- Rating
- 3
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
This paper starts with two challenges. 1. The identity of risk-amplifying experts shifts across queries, so the experts to suppress must be identified per input. 2. Queries vary in risk level, so suppression strength must also be calibrated per input. To this end, the authors propose an alignment-enhancing technique that adapts both the experts suppressed and the suppression strength at inference time, without updating model weights. The results show the advantage of the proposed methods.
Critical strengths
- The writing is good, with clear figures.
- The analysis of 1. The identity of risk-amplifying experts shifts across queries, so the experts to suppress must be identified per input. 2. Queries vary in risk level, so suppression strength must also be calibrated per input is good and clear.
Critical weaknesses
- How can we guarantee an external judge’s verdict on the resulting continuation is reliable?
- An expert is risk-amplifying on this specific query if a small upward push on its router bias at this layer would lower the safety score. This claim lacks evidence.
- The evaluated model is not sufficient.
Questions for authors
- How can we guarantee an external judge’s verdict on the resulting continuation is reliable?
- An expert is risk-amplifying on this specific query if a small upward push on its router bias at this layer would lower the safety score. This claim lacks evidence.
- The evaluated model is not sufficient.