Visual Autoregressive Models

Follow this topic Ask about this topic

Visual autoregressive models represent images or videos as discrete tokens and predict them step by step, often moving from coarse structure to finer detail. Common approaches use hierarchical tokenizers and transformer decoders, with methods such as next-scale prediction, flexible token ordering, and parallel or speculative decoding. They are used for image and video generation as well as editing, super-resolution, reconstruction, and visual classification, with efficiency improvements aimed at reducing decoding time and memory use. A central challenge is balancing speed and controllability with global coherence and fine detail, since errors in early predictions can propagate through later stages.

Latest papers

56 papers2 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

HGA-Quant: Heterogeneity-Aware Quantization for AutoRegressive Visual Generation Models
Oct 11, 2026Xinmeng Liu, Xianglong Yan, Kaicheng Yang +2Post-Training Quantization Neural Network Quantization
Shanghai Jiao Tong University · Xi’an University of Electronic Science and Technology
Pretraining and Test-Time Training Budgets Matter for ARC: A Controlled Study of VARC and TRM
Oct 9, 2026Adrien Deliege, Marc Van DroogenbroeckRecursive Reasoning Models Abstract Reasoning
Montefiore Institute of Electrical Engineering and Computer Science, University of Liège, Liège, Belgium
Edit-VAR: Taming Visual Autoregressive Model for Precise Video Editing
Sep 18, 2026Chongbo Zhao, Jiangming Wang, Xilai Wang +5Video Editing Instruction-Based Video Editing
Sun Yat-sen University · South China University of Technology · Tsinghua University +3
Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling
Sep 10, 2026Meimingwei Li, Stefan Andreas Baumann, Felix Krause +1Visual Autoregressive Models T2I Generation
Munich Center for Machine Learning (MCML)
Attribute Token Arithmetic: Disentangled and Continuous Semantic Control for Visual Autoregressive Models
Aug 28, 2026Xindi Yang, Yicheng Wu, Cheng Zhang +2Controllable Image Generation Visual Autoregressive Models
Faculty of IT, Monash University
VISTA: Test-Time Compositional Alignment for Visual Autoregressive Generation
Aug 23, 2026Hossein Shahabadi, Niki Sepasian, Mahdieh Soleymani BaghshahVisual Autoregressive Models Compositional Image Generation
Department of Computer Engineering, Sharif University of Technology
Accelerating Visual On-Policy Distillation with Batched Speculative Jacobi Rollouts
Aug 18, 2026Bingqi Shan, Zhehao Yu, Kenhong Lin +1Speculative Decoding Visual Autoregressive Models
Shenzhen Key Laboratory of Internet Information Collaboration Harbin Institute of Technology, Shenzhen Shenzhen, China
Beyond Text Conditioning: A Systematic Study of MLLM-DiT Fusion for Video Generation
Aug 14, 2026Yanbo Ding, Yijia Fan, Caihua Shan +8Multimodal Video Generation T2V Generation
Chinese Academy of Sciences · Microsoft Research · Sun Yat-sen University +3
Model the Edit, Not the Image: Visual Autoregressive Editing from a Source-Centric Perspective
Aug 10, 2026Hongyi Fang, Chuwen Xie, Benjia Zhou +6Text-Guided Image Editing Visual Autoregressive Models
Beijing Institute of Technology, Zhuhai · Zhejiang University · Tencent +1
SynVAR: Synergizing Spatial and Semantic Alignment in Visual Autoregressive Model
Aug 8, 2026Zhennan Chen, Tianxing Shi, Pengcheng Xu +5Visual Autoregressive Models Autoregressive Image Generation
State Key Laboratory for Novel Software Technology, Nanjing University
VARPose: Flexible 2D Pose Densification via Visual Autoregressive Modeling for Enhanced 3D Lifting
Aug 3, 2026Kaiyuan Pu, Tiantian Yang, Dan ZengVisual Autoregressive Models 3D Human Pose Estimation
School of Artificial Intelligence, Sun Yat-sen University Zhuhai, China · The Technology Innovation Center for Collaborative Applications of Natural Resources Data in GBA, MNR Guangzhou, China
Detail Continuation over a Trustworthy Coarse Scale for Autoregressive Super-Resolution
Aug 3, 2026Hongyi Fang, Jiahui Wu, Yichen Yue +2Visual Autoregressive Models Autoregressive Image Super-Resolution
Sun Yat-Sen University Zhuhai, China · Beijing Institute of Technology Zhuhai, China · Technology Innovation Center for Collaborative Applications of Natural Resources Data in GBA, Ministry of Natural Resources Guangzhou, China
Next-Frequency Autoregressive Modeling for Remote Sensing Change Detection
Jul 28, 2026Luqi Gong, Yue Chen, Qixin Xie +3Visual Autoregressive Models Frequency-Domain Feature Learning
