Question

Graph visualization using Tesnorboard !

Answer 1

Tensorboard 

• TensorBoard’s Graphs dashboard is a powerful tool for examining your TensorFlow model.
• It visualize the training parameters, metrics or any statics of neural network.
• You can quickly view a conceptual graph of your model’s structure and ensure it matches your intended design.

Views in Tensorboard 

• Scalar Dashboard

• Histogram Dashboard

• Distribution Dashboard

• Image Dashboard

• Audio Dashboard

• Graph Explorer

• Embedding Projector

• Text Dashboard

Let's take a look of some syntax of tensorboard

#To see version of tensor flow 

import tensorflow as tf

print("TensorFlow version: ", tf.__version__)

Output

TensorFlow version: 1.15.0

Simple Example to see Tensorboard

#multiplication of two numbers 

import tensorflow as tf

a = tf.constant(4.0)

b = tf.constant(6.0)

c = a * b

# eveluate the tensor

sess = tf.Session()

print (sess.run(c))

#A file that write data from Tensorboard 

writer=tf.summary.FileWriter("./logs",sess.graph)

sess.run(c)

writer.close()

sess.close()

 Output

24.0

# Load the TensorBoard notebook extension.
%load_ext tensorboard

Note:-This will load/Launch  the Tensorboard

#Start TensorBoard and wait a few seconds for the UI to load. 

%tensorboard --logdir logs 

Output

Note:- By default, TensorBoard displays the op-level graph.

Note:-graph is inverted; data flows from bottom to top,compared to the code.

# Clear any logs from previous runs

!rm -rf ./logs/  

Keras model using Tensorboard

Simple sequential model

# Load the TensorBoard notebook extension.

%load_ext tensorboard