zle-fra

opusTCv20210807_transformer-big_2022-03-14.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): bel bel_Latn orv_Cyrl rus ukr
• target language(s): fra
• raw source language(s): bel orv rus ukr
• raw target language(s): fra
• model: transformer-big
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• download: opusTCv20210807_transformer-big_2022-03-14.zip
• test set translations: opusTCv20210807_transformer-big_2022-03-14.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-03-14.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newstest2012.rus-fra	22.4	0.51563	3003	78011	1.000
newstest2013.rus-fra	26.2	0.53771	3000	70037	0.999
Tatoeba-test-v2021-08-07.bel-fra	26.8	0.47204	283	2005	1.000
Tatoeba-test-v2021-08-07.multi-fra	50.8	0.67309	10000	66671	0.982
Tatoeba-test-v2021-08-07.orv-fra	4.5	0.18647	37	290	1.000
Tatoeba-test-v2021-08-07.rus-fra	46.0	0.64489	11490	80573	0.980
Tatoeba-test-v2021-08-07.ukr-fra	44.6	0.63086	10035	63222	0.996

opusTCv20210807_transformer-big_2022-03-19.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): bel bel_Latn orv_Cyrl rus ukr
• target language(s): fra
• raw source language(s): bel orv rus ukr
• raw target language(s): fra
• model: transformer-big
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• download: opusTCv20210807_transformer-big_2022-03-19.zip
• test set translations: opusTCv20210807_transformer-big_2022-03-19.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-03-19.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newstest2012.rus-fra	25.3	0.53653	3003	78011	0.993
newstest2013.rus-fra	29.4	0.56073	3000	70037	0.994
Tatoeba-test-v2021-08-07.bel-fra	43.5	0.63052	283	2005	1.000
Tatoeba-test-v2021-08-07.multi-fra	52.1	0.68244	10000	66671	0.979
Tatoeba-test-v2021-08-07.orv-fra	9.6	0.24958	37	290	0.951
Tatoeba-test-v2021-08-07.rus-fra	51.6	0.68606	11490	80573	0.974
Tatoeba-test-v2021-08-07.ukr-fra	50.5	0.67740	10035	63222	0.982

opusTCv20210807_transformer-big_2022-03-23.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): bel bel_Latn orv_Cyrl rus ukr
• target language(s): fra
• raw source language(s): bel orv rus ukr
• raw target language(s): fra
• model: transformer-big
• pre-processing: normalization + SentencePiece (spm32k,spm32k)
• download: opusTCv20210807_transformer-big_2022-03-23.zip
• test set translations: opusTCv20210807_transformer-big_2022-03-23.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-03-23.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newstest2012.rus-fra	25.3	0.53679	3003	78011	0.990
newstest2013.rus-fra	29.7	0.56211	3000	70037	0.990
Tatoeba-test-v2021-08-07.bel-fra	46.4	0.65415	283	2005	1.000
Tatoeba-test-v2021-08-07.multi-fra	52.4	0.68422	10000	66671	0.978
Tatoeba-test-v2021-08-07.orv-fra	6.3	0.22337	37	290	0.902
Tatoeba-test-v2021-08-07.rus-fra	51.8	0.68699	11490	80573	0.973
Tatoeba-test-v2021-08-07.ukr-fra	50.7	0.67887	10035	63222	0.980