Merge-of-Thought Distillation

• URL: http://arxiv.org/abs/2509.08814v3
• Date: Thu, 16 Oct 2025 15:43:35 GMT
• Title: Merge-of-Thought Distillation
• Authors: Zhanming Shen, Zeyu Qin, Zenan Huang, Hao Chen, Jiaqi Hu, Yihong Zhuang, Guoshan Lu, Gang Chen, Junbo Zhao,
• Abstract summary: Merge-of-Thought Distillation (MoT) is a lightweight framework that alternates between teacher-specific supervised fine-tuning branches and weight-space merging the resulting student variants.<n>On competition math benchmarks, applying MoT to a Qwen3-14B student surpasses strong models including Deepseek-R1, Qwen3-32B, and OpenAI-O1.<n>MoT consistently outperforms the best single-teacher distillation, improves general reasoning beyond mathematics, and shows robustness to distribution-shifted and peer-level teachers.
• Score: 23.53356244978525
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: Efficient reasoning distillation for long chain-of-thought (CoT) models is increasingly constrained by the assumption of a single oracle teacher, despite the practical availability of multiple candidate teachers and growing CoT corpora. We revisit teacher selection and observe that different students have different "best teachers," and even for the same student, the best teacher can vary across datasets. Therefore, to unify multiple teachers' reasoning abilities into a student to overcome conflicts among various teachers' supervision, we propose Merge-of-Thought Distillation (MoT), a lightweight framework that alternates between teacher-specific supervised fine-tuning branches and weight-space merging of the resulting student variants. On competition math benchmarks, using only about 200 CoT samples, applying MoT to a Qwen3-14B student surpasses strong models including Deepseek-R1, Qwen3-32B, and OpenAI-O1, demonstrating substantial gains. Besides, MoT consistently outperforms the best single-teacher distillation, improves general reasoning beyond mathematics while reducing catastrophic forgetting, and shows robustness to distribution-shifted and peer-level teachers. Finally, we have demonstrated MoT possesses consensus CoT by eliminating teacher-specific inductive biases and inter-teacher conflicts while repeatedly reinforcing the learning of consensus reasoning features. These results position MoT as a simple, effective route to efficiently distilling long CoT capabilities from diverse teachers into compact students.

