大模型推理能力

TrajSelector: Harnessing Latent Representations for Efficient and Effective Best-of-N in Large Reasoning Model

Bin Yu, Xinming Wang, Shijie Lian, Haotian Li, Changti Wu, Ruina Hu, Bailing Wang, Yuliang Wei, Kai Chen

机构 * Harbin Institute of Technology（哈尔滨工业大学） ； Zhongguancun Academy（中关村学院） ； Zhongguancun Institute of Artificial Intelligence（中关村人工智能研究院） ； Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所） ； Huazhong University of Science and Technology（华中科技大学） ； East China Normal University（东华大学） 纠错

专题命中 测试时计算 ：reasoning(title,abstract)；verifier(abstract)；分类 cs.CL

Comments 13 pages, 6 figures. Project website: https://zgca-ai4edu.github.io/TrajSelector

CAPO: Towards Enhancing LLM Reasoning through Generative Credit Assignment

Guofu Xie, Yunsheng Shi, Hongtao Tian, Ting Yao, Xiao Zhang

机构 * Gaoling School of Artificial Intelligence, Renmin University of China（人工智能学院，中国人民大学） ； Tencent Inc（腾讯公司） 纠错

专题命中 测试时计算 ：reasoning(title,abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Work in progress

CorrSteer: Generation-Time LLM Steering via Correlated Sparse Autoencoder Features

Seonglae Cho, Zekun Wu, Adriano Koshiyama

机构 * University College London（伦敦大学学院） 纠错

专题命中 测试时计算 ：reasoning(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 42 pages, 9 tables

Mapping from Meaning: Addressing the Miscalibration of Prompt-Sensitive Language Models

Kyle Cox, Jiawei Xu, Yikun Han, Rong Xu, Tianhao Li, Chi-Yang Hsu, Tianlong Chen, Walter Gerych, Ying Ding

专题命中 测试时计算 ：reasoning(abstract)；分类 cs.CL、cs.LG

Journal ref Proceedings of the AAAI Conference on Artificial Intelligence. 39, 22 (Apr. 2025), 23696-23703

Every Rollout Counts: Optimal Resource Allocation for Efficient Test-Time Scaling

Xinglin Wang, Yiwei Li, Shaoxiong Feng, Peiwen Yuan, Yueqi Zhang, Jiayi Shi, Chuyi Tan, Boyuan Pan, Yao Hu, Kan Li

机构 * School of Computer Science, Beijing Institute of Technology（计算机学院，北京理工大学） ； Xiaohongshu Inc（小红书公司） 纠错

专题命中 测试时计算 ：reasoning(abstract)；分类 cs.AI、cs.LG

Comments Accepted at NeurIPS2025

Seeing but Not Believing: Probing the Disconnect Between Visual Attention and Answer Correctness in VLMs

Zhining Liu, Ziyi Chen, Hui Liu, Chen Luo, Xianfeng Tang, Suhang Wang, Joy Zeng, Zhenwei Dai, Zhan Shi, Tianxin Wei, Benoit Dumoulin, Hanghang Tong

机构 * University of Illinois Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校） ； Amazon（亚马逊） ； Penn State University（宾夕法尼亚州立大学） 纠错

专题命中 测试时计算 ：reasoning(abstract)；分类 cs.AI

Comments 21 pages, 10 figures, 6 tables

Lost at the Beginning of Reasoning

Baohao Liao, Xinyi Chen, Sara Rajaee, Yuhui Xu, Christian Herold, Anders Søgaard, Maarten de Rijke, Christof Monz

机构 * University of Amsterdam（阿姆斯特丹大学） ； eBay Inc.（eBay公司） ； Salesforce AI Research（Salesforce人工智能研究） ； Copenhagen University（哥本哈根大学） 纠错

专题命中 复杂问题求解 ：reasoning(title,abstract)；chain-of-thought(abstract)；CoT(abstract)；self-correction(abstract)

Comments remove the benchmark part. (10 pages, 6 figures, 5 tables)

