{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 5238,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0057273768613974796,
      "grad_norm": 69.81166336806189,
      "learning_rate": 5e-06,
      "loss": 0.7165,
      "step": 10
    },
    {
      "epoch": 0.011454753722794959,
      "grad_norm": 1.0213242744191677,
      "learning_rate": 5e-06,
      "loss": 0.6046,
      "step": 20
    },
    {
      "epoch": 0.01718213058419244,
      "grad_norm": 0.9449746812829907,
      "learning_rate": 5e-06,
      "loss": 0.5629,
      "step": 30
    },
    {
      "epoch": 0.022909507445589918,
      "grad_norm": 0.8701146894026842,
      "learning_rate": 5e-06,
      "loss": 0.5346,
      "step": 40
    },
    {
      "epoch": 0.0286368843069874,
      "grad_norm": 0.9523022991510063,
      "learning_rate": 5e-06,
      "loss": 0.5263,
      "step": 50
    },
    {
      "epoch": 0.03436426116838488,
      "grad_norm": 0.8669421913033644,
      "learning_rate": 5e-06,
      "loss": 0.508,
      "step": 60
    },
    {
      "epoch": 0.04009163802978236,
      "grad_norm": 0.6655121535600234,
      "learning_rate": 5e-06,
      "loss": 0.5057,
      "step": 70
    },
    {
      "epoch": 0.045819014891179836,
      "grad_norm": 0.5907864190703314,
      "learning_rate": 5e-06,
      "loss": 0.5001,
      "step": 80
    },
    {
      "epoch": 0.05154639175257732,
      "grad_norm": 0.614544161319113,
      "learning_rate": 5e-06,
      "loss": 0.5063,
      "step": 90
    },
    {
      "epoch": 0.0572737686139748,
      "grad_norm": 0.5397294149815204,
      "learning_rate": 5e-06,
      "loss": 0.4995,
      "step": 100
    },
    {
      "epoch": 0.06300114547537228,
      "grad_norm": 0.5645343469851593,
      "learning_rate": 5e-06,
      "loss": 0.5017,
      "step": 110
    },
    {
      "epoch": 0.06872852233676977,
      "grad_norm": 0.5690627599199714,
      "learning_rate": 5e-06,
      "loss": 0.4896,
      "step": 120
    },
    {
      "epoch": 0.07445589919816724,
      "grad_norm": 0.5699342648205336,
      "learning_rate": 5e-06,
      "loss": 0.4921,
      "step": 130
    },
    {
      "epoch": 0.08018327605956473,
      "grad_norm": 0.5710376129366055,
      "learning_rate": 5e-06,
      "loss": 0.4872,
      "step": 140
    },
    {
      "epoch": 0.0859106529209622,
      "grad_norm": 0.5227080703520162,
      "learning_rate": 5e-06,
      "loss": 0.477,
      "step": 150
    },
    {
      "epoch": 0.09163802978235967,
      "grad_norm": 0.5746214024823622,
      "learning_rate": 5e-06,
      "loss": 0.4923,
      "step": 160
    },
    {
      "epoch": 0.09736540664375716,
      "grad_norm": 0.593169674819385,
      "learning_rate": 5e-06,
      "loss": 0.4889,
      "step": 170
    },
    {
      "epoch": 0.10309278350515463,
      "grad_norm": 0.5370925645974212,
      "learning_rate": 5e-06,
      "loss": 0.4905,
      "step": 180
    },
    {
      "epoch": 0.10882016036655212,
      "grad_norm": 0.5343324536740028,
      "learning_rate": 5e-06,
      "loss": 0.4792,
      "step": 190
    },
    {
      "epoch": 0.1145475372279496,
      "grad_norm": 0.5161651942536712,
      "learning_rate": 5e-06,
      "loss": 0.4748,
      "step": 200
    },
    {
      "epoch": 0.12027491408934708,
      "grad_norm": 0.5441222202912794,
      "learning_rate": 5e-06,
      "loss": 0.4759,
      "step": 210
    },
    {
      "epoch": 0.12600229095074456,
      "grad_norm": 0.5327482809839887,
      "learning_rate": 5e-06,
      "loss": 0.4794,
      "step": 220
    },
    {
      "epoch": 0.13172966781214204,
      "grad_norm": 0.5225815889111504,
      "learning_rate": 5e-06,
      "loss": 0.4766,
      "step": 230
    },
    {
      "epoch": 0.13745704467353953,
      "grad_norm": 0.5070284078877997,
      "learning_rate": 5e-06,
      "loss": 0.4709,
      "step": 240
    },
    {
      "epoch": 0.143184421534937,
      "grad_norm": 0.517606520993401,
      "learning_rate": 5e-06,
      "loss": 0.4712,
      "step": 250
    },
    {
      "epoch": 0.14891179839633448,
      "grad_norm": 0.5320984686529744,
      "learning_rate": 5e-06,
      "loss": 0.4695,
      "step": 260
    },
    {
      "epoch": 0.15463917525773196,
      "grad_norm": 0.529639868170382,
      "learning_rate": 5e-06,
      "loss": 0.4655,
      "step": 270
    },
    {
      "epoch": 0.16036655211912945,
      "grad_norm": 0.575168826972108,
      "learning_rate": 5e-06,
      "loss": 0.4746,
      "step": 280
    },
    {
      "epoch": 0.1660939289805269,
      "grad_norm": 0.5001765588738531,
      "learning_rate": 5e-06,
      "loss": 0.4647,
      "step": 290
    },
    {
      "epoch": 0.1718213058419244,
      "grad_norm": 0.5396012723001057,
      "learning_rate": 5e-06,
      "loss": 0.4659,
      "step": 300
    },
    {
      "epoch": 0.1775486827033219,
      "grad_norm": 0.5095325332822108,
      "learning_rate": 5e-06,
      "loss": 0.4579,
      "step": 310
    },
    {
      "epoch": 0.18327605956471935,
      "grad_norm": 0.5122121536362334,
      "learning_rate": 5e-06,
      "loss": 0.4576,
      "step": 320
    },
    {
      "epoch": 0.18900343642611683,
      "grad_norm": 0.5329548850391604,
      "learning_rate": 5e-06,
      "loss": 0.4629,
      "step": 330
    },
    {
      "epoch": 0.19473081328751432,
      "grad_norm": 0.5042536803872275,
      "learning_rate": 5e-06,
      "loss": 0.4605,
      "step": 340
    },
    {
      "epoch": 0.2004581901489118,
      "grad_norm": 0.5300288779122528,
      "learning_rate": 5e-06,
      "loss": 0.4721,
      "step": 350
    },
    {
      "epoch": 0.20618556701030927,
      "grad_norm": 0.5190574562315794,
      "learning_rate": 5e-06,
      "loss": 0.4609,
      "step": 360
    },
    {
      "epoch": 0.21191294387170675,
      "grad_norm": 0.5241199686918058,
      "learning_rate": 5e-06,
      "loss": 0.4526,
      "step": 370
    },
    {
      "epoch": 0.21764032073310424,
      "grad_norm": 0.5233221069855917,
      "learning_rate": 5e-06,
      "loss": 0.4618,
      "step": 380
    },
    {
      "epoch": 0.22336769759450173,
      "grad_norm": 0.5210038372014036,
      "learning_rate": 5e-06,
      "loss": 0.4539,
      "step": 390
    },
    {
      "epoch": 0.2290950744558992,
      "grad_norm": 0.553052125634231,
      "learning_rate": 5e-06,
      "loss": 0.4655,
      "step": 400
    },
    {
      "epoch": 0.23482245131729668,
      "grad_norm": 0.5402131075813968,
      "learning_rate": 5e-06,
      "loss": 0.4592,
      "step": 410
    },
    {
      "epoch": 0.24054982817869416,
      "grad_norm": 0.547123856227283,
      "learning_rate": 5e-06,
      "loss": 0.4661,
      "step": 420
    },
    {
      "epoch": 0.24627720504009165,
      "grad_norm": 0.5398652887149104,
      "learning_rate": 5e-06,
      "loss": 0.4682,
      "step": 430
    },
    {
      "epoch": 0.2520045819014891,
      "grad_norm": 0.546075973932548,
      "learning_rate": 5e-06,
      "loss": 0.4628,
      "step": 440
    },
    {
      "epoch": 0.25773195876288657,
      "grad_norm": 0.5049452018718911,
      "learning_rate": 5e-06,
      "loss": 0.4608,
      "step": 450
    },
    {
      "epoch": 0.2634593356242841,
      "grad_norm": 0.5290358743639263,
      "learning_rate": 5e-06,
      "loss": 0.4455,
      "step": 460
    },
    {
      "epoch": 0.26918671248568155,
      "grad_norm": 0.5765812894544023,
      "learning_rate": 5e-06,
      "loss": 0.463,
      "step": 470
    },
    {
      "epoch": 0.27491408934707906,
      "grad_norm": 0.5447055932241692,
      "learning_rate": 5e-06,
      "loss": 0.4433,
      "step": 480
    },
    {
      "epoch": 0.2806414662084765,
      "grad_norm": 0.5289823549926729,
      "learning_rate": 5e-06,
      "loss": 0.4431,
      "step": 490
    },
    {
      "epoch": 0.286368843069874,
      "grad_norm": 0.4980084292593178,
      "learning_rate": 5e-06,
      "loss": 0.456,
      "step": 500
    },
    {
      "epoch": 0.2920962199312715,
      "grad_norm": 0.5522447921128621,
      "learning_rate": 5e-06,
      "loss": 0.4652,
      "step": 510
    },
    {
      "epoch": 0.29782359679266895,
      "grad_norm": 0.5296189595956464,
      "learning_rate": 5e-06,
      "loss": 0.4297,
      "step": 520
    },
    {
      "epoch": 0.3035509736540664,
      "grad_norm": 0.5271978881360857,
      "learning_rate": 5e-06,
      "loss": 0.4629,
      "step": 530
    },
    {
      "epoch": 0.30927835051546393,
      "grad_norm": 0.5071706509511955,
      "learning_rate": 5e-06,
      "loss": 0.4533,
      "step": 540
    },
    {
      "epoch": 0.3150057273768614,
      "grad_norm": 0.558135676822065,
      "learning_rate": 5e-06,
      "loss": 0.4513,
      "step": 550
    },
    {
      "epoch": 0.3207331042382589,
      "grad_norm": 0.5784740135003561,
      "learning_rate": 5e-06,
      "loss": 0.4604,
      "step": 560
    },
    {
      "epoch": 0.32646048109965636,
      "grad_norm": 0.5335495547019294,
      "learning_rate": 5e-06,
      "loss": 0.4605,
      "step": 570
    },
    {
      "epoch": 0.3321878579610538,
      "grad_norm": 0.5489863296663943,
      "learning_rate": 5e-06,
      "loss": 0.4506,
      "step": 580
    },
    {
      "epoch": 0.33791523482245134,
      "grad_norm": 0.535282941674749,
      "learning_rate": 5e-06,
      "loss": 0.444,
      "step": 590
    },
    {
      "epoch": 0.3436426116838488,
      "grad_norm": 0.5147273315285874,
      "learning_rate": 5e-06,
      "loss": 0.4516,
      "step": 600
    },
    {
      "epoch": 0.34936998854524626,
      "grad_norm": 0.50077282871092,
      "learning_rate": 5e-06,
      "loss": 0.4581,
      "step": 610
    },
    {
      "epoch": 0.3550973654066438,
      "grad_norm": 0.5415059448469354,
      "learning_rate": 5e-06,
      "loss": 0.4602,
      "step": 620
    },
    {
      "epoch": 0.36082474226804123,
      "grad_norm": 0.5362372324428211,
      "learning_rate": 5e-06,
      "loss": 0.4513,
      "step": 630
    },
    {
      "epoch": 0.3665521191294387,
      "grad_norm": 0.5417339363013141,
      "learning_rate": 5e-06,
      "loss": 0.4536,
      "step": 640
    },
    {
      "epoch": 0.3722794959908362,
      "grad_norm": 0.5368661394039848,
      "learning_rate": 5e-06,
      "loss": 0.4625,
      "step": 650
    },
    {
      "epoch": 0.37800687285223367,
      "grad_norm": 0.48954972250964995,
      "learning_rate": 5e-06,
      "loss": 0.4437,
