Indian Institute of Technology Bombay

Vision Transformer and models like it use a lot of computation and memory when processing images. New work modifies these architectures to run more efficiently while adopting helpful properties from convolutions.