| ID | Language | Benchmark | Output | bleu |
|---|---|---|---|---|
| 0 | eng-fra | flores101-devtest | show | 47.6 |
| 1 | eng-fra | multi30k_test_2016_flickr | show | 48.9 |
| 2 | eng-fra | multi30k_test_2017_flickr | show | 49.1 |
| 3 | eng-fra | multi30k_test_2017_mscoco | show | 51.4 |
| 4 | eng-fra | multi30k_test_2018_flickr | show | 41.9 |
| 5 | eng-fra | newsdiscusstest2015 | show | 38.4 |
| 6 | eng-fra | newssyscomb2009 | show | 29.2 |
| 7 | eng-fra | newstest2009 | show | 30.0 |
| 8 | eng-fra | newstest2010 | show | 31.6 |
| 9 | eng-fra | newstest2011 | show | 34.8 |
| 10 | eng-fra | newstest2012 | show | 32.1 |
| 11 | eng-fra | newstest2013 | show | 34.1 |
| 12 | eng-fra | newstest2014 | show | 38.8 |
| 13 | eng-fra | tatoeba-test-v2020-07-28 | show | 50.6 |
| 14 | eng-fra | tatoeba-test-v2021-03-30 | show | 51.2 |
| 15 | eng-fra | tatoeba-test-v2021-08-07 | show | 51.7 |
| 16 | eng-fra | tico19-test | show | 36.4 |
| average | 41.047 |