opus1m+bt-tuned4eng2bel-2021-04-16.yml

release: eng-zle/opus1m+bt-tuned4eng2bel-2021-04-16.zip
release-date: 2021-04-16
dataset-name: opus1m+bt-tuned4eng2bel
modeltype: transformer-align
vocabulary:
   source: opus1m+bt-tuned4eng2bel.spm32k-spm32k.vocab.yml
   target: opus1m+bt-tuned4eng2bel.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:
   - eng
target-languages:
   - bel
use-target-labels:
   - ">>bel<<"
   - ">>bel_Latn<<"
training-data:
   eng-bel: Tatoeba-train (154954) wiki.aa (966635) wiki.ab (894574) wikibooks.aa (1629) wikiquote.aa (1489) wikisource.aa (154050) 
   eng-bel_Latn: Tatoeba-train (2570) 
validation-data:
   bel-eng: Tatoeba-dev, 4241
   bel_Latn-eng: Tatoeba-dev, 12
   total-size-shuffled: 1012
   devset-selected: top 1012  lines of Tatoeba-dev.eng-bel.src.shuffled
test-data:
   Tatoeba-test.eng-bel: 2500/16237
   Tatoeba-test.eng-multi: 10000/63596
BLEU-scores:
   Tatoeba-test.eng-bel: 24.1
   Tatoeba-test.eng-multi: 4.1
chr-F-scores:
   Tatoeba-test.eng-bel: 0.502
   Tatoeba-test.eng-multi: 0.206