{
  "best_metric": 0.15666962437148774,
  "best_model_checkpoint": "results15/checkpoint-3000",
  "epoch": 0.256,
  "eval_steps": 500,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0064,
      "grad_norm": 4.8001484870910645,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 1.3393,
      "step": 100
    },
    {
      "epoch": 0.0128,
      "grad_norm": 3.706824779510498,
      "learning_rate": 3.920000000000001e-06,
      "loss": 0.5176,
      "step": 200
    },
    {
      "epoch": 0.0192,
      "grad_norm": 3.8430540561676025,
      "learning_rate": 5.92e-06,
      "loss": 0.2587,
      "step": 300
    },
    {
      "epoch": 0.0256,
      "grad_norm": 3.9047493934631348,
      "learning_rate": 7.92e-06,
      "loss": 0.2493,
      "step": 400
    },
    {
      "epoch": 0.032,
      "grad_norm": 3.7892723083496094,
      "learning_rate": 9.920000000000002e-06,
      "loss": 0.2446,
      "step": 500
    },
    {
      "epoch": 0.032,
      "eval_test1_cer": 0.1596436118028828,
      "eval_test1_cer_norm": 0.14118874332848133,
      "eval_test1_loss": 0.22201526165008545,
      "eval_test1_runtime": 1157.1217,
      "eval_test1_samples_per_second": 2.161,
      "eval_test1_steps_per_second": 0.068,
      "eval_test1_wer": 0.3501922508133688,
      "eval_test1_wer_norm": 0.27503187357310166,
      "step": 500
    },
    {
      "epoch": 0.032,
      "eval_test2_cer": 0.34475821649329114,
      "eval_test2_cer_norm": 0.32302360141569386,
      "eval_test2_loss": 0.3860654830932617,
      "eval_test2_runtime": 1327.3073,
      "eval_test2_samples_per_second": 1.884,
      "eval_test2_steps_per_second": 0.06,
      "eval_test2_wer": 0.5888413010765134,
      "eval_test2_wer_norm": 0.5090455757856894,
      "step": 500
    },
    {
      "epoch": 0.0384,
      "grad_norm": 4.102191925048828,
      "learning_rate": 9.936528925619836e-06,
      "loss": 0.2387,
      "step": 600
    },
    {
      "epoch": 0.0448,
      "grad_norm": 3.395754337310791,
      "learning_rate": 9.870413223140496e-06,
      "loss": 0.2333,
      "step": 700
    },
    {
      "epoch": 0.0512,
      "grad_norm": 3.3210232257843018,
      "learning_rate": 9.804297520661158e-06,
      "loss": 0.2296,
      "step": 800
    },
    {
      "epoch": 0.0576,
      "grad_norm": 3.533163070678711,
      "learning_rate": 9.738181818181818e-06,
      "loss": 0.2235,
      "step": 900
    },
    {
      "epoch": 0.064,
      "grad_norm": 3.154200315475464,
      "learning_rate": 9.67206611570248e-06,
      "loss": 0.2268,
      "step": 1000
    },
    {
      "epoch": 0.064,
      "eval_test1_cer": 0.06988151837597162,
      "eval_test1_cer_norm": 0.05281416787967006,
      "eval_test1_loss": 0.2057729810476303,
      "eval_test1_runtime": 1040.3745,
      "eval_test1_samples_per_second": 2.403,
      "eval_test1_steps_per_second": 0.076,
      "eval_test1_wer": 0.20041407867494823,
      "eval_test1_wer_norm": 0.1341062057105583,
      "step": 1000
    },
    {
      "epoch": 0.064,
      "eval_test2_cer": 0.17753561736770693,
      "eval_test2_cer_norm": 0.1456756809604818,
      "eval_test2_loss": 0.35842248797416687,
      "eval_test2_runtime": 1152.2398,
      "eval_test2_samples_per_second": 2.17,
      "eval_test2_steps_per_second": 0.069,
      "eval_test2_wer": 0.36184743604583863,
      "eval_test2_wer_norm": 0.28771889133712164,
