语言大模型 / LLM - arXivDaily 专题

2510.02483 2025-10-06 cs.LG cs.AI cs.CL 88%

Litespark Technical Report: High-Throughput, Energy-Efficient LLM Training Framework

Nii Osae Osae Dade, Moinul Hossain Rahat

机构 * Mindbeam AI

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

Comments 14 pages

2508.04096 2025-08-07 cs.SD eess.AS 88%

Efficient Scaling for LLM-based ASR

Bingshen Mu, Yiwen Shao, Kun Wei, Dong Yu, Lei Xie

机构 * Tencent AI Lab（腾讯人工智能实验室）

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

Comments Accepted by ASRU 2025

2306.15933 2025-08-05 cs.CL cs.AI cs.LG 88%

You Can Generate It Again: Data-to-Text Generation with Verification and Correction Prompting

Xuan Ren, Zeyu Zhang, Lingqiao Liu

机构 * AIML, University of Adelaide（人工智能实验室，阿德莱德大学）

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

2507.12000 2025-07-18 eess.SP 88%

DSSD: Efficient Edge-Device LLM Deployment and Collaborative Inference via Distributed Split Speculative Decoding

Jiahong Ning, Ce Zheng, Tingting Yang

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

Comments ICML 2025

2507.11959 2025-07-17 cs.CL cs.AI cs.LG 88%

PoTPTQ: A Two-step Power-of-Two Post-training for LLMs

Xinyu Wang, Vahid Partovi Nia, Peng Lu, Jerry Huang, Xiao-Wen Chang, Boxing Chen, Yufei Cui

机构 * McGill University, Canada（麦吉尔大学）； Huawei Noah's Ark Lab, Canada（华为诺亚实验室）； Mila – Quebec AI Institute, Canada（魁北克AI研究院）

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

Comments Accepted at ECAI 2025 (European Conference on Artificial Intelligence)

2506.06698 2025-06-10 cs.AI cs.CL cs.CV cs.LG 88%

Contextual Experience Replay for Self-Improvement of Language Agents

Yitao Liu, Chenglei Si, Karthik Narasimhan, Shunyu Yao

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

Comments Accepted to ACL 2025. 20 pages

2410.02958 2025-06-09 cs.LG cs.AI cs.CL cs.MA 88%

AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML

Patara Trirat, Wonyong Jeong, Sung Ju Hwang

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

Comments ICML 2025, Project Page: https://deepauto-ai.github.io/automl-agent

2506.02208 2025-06-04 cs.LG cs.AI cs.CL 88%

KDRL: Post-Training Reasoning LLMs via Unified Knowledge Distillation and Reinforcement Learning

Hongling Xu, Qi Zhu, Heyuan Deng, Jinpeng Li, Lu Hou, Yasheng Wang, Lifeng Shang, Ruifeng Xu, Fei Mi

机构 * Harbin Institute of Technology (Shenzhen)（哈尔滨工业大学（深圳））； Huawei Noah’s Ark Lab（华为诺亚实验室）

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

2412.03621 2025-06-02 cs.NI 88%

JPPO++: Joint Power and Denoising-inspired Prompt Optimization for Mobile LLM Services

Feiran You, Hongyang Du, Kaibin Huang, Abbas Jamalipour

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

Comments arXiv admin note: text overlap with arXiv:2411.18010

2502.11882 2025-05-29 cs.AI cs.CL cs.HC cs.LG cs.MA 88%

Leveraging Dual Process Theory in Language Agent Framework for Real-time Simultaneous Human-AI Collaboration

Shao Zhang, Xihuai Wang, Wenhao Zhang, Chaoran Li, Junru Song, Tingyu Li, Lin Qiu, Xuezhi Cao, Xunliang Cai, Wen Yao, Weinan Zhang, Xinbing Wang, Ying Wen

机构 * Shanghai Jiao Tong University（上海交通大学）； Meituan（美团）； Intelligent Game and Decision Laboratory（智能游戏与决策实验室）

