Machine-Generated Text Detection on HellaSwag In-Domain (IND-MGT)
96.3AccuracyDEER
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DEERBackbone=RoBERTa-base, Detection Methodology Category=Domain generalization methods2025.11 | 96.3 | 96.35 | |
| MGT-PrismBackbone=RoBERTa-base, Detection Methodology Category=Domain generalization methods2025.11 | 96.28 | 96.23 | |
| PeCoLABackbone=RoBERTa-base, Detection Methodology Category=Model-based methods2025.11 | 96.04 | 96.02 | |
| MSCLBackbone=RoBERTa-base, Detection Methodology Category=Domain generalization methods2025.11 | 95.87 | 95.76 | |
| EAGLEBackbone=RoBERTa-base, Detection Methodology Category=Model-based methods2025.11 | 95.8 | 95.69 | |
| RobertaBackbone=RoBERTa-base, Detection Methodology Category=Model-based methods2025.11 | 95.24 | 95.35 | |
| TACITDetection Methodology Category=Domain generalization methods2025.11 | 94.13 | 94.18 | |
| GhostbusterDetection Methodology Category=Model-based methods2025.11 | 90.79 | 90.79 | |
| MoSEsDetection Methodology Category=Model-based methods2025.11 | 87.87 | 88.6 | |
| Fast-DetectGPTBackbone=GPT-2 Small (124M), Detection Methodology Category=Metric-based methods2025.11 | 78.79 | 77.43 | |
| BinocularsBackbone=GPT-2 Small (124M), Detection Methodology Category=Metric-based methods2025.11 | 72.06 | 73.84 | |
| ImBDDetection Methodology Category=Model-based methods2025.11 | 50 | 66.67 |