llama.cpp

llama : use LLM_KV for quantization_version & file_type

#24802

Merged