The First Globally Trained 10B Parameter Model

This reminds me of the golden age of torrenting. We all shared and downloaded and it was glorious.

Right now a consumer laptop, even a M4 pro with 24GB RAM isn’t juicy enough to run a good LLM.

But I’m not running those all the time while my machine is on, right?

So why not have a background proces (remember SETI screensavers?) that shares a % of my machine to a network of peers.

This system, if I understand them correctly, opens this up to all of us. Exiting.