metadata
dataset_info:
features:
- name: original_prompt
dtype: string
- name: original_image
dtype: image
- name: edit_prompt
dtype: string
- name: edited_prompt
dtype: string
- name: edited_image
dtype: image
- name: adversarial_image
dtype: image
splits:
- name: train
num_bytes: 2287964054
num_examples: 5000
download_size: 2287986553
dataset_size: 2287964054
configs:
- config_name: default
data_files:
- split: train
path: data/train-*