Step-wise Rubric Rewards for LLM Reasoning
cs.LG
Submitted: 2026-05-17
Updated: 2026-10-30
Comments: NeurIPS 2026; Code available at https://github.com/akarinmoe/SRaR
Code: https://github.com/akarinmoe/SRaR
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Reward and Guidance through Rubrics: Promoting Exploration to Improve Multi-Domain Reasoning
- BabyVision: Visual Reasoning Beyond Language
- Interleaved Latent Visual Reasoning with Selective Perceptual Modeling
- Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domains
- Improve Mathematical Reasoning in Language Models by Automated Process Supervision
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- From Context to Skills: Can Language Models Learn from Context Skillfully?
- LongCat-Next: Lexicalizing Modalities as Discrete Tokens
- GRPO-VPS: Enhancing Group Relative Policy Optimization with Verifiable Process Supervision for Effective Reasoning
- Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations
- Grouter: Decoupling Routing from Representation for Accelerated MoE Training
- Qwen3 Technical Report
- DAPO: An Open-Source LLM Reinforcement Learning System at Scale
- MENTOR: Efficient Multimodal-Conditioned Tuning for Autoregressive Vision Generation Models
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis