{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9997268505872713,
  "eval_steps": 500,
  "global_step": 915,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0010925976509150504,
      "grad_norm": 22.730928776382456,
      "learning_rate": 1.0869565217391305e-07,
      "loss": 1.3223,
      "step": 1
    },
    {
      "epoch": 0.005462988254575253,
      "grad_norm": 20.891256691927374,
      "learning_rate": 5.434782608695653e-07,
      "loss": 1.3066,
      "step": 5
    },
    {
      "epoch": 0.010925976509150505,
      "grad_norm": 8.917157931216027,
      "learning_rate": 1.0869565217391306e-06,
      "loss": 1.1943,
      "step": 10
    },
    {
      "epoch": 0.01638896476372576,
      "grad_norm": 7.290952602788098,
      "learning_rate": 1.6304347826086957e-06,
      "loss": 1.0279,
      "step": 15
    },
    {
      "epoch": 0.02185195301830101,
      "grad_norm": 3.006256144496407,
      "learning_rate": 2.173913043478261e-06,
      "loss": 0.9049,
      "step": 20
    },
    {
      "epoch": 0.027314941272876262,
      "grad_norm": 2.371134339375521,
      "learning_rate": 2.7173913043478263e-06,
      "loss": 0.8647,
      "step": 25
    },
    {
      "epoch": 0.03277792952745152,
      "grad_norm": 2.1504800702490865,
      "learning_rate": 3.2608695652173914e-06,
      "loss": 0.833,
      "step": 30
    },
    {
      "epoch": 0.03824091778202677,
      "grad_norm": 2.2317280108941,
      "learning_rate": 3.804347826086957e-06,
      "loss": 0.807,
      "step": 35
    },
    {
      "epoch": 0.04370390603660202,
      "grad_norm": 2.2488850758737255,
      "learning_rate": 4.347826086956522e-06,
      "loss": 0.7919,
      "step": 40
    },
    {
      "epoch": 0.04916689429117727,
      "grad_norm": 2.3580705798318484,
      "learning_rate": 4.891304347826087e-06,
      "loss": 0.7761,
      "step": 45
    },
    {
      "epoch": 0.054629882545752524,
      "grad_norm": 2.4373495899440414,
      "learning_rate": 5.4347826086956525e-06,
      "loss": 0.7614,
      "step": 50
    },
    {
      "epoch": 0.060092870800327776,
      "grad_norm": 2.4135433324014763,
      "learning_rate": 5.978260869565218e-06,
      "loss": 0.7458,
      "step": 55
    },
    {
      "epoch": 0.06555585905490303,
      "grad_norm": 2.81358704818424,
      "learning_rate": 6.521739130434783e-06,
      "loss": 0.7291,
      "step": 60
    },
    {
      "epoch": 0.07101884730947829,
      "grad_norm": 2.438627231166693,
      "learning_rate": 7.065217391304349e-06,
      "loss": 0.72,
      "step": 65
    },
    {
      "epoch": 0.07648183556405354,
      "grad_norm": 2.1869040708927177,
      "learning_rate": 7.608695652173914e-06,
      "loss": 0.7141,
      "step": 70
    },
    {
      "epoch": 0.08194482381862879,
      "grad_norm": 2.184483160699592,
      "learning_rate": 8.15217391304348e-06,
      "loss": 0.6994,
      "step": 75
    },
    {
      "epoch": 0.08740781207320404,
      "grad_norm": 2.3203787824757507,
      "learning_rate": 8.695652173913044e-06,
      "loss": 0.7019,
      "step": 80
    },
    {
      "epoch": 0.0928708003277793,
      "grad_norm": 2.3068506646214915,
      "learning_rate": 9.23913043478261e-06,
      "loss": 0.6958,
      "step": 85
    },
    {
      "epoch": 0.09833378858235454,
      "grad_norm": 2.257062695334498,
      "learning_rate": 9.782608695652175e-06,
      "loss": 0.6842,
      "step": 90
    },
    {
      "epoch": 0.1037967768369298,
      "grad_norm": 2.400983485932064,
      "learning_rate": 9.999672147992806e-06,
      "loss": 0.6914,
      "step": 95
    },
    {
      "epoch": 0.10925976509150505,
      "grad_norm": 2.324571446077097,
      "learning_rate": 9.997668763647963e-06,
      "loss": 0.6867,
      "step": 100
    },
    {
      "epoch": 0.1147227533460803,
      "grad_norm": 2.137443230070172,
      "learning_rate": 9.993844863856281e-06,
      "loss": 0.689,
      "step": 105
    },
    {
      "epoch": 0.12018574160065555,
      "grad_norm": 2.0564600115869585,
      "learning_rate": 9.988201841560945e-06,
