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

How to Backdoor Image Knowledge Distillation

Knowledge distillation is widely used to transfer behavior from a large teacher model to a smaller student. It is often assumed to be safe when the teacher is clean, because classic backdoor attacks rely on poisoned labels and triggers in supervised training, whereas distillation trains the student to match a teacher's outputs. We show that this assumption can fail when the distillation dataset itself is poisoned. Our attack injects triggered and manipulated images that a clean teacher already predicts as an attacker chosen target label, which causes the student to learn a backdoor even though the teacher remains unaffected. We evaluate this threat across multiple manipulation strategies, including targeted adversarial perturbations and targeted GAN based class transitions, and study how distillation settings influence both accuracy and attack success. We show that the attack remains effective even at a 10% poisoning rates. The results demonstrate that a clean teacher alone is not a sufficient safeguard: poisoned distillation data can produce a strongly backdoored student while maintaining competitive performance on clean images. These findings show that the integrity and provenance of distillation data are part of the security boundary of data intensive KD pipelines, even when the teacher itself is trusted.

cs.CR↗
arXiv2026

KDFlow: A User-Friendly and Efficient Knowledge Distillation Framework for Large Language Models

Knowledge distillation (KD) is widely used to compress and post-train large language models (LLMs), yet many existing frameworks execute teacher inference with the same training-oriented backend as student optimization, leading to suboptimal efficiency. In this paper, we propose KDFlow, a novel framework for LLM distillation that features a decoupled architecture and employs SGLang for teacher inference. KDFlow combines SGLang for teacher inference with PyTorch FSDP2 for student optimization, allowing each model to run on a backend tailored to its workload. To enable efficient full-vocabulary distillation in this decoupled architecture, KDFlow transfers the teacher's final hidden states via Ray's object store and recomputes teacher logits on each student worker using a frozen copy of the teacher's output head. Furthermore, our framework supports both off-policy and on-policy distillation and incorporates cross-tokenizer algorithms through highly extensible and user-friendly APIs. Experiments show that KDFlow achieves a 1.44$\times$ to 6.36$\times$ speedup over MS-SWIFT in off-policy distillation and a 1.43$\times$ to 1.75$\times$ speedup over verl in on-policy distillation. KDFlow further scales to 64 GPUs, achieving 3.68$\times$ and 2.52$\times$ strong-scaling speedups in two representative model configurations. The code and documentation are publicly available.

cs.CL↗
arXiv2026

PatchFormer: A Patch-Based Time Series Foundation Model with Hierarchical Masked Reconstruction and Cross-Domain Transfer Learning for Zero-Shot Multi-Horizon Forecasting

Time series forecasting is a fundamental problem with applications in climate, energy, healthcare, and finance. Many existing approaches require domain-specific feature engineering and substantial labeled data for each task. We introduce PatchFormer, a patch-based time series foundation model that uses hierarchical masked reconstruction for self-supervised pretraining and lightweight adapters for efficient transfer. PatchFormer segments time series into patches and learns multiscale temporal representations with learnable aggregation across temporal scales. Pretraining uses masked patch reconstruction with dynamic masking and objectives that encourage both local accuracy and global consistency, followed by cross-domain knowledge distillation. Experiments on 24 benchmark datasets spanning weather, energy, traffic, finance, and healthcare demonstrate state-of-the-art zero-shot multi-horizon forecasting, reducing mean squared error by 27.3 percent relative to strong baselines while requiring 94 percent less task-specific training data. The model exhibits near log-linear scaling with more pretraining data up to 100 billion points and processes length-512 sequences 3.8x faster than full-sequence transformers.

cs.LG↗
arXiv2026

Continual Distillation Learning for Rehearsal-Free Class-Incremental Learning via Decoupled Prompting

Prompt-based continual learning has shown strong performance in rehearsal-free class-incremental learning by adapting learnable prompts while freezing a pre-trained Vision Transformer (ViT) backbone. However, the effect of backbone scale remains underexplored. We observe that larger ViT backbones consistently yield better continual learning performance, which motivates us to study how to transfer such capability from a larger model to a smaller one. In this paper, we introduce Continual Distillation Learning (CDL), a new setting for knowledge distillation in rehearsal-free prompt-based continual learning. We show that conventional distillation methods provide only limited gains in CDL, mainly because task-specific prompts are forced to encode both continual adaptation and distillation knowledge, while lacking a persistent mechanism for cross-task knowledge transfer. To address this problem, we propose Decoupled Continual Distillation Learning (D-CDL), which introduces persistent Knowledge-Distillation prompts (KD-prompts) and a dedicated KD branch to explicitly decouple distillation from task adaptation. The proposed KD-prompts are propagated across tasks as a global carrier of teacher knowledge, while the original prompts remain responsible for continual learning. D-CDL is simple, general, and can be integrated into various prompt-based continual learning frameworks. Extensive experiments on Split CIFAR-100 and Split ImageNet-R across four representative continual learning methods show that D-CDL consistently outperforms existing distillation baselines and substantially improves student performance under different teacher-student settings.

