Trait relearn::envs::meta::MetaFeedback

source · [−]
pub trait MetaFeedback {
    type Inner: Clone + Send;
    type Outer: Clone + Send;

    fn neutral_outer() -> Self::Outer;
    fn into_inner_outer(self) -> (Self::Inner, Self::Outer);
}
Expand description

A feedback type that can be decomposed into an inner and outer space for a meta environment.

Required Associated Types

Required Methods

Neutral outer feedback that does not indicate good or bad behaviour.

Split the feedback into inner and outer environment feedback

Implementors

Reward feedback is always replicated in both the inner and outer environments.

This is the structure of RL-Squared meta reinforcement learning.