Spoofing Attack Robustness on BookSum
0.9552AUCDualGuard
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DualGuardBase LLM=OPT-1.3B2025.12 | 0.9552 | 77.84 | 86.93 | 0.9656 | |
| DualGuardBase LLM=Llama3.1-8B-Instruct2025.12 | 0.9354 | 56.55 | 74.48 | 0.9303 | |
| DIPmarkBase LLM=OPT-1.3B2025.12 | 0.5375 | 9.73 | 14.59 | 0.5363 | |
| AARBase LLM=OPT-1.3B2025.12 | 0.5308 | 9.77 | 17.82 | 0.6571 | |
| SWEETBase LLM=OPT-1.3B2025.12 | 0.5136 | 7.98 | 11.17 | 0.7492 | |
| AARBase LLM=Llama3.1-8B-Instruct2025.12 | 0.5073 | 4.03 | 10.74 | 0.6201 | |
| DIPmarkBase LLM=Llama3.1-8B-Instruct2025.12 | 0.5006 | 8.45 | 11.97 | 0.5259 | |
| SynthIDBase LLM=Llama3.1-8B-Instruct2025.12 | 0.4941 | 3.9 | 6.49 | 0.5813 | |
| XSIRBase LLM=Llama3.1-8B-Instruct2025.12 | 0.4921 | 6.45 | 9.03 | 0.6815 | |
| UnbiasedBase LLM=Llama3.1-8B-Instruct2025.12 | 0.4914 | 4.84 | 9.68 | 0.501 | |
| KGWBase LLM=Llama3.1-8B-Instruct2025.12 | 0.4842 | 6.88 | 8.13 | 0.692 | |
| EWDBase LLM=OPT-1.3B2025.12 | 0.4733 | 2.78 | 6.67 | 0.7277 | |
| UnbiasedBase LLM=OPT-1.3B2025.12 | 0.4729 | 4.55 | 9.11 | 0.4945 | |
| KGWBase LLM=OPT-1.3B2025.12 | 0.4613 | 6.11 | 8.33 | 0.7195 | |
| SynthIDBase LLM=OPT-1.3B2025.12 | 0.45 | 5.43 | 13.04 | 0.6029 | |
| EWDBase LLM=Llama3.1-8B-Instruct2025.12 | 0.4354 | 4.96 | 7.8 | 0.6812 | |
| SWEETBase LLM=Llama3.1-8B-Instruct2025.12 | 0.4269 | 2 | 6.67 | 0.6654 | |
| SIRBase LLM=OPT-1.3B2025.12 | 0.419 | 3.3 | 6.59 | 0.6748 | |
| XSIRBase LLM=OPT-1.3B2025.12 | 0.3882 | 1.08 | 6.49 | 0.6741 | |
| SIRBase LLM=Llama3.1-8B-Instruct2025.12 | 0.3378 | 3.12 | 5.63 | 0.5702 |