metadata
dataset_info:
features:
- name: image
dtype: image
- name: id
dtype: string
- name: long_answer_type
dtype: string
- name: question
dtype: string
- name: short_answer
dtype: string
- name: candidates
sequence: string
- name: pos_idx
dtype: int64
- name: bounding_box
sequence: int64
splits:
- name: train
num_bytes: 27536011591
num_examples: 100000
- name: test
num_bytes: 838151428
num_examples: 3000
download_size: 28302125366
dataset_size: 28374163019
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*