Hacker News Reader

Will this strategy work on lower memory unified memory machines like Apple Silicon or AMD?

There are solutions already for Mac , check out ISTA DAs lab hugginface, I got one of their qwen next models working pretty fast on m2 ultra