Distributional Reward Calibration and Budget-Neutral Credit Assignment for Multimodal RLVR
We revisit advantage estimation in Group Relative Policy Optimization (GRPO) and observe that any token-level advantage admits a two-stage factorization: a \emph{sample-level baselining} step that converts scalar verifier rewards into a comparable sequence-level signal, and a \emph{token-level credit assignment} step t...
Lang Chen, Zong-Xin Xu, Bo-Rui Wang et al.
· 0 citations