🌟 Welcome to the repository for our paper titled "CyBond Punk: Rethinking Message Passing Mechanism via Graph Edge Space".

Our implementation is based on PyTorch Geometric that are provided under MIT License. All experiments were run in a shared computing cluster environment with varying CPU and GPU architectures. These involved a A100 (40GB) GPUs. The resource budget for each experiment was 1 GPU, between 4 and 6 CPUs, and up to 32GB system RAM.

- CyBond-Punk
  - eval_edge.py         # Evaluation script for models with edge information
  - model_edge.py        # Model definition with edge information
  - model_noedge.py      # Model definition without edge information
  - requirements.txt     # List of dependencies
  - train_edge.py        # Training script for models with edge information
  - train_noedge.py      # Training script for models without edge information
  - utils.py             # Utility functions
  - README.md            # This readme file

To get started with CyBond-Punk, follow these installation instructions:

Ensure you have the following software installed:

• Python 3.8 or higher
• pip (Python package installer)

Clone this repository to your local machine using:

git clone https://github.com/yourusername/CyBond-Punk.git
cd CyBond-Punk

Install the required Python packages:

pip install -r requirements.txt

Since our implementation relies on PyTorch Geometric, you'll need to install it separately. Use the following command to install PyTorch Geometric along with the CUDA support suitable for your system:

pip install torch-scatter torch-sparse torch-cluster torch-spline-conv torch-geometric

After installing the necessary dependencies, you can run the various scripts provided in this repository:

To train the model with edge information: (such as ZINC and LRGB datasets)

python train_edge.py

To train the model without edge information: (such as TUDataset datasets)

python train_noedge.py

To evaluate the model with edge information:

python eval_edge.py

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

This work was made possible thanks to the resources provided by the shared computing cluster environment.

We welcome contributions! Please read our contributing guidelines for more details.

If you have any questions or issues, please open an issue in this repository or contact us at [email protected].

Happy coding! 🚀