- dataset: opusTCv20210807+bt
- model: transformer-big
- source language(s): eng
- target language(s): cat oci spa
- raw source language(s): eng
- raw target language(s): cat oci spa
- model: transformer-big
- pre-processing: normalization + SentencePiece (spm32k,spm32k)
- a sentence initial language token is required in the form of
>>id<<(id = valid target language ID) - valid language labels:
- download: opusTCv20210807+bt_transformer-big_2022-03-13.zip
- test set translations: opusTCv20210807+bt_transformer-big_2022-03-13.test.txt
- test set scores: opusTCv20210807+bt_transformer-big_2022-03-13.eval.txt
| testset | BLEU | chr-F | #sent | #words | BP |
|---|---|---|---|---|---|
| newssyscomb2009.eng-spa | 31.5 | 0.58468 | 502 | 12506 | 0.997 |
| news-test2008.eng-spa | 29.7 | 0.56500 | 2051 | 52596 | 1.000 |
| newstest2009.eng-spa | 30.5 | 0.57968 | 2525 | 68114 | 1.000 |
| newstest2010.eng-spa | 37.5 | 0.62318 | 2489 | 65522 | 1.000 |
| newstest2011.eng-spa | 39.1 | 0.62409 | 3003 | 79476 | 0.983 |
| newstest2012.eng-spa | 39.6 | 0.62962 | 3003 | 79006 | 0.970 |
| newstest2013.eng-spa | 35.8 | 0.60269 | 3000 | 70528 | 0.961 |
| Tatoeba-test-v2021-08-07.eng-cat | 47.8 | 0.66475 | 1631 | 12342 | 0.991 |
| Tatoeba-test-v2021-08-07.eng-multi | 55.5 | 0.72143 | 10000 | 79854 | 0.987 |
| Tatoeba-test-v2021-08-07.eng-oci | 7.7 | 0.29437 | 841 | 5219 | 0.899 |
| Tatoeba-test-v2021-08-07.eng-spa | 56.7 | 0.73479 | 16583 | 134681 | 0.988 |
| tico19-test.eng-spa | 52.4 | 0.73122 | 2100 | 66591 | 0.939 |