ConceptAI & ML
Distribution sharpening in RLHF
The process by which reinforcement learning with human feedback makes model output distributions more peaked around favored modes, often leading to overconfidence.
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.