step": 840
    },
    {
      "epoch": 0.5771515871668647,
      "grad_norm": 0.01720350980758667,
      "learning_rate": 4.544213545170741e-05,
      "loss": 0.0018,
      "step": 850
    },
    {
      "epoch": 0.583941605839416,
      "grad_norm": 0.0026177107356488705,
      "learning_rate": 4.4262956337730385e-05,
      "loss": 0.0033,
      "step": 860
    },
    {
      "epoch": 0.5907316245119674,
      "grad_norm": 0.010454360395669937,
      "learning_rate": 4.3087002268897434e-05,
      "loss": 0.0022,
      "step": 870
    },
    {
      "epoch": 0.5975216431845187,
      "grad_norm": 0.007146151270717382,
      "learning_rate": 4.191493430082929e-05,
      "loss": 0.0037,
      "step": 880
    },
    {
      "epoch": 0.6043116618570701,
      "grad_norm": 0.0408344529569149,
      "learning_rate": 4.074741130459813e-05,
      "loss": 0.0022,
      "step": 890
    },
    {
      "epoch": 0.6111016805296214,
      "grad_norm": 0.009422015398740768,
      "learning_rate": 3.9585089596347094e-05,
      "loss": 0.0003,
      "step": 900
    },
    {
      "epoch": 0.6178916992021728,
      "grad_norm": 0.005031352397054434,
      "learning_rate": 3.842862256834613e-05,
      "loss": 0.0032,
      "step": 910
    },
    {
      "epoch": 0.6246817178747242,
      "grad_norm": 0.007191929966211319,
      "learning_rate": 3.727866032169127e-05,
      "loss": 0.0011,
      "step": 920
    },
    {
      "epoch": 0.6314717365472755,
      "grad_norm": 0.006164975464344025,
      "learning_rate": 3.6135849300854225e-05,
      "loss": 0.0003,
      "step": 930
    },
    {
      "epoch": 0.6382617552198269,
      "grad_norm": 0.005800315178930759,
      "learning_rate": 3.500083193028741e-05,
      "loss": 0.0026,
      "step": 940
    },
    {
      "epoch": 0.6450517738923782,
      "grad_norm": 0.010351837612688541,
      "learning_rate": 3.387424625328892e-05,
      "loss": 0.0004,
      "step": 950
    },
    {
      "epoch": 0.6518417925649296,
      "grad_norm": 0.007848144508898258,
      "learning_rate": 3.275672557333015e-05,
      "loss": 0.0026,
      "step": 960
    },
    {
      "epoch": 0.6586318112374809,
      "grad_norm": 0.012394148856401443,
      "learning_rate": 3.164889809804808e-05,
      "loss": 0.0003,
      "step": 970
    },
    {
      "epoch": 0.6654218299100323,
      "grad_norm": 0.003295676549896598,
      "learning_rate": 3.055138658610202e-05,
      "loss": 0.0023,
      "step": 980
    },
    {
      "epoch": 0.6722118485825836,
      "grad_norm": 0.006249323021620512,
      "learning_rate": 2.946480799709358e-05,
      "loss": 0.002,
      "step": 990
    },
    {
      "epoch": 0.679001867255135,
      "grad_norm": 0.011923151090741158,
      "learning_rate": 2.83897731447464e-05,
      "loss": 0.0083,
      "step": 1000
    },
    {
      "epoch": 0.6857918859276863,
      "grad_norm": 0.24134011566638947,
      "learning_rate": 2.7326886353540916e-05,
      "loss": 0.007,
      "step": 1010
    },
    {
      "epoch": 0.6925819046002376,
      "grad_norm": 0.5382188558578491,
      "learning_rate": 2.627674511899678e-05,
      "loss": 0.0035,
      "step": 1020
    },
    {
      "epoch": 0.699371923272789,
      "grad_norm": 0.023413360118865967,
      "learning_rate": 2.5239939771794563e-05,
      "loss": 0.0007,
      "step": 1030
    },
    {
      "epoch": 0.7061619419453403,
      "grad_norm": 0.06131948158144951,
      "learning_rate": 2.421705314592452e-05,
      "loss": 0.0006,
      "step": 1040
    },
    {
      "epoch": 0.7129519606178917,
      "grad_norm": 0.3948290944099426,
      "learning_rate": 2.3208660251050158e-05,
      "loss": 0.0047,
      "step": 1050
    },
    {
      "epoch": 0.719741979290443,
      "grad_norm": 0.007734253071248531,
