C3_W1_Lab_1 keras tuner demo

Hello everyone,

When re running the notebook lab I get different results each time. And sometimes, the number of units found is higher than the one used in the baseline model.

First question: can we provide a seed for keras to reproduce results?
Second question: anyone got the same problem?


There is a seed parameter in kt.Hyperband constructor. Set this to get the same output configuration as the result.

Don’t forget to upgrade keras-tuner to 1.1.3 using pip install -U keras-tuner