Big Bidirectional Insertion Representations for Documents
This addresses the challenge of long-form text generation for document-level translation tasks, though it appears incremental in nature.
The paper tackles document-level translation by scaling up insertion-based models to handle long-form documents, achieving a +4.3 BLEU improvement on WMT'19 English→German translation compared to the Insertion Transformer baseline.
The Insertion Transformer is well suited for long form text generation due to its parallel generation capabilities, requiring $O(\log_2 n)$ generation steps to generate $n$ tokens. However, modeling long sequences is difficult, as there is more ambiguity captured in the attention mechanism. This work proposes the Big Bidirectional Insertion Representations for Documents (Big BIRD), an insertion-based model for document-level translation tasks. We scale up the insertion-based models to long form documents. Our key contribution is introducing sentence alignment via sentence-positional embeddings between the source and target document. We show an improvement of +4.3 BLEU on the WMT'19 English$\rightarrow$German document-level translation task compared with the Insertion Transformer baseline.