1 paper
Nicholas Foley, Devin Marinelli, Donny Moore +2
In standard attention, three separately learned projections decide how strongly a token attends to each neighbor (WQ, WK) and how the attended features are transformed before…