opus-2020-07-26.yml

release: afa-eng/opus-2020-07-26.zip
release-date: 2020-07-26
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:
   - acm
   - afb
   - amh
   - apc
   - ara
   - arq
   - ary
   - arz
   - hau
   - heb
   - kab
   - mlt
   - rif
   - shy
   - som
   - tir
target-languages:
   - eng
test-data:
   Tatoeba-test.amh-eng: 190/802
   Tatoeba-test.ara-eng: 10000/62449
   Tatoeba-test.hau-eng: 56/378
   Tatoeba-test.heb-eng: 10000/62332
   Tatoeba-test.kab-eng: 10000/49297
   Tatoeba-test.mlt-eng: 203/935
   Tatoeba-test.multi-eng: 10000/58147
   Tatoeba-test.rif-eng: 53/198
   Tatoeba-test.shy-eng: 16/78
   Tatoeba-test.som-eng: 1/2
   Tatoeba-test.tir-eng: 69/426
BLEU-scores:
   Tatoeba-test.amh-eng: 35.1
   Tatoeba-test.ara-eng: 34.5
   Tatoeba-test.hau-eng: 11.5
   Tatoeba-test.heb-eng: 41.2
   Tatoeba-test.kab-eng: 3.7
   Tatoeba-test.mlt-eng: 40.5
   Tatoeba-test.multi-eng: 25.4
   Tatoeba-test.rif-eng: 3.5
   Tatoeba-test.shy-eng: 1.4
   Tatoeba-test.som-eng: 16.2
   Tatoeba-test.tir-eng: 13.9
chr-F-scores:
   Tatoeba-test.amh-eng: 0.553
   Tatoeba-test.ara-eng: 0.526
   Tatoeba-test.hau-eng: 0.307
   Tatoeba-test.heb-eng: 0.578
   Tatoeba-test.kab-eng: 0.203
   Tatoeba-test.mlt-eng: 0.586
   Tatoeba-test.multi-eng: 0.450
   Tatoeba-test.rif-eng: 0.143
   Tatoeba-test.shy-eng: 0.141
   Tatoeba-test.som-eng: 0.285
   Tatoeba-test.tir-eng: 0.337