{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 19.995409685563462,
  "eval_steps": 500,
  "global_step": 21780,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.09180628873077806,
      "grad_norm": 0.289526104927063,
      "learning_rate": 3e-05,
      "loss": 1.7578,
      "step": 100
    },
    {
      "epoch": 0.18361257746155613,
      "grad_norm": 0.27366071939468384,
      "learning_rate": 3e-05,
      "loss": 1.6025,
      "step": 200
    },
    {
      "epoch": 0.2754188661923342,
      "grad_norm": 0.2666202485561371,
      "learning_rate": 3e-05,
      "loss": 1.5821,
      "step": 300
    },
    {
      "epoch": 0.36722515492311225,
      "grad_norm": 0.25359421968460083,
      "learning_rate": 3e-05,
      "loss": 1.5884,
      "step": 400
    },
    {
      "epoch": 0.4590314436538903,
      "grad_norm": 0.26042038202285767,
      "learning_rate": 3e-05,
      "loss": 1.5873,
      "step": 500
    },
    {
      "epoch": 0.5508377323846684,
      "grad_norm": 0.25545641779899597,
      "learning_rate": 3e-05,
      "loss": 1.5861,
      "step": 600
    },
    {
      "epoch": 0.6426440211154464,
      "grad_norm": 0.2685534358024597,
      "learning_rate": 3e-05,
      "loss": 1.5889,
      "step": 700
    },
    {
      "epoch": 0.7344503098462245,
      "grad_norm": 0.30292272567749023,
      "learning_rate": 3e-05,
      "loss": 1.6001,
      "step": 800
    },
    {
      "epoch": 0.8262565985770025,
      "grad_norm": 0.3182919919490814,
      "learning_rate": 3e-05,
      "loss": 1.5712,
      "step": 900
    },
    {
      "epoch": 0.9180628873077806,
      "grad_norm": 0.31249287724494934,
      "learning_rate": 3e-05,
      "loss": 1.584,
      "step": 1000
    },
    {
      "epoch": 0.999770484278173,
      "eval_accuracy": 0.6762416302765648,
      "eval_loss": 1.5102524757385254,
      "eval_runtime": 8.9715,
      "eval_samples_per_second": 55.732,
      "eval_steps_per_second": 7.022,
      "step": 1089
    },
    {
      "epoch": 1.0098691760385587,
      "grad_norm": 0.2788257300853729,
      "learning_rate": 3e-05,
      "loss": 1.5741,
      "step": 1100
    },
    {
      "epoch": 1.1016754647693368,
      "grad_norm": 0.3435765504837036,
      "learning_rate": 3e-05,
      "loss": 1.5538,
      "step": 1200
    },
    {
      "epoch": 1.1934817535001148,
      "grad_norm": 0.42897695302963257,
      "learning_rate": 3e-05,
      "loss": 1.548,
      "step": 1300
    },
    {
      "epoch": 1.2852880422308928,
      "grad_norm": 0.3922116160392761,
      "learning_rate": 3e-05,
      "loss": 1.5439,
      "step": 1400
    },
    {
      "epoch": 1.377094330961671,
      "grad_norm": 0.39680397510528564,
      "learning_rate": 3e-05,
      "loss": 1.5422,
      "step": 1500
    },
    {
      "epoch": 1.468900619692449,
      "grad_norm": 0.42543351650238037,
      "learning_rate": 3e-05,
      "loss": 1.5569,
      "step": 1600
    },
    {
      "epoch": 1.560706908423227,
      "grad_norm": 0.41228362917900085,
      "learning_rate": 3e-05,
      "loss": 1.5229,
      "step": 1700
    },
    {
      "epoch": 1.652513197154005,
      "grad_norm": 0.46888694167137146,
      "learning_rate": 3e-05,
      "loss": 1.5442,
      "step": 1800
    },
    {
      "epoch": 1.744319485884783,
      "grad_norm": 0.4511169195175171,
      "learning_rate": 3e-05,
      "loss": 1.538,
      "step": 1900
    },
    {
      "epoch": 1.836125774615561,
      "grad_norm": 0.46713030338287354,
      "learning_rate": 3e-05,
      "loss": 1.5343,
      "step": 2000
    },
    {
      "epoch": 1.9279320633463393,
      "grad_norm": 0.5275651812553406,
      "learning_rate": 3e-05,
      "loss": 1.5504,
      "step": 2100
    },
    {
      "epoch": 1.999540968556346,
      "eval_accuracy": 0.6791033478893741,
      "eval_loss": 1.4772261381149292,
      "eval_runtime": 9.0172,
      "eval_samples_per_second": 55.45,
      "eval_steps_per_second": 6.987,
      "step": 2178
    },
    {
      "epoch": 2.0197383520771175,
      "grad_norm": 0.47441884875297546,
      "learning_rate": 3e-05,
      "loss": 1.5331,
      "step": 2200
    },
    {
      "epoch": 2.1115446408078955,
      "grad_norm": 0.47345635294914246,
      "learning_rate": 3e-05,
      "loss": 1.4902,
      "step": 2300
    },
    {
      "epoch": 2.2033509295386735,
      "grad_norm": 0.592490017414093,
      "learning_rate": 3e-05,
      "loss": 1.4996,
      "step": 2400
    },
    {
      "epoch": 2.2951572182694515,
      "grad_norm": 0.5280076265335083,
      "learning_rate": 3e-05,
      "loss": 1.4949,
      "step": 2500
    },
    {
      "epoch": 2.3869635070002295,
      "grad_norm": 0.5791444182395935,
      "learning_rate": 3e-05,
      "loss": 1.499,
      "step": 2600
    },
    {
      "epoch": 2.4787697957310075,
      "grad_norm": 0.5848264098167419,
      "learning_rate": 3e-05,
      "loss": 1.479,
      "step": 2700
    },
    {
      "epoch": 2.5705760844617855,
      "grad_norm": 0.5598397254943848,
