检索结果-内蒙古大学图书馆

Addressing Activation Outliers in LLMs: A Systematic Review of Post-Training Quantization Techniques

IEEE Access 2025年 13卷 81917-81932页

作者： Patrik Czakó Gábor Kertész Sándor Szénási Doctoral School of Applied Informatics and Applied Mathematics Óbuda University Budapest Hungary John von Neumann Faculty of Informatics Óbuda University Budapest Hungary Laboratory of Parallel and Distributed Systems Institute for Computer Science and Control (SZTAKI) Hungarian Research Network (HUN-REN) Budapest Hungary Faculty of Economics and Informatics J. Selye University Komárno Slovakia

Large Language Models (LLMs) have transformed natural language processing, yet their deployment remains challenging due to substantial computational, memory, and energy demands. Post-training quantization has emerged as a key strategy for enabling efficient inference, particularly in resource-constrained settings. This systematic review focuses on weight-activation quantization, with a unique emphasis on the emergent outlier phenomenon in LLM activations. This work evaluates recent techniques that mitigate activation outliers and improve quantization efficiency, distinguishing itself from prior reviews. Using the PRISMA methodology, we examine 52 recent studies to uncover key trends and evaluate the effectiveness of different approaches. By synthesizing insights from these works, this review presents a diverse set of techniques and their implications for activation quantization, laying the groundwork for future research and practical advancements in LLM deployment.

关键词： Quantization (signal) Computational modeling Throughput Arithmetic Decoding Adaptation models Transformers Systematics Systematic literature review Large language models

来源：评论

学校读者我要写书评

暂无评论

Grid application monitoring and debugging using the mercury monitoring system

Grid application monitoring and debugging using the mercury ...

引用

European Grid Conference on Advances in Grid Computing - EGC 2005

作者： Gombás, Gábor Marosi, Csaba Attila Balaton, Zoltán MTA SZTAKI Laboratory of Parallel and Distributed Systems P.O. Box 63 H-1518 Budapest Hungary

The Mercury Monitoring System is a generic GMA-compatible grid monitoring framework. In this paper we present how higher level application monitoring and debugging facilities can be built on top of Mercury. Currently two higher level application monitoring systems are implemented using Mercury: one is GRM/PROVE and the other is a stand-alone MPI wrapper library that can be used to learn about the communication patterns of MPI programs. Remote debugging of applications has also been implemented on top of Mercury by using the remote debugging facilities of gdb (the GNU Debugger) which is also presented in this paper. © Springer-Verlag Berlin Heidelberg 2005.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Workflow support for complex grid applications: Integrated and portal solutions

Workflow support for complex grid applications: Integrated a...

引用

2nd European AcrossGrids Conference, AxGrids 2004

作者： Lovas, Róbert Dózsa, Gábor Kacsuk, Péter Podhorszki, Norbert Drótos, Dániel MTA SZTAKI Laboratory of Parallel and Distributed Systems P.O. Box 63 Budapest Hungary

ISBN: (纸本)3540228888

In this paper we present a workflow solution to support graphically the design, execution, monitoring, and performance visualisation of complex grid applications. The described workflow concept can provide interoperability among different types of legacy applications on heterogeneous computational platforms, such as Condor or Globus based grids. The major design and implementation issues concerning the integration of Condor tools, Mercury grid monitoring infrastructure, PROVE performance visualisation tool, and the new workflow layer of P-GRADE are discussed in two scenarios. The integrated version of P-GRADE represents the thick client concept, while the portal version needs only a thin client and can be accessed by a standard web browser. To illustrate the application of our approach in the grid, an ultra-short range weather prediction system is presented that can be executed in a grid testbed and visualised not only at workflow level but at the level of individual parallel jobs, too. © Springer-Verlag Berlin Heidelberg 2004.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Matching matchings

Matching matchings

引用

2013 IEEE 3rd Eastern European Regional Conference on the Engineering of Computer Based systems, ECBS-EERC 2013

作者： Bacsó, Gábor Keszler, Anita Tuza, Zsolt Laboratory of Parallel and Distributed Systems MTA SZTAKI Hungary Distributed Events Analysis Research Group MTA SZTAKI Hungary Rényi Institute Budapest Hungary University of Pannonia Veszprém Hungary

ISBN: (纸本)9780769550640

This paper presents the first steps toward a graph comparison method based on matching matchings, or in other words, comparison of independent edge sets in graphs. The novelty of our approach is to use matchings for calculating distance of graphs in case of edge-colored graphs. This idea can be used as a preprocessing step of graph querying applications, to speed up exact and inexact graph matching methods. We introduce the notion of colored matchings and prove some properties of them in edge colored complete graphs and complete bipartite graphs in case of two colors. © 2013 IEEE.

