---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: elements
    sequence:
    - name: role
      dtype: string
    - name: text
      dtype: string
  splits:
  - name: train
    num_bytes: 188454566523.0
    num_examples: 2000000
  download_size: 188179589916
  dataset_size: 188454566523.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---