WMT
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
WMT En-Ro 16 (test)
38BLEU
5
WMT De→En Top 30% 2019
0.883Pearson Correlation (|r|)
5
WMT De→En 2019 (All)
0.951Pearson Correlation (|r|)
5
WMT En→De 2019 (Top 30%)
0.974Pearson Correlation (|r|)
5
WMT En→De 2019 (all)
0.991|r|
5
WMT Chinese-English (Zh-En) 2019 20.4M sentence pairs (test)
27.61BLEU
5
WMT Ro-En 2016 (newstest)
39.1BLEU Score
5
WMT x86 d=16 1.5.0 (test)
1.449Inference Time (s)
4
WMT23 en-de AIRC translations (test)
83.3AUC
4
WMT23 en-de Lan-BridgeMT translations (test)
0.801AUC
4
en-de GPT4-5shot translations WMT23 (test)
0.715AUC
4
WMT en-cs 23
16.24BLEU
4
WMT '24
6.4Difficulty Score
4
WMT14
71AUC
4
WMT'14
17.55BLEU
4
WMT ZH
16.67BLEU
4
WMT TR
14.35BLEU
4
WMT FR
24.82BLEU
4
WMT DE
23.02BLEU
4
Average (wmt19_de, wmt14_fr, wmt19_ru, wmt14_es)
20.91BLEU Score
4
WMT es 14
12.78BLEU
4
WMT ru 19
18.55BLEU
4
WMT fr 14
25.27BLEU
4
WMT de 19
27.06BLEU Score
4
WMT German-Romansh 2024++ (test)
0.47Fluency
4