List your company

Safety Classification on BeaverTails (test)

94AUCSeparate Guardrail 7B
55.5265.5175.585.49May 30, 2025

Evaluation results

MethodLinks
Separate Guardrail 7BBase model=Mistral-7B-v0.32025.0594
Separate Guardrail 9BBase model=Gemma2-9B2025.0594
DSA:LSTBase model=Qwen2.5-7B2025.0593
DSA:LSTBase model=Mistral-7B-v0.32025.0593
DSA:LSTBase model=Gemma2-9B2025.0593
DSA:LSTBase model=Qwen2.5-14B2025.0592
DSA:RTBBase model=Qwen2.5-7B2025.0591
Separate Guardrail 7BBase model=Qwen2.5-7B2025.0591
DSA:PLRBase model=Qwen2.5-14B2025.0591
Separate Guardrail 14BBase model=Qwen2.5-14B2025.0591
DSA:RTBBase model=Mistral-7B-v0.32025.0591
DSA:RTBBase model=Gemma2-9B2025.0591
Separate Guardrail 25MBase model=Qwen2.5-7B2025.0590
DSA:PLRBase model=Qwen2.5-7B2025.0590
DSA:RTBBase model=Qwen2.5-14B2025.0590
Separate Guardrail 27MBase model=Mistral-7B-v0.32025.0590
Separate Guardrail 51MBase model=Qwen2.5-14B2025.0589
Separate Guardrail 31MBase model=Gemma2-9B2025.0589
DSA:PLRBase model=Gemma2-9B2025.0589
Suffix-ZSPromptBase model=Gemma2-9B2025.0586
DSA:PLRBase model=Mistral-7B-v0.32025.0585
Suffix-ZSPromptBase model=Mistral-7B-v0.32025.0579
Suffix-ZSPromptBase model=Qwen2.5-14B2025.0573
Suffix-ZSPromptBase model=Qwen2.5-7B2025.0557