metadata
dataset_info:
features:
- name: image_path
dtype: string
- name: caption
dtype: string
- name: question
dtype: string
- name: image
dtype: image
splits:
- name: train
num_bytes: 100028841
num_examples: 999
download_size: 99402588
dataset_size: 100028841
configs:
- config_name: default
data_files:
- split: train
path: data/train-*