{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9997846452029719,
  "eval_steps": 500,
  "global_step": 3714,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00026919349628512973,
      "learning_rate": 5e-09,
      "loss": 9.3094,
      "step": 1
    },
    {
      "epoch": 0.005383869925702595,
      "learning_rate": 1e-07,
      "loss": 9.7938,
      "step": 20
    },
    {
      "epoch": 0.01076773985140519,
      "learning_rate": 2e-07,
      "loss": 9.4725,
      "step": 40
    },
    {
      "epoch": 0.016151609777107785,
      "learning_rate": 3e-07,
      "loss": 8.3855,
      "step": 60
    },
    {
      "epoch": 0.02153547970281038,
      "learning_rate": 4e-07,
      "loss": 7.0557,
      "step": 80
    },
    {
      "epoch": 0.026919349628512976,
      "learning_rate": 5e-07,
      "loss": 6.2559,
      "step": 100
    },
    {
      "epoch": 0.03230321955421557,
      "learning_rate": 6e-07,
      "loss": 5.941,
      "step": 120
    },
    {
      "epoch": 0.03768708947991817,
      "learning_rate": 7e-07,
      "loss": 5.657,
      "step": 140
    },
    {
      "epoch": 0.04307095940562076,
      "learning_rate": 8e-07,
      "loss": 5.3831,
      "step": 160
    },
    {
      "epoch": 0.048454829331323356,
      "learning_rate": 9e-07,
      "loss": 5.476,
      "step": 180
    },
    {
      "epoch": 0.05383869925702595,
      "learning_rate": 1e-06,
      "loss": 5.081,
      "step": 200
    },
    {
      "epoch": 0.05922256918272854,
      "learning_rate": 9.99696069161151e-07,
      "loss": 4.9985,
      "step": 220
    },
    {
      "epoch": 0.06460643910843114,
      "learning_rate": 9.993921383223017e-07,
      "loss": 5.0411,
      "step": 240
    },
    {
      "epoch": 0.06999030903413374,
      "learning_rate": 9.990882074834525e-07,
      "loss": 5.0192,
      "step": 260
    },
    {
      "epoch": 0.07537417895983634,
      "learning_rate": 9.987842766446035e-07,
      "loss": 5.0827,
      "step": 280
    },
    {
      "epoch": 0.08075804888553892,
      "learning_rate": 9.984803458057545e-07,
      "loss": 4.9614,
      "step": 300
    },
    {
      "epoch": 0.08614191881124152,
      "learning_rate": 9.981764149669053e-07,
      "loss": 5.0168,
      "step": 320
    },
    {
      "epoch": 0.09152578873694411,
      "learning_rate": 9.97872484128056e-07,
      "loss": 5.0651,
      "step": 340
    },
    {
      "epoch": 0.09690965866264671,
      "learning_rate": 9.97568553289207e-07,
      "loss": 5.0639,
      "step": 360
    },
    {
      "epoch": 0.10229352858834931,
      "learning_rate": 9.972646224503578e-07,
      "loss": 4.8337,
      "step": 380
    },
    {
      "epoch": 0.1076773985140519,
      "learning_rate": 9.969606916115088e-07,
      "loss": 4.7206,
      "step": 400
    },
    {
      "epoch": 0.11306126843975449,
      "learning_rate": 9.966567607726596e-07,
      "loss": 4.8773,
      "step": 420
    },
    {
      "epoch": 0.11844513836545709,
      "learning_rate": 9.963528299338106e-07,
      "loss": 4.7987,
      "step": 440
    },
    {
      "epoch": 0.12382900829115968,
      "learning_rate": 9.960488990949614e-07,
      "loss": 4.7167,
      "step": 460
    },
    {
      "epoch": 0.12921287821686228,
      "learning_rate": 9.957449682561124e-07,
      "loss": 4.6926,
      "step": 480
    },
    {
      "epoch": 0.13459674814256486,
      "learning_rate": 9.954410374172632e-07,
      "loss": 4.8212,
      "step": 500
    },
    {
      "epoch": 0.13998061806826748,
      "learning_rate": 9.95137106578414e-07,
      "loss": 4.7605,
      "step": 520
    },
    {
