This repo will contain various neural networks implementations based only on numpy.
The point of this project is to get some practice by working on the building blocks of
deep learning, so that I get a decent understanding of the fundations of DL.
This proper understanding of the fundations will be needed for creating new models during m PhD.
For now implementation is not meant to be modular but will soon be (writting good DL libraries is not a reasonable goal for now).
The kinds of architectures I want to implement are:
- 5 layers MLP
- Simple CNN
- AlexNet
- LSTM