Less is More: Learning Graph Tasks with Just LLMs

TL;DR

This paper demonstrates that large language models can learn to solve graph reasoning tasks effectively using simple chain-of-thought prompting, without the need for specialized graph encoders or complex serialization methods.

Contribution

It shows that small LLMs can learn and generalize graph tasks through instructive chain-of-thought training, challenging the need for specialized graph encoding models.

Findings

Small LLMs can solve graph tasks with chain-of-thought training

LLMs generalize to unseen graph structures and tasks

No need for specialized graph encoders

Abstract

For large language models (LLMs), reasoning over graphs could help solve many problems. Prior work has tried to improve LLM graph reasoning by examining how best to serialize graphs as text and by combining GNNs and LLMs. However, the merits of such approaches remain unclear, so we empirically answer the following research questions: (1) Can LLMs learn to solve fundamental graph tasks without specialized graph encoding models?, (2) Can LLMs generalize learned solutions to unseen graph structures or tasks?, and (3) What are the merits of competing approaches to learn graph tasks? We show that even small LLMs can learn to solve graph tasks by training them with instructive chain-of-thought solutions, and this training generalizes, without specialized graph encoders, to new tasks and graph structures.