Hang Chen 💪

Hang ChenHáng Chén

(he/him)

Postdoctoral Research Fellow at Nanyang Technological University

College of Computing & Data Science, Nanyang Technological University

Professional Summary

Hang Chen received his Ph.D. from the Faculty of Electronic and Information Engineering at Xi’an Jiaotong University in March 2026. In July 2026, he joined the College of Computing & Data Science (CCDS) at Nanyang Technological University (NTU) as a Postdoctoral Research Fellow, working with Prof. Wenya Wang. His research is rooted in causal representation and structural analysis in machine learning. Recently, his focus has shifted towards applying the mechanistic interpretability of Large Language Models (LLMs) to guide and control post-training processes, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL).

📄 CV: English | 中文

Education

Ph.D. in Computer Science

2021-09-01
2026-03-30

Xi'an Jiaotong University

B.S. in Computer Science

2016-09-01
2020-06-30

Xi'an Jiaotong University

Interests

Mechanistic Interpretability Post-training Large Language Models Causality Sentiment Analysis
📚 My Research

My research focuses on making Mechanistic Interpretability (MI) practically useful for LLM post-training: integrating interpretability with SFT and RL, and using mechanistic analysis to diagnose and guide post-training. Representative questions include identifying conflicts when fine-tuning on multiple tasks, anticipating a model’s post-SFT mechanisms before training, and using interpretability signals to supervise fine-tuning. Previously, I worked on causal mechanisms of language models, mathematical models of emergence, and downstream tasks such as affective reasoning. 😃

Featured Publications
Beyond Static Interpretability: Anticipating Post-SFT Mechanisms from Pre-SFT Parameters for Better Tuning featured image

Beyond Static Interpretability: Anticipating Post-SFT Mechanisms from Pre-SFT Parameters for Better Tuning

Mechanistic Localization bridges mechanistic interpretability and post-training optimization by isolating critical parameters via interpretative approaches and then guiding …

hang-chen-jiaying-zhu-wenya-wang
•
CLUE: Conflict-guided Localization for LLM Unlearning Framework featured image

CLUE: Conflict-guided Localization for LLM Unlearning Framework

The LLM unlearning aims to eliminate the influence of undesirable data without affecting causally unrelated information. This process typically involves using a forget set to …

hang-chen-jiaying-zhu-xinyu-yang-wenya-wang
•
Rethinking Circuit Completeness in Language Models: AND, OR, and ADDER Gates featured image

Rethinking Circuit Completeness in Language Models: AND, OR, and ADDER Gates

Circuit discovery has gradually become one of the prominent methods for mechanistic interpretability, and research on circuit completeness has also garnered increasing attention. …

hang-chen-jiaying-zhu-xinyu-yang-wenya-wang
•
Towards Causal Relationship in indefinite data: New Datasets and Baseline Model featured image

Towards Causal Relationship in indefinite data: New Datasets and Baseline Model

The cross-fertilization of deep learning and causal discovery has given birth to broader causal data forms, involving multi-structured data like the Netsim dataset, and complex …

hang-chen-xinyu-yang-keqing-du
•
Recent Publications
(2026). Navigating by Old Maps: The Pitfalls of Static Mechanistic Localization in LLM Post-Training. NeurIPS.
(2026). CLUE: Conflict-guided Localization for LLM Unlearning Framework. ICLR.
(2026). Skill Path: Unveiling Language Skills from Circuit Graphs. AAAI (Oral).
(2025). Rethinking Circuit Completeness in Language Models: AND, OR, and ADDER Gates. Neurips.
Recent News
Starting a New Chapter: Joining NTU CCDS as a Postdoc! featured image

Starting a New Chapter: Joining NTU CCDS as a Postdoc!

Having already collaborated with Prof. Wenya Wang on four papers exploring LLM mechanistic interpretability, it is absolutely fantastic to finally be here in person. For my …

avatar
Hang Chen
•
Successfully Earned My Ph.D. from the Faculty of Electronic and Information Engineering at Xi'an Jiaotong University featured image

Successfully Earned My Ph.D. from the Faculty of Electronic and Information Engineering at Xi'an Jiaotong University

Since my initial encounter with machine learning in 2021, my Ph.D. journey has led me to sequentially explore affective computing, causal analysis, probabilistic methods, language …

avatar
Hang Chen
•