Shypd.aiThe largest open index of AI toolsJittorGeometric is an open-source graph machine learning library built on the Jittor framework. It provides comprehensive support for Graph Neural Networks (GNNs) research and applications, featuring enhanced performance, flexibility, and scalability.
Trending
About
JittorGeometric 2.0 is a state-of-the-art graph machine learning library built on the Jittor framework, a Chinese-developed deep learning library. It offers comprehensive support for Graph Neural Networks (GNNs) research and applications, emphasizing enhanced performance, flexibility, and scalability. Key features include Just-In-Time (JIT) compilation for dynamic code modification, optimized sparse operations with CuSparse acceleration, and a comprehensive model zoo with over 40 implemented GNN models covering classic, spectral, dynamic, molecular, and transformer-based architectures. The library also provides rich dataset support for popular graph datasets like Planetoid and OGB. Version 2.0 introduces distributed training capabilities, dynamic graph processing, mini-batch support for large-scale graphs, and GNN support for NPUs (Ascend-GNN).
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending
