语言大模型 / LLM - arXivDaily 专题

2408.09657 2024-08-20 cs.SE 88%

Impact of Large Language Models of Code on Fault Localization

Suhwan Ji, Sanghwa Lee, Changsup Lee, Hyeonseung Im, Yo-Sub Han

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

2408.09017 2024-08-20 cs.IR 88%

Meta Knowledge for Retrieval Augmented Large Language Models

Laurent Mombaerts, Terry Ding, Adi Banerjee, Florian Felice, Jonathan Taws, Tarik Borogovac

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

Comments Accepted in Workshop on Generative AI for Recommender Systems and Personalization, KDD 2024

2312.15599 2024-07-31 cs.IR 88%

Preliminary Study on Incremental Learning for Large Language Model-based Recommender Systems

Tianhao Shi, Yang Zhang, Zhijian Xu, Chong Chen, Fuli Feng, Xiangnan He, Qi Tian

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

Comments accepted in the short paper track of the 2024 ACM International Conference on Information and Knowledge Management (CIKM 2024)

2407.15087 2024-07-23 cs.CV 88%

Navigation Instruction Generation with BEV Perception and Large Language Models

Sheng Fan, Rui Liu, Wenguan Wang, Yi Yang

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

Comments ECCV 2024; Project Page: https://github.com/FanScy/BEVInstructor

2407.11422 2024-07-17 cs.CV 88%

Reflective Instruction Tuning: Mitigating Hallucinations in Large Vision-Language Models

Jinrui Zhang, Teng Wang, Haigang Zhang, Ping Lu, Feng Zheng

专题命中指令微调：language model(title,abstract)；instruction tuning(title,abstract)

Comments To appear at ECCV2024

2406.05365 2024-06-25 cs.CL cs.AI cs.LG 88%

CaLM: Contrasting Large and Small Language Models to Verify Grounded Generation

I-Hung Hsu, Zifeng Wang, Long T. Le, Lesly Miculicich, Nanyun Peng, Chen-Yu Lee, Tomas Pfister

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

Comments ACL 2024 Camera Ready Version

2312.02546 2024-05-30 cs.CV 88%

Machine Vision Therapy: Multimodal Large Language Models Can Enhance Visual Robustness via Denoising In-Context Learning

Zhuo Huang, Chang Liu, Yinpeng Dong, Hang Su, Shibao Zheng, Tongliang Liu

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

Comments ICML 2024

2405.03066 2024-05-24 cs.ET 88%

A scoping review of using Large Language Models (LLMs) to investigate Electronic Health Records (EHRs)

Lingyao Li, Jiayan Zhou, Zhenxiang Gao, Wenyue Hua, Lizhou Fan, Huizi Yu, Loni Hagen, Yongfeng Zhang, Themistocles L. Assimes, Libby Hemphill, Siyuan Ma

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

2308.11103 2024-05-21 cs.CL cs.AI cs.IR cs.LG 88%

Anonymity at Risk? Assessing Re-Identification Capabilities of Large Language Models

Alex Nyffenegger, Matthias Stürmer, Joel Niklaus

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

Comments Accepted to NAACL Findings 2024

2312.02233 2024-05-13 cs.CV 88%

MedXChat: A Unified Multimodal Large Language Model Framework towards CXRs Understanding and Generation

Ling Yang, Zhanyu Wang, Zhenghao Chen, Xinyu Liang, Luping Zhou

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

2310.10221 2024-02-26 cs.RO cs.CV 88%

RoboLLM: Robotic Vision Tasks Grounded on Multimodal Large Language Models

Zijun Long, George Killick, Richard McCreadie, Gerardo Aragon Camarasa

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

2401.05702 2024-01-12 cs.CV 88%

Video Anomaly Detection and Explanation via Large Language Models

Hui Lv, Qianru Sun

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

Comments 9 pages, 6 figures

2401.03217 2024-01-09 cs.RO cs.HC 88%

Understanding Large-Language Model (LLM)-powered Human-Robot Interaction

Callie Y. Kim, Christine P. Lee, Bilge Mutlu

专题命中指令微调：LLM(title,abstract)；language model(title,abstract)

Comments 10 pages, 4 figures. Callie Y. Kim and Christine P. Lee contributed equally to the work. To be published in Proceedings of the 2024 ACM/IEEE International Conference on Human-Robot Interaction (HRI '24), March 11--14, 2024, Boulder, CO, USA

