metadata
dataset_info:
features:
- name: sample_id
dtype: string
- name: img_path
dtype: binary
- name: dataset_name
dtype: string
- name: ocr_bboxes
struct:
- name: word
sequence:
sequence: int64
- name: ocr_predictions
struct:
- name: word
sequence: string
- name: source_text
dtype: string
- name: summary_text
dtype: string
- name: table_xml_path
dtype: string
splits:
- name: train
num_bytes: 287281984784
num_examples: 223008
download_size: 281987898601
dataset_size: 287281984784
configs:
- config_name: default
data_files:
- split: train
path: data/train-*