sla-por

opusTCv20210807_transformer-big_2022-09-15.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): bel bel_Latn bos_Cyrl bos_Latn bul ces dsb hbs hbs_Cyrl hrv hsb mkd pol rus slv srp_Cyrl srp_Latn ukr
• target language(s): pob por
• raw source language(s): bel bos bul ces dsb hbs hrv hsb mkd pol rus slv srp ukr
• raw target language(s): pob por
• 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_2022-09-15.zip
• test set translations: opusTCv20210807_transformer-big_2022-09-15.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-09-15.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
Tatoeba-test-v2021-08-07.bel-por	65.6	0.74422	3	21	1.000
Tatoeba-test-v2021-08-07.bul-por	40.7	0.61337	169	1117	0.960
Tatoeba-test-v2021-08-07.ces-por	50.2	0.70326	133	919	0.968
Tatoeba-test-v2021-08-07.dsb-por	100.0	10.00000	1	5	1.000
Tatoeba-test-v2021-08-07.hbs-por	42.0	0.59319	107	626	0.932
Tatoeba-test-v2021-08-07.hrv-por	40.9	0.58407	105	612	0.931
Tatoeba-test-v2021-08-07.hsb-por	100.0	10.00000	1	5	1.000
Tatoeba-test-v2021-08-07.mkd-por	47.2	0.64919	3	21	1.000
Tatoeba-test-v2021-08-07.multi-multi	43.4	0.63921	10000	71535	0.977
Tatoeba-test-v2021-08-07.pol-por	43.0	0.63610	705	5059	0.971
Tatoeba-test-v2021-08-07.rus-por	42.2	0.63384	10000	74705	0.964
Tatoeba-test-v2021-08-07.slv-por	49.5	0.66102	6	28	0.805
Tatoeba-test-v2021-08-07.srp_Cyrl-por	82.5	0.86369	2	14	1.000
Tatoeba-test-v2021-08-07.ukr-por	43.7	0.64514	3372	21301	1.000