Shawn Lee

About Me

I'm Sihyun (Shawn) Lee, an undergraduate at Carnegie Mellon University's School of Computer Science majoring in CS + ML/Systems.

I'll be joining Jump Trading next summer as an intern in Chicago.

I also have a wide range of personal projects and research, which has allowed me to self-acquire many technical skills and competency in developing production-level applications that solve real-life problems for hundreds of people to use.

I enjoy making music and videos in my free time.

Feel free to learn more about me through my bio, or visit my blog!

Bio My blog Resume

Experience

Jump Trading

SWE Intern | Incoming | Chicago, IL

Bodo.ai

SWE Intern | Summer 2024 | Pittsburgh, PA

CMU Language Technologies Institute

Undergrad Researcher | Summer 2023 | Pittsburgh, PA

Published Papers

Think Twice: Perspective-Taking Improves Large Language Models' Theory-of-Mind Capabilities

Alex Wilf, Sihyun Shawn Lee, Paul Pu Liang, Louis-Philippe Morency (2023)

Working with Alex and Prof. Morency in CMU's Language Technologies Institute, I was able to develop a simple prompting strategy that massively improves the theory-of-mind capabilities of large language models — a challenge that has long eluded these systems.

SimToM is a novel LLM prompting strategy that significantly enhances the theory-of-mind capabilities in LLMs. The approach is not only simple and effective, but also universally applicable across various models without the need for extra training. Remarkably, it achieves an improvement of up to +29.5% absolute accuracy over the current state-of-the-art in several theory-of-mind benchmarks.

The paper was accepted as a main conference paper at ACL 2024.

Raphael Mansuy wrote a blog post on our paper -- be sure to check it out!

ACL Print arXiv Preprint Blog Post