语言大模型 / LLM - arXivDaily 专题

2405.06270 2025-07-28 cs.LG cs.AI cs.CL 80%

XAI4LLM. Let Machine Learning Models and LLMs Collaborate for Enhanced In-Context Learning in Healthcare

Fatemeh Nazary, Yashar Deldjoo, Tommaso Di Noia, Eugenio di Sciascio

机构 * Polytechnic University of Bari（巴里理工大学）

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

2410.06238 2025-07-15 cs.LG cs.AI cs.CL 80%

EVOLvE: Evaluating and Optimizing LLMs For In-Context Exploration

Allen Nie, Yi Su, Bo Chang, Jonathan N. Lee, Ed H. Chi, Quoc V. Le, Minmin Chen

机构 * Stanford University（斯坦福大学）； Google DeepMind（谷歌DeepMind）

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. Published at ICML 2025

2507.06416 2025-07-10 eess.SY cs.SY 80%

Voltage Regulation in Distribution Systems with Data Center Loads

Yize Chen, Baosen Zhang

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

Comments Code available at https://github.com/chennnnnyize/voltage-regulation-with-data-centers

2506.08572 2025-07-08 cs.LG cs.AI cs.CL stat.ML 80%

The Geometries of Truth Are Orthogonal Across Tasks

Waiss Azizian, Michael Kirchhof, Eugene Ndiaye, Louis Bethune, Michal Klein, Pierre Ablin, Marco Cuturi

机构 * Apple（苹果公司）

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

2507.00054 2025-07-02 cs.AI cs.CL cs.LG 80%

Enhancing Reasoning Capabilities in SLMs with Reward Guided Dataset Distillation

Shreyansh Padarha

机构 * Shreyansh Padarha（独立研究者）

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

Comments 17 Pages, 7 figures

2503.10135 2025-07-01 cs.CL cs.AI cs.LG 80%

Gumiho: A Hybrid Architecture to Prioritize Early Tokens in Speculative Decoding

Jinze Li, Yixing Xu, Haiduo Huang, Xuanwu Yin, Dong Li, Edith C. H. Ngai, Emad Barsoum

机构 * Advanced Micro Devices, Inc., Beijing, China（Advanced Micro Devices公司，北京）； Department of Electrical and Electronic Engineering, The University of Hong Kong（香港大学电子与电气工程系）； Institute of Artificial Intelligence and Robotics, Xi’an Jiaotong University（西安交通大学人工智能与机器人研究所）

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 42nd International Conference on Machine Learning (ICML 2025). Code: https://github.com/AMD-AIG-AIMA/Gumiho

2506.21263 2025-06-27 cs.LG cs.AI cs.CL 80%

DiLoCoX: A Low-Communication Large-Scale Training Framework for Decentralized Cluster

Ji Qi, WenPeng Zhu, Li Li, Ming Wu, YingJun Wu, Wu He, Xun Gao, Jason Zeng, Michael Heinrich

机构 * Zero Gravity Labs（零重力实验室）

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

2506.16043 2025-06-23 cs.CL cs.AI cs.LG 80%

DynScaling: Efficient Verifier-free Inference Scaling via Dynamic and Integrated Sampling

Fei Wang, Xingchen Wan, Ruoxi Sun, Jiefeng Chen, Sercan Ö. Arık

机构 * Google（谷歌）

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

2506.09099 2025-06-19 cs.LG cs.AI cs.CL 80%

Too Big to Think: Capacity, Memorization, and Generalization in Pre-Trained Transformers

Joshua Barron, Devin White

机构 * Amazon（亚马逊）

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

Comments Accepted for oral presentation to Tiny Titans: The next wave of On-Device Learning for Foundational Models Workshop at the 42nd International Conference on Machine Learning

2506.06166 2025-06-09 cs.LG cs.AI cs.CL cs.CY cs.HC 80%

The Lock-in Hypothesis: Stagnation by Algorithm

Tianyi Alex Qiu, Zhonghao He, Tejasveer Chugh, Max Kleiman-Weiner

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 ICML 2025, 46 pages

2410.22118 2025-06-09 cs.CL cs.AI cs.LG 80%

The Impact of Inference Acceleration on Bias of LLMs

Elisabeth Kirsten, Ivan Habernal, Vedant Nanda, Muhammad Bilal Zafar

机构 * Ruhr University Bochum（博德姆鲁尔大学）； Aleph Alpha（Aleph Alpha公司）； UAR Research Center for Trustworthy Data Science and Security（UAR可信数据科学与安全研究中心）

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

2506.04566 2025-06-06 cs.LG cs.AI cs.CL cs.CR 80%

Clustering and Median Aggregation Improve Differentially Private Inference

Kareem Amin, Salman Avestimehr, Sara Babakniya, Alex Bie, Weiwei Kong, Natalia Ponomareva, Umar Syed

