Getting non-tensor members from C++ code

Hey folks. Suppose I have the following model in python:

@torch.jit.script
class ModelProperties:
    def __init__( self, state: int, time: float ):
        self.state = state
        self.time = time

class MyModel(nn.Module):
    def __init__self, props: ModelProperties):
        self.props_ = props

Now, I can save it to torchscript using torch.jit.script, and load it as a torch::jit::script::Module. Is there away to access that self.props_ field from the C++ code? I see that there is API for getting “parameters”, but I don’t see a way to get or set the non-parameter members. Is that a thing that can be done, or do I have to use parameter objects for everything I might want to change?