Searcharxiv⌕ Search

method

Knowledge distillation

Knowledge distillation: explore 54 source-linked works published from 2020 to 2026, with original documents and citations.

Search within this collectionFollow this collection with RSS

Coverage and selection

Includes records with this source-supplied label or an explicit phrase match in their metadata. Matches indicate a mention, not proof that a paper uses a method or tests a material. Source versions are consolidated by DOI.

Sources: arxiv. Collection updated 2026-09-16. Counts describe this index, not the complete source archives.

arXiv2026

A Token is Worth over 1,000 Tokens: Efficient Knowledge Distillation through Low-Rank Clone

Training high-performing Small Language Models (SLMs) remains costly, even with knowledge distillation and pruning from larger teacher models. Existing work often faces three key challenges: (1) information loss from hard pruning, (2) inefficient alignment of representations, and (3) underutilization of informative activations, particularly from Feed-Forward Networks (FFNs). To address these challenges, we introduce Low-Rank Clone (LRC), an efficient pre-training method that constructs SLMs aspiring to behavioral equivalence with strong teacher models. LRC trains a set of low-rank projection matrices that jointly enable soft pruning by compressing teacher weights, and activation clone by aligning student activations, including FFN signals, with those of the teacher. This unified design maximizes knowledge transfer while removing the need for explicit alignment modules. Extensive experiments with open-source teachers (e.g., Llama-3.2-3B-Instruct, Qwen2.5-3B/7B-Instruct) show that LRC matches or surpasses state-of-the-art models trained on trillions of tokens--while using only 20B tokens, achieving over 1,000x training efficiency. Our codes and model checkpoints are available at https://github.com/CURRENTF/LowRankClone and https://huggingface.co/collections/JitaiHao/low-rank-clone-lrc-6828389e96a93f1d4219dfaf.

cs.CL↗
arXiv2026

Edit Knowledge, Not Just Facts via Multi-Step Reasoning over Background Stories

Enabling artificial intelligence systems, particularly large language models, to update knowledge and flexibly apply it during reasoning remains a central challenge. Existing knowledge editing approaches emphasize atomic facts, improving factual recall but often failing to integrate updated information into a coherent framework usable across contexts. In this work, we argue that knowledge update is fundamentally a reasoning problem rather than a memorization problem. Consequently, a model should be trained in situations where the new information is instrumental to solving a task, combined with pre-existing knowledge, and exercised through multi-step reasoning. Based on this insight, we propose a training strategy based on three principles. First, new knowledge is introduced as a coherent background story that contextualizes novel facts and explains their relation to existing knowledge. Second, models are trained using self-generated multi-hop questions that require multi-step reasoning involving the new information. Third, training is done using knowledge distillation, forcing a student model to internalize the teacher's reasoning behavior without access to the novel information. Experiments show that models trained with this strategy effectively leverage newly acquired knowledge during reasoning and achieve remarkable performance on challenging questions that require combining multiple new facts.

cs.AI↗
arXiv2026

What Do Students Learn? A Feature-Level Analysis of Dark Knowledge

Knowledge Distillation (KD) is a powerful tool for model compression, yet the precise mechanisms by which student models acquire feature representations remain underexplored. In this work, we analyze student feature learning using the Interaction Tensor framework. Our analysis reveals that effective KD acts as a regularizer that prunes low-frequency, sample-specific features, encouraging the student to rely on a compact set of highly reusable features. Crucially, we observe that the dataset-level confusion matrix contains structural information analogous to the teacher's "Dark Knowledge." Leveraging this insight, we propose Confusion Distillation (CD), a teacher-free self-distillation method that utilizes the model's own evolving )confusion patterns as dynamic soft targets. CD achieves competitive performance on ResNet-34 and ResNet-50 for CIFAR-100, outperforming existing self-distillation methods like CS-KD and PS-KD by 1.2% while offering a computationally efficient alternative to standard KD.

cs.LG↗
arXiv2026

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.

