How alisakour/conditional-nca-digits works
This small research repository presents a conditional neural cellular automaton (NCA) for growing class-specific handwritten digits. The documented workflow is to condition the model on a digit class and generate a digit through morphogenesis; the README describes growth from a single pixel. The notebook is the only implementation source shown, and the repository also contains a trained checkpoint. No source excerpts were provided, so internal stages and the checkpoint’s loading path cannot be verified.
Main entry points: conditional-nca-implementation.ipynb NCA model · c_nca_mnist.pth Trained weights
Components 5
Conditional NCA
- Class condition
conditional-nca-implementation.ipynb - NCA model
conditional-nca-implementation.ipynb - Trained weights
c_nca_mnist.pth - Digit growth
conditional-nca-implementation.ipynb
Other components
- User
Connections 4
- User → Class condition: selects class
README.md - Class condition → NCA model: conditions
README.md - NCA model → Digit growth: generates
README.md - Trained weights → NCA model: weights