AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Rényi Diffusion Models

Yirong Shen, Lu GAN, Cong Ling

ICML 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
replace KL-based ELBO with Renyi objective (alpha=0.5) (ablation)eps-prediction diffusion, cosine schedule · CIFAR-10NLL (bits/dim): 2.74 -> 2.48Improved
replace KL-based ELBO with Renyi objective (ablation)eps-prediction diffusion, cosine schedule · ImageNet-64NLL (bits/dim): 3.57 -> 3.29Improved
replace KL with Renyi objective without weighted loss (ablation)eps-prediction diffusion · CIFAR-10FID: 6.31 -> 6.33 (DDPM sampler)No clear effect
reduce Renyi order alpha from 1 toward 0.5 (ablation)Renyi diffusion · CIFAR-10NLL (bits/dim): 2.74 -> 2.48Improved
reduce alpha to very small values (0.0001) (ablation)Renyi diffusion · CIFAR-10FID: 6.06 -> 28.79 (NLL 2.74 -> 2.29)Hurt
add time-dependent weighted loss to KL baseline (ablation)eps-prediction baseline · CIFAR-10FID / NLL: FID 6.31 -> 2.53; NLL 2.74 -> 3.97Mixed
add weighted loss to Renyi objective (ablation)Renyi (eps) · CIFAR-10FID / NLL: FID 6.33 -> 2.51; NLL 2.48 -> 2.66Mixed
add importance sampling with Fourier features to Renyi + weighted loss (ablation)Renyi (eps) + WL · CIFAR-10FID / NLL: FID 2.51 -> 2.27; NLL 2.66 -> 2.24Improved
replace cosine with adaptive noise schedule for Renyi (eps) (ablation)Renyi (eps) · CIFAR-10NLL (bits/dim): 2.13 -> 2.10Improved
use v-prediction with Renyi objective (ablation)Renyi diffusion · CIFAR-10NLL (bits/dim): 2.10 -> 2.04Improved
train with Renyi bound instead of separate FID- and NLL-favorable regimesRDM · CIFAR-10NLL and FID: NLL 2.09, FID 2.08Improved
Renyi tighter bound shortens trainingRDM · training iterationsiterations to converge: approx 770k vs 2M (i-DODE, W-PCDM) and 10M (VDM)Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 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