用结构化模板和难度分级训练，提升大模型解决难题的能力
Can Structured Templates Facilitate LLMs in Tackling Harder Tasks? : An Exploration of Scaling Laws by Difficulty
- 设计结构化解题模板与渐进难度课程，显式教模型推理流程
- 在GSM8K、AIME24等数据集上，难题准确率显著提升，效率更高
- 适合需要强逻辑推理的数学、编程等复杂任务研究者使用
结构化程序推理对大型语言模型（LLM）至关重要，尤其在数学任务中。尽管后训练方法提升了性能，但在复杂任务中仍难以捕捉深层程序逻辑。本文首次揭示一种按难度变化的缩放规律：随着训练数据复杂度上升，模型表现呈倒U型曲线——过多低难度数据会阻碍抽象能力，而高难度数据能显著增强推理能力。为此，提出结构化解题模板（SST）框架，包含三部分：（1）使用结构化解题链进行微调，并采用动态加权损失优先优化程序逻辑；（2）推理时注入解题模板作为认知支架；（3）集成课程微调，显式教会模型自规划-执行-自修正。在GSM8K、AIME24及新构建的Dynamic En数据集上的实验表明，SST显著提升准确率与效率，尤其在高难度问题上优势明显。
原文摘要 · Abstract (English)
Structured, procedural reasoning is essential for Large Language Models (LLMs), especially in mathematics. While post-training methods have improved LLM performance, they still fall short in capturing deep procedural logic on complex tasks. To tackle the issue, in this paper, we first investigate this limitation and uncover a novel finding: a Scaling Law by Difficulty, which reveals that model performance follows a U-shaped curve with respect to training data complexity -- excessive low-difficulty data impedes abstraction, while high-difficulty data significantly enhances reasoning ability. Motivated by this, we propose the Structured Solution Template (SST) framework, which uses solution templates and a curriculum of varied difficulty to explicitly teach procedural reasoning. Specifically, SST comprises (1) fine-tuning with structured solution-template chains and dynamically weighted loss to prioritize procedural logic, (2) prompt-time injection of solution templates as cognitive scaffolds to guide inference, and (3) integrated curriculum fine-tuning that explicitly teaches the model to self-plan - execute - self-correct. Experiments on GSM8K, AIME24, and new Dynamic En benchmark show that SST significantly improves both accuracy and efficiency, especially on harder problems.
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。