语言大模型 / LLM

Feature Adaptation of Pre-Trained Language Models across Languages and Domains with Robust Self-Training

Hai Ye, Qingyu Tan, Ruidan He, Juntao Li, Hwee Tou Ng, Lidong Bing

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

Comments To appear at EMNLP 2020. 14 pages. Code is available at: https://github.com/oceanypt/CFd

Predicting metrical patterns in Spanish poetry with language models

Javier de la Rosa, Salvador Ros, Elena González-Blanco

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

Comments LXAI Workshop @ NeurIPS 2020

Reducing Non-Normative Text Generation from Language Models

Xiangyu Peng, Siyan Li, Spencer Frazier, Mark Riedl

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

StyleDGPT: Stylized Response Generation with Pre-trained Language Models

Ze Yang, Wei Wu, Can Xu, Xinnian Liang, Jiaqi Bai, Liran Wang, Wei Wang, Zhoujun Li

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

Comments Findings of EMNLP2020

Retrofitting Structure-aware Transformer Language Model for End Tasks

Hao Fei, Yafeng Ren, Donghong Ji

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

Comments Accepted as long paper in EMNLP2020 main proceeding

Attention Flows: Analyzing and Comparing Attention Mechanisms in Language Models

Joseph F DeRose, Jiayao Wang, Matthew Berger

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

Comments 11 pages, 12 figures, to be published in IEEE Transactions on Visualization and Computer Graphics

El Departamento de Nosotros: How Machine Translated Corpora Affects Language Models in MRC Tasks

Maria Khvalchik, Mikhail Galkin

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

Differentiable Language Model Adversarial Attacks on Categorical Sequence Classifiers

I. Fursov, A. Zaytsev, N. Kluchnikov, A. Kravchenko, E. Burnaev

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

Adversarial Training for Large Neural Language Models

Xiaodong Liu, Hao Cheng, Pengcheng He, Weizhu Chen, Yu Wang, Hoifung Poon, Jianfeng Gao

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

Comments 13 pages, 9 tables, 2 figures

Deepening Hidden Representations from Pre-trained Language Models

Junjie Yang, Hai Zhao

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

FineText: Text Classification via Attention-based Language Model Fine-tuning

Yunzhe Tao, Saurabh Gupta, Satyapriya Krishna, Xiong Zhou, Orchid Majumder, Vineet Khare

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

The merits of Universal Language Model Fine-tuning for Small Datasets -- a case with Dutch book reviews

Benjamin van der Burgh, Suzan Verberne

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

Comments 5 pages, 2 figures

Fine-tuning Pre-Trained Transformer Language Models to Distantly Supervised Relation Extraction

Christoph Alt, Marc Hübner, Leonhard Hennig

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

Comments To appear in Proceedings of ACL 2019 (11 pages). arXiv admin note: text overlap with arXiv:1906.03088

Fine-tuning of Language Models with Discriminator

Vadim Popov, Mikhail Kudinov

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

Practical Text Classification With Large Pre-Trained Language Models

Neel Kant, Raul Puri, Nikolai Yakovenko, Bryan Catanzaro

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

Comments 8 pages, submitted to AAAI 2019

Infusing fine-grained visual knowledge to Vision-Language Models

Nikolaos-Antonios Ypsilantis, Kaifeng Chen, André Araujo, Ondřej Chum

机构 * VRG, FEE, Czech Technical University in Prague（捷克布拉格技术大学） 纠错

专题命中 指令微调 ：language model(title,abstract)；foundation model(comments)

Comments ICCVW 2025 accepted paper. Workshop name: "What is Next in Multimodal Foundation Models?"

Dynamic Gradient Sparsification Training for Few-Shot Fine-tuning of CT Lymph Node Segmentation Foundation Model

Zihao Luo, Zijun Gao, Wenjun Liao, Shichuan Zhang, Guotai Wang, Xiangde Luo

专题命中 指令微调 ：foundation model(title,abstract)

Comments 10 pages, 3 figures, 2 tables, and the lymph node segmentation foundation model code and pretrained model are available

CaLoRAify: Calorie Estimation with Visual-Text Pairing and LoRA-Driven Visual Language Models

Dongyu Yao, Keling Yao, Junhong Zhou, Yinghao Zhang

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

Comments Disclaimer: This work is part of a course project and reflects ongoing exploration in the field of vision-language models and calorie estimation. Findings and conclusions are subject to further validation and refinement

Generative Visual Instruction Tuning

Jefferson Hernandez, Ruben Villegas, Vicente Ordonez

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

Comments Add more results using task tokens, expand the introduction and related work FIX: error in LLM-as-judge evaluation that was over-inflating the results

Can Visual Foundation Models Achieve Long-term Point Tracking?

