Graph Neural Networks (PyG). Node/graph classification, link prediction, GCN, GAT, GraphSAGE, heterogeneous graphs, molecular property prediction, for geometric deep learning.
8.3
Rating
0
Installs
Machine Learning
Category
Exceptional skill for Graph Neural Networks with PyTorch Geometric. The description clearly indicates capabilities (node/graph classification, link prediction, various GNN architectures, heterogeneous graphs, molecular prediction). SKILL.md provides comprehensive task knowledge including installation, core concepts (Data structure, edge indexing, mini-batching), multiple GNN implementations (GCN, GAT, GraphSAGE), custom layer creation via MessagePassing, dataset handling (built-in and custom), complete training workflows for different scenarios (node classification, graph classification, large-scale with neighbor sampling), and advanced features (heterogeneous graphs, transforms, explainability, pooling). Structure is excellent with logical progression from basics to advanced topics, clear code examples throughout, and well-organized reference to bundled documentation files. Novelty is high: GNN development requires specialized knowledge of graph representations, message passing paradigms, batching strategies for irregular structures, and PyG-specific APIs that would be token-intensive and error-prone for a CLI agent to implement from scratch. The skill meaningfully reduces complexity for geometric deep learning tasks. Minor point: could benefit from more explicit troubleshooting section, but overall extremely well-executed.
K-Dense-AI
Skill Author
Loading SKILL.md…
