[MPT-30b]( https://huggingface.co/mosaicml/mpt-30b), the lastest model from Mosaic is setting benchmarks for being the current best single GPU LLM outthere. Would be really cool to see mpt-30b & mpt-30b-instruct support by vLLM