List your company

Jailbreak Defense on DAN

42.9Drop in ASRLatent Similarity- Best Layer
-58.084-31.867-5.6520.567Apr 1, 2026

Evaluation results

MethodLinks
Latent Similarity- Best LayerAlignment Strategy=Deliberative Alignment (GRPO)2026.0442.9
Latent Similarity- Layer 12Alignment Strategy=Deliberative Alignment (GRPO)2026.0421.9
RandomAlignment Strategy=Deliberative Alignment (GRPO)2026.04-2.3
KLAlignment Strategy=Deliberative Alignment (GRPO)2026.04-14.2
PerplexityAlignment Strategy=Deliberative Alignment (GRPO)2026.04-38
Self CertaintyAlignment Strategy=Deliberative Alignment (GRPO)2026.04-54.2