Initialize scaling parameter to prevent NaNs in scratch training

[Astro36]

Description

The self.scaling parameter is created using torch.empty and is then used without being initialized.
This is not a problem during finetuning, because pretrained checkpoints provide valid scaling parameters.
But when training from scratch, these undefined values may occasionally cause numerical instability and lead to NaN outputs.
Initializing self.scaling with torch.randn ensures proper initial values and improves training stability.

List of changes

• Initialize self.scaling with torch.randn.

For reviewers

• No functionality change, just fixed parameter initialization.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.