AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

When Scores Learn Geometry: Rate Separations under the Manifold Hypothesis

Xiang Li, Zebang Shen, Ya-Ping Hsieh, Niao He

ICLR 2026 · Image generation › Fast diffusion sampling · approach: Diffusion theory and design space

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
scale unconditional score in corrector step by sigma^alpha (Tempered Score)PC sampler, Stable Diffusion 1.5 · Furniture promptP-sim: 29.40 -> 30.20Improved
scale unconditional score in corrector step by sigma^alpha (Tempered Score)PC sampler, Stable Diffusion 1.5 · Car promptI-sim (diversity): 87.20 -> 87.14Improved
Tempered Score vs DDPMStable Diffusion 1.5 · Architecture promptP-sim: 27.36 -> 27.32No clear effect
Tempered Score with alpha=1 untuned across corrector steps (ablation)PC sampler · Furniture/Car/Architecture, 5-30 stepsP-sim and I-sim: e.g. 81.24 -> 80.95 I-sim at 10 stepsImproved
Tempered Score Langevin vs standard Langevinscore model on ellipse · Ellipse syntheticdistribution recoveryImproved

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