{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9991537376586743,
  "eval_steps": 500,
  "global_step": 1329,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.022566995768688293,
      "grad_norm": 3.9500676023225703,
      "learning_rate": 3.7593984962406015e-07,
      "loss": 0.8582,
      "step": 10
    },
    {
      "epoch": 0.045133991537376586,
      "grad_norm": 1.395677808116374,
      "learning_rate": 7.518796992481203e-07,
      "loss": 0.8024,
      "step": 20
    },
    {
      "epoch": 0.06770098730606489,
      "grad_norm": 0.9062193282306004,
      "learning_rate": 1.1278195488721805e-06,
      "loss": 0.7357,
      "step": 30
    },
    {
      "epoch": 0.09026798307475317,
      "grad_norm": 0.8282785482347225,
      "learning_rate": 1.5037593984962406e-06,
      "loss": 0.6898,
      "step": 40
    },
    {
      "epoch": 0.11283497884344147,
      "grad_norm": 0.701981062657255,
      "learning_rate": 1.8796992481203007e-06,
      "loss": 0.6707,
      "step": 50
    },
    {
      "epoch": 0.13540197461212977,
      "grad_norm": 0.640628105313228,
      "learning_rate": 2.255639097744361e-06,
      "loss": 0.6616,
      "step": 60
    },
    {
      "epoch": 0.15796897038081806,
      "grad_norm": 0.6794950494897954,
      "learning_rate": 2.631578947368421e-06,
      "loss": 0.6518,
      "step": 70
    },
    {
      "epoch": 0.18053596614950634,
      "grad_norm": 0.7767981323221586,
      "learning_rate": 3.007518796992481e-06,
      "loss": 0.6496,
      "step": 80
    },
    {
      "epoch": 0.20310296191819463,
      "grad_norm": 0.7882061395786479,
      "learning_rate": 3.3834586466165413e-06,
      "loss": 0.6451,
      "step": 90
    },
    {
      "epoch": 0.22566995768688294,
      "grad_norm": 0.9439189657933283,
      "learning_rate": 3.7593984962406014e-06,
      "loss": 0.6417,
      "step": 100
    },
    {
      "epoch": 0.24823695345557123,
      "grad_norm": 0.7346506648254489,
      "learning_rate": 4.135338345864662e-06,
      "loss": 0.6314,
      "step": 110
    },
    {
      "epoch": 0.27080394922425954,
      "grad_norm": 0.6595327999279074,
      "learning_rate": 4.511278195488722e-06,
      "loss": 0.6293,
      "step": 120
    },
    {
      "epoch": 0.2933709449929478,
      "grad_norm": 0.8336967878084969,
      "learning_rate": 4.887218045112782e-06,
      "loss": 0.6301,
      "step": 130
    },
    {
      "epoch": 0.3159379407616361,
      "grad_norm": 1.0637113778881198,
      "learning_rate": 4.970735785953177e-06,
      "loss": 0.6245,
      "step": 140
    },
    {
      "epoch": 0.3385049365303244,
      "grad_norm": 0.6759376136072178,
      "learning_rate": 4.928929765886288e-06,
      "loss": 0.6245,
      "step": 150
    },
    {
      "epoch": 0.3610719322990127,
      "grad_norm": 1.0169288303598698,
      "learning_rate": 4.887123745819398e-06,
      "loss": 0.6261,
      "step": 160
    },
    {
      "epoch": 0.383638928067701,
      "grad_norm": 0.7994151168318469,
      "learning_rate": 4.845317725752509e-06,
      "loss": 0.621,
      "step": 170
    },
    {
      "epoch": 0.40620592383638926,
      "grad_norm": 0.7288471424905668,
      "learning_rate": 4.803511705685619e-06,
      "loss": 0.6169,
      "step": 180
    },
    {
      "epoch": 0.4287729196050776,
      "grad_norm": 0.6569051029133128,
      "learning_rate": 4.7617056856187295e-06,
      "loss": 0.6107,
      "step": 190
    },
