An activation layer acts on every output from a layer, then that output connects to the next layer in whatever way it would've absent the activation layer. So it might be good to have the activation layer shown as the full width of the preceding layer, and it might be good to show connection lines from the activation layer is if it were the previous layer (this only really matters, I think, for dense layers.)
An activation layer acts on every output from a layer, then that output connects to the next layer in whatever way it would've absent the activation layer. So it might be good to have the activation layer shown as the full width of the preceding layer, and it might be good to show connection lines from the activation layer is if it were the previous layer (this only really matters, I think, for dense layers.)