{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 49.44692737430167,
  "eval_steps": 100,
  "global_step": 4450,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.111731843575419,
      "grad_norm": 0.48533013463020325,
      "learning_rate": 0.00028199999999999997,
      "loss": 11.3881,
      "step": 100
    },
    {
      "epoch": 1.111731843575419,
      "eval_cer": 0.9926923478442427,
      "eval_loss": 3.8517303466796875,
      "eval_runtime": 12.8394,
      "eval_samples_per_second": 22.197,
      "eval_steps_per_second": 2.804,
      "step": 100
    },
    {
      "epoch": 2.223463687150838,
      "grad_norm": 2.8934638500213623,
      "learning_rate": 0.0002935172413793103,
      "loss": 3.9077,
      "step": 200
    },
    {
      "epoch": 2.223463687150838,
      "eval_cer": 0.9924835577826495,
      "eval_loss": 3.6587204933166504,
      "eval_runtime": 12.8147,
      "eval_samples_per_second": 22.24,
      "eval_steps_per_second": 2.809,
      "step": 200
    },
    {
      "epoch": 3.335195530726257,
      "grad_norm": 2.12126088142395,
      "learning_rate": 0.0002866206896551724,
      "loss": 3.8977,
      "step": 300
    },
    {
      "epoch": 3.335195530726257,
      "eval_cer": 0.9926923478442427,
      "eval_loss": 3.657623291015625,
      "eval_runtime": 12.8572,
      "eval_samples_per_second": 22.167,
      "eval_steps_per_second": 2.8,
      "step": 300
    },
    {
      "epoch": 4.446927374301676,
      "grad_norm": 1.5865833759307861,
      "learning_rate": 0.00027972413793103445,
      "loss": 3.8192,
      "step": 400
    },
    {
      "epoch": 4.446927374301676,
      "eval_cer": 0.9929011379058357,
      "eval_loss": 3.663019895553589,
      "eval_runtime": 12.8178,
      "eval_samples_per_second": 22.235,
      "eval_steps_per_second": 2.809,
      "step": 400
    },
    {
      "epoch": 5.558659217877095,
      "grad_norm": 0.37739208340644836,
      "learning_rate": 0.00027282758620689656,
      "loss": 3.7035,
      "step": 500
    },
    {
      "epoch": 5.558659217877095,
      "eval_cer": 0.9789122037791002,
      "eval_loss": 3.678802967071533,
      "eval_runtime": 12.8176,
      "eval_samples_per_second": 22.235,
      "eval_steps_per_second": 2.809,
      "step": 500
    },
    {
      "epoch": 6.670391061452514,
      "grad_norm": 0.6414264440536499,
      "learning_rate": 0.00026593103448275857,
      "loss": 3.6207,
      "step": 600
    },
    {
      "epoch": 6.670391061452514,
      "eval_cer": 0.9543793715419147,
      "eval_loss": 3.6347501277923584,
      "eval_runtime": 12.7764,
      "eval_samples_per_second": 22.307,
      "eval_steps_per_second": 2.818,
      "step": 600
    },
    {
      "epoch": 7.782122905027933,
      "grad_norm": 0.7569805979728699,
      "learning_rate": 0.0002590344827586207,
      "loss": 3.5701,
      "step": 700
    },
    {
      "epoch": 7.782122905027933,
      "eval_cer": 0.9704562062845808,
      "eval_loss": 3.594348669052124,
      "eval_runtime": 12.8842,
      "eval_samples_per_second": 22.12,
      "eval_steps_per_second": 2.794,
      "step": 700
    },
    {
      "epoch": 8.893854748603353,
      "grad_norm": 0.617978572845459,
      "learning_rate": 0.00025213793103448273,
      "loss": 3.5264,
      "step": 800
    },
    {
      "epoch": 8.893854748603353,
      "eval_cer": 0.9584507777429795,
      "eval_loss": 3.557063579559326,
      "eval_runtime": 12.7852,
      "eval_samples_per_second": 22.291,
      "eval_steps_per_second": 2.816,
      "step": 800
    },
    {
      "epoch": 10.0,
      "grad_norm": 1.1942850351333618,
      "learning_rate": 0.00024524137931034484,
      "loss": 3.3878,
      "step": 900
    },
    {
      "epoch": 10.0,
      "eval_cer": 0.961895813759265,
      "eval_loss": 3.5423524379730225,
      "eval_runtime": 12.8758,
      "eval_samples_per_second": 22.135,
      "eval_steps_per_second": 2.796,
