Decoder-only Streaming Transformer for Simultaneous Translation

TL;DR

This paper introduces the first Decoder-only Streaming Transformer for simultaneous translation, leveraging a novel streaming self-attention mechanism to improve translation quality and achieve state-of-the-art results.

Contribution

It proposes a new Decoder-only architecture for SiMT, including a streaming self-attention mechanism and position encoding strategy, addressing training and inference challenges.

Findings

Achieves state-of-the-art performance on three translation tasks.

Demonstrates effective translation policy assessment with SSA.

Outperforms existing Encoder-Decoder SiMT models.

Abstract

Simultaneous Machine Translation (SiMT) generates translation while reading source tokens, essentially producing the target prefix based on the source prefix. To achieve good performance, it leverages the relationship between source and target prefixes to exact a policy to guide the generation of translations. Although existing SiMT methods primarily focus on the Encoder-Decoder architecture, we explore the potential of Decoder-only architecture, owing to its superior performance in various tasks and its inherent compatibility with SiMT. However, directly applying the Decoder-only architecture to SiMT poses challenges in terms of training and inference. To alleviate the above problems, we propose the first Decoder-only SiMT model, named Decoder-only Streaming Transformer (DST). Specifically, DST separately encodes the positions of the source and target prefixes, ensuring that the position of the target prefix remains unaffected by the expansion of the source prefix. Furthermore, we propose a Streaming Self-Attention (SSA) mechanism tailored for the Decoder-only architecture. It is capable of obtaining translation policy by assessing the sufficiency of input source information and integrating with the soft-attention mechanism to generate translations. Experiments demonstrate that our approach achieves state-of-the-art performance on three translation tasks.