linear attention

A variant of attention mechanisms that scales linearly with input size, improving efficiency in processing sequences without compromising on performance.

11 papers