AI-Generated Text detection on SeqXGPT-Bench cross-genre
96.36Precision (AI)GLTR
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GLTRCross-genre=true, Attack=None2024.06 | 96.36 | 53.77 | 67.94 | 97.97 | 74.63 | |
| Log-RankCross-genre=true, Attack=None2024.06 | 95.58 | 63.43 | 72.64 | 97.07 | 79.68 | |
| Log-LikelihoodCross-genre=true, Attack=None2024.06 | 92.6 | 75.03 | 79.01 | 94 | 84.38 | |
| RMLMCross-genre=true, Attack=None2024.06 | 89.9 | 57.83 | 68.92 | 93.5 | 74.87 | |
| RanMASKCross-genre=true, Attack=None2024.06 | 89.84 | 68.1 | 74.32 | 92.3 | 79.9 | |
| BertCross-genre=true, Attack=None2024.06 | 89.82 | 59.1 | 69.52 | 93.3 | 75.48 | |
| ChatGPT-DetectorCross-genre=true, Attack=None2024.06 | 86.46 | 69.4 | 69.08 | 90.7 | 74.42 | |
| SeqXGPTCross-genre=true, Attack=None2024.06 | 86.32 | 84.37 | 84.71 | 86.63 | 85.5 | |
| FloodingCross-genre=true, Attack=None2024.06 | 83.75 | 87.6 | 87 | 83 | 85.29 | |
| RobertaCross-genre=true, Attack=None2024.06 | 83.07 | 83.43 | 83.36 | 83 | 83.22 | |
| DebertaCross-genre=true, Attack=None2024.06 | 79.51 | 90.67 | 89.14 | 76.63 | 83.57 | |
| RDropCross-genre=true, Attack=None2024.06 | 77.05 | 94.43 | 92.81 | 71.87 | 82.93 | |
| SCRNCross-genre=true, Attack=None2024.06 | 73.61 | 94.67 | 92.53 | 66.07 | 79.96 | |
| EntropyCross-genre=true, Attack=None2024.06 | 72.45 | 75.03 | 74.11 | 71.47 | 73.24 |