spa-zle

opusTCv20210807_transformer-big_2022-03-14.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): spa
• target language(s): bel bel_Latn orv_Cyrl rue rus ukr
• raw source language(s): spa
• raw target language(s): bel orv rue rus ukr
• 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_2022-03-14.zip
• test set translations: opusTCv20210807_transformer-big_2022-03-14.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-03-14.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newstest2012.spa-rus	23.9	0.52017	3003	64830	0.990
newstest2013.spa-rus	26.1	0.53548	3000	58560	0.993
Tatoeba-test-v2021-08-07.spa-bel	23.3	0.50542	205	1259	1.000
Tatoeba-test-v2021-08-07.spa-bel_Latn	6.6	1.174	1	8	1.000
Tatoeba-test-v2021-08-07.spa-multi	46.2	0.66158	10000	59935	0.994
Tatoeba-test-v2021-08-07.spa-orv	0.9	0.13927	33	142	1.000
Tatoeba-test-v2021-08-07.spa-rue	1.6	0.18533	97	319	1.000
Tatoeba-test-v2021-08-07.spa-rus	47.1	0.67116	10506	69028	0.998
Tatoeba-test-v2021-08-07.spa-ukr	39.0	0.61030	10115	54407	1.000

opusTCv20210807_transformer-big_2022-03-17.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): spa
• target language(s): bel bel_Latn orv_Cyrl rue rus ukr
• raw source language(s): spa
• raw target language(s): bel orv rue rus ukr
• 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_2022-03-17.zip
• test set translations: opusTCv20210807_transformer-big_2022-03-17.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-03-17.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newstest2012.spa-rus	23.9	0.52017	3003	64830	0.990
newstest2013.spa-rus	26.1	0.53548	3000	58560	0.993
Tatoeba-test-v2021-08-07.spa-bel	23.3	0.50542	205	1259	1.000
Tatoeba-test-v2021-08-07.spa-bel_Latn	6.6	1.174	1	8	1.000
Tatoeba-test-v2021-08-07.spa-multi	46.5	0.66422	10000	59935	0.994
Tatoeba-test-v2021-08-07.spa-orv	0.9	0.13927	33	142	1.000
Tatoeba-test-v2021-08-07.spa-rue	1.6	0.18533	97	319	1.000
Tatoeba-test-v2021-08-07.spa-rus	47.1	0.67116	10506	69028	0.998
Tatoeba-test-v2021-08-07.spa-ukr	39.0	0.61030	10115	54407	1.000

opusTCv20210807_transformer-big_2022-03-23.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): spa
• target language(s): bel bel_Latn orv_Cyrl rue rus ukr
• raw source language(s): spa
• raw target language(s): bel orv rue rus ukr
• 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_2022-03-23.zip
• test set translations: opusTCv20210807_transformer-big_2022-03-23.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-03-23.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
newstest2012.spa-rus	24.6	0.52441	3003	64830	0.985
newstest2013.spa-rus	27.0	0.54250	3000	58560	0.984
Tatoeba-test-v2021-08-07.spa-bel	27.4	0.54428	205	1259	0.998
Tatoeba-test-v2021-08-07.spa-bel_Latn	6.6	1.174	1	8	1.000
Tatoeba-test-v2021-08-07.spa-multi	46.5	0.66543	10000	59935	0.993
Tatoeba-test-v2021-08-07.spa-orv	0.9	0.13646	33	142	1.000
Tatoeba-test-v2021-08-07.spa-rue	1.5	0.17914	97	319	1.000
Tatoeba-test-v2021-08-07.spa-rus	48.5	0.68233	10506	69028	0.992
Tatoeba-test-v2021-08-07.spa-ukr	42.1	0.63405	10115	54407	0.999