trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 80,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.625,
      "grad_norm": 35.632633209228516,
      "learning_rate": 9.620253164556963e-06,
      "loss": 3.8394,
      "step": 10
    },
    {
      "epoch": 1.0,
      "eval_cer": 0.6162631650936944,
      "eval_loss": 1.9695916175842285,
      "eval_runtime": 53.8754,
      "eval_samples_per_second": 14.051,
      "eval_steps_per_second": 0.056,
      "eval_wer": 0.9849082490138913,
      "step": 16
    },
    {
      "epoch": 1.25,
      "grad_norm": 14.219266891479492,
      "learning_rate": 8.354430379746837e-06,
      "loss": 2.1751,
      "step": 20
    },
    {
      "epoch": 1.875,
      "grad_norm": 9.090901374816895,
      "learning_rate": 7.08860759493671e-06,
      "loss": 1.3507,
      "step": 30
    },
    {
      "epoch": 2.0,
      "eval_cer": 0.41071672821775407,
      "eval_loss": 1.0905992984771729,
      "eval_runtime": 54.5857,
      "eval_samples_per_second": 13.868,
      "eval_steps_per_second": 0.055,
      "eval_wer": 0.8955582232893158,
      "step": 32
    },
    {
      "epoch": 2.5,
      "grad_norm": 4.849871635437012,
      "learning_rate": 5.8227848101265824e-06,
      "loss": 0.9898,
      "step": 40
    },
    {
      "epoch": 3.0,
      "eval_cer": 0.2880590890439064,
      "eval_loss": 0.7982303500175476,
      "eval_runtime": 54.2612,
      "eval_samples_per_second": 13.951,
      "eval_steps_per_second": 0.055,
      "eval_wer": 0.7871720116618076,
      "step": 48
    },
    {
      "epoch": 3.125,
      "grad_norm": 3.600881576538086,
      "learning_rate": 4.556962025316456e-06,
      "loss": 0.8102,
      "step": 50
    },
    {
      "epoch": 3.75,
      "grad_norm": 3.1939327716827393,
      "learning_rate": 3.2911392405063294e-06,
      "loss": 0.6834,
      "step": 60
    },
    {
      "epoch": 4.0,
      "eval_cer": 0.22661058678703325,
      "eval_loss": 0.6941161155700684,
      "eval_runtime": 51.0421,
      "eval_samples_per_second": 14.831,
      "eval_steps_per_second": 0.059,
      "eval_wer": 0.7153146973074944,
      "step": 64
    },
    {
      "epoch": 4.375,
      "grad_norm": 2.299682378768921,
      "learning_rate": 2.0253164556962026e-06,
      "loss": 0.636,
      "step": 70
    },
    {
      "epoch": 5.0,
      "grad_norm": 2.070575714111328,
      "learning_rate": 7.59493670886076e-07,
      "loss": 0.5981,
      "step": 80
    },
    {
      "epoch": 5.0,
      "eval_cer": 0.21645465736561345,
      "eval_loss": 0.6611320376396179,
      "eval_runtime": 51.3922,
      "eval_samples_per_second": 14.73,
      "eval_steps_per_second": 0.058,
      "eval_wer": 0.6882181444006173,
      "step": 80
    },
    {
      "epoch": 5.0,
      "step": 80,
      "total_flos": 1.9924990745190072e+18,
      "train_loss": 1.3853384613990785,
      "train_runtime": 2046.8169,
      "train_samples_per_second": 14.442,
      "train_steps_per_second": 0.039
    }
  ],
  "logging_steps": 10,
  "max_steps": 80,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.9924990745190072e+18,
  "train_batch_size": 192,
  "trial_name": null,
  "trial_params": null
}