AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

RNE: plug-and-play diffusion inference-time control and energy-based training

Jiajun He, José Miguel Hernández-Lobato, Yuanqi Du, Francisco Vargas

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Training-free diffusion acceleration · 18 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
replace summed scores with RNC sampling and target processSum score · Multi-target SBDDbetter than known: 0.345 -> 0.649Improved
use RNC with c_b=0.2 instead of FKCFKC · Multi-target SBDDbetter than known: 0.608 -> 0.649Improved
use RNC with c_a=1, c_b=0 instead of FKCFKC · Multi-target SBDDbetter than known: 0.608 -> 0.589No clear effect
replace summed scores with FKC/RNC SMCSum score · Multi-target SBDDdiversity: 0.881 -> 0.814Hurt
add RNC SMC to reward-gradient guidanceguidance without SMC · Maze stitching task 4success rate: 0.288 -> 1.000Improved
add RNE regularisation to thermodynamic integrationTI without RNE · ALDP solvation free energyfree energy estimate: 27.30 -> 29.28Improved
add RNE regularisation to denoising score matchingDSM · 2D GMM densitylearned densityImproved
add RNE regularisation or dual score matching to DSMDSM · 100D GMMlearned energy accuracyImproved
choose c_a+c_b near 1 (within 1 +/- 0.2) in RNC (ablation)RNC · ALDP annealingimportance weight variance: c_a+c_b within 1 +/- 0.2Improved
use general (c_a,c_b) instead of c_a=1, c_b=0 (ablation)RNC · ALDP annealingW2 and TVDImproved
increase batch size from 100 to 5000 particles with RNC instead of FKC (ablation)FKC · ALDP annealingW2 distanceImproved
add RNE regularisation to DSMDSM · ALDP diffusion qualitysample qualityNo clear effect

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