# How can I do hyperparameter optimization for a CNN-LSTM neural network?

I have built a CNN-LSTM neural network with 2 inputs and 2 outputs in Keras. I trained the network with model.fit_generator() (and not model.fit()), to load just parts of the training data when needed, because the training data is too large to load at once.

After the training the model was not working. So I checked training data (before and after augmentation). The training data are correct. So I thought the reason why the model does not work must be that I have not found the optimal hyperparameters yet.

But how can I do hyperparameter optimization on a network with multiple inputs and outputs and trained with model.fit_generator()? All I can find online is hyperparameter optimization of networks with a single input and single output and trained with model.fit().

• The model not working does not mean that the hyperparameters is the problem. More often than not even with wrong hyperparameters a model should work but just not work that well. What data are you using? Maybe you can try using a more simple LSTM model first before trying a CNN LSTM model. – Clement Hui Jan 9 '20 at 15:08