{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 50,
  "global_step": 11436,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 0.24680212140083313,
      "learning_rate": 4.979447262550289e-05,
      "loss": 2.1164,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.1798873394727707,
      "learning_rate": 4.957582648242085e-05,
      "loss": 2.108,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.12875229120254517,
      "learning_rate": 4.9357180339338815e-05,
      "loss": 2.1164,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.13966438174247742,
      "learning_rate": 4.913853419625678e-05,
      "loss": 1.975,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.14950478076934814,
      "learning_rate": 4.891988805317474e-05,
      "loss": 2.0598,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.11292492598295212,
      "learning_rate": 4.8701241910092706e-05,
      "loss": 2.0375,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.15274354815483093,
      "learning_rate": 4.848259576701068e-05,
      "loss": 2.0463,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.13233616948127747,
      "learning_rate": 4.8263949623928634e-05,
      "loss": 2.0382,
      "step": 400
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.11243890225887299,
      "learning_rate": 4.80453034808466e-05,
      "loss": 2.0239,
      "step": 450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.12158068269491196,
      "learning_rate": 4.782665733776456e-05,
      "loss": 2.0376,
      "step": 500
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.14348222315311432,
      "learning_rate": 4.7608011194682526e-05,
      "loss": 2.0658,
      "step": 550
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.1698862612247467,
      "learning_rate": 4.738936505160049e-05,
      "loss": 2.0811,
      "step": 600
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.18587234616279602,
      "learning_rate": 4.717071890851846e-05,
      "loss": 2.0649,
      "step": 650
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.1404772698879242,
      "learning_rate": 4.6952072765436424e-05,
      "loss": 2.0424,
      "step": 700
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.13483217358589172,
      "learning_rate": 4.673342662235438e-05,
      "loss": 2.0892,
      "step": 750
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.12589296698570251,
      "learning_rate": 4.6514780479272345e-05,
      "loss": 1.9907,
      "step": 800
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.11405760794878006,
      "learning_rate": 4.629613433619031e-05,
      "loss": 2.0512,
      "step": 850
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.14347495138645172,
      "learning_rate": 4.607748819310827e-05,
      "loss": 2.0381,
      "step": 900
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.14272941648960114,
      "learning_rate": 4.5858842050026244e-05,
      "loss": 2.0089,
      "step": 950
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.11842367798089981,
      "learning_rate": 4.564019590694421e-05,
      "loss": 2.0551,
      "step": 1000
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.14809106290340424,
      "learning_rate": 4.542154976386217e-05,
      "loss": 2.0635,
      "step": 1050
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.17582620680332184,
      "learning_rate": 4.520290362078013e-05,
      "loss": 2.0037,
      "step": 1100
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.15331187844276428,
      "learning_rate": 4.498425747769809e-05,
      "loss": 2.0176,
      "step": 1150
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.1289476752281189,
      "learning_rate": 4.476561133461606e-05,
      "loss": 1.9955,
      "step": 1200
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.12641307711601257,
      "learning_rate": 4.454696519153402e-05,
      "loss": 2.0665,
      "step": 1250
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.1526792198419571,
      "learning_rate": 4.432831904845199e-05,
      "loss": 2.0381,
      "step": 1300
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.15751318633556366,
      "learning_rate": 4.4109672905369955e-05,
      "loss": 2.1066,
      "step": 1350
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.16713933646678925,
      "learning_rate": 4.389102676228792e-05,
      "loss": 2.0773,
      "step": 1400
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1296485811471939,
      "learning_rate": 4.3672380619205876e-05,
      "loss": 2.0354,
      "step": 1450
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.1575527936220169,
      "learning_rate": 4.345373447612384e-05,
      "loss": 2.057,
      "step": 1500
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.14888359606266022,
      "learning_rate": 4.3235088333041804e-05,
      "loss": 2.0769,
      "step": 1550
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.13250517845153809,
      "learning_rate": 4.3016442189959775e-05,
      "loss": 2.0548,
      "step": 1600
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.11338368058204651,
      "learning_rate": 4.279779604687774e-05,
      "loss": 2.0733,
      "step": 1650
