AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling

Meimingwei Li, Stefan Andreas Baumann, Felix Krause, Bjorn Ommer

ECCV 2026 · Image generation › Autoregressive and latent generation · approach: Next-scale and token autoregressive images · 0 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
add autoregressive intra-scale sampling (causal refiner) to VARVAR-d16 · ImageNet 256x256FID: 3.30 -> 2.81Improved
use bidirectional parallel refiner with independent sampling instead of AR (ablation)VAR-d16 + refiner · ImageNet 256x256FID: 3.30 -> 3.15No clear effect
continue training the base model for 30 more epochs (ablation)VAR-d16 · ImageNet 256x256FID: 3.30 -> 3.12No clear effect
increase refiner depth beyond 2 blocks (ablation)Logit Refiner · ImageNet 256x256FID: 2.81 -> 2.82 (d_r=4); 2.81 (d_r=8)No clear effect
use depth-0 refiner (AR input projection and head only) (ablation)Logit Refiner · ImageNet 256x256FID: 3.30 -> 3.02Improved
also train the base backbone jointly with the refiner (ablation)Logit Refiner · ImageNet 256x256FID: 2.81 -> 2.72Improved
apply refiner at scale 2^2 (ablation)Logit Refiner · ImageNet 256x256FID: 2.98 -> 2.81Improved
apply refiner at scale 8^2 (ablation)Logit Refiner · ImageNet 256x256FID: 2.81 -> 2.81No clear effect
change intra-scale token ordering (ablation)Logit Refiner · ImageNet 256x256FID: 2.81 vs 2.83-2.86No clear effect
add refiner for VAR-d24 versus larger VAR-d30VAR-d24 + refiner · ImageNet 256x256FID: 1.92 -> 1.83Improved
add refinerVAR-d24 · ImageNet 256x256Inception Score: 312.9 -> 288.2Hurt
add refiner without CFG on small modelVAR-d16 · ImageNet 256x256 (no CFG)FID: 3.44 -> 3.41No clear effect

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