---
dataset_info:
  features:
  - name: messages
    sequence:
    - name: role
      dtype: string
    - name: content
      sequence:
      - name: type
        dtype: string
      - name: text
        dtype: string
      - name: image
        dtype: image
  splits:
  - name: train
    num_bytes: 5584537898.28
    num_examples: 16494
  download_size: 5481920052
  dataset_size: 5584537898.28
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---