    {
      "epoch": 0.4513399153737659,
      "grad_norm": 0.6306476938762344,
      "learning_rate": 4.71989966555184e-06,
      "loss": 0.6166,
      "step": 200
    },
    {
      "epoch": 0.47390691114245415,
      "grad_norm": 0.6011927260340564,
      "learning_rate": 4.67809364548495e-06,
      "loss": 0.6128,
      "step": 210
    },
    {
      "epoch": 0.49647390691114246,
      "grad_norm": 0.6568818099792321,
      "learning_rate": 4.63628762541806e-06,
      "loss": 0.6041,
      "step": 220
    },
    {
      "epoch": 0.5190409026798307,
      "grad_norm": 0.6201063357053574,
      "learning_rate": 4.594481605351171e-06,
      "loss": 0.6126,
      "step": 230
    },
    {
      "epoch": 0.5416078984485191,
      "grad_norm": 0.5804525175003069,
      "learning_rate": 4.5526755852842815e-06,
      "loss": 0.611,
      "step": 240
    },
    {
      "epoch": 0.5641748942172073,
      "grad_norm": 0.642468562835698,
      "learning_rate": 4.510869565217392e-06,
      "loss": 0.6049,
      "step": 250
    },
    {
      "epoch": 0.5867418899858956,
      "grad_norm": 0.6897016737786636,
      "learning_rate": 4.469063545150502e-06,
      "loss": 0.6073,
      "step": 260
    },
    {
      "epoch": 0.609308885754584,
      "grad_norm": 0.6770865024228586,
      "learning_rate": 4.427257525083612e-06,
      "loss": 0.6025,
      "step": 270
    },
    {
      "epoch": 0.6318758815232722,
      "grad_norm": 0.6074629511796228,
      "learning_rate": 4.3854515050167225e-06,
      "loss": 0.609,
      "step": 280
    },
    {
      "epoch": 0.6544428772919605,
      "grad_norm": 0.6220418776986034,
      "learning_rate": 4.3436454849498336e-06,
      "loss": 0.6021,
      "step": 290
    },
    {
      "epoch": 0.6770098730606487,
      "grad_norm": 0.7074012275628311,
      "learning_rate": 4.301839464882944e-06,
      "loss": 0.6081,
      "step": 300
    },
    {
      "epoch": 0.6995768688293371,
      "grad_norm": 0.5964332959901174,
      "learning_rate": 4.260033444816054e-06,
      "loss": 0.6011,
      "step": 310
    },
    {
      "epoch": 0.7221438645980254,
      "grad_norm": 0.6259056006758187,
      "learning_rate": 4.218227424749164e-06,
      "loss": 0.6042,
      "step": 320
    },
    {
      "epoch": 0.7447108603667136,
      "grad_norm": 0.5493942038048834,
      "learning_rate": 4.1764214046822745e-06,
      "loss": 0.5995,
      "step": 330
    },
    {
      "epoch": 0.767277856135402,
      "grad_norm": 0.6181685023195539,
      "learning_rate": 4.134615384615385e-06,
      "loss": 0.5987,
      "step": 340
    },
    {
      "epoch": 0.7898448519040903,
      "grad_norm": 0.588036854641419,
      "learning_rate": 4.092809364548495e-06,
      "loss": 0.6,
      "step": 350
    },
    {
      "epoch": 0.8124118476727785,
      "grad_norm": 0.6828867431467442,
      "learning_rate": 4.051003344481605e-06,
      "loss": 0.5989,
      "step": 360
    },
    {
      "epoch": 0.8349788434414669,
      "grad_norm": 0.6024684614349975,
      "learning_rate": 4.0091973244147155e-06,
      "loss": 0.6059,
      "step": 370
    },
    {
      "epoch": 0.8575458392101551,
      "grad_norm": 0.5987044058560508,
      "learning_rate": 3.967391304347827e-06,
      "loss": 0.6023,
      "step": 380
    },
    {
      "epoch": 0.8801128349788434,
      "grad_norm": 0.6031854798342579,
      "learning_rate": 3.925585284280937e-06,
      "loss": 0.5963,
      "step": 390
