Hacker News new | past | comments | ask | show | jobs | submit login

It is used as a predictive model of the environment for model-based RL. I.e. agents can predict consequences of their actions.





Oh, I see. I was somehow under the impression that the simulation was the game the RL agent learns to play (which kinda seemed nonsensical).



Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: