def testGetVariableGivenNameScoped(self):
with self.test_session():
with variable_scope.variable_scope('A'):
a = variables_lib2.variable('a', [5])
b = variables_lib2.variable('b', [5])
self.assertEquals([a], variables_lib2.get_variables_by_name('a'))
self.assertEquals([b], variables_lib2.get_variables_by_name('b'))
def testEmptyUpdateOps(self):
with ops.Graph().as_default():
random_seed.set_random_seed(0)
tf_inputs = constant_op.constant(self._inputs, dtype=dtypes.float32)
tf_labels = constant_op.constant(self._labels, dtype=dtypes.float32)
tf_predictions = batchnorm_classifier(tf_inputs)
loss_ops.log_loss(tf_predictions, tf_labels)
total_loss = loss_ops.get_total_loss()
optimizer = gradient_descent.GradientDescentOptimizer(learning_rate=1.0)
train_op = training.create_train_op(total_loss, optimizer, update_ops=[])
moving_mean = variables_lib.get_variables_by_name('moving_mean')[0]
moving_variance = variables_lib.get_variables_by_name('moving_variance')[
0]
with session_lib.Session() as sess:
# Initialize all variables
sess.run(variables_lib2.global_variables_initializer())
mean, variance = sess.run([moving_mean, moving_variance])
# After initialization moving_mean == 0 and moving_variance == 1.
self.assertAllClose(mean, [0] * 4)
self.assertAllClose(variance, [1] * 4)
for _ in range(10):
sess.run([train_op])
mean = moving_mean.eval()
variance = moving_variance.eval()
# Since we skip update_ops the moving_vars are not updated.
self.assertAllClose(mean, [0] * 4)
self.assertAllClose(variance, [1] * 4)
def testCreateVariables(self):
height, width = 3, 3
images = random_ops.random_uniform((5, height, width, 3), seed=1)
normalization.instance_norm(images, center=True, scale=True)
beta = contrib_variables.get_variables_by_name('beta')[0]
gamma = contrib_variables.get_variables_by_name('gamma')[0]
self.assertEqual('InstanceNorm/beta', beta.op.name)
self.assertEqual('InstanceNorm/gamma', gamma.op.name)
def testTrainAllVarsHasLowerLossThanTrainSubsetOfVars(self):
logdir1 = os.path.join(tempfile.mkdtemp(prefix=self.get_temp_dir()),
'tmp_logs1')
# First, train only the weights of the model.
with ops.Graph().as_default():
random_seed.set_random_seed(0)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(
learning_rate=1.0)
weights = variables_lib2.get_variables_by_name('weights')
train_op = learning.create_train_op(total_loss,
optimizer,
variables_to_train=weights)
loss = learning.train(train_op,
logdir1,
number_of_steps=200,
log_every_n_steps=10)
self.assertGreater(loss, .015)
self.assertLess(loss, .05)
# Next, train the biases of the model.
with ops.Graph().as_default():
random_seed.set_random_seed(1)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(
learning_rate=1.0)
biases = variables_lib2.get_variables_by_name('biases')
train_op = learning.create_train_op(total_loss,
optimizer,
variables_to_train=biases)
loss = learning.train(train_op,
logdir1,
number_of_steps=300,
log_every_n_steps=10)
self.assertGreater(loss, .015)
self.assertLess(loss, .05)
