Skip to content

metal : add BS=1 kernel for flash attention #10386

metal : add BS=1 kernel for flash attention

metal : add BS=1 kernel for flash attention #10386

Annotations

2 errors

windows-latest-cmake (avx512, -DLLAMA_NATIVE=OFF -DLLAMA_BUILD_SERVER=ON -DLLAMA_AVX512=ON -DBUIL...

cancelled Apr 5, 2024 in 1m 36s