| ID | Language | Benchmark | Output | bleu |
|---|---|---|---|---|
| 0 | eng-deu | flores101-devtest | show | 33.4 |
| 1 | eng-deu | multi30k_task2_test_2016 | show | 2.6 |
| 2 | eng-deu | multi30k_test_2016_flickr | show | 34.2 |
| 3 | eng-deu | multi30k_test_2017_flickr | show | 33.3 |
| 4 | eng-deu | multi30k_test_2017_mscoco | show | 28.3 |
| 5 | eng-deu | multi30k_test_2018_flickr | show | 28.9 |
| 6 | eng-deu | newssyscomb2009 | show | 21.6 |
| 7 | eng-deu | newstest2009 | show | 21.3 |
| 8 | eng-deu | newstest2010 | show | 23.7 |
| 9 | eng-deu | newstest2011 | show | 20.9 |
| 10 | eng-deu | newstest2012 | show | 21.6 |
| 11 | eng-deu | newstest2013 | show | 25.3 |
| 12 | eng-deu | newstest2014 | show | 25.7 |
| 13 | eng-deu | newstest2015 | show | 29.7 |
| 14 | eng-deu | newstest2016 | show | 34.7 |
| 15 | eng-deu | newstest2017 | show | 28.2 |
| 16 | eng-deu | newstest2018 | show | 41.6 |
| 17 | eng-deu | newstest2019 | show | 37.5 |
| 18 | eng-deu | newstest2020 | show | 28.8 |
| 19 | eng-deu | newstestB2020 | show | 28.1 |
| 20 | eng-deu | tatoeba-test-v2020-07-28 | show | 43.1 |
| 21 | eng-deu | tatoeba-test-v2021-03-30 | show | 42.1 |
| 22 | eng-deu | tatoeba-test-v2021-08-07 | show | 41.2 |
| average | 29.383 |