Hacker News new | ask | show | jobs
by dchest 813 days ago
Usually around 5 GB for a 7B 4-bit quantized model.