Distractor Injection Attacks on Large Reasoning Models: Characterization and Defense

Zhehao Zhang, Weijie Xu, Shixian Cui, Chandan K. Reddy

机构 * Amazon（亚马逊） 纠错

专题命中 复杂问题求解 ：reasoning(title,abstract)；chain-of-thought(abstract)；CoT(abstract)；分类 cs.AI

Comments 29 pages, 9 tables, 4 figures

Reasoning Distillation and Structural Alignment for Improved Code Generation

Amir Jalilifard, Anderson de Rezende Rocha, Marcos Medeiros Raimundo

机构 * Universidade Estadual de Campinas（坎皮纳斯州立大学） 纠错

专题命中 复杂问题求解 ：reasoning(title,abstract)；分类 cs.CL、cs.AI、cs.LG

Integrating Performance Tools in Model Reasoning for GPU Kernel Optimization

Daniel Nichols, Konstantinos Parasyris, Charles Jekel, Abhinav Bhatele, Harshitha Menon

专题命中 复杂问题求解 ：reasoning(title,abstract)；chain-of-thought(abstract)

From Sequence to Structure: Uncovering Substructure Reasoning in Transformers

Xinnan Dai, Kai Yang, Jay Revolinsky, Kai Guo, Aoran Wang, Bohang Zhang, Jiliang Tang

机构 * Michigan State University（密歇根州立大学） ； Peking University（北京大学） ； University of Luxembourg（卢森堡大学） 纠错

专题命中 复杂问题求解 ：reasoning(title,abstract)；分类 cs.CL、cs.AI

Comments Camera Ready version for Neurips 2025

Towards Interpretable and Trustworthy Time Series Reasoning: A BlueSky Vision

Kanghui Ning, Zijie Pan, Yushan Jiang, Anderson Schneider, Yuriy Nevmyvaka, Dongjin Song

机构 * School of Computing University of Connecticut Storrs, CT（计算学院 美国康涅狄格大学 斯托尔斯分校） ； Department of Machine Learning Research Morgan Stanley New York, NY（机器学习研究部 花旗集团 新 York） 纠错

专题命中 复杂问题求解 ：reasoning(title,abstract)；分类 cs.LG

Thinking Out Loud: Do Reasoning Models Know When They're Right?

Qingcheng Zeng, Weihao Xuan, Leyang Cui, Rob Voigt

机构 * Northwestern University（西北大学） ； The University of Tokyo（东京大学） ； RIKEN AIP（理化学研究所AIP） ； Westlake University（西湖大学） ； University of California, Davis（加州大学戴维斯分校） 纠错

专题命中 复杂问题求解 ：reasoning(title,abstract)；分类 cs.CL

Comments EMNLP 2025

Enhancing Compositional Reasoning in CLIP via Reconstruction and Alignment of Text Descriptions

Jihoon Kwon, Kyle Min, Jy-yong Sohn

机构 * Seoul National University（首尔国立大学） ； Oracle（Oracle公司） ； Yonsei University（延世大学） ； Intel Labs（英特尔实验室） 纠错

专题命中 复杂问题求解 ：reasoning(title,abstract)；分类 cs.AI

Comments Accepted at NeurIPS 2025 (poster). This is the camera-ready version

LANPO: Bootstrapping Language and Numerical Feedback for Reinforcement Learning in LLMs

Ang Li, Yifei Wang, Zhihang Yuan, Stefanie Jegelka, Yisen Wang

机构 * MIT（麻省理工学院） 纠错

专题命中 复杂问题求解 ：reasoning(abstract)；self-correction(abstract)；分类 cs.AI、cs.LG

Evaluating Prompting Strategies and Large Language Models in Systematic Literature Review Screening: Relevance and Task-Stage Classification

Binglan Han, Anuradha Mathrani, Teo Susnjak

机构 * School of Mathematical and Computational Sciences（数学与计算科学学院） 纠错

专题命中 复杂问题求解 ：chain-of-thought(abstract)；CoT(abstract)；分类 cs.CL、cs.AI

PrediQL: Automated Testing of GraphQL APIs with LLMs

Shaolun Liu, Sina Marefat, Omar Tsai, Yu Chen, Zecheng Deng, Jia Wang, Mohammad A. Tayebi

