Run large language models like BLOOM-176B collaboratively — folding@home style distributed LLM inference.