CGCNN2

As the original Crystal Graph Convolutional Neural Networks (CGCNN) repository is no longer actively maintained, this repository is a reproduction of CGCNN by Xie et al. It includes necessary updates for deprecated components and a few additional functions to ensure smooth operation. Despite its age, CGCNN remains a straightforward and fast deep learning framework that is easy to learn and use.

The package provides the following major functions:

• Training a CGCNN model using a custom dataset.
• Predicting material properties with a pre-trained CGCNN model.
• Fine-tuning a pre-trained CGCNN model on a new dataset.
• Extracting structural features as descriptors for downstream tasks.

Installation

Make sure you have a Python interpreter, version 3.11 to 3.14. Then, you can simply install cgcnn2 from PyPI using pip:

pip install cgcnn2

or, if you manage your project with uv:

uv add cgcnn2

If you'd like to use the latest unreleased version on the main branch, you can install it directly from GitHub:

pip install git+https://github.com/jcwang587/cgcnn2@main

Get Started

There are entry points for training, predicting, and fine-tuning CGCNN models. For example, to explore the usage of the provided training script cgcnn-tr, you can use the --help option of the command:

cgcnn-tr --help

Similarly, you can access the predicting and fine-tuning functionalities through cgcnn-pr and cgcnn-ft commands. A detailed user guide documentation is available at: https://jcwang.dev/cgcnn2/

References

The original paper describes the CGCNN framework in detail:

@article{PhysRevLett2018,
  title = {Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties},
  author = {Xie, Tian and Grossman, Jeffrey C.},
  journal = {Phys. Rev. Lett.},
  volume = {120},
  issue = {14},
  pages = {145301},
  numpages = {6},
  year = {2018},
  month = {Apr},
  publisher = {American Physical Society},
  doi = {10.1103/PhysRevLett.120.145301},
  url = {https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.120.145301}
}