trainer_state.json

{
  "best_metric": 4.924107074737549,
  "best_model_checkpoint": "ner-bert-ingredients/checkpoint-20",
  "epoch": 50.0,
  "eval_steps": 10,
  "global_step": 50,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 10.0,
      "learning_rate": 4e-05,
      "loss": 0.0466,
      "step": 10
    },
    {
      "epoch": 10.0,
      "eval_loss": 4.74609375,
      "eval_runtime": 0.0189,
      "eval_samples_per_second": 52.922,
      "eval_steps_per_second": 52.922,
      "step": 10
    },
    {
      "epoch": 20.0,
      "learning_rate": 4e-05,
      "loss": 0.0213,
      "step": 20
    },
    {
      "epoch": 20.0,
      "eval_loss": 4.924107074737549,
      "eval_runtime": 0.0183,
      "eval_samples_per_second": 54.52,
      "eval_steps_per_second": 54.52,
      "step": 20
    },
    {
      "epoch": 30.0,
      "learning_rate": 3e-05,
      "loss": 0.01,
      "step": 30
    },
    {
      "epoch": 30.0,
      "eval_loss": 5.040736675262451,
      "eval_runtime": 0.0178,
      "eval_samples_per_second": 56.023,
      "eval_steps_per_second": 56.023,
      "step": 30
    },
    {
      "epoch": 40.0,
      "learning_rate": 2e-05,
      "loss": 0.0063,
      "step": 40
    },
    {
      "epoch": 40.0,
      "eval_loss": 5.1612725257873535,
      "eval_runtime": 0.0185,
      "eval_samples_per_second": 54.098,
      "eval_steps_per_second": 54.098,
      "step": 40
    },
    {
      "epoch": 50.0,
      "learning_rate": 4e-05,
      "loss": 0.0043,
      "step": 50
    },
    {
      "epoch": 50.0,
      "eval_loss": 5.35546875,
      "eval_runtime": 0.0182,
      "eval_samples_per_second": 55.091,
      "eval_steps_per_second": 55.091,
      "step": 50
    },
    {
      "epoch": 50.0,
      "step": 50,
      "total_flos": 32856154788600.0,
      "train_loss": 0.0008505997806787491,
      "train_runtime": 23.3473,
      "train_samples_per_second": 19.274,
      "train_steps_per_second": 2.142
    }
  ],
  "logging_steps": 10,
  "max_steps": 50,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 50,
  "save_steps": 10,
  "total_flos": 32856154788600.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}