      "learning_rate": 3e-05,
      "loss": 1.4938,
      "step": 2800
    },
    {
      "epoch": 2.6623823731925635,
      "grad_norm": 0.5774019360542297,
      "learning_rate": 3e-05,
      "loss": 1.4884,
      "step": 2900
    },
    {
      "epoch": 2.754188661923342,
      "grad_norm": 0.6278976202011108,
      "learning_rate": 3e-05,
      "loss": 1.5027,
      "step": 3000
    },
    {
      "epoch": 2.84599495065412,
      "grad_norm": 0.5700748562812805,
      "learning_rate": 3e-05,
      "loss": 1.4777,
      "step": 3100
    },
    {
      "epoch": 2.937801239384898,
      "grad_norm": 0.6338950395584106,
      "learning_rate": 3e-05,
      "loss": 1.4842,
      "step": 3200
    },
    {
      "epoch": 2.9993114528345193,
      "eval_accuracy": 0.6811615720524018,
      "eval_loss": 1.4502207040786743,
      "eval_runtime": 9.3503,
      "eval_samples_per_second": 53.474,
      "eval_steps_per_second": 6.738,
      "step": 3267
    },
    {
      "epoch": 3.029607528115676,
      "grad_norm": 0.6205599904060364,
      "learning_rate": 3e-05,
      "loss": 1.4695,
      "step": 3300
    },
    {
      "epoch": 3.121413816846454,
      "grad_norm": 0.6470921635627747,
      "learning_rate": 3e-05,
      "loss": 1.4401,
      "step": 3400
    },
    {
      "epoch": 3.213220105577232,
      "grad_norm": 0.8113517761230469,
      "learning_rate": 3e-05,
      "loss": 1.4312,
      "step": 3500
    },
    {
      "epoch": 3.30502639430801,
      "grad_norm": 0.6976670026779175,
      "learning_rate": 3e-05,
      "loss": 1.4548,
      "step": 3600
    },
    {
      "epoch": 3.396832683038788,
      "grad_norm": 0.7569802403450012,
      "learning_rate": 3e-05,
      "loss": 1.4447,
      "step": 3700
    },
    {
      "epoch": 3.488638971769566,
      "grad_norm": 0.8047822117805481,
      "learning_rate": 3e-05,
      "loss": 1.434,
      "step": 3800
    },
    {
      "epoch": 3.580445260500344,
      "grad_norm": 0.710166871547699,
      "learning_rate": 3e-05,
      "loss": 1.4283,
      "step": 3900
    },
    {
      "epoch": 3.672251549231122,
      "grad_norm": 0.7864311933517456,
      "learning_rate": 3e-05,
      "loss": 1.4493,
      "step": 4000
    },
    {
      "epoch": 3.7640578379619005,
      "grad_norm": 0.7331141829490662,
      "learning_rate": 3e-05,
      "loss": 1.4259,
      "step": 4100
    },
    {
      "epoch": 3.8558641266926785,
      "grad_norm": 0.7041341662406921,
      "learning_rate": 3e-05,
      "loss": 1.4316,
      "step": 4200
    },
    {
      "epoch": 3.9476704154234565,
      "grad_norm": 0.6956498622894287,
      "learning_rate": 3e-05,
      "loss": 1.427,
      "step": 4300
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6837205240174673,
      "eval_loss": 1.4199937582015991,
      "eval_runtime": 9.1672,
      "eval_samples_per_second": 54.542,
      "eval_steps_per_second": 6.872,
      "step": 4357
    },
    {
      "epoch": 4.039476704154235,
      "grad_norm": 0.7389739751815796,
      "learning_rate": 3e-05,
      "loss": 1.4099,
      "step": 4400
    },
    {
      "epoch": 4.131282992885013,
      "grad_norm": 0.7883840203285217,
      "learning_rate": 3e-05,
      "loss": 1.3709,
      "step": 4500
    },
    {
      "epoch": 4.223089281615791,
      "grad_norm": 0.7341814041137695,
      "learning_rate": 3e-05,
      "loss": 1.3743,
      "step": 4600
    },
    {
      "epoch": 4.314895570346569,
      "grad_norm": 0.9007183909416199,
      "learning_rate": 3e-05,
      "loss": 1.3929,
      "step": 4700
    },
    {
      "epoch": 4.406701859077347,
      "grad_norm": 0.8208268284797668,
      "learning_rate": 3e-05,
      "loss": 1.3843,
      "step": 4800
    },
    {
      "epoch": 4.498508147808125,
      "grad_norm": 0.7786016464233398,
      "learning_rate": 3e-05,
      "loss": 1.3789,
      "step": 4900
    },
    {
      "epoch": 4.590314436538903,
      "grad_norm": 0.9414695501327515,
      "learning_rate": 3e-05,
      "loss": 1.3812,
      "step": 5000
    },
    {
      "epoch": 4.682120725269681,
      "grad_norm": 0.7854552268981934,
      "learning_rate": 3e-05,
      "loss": 1.3851,
      "step": 5100
    },
    {
      "epoch": 4.773927014000459,
      "grad_norm": 0.8319596648216248,
      "learning_rate": 3e-05,
      "loss": 1.3846,
      "step": 5200
    },
    {
      "epoch": 4.865733302731237,
      "grad_norm": 0.8832118511199951,
      "learning_rate": 3e-05,
      "loss": 1.3845,
      "step": 5300
    },
    {
      "epoch": 4.957539591462015,
      "grad_norm": 0.8607555627822876,
      "learning_rate": 3e-05,
      "loss": 1.3827,
      "step": 5400
    },
    {
      "epoch": 4.999770484278173,
      "eval_accuracy": 0.6860465793304221,
      "eval_loss": 1.3910759687423706,
      "eval_runtime": 8.9399,
      "eval_samples_per_second": 55.929,
      "eval_steps_per_second": 7.047,
      "step": 5446
    },
    {
      "epoch": 5.049345880192793,
