view article Article Performant local mixture-of-experts CPU inference with GPU acceleration in llama.cpp Jan 30 • 12