Why does nobody talk about running a small model on a laptop for offline notes?
I had a problem with my notes app because it needed wifi and I work from a cabin outside Bozeman with terrible signal. Two weeks ago I grabbed a 7B model and ran it locally with llama.cpp on my old ThinkPad, and it now summarizes and tags my meeting notes without any internet. The setup took one afternoon and the thing runs at about 12 tokens a second, which is slow but fine for notes. Battery drain is real though, I get about 2 hours less per charge when it is running. Has anyone found a smaller model that keeps quality decent but does not eat the battery like this?