1 paper · 1 filter
Denys Pushkin, Luis Barba
In recent years, SGD and its variants have become the standard tool to train Deep Neural Networks. In this paper, we focus on the recently proposed variant Lookahead, which improve…