      "epoch": 0.14536448799397006,
      "learning_rate": 9.94833175739565e-07,
      "loss": 4.8473,
      "step": 540
    },
    {
      "epoch": 0.15074835791967267,
      "learning_rate": 9.94529244900716e-07,
      "loss": 4.735,
      "step": 560
    },
    {
      "epoch": 0.15613222784537525,
      "learning_rate": 9.942253140618667e-07,
      "loss": 4.7372,
      "step": 580
    },
    {
      "epoch": 0.16151609777107784,
      "learning_rate": 9.939213832230175e-07,
      "loss": 4.8833,
      "step": 600
    },
    {
      "epoch": 0.16689996769678045,
      "learning_rate": 9.936174523841685e-07,
      "loss": 4.742,
      "step": 620
    },
    {
      "epoch": 0.17228383762248303,
      "learning_rate": 9.933135215453195e-07,
      "loss": 4.5842,
      "step": 640
    },
    {
      "epoch": 0.17766770754818564,
      "learning_rate": 9.930095907064703e-07,
      "loss": 4.7701,
      "step": 660
    },
    {
      "epoch": 0.18305157747388823,
      "learning_rate": 9.92705659867621e-07,
      "loss": 4.7634,
      "step": 680
    },
    {
      "epoch": 0.18843544739959084,
      "learning_rate": 9.92401729028772e-07,
      "loss": 4.6502,
      "step": 700
    },
    {
      "epoch": 0.19381931732529342,
      "learning_rate": 9.92097798189923e-07,
      "loss": 4.6672,
      "step": 720
    },
    {
      "epoch": 0.199203187250996,
      "learning_rate": 9.917938673510738e-07,
      "loss": 4.75,
      "step": 740
    },
    {
      "epoch": 0.20458705717669862,
      "learning_rate": 9.914899365122246e-07,
      "loss": 4.5932,
      "step": 760
    },
    {
      "epoch": 0.2099709271024012,
      "learning_rate": 9.911860056733756e-07,
      "loss": 4.6468,
      "step": 780
    },
    {
      "epoch": 0.2153547970281038,
      "learning_rate": 9.908820748345266e-07,
      "loss": 4.5592,
      "step": 800
    },
    {
      "epoch": 0.2207386669538064,
      "learning_rate": 9.905781439956774e-07,
      "loss": 4.6977,
      "step": 820
    },
    {
      "epoch": 0.22612253687950898,
      "learning_rate": 9.902742131568282e-07,
      "loss": 4.6447,
      "step": 840
    },
    {
      "epoch": 0.2315064068052116,
      "learning_rate": 9.899702823179792e-07,
      "loss": 4.6937,
      "step": 860
    },
    {
      "epoch": 0.23689027673091417,
      "learning_rate": 9.8966635147913e-07,
      "loss": 4.7408,
      "step": 880
    },
    {
      "epoch": 0.24227414665661678,
      "learning_rate": 9.89362420640281e-07,
      "loss": 4.6969,
      "step": 900
    },
    {
      "epoch": 0.24765801658231937,
      "learning_rate": 9.890584898014317e-07,
      "loss": 4.6467,
      "step": 920
    },
    {
      "epoch": 0.25304188650802195,
      "learning_rate": 9.887545589625827e-07,
      "loss": 4.6131,
      "step": 940
    },
    {
      "epoch": 0.25842575643372456,
      "learning_rate": 9.884506281237335e-07,
      "loss": 4.6146,
      "step": 960
    },
    {
      "epoch": 0.2638096263594272,
      "learning_rate": 9.881466972848845e-07,
      "loss": 4.6659,
      "step": 980
    },
    {
      "epoch": 0.26919349628512973,
      "learning_rate": 9.878427664460353e-07,
      "loss": 4.6548,
      "step": 1000
    },
    {
      "epoch": 0.27457736621083234,
      "learning_rate": 9.87538835607186e-07,
      "loss": 4.5025,
      "step": 1020
    },
    {
      "epoch": 0.27996123613653495,
      "learning_rate": 9.87234904768337e-07,
      "loss": 4.5433,
      "step": 1040
    },
    {
      "epoch": 0.28534510606223756,
      "learning_rate": 9.86930973929488e-07,
      "loss": 4.6703,
      "step": 1060
