Tuning summary does not agree with given search space

[rafaelstevenson]

Hello, I am currently using keras tuner, and it seems there is a problem where i used a for loop to define the number of hidden layers as a search space, but the model results summary does not agree with the given 'num_layers' as when 'num_layers'==3 for example, why there are still hyper parameters for 6 hidden layers as i show in the screenshot.

If anybody could help or know the solution would be truly appreciated, thank you

[haifeng-jin]

The tuner would tune all the hps that it saw in the previous trials.
Unless the hp is considered not active, which is decided by the conditional_scope.
https://keras.io/api/keras_tuner/hyperparameters/#conditional_scope-method