{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1920,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.015625,
      "grad_norm": 10.444775818294296,
      "learning_rate": 5.208333333333334e-07,
      "loss": 0.8468,
      "step": 10
    },
    {
      "epoch": 0.03125,
      "grad_norm": 3.3931220262488844,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 0.7571,
      "step": 20
    },
    {
      "epoch": 0.046875,
      "grad_norm": 1.725977917785408,
      "learning_rate": 1.5625e-06,
      "loss": 0.6887,
      "step": 30
    },
    {
      "epoch": 0.0625,
      "grad_norm": 1.5579846393216565,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.6419,
      "step": 40
    },
    {
      "epoch": 0.078125,
      "grad_norm": 1.7208975007176577,
      "learning_rate": 2.604166666666667e-06,
      "loss": 0.6117,
      "step": 50
    },
    {
      "epoch": 0.09375,
      "grad_norm": 1.7694674612194612,
      "learning_rate": 3.125e-06,
      "loss": 0.5912,
      "step": 60
    },
    {
      "epoch": 0.109375,
      "grad_norm": 2.038979400755539,
      "learning_rate": 3.6458333333333333e-06,
      "loss": 0.5743,
      "step": 70
    },
    {
      "epoch": 0.125,
      "grad_norm": 1.758888705110013,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.5807,
      "step": 80
    },
    {
      "epoch": 0.140625,
      "grad_norm": 1.747787322446748,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.561,
      "step": 90
    },
    {
      "epoch": 0.15625,
      "grad_norm": 2.023155826824772,
      "learning_rate": 4.999946602507342e-06,
      "loss": 0.5596,
      "step": 100
    },
    {
      "epoch": 0.171875,
      "grad_norm": 2.7065268999883747,
      "learning_rate": 4.999345909821495e-06,
      "loss": 0.5426,
      "step": 110
    },
    {
      "epoch": 0.1875,
      "grad_norm": 3.0442036812906323,
      "learning_rate": 4.998077956371428e-06,
      "loss": 0.5514,
      "step": 120
    },
    {
      "epoch": 0.203125,
      "grad_norm": 2.569524644395966,
      "learning_rate": 4.996143118291191e-06,
      "loss": 0.5499,
      "step": 130
    },
    {
      "epoch": 0.21875,
      "grad_norm": 2.4274788708639226,
      "learning_rate": 4.993541969543877e-06,
      "loss": 0.535,
      "step": 140
    },
    {
      "epoch": 0.234375,
      "grad_norm": 2.242979877814027,
      "learning_rate": 4.990275281751359e-06,
      "loss": 0.5295,
      "step": 150
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.085717253543815,
      "learning_rate": 4.9863440239653865e-06,
      "loss": 0.5394,
      "step": 160
    },
    {
      "epoch": 0.265625,
      "grad_norm": 1.980124498533203,
      "learning_rate": 4.981749362380126e-06,
      "loss": 0.5318,
      "step": 170
    },
    {
      "epoch": 0.28125,
      "grad_norm": 1.4291512505399062,
      "learning_rate": 4.976492659986207e-06,
      "loss": 0.5273,
      "step": 180
    },
    {
      "epoch": 0.296875,
      "grad_norm": 2.1782127342393456,
      "learning_rate": 4.9705754761663995e-06,
      "loss": 0.5262,
      "step": 190
    },
    {
      "epoch": 0.3125,
      "grad_norm": 1.8726553071551566,
      "learning_rate": 4.9639995662330295e-06,
      "loss": 0.5156,
      "step": 200
    },
    {
      "epoch": 0.328125,
      "grad_norm": 1.7397260041367604,
      "learning_rate": 4.956766880907269e-06,
      "loss": 0.5138,
      "step": 210
    },
    {
      "epoch": 0.34375,
      "grad_norm": 1.9152193087219476,
      "learning_rate": 4.948879565740459e-06,
      "loss": 0.5117,
      "step": 220
    },
    {
      "epoch": 0.359375,
      "grad_norm": 1.909800181545602,
      "learning_rate": 4.940339960477644e-06,
      "loss": 0.5158,
      "step": 230
    },
    {
      "epoch": 0.375,
      "grad_norm": 2.077990164198304,
      "learning_rate": 4.931150598363494e-06,
      "loss": 0.5056,
      "step": 240
    },
    {
      "epoch": 0.390625,
      "grad_norm": 1.5538564815058695,
      "learning_rate": 4.921314205390822e-06,
