trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 1410,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.55,
      "learning_rate": 1.2907801418439719e-05,
      "loss": 0.0321,
      "step": 500
    },
    {
      "epoch": 7.09,
      "learning_rate": 5.815602836879432e-06,
      "loss": 0.0039,
      "step": 1000
    },
    {
      "epoch": 10.0,
      "step": 1410,
      "total_flos": 2367734848668648.0,
      "train_loss": 0.01299606111455471,
      "train_runtime": 444.0291,
      "train_samples_per_second": 50.672,
      "train_steps_per_second": 3.175
    }
  ],
  "max_steps": 1410,
  "num_train_epochs": 10,
  "total_flos": 2367734848668648.0,
  "trial_name": null,
  "trial_params": null
}