Hi! I am Haoxiang Wang (王浩翔), an incoming Ph.D. student at the Wangxuan Institute of Computer Technology, Peking University. I received my B.S. in Computer Science from the School of EECS, Peking University in 2026.
Recently, my research has focused on evaluating the dynamic capability boundaries of large language models, especially how to move beyond fixed benchmarks and static worst-case attacks toward skill-guided search and calibrated difficulty estimation.
Research Interests
Large Language Models (LLM) AI Safety Reinforcement Learning (RL) Multi-modal Learning
News
- May 2026 Our work Beyond Fixed Benchmarks and Worst-Case Attacks: Dynamic Boundary Evaluation for Language Models is currently under submission to NeurIPS 2026.
- May 2026 My undergraduate thesis on dynamic boundary evaluation for language models, titled 大语言模型动态能力边界评测：技能引导搜索与难度校准, was selected as an Outstanding Undergraduate Thesis.
- Jan 2026 Serving as a reviewer for ACL Rolling Review (ARR).
- Oct 2025 Awarded the Zhiban Scholarship (智班奖学金), Peking University.
- Sep 2025 Our paper SPTI has been accepted to NeurIPS 2025!