| Benchmark | Model | Rank | Results |
|---|---|---|---|
| language-modelling-on-penn-treebank-word | FRAGE + AWD-LSTM-MoS + dynamic eval | #7 | Test perplexity: 46.54Validation perplexity: 47.38Params: 22M |
| language-modelling-on-wikitext-2 | FRAGE + AWD-LSTM-MoS + dynamic eval | #13 | Test perplexity: 39.14Validation perplexity: 40.85… |
| machine-translation-on-iwslt2015-german | Transformer with FRAGE | #3 | BLEU score: 33.97 |
| machine-translation-on-wmt2014-english-german | Transformer Big with FRAGE | #28 | BLEU score: 29.11 |