2312.16217 2023-12-29 cs.CV cs.RO 88%

ManipLLM: Embodied Multimodal Large Language Model for Object-Centric Robotic Manipulation

Xiaoqi Li, Mingxu Zhang, Yiran Geng, Haoran Geng, Yuxing Long, Yan Shen, Renrui Zhang, Jiaming Liu, Hao Dong

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

2304.09960 2023-09-15 cs.CL cs.AI cs.LG 88%

A Latent Space Theory for Emergent Abilities in Large Language Models

Hui Jiang

专题命中指令微调：large language model(title)；language model(title)；prompting(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 17 pages, 3 figures

2306.00409 2023-08-23 cs.CV 88%

Adapting Pre-trained Language Models to Vision-Language Tasks via Dynamic Visual Prompting

Shubin Huang, Qiong Wu, Yiyi Zhou, Weijie Chen, Rongsheng Zhang, Xiaoshuai Sun, Rongrong Ji

专题命中指令微调：language model(title,abstract)；prompting(title,abstract)

2307.13896 2023-07-27 cs.DC 88%

Low-Parameter Federated Learning with Large Language Models

Jingang Jiang, Xiangyang Liu, Chenyou Fan

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

2303.14292 2023-03-28 cs.HC 88%

Chat2VIS: Fine-Tuning Data Visualisations using Multilingual Natural Language Text and Pre-Trained Large Language Models

Paula Maddigan, Teo Susnjak

专题命中指令微调：large language model(title,abstract)；language model(title,abstract)

2106.09685 2021-10-19 cs.CL cs.AI cs.LG 88%

LoRA: Low-Rank Adaptation of Large Language Models

Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen

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

Comments Draft V2 includes better baselines, experiments on GLUE, and more on adapter latency

2503.19469 2025-03-31 cs.CL cs.AI 87%

Enhancing Small Language Models for Cross-Lingual Generalized Zero-Shot Classification with Soft Prompt Tuning

Fred Philippy, Siwen Guo, Cedric Lothritz, Jacques Klein, Tegawendé F. Bissyandé

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

Comments Workshop on Language Models for Underserved Communities (co-located with NAACL 2025)

2605.25981 2026-08-27 cs.CL 版本更新 87%

AgentDiff: Meaning-Bearing Rewrites Trigger Deeper Divergence than Presentation Changes in LLM Agents

LLM 代理何时对表面噪声与语义噪声做出不同处理？一项基于 68 个单元格的测量研究及留出轨迹级验证

Liyun Zhang, Jiayi Guo

机构 * School of Information and Software Engineering, UESTC（信息与软件工程学院，电子科技大学）； Jacobs School of Engineering, UC San Diego（工程学院，加州大学圣地亚哥分校）

专题命中指令微调：LLM(title,title_cn)；分类 cs.CL

AI总结本研究通过 68 个单元格的测量实验，发现大语言模型驱动的思维链和 ReAct 代理对语义扰动（如释义、同义词）比表现扰动（如格式、重排序）更敏感，并基于留出模型和轨迹级机制分析提出了“隐蔽发散”解释。

2608.17310 2026-08-24 cs.LG 版本更新 87%

Agentic ESOpt: Fine-Tuning Long-Horizon LLM Agents with Minimal GPU Requirements

Agentic ESOpt：以最小GPU资源微调长视野大语言模型智能体

Zhi Zheng, Rongsheng Chen, Yunpeng Ba, Zhenkun Wang, Yee Whye Teh, Wee Sun Lee

机构 * National University of Singapore（新加坡国立大学）

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

AI总结本文提出Agentic ESOpt框架，用进化策略微调长视野LLM智能体，仅需最小GPU资源，在WebArena-Lite上使Qwen-3.5-27B性能提升6.69%，提示-参数协同演化在多数设置中优于基准。

2603.17450 2026-08-13 cs.IR cs.AI 版本更新 87%

VLM2Rec: Resolving Modality Collapse in Vision-Language Model Embedders for Multimodal Sequential Recommendation