      "step": 900
    },
    {
      "epoch": 11.111731843575418,
      "grad_norm": 0.859997570514679,
      "learning_rate": 0.00023834482758620687,
      "loss": 3.2684,
      "step": 1000
    },
    {
      "epoch": 11.111731843575418,
      "eval_cer": 0.9471761144169537,
      "eval_loss": 3.390798568725586,
      "eval_runtime": 12.7862,
      "eval_samples_per_second": 22.29,
      "eval_steps_per_second": 2.816,
      "step": 1000
    },
    {
      "epoch": 12.223463687150838,
      "grad_norm": 2.701083183288574,
      "learning_rate": 0.00023144827586206896,
      "loss": 3.0432,
      "step": 1100
    },
    {
      "epoch": 12.223463687150838,
      "eval_cer": 0.8914291679716045,
      "eval_loss": 3.5293474197387695,
      "eval_runtime": 12.8801,
      "eval_samples_per_second": 22.127,
      "eval_steps_per_second": 2.795,
      "step": 1100
    },
    {
      "epoch": 13.335195530726256,
      "grad_norm": 1.2649680376052856,
      "learning_rate": 0.00022455172413793101,
      "loss": 2.7166,
      "step": 1200
    },
    {
      "epoch": 13.335195530726256,
      "eval_cer": 0.8706545568430942,
      "eval_loss": 2.8964927196502686,
      "eval_runtime": 12.8464,
      "eval_samples_per_second": 22.185,
      "eval_steps_per_second": 2.802,
      "step": 1200
    },
    {
      "epoch": 14.446927374301676,
      "grad_norm": 2.458191156387329,
      "learning_rate": 0.0002176551724137931,
      "loss": 2.372,
      "step": 1300
    },
    {
      "epoch": 14.446927374301676,
      "eval_cer": 0.8353690364338657,
      "eval_loss": 2.9027018547058105,
      "eval_runtime": 12.8524,
      "eval_samples_per_second": 22.175,
      "eval_steps_per_second": 2.801,
      "step": 1300
    },
    {
      "epoch": 15.558659217877095,
      "grad_norm": 2.145862102508545,
      "learning_rate": 0.00021075862068965515,
      "loss": 2.1382,
      "step": 1400
    },
    {
      "epoch": 15.558659217877095,
      "eval_cer": 0.8355778264954589,
      "eval_loss": 2.767353057861328,
      "eval_runtime": 12.8847,
      "eval_samples_per_second": 22.119,
      "eval_steps_per_second": 2.794,
      "step": 1400
    },
    {
      "epoch": 16.670391061452513,
      "grad_norm": 2.6955654621124268,
      "learning_rate": 0.0002038620689655172,
      "loss": 1.9442,
      "step": 1500
    },
    {
      "epoch": 16.670391061452513,
      "eval_cer": 0.8006054911786199,
      "eval_loss": 2.9376468658447266,
      "eval_runtime": 12.8501,
      "eval_samples_per_second": 22.179,
      "eval_steps_per_second": 2.802,
      "step": 1500
    },
    {
      "epoch": 17.782122905027933,
      "grad_norm": 1.8743767738342285,
      "learning_rate": 0.0001969655172413793,
      "loss": 1.7881,
      "step": 1600
    },
    {
      "epoch": 17.782122905027933,
      "eval_cer": 0.722831193235202,
      "eval_loss": 2.7831432819366455,
      "eval_runtime": 12.7796,
      "eval_samples_per_second": 22.301,
      "eval_steps_per_second": 2.817,
      "step": 1600
    },
    {
      "epoch": 18.893854748603353,
      "grad_norm": 2.472550392150879,
      "learning_rate": 0.00019006896551724138,
      "loss": 1.6766,
      "step": 1700
    },
    {
      "epoch": 18.893854748603353,
      "eval_cer": 0.7167762814490031,
      "eval_loss": 2.8390402793884277,
      "eval_runtime": 12.8876,
      "eval_samples_per_second": 22.114,
      "eval_steps_per_second": 2.793,
      "step": 1700
    },
    {
      "epoch": 20.0,
      "grad_norm": 8.466638565063477,
      "learning_rate": 0.00018317241379310343,
      "loss": 1.5416,
      "step": 1800
    },
    {
      "epoch": 20.0,
      "eval_cer": 0.730452030483349,
      "eval_loss": 3.0157971382141113,
      "eval_runtime": 12.7638,
      "eval_samples_per_second": 22.329,
      "eval_steps_per_second": 2.82,
      "step": 1800
    },
    {
      "epoch": 21.11173184357542,
