Natural Language Inference on AoE NLI (test)
81.6AccuracySUA-TR
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SUA-TRObjective=(15)2026.04 | 81.6 | 77.9 | 0.041 | 0.762 | 87.9 | |
| Standard2026.04 | 81.1 | 69.3 | 0.109 | 0.644 | 84.1 | |
| Temp. Scale2026.04 | 81.1 | 69.3 | 0.056 | 0.649 | 84.4 | |
| Self-Consist.2026.04 | 81.1 | 69.3 | — | 0.672 | 85.8 | |
| AT2026.04 | 80.3 | 75.1 | 0.128 | 0.651 | 84.8 |