AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Wavelet Flow: Fast Training of High Resolution Normalizing Flows

Jason J. Yu, Konstantinos G. Derpanis, Marcus A Brubaker

NeurIPS 2020 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Normalizing flow training and design · 23 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
wavelet-based multi-scale conditional flow instead of GlowWavelet Flow · ImageNet 64x64bits per dimension: 3.81 -> 3.78Improved
wavelet-based multi-scale conditional flow instead of GlowWavelet Flow · LSUN bedroom 64x64bits per dimension: 2.38 -> 2.41Hurt
wavelet-based flow instead of Glow (training cost)Wavelet Flow · CelebA-HQ 256x256seconds per image: 1.79 -> 0.0147Improved
wavelet-based flow instead of Glow (total training)Wavelet Flow · LSUN bedroom 64x64training GPU hours: 3,700 -> 572Improved
wavelet-based flow instead of Glow (sample quality)Wavelet Flow · LSUN bedroom 64x64FID: 60.03 -> 121.20Hurt
use affine instead of additive coupling (ablation)Wavelet Flow · CelebA-HQ 256x256 5-bitbits per dimension: 1.12 -> 0.94Improved
use affine instead of additive coupling for annealed samples (ablation)Wavelet Flow · CelebA-HQimage qualityNo clear effect
truncate 64x64 model to evaluate at 32x32 (ablation)Wavelet Flow · ImageNet 32x32bits per dimension: 4.08 -> 4.16Hurt
use 5-bit instead of 8-bit data to improve samplesWavelet Flow · CelebA-HQsample qualityNo clear effect
add multi-scale MCMC annealing at T=0.97 (ablation)Wavelet Flow · CelebA-HQsample qualityImproved

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