AI Paper Reading Club

​AI Paper Reading Club is a small, informal gathering for researchers and engineers in SF to discuss recent papers, experiments, and ideas in AI in person. The goal is to bring technical discourse offline and into a more thoughtful, conversational setting.

​For our first session, we’ll be discussing our recent release, Aurora: a new optimizer that explores failure modes in Muon and introduces a leverage-aware approach to improving training efficiency in frontier-scale models.

​Read the paper here: Aurora: A Leverage-Aware Optimizer for Rectangular Matrices

​We’ll kick things off with brief introductions and a discussion of the paper, then keep things fairly informal and discussion-oriented from there. Snacks and beverages will be provided.

​Moving forward, we’d love for this to become a recurring space for discussing interesting papers, experiments, and ideas across AI research more broadly—including work beyond our own! We’re very open to suggestions, and if you’d like to collaborate on or host a future session, please reach out.