opus-2021-02-19.yml

release: spa-zho/opus-2021-02-19.zip
release-date: 2021-02-19
dataset-name: opus
modeltype: transformer
pre-processing: normalization + SentencePiece (spm32k,spm32k)
subwords:
   source: spm32k
   target: spm32k
subword-models:
   source: source.spm
   target: target.spm
use-target-labels:
   - ">>cjy_Hans<<"
   - ">>cjy_Hant<<"
   - ">>cmn<<"
   - ">>cmn_Hans<<"
   - ">>cmn_Hant<<"
   - ">>hsn<<"
   - ">>hsn_Hani<<"
   - ">>lzh<<"
   - ">>nan<<"
   - ">>wuu<<"
   - ">>yue_Hans<<"
   - ">>yue_Hant<<"
source-languages:
   - spa
target-languages:
   - cjy
   - cmn
   - hsn
   - lzh
   - nan
   - wuu
   - yue
training-data:
   spa-cmn: Tatoeba-train (286314) 
   spa-cmn_Hans: Tatoeba-train (32075453) 
   spa-cmn_Hant: Tatoeba-train (6041422) 
   spa-nan: Tatoeba-train (77) 
   spa-yue_Hans: Tatoeba-train (347) 
   spa-yue_Hant: Tatoeba-train (59293) 
validation-data:
   spa-zho: Tatoeba-dev, 6921
   total-size-shuffled: 6916
   devset-selected: top 5000  lines of Tatoeba-dev.src.shuffled!
test-data:
   Tatoeba-test.spa-zho: 2500/22762
BLEU-scores:
   Tatoeba-test.spa-zho: 38.8
chr-F-scores:
   Tatoeba-test.spa-zho: 0.324