WMT
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
WMT16 En-De
56.67Accuracy
2
WMT En-Fr 14
62.41Accuracy
2
WMT en-de 14
0.848Acceptance Rate
2
WMT Zh-En (subset of 600 samples) 2022
0.46Kendall Correlation
2
WMT newstest 2016
35.37BLEU
2
WMT newstest 2011
23.27BLEU
2
WMT news 2010 (test)
25.26BLEU
2
WMT Bio Average 2020
5.59MAE
2
WMT Bio zh-en 2020
5.18MAE
2
WMT Bio (ru-en) 2020
5.51MAE
2
WMT Bio (it-en) 2020
5.9MAE
2
WMT Bio (es-en) 2020
6.23MAE
2
WMT Bio en-zh 2020
2.17MAE
2
WMT Bio (en-ru) 2020
4.16MAE
2
WMT Bio en-it 2020
5.94MAE
2
WMT Bio en-es 2020
6.31MAE
2
WMT Bio (en-de) 2020
5.69MAE
2
WMT Bio de-en 2020
5.41MAE
2
WMT fr-en 2014 (test)
35.56BLEU Score
2
WMT Lithuanian-English 19
35.6BLEU
2
WMT Finnish-English (fi-en) 19
31.9BLEU
2
WMT German-English (de-en) 19
42.8BLEU
2
WMT20 EN-JA (test)
20.3BLEU
2
WMT English-German Permuted input '17 newstest14 (test)
83.7EoE
2
WMT 2018 (test)
21.9Score (En->Et)
2