{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.149184149184149,
  "eval_steps": 500,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014344629729245113,
      "grad_norm": 2.203942058202339,
      "learning_rate": 4.285714285714286e-06,
      "loss": 1.1704,
      "step": 10
    },
    {
      "epoch": 0.028689259458490227,
      "grad_norm": 1.1497467131489452,
      "learning_rate": 9.047619047619047e-06,
      "loss": 0.9166,
      "step": 20
    },
    {
      "epoch": 0.04303388918773534,
      "grad_norm": 0.8889709375752792,
      "learning_rate": 1.3809523809523811e-05,
      "loss": 0.7948,
      "step": 30
    },
    {
      "epoch": 0.05737851891698045,
      "grad_norm": 0.9291104053668569,
      "learning_rate": 1.8571428571428572e-05,
      "loss": 0.7447,
      "step": 40
    },
    {
      "epoch": 0.07172314864622557,
      "grad_norm": 1.1881747732049386,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.729,
      "step": 50
    },
    {
      "epoch": 0.08606777837547068,
      "grad_norm": 0.9032703792102055,
      "learning_rate": 2.8095238095238096e-05,
      "loss": 0.7177,
      "step": 60
    },
    {
      "epoch": 0.1004124081047158,
      "grad_norm": 0.9447848465031972,
      "learning_rate": 3.285714285714286e-05,
      "loss": 0.7048,
      "step": 70
    },
    {
      "epoch": 0.1147570378339609,
      "grad_norm": 0.9807912049624671,
      "learning_rate": 3.761904761904762e-05,
      "loss": 0.7048,
      "step": 80
    },
    {
      "epoch": 0.129101667563206,
      "grad_norm": 0.8678523409043055,
      "learning_rate": 4.2380952380952385e-05,
      "loss": 0.7065,
      "step": 90
    },
    {
      "epoch": 0.14344629729245115,
      "grad_norm": 1.095952408037045,
      "learning_rate": 4.714285714285714e-05,
      "loss": 0.7098,
      "step": 100
    },
    {
      "epoch": 0.15779092702169625,
      "grad_norm": 0.932556439833824,
      "learning_rate": 5.1904761904761913e-05,
      "loss": 0.7047,
      "step": 110
    },
    {
      "epoch": 0.17213555675094136,
      "grad_norm": 0.716991228951169,
      "learning_rate": 5.666666666666667e-05,
      "loss": 0.704,
      "step": 120
    },
    {
      "epoch": 0.1864801864801865,
      "grad_norm": 0.7487957467977157,
      "learning_rate": 6.142857142857143e-05,
      "loss": 0.7013,
      "step": 130
    },
    {
      "epoch": 0.2008248162094316,
      "grad_norm": 0.7206473376735171,
      "learning_rate": 6.619047619047619e-05,
      "loss": 0.7038,
      "step": 140
    },
    {
      "epoch": 0.2151694459386767,
      "grad_norm": 0.6694694288559293,
      "learning_rate": 7.095238095238096e-05,
      "loss": 0.7119,
      "step": 150
    },
    {
      "epoch": 0.2295140756679218,
      "grad_norm": 0.7219727300855844,
      "learning_rate": 7.571428571428571e-05,
      "loss": 0.7068,
      "step": 160
    },
    {
      "epoch": 0.24385870539716695,
      "grad_norm": 0.6818226409038356,
      "learning_rate": 8.047619047619048e-05,
      "loss": 0.7088,
      "step": 170
    },
    {
      "epoch": 0.258203335126412,
      "grad_norm": 0.6977379507391648,
      "learning_rate": 8.523809523809524e-05,
      "loss": 0.7032,
      "step": 180
    },
    {
      "epoch": 0.2725479648556572,
      "grad_norm": 0.6553925195793743,
