大模型推理能力

Towards an Atlas of Cultural Commonsense for Machine Reasoning

Anurag Acharya, Kartik Talamadupula, Mark A Finlayson

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

Comments 9 pages, 9 figures

Incorporating Pragmatic Reasoning Communication into Emergent Language

Yipeng Kang, Tonghan Wang, Gerard de Melo

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

Comments 9 pages. Accepted as a spotlight paper to NeurIPS 2020

ExplanationLP: Abductive Reasoning for Explainable Science Question Answering

Mokanarangan Thayaparan, Marco Valentino, André Freitas

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

Scalable Multi-Hop Relational Reasoning for Knowledge-Aware Question Answering

Yanlin Feng, Xinyue Chen, Bill Yuchen Lin, Peifeng Wang, Jun Yan, Xiang Ren

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

Comments Accepted to EMNLP 2020. Project page: https://github.com/INK-USC/MHGRN

Enabling Robots to Understand Incomplete Natural Language Instructions Using Commonsense Reasoning

Haonan Chen, Hao Tan, Alan Kuntz, Mohit Bansal, Ron Alterovitz

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

Comments 7 pages, 4 figures, ICRA 2020

Contrastive Self-Supervised Learning for Commonsense Reasoning

Tassilo Klein, Moin Nabi

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

Comments To appear at ACL2020

Pre-training Is (Almost) All You Need: An Application to Commonsense Reasoning

Alexandre Tamborrino, Nicola Pellicano, Baptiste Pannier, Pascal Voitot, Louise Naudin

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

Comments Accepted at ACL 2020

Abductive Reasoning as Self-Supervision for Common Sense Question Answering

Sathyanarayanan N. Aakur, Sudeep Sarkar

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

Comments 8 Pages, 4 figures, 4 tables

Adapting Meta Knowledge Graph Information for Multi-Hop Reasoning over Few-Shot Relations

Xin Lv, Yuxian Gu, Xu Han, Lei Hou, Juanzi Li, Zhiyuan Liu

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

Probabilistic Recursive Reasoning for Multi-Agent Reinforcement Learning

Ying Wen, Yaodong Yang, Rui Luo, Jun Wang, Wei Pan

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

Comments ICLR 2019

From Machine Learning to Machine Reasoning

Leon Bottou

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

Comments 15 pages - fix broken pagination in v2

FAU at ImageCLEF 2026 Task on Multimodal Reasoning Robust Candidate Scoring and Concise Multilingual Visual Answering

FAU参加2026年ImageCLEF多模态推理任务：鲁棒候选评分与简洁多语种视觉问答

Mohamed Basem, Vincent Christlein

机构 * Friedrich-Alexander-Universität Erlangen-Nürnberg（弗里德里希-亚历山大-埃尔兰根-纽伦堡大学） 纠错

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.LG

AI总结 FAU提交的多模态推理系统，未做任务特定模型训练，在2026年ImageCLEF竞赛中，获Visual MCQ第三名、Visual OpenQA第一名，凸显推理工程的实用价值。

Comments 16 pages, 3 figures, 7 tables. CLEF 2026 Working Notes, ImageCLEF 2026 Multimodal Reasoning Task

MindClaw: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention

MindClaw: 用于精确干预的闭环具身心理状态推理

Ruoxuan Zhang, Qiaoqiao Wan, Zhengguang Wang, Chenghao Yu, Hongxia Xie, Jianlong Fu, Wen-Huang Cheng

机构 * Jilin University（吉林大学） ； Microsoft Asia（微软亚洲） ； National Taiwan University（国立台湾大学） 纠错

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.AI

AI总结 提出MindClaw框架，通过闭环具身心理状态推理实现精确干预，结合多源输入、信念记忆、认知触发技能和动作生成，在动态环境中优化干预时机。

Comments Extended version of the CVPR 2026 paper *MindPower: Enabling Theory-of-Mind Reasoning in VLM-based Embodied Agents*. This work is in progress

Position: Stop Anthropomorphizing Intermediate Tokens as Reasoning/Thinking Traces!

