whisper.cpp
aef59f48 - CUDA: FA support for Deepseek (Ampere or newer) (llama/13306)

Commit
234 days ago
CUDA: FA support for Deepseek (Ampere or newer) (llama/13306) * CUDA: FA support for Deepseek (Ampere or newer) * do loop unrolling via C++ template
Committer
Parents
Loading