public save(path as Path, includeOptimizer as Boolean = true) as Void
Saves architecture, exact parameter bits and optional training state to KNN 1.0. Built-in stages include convolutions, transposed convolutions, pooling, Dropout and activations. Custom classes are explicitly rejected. GPU parameters are synchronized; the selected device is not persisted. The model must be built and must not be mutated concurrently. Publication replaces an older checkpoint atomically only after a complete write.
| Parameter | Description |
|---|---|
path | Destination, normally ending in .knn. No extension also uses KNN. |
includeOptimizer | False creates an uncompiled inference checkpoint. |
model.save(FilePath("MnistCNN.knn"))