trainer_state.json

{
  "best_metric": 0.9813176703701083,
  "best_model_checkpoint": "../result/valentwin-academic_papers-n-100-hn-10-selective-neg-lr-3e5-bs-512-20240421",
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 576,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.2834,
      "step": 192
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9813176703701083,
      "step": 192
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.999999999999999e-06,
      "loss": 0.0286,
      "step": 384
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9725426367560682,
      "step": 384
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 0.0089,
      "step": 576
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9765763215625222,
      "step": 576
    },
    {
      "epoch": 3.0,
      "step": 576,
      "total_flos": 0,
      "train_runtime": 1380.2265,
      "train_samples_per_second": 0.417
    }
  ],
  "logging_steps": 500,
  "max_steps": 576,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 0,
  "train_batch_size": null,
  "trial_name": null,
  "trial_params": null
}