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

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

Targeting cholesterol trafficking to mitigate axonal degeneration in hereditary spastic paraplegia

维普期刊数据库评论

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5期20卷 1397-1398页

作者： Zhenyu Chen xue-Jun li Department of Biomedical Sciences University of Illinois College of Medicine RockfordRockfordILUSA Department of Bioengineering University of Illinois at ChicagoChicagoILUSA

Axonal degeneration underlies many debilitating diseases including hereditary spastic paraplegia(HSP),a genetically and clinically diverse group of disorders characterized by spasticity and weakness of the lower *** is one significant cause of chronic neurodisability due to the lack of effective treatments and a wide range of onset ages from early childhood to 70 years.

关键词： degeneration diseases hereditary

Turbulent Agglomeration of Microparticles in a Cylinder Wake Flow Using LES-DEM:Focusing on the Effect of the Reynolds Number

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 34-49页

作者： WANG Shuang MU lin li xue XIE Jun DONG Ming School of Energy and Power Engineering Dalian University of Technology School of Optical Information and Energy Engineering Wuhan Institute of Technology School of Energy and Environment Shenyang Aerospace University

Turbulent agglomeration is viewed as a promising technology for enhancing fine particle removal *** better understand particle transport,agglomeration behaviors,and fluid-particle interactions,we numerically explored these phenomena under cylindrical vortex wake influence using a coupled large eddy simulation and discrete element method（LES-DEM） *** validity of the LES approach was verified by comparison with available direct numerical simulation（DNS） *** adopted the Johnson-Kendall-Roberts（JKR） contact model for particle-particle *** particle dispersion and agglomeration characteristics of particles with different diameters（dp=2-20 μm） in the laminar and transition of shear layer（TrSL） flow regimes were *** particles were concentrated at the vortex centers,while larger particles accumulated around the *** agglomeration efficiency exhibited an M-shaped profile spanwise（y-direction）.With increasing Reynolds number,the agglomeration efficiency and turbulence intensity *** particle agglomeration efficiency peaks at a certain Reynolds ***,at higher Reynolds numbers,reducing the residence time of particles in the flow field decreases the agglomeration efficiency.

关键词： discrete element method large eddy simulation microparticle particle agglomeration

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Correction

International Journal of Automation and computing 2021年第6期18卷 1045-1045页

作者： ao-xue li Ke-Xin Zhang li-Wei Wang The Key Laboratory of Machine Perception(MOE) School of Electronics Engineering and Computer SciencePeking UniversityBeijing 100871China

Correction to:Zero-shot Fine-grained Classification by Deep Feature Learning with Semantics DOI:10.1007/s11633-019-1177-8 Authors:ao-xue li,Ke-Xin Zhang,li-Wei Wang The article Zero-shot Fine-grained Classification by Deep Feature Learning with Semantics written by ao-xue li,Ke-Xin Zhang and li-Wei Wang,was originally published on vol.16,no.5 of International Journal of Automation and Computing without Open *** publication,the authors decided to opt for Open Choice and to make the article an Open Access publication.

关键词： Open DOI Access make

A Multi-Objective Particle Swarm Optimization Algorithm Based on Decomposition and Multi-Selection Strategy

维普期刊数据库评论

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 997-1026页

作者： li Ma Cai Dai Xingsi xue Cheng Peng School of Computer Science Shaanxi Normal UniversityXi’an710119China Fujian Provincial Key Laboratory of Big Data Mining and Applications Fujian University of TechnologyFuzhou350118China Information Construction and Management Center and Institute of Artificial Intelligence and Educational New Productivity Ningxia Normal UniversityGuyuan756099China

The multi-objective particle swarm optimization algorithm(MOPSO)is widely used to solve multi-objective optimization *** the article,amulti-objective particle swarm optimization algorithmbased on decomposition and multi-selection strategy is proposed to improve the search ***,two update strategies based on decomposition are used to update the evolving population and external archive,***,a multiselection strategy is *** first strategy is for the subspace without a non-dominated *** the neighbor particles,the particle with the smallest penalty-based boundary intersection value is selected as the global optimal solution and the particle far away fromthe search particle and the global optimal solution is selected as the personal optimal solution to enhance global *** second strategy is for the subspace with a non-dominated *** the neighbor particles,two particles are randomly selected,one as the global optimal solution and the other as the personal optimal solution,to enhance local *** third strategy is for Pareto optimal front(PF)discontinuity,which is identified by the cumulative number of iterations of the subspace without non-dominated *** the subsequent iteration,a new probability distribution is used to select from the remaining subspaces to ***,an adaptive inertia weight update strategy based on the dominated degree is designed to further improve the search ***,the proposed algorithmis compared with fivemulti-objective particle swarm optimization algorithms and five multi-objective evolutionary algorithms on 22 test *** results show that the proposed algorithm has better performance.

