About Me

I’m a second-year CS Ph.D. student at Rice University, advised by Prof. Hanjie Chen. I’m interested in interpretable machine learning—especially understanding how language models behave and process information to make decisions. Before Rice, I earned my master’s at Carnegie Mellon University.

Publications

Steering Information Utility in Key-Value Memory for Language Model Post-Training
Chunyuan Deng, Ruidi Chang, Hanjie Chen
Advances in Neural Information Processing Systems (NeurIPS), 2025.
Learning Distribution-wise Control in Representation Space for Language Models
Chunyuan Deng, Ruidi Chang, Hanjie Chen
International Conference on Machine Learning (ICML), 2025.
SAFR: Neuron Redistribution for Interpretability
Ruidi Chang, Chunyuan Deng, Hanjie Chen
North American Chapter of the Association for Computational Linguistics (NAACL) Findings, 2025.
🧠 Neurons often mix too many features (superposition) — making models a black box.

🎯 SAFR strategically redistributes neurons:
- Monosemantic for important tokens
- Polysemantic where interactions matter
SAFR improves interpretability.
The Generalization Ridge: Information Flow in Natural Language Generation
Ruidi Chang, Chunyuan Deng, Hanjie Chen
arXiv preprint, 2025
🔍 Models aren’t just monotonically better deeper, they show a ridge of generalization in intermidiate layers.

✂️ InfoRidge is an information-theoretic lens to trace how predictive information flows across depth.
- Predictive information peaks in upper-middle layers (the “ridge”), where models exhibit stronger generalization behavior, then drops in final layers.
- Residual scaling probes show that under distribution shift, models downweight deep layers and rely more on ridge layers.
InfoRidge identifies the generalization ridge — an intermidiate-layer peak in predictive information.
Large Language Model Based Multi-Agents: A Survey of Progress and Challenges
Taicheng Guo, Xiuying Chen, Yaqi Wang, Ruidi Chang, Shichao Pei, Nitesh V. Chawla, Olaf Wiest, Xiangliang Zhang
International Joint Conference on Artificial Intelligence (IJCAI), 2024.
Language Models are Symbolic Learners in Arithmetic
Chunyuan Deng, Zhiqi Li, Roy Xie, Ruidi Chang, Hanjie Chen
In submission (Transactions on Machine Learning Research, TMLR).

Services

Reviewer: ACL ARR 2025 Feb, IEEE BigData 2025, EMNLP BlackboxNLP Workshop 2025/2024, COLM 2025 Workshop XLLM-Reason-Plan, COLING 2025
Volunteer: EMNLP BlackboxNLP Workshop 2024
Mentoring: Nursultan Asilbekov (SURF Program)

Last update: 10/2025