| ID | Language | Benchmark | Output | bleu |
|---|---|---|---|---|
| 0 | deu-fra | flores101-devtest | show | 33.8 |
| 1 | deu-fra | multi30k_test_2016_flickr | show | 36.5 |
| 2 | deu-fra | multi30k_test_2017_flickr | show | 34.5 |
| 3 | deu-fra | multi30k_test_2017_mscoco | show | 32 |
| 4 | deu-fra | multi30k_test_2018_flickr | show | 29.2 |
| 5 | deu-fra | newssyscomb2009 | show | 26.5 |
| 6 | deu-fra | newstest2009 | show | 25.3 |
| 7 | deu-fra | newstest2010 | show | 27.8 |
| 8 | deu-fra | newstest2011 | show | 26.6 |
| 9 | deu-fra | newstest2012 | show | 27.2 |
| 10 | deu-fra | newstest2013 | show | 29.1 |
| 11 | deu-fra | newstest2019 | show | 32.6 |
| 12 | deu-fra | newstest2020 | show | 29.4 |
| 13 | deu-fra | newstest2021 | show | 26.5 |
| 14 | deu-fra | tatoeba-test-v2020-07-28 | show | 49.4 |
| 15 | deu-fra | tatoeba-test-v2021-03-30 | show | 50.2 |
| 16 | deu-fra | tatoeba-test-v2021-08-07 | show | 50.3 |
| average | 33.347 |