Knowledge Distillation

Ask about this topic

Overview

Knowledge distillation studies how a student model can learn from a teacher’s outputs, internal representations, examples, or behavior, often to reduce inference cost or transfer capabilities. Common approaches match softened class probabilities, hidden features, attention or relational structure, and use teacher-generated data; newer language-model methods also train on student-generated rollouts. The field spans model compression, cross-architecture and multimodal transfer, and settings such as federated learning, robotics, and low-resource tasks. A central challenge is choosing supervision that fits the student and remains reliable: stronger or denser teacher signals can help, but can also cause negative transfer, forgetting, or loss of capabilities not captured by the benchmark.

Latest papers

CardsList

834 papers164 papers in the last four weeks, up 95% on the four weeks before. 0.9% of all new papers.

Aug 29, 2026

Knowledge Distillation under Teacher Misspecification: An Order-Parameter Analysis of the Gap between Teacher Mimicry and Task Performance

Knowledge distillation trains a small student model to reproduce the outputs of a large teacher model, and its progress is typically monitored through the teacher--student discrepancy. The quantity of ultimate interest, however, is the student's error with respect to the true task. We study the relation between these two objectives in a minimal three-party model, a true teacher (generative model), a teacher, and a student, all soft committee machines, in which the true teacher contains a shared latent factor that the teacher cannot represent, with mismatch strength controlled by a single scalar \dmiss\dmiss. Within an order-parameter description of online distillation, and exploiting closed-form (arcsine-type) expressions for all errors under error-function activations, we prove that the learning dynamics and the distillation error \Ets\Ets are exactly invariant to \dmiss\dmiss, whereas the true error \Etzs\Etzs and the gap Δ=\Etzs−\EtsΔ=\Etzs-\Ets are strictly increasing in \dmiss\dmiss, with a rate that is amplified linearly by the complexity M0M_0 of the true teacher. Numerical phase diagrams over the plane spanned by true-teacher complexity and student capacity confirm the predicted deformation: the contours of \Ets\Ets do not move while the landscape of \Etzs\Etzs rises systematically, and a teacher-miss regime, where mimicry succeeds but the task fails, expands with \dmiss\dmiss. The results give a quantitative warning against evaluating distillation solely through teacher-mimicry metrics and identify the gap ΔΔ as a minimal diagnostic for distinguishing teacher-miss from capacity-limited failure.

Kazuyuki Hara, Hideitsu Hino

Nihon University · The Institute of Statistical Mathematics

Knowledge DistillationModel MisspecificationStatistical Physics of Learning
AskPDF
Aug 29, 2026

Temperature-Adaptive Transformed Teacher Matching

Temperature scaling is a core component of knowledge distillation, yet its role and effect are still not fully understood. Transformed Teacher Matching (TTM) clarifies the role of temperature scaling by applying it only to the teacher distribution and interpreting the resulting objective as standard distillation with an implicit Rényi entropy regularization on the student. However, TTM still relies on a fixed temperature and does not specify how the teacher-side temperature should be adapted for individual samples. In this paper, we introduce a sample-wise inverse-temperature update for TTM by locally minimizing the Kullback-Leibler divergence between the temperature-scaled teacher distribution and the student's prediction. We derive closed-form first and second derivatives with respect to the inverse temperature, and show that they can be expressed using variance and covariance statistics of centered teacher and student logits under the transformed teacher weighting. This yields an efficient curvature-aware update that requires one softmax evaluation and a constant number of class-wise weighted sums. Experiments on standard image classification distillation benchmarks show that our temperature adaptation generally improves TTM and WTTM, while remaining competitive with or outperforming prior temperature-adaptive distillation baselines.

Hiroaki Aizawa, Yoshikazu Hayashi

Graduate School of Advanced Science and Engineering Hiroshima University, Japan · Intelligent Production Technology Research & Development Center for Aerospace Gifu University, Japan

Knowledge DistillationAdaptive Knowledge Distillation
AskPDF
Aug 29, 2026

Explainable Multi-Loss Distillation Framework for Efficient and Interpretable Shrimp Disease Text Classification

Shrimp disease classification has become an urgent issue due to its significant impact on the import-export output of producing countries, particularly Vietnam. Most existing studies focus on image-based classification, which typically operates at the late stage of disease manifestation. Therefore, text-based classification has the potential to enable early and timely disease detection. To address this limitation, we introduce the SALT (Shrimp disease text Analysis with multi-Loss disTillation) framework, which incorporates explainability analysis using Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP) to evaluate model predictions and interpret the learned linguistic features. Experimental results demonstrate that SALT achieves competitive performance across multiple distillation objectives, outperforming supervised baselines while providing a favorable trade-off between predictive performance and computational efficiency. Moreover, it exhibits strong explainability, accurately identifying key linguistic features and semantic patterns relevant to disease descriptions. These findings highlight the potential of knowledge distillation-based text classification for future applications in early shrimp disease diagnosis and related research directions.

Anh Nguyen Quynh, Khang Nguyen Quoc, Luyl-Da Quach

Explainable Text ClassificationText ClassificationKnowledge Distillation
AskPDF
Aug 28, 2026

VISTA: Verifier-Informed Student-to-Teacher Adaptation for On-Policy Self-Distillation

On-policy self-distillation (OPSD) improves reasoning by training a problem-only student on its own rollouts using dense token-level supervision from a privileged teacher that also sees a reference solution. However, standard OPSD treats the teacher distribution as a fixed target along the student's rollout and updates only the student %, although -- even though privileged conditioning does not guarantee that the teacher always provides the most appropriate target for problem-only reasoning. This one-way supervision can therefore misdirect the student when the teacher distribution is misaligned with valid student reasoning. We therefore introduce Verifier-Informed Student-to-Teacher Adaptation (VISTA), which preserves the standard OPSD student update while using outcome-verified rollouts to adapt the teacher toward the student distribution. Within each verified rollout, VISTA further restricts this adaptation to the top-kk positions with the largest teacher--student KL divergence. Notably, VISTA reuses the rollout and loss function from standard OPSD, introducing no additional sampling or separate reward objective. Across AIME24, AIME25, and HMMT25 with Qwen3 models at 1.7B, 4B, and 8B, VISTA achieves the highest Avg@12 at every scale, improving over OPSD by 0.60.6, 0.70.7, and 2.12.1 points, respectively. These results demonstrate the value of student supervision from outcome-verified rollouts and highlight student-to-teacher adaptation as a promising direction for OPSD.

