Hi, your repository is really awesome! and I would like to participate by adding pre-train models.
I already trained a PPO model on BattleZone using your tools and I think I got a nice version. It could be improve because the AI have hard time to aim harder enemies that are coming later during the game. I have a good 33 sec video that show well the good aiming and search of blue enemies.
I have the model and video ready, but I saw that you prefer to open issue to discuss before sending an PR.
What would you like to check with me before I build a PR?
Should I create a PR before or after submitting to Hugging Face?