Week 1, Assignment 1, Ex3, No 'dense layer' implemented?

Hi, in Figure 4 showing a single LTSM cell, it is mentioned that ‘softmax’ includes a dense layer.

However, in Ex3, only a softmax function is implemented, as shown below in utils.py:

I wonder what this dense layer is? Below is my interpretation: for a given time step , the predicted output is of dimensions (n_y, m), where softmax gives n_y probabilities for each of the m training examples. Not sure what a dense layer here would serve?

Thanks for any clarification.

The dense layer serves to learn the Wy weight matrix and by, so that the predictions are useful.
Your markup seems correct.