大模型推理能力 - arXivDaily 专题

2510.11842 2025-10-15 cs.LG cs.CL 62%

Balancing Synthetic Data and Replay for Enhancing Task-Specific Capabilities

Urs Spiegelhalter, Jörg K. H. Franke, Frank Hutter

机构 * University of Freiburg（弗赖堡大学）； ELLIS Institute Tübingen（图宾根ELLIS研究所）； Open-Sci Collective（开放科学集体）； LAION ； Prior Labs（Prior实验室）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.LG

Comments Presented at 39th Conference on Neural Information Processing Systems (NeurIPS 2025) Workshop on Continual and Compatible Foundation Model Updates (CCFM)

2507.04295 2025-10-15 cs.CY cs.AI cs.CL cs.HC 62%

LearnLens: LLM-Enabled Personalised, Curriculum-Grounded Feedback with Educators in the Loop

Runcong Zhao, Artem Bobrov, Jiazheng Li, Cesare Aloisi, Yulan He

机构 * King’s College London（伦敦国王学院）； AQA

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments EMNLP 2025

2412.12004 2025-10-15 cs.CL cs.LG 62%

The Open Source Advantage in Large Language Models (LLMs)

Jiya Manchanda, Laura Boettcher, Matheus Westphalen, Jasser Jasser

机构 * Department of Philosophy and Religion（哲学与宗教系）； Rollins College（罗林斯学院）； Department of Mathematics and Computer Science（数学与计算机科学系）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.LG

Comments 9 pages, 1 figure

2509.05396 2025-10-14 cs.CL cs.AI cs.MA 62%

Talk Isn't Always Cheap: Understanding Failure Modes in Multi-Agent Debate

Andrea Wynn, Harsh Satija, Gillian Hadfield

机构 * Johns Hopkins University（约翰霍普金斯大学）； Vector Institute（向量研究所）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments ICML MAS Workshop 2025

2510.11292 2025-10-14 cs.LG cs.AI 62%

LouisKV: Efficient KV Cache Retrieval for Long Input-Output Sequences

Wenbo Wu, Qingyi Si, Xiurui Pan, Ye Wang, Jie Zhang

机构 * Peking University（北京大学）； Huawei Technologies Ltd.（华为技术有限公司）； Chongqing University of Post and Telecommunications（重庆邮电大学）

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

2510.10398 2025-10-14 cs.CL cs.AI 62%

STEAM: A Semantic-Level Knowledge Editing Framework for Large Language Models

Geunyeong Jeong, Juoh Sun, Seonghee Lee, Harksoo Kim

机构 * Konkuk University（韩国康康大学）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Accepted to EMNLP 2025 (Findings)

2510.09893 2025-10-14 cs.CL cs.LG 62%

HIPPD: Brain-Inspired Hierarchical Information Processing for Personality Detection

Guanming Chen, Lingzhi Shen, Xiaohao Cai, Imran Razzak, Shoaib Jameel

机构 * School of Electronics and Computer Science University of Southampton（电子与计算机科学学院英国南安普顿大学）； Department of Computational Biology Mohamed bin Zayed University of Artificial Intelligence（计算生物学系摩洛哥本·扎耶德人工智能大学）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.LG

2509.13380 2025-10-14 cs.RO cs.AI cs.LG cs.MA cs.SY eess.SY 62%

ASTREA: Introducing Agentic Intelligence for Orbital Thermal Autonomy

Alejandro D. Mousist

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

Comments Accepted for presentation at the European Space Agency's AI Start 2025 Conference (see https://atpi.eventsair.com/ai-star-2025/)

2509.07558 2025-10-14 cs.LG cs.AI 62%

VL Norm: Rethink Loss Aggregation in RLVR

Zhiyuan He, Xufang Luo, Yike Zhang, Yuqing Yang, Lili Qiu

机构 * Microsoft Research（微软研究院）； Tsinghua University（清华大学）

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

2510.07488 2025-10-13 cs.CL cs.AI 62%

Can Lessons From Human Teams Be Applied to Multi-Agent Systems? The Role of Structure, Diversity, and Interaction Dynamics

