Recent Large Language Models (LLMs) have demonstrated strong performance in translation without needing to be finetuned on additional parallel corpora. However, they still underperform for low-resource language pairs. Previous works have focused on mitigating this issue by leveraging relevant few-shot examples or external resources such as dictionaries or grammar books, making models heavily reliant on these nonparametric sources of information. In this paper, we propose a novel method named IntGrad MT that focuses on fully exploiting an LLM's inherent translation capability. IntGrad MT achieves this by constructing a chain of few-shot examples, each consisting of a source sentence and the model's own translation, that rise incrementally in difficulty. IntGrad MT employs two techniques: Sentence Interpolation, which generates a sequence of sentences that gradually change from an easy sentence to translate to a difficult one, and Gradual MT, which sequentially translates this chain using translations of earlier sentences as few-shot examples for the translation of subsequent ones. With this approach, we observe a substantial enhancement in the xCOMET scores of various LLMs for multiple languages, especially in low-resource languages such as Hindi(8.26), Swahili(7.10), Bengali(6.97) and Marathi(13.03). Our approach presents a practical way of enhancing LLMs' performance without extra training.

翻译：近期的大语言模型（LLM）在无需额外平行语料微调的情况下，已展现出强大的翻译性能。然而，对于低资源语言对，其表现仍不尽如人意。先前的研究主要通过利用相关的少样本示例或外部资源（如词典或语法书）来缓解此问题，这使得模型严重依赖于这些非参数化的信息来源。本文提出了一种名为 IntGrad MT 的新方法，旨在充分挖掘 LLM 固有的翻译能力。IntGrad MT 通过构建一个由少样本示例组成的链来实现这一目标，其中每个示例包含一个源语句及其对应的模型自身译文，且难度逐步递增。IntGrad MT 采用两种技术：句子插值（Sentence Interpolation），用于生成一系列句子，使其从易于翻译的句子逐渐过渡到难以翻译的句子；以及渐进式机器翻译（Gradual MT），该技术利用先前句子的译文作为少样本示例，依次翻译此链中的后续句子。通过这种方法，我们观察到多种 LLM 在多个语言上的 xCOMET 分数显著提升，尤其是在印地语（8.26）、斯瓦希里语（7.10）、孟加拉语（6.97）和马拉地语（13.03）等低资源语言上。我们的方法为在不进行额外训练的情况下提升 LLM 的性能提供了一种实用途径。