Any chance of some updated quants for IQ2_XS? (this is the sweet spot for fitting a 70b model on a single 24GB GPU)
· Sign up or log in to comment