CoT Distillation

CoT: Chain-of-Thought

Follow this topic Ask about this topic

Chain-of-thought distillation trains a student model using a teacher’s intermediate reasoning, aiming to transfer problem-solving behavior as well as correct answers. Common approaches include supervised fine-tuning on teacher-generated traces, matching teacher and student token distributions during student rollouts, and compressing or selecting traces to fit the learner. Methods are tested on mathematical reasoning benchmarks such as GSM8K, MATH, and AIME, as well as code, knowledge, and multimodal tasks. A central tension is that detailed traces can provide useful supervision but may be costly, incorrect, or mismatched to what a student can learn; answer accuracy alone also does not guarantee faithful reasoning.

Latest papers

80 papers13 papers in the last four weeks, up 225% on the four weeks before. 0.1% of all new papers.

Apr 17, 2026

Improving Reasoning Capabilities in Small Models through Mixture-of-Layers Distillation with Stepwise Attention on Key Information

The significant computational demands of large language models have increased interest in distilling reasoning abilities into smaller models via Chain-of-Thought (CoT) distillation. Current CoT distillation methods mainly focus on transferring teacher-generated rationales for complex reasoning to student models. However, they do not adequately explore teachers' dynamic attention toward critical information during reasoning. We find that language models exhibit progressive attention shifts towards key information during reasoning, which implies essential clues for drawing conclusions. Building on this observation and analysis, we introduce a novel CoT distillation framework that transfers the teacher's stepwise attention on key information to the student model. This establishes structured guidance for the student's progressive concentration on key information during reasoning. More importantly, we develop a Mixture of Layers module enabling dynamic alignment that adapts to different layers between the teacher and student. Our method achieves consistent performance improvements across multiple mathematical and commonsense reasoning datasets. To our knowledge, it is the first method to leverage stepwise attention within CoT distillation to improve small model reasoning.

Yao Chen, Jiawei Sheng, Wenyuan Zhang +1

Institute of Information Engineering, Chinese Academy of Sciences · School of Cyber Security, University of Chinese Academy of Sciences

CoT Distillation Small Language Models Transfer Learning Language Model Distillation Attention Mechanisms Knowledge Distillation

Apr 10, 2026

Revisiting the Capacity Gap in Chain-of-Thought Distillation from a Practical Perspective

Chain-of-thought (CoT) distillation transfers reasoning behaviors from a strong teacher to a smaller student, but prior work reports a capacity gap: distillation may fail when the teacher-student capability mismatch is large. We revisit the capacity gap from a practical perspective by re-examining commonly used experimental settings. Notably, we find that CoT distillation often degrades performance compared to the student's pre-distillation baseline, and that some settings used in prior work, while suitable for establishing the capacity gap as a phenomenon, do not reflect realistic deployment scenarios. Complementing prior work that establishes the capacity gap, we evaluate its practical impact under more realistic settings and find that it does not consistently dominate; stronger teachers tend to be preferable when candidate teachers differ substantially in performance. Our results offer practical guidance for selecting teacher-student pairs in CoT distillation.

Tokio Kajitsuka, Ukyo Honda, Sho Takase

The University of Tokyo, Tokyo, Japan · CyberAgent, Tokyo, Japan

CoT Distillation Reasoning Distillation Teacher-Student Learning Knowledge Distillation

Apr 7, 2026

State-of-the-Art Arabic Language Modeling with Sparse MoE Fine-Tuning and Chain-of-Thought Distillation

