run-1/checkpoint-4210/trainer_state.json

{
  "best_metric": 0.8211009174311926,
  "best_model_checkpoint": "Tech-oriented/best_model_bert_uncasedbert-base-uncased-finetuned-sst2/run-1/checkpoint-4210",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 4210,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 19.184335708618164,
      "learning_rate": 6.400602545895411e-07,
      "loss": 0.684,
      "step": 4209
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8211009174311926,
      "eval_loss": 0.5808640718460083,
      "eval_runtime": 2.4176,
      "eval_samples_per_second": 360.683,
      "eval_steps_per_second": 22.749,
      "step": 4210
    }
  ],
  "logging_steps": 4209,
  "max_steps": 8420,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "total_flos": 1218763297309440.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 1.2798165147575246e-06,
    "num_train_epochs": 2,
    "per_device_train_batch_size": 16,
    "seed": 24,
    "weight_decay": 0.022271875197444224
  }
}