mlsanityData to 7 Oct 2026

HEAL: Hindsight Entropy-Assisted Learning for Reasoning Distillation

arXiv 2603.10359 ↗First posted 11 Mar 2026cs.AI cs.LGcited by 2

On the map: LLM reasoning → Test-time compute and efficient thinking

Abstract (arXiv)

Distilling reasoning capabilities from Large Reasoning Models (LRMs) into smaller models is typically constrained by the limitations of rejection sampling. Standard methods treat the teacher as a static filter, discarding complex "corner-case" problems where the teacher fails to explore valid solutions independently, thereby creating an artificial "Teacher Ceiling" for the student. In this work, we propose Hindsight Entropy-Assisted Learning (HEAL), an RL-free framework designed to bridge this reasoning gap. Drawing on the educational theory of the Zone of Proximal Development (ZPD), HEAL synergizes three core modules: (1) Guided Entropy-Assisted Repair (GEAR), an active intervention mechanism that detects critical reasoning breakpoints via entropy dynamics and injects targeted hindsight hints to repair broken trajectories; (2) Perplexity-Uncertainty Ratio Estimator (PURE), a ratio-based filtering heuristic that reduces high-anomaly shortcut-like rationales; and (3) Progressive Answer-guided Curriculum Evolution (PACE), a three-stage distillation strategy that organizes training from foundational alignment to hard-case adaptation. Extensive experiments on multiple benchmarks demonstrate that HEAL significantly outperforms traditional SFT distillation and other baselines.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 9 of its 31 references that are in the map. First posted on arXiv 11 Mar 2026.

Builds on

9 of 31 references.

References · 9

  1. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,748
  2. 2025LIMO: Less is More for Reasoningcited by 488
  3. 2026Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Modelscited by 386
  4. 2026Reinforcement Learning via Self-Distillationcited by 285
  5. 2026Self-Distillation Enables Continual Learningcited by 212
  6. 2025EDGE-GRPO: Entropy-Driven GRPO with Guided Error Correction for Advantage Diversitycited by 41
  7. 2025GHPO: Adaptive Guidance for Stable and Efficient LLM Reinforcement Learningcited by 33
  8. 2025Why Distillation can Outperform Zero-RL: The Role of Flexible Reasoningcited by 15
  9. 2025Reinforcement Learning vs. Distillation: Understanding Accuracy and Capability in LLM Reasoningcited by 14

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026Answer-Conditioned Chains of Thought Degrade Verifiable-Reasoning Distillation in Large Language Modelscited by 5
  2. 2026Teaching the Way, Not the Answer: Privileged Tutoring Distillation for Multimodal Policy Optimizationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Learning from Repaired Reasoning: Root-Cause-Guided On-Policy Distillation
  2. 2026E2-OPSD: Taming Entropy Overshoot in On-Policy Self-Distillation
  3. 2026GateKD: Confidence-Gated Closed-Loop Distillation for Robust Reasoning
  4. 2026Entropy-Preserving Supervised Fine-Tuning via Adaptive Self-Distillation for Large Reasoning Models
  5. 2026Respecting Self-Uncertainty in On-Policy Self-Distillation for Efficient LLM Reasoning
  6. 2025Entropy-Guided Reasoning Compression
  7. 2025Entropy-Guided Loop: Achieving Reasoning through Uncertainty-Aware Generation
  8. 2026ERR+: Sequential Entropy Resolution for Efficient and Decisive LLM Reasoning