让模型自动选算术用代码、抽象推理用文本，提升推理效率与准确率。
Agentic-R1: Distilled Dual-Strategy Reasoning
- 融合多教师策略，训练统一学生模型动态选择推理方式。
- 在计算密集型与标准基准上均实现更高准确率，最高提升12.3%。
- 适合需要高效精准推理的应用，如数学题求解与复杂逻辑判断。
当前长链式思维（long-CoT）模型在数学推理上表现优异，但依赖缓慢且易出错的自然语言追踪。工具增强型代理通过代码执行解决算术问题，但在复杂逻辑任务中常失效。我们提出一种微调框架 DualDistill，将多个教师模型的互补推理策略蒸馏到统一的学生模型中。基于此方法，我们训练了 Agentic-R1，该模型能根据查询动态选择最优策略：对算术和算法问题调用工具，对抽象问题采用文本推理。我们的方法在多种任务上提升准确率，包括计算密集型与标准基准，验证了多策略蒸馏在实现鲁棒高效推理中的有效性。项目开源地址：https://github.com/StigLidu/DualDistill。
原文摘要 · Abstract (English)
Current long chain-of-thought (long-CoT) models excel at mathematical reasoning but rely on slow and error-prone natural language traces. Tool-augmented agents address arithmetic via code execution, but often falter on complex logical tasks. We introduce a fine-tuning framework, DualDistill, that distills complementary reasoning strategies from multiple teachers into a unified student model. Using this approach, we train Agentic-R1, which dynamically selects the optimal strategy for each query, invoking tools for arithmetic and algorithmic problems, and using text-based reasoning for abstract ones. Our method improves accuracy across a range of tasks, including both computation-intensive and standard benchmarks, demonstrating the effectiveness of multi-strategy distillation in achieving robust and efficient reasoning. Our project is available at https://github.com/StigLidu/DualDistill
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。