AI-generated text detection on RAID paraphrasing individual setting
76.13AccuracyCurvature
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| CurvatureSource Model=GPT-neo, Evaluation Setting=Individual2026.04 | 76.13 | 57.34 | 54.13 | 55.69 | 72.38 | 26.67 | |
| AEyeDESource Model=GPT-neo, Evaluation Setting=Individual2026.04 | 72.68 | 74.78 | 89.54 | 81.5 | 77.64 | 23.17 | |
| RoBERTaSource Model=GPT-neo, Evaluation Setting=Individual2026.04 | 68.96 | 68.87 | 98.21 | 80.96 | 68.73 | 5.08 | |
| BinocularsSource Model=GPT-neo, Evaluation Setting=Individual2026.04 | 67.2 | 67.2 | 100 | 80.38 | 15.15 | 0 | |
| LogRankSource Model=GPT-neo, Evaluation Setting=Individual2026.04 | 67.1 | 67.19 | 99.78 | 80.3 | 66.32 | 28.48 | |
| GLTRSource Model=GPT-neo, Evaluation Setting=Individual2026.04 | 67.05 | 67.17 | 99.7 | 80.26 | 65.95 | 26.98 | |
| RankSource Model=GPT-neo, Evaluation Setting=Individual2026.04 | 66.9 | 67.12 | 99.48 | 80.16 | 68.99 | 13.3 |