This paper introduces Arabic-DeepSeek-R1, an application-driven open-source Arabic LLM that leverages a sparse MoE backbone to address the digital equity gap for under-represented languages, and establishes a new SOTA across the entire Open Arabic LLM Leaderboard (OALL). Our four-phase CoT distillation scheme integrates Arabic-specific linguistic verification and regional ethical norms into a 372M-token, contamination-controlled 80/20 Arabic-English training mixture. Arabic-DeepSeek-R1 achieves the highest average score across the seven-benchmark OALL suite while establishing SOTA or near-SOTA, including dominant results on grammar-focused MadinahQA (surpassing both GPT-5.1 and the OALL leader by substantial margins), safety-oriented AraTrust, multi-ability AlGhafa, and retrieval-augmented ALRAGE. Our results indicate that the combination of sparse MoE architecture, culturally-informed CoT distillation with explicit Arabic linguistic checks, and strategic bilingual data curation enables an open-source adapted model to systematically outperform the proprietary frontier system GPT-5.1 on the majority of benchmarks evaluating comprehensive language-specific tasks: the first such demonstration for Arabic LLMs. These findings indicate that much of Arabic's performance deficit in current LLM ecosystems stems from under-specialization rather than architectural limitations, and that parameter-efficient adaptation of open reasoning models can yield breakthrough SOTA performance without industrial-scale pretraining costs. Arabic-DeepSeek-R1 establishes a validated and replicable framework for sovereign and domain-specific language technologies, demonstrating that strategic, culturally-grounded adaptation of sparse MoE backbones offers a viable and cost-effective pathway to achieving record-breaking performance across standardized benchmarks for low-resource languages.

Navan Preet Singh, Anurag Garikipati, Ahmed Abulkhair +6

Forta, Houston, TX · Incept Labs, Houston, TX · Titan Holdings, San Francisco, CA

Arabic Language Models Arabic NLP LLM Fine-Tuning Low-Resource Language Processing Language Model Distillation CoT Distillation

Apr 3, 2026

Student-in-the-Loop Chain-of-Thought Distillation via Generation-Time Selection

Large reasoning models achieve strong performance on complex tasks through long chain-of-thought (CoT) trajectories, but directly transferring such reasoning processes to smaller models remains challenging. A key difficulty is that not all teacher-generated reasoning trajectories are suitable for student learning. Existing approaches typically rely on post-hoc filtering, selecting trajectories after full generation based on heuristic criteria. However, such methods cannot control the generation process itself and may still produce reasoning paths that lie outside the student's learning capacity. To address this limitation, we propose Gen-SSD (Generation-time Self-Selection Distillation), a student-in-the-loop framework that performs generation-time selection. Instead of passively consuming complete trajectories, the student evaluates candidate continuations during the teacher's sampling process, guiding the expansion of only learnable reasoning paths and enabling early pruning of unhelpful branches. Experiments on mathematical reasoning benchmarks demonstrate that Gen-SSD consistently outperforms standard knowledge distillation and recent baselines, with improvements of around 5.9 points over Standard KD and up to 4.7 points over other baselines. Further analysis shows that Gen-SSD produces more stable and learnable reasoning trajectories, highlighting the importance of incorporating supervision during generation for effective distillation.

Chaoqun He, Yingfa Chen, Chaojun Xiao +2

Tsinghua University

Language Model Distillation CoT Distillation Knowledge Distillation Selective Knowledge Distillation

Mar 20, 2026

Grounded Chess Reasoning in Language Models via Master Distillation

Language models often lack grounded reasoning capabilities in specialized domains where training data is scarce but bespoke systems excel. We introduce a general framework for distilling expert system reasoning into natural language chain-of-thought explanations, enabling compact models to acquire domain expertise and the ability to generate faithful, grounded explanations. Rather than distilling only final outputs, we capture the full reasoning process, transforming opaque expert computations into transparent, step-by-step explanations. We demonstrate this approach in chess, a canonical reasoning domain where language models continue to underperform. Our 4B parameter model, C1, advances from a near-zero baseline to 48.1% accuracy, outperforming all open-source models and most frontier proprietary systems. Notably, C1 surpasses its distillation teacher and generates solutions in two orders of magnitude fewer tokens than baselines. Unlike prior neural chess approaches that predict only best moves, C1 generates explainable solutions revealing strategic reasoning. Our pipeline combines supervised fine-tuning and reinforcement learning with theme-balanced data sampling for comprehensive tactical coverage. Master Distillation demonstrates how to inject expert-level knowledge into compact models for under-optimized domains, offering a recipe for unlocking RLVR where LLMs lack sufficient base capabilities.

