语言大模型 / LLM - arXivDaily 专题

2410.03804 2025-04-04 cs.CL cs.AI cs.LG 80%

Mixture of Attentions For Speculative Decoding

Matthieu Zimmer, Milan Gritta, Gerasimos Lampouras, Haitham Bou Ammar, Jun Wang

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

Comments Accepted at International Conference on Learning Representations (ICLR 2025)

2408.02549 2025-03-25 eess.SY cs.SY 80%

Generative AI as a Service in 6G Edge-Cloud: Generation Task Offloading by In-context Learning

Hao Zhou, Chengming Hu, Dun Yuan, Ye Yuan, Di Wu, Xue Liu, Zhu Han, Charlie Zhang

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

Comments This paper has been accepted by IEEE Wireless Communications Letters

2402.05868 2025-03-24 cs.CL cs.AI cs.CR cs.IR cs.LG 80%

EmojiPrompt: Generative Prompt Obfuscation for Privacy-Preserving Communication with Cloud-based LLMs

Sam Lin, Wenyue Hua, Zhenting Wang, Mingyu Jin, Lizhou Fan, Yongfeng Zhang

机构 * Rutgers University（罗格斯大学）； University of Michigan（密歇根大学）

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

Comments Accepted to the 2025 Annual Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL 2025)

2406.18678 2025-03-05 cs.LG cs.AI cs.CL 80%

Few-shot Personalization of LLMs with Mis-aligned Responses

Jaehyung Kim, Yiming Yang

机构 * Yonsei University（延世大学）； Carnegie Mellon University（卡内基梅隆大学）

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

Comments NAACL 25 (main, long), 32 pages

2405.01229 2025-03-04 cs.LG cs.AI cs.CL cs.CR math.OC 80%

Boosting Jailbreak Attack with Momentum

Yihao Zhang, Zeming Wei

机构 * Peking University（北京大学）； UC Berkeley（加州大学伯克利分校）

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

Comments Accepted by ICASSP 2025

2410.11261 2025-02-27 cs.LG cs.AI cs.CL 80%

Beyond Linear Approximations: A Novel Pruning Approach for Attention Matrix

Yingyu Liang, Jiangxuan Long, Zhenmei Shi, Zhao Song, Yufa Zhou

机构 * The University of Hong Kong（香港大学）； University of Wisconsin-Madison（威斯康星大学麦迪逊分校）； South China University of Technology（华南理工大学）； The Simons Institute for the Theory of Computing（西蒙斯理论计算研究所）； University of Pennsylvania（宾夕法尼亚大学）

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

Comments ICLR 2025

2406.18665 2025-02-25 cs.LG cs.AI cs.CL 80%

RouteLLM: Learning to Route LLMs with Preference Data

Isaac Ong, Amjad Almahairi, Vincent Wu, Wei-Lin Chiang, Tianhao Wu, Joseph E. Gonzalez, M Waleed Kadous, Ion Stoica

机构 * UC Berkeley（加州大学伯克利分校）； Anyscale ； Canva

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

2502.15816 2025-02-25 cs.DC 80%

GenAI at the Edge: Comprehensive Survey on Empowering Edge Devices

Mozhgan Navardi, Romina Aalishah, Yuzhe Fu, Yueqian Lin, Hai Li, Yiran Chen, Tinoosh Mohsenin

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

Comments AAAI 2025 Spring Symposium Series (SSS), GenAI@Edge: Empowering Generative AI at the Edge Symposium

2502.15734 2025-02-25 cs.DC cs.AI cs.CL cs.LG cs.OS 80%

Cache-Craft: Managing Chunk-Caches for Efficient Retrieval-Augmented Generation

Shubham Agarwal, Sai Sundaresan, Subrata Mitra, Debabrata Mahapatra, Archit Gupta, Rounak Sharma, Nirmal Joshua Kapu, Tong Yu, Shiv Saini

机构 * Adobe Research（奥多比研究院）； IIT Bombay（印度理工学院孟买分校）； IIT Kanpur（印度理工学院坎普尔分校）

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

Comments Accepted at SIGMOD 2025

2502.13502 2025-02-25 cs.CL cs.AI cs.LG 80%

PLDR-LLMs Learn A Generalizable Tensor Operator That Can Replace Its Own Deep Neural Net At Inference

Burc Gokden

机构 * Fromthesky Research Labs LLC（弗罗姆斯基研究实验室有限责任公司）

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

Comments 15 pages, 1 figure, 12 tables, more ablation data included

2502.08172 2025-02-13 cs.SE 80%

Intention is All You Need: Refining Your Code from Your Intention

Qi Guo, Xiaofei Xie, Shangqing Liu, Ming Hu, Xiaohong Li, Lei Bu

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

2502.06844 2025-02-12 cs.LG cs.AI cs.CL 80%

Exploring Model Invariance with Discrete Search for Ultra-Low-Bit Quantization

Yuqiao Wen, Yanshuai Cao, Lili Mou

机构 * University of Alberta（阿尔伯塔大学）； RBC Borealis ； Alberta Machine Intelligence Institute (Amii)（阿尔伯塔机器智能研究所（Amii））