cs.CL↗
arXiv2026

A Closed-Loop Evaluation of Capability Loss and Recovery in Compressed Driving Policies

Many automobile and mobility companies deploy learned driving policies on embedded computers with limited memory and power. Pruning, knowledge distillation, and quantization are the standard methods to reduce the size and the inference cost of these policies. However, these methods are commonly assessed by aggregate numerical scores, and such scores may not reflect the ability of the policy to drive safely when interacting with other road users. In this study, we propose a stage-wise closed-loop evaluation approach to follow a driving policy through a compression pipeline. We formulate the driving task as a partially observable Markov decision process (POMDP) and train a belief-state policy with proximal policy optimization (PPO) in Gym-Duckietown. We then extract the actor, compress it one stage at a time, and evaluate it on five driving curricula. We show that structured pruning is the stage at which the driving capability is first lost. Meanwhile, distillation improves the pruned actor, but the improvement is limited by its rehearsal data. Integer quantization of the improved actor loses some of the curricula that require the vehicle to stop and then resume. Interestingly, the same procedure on the unpruned actor preserves all five curricula. Our study thus provides an empirical analysis aiming to answer the currently active discussions on how to accept a compressed driving policy, so as to achieve a safe and statistically reliable deployment of automated driving functions.

cs.AI↗
arXiv2026

Position Matters: Feature Inversion Attacks in ViT Split Inference with Token Reduction and Shuffling

Vision Transformers (ViTs) are increasingly used in split-inference systems, where edge devices transmit intermediate token representations to a remote cloud. In this setting, token reduction lowers computation and communication costs, while token shuffling disrupts the spatial organization of the transmitted tokens, potentially limiting information leakage. However, their privacy benefits remain unclear against feature inversion attacks, which attempt to reconstruct the input from the transmitted embeddings. In this work, we show that, despite disrupting the spatial structure required by conventional reconstruction attacks, transmitted token embeddings retain substantial positional information. Based on this observation, we introduce the Spatially Aligned Reconstruction Attack (SARA), a unified pipeline that predicts token positions, restores their spatial layout, reconstructs missing embeddings using a feature-space masked autoencoder, and recovers the input image. Our results demonstrate that token shuffling provides only apparent privacy, as SARA largely reconstructs the original token organization. Token reduction offers stronger protection, but significant leakage persists when the retained tokens preserve sufficient semantic and positional information. Finally, we introduce a lightweight edge-side defense that removes positional embeddings and progressively adapts the edge-side transformer blocks through knowledge distillation. It substantially reduces attack performance against SARA, while preserving downstream task accuracy and requiring no changes to the cloud-side model.

cs.CR↗
arXiv2026

Knowledge Distillation During Mid-Training Favors Reasoning over Factual Recall

Logit-based knowledge distillation (KD) is used to train smaller language models (LMs) via supervision from stronger teachers, but whether its benefits are consistent across training stages remains unclear. Through controlled experiments, we find that forward Kullback-Leibler (KL) distillation--the standard KD formulation--with post-trained teachers behaves fundamentally differently during mid-training, an intermediate phase of self-supervised learning on curated corpora. Surprisingly, while forward KD simultaneously improves reasoning and factual recall during pre-training relative to standard next-token prediction (NTP), it instead slows factual recall acquisition during mid-training despite continued reasoning gains. We trace this stage dependence to an asymmetry in teacher confidence across data domains and the student's evolving knowledge state: teachers are more confident on procedural than knowledge-intensive data, while students acquire low-entropy factual knowledge earlier in training. To mitigate this imbalance, we propose Switch Distillation, a simple mid-training objective that distills on tokens where the teacher is confident, using teacher predictive entropy as a lightweight routing signal, and otherwise falls back to cross-entropy. Switch Distillation consistently outperforms existing distillation objectives across teacher sizes. Relative to standard NTP, it achieves 1.61-1.71x the reasoning performance and 1.13-1.19x the knowledge and commonsense performance while preserving 96.7-96.8% of factual recall. Crucially, these benefits persist after post-training: Switch Distillation closes the factual recall gap while maintaining 1.25-1.32x and 1.13-1.20x gains in reasoning and knowledge and commonsense, respectively.

