{
  "best_metric": 0.8910550458715596,
  "best_model_checkpoint": "electra-distilled-sst\\run-1\\checkpoint-1054",
  "epoch": 4.0,
  "eval_steps": 500,
  "global_step": 2108,
  "is_hyper_param_search": true,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "grad_norm": 66.74252319335938,
      "learning_rate": 0.0002000029100592468,
      "loss": 2.5183,
      "step": 527
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.875,
      "eval_loss": 2.3820443153381348,
      "eval_runtime": 1.4493,
      "eval_samples_per_second": 601.689,
      "eval_steps_per_second": 4.83,
      "step": 527
    },
    {
      "epoch": 2.0,
      "grad_norm": 67.3247299194336,
      "learning_rate": 0.00015007323994971434,
      "loss": 1.237,
      "step": 1054
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8910550458715596,
      "eval_loss": 2.0578901767730713,
      "eval_runtime": 1.4579,
      "eval_samples_per_second": 598.133,
      "eval_steps_per_second": 4.802,
      "step": 1054
    },
    {
      "epoch": 3.0,
      "grad_norm": 9.862482070922852,
      "learning_rate": 0.00010014356984018184,
      "loss": 0.8517,
      "step": 1581
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8841743119266054,
      "eval_loss": 2.112179756164551,
      "eval_runtime": 1.4398,
      "eval_samples_per_second": 605.639,
      "eval_steps_per_second": 4.862,
      "step": 1581
    },
    {
      "epoch": 4.0,
      "grad_norm": 9.713326454162598,
      "learning_rate": 5.021389973064936e-05,
      "loss": 0.62,
      "step": 2108
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8784403669724771,
      "eval_loss": 2.1044843196868896,
      "eval_runtime": 1.4493,
      "eval_samples_per_second": 601.685,
      "eval_steps_per_second": 4.83,
      "step": 2108
    }
  ],
  "logging_steps": 500,
  "max_steps": 2635,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 748818980000436.0,
  "train_batch_size": 128,
  "trial_name": null,
  "trial_params": {
    "alpha": 0.15523880635662113,
    "learning_rate": 0.0002496483505476624,
    "num_train_epochs": 5,
    "temperature": 16
  }
}