      "step": 660
    },
    {
      "epoch": 0.3837342497136312,
      "grad_norm": 0.5655597381472477,
      "learning_rate": 5e-06,
      "loss": 0.4559,
      "step": 670
    },
    {
      "epoch": 0.38946162657502864,
      "grad_norm": 0.5042748966559669,
      "learning_rate": 5e-06,
      "loss": 0.4444,
      "step": 680
    },
    {
      "epoch": 0.3951890034364261,
      "grad_norm": 0.502096112493101,
      "learning_rate": 5e-06,
      "loss": 0.4437,
      "step": 690
    },
    {
      "epoch": 0.4009163802978236,
      "grad_norm": 0.5154503882668929,
      "learning_rate": 5e-06,
      "loss": 0.4517,
      "step": 700
    },
    {
      "epoch": 0.4066437571592211,
      "grad_norm": 0.5010738869784289,
      "learning_rate": 5e-06,
      "loss": 0.4443,
      "step": 710
    },
    {
      "epoch": 0.41237113402061853,
      "grad_norm": 0.5262171503893435,
      "learning_rate": 5e-06,
      "loss": 0.4416,
      "step": 720
    },
    {
      "epoch": 0.41809851088201605,
      "grad_norm": 0.5413323628955329,
      "learning_rate": 5e-06,
      "loss": 0.453,
      "step": 730
    },
    {
      "epoch": 0.4238258877434135,
      "grad_norm": 0.5117489213181463,
      "learning_rate": 5e-06,
      "loss": 0.4506,
      "step": 740
    },
    {
      "epoch": 0.42955326460481097,
      "grad_norm": 0.5423772098810782,
      "learning_rate": 5e-06,
      "loss": 0.4394,
      "step": 750
    },
    {
      "epoch": 0.4352806414662085,
      "grad_norm": 0.5142331037751853,
      "learning_rate": 5e-06,
      "loss": 0.4391,
      "step": 760
    },
    {
      "epoch": 0.44100801832760594,
      "grad_norm": 0.5185745909865896,
      "learning_rate": 5e-06,
      "loss": 0.4569,
      "step": 770
    },
    {
      "epoch": 0.44673539518900346,
      "grad_norm": 0.5250033250072343,
      "learning_rate": 5e-06,
      "loss": 0.453,
      "step": 780
    },
    {
      "epoch": 0.4524627720504009,
      "grad_norm": 0.5273208537854123,
      "learning_rate": 5e-06,
      "loss": 0.4532,
      "step": 790
    },
    {
      "epoch": 0.4581901489117984,
      "grad_norm": 0.5219200094883771,
      "learning_rate": 5e-06,
      "loss": 0.4534,
      "step": 800
    },
    {
      "epoch": 0.4639175257731959,
      "grad_norm": 0.5265307700538928,
      "learning_rate": 5e-06,
      "loss": 0.4467,
      "step": 810
    },
    {
      "epoch": 0.46964490263459335,
      "grad_norm": 0.5138527259324056,
      "learning_rate": 5e-06,
      "loss": 0.4427,
      "step": 820
    },
    {
      "epoch": 0.4753722794959908,
      "grad_norm": 0.530301258016885,
      "learning_rate": 5e-06,
      "loss": 0.45,
      "step": 830
    },
    {
      "epoch": 0.48109965635738833,
      "grad_norm": 0.523291306479459,
      "learning_rate": 5e-06,
      "loss": 0.448,
      "step": 840
    },
    {
      "epoch": 0.4868270332187858,
      "grad_norm": 0.5425078145535479,
      "learning_rate": 5e-06,
      "loss": 0.4444,
      "step": 850
    },
    {
      "epoch": 0.4925544100801833,
      "grad_norm": 0.5486979687540795,
      "learning_rate": 5e-06,
      "loss": 0.4506,
      "step": 860
    },
    {
      "epoch": 0.49828178694158076,
      "grad_norm": 0.5073997763272261,
      "learning_rate": 5e-06,
      "loss": 0.4498,
      "step": 870
    },
    {
      "epoch": 0.5040091638029782,
      "grad_norm": 0.5025459977891622,
      "learning_rate": 5e-06,
      "loss": 0.4347,
      "step": 880
    },
    {
      "epoch": 0.5097365406643757,
      "grad_norm": 0.5133648493490751,
      "learning_rate": 5e-06,
      "loss": 0.4428,
      "step": 890
    },
    {
      "epoch": 0.5154639175257731,
      "grad_norm": 0.4885863471136841,
      "learning_rate": 5e-06,
      "loss": 0.4478,
      "step": 900
    },
    {
      "epoch": 0.5211912943871707,
      "grad_norm": 0.5425525752294759,
      "learning_rate": 5e-06,
      "loss": 0.4439,
      "step": 910
    },
    {
      "epoch": 0.5269186712485682,
      "grad_norm": 0.5506692407890452,
      "learning_rate": 5e-06,
      "loss": 0.4491,
      "step": 920
    },
    {
      "epoch": 0.5326460481099656,
      "grad_norm": 0.5478953525966641,
      "learning_rate": 5e-06,
      "loss": 0.4539,
      "step": 930
    },
    {
      "epoch": 0.5383734249713631,
      "grad_norm": 0.4860248929907917,
      "learning_rate": 5e-06,
      "loss": 0.4436,
      "step": 940
    },
    {
      "epoch": 0.5441008018327605,
      "grad_norm": 0.5381987297241178,
      "learning_rate": 5e-06,
      "loss": 0.4378,
      "step": 950
    },
    {
      "epoch": 0.5498281786941581,
      "grad_norm": 0.5164585071291472,
      "learning_rate": 5e-06,
      "loss": 0.438,
      "step": 960
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.5478408716601747,
      "learning_rate": 5e-06,
      "loss": 0.4429,
      "step": 970
    },
    {
      "epoch": 0.561282932416953,
      "grad_norm": 0.5319168815859161,
      "learning_rate": 5e-06,
      "loss": 0.454,
      "step": 980
    },
    {
      "epoch": 0.5670103092783505,
      "grad_norm": 0.5049587251570472,
      "learning_rate": 5e-06,
      "loss": 0.4469,
      "step": 990
    },
    {
      "epoch": 0.572737686139748,
      "grad_norm": 0.5319440597432152,
      "learning_rate": 5e-06,
      "loss": 0.437,
      "step": 1000
    },
    {
      "epoch": 0.5784650630011455,
      "grad_norm": 0.511281381398782,
      "learning_rate": 5e-06,
      "loss": 0.4382,
      "step": 1010
    },
    {
      "epoch": 0.584192439862543,
      "grad_norm": 0.5194710291745682,
      "learning_rate": 5e-06,
      "loss": 0.4559,
      "step": 1020
    },
    {
      "epoch": 0.5899198167239404,
      "grad_norm": 0.5074370643679889,
      "learning_rate": 5e-06,
      "loss": 0.4588,
      "step": 1030
    },
    {
      "epoch": 0.5956471935853379,
      "grad_norm": 0.5191559040020144,
      "learning_rate": 5e-06,
      "loss": 0.4377,
      "step": 1040
    },
    {
      "epoch": 0.6013745704467354,
      "grad_norm": 0.5476079411315063,
      "learning_rate": 5e-06,
      "loss": 0.4449,
      "step": 1050
    },
    {
      "epoch": 0.6071019473081328,
      "grad_norm": 0.5673160064722429,
      "learning_rate": 5e-06,
      "loss": 0.4352,
      "step": 1060
    },
    {
      "epoch": 0.6128293241695304,
      "grad_norm": 0.5065625800440946,
      "learning_rate": 5e-06,
      "loss": 0.441,
      "step": 1070
    },
    {
      "epoch": 0.6185567010309279,
      "grad_norm": 0.5512625856761683,
      "learning_rate": 5e-06,
      "loss": 0.4429,
      "step": 1080
    },
    {
      "epoch": 0.6242840778923253,
      "grad_norm": 0.552355840658565,
      "learning_rate": 5e-06,
      "loss": 0.4434,
      "step": 1090
    },
    {
      "epoch": 0.6300114547537228,
      "grad_norm": 0.487614534934624,
      "learning_rate": 5e-06,
      "loss": 0.4292,
      "step": 1100
    },
    {
      "epoch": 0.6357388316151202,
      "grad_norm": 0.512636901651117,
      "learning_rate": 5e-06,
      "loss": 0.4441,
      "step": 1110
    },
    {
      "epoch": 0.6414662084765178,
      "grad_norm": 0.5128771226016339,
      "learning_rate": 5e-06,
      "loss": 0.4386,
      "step": 1120
    },
    {
      "epoch": 0.6471935853379153,
      "grad_norm": 0.501226424633001,
      "learning_rate": 5e-06,
      "loss": 0.4277,
      "step": 1130
    },
    {
      "epoch": 0.6529209621993127,
      "grad_norm": 0.5370012590547919,
      "learning_rate": 5e-06,
      "loss": 0.4444,
      "step": 1140
    },
    {
      "epoch": 0.6586483390607102,
      "grad_norm": 0.5136199788898852,
      "learning_rate": 5e-06,
      "loss": 0.4441,
      "step": 1150
    },
    {
      "epoch": 0.6643757159221076,
      "grad_norm": 0.4964737141886483,
      "learning_rate": 5e-06,
      "loss": 0.4488,
      "step": 1160
    },
    {
      "epoch": 0.6701030927835051,
      "grad_norm": 0.5516171001785463,
      "learning_rate": 5e-06,
      "loss": 0.4425,
      "step": 1170
    },
    {
      "epoch": 0.6758304696449027,
      "grad_norm": 0.5495702116230217,
      "learning_rate": 5e-06,
      "loss": 0.4408,
      "step": 1180
    },
    {
      "epoch": 0.6815578465063001,
      "grad_norm": 0.5459174578575696,
      "learning_rate": 5e-06,
      "loss": 0.4407,
      "step": 1190
    },
    {
      "epoch": 0.6872852233676976,
      "grad_norm": 0.5345898636386911,
      "learning_rate": 5e-06,
      "loss": 0.4537,
      "step": 1200
    },
    {
      "epoch": 0.693012600229095,
      "grad_norm": 0.5243198575640295,
      "learning_rate": 5e-06,
      "loss": 0.4406,
      "step": 1210
    },
    {
      "epoch": 0.6987399770904925,
      "grad_norm": 0.5129227421288873,
      "learning_rate": 5e-06,
      "loss": 0.4317,
      "step": 1220
    },
    {
      "epoch": 0.7044673539518901,
      "grad_norm": 0.5260957634672794,
      "learning_rate": 5e-06,
      "loss": 0.4273,
      "step": 1230
    },
    {
      "epoch": 0.7101947308132875,
      "grad_norm": 0.5437290419224033,
      "learning_rate": 5e-06,
      "loss": 0.4259,
      "step": 1240
    },
    {
      "epoch": 0.715922107674685,
      "grad_norm": 0.5347492902940403,
      "learning_rate": 5e-06,
      "loss": 0.4431,
      "step": 1250
    },
    {
      "epoch": 0.7216494845360825,
      "grad_norm": 0.5263261030228884,
      "learning_rate": 5e-06,
      "loss": 0.4345,
      "step": 1260
    },
    {
      "epoch": 0.7273768613974799,
      "grad_norm": 0.5084395646880466,
      "learning_rate": 5e-06,
      "loss": 0.4402,
      "step": 1270
    },
    {
      "epoch": 0.7331042382588774,
      "grad_norm": 0.5069357119145561,
      "learning_rate": 5e-06,
      "loss": 0.4193,
      "step": 1280
    },
    {
      "epoch": 0.738831615120275,
      "grad_norm": 0.5400403941956701,
      "learning_rate": 5e-06,
      "loss": 0.4398,
      "step": 1290
    },
    {
      "epoch": 0.7445589919816724,
      "grad_norm": 0.5233065624416613,
      "learning_rate": 5e-06,
      "loss": 0.4415,
      "step": 1300
    },
    {
      "epoch": 0.7502863688430699,
      "grad_norm": 0.5423286351815177,
      "learning_rate": 5e-06,
      "loss": 0.4455,
      "step": 1310
    },
    {
      "epoch": 0.7560137457044673,
      "grad_norm": 0.5109684016877989,
      "learning_rate": 5e-06,
      "loss": 0.4424,
      "step": 1320
    },
    {
      "epoch": 0.7617411225658648,
