avatar

Haiyan Zhao

Ph.D. Student
New Jersey Institute of Technology
hz54@njit.edu


About Me

Hi there! 👋 I am a third-year graduate student at the Department of Computer Science, New Jersey Institute of Technology (NJIT). Currently, I am working closely with Prof. Mengnan Du on model interpretability. My research interests lie in representation understanding, with an emphasis on activation explanations with large language models (LLMs). My recent work focuses on activation inversion and generating natural language explanations for internal features in LLMs.

🚀 I am actively seeking research internship opportunities and welcome collaborations on LLM interpretability. Feel free to reach out! 📬

Research Interests

XAI, LLM Alignment

News

  • [06/2024] Talk @ IEEE ICHI: Explaining training samples and enhancing trustworthiness with LLM
  • [10/2023] Invited talk @ NJIT FinTech Seminar: Explainability on LLMs
  • [10/2023] Invited talk @ MLLM-AI: Explainability on LLMs
  • [10/2023] Invited talk @ UberAI: Explainability on LLMs

Selected Publications

  1. arXiv
    Universal Activation Verbalizer: A Unified Framework for Cross-Model Activation Explanation
    Haiyan Zhao, Zirui He, Guanchu Wang, Ali Payani, Yingcong Li, Mengnan Du
    preprint
    PDF
  2. arXiv
    SAEExplainer: Interpreting SAE Features with Activation-Guided Preference Optimization
    Jingyi He, Haiyan Zhao, Ruxue Shi, Yanguang Liu, Xin Wang, Fei Sun, Mengnan Du
    preprint
    PDF
  3. arXiv
    Rep2Text: Decoding Full Text from a Single LLM Token Representation
    Haiyan Zhao, Zirui He, Fan Yang, Ali Payani, Mengnan Du
    preprint
    PDF
  4. arXiv
    MemLens: Uncovering Memorization in LLMs with Activation Trajectories
    Zirui He, Haiyan Zhao, Ali Payani and Mengnan Du
    preprint
    PDF
  5. EACL
    Denoising Concept Vectors with Sparse Autoencoders for Improved Language Model Steering
    Haiyan Zhao, Xuansheng Wu, Fan Yang, Bo Shen, Ninghao Liu and Mengnan Du
    EACL Findings 2026
    PDF
  6. EMNLP
    A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Models
    Dong Shu*, Xuansheng Wu*, Haiyan Zhao*, Daking Rai, Ziyu Yao, Ninghao Liu and Mengnan Du
    EMNLP Findings 2025
    PDF
  7. ECML
    SAIF: A Sparse Autoencoder Framework for Interpreting and Steering Instruction Following of Language Models
    Zirui He*, Haiyan Zhao*, Yiran Qiao, Fan Yang, Ali Payani, Jing Ma and Mengnan Du
    ECML PKDD 2026
    PDF
  8. EMNLP
    Large Vision-Language Model Alignment and Misalignment: A Survey Through the Lens of Explainability
    Dong Shu, Haiyan Zhao, Jingyu Hu, Weiru Liu, Lu Cheng, and Mengnan Du
    EMNLP Findings 2025
    PDF
  9. ICLR
    Beyond Single Concept Vector: Modeling Concept Subspace in LLMs with Gaussian Distribution
    Haiyan Zhao, Heng Zhao, Bo Shen, Ali Payani, Fan Yang, Mengnan Du
    ICLR 2025
    PDF Code Project Page
  10. TKDD
    Usable XAI: 10 Strategies Towards Exploiting Explainability in the LLM Era
    Xuansheng Wu*, Haiyan Zhao*, Yaochen Zhu*, Yucheng Shi*, Fan Yang, Tianming Liu, Xiaoming Zhai, Wenlin Yao, Jundong Li, Mengnan Du, Ninghao Liu
    ACM TKDD 2026
    PDF Code
  11. SIGKDD
    Towards Uncovering How Large Language Model Works: An Explainability Perspective
    Haiyan Zhao, Fan Yang, Bo Shen, Himabindu Lakkaraju, Mengnan Du
    SIGKDD Explorations 2025
    PDF
  12. TIST
    Explainability for Large Language Models: A Survey
    Haiyan Zhao, Hanjie Chen, Fan Yang, Ninghao Liu, Huiqi Deng, Hengyi Cai, Shuaiqiang Wang, Dawei Yin, Mengnan Du
    ACM TIST 2024
    PDF Code

Teaching

Services


Powered by Jekyll and Minimal Light theme.