AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

FastFLUX: Pruning FLUX with Block-wise Replacement and Sandwich Training

Fuhan Cai, Yong Guo, Jie Li, Wenbo Li, Jian Chen, Xiangzhong Fang

AAAI 2026 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 3 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 Sandwich Training (LoRA on neighbouring blocks) (ablation)FastFLUX-10% · HPS v2HPS v2 score: 27.99 -> 28.58Improved
add BRLL linear replacement (ablation)FastFLUX · HPS v2HPS v2 score: 28.25 -> 28.58Improved
use 2 LoRA-enhanced layers instead of 0 (ablation)Sandwich Training · HPS v2HPS v2 score: +0.60Improved
use 4 LoRA-enhanced layers instead of 2 (ablation)Sandwich Training · HPS v2HPS v2 score: +0.03No clear effect
use importance-guided order instead of Start2End (ablation)FastFLUX · HPS v2HPS v2 score: 24.63 -> 28.58Improved
use importance-guided order instead of End2Start (ablation)FastFLUX · HPS v2HPS v2 score: 28.15 -> 28.58Improved
replace L1-norm pruning with FastFLUX at similar latency (12% vs 10%)FLUX.1-dev · HPS v2HPS v2 score: 28.20 -> 28.58Improved
replace L1-norm pruning with FastFLUX at similar latency (27% vs 20%)FLUX.1-dev · HPS v2HPS v2 score: 27.17 -> 27.76Improved
prune 10% of blocks with FastFLUXFLUX.1-dev · HPS v2HPS v2 score: 28.32 -> 28.58Improved
replace L1-norm pruning with FastFLUX at similar latency (27% vs 20%)FLUX.1-dev · GenEvaloverall score: 0.47 -> 0.53Improved
prune 10% with FastFLUX vs L1 pruning 12%FLUX.1-dev · GenEvaloverall score: 0.60 -> 0.60No clear effect
increase pruning ratio from 10% to 30%FastFLUX · HPS v2HPS v2 score: 28.58 -> 26.70Hurt

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