---
dataset_info:
  features:
  - name: id
    dtype: string
  - name: image
    dtype: string
  - name: conversations
    list:
    - name: from
      dtype: string
    - name: value
      dtype: string
  splits:
  - name: pretrain_json
    num_bytes: 93970446
    num_examples: 558128
  download_size: 36011763
  dataset_size: 93970446
configs:
- config_name: default
  data_files:
  - split: pretrain_json
    path: data/pretrain_json-*
---