cs.CL↗
arXiv2026

Convergence Theory of Knowledge Distillation in Asynchronous P2P Gossip Learning Network

Decentralized, serverless learning increasingly connects devices running different architectures, where the standard tool, decentralized SGD, is undefined as models with different parameter counts cannot be averaged. Knowledge distillation (KD) exchanges soft predictions rather than weights and sidesteps this obstacle, yet convergence theory for fully decentralized, asynchronous peer-to-peer (P2P) KD is lacking. We provide one, relocating consensus from parameter space to function (output) space: a KD event is a geometric contraction operator in logit space on the peers' predictive distributions, which we analyse in the Hilbert space of predictions on a reference measure. Under standard smoothness/variance assumptions and two realizability assumptions, one bridging parameter SGD to the functional step and one controlling restricted task/KD alignment, the time-averaged functional stationarity and function-space disagreement converge at rate $O(1/(ηT))$ to an $O(η)+O(B_f^2)+O(ζ_f^2)$ neighbourhood. Here $B_f$ is the distance from the task optimum to the peers' reachable classes and $ζ_f$ measures persistent local-task heterogeneity. Across homogeneous, width-heterogeneous, and mixed-family networks of the experiments, KD contracts function disagreement by $40-61\times$, while isolated training does not. The sampled stationarity diagnostic has late transient exponents $0.99-1.90$ on the shared-skeleton main runs, and the four-point step-size sweep exhibits the predicted transient: neighbourhood tradeoff.

cs.LG↗
arXiv2026

Contrastive Knowledge Distillation for Anomaly Detection in Multi-Illumination/Focus Display Images

In this paper, we tackle automatic anomaly detection in multi-illumination and multi-focus display images. The minute defects on the display surface are hard to spot out in RGB images and by a model trained with only normal data. To address this, we propose a novel contrastive learning scheme for knowledge distillation-based anomaly detection. In our framework, Multiresolution Knowledge Distillation (MKD) is adopted as a baseline, which operates by measuring feature similarities between the teacher and student networks. Based on MKD, we propose a novel contrastive learning method, namely Multiresolution Contrastive Distillation (MCD), which does not require positive/negative pairs with an anchor but operates by pulling/pushing the distance between the teacher and student features. Furthermore, we propose the blending module that transforms and aggregate multi-channel information to the three-channel input layer of MCD. Our proposed method significantly outperforms competitive state-of-the-art methods in both AUROC and accuracy metrics on the collected Multi-illumination and Multi-focus display image dataset for Anomaly Detection (MMdAD).

cs.CV↗
arXiv2026

Is Knowledge Distillation Actually Greener? A Case Study in Machine Translation

Knowledge distillation (KD) is a technique to compress a larger teacher system into a smaller student. In machine translation, KD is commonly evaluated through translation quality and inference efficiency, without jointly accounting for the environmental costs of producing and deploying the distilled system. We evaluate representative KD methods both on bespoke MT models and LLMs, by considering both translation quality and computational cost, using the Machine Learning Life Cycle Assessment tool, which accounts for costs throughout the KD model life cycle. Our key finding is that the deployment volume required to amortize KD is serving-dependent and can shift by several orders of magnitude under batching. We include actionable guidance for selecting, developing, and evaluating KD methods under quality and compute-induced constraints.

cs.CL↗
arXiv2026

Domain Mixture Design via Log-Likelihood Differences for Aligning Language Models with a Target Model

Instead of directly distilling a language model, this study addresses the problem of aligning a base model with a target model in distribution by designing the domain mixture of training data for pretraining or continued pretraining as a fixed training recipe. We propose a method for determining domain weights by viewing models as points in log-likelihood space and aligning the training update direction with the direction toward the target model. Experiments with NanoGPT show that the proposed method consistently reduces the KL divergence between the trained base model and the target model relative to training with Pile-original weighting. Although knowledge distillation remains more effective when available, the proposed method achieves meaningful alignment, and downstream task performance also tends to become closer to that of the target model.