      "learning_rate": 9e-05,
      "loss": 0.7081,
      "step": 190
    },
    {
      "epoch": 0.2868925945849023,
      "grad_norm": 0.5491527719782942,
      "learning_rate": 9.476190476190476e-05,
      "loss": 0.7079,
      "step": 200
    },
    {
      "epoch": 0.3012372243141474,
      "grad_norm": 0.6640413812930087,
      "learning_rate": 9.952380952380953e-05,
      "loss": 0.7153,
      "step": 210
    },
    {
      "epoch": 0.3155818540433925,
      "grad_norm": 0.5699857338148432,
      "learning_rate": 9.999436939807164e-05,
      "loss": 0.7235,
      "step": 220
    },
    {
      "epoch": 0.3299264837726376,
      "grad_norm": 0.6222679010251994,
      "learning_rate": 9.99749072170404e-05,
      "loss": 0.7215,
      "step": 230
    },
    {
      "epoch": 0.3442711135018827,
      "grad_norm": 0.529203333569492,
      "learning_rate": 9.994154935353517e-05,
      "loss": 0.7074,
      "step": 240
    },
    {
      "epoch": 0.3586157432311278,
      "grad_norm": 0.524281983121501,
      "learning_rate": 9.98943050828164e-05,
      "loss": 0.7086,
      "step": 250
    },
    {
      "epoch": 0.372960372960373,
      "grad_norm": 0.5037155864766709,
      "learning_rate": 9.983318754130435e-05,
      "loss": 0.7004,
      "step": 260
    },
    {
      "epoch": 0.3873050026896181,
      "grad_norm": 0.5619940203814628,
      "learning_rate": 9.975821372292653e-05,
      "loss": 0.6961,
      "step": 270
    },
    {
      "epoch": 0.4016496324188632,
      "grad_norm": 0.46139196683094363,
      "learning_rate": 9.966940447439245e-05,
      "loss": 0.6914,
      "step": 280
    },
    {
      "epoch": 0.4159942621481083,
      "grad_norm": 0.5701814263114097,
      "learning_rate": 9.956678448939718e-05,
      "loss": 0.6805,
      "step": 290
    },
    {
      "epoch": 0.4303388918773534,
      "grad_norm": 0.5822888434573161,
      "learning_rate": 9.945038230175509e-05,
      "loss": 0.6918,
      "step": 300
    },
    {
      "epoch": 0.4446835216065985,
      "grad_norm": 0.44360325798902245,
      "learning_rate": 9.932023027746602e-05,
      "loss": 0.6798,
      "step": 310
    },
    {
      "epoch": 0.4590281513358436,
      "grad_norm": 0.4584389136933647,
      "learning_rate": 9.917636460571578e-05,
      "loss": 0.6848,
      "step": 320
    },
    {
      "epoch": 0.47337278106508873,
      "grad_norm": 0.42186486444358706,
      "learning_rate": 9.901882528881363e-05,
      "loss": 0.6856,
      "step": 330
    },
    {
      "epoch": 0.4877174107943339,
      "grad_norm": 0.4547332297314441,
      "learning_rate": 9.884765613106948e-05,
      "loss": 0.6761,
      "step": 340
    },
    {
      "epoch": 0.502062040523579,
      "grad_norm": 0.5669430722485119,
      "learning_rate": 9.866290472661406e-05,
      "loss": 0.6795,
      "step": 350
    },
    {
      "epoch": 0.516406670252824,
      "grad_norm": 0.4480667658595798,
      "learning_rate": 9.846462244616508e-05,
      "loss": 0.6701,
      "step": 360
    },
    {
      "epoch": 0.5307512999820692,
      "grad_norm": 0.4330593851481768,
      "learning_rate": 9.825286442274357e-05,
      "loss": 0.6785,
      "step": 370
    },
    {
      "epoch": 0.5450959297113144,
      "grad_norm": 0.43973236782853137,
