Is there a convenient method of making sure all tensors have been moved to the correct device on a CPU only machine?

When debugging my code on a CPU only machine, is there a convenient way to check all required tensors have been moved to the correct device prior to using the code to train the model on a machine with GPU access? I was wondering if there was maybe some kind of ‘dummy’ CPU device I could move the tensors to that would throw a similar error.

For context I am developing a model on my local machine which will be trained on a HPC cluster I have access to. However, while possible, its inconvenient to debug the code on this cluster.