StructReward: Efficient Structured Process Rewards for Self-Correcting Multimodal Reasoning
Yifan Li, Ruxin Sun, Tongzhou Zhao
cs.AI
Submitted: 2026-08-11
Updated: 2026-08-12
Comments: 9 pages, 3 figures. Yifan Li and Ruxin Sun contributed equally
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Qwen3-VL Technical Report
- Constitutional AI: Harmlessness from AI Feedback
- Unveiling Chain of Step Reasoning for Vision-Language Models with Fine-grained Rewards
- Advancing Multimodal Reasoning: From Optimized Cold Start to Staged Reinforcement Learning
- VRPRM: Process Reward Modeling via Visual Reasoning
- RM-R1: Reward Modeling as Reasoning
- Training Verifiers to Solve Math Word Problems
- Process Reinforcement through Implicit Rewards
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles
- Sherlock: Self-Correcting Reasoning in Vision-Language Models
- MM-PRM: Enhancing Multimodal Mathematical Reasoning with Scalable Step-Level Supervision
- Vision-R1: Incentivizing Reasoning Capability in Multimodal Large Language Models
- PuzzleCraft: Exploration-Aware Curriculum Learning for Puzzle-Based RLVR in VLMs
- Process Reward Models That Think
- VL-RewardBench: A Challenging Benchmark for Vision-Language Generative Reward Models
- UniM: A Unified Any-to-Any Interleaved Multimodal Benchmark
- Let's Verify Step by Step
- NoisyRollout: Reinforcing Visual Reasoning with Data Augmentation
- SPARK: Synergistic Policy And Reward Co-Evolving Framework
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems