LLM-generated text detection on RAID Wikipedia-related
99.94GPT-4 ScoreSearchLLM
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SearchLLMBase detector=Binoculars2026.01 | 99.94 | 99.99 | 99.89 | 99.99 | 99.99 | |
| Binoculars2026.01 | 99.87 | 99.99 | 99.77 | 99.99 | 99.98 | |
| SearchLLMBase detector=Longformer2026.01 | 99.79 | 99.96 | 97.94 | 99.75 | 99.66 | |
| SearchLLMBase detector=RADAR2026.01 | 99.49 | 99.82 | 97.93 | 99.79 | 99.7 | |
| Longformer2026.01 | 99.32 | 99.7 | 95.05 | 99.16 | 98.95 | |
| RADAR2026.01 | 99.06 | 99.68 | 95.99 | 99.62 | 99.47 | |
| SearchLLMBase detector=Detect2026.01 | 93.79 | 95.46 | 87.93 | 92.33 | 93.99 | |
| Detect2026.01 | 90.22 | 91.67 | 76.29 | 85.42 | 88.68 |