- dataset: opus
- model: transformer-align
- source language(s): cmn lzh nan yue
- target language(s): heb
- 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 |
|---|---|---|---|---|---|
| Tatoeba-test.heb-zho | 28.4 | 0.472 | 708 | 3654 | 0.995 |