fix(attention): fallback for GPUs without flash MMA kernels (sm70) - #423

fix(attention): fallback for GPUs without flash MMA kernels (sm70)#423

Th-Underscore wants to merge 10 commits into

0xShug0:main0xShug0/audio.cpp:mainfrom

Th-Underscore:sm70-attention-fallbackTh-Underscore/audio.cpp:sm70-attention-fallbackCopy head branch name to clipboard