About Me
I am Haoming Xu (徐浩铭), a graduate student in the combined Master’s and Ph.D. program in Artificial Intelligence at Zhejiang University, advised by Ningyu Zhang and Shumin Deng. I received my bachelor’s degree in Computer Science and Technology from Harbin Institute of Technology in 2025.
My research focuses on LLM Belief and Trustworthy Large Language Models (大模型 Belief 和可信大模型). I study how language models maintain, update, and isolate beliefs across contexts, and how to diagnose and improve their truthfulness and consistency. My work also explores knowledge editing, machine unlearning, and hallucination mitigation.
I welcome collaborations on LLM belief and trustworthy AI. Please contact me at haomingxu2003@gmail.com.
View my Chinese CV · Download PDF
📝 Publications
You can find my publications on Google Scholar.
2026
EMNLP 2026 Main
Haoming Xu, Weihong Xu, Zongrui Li, Mengru Wang, Yunzhi Yao, Chiyu Wu, Jin Shang, Yu Gong, Shumin Deng
We introduce Contextual Belief Management (CBM) and BeliefTrack to diagnose failures in belief retention, updating, and isolation. Belief-state rewards for reinforcement learning reduce the average failure rate by 70.9%.
Nature Machine Intelligence (NMI), 2026
Ningyu Zhang, Yunzhi Yao, Jiaxin Qin, Haoming Xu, Yuqi Zhu, Zeping Yu, Mengru Wang, Yuqi Tang, Jia-Chen Gu, Shumin Deng, Huajun Chen
A systematic analysis of the challenges and limitations of knowledge editing for reasoning models, with directions for principled methods that address reasoning capabilities.
Haoming Xu, Ningyuan Zhao, Yunzhi Yao, Weihong Xu, Hongru Wang, Xinle Deng, Shumin Deng, Jeff Z. Pan, Huajun Chen, Ningyu Zhang
We introduce Neighborhood Consistency Belief (NCB) to diagnose fragile beliefs in LLMs and a SAT method that improves robustness by approximately 30%.
2025
Yunzhi Yao, Jiaxin Qin, Ningyu Zhang, Haoming Xu, Yuqi Zhu, Zeping Yu, Mengru Wang, Yuqi Tang, Jia-Chen Gu, Shumin Deng, Nanyun Peng, Huajun Chen
Revisiting knowledge editing methods for large language models in the reasoning era.
Jizhan Fang, Xinle Deng, Haoming Xu, Ziyan Jiang, Yuqi Tang, Ziwen Xu, Shumin Deng, Yunzhi Yao, Mengru Wang, Shuofei Qiao, Huajun Chen, Ningyu Zhang
A lightweight and efficient memory-augmented generation framework for large language models.
Ziwen Xu, Shuxun Wang, Kewei Xu, Haoming Xu, Mengru Wang, Xinle Deng, Yunzhi Yao, Guozhou Zheng, Huajun Chen, Ningyu Zhang
An easy-to-use steering framework for editing large language models with improved performance.
Haoming Xu, Shuxun Wang, Yanqiu Zhao, Yi Zhong, Ziyan Jiang, Ningyuan Zhao, Shumin Deng, Huajun Chen, Ningyu Zhang
We propose an unlearning approach via model merging for SemEval-2025 Task 4. We achieved an excellent second-place finish.
Haoming Xu, Ningyuan Zhao, Liming Yang, Sendong Zhao, Shumin Deng, Mengru Wang, Bryan Hooi, Nay Oo, Huajun Chen, Ningyu Zhang
A novel framework for machine unlearning in large language models through a learning-based approach.
2024
Chenxi Wang, Xiang Chen, Ningyu Zhang, Bozhong Tian, Haoming Xu, Shumin Deng, Huajun Chen
A dynamic correction decoding strategy for multimodal large language models to mitigate hallucinations.
Zijian Li, Sendong Zhao, Haochun Wang, Haoming Xu, Bing Qin, Ting Liu
A comprehensive benchmark for Chinese medical complex open-question answering systems.
2023
Haochun Wang, Sendong Zhao, Zewen Qiang, Zijian Li, Nuwa Xi, Yanrui Du, MuZhen Cai, Haoqiang Guo, Yuhan Chen, Haoming Xu, Bing Qin, Ting Liu
Integrating structured medical knowledge bases to improve reliable response generation in Chinese medical QA systems.
🚀 Projects
An easy-to-use steering framework for large language models with improved editing capabilities and user experience.
As a core maintainer, I integrated DPO and QLoRA alignment and fine-tuning modules, contributed to the framework architecture, and optimized MEMIT and ROME for performance and extensibility across model architectures.
An ICLR 2026 project for efficient long-term memory management and reduced redundancy in long-context scenarios. I contributed to Memory repository baselines and ongoing maintenance.
📖 Education
Artificial Intelligence · Combined Master’s and Ph.D. Program
Zhejiang University, September 2025 - June 2030 (expected)
Advisors: Ningyu Zhang and Shumin Deng
Bachelor of Computer Science and Technology
School of Honors, Harbin Institute of Technology, September 2021 - June 2025
🏆 Honors & Competitions
- Second place, SemEval-2025 Task 4: LLM Unlearning. Led the team in combining unlearning methods and model merging to balance forgetting and retention.
- Global Second Prize, ASC24 Student Supercomputer Challenge. Optimized LLM inference with vLLM, quantization, and early stopping to improve throughput in a supercomputing environment.
- Jiang Chen Scholarship and Zhejiang Provincial Xinmiao Talent Program.
💻 Internship Experience
- Algorithm Intern | June 2026 - present
Alibaba Security AGI Lab · Yuvion Team, China - Research Intern | July 2023 - June 2024
Research Center for Social Computing and Interactive Robotics (SCIR), Harbin Institute of Technology, Harbin, China - AI Technology Intern | July 2022 - August 2022
LEJU Robotics, Suzhou, China