      "loss": 0.5059,
      "step": 250
    },
    {
      "epoch": 0.40625,
      "grad_norm": 1.8861482326610306,
      "learning_rate": 4.910833699491932e-06,
      "loss": 0.4997,
      "step": 260
    },
    {
      "epoch": 0.421875,
      "grad_norm": 2.22030396642517,
      "learning_rate": 4.899712189673023e-06,
      "loss": 0.5034,
      "step": 270
    },
    {
      "epoch": 0.4375,
      "grad_norm": 2.1138662624316003,
      "learning_rate": 4.8879529750919105e-06,
      "loss": 0.5035,
      "step": 280
    },
    {
      "epoch": 0.453125,
      "grad_norm": 3.704875355592209,
      "learning_rate": 4.875559544079348e-06,
      "loss": 0.5024,
      "step": 290
    },
    {
      "epoch": 0.46875,
      "grad_norm": 2.4124963571250917,
      "learning_rate": 4.862535573104217e-06,
      "loss": 0.49,
      "step": 300
    },
    {
      "epoch": 0.484375,
      "grad_norm": 2.6383521452329703,
      "learning_rate": 4.848884925682926e-06,
      "loss": 0.494,
      "step": 310
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.7798577923861036,
      "learning_rate": 4.8346116512333045e-06,
      "loss": 0.4978,
      "step": 320
    },
    {
      "epoch": 0.515625,
      "grad_norm": 2.5016539728327105,
      "learning_rate": 4.819719983873357e-06,
      "loss": 0.4959,
      "step": 330
    },
    {
      "epoch": 0.53125,
      "grad_norm": 2.3486985533520395,
      "learning_rate": 4.804214341165228e-06,
      "loss": 0.4933,
      "step": 340
    },
    {
      "epoch": 0.546875,
      "grad_norm": 2.1224863685900113,
      "learning_rate": 4.788099322804749e-06,
      "loss": 0.4835,
      "step": 350
    },
    {
      "epoch": 0.5625,
      "grad_norm": 2.1137649633214473,
      "learning_rate": 4.771379709256953e-06,
      "loss": 0.4899,
      "step": 360
    },
    {
      "epoch": 0.578125,
      "grad_norm": 1.8070271600662309,
      "learning_rate": 4.754060460337963e-06,
      "loss": 0.4904,
      "step": 370
    },
    {
      "epoch": 0.59375,
      "grad_norm": 2.229961084900482,
      "learning_rate": 4.7361467137436846e-06,
      "loss": 0.4909,
      "step": 380
    },
    {
      "epoch": 0.609375,
      "grad_norm": 2.5450078236200255,
      "learning_rate": 4.7176437835257225e-06,
      "loss": 0.4878,
      "step": 390
    },
    {
      "epoch": 0.625,
      "grad_norm": 1.6490727233459694,
      "learning_rate": 4.698557158514988e-06,
      "loss": 0.5006,
      "step": 400
    },
    {
      "epoch": 0.640625,
      "grad_norm": 1.5461965981797419,
      "learning_rate": 4.678892500693451e-06,
      "loss": 0.493,
      "step": 410
    },
    {
      "epoch": 0.65625,
      "grad_norm": 1.5503545160303023,
      "learning_rate": 4.6586556435145404e-06,
      "loss": 0.4815,
      "step": 420
    },
    {
      "epoch": 0.671875,
      "grad_norm": 1.4158769400928066,
      "learning_rate": 4.637852590172665e-06,
      "loss": 0.4936,
      "step": 430
    },
    {
      "epoch": 0.6875,
      "grad_norm": 1.4477782106737869,
      "learning_rate": 4.616489511822384e-06,
      "loss": 0.4885,
      "step": 440
    },
    {
      "epoch": 0.703125,
      "grad_norm": 1.6879117554810668,
      "learning_rate": 4.59457274574776e-06,
      "loss": 0.4937,
      "step": 450
    },
    {
      "epoch": 0.71875,
      "grad_norm": 1.5561703988248663,
      "learning_rate": 4.572108793482425e-06,
      "loss": 0.4834,
      "step": 460
    },
    {
      "epoch": 0.734375,
      "grad_norm": 1.4814626689757864,
      "learning_rate": 4.549104318880919e-06,
      "loss": 0.4955,
      "step": 470
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3493469423874893,
      "learning_rate": 4.525566146141886e-06,
      "loss": 0.4839,
      "step": 480
    },
    {
      "epoch": 0.765625,
      "grad_norm": 1.5247306592093048,
      "learning_rate": 4.501501257783692e-06,
      "loss": 0.482,
      "step": 490
    },
    {
      "epoch": 0.78125,
      "grad_norm": 1.8021809188964066,
      "learning_rate": 4.47691679257309e-06,