    },
    {
      "epoch": 0.2907289759879401,
      "learning_rate": 9.866270430906388e-07,
      "loss": 4.4159,
      "step": 1080
    },
    {
      "epoch": 0.29611284591364273,
      "learning_rate": 9.863231122517896e-07,
      "loss": 4.5891,
      "step": 1100
    },
    {
      "epoch": 0.30149671583934534,
      "learning_rate": 9.860191814129406e-07,
      "loss": 4.5233,
      "step": 1120
    },
    {
      "epoch": 0.3068805857650479,
      "learning_rate": 9.857152505740916e-07,
      "loss": 4.48,
      "step": 1140
    },
    {
      "epoch": 0.3122644556907505,
      "learning_rate": 9.854113197352424e-07,
      "loss": 4.651,
      "step": 1160
    },
    {
      "epoch": 0.3176483256164531,
      "learning_rate": 9.851073888963932e-07,
      "loss": 4.508,
      "step": 1180
    },
    {
      "epoch": 0.3230321955421557,
      "learning_rate": 9.848034580575442e-07,
      "loss": 4.361,
      "step": 1200
    },
    {
      "epoch": 0.3284160654678583,
      "learning_rate": 9.844995272186952e-07,
      "loss": 4.3891,
      "step": 1220
    },
    {
      "epoch": 0.3337999353935609,
      "learning_rate": 9.84195596379846e-07,
      "loss": 4.3769,
      "step": 1240
    },
    {
      "epoch": 0.3391838053192635,
      "learning_rate": 9.838916655409967e-07,
      "loss": 4.4891,
      "step": 1260
    },
    {
      "epoch": 0.34456767524496607,
      "learning_rate": 9.835877347021477e-07,
      "loss": 4.4237,
      "step": 1280
    },
    {
      "epoch": 0.3499515451706687,
      "learning_rate": 9.832838038632987e-07,
      "loss": 4.3194,
      "step": 1300
    },
    {
      "epoch": 0.3553354150963713,
      "learning_rate": 9.829798730244495e-07,
      "loss": 4.4627,
      "step": 1320
    },
    {
      "epoch": 0.36071928502207384,
      "learning_rate": 9.826759421856003e-07,
      "loss": 4.3827,
      "step": 1340
    },
    {
      "epoch": 0.36610315494777645,
      "learning_rate": 9.823720113467513e-07,
      "loss": 4.3972,
      "step": 1360
    },
    {
      "epoch": 0.37148702487347907,
      "learning_rate": 9.820680805079023e-07,
      "loss": 4.3642,
      "step": 1380
    },
    {
      "epoch": 0.3768708947991817,
      "learning_rate": 9.81764149669053e-07,
      "loss": 4.3781,
      "step": 1400
    },
    {
      "epoch": 0.38225476472488423,
      "learning_rate": 9.814602188302038e-07,
      "loss": 4.5385,
      "step": 1420
    },
    {
      "epoch": 0.38763863465058684,
      "learning_rate": 9.811562879913546e-07,
      "loss": 4.3083,
      "step": 1440
    },
    {
      "epoch": 0.39302250457628946,
      "learning_rate": 9.808523571525058e-07,
      "loss": 4.3722,
      "step": 1460
    },
    {
      "epoch": 0.398406374501992,
      "learning_rate": 9.805484263136566e-07,
      "loss": 4.3383,
      "step": 1480
    },
    {
      "epoch": 0.4037902444276946,
      "learning_rate": 9.802444954748074e-07,
      "loss": 4.3492,
      "step": 1500
    },
    {
      "epoch": 0.40917411435339723,
      "learning_rate": 9.799405646359582e-07,
      "loss": 4.3363,
      "step": 1520
    },
    {
      "epoch": 0.4145579842790998,
      "learning_rate": 9.796366337971094e-07,
      "loss": 4.365,
      "step": 1540
    },
    {
      "epoch": 0.4199418542048024,
      "learning_rate": 9.793327029582602e-07,
      "loss": 4.3037,
      "step": 1560
    },
    {
      "epoch": 0.425325724130505,
      "learning_rate": 9.79028772119411e-07,
      "loss": 4.4482,
      "step": 1580
    },
    {
      "epoch": 0.4307095940562076,
      "learning_rate": 9.787248412805617e-07,