      "loss": 0.6688,
      "step": 110
    },
    {
      "epoch": 0.12564872985523082,
      "grad_norm": 2.139005852684594,
      "learning_rate": 9.980741752362222e-06,
      "loss": 0.6711,
      "step": 115
    },
    {
      "epoch": 0.13111171810980607,
      "grad_norm": 2.530899194975221,
      "learning_rate": 9.971467313768668e-06,
      "loss": 0.6681,
      "step": 120
    },
    {
      "epoch": 0.13657470636438132,
      "grad_norm": 2.1541912419217466,
      "learning_rate": 9.96038190420721e-06,
      "loss": 0.6624,
      "step": 125
    },
    {
      "epoch": 0.14203769461895657,
      "grad_norm": 2.2266155964762646,
      "learning_rate": 9.947489561792475e-06,
      "loss": 0.6678,
      "step": 130
    },
    {
      "epoch": 0.14750068287353182,
      "grad_norm": 2.0151113143541215,
      "learning_rate": 9.932794982855818e-06,
      "loss": 0.6576,
      "step": 135
    },
    {
      "epoch": 0.15296367112810708,
      "grad_norm": 2.122951721690571,
      "learning_rate": 9.916303520234573e-06,
      "loss": 0.6518,
      "step": 140
    },
    {
      "epoch": 0.15842665938268233,
      "grad_norm": 2.073005839186561,
      "learning_rate": 9.898021181322158e-06,
      "loss": 0.6467,
      "step": 145
    },
    {
      "epoch": 0.16388964763725758,
      "grad_norm": 2.122513038916697,
      "learning_rate": 9.877954625879746e-06,
      "loss": 0.6422,
      "step": 150
    },
    {
      "epoch": 0.16935263589183283,
      "grad_norm": 1.938211750086781,
      "learning_rate": 9.8561111636103e-06,
      "loss": 0.6313,
      "step": 155
    },
    {
      "epoch": 0.17481562414640808,
      "grad_norm": 2.1510416920205997,
      "learning_rate": 9.832498751495832e-06,
      "loss": 0.637,
      "step": 160
    },
    {
      "epoch": 0.18027861240098333,
      "grad_norm": 2.368980517738678,
      "learning_rate": 9.807125990898905e-06,
      "loss": 0.636,
      "step": 165
    },
    {
      "epoch": 0.1857416006555586,
      "grad_norm": 2.2294247264281553,
      "learning_rate": 9.780002124429377e-06,
      "loss": 0.6378,
      "step": 170
    },
    {
      "epoch": 0.19120458891013384,
      "grad_norm": 2.044434482857163,
      "learning_rate": 9.75113703257758e-06,
      "loss": 0.6299,
      "step": 175
    },
    {
      "epoch": 0.1966675771647091,
      "grad_norm": 2.273754929604351,
      "learning_rate": 9.720541230115113e-06,
      "loss": 0.6278,
      "step": 180
    },
    {
      "epoch": 0.20213056541928434,
      "grad_norm": 1.9049721088945737,
      "learning_rate": 9.688225862264604e-06,
      "loss": 0.6215,
      "step": 185
    },
    {
      "epoch": 0.2075935536738596,
      "grad_norm": 2.2956911916513936,
      "learning_rate": 9.654202700639805e-06,
      "loss": 0.6114,
      "step": 190
    },
    {
      "epoch": 0.21305654192843484,
      "grad_norm": 2.047625547879147,
      "learning_rate": 9.61848413895751e-06,
      "loss": 0.6272,
      "step": 195
    },
    {
      "epoch": 0.2185195301830101,
      "grad_norm": 2.144882455415688,
      "learning_rate": 9.581083188522862e-06,
      "loss": 0.6146,
      "step": 200
    },
    {
      "epoch": 0.22398251843758535,
      "grad_norm": 1.882314204150727,
      "learning_rate": 9.542013473489683e-06,
      "loss": 0.6115,
      "step": 205
    },
    {
      "epoch": 0.2294455066921606,
      "grad_norm": 2.0782768044314746,
      "learning_rate": 9.501289225897566e-06,
      "loss": 0.5965,
      "step": 210
    },
    {
      "epoch": 0.23490849494673585,
      "grad_norm": 1.9514474259034185,
      "learning_rate": 9.458925280487532e-06,
      "loss": 0.6176,
      "step": 215
    },
    {
      "epoch": 0.2403714832013111,
      "grad_norm": 1.8865949054260884,
      "learning_rate": 9.414937069298125e-06,
      "loss": 0.6018,
      "step": 220
    },
    {
      "epoch": 0.24583447145588638,
      "grad_norm": 2.0683179915172962,
      "learning_rate": 9.369340616043948e-06,
      "loss": 0.6105,
      "step": 225
    },
    {
      "epoch": 0.25129745971046163,
      "grad_norm": 2.004631261000376,
      "learning_rate": 9.322152530278658e-06,
      "loss": 0.5931,
      "step": 230
    },
    {
      "epoch": 0.2567604479650369,
      "grad_norm": 1.9653795566834225,
      "learning_rate": 9.273390001344544e-06,
      "loss": 0.5878,
      "step": 235
    },
    {
      "epoch": 0.26222343621961214,
      "grad_norm": 2.0291055623314715,
      "learning_rate": 9.223070792110927e-06,
      "loss": 0.5827,
      "step": 240
    },
    {
      "epoch": 0.2676864244741874,
      "grad_norm": 1.9570477475533896,
      "learning_rate": 9.17121323250362e-06,
      "loss": 0.5943,
      "step": 245
    },
    {
      "epoch": 0.27314941272876264,
      "grad_norm": 2.1357120267635463,
      "learning_rate": 9.117836212827839e-06,
      "loss": 0.571,
      "step": 250
    },
    {
      "epoch": 0.2786124009833379,
      "grad_norm": 1.9807416332273953,
      "learning_rate": 9.062959176886967e-06,
      "loss": 0.5895,
      "step": 255
    },
    {
      "epoch": 0.28407538923791315,
      "grad_norm": 1.8679007980966873,
      "learning_rate": 9.006602114899711e-06,
      "loss": 0.5837,
      "step": 260
    },
    {
      "epoch": 0.2895383774924884,
      "grad_norm": 2.159508129093504,
      "learning_rate": 8.948785556218202e-06,
      "loss": 0.5872,
      "step": 265
    },
    {
      "epoch": 0.29500136574706365,
      "grad_norm": 2.011039338549636,
      "learning_rate": 8.88953056184971e-06,
      "loss": 0.5656,
      "step": 270
    },
    {
      "epoch": 0.3004643540016389,
      "grad_norm": 1.8497602272423903,
      "learning_rate": 8.828858716784692e-06,
      "loss": 0.5607,
      "step": 275
    },
    {
      "epoch": 0.30592734225621415,
      "grad_norm": 1.8697548121419354,
      "learning_rate": 8.766792122133949e-06,
      "loss": 0.569,
      "step": 280
    },
    {
      "epoch": 0.3113903305107894,
      "grad_norm": 2.023364352460863,
      "learning_rate": 8.703353387077813e-06,
      "loss": 0.5694,
      "step": 285
    },
    {
      "epoch": 0.31685331876536466,
      "grad_norm": 1.9966954532417194,
      "learning_rate": 8.638565620630219e-06,
      "loss": 0.5485,
      "step": 290
    },
    {
      "epoch": 0.3223163070199399,
      "grad_norm": 2.118048147240604,
      "learning_rate": 8.572452423220717e-06,
      "loss": 0.5544,
      "step": 295
    },
    {
      "epoch": 0.32777929527451516,
      "grad_norm": 1.9646992858208574,
      "learning_rate": 8.505037878097481e-06,
      "loss": 0.5585,
      "step": 300
    },
    {
      "epoch": 0.3332422835290904,
      "grad_norm": 1.842699727967433,
      "learning_rate": 8.436346542554432e-06,
      "loss": 0.5525,
      "step": 305
    },
    {
      "epoch": 0.33870527178366566,
      "grad_norm": 2.1516397434897,
      "learning_rate": 8.366403438985675e-06,
      "loss": 0.5598,
      "step": 310
    },
    {
      "epoch": 0.3441682600382409,
      "grad_norm": 1.935829862534685,
      "learning_rate": 8.295234045770524e-06,
      "loss": 0.5474,
      "step": 315
    },
    {
      "epoch": 0.34963124829281617,
      "grad_norm": 2.0947580703115,
      "learning_rate": 8.222864287992419e-06,
      "loss": 0.5444,
      "step": 320
    },
    {
      "epoch": 0.3550942365473914,
      "grad_norm": 2.0230909691945462,
      "learning_rate": 8.149320527995111e-06,
      "loss": 0.5367,
      "step": 325
    },
    {
      "epoch": 0.36055722480196667,
      "grad_norm": 1.9422498738110598,
      "learning_rate": 8.074629555779586e-06,
      "loss": 0.5374,
      "step": 330
    },
    {
      "epoch": 0.3660202130565419,
      "grad_norm": 2.017157014853384,
      "learning_rate": 7.998818579245183e-06,
      "loss": 0.538,
      "step": 335
    },
    {
      "epoch": 0.3714832013111172,
      "grad_norm": 2.0200129213489952,
      "learning_rate": 7.9219152142785e-06,
      "loss": 0.5261,
      "step": 340
    },
    {
      "epoch": 0.3769461895656924,
      "grad_norm": 1.9025863914724077,
      "learning_rate": 7.843947474693665e-06,
      "loss": 0.5272,
