Hi, I'm a PhD student at George Mason University advised by Prof. Ziyu Yao. I’m interested in understanding how language models work and applying those insights to practical applications such as improving model capabilities and controllability.
I am actively seeking a Summer 2026 internship. My recent work centers around mechanistic interpretability (MI) of language models, aiming to better understand how model perform reasoning and to leverage those insights to improve their reasoning capabilities.
Resume / Twitter / Google Scholar / LinkedIn
News & Updates
Upcoming events:
- Attending NeurIPS (San Diego) to present my work on Balanced Parentheses Errors.
Past news:
- We organized The First Workshop on the Application of LLM Explainability to Reasoning and Planning at COLM2025.
- Released the preprint for our survey paper on Mechanistic Interpretability.
- Conducted Tutorial on Mechanistic Interpretability for Language Models at ICML2025.
- Our paper, Failure by Interference: Language Models Make Balanced Parentheses Errors When Faulty Mechanisms Overshadow Sound Ones got accepted in NeurIPS 2025.
- Our paper, All for one: Llms solve mental math at the last token with information transferred from other tokens got accepted in EMNLP 2025.
Papers
NeurIPS 2025
An Failure by Interference: Language Models Make Balanced Parentheses Errors When Faulty Mechanisms Overshadow Sound Ones. (Paper)
EMNLP 2025
All for one: Llms solve mental math at the last token with information transferred from other tokens. (Paper)
Pre-print 2025
A Practical Review of Mechanistic Interpretability for Transformer-Based Language Models. (Paper)
EMNLP 2025
A survey on sparse autoencoders: Interpreting the internal mechanisms of large language models. (Paper)
ACL 2024
An investigation of neuron activation as a unified lens to explain chain-of-thought eliciting arithmetic reasoning of llms. (Paper)
Pre-print 2024
Understanding the Effect of Algorithm Transparency of Model Explanations in Text-to-SQL Semantic Parsing. (Paper)
ACL 2023
Improving generalization in language model-based text-to-SQL semantic parsing: Two simple semantic boundary-based techniques. (Paper)
AAAI 2023
Explaining large language model-based neural semantic parsers (student abstract). (Paper)
Talks
ICML 2025
Invited Talk (George Washington Uni)
Mechanistic Interpretability of Language Models at George Washington University.