VLM2Rec: 解决视觉语言模型嵌入器在多模态序列推荐中的模态崩溃问题

Junyoung Kim, Woojoo Kim, Wonbin Kweon, Jaehyung Lim, Dongha Kim, Hwanjo Yu

机构 * Pohang University of Science

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

AI总结本文提出VLM2Rec框架，通过弱模态惩罚对比学习和跨模态关系拓扑正则化，解决多模态序列推荐中模态崩溃问题，提升推荐准确性和鲁棒性。

2607.04412 2026-08-11 cs.AI 版本更新 87%

LLM-as-a-Tutor: Policy-Aware Prompt Adaptation for Non-Verifiable RL

大语言模型作为导师：不可验证强化学习中的策略感知提示适应

Yujin Kim, Namgyu Ho, Sangmin Hwang, Joonkee Kim, Yongjin Yang, Sangmin Bae, Seungone Kim, Jaehun Jung, Se-Young Yun, Hwanjun Song

机构 * KAIST（韩国科学技术院）； Upstage ； University of Toronto（多伦多大学）； Carnegie Mellon University（卡内基梅隆大学）； NVIDIA（英伟达）

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

AI总结针对不可验证强化学习中训练提示静态导致的问题，提出LLM-as-a-Tutor框架，让大语言模型从评判扩展为导师，通过对比策略展开检测无挑战性提示并添加约束，提升性能。

2608.07226 2026-08-10 cs.AR cs.LG 新提交 87%

Dual-Node NVIDIA DGX Spark over Tailscale: A Remote-Access Testbed for Distributed LLM Training and Cyber-Threat-Intelligence Fine-Tuning

基于Tailscale的双节点NVIDIA DGX Spark：用于分布式大语言模型训练和网络威胁情报微调的远程访问测试平台

Vasanth Iyer

专题命中指令微调：LLM(title,abstract)；SFT(abstract,abstract_cn)；pretraining(abstract)；分类 cs.LG

AI总结本研究展示了双节点NVIDIA DGX Spark通过Tailscale VPN实现分布式NanoChat预训练，构建网络安全微调数据集并验证其效果，同时该集群可支持研究与教学，证明适度本地基础设施的可行性。

Comments 12 pages, 3 figures, 4 tables. Deployment runbook and code: https://github.com/viyer-research/dual-dgx-spark-tailscale-testbed

2608.01437 2026-08-04 cs.AI 新提交 87%

Beyond Routing Saturation: A Long-Horizon Class-Incremental Perspective on Expert Routing in Multimodal Continual Instruction Tuning

超越路由饱和：多模态持续指令调优中专家路由的长程类增量视角

Huiyu Yi, Yongqi Xu, Bogang Zhang, Dunwei Tu, Xu Zhiming, Zhen-Hao Xie, Baile Xu, Furao Shen

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

AI总结该研究针对多模态持续指令调优中专家路由的任务识别问题，构建含34个任务的FLEX基准，提出MCIL框架，适配CIL方法后使LoRA匹配和MacroScore分别提升最高16.3个百分点、4.6个点

2607.18232 2026-07-23 cs.CL 版本更新 87%

It's Not What You Say, It's How You Say It: Evaluating LLM Responses to Expressions of Belief

重要的不是你说了什么，而是你怎么说：评估大语言模型对信念表达的回应

Kevin Du, Clara Kümpel, Michelle Wastl, Alex Warstadt

机构 * ETH Zürich（苏黎世联邦理工学院）； University of Zurich（苏黎世大学）； UC San Diego（加州大学圣地亚哥分校）

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

AI总结研究用户信念表达形式对大语言模型的影响，引入基于语言维度的类型学，生成可控查询对，评估不同架构、规模和训练阶段的16个LLMs，发现响应行为差异及能显著说服模型的特定信念表达，揭示语言框架对模型情境整合的影响。

Comments Published at ACL 2026

2606.14530 2026-07-17 cs.LG 版本更新 87%

Code Correctness Is Linearly Decodable from LLM Hidden States Before Generation

LLM隐藏状态中的代码正确性信号：生成前探测与修复几何

Carlo Di Cicco

机构 * Independent researcher（独立研究员）

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

AI总结本文通过残差化方法，发现Qwen3-4B-Instruct模型在生成前隐藏状态可线性解码代码正确性（AUC 0.931），但修复成功的方向性信号在控制上下文协变量后消失，揭示了方法学上的正负结果。

