●builderYou can leverage this architecture to build agents capable of both predicting game states and executing complex native controls.
●researcherThis provides a framework for studying the intersection of generative world models and task-driven reinforcement learning.