atlas

Research map / Image generation / Fast diffusion sampling

Soft-IntroVAE: Analyzing and Improving the Introspective Variational Autoencoder

CVPR 2021 · Image generation › Fast diffusion sampling · approach: Distributional control of diffusion · 45 citations (OpenAlex)

Check your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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 IntroVAE's hard-margin hinge loss with a soft exponential ELBO penaltyIntroVAE · 8 GaussiansKL: 2.53$\pm$1.07 -> 1.25$\pm$0.11Helped
Replace IntroVAE's hard-margin hinge loss with a soft exponential ELBO penaltyIntroVAE · SpiralKL: 8.38$\pm$0.45 -> 8.13$\pm$0.3Helped
Replace IntroVAE's hard-margin hinge loss with a soft exponential ELBO penaltyIntroVAE · CheckerboardKL: 19.03$\pm$0.34 -> 20.27$\pm$0.21Hurt
Replace IntroVAE's hard-margin hinge loss with a soft exponential ELBO penaltyIntroVAE · RingsKL: 10.21$\pm$0.49 -> 9.18$\pm$0.33Helped
Replace IntroVAE's hard-margin hinge loss with a soft exponential ELBO penaltyIntroVAE · CheckerboardgnELBO: 8.54$\pm$0.11 -> 4.47$\pm$0.29Helped
Use class conditioningSoft-IntroVAE · CIFAR-10FID: 4.6 -> 4.07Helped
Use unsupervised Soft-IntroVAE content transfer instead of supervised LORDCars3D content transfer · Cars3DLPIPS: 0.078 -> 0.084Hurt

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