1 paper · 1 filter
Xinglong Sun, Barath Lakshmanan, Maying Shen +3
As we push the boundaries of performance in various vision tasks, the models grow in size correspondingly. To keep up with this growth, we need very aggressive pruning techniques f…