关键词： Multi-objective optimization multi-objective particle swarm optimization decomposition multi-selection strategy

Modular assembly of Co4-thiacalix[4]arene units and Keggin-type SiW12O40 for visible-light photothermal conversion

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Tungsten 2025年第1期7卷 112-119页

作者： Jia-Hui li Qing Zhao Dong-ao Mao Kun Zhou Bao-Kuan Chen Yan-Feng Bi School of Petrochemical Engineering Liaoning Petrochemical University

The construction of metal-organic coordination clusters（MCCs） using modular secondary building units is an important method for preparing functional *** this study,two new MCCs modularly assembled from Keggin-type {SiW12O40}H4TC4A=p-tert-Butylthiacalix[4]arene(SiW12) units and thiacalixarene-supported Co4-thiacalix[4]arene(Co4-TC4A polynuclear second-building units（PSBUs,H4TC4A=p-tert-Butylthiacalix[4]arene） are *** two MCCs,abbreviated as Co4-SiW12and Co8-SiW12,were structurally characterized by single-crystal X-ray *** Co4-SiW12is a supramolecular structure involving cation-anion interactions between Co4-TC4A PSBUs and ***8-SiW12features "barbell-like" shapes with two Co4-TC4A coordinatively linked to a Keggin-type SiW12unit on opposite via W-O_t-Co *** coordination of Co4-TC4A to SiW12in Co8-SiW12resulted in significantly improved photothermal conversion performances compared with Co4-SiW12and pure SiW12or ***-gap determination and photophysical analysis combined with density functional theory calculations indicated the binding mode between Co4-TC4A PSBUs and SiW12,which considerably influences the way of electron transport and the conversion ability of visible light to heat.

关键词： Modular construction Photothermal conversion Polyoxometalate Thiacalixarene

The complex roles of m^(6)A modifications in neural stem cell proliferation, differentiation, and self-renewal and implications for memory 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6期20卷 1582-1598页

作者： Yanxi li Jing xue Yuejia Ma Ke Ye xue Zhao Fangliang Ge Feifei Zheng Lulu liu Xu Gao Dayong Wang Qing Xia Department of Biochemistry and Molecular Biology College of Basic Medical SciencesHarbin Medical UniversityHarbinHeilongjiang ProvinceChina College of Basic Medical Sciences Harbin Medical UniversityHarbinHeilongjiang ProvinceChina Basic Medical Institute Heilongjiang Academy of Medical SciencesHarbinHeilongjiang ProvinceChina Key Laboratory of Heilongjiang Province for Genetically Modified Animals Harbin Medical UniversityHarbinHeilongjiang ProvinceChina Translational Medicine Research and Cooperation Center of Northern China Heilongjiang Academy of Medical SciencesHarbinHeilongjiang ProvinceChina Dongzhimen Hospital Beijing University of Chinese MedicineBeijingChina

