opus+bt-2021-04-30.yml

release: ido-eng/opus+bt-2021-04-30.zip
release-date: 2021-04-30
dataset-name: opus+bt
modeltype: transformer-align
vocabulary:
   source: opus+bt.spm32k-spm32k.vocab.yml
   target: opus+bt.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:
   - ido
target-languages:
   - eng
training-data:
   ido-eng: Tatoeba-train (244) wikibooks.aa (992137) wikinews.aa (457241) wikipedia.aa (982423) wikipedia.ab (982571) wikipedia.ac (982558) wikipedia.ad (982417) wikiquote.aa (996971) 
   ido_Latn-eng: Tatoeba-train (35310) 
validation-data:
   eng-ido: Tatoeba-dev, 1000
   eng-ido_Latn: Tatoeba-dev, 1000
   total-size-shuffled: 996
   devset-selected: top 996  lines of Tatoeba-dev.src.shuffled!
test-data:
   Tatoeba-test.ido_Latn-eng: 1967/14114
   Tatoeba-test.ido-eng: 1968/14126
BLEU-scores:
   Tatoeba-test.ido_Latn-eng: 31.0
   Tatoeba-test.ido-eng: 31.2
chr-F-scores:
   Tatoeba-test.ido_Latn-eng: 0.495
   Tatoeba-test.ido-eng: 0.499