      "grad_norm": 0.8777875900268555,
      "learning_rate": 3e-05,
      "loss": 1.3512,
      "step": 5500
    },
    {
      "epoch": 5.141152168923571,
      "grad_norm": 0.9193658232688904,
      "learning_rate": 3e-05,
      "loss": 1.3124,
      "step": 5600
    },
    {
      "epoch": 5.232958457654349,
      "grad_norm": 0.9822832345962524,
      "learning_rate": 3e-05,
      "loss": 1.3189,
      "step": 5700
    },
    {
      "epoch": 5.324764746385127,
      "grad_norm": 0.9231218099594116,
      "learning_rate": 3e-05,
      "loss": 1.3321,
      "step": 5800
    },
    {
      "epoch": 5.416571035115905,
      "grad_norm": 0.961618185043335,
      "learning_rate": 3e-05,
      "loss": 1.3275,
      "step": 5900
    },
    {
      "epoch": 5.508377323846684,
      "grad_norm": 1.1759928464889526,
      "learning_rate": 3e-05,
      "loss": 1.3301,
      "step": 6000
    },
    {
      "epoch": 5.600183612577462,
      "grad_norm": 1.0055111646652222,
      "learning_rate": 3e-05,
      "loss": 1.3261,
      "step": 6100
    },
    {
      "epoch": 5.69198990130824,
      "grad_norm": 0.9605348110198975,
      "learning_rate": 3e-05,
      "loss": 1.3329,
      "step": 6200
    },
    {
      "epoch": 5.783796190039018,
      "grad_norm": 1.0969476699829102,
      "learning_rate": 3e-05,
      "loss": 1.3347,
      "step": 6300
    },
    {
      "epoch": 5.875602478769796,
      "grad_norm": 0.9841852188110352,
      "learning_rate": 3e-05,
      "loss": 1.3215,
      "step": 6400
    },
    {
      "epoch": 5.967408767500574,
      "grad_norm": 1.0173933506011963,
      "learning_rate": 3e-05,
      "loss": 1.3425,
      "step": 6500
    },
    {
      "epoch": 5.999540968556346,
      "eval_accuracy": 0.6886841339155749,
      "eval_loss": 1.3614963293075562,
      "eval_runtime": 8.9983,
      "eval_samples_per_second": 55.566,
      "eval_steps_per_second": 7.001,
      "step": 6535
    },
    {
      "epoch": 6.059215056231352,
      "grad_norm": 0.9646373987197876,
      "learning_rate": 3e-05,
      "loss": 1.287,
      "step": 6600
    },
    {
      "epoch": 6.15102134496213,
      "grad_norm": 0.8929613828659058,
      "learning_rate": 3e-05,
      "loss": 1.2574,
      "step": 6700
    },
    {
      "epoch": 6.242827633692908,
      "grad_norm": 1.285346508026123,
      "learning_rate": 3e-05,
      "loss": 1.2762,
      "step": 6800
    },
    {
      "epoch": 6.334633922423686,
      "grad_norm": 1.102123498916626,
      "learning_rate": 3e-05,
      "loss": 1.2667,
      "step": 6900
    },
    {
      "epoch": 6.426440211154464,
      "grad_norm": 1.021745204925537,
      "learning_rate": 3e-05,
      "loss": 1.2698,
      "step": 7000
    },
    {
      "epoch": 6.518246499885242,
      "grad_norm": 1.1759482622146606,
      "learning_rate": 3e-05,
      "loss": 1.2781,
      "step": 7100
    },
    {
      "epoch": 6.61005278861602,
      "grad_norm": 1.2193723917007446,
      "learning_rate": 3e-05,
      "loss": 1.2784,
      "step": 7200
    },
    {
      "epoch": 6.701859077346798,
      "grad_norm": 1.1053309440612793,
      "learning_rate": 3e-05,
      "loss": 1.2732,
      "step": 7300
    },
    {
      "epoch": 6.793665366077576,
      "grad_norm": 1.7023396492004395,
      "learning_rate": 3e-05,
      "loss": 1.269,
      "step": 7400
    },
    {
      "epoch": 6.885471654808354,
      "grad_norm": 1.0934760570526123,
      "learning_rate": 3e-05,
      "loss": 1.2691,
      "step": 7500
    },
    {
      "epoch": 6.977277943539132,
      "grad_norm": 1.0586143732070923,
      "learning_rate": 3e-05,
      "loss": 1.2738,
      "step": 7600
    },
    {
      "epoch": 6.999311452834519,
      "eval_accuracy": 0.6910072780203784,
      "eval_loss": 1.3299835920333862,
      "eval_runtime": 9.0186,
      "eval_samples_per_second": 55.441,
      "eval_steps_per_second": 6.986,
      "step": 7624
    },
    {
      "epoch": 7.06908423226991,
      "grad_norm": 1.2295095920562744,
      "learning_rate": 3e-05,
      "loss": 1.2221,
      "step": 7700
    },
    {
      "epoch": 7.160890521000688,
      "grad_norm": 1.0440340042114258,
      "learning_rate": 3e-05,
      "loss": 1.2119,
      "step": 7800
    },
    {
      "epoch": 7.252696809731467,
      "grad_norm": 1.1318169832229614,
      "learning_rate": 3e-05,
      "loss": 1.2061,
      "step": 7900
    },
    {
      "epoch": 7.344503098462245,
      "grad_norm": 1.2543174028396606,
      "learning_rate": 3e-05,
      "loss": 1.2222,
      "step": 8000
    },
    {
      "epoch": 7.436309387193023,
      "grad_norm": 1.098528504371643,
      "learning_rate": 3e-05,
      "loss": 1.211,
      "step": 8100
    },
    {
      "epoch": 7.528115675923801,
      "grad_norm": 1.5505329370498657,
      "learning_rate": 3e-05,
      "loss": 1.2313,
      "step": 8200
    },
    {
      "epoch": 7.619921964654579,
      "grad_norm": 1.2159889936447144,
      "learning_rate": 3e-05,
      "loss": 1.2155,
      "step": 8300
