Question Answering on TyDiQA GoldP
89.4F1 ScoreByT5
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ByT5Scale=XXL, Evaluation Protocol=In-language multitask2021.05 | 89.4 | 81.4 | |
| mT5Scale=XXL, Evaluation Protocol=In-language multitask2021.05 | 88.7 | 79.5 | |
| ByT5Scale=XL, Evaluation Protocol=In-language multitask2021.05 | 88 | 79.3 | |
| ByT5Scale=Large, Evaluation Protocol=In-language multitask2021.05 | 87.7 | 79.2 | |
| mT5Scale=XL, Evaluation Protocol=In-language multitask2021.05 | 87.6 | 78.4 | |
| ByT5Scale=Base, Evaluation Protocol=In-language multitask2021.05 | 86.4 | 78 | |
| mT5Scale=Large, Evaluation Protocol=In-language multitask2021.05 | 85.3 | 75.3 | |
| mT5Scale=XXL, Evaluation Protocol=Translate-train2021.05 | 83.3 | 69.4 | |
| ByT5Scale=XXL, Evaluation Protocol=Translate-train2021.05 | 83.2 | 69.6 | |
| ByT5Scale=Small, Evaluation Protocol=In-language multitask2021.05 | 82.6 | 73.6 | |
| mT5Scale=Base, Evaluation Protocol=In-language multitask2021.05 | 81.7 | 71.2 | |
| ByT5Scale=XL, Evaluation Protocol=Translate-train2021.05 | 81.5 | 67.6 | |
| ByT5Scale=Large, Evaluation Protocol=Translate-train2021.05 | 80.1 | 66.4 | |
| mT5Scale=XL, Evaluation Protocol=Translate-train2021.05 | 80.1 | 65 | |
| mT5Scale=Small, Evaluation Protocol=In-language multitask2021.05 | 75.9 | 64.8 | |
| mT5Scale=Large, Evaluation Protocol=Translate-train2021.05 | 75.8 | 60.2 | |
| ByT5Scale=Base, Evaluation Protocol=Translate-train2021.05 | 75.6 | 61.7 | |
| mT5Scale=Base, Evaluation Protocol=Translate-train2021.05 | 66.4 | 51 | |
| ByT5Scale=Small, Evaluation Protocol=Translate-train2021.05 | 64.2 | 50.6 | |
| mT5Scale=Small, Evaluation Protocol=Translate-train2021.05 | 49.8 | 35.6 |