About Me
👋 Hi, I’m Junzhuo Li
I am a Ph.D. student at Hong Kong University of Science and Technology (Guangzhou) 🎓, advised by Prof.Xuming Hu (since 2024).
Before that, I received my M.Sc. degree from Tianjin University under the guidance of Prof. Deyi Xiong.
My research interests lie in Natural Language Processing (NLP) and Large Language Models (LLMs), focusing on:
- 🔎 Mixture-of-Experts (MoE) models – interpretability, scaling laws, and efficient training
- 🧩 Knowledge attribution and transfer – tracing and controlling knowledge in LLMs
- ⚡ Efficient training & inference – distillation, routing, compute allocation
If you’re interested in collaborating 🤝, feel free to reach out via email: jzli[at]tju[dot]edu[dot]cn or jz.li[at]connect.hkust-gz.edu.cn
🔔 News
- [08/2025] 🎉 Our paper Multi-Domain MoE Adaptation and Internal Chain-of-Thought was accepted to EMNLP 2025 (Suzhou).
- [05/2025] 🎉 Our paper Decoding Knowledge Attribution in Mixture-of-Experts was accepted to ACL 2025 (Vienna).
- [01/2024] 🎓 Graduated from Tianjin University (M.Sc.).
- [10/2023] Our papers on Privacy Neurons (DEPN) and Cross-lingual Knowledge Transfer were accepted to EMNLP 2023.
📄 Publications
Junzhuo Li, Bo Wang, Xiuze Zhou, and Xuming Hu.
Dynamic Expert Specialization: Towards Catastrophic Forgetting-Free Multi-Domain MoE Adaptation
EMNLP 2025 (Long Papers), Suzhou, China.Zhipeng Yang, Junzhuo Li, Siyu Xia, and Xuming Hu.
Internal Chain-of-Thought: Empirical Evidence for Layer-wise Subtask Scheduling in LLMs
EMNLP 2025 (Long Papers), Suzhou, China.Junzhuo Li, Bo Wang, Xiuze Zhou, Peijie Jiang, Jia Liu, and Xuming Hu.
Decoding Knowledge Attribution in Mixture-of-Experts: A Framework of Basic-Refinement Collaboration and Efficiency Analysis
ACL 2025 (Long Papers), Vienna, Austria.Yanggan Gu, Junzhuo Li, Sirui Huang, Xin Zou, Zhenghua Li, and Xuming Hu.
Capturing Nuanced Preferences: Preference-Aligned Distillation for Small Language Models
Findings of ACL 2025, Vienna, Austria.Xinwei Wu, Junzhuo Li, Minghui Xu, Weilong Dong, Shuangzhi Wu, Chao Bian, and Deyi Xiong.
DEPN: Detecting and Editing Privacy Neurons in Pretrained Language Models
EMNLP 2023 (Long Papers), Singapore.Shaoyang Xu, Junzhuo Li, and Deyi Xiong.
Language Representation Projection: Can We Transfer Factual Knowledge across Languages in Multilingual Language Models?
EMNLP 2023 (Short Papers), Singapore.Chuang Liu, Junzhuo Li, and Deyi Xiong.
Tab-CQA: A Tabular Conversational Question Answering Dataset on Financial Reports
ACL 2023 (Industry Track), Toronto, Canada.Junzhuo Li and Deyi Xiong.
KaFSP: Knowledge-Aware Fuzzy Semantic Parsing for Conversational Question Answering over a Large-Scale Knowledge Base
ACL 2022 (Long Papers), Dublin, Ireland.
💼 Experience
- Research Intern @ Tongyi Lab (Institute of Intelligent Computing) 🧠 (2025.8 – present)
- Research Intern @ Ant Group 🐜 (2024.12 – 2025.6)
- Recommended Algorithm Intern @ Meituan 🍜 (2023.6 – 2023.7)
- Research Intern @ Bytedance 🎵 (2022.1 – 2022.9)
🎓 Education
- Ph.D. in Artificial Intelligence, Hong Kong University of Science and Technology (Guangzhou) (2024 – )
- M.Sc. in Computer Technology, Tianjin University (2021 – 2024.1)
- B.S. in Computer Science and Technology, Zhengzhou University (2017 – 2021)
⭐ Outside research, I enjoy fitness 🏋️ and tennis 🎾 to stay balanced and energetic!