Support for target and draft models on different GPUs

Hi there, Thanks for your work on SpecForge. Does it support running the target model and draft model on different GPUs (e.g., GPU 0 and GPU 1) or across different machines? Best regards