metadata
dataset_info:
features:
- name: question
dtype: string
- name: docId
dtype: int64
- name: answers
sequence: string
- name: data_split
dtype: string
- name: bounding_boxes
sequence:
sequence: int64
- name: word_list
sequence: string
- name: image_raw
dtype: image
- name: ground_truth
dtype: string
splits:
- name: train
num_bytes: 324385610
num_examples: 2000
- name: valid
num_bytes: 207926530
num_examples: 400
download_size: 0
dataset_size: 532312140
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: valid
path: data/valid-*