      "grad_norm": 0.527654171024067,
      "learning_rate": 5e-06,
      "loss": 0.4379,
      "step": 1330
    },
    {
      "epoch": 0.7674684994272624,
      "grad_norm": 0.4948196366852471,
      "learning_rate": 5e-06,
      "loss": 0.4409,
      "step": 1340
    },
    {
      "epoch": 0.7731958762886598,
      "grad_norm": 0.5572218648877072,
      "learning_rate": 5e-06,
      "loss": 0.442,
      "step": 1350
    },
    {
      "epoch": 0.7789232531500573,
      "grad_norm": 0.4669728936660895,
      "learning_rate": 5e-06,
      "loss": 0.4194,
      "step": 1360
    },
    {
      "epoch": 0.7846506300114547,
      "grad_norm": 0.5428733395212945,
      "learning_rate": 5e-06,
      "loss": 0.425,
      "step": 1370
    },
    {
      "epoch": 0.7903780068728522,
      "grad_norm": 0.4985083493758901,
      "learning_rate": 5e-06,
      "loss": 0.428,
      "step": 1380
    },
    {
      "epoch": 0.7961053837342497,
      "grad_norm": 0.5127694485606286,
      "learning_rate": 5e-06,
      "loss": 0.4318,
      "step": 1390
    },
    {
      "epoch": 0.8018327605956472,
      "grad_norm": 0.5429586838215733,
      "learning_rate": 5e-06,
      "loss": 0.429,
      "step": 1400
    },
    {
      "epoch": 0.8075601374570447,
      "grad_norm": 0.5220082918214546,
      "learning_rate": 5e-06,
      "loss": 0.4359,
      "step": 1410
    },
    {
      "epoch": 0.8132875143184422,
      "grad_norm": 0.4999143755354932,
      "learning_rate": 5e-06,
      "loss": 0.4344,
      "step": 1420
    },
    {
      "epoch": 0.8190148911798396,
      "grad_norm": 0.5066162466504854,
      "learning_rate": 5e-06,
      "loss": 0.4341,
      "step": 1430
    },
    {
      "epoch": 0.8247422680412371,
      "grad_norm": 0.5175621795775882,
      "learning_rate": 5e-06,
      "loss": 0.4383,
      "step": 1440
    },
    {
      "epoch": 0.8304696449026346,
      "grad_norm": 0.46825138868237176,
      "learning_rate": 5e-06,
      "loss": 0.4186,
      "step": 1450
    },
    {
      "epoch": 0.8361970217640321,
      "grad_norm": 0.49668507800036343,
      "learning_rate": 5e-06,
      "loss": 0.4352,
      "step": 1460
    },
    {
      "epoch": 0.8419243986254296,
      "grad_norm": 0.5076029260875138,
      "learning_rate": 5e-06,
      "loss": 0.4418,
      "step": 1470
    },
    {
      "epoch": 0.847651775486827,
      "grad_norm": 0.5255466837310667,
      "learning_rate": 5e-06,
      "loss": 0.4318,
      "step": 1480
    },
    {
      "epoch": 0.8533791523482245,
      "grad_norm": 0.524332739040499,
      "learning_rate": 5e-06,
      "loss": 0.4348,
      "step": 1490
    },
    {
      "epoch": 0.8591065292096219,
      "grad_norm": 0.49048587533812715,
      "learning_rate": 5e-06,
      "loss": 0.4335,
      "step": 1500
    },
    {
      "epoch": 0.8648339060710195,
      "grad_norm": 0.5274633464517493,
      "learning_rate": 5e-06,
      "loss": 0.4302,
      "step": 1510
    },
    {
      "epoch": 0.870561282932417,
      "grad_norm": 0.5216418463132072,
      "learning_rate": 5e-06,
      "loss": 0.429,
      "step": 1520
    },
    {
      "epoch": 0.8762886597938144,
      "grad_norm": 0.51978258763271,
      "learning_rate": 5e-06,
      "loss": 0.4348,
      "step": 1530
    },
    {
      "epoch": 0.8820160366552119,
      "grad_norm": 0.5438623460564584,
      "learning_rate": 5e-06,
      "loss": 0.4454,
      "step": 1540
    },
    {
      "epoch": 0.8877434135166093,
      "grad_norm": 0.4993963585199609,
      "learning_rate": 5e-06,
      "loss": 0.4245,
      "step": 1550
    },
    {
      "epoch": 0.8934707903780069,
      "grad_norm": 0.5430194121475598,
      "learning_rate": 5e-06,
      "loss": 0.4456,
      "step": 1560
    },
    {
      "epoch": 0.8991981672394044,
      "grad_norm": 0.5336061227789548,
      "learning_rate": 5e-06,
      "loss": 0.4225,
      "step": 1570
    },
    {
      "epoch": 0.9049255441008018,
      "grad_norm": 0.4974230293054768,
      "learning_rate": 5e-06,
      "loss": 0.4248,
      "step": 1580
    },
    {
      "epoch": 0.9106529209621993,
      "grad_norm": 0.5128982229920478,
      "learning_rate": 5e-06,
      "loss": 0.4338,
      "step": 1590
    },
    {
      "epoch": 0.9163802978235968,
      "grad_norm": 0.5231256959340072,
      "learning_rate": 5e-06,
      "loss": 0.435,
      "step": 1600
    },
    {
      "epoch": 0.9221076746849943,
      "grad_norm": 0.4918096317536555,
      "learning_rate": 5e-06,
      "loss": 0.4438,
      "step": 1610
    },
    {
      "epoch": 0.9278350515463918,
      "grad_norm": 0.5278974812779116,
      "learning_rate": 5e-06,
      "loss": 0.4313,
      "step": 1620
    },
    {
      "epoch": 0.9335624284077892,
      "grad_norm": 0.4921635557471188,
      "learning_rate": 5e-06,
      "loss": 0.4208,
      "step": 1630
    },
    {
      "epoch": 0.9392898052691867,
      "grad_norm": 0.5251247147790307,
      "learning_rate": 5e-06,
      "loss": 0.4254,
      "step": 1640
    },
    {
      "epoch": 0.9450171821305842,
      "grad_norm": 0.4710846491151496,
      "learning_rate": 5e-06,
      "loss": 0.4357,
      "step": 1650
    },
    {
      "epoch": 0.9507445589919816,
      "grad_norm": 0.5545034160352649,
      "learning_rate": 5e-06,
      "loss": 0.4296,
      "step": 1660
    },
    {
      "epoch": 0.9564719358533792,
      "grad_norm": 0.5200683652411496,
      "learning_rate": 5e-06,
      "loss": 0.432,
      "step": 1670
    },
    {
      "epoch": 0.9621993127147767,
      "grad_norm": 0.5229813016973693,
      "learning_rate": 5e-06,
      "loss": 0.4374,
      "step": 1680
    },
    {
      "epoch": 0.9679266895761741,
      "grad_norm": 0.4819952616802862,
      "learning_rate": 5e-06,
      "loss": 0.4137,
      "step": 1690
    },
    {
      "epoch": 0.9736540664375716,
      "grad_norm": 0.5337445133162833,
      "learning_rate": 5e-06,
      "loss": 0.4347,
      "step": 1700
    },
    {
      "epoch": 0.979381443298969,
      "grad_norm": 0.5207169222218365,
      "learning_rate": 5e-06,
      "loss": 0.4234,
      "step": 1710
    },
    {
      "epoch": 0.9851088201603666,
      "grad_norm": 0.5235455233723544,
      "learning_rate": 5e-06,
      "loss": 0.433,
      "step": 1720
    },
    {
      "epoch": 0.9908361970217641,
      "grad_norm": 0.5297773672827779,
      "learning_rate": 5e-06,
      "loss": 0.4322,
      "step": 1730
    },
    {
      "epoch": 0.9965635738831615,
      "grad_norm": 0.5100360279338871,
      "learning_rate": 5e-06,
      "loss": 0.429,
      "step": 1740
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.4306947588920593,
      "eval_runtime": 295.9271,
      "eval_samples_per_second": 39.746,
      "eval_steps_per_second": 0.622,
      "step": 1746
    },
    {
      "epoch": 1.002290950744559,
      "grad_norm": 0.5164564401791184,
      "learning_rate": 5e-06,
      "loss": 0.4117,
      "step": 1750
    },
    {
      "epoch": 1.0080183276059564,
      "grad_norm": 0.5373967670552326,
      "learning_rate": 5e-06,
      "loss": 0.3742,
      "step": 1760
    },
    {
      "epoch": 1.013745704467354,
      "grad_norm": 0.5049268545352126,
      "learning_rate": 5e-06,
      "loss": 0.3779,
      "step": 1770
    },
    {
      "epoch": 1.0194730813287514,
      "grad_norm": 0.5061218550685067,
      "learning_rate": 5e-06,
      "loss": 0.3702,
      "step": 1780
    },
    {
      "epoch": 1.0252004581901488,
      "grad_norm": 0.4986641460518763,
      "learning_rate": 5e-06,
      "loss": 0.3711,
      "step": 1790
    },
    {
      "epoch": 1.0309278350515463,
      "grad_norm": 0.5053371876016091,
      "learning_rate": 5e-06,
      "loss": 0.3663,
      "step": 1800
    },
    {
      "epoch": 1.036655211912944,
      "grad_norm": 0.5638955186469768,
      "learning_rate": 5e-06,
      "loss": 0.3785,
      "step": 1810
    },
    {
      "epoch": 1.0423825887743414,
      "grad_norm": 0.5203865221430506,
      "learning_rate": 5e-06,
      "loss": 0.37,
      "step": 1820
    },
    {
      "epoch": 1.0481099656357389,
      "grad_norm": 0.5084938620530141,
      "learning_rate": 5e-06,
      "loss": 0.3773,
      "step": 1830
    },
    {
      "epoch": 1.0538373424971363,
      "grad_norm": 0.49981188453102293,
      "learning_rate": 5e-06,
      "loss": 0.3795,
      "step": 1840
    },
    {
      "epoch": 1.0595647193585338,
      "grad_norm": 0.4982217537194922,
      "learning_rate": 5e-06,
      "loss": 0.368,
      "step": 1850
    },
    {
      "epoch": 1.0652920962199313,
      "grad_norm": 0.5082225744208482,
      "learning_rate": 5e-06,
      "loss": 0.3726,
      "step": 1860
    },
    {
      "epoch": 1.0710194730813287,
      "grad_norm": 0.5263843366712018,
      "learning_rate": 5e-06,
      "loss": 0.3806,
      "step": 1870
    },
    {
      "epoch": 1.0767468499427262,
      "grad_norm": 0.48898589739857184,
      "learning_rate": 5e-06,
      "loss": 0.3719,
      "step": 1880
    },
    {
      "epoch": 1.0824742268041236,
      "grad_norm": 0.5096088003209625,
      "learning_rate": 5e-06,
      "loss": 0.373,
      "step": 1890
    },
    {
      "epoch": 1.088201603665521,
      "grad_norm": 0.48705936220014545,
      "learning_rate": 5e-06,
      "loss": 0.3844,
      "step": 1900
    },
    {
      "epoch": 1.0939289805269188,
      "grad_norm": 0.49298990949849025,
      "learning_rate": 5e-06,
      "loss": 0.3734,
      "step": 1910
    },
    {
      "epoch": 1.0996563573883162,
      "grad_norm": 0.5169227323207669,
      "learning_rate": 5e-06,
      "loss": 0.3794,
      "step": 1920
    },
    {
      "epoch": 1.1053837342497137,
      "grad_norm": 0.502144264943404,
      "learning_rate": 5e-06,
      "loss": 0.3811,
      "step": 1930
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 0.47522199328541975,
      "learning_rate": 5e-06,
      "loss": 0.3673,
      "step": 1940
    },
    {
      "epoch": 1.1168384879725086,
      "grad_norm": 0.49289512210078507,
      "learning_rate": 5e-06,
      "loss": 0.3723,
      "step": 1950
    },
    {
      "epoch": 1.122565864833906,
      "grad_norm": 0.5179394316744345,
      "learning_rate": 5e-06,
      "loss": 0.3716,
      "step": 1960
    },
    {
      "epoch": 1.1282932416953035,
      "grad_norm": 0.4991350954847643,
      "learning_rate": 5e-06,
      "loss": 0.3757,
      "step": 1970
    },
    {
      "epoch": 1.134020618556701,
      "grad_norm": 0.5215898643906023,
      "learning_rate": 5e-06,
      "loss": 0.3811,
      "step": 1980
    },
    {
      "epoch": 1.1397479954180985,
      "grad_norm": 0.5248312587556514,
      "learning_rate": 5e-06,
      "loss": 0.3834,
      "step": 1990
    },
    {
      "epoch": 1.145475372279496,
      "grad_norm": 0.5112191261184691,
      "learning_rate": 5e-06,
      "loss": 0.3731,
      "step": 2000
    },
    {
      "epoch": 1.1512027491408934,
      "grad_norm": 0.49244426705823124,
      "learning_rate": 5e-06,
      "loss": 0.3782,
      "step": 2010
    },
    {
      "epoch": 1.1569301260022908,
      "grad_norm": 0.5524056534176885,
      "learning_rate": 5e-06,
      "loss": 0.3751,
      "step": 2020
    },
    {
      "epoch": 1.1626575028636885,
      "grad_norm": 0.522166484229405,
      "learning_rate": 5e-06,
      "loss": 0.3761,
      "step": 2030
    },
    {
      "epoch": 1.168384879725086,
      "grad_norm": 0.4995069364287706,
      "learning_rate": 5e-06,
      "loss": 0.3741,
      "step": 2040
    },
    {
      "epoch": 1.1741122565864834,
      "grad_norm": 0.510902706338532,
      "learning_rate": 5e-06,
      "loss": 0.3743,
      "step": 2050
    },
    {
      "epoch": 1.179839633447881,
      "grad_norm": 0.5065453487736599,
      "learning_rate": 5e-06,
      "loss": 0.3722,
      "step": 2060
    },
    {
      "epoch": 1.1855670103092784,
      "grad_norm": 0.5151805060650854,
      "learning_rate": 5e-06,
      "loss": 0.375,
      "step": 2070
    },
    {
      "epoch": 1.1912943871706758,
      "grad_norm": 0.49831260628359125,
      "learning_rate": 5e-06,
      "loss": 0.3808,
      "step": 2080
    },
    {
      "epoch": 1.1970217640320733,
      "grad_norm": 0.5121714439577272,
      "learning_rate": 5e-06,
      "loss": 0.3698,
      "step": 2090
    },
    {
      "epoch": 1.2027491408934707,
      "grad_norm": 0.4912963388060686,
      "learning_rate": 5e-06,
      "loss": 0.3699,
      "step": 2100
    },
    {
      "epoch": 1.2084765177548682,
      "grad_norm": 0.48571233935935587,
      "learning_rate": 5e-06,
      "loss": 0.378,
      "step": 2110
    },
    {
      "epoch": 1.2142038946162657,
      "grad_norm": 0.4821158492894926,
      "learning_rate": 5e-06,
      "loss": 0.3742,
      "step": 2120
    },
    {
      "epoch": 1.2199312714776633,
      "grad_norm": 0.4873415658805343,
      "learning_rate": 5e-06,
      "loss": 0.3757,
      "step": 2130
    },
    {
      "epoch": 1.2256586483390608,
      "grad_norm": 0.535296961321234,
      "learning_rate": 5e-06,
      "loss": 0.3803,
      "step": 2140
    },
    {
      "epoch": 1.2313860252004583,
      "grad_norm": 0.5158442516232742,
      "learning_rate": 5e-06,
      "loss": 0.3799,
      "step": 2150
    },
    {
      "epoch": 1.2371134020618557,
      "grad_norm": 0.5188127574002888,
      "learning_rate": 5e-06,
      "loss": 0.3799,
      "step": 2160
    },
    {
      "epoch": 1.2428407789232532,
      "grad_norm": 0.5197089286594105,
      "learning_rate": 5e-06,
      "loss": 0.3807,
      "step": 2170
    },
    {
      "epoch": 1.2485681557846506,
      "grad_norm": 0.503291581955108,
      "learning_rate": 5e-06,
      "loss": 0.376,
      "step": 2180
    },
    {
      "epoch": 1.254295532646048,
      "grad_norm": 0.5077499589009502,
      "learning_rate": 5e-06,
      "loss": 0.3723,
      "step": 2190
    },
    {
      "epoch": 1.2600229095074456,
      "grad_norm": 0.5332500891173507,
      "learning_rate": 5e-06,
      "loss": 0.3789,
      "step": 2200
    },
    {
      "epoch": 1.265750286368843,
      "grad_norm": 0.5060162257976599,
      "learning_rate": 5e-06,
      "loss": 0.3801,
      "step": 2210
    },
    {
      "epoch": 1.2714776632302405,
      "grad_norm": 0.4963592770753574,
      "learning_rate": 5e-06,
      "loss": 0.3714,
      "step": 2220
    },
    {
      "epoch": 1.277205040091638,
      "grad_norm": 0.4813664865465453,
      "learning_rate": 5e-06,
      "loss": 0.373,
      "step": 2230
    },
    {
      "epoch": 1.2829324169530354,
      "grad_norm": 0.5399021996788194,
      "learning_rate": 5e-06,
      "loss": 0.3847,
      "step": 2240
    },
    {
      "epoch": 1.2886597938144329,
      "grad_norm": 0.5213132074977475,
      "learning_rate": 5e-06,
      "loss": 0.3707,
      "step": 2250
    },
    {
      "epoch": 1.2943871706758305,
      "grad_norm": 0.5096095310985154,
      "learning_rate": 5e-06,
      "loss": 0.3735,
      "step": 2260
    },
    {
      "epoch": 1.300114547537228,
      "grad_norm": 0.5046230839401389,
      "learning_rate": 5e-06,
      "loss": 0.3777,
      "step": 2270
    },
    {
      "epoch": 1.3058419243986255,
      "grad_norm": 0.49606318588063886,
      "learning_rate": 5e-06,
      "loss": 0.3802,
      "step": 2280
    },
    {
      "epoch": 1.311569301260023,
      "grad_norm": 0.48618155936697477,
      "learning_rate": 5e-06,
      "loss": 0.3642,
      "step": 2290
    },
    {
      "epoch": 1.3172966781214204,
      "grad_norm": 0.5131608728495509,
      "learning_rate": 5e-06,
      "loss": 0.3752,
      "step": 2300
    },
    {
      "epoch": 1.3230240549828178,
      "grad_norm": 0.5398006873457784,
      "learning_rate": 5e-06,
      "loss": 0.3771,
      "step": 2310
    },
    {
      "epoch": 1.3287514318442153,
      "grad_norm": 0.5147920738496441,
      "learning_rate": 5e-06,
      "loss": 0.3701,
      "step": 2320
    },
    {
      "epoch": 1.3344788087056128,
      "grad_norm": 0.5034049600590188,
      "learning_rate": 5e-06,
      "loss": 0.3661,
      "step": 2330
    },
    {
      "epoch": 1.3402061855670104,
      "grad_norm": 0.49524063841373434,
      "learning_rate": 5e-06,
      "loss": 0.3659,
      "step": 2340
    },
    {
      "epoch": 1.345933562428408,
      "grad_norm": 0.5167698042786277,
      "learning_rate": 5e-06,
      "loss": 0.369,
      "step": 2350
    },
    {
      "epoch": 1.3516609392898054,
      "grad_norm": 0.49269430039048795,
      "learning_rate": 5e-06,
      "loss": 0.3773,
      "step": 2360
    },
    {
      "epoch": 1.3573883161512028,
      "grad_norm": 0.5168560438384152,
      "learning_rate": 5e-06,
      "loss": 0.3743,
      "step": 2370
    },
    {
      "epoch": 1.3631156930126003,
      "grad_norm": 0.4932882563190922,
      "learning_rate": 5e-06,
      "loss": 0.3816,
      "step": 2380
    },
    {
      "epoch": 1.3688430698739977,
      "grad_norm": 0.5282260861884785,
      "learning_rate": 5e-06,
      "loss": 0.3844,
      "step": 2390
    },
    {
      "epoch": 1.3745704467353952,
      "grad_norm": 0.5019717962696827,
      "learning_rate": 5e-06,
      "loss": 0.3672,
      "step": 2400
    },
    {
      "epoch": 1.3802978235967927,
      "grad_norm": 0.4895383626409957,
      "learning_rate": 5e-06,
      "loss": 0.373,
      "step": 2410
    },
    {
      "epoch": 1.38602520045819,
      "grad_norm": 0.4802171696456841,
      "learning_rate": 5e-06,
      "loss": 0.3674,
      "step": 2420
    },
    {
      "epoch": 1.3917525773195876,
      "grad_norm": 0.47543500312582,
      "learning_rate": 5e-06,
      "loss": 0.3725,
      "step": 2430
    },
    {
      "epoch": 1.397479954180985,
      "grad_norm": 0.5058875019724552,
      "learning_rate": 5e-06,
      "loss": 0.3721,
      "step": 2440
    },
    {
      "epoch": 1.4032073310423825,
      "grad_norm": 0.4993814896944319,
      "learning_rate": 5e-06,
      "loss": 0.3779,
      "step": 2450
    },
    {
      "epoch": 1.40893470790378,
      "grad_norm": 0.5109971377219812,
      "learning_rate": 5e-06,
      "loss": 0.3787,
      "step": 2460
    },
    {
      "epoch": 1.4146620847651776,
      "grad_norm": 0.5139034920537475,
      "learning_rate": 5e-06,
      "loss": 0.3728,
      "step": 2470
    },
    {
      "epoch": 1.420389461626575,
      "grad_norm": 0.5093772065662143,
      "learning_rate": 5e-06,
      "loss": 0.3723,
      "step": 2480
    },
    {
      "epoch": 1.4261168384879725,
      "grad_norm": 0.4971098961659199,
      "learning_rate": 5e-06,
      "loss": 0.3707,
      "step": 2490
    },
    {
      "epoch": 1.43184421534937,
      "grad_norm": 0.4955169015948491,
      "learning_rate": 5e-06,
      "loss": 0.3815,
      "step": 2500
    },
    {
      "epoch": 1.4375715922107675,
      "grad_norm": 0.5057004758414246,
      "learning_rate": 5e-06,
      "loss": 0.369,
      "step": 2510
    },
    {
      "epoch": 1.443298969072165,
      "grad_norm": 0.5371969106259902,
      "learning_rate": 5e-06,
      "loss": 0.3725,
      "step": 2520
    },
    {
      "epoch": 1.4490263459335624,
      "grad_norm": 0.5043182854139414,
      "learning_rate": 5e-06,
      "loss": 0.3625,
      "step": 2530
    },
    {
      "epoch": 1.4547537227949598,
      "grad_norm": 0.5039396765453658,
      "learning_rate": 5e-06,
      "loss": 0.3735,
      "step": 2540
    },
    {
      "epoch": 1.4604810996563573,
      "grad_norm": 0.4643231312606545,
      "learning_rate": 5e-06,
      "loss": 0.3714,
      "step": 2550
    },
    {
      "epoch": 1.466208476517755,
      "grad_norm": 0.5122652558271147,
      "learning_rate": 5e-06,
      "loss": 0.3737,
      "step": 2560
    },
    {
      "epoch": 1.4719358533791524,
      "grad_norm": 0.4599727947671919,
      "learning_rate": 5e-06,
      "loss": 0.381,
      "step": 2570
    },
    {
      "epoch": 1.47766323024055,
      "grad_norm": 0.47595731930660123,
      "learning_rate": 5e-06,
      "loss": 0.3671,
      "step": 2580
    },
    {
      "epoch": 1.4833906071019474,
      "grad_norm": 0.49427201740882865,
      "learning_rate": 5e-06,
      "loss": 0.3746,
      "step": 2590
    },
    {
      "epoch": 1.4891179839633448,
      "grad_norm": 0.5160559476566308,
      "learning_rate": 5e-06,
      "loss": 0.3732,
      "step": 2600
    },
    {
      "epoch": 1.4948453608247423,
      "grad_norm": 0.5175281975939867,
      "learning_rate": 5e-06,
      "loss": 0.3757,
      "step": 2610
    },
    {
      "epoch": 1.5005727376861397,
      "grad_norm": 0.4918105583749727,
      "learning_rate": 5e-06,
      "loss": 0.3804,
      "step": 2620
    },
    {
      "epoch": 1.5063001145475372,
      "grad_norm": 0.5233131734973828,
      "learning_rate": 5e-06,
      "loss": 0.3808,
      "step": 2630
    },
    {
      "epoch": 1.5120274914089347,
      "grad_norm": 0.5032462746632174,