Görkay Aydemir, Weidi Xie, Fatma Güney

专题命中 指令微调 ：foundation model(title,abstract)

Comments ECCV 2024 - Emergent Visual Abilities and Limits of Foundation Models (EVAL-FoMo) Workshop

Rethinking Semi-supervised Learning with Language Models

Zhengxiang Shi, Francesco Tonolini, Nikolaos Aletras, Emine Yilmaz, Gabriella Kazai, Yunlong Jiao

专题命中 指令微调 ：language model(title)；分类 cs.CL、cs.AI、cs.LG；pretraining(comments)

Comments Findings of ACL 2023. Code is available at https://github.com/amzn/pretraining-or-self-training

Knowledge-Graph-Gated Defactualization for Style-Controllable and Fact-Preserving Generation in Agentic Conversational AI

面向智能体对话AI的风格可控且事实保留生成的知识图谱门控去事实化方法

Tanmay Kumar Shrivastava, Darsh Rohit Nandu, Rajesh Kumar Mundotiya

机构 * Indian Institute of Technology (IIT) Bhilai（印度比莱印度理工学院） 纠错

专题命中 指令微调 ：LLM(abstract_cn)；large language model(abstract)；language model(abstract)；分类 cs.CL、cs.AI

AI总结 针对智能体对话AI的事实保留与风格可控生成需求，提出DSR知识工程框架，结合KG与激活引导，在LLaMA系列模型上验证其可提升实体恢复率并保持风格控制，无需微调。

Task Specialization Fine-Tuning for Contextual Reinforcement Learning

上下文强化学习的任务专业化微调

Jianan Zhou, Jung-Hoon Cho, Tianyue Zhou, Han Zheng, Jie Zhang, Roy Dong, Yining Ma, Cathy Wu

机构 * Nanyang Technological University（南洋理工大学） ； MIT（麻省理工学院） ； UIUC（伊利诺伊大学厄巴纳-香槟分校） 纠错

专题命中 指令微调 ：LLM(abstract,abstract_cn)；pretraining(abstract)；分类 cs.AI、cs.LG

AI总结 针对上下文强化学习的任务专业化微调难题，提出TSFT框架，通过整数线性规划分配微调预算，在多领域实验中提升了任务覆盖性能。

Palmyra x6 Technical Report: An Agentic, Tool-Use Model Post-Trained via Anchored Supervised Fine-Tuning

Palmyra x6技术报告：一种通过锚定监督微调进行后训练的具工具使用能力的智能体模型

Peng Du, Kiran Kamble, Rakshith Vasudev, Zhizhuo Yang, Rohith Nadimpally, Arjun Krishna, Waseem Alshikh, Daniel M. Bikel

机构 * Writer AI Research, Writer, Inc.（Writer AI研究院，Writer公司） 纠错

专题命中 指令微调 ：large language model(abstract)；language model(abstract)；post-training(abstract)；分类 cs.CL、cs.AI

AI总结 Palmyra x6是一款针对企业级智能体任务优化的大语言模型，通过锚定监督微调后训练构建，在公开基准测试及偏见安全评估中表现优异。

Comments 12 pages

CASA: Classification Augmented with Safety Attention for Robust Multimodal Alignment

通过条件解码实现鲁棒的多模态安全性

Anurag Kumar, Raghuveer Peri, Jon Burnsky, Alexandru Nelus, Rohit Paturi, Srikanth Vishnubhotla, Yanjun Qi

机构 * The Ohio State University（俄亥俄州立大学） ； AWS（亚马逊云服务） 纠错

专题命中 指令微调 ：LLM(abstract,abstract_cn)；language model(abstract)；分类 cs.AI、cs.LG

AI总结 本文提出CASA方法，通过内部表示预测安全令牌以提升多模态大语言模型的安全性，实验显示其在多种基准上显著降低攻击成功率，同时保持良性输入的实用性。

Comments 9 pages + Appendix section

Latent-LoRA: Compact Latent-Space Adapters with Gradient-Free Routing for Continual Learning

Latent-LoRA：用于持续学习的具有无梯度路由的紧凑潜在空间适配器

Reza Rahimi Azghan, Gautham Krishna Gudur, Giulia Pedrielli, Pavan Turaga, Hassan Ghasemzadeh

机构 * College of Health Solutions, Arizona State University（亚利桑那州立大学健康解决方案学院） ； Department of Electrical and Computer Engineering, The University of Texas at Austin（德克萨斯大学奥斯汀分校电气与计算机工程系） ； School of Computing and Augmented Intelligence, Arizona State University（亚利桑那州立大学计算与增强智能学院） ； The GAME School, Arizona State University（亚利桑那州立大学GAME学院） 纠错

