metadata
dataset_info:
features:
- name: image
dtype: image
- name: filename
dtype: string
- name: query
list: string
- name: message
list: string
- name: caption
struct:
- name: anti_aesthetic_caption
dtype: string
- name: anti_aesthetic_elements
list: string
- name: clean_caption
dtype: string
- name: thinking
dtype: string
- name: clean_image
dtype: image
- name: image_reward
list: float64
- name: human_score
dtype: float64
splits:
- name: train
num_bytes: 637297012.856
num_examples: 6272
download_size: 574271486
dataset_size: 637297012.856
configs:
- config_name: default
data_files:
- split: train
path: data/train-*