WMT
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
WMT (document-level) Low Resource Zh→X 24++
81.59COMET-22
12
WMT De⇒En 2021 (test)
32.26BLEU
12
WMT 2023 (test)
0.588MAE (EN→DE)
12
WMT ZH -> EN
60.7Pairwise Accuracy
12
WMT EN -> DE
51.6Pairwise Acc
12
WMT 24
0.606Pearson Correlation
12
WMT EN–TR 24
57.42chrF++
12
WMT En-Ja 2023 (test)
88.5COMET
12
WMT JA-EN 22
84COMET
12
WMT EN-ZH 22
88.2COMET
12
WMT 2023 (test)
82.3COMET
12
WMT ru-en
43.51BLEU
12
WMT cs-en
48.67BLEU
12
WMT newstest 2019
43.43BLEU
12
WMT English-to-German Average newstest2010-2016 2016
27.31BLEU
12
WMT English-to-German 2016 (newstest2015)
30.6BLEU
12
WMT English-to-German 2016 (newstest2014)
28.28BLEU
12
WMT English-to-German 2016 (newstest2013)
26.98BLEU
12
WMT English-to-German newstest2012 2016
23.09BLEU
12
WMT English-to-German 2016 (newstest2011)
22.77BLEU
12
WMT English-to-German 2016 (newstest2010)
0.249BLEU
12
WMT20 De-En (test)
44.73d-BLEU
12
WMT17 (test)
0.653Kendall Tau
12
WMT En-De sacreBLEU (test)
29.65SacreBLEU
12
WMT EN-FR_FR 24++
0.652SLI Iteration 1
11