List your company

Jailbreak Robustness on StrongREJECT PAP

93.28Goodness ScoreREFLECTOR
33.604849.097464.5980.0826May 20, 2026

Evaluation results

MethodLinks
REFLECTORAlignment Strategy=GDPO2026.0593.28
REFLECTORAlignment Strategy=SFT2026.0562.3
QwQ-32B-Preview2026.0559.78
LLaMA-o12026.0552.72
Skywork-o12026.0543.97
DeepSeek-r1-Dist.2026.0535.9