    },
    {
      "epoch": 0.9026798307475318,
      "grad_norm": 0.5784643272502593,
      "learning_rate": 3.883779264214047e-06,
      "loss": 0.5984,
      "step": 400
    },
    {
      "epoch": 0.92524682651622,
      "grad_norm": 0.5828958217828615,
      "learning_rate": 3.841973244147157e-06,
      "loss": 0.5963,
      "step": 410
    },
    {
      "epoch": 0.9478138222849083,
      "grad_norm": 0.6221793656032795,
      "learning_rate": 3.800167224080268e-06,
      "loss": 0.5968,
      "step": 420
    },
    {
      "epoch": 0.9703808180535967,
      "grad_norm": 0.5425155387264896,
      "learning_rate": 3.7583612040133782e-06,
      "loss": 0.5907,
      "step": 430
    },
    {
      "epoch": 0.9929478138222849,
      "grad_norm": 0.5548540193834617,
      "learning_rate": 3.7165551839464885e-06,
      "loss": 0.5953,
      "step": 440
    },
    {
      "epoch": 0.9997179125528914,
      "eval_loss": 0.5933970212936401,
      "eval_runtime": 687.5019,
      "eval_samples_per_second": 17.369,
      "eval_steps_per_second": 0.544,
      "step": 443
    },
    {
      "epoch": 1.0155148095909732,
      "grad_norm": 0.663986999357991,
      "learning_rate": 3.674749163879599e-06,
      "loss": 0.6102,
      "step": 450
    },
    {
      "epoch": 1.0380818053596614,
      "grad_norm": 0.6073973469870063,
      "learning_rate": 3.6329431438127094e-06,
      "loss": 0.5402,
      "step": 460
    },
    {
      "epoch": 1.0606488011283497,
      "grad_norm": 0.5871743515341908,
      "learning_rate": 3.5911371237458196e-06,
      "loss": 0.5553,
      "step": 470
    },
    {
      "epoch": 1.0832157968970382,
      "grad_norm": 0.5998465628010546,
      "learning_rate": 3.5493311036789303e-06,
      "loss": 0.5428,
      "step": 480
    },
    {
      "epoch": 1.1057827926657264,
      "grad_norm": 0.631113957346967,
      "learning_rate": 3.5075250836120405e-06,
      "loss": 0.553,
      "step": 490
    },
    {
      "epoch": 1.1283497884344147,
      "grad_norm": 0.5595994504229799,
      "learning_rate": 3.465719063545151e-06,
      "loss": 0.5511,
      "step": 500
    },
    {
      "epoch": 1.150916784203103,
      "grad_norm": 0.5718415684682732,
      "learning_rate": 3.4239130434782614e-06,
      "loss": 0.5519,
      "step": 510
    },
    {
      "epoch": 1.1734837799717912,
      "grad_norm": 0.5565246715923281,
      "learning_rate": 3.3821070234113717e-06,
      "loss": 0.5441,
      "step": 520
    },
    {
      "epoch": 1.1960507757404795,
      "grad_norm": 0.5740404698821812,
      "learning_rate": 3.3403010033444823e-06,
      "loss": 0.5498,
      "step": 530
    },
    {
      "epoch": 1.2186177715091677,
      "grad_norm": 0.6406151182955427,
      "learning_rate": 3.298494983277592e-06,
      "loss": 0.5472,
      "step": 540
    },
    {
      "epoch": 1.2411847672778562,
      "grad_norm": 1.0247459934962126,
      "learning_rate": 3.2566889632107024e-06,
      "loss": 0.5519,
      "step": 550
    },
    {
      "epoch": 1.2637517630465445,
      "grad_norm": 0.6137317588735225,
      "learning_rate": 3.2148829431438126e-06,
      "loss": 0.5537,
      "step": 560
    },
    {
      "epoch": 1.2863187588152327,
      "grad_norm": 0.5803805149684825,
      "learning_rate": 3.1730769230769233e-06,
      "loss": 0.5507,
      "step": 570
    },
    {
      "epoch": 1.308885754583921,
      "grad_norm": 0.614582832632458,
