What is the use of call method and when to use it?

[everyday1]

I am having a hardtime understanding call methods in python. I have this class object

class Attention(tf.keras.Model):
    def __init__(self, units):
        super(Attention, self).__init__()
        self.W1 = tf.keras.layers.Dense(units)
        self.W2 = tf.keras.layers.Dense(units)
        self.V = tf.keras.layers.Dense(1)

    def call(self, features, hidden):
        hidden_with_time_axis = tf.expand_dims(hidden, 1)
        score = tf.nn.tanh(self.W1(features) + self.W2(hidden_with_time_axis))
        attention_weights = tf.nn.softmax(self.V(score), axis=1)
        context_vector = attention_weights * features
        context_vector = tf.reduce_sum(context_vector, axis=1)

        return context_vector, attention_weights

In this class object two Dense layer attributes self.W1 and self.W1 were created and then a features and a hidden_with_time_axis attribute is called on W1 and W2.

I know that when we initiate the line self.W1 = tf.keras.layers.Dense(units) its creating a Dense neural network layer of n units. But under call method they are doing this self.W1(features). What is the purpose of this and what is the intentional behaviour?

I tried to make a sample class object for understanding classes

class Foo:
    def __init__(self, units):
        self.units=units
    def __call__(self):
        print ('called '+self.units)


a=Foo(3)
b=Foo(a)

But I still dont understand the use of a call object and when and where to use it? Can anyone explain it in a simple manner?.

[ichabod801]

It's a way to create functions out of class instances.

In the expression bar = foo(), the parentheses are an operator. Specifically, they are the function call operator. Normally you only see this with functions:

def foo(x):
    return 2 * x + 1
bar = foo(5)
#        ^ function call operator

The __call__ method of a class allows you to override that operator, just as the __add__ method allows you to override the + operator. This allows your class to operate as if it were a function. A typical example is to make an efficient factorial function by storing previously calculated values:

class Factorial(object):

    def __init__(self):
        self.facts = [1, 1]

    def __call__(self, x):
        try:
            return self.facts[x]
        except IndexError:
            while len(self.facts) <= x:
                self.facts.append(self.facts[-1] * len(self.facts))
            return self.facts[-1]

You can now create an instance of the factorial object and use it as a function:

Output:
>>> factorial = Factorial()
>>> factorial(5)    # adds items to factorial.facts
120
>>> factorial(3)    # Use factorial.facts[3]
6