metadata
dataset_info:
features:
- name: image
dtype: image
- name: question
dtype: string
- name: answer
dtype: string
- name: ground_truth
dtype: string
splits:
- name: train
num_bytes: 1224460931.6
num_examples: 3200
- name: test
num_bytes: 153123921
num_examples: 400
download_size: 1324055034
dataset_size: 1377584852.6
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*