Datasets:

Formats:
parquet
ArXiv:
Libraries:
Datasets
Dask
License:
Dataset Viewer
Auto-converted to Parquet
are_different
bool
2 classes
best_image_uid
stringlengths
4
36
caption
stringlengths
0
1.6k
created_at
timestamp[ns]date
2023-02-18 21:11:11
2023-06-24 23:18:02
has_label
bool
2 classes
image_0_uid
stringlengths
36
36
image_0_url
stringlengths
114
114
image_1_uid
stringlengths
36
36
image_1_url
stringlengths
114
114
jpg_0
unknown
jpg_1
unknown
label_0
float64
0
1
label_1
float64
0
1
model_0
stringclasses
14 values
model_1
stringclasses
13 values
ranking_id
int64
40.2k
1.32M
user_id
int64
1
9.31k
num_example_per_prompt
int64
1
3.18k
__index_level_0__
int64
415
1M
true
c13f7ea3-6e2d-4c27-adb6-0ed2a7e9162c
crazy frog, on one wheel, motorcycle, dead
2023-03-30T20:35:41.587000
true
c13f7ea3-6e2d-4c27-adb6-0ed2a7e9162c
https://text-to-image-hu…0ed2a7e9162c.png
21ebf325-e6e1-415a-9ed5-81cba84223b1
https://text-to-image-hu…81cba84223b1.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
1
0
yuvalkirstain/dreamlike-photoreal-2-flax
stable-diffusion-xl-beta-v2-2-2
717,275
6,643
92
415
true
e304e68b-2ce5-4671-89b7-4f232b40802b
"a cat sitting on top of a helloween pumpkin, cementry in background, dramatic purple lighting, circ(...TRUNCATED)
2023-03-22T18:52:29.394000
true
5ff8ac58-aa89-4a49-901f-41e8bc039a3a
https://text-to-image-hu…41e8bc039a3a.png
e304e68b-2ce5-4671-89b7-4f232b40802b
https://text-to-image-hu…4f232b40802b.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
0
1
stable-diffusion-xl-v2-2
stable-diffusion-xl-beta-v2-2-2
427,867
5,080
65
490
true
11254699-b4ca-4417-936f-f558dc80a119
man holding two urumi, one on each head
2023-03-21T01:02:43.866000
true
11254699-b4ca-4417-936f-f558dc80a119
https://text-to-image-hu…f558dc80a119.png
c2207166-a7ef-45b7-bab2-bc7b85dbfedc
https://text-to-image-hu…bc7b85dbfedc.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
1
0
yuvalkirstain/dreamlike-photoreal-2-flax
stable-diffusion-xl-beta-v2-2-2
257,929
3,602
12
616
false
none
"Draw an undirected graph with 11 nodes and 10 edges, where each edge connects two nodes. The edge c(...TRUNCATED)
2023-03-25T13:45:08.343000
false
e8279446-43c7-4c2a-a7f3-36249bc5dedc
https://text-to-image-hu…36249bc5dedc.png
2d676924-293a-4182-9f12-84fae81a7cef
https://text-to-image-hu…84fae81a7cef.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
0.5
0.5
stable-diffusion-xl-v2-2
stable-diffusion-xl-v2-2
577,813
5,921
9
669
true
73958b44-e34f-46a5-ac0f-fefaaaf4c768
photo an attractive woman gunslinger standing in a scifi-fantasy wilderness
2023-03-20T08:34:36.768000
true
9016da1d-c3df-4a5b-8284-271637b174de
https://text-to-image-hu…271637b174de.png
73958b44-e34f-46a5-ac0f-fefaaaf4c768
https://text-to-image-hu…fefaaaf4c768.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
0
1
stable-diffusion-xl-beta-v2-2-2
yuvalkirstain/dreamlike-photoreal-2-flax
244,154
3,520
202
972
true
5bc0d56a-f7cd-4808-b60e-4f9dc3bc6d2a
london skyline 1908
2023-03-31T15:53:13.140000
true
5bc0d56a-f7cd-4808-b60e-4f9dc3bc6d2a
https://text-to-image-hu…4f9dc3bc6d2a.png
08f86030-1600-4b69-8983-abcbf4582b85
https://text-to-image-hu…abcbf4582b85.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
1
0
stable-diffusion-xl-beta-v2-2-2
stable-diffusion-xl-beta-v2-2-3
731,534
6,725
23
1,312
true
d5f1b960-38e3-433a-97e8-a4aa100861fd
tiling flower
2023-03-25T20:50:40.298000
true
d5f1b960-38e3-433a-97e8-a4aa100861fd
https://text-to-image-hu…a4aa100861fd.png
d0970c84-d6e4-4c8c-9679-e463837e3e7e
https://text-to-image-hu…e463837e3e7e.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
1
0
stable-diffusion-xl-v2-2
stable-diffusion-xl-beta-v2-2-2
593,096
5,988
17
1,359
false
ed3fd208-44d7-4a0f-8d4f-c2d8598cd15b
"self-replicating space squid perched on the edge of greatness looks over it's vast kingdom and wond(...TRUNCATED)
2023-03-22T22:23:13.415000
true
ed3fd208-44d7-4a0f-8d4f-c2d8598cd15b
https://text-to-image-hu…c2d8598cd15b.png
0de26388-277f-43ea-820e-59c8ec98fd1e
https://text-to-image-hu…59c8ec98fd1e.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
1
0
stable-diffusion-xl-v2-2
stable-diffusion-xl-v2-2
439,285
4,464
14
2,045
true
03609e67-7dc5-49a4-a9f9-96158daa4a09
"close-up 3d render of hyper real iridescent translucent gelatinous wrinkled cloth, satin, silk, blu(...TRUNCATED)
2023-05-20T06:08:26.184000
true
03609e67-7dc5-49a4-a9f9-96158daa4a09
https://text-to-image-hu…96158daa4a09.png
02e8c8f3-2ef6-4237-9226-f1eeb698882f
https://text-to-image-hu…f1eeb698882f.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
1
0
yuvalkirstain/dreamlike-photoreal-2-flax
stable-diffusion-xl-beta-v2-2-2
1,138,668
7,659
15
2,328
true
none
ink simplistic drawing of a skull in a crown
2023-06-17T17:35:38.273000
false
00173272-403b-4bf7-b5d4-9f03aadfde3a
https://text-to-image-hu…9f03aadfde3a.png
993b3ae9-71da-4dae-9fcf-dbc9940f7ab0
https://text-to-image-hu…dbc9940f7ab0.png
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
"/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAIBAQEBAQIBAQECAgICAgQDAgICAgUEBAMEBgUGBgYFBgYGBwkIBgcJBwYGCAsICQo(...TRUNCATED)
0.5
0.5
stable-diffusion-xl-beta-v2-2-5-b
stable-diffusion-xl-beta-v2-2-2
1,294,847
6,688
8
2,687
End of preview. Expand in Data Studio