      "learning_rate": 3.1312709030100335e-06,
      "loss": 0.5505,
      "step": 580
    },
    {
      "epoch": 1.3314527503526092,
      "grad_norm": 0.5795519009889963,
      "learning_rate": 3.0894648829431438e-06,
      "loss": 0.5501,
      "step": 590
    },
    {
      "epoch": 1.3540197461212977,
      "grad_norm": 0.556651253096743,
      "learning_rate": 3.0476588628762544e-06,
      "loss": 0.5557,
      "step": 600
    },
    {
      "epoch": 1.376586741889986,
      "grad_norm": 0.5720074610909325,
      "learning_rate": 3.0058528428093647e-06,
      "loss": 0.5473,
      "step": 610
    },
    {
      "epoch": 1.3991537376586742,
      "grad_norm": 0.5813836681865406,
      "learning_rate": 2.964046822742475e-06,
      "loss": 0.5517,
      "step": 620
    },
    {
      "epoch": 1.4217207334273625,
      "grad_norm": 0.5523421119994065,
      "learning_rate": 2.9222408026755856e-06,
      "loss": 0.5495,
      "step": 630
    },
    {
      "epoch": 1.4442877291960508,
      "grad_norm": 0.5617619307728544,
      "learning_rate": 2.880434782608696e-06,
      "loss": 0.5488,
      "step": 640
    },
    {
      "epoch": 1.466854724964739,
      "grad_norm": 0.6071916695080162,
      "learning_rate": 2.8386287625418065e-06,
      "loss": 0.5496,
      "step": 650
    },
    {
      "epoch": 1.4894217207334273,
      "grad_norm": 0.6034608381516856,
      "learning_rate": 2.7968227424749167e-06,
      "loss": 0.5532,
      "step": 660
    },
    {
      "epoch": 1.5119887165021155,
      "grad_norm": 0.5806505121518643,
      "learning_rate": 2.755016722408027e-06,
      "loss": 0.5478,
      "step": 670
    },
    {
      "epoch": 1.5345557122708038,
      "grad_norm": 0.5823380773786808,
      "learning_rate": 2.7132107023411376e-06,
      "loss": 0.542,
      "step": 680
    },
    {
      "epoch": 1.5571227080394923,
      "grad_norm": 0.591931176130096,
      "learning_rate": 2.671404682274248e-06,
      "loss": 0.5436,
      "step": 690
    },
    {
      "epoch": 1.5796897038081805,
      "grad_norm": 0.575757448424994,
      "learning_rate": 2.629598662207358e-06,
      "loss": 0.5479,
      "step": 700
    },
    {
      "epoch": 1.6022566995768688,
      "grad_norm": 0.5440261614938245,
      "learning_rate": 2.5877926421404688e-06,
      "loss": 0.5505,
      "step": 710
    },
    {
      "epoch": 1.6248236953455573,
      "grad_norm": 0.5866164019825096,
      "learning_rate": 2.545986622073579e-06,
      "loss": 0.5583,
      "step": 720
    },
    {
      "epoch": 1.6473906911142455,
      "grad_norm": 0.574480308049467,
      "learning_rate": 2.5041806020066893e-06,
      "loss": 0.5414,
      "step": 730
    },
    {
      "epoch": 1.6699576868829338,
      "grad_norm": 0.5987421971532259,
      "learning_rate": 2.4623745819397995e-06,
      "loss": 0.5567,
      "step": 740
    },
    {
      "epoch": 1.692524682651622,
      "grad_norm": 0.5606374529736117,
      "learning_rate": 2.4205685618729097e-06,
      "loss": 0.5481,
      "step": 750
    },
    {
      "epoch": 1.7150916784203103,
      "grad_norm": 0.5589529879882318,
      "learning_rate": 2.3787625418060204e-06,
      "loss": 0.547,
      "step": 760
    },
    {
      "epoch": 1.7376586741889986,
      "grad_norm": 0.537042270204007,
      "learning_rate": 2.3369565217391307e-06,
      "loss": 0.5472,
      "step": 770
    },
    {
      "epoch": 1.7602256699576868,
      "grad_norm": 0.5692518598971062,