      "loss": 0.4758,
      "step": 500
    },
    {
      "epoch": 0.796875,
      "grad_norm": 1.3217160484942474,
      "learning_rate": 4.451820043407527e-06,
      "loss": 0.4788,
      "step": 510
    },
    {
      "epoch": 0.8125,
      "grad_norm": 1.5947481734921138,
      "learning_rate": 4.426218455151733e-06,
      "loss": 0.4722,
      "step": 520
    },
    {
      "epoch": 0.828125,
      "grad_norm": 1.3218456009113562,
      "learning_rate": 4.400119622429226e-06,
      "loss": 0.4792,
      "step": 530
    },
    {
      "epoch": 0.84375,
      "grad_norm": 1.4119902047840658,
      "learning_rate": 4.3735312873693905e-06,
      "loss": 0.4836,
      "step": 540
    },
    {
      "epoch": 0.859375,
      "grad_norm": 1.2667791953694767,
      "learning_rate": 4.346461337310805e-06,
      "loss": 0.4827,
      "step": 550
    },
    {
      "epoch": 0.875,
      "grad_norm": 1.9380616838070583,
      "learning_rate": 4.31891780246149e-06,
      "loss": 0.4798,
      "step": 560
    },
    {
      "epoch": 0.890625,
      "grad_norm": 1.906273428173042,
      "learning_rate": 4.290908853516771e-06,
      "loss": 0.4764,
      "step": 570
    },
    {
      "epoch": 0.90625,
      "grad_norm": 2.1397738699102526,
      "learning_rate": 4.262442799235474e-06,
      "loss": 0.4698,
      "step": 580
    },
    {
      "epoch": 0.921875,
      "grad_norm": 1.4439475222238531,
      "learning_rate": 4.233528083975155e-06,
      "loss": 0.4702,
      "step": 590
    },
    {
      "epoch": 0.9375,
      "grad_norm": 1.7063789221261367,
      "learning_rate": 4.204173285187117e-06,
      "loss": 0.466,
      "step": 600
    },
    {
      "epoch": 0.953125,
      "grad_norm": 1.5514623726199628,
      "learning_rate": 4.1743871108719334e-06,
      "loss": 0.4696,
      "step": 610
    },
    {
      "epoch": 0.96875,
      "grad_norm": 1.4916007625639955,
      "learning_rate": 4.1441783969962506e-06,
      "loss": 0.4708,
      "step": 620
    },
    {
      "epoch": 0.984375,
      "grad_norm": 2.36511645668308,
      "learning_rate": 4.113556104871631e-06,
      "loss": 0.4684,
      "step": 630
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6210452179231138,
      "learning_rate": 4.082529318496206e-06,
      "loss": 0.4706,
      "step": 640
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.4656927287578583,
      "eval_runtime": 59.1544,
      "eval_samples_per_second": 291.441,
      "eval_steps_per_second": 1.15,
      "step": 640
    },
    {
      "epoch": 1.015625,
      "grad_norm": 1.8958635048318864,
      "learning_rate": 4.051107241859944e-06,
      "loss": 0.3837,
      "step": 650
    },
    {
      "epoch": 1.03125,
      "grad_norm": 1.4711121127809,
      "learning_rate": 4.019299196214315e-06,
      "loss": 0.3803,
      "step": 660
    },
    {
      "epoch": 1.046875,
      "grad_norm": 1.389098603274721,
      "learning_rate": 3.987114617307176e-06,
      "loss": 0.3742,
      "step": 670
    },
    {
      "epoch": 1.0625,
      "grad_norm": 1.4934816227614662,
      "learning_rate": 3.954563052583687e-06,
      "loss": 0.3877,
      "step": 680
    },
    {
      "epoch": 1.078125,
      "grad_norm": 2.191093614414856,
      "learning_rate": 3.92165415835409e-06,
      "loss": 0.3805,
      "step": 690
    },
    {
      "epoch": 1.09375,
      "grad_norm": 2.0750682076780134,
      "learning_rate": 3.888397696929208e-06,
      "loss": 0.377,
      "step": 700
    },
    {
      "epoch": 1.109375,
      "grad_norm": 2.0456715879045366,
      "learning_rate": 3.854803533724477e-06,
      "loss": 0.378,
      "step": 710
    },
    {
      "epoch": 1.125,
      "grad_norm": 1.9908156741899143,
      "learning_rate": 3.820881634333416e-06,
      "loss": 0.3815,
      "step": 720
    },
    {
      "epoch": 1.140625,
      "grad_norm": 1.434003143563507,
      "learning_rate": 3.786642061571356e-06,
      "loss": 0.3801,
      "step": 730
    },
    {
      "epoch": 1.15625,
      "grad_norm": 1.922816409585947,
