trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 1050,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.43,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 1.1647,
      "step": 500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 0.7238,
      "step": 1000
    },
    {
      "epoch": 3.0,
      "step": 1050,
      "total_flos": 2340364967608320.0,
      "train_loss": 0.9286957077752976,
      "train_runtime": 364.5018,
      "train_samples_per_second": 28.806,
      "train_steps_per_second": 2.881
    }
  ],
  "max_steps": 1050,
  "num_train_epochs": 3,
  "total_flos": 2340364967608320.0,
  "trial_name": null,
  "trial_params": null
}