Skip to content
Benchmaker

Glossary · Use and training pipelines

KL-regularized objective

The RLHF-style training objective: maximize expected reward minus β times the KL divergence from a reference policy. The KL term limits policy drift and is the natural exposure axis for this optimizer.

Also called: KL penalty

Defined in

Handbook · RegimeEvaluators inside training loopsRLHF is a pipeline, not an evaluator. Where the learned judge sits in RLHF, RLAIF, RLVR and DPO, and what it is exposed to.
RLHF
Reinforcement learning from human feedback: a training pipeline that consumes preference data, usually via a learned reward model and policy optimization. Not itself an evaluator family.
Exposure budget
How much optimization pressure an evaluator absorbs, recorded in the optimizer's own unit (candidates, KL, developer iterations, queries, self-modification events), with the induced shift D(Q_b ‖ P_0) where measurable. There is no universal unit.

Where it comes up

See it in the full glossary →