spa-eng

opus-2021-02-19.zip

• dataset: opus
• model: transformer
• source language(s): spa
• target language(s): eng
• model: transformer
• 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.spa-eng	30.6	0.570	502	11821	1.000
news-test2008.spa-eng	27.9	0.553	2051	49380	1.000
newstest2009.spa-eng	30.4	0.572	2525	65402	0.986
newstest2010.spa-eng	36.1	0.614	2489	61724	1.000
newstest2011.spa-eng	34.2	0.599	3003	74681	1.000
newstest2012.spa-eng	37.9	0.624	3003	72812	1.000
newstest2013.spa-eng	35.3	0.609	3000	64505	1.000
Tatoeba-test.spa-eng	59.6	0.739	10000	79376	0.974
tico19-test.spa-eng	48.6	0.736	2100	56339	1.000

opus+bt-2021-04-30.zip

• dataset: opus+bt
• model: transformer-align
• source language(s): spa
• 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.spa-eng	30.6	0.571	502	11821	0.995
news-test2008.spa-eng	27.9	0.553	2051	49380	1.000
newstest2009.spa-eng	30.3	0.571	2525	65402	0.984
newstest2010.spa-eng	36.2	0.616	2489	61724	0.998
newstest2011.spa-eng	34.3	0.598	3003	74681	1.000
newstest2012.spa-eng	38.0	0.624	3003	72812	1.000
newstest2013.spa-eng	35.6	0.610	3000	64505	1.000
Tatoeba-test.spa-eng	59.7	0.740	10000	79376	0.974
tico19-test.spa-eng	49.3	0.738	2100	56339	1.000