语言大模型 / LLM - arXivDaily 专题

2312.11795 2023-12-20 cs.CL 77%

MELO: Enhancing Model Editing with Neuron-Indexed Dynamic LoRA

Lang Yu, Qin Chen, Jie Zhou, Liang He

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 In Proceedings of The 38th Annual AAAI Conference on Artificial Intelligence

2311.10961 2023-11-21 cs.CL 77%

Journey of Hallucination-minimized Generative AI Solutions for Financial Decision Makers

Sohini Roychowdhury

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 4 pages, 2 Figures

2311.07897 2023-11-15 cs.CL 77%

CPopQA: Ranking Cultural Concept Popularity by LLMs

Ming Jiang, Mansi Joshi

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

2309.13136 2023-09-26 cs.CV cs.AI 77%

Contextual Emotion Estimation from Image Captions

Vera Yang, Archita Srivastava, Yasaman Etesam, Chuxuan Zhang, Angelica Lim

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 to ACII 2023. Project page: http://rosielab.github.io/emotion-captions/

2309.11478 2023-09-21 cs.AI 77%

Fictional Worlds, Real Connections: Developing Community Storytelling Social Chatbots through LLMs

Yuqian Sun, Hanyi Wang, Pok Man Chan, Morteza Tabibi, Yan Zhang, Huan Lu, Yuheng Chen, Chang Hee Lee, Ali Asadipour

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

2309.08594 2023-09-18 cs.CL 77%

"Merge Conflicts!" Exploring the Impacts of External Distractors to Parametric Knowledge Graphs

Cheng Qian, Xinran Zhao, Sherry Tongshuang Wu

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

2308.12562 2023-08-25 cs.LG stat.ML 77%

Variational Information Pursuit with Large Language and Multimodal Models for Interpretable Predictions

Kwan Ho Ryan Chan, Aditya Chattopadhyay, Benjamin David Haeffele, Rene Vidal

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

2307.03987 2023-08-15 cs.CL 77%

A Stitch in Time Saves Nine: Detecting and Mitigating Hallucinations of LLMs by Validating Low-Confidence Generation

Neeraj Varshney, Wenlin Yao, Hongming Zhang, Jianshu Chen, Dong Yu

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 update to include additional experiments

2306.13865 2023-06-27 cs.CL 77%

IERL: Interpretable Ensemble Representation Learning -- Combining CrowdSourced Knowledge and Distributed Semantic Representations

Yuxin Zi, Kaushik Roy, Vignesh Narayanan, Manas Gaur, Amit Sheth

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 Accepted for publication at the KDD workshop on Knowledge-infused Machine Learning, 2023

2305.10838 2023-06-06 cs.LG cs.PL 77%

ProgSG: Cross-Modality Representation Learning for Programs in Electronic Design Automation

Yunsheng Bai, Atefeh Sohrabizadeh, Zongyue Qin, Ziniu Hu, Yizhou Sun, Jason Cong

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 Requires further polishing

2302.02029 2023-02-07 cs.CL 77%

Towards Few-Shot Identification of Morality Frames using In-Context Learning

Shamik Roy, Nishanth Sridhar Nakshatri, Dan Goldwasser

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 Accepted to the 5th Workshop on NLP and CSS at EMNLP 2022

2212.13371 2022-12-29 cs.AI cs.HC econ.GN q-fin.EC 77%

Measuring an artificial intelligence agent's trust in humans using machine incentives

Tim Johnson, Nick Obradovich

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

2206.13757 2022-06-29 cs.CL cs.CY 77%

Flexible text generation for counterfactual fairness probing

Zee Fryer, Vera Axelrod, Ben Packer, Alex Beutel, Jilin Chen, Kellie Webster

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

2402.12264 2025-05-22 cs.LG cs.AI cs.CL stat.ML 76%

Uncertainty quantification in fine-tuned LLMs using LoRA ensembles

Oleksandr Balabanov, Hampus Linander

机构 * Stockholm University Department of Physics（斯德哥尔摩大学物理系）； Department of Mathematical Sciences（数学科学系）； Chalmers university of technology & University of Gothenburg（楚德勒斯技术大学及哥德堡大学）； VERSES Research Lab（VERSES研究实验室）