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.13314321637153625,
      "learning_rate": 4.25791499037957e-05,
      "loss": 2.0927,
      "step": 1700
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.13528186082839966,
      "learning_rate": 4.2360503760713666e-05,
      "loss": 2.0358,
      "step": 1750
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.123367078602314,
      "learning_rate": 4.2141857617631624e-05,
      "loss": 2.122,
      "step": 1800
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.1488184630870819,
      "learning_rate": 4.192321147454959e-05,
      "loss": 2.0561,
      "step": 1850
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.16916592419147491,
      "learning_rate": 4.170456533146755e-05,
      "loss": 2.0353,
      "step": 1900
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.14417660236358643,
      "learning_rate": 4.148591918838552e-05,
      "loss": 1.9982,
      "step": 1950
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.12554995715618134,
      "learning_rate": 4.1267273045303486e-05,
      "loss": 2.1305,
      "step": 2000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.1197732612490654,
      "learning_rate": 4.104862690222145e-05,
      "loss": 2.0228,
      "step": 2050
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.11664281785488129,
      "learning_rate": 4.0829980759139414e-05,
      "loss": 2.0536,
      "step": 2100
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1292441338300705,
      "learning_rate": 4.061133461605737e-05,
      "loss": 2.072,
      "step": 2150
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.1317589282989502,
      "learning_rate": 4.0392688472975335e-05,
      "loss": 2.0793,
      "step": 2200
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.14351141452789307,
      "learning_rate": 4.0174042329893305e-05,
      "loss": 2.0527,
      "step": 2250
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.1438673734664917,
      "learning_rate": 3.995539618681127e-05,
      "loss": 2.0586,
      "step": 2300
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.13835753500461578,
      "learning_rate": 3.973675004372923e-05,
      "loss": 2.0294,
      "step": 2350
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.110976442694664,
      "learning_rate": 3.95181039006472e-05,
      "loss": 2.0917,
      "step": 2400
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.11097148805856705,
      "learning_rate": 3.9299457757565154e-05,
      "loss": 2.0185,
      "step": 2450
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.173320472240448,
      "learning_rate": 3.908081161448312e-05,
      "loss": 2.0346,
      "step": 2500
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.10903707146644592,
      "learning_rate": 3.886216547140108e-05,
      "loss": 2.0342,
      "step": 2550
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.1535567194223404,
      "learning_rate": 3.864351932831905e-05,
      "loss": 2.0104,
      "step": 2600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.13500377535820007,
      "learning_rate": 3.842487318523702e-05,
      "loss": 2.0813,
      "step": 2650
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.1488364040851593,
      "learning_rate": 3.820622704215498e-05,
      "loss": 2.0034,
      "step": 2700
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.14241208136081696,
      "learning_rate": 3.7987580899072945e-05,
      "loss": 2.0316,
      "step": 2750
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.12958261370658875,
      "learning_rate": 3.77689347559909e-05,
      "loss": 2.0108,
      "step": 2800
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12320134788751602,
      "learning_rate": 3.7550288612908866e-05,
      "loss": 1.9801,
      "step": 2850
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.12050608545541763,
      "learning_rate": 3.7331642469826836e-05,
      "loss": 2.0882,
      "step": 2900
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1593894511461258,
      "learning_rate": 3.71129963267448e-05,
      "loss": 2.0175,
      "step": 2950
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.13536041975021362,
      "learning_rate": 3.6894350183662764e-05,
      "loss": 2.0235,
      "step": 3000
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.16139154136180878,
      "learning_rate": 3.667570404058073e-05,
      "loss": 2.0379,
      "step": 3050
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.14355123043060303,
      "learning_rate": 3.645705789749869e-05,
      "loss": 2.0232,
      "step": 3100
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1250680685043335,
      "learning_rate": 3.623841175441665e-05,
      "loss": 2.0238,
      "step": 3150
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.16705290973186493,
      "learning_rate": 3.601976561133461e-05,
      "loss": 2.0823,
      "step": 3200
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.1477033495903015,
      "learning_rate": 3.5801119468252584e-05,
      "loss": 2.0459,
      "step": 3250
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.15941354632377625,
      "learning_rate": 3.558247332517055e-05,
      "loss": 2.0443,
