trainer_state.json

{
  "best_metric": 0.9402330079719138,
  "best_model_checkpoint": "model_saves/roberta-large_lemon-spell_5k_6_p3/checkpoint-268",
  "epoch": 4.0,
  "global_step": 1072,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9402330079719138,
      "eval_loss": 0.42651739716529846,
      "eval_runtime": 3.4029,
      "eval_samples_per_second": 1287.737,
      "eval_steps_per_second": 10.285,
      "step": 268
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.940137971131162,
      "eval_loss": 0.43512868881225586,
      "eval_runtime": 3.4041,
      "eval_samples_per_second": 1287.267,
      "eval_steps_per_second": 10.282,
      "step": 536
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9395677500866513,
      "eval_loss": 0.4573929011821747,
      "eval_runtime": 3.3976,
      "eval_samples_per_second": 1289.716,
      "eval_steps_per_second": 10.301,
      "step": 804
    },
    {
      "epoch": 3.73,
      "learning_rate": 1e-05,
      "loss": 0.3181,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9387459609342681,
      "eval_loss": 0.49689212441444397,
      "eval_runtime": 3.39,
      "eval_samples_per_second": 1292.627,
      "eval_steps_per_second": 10.324,
      "step": 1072
    },
    {
      "epoch": 4.0,
      "step": 1072,
      "total_flos": 1.3860081393205248e+16,
      "train_loss": 0.31256330547048083,
      "train_runtime": 318.4127,
      "train_samples_per_second": 1616.016,
      "train_steps_per_second": 12.625
    }
  ],
  "max_steps": 4020,
  "num_train_epochs": 15,
  "total_flos": 1.3860081393205248e+16,
  "trial_name": null,
  "trial_params": null
}