语言大模型 / LLM - arXivDaily 专题

1905.10949 2019-05-28 cs.LG cs.CL stat.ML 62%

QuesNet: A Unified Representation for Heterogeneous Test Questions

Yu Yin, Qi Liu, Zhenya Huang, Enhong Chen, Wei Tong, Shijin Wang, Yu Su

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

1902.09969 2019-02-27 cs.CV cs.CL cs.LG 62%

Using Deep Object Features for Image Descriptions

Ashutosh Mishra, Marcus Liwicki

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

Comments arXiv admin note: text overlap with arXiv:1411.2539, arXiv:1609.06647 by other authors

1811.11041 2018-11-28 cs.CL cs.AI math.CT 62%

Translating and Evolving: Towards a Model of Language Change in DisCoCat

Tai-Danae Bradley, Martha Lewis, Jade Master, Brad Theilman

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

Comments In Proceedings CAPNS 2018, arXiv:1811.02701

Journal ref EPTCS 283, 2018, pp. 50-61

1709.00028 2018-09-11 cs.CL cs.LG 62%

Glyph-aware Embedding of Chinese Characters

Falcon Z. Dai, Zheng Cai

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

Comments Workshop on Subword and Character level models in NLP at EMNLP 2017. Source code available

1808.03793 2018-08-14 cs.IR cs.CL cs.LG 62%

Document Informed Neural Autoregressive Topic Models

Pankaj Gupta, Florian Buettner, Hinrich Schütze

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

1807.11582 2018-08-07 cs.CL cs.LG stat.ML 62%

A Hierarchical Approach to Neural Context-Aware Modeling

Patrick Huber, Jan Niehues, Alex Waibel

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

Comments 8 pages, 2 figures, 1 table

1805.00314 2018-05-02 cs.CV cs.AI cs.CL 62%

Object Counts! Bringing Explicit Detections Back into Image Captioning

Josiah Wang, Pranava Madhyastha, Lucia Specia

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

Comments Please cite: In Proceedings of 2018 Conference of the North American Chapter of the Association for Computational Linguistics (NAACL 2018)

1704.04664 2018-03-12 cs.AI cs.CL cs.RO 62%

Online Spatial Concept and Lexical Acquisition with Simultaneous Localization and Mapping

Akira Taniguchi, Yoshinobu Hagiwara, Tadahiro Taniguchi, Tetsunari Inamura

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

Comments This paper was accepted in the 2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2017)

1701.04313 2018-02-14 cs.CL cs.IR cs.LG cs.NE 62%

End-to-End ASR-free Keyword Search from Speech

Kartik Audhkhasi, Andrew Rosenberg, Abhinav Sethy, Bhuvana Ramabhadran, Brian Kingsbury

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

Comments Published in the IEEE 2017 International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2017), scheduled for 5-9 March 2017 in New Orleans, Louisiana, USA

1706.02496 2017-06-09 stat.ML cs.CL cs.LG 62%

Context encoders as a simple but powerful extension of word2vec

Franziska Horn

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

Comments ACL 2017 2nd Workshop on Representation Learning for NLP

1511.06349 2017-03-01 cs.LG cs.CL 62%

Generating Sentences from a Continuous Space

Samuel R. Bowman, Luke Vilnis, Oriol Vinyals, Andrew M. Dai, Rafal Jozefowicz, Samy Bengio

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

Comments First two authors contributed equally. Work was done when all authors were at Google, Inc

Journal ref SIGNLL Conference on Computational Natural Language Learning (CONLL), 2016

1701.00851 2017-01-05 cs.CL cs.LG 62%

Unsupervised neural and Bayesian models for zero-resource speech processing

Herman Kamper

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

Comments PhD thesis, University of Edinburgh, 107 pages, submitted and accepted 2016

