A Simple but Tough-to-Beat Data Augmentation Approach for Natural Language Understanding and Generation

Benchmark Model Rank Results
machine-translation-on-iwslt2014-germanCutoff#8BLEU score: 37.6