trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 1140,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.39,
      "learning_rate": 1.6842105263157893e-05,
      "loss": 0.1909,
      "step": 500
    },
    {
      "epoch": 8.77,
      "learning_rate": 3.6842105263157892e-06,
      "loss": 0.0196,
      "step": 1000
    },
    {
      "epoch": 10.0,
      "step": 1140,
      "total_flos": 931250075040000.0,
      "train_loss": 0.09304492578171847,
      "train_runtime": 41.8899,
      "train_samples_per_second": 432.562,
      "train_steps_per_second": 27.214
    }
  ],
  "max_steps": 1140,
  "num_train_epochs": 10,
  "total_flos": 931250075040000.0,
  "trial_name": null,
  "trial_params": null
}