Neurora/opus-tatoeba-ces-eng

Opus Tatoeba | Czech -> English

• dataset: opus
• model: transformer
• source language(s): ces
• 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.ces-eng	27.7	0.551	502	11821	0.971
newstest2009.ces-eng	27.2	0.550	2525	65402	0.970
newstest2010.ces-eng	27.3	0.559	2489	61724	0.978
newstest2011.ces-eng	28.0	0.557	3003	74681	0.990
newstest2012.ces-eng	27.2	0.552	3003	72812	1.000
newstest2013.ces-eng	30.7	0.572	3000	64505	1.000
newstest2014-csen.ces-eng	34.2	0.614	3003	68065	0.999
newstest2015-encs.ces-eng	30.7	0.568	2656	53572	0.975
newstest2016-encs.ces-eng	32.4	0.589	2999	64670	0.998
newstest2017-encs.ces-eng	28.9	0.559	3005	61725	0.996
newstest2018-encs.ces-eng	30.4	0.568	2983	63496	0.991
Tatoeba-test.ces-eng	56.9	0.719	10000	75376	0.962