      "step": 1000
    },
    {
      "epoch": 0.0704,
      "grad_norm": 4.132891654968262,
      "learning_rate": 9.605950413223142e-06,
      "loss": 0.2234,
      "step": 1100
    },
    {
      "epoch": 0.0768,
      "grad_norm": 3.4513018131256104,
      "learning_rate": 9.539834710743802e-06,
      "loss": 0.2218,
      "step": 1200
    },
    {
      "epoch": 0.0832,
      "grad_norm": 3.2677953243255615,
      "learning_rate": 9.473719008264464e-06,
      "loss": 0.2178,
      "step": 1300
    },
    {
      "epoch": 0.0896,
      "grad_norm": 2.8952624797821045,
      "learning_rate": 9.407603305785124e-06,
      "loss": 0.2161,
      "step": 1400
    },
    {
      "epoch": 0.096,
      "grad_norm": 3.3220458030700684,
      "learning_rate": 9.341487603305786e-06,
      "loss": 0.209,
      "step": 1500
    },
    {
      "epoch": 0.096,
      "eval_test1_cer": 0.061481208965361106,
      "eval_test1_cer_norm": 0.04084910237748666,
      "eval_test1_loss": 0.19823798537254333,
      "eval_test1_runtime": 1030.8083,
      "eval_test1_samples_per_second": 2.425,
      "eval_test1_steps_per_second": 0.077,
      "eval_test1_wer": 0.1774622892635315,
      "eval_test1_wer_norm": 0.11234322649509296,
      "step": 1500
    },
    {
      "epoch": 0.096,
      "eval_test2_cer": 0.11561981757877281,
      "eval_test2_cer_norm": 0.09326176648872725,
      "eval_test2_loss": 0.34961584210395813,
      "eval_test2_runtime": 1044.2589,
      "eval_test2_samples_per_second": 2.394,
      "eval_test2_steps_per_second": 0.076,
      "eval_test2_wer": 0.25847899062391483,
      "eval_test2_wer_norm": 0.18728980633190306,
      "step": 1500
    },
    {
      "epoch": 0.1024,
      "grad_norm": 3.133871555328369,
      "learning_rate": 9.275371900826448e-06,
      "loss": 0.2173,
      "step": 1600
    },
    {
      "epoch": 0.1088,
      "grad_norm": 3.023627758026123,
      "learning_rate": 9.209917355371902e-06,
      "loss": 0.2098,
      "step": 1700
    },
    {
      "epoch": 0.1152,
      "grad_norm": 3.384046792984009,
      "learning_rate": 9.143801652892564e-06,
      "loss": 0.2085,
      "step": 1800
    },
    {
      "epoch": 0.1216,
      "grad_norm": 4.177057266235352,
      "learning_rate": 9.077685950413224e-06,
      "loss": 0.2057,
      "step": 1900
    },
    {
      "epoch": 0.128,
      "grad_norm": 3.657303810119629,
      "learning_rate": 9.011570247933886e-06,
      "loss": 0.2073,
      "step": 2000
    },
    {
      "epoch": 0.128,
      "eval_test1_cer": 0.06908912534903026,
      "eval_test1_cer_norm": 0.04828723920426977,
      "eval_test1_loss": 0.18983058631420135,
      "eval_test1_runtime": 1097.0314,
      "eval_test1_samples_per_second": 2.279,
      "eval_test1_steps_per_second": 0.072,
      "eval_test1_wer": 0.1898846495119787,
      "eval_test1_wer_norm": 0.128057639280102,
      "step": 2000
    },
    {
      "epoch": 0.128,
      "eval_test2_cer": 0.1378665385195236,
      "eval_test2_cer_norm": 0.1097016092763145,
      "eval_test2_loss": 0.33988869190216064,
      "eval_test2_runtime": 1142.2555,
      "eval_test2_samples_per_second": 2.189,
      "eval_test2_steps_per_second": 0.069,
      "eval_test2_wer": 0.2811957402477139,
      "eval_test2_wer_norm": 0.21402064246781863,
      "step": 2000
    },
    {
      "epoch": 0.1344,
      "grad_norm": 2.9096293449401855,
