检索结果-南通市图书馆

Optimizing Fine-Tuning in Quantized Language Models:An In-Depth Analysis of Key Variables

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2025年第1期82卷 307-325页

作者： Ao Shen Zhiquan lai Dongsheng Li Xiaoyu Hu National Key Laboratory of Parallel and Distributed Computing National University of Defense TechnologyChangsha410073China Strategic Assessments and Consultation Institute Academy of Military ScienceBeijing100091China

Large-scale Language Models(LLMs)have achieved significant breakthroughs in Natural Language Processing(NLP),driven by the pre-training and fine-tuning *** this approach allows models to specialize in specific tasks with reduced training costs,the substantial memory requirements during fine-tuning present a barrier to broader ***-Efficient Fine-Tuning(PEFT)techniques,such as Low-Rank Adaptation(LoRA),and parameter quantization methods have emerged as solutions to address these challenges by optimizing memory usage and computational *** these,QLoRA,which combines PEFT and quantization,has demonstrated notable success in reducing memory footprints during fine-tuning,prompting the development of various QLoRA *** these advancements,the quantitative impact of key variables on the fine-tuning performance of quantized LLMs remains *** study presents a comprehensive analysis of these key variables,focusing on their influence across different layer types and depths within LLM *** investigation uncovers several critical findings:(1)Larger layers,such as MLP layers,can maintain performance despite reductions in adapter rank,while smaller layers,like self-attention layers,aremore sensitive to such changes;(2)The effectiveness of balancing factors depends more on specific values rather than layer type or depth;(3)In quantization-aware fine-tuning,larger layers can effectively utilize smaller adapters,whereas smaller layers struggle to do *** insights suggest that layer type is a more significant determinant of fine-tuning success than layer depth when optimizing quantized ***,for the same discount of trainable parameters,reducing the trainable parameters in a larger layer is more effective in preserving fine-tuning accuracy than in a smaller *** study provides valuable guidance for more efficient fine-tuning strategies and opens avenues for further research into optimizing LLM

关键词： Large-scale Language Model Parameter-Efficient Fine-Tuning parameter quantization key variable trainable parameters experimental analysis

Structural Modal Parameter Recognition and Related Damage Identification Methods under Environmental Excitations: A Review

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Structural Durability & Health Monitoring 2025年第1期19卷 25-54页

作者： Chao Zhang Shang-Xi lai Hua-Ping Wang School of Civil Engineering and Mechanics Lanzhou UniversityLanzhou730000China Key Laboratory of Special Functional Materials and Structural Design Ministry of EducationLanzhou UniversityLanzhou730000China

Modal parameters can accurately characterize the structural dynamic properties and assess the physical state of the ***,it is particularly signiﬁcant to identify the structural modal parameters according to the monitoring data information in the structural health monitoring(SHM)system,so as to provide a scientiﬁc basis for structural damage identiﬁcation and dynamic model modiﬁ*** view of this,this paper reviews methods for identifying structural modal parameters under environmental excitation and brieﬂy describes how to identify structural damages based on the derived modal *** paper primarily introduces data-driven modal parameter recognition methods(e.g.,time-domain,frequency-domain,and time-frequency-domain methods,etc.),brieﬂy describes damage identiﬁcation methods based on the variations of modal parameters(e.g.,natural frequency,modal shapes,and curvature modal shapes,etc.)and modal validation methods(e.g.,Stability Diagram and Modal Assurance Criterion,etc.).The current status of the application of artiﬁcial intelligence(AI)methods in the direction of modal parameter recognition and damage identiﬁcation is further *** on the pre-vious analysis,the main development trends of structural modal parameter recognition and damage identiﬁcation methods are given to provide scientiﬁc references for the optimized design and functional upgrading of SHM systems.

关键词： Structural health monitoring data information modal parameters damage identiﬁcation AI method

Enhancing Deepfake Detection:Proactive Forensics Techniques Using Digital Watermarking

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2025年第1期82卷 73-102页

作者： Zhimao lai Saad Arif Cong Feng Guangjun Liao Chuntao Wang School of Automation Guangdong University and TechnologyGuangzhou510006China School of Immigration Administration(Guangzhou) China People’s Police UniversityGuangzhou510663China Department of Mechanical Engineering College of EngineeringKing Faisal UniversityAl Ahsa31982Saudi Arabia Cyber Police Division Guangzhou Public Security BureauGuangzhou510030China School of Forensic Science and Technology Guangdong Police CollegeGuangzhou510320China College of Mathematics and Informatics South China Agricultural UniversityGuangzhou510642China

