Home Projects

Megan Wei

I'm a Masters of Engineering student at MIT, advised by Professor Joshua Tenenbaum in the Computational Cognitive Science Group. Currently, I'm working on building compositional diffusion models to generate scenes with complex object relations and properties.

I completed my Bachelor of Science in Computer Science with a minor in Business Analytics at MIT. During my undergrad, I worked with Professor Antonio Torralba on building VirtualHome and Professor Joshua Tenenbaum on comparing large language models to humans on several reasoning tasks.

My main research goal is to bridge the gap between human and machine multi-modal reasoning and perception.

       

profile photo
Research
gpt Structured, flexible, and robust: benchmarking and improving large language models towards more human-like behavior in out-of-distribution reasoning tasks
Katherine M. Collins*, Catherine Wong*, Jiahai Feng, Megan Wei, Joshua B. Tenenbaum
CogSci, 2022
Paper  /  Website  /  Code

We introduce a challenge benchmark for comparing humans and large language models on increasingly constrained planning and explanation tasks, and a new hybrid Parse-and-Solve model that translates language into formal goal programs to improve planning robustness.

Initiatives
wai MITxHarvard Women in AI
YouTube  /  Twitter  /  LinkedIn
Co-President and Founding Team
Started and grew a group of now 600 MIT and Harvard undergrads to postdocs. Spearheaded new initiatives: interview series, paper reading sessions, mentorship, elementary to high school outreach, marketing, corporate sponsorships, and professional development.

Website template credits: John Barron