Python : Plotting Architecture of Tensorflow Keras model

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

Plotting an architecture of Tensorflow Keras model by plot_model()

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.

1 2

pip install graphviz pip install pydot

cs

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

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

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')   Colored by Color Scripter

cs

As expected, we can get the following visualization.

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

1

conda install graphviz

cs