arXiv subjects

Jurn-Gyu Park

Publications and source records attributed to Jurn-Gyu Park.

5 recordsLinked to original sources

arXiv2026

Energy- and Memory-Efficient PEFT Methods for Personalized On-Device SLMs on Consumer GPUs

Despite rapid advances in large language models (LLMs), deploying and personalizing them on resource-constrained devices remains impractical due to high VRAM, time, and energy costs. Parameter-Efficient Fine-Tuning (PEFT) of Small Language Models (SLMs) offers a promising alternative, yet few studies compare PEFT methods across architectures using both general and personalization benchmarks while accounting for energy consumption. We compare five fine-tuning approaches (Full Fine-Tuning, LoRA, LoRA+, QLoRA, and BitFit) on four SLMs from two families (Transformer-based: TinyLlama-1.1B, Qwen3-1.7B; SSM-based: Mamba-1.4B, Mamba-2-1.3B) across three GLUE tasks (SST-2, QNLI, STS-B) and three LaMP personalization tasks (LaMP-1, LaMP-2, LaMP-3). Each configuration is evaluated with the energy-focused NetScore-E and the memory-focused NetScore-M, the two variants that reflect the constraints binding on-device deployment. Methods are selected with a strict energy-first rule (highest NetScore-E, ties broken by NetScore#). LoRA+ achieves the highest NetScore-E in 19 of 24 configurations and the highest NetScore-M in 13 of 24, and is the selected method in 18 of 24. QLoRA, available only for the Transformer models, cuts peak finetuning VRAM by up to 3.9x relative to LoRA and therefore takes the best NetScore-M in 5 of the 12 Transformer configurations, although its de-quantization overhead leaves it selected in only one of them once energy decides. BitFit and full fine-tuning are almost never competitive on either variant, and TinyLlama-1.1B leads the energy-focused NetScore-E on five of the six benchmarks and the memory-focused NetScore-M on four. These results show that compact SLMs paired with PEFT provide a practical, energy-aware path to personalized on-device deployment, with the optimal method set by the dominant constraint: LoRA+ for energy and QLoRA for memory.

cs.CL↗

arXiv2026

Energy-Efficient GPU DVFS for Fine-Tuning of SLMs on Resource-constrained Embedded Devices

Dynamic Voltage Frequency Scaling (DVFS) on resource-constrained embedded GPU platforms is essential for energy-efficient small language model (SLM) fine-tuning, as privacy- and personalization-driven adaptation increasingly requires local execution and involves repeated forward-backward optimization over many mini-batches, making it substantially more time- and energy-intensive than single-pass inference. To this end, 1) we first characterize the fine-tuning behavior of representative encoder-only SLMs of BERT variants, and autoregressive decoder-only SLMs of Pythia variants on GLUE benchmarks. In addition to the characterizations, 2) we propose a simple yet effective ML-based model selection that selects energy-optimal GPU DVFS settings on resource-constrained embedded platforms. Our results on NVIDIA Jetson AGX Orin demonstrate average 13.11% energy savings (up to 26.73%) over MAXN Mode 0, which has no explicit power cap.

cs.PF↗

arXiv2026

Efficient PEFT Methods with Adaptive Checkpointing for Vision Models and VLMs on Resource Constrained Consumer-GPUs

Modern pretrained vision models achieve strong accuracy but demand substantial GPU memory for fine-tuning, making edge deployment impractical. This paper compares five parameter-efficient fine-tuning (PEFT) methods (Full FT, LoRA, AdaLoRA, QLoRA, BitFit) on Transformers- (ViT-Small, TinyViT) and Mamba-based vision backbones (Vim-Small, MambaVision-T) under an on-device VRAM budget (e.g., 2 GB), together with three gradient-checkpointing strategies (none, static, and a proposed memory-budget-aware adaptive algorithm); and we evaluate three families of foundation-model baselines: zero-shot contrastive vision language models (OpenCLIP, SigLIP), self-supervised vision backbones with lightweight evaluation protocols (DINOv2), and autoregressive VLMs for prompt-based classification (PaliGemma, MobileVLM, SmolVLM). Experiments on CIFAR-100 and DTD report accuracy, training time, energy, and the NetScore family of multi-objective metrics, which we extend with two deployment-aware variants. QLoRA and BitFit cut energy 20-30% at a 1-2% accuracy cost; the adaptive algorithm reduces peak memory 43-79% with 9-30% energy overhead. DINOv2 surpasses fine-tuned models on CIFAR-100 (0.917 vs. 0.897) at a fraction of the energy, while small autoregressive VLMs remain uncompetitive.

cs.CV↗

arXiv2025

Energy-Efficient Vision Transformer Inference for Edge-AI Deployment

The growing deployment of Vision Transformers (ViTs) on energy-constrained devices requires evaluation methods that go beyond accuracy alone. We present a two-stage pipeline for assessing ViT energy efficiency that combines device-agnostic model selection with device-related measurements. We benchmark 13 ViT models on ImageNet-1K and CIFAR-10, running inference on NVIDIA Jetson TX2 (edge device) and an NVIDIA RTX 3050 (mobile GPU). The device-agnostic stage uses the NetScore metric for screening; the device-related stage ranks models with the Sustainable Accuracy Metric (SAM). Results show that hybrid models such as LeViT_Conv_192 reduce energy by up to 53% on TX2 relative to a ViT baseline (e.g., SAM5=1.44 on TX2/CIFAR-10), while distilled models such as TinyViT-11M_Distilled excel on the mobile GPU (e.g., SAM5=1.72 on RTX 3050/CIFAR-10 and SAM5=0.76 on RTX 3050/ImageNet-1K).

cs.LG↗

arXiv2025

Efficient-Husformer: Efficient Multimodal Transformer Hyperparameter Optimization for Stress and Cognitive Loads

Transformer-based models have gained considerable attention in the field of physiological signal analysis. They leverage long-range dependencies and complex patterns in temporal signals, allowing them to achieve performance superior to traditional RNN and CNN models. However, they require high computational intensity and memory demands. In this work, we present Efficient-Husformer, a novel Transformer-based architecture developed with hyperparameter optimization (HPO) for multi-class stress detection across two multimodal physiological datasets (WESAD and CogLoad). The main contributions of this work are: (1) the design of a structured search space, targeting effective hyperparameter optimization; (2) a comprehensive ablation study evaluating the impact of architectural decisions; (3) consistent performance improvements over the original Husformer, with the best configuration achieving an accuracy of 88.41 and 92.61 (improvements of 13.83% and 6.98%) on WESAD and CogLoad datasets, respectively. The best-performing configuration is achieved with the (L + dm) or (L + FFN) modality combinations, using a single layer, 3 attention heads, a model dimension of 18/30, and FFN dimension of 120/30, resulting in a compact model with only about 30k parameters.

cs.LG↗