llama.cpp

982e3472 - quantize : fix minor logic flaw in --tensor-type (#14572)