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

AI总结 研究针对大语言模型持续学习的灾难性遗忘问题，提出Latent-LoRA系统。利用冻结嵌入层池化令牌嵌入分离任务分布，无需可训练路由组件，通过奇异值分解和正交正则化控制任务干扰，实验显示其有近零遗忘的最优性能。

AgentOmnia: Scaling Agentic Models for Full-Scenario Applications

AgentOmnia：用于全场景应用的规模化智能体模型

Hao Jiang, Gangtao Xin, Yingdi Huang, Guojie Zhu, Jiangshan Zhang, Xinyuan Lin, Yunkun Xu, Chengyu Shen, Wenlong Fei, Jiawei Li, Yujie Fu, Sichen Kang, Tingyu Xie, Yedi Hu, Jingren Zhang, Hongcheng Gao, Jianshu Zeng, Chong Chen, Chang Guo, Chao Feng, Feng Wang, Fulin Lin, Jinchao Ma, Lang Mei, Li Huang, Liyan Liu, Qing He, Shuting Tao, Siyu Mo, Xiangnan Chen, Xiaohan Yu, Xiaoyang Li, Yanheng Hou, Yanyu Wu, Zhihan Yang, Wentao Zhang, Yang Gao, Zhao Cao

专题命中 指令微调 ：large language model(abstract)；language model(abstract)；post-training(abstract)；分类 cs.CL、cs.AI

AI总结 研究大型语言模型智能体全场景扩展问题，提出AgentOmnia框架，结合多种技术构建环境、工具和任务，通过多种方式支持训练后处理，能将多个基准测试通过率和宏平均率大幅提高，实现广泛改进并为自我进化提供证据。

Comments 69 pages, 18 figures, 13 tables

The Blessing of Dimensionality: How Near-Orthogonality in High-Dimensional Spaces Explains Temporal Portability

维度的祝福：高维空间中的近正交性如何解释时间可移植性

Abigail Woodring, Adrian Chan, Rana Muhammad Shahroz Khan, Sukwon Yun, Chau-Wai Wong, Tianlong Chen

机构 * NC State University（北卡罗来纳州立大学） ； University of North Carolina at Chapel Hill（北卡罗来纳大学教堂山分校） 纠错

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

AI总结 研究探讨PortLLM在持续预训练中LoRA补丁的长期时间可移植性及有效性。通过对Mistral、Gemma和Qwen基础模型进行实证研究，并提供理论分析，发现其可移植性持久，高维向量近正交性是关键，还展示了损失景观几何视角。

Rooted Absorbed Prefix Trajectory Balance with Submodular Replay for GFlowNet Training

基于子模重放的根吸收前缀轨迹平衡用于GFlowNet训练

Xi Wang, Wenbo Lu, Shengjie Wang

机构 * Courant Institute School of Mathematics, Computing, and Data Science, New York University（纽约大学Courant研究所数学、计算与数据科学学院） ； Courant Institute School of Mathematics, Computing（纽约大学Courant研究所数学、计算与数据科学学院） ； Data Science, New York University（纽约大学数据科学学院） 纠错

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

AI总结 针对GFlowNet的模式坍塌问题，提出RapTB目标函数（通过根锚定子轨迹监督和吸收后缀备份提供密集前缀学习信号）和SubM子模重放策略（促进高奖励和多样性），在分子生成等任务中提升优化性能和多样性。

Journal ref Forty-third International Conference on Machine Learning (ICML 2026)

PASs-MoE: Mitigating Misaligned Co-drift among Router and Experts via Pathway Activation Subspaces for Continual Learning

PASs-MoE：通过路径激活子空间减轻路由器与专家之间的错位协同漂移以进行持续学习

Zhiyan Hou, Haiyun Guo, Haokai Ma, Yandu Sun, Yonghui Yang, Jinqiao Wang

机构 * Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所） ； University of Chinese Academy of Sciences（中国科学院大学） ； National University of Singapore（新加坡国立大学） ； Southeast University, Nanjing, China（南京东南大学） ； Wuhan AI Research, Wuhan, China（武汉人工智能研究院） 纠错

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

AI总结 研究持续指令调整中多模态大语言模型的问题，提出基于路径激活子空间的固定容量PASs - MoE - LoRA方法，含PAS引导的重新加权和PAS感知的秩稳定，实验表明该方法在准确性和抗遗忘性上优于基线和变体且不增参数。

Comments Published in the Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), Volume 1: Long Papers. 14 pages. Code is available at https://github.com/yueluoshuangtian/PASs-MoE

Journal ref Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 31959--31972, San Diego, California, United States, July 2026. Association for Computational Linguistics