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

Comments Accepted by ACL 2025 Main. Camera Ready Version

2504.07428 2025-04-11 cs.IT cs.NI math.IT 88%

Task-oriented Age of Information for Remote Inference with Hybrid Language Models

Shuying Gan, Xijun Wang, Chenyuan Feng, Chao Xu, Howard H. Yang, Xiang Chen, Tony Q. S. Quek

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

Comments accepted by ICCCS 2025

2504.03360 2025-04-07 cs.CY cs.AI cs.CL cs.LG 88%

Sustainable LLM Inference for Edge AI: Evaluating Quantized LLMs for Energy Efficiency, Output Accuracy, and Inference Latency

Erik Johannes Husom, Arda Goknil, Merve Astekin, Lwin Khin Shar, Andre Kåsen, Sagar Sen, Benedikt Andreas Mithassel, Ahmet Soylu

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

Comments 30 pages, 14 figures

2503.17793 2025-03-25 cs.LG cs.AI cs.CL 88%

Every Sample Matters: Leveraging Mixture-of-Experts and High-Quality Data for Efficient and Accurate Code LLM

Codefuse, Ling Team, :, Wenting Cai, Yuchen Cao, Chaoyu Chen, Chen Chen, Siba Chen, Qing Cui, Peng Di, Junpeng Fang, Zi Gong, Ting Guo, Zhengyu He, Yang Huang, Cong Li, Jianguo Li, Zheng Li, Shijie Lian, BingChang Liu, Songshan Luo, Shuo Mao, Min Shen, Jian Wu, Jiaolong Yang, Wenjie Yang, Tong Ye, Hang Yu, Wei Zhang, Zhenduo Zhang, Hailin Zhao, Xunjin Zheng, Jun Zhou

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

Comments 20 pages, 6 figures

2503.05005 2025-03-12 cs.CL cs.AI cs.LG 88%

Balcony: A Lightweight Approach to Dynamic Inference of Generative Language Models

Benyamin Jamialahmadi, Parsa Kavehzadeh, Mehdi Rezagholizadeh, Parsa Farinneya, Hossein Rajabzadeh, Aref Jafari, Boxing Chen, Marzieh S. Tahaei

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

2501.16937 2025-03-03 cs.LG cs.AI cs.CL 88%

TAID: Temporally Adaptive Interpolated Distillation for Efficient Knowledge Transfer in Language Models

Makoto Shing, Kou Misaki, Han Bao, Sho Yokoi, Takuya Akiba

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

Comments To appear at the 13th International Conference on Learning Representations (ICLR 2025) as a Spotlight presentation

2405.16406 2025-02-21 cs.LG cs.AI cs.CL cs.CV 88%

SpinQuant: LLM quantization with learned rotations

Zechun Liu, Changsheng Zhao, Igor Fedorov, Bilge Soran, Dhruv Choudhary, Raghuraman Krishnamoorthi, Vikas Chandra, Yuandong Tian, Tijmen Blankevoort

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

Comments ICLR 2025

2502.04428 2025-02-10 cs.CL cs.AI cs.LG 88%

Confident or Seek Stronger: Exploring Uncertainty-Based On-device LLM Routing From Benchmarking to Generalization

Yu-Neng Chuang, Leisheng Yu, Guanchu Wang, Lizhe Zhang, Zirui Liu, Xuanting Cai, Yang Sui, Vladimir Braverman, Xia Hu

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

2501.14371 2025-01-27 cs.CL cs.AI cs.LG 88%

DRESSing Up LLM: Efficient Stylized Question-Answering via Style Subspace Editing

Xinyu Ma, Yifeng Xu, Yang Lin, Tianlong Wang, Xu Chu, Xin Gao, Junfeng Zhao, Yasha Wang

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

Comments ICLR 2025 Accepted

2412.16117 2024-12-23 cs.CV 88%

PruneVid: Visual Token Pruning for Efficient Video Large Language Models

Xiaohu Huang, Hao Zhou, Kai Han

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

Comments Efficient Video Large Language Models

2404.13082 2024-11-21 cs.CL cs.AI cs.LG 88%

Efficient Contextual LLM Cascades through Budget-Constrained Policy Learning

Xuechen Zhang, Zijian Huang, Ege Onur Taga, Carlee Joe-Wong, Samet Oymak, Jiasi Chen

