语言大模型 / LLM - arXivDaily 专题

2410.17259 2024-10-24 cs.NI cs.IT cs.LG math.IT 90%

Large Language Models for Knowledge-Free Network Management: Feasibility Study and Opportunities

Hoon Lee, Mintae Kim, Seunghwan Baek, Namyoon Lee, Merouane Debbah, Inkyu Lee

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

2403.00067 2024-10-22 cs.CL 90%

Query-OPT: Optimizing Inference of Large Language Models via Multi-Query Instructions in Meeting Summarization

Md Tahmid Rahman Laskar, Elena Khasanova, Xue-Yong Fu, Cheng Chen, Shashi Bhushan TN

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

Comments Accepted at EMNLP 2024 (Industry Track)

2410.12462 2024-10-17 cs.CL 90%

Bridging the Language Gaps in Large Language Models with Inference-Time Cross-Lingual Intervention

Weixuan Wang, Minghao Wu, Barry Haddow, Alexandra Birch

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

2311.09799 2024-10-15 cs.CL 90%

How Far Can We Extract Diverse Perspectives from Large Language Models?

Shirley Anugrah Hayati, Minhwa Lee, Dheeraj Rajagopal, Dongyeop Kang

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

Comments Accepted at EMNLP 2024 Main Conference

2409.12740 2024-09-20 cs.IR cs.AI 90%

HLLM: Enhancing Sequential Recommendations via Hierarchical Large Language Models for Item and User Modeling

Junyi Chen, Lu Chi, Bingyue Peng, Zehuan Yuan

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

2408.04413 2024-08-09 cs.LG cs.AR 90%

Deeploy: Enabling Energy-Efficient Deployment of Small Language Models On Heterogeneous Microcontrollers

Moritz Scherer, Luka Macan, Victor Jung, Philip Wiese, Luca Bompani, Alessio Burrello, Francesco Conti, Luca Benini

专题命中效率与部署：language model(title,abstract)；small language model(title,abstract)；foundation model(abstract)；SLM(abstract)

Comments Accepted for publication at ESWEEK - CASES 2024

2406.04640 2024-06-10 cs.LG 90%

LinkGPT: Teaching Large Language Models To Predict Missing Links

Zhongmou He, Jing Zhu, Shengyi Qian, Joyce Chai, Danai Koutra

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

2310.09497 2024-05-31 cs.IR cs.AI 90%

A Setwise Approach for Effective and Highly Efficient Zero-shot Ranking with Large Language Models

Shengyao Zhuang, Honglei Zhuang, Bevan Koopman, Guido Zuccon

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

Comments SIGIR2024 full paper

2402.10866 2024-05-29 cs.CL 90%

EcoRank: Budget-Constrained Text Re-ranking Using Large Language Models

Muhammad Shihab Rashid, Jannat Ara Meem, Yue Dong, Vagelis Hristidis

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

Comments Accepted to Findings of ACL 24

2312.15224 2024-01-10 cs.AI cs.HC 90%

LLM-Powered Hierarchical Language Agent for Real-time Human-AI Coordination

Jijia Liu, Chao Yu, Jiaxuan Gao, Yuqing Xie, Qingmin Liao, Yi Wu, Yu Wang

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

Comments This paper is accpeted by AAMAS 2024. More demonstrations can be seen on our website https://sites.google.com/view/overcooked-hla/

2310.08908 2023-10-16 cs.CL 90%

Human-in-the-loop Machine Translation with Large Language Model

Xinyi Yang, Runzhe Zhan, Derek F. Wong, Junchao Wu, Lidia S. Chao

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

Comments Accepted to MT Summit 2023

2309.05951 2023-09-13 cs.CL 90%

Balanced and Explainable Social Media Analysis for Public Health with Large Language Models

Yan Jiang, Ruihong Qiu, Yi Zhang, Peng-Fei Zhang

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

Comments arXiv admin note: text overlap with arXiv:2309.04213

2305.11176 2023-05-25 cs.RO cs.AI 90%

Instruct2Act: Mapping Multi-modality Instructions to Robotic Actions with Large Language Model

Siyuan Huang, Zhengkai Jiang, Hao Dong, Yu Qiao, Peng Gao, Hongsheng Li

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

2211.14133 2023-05-16 cs.LG 90%

PipeFisher: Efficient Training of Large Language Models Using Pipelining and Fisher Information Matrices

Kazuki Osawa, Shigang Li, Torsten Hoefler

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

Comments MLSys 2023

2210.03871 2022-10-11 cs.CL 90%

Data-Efficiency with a Single GPU: An Exploration of Transfer Methods for Small Language Models

