Heya! I'm taking a machine learning module and we're currently using a simple neural network with neurons and layers to work towards plotting three shapes: a circle, a circle with a hole in it (a donut) and a 'scream' (a circle with two holes for 'eyes' and one for its 'mouth'. I have some code that I used to visualise this neural network but currently its not visualising the circle very well - there are a lot of points and its not very round. Furthermore, the other two shapes are not developing any holes in them and are just coming out as filled in circles. I'm pretty sure I may need to tweak things such as the number of layers, nodes, steps and learning rate but I'm not sure. This is what I have:

the other two shapes provide similar results.