metadata
dataset_info:
features:
- name: id
dtype: int64
- name: image
dtype: image
- name: conditioning_image
dtype: image
- name: text
dtype: string
- name: gpt4o_caption
dtype: string
- name: general_caption
dtype: string
- name: scene_person_descriptions
sequence: string
- name: num_people
dtype: int32
- name: annotations_json
dtype: string
splits:
- name: train
num_bytes: 34278914557
num_examples: 20000
download_size: 34094752883
dataset_size: 34278914557
configs:
- config_name: default
data_files:
- split: train
path: data/train-*