Ease of usage version

Ton of new features:

• CrossEntropy loss
• Model saving and loading
• CLI model defining and training
• CSV dataset loading
• Actual documentation
• pip deployment

I skipped 0.4 to fight bugs on test.pypi and ordinary pypi

Lowkey usable version

In this release, several useful features were added such as ReLU activation and batch shuffling

Mathematically correct release

Fix of math bugs, after discovering error in convergence. The bugs were discovered almost by accident, but were sucessfully anihilated with the help of two separate ML blogs.

First working release

First release which converges sucessfully.

Activation functions:

• Sigmoid
• TODO:
  • ReLU, Tanh, ...

Loss functions:

• Mean Squared Error
• TODO:
  • Cross Entropy