analiza aiqbench
The Logit Refiner addresses a fundamental bottleneck in Visual Autoregressive Models by correcting the mean-field approximation inherent in parallel decoding, which often leads to spatially incoherent image generation. This lightweight module restores intra-scale dependencies through sequential sampling without requiring retraining or significant additional parameters, allowing smaller models to outperform larger baselines. The method demonstrates that improving generation quality relies on fixing the decoding rule rather than simply scaling model capacity.
Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling
arXiv:2609.11804v1 Announce Type: cross Abstract: Visual Autoregressive Models (VAR) generate images through next-scale prediction, producing all tokens within each scale in parallel. We show that this parallel decoding constitutes a mean-field-style approximation that discards spatial dependencies among same-scale tokens, causing locally incoherent samples regardless of backbone capacity -- a limitation of the decoding rule. Addressing this limitation, we introduce the Logit Refiner, a lightweight autoregressive module that restores intra-scale dependencies by sequentially sampling tokens conditioned on frozen backbone features. Adding only ~10% parameters and less than 5% of the base model's training compute, it plugs into any pretrained VAR checkpoint without retraining. Controlled ablations isolate joint intra-scale sampling -- rather than additional capacity or training -- as the critical ingredient. Across backbones from 310M to 2B parameters on class-conditional ImageNet 256x256, the refiner consistently improves generation quality, enabling a 1.1B-parameter model to surpass one twice its size. The approach further generalizes to text-to-image generation, confirming that the mean-field bottleneck persists across VAR variants and is effectively alleviated by our method. Project page: https://compvis.github.io/logit-refiner/
Abstract:Visual Autoregressive Models (VAR) generate images through next-scale prediction, producing all tokens within each scale in parallel. We show that this parallel decoding constitutes a mean-field-style approximation that discards spatial dependencies among same-scale tokens, causing locally incoherent samples regardless of backbone capacity -- a limitation of the decoding rule. Addressing this limitation, we introduce the Logit Refiner, a lightweight autoregressive module that restores intra-scale dependencies by sequentially sampling tokens conditioned on frozen backbone features. Adding only ~10% parameters and less than 5% of the base model's training compute, it plugs into any pretrained VAR checkpoint without retraining. Controlled ablations isolate joint intra-scale sampling -- rather than additional capacity or training -- as the critical ingredient. Across backbones from 310M to 2B parameters on class-conditional ImageNet 256x256, the refiner consistently improves generation quality, enabling a 1.1B-parameter model to surpass one twice its size. The approach further generalizes to text-to-image generation, confirming that the mean-field bottleneck persists across VAR variants and is effectively alleviated by our method. Project page: this https URL
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
arXiv:2609.11804 [cs.CV]
(or arXiv:2609.11804v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2609.11804 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 273 cuvinte
Subiecte conexe
Edit the Bits, Diff the Codes: Bitwise Residual Editing for Visual Autoregressive Models
arXiv · 12 iun. 2026
Overconfidence and Calibration in Medical VQA: Empirical Findings and Hallucination-Aware Mitigation
arXiv · 1 oct. 2026
Evidence Recomposition and Predictive Context Residualization for Visual Attribution in Multimodal Large Language Models
arXiv · 18 aug. 2026
AsyncLane: Decoupling Refinement from Advancement in Diffusion Language Model Decoding
arXiv · 9 iun. 2026