README.md

---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: ground_truth
    struct:
    - name: gt_parse
      list:
      - name: question
        dtype: string
      - name: answer
        dtype: string
  - name: pixel_values
    sequence:
      sequence:
        sequence: float32
  - name: labels
    sequence: int64
  - name: target_sequence
    dtype: string
  splits:
  - name: train
    num_bytes: 816349208.4
    num_examples: 90
  - name: test
    num_bytes: 90705467.6
    num_examples: 10
  download_size: 89805128
  dataset_size: 907054676.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
---