Raphaël Sarfati

Text-Trained LLMs Can Zero-Shot Extrapolate PDE Dynamics

Density estimation with LLMs: a geometric investigation of in-context learning trajectories

Lines of Thought in Large Language Models