      "learning_rate": 2.2951505016722413e-06,
      "loss": 0.5434,
      "step": 780
    },
    {
      "epoch": 1.782792665726375,
      "grad_norm": 0.548945107183346,
      "learning_rate": 2.253344481605351e-06,
      "loss": 0.5464,
      "step": 790
    },
    {
      "epoch": 1.8053596614950633,
      "grad_norm": 0.5612980211542685,
      "learning_rate": 2.211538461538462e-06,
      "loss": 0.5412,
      "step": 800
    },
    {
      "epoch": 1.8279266572637518,
      "grad_norm": 0.5352386025237578,
      "learning_rate": 2.169732441471572e-06,
      "loss": 0.5519,
      "step": 810
    },
    {
      "epoch": 1.85049365303244,
      "grad_norm": 0.5369069746509202,
      "learning_rate": 2.1279264214046823e-06,
      "loss": 0.5388,
      "step": 820
    },
    {
      "epoch": 1.8730606488011283,
      "grad_norm": 0.5357053149375428,
      "learning_rate": 2.086120401337793e-06,
      "loss": 0.5507,
      "step": 830
    },
    {
      "epoch": 1.8956276445698168,
      "grad_norm": 0.722556948186832,
      "learning_rate": 2.044314381270903e-06,
      "loss": 0.5508,
      "step": 840
    },
    {
      "epoch": 1.918194640338505,
      "grad_norm": 0.5648614764610744,
      "learning_rate": 2.0025083612040134e-06,
      "loss": 0.5383,
      "step": 850
    },
    {
      "epoch": 1.9407616361071933,
      "grad_norm": 0.5449086056577778,
      "learning_rate": 1.960702341137124e-06,
      "loss": 0.5434,
      "step": 860
    },
    {
      "epoch": 1.9633286318758816,
      "grad_norm": 0.5439248073291871,
      "learning_rate": 1.9188963210702343e-06,
      "loss": 0.5501,
      "step": 870
    },
    {
      "epoch": 1.9858956276445698,
      "grad_norm": 0.5668278705849349,
      "learning_rate": 1.8770903010033448e-06,
      "loss": 0.5476,
      "step": 880
    },
    {
      "epoch": 1.9994358251057829,
      "eval_loss": 0.5841257572174072,
      "eval_runtime": 685.8839,
      "eval_samples_per_second": 17.41,
      "eval_steps_per_second": 0.545,
      "step": 886
    },
    {
      "epoch": 2.008462623413258,
      "grad_norm": 0.8419949483787798,
      "learning_rate": 1.8352842809364548e-06,
      "loss": 0.5793,
      "step": 890
    },
    {
      "epoch": 2.0310296191819464,
      "grad_norm": 0.5856594136748535,
      "learning_rate": 1.7934782608695653e-06,
      "loss": 0.5034,
      "step": 900
    },
    {
      "epoch": 2.0535966149506346,
      "grad_norm": 0.5827928398358422,
      "learning_rate": 1.7516722408026757e-06,
      "loss": 0.5061,
      "step": 910
    },
    {
      "epoch": 2.076163610719323,
      "grad_norm": 0.5599155876387997,
      "learning_rate": 1.709866220735786e-06,
      "loss": 0.4983,
      "step": 920
    },
    {
      "epoch": 2.098730606488011,
      "grad_norm": 0.5470260712158903,
      "learning_rate": 1.6680602006688964e-06,
      "loss": 0.5052,
      "step": 930
    },
    {
      "epoch": 2.1212976022566994,
      "grad_norm": 0.5758776710250274,
      "learning_rate": 1.6262541806020069e-06,
      "loss": 0.5089,
      "step": 940
    },
    {
      "epoch": 2.143864598025388,
      "grad_norm": 0.5755610731770497,
      "learning_rate": 1.5844481605351173e-06,
      "loss": 0.5099,
      "step": 950
    },
    {
      "epoch": 2.1664315937940763,
      "grad_norm": 0.5688878225446357,
      "learning_rate": 1.5426421404682276e-06,
      "loss": 0.5059,
      "step": 960
    },
    {
