LLM4GNAS: A Large Language Model Based Toolkit for Graph Neural Architecture Search

TL;DR

LLM4GNAS introduces a toolkit leveraging Large Language Models to automate and adapt Graph Neural Architecture Search, reducing manual effort and improving performance on various graph learning tasks.

Contribution

This work presents a novel LLM-based toolkit for GNAS that simplifies adaptation to new search spaces and enhances GNN design through prompt-based modifications.

Findings

Outperforms existing GNAS methods on multiple graph tasks.

Reduces manual intervention in GNAS algorithm adaptation.

Demonstrates robustness and extensibility of the LLM4GNAS toolkit.

Abstract

Graph Neural Architecture Search (GNAS) facilitates the automatic design of Graph Neural Networks (GNNs) tailored to specific downstream graph learning tasks. However, existing GNAS approaches often require manual adaptation to new graph search spaces, necessitating substantial code optimization and domain-specific knowledge. To address this challenge, we present LLM4GNAS, a toolkit for GNAS that leverages the generative capabilities of Large Language Models (LLMs). LLM4GNAS includes an algorithm library for graph neural architecture search algorithms based on LLMs, enabling the adaptation of GNAS methods to new search spaces through the modification of LLM prompts. This approach reduces the need for manual intervention in algorithm adaptation and code modification. The LLM4GNAS toolkit is extensible and robust, incorporating LLM-enhanced graph feature engineering, LLM-enhanced graph neural architecture search, and LLM-enhanced hyperparameter optimization. Experimental results indicate that LLM4GNAS outperforms existing GNAS methods on tasks involving both homogeneous and heterogeneous graphs.