| ID | Language | Benchmark | Output | spbleu |
|---|---|---|---|---|
| 0 | eng-deu | flores101-devtest | show | 47.7 |
| 1 | eng-deu | flores200-devtest | show | 47.7 |
| 2 | eng-deu | multi30k_task2_test_2016 | show | 5.8 |
| 3 | eng-deu | multi30k_test_2016_flickr | show | 47.2 |
| 4 | eng-deu | multi30k_test_2017_flickr | show | 48.2 |
| 5 | eng-deu | multi30k_test_2017_mscoco | show | 42.2 |
| 6 | eng-deu | multi30k_test_2018_flickr | show | 44.8 |
| 7 | eng-deu | newssyscomb2009 | show | 32.2 |
| 8 | eng-deu | newstest2009 | show | 30.9 |
| 9 | eng-deu | newstest2010 | show | 34.3 |
| 10 | eng-deu | newstest2011 | show | 31.5 |
| 11 | eng-deu | newstest2012 | show | 35 |
| 12 | eng-deu | newstest2013 | show | 38.9 |
| 13 | eng-deu | newstest2014 | show | 43.3 |
| 14 | eng-deu | newstest2015 | show | 47.6 |
| 15 | eng-deu | newstest2016 | show | 47.3 |
| 16 | eng-deu | newstest2017 | show | 46 |
| 17 | eng-deu | newstest2018 | show | 54.6 |
| 18 | eng-deu | newstest2019 | show | 48.5 |
| 19 | eng-deu | newstest2020 | show | 42.9 |
| 20 | eng-deu | newstestB2020 | show | 43 |
| 21 | eng-deu | tatoeba-test-v2020-07-28 | show | 48.5 |
| 22 | eng-deu | tatoeba-test-v2021-03-30 | show | 47.6 |
| 23 | eng-deu | tatoeba-test-v2021-08-07 | show | 47.5 |
| average | 41.800 |