Setting model parameters in optimizer

Is there a way of specifying the model’s parameters that one intends to optimize within pyro.optim optimizers, just like in PyTorch? This is especially important if some layers of a guide network are supposed to be frozen… I’ve tried many ways of passing these parameters but different errors have been thrown up. Some help would be really appreciated!

if i understand what you want you can use the logic in per_param_args in this tutorial