trainer_state.json

{
  "best_metric": 2.426845073699951,
  "best_model_checkpoint": "output/arash/checkpoint-21",
  "epoch": 1.0,
  "global_step": 21,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.24,
      "learning_rate": 0.00011888735840752609,
      "loss": 3.4831,
      "step": 5
    },
    {
      "epoch": 0.48,
      "learning_rate": 7.372648442002871e-05,
      "loss": 2.7484,
      "step": 10
    },
    {
      "epoch": 0.71,
      "learning_rate": 2.5828599592490882e-05,
      "loss": 2.7318,
      "step": 15
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.662053209561833e-07,
      "loss": 2.6885,
      "step": 20
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.426845073699951,
      "eval_runtime": 1.2965,
      "eval_samples_per_second": 22.367,
      "eval_steps_per_second": 3.085,
      "step": 21
    }
  ],
  "max_steps": 21,
  "num_train_epochs": 1,
  "total_flos": 21817884672000.0,
  "trial_name": null,
  "trial_params": null
}