      "learning_rate": 3.752094972490348e-06,
      "loss": 0.3774,
      "step": 740
    },
    {
      "epoch": 1.171875,
      "grad_norm": 1.2664235758845404,
      "learning_rate": 3.717250615366108e-06,
      "loss": 0.3821,
      "step": 750
    },
    {
      "epoch": 1.1875,
      "grad_norm": 1.4529409619288622,
      "learning_rate": 3.6821193266578976e-06,
      "loss": 0.3806,
      "step": 760
    },
    {
      "epoch": 1.203125,
      "grad_norm": 1.4048451043641628,
      "learning_rate": 3.6467115279422523e-06,
      "loss": 0.3768,
      "step": 770
    },
    {
      "epoch": 1.21875,
      "grad_norm": 1.5780951905511307,
      "learning_rate": 3.611037722821452e-06,
      "loss": 0.3849,
      "step": 780
    },
    {
      "epoch": 1.234375,
      "grad_norm": 1.3754990978690818,
      "learning_rate": 3.5751084938076697e-06,
      "loss": 0.3819,
      "step": 790
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.7832805601394666,
      "learning_rate": 3.5389344991836977e-06,
      "loss": 0.3815,
      "step": 800
    },
    {
      "epoch": 1.265625,
      "grad_norm": 1.5731803619002744,
      "learning_rate": 3.5025264698412127e-06,
      "loss": 0.3769,
      "step": 810
    },
    {
      "epoch": 1.28125,
      "grad_norm": 1.4508731895019338,
      "learning_rate": 3.4658952060974858e-06,
      "loss": 0.386,
      "step": 820
    },
    {
      "epoch": 1.296875,
      "grad_norm": 1.3310110347445678,
      "learning_rate": 3.4290515744915135e-06,
      "loss": 0.3793,
      "step": 830
    },
    {
      "epoch": 1.3125,
      "grad_norm": 1.5870629040014994,
      "learning_rate": 3.3920065045604874e-06,
      "loss": 0.3815,
      "step": 840
    },
    {
      "epoch": 1.328125,
      "grad_norm": 1.4428244899236522,
      "learning_rate": 3.3547709855975908e-06,
      "loss": 0.3812,
      "step": 850
    },
    {
      "epoch": 1.34375,
      "grad_norm": 1.4623661323810098,
      "learning_rate": 3.317356063392059e-06,
      "loss": 0.3759,
      "step": 860
    },
    {
      "epoch": 1.359375,
      "grad_norm": 1.4541598774189346,
      "learning_rate": 3.2797728369524878e-06,
      "loss": 0.3863,
      "step": 870
    },
    {
      "epoch": 1.375,
      "grad_norm": 1.7531641958321784,
      "learning_rate": 3.242032455214346e-06,
      "loss": 0.3825,
      "step": 880
    },
    {
      "epoch": 1.390625,
      "grad_norm": 2.0248369297536164,
      "learning_rate": 3.2041461137326825e-06,
      "loss": 0.3808,
      "step": 890
    },
    {
      "epoch": 1.40625,
      "grad_norm": 1.7114859453687477,
      "learning_rate": 3.166125051361007e-06,
      "loss": 0.3813,
      "step": 900
    },
    {
      "epoch": 1.421875,
      "grad_norm": 1.5182934800219041,
      "learning_rate": 3.127980546917318e-06,
      "loss": 0.3749,
      "step": 910
    },
    {
      "epoch": 1.4375,
      "grad_norm": 1.2810157768825081,
      "learning_rate": 3.089723915838283e-06,
      "loss": 0.3798,
      "step": 920
    },
    {
      "epoch": 1.453125,
      "grad_norm": 1.3056749068330433,
      "learning_rate": 3.051366506822554e-06,
      "loss": 0.3824,
      "step": 930
    },
    {
      "epoch": 1.46875,
      "grad_norm": 1.339776576173714,
      "learning_rate": 3.0129196984642084e-06,
      "loss": 0.3819,
      "step": 940
    },
    {
      "epoch": 1.484375,
      "grad_norm": 1.3275514822294423,
      "learning_rate": 2.9743948958773373e-06,
      "loss": 0.3804,
      "step": 950
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3861961679351784,
      "learning_rate": 2.9358035273127484e-06,
      "loss": 0.3876,
      "step": 960
    },
    {
      "epoch": 1.515625,
      "grad_norm": 1.4583205842845002,
      "learning_rate": 2.8971570407678222e-06,
      "loss": 0.3811,
      "step": 970
    },
    {
      "epoch": 1.53125,
      "grad_norm": 1.5523942903859895,
      "learning_rate": 2.8584669005904993e-06,
      "loss": 0.3764,
      "step": 980
    },
    {
      "epoch": 1.546875,
