语言大模型 / LLM - arXivDaily 专题

2506.16500 2025-06-23 cs.LG 79%

SparseLoRA: Accelerating LLM Fine-Tuning with Contextual Sparsity

Samir Khaki, Xiuyu Li, Junxian Guo, Ligeng Zhu, Chenfeng Xu, Konstantinos N. Plataniotis, Amir Yazdanbakhsh, Kurt Keutzer, Song Han, Zhijian Liu

机构 * University of Toronto（多伦多大学）； MIT（麻省理工学院）； Google DeepMind（谷歌DeepMind）

专题命中指令微调：LLM(title,abstract)；分类 cs.LG

Comments ICML 2025. The first three authors contributed equally to this work. Project page: https://z-lab.ai/projects/sparselora

2503.01807 2025-06-23 cs.CL 79%

Large-Scale Data Selection for Instruction Tuning

Hamish Ivison, Muru Zhang, Faeze Brahman, Pang Wei Koh, Pradeep Dasigi

机构 * University of Washington（华盛顿大学）； Allen Institute for AI（人工智能研究院）； USC（南加州大学）

专题命中指令微调：instruction tuning(title)；language model(abstract)；分类 cs.CL

Comments Updated, new baselines, removed some typos

2506.14382 2025-06-18 cs.CV cs.AI 79%

DepthSeg: Depth prompting in remote sensing semantic segmentation

Ning Zhou, Shanxiong Chen, Mingting Zhou, Haigang Sui, Lieyun Hu, Han Li, Li Hua, Qiming Zhou

机构 * Wuhan University（武汉大学）； Changjiang Spatial Information Technology Engineering Co., Ltd（长江空间信息技术工程有限公司）； Huazhong Agr. University（华中农大）； Hong Kong Baptist University（香港 Baptist 大学）

专题命中指令微调：prompting(title,abstract)；分类 cs.AI

2410.19406 2025-06-17 cs.LG 79%

An Auditing Test To Detect Behavioral Shift in Language Models

Leo Richter, Xuanli He, Pasquale Minervini, Matt J. Kusner

机构 * UCL Centre for Artificial Intelligence, University College London（伦敦大学学院人工智能中心，大学学院伦敦）； School of Informatics, University of Edinburgh（信息学院，爱丁堡大学）； Miniml.AI, United Kingdom（Miniml.AI，英国）； Polytechnique Montréal, Canada（蒙特利尔理工学院，加拿大）； Mila – Quebec AI Institute, Canada（魁北克人工智能研究所，加拿大）

专题命中指令微调：language model(title,abstract)；分类 cs.LG

Comments 25 pages, 12 figures

Journal ref International Conference on Learning Representations (ICLR) 2025

2506.12213 2025-06-17 cs.LG cs.DC 79%

Fed-HeLLo: Efficient Federated Foundation Model Fine-Tuning with Heterogeneous LoRA Allocation

Zikai Zhang, Ping Liu, Jiahao Xu, Rui Hu

机构 * Department of Computer Science and Engineering, University of Nevada, Reno（计算机科学与工程系，内华达大学雷诺分校）

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

Comments Accepted to TNNLS 2025

2506.05176 2025-06-12 cs.CL 79%

Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models

Yanzhao Zhang, Mingxin Li, Dingkun Long, Xin Zhang, Huan Lin, Baosong Yang, Pengjun Xie, An Yang, Dayiheng Liu, Junyang Lin, Fei Huang, Jingren Zhou

机构 * Tongyi Lab（通义实验室）； Alibaba Group（阿里巴巴集团）

专题命中指令微调：foundation model(title,abstract)；分类 cs.CL

2409.06985 2025-06-10 cs.LG 79%

Unveiling Markov Heads in Pretrained Language Models for Offline Reinforcement Learning

Wenhao Zhao, Qiushui Xu, Linjie Xu, Lei Song, Jinyu Wang, Chunlai Zhou, Jiang Bian

专题命中指令微调：language model(title,abstract)；分类 cs.LG

Comments The first two authors contributed equally. The order was decided by a dice roll

2506.05027 2025-06-06 cs.LG 79%

Tuning the Right Foundation Models is What you Need for Partial Label Learning

Kuang He, Wei Tang, Tong Wei, Min-Ling Zhang

机构 * School of Computer Science and Engineering, Southeast University, Nanjing 210096, China（南京大学计算机科学与工程学院）； Key Lab. of Computer Network and Information Integration (Southeast University), MoE, China（计算机网络与信息集成重点实验室（东南大学））

