Name		Name	Last commit message	Last commit date
parent directory ..
README.md		README.md
opus+bt-2021-04-30.yml		opus+bt-2021-04-30.yml
opus+bt.yml		opus+bt.yml
opus-2020-07-03.yml		opus-2020-07-03.yml
opus-2021-02-19.yml		opus-2021-02-19.yml
opus-2021-02-22.yml		opus-2021-02-22.yml
opus.yml		opus.yml

README.md

opus-2020-07-03.zip

dataset: opus
model: transformer
source language(s): deu
target language(s): eng
model: transformer
pre-processing: normalization + SentencePiece (spm32k,spm32k)
download: opus-2020-07-03.zip
test set translations: opus-2020-07-03.test.txt
test set scores: opus-2020-07-03.eval.txt

Benchmarks

testset	BLEU	chr-F
newssyscomb2009-deueng.deu.eng	28.6	0.553
news-test2008-deueng.deu.eng	27.9	0.549
newstest2009-deueng.deu.eng	26.8	0.544
newstest2010-deueng.deu.eng	30.5	0.587
newstest2011-deueng.deu.eng	27.5	0.556
newstest2012-deueng.deu.eng	29.2	0.569
newstest2013-deueng.deu.eng	32.2	0.584
newstest2014-deen-deueng.deu.eng	34.0	0.600
newstest2015-ende-deueng.deu.eng	34.2	0.599
newstest2016-ende-deueng.deu.eng	40.7	0.652
newstest2017-ende-deueng.deu.eng	35.9	0.612
newstest2018-ende-deueng.deu.eng	44.1	0.670
newstest2019-deen-deueng.deu.eng	40.0	0.642
Tatoeba-test.deu.eng	53.1	0.693

opus-2021-02-12.zip

dataset: opus
model: transformer
source language(s): deu
target language(s): eng
model: transformer
pre-processing: normalization + SentencePiece (spm32k,spm32k)
download: opus-2021-02-12.zip
test set translations: opus-2021-02-12.test.txt
test set scores: opus-2021-02-12.eval.txt

Benchmarks

testset	BLEU	chr-F
newssyscomb2009-deueng.deu.eng	28.8	0.557
news-test2008-deueng.deu.eng	28.0	0.551
newstest2009-deueng.deu.eng	27.1	0.546
newstest2010-deueng.deu.eng	30.8	0.588
newstest2011-deueng.deu.eng	27.5	0.556
newstest2012-deueng.deu.eng	29.4	0.570
newstest2013-deueng.deu.eng	32.3	0.585
newstest2014-deen-deueng.deu.eng	34.6	0.603
newstest2015-ende-deueng.deu.eng	34.6	0.602
newstest2016-ende-deueng.deu.eng	40.9	0.653
newstest2017-ende-deueng.deu.eng	36.3	0.615
newstest2018-ende-deueng.deu.eng	44.6	0.674
newstest2019-deen-deueng.deu.eng	40.5	0.645
Tatoeba-test.deu.eng	53.4	0.696
Tatoeba-test.deu-eng.deu.eng	53.4	0.696

opus-2021-02-17.zip

dataset: opus
model: transformer-align
source language(s): deu
target language(s): eng
model: transformer-align
pre-processing: normalization + SentencePiece (spm32k,spm32k)
download: opus-2021-02-17.zip
test set translations: opus-2021-02-17.test.txt
test set scores: opus-2021-02-17.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newssyscomb2009.deu-eng	29.0	0.555	502	11821	0.997
news-test2008.deu-eng	27.8	0.549	2051	49380	1.000
newstest2009.deu-eng	27.1	0.546	2525	65402	0.985
newstest2010.deu-eng	30.6	0.587	2489	61724	1.000
newstest2011.deu-eng	27.6	0.556	3003	74681	1.000
newstest2012.deu-eng	29.1	0.569	3003	72812	1.000
newstest2013.deu-eng	32.3	0.584	3000	64505	1.000
newstest2014-deen.deu-eng	34.2	0.601	3003	67337	0.981
newstest2015-ende.deu-eng	34.3	0.600	2169	46443	0.980
newstest2016-ende.deu-eng	40.4	0.650	2999	64126	1.000
newstest2017-ende.deu-eng	35.9	0.613	3004	64399	0.987
newstest2018-ende.deu-eng	44.1	0.670	2998	67013	0.983
newstest2019-deen.deu-eng	40.2	0.644	2000	39282	1.000
Tatoeba-test.deu-eng	53.2	0.695	10000	81233	0.974