cs.CV↗
arXiv2026

Ensemble-Guided Distillation for Compact and Robust Acoustic Scene Classification on Edge Devices

We present a compact, quantization-ready acoustic scene classification (ASC) framework that couples an efficient student network with a learned teacher ensemble and knowledge distillation. The student backbone uses stacked depthwise-separable "expand-depthwise-project" blocks with global response normalization to stabilize training and improve robustness to device and noise variability, while a global pooling head yields class logits for efficient edge inference. To inject richer inductive bias, we assemble a diverse set of teacher models and learn two complementary fusion heads: z1, which predicts per-teacher mixture weights using a student-style backbone, and z2, a lightweight MLP that performs per-class logit fusion. The student is distilled from the ensemble via temperature-scaled soft targets combined with hard labels, enabling it to approximate the ensemble's decision geometry with a single compact model. Evaluated on the TAU Urban Acoustic Scenes 2022 Mobile benchmark, our approach achieves state-of-the-art (SOTA) results on the TAU dataset under matched edge-deployment constraints, demonstrating strong performance and practicality for mobile ASC.

cs.SD↗
arXiv2026

Distilling Vision-Language Models for On-Device Fire Understanding

Vision-language models (VLMs) offer a promising alternative to conventional fire detection systems by reasoning about the semantic context of a scene and thus reducing false alarms, yet their large model size makes deployment on embedded fire sensors impractical. In this paper, we study how domain-specialized VLMs can be compressed for fully on-device deployment without losing the safety-critical behavior required for fire detection. We develop a teacher-student knowledge distillation framework in which large VLMs fine-tuned for fire understanding can be distilled into lightweight students. Experiments across multiple VLM families and model scales show that compact students preserve most of their teachers' fire-understanding capability. We further deploy the distilled models on our commercial Detectium fire detection sensor and jointly evaluate reasoning accuracy, latency, and memory usage. The results show that compression and deployment affect not only accuracy but also model failure modes, with Qwen2.5-0.5B providing the strongest overall deployment trade-off. Our findings provide broader guidance for deploying domain-specialized VLMs in resource-constrained, safety-critical settings.

cs.AI↗
arXiv2026

The Struggle Between Continuation and Refusal: A Mechanistic Analysis of the Continuation-Triggered Jailbreak in LLMs

With the rapid advancement of large language models (LLMs), the safety of LLMs has become a critical concern. Despite significant efforts in safety alignment, current LLMs remain vulnerable to jailbreaking attacks. However, the root causes of such vulnerabilities are still poorly understood, necessitating a rigorous investigation into jailbreak mechanisms across both academic and industrial communities. In this work, we focus on a continuation-triggered jailbreak phenomenon, whereby simply relocating a continuation-triggered instruction suffix can substantially increase jailbreak success rates. To uncover the intrinsic mechanisms of this phenomenon, we conduct a comprehensive mechanistic interpretability analysis at the level of attention heads. Through causal interventions and activation scaling, we show that this jailbreak behavior primarily arises from an inherent competition between the model's intrinsic continuation drive and the safety defenses acquired through alignment training. Furthermore, we perform a detailed behavioral analysis of the identified safety-critical attention heads, revealing notable differences in the behaviors of safety heads across different model architectures. Grounded in these mechanistic findings, we propose Head Competition Steering (HCS), a mechanistically grounded inference-time strategy that explicitly leverages the competition between safety heads and continuation heads to suppress harmful generation, and further distill its behavioral signal into a student model via knowledge distillation, achieving inference-time safety improvements without additional computational overhead.

