Skip to content

Topic

Model-based guardrails

The practice of using one language model to evaluate or gate another model's outputs and actions, in place of a human reviewer or a fixed rule set.

Current clusters