Concurrent GPU and NPU inference on IMX 8M Plus

Hello, I would like to run 1 neural network on the VIP8000 Nano NPU and 1 network on the G7000UL GPU. Currently I am able to choose which accelerator to target through the USE_GPU_INFERENCE environment variable. Is it possible to override this environment variable and use both accelerators in an application running in the same environment? Thanks :)