Artificial Intelligence By Example

Saving the model

Saving the model, in this case, consists in saving three files.

First, we will discuss the model file.

model.json, saved in the following code, contains serialized data describing the model itself without weights. It contains the parameters and options of each layer. This information is very useful to fine-tune the model:

# serialize model to JSONmodel_json = classifier.to_json()with open("model.json", "w") as json_file:json_file.write(model_json)

More details on this file are provided in the next section, in the Loading the model section.

• Now we will discuss the weights file.

To understand a CNN model or any other type of model, having access to the weights can provide useful information when nothing works or to solve ...