Shiyuan Huang is a Ph.D. student in Computer Science and Engineering at UC Santa Cruz, advised by Dr. Leilani Gilpin and Dr. Ian Lane. His research focuses on the explainability and trustworthiness of large language models (LLMs), aiming to uncover their internal mechanisms to help users better understand and trust them.

Currently, he investigates how likelihood and uncertainty quantification in language model generation can be used to evaluate models’ knowledge boundaries and abstention behavior. He also leverages these measures to explain why LLMs perform well or poorly on specific tasks, with the broader goal of improving their reliability and interpretability.