Related papers

• "The Whole Is Greater Than the Sum of Its Parts": A Compatibility-Aware Multi-Teacher CoT Distillation Framework [16.96094045628127]
  Chain-of-Thought (CoT) reasoning empowers Large Language Models (LLMs) with remarkable capabilities but typically requires prohibitive parameter scales.<n>CoT distillation has emerged as a promising paradigm to transfer reasoning prowess into compact Student Models (SLMs)<n>We introduce COMPACT, a framework that adaptively fuses supervisions from different teachers by dynamically weighting teacher gradients.
  arXiv  Detail & Related papers  (2026-01-20T14:05:19Z)
• Long-Chain Reasoning Distillation via Adaptive Prefix Alignment [57.130176131042965]
  We propose a framework that exploits teacher CoTs for distillation through adaptive prefix alignment.<n>P-ALIGN adaptively truncates teacher-generated reasoning trajectories by determining whether the remaining suffix is concise.<n>Experiments on multiple mathematical reasoning benchmarks demonstrate that P-ALIGN outperforms all baselines by over 3%.
  arXiv  Detail & Related papers  (2026-01-15T04:40:45Z)
• Enhancing Long-Chain Reasoning Distillation through Error-Aware Self-Reflection [64.73809794561305]
  errOr-aware self-ReflectION (ORION) is a framework that refines teacher CoTs through an Error-Aware Reflection process.<n> Experiments on multiple mathematical reasoning benchmarks demonstrate that ORION consistently improves performance by more than 2% over all baselines.
  arXiv  Detail & Related papers  (2025-05-28T08:57:03Z)
• Unveiling the Key Factors for Distilling Chain-of-Thought Reasoning [33.02060729778806]
  This study examines the factors influencing Chain-of-Thought (CoT) distillation in Small Language Models (SLMs)<n>We find that SLMs exhibit a non-monotonic relationship with granularity, with stronger models benefiting from finer-grained reasoning and weaker models performing better with simpler CoT supervision.<n>These findings emphasize the need to tailor CoT strategies to specific student model, offering actionable insights for optimizing CoT distillation in SLMs.
  arXiv  Detail & Related papers  (2025-02-25T09:08:45Z)
• Switching Temporary Teachers for Semi-Supervised Semantic Segmentation [45.20519672287495]
  The teacher-student framework, prevalent in semi-supervised semantic segmentation, mainly employs the exponential moving average (EMA) to update a single teacher's weights based on the student's. This paper introduces Dual Teacher, a simple yet effective approach that employs dual temporary teachers aiming to alleviate the coupling problem for the student.
  arXiv  Detail & Related papers  (2023-10-28T08:49:16Z)
• Hybrid Distillation: Connecting Masked Autoencoders with Contrastive Learners [102.20090188997301]
  We explore how to obtain a model that combines Contrastive Learning (CL) and Masked Image Modeling (MIM) strengths. In order to better obtain both discrimination and diversity, we propose a simple but effective Hybrid Distillation strategy. Experiment results prove that Hybrid Distill can achieve superior performance on different benchmarks.
  arXiv  Detail & Related papers  (2023-06-28T02:19:35Z)
• On student-teacher deviations in distillation: does it pay to disobey? [54.908344098305804]
  Knowledge distillation has been widely used to improve the test accuracy of a "student" network. Despite being trained to fit the teacher's probabilities, the student may not only significantly deviate from the teacher probabilities, but may also outdo the teacher in performance.
  arXiv  Detail & Related papers  (2023-01-30T14:25:02Z)
• Faculty Distillation with Optimal Transport [53.69235109551099]
  We propose to link teacher's task and student's task by optimal transport. Based on the semantic relationship between their label spaces, we can bridge the support gap between output distributions. Experiments under various settings demonstrate the succinctness and versatility of our method.
  arXiv  Detail & Related papers  (2022-04-25T09:34:37Z)
• Confidence-Aware Multi-Teacher Knowledge Distillation [12.938478021855245]
  Confidence-Aware Multi-teacher Knowledge Distillation (CA-MKD) is proposed. It adaptively assigns sample-wise reliability for each teacher prediction with the help of ground-truth labels. Our CA-MKD consistently outperforms all compared state-of-the-art methods across various teacher-student architectures.
  arXiv  Detail & Related papers  (2021-12-30T11:00:49Z)
• Learning to Teach with Student Feedback [67.41261090761834]
  Interactive Knowledge Distillation (IKD) allows the teacher to learn to teach from the feedback of the student. IKD trains the teacher model to generate specific soft target at each training step for a certain student. Joint optimization for both teacher and student is achieved by two iterative steps.
  arXiv  Detail & Related papers  (2021-09-10T03:01:01Z)
• Representation Consolidation for Training Expert Students [54.90754502493968]
  We show that a multi-head, multi-task distillation method is sufficient to consolidate representations from task-specific teacher(s) and improve downstream performance. Our method can also combine the representational knowledge of multiple teachers trained on one or multiple domains into a single model.
  arXiv  Detail & Related papers  (2021-07-16T17:58:18Z)
• Teacher's pet: understanding and mitigating biases in distillation [61.44867470297283]
  Several works have shown that distillation significantly boosts the student's overall performance. However, are these gains uniform across all data subgroups? We show that distillation can harm performance on certain subgroups. We present techniques which soften the teacher influence for subgroups where it is less reliable.
  arXiv  Detail & Related papers  (2021-06-19T13:06:25Z)
• One Teacher is Enough? Pre-trained Language Model Distillation from Multiple Teachers [54.146208195806636]
  We propose a multi-teacher knowledge distillation framework named MT-BERT for pre-trained language model compression. We show that MT-BERT can train high-quality student model from multiple teacher PLMs. Experiments on three benchmark datasets validate the effectiveness of MT-BERT in compressing PLMs.
  arXiv  Detail & Related papers  (2021-06-02T08:42:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.