A method to increase the speed and lower the memory footprint of existing vision transformers. - View it on GitHub
Star
975
Rank
35402