      "step": 3300
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1715698093175888,
      "learning_rate": 3.536382718208851e-05,
      "loss": 2.0891,
      "step": 3350
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.29363158345222473,
      "learning_rate": 3.5145181039006475e-05,
      "loss": 2.0438,
      "step": 3400
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1486337035894394,
      "learning_rate": 3.492653489592444e-05,
      "loss": 2.0515,
      "step": 3450
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.12089661508798599,
      "learning_rate": 3.4707888752842396e-05,
      "loss": 2.0669,
      "step": 3500
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3512625992298126,
      "learning_rate": 3.448924260976037e-05,
      "loss": 2.0875,
      "step": 3550
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.15912899374961853,
      "learning_rate": 3.427059646667833e-05,
      "loss": 2.0583,
      "step": 3600
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.14142446219921112,
      "learning_rate": 3.4051950323596295e-05,
      "loss": 2.0666,
      "step": 3650
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.13084927201271057,
      "learning_rate": 3.383330418051426e-05,
      "loss": 2.1078,
      "step": 3700
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1620977371931076,
      "learning_rate": 3.361465803743222e-05,
      "loss": 2.0184,
      "step": 3750
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18495036661624908,
      "learning_rate": 3.3396011894350187e-05,
      "loss": 2.0117,
      "step": 3800
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.14189182221889496,
      "learning_rate": 3.3177365751268144e-05,
      "loss": 2.0641,
      "step": 3850
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.16044092178344727,
      "learning_rate": 3.2958719608186114e-05,
      "loss": 2.0496,
      "step": 3900
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12184160947799683,
      "learning_rate": 3.274007346510408e-05,
      "loss": 2.0343,
      "step": 3950
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.10508805513381958,
      "learning_rate": 3.252142732202204e-05,
      "loss": 1.9982,
      "step": 4000
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.14695701003074646,
      "learning_rate": 3.2302781178940006e-05,
      "loss": 2.0163,
      "step": 4050
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2915743291378021,
      "learning_rate": 3.208413503585797e-05,
      "loss": 2.0559,
      "step": 4100
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.16513386368751526,
      "learning_rate": 3.1865488892775934e-05,
      "loss": 2.0771,
      "step": 4150
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.11002013087272644,
      "learning_rate": 3.16468427496939e-05,
      "loss": 2.0516,
      "step": 4200
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1383182853460312,
      "learning_rate": 3.142819660661186e-05,
      "loss": 2.0783,
      "step": 4250
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.12230634689331055,
      "learning_rate": 3.1209550463529826e-05,
      "loss": 2.0521,
      "step": 4300
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.15650221705436707,
      "learning_rate": 3.099090432044779e-05,
      "loss": 2.0358,
      "step": 4350
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.16002853214740753,
      "learning_rate": 3.0772258177365753e-05,
      "loss": 2.0217,
      "step": 4400
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.13085784018039703,
      "learning_rate": 3.055361203428372e-05,
      "loss": 2.0019,
      "step": 4450
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.11252173781394958,
      "learning_rate": 3.0334965891201685e-05,
      "loss": 2.0142,
      "step": 4500
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.13878273963928223,
      "learning_rate": 3.0116319748119642e-05,
      "loss": 2.0507,
      "step": 4550
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.13026192784309387,
      "learning_rate": 2.9897673605037606e-05,
      "loss": 2.0927,
      "step": 4600
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.12136498093605042,
      "learning_rate": 2.9679027461955573e-05,
      "loss": 2.0314,
      "step": 4650
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1311144381761551,
      "learning_rate": 2.9460381318873537e-05,
      "loss": 2.0167,
      "step": 4700
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.13698582351207733,
      "learning_rate": 2.92417351757915e-05,
      "loss": 2.0603,
      "step": 4750
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.13138704001903534,
      "learning_rate": 2.9023089032709465e-05,
      "loss": 2.0659,
      "step": 4800
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.13684587180614471,
      "learning_rate": 2.8804442889627432e-05,
      "loss": 2.0426,
      "step": 4850
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.15447795391082764,
      "learning_rate": 2.858579674654539e-05,
      "loss": 2.0724,
      "step": 4900
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2186385840177536,