      "loss": 4.3282,
      "step": 1600
    },
    {
      "epoch": 0.4360934639819102,
      "learning_rate": 9.78420910441713e-07,
      "loss": 4.1645,
      "step": 1620
    },
    {
      "epoch": 0.4414773339076128,
      "learning_rate": 9.781169796028637e-07,
      "loss": 4.2391,
      "step": 1640
    },
    {
      "epoch": 0.4468612038333154,
      "learning_rate": 9.778130487640145e-07,
      "loss": 4.3259,
      "step": 1660
    },
    {
      "epoch": 0.45224507375901796,
      "learning_rate": 9.775091179251653e-07,
      "loss": 4.3063,
      "step": 1680
    },
    {
      "epoch": 0.45762894368472057,
      "learning_rate": 9.772051870863165e-07,
      "loss": 4.1056,
      "step": 1700
    },
    {
      "epoch": 0.4630128136104232,
      "learning_rate": 9.769012562474673e-07,
      "loss": 4.3614,
      "step": 1720
    },
    {
      "epoch": 0.4683966835361258,
      "learning_rate": 9.76597325408618e-07,
      "loss": 4.2459,
      "step": 1740
    },
    {
      "epoch": 0.47378055346182835,
      "learning_rate": 9.762933945697688e-07,
      "loss": 4.3372,
      "step": 1760
    },
    {
      "epoch": 0.47916442338753096,
      "learning_rate": 9.7598946373092e-07,
      "loss": 4.2036,
      "step": 1780
    },
    {
      "epoch": 0.48454829331323357,
      "learning_rate": 9.756855328920708e-07,
      "loss": 4.2001,
      "step": 1800
    },
    {
      "epoch": 0.4899321632389361,
      "learning_rate": 9.753816020532216e-07,
      "loss": 4.331,
      "step": 1820
    },
    {
      "epoch": 0.49531603316463874,
      "learning_rate": 9.750776712143724e-07,
      "loss": 4.2512,
      "step": 1840
    },
    {
      "epoch": 0.5006999030903413,
      "learning_rate": 9.747737403755234e-07,
      "loss": 4.1834,
      "step": 1860
    },
    {
      "epoch": 0.5060837730160439,
      "learning_rate": 9.744698095366744e-07,
      "loss": 4.3188,
      "step": 1880
    },
    {
      "epoch": 0.5114676429417465,
      "learning_rate": 9.741658786978252e-07,
      "loss": 4.4222,
      "step": 1900
    },
    {
      "epoch": 0.5168515128674491,
      "learning_rate": 9.73861947858976e-07,
      "loss": 4.0525,
      "step": 1920
    },
    {
      "epoch": 0.5222353827931517,
      "learning_rate": 9.73558017020127e-07,
      "loss": 4.137,
      "step": 1940
    },
    {
      "epoch": 0.5276192527188543,
      "learning_rate": 9.73254086181278e-07,
      "loss": 4.1998,
      "step": 1960
    },
    {
      "epoch": 0.533003122644557,
      "learning_rate": 9.729501553424287e-07,
      "loss": 4.0201,
      "step": 1980
    },
    {
      "epoch": 0.5383869925702595,
      "learning_rate": 9.726462245035795e-07,
      "loss": 4.1648,
      "step": 2000
    },
    {
      "epoch": 0.5437708624959621,
      "learning_rate": 9.723422936647305e-07,
      "loss": 4.0576,
      "step": 2020
    },
    {
      "epoch": 0.5491547324216647,
      "learning_rate": 9.720383628258815e-07,
      "loss": 4.1884,
      "step": 2040
    },
    {
      "epoch": 0.5545386023473673,
      "learning_rate": 9.717344319870323e-07,
      "loss": 4.0947,
      "step": 2060
    },
    {
      "epoch": 0.5599224722730699,
      "learning_rate": 9.71430501148183e-07,
      "loss": 4.1478,
      "step": 2080
    },
    {
      "epoch": 0.5653063421987725,
      "learning_rate": 9.71126570309334e-07,
      "loss": 4.2073,
      "step": 2100
    },
    {
      "epoch": 0.5706902121244751,
      "learning_rate": 9.70822639470485e-07,
      "loss": 4.2448,
      "step": 2120
    },
    {
      "epoch": 0.5760740820501776,
      "learning_rate": 9.705187086316358e-07,
