run-3/checkpoint-1684/trainer_state.json

{
  "best_metric": 0.8899082568807339,
  "best_model_checkpoint": "bert-base-uncased-finetuned-sst2/run-3/checkpoint-1684",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 1684,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.3,
      "grad_norm": 7.637521743774414,
      "learning_rate": 3.0568943955117345e-06,
      "loss": 0.5094,
      "step": 500
    },
    {
      "epoch": 0.59,
      "grad_norm": 104.65070343017578,
      "learning_rate": 2.721119527814927e-06,
      "loss": 0.4332,
      "step": 1000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.10155901312828064,
      "learning_rate": 2.3853446601181197e-06,
      "loss": 0.4318,
      "step": 1500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8899082568807339,
      "eval_loss": 0.4135288596153259,
      "eval_runtime": 2.3891,
      "eval_samples_per_second": 364.998,
      "eval_steps_per_second": 23.022,
      "step": 1684
    }
  ],
  "logging_steps": 500,
  "max_steps": 5052,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 74152095398880.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 3.3926692632085418e-06,
    "num_train_epochs": 3,
    "per_device_train_batch_size": 4,
    "seed": 6
  }
}