metadata
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
dataset_info:
features:
- name: scene_id
dtype: string
- name: before
dtype: binary
- name: after
dtype: binary
- name: water_mask
dtype: binary
- name: prompt
dtype: string
splits:
- name: train
num_bytes: 251969930336
num_examples: 64000
download_size: 172674106660
dataset_size: 251969930336