arXiv cs.CL· Wen Yang, Junhong Wu, Chong Li, Chengqing Zong, Jiajun Zhang·· 13 小时前AI 评分31
以英语为中心的 LLM 跨语言数学推理迁移效果如何？
How Well Does Mathematical Reasoning Transfer Across Languages in English-Centric LLMs?
AI 导读
研究评估了以英语为中心的 LLM 在多语言数学推理基准上的跨语言迁移，发现迁移效果随初始模型、目标语言和训练范式差异很大。更强的英语初始化并不必然带来更高的相对迁移效率，即使其多语言推理准确率更高。最大提升来自突破纯英语监督：加入一种平行语言即可带来显著增益，继续增加语言收益更小但稳定。
正文 · 原文
Abstract:Recent advances in Reinforcement Post-Training (RPT) have substantially improved the mathematical reasoning capabilities of large language models (LLMs), but it remains unclear how well such gains transfer across languages. In this work, we study cross-lingual mathematical reasoning transfer in English-centric LLMs under controlled multilingual evaluation. We first evaluate a range of English-centric reasoning models on multilingual mathematical reasoning benchmarks and find that transfer is highly variable across initial model choice, target language, and training paradigm. Through controlled comparative experiments, we further find that stronger English-centric initialization does not necessarily lead to stronger relative transfer efficiency, even when it yields higher multilingual reasoning accuracy. Finally, we show that the largest improvement comes from moving beyond English-only supervision: adding a single parallel language yields a substantial gain, while further gains from additional languages are smaller but consistent. Overall, our results highlight the need to evaluate multilingual mathematical reasoning directly, using both absolute performance and relative transfer-efficiency metrics.
| Comments: | Camera ready version for EMNLP 2026 Main |
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2510.02272 [cs.CL] |
| (or arXiv:2510.02272v2 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2510.02272 arXiv-issued DOI via DataCite |
Submission history
From: Wen Yang [view email]
[v1]
Thu, 2 Oct 2025 17:49:49 UTC (252 KB)
[v2]
Thu, 22 Oct 2026 16:06:00 UTC (272 KB)
来源：arXiv cs.CL · arxiv.org