The reason for the embedding net

[Mirorrn]

Hello,

I am not quite clear what the embedding net is good for. According to the code, it turns a 2-dimensional input vector (x, y) into a 16-dimensional vector (latent space).

Many thanks in advance for any insights.