opus-2020-06-28.yml

release: eng-sem/opus-2020-06-28.zip
release-date: 2020-06-28
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
target-languages:
   - acm
   - afb
   - amh
   - apc
   - ara
   - arq
   - ary
   - arz
   - heb
   - mlt
   - phn
   - tir
   - tmr
use-target-labels:
   - >>acm<<
   - >>afb<<
   - >>amh<<
   - >>apc<<
   - >>apc_Latn<<
   - >>ara<<
   - >>ara_Latn<<
   - >>arq<<
   - >>arq_Latn<<
   - >>ary<<
   - >>arz<<
   - >>heb<<
   - >>mlt<<
   - >>phn_Phnx<<
   - >>tir<<
   - >>tmr_Hebr<<
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: 0/0
   - Tatoeba-test.eng-phn: 5/28
   - Tatoeba-test.eng-tir: 69/305
   - Tatoeba-test.eng-tmr: 19/76
BLEU-scores:
   - Tatoeba-test.eng-amh: 10.8
   - Tatoeba-test.eng-ara: 12.6
   - Tatoeba-test.eng-heb: 32.6
   - Tatoeba-test.eng-mlt: 17.9
   - Tatoeba-test.eng-multi: 22.8
   - Tatoeba-test.eng-phn: 0.5
   - Tatoeba-test.eng-tir: 2.5
   - Tatoeba-test.eng-tmr: 0.8
chr-F-scores:
   - Tatoeba-test.eng-amh: 0.499
   - Tatoeba-test.eng-ara: 0.421
   - Tatoeba-test.eng-heb: 0.557
   - Tatoeba-test.eng-mlt: 0.552
   - Tatoeba-test.eng-multi: 0.487
   - Tatoeba-test.eng-phn: 0.003
   - Tatoeba-test.eng-tir: 0.239
   - Tatoeba-test.eng-tmr: 0.003