metadata
dataset_info:
features:
- name: scene_name
dtype: string
- name: ground_truth
dtype: image
- name: caption
dtype: string
- name: conditioning_images_one
dtype: image
- name: conditioning_images_two
dtype: image
- name: reference_image
dtype: string
- name: prescan_images
dtype: image
splits:
- name: train
num_bytes: 20869224
num_examples: 14
download_size: 20831985
dataset_size: 20869224
configs:
- config_name: default
data_files:
- split: train
path: data/train-*