why you make only r32 rank? why not r128 or even r256?
I didn’t see any real improvement with this model at higher ranks. It also has slower inference.
· Sign up or log in to comment