      "epoch": 2.1889985895627646,
      "grad_norm": 0.5484653799292355,
      "learning_rate": 1.500836120401338e-06,
      "loss": 0.5027,
      "step": 970
    },
    {
      "epoch": 2.211565585331453,
      "grad_norm": 0.5855753779073777,
      "learning_rate": 1.4590301003344485e-06,
      "loss": 0.497,
      "step": 980
    },
    {
      "epoch": 2.234132581100141,
      "grad_norm": 0.5846394849624436,
      "learning_rate": 1.4172240802675585e-06,
      "loss": 0.4967,
      "step": 990
    },
    {
      "epoch": 2.2566995768688294,
      "grad_norm": 0.6020064410941347,
      "learning_rate": 1.375418060200669e-06,
      "loss": 0.5061,
      "step": 1000
    },
    {
      "epoch": 2.2792665726375176,
      "grad_norm": 0.5674319171403842,
      "learning_rate": 1.3336120401337794e-06,
      "loss": 0.5119,
      "step": 1010
    },
    {
      "epoch": 2.301833568406206,
      "grad_norm": 0.5608300219896133,
      "learning_rate": 1.2918060200668896e-06,
      "loss": 0.5057,
      "step": 1020
    },
    {
      "epoch": 2.324400564174894,
      "grad_norm": 0.5902005590764662,
      "learning_rate": 1.25e-06,
      "loss": 0.5096,
      "step": 1030
    },
    {
      "epoch": 2.3469675599435824,
      "grad_norm": 0.546284085451993,
      "learning_rate": 1.2081939799331106e-06,
      "loss": 0.5069,
      "step": 1040
    },
    {
      "epoch": 2.3695345557122707,
      "grad_norm": 0.5498516301636015,
      "learning_rate": 1.1663879598662208e-06,
      "loss": 0.5041,
      "step": 1050
    },
    {
      "epoch": 2.392101551480959,
      "grad_norm": 0.5351775325473189,
      "learning_rate": 1.124581939799331e-06,
      "loss": 0.5032,
      "step": 1060
    },
    {
      "epoch": 2.414668547249647,
      "grad_norm": 0.5483556296524353,
      "learning_rate": 1.0827759197324415e-06,
      "loss": 0.5035,
      "step": 1070
    },
    {
      "epoch": 2.4372355430183354,
      "grad_norm": 0.5646775870143284,
      "learning_rate": 1.040969899665552e-06,
      "loss": 0.5069,
      "step": 1080
    },
    {
      "epoch": 2.459802538787024,
      "grad_norm": 0.5302435040525055,
      "learning_rate": 9.991638795986624e-07,
      "loss": 0.505,
      "step": 1090
    },
    {
      "epoch": 2.4823695345557124,
      "grad_norm": 0.5774882345635233,
      "learning_rate": 9.573578595317726e-07,
      "loss": 0.5074,
      "step": 1100
    },
    {
      "epoch": 2.5049365303244007,
      "grad_norm": 0.5389744005535605,
      "learning_rate": 9.15551839464883e-07,
      "loss": 0.5107,
      "step": 1110
    },
    {
      "epoch": 2.527503526093089,
      "grad_norm": 0.5359062536764768,
      "learning_rate": 8.737458193979933e-07,
      "loss": 0.5001,
      "step": 1120
    },
    {
      "epoch": 2.550070521861777,
      "grad_norm": 0.5329301854453081,
      "learning_rate": 8.319397993311038e-07,
      "loss": 0.5053,
      "step": 1130
    },
    {
      "epoch": 2.5726375176304654,
      "grad_norm": 0.5469111484464729,
      "learning_rate": 7.901337792642141e-07,
      "loss": 0.5034,
      "step": 1140
    },
    {
      "epoch": 2.5952045133991537,
      "grad_norm": 0.5376793002418263,
      "learning_rate": 7.483277591973246e-07,
      "loss": 0.5038,
      "step": 1150
    },
    {
      "epoch": 2.617771509167842,
      "grad_norm": 0.5321398147289212,
      "learning_rate": 7.065217391304348e-07,
      "loss": 0.5072,
      "step": 1160
