Beyond Prompt Count: How Data Shapes Transfer in On-Policy Distillation

arXiv:2609.37377 · cs.AI · Submitted 2026-09-29 · Read on arXiv

cs.AI

Submitted: 2026-09-29

Updated: 2026-09-29

Code: https://github.com/wyy-1112/dissecting-opd

Terminology

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

Sources

• Training Verifiers to Solve Math Word Problems
• Process Reinforcement through Implicit Rewards
• DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence
• MegaScience: Pushing the Frontiers of Post-Training Datasets for Science Reasoning
• Rethinking On-Policy Distillation of Large Language Models II: One Training Example
• JustRL: Scaling a 1.5B LLM with a Simple RL Recipe
• Scaling Reasoning Efficiently via Relaxed On-Policy Distillation
• KAT-Coder-V2 Technical Report
• Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe
• STEPS: Selective On-Policy Self-Distillation for Reasoning
• Consolidating RLVR Capabilities Across Domains: A Deep Dive into Fusion Paradigms
• MiMo-V2-Flash Technical Report
• Qwen3 Technical Report
• GLM-5: from Vibe Coding to Agentic Engineering
• Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models
• HY-MT1.5 Technical Report
• Instruction-Following Evaluation for Large Language Models

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