Hey, I'm brand new to ML-Agents and only slightly experienced with AI. My master thesis is on solving a social card game with reinforcement learning and I was thinking about creating the environment on Unity and leveraging ML-Agents for this task. My question is, is this feasible? Most of the examples I've seen take in visual observations and the decisions are physical movements of an agent, but in my case it would just be a 2D top down view of a board and an agent making a decision when it comes to its turn based on the environment and actions of the other players up until that point (the opponents will be heuristically hard coded with how they play, with self play being implemented once the agent becomes good enough). If it is feasible and something similar has been done before in a tutorial (or the closest thing that I can adapt for my use case), I would really appreciate a direction to it being implemented. Thanks in advance unity gang!