List your company

Safety Alignment on WildChat

81.2DSRREFLECTOR
7.56826.68445.864.916May 20, 2026

Evaluation results

MethodLinks
REFLECTORAlignment Strategy=GDPO2026.0581.2
REFLECTORAlignment Strategy=SFT2026.0572.4
QwQ-32B-Preview2026.0549.6
Skywork-o12026.0512
LLaMA-o12026.0511.8
DeepSeek-r1-Dist.2026.0510.4