MMODELYST
People/Hamed Hassani
PPL

Hamed Hassani

Papers (4)
Feature Learning in Linear-Width Two-Layer Networks: Two vs. One Step of Gradient DescentInfoSFT: Learn More and Forget Less with Information-Aware Token WeightingContinuous First, Discrete Later: VQ-VAEs Without Dimensional CollapseRisk-Controlled Post-Processing of Decision Policies