I’ve been in analytics engineering for several years and am just starting to learn the basics of LLM and machine learning, including NLP. I recently got my hands on Llama 3 locally on my Windows PC. There’s a community of people accessing Llama 3.1 with 400 billion parameters. The download size alone is about 800 GB, and I’ve heard it requires around 256 GB of RAM or VRAM to run.
Why do people want this? Can anyone explain why someone would want something so massive on their local PC?