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

2506.03566 2025-06-05 cs.CL cs.AI cs.LG 80%

POSS: Position Specialist Generates Better Draft for Speculative Decoding

Langlin Huang, Chengsong Huang, Jixuan Leng, Di Huang, Jiaxin Huang

机构 * Washington University in St. Louis（华盛顿大学圣路易斯分校）； 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

2504.00030 2025-06-05 cs.CL cs.AI cs.LG 80%

Token-Driven GammaTune: Adaptive Calibration for Enhanced Speculative Decoding

Aayush Gautam, Susav Shrestha, Narasimha Reddy

机构 * Department of Electrical and Computer Engineering（电气与计算机工程系）

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 6 pages, 2 figures, 1 table

2410.09300 2025-06-05 cs.CL cs.AI cs.LG 80%

Nudging: Inference-time Alignment of LLMs via Guided Decoding

Yu Fei, Yasaman Razeghi, Sameer Singh

机构 * Department of Computer Science University of California Irvine（计算机科学系，加州大学伊文斯顿分校）

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 to ACL 2025 (main)

2506.01151 2025-06-03 cs.LG cs.AI cs.CL 80%

Earley-Driven Dynamic Pruning for Efficient Structured Decoding

Xintong Sun, Chi Wei, Minghao Tian, Shiwen Ni

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 ICML2025 poster

2506.01118 2025-06-03 cs.CV 80%

Revolutionizing Radiology Workflow with Factual and Efficient CXR Report Generation

Pimchanok Sukjai, Apiradee Boonmee

机构 * Kasem Bundit University（Kasem Bundit大学）

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

2501.15056 2025-06-03 cs.AI cs.CL cs.HC cs.LG 80%

Feedback-Aware Monte Carlo Tree Search for Efficient Information Seeking in Goal-Oriented Conversations

Harshita Chopra, Chirag Shah

机构 * University of Washington（华盛顿大学）

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

2503.07067 2025-06-02 cs.CL cs.AI cs.LG 80%

DistiLLM-2: A Contrastive Approach Boosts the Distillation of LLMs

Jongwoo Ko, Tianyi Chen, Sungnyun Kim, Tianyu Ding, Luming Liang, Ilya Zharkov, Se-Young Yun

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 ICML2025 Spotlight

2407.00066 2025-06-02 cs.DC cs.AI cs.CL cs.LG 80%

Compress then Serve: Serving Thousands of LoRA Adapters with Little Overhead

Rickard Brüel-Gabrielsson, Jiacheng Zhu, Onkar Bhardwaj, Leshem Choshen, Kristjan Greenewald, Mikhail Yurochkin, Justin Solomon

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

2502.16880 2025-05-27 cs.CL cs.AI cs.LG 80%

CORAL: Learning Consistent Representations across Multi-step Training with Lighter Speculative Drafter

Yepeng Weng, Dianwen Mei, Huishi Qiu, Xujie Chen, Li Liu, Jiang Tian, Zhongchao Shi

机构 * AI Lab, Lenovo Research（联想研究院人工智能实验室）

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

Comments Accepted to ACL 2025 main conference

2505.18413 2025-05-27 cs.LG cs.AI cs.CL cs.CV 80%

LatentLLM: Attention-Aware Joint Tensor Compression

Toshiaki Koike-Akino, Xiangyu Chen, Jing Liu, Ye Wang, Pu, Wang, Matthew Brand

机构 * Mitsubishi Electric Research Laboratories (MERL)（三菱电机研究实验室）

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

Comments 37 pages, 16 figures

2505.16210 2025-05-23 cs.LG cs.AI cs.CL 80%

NQKV: A KV Cache Quantization Scheme Based on Normal Distribution Characteristics

Zhihang Cai, Xingjun Zhang, Zhendong Tan, Zheng Wei

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 11 pages, 9 figures

2505.15753 2025-05-22 cs.CR cs.AI cs.CL cs.LG 80%

Scalable Defense against In-the-wild Jailbreaking Attacks with Safety Context Retrieval

Taiye Chen, Zeming Wei, Ang Li, Yisen Wang

机构 * School of EECS, Peking University（电子工程系，北京大学）； School of Mathematical Sciences, Peking University（数学系，北京大学）； State Key Lab of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University（通用人工智能国家重点实验室，智能科学与技术学校，北京大学）； Institute for Artificial Intelligence, Peking University（人工智能研究院，北京大学）

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

2410.09083 2025-05-21 cs.AI cs.CL cs.CV cs.LG 80%

Evaluating the Correctness of Inference Patterns Used by LLMs for Judgment

Lu Chen, Yuxuan Huang, Yixing Li, Dongrui Liu, Qihan Ren, Shuai Zhao, Kun Kuang, Zilong Zheng, Quanshi Zhang

机构 * Shanghai Jiao Tong University（上海交通大学）； The Chinese University of Hong Kong（香港中文大学）； Shanghai Artificial Intelligence Laboratory（上海人工智能实验室）； Zhejiang University（浙江大学）； Beijing Institute for General Artificial Intelligence（北京一般人工智能研究所）

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

2504.01281 2025-05-21 cs.LG cs.AI cs.CL cs.IR 80%

Scaling Test-Time Inference with Policy-Optimized, Dynamic Retrieval-Augmented Generation via KV Caching and Decoding

Sakhinana Sagar Srinivas, Akash Das, Shivam Gupta, Venkataramana Runkana

机构 * Tata Research Development（塔塔研究与发展中心）； Design Center, Bangalore（班加罗尔设计中心）

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

2404.11865 2025-05-19 cs.CV 80%

From Image to Video, what do we need in multimodal LLMs?

