Abstract: The quadratic computation complexity of self-attention has been a persistent challenge when applying Transformer models to vision tasks. Linear attention, on the other hand, offers a much ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results