Module juice::layers::utility[][src]

Expand description

Provides various helpful layers, which might be not directly related to neural networks in general.

These layers do not have to necesarrely manipulate the data flowing through them and might have no effect on the Networks’ capabilities to learn (e.g. loging) but obey all the rules of a Layer. The type of these layers can vary a lot. From data normalization to specific data access layers for e.g. a database like LevelDB.


pub use self::flatten::Flatten;
pub use self::reshape::Reshape;
pub use self::reshape::ReshapeConfig;


Flattens the bottom Blob into a simpler top Blob.

Utility layer to give a tensor another shape.