Disregarding model.compile()

[franciscogmm]

Posted this on https://threads-iiith.quora.com/Neuro-Evolution-with-Flappy-Bird-Genetic-Evolution-on-Neural-Networks:

Thanks for this!! I thought I’d get stuck with Pybrain because I never figured out how to do the same with Keras. I didn’t know that you could disregard the optimizer and cost function even if it is implemented.

Question re the quote below:

“Often you would also be required to define a loss function and an optimizer to train the network via back-propagation (but here we don’t need to perform back-propagation, so we can just ignore this part).”

Why did you still define the sgd and mode.comile()?

Is it because you’re not calling .fit() before the .predict()? Thanks!

[AroMorin]

I think, and I can be completely wrong here, that Keras doesn't actually go ahead and make the network UNLESS you compile it.