Toward Plasticity-Preserving KL Regularization for Capability Retention in LLM Reinforcement Learning
Li Wang, Xiaodong Lu, Xiaohan Wang, Jiajun Chai, Wei Lin, Tianhao Peng, Guojun Yin
cs.LG, cs.CL
Submitted: 2026-08-03
Code: https://github.com/Lumina04/CoKL
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
- Language Models that Think, Chat Better
- Advancing General-Purpose Reasoning Models with Modular Gradient Surgery
- The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models
- Unlocking Reasoning Capabilities in LLMs via Reinforcement Learning Exploration
- Nemotron-Math: Efficient Long-Context Distillation of Mathematical Reasoning from Multi-Mode Supervision
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- KL-Regularized Reinforcement Learning is Designed to Mode Collapse
- Measuring Mathematical Problem Solving With the MATH Dataset
- Whatever Remains Must Be True: Filtering Drives Reasoning in LLMs, Shaping Diversity
- SAGE: Shaping Anchors for Guided Exploration in RLVR of LLMs
- Omni-Thinker: Scaling Multi-Task RL in LLMs with Hybrid Reward and Task Scheduling
- The Choice of Divergence: A Neglected Key to Mitigating Diversity Collapse in Reinforcement Learning with Verifiable Reward
- Can One Domain Help Others? A Data-Centric Study on Multi-Domain Reasoning via Reinforcement Learning
- expo: Exploration-prioritized policy optimization via adaptive kl regulation and gaussian curriculum sampling
- ResRL: Boosting LLM Reasoning via Negative Sample Projection Residual Reinforcement Learning
- ResT: Reshaping Token-Level Policy Gradients for Tool-Use Large Language Models
- Skywork-Reward: Bag of Tricks for Reward Modeling in LLMs
- Understanding R1-Zero-Like Training: A Critical Perspective
- Uniform-Correct Policy Optimization: Breaking RLVR's Indifference to Diversity
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks