[restart] [share link] [json] select language: [swap] [compare scores] [compare models] [map] [release history] [uploads]

OPUS-MT Dashboard

Model Scores (comparing between OPUS-MT and external models)

IDBenchmark (bleu)OutputOPUS-MTbleuexternalbleuDiff
0flores101-devtestcompareeng-deu/opus...2021-12-0840facebook/wmt19-en-de40.9-0.9
1flores200-devtestcompareeng-deu/opus...2021-12-0840facebook/wmt19-en-de40.9-0.9
2generaltest2022compareeng-deu/opus...2021-12-0846.4facebook/nll...illed-1.3B43.7 2.7
3multi30k_task2_test_2016comparegmw-gmw/opus...2022-08-112.7facebook/wmt19-en-de3.2-0.5
4multi30k_test_2016_flickrcompareen-de/opus-2020-02-2636.3facebook/wmt19-en-de39.7-3.4
5multi30k_test_2017_flickrcompareeng-deu/opus-2021-02-2235.8facebook/wmt19-en-de39.5-3.7
6multi30k_test_2017_mscococomparegmw-deu+eng+...2024-05-3030.7facebook/wmt19-en-de34.2-3.5
7multi30k_test_2018_flickrcompareeng-deu/opus-2021-02-1931.6facebook/wmt19-en-de36.7-5.1
8newssyscomb2009compareeng-deu/opus...2021-12-0824.2facebook/wmt19-en-de25.5-1.3
9newstest2008compareeng-deu/opus...2021-12-0824.7facebook/nllb-200-3.3B23.1 1.6
10newstest2009compareeng-deu/opus...2021-12-0823.6facebook/wmt19-en-de24.6-1.0
11newstest2010compareeng-deu/opus...2021-12-0826.8facebook/wmt19-en-de28-1.2
12newstest2011compareeng-deu/opus...2021-12-0824facebook/wmt19-en-de25.3-1.3
13newstest2012compareeng-deu/opus...2021-12-0824.7facebook/wmt19-en-de29.4-4.7
14newstest2013compareeng-deu/opus...2021-12-0828.3facebook/wmt19-en-de32.8-4.5
15newstest2014compareeng-deu/opus...2021-12-0830.4facebook/wmt19-en-de35-4.6
16newstest2015compareeng-deu/opus...2021-12-0833.3facebook/wmt19-en-de40.7-7.4
17newstest2016compareeng-deu/opus...2021-12-0839.8facebook/wmt19-en-de40.6-0.8
18newstest2017compareeng-deu/opus...2021-12-0831.9facebook/wmt19-en-de38.7-6.8
19newstest2018compareeng-deu/opus...2021-12-0848.8gsarti/opus-...-big-en-de48.6 0.2
20newstest2019compareeng-deu/opus...2021-12-0844.6gsarti/opus-...-big-en-de44.3 0.3
21newstest2020compareeng-deu/opus...2021-12-0834.6facebook/wmt19-en-de36.6-2.0
22newstest2021compareeng-deu/opus...2021-12-0852.8facebook/nll...illed-1.3B49.8 3.0
23newstestALL2020compareeng-deu/opus...2021-12-0847.9facebook/nllb-200-1.3B42 5.9
24newstestB2020compareeng-deu/opus...2021-12-0834.4facebook/wmt19-en-de36.8-2.4
25ntrex128compareeng-deu/opus...2021-12-0831.7facebook/nll...illed-1.3B30.4 1.3
26tatoeba-test-v2020-07-28compareen-de/opus-2020-02-2646.7facebook/nll...illed-1.3B46.5 0.2
27tatoeba-test-v2021-03-30compareen-de/opus-2020-02-2645.5facebook/nll...illed-1.3B45.5 0.0
28tatoeba-test-v2021-08-07compareen-de/opus-2020-02-2644.7facebook/nll...illed-1.3B44.8-0.1
average34.736.1-1.4