A simple neural network library, writen in python for educational purposes. It provides a simple structure to promote researchers in developing their own layers and functions. Advanced tools such as Tensorflow and Pytorch deduce gradients automaticaly by following variable calculatons. On the contrary python_nn requieres for the user to derive the gradients. It also includes a framework to test the derivations.
Simple installation using pip:
$ pip install git+https://github.com/theo2021/Python_NN.gitA simple jupyter example can be found here There are plenty more options, more examples will be coming.
The code was developed during the course Deep Learning in Data Science.
Please send me a feedback about the project in theo2021@gmail.com .