Show HN: Run Llama3.1 405B on a 8GB VRAM challenge [video]

Not fast, not a good fit for realtime interaction applications, but for offline data processing cases, it works perfectly. I use it for my offline tasks.