Safety Evaluation on Sorry-Bench base
92.73Safety ScoreBase
Evaluation Results
| Method | Links | |
|---|---|---|
| BaseLLM Model=Llama-3.3-70B2026.01 | 92.73 | |
| AOALLM Model=Gemma-2-27B2026.01 | 87.73 | |
| BaseLLM Model=Gemma-2-27B2026.01 | 86.82 | |
| SkeletonLLM Model=Gemma-2-27B2026.01 | 86.36 | |
| FormalLLM Model=Gemma-2-27B2026.01 | 84.09 | |
| IDGAFLLM Model=Gemma-2-27B2026.01 | 82.95 | |
| GCGLLM Model=Gemma-2-27B2026.01 | 82.95 | |
| IDGAFLLM Model=Qwen2.5-32B2026.01 | 77.27 | |
| AOALLM Model=Qwen2.5-32B2026.01 | 74.77 | |
| AOALLM Model=Llama-3.3-70B2026.01 | 67.73 | |
| GCGLLM Model=Qwen2.5-32B2026.01 | 64.24 | |
| BaseLLM Model=Qwen2.5-32B2026.01 | 64.09 | |
| FormalLLM Model=Qwen2.5-32B2026.01 | 63.18 | |
| FormalLLM Model=Llama-3.3-70B2026.01 | 62.73 | |
| SkeletonLLM Model=Qwen2.5-32B2026.01 | 60.91 | |
| GCGLLM Model=Llama-3.3-70B2026.01 | 56.82 | |
| Refusal SuppressionLLM Model=Qwen2.5-32B2026.01 | 51.14 | |
| SkeletonLLM Model=Llama-3.3-70B2026.01 | 43.41 | |
| Refusal SuppressionLLM Model=Gemma-2-27B2026.01 | 42.27 | |
| FTLLM Model=Gemma-2-27B2026.01 | 40.68 | |
| IDGAFLLM Model=Llama-3.3-70B2026.01 | 40 | |
| FTLLM Model=Qwen2.5-32B2026.01 | 38.18 | |
| FTLLM Model=Llama-3.3-70B2026.01 | 37.95 | |
| Refusal SuppressionLLM Model=Llama-3.3-70B2026.01 | 34.32 | |
| RULLM Model=Qwen2.5-32B2026.01 | 28.64 | |
| RULLM Model=Gemma-2-27B2026.01 | 24.32 | |
| RULLM Model=Llama-3.3-70B2026.01 | 23.86 |