Comment by om8

Could've been better if GGUF implemented QTIP format. GGUF representation is a major limitation for llama.cpp quantization performance