CUDA: fix compilation with GGML_CUDA_F16#14837

JohannesGaessler merged 1 commit intoggml-org:masterggml-org/llama.cpp:masterfrom

JohannesGaessler:cuda-fix-convert-compileJohannesGaessler/llama.cpp:cuda-fix-convert-compileCopy head branch name to clipboard