1612.03769 2016-12-14 cs.CL cs.AI 62%

Context-aware Sentiment Word Identification: sentiword2vec

Yushi Yao, Guangjian Li

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

Comments 15 pages

1602.08952 2016-06-09 cs.CL cs.LG 62%

Representation of linguistic form and function in recurrent neural networks

Ákos Kádár, Grzegorz Chrupała, Afra Alishahi

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

1504.06665 2015-04-29 cs.CL cs.AI 62%

Using Syntax-Based Machine Translation to Parse English into Abstract Meaning Representation

Michael Pust, Ulf Hermjakob, Kevin Knight, Daniel Marcu, Jonathan May

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

Comments 10 pages, 8 figures

1206.6423 2012-07-02 cs.CL cs.LG cs.RO 62%

A Joint Model of Language and Perception for Grounded Attribute Learning

Cynthia Matuszek, Nicholas FitzGerald, Luke Zettlemoyer, Liefeng Bo, Dieter Fox

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

Comments Appears in Proceedings of the 29th International Conference on Machine Learning (ICML 2012)

1102.5185 2011-02-28 cs.CL cs.AI 62%

Universal Higher Order Grammar

Victor Gluzberg

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

Comments 48 pages

2606.31208 2026-07-01 cs.LG cs.CR 新提交 61%

Probing Memorization of Tabular In-Context Learning

探测表格上下文学习中的记忆化

Francesco Capano, Jonas Böhler

机构 * SAP SE（SAP公司）

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

AI总结提出ICLMEM框架探测表格基础模型在上下文学习中的参数记忆化，发现低基数任务中存在中等记忆信号，但实际训练条件下信号消失。

Comments Accepted at 2nd ICML Workshop on Foundation Models for Structured Data, 2026

2605.31562 2026-06-01 cs.LG 61%

Effective Biological Representation Learning by Masking Gene Expression

通过掩码基因表达实现有效的生物表示学习

Kian Kenyon-Dean, Alina Selega, Ihab Bendidi, Jordan M. Sorokin, Luca Bertinetto, David Errington, Hayley Donnella, Oren Kraus

机构 * Recursion ； Valence Labs ； École Normale Supérieure PSL

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

AI总结提出自监督模型TxFM，采用掩码自编码方法处理RNA-seq数据，通过消融研究确定关键架构，并在精心策划的DiverseRNA-1.4M数据集上训练，获得优于大规模基础模型的基因表示。

Comments 31 pages, 11 figures. Preprint; presented at ICLR 2026 2nd Workshop on Foundation Models for Science: Real-World Impact and Science-First Design

2512.24842 2026-01-01 cs.CL stat.ML 61%

Triangulation as an Acceptance Rule for Multilingual Mechanistic Interpretability

三角化作为多语言机制可解释性的接受规则

Yanan Long

机构 * StickFlux Labs（StickFlux实验室）

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

AI总结本文提出三角化作为多语言机制可解释性的接受规则，通过因果标准验证模型行为的必要性、充分性和不变性，以提高跨语言解释的可靠性。

Comments NeurIPS 2025 Workshop Evaluating the Evolving LLM Lifecycle: Benchmarks, Emergent Abilities, and Scaling

2510.26777 2025-10-31 cs.LG 61%

Pre-trained Forecasting Models: Strong Zero-Shot Feature Extractors for Time Series Classification

Andreas Auer, Daniel Klotz, Sebastinan Böck, Sepp Hochreiter

机构 * NXAI GmbH（NXAI公司）； ELLIS Unit, LIT AI Lab, Institute for Machine Learning, JKU Linz（ELLIS单位、LIT AI实验室、机器学习研究所）； Interdisciplinary Transformation University Austria（跨学科转型大学奥地利）

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

Comments NeurIPS 2025 Workshop on Recent Advances in Time Series Foundation Models (BERT2S)

2501.04234 2025-01-09 stat.ML cs.LG stat.AP 61%

Statistical Uncertainty Quantification for Aggregate Performance Metrics in Machine Learning Benchmarks

Rachel Longjohn, Giri Gopalan, Emily Casleton

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

Comments LA-UR-24-25289; presented at the Workshop on Statistical Frontiers in LLMs and Foundation Models at NeurIPS 2024

