RuAG: Learned-rule-augmented Generation for Large Language Models

TL;DR

RuAG introduces a framework that automatically extracts interpretable logic rules from data using LLMs and MCTS, enhancing large language models' reasoning across diverse tasks by injecting targeted knowledge.

Contribution

The paper presents a novel method combining LLMs, Monte Carlo Tree Search, and logic rule distillation to improve reasoning in LLMs, addressing contextual window limitations.

Findings

Effective knowledge injection improves reasoning performance.

Applicable across NLP, time-series, and industrial tasks.

Demonstrates significant enhancement over baseline models.

Abstract

In-context learning (ICL) and Retrieval-Augmented Generation (RAG) have gained attention for their ability to enhance LLMs' reasoning by incorporating external knowledge but suffer from limited contextual window size, leading to insufficient information injection. To this end, we propose a novel framework, RuAG, to automatically distill large volumes of offline data into interpretable first-order logic rules, which are injected into LLMs to boost their reasoning capabilities. Our method begins by formulating the search process relying on LLMs' commonsense, where LLMs automatically define head and body predicates. Then, RuAG applies Monte Carlo Tree Search (MCTS) to address the combinational searching space and efficiently discover logic rules from data. The resulting logic rules are translated into natural language, allowing targeted knowledge injection and seamless integration into LLM prompts for LLM's downstream task reasoning. We evaluate our framework on public and private industrial tasks, including natural language processing, time-series, decision-making, and industrial tasks, demonstrating its effectiveness in enhancing LLM's capability over diverse tasks.