# Finally, train both weights and bias to get lower loss.
with ops.Graph().as_default():
random_seed.set_random_seed(2)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(
learning_rate=1.0)
train_op = learning.create_train_op(total_loss, optimizer)
loss = learning.train(train_op,
logdir1,
number_of_steps=400,
log_every_n_steps=10)
self.assertIsNotNone(loss)
self.assertLess(loss, .015)
def testReuseVariables(self):
height, width = 3, 3
images = random_ops.random_uniform((5, height, width, 3), seed=1)
normalization.instance_norm(images, scale=True, scope='IN')
normalization.instance_norm(images, scale=True, scope='IN', reuse=True)
beta = contrib_variables.get_variables_by_name('beta')
gamma = contrib_variables.get_variables_by_name('gamma')
self.assertEqual(1, len(beta))
self.assertEqual(1, len(gamma))
def testCreateOpNoScaleCenter(self):
height, width, groups = 3, 3, 7
images = random_ops.random_uniform(
(5, height, width, 3*groups), dtype=dtypes.float32, seed=1)
output = normalization.group_norm(images, groups=groups, center=False,
scale=False)
self.assertListEqual([5, height, width, 3*groups], output.shape.as_list())
self.assertEqual(0, len(contrib_variables.get_variables_by_name('beta')))
self.assertEqual(0, len(contrib_variables.get_variables_by_name('gamma')))
def testGetVariableWithoutScope(self):
with self.test_session():
a = variables_lib2.variable('a', [5])
fooa = variables_lib2.variable('fooa', [5])
b_a = variables_lib2.variable('B/a', [5])
matched_variables = variables_lib2.get_variables_by_name('a')
self.assertEquals([a, b_a], matched_variables)
matched_variables = variables_lib2.get_variables_by_name('fooa')
self.assertEquals([fooa], matched_variables)
def testCreateOpNoScaleCenter(self):
height, width = 3, 3
images = random_ops.random_uniform(
(5, height, width, 3), dtype=dtypes.float64, seed=1)
output = normalization.instance_norm(images, center=False, scale=False)
self.assertStartsWith(
output.op.name, 'InstanceNorm/instancenorm')
self.assertListEqual([5, height, width, 3], output.shape.as_list())
self.assertEqual(0, len(contrib_variables.get_variables_by_name('beta')))
self.assertEqual(0, len(contrib_variables.get_variables_by_name('gamma')))
def testCreateVariables_NCHW(self):
height, width, groups = 3, 3, 4
images = random_ops.random_uniform((5, 2*groups, height, width), seed=1)
normalization.group_norm(images, groups=4,
channels_axis=-3, reduction_axes=(-2, -1),
center=True, scale=True)
beta = contrib_variables.get_variables_by_name('beta')[0]
gamma = contrib_variables.get_variables_by_name('gamma')[0]
self.assertEqual('GroupNorm/beta', beta.op.name)
self.assertEqual('GroupNorm/gamma', gamma.op.name)
def testGetVariableWithScope(self):
with self.test_session():
with variable_scope.variable_scope('A'):
a = variables_lib2.variable('a', [5])
fooa = variables_lib2.variable('fooa', [5])
with variable_scope.variable_scope('B'):
a2 = variables_lib2.variable('a', [5])
matched_variables = variables_lib2.get_variables_by_name('a')
self.assertEquals([a, a2], matched_variables)
matched_variables = variables_lib2.get_variables_by_name('fooa')
self.assertEquals([fooa], matched_variables)
matched_variables = variables_lib2.get_variables_by_name('/a')
self.assertEquals([], matched_variables)
matched_variables = variables_lib2.get_variables_by_name('a', scope='A')
self.assertEquals([a], matched_variables)
def testTrainAllVarsHasLowerLossThanTrainSubsetOfVars(self):
logdir1 = os.path.join(
tempfile.mkdtemp(prefix=self.get_temp_dir()), 'tmp_logs1')
# First, train only the weights of the model.
with ops.Graph().as_default():
random_seed.set_random_seed(0)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(learning_rate=1.0)
weights = variables_lib2.get_variables_by_name('weights')
train_op = learning.create_train_op(
total_loss, optimizer, variables_to_train=weights)
loss = learning.train(
train_op, logdir1, number_of_steps=200, log_every_n_steps=10)
self.assertGreater(loss, .015)
self.assertLess(loss, .05)
# Next, train the biases of the model.
with ops.Graph().as_default():
random_seed.set_random_seed(1)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(learning_rate=1.0)
biases = variables_lib2.get_variables_by_name('biases')
train_op = learning.create_train_op(
total_loss, optimizer, variables_to_train=biases)
loss = learning.train(
train_op, logdir1, number_of_steps=300, log_every_n_steps=10)
self.assertGreater(loss, .015)
self.assertLess(loss, .05)
# Finally, train both weights and bias to get lower loss.
with ops.Graph().as_default():
random_seed.set_random_seed(2)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(learning_rate=1.0)
train_op = learning.create_train_op(total_loss, optimizer)
loss = learning.train(
train_op, logdir1, number_of_steps=400, log_every_n_steps=10)
self.assertIsNotNone(loss)
self.assertLess(loss, .015)
def testUseUpdateOps(self):
with ops.Graph().as_default():
random_seed.set_random_seed(0)
tf_inputs = constant_op.constant(self._inputs,
dtype=dtypes.float32)
tf_labels = constant_op.constant(self._labels,
dtype=dtypes.float32)
expected_mean = np.mean(self._inputs, axis=(0))
expected_var = np.var(self._inputs, axis=(0))
expected_var = self._addBesselsCorrection(16, expected_var)
tf_predictions = BatchNormClassifier(tf_inputs)
loss_ops.log_loss(tf_predictions, tf_labels)
total_loss = loss_ops.get_total_loss()
optimizer = gradient_descent.GradientDescentOptimizer(
learning_rate=1.0)
train_op = learning.create_train_op(total_loss, optimizer)
moving_mean = variables_lib2.get_variables_by_name(
'moving_mean')[0]
moving_variance = variables_lib2.get_variables_by_name(
'moving_variance')[0]
with session.Session() as sess:
# Initialize all variables
sess.run(variables_lib.global_variables_initializer())
mean, variance = sess.run([moving_mean, moving_variance])
