Nobel Dhar, Md Romyull Islam, Xuechen Zhang +4cs.DC cs.LG
Deploying large language models on edge devices is increasingly limited by a widening gap between model size and available memory. Existing approaches such as quantization, smaller models, and offloading can raise the effective memory limit, but they still assume that the model can be compressed or partitioned to fit within some budget. We target the harder model-exceeds-memory setting, in which the model remains larger than resident memory throughout execution and storage becomes an active source of weights on the critical path. We observe that MLP activity during autoregressive decoding has strong temporal locality: approximately 82-85% of active neurons persist from one token to the next. This means that most sparse weights needed for the current token are already resident, and only the newly needed rows must be fetched from storage. We present NeuroPrefetcher, a storage-backed LLM inference system that exploits this property through predictive delta prefetching. After layer 0, a single GPU-resident predictor, occupying 2.86% of base model parameters, predicts sparse activity for all downstream MLP layers in one forward pass. The runtime compares these predictions against resident GPU buffers and issues application-scheduled NVMe reads only for incoming delta rows, replacing reactive operating-system demand paging with explicit, model-aware weight movement. On real unified-memory edge hardware, NeuroPrefetcher achieves 7.9-12.0x speedup over llama.cpp across constrained memory budgets.
Model compression is critical for deploying networks on resource-constrained edge devices. While pruning-based methods can significantly reduce model size, they often suffer from abrupt performance collapse beyond a sparsity thresh-old, making it difficult to identify the feasible compression limit of the model. To address this challenge, we propose a boundary-Learning reverse regrowth framework, BRIDGE, that reformulates compression as a constructive boundary-search problem. Unlike forward pruning, our method first drives the model to an extremely sparse state to expose the collapse region, and then selectively regenerates the critical structure to restore performance. The proposed framework employs a hierarchical regeneration strategy, including coarse-grained layer selection and fine-grained regeneration parameter selection, to accurately identify which parameters require recovery. Experiments show that our method can recover models from the brink of collapse on both CNNs and Transformer architectures, demonstrating its architecture in-dependence. BRIDGE achieves a performance improvement of up to 1.49% in unstructured pruning and up to 4.77% in structured pruning. These results demonstrate that reverse regeneration can effectively extend the compression limit while maintaining stable performance. The source code is available at https://github.com/EnumaCaliber/BRIDGE.
Lookup table (LUT)-based image denoising methods have attracted increasing attention due to their high efficiency and hardware-friendly properties. However, existing RGB-LUT approaches require three identical LUTs to process RGB channels in parallel, resulting in large on-chip SRAM consumption. A simple alternative is to apply LUT processing only to the luminance (Y) channel in the YUV color space to reduce memory usage. However, this naive strategy leads to degraded restoration quality, since ignoring the chrominance (UV) channels introduces color distortion and residual artifacts. In this work, we propose Hybrid-LUT, a YUV-based asymmetric channel-processing framework that combines LUT and filtering in a unified design. Specifically, a multi-band LUT branch with pixel-level weight fusion is applied to the Y channel to recover fine textures, while lightweight filtering is used for the UV channels to maintain color consistency. This design reduces LUT storage by two-thirds compared with RGB-LUT methods while maintaining the same runtime throughput. Extensive experiments show that Hybrid-LUT achieves state-of-the-art (SOTA) performance across multiple benchmarks with only 421 KB of storage. In particular, our method surpasses existing LUT-based denoising approaches by at least 0.63 dB CPSNR on real-world datasets, demonstrating its effectiveness for image denoising on resource-constrained edge devices. The project is available at https://github.com/Ai-ZL/Hybrid-LUT .
On-device speech emotion recognition (SER) is critical for real-time applications, yet large self-supervised models that excel at SER are too costly for edge devices. Multi-teacher knowledge distillation can compress them into a lightweight student, but two challenges remain: teacher reliability varies across batches, and logit-level distillation ignores inter-sample relational structure. We propose Adaptive Multi-teacher Relational Distillation (AMRD) to address both. A one-class SVM on each teacher's logit similarity matrix assigns per-batch weights favoring more coherent teachers. A relational distillation loss aligns teacher and student similarity matrices, capturing structure that logit matching misses. On IEMOCAP and CREMA-D datasets across four student architectures, AMRD outperforms single-teacher distillation baselines in most settings, and ablations confirm both components yield complementary gains.
