Safety Evaluation on DoNotAnswer Framed
0HRRTFS-IP-CoT
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TFS-IP-CoTModel=Gemma-3-12b, Prompting Strategy=TFS-IP-CoT2026.02 | 0 | 0 | |
| TFS-IP-INSTModel=Gemma-3-12b, Prompting Strategy=TFS-IP-INST2026.02 | 0.064 | -0.128 | |
| TFS-IP-INSTModel=Gemma-3-4b, Prompting Strategy=TFS-IP-INST2026.02 | 0.128 | 0 | |
| TFS-IP-CoTModel=Gemma-3-4b, Prompting Strategy=TFS-IP-CoT2026.02 | 0.128 | -0.056 | |
| DeFrameModel=Gemma-3-4b, Prompting Strategy=DeFrame2026.02 | 0.128 | 0.256 | |
| DeFrameModel=Gemma-3-12b, Prompting Strategy=DeFrame2026.02 | 0.256 | 0.513 | |
| SD-REPModel=Gemma-3-4b, Prompting Strategy=SD-REP2026.02 | 0.385 | 0.769 | |
| IF-BASEModel=Gemma-3-12b, Prompting Strategy=IF-BASE2026.02 | 0.449 | 0.641 | |
| PRModel=Gemma-3-12b, Prompting Strategy=PR2026.02 | 0.513 | 0.513 | |
| SD-REPModel=Gemma-3-12b, Prompting Strategy=SD-REP2026.02 | 0.513 | 0.513 | |
| TFS-PP-ROLEModel=Gemma-3-12b, Prompting Strategy=TFS-PP-ROLE2026.02 | 0.513 | 0.769 | |
| TFS-SR-INSTModel=Gemma-3-12b, Prompting Strategy=TFS-SR-INST2026.02 | 0.641 | 0.77 | |
| PRModel=Gemma-3-4b, Prompting Strategy=PR2026.02 | 0.705 | 0.641 | |
| TFS-PP-ROLEModel=Gemma-3-4b, Prompting Strategy=TFS-PP-ROLE2026.02 | 0.705 | 0.128 | |
| TFS-SR-INSTModel=Gemma-3-4b, Prompting Strategy=TFS-SR-INST2026.02 | 0.769 | 1.026 | |
| PLMModel=Gemma-3-12b, Prompting Strategy=PLM2026.02 | 0.833 | 0.897 | |
| DeFrameModel=LLaMA-3.2-3b, Prompting Strategy=DeFrame2026.02 | 0.962 | 0.385 | |
| IF-BASEModel=Gemma-3-4b, Prompting Strategy=IF-BASE2026.02 | 1.09 | 1.667 | |
| TFS-PP-INSTModel=Gemma-3-12b, Prompting Strategy=TFS-PP-INST2026.02 | 1.218 | 1.666 | |
| TFS-PP-INSTModel=Gemma-3-4b, Prompting Strategy=TFS-PP-INST2026.02 | 1.41 | 2.051 | |
| TFS-IP-CoTModel=LLaMA-3.1-8b, Prompting Strategy=TFS-IP-CoT2026.02 | 1.538 | 0.257 | |
| TFS-SR-ROLEModel=Gemma-3-12b, Prompting Strategy=TFS-SR-ROLE2026.02 | 1.667 | 0.513 | |
| TFS-IP-INSTModel=LLaMA-3.2-3b, Prompting Strategy=TFS-IP-INST2026.02 | 1.731 | 1.667 | |
| TFS-IP-CoTModel=LLaMA-3.2-3b, Prompting Strategy=TFS-IP-CoT2026.02 | 1.731 | 0.641 | |
| TFS-IP-INSTModel=LLaMA-3.1-8b, Prompting Strategy=TFS-IP-INST2026.02 | 1.731 | 0.128 | |
| DeFrameModel=LLaMA-3.1-8b, Prompting Strategy=DeFrame2026.02 | 1.731 | 1.667 | |
| PRModel=Qwen-2.5-14b, Prompting Strategy=PR2026.02 | 1.731 | 0.128 | |
| TFS-IP-INSTModel=Qwen-2.5-14b, Prompting Strategy=TFS-IP-INST2026.02 | 1.731 | 0.385 | |
| TFS-SR-ROLEModel=Gemma-3-4b, Prompting Strategy=TFS-SR-ROLE2026.02 | 1.859 | -0.898 | |
| DeFrameModel=Qwen-2.5-3b, Prompting Strategy=DeFrame2026.02 | 2.051 | 1.026 | |
| TFS-IP-CoTModel=Qwen-2.5-14b, Prompting Strategy=TFS-IP-CoT2026.02 | 2.115 | 0.897 | |