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

2404.06910 2024-07-22 cs.CL cs.AI cs.LG 88%

Superposition Prompting: Improving and Accelerating Retrieval-Augmented Generation

Thomas Merth, Qichen Fu, Mohammad Rastegari, Mahyar Najibi

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

2312.12009 2024-06-27 cs.CL cs.AI cs.LG 88%

Active Preference Inference using Language Models and Probabilistic Reasoning

Wasu Top Piriyakulkij, Volodymyr Kuleshov, Kevin Ellis

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

2402.04291 2024-06-19 cs.LG cs.AI cs.CL 88%

BiLLM: Pushing the Limit of Post-Training Quantization for LLMs

Wei Huang, Yangdong Liu, Haotong Qin, Ying Li, Shiming Zhang, Xianglong Liu, Michele Magno, Xiaojuan Qi

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

Comments 19 pages

2308.02582 2023-08-10 cs.CL cs.AI cs.LG 88%

Adapt and Decompose: Efficient Generalization of Text-to-SQL via Domain Adapted Least-To-Most Prompting

Aseem Arora, Shabbirhussain Bhaisaheb, Harshit Nigam, Manasi Patwardhan, Lovekesh Vig, Gautam Shroff

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

Comments 22 Pages

2604.06013 2026-04-08 cs.AI cs.CL 88%

Epistemic Blinding: An Inference-Time Protocol for Auditing Prior Contamination in LLM-Assisted Analysis

知识盲区：一种用于审计LLM辅助分析中先验污染的推理时协议

Michael Cuccarese

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

AI总结本文提出了一种在代理系统中用于审计LLM辅助分析中先验污染的推理时协议，通过替换实体标识符以匿名代码并比较输出与未盲对照组，以恢复审计性。

Comments code and LLM skill at: https://github.com/mcuccarese/epistemic-blinding 7 pages 3 figures

2403.16809 2024-03-26 eess.SY cs.AI cs.LG cs.SY 88%

An LLM-Based Digital Twin for Optimizing Human-in-the Loop Systems

Hanqing Yang, Marie Siew, Carlee Joe-Wong

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

Comments Accepted at International Workshop on Foundation Models for Cyber-Physical Systems & Internet of Things (FMSys) 2024, Co-located at CPS-IoT Week 2024

2608.17234 2026-08-19 cs.CR cs.AI 新提交 88%

COMIC: Reference-Aware Safety Gating for Multimodal Large Language Models

COMIC：面向多模态大语言模型的参考感知安全门控

Md Abdullahil Oaphy, Anhao Xiang, Zongxing Xie, Huayue Gu, Chenyu Wang, Honghui Xu

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

AI总结本研究针对多模态大语言模型的参考依赖型安全缺陷，提出COMIC参考感知安全门控，通过解析操作-目标对评估安全性，提升了模型鲁棒性且保留良性效用。

2601.22170 2026-08-19 math.NA cs.LG cs.NA stat.ML 版本更新 88%

Large Language Models: A Mathematical Formulation

大语言模型：数学表述

Ricardo Baptista, Andrew Stuart, Son Tran

机构 * Statistical Sciences, University of Toronto（多伦多大学统计科学系）； Vector Institute（向量研究所）； Amazon Search（亚马逊搜索）； Computing and Mathematical Sciences, California Institute of Technology（加州理工学院计算与数学科学系）

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

AI总结本文提出了一种大语言模型的数学框架，涵盖文本编码、模型架构、学习过程及部署方法，旨在提升算法的准确性、效率和鲁棒性。

Comments 56 pages, 2 figures

2608.16336 2026-08-18 cs.AR cs.DC cs.LG 新提交 88%

Beyond Binary Priorities: Multi-Tier SLA Scheduling for Large Language Model Serving

