TensorFlow visualization tool TensorBoard default and custom graphs

I. Chart

Graph: Data (Tensor Tenrsor) + Operation (Node Operation) (Static)

The diagram can be used: 1. default diagram; 2. customized diagram.

1. Default Chart

A way to view the default diagram:

1. Call method: tf.get_default_graph()
2. View properties: .graph

1, call the method to view the default map properties

# Method I: Invocation method
    default = tf.get_default_graph()
    print('default:', default)

2, .graph view graph properties

# Method 2: View Properties
    # View node properties
    print('Attributes of a:', )
    print('Attributes of c:', )
    # View session properties
    print('Graph properties for session sess:', )

The reason you can notice that these graphs all have the same address is because they all use the default graph by default.

coding

# View Default Chart
def View_Graph():
    # Method I: Invocation method
    default = tf.get_default_graph()
    print('default:', default)   
    # Method 2: View Properties
    # View node properties
    print('Attributes of a:', )
    print('Properties of c:', )
    # View session properties
    print('Graph properties for session sess:', )

2. Customized diagrams (creation of diagrams)

1、Create customized diagram

# 1 Create a customized diagram
    new_graph = ()
    print(new_graph)

2、Create static diagram

 # 2 Create static graphs (tensor and nodes)
    with new_graph.as_default():
        a = (10)
        b = (20)
        c = a + b
        print(c)

3. Open session (run)

# 3 Open dialog (running)
    with (graph=new_graph) as sess:
        print('c=', (c))

4、View customized diagram

# 4 Viewing Customized Diagrams
    View_Graph(a, b, c, sess)

# View Figure
def View_Graph(a, b, c, sess):
    # Method 1: Invocation method
    default = tf.get_default_graph()
    print('default:', default)
 
    # Method 2: View Properties
    # View node properties
    print('Attributes of a:', )
    print('Properties of c:', )
    # View session properties
    print('Graph properties for session sess:', )

coding

# Customization charts
def Create_myGraph():
    # 1 Create a customized diagram
    new_graph = ()
    print(new_graph)
    
    # 2 Create static graphs (tensor and nodes)
    with new_graph.as_default():
        a = (10)
        b = (20)
        c = a + b
        print(c)
    
    # 3 Open dialog (running)
    with (graph=new_graph) as sess:
        print('c=', (c))
 
    # 4 Viewing Customized Diagrams
    View_Graph(a, b, c, sess)

II. TensorBoard Visualization

1、Visualization processing

 (path, graph=)

# Visualization
        ("C:\\Users\\Administrator\\Desktop\\summary", graph=)            #path                                            seek

2. Open TensorBoard

Operate in cmd:

1. Move to the front of the folder first

cd C://Users//Administrator//Desktop

2. Open TensorBoard (get data from file)

tensorboard --logdir=summary

3. Open the given URL

http://localhost:6006/ (URL given in cmd)

Visualization results are obtained:

master code

import tensorflow as tf
# Create the TensorFlow framework
def Create_Tensorflow():
    # Chart (static)
    a = (2)  # Data 1 (tensor)
    b = (6)  # Data 2 (tensor)
    c = a + b  # Operations (nodes)
    # Sessions (executive)
    with () as sess:
        print('c=', (c))
        # Visualization
        ("C:\\Users\\Administrator\\Desktop\\summary", graph=) 
    # View Default Chart
    View_Graph(a, b, c, sess) 
# View Figure
def View_Graph(a, b, c, sess):
    # Method I: Invocation method
    default = tf.get_default_graph()
    print('default:', default) 
    # Method 2: View Properties
    # View node properties
    print('Attributes of a:', )
    print('Attributes of c:', )
    # View session properties
    print('Graph properties for session sess:', ) 
# Customized charts
def Create_myGraph():
    # 1 Create a customized diagram
    new_graph = ()
    print(new_graph)    
    # 2 Create static graphs (tensor and nodes)
    with new_graph.as_default():
        a = (10)
        b = (20)
        c = a + b
        print(c)   
    # 3 Open dialog (running)
    with (graph=new_graph) as sess:
        print('c=', (c)) 
    # 4 Viewing Customized Diagrams
    View_Graph(a, b, c, sess) 
if __name__ == '__main__':
    # Create the TensorFlow framework
    Create_Tensorflow() 
    # Create custom diagrams
    Create_myGraph()

Above is the TensorFlow visualization tool TensorBoard default graph and custom graph of the details, more information about TensorFlow visualization TensorBoard tool please pay attention to my other related articles!