    },
    {
      "epoch": 7.711728253385357,
      "grad_norm": 1.2545368671417236,
      "learning_rate": 3e-05,
      "loss": 1.2125,
      "step": 8400
    },
    {
      "epoch": 7.803534542116135,
      "grad_norm": 1.0893586874008179,
      "learning_rate": 3e-05,
      "loss": 1.2067,
      "step": 8500
    },
    {
      "epoch": 7.895340830846913,
      "grad_norm": 1.2962942123413086,
      "learning_rate": 3e-05,
      "loss": 1.2123,
      "step": 8600
    },
    {
      "epoch": 7.987147119577691,
      "grad_norm": 1.0764884948730469,
      "learning_rate": 3e-05,
      "loss": 1.2283,
      "step": 8700
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.6934876273653566,
      "eval_loss": 1.3020232915878296,
      "eval_runtime": 8.9433,
      "eval_samples_per_second": 55.908,
      "eval_steps_per_second": 7.044,
      "step": 8714
    },
    {
      "epoch": 8.07895340830847,
      "grad_norm": 1.2428829669952393,
      "learning_rate": 3e-05,
      "loss": 1.1629,
      "step": 8800
    },
    {
      "epoch": 8.170759697039248,
      "grad_norm": 1.19692862033844,
      "learning_rate": 3e-05,
      "loss": 1.1448,
      "step": 8900
    },
    {
      "epoch": 8.262565985770026,
      "grad_norm": 1.3424954414367676,
      "learning_rate": 3e-05,
      "loss": 1.1617,
      "step": 9000
    },
    {
      "epoch": 8.354372274500804,
      "grad_norm": 1.3499901294708252,
      "learning_rate": 3e-05,
      "loss": 1.1564,
      "step": 9100
    },
    {
      "epoch": 8.446178563231582,
      "grad_norm": 1.183600664138794,
      "learning_rate": 3e-05,
      "loss": 1.1524,
      "step": 9200
    },
    {
      "epoch": 8.53798485196236,
      "grad_norm": 1.3151459693908691,
      "learning_rate": 3e-05,
      "loss": 1.1499,
      "step": 9300
    },
    {
      "epoch": 8.629791140693138,
      "grad_norm": 1.3484901189804077,
      "learning_rate": 3e-05,
      "loss": 1.1716,
      "step": 9400
    },
    {
      "epoch": 8.721597429423916,
      "grad_norm": 1.326663851737976,
      "learning_rate": 3e-05,
      "loss": 1.1665,
      "step": 9500
    },
    {
      "epoch": 8.813403718154694,
      "grad_norm": 1.315091609954834,
      "learning_rate": 3e-05,
      "loss": 1.1713,
      "step": 9600
    },
    {
      "epoch": 8.905210006885472,
      "grad_norm": 1.3116127252578735,
      "learning_rate": 3e-05,
      "loss": 1.1669,
      "step": 9700
    },
    {
      "epoch": 8.99701629561625,
      "grad_norm": 1.4249849319458008,
      "learning_rate": 3e-05,
      "loss": 1.1788,
      "step": 9800
    },
    {
      "epoch": 8.999770484278173,
      "eval_accuracy": 0.6963406113537118,
      "eval_loss": 1.272213339805603,
      "eval_runtime": 8.9482,
      "eval_samples_per_second": 55.877,
      "eval_steps_per_second": 7.041,
      "step": 9803
    },
    {
      "epoch": 9.088822584347028,
      "grad_norm": 1.412617802619934,
      "learning_rate": 3e-05,
      "loss": 1.0912,
      "step": 9900
    },
    {
      "epoch": 9.180628873077806,
      "grad_norm": 1.2806413173675537,
      "learning_rate": 3e-05,
      "loss": 1.1013,
      "step": 10000
    },
    {
      "epoch": 9.272435161808584,
      "grad_norm": 1.8053137063980103,
      "learning_rate": 3e-05,
      "loss": 1.1071,
      "step": 10100
    },
    {
      "epoch": 9.364241450539362,
      "grad_norm": 1.352771520614624,
      "learning_rate": 3e-05,
      "loss": 1.1043,
      "step": 10200
    },
    {
      "epoch": 9.45604773927014,
      "grad_norm": 1.5698919296264648,
      "learning_rate": 3e-05,
      "loss": 1.1187,
      "step": 10300
    },
    {
      "epoch": 9.547854028000918,
      "grad_norm": 1.4473572969436646,
      "learning_rate": 3e-05,
      "loss": 1.0991,
      "step": 10400
    },
    {
      "epoch": 9.639660316731696,
      "grad_norm": 1.5458990335464478,
      "learning_rate": 3e-05,
      "loss": 1.1168,
      "step": 10500
    },
    {
      "epoch": 9.731466605462474,
      "grad_norm": 1.3577615022659302,
      "learning_rate": 3e-05,
      "loss": 1.1081,
      "step": 10600
    },
    {
      "epoch": 9.823272894193252,
      "grad_norm": 1.6055794954299927,
      "learning_rate": 3e-05,
      "loss": 1.1117,
      "step": 10700
    },
    {
      "epoch": 9.91507918292403,
      "grad_norm": 1.5611170530319214,
      "learning_rate": 3e-05,
      "loss": 1.1156,
      "step": 10800
    },
    {
      "epoch": 9.999540968556346,
      "eval_accuracy": 0.6990072780203784,
      "eval_loss": 1.2414618730545044,
      "eval_runtime": 9.5021,
      "eval_samples_per_second": 52.62,
      "eval_steps_per_second": 6.63,
      "step": 10892
    },
    {
      "epoch": 10.006885471654808,
      "grad_norm": 1.4584790468215942,
      "learning_rate": 3e-05,
      "loss": 1.1099,
      "step": 10900
    },
    {
      "epoch": 10.098691760385586,
      "grad_norm": 1.4327212572097778,
      "learning_rate": 3e-05,
      "loss": 1.039,
