llama.cpp

ggml-quants : fix avx2 iq1_s vec_dot when compiled with gcc

#5742

Merged