run-5/checkpoint-2138/trainer_state.json

{
  "best_metric": 0.4778162027316995,
  "best_model_checkpoint": "distilbert-base-uncased-finetuned-cola/run-5/checkpoint-2138",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 2138,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4677268475210477,
      "grad_norm": 4.435796737670898,
      "learning_rate": 1.7076894081996187e-05,
      "loss": 0.5421,
      "step": 500
    },
    {
      "epoch": 0.9354536950420954,
      "grad_norm": 13.997383117675781,
      "learning_rate": 1.3922683871062276e-05,
      "loss": 0.5008,
      "step": 1000
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.4736253321170807,
      "eval_matthews_correlation": 0.45008318903303624,
      "eval_runtime": 0.7693,
      "eval_samples_per_second": 1355.736,
      "eval_steps_per_second": 85.79,
      "step": 1069
    },
    {
      "epoch": 1.4031805425631432,
      "grad_norm": 34.2601432800293,
      "learning_rate": 1.0768473660128366e-05,
      "loss": 0.3575,
      "step": 1500
    },
    {
      "epoch": 1.8709073900841908,
      "grad_norm": 39.75000762939453,
      "learning_rate": 7.6142634491944576e-06,
      "loss": 0.3442,
      "step": 2000
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.6032345294952393,
      "eval_matthews_correlation": 0.4778162027316995,
      "eval_runtime": 0.9252,
      "eval_samples_per_second": 1127.3,
      "eval_steps_per_second": 71.334,
      "step": 2138
    }
  ],
  "logging_steps": 500,
  "max_steps": 3207,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 75157137960120.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 2.0231104292930096e-05,
    "num_train_epochs": 3,
    "per_device_train_batch_size": 8,
    "seed": 39
  }
}