opus2m-2020-08-12.yml

release: nic-eng/opus2m-2020-08-12.zip
release-date: 2020-08-12
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:
   - bam
   - ewe
   - fuc
   - fuv
   - ibo
   - kin
   - lin
   - lug
   - nya
   - run
   - sag
   - sna
   - swh
   - toi
   - tso
   - umb
   - wol
   - xho
   - yor
   - zul
target-languages:
   - eng
test-data:
   - Tatoeba-test.bam-eng: 7/13
   - Tatoeba-test.ewe-eng: 6/25
   - Tatoeba-test.ful-eng: 41/153
   - Tatoeba-test.ibo-eng: 21/126
   - Tatoeba-test.kin-eng: 17/93
   - Tatoeba-test.lin-eng: 28/204
   - Tatoeba-test.lug-eng: 2/11
   - Tatoeba-test.multi-eng: 2550/12391
   - Tatoeba-test.nya-eng: 22/106
   - Tatoeba-test.run-eng: 1703/8221
   - Tatoeba-test.sag-eng: 11/31
   - Tatoeba-test.sna-eng: 41/191
   - Tatoeba-test.swa-eng: 386/1993
   - Tatoeba-test.toi-eng: 2/11
   - Tatoeba-test.tso-eng: 3/12
   - Tatoeba-test.umb-eng: 32/93
   - Tatoeba-test.wol-eng: 7/23
   - Tatoeba-test.xho-eng: 152/783
   - Tatoeba-test.yor-eng: 35/163
   - Tatoeba-test.zul-eng: 34/139
BLEU-scores:
   - Tatoeba-test.bam-eng: 2.4
   - Tatoeba-test.ewe-eng: 10.3
   - Tatoeba-test.ful-eng: 1.2
   - Tatoeba-test.ibo-eng: 7.5
   - Tatoeba-test.kin-eng: 30.7
   - Tatoeba-test.lin-eng: 3.1
   - Tatoeba-test.lug-eng: 3.1
   - Tatoeba-test.multi-eng: 21.3
   - Tatoeba-test.nya-eng: 31.6
   - Tatoeba-test.run-eng: 24.9
   - Tatoeba-test.sag-eng: 5.2
   - Tatoeba-test.sna-eng: 20.1
   - Tatoeba-test.swa-eng: 4.6
   - Tatoeba-test.toi-eng: 4.8
   - Tatoeba-test.tso-eng: 100.0
   - Tatoeba-test.umb-eng: 9.0
   - Tatoeba-test.wol-eng: 14.0
   - Tatoeba-test.xho-eng: 38.2
   - Tatoeba-test.yor-eng: 21.2
   - Tatoeba-test.zul-eng: 42.3
chr-F-scores:
   - Tatoeba-test.bam-eng: 0.090
   - Tatoeba-test.ewe-eng: 0.384
   - Tatoeba-test.ful-eng: 0.114
   - Tatoeba-test.ibo-eng: 0.197
   - Tatoeba-test.kin-eng: 0.481
   - Tatoeba-test.lin-eng: 0.185
   - Tatoeba-test.lug-eng: 0.261
   - Tatoeba-test.multi-eng: 0.377
   - Tatoeba-test.nya-eng: 0.502
   - Tatoeba-test.run-eng: 0.420
   - Tatoeba-test.sag-eng: 0.231
   - Tatoeba-test.sna-eng: 0.374
   - Tatoeba-test.swa-eng: 0.191
   - Tatoeba-test.toi-eng: 0.122
   - Tatoeba-test.tso-eng: 1.000
   - Tatoeba-test.umb-eng: 0.246
   - Tatoeba-test.wol-eng: 0.212
   - Tatoeba-test.xho-eng: 0.558
   - Tatoeba-test.yor-eng: 0.364
   - Tatoeba-test.zul-eng: 0.589