Alon Albalak, Akshat Shrivastava, Chinnadhurai Sankar, Adithya Sagar, Mike Ross

专题命中效率与部署：language model(title,abstract)；small language model(title)；large language model(abstract)；instruction tuning(abstract)

2305.17126 2024-03-12 cs.LG cs.AI cs.CL stat.ML 90%

Large Language Models as Tool Makers

Tianle Cai, Xuezhi Wang, Tengyu Ma, Xinyun Chen, Denny Zhou

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

Comments Code available at https://github.com/ctlllll/LLM-ToolMaker

2602.00663 2026-08-24 cs.AI cs.LG q-bio.BM 版本更新 90%

SEISMO: Explanation-Aware, Trajectory-Conditioned LLM Agents for Sample-Efficient Molecular Optimisation

SEISMO：通过轨迹感知的LLM代理提高分子优化的样本效率

Fabian P. Krüger, Andrea Hunklinger, Adrian Wolny, Tim J. Adler, Igor Tetko, Santiago David Villalba

机构 * Technical University of Munich, Germany（慕尼黑技术大学，德国）； TUM School of Computation, Information（TUM计算、信息学院）； Technology, Department of Mathematics（技术学院，数学系）； Helmholtz Munich – German Research Center for Environmental Health (GmbH), Institute of Structural Biology, Molecular Targets（海德堡慕尼黑德国环境健康研究所以及结构生物学研究所，分子靶点）； Therapeutics Center, 85764 Neuherberg, Germany（治疗中心，德国新赫尔伯格85764）； Machine Learning Research（机器学习研究）

专题命中效率与部署：LLM(title,title_cn)；分类 cs.AI、cs.LG

AI总结 SEISMO通过轨迹感知的LLM代理实现高效的分子优化，利用结构化反馈提升样本效率。

Comments Fabian P. Krüger and Andrea Hunklinger contributed equally to this work

2608.17556 2026-08-19 cs.CR cs.CL cs.LG 新提交 90%

Reflex-Guard: A Low-Latency Guardrail for LLM Prompt Safety Using Dense Semantic Embeddings

Reflex-Guard：一种使用稠密语义嵌入的低延迟大语言模型提示安全护栏

Istiaque Ahmed, Afia Anjum Borsha, Ranat Das Prangon, Abu-fuad Ahmad, Thi Hong Tran

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

AI总结 Reflex-Guard是一种本地运行的轻量型LLM提示安全护栏，采用稠密语义嵌入与快速分类器，延迟低至37.6毫秒，召回率达95.9%，性能优于现有基线，可高效检测各类越狱攻击。

2601.18899 2026-08-19 cs.CL cs.AI cs.SD 版本更新 90%

Language Family Matters: Evaluating LLM-Based ASR Across Linguistic Boundaries

语言家族至关重要：评估基于LLM的ASR跨语言边界

Yuchen Zhang, Ravi Shekhar, Haralambos Mouratidis

机构 * Institute for Analytics and Data Science, University of Essex（埃塞克斯大学分析与数据科学研究所）； School of Computer Science and Electronic Engineering, University of Essex（埃塞克斯大学计算机科学与电子工程学院）

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

AI总结本文提出基于语言家族的连接器共享策略，通过减少参数数量并提升跨领域泛化能力，实现更高效的多语言ASR部署。

Comments Accepted by EACL'26 main

2608.15602 2026-08-18 cs.LG cs.AI 新提交 90%

FluxBin: Flexible LUT-based Ultra-low-bit LLM Inference by Algorithm-Kernel Synergy

FluxBin：基于灵活查找表的超低位大语言模型推理——算法与内核协同设计

Qingyao Yang, Runming Yang, He Xiao, Wendong Xu, Junyu Chen, Haobo Liu, Chenchen Ding, Ruihan Hu, Yik-Chung Wu, Ngai Wong

机构 * The University of Hong Kong（香港大学）； Harbin Institute of Technology（哈尔滨工业大学）

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

AI总结本研究针对二值量化LLM推理的硬件内核缺失问题，提出FluxBin算法-内核协同设计，实现超低位推理的加速、能效提升与内存压缩，可高效部署70B规模模型。

2605.20254 2026-08-18 cs.IR cs.AI cs.CV cs.LG 版本更新 90%

Efficient Table QA via TableGrid Navigation and Progressive Inference Prompting

通过表格网格导航和逐步推理提示实现高效的表格问答

Amritansh Maurya, Navjot Singh, Mohammed Javed, Omar Moured

机构 * Vision Intelligence Lab, IIIT Allahabad, Prayagraj, India（视觉智能实验室，印度拉贾斯坦邦阿拉哈巴德）

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

AI总结本文提出了一种无需训练的表格问答方法，通过TableGrid导航和Progressive Inference Prompting框架，提升了表格问答的精度和效率，并在多个数据集上验证了其有效性。

