metadata
dataset_info:
features:
- name: multiple_choice_answer
dtype: string
- name: question
dtype: string
- name: image
dtype: image
splits:
- name: train
num_bytes: 16427957.954119343
num_examples: 100
download_size: 16332826
dataset_size: 16427957.954119343
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
A small VQAv2 Dataset For Testing Script or Notebook on Google Collab. Use split='train'.