trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.560819462227913,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.64,
      "learning_rate": 1.573196756295348e-05,
      "loss": 1.2457,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.8522860407829285,
      "eval_runtime": 1521.335,
      "eval_samples_per_second": 3.064,
      "eval_steps_per_second": 0.383,
      "step": 781
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.1463935125906958e-05,
      "loss": 0.9332,
      "step": 1000
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.195902688860435e-06,
      "loss": 0.8339,
      "step": 1500
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.7283812761306763,
      "eval_runtime": 1308.1173,
      "eval_samples_per_second": 3.563,
      "eval_steps_per_second": 0.446,
      "step": 1562
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.927870251813914e-06,
      "loss": 0.7828,
      "step": 2000
    }
  ],
  "max_steps": 2343,
  "num_train_epochs": 3,
  "total_flos": 2070920593170000.0,
  "trial_name": null,
  "trial_params": null
}