metadata
dataset_info:
features:
- name: task
dtype: string
- name: image_path
dtype: string
- name: input_image
dtype: image
- name: edited_image
dtype: image
- name: instruction
dtype: string
- name: caption
sequence: string
- name: mask
dtype: 'null'
splits:
- name: train
num_bytes: 77843676
num_examples: 89
download_size: 77848979
dataset_size: 77843676
configs:
- config_name: default
data_files:
- split: train
path: data/train-*