vfleaking / max-margin

Code for the paper "Gradient Descent Maximizes the Margin of Homogeneous Neural Networks"

Geek Repo

Github PK Tool