      "step": 345
    },
    {
      "epoch": 0.3824091778202677,
      "grad_norm": 1.98859186872293,
      "learning_rate": 7.764943762027675e-06,
      "loss": 0.5271,
      "step": 350
    },
    {
      "epoch": 0.38787216607484293,
      "grad_norm": 1.9641160396193444,
      "learning_rate": 7.68493285519447e-06,
      "loss": 0.5156,
      "step": 355
    },
    {
      "epoch": 0.3933351543294182,
      "grad_norm": 1.9337347711635438,
      "learning_rate": 7.603943900001567e-06,
      "loss": 0.5152,
      "step": 360
    },
    {
      "epoch": 0.39879814258399343,
      "grad_norm": 1.9734924840604742,
      "learning_rate": 7.522006398533022e-06,
      "loss": 0.5157,
      "step": 365
    },
    {
      "epoch": 0.4042611308385687,
      "grad_norm": 2.3339176407449904,
      "learning_rate": 7.439150198402619e-06,
      "loss": 0.5162,
      "step": 370
    },
    {
      "epoch": 0.40972411909314393,
      "grad_norm": 2.2975439038971732,
      "learning_rate": 7.355405481881207e-06,
      "loss": 0.5157,
      "step": 375
    },
    {
      "epoch": 0.4151871073477192,
      "grad_norm": 1.971313224728527,
      "learning_rate": 7.270802754902092e-06,
      "loss": 0.5014,
      "step": 380
    },
    {
      "epoch": 0.42065009560229444,
      "grad_norm": 1.8926014206142687,
      "learning_rate": 7.185372835948574e-06,
      "loss": 0.5082,
      "step": 385
    },
    {
      "epoch": 0.4261130838568697,
      "grad_norm": 2.0424297512212677,
      "learning_rate": 7.0991468448275914e-06,
      "loss": 0.4969,
      "step": 390
    },
    {
      "epoch": 0.43157607211144494,
      "grad_norm": 1.9075652902136018,
      "learning_rate": 7.0121561913336255e-06,
      "loss": 0.5025,
      "step": 395
    },
    {
      "epoch": 0.4370390603660202,
      "grad_norm": 1.972187929750843,
      "learning_rate": 6.924432563806962e-06,
      "loss": 0.4805,
      "step": 400
    },
    {
      "epoch": 0.44250204862059545,
      "grad_norm": 2.242916059633561,
      "learning_rate": 6.836007917590487e-06,
      "loss": 0.503,
      "step": 405
    },
    {
      "epoch": 0.4479650368751707,
      "grad_norm": 2.0649875803912274,
      "learning_rate": 6.746914463389216e-06,
      "loss": 0.4901,
      "step": 410
    },
    {
      "epoch": 0.45342802512974595,
      "grad_norm": 1.8063638082708948,
      "learning_rate": 6.65718465553681e-06,
      "loss": 0.4755,
      "step": 415
    },
    {
      "epoch": 0.4588910133843212,
      "grad_norm": 1.969494410176558,
      "learning_rate": 6.566851180173344e-06,
      "loss": 0.4874,
      "step": 420
    },
    {
      "epoch": 0.46435400163889645,
      "grad_norm": 1.8970828426082733,
      "learning_rate": 6.475946943338616e-06,
      "loss": 0.4766,
      "step": 425
    },
    {
      "epoch": 0.4698169898934717,
      "grad_norm": 1.9623505460334127,
      "learning_rate": 6.384505058985388e-06,
      "loss": 0.487,
      "step": 430
    },
    {
      "epoch": 0.47527997814804696,
      "grad_norm": 2.016298870850614,
      "learning_rate": 6.292558836916856e-06,
      "loss": 0.4684,
      "step": 435
    },
    {
      "epoch": 0.4807429664026222,
      "grad_norm": 1.9782532484829365,
      "learning_rate": 6.200141770652791e-06,
      "loss": 0.4655,
      "step": 440
    },
    {
      "epoch": 0.4862059546571975,
      "grad_norm": 1.9875880346692325,
      "learning_rate": 6.107287525228764e-06,
      "loss": 0.4829,
      "step": 445
    },
    {
      "epoch": 0.49166894291177277,
      "grad_norm": 2.0320364233161836,
      "learning_rate": 6.014029924932874e-06,
      "loss": 0.4753,
      "step": 450
    },
    {
      "epoch": 0.497131931166348,
      "grad_norm": 2.025675106544592,
      "learning_rate": 5.920402940984483e-06,
      "loss": 0.4673,
      "step": 455
    },
    {
      "epoch": 0.5025949194209233,
      "grad_norm": 1.921954531735398,
      "learning_rate": 5.826440679159424e-06,
      "loss": 0.4532,
      "step": 460
    },
    {
      "epoch": 0.5080579076754985,
      "grad_norm": 2.01066093050374,