2407.10683 2024-07-16 cs.CV cs.AI 61%

Addressing Image Hallucination in Text-to-Image Generation through Factual Image Retrieval

Youngsun Lim, Hyunjung Shim

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

Comments This paper has been accepted for oral presentation at the IJCAI 2024 Workshop on Trustworthy Interactive Decision-Making with Foundation Models

2608.22786 2026-08-25 cs.LG 新提交 57%

ReCoG: Reciprocal Co-Evolution for Multimodal Graph Learning

ReCoG：面向多模态图学习的互惠协同演化

Rui Xue, Tianfu Wu

机构 * North Carolina State University（北卡罗来纳州立大学）

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

AI总结 ReCoG是将图结构学习与多模态表示学习紧密耦合的新范式，在节点分类和链接预测基准中优于多模态图结构学习基线，证明了结构与语义协同演化的重要性。

2608.22358 2026-08-25 cs.LG physics.geo-ph 新提交 57%

Tracing the Unlabeled Storm: Cross-Variable Transfer in a Lagrangian Atmospheric JEPA Framework

追踪未标记的风暴：拉格朗日大气JEPA框架中的跨变量迁移

K M Anirudh, S Sandeep, Hariprasad Kodamana

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

AI总结该研究提出跨变量代理学习方法，基于M-JEPA在无降水监督下预训练，实现了优于ECMWF集合的季风降水预测，为大气表示迁移提供诊断框架。

Comments 8 pages, 3 figures, plus supplementary material

2608.05154 2026-08-25 cs.CL cs.CV 版本更新 57%

RIG-RoPE: Relation-Stratified Multimodal Attention with Instance-Local Rotary Geometry and Representation-Aware Traversal Coordinates

RIG-RoPE：具有时长感知时间坐标的关系与实例门控旋转位置编码

Donggen Li

机构 * Sichuan University（四川大学）

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

AI总结本研究针对多模态场景中静态多维位置分配的局限，提出RIG-RoPE机制，通过关系与实例门控及时长感知时间坐标优化位置编码，未增加可学习参数，确立了相关公式与验证路径。

Comments 26 pages, 2 figures, 4 tables. This revision adds matched, inference-only Qwen2-VL-2B native/RIG L1 mechanism evidence: exact text-only equivalence, exact RIG Gauge invariance, native Gauge sensitivity, and retained same-instance spatial effects. Task-level Qwen benchmarks, training gains, stable task improvement, universality, and empirical superiority remain unestablished

2605.23262 2026-08-25 cs.AI 版本更新 57%

Designing Benchmarks for Knowledge Work

知识工作的设计与报告基准

Yining Hua, Hongbin Na, Cyrus Ayubcha, Levi Lian

机构 * Harvard University（哈佛大学）； University of Technology Sydney（悉尼科技大学）； Stanford University（斯坦福大学）； Raycaster AI

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

AI总结针对当前知识工作评估基准与真实部署脱节的问题，提出三步法（定义工作活动、指定测试设置、评分工作产品）来明确基准任务与工作主张的对应关系，并通过三个案例分析展示设计选择如何影响可支持的工作主张。

Comments 18 pages. This replacement updates the title and revises the contribution from a three-step reporting approach to a four-field work-centered benchmark representation (represented activity, tested setting, required work product, and evaluated result). Author affiliations now include Agent Evaluation Science, Inc., New York, NY 10001

2602.14681 2026-08-25 cs.MA cs.AI 版本更新 57%

ST-EVO: Towards Generative Spatio-Temporal Evolution of Multi-Agent Communication Topologies

ST-EVO：迈向多智能体通信拓扑的生成时空演化

Xingjian Wu, Xvyuan Liu, Junkai Lu, Siyuan Wang, Xiangfei Qiu, Yang Shu, Jilin Hu, Chenjuan Guo, Bin Yang