cs.AI↗
arXiv2026

Importance-Aware Low-Rank Distillation of Diffusion Transformers

Diffusion Transformers (DiTs) have emerged as a dominant architecture for high-quality text-to-image generation, yet their scale poses challenges for efficient deployment. While truncated singular value decomposition (SVD) is a principled tool for parameter reduction, evidence from large language models (LLMs) suggests that naive low-rank approximation can cause catastrophic failure. In contrast, we find that truncated SVD in DiTs produces smooth degradation even under substantial global compression, with redundancy distributed across projection matrices throughout the whole network rather than concentrated in a few transformer blocks. Building on these insights, we introduce SVDtrunc, a two-step block-level compression scheme, first allocating ranks across blocks and compressing the least important ones via truncated SVD under a global parameter budget, and then fine-tuning all blocks with modular knowledge distillation and a rectified-flow objective. We apply SVDtrunc to FLUX.dev across compression levels ranging from 40-90% of the original parameter count. Across three benchmarks, GenEval, HPSv2, and DPG, we outperform all competing approaches. Notably, and in contrast to prior work, we retain near-full performance at 68% and remain competitive even at 57% of the original parameter budget. Furthermore, we show that SVDtrunc complements step distillation and achieves strong results even without fine-tuning, positioning it as a practical continuation of efficiency improvements beyond diffusion step reduction for large-scale generative models. Project page: https://vislearn.github.io/SVDtrunc/

cs.CV↗
arXiv2026

Persistent Teacher Anchoring for Tool-Using Agents

Distillation is common in LLM post-training, where on-policy knowledge distillation (OPKD) uses student-generated trajectories to prepare the student for downstream RL. At each state, the student matches a next-token distribution supplied by the teacher. As the rollout enters states the teacher would not visit, the teacher-student distribution gap can accumulate. In tool use, this gap becomes consequential because student-written calls execute before supervision and their observations shape later prefixes. Proposer-verifier generation addresses this drift by letting the teacher decide which student-proposed text is retained during generation. Existing formulations govern text but leave tool execution outside their scope. We propose Persistent Teacher Anchoring (PTA), a student-induced but teacher-committed rollout construction. PTA retains chunk-level verification and adds turn-level commitment, allowing a call to reach the environment only after the teacher has verified the entire turn. Treating verified chunks as atomic generation units, we introduce persistent lookahead, which fills idle rollout capacity by advancing future samples and carrying unfinished ones across student updates under the fixed verifier. Across Search-R1-style retrieval and DeepEyes-style perception RL, applying PTA before downstream RL improves macro best@4 by 2.5 and 2.8 points over OPKD under the same downstream RL budget, while lookahead improves throughput by 24%.

cs.LG↗
arXiv2026

Lightweight Vision Transformer Compression for On-Device Plant Disease Detection in Resource-Constrained Agricultural Field Conditions

Chilli (Capsicum annuum) is one of India's most economically significant crops, yet its productivity is persistently threatened by diseases that are difficult to identify without expert intervention. While Vision Transformers (ViTs) have achieved high classification accuracy, their large computational footprint makes deployment on resource constrained devices challenging. Existing compression approaches typically address pruning, quantization, and knowledge distillation in isolation, leaving the potential benefits and interactions of their combined application insufficiently explored. We propose a unified Vision Transformer compression framework that combines Hessian-Balanced Adaptive Block Pruning (H-BAC), guided by second-order sensitivity estimation, with quantization and attention-based knowledge distillation. To systematically identify the most effective configuration within each compression family, each technique is first evaluated independently through controlled ablation studies, after which the best-performing components are integrated into a sequential deployment pipeline tailored to real-world agricultural constraints. On a chilli 3-class village-split dataset with a genuine cross-village, cross-device out-of-distribution test split, the resulting compressed models match or exceed the 95.13% FP32 baseline's accuracy, alongside 74-98% model size reduction, and the fully integrated compression pipeline achieves a 54.5x size reduction (327.42 MB to 6.01 MB) at 95.13 +/- 2.32% accuracy across four tested configurations. A direct comparison further reveals that, on this dataset, a directly-trained student of the same final size, without pruning or distillation, reaches comparable accuracy of 94.87%, at the same 6.01 MB INT8 size, indicating where H-BAC and knowledge distillation are, and are not yet shown to be, worth their computational cost.

cs.CV↗
arXiv2026

CrisisKD: Five-Stage Knowledge Distillation for Aspect-Level Sentiment and Emotion Analysis in Crisis Discourse

