metadata
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
num_examples: 20000
download_size: 3978058364
dataset_size: 3979598471
configs:
- config_name: default
data_files:
- split: train
path: data/train-*