model-based

Definition

An approach that keeps an explicit model of how the environment responds and uses it to plan or to generate imagined experience. In reinforcement learning it contrasts with model-free methods, which learn a policy directly from trial and error: model-based agents are typically far more sample-efficient because rollouts inside the model are cheap, but they inherit whatever the model gets wrong.

Explainers using this term

• Paper Deep Dive — Large Discovery Models: giving an LLM a value signal for what to try nextLarge Discovery Models: Empirically-grounded Model-Based Open-Ended Search