Zewen Ding, Zezhong Wu, Zhou Tao +5

University of Science and Technology of China · State Key Laboratory of Cognitive Intelligence

On-Policy Self-DistillationTeacher-Student LearningKnowledge DistillationMathematical Reasoning
AskPDF
Aug 28, 2026

When Teacher Guidance Misleads: Reward-Aligned On-Policy Distillation

On-policy distillation (OPD) has recently emerged as a popular post-training paradigm for large language models (LLMs), providing an efficient way to transfer the knowledge and capabilities of teacher models into student models. However, teacher guidance on student-generated prefixes is not always reliable. Training should optimize the model to generate responses that are more likely to be correct, or equivalently, to get higher outcome rewards. But during OPD, the teacher model may provide guidance that discourages the student from moving toward correct trajectories or moves the student toward incorrect ones, which is misaligned with outcome reward. Such misaligned guidance is unreliable, as it would mislead the optimization process and ultimately degrade model performance. To mitigate misaligned teacher guidance, we propose Reward-Aligned On-Policy Distillation (RA-OPD). The key insight is to keep only trajectories whose induced updates move the student toward correct trajectories or discourage the student from moving toward incorrect ones. Specifically, for each sampled trajectory, RA-OPD checks whether its trajectory-level distillation return is consistent with its outcome reward and then filters out the misaligned trajectories. RA-OPD selects more reliable trajectories to improve student model performance without requiring additional computational cost. We evaluate RA-OPD on math and code benchmarks using models from the Qwen3 family and the DeepSeek-R1 family. Across seven math benchmarks and three code benchmarks, RA-OPD significantly outperforms standard OPD and other tested OPD variants.

Siyuan Gan, Yuhan Li, Xiran Wang +5

State Key Laboratory of Novel Software Technology, Nanjing University, Nanjing, China · Shanghai Artificial Intelligence Laboratory, Shanghai, China

Language Model DistillationOn-Policy DistillationKnowledge Distillation
AskPDF
Aug 28, 2026

SpikeOPD: Stable On-Policy Distillation for Autoregressive Spiking Language Models

Spiking neural networks (SNNs) offer a path to energy-efficient language modeling through sparse encoding and event-driven computation, but training capable spiking language models from scratch remains difficult. A practical alternative is ANN-to-SNN migration through knowledge distillation (KD), where a pretrained artificial neural network (ANN) teacher supervises an SNN student. Existing migration approaches distill on fixed corpus prefixes, whereas autoregressive inference conditions on self-generated prefixes, creating prefix-source mismatch. It manifests as output-policy mismatch with the ANN teacher and internal spiking-dynamics drift between self-generated and matched corpus prefixes. On-policy distillation (OPD) offers a natural way to mitigate both manifestations by continuing teacher supervision on self-generated prefixes. We evaluate a teacher-only full-KL variant, Vanilla OPD, via a controlled stress test and observe it may suffer from delayed rollout-feedback collapse. This result shows that on-policy coverage alone does not ensure stable adaptation. Motivated by these findings, we propose SpikeOPD, a stable on-policy distillation framework for autoregressive SNNs that learns from self-generated prefixes while maintaining rollout stability. It applies full-KL teacher correction to reduce output-policy mismatch, while matched-prefix policy anchoring constrains policy departure from the frozen reference SNN on the same prefixes. Layerwise spike regularization further limits firing-rate deviations during on-policy adaptation. Across three model scales, SpikeOPD improves average accuracy over the corresponding KD SNNs by 0.8, 1.7, and 2.9 points at 0.125B, 0.35B, and 1.3B, respectively, while preserving their sparse-compute profiles.

Enqiao Lu, Xingrui Yu, Yiwei Fu +7

The Chinese University of Hong Kong, Shenzhen · Agency for Science, Technology and Research (A*STAR), Singapore · Nanyang Technological University, Singapore +3

Spiking Language ModelsSpiking Neural NetworksOn-Policy DistillationKnowledge DistillationLanguage ModelingAutoregressive Language Modeling
AskPDF
Aug 27, 2026

Below the Noise Floor: Bimodal Seed Collapse and Distinct Failure Modes in Small-Model Knowledge Distillation

Function routing -- selecting the correct API call from a fixed catalog given a natural-language request -- is a deployment problem where small students are attractive but knowledge distillation gains are typically reported single-seed, at scales where seed variance is unknown. On a 740-instance healthcare API routing task with a 1.5B Qwen student and a 20B teacher, we compare eight KD variants against supervised cross-entropy, using three to six seeds for key configurations. We find: (i) per-seed standard deviation ranges from 2.8 to 48.7 percentage points, swallowing every claimed KD gain below five points; (ii) three of seven KD variants exhibit bimodal collapse, with at least one in three to five seeds falling below 55% accuracy while the others train normally, and a fourth showing elevated variance; (iii) collapse has distinct modes -- wrong-function selection for ce_kd and ce_paraphrase, and a previously undocumented output-truncation mode for reasoning_kd, where the model emits reasoning but terminates before producing a function name (0.9% accuracy); (iv) only progressive_kd and rank_kd avoid collapse across observed seeds, with sigma <= 3.9 pp; (v) a naive cross-split +3.78 pp gain from input enrichment reverses to -2.70 pp under controlled within-split multi-seed re-testing. Single-seed evaluation is therefore unable to detect central failure modes in small-model KD.

