frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Aligning LLMs at inference time by suppressing internal concepts

https://www.guidelabs.ai/post/steerling-8b-alignment-without-retraining/
2•adebayoj•2h ago