def test_summary(self):
class ToString(object):
def __init__(self):
self.contents = ''
def __call__(self, msg):
self.contents += msg + '\n'
# Single-io
model = testing_utils.SmallSubclassMLP(num_hidden=32,
num_classes=4,
use_bn=True,
use_dp=True)
model(np.ones((3, 4))) # need to build model first
print_fn = ToString()
model.summary(print_fn=print_fn)
self.assertTrue('Trainable params: 356' in print_fn.contents)
# Multi-io
model = model_util.get_multi_io_subclass_model(num_classes=(5, 6),
use_bn=True,
use_dp=True)
model([np.ones((3, 4)), np.ones((3, 4))]) # need to build model first
print_fn = ToString()
model.summary(print_fn=print_fn)
self.assertTrue('Trainable params: 587' in print_fn.contents)
def __init__(self, num_classes=2):
super(NestedTestModel1, self).__init__(name='nested_model_1')
self.num_classes = num_classes
self.dense1 = keras.layers.Dense(32, activation='relu')
self.dense2 = keras.layers.Dense(num_classes, activation='relu')
self.bn = keras.layers.BatchNormalization()
self.test_net = testing_utils.SmallSubclassMLP(num_hidden=32,
num_classes=4,
use_bn=True,
use_dp=True)
def test_trainable_custom_model_false(self):
"""Tests that overall False trainable status of Model is preserved."""
# Set all layers to *not* be trainable.
model = testing_utils.SmallSubclassMLP(1, 4, trainable=False)
model.compile(loss='mse', optimizer='rmsprop')
self._train_model(model, use_dataset=False)
loaded = self._save_and_load(model)
self._test_evaluation(model, loaded)
self.assertEmpty(model.trainable_variables)
self.assertEmpty(loaded.trainable_variables)
def test_invalid_input_shape_build(self):
num_classes = 2
input_dim = 50
model = testing_utils.SmallSubclassMLP(
num_hidden=32, num_classes=num_classes, use_dp=True, use_bn=True)
self.assertFalse(model.built, 'Model should not have been built')
self.assertFalse(model.weights, ('Model should have no weights since it '
'has not been built.'))
with self.assertRaisesRegexp(
ValueError, 'input shape is not one of the valid types'):
model.build(input_shape=tensor_shape.Dimension(input_dim))
def test_single_io_workflow_with_tensors(self):
num_classes = 2
num_samples = 10
input_dim = 50
with ops.Graph().as_default(), self.cached_session():
model = testing_utils.SmallSubclassMLP(
num_hidden=32, num_classes=num_classes, use_dp=True, use_bn=True)
model.compile(loss='mse', optimizer='rmsprop')
x = array_ops.ones((num_samples, input_dim))
y = array_ops.zeros((num_samples, num_classes))
model.fit(x, y, epochs=2, steps_per_epoch=10, verbose=0)
_ = model.evaluate(steps=10, verbose=0)
def test_single_io_dimension_subclass_build(self):
num_classes = 2
input_dim = tensor_shape.Dimension(50)
batch_size = tensor_shape.Dimension(None)
model = testing_utils.SmallSubclassMLP(
num_hidden=32, num_classes=num_classes, use_dp=True, use_bn=True)
self.assertFalse(model.built, 'Model should not have been built')
self.assertFalse(model.weights, ('Model should have no weights since it '
'has not been built.'))
model.build(input_shape=(batch_size, input_dim))
self.assertTrue(model.weights, ('Model should have weights now that it '
'has been properly built.'))
self.assertTrue(model.built, 'Model should be built after calling `build`.')
model(array_ops.ones((32, input_dim)))
def test_single_io_workflow_with_np_arrays(self):
num_classes = 2
num_samples = 100
input_dim = 50
model = testing_utils.SmallSubclassMLP(
num_hidden=32, num_classes=num_classes, use_dp=True, use_bn=True)
model.compile(
loss='mse',
optimizer='rmsprop',
metrics=['acc', keras.metrics.CategoricalAccuracy()],
run_eagerly=testing_utils.should_run_eagerly(),
experimental_run_tf_function=testing_utils.should_run_tf_function())
x = np.ones((num_samples, input_dim))
y = np.zeros((num_samples, num_classes))
model.fit(x, y, epochs=2, batch_size=32, verbose=0)
_ = model.evaluate(x, y, verbose=0)
def test_single_io_workflow_with_datasets(self):
num_classes = 2
num_samples = 10
input_dim = 50
with self.cached_session():
model = testing_utils.SmallSubclassMLP(num_hidden=32,
num_classes=num_classes,
use_dp=True,
use_bn=True)
model.compile(loss='mse',
optimizer='rmsprop',
run_eagerly=testing_utils.should_run_eagerly())
x = np.ones((num_samples, input_dim), dtype=np.float32)
y = np.zeros((num_samples, num_classes), dtype=np.float32)
dataset = dataset_ops.Dataset.from_tensor_slices((x, y))
dataset = dataset.repeat(100)
dataset = dataset.batch(10)
model.fit(dataset, epochs=2, steps_per_epoch=10, verbose=0)
_ = model.evaluate(dataset, steps=10, verbose=0)
