Binary human vs. AI classification on NarraBench (test)
99.9Macro F1 ScoreModernBERT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ModernBERT2026.04 | 99.9 | 100 | |
| Stylometric+XGBSize=1442026.04 | 99.8 | 99.9 | |
| TF-IDF+XGBSize=5,0002026.04 | 99.7 | 99.9 | |
| Narr. + StyleSize=3042026.04 | 96 | 98.2 | |
| NarrativeSize=2572026.04 | 93.2 | 95.9 | |
| Core+FPSize=1012026.04 | 91.1 | 93.4 | |
| Style OnlySize=392026.04 | 85.8 | 86.7 | |
| Core OnlySize=302026.04 | 84.8 | 82.8 | |
| Binoculars2026.04 | 55.9 | 40.4 |