      "learning_rate": 5.732177367366176e-06,
      "loss": 0.4658,
      "step": 465
    },
    {
      "epoch": 0.5135208959300738,
      "grad_norm": 1.8422653149296644,
      "learning_rate": 5.63764734317758e-06,
      "loss": 0.4529,
      "step": 470
    },
    {
      "epoch": 0.518983884184649,
      "grad_norm": 1.9255817468742156,
      "learning_rate": 5.542885041322577e-06,
      "loss": 0.4561,
      "step": 475
    },
    {
      "epoch": 0.5244468724392243,
      "grad_norm": 1.951977869616497,
      "learning_rate": 5.447924981142578e-06,
      "loss": 0.4565,
      "step": 480
    },
    {
      "epoch": 0.5299098606937995,
      "grad_norm": 1.9554258723769773,
      "learning_rate": 5.3528017540169974e-06,
      "loss": 0.4475,
      "step": 485
    },
    {
      "epoch": 0.5353728489483748,
      "grad_norm": 1.9102850022081725,
      "learning_rate": 5.25755001076255e-06,
      "loss": 0.4497,
      "step": 490
    },
    {
      "epoch": 0.54083583720295,
      "grad_norm": 2.239195466962009,
      "learning_rate": 5.1622044490108994e-06,
      "loss": 0.4454,
      "step": 495
    },
    {
      "epoch": 0.5462988254575253,
      "grad_norm": 1.8827718246138567,
      "learning_rate": 5.066799800569247e-06,
      "loss": 0.4404,
      "step": 500
    },
    {
      "epoch": 0.5517618137121005,
      "grad_norm": 1.9069058786965245,
      "learning_rate": 4.971370818768475e-06,
      "loss": 0.4236,
      "step": 505
    },
    {
      "epoch": 0.5572248019666758,
      "grad_norm": 1.855279389571921,
      "learning_rate": 4.875952265803452e-06,
      "loss": 0.4403,
      "step": 510
    },
    {
      "epoch": 0.562687790221251,
      "grad_norm": 1.8972542181259846,
      "learning_rate": 4.780578900070104e-06,
      "loss": 0.4372,
      "step": 515
    },
    {
      "epoch": 0.5681507784758263,
      "grad_norm": 1.844147614606819,
      "learning_rate": 4.685285463503867e-06,
      "loss": 0.4331,
      "step": 520
    },
    {
      "epoch": 0.5736137667304015,
      "grad_norm": 1.9625504078338127,
      "learning_rate": 4.5901066689241514e-06,
      "loss": 0.4292,
      "step": 525
    },
    {
      "epoch": 0.5790767549849768,
      "grad_norm": 2.2025005649073433,
      "learning_rate": 4.49507718738939e-06,
      "loss": 0.4293,
      "step": 530
    },
    {
      "epoch": 0.584539743239552,
      "grad_norm": 1.9450789672739341,
      "learning_rate": 4.400231635567319e-06,
      "loss": 0.4192,
      "step": 535
    },
    {
      "epoch": 0.5900027314941273,
      "grad_norm": 1.9764711012724976,
      "learning_rate": 4.305604563125061e-06,
      "loss": 0.4182,
      "step": 540
    },
    {
      "epoch": 0.5954657197487025,
      "grad_norm": 2.1198793543903194,
      "learning_rate": 4.211230440143604e-06,
      "loss": 0.4114,
      "step": 545
    },
    {
      "epoch": 0.6009287080032778,
      "grad_norm": 3.1276187198521463,
      "learning_rate": 4.117143644561306e-06,
      "loss": 0.4166,
      "step": 550
    },
    {
      "epoch": 0.606391696257853,
      "grad_norm": 1.8794840574959752,
      "learning_rate": 4.023378449650928e-06,
      "loss": 0.4076,
      "step": 555
    },
    {
      "epoch": 0.6118546845124283,
      "grad_norm": 1.9139438651443652,
      "learning_rate": 3.929969011534811e-06,
      "loss": 0.4218,
      "step": 560
    },
    {
      "epoch": 0.6173176727670036,
      "grad_norm": 1.861741261098825,
      "learning_rate": 3.83694935674272e-06,
      "loss": 0.4169,
      "step": 565
    },
    {
      "epoch": 0.6227806610215788,
      "grad_norm": 1.8247668127301933,
      "learning_rate": 3.7443533698169186e-06,
      "loss": 0.4132,
      "step": 570
    },
    {
      "epoch": 0.6282436492761541,
      "grad_norm": 1.9796553518989775,
      "learning_rate": 3.652214780968926e-06,
      "loss": 0.3996,
      "step": 575
    },
    {
      "epoch": 0.6337066375307293,
      "grad_norm": 1.823241119094308,
      "learning_rate": 3.5605671537925264e-06,
      "loss": 0.4088,
      "step": 580
    },
    {
