itc-cel

opusTCv20210807_transformer-big_2022-08-12.zip

• dataset: opusTCv20210807
• model: transformer-big
• source language(s): fra ita lad_Latn lat_Latn mol pob por ron spa
• target language(s): bre cor cym gla gle
• raw source language(s): fra ita lad lat mol pob por ron spa
• raw target language(s): bre cor cym gla gle
• 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_2022-08-12.zip
• test set translations: opusTCv20210807_transformer-big_2022-08-12.test.txt
• test set scores: opusTCv20210807_transformer-big_2022-08-12.eval.txt

Benchmarks

testset	BLEU	chr-F	#sent	#words	BP
Tatoeba-test-v2021-08-07.fra-bre	2.7	0.18566	2494	14371	0.833
Tatoeba-test-v2021-08-07.fra-cor	0.9	0.12500	555	3196	1.000
Tatoeba-test-v2021-08-07.fra-cym	22.6	0.41011	29	161	0.909
Tatoeba-test-v2021-08-07.fra-gla	3.2	0.21036	173	1167	0.981
Tatoeba-test-v2021-08-07.fra-gle	14.2	0.37869	94	896	0.909
Tatoeba-test-v2021-08-07.ita-bre	6.0	0.24925	38	210	0.835
Tatoeba-test-v2021-08-07.ita-cor	0.5	0.12310	287	1538	1.000
Tatoeba-test-v2021-08-07.ita-cym	9.2	0.36767	59	325	0.873
Tatoeba-test-v2021-08-07.ita-gla	2.8	0.18699	173	1069	1.000
Tatoeba-test-v2021-08-07.lad-gle	7.1	0.30837	3	16	1.000
Tatoeba-test-v2021-08-07.lat-cym	4.6	0.16859	2	23	0.587
Tatoeba-test-v2021-08-07.multi-multi	3.9	0.18800	4564	27201	0.938
Tatoeba-test-v2021-08-07.por-cor	0.3	0.12570	63	392	1.000
Tatoeba-test-v2021-08-07.por-gla	3.7	0.17568	65	444	0.891
Tatoeba-test-v2021-08-07.ron-gle	23.6	0.44181	1	5	1.000
Tatoeba-test-v2021-08-07.spa-cor	0.3	0.11898	206	1174	1.000
Tatoeba-test-v2021-08-07.spa-cym	15.8	0.42490	22	137	1.000
Tatoeba-test-v2021-08-07.spa-gla	4.6	0.20008	289	2014	0.970
Tatoeba-test-v2021-08-07.spa-gle	15.8	0.34037	16	101	1.000