      "loss": 4.1576,
      "step": 2140
    },
    {
      "epoch": 0.5814579519758802,
      "learning_rate": 9.702147777927866e-07,
      "loss": 3.9818,
      "step": 2160
    },
    {
      "epoch": 0.5868418219015828,
      "learning_rate": 9.699108469539376e-07,
      "loss": 4.1782,
      "step": 2180
    },
    {
      "epoch": 0.5922256918272855,
      "learning_rate": 9.696069161150886e-07,
      "loss": 4.1193,
      "step": 2200
    },
    {
      "epoch": 0.5976095617529881,
      "learning_rate": 9.693029852762394e-07,
      "loss": 3.8744,
      "step": 2220
    },
    {
      "epoch": 0.6029934316786907,
      "learning_rate": 9.689990544373901e-07,
      "loss": 3.9768,
      "step": 2240
    },
    {
      "epoch": 0.6083773016043932,
      "learning_rate": 9.686951235985411e-07,
      "loss": 4.0686,
      "step": 2260
    },
    {
      "epoch": 0.6137611715300958,
      "learning_rate": 9.68391192759692e-07,
      "loss": 4.1936,
      "step": 2280
    },
    {
      "epoch": 0.6191450414557984,
      "learning_rate": 9.68087261920843e-07,
      "loss": 4.0601,
      "step": 2300
    },
    {
      "epoch": 0.624528911381501,
      "learning_rate": 9.677833310819937e-07,
      "loss": 4.1562,
      "step": 2320
    },
    {
      "epoch": 0.6299127813072036,
      "learning_rate": 9.674794002431447e-07,
      "loss": 4.1972,
      "step": 2340
    },
    {
      "epoch": 0.6352966512329062,
      "learning_rate": 9.671754694042955e-07,
      "loss": 4.1596,
      "step": 2360
    },
    {
      "epoch": 0.6406805211586089,
      "learning_rate": 9.668715385654465e-07,
      "loss": 4.1037,
      "step": 2380
    },
    {
      "epoch": 0.6460643910843114,
      "learning_rate": 9.665676077265973e-07,
      "loss": 4.1364,
      "step": 2400
    },
    {
      "epoch": 0.651448261010014,
      "learning_rate": 9.662636768877483e-07,
      "loss": 3.9431,
      "step": 2420
    },
    {
      "epoch": 0.6568321309357166,
      "learning_rate": 9.65959746048899e-07,
      "loss": 4.0679,
      "step": 2440
    },
    {
      "epoch": 0.6622160008614192,
      "learning_rate": 9.6565581521005e-07,
      "loss": 4.0304,
      "step": 2460
    },
    {
      "epoch": 0.6675998707871218,
      "learning_rate": 9.653518843712008e-07,
      "loss": 4.1289,
      "step": 2480
    },
    {
      "epoch": 0.6729837407128244,
      "learning_rate": 9.650479535323516e-07,
      "loss": 4.1471,
      "step": 2500
    },
    {
      "epoch": 0.678367610638527,
      "learning_rate": 9.647440226935026e-07,
      "loss": 4.0465,
      "step": 2520
    },
    {
      "epoch": 0.6837514805642295,
      "learning_rate": 9.644400918546536e-07,
      "loss": 4.0011,
      "step": 2540
    },
    {
      "epoch": 0.6891353504899321,
      "learning_rate": 9.641361610158044e-07,
      "loss": 4.0198,
      "step": 2560
    },
    {
      "epoch": 0.6945192204156347,
      "learning_rate": 9.638322301769551e-07,
      "loss": 4.1341,
      "step": 2580
    },
    {
      "epoch": 0.6999030903413374,
      "learning_rate": 9.635282993381061e-07,
      "loss": 3.9473,
      "step": 2600
    },
    {
      "epoch": 0.70528696026704,
      "learning_rate": 9.632243684992571e-07,
      "loss": 3.9478,
      "step": 2620
    },
    {
      "epoch": 0.7106708301927426,
      "learning_rate": 9.62920437660408e-07,
      "loss": 4.1796,
      "step": 2640
    },
    {
      "epoch": 0.7160547001184452,
      "learning_rate": 9.626165068215587e-07,
      "loss": 4.0625,
      "step": 2660
    },
    {
      "epoch": 0.7214385700441477,
