A method to increase the speed and lower the memory footprint of existing vision transformers. - View it on GitHub
Star
950
Rank
35692