trainer_state.json

{
  "best_metric": 0.9415858853520276,
  "best_model_checkpoint": "model_saves/deberta-large_lemon-spell_5k_3_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": 0.9415858853520276,
      "eval_loss": 0.41651687026023865,
      "eval_runtime": 4.6276,
      "eval_samples_per_second": 946.933,
      "eval_steps_per_second": 7.563,
      "step": 268
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9410380259171055,
      "eval_loss": 0.4360532760620117,
      "eval_runtime": 4.6352,
      "eval_samples_per_second": 945.374,
      "eval_steps_per_second": 7.551,
      "step": 536
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9401826943503393,
      "eval_loss": 0.4828942120075226,
      "eval_runtime": 4.6438,
      "eval_samples_per_second": 943.624,
      "eval_steps_per_second": 7.537,
      "step": 804
    },
    {
      "epoch": 3.73,
      "learning_rate": 1e-05,
      "loss": 0.256,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.94000380147363,
      "eval_loss": 0.5373597741127014,
      "eval_runtime": 4.6325,
      "eval_samples_per_second": 945.922,
      "eval_steps_per_second": 7.555,
      "step": 1072
    },
    {
      "epoch": 4.0,
      "step": 1072,
      "total_flos": 1.6078384287186944e+16,
      "train_loss": 0.24955286285770473,
      "train_runtime": 463.7536,
      "train_samples_per_second": 1109.555,
      "train_steps_per_second": 8.668
    }
  ],
  "max_steps": 4020,
  "num_train_epochs": 15,
  "total_flos": 1.6078384287186944e+16,
  "trial_name": null,
  "trial_params": null
}