      "learning_rate": 9.802768953634388e-05,
      "loss": 0.6692,
      "step": 380
    },
    {
      "epoch": 0.5594405594405595,
      "grad_norm": 0.4800141760306811,
      "learning_rate": 9.778916039756193e-05,
      "loss": 0.6644,
      "step": 390
    },
    {
      "epoch": 0.5737851891698046,
      "grad_norm": 0.4370272358079981,
      "learning_rate": 9.753734333018616e-05,
      "loss": 0.6582,
      "step": 400
    },
    {
      "epoch": 0.5881298188990497,
      "grad_norm": 0.46636751943765076,
      "learning_rate": 9.727230835275598e-05,
      "loss": 0.6606,
      "step": 410
    },
    {
      "epoch": 0.6024744486282948,
      "grad_norm": 0.44398226842129596,
      "learning_rate": 9.699412915909284e-05,
      "loss": 0.6591,
      "step": 420
    },
    {
      "epoch": 0.6168190783575399,
      "grad_norm": 0.417529788133883,
      "learning_rate": 9.670288309780953e-05,
      "loss": 0.6502,
      "step": 430
    },
    {
      "epoch": 0.631163708086785,
      "grad_norm": 0.425244481854258,
      "learning_rate": 9.639865115080304e-05,
      "loss": 0.6601,
      "step": 440
    },
    {
      "epoch": 0.6455083378160301,
      "grad_norm": 0.4266753216646151,
      "learning_rate": 9.608151791073737e-05,
      "loss": 0.6552,
      "step": 450
    },
    {
      "epoch": 0.6598529675452752,
      "grad_norm": 0.4060939108779214,
      "learning_rate": 9.575157155752222e-05,
      "loss": 0.6483,
      "step": 460
    },
    {
      "epoch": 0.6741975972745203,
      "grad_norm": 0.44733731016902734,
      "learning_rate": 9.54089038337943e-05,
      "loss": 0.6414,
      "step": 470
    },
    {
      "epoch": 0.6885422270037654,
      "grad_norm": 0.4343263599516913,
      "learning_rate": 9.5053610019408e-05,
      "loss": 0.6452,
      "step": 480
    },
    {
      "epoch": 0.7028868567330105,
      "grad_norm": 0.40484892671799194,
      "learning_rate": 9.468578890494256e-05,
      "loss": 0.6399,
      "step": 490
    },
    {
      "epoch": 0.7172314864622557,
      "grad_norm": 0.3977086895873736,
      "learning_rate": 9.430554276423292e-05,
      "loss": 0.6427,
      "step": 500
    },
    {
      "epoch": 0.7315761161915008,
      "grad_norm": 0.4059393728648637,
      "learning_rate": 9.391297732593229e-05,
      "loss": 0.6405,
      "step": 510
    },
    {
      "epoch": 0.745920745920746,
      "grad_norm": 0.3918015124178049,
      "learning_rate": 9.350820174411386e-05,
      "loss": 0.6428,
      "step": 520
    },
    {
      "epoch": 0.7602653756499911,
      "grad_norm": 0.397284227604159,
      "learning_rate": 9.309132856792023e-05,
      "loss": 0.6379,
      "step": 530
    },
    {
      "epoch": 0.7746100053792362,
      "grad_norm": 0.423979379524061,
      "learning_rate": 9.266247371026873e-05,
      "loss": 0.6346,
      "step": 540
    },
    {
      "epoch": 0.7889546351084813,
      "grad_norm": 0.3974744664267281,
      "learning_rate": 9.222175641562143e-05,
      "loss": 0.6389,
      "step": 550
    },
    {
      "epoch": 0.8032992648377264,
      "grad_norm": 0.40660189053377105,
      "learning_rate": 9.176929922682891e-05,
      "loss": 0.6328,
      "step": 560
    },
    {
      "epoch": 0.8176438945669715,
      "grad_norm": 0.39183846193855565,
