/GNAS-MP

Pytorch Implementation of Rethinking Graph Neural Architecture Search from Message-passing (CVPR2021)

Primary LanguageJupyter Notebook

Rethinking Graph Neural Architecture Search from Message-passing

Getting Started

0. Prerequisites

  • Linux
  • NVIDIA GPU + CUDA CuDNN

1. Setup Python Environment

# clone Github repo
conda install git
git clone https://github.com/phython96/GNAS-MP.git
cd GNAS-MP

# Install python environment
conda env create -f environment_gpu.yml
conda activate gnasmp

2. Download datasets

The datasets are provided by project benchmarking-gnns, you can click here to download all the required datasets.

3. Search Architectures

python scripts/search_molecules_zinc.sh [gpu_id]

4. Train & Test

python scripts/train_molecules_zinc.sh [gpu_id] '[path_to_genotypes]/example.yaml'

Reference

@inproceedings{cai2021rethinking,
  title={Rethinking Graph Neural Architecture Search from Message-passing},
  author={Cai, Shaofei and Li, Liang and Deng, Jincan and Zhang, Beichen and Zha, Zheng-Jun and Su, Li and Huang, Qingming},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={6657--6666},
  year={2021}
}