[论文] Multilingual Reasoning Cascades Need More Context

论文概要

研究领域: NLP 作者: Arnav Mazumder, Dengjia Zhang, Shuyue Stella Li, Yulia Tsvetkov, Niyati Bafna 发布时间: 2026-06-25 arXiv: 2606.27306

中文摘要

推理的翻译级联将查询从其他语言翻译为英语，用英语推理，再将答案翻译回原始语言。这是多语言推理的一种有竞争力的方法，但结构上存在信息损失，因为每一阶段都会丢弃后续阶段可能需要的信息，包括文化 grounding、语域和消歧线索。我们考察了一种简单且无需训练的干预措施的效益：上下文感知翻译级联，它额外将原始问题、英语翻译问题和推理轨迹提供给最终翻译模块的上下文。我们在九个多语言基准上进行评估，涵盖各种任务类型、三种骨干模型和 285 种高、中、低资源语言，并证明在开放式生成任务上跨模型和资源制度均有显著增益。我们表明原始语言问题携带了大部分有益的上下文。我们的研究强调需要更好地设计机器翻译级联中的信息流以缓解错误传播，并提供了一种简单且可操作的默认策略：在流水线末端保留原始用户问题。

原文摘要

Translation cascades for reasoning translate the query from another language to English, reason in English, and translate the answer back to the original language. This is a competitive approach to multilingual reasoning, but structurally lossy, since each stage discards information later stages may need, including cues for cultural grounding, register, and disambiguation. We examine the benefits of a simple and training-free intervention: a context-aware translation cascade, which additionally provides the original question, the English translated question, and the reasoning trace to the context of the final translation module. We evaluate gains across nine multilingual benchmarks including various task types, three backbone models, and 285 high-, mid-, and low-resource languages, and dem...

--- *自动采集于 2026-06-28*

#论文 #arXiv #NLP #小凯