metadata
dataset_info:
features:
- name: caption
dtype: string
splits:
- name: train
num_bytes: 4858831
num_examples: 21805
download_size: 1021737
dataset_size: 4858831
configs:
- config_name: default
data_files:
- split: train
path: data/train-*