The Importance of Max Pooling in NLP Architecture

Max pooling is when we take the maximum of those four blocks or it can be four blocks or nine blocks or like Whatever you can fiddle around with sort of how big the pool is there. The idea is that you take A bunch of pixels to get you pull them together And then you select just one from that pool by some kind of selection criteria So instead of passing all four numbers onto the next round You take that one number and that's the only thing that moves on.