claim
active
claim:understanding-model-internals-is-crucial-for-identifying-shortcomings-and-guiding-safety-research-steering-can-aid-auditing-and-red-teamingUnderstanding model internals is crucial for identifying shortcomings and guiding safety research; steering can aid auditing and red-teaming
Safety motivation.
Source paper
extracted_fromNeighborhood — ranked by edge-count
Papers (1)
paper
Institutes (1)
institute
- Haize LabscitesRed-teaming partner used to demonstrate steering for jailbreak elicitation.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Addresses skeptical alternative that reports reflect only conversational content
- Not enough to trust medical responses without experts.
- Speculation about false positives.
- Central interpretive claim organizing the entire paper's results
- General technique of modifying activations to control model behavior.
- Practical implication drawn from the prosocial persona paradox finding
- Limitations.
- The method can steer the model in both positive and negative directions on the target semantic.