trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 411,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.73,
      "learning_rate": 4.21409214092141e-05,
      "loss": 1.6779,
      "step": 100
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.859078590785908e-05,
      "loss": 1.0913,
      "step": 200
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.5040650406504067e-05,
      "loss": 0.7497,
      "step": 300
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.4905149051490515e-06,
      "loss": 0.4489,
      "step": 400
    },
    {
      "epoch": 3.0,
      "step": 411,
      "total_flos": 633388750422912.0,
      "train_loss": 0.9768911893060318,
      "train_runtime": 400.4654,
      "train_samples_per_second": 65.429,
      "train_steps_per_second": 1.026
    }
  ],
  "max_steps": 411,
  "num_train_epochs": 3,
  "total_flos": 633388750422912.0,
  "trial_name": null,
  "trial_params": null
}