Dipto Sumit, Sakib Ul Haque, Farig Sadeque

Department of Computer Science and Engineering BRAC University, Bangladesh

Knowledge DistillationLanguage Model DistillationTool Use
AskPDF
Aug 27, 2026

Consolidating RLVR Capabilities Across Domains: A Deep Dive into Fusion Paradigms

Reinforcement learning with verifiable rewards (RLVR) improves specific capabilities of large language models, but covering multiple capabilities often involves training separate domain experts and subsequently consolidating them. We organize three fusion paradigms by the artifacts they reuse: Merge combines expert task vectors, Mix RL pools their datasets, and multi-teacher on-policy distillation (MOPD) uses both. Because they have largely been studied in isolation, how they compare and how to choose among them remain unclear. We compare all three using shared experts and data across model scales and a multi-domain benchmark suite. Although their average performance differs by at most 1.4 points, the gap reaches 8.6 points on a single benchmark, with domain-level variation tracking cross-domain relations visible in task-vector geometry. Training dynamics expose distinct constraints: Mix RL depends on domain mixture proportions, MOPD remains bounded by its teachers, and Merge compresses all expert updates into one. All three improve single-sample accuracy without measurable gains in solution coverage or losses in held-out capabilities. These results yield a practical guideline: use Merge when experts already exist and cheap fusion is paramount; Mix RL when training a unified model without experts, with domain proportions adjusted for cross-domain transfer; and MOPD when preserving domain-specific gains matters more than surpassing teachers or minimizing end-to-end cost.

Siye Wu, Kai Yang, Yuchen Cai +8

Fudan University · LLM Department, Tencent

Reinforcement Learning with Verifiable RewardsLanguage Model MergingTask Vector MergingMulti-Task RLKnowledge Distillation
AskPDF
Aug 27, 2026

Video-OPSD: Exploiting Privileged Visual Evidence for On-Policy Self-Distillation in Video Large Language Models

On-policy self-distillation (OPSD) has recently emerged as an effective post-training paradigm that improves policy optimization through dense token-level supervision from a privileged self-teacher. Despite its promise, OPSD remains largely underexplored for Video Large Language Models (Video-LLMs). Existing methods typically construct privileged teachers by augmenting their context with additional information while keeping the primary input unchanged for both teacher and student. Video reasoning, however, offers a distinct source of privileged supervision within the primary input itself: long videos contain substantial temporal redundancy, and only a small subset of frames provides the evidence necessary to answer a question. Building on this observation, we present Video-OPSD\textbf{Video-OPSD}, an OPSD framework that exploits privileged visual evidence for both self-teacher construction and knowledge transfer. First, our Evidence-Grounded Self-Teacher conditions the teacher exclusively on annotated evidence frames while the student continues to reason over the complete video. This focused visual input enables the teacher to provide more informative supervision. Second, our Evidence-Guided Token Optimization adaptively weights token-level distillation according to each reasoning token's reliance on privileged visual evidence, thereby emphasizing perceptually grounded reasoning. Experiments across video understanding and reasoning benchmarks show that Video-OPSD\textbf{Video-OPSD} consistently improves upon Standard OPSD across multiple backbones and achieves performance comparable to GRPO while requiring substantially less training time, establishing an effective and efficient post-training approach for Video-LLMs.

Ziyue Wang, Shiqi Huang, Weiwen Xu +2

School of Electrical and Electronic Engineering, Nanyang Technological University · The Chinese University of Hong Kong

Video-Language ModelsOn-Policy Self-DistillationKnowledge DistillationVideo ReasoningTemporal Video Understanding
AskPDF
Aug 27, 2026

Scaling Model-Generated Distillation Data Can Make Latent Teacher Traits More Recoverable

Scaling model-generated data is usually viewed as improving distillation: more examples should increase coverage, reduce noise, and produce stronger students. We show a second effect: larger datasets can make subtle teacher-specific signals easier to detect in the trained student, even when examples are off-task and never mention the trait. In a controlled setup inspired by subliminal learning, a teacher induced to express a target trait generates restricted off-task data, such as number-only completions. Students trained on different amounts of independent off-task data are evaluated in a separate domain, with matched no-trait controls isolating target-specific transfer. Our main finding is that larger independent datasets make the teacher's induced trait stand out more clearly in the student's later behavior. Other plausible traits may also strengthen with scale, but the target usually grows more. When the small-scale student already favors the target, scaling mainly amplifies that behavior; when it favors a related or salient alternative, more data can shift behavior toward the intended trait. Analyses of learned LoRA updates show a parallel trend. These effects appear across model families, trait types, multi-trait settings, and cross-model transfer. Our results suggest that scaling generated distillation data should be paired with trait-aware curation and evaluation, even when the data appears off-task or benign.

Zhichen Dong, Zhixuan Liu, Yuyu Fan +3

Shanghai Jiao Tong University · Shanghai Artificial Intelligence Laboratory · Fudan University

Language Model DistillationKnowledge Distillation
AskPDF
Aug 27, 2026

Recovering General Capabilities via Uncertainty-Calibrated Multi-Teacher On-Policy Distillation

Specializing large language models to vertical domains improves domain-specific behavior but often degrades general capabilities. We study this trade-off in Multi-Teacher On-Policy Distillation (MOPD), where a specialized model learns from domain and general teachers on its own sampled trajectories. Standard MOPD faces two limitations: ordinary on-policy sampling rarely exposes tokens with large positive teacher--student advantages, and advantage sign alone does not establish whether the proposed update direction is reliable. We propose Uncertainty-Calibrated MOPD (UCMOPD), which addresses these limitations through two complementary mechanisms. Golden-Gain Enhancement combines higher-temperature exploration with a standard-temperature anchor and retains trajectories whose positive learning signal matches or exceeds the prompt-specific anchor. Teacher-Endorsement Filtering then uses centered log-likelihood (CLL) to estimate each retained token's plausibility relative to the teacher's uncertainty and probabilistically preserves updates whose directions are supported by that endorsement. Across role-playing and medical-domain specialization, UCMOPD improves the general-capability average over standard MOPD by 4.48%4.48\% and 7.86%7.86\%, respectively, while maintaining vertical-domain performance. Component ablations and diagnostic analyses support the intended roles of the two mechanisms: exposing and selecting stronger positive signals at the trajectory level and validating update directions through teacher endorsement at the token level.