      "step": 11000
    },
    {
      "epoch": 10.190498049116364,
      "grad_norm": 1.4160873889923096,
      "learning_rate": 3e-05,
      "loss": 1.0578,
      "step": 11100
    },
    {
      "epoch": 10.282304337847142,
      "grad_norm": 1.506165862083435,
      "learning_rate": 3e-05,
      "loss": 1.0481,
      "step": 11200
    },
    {
      "epoch": 10.37411062657792,
      "grad_norm": 1.6476013660430908,
      "learning_rate": 3e-05,
      "loss": 1.0582,
      "step": 11300
    },
    {
      "epoch": 10.465916915308698,
      "grad_norm": 1.4203314781188965,
      "learning_rate": 3e-05,
      "loss": 1.0615,
      "step": 11400
    },
    {
      "epoch": 10.557723204039476,
      "grad_norm": 1.591191053390503,
      "learning_rate": 3e-05,
      "loss": 1.0654,
      "step": 11500
    },
    {
      "epoch": 10.649529492770254,
      "grad_norm": 1.552139401435852,
      "learning_rate": 3e-05,
      "loss": 1.043,
      "step": 11600
    },
    {
      "epoch": 10.741335781501032,
      "grad_norm": 1.5005476474761963,
      "learning_rate": 3e-05,
      "loss": 1.0518,
      "step": 11700
    },
    {
      "epoch": 10.83314207023181,
      "grad_norm": 1.6541969776153564,
      "learning_rate": 3e-05,
      "loss": 1.0617,
      "step": 11800
    },
    {
      "epoch": 10.924948358962588,
      "grad_norm": 1.4498178958892822,
      "learning_rate": 3e-05,
      "loss": 1.0526,
      "step": 11900
    },
    {
      "epoch": 10.999311452834519,
      "eval_accuracy": 0.7014876273653566,
      "eval_loss": 1.2131479978561401,
      "eval_runtime": 9.02,
      "eval_samples_per_second": 55.432,
      "eval_steps_per_second": 6.984,
      "step": 11981
    },
    {
      "epoch": 11.016754647693366,
      "grad_norm": 1.384069561958313,
      "learning_rate": 3e-05,
      "loss": 1.0554,
      "step": 12000
    },
    {
      "epoch": 11.108560936424144,
      "grad_norm": 1.4845672845840454,
      "learning_rate": 3e-05,
      "loss": 0.9886,
      "step": 12100
    },
    {
      "epoch": 11.200367225154924,
      "grad_norm": 1.7744626998901367,
      "learning_rate": 3e-05,
      "loss": 0.9969,
      "step": 12200
    },
    {
      "epoch": 11.292173513885702,
      "grad_norm": 1.6337647438049316,
      "learning_rate": 3e-05,
      "loss": 0.9899,
      "step": 12300
    },
    {
      "epoch": 11.38397980261648,
      "grad_norm": 2.003005266189575,
      "learning_rate": 3e-05,
      "loss": 1.0111,
      "step": 12400
    },
    {
      "epoch": 11.475786091347258,
      "grad_norm": 1.968371033668518,
      "learning_rate": 3e-05,
      "loss": 1.0012,
      "step": 12500
    },
    {
      "epoch": 11.567592380078036,
      "grad_norm": 1.6538879871368408,
      "learning_rate": 3e-05,
      "loss": 0.9972,
      "step": 12600
    },
    {
      "epoch": 11.659398668808814,
      "grad_norm": 1.6392265558242798,
      "learning_rate": 3e-05,
      "loss": 1.0084,
      "step": 12700
    },
    {
      "epoch": 11.751204957539592,
      "grad_norm": 1.7361793518066406,
      "learning_rate": 3e-05,
      "loss": 1.0109,
      "step": 12800
    },
    {
      "epoch": 11.84301124627037,
      "grad_norm": 1.4300850629806519,
      "learning_rate": 3e-05,
      "loss": 1.0163,
      "step": 12900
    },
    {
      "epoch": 11.934817535001148,
      "grad_norm": 1.6984518766403198,
      "learning_rate": 3e-05,
      "loss": 1.0146,
      "step": 13000
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.7045123726346434,
      "eval_loss": 1.1803950071334839,
      "eval_runtime": 8.9699,
      "eval_samples_per_second": 55.742,
      "eval_steps_per_second": 7.024,
      "step": 13071
    },
    {
      "epoch": 12.026623823731926,
      "grad_norm": 1.52531898021698,
      "learning_rate": 3e-05,
      "loss": 0.9803,
      "step": 13100
    },
    {
      "epoch": 12.118430112462704,
      "grad_norm": 1.8437868356704712,
      "learning_rate": 3e-05,
      "loss": 0.9362,
      "step": 13200
    },
    {
      "epoch": 12.210236401193482,
      "grad_norm": 1.7236285209655762,
      "learning_rate": 3e-05,
      "loss": 0.9476,
      "step": 13300
    },
    {
      "epoch": 12.30204268992426,
      "grad_norm": 1.7923431396484375,
      "learning_rate": 3e-05,
      "loss": 0.9473,
      "step": 13400
    },
    {
      "epoch": 12.393848978655038,
      "grad_norm": 1.9459409713745117,
      "learning_rate": 3e-05,
      "loss": 0.9521,
      "step": 13500
    },
    {
      "epoch": 12.485655267385816,
      "grad_norm": 1.8831307888031006,
      "learning_rate": 3e-05,
      "loss": 0.942,
      "step": 13600
    },
    {
      "epoch": 12.577461556116594,
      "grad_norm": 1.629230260848999,
      "learning_rate": 3e-05,
      "loss": 0.9558,
      "step": 13700
    },
    {
      "epoch": 12.669267844847372,
      "grad_norm": 1.5318315029144287,
      "learning_rate": 3e-05,
      "loss": 0.9525,
      "step": 13800
    },
    {
      "epoch": 12.76107413357815,
