Hi, How would you go about training 2 sequential actions. For example, if you had to train a bot to get a key and open the door. Going to find the key is one task and then finding the door is the second. Would I just ignore rewards on the door if they don't have the key yet? Thoughts? Thanks!