# SoftCFG: Uncertainty-guided Stable Guidance for Visual Autoregressive Model

Source: [SoftCFG: Uncertainty-guided Stable Guidance for Visual Autoregressive Model](<https://www.sota2.com/research/paper/68fa7bd21e209131cea8ab32>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2025-10-26T05:49:23.746Z

Authors: Dongli Xu, Aleksei Tiulpin, Matthew B. Blaschko

Published: 2025-10-01T15:04:00.000Z

## Abstract

Autoregressive (AR) models have emerged as powerful tools for image generation by modeling images as sequences of discrete tokens. While Classifier-Free Guidance (CFG) has been adopted to improve conditional generation, its application in AR models faces two key issues: guidance diminishing, where the conditional-unconditional gap quickly vanishes as decoding progresses, and over-guidance, where strong conditions distort visual coherence. To address these challenges, we propose SoftCFG, an uncertainty-guided inference method that distributes adaptive perturbations across all tokens in the sequence. The key idea behind SoftCFG is to let each generated token contribute certainty-weighted guidance, ensuring that the signal persists across steps while resolving conflicts between text guidance and visual context. To further stabilize long-sequence generation, we introduce Step Normalization, which bounds cumulative perturbations of SoftCFG. Our method is training-free, model-agnostic, and seamlessly integrates with existing AR pipelines. Experiments show that SoftCFG significantly improves image quality over standard CFG and achieves state-of-the-art FID on ImageNet 256\*256 among autoregressive models.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2510.00996>)

[PDF](<https://arxiv.org/pdf/2510.00996.pdf>)