opus-2021-02-19.zip

dataset: opus
model: transformer-align
source language(s): deu
target language(s): eng
model: transformer-align
pre-processing: normalization + SentencePiece (spm32k,spm32k)
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
newssyscomb2009.deu-eng	28.9	0.554	502	11821	0.994
news-test2008.deu-eng	27.8	0.550	2051	49380	1.000
newstest2009.deu-eng	27.1	0.545	2525	65402	0.984
newstest2010.deu-eng	30.6	0.587	2489	61724	1.000
newstest2011.deu-eng	27.7	0.557	3003	74681	1.000
newstest2012.deu-eng	29.3	0.570	3003	72812	1.000
newstest2013.deu-eng	32.4	0.584	3000	64505	1.000
newstest2014-deen.deu-eng	34.4	0.602	3003	67337	0.980
newstest2015-ende.deu-eng	34.3	0.601	2169	46443	0.979
newstest2016-ende.deu-eng	40.4	0.651	2999	64126	1.000
newstest2017-ende.deu-eng	36.1	0.613	3004	64399	0.987
newstest2018-ende.deu-eng	44.4	0.671	2998	67013	0.985
newstest2019-deen.deu-eng	40.3	0.645	2000	39282	1.000
Tatoeba-test.deu-eng	53.3	0.696	10000	81233	0.974

opus-2021-02-22.zip

dataset: opus
model: transformer-align
source language(s): deu
target language(s): eng
model: transformer-align
pre-processing: normalization + SentencePiece (spm32k,spm32k)
download: opus-2021-02-22.zip
test set translations: opus-2021-02-22.test.txt
test set scores: opus-2021-02-22.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newssyscomb2009.deu-eng	28.9	0.555	502	11821	0.997
news-test2008.deu-eng	28.1	0.551	2051	49380	1.000
newstest2009.deu-eng	27.1	0.546	2525	65402	0.985
newstest2010.deu-eng	30.7	0.588	2489	61724	1.000
newstest2011.deu-eng	27.6	0.557	3003	74681	1.000
newstest2012.deu-eng	29.3	0.571	3003	72812	1.000
newstest2013.deu-eng	32.2	0.584	3000	64505	1.000
newstest2014-deen.deu-eng	34.3	0.602	3003	67337	0.980
newstest2015-ende.deu-eng	34.5	0.603	2169	46443	0.980
newstest2016-ende.deu-eng	40.6	0.652	2999	64126	1.000
newstest2017-ende.deu-eng	36.2	0.614	3004	64399	0.987
newstest2018-ende.deu-eng	44.3	0.672	2998	67013	0.983
newstest2019-deen.deu-eng	41.0	0.649	2000	39282	1.000
Tatoeba-test.deu-eng	53.5	0.697	10000	81233	0.974

opus+bt-2021-04-30.zip

dataset: opus+bt
model: transformer-align
source language(s): deu
target language(s): eng
model: transformer-align
pre-processing: normalization + SentencePiece (spm32k,spm32k)
download: opus+bt-2021-04-30.zip
test set translations: opus+bt-2021-04-30.test.txt
test set scores: opus+bt-2021-04-30.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newssyscomb2009.deu-eng	29.0	0.555	502	11821	0.996
news-test2008.deu-eng	27.8	0.549	2051	49380	1.000
newstest2009.deu-eng	26.7	0.544	2525	65402	0.983
newstest2010.deu-eng	30.4	0.585	2489	61724	1.000
newstest2011.deu-eng	27.2	0.554	3003	74681	1.000
newstest2012.deu-eng	29.0	0.568	3003	72812	1.000
newstest2013.deu-eng	32.0	0.582	3000	64505	1.000
newstest2014-deen.deu-eng	34.2	0.601	3003	67337	0.980
newstest2015-ende.deu-eng	34.4	0.601	2169	46443	0.979
newstest2016-ende.deu-eng	40.9	0.653	2999	64126	1.000
newstest2017-ende.deu-eng	36.1	0.613	3004	64399	0.984
newstest2018-ende.deu-eng	44.6	0.672	2998	67013	0.982
newstest2019-deen.deu-eng	40.3	0.645	2000	39282	1.000
Tatoeba-test.deu-eng	53.1	0.694	10000	81233	0.975