语言大模型 / LLM - arXivDaily 专题

2506.07735 2025-06-10 cs.LG cs.CV 77%

Language Embedding Meets Dynamic Graph: A New Exploration for Neural Architecture Representation Learning

Haizhao Jing, Haokui Zhang, Zhenhao Shang, Rong Xiao, Peng Wang, Yanning Zhang

机构 * Northwestern Polytechnical University（西北工业大学）； Intellifusion

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.LG

Comments 9 pages, 3 figures

2411.12584 2025-06-10 cs.CV cs.AI 77%

Leveraging MLLM Embeddings and Attribute Smoothing for Compositional Zero-Shot Learning

Xudong Yan, Songhe Feng, Yang Zhang, Jian Yang, Yueguan Lin, Haojun Fei

机构 * School of Computer Science and Technology, Beijing Jiaotong University（计算机科学与技术学院，北京交通大学）； Qifu Technology（启福科技）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.AI

Comments Accepted by IJCAI 2025

2506.04290 2025-06-10 q-fin.RM cs.LG 77%

Interpretable LLMs for Credit Risk: A Systematic Review and Taxonomy

Muhammed Golec, Maha AlabdulJalil

机构 * School of Electronic Engineering and Computer Science, Queen Mary University of London（伦敦大学电子工程与计算机科学学院）； Institute for Data Science and Artificial Intelligence, Boğaziçi University（博雅大学数据科学与人工智能研究所）； College of Science Computer Science Department, Kuwait University（科威特大学科学学院计算机科学系）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.LG

Comments 20 pages, 6 figures, updated author affiliation and removed prior submission note

2411.07965 2025-06-04 cs.CL 77%

SHARP: Unlocking Interactive Hallucination via Stance Transfer in Role-Playing LLMs

Chuyi Kong, Ziyang Luo, Hongzhan Lin, Zhiyuan Fan, Yaxin Fan, Yuxi Sun, Jing Ma

专题命中知识编辑与模型理解：large language model(abstract)；language model(abstract)；post-training(abstract)；分类 cs.CL

Comments 28 pages, unfortunately accepted to findings with Meta 4, acknowledge and apologize to the reviewers and area chair who support our work in the discussion period

Journal ref ACL 2025 Finding

2408.08769 2025-06-04 cs.CL 77%

Lower Layers Matter: Alleviating Hallucination via Multi-Layer Fusion Contrastive Decoding with Truthfulness Refocused

Dingwei Chen, Feiteng Fang, Shiwen Ni, Feng Liang, Xiping Hu, Ahmadreza Argha, Hamid Alinejad-Rokny, Min Yang, Chengming Li

机构 * Sun Yat-Sen University（中山大学）； Shenzhen MSU-BIT University（深圳MSU-BIT大学）； UNSW, Sydney, NSW 2052, Australia（新南威尔士大学，悉尼，新南威尔士州，2052，澳大利亚）； Shenzhen Key Laboratory for High Performance Data Mining, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences（深圳高性能数据挖掘关键实验室，深圳先进技术研究院，中国科学院）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

2412.18069 2025-06-03 cs.CL 77%

Improving Factuality with Explicit Working Memory

Mingda Chen, Yang Li, Karthik Padthe, Rulin Shao, Alicia Sun, Luke Zettlemoyer, Gargi Ghosh, Wen-tau Yih

机构 * Meta FAIR

专题命中知识编辑与模型理解：large language model(abstract)；language model(abstract)；prompting(abstract)；分类 cs.CL

Comments ACL 2025 Camera Ready

2506.00823 2025-06-03 cs.CL 77%

Probing the Geometry of Truth: Consistency and Generalization of Truth Directions in LLMs Across Logical Transformations and Question Answering Tasks

Yuntai Bao, Xuhong Zhang, Tianyu Du, Xinkui Zhao, Zhengwen Feng, Hao Peng, Jianwei Yin

机构 * Zhejiang University（浙江大学）； Zhejiang Normal University（浙江师范大学）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

Comments 19 pages, 16 figures; accepted to Findings of ACL 2025

2506.00274 2025-06-03 cs.CR cs.AI 77%

Chances and Challenges of the Model Context Protocol in Digital Forensics and Incident Response