      "epoch": 0.6391696257853046,
      "grad_norm": 1.881819370184891,
      "learning_rate": 3.469443873037457e-06,
      "loss": 0.4073,
      "step": 585
    },
    {
      "epoch": 0.6446326140398798,
      "grad_norm": 1.947962800229108,
      "learning_rate": 3.3788781324482446e-06,
      "loss": 0.4067,
      "step": 590
    },
    {
      "epoch": 0.6500956022944551,
      "grad_norm": 1.9214089256240032,
      "learning_rate": 3.2889029226726286e-06,
      "loss": 0.3996,
      "step": 595
    },
    {
      "epoch": 0.6555585905490303,
      "grad_norm": 1.9543022070758702,
      "learning_rate": 3.1995510192439587e-06,
      "loss": 0.3982,
      "step": 600
    },
    {
      "epoch": 0.6610215788036056,
      "grad_norm": 1.9173316158851936,
      "learning_rate": 3.1108549706419553e-06,
      "loss": 0.4,
      "step": 605
    },
    {
      "epoch": 0.6664845670581808,
      "grad_norm": 1.8159793606682484,
      "learning_rate": 3.022847086436176e-06,
      "loss": 0.3988,
      "step": 610
    },
    {
      "epoch": 0.6719475553127561,
      "grad_norm": 1.9089073637989584,
      "learning_rate": 2.9355594255165186e-06,
      "loss": 0.3871,
      "step": 615
    },
    {
      "epoch": 0.6774105435673313,
      "grad_norm": 1.8914866485656172,
      "learning_rate": 2.8490237844150335e-06,
      "loss": 0.3885,
      "step": 620
    },
    {
      "epoch": 0.6828735318219066,
      "grad_norm": 1.9221955744591699,
      "learning_rate": 2.763271685723311e-06,
      "loss": 0.3818,
      "step": 625
    },
    {
      "epoch": 0.6883365200764818,
      "grad_norm": 1.8733166396673193,
      "learning_rate": 2.6783343666096445e-06,
      "loss": 0.3874,
      "step": 630
    },
    {
      "epoch": 0.6937995083310571,
      "grad_norm": 2.134783287581667,
      "learning_rate": 2.5942427674401936e-06,
      "loss": 0.3899,
      "step": 635
    },
    {
      "epoch": 0.6992624965856323,
      "grad_norm": 1.8202198608461309,
      "learning_rate": 2.5110275205082224e-06,
      "loss": 0.3839,
      "step": 640
    },
    {
      "epoch": 0.7047254848402076,
      "grad_norm": 1.9005906531792638,
      "learning_rate": 2.428718938875607e-06,
      "loss": 0.3812,
      "step": 645
    },
    {
      "epoch": 0.7101884730947828,
      "grad_norm": 1.8656177124830933,
      "learning_rate": 2.3473470053305954e-06,
      "loss": 0.3805,
      "step": 650
    },
    {
      "epoch": 0.7156514613493581,
      "grad_norm": 1.9350473980336609,
      "learning_rate": 2.266941361465886e-06,
      "loss": 0.381,
      "step": 655
    },
    {
      "epoch": 0.7211144496039333,
      "grad_norm": 1.8146852411399312,
      "learning_rate": 2.187531296881017e-06,
      "loss": 0.3751,
      "step": 660
    },
    {
      "epoch": 0.7265774378585086,
      "grad_norm": 1.8246829733561232,
      "learning_rate": 2.1091457385129267e-06,
      "loss": 0.3752,
      "step": 665
    },
    {
      "epoch": 0.7320404261130838,
      "grad_norm": 1.8550222595976351,
      "learning_rate": 2.031813240098686e-06,
      "loss": 0.3721,
      "step": 670
    },
    {
      "epoch": 0.7375034143676591,
      "grad_norm": 1.8354415599143141,
      "learning_rate": 1.9555619717741248e-06,
      "loss": 0.3733,
      "step": 675
    },
    {
      "epoch": 0.7429664026222343,
      "grad_norm": 1.9427997534753285,
      "learning_rate": 1.8804197098122168e-06,
      "loss": 0.3745,
      "step": 680
    },
    {
      "epoch": 0.7484293908768096,
      "grad_norm": 1.9154466908271706,
      "learning_rate": 1.8064138265049457e-06,
      "loss": 0.3743,
      "step": 685
    },
    {
      "epoch": 0.7538923791313848,
      "grad_norm": 1.8117953280714802,
      "learning_rate": 1.7335712801923015e-06,
      "loss": 0.3707,
      "step": 690
    },
    {
      "epoch": 0.7593553673859601,
      "grad_norm": 1.894844068057027,
      "learning_rate": 1.6619186054421088e-06,
      "loss": 0.3677,
      "step": 695
    },
    {
      "epoch": 0.7648183556405354,
      "grad_norm": 1.8949515530814003,