      "learning_rate": 2.2215327949269886e-05,
      "loss": 0.0015,
      "step": 1060
    },
    {
      "epoch": 0.7265319979629944,
      "grad_norm": 0.006417641416192055,
      "learning_rate": 2.123761463645878e-05,
      "loss": 0.0003,
      "step": 1070
    },
    {
      "epoch": 0.7333220166355457,
      "grad_norm": 0.007732924073934555,
      "learning_rate": 2.0276069928369746e-05,
      "loss": 0.0003,
      "step": 1080
    },
    {
      "epoch": 0.7401120353080971,
      "grad_norm": 0.005477888975292444,
      "learning_rate": 1.933123435167018e-05,
      "loss": 0.0015,
      "step": 1090
    },
    {
      "epoch": 0.7469020539806485,
      "grad_norm": 0.0037697781808674335,
      "learning_rate": 1.840363904008819e-05,
      "loss": 0.0003,
      "step": 1100
    },
    {
      "epoch": 0.7536920726531998,
      "grad_norm": 0.007240783888846636,
      "learning_rate": 1.7493805435838874e-05,
      "loss": 0.0002,
      "step": 1110
    },
    {
      "epoch": 0.7604820913257512,
      "grad_norm": 0.008032640442252159,
      "learning_rate": 1.660224499649861e-05,
      "loss": 0.0004,
      "step": 1120
    },
    {
      "epoch": 0.7672721099983025,
      "grad_norm": 0.47604888677597046,
      "learning_rate": 1.5729458907492327e-05,
      "loss": 0.0025,
      "step": 1130
    },
    {
      "epoch": 0.7740621286708539,
      "grad_norm": 0.003562426194548607,
      "learning_rate": 1.4875937800354988e-05,
      "loss": 0.0022,
      "step": 1140
    },
    {
      "epoch": 0.7808521473434052,
      "grad_norm": 0.007144240662455559,
      "learning_rate": 1.404216147692598e-05,
      "loss": 0.0002,
      "step": 1150
    },
    {
      "epoch": 0.7876421660159565,
      "grad_norm": 0.007316683419048786,
      "learning_rate": 1.3228598639631418e-05,
      "loss": 0.0043,
      "step": 1160
    },
    {
      "epoch": 0.7944321846885078,
      "grad_norm": 0.005215325392782688,
      "learning_rate": 1.2435706628005767e-05,
      "loss": 0.0002,
      "step": 1170
    },
    {
      "epoch": 0.8012222033610592,
      "grad_norm": 0.00847008265554905,
      "learning_rate": 1.1663931161601188e-05,
      "loss": 0.0027,
      "step": 1180
    },
    {
      "epoch": 0.8080122220336106,
      "grad_norm": 0.007614725735038519,
      "learning_rate": 1.0913706089428932e-05,
      "loss": 0.0003,
      "step": 1190
    },
    {
      "epoch": 0.8148022407061619,
      "grad_norm": 0.006003544665873051,
      "learning_rate": 1.0185453146073604e-05,
      "loss": 0.0002,
      "step": 1200
    },
    {
      "epoch": 0.8215922593787133,
      "grad_norm": 0.017720144242048264,
      "learning_rate": 9.479581714617668e-06,
      "loss": 0.0003,
      "step": 1210
    },
    {
      "epoch": 0.8283822780512646,
      "grad_norm": 0.004918493330478668,
      "learning_rate": 8.796488596509133e-06,
      "loss": 0.0014,
      "step": 1220
    },
    {
      "epoch": 0.835172296723816,
      "grad_norm": 0.006771822925657034,
      "learning_rate": 8.136557788501903e-06,
      "loss": 0.0033,
      "step": 1230
    },
    {
      "epoch": 0.8419623153963673,
      "grad_norm": 0.00858241692185402,
      "learning_rate": 7.500160266794371e-06,
      "loss": 0.0002,
      "step": 1240
    },
    {
      "epoch": 0.8487523340689187,
      "grad_norm": 0.007278708275407553,
      "learning_rate": 6.887653778487307e-06,
      "loss": 0.0018,
      "step": 1250
    },
    {
      "epoch": 0.8555423527414701,
      "grad_norm": 0.0077684009447693825,
      "learning_rate": 6.2993826404783965e-06,
      "loss": 0.0002,
      "step": 1260
    },
    {
      "epoch": 0.8623323714140214,
      "grad_norm": 0.0030015786178410053,
      "learning_rate": 5.735677545906626e-06,
