ArchiveSequencesAbout

QuestionsEventsShortformAlignment ForumAF Comments

HomeFeaturedAllTagsRecent Comments

Phu Hoang

Karma: 3

AllPosts Comments

Can we teach a model to encode a semantic feature on a chosen manifold in just three channels?

Phu Hoang27 Jul 2026 8:00 UTC

4 points

0 comments17 min readLW link