Why TorchScript module does not take less GPU memory than Pytorch model?

I didn’t know about it.
Most of my attempts to use torchscript to speed up the forward pass showed no improvement at all.

Are there other kind of situations besides pointwise ops?