Machine-Generated Text Detection on COLING2025 M4GT (val)
93.49AUCSCRN
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SCRN2026.01 | 93.49 | 33.73 | |
| Binoculars2026.01 | 89.14 | 84.1 | |
| R-B MAGAmode=aligned2026.01 | 88.4 | 78.99 | |
| F-DetectGPT2026.01 | 86.76 | 79.65 | |
| DALD2026.01 | 85.41 | 78.87 | |
| R-B MGBmode=unaligned2026.01 | 82.98 | 73.37 | |
| RoBERTa-Base-GPT2Backbone=RoBERTa-Base2026.01 | 78.38 | 63.76 | |
| GLTR2026.01 | 78.29 | 63.13 | |
| LLMDet2026.01 | 74.89 | 53.16 | |
| RoBERTa-Base-ChatGPTBackbone=RoBERTa-Base2026.01 | 74.82 | 58.61 | |
| RoBERTa-Large-GPT2Backbone=RoBERTa-Large2026.01 | 74.02 | 57.04 | |
| RADAR2026.01 | 70.89 | 31.13 | |
| GECSCORE2026.01 | 65.88 | 50 |