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

Comments The code can be found at \url{https://github.com/SEU-hk/PartialCLIP}

2506.00630 2025-06-03 cs.LG 79%

Probabilistic Forecasting for Building Energy Systems using Time-Series Foundation Models

Young Jin Park, Francois Germain, Jing Liu, Ye Wang, Toshiaki Koike-Akino, Gordon Wichern, Navid Azizan, Christopher R. Laughman, Ankush Chakrabarty

机构 * IEEE

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

Comments Preliminary version appeared in NeurIPS TSALM Workshop: https://neurips.cc/virtual/2024/103019

2505.24088 2025-06-02 cs.LG cs.CV 79%

Proxy-FDA: Proxy-based Feature Distribution Alignment for Fine-tuning Vision Foundation Models without Forgetting

Chen Huang, Skyler Seto, Hadi Pouransari, Mehrdad Farajtabar, Raviteja Vemulapalli, Fartash Faghri, Oncel Tuzel, Barry-John Theobald, Josh Susskind

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

Comments ICML 2025

2502.15678 2025-06-02 cs.LG 79%

Testing the Limits of Fine-Tuning for Improving Visual Cognition in Vision Language Models

Luca M. Schulze Buschoff, Konstantinos Voudouris, Elif Akata, Matthias Bethge, Joshua B. Tenenbaum, Eric Schulz

专题命中指令微调：language model(title,abstract)；分类 cs.LG

2505.23627 2025-05-30 cs.LG 79%

Prompting Whisper for Improved Verbatim Transcription and End-to-end Miscue Detection

Griffin Dietz Smith, Dianna Yee, Jennifer King Chen, Leah Findlater

机构 * Apple（苹果公司）

专题命中指令微调：prompting(title,abstract)；分类 cs.LG

Comments Interspeech 2025

2410.02681 2025-05-30 cs.LG 79%

Understanding and Mitigating Miscalibration in Prompt Tuning for Vision-Language Models

Shuoyuan Wang, Yixuan Li, Hongxin Wei

机构 * Shuoyuan Wang（王舒元）； Yixuan Li（李奕轩）； Hongxin Wei（魏宏鑫）

专题命中指令微调：language model(title,abstract)；分类 cs.LG

Comments Accepted by ICML 2025

2505.22067 2025-05-29 cs.CV cs.AI cs.RO 79%

From Failures to Fixes: LLM-Driven Scenario Repair for Self-Evolving Autonomous Driving

Xinyu Xia, Xingjun Ma, Yunfeng Hu, Ting Qu, Hong Chen, Xun Gong

机构 * Jilin University（吉林大学）； Fudan University（复旦大学）； Tongji University（同济大学）

专题命中指令微调：LLM(title,abstract)；分类 cs.AI

2505.16540 2025-05-23 cs.CV cs.AI 79%

TextureSAM: Towards a Texture Aware Foundation Model for Segmentation

Inbal Cohen, Boaz Meivar, Peihan Tu, Shai Avidan, Gal Oren

机构 * Tel Aviv University（特拉维夫大学）； University of Maryland, College Park（马里兰大学College Park分校）； Stanford University（斯坦福大学）； Technion（技术学院）

专题命中指令微调：foundation model(title,abstract)；分类 cs.AI

2505.16118 2025-05-23 cs.CL stat.AP 79%

Semiotic Reconstruction of Destination Expectation Constructs An LLM-Driven Computational Paradigm for Social Media Tourism Analytics

Haotian Lan, Yao Gao, Yujun Cheng, Wei Yuan, Kun Wang

专题命中指令微调：LLM(title,abstract)；分类 cs.CL

Comments 33 pages, 6 figures

2505.15506 2025-05-22 cs.CV cs.LG 79%

Prompt Tuning Vision Language Models with Margin Regularizer for Few-Shot Learning under Distribution Shifts

Debarshi Brahma, Anuska Roy, Soma Biswas

机构 * Indian Institute of Science（印度科学研究院）

专题命中指令微调：language model(title)；foundation model(abstract)；分类 cs.LG

Comments Published in TMLR (2025)

2505.14646 2025-05-21 cs.CV cs.AI 79%

CAD-Coder: An Open-Source Vision-Language Model for Computer-Aided Design Code Generation

Anna C. Doris, Md Ferdous Alam, Amin Heyrani Nobari, Faez Ahmed

机构 * Massachusetts Institute of Technology（麻省理工学院）

专题命中指令微调：language model(title,abstract)；分类 cs.AI

2505.13150 2025-05-20 cs.LG 79%

Zero-Shot Adaptation of Behavioral Foundation Models to Unseen Dynamics

Maksim Bobrin, Ilya Zisman, Alexander Nikulin, Vladislav Kurenkov, Dmitry Dylov

机构 * Applied AI Institute（应用人工智能研究所）； Computational Imaging Lab（计算成像实验室）； MSU（莫斯科国立大学）； Innopolis University（因诺波利斯大学）

