trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "global_step": 310,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 3.23,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 7.9382,
      "step": 100
    },
    {
      "epoch": 6.45,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 3.6371,
      "step": 200
    },
    {
      "epoch": 9.68,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 3.1273,
      "step": 300
    },
    {
      "epoch": 10.0,
      "step": 310,
      "total_flos": 2.1035435929717018e+18,
      "train_loss": 4.839959698338663,
      "train_runtime": 859.0241,
      "train_samples_per_second": 11.525,
      "train_steps_per_second": 0.361
    }
  ],
  "max_steps": 310,
  "num_train_epochs": 10,
  "total_flos": 2.1035435929717018e+18,
  "trial_name": null,
  "trial_params": null
}