Identifying the target of emotional words or phrases in crisis situations, especially health-related ones, is important for understanding public concerns across cultural and linguistic contexts. We propose CrisisKD, a five-stage teacher--student knowledge distillation framework for aspect-level sentiment and emotion analysis on unannotated social media data. A teacher LLM generates aspect-level labels and reasoning traces that supervise a smaller student model across aspect extraction, syntactic parsing, opinion extraction, sentiment classification, and emotion classification. Using this framework, we construct and release a dataset containing 50,615 aspect-level labels, together with the annotation and fine-tuning scripts as open-source resources. The resulting student supports end-to-end ABSA and emotion detection at substantially lower inference cost than the teacher. On a manually annotated 500-tweet gold set, the 5-task Qwen2.5-7B student improves over the untuned model by 7.9 F1 points on aspect extraction, 17.0 points on emotion accuracy, and 6.5 points on sentiment accuracy. On the external ABEA benchmark, CrisisKD improves the same-model Qwen2.5-7B ICL baseline by 2.8 F1 points on ATE and 3.8 F1 points on joint ATE+AEC.

cs.CL↗
arXiv2026

ALRA: Adaptive Local Relational Alignment for Logit-Based Pre-training Distillation of Autoregressive Language Models

Logit-based knowledge distillation for autoregressive language models usually aligns teacher and student next-token distributions over the entire vocabulary. However, this global objective overlooks relative preferences among likely token alternatives. Existing local approaches often select candidate tokens from either the teacher or the student alone. Teacher-only selection can miss tokens that the student considers likely, while student-only selection can rely on an inaccurate ranking early in training. We propose Adaptive Local Relational Alignment (ALRA), a position-specific framework combining student proposals with teacher guidance. At each valid prediction position, the student proposes likely tokens, while the teacher's most probable token is included as an anchor. ALRA adjusts the number of selected tokens according to how broadly the teacher distributes probability within this candidate set relative to the current batch. Adaptive Local Divergence retains the mass-matching term and separately matches the relative token distributions within the selected and remaining vocabulary regions. Unlike the exact full-vocabulary decomposition, it replaces the teacher-mass coefficients of the two conditional terms with unit coefficients, preventing either term from being downweighted solely because its region has low teacher probability. Student-Weighted Pairwise Relational Alignment emphasizes high-probability token pairs with small student probability gaps and gives less weight to unlikely or clearly separated pairs. Experiments on The Pile with randomly initialized 200M- and 500M-parameter students across nine zero-shot benchmarks yield average accuracies of 36.62% and 37.40%. ALRA exceeds the strongest competing distillation baseline by 0.94 and 0.83 percentage points and improves over pre-training without distillation by 2.31 and 2.91 points, respectively.

stat.ML↗
arXiv2026

Discriminative and Consistent Representation Distillation

Knowledge Distillation (KD) transfers knowledge from a large teacher to a smaller student model. While contrastive objectives have proven effective for learning structured representations in self-supervised settings, their use in distillation is hindered by two practical shortcomings: the reliance on external memory banks for negative sampling, and fixed temperature hyperparameters that limit adaptability across training stages and teacher-student pairs. We therefore propose Discriminative and Consistent Representation Distillation (DCD), which combines contrastive instance discrimination with a consistency regularization term over the cross-model similarity matrix. The contrastive term aligns each student representation with its teacher counterpart, while the consistency term penalizes asymmetry between the row-normalized and column-normalized views of that matrix, constraining the off-diagonal structure that instance discrimination alone leaves free; we show that it vanishes precisely when this matrix is symmetric. We further introduce an efficient in-batch sampling that eliminates external memory banks, and learnable scale and bias parameters that adapt during training to control the sharpness and offset of the distillation signal. The method matches the training speed of standard KD while adding only 66K additional parameters. Through extensive experiments on CIFAR-100, ImageNet, and MS-COCO, together with cross-dataset transfer to STL-10 and Tiny ImageNet, we show that our approach achieves competitive performance in classification, object detection, and transfer, while substantially reducing memory consumption and training time compared to existing contrastive distillation methods.

cs.CV↗
arXiv2026

Uniformity First: Uniformity-aware Test-time Adaptation of Vision-language Models against Image Corruption