Zhenwei Tang, Qianfeng Wen, Seth Grief-Albert +4

Department of Computer Science, University of Toronto · Queen’s University · Coolwei AI Lab

LLM Grounding CoT Distillation Language Model Distillation Chess LLM Reasoning

Mar 15, 2026

Distilling Reasoning Without Knowledge: A Framework for Reliable LLMs

Fact-seeking question answering with large language models (LLMs) remains unreliable when answers depend on up-to-date or conflicting information. Although retrieval-augmented and tool-using LLMs reduce hallucinations, they often rely on implicit planning, leading to inefficient tool usage. We propose a modular framework that explicitly separates planning from factual retrieval and answer synthesis. A lightweight student planner is trained via a teacher-student framework to generate structured decompositions consisting of abstract reasoning steps and searchable fact requests. The supervision signals contain only planning traces and fact requests, without providing factual answers or retrieved evidence. At inference, the planner produces plans, while prompt-engineered modules perform retrieval and response synthesis. We evaluate the proposed framework on SEAL-0, an extremely challenging benchmark for search-augmented LLMs. Results show that supervised planning improves both accuracy and latency compared to monolithic reasoning models and prompt-based tool-augmented frameworks, demonstrating that explicitly learned planning structures are essential for reliable fact-seeking LLMs.

Auksarapak Kietkajornrit, Jad Tarifi, Nima Asgharbeygi

Integral AI

LLM Planning LLM Reliability Question Answering Agentic RAG CoT Distillation

Mar 11, 2026

HEAL: Hindsight Entropy-Assisted Learning for Reasoning Distillation

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.

Wenjing Zhang, Jiangze Yan, Jieyun Huang +7

Data Science & AI Research Institute, China Unicom · Unicom Data Intelligence, China Unicom

CoT Distillation Language Model Distillation LLM Reasoning Large Reasoning Models

Mar 5, 2026

CRISP: Compressed Reasoning via Iterative Self-Policy Distillation

Reasoning models often generate far more tokens than a task requires, which raises inference cost and can compound errors. We introduce CRISP (Compressed Reasoning via Iterative Self-Policy Distillation), an on-policy self-distillation method that teaches a model to reason more concisely by distilling its own concise behavior back into itself. The method uses a single idea: condition the same model on a "be concise" instruction to obtain teacher logits, then minimize the per-token reverse KL divergence between the student and this teacher on the student's own rollouts. It requires no ground-truth answers, no token budgets, and no difficulty estimators. The reverse-KL objective is naturally difficulty-adaptive: it compresses easy problems aggressively while preserving the reasoning steps that hard problems require. On Qwen3-14B, CRISP cuts reasoning length by up to 56% on MATH-500 and 38% on the harder AIME 2024, while improving MATH-500 accuracy by up to 3.3 points over the base model and holding AIME 2024 accuracy within about one point. This behavior generalizes across model sizes and families: Qwen3-8B shows the same compression with accuracy preserved, and DeepSeek-R1-Distill-Llama-8B improves accuracy on all five benchmarks while shortening its responses. General capabilities are preserved across all three models. Code is available at https://github.com/HJSang/OPSD_Reasoning_Compression.

Hejian Sang, Yuanda Xu, Zhengze Zhou +3

CoT Distillation Efficient Language Model Reasoning On-Policy Distillation CoT Compression LLM Reasoning

Feb 27, 2026

Learning from Partial Chain-of-Thought via Truncated-Reasoning Self-Distillation

Reasoning-oriented language models achieve strong performance by generating long chain-of-thought traces at inference time. However, this capability comes with substantial and often excessive computational cost, which can materialize in redundant or inefficient reasoning. We study this setting and introduce Truncated-Reasoning Self-Distillation (TRSD), a lightweight post-training procedure that encourages models to produce correct predictions from partial reasoning traces. In TRSD, a frozen teacher model first generates a full reasoning trace and evaluates the corresponding answer distribution conditioned on the prompt and the complete reasoning to construct a synthetic training target. A student model with the same architecture is then trained to match the teacher's answer distribution while being conditioned only on a truncated prefix of its reasoning trace. Across multiple reasoning benchmarks and token budgets, we demonstrate that TRSD improves robustness to truncated inference, with far reduced accuracy tradeoffs when applied to a diverse set of reasoning models. Moreover, although never explicitly regularized for shorter generation during training, we also find that TRSD-trained models inherently output shorter reasoning traces without truncation, significantly reducing inference-time costs even without artificial interventions.