Suyuan Huang, Haoxin Zhang, Linqing Zhong, Honggu Chen, Yan Gao, Yao Hu, Zengchang Qin

机构 * Intelligent Computing and Machine Learning Lab, School of ASEE, Beihang University（北京航空航天大学自动化学院智能计算与机器学习实验室）； Xiaohongshu（小红书）； School of Sino-French Engineer, Beihang University（北京航空航天大学中法工程师学院）； College of Engineering and Computer Science, VinUniversity（Vin大学工程与计算机科学学院）

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

2505.08823 2025-05-15 cs.LG cs.AI cs.CL 80%

An Extra RMSNorm is All You Need for Fine Tuning to 1.58 Bits

Cody Steinmetz, Gavin Childress, Aaron Herbst, Gavin Jones, Jasdeep Singh, Eli Vang, Keagan Weinstock

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

2505.07289 2025-05-13 cs.CL cs.AI cs.LG 80%

Semantic Retention and Extreme Compression in LLMs: Can We Have Both?

Stanislas Laborde, Martin Cousseau, Antoun Yaacoub, Lionel Prevost

机构 * ESIEA Lab（ESIEA实验室）； ESIEA

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 publication in the Proceedings of the 2025 International Joint Conference on Neural Networks (IJCNN); this arXiv version includes an appendix with 6 result tables; 10 pages, 15 figures, 7 tables

2505.05190 2025-05-13 cs.LG cs.AI cs.CL cs.CR 80%

Revealing Weaknesses in Text Watermarking Through Self-Information Rewrite Attacks

Yixin Cheng, Hongcheng Guo, Yangming Li, Leonid Sigal

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 ICML 2025 Accpeted

1. 效率与部署 22531 篇

XAI4LLM. Let Machine Learning Models and LLMs Collaborate for Enhanced In-Context Learning in Healthcare

EVOLvE: Evaluating and Optimizing LLMs For In-Context Exploration

Voltage Regulation in Distribution Systems with Data Center Loads

The Geometries of Truth Are Orthogonal Across Tasks

Enhancing Reasoning Capabilities in SLMs with Reward Guided Dataset Distillation

Gumiho: A Hybrid Architecture to Prioritize Early Tokens in Speculative Decoding

DiLoCoX: A Low-Communication Large-Scale Training Framework for Decentralized Cluster

DynScaling: Efficient Verifier-free Inference Scaling via Dynamic and Integrated Sampling

Too Big to Think: Capacity, Memorization, and Generalization in Pre-Trained Transformers

The Lock-in Hypothesis: Stagnation by Algorithm

The Impact of Inference Acceleration on Bias of LLMs

Clustering and Median Aggregation Improve Differentially Private Inference

POSS: Position Specialist Generates Better Draft for Speculative Decoding

Token-Driven GammaTune: Adaptive Calibration for Enhanced Speculative Decoding

Nudging: Inference-time Alignment of LLMs via Guided Decoding

Earley-Driven Dynamic Pruning for Efficient Structured Decoding

Revolutionizing Radiology Workflow with Factual and Efficient CXR Report Generation

Feedback-Aware Monte Carlo Tree Search for Efficient Information Seeking in Goal-Oriented Conversations

DistiLLM-2: A Contrastive Approach Boosts the Distillation of LLMs

Compress then Serve: Serving Thousands of LoRA Adapters with Little Overhead

CORAL: Learning Consistent Representations across Multi-step Training with Lighter Speculative Drafter

LatentLLM: Attention-Aware Joint Tensor Compression

NQKV: A KV Cache Quantization Scheme Based on Normal Distribution Characteristics

Scalable Defense against In-the-wild Jailbreaking Attacks with Safety Context Retrieval

Evaluating the Correctness of Inference Patterns Used by LLMs for Judgment

Scaling Test-Time Inference with Policy-Optimized, Dynamic Retrieval-Augmented Generation via KV Caching and Decoding

From Image to Video, what do we need in multimodal LLMs?

An Extra RMSNorm is All You Need for Fine Tuning to 1.58 Bits

Semantic Retention and Extreme Compression in LLMs: Can We Have Both?

Revealing Weaknesses in Text Watermarking Through Self-Information Rewrite Attacks