CUDA: mul_mat_q=true as default for llama_context_params#2912

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

JohannesGaessler:cuda-mmq-default-2JohannesGaessler/llama.cpp:cuda-mmq-default-2Copy head branch name to clipboard