Skip to content

A short and easy PyTorch implementation of E(n) Equivariant Graph Neural Networks

License

Notifications You must be signed in to change notification settings

bayesgroup/equivariant-gnn-ddd-hse

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Simple implementation of Equivariant GNN

Open In Colab

  • A short implementation of E(n) Equivariant Graph Neural Networks for HOMO energy prediction.
  • Just 50 lines of code;
  • The implementation is based on pure PyTorch & Numpy, it has no external packages (like PyTorch-geometric).
  • Closely matches the Mean Absolute Error reported in the paper.

About

A short and easy PyTorch implementation of E(n) Equivariant Graph Neural Networks

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 82.4%
  • Jupyter Notebook 17.6%