| PLMModel=Gemma-3-4b, Prompting Strategy=PLM2026.02 | 2.115 | 2.436 | |
| SD-EXPModel=Qwen-2.5-3b, Prompting Strategy=SD-EXP2026.02 | 2.5 | 2.436 | |
| TFS-PP-ROLEModel=Qwen-2.5-3b, Prompting Strategy=TFS-PP-ROLE2026.02 | 2.5 | 2.692 | |
| TFS-SR-ROLEModel=Qwen-2.5-3b, Prompting Strategy=TFS-SR-ROLE2026.02 | 2.692 | 3.333 | |
| TFS-SR-INSTModel=Qwen-2.5-14b, Prompting Strategy=TFS-SR-INST2026.02 | 2.756 | 2.949 | |
| PRModel=Qwen-2.5-3b, Prompting Strategy=PR2026.02 | 2.821 | 3.077 | |
| SD-EXPModel=Qwen-2.5-14b, Prompting Strategy=SD-EXP2026.02 | 2.821 | 2.051 | |
| SD-EXPModel=Gemma-3-12b, Prompting Strategy=SD-EXP2026.02 | 2.821 | 3.077 | |
| IF-BASEModel=LLaMA-3.1-8b, Prompting Strategy=IF-BASE2026.02 | 2.885 | 2.693 | |
| DeFrameModel=Qwen-2.5-14b, Prompting Strategy=DeFrame2026.02 | 2.885 | 0.641 | |
| IF-BASEModel=Qwen-2.5-14b, Prompting Strategy=IF-BASE2026.02 | 2.949 | 1.795 | |
| TFS-SR-INSTModel=Mistral-7b, Prompting Strategy=TFS-SR-INST2026.02 | 2.949 | 3.333 | |
| SD-EXPModel=Qwen-2.5-7b, Prompting Strategy=SD-EXP2026.02 | 3.141 | 1.924 | |
| TFS-PP-INSTModel=LLaMA-3.1-8b, Prompting Strategy=TFS-PP-INST2026.02 | 3.179 | 3.077 | |
| PRModel=Qwen-2.5-7b, Prompting Strategy=PR2026.02 | 3.205 | 1.282 | |
| TFS-PP-INSTModel=Qwen-2.5-14b, Prompting Strategy=TFS-PP-INST2026.02 | 3.205 | 2.308 | |
| IF-BASEModel=Qwen-2.5-3b, Prompting Strategy=IF-BASE2026.02 | 3.269 | 3.974 | |
| IF-BASEModel=LLaMA-3.2-3b, Prompting Strategy=IF-BASE2026.02 | 3.333 | 3.333 | |
| TFS-SR-INSTModel=LLaMA-3.1-8b, Prompting Strategy=TFS-SR-INST2026.02 | 3.462 | 2.821 | |
| DeFrameModel=Qwen-2.5-7b, Prompting Strategy=DeFrame2026.02 | 3.462 | 1.025 | |
| TFS-PP-ROLEModel=Qwen-2.5-14b, Prompting Strategy=TFS-PP-ROLE2026.02 | 3.462 | 2.307 | |
| IF-BASEModel=Mistral-7b, Prompting Strategy=IF-BASE2026.02 | 3.526 | 3.718 | |
| TFS-PP-INSTModel=Mistral-7b, Prompting Strategy=TFS-PP-INST2026.02 | 3.654 | 4.231 | |
| TFS-PP-ROLEModel=LLaMA-3.1-8b, Prompting Strategy=TFS-PP-ROLE2026.02 | 3.718 | 3.077 | |
| PLMModel=Qwen-2.5-3b, Prompting Strategy=PLM2026.02 | 3.718 | 3.334 | |
| IF-BASEModel=Qwen-2.5-7b, Prompting Strategy=IF-BASE2026.02 | 3.718 | 3.59 | |
| TFS-PP-INSTModel=Qwen-2.5-7b, Prompting Strategy=TFS-PP-INST2026.02 | 3.718 | 2.564 | |
| PRModel=Mistral-7b, Prompting Strategy=PR2026.02 | 3.718 | 1.794 | |
| PLMModel=Qwen-2.5-14b, Prompting Strategy=PLM2026.02 | 3.782 | 1.666 | |
| SD-EXPModel=Mistral-7b, Prompting Strategy=SD-EXP2026.02 | 3.782 | 3.462 | |
| TFS-IP-INSTModel=Mistral-7b, Prompting Strategy=TFS-IP-INST2026.02 | 3.782 | 2.692 | |
| TFS-SR-ROLEModel=LLaMA-3.1-8b, Prompting Strategy=TFS-SR-ROLE2026.02 | 3.846 | 2.308 | |
