trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "global_step": 2655,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.94,
      "learning_rate": 2.4350282485875707e-05,
      "loss": 0.0829,
      "step": 500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.870056497175141e-05,
      "loss": 0.0311,
      "step": 1000
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.3050847457627118e-05,
      "loss": 0.0178,
      "step": 1500
    },
    {
      "epoch": 3.77,
      "learning_rate": 7.401129943502825e-06,
      "loss": 0.0113,
      "step": 2000
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.7514124293785311e-06,
      "loss": 0.0062,
      "step": 2500
    },
    {
      "epoch": 5.0,
      "step": 2655,
      "total_flos": 3189153922083888.0,
      "train_loss": 0.028442844630634716,
      "train_runtime": 487.3991,
      "train_samples_per_second": 87.033,
      "train_steps_per_second": 5.447
    }
  ],
  "max_steps": 2655,
  "num_train_epochs": 5,
  "total_flos": 3189153922083888.0,
  "trial_name": null,
  "trial_params": null
}