HanoiT: Enhancing Context-aware Translation via Selective Context

Context-aware neural machine translation aims to use the document-level context improve quality. However, not all words in are helpful. The irrelevant or trivial may bring some noise and distract model from learning relationship between current sentence auxiliary context. To mitigate this problem, we propose a novel end-to-end encoder-decoder with layer-wise selection mechanism sift refine long document verify effectiveness of our method, extensive experiments extra quantitative analysis conducted on four benchmarks. experimental results demonstrate that significantly outperforms previous models datasets via soft mechanism.