Natural Language Inference on MultiNLI Noise (test)
87.57AccuracyRoBERTa
Evaluation Results
| Method | Links | |
|---|---|---|
| RoBERTaMitigation Approach=None2026.04 | 87.57 | |
| Hybrid MitigationMitigation Approach=Hybrid, Base Model=ELECTRA-small2026.04 | 82.18 | |
| GPT-4o-miniMitigation Approach=None2026.04 | 81.85 | |
| AugmentationMitigation Approach=Augmentation, Base Model=ELECTRA-small2026.04 | 81.61 | |
| PreprocessingMitigation Approach=Preprocessing, Base Model=ELECTRA-small2026.04 | 77.09 | |
| ELECTRA-smallMitigation Approach=None2026.04 | 75.54 | |
| GPT-3.5Mitigation Approach=None2026.04 | 64.16 |