{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.383378016085791,
  "eval_steps": 500,
  "global_step": 3500,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_accuracy": 0.6661079525947571,
      "eval_loss": 0.5971955060958862,
      "eval_runtime": 15.5437,
      "eval_samples_per_second": 191.91,
      "eval_steps_per_second": 12.031,
      "step": 373
    },
    {
      "epoch": 1.3404825737265416,
      "grad_norm": 10.565417289733887,
      "learning_rate": 1.8562171591313335e-05,
      "loss": 0.6088,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.68890380859375,
      "eval_loss": 0.554870069026947,
      "eval_runtime": 20.9818,
      "eval_samples_per_second": 142.171,
      "eval_steps_per_second": 8.912,
      "step": 746
    },
    {
      "epoch": 2.680965147453083,
      "grad_norm": 12.369528770446777,
      "learning_rate": 1.5688770416187432e-05,
      "loss": 0.4913,
      "step": 1000
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.6895742416381836,
      "eval_loss": 0.6084151268005371,
      "eval_runtime": 21.3765,
      "eval_samples_per_second": 139.546,
      "eval_steps_per_second": 8.748,
      "step": 1119
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.6959436535835266,
      "eval_loss": 0.6611299514770508,
      "eval_runtime": 14.7723,
      "eval_samples_per_second": 201.931,
      "eval_steps_per_second": 12.659,
      "step": 1492
    },
    {
      "epoch": 4.021447721179625,
      "grad_norm": 14.231189727783203,
      "learning_rate": 1.2815369241061527e-05,
      "loss": 0.3834,
      "step": 1500
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.7039892673492432,
      "eval_loss": 0.6887115240097046,
      "eval_runtime": 8.0268,
      "eval_samples_per_second": 371.63,
      "eval_steps_per_second": 23.297,
      "step": 1865
    },
    {
      "epoch": 5.361930294906166,
      "grad_norm": 17.718364715576172,
      "learning_rate": 9.941968065935624e-06,
      "loss": 0.2616,
      "step": 2000
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.7019778490066528,
      "eval_loss": 0.7845392227172852,
      "eval_runtime": 57.529,
      "eval_samples_per_second": 51.852,
      "eval_steps_per_second": 3.251,
      "step": 2238
    },
    {
      "epoch": 6.702412868632708,
      "grad_norm": 17.930471420288086,
      "learning_rate": 7.0685668908097215e-06,
      "loss": 0.1949,
      "step": 2500
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.6979550719261169,
      "eval_loss": 0.8831968307495117,
      "eval_runtime": 50.8363,
      "eval_samples_per_second": 58.679,
      "eval_steps_per_second": 3.678,
      "step": 2611
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.7003017067909241,
      "eval_loss": 0.9950036406517029,
      "eval_runtime": 50.3735,
      "eval_samples_per_second": 59.218,
      "eval_steps_per_second": 3.712,
      "step": 2984
    },
    {
      "epoch": 8.04289544235925,
      "grad_norm": 12.573803901672363,
      "learning_rate": 4.1951657156838185e-06,
      "loss": 0.1515,
      "step": 3000
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.6962789297103882,
      "eval_loss": 1.03825843334198,
      "eval_runtime": 58.6383,
      "eval_samples_per_second": 50.871,
      "eval_steps_per_second": 3.189,
      "step": 3357
    },
    {
      "epoch": 9.383378016085791,
      "grad_norm": 25.479394912719727,
      "learning_rate": 1.3217645405579155e-06,
      "loss": 0.1231,
      "step": 3500
    }
  ],
  "logging_steps": 500,
  "max_steps": 3730,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5252134618773024.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 2.1435572766439238e-05,
    "per_device_train_batch_size": 32
  }
}