A day with .Net

My day to day experince in .net

Visualise Computational Graphs with Tensorboard and Tensorflow

Posted by vivekcek on August 6, 2017

In this post i will show , how you can visualise computational graphs created by tensorflow bu using tensorboard.

I am using visual studio for tensorflow development in windows. Refer my previous blog to setup tensor flow in windows Installing tensorflow in windows

Hope the reader has some basic understanding of tensorflow. Tensorflow execute every operation by building computational graphs. To visualize such graph we use tensorboard.

Tensorboard is a visualisation tool that will be installed as a part of tensorflow installation.

Write below code in visual studio. In this code we declare four constants, then we multiply ‘a’ and b’ after that we divide ‘c’ and ‘d’ then result of both were added to produce the final output.

import tensorflow as tf

a=tf.constant(6,name="a")
b=tf.constant(3,name="b")
c=tf.constant(10,name="c")
d=tf.constant(5,name="d")

mul=tf.multiply(a,b,name="mul")
div=tf.div(c,d,name="div")

addn=tf.add_n([mul,div],name="addn")

sess=tf.Session()
output =sess.run(addn)
print(output)
writer=tf.summary.FileWriter('./visual',sess.graph)
writer.close()
sess.close()

These line actually help us to create graphs for above computation. Here “visual” is a folder created in your running directory.

writer=tf.summary.FileWriter('./visual',sess.graph)
writer.close()

Now go to your running directory and execute below command.

tensorboard --logdir="visual"

Now browse to the link and see the graph.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s