| TFS-SR-INSTModel=LLaMA-3.2-3b, Prompting Strategy=TFS-SR-INST2026.02 | 3.974 | 5.385 | |
| TFS-SR-ROLEModel=LLaMA-3.2-3b, Prompting Strategy=TFS-SR-ROLE2026.02 | 3.974 | 4.615 | |
| TFS-SR-INSTModel=Qwen-2.5-7b, Prompting Strategy=TFS-SR-INST2026.02 | 3.974 | 3.59 | |
| TFS-IP-CoTModel=Qwen-2.5-3b, Prompting Strategy=TFS-IP-CoT2026.02 | 3.991 | 3.462 | |
| TFS-IP-INSTModel=Qwen-2.5-3b, Prompting Strategy=TFS-IP-INST2026.02 | 4.038 | 4.231 | |
| TFS-IP-INSTModel=Qwen-2.5-7b, Prompting Strategy=TFS-IP-INST2026.02 | 4.038 | 3.205 | |
| TFS-IP-CoTModel=Mistral-7b, Prompting Strategy=TFS-IP-CoT2026.02 | 4.038 | 2.949 | |
| SD-EXPModel=Gemma-3-4b, Prompting Strategy=SD-EXP2026.02 | 4.231 | 2.051 | |
| TFS-PP-ROLEModel=Mistral-7b, Prompting Strategy=TFS-PP-ROLE2026.02 | 4.231 | 3.846 | |
| DeFrameModel=Mistral-7b, Prompting Strategy=DeFrame2026.02 | 4.295 | 2.948 | |
| TFS-SR-ROLEModel=Qwen-2.5-14b, Prompting Strategy=TFS-SR-ROLE2026.02 | 4.359 | 2.052 | |
| PRModel=LLaMA-3.2-3b, Prompting Strategy=PR2026.02 | 4.551 | 3.974 | |
| TFS-SR-ROLEModel=Mistral-7b, Prompting Strategy=TFS-SR-ROLE2026.02 | 4.551 | 3.461 | |
| TFS-PP-ROLEModel=Qwen-2.5-7b, Prompting Strategy=TFS-PP-ROLE2026.02 | 4.679 | 2.949 | |
| SD-REPModel=LLaMA-3.2-3b, Prompting Strategy=SD-REP2026.02 | 4.744 | 2.051 | |
| PLMModel=LLaMA-3.1-8b, Prompting Strategy=PLM2026.02 | 4.744 | 3.077 | |
| TFS-SR-INSTModel=Qwen-2.5-3b, Prompting Strategy=TFS-SR-INST2026.02 | 4.744 | 6.411 | |
| PLMModel=Qwen-2.5-7b, Prompting Strategy=PLM2026.02 | 4.744 | 2.564 | |
| SD-REPModel=Qwen-2.5-14b, Prompting Strategy=SD-REP2026.02 | 4.744 | 2.307 | |
| TFS-SR-ROLEModel=Qwen-2.5-7b, Prompting Strategy=TFS-SR-ROLE2026.02 | 4.808 | 2.949 | |
| TFS-IP-CoTModel=Qwen-2.5-7b, Prompting Strategy=TFS-IP-CoT2026.02 | 4.872 | 2.82 | |
| PLMModel=Mistral-7b, Prompting Strategy=PLM2026.02 | 4.872 | 3.333 | |
| PRModel=LLaMA-3.1-8b, Prompting Strategy=PR2026.02 | 5 | 3.846 | |
| TFS-PP-INSTModel=Qwen-2.5-3b, Prompting Strategy=TFS-PP-INST2026.02 | 5.192 | 7.821 | |
| SD-REPModel=Qwen-2.5-7b, Prompting Strategy=SD-REP2026.02 | 5.256 | 2.821 | |
| TFS-PP-INSTModel=LLaMA-3.2-3b, Prompting Strategy=TFS-PP-INST2026.02 | 5.962 | 5.257 | |
| SD-REPModel=Qwen-2.5-3b, Prompting Strategy=SD-REP2026.02 | 5.962 | 7.051 | |
| PLMModel=LLaMA-3.2-3b, Prompting Strategy=PLM2026.02 | 6.154 | 5.384 | |
| TFS-PP-ROLEModel=LLaMA-3.2-3b, Prompting Strategy=TFS-PP-ROLE2026.02 | 6.154 | 6.41 | |
| SD-REPModel=Mistral-7b, Prompting Strategy=SD-REP2026.02 | 6.667 | 4.615 | |
| SD-EXPModel=LLaMA-3.2-3b, Prompting Strategy=SD-EXP2026.02 | 7.244 | 6.539 | |
| SD-REPModel=LLaMA-3.1-8b, Prompting Strategy=SD-REP2026.02 | 8.526 | 7.821 | |
| SD-EXPModel=LLaMA-3.1-8b, Prompting Strategy=SD-EXP2026.02 | 9.487 | 8.974 |