Skip to content

Latest commit

 

History

History

sem-fiu

README.md

opus-2021-02-13.zip

  • dataset: opus
  • model: transformer
  • source language(s): amh ara arc hbo heb jpa mlt oar phn syc syr tig tir tmr
  • target language(s): chm est fin fkv hun izh kom krl liv mdf myv olo sma sme udm vep
  • 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-2021-02-13.zip
  • test set translations: opus-2021-02-13.test.txt
  • test set scores: opus-2021-02-13.eval.txt

Benchmarks

testset BLEU chr-F
Tatoeba-test.ara-fin.ara.fin 64.3 0.760
Tatoeba-test.ara-hun.ara.hun 31.5 0.513
Tatoeba-test.heb-fin.heb.fin 33.8 0.589
Tatoeba-test.heb-hun.heb.hun 26.4 0.520
Tatoeba-test.multi.multi 30.0 0.545
Tatoeba-test.tmr-hun.tmr.hun 6.7 0.059

opus-2021-02-19.zip

  • dataset: opus
  • model: transformer
  • source language(s): ara arq arz heb jpa tmr
  • target language(s): fin hun
  • 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: >>hun<< >>fin<<
  • 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.ara-fin 64.3 0.760 7 34 1.000
Tatoeba-test.ara-hun 31.5 0.513 93 482 1.000
Tatoeba-test.arq-hun 8.1 0.148 1 6 1.000
Tatoeba-test.heb-fin 34.0 0.590 212 1302 0.911
Tatoeba-test.heb-hun 26.3 0.519 401 2177 0.986
Tatoeba-test.jpa-hun 6.4 0.113 2 6 1.000
Tatoeba-test.multi-multi 30.0 0.545 718 4013 0.966
Tatoeba-test.tmr-hun 6.7 0.059 5 16 1.000

opus-tuned4ara2fin-2021-03-03.zip

Benchmarks

testset BLEU chr-F #sent #words BP
Tatoeba-test.ara-fin 69.5 0.772 7 34 1.000
Tatoeba-test.ara-hun 3.0 0.162 93 482 1.000
Tatoeba-test.heb-fin 26.4 0.519 212 1302 0.902
Tatoeba-test.heb-hun 1.1 0.153 401 2179 1.000
Tatoeba-test.jpa-hun 9.5 0.110 2 6 1.000
Tatoeba-test.multi-multi 10.4 0.293 718 4013 1.000
Tatoeba-test.tmr-hun 4.2 0.044 5 16 1.000