metadata
dataset_info:
features:
- name: task
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
- name: image
dtype: image
splits:
- name: train
num_bytes: 932863.75
num_examples: 4482
download_size: 9450152
dataset_size: 932863.75
configs:
- config_name: default
data_files:
- split: train
path: data/train-*