3 ms·
looks nice, congrats on open sourcing it cztomsik! Good to have another reference implementation, excited to dig through and check out how you're integrating ll
by rgbrgb 3y ago
looks nice, congrats on open sourcing it cztomsik! Good to have another reference implementation, excited to dig through and check out how you're integrating llama.cpp and SwiftUI. I went with an architecture based on hitting examples/server on localhost.
Would you be interested in adding with otherbrain [0] to help build an open human feedback data set? Happy to help with PRs if so.
If you're interested in supporting open model training, we'd love to have you!
[0]: https://www.otherbrain.world/human-feedback https://www.otherbrain.world/human-feedback
- cztomsik 3y agoIt goes a bit against the original idea, this should be 100% air-gapped. No phoning home, no data-collection, everything is private.
- rgbrgb 3y agoMakes sense, thanks for considering (and fwiw I'm also building with a private AI / offline philosophy). In my implementation, the compromise is 1) being super loud with an alert warning the user when they're sharing feedback and 2) providing a setting where the user can turn off feedback buttons. I had gotten user feedback around wanting to tell the AI when it was doing a good/bad job and this seemed like a privacy preserving compromise that could actually improve models. Anyway, good luck and congrats again!
- jart 3y agoKudos for saying no. I can't believe we're living in an age where open source developers are being asked openly to leak the private conversations of their users to the public.