关键词： Graph theory

来源：评论

学校读者我要写书评

暂无评论

WS-PGRADE: Supporting parameter sweep applications in workflows

WS-PGRADE: Supporting parameter sweep applications in workfl...

引用

Workshop on Workflows in Support of Large-Scale Science, WORKS

作者： Peter Kacsuk Krisztian Karoczkai Gabor Hermann Gergely Sipos Jozsef Kovacs Parallel and Distributed Systems Laboratory MTA SZTAKI Budapest Hungary

Despite majority of production level grid applications are either parameter studies or workflows some of them cannot fit into these categories and implement more complex structures. Such an application class was found in the CancerGrid project, where several parameter study simulation must be connected together into a workflow and must be executed on various production grids. The paper introduces the structure of parameter study based grid applications and describes the WS-PGRADE system which provides support for the development and execution of this category of applications. The data generator, data collector and data processor components of WS-PGRADE provide easy to understand, flexible, though very powerful mechanisms to express involved grid application structures. The grid connector layer of WS-PGRADE can distribute application components to Desktop Grids, service grids, Web Service networks and local resources, thus can find the most efficient location to execute calculations. By using WS-PGRADE, the members of the CancerGrid consortium are able to create and execute grid applications that consist of millions of jobs.

关键词： Portals Engines Mesh generation Web services Feedback Laboratories Production systems Connectors Collaborative tools Collaborative work

来源：评论

学校读者我要写书评

暂无评论

Brokering Solution for Science Gateways Using Multiple distributed Computing Infrastructures

Brokering Solution for Science Gateways Using Multiple Distr...

引用

International Workshop on Science Gateways (IWSG)

作者： Krisztian Karoczkai Attila Kertesz Peter Kacsuk MTA SZTAKI Laboratory of Parallel and Distributed Systems Budapest Hungary

The ever growing number of computation-intensive applications calls for the interoperation of distributed infrastructures such as Clouds, Grids and private clusters. The European SHIWA and ER-flow projects have been initiated to enable the combination of heterogeneous scientific workflows, and to execute them in a large-scale system consisting of multiple distributed Computing Infrastructures including Grids and Clouds. One of the resource management challenges of these projects called parameter study job scheduling. A parameter study job of a workflow has a large number of input files to be consumed by independent job instances. In this paper we propose a meta-brokering solution for these workflows to be applied in science gateways. To cope with the high uncertainty and unpredictable load of the utilized infrastructures, we rely on resource priority services. These tools are capable of determining and dynamically updating priorities of the available infrastructures to be selected for job instances.

关键词： Logic gates Time measurement distributed computing Europe Concrete Resource management Workflow management software

来源：评论

学校读者我要写书评

暂无评论

Machine learning methods in Smartphone-Based Activity Recognition

Machine learning methods in Smartphone-Based Activity Recogn...

引用

International Symposium on Applied Computational Intelligence and Informatics ( SACI)

作者： Istvan Pintye Laboratory of Parallel and Distributed Systems Institute for Computer Science and Control (SZTAKI) Hungary

ISBN: (数字)9781728173771

ISBN: (纸本)9781728173788

In this paper, we present a system for human physical Activity Recognition (AR) using smartphone with embedded sensors. This paper addresses the question whether there is a comfortable way to predict human activities based on collected data from smartphone embedded gyroscope and accelerometer. Computational background of this work based on self-learning machine learning methods. In order to train the machine learning algorithms, The University of California, Irvine (UCI) dataset was used and the different models were compared. After selecting the best model further modifications were suggested in order to improve the accuracy of the model. At the end 96.88% accuracy was reached.