      "learning_rate": 5e-06,
      "loss": 0.3786,
      "step": 2640
    },
    {
      "epoch": 1.5177548682703321,
      "grad_norm": 0.5238794388448361,
      "learning_rate": 5e-06,
      "loss": 0.3787,
      "step": 2650
    },
    {
      "epoch": 1.5234822451317296,
      "grad_norm": 0.5075967240987918,
      "learning_rate": 5e-06,
      "loss": 0.3813,
      "step": 2660
    },
    {
      "epoch": 1.529209621993127,
      "grad_norm": 0.5094033608152486,
      "learning_rate": 5e-06,
      "loss": 0.388,
      "step": 2670
    },
    {
      "epoch": 1.5349369988545245,
      "grad_norm": 0.4937138678412315,
      "learning_rate": 5e-06,
      "loss": 0.3774,
      "step": 2680
    },
    {
      "epoch": 1.540664375715922,
      "grad_norm": 0.5074829225323386,
      "learning_rate": 5e-06,
      "loss": 0.3804,
      "step": 2690
    },
    {
      "epoch": 1.5463917525773194,
      "grad_norm": 0.5311200275972803,
      "learning_rate": 5e-06,
      "loss": 0.3715,
      "step": 2700
    },
    {
      "epoch": 1.552119129438717,
      "grad_norm": 0.47102047541148007,
      "learning_rate": 5e-06,
      "loss": 0.3726,
      "step": 2710
    },
    {
      "epoch": 1.5578465063001146,
      "grad_norm": 0.5071401413917036,
      "learning_rate": 5e-06,
      "loss": 0.3708,
      "step": 2720
    },
    {
      "epoch": 1.563573883161512,
      "grad_norm": 0.529109086962967,
      "learning_rate": 5e-06,
      "loss": 0.3802,
      "step": 2730
    },
    {
      "epoch": 1.5693012600229095,
      "grad_norm": 0.4921183498105907,
      "learning_rate": 5e-06,
      "loss": 0.3709,
      "step": 2740
    },
    {
      "epoch": 1.575028636884307,
      "grad_norm": 0.5098013107410593,
      "learning_rate": 5e-06,
      "loss": 0.3807,
      "step": 2750
    },
    {
      "epoch": 1.5807560137457046,
      "grad_norm": 0.5153311664282273,
      "learning_rate": 5e-06,
      "loss": 0.3778,
      "step": 2760
    },
    {
      "epoch": 1.586483390607102,
      "grad_norm": 0.4972397902001517,
      "learning_rate": 5e-06,
      "loss": 0.3694,
      "step": 2770
    },
    {
      "epoch": 1.5922107674684995,
      "grad_norm": 0.5119592554713294,
      "learning_rate": 5e-06,
      "loss": 0.3664,
      "step": 2780
    },
    {
      "epoch": 1.597938144329897,
      "grad_norm": 0.5267592970470695,
      "learning_rate": 5e-06,
      "loss": 0.3782,
      "step": 2790
    },
    {
      "epoch": 1.6036655211912945,
      "grad_norm": 0.5009393372942625,
      "learning_rate": 5e-06,
      "loss": 0.3809,
      "step": 2800
    },
    {
      "epoch": 1.609392898052692,
      "grad_norm": 0.5199048188264086,
      "learning_rate": 5e-06,
      "loss": 0.3839,
      "step": 2810
    },
    {
      "epoch": 1.6151202749140894,
      "grad_norm": 0.5236842623961477,
      "learning_rate": 5e-06,
      "loss": 0.3749,
      "step": 2820
    },
    {
      "epoch": 1.6208476517754868,
      "grad_norm": 0.5159942734966763,
      "learning_rate": 5e-06,
      "loss": 0.3842,
      "step": 2830
    },
    {
      "epoch": 1.6265750286368843,
      "grad_norm": 0.4903857233492226,
      "learning_rate": 5e-06,
      "loss": 0.3762,
      "step": 2840
    },
    {
      "epoch": 1.6323024054982818,
      "grad_norm": 0.4881518724661408,
      "learning_rate": 5e-06,
      "loss": 0.3813,
      "step": 2850
    },
    {
      "epoch": 1.6380297823596792,
      "grad_norm": 0.5231546664688862,
      "learning_rate": 5e-06,
      "loss": 0.373,
      "step": 2860
    },
    {
      "epoch": 1.6437571592210767,
      "grad_norm": 0.49484922058601294,
      "learning_rate": 5e-06,
      "loss": 0.3715,
      "step": 2870
    },
    {
      "epoch": 1.6494845360824741,
      "grad_norm": 0.5254331745353574,
      "learning_rate": 5e-06,
      "loss": 0.3724,
      "step": 2880
    },
    {
      "epoch": 1.6552119129438716,
      "grad_norm": 0.5095482545887045,
      "learning_rate": 5e-06,
      "loss": 0.3795,
      "step": 2890
    },
    {
      "epoch": 1.660939289805269,
      "grad_norm": 0.4741172756115684,
      "learning_rate": 5e-06,
      "loss": 0.3795,
      "step": 2900
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 0.49234693166281435,
      "learning_rate": 5e-06,
      "loss": 0.373,
      "step": 2910
    },
    {
      "epoch": 1.6723940435280642,
      "grad_norm": 0.49696468106227953,
      "learning_rate": 5e-06,
      "loss": 0.3758,
      "step": 2920
    },
    {
      "epoch": 1.6781214203894617,
      "grad_norm": 0.5077092844151242,
      "learning_rate": 5e-06,
      "loss": 0.3731,
      "step": 2930
    },
    {
      "epoch": 1.6838487972508591,
      "grad_norm": 0.5147506102086381,
      "learning_rate": 5e-06,
      "loss": 0.3757,
      "step": 2940
    },
    {
      "epoch": 1.6895761741122566,
      "grad_norm": 0.4962203469789613,
      "learning_rate": 5e-06,
      "loss": 0.3807,
      "step": 2950
    },
    {
      "epoch": 1.695303550973654,
      "grad_norm": 0.5215405834502828,
      "learning_rate": 5e-06,
      "loss": 0.3817,
      "step": 2960
    },
    {
      "epoch": 1.7010309278350515,
      "grad_norm": 0.5291514188360492,
      "learning_rate": 5e-06,
      "loss": 0.3733,
      "step": 2970
    },
    {
      "epoch": 1.7067583046964492,
      "grad_norm": 0.5152001522839965,
      "learning_rate": 5e-06,
      "loss": 0.3713,
      "step": 2980
    },
    {
      "epoch": 1.7124856815578466,
      "grad_norm": 0.4943426681496219,
      "learning_rate": 5e-06,
      "loss": 0.3691,
      "step": 2990
    },
    {
      "epoch": 1.718213058419244,
      "grad_norm": 0.5036953905099308,
      "learning_rate": 5e-06,
      "loss": 0.3732,
      "step": 3000
    },
    {
      "epoch": 1.7239404352806416,
      "grad_norm": 0.48232912691583646,
      "learning_rate": 5e-06,
      "loss": 0.3712,
      "step": 3010
    },
    {
      "epoch": 1.729667812142039,
      "grad_norm": 0.49927832455696153,
      "learning_rate": 5e-06,
      "loss": 0.3795,
      "step": 3020
    },
    {
      "epoch": 1.7353951890034365,
      "grad_norm": 0.5113424281726278,
      "learning_rate": 5e-06,
      "loss": 0.3702,
      "step": 3030
    },
    {
      "epoch": 1.741122565864834,
      "grad_norm": 0.4990266194355831,
      "learning_rate": 5e-06,
      "loss": 0.3685,
      "step": 3040
    },
    {
      "epoch": 1.7468499427262314,
      "grad_norm": 0.4950858312514841,
      "learning_rate": 5e-06,
      "loss": 0.3708,
      "step": 3050
    },
    {
      "epoch": 1.7525773195876289,
      "grad_norm": 0.505648882916022,
      "learning_rate": 5e-06,
      "loss": 0.3669,
      "step": 3060
    },
    {
      "epoch": 1.7583046964490263,
      "grad_norm": 0.5033432169804157,
      "learning_rate": 5e-06,
      "loss": 0.3752,
      "step": 3070
    },
    {
      "epoch": 1.7640320733104238,
      "grad_norm": 0.4941530503640251,
      "learning_rate": 5e-06,
      "loss": 0.3774,
      "step": 3080
    },
    {
      "epoch": 1.7697594501718212,
      "grad_norm": 0.47374323765927717,
      "learning_rate": 5e-06,
      "loss": 0.369,
      "step": 3090
    },
    {
      "epoch": 1.7754868270332187,
      "grad_norm": 0.5114072137457846,
      "learning_rate": 5e-06,
      "loss": 0.3744,
      "step": 3100
    },
    {
      "epoch": 1.7812142038946162,
      "grad_norm": 0.5714812909761224,
      "learning_rate": 5e-06,
      "loss": 0.3728,
      "step": 3110
    },
    {
      "epoch": 1.7869415807560136,
      "grad_norm": 0.5031460002692415,
      "learning_rate": 5e-06,
      "loss": 0.3586,
      "step": 3120
    },
    {
      "epoch": 1.792668957617411,
      "grad_norm": 0.5508643603790662,
      "learning_rate": 5e-06,
      "loss": 0.3775,
      "step": 3130
    },
    {
      "epoch": 1.7983963344788088,
      "grad_norm": 0.5128236251176556,
      "learning_rate": 5e-06,
      "loss": 0.3761,
      "step": 3140
    },
    {
      "epoch": 1.8041237113402062,
      "grad_norm": 0.4976776485027551,
      "learning_rate": 5e-06,
      "loss": 0.3711,
      "step": 3150
    },
    {
      "epoch": 1.8098510882016037,
      "grad_norm": 0.517111964713008,
      "learning_rate": 5e-06,
      "loss": 0.3761,
      "step": 3160
    },
    {
      "epoch": 1.8155784650630011,
      "grad_norm": 0.49528104941202294,
      "learning_rate": 5e-06,
      "loss": 0.3805,
      "step": 3170
    },
    {
      "epoch": 1.8213058419243986,
      "grad_norm": 0.49140254078298945,
      "learning_rate": 5e-06,
      "loss": 0.3765,
      "step": 3180
    },
    {
      "epoch": 1.827033218785796,
      "grad_norm": 0.507350935835637,
      "learning_rate": 5e-06,
      "loss": 0.3658,
      "step": 3190
    },
    {
      "epoch": 1.8327605956471937,
      "grad_norm": 0.514856233646019,
      "learning_rate": 5e-06,
      "loss": 0.3791,
      "step": 3200
    },
    {
      "epoch": 1.8384879725085912,
      "grad_norm": 0.48761431113920733,
      "learning_rate": 5e-06,
      "loss": 0.3719,
      "step": 3210
    },
    {
      "epoch": 1.8442153493699887,
      "grad_norm": 0.4992391499981669,
      "learning_rate": 5e-06,
      "loss": 0.3787,
      "step": 3220
    },
    {
      "epoch": 1.8499427262313861,
      "grad_norm": 0.49038499742537395,
      "learning_rate": 5e-06,
      "loss": 0.3738,
      "step": 3230
    },
    {
      "epoch": 1.8556701030927836,
      "grad_norm": 0.48881972116025446,
      "learning_rate": 5e-06,
      "loss": 0.3848,
      "step": 3240
    },
    {
      "epoch": 1.861397479954181,
      "grad_norm": 0.5075760745210575,
      "learning_rate": 5e-06,
      "loss": 0.3706,
      "step": 3250
    },
    {
      "epoch": 1.8671248568155785,
      "grad_norm": 0.5261001403985256,
      "learning_rate": 5e-06,
      "loss": 0.3695,
      "step": 3260
    },
    {
      "epoch": 1.872852233676976,
      "grad_norm": 0.5143555207892127,
      "learning_rate": 5e-06,
      "loss": 0.371,
      "step": 3270
    },
    {
      "epoch": 1.8785796105383734,
      "grad_norm": 0.516416280940184,
      "learning_rate": 5e-06,
      "loss": 0.3724,
      "step": 3280
    },
    {
      "epoch": 1.8843069873997709,
      "grad_norm": 0.5089108000531376,
      "learning_rate": 5e-06,
      "loss": 0.3722,
      "step": 3290
    },
    {
      "epoch": 1.8900343642611683,
      "grad_norm": 0.4931941440250671,
      "learning_rate": 5e-06,
