llama.cpp

75c91de6 - CUDA: add bilinear interpolation for upscale (#14563)