With the rapid advancement of visual generative models such as Generative Adversarial Networks(GANs)and stable Diffusion,the creation of highly realistic Deepfake through automated forgery has significantly *** paper examines the advancements inDeepfake detection and defense technologies,emphasizing the shift from passive detection methods to proactive digital watermarking *** detection methods,which involve extracting features from images or videos to identify forgeries,encounter challenges such as poor performance against unknown manipulation techniques and susceptibility to counter-forensic *** contrast,proactive digital watermarking techniques embed specificmarkers into images or videos,facilitating real-time detection and traceability,thereby providing a preemptive defense againstDeepfake *** offer a comprehensive analysis of digitalwatermarking-based forensic techniques,discussing their advantages over passivemethods and highlighting four key benefits:real-time detection,embedded defense,resistance to tampering,and provision of legal ***,the paper identifies gaps in the literature concerning proactive forensic techniques and suggests future research directions,including cross-domain watermarking and adaptive watermarking *** systematically classifying and comparing existing techniques,this review aims to contribute valuable insights for the development of more effective proactive defense strategies in Deepfake forensics.

关键词： Deepfake proactive forensics digital watermarking traceability detection techniques

Pathogenesis, diagnosis, and treatment of epilepsy: electromagnetic stimulation-mediated neuromodulation therapy and new technologies

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Neural Regeneration Research 2025年第4期20卷 917-935页

作者： Dian Jiao lai Xu Zhen Gu Hua Yan Dingding Shen Xiaosong Gu Academy of Medical Engineering and Translational Medicine Tianjin UniversityTianjinChina Key Laboratory of Advanced Drug Delivery Systems of Zhejiang Province College of Pharmaceutical SciencesZhejiang UniversityHangzhouZhejiang ProvinceChina Affiliated Hospital of Nantong University NantongJiangsu ProvinceChina

Epilepsy is a severe,relapsing,and multifactorial neurological *** regarding the accurate diagnosis,prognosis,and in-depth pathogenesis are crucial for the precise and effective treatment of *** pathogenesis of epilepsy is complex and involves alterations in variables such as gene expression,protein expression,ion channel activity,energy metabolites,and gut microbiota *** results are lacking for conventional treatments for *** resection of lesions,drug therapy,and non-drug interventions are mainly used in clinical practice to treat pain associated with ***-pharmacological treatments,such as a ketogenic diet,gene therapy for nerve regeneration,and neural regulation,are currently areas of research *** review provides a comprehensive overview of the pathogenesis,diagnostic methods,and treatments of *** also elaborates on the theoretical basis,treatment modes,and effects of invasive nerve stimulation in neurotherapy,including percutaneous vagus nerve stimulation,deep brain electrical stimulation,repetitive nerve electrical stimulation,in addition to non-invasive transcranial magnetic stimulation and transcranial direct current *** studies have shown that electromagnetic stimulation-mediated neuromodulation therapy can markedly improve neurological function and reduce the frequency of epileptic ***,many new technologies for the diagnosis and treatment of epilepsy are being ***,current research is mainly focused on analyzing patients’clinical manifestations and exploring relevant diagnostic and treatment methods to study the pathogenesis at a molecular level,which has led to a lack of consensus regarding the mechanisms related to the disease.

关键词： diagnosis drug treatment electroencephalography epilepsy monitoring epilepsy nerve regeneration neurostimulation non-drug interventions pathogenesis prediction

Experimental Study on Energy Storage Characteristics of Sintered Ore Particle Packed Beds

在线全文

学校读者我要写书评

暂无评论

Journal of Thermal Science 2025年第1期34卷 242-253页

作者： lai Zhenya DING Liwei LYU Hongkun HOU chenglong CHEN Jiaying GUO Xutao HAN Gaoyan ZHANG Kang State Grid Zhejiang Electric Power Research Institute State Key Laboratory of Clean Energy Utilization Zhejiang University

The particle packed bed energy storage system has advantages such as low costs and wide temperature ranges,which can be combined with solar thermal power generation systems to solve the inherent volatility and discontinuity of renewable *** new materials with low costs and excellent storage performances is one of the eternal research hotspots in the field of energy *** paper innovatively uses sintered ore particles as energy storage material and studies the effect of particle size on the airflow resistance characteristics,energy storage characteristics,and thermocline evolution characteristics of the packed bed through thermal energy storage *** results indicate that for the particles in the macro scale,the smaller the particle,the lower the absolute permeability of the bed and the greater the airflow *** packed bed with smaller particles has a larger specific surface area,larger bulk mass,and smaller bed ***,the packed beds with smaller particles have better thermocline characteristics,less irreversible loss,and can achieve higher thermal efficiency and higher exergy efficiency in the heat storage *** cycle thermal efficiency in packed beds with 25-40 mm,16-25 mm,and 10-16 mm particles is 53.58%,56.27%,and 57.60%,respectively,and the cycle exergy efficiency is 61.81%,69.25%,and 74.13%,***,this paper also studies the effect of discharging airflow rates on thermal storage *** experimental results indicate that suitable discharging strategies should be selected based on different heat demands.

