Natural Language Inference on SNLI hard 1.0 (test)
84.48AccuracyBERT-base + PoE w/ Seq-Z
Evaluation results
| Method | Links | |
|---|---|---|
| BERT-base + PoE w/ Seq-ZBackbone=BERT-base, Debiasing Strategy=PoE, Training Data=Seq-Z Z(DG+ -> Z(DSNLI))2022.03 | 84.48 | |
| DFLTraining Data=SNLI, Debiasing Strategy=Debiased Focal Loss2022.03 | 83.01 | |
| BERT-base + PoE w/ DSNLIBackbone=BERT-base, Debiasing Strategy=PoE, Training Data=DSNLI2022.03 | 82.92 | |
| BERT-base w/ Seq-ZBackbone=BERT-base, Debiasing Dataset=Seq-Z Z(DG -> Z(DSNLI))2022.03 | 82.82 | |
| BERT-base w/ Par-ZBackbone=BERT-base, Debiasing Dataset=Par-Z Z(DSNLI) U Z(DG+)2022.03 | 82.81 | |
| PoETraining Data=SNLI, Debiasing Strategy=Product of Experts2022.03 | 82.15 | |
| BERT-base w/ Z-AugBackbone=BERT-base, Debiasing Dataset=Z-Aug Z(DG + DSNLI)2022.03 | 81.78 | |
| BERT-base baselineBackbone=BERT-base, Training Data=DSNLI2022.03 | 80.34 | |
| BERT teacherRole=Teacher, Architecture=BERT2023.05 | 80.31 | |
| CRvNNModel Type=Sequence Encoder2023.11 | 70.6 | |
| OMModel Type=Sequence Encoder2023.11 | 70.6 | |
| BT-GRCModel Type=Sequence Encoder2023.11 | 70.5 | |
| RIR-EBT-GRCModel Type=Sequence Encoder2023.11 | 70.4 | |
| BT-GRC OSModel Type=Sequence Encoder2023.11 | 70.3 | |
| EBT-GRCModel Type=Sequence Encoder2023.11 | 69.9 | |
| RIR-GRCModel Type=Sequence Encoder2023.11 | 69.6 | |
| BBT-GRCModel Type=Sequence Encoder2023.11 | 69.2 | |
| Ens. AdvClsTraining Data=SNLI, Debiasing Strategy=Ensemble Adversarial Classifier2022.03 | 67.42 | |
| AdvClsTraining Data=SNLI, Debiasing Strategy=Adversarial Classifier2022.03 | 66.27 | |
| DTA with DMURole=Student, Architecture=TinyBERT2023.05 | 66.04 | |
| DMURole=Student, Architecture=TinyBERT2023.05 | 65.94 | |
| DTARole=Student, Architecture=TinyBERT2023.05 | 60.26 | |
| KD + SmoothingRole=Student, Architecture=TinyBERT2023.05 | 60.07 | |
| KD (standard distillation)Role=Student, Architecture=TinyBERT2023.05 | 60.02 | |
| TinyBERT baselineRole=Student, Architecture=TinyBERT2023.05 | 56.98 | |
| Baseline w/ labelled aug. dataRole=Student, Architecture=TinyBERT2023.05 | 56.73 | |
| JTTRole=Student, Architecture=TinyBERT2023.05 | 55.93 |