      "loss": 0.3677,
      "step": 3300
    },
    {
      "epoch": 1.8957617411225658,
      "grad_norm": 0.5066849777964215,
      "learning_rate": 5e-06,
      "loss": 0.3751,
      "step": 3310
    },
    {
      "epoch": 1.9014891179839633,
      "grad_norm": 0.4924900296297563,
      "learning_rate": 5e-06,
      "loss": 0.374,
      "step": 3320
    },
    {
      "epoch": 1.9072164948453607,
      "grad_norm": 0.48837230848342283,
      "learning_rate": 5e-06,
      "loss": 0.3633,
      "step": 3330
    },
    {
      "epoch": 1.9129438717067582,
      "grad_norm": 0.5068824941492155,
      "learning_rate": 5e-06,
      "loss": 0.3786,
      "step": 3340
    },
    {
      "epoch": 1.9186712485681556,
      "grad_norm": 0.47874082053083117,
      "learning_rate": 5e-06,
      "loss": 0.3716,
      "step": 3350
    },
    {
      "epoch": 1.9243986254295533,
      "grad_norm": 0.5084759163319922,
      "learning_rate": 5e-06,
      "loss": 0.3661,
      "step": 3360
    },
    {
      "epoch": 1.9301260022909508,
      "grad_norm": 0.495142196557906,
      "learning_rate": 5e-06,
      "loss": 0.3681,
      "step": 3370
    },
    {
      "epoch": 1.9358533791523482,
      "grad_norm": 0.4894249690522435,
      "learning_rate": 5e-06,
      "loss": 0.3718,
      "step": 3380
    },
    {
      "epoch": 1.9415807560137457,
      "grad_norm": 0.5400470537238534,
      "learning_rate": 5e-06,
      "loss": 0.3788,
      "step": 3390
    },
    {
      "epoch": 1.9473081328751431,
      "grad_norm": 0.49882507022906875,
      "learning_rate": 5e-06,
      "loss": 0.3685,
      "step": 3400
    },
    {
      "epoch": 1.9530355097365406,
      "grad_norm": 0.4898539601683119,
      "learning_rate": 5e-06,
      "loss": 0.3737,
      "step": 3410
    },
    {
      "epoch": 1.9587628865979383,
      "grad_norm": 0.5065048271839747,
      "learning_rate": 5e-06,
      "loss": 0.3698,
      "step": 3420
    },
    {
      "epoch": 1.9644902634593358,
      "grad_norm": 0.493065120502946,
      "learning_rate": 5e-06,
      "loss": 0.3764,
      "step": 3430
    },
    {
      "epoch": 1.9702176403207332,
      "grad_norm": 0.4855908493410204,
      "learning_rate": 5e-06,
      "loss": 0.3701,
      "step": 3440
    },
    {
      "epoch": 1.9759450171821307,
      "grad_norm": 0.5073814840391613,
      "learning_rate": 5e-06,
      "loss": 0.3686,
      "step": 3450
    },
    {
      "epoch": 1.9816723940435281,
      "grad_norm": 0.5150533996226417,
      "learning_rate": 5e-06,
      "loss": 0.3728,
      "step": 3460
    },
    {
      "epoch": 1.9873997709049256,
      "grad_norm": 0.5168240656639432,
      "learning_rate": 5e-06,
      "loss": 0.3725,
      "step": 3470
    },
    {
      "epoch": 1.993127147766323,
      "grad_norm": 0.5067726637599477,
      "learning_rate": 5e-06,
      "loss": 0.378,
      "step": 3480
    },
    {
      "epoch": 1.9988545246277205,
      "grad_norm": 0.5310084366302013,
      "learning_rate": 5e-06,
      "loss": 0.3755,
      "step": 3490
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.4207070767879486,
      "eval_runtime": 290.7303,
      "eval_samples_per_second": 40.457,
      "eval_steps_per_second": 0.633,
      "step": 3492
    },
    {
      "epoch": 2.004581901489118,
      "grad_norm": 0.5582678235559352,
      "learning_rate": 5e-06,
      "loss": 0.3263,
      "step": 3500
    },
    {
      "epoch": 2.0103092783505154,
      "grad_norm": 0.4945423992218601,
      "learning_rate": 5e-06,
      "loss": 0.3121,
      "step": 3510
    },
    {
      "epoch": 2.016036655211913,
      "grad_norm": 0.503862666771499,
      "learning_rate": 5e-06,
      "loss": 0.3064,
      "step": 3520
    },
    {
      "epoch": 2.0217640320733103,
      "grad_norm": 0.5070969658417839,
      "learning_rate": 5e-06,
      "loss": 0.3114,
      "step": 3530
    },
    {
      "epoch": 2.027491408934708,
      "grad_norm": 0.4962584920317651,
      "learning_rate": 5e-06,
      "loss": 0.309,
      "step": 3540
    },
    {
      "epoch": 2.0332187857961053,
      "grad_norm": 0.5302688037121702,
      "learning_rate": 5e-06,
      "loss": 0.3152,
      "step": 3550
    },
    {
      "epoch": 2.0389461626575027,
      "grad_norm": 0.5169447575631415,
      "learning_rate": 5e-06,
      "loss": 0.3137,
      "step": 3560
    },
    {
      "epoch": 2.0446735395189,
      "grad_norm": 0.5127605730394688,
      "learning_rate": 5e-06,
      "loss": 0.3142,
      "step": 3570
    },
    {
      "epoch": 2.0504009163802976,
      "grad_norm": 0.49179890084778705,
      "learning_rate": 5e-06,
      "loss": 0.3138,
      "step": 3580
    },
    {
      "epoch": 2.056128293241695,
      "grad_norm": 0.5228610196241354,
      "learning_rate": 5e-06,
      "loss": 0.3272,
      "step": 3590
    },
    {
      "epoch": 2.0618556701030926,
      "grad_norm": 0.5330298953216434,
      "learning_rate": 5e-06,
      "loss": 0.3217,
      "step": 3600
    },
    {
      "epoch": 2.0675830469644905,
      "grad_norm": 0.515621478476054,
      "learning_rate": 5e-06,
      "loss": 0.3126,
      "step": 3610
    },
    {
      "epoch": 2.073310423825888,
      "grad_norm": 0.48330918450825416,
      "learning_rate": 5e-06,
      "loss": 0.3128,
      "step": 3620
    },
    {
      "epoch": 2.0790378006872854,
      "grad_norm": 0.5133555477218072,
      "learning_rate": 5e-06,
      "loss": 0.3103,
      "step": 3630
    },
    {
      "epoch": 2.084765177548683,
      "grad_norm": 0.4770491329819935,
      "learning_rate": 5e-06,
      "loss": 0.3122,
      "step": 3640
    },
    {
      "epoch": 2.0904925544100803,
      "grad_norm": 0.4964810429371462,
      "learning_rate": 5e-06,
      "loss": 0.3241,
      "step": 3650
    },
    {
      "epoch": 2.0962199312714778,
      "grad_norm": 0.490969087167152,
      "learning_rate": 5e-06,
      "loss": 0.3183,
      "step": 3660
    },
    {
      "epoch": 2.1019473081328752,
      "grad_norm": 0.5043492115565404,
      "learning_rate": 5e-06,
      "loss": 0.3142,
      "step": 3670
    },
    {
      "epoch": 2.1076746849942727,
      "grad_norm": 0.5249832358572287,
      "learning_rate": 5e-06,
      "loss": 0.3164,
      "step": 3680
    },
    {
      "epoch": 2.11340206185567,
      "grad_norm": 0.47196104358453034,
      "learning_rate": 5e-06,
      "loss": 0.3121,
      "step": 3690
    },
    {
      "epoch": 2.1191294387170676,
      "grad_norm": 0.5025207771066358,
      "learning_rate": 5e-06,
      "loss": 0.3134,
      "step": 3700
    },
    {
      "epoch": 2.124856815578465,
      "grad_norm": 0.512306750789012,
      "learning_rate": 5e-06,
      "loss": 0.3101,
      "step": 3710
    },
    {
      "epoch": 2.1305841924398625,
      "grad_norm": 0.48884001780131237,
      "learning_rate": 5e-06,
      "loss": 0.3188,
      "step": 3720
    },
    {
      "epoch": 2.13631156930126,
      "grad_norm": 0.4926488293937944,
      "learning_rate": 5e-06,
      "loss": 0.3153,
      "step": 3730
    },
    {
      "epoch": 2.1420389461626574,
      "grad_norm": 0.47726922330881655,
      "learning_rate": 5e-06,
      "loss": 0.3123,
      "step": 3740
    },
    {
      "epoch": 2.147766323024055,
      "grad_norm": 0.5330351768407757,
      "learning_rate": 5e-06,
      "loss": 0.3205,
      "step": 3750
    },
    {
      "epoch": 2.1534936998854524,
      "grad_norm": 0.4985016475823844,
      "learning_rate": 5e-06,
      "loss": 0.3139,
      "step": 3760
    },
    {
      "epoch": 2.15922107674685,
      "grad_norm": 0.5358542806513212,
      "learning_rate": 5e-06,
      "loss": 0.3182,
      "step": 3770
    },
    {
      "epoch": 2.1649484536082473,
      "grad_norm": 0.5021134280557785,
      "learning_rate": 5e-06,
      "loss": 0.317,
      "step": 3780
    },
    {
      "epoch": 2.1706758304696447,
      "grad_norm": 0.48209366607788223,
      "learning_rate": 5e-06,
      "loss": 0.3232,
      "step": 3790
    },
    {
      "epoch": 2.176403207331042,
      "grad_norm": 0.5074471012148579,
      "learning_rate": 5e-06,
      "loss": 0.315,
      "step": 3800
    },
    {
      "epoch": 2.1821305841924397,
      "grad_norm": 0.511716111734798,
      "learning_rate": 5e-06,
      "loss": 0.3157,
      "step": 3810
    },
    {
      "epoch": 2.1878579610538376,
      "grad_norm": 0.48837509446626515,
      "learning_rate": 5e-06,
      "loss": 0.3126,
      "step": 3820
    },
    {
      "epoch": 2.193585337915235,
      "grad_norm": 0.5257347682730206,
      "learning_rate": 5e-06,
      "loss": 0.3128,
      "step": 3830
    },
    {
      "epoch": 2.1993127147766325,
      "grad_norm": 0.5076959903128635,
      "learning_rate": 5e-06,
      "loss": 0.3197,
      "step": 3840
    },
    {
      "epoch": 2.20504009163803,
      "grad_norm": 0.4979642794443462,
      "learning_rate": 5e-06,
      "loss": 0.3167,
      "step": 3850
    },
    {
      "epoch": 2.2107674684994274,
      "grad_norm": 0.535375199212662,
      "learning_rate": 5e-06,
      "loss": 0.3221,
      "step": 3860
    },
    {
      "epoch": 2.216494845360825,
      "grad_norm": 0.515160036407322,
      "learning_rate": 5e-06,
      "loss": 0.3209,
      "step": 3870
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 0.4951254251771504,
      "learning_rate": 5e-06,
      "loss": 0.3201,
      "step": 3880
    },
    {
      "epoch": 2.22794959908362,
      "grad_norm": 0.521169428795531,
      "learning_rate": 5e-06,
      "loss": 0.3239,
      "step": 3890
    },
    {
      "epoch": 2.2336769759450172,
      "grad_norm": 0.49666401639454866,
      "learning_rate": 5e-06,
      "loss": 0.3137,
      "step": 3900
    },
    {
      "epoch": 2.2394043528064147,
      "grad_norm": 0.5283758992360169,
      "learning_rate": 5e-06,
      "loss": 0.3157,
      "step": 3910
    },
    {
      "epoch": 2.245131729667812,
      "grad_norm": 0.5283131673253361,
      "learning_rate": 5e-06,
      "loss": 0.3228,
      "step": 3920
    },
    {
      "epoch": 2.2508591065292096,
      "grad_norm": 0.4927867619871094,
      "learning_rate": 5e-06,
      "loss": 0.3123,
      "step": 3930
    },
    {
      "epoch": 2.256586483390607,
      "grad_norm": 0.4748812879634188,
      "learning_rate": 5e-06,
      "loss": 0.3135,
      "step": 3940
    },
    {
      "epoch": 2.2623138602520045,
      "grad_norm": 0.4893718894473708,
      "learning_rate": 5e-06,
      "loss": 0.3166,
