def testBuildPreLogitsCifarModel(self):
batch_size = 5
height, width = 32, 32
num_classes = None
inputs = tf.random_uniform((batch_size, height, width, 3))
tf.train.create_global_step()
with slim.arg_scope(nasnet.nasnet_cifar_arg_scope()):
net, end_points = nasnet.build_nasnet_cifar(inputs, num_classes)
self.assertFalse('AuxLogits' in end_points)
self.assertFalse('Predictions' in end_points)
self.assertTrue(net.op.name.startswith('final_layer/Mean'))
self.assertListEqual(net.get_shape().as_list(), [batch_size, 768])
def testOverrideHParamsCifarModel(self):
batch_size = 5
height, width = 32, 32
num_classes = 10
inputs = tf.random_uniform((batch_size, height, width, 3))
tf.train.create_global_step()
config = nasnet.cifar_config()
config.set_hparam('data_format', 'NCHW')
with slim.arg_scope(nasnet.nasnet_cifar_arg_scope()):
_, end_points = nasnet.build_nasnet_cifar(inputs,
num_classes,
config=config)
self.assertListEqual(end_points['Stem'].shape.as_list(),
[batch_size, 96, 32, 32])
def testNoAuxHeadCifarModel(self):
batch_size = 5
height, width = 32, 32
num_classes = 10
for use_aux_head in (True, False):
tf.reset_default_graph()
inputs = tf.random_uniform((batch_size, height, width, 3))
tf.train.create_global_step()
config = nasnet.cifar_config()
config.set_hparam('use_aux_head', int(use_aux_head))
with slim.arg_scope(nasnet.nasnet_cifar_arg_scope()):
_, end_points = nasnet.build_nasnet_cifar(inputs,
num_classes,
config=config)
self.assertEqual('AuxLogits' in end_points, use_aux_head)
def testBuildLogitsCifarModel(self):
batch_size = 5
height, width = 32, 32
num_classes = 10
inputs = tf.random_uniform((batch_size, height, width, 3))
tf.train.create_global_step()
with slim.arg_scope(nasnet.nasnet_cifar_arg_scope()):
logits, end_points = nasnet.build_nasnet_cifar(inputs, num_classes)
auxlogits = end_points['AuxLogits']
predictions = end_points['Predictions']
self.assertListEqual(auxlogits.get_shape().as_list(),
[batch_size, num_classes])
self.assertListEqual(logits.get_shape().as_list(),
[batch_size, num_classes])
self.assertListEqual(predictions.get_shape().as_list(),
[batch_size, num_classes])
def testUseBoundedAcitvationCifarModel(self):
batch_size = 1
height, width = 32, 32
num_classes = 10
for use_bounded_activation in (True, False):
tf.reset_default_graph()
inputs = tf.random_uniform((batch_size, height, width, 3))
config = nasnet.cifar_config()
config.set_hparam('use_bounded_activation', use_bounded_activation)
with slim.arg_scope(nasnet.nasnet_cifar_arg_scope()):
_, _ = nasnet.build_nasnet_cifar(inputs,
num_classes,
config=config)
for node in tf.get_default_graph().as_graph_def().node:
if node.op.startswith('Relu'):
self.assertEqual(node.op == 'Relu6',
use_bounded_activation)
def testAllEndPointsShapesCifarModel(self):
batch_size = 5
height, width = 32, 32
num_classes = 10
inputs = tf.random_uniform((batch_size, height, width, 3))
tf.train.create_global_step()
with slim.arg_scope(nasnet.nasnet_cifar_arg_scope()):
_, end_points = nasnet.build_nasnet_cifar(inputs, num_classes)
endpoints_shapes = {
'Stem': [batch_size, 32, 32, 96],
'Cell_0': [batch_size, 32, 32, 192],
'Cell_1': [batch_size, 32, 32, 192],
'Cell_2': [batch_size, 32, 32, 192],
'Cell_3': [batch_size, 32, 32, 192],
'Cell_4': [batch_size, 32, 32, 192],
'Cell_5': [batch_size, 32, 32, 192],
'Cell_6': [batch_size, 16, 16, 384],
'Cell_7': [batch_size, 16, 16, 384],
'Cell_8': [batch_size, 16, 16, 384],
'Cell_9': [batch_size, 16, 16, 384],
'Cell_10': [batch_size, 16, 16, 384],
'Cell_11': [batch_size, 16, 16, 384],
'Cell_12': [batch_size, 8, 8, 768],
'Cell_13': [batch_size, 8, 8, 768],
'Cell_14': [batch_size, 8, 8, 768],
'Cell_15': [batch_size, 8, 8, 768],
'Cell_16': [batch_size, 8, 8, 768],
'Cell_17': [batch_size, 8, 8, 768],
'Reduction_Cell_0': [batch_size, 16, 16, 256],
'Reduction_Cell_1': [batch_size, 8, 8, 512],
'global_pool': [batch_size, 768],
# Logits and predictions
'AuxLogits': [batch_size, num_classes],
'Logits': [batch_size, num_classes],
'Predictions': [batch_size, num_classes]
}
self.assertItemsEqual(endpoints_shapes.keys(), end_points.keys())
for endpoint_name in endpoints_shapes:
tf.logging.info('Endpoint name: {}'.format(endpoint_name))
expected_shape = endpoints_shapes[endpoint_name]
self.assertTrue(endpoint_name in end_points)
self.assertListEqual(
end_points[endpoint_name].get_shape().as_list(),
expected_shape)