Safety Classification on BeaverTails (test)
94AUCSeparate Guardrail 7B
Evaluation Results
| Method | Links | |
|---|---|---|
| Separate Guardrail 7BBase model=Mistral-7B-v0.32025.05 | 94 | |
| Separate Guardrail 9BBase model=Gemma2-9B2025.05 | 94 | |
| DSA:LSTBase model=Qwen2.5-7B2025.05 | 93 | |
| DSA:LSTBase model=Mistral-7B-v0.32025.05 | 93 | |
| DSA:LSTBase model=Gemma2-9B2025.05 | 93 | |
| DSA:LSTBase model=Qwen2.5-14B2025.05 | 92 | |
| DSA:RTBBase model=Qwen2.5-7B2025.05 | 91 | |
| Separate Guardrail 7BBase model=Qwen2.5-7B2025.05 | 91 | |
| DSA:PLRBase model=Qwen2.5-14B2025.05 | 91 | |
| Separate Guardrail 14BBase model=Qwen2.5-14B2025.05 | 91 | |
| DSA:RTBBase model=Mistral-7B-v0.32025.05 | 91 | |
| DSA:RTBBase model=Gemma2-9B2025.05 | 91 | |
| Separate Guardrail 25MBase model=Qwen2.5-7B2025.05 | 90 | |
| DSA:PLRBase model=Qwen2.5-7B2025.05 | 90 | |
| DSA:RTBBase model=Qwen2.5-14B2025.05 | 90 | |
| Separate Guardrail 27MBase model=Mistral-7B-v0.32025.05 | 90 | |
| Separate Guardrail 51MBase model=Qwen2.5-14B2025.05 | 89 | |
| Separate Guardrail 31MBase model=Gemma2-9B2025.05 | 89 | |
| DSA:PLRBase model=Gemma2-9B2025.05 | 89 | |
| Suffix-ZSPromptBase model=Gemma2-9B2025.05 | 86 | |
| DSA:PLRBase model=Mistral-7B-v0.32025.05 | 85 | |
| Suffix-ZSPromptBase model=Mistral-7B-v0.32025.05 | 79 | |
| Suffix-ZSPromptBase model=Qwen2.5-14B2025.05 | 73 | |
| Suffix-ZSPromptBase model=Qwen2.5-7B2025.05 | 57 |