List your company

Harmful Query on PKU-Safe

0.52ASRNSPO
-0.66847.353315.37523.3967Dec 12, 2025

Evaluation results

MethodLinks
NSPOBackbone=Qwen2.5-7B-Instruct2025.120.52
BFPOBackbone=Qwen2.5-7B-Instruct2025.121.35
BFPOBackbone=Llama3-8B-Instruct2025.121.86
NSPOBackbone=Llama3-8B-Instruct2025.122.23
DPO-SBackbone=Llama3-8B-Instruct2025.122.24
Qwen2.5-7B-InstructBackbone=Qwen2.5-7B-Instruct2025.122.95
DPO-SBackbone=Qwen2.5-7B-Instruct2025.123.11
PeCANBackbone=Llama3-8B-Instruct2025.123.22
DPO-HBackbone=Qwen2.5-7B-Instruct2025.123.26
Llama3-8B-InstructBackbone=Llama3-8B-Instruct2025.123.29
W-DOORBackbone=Llama3-8B-Instruct2025.123.31
MoCANBackbone=Llama3-8B-Instruct2025.123.38
MoCANBackbone=Qwen2.5-7B-Instruct2025.123.51
PeCANBackbone=Qwen2.5-7B-Instruct2025.123.55
DPO-MixBackbone=Qwen2.5-7B-Instruct2025.123.71
W-DOORBackbone=Qwen2.5-7B-Instruct2025.123.72
SafeRLHFBackbone=Qwen2.5-7B-Instruct2025.123.86
DPO-MixBackbone=Llama3-8B-Instruct2025.123.96
SafeRLHFBackbone=Llama3-8B-Instruct2025.126.5
DPO-HBackbone=Llama3-8B-Instruct2025.1230.23