关键词： airflow resistance particle packed bed thermal energy storage thermocline evolution

Meningeal lymphatic vessel crosstalk with central nervous system immune cells in aging and neurodegenerative diseases

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Neural Regeneration Research 2025年第3期20卷 763-778页

作者： Minghuang Gao Xinyue Wang Shijie Su Weicheng Feng Yaona lai Kongli Huang Dandan Cao Qi Wang Science and Technology Innovation Center Guangzhou University of Chinese MedicineGuangzhouGuangdong ProvinceChina Institute of Clinical Pharmacology Guangzhou University of Chinese MedicineGuangzhouGuangdong ProvinceChina The Sixth Affiliated Hospital Sun Yat-sen UniversityGuangzhouGuangdong ProvinceChina

Meningeal lymphatic vessels form a relationship between the nervous system and periphery, which is relevant in both health and disease. Meningeal lymphatic vessels not only play a key role in the drainage of brain metabolites but also contribute to antigen delivery and immune cell activation. The advent of novel genomic technologies has enabled rapid progress in the characterization of myeloid and lymphoid cells and their interactions with meningeal lymphatic vessels within the central nervous system. In this review, we provide an overview of the multifaceted roles of meningeal lymphatic vessels within the context of the central nervous system immune network, highlighting recent discoveries on the immunological niche provided by meningeal lymphatic vessels. Furthermore, we delve into the mechanisms of crosstalk between meningeal lymphatic vessels and immune cells in the central nervous system under both homeostatic conditions and neurodegenerative diseases, discussing how these interactions shape the pathological outcomes. Regulation of meningeal lymphatic vessel function and structure can influence lymphatic drainage, cerebrospinal fluid-borne immune modulators, and immune cell populations in aging and neurodegenerative disorders, thereby playing a key role in shaping meningeal and brain parenchyma immunity.

关键词： central nervous system meningeal lymphatic vessels immunity myeloid cells lymphatic cells neurodegenerative disease

Establishment of human cerebral organoid systems to model early neural development and assess the central neurotoxicity of environmental toxins

在线全文

学校读者我要写书评

暂无评论

Neural Regeneration Research 2025年第1期20卷 242-252页

作者： Daiyu Hu Yuanqing Cao chenglin Cai Guangming Wang Min Zhou Luying Peng Yantao Fan Qiong lai Zhengliang Gao Fundamental Research Center Shanghai Yangzhi Rehabilitation Hospital(Shanghai Sunshine Rehabilitation Center)School of MedicineTongji UniversityShanghaiChina Institute of Geriatrics(Shanghai University) Affiliated Nantong Hospital of Shanghai University(The Sixth People’s Hospital of Nantong)Shanghai University School of MedicineNantongJiangsu ProvinceChina Shanghai Engineering Research Center of Organ Repair Shanghai University School of MedicineShanghaiChina Key Laboratory of Arrhythmias Ministry of EducationShanghai East HospitalSchool of MedicineTongji UniversityShanghaiChina

Human brain development is a complex process,and animal models often have significant *** address this,researchers have developed pluripotent stem cell-derived three-dimensional structures,known as brain-like organoids,to more accurately model early human brain development and *** enable more consistent and intuitive reproduction of early brain development,in this study,we incorporated forebrain organoid culture technology into the traditional unguided method of brain organoid *** involved embedding organoids in matrigel for only 7 days during the rapid expansion phase of the neural epithelium and then removing them from the matrigel for further cultivation,resulting in a new type of human brain organoid *** cerebral organoid system replicated the temporospatial characteristics of early human brain development,including neuroepithelium derivation,neural progenitor cell production and maintenance,neuron differentiation and migration,and cortical layer patterning and formation,providing more consistent and reproducible organoids for developmental modeling and toxicology *** a proof of concept,we applied the heavy metal cadmium to this newly improved organoid system to test whether it could be used to evaluate the neurotoxicity of environmental *** organoids exposed to cadmium for 7 or 14 days manifested severe damage and abnormalities in their neurodevelopmental patterns,including bursts of cortical cell death and premature *** exposure caused progressive depletion of neural progenitor cells and loss of organoid integrity,accompanied by compensatory cell proliferation at ectopic *** convenience,flexibility,and controllability of this newly developed organoid platform make it a powerful and affordable alternative to animal models for use in neurodevelopmental,neurological,and neurotoxicological studies.

