Fri, Apr 19, 2024The llama.cpp developers and maintainers are rocking and rolling. Yesterday at night, I updated the code, and it’s so fast. I changed the backend from CUBLAS to CUDA, and it’s much faster. Great job, guys!Next - Previous