run-4/checkpoint-500/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.6737967914438503,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7254441976547241,
      "eval_loss": 0.515117347240448,
      "eval_runtime": 9.0075,
      "eval_samples_per_second": 331.167,
      "eval_steps_per_second": 20.76,
      "step": 187
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7492457032203674,
      "eval_loss": 0.4793011546134949,
      "eval_runtime": 9.1228,
      "eval_samples_per_second": 326.983,
      "eval_steps_per_second": 20.498,
      "step": 374
    },
    {
      "epoch": 2.6737967914438503,
      "grad_norm": 12.898003578186035,
      "learning_rate": 2.66142701656233e-06,
      "loss": 0.5477,
      "step": 500
    }
  ],
  "logging_steps": 500,
  "max_steps": 1870,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 5510487892159476.0,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 3.6327507452347134e-06,
    "per_device_train_batch_size": 64
  }
}