run-1/checkpoint-102/trainer_state.json

{
  "best_metric": 0.6329721808433533,
  "best_model_checkpoint": "BERT-WMM/run-1/checkpoint-102",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 102,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7056338028169014,
      "eval_loss": 0.7160280346870422,
      "eval_runtime": 2.4586,
      "eval_samples_per_second": 866.36,
      "eval_steps_per_second": 54.503,
      "step": 34
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7389671361502348,
      "eval_loss": 0.6426944732666016,
      "eval_runtime": 2.5092,
      "eval_samples_per_second": 848.874,
      "eval_steps_per_second": 53.403,
      "step": 68
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7394366197183099,
      "eval_loss": 0.6329721808433533,
      "eval_runtime": 2.4846,
      "eval_samples_per_second": 857.288,
      "eval_steps_per_second": 53.933,
      "step": 102
    }
  ],
  "logging_steps": 500,
  "max_steps": 102,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 0,
  "train_batch_size": 256,
  "trial_name": null,
  "trial_params": {
    "per_device_train_batch_size": 256
  }
}