Pymc pytorch ops assistance

Hi, I am a nuclear engineering graduate student, and I am attempting to use pymc for some of my research. I am particularly interested in using pymc’s variational inference capabilites to infer parameters of my pytorch models. Could someone please help me understand the current state of integrating pytorch models with pymc? I have tried searching on my own, but am still very new to all of the associated lingo when trying to use the two together.