PatchWorld: Gradient-Free Optimization of Executable World Models

ArXi:2605.30880v1 Announce Type: cross Text-agent environments are typically modeled as partially observable Marko decision processes (POMDPs), assuming that the simulator's latent state and transition dynamics are hidden from the agent. Yet little work has examined whether executable code can be induced to serve as a world model for prediction and planning under partial observability. We