Comments Accepted for Presentation in ICDAR 2026, Vienna, Austria

2506.07449 2026-08-18 cs.IR cs.AI cs.CL 版本更新 90%

LlamaRec-LKG-RAG: A Single-Pass, Learnable Knowledge Graph-RAG Framework for LLM-Based Ranking

LlamaRec-LKG-RAG：一种用于基于大语言模型的排序的单遍可学习知识图谱-RAG框架

Vahid Azizi, Fatemeh Koochaki

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

AI总结该研究提出LlamaRec-LKG-RAG框架，将个性化知识图谱上下文集成到LLM推荐排序中，经ML-100K等数据集实验，在MRR等指标上较LlamaRec取得提升，为知识感知个性化推荐奠定基础。

2608.10037 2026-08-12 cs.LG cs.AI 新提交 90%

DOCSCHISEL: Adaptive Tool Documentation Optimization Framework for LLM Agents

DOCSCHISEL：面向大语言模型智能体的自适应工具文档优化框架

You Lu, Kun Zhang, Bihuan Chen, Xin Peng

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

AI总结该研究针对LLM智能体工具文档的异质性与泛化问题，提出DocsChisel自适应优化框架，经实验较原始文档及EasyTool、DRAFT基线大幅提升任务成功率，且开销有限。

2608.07498 2026-08-11 cs.HC cs.AI cs.LG cs.MA 新提交 90%

Knowing You Is Everything: LLM Agents Achieve Near-Perfect Profile-Consistent Reaction Prediction in Social Media Simulation

了解你就是一切：LLM智能体在社交媒体模拟中实现近乎完美的个人资料一致反应预测

Ljubisa Bojic, Ljiljana Matic, Joerg Matthes, Milan Cabarkapa, Bojana Dinic, Jue Wang

专题命中效率与部署：LLM(title,title_cn)；分类 cs.AI、cs.LG

AI总结本研究通过基准测试12种LLM配置，发现其在社交媒体反应预测中表现优异，可用于推荐系统压力测试，同时也揭示了大规模合成智能体群对公众意见的潜在威胁。

Comments 28 pages, 6 figures

2607.21985 2026-08-11 cs.DC cs.AI cs.AR cs.LG 版本更新 90%

Unified Static-Dynamic Pruning for Efficient LLM Inference

用于高效大语言模型推理的统一静态-动态剪枝

Jinhyeok Kim, Yejoon Lee, Jaeyoung Do

机构 * Seoul National University（首尔国立大学）

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

AI总结研究针对大语言模型推理瓶颈，提出统一静态-动态剪枝框架SPDP，集成非结构化SP与输入自适应DP，设计新格式和内核，经评估其加速效果显著，推进推理效率-质量前沿，提升大规模LLM服务的吞吐量和性能功耗比。

Comments Proceedings of the VLDB Endowment (PVLDB), Volume 19, Issue 11, 2026

2608.05651 2026-08-07 cs.CL cs.AI cs.NE 新提交 90%

Relay, Don't Route: Adaptive Population Handoff for Cost-Efficient LLM-Driven Evolution

中继而非路由：面向成本高效的大语言模型驱动演化的自适应种群交接

Sichun Luo, Yi Huang, Guanzhi Deng, Haibo Wang, Haochen Luo, Lei Li, Zefa Hu, Junlan Feng, Qi Liu

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

AI总结针对LLM驱动演化全程用强模型成本高的问题，提出无需训练的自适应种群交接框架，通过廉价模型探索、中继增益触发交接，在12个测试设置中11个获最高平均分数，性能优于基线。

2608.05303 2026-08-07 cs.AR cs.CL cs.LG 新提交 90%

EdgeXpert: An Edge Device for Memory-Efficient LLM Inference with Mixture-of-Experts and Speculative Decoding

EdgeXpert：一种基于混合专家与推测解码的内存高效型大语言模型推理边缘设备

Sangwoo Ha, Hyunwoo Seo, Yurim Jo, Youngjin Moon, Hoi-Jun Yoo

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

AI总结 EdgeXpert是一款软硬件协同设计的LLM加速器，通过优化预填充与解码阶段的专家处理，解决推测解码与MoE的兼容性问题，在降低延迟与能耗的同时维持精度，适用于边缘设备的LLM推理。

Comments Accepted at the 59th IEEE/ACM International Symposium on Microarchitecture (MICRO 2026)

2608.03036 2026-08-05 cs.SE cs.AI cs.LG 新提交 90%

LLM Serving in the Wild: An Empirical Study of Frameworks, Methods, and System Designs