      "learning_rate": 8.945454545454546e-06,
      "loss": 0.208,
      "step": 2100
    },
    {
      "epoch": 0.1408,
      "grad_norm": 3.0557897090911865,
      "learning_rate": 8.879338842975208e-06,
      "loss": 0.2146,
      "step": 2200
    },
    {
      "epoch": 0.1472,
      "grad_norm": 3.3298580646514893,
      "learning_rate": 8.81322314049587e-06,
      "loss": 0.2035,
      "step": 2300
    },
    {
      "epoch": 0.1536,
      "grad_norm": 2.332103729248047,
      "learning_rate": 8.74710743801653e-06,
      "loss": 0.196,
      "step": 2400
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.920443534851074,
      "learning_rate": 8.680991735537191e-06,
      "loss": 0.1982,
      "step": 2500
    },
    {
      "epoch": 0.16,
      "eval_test1_cer": 0.06446211606671194,
      "eval_test1_cer_norm": 0.04428918000970403,
      "eval_test1_loss": 0.18459410965442657,
      "eval_test1_runtime": 1086.8024,
      "eval_test1_samples_per_second": 2.3,
      "eval_test1_steps_per_second": 0.073,
      "eval_test1_wer": 0.17775805974563738,
      "eval_test1_wer_norm": 0.11643490378628399,
      "step": 2500
    },
    {
      "epoch": 0.16,
      "eval_test2_cer": 0.11876224935926429,
      "eval_test2_cer_norm": 0.09257249564928335,
      "eval_test2_loss": 0.33487755060195923,
      "eval_test2_runtime": 1110.7242,
      "eval_test2_samples_per_second": 2.251,
      "eval_test2_steps_per_second": 0.071,
      "eval_test2_wer": 0.2543986572519968,
      "eval_test2_wer_norm": 0.1844195755537516,
      "step": 2500
    },
    {
      "epoch": 0.1664,
      "grad_norm": 2.761300802230835,
      "learning_rate": 8.614876033057852e-06,
      "loss": 0.1969,
      "step": 2600
    },
    {
      "epoch": 0.1728,
      "grad_norm": 2.2639405727386475,
      "learning_rate": 8.548760330578513e-06,
      "loss": 0.2011,
      "step": 2700
    },
    {
      "epoch": 0.1792,
      "grad_norm": 2.7103681564331055,
      "learning_rate": 8.482644628099175e-06,
      "loss": 0.1911,
      "step": 2800
    },
    {
      "epoch": 0.1856,
      "grad_norm": 3.4355201721191406,
      "learning_rate": 8.416528925619835e-06,
      "loss": 0.1915,
      "step": 2900
    },
    {
      "epoch": 0.192,
      "grad_norm": 2.784461259841919,
      "learning_rate": 8.350413223140497e-06,
      "loss": 0.1962,
      "step": 3000
    },
    {
      "epoch": 0.192,
      "eval_test1_cer": 0.052505471285186026,
      "eval_test1_cer_norm": 0.03515769044153324,
      "eval_test1_loss": 0.17998091876506805,
      "eval_test1_runtime": 1020.463,
      "eval_test1_samples_per_second": 2.45,
      "eval_test1_steps_per_second": 0.077,
      "eval_test1_wer": 0.15666962437148774,
      "eval_test1_wer_norm": 0.09473122424170546,
      "step": 3000
    },
    {
      "epoch": 0.192,
      "eval_test2_cer": 0.11729232624755012,
      "eval_test2_cer_norm": 0.09097886236092372,
      "eval_test2_loss": 0.3286541700363159,
      "eval_test2_runtime": 1052.1877,
      "eval_test2_samples_per_second": 2.376,
      "eval_test2_steps_per_second": 0.075,
      "eval_test2_wer": 0.2477427943048964,
      "eval_test2_wer_norm": 0.17702655688275543,
      "step": 3000
    },
    {
      "epoch": 0.1984,
      "grad_norm": 2.9920506477355957,
      "learning_rate": 8.284297520661157e-06,
      "loss": 0.1916,
      "step": 3100
    },
