opus-2020-10-04.yml

release: taw-taw/opus-2020-10-04.zip
release-date: 2020-10-04
dataset-name: opus
modeltype: transformer
pre-processing: normalization + SentencePiece (spm32k,spm32k)
subwords:
   source: spm32k
   target: spm32k
subword-models:
   source: source.spm
   target: target.spm
source-languages:
   - eng
   - lao
   - tha
target-languages:
   - eng
   - lao
   - tha
use-target-labels:
   - ">>eng<<"
   - ">>lao<<"
   - ">>tha<<"
test-data:
   Tatoeba-test.eng-lao: 36/44
   Tatoeba-test.eng-tha: 1145/1338
   Tatoeba-test.lao-eng: 36/182
   Tatoeba-test.multi-multi: 2362/7679
   Tatoeba-test.tha-eng: 1145/6115
BLEU-scores:
   Tatoeba-test.eng-lao: 0.2
   Tatoeba-test.eng-tha: 2.3
   Tatoeba-test.lao-eng: 4.5
   Tatoeba-test.multi-multi: 32.0
   Tatoeba-test.tha-eng: 40.5
chr-F-scores:
   Tatoeba-test.eng-lao: 0.304
   Tatoeba-test.eng-tha: 0.516
   Tatoeba-test.lao-eng: 0.200
   Tatoeba-test.multi-multi: 0.541
   Tatoeba-test.tha-eng: 0.589