Pre-trained vision-language models, such as contrastive language-image pre-training (CLIP), have demonstrated a remarkable generalizability, enabling a wide range of applications, including zero-shot classification. However, vision-language models still struggle to handle distribution shifts, where input samples have large gaps from training ones. We found that CLIP is especially vulnerable to image corruption, a type of realistic distribution shift caused by sensor conditions such as weather, light, or noise. Collecting a new dataset from a test distribution for fine-tuning is highly costly since image corruption occurs unexpectedly and has a wide variety of types. Thus, we investigate test-time adaptation (TTA) of zero-shot classification, which enables on-the-fly adaptation to the test distribution with unlabeled test data. Existing TTA methods for CLIP mainly focus on modifying image and text embeddings or predictions to address distribution shifts. Although these methods can adapt to domain shifts, such as out-of-distribution or different renditions in input images, they fail to adapt to distribution shifts beyond domain shifts, e.g., image corruption. We found that uniformity of image embeddings, which is related to the amount of information, is a key factor that differentiates domain shifts and other distribution shifts. To enable adaptation to image corruption, we propose a novel method called uniformity-aware information-balanced TTA (UnInfo). To address distribution shifts, we introduce uniformity-aware confidence maximization, information-aware loss balancing, and knowledge distillation from the exponential moving average (EMA) teacher. Through experiments, we demonstrate that our UnInfo improves accuracy under image corruption by retaining information in terms of uniformity. The code is available at https://github.com/kzkadc/uninfo.

cs.CV↗
arXiv2026

Breaking the Geometric Bottleneck: Contrastive Expansion in Asymmetric Cross-Modal Distillation

Knowledge distillation between asymmetric architectures often induces severe geometric constraints on the learned representation space. We investigate dimensional collapse when distilling global Vision Transformers into capacity-constrained, local-receptive-field CNNs (0.5M-8.0M parameters). Using strictly centered SVD and Shannon Entropy Effective Rank, we confirm capacity-agnostic collapse under cosine distillation: a CLIP ViT-B/32 Teacher exhibits Effective Rank 88.68 on CIFAR-10, while all cosine-distilled students collapse to ~17 regardless of parameter count. An auxiliary InfoNCE objective expands this to ~41 dimensions. Critically, we ask whether this expansion is functionally useful. Multi-seed linear-probe evaluation shows InfoNCE expansion degrades downstream accuracy by 15-18 points relative to the collapsed baseline, despite more than doubling Effective Rank. A class-structure decomposition traces this to signal dilution: InfoNCE's class-blind uniformity pressure weakens class-discriminative structure in the original dimensions while adding only weakly relevant structure elsewhere. We then test a label-aware alternative, Supervised Contrastive distillation. On CIFAR-100, where we swept student capacity directly, its Effective Rank is invariant to capacity; on CIFAR-10, at a single tested width, it settles to a lower rank while matching baseline accuracy. Sweeping temperature instead of capacity, rank and downstream accuracy increase together monotonically. These results show Effective Rank alone is not a reliable proxy for representation quality: whether expansion helps or harms downstream performance depends on whether the driving objective is label-aware, not the magnitude of expansion itself.

cs.CV↗
arXiv2026

ProgResViT: Progressive Resolution and Width for Adaptive Vision Transformers

Vision Transformers (ViTs) typically process every image using a fixed input resolution and model width, even though many images can be classified with substantially less computation. We introduce ProgResViT, an input-adaptive ViT that performs inference progressively across multiple rounds. The first round processes a low-resolution image with a narrow subnetwork. Inference terminates when the prediction is sufficiently confident; otherwise, the model reuses the representations produced in the current round and proceeds with a higher-resolution input and a wider subnetwork to refine its prediction. As all rounds share a single backbone, we propose Progress-Conditioned Soft Gating (PSG), which conditions token fusion and layer outputs on the current round, block, and input resolution. On image classification, applying ProgResViT to DeiT yields better accuracy-compute trade-offs than adaptive-width, adaptive-depth, and dynamic-token baselines. With knowledge distillation, a DeiT-based ProgResViT achieves 84.9% top-1 accuracy, slightly exceeding the reported DeiT-III-S accuracy under a comparable evaluation setting. We show that the same design also provides favorable accuracy-compute trade-offs for self-supervised DINO representations and downstream semantic segmentation. Code is available at https://github.com/ds-kiel/ProgResViT.

cs.CV↗
arXiv2026

Relational Representation Distillation

Knowledge distillation transfers knowledge from large teacher models to more compact student networks. The standard approach minimizes the Kullback-Leibler (KL) divergence between the probabilistic outputs of the teacher and student, aligning predictions but neglecting the structural relationships encoded within the teacher's internal representations. Recent advances have adopted contrastive learning objectives to address this limitation; however, such instance-discrimination-based methods induce a "class collision problem", in which semantically related samples are inappropriately pushed apart despite belonging to similar classes. To overcome this, we propose Relational Representation Distillation (RRD) that preserves the relative relationships among instances rather than enforcing absolute separation. Our method introduces separate temperature parameters for teacher and student distributions, with a sharper teacher (low $τ_t$) emphasizing primary relationships and a softer student (high $τ_s$) maintaining secondary similarities. This dual-temperature formulation creates an implicit information bottleneck that preserves fine-grained relational structure while avoiding the over-separation characteristic of contrastive losses. We establish theoretical connections showing that InfoNCE emerges as a limiting case of our objective when $τ_t \rightarrow 0$, and empirically demonstrate that this relaxed formulation yields superior relational alignment and generalization across classification and detection tasks.

