# Reinforcement Learning (Outline) * Best examples of RL so far * alpha zero * mu zero * Put in a reminder of the quotes from the Deep mind team * What are the pros of RL? What does it provide us with that DL/LLMs on their own will not? * What are the constraints that may not often be talked about? * [Requires a game engine](https://twitter.com/pfau/status/1680152841321627648?s=46) ### To include * Alpha go * Is it because the space of the board is already defined? What about if you don't even know what the board is...? (if you are trying to discover natural laws for instance) * [Requires a game engine](https://twitter.com/pfau/status/1680152841321627648?s=46) * Same goes for breakout. This game has a clear loss function, clear rules. What if you had to learn the rules and the loss function…? Again, use feynmans quote of chess board * --- Date: 20230723 Links to: Tags: References: * []()