Multimodal learning & computer vision

Gengwei Zhang

张耕维/David

Postdoctoral Researcher

Department of Computer Science
University of North Carolina at Chapel Hill

01 /

About Me

I am a Postdoctoral Research Associate in the Department of Computer Science, University of North Carolina at Chapel Hill, hosted by Prof. Tianlong Chen. I obtained my Ph.D. from the Faculty of Engineering and Information Technology at the University of Technology Sydney (UTS), supervised by Prof. Yunchao Wei and Prof. Ling Chen.

I received my bachelor's degree in School of Data and Computer Science (currently School of Computer Science and Engineering) from Sun Yat-Sen University (SYSU). Before UTS, I spent a wonderful year working with Prof. Xiaodan Liang at SYSU as a research assistant.

My research focuses on multimodal learning, with particular interests in multimodal reasoning and image and video generation. I also work on continual learning and adapting pre-trained models to new tasks.

Explore my publications

Latest updates

2026.03
Our paper Understanding the Role of Hallucination in Reinforcement Post-Training of Multimodal Reasoning Models is accepted by CVPR 2026.
2026.03
Our paper M4V: Multimodal Mamba for Efficient Text-to-Video Generation is accepted by CVPR 2026. The code is available.
2025.12
I will join UNC as a postdoc.
2025.11
I received my Ph.D. degree!

Earlier updatesFewer updates

2024.08
A substantial improvement of our SLCA is available: SLCA++: Unleash the Power of Sequential Fine-tuning for Continual Learning with Pre-training. The code is also available.
2023.07
Our paper “SLCA: Slow Learner with Classifier Alignment for Continual Learning on a Pre-trained Model” has been accepted by ICCV 2023, and the code is available.
2022.12
Our paper “Mask Matching Transformer for Few-Shot Segmentation” has been accepted by NeurIPS 2022, and the code is available.
2021.12
Our paper “Few-Shot Segmentation via Cycle-Consistent Transformer” has been accepted by NeurIPS 2021, and the code is available.

02 /

Research directions

Reasoning, generating, and learning across tasks.

Multimodal reasoning

Understanding and improving reasoning in multimodal models through reinforcement post-training.

Explore publications

Visual generation

Generating images and videos with flexible and efficient generative models.

Explore publications

Continual learning

Adapting pre-trained models to new tasks while retaining prior knowledge.

Explore publications

Multimodal LearningMultimodal reasoning · Reinforcement post-training · Image and video generation

Continual LearningPre-trained model adaptation · Sequential fine-tuning · Knowledge retention

03 /

Selected publications

Full list on Google Scholar

A selection of my work. * Equal contribution.

CVPR2026Video Generation

M4V: Multimodal Mamba for Efficient Text-to-Video Generation

Jiancheng Huang^*, Gengwei Zhang^*, Zequn Jie, Siyu Jiao, Yinlong Qian, Ling Chen, Yunchao Wei, Lin Ma

Multimodal Learning

Paper Code Project

CVPR2026Multimodal Reasoning

Understanding the Role of Hallucination in Reinforcement Post-Training of Multimodal Reasoning Models

Gengwei Zhang, Jie Peng, Zhen Tan, Mufan Qiu, Hossein Nourkhiz Mahjoub, Vaishnav Tadiparthi, Kwonjoon Lee, Yanyong Zhang, Tianlong Chen

Reinforcement Learning

Paper

NeurIPS2025Image Generation

FlexVAR: Flexible Visual Autoregressive Modeling without Residual Prediction

Siyu Jiao, Gengwei Zhang, Yinlong Qian, Jiancheng Huang, Yao Zhao, Humphrey Shi, Lin Ma, Yunchao Wei, Zequn Jie

Autoregressive Modeling

Paper Code

Preprint2024Continual Learning

SLCA++: Unleash the Power of Sequential Fine-tuning for Continual Learning with Pre-training

Gengwei Zhang^*, Liyuan Wang^*, Guoliang Kang, Ling Chen, Yunchao Wei

Pre-trained Models

Paper Code

ICCV2023Continual Learning

SLCA: Slow Learner with Classifier Alignment for Continual Learning on a Pre-trained Model

Gengwei Zhang^*, Liyuan Wang^*, Guoliang Kang, Ling Chen, Yunchao Wei

Image Classification

Paper Code

CVPR2022Continual Learning

Continual Object Detection via Prototypical Task Correlation Guided Gating Mechanism

Binbin Yang, Xinchi Deng, Han Shi, Changlin Li, Gengwei Zhang, Hang Xu, Shen Zhao, Liang Lin, Xiaodan Liang

Object Detection

Paper

NeurIPS2022Few-shot Learning

Mask Matching Transformer for Few-Shot Segmentation

Siyu Jiao, Gengwei Zhang, Shant Navasardyan, Ling Chen, Yao Zhao, Yunchao Wei, Humphrey Shi

Segmentation

Paper Code

ICLR2021AutoML

Loss Function Discovery for Object Detection via Convergence-Simulation Driven Search

Peidong Liu^*, Gengwei Zhang^*, Bochao Wang, Hang Xu, Xiaodan Liang, Yong Jiang, Zhenguo Li

Object Detection

Paper

AAAI2021AutoML

Ada-Segment: Automated Multi-loss Adaptation for Panoptic Segmentation

Gengwei Zhang, Yiming Gao, Hang Xu, Hao Zhang, Zhenguo Li, Xiaodan Liang

Segmentation

Paper

NeurIPS2021Few-shot Learning

Few-Shot Segmentation via Cycle-Consistent Transformer

Gengwei Zhang, Guoliang Kang, Yi Yang, Yunchao Wei

Segmentation

Paper Code

CVPR2020Graph Reasoning

Bidirectional Graph Reasoning Network for Panoptic Segmentation

Yangxin Wu^*, Gengwei Zhang^*, Yiming Gao, Xiajun Deng, Ke Gong, Xiaodan Liang, Liang Lin

Segmentation

Paper

NeurIPS2020AutoML

Auto-Panoptic: Cooperative Multi-Component Architecture Search for Panoptic Segmentation

Yangxin Wu, Gengwei Zhang, Hang Xu, Xiaodan Liang, Liang Lin

Segmentation

Paper

Get in touch

Let’s talk research.

For research conversations and academic inquiries.

[email protected]