Text-to-Image Generation on DrawBench

Metric: Human Preference Alignement (HPSv2) (lower is better)

LeaderboardDataset

Loading chart...

Results

Submit a result

Sort:

#	Model↕	Human Preference Alignement (HPSv2)▲	Extra Data	Paper	Date↕	Code
1	LCM	0.261	No	Latent Consistency Models: Synthesizing High-Res...	2023-10-06	Code
2	Stable Diffusion 1.5	0.2646	No	High-Resolution Image Synthesis with Latent Diff...	2021-12-20	Code
3	Stable Diffusion 1.5 (DPO)	0.2672	No	Diffusion Model Alignment Using Direct Preferenc...	2023-11-21	Code
4	Stable Diffusion 1.5 (DDPO)	0.2673	No	Training Diffusion Models with Reinforcement Lea...	2023-05-22	Code
5	Stable Diffusion 1.5 (Curriculum DPO)	0.2681	No	Curriculum Direct Preference Optimization for Di...	2024-05-22	Code
6	LCM (DDPO)	0.2803	No	Training Diffusion Models with Reinforcement Lea...	2023-05-22	Code
7	LCM (DPO)	0.2814	No	Diffusion Model Alignment Using Direct Preferenc...	2023-11-21	Code
8	LCM (Curriculum DPO)	0.2851	No	Curriculum Direct Preference Optimization for Di...	2024-05-22	Code

#1LCMSOTA
0.261
Human Preference Alignement (HPSv2)· 2023-10-06
Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference Code
#2Stable Diffusion 1.5SOTA
0.2646
Human Preference Alignement (HPSv2)· 2021-12-20
High-Resolution Image Synthesis with Latent Diffusion Models Code
#3Stable Diffusion 1.5 (DPO)
0.2672
Human Preference Alignement (HPSv2)· 2023-11-21
Diffusion Model Alignment Using Direct Preference Optimization Code
#4Stable Diffusion 1.5 (DDPO)
0.2673
Human Preference Alignement (HPSv2)· 2023-05-22
Training Diffusion Models with Reinforcement Learning Code
#5Stable Diffusion 1.5 (Curriculum DPO)
0.2681
Human Preference Alignement (HPSv2)· 2024-05-22
Curriculum Direct Preference Optimization for Diffusion and Consistency Models Code
#6LCM (DDPO)
0.2803
Human Preference Alignement (HPSv2)· 2023-05-22
Training Diffusion Models with Reinforcement Learning Code
#7LCM (DPO)
0.2814
Human Preference Alignement (HPSv2)· 2023-11-21
Diffusion Model Alignment Using Direct Preference Optimization Code
#8LCM (Curriculum DPO)
0.2851
Human Preference Alignement (HPSv2)· 2024-05-22
Curriculum Direct Preference Optimization for Diffusion and Consistency Models Code