cs.CL↗
arXiv2026

Uncertainty-Aware Trajectory Forecasting from Imperfect Tracking

Most trajectory forecasting models are trained on clean annotated histories, and are often evaluated under the same idealized assumption, although practical deployments rely on trajectories produced by imperfect multi-object trackers. The real-world observations exhibit localization jitter, missed or unstable detections, and data-association ambiguity, which are usually either ignored or removed through denoising. This paper instead treats tracking-derived reliability cues as an informative signal to be propagated to the predictor. We propose a plug-in uncertainty-aware formulation in which each observed state is encoded as an uncertain state representation, modeled by a Gaussian distribution whose covariance combines detection-level localization uncertainty and association-level ambiguity through the law of total variance. Existing backbones are adapted with minimal architectural changes: input trajectories are represented as Gaussian observations, and predicted trajectories are produced as Gaussian forecasts rather than deterministic coordinates. To train predictors that remain robust under structured observation noise, we combine temporally correlated Ornstein-Uhlenbeck perturbations with response-based knowledge distillation from a teacher trained on clean trajectories. Experiments on Oxford Town Centre and VIRAT using real tracker outputs, together with a complementary ETH/UCY pseudo-detection protocol, show that the proposed formulation improves displacement accuracy and the reliability-sharpness trade-off of probabilistic forecasts.

cs.CV↗
arXiv2026

CRAD: Class-wise Reliability-Aware Distillation for Decentralized Heterogeneous Federated Learning

Conventional federated learning (FL) relies on parameter averaging, which forces clients to be doubly homogeneous: it demands an identical architecture and degrades under non-IID data. Real-world deployments usually break both assumptions. We sidestep both by building a decentralized knowledge distillation framework in which each client evaluates its peers' model snapshots on its own local data and distills from the resulting soft predictions. Because knowledge is transferred through the shared class posterior, clients are free to run different architectures; and because every teacher is evaluated on the student's own device, raw data never leaves the client, with no central server or public dataset required. Within this setting, we identify and address an under-examined problem: how to combine the peer teacher predictions. Existing methods, like uniform averaging, ignore how knowledge reliability varies across teachers and classes. We propose Class-wise Reliability-Aware Distillation (CRAD), which, per class, first discards teachers that disagree with the peer consensus and then takes a weighted average of the rest, weighting each teacher by its per-class reliability (precision, or inverse variance). Since the variance of an accuracy from $n$ samples scales as $1/n$, support enters automatically: among the teachers that survive filtering, a teacher is trusted for a class to the degree that it is both accurate and well-evidenced for it. On three image-classification benchmarks (CIFAR-10, CIFAR-100, and PathMNIST colon pathology), across heterogeneous architectures under severe non-IID skew, CRAD consistently outperforms competing methods in global accuracy.

cs.LG↗
arXiv2026

Robustness of Vision Language Models Against Split-Image Harmful Input Attacks

Vision-Language Models (VLMs) are now a core part of modern AI. Recent work proposed several visual jailbreak attacks using single/ holistic images. However, contemporary VLMs demonstrate strong robustness against such attacks due to extensive safety alignment through preference optimization, e.g., reinforcement learning from human feedback (RLHF). In this work, we identify a new vulnerability: while VLM pretraining and instruction tuning generalize well to split-image inputs, safety alignment is typically performed only on holistic images and does not account for harmful semantics distributed across multiple image fragments. Consequently, VLMs often fail to detect and reject harmful split-image inputs, in which unsafe cues emerge only upon combining images. We introduce novel split-image visual jailbreak attacks (\textbf{SIVA}) that exploit this misalignment. Unlike prior optimization-based attacks, which exhibit poor black-box transferability due to architectural and prior mismatches across models, our attacks evolve in progressive phases from naive splitting to an adaptive white-box attack, culminating in a black-box transfer attack. Our strongest strategy leverages a novel adversarial knowledge distillation \textbf{(Adv-KD)} algorithm to substantially improve cross-model transferability. Evaluations on four state-of-the-art modern VLMs and three jailbreak datasets demonstrate that our strongest attack achieves up to 44% higher transfer success than existing baselines. Lastly, we propose efficient ways to address this critical vulnerability in the current VLM safety alignment.

cs.CV↗
arXiv2026

ACTD: Anchor-Based Cross-Tokenizer Distillation with Residual Regularization

Knowledge distillation effectively transfers reasoning capabilities from large language models to lightweight student models. To enable knowledge transfer across disparate model families, researchers increasingly explore cross-tokenizer distillation. However, cross-tokenizer distillation remains challenging due to vocabulary and sequence misalignment, while approximate vocabulary alignment can introduce additional noise into distillation. To address these challenges, we propose Anchor-Based Cross-Tokenizer Distillation with Residual Regularization (ACTD). ACTD bridges structural heterogeneity through vocabulary and sequence alignment, while mitigating alignment noise via a novel anchor loss with residual regularization. We further extend this framework to a multi-teacher setting. Evaluated across five reasoning benchmarks with three distinct teacher models, ACTD achieves state-of-the-art performance. Moreover, its multi-teacher extension outperforms the strongest single-teacher and multi-teacher baselines, further demonstrating the robustness of our method.

cs.CL↗
arXiv2026

DISTAL: Distillation and Self-Supervised Pretraining for Structure-Agnostic Materials Property Prediction

Materials property prediction remains difficult in low-data settings, where many target properties are supported by only a limited number of labeled samples. Models with the strongest predictive accuracy often depend on crystal structures, which restricts their use in early-stage screening when structural information is limited or unavailable. To address this challenge, we propose DISTAL, a dual-prior framework for structure-agnostic materials property prediction that combines self-supervised compositional pretraining with structure-aware knowledge distillation. DISTAL first learns transferable compositional representations from a large virtual composition space using 145 composition-derived descriptors. It then distills structural knowledge from a pretrained ALIGNN teacher into a composition-conditioned student. This setting allows structural priors to be used during training without requiring structural inputs at inference. By integrating explicit compositional descriptors, pretrained latent features, and distilled structural features within a unified prediction pipeline, DISTAL captures complementary signals that are difficult to recover from any single representation alone. Across 39 benchmark tasks, the best-performing multimodal configuration combines all three signals, and improves over the reference benchmark on 37 tasks. DISTAL achieves the strongest overall performance among all evaluated feature combinations. These results indicate that compositional pretraining and structural distillation provide complementary priors and offer a practical route to robust composition-only prediction in small-data materials informatics. The source code and the pre-trained models are anonymously available at: https://osf.io/eq96d/overview?view_only=451617f42f7849e08750bd1852b48980 and will be released at the official link after acceptance.

cs.LG↗
arXiv2026

Unleashing the Potential of Mamba: Boosting a LiDAR 3D Sparse Detector by Using Cross-Model Knowledge Distillation

The LiDAR 3D object detector that balances accuracy and speed is crucial for achieving real-time perception in autonomous driving. However, many existing LiDAR detection models rely on complex feature transformations, leading to poor real-time performance and high resource consumption, which limits their practical effectiveness. In this work, we propose a Faster LiDAR 3D object detection framework that Adaptively aligns Sparse voxels to enable efficient heterogeneous knowledge Distillation, called FASD. We aim to distill the Transformer's sequence modeling capability into Mamba models, significantly boosting accuracy through knowledge transfer. Specifically, we first design a cross-model knowledge distillation architecture to convey the global contextual understanding capabilities of the Transformer to Mamba. The Transformer-based teacher model employs a scale-adaptive attention mechanism to enhance multi-scale fusion. In contrast, the Mamba-based student model leverages feature alignment through spatial alignment adapters, supervised with latent-space features and span-head logit distributions, leading to improved performance and efficiency. We evaluated FASD on the Waymo and nuScenes datasets, achieving up to a 2x reduction in FLOPs and a 4x reduction in memory consumption, while improving baseline performance by 1-2 percentage points and maintaining high deployment efficiency.

