metadata
dataset_info:
features:
- name: image
dtype: image
- name: canny_edges
dtype: image
- name: caption
dtype: string
- name: captions
sequence: string
splits:
- name: train
num_bytes: 2656560746.5
num_examples: 20638
download_size: 2648738157
dataset_size: 2656560746.5
configs:
- config_name: default
data_files:
- split: train
path: data/train-*