mlsanityData to 7 Oct 2026

Long-Short Chain-of-Thought Mixture Supervised Fine-Tuning Eliciting Efficient Reasoning in Large Language Models

arXiv 2505.03469 ↗First posted 6 May 2025cs.CLcited by 37

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

Abstract (arXiv)

Recent advances in large language models have demonstrated that Supervised Fine-Tuning (SFT) with Chain-of-Thought (CoT) reasoning data distilled from large reasoning models (e.g., DeepSeek R1) can effectively transfer reasoning capabilities to non-reasoning models. However, models fine-tuned with this approach inherit the "overthinking" problem from teacher models, producing verbose and redundant reasoning chains during inference. To address this challenge, we propose Long-Short Chain-of-Thought Mixture Supervised Fine-Tuning (LS-Mixture SFT), which combines long CoT reasoning dataset with their short counterparts obtained through structure-preserved rewriting. Our experiments demonstrate that models trained using the LS-Mixture SFT method, compared to those trained with direct SFT, achieved an average accuracy improvement of 2.3% across various benchmarks while substantially reducing model response length by approximately 47.61%. This work offers an approach to endow non-reasoning models with reasoning capabilities through supervised fine-tuning while avoiding the inherent overthinking problems inherited from teacher models, thereby enabling efficient reasoning in the fine-tuned models.

Key facts. Cited by 37 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 26 of its 41 references that are in the map. First posted on arXiv 6 May 2025.

Builds on

26 of 41 references.

Closest ancestors · 15

Same direction, newest first

  1. 2025Efficient Reasoning for LLMs through Speculative Chain-of-Thought27 Apr
  2. 2025Thought Manipulation: External Thought Can Be Efficient for Large Reasoning Models18 Apr
  3. 2025A Survey on Test-Time Scaling in Large Language Models: What, How, Where, and How Well?31 Mar
  4. 2025Think Twice: Enhancing LLM Reasoning by Scaling Multi-round Test-time Thinking25 Mar
  5. 2025Stop Overthinking: A Survey on Efficient Reasoning for Large Language Models20 Mar
  6. 2025L1: Controlling How Long A Reasoning Model Thinks With Reinforcement Learning6 Mar
  7. 2025DAST: Difficulty-Adaptive Slow-Thinking for Large Reasoning Models6 Mar
  8. 2025Efficient Test-Time Scaling via Self-Calibration25 Feb
  9. 2025Demystifying Long Chain-of-Thought Reasoning in LLMs5 Feb
  10. 2025O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning22 Jan
  11. 2024Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs30 Dec
  12. 2024C3oT: Generating Shorter Chain-of-Thought without Compromising Effectiveness16 Dec
3 more
  1. 2024Imitate, Explore, and Self-Improve: A Reproduction Report on Slow-thinking Reasoning Systems12 Dec
  2. 2024O1 Replication Journey -- Part 2: Surpassing O1-preview through Simple Distillation, Big Progress or Bitter Lesson?25 Nov
  3. 2024O1 Replication Journey: A Strategic Progress Report -- Part 18 Oct

Other references · 11

  1. 2024RouteLLM: Learning to Route LLMs with Preference Datacited by 633
  2. 2025Chain of Draft: Thinking Faster by Writing Lesscited by 290
  3. 2024Chain of Thought Empowers Transformers to Solve Inherently Serial Problemscited by 256
  4. 2023Navigate through Enigmatic Labyrinth A Survey of Chain of Thought Reasoning: Advances, Frontiers and Futurecited by 184
  5. 2025Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyondcited by 175
  6. 2025Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketchingcited by 116
  7. 2023SciEval: A Multi-Level Large Language Model Evaluation Benchmark for Scientific Researchcited by 112
  8. 2025An Empirical Study on Eliciting and Improving R1-like Reasoning Modelscited by 78
  9. 2025100 Days After DeepSeek-R1: A Survey on Replication Studies and More Directions for Reasoning Language Modelscited by 29
  10. 2024MuMath-Code: Combining Tool-Use Large Language Models with Multi-perspective Data Augmentation for Mathematical Reasoningcited by 22
  11. 2024ARES: Alternating Reinforcement Learning and Supervised Fine-Tuning for Enhanced Multi-Modal Chain-of-Thought Reasoning Through Diverse AI Feedbackcited by 13

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 23; 2026: 14.

37 papers.