      "learning_rate": 9.130522795105676e-05,
      "loss": 0.6332,
      "step": 570
    },
    {
      "epoch": 0.8319885242962166,
      "grad_norm": 0.3794359883589157,
      "learning_rate": 9.082967162480459e-05,
      "loss": 0.6248,
      "step": 580
    },
    {
      "epoch": 0.8463331540254617,
      "grad_norm": 0.3757192118653646,
      "learning_rate": 9.034276247802688e-05,
      "loss": 0.6204,
      "step": 590
    },
    {
      "epoch": 0.8606777837547068,
      "grad_norm": 0.4344414187841982,
      "learning_rate": 8.984463589736614e-05,
      "loss": 0.6253,
      "step": 600
    },
    {
      "epoch": 0.8750224134839519,
      "grad_norm": 0.4124738293723295,
      "learning_rate": 8.933543038850816e-05,
      "loss": 0.6254,
      "step": 610
    },
    {
      "epoch": 0.889367043213197,
      "grad_norm": 0.3729179255559684,
      "learning_rate": 8.881528753767007e-05,
      "loss": 0.6225,
      "step": 620
    },
    {
      "epoch": 0.9037116729424421,
      "grad_norm": 0.3999425136973519,
      "learning_rate": 8.82843519722319e-05,
      "loss": 0.6219,
      "step": 630
    },
    {
      "epoch": 0.9180563026716873,
      "grad_norm": 0.3613041053584094,
      "learning_rate": 8.774277132052237e-05,
      "loss": 0.6151,
      "step": 640
    },
    {
      "epoch": 0.9324009324009324,
      "grad_norm": 0.3606937460298364,
      "learning_rate": 8.719069617077046e-05,
      "loss": 0.6216,
      "step": 650
    },
    {
      "epoch": 0.9467455621301775,
      "grad_norm": 0.42662552468134024,
      "learning_rate": 8.662828002923378e-05,
      "loss": 0.6179,
      "step": 660
    },
    {
      "epoch": 0.9610901918594227,
      "grad_norm": 0.3630873900729852,
      "learning_rate": 8.605567927751576e-05,
      "loss": 0.6171,
      "step": 670
    },
    {
      "epoch": 0.9754348215886678,
      "grad_norm": 0.3905293019503008,
      "learning_rate": 8.547305312908318e-05,
      "loss": 0.6085,
      "step": 680
    },
    {
      "epoch": 0.9897794513179129,
      "grad_norm": 0.36834813483218043,
      "learning_rate": 8.48805635849964e-05,
      "loss": 0.6074,
      "step": 690
    },
    {
      "epoch": 1.002868925945849,
      "grad_norm": 0.39960980145566394,
      "learning_rate": 8.427837538886437e-05,
      "loss": 0.5784,
      "step": 700
    },
    {
      "epoch": 1.017213555675094,
      "grad_norm": 0.38333026599469155,
      "learning_rate": 8.366665598103727e-05,
      "loss": 0.4367,
      "step": 710
    },
    {
      "epoch": 1.0315581854043392,
      "grad_norm": 0.3516312791289111,
      "learning_rate": 8.304557545204908e-05,
      "loss": 0.4364,
      "step": 720
    },
    {
      "epoch": 1.0459028151335843,
      "grad_norm": 0.33320216950958276,
      "learning_rate": 8.241530649532339e-05,
      "loss": 0.4252,
      "step": 730
    },
    {
      "epoch": 1.0602474448628294,
      "grad_norm": 0.3367270532117864,
      "learning_rate": 8.177602435915546e-05,
      "loss": 0.4245,
      "step": 740
    },
    {
      "epoch": 1.0745920745920745,
      "grad_norm": 0.36314384787697035,
      "learning_rate": 8.11279067979839e-05,
      "loss": 0.4172,
      "step": 750
    },
    {
      "epoch": 1.0889367043213196,
      "grad_norm": 0.3331116126500675,