cs.CV↗
arXiv2026

Personalized Treatment Outcome Prediction from Scarce Data via Dual-Channel Knowledge Distillation and Adaptive Fusion

Personalized treatment outcome prediction based on trial data for small-sample and rare patient groups is a critical task in precision medicine. However, the high cost and scarcity of trial data limit the prediction performance. To address this issue, we propose a cross-fidelity knowledge distillation and adaptive fusion network (CFKD-AFN), which leverages abundant but low-fidelity simulation data to enhance the prediction on scarce but high-fidelity trial data. CFKD-AFN incorporates a dual-channel knowledge distillation module to extract complementary knowledge from the low-fidelity model, along with an attention-guided fusion module to adaptively integrate multi-source information. Experiments on chronic obstructive pulmonary disease show that CFKD-AFN reduces the mean squared error by 6.67% ~ 74.55% and the mean absolute percentage error by 1.43% ~ 51.54% compared to the evaluated competing methods and remains robust to varying high-fidelity dataset sizes. Furthermore, we extend the CFKD-AFN framework to an interpretable variant for exploratory analysis of feature-attribution patterns associated with treatment outcomes.

cs.LG↗
Compare source metadata on this page
WorkPublishedSource identifierSource
A Token is Worth over 1,000 Tokens: Efficient Knowledge Distillation through Low-Rank Clone2026-09-012505.12781arxiv
Edit Knowledge, Not Just Facts via Multi-Step Reasoning over Background Stories2026-09-012602.02028arxiv
What Do Students Learn? A Feature-Level Analysis of Dark Knowledge2026-09-012606.03052arxiv
SPEAR: Distilling Domain-Adaptive Reasoning Skeletons via Sequential Symbolic Alignment in Reinforcement Learning2026-10-102608.26550arxiv
A Closed-Loop Evaluation of Capability Loss and Recovery in Compressed Driving Policies2026-09-012609.00718arxiv
Position Matters: Feature Inversion Attacks in ViT Split Inference with Token Reduction and Shuffling2026-09-172609.01232arxiv
Knowledge Distillation During Mid-Training Favors Reasoning over Factual Recall2026-09-202609.01532arxiv
Convergence Theory of Knowledge Distillation in Asynchronous P2P Gossip Learning Network2026-09-012609.01952arxiv
Contrastive Knowledge Distillation for Anomaly Detection in Multi-Illumination/Focus Display Images2026-09-012609.05520arxiv
Is Knowledge Distillation Actually Greener? A Case Study in Machine Translation2026-08-312602.09691arxiv
Domain Mixture Design via Log-Likelihood Differences for Aligning Language Models with a Target Model2026-08-312603.16622arxiv
Uncertainty-Aware Trajectory Forecasting from Imperfect Tracking2026-08-312608.30899arxiv
CRAD: Class-wise Reliability-Aware Distillation for Decentralized Heterogeneous Federated Learning2026-08-312609.00446arxiv
Robustness of Vision Language Models Against Split-Image Harmful Input Attacks2026-08-302602.08136arxiv
ACTD: Anchor-Based Cross-Tokenizer Distillation with Residual Regularization2026-08-302608.29662arxiv
DISTAL: Distillation and Self-Supervised Pretraining for Structure-Agnostic Materials Property Prediction2026-08-302609.00059arxiv
Unleashing the Potential of Mamba: Boosting a LiDAR 3D Sparse Detector by Using Cross-Model Knowledge Distillation2026-08-292409.11018arxiv
Personalized Treatment Outcome Prediction from Scarce Data via Dual-Channel Knowledge Distillation and Adaptive Fusion2026-08-292510.26444arxiv

These are bibliographic comparisons, not experimental rankings. Follow the original document for methods and conditions.