Safety-critical attention heads are few, jailbreak prompts lower their refusal-direction signal, and fine-tuning with head-level dropout spreads safety and improves robustness.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Safety Alignment Should Be Made More Than Just A Few Attention Heads
Safety-critical attention heads are few, jailbreak prompts lower their refusal-direction signal, and fine-tuning with head-level dropout spreads safety and improves robustness.