opus-2021-02-18.yml

release: jpn-rus/opus-2021-02-18.zip
release-date: 2021-02-18
dataset-name: opus
modeltype: transformer-align
pre-processing: normalization + SentencePiece (spm32k,spm32k)
subwords:
   - source: spm32k
   - target: spm32k
subword-models:
   - source: source.spm
   - target: target.spm
source-languages:
   - jpn
target-languages:
   - rus
training-data:
   - jpn-rus: Tatoeba-train (963348) 
   - jpn_Bopo-rus: Tatoeba-train (3080) 
   - jpn_Hani-rus: Tatoeba-train (246632) 
   - jpn_Hira-rus: Tatoeba-train (1344304) 
   - jpn_Kana-rus: Tatoeba-train (228327) 
   - jpn_Latn-rus: Tatoeba-train (16258) 
   - jpn_Yiii-rus: Tatoeba-train (6031) 
validation-data:
   - jpn-rus: Tatoeba-dev, 15301
   - total size of shuffled dev data: 15301
   - devset = top 5000  lines of Tatoeba-dev.src.shuffled!
test-data:
   - Tatoeba-test.jpn-rus: 10000/70820
BLEU-scores:
   - Tatoeba-test.jpn-rus: 23.2
chr-F-scores:
   - Tatoeba-test.jpn-rus: 0.441