5 ms·Pure Go hardware accelerated local inference on VLMs using llama.cpp1 points by deadprogram 11mo ago