trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 2040,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.49,
      "learning_rate": 4.898708324437183e-05,
      "loss": 1.8409,
      "step": 500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.328609502502189e-05,
      "loss": 1.189,
      "step": 1000
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.758510680567194e-05,
      "loss": 0.9051,
      "step": 1500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.8841185863219934e-06,
      "loss": 0.8443,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "step": 2040,
      "total_flos": 70197983838208.0,
      "train_loss": 1.1858560001148897,
      "train_runtime": 328.5139,
      "train_samples_per_second": 794.755,
      "train_steps_per_second": 6.21
    }
  ],
  "max_steps": 2040,
  "num_train_epochs": 2,
  "total_flos": 70197983838208.0,
  "trial_name": null,
  "trial_params": null
}