Rethinking and Improving Natural Language Generation with Layer-Wise Multi-View Decoding

TL;DR

This paper introduces layer-wise multi-view decoding for sequence-to-sequence models, enhancing information extraction from encoder layers, addressing hierarchy bypassing, and achieving state-of-the-art results across diverse NLP tasks.

Contribution

It proposes a novel multi-view decoding method that incorporates multiple encoder layers for each decoder layer, improving performance and addressing hierarchy bypassing in deep models.

Findings

Addresses hierarchy bypassing problem effectively.

Achieves state-of-the-art results on ten benchmark datasets.

Requires negligible additional parameters.

Abstract

In sequence-to-sequence learning, e.g., natural language generation, the decoder relies on the attention mechanism to efficiently extract information from the encoder. While it is common practice to draw information from only the last encoder layer, recent work has proposed to use representations from different encoder layers for diversified levels of information. Nonetheless, the decoder still obtains only a single view of the source sequences, which might lead to insufficient training of the encoder layer stack due to the hierarchy bypassing problem. In this work, we propose layer-wise multi-view decoding, where for each decoder layer, together with the representations from the last encoder layer, which serve as a global view, those from other encoder layers are supplemented for a stereoscopic view of the source sequences. Systematic experiments and analyses show that we successfully address the hierarchy bypassing problem, require almost negligible parameter increase, and substantially improve the performance of sequence-to-sequence learning with deep representations on five diverse tasks, i.e., machine translation, abstractive summarization, image captioning, video captioning, medical report generation, and paraphrase generation. In particular, our approach achieves new state-of-the-art results on ten benchmark datasets, including a low-resource machine translation dataset and two low-resource medical report generation datasets.