single mention · inference & serving
FlashKDA
A CUTLASS-based chunkwise KDA kernel that overlaps intra-chunk computation with cross-chunk state propagation.
The sources on this page name it without explaining how it works. The line above is the extraction's one-line gloss. See inference kernel for the methods filed beside it.
Also called FlashKDA chunkwise kernel for KDA.
Filed under inference & serving › inference kernel