John Cooper
PhD Candidate, Computer Science
UW Madison, advised by Fred Sala
Email: jfcooper2@cs.wisc.edu · GitHub · Google Scholar · CV
About Me
Hello! I’m a 3rd year PhD student at UW Madison in Fred Sala’s group working on understanding the expressivity/limitations of hybrid models, RL, and steering. I also have collaborated with Dimitris Papailiopoulos investigating the expressivity and properties of transformers. Previously, I was an undergraduate at the University of New Mexico.
I am particularly fond of theoretical machine learning and mechanistic interpretability. Check out some of my recent projects!
News
- December, 2025: I’m starting to work on my own blog! It probably won’t be very consistent from post to post, but maybe you’ll find something interesting eventually!
- November, 2025: Dyah Adila and I posted a new blog about the similarities/differences between fine-tuning and steering. Take a look!
Recent Publications
-
Weak-to-Strong Generalization Through the Data-Centric Lens, ICLR 2026
Changho Shin, John Cooper, Fred Sala -
Everything Everywhere All at Once: LLMs can In-Context Learn Multiple Tasks in Superposition, ICML 2025 Spotlight
Zheyang Xiong, Ziyang Cai, John Cooper, Albert Ge, Vasilis Papageorgiou, Zack Sifakis, Angeliki Giannou, Ziqian Lin, Liu Yang, Saurabh Agarwal, Grigorios Chrysos, Samet Oymak, Kangwook Lee, Dimitris Papailiopoulos
Pre-print Publications
- R&B: Domain Regrouping and Data Mixture Balancing for Efficient Foundation Model Training
Albert Ge, Tzu-Heng Huang, John Cooper, Avi Trost, Ziyi Chu, Satya Sai Srinath Namburi GNVV, Ziyang Cai, Kendall Park, Nicholas Roberts, Fred Sala
Teaching
- Teaching assistant for CS 354 (Machine Organization and Programming)
Fall 2022-Fall 2023
Service
Reviewed for ICML 2024, TMLR 2024, ICLR 2025, 2026.