I have a framework 13, but I couldn't imagine running a local llm on it, how do you do it? Do you have a eGPU?