Skip to content

perf(krea2): keep the K/V heads unexpanded where a fused kernel takes them - #174

Draft
Pfannkuchensack wants to merge 1 commit into
feat/krea2-cudnn-attentionfrom
feat/krea2-native-gqa
Draft

perf(krea2): keep the K/V heads unexpanded where a fused kernel takes them#174
Pfannkuchensack wants to merge 1 commit into
feat/krea2-cudnn-attentionfrom
feat/krea2-native-gqa