def main():
# load data
datasets = load_data('mnist.pkl.gz')
train_set_x, train_set_y = datasets[0]
valid_set_x, valid_set_y = datasets[1]
#print train_set_x.shape
"""
def main():
# load data
datasets = load_data("mnist.pkl.gz")
train_set_x, train_set_y = datasets[0]
valid_set_x, valid_set_y = datasets[1]
# make neural network
layers = []
layers.append(ImageInput(out_sx=28, out_sy=28, out_depth=1))
layers.append(LeNetConvPoolLayer(out_depth=7, filter_size=5))
layers.append(LeNetConvPoolLayer(out_depth=3, filter_size=5))
layers.append(Flattern())
layers.append(LogisticRegression(n_out=10))
# compile model and train
model = Model(layers)
model.fit(train_set_x, train_set_y, validation_data=[valid_set_x, valid_set_y])