Ziyuan Liu, Jiao Ou, Jian Liang +2

Kuaishou Technology Beijing, China

Uncertainty-Aware Knowledge DistillationMulti-Teacher Knowledge DistillationOn-Policy DistillationKnowledge Distillation
AskPDF
Aug 27, 2026

SPEAR: Distilling Domain-Adaptive Reasoning Skeletons via Sequential Symbolic Alignment in Reinforcement Learning

Reinforcement learning-based knowledge distillation has the potential to transfer complex reasoning from teacher to student models, yet it currently faces a critical dilemma: researchers must choose between sparse outcome-based rewards, which provide insufficient logical guidance, or expensive neural Process Reward Models (PRMs) for dense signals. We resolve this by introducing SPEAR (Symbolic Process Evaluation and Alignment Reward), a training-free and plug-and-play process reward method for sequence-level on-policy distillation. SPEAR projects natural-language reasoning traces into domain-adaptive symbolic milestones, providing an efficient proxy for process-level reasoning alignment. By utilizing the longest common subsequence (LCS) to align student explorations with teacher milestones, SPEAR provides a dense, order-aware reward signal that aligns logical consistency without the need for an external neural verifier. Our experiments across math, science, and commonsense reasoning tasks demonstrate that SPEAR effectively bridges the reasoning gap between student and teacher models via sequence-level distillation with efficient dense process rewards. Our code and data are available at: https://github.com/zhuochunli/SPEAR.

Zhuochun Li, Yuelyu Ji, Yiming Zeng +1

University of Pittsburgh, Pittsburgh, USA · University of Connecticut, Storrs, USA

Process Reward ModelsRL for Language Model ReasoningOn-Policy DistillationKnowledge Distillation
AskPDF
Aug 26, 2026

DualOPSD: Closing the Feedback Loop in On-Policy Self-Distillation

On-policy self-distillation (OPSD) combines student generated trajectories with dense supervision from a privileged teacher of the same model. Its pointwise clipping protects the student from locally incompatible teacher directions, but the fixed teacher receives no feedback about the supervision the student rejects, which limits effective learning from the teacher. We introduce DualOPSD, which alternates updates in a coupled teacher-student system. The student retains OPSD's clipped forward KL objective. The privileged teacher adapts to the updated student after each student step. A residual target based on the frozen initial model preserves the effect of privileged context during teacher adaptation. Across Qwen3-8B/4B/1.7B and three math benchmarks, DualOPSD improves avg@12 accuracy over OPSD by 9.83 percentage points. Further diagnostic analyses show lower bidirectional KL, less persistent mismatch, and retained sensitivity to privileged reference. Together, these results demonstrate that DualOPSD can be a more effective form of privileged self-distillation. Code is available at: https://github.com/on1262/DualOPSD.

Yutong Chen, Guangfu Guo, Zhichao Xu +1

Department of Computer Science, Clemson University, 105 Sikes Hall, Clemson, SC 29634, USA · Department of Computer Science, University of Utah

Adaptive Knowledge DistillationOn-Policy Self-DistillationLanguage Model DistillationKnowledge Distillation
AskPDF
Aug 26, 2026

Socialized Detector Learning: Trajectory-Guided and Reciprocal Distillation for Heterogeneous Object Detectors

Object detection knowledge is fragmented across independently trained, heterogeneous detectors with complementary category supports. In socialized learning, this knowledge resides in a society, and learning aims to evolve the society collectively through exchange. However, aggregation-based socialization does not explicitly plan transfer order, whereas progressive multi-teacher distillation considers order but remains a one-way student enhancement in a shared category space. Building on Socialized Learning, we formulate Socialized Detector Learning (SDL) for heterogeneous, category-specialized object detectors and propose Trajectory-Guided and Reciprocal Distillation (TGRD).TGRD estimates directed operational Inter-Detector Transfer Difficulty (IDTD) from held-out feature-alignment residuals, precomputes a fixed score table, and greedily constructs a carrier trajectory. Along the trajectory, knowledge is progressively consolidated into a union-category carrier and then returned to experts through reciprocal transfer. A conditional proxy-certificate analysis shows that, under stated assumptions, the progressive certificate is no larger than an aggregated-target counterpart. On MS COCO with four heterogeneous experts and two carrier initializations, final carriers outperform epoch-matched simultaneous aggregation controls by 2.6 AP in both settings. Reciprocal detectors attain 20.8--28.4 AP on previously unsupported categories while remaining within 1.3 AP of original expert-specific performance. These results support order-aware progressive consolidation followed by reciprocal transfer as a viable mechanism for detector-society evolution.

Weihao Li, Yunqi Zhu, Zhihe Fan +5

School of New Media and Communication, Tianjin University, Tianjin, China · School of Computer Science and Engineering, University of New South Wales, Sydney, Australia · School of Sports Training, Tianjin University of Sport, Tianjin, China +5

Object DetectionKnowledge DistillationMulti-Teacher Knowledge Distillation
AskPDF
Aug 26, 2026

Why Does Graph Learning Fail to Fully Benefit from a Text Teacher?

Graph neural networks (GNNs) are widely used to represent complex interactions and relationships among entities. We investigate a multimodal model that combines two complementary ideas: a self-supervised method that enables a GNN encoder pretrained on one dataset to operate directly on another dataset with a different node-feature dimensionality, without rebuilding the model or realigning the data; and an alternating optimization method that updates a language-model module in an E-step and a GNN module in an M-step, rather than jointly training a large language model and a GNN end to end on a large graph. Despite expectations, the combined model did not sufficiently improve predictive performance. We identify six factors: (1) an external anchor in the E-step has a strength-safety trade-off: a weak anchor has little effect, whereas an overly strong anchor can damage the graph representation; (2) the knowledge of the E-step teacher is not injected directly into the GCN embedding Z; (3) the representation space constructed in the M-step is not optimized for the same objective as the E-step teacher space, resulting in a compromise representation for target classification; (4) GCN propagation averages a node's own textual information with information from its neighbors; (5) cosine alignment does not guarantee axes that are discriminative for classification, so stronger geometric alignment with the E-step text anchor need not sufficiently improve the target decision boundary or classification performance; and (6) the force that preserves the source-side self-supervised geometry in the M-step conflicts with the force that moves the representation toward the E-step teacher. We support these observations through a staged set of experiments that varies the influence of the E-step.

