trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 1,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 3.7033,
      "step": 1
    },
    {
      "epoch": 1.0,
      "eval_loss": 3.499124526977539,
      "eval_runtime": 441.0593,
      "eval_samples_per_second": 2.267,
      "eval_steps_per_second": 0.143,
      "eval_wer": 122.36095346197501,
      "step": 1
    },
    {
      "epoch": 1.0,
      "step": 1,
      "total_flos": 1575608647680000.0,
      "train_loss": 3.703259229660034,
      "train_runtime": 623.5929,
      "train_samples_per_second": 0.103,
      "train_steps_per_second": 0.002
    }
  ],
  "max_steps": 1,
  "num_train_epochs": 9223372036854775807,
  "total_flos": 1575608647680000.0,
  "trial_name": null,
  "trial_params": null
}