{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.347593582887701,
  "eval_steps": 500,
  "global_step": 1000,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.7576265335083008,
      "eval_loss": 0.4841073751449585,
      "eval_runtime": 9.3726,
      "eval_samples_per_second": 318.268,
      "eval_steps_per_second": 19.952,
      "step": 187
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7961783409118652,
      "eval_loss": 0.49574801325798035,
      "eval_runtime": 9.44,
      "eval_samples_per_second": 315.995,
      "eval_steps_per_second": 19.809,
      "step": 374
    },
    {
      "epoch": 2.6737967914438503,
      "grad_norm": 15.436814308166504,
      "learning_rate": 2.2676392198891078e-05,
      "loss": 0.4296,
      "step": 500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7998659014701843,
      "eval_loss": 0.4749740660190582,
      "eval_runtime": 9.2752,
      "eval_samples_per_second": 321.61,
      "eval_steps_per_second": 20.161,
      "step": 561
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8008716106414795,
      "eval_loss": 0.6543935537338257,
      "eval_runtime": 9.3222,
      "eval_samples_per_second": 319.988,
      "eval_steps_per_second": 20.06,
      "step": 748
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8022125363349915,
      "eval_loss": 0.8691691160202026,
      "eval_runtime": 9.7075,
      "eval_samples_per_second": 307.287,
      "eval_steps_per_second": 19.263,
      "step": 935
    },
    {
      "epoch": 5.347593582887701,
      "grad_norm": 26.959753036499023,
      "learning_rate": 1.4400336651850539e-05,
      "loss": 0.1024,
      "step": 1000
    }
  ],
  "logging_steps": 500,
  "max_steps": 1870,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 1.1927468911210596e+16,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 3.095244774593162e-05,
    "per_device_train_batch_size": 64
  }
}