Skip to content

Latest commit

 

History

History
75 lines (63 loc) · 3.46 KB

README.md

File metadata and controls

75 lines (63 loc) · 3.46 KB

opus-2020-10-04.zip

  • dataset: opus
  • model: transformer
  • source language(s): bre cor cym eng gla gle glv
  • target language(s): bre cor cym eng gla gle glv
  • model: transformer
  • pre-processing: normalization + SentencePiece (spm32k,spm32k)
  • a sentence initial language token is required in the form of >>id<< (id = valid target language ID)
  • download: opus-2020-10-04.zip
  • test set translations: opus-2020-10-04.test.txt
  • test set scores: opus-2020-10-04.eval.txt

Benchmarks

testset BLEU chr-F
Tatoeba-test.bre-eng.bre.eng 18.0 0.378
Tatoeba-test.cor-cym.cor.cym 8.2 0.312
Tatoeba-test.cor-eng.cor.eng 2.7 0.170
Tatoeba-test.cym-cor.cym.cor 4.3 0.086
Tatoeba-test.cym-eng.cym.eng 39.7 0.574
Tatoeba-test.cym-gla.cym.gla 7.8 0.177
Tatoeba-test.cym-gle.cym.gle 19.7 0.437
Tatoeba-test.cym-glv.cym.glv 9.9 0.269
Tatoeba-test.eng-bre.eng.bre 9.6 0.330
Tatoeba-test.eng-cor.eng.cor 0.3 0.085
Tatoeba-test.eng-cym.eng.cym 30.4 0.539
Tatoeba-test.eng-gla.eng.gla 8.3 0.316
Tatoeba-test.eng-gle.eng.gle 34.2 0.567
Tatoeba-test.eng-glv.eng.glv 11.3 0.372
Tatoeba-test.gla-cym.gla.cym 6.3 0.407
Tatoeba-test.gla-eng.gla.eng 15.2 0.324
Tatoeba-test.gle-cym.gle.cym 15.9 0.424
Tatoeba-test.gle-eng.gle.eng 47.0 0.637
Tatoeba-test.gle-glv.gle.glv 17.4 0.476
Tatoeba-test.glv-cym.glv.cym 7.4 0.271
Tatoeba-test.glv-eng.glv.eng 18.1 0.438
Tatoeba-test.glv-gle.glv.gle 48.3 0.587
Tatoeba-test.multi.multi 20.5 0.357

opus-2021-02-19.zip

  • dataset: opus
  • model: transformer
  • source language(s): cor cym gla gle glv
  • target language(s): cor cym gla gle glv
  • model: transformer
  • 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: >>eng<< >>cym<< >>glv<< >>gle<< >>gla<< >>bre<< >>cor<<
  • download: opus-2021-02-19.zip
  • test set translations: opus-2021-02-19.test.txt
  • test set scores: opus-2021-02-19.eval.txt

Benchmarks

testset BLEU chr-F #sent #words BP
Tatoeba-test.cor-cym 8.2 0.312 3 19 1.000
Tatoeba-test.cym-cor 4.3 0.086 3 18 0.607
Tatoeba-test.cym-gla 7.8 0.195 1 7 1.000
Tatoeba-test.cym-gle 19.7 0.437 1 11 1.000
Tatoeba-test.cym-glv 9.9 0.269 11 60 1.000
Tatoeba-test.gla-cym 8.6 0.404 1 6 1.000
Tatoeba-test.gle-cym 15.9 0.424 1 13 0.834
Tatoeba-test.gle-glv 17.4 0.476 3 14 1.000
Tatoeba-test.glv-cym 7.4 0.271 11 71 0.986
Tatoeba-test.glv-gle 48.3 0.587 3 15 1.000
Tatoeba-test.multi-multi 14.4 0.353 766 4334 0.955