      "step": 3950
    },
    {
      "epoch": 2.268041237113402,
      "grad_norm": 0.5104857240717058,
      "learning_rate": 5e-06,
      "loss": 0.3202,
      "step": 3960
    },
    {
      "epoch": 2.2737686139747995,
      "grad_norm": 0.5013495951931499,
      "learning_rate": 5e-06,
      "loss": 0.323,
      "step": 3970
    },
    {
      "epoch": 2.279495990836197,
      "grad_norm": 0.501822158580599,
      "learning_rate": 5e-06,
      "loss": 0.3156,
      "step": 3980
    },
    {
      "epoch": 2.2852233676975944,
      "grad_norm": 0.5283371610492251,
      "learning_rate": 5e-06,
      "loss": 0.319,
      "step": 3990
    },
    {
      "epoch": 2.290950744558992,
      "grad_norm": 0.49904707487464617,
      "learning_rate": 5e-06,
      "loss": 0.3208,
      "step": 4000
    },
    {
      "epoch": 2.2966781214203893,
      "grad_norm": 0.4853428332577101,
      "learning_rate": 5e-06,
      "loss": 0.3151,
      "step": 4010
    },
    {
      "epoch": 2.3024054982817868,
      "grad_norm": 0.49871404661053265,
      "learning_rate": 5e-06,
      "loss": 0.3198,
      "step": 4020
    },
    {
      "epoch": 2.308132875143184,
      "grad_norm": 0.48059451244658824,
      "learning_rate": 5e-06,
      "loss": 0.3104,
      "step": 4030
    },
    {
      "epoch": 2.3138602520045817,
      "grad_norm": 0.4751556595576501,
      "learning_rate": 5e-06,
      "loss": 0.3223,
      "step": 4040
    },
    {
      "epoch": 2.319587628865979,
      "grad_norm": 0.5264389154058611,
      "learning_rate": 5e-06,
      "loss": 0.3183,
      "step": 4050
    },
    {
      "epoch": 2.325315005727377,
      "grad_norm": 0.5024196062279519,
      "learning_rate": 5e-06,
      "loss": 0.3125,
      "step": 4060
    },
    {
      "epoch": 2.3310423825887745,
      "grad_norm": 0.5111784195864709,
      "learning_rate": 5e-06,
      "loss": 0.3166,
      "step": 4070
    },
    {
      "epoch": 2.336769759450172,
      "grad_norm": 0.4887735479921481,
      "learning_rate": 5e-06,
      "loss": 0.3241,
      "step": 4080
    },
    {
      "epoch": 2.3424971363115694,
      "grad_norm": 0.5330627179549918,
      "learning_rate": 5e-06,
      "loss": 0.3267,
      "step": 4090
    },
    {
      "epoch": 2.348224513172967,
      "grad_norm": 0.47798882316915986,
      "learning_rate": 5e-06,
      "loss": 0.3152,
      "step": 4100
    },
    {
      "epoch": 2.3539518900343643,
      "grad_norm": 0.5191521422457772,
      "learning_rate": 5e-06,
      "loss": 0.3293,
      "step": 4110
    },
    {
      "epoch": 2.359679266895762,
      "grad_norm": 0.5125909905451724,
      "learning_rate": 5e-06,
      "loss": 0.3134,
      "step": 4120
    },
    {
      "epoch": 2.3654066437571593,
      "grad_norm": 0.4886207764110825,
      "learning_rate": 5e-06,
      "loss": 0.3217,
      "step": 4130
    },
    {
      "epoch": 2.3711340206185567,
      "grad_norm": 0.5044443936607491,
      "learning_rate": 5e-06,
      "loss": 0.3179,
      "step": 4140
    },
    {
      "epoch": 2.376861397479954,
      "grad_norm": 0.5120039487905359,
      "learning_rate": 5e-06,
      "loss": 0.321,
      "step": 4150
    },
    {
      "epoch": 2.3825887743413516,
      "grad_norm": 0.489165925771695,
      "learning_rate": 5e-06,
      "loss": 0.3108,
      "step": 4160
    },
    {
      "epoch": 2.388316151202749,
      "grad_norm": 0.5058038370694165,
      "learning_rate": 5e-06,
      "loss": 0.3217,
      "step": 4170
    },
    {
      "epoch": 2.3940435280641466,
      "grad_norm": 0.5098394942280727,
      "learning_rate": 5e-06,
      "loss": 0.3162,
      "step": 4180
    },
    {
      "epoch": 2.399770904925544,
      "grad_norm": 0.5143275541095652,
      "learning_rate": 5e-06,
      "loss": 0.3206,
      "step": 4190
    },
    {
      "epoch": 2.4054982817869415,
      "grad_norm": 0.48723979179724397,
      "learning_rate": 5e-06,
      "loss": 0.3227,
      "step": 4200
    },
    {
      "epoch": 2.411225658648339,
      "grad_norm": 0.49153880143502765,
      "learning_rate": 5e-06,
      "loss": 0.327,
      "step": 4210
    },
    {
      "epoch": 2.4169530355097364,
      "grad_norm": 0.4990389931982443,
      "learning_rate": 5e-06,
      "loss": 0.3145,
      "step": 4220
    },
    {
      "epoch": 2.422680412371134,
      "grad_norm": 0.5007707010716869,
      "learning_rate": 5e-06,
      "loss": 0.3192,
      "step": 4230
    },
    {
      "epoch": 2.4284077892325313,
      "grad_norm": 0.4842447385057361,
      "learning_rate": 5e-06,
      "loss": 0.3187,
      "step": 4240
    },
    {
      "epoch": 2.434135166093929,
      "grad_norm": 0.5282173056004758,
      "learning_rate": 5e-06,
      "loss": 0.3248,
      "step": 4250
    },
    {
      "epoch": 2.4398625429553267,
      "grad_norm": 0.5242920164717728,
      "learning_rate": 5e-06,
      "loss": 0.3276,
      "step": 4260
    },
    {
      "epoch": 2.445589919816724,
      "grad_norm": 0.5256084243387356,
      "learning_rate": 5e-06,
      "loss": 0.3076,
      "step": 4270
    },
    {
      "epoch": 2.4513172966781216,
      "grad_norm": 0.49717599592896866,
      "learning_rate": 5e-06,
      "loss": 0.3183,
      "step": 4280
    },
    {
      "epoch": 2.457044673539519,
      "grad_norm": 0.4917289115217294,
      "learning_rate": 5e-06,
      "loss": 0.3169,
      "step": 4290
    },
    {
      "epoch": 2.4627720504009165,
      "grad_norm": 0.48696947359252235,
      "learning_rate": 5e-06,
      "loss": 0.3181,
      "step": 4300
    },
    {
      "epoch": 2.468499427262314,
      "grad_norm": 0.5066247814081654,
      "learning_rate": 5e-06,
      "loss": 0.3214,
      "step": 4310
    },
    {
      "epoch": 2.4742268041237114,
      "grad_norm": 0.5116828561516821,
      "learning_rate": 5e-06,
      "loss": 0.3232,
      "step": 4320
    },
    {
      "epoch": 2.479954180985109,
      "grad_norm": 0.4721726863796144,
      "learning_rate": 5e-06,
      "loss": 0.3203,
      "step": 4330
    },
    {
      "epoch": 2.4856815578465064,
      "grad_norm": 0.5140261292884363,
      "learning_rate": 5e-06,
      "loss": 0.3205,
      "step": 4340
    },
    {
      "epoch": 2.491408934707904,
      "grad_norm": 0.4995754907401082,
      "learning_rate": 5e-06,
      "loss": 0.3258,
      "step": 4350
    },
    {
      "epoch": 2.4971363115693013,
      "grad_norm": 0.5041823551282999,
      "learning_rate": 5e-06,
      "loss": 0.3194,
      "step": 4360
    },
    {
      "epoch": 2.5028636884306987,
      "grad_norm": 0.49384248849165524,
      "learning_rate": 5e-06,
      "loss": 0.3225,
      "step": 4370
    },
    {
      "epoch": 2.508591065292096,
      "grad_norm": 0.4889201995538099,
      "learning_rate": 5e-06,
      "loss": 0.3205,
      "step": 4380
    },
    {
      "epoch": 2.5143184421534936,
      "grad_norm": 0.47694834539706893,
      "learning_rate": 5e-06,
      "loss": 0.3175,
      "step": 4390
    },
    {
      "epoch": 2.520045819014891,
      "grad_norm": 0.49461329366934864,
      "learning_rate": 5e-06,
      "loss": 0.3223,
      "step": 4400
    },
    {
      "epoch": 2.5257731958762886,
      "grad_norm": 0.5125961099921013,
      "learning_rate": 5e-06,
      "loss": 0.323,
      "step": 4410
    },
    {
      "epoch": 2.531500572737686,
      "grad_norm": 0.48971922047461597,
      "learning_rate": 5e-06,
      "loss": 0.3201,
      "step": 4420
    },
    {
      "epoch": 2.5372279495990835,
      "grad_norm": 0.47489099276070756,
      "learning_rate": 5e-06,
      "loss": 0.3227,
      "step": 4430
    },
    {
      "epoch": 2.542955326460481,
      "grad_norm": 0.4907691648649352,
      "learning_rate": 5e-06,
      "loss": 0.3201,
      "step": 4440
    },
    {
      "epoch": 2.5486827033218784,
      "grad_norm": 0.5026363282224567,
      "learning_rate": 5e-06,
      "loss": 0.3196,
      "step": 4450
    },
    {
      "epoch": 2.554410080183276,
      "grad_norm": 0.5176119238693507,
      "learning_rate": 5e-06,
      "loss": 0.3243,
      "step": 4460
    },
    {
      "epoch": 2.5601374570446733,
      "grad_norm": 0.5085068635390583,
      "learning_rate": 5e-06,
      "loss": 0.3183,
      "step": 4470
    },
    {
      "epoch": 2.565864833906071,
      "grad_norm": 0.472068765168504,
      "learning_rate": 5e-06,
      "loss": 0.3172,
      "step": 4480
    },
    {
      "epoch": 2.5715922107674682,
      "grad_norm": 0.5107727781006867,
      "learning_rate": 5e-06,
      "loss": 0.3276,
      "step": 4490
    },
    {
      "epoch": 2.5773195876288657,
      "grad_norm": 0.49576533772806247,
      "learning_rate": 5e-06,
      "loss": 0.3197,
      "step": 4500
    },
    {
      "epoch": 2.5830469644902636,
      "grad_norm": 0.5147069600733026,
      "learning_rate": 5e-06,
      "loss": 0.3303,
      "step": 4510
    },
    {
      "epoch": 2.588774341351661,
      "grad_norm": 0.5410222797339587,
      "learning_rate": 5e-06,
      "loss": 0.3213,
      "step": 4520
    },
    {
      "epoch": 2.5945017182130585,
      "grad_norm": 0.5129756441436896,
      "learning_rate": 5e-06,
      "loss": 0.3159,
      "step": 4530
    },
    {
      "epoch": 2.600229095074456,
      "grad_norm": 0.5090769089905053,
      "learning_rate": 5e-06,
      "loss": 0.3198,
      "step": 4540
    },
    {
      "epoch": 2.6059564719358534,
      "grad_norm": 0.5312483213444277,
      "learning_rate": 5e-06,
      "loss": 0.3239,
      "step": 4550
    },
    {
      "epoch": 2.611683848797251,
      "grad_norm": 0.4990401487796137,
      "learning_rate": 5e-06,
      "loss": 0.3165,
      "step": 4560
    },
    {
      "epoch": 2.6174112256586484,
      "grad_norm": 0.4906139970113933,
      "learning_rate": 5e-06,
      "loss": 0.3162,
      "step": 4570
    },
    {
      "epoch": 2.623138602520046,
      "grad_norm": 0.49275236064069183,
      "learning_rate": 5e-06,
      "loss": 0.3162,
      "step": 4580
    },
    {
      "epoch": 2.6288659793814433,
      "grad_norm": 0.505876180763233,
      "learning_rate": 5e-06,
      "loss": 0.3173,
      "step": 4590
    },
    {
      "epoch": 2.6345933562428407,
      "grad_norm": 0.5034109210608266,
      "learning_rate": 5e-06,
      "loss": 0.327,
      "step": 4600
    },
    {
      "epoch": 2.640320733104238,
      "grad_norm": 0.5125105668846632,
      "learning_rate": 5e-06,
      "loss": 0.3201,
      "step": 4610
    },
    {