# After initialization moving_mean == 0 and moving_variance == 1.
self.assertAllClose(mean, [0] * 4)
self.assertAllClose(variance, [1] * 4)
for _ in range(10):
sess.run([train_op])
mean = moving_mean.eval()
variance = moving_variance.eval()
# After 10 updates with decay 0.1 moving_mean == expected_mean and
# moving_variance == expected_var.
self.assertAllClose(mean, expected_mean)
self.assertAllClose(variance, expected_var)
def testUseUpdateOps(self):
with ops.Graph().as_default():
random_seed.set_random_seed(0)
tf_inputs = constant_op.constant(self._inputs, dtype=dtypes.float32)
tf_labels = constant_op.constant(self._labels, dtype=dtypes.float32)
expected_mean = np.mean(self._inputs, axis=(0))
expected_var = np.var(self._inputs, axis=(0))
expected_var = self._addBesselsCorrection(16, expected_var)
tf_predictions = BatchNormClassifier(tf_inputs)
loss_ops.log_loss(tf_predictions, tf_labels)
total_loss = loss_ops.get_total_loss()
optimizer = gradient_descent.GradientDescentOptimizer(learning_rate=1.0)
train_op = learning.create_train_op(total_loss, optimizer)
moving_mean = variables_lib2.get_variables_by_name('moving_mean')[0]
moving_variance = variables_lib2.get_variables_by_name('moving_variance')[
0]
with session.Session() as sess:
# Initialize all variables
sess.run(variables_lib.global_variables_initializer())
mean, variance = sess.run([moving_mean, moving_variance])
# After initialization moving_mean == 0 and moving_variance == 1.
self.assertAllClose(mean, [0] * 4)
self.assertAllClose(variance, [1] * 4)
for _ in range(10):
sess.run([train_op])
mean = moving_mean.eval()
variance = moving_variance.eval()
# After 10 updates with decay 0.1 moving_mean == expected_mean and
# moving_variance == expected_var.
self.assertAllClose(mean, expected_mean)
self.assertAllClose(variance, expected_var)
def testTrainAllVarsHasLowerLossThanTrainSubsetOfVars(self):
logdir = os.path.join(self.get_temp_dir(), 'tmp_logs3/')
if gfile.Exists(logdir): # For running on jenkins.
gfile.DeleteRecursively(logdir)
# First, train only the weights of the model.
with ops.Graph().as_default():
random_seed.set_random_seed(0)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(learning_rate=1.0)
weights = variables_lib.get_variables_by_name('weights')
train_op = training.create_train_op(
total_loss, optimizer, variables_to_train=weights)
saver = saver_lib.Saver()
loss = training.train(
train_op,
logdir,
hooks=[
basic_session_run_hooks.CheckpointSaverHook(
logdir, save_steps=1, saver=saver),
basic_session_run_hooks.StopAtStepHook(num_steps=200),
])
self.assertGreater(loss, .015)
self.assertLess(loss, .05)
# Next, train the biases of the model.
with ops.Graph().as_default():
random_seed.set_random_seed(1)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(learning_rate=1.0)
biases = variables_lib.get_variables_by_name('biases')
train_op = training.create_train_op(
total_loss, optimizer, variables_to_train=biases)
saver = saver_lib.Saver()
loss = training.train(
train_op,
logdir,
hooks=[
basic_session_run_hooks.CheckpointSaverHook(
logdir, save_steps=1, saver=saver),
basic_session_run_hooks.StopAtStepHook(num_steps=300),
])
self.assertGreater(loss, .015)
self.assertLess(loss, .05)
# Finally, train both weights and bias to get lower loss.
with ops.Graph().as_default():
random_seed.set_random_seed(2)
total_loss = self.ModelLoss()
optimizer = gradient_descent.GradientDescentOptimizer(learning_rate=1.0)
train_op = training.create_train_op(total_loss, optimizer)
saver = saver_lib.Saver()
loss = training.train(
train_op,
logdir,
hooks=[
basic_session_run_hooks.CheckpointSaverHook(
logdir, save_steps=1, saver=saver),
basic_session_run_hooks.StopAtStepHook(num_steps=400),
])
self.assertIsNotNone(loss)
self.assertLess(loss, .015)
