AI Paper Atlas

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

Closing the Dequantization Gap: PixelCNN as a Single-Layer Flow

Didrik Nielsen, Ole Winther

NeurIPS 2020 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Expressive invertible architectures · 2 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
Train with ELBO instead of exact likelihood (evaluated exactly) (ablation)PixelCNN++ flow · CIFAR-10bits/dim: 2.924 -> 2.944Hurt
Train with ELBO instead of exact likelihood (evaluated exactly) (ablation)PixelCNN (Quad.) flow · CIFAR-10bits/dim: 3.090 -> 3.104Hurt
Train with ELBO instead of exact likelihood (ablation)PixelCNN (bin cond.) · CIFAR-10bits/dim: 3.141 -> 3.141No clear effect
Remove bin conditioning (uniform dequantization) (ablation)PixelCNN++ flow · CIFAR-10bits/dim: 2.924 -> 3.112Hurt
Remove bin conditioning (uniform dequantization) (ablation)PixelCNN flow · CIFAR-10bits/dim: 3.141 -> 3.248Hurt
Evaluate with IWBO(1000) instead of ELBO (ablation)PixelCNN++ (bin cond., ELBO-trained) · CIFAR-10bits/dim: 2.993 -> 2.974Improved
Evaluate exact likelihood instead of ELBO (ablation)PixelCNN++ (bin cond., ELBO-trained) · CIFAR-10bits/dim: 2.993 -> 2.944Improved
Stack 2 PixelCNN layers instead of 1 (ablation)PixelCNN · CIFAR-10bits/dim: 3.14 -> 3.07Improved
Stack 4 PixelCNN layers without dropout (ablation)PixelCNN · CIFAR-10bits/dim: 3.07 -> 3.09No clear effect
Add dropout 0.2 to 4-layer stack (ablation)PixelCNN · CIFAR-10bits/dim: 3.09 -> 3.02Improved
Use quadratic splines instead of linear splines (ablation)PixelCNN · CIFAR-10bits/dim: 3.14 -> 3.09Improved
Combine PixelCNN (Quad.) with PixelCNN++ in a two-layer flow (ablation)PixelCNN++ · CIFAR-10bits/dim: 2.924 -> 2.906Improved

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