Gianluigi Silvestri, Edoardo Cetin

Radboud University · Sakana AI

Efficient Language Model Reasoning CoT Distillation Self-Distillation

Feb 25, 2026

D-COT: Disciplined Chain-of-Thought Learning for Efficient Reasoning in Small Language Models

Chain-of-Thought (CoT) distillation from Large Language Models (LLMs) often induces "overthinking" in Small Language Models (SLMs), leading to performance degradation and excessive token consumption. In this study, we propose Disciplined Chain-of-Thought (D-CoT), a novel framework that enforces a structured reasoning process using control tags -- such as <TEMP_LOW> for fact-checking and <TEMP_HIGH> for multi-perspective exploration -- as auxiliary scaffolding during training. By optimizing the CoT trajectory, D-CoT suppresses reasoning drift and simultaneously achieves token reduction and performance improvement. We demonstrate the efficacy of our approach on Qwen3-8B: with only 5,000 training samples, D-CoT significantly boosts accuracy on GPQA-diamond by 9.9% and MMLU-Pro (0-shot) by 9.1%, while drastically reducing computational costs. Furthermore, we confirm that the model internalizes this disciplined thought structure, maintaining high performance even without explicit control tags during inference.

Shunsuke Ubukata

Toyo University

Efficient Language Model Reasoning CoT Distillation Language Model Distillation Overthinking in Language Models

Feb 25, 2026

Explain in Your Own Words: Improving Reasoning via Token-Selective Dual Knowledge Distillation

Knowledge Distillation (KD) can transfer the reasoning abilities of large models to smaller ones, which can reduce the costs to generate Chain-of-Thoughts for reasoning tasks. KD methods typically ask the student to mimic the teacher's distribution over the entire output. However, a student with limited capacity can be overwhelmed by such extensive supervision causing a distribution mismatch, especially in complex reasoning tasks. We propose Token-Selective Dual Knowledge Distillation (TSD-KD), a framework for student-centric distillation. TSD-KD focuses on distilling important tokens for reasoning and encourages the student to explain reasoning in its own words. TSD-KD combines indirect and direct distillation. Indirect distillation uses a weak form of feedback based on preference ranking. The student proposes candidate responses generated on its own; the teacher re-ranks those candidates as indirect feedback without enforcing its entire distribution. Direct distillation uses distribution matching; however, it selectively distills tokens based on the relative confidence between teacher and student. Finally, we add entropy regularization to maintain the student's confidence during distillation. Overall, our method provides the student with targeted and indirect feedback to support its own reasoning process and to facilitate self-improvement. The experiments show the state-of-the-art performance of TSD-KD on 10 challenging reasoning benchmarks, outperforming the baseline and runner-up in accuracy by up to 54.4% and 40.3%, respectively. Notably, a student trained by TSD-KD even outperformed its own teacher model in four cases by up to 20.3%. The source code is available at https://github.com/kmswin1/TSD-KD.

Minsang Kim, Seung Jun Baek

Korea University · SK Telecom

Language Model Distillation Selective Knowledge Distillation CoT Distillation LLM Reasoning

Feb 24, 2026

GATES: Self-Distillation under Privileged Context with Consensus Gating

We study self-distillation in settings where supervision is unreliable: there are no ground truth labels, verifiable rewards, or external graders to evaluate answers. We focus on document-grounded question answering with asymmetric context, where a single model serves as both tutor (with access to a relevant source document during training) and student (answering from the question alone at test time). Rather than assuming tutor correctness, we derive supervision online from tutor consensus by sampling multiple document-grounded reasoning traces and using agreement to gate learning. Conditioned on this reliability signal, we distill knowledge through full tutor reasoning trajectories (not just final answers), providing a dense and stable learning signal. Empirically, this consensus-gated trajectory distillation substantially improves transfer to the document-free student. Held-out in-domain accuracy under asymmetric evaluation improves from 46.0% to 62.0%, and average (maj@8) accuracy on public document-free math benchmarks improves from 20.2% to 35.4%.

