About
I am a PhD student at Berkeley AI Research, advised by Alexei A. Efros (2023-present). I aim to characterize emergent structures that neural networks develop and build phenomenological models that make testable predictions, drawing inspiration from physics, biology, and neuroscience. Some of my current interests include the science of scaling, representation universality, weight space geometry, learning dynamics, and how data shapes model behavior. My research is supported by the DOE Computational Science Graduate Fellowship.
Previously, I graduated from Northwestern in 2023 with a BS in Computer Science. During my undergrad, I had the pleasure of working with many wonderful mentors: Pietro Perona, Aggelos Katsaggelos, Jennifer J. Sun, Vibhav Vineet, and Neel Joshi.
Recent News
-
New preprint released: Base Models Can Reason By Taking a Cue From Training Data
-
Neuron Populations Exhibit Divergent Selectivity with Scale received the Best Paper Award at the COLM 2026 Scientific Understanding of Foundation Models workshop.
-
Neuron Populations Exhibit Divergent Selectivity with Scale accepted to NeurIPS 2026.
-
New preprint released: Neuron Populations Exhibit Divergent Selectivity with Scale
-
Co-organized the How Do Vision Models Work workshop at CVPR.
-
CS Seminar talk at Northwestern University.
Selected Publications
See my Google Scholar for the full list of publications.
Neuron Populations Exhibit Divergent Selectivity with Scale
YB, AE, and YG jointly advised this work.
NeurIPS 2026
Best Paper Award, COLM 2026 Scientific Understanding of Foundation Models Workshop
Vision Transformers Don't Need Trained Registers
NeurIPS 2025 (Spotlight, top 3%)
Interpreting the Weight Space of Customized Diffusion Models
NeurIPS 2024
