README.md

This repository contains the official implementation of the graph sampling method presented in "GRAPES: Learning to Sample Graphs for Scalable Graph Neural Networks".

Instructions

1. Install dependencies

Create a conda environment with the provided file, then activate it:

conda env create -f environment.yml
conda activate grapes

2. Train a model

Run the following to train a GCN classifier on the Cora dataset:

python main.py

We provide configuration files to reproduce the results in our experiments with all datasets. To use them, run:

python main.py --config_file=configs/<dataset>.txt

Replacing <dataset> with the name of the dataset.

3. Inspect results on W&B

Logging on Weights & Biases is enabled by default. Results will be logged to a project with name gflow-sampling. To disable this, add the flag --log_wandb=False.

---

Baselines and Data Analysis

For the baseline implementation anad data analysis, please check out the following repos:

• GraphSAINT: https://github.com/dfdazac/grapes/blob/main/graphsaint.py
• LADIES & FastGCN: https://anonymous.4open.science/r/LADIES-9589
• GAS: https://anonymous.4open.science/r/pyg_autoscale-2A4C
• AS-GCN: https://anonymous.4open.science/r/as-gcn-B0FA
• Data Analysis: https://anonymous.4open.science/r/GRAPES-plots-and-analyses-F4B6