      "learning_rate": 8.04711340229654e-05,
      "loss": 0.4117,
      "step": 760
    },
    {
      "epoch": 1.1032813340505647,
      "grad_norm": 0.3316800823164157,
      "learning_rate": 7.980588865186649e-05,
      "loss": 0.4217,
      "step": 770
    },
    {
      "epoch": 1.11762596377981,
      "grad_norm": 0.37471291784176136,
      "learning_rate": 7.913235565828613e-05,
      "loss": 0.4178,
      "step": 780
    },
    {
      "epoch": 1.1319705935090552,
      "grad_norm": 0.3734129628024853,
      "learning_rate": 7.845072232022311e-05,
      "loss": 0.4173,
      "step": 790
    },
    {
      "epoch": 1.1463152232383003,
      "grad_norm": 0.33461223045242555,
      "learning_rate": 7.776117816800288e-05,
      "loss": 0.4258,
      "step": 800
    },
    {
      "epoch": 1.1606598529675454,
      "grad_norm": 0.3149294832232542,
      "learning_rate": 7.706391493157805e-05,
      "loss": 0.4194,
      "step": 810
    },
    {
      "epoch": 1.1750044826967905,
      "grad_norm": 0.33310067113179365,
      "learning_rate": 7.635912648721718e-05,
      "loss": 0.4159,
      "step": 820
    },
    {
      "epoch": 1.1893491124260356,
      "grad_norm": 0.34107614209430825,
      "learning_rate": 7.564700880359696e-05,
      "loss": 0.4193,
      "step": 830
    },
    {
      "epoch": 1.2036937421552807,
      "grad_norm": 0.3033843674878663,
      "learning_rate": 7.492775988731243e-05,
      "loss": 0.4096,
      "step": 840
    },
    {
      "epoch": 1.2180383718845258,
      "grad_norm": 0.36776234094697624,
      "learning_rate": 7.420157972782063e-05,
      "loss": 0.4179,
      "step": 850
    },
    {
      "epoch": 1.232383001613771,
      "grad_norm": 0.32016874059260647,
      "learning_rate": 7.346867024183291e-05,
      "loss": 0.4178,
      "step": 860
    },
    {
      "epoch": 1.246727631343016,
      "grad_norm": 0.3461545416340315,
      "learning_rate": 7.272923521717133e-05,
      "loss": 0.4178,
      "step": 870
    },
    {
      "epoch": 1.2610722610722611,
      "grad_norm": 0.3239854507991741,
      "learning_rate": 7.198348025610481e-05,
      "loss": 0.4173,
      "step": 880
    },
    {
      "epoch": 1.2754168908015062,
      "grad_norm": 0.35967022774745094,
      "learning_rate": 7.12316127181808e-05,
      "loss": 0.4128,
      "step": 890
    },
    {
      "epoch": 1.2897615205307513,
      "grad_norm": 0.33647912683040776,
      "learning_rate": 7.047384166256815e-05,
      "loss": 0.4206,
      "step": 900
    },
    {
      "epoch": 1.3041061502599964,
      "grad_norm": 0.34026212090737123,
      "learning_rate": 6.971037778992775e-05,
      "loss": 0.4196,
      "step": 910
    },
    {
      "epoch": 1.3184507799892415,
      "grad_norm": 0.31786823344694176,
      "learning_rate": 6.894143338382639e-05,
      "loss": 0.4217,
      "step": 920
    },
    {
      "epoch": 1.3327954097184866,
      "grad_norm": 0.3189700837737566,
      "learning_rate": 6.81672222517107e-05,
      "loss": 0.4225,
      "step": 930
    },
    {
      "epoch": 1.3471400394477318,
      "grad_norm": 0.3275473534649835,
      "learning_rate": 6.73879596654573e-05,
      "loss": 0.4119,
      "step": 940
    },
    {
      "epoch": 1.3614846691769769,
      "grad_norm": 0.3420058843642656,
      "learning_rate": 6.660386230151571e-05,