Fumiaki Kimino, Ryoma Sato

SOKENDAI · National Institute of Informatics · National Institute of Informatics, SOKENDAI

Graph Representation LearningText-Attributed Graph LearningLLMs for Graph LearningKnowledge DistillationGraph SSL
AskPDF
Aug 26, 2026

A Token-Level Analysis of Sampled-Token Reverse-KL On-Policy Distillation

On-policy distillation (OPD) supervises a student on its own trajectories with token-level signals from a frozen teacher, yet how a sampled loss allocates updates across tokens remains poorly understood. We analyze the gradient of the per-token K2 estimator of reverse KL with respect to the student logits. The ℓ1\ell_1 norm of this gradient factorizes into the absolute teacher--student log-probability gap and a student-side softmax factor that grows as the sampled token becomes less likely under the student. In our math-distillation runs, these per-token norms are highly non-uniform: low-student-probability tokens account for a disproportionate share of their sum and are also enriched in large teacher--student gaps. As a lightweight intervention suggested by this analysis, we study Surprise-aware Reweighting (SuRe), a detached, bounded weighting rule that further amplifies this existing allocation. Across two Qwen3 student scales, SuRe improves several math metrics over vanilla OPD and shows no clear degradation on the selected out-of-domain benchmarks. Our primary contribution is therefore a gradient-level characterization of reverse-KL OPD trained with the K2 estimator, with SuRe as one empirical instantiation.

Bing Shao, Jiazheng Zhang, Long Ma +13

College of Computer Science and Artificial Intelligence, Fudan University

Language Model DistillationOn-Policy DistillationKnowledge DistillationKullback-Leibler Divergence
AskPDF
Aug 25, 2026

OPDSearch+: On-Policy Distillation with RL Refinement for Search-Augmented Reasoning

Search-augmented reasoning remains difficult for small language models. On-policy distillation (OPD) from trained teachers offers a promising direction, but suffers from two issues: (1) high-quality multi-turn search trajectories depend on dynamic retriever responses, making SFT data prohibitively expensive to collect at scale; (2) task-specifically trained teachers incur substantial training cost, while directly applying OPD with an off-the-shelf teacher without task-specific fine-tuning constrains the student to the teacher's performance ceiling and suffers from severe training instability. We propose OPDSearch+, the first distillation paradigm that requires no teacher fine-tuning for search-augmented reasoning. We investigate the role of a frozen off-the-shelf instruct model as the teacher in on-policy distillation, and reveal a key insight: the teacher reshapes the student's policy distribution so that subsequent RL converges to a superior solution that RL alone cannot reach. In stage one, the student interacts with a live search engine and is distilled via a per-position forward KL objective, transferring reasoning decomposition and evidence integration skills without any task-specific teacher training. In stage two, RL refines the distilled student from a richer behavioral foundation, achieving performance that RL alone cannot reach from scratch. Across seven QA benchmarks, OPDSearch+ with a 3B model consistently outperforms all prior 3B RL baselines, achieving gains of 13.1% on HotpotQA and 8.5% on 2WikiMultihopQA.

Qinglin Ye, Zhiyuan Gu, Jingjie Xia +6

University of Chinese Academy of Sciences · Institute of Computing Technology, Chinese Academy of Sciences · Institute of Automation, Chinese Academy of Sciences +5

On-Policy DistillationLLM ReasoningKnowledge DistillationRL for Language Model ReasoningMulti-Hop QA
AskPDF
Aug 24, 2026

PROOF-Gen: From Optimized Data to Better Distillation

Supervised fine-tuning on teacher-generated trajectories is the standard first stage for distilling tool-calling capabilities into deployable models. Post-training pipelines that drive shipped tool-calling agents re-run this stage on a daily or weekly cadence, paying the frontier-teacher cost each cycle, yet the mechanism is generate-and-filter (keep the teacher's passing trajectories, discard the rest) and each cycle leaves behind the same hard scenarios because failures supply no signal. On τ2-bench, 57% of teacher trials fail, two-thirds of them near-misses (most tool calls correct, undone by one decisive error). We introduce PROOF-Gen (Per-scenario Reflective Optimization to Overcome FailedGeneration), which recovers golden trajectories from these failures via per-scenario prompt optimization. For each failed task, a reflector analyzes the execution trace and evaluation feedback, then writes corrective guidance that steers the teacher to a passing trajectory. The guidance is stripped before training, so the student learns from clean demonstrations with no task-specific scaffold. On τ2-bench, per-scenario optimization recovers 93% of failed scenarios. Fine-tuned on the combined data, Qwen3-4B-Instruct-2507 improves from Pass^1=0.132 to 0.529 and Gemma 4 E4B-it gains +7.2pp on BFCL v4 multi-turn. In a deployed pipeline, the method lifts trajectory quality by +6.3pp goal completion and transfers to a deployed on-device model (+1.5pp goal completion; +1.7 to +5.0pp across response-quality metrics), with positive transfer in every locale (non-English average +1.48pp).

Anh Ta, Junjie Zhu, Shahin Shayandeh

Apple

LLM Agent TrainingLLM Agent Trajectory SynthesisKnowledge DistillationTool UsePrompt OptimizationSupervised Fine-Tuning
AskPDF
Aug 24, 2026

Spatiotemporal Distillation via Recurrent Bottlenecks for Aortic Tracking

Cardiac cine-MRI serves as a direct visual indicator of cardiovascular hemodynamics by capturing the continuous wall motion of the aorta. Quantifying these dynamic structural changes across the cardiac cycle is essential for measuring aortic distensibility, a primary marker of arterial stiffness. However, standard 2D segmentation networks focus on each frame independently. Consequently, when rapid systolic flow temporarily obscures the aorta's boundaries, this lack of continuous context results in frame-to-frame tracking dropouts and boundary inconsistencies. Spatiotemporal (2D+t2\text{D}+t) networks can enforce temporal consistency across the sequence but suffer from a scarcity of expert annotations. To address this, we present a semi-supervised spatiotemporal (2D2\text{D} to 2D+t2\text{D}+t) knowledge distillation framework exploiting the cardiac cycle. The framework distills a spatial teacher's expertise into a spatiotemporal student network by executing a dynamic latent interception, pairing a recurrent spatiotemporal bottleneck with a residual spatial bypass. Our model selection strategy applies a baseline validation threshold (DSC≥0.50\text{DSC} \ge 0.50) prior to selecting the epoch that maximizes anatomical consistency. This strategy enables the spatiotemporal student model to achieve superior surface tracking accuracy (NSD@1mm=92.3%±0.2%\text{NSD@1mm} = 92.3\% \pm 0.2\%) and high structural reliability (Frac2CC=99.2%±0.6%\text{Frac}_{2\text{CC}} = 99.2\% \pm 0.6\%), reducing population-wide structural anomalies by over 56% compared to a 2D nnU-Net baseline.

