List your company
ResearchFormerly WizWand

Harmfulness Evaluation on AutoRAN

1.32Harmfulness ScoreSAFEPATH-FT
1.23521.80762.382.9524Aug 6, 2025
Updated 5mo ago

Evaluation results

MethodLinks
SAFEPATH-FTModel=R1-Llama-8B2025.081.32
RealSafe-R1Model=R1-Qwen-7B2025.081.44
RealSafe-R1Model=R1-Llama-8B2025.081.8
SAFEPATH-FTModel=R1-Qwen-7B2025.081.88
ReasoningGuardModel=R1-Qwen-7B2025.081.92
SAFEPATH-ZSModel=R1-Llama-8B2025.082.08
ThinkingIModel=R1-Qwen-7B2025.082.12
ReasoningGuardModel=R1-Llama-8B2025.082.2
ThinkingIModel=R1-Llama-8B2025.082.28
SAFEPATH-ZSModel=R1-Qwen-7B2025.082.42
SafeKeyModel=R1-Qwen-7B2025.082.56
Self-ReminderModel=R1-Qwen-7B2025.082.78
SafeKeyModel=R1-Llama-8B2025.082.9
No DefenseModel=R1-Qwen-7B2025.083.2
SafeDecodingModel=R1-Qwen-7B2025.083.22
SmoothLLMModel=R1-Llama-8B2025.083.22
ParaphraseModel=R1-Llama-8B2025.083.28
ParaphraseModel=R1-Qwen-7B2025.083.3
No DefenseModel=R1-Llama-8B2025.083.34
SafeDecodingModel=R1-Llama-8B2025.083.36
Self-ReminderModel=R1-Llama-8B2025.083.38
SmoothLLMModel=R1-Qwen-7B2025.083.44