EMNLP Findings2024
Breaking Language Barriers in Multilingual Mathematical Reasoning: Insights and Observations
Nuo Chen, Zinan Zheng, Ning Wu, Ming Gong, Dongmei Zhang, Jia Li
摘要
Existing research predominantly focuses on developing powerful large language models (LLMs) for mathematical reasoning within monolingual languages, with few explorations in preserving efficacy in a multilingual context. To bridge this gap, this paper pioneers exploring and training powerful Multilingual Math Reasoning (xMR) LLMs. Firstly, by utilizing translation, we construct the first multilingual math reasoning instruction dataset, MGSM8KInstruct, encompassing ten distinct languages, thus addressing the issue of training data scarcity in xMR tasks. Based on the collected dataset, we propose different training strategies to build powerful xMR LLMs, named MathOctopus, which notably outperform conventional open-source LLMs and exhibit superiority over ChatGPT in few-shot scenarios. Notably, MathOctopus-13B reaches 47.6% accuracy which exceeds ChatGPT 46.3% on MGSM testset. Beyond remarkable results, we unearth several pivotal observations and insights: (1) When extending the rejection sampling strategy to the multilingual context, it proves effective for model performances, albeit limited. (2) Employing parallel corpora for math Supervised Fine-Tuning (SFT) across multiple languages not only significantly enhances model performance multilingually and elevates their monolingual performance. This indicates that crafting multilingual corpora can be regarded as a vital strategy for enhancing model performance in a specific language, especially in mathematical reasoning tasks. For instance, MathOctopus-7B improves its counterparts that trained on English from 42.4% to 50.8% on the GSM8K test set. James decides to run 3 sprints 3 times a week. He runs 60 meters each sprint. How many total meters does he run a week? English 詹姆斯决定每周跑 3 次 3 段冲刺,每段冲刺跑 60 米。他 每周一共跑多少米? Chinese James beschließt, 3-mal pro Woche 3 Kurzstreckenläufe zu absolvieren. Pro Kurzstreckenlauf läuft er 60 Meter. Wie viele Meter insgesamt läuft er pro Woche? German French Japanese …… He sprints 33=<<33=9>>9 time. So he runs 960=<<960 =540>>540 meters. English Chinese German French Japanese …… MathOctopus Question Answer 詹姆斯一共冲刺 33=<<33=9>>9 次。 所以他每周一共跑 960=<<960=540>>540米。 James sprintet 33=<<33=9>>9 Mal. Also läuft er 960= <<960=540>>540 Meter.
问 Lune
尽管问,回答会引用原文。
Lune 会读完全文和相关的 Paper,每个回答都附上所依据的原文段落。
引用它的顶会 Paper33
- XCOT: Cross-lingual Instruction Tuning for Cross-lingual Chain-of-Thought ReasoningLinzheng Chai, Jian Yang, Tao Sun, Hongcheng Guo 等AAAI 2025 · 70 次引用
- Linguistic Generalizability of Test-Time Scaling in Mathematical ReasoningGuijin Son, Jiwoo Hong, Hyunwoo Ko, James ThorneACL 2025 · 36 次引用
- Multilingual Routing in Mixture-of-ExpertsLucas Bandarkar, Chenyuan Yang, Mohsen Fayyaz, Junlin Hu 等ICLR 2026 · 34 次引用
- MindMerger: Efficiently Boosting LLM Reasoning in non-English LanguagesZixian Huang, Wenhao Zhu, Gong Cheng, Lei Li 等NeurIPS 2024 · 31 次引用
- SimpleStrat: Diversifying Language Model Generation with StratificationJustin Wong, Yury Orlovskiy, Alexander Shypula, Michael Luo 等NeurIPS 2025 · 17 次引用
它引用的 Paper24
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah 等NeurIPS 2020 · 64,255 次引用
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma 等NeurIPS 2022 · 22,562 次引用
- LoRA: Low-Rank Adaptation of Large Language ModelsEdward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu 等ICLR 2022 · 18,833 次引用
- Multitask Prompted Training Enables Zero-Shot Task GeneralizationVictor Sanh, Albert Webson, Colin Raffel, Stephen H. Bach 等ICLR 2022 · 1,976 次引用
- STaR: Bootstrapping Reasoning With ReasoningEric Zelikman, Yuhuai Wu, Jesse Mu, Noah D. GoodmanNeurIPS 2022 · 1,126 次引用
相关 Paper
- mCoT: Multilingual Instruction Tuning for Reasoning Consistency in Language ModelsHuiyuan Lai, Malvina NissimACL 2024 · 5 次引用
- MAmmoTH: Building Math Generalist Models through Hybrid Instruction TuningXiang Yue, Xingwei Qu, Ge Zhang, Yao Fu 等ICLR 2024 · 558 次引用
- WizardMath: Empowering Mathematical Reasoning for Large Language Models via Reinforced Evol-InstructHaipeng Luo, Qingfeng Sun, Can Xu, Pu Zhao 等ICLR 2025
- MathGenie: Generating Synthetic Data with Question Back-translation for Enhancing Mathematical Reasoning of LLMsZimu Lu, Aojun Zhou, Houxing Ren, Ke Wang 等ACL 2024 · 11 次引用
- MMATH: A Multilingual Benchmark for Mathematical ReasoningWenyang Luo, Wayne Xin Zhao, Jing Sha, Shijin Wang 等EMNLP Findings 2025
