MBZUAI releases SR²AM, an agentic LLM framework that uses a learned configurator to regulate future-state simulations

The framework uses reinforcement learning to optimize planning depth