Pick-a-Pic v2 · FiFA Filtered Subsets

These subsets were produced by filtering the original Pick-a-Pic v2 dataset using FiFA, a data filtering algorithm proposed in the paper Automated Filtering of Human Feedback Data for Aligning Text-to-Image Diffusion Models.

Overview

The filtering process is based on three key metrics:

  1. Preference Margin: Estimated using PickScore
  2. Text Quality: Estimated through LLM scoring
  3. Text Diversity: Estimated using K-NN distance

Dataset Configurations

This dataset provides several configurations, each corresponding to a different number of filtered triplets selected by the automated FiFA algorithm. You can choose from FiFA-500, FiFA-1k, FiFA-5k, FiFA-10k, FiFA-20k, FiFA-50k, and FiFA-100k, depending on your needs. The only difference between these configurations is the number of examples included; all are filtered using the same FiFA method. We recommend using the FiFA-5k configuration as the default, as it generally works best for Stable Diffusion training.

Quick Start

from datasets import load_dataset

# Load a specific configuration
dataset = load_dataset("Dragonjinny/FiFA-pickapic-v2", "FiFA-5k", split="train")

# Access the data
import io
from PIL import Image

for example in dataset:
    caption = example["caption"]  # The prompt text
    jpg_0 = example["jpg_0"]      # First image (bytes)
    jpg_1 = example["jpg_1"]      # Second image (bytes)
    label_0 = example["label_0"]  # Binary label (0 or 1) indicating which image is preferred
    
    # Convert bytes to PIL Images
    image1 = Image.open(io.BytesIO(jpg_0)).convert("RGB")
    image2 = Image.open(io.BytesIO(jpg_1)).convert("RGB")
    
    # Now you can work with the images
    print(f"Caption: {caption}")
    print(f"Preferred image: {'jpg_0' if label_0 == 1 else 'jpg_1'}")
    # image1.show()  # Display the first image
    # image2.show()  # Display the second image

Data Format

Each example contains:

  • caption: The prompt text
  • jpg_0: First image (bytes)
  • jpg_1: Second image (bytes)
  • label_0: Binary label (0 or 1) indicating which image is preferred
  • __index_level_0__ : Unique ID for each data point

Citation

If you use this dataset in your research, please cite our paper:

@inproceedings{
yang2025automated,
title={Automated Filtering of Human Feedback Data for Aligning Text-to-Image Diffusion Models},
author={Yongjin Yang and Sihyeon Kim and Hojung Jung and Sangmin Bae and SangMook Kim and Se-Young Yun and Kimin Lee},
booktitle={The Thirteenth International Conference on Learning Representations},
year={2025},
url={https://openreview.net/forum?id=8jvVNPHtVJ}
}

License

This dataset is licensed under MIT License, following the license of the original Pick-a-Pic v2 dataset.

Downloads last month
320

Data Sourcing report

powered
by Spawning.ai

No elements in this dataset have been identified as either opted-out, or opted-in, by their creator.