Plotting an Architecture of Tensorflow Keras Model by plot_model()

This post shows how to plot an architecture flow of Tensorflow Keras model by using plot_model() function.

To plot the architecture flow of Tensorflow Keras model by using plot_model() function, two packages should be installed: graphviz and pydot.

To install these packages, type the following commands in the anaconda prompt.

pip install graphviz
pip install pydot

In the Jupyter notebook, we can use plot_model() function to draw an architecture of the following Keras model.

from keras import layers, models
from keras.utils.vis_utils import plot_model
 
#1. Modeling
model = models.Sequential()
 
model.add(layers.Dense(128, activation='relu', 
                       input_shape=(256,),       name='HiddenLayer01'))
model.add(layers.Dense(64 , activation='relu',   name='HiddenLayer02'))
model.add(layers.Dense(32 , activation='relu',   name='HiddenLayer03'))
model.add(layers.Dense(1  , activation='softmax',name='OuputLayer'))
 
model.compile(loss='categorical_crossentropy', 
              optimizer = 'adam', metrics=['accuracy'])
 
#2. Visualizing
plot_model(model, show_shapes=True, to_file='model.png')

As expected, we can get the following visualization.

If you happen to encounter an error regarding the installation, one possible solution is to use the following conda command in the anaconda prompt.

conda install graphviz

Visit SH Fintech Modeling for additional insights on this topic: https://kiandlee.blogspot.com/2022/12/python-plotting-architecture-of.html.