专题命中效率与部署：large language model(abstract)；language model(abstract)；post-training(abstract)；分类 cs.CL、cs.AI、cs.LG

2502.05694 2025-02-11 cs.CL cs.AI cs.LG 80%

Zero-Shot End-to-End Relation Extraction in Chinese: A Comparative Study of Gemini, LLaMA and ChatGPT

Shaoshuai Du, Yiyi Tao, Yixian Shen, Hang Zhang, Yanxin Shen, Xinyu Qiu, Chuanqi Shi

机构 * University of Amsterdam（阿姆斯特丹大学）； Johns Hopkins University（约翰斯·霍普金斯大学）； University of California San Diego（加利福尼亚大学圣迭戈分校）； Simon Fraser University（西蒙菲莎大学）； Northeastern University（东北大学）

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

2409.14175 2025-02-05 cs.CL cs.AI cs.LG 80%

QMOS: Enhancing LLMs for Telecommunication with Question Masked loss and Option Shuffling

Blessed Guda, Gabrial Zencha Ashungafac, Lawrence Francis, Carlee Joe-Wong

机构 * College of Engineering, Carnegie Mellon University（卡内基梅隆大学工程学院）

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

Journal ref IEEE Globecom Workshop 2024

2412.18644 2025-01-29 cs.CL cs.AI cs.LG 80%

DynaGRAG | Exploring the Topology of Information for Advancing Language Understanding and Generation in Graph Retrieval-Augmented Generation

Karishma Thakrar

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

2501.06252 2025-01-27 cs.LG cs.AI cs.CL 80%

Transformer-Squared: Self-adaptive LLMs

Qi Sun, Edoardo Cetin, Yujin Tang

机构 * Sakana AI（萨卡纳人工智能公司）； Institute of Science Tokyo（东京科学大学）

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

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

2412.15283 2024-12-23 cs.CL cs.AI cs.LG 80%

Channel Merging: Preserving Specialization for Merged Experts

Mingyang Zhang, Jing Liu, Ganggui Ding, Xinyi Yu, Linlin Ou, Bohan Zhuang

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

Comments accepted by AAAI 2025

2412.02602 2024-12-04 cs.CL cs.AI cs.CY cs.LG 80%

CEGI: Measuring the trade-off between efficiency and carbon emissions for SLMs and VLMs

Abhas Kumar, Kapil Pathak, Rajesh Kavuru, Prabhakar Srinivasan

机构 * Synechron（赛克龙公司）

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

2411.13157 2024-11-28 cs.CL cs.AI cs.LG 80%

Closer Look at Efficient Inference Methods: A Survey of Speculative Decoding

Hyun Ryu, Eric Kim

机构 * KAIST（韩国科学技术院）； Georgia Institute of Technology（佐治亚理工学院）

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

2409.00119 2024-11-05 cs.LG cs.AI cs.CL 80%

3-in-1: 2D Rotary Adaptation for Efficient Finetuning, Efficient Batching and Composability

Baohao Liao, Christof Monz

机构 * University of Amsterdam（阿姆斯特丹大学）； eBay Inc.（eBay公司）

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

Comments Accepted to NeurIPS 2024. Code: https://github.com/BaohaoLiao/road

2312.02119 2024-11-01 cs.LG cs.AI cs.CL cs.CR stat.ML 80%

Tree of Attacks: Jailbreaking Black-Box LLMs Automatically

Anay Mehrotra, Manolis Zampetakis, Paul Kassianik, Blaine Nelson, Hyrum Anderson, Yaron Singer, Amin Karbasi

机构 * Yale University（耶鲁大学）； Robust Intelligence ； Google Research（谷歌研究院）

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

Comments Accepted for presentation at NeurIPS 2024. Code: https://github.com/RICommunity/TAP

2410.18952 2024-10-31 cs.CL cs.AI cs.LG 80%

Dynamic Vocabulary Pruning in Early-Exit LLMs

Jort Vincenti, Karim Abdel Sadek, Joan Velja, Matteo Nulli, Metod Jazbec

机构 * University of Amsterdam（阿姆斯特丹大学）； Krueger AI Safety Lab (KASL)（克鲁格人工智能安全实验室（KASL））； UvA-Bosch Delta Lab（阿姆斯特丹大学-博世三角洲实验室）

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

Journal ref NeurIPS 2024 ENLSP Workshop

2405.16122 2024-10-30 cs.AI cs.CL cs.LG stat.ML 80%

Prompt Optimization with EASE? Efficient Ordering-aware Automated Selection of Exemplars

Zhaoxuan Wu, Xiaoqiang Lin, Zhongxiang Dai, Wenyang Hu, Yao Shu, See-Kiong Ng, Patrick Jaillet, Bryan Kian Hsiang Low

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

Comments 28 pages, 1 figure, 35 tables

2406.17415 2024-10-29 cs.CL cs.AI cs.LG 80%

Layer-Wise Quantization: A Pragmatic and Effective Method for Quantizing LLMs Beyond Integer Bit-Levels

