alt Hacker NewsIs the actor trained on world model's mistakes as well? If so, it should learn to force hallucinations that force it into an easier state.
Is the actor trained on world model's mistakes as well? If so, it should learn to force hallucinations that force it into an easier state.
Yes, there were some hallucinations, e.g: in breakout hitting a corner would destroy more blocks than expected, in doom-lite it could shoot from miles away and get a kill. The fixed was add some noise to the input frame and keep the target clean.