      "learning_rate": 2.8367150603463356e-05,
      "loss": 2.0357,
      "step": 4950
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.12155057489871979,
      "learning_rate": 2.814850446038132e-05,
      "loss": 2.062,
      "step": 5000
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28663524985313416,
      "learning_rate": 2.7929858317299284e-05,
      "loss": 1.9945,
      "step": 5050
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.16327697038650513,
      "learning_rate": 2.7711212174217248e-05,
      "loss": 2.0262,
      "step": 5100
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.11811967194080353,
      "learning_rate": 2.7492566031135215e-05,
      "loss": 1.9902,
      "step": 5150
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.14936485886573792,
      "learning_rate": 2.727391988805318e-05,
      "loss": 1.9835,
      "step": 5200
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.12473815679550171,
      "learning_rate": 2.7055273744971136e-05,
      "loss": 2.0101,
      "step": 5250
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.12709933519363403,
      "learning_rate": 2.6836627601889104e-05,
      "loss": 2.0394,
      "step": 5300
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2092856615781784,
      "learning_rate": 2.6617981458807068e-05,
      "loss": 2.0482,
      "step": 5350
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.10988406836986542,
      "learning_rate": 2.639933531572503e-05,
      "loss": 2.0043,
      "step": 5400
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.11755118519067764,
      "learning_rate": 2.6180689172642995e-05,
      "loss": 2.04,
      "step": 5450
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.12184648215770721,
      "learning_rate": 2.5962043029560963e-05,
      "loss": 2.0741,
      "step": 5500
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19833241403102875,
      "learning_rate": 2.574339688647892e-05,
      "loss": 2.0551,
      "step": 5550
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.1213446855545044,
      "learning_rate": 2.5524750743396887e-05,
      "loss": 2.0349,
      "step": 5600
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.10714009404182434,
      "learning_rate": 2.530610460031485e-05,
      "loss": 1.9495,
      "step": 5650
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.11714328825473785,
      "learning_rate": 2.5087458457232815e-05,
      "loss": 2.0051,
      "step": 5700
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.14028282463550568,
      "learning_rate": 2.486881231415078e-05,
      "loss": 2.0059,
      "step": 5750
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.11705906689167023,
      "learning_rate": 2.4650166171068746e-05,
      "loss": 2.0374,
      "step": 5800
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.15596671402454376,
      "learning_rate": 2.4431520027986707e-05,
      "loss": 2.0238,
      "step": 5850
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.13032393157482147,
      "learning_rate": 2.421287388490467e-05,
      "loss": 2.0565,
      "step": 5900
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.14617374539375305,
      "learning_rate": 2.3994227741822638e-05,
      "loss": 2.0515,
      "step": 5950
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.156155064702034,
      "learning_rate": 2.37755815987406e-05,
      "loss": 2.0327,
      "step": 6000
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.11759106069803238,
      "learning_rate": 2.3556935455658562e-05,
      "loss": 2.0721,
      "step": 6050
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1545974314212799,
      "learning_rate": 2.3338289312576526e-05,
      "loss": 1.9866,
      "step": 6100
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.13251113891601562,
      "learning_rate": 2.3119643169494493e-05,
      "loss": 2.0147,
      "step": 6150
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1533696949481964,
      "learning_rate": 2.2900997026412454e-05,
      "loss": 2.0267,
      "step": 6200
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.12238262593746185,
      "learning_rate": 2.2682350883330418e-05,
      "loss": 2.0202,
      "step": 6250
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.13370424509048462,
      "learning_rate": 2.2463704740248385e-05,
      "loss": 2.0542,
      "step": 6300
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.135158509016037,
      "learning_rate": 2.2245058597166346e-05,
      "loss": 2.0228,
      "step": 6350
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1633879691362381,
      "learning_rate": 2.202641245408431e-05,
      "loss": 2.014,
      "step": 6400
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.14239120483398438,
      "learning_rate": 2.1807766311002277e-05,
      "loss": 2.1116,
      "step": 6450
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.15084555745124817,
      "learning_rate": 2.1589120167920237e-05,
      "loss": 2.0803,
      "step": 6500
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.12641365826129913,
      "learning_rate": 2.13704740248382e-05,
      "loss": 2.0501,
      "step": 6550