      "grad_norm": 1.365188486045966,
      "learning_rate": 2.8197445840784173e-06,
      "loss": 0.3706,
      "step": 990
    },
    {
      "epoch": 1.5625,
      "grad_norm": 1.5063105750252836,
      "learning_rate": 2.781001578074217e-06,
      "loss": 0.3806,
      "step": 1000
    },
    {
      "epoch": 1.578125,
      "grad_norm": 1.4218424948694481,
      "learning_rate": 2.7422493755580052e-06,
      "loss": 0.3787,
      "step": 1010
    },
    {
      "epoch": 1.59375,
      "grad_norm": 1.263189980212751,
      "learning_rate": 2.703499472238004e-06,
      "loss": 0.3726,
      "step": 1020
    },
    {
      "epoch": 1.609375,
      "grad_norm": 1.3846905506410119,
      "learning_rate": 2.664763363140382e-06,
      "loss": 0.3814,
      "step": 1030
    },
    {
      "epoch": 1.625,
      "grad_norm": 1.2445425042616887,
      "learning_rate": 2.6260525391993027e-06,
      "loss": 0.3759,
      "step": 1040
    },
    {
      "epoch": 1.640625,
      "grad_norm": 1.3055007938596355,
      "learning_rate": 2.5873784838481764e-06,
      "loss": 0.3762,
      "step": 1050
    },
    {
      "epoch": 1.65625,
      "grad_norm": 1.3144625349610841,
      "learning_rate": 2.548752669613132e-06,
      "loss": 0.3796,
      "step": 1060
    },
    {
      "epoch": 1.671875,
      "grad_norm": 1.3350371737116375,
      "learning_rate": 2.510186554709741e-06,
      "loss": 0.3815,
      "step": 1070
    },
    {
      "epoch": 1.6875,
      "grad_norm": 1.566780577726475,
      "learning_rate": 2.471691579643968e-06,
      "loss": 0.3708,
      "step": 1080
    },
    {
      "epoch": 1.703125,
      "grad_norm": 1.2771965688411218,
      "learning_rate": 2.4332791638183935e-06,
      "loss": 0.3741,
      "step": 1090
    },
    {
      "epoch": 1.71875,
      "grad_norm": 1.3207787835503493,
      "learning_rate": 2.3949607021446774e-06,
      "loss": 0.3775,
      "step": 1100
    },
    {
      "epoch": 1.734375,
      "grad_norm": 1.263046733536951,
      "learning_rate": 2.3567475616633046e-06,
      "loss": 0.3742,
      "step": 1110
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.3161682150839322,
      "learning_rate": 2.318651078171589e-06,
      "loss": 0.3771,
      "step": 1120
    },
    {
      "epoch": 1.765625,
      "grad_norm": 1.2308341651874004,
      "learning_rate": 2.2806825528609457e-06,
      "loss": 0.3788,
      "step": 1130
    },
    {
      "epoch": 1.78125,
      "grad_norm": 1.3859163680872322,
      "learning_rate": 2.2428532489644368e-06,
      "loss": 0.3718,
      "step": 1140
    },
    {
      "epoch": 1.796875,
      "grad_norm": 1.389479577560998,
      "learning_rate": 2.2051743884155636e-06,
      "loss": 0.3811,
      "step": 1150
    },
    {
      "epoch": 1.8125,
      "grad_norm": 1.179457462552155,
      "learning_rate": 2.167657148519328e-06,
      "loss": 0.3709,
      "step": 1160
    },
    {
      "epoch": 1.828125,
      "grad_norm": 1.1984315597042852,
      "learning_rate": 2.1303126586365174e-06,
      "loss": 0.3746,
      "step": 1170
    },
    {
      "epoch": 1.84375,
      "grad_norm": 1.3778930599081922,
      "learning_rate": 2.093151996882217e-06,
      "loss": 0.3701,
      "step": 1180
    },
    {
      "epoch": 1.859375,
      "grad_norm": 1.393001774481534,
      "learning_rate": 2.0561861868395303e-06,
      "loss": 0.3777,
      "step": 1190
    },
    {
      "epoch": 1.875,
      "grad_norm": 1.255228122839463,
      "learning_rate": 2.0194261942894625e-06,
      "loss": 0.3723,
      "step": 1200
    },
    {
      "epoch": 1.890625,
      "grad_norm": 1.3414771695904,
      "learning_rate": 1.982882923957969e-06,
      "loss": 0.3788,
      "step": 1210
    },
    {
      "epoch": 1.90625,
      "grad_norm": 1.3315196766376038,
      "learning_rate": 1.9465672162811004e-06,
      "loss": 0.3749,
      "step": 1220
    },
    {
      "epoch": 1.921875,
      "grad_norm": 1.196593458985567,
      "learning_rate": 1.9104898441892223e-06,
      "loss": 0.3699,
