tensorflow: ValueError: name for name_scope must be a string when Building up a Custom Model

Please make sure that this is a bug. As per our GitHub Policy, we only address code/doc bugs, performance issues, feature requests and build/installation issues on GitHub. tag:bug_template

System information

• Have I written custom code (as opposed to using a stock example script provided in TensorFlow): yes
• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Windows 7
• Mobile device (e.g. iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device:
• TensorFlow installed from (source or binary):
• TensorFlow version (use command below): 2.0.0
• Python version: 3.7
• Bazel version (if compiling from source):
• GCC/Compiler version (if compiling from source):
• CUDA/cuDNN version: 10.1
• GPU model and memory:

You can collect some of this information using our environment capture script You can also obtain the TensorFlow version with: 1. TF 1.0: python -c "import tensorflow as tf; print(tf.GIT_VERSION, tf.VERSION)" 2. TF 2.0: python -c "import tensorflow as tf; print(tf.version.GIT_VERSION, tf.version.VERSION)"

Hi,

I tried to follow tf’s doc to build up a model as follows: https://www.tensorflow.org/guide/keras/custom_layers_and_models

`

class CBR(layers.Layer):
	'''
	Convolution + Batch normalisation + Relu
	'''
	def __int__(self, filterNum, kSize, strSize, padMode, name='cbr', **kwargs):
		super(CBR, self).__init__(name=name, **kwargs)
		self.conv3D = layers.Conv3D(filters=filterNum, kernel_size=kSize, strides=strSize, padding=padMode, data_format='channels_first')
		self.BN = layers.BatchNormalization(axis=1)
	def call(self, inputs):
		x = self.conv3D(inputs)
		x=self.BN(x)
		return layers.Relu(x)
		
class SimpleUNet(tf.keras.Model, layers.Layer):
	'''
	Serialise basic units so as to build up a double-layered encoder-decoder U-Net
	Input:
		inDim: [mbSize, modaility/channel, tensor dimensions]
		classNum: background included
		name: name for the net
		inputs: 5D tf tensor of [mbSize, modaility/channel, tensor dimensions]. Inputs must be organised into channel first
	Returns:
		outputs: 5D tf tensor of [mbSize, classNum, tensor dimensions]
	'''
	def __init__(self, inDim, classNum, name='SimpleUNet', **kwargs):
		super(SimpleUNet, self).__init__(name=name, **kwargs)
		self.inDim = inDim
		self.classNum = classNum
		dimEnSt1End = np.array(inDim[1:])-2-2
		dimEnSt2Ed = dimEnSt1End/2-2-2
		dimBridgeEnd = (dimEnSt2Ed/2-2-2)*2
		dimDEStd1End = (dimBridgeEnd-2-2)*2
		outDim = dimDEStd1End-2-2-2
		temp = ((dimEnSt2Ed - dimBridgeEnd)/2).astype('int32')
		crop3d1 = tuple(np.tile(temp, (2, 1)).T)
		temp = ((dimEnSt1End - dimDEStd1End)/2).astype('int32')
		crop3d2 = tuple(np.tile(temp, (2, 1)).T)
		# list of basic units used in the model
		self.en_st1_cbr1 = CBR(32, 3, 1, 'valid')
		self.en_st1_cbr2 = CBR(64, 3, 1, 'valid')
		self.en_st2_mp = layers.MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2), padding='valid', data_format='channels_first')
		self.en_st2_cbr1 = CBR(64, 3, 1, 'valid')
		self.en_st2_cbr2 = CBR(128, 3, 1, 'valid')
		self.bridge_mp = layers.MaxPooling3D(pool_size=(2, 2, 2), strides=(2, 2, 2), padding='valid', data_format='channels_first')
		self.bridge_cbr1 = CBR(128, 3, 1, 'valid')
		self.bridge_cbr2 = CBR(256, 3, 1, 'valid')    
		self.bridge_tconv1 = layers.Conv3DTranspose(256, 2, strides=2, padding='valid', data_format='channels_first')
		self.de_3dcrop1 = layers.Cropping3D(crop3d1, data_format='channels_first')
		self.de_st1_cbr1 = CBR(256, 3, 1, 'valid')
		self.de_st1_cbr2 = CBR(128, 3, 1, 'valid')    
		self.de_st1_tconv1 = layers.Conv3DTranspose(128, 2, strides=2, padding='valid', data_format='channels_first')
		self.de_3dcrop2 = layers.Cropping3D(crop3d2, data_format='channels_first')
		self.de_st2_cbr1 = CBR(64, 3, 1, 'valid')
		self.de_st2_cbr2 = CBR(64, 3, 1, 'valid') 
		self.final_conv3D = layers.Conv3D(filters=self.classNum, kernel_size=3, strides=1, padding='valid', data_format='channels_first')
				
	def call(self, inputs):
		x = self.en_st1_cbr1(inputs)
		xEnSt1End = self.en_st1_cbr2(x)
		x= self.en_st2_mp(xEnSt1End)
		x= self.en_st2_cbr1(x)
		xEnSt2Ed = self.en_st2_cbr2(x)
		x = self.bridge_mp(xEnSt2Ed)
		x = self.bridge_cbr1(x)
		x = self.bridge_cbr2(x)
		xBridgeEnd = self.bridge_tconv1(x)
		xCrop1 = self.de_3dcrop1(xEnSt2Ed)
		x = layers.Concatenate([xBridgeEnd, xCrop1], axis=1)
		x = self.de_st1_cbr1(x)
		x = self.de_st1_cbr2(x)
		xDeSt1End = self.de_st1_tconv1(x)
		xCrop2 = self.de_3dcrop2(xEnSt1End)
		x = layers.Concatenate([xDeSt1End, xCrop2], axis=1)
		x = self.de_st2_cbr1(x)
		x = self.de_st2_cbr2(x)
		x = self.final_conv3D(x)
		outputs = activations.softmax(x, axis=1)
		
		return outputs

`

Then I initialised it, and tried to build it by calling SUNet.build classNum = 3 mbSize = 16 inDim = [4, 64, 64, 64] SUNet = SimpleUNet(inDim, classNum) SUNet.build(input_shape=inDim) I strictly followed the example given in tf’s doc, but an error was raised when building up it ValueError: name for name_scope must be a string. It occurred here when CBR is called for the first time:

def __int__(self, filterNum, kSize, strSize, padMode, name='cbr', **kwargs): super(CBR, self).__init__(name=name, **kwargs)

I cannot figure out any syntactic mistake. Could anyone give me a hand? Thanks a lot. Or, is the model cannot be built at this moment until it is actually used when being called in the training?