refactor(attention): move linear-attention metadata into a LinearAttnBackend - #837

refactor(attention): move linear-attention metadata into a LinearAttnBackend#837

jason-fxz merged 3 commits into

mainFlashML-org/FreeToken:mainfrom

refactor/linear-attn-backendFlashML-org/FreeToken:refactor/linear-attn-backendCopy head branch name to clipboard