Research

Fast optimizers. Principled optimizers.

Spring 2025: wrote my thesis as a guide to understanding Muon, and to advocate spectrally regulating weights.

Fall 2024: co-created Muon, which was born out of the modular duality framework with Jeremy Bernstein.

Selected Publications

Training Transformers with Enforced Lipschitz Constants (2025)

Duality, Weight Decay, and Metrized Deep Learning (2025)

Modular Duality in Deep Learning (2024)

Old Optimizer, New Norm: An Anthology (2024)

An Assessment of Model-on-Model Deception (2024)

Bridging Autoregressive Neural Networks and Tensor Networks for Quantum Many-Body Simulation (2023)

Who Are the Gatekeepers? An Examination of Diversity in INFORMS Journal Editorial Boards (2021)