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18917310237884521,
      "learning_rate": 2.115182788175617e-05,
      "loss": 2.0069,
      "step": 6600
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1685107797384262,
      "learning_rate": 2.0933181738674133e-05,
      "loss": 2.0559,
      "step": 6650
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1830686628818512,
      "learning_rate": 2.0714535595592093e-05,
      "loss": 2.0366,
      "step": 6700
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.11661963164806366,
      "learning_rate": 2.0495889452510057e-05,
      "loss": 2.0451,
      "step": 6750
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.15775232017040253,
      "learning_rate": 2.0277243309428024e-05,
      "loss": 2.0785,
      "step": 6800
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1487419754266739,
      "learning_rate": 2.0058597166345985e-05,
      "loss": 2.0231,
      "step": 6850
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.11676029115915298,
      "learning_rate": 1.983995102326395e-05,
      "loss": 2.0444,
      "step": 6900
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.16297248005867004,
      "learning_rate": 1.9621304880181916e-05,
      "loss": 2.0288,
      "step": 6950
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.10936591029167175,
      "learning_rate": 1.940265873709988e-05,
      "loss": 2.0973,
      "step": 7000
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.13290850818157196,
      "learning_rate": 1.918401259401784e-05,
      "loss": 2.0536,
      "step": 7050
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1514565795660019,
      "learning_rate": 1.8965366450935808e-05,
      "loss": 1.9953,
      "step": 7100
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.14755938947200775,
      "learning_rate": 1.874672030785377e-05,
      "loss": 2.0123,
      "step": 7150
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.14362554252147675,
      "learning_rate": 1.8528074164771732e-05,
      "loss": 2.0726,
      "step": 7200
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.13935257494449615,
      "learning_rate": 1.83094280216897e-05,
      "loss": 2.0586,
      "step": 7250
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17782963812351227,
      "learning_rate": 1.8090781878607663e-05,
      "loss": 2.0021,
      "step": 7300
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.11529221385717392,
      "learning_rate": 1.7872135735525627e-05,
      "loss": 2.0405,
      "step": 7350
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20195411145687103,
      "learning_rate": 1.765348959244359e-05,
      "loss": 2.0568,
      "step": 7400
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1562044322490692,
      "learning_rate": 1.7434843449361555e-05,
      "loss": 2.0669,
      "step": 7450
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1276342272758484,
      "learning_rate": 1.721619730627952e-05,
      "loss": 2.0323,
      "step": 7500
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12369738519191742,
      "learning_rate": 1.699755116319748e-05,
      "loss": 2.0451,
      "step": 7550
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.12338880449533463,
      "learning_rate": 1.6778905020115447e-05,
      "loss": 2.0672,
      "step": 7600
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1343850940465927,
      "learning_rate": 1.656025887703341e-05,
      "loss": 2.0319,
      "step": 7650
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.11484856903553009,
      "learning_rate": 1.6341612733951375e-05,
      "loss": 2.0128,
      "step": 7700
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1755642592906952,
      "learning_rate": 1.612296659086934e-05,
      "loss": 2.0056,
      "step": 7750
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1479136049747467,
      "learning_rate": 1.5904320447787302e-05,
      "loss": 2.0359,
      "step": 7800
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.14136551320552826,
      "learning_rate": 1.5685674304705266e-05,
      "loss": 2.0306,
      "step": 7850
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24016603827476501,
      "learning_rate": 1.546702816162323e-05,
      "loss": 2.0095,
      "step": 7900
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.1085515171289444,
      "learning_rate": 1.5248382018541194e-05,
      "loss": 2.0747,
      "step": 7950
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.10604669898748398,
      "learning_rate": 1.5029735875459158e-05,
      "loss": 1.9993,
      "step": 8000
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.12853656709194183,
      "learning_rate": 1.481108973237712e-05,
      "loss": 2.08,
      "step": 8050
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.13124075531959534,
      "learning_rate": 1.4592443589295086e-05,
      "loss": 2.0017,
      "step": 8100
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.12069742381572723,
      "learning_rate": 1.437379744621305e-05,
      "loss": 2.0255,
      "step": 8150
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.17367246747016907,
      "learning_rate": 1.4155151303131015e-05,
      "loss": 1.977,
