trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 2692,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.37,
      "learning_rate": 2.448365527488856e-05,
      "loss": 0.345,
      "step": 500
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.891158989598811e-05,
      "loss": 0.1299,
      "step": 1000
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.3339524517087667e-05,
      "loss": 0.0899,
      "step": 1500
    },
    {
      "epoch": 1.49,
      "learning_rate": 7.767459138187223e-06,
      "loss": 0.0673,
      "step": 2000
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.2065378900445768e-06,
      "loss": 0.0638,
      "step": 2500
    },
    {
      "epoch": 2.0,
      "step": 2692,
      "total_flos": 4.664909366506291e+16,
      "train_loss": 0.13270232688087588,
      "train_runtime": 3759.1301,
      "train_samples_per_second": 5.726,
      "train_steps_per_second": 0.716
    }
  ],
  "max_steps": 2692,
  "num_train_epochs": 2,
  "total_flos": 4.664909366506291e+16,
  "trial_name": null,
  "trial_params": null
}