      "grad_norm": 1.611336588859558,
      "learning_rate": 3e-05,
      "loss": 0.9619,
      "step": 13900
    },
    {
      "epoch": 12.852880422308928,
      "grad_norm": 1.6721709966659546,
      "learning_rate": 3e-05,
      "loss": 0.9619,
      "step": 14000
    },
    {
      "epoch": 12.944686711039706,
      "grad_norm": 1.8074623346328735,
      "learning_rate": 3e-05,
      "loss": 0.9613,
      "step": 14100
    },
    {
      "epoch": 12.999770484278173,
      "eval_accuracy": 0.7071382823871907,
      "eval_loss": 1.1507638692855835,
      "eval_runtime": 8.9983,
      "eval_samples_per_second": 55.566,
      "eval_steps_per_second": 7.001,
      "step": 14160
    },
    {
      "epoch": 13.036492999770484,
      "grad_norm": 2.195594549179077,
      "learning_rate": 3e-05,
      "loss": 0.9259,
      "step": 14200
    },
    {
      "epoch": 13.128299288501262,
      "grad_norm": 1.8173458576202393,
      "learning_rate": 3e-05,
      "loss": 0.8902,
      "step": 14300
    },
    {
      "epoch": 13.22010557723204,
      "grad_norm": 1.7481939792633057,
      "learning_rate": 3e-05,
      "loss": 0.8859,
      "step": 14400
    },
    {
      "epoch": 13.311911865962818,
      "grad_norm": 1.938438892364502,
      "learning_rate": 3e-05,
      "loss": 0.899,
      "step": 14500
    },
    {
      "epoch": 13.403718154693596,
      "grad_norm": 1.8565011024475098,
      "learning_rate": 3e-05,
      "loss": 0.889,
      "step": 14600
    },
    {
      "epoch": 13.495524443424374,
      "grad_norm": 1.6509944200515747,
      "learning_rate": 3e-05,
      "loss": 0.9182,
      "step": 14700
    },
    {
      "epoch": 13.587330732155152,
      "grad_norm": 1.9225726127624512,
      "learning_rate": 3e-05,
      "loss": 0.9091,
      "step": 14800
    },
    {
      "epoch": 13.67913702088593,
      "grad_norm": 1.7917280197143555,
      "learning_rate": 3e-05,
      "loss": 0.9137,
      "step": 14900
    },
    {
      "epoch": 13.770943309616708,
      "grad_norm": 2.0736453533172607,
      "learning_rate": 3e-05,
      "loss": 0.904,
      "step": 15000
    },
    {
      "epoch": 13.862749598347486,
      "grad_norm": 2.1191747188568115,
      "learning_rate": 3e-05,
      "loss": 0.9146,
      "step": 15100
    },
    {
      "epoch": 13.954555887078264,
      "grad_norm": 1.8331027030944824,
      "learning_rate": 3e-05,
      "loss": 0.9109,
      "step": 15200
    },
    {
      "epoch": 13.999540968556346,
      "eval_accuracy": 0.7097409024745269,
      "eval_loss": 1.12144935131073,
      "eval_runtime": 8.9899,
      "eval_samples_per_second": 55.618,
      "eval_steps_per_second": 7.008,
      "step": 15249
    },
    {
      "epoch": 14.046362175809042,
      "grad_norm": 1.818524718284607,
      "learning_rate": 3e-05,
      "loss": 0.8787,
      "step": 15300
    },
    {
      "epoch": 14.13816846453982,
      "grad_norm": 1.9324177503585815,
      "learning_rate": 3e-05,
      "loss": 0.8487,
      "step": 15400
    },
    {
      "epoch": 14.229974753270598,
      "grad_norm": 1.952480435371399,
      "learning_rate": 3e-05,
      "loss": 0.8526,
      "step": 15500
    },
    {
      "epoch": 14.321781042001376,
      "grad_norm": 1.9058892726898193,
      "learning_rate": 3e-05,
      "loss": 0.8433,
      "step": 15600
    },
    {
      "epoch": 14.413587330732156,
      "grad_norm": 1.9198521375656128,
      "learning_rate": 3e-05,
      "loss": 0.8591,
      "step": 15700
    },
    {
      "epoch": 14.505393619462934,
      "grad_norm": 2.374208927154541,
      "learning_rate": 3e-05,
      "loss": 0.8643,
      "step": 15800
    },
    {
      "epoch": 14.597199908193712,
      "grad_norm": 1.8864604234695435,
      "learning_rate": 3e-05,
      "loss": 0.8623,
      "step": 15900
    },
    {
      "epoch": 14.68900619692449,
      "grad_norm": 1.9877722263336182,
      "learning_rate": 3e-05,
      "loss": 0.8569,
      "step": 16000
    },
    {
      "epoch": 14.780812485655268,
      "grad_norm": 2.204672336578369,
      "learning_rate": 3e-05,
      "loss": 0.8629,
      "step": 16100
    },
    {
      "epoch": 14.872618774386046,
      "grad_norm": 1.790323257446289,
      "learning_rate": 3e-05,
      "loss": 0.8542,
      "step": 16200
    },
    {
      "epoch": 14.964425063116824,
      "grad_norm": 1.8623679876327515,
      "learning_rate": 3e-05,
      "loss": 0.8566,
      "step": 16300
    },
    {
      "epoch": 14.999311452834519,
      "eval_accuracy": 0.7127714701601164,
      "eval_loss": 1.0913478136062622,
      "eval_runtime": 8.9695,
      "eval_samples_per_second": 55.745,
      "eval_steps_per_second": 7.024,
      "step": 16338
    },
    {
      "epoch": 15.056231351847602,
      "grad_norm": 1.6198936700820923,
      "learning_rate": 3e-05,
      "loss": 0.8233,
      "step": 16400
    },
    {
      "epoch": 15.14803764057838,
      "grad_norm": 2.117966413497925,
      "learning_rate": 3e-05,
      "loss": 0.8004,
