Paraphrase Identification on QQP, PAWS, and PIT (In-distribution test)
92AUROC (QQP->QQP)ROBERTa
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ROBERTa2022.10 | 92 | 98.7 | 90.2 | 82 | 80 | 88.6 | |
| GAPX2022.10 | 90.7 | 98.1 | 87.4 | 72.5 | 79.2 | 85.6 | |
| BART2022.10 | 90.6 | 98.2 | 89.1 | 78.2 | 77.9 | 86.8 | |
| BERT2022.10 | 90 | 97.6 | 85.1 | 77.3 | 73.4 | 84.7 | |
| BERT+EPEntropy Penalty=true2022.10 | 89.8 | 95.8 | 82.7 | 76.1 | 73.7 | 83.6 | |
| IDP2022.10 | 88.1 | 95.4 | 86.8 | 77.9 | 74.6 | 84.6 | |
| OODP2022.10 | 79.6 | 84.6 | 73.6 | 72.8 | 76.4 | 77.4 | |
| GAP2022.10 | 78.9 | 92.3 | 84.5 | 72.3 | 76.4 | 80.9 | |
| BILSTM2022.10 | 64.6 | 49.6 | 50.9 | 52.8 | 49 | 53.4 | |
| BOW2022.10 | 60.9 | 57.7 | 47.7 | 47.4 | 61.3 | 55 |