#artificial intelligence
May 2025
R3: Robust Rubric-Agnostic Reward Models
R3 is introduced, a novel reward modeling framework that is rubric-agnostic, generalizable across evaluation dimensions, and provides interpretable, reasoned score assignments, enabling more transparent and flexible evaluation of language models.
David Anugraha, Zi-Lu Tang, Lester James Validad Miranda et al.
· Trans. Mach. Learn. Res. · 23 citations