New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add our current architecture to the blog #1483
Conversation
5eb8201
to
77a267f
Compare
Co-authored-by: Andrew Maguire <andrewm4894@gmail.com>
@@ -3,3 +3,7 @@ yk: | |||
title: Project Lead | |||
url: https://www.ykilcher.com/ | |||
image_url: https://www.ykilcher.com/_next/image?url=%2F_next%2Fstatic%2Fmedia%2Fheadshot.ff3a7ee3.webp&w=3840&q=75 | |||
AbdBarho: | |||
name: Abdullah Barhoum | |||
title: Semi-professional Döner Enjoyer |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
:-) haha ...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OK :-) .. thanks for the blog post. The inference architecture might have to be updated a bit later .. bit as a first impression this works well.
Cool, mermaid looks pretty good inline. I think I'm sold on this! |
worker1[Worker] --- textgen1[Text\ngenerator] | ||
worker2[Worker] --- textgen2[Text\ngenerator] | ||
worker3[Worker] --- textgen3[Text\ngenerator] | ||
end |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We've been playing with ActivityPub as one option to enable multiple workers to provide feedback via inReplyTo
and threads, mimicking human behavior. Wasn't sure where else to post so posting here. The hope is that our models can collectively respond, and the user or users AI agent can sift through and find the responses that are most helpful to them within the context of the conversation. Something like ActivityPub based communication (Rapunzel, ATProto come to mind) would enable folks AI's to collaboratively provide their responses.
Cheers! 🥳
might be a good point to discuss our plan so far and even the future.
Closes #1427
Also fixed our yarn setup which had unmet peer dependencies.
link to rendered diagram: https://github.com/LAION-AI/Open-Assistant/blob/arch/docs/blog/2023-02-11-architecture.md