Document-Level Machine Translation on TED15 Zh-En 2010-2013 (test)
31.4d-BLEUDOCmT5-25-Large
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DOCmT5-25-LargeCategory=Cross-Lingual2021.12 | 31.4 | — | — | |
| DOCmT5-25Category=Cross-Lingual2021.12 | 30.99 | — | — | |
| DOCmT5-5-LargeCategory=Cross-Lingual2021.12 | 30.52 | — | — | |
| Doc-mBART25 + NRETM PeBase Model=Doc-mBART25, Constraint=Pe2021.07 | 30.4 | 24.9 | 100 | |
| Doc-mBART25Model Type=Document-level pre-trained2021.07 | 29.6 | 23.8 | 98.7 | |
| mBARTCategory=Previous Systems2021.12 | 29.6 | — | — | |
| DOCmT5-5Category=Cross-Lingual2021.12 | 28.97 | — | — | |
| Sen-mBART25Model Type=Sentence-level pre-trained2021.07 | 28.4 | — | — | |
| MARGECategory=Previous Systems2021.12 | 28.4 | — | — | |
| mT5 w. DocNMTCategory=Cross-Lingual2021.12 | 26.17 | — | — | |
| mT5 w. DocTLMCategory=Cross-Lingual2021.12 | 25.87 | — | — | |
| mT5Category=Mono-Lingual2021.12 | 24.24 | — | — | |
| mT5 w. cont-5langsCategory=Mono-Lingual2021.12 | 24.22 | — | — | |
| HANModel Type=Non-pretraining approach2021.07 | 24 | — | — | |
| HANCategory=Previous Systems2021.12 | 24 | — | — | |
| mT5 w. DrCategory=Mono-Lingual, Objective=Document-Reordering (Dr)2021.12 | 23.75 | — | — |