    {
      "epoch": 0.2048,
      "grad_norm": 2.529097318649292,
      "learning_rate": 8.21818181818182e-06,
      "loss": 0.1909,
      "step": 3200
    },
    {
      "epoch": 0.2112,
      "grad_norm": 2.646070718765259,
      "learning_rate": 8.152066115702481e-06,
      "loss": 0.1904,
      "step": 3300
    },
    {
      "epoch": 0.2176,
      "grad_norm": 3.047825336456299,
      "learning_rate": 8.085950413223141e-06,
      "loss": 0.1935,
      "step": 3400
    },
    {
      "epoch": 0.224,
      "grad_norm": 2.928596258163452,
      "learning_rate": 8.019834710743803e-06,
      "loss": 0.1943,
      "step": 3500
    },
    {
      "epoch": 0.224,
      "eval_test1_cer": 0.06426873443513698,
      "eval_test1_cer_norm": 0.041746724890829694,
      "eval_test1_loss": 0.17650315165519714,
      "eval_test1_runtime": 1003.7308,
      "eval_test1_samples_per_second": 2.491,
      "eval_test1_steps_per_second": 0.079,
      "eval_test1_wer": 0.17329192546583852,
      "eval_test1_wer_norm": 0.10694695644439173,
      "step": 3500
    },
    {
      "epoch": 0.224,
      "eval_test2_cer": 0.13343792401628224,
      "eval_test2_cer_norm": 0.09719207680725836,
      "eval_test2_loss": 0.32644087076187134,
      "eval_test2_runtime": 1081.9218,
      "eval_test2_samples_per_second": 2.311,
      "eval_test2_steps_per_second": 0.073,
      "eval_test2_wer": 0.26533742331288346,
      "eval_test2_wer_norm": 0.1966542966484982,
      "step": 3500
    },
    {
      "epoch": 0.2304,
      "grad_norm": 2.7711305618286133,
      "learning_rate": 7.953719008264463e-06,
      "loss": 0.196,
      "step": 3600
    },
    {
      "epoch": 0.2368,
      "grad_norm": 2.9088964462280273,
      "learning_rate": 7.888264462809919e-06,
      "loss": 0.1899,
      "step": 3700
    },
    {
      "epoch": 0.2432,
      "grad_norm": 2.884896993637085,
      "learning_rate": 7.822148760330579e-06,
      "loss": 0.1938,
      "step": 3800
    },
    {
      "epoch": 0.2496,
      "grad_norm": 2.511730432510376,
      "learning_rate": 7.75603305785124e-06,
      "loss": 0.1872,
      "step": 3900
    },
    {
      "epoch": 0.256,
      "grad_norm": 2.930680990219116,
      "learning_rate": 7.689917355371903e-06,
      "loss": 0.1894,
      "step": 4000
    },
    {
      "epoch": 0.256,
      "eval_test1_cer": 0.071758735189797,
      "eval_test1_cer_norm": 0.047394468704512374,
      "eval_test1_loss": 0.1755101978778839,
      "eval_test1_runtime": 1295.9749,
      "eval_test1_samples_per_second": 1.929,
      "eval_test1_steps_per_second": 0.061,
      "eval_test1_wer": 0.18352558414670217,
      "eval_test1_wer_norm": 0.12132712663444718,
      "step": 4000
    },
    {
      "epoch": 0.256,
      "eval_test2_cer": 0.13147802653399668,
      "eval_test2_cer_norm": 0.10113216401712911,
      "eval_test2_loss": 0.32261422276496887,
      "eval_test2_runtime": 1380.4258,
      "eval_test2_samples_per_second": 1.811,
      "eval_test2_steps_per_second": 0.057,
      "eval_test2_wer": 0.2693888181502489,
      "eval_test2_wer_norm": 0.19903165951524993,
      "step": 4000
    }
  ],
  "logging_steps": 100,
  "max_steps": 15625,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 9223372036854775807,
  "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": 1.3063742816256e+20,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}