I am a final-year computer science Ph.D. candidate at University of Toronto working with Yang Xu in the Computational Linguistics Group. I am broadly interested in natural language processing, machine learning, and computational linguistics.

My current research is focused on unveiling the mechanistic principles of large language models in solving real-world problems. In particular, I am deeply interested in understanding why language models sometimes generate certain types of undesired outputs, such as non-factual hallucinations and adversarially triggered harmful contents. This line of research has the potential to enhance the safety and transparency of state-of-the-art generative AI systems.

During my PhD, I studied computational modeling of creative language use such as metaphor and idiom.

For more information, you can find my CV here.

Contact: jadeleiyu [at] cs [dot] toronto [dot] edu