Hello, Exercise 5 doesn’t want us to specify a regularizer for the output layer. Is there a particular reason for that? Thanks!
Hi @chaohan, I think it’s a good learning opportunity for you to try adding regularizer to wherever you want to try. Section 7 demonstrated how to look for the best lambda value for the first 2 layers, keep those values, then you may run a test looking for the best lambda value for just the last layer.
Hi @rmwkwok , I have a follow-up question: it looks like we can keep a lambda value for a given layer and look for the best lambda for other layers. Is this the recommended way to identify the best lambda value(s) for a model?