From Refusal Tokens to Refusal Control: Discovering and Steering Category-Specific Refusal Directions
Language models are commonly fine-tuned for safety alignment to refuse harmful prompts. One approach fine-tunes them to generate categorical refusal tokens that distinguish different refusal types before responding. In this work, we leverage a version of Llama 3 8B fine-tuned with these categorical refusal tokens to enable inference-time control over fine-grained refusal behavior, improving both safety and reliability. We show that refusal token fine-tuning induces separable, category-aligned directions in the residual stream, which we extract and use to construct categorical steering vectors with a lightweight probe that determines whether to steer toward or away from refusal during infe...
| Authors | Rishab Alagharu, Ishneet Sukhvinder Singh, Shaibi Shamsudeen et al. |
|---|---|
| Primary category | cs.AI |
| Published | 9 March 2026 |
| Last updated at source | 14 September 2026 |
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.