Extracting component means and convariances from mixture model

Question

I am currently trying to write a simple multivariate gaussian mixture model using tensorflow probability. Specifically, I have some 2-dimensional input and 2-dimensional output