Author: Jakub Cisło
http://cislo.net.pl
jakub@cislo.net.pl
Implementation of neural network in Python using Numpy and Autograd.
Method: transformation matrix and stochastic gradient descent
Required: python2
, numpy
, matplotlib
, autograd
, MNIST and CIFAR-10 datasets
##Usage:
python -i mnist.py
python -i cifar.py
###Interactive commands
create(layers=[28*28, 100, 10], batch_size=32, dropout=0.1)
- create a new modellearn()
- start the model learningsave(name)
- save the model to fileload(name)
- load the model from fileinfo()
- show info and statistics about the modelbest()
- show best images from each categoryworst()
- show worst images from each category