Cross-lingual Fact-to-Text generation on XALIGN (test)
40.6Score (hi)mT5
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| mT5Model=mT52022.02 | 40.6 | 20.2 | 11.4 | 13.6 | 43.7 | 16.6 | 45.3 | 8.7 | 25 | |
| TransformerModel=Transformer2022.02 | 35.4 | 17.3 | 6.9 | 8.8 | 38.8 | 13.2 | 35.6 | 3.1 | 19.9 | |
| GAT+TransformerModel=GAT+Trans.2022.02 | 29.5 | 17.9 | 4.9 | 7.2 | 40.3 | 11.3 | 30.2 | 5.1 | 18.3 | |
| Naïve translation baselineModel=Baseline2022.02 | 2.7 | 2 | 1 | 1.7 | 1.01 | 0.6 | 2.7 | 0.4 | 1.5 |