ConceptAI & ML
Undoing Implicit regularization
Adding an explicit loss term to the training algorithm that mirrors the bad second term of the offline RL regularizer to restore scaling benefits.
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
3The lectures that define it or assume it, and the concepts next to it.