metadata
dataset_info:
features:
- name: caption
dtype: string
- name: image
dtype: image
splits:
- name: train
num_bytes: 19469971483.616
num_examples: 19928
download_size: 19620338126
dataset_size: 19469971483.616
configs:
- config_name: default
data_files:
- split: train
path: data/train-*