Big changes!

The Network wrapper has been completely rewritten from the ground up, fixing some major bugs. The wrapper is also a lot more flexible now, making it easier to integrate new components in the future without having to rewrite any internals.
Validation testing also supported when using network wrapper.
More flexible and accurate Accuracy/Loss reporting.
Fixed bug with Dropout layers when trying to get inference from the model.
Added new Loss functions and activation layers to support a larger variety of models like classification, linear regression, binary logistic regression.

Provide feedback