通过原则性合成逻辑语料库增强LLM的推理能力
Enhancing Reasoning Capabilities of LLMs via Principled Synthetic Logic Corpus
AI总结：
针对LLM推理能力不足的问题，本文提出附加逻辑训练（ALT）方法，并基于符号逻辑理论构建了高质量多步演绎合成语料库FLD×2。实验表明，该方法显著提升了LLaMA-3.1-70B等先进模型在逻辑、数学、编程及BBH基准上的表现。
摘要：
大型语言模型（LLMs）能够解决广泛的任务，但它们在推理方面仍存在困难。为解决此问题，我们提出Additional Logic Training（ALT），旨在通过程序生成的逻辑推理样本来增强LLM的推理能力。我们首先结合符号逻辑理论与先前的经验见解，确立了设计高质量样本的原则。然后，基于这些原则，我们构建了一个名为Formal Logic Deduction Diverse（FLD×2）的合成语料库，其中包含大量多步演绎样本，具有未知事实、多样化推理规则、多样化语言表达和具有挑战性的干扰项。最后，我们实证表明，在FLD×2上进行ALT能显著增强最先进LLM（包括LLaMA-3.1-70B）的推理能力。提升包括在逻辑推理基准上最高提升30分，在数学和编程基准上最高提升10分，以及在基准套件BBH上提升5分。
英文摘要：
Large language models (LLMs) are capable of solving a wide range of tasks, yet they have struggled with reasoning. To address this, we propose $\textbf{Additional Logic Training (ALT)}$, which aims to enhance LLMs' reasoning capabilities by program-generated logical reasoning samples. We first establish principles for designing high-quality samples by integrating symbolic logic theory and previous empirical insights. Then, based on these principles, we construct a synthetic corpus named $\textbf{Formal Logic Deduction Diverse}$ ($\textbf{FLD}$$_{\times 2}$), comprising numerous samples of multi-step deduction with unknown facts, diverse reasoning rules, diverse linguistic expressions, and challenging distractors. Finally, we empirically show that ALT on FLD$_{\times2}$ substantially enhances the reasoning capabilities of state-of-the-art LLMs, including LLaMA-3.1-70B. Improvements include gains of up to 30 points on logical reasoning benchmarks, up to 10 points on math and coding benchmarks, and 5 points on the benchmark suite BBH.
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。