      "loss": 0.4153,
      "step": 950
    },
    {
      "epoch": 1.375829298906222,
      "grad_norm": 0.36451277332483983,
      "learning_rate": 6.581514818066088e-05,
      "loss": 0.4176,
      "step": 960
    },
    {
      "epoch": 1.390173928635467,
      "grad_norm": 0.3392630329016263,
      "learning_rate": 6.502203660737169e-05,
      "loss": 0.4146,
      "step": 970
    },
    {
      "epoch": 1.4045185583647122,
      "grad_norm": 0.32756084558577897,
      "learning_rate": 6.422474810885278e-05,
      "loss": 0.4093,
      "step": 980
    },
    {
      "epoch": 1.4188631880939573,
      "grad_norm": 0.31495606064539233,
      "learning_rate": 6.342350437371614e-05,
      "loss": 0.4127,
      "step": 990
    },
    {
      "epoch": 1.4332078178232024,
      "grad_norm": 0.3106906576340927,
      "learning_rate": 6.26185281903399e-05,
      "loss": 0.4109,
      "step": 1000
    },
    {
      "epoch": 1.4475524475524475,
      "grad_norm": 0.34231901054792907,
      "learning_rate": 6.181004338492141e-05,
      "loss": 0.4115,
      "step": 1010
    },
    {
      "epoch": 1.4618970772816926,
      "grad_norm": 0.2998753745826404,
      "learning_rate": 6.09982747592415e-05,
      "loss": 0.4063,
      "step": 1020
    },
    {
      "epoch": 1.4762417070109377,
      "grad_norm": 0.3030265142945988,
      "learning_rate": 6.018344802815778e-05,
      "loss": 0.4096,
      "step": 1030
    },
    {
      "epoch": 1.4905863367401828,
      "grad_norm": 0.3209761231799724,
      "learning_rate": 5.936578975684378e-05,
      "loss": 0.4115,
      "step": 1040
    },
    {
      "epoch": 1.504930966469428,
      "grad_norm": 0.3117296901575659,
      "learning_rate": 5.854552729779184e-05,
      "loss": 0.4083,
      "step": 1050
    },
    {
      "epoch": 1.519275596198673,
      "grad_norm": 0.32123211938760937,
      "learning_rate": 5.772288872759702e-05,
      "loss": 0.4068,
      "step": 1060
    },
    {
      "epoch": 1.5336202259279181,
      "grad_norm": 0.32646003873154356,
      "learning_rate": 5.6898102783539665e-05,
      "loss": 0.4037,
      "step": 1070
    },
    {
      "epoch": 1.5479648556571632,
      "grad_norm": 0.31930171711882954,
      "learning_rate": 5.607139879998427e-05,
      "loss": 0.4052,
      "step": 1080
    },
    {
      "epoch": 1.5623094853864083,
      "grad_norm": 0.2987369966742978,
      "learning_rate": 5.524300664461235e-05,
      "loss": 0.4041,
      "step": 1090
    },
    {
      "epoch": 1.5766541151156535,
      "grad_norm": 0.29715944144054796,
      "learning_rate": 5.441315665450697e-05,
      "loss": 0.408,
      "step": 1100
    },
    {
      "epoch": 1.5909987448448986,
      "grad_norm": 0.300825472928675,
      "learning_rate": 5.3582079572106794e-05,
      "loss": 0.3996,
      "step": 1110
    },
    {
      "epoch": 1.6053433745741437,
      "grad_norm": 0.321003899975276,
      "learning_rate": 5.275000648104743e-05,
      "loss": 0.4012,
      "step": 1120
    },
    {
      "epoch": 1.6196880043033888,
      "grad_norm": 0.3068035924351862,
      "learning_rate": 5.191716874190785e-05,
      "loss": 0.4056,
      "step": 1130
    },
    {
      "epoch": 1.6340326340326339,
      "grad_norm": 0.3172811217753757,
