语言大模型 / LLM

VitaLLM: A Versatile and Tiny Accelerator for Mixed-Precision LLM Inference on Edge Devices

VitaLLM：一种 versatile 且 tiny 的混合精度 LLM 推理边缘设备加速器

Zi-Wei Lin, Tian-Sheuan Chang

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 VitaLLM 通过 ternary 权重模型和高效计算核心，实现边缘设备上的高效混合精度 LLM 推理，减少 KV 通信并提升利用效率。

Comments accepted in ISCAS 2026

RCW-CIM: A Digital CIM-based LLM Accelerator with Read-Compute/Write

RCW-CIM: 一种基于数字计算-in-内存的LLM加速器及读-计算/写架构

Yan-Cheng Guo, Tian-Sheuan Chang, Jian-Wei Su

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文提出RCW-CIM架构，通过减少权重更新延迟和非线性操作融合，实现LLM加速，使Llama2-7B模型解码延迟降低21.59%，并提升能效。

Comments accepted in ISCAS 2026

Recommendations for Efficient and Responsible LLM Adoption within Industrial Software Development

工业软件开发中高效负责任的LLM采用建议

Krishna Ronanki, Beatriz Cabrero-Daniel, Tomas Herda, Stefan Sitkovich, Jennifer Horkoff, Christian Berger

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文提出七条可行建议，旨在解决工业软件环境中高效且负责任地采用LLM的实践缺口，通过多案例研究和问卷调查验证了建议的相关性。

Comments Accepted for publication in the Information and Software Technology Journal

Black-Box Skill Stealing Attack from Proprietary LLM Agents: An Empirical Study

从专有LLM代理中黑盒技能窃取攻击：实证研究

Zihan Wang, Rui Zhang, Yu Liu, Chi Liu, Qingchuan Zhao, Hongwei Li, Guowen Xu

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文实证研究了针对LLM代理系统的黑盒技能窃取攻击，发现技能窃取比传统提示窃取更具危害性，通过自动化攻击生成框架评估了商业代理平台和LLM，提出防御机制以降低技能泄露风险。

Comments Preprint

A Task Decomposition and Planning Framework for Efficient LLM Inference in AI-Enabled WiFi-Offload Networks

一种用于AI赋能Wi-Fi卸载网络中高效LLM推理的任务分解与规划框架

Mingqi Han, Xinghua Sun

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文提出一种用户-边缘协作框架，通过LLM规划器实现任务分解与子任务难度预测，优化多用户多边缘Wi-Fi网络中的子任务分配与执行，降低延迟并提升整体性能。

Comments 7 pages, 4 figures, conference version

Break the Optimization Barrier of LLM-Enhanced Recommenders: A Theoretical Analysis and Practical Framework

突破LLM增强推荐系统优化障碍：理论分析与实践框架

Zhangchi Zhu, Wei Zhang

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文通过理论分析和实践框架，解决LLM增强推荐系统中优化障碍问题，提出TF-LLMER框架，通过嵌入规范化和Rec-PCA方法提升推荐性能。

DeepParse: Hybrid Log Parsing with LLM-Synthesized Regex Masks

DeepParse: 基于LLM合成的混合日志解析

Amir Shetaia, Sean Kauffman

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文提出DeepParse，通过LLM提取日志变量模式并结合Drain算法实现高效准确的日志解析，提升提取精度与一致性，降低误报并提升效率。

UniEP: Unified Expert-Parallel MoE MegaKernel for LLM Training

UniEP：统一的专家并行MoE MegaKernel用于LLM训练

Size Zheng, Xuegui Zheng, Li-wen Chang, Jidong Zhai

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文提出UniEP，通过统一专家并行优化策略，将MoE通信与计算融合为MegaKernel，实现自动化适应性调整，提升LLM训练效率并保持精度。

CHICO-Agent: An LLM Agent for the Cross-layer Optimization of 2.5D and 3D Chiplet-based Systems

CHICO-Agent：一种用于2.5D和3D芯片片上系统的跨层优化LLM代理

Qihang Wu, Aman Arora, Vidya A. Chhabria

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 针对2.5D/3D芯片片上系统设计复杂性问题，提出CHICO-Agent框架，通过LLM驱动优化，降低成本并提供可解释的审计跟踪。

PlanCompiler: A Deterministic Compilation Architecture for Structured Multi-Step LLM Pipelines

PlanCompiler：一种确定性编译架构用于结构化多步骤LLM流水线

Pranav Harikumar

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 PlanCompiler通过类型节点注册表、静态图验证和确定性编译分离规划与执行，提升结构化LLM流水线的可靠性与效率，实现高成功率和成本效益。

Comments 31 pages, 1 figure, 7 tables, includes appendices and reproduction details

TRACE: Evaluating Execution Efficiency of LLM-Based Code Translation

TRACE: 评估基于LLM的代码翻译的执行效率

Zhihao Gong, Zeyu Sun, Dong Huang, Qingyuan Liang, Jie M. Zhang, Dan Hao

专题命中 效率与部署 ：LLM(title,title_cn)；large language model(abstract)；language model(abstract)