Jan-Niclas Hilgert, Carlo Jakobs, Michael Külper, Martin Lambertz, Axel Mahr, Elmar Padilla

机构 * Fraunhofer FKIE（弗劳恩霍夫研究所）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.AI

2505.24650 2025-06-02 cs.CE cs.LG q-fin.ST 77%

Beyond the Black Box: Interpretability of LLMs in Finance

Hariom Tatsat, Ariye Shater

机构 * Barclays（巴克莱银行）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.LG

Comments 28 pages, 15 figures

2505.21701 2025-06-02 cs.CL 77%

Do We Know What LLMs Don't Know? A Study of Consistency in Knowledge Probing

Raoyuan Zhao, Abdullatif Köksal, Ali Modarressi, Michael A. Hedderich, Hinrich Schütze

机构 * LMU Munich（慕尼黑莱布尼茨大学）； Munich Center for Machine Learning (MCML)（慕尼黑机器学习中心）

专题命中知识编辑与模型理解：large language model(abstract)；language model(abstract)；prompting(abstract)；分类 cs.CL

2505.23410 2025-05-30 cs.CL 77%

From Parameters to Prompts: Understanding and Mitigating the Factuality Gap between Fine-Tuned LLMs

Xuan Gong, Hanbo Huang, Shiyu Liang

机构 * Tongji University（同济大学）； Shanghai Jiao Tong University（上海交通大学）

专题命中知识编辑与模型理解：large language model(abstract)；language model(abstract)；prompting(abstract)；分类 cs.CL

Comments The code of this paper will be released soon

2505.23223 2025-05-30 cs.LG 77%

Daunce: Data Attribution through Uncertainty Estimation

Xingyuan Pan, Chenlu Ye, Joseph Melkonian, Jiaqi W. Ma, Tong Zhang

机构 * University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）； Womp Labs（Womp实验室）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.LG

2412.04141 2025-05-30 cs.CL 77%

Reducing Tool Hallucination via Reliability Alignment

Hongshen Xu, Zichen Zhu, Lei Pan, Zihan Wang, Su Zhu, Da Ma, Ruisheng Cao, Lu Chen, Kai Yu

机构 * X-LANCE Lab, School of Computer Science, Shanghai Jiao Tong University, Shanghai, China.（上海交通大学计算机科学学院X-LANCE实验室）； MoE Key Lab of Artificial Intelligence, Shanghai, China.（人工智能MoE重点实验室）； Jiangsu Key Lab of Language Computing, Suzhou, China.（江苏语言计算重点实验室）； Suzhou Laboratory, Suzhou, China.（苏州实验室）； AISpeech Co., Ltd., Suzhou, China.（AISpeech公司）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

2505.22298 2025-05-29 cs.CL 77%

Adaptive Detoxification: Safeguarding General Capabilities of LLMs through Toxicity-Aware Knowledge Editing

Yifan Lu, Jing Li, Yigeng Zhou, Yihui Zhang, Wenya Wang, Xiucheng Li, Meishan Zhang, Fangming Liu, Jun Yu, Min Zhang

机构 * Harbin Institute of Technology, Shenzhen, China（哈尔滨工业大学（深圳））； Nanyang Technological University（南洋理工大学）； Peng Cheng Laboratory（鹏城实验室）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

Comments ACL 2025 Findings

2502.11100 2025-05-29 cs.CL 77%

Towards Achieving Concept Completeness for Textual Concept Bottleneck Models

Milan Bhan, Yann Choho, Pierre Moreau, Jean-Noel Vittaut, Nicolas Chesneau, Marie-Jeanne Lesot

机构 * Sorbonne Université, CNRS, LIP6（索邦大学、国家科学研究中心、LIP6）； Ekimetrics

专题命中知识编辑与模型理解：LLM(abstract)；language model(abstract)；small language model(abstract)；分类 cs.CL

2505.15970 2025-05-23 cs.CV cs.LG 77%

Analyzing Hierarchical Structure in Vision Models with Sparse Autoencoders

Matthew Lyle Olson, Musashi Hinck, Neale Ratzlaff, Changbai Li, Phillip Howard, Vasudev Lal, Shao-Yen Tseng