Dexter Wen Jie Teo, Nairouz Shehata, Herve Lombaert

College of Computing and Data Science, Nanyang Technological University, Singapore · Department of Computer and Software Engineering, Polytechnique Montreal, Canada

Medical Image SegmentationCardiac MRISemi-Supervised Medical Image SegmentationKnowledge Distillation
AskPDF
Aug 24, 2026

Too much of a good thing -- when knowledge distillation promotes overfitting, and how to avoid it

The growing size of Convolutional Neural Networks has led to increasingly large and costly models. Knowledge Distillation (KD) addresses this by transferring knowledge from a large network (teacher) to a small one (student), also reducing the training data required. KD is traditionally applied only at the network's final output. However, its behaviour when applied at intermediate network layers has received little attention. This raises the question of whether intermediate block-wise KD, which provides supervision throughout the network, could offer an advantage under specific conditions, such as few instances per class, which is common in fine-grained datasets. This work proposes a student design based on simple, homogeneous blocks mirroring those of the teacher, distilling knowledge between corresponding blocks. Across eleven datasets, we show that on classic datasets, distilling only the last block is sufficient -- and often best--, whereas fine-grained, data-scarce settings benefit substantially from intermediate supervision, with even a single additional distillation point narrowing the gap considerably. We further study how this supervision should be guided, exploring configurations of varying granularity and informed by an explainability analysis based on attention maps, Centered Kernel Alignment, and Grad-CAM, alongside the impact of teacher and student fine-tuning strategies. This work shows that intermediate block-wise distillation, guided appropriately, is key to building compact data-efficient models without sacrificing accuracy.

Irene Trigueros-Lorca, Leonardo Concepción, Christian Wagner +2

Andalusian Research Institute in Data Science and Computational Intelligence, 18016 Granada, Spain · Lab for Uncertainty in Data and Decision Making (LUCID), School of Computer Science, University of Nottingham, NG7 2RD Nottingham, UK · Department of Computer Science and Artificial Intelligence, School of Computer Science and Telecommunications Engineering (ETSIIT), University of Granada, 18071 Granada, Spain

Knowledge DistillationFeature-Based Knowledge DistillationData-Efficient Learning
AskPDF
Aug 24, 2026

Interpretable AI with Local Distillation

Modern AI models such as tabular foundation models and gradient-boosted ensembles can outpredict classical methods, but provide little basis for reasoning about their predictions. High-stakes decisions call for models that are both accurate and interpretable as built. Local linear modeling offers a path forward: a smooth regression function is locally well approximated by a linear one, allowing a linear fit near each query point to achieve high accuracy without sacrificing transparency. The challenges lie in learning what is "local" and developing statistical tools for interpretation. Here, we propose local distillation, in which a black-box "teacher" guides a regularized linear "student" model at each query point. The teacher (1) defines locality by upweighting training observations with similar predicted outcomes, and (2) anchors the fit with its prediction at the query point, included as a pseudo-observation whose weight is estimated from the data. For interpretation, we add a small amount of Gaussian randomization to the local objective and use refits to assess stability: selection frequencies identify reliable features at a query point, and clustering the randomized fits identifies stable subgroups across the data. Under the lasso penalty, we prove that this randomization yields feature-selection probabilities that are stable under small perturbations of the training responses. Across 17 benchmark datasets, local distillation nearly matches its AI teacher's accuracy while producing a sparse linear model at each test point. In a high-dimensional cancer gene expression example, the framework identifies patient subgroups whose local models use different genes; this heterogeneity is invisible to a global linear model, and difficult to surface in a black-box model.

Erin Craig, Yiling Huang, Snigdha Panigrahi

Department of Biostatistics, University of Michigan · Department of Statistics, University of Michigan

Surrogate ModelingInterpretable MLKnowledge Distillation
AskPDF
Aug 24, 2026

Cross-Domain, Multi-Task Data-to-Text Generation without In-Domain Training Data

Structured data exists in many forms (tables, knowledge graphs, charts, and time series), and converting it into text may involve different generation tasks. However, most prior work on data-to-text (D2T) generation has focused on specific tasks and datasets, relying either on task-specific training data or on the zero-shot capabilities of large language models. We study cross-domain D2T generation in a setting where neither in-domain training text nor test references are available, and where domains, generation goals, and input structures vary substantially. We compare data-driven knowledge distillation (DDKD) against zero-shot inference and fine-tuning on out-of-domain D2T data, and introduce structure-preserving augmentation via structural subsampling and perturbation. Experiments on five benchmarks show that, at constant model size (1.7B parameters), DDKD consistently outperforms both fine-tuning and zero-shot inference. Moreover, the resulting small models outperform a much larger finetuned model on two of the five domains, achieving comparable performance on the remaining three. We further construct QUINTD-5, a fivefold extension of QUINTD-1, and show that simply scaling real target-domain inputs yields only modest gains, whereas our augmentation strategy remains more effective and more cost-efficient for cross-domain distillation.