      "learning_rate": 1.5914819033841843e-06,
      "loss": 0.3639,
      "step": 700
    },
    {
      "epoch": 0.7702813438951106,
      "grad_norm": 1.7568307263853036,
      "learning_rate": 1.5222868322024092e-06,
      "loss": 0.3586,
      "step": 705
    },
    {
      "epoch": 0.7757443321496859,
      "grad_norm": 1.759178792643852,
      "learning_rate": 1.4543585977881513e-06,
      "loss": 0.35,
      "step": 710
    },
    {
      "epoch": 0.7812073204042611,
      "grad_norm": 1.8973466785062572,
      "learning_rate": 1.3877219445584283e-06,
      "loss": 0.3593,
      "step": 715
    },
    {
      "epoch": 0.7866703086588364,
      "grad_norm": 1.9605293056514521,
      "learning_rate": 1.3224011464422032e-06,
      "loss": 0.3621,
      "step": 720
    },
    {
      "epoch": 0.7921332969134116,
      "grad_norm": 1.812660380797101,
      "learning_rate": 1.2584199980380358e-06,
      "loss": 0.3599,
      "step": 725
    },
    {
      "epoch": 0.7975962851679869,
      "grad_norm": 1.8356228496720945,
      "learning_rate": 1.1958018059463577e-06,
      "loss": 0.3605,
      "step": 730
    },
    {
      "epoch": 0.8030592734225621,
      "grad_norm": 1.691309969533953,
      "learning_rate": 1.1345693802795177e-06,
      "loss": 0.3583,
      "step": 735
    },
    {
      "epoch": 0.8085222616771374,
      "grad_norm": 1.8242600919765277,
      "learning_rate": 1.0747450263526576e-06,
      "loss": 0.3598,
      "step": 740
    },
    {
      "epoch": 0.8139852499317126,
      "grad_norm": 1.7765876082507979,
      "learning_rate": 1.0163505365585086e-06,
      "loss": 0.3493,
      "step": 745
    },
    {
      "epoch": 0.8194482381862879,
      "grad_norm": 1.8095739075310635,
      "learning_rate": 9.594071824289986e-07,
      "loss": 0.3571,
      "step": 750
    },
    {
      "epoch": 0.8249112264408631,
      "grad_norm": 1.7811537133839481,
      "learning_rate": 9.039357068866178e-07,
      "loss": 0.3518,
      "step": 755
    },
    {
      "epoch": 0.8303742146954384,
      "grad_norm": 1.7625847558197685,
      "learning_rate": 8.49956316688329e-07,
      "loss": 0.3518,
      "step": 760
    },
    {
      "epoch": 0.8358372029500136,
      "grad_norm": 1.8189872840713601,
      "learning_rate": 7.974886750647887e-07,
      "loss": 0.3381,
      "step": 765
    },
    {
      "epoch": 0.8413001912045889,
      "grad_norm": 1.8024026803759359,
      "learning_rate": 7.465518945575789e-07,
      "loss": 0.345,
      "step": 770
    },
    {
      "epoch": 0.8467631794591641,
      "grad_norm": 1.895342687896103,
      "learning_rate": 6.971645300570084e-07,
      "loss": 0.3464,
      "step": 775
    },
    {
      "epoch": 0.8522261677137394,
      "grad_norm": 1.796319045444646,
      "learning_rate": 6.49344572043083e-07,
      "loss": 0.3477,
      "step": 780
    },
    {
      "epoch": 0.8576891559683146,
      "grad_norm": 1.8845046027721593,
      "learning_rate": 6.031094400320498e-07,
      "loss": 0.3443,
      "step": 785
    },
    {
      "epoch": 0.8631521442228899,
      "grad_norm": 1.724637234062916,
      "learning_rate": 5.584759762309422e-07,
      "loss": 0.3367,
      "step": 790
    },
    {
      "epoch": 0.8686151324774651,
      "grad_norm": 1.8964510379949302,
      "learning_rate": 5.154604394024254e-07,
      "loss": 0.3396,
      "step": 795
    },
    {
      "epoch": 0.8740781207320404,
      "grad_norm": 1.6763296815328177,
      "learning_rate": 4.7407849894216097e-07,
      "loss": 0.3348,
      "step": 800
    },
    {
      "epoch": 0.8795411089866156,
      "grad_norm": 1.7128909727748394,
      "learning_rate": 4.343452291708783e-07,
      "loss": 0.343,
      "step": 805
    },
    {
      "epoch": 0.8850040972411909,
      "grad_norm": 1.8153223119786095,
      "learning_rate": 3.962751038431961e-07,
      "loss": 0.3405,
      "step": 810
    },
    {
      "epoch": 0.8904670854957661,
      "grad_norm": 1.7625733448640701,
      "learning_rate": 3.5988199087523987e-07,
      "loss": 0.3396,
      "step": 815
    },
    {