关键词：

来源：评论

学校读者我要写书评

暂无评论

LCG-2 resource broker support in P-GRADE portal

LCG-2 resource broker support in P-GRADE portal

引用

MIPRO 2006 - 29th International Convention: Microelectronics, Electronics and Electronic Technologies, MEET, Hypermedia and Grid systems HGS

作者： Tóth, Adrián Kacsuk, Peter Department of Information Science University of Miskolc H-3515 Miskolc Egyetemváros Hungary MTA SZTAKI Laboratory of Parallel and Distributed Systems P.O. Box 63 H-1518 Budapest Hungary

ISBN: (纸本)9532330186

The P-GRADE portal is a workflow-oriented Grid portal that originally served Globus-2 Grids. The Portal introduced an easy-to-use workflow-centric interface on top of the job oriented Globus systems. Because Globus does not contain any resource broker, users of the P-GRADE Portal had to map workflow components onto physical resources manually. However, just recently, most of the production Grids have moved from pure Globus middleware to more advanced solutions that include broker services as well. This tendency is clearly amplified by the emergence of LCG-2 middleware, the results of EGEE, the biggest Grid project of the World. The paper describes the broker-enabled version of the P-GRADE Portal. This portal version is capable to submit workflow components into LCG2 based Grids using the Resource Broker service. This support makes the usage of production Grids significantly easier since end-users are relived from the complicated task of resource selection. © 2006 by MIPRO All rights reserved.

关键词： Middleware

来源：评论

学校读者我要写书评

暂无评论

A Workload-Aware Energy Model for Virtual Machine Migration

A Workload-Aware Energy Model for Virtual Machine Migration

引用

IEEE International Conference on Cluster Computing

作者： Vincenzo De Maio Gabor Kecskemeti Radu Prodan Institute of Computer Science University of Innsbruck MTA SZTAKI Laboratory of Parallel and Distributed Systems

Energy consumption has become a significant issue for data centres. Assessing their consumption requires precise and detailed models. In the latter years, many models have been proposed, but most of them either do not consider energy consumption related to virtual machine migration or do not consider the variation of the workload on (1) the virtual machines (VM) and (2) the physical machines hosting the VMs. In this paper, we show that omitting migration and workload variation from the models could lead to misleading consumption estimates. Then, we propose a new model for data centre energy consumption that takes into account the previously omitted model parameters and provides accurate energy consumption predictions for paravirtualised virtual machines running on homogeneous hosts. The new model's accuracy is evaluated with a comprehensive set of operational scenarios. With the use of these scenarios we present a comparative analysis of our model with similar state-of-the-art models for energy consumption of VM Migration, showing an improvement up to 24% in accuracy of prediction.

关键词： Energy consumption Power demand Memory management Computational modeling Virtual machining Predictive models Hardware

来源：评论

学校读者我要写书评

暂无评论

On the algorithms of the grid-based EMMIL E-marketplace model

On the algorithms of the grid-based EMMIL E-marketplace mode...

引用

28th International Convention Proceedings: Microelectronics, Electronics and Electronic Technologies MEET, Hypermedia and Grid systems HGS 2OO5

作者： Bruckner, Lívia Kacsukné Hermann, Gábor Dept. of Methodology and Information Systems International Business School 2-4 Tárogató út Budapest 1021 Hungary Laboratory of Parallel and Distributed Systems MTA SZTAKI 13-17. Kende u Budapest 1111 Hungary

ISBN: (纸本)9532330119

Previous publications of the authors have described the concept of the Grid based EMMIL emarketplace model, a new business model powered by Grid technology to improve supply chain efficiency. EMMIL. E-Marketplace Model Integrated with Logistics is a new model that brings together three entity types. seller buyer and third party logistics service provider - in the same transaction optimising the total costs and maximising the revenue. After summarising the main features of the business model and the three sided auction algorithms this paper will focus on the algorithmic questions of the repetitive optimisation showing the way how the Marketplace Support Grid can help to achieve the optimal solution within a short time frame that is necessary for the on-line business environment.

关键词： Supply chains

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：