Razvan-Gabriel Dumitru, Vikas Yadav, Rishabh Maheshwary, Paul-Ioan Clotan, Sathwik Tejaswi Madhusudhan, Mihai Surdeanu

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

2404.16710 2024-10-21 cs.CL cs.AI cs.LG 80%

LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding

Mostafa Elhoushi, Akshat Shrivastava, Diana Liskovich, Basil Hosmer, Bram Wasti, Liangzhen Lai, Anas Mahmoud, Bilge Acun, Saurabh Agarwal, Ahmed Roman, Ahmed A Aly, Beidi Chen, Carole-Jean Wu

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

Comments ACL 2024

2406.15786 2024-10-18 cs.LG cs.AI cs.CL 80%

What Matters in Transformers? Not All Attention is Needed

Shwai He, Guoheng Sun, Zheyu Shen, Ang Li

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

Comments 15 pages, 13 figures, 6 tables

2406.11890 2024-10-16 cs.LG cs.AI cs.CL 80%

Unraveling the Mechanics of Learning-Based Demonstration Selection for In-Context Learning

Hui Liu, Wenya Wang, Hao Sun, Chris Xing Tian, Chenqi Kong, Xin Dong, Haoliang Li

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

Comments 17 pages, 7 figures and 9 tables

2403.02966 2024-10-10 cs.CL cs.AI cs.LG 80%

Evidence-Focused Fact Summarization for Knowledge-Augmented Zero-Shot Question Answering

Sungho Ko, Hyunjin Cho, Hyungjoo Chae, Jinyoung Yeo, Dongha Lee

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

2410.05603 2024-10-10 cs.LG cs.AI cs.CL 80%

Everything Everywhere All at Once: LLMs can In-Context Learn Multiple Tasks in Superposition

Zheyang Xiong, Ziyang Cai, John Cooper, Albert Ge, Vasilis Papageorgiou, Zack Sifakis, Angeliki Giannou, Ziqian Lin, Liu Yang, Saurabh Agarwal, Grigorios G Chrysos, Samet Oymak, Kangwook Lee, Dimitris Papailiopoulos

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

2309.05516 2024-10-10 cs.CL cs.AI cs.LG 80%

Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs

Wenhua Cheng, Weiwei Zhang, Haihao Shen, Yiyang Cai, Xin He, Kaokao Lv, Yi Liu

专题命中效率与部署：large language model(abstract)；language model(abstract)；post-training(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments EMNLP24 Findings

1. 效率与部署 22531 篇

Mixture of Attentions For Speculative Decoding

Generative AI as a Service in 6G Edge-Cloud: Generation Task Offloading by In-context Learning

EmojiPrompt: Generative Prompt Obfuscation for Privacy-Preserving Communication with Cloud-based LLMs

Few-shot Personalization of LLMs with Mis-aligned Responses

Boosting Jailbreak Attack with Momentum

Beyond Linear Approximations: A Novel Pruning Approach for Attention Matrix

RouteLLM: Learning to Route LLMs with Preference Data

GenAI at the Edge: Comprehensive Survey on Empowering Edge Devices

Cache-Craft: Managing Chunk-Caches for Efficient Retrieval-Augmented Generation

PLDR-LLMs Learn A Generalizable Tensor Operator That Can Replace Its Own Deep Neural Net At Inference

Intention is All You Need: Refining Your Code from Your Intention

Exploring Model Invariance with Discrete Search for Ultra-Low-Bit Quantization

Zero-Shot End-to-End Relation Extraction in Chinese: A Comparative Study of Gemini, LLaMA and ChatGPT

QMOS: Enhancing LLMs for Telecommunication with Question Masked loss and Option Shuffling

DynaGRAG | Exploring the Topology of Information for Advancing Language Understanding and Generation in Graph Retrieval-Augmented Generation

Transformer-Squared: Self-adaptive LLMs

Channel Merging: Preserving Specialization for Merged Experts

CEGI: Measuring the trade-off between efficiency and carbon emissions for SLMs and VLMs

Closer Look at Efficient Inference Methods: A Survey of Speculative Decoding

3-in-1: 2D Rotary Adaptation for Efficient Finetuning, Efficient Batching and Composability

Tree of Attacks: Jailbreaking Black-Box LLMs Automatically

Dynamic Vocabulary Pruning in Early-Exit LLMs

Prompt Optimization with EASE? Efficient Ordering-aware Automated Selection of Exemplars

Layer-Wise Quantization: A Pragmatic and Effective Method for Quantizing LLMs Beyond Integer Bit-Levels

LayerSkip: Enabling Early Exit Inference and Self-Speculative Decoding

What Matters in Transformers? Not All Attention is Needed

Unraveling the Mechanics of Learning-Based Demonstration Selection for In-Context Learning

Evidence-Focused Fact Summarization for Knowledge-Augmented Zero-Shot Question Answering

Everything Everywhere All at Once: LLMs can In-Context Learn Multiple Tasks in Superposition

Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs