Yes, I do - attached is a grab of the base task. I also ran trains.get_task(*params).id to make sure the task id's matched, in case a did something stupid. They did some I'm guessing it has something to do with the way I am implementing the optimization itself, but all my code is boilerplate for my toy example.