---
dataset_info:
  features:
  - name: image
    dtype: image
  - name: caption
    dtype: string
  - name: annotations
    sequence:
    - name: illegibility
      dtype: bool
    - name: language
      dtype: string
    - name: polygon
      sequence:
        sequence: int64
    - name: pos
      dtype: int64
    - name: text
      dtype: string
    - name: valid
      dtype: bool
    - name: rec_score
      dtype: float64
  splits:
  - name: train
    num_bytes: 1039767209.2
    num_examples: 4168
  download_size: 973143467
  dataset_size: 1039767209.2
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---