llama.cpp

651adf4b - gguf_convert_endian.py: implement byteswapping for q4_k and q6_k (#11349)