---
dataset_info:
  features:
  - name: prompt
    dtype: string
  - name: chosen
    dtype: string
  - name: rejected
    dtype: string
  - name: chosen_pi_score
    dtype: float64
  - name: rejected_pi_score
    dtype: float64
  - name: agree
    dtype: int64
  splits:
  - name: train
    num_bytes: 850804
    num_examples: 500
  download_size: 94829
  dataset_size: 850804
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---