Yifei Song, Kun Efimov-Zhang, Claire Gardent

CNRS/LORIA and Université de Lorraine

Text GenerationCross-Domain Transfer LearningKnowledge DistillationData Augmentation
AskPDF
Aug 24, 2026

SelFusion: Self-distillation for Diffusion Language Models

Diffusion language models (DLMs) alleviate the inherent latency bottleneck of autoregressive (AR) large language models (LLMs), but their degraded generation quality limits practical applicability. Although knowledge distillation (KD) can be a promising direction for improving performance, we empirically find that naively applying conventional KD yields only marginal gains, or even degrades generation quality. Based on these observations, we propose a novel self-distillation framework for DLMs, namely SelFusion. To enable effective KD without an external teacher model, SelFusion performs two forward passes with different masking levels, defining the hard mode with a larger masking probability and the easy mode with a smaller masking probability. However, the easy mode is not always more accurate than the hard mode and can be overconfident on incorrect tokens. Thus, we introduce bidirectional KD between the two modes, which can dynamically determine the distillation direction based on token-level correctness. Experimental results on instruction-following tasks show that the proposed self-distillation substantially outperforms other KD methods with external LLM and DLM teachers. In many configurations, the student trained with SelFusion even surpasses the performance of the LLM teacher, providing a practical path toward improving DLM generation quality. Source code can be found at https://github.com/scai-research/SelFusion_official

Hyeongsoo Lim, Jinyoung Kim, Eunseo Seo +2

Department of Artificial Intelligence, Chung-Ang University

Diffusion Language Model TrainingDiffusion Language ModelsDiffusion Language Model DistillationSelf-DistillationKnowledge Distillation
AskPDF
Aug 24, 2026

Thinking at the Right Size: Amortized Distillation Across Post-Trained LLMs

Practical deployment of large language models (LLMs) requires families of post-trained variants---instruction-tuned, reasoning-tuned, and chat-style models---each at multiple sizes to meet diverse latency and memory budgets. Producing each (variant, size) pair independently is prohibitive, so model families typically span only a handful of coarse-grained sizes per post-trained variant. Boomerang distillation (Kangaslahti et al., 2026) reduces this cost along the size axis for base models. Through model size interpolation, it constructs models of intermediate sizes from a single teacher-student pair without additional training. However, it still treats each post-trained variant as a separate object of optimization. We introduce ADAPT---Amortized Distillation Across Post-Trained LLMs---a framework for amortizing distillation across both axes of a model family: size and post-training variant, producing L×KL \times K models for LL interpolated sizes across KK post-trained variants with a single distillation run. ADAPT combines two components. First, a two-phase distillation procedure constructs post-trained students through pre-training alignment and supervised fine-tuning distillation, enabling smooth size--performance interpolation on generation and reasoning tasks. Second, weight-delta initialization approximates this construction across post-trained variants by transferring the distillation-induced weight change from the base model to students initialized from different post-trained variants. The resulting continuum of interpolated models also enables adaptive model-size selection at inference time, improving the compute--accuracy trade-off for long-form reasoning tasks.

Yan Zhou, Sara Kangaslahti, Jonathan Geuter +4

Harvard University · Kempner Institute · IST Austria

Language Model DistillationKnowledge DistillationLanguage Model Post-Training
AskPDF
Aug 24, 2026

DIAG: Diagnostic Iterative Alignment and Generation for Data-Efficient Mathematical Preference Distillation

Iterative preference optimization is essential for aligning Large Language Models on mathematical reasoning tasks, yet its efficiency is often throttled by signal scarcity: as the model improves, static problem sets become increasingly mismatched to the model's evolving competence, producing rollouts that are either too easy or too hard and therefore non-informative, which leads to a scarcity of valid preference pairs. We propose DIAG, a Diagnostic Iterative Alignment and Generation framework that adaptively reshapes the practice distribution to increase informative supervision and focus training near the student's current competence boundary. DIAG consists of two phases: (1) diagnosing valid preference-pair yield to calibrate the exploration-exploitation trade-off and allocate topic quotas via an Empirical Bayes shrinkage estimator, thereby prioritizing high-yield concepts; and (2) generating targeted practice, where a teacher synthesizes variants from the student's failure traces. We further provide a theoretical view interpreting DIAG as a teacher-mediated approximation to KL-regularized reweighting of the practice distribution toward the student's competence boundary, where valid preference-pair yield is maximized. Experiments show that DIAG boosts yield across iterations and delivers stronger reasoning performance under an iso-effective training budget, demonstrating that it can distill more informative preference supervision for mathematical reasoning.

Guhan Chen, Songtao Tian, Bohan Li +3

Tsinghua University · Kyoto University · Nanjing University

LLM AlignmentPreference OptimizationLLM Mathematical ReasoningKnowledge Distillation
AskPDF
Aug 23, 2026

DiD It in 87 Minutes: A Label-Free Softmax-to-Linear Adaptation of Vision Transformers for Object Detection

While linear attention is a compelling mechanism for high-resolution object detection due to its reduced cost for global token mixing, converting the Softmax-attention ViT backbone of a trained detector into a linear-attention one is not a trivial drop-in replacement. Directly swapping the attention operator leads to severe performance degradation, and generic label-free distillation, though effective for classification, often fails on detection tasks. We argue that the central challenge is detector-interface preservation: the converted backbone must reproduce the exact feature tensors expected by the fixed downstream detector, rather than merely imitating internal Softmax hidden states. To address this, we introduce Detector-Interface Distillation (DiD), a label-free conversion method that exclusively trains the linear-attention backbone by aligning detector-facing interface tensors with those of a frozen Softmax teacher. On DOTA-v1.5, DiD substantially outperforms established baselines and matches supervised, fully trained linear models. Adaptation completes in roughly 87 minutes on 4 GPUs, and the linearized backbone cuts inference latency by ~62% and peak memory by ~49%. We hope our findings offer the community a simple, label-free route to reusing trained Softmax detectors as efficient linear ones, and encourage interface-aware objectives in future architecture-conversion work.

Huaiyuan Qin, Gabriel James Goenawan, Zihang Lin +2

Institute of Advanced Intelligence and Computing (IAIC), A*STAR, Singapore · Nanyang Technological University

Efficient Object DetectionEfficient ViTsKnowledge DistillationLinear Attention
AskPDF
Aug 23, 2026

More accurate behavioral predictions with hybrid Bayesian-connectionist models

Researchers must often choose between Bayesian or neural network models of behavior, two paradigms with complementary strengths and weaknesses. An ideal paradigm would facilitate testing many kinds of representations and inductive biases; Bayesian models make this easy, while neural networks do not. Similarly, an ideal paradigm would avoid over-simplifications; neural networks make this easy, while Bayesian models do not. Here, we introduce Bayesian distillation with Behavioral Tuning (BBT) as an approach to getting the best of both traditions. BBT offers a simple recipe for model building: first, a neural network is trained to mimic a Bayesian model through synthetic data, and second, the network is fine-tuned on human behavior to capture additional structure and nuance. Across four case studies in human concept learning, we find that BBT outperforms traditional approaches at predicting human behavior while also revealing psychological insights, resulting in models that can both mimic Bayesian priors and capture heuristics and biases that violate simple modeling assumptions.

