Computation graphs will only be defined; in order to execute the computation graph, we use TensorFlow sessions:

sess = tf.Session()

We can create the session for our computation graph using the tf.Session() method, which will allocate the memory for storing the current value of the variable. After creating the session, we can execute our graph with the sess.run() method.

In order to run anything in TensorFlow, we need to start the TensorFlow session for an instance; please refer to the code:

import tensorflow as tf
a = tf.multiply(2,3)
print(a)

It will print a TensorFlow object instead of 6. As already said, whenever we import TensorFlow a default computation graph will automatically be created and all nodes a that we created will get attached to the graph. In order to execute the graph, we need to initialize a TensorFlow session as follows:

#Import tensorflow 
import tensorflow as tf

#Initialize variables
a = tf.multiply(2,3)

#create tensorflow session for executing the session
with tf.Session() as sess:
  #run the session
  print(sess.run(a))

The preceding code will print 6.