Filip Pawlicki, Marcel Kańduła, Marcin Pucek +1cs.CV
Recent Video Super-Resolution (VSR) methods rely heavily on transformers and explicit optical flow, creating computational overhead and custom operations that hinder deployment on hardware accelerators like TensorRT. To address this, we introduce NanoVSR, a scalable, fully convolutional architecture designed for resource-constrained edge devices. Using structural reparameterization, NanoVSR collapses into standard convolutions during inference, ensuring seamless hardware compatibility and negligible runtime overhead. Furthermore, despite lacking explicit motion compensation, it maintains competitive restoration quality by implicitly learning spatio-temporal alignments through progressive training. Evaluated on the REDS4 benchmark, NanoVSR demonstrates an exceptional balance between accuracy and computational efficiency, significantly improving the trade-off for compact architectures. Our NanoVSR-644k baseline yields 28.64 dB PSNR while delivering 27.2 FPS on the NVIDIA Jetson Orin NX 16GB (25W), offering massive speed gains over heavier models. The scaled NanoVSR-1.7M variant reaches 29.15 dB with a throughput of 19.58 FPS, providing superior, edge-optimized upscaling. Code is available at https://github.com/filippawlicki/nanovsr.
Test-time adaptation (TTA) mitigates domain shifts by using incoming test data to update a model on the fly. The majority of TTA methods require resource-intensive backpropagation (BP) for model updates, particularly demanding large memory sizes, which makes it infeasible to deploy them on resource-limited devices (e.g., edge devices). To address this issue, we integrate two different techniques, zeroth-order optimization (ZOO) and model merging, under the recently established cross-device collaborative TTA (CDC-TTA) framework, where the system is composed of a mixture of resource-abundant and resource-limited devices, and the model information (e.g., model weights obtained on each device) is shared across the devices. Our method is executable on resource-limited devices by introducing ZOO, which requires only forward processing and bypasses the resource-intensive BP optimization. Concurrently, to mitigate the high-dimensional optimization difficulty caused by the side effect of ZOO, we incorporate model merging of the shared multiple models and set the merge coefficients as the optimization objective, which successfully reduces the optimization dimension. In addition, to enhance the synergistic combination of ZOO and model merging, we propose a unique preprocessing strategy that trims intra-model non-influential weights and reduces the inter-model information redundancy. We empirically confirmed the effectiveness of our method using common corruption and style-transferred image benchmarks.
Recent advances in speech separation (SS) have led to compact front-end models with small parameter sizes, yet their high computational cost remains a major barrier for deployment on edge devices. To address this, we propose TF-MoE, a sparse Mixture-of-Experts (MoE) framework that enhances model capacity with almost no increase in inference cost. Our method introduces dynamic expert specialization in time and frequency dimensions through alternating time-wise and frequency-wise MoE modules, each dynamically selecting experts per frame or mel band. Built upon a mel-band-splitting Conformer backbone, TF-MoE achieves strong performance on SS tasks under low-compute settings. Experimental results demonstrate that TF-MoE consistently improves separation performance under computation cost constraints, outperforming BSRNN by +3.8 dB SDR on Libri2Mix with comparable 4.1 GMACs/s inference cost. This positions TF-MoE as a promising candidate for edge-device deployment.
Fine-tuning of Large Language Models (LLMs) using Low-Rank Adaptation (LoRA) on an end-user's data offers personalized experiences while keeping data private, but faces severe memory constraints on consumer hardware. Peak memory during fine-tuning often exceeds device limits, especially for models with billions of parameters and long-context training data. This paper introduces a suite of complementary techniques to reduce memory footprint without sacrificing model quality: (1) base model quantization with on-the-fly dequantization, (2) memory-efficient checkpointing combining selective activation caching and disk offloading, (3) softmax approximation using semantically relevant token subsets, and (4) logits masking. Experiments on Llama-3.2 3B and Qwen-2.5 3B demonstrate up to $26\times$ and $28\times$ reduction in peak memory, enabling fine-tuning on resource-constrained devices.
