opus2m-2020-08-01.yml

release: eng-sem/opus2m-2020-08-01.zip
release-date: 2020-08-01
dataset-name: opus2m
modeltype: transformer
pre-processing: normalization + SentencePiece (spm32k,spm32k)
subwords:
   source: spm32k
   target: spm32k
subword-models:
   source: source.spm
   target: target.spm
source-languages:
   - eng
target-languages:
   - acm
   - afb
   - amh
   - apc
   - ara
   - arq
   - ary
   - arz
   - heb
   - mlt
   - tir
use-target-labels:
   - ">>acm<<"
   - ">>afb<<"
   - ">>amh<<"
   - ">>apc<<"
   - ">>ara<<"
   - ">>arq<<"
   - ">>ary<<"
   - ">>arz<<"
   - ">>heb<<"
   - ">>mlt<<"
   - ">>tir<<"
test-data:
   Tatoeba-test.eng-amh: 190/554
   Tatoeba-test.eng-ara: 10000/50193
   Tatoeba-test.eng-heb: 10000/49113
   Tatoeba-test.eng-mlt: 203/673
   Tatoeba-test.eng-multi: 10000/49328
   Tatoeba-test.eng-tir: 69/305
BLEU-scores:
   Tatoeba-test.eng-amh: 11.2
   Tatoeba-test.eng-ara: 12.7
   Tatoeba-test.eng-heb: 33.8
   Tatoeba-test.eng-mlt: 18.7
   Tatoeba-test.eng-multi: 23.5
   Tatoeba-test.eng-tir: 2.7
chr-F-scores:
   Tatoeba-test.eng-amh: 0.480
   Tatoeba-test.eng-ara: 0.417
   Tatoeba-test.eng-heb: 0.564
   Tatoeba-test.eng-mlt: 0.554
   Tatoeba-test.eng-multi: 0.486
   Tatoeba-test.eng-tir: 0.248