Run vLLM on one or two Nvidia DGX Spark servers - View it on GitHub