Comment by abhimskywalker

"The recent change also means you can run multiple LLaMA ./main processes at the same time, and they'll all share the same memory resources." So this could have a main and multiple sub-worker llm processes possibly collaborating while sharing same memory footprint?