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

2405.06836 2025-05-20 q-bio.BM cs.LG 79%

Improving Targeted Molecule Generation through Language Model Fine-Tuning Via Reinforcement Learning

Salma J. Ahmed, Emad A. Mohammed

机构 * Department of Physics and Computer Science（物理与计算机科学系）

专题命中指令微调：language model(title,abstract)；分类 cs.LG

2503.07851 2025-05-19 cs.LG cs.CV cs.IT math.IT stat.ML 79%

TwinTURBO: Semi-Supervised Fine-Tuning of Foundation Models via Mutual Information Decompositions for Downstream Task and Latent Spaces

Guillaume Quétant, Pavlo Molchanov, Slava Voloshynovskiy

机构 * Dept. of Computer Science University of Geneva（计算机科学系日内瓦大学）； NVIDIA

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

2411.15279 2025-05-07 cs.LG cs.GR 79%

Don't Mesh with Me: Generating Constructive Solid Geometry Instead of Meshes by Fine-Tuning a Code-Generation LLM

Maximilian Mews, Ansar Aynetdinov, Vivian Schiller, Peter Eisert, Alan Akbik

机构 * Humboldt-Universität zu Berlin（柏林洪堡大学）

专题命中指令微调：LLM(title,abstract)；分类 cs.LG

Comments Accepted to the AI for Content Creation Workshop at CVPR 2025

2504.08685 2025-05-06 cs.CV cs.AI 79%

Seaweed-7B: Cost-Effective Training of Video Generation Foundation Model

Team Seawead, Ceyuan Yang, Zhijie Lin, Yang Zhao, Shanchuan Lin, Zhibei Ma, Haoyuan Guo, Hao Chen, Lu Qi, Sen Wang, Feng Cheng, Feilong Zuo, Xuejiao Zeng, Ziyan Yang, Fangyuan Kong, Meng Wei, Zhiwu Qing, Fei Xiao, Tuyen Hoang, Siyu Zhang, Peihao Zhu, Qi Zhao, Jiangqiao Yan, Liangke Gui, Sheng Bi, Jiashi Li, Yuxi Ren, Rui Wang, Huixia Li, Xuefeng Xiao, Shu Liu, Feng Ling, Heng Zhang, Houmin Wei, Huafeng Kuang, Jerry Duncan, Junda Zhang, Junru Zheng, Li Sun, Manlin Zhang, Renfei Sun, Xiaobin Zhuang, Xiaojie Li, Xin Xia, Xuyan Chi, Yanghua Peng, Yuping Wang, Yuxuan Wang, Zhongkai Zhao, Zhuo Chen, Zuquan Song, Zhenheng Yang, Jiashi Feng, Jianchao Yang, Lu Jiang

机构 * ByteDance Seed（字节跳动种子）

专题命中指令微调：foundation model(title,abstract)；分类 cs.AI

Comments Technical report (some typos fixed)

2505.01790 2025-05-06 cs.CV cs.CL cs.MM 79%

Enhancing the Learning Experience: Using Vision-Language Models to Generate Questions for Educational Videos

Markos Stamatakis, Joshua Berger, Christian Wartena, Ralph Ewerth, Anett Hoppe

机构 * TIB – Leibniz Information Centre for Science and Technology（蒂宾根-莱比锡信息科学与技术研究中心）； Hochschule Hannover – Data（汉诺威高等学院-数据）； H Institute for Applied Data Science（汉诺威应用数据科学研究所）； L3S Research Center – Leibniz University Hannover（L3S研究中心-汉诺威莱比锡大学）

专题命中指令微调：language model(title,abstract)；分类 cs.CL

Comments 12 pages (excluding references), 8 tables, 1 equation

2505.01075 2025-05-05 cs.LG 79%

Federated Adapter on Foundation Models: An Out-Of-Distribution Approach

Yiyuan Yang, Guodong Long, Tianyi Zhou, Qinghua Lu, Shanshan Ye, Jing Jiang

机构 * University of Technology Sydney（悉尼技术大学）； University of Maryland（马里兰大学）； Data61, CSIRO（数据61，澳大利亚联邦科学与工业研究组织）

