trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.995850622406639,
  "global_step": 300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.33,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.9021,
      "step": 100
    },
    {
      "epoch": 6.66,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.5015,
      "step": 200
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.0,
      "loss": 0.3757,
      "step": 300
    },
    {
      "epoch": 10.0,
      "step": 300,
      "total_flos": 1317332914237440.0,
      "train_loss": 0.5930885314941406,
      "train_runtime": 183.6839,
      "train_samples_per_second": 26.241,
      "train_steps_per_second": 1.633
    }
  ],
  "max_steps": 300,
  "num_train_epochs": 10,
  "total_flos": 1317332914237440.0,
  "trial_name": null,
  "trial_params": null
}