def testConvBatchNormArgument(self):
st = self.input_layer.sequential()
st.reshape([DIM_SAME, DIM_SAME, DIM_SAME, 1])
st.conv2d(3, 2,
batch_normalize=prettytensor.BatchNormalizationArguments(
scale_after_normalization=False))
self.assertEqual(2,
len(tf.get_collection(prettytensor.GraphKeys.UPDATE_OPS)))
self.assertTrue(tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, '.*/beta'))
self.assertFalse(tf.get_collection(
tf.GraphKeys.GLOBAL_VARIABLES, '.*/gamma'))
self.assertTrue(tf.get_collection(
tf.GraphKeys.GLOBAL_VARIABLES, '.*/moving_variance'))
self.assertTrue(tf.get_collection(
tf.GraphKeys.GLOBAL_VARIABLES, '.*/moving_mean'))
def testConvBatchNorm(self):
st = self.input_layer.sequential()
st.reshape([DIM_SAME, DIM_SAME, DIM_SAME, 1])
with prettytensor.defaults_scope(
batch_normalize=prettytensor.BatchNormalizationArguments(
learned_moments_update_rate=0.0003,
variance_epsilon=0.001,
scale_after_normalization=True)):
st.conv2d(3, 2)
self.assertEqual(2,
len(tf.get_collection(prettytensor.GraphKeys.UPDATE_OPS)))
self.assertTrue(tf.get_collection(tf.GraphKeys.GLOBAL_VARIABLES, '.*/beta'))
self.assertTrue(tf.get_collection(
tf.GraphKeys.GLOBAL_VARIABLES, '.*/gamma'))
self.assertTrue(tf.get_collection(
tf.GraphKeys.GLOBAL_VARIABLES, '.*/moving_variance'))
self.assertTrue(tf.get_collection(
tf.GraphKeys.GLOBAL_VARIABLES, '.*/moving_mean'))
return x_train, y_train
img_size = 28
img_size_flat = img_size * img_size
img_shape = (img_size, img_size)
num_channels = 1
num_classes = 10
x = tf.placeholder(tf.float32, shape=[None, img_size_flat], name='x')
x_image = tf.reshape(x, [-1, img_size, img_size, num_channels])
y_true = tf.placeholder(tf.float32, shape=[None, 10], name='y_true')
y_true_cls = tf.argmax(y_true, dimension=1)
x_wrap = pt.wrap(x_image)
norm = pt.BatchNormalizationArguments(scale_after_normalization=True)
with pt.defaults_scope(activation_fn=tf.nn.relu):
y_pred, loss = x_wrap.\
conv2d(kernel=5, depth=16, name='conv1', batch_normalize=norm).\
max_pool(kernel=2, stride=2).\
conv2d(kernel=5, depth=36, name='conv2', batch_normalize=norm).\
max_pool(kernel=2, stride=2).\
flatten().\
fully_connected(size=128, name='fc1').\
softmax_classifier(num_classes=num_classes, labels=y_true)
optimizer = tf.train.AdamOptimizer(learning_rate=1e-4).minimize(loss)
y_pred_cls = tf.argmax(y_pred, dimension=1)
correct_prediction = tf.equal(y_pred_cls, y_true_cls)
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
