The term world model is used quite often recently, but it s still not very well defined. Depending on who is speaking, it can mean an action-conditioned video, generator, an explicit geometry/physics simulator, a learned dynamics model, used for planning, a representation predictor such as JEPA, or structure, discovered inside a network trained for something else.This is my attempt to understand it. It s free and open source - so if you have any suggestions / improvements to make, please let me know / contribute a pr:https://github.com/NiluK/worldmodels101