Model-based RL Algorithms

[lmc19970711]

Can you support more model-based algorithms, such as Dreamer, World model, etc.?

[PaParaZz1]

Yes, we are working on enriching our model-based RL algorithm zoo, we also opened source a new repo about a curated list of resources of mbrl several days ago.

We are preparing steve and mcts-based method such as MuZero, and we also welcome more collaborators, can you offer more detailed thoughts, such as:

• paper link and related code
• evaluation environment and target performance (e.g.: reward and running time)
• how can you participate in this work
• other funny tricks and detailed comments