Let's say I am building my neural network architecture by subclassing tensorflow's Model class. A standard way to do this would be to:
class IdentityBlock(tf.keras.Model):
def __init__(self, filters, kernel_size, **kwargs):
super().__init__(**kwargs)
self.conv1 = tf.keras.layers.Conv2D(filters, kernel_size, padding="same")
self.bn1 = tf.keras.layers.BatchNormalization()
self.act = tf.keras.layers.Activation("relu")
self.add = tf.keras.layers.Add()
def call(self, input_tensor):
x = self.conv1(input_tensor)
x = self.bn1(x)
x = self.act(x)
x = self.add([x, input_tensor])
return x
Which I understand, we subclass the Model class and any argument not specified has the default value in **kwargs. Now I also saw this option which is confusing me:
class IdentityBlock(tf.keras.Model):
def __init__(self, filters, kernel_size):
super(IdentityBlock, self).__init__(name='')
self.conv1 = tf.keras.layers.Conv2D(filters, kernel_size, padding="same")
self.bn1 = tf.keras.layers.BatchNormalization()
self.act = tf.keras.layers.Activation("relu")
self.add = tf.keras.layers.Add()
def call(self, input_tensor):
x = self.conv1(input_tensor)
x = self.bn1(x)
x = self.act(x)
x = self.add([x, input_tensor])
return x
Why would we pass IdentityBlock and self in super(), and also not put **kwargs in the __init__ parameters? What is the difference between these two subclassing methods, and if there isn't one, how are these the equivalent?