How to learn mlagents

Hello, I have already installed mlagents and trained my agents. I would like to use my own reinforcement learning algorithm to train multi-agent systems, but I have no idea at the moment. I would like to ask for your advice. thank you very much indeed