I am implementing the example found here: https://pyro.ai/examples/dkl.html. I am having trouble saving and loading the model.
After I trained the
gpmodule, I used
torch.save(gpmodule.state_dict(), "model.p") to save the parameters for another session. However, when I enter another session, initialize a
gpmodule object, and use
gpmodule.load_state_dict() to load, the model performs as if it was randomly initialized.
Am I saving the model parameters incorrectly?