立场：停止将中间令牌拟人化为推理/思考痕迹！

Subbarao Kambhampati, Karthik Valmeekam, Siddhant Bhambri, Vardhan Palod, Lucas Saldyt, Kaya Stechly, Soumya Rani Samineni, Durgesh Kalwar, Upasana Biswas

机构 * University of California, Berkeley（加州大学伯克利分校） 纠错

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.AI

AI总结 本文论证将模型生成的中间令牌拟人化为“推理痕迹”或“思考痕迹”具有误导性，呼吁社区避免此类拟人化。

Comments Appears in ICML 2026. [This is a fork of v1. This fork, while overlapping with v1 in background section, differs both in the overall focus as well as the specific argument against anthropomorphization of reasoning traces]

Journal ref ICML 2026

OSS-CRS: Liberating AIxCC Cyber Reasoning Systems for Real-World Open-Source Security

OSS-CRS：解放AIxCC网络推理系统以应对现实中的开源安全

Andrew Chin, Dongkwan Kim, Yu-Fu Fu, Fabian Fleischer, Youngjoon Kim, HyungSeok Han, Cen Zhang, Brian Junekyu Lee, Hanqing Zhao, Taesoo Kim

机构 * Georgia Institute of Technology（佐治亚理工学院） ； Microsoft（微软） 纠错

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.AI

AI总结 OSS-CRS提出一个可本地部署的框架，用于运行和结合CRS技术，针对真实开源项目进行安全分析，并实现预算感知的资源管理。

Comments Version 1.1 (March 2026), OSS-CRS: an open-source framework for porting, deploying, and composing AIxCC cyber reasoning systems. Project page: https://github.com/ossf/oss-crs

Geometric Reasoning in the Embedding Space

嵌入空间中的几何推理

Jan Hůla, David Mojžíšek, Jiří Janeček, David Herel, Mikoláš Janota

机构 * University of Ostrava（奥斯特拉瓦大学） ； Czech Technical University in Prague（布拉格捷克技术大学） 纠错

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.LG

AI总结 本文提出通过图神经网络和变换器在嵌入空间中进行几何推理，证明其能有效恢复网格结构并优于变换器。

Comments published version of the article in English

Journal ref Mojzisek, D., Hula, J., Janecek, J., Herel, D., & Janota, M. (2025). Geometric Reasoning in the Embedding Space. Machine Learning and Knowledge Extraction, 7(3), 93

Monitor-Generate-Verify (MGV): Formalising Metacognitive Theory for Language Model Reasoning

Monitor-Generate-Verify (MGV): 形式化语言模型推理的元认知理论

Nick Oh, Fernand Gobet

机构 * socius labs ； Centre for Philosophy of Natural and Social Science (CPNSS)（哲学自然与社会科学中心） ； London School of Economics（伦敦经济学院） 纠错

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.AI

AI总结 MGV框架通过引入监控机制，弥补生成-验证范式中监控过程的缺失，以提升语言模型推理的准确性和鲁棒性。

Comments Presented at the Workshop on the Foundations of Reasoning in Language Models at NeurIPS 2025 (non-archival)

Large Language Models Reasoning Abilities Under Non-Ideal Conditions After RL-Fine-Tuning

Chang Tian, Matthew B. Blaschko, Mingzhe Xing, Xiuxing Li, Yinliang Yue, Marie-Francine Moens

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.AI

Comments large language models, large vision-language model, reasoning, non-ideal conditions, reinforcement learning

ReasoningRec: Bridging Personalized Recommendations and Human-Interpretable Explanations through LLM Reasoning

Millennium Bismay, Xiangjue Dong, James Caverlee

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.AI

Comments Large Language Model, Recommendation, Human-Interpretable Reasoning, Personalization

Code Prompting Elicits Conditional Reasoning Abilities in Text+Code LLMs

Haritz Puerto, Martin Tutek, Somak Aditya, Xiaodan Zhu, Iryna Gurevych

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.CL

Comments EMNLP Main 2024. Code, prompt templates, prompts, and outputs are publicly available at https://github.com/UKPLab/arxiv2024-conditional-reasoning-llms

CURIE: An Iterative Querying Approach for Reasoning About Situations

Dheeraj Rajagopal, Aman Madaan, Niket Tandon, Yiming Yang, Shrimai Prabhumoye, Abhilasha Ravichander, Peter Clark, Eduard Hovy

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.CL

Comments This paper builds upon EIGEN (arXiv:2010.11764) and proposes a general framework for situational reasoning

Non-Monotonic Reasoning and Story Comprehension

Irene-Anna Diakidoy, Antonis Kakas, Loizos Michael, Rob Miller

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.AI

Journal ref Proceedings of the 15th International Workshop on Non-Monotonic Reasoning (NMR 2014), Vienna, 1719 July, 2014

World-to-Wrist: Task-Conditioned Future Wrist Modeling for Fine-Grained Robot Manipulation

世界到手腕：面向精细机器人操作的任务条件未来手腕建模

Yuhao Pan, Haosong Peng, Zhengshen Zhang, Zhengyang Yan, Yalun Dai, Fushuo Huo, Chujie Wang, Tianyu Qi, Xiucheng Wang, Nan Cheng, Wenchao Xu