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

Comments Accepted for ICLR2025 Workshop "Quantify Uncertainty and Hallucination in Foundation Models: The Next Frontier in Reliable AI"

2607.16259 2026-07-21 cs.LG cs.CL stat.ML 新提交 76%

Quantifying Ranking Uncertainty in LLM Benchmarks

量化语言模型基准测试中的排名不确定性

Bitya Neuhof, Yuval Benjamini

专题命中知识编辑与模型理解：LLM(title)；分类 cs.CL、cs.LG

AI总结研究量化语言模型基准测试中排名不确定性问题，通过汇总成对假设检验来实现，分析了知识评估基准MMLU的不确定性来源并展示如何修改假设检验，指出MMLU各主题排名变异性大，比较模型时应考虑。

2606.26783 2026-07-08 cs.LG cs.CL 新提交 76%

Reproducibility Study of "AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models"

可重复性研究："AlphaEdit: 面向语言模型的零空间约束知识编辑"

Ananth K Suresh, Arya Hariharan

机构 * Independent（独立研究者）

专题命中知识编辑与模型理解：language model(title)；分类 cs.CL、cs.LG

AI总结本研究复现了AlphaEdit知识编辑方法，发现其在原始设置下结果可复现，但扩展到新架构和大量顺序编辑时性能下降，表明其理论保证有边界。

Comments 21 pages, 2 figures

2605.19462 2026-05-20 cs.LG cs.AI 76%

Quantifying the Pre-training Dividend: Generative versus Latent Self-Supervised Learning for Time Series Foundation Models

量化预训练红利：生成与潜在自监督学习在时间序列基础模型中的应用

Noam Major, Kathy Razmadze, Yoli Shavit

机构 * Faculty of Engineering, Bar-Ilan University（巴伊兰大学工程学院）

专题命中知识编辑与模型理解：foundation model(title)；分类 cs.AI、cs.LG

AI总结本文研究了自监督学习在时间序列中的应用，比较了生成范式与潜在对齐架构，发现预训练红利在异常检测和分类任务中显著提升，但在预测任务中效果有限，同时表明表示质量与数据来源无关，且在适度的架构深度下趋于稳定。

2604.08645 2026-04-13 cs.CV cs.AI cs.LG cs.RO 76%

3D-VCD: Hallucination Mitigation in 3D-LLM Embodied Agents through Visual Contrastive Decoding

3D-VCD：通过视觉对比解码缓解3D-LLM具身代理中的幻觉

Makanjuola Ogunleye, Eman Abdelrahman, Ismini Lourentzou

机构 * Virginia Tech（弗吉尼亚理工大学）； University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）

专题命中知识编辑与模型理解：LLM(title)；分类 cs.AI、cs.LG

AI总结本文提出3D-VCD，一种用于缓解3D具身代理幻觉的视觉对比解码框架，通过构造扭曲的3D场景图来提升 grounded 推理能力，实验表明其在3D-POPE和HEAL基准上有效。

Comments 8 pages, 6 figures, Accepted at IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026

2603.04972 2026-03-06 cs.LG cs.CL 76%

Functionality-Oriented LLM Merging on the Fisher--Rao Manifold

面向功能的LLM在Fisher-Rao流形上的融合

Jiayu Wang, Zuojun Ye, Wenpeng Yin

机构 * Pennsylvania State University（宾夕法尼亚州立大学）

专题命中知识编辑与模型理解：LLM(title)；分类 cs.CL、cs.LG

AI总结本文提出在Fisher-Rao流形上计算加权Karcher均值，以实现更稳定的LLM融合，提升模型性能。

Comments 9 pages, 2 figures

2602.06129 2026-02-09 cs.LG cs.AI 76%

Urban Spatio-Temporal Foundation Models for Climate-Resilient Housing: Scaling Diffusion Transformers for Disaster Risk Prediction

城市时空基础模型用于气候韧性住房：扩散变换器的扩展用于灾害风险预测

Olaf Yunus Laitinen Imanov, Derya Umut Kulali, Taner Yilmaz

机构 * Technical University of Denmark（技术大学）； Eskisehir Technical University（埃斯基谢普大学）； Afyon Kocatepe University（阿夫yon卡奥塔佩大学）

专题命中知识编辑与模型理解：foundation model(title)；分类 cs.AI、cs.LG

AI总结本文提出Skjold-DiT模型，通过整合时空城市数据预测建筑气候风险，并结合交通网络结构提升灾害响应能力。

Comments 10 pages, 5 figures. Submitted to IEEE Transactions on Intelligent Vehicles

2601.16766 2026-01-26 cs.CL cs.AI 76%

Do LLM hallucination detectors suffer from low-resource effect?