      "step": 16500
    },
    {
      "epoch": 15.239843929309158,
      "grad_norm": 1.9046192169189453,
      "learning_rate": 3e-05,
      "loss": 0.783,
      "step": 16600
    },
    {
      "epoch": 15.331650218039936,
      "grad_norm": 1.8354123830795288,
      "learning_rate": 3e-05,
      "loss": 0.8123,
      "step": 16700
    },
    {
      "epoch": 15.423456506770714,
      "grad_norm": 1.8810902833938599,
      "learning_rate": 3e-05,
      "loss": 0.8062,
      "step": 16800
    },
    {
      "epoch": 15.515262795501492,
      "grad_norm": 2.2442831993103027,
      "learning_rate": 3e-05,
      "loss": 0.8121,
      "step": 16900
    },
    {
      "epoch": 15.60706908423227,
      "grad_norm": 2.308647394180298,
      "learning_rate": 3e-05,
      "loss": 0.8155,
      "step": 17000
    },
    {
      "epoch": 15.698875372963048,
      "grad_norm": 2.2714340686798096,
      "learning_rate": 3e-05,
      "loss": 0.8211,
      "step": 17100
    },
    {
      "epoch": 15.790681661693826,
      "grad_norm": 1.9850467443466187,
      "learning_rate": 3e-05,
      "loss": 0.8054,
      "step": 17200
    },
    {
      "epoch": 15.882487950424604,
      "grad_norm": 2.5280234813690186,
      "learning_rate": 3e-05,
      "loss": 0.8198,
      "step": 17300
    },
    {
      "epoch": 15.974294239155382,
      "grad_norm": 2.184380292892456,
      "learning_rate": 3e-05,
      "loss": 0.8307,
      "step": 17400
    },
    {
      "epoch": 16.0,
      "eval_accuracy": 0.7155633187772926,
      "eval_loss": 1.0599175691604614,
      "eval_runtime": 8.9521,
      "eval_samples_per_second": 55.853,
      "eval_steps_per_second": 7.037,
      "step": 17428
    },
    {
      "epoch": 16.06610052788616,
      "grad_norm": 2.437701463699341,
      "learning_rate": 3e-05,
      "loss": 0.7609,
      "step": 17500
    },
    {
      "epoch": 16.15790681661694,
      "grad_norm": 2.636090040206909,
      "learning_rate": 3e-05,
      "loss": 0.758,
      "step": 17600
    },
    {
      "epoch": 16.249713105347716,
      "grad_norm": 2.1846566200256348,
      "learning_rate": 3e-05,
      "loss": 0.7439,
      "step": 17700
    },
    {
      "epoch": 16.341519394078496,
      "grad_norm": 2.1148085594177246,
      "learning_rate": 3e-05,
      "loss": 0.7568,
      "step": 17800
    },
    {
      "epoch": 16.433325682809272,
      "grad_norm": 1.8323599100112915,
      "learning_rate": 3e-05,
      "loss": 0.7678,
      "step": 17900
    },
    {
      "epoch": 16.525131971540052,
      "grad_norm": 2.67404842376709,
      "learning_rate": 3e-05,
      "loss": 0.7719,
      "step": 18000
    },
    {
      "epoch": 16.616938260270828,
      "grad_norm": 2.3159210681915283,
      "learning_rate": 3e-05,
      "loss": 0.78,
      "step": 18100
    },
    {
      "epoch": 16.708744549001608,
      "grad_norm": 1.924141526222229,
      "learning_rate": 3e-05,
      "loss": 0.7774,
      "step": 18200
    },
    {
      "epoch": 16.800550837732384,
      "grad_norm": 1.9718719720840454,
      "learning_rate": 3e-05,
      "loss": 0.7714,
      "step": 18300
    },
    {
      "epoch": 16.892357126463164,
      "grad_norm": 2.0986855030059814,
      "learning_rate": 3e-05,
      "loss": 0.7861,
      "step": 18400
    },
    {
      "epoch": 16.98416341519394,
      "grad_norm": 2.2935447692871094,
      "learning_rate": 3e-05,
      "loss": 0.7803,
      "step": 18500
    },
    {
      "epoch": 16.999770484278173,
      "eval_accuracy": 0.7184133915574964,
      "eval_loss": 1.028311014175415,
      "eval_runtime": 8.9448,
      "eval_samples_per_second": 55.898,
      "eval_steps_per_second": 7.043,
      "step": 18517
    },
    {
      "epoch": 17.07596970392472,
      "grad_norm": 2.364075183868408,
      "learning_rate": 3e-05,
      "loss": 0.7264,
      "step": 18600
    },
    {
      "epoch": 17.167775992655496,
      "grad_norm": 2.1636979579925537,
      "learning_rate": 3e-05,
      "loss": 0.7038,
      "step": 18700
    },
    {
      "epoch": 17.259582281386276,
      "grad_norm": 2.135673761367798,
      "learning_rate": 3e-05,
      "loss": 0.7136,
      "step": 18800
    },
    {
      "epoch": 17.351388570117052,
      "grad_norm": 2.1516411304473877,
      "learning_rate": 3e-05,
      "loss": 0.7231,
      "step": 18900
    },
    {
      "epoch": 17.443194858847832,
      "grad_norm": 2.499406337738037,
      "learning_rate": 3e-05,
      "loss": 0.7302,
      "step": 19000
    },
    {
      "epoch": 17.53500114757861,
      "grad_norm": 2.455547332763672,
      "learning_rate": 3e-05,
      "loss": 0.7407,
      "step": 19100
    },
    {
      "epoch": 17.626807436309388,
      "grad_norm": 2.248194932937622,
      "learning_rate": 3e-05,
      "loss": 0.722,
      "step": 19200
    },
    {
      "epoch": 17.718613725040164,
      "grad_norm": 2.3520660400390625,
      "learning_rate": 3e-05,
      "loss": 0.7291,
      "step": 19300
    },
    {
      "epoch": 17.810420013770944,
