I have spent couple of hours on weekend to understand the concepts of tensorflow and would like to contribute in simplest form as most of the articles I read were too complicated and took me a while to understand.

A tensor consists of a set of primitive values shaped into an array of any number of dimensions. To make it more simple, consider it as an array of data.


Tensorflow is the flow of data(tensor) between different nodes(operation) of graph.

Installation :

pre-requisite : Python 3.5

pip install tensorflow

Building Blocks

Constant : You can store your data and cannot be altered(as the word says)

constant1 = tf.constant([2])
constant2 = tf.constant([2,4])

Variable : data can be altered

var_data = tf.Variable(5)  // var_data = 5
new_value = tf.add(var_data,var_data) // new_value = 10
update = tf.assign(var_data,new_value) // var_data =10

You need to initialize all variables using below function before running it using session(will discuss later).


Placeholder : Same as variable but data would be assigned later in session. Consider scenario, you have a formulae by no data yet.

x = tf.placeholder(tf.float32)
y = x*x

with tf.Session() as session:
    result = session.run(y,{x:10})
    print(result) // 100