AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Fast Sampling Through The Reuse Of Attention Maps In Diffusion Models

Rosco Hunter, Łukasz Dudziak, Mohamed S. Abdelfattah, Abhinav Mehrotra, Sourav Bhattacharya, Hongkai Wen

IJCV 2025 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 7 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add perturbation search on top of HURRY (PHAST) (ablation)HURRY, 10-step DPM++ with 3 reuse steps · ImageNet promptsPSNR: 25.9 -> 27.2Improved
reuse attention maps instead of reducing steps to 1320-step DDIM, SD1.5 · MS-COCO 2017PSNR: 17.50 -> 25.90Improved
reuse attention maps instead of reducing steps to 13SDXL · PartiPromptsPSNR: 20.30 -> 29.92Improved
reuse attention maps instead of reducing steps to 13SD1.5 DDIM · MS-COCO 2017FID: 28.22 -> 29.18No clear effect
reuse attention maps instead of reducing steps to 13SD1.5 DDIM · MS-COCO 2014FID: 18.71 -> 17.62Improved
reuse attention maps instead of reducing steps to 13CFG-distilled SD1.5 DDIM · MS-COCO 2017PSNR: 17.68 -> 28.63Improved
store cached attention maps in FP16 (ablation)PHAST, SD1.5 DDIM · MS-COCO 2017PSNR: 25.90 -> 25.95No clear effect
store cached attention maps in INT8 (ablation)PHAST, CFG-distilled DPM++ · MS-COCO 2017PSNR: 25.34 -> 25.35No clear effect
reuse features (attention block outputs) instead of attention maps (ablation)PHAST, DDIM, 10 reuse steps · ImageNet promptsPSNR: 27.10 -> 25.05Hurt
reuse features instead of attention maps (ablation)HURRY, DPM++, 10 reuse steps · ImageNet promptsPSNR: 22.49 -> 20.21Hurt
use DPM++ instead of DDIM samplerPHAST, SD1.5 · MS-COCO 2017PSNR: 25.90 -> 23.02Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy