trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "eval_steps": 100,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "learning_rate": 4.6000000000000004e-07,
      "loss": 1.7988,
      "step": 25
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.600000000000001e-07,
      "loss": 1.6021,
      "step": 50
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.46e-06,
      "loss": 1.3669,
      "step": 75
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.9600000000000003e-06,
      "loss": 1.2085,
      "step": 100
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.0991923809051514,
      "eval_runtime": 230.1115,
      "eval_samples_per_second": 1.738,
      "eval_steps_per_second": 0.109,
      "eval_wer": 1.0023889154323937,
      "step": 100
    },
    {
      "epoch": 2.5,
      "learning_rate": 2.46e-06,
      "loss": 0.984,
      "step": 125
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.96e-06,
      "loss": 0.7067,
      "step": 150
    },
    {
      "epoch": 3.5,
      "learning_rate": 3.46e-06,
      "loss": 0.4651,
      "step": 175
    },
    {
      "epoch": 4.0,
      "learning_rate": 3.96e-06,
      "loss": 0.3129,
      "step": 200
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.2795320451259613,
      "eval_runtime": 194.3318,
      "eval_samples_per_second": 2.058,
      "eval_steps_per_second": 0.129,
      "eval_wer": 0.8767319636884854,
      "step": 200
    },
    {
      "epoch": 4.0,
      "step": 200,
      "total_flos": 1.846946562048e+18,
      "train_loss": 1.0556322383880614,
      "train_runtime": 2067.8737,
      "train_samples_per_second": 3.095,
      "train_steps_per_second": 0.097
    }
  ],
  "logging_steps": 25,
  "max_steps": 200,
  "num_train_epochs": 4,
  "save_steps": 100,
  "total_flos": 1.846946562048e+18,
  "trial_name": null,
  "trial_params": null
}