---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: qas
    list:
    - name: question
      dtype: string
    - name: answers
      list: string
  splits:
  - name: train
    num_bytes: 2951305494.152
    num_examples: 18921
  download_size: 2995367927
  dataset_size: 2951305494.152
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---

https://arxiv.org/abs/1905.13648