      "step": 8200
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.17460305988788605,
      "learning_rate": 1.3936505160048976e-05,
      "loss": 2.0414,
      "step": 8250
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.11902675032615662,
      "learning_rate": 1.3717859016966941e-05,
      "loss": 2.0401,
      "step": 8300
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1274511069059372,
      "learning_rate": 1.3499212873884905e-05,
      "loss": 2.0513,
      "step": 8350
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.12478269636631012,
      "learning_rate": 1.3280566730802868e-05,
      "loss": 2.0096,
      "step": 8400
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.12884369492530823,
      "learning_rate": 1.3061920587720833e-05,
      "loss": 2.0188,
      "step": 8450
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.13595916330814362,
      "learning_rate": 1.2843274444638797e-05,
      "loss": 2.0259,
      "step": 8500
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.17813335359096527,
      "learning_rate": 1.2624628301556763e-05,
      "loss": 2.0578,
      "step": 8550
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1450045257806778,
      "learning_rate": 1.2405982158474727e-05,
      "loss": 2.0138,
      "step": 8600
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1711457371711731,
      "learning_rate": 1.2187336015392689e-05,
      "loss": 1.9998,
      "step": 8650
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.13330256938934326,
      "learning_rate": 1.1968689872310653e-05,
      "loss": 1.9919,
      "step": 8700
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.1473345309495926,
      "learning_rate": 1.1750043729228617e-05,
      "loss": 2.071,
      "step": 8750
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.16961540281772614,
      "learning_rate": 1.153139758614658e-05,
      "loss": 2.0439,
      "step": 8800
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.14884643256664276,
      "learning_rate": 1.1312751443064546e-05,
      "loss": 2.066,
      "step": 8850
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17238876223564148,
      "learning_rate": 1.1094105299982508e-05,
      "loss": 1.9937,
      "step": 8900
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.18265317380428314,
      "learning_rate": 1.0875459156900472e-05,
      "loss": 2.0032,
      "step": 8950
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.14806517958641052,
      "learning_rate": 1.0656813013818436e-05,
      "loss": 2.0888,
      "step": 9000
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.12588676810264587,
      "learning_rate": 1.04381668707364e-05,
      "loss": 2.0247,
      "step": 9050
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1755397915840149,
      "learning_rate": 1.0219520727654366e-05,
      "loss": 2.0018,
      "step": 9100
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.15915009379386902,
      "learning_rate": 1.0000874584572328e-05,
      "loss": 2.0219,
      "step": 9150
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.15032340586185455,
      "learning_rate": 9.782228441490293e-06,
      "loss": 2.0401,
      "step": 9200
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.1431685835123062,
      "learning_rate": 9.563582298408257e-06,
      "loss": 2.1078,
      "step": 9250
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16817353665828705,
      "learning_rate": 9.34493615532622e-06,
      "loss": 2.0233,
      "step": 9300
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.11254491657018661,
      "learning_rate": 9.126290012244185e-06,
      "loss": 2.0385,
      "step": 9350
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1725805401802063,
      "learning_rate": 8.907643869162147e-06,
      "loss": 2.0767,
      "step": 9400
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.15062490105628967,
      "learning_rate": 8.688997726080113e-06,
      "loss": 2.0677,
      "step": 9450
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.09490929543972015,
      "learning_rate": 8.470351582998077e-06,
      "loss": 2.059,
      "step": 9500
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.13435165584087372,
      "learning_rate": 8.25170543991604e-06,
      "loss": 2.0475,
      "step": 9550
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.17339274287223816,
      "learning_rate": 8.033059296834005e-06,
      "loss": 2.0018,
      "step": 9600
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.18383900821208954,
      "learning_rate": 7.814413153751967e-06,
      "loss": 2.0207,
      "step": 9650
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.15471112728118896,
      "learning_rate": 7.5957670106699325e-06,
      "loss": 2.1015,
      "step": 9700
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.13521017134189606,
      "learning_rate": 7.3771208675878956e-06,
      "loss": 2.0512,
      "step": 9750
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.15841689705848694,
      "learning_rate": 7.15847472450586e-06,
      "loss": 2.0085,
      "step": 9800
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.16433627903461456,
