Week 2 emojify_v2 softmax error

Course Q&A Deep Learning Specialization Sequence Models
1

I get the following error if I just pass the dense layer X to softmax:

NotImplementedError: Cannot convert a symbolic Tensor (dense_25/BiasAdd:0) to a numpy array. This error may indicate that you're trying to pass a Tensor to a NumPy call, which is not supported

I tried converting X to an ndarray first and then passing it to softmax but it still does not like the input. It seems be unhappy about the shape of the input (None, 5):

TypeError: expected sequence object with len >= 0 or a single integer

I assume I am missing some simple detail but haven’t been able to figure it out yet.

2

I presume you’re working on this:

image
image830×215 11.5 KB

First you compute z, then you use softmax(z).

3

Or, for the v2 model, you create an Activation() layer, then pass (X) to it. This uses the same style of syntax you use for the Dense layer.

4

The questions is for the def Emojify_V2(input_shape, ...): model and your answer helped. I didn’t realize tensorflow was to provide the function.