    },
    {
      "epoch": 2.64033850493653,
      "grad_norm": 0.5325680399356436,
      "learning_rate": 6.647157190635452e-07,
      "loss": 0.4955,
      "step": 1170
    },
    {
      "epoch": 2.6629055007052185,
      "grad_norm": 0.6374259282175002,
      "learning_rate": 6.229096989966555e-07,
      "loss": 0.5066,
      "step": 1180
    },
    {
      "epoch": 2.685472496473907,
      "grad_norm": 0.5462731372886629,
      "learning_rate": 5.81103678929766e-07,
      "loss": 0.5071,
      "step": 1190
    },
    {
      "epoch": 2.7080394922425954,
      "grad_norm": 0.5385696882283996,
      "learning_rate": 5.392976588628763e-07,
      "loss": 0.5054,
      "step": 1200
    },
    {
      "epoch": 2.7306064880112837,
      "grad_norm": 0.5456766485564613,
      "learning_rate": 4.974916387959867e-07,
      "loss": 0.5061,
      "step": 1210
    },
    {
      "epoch": 2.753173483779972,
      "grad_norm": 0.5339029579122799,
      "learning_rate": 4.55685618729097e-07,
      "loss": 0.5024,
      "step": 1220
    },
    {
      "epoch": 2.77574047954866,
      "grad_norm": 0.5252848686524678,
      "learning_rate": 4.138795986622074e-07,
      "loss": 0.5092,
      "step": 1230
    },
    {
      "epoch": 2.7983074753173485,
      "grad_norm": 0.5250218608560205,
      "learning_rate": 3.7207357859531776e-07,
      "loss": 0.5074,
      "step": 1240
    },
    {
      "epoch": 2.8208744710860367,
      "grad_norm": 0.5267485139063955,
      "learning_rate": 3.302675585284281e-07,
      "loss": 0.507,
      "step": 1250
    },
    {
      "epoch": 2.843441466854725,
      "grad_norm": 0.5433190534773383,
      "learning_rate": 2.884615384615385e-07,
      "loss": 0.5085,
      "step": 1260
    },
    {
      "epoch": 2.8660084626234132,
      "grad_norm": 0.5174335750907513,
      "learning_rate": 2.4665551839464886e-07,
      "loss": 0.5057,
      "step": 1270
    },
    {
      "epoch": 2.8885754583921015,
      "grad_norm": 0.5152650038198892,
      "learning_rate": 2.048494983277592e-07,
      "loss": 0.503,
      "step": 1280
    },
    {
      "epoch": 2.9111424541607898,
      "grad_norm": 0.5238413421157163,
      "learning_rate": 1.6304347826086958e-07,
      "loss": 0.507,
      "step": 1290
    },
    {
      "epoch": 2.933709449929478,
      "grad_norm": 0.5532403905639468,
      "learning_rate": 1.2123745819397995e-07,
      "loss": 0.502,
      "step": 1300
    },
    {
      "epoch": 2.9562764456981663,
      "grad_norm": 0.5439654700776371,
      "learning_rate": 7.943143812709031e-08,
      "loss": 0.506,
      "step": 1310
    },
    {
      "epoch": 2.9788434414668545,
      "grad_norm": 0.5300109389569346,
      "learning_rate": 3.762541806020067e-08,
      "loss": 0.5015,
      "step": 1320
    },
    {
      "epoch": 2.9991537376586743,
      "eval_loss": 0.5893023610115051,
      "eval_runtime": 684.3576,
      "eval_samples_per_second": 17.448,
      "eval_steps_per_second": 0.546,
      "step": 1329
    },
    {
      "epoch": 2.9991537376586743,
      "step": 1329,
      "total_flos": 5064195066298368.0,
      "train_loss": 0.5616077652006243,
      "train_runtime": 120588.8327,
      "train_samples_per_second": 5.644,
      "train_steps_per_second": 0.011
    }
  ],
  "logging_steps": 10,
  "max_steps": 1329,
  "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": 5064195066298368.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}