Flash-OPD: Fast On-Policy Distillation

arXiv:2610.06105 · cs.LG · Submitted 2026-10-05 · Read on arXiv

cs.LG

Submitted: 2026-10-05

Updated: 2026-10-05

Comments: 21 pages, 9 figures

Code: https://github.com/Onedean/Flash-OPD

Project page: https://onedean.github.io/Flash-OPD-Site

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

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

Sources

• Learning to Foresee: Unveiling the Unlocking Efficiency of On-Policy Distillation
• Revisiting On-Policy Distillation: Empirical Failure Modes and Simple Fixes
• Distilling the Knowledge in a Neural Network
• Reinforcement Learning via Self-Distillation
• Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?
• Scaling Reasoning Efficiently via Relaxed On-Policy Distillation
• Rethinking On-Policy Distillation of Large Language Models: Phenomenology, Mechanism, and Recipe
• ADWIN: Adaptive Windows for Horizon-Aware On-Policy Distillation
• CRISP: Compressed Reasoning via Iterative Self-Policy Distillation
• DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
• Self-Distillation Enables Continual Learning
• Demystifying On-Policy Distillation: Roles, Pathologies, and Regulations
• Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation
• DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression
• Learning beyond Teacher: Generalized On-Policy Distillation with Reward Extrapolation
• Accordion-Thinking: Self-Regulated Step Summaries for Efficient and Readable LLM Reasoning
• Prune-OPD: Efficient and Reliable On-Policy Distillation for Long-Horizon Reasoning
• Online Experiential Learning for Language Models
• GLM-5: from Vibe Coding to Agentic Engineering
• Are Full Rollouts Necessary for On-Policy Distillation?

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