Natural Language Inference on SNLI Noise variant (test)
89.77AccuracyRoBERTa
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| RoBERTaMitigation Approach=None2026.04 | 89.77 | — | |
| AugmentationMitigation Approach=Augmentation, Base Model=ELECTRA-small2026.04 | 89.44 | — | |
| ELECTRA (Hybrid)Training Configuration=Hybrid, Evaluation Protocol=Fine-tuned2026.04 | 89.26 | 0.0001 | |
| Hybrid MitigationMitigation Approach=Hybrid, Base Model=ELECTRA-small2026.04 | 89.26 | — | |
| GPT-4o-miniEvaluation Protocol=Zero-shot2026.04 | 86.26 | — | |
| GPT-4o-miniMitigation Approach=None2026.04 | 86.26 | — | |
| PreprocessingMitigation Approach=Preprocessing, Base Model=ELECTRA-small2026.04 | 82.36 | — | |
| ELECTRA (Baseline)Training Configuration=Baseline, Evaluation Protocol=Fine-tuned2026.04 | 80.19 | 0.0001 | |
| ELECTRA-smallMitigation Approach=None2026.04 | 80.19 | — | |
| GPT-3.5Mitigation Approach=None2026.04 | 63.55 | — |