大型语言模型的幻觉检测器是否受到低资源效应影响？

Debtanu Datta, Mohan Kishore Chilukuri, Yash Kumar, Saptarshi Ghosh, Muhammad Bilal Zafar

专题命中知识编辑与模型理解：LLM(title)；分类 cs.CL、cs.AI

AI总结研究发现，幻觉检测器在低资源语言中表现较任务本身更稳健，可能因内部机制编码了不确定性信号。

Comments Accepted at EACL 2026 (Main)

2509.24793 2025-12-18 cs.SD cs.AI cs.LG eess.AS 76%

Sparse Autoencoders Make Audio Foundation Models more Explainable

稀疏自编码器使音频基础模型更加可解释

Théo Mariotte, Martin Lebourdais, Antonio Almudévar, Marie Tahon, Alfonso Ortega, Nicolas Dugué

机构 * LIUM, Le Mans Université（利姆大学LIUM）； VivoLab, I3A, University of Zaragoza（瓦沃拉实验室、I3A、萨拉戈萨大学）

专题命中知识编辑与模型理解：foundation model(title)；分类 cs.AI、cs.LG

AI总结本研究利用稀疏自编码器分析音频预训练模型的隐藏表示，揭示其在歌唱技巧分类中的可解释性提升及对自监督学习系统的作用。

Comments 5 pages, 5 figures, 1 table, submitted to ICASSP 2026

2510.12699 2025-10-15 cs.CL cs.AI 76%

Generation Space Size: Understanding and Calibrating Open-Endedness of LLM Generations

Sunny Yu, Ahmad Jabbar, Robert Hawkins, Dan Jurafsky, Myra Cheng

机构 * Department of Computer Science, Stanford University（计算机科学系，斯坦福大学）； Department of Linguistics, Stanford University（语言学系，斯坦福大学）

专题命中知识编辑与模型理解：LLM(title)；分类 cs.CL、cs.AI

2505.15038 2025-07-31 cs.CL cs.AI 76%

Denoising Concept Vectors with Sparse Autoencoders for Improved Language Model Steering

Haiyan Zhao, Xuansheng Wu, Fan Yang, Bo Shen, Ninghao Liu, Mengnan Du

机构 * New Jersey Institute of Technology（新泽西理工学院）； University of Georgia（佐治亚大学）； Wake Forest University（威克森林大学）

专题命中知识编辑与模型理解：language model(title)；分类 cs.CL、cs.AI

Comments 12 pages, 4 figures, 4 tables

2507.10015 2025-07-18 cs.CV cs.AI cs.LG 76%

(Almost) Free Modality Stitching of Foundation Models

Jaisidh Singh, Diganta Misra, Boris Knyazev, Antonio Orvieto

机构 * University of Tübingen（图宾根大学）； Zuse School ELIZA（Zuse学校ELIZA）； ELLIS Institute Tübingen（图宾根ELLIS研究所）； MPI-IS Tübingen（图宾根MPI-IS研究所）； SAIT AI Lab Montréal（蒙特利尔SAIT人工智能实验室）； Tübingen AI Center（图宾根人工智能中心）

专题命中知识编辑与模型理解：foundation model(title)；分类 cs.AI、cs.LG

Comments Pre-print

2408.08852 2025-06-18 cs.AI cs.LG 76%

A Unified Framework for Next-Gen Urban Forecasting via LLM-driven Dependency Retrieval and GeoTransformer

Yuhao Jia, Zile Wu, Shengao Yi, Yifei Sun, Xiao Huang

机构 * Emory University, University of Pennsylvania（埃默里大学和宾夕法尼亚大学）； University of Pennsylvania（宾夕法尼亚大学）； Emory University（埃默里大学）

专题命中知识编辑与模型理解：LLM(title)；分类 cs.AI、cs.LG

2506.02481 2025-06-04 cs.CL cs.AI 76%

Do Language Models Think Consistently? A Study of Value Preferences Across Varying Response Lengths