      "step": 1230
    },
    {
      "epoch": 1.9375,
      "grad_norm": 1.2207357045658285,
      "learning_rate": 1.8746615099112667e-06,
      "loss": 0.3675,
      "step": 1240
    },
    {
      "epoch": 1.953125,
      "grad_norm": 1.2040449871674417,
      "learning_rate": 1.8390928417999492e-06,
      "loss": 0.3738,
      "step": 1250
    },
    {
      "epoch": 1.96875,
      "grad_norm": 1.506663538304499,
      "learning_rate": 1.803794391178908e-06,
      "loss": 0.3733,
      "step": 1260
    },
    {
      "epoch": 1.984375,
      "grad_norm": 1.3210143508850016,
      "learning_rate": 1.7687766292126865e-06,
      "loss": 0.3742,
      "step": 1270
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3003812635061929,
      "learning_rate": 1.7340499438004995e-06,
      "loss": 0.369,
      "step": 1280
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.4540603756904602,
      "eval_runtime": 68.5046,
      "eval_samples_per_second": 251.662,
      "eval_steps_per_second": 0.993,
      "step": 1280
    },
    {
      "epoch": 2.015625,
      "grad_norm": 1.799529367757516,
      "learning_rate": 1.6996246364946986e-06,
      "loss": 0.2851,
      "step": 1290
    },
    {
      "epoch": 2.03125,
      "grad_norm": 1.4376838261754665,
      "learning_rate": 1.665510919444851e-06,
      "loss": 0.2857,
      "step": 1300
    },
    {
      "epoch": 2.046875,
      "grad_norm": 1.3710510542531367,
      "learning_rate": 1.6317189123683428e-06,
      "loss": 0.2876,
      "step": 1310
    },
    {
      "epoch": 2.0625,
      "grad_norm": 1.4736988163726463,
      "learning_rate": 1.5982586395483984e-06,
      "loss": 0.2851,
      "step": 1320
    },
    {
      "epoch": 2.078125,
      "grad_norm": 1.3865134466434497,
      "learning_rate": 1.5651400268604063e-06,
      "loss": 0.2825,
      "step": 1330
    },
    {
      "epoch": 2.09375,
      "grad_norm": 1.3685250396007145,
      "learning_rate": 1.5323728988274513e-06,
      "loss": 0.2823,
      "step": 1340
    },
    {
      "epoch": 2.109375,
      "grad_norm": 1.3255774882469817,
      "learning_rate": 1.4999669757058956e-06,
      "loss": 0.2829,
      "step": 1350
    },
    {
      "epoch": 2.125,
      "grad_norm": 1.3462861294111625,
      "learning_rate": 1.4679318706019011e-06,
      "loss": 0.2861,
      "step": 1360
    },
    {
      "epoch": 2.140625,
      "grad_norm": 1.2761171662988255,
      "learning_rate": 1.4362770866197365e-06,
      "loss": 0.2798,
      "step": 1370
    },
    {
      "epoch": 2.15625,
      "grad_norm": 1.2996766419653487,
      "learning_rate": 1.405012014042708e-06,
      "loss": 0.2831,
      "step": 1380
    },
    {
      "epoch": 2.171875,
      "grad_norm": 1.2756611803725677,
      "learning_rate": 1.3741459275475718e-06,
      "loss": 0.2842,
      "step": 1390
    },
    {
      "epoch": 2.1875,
      "grad_norm": 1.3788917408268033,
      "learning_rate": 1.3436879834532237e-06,
      "loss": 0.2883,
      "step": 1400
    },
    {
      "epoch": 2.203125,
      "grad_norm": 1.3457810994006327,
      "learning_rate": 1.3136472170045173e-06,
      "loss": 0.2796,
      "step": 1410
    },
    {
      "epoch": 2.21875,
      "grad_norm": 1.3527803102732727,
      "learning_rate": 1.2840325396919851e-06,
      "loss": 0.2809,
      "step": 1420
    },
    {
      "epoch": 2.234375,
      "grad_norm": 1.3139656273448967,
      "learning_rate": 1.2548527366082746e-06,
      "loss": 0.2899,
      "step": 1430
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3171500244547871,
      "learning_rate": 1.226116463842083e-06,
      "loss": 0.2816,
      "step": 1440
    },
    {
      "epoch": 2.265625,
      "grad_norm": 1.321106017997469,
      "learning_rate": 1.1978322459103558e-06,
      "loss": 0.2866,
      "step": 1450
    },
    {
      "epoch": 2.28125,
      "grad_norm": 1.3880168746621624,
      "learning_rate": 1.1700084732295222e-06,
      "loss": 0.2789,
      "step": 1460
    },
    {
      "epoch": 2.296875,
