Does PLM support multi-GPU inference?

Hello,

Thank you for your great work.

I'm currently using apps/plm/generate.py for PLM model inference and noticed that it appears to only support single-GPU operation. For larger models, having multi-GPU support would be highly beneficial.

Could you please advise if:

There are existing multi-GPU inference capabilities that I might have missed in the implementation?

If not, are there plans to add this functionality in future releases?

Are there recommended workarounds or modifications to enable multi-GPU inference with the current codebase?