      "epoch": 0.8959300737503414,
      "grad_norm": 1.8570474171994868,
      "learning_rate": 3.2517914729292446e-07,
      "loss": 0.3391,
      "step": 820
    },
    {
      "epoch": 0.9013930620049166,
      "grad_norm": 1.8466215461741358,
      "learning_rate": 2.921792144027813e-07,
      "loss": 0.3407,
      "step": 825
    },
    {
      "epoch": 0.9068560502594919,
      "grad_norm": 1.7424828950437026,
      "learning_rate": 2.6089421318707785e-07,
      "loss": 0.3365,
      "step": 830
    },
    {
      "epoch": 0.9123190385140671,
      "grad_norm": 1.8348358804511151,
      "learning_rate": 2.3133553992488845e-07,
      "loss": 0.3453,
      "step": 835
    },
    {
      "epoch": 0.9177820267686424,
      "grad_norm": 1.8267323618432934,
      "learning_rate": 2.0351396204074946e-07,
      "loss": 0.337,
      "step": 840
    },
    {
      "epoch": 0.9232450150232177,
      "grad_norm": 1.6632279811187767,
      "learning_rate": 1.7743961418237144e-07,
      "loss": 0.3341,
      "step": 845
    },
    {
      "epoch": 0.9287080032777929,
      "grad_norm": 1.8291284979726596,
      "learning_rate": 1.5312199452886145e-07,
      "loss": 0.332,
      "step": 850
    },
    {
      "epoch": 0.9341709915323682,
      "grad_norm": 1.7047344118755032,
      "learning_rate": 1.3056996133079924e-07,
      "loss": 0.3375,
      "step": 855
    },
    {
      "epoch": 0.9396339797869434,
      "grad_norm": 1.745785425026002,
      "learning_rate": 1.0979172968340667e-07,
      "loss": 0.3321,
      "step": 860
    },
    {
      "epoch": 0.9450969680415187,
      "grad_norm": 1.8776860033466447,
      "learning_rate": 9.079486853402097e-08,
      "loss": 0.34,
      "step": 865
    },
    {
      "epoch": 0.9505599562960939,
      "grad_norm": 1.7269678812008018,
      "learning_rate": 7.358629792492522e-08,
      "loss": 0.342,
      "step": 870
    },
    {
      "epoch": 0.9560229445506692,
      "grad_norm": 1.7727161663929172,
      "learning_rate": 5.8172286472566456e-08,
      "loss": 0.3391,
      "step": 875
    },
    {
      "epoch": 0.9614859328052444,
      "grad_norm": 1.7673439074550763,
      "learning_rate": 4.455844908407059e-08,
      "loss": 0.3416,
      "step": 880
    },
    {
      "epoch": 0.9669489210598197,
      "grad_norm": 1.7746641186916818,
      "learning_rate": 3.27497449118791e-08,
      "loss": 0.3268,
      "step": 885
    },
    {
      "epoch": 0.972411909314395,
      "grad_norm": 1.858186055296109,
      "learning_rate": 2.275047554726795e-08,
      "loss": 0.3371,
      "step": 890
    },
    {
      "epoch": 0.9778748975689703,
      "grad_norm": 1.6930712261364957,
      "learning_rate": 1.4564283453392669e-08,
      "loss": 0.3425,
      "step": 895
    },
    {
      "epoch": 0.9833378858235455,
      "grad_norm": 1.778838215839912,
      "learning_rate": 8.194150638438093e-09,
      "loss": 0.3315,
      "step": 900
    },
    {
      "epoch": 0.9888008740781208,
      "grad_norm": 1.8180837953030582,
      "learning_rate": 3.6423975693533618e-09,
      "loss": 0.3404,
      "step": 905
    },
    {
      "epoch": 0.994263862332696,
      "grad_norm": 1.7628508134053706,
      "learning_rate": 9.106823265681019e-10,
      "loss": 0.335,
      "step": 910
    },
    {
      "epoch": 0.9997268505872713,
      "grad_norm": 1.7810702627412893,
      "learning_rate": 0.0,
      "loss": 0.3383,
      "step": 915
    },
    {
      "epoch": 0.9997268505872713,
      "eval_loss": 0.2970966100692749,
      "eval_runtime": 1.1854,
      "eval_samples_per_second": 1.687,
      "eval_steps_per_second": 0.844,
      "step": 915
    },
    {
      "epoch": 0.9997268505872713,
      "step": 915,
      "total_flos": 191530040033280.0,
      "train_loss": 0.4996221627042593,
      "train_runtime": 20787.7504,
      "train_samples_per_second": 1.409,
      "train_steps_per_second": 0.044
    }
  ],
  "logging_steps": 5,
  "max_steps": 915,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 191530040033280.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}