Detection of LLM-Generated Text on WritingPrompts GPT-J-6B
97.6AUROCSurpMark ref-attack
Evaluation results
| Method | Links | |
|---|---|---|
| SurpMark ref-attackAdversarial Attack Type=Original2025.10 | 97.6 | |
| SurpMark test-attackAdversarial Attack Type=Original2025.10 | 97.6 | |
| SurpMark both-attackAdversarial Attack Type=Original2025.10 | 97.6 | |
| Lastde++Adversarial Attack Type=Original2025.10 | 96.96 | |
| SurpMark test-attackAdversarial Attack Type=Attack 12025.10 | 95.62 | |
| SurpMark ref-attackAdversarial Attack Type=Attack 12025.10 | 95.06 | |
| SurpMark ref-attackAdversarial Attack Type=Attack 22025.10 | 94.67 | |
| SurpMark both-attackAdversarial Attack Type=Attack 12025.10 | 94.3 | |
| Fast-DetectGPTAdversarial Attack Type=Original2025.10 | 93.8 | |
| SurpMark both-attackAdversarial Attack Type=Attack 22025.10 | 92.74 | |
| SurpMark test-attackAdversarial Attack Type=Attack 22025.10 | 92.59 | |
| Fast-DetectGPTAdversarial Attack Type=Attack 12025.10 | 85.95 | |
| Lastde++Adversarial Attack Type=Attack 22025.10 | 85.42 | |
| Lastde++Adversarial Attack Type=Attack 12025.10 | 84.24 | |
| Fast-DetectGPTAdversarial Attack Type=Attack 22025.10 | 79.26 |