Seong Joon Oh

Latest

• AdamP: Slowing Down the Slowdown for Momentum Optimizers on Scale-invariant Weights