I don’t think this post really needs a body, but here we are. Thanks for any help.
hi, I think everything is possible, I’ve loaded a trained model into my scene, and removed the endepisode call from my script. so basically the agent now only has one chance to achieve the goal, if it fails, there is no way to reset the scene because end episode is removed. but if this is what you are after, then fine.
yeah i was thinking about that i was hoping to do it without pulling out code. I was thinking about having more control what happens when the episode should end but I can probably just overwrite the method.