{
  "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": 0.7696949243545532,
      "eval_loss": 0.4614783227443695,
      "eval_runtime": 8.9518,
      "eval_samples_per_second": 333.23,
      "eval_steps_per_second": 20.89,
      "step": 187
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8045591711997986,
      "eval_loss": 0.4153314232826233,
      "eval_runtime": 9.0157,
      "eval_samples_per_second": 330.868,
      "eval_steps_per_second": 20.742,
      "step": 374
    },
    {
      "epoch": 2.6737967914438503,
      "grad_norm": 13.34338665008545,
      "learning_rate": 9.020650824026705e-06,
      "loss": 0.4417,
      "step": 500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8132752180099487,
      "eval_loss": 0.45979902148246765,
      "eval_runtime": 8.9278,
      "eval_samples_per_second": 334.124,
      "eval_steps_per_second": 20.946,
      "step": 561
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8209855556488037,
      "eval_loss": 0.49998581409454346,
      "eval_runtime": 9.0014,
      "eval_samples_per_second": 331.392,
      "eval_steps_per_second": 20.775,
      "step": 748
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8307073712348938,
      "eval_loss": 0.5820282101631165,
      "eval_runtime": 9.2967,
      "eval_samples_per_second": 320.868,
      "eval_steps_per_second": 20.115,
      "step": 935
    },
    {
      "epoch": 5.347593582887701,
      "grad_norm": 13.978471755981445,
      "learning_rate": 5.728442494089952e-06,
      "loss": 0.1523,
      "step": 1000
    }
  ],
  "logging_steps": 500,
  "max_steps": 1870,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 1.0970385915579516e+16,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 1.231285915396346e-05,
    "per_device_train_batch_size": 64
  }
}