checkpoint-120/trainer_state.json

{
  "best_metric": 0.8277802155149765,
  "best_model_checkpoint": "distilhubert-finetuned-mixed-data/checkpoint-100",
  "epoch": 26.666666666666668,
  "eval_steps": 100,
  "global_step": 120,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 11.11111111111111,
      "grad_norm": 2.099414587020874,
      "learning_rate": 0.00021731987703006933,
      "loss": 0.7753,
      "step": 50
    },
    {
      "epoch": 22.22222222222222,
      "grad_norm": 0.09225956350564957,
      "learning_rate": 2.4676828288059558e-05,
      "loss": 0.3691,
      "step": 100
    },
    {
      "epoch": 22.22222222222222,
      "eval_accuracy": 0.8241758241758241,
      "eval_confusion_matrix": [
        [
          51,
          10,
          0,
          2
        ],
        [
          5,
          44,
          9,
          0
        ],
        [
          1,
          14,
          67,
          0
        ],
        [
          7,
          0,
          0,
          63
        ]
      ],
      "eval_f1": 0.8277802155149765,
      "eval_loss": 0.7942458391189575,
      "eval_precision": 0.8346819204947628,
      "eval_recall": 0.8241758241758241,
      "eval_runtime": 3.7898,
      "eval_samples_per_second": 72.036,
      "eval_steps_per_second": 0.792,
      "step": 100
    }
  ],
  "logging_steps": 50,
  "max_steps": 120,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 30,
  "save_steps": 100,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.001
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 6.619818670848e+16,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": null
}