Rasika Muralidharan, Haewoon Kwak, Jisun An

机构 * Indiana University Bloomington（印第安纳大学布卢明顿分校）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Under Review at ARR

2510.08203 2025-10-10 cs.CL cs.AI 62%

Memory Retrieval and Consolidation in Large Language Models through Function Tokens

Shaohua Zhang, Yuan Lin, Hang Li

机构 * ByteDance Seed（字节跳动种子）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

2508.01696 2025-10-10 cs.CL cs.AI 62%

CoCoA: Collaborative Chain-of-Agents for Parametric-Retrieved Knowledge Synergy

Yi Jiang, Sendong Zhao, Jianbo Li, Haochun Wang, Lizhe Zhang, Yan Liu, Bing Qin

机构 * Harbin Institute of Technology（哈尔滨工业大学）； China Mobile Group Heilongjiang Co.,Ltd（中国移动集团黑龙江公司）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments code available at https://github.com/liunian-Jay/CoCoA

2506.19823 2025-10-08 cs.LG cs.AI 62%

Persona Features Control Emergent Misalignment

Miles Wang, Tom Dupré la Tour, Olivia Watkins, Alex Makelov, Ryan A. Chi, Samuel Miserendino, Jeffrey Wang, Achyuta Rajaram, Johannes Heidecke, Tejal Patwardhan, Dan Mossing

机构 * OpenAI

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

2510.04311 2025-10-07 cs.AI cs.LG 62%

On the Importance of Task Complexity in Evaluating LLM-Based Multi-Agent Systems

Bohan Tang, Huidong Liang, Keyue Jiang, Xiaowen Dong

机构 * University of Oxford（牛津大学）； University College London（伦敦大学学院）

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

2510.04196 2025-10-07 cs.AI cs.LG 62%

COSMO-RL: Towards Trustworthy LMRMs via Joint Safety and Stability

Yizhuo Ding, Mingkang Chen, Qiuhua Liu, Fenghua Weng, Wanying Qu, Yue Yang, Yugang Jiang, Zuxuan Wu, Yanwei Fu, Wenqi Shao

机构 * Fudan University（复旦大学）； Shanghai AI Laboratory（上海人工智能实验室）； The University of Hong Kong（香港大学）； Shenzhen University（深圳大学）； ShanghaiTech University（上海交通大学）

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

2510.03521 2025-10-07 cs.CL cs.AI 62%

Identifying Financial Risk Information Using RAG with a Contrastive Insight

Ali Elahi

机构 * Department of Computer Science University of Illinois Chicago（伊利诺伊大学芝加哥分校计算机科学系）； Surlamer Investments（Surlamer投资公司）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments 7 pages, 1 figure, Workshop on Generative AI in Finance, NeurIPS 2025

2510.03442 2025-10-07 cs.LG cs.AI cs.MA 62%

The Argument is the Explanation: Structured Argumentation for Trust in Agents

Ege Cakar, Per Ola Kristensson

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

Comments 8 pages, 4 figures, 6 tables, submitted to IAAI-26

2510.02200 2025-10-03 cs.CL cs.AI 62%

ARUQULA -- An LLM based Text2SPARQL Approach using ReAct and Knowledge Graph Exploration Utilities

Felix Brei, Lorenz Bühmann, Johannes Frey, Daniel Gerber, Lars-Peter Meyer, Claus Stadler, Kirill Bulert

机构 * Institute for Applied Informatics at Leipzig University（莱比锡大学应用信息学院）； Leipzig University（莱比锡大学）； Chemnitz Technical University（Chemnitz技术大学）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments peer reviewed publication at Text2SPARQL Workshop @ ESWC 2025

