Hi I'm currently trying to integrate imitation learning in a preexisting project using ml-agents 0.13. It's a space shooter with a time travel component in a single level. The player's and enemies actions are recorded during the game and when the player looses the level reboots. During that reboot a bot is added that replays the actions the player took previously and the enemies replay the same actions they did previously, if the enemies run out of actions an AI takes over and if the player can save the bot an other AI takes over. I'd like to replace the AI that takes over the bot by an online imitation learning AI but I'm running into a few problems: 1. Online learning has been depreciated, I've been trying to get around it using GAIL but I'm not sure if it could work. 2. Gail training is accelerated which I do not want for an attempted online training, can i get around that ? Thank you for time, I hope you have a good day.