Skip to content
Benchmaker

Glossary · Use and training pipelines

Training reward (use)

3Regime

A use in which the evaluator's signal enters parameter updates, as in RLHF or RLVR. Of the four uses of a judge (reporting, selection, search, training), it applies the most optimization pressure.

Also called: training use

Defined in

Handbook · RegimeUse regimes: same judge, four jobsMeasure, optimize or monitor: an evaluator changes meaning when its signal enters a different loop, and signals migrate between loops.
Use regime
The causal role a signal plays: measure (supports inference), optimize (shapes the candidate or policy distribution), or monitor-control (detects and intervenes in deployment).
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.
RLVR
Reinforcement learning with automatically verifiable reward signals, usually exact, executable, or formal. Grounded but incomplete.

See it in the full glossary →