AI总结 本文提出TRACE基准，评估LLM翻译代码的执行效率，发现正确性不等于效率，23.5%的正确翻译存在显著低效，提示需关注效率意识。

Comments I wrongly uploaded twice the same paper; see arXiv:2508.11468

Failure-Aware Enhancements for Large Language Model (LLM) Code Generation: An Empirical Study on Decision Framework

面向大语言模型（LLM）代码生成的失败意识增强：决策框架的实证研究

Jianru Shen, Zedong Peng, Lucy Owen

专题命中 效率与部署 ：large language model(title,abstract)；language model(title,abstract)；LLM(title)；prompting(abstract)

AI总结 本文通过实证研究，提出了一种基于失败特征的决策框架，用于指导大语言模型代码生成中的增强策略选择。

Comments Accepted at SANER 2026

Evaluating the Impact of Post-Training Quantization on Large Language Models for Code Generation

评估后训练量化对代码生成大语言模型的影响

Alessandro Giagnorio, Antonio Mastropaolo, Saima Afrin, Massimiliano Di Penta, Gabriele Bavota

专题命中 效率与部署 ：large language model(title,abstract)；language model(title,abstract)；post-training(title)；LLM(abstract)

AI总结 本文研究了大语言模型在代码生成中的量化影响，通过更大型的模型和最新量化技术，发现4位精度可显著降低内存占用而不影响性能。

TZ-LLM: Protecting On-Device Large Language Models with Arm TrustZone

Xunjie Wang, Jiacheng Shi, Zihan Zhao, Yang Yu, Zhichao Hua, Jinyu Gu

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

NAMO-LLM: Efficient Navigation Among Movable Obstacles with Large Language Model Guidance

Yuqing Zhang, Yiannis Kantaros

机构 * Department of Electrical and Systems Engineering, McKelvey School of Engineering, Washington University in St.Louis（电气与系统工程系，麦凯利工程学院，华盛顿大学圣路易斯分校） 纠错

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

Comments 9 pages, 6 figures

Large Language Models, and LLM-Based Agents, Should Be Used to Enhance the Digital Public Sphere

Seth Lazar, Luke Thorburn, Tian Jin, Luca Belli

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

Toward Edge General Intelligence with Multiple-Large Language Model (Multi-LLM): Architecture, Trust, and Orchestration

Haoxiang Luo, Yinqiu Liu, Ruichen Zhang, Jiacheng Wang, Gang Sun, Dusit Niyato, Hongfang Yu, Zehui Xiong, Xianbin Wang, Xuemin Shen

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

FAS-LLM: Large Language Model-Based Channel Prediction for OTFS-Enabled Satellite-FAS Links

Halvin Yang, Sangarapillai Lambotharan, Mahsa Derakhshani

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

Comments 12 pages, 8 figures, submitted to JSAC

DART-LLM: Dependency-Aware Multi-Robot Task Decomposition and Execution using Large Language Models

Yongdong Wang, Runze Xiao, Jun Younes Louhi Kasahara, Ryosuke Yajima, Keiji Nagatani, Atsushi Yamashita, Hajime Asama

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

Comments The work was first submitted to an IEEE conference on September 15, 2024

Botfip-LLM: An Enhanced Multimodal Scientific Computing Framework Leveraging Knowledge Distillation from Large Language Models

Tianhao Chen, Pengbo Xu, Pengbo Xu

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

LLM-Slice: Dedicated Wireless Network Slicing for Large Language Models

Boyi Liu, Jingwen Tong, Jun Zhang

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

LLM in a flash: Efficient Large Language Model Inference with Limited Memory

Keivan Alizadeh, Iman Mirzadeh, Dmitry Belenko, Karen Khatamifard, Minsik Cho, Carlo C Del Mundo, Mohammad Rastegari, Mehrdad Farajtabar

专题命中 效率与部署 ：large language model(title,abstract)；language model(title,abstract)；LLM(title)；分类 cs.CL、cs.AI、cs.LG

Comments ACL 2024

Hierarchical Micro-Segmentations for Zero-Trust Services via Large Language Model (LLM)-enhanced Graph Diffusion

Yinqiu Liu, Guangyuan Liu, Hongyang Du, Dusit Niyato, Jiawen Kang, Zehui Xiong, Dong In Kim, Xuemin Shen

专题命中 效率与部署 ：LLM(title,abstract)；large language model(title,abstract)；language model(title,abstract)

Comments 13 pages

FinGPT-HPC: Efficient Pretraining and Finetuning Large Language Models for Financial Applications with High-Performance Computing

Xiao-Yang Liu, Jie Zhang, Guoxuan Wang, Weiqing Tong, Anwar Walid

专题命中 效率与部署 ：large language model(title,abstract)；language model(title,abstract)；pretraining(title)；分类 cs.CL、cs.AI、cs.LG

Activation Sparsity Opportunities for Compressing General Large Language Models

Nobel Dhar, Bobin Deng, Md Romyull Islam, Kazi Fahim Ahmad Nasif, Liang Zhao, Kun Suo