专题命中知识编辑与模型理解：LLM(abstract)；分类 cs.AI

AI总结 ST-EVO从时空视角提出多智能体通信拓扑的生成方法，通过紧凑调度器实现对话层面的通信调度，并具备自我反馈能力，实验显示其在多个基准测试中性能优越。

2608.19863 2026-08-21 eess.AS cs.AI cs.SD 新提交 57%

Listening Forward: Next Patch Embedding Prediction Enables Scalable Audio Learners

向前听：下一个补丁嵌入预测可实现可扩展音频学习器

Umberto Cappellazzo, Xubo Liu, Stavros Petridis, Maja Pantic

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

AI总结该研究提出自监督框架NAPE，通过因果Transformer预测下一个音频补丁嵌入，在六个音频语音基准上实现先进性能，可扩展编码器规模，获强线性探测结果，无显式监督时生成结构化注意力模式。

Comments Project website: https://umbertocappellazzo.github.io/nape

2604.03634 2026-08-21 cs.LG cs.IT eess.SP math.IT 57%

Algebraic Diversity: Group-Theoretic Spectral Estimation from Single Observations

代数多样性：从单次观测进行群论谱估计

Mitchell A. Thornton

机构 * Richardson, TX 75080 USA（美国德克萨斯州里奇蒙德市75080号）

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

AI总结本文通过群论方法揭示了单次观测下的谱估计问题，证明了时间平均是退化群作用的特例，并展示了群平均估计与多快门协方差估计的等效性，同时统一了DFT、DCT和KLT等变换。

Comments 41 pages, 14 figures. v3: Retracted six findings in Section 11, transformer application, due to error in spectral concentration metric. Corrected results deferred to separate publication. Remark added after Conjecture 23 on orbit-structure bias in psi criterion. v4: new result blind group matching; v5: updated metrics; v6: New Theorem 7 on MLE, added Nitzberg citation; v7:"group gain" update

1. 知识编辑与模型理解 7583 篇

QuesNet: A Unified Representation for Heterogeneous Test Questions

Using Deep Object Features for Image Descriptions

Translating and Evolving: Towards a Model of Language Change in DisCoCat

Glyph-aware Embedding of Chinese Characters

Document Informed Neural Autoregressive Topic Models

A Hierarchical Approach to Neural Context-Aware Modeling

Object Counts! Bringing Explicit Detections Back into Image Captioning

Online Spatial Concept and Lexical Acquisition with Simultaneous Localization and Mapping

End-to-End ASR-free Keyword Search from Speech

Context encoders as a simple but powerful extension of word2vec

Generating Sentences from a Continuous Space

Unsupervised neural and Bayesian models for zero-resource speech processing

Context-aware Sentiment Word Identification: sentiword2vec

Representation of linguistic form and function in recurrent neural networks

Using Syntax-Based Machine Translation to Parse English into Abstract Meaning Representation

A Joint Model of Language and Perception for Grounded Attribute Learning

Universal Higher Order Grammar

Probing Memorization of Tabular In-Context Learning

Effective Biological Representation Learning by Masking Gene Expression

Triangulation as an Acceptance Rule for Multilingual Mechanistic Interpretability

Pre-trained Forecasting Models: Strong Zero-Shot Feature Extractors for Time Series Classification

Statistical Uncertainty Quantification for Aggregate Performance Metrics in Machine Learning Benchmarks

Addressing Image Hallucination in Text-to-Image Generation through Factual Image Retrieval

ReCoG: Reciprocal Co-Evolution for Multimodal Graph Learning

Tracing the Unlabeled Storm: Cross-Variable Transfer in a Lagrangian Atmospheric JEPA Framework

RIG-RoPE: Relation-Stratified Multimodal Attention with Instance-Local Rotary Geometry and Representation-Aware Traversal Coordinates

Designing Benchmarks for Knowledge Work

ST-EVO: Towards Generative Spatio-Temporal Evolution of Multi-Agent Communication Topologies

Listening Forward: Next Patch Embedding Prediction Enables Scalable Audio Learners

Algebraic Diversity: Group-Theoretic Spectral Estimation from Single Observations