The problem about the model.prune() (and other like this)

[savefishs]

   This is a issues that were found when using the functions which is model.prune() & model.prune_intput().

the issue is for GPU: if you finish training a model and want to use prune() or prune_input() ,prune_node() ，they create the new model and return ,but the functions did'nt move the model to the select device (so the model now is in cpu) . so the code will report :