Go apps can directly integrate llama.cpp for HW accelerated local inference github.com 1 points by deadprogram 4 hours ago