trainer_state.json

{
  "best_metric": 0.9414014020729212,
  "best_model_checkpoint": "model_saves/deberta-large_spell_10k_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": 0.9414014020729212,
      "eval_loss": 0.42718926072120667,
      "eval_runtime": 4.6994,
      "eval_samples_per_second": 932.461,
      "eval_steps_per_second": 7.448,
      "step": 268
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9404510336654033,
      "eval_loss": 0.44841113686561584,
      "eval_runtime": 4.6847,
      "eval_samples_per_second": 935.388,
      "eval_steps_per_second": 7.471,
      "step": 536
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9399367166448641,
      "eval_loss": 0.4899842143058777,
      "eval_runtime": 4.6881,
      "eval_samples_per_second": 934.704,
      "eval_steps_per_second": 7.466,
      "step": 804
    },
    {
      "epoch": 3.73,
      "learning_rate": 1e-05,
      "loss": 0.2471,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9396460157202116,
      "eval_loss": 0.534565269947052,
      "eval_runtime": 4.6788,
      "eval_samples_per_second": 936.574,
      "eval_steps_per_second": 7.481,
      "step": 1072
    },
    {
      "epoch": 4.0,
      "step": 1072,
      "total_flos": 1.6324310893330432e+16,
      "train_loss": 0.24090948923310238,
      "train_runtime": 462.7295,
      "train_samples_per_second": 1112.01,
      "train_steps_per_second": 8.688
    }
  ],
  "max_steps": 4020,
  "num_train_epochs": 15,
  "total_flos": 1.6324310893330432e+16,
  "trial_name": null,
  "trial_params": null
}