deep_nn_model_with_only_python_100%_test_accuracy

deep learning model with only python and numpy with test accuracy 99 % on mnist dataset and different optimization choices

model description

the model implemented in just one function because i thought it will be helpful to see the whole training process in one place

model features

1 - binary model 2- contain three different optimization methods (gd , rms_prob , adam) 3- contain l2 regularization 4- different choices for activations function (sigmoid , relu , tanh) 5- predict function

what you can improve

1- you can rebuild the model with classes to be more structured 2- you can add softmax activation in case you want use it for multiclasses i didn,t implement it because i thought there is more important think to focus on