OPUS-MT-train/work-spm/en-de/README.md
2020-01-10 16:45:42 +02:00

27 lines
1.1 KiB
Markdown

# opus-2019-12-18.zip
* dataset: opus
* model: transformer-align
* pre-processing: normalization + SentencePiece
* download: [opus-2019-12-18.zip](https://object.pouta.csc.fi/OPUS-MT-models/en-de/opus-2019-12-18.zip)
* test set translations: [opus-2019-12-18.test.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-de/opus-2019-12-18.test.txt)
* test set scores: [opus-2019-12-18.eval.txt](https://object.pouta.csc.fi/OPUS-MT-models/en-de/opus-2019-12-18.eval.txt)
## Benchmarks
| testset | BLEU | chr-F |
|-----------------------|-------|-------|
| newssyscomb2009.en.de | 23.1 | 0.535 |
| news-test2008.en.de | 23.5 | 0.529 |
| newstest2009.en.de | 22.2 | 0.529 |
| newstest2010.en.de | 24.9 | 0.543 |
| newstest2011.en.de | 22.3 | 0.522 |
| newstest2012.en.de | 23.0 | 0.523 |
| newstest2013.en.de | 26.6 | 0.550 |
| newstest2015-ende.en.de | 31.5 | 0.593 |
| newstest2016-ende.en.de | 36.8 | 0.633 |
| newstest2017-ende.en.de | 29.9 | 0.583 |
| newstest2018-ende.en.de | 44.8 | 0.685 |
| newstest2019-ende.en.de | 41.6 | 0.655 |
| Tatoeba.en.de | 46.8 | 0.655 |