超越二元优先级：面向大语言模型服务的多层级SLA调度

Anders Vestrum, Arya Raeesi, Hanna Roed

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

AI总结本研究扩展Llumnix的优先级模型支持任意层级，在Vidur模拟器中对比多基线，发现四层优先级调度成本效益最优，10层时仍能维持收益且无尾部延迟崩溃。

Comments 13 pages, 9 figures, 4 tables

2608.11742 2026-08-13 cs.CL 新提交 88%

Ripple-Pivot Search: Active Parallel Decoding for Diffusion Large Language Models

波纹枢轴搜索：扩散大语言模型的主动并行解码

Yushi Ye, Xu Chen, Haoyun Jiang, Jinsong Lan, Haihong Tang, Bo Han, Ivor Tsang, Yanfeng Wang, Bo Zheng, Jiangchao Yao

机构 * Cooperative Medianet Innovation Center, Shanghai Jiao Tong University（上海交通大学合作中位数网络创新中心）； Alibaba Group TMLR Group（阿里巴巴集团TMLR团队）； Hong Kong Baptist University（香港浸会大学）； A*STAR CFAR and Nanyang Technological University（新加坡科研局计算科学中心和南洋理工大学）； School of Artificial Intelligence, Shanghai Jiao Tong University（上海交通大学人工智能学院）

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

AI总结提出RPS方法，利用dLLM解码的波纹效应，在保持生成质量的同时，大幅提升解码速度与准确率，适用于推理及代码生成任务。

1. 效率与部署 22368 篇

Litespark Technical Report: High-Throughput, Energy-Efficient LLM Training Framework

Efficient Scaling for LLM-based ASR

You Can Generate It Again: Data-to-Text Generation with Verification and Correction Prompting

DSSD: Efficient Edge-Device LLM Deployment and Collaborative Inference via Distributed Split Speculative Decoding

PoTPTQ: A Two-step Power-of-Two Post-training for LLMs

Contextual Experience Replay for Self-Improvement of Language Agents

AutoML-Agent: A Multi-Agent LLM Framework for Full-Pipeline AutoML

KDRL: Post-Training Reasoning LLMs via Unified Knowledge Distillation and Reinforcement Learning

JPPO++: Joint Power and Denoising-inspired Prompt Optimization for Mobile LLM Services

Leveraging Dual Process Theory in Language Agent Framework for Real-time Simultaneous Human-AI Collaboration

Task-oriented Age of Information for Remote Inference with Hybrid Language Models

Sustainable LLM Inference for Edge AI: Evaluating Quantized LLMs for Energy Efficiency, Output Accuracy, and Inference Latency

Every Sample Matters: Leveraging Mixture-of-Experts and High-Quality Data for Efficient and Accurate Code LLM

Balcony: A Lightweight Approach to Dynamic Inference of Generative Language Models

TAID: Temporally Adaptive Interpolated Distillation for Efficient Knowledge Transfer in Language Models

SpinQuant: LLM quantization with learned rotations

Confident or Seek Stronger: Exploring Uncertainty-Based On-device LLM Routing From Benchmarking to Generalization

DRESSing Up LLM: Efficient Stylized Question-Answering via Style Subspace Editing

PruneVid: Visual Token Pruning for Efficient Video Large Language Models

Efficient Contextual LLM Cascades through Budget-Constrained Policy Learning

Superposition Prompting: Improving and Accelerating Retrieval-Augmented Generation

Active Preference Inference using Language Models and Probabilistic Reasoning

BiLLM: Pushing the Limit of Post-Training Quantization for LLMs

Adapt and Decompose: Efficient Generalization of Text-to-SQL via Domain Adapted Least-To-Most Prompting

Epistemic Blinding: An Inference-Time Protocol for Auditing Prior Contamination in LLM-Assisted Analysis

An LLM-Based Digital Twin for Optimizing Human-in-the Loop Systems

COMIC: Reference-Aware Safety Gating for Multimodal Large Language Models

Large Language Models: A Mathematical Formulation

Beyond Binary Priorities: Multi-Tier SLA Scheduling for Large Language Model Serving

Ripple-Pivot Search: Active Parallel Decoding for Diffusion Large Language Models