---
dataset_info:
  features:
  - name: caption
    dtype: string
  - name: image_tokens
    dtype: string
  - name: user
    dtype: string
  - name: assistant
    dtype: string
  - name: conversations
    list:
    - name: content
      dtype: string
    - name: role
      dtype: string
  splits:
  - name: train
    num_bytes: 10900258198
    num_examples: 519288
  download_size: 4159919663
  dataset_size: 10900258198
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---