      "grad_norm": 1.297165181118468,
      "learning_rate": 1.1426533996265006e-06,
      "loss": 0.2818,
      "step": 1470
    },
    {
      "epoch": 2.3125,
      "grad_norm": 1.2914280928037412,
      "learning_rate": 1.1157751398902338e-06,
      "loss": 0.2856,
      "step": 1480
    },
    {
      "epoch": 2.328125,
      "grad_norm": 1.374705202612009,
      "learning_rate": 1.0893816673644643e-06,
      "loss": 0.2836,
      "step": 1490
    },
    {
      "epoch": 2.34375,
      "grad_norm": 1.3661463800498177,
      "learning_rate": 1.0634808115824667e-06,
      "loss": 0.2822,
      "step": 1500
    },
    {
      "epoch": 2.359375,
      "grad_norm": 1.285521847136074,
      "learning_rate": 1.0380802559444444e-06,
      "loss": 0.2788,
      "step": 1510
    },
    {
      "epoch": 2.375,
      "grad_norm": 1.3609559873792854,
      "learning_rate": 1.013187535438278e-06,
      "loss": 0.2815,
      "step": 1520
    },
    {
      "epoch": 2.390625,
      "grad_norm": 1.359498367844339,
      "learning_rate": 9.888100344042926e-07,
      "loss": 0.279,
      "step": 1530
    },
    {
      "epoch": 2.40625,
      "grad_norm": 1.2578896443023726,
      "learning_rate": 9.649549843447213e-07,
      "loss": 0.2837,
      "step": 1540
    },
    {
      "epoch": 2.421875,
      "grad_norm": 1.476169547857346,
      "learning_rate": 9.416294617785033e-07,
      "loss": 0.2846,
      "step": 1550
    },
    {
      "epoch": 2.4375,
      "grad_norm": 1.3700496792517187,
      "learning_rate": 9.188403861420614e-07,
      "loss": 0.2841,
      "step": 1560
    },
    {
      "epoch": 2.453125,
      "grad_norm": 1.278136583382333,
      "learning_rate": 8.965945177366716e-07,
      "loss": 0.2802,
      "step": 1570
    },
    {
      "epoch": 2.46875,
      "grad_norm": 1.321201564334253,
      "learning_rate": 8.74898455723044e-07,
      "loss": 0.2854,
      "step": 1580
    },
    {
      "epoch": 2.484375,
      "grad_norm": 1.3469922385355766,
      "learning_rate": 8.537586361637059e-07,
      "loss": 0.2815,
      "step": 1590
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3453576727439425,
      "learning_rate": 8.331813301137644e-07,
      "loss": 0.2864,
      "step": 1600
    },
    {
      "epoch": 2.515625,
      "grad_norm": 1.2960156031598227,
      "learning_rate": 8.131726417606181e-07,
      "loss": 0.2739,
      "step": 1610
    },
    {
      "epoch": 2.53125,
      "grad_norm": 1.3651173071468319,
      "learning_rate": 7.937385066131745e-07,
      "loss": 0.2795,
      "step": 1620
    },
    {
      "epoch": 2.546875,
      "grad_norm": 1.3217889886043823,
      "learning_rate": 7.748846897410985e-07,
      "loss": 0.2734,
      "step": 1630
    },
    {
      "epoch": 2.5625,
      "grad_norm": 1.3037202509770986,
      "learning_rate": 7.566167840646245e-07,
      "loss": 0.2847,
      "step": 1640
    },
    {
      "epoch": 2.578125,
      "grad_norm": 1.2986387865278517,
      "learning_rate": 7.389402086954368e-07,
      "loss": 0.2948,
      "step": 1650
    },
    {
      "epoch": 2.59375,
      "grad_norm": 1.2549602691888244,
      "learning_rate": 7.218602073291095e-07,
      "loss": 0.2842,
      "step": 1660
    },
    {
      "epoch": 2.609375,
      "grad_norm": 1.2411990452005732,
      "learning_rate": 7.053818466895791e-07,
      "loss": 0.279,
      "step": 1670
    },
    {
      "epoch": 2.625,
      "grad_norm": 1.2694405064326646,
      "learning_rate": 6.895100150261206e-07,
      "loss": 0.2868,
      "step": 1680
    },
    {
      "epoch": 2.640625,
      "grad_norm": 1.3041289401671792,
      "learning_rate": 6.742494206632659e-07,
      "loss": 0.285,
      "step": 1690
    },
    {
      "epoch": 2.65625,
      "grad_norm": 1.2477084403906091,
      "learning_rate": 6.596045906040921e-07,
      "loss": 0.2823,
      "step": 1700
    },
    {
      "epoch": 2.671875,
      "grad_norm": 1.2407685713907555,
      "learning_rate": 6.455798691873041e-07,
      "loss": 0.2845,
      "step": 1710
