llama.cpp
cuda : optimize argmax
#10441
Merged

Loading