problem when loading a keras model saved through a wrapper class

January 2, 2023

I have the following wrapper class:

from tensorflow.keras.models import Sequential
class NeuralNet(Sequential):
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

I can fit and save the model without problems, but when I try to load it:

from tensorflow.keras.models import load_model
model = load_model('model.h5')

I get:

--> 296     raise ValueError('Unknown ' + printable_module_name + ': ' + class_name)
    297 
    298   cls_config = config['config']

ValueError: Unknown layer: NeuralNet

I’d like to find a way to solve this error keeping the wrapper class.

>Solution :

you can also define a custom_objects dictionary as a class attribute in the NeuralNet class and pass it to the load_model function like this:

from tensorflow.keras.models import Sequential

class NeuralNet(Sequential):
    custom_objects = {"NeuralNet": NeuralNet}

    def __init__(self, **kwargs):
        super().__init__(**kwargs)

model = load_model('model.h5', custom_objects=NeuralNet.custom_objects)