Concept
Generative reward model
A generative reward model is a reward model that judges an output by generating text (a critique, a verdict, or a step-by-step assessment) rather than emitting a single scalar score. Think of it like a classic reward model in RLHF, except instead of mapping a response to…
The rest of “Generative reward model” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→