Self plug: run llama.cpp as an inference server on a spot instance anywhere: https://cedana.readthedocs.io/en/latest/examples.html#runnin...