      "learning_rate": 9.623125759827097e-07,
      "loss": 3.9337,
      "step": 2680
    },
    {
      "epoch": 0.7268224399698503,
      "learning_rate": 9.620086451438607e-07,
      "loss": 4.0254,
      "step": 2700
    },
    {
      "epoch": 0.7322063098955529,
      "learning_rate": 9.617047143050115e-07,
      "loss": 4.0872,
      "step": 2720
    },
    {
      "epoch": 0.7375901798212555,
      "learning_rate": 9.614007834661623e-07,
      "loss": 4.0988,
      "step": 2740
    },
    {
      "epoch": 0.7429740497469581,
      "learning_rate": 9.610968526273132e-07,
      "loss": 3.9557,
      "step": 2760
    },
    {
      "epoch": 0.7483579196726607,
      "learning_rate": 9.60792921788464e-07,
      "loss": 4.2914,
      "step": 2780
    },
    {
      "epoch": 0.7537417895983634,
      "learning_rate": 9.60488990949615e-07,
      "loss": 4.1434,
      "step": 2800
    },
    {
      "epoch": 0.7591256595240659,
      "learning_rate": 9.601850601107658e-07,
      "loss": 3.8402,
      "step": 2820
    },
    {
      "epoch": 0.7645095294497685,
      "learning_rate": 9.598811292719168e-07,
      "loss": 3.9003,
      "step": 2840
    },
    {
      "epoch": 0.7698933993754711,
      "learning_rate": 9.595771984330676e-07,
      "loss": 3.8761,
      "step": 2860
    },
    {
      "epoch": 0.7752772693011737,
      "learning_rate": 9.592732675942186e-07,
      "loss": 3.9404,
      "step": 2880
    },
    {
      "epoch": 0.7806611392268763,
      "learning_rate": 9.589693367553694e-07,
      "loss": 3.859,
      "step": 2900
    },
    {
      "epoch": 0.7860450091525789,
      "learning_rate": 9.586654059165201e-07,
      "loss": 3.9817,
      "step": 2920
    },
    {
      "epoch": 0.7914288790782815,
      "learning_rate": 9.583614750776711e-07,
      "loss": 4.0039,
      "step": 2940
    },
    {
      "epoch": 0.796812749003984,
      "learning_rate": 9.580575442388221e-07,
      "loss": 4.0198,
      "step": 2960
    },
    {
      "epoch": 0.8021966189296866,
      "learning_rate": 9.57753613399973e-07,
      "loss": 3.945,
      "step": 2980
    },
    {
      "epoch": 0.8075804888553892,
      "learning_rate": 9.574496825611237e-07,
      "loss": 3.9635,
      "step": 3000
    },
    {
      "epoch": 0.8129643587810919,
      "learning_rate": 9.571457517222747e-07,
      "loss": 3.8891,
      "step": 3020
    },
    {
      "epoch": 0.8183482287067945,
      "learning_rate": 9.568418208834257e-07,
      "loss": 3.9844,
      "step": 3040
    },
    {
      "epoch": 0.8237320986324971,
      "learning_rate": 9.565378900445765e-07,
      "loss": 4.0569,
      "step": 3060
    },
    {
      "epoch": 0.8291159685581996,
      "learning_rate": 9.562339592057273e-07,
      "loss": 3.8665,
      "step": 3080
    },
    {
      "epoch": 0.8344998384839022,
      "learning_rate": 9.559300283668782e-07,
      "loss": 3.9799,
      "step": 3100
    },
    {
      "epoch": 0.8398837084096048,
      "learning_rate": 9.556260975280292e-07,
      "loss": 4.0793,
      "step": 3120
    },
    {
      "epoch": 0.8452675783353074,
      "learning_rate": 9.5532216668918e-07,
      "loss": 3.941,
      "step": 3140
    },
    {
      "epoch": 0.85065144826101,
      "learning_rate": 9.550182358503308e-07,
      "loss": 3.9981,
      "step": 3160
    },
    {
      "epoch": 0.8560353181867126,
      "learning_rate": 9.547143050114818e-07,
      "loss": 3.9694,
      "step": 3180
    },
    {
      "epoch": 0.8614191881124152,
      "learning_rate": 9.544103741726328e-07,
      "loss": 3.9918,
      "step": 3200
    },
    {
      "epoch": 0.8668030580381177,