      "learning_rate": 5.1083797927879896e-05,
      "loss": 0.3995,
      "step": 1140
    },
    {
      "epoch": 1.648377263761879,
      "grad_norm": 0.3135234713854417,
      "learning_rate": 5.025012576037855e-05,
      "loss": 0.4058,
      "step": 1150
    },
    {
      "epoch": 1.6627218934911243,
      "grad_norm": 0.2927786895264955,
      "learning_rate": 4.9416384044611124e-05,
      "loss": 0.4015,
      "step": 1160
    },
    {
      "epoch": 1.6770665232203694,
      "grad_norm": 0.29358825034715325,
      "learning_rate": 4.858280460512302e-05,
      "loss": 0.4013,
      "step": 1170
    },
    {
      "epoch": 1.6914111529496145,
      "grad_norm": 0.3100930655471993,
      "learning_rate": 4.7749619221338227e-05,
      "loss": 0.4009,
      "step": 1180
    },
    {
      "epoch": 1.7057557826788596,
      "grad_norm": 0.307169278710133,
      "learning_rate": 4.691705956311225e-05,
      "loss": 0.3992,
      "step": 1190
    },
    {
      "epoch": 1.7201004124081047,
      "grad_norm": 0.29726825931152523,
      "learning_rate": 4.608535712631566e-05,
      "loss": 0.3941,
      "step": 1200
    },
    {
      "epoch": 1.7344450421373498,
      "grad_norm": 0.2930544538896363,
      "learning_rate": 4.525474316846581e-05,
      "loss": 0.3916,
      "step": 1210
    },
    {
      "epoch": 1.748789671866595,
      "grad_norm": 0.3040077165902238,
      "learning_rate": 4.4425448644425066e-05,
      "loss": 0.3923,
      "step": 1220
    },
    {
      "epoch": 1.76313430159584,
      "grad_norm": 0.29459907021508275,
      "learning_rate": 4.359770414218296e-05,
      "loss": 0.3875,
      "step": 1230
    },
    {
      "epoch": 1.7774789313250852,
      "grad_norm": 0.30065987027605934,
      "learning_rate": 4.2771739818740565e-05,
      "loss": 0.3961,
      "step": 1240
    },
    {
      "epoch": 1.7918235610543303,
      "grad_norm": 0.2920496264668077,
      "learning_rate": 4.194778533611451e-05,
      "loss": 0.3852,
      "step": 1250
    },
    {
      "epoch": 1.8061681907835754,
      "grad_norm": 0.2888363234117279,
      "learning_rate": 4.112606979747881e-05,
      "loss": 0.3937,
      "step": 1260
    },
    {
      "epoch": 1.8205128205128205,
      "grad_norm": 0.30497579196679886,
      "learning_rate": 4.030682168346192e-05,
      "loss": 0.3817,
      "step": 1270
    },
    {
      "epoch": 1.8348574502420656,
      "grad_norm": 0.28969651963041293,
      "learning_rate": 3.949026878861704e-05,
      "loss": 0.383,
      "step": 1280
    },
    {
      "epoch": 1.8492020799713107,
      "grad_norm": 0.28259658296458734,
      "learning_rate": 3.867663815808303e-05,
      "loss": 0.3832,
      "step": 1290
    },
    {
      "epoch": 1.8635467097005558,
      "grad_norm": 0.30667496660431287,
      "learning_rate": 3.78661560244539e-05,
      "loss": 0.3833,
      "step": 1300
    },
    {
      "epoch": 1.8778913394298011,
      "grad_norm": 0.3058644823146065,
      "learning_rate": 3.705904774487396e-05,
      "loss": 0.3794,
      "step": 1310
    },
    {
      "epoch": 1.8922359691590462,
      "grad_norm": 0.2990291886718597,
      "learning_rate": 3.6255537738376706e-05,
      "loss": 0.3874,
      "step": 1320
    },
    {
      "epoch": 1.9065805988882913,
      "grad_norm": 0.2863080421019278,