Alex Stein, Furong Huang, Tom Goldstein

University of Maryland, College Park

Knowledge Distillation On-Policy Self-Distillation Context Distillation CoT Distillation Document QA

Feb 16, 2026

Measuring and Mitigating Post-hoc Rationalization in Reverse Chain-of-Thought Generation

Reverse Chain-of-Thought Generation (RCG) synthesizes reasoning traces from query-answer pairs, but answer-visible generation can justify a pre-committed answer rather than derive it. This post-hoc rationalization creates a train-inference mismatch because student models are trained on answer-conditioned traces but must reason without answer access at inference time. We quantify this mismatch through lexical, trajectory, and probabilistic anchoring, measuring surface overlap, answer-conditioned generation dynamics, and answer recoverability from the trace, respectively. We find that semantic suppression, a seemingly intuitive mitigation, reduces lexical overlap but increases trajectory anchoring: avoiding the answer requires continually tracking it, thereby strengthening its influence on generation. We therefore propose Structural Skeleton-guided Reasoning (SSR), which replaces suppression with structural decoupling by first generating an abstract functional skeleton and then using it to guide the full reasoning trace. Anchoring analyses show that SSR reduces all three forms of answer dependence. Across in-domain and out-of-distribution benchmarks, its distilled variant, SSR-D, improves performance by up to 10 points over suppression baselines and better preserves out-of-distribution performance. Code is available at https://github.com/viniferagy/SSR.

Guangyue Peng, Zongchao Chen, Wen Luo +9

State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University · Nanbeige Lab, BOSS Zhipin · University of Electronic Science and Technology of China

CoT Distillation CoT Reasoning Reasoning Trace Analysis

Feb 5, 2026

BRIDGE: Bridging Reasoning In Distillation Gap Elimination via Structure-Aware Masking

Chain-of-Thought (CoT) reasoning has significantly improved LLMs' mathematical problem-solving capabilities, but distilling such capabilities into smaller models remains challenging due to the capacity mismatch between verbose teachers and compact students. Directly copying teachers' lengthy reasoning chains causes capacity overload, resulting in truncated outputs or repetitive failure. Existing remedies each sacrifice a critical property of CoT: implicit reasoning methods (e.g., compressing reasoning into hidden states) trade away interpretability and verifiability, while heuristic compression strategies (e.g., random step pruning) destroy logical integrity. To address this, we propose BRIDGE, a curriculum framework that first establishes structural understanding via masked reconstruction, then uses GRPO-based reinforcement learning to guide students in self-discovering the optimal balance between accuracy and brevity, and finally internalizes complex reasoning through teacher-guided rewriting on failure cases. On GSM8K, BRIDGE enables Qwen2.5-3B to achieve 11.29% accuracy improvement and 27.4% token reduction over the original model, outperforming instruction-tuned variants and distillation baselines. Zero-shot transfer experiments on SVAMP and MATH-500 further confirm the generalization of internalized reasoning. Our code and model checkpoints are publicly available at https://github.com/Applied-Machine-Learning-Lab/SDM2026_BRIDGE and https://huggingface.co/bowen0815/BRIDGE.

Bowen Yu, Sheng Zhang, Binhao Wang +8

City University of Hong Kong · Mohamed Bin Zayed University of Artificial Intelligence

CoT Distillation Mathematical Reasoning Teacher-Student Learning CoT Compression

Feb 3, 2026

Distilling LLM Reasoning into Graph of Concept Predictors

Deploying Large Language Models (LLMs) for discriminative workloads is often limited by inference latency, compute, and API costs at scale. Active distillation reduces these costs by querying an LLM oracle to train small discriminative students, but most pipelines distill only final labels, discarding intermediate reasoning signals and offering limited diagnostics of what reasoning is missing and where errors arise. We propose Graph of Concept Predictors (GCP), a reasoning-aware active distillation framework in which the teacher's reasoning is elicited as a directed acyclic graph of intermediate concepts and mirrored in the student. GCP enhances sample efficiency through a graph-aware acquisition strategy that weights per-concept uncertainty, gradient diversity, and coverage by node centrality. Additionally, it improves training stability and efficiency by performing targeted sub-module retraining, which attributes downstream loss to specific concept predictors and updates only the most influential modules. Experiments on eight NLP classification benchmarks demonstrate that GCP enhances performance under limited annotation budgets while yielding more interpretable and controllable training dynamics. Code is available at https://github.com/Ziyang-Yu/GCP.

