Added test inference from exllamav2, added gguf-py from llama.cpp

initial commit