← 研踪 PaperTrailarXiv:2611.24005 · v1
How Reward Gain Imbalance Shapes Ranking: Rethinking Rubric-Based Reward Aggregation
首次提交 2026-11-18 15:56
版本更新 2026-11-18 15:56
雷达观测 2026-11-19 19:07
摘要
Rubric-based reinforcement learning constructs rewards for aligning large language models with human preferences, typically by equally aggregating continuous scores that a judge assigns to individual criteria. However, scoring heterogeneity across criteria can distort candidate rankings. Existing work has focused on improving judges, learning criterion weights, or refining aggregation, whereas the interaction between scoring heterogeneity and ranking remains insufficiently understood. We formalize this heterogeneity as \emph{reward gain imbalance}: each criterion enters the aggregated reward with its own reward gain, the gap between its expected scores on satisfied and unsatisfied responses. Counterintuitively, our experiments show that removing this imbalance, even with ground-truth labels, lowers ranking accuracy. To explain this, we derive an exact decomposition of the reward difference between any two candidates into a base satisfaction signal, a reward gain imbalance term, and a residual term, showing that reward gain imbalance can attenuate or amplify the base signal, and finding that misrankings usually concentrate on pairs where it attenuates. Guided by this analysis, we propose the adaptive \textbf{D}ual-\textbf{C}enter \textbf{C}lipping aggregation (DCC), which bounds each criterion's contribution to the reward difference instead of simply removing the imbalance. Using only the scores of the current candidate group, it determines a clipping interval that imposes a common upper bound on per-criterion score differences between candidates, limiting the excessive influence of extreme scores while preserving continuous judgment information....
观测到的版本
相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。
v1 · 2026-11-18 15:56 · 最新观测版本
How Reward Gain Imbalance Shapes Ranking: Rethinking Rubric-Based Reward Aggregation
Rubric-based reinforcement learning constructs rewards for aligning large language models with human preferences, typically by equally aggregating continuous scores that a judge assigns to individual criteria. However, scoring heterogeneity across criteria can distort candidate rankings. Existing work has focused on improving judges, learning criterion weights, or refining aggregation, whereas the interaction between scoring heterogeneity and ranking remains insufficiently understood. We formalize this heterogeneity as \emph{reward gain imbalance}: each criterion enters the aggregated reward with its own reward gain, the gap between its expected scores on satisfied and unsatisfied responses. Counterintuitively, our experiments show that removing this imbalance, even with ground-truth labels, lowers ranking accuracy. To explain this, we derive an exact decomposition of the reward difference between any two candidates into a base satisfaction signal, a reward gain imbalance term, and a residual term, showing that reward gain imbalance can attenuate or amplify the base signal, and finding that misrankings usually concentrate on pairs where it attenuates. Guided by this analysis, we propose the adaptive \textbf{D}ual-\textbf{C}enter \textbf{C}lipping aggregation (DCC), which bounds each criterion's contribution to the reward difference instead of simply removing the imbalance. Using only the scores of the current candidate group, it determines a clipping interval that imposes a common upper bound on per-criterion score differences between candidates, limiting the excessive influence of extreme scores while preserving continuous judgment information....
查看此版本解读 →查看此版本原文 ↗