When Does Exploration Prevent Collapse? Three Regimes and the Path of the Reward in Self-Consistency Training
cs.LG
Submitted: 2026-10-31
Updated: 2026-10-31
Terminology
Sources
- GRPO, Dr. GRPO, and DAPO Are Three Operations on One Number: The Group-Standard-Deviation Identity
- Uncalibrated Reasoning: GRPO Induces Overconfidence for Stochastic Outcomes
- Trust Your Instincts: Confidence-Driven Test-Time RL for Vision-Language-Action Models
- Group Entropy-Controlled Policy Optimization
- Certified Self-Consistency: Statistical Guarantees and Test-Time Training for Reliable Reasoning in LLMs
- The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models
- Collective Dynamics of Self-Reinforcement: How Cross-Task Coupling Drives Policy Collapse in Unsupervised RLVR
- Distribution-Aware Reward Estimation for Test-Time Reinforcement Learning
- Why GRPO Needs Normalization: A Local-Curvature Perspective on Adaptive Gradients
- REINFORCE++: Stabilizing Critic-Free Policy Optimization with Global Advantage Normalization
- Exploration-Preserving Policy Optimization
- Jointly Reinforcing Diversity and Quality in Language Model Generations
- Detecting and Mitigating the Correct-Answer Extinction Window in Test-Time Reinforcement Learning with Majority Voting
- ETTRL: Balancing Exploration and Exploitation in LLM Test-Time Reinforcement Learning Via Entropy Mechanism
- VLA-RL: Towards Masterful and General Robotic Manipulation with Scalable Reinforcement Learning
- CoRE: Consensus Rewards via Equilibrium for Test-Time Reinforcement Learning
- Reinforcement Learning with Verifiable Rewards: GRPO's Effective Loss, Dynamics, and Success Amplification
- Maximizing Confidence Alone Improves Reasoning
- Survive or Collapse: The Asymmetric Roles of Data Gating and Reward Grounding in Self-Play RL
- Proximal Policy Optimization Algorithms
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