5 papers
Grounding the Score: Explicit Visual Premise Verification for Reliable Vision-Language Process Reward Models
Junxin Wang, Dai Guan, Weijie Qiu +7
Vision-language process reward models (VL-PRMs) are increasingly used to score intermediate reasoning steps and rerank candidates under test-time scaling. However, they often funct…
Rationale Matters: Learning Transferable Rubrics via Proxy-Guided Critique for VLM Reward Models
Weijie Qiu, Dai Guan, Junxin Wang +6
Generative reward models (GRMs) for vision-language models (VLMs) often evaluate outputs via a three-stage pipeline: rubric generation, criterion-based scoring, and a final verdict…
CLIPO: Contrastive Learning in Policy Optimization Generalizes RLVR
Sijia Cui, Pengyu Cheng, Jiajun Song +6
Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced the reasoning capacity of Large Language Models (LLMs). However, RLVR solely relies on final answer…
Open Rubric System: Scaling Reinforcement Learning with Pairwise Adaptive Rubric
Ruipeng Jia, Yunyi Yang, Wen Wang +7
Scalar reward models compress multi-dimensional human preferences into a single opaque score, creating an information bottleneck that often leads to brittleness and reward hacking…
Writing-Zero: Bridge the Gap Between Non-verifiable Tasks and Verifiable Rewards
Ruipeng Jia, Yunyi Yang, Yongbo Gai +5
Reinforcement learning with verifiable rewards (RLVR) has enabled large language models (LLMs) to achieve remarkable breakthroughs in reasoning tasks with objective ground-truth an…