Abstract We explore building generative neural network models of popular reinforcement learning environments. Our world model can be trained quickly in an unsupervised manner to learn a compressed spatial and temporal representation of the environment. By using features extracted from the world model as inputs to an agent, we can train a very compact and simple policy that can solve the required t
![World Models](https://cdn-ak-scissors.b.st-hatena.com/image/square/ef0c1e7d5f2d03c373f94a69b0fae2f0f2bd2b47/height=288;version=1;width=512/https%3A%2F%2Fworldmodels.github.io%2Fassets%2Fworld_models_card_both.png)