      "grad_norm": 2.14225172996521,
      "learning_rate": 0.0001762758620689655,
      "loss": 1.4251,
      "step": 1900
    },
    {
      "epoch": 21.11173184357542,
      "eval_cer": 0.6958972752896962,
      "eval_loss": 2.983879327774048,
      "eval_runtime": 12.9131,
      "eval_samples_per_second": 22.071,
      "eval_steps_per_second": 2.788,
      "step": 1900
    },
    {
      "epoch": 22.223463687150836,
      "grad_norm": 4.268678188323975,
      "learning_rate": 0.00016937931034482757,
      "loss": 1.32,
      "step": 2000
    },
    {
      "epoch": 22.223463687150836,
      "eval_cer": 0.7089466541392629,
      "eval_loss": 2.9795498847961426,
      "eval_runtime": 13.1212,
      "eval_samples_per_second": 21.721,
      "eval_steps_per_second": 2.744,
      "step": 2000
    },
    {
      "epoch": 23.335195530726256,
      "grad_norm": 2.3187201023101807,
      "learning_rate": 0.00016248275862068966,
      "loss": 1.2355,
      "step": 2100
    },
    {
      "epoch": 23.335195530726256,
      "eval_cer": 0.7216828478964401,
      "eval_loss": 3.145298957824707,
      "eval_runtime": 12.8542,
      "eval_samples_per_second": 22.172,
      "eval_steps_per_second": 2.801,
      "step": 2100
    },
    {
      "epoch": 24.446927374301676,
      "grad_norm": 4.309772968292236,
      "learning_rate": 0.00015558620689655171,
      "loss": 1.1704,
      "step": 2200
    },
    {
      "epoch": 24.446927374301676,
      "eval_cer": 0.7027873473222674,
      "eval_loss": 3.060805320739746,
      "eval_runtime": 12.8156,
      "eval_samples_per_second": 22.238,
      "eval_steps_per_second": 2.809,
      "step": 2200
    },
    {
      "epoch": 25.558659217877096,
      "grad_norm": 2.0230820178985596,
      "learning_rate": 0.00014868965517241377,
      "loss": 1.0789,
      "step": 2300
    },
    {
      "epoch": 25.558659217877096,
      "eval_cer": 0.7133312454327174,
      "eval_loss": 3.070970058441162,
      "eval_runtime": 12.8729,
      "eval_samples_per_second": 22.139,
      "eval_steps_per_second": 2.797,
      "step": 2300
    },
    {
      "epoch": 26.670391061452513,
      "grad_norm": 3.8290631771087646,
      "learning_rate": 0.00014179310344827585,
      "loss": 1.0109,
      "step": 2400
    },
    {
      "epoch": 26.670391061452513,
      "eval_cer": 0.7095730243240421,
      "eval_loss": 3.1124370098114014,
      "eval_runtime": 12.8909,
      "eval_samples_per_second": 22.109,
      "eval_steps_per_second": 2.793,
      "step": 2400
    },
    {
      "epoch": 27.782122905027933,
      "grad_norm": 2.3735196590423584,
      "learning_rate": 0.0001348965517241379,
      "loss": 0.9362,
      "step": 2500
    },
    {
      "epoch": 27.782122905027933,
      "eval_cer": 0.7221004280196263,
      "eval_loss": 3.4384732246398926,
      "eval_runtime": 12.8512,
      "eval_samples_per_second": 22.177,
      "eval_steps_per_second": 2.801,
      "step": 2500
    },
    {
      "epoch": 28.893854748603353,
      "grad_norm": 2.767505645751953,
      "learning_rate": 0.000128,
      "loss": 0.8671,
      "step": 2600
    },
    {
      "epoch": 28.893854748603353,
      "eval_cer": 0.7152103559870551,
      "eval_loss": 3.3170225620269775,
      "eval_runtime": 12.9113,
      "eval_samples_per_second": 22.074,
      "eval_steps_per_second": 2.788,
      "step": 2600
    },
    {
      "epoch": 30.0,
      "grad_norm": 7.469735622406006,
      "learning_rate": 0.00012110344827586206,
      "loss": 0.831,
      "step": 2700
    },
    {
      "epoch": 30.0,
      "eval_cer": 0.7012214218603194,
      "eval_loss": 3.2871029376983643,
      "eval_runtime": 12.7981,
      "eval_samples_per_second": 22.269,
      "eval_steps_per_second": 2.813,
      "step": 2700
    },
    {
      "epoch": 31.11173184357542,
      "grad_norm": 3.1189279556274414,
      "learning_rate": 0.00011420689655172412,
      "loss": 0.7467,
