Question Answering on SQuAD 2.0 Sep 9, 2018 (test)
86.9EMHuman
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HumanModel Type=Pipeline2018.10 | 86.9 | 89.5 | |
| Human2018.10 | 86.9 | 89.5 | |
| BERT_LARGEmodel_type=Single2018.10 | 80 | 83.1 | |
| MIR-MRC (F-Net)model_type=Single2018.10 | 74.8 | 78 | |
| nlnetmodel_type=Single2018.10 | 74.2 | 77.1 | |
| RMR+ELMo+VerifierModel Type=Pipeline2018.10 | 71.7 | 74.2 | |
| U-NetModel Type=Ensemble2018.10 | 71.5 | 75 | |
| unetmodel_type=Ensemble2018.10 | 71.4 | 74.9 | |
| SLQA+model_type=Single2018.10 | 71.4 | 74.4 | |
| SANModel Type=Ensemble2018.10 | 71.3 | 73.7 | |
| FusionNet++Model Type=Ensemble2018.10 | 70.3 | 72.6 | |
| U-NetModel Type=End-to-end2018.10 | 69.2 | 72.6 | |
| SANModel Type=End-to-end2018.10 | 68.6 | 71.4 | |
| VS3-NetModel Type=End-to-end2018.10 | 68.4 | 71.3 | |
| FusionNet++Model Type=End-to-end2018.10 | 66.6 | 69.6 | |
| DocQAModel Type=End-to-end2018.10 | 63.4 | 66.3 | |
| BNA*Model Type=End-to-end2018.10 | 59.2 | 62.1 |