Inderjeet Nair, Lu Wang

机构 * University of Michigan（密歇根大学）

专题命中知识编辑与模型理解：language model(title)；分类 cs.CL、cs.AI

2505.20487 2025-05-28 cs.CL cs.AI 76%

InFact: Informativeness Alignment for Improved LLM Factuality

Roi Cohen, Russa Biswas, Gerard de Melo

机构 * Hasso Plattner Institute University of Potsdam（霍普夫纳研究所波茨坦大学）； Dept. of Computer Science Aalborg University（计算机科学系奥尔堡大学）

专题命中知识编辑与模型理解：LLM(title)；分类 cs.CL、cs.AI

2412.11333 2025-05-27 cs.CL cs.AI 76%

Segment-Level Diffusion: A Framework for Controllable Long-Form Generation with Diffusion Language Models

Xiaochen Zhu, Georgi Karadzhov, Chenxi Whitehouse, Andreas Vlachos

机构 * University of Cambridge（剑桥大学）

专题命中知识编辑与模型理解：language model(title)；分类 cs.CL、cs.AI

Comments 9 pages (main body), 3 figures (main body), ACL 2025 Main

2410.19889 2024-10-29 cs.CL cs.LG 76%

Ensembling Finetuned Language Models for Text Classification

Sebastian Pineda Arango, Maciej Janowski, Lennart Purucker, Arber Zela, Frank Hutter, Josif Grabocka

专题命中知识编辑与模型理解：language model(title)；分类 cs.CL、cs.LG

Comments Workshop on Fine-Tuning in Modern Machine Learning @ NeurIPS 2024. arXiv admin note: text overlap with arXiv:2410.04520

1. 知识编辑与模型理解 7583 篇

MELO: Enhancing Model Editing with Neuron-Indexed Dynamic LoRA

Journey of Hallucination-minimized Generative AI Solutions for Financial Decision Makers

CPopQA: Ranking Cultural Concept Popularity by LLMs

Contextual Emotion Estimation from Image Captions

Fictional Worlds, Real Connections: Developing Community Storytelling Social Chatbots through LLMs

"Merge Conflicts!" Exploring the Impacts of External Distractors to Parametric Knowledge Graphs

Variational Information Pursuit with Large Language and Multimodal Models for Interpretable Predictions

A Stitch in Time Saves Nine: Detecting and Mitigating Hallucinations of LLMs by Validating Low-Confidence Generation

IERL: Interpretable Ensemble Representation Learning -- Combining CrowdSourced Knowledge and Distributed Semantic Representations

ProgSG: Cross-Modality Representation Learning for Programs in Electronic Design Automation

Towards Few-Shot Identification of Morality Frames using In-Context Learning

Measuring an artificial intelligence agent's trust in humans using machine incentives

Flexible text generation for counterfactual fairness probing

Uncertainty quantification in fine-tuned LLMs using LoRA ensembles

Quantifying Ranking Uncertainty in LLM Benchmarks

Reproducibility Study of "AlphaEdit: Null-Space Constrained Knowledge Editing for Language Models"

Quantifying the Pre-training Dividend: Generative versus Latent Self-Supervised Learning for Time Series Foundation Models

3D-VCD: Hallucination Mitigation in 3D-LLM Embodied Agents through Visual Contrastive Decoding

Functionality-Oriented LLM Merging on the Fisher--Rao Manifold

Urban Spatio-Temporal Foundation Models for Climate-Resilient Housing: Scaling Diffusion Transformers for Disaster Risk Prediction

Do LLM hallucination detectors suffer from low-resource effect?

Sparse Autoencoders Make Audio Foundation Models more Explainable

Generation Space Size: Understanding and Calibrating Open-Endedness of LLM Generations

Denoising Concept Vectors with Sparse Autoencoders for Improved Language Model Steering

(Almost) Free Modality Stitching of Foundation Models

A Unified Framework for Next-Gen Urban Forecasting via LLM-driven Dependency Retrieval and GeoTransformer

Do Language Models Think Consistently? A Study of Value Preferences Across Varying Response Lengths

InFact: Informativeness Alignment for Improved LLM Factuality

Segment-Level Diffusion: A Framework for Controllable Long-Form Generation with Diffusion Language Models

Ensembling Finetuned Language Models for Text Classification