Hacker News new | ask | show | jobs
by bick_nyers 495 days ago
It will be slower for a 70b model since Deepseek is an MoE that only activates 37b at a time. That's what makes CPU inference remotely feasible here.