trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 1150,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.17,
      "learning_rate": 1.1304347826086957e-05,
      "loss": 0.4267,
      "step": 500
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 0.1886,
      "step": 1000
    },
    {
      "epoch": 5.0,
      "step": 1150,
      "total_flos": 2412728377651200.0,
      "train_loss": 0.2803073907935101,
      "train_runtime": 6011.5008,
      "train_samples_per_second": 3.051,
      "train_steps_per_second": 0.191
    }
  ],
  "max_steps": 1150,
  "num_train_epochs": 5,
  "total_flos": 2412728377651200.0,
  "trial_name": null,
  "trial_params": null
}