tensorflow 2 models at the same time (each gets 2 GPUs), is that possible?

Question

How can I train 2 LSTM models at the same time, providing each of them 2 own GPUs (when I have 4 in a workstation)?

I currently only know how to train one. Is there a