      "learning_rate": 3.545584942348426e-05,
      "loss": 0.3778,
      "step": 1330
    },
    {
      "epoch": 1.9209252286175365,
      "grad_norm": 0.2900728201465024,
      "learning_rate": 3.466020515608525e-05,
      "loss": 0.3755,
      "step": 1340
    },
    {
      "epoch": 1.9352698583467816,
      "grad_norm": 0.28225638086220933,
      "learning_rate": 3.386882616760794e-05,
      "loss": 0.3794,
      "step": 1350
    },
    {
      "epoch": 1.9496144880760267,
      "grad_norm": 0.29192196316916985,
      "learning_rate": 3.30819325035062e-05,
      "loss": 0.378,
      "step": 1360
    },
    {
      "epoch": 1.9639591178052718,
      "grad_norm": 0.28745199551725825,
      "learning_rate": 3.229974296207513e-05,
      "loss": 0.3812,
      "step": 1370
    },
    {
      "epoch": 1.9783037475345169,
      "grad_norm": 0.29973956860014883,
      "learning_rate": 3.152247503361353e-05,
      "loss": 0.3724,
      "step": 1380
    },
    {
      "epoch": 1.992648377263762,
      "grad_norm": 0.31079745266426567,
      "learning_rate": 3.075034483994997e-05,
      "loss": 0.3718,
      "step": 1390
    },
    {
      "epoch": 2.005737851891698,
      "grad_norm": 0.33118419222329254,
      "learning_rate": 2.998356707434947e-05,
      "loss": 0.2803,
      "step": 1400
    },
    {
      "epoch": 2.020082481620943,
      "grad_norm": 0.2667436179131777,
      "learning_rate": 2.9222354941817375e-05,
      "loss": 0.1535,
      "step": 1410
    },
    {
      "epoch": 2.034427111350188,
      "grad_norm": 0.2667296842605953,
      "learning_rate": 2.846692009981693e-05,
      "loss": 0.1438,
      "step": 1420
    },
    {
      "epoch": 2.0487717410794333,
      "grad_norm": 0.2687925364002219,
      "learning_rate": 2.771747259941734e-05,
      "loss": 0.1383,
      "step": 1430
    },
    {
      "epoch": 2.0631163708086784,
      "grad_norm": 0.258692098758459,
      "learning_rate": 2.6974220826888374e-05,
      "loss": 0.1348,
      "step": 1440
    },
    {
      "epoch": 2.0774610005379235,
      "grad_norm": 0.27396537684578004,
      "learning_rate": 2.623737144575787e-05,
      "loss": 0.1371,
      "step": 1450
    },
    {
      "epoch": 2.0918056302671686,
      "grad_norm": 0.26487337871277594,
      "learning_rate": 2.5507129339348335e-05,
      "loss": 0.132,
      "step": 1460
    },
    {
      "epoch": 2.1061502599964137,
      "grad_norm": 0.26513450922087345,
      "learning_rate": 2.478369755380839e-05,
      "loss": 0.1335,
      "step": 1470
    },
    {
      "epoch": 2.120494889725659,
      "grad_norm": 0.25972172147117933,
      "learning_rate": 2.406727724165524e-05,
      "loss": 0.134,
      "step": 1480
    },
    {
      "epoch": 2.134839519454904,
      "grad_norm": 0.2524599507939378,
      "learning_rate": 2.3358067605843537e-05,
      "loss": 0.1307,
      "step": 1490
    },
    {
      "epoch": 2.149184149184149,
      "grad_norm": 0.8197351837947908,
      "learning_rate": 2.2656265844376367e-05,
      "loss": 0.1332,
      "step": 1500
    }
  ],
  "logging_steps": 10,
  "max_steps": 2094,
  "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": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1017570026586112.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}