Models, reasoning & training · Concept

RLVR

Also called: reinforcement learning with verifiable rewards

Reinforcement learning with verifiable rewards, such as outcomes checked by tests or objective graders.