专题命中知识编辑与模型理解：large language model(abstract)；language model(abstract)；foundation model(abstract)；分类 cs.LG

Comments (Oral) CVPR 2025 Workshop on Mechanistic Interpretability for Vision. Authors 1 and 2 contributed equally

2410.12924 2025-05-21 cs.CL 77%

Interpreting token compositionality in LLMs: A robustness analysis

Nura Aljaafari, Danilo S. Carvalho, André Freitas

机构 * Department of Computer Science, University of Manchester（曼彻斯特大学计算机科学系）； Idiap Research Institute（Idiap研究机构）； National Biomarker Centre, CRUK-MI, Univ. of Manchester（国家生物标记中心，CRUK-MI，曼彻斯特大学）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

Comments 23 pages, 3 Figures, 14 tables

2505.12265 2025-05-20 cs.CL 77%

Learning Auxiliary Tasks Improves Reference-Free Hallucination Detection in Open-Domain Long-Form Generation

Chengwei Qin, Wenxuan Zhou, Karthik Abinav Sankararaman, Nanshu Wang, Tengyu Xu, Alexander Radovic, Eryk Helenowski, Arya Talebzadeh, Aditya Tayade, Sinong Wang, Shafiq Joty, Han Fang, Hao Ma

机构 * HKUST(GZ)（香港科技大学（广州））； NTU（国立新加坡大学）； GenAI, Meta（生成式人工智能，Meta）

专题命中知识编辑与模型理解：large language model(abstract)；language model(abstract)；prompting(abstract)；分类 cs.CL

2505.07309 2025-05-13 cs.LG 77%

Uncertainty Profiles for LLMs: Uncertainty Source Decomposition and Adaptive Model-Metric Selection

Pei-Fu Guo, Yun-Da Tsai, Shou-De Lin

机构 * National Taiwan University（国立台湾大学）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.LG

2505.06224 2025-05-12 cs.LG 77%

Towards a Unified Representation Evaluation Framework Beyond Downstream Tasks

Christos Plachouras, Julien Guinot, George Fazekas, Elio Quinton, Emmanouil Benetos, Johan Pauwels

机构 * Queen Mary University of London（伦敦女王学院）； Universal Music Group（环球音乐集团）

专题命中知识编辑与模型理解：foundation model(abstract)；pretraining(abstract)；prompting(abstract)；分类 cs.LG

Comments Accepted at IJCNN 2025

2505.04844 2025-05-09 cs.CL 77%

Osiris: A Lightweight Open-Source Hallucination Detection System

Alex Shan, John Bauer, Christopher D. Manning

机构 * Stanford University（斯坦福大学）； Stanford HAI（斯坦福HAI）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

Comments Stanford 191W

2505.03547 2025-05-07 cs.AI 77%

STORY2GAME: Generating (Almost) Everything in an Interactive Fiction Game

Eric Zhou, Shreyas Basavatia, Moontashir Siam, Zexin Chen, Mark O. Riedl

机构 * School of Interactive Computing, College of Computing（交互计算学院、计算机学院）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.AI

2409.07394 2025-04-30 cs.CL 77%

AdaCAD: Adaptively Decoding to Balance Conflicts between Contextual and Parametric Knowledge

Han Wang, Archiki Prasad, Elias Stengel-Eskin, Mohit Bansal

机构 * UNC Chapel Hill（北卡罗来纳大学教堂山分校）

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

Comments NAACL 2025 (main conference), Code: https://github.com/HanNight/AdaCAD

2504.11104 2025-04-16 cs.CL cs.CV cs.CY 77%

Using LLMs as prompt modifier to avoid biases in AI image generators

René Peinl

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

2504.02395 2025-04-04 cs.CL 77%

The quasi-semantic competence of LLMs: a case study on the part-whole relation

Mattia Proietti, Alessandro Lenci

专题命中知识编辑与模型理解：large language model(abstract)；language model(abstract)；prompting(abstract)；分类 cs.CL

