Home > Chapter 10: On-policy Control with Approximation

rlai.gpi.state_action_value.function_approximation.FunctionApproximationPolicy

Policy for use with function approximation methods. This is effectively an interface to the underlying function
    approximation estimator and its reward model, which are accessed by indexing the policy with a state (e.g., a call
    like `agent.pi[state]`), which returns an action-probability dictionary.