ConceptAI & ML
Overoptimization
Exploiting the reward model after a while such that the policy initially gets better, but later on gets worse due to distribution shift.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Lecturers
1Connections
4The lectures that define it or assume it, and the concepts next to it.