Deploy: How to deploy NLU and Policies on different GPUs?

python:3.9
rasa:3.1.0

After the rasa model is trained, the NLU model and the Policies model are in one tar.gz. How to deploy the NLU model and the Policies model on different GPUs? For example, the NLU model is deployed on GPU0, and the Policies model is deployed on GPU1, thanks!

You need to train the NLU and core models separately using rasa train.