2510.01293 2025-10-03 cs.AI cs.LG 62%

Cyber Academia-Chemical Engineering (CA-ChemE): A Living Digital Town for Self-Directed Research Evolution and Emergent Scientific Discovery

Zekun Jiang, Chunming Xu, Tianhang Zhou

机构 * State Key Laboratory of Heavy Oil Processing, College of Carbon Neutrality Future Technology, China University of Petroleum (Beijing), Beijing 102249, China（石油加工国家重点实验室，碳中和未来技术学院，中国石油大学（北京），北京102249）

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

2510.00499 2025-10-03 cs.CL cs.AI 62%

MOSS-Speech: Towards True Speech-to-Speech Models Without Text Guidance

Xingjian Zhao, Zhe Xu, Qinyuan Cheng, Zhaoye Fei, Luozhijie Jin, Yang Wang, Hanfu Chen, Yaozhou Jiang, Qinghui Gao, Ke Chen, Ruixiao Li, Mingshu Chen, Ruiming Wang, Wenbo Zhang, Yiyang Zhang, Donghua Yu, Yang Gao, Xiaogui Yang, Yitian Gong, Yuanfan Xu, Yaqian Zhou, Xuanjing Huang, Xipeng Qiu

机构 * SII OpenMOSS Team（SII开放MOSS团队）； Shanghai Innovation Institute（上海创新研究院）； Fudan University（复旦大学）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

2508.02961 2025-10-03 cs.AI cs.CL cs.CR 62%

Defend LLMs Through Self-Consciousness

Boshi Huang, Fabio Nonato de Paula

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments company requests to withdraw

2509.24866 2025-10-02 cs.CL cs.AI 62%

Metaphor identification using large language models: A comparison of RAG, prompt engineering, and fine-tuning

Matteo Fuoli, Weihang Huang, Jeannette Littlemore, Sarah Turner, Ellen Wilding

机构 * Department of Linguistics and Communication（语言学与传播学系）； University of Birmingham（伯明翰大学）； Centre for Arts, Memory & Communities（艺术、记忆与社区中心）； Coventry University（考文垂大学）

专题命中其他推理：chain-of-thought(abstract)；分类 cs.CL、cs.AI

2509.26507 2025-10-01 cs.NE cs.AI cs.LG stat.ML 62%

The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain

Adrian Kosowski, Przemysław Uznański, Jan Chorowski, Zuzanna Stamirowska, Michał Bartoszkiewicz

机构 * Pathway

专题命中其他推理：reasoning(abstract)；分类 cs.AI、cs.LG

Comments Code available at: https://github.com/pathwaycom/bdh Accompanying blog: https://pathway.com/research/bdh

2509.25591 2025-10-01 cs.AI cs.CL q-bio.OT 62%

Building the EHR Foundation Model via Next Event Prediction

Zekai Chen, Arda Pekis, Kevin Brown

机构 * Standard Model Biomedicine（标准模型生物医学）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

2508.19294 2025-10-01 cs.CV cs.AI cs.CL 62%

Object Detection with Multimodal Large Vision-Language Models: An In-depth Review

Ranjan Sapkota, Manoj Karkee

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments First Peer Reviewed Review Paper for Object Detection with Vision-Language Models (VLMs)

Journal ref Information Fusion, 2025

2509.25176 2025-09-30 cs.LG cs.CL 62%

SIRI: Scaling Iterative Reinforcement Learning with Interleaved Compression

Haoming Wen, Yushi Bai, Juanzi Li, Jie Tang

机构 * Tsinghua University（清华大学）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.LG

Comments In submission

2509.25131 2025-09-30 cs.SD cs.AI cs.CL cs.CV cs.MM 62%

MGM-Omni: Scaling Omni LLMs to Personalized Long-Horizon Speech

Chengyao Wang, Zhisheng Zhong, Bohao Peng, Senqiao Yang, Yuqi Liu, Haokun Gui, Bin Xia, Jingyao Li, Bei Yu, Jiaya Jia

