no code implementations • 14 Dec 2021 • Jiahao Dai, Jing-Wen Yi, Li Chai
The contributions of this paper are three: (i) For the one-tap memory scheme, we demonstrate that the state deviation memory is useless for the optimal convergence.