关键词： cadmium cell death cell proliferation cortical development environmental toxins neural progenitor cells neurogenesis neurotoxicology organoids stem cells

Molecular simulation study of the microstructures and properties of pyridinium ionic liquid[HPy][BF_(4)]mixed with acetonitrile

在线全文

学校读者我要写书评

暂无评论

原子与分子物理学报 2025年第4期42卷 27-32页

作者： XU Jian-Qiang MA Zhao-Peng cheng Si LIU Zhi-Cong ZHU Guang-lai Department of Physics School of Physics and Electronic InformationAnhui Normal UniversityWuhu 241002China

The microstructures and thermodynamic properties of mixed systems comprising pyridinium ionic liquid[HPy][BF_(4)]and acetonitrile at different mole fractions were studied using molecular dynamics simulation in this *** following properties were determined:density,self-diffusion coefficient,excess molar volume,and radial distribution *** results show that with an increase in the mole fraction of[HPy][BF_(4)],the self-diffusion coefficient ***,the excess molar volume initially decreases,reaches a minimum,and then *** rules of radial distribution functions(RDFs)of characteristic atoms are *** increasing the mole fraction of[HPy][BF_(4)],the first peak of the RDFs of HA1-F decreases,while that of CT6-CT6 rises at first and then *** indicates that the solvent molecules affect the polar and non-polar regions of[HPy][BF_(4)]differently.

关键词： Pyridinium ionic liquids Thermodynamic properties Molecular dynamics simulation Radial distribution functions

Bifunctional bridging capping layer enables 24.5% efficiency of perovskite solar cells with polymer-based hole transport materials

在线全文

学校读者我要写书评

暂无评论

Science China Chemistry 2025年第1期68卷 350-359页

作者： Can Zhu Yiyang Wang Lei Meng Beibei Qiu Jing Li Shucheng Qin Ke Hu Xin Jiang Wenbin lai Minchao Liu Zhe Liu Chenxing Lu Jinyuan Zhang Yongfang Li Beijing National Laboratory for Molecular Sciences CAS Key Laboratory of Organic Solids Institute of Chemistry Chinese Academy of Sciences School of Chemical Science University of Chinese Academy of Sciences Key Laboratory of Solid State Optoelectronic Devices of Zhejiang Province College of Physics and Electronic Information Engineering Zhejiang Normal University Key Laboratory of Photochemical Conversion and Optoelectronic Materials Technical Institute of Physics and ChemistryChinese Academy of Sciences Laboratory of Advanced Optoelectronic Materials College of Chemistry Chemical Engineering and Materials ScienceSoochow University

Developing a bridge capping layer between perovskite and hole transport layer materials（HTMs） in the n-i-p perovskite solar cells（pero-SCs） is an effective approach to modify the morphology of HTMs and passivate the perovskite ***, we select the quinoxaline-based bifunctional passivation agent, quinoxalin-6-yl-methylamine hydrochloride（Qx MACl）,as the bridging layer, and a D-A copolymer PBQ12 containing the same quinoxaline unit as an HTM for the n-i-p pero-SCs. Due to the π-π stacking among the common quinoxaline units in the bridge layer and HTM, Qx MACl induces the π-π stacking of the PBQ12 film and improves the film morphology of HTMs with better conductivity. Additionally, Qx MACl can effectively passivate the perovskite surface, and PBQ12 possesses appropriate energy levels and high hole mobility. The pero-SCs based on FAPbI3with PBQ12/Qx MACl treatment showed a higher power conversion efficiency（PCE） of 24.05% and outstanding stability, maintaining 95.4% and 92.1% of its initial PCE after 750 h of storage and after over 800 h of thermal annealing at 85 °C, respectively. To further enhance the PCE of the PBQ12/Qx MACl-based devices, we developed a non-metal ion dopant for the PBQ12 HTM. Through trace doping of PBQ12 HTM by the non-metal ion dopant, the PCE of the PBQ12/Qx MACl-based devices reached 25.24%（the calibrated PCE of 24.55% by the National Institute of Metrology, China）.

关键词： D-A copolymer, hole transport materials, bifunctional bridge capping layer, passivation perovskite solar cells