What is Tensor
In simple language we can call as multi dimensional array. Let us try to understand different dimensions
Zero dimension : It is a single point or a scalar value
One dimension : It is a line , it consist of zero dimension points
Two dimensions : It is a matrix ,
N dimensions : Tensor
What is graph and what is session
Tensors flow across a graph consist of operations. Here tensors flows through each node or operation under a session. The graph consist of nodes or operations and tensors are our inputs or resultants out of the nodes.
Each session is a process flow across the graph using nodes and tensors.
Having said all these complex words to understand let us first start with some examples.
Importing library
import tensorflow as tf
Creating graph
import tensorflow as tf
my_graph = tf.Graph()
with my_graph.as_default():
a = tf.constant([50], name = 'my_const_a')
Here we have added one constant. Now we will create session and see the result.
import tensorflow as tf
my_graph = tf.Graph()
with my_graph.as_default():
a = tf.constant([50], name = 'my_const_a')
my_sess = tf.Session(graph = my_graph)
result = my_sess.run(a)
print(result)
sess.close()
Output
[50]
Adding
import tensorflow as tf
my_graph = tf.Graph()
with my_graph.as_default():
a = tf.constant([12], name = 'my_const_a')
b = tf.constant([13], name = 'my_const_b')
c=tf.add(a,b)
sess = tf.Session(graph = my_graph)
result = sess.run(c)
print(result)
sess.close()
Output
[25]
We will get shape of different dimensions
import tensorflow as tf
my_graph = tf.Graph()
with my_graph.as_default():
my_scalar = tf.constant(5)
my_vector = tf.constant([1,2,3])
my_matrix = tf.constant([[1,2,3],[4,5,6],[7,8,9]])
my_tensor = tf.constant( [ [[1,2,3],[4,5,6],[7,8,9]] ,
[[4,5,6],[1,2,3],[7,8,9]] , [[7,8,9],[4,5,6],[1,2,3]] ] )
print(my_scalar.shape)
print(my_vector.shape)
print(my_matrix.shape)
print(my_tensor.shape)
Output
()
(3,)
(3, 3)
(3, 3, 3)
Now let us multiply two matrix by using matmul()
import tensorflow as tf
my_graph = tf.Graph()
with my_graph.as_default():
my_matrix1=tf.constant([[1,2,3],[4,5,6],[7,8,9]])
my_matrix2=tf.constant([[11,12,13],[14,15,16],[17,18,19]])
my_result=tf.matmul(my_matrix1,my_matrix2)
with tf.Session(graph = my_graph) as my_sess:
result=my_sess.run(my_result)
print(result)
Output
[[ 90 96 102]
[216 231 246]
[342 366 390]]
« Tensorflow Installation
← Subscribe to our YouTube Channel here