AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Overshoot and Shrinkage in Classifier-Free Guidance: From Theory to Practice

Krunoslav Lehman Pavasovic, Jakob Verbeek, Giulio Biroli, Marc Mezard

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Guidance for conditional diffusion

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 power-law nonlinearity (alpha=0.9) to standard CFGstandard CFG · EDM2-S (class-cond, ImageNet-1K 512)FID: 2.29 -> 1.93Improved
add power-law nonlinearity to standard CFGstandard CFG · DiT/XL-2 (class-cond, ImageNet-1K 256)FID: 2.27 -> 2.05Improved
add power-law nonlinearity to standard CFGstandard CFG · Diff. MMDiT (T2IM, CC12M)FID: 8.58 -> 8.11Improved
add power-law nonlinearity to standard CFGstandard CFG · FM MMDiT (T2IM, COCO)FID: 5.20 -> 4.81Improved
combine power-law with CADSCADS · EDM2-S (class-cond, ImageNet-1K 512)FID: 1.60 -> 1.52Improved
combine power-law with limited-interval guidanceLimited interval · EDM2-S (class-cond, ImageNet-1K 512)FID: 1.87 -> 1.73Improved
combine power-law with limited-interval guidanceLimited interval · EDM2-S (class-cond, ImageNet-1K 512)precision: 0.760 -> 0.752Hurt
add power-law nonlinearity to standard CFGstandard CFG · FM MMDiT (T2IM, COCO)precision: 0.629 -> 0.621Hurt
add power-law nonlinearity to standard CFGstandard CFG · Diff. MMDiT (T2IM, CC12M)recall: 0.569 -> 0.553Hurt
increase alpha of power-law CFG (ablation)power-law CFG · EDM2-S ImageNet-1K 512FIDImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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