- dataset: opusTCv20210807
- model: transformer-big
- source language(s): ces pol rus slv ukr
- target language(s): kor
- raw source language(s): ces pol rus slv ukr
- raw target language(s): kor
- model: transformer-big
- pre-processing: normalization + SentencePiece (spm32k,spm32k)
- download: opusTCv20210807-sepvoc_transformer-big_2022-09-15.zip
- test set translations: opusTCv20210807-sepvoc_transformer-big_2022-09-15.test.txt
- test set scores: opusTCv20210807-sepvoc_transformer-big_2022-09-15.eval.txt
| testset | BLEU | chr-F | #sent | #words | BP |
|---|---|---|---|---|---|
| Tatoeba-test-v2021-08-07.ces-kor | 15.3 | 0.40399 | 14 | 58 | 1.000 |
| Tatoeba-test-v2021-08-07.multi-kor | 14.0 | 0.32889 | 345 | 1757 | 0.990 |
| Tatoeba-test-v2021-08-07.pol-kor | 9.7 | 0.27479 | 75 | 417 | 0.978 |
| Tatoeba-test-v2021-08-07.rus-kor | 15.7 | 0.33704 | 221 | 1125 | 0.990 |
| Tatoeba-test-v2021-08-07.slv-kor | 6.5 | 0.21775 | 7 | 34 | 0.842 |
| Tatoeba-test-v2021-08-07.ukr-kor | 21.7 | 0.44803 | 28 | 123 | 1.000 |