Natural Language Inference on CB (Statistical Analysis)
87.5Final AccuracyTRAS**
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| TRAS**Configuration=MCSA + TR2025.07 | 87.5 | 0.006 | 2.713 | 79.3 | 80.3 | |
| Baseline + TR*Regularization=Textual Regularization from iter. 02025.07 | 85.3 | 0.014 | 2.047 | 79.3 | 80.3 | |
| Baseline + MCSA*Aggregation samples (K)=22025.07 | 82.7 | 0.029 | 1.659 | 79.3 | 80.3 | |
| PromptAgentstatus=Baseline2025.07 | 78.7 | — | — | 79.3 | 80.3 |