Mixture-of-Experts (MoE) models activate only a sparse subset of experts per token, yet consecutive tokens frequently activate different experts -- causing constant weight swapping between slow storage and fast memory on edge devices. Existing remedies are either system-level (caching heuristics) or post-hoc (router fine-tuning), leaving the root cause unchanged during pretraining. We propose StickyMoE, a differentiable routing consistency loss that penalises abrupt expert switches between adjacent tokens, encouraging the router to maintain the same expert assignment across semantically coherent spans. StickyMoE requires no architectural changes, adds a single hyperparameter lambda, and unlike post-hoc methods, allows expert representations and routing decisions to co-adapt from the first training step. Experiments on small-scale MoE language models show that StickyMoE reduces the expert switch rate by up to 60% with less than 4% perplexity degradation, Pareto-dominating post-hoc fine-tuning on the quality-locality frontier. Routing temporal locality is most efficiently instilled at training time.
The rise of edge-based machine learning has enabled distributed adaptation of language models across mobile and IoT devices, offering privacy preservation and real-time responsiveness. However, distributed fine-tuning of language models on untrusted or heterogeneous edge nodes introduces new vulnerabilities. Compromised or unreliable devices can inject poisoned updates, leading to stealthy model manipulation or convergence degradation. Classical defenses such as robust aggregation or temporal anomaly detection operate on a single global model and are therefore limited in detecting coordinated or persistent poisoning. This work proposes a new system-level defense based on model multiplicity. Instead of maintaining one global model, the system rotates or concurrently trains multiple small language models (e.g., DistilGPT-2), each updated by independently sampled subsets of edge nodes. These models evolve under distinct training trajectories, creating multiple independent views of the same distributed population. Divergence between models quantified through gradient similarity, loss evolution, or parameter variance serves as a signal of anomalous or adversarial behavior. When one model deviates significantly from the ensemble mean, the system flags its contributing nodes for isolation or re-weighting. We implement this framework and evaluate it on edge-scale simulations of Small Language Model (SLM) training under varying heterogeneity and attack conditions. Results show that model multiplicity enables earlier and more reliable detection of poisoning compared to classical single-model defenses such as Flanders and Robust methods. Our findings demonstrate that diversity in model evolution can serve as a practical and effective defense mechanism for secure distributed learning on resource-constrained edge devices.
Changyu Li, Shuanghong Huang, Jiashen Liu +5cs.LG cs.AI
Federated fine-tuning provides a practical route to adapt large language models (LLMs) on edge devices without centralizing private data, yet in mobile deployments the training wall-clock is often bottlenecked by straggler-limited uplink communication under heterogeneous bandwidth and intermittent participation. Although parameter-efficient fine-tuning (PEFT) reduces trainable parameters, per-round payloads remain prohibitive in non-IID regimes, where uniform compression can discard rare but task-critical signals. We propose Fed-FSTQ, a Fisher-guided token quantization system primitive for communication-efficient federated LLM fine-tuning. Fed-FSTQ employs a lightweight Fisher proxy to estimate token sensitivity, coupling importance-aware token selection with non-uniform mixed-precision quantization to allocate higher fidelity to informative evidence while suppressing redundant transmission. The method is model-agnostic, serves as a drop-in module for standard federated PEFT pipelines, e.g., LoRA, without modifying the server aggregation rule, and supports bandwidth-heterogeneous clients via compact sparse message packing. Experiments on multilingual QA and medical QA under non-IID partitions show that Fed-FSTQ reduces cumulative uplink traffic required to reach a fixed quality threshold by 46x relative to a standard LoRA baseline, and improves end-to-end wall-clock time-to-accuracy by 52%. Furthermore, enabling Fisher-guided token reduction at inference yields up to a 1.55x end-to-end speedup on NVIDIA Jetson-class edge devices, demonstrating deployability under tight resource constraints.