IWSLT
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
iwslt En->Nl 2014 (test)
10.1Tokenized BLEU
3
IWSLT (train)
3Train Perplexity
3
IWSLT (val)
4.86Perplexity (Validation)
3
IWSLT EN-DE 2014
29.14BLEU
3
IWSLT Nl-En 2014 (test)
34.8BLEU
3
IWSLT Ro-En 2014 (test)
33.49BLEU
3
IWSLT En-Ar 2017 (test)
15.05BLEU
3
IWSLT Zh-En 2015 (test)
21.45BLEU
3
IWSLT En-Zh 2015 (test)
23.01BLEU
3
IWSLT Vi-En 2015 (test)
32.99BLEU
3
IWSLT DE-EN
34.5BLEU Score
3
IWSLT Fr-En 2016 (test)
34.72BLEU Score
3
IWSLT De-En 2016 (test)
29.33BLEU
3
IWSLT En-De 2016 (test)
23.97BLEU
3
IWSLT IF short track (zh) 2026 (official)
0.794TRANS-COMET
2
IWSLT IF short track (de) 2026 (official)
0.765TRANS-COMET
2
IWSLT IF short track (it) 2026 (official)
0.763TRANS-COMET
2
IWSLT IF short track (en) 2026 (official)
53.1QA-BERTScore
2
IWSLT short-form EN-ZH 2026
91.5Accuracy
2
IWSLT short-form EN-ZH 2026
0.52BERTScore
2
IWSLT short-form EN-ZH 2026
0.777COMET Score
2
IWSLT short-form EN-IT 2026
0.527BERTScore
2
IWSLT short-form EN-IT 2026
74.2COMET
2
IWSLT short-form EN-DE 2026
76.2Accuracy
2
IWSLT short-form EN-DE 2026
50.5BERTScore
2