Natural Language Inference on MNLI (all combined)
85.98AccuracyDTAens
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DTAensBackbone=DeBERTa, Hyper-parameter tuning on MNLI=false, Zero-shot=true2023.05 | 85.98 | 1.31 | |
| KDBackbone=DeBERTa, Hyper-parameter tuning on MNLI=false, Zero-shot=true2023.05 | 85.04 | 0.37 | |
| LIRExBackbone=RoBERTa, Hyper-parameter tuning on MNLI=false, Zero-shot=true2023.05 | 79.82 | -0.1 | |
| NILEBackbone=RoBERTa, Hyper-parameter tuning on MNLI=false, Zero-shot=true2023.05 | 77.15 | -2.14 | |
| DTAensBackbone=BERT, Hyper-parameter tuning on MNLI=false, Zero-shot=true2023.05 | 76.43 | 1.44 | |
| KDBackbone=BERT, Hyper-parameter tuning on MNLI=false, Zero-shot=true2023.05 | 75.46 | 0.47 | |
| Debiased Focal LossBackbone=BERT, Hyper-parameter tuning on MNLI=true, Zero-shot=true2023.05 | 73.79 | -0.4 | |
| Product of ExpertsBackbone=BERT, Hyper-parameter tuning on MNLI=true, Zero-shot=true2023.05 | 73.55 | -0.64 | |
| Rationale supervisionBackbone=BERT, Hyper-parameter tuning on MNLI=false, Zero-shot=true2023.05 | 73.28 | 0.87 | |
| Ensemble adversariesBackbone=LSTM, Hyper-parameter tuning on MNLI=true, Zero-shot=true2023.05 | 53.49 | 0.35 | |
| Hyp-only adversaryBackbone=LSTM, Hyper-parameter tuning on MNLI=true, Zero-shot=true2023.05 | 48.24 | 1.52 | |
| Negative samplingBackbone=LSTM, Hyper-parameter tuning on MNLI=true, Zero-shot=true2023.05 | 43.71 | -3.01 |