提出新框架评估跨语言程序思维推理，提升多语种大模型推理能力。
Towards Better Understanding of Program-of-Thought Reasoning in Cross-Lingual and Multilingual Environments
- 分离多语言推理与代码执行，解耦问题理解与计算过程
- 微调后程序思维比链式思维在多语言上表现更好
- 推理质量与答案准确率强相关，可作推理优化参考
多步推理对大语言模型至关重要，但多语言表现仍存挑战。链式思维提示虽能提升推理能力，但在非英语语言中因推理与执行纠缠而受限。程序思维提示将推理与执行分离，提供新路径，但难点在于从非英语问题生成正确程序。本文提出评估框架，将多语言推理与代码执行解耦，考察（i）微调对问题-推理对齐的影响，（ii）推理质量对答案正确性的作用。结果表明，程序思维微调显著提升多语言推理能力，优于链式思维微调模型。进一步发现推理质量（以代码质量衡量）与答案准确率高度相关，凸显其作为测试时性能提升策略的潜力。
原文摘要 · Abstract (English)
Multi-step reasoning is essential for large language models (LLMs), yet multilingual performance remains challenging. While Chain-of-Thought (CoT) prompting improves reasoning, it struggles with non-English languages due to the entanglement of reasoning and execution. Program-of-Thought (PoT) prompting separates reasoning from execution, offering a promising alternative but shifting the challenge to generating programs from non-English questions. We propose a framework to evaluate PoT by separating multilingual reasoning from code execution to examine (i) the impact of fine-tuning on question-reasoning alignment and (ii) how reasoning quality affects answer correctness. Our findings demonstrate that PoT fine-tuning substantially enhances multilingual reasoning, outperforming CoT fine-tuned models. We further demonstrate a strong correlation between reasoning quality (measured through code quality) and answer accuracy, highlighting its potential as a test-time performance improvement heuristic.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。