arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.04764cs.CL
收藏

超越多示例翻译:为低资源机器翻译扩展上下文演示

Beyond Many-Shot Translation: Scaling In-Context Demonstrations For Low-Resource Machine Translation

Institute of Information Science, Academia Sinica(台湾“中央研究院”信息科学研究所);National Taiwan University of Science and Technology(台湾科技大学);Ecole Centrale de Marseille(马赛中央理工学院);National University of Singapore(新加坡国立大学)

Luis Frentzen Salim, Esteban Carlin, Alexandre Morinvil, Xi Ai, Lun-Wei Ku

AI总结:

本研究探索了通过扩展上下文长度来改进低资源机器翻译的ICL方法,发现增加上下文可提升性能但存在饱和点,且不同语料库类型对结果影响显著。

摘要:

构建低资源语言的机器翻译(MT)系统具有显著的难度,这主要是由于高质量数据的稀缺性。尽管大型语言模型(LLMs)已经提升了MT系统的性能,但将它们适应于较少代表的语言仍然具有挑战性。上下文学习(ICL)可能为适应LLMs进行低资源MT提供新的方法,通过在推理时对模型进行演示条件。在本研究中,我们探索将低资源机器翻译的ICL扩展到数千个示例,使用长上下文模型。我们将上下文标记预算扩展到100万标记,并比较三种类型的训练语料库作为上下文监督:单语无监督数据、指令式数据以及平行数据(英语-目标语言和印尼语-目标语言)。我们的实验结果表明,额外上下文带来的收益迅速饱和,并且在接近最大上下文窗口时可能会下降,其扩展行为强烈依赖于语料库类型。值得注意的是,某些形式的单语监督可以与平行数据竞争,尽管后者提供了额外的监督。总体而言,我们的结果刻画了长上下文ICL在低资源MT中的有效限制和语料库类型敏感性,强调更大的上下文窗口并不一定带来成比例的质量提升。

英文摘要:

Building machine translation (MT) systems for low-resource languages is notably difficult due to the scarcity of high-quality data. Although Large Language Models (LLMs) have improved MT system performance, adapting them to lesser-represented languages remains challenging. In-context learning (ICL) may offer novel ways to adapt LLMs for low-resource MT by conditioning models on demonstration at inference time. In this study, we explore scaling low-resource machine translation ICL beyond the few-shot setting to thousands of examples with long-context models. We scale in-context token budget to 1M tokens and compare three types of training corpora used as in-context supervision: monolingual unsupervised data, instruction-style data, and parallel data (English--target and Indonesian--target). Our experiments on Javanese and Sundanese show that gains from additional context saturate quickly and can degrade near the maximum context window, with scaling behavior strongly dependent on corpus type. Notably, some forms of monolingual supervision can be competitive with parallel data, despite the latter offering additional supervision. Overall, our results characterize the effective limits and corpus-type sensitivity of long-context ICL for low-resource MT, highlighting that larger context windows do not necessarily yield proportional quality gains.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