      "grad_norm": 2.1547889709472656,
      "learning_rate": 3e-05,
      "loss": 0.7317,
      "step": 19400
    },
    {
      "epoch": 17.90222630250172,
      "grad_norm": 2.608548402786255,
      "learning_rate": 3e-05,
      "loss": 0.738,
      "step": 19500
    },
    {
      "epoch": 17.9940325912325,
      "grad_norm": 2.2248220443725586,
      "learning_rate": 3e-05,
      "loss": 0.7486,
      "step": 19600
    },
    {
      "epoch": 17.999540968556346,
      "eval_accuracy": 0.7214614264919942,
      "eval_loss": 0.9996564984321594,
      "eval_runtime": 8.9645,
      "eval_samples_per_second": 55.776,
      "eval_steps_per_second": 7.028,
      "step": 19606
    },
    {
      "epoch": 18.085838879963276,
      "grad_norm": 2.197584867477417,
      "learning_rate": 3e-05,
      "loss": 0.6731,
      "step": 19700
    },
    {
      "epoch": 18.177645168694056,
      "grad_norm": 2.392916440963745,
      "learning_rate": 3e-05,
      "loss": 0.676,
      "step": 19800
    },
    {
      "epoch": 18.269451457424832,
      "grad_norm": 2.4115874767303467,
      "learning_rate": 3e-05,
      "loss": 0.691,
      "step": 19900
    },
    {
      "epoch": 18.361257746155612,
      "grad_norm": 2.320349931716919,
      "learning_rate": 3e-05,
      "loss": 0.677,
      "step": 20000
    },
    {
      "epoch": 18.45306403488639,
      "grad_norm": 2.2987887859344482,
      "learning_rate": 3e-05,
      "loss": 0.6857,
      "step": 20100
    },
    {
      "epoch": 18.544870323617168,
      "grad_norm": 2.541984796524048,
      "learning_rate": 3e-05,
      "loss": 0.6787,
      "step": 20200
    },
    {
      "epoch": 18.636676612347944,
      "grad_norm": 2.0782082080841064,
      "learning_rate": 3e-05,
      "loss": 0.6973,
      "step": 20300
    },
    {
      "epoch": 18.728482901078724,
      "grad_norm": 2.4935009479522705,
      "learning_rate": 3e-05,
      "loss": 0.7083,
      "step": 20400
    },
    {
      "epoch": 18.8202891898095,
      "grad_norm": 2.8205904960632324,
      "learning_rate": 3e-05,
      "loss": 0.6872,
      "step": 20500
    },
    {
      "epoch": 18.91209547854028,
      "grad_norm": 2.335952043533325,
      "learning_rate": 3e-05,
      "loss": 0.6992,
      "step": 20600
    },
    {
      "epoch": 18.99931145283452,
      "eval_accuracy": 0.7237583697234352,
      "eval_loss": 0.971889078617096,
      "eval_runtime": 9.1348,
      "eval_samples_per_second": 54.736,
      "eval_steps_per_second": 6.897,
      "step": 20695
    },
    {
      "epoch": 19.003901767271056,
      "grad_norm": 1.9122214317321777,
      "learning_rate": 3e-05,
      "loss": 0.7018,
      "step": 20700
    },
    {
      "epoch": 19.095708056001836,
      "grad_norm": 2.1178972721099854,
      "learning_rate": 3e-05,
      "loss": 0.6337,
      "step": 20800
    },
    {
      "epoch": 19.187514344732612,
      "grad_norm": 2.1954286098480225,
      "learning_rate": 3e-05,
      "loss": 0.6294,
      "step": 20900
    },
    {
      "epoch": 19.279320633463392,
      "grad_norm": 2.2881522178649902,
      "learning_rate": 3e-05,
      "loss": 0.6436,
      "step": 21000
    },
    {
      "epoch": 19.371126922194172,
      "grad_norm": 2.2738537788391113,
      "learning_rate": 3e-05,
      "loss": 0.6584,
      "step": 21100
    },
    {
      "epoch": 19.462933210924948,
      "grad_norm": 2.3467330932617188,
      "learning_rate": 3e-05,
      "loss": 0.655,
      "step": 21200
    },
    {
      "epoch": 19.554739499655728,
      "grad_norm": 2.7984132766723633,
      "learning_rate": 3e-05,
      "loss": 0.647,
      "step": 21300
    },
    {
      "epoch": 19.646545788386504,
      "grad_norm": 2.397935152053833,
      "learning_rate": 3e-05,
      "loss": 0.6642,
      "step": 21400
    },
    {
      "epoch": 19.738352077117284,
      "grad_norm": 2.7952253818511963,
      "learning_rate": 3e-05,
      "loss": 0.6604,
      "step": 21500
    },
    {
      "epoch": 19.83015836584806,
      "grad_norm": 2.212345600128174,
      "learning_rate": 3e-05,
      "loss": 0.6598,
      "step": 21600
    },
    {
      "epoch": 19.92196465457884,
      "grad_norm": 2.5237057209014893,
      "learning_rate": 3e-05,
      "loss": 0.6632,
      "step": 21700
    },
    {
      "epoch": 19.995409685563462,
      "eval_accuracy": 0.7263318777292577,
      "eval_loss": 0.9440018534660339,
      "eval_runtime": 8.9833,
      "eval_samples_per_second": 55.659,
      "eval_steps_per_second": 7.013,
      "step": 21780
    },
    {
      "epoch": 19.995409685563462,
      "step": 21780,
      "total_flos": 2.2953223726028554e+18,
      "train_loss": 1.0977098983400344,
      "train_runtime": 46950.9086,
      "train_samples_per_second": 14.847,
      "train_steps_per_second": 0.464
    }
  ],
  "logging_steps": 100,
  "max_steps": 21780,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 2.2953223726028554e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}