zho-gmq

opusTCv20210807_transformer-big_2022-07-23.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): cmn cmn_Hans cmn_Hant nan wuu yue_Hans yue_Hant
• target language(s): dan fao isl nno nob swe
• raw source language(s): cmn nan wuu yue
• raw target language(s): dan fao isl nno nob swe
• model: transformer-big
• 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:
• download: opusTCv20210807_transformer-big_2022-07-23.zip
• test set translations: opusTCv20210807_transformer-big_2022-07-23.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-07-23.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
Tatoeba-test-v2021-08-07.cmn-dan	100.0	10.00000	1	20	1.000
Tatoeba-test-v2021-08-07.cmn_Hans-dan	48.3	0.64401	66	404	0.975
Tatoeba-test-v2021-08-07.cmn_Hans-isl	27.5	0.50260	7	41	0.950
Tatoeba-test-v2021-08-07.cmn_Hans-nno	4.2	0.16364	3	29	0.727
Tatoeba-test-v2021-08-07.cmn_Hans-nob	34.4	0.54056	91	738	0.933
Tatoeba-test-v2021-08-07.cmn_Hans-swe	40.2	0.57333	478	2987	0.849
Tatoeba-test-v2021-08-07.cmn_Hant-dan	49.2	0.64658	91	591	0.964
Tatoeba-test-v2021-08-07.cmn_Hant-isl	4.7	0.31555	5	39	0.834
Tatoeba-test-v2021-08-07.cmn_Hant-nno	5.8	0.25861	3	15	1.000
Tatoeba-test-v2021-08-07.cmn_Hant-nob	35.0	0.53171	104	795	0.859
Tatoeba-test-v2021-08-07.cmn_Hant-swe	42.5	0.60566	567	3185	0.886
Tatoeba-test-v2021-08-07.cmn-swe	86.8	0.78451	4	50	1.000
Tatoeba-test-v2021-08-07.multi-multi	41.2	0.58304	1421	8921	0.884
Tatoeba-test-v2021-08-07.zho-dan	50.4	0.64986	158	1015	0.969
Tatoeba-test-v2021-08-07.zho-fao	0.2	0.12183	1	27	0.030
Tatoeba-test-v2021-08-07.zho-isl	16.1	0.41961	12	80	0.895
Tatoeba-test-v2021-08-07.zho-nor	33.9	0.52814	201	1577	0.893
Tatoeba-test-v2021-08-07.zho-swe	42.0	0.59048	1049	6222	0.870