Comments 7 pages, 4 tables. Code, data, and analysis scripts available at https://github.com/CarloDiCicco/ReasoningLab

2603.24787 2026-07-15 cs.AI 版本更新 87%

ReLope: KL-Regularized LoRA Probes for Multimodal LLM Routing

ReLope：KL正则化的LoRA探针用于多模态大语言模型路由

Yaopei Zeng, Congchao Wang, Blake JianHang Chen, Lu Lin

机构 * Pennsylvania State University（宾夕法尼亚州立大学）； Amazon AGI（亚马逊AGI）； Google（谷歌）

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

AI总结针对多模态大语言模型路由中探针性能下降问题，提出Attention Probe和KL正则化的LoRA探针ReLope，通过改进隐藏状态质量提升路由效果。

2505.12682 2026-07-15 cs.LG 版本更新 87%

RAFP: Identifying LLM Lineages via Rare-Region Fingerprints

RAFP：通过稀有区域指纹识别大语言模型谱系

Yun-Yun Tsai, Jia Hao Liang, Chuan Guo, Junfeng Yang, Laurens van der Maaten

机构 * Department of Computer Science, Columbia University, USA（哥伦比亚大学计算机科学系）； Meta, USA（Meta公司）

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

AI总结针对大语言模型所有权验证需求，提出RAFP框架，利用稀有区域指纹识别模型谱系。该方法非侵入性，通过离散梯度优化构建指纹，理论分析表明其在微调下似然变化有界，实验显示在黑盒设置中性能优于基线。

Comments 16 pages

1. 指令微调 11692 篇

Impact of Large Language Models of Code on Fault Localization

Meta Knowledge for Retrieval Augmented Large Language Models

Preliminary Study on Incremental Learning for Large Language Model-based Recommender Systems

Navigation Instruction Generation with BEV Perception and Large Language Models

Reflective Instruction Tuning: Mitigating Hallucinations in Large Vision-Language Models

CaLM: Contrasting Large and Small Language Models to Verify Grounded Generation

Machine Vision Therapy: Multimodal Large Language Models Can Enhance Visual Robustness via Denoising In-Context Learning

A scoping review of using Large Language Models (LLMs) to investigate Electronic Health Records (EHRs)

Anonymity at Risk? Assessing Re-Identification Capabilities of Large Language Models

MedXChat: A Unified Multimodal Large Language Model Framework towards CXRs Understanding and Generation

RoboLLM: Robotic Vision Tasks Grounded on Multimodal Large Language Models

Video Anomaly Detection and Explanation via Large Language Models

Understanding Large-Language Model (LLM)-powered Human-Robot Interaction

ManipLLM: Embodied Multimodal Large Language Model for Object-Centric Robotic Manipulation

A Latent Space Theory for Emergent Abilities in Large Language Models

Adapting Pre-trained Language Models to Vision-Language Tasks via Dynamic Visual Prompting

Low-Parameter Federated Learning with Large Language Models

Chat2VIS: Fine-Tuning Data Visualisations using Multilingual Natural Language Text and Pre-Trained Large Language Models

LoRA: Low-Rank Adaptation of Large Language Models

Enhancing Small Language Models for Cross-Lingual Generalized Zero-Shot Classification with Soft Prompt Tuning

AgentDiff: Meaning-Bearing Rewrites Trigger Deeper Divergence than Presentation Changes in LLM Agents

Agentic ESOpt: Fine-Tuning Long-Horizon LLM Agents with Minimal GPU Requirements

VLM2Rec: Resolving Modality Collapse in Vision-Language Model Embedders for Multimodal Sequential Recommendation

LLM-as-a-Tutor: Policy-Aware Prompt Adaptation for Non-Verifiable RL

Dual-Node NVIDIA DGX Spark over Tailscale: A Remote-Access Testbed for Distributed LLM Training and Cyber-Threat-Intelligence Fine-Tuning

Beyond Routing Saturation: A Long-Horizon Class-Incremental Perspective on Expert Routing in Multimodal Continual Instruction Tuning

It's Not What You Say, It's How You Say It: Evaluating LLM Responses to Expressions of Belief

Code Correctness Is Linearly Decodable from LLM Hidden States Before Generation

ReLope: KL-Regularized LoRA Probes for Multimodal LLM Routing

RAFP: Identifying LLM Lineages via Rare-Region Fingerprints