# Learning with Rare Success but Rich Feedback via Reflection-Enhanced Self-Distillation

arXiv: https://arxiv.org/abs/2605.12741 · HTML: https://mlsanity.com/paper/2605.12741/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in On-policy and self-distillation for LLMs, one of 12 directions in Knowledge distillation.
- Builds on 29 of its 30 references that are in the map.
- First posted on arXiv 12 May 2026.

## Placed in

- Knowledge distillation / [On-policy and self-distillation for LLMs](https://mlsanity.com/distill/on-policy-and-self-distillation-for-llms/)

## Builds on (most cited first, 29 of 29)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter](https://mlsanity.com/paper/1910.01108/) (2019)
- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://mlsanity.com/paper/2303.11366/) (2023)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [Gemma 2: Improving Open Language Models at a Practical Size](https://mlsanity.com/paper/2408.00118/) (2024)
- [Instruction-Following Evaluation for Large Language Models](https://mlsanity.com/paper/2311.07911/) (2023)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [On-Policy Distillation of Language Models: Learning from Self-Generated Mistakes](https://mlsanity.com/paper/2306.13649/) (2023)
- [Sequence-Level Knowledge Distillation](https://mlsanity.com/paper/1606.07947/) (2016)
- [Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models](https://mlsanity.com/paper/2401.01335/) (2024)
- [Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models](https://mlsanity.com/paper/2601.18734/) (2026)
- [Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models](https://mlsanity.com/paper/2510.04618/) (2025)
- [Reinforcement Learning via Self-Distillation](https://mlsanity.com/paper/2601.20802/) (2026)
- [Is DPO Superior to PPO for LLM Alignment? A Comprehensive Study](https://mlsanity.com/paper/2404.10719/) (2024)
- [MiMo-V2-Flash Technical Report](https://mlsanity.com/paper/2601.02780/) (2026)
- [On-Policy Context Distillation for Language Models](https://mlsanity.com/paper/2602.12275/) (2026)
- [RL's Razor: Why Online Reinforcement Learning Forgets Less](https://mlsanity.com/paper/2509.04259/) (2025)
- [DistiLLM: Towards Streamlined Distillation for Large Language Models](https://mlsanity.com/paper/2402.03898/) (2024)
- [Self-Distilled RLVR](https://mlsanity.com/paper/2604.03128/) (2026)
- [Self-Discover: Large Language Models Self-Compose Reasoning Structures](https://mlsanity.com/paper/2402.03620/) (2024)
- [Why Does Self-Distillation (Sometimes) Degrade the Reasoning Capability of LLMs?](https://mlsanity.com/paper/2603.24472/) (2026)
- [Retaining by Doing: The Role of On-Policy Data in Mitigating Forgetting](https://mlsanity.com/paper/2510.18874/) (2025)
- [Privileged Information Distillation for Language Models](https://mlsanity.com/paper/2602.04942/) (2026)
- [Answering Questions by Meta-Reasoning over Multiple Chains of Thought](https://mlsanity.com/paper/2304.13007/) (2023)
- [Unifying Group-Relative and Self-Distillation Policy Optimization via Sample Routing](https://mlsanity.com/paper/2604.02288/) (2026)
- [Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training](https://mlsanity.com/paper/2507.05386/) (2025)
- [RL Grokking Recipe: How Does RL Unlock and Transfer New Algorithms in LLMs?](https://mlsanity.com/paper/2509.21016/) (2025)

## Built on by (most cited first, 8 of 8)

- [Latent On-Policy Self-Distillation](https://mlsanity.com/paper/2608.13040/) (2026)
- [Revisit to Segment: Working Memory Distillation for Reasoning Segmentation](https://mlsanity.com/paper/2609.34863/) (2026)
- [Rubric-Aware On-Policy Self-Distillation for LLM Personalization](https://mlsanity.com/paper/2609.35262/) (2026)
- [Shockingly Simple Self-retrospection Improves Agentic Models Without RL](https://mlsanity.com/paper/2609.35741/) (2026)
- [Train Ahead, Distill Back: Bootstrapping On-Policy Self-Distillation for Large Language Models](https://mlsanity.com/paper/2609.37132/) (2026)
- [Privileged Context as Drift in On-Policy Self-Distillation](https://mlsanity.com/paper/2610.07842/) (2026)
- [ReTeach: Building a Self-Teacher through Multi-Round Reflection and Retry](https://mlsanity.com/paper/2610.11529/) (2026)
- [Self-Distillation with Verbal Regularization](https://mlsanity.com/paper/2610.16354/) (2026)

## Abstract (arXiv)

Enabling Large Language Models (LLMs) to continuously improve from environmental interactions is a central challenge in post-training. While on-policy self-distillation offers a promising paradigm, existing methods predominantly treat environmental feedback as a passive conditioning signal. Consequently, they heavily rely on successful demonstrations and struggle to learn in rare-success regimes. To bridge this gap, we introduce Reflection-Enhanced Self-Distillation (RESD), a framework that transforms raw failure feedback into an active source of corrective supervision. Instead of passively appending feedback, RESD interprets failed trajectories by generating retrospective reflections to diagnose local errors, and curates a persistent global playbook to preserve reusable lessons across training steps. The enriched context enables the self-teacher to provide actionable token-level supervision even in the absence of successful rollouts. Empirical evaluations on multiple continual learning tasks demonstrate that RESD substantially outperforms standard self-distillation baselines. Furthermore, RESD achieves significantly faster early-stage improvement than GRPO with 8× samples using only a single rollout per prompt, highlighting its superior interaction efficiency.