trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 49.98360655737705,
  "global_step": 2400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 10.41,
      "learning_rate": 3.958333333333333e-05,
      "loss": 2.9289,
      "step": 500
    },
    {
      "epoch": 20.82,
      "learning_rate": 2.916666666666667e-05,
      "loss": 2.5801,
      "step": 1000
    },
    {
      "epoch": 31.25,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 2.3955,
      "step": 1500
    },
    {
      "epoch": 41.66,
      "learning_rate": 8.333333333333334e-06,
      "loss": 2.2678,
      "step": 2000
    },
    {
      "epoch": 49.98,
      "step": 2400,
      "total_flos": 2.9654605052706816e+17,
      "train_loss": 2.4875941975911458,
      "train_runtime": 37059.0269,
      "train_samples_per_second": 84.247,
      "train_steps_per_second": 0.065
    }
  ],
  "max_steps": 2400,
  "num_train_epochs": 50,
  "total_flos": 2.9654605052706816e+17,
  "trial_name": null,
  "trial_params": null
}