Sort:

By area:

  1. 2025Learning When to Think: Shaping Adaptive Reasoning in R1-Style Models via Multi-Stage RLcited by 59
  2. 2025SwS: Self-aware Weakness-driven Problem Synthesis in Reinforcement Learning for LLM Reasoningcited by 39
  3. 2025Don't Overthink It: A Survey of Efficient R1-style Large Reasoning Modelscited by 36
  4. 2025Reasoning on a Budget: A Survey of Adaptive and Controllable Test-Time Compute in LLMscited by 31
  5. 2025Promoting Efficient Reasoning with Verifiable Stepwise Rewardcited by 27
  6. 2025Mini-Omni-Reasoner: Token-Level Thinking-in-Speaking in Large Speech Modelscited by 20
  7. 2025ARES: Multimodal Adaptive Reasoning via Difficulty-Aware Token-Level Entropy Shapingcited by 16
  8. 2025Concise Reasoning, Big Gains: Pruning Long Reasoning Trace with Difficulty-Aware Promptingcited by 13
  9. 2025From Efficiency to Adaptivity: A Deeper Look at Adaptive Reasoning in Large Language Modelscited by 11
  10. 2025The First Impression Problem: Internal Bias Triggers Overthinking in Reasoning Modelscited by 10
  11. 2026Neural Chain-of-Thought Search: Searching the Optimal Reasoning Path to Enhance Large Language Modelscited by 7
  12. 2025How Well Do Large Reasoning Models Translate? A Comprehensive Evaluation for Multi-Domain Machine Translationcited by 5
25 more
  1. 2025PEAR: Phase Entropy Aware Reward for Efficient Reasoningcited by 5
  2. 2025CaughtCheating: Is Your MLLM a Good Cheating Detective? Exploring the Boundary of Visual Perception and Reasoningcited by 4
  3. 2025Learning to Ponder: Adaptive Reasoning in Latent Spacecited by 4
  4. 2025Putting on the Thinking Hats: A Survey on Chain of Thought Fine-tuning from the Perspective of Human Reasoning Mechanismcited by 4
  5. 2025One Token Embedding Is Enough to Deadlock Your Large Reasoning Modelcited by 4
  6. 2025Rethinking Supervised Fine-Tuning: Emphasizing Key Answer Tokens for Improved LLM Accuracycited by 4
  7. 2026ConPress: Learning Efficient Reasoning from Multi-Question Contextual Pressurecited by 4
  8. 2025The Imitation Game: Turing Machine Imitator is Length Generalizable Reasonercited by 3
  9. 2025TrajSelector: Harnessing Latent Representations for Efficient and Effective Best-of-N in Large Reasoning Modelcited by 3
  10. 2026ReEfBench: Quantifying the Reasoning Efficiency of LLMscited by 3
  11. 2026Draft-Thinking: Learning Efficient Reasoning in Long Chain-of-Thought LLMscited by 3
  12. 2026ReasonAlloc: Hierarchical Decoding-Time KV Cache Budget Allocation for Reasoning Modelscited by 3
  13. 2025The Valley of Code Reasoning: Scaling Knowledge Distillation of Large Language Modelscited by 2
  14. 2025Efficient Reasoning via Thought-Training and Thought-Free Inferencecited by 2
  15. 2026Stop Unnecessary Reflection: Training LRMs for Efficient Reasoning with Adaptive Reflection and Length Coordinated Penaltycited by 2
  16. 2025Focused Chain-of-Thought: Efficient LLM Reasoning via Structured Input Informationcited by 1
  17. 2026QA-Merging: Query-Adaptive Reasoning via Layer Selective Model Mergingcited by 1
  18. 2026GRLO: Towards Generalizable Reinforcement Learning in Open-Ended Environments from Zerocited by 1
  19. 2025Enhancing LLM Reasoning via Non-Human-Like Reasoning Path Preference Optimizationcited by 0
  20. 2026Compute Where it Counts: Self Optimizing Language Modelscited by 0
  21. 2026Data Difficulty and the Generalization--Extrapolation Tradeoff in LLM Fine-Tuningcited by 0
  22. 2026D2Evo: Dual Difficulty-Aware Self-Evolution for Data-Efficient Reinforcement Learningcited by 0
  23. 2026GRIP: Granular Reward-Guided Parameter Interpolation for Efficient Reasoningcited by 0
  24. 2026Revisiting Complete Reasoning Traces for Post-Trainingcited by 0
  25. 2026Efficient Reasoning via Constrained Optimization in Latent Spacecited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025QFFT, Question-Free Fine-Tuning for Adaptive Reasoning
  2. 2026Zipping the Thought: When and How Compressed Reasoning Data Works in LLM Post-Training
  3. 2025Thinking Preference Optimization
  4. 2024Fine-Tuning on Diverse Reasoning Chains Drives Within-Inference CoT Refinement in LLMs
  5. 2024ReFT: Reasoning with Reinforced Fine-Tuning
  6. 2022Large Language Models Are Reasoning Teachers
  7. 2026On the Role of Reasoning Patterns in the Generalization Discrepancy of Long Chain-of-Thought Supervised Fine-Tuning
  8. 2025Mixture of Reasonings: Teach Large Language Models to Reason with Adaptive Strategies