metadata
dataset_info:
features:
- name: image
dtype:
image:
decode: false
- name: qa
list:
- name: question
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 4425372570
num_examples: 200000
download_size: 4272488137
dataset_size: 4425372570
configs:
- config_name: default
data_files:
- split: train
path: data/train-*