trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 2165,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.15,
      "learning_rate": 2.30715935334873e-05,
      "loss": 0.0818,
      "step": 500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.6143187066974596e-05,
      "loss": 0.0239,
      "step": 1000
    },
    {
      "epoch": 3.46,
      "learning_rate": 9.214780600461894e-06,
      "loss": 0.0117,
      "step": 1500
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.2863741339491914e-06,
      "loss": 0.0043,
      "step": 2000
    },
    {
      "epoch": 5.0,
      "step": 2165,
      "total_flos": 4753581654083658.0,
      "train_loss": 0.028315420073669998,
      "train_runtime": 457.8224,
      "train_samples_per_second": 75.521,
      "train_steps_per_second": 4.729
    }
  ],
  "max_steps": 2165,
  "num_train_epochs": 5,
  "total_flos": 4753581654083658.0,
  "trial_name": null,
  "trial_params": null
}