trainer_state.json

{
  "best_metric": 0.940070886302396,
  "best_model_checkpoint": "model_saves/roberta-large_lemon-spell_5k_4_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": 0.940070886302396,
      "eval_loss": 0.42090117931365967,
      "eval_runtime": 3.3888,
      "eval_samples_per_second": 1293.092,
      "eval_steps_per_second": 10.328,
      "step": 268
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9392379163452186,
      "eval_loss": 0.443440705537796,
      "eval_runtime": 3.383,
      "eval_samples_per_second": 1295.312,
      "eval_steps_per_second": 10.346,
      "step": 536
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9395174364650768,
      "eval_loss": 0.4689682424068451,
      "eval_runtime": 3.3773,
      "eval_samples_per_second": 1297.495,
      "eval_steps_per_second": 10.363,
      "step": 804
    },
    {
      "epoch": 3.73,
      "learning_rate": 1e-05,
      "loss": 0.2919,
      "step": 1000
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.937823544538736,
      "eval_loss": 0.5258461833000183,
      "eval_runtime": 3.3916,
      "eval_samples_per_second": 1292.024,
      "eval_steps_per_second": 10.32,
      "step": 1072
    },
    {
      "epoch": 4.0,
      "step": 1072,
      "total_flos": 1.3870742575775744e+16,
      "train_loss": 0.2865001568153723,
      "train_runtime": 320.3001,
      "train_samples_per_second": 1606.493,
      "train_steps_per_second": 12.551
    }
  ],
  "max_steps": 4020,
  "num_train_epochs": 15,
  "total_flos": 1.3870742575775744e+16,
  "trial_name": null,
  "trial_params": null
}