通过测试时扩展实现跨语言推理
Crosslingual Reasoning through Test-Time Scaling
AI总结：
本文研究以英语为中心的推理语言模型通过测试时扩展实现跨语言泛化的潜力与机制，发现其能提升多语言数学推理、遵循“引用并思考”模式，并指出高资源语言推理更优而低资源及域外推理仍需改进。
摘要：
大型语言模型的推理能力主要针对英语进行研究，即使预训练模型是多语言的。在本工作中，我们研究了英语推理微调与长思维链（CoTs）在多大程度上能够跨语言泛化。首先，我们发现，扩大以英语为中心的推理语言模型（RLMs）的推理计算量，能够改善包括低资源语言在内的多种语言上的多语言数学推理能力，其改善程度甚至使这些模型超过了规模为其两倍的模型。其次，我们揭示，虽然以英语为中心的RLM的思维链自然以英语为主，但它们始终遵循一种“引用并思考”的模式来推理被引用的非英语输入。第三，我们发现了一种控制长思维链推理语言的有效策略，并观察到模型在高资源语言中推理得更好且更高效。最后，我们观察到域外推理泛化能力较差，尤其是从STEM领域到文化常识知识，即使对英语也是如此。总体而言，我们展示了英语推理测试时扩展的跨语言泛化潜力，研究了其机制，并概述了其局限性。我们得出结论：实践者应让以英语为中心的RLM在高资源语言中进行推理，同时需要进一步的工作来改善低资源语言和域外情境中的推理。
英文摘要：
Reasoning capabilities of large language models are primarily studied for English, even when pretrained models are multilingual. In this work, we investigate to what extent English reasoning finetuning with long chain-of-thoughts (CoTs) can generalize across languages. First, we find that scaling up inference compute for English-centric reasoning language models (RLMs) improves multilingual mathematical reasoning across many languages including low-resource languages, to an extent where they outperform models twice their size. Second, we reveal that while English-centric RLM's CoTs are naturally predominantly English, they consistently follow a quote-and-think pattern to reason about quoted non-English inputs. Third, we discover an effective strategy to control the language of long CoT reasoning, and we observe that models reason better and more efficiently in high-resource languages. Finally, we observe poor out-of-domain reasoning generalization, in particular from STEM to cultural commonsense knowledge, even for English. Overall, we demonstrate the potentials, study the mechanisms and outline the limitations of crosslingual generalization of English reasoning test-time scaling. We conclude that practitioners should let English-centric RLMs reason in high-resource languages, while further work is needed to improve reasoning in low-resource languages and out-of-domain contexts.
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。