Ziyang Yu, Liang Zhao

Department of Computer Science Emory University Atlanta, USA

CoT Distillation Active Learning Text Classification Language Model Distillation

Dec 24, 2025

Distilling the Essence: Efficient Reasoning Distillation via Sequence Truncation

Distilling the capabilities from a large reasoning model (LRM) to a smaller student model often involves training on substantial amounts of reasoning data. However, knowledge distillation (KD) over lengthy sequences with prompt (P), chain-of-thought (CoT), and answer (A) sections makes the process computationally expensive. In this work, we investigate how the allocation of supervision across different sections (P, CoT, A) affects student performance. Our analysis shows that selective KD over only the CoT tokens can be effective when the prompt and answer information is encompassed by it. Building on this insight, we establish a truncation protocol to quantify computation-quality tradeoffs as a function of sequence length. We observe that beyond a specific length, longer training sequences provide marginal returns for downstream performance but require substantially higher memory and FLOPs. To this end, training on only the first

50\%

of tokens of every training sequence can retain, on average,

\approx91\%

of full-sequence performance on math benchmarks while reducing training time, memory usage, and FLOPs by about

50\%

each. Codes are available at https://github.com/weiruichen01/distilling-the-essence.

Wei-Rui Chen, Vignesh Kothapalli, Ata Fatahibaarzi +5

1The University of British Columbia · 2LinkedIn · 3Canada Research Chair in NLP and ML

CoT Distillation Selective Knowledge Distillation Efficient Language Model Reasoning Efficient Language Model Training Teacher-Student Learning Knowledge Distillation+2

Nov 14, 2025

From Hindsight to Foresight: Self-Encouraged Hindsight Distillation for Knowledge-based Visual Question Answering

Knowledge-based Visual Question Answering (KBVQA) necessitates external knowledge incorporation beyond cross-modal understanding. Existing KBVQA methods either utilize implicit knowledge in multimodal large language models via in-context learning or explicit knowledge via retrieval augmented generation. However, their reasoning processes remain implicit, without explicit multi-step trajectories. To address this gap, we propose a Self-Encouraged Hindsight Distillation Reasoning (HinD) framework, aiming at eliciting reasoning ability inside the MLLM by constructing a Hindsight Teacher with privileged information to teach the Foresight Student. First, we construct the Hindsight Teacher by prompting the MLLM with the reasoning target as privileged information to complete the reasoning process, obtaining Hindsight-Zero training data. Then, the Foresight Student, without knowing the answer, learns the golden trajectories from Hindsight in two ways: (1) Hindsight Distillation Fine-Tuning to self-distill the Hindsight-Zero into a modularized Chain-of-Thought Generator and a Knowledge Generator for sequential steps and discrete facts generation, respectively; (2) Knowledge Encouragement Preference Optimization to encourage the under-confident but relevant knowledge inside the MLLM and suppress the over-confident but irrelevant one. Experiments on OK-VQA and A-OKVQA validate the effectiveness of HinD, showing that HinD with 7-8B MLLM achieves superior performance without commercial model APIs or retrieved knowledge.

Yu Zhao, Ying Zhang, Xuhui Sui +4

College of Computer Science, VCIP, DISSec, Nankai University, Tianjin, China · School of Software, Tiangong University, Tianjin, China · Shenzhen Campus of Sun Yat-sen University, Shenzhen, China +1

Visual Question Answering CoT Distillation Large Vision-Language Models VLM Reasoning Knowledge-Based VQA Knowledge Distillation

Sep 30, 2025

Thinking-Free Policy Initialization Makes Distilled Reasoning Models More Effective and Efficient Reasoners

Reinforcement Learning with Verifiable Reward (RLVR) effectively solves complex tasks but demands extremely long context lengths during training, leading to substantial computational costs. While multi-stage training can partially mitigate this, starting with overly short contexts often causes irreversible performance degradation, ultimately failing to reduce overall training compute significantly. In this paper, we introduce Thinking-Free Policy Initialization (TFPI), a simple yet effective adaptation to RLVR that bridges long Chain-of-Thought (CoT) distillation and standard RLVR. TFPI employs a simple ThinkingFree operation, explicitly discarding the thinking content via a direct </think> append, to reduce token usage during inference. Training with ThinkingFree-adapted inputs improves performance and lowers token consumption, even in the original slow-thinking mode. Extensive experiments across various benchmarks have shown that TFPI accelerates RL convergence, achieves a higher performance ceiling, and yields more token-efficient reasoning models without specialized rewards or complex training designs. With TFPI only, we train a 4B model to reach 89.0% accuracy on AIME24 and 65.5% on LiveCodeBench using less than 4K H20 hours.

