LLMs Are Zero-Shot Context-Aware Simultaneous Translators

TL;DR

This paper demonstrates that open-source large language models can perform zero-shot simultaneous translation effectively, especially with minimal background info, highlighting their potential for multilingual, context-aware translation without extensive training.

Contribution

It shows that open-source LLMs can achieve competitive zero-shot simultaneous translation and improve with minimal background info, reducing the need for resource-intensive training.

Findings

LLMs perform on par or better than state-of-the-art in zero-shot SiMT

Minimal background info enhances translation performance

LLMs enable resource-efficient, multilingual, context-aware SiMT systems

Abstract

The advent of transformers has fueled progress in machine translation. More recently large language models (LLMs) have come to the spotlight thanks to their generality and strong performance in a wide range of language tasks, including translation. Here we show that open-source LLMs perform on par with or better than some state-of-the-art baselines in simultaneous machine translation (SiMT) tasks, zero-shot. We also demonstrate that injection of minimal background information, which is easy with an LLM, brings further performance gains, especially on challenging technical subject-matter. This highlights LLMs' potential for building next generation of massively multilingual, context-aware and terminologically accurate SiMT systems that require no resource-intensive training or fine-tuning.