Jingcheng (Frank) Niu
Hi! I am a Postdoctoral researcher at the UKP Lab @TU Darmstadt, led by Professor Iryna Gurevych.
Previously, I did my PhD at the Computational Linguistics Group @UofT. My advisor is Professor Gerald Penn. ηιη¨ is my name in Chinese.
I am interested in LLM Interpretability. You can find my research on the List of Publications.
Curriculum Vitae Scholar GitHub Twitter Linkedin
Feel free to email me about my research! I'm also happy to discuss research-related topics, grad school application, or careers in AI/NLP. Email: firstname dot lastname at tu-darmstadt.de.
News
- [May 2026] Paper All Circuits Lead to Rome: Rethinking Functional Anisotropy in Circuit and Sheaf Discovery for LLMs is accepted to ICML! ππ
- [Mar 2026] Paper Hypothesis-Driven Feature Manifold Analysis in LLMs via Supervised Multi-Dimensional Scaling is accepted to TMLR! ππ
- [Sep 2025] Paper Illusion or Algorithm? Investigating Memorization, Emergence, and Symbolic Processing in In-Context Learning is accepted to TMLR!
- [Aug 2025] 2 papers are accepted at EMNLP 2025 main conference! ππ
- [July 2025] Paper Llama See, Llama Do: A Mechanistic Perspective on Contextual Entrainment and Distraction in LLMs won the Outstanding Paper Award at ACL 2025!
- [Jan 2025] I am joining the UKP Lab as a postdoctoral fellow.
Selected Publications
- All Circuits Lead to Rome: Rethinking Functional Anisotropy in Circuit and Sheaf Discovery for LLMsXi Chen*, Mingyu Jin*, Jingcheng Niu*, Yutong Yin, Jinman Zhao, Bangwei Guo, Dimitris N. Metaxas, Zhaoran Wang, Yutao Yue and Gerald Penn.ICML 2026
- Illusion or Algorithm? Investigating Memorization, Emergence, and Symbolic Processing in In-Context LearningTMLR 2025
(J2C Certification) - Llama See, Llama Do: A Mechanistic Perspective on Contextual Entrainment and Distraction in LLMsACL 2025
(Outstanding Paper Award) - What does the Knowledge Neuron Thesis Have to do with Knowledge?ICLR 2024
(spotlight)