opusTCv20210807_transformer-big_2022-07-22.yml

release: zlw-fin/opusTCv20210807_transformer-big_2022-07-22.zip
release-date: 2022-07-22
dataset-name: opusTCv20210807
modeltype: transformer-big
vocabulary:
   source: opusTCv20210807.spm32k-spm32k.vocab.yml
   target: opusTCv20210807.spm32k-spm32k.vocab.yml
pre-processing: normalization + SentencePiece (spm32k,spm32k)
subwords:
   source: spm32k
   target: spm32k
subword-models:
   source: source.spm
   target: target.spm
source-languages:
   - ces
   - pol
target-languages:
   - fin
raw-source-languages:
   - ces
   - pol
raw-target-languages:
   - fin
training-data:
   ces-fin: Tatoeba-train-v2021-08-07.ces-fin.strict (43470575) 
   pol-fin: Tatoeba-train-v2021-08-07.fin-pol.strict (42191613) 
validation-data:
   ces-fin: Tatoeba-dev-v2021-08-07, 1001
   fin-pol: Tatoeba-dev-v2021-08-07, 926
   total-size-shuffled: 1923
   devset-selected: top 1923  lines of Tatoeba-dev-v2021-08-07.src.shuffled
test-data:
   Tatoeba-test-v2021-08-07.ces-fin: 88/408
   Tatoeba-test-v2021-08-07.multi-fin: 697/3701
   Tatoeba-test-v2021-08-07.pol-fin: 609/3293
BLEU-scores:
   Tatoeba-test-v2021-08-07.ces-fin: 62.2
   Tatoeba-test-v2021-08-07.multi-fin: 46.3
   Tatoeba-test-v2021-08-07.pol-fin: 44.4
chr-F-scores:
   Tatoeba-test-v2021-08-07.ces-fin: 0.72184
   Tatoeba-test-v2021-08-07.multi-fin: 0.66927
   Tatoeba-test-v2021-08-07.pol-fin: 0.66230