      "epoch": 2.6460481099656357,
      "grad_norm": 0.4986777563152259,
      "learning_rate": 5e-06,
      "loss": 0.3243,
      "step": 4620
    },
    {
      "epoch": 2.651775486827033,
      "grad_norm": 0.5028807375636888,
      "learning_rate": 5e-06,
      "loss": 0.313,
      "step": 4630
    },
    {
      "epoch": 2.6575028636884306,
      "grad_norm": 0.5027638307219331,
      "learning_rate": 5e-06,
      "loss": 0.3251,
      "step": 4640
    },
    {
      "epoch": 2.663230240549828,
      "grad_norm": 0.5118793036289471,
      "learning_rate": 5e-06,
      "loss": 0.3171,
      "step": 4650
    },
    {
      "epoch": 2.6689576174112255,
      "grad_norm": 0.5079584260143071,
      "learning_rate": 5e-06,
      "loss": 0.3196,
      "step": 4660
    },
    {
      "epoch": 2.6746849942726234,
      "grad_norm": 0.5098308127405584,
      "learning_rate": 5e-06,
      "loss": 0.33,
      "step": 4670
    },
    {
      "epoch": 2.680412371134021,
      "grad_norm": 0.5028460889046886,
      "learning_rate": 5e-06,
      "loss": 0.3233,
      "step": 4680
    },
    {
      "epoch": 2.6861397479954183,
      "grad_norm": 0.4950860630453671,
      "learning_rate": 5e-06,
      "loss": 0.3218,
      "step": 4690
    },
    {
      "epoch": 2.691867124856816,
      "grad_norm": 0.5363041499133996,
      "learning_rate": 5e-06,
      "loss": 0.3115,
      "step": 4700
    },
    {
      "epoch": 2.6975945017182132,
      "grad_norm": 0.47848642716067413,
      "learning_rate": 5e-06,
      "loss": 0.3229,
      "step": 4710
    },
    {
      "epoch": 2.7033218785796107,
      "grad_norm": 0.5132570731802132,
      "learning_rate": 5e-06,
      "loss": 0.3176,
      "step": 4720
    },
    {
      "epoch": 2.709049255441008,
      "grad_norm": 0.5123910169044057,
      "learning_rate": 5e-06,
      "loss": 0.3216,
      "step": 4730
    },
    {
      "epoch": 2.7147766323024056,
      "grad_norm": 0.49140644629093294,
      "learning_rate": 5e-06,
      "loss": 0.3255,
      "step": 4740
    },
    {
      "epoch": 2.720504009163803,
      "grad_norm": 0.4903514464262286,
      "learning_rate": 5e-06,
      "loss": 0.3221,
      "step": 4750
    },
    {
      "epoch": 2.7262313860252005,
      "grad_norm": 0.5202137394656217,
      "learning_rate": 5e-06,
      "loss": 0.3188,
      "step": 4760
    },
    {
      "epoch": 2.731958762886598,
      "grad_norm": 0.5189552209211002,
      "learning_rate": 5e-06,
      "loss": 0.3268,
      "step": 4770
    },
    {
      "epoch": 2.7376861397479955,
      "grad_norm": 0.4863895418886132,
      "learning_rate": 5e-06,
      "loss": 0.3146,
      "step": 4780
    },
    {
      "epoch": 2.743413516609393,
      "grad_norm": 0.4963628119686479,
      "learning_rate": 5e-06,
      "loss": 0.3306,
      "step": 4790
    },
    {
      "epoch": 2.7491408934707904,
      "grad_norm": 0.4780341556691321,
      "learning_rate": 5e-06,
      "loss": 0.3183,
      "step": 4800
    },
    {
      "epoch": 2.754868270332188,
      "grad_norm": 0.47632144883855376,
      "learning_rate": 5e-06,
      "loss": 0.3102,
      "step": 4810
    },
    {
      "epoch": 2.7605956471935853,
      "grad_norm": 0.5229861996836561,
      "learning_rate": 5e-06,
      "loss": 0.3296,
      "step": 4820
    },
    {
      "epoch": 2.7663230240549828,
      "grad_norm": 0.5062883215395301,
      "learning_rate": 5e-06,
      "loss": 0.3233,
      "step": 4830
    },
    {
      "epoch": 2.77205040091638,
      "grad_norm": 0.4943950877672053,
      "learning_rate": 5e-06,
      "loss": 0.3273,
      "step": 4840
    },
    {
      "epoch": 2.7777777777777777,
      "grad_norm": 0.5098452348979725,
      "learning_rate": 5e-06,
      "loss": 0.3169,
      "step": 4850
    },
    {
      "epoch": 2.783505154639175,
      "grad_norm": 0.5014610096004497,
      "learning_rate": 5e-06,
      "loss": 0.3232,
      "step": 4860
    },
    {
      "epoch": 2.7892325315005726,
      "grad_norm": 0.5013602455951724,
      "learning_rate": 5e-06,
      "loss": 0.3183,
      "step": 4870
    },
    {
      "epoch": 2.79495990836197,
      "grad_norm": 0.53869757575354,
      "learning_rate": 5e-06,
      "loss": 0.3251,
      "step": 4880
    },
    {
      "epoch": 2.8006872852233675,
      "grad_norm": 0.4898326008152037,
      "learning_rate": 5e-06,
      "loss": 0.3231,
      "step": 4890
    },
    {
      "epoch": 2.806414662084765,
      "grad_norm": 0.4810606140091934,
      "learning_rate": 5e-06,
      "loss": 0.3174,
      "step": 4900
    },
    {
      "epoch": 2.8121420389461624,
      "grad_norm": 0.5170925060870262,
      "learning_rate": 5e-06,
      "loss": 0.3206,
      "step": 4910
    },
    {
      "epoch": 2.81786941580756,
      "grad_norm": 0.4980004979531469,
      "learning_rate": 5e-06,
      "loss": 0.3271,
      "step": 4920
    },
    {
      "epoch": 2.8235967926689574,
      "grad_norm": 0.4751276576064784,
      "learning_rate": 5e-06,
      "loss": 0.3255,
      "step": 4930
    },
    {
      "epoch": 2.8293241695303553,
      "grad_norm": 0.4858328987324905,
      "learning_rate": 5e-06,
      "loss": 0.3199,
      "step": 4940
    },
    {
      "epoch": 2.8350515463917527,
      "grad_norm": 0.5147620527991269,
      "learning_rate": 5e-06,
      "loss": 0.3287,
      "step": 4950
    },
    {
      "epoch": 2.84077892325315,
      "grad_norm": 0.5149232879792602,
      "learning_rate": 5e-06,
      "loss": 0.3274,
      "step": 4960
    },
    {
      "epoch": 2.8465063001145476,
      "grad_norm": 0.5008258037850903,
      "learning_rate": 5e-06,
      "loss": 0.3166,
      "step": 4970
    },
    {
      "epoch": 2.852233676975945,
      "grad_norm": 0.5022192392797802,
      "learning_rate": 5e-06,
      "loss": 0.322,
      "step": 4980
    },
    {
      "epoch": 2.8579610538373426,
      "grad_norm": 0.47652790071973783,
      "learning_rate": 5e-06,
      "loss": 0.3189,
      "step": 4990
    },
    {
      "epoch": 2.86368843069874,
      "grad_norm": 0.4759345790313986,
      "learning_rate": 5e-06,
      "loss": 0.3215,
      "step": 5000
    },
    {
      "epoch": 2.8694158075601375,
      "grad_norm": 0.5072028525553413,
      "learning_rate": 5e-06,
      "loss": 0.3293,
      "step": 5010
    },
    {
      "epoch": 2.875143184421535,
      "grad_norm": 0.5143722957913531,
      "learning_rate": 5e-06,
      "loss": 0.3237,
      "step": 5020
    },
    {
      "epoch": 2.8808705612829324,
      "grad_norm": 0.5041664264008966,
      "learning_rate": 5e-06,
      "loss": 0.3171,
      "step": 5030
    },
    {
      "epoch": 2.88659793814433,
      "grad_norm": 0.49156126691930446,
      "learning_rate": 5e-06,
      "loss": 0.3153,
      "step": 5040
    },
    {
      "epoch": 2.8923253150057273,
      "grad_norm": 0.49867336387580574,
      "learning_rate": 5e-06,
      "loss": 0.3263,
      "step": 5050
    },
    {
      "epoch": 2.8980526918671248,
      "grad_norm": 0.46719115878858214,
      "learning_rate": 5e-06,
      "loss": 0.3214,
      "step": 5060
    },
    {
      "epoch": 2.9037800687285222,
      "grad_norm": 0.48603049910129803,
      "learning_rate": 5e-06,
      "loss": 0.3192,
      "step": 5070
    },
    {
      "epoch": 2.9095074455899197,
      "grad_norm": 0.5044239151316462,
      "learning_rate": 5e-06,
      "loss": 0.3218,
      "step": 5080
    },
    {
      "epoch": 2.915234822451317,
      "grad_norm": 0.4870384358690395,
      "learning_rate": 5e-06,
      "loss": 0.318,
      "step": 5090
    },
    {
      "epoch": 2.9209621993127146,
      "grad_norm": 0.512106578999376,
      "learning_rate": 5e-06,
      "loss": 0.3194,
      "step": 5100
    },
    {
      "epoch": 2.9266895761741125,
      "grad_norm": 0.5023395866834555,
      "learning_rate": 5e-06,
      "loss": 0.3297,
      "step": 5110
    },
    {
      "epoch": 2.93241695303551,
      "grad_norm": 0.5257957417842357,
      "learning_rate": 5e-06,
      "loss": 0.3246,
      "step": 5120
    },
    {
      "epoch": 2.9381443298969074,
      "grad_norm": 0.48614775617490547,
      "learning_rate": 5e-06,
      "loss": 0.318,
      "step": 5130
    },
    {
      "epoch": 2.943871706758305,
      "grad_norm": 0.49942615792365136,
      "learning_rate": 5e-06,
      "loss": 0.3196,
      "step": 5140
    },
    {
      "epoch": 2.9495990836197024,
      "grad_norm": 0.49083897626357387,
      "learning_rate": 5e-06,
      "loss": 0.3129,
      "step": 5150
    },
    {
      "epoch": 2.9553264604811,
      "grad_norm": 0.49254139484612786,
      "learning_rate": 5e-06,
      "loss": 0.3247,
      "step": 5160
    },
    {
      "epoch": 2.9610538373424973,
      "grad_norm": 0.5090239638844181,
      "learning_rate": 5e-06,
      "loss": 0.3198,
      "step": 5170
    },
    {
      "epoch": 2.9667812142038947,
      "grad_norm": 0.504113575432436,
      "learning_rate": 5e-06,
      "loss": 0.3184,
      "step": 5180
    },
    {
      "epoch": 2.972508591065292,
      "grad_norm": 0.500020414647489,
      "learning_rate": 5e-06,
      "loss": 0.3167,
      "step": 5190
    },
    {
      "epoch": 2.9782359679266897,
      "grad_norm": 0.5217079453004775,
      "learning_rate": 5e-06,
      "loss": 0.3263,
      "step": 5200
    },
    {
      "epoch": 2.983963344788087,
      "grad_norm": 0.47426209559251875,
      "learning_rate": 5e-06,
      "loss": 0.3185,
      "step": 5210
    },
    {
      "epoch": 2.9896907216494846,
      "grad_norm": 0.5033657036471195,
      "learning_rate": 5e-06,
      "loss": 0.3159,
      "step": 5220
    },
    {
      "epoch": 2.995418098510882,
      "grad_norm": 0.49409722412965856,
      "learning_rate": 5e-06,
      "loss": 0.3266,
      "step": 5230
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.42773160338401794,
      "eval_runtime": 278.3004,
      "eval_samples_per_second": 42.264,
      "eval_steps_per_second": 0.661,
      "step": 5238
    },
    {
      "epoch": 3.0,
      "step": 5238,
      "total_flos": 2745782658662400.0,
      "train_loss": 0.38240701185468956,
      "train_runtime": 39622.8086,
      "train_samples_per_second": 16.92,
      "train_steps_per_second": 0.132
    }
  ],
  "logging_steps": 10,
  "max_steps": 5238,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "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": 2745782658662400.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}