Question Answering on MULTIHIERTT (test)
83.12EMHuman Expert Performance
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Human Expert Performance2022.06 | 83.12 | 87.03 | |
| MT2NetBackbone=ROBERTa-large2022.06 | 36.22 | 38.43 | |
| MT2NetBackbone=ROBERTa-base2022.06 | 34.32 | 36.17 | |
| MT2NetBackbone=BERT-large2022.06 | 33.25 | 34.98 | |
| MT2NetBackbone=BERT-base2022.06 | 32.07 | 33.67 | |
| FinQANetBackbone=ROBERTa-large2022.06 | 31.72 | 33.6 | |
| Facts Retrieving + Seq2Prog2022.06 | 24.58 | 26.3 | |
| TAGOPBackbone=ROBERTa-large2022.06 | 17.81 | 19.35 | |
| Facts Retrieving + NumNet2022.06 | 10.77 | 12.02 | |
| Facts Retrieving + TAPAS2022.06 | 7.67 | 10.04 | |
| Longformer + Reasoning2022.06 | 2.86 | 6.23 |