Is the Attention mechanism (layer) implemented in the dnn module
Wanted to check if the Attentionlayer operation implemented as a layer in the dnn module? Link to the "Attention is all you Need" paper https://papers.nips.cc/paper/7181-att...
do you know any computer-vision related model using it ?
(imho it's more used for text, like in the mentioned paper)
The paper uses this attention mechanism for language task but it seems like it can be used for vision tasks as well.