llama.cpp

correction of the attn.v.weight quantization for IQ3_XS

#6209

Merged