SW-9298: Reduce maximum number of neurons per layer in examples from 100 to 80 for go and lab support.

[maxwell-ashurst]

Reduces maximum number of neurons per layer from 100 to 80 for all examples. Lab and go can only support up to 80 neurons per layer, so this allows the examples to work on these devices. We could probably just add a note instead to let customers know they can do this on lab/go instead of changing the existing examples. No significant drop in performance when reducing the layer sizes, so no strong preference for which we choose to do from me.

[paulaltin]

Lab and go can only support up to 80 neurons per layer

Note that is only true for the 2-slot MIM. The 3-slot layout, which we expect to release at the same time as NN on Lab/Go, will probably only support 50 or 60 neurons per layer. This might affect the decision on whether to change the examples and/or add a note.

[MICHELA-BENEDETTI]

Good to go from my end!

Just a couple of minor things:

• if possible can we have all the cells' number to be sequential to show we run the cell in the order to follow the notebook? For example in the classification notebook is goes from 1,15, 18, 21 and back to 5.
• Can we set the random seed in the signal_id notebook?