Debugging your model
Complex deep learning models are prone to error. With millions of parameters, there are a number things that can go wrong. Luckily, the field has developed a number of useful tools to improve model performance. In this section, we will introduce the most useful tools that you can use to debug and improve your model.
Hyperparameter search with Hyperas
Manually tuning the hyperparameters of a neural network can be a tedious task. Despite you possibly having some intuition about what works and what does not, there are no hard rules to apply when it comes to tuning hyperparameters. This is why practitioners with lots of computing power on hand use automatic hyperparameter search. After all, hyperparameters form a search space just like the model's parameters do. The difference is that we cannot apply backpropagation to them and cannot take derivatives of them. We can still apply all non-gradient based optimization algorithms to them.
There are a number of different hyperparameter...
