Mike Zhou
mikezhou [at] seas [dot] upenn [dot] edu
My name is Mike Zhou and I am currently a fourth year undergraduate studying computer science and mathematics at the University of Pennsylvania. I am interested in deep learning, natural langauge processing, and arithmetic combinatorics. I am currently an undergraduate researcher at Penn’s Cognitive Computation Group, where I am working with Professor Dan Roth on various aspects of machine translation and language model reasoning. Before that, I worked on the learnability of n-gram models at Rycolab.
Within NLP, I am interested in language model reasoning. I am interested in where models parametrically store their knowledge, and why some prompts (like CoT and self consistency) elicit reasoning better than others. On my free time, I also like to think about problems related to arithmetic combinatorics and graph theory.
Click here to learn more about me, or check out my notes here.
latest posts
Oct 21, 2024 | Hardy-Ramanujan with probabilistic method |
---|---|
Sep 16, 2024 | Minimum number of lines from non-collinear points |
selected publications
- Can Transformers Learn n-gram Language Models?In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing , 2024