Question Answering on QUOREF (test)
80.61EMTASE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TASEEncoder=CorefROBERTa2020.04 | 80.61 | 86.7 | |
| TASEEncoder=ROBERTa2020.04 | 79.66 | 86.13 | |
| CorefROBERTaModel Scale=LARGE2020.04 | 75.8 | 82.81 | |
| CorefRoBERTa_LARGEbackbone=RoBERTa-Large2020.04 | 75.8 | 82.81 | |
| RoBERTa_LARGEdescription=stronger baseline with reasoning framework2020.04 | 75.56 | 82.11 | |
| ROBERTa-MT2020.04 | 72.61 | 80.68 | |
| RoBERTa-MTtraining=multi-task pre-training, source=official leaderboard2020.04 | 72.61 | 80.68 | |
| CorefBERT_LARGEbackbone=BERT-Large2020.04 | 70.67 | 76.89 | |
| BERT_LARGEdescription=stronger baseline with reasoning framework2020.04 | 67.24 | 74 | |
| CorefBERT_Basebackbone=BERT-Base2020.04 | 66.22 | 72.96 | |
| XLNet2020.04 | 61.88 | 71.51 | |
| BERT_BASEdescription=stronger baseline with reasoning framework2020.04 | 61.37 | 68.56 | |
| BERT_BASEdescription=simple fine-tuning2020.04 | 59.28 | 66.39 | |
| QANet+BERT_BASEsource=Dasigi et al. (2019)2020.04 | 42.41 | 47.2 | |
| QANetsource=Dasigi et al. (2019)2020.04 | 34.17 | 38.9 |