源文预热多轮对话有助于大语言模型翻译文档
Source-primed Multi-turn Conversation Helps Large Language Models Translate Documents
- University of Zurich（苏黎世大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文提出利用多轮对话保留上下文并复用 KV cache 的文档翻译方法，且在翻译前提供完整源文档进行源文预热，实验证明其优于单轮全文翻译和独立片段翻译。
AI中文摘要：
LLM 为真正简单的文档级机器翻译铺平了道路，但漏译错误等挑战仍然存在。本文研究一种处理文档级机器翻译的简单方法，即以多轮对话方式利用先前上下文。具体而言，该方法将文档分解为若干片段，并在保留先前轮次的同时迭代翻译它们，无需额外训练即可保证译文连贯，还能完全复用先前轮次的 KV cache，从而最大限度降低计算开销。我们进一步提出一种“source-primed”（源文预热）方法，在多轮翻译之前先提供完整源文档。实验表明，在具有代表性的 LLM 上，依据多项自动指标，这种多轮方法优于单轮翻译整篇文档以及独立翻译每个片段，为使用 LLM 进行文档级翻译建立了一个强基线。
英文摘要：
LLMs have paved the way for truly simple document-level machine translation, but challenges such as omission errors remain. In this paper, we study a simple method for handling document-level machine translation, by leveraging previous contexts in a multi-turn conversational manner. Specifically, by decomposing documents into segments and iteratively translating them while maintaining previous turns, this method ensures coherent translations without additional training, and can fully re-use the KV cache of previous turns thus minimizing computational overhead. We further propose a `source-primed' method that first provides the whole source document before multi-turn translation. We empirically show this multi-turn method outperforms both translating entire documents in a single turn and translating each segment independently according to multiple automatic metrics in representative LLMs, establishing a strong baseline for document-level translation using LLMs.