Research Center for Space Computing System, Zhejiang Lab, Hangzhou 311121, China · State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications, Beijing 100876, China · College of Economics and Management, South China Agricultural University, Guangzhou 510642, China
UniGen-AR: Unifying Visual Generation with Auto-Regressive Modeling
Jul 27, 2026Zhipeng Bao, Zhen Zhu, Nupur Kumari +4Unified Multimodal Models Visual Autoregressive Models
1Carnegie Mellon University · University of Illinois Urbana-Champaign · 3Toyota Research Institute
Structure-Detail Decoupled Autoregressive Generation for Fast and High-Fidelity Virtual Try-On
Jul 13, 2026Lu Yang, Xiaonan Hu, Yanan Li +3Virtual Try-On Visual Autoregressive Models
Huazhong University of Science and Technology, China · Wuhan Institute of Technology, China · Xiaomi EV, China
MEPA: Multi-Scale Representation Alignment for Visual Autoregressive Modeling with Mixture of Experts
Jul 1, 2026Nuoyan Zhou, Zhijun Tu, Lei Yu +4Visual Autoregressive Models Autoregressive Image Generation
State Key Laboratory of Integrated Services Networks, Xidian University, Xi’an, China · † This work was done during an internship at Huawei Technologies Co., Ltd. · Huawei Technologies Co., Ltd.
DeVAR: Low-Dose CT Denoising via Visual Autoregressive Modeling
Jun 26, 2026Xizhuo Zhang, Yannian Gu, Zhongzhen Huang +2Visual Autoregressive Models Computed Tomography
Shanghai Jiao Tong University, China · SenseTime Research, China · Shanghai Innovation Institute, China
Data Provenance for Image Auto-Regressive Generation
Jun 22, 2026Bihe Zhao, Louis Kerner, Michel Meintz +3Visual Autoregressive Models Autoregressive Image Generation
CISPA Helmholtz Center for Information Security
Training-Free Semantic Correction for Autoregressive Visual Models
Jun 21, 2026Junhao Chen, Chanyu Zhu, Zheqi Lv +2Visual Autoregressive Models Autoregressive Image Generation
Zhejiang University · Shandong University
SSD: Spatially Speculative Decoding Accelerates Autoregressive Image Generation
Jun 18, 2026Shilong Xiang, Zirui Zhang, Lijun Yu +1Visual Autoregressive Models Speculative Decoding
Rutgers University
Shift-and-Sum Quantization for Visual Autoregressive Models
Jun 15, 2026Jaehyeon Moon, Bumsub HamVisual Autoregressive Models Autoregressive Image Generation
1Yonsei University · 2Articron
SACE: Concept Erasure at the Semantic Singularity in Visual Autoregressive Models
Jun 14, 2026Siya Yang, Nanxiang Jiang, Zhaoxin Fan +1Visual Autoregressive Models T2I Generation
1Hefei University of Technology · 2Beihang University
Edit the Bits, Diff the Codes: Bitwise Residual Editing for Visual Autoregressive Models
Jun 11, 2026Shengqiang Zhang, Ruotong Liao, Volker Tresp +2Text-Guided Image Editing Visual Autoregressive Models
LMU Munich & Munich Center for Machine Learning (MCML)
HACK++: Towards More Effective Head-Aware Key-Value Compression for Efficient Visual Autoregressive Modeling
Jun 6, 2026Ziran Qin, Yuchen Jiang, Mingbao Lin +4KV Caching Visual Autoregressive Models
Shanghai Jiao Tong University, Shanghai, China · Rakuten, Singapore · Tsinghua University, Beijing, China
Knowledge Distillation for Visual Autoregressive Models
Jun 4, 2026Elia Peruzzo, Aritra Bhowmik, Guillaume Sautiere +2Visual Autoregressive Models Autoregressive Image Generation
Qualcomm AI Research∗ · Qualcomm AI Research · University of Technology Nuremberg
Parallel Jacobi Decoding for Fast Autoregressive Image Generation
Jun 4, 2026Boya Liao, Ying Li, Siyong Jian +1Visual Autoregressive Models Autoregressive Image Generation
Westlake University
Where to Refine, When to Stop: Rethinking Redundancy via Latent Discrepancy for Efficient Visual Autoregressive Generation
May 29, 2026Changwang Mei, Peisong Wang, Zekun Li +7Efficient Neural Network Inference Visual Autoregressive Models
Nanjing University of Science and Technology · C2DL, Institute of Automation, Chinese Academy of Sciences · Beijing Academy of Artificial Intelligence +2
VPG: Visual Prefix Guidance for Autoregressive Image and Video Generation
May 28, 2026Xinyao Liao, Qiyuan He, Yicong Li +4Visual Autoregressive Models Autoregressive Image Generation
National University of Singapore · Huazhong University of Science & Technology
Adversarial Error Correction for Visual Autoregressive Generation
May 24, 2026Ligong Bi, Tao Huang, Jianyuan Guo +1Visual Autoregressive Models Autoregressive Image Generation
1Shanghai Jiao Tong University · 2City University of Hong Kong · 3The University of Sydney
CPC-VAR:Continual Personalized and Compositional Generation in Visual Autoregressive Models
May 19, 2026Junhao Li, Xinhao Zhong, Yi sun +4Visual Autoregressive Models Subject-Driven Image Generation
Harbin Institute of Technology, Shenzhen · South China University of Technology · Peng Cheng Laboratory +1

