Fetching the paper…

Linear attention is (maybe) all you need (to understand transformer optimization) · Around