Brenden M. Lake, Akshay K. Jagadish, Guangyuan Jiang

Departments of Computer Science and Psychology, Princeton University · Princeton AI Lab, Princeton University · Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology

Computational Cognitive ModelingHuman Behavior ModelingBayesian Neural NetworksKnowledge DistillationBayesian Inference
AskPDF
Aug 22, 2026

Close Shortcut Wins Long: Seeking Diverse and Stable Generators for Data-Free Knowledge Distillation

Data-Free Knowledge Distillation (DFKD) preserves privacy by transferring knowledge without real data access. However, existing generator-based DFKD methods suffer from over-reliance on teacher preferences and pattern collapse, exhibiting "generative shortcut learning" in the frequency domain: dependent on specific frequency components and frequency positions, resulting in inconsistent synthetic image quality and class diversity. In this paper, we propose a CSWL framework aimed at introducing insights from the frequency domain perspective to improve generator diversity and training stability to Close the phenomenon of Shortcut learning to Win in the Longer term. To address the issue of generative shortcut learning, we introduce frequency-domain augmentation at the feature level, encouraging the generator to attend to the full frequency spectrum and thereby suppress shortcut learning behavior. To tackle training instability, we propose a Cross-Stage Frequency Reconstruction (CSFR) auxiliary task, which implicitly constructs an Exponential Moving Average (EMA) mechanism to promote long-term optimization and stability. Extensive experiments, including downstream tasks and various image recognition datasets at multiple resolutions, validate the effectiveness of CSWL in improving both diversity and stability from the frequency view.

Kailin Lyu, Zherui Zhang, Junhao Dong +11

Institute of Automation, Chinese Academy of Sciences, China · Zhongguancun Academy, China

Data-Free Knowledge DistillationKnowledge Distillation
AskPDF
Aug 21, 2026

STAR-OPD: Structured Aspect-Cascade-Aware On-Policy Reward Distillation for ABSA Quadruple Extraction

Aspect-based sentiment analysis (ABSA) quadruple extraction requires jointly predicting target, aspect, opinion, and sentiment over reviews that often contain multiple fine-grained sentiment tuples. While large chain-of-thought (CoT) models perform well on this task, distilling them into smaller deployable models remains difficult. We identify a task-specific failure mode in distilled ABSA extraction: student errors at the target-aspect interface create structurally invalid states, such as broken target-aspect bindings and hallucinated targets, which then corrupt downstream predictions. Conventional off-policy distillation is poorly suited to this setting because it trains only on teacher-generated trajectories and provides little supervision on the student-induced structural states that dominate inference. To address this mismatch, we propose STAR-OPD (STructured Aspect-cascade-aware On-Policy Reward Distillation), which builds on generic on-policy distillation and instantiates it for ABSA quadruple extraction with cascade-aware, set-structured rewards. STAR-OPD trains on student rollouts and applies set-structured rewards that directly target binding consistency, target grounding, and fine-grained aspect disambiguation. Experiments on E-ABSA20K and SemEval-2014 show that STAR-OPD consistently outperforms off-policy and general on-policy baselines, reduces target hallucination, and substantially improves performance on structurally hard cases. With Qwen3-4B, STAR-OPD substantially narrows the student-teacher gap while improving inference efficiency, highlighting the importance of on-policy structural correction for distilled ABSA extraction.

Tong Sun, Mingyang Ma, Jiayang Yu

Alibaba International Digital Commerce Group

RL for Knowledge DistillationAspect-Based Sentiment AnalysisKnowledge DistillationOn-Policy Distillation
AskPDF
Aug 21, 2026

ForeTime-VLA: Causal Future-Token Distillation from a World Action Model for Conveyor-Belt Manipulation

Manipulating moving objects requires a policy to anticipate contact events, yet vision-language-action (VLA) policies are commonly fine-tuned from the current observation alone. World action models (WAMs) learn predictive dynamics, but running a video-scale teacher or explicitly imagining future frames at deployment is costly. We introduce ForeTime-VLA, a dense pi0.5 policy that distills a future-aware, action-equivalent representation from a frozen Fast-WAM-derived teacher while remaining causal at inference. Offline, current and future video latents are compressed into a whitened 64-D target. Online, an eight-frame history encoder predicts this target together with manipulation phase and normalized time-to-transition. Four future tokens and one phase token condition the VLM prefix, while the predicted future and transition horizon condition the action expert. Training retains the original flow-matching action target and adds cosine, relational geometry, phase, time-to-transition, and action-equivalence objectives. On a deduplicated conveyor-belt dataset, we compare 40k-step checkpoints on 768 matched windows per split. Test MAE decreases from 0.134119 to 0.130593 (2.63%; paired-bootstrap 95% CI: 0.82-4.48% improvement), and test L2 decreases by 3.02%, at a 2.46-2.93% latency cost. In quantitative real-robot evaluation, ForeTime-VLA achieves 81.1% stationary and 58.9% slow-moving grasp success, exceeding the next-best reference by 12.2 and 22.2 percentage points, respectively. Across three belt speeds, it completes 44/90 grasps versus 23/90 for pi0.5, including 11/30 versus 2/30 at fast speed. The agreement between offline orientation gains and reduced real-robot contact-pose failures supports causal future-token distillation as an effective way to improve dynamic manipulation without deploying the world-model teacher.

Siyuan Ma, Yutian Zhang, Boshi Zhang +4

Tsinghua University, Beijing, China · Shanghai Artificial Intelligence Laboratory, Shanghai, China · Harbin Institute of Technology, Harbin, China +1

Visuomotor Policy LearningVision-Language-Action ModelsKnowledge Distillation
AskPDF