{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.361930294906166,
  "eval_steps": 500,
  "global_step": 4000,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.6702412868632708,
      "grad_norm": 13.834343910217285,
      "learning_rate": 1.8689758651553552e-05,
      "loss": 0.5491,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.762319803237915,
      "eval_loss": 0.4773792326450348,
      "eval_runtime": 8.5342,
      "eval_samples_per_second": 349.533,
      "eval_steps_per_second": 21.912,
      "step": 746
    },
    {
      "epoch": 1.3404825737265416,
      "grad_norm": 12.70506477355957,
      "learning_rate": 1.7347103576010912e-05,
      "loss": 0.4116,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7696949243545532,
      "eval_loss": 0.5922191739082336,
      "eval_runtime": 8.5618,
      "eval_samples_per_second": 348.409,
      "eval_steps_per_second": 21.841,
      "step": 1492
    },
    {
      "epoch": 2.0107238605898123,
      "grad_norm": 100.83161163330078,
      "learning_rate": 1.6004448500468272e-05,
      "loss": 0.2993,
      "step": 1500
    },
    {
      "epoch": 2.680965147453083,
      "grad_norm": 251.75213623046875,
      "learning_rate": 1.4661793424925633e-05,
      "loss": 0.1136,
      "step": 2000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7676835656166077,
      "eval_loss": 0.9344700574874878,
      "eval_runtime": 8.6072,
      "eval_samples_per_second": 346.57,
      "eval_steps_per_second": 21.726,
      "step": 2238
    },
    {
      "epoch": 3.351206434316354,
      "grad_norm": 17.935895919799805,
      "learning_rate": 1.3319138349382991e-05,
      "loss": 0.1,
      "step": 2500
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.777405321598053,
      "eval_loss": 1.230825662612915,
      "eval_runtime": 8.5403,
      "eval_samples_per_second": 349.285,
      "eval_steps_per_second": 21.896,
      "step": 2984
    },
    {
      "epoch": 4.021447721179625,
      "grad_norm": 0.8948413729667664,
      "learning_rate": 1.1976483273840351e-05,
      "loss": 0.0715,
      "step": 3000
    },
    {
      "epoch": 4.6916890080428955,
      "grad_norm": 0.7082040309906006,
      "learning_rate": 1.063382819829771e-05,
      "loss": 0.0294,
      "step": 3500
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.7763996124267578,
      "eval_loss": 1.194653034210205,
      "eval_runtime": 8.5079,
      "eval_samples_per_second": 350.617,
      "eval_steps_per_second": 21.98,
      "step": 3730
    },
    {
      "epoch": 5.361930294906166,
      "grad_norm": 0.04226335510611534,
      "learning_rate": 9.29117312275507e-06,
      "loss": 0.025,
      "step": 4000
    }
  ],
  "logging_steps": 500,
  "max_steps": 7460,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 9475245428463684.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 2.0032413727096193e-05,
    "per_device_train_batch_size": 16
  }
}