---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: conversations
    list:
    - name: role
      dtype: string
    - name: content
      dtype: string
  splits:
  - name: train
    num_bytes: 2979915678.5
    num_examples: 20212
  download_size: 2550161782
  dataset_size: 2979915678.5
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---