2504.00029 2025-04-02 cs.SE cs.AI 77%

Generating Structured Plan Representation of Procedures with LLMs

Deepeka Garg, Sihan Zeng, Sumitra Ganesh, Leo Ardon

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.AI

2407.15441 2025-04-01 cs.CL 77%

Developing a Reliable, Fast, General-Purpose Hallucination Detection and Mitigation Service

Song Wang, Xun Wang, Jie Mei, Yujia Xie, Sean Muarray, Zhang Li, Lingfeng Wu, Si-Qing Chen, Wayne Xiong

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

2411.05609 2025-03-17 cs.CV cs.LG 77%

A Two-Step Concept-Based Approach for Enhanced Interpretability and Trust in Skin Lesion Diagnosis

Cristiano Patrício, Luís F. Teixeira, João C. Neves

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.LG

Comments Published in the Computational and Structural Biotechnology Journal

2405.21030 2025-03-17 cs.AI 77%

Standards for Belief Representations in LLMs

Daniel A. Herrmann, Benjamin A. Levinstein

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.AI

Journal ref Minds and Machines 35:5 (2025) 1-25

2405.03279 2025-03-17 cs.CL 77%

Lifelong Knowledge Editing for LLMs with Retrieval-Augmented Continuous Prompt Learning

Qizhou Chen, Taolin Zhang, Xiaofeng He, Dongyang Li, Chengyu Wang, Longtao Huang, Hui Xue

专题命中知识编辑与模型理解：LLM(abstract)；large language model(abstract)；language model(abstract)；分类 cs.CL

Comments EMNLP 2024 main

1. 知识编辑与模型理解 7552 篇

Language Embedding Meets Dynamic Graph: A New Exploration for Neural Architecture Representation Learning

Leveraging MLLM Embeddings and Attribute Smoothing for Compositional Zero-Shot Learning

Interpretable LLMs for Credit Risk: A Systematic Review and Taxonomy

SHARP: Unlocking Interactive Hallucination via Stance Transfer in Role-Playing LLMs

Lower Layers Matter: Alleviating Hallucination via Multi-Layer Fusion Contrastive Decoding with Truthfulness Refocused

Improving Factuality with Explicit Working Memory

Probing the Geometry of Truth: Consistency and Generalization of Truth Directions in LLMs Across Logical Transformations and Question Answering Tasks

Chances and Challenges of the Model Context Protocol in Digital Forensics and Incident Response

Beyond the Black Box: Interpretability of LLMs in Finance

Do We Know What LLMs Don't Know? A Study of Consistency in Knowledge Probing

From Parameters to Prompts: Understanding and Mitigating the Factuality Gap between Fine-Tuned LLMs

Daunce: Data Attribution through Uncertainty Estimation

Reducing Tool Hallucination via Reliability Alignment

Adaptive Detoxification: Safeguarding General Capabilities of LLMs through Toxicity-Aware Knowledge Editing

Towards Achieving Concept Completeness for Textual Concept Bottleneck Models

Analyzing Hierarchical Structure in Vision Models with Sparse Autoencoders

Interpreting token compositionality in LLMs: A robustness analysis

Learning Auxiliary Tasks Improves Reference-Free Hallucination Detection in Open-Domain Long-Form Generation

Uncertainty Profiles for LLMs: Uncertainty Source Decomposition and Adaptive Model-Metric Selection

Towards a Unified Representation Evaluation Framework Beyond Downstream Tasks

Osiris: A Lightweight Open-Source Hallucination Detection System

STORY2GAME: Generating (Almost) Everything in an Interactive Fiction Game

AdaCAD: Adaptively Decoding to Balance Conflicts between Contextual and Parametric Knowledge

Using LLMs as prompt modifier to avoid biases in AI image generators

The quasi-semantic competence of LLMs: a case study on the part-whole relation

Generating Structured Plan Representation of Procedures with LLMs

Developing a Reliable, Fast, General-Purpose Hallucination Detection and Mitigation Service

A Two-Step Concept-Based Approach for Enhanced Interpretability and Trust in Skin Lesion Diagnosis

Standards for Belief Representations in LLMs

Lifelong Knowledge Editing for LLMs with Retrieval-Augmented Continuous Prompt Learning