fin-gmq

opus4m+btTCv20210807-2022-01-19.zip

• dataset: opus4m+btTCv20210807
• model: transformer
• source language(s): fin
• target language(s): dan isl nno nob swe
• raw source language(s): fin
• raw target language(s): dan isl nno nob swe
• 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: >>dan<< >>isl<< >>nno<< >>nob<< >>swe<<
• download: opus4m+btTCv20210807-2022-01-19.zip
• test set translations: opus4m+btTCv20210807-2022-01-19.test.txt
• test set scores: opus4m+btTCv20210807-2022-01-19.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
fiskmo_testset.fin-swe	28.5	0.611	523	9479	1.000
Tatoeba-test-v2021-08-07.fin-dan	53.7	0.698	2665	16668	0.997
Tatoeba-test-v2021-08-07.fin-isl	100.0	1.000	2	8	1.000
Tatoeba-test-v2021-08-07.fin-multi	50.9	0.691	5155	31444	1.000
Tatoeba-test-v2021-08-07.fin-nor	46.7	0.679	2488	14764	1.000
Tatoeba-test-v2021-08-07.fin-swe	51.9	0.686	2841	19109	0.966