      "loss": 0.0006,
      "step": 1270
    },
    {
      "epoch": 0.8691223900865728,
      "grad_norm": 0.34356826543807983,
      "learning_rate": 5.196855378254989e-06,
      "loss": 0.0017,
      "step": 1280
    },
    {
      "epoch": 0.8759124087591241,
      "grad_norm": 0.006315717939287424,
      "learning_rate": 4.683219033216402e-06,
      "loss": 0.0002,
      "step": 1290
    },
    {
      "epoch": 0.8827024274316755,
      "grad_norm": 0.007271610666066408,
      "learning_rate": 4.1950572484226345e-06,
      "loss": 0.0002,
      "step": 1300
    },
    {
      "epoch": 0.8894924461042268,
      "grad_norm": 0.0047674220986664295,
      "learning_rate": 3.7326444411321547e-06,
      "loss": 0.0002,
      "step": 1310
    },
    {
      "epoch": 0.8962824647767781,
      "grad_norm": 0.003703842405229807,
      "learning_rate": 3.2962405539681217e-06,
      "loss": 0.0002,
      "step": 1320
    },
    {
      "epoch": 0.9030724834493294,
      "grad_norm": 0.00623746681958437,
      "learning_rate": 2.8860909087931543e-06,
      "loss": 0.0006,
      "step": 1330
    },
    {
      "epoch": 0.9098625021218808,
      "grad_norm": 0.005532027687877417,
      "learning_rate": 2.5024260688030987e-06,
      "loss": 0.0001,
      "step": 1340
    },
    {
      "epoch": 0.9166525207944322,
      "grad_norm": 0.009211459197103977,
      "learning_rate": 2.145461708917312e-06,
      "loss": 0.0028,
      "step": 1350
    },
    {
      "epoch": 0.9234425394669835,
      "grad_norm": 0.0028671075124293566,
      "learning_rate": 1.8153984945382452e-06,
      "loss": 0.0002,
      "step": 1360
    },
    {
      "epoch": 0.9302325581395349,
      "grad_norm": 0.006661532912403345,
      "learning_rate": 1.512421968748623e-06,
      "loss": 0.0003,
      "step": 1370
    },
    {
      "epoch": 0.9370225768120862,
      "grad_norm": 0.004116351250559092,
      "learning_rate": 1.2367024480094691e-06,
      "loss": 0.0006,
      "step": 1380
    },
    {
      "epoch": 0.9438125954846376,
      "grad_norm": 0.006897938437759876,
      "learning_rate": 9.883949264176962e-07,
      "loss": 0.0022,
      "step": 1390
    },
    {
      "epoch": 0.9506026141571889,
      "grad_norm": 0.004066048189997673,
      "learning_rate": 7.676389885771518e-07,
      "loss": 0.0002,
      "step": 1400
    },
    {
      "epoch": 0.9573926328297403,
      "grad_norm": 0.006368360482156277,
      "learning_rate": 5.745587311318968e-07,
      "loss": 0.0002,
      "step": 1410
    },
    {
      "epoch": 0.9641826515022917,
      "grad_norm": 0.005802372936159372,
      "learning_rate": 4.0926269300603503e-07,
      "loss": 0.0002,
      "step": 1420
    },
    {
      "epoch": 0.970972670174843,
      "grad_norm": 0.014351447112858295,
      "learning_rate": 2.71843794389226e-07,
      "loss": 0.0002,
      "step": 1430
    },
    {
      "epoch": 0.9777626888473944,
      "grad_norm": 0.003448351053521037,
      "learning_rate": 1.623792845020955e-07,
      "loss": 0.0022,
      "step": 1440
    },
    {
      "epoch": 0.9845527075199457,
      "grad_norm": 0.003692623693495989,
      "learning_rate": 8.093069817109445e-08,
      "loss": 0.0002,
      "step": 1450
    },
    {
      "epoch": 0.9913427261924971,
      "grad_norm": 0.004655805416405201,
      "learning_rate": 2.7543821237030475e-08,
      "loss": 0.0002,
      "step": 1460
    },
    {
      "epoch": 0.9981327448650483,
      "grad_norm": 0.005727958399802446,
      "learning_rate": 2.2486648168396075e-09,
      "loss": 0.0002,
      "step": 1470
    }
  ],
  "logging_steps": 10,
  "max_steps": 1473,
  "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": 1.364419674648871e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}