Multi30K
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Multi30K (test)
60.6BLEU-4
139
Multi30K 2016 (test)
42.7BLEU
52
Multi30k En-De 2017 (test)
62.8METEOR
45
Multi30k (test)
97.8Recall (EN)
35
Multi30K eng → deu 2017 (test)
41.8BLEU
31
Multi30K eng → deu 2016 (test)
47BLEU
31
Multi30k En-Fr 2017 (test)
76.61METEOR
31
Multi30K eng → fra 2017 (test)
62.1BLEU
30
Multi30k En-Fr 2016 (test)
81.75METEOR Score
30
Multi30k En→Fr v1 2017 (test)
56BLEU
30
Multi30K En → De (test)
57.2METEOR
26
Multi30K 2016 (test)
63.7BLEU
26
Multi30k Task1 (en-de)
41.96BLEU Score
26
Multi30k Task1 en-fr
62.12BLEU Score
25
Multi30k M30kT (test)
50.77BLEU Score
19
Multi30k En-Cs 2018 (test)
39.9BLEU
18
Multi30k En-Cs 2016 (test)
0.414BLEU
18
Multi30k WMT17 (test)
42BLEU
16
Multi30K MSCOCO (eng → fra)
55.3BLEU
15
Multi30K eng → fra 2016 (test)
67.5BLEU
15
Multi30K MSCOCO eng → deu
38.5BLEU
15
Multi30K MSCOCO
30.7BLEU
12
Multi30k 2016 (test)
44.6BLEU
11
Multi30K 2018 (test)
0.395BLEU
10
Multi30K En → Fr (test)
56.9BLEU
9