Large Catapults in Momentum Gradient Descent with Warmup: An Empirical Study

Abstract

Although gradient descent with momentum is widely used in modern deep learning, a concrete understanding of its effects on the training trajectory still remains elusive. In this work, we empirically show that momentum gradient descent with a large learning rate and learning rate warmup displays large catapults, driving the iterates towards flatter minima than those found by gradient descent. We then provide empirical evidence and theoretical intuition that the large catapult is caused by momentum ``amplifying’’ the self-stabilization (Damian et al., 2023).

Junghyun Lee

PhD Student

PhD student at GSAI, KAIST, jointly advised by Prof. Se-Young Yun and Prof. Chulhee Yun. Interested in mathematical and theoretical AI, where I am interested in explaining the recent success of machine/deep learning algorithms and designing mathematically well-grounded algorithms for various scenarios.