      "learning_rate": 9.541064433337836e-07,
      "loss": 3.8542,
      "step": 3220
    },
    {
      "epoch": 0.8721869279638204,
      "learning_rate": 9.538025124949344e-07,
      "loss": 3.7751,
      "step": 3240
    },
    {
      "epoch": 0.877570797889523,
      "learning_rate": 9.534985816560854e-07,
      "loss": 3.9786,
      "step": 3260
    },
    {
      "epoch": 0.8829546678152256,
      "learning_rate": 9.531946508172361e-07,
      "loss": 3.9525,
      "step": 3280
    },
    {
      "epoch": 0.8883385377409282,
      "learning_rate": 9.528907199783871e-07,
      "loss": 3.8379,
      "step": 3300
    },
    {
      "epoch": 0.8937224076666308,
      "learning_rate": 9.52586789139538e-07,
      "loss": 3.9449,
      "step": 3320
    },
    {
      "epoch": 0.8991062775923334,
      "learning_rate": 9.522828583006888e-07,
      "loss": 3.891,
      "step": 3340
    },
    {
      "epoch": 0.9044901475180359,
      "learning_rate": 9.519789274618397e-07,
      "loss": 3.8922,
      "step": 3360
    },
    {
      "epoch": 0.9098740174437385,
      "learning_rate": 9.516749966229907e-07,
      "loss": 3.9657,
      "step": 3380
    },
    {
      "epoch": 0.9152578873694411,
      "learning_rate": 9.513710657841416e-07,
      "loss": 3.9795,
      "step": 3400
    },
    {
      "epoch": 0.9206417572951437,
      "learning_rate": 9.510671349452924e-07,
      "loss": 3.9513,
      "step": 3420
    },
    {
      "epoch": 0.9260256272208464,
      "learning_rate": 9.507632041064432e-07,
      "loss": 4.0202,
      "step": 3440
    },
    {
      "epoch": 0.931409497146549,
      "learning_rate": 9.504592732675942e-07,
      "loss": 3.9291,
      "step": 3460
    },
    {
      "epoch": 0.9367933670722516,
      "learning_rate": 9.501553424287451e-07,
      "loss": 4.1216,
      "step": 3480
    },
    {
      "epoch": 0.9421772369979541,
      "learning_rate": 9.498514115898959e-07,
      "loss": 3.9733,
      "step": 3500
    },
    {
      "epoch": 0.9475611069236567,
      "learning_rate": 9.495474807510468e-07,
      "loss": 3.8097,
      "step": 3520
    },
    {
      "epoch": 0.9529449768493593,
      "learning_rate": 9.492435499121978e-07,
      "loss": 3.8426,
      "step": 3540
    },
    {
      "epoch": 0.9583288467750619,
      "learning_rate": 9.489396190733485e-07,
      "loss": 3.9829,
      "step": 3560
    },
    {
      "epoch": 0.9637127167007645,
      "learning_rate": 9.486356882344995e-07,
      "loss": 3.646,
      "step": 3580
    },
    {
      "epoch": 0.9690965866264671,
      "learning_rate": 9.483317573956504e-07,
      "loss": 3.7726,
      "step": 3600
    },
    {
      "epoch": 0.9744804565521697,
      "learning_rate": 9.480278265568012e-07,
      "loss": 3.7979,
      "step": 3620
    },
    {
      "epoch": 0.9798643264778722,
      "learning_rate": 9.477238957179521e-07,
      "loss": 3.8609,
      "step": 3640
    },
    {
      "epoch": 0.9852481964035749,
      "learning_rate": 9.47419964879103e-07,
      "loss": 3.7803,
      "step": 3660
    },
    {
      "epoch": 0.9906320663292775,
      "learning_rate": 9.471160340402539e-07,
      "loss": 3.9751,
      "step": 3680
    },
    {
      "epoch": 0.9960159362549801,
      "learning_rate": 9.468121032014048e-07,
      "loss": 3.7751,
      "step": 3700
    }
  ],
  "logging_steps": 20,
  "max_steps": 59424,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 16,
  "save_steps": 10000.0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.454634472726528e+16,
  "train_batch_size": 5,
  "trial_name": null,
  "trial_params": null
}