| Benchmark | Model | Rank | Results |
|---|---|---|---|
| abstractive-text-summarization-on-cnn-daily-mail | Subformer-base | – | ROUGE-1: 40.9ROUGE-2: 18.3ROUGE-L: 37.7 |
| language-modelling-on-wikitext-103 | Subformer | – | Test perplexity: 20.39Number of params: 96M |
| machine-translation-on-wmt2014-english-german | Subformer-xlarge | – | BLEU score: 29.3 |