专题命中 效率与部署 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract,comments)；small language model(abstract)

Comments pp. 1-9, doi: 10.1109/IPCCC59868.2024.10850382. keywords: {Accuracy;Prefetching;Large language models;Computational modeling;Companies;Transformers;User experience;Time factors;Tuning;Guidelines;Large Language Models (LLMs);AI Compression;Activation Sparsity;Edge LLM},

Journal ref 2024 IEEE International Performance, Computing, and Communications Conference (IPCCC), Orlando, FL, USA, 2024

Wasted large language models: A life cycle thinking approach

被浪费的大语言模型：一种生命周期思维方法

Erik Johannes Husom, Maria Emine Nylund, Ophelia Prillard

专题命中 效率与部署 ：large language model(title,abstract)；language model(title,abstract)；LLM(abstract)；分类 cs.LG

AI总结 该研究将生命周期思维与欧盟废弃物层级应用于LLMs，提出通过预防、再利用等措施减少LLMs废弃物及环境影响，强调预防不必要使用的重要性。

Comments 5 pages, 1 figures. Accepted at the 2nd International Workshop on Low Carbon Computing (LOCO 2026), Lancaster University, United Kingdom, 10-11 September 2026. Part of the LOCO 2026 proceedings, arXiv: LOCO2026/P14

Toward Federated Large Language Models in Medicine: A Parameter-Efficient Framework for Privacy-Preserving, Multi-Institutional Adaptation

一种联邦学习和参数高效的大型语言模型在医学中的训练框架

Anran Li, Yuanyuan Chen, Wenjun Long, Yu Yin, Yan Hu, Hyunjae Kim, Weipeng Zhou, Yujia Zhou, Hongyi Peng, Yang Ren, Xuguang Ai, Zhenyue Qin, Ming Hu, Xiaoxiao Li, Han Yu, Yih-Chung Tham, Lucila Ohno-Machado, Hua Xu, Qingyu Chen

专题命中 效率与部署 ：large language model(title,abstract)；language model(title,abstract)；LLM(summary_cn,abstract)；分类 cs.CL

AI总结 本文提出Fed-MedLoRA框架，通过低秩适配器参数实现医学LLM的联邦学习，提升跨机构异质性下的模型收敛性和泛化能力。

Comments 41 pages, 11 tables, 3 figures; Just accepted

Benchmarking Trustworthiness of SLMs: Pre-trained vs. Compressed

小型语言模型（SLM）的可信性基准测试：预训练模型与压缩模型的对比

Haokun Lin, Kaijie Zhu, Haobo Xu, Yichen Wu, Zhichao Lu, Qingfu Zhang, Zhenan Sun

机构 * Institute of Automation, CAS（中国科学院自动化研究所） ； Tsinghua University（清华大学） ； Harvard Medical School（哈佛医学院） ； City University of Hong Kong（香港城市大学） 纠错

专题命中 效率与部署 ：SLM(title_cn,summary_cn)；LLM(abstract_cn)；large language model(abstract)；language model(abstract)

AI总结 本研究评估SLM的可信性，发现量化压缩预训练模型比剪枝更能保留可信性，且量化压缩可靠大模型得到的SLM比从头训练的小型模型更优，知识蒸馏可进一步提升其可靠性。

Comments Published in IJCNN 2026

Abstract representational geometry supports inference in large language models

抽象表征几何支持大型语言模型中的推理

Yunan Zeng, Yuwang Wang

机构 * College of Future Information Technology, Fudan University（复旦大学未来信息科技学院） ； Beijing National Research Center for Information Science and Technology, Tsinghua University（北京信息科学与技术国家研究中心，清华大学） 纠错

专题命中 效率与部署 ：large language model(title,abstract)；language model(title,abstract)；LLM(summary_cn,abstract_cn)；分类 cs.AI

AI总结 本研究通过文本版情境反转学习范式，比较人类与LLM的行为和表征，发现LLM内部状态在推理时形成类似海马体的抽象几何结构，且该结构分层组织，高层次表征与推理相关。

Efficient and Stealthy Jailbreak Attacks via Adversarial Prompt Distillation from LLMs to SLMs

通过从大语言模型到小语言模型的对抗性提示蒸馏实现高效且隐蔽的越狱攻击

Xiang Li, Chong Zhang, Jia Wang, Fangyu Wu, Yushi Li, Xiaobo Jin

机构 * Xi’an Jiaotong-Liverpool University（西安交通大学利物浦大学） ； The Chinese University of Hong Kong（香港中文大学） ； University of Liverpool（利物浦大学） 纠错

专题命中 效率与部署 ：LLM(summary_cn,abstract)；SLM(summary_cn,abstract_cn)；large language model(abstract)；language model(abstract)

AI总结 提出对抗性提示蒸馏（APD）框架，将LLM的越狱能力迁移到SLM，实现高效低资源攻击，在GPT-4上达到96.4%攻击成功率，速度提升3.7倍，参数减少11.3倍。

Comments 24 pages, 3 figures