Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-Training

arXiv:2605.12483 · cs.LG, cs.AI · Submitted 2026-05-12 · Read on arXiv

cs.LG, cs.AI

Submitted: 2026-05-12

Updated: 2026-10-12

Terminology

• Distillation — 3× in this paper · explained in 2 episodes
• GRPO — 3× in this paper · explained in 6 episodes
• On-Policy Distillation — 3× in this paper · explained in 3 episodes
• Ablation — 1× in this paper · explained in 1 episode

Sources

• Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
• Rubric-based On-policy Distillation
• Flow-OPD: On-Policy Distillation for Flow Matching Models
• Specializing Smaller Language Models towards Multi-Step Reasoning
• The Llama 3 Herd of Models
• Self-Distillation Zero: Self-Revision Turns Binary Rewards into Dense Supervision
• Distilling the Knowledge in a Neural Network
• Uni-OPD: Unifying On-Policy Distillation with a Dual-Perspective Recipe
• Beyond Correctness: Learning Robust Reasoning via Transfer
• Explanations from Large Language Models Make Small Reasoners Better
• Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe
• ORBIT: On-policy Exploration-Exploitation for Controllable Multi-Budget Reasoning
• Knowledge Distillation with Training Wheels
• CRISP: Compressed Reasoning via Iterative Self-Policy Distillation
• Proximal Policy Optimization Algorithms
• DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
• Self-Distillation Enables Continual Learning
• HybridFlow: A Flexible and Efficient RLHF Framework
• A Survey of On-Policy Distillation for Large Language Models
• Skill-SD: Skill-Conditioned Self-Distillation for Multi-turn LLM Agents

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