metadata
dataset_info:
features:
- name: Filename
dtype: string
- name: Aesthetics score
dtype: float64
- name: Artifact score
dtype: float64
- name: Misalignment score
dtype: float64
- name: Overall score
dtype: float64
- name: Artifact heatmap
sequence:
sequence:
sequence: int64
- name: Misalignment heatmap
sequence:
sequence:
sequence: int64
- name: Misalignment token label
dtype: string
- name: prompt
dtype: string
- name: image_preferred
dtype: binary
- name: image_not_preferred
dtype: binary
splits:
- name: train
num_bytes: 595869460
num_examples: 90
download_size: 30870032
dataset_size: 595869460
configs:
- config_name: default
data_files:
- split: train
path: data/train-*