This project leverages a reproducible devcontainer environment, making it easy to set up and run on any machine with Docker and Visual Studio Code (or another compatible editor). By comparing three state-of-the-art GNN architectures (GCN, GAT, and GIN), the project provides insights into their relative performance in a regression task.
machine-learning deep-learning reproducible-research devcontainer model-comparison graph-neural-networks pytorch-geometric gnn molecular-engineering qm9-dataset
-
Updated
Mar 3, 2026 - Python