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Code is available at https://github.com/dvlab-research/MGM-Omni

2509.24356 2025-09-30 cs.CL cs.AI 62%

Beyond Repetition: Text Simplification and Curriculum Learning for Data-Constrained Pretraining

Matthew Theodore Roque, Dan John Velasco

机构 * Samsung R&D Institute Philippines（三星菲律宾研发中心）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

Comments To be published in BabyLM Workshop at EMNLP 2025

2407.03779 2025-09-30 cs.CL cs.LG 62%

Sheaf Discovery with Joint Computation Graph Pruning and Flexible Granularity

Lei Yu, Jingcheng Niu, Zining Zhu, Xi Chen, Gerald Penn

机构 * University of Toronto（多伦多大学）； UKP Lab（UKP实验室）； Stevens Institute of Technology（斯蒂文斯理工学院）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.LG

2509.23755 2025-09-30 cs.CL cs.AI 62%

Understanding Textual Capability Degradation in Speech LLMs via Parameter Importance Analysis

Chao Wang, Rui-Chen Zheng, Yang Ai, Zhen-Hua Ling

机构 * National Engineering Research Center of Speech and Language Information Processing（语音与语言信息处理国家级工程研究中心）； University of Science and Technology of China（中国科学技术大学）

专题命中其他推理：reasoning(abstract)；分类 cs.CL、cs.AI

1. 其他推理 5813 篇

Balancing Synthetic Data and Replay for Enhancing Task-Specific Capabilities

LearnLens: LLM-Enabled Personalised, Curriculum-Grounded Feedback with Educators in the Loop

The Open Source Advantage in Large Language Models (LLMs)

Talk Isn't Always Cheap: Understanding Failure Modes in Multi-Agent Debate

LouisKV: Efficient KV Cache Retrieval for Long Input-Output Sequences

STEAM: A Semantic-Level Knowledge Editing Framework for Large Language Models

HIPPD: Brain-Inspired Hierarchical Information Processing for Personality Detection

ASTREA: Introducing Agentic Intelligence for Orbital Thermal Autonomy

VL Norm: Rethink Loss Aggregation in RLVR

Can Lessons From Human Teams Be Applied to Multi-Agent Systems? The Role of Structure, Diversity, and Interaction Dynamics

Memory Retrieval and Consolidation in Large Language Models through Function Tokens

CoCoA: Collaborative Chain-of-Agents for Parametric-Retrieved Knowledge Synergy

Persona Features Control Emergent Misalignment

On the Importance of Task Complexity in Evaluating LLM-Based Multi-Agent Systems

COSMO-RL: Towards Trustworthy LMRMs via Joint Safety and Stability

Identifying Financial Risk Information Using RAG with a Contrastive Insight

The Argument is the Explanation: Structured Argumentation for Trust in Agents

ARUQULA -- An LLM based Text2SPARQL Approach using ReAct and Knowledge Graph Exploration Utilities

Cyber Academia-Chemical Engineering (CA-ChemE): A Living Digital Town for Self-Directed Research Evolution and Emergent Scientific Discovery

MOSS-Speech: Towards True Speech-to-Speech Models Without Text Guidance

Defend LLMs Through Self-Consciousness

Metaphor identification using large language models: A comparison of RAG, prompt engineering, and fine-tuning

The Dragon Hatchling: The Missing Link between the Transformer and Models of the Brain

Building the EHR Foundation Model via Next Event Prediction

Object Detection with Multimodal Large Vision-Language Models: An In-depth Review

SIRI: Scaling Iterative Reinforcement Learning with Interleaved Compression

MGM-Omni: Scaling Omni LLMs to Personalized Long-Horizon Speech

Beyond Repetition: Text Simplification and Curriculum Learning for Data-Constrained Pretraining

Sheaf Discovery with Joint Computation Graph Pruning and Flexible Granularity

Understanding Textual Capability Degradation in Speech LLMs via Parameter Importance Analysis