Skip to content

Latest commit

 

History

History

aav-fiu

README.md

opus-2021-02-10.zip

  • dataset: opus
  • model: transformer
  • source language(s): hoc kha khm mnw sat vie
  • 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-2021-02-10.zip
  • test set translations: opus-2021-02-10.test.txt
  • test set scores: opus-2021-02-10.eval.txt

Benchmarks

testset BLEU chr-F
newsdev2015-enfi-engfin.eng.fin 16.1 0.494
newsdev2018-enet-engest.eng.est 16.7 0.490
newssyscomb2009-enghun.eng.hun 14.2 0.456
newstest2009-enghun.eng.hun 13.7 0.444
newstest2015-enfi-engfin.eng.fin 17.3 0.504
newstest2016-enfi-engfin.eng.fin 18.7 0.516
newstest2017-enfi-engfin.eng.fin 20.7 0.536
newstest2018-enet-engest.eng.est 17.8 0.502
newstest2018-enfi-engfin.eng.fin 13.9 0.475
newstest2019-enfi-engfin.eng.fin 18.4 0.494
newstestB2016-enfi-engfin.eng.fin 15.0 0.487
newstestB2017-enfi-engfin.eng.fin 16.9 0.502
Tatoeba-test.khm-fin.khm.fin 3.7 0.141
Tatoeba-test.multi.multi 31.1 0.539
Tatoeba-test.vie-fin.vie.fin 54.3 0.645
Tatoeba-test.vie-hun.vie.hun 21.3 0.454

opus-2021-02-19.zip

  • dataset: opus
  • model: transformer
  • source language(s): eng hoc kha khm mnw vie
  • target language(s): eng est fin fkv hun izh kom kpv krl liv mdf mhr mrj myv sma sme udm vro
  • 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: >>fin<< >>eng<< >>hun<< >>sme<< >>est<< >>mhr<< >>udm<< >>mrj<<
  • 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
newsdev2015-enfi.eng-fin 16.1 0.494 1500 23375 1.000
newsdev2018-enet.eng-est 16.7 0.490 2000 34508 0.995
newssyscomb2009.eng-hun 14.2 0.456 502 9733 1.000
newstest2009.eng-hun 13.7 0.444 2525 54965 0.993
newstest2015-enfi.eng-fin 17.3 0.504 1370 19968 1.000
newstest2016-enfi.eng-fin 18.7 0.516 3000 48116 0.981
newstest2017-enfi.eng-fin 20.7 0.536 3002 45718 0.989
newstest2018-enet.eng-est 17.8 0.502 2000 36236 0.988
newstest2018-enfi.eng-fin 13.9 0.474 3000 45475 1.000
newstest2019-enfi.eng-fin 18.4 0.494 1997 38369 0.949
newstestB2016-enfi.eng-fin 15.0 0.487 3000 45766 1.000
newstestB2017-enfi.eng-fin 16.9 0.502 3002 45506 0.994
Tatoeba-test.khm-fin 3.7 0.141 5 22 1.000
Tatoeba-test.multi-multi 31.1 0.539 10000 61725 0.966
Tatoeba-test.vie-fin 54.3 0.645 3 14 0.926
Tatoeba-test.vie-hun 21.3 0.454 29 176 1.000

opus4m+btTCv20210807-2021-09-30.zip

Benchmarks

testset BLEU chr-F #sent #words BP
Tatoeba-test-v2021-08-07.khm-fin 39.6 0.443 5 22 0.905
Tatoeba-test-v2021-08-07.multi-multi 23.6 0.506 37 212 1.000
Tatoeba-test-v2021-08-07.vie-fin 45.9 0.645 3 14 1.000
Tatoeba-test-v2021-08-07.vie-hun 19.5 0.499 29 176 1.000