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

2409.09894 2025-05-01 cs.LG econ.EM stat.ME stat.ML 79%

Estimating Wage Disparities Using Foundation Models

Keyon Vafa, Susan Athey, David M. Blei

机构 * Harvard University（哈佛大学）； Stanford University（斯坦福大学）； Columbia University（哥伦比亚大学）

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

2410.03258 2025-04-29 cs.CL 79%

Adaptive BPE Tokenization for Enhanced Vocabulary Adaptation in Finetuning Pretrained Language Models

Gunjan Balde, Soumyadeep Roy, Mainack Mondal, Niloy Ganguly

机构 * Indian Institute of Technology Kharagpur（印度理工学院卡拉格浦分校）

专题命中指令微调：language model(title,abstract)；分类 cs.CL

Comments 11 pages. Accepted at EMNLP Findings 2024 (The 2024 Conference on Empirical Methods in Natural Language Processing)

2504.12471 2025-04-18 cs.LG cs.DC cs.PF 79%

You Don't Need All Attentions: Distributed Dynamic Fine-Tuning for Foundation Models

Shiwei Ding, Lan Zhang, Zhenlin Wang, Giuseppe Ateniese, Xiaoyong Yuan

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

2504.10850 2025-04-16 cs.LG cs.CR 79%

How to Enhance Downstream Adversarial Robustness (almost) without Touching the Pre-Trained Foundation Model?

Meiqi Liu, Zhuoqun Huang, Yue Xing

专题命中指令微调：foundation model(title,abstract)；分类 cs.LG

Comments 22 pages, 2 figures, 12 tables. Include 10 pages of appendices

2503.14936 2025-04-16 cs.SE cs.HC cs.LG 79%

Enhancing Code LLM Training with Programmer Attention

Yifan Zhang, Chen Huang, Zachary Karas, Dung Thuy Nguyen, Kevin Leach, Yu Huang

专题命中指令微调：LLM(title,abstract)；分类 cs.LG

1. 指令微调 11653 篇

SparseLoRA: Accelerating LLM Fine-Tuning with Contextual Sparsity

Large-Scale Data Selection for Instruction Tuning

DepthSeg: Depth prompting in remote sensing semantic segmentation

An Auditing Test To Detect Behavioral Shift in Language Models

Fed-HeLLo: Efficient Federated Foundation Model Fine-Tuning with Heterogeneous LoRA Allocation

Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models

Unveiling Markov Heads in Pretrained Language Models for Offline Reinforcement Learning

Tuning the Right Foundation Models is What you Need for Partial Label Learning

Probabilistic Forecasting for Building Energy Systems using Time-Series Foundation Models

Proxy-FDA: Proxy-based Feature Distribution Alignment for Fine-tuning Vision Foundation Models without Forgetting

Testing the Limits of Fine-Tuning for Improving Visual Cognition in Vision Language Models

Prompting Whisper for Improved Verbatim Transcription and End-to-end Miscue Detection

Understanding and Mitigating Miscalibration in Prompt Tuning for Vision-Language Models

From Failures to Fixes: LLM-Driven Scenario Repair for Self-Evolving Autonomous Driving

TextureSAM: Towards a Texture Aware Foundation Model for Segmentation

Semiotic Reconstruction of Destination Expectation Constructs An LLM-Driven Computational Paradigm for Social Media Tourism Analytics

Prompt Tuning Vision Language Models with Margin Regularizer for Few-Shot Learning under Distribution Shifts

CAD-Coder: An Open-Source Vision-Language Model for Computer-Aided Design Code Generation

Zero-Shot Adaptation of Behavioral Foundation Models to Unseen Dynamics

Improving Targeted Molecule Generation through Language Model Fine-Tuning Via Reinforcement Learning

TwinTURBO: Semi-Supervised Fine-Tuning of Foundation Models via Mutual Information Decompositions for Downstream Task and Latent Spaces

Don't Mesh with Me: Generating Constructive Solid Geometry Instead of Meshes by Fine-Tuning a Code-Generation LLM

Seaweed-7B: Cost-Effective Training of Video Generation Foundation Model

Enhancing the Learning Experience: Using Vision-Language Models to Generate Questions for Educational Videos

Federated Adapter on Foundation Models: An Out-Of-Distribution Approach

Estimating Wage Disparities Using Foundation Models

Adaptive BPE Tokenization for Enhanced Vocabulary Adaptation in Finetuning Pretrained Language Models

You Don't Need All Attentions: Distributed Dynamic Fine-Tuning for Foundation Models

How to Enhance Downstream Adversarial Robustness (almost) without Touching the Pre-Trained Foundation Model?

Enhancing Code LLM Training with Programmer Attention