    },
    {
      "epoch": 2.6875,
      "grad_norm": 1.2726899580839428,
      "learning_rate": 6.32179416798501e-07,
      "loss": 0.2891,
      "step": 1720
    },
    {
      "epoch": 2.703125,
      "grad_norm": 1.2629501555738192,
      "learning_rate": 6.194072086360125e-07,
      "loss": 0.2779,
      "step": 1730
    },
    {
      "epoch": 2.71875,
      "grad_norm": 1.3031032470418795,
      "learning_rate": 6.072670335316676e-07,
      "loss": 0.2793,
      "step": 1740
    },
    {
      "epoch": 2.734375,
      "grad_norm": 1.2682201497592243,
      "learning_rate": 5.957624928268528e-07,
      "loss": 0.2855,
      "step": 1750
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3301507903535867,
      "learning_rate": 5.848969993041866e-07,
      "loss": 0.2788,
      "step": 1760
    },
    {
      "epoch": 2.765625,
      "grad_norm": 1.2869688012601592,
      "learning_rate": 5.746737761751279e-07,
      "loss": 0.2843,
      "step": 1770
    },
    {
      "epoch": 2.78125,
      "grad_norm": 1.304345949854902,
      "learning_rate": 5.650958561238212e-07,
      "loss": 0.2848,
      "step": 1780
    },
    {
      "epoch": 2.796875,
      "grad_norm": 1.2612206521119516,
      "learning_rate": 5.561660804074635e-07,
      "loss": 0.2784,
      "step": 1790
    },
    {
      "epoch": 2.8125,
      "grad_norm": 1.21963172307497,
      "learning_rate": 5.478870980134524e-07,
      "loss": 0.2787,
      "step": 1800
    },
    {
      "epoch": 2.828125,
      "grad_norm": 1.2820291216583481,
      "learning_rate": 5.402613648735749e-07,
      "loss": 0.2827,
      "step": 1810
    },
    {
      "epoch": 2.84375,
      "grad_norm": 1.270150906269018,
      "learning_rate": 5.332911431354621e-07,
      "loss": 0.2753,
      "step": 1820
    },
    {
      "epoch": 2.859375,
      "grad_norm": 1.2373120076036523,
      "learning_rate": 5.269785004915327e-07,
      "loss": 0.2842,
      "step": 1830
    },
    {
      "epoch": 2.875,
      "grad_norm": 1.3257054430640673,
      "learning_rate": 5.213253095656177e-07,
      "loss": 0.2794,
      "step": 1840
    },
    {
      "epoch": 2.890625,
      "grad_norm": 1.2939668010311147,
      "learning_rate": 5.16333247357453e-07,
      "loss": 0.285,
      "step": 1850
    },
    {
      "epoch": 2.90625,
      "grad_norm": 1.2762831271442217,
      "learning_rate": 5.120037947452043e-07,
      "loss": 0.2808,
      "step": 1860
    },
    {
      "epoch": 2.921875,
      "grad_norm": 1.2672944555649501,
      "learning_rate": 5.083382360461682e-07,
      "loss": 0.2828,
      "step": 1870
    },
    {
      "epoch": 2.9375,
      "grad_norm": 1.253289043826509,
      "learning_rate": 5.05337658635785e-07,
      "loss": 0.2859,
      "step": 1880
    },
    {
      "epoch": 2.953125,
      "grad_norm": 1.3369028630930568,
      "learning_rate": 5.030029526250719e-07,
      "loss": 0.2758,
      "step": 1890
    },
    {
      "epoch": 2.96875,
      "grad_norm": 1.2821831872418814,
      "learning_rate": 5.013348105965748e-07,
      "loss": 0.2853,
      "step": 1900
    },
    {
      "epoch": 2.984375,
      "grad_norm": 1.350766870220701,
      "learning_rate": 5.003337273989165e-07,
      "loss": 0.2841,
      "step": 1910
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.2850731256398582,
      "learning_rate": 5e-07,
      "loss": 0.287,
      "step": 1920
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.4800361096858978,
      "eval_runtime": 67.4827,
      "eval_samples_per_second": 255.473,
      "eval_steps_per_second": 1.008,
      "step": 1920
    },
    {
      "epoch": 3.0,
      "step": 1920,
      "total_flos": 3216071511244800.0,
      "train_loss": 0.3934497516602278,
      "train_runtime": 11864.3784,
      "train_samples_per_second": 82.821,
      "train_steps_per_second": 0.162
    }
  ],
  "logging_steps": 10,
  "max_steps": 1920,
  "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": 3216071511244800.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}