- dataset: opus
- model: transformer-align
- source language(s): rus
- target language(s): fra
- model: transformer-align
- pre-processing: normalization + SentencePiece (spm32k,spm32k)
- download: opus-2021-05-16.zip
- test set translations: opus-2021-05-16.test.txt
- test set scores: opus-2021-05-16.eval.txt
| testset | BLEU | chr-F | #sent | #words | BP |
|---|---|---|---|---|---|
| newstest2012.rus-fra | 18.3 | 0.485 | 3003 | 78011 | 1.000 |
| newstest2013.rus-fra | 21.5 | 0.499 | 3000 | 70037 | 1.000 |
| Tatoeba-test.rus-fra | 44.3 | 0.617 | 10000 | 70132 | 0.975 |