Xin Xu, Clive Bai, Kai Yang +4

LLM Department, Tencent · The Hong Kong University of Science and Technology · The University of Hong Kong

Efficient Language Model Reasoning RL for Language Model Reasoning Reinforcement Learning with Verifiable Rewards CoT Distillation Language Model Distillation

Sep 26, 2025

Scale or Reason? A Compute-Equivalent Analysis of Reasoning Distillation

Distilling reasoning traces from strong teacher models has become the standard recipe for building capable small language models. Yet reasoning traces are 5-20

\times

longer than standard instruction fine-tuning (IFT) outputs, meaning every practitioner who chooses reasoning distillation implicitly forgoes training a larger IFT model on the same compute budget. Whether this trade-off is worthwhile remains unaddressed. We study it with a controlled experiment: a single teacher generates paired IFT and reasoning outputs for identical prompts by toggling only its reasoning mode, isolating supervision format as the sole variable. Training students at five scales (0.5B to 14B) and evaluating on 18 benchmarks, we find that at matched FLOPs, IFT lies on or near the Pareto frontier across the majority of configurations. Reasoning reaches the Pareto frontier only on open-ended tasks at 7B and above. Even there, a sequential curriculum mixing just 25-50% reasoning data with IFT captures most of the accuracy benefit at far lower compute cost.

Nicolas Boizard, Hippolyte Gisserot-Boukhlef, Kevin El Haddad +2

1Diabolocom · 5MICS, CentraleSupélec, Université Paris-Saclay · 2Artefact Research Center +2

CoT Distillation LLM Fine-Tuning Efficient Language Model Reasoning Efficient Language Model Training Language Model Distillation Teacher-Student Learning

Aug 13, 2025

Beyond Scaling Law: A Data-Efficient Distillation Framework for Reasoning

Large language models (LLMs) demonstrate remarkable reasoning capabilities in tasks such as algorithmic coding and mathematical problem-solving. Recent methods have improved reasoning through expanded corpus and multistage training combining reinforcement learning and supervised fine-tuning. Although some methods suggest that small but targeted dataset can incentivize reasoning via only distillation, a reasoning scaling laws is still taking shape, increasing computational costs. To address this, we propose a data-efficient distillation framework (DED) that optimizes the Pareto frontier of reasoning distillation. Inspired by the on-policy learning and diverse roll-out strategies of reinforcement learning, the key idea of our approach is threefold: (1) We identify that benchmark scores alone do not determine an effective teacher model. Through comprehensive comparisons of leading reasoning LLMs, we develop a method to select an optimal teacher model. (2) While scaling distillation can enhance reasoning, it often degrades out-of-domain performance. A carefully curated, smaller corpus achieves a balanced trade-off between in-domain and out-of-domain capabilities. (3) Diverse reasoning trajectories encourage the student model to develop robust reasoning skills. We validate our method through evaluations on mathematical reasoning (AIME 2024/2025, MATH-500) and code generation (LiveCodeBench), achieving state-of-the-art results with only 0.8k carefully curated examples, bypassing the need for extensive scaling. Our systematic analysis demonstrates that DED outperforms existing methods by considering factors beyond superficial hardness, token length, or teacher model capability. This work offers a practical and efficient pathway to advanced reasoning while preserving general capabilities.

Xiaojun Wu, Xiaoguang Jiang, Huiyang Li +11

1Zhongxing Telecom Equipment(ZTE), China · 2China Mobile Communications Group Co Ltd

CoT Distillation Language Model Distillation Algorithmic Reasoning LLM Reasoning Data-Efficient Learning