---
dataset_info:
  features:
  - name: instruction
    dtype: string
  - name: prefer
    dtype: string
  - name: rejected
    dtype: string
  - name: images
    list:
    - name: bytes
      dtype: binary
    - name: path
      dtype: 'null'
  splits:
  - name: train
    num_bytes: 2112198428
    num_examples: 49972
  download_size: 2089954601
  dataset_size: 2112198428
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---