Hi! I'm a fourth-year undergraduate at Princeton University studying computer science and math. Right now, I'm particularly excited about mechanistic interpretability, the study of using neural network weights to understand their behavior and internal representations.
Previously, I've spent a lot of time with language models — figuring out where they store facts, turning them into search engines, and using them to convert requests into actions in Google Assistant.
Other things I find fascinating or fun include tech policy, mechanism design, bioinformatics, all sorts of great writing, bouldering, BJJ, and cardistry. You can reach me at [x]@princeton.edu, where [x] is replaced by mwtang.