机构 * The Hong Kong University of Science and Technology（香港科技大学） ； National University of Singapore（新加坡国立大学） ； Nanyang Technological University（南洋理工大学） ； Wuhan University（武汉大学） ； Sun Yat-sen University（中山大学） ； Xidian University（西安电子科技大学） ； Southeast University（东南大学） 纠错

专题命中 其他推理 ：CoT(summary_cn,abstract)

AI总结 本研究提出W2-VLA模型，通过任务条件未来手腕建模结合W2-CoT辅助监督，在LIBERO等数据集及真实任务中提升了机器人精细接触敏感操作能力，动作生成速率超80Hz。

Why Thinking Hurts: Diagnosing and Rectifying Linguistic Inertia in Large Language Models for Recommendation

为什么思考有害：诊断并纠正大型语言模型在推荐中的语言惯性

Luankang Zhang, Yonghao Huang, Hang Lv, Xuyang Zhi, Mingjia Yin, Yuyang Ye, Wei Guo, Hao Wang, Enhong Chen

专题命中 其他推理 ：CoT(abstract,abstract_cn)；reasoning(abstract)；chain-of-thought(abstract)

AI总结 本文发现链式思维推理在推荐模型中会导致性能下降，归因于语言惯性，并提出无需训练的语言惯性校准解码（LICD）框架，通过推理链压缩和偏差相减对比推理来缓解该问题。

OpenRLHF: An Easy-to-use, Scalable and High-performance RLHF Framework

Jian Hu, Xibin Wu, Wei Shen, Jason Klein Liu, Zilin Zhu, Weixun Wang, Songlin Jiang, Haoran Wang, Hao Chen, Bin Chen, Weikai Fang, Xianyu, Yu Cao, Haotian Xu, Yiming Liu

机构 * Team Project Leader（团队项目负责人） 纠错

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

Comments update template

Do Efficient Transformers Really Save Computation?

Kai Yang, Jan Ackermann, Zhenyu He, Guhao Feng, Bohang Zhang, Yunzhen Feng, Qiwei Ye, Di He, Liwei Wang

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

Comments 20 pages, ICML 2024 Camera Ready Version

Rethinking harmless refusals when fine-tuning foundation models

Florin Pop, Judd Rosenblatt, Diogo Schwerz de Lucena, Michael Vaiana

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

Comments ICLR 2024 AGI Workshop Poster

A Study on the Calibration of In-context Learning

Hanlin Zhang, Yi-Fan Zhang, Yaodong Yu, Dhruv Madeka, Dean Foster, Eric Xing, Himabindu Lakkaraju, Sham Kakade

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

Comments NAACL 2024

Scale or Reason? A Compute-Equivalent Analysis of Reasoning Distillation

规模还是推理？推理蒸馏的计算等价分析

Nicolas Boizard, Hippolyte Gisserot-Boukhlef, Kevin El Haddad, Céline Hudelot, Pierre Colombo

机构 * Diabolocom ； Artefact Research Center ； Equall ； ISIA Lab, University of Mons（ISIA实验室，蒙斯大学） ； MICS, CentraleSupélec, Université Paris-Saclay（MICS，CentraleSupélec，巴黎萨克雷大学） 纠错

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.CL

AI总结 通过控制实验，在相同计算预算下比较推理蒸馏与标准指令微调，发现指令微调在多数配置下处于帕累托前沿，而推理蒸馏仅在7B以上开放任务中有效，混合25-50%推理数据可低成本获得大部分收益。

SEER: Long-Context Reasoning via Selective Visual-Text Compression

SEER：基于选择性视觉-文本压缩的长上下文推理

Jiawei Xu, Zhilin Zhai, Jinrui Fang, Ruohan Xu, Mingfei Lu, Yi Zhang, Guanchu Wang, Tianlong Chen, Ying Ding

机构 * The University of North Carolina at Chapel Hill（北卡罗来纳大学教堂山分校） ； University of Cambridge（剑桥大学） ； University of Technology Sydney（悉尼科技大学） ； The University of North Carolina at Charlotte（北卡罗来纳大学夏洛特分校） ； The University of Texas at Austin（德克萨斯大学奥斯汀分校） 纠错

专题命中 其他推理 ：reasoning(title,abstract)；分类 cs.CL

AI总结 SEER是结合视觉压缩效率与文本推理精度的框架，经监督微调后在LongBench等长上下文基准测试中，准确率优于Glyph-9B、Qwen3-8B等基线模型，可提升提取精度并保留提示token节省量。

Comments COLM 2026, Third Conference on Language Modeling