paperarXivTrust 82 · PrimaryPublished 1mo agoLive · 3mo ago
Refusal as a single linear direction
Evidence that model refusals are mediated by one interpretable direction in activation space.
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- LinkedLinked via unknownHallucination →