AI Paper Atlas

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

Transporting Quantiles to Codebook: Scalable Vector Quantization without Codebook Collapse

Yuwei Zeng, Zekun Shi

NeurIPS 2026 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Structured latent VAEs · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes a scalable vector quantization method that formulates neural quantization as joint modeling and clustering to avoid codebook collapse in VQ-VAEs.

Problem

VQ-VAE discrete representation learning suffers from codebook collapse and inefficient code utilization, which hurts autoregressive generative models that rely on it.

Approach

Reformulates neural quantization as a joint modeling and clustering process (transporting quantiles to the codebook) as a simple alternative to standard vector quantization.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace learned/nearest-neighbor VQ codebook with flow-transported quantile codebookVQ-GAN tokenizer · CIFAR-10 reconstruction (d=14, |C|=16384)codebook utilization (%): 2.8 -> 100.0Improved
replace VQ-GAN quantizer with FlowVQVQ-GAN tokenizer · CIFAR-10 reconstruction (d=14, |C|=16384)rFID: 1.65 -> 1.40Improved
increase codebook size from 1024 to 10240 with fixed latent dim in standard VQVQ-GAN · CIFAR-10 reconstructioncodebook utilization (%): 45.6 -> 2.7Hurt
add auxiliary entropy loss (ablation)LFQ · CIFAR-10 reconstruction (d=14, |C|=16384)codebook utilization (%): 0.8 -> 100.0Improved
add auxiliary entropy loss (ablation)FlowVQ · CIFAR-10 reconstruction (d=14, |C|=16384)rFID: 1.40 -> 1.39No clear effect
add self-correction toward Voronoi centroids (ablation)FlowVQ transport codebook · CIFAR-10FID: 1.49 -> 1.52Hurt
add self-correction toward Voronoi centroids (ablation)FlowVQ transport codebook · MP-20match rate (%): 88.15 -> 90.81Improved
add online correction on top of self-correction (ablation)FlowVQ · CIFAR-10FID: 1.52 -> 1.40Improved
add online correction on top of self-correction (ablation)FlowVQ · MP-20match rate (%): 90.81 -> 90.66No clear effect
strengthen entropy regularization (epsilon 10 to 1) (ablation)k-means-initialized OptVQ · MP-20match rate (%): 17.85 -> 53.27Mixed
replace k-means-initialized VQ-VAE with FlowVQcrystal tokenizer · MP-20match rate (%): 78.38 -> 90.81Improved
replace k-means-initialized VQ-VAE with FlowVQrobot action tokenizer · DROID action chunksMSE (e-4): 1.88 -> 1.71Improved

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.

Experimental setup

Builds on: VQ-VAE

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