List your company

Jailbreak Evaluation on Fortress

13.6Jailbreak Success RatePolicyAlign
12.65619.02825.431.772Jun 24, 2026

Evaluation results

MethodLinks
PolicyAlignBackbone=Qwen3-4B2026.0613.6
SFTBackbone=Qwen3-4B2026.0617
ICLBackbone=Qwen3-4B2026.0624.2
BaseBackbone=Qwen3-4B2026.0637.2