run-3/checkpoint-842/trainer_state.json

{
  "best_metric": 0.9174311926605505,
  "best_model_checkpoint": "bert-base-uncased-finetuned-sst2/run-3/checkpoint-842",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 842,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8784403669724771,
      "eval_loss": 0.32404395937919617,
      "eval_runtime": 2.2657,
      "eval_samples_per_second": 384.875,
      "eval_steps_per_second": 24.275,
      "step": 421
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.08605317771434784,
      "learning_rate": 3.2162650675262254e-05,
      "loss": 0.3042,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9174311926605505,
      "eval_loss": 0.2911108732223511,
      "eval_runtime": 2.3208,
      "eval_samples_per_second": 375.734,
      "eval_steps_per_second": 23.699,
      "step": 842
    }
  ],
  "logging_steps": 500,
  "max_steps": 1263,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 146782052231400.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 5.323909279535547e-05,
    "num_train_epochs": 3,
    "per_device_train_batch_size": 16,
    "seed": 8
  }
}