Skip to content
Benchmaker

Glossary · Use and training pipelines

RLAIF

RL-style alignment using AI-generated feedback instead of, or alongside, human feedback.

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.

See it in the full glossary →