trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 1560,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.6,
      "learning_rate": 1.3589743589743592e-05,
      "loss": 0.6095,
      "step": 500
    },
    {
      "epoch": 3.21,
      "learning_rate": 7.17948717948718e-06,
      "loss": 0.3272,
      "step": 1000
    },
    {
      "epoch": 4.81,
      "learning_rate": 7.692307692307694e-07,
      "loss": 0.0972,
      "step": 1500
    },
    {
      "epoch": 5.0,
      "step": 1560,
      "total_flos": 813286136217600.0,
      "train_loss": 0.3344010169689472,
      "train_runtime": 98.9629,
      "train_samples_per_second": 125.805,
      "train_steps_per_second": 15.763
    }
  ],
  "max_steps": 1560,
  "num_train_epochs": 5,
  "total_flos": 813286136217600.0,
  "trial_name": null,
  "trial_params": null
}