Using a mac mini (2012) for LLM inference with KoboldCpp

… especially when your inference server is a potato. I have a 2012 Intel Mac mini. It is one of those machines that still works surprisingly well, provided you don't keep asking it to behave like some