野外环境下的大语言模型服务：框架、方法与系统设计的实证研究

Forough Majidi, Mohammad Mehdi Morovati, Foutse Khomh, Heng Li

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

AI总结本研究通过实证分析开源软件系统中5种LLM服务框架的应用情况，明确了框架使用特点、常用服务方法及应用场景，为相关人员提供了实践见解。

2608.02680 2026-08-05 cs.SE cs.AI cs.LG 新提交 90%

TraceCompiler: Skill-Guided Mining and Compilation of LLM Agent Traces into Mostly Deterministic Workflows

TraceCompiler：技能引导的LLM智能体轨迹挖掘与编译为近确定性工作流

Salma El Yadouni, Guanyi Li

专题命中效率与部署：LLM(title,title_cn)；分类 cs.AI、cs.LG

AI总结本文提出TraceCompiler系统，通过技能引导挖掘含噪声的LLM智能体轨迹并编译为近确定性工作流，在T1、AppWorld数据集上验证了依赖恢复的高精度，实现了Venmo资金请求意图的调用减少。

Comments 17 pages, 4 figures, 5 tables

2606.01774 2026-08-05 cs.LG cs.AI 版本更新 90%

FLARE: Diffusion for Hybrid Language Model

FLARE: 混合语言模型的扩散方法

Yuchen Zhu, Jing Shi, Chongjian Ge, Hao Tan, Yiran Xu, Wanrong Zhu, Jason Kuen, Koustava Goswami, Rajiv Jain, Yongxin Chen, Molei Tao, Jiuxiang Gu

机构 * Adobe Research（Adobe研究院）； Georgia Institute of Technology（佐治亚理工学院）

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

AI总结提出FLARE框架，通过结合自回归和扩散目标、硬件感知内核和统一推理，将混合注意力LLM转换为支持并行解码的扩散模型，在保持能力的同时提升吞吐量。

1. 效率与部署 22313 篇

Large Language Models for Knowledge-Free Network Management: Feasibility Study and Opportunities

Query-OPT: Optimizing Inference of Large Language Models via Multi-Query Instructions in Meeting Summarization

Bridging the Language Gaps in Large Language Models with Inference-Time Cross-Lingual Intervention

How Far Can We Extract Diverse Perspectives from Large Language Models?

HLLM: Enhancing Sequential Recommendations via Hierarchical Large Language Models for Item and User Modeling

Deeploy: Enabling Energy-Efficient Deployment of Small Language Models On Heterogeneous Microcontrollers

LinkGPT: Teaching Large Language Models To Predict Missing Links

A Setwise Approach for Effective and Highly Efficient Zero-shot Ranking with Large Language Models

EcoRank: Budget-Constrained Text Re-ranking Using Large Language Models

LLM-Powered Hierarchical Language Agent for Real-time Human-AI Coordination

Human-in-the-loop Machine Translation with Large Language Model

Balanced and Explainable Social Media Analysis for Public Health with Large Language Models

Instruct2Act: Mapping Multi-modality Instructions to Robotic Actions with Large Language Model

PipeFisher: Efficient Training of Large Language Models Using Pipelining and Fisher Information Matrices

Data-Efficiency with a Single GPU: An Exploration of Transfer Methods for Small Language Models

Large Language Models as Tool Makers

SEISMO: Explanation-Aware, Trajectory-Conditioned LLM Agents for Sample-Efficient Molecular Optimisation

Reflex-Guard: A Low-Latency Guardrail for LLM Prompt Safety Using Dense Semantic Embeddings

Language Family Matters: Evaluating LLM-Based ASR Across Linguistic Boundaries

FluxBin: Flexible LUT-based Ultra-low-bit LLM Inference by Algorithm-Kernel Synergy

Efficient Table QA via TableGrid Navigation and Progressive Inference Prompting

LlamaRec-LKG-RAG: A Single-Pass, Learnable Knowledge Graph-RAG Framework for LLM-Based Ranking

DOCSCHISEL: Adaptive Tool Documentation Optimization Framework for LLM Agents

Knowing You Is Everything: LLM Agents Achieve Near-Perfect Profile-Consistent Reaction Prediction in Social Media Simulation

Unified Static-Dynamic Pruning for Efficient LLM Inference

Relay, Don't Route: Adaptive Population Handoff for Cost-Efficient LLM-Driven Evolution

EdgeXpert: An Edge Device for Memory-Efficient LLM Inference with Mixture-of-Experts and Speculative Decoding

LLM Serving in the Wild: An Empirical Study of Frameworks, Methods, and System Designs

TraceCompiler: Skill-Guided Mining and Compilation of LLM Agent Traces into Mostly Deterministic Workflows

FLARE: Diffusion for Hybrid Language Model