0.6.0

This release adds major neural-solver backend support and improves cross-platform compatibility.

Highlights

• Added TensorFlow support for neural solvers:
  • pca_gm
  • ipca_gm
  • cie
  • ngm
• Added MindSpore support for neural solvers:
  • pca_gm
  • ipca_gm
  • cie
  • ngm
• Added backend-native pretrained model loading for TensorFlow and MindSpore
• Added an offline converter for generating TensorFlow and MindSpore pretrained weights from PyTorch checkpoints
• Added RDKit graph matching examples and documentation
• Improved dataset links and CI compatibility
• Improved neural solver tests across backends

Since 0.5.5

Key changes include:

• TensorFlow and MindSpore neural solver implementations
• TensorFlow and MindSpore pretrained checkpoint support
• RDKit graph matching examples
• CI and test stability improvements
• README and documentation updates

Full Changelog: 0.5.5...0.6.0