run-3/checkpoint-1000/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.680965147453083,
  "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": 0.6443178057670593,
      "eval_loss": 0.6200888156890869,
      "eval_runtime": 9.4558,
      "eval_samples_per_second": 315.469,
      "eval_steps_per_second": 19.776,
      "step": 373
    },
    {
      "epoch": 1.3404825737265416,
      "grad_norm": 11.498894691467285,
      "learning_rate": 1.0365748013479516e-05,
      "loss": 0.6738,
      "step": 500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.7539389729499817,
      "eval_loss": 0.47277966141700745,
      "eval_runtime": 9.3806,
      "eval_samples_per_second": 317.998,
      "eval_steps_per_second": 19.935,
      "step": 746
    },
    {
      "epoch": 2.680965147453083,
      "grad_norm": 9.671843528747559,
      "learning_rate": 8.761143057832532e-06,
      "loss": 0.4763,
      "step": 1000
    }
  ],
  "logging_steps": 500,
  "max_steps": 3730,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 5616578454487560.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": {
    "learning_rate": 1.1970352969126499e-05,
    "per_device_train_batch_size": 32
  }
}