---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: ocr
    list:
    - name: box
      sequence: float64
    - name: cls
      dtype: int64
    - name: word
      dtype: string
  splits:
  - name: train
    num_bytes: 3979598471.0
    num_examples: 20000
  download_size: 3978058364
  dataset_size: 3979598471.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---