The Graph Neural Network Playground is an interactive tool designed to help users visualize and experiment with Graph Neural Networks (GNNs). Built using TypeScript and d3.js, it provides an intuitive interface to train GNNs on custom datasets and observe model performance in real-time.
Your feedback is highly appreciated! If you have any questions, feature requests, or encounter any bugs, please use the GitHub issues section to report them.
- Custom Data Input: Upload your own graph-based data for training.
- Real-Time Training & Feedback: Visualize how the model's performance improves as it trains, and see the loss/error rate diminish over time.
- Model Persistence: Save the trained model weights for future use or download them directly from the app.
-
Step 1: Upload Your Material Crystal Data
- Upload your custom graph data to train a GNN model on it.
-
Step 2: Train and Evaluate the Model
- Monitor the loss/error rate as the model trains. The playground provides real-time feedback to help you understand model performance.
-
Step 3: Save Model Weights
- Once the model has converged, you can save the trained weights for future use or download them directly.
Happy experimenting with Graph Neural Networks!