llama.cpp
CUDA: fix build error from ambiguous __half conversions in conv2d
#15690
Merged
Merged