Model1

simple keras model

###########Import
import keras
from keras.callbacks import ModelCheckpoint
from keras import models
from keras import layers
keras.__version__

###########Load MNIST
from keras.datasets import mnist
from keras.utils import to_categorical

(train_images, train_labels), (test_images, test_labels) = mnist.load_data()

train_images = train_images.reshape((60000, 28 * 28))
train_images = train_images.astype('float32') / 255

test_images = test_images.reshape((10000, 28 * 28))
test_images = test_images.astype('float32') / 255

train_labels = to_categorical(train_labels)
test_labels = to_categorical(test_labels)

###########Network1
network1 = models.Sequential()
network1.add(layers.Dense(10, activation='softmax', input_shape=(28 * 28,)))
network1.compile(optimizer='rmsprop',loss='categorical_crossentropy', metrics=['accuracy'])

###########Network2
network2 = models.Sequential()
network2.add(layers.Dense(512, activation='relu', input_shape=(28 * 28,)))
network2.add(layers.Dense(512, activation='relu', input_shape=(512,)))
network2.add(layers.Dense(10, activation='softmax'))
network2.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])

###########Fit both
result1 = network1.fit(train_images, train_labels, epochs= 100, batch_size=1024, validation_split = 0.1)
result1 = network2.fit(train_images, train_labels, epochs= 100, batch_size=1024, validation_split = 0.1)

###########Test
n1_test_loss, n1_test_acc = network1.evaluate(test_images, test_labels)
print('n1_test_acc:', n1_test_acc)
print('n1_test_loss:', n1_test_loss)

n2_test_loss, n2_test_acc = network2.evaluate(test_images, test_labels)
print('n2_test_acc:', n2_test_acc)
print('n2_test_loss:', n2_test_loss)