      "step": 2800
    },
    {
      "epoch": 31.11173184357542,
      "eval_cer": 0.7047708529074016,
      "eval_loss": 3.5182864665985107,
      "eval_runtime": 12.9125,
      "eval_samples_per_second": 22.072,
      "eval_steps_per_second": 2.788,
      "step": 2800
    },
    {
      "epoch": 32.22346368715084,
      "grad_norm": 14.725250244140625,
      "learning_rate": 0.0001073103448275862,
      "loss": 0.7128,
      "step": 2900
    },
    {
      "epoch": 32.22346368715084,
      "eval_cer": 0.7088422591084664,
      "eval_loss": 3.3886497020721436,
      "eval_runtime": 12.9242,
      "eval_samples_per_second": 22.052,
      "eval_steps_per_second": 2.785,
      "step": 2900
    },
    {
      "epoch": 33.33519553072626,
      "grad_norm": 2.3201019763946533,
      "learning_rate": 0.00010041379310344826,
      "loss": 0.6815,
      "step": 3000
    },
    {
      "epoch": 33.33519553072626,
      "eval_cer": 0.7230399832967951,
      "eval_loss": 3.731771230697632,
      "eval_runtime": 12.9081,
      "eval_samples_per_second": 22.079,
      "eval_steps_per_second": 2.789,
      "step": 3000
    },
    {
      "epoch": 34.44692737430167,
      "grad_norm": 4.227590084075928,
      "learning_rate": 9.351724137931033e-05,
      "loss": 0.6452,
      "step": 3100
    },
    {
      "epoch": 34.44692737430167,
      "eval_cer": 0.732017955945297,
      "eval_loss": 3.7759644985198975,
      "eval_runtime": 12.9044,
      "eval_samples_per_second": 22.085,
      "eval_steps_per_second": 2.79,
      "step": 3100
    },
    {
      "epoch": 35.55865921787709,
      "grad_norm": 1.658971905708313,
      "learning_rate": 8.66206896551724e-05,
      "loss": 0.6052,
      "step": 3200
    },
    {
      "epoch": 35.55865921787709,
      "eval_cer": 0.7280509447750287,
      "eval_loss": 3.8133554458618164,
      "eval_runtime": 12.9295,
      "eval_samples_per_second": 22.043,
      "eval_steps_per_second": 2.784,
      "step": 3200
    },
    {
      "epoch": 36.67039106145251,
      "grad_norm": 3.6614513397216797,
      "learning_rate": 7.972413793103447e-05,
      "loss": 0.5779,
      "step": 3300
    },
    {
      "epoch": 36.67039106145251,
      "eval_cer": 0.6808643908549953,
      "eval_loss": 3.7408699989318848,
      "eval_runtime": 13.0708,
      "eval_samples_per_second": 21.804,
      "eval_steps_per_second": 2.754,
      "step": 3300
    },
    {
      "epoch": 37.78212290502793,
      "grad_norm": 1.5677216053009033,
      "learning_rate": 7.282758620689654e-05,
      "loss": 0.5343,
      "step": 3400
    },
    {
      "epoch": 37.78212290502793,
      "eval_cer": 0.6906775237498695,
      "eval_loss": 3.825223922729492,
      "eval_runtime": 12.9331,
      "eval_samples_per_second": 22.036,
      "eval_steps_per_second": 2.784,
      "step": 3400
    },
    {
      "epoch": 38.89385474860335,
      "grad_norm": 2.8445885181427,
      "learning_rate": 6.593103448275861e-05,
      "loss": 0.5056,
      "step": 3500
    },
    {
      "epoch": 38.89385474860335,
      "eval_cer": 0.6964192504436789,
      "eval_loss": 3.9913222789764404,
      "eval_runtime": 12.9414,
      "eval_samples_per_second": 22.022,
      "eval_steps_per_second": 2.782,
      "step": 3500
    },
    {
      "epoch": 40.0,
      "grad_norm": 12.087656021118164,
      "learning_rate": 5.903448275862069e-05,
      "loss": 0.5017,
      "step": 3600
    },
    {
      "epoch": 40.0,
      "eval_cer": 0.6999686814907611,
      "eval_loss": 4.1249589920043945,
      "eval_runtime": 12.9858,
      "eval_samples_per_second": 21.947,
      "eval_steps_per_second": 2.772,
      "step": 3600
    },
    {
      "epoch": 41.11173184357542,
      "grad_norm": 1.838942527770996,
      "learning_rate": 5.213793103448276e-05,
      "loss": 0.4592,
      "step": 3700
    },
    {
      "epoch": 41.11173184357542,
      "eval_cer": 0.7221004280196263,
      "eval_loss": 4.20962381362915,
      "eval_runtime": 12.8242,
      "eval_samples_per_second": 22.224,
      "eval_steps_per_second": 2.807,
      "step": 3700
    },
    {
      "epoch": 42.22346368715084,
      "grad_norm": 7.289366245269775,
      "learning_rate": 4.524137931034483e-05,
      "loss": 0.4478,
      "step": 3800
    },
    {
      "epoch": 42.22346368715084,
      "eval_cer": 0.7031005324146571,
      "eval_loss": 4.215978622436523,
      "eval_runtime": 12.8764,
      "eval_samples_per_second": 22.134,
      "eval_steps_per_second": 2.796,
      "step": 3800
    },
    {
      "epoch": 43.33519553072626,
      "grad_norm": 1.9154551029205322,
      "learning_rate": 3.834482758620689e-05,
      "loss": 0.4172,
      "step": 3900
    },
    {
      "epoch": 43.33519553072626,
      "eval_cer": 0.7085290740160768,
      "eval_loss": 4.245218276977539,
      "eval_runtime": 12.9248,
      "eval_samples_per_second": 22.051,
      "eval_steps_per_second": 2.785,
      "step": 3900
    },
    {
      "epoch": 44.44692737430167,
      "grad_norm": 6.6539154052734375,
      "learning_rate": 3.151724137931034e-05,
      "loss": 0.4018,
      "step": 4000
    },
    {
      "epoch": 44.44692737430167,
      "eval_cer": 0.7155235410794446,
      "eval_loss": 4.3074631690979,
      "eval_runtime": 12.9104,
      "eval_samples_per_second": 22.075,
      "eval_steps_per_second": 2.788,
      "step": 4000
    },
    {
      "epoch": 45.55865921787709,
      "grad_norm": 1.8227887153625488,
      "learning_rate": 2.4620689655172412e-05,
      "loss": 0.3854,
      "step": 4100
    },
    {
      "epoch": 45.55865921787709,
      "eval_cer": 0.7094686292932456,
      "eval_loss": 4.359448432922363,
      "eval_runtime": 12.9111,
      "eval_samples_per_second": 22.074,
      "eval_steps_per_second": 2.788,
      "step": 4100
    },
    {
      "epoch": 46.67039106145251,
      "grad_norm": 3.2107598781585693,
      "learning_rate": 1.7724137931034482e-05,
      "loss": 0.3771,
      "step": 4200
    },
    {
      "epoch": 46.67039106145251,
      "eval_cer": 0.7133312454327174,
      "eval_loss": 4.366276741027832,
      "eval_runtime": 12.9048,
      "eval_samples_per_second": 22.085,
      "eval_steps_per_second": 2.79,
      "step": 4200
    },
    {
      "epoch": 47.78212290502793,
      "grad_norm": 1.2981507778167725,
      "learning_rate": 1.082758620689655e-05,
      "loss": 0.3553,
      "step": 4300
    },
    {
      "epoch": 47.78212290502793,
      "eval_cer": 0.7135400354943104,
      "eval_loss": 4.419826030731201,
      "eval_runtime": 12.9283,
      "eval_samples_per_second": 22.045,
      "eval_steps_per_second": 2.785,
      "step": 4300
    },
    {
      "epoch": 48.89385474860335,
      "grad_norm": 2.526685953140259,
      "learning_rate": 3.93103448275862e-06,
      "loss": 0.3565,
      "step": 4400
    },
    {
      "epoch": 48.89385474860335,
      "eval_cer": 0.7108257646936006,
      "eval_loss": 4.4135847091674805,
      "eval_runtime": 12.8772,
      "eval_samples_per_second": 22.132,
      "eval_steps_per_second": 2.796,
      "step": 4400
    },
    {
      "epoch": 49.44692737430167,
      "step": 4450,
      "total_flos": 9.669635712078565e+18,
      "train_loss": 1.799734133602528,
      "train_runtime": 7166.9702,
      "train_samples_per_second": 19.897,
      "train_steps_per_second": 0.621
    },
    {
      "epoch": 49.44692737430167,
      "eval_cer": 0.7118697150015659,
      "eval_loss": 4.406894207000732,
      "eval_runtime": 12.8787,
      "eval_samples_per_second": 22.13,
      "eval_steps_per_second": 2.795,
      "step": 4450
    }
  ],
  "logging_steps": 100,
  "max_steps": 4450,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "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": 9.669635712078565e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}