cs.CV↗
arXiv2026

SARTM: Segment Any RGB Thermal Model with Language aided Distillation

The recent Segment Anything Model (SAM) demonstrates strong instance segmentation performance across various downstream tasks. However, SAM is trained solely on RGB data, limiting its direct applicability to RGB-thermal (RGB-T) semantic segmentation. Given that RGB-T provides a robust solution for scene understanding in adverse weather and lighting conditions, such as low light and overexposure, we propose a novel framework, SARTM, which customizes the powerful SAM for RGB-T semantic segmentation. Our key idea is to unleash the potential of SAM while introduce semantic understanding modules for RGB-T data pairs. Specifically, our framework first involves fine-tuning the original SAM by adding extra LoRA layers, aiming at preserving SAM's strong generalization and segmentation capabilities for downstream tasks. Secondly, we introduce language information as guidance for training our SARTM. To address cross-modal inconsistencies, we introduce a Cross-Modal Knowledge Distillation(CMKD) module that effectively achieves modality adaptation while maintaining its generalization capabilities. This semantic module enables the minimization of modality gaps and alleviates semantic ambiguity, facilitating the combination of any modality under any visual conditions. Furthermore, we enhance the segmentation performance by adjusting the segmentation head of SAM and incorporating an auxiliary semantic segmentation head, which integrates multi-scale features for effective fusion. Extensive experiments are conducted across three multi-modal RGBT semantic segmentation benchmarks: MFNET, PST900, and FMB. Both quantitative and qualitative results consistently demonstrate that the proposed SARTM significantly outperforms state-of-the-art approaches across a variety of conditions. Code and pre-trained weights can be found at https://github.com/wahaha-debug/SARTM.

cs.CV↗
Compare source metadata on this page
WorkPublishedSource identifierSource
How to Backdoor Image Knowledge Distillation2026-10-222504.21323arxiv
KDFlow: A User-Friendly and Efficient Knowledge Distillation Framework for Large Language Models2026-09-072603.01875arxiv
PatchFormer: A Patch-Based Time Series Foundation Model with Hierarchical Masked Reconstruction and Cross-Domain Transfer Learning for Zero-Shot Multi-Horizon Forecasting2026-09-062601.20845arxiv
Continual Distillation Learning for Rehearsal-Free Class-Incremental Learning via Decoupled Prompting2026-09-052407.13911arxiv
Ensemble-Guided Distillation for Compact and Robust Acoustic Scene Classification on Edge Devices2026-09-052512.13905arxiv
Distilling Vision-Language Models for On-Device Fire Understanding2026-09-052609.05782arxiv
The Struggle Between Continuation and Refusal: A Mechanistic Analysis of the Continuation-Triggered Jailbreak in LLMs2026-09-042603.08234arxiv
Importance-Aware Low-Rank Distillation of Diffusion Transformers2026-09-042609.04646arxiv
Persistent Teacher Anchoring for Tool-Using Agents2026-09-042609.04773arxiv
Lightweight Vision Transformer Compression for On-Device Plant Disease Detection in Resource-Constrained Agricultural Field Conditions2026-09-042609.05334arxiv
CrisisKD: Five-Stage Knowledge Distillation for Aspect-Level Sentiment and Emotion Analysis in Crisis Discourse2026-09-042609.05757arxiv
ALRA: Adaptive Local Relational Alignment for Logit-Based Pre-training Distillation of Autoregressive Language Models2026-09-032609.03355arxiv
Discriminative and Consistent Representation Distillation2026-09-022407.11802arxiv
Uniformity First: Uniformity-aware Test-time Adaptation of Vision-language Models against Image Corruption2026-09-022505.12912arxiv
Breaking the Geometric Bottleneck: Contrastive Expansion in Asymmetric Cross-Modal Distillation2026-09-022603.06698arxiv
ProgResViT: Progressive Resolution and Width for Adaptive Vision Transformers2026-09-022609.03216arxiv
Relational Representation Distillation2026-09-012407.12073arxiv
SARTM: Segment Any RGB Thermal Model with Language aided Distillation2026-09-012505.01950arxiv

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