metadata
dataset_info:
features:
- name: question
dtype: string
- name: questionId
dtype: string
- name: question_types
dtype: string
- name: image
struct:
- name: bytes
dtype: binary
- name: path
dtype: string
- name: answer
struct:
- name: DOCVQA
dtype: string
splits:
- name: train
num_bytes: 1130015136
num_examples: 30205
download_size: 1090316316
dataset_size: 1130015136
configs:
- config_name: default
data_files:
- split: train
path: data/train-*