I had not been able to locate examples for how to create a neural network whose weights/bias are drawn from a probability distribution, using only the main pytorch library, without reliance on additional libraries such as
pyro. Thanks to the help from this community, @krypticmouse in particular, I was able to cobble together a working example which I’ve posted on Medium.
I hope others might find it useful, and welcome suggestions on how to improve what’s been published as I’m still learning best practices on this subject matter.
Here is a link to the article: Probabilistic Neural Network with Pytorch