      "learning_rate": 6.939828581423824e-06,
      "loss": 2.0714,
      "step": 9850
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23176471889019012,
      "learning_rate": 6.721182438341787e-06,
      "loss": 2.0268,
      "step": 9900
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.16546279191970825,
      "learning_rate": 6.502536295259752e-06,
      "loss": 2.0405,
      "step": 9950
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.1396203637123108,
      "learning_rate": 6.283890152177715e-06,
      "loss": 2.0846,
      "step": 10000
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.13313227891921997,
      "learning_rate": 6.06524400909568e-06,
      "loss": 2.0844,
      "step": 10050
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.14923414587974548,
      "learning_rate": 5.846597866013644e-06,
      "loss": 2.0501,
      "step": 10100
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.12807855010032654,
      "learning_rate": 5.627951722931608e-06,
      "loss": 2.0571,
      "step": 10150
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.15558409690856934,
      "learning_rate": 5.4093055798495716e-06,
      "loss": 2.0353,
      "step": 10200
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1312248706817627,
      "learning_rate": 5.1906594367675355e-06,
      "loss": 2.0441,
      "step": 10250
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.15154685080051422,
      "learning_rate": 4.9720132936855e-06,
      "loss": 2.0675,
      "step": 10300
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1447318196296692,
      "learning_rate": 4.753367150603463e-06,
      "loss": 1.9998,
      "step": 10350
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1512334793806076,
      "learning_rate": 4.534721007521427e-06,
      "loss": 2.0112,
      "step": 10400
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.13242949545383453,
      "learning_rate": 4.316074864439391e-06,
      "loss": 2.0191,
      "step": 10450
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.13372650742530823,
      "learning_rate": 4.097428721357356e-06,
      "loss": 2.0326,
      "step": 10500
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1071472018957138,
      "learning_rate": 3.87878257827532e-06,
      "loss": 2.0598,
      "step": 10550
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.14204634726047516,
      "learning_rate": 3.6601364351932836e-06,
      "loss": 2.0241,
      "step": 10600
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.11200432479381561,
      "learning_rate": 3.441490292111247e-06,
      "loss": 1.9981,
      "step": 10650
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17016327381134033,
      "learning_rate": 3.222844149029211e-06,
      "loss": 1.9943,
      "step": 10700
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.15685972571372986,
      "learning_rate": 3.004198005947175e-06,
      "loss": 2.0516,
      "step": 10750
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1390787661075592,
      "learning_rate": 2.7855518628651393e-06,
      "loss": 2.0148,
      "step": 10800
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1561822146177292,
      "learning_rate": 2.566905719783103e-06,
      "loss": 1.9953,
      "step": 10850
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1486501395702362,
      "learning_rate": 2.348259576701067e-06,
      "loss": 2.0293,
      "step": 10900
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.12501879036426544,
      "learning_rate": 2.129613433619031e-06,
      "loss": 2.0506,
      "step": 10950
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.12802496552467346,
      "learning_rate": 1.9109672905369953e-06,
      "loss": 1.9748,
      "step": 11000
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.11280578374862671,
      "learning_rate": 1.6923211474549588e-06,
      "loss": 2.0521,
      "step": 11050
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.13323938846588135,
      "learning_rate": 1.473675004372923e-06,
      "loss": 2.0329,
      "step": 11100
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1350185126066208,
      "learning_rate": 1.2550288612908868e-06,
      "loss": 2.0562,
      "step": 11150
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.14205139875411987,
      "learning_rate": 1.036382718208851e-06,
      "loss": 2.0528,
      "step": 11200
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.19515833258628845,
      "learning_rate": 8.177365751268147e-07,
      "loss": 1.9489,
      "step": 11250
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.15463581681251526,
      "learning_rate": 5.990904320447787e-07,
      "loss": 2.0119,
      "step": 11300
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.13750217854976654,
      "learning_rate": 3.8044428896274276e-07,
      "loss": 2.0552,
      "step": 11350
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1413380652666092,
      "learning_rate": 1.6179814588070666e-07,
      "loss": 2.0558,
      "step": 11400
    }
  ],
  "logging_steps": 50,
  "max_steps": 11436,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 2.4504387448594235e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}