N6-methyladenosine(m^(6)A), the most prevalent and conserved RNA modification in eukaryotic cells, profoundly influences virtually all aspects of mRNA metabolism. mRNA plays crucial roles in neural stem cell genesis and neural regeneration, where it is highly concentrated and actively involved in these processes. Changes in m^(6)A modification levels and the expression levels of related enzymatic proteins can lead to neurological dysfunction and contribute to the development of neurological diseases. Furthermore, the proliferation and differentiation of neural stem cells, as well as nerve regeneration, are intimately linked to memory function and neurodegenerative diseases. This paper presents a comprehensive review of the roles of m^(6)A in neural stem cell proliferation, differentiation, and self-renewal, as well as its implications in memory and neurodegenerative diseases. m^(6)A has demonstrated divergent effects on the proliferation and differentiation of neural stem cells. These observed contradictions may arise from the time-specific nature of m^(6)A and its differential impact on neural stem cells across various stages of development. Similarly, the diverse effects of m^(6)A on distinct types of memory could be attributed to the involvement of specific brain regions in memory formation and recall. Inconsistencies in m^(6)A levels across different models of neurodegenerative disease, particularly Alzheimer's disease and Parkinson's disease, suggest that these disparities are linked to variations in the affected brain regions. Notably, the opposing changes in m^(6)A levels observed in Parkinson's disease models exposed to manganese compared to normal Parkinson's disease models further underscore the complexity of m^(6)A's role in neurodegenerative processes. The roles of m^(6)A in neural stem cell proliferation, differentiation, and self-renewal, and its implications in memory and neurodegenerative diseases, appear contradictory. These inconsistencies may be a

关键词： Alzheimer's disease cell self-renewal central nervous system memory microglia nerve regeneration neurodegenerative diseases neurogenesis RNA methylation

Observation of acoustic hybrid topological phases induced by the p-d orbital interactions

在线全文

学校读者我要写书评

暂无评论

Science China(Physics,Mechanics & Astronomy) 2025年第1期Mechanics & Astronomy) .卷 132-139页

作者： Xiao Xiang Feng Gao Yu-Gui Peng Peng Wu Zong-lin li xue-Feng Zhu School of Physics and Innovation Institute Huazhong University of Science and Technology

In spinless systems,growing attention has recently been attracted to synthetic gauge fields,which redefine the fundamental crystal symmetries by utilizing the projective algebraic ***,synthetic gauge fields have been commonly investigated in single orbital systems,and explorations on hybrid orbitals remain elusive in ***,we report the experimental observation of hybrid topological phases induced by hybrid synthetic gauge flux,which is formed by the interaction between p and d orbitals embedded in acoustic *** breaking the translation symmetries of Lxand/or Ly,we unambiguously demonstrate the first-order M?bius edge states and higher-order corner *** work establishes a platform for seeking exotic topological phases induced by hybrid-orbitals interactions and initializing the framework of hybrid-orbitalsrelated topological *** applications can be anticipated in scenarios such as high-quality sensing and robust wave trapping due to the stepwise energy distribution of the hybrid topological phase.

关键词： topological insulator synthetic gauge field projective symmetry corner states

Influence of introducing Zr,Ti,Nb and Ce elements on externally solidified crystals and mechanical properties of high-pressure die-casting Al–Si alloy

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

International Journal of Minerals,Metallurgy and Materials 2025年第1期32卷 147-153页

作者： Junjie li Wenbo Yu Zhenyu Sun Weichen Zheng liangwei Zhang Yanling xue Wenning liu Shoumei Xiong Center of Materials Science and Engineering School of Mechanical and Electronic Control EngineeringBeijing Jiaotong UniversityBeijing 100044China Traffic&Transportation Engineering Central South UniversityChangsha 410075China Shanghai Synchrotron Radiation Facility Shanghai Advanced Research InstituteChinese Academy of SciencesShanghai 201204China School of Materials Science and Engineering Tsinghua UniversityBeijing 100084China

High pressure die casting(HPDC)AlSi10Mn Mg alloy castings are widely used in the automobile *** can optimize the mechanical properties of castings through heat treatment,while the release of thermal stress arouses the deformation of large integrated ***,the development of non-heat treatment Al alloys is becoming the hot *** addition,HPDC contains externally solidified crystals(ESCs),which are detrimental to the mechanical properties of *** achieve high strength and toughness of non-heat treatment die-casting Al-Si alloy,we used AlSi9Mn alloy as matrix with the introduction of Zr,Ti,Nb,and *** influences on ESCs and mechanical properties were systematically investigated through three-dimensional reconstruction and thermodynamic *** results reveal that the addition of Ti increased ESCs'size and porosity,while the introduction of Nb refined ESCs and decreased ***,large-sized Al_3(Zr,Ti)phases formed and degraded the mechanical *** introduction of Ce resulted in the poisoning effect and reduced mechanical properties.

关键词： aluminium alloy high-pressure die-casting externally solidified crystals non-heat treatment