Oct 11, 2026J/K move · Enter open · S save

HGA-Quant: Heterogeneity-Aware Quantization for AutoRegressive Visual Generation Models

Xinmeng Liu, Xianglong Yan, Kaicheng Yang, Weisheng Dong+1

Shanghai Jiao Tong University · Xi’an University of Electronic Science and Technology

Post-Training Quantization Neural Network Quantization Low-Bit Quantization Activation Quantization Visual Autoregressive Models Mixed-Precision Quantization

Open paper AskPDF arXiv

Recently, AutoRegressive Visual Generation (ARVG) models have emerged as a competitive paradigm for high-quality image synthesis. However, their substantial computational and memory demands make them difficult to deploy on resource-constrained edge devices. Post-Training Quantization (PTQ) provides a promising solution to reduce the computational and memory costs of ARVG models, yet existing methods often suffer from substantial image quality degradation under low-bit quantization. In this paper, we propose a training-free PTQ framework for ARVG, consisting of Tail-Aware Bounded-init Grid Refinement (T-BGR) and Coarse-to-Fine Activation Precision Scheduling (CAPS). T-BGR introduces a tail-aware reconstruction objective for weight quantizer calibration, assigning greater importance to distribution-tail coefficients and preserving fine-grained information without introducing additional parameters or high-precision branches. CAPS performs coarse-to-fine activation precision scheduling by jointly considering the importance and computational cost of different generation stages, enabling efficient mixed-precision allocation under a given bit budget. Extensive experiments on multiple ARVG backbones demonstrate that our framework consistently improves robustness under aggressive low-bit quantization. Notably, on Infinity, our framework achieves near full-precision image generation quality at an effective W4A5.6 activation precision, while existing methods at W4A6 still exhibit noticeable quality degradation or inconsistent performance. Code is available at https://github.com/lxm-lc/HGA-Quant.