HG-Adapter: Improving Pre-Trained Heterogeneous Graph Neural Networks with Dual Adapters

September 18, 2024 ยท View on GitHub

This repository contains the reference code for the manuscript ``HG-Adapter: Improving Pre-Trained Heterogeneous Graph Neural Networks with Dual Adapters"

Contents

  1. Installation
  2. Preparation
  3. Training

Installation

  • pip install -r requirements.txt
  • Unzip the datasets (can be found in ./dataset/)
  • Unzip the pre-trained embedding (can be found in ./pre_trained_embedding)

Preparation

Important args:

  • --upload_pa Test checkpoints to reproduce the results
  • --dataset Heterogeneous graph: ACM, Yelp, DBLP, Aminer
  • --custom_key Node: node classification

Training

python main.py