1 paper
Erik Gösche, Reza Eghbali, Florian Knoll +1
Transformer-based networks applied to image patches have achieved cutting-edge performance in many vision tasks. However, lacking the built-in bias of convolutional neural networks…