Conversational Question Answering on CoQA official (test)
88.8Overall F1Human
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Human2018.12 | 88.8 | 90.2 | 88.4 | 89.8 | 88.6 | 89.9 | 86.7 | 88.1 | — | — | |
| Human2018.10 | 88.8 | 90.2 | 88.4 | 89.8 | 88.6 | 89.9 | 86.7 | 88.1 | — | — | |
| SDNetensemble=true2018.12 | 79.3 | 78.7 | 77.1 | 80.2 | 81.9 | 85.2 | 72.3 | 79.7 | — | — | |
| SDNetensemble=false2018.12 | 76.6 | 75.4 | 73.9 | 77.1 | 80.3 | 83.1 | 69.8 | 76.8 | — | — | |
| FlowQA2018.12 | 75 | 73.7 | 71.6 | 76.8 | 79 | 80.2 | 67.8 | 76.1 | — | — | |
| FLOWQAN-Ans=12018.10 | 75 | 73.7 | 71.6 | 76.8 | 79 | 80.2 | 67.8 | 76.1 | — | — | |
| BiDAF++2018.12 | 67.8 | 66.5 | 65.7 | 70.2 | 71.6 | 72.6 | 60.8 | 67.1 | — | — | |
| BiDAF++N-ctx=32018.10 | 67.8 | 66.5 | 65.7 | 70.2 | 71.6 | 72.6 | 60.8 | 67.1 | — | — | |
| BiDAF++context window=3-ctx2018.09 | 67.8 | — | — | — | — | — | — | — | 69.4 | 63.8 | |
| DrQA+PGNet2018.12 | 65.1 | 64.2 | 63.7 | 67.1 | 68.3 | 71.4 | 57.8 | 63.1 | — | — | |
| DrQA + PGNetN-ctx=12018.10 | 65.1 | 64.2 | 63.7 | 67.1 | 68.3 | 71.4 | 57.8 | 63.1 | — | — | |
| DrQA + PGNettype=Abstractive2018.09 | 65.1 | — | — | — | — | — | — | — | 67 | 60.4 | |
| DrQA2018.12 | 52.6 | 46.7 | 53.9 | 54.1 | 57.8 | 59.4 | 45 | 51 | — | — | |
| DrQAN-ctx=12018.10 | 52.6 | 46.7 | 53.9 | 54.1 | 57.8 | 59.4 | 45 | 51 | — | — | |
| DrQAtype=Extractive2018.09 | 52.6 | — | — | — | — | — | — | — | 54.5 | 47.9 | |
| PGNet2018.12 | 44.1 | 49 | 43.3 | 47.5 | 47.5 | 45.1 | 38.6 | 38.1 | — | — | |
| PGNetN-ctx=12018.10 | 44.1 | 49 | 43.3 | 47.5 | 47.5 | 45.1 | 38.6 | 38.1 | — | — |