llama.cpp

4755afd1 - llama : fix integer overflow during quantization (#6063)