opus-tatoeba-en-ro / README.md
tiedeman's picture
Initial commit
6c507fe
|
raw
history blame
2.37 kB
metadata
language:
  - en
  - ro
tags:
  - translation
license: apache-2.0

en-ro

  • source group: English

  • target group: Romanian

  • OPUS readme: eng-ron

  • model: transformer-align

  • source language(s): eng

  • target language(s): mol ron

  • model: transformer-align

  • 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 original weights: opus+bt-2021-03-07.zip

  • test set translations: opus+bt-2021-03-07.test.txt

  • test set scores: opus+bt-2021-03-07.eval.txt

Benchmarks

testset BLEU chr-F #sent #words BP
newsdev2016-enro.eng-ron 33.5 0.610 1999 51566 0.984
newstest2016-enro.eng-ron 31.7 0.591 1999 49094 0.998
Tatoeba-test.eng-ron 46.9 0.678 5000 36851 0.983

System Info: