---
dataset_info:
  features:
  - name: images
    list: image
  - name: messages
    list:
    - name: content
      list: json
    - name: role
      dtype: string
  splits:
  - name: train
    num_bytes: 3266962724
    num_examples: 27138
  download_size: 3266919818
  dataset_size: 3266962724
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---