Returned xlora_scalings should be copied to the same device as the model

[data-hound]

https://github.com/EricLBuehler/xlora/blob/8217992c8dcf5a1c025749a48e665c59cc0b6350/src/xlora/xlora_insertion.py#L65C16-L65C30

The xlora_scalings generated as:

xlora_scalings: Tensor = self.model.internal_xlora_scalings[:, :, self.layer_number, :]

should be put on the same device as where the model exists to enable return x * scalings in https://github.com/EricLBuehler/xlora/blob/8217992c8dcf5a1c025749a48e665c59cc0b6350/src/xlora/xlora_insertion.py#L40C9-L40C28

[XDUWen]

same error,have you solved it?