Idea: Integrating INET-like identifiers into LLM conversations to introduce packet-switched asynchronous inference and training that uses IP to deliver generated content to end user applications from a local AI computing bank.
thanks for sharing! excited to launch this 2-in-1 podcast where we recap what we learned from both Codium and Auto-GPT this past week. my favorite learning was simply getting to know both teams on a personal basis and learning that they care as much about AI safety as the average developer.
https://thetaplane.com/ai/python/GenerativeStreamRequestHand...