vLLM on AMD GPUs and Google TPUs

This YouTube talk is overlaid with portable W3C Web Annotations connecting it to its canonical devreal.ai graph nodes.