Natural Language Understanding on GLUE (SST-2, QNLI, QQP, MNLI, STS-B)
96.67SST-2 AccuracyModernBERT
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ModernBERTModel Size=Large2026.02 | 96.67 | 94.82 | 89.73 | 90.14 | 91.33 | |
| RexBERTModel Size=Large2026.02 | 95.76 | 94 | 89.84 | 92.06 | 91.8 | |
| ModernBERTModel Size=Base2026.02 | 94.72 | 90.88 | 89.26 | 88.71 | 85.92 | |
| RexBERTModel Size=Base2026.02 | 94.27 | 92.75 | 89.28 | 87.23 | 89.55 | |
| RexBERTModel Size=Mini2026.02 | 92.78 | 91.51 | 88.52 | 85.23 | 88.53 | |
| RexBERTModel Size=Micro2026.02 | 90.37 | 85.59 | 85.9 | 78.26 | 83 | |
| DistilBERTModel Size=Mini2026.02 | 89.23 | 87.99 | 87.42 | 80.75 | 84.84 | |
| ettin-17mModel Size=Micro2026.02 | 88.42 | 85.3 | 85.54 | 77.66 | 82.62 |