通过语言混合的推理模型推动多语言推理
Pushing on Multilingual Reasoning Models with Language-Mixed Chain-of-Thought
AI总结：
本文提出语言混合的推理方法，通过多语言推理提升模型性能，展示在韩国案例研究中达到最先进的表现。
摘要：
最近的前沿模型利用长链式推理来探索上下文中的解决方案空间，从而实现更强的性能。虽然许多研究致力于通过蒸馏构建更小但能力更强的模型，但大多数研究集中在英语上，关于语言特定推理的了解仍十分有限。为填补这一空白，我们首先引入了**语言混合CoT**，这是一种在英语和目标语言之间切换的推理模式，利用英语作为锚点以提升推理能力，同时最小化翻译伪影。作为韩国案例研究，我们整理了**Yi-Sang**：579万个本地韩语提示来自网络问答、考试、STEM和代码；370万个由Qwen3-32B生成的长推理轨迹；以及一个针对性的26万个高产出子集。我们训练了九个模型（4B-35B）在六个家族（Qwen2.5、Llama-3.1、Gemma-3等）中。我们的最佳模型，**KO-REAson-35B**，在整体平均分数（64.0±25）上达到最先进的性能，排名第一在9个基准中的5个，其余排名第二。较小和中等规模的模型也受益匪浅，评估的九个基准的平均改进为+18.6分。消融实验显示**语言混合CoT**比单语CoT更有效，同时也带来了跨语言和多模态的性能提升。我们发布我们的数据整理管道、评估系统、数据集和模型以促进语言特定推理的研究。数据和模型收集：https://huggingface.co/KOREAson。
英文摘要：
Recent frontier models employ long chain-of-thought reasoning to explore solution spaces in context and achieve stonger performance. While many works study distillation to build smaller yet capable models, most focus on English and little is known about language-specific reasoning. To bridge this gap, we first introduct **Language-Mixed CoT**, a reasoning schema that switches between English and a target language, using English as an anchor to excel in reasoning while minimizing translation artificats. As a Korean case study, we curate **Yi-Sang**: 5.79M native-Korean prompts from web Q&A, exams, STEM, and code; 3.7M long reasoning traces generated from Qwen3-32B; and a targeted 260k high-yield subset. We train ninve models (4B-35B) across six families (Qwen2.5, Llama-3.1, Gemma-3, etc). Our best model, **KO-REAson-35B**, achieves state-of-the-art performance, with the highest overall average score (64.0 \pm 25), ranking first on 5/9 benchmarks and second on the remainder. Samller and mid-sized models also benefit substantially, with an average improvement of +18.6 points across teh evaluated nine benchmarks. Ablations show **Language-Mixed CoT** is more effective than monolingual CoT, also resulting in cross-lingual and mult-modal performance gains. We release our data-curation pipeline, evaluation system, datasets, and models to advance research on language-specific reasoning. Data and model collection: https://huggingface.co/KOREAson.
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。