专题命中 复杂问题求解 ：reasoning(abstract)；self-correction(abstract)

Comments 8 pages, two columns

Agentic Design of Compositional Machines

Wenqian Zhang, Weiyang Liu, Zhen Liu

机构 * The Chinese University of Hong Kong (Shenzhen)（香港中文大学（深圳）） 纠错

专题命中 复杂问题求解 ：reasoning(abstract)；分类 cs.CL、cs.AI、cs.LG

Comments 75 pages, 31 figures, Project Page: https://besiegefield.github.io

Urban-R1: Reinforced MLLMs Mitigate Geospatial Biases for Urban General Intelligence

Qiongyan Wang, Xingchen Zou, Yutian Jiang, Haomin Wen, Jiaheng Wei, Qingsong Wen, Yuxuan Liang

机构 * The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州）） ； Carnegie Mellon University（卡内基梅隆大学） ； Squirrel Ai Learning 纠错

专题命中 复杂问题求解 ：reasoning(abstract)；分类 cs.AI、cs.LG

End-to-End Argument Mining through Autoregressive Argumentative Structure Prediction

Nilmadhab Das, Vishal Vaibhav, Yash Sunil Choudhary, V. Vijaya Saradhi, Ashish Anand

机构 * Applied Machine Learning (AMaL) Lab（应用机器学习（AMaL）实验室） ； Department of Computer Science and Engineering, Indian Institute of Technology (IIT), Guwahati, India（计算机科学与工程系，印度理工学院（IIT），古瓦哈蒂，印度） ； North Eastern Regional Institute of Science and Technology (NERIST), Arunachal Pradesh, India（北东部科学与技术研究所（NERIST），阿鲁纳恰尔邦，印度） 纠错

专题命中 复杂问题求解 ：reasoning(abstract)；分类 cs.CL、cs.AI

Comments Accepted version. To appear in IJCNN 2025

Stratos: An End-to-End Distillation Pipeline for Customized LLMs under Distributed Cloud Environments

Ziming Dai, Tuo Zhang, Fei Gao, Xingyi Cai, Xiaofei Wang, Cheng Zhang, Wenyu Wang, Chengjie Zang

专题命中 复杂问题求解 ：reasoning(abstract)；分类 cs.AI、cs.LG

Enhancing Efficiency and Exploration in Reinforcement Learning for LLMs

Mengqi Liao, Xiangyu Xi, Ruinian Chen, Jia Leng, Yangen Hu, Ke Zeng, Shuai Liu, Huaiyu Wan

机构 * School of Computer Science and Technology, Beijing Jiaotong University（计算机科学与技术学院，北京交通大学） ； Meituan（美团） ； Beijing Key Laboratory of Traffic Data Mining and Embodied Intelligence（北京交通数据挖掘与具身智能重点实验室） 纠错

专题命中 复杂问题求解 ：reasoning(abstract)；分类 cs.CL、cs.LG

Comments Accept by EMNLP 2025 main

The Atomic Instruction Gap: Instruction-Tuned LLMs Struggle with Simple, Self-Contained Directives

Henry Lim, Kwan Hui Lim

机构 * Singapore University of Technology and Design（新加坡科技设计大学） 纠错

专题命中 复杂问题求解 ：reasoning(abstract)；分类 cs.CL

Comments 11 pages, 1 figure, 8 tables

Real-Time World Crafting: Generating Structured Game Behaviors from Natural Language with Large Language Models

Austin Drake, Hang Dong

机构 * University of Exeter Department of Computer Science（埃克塞特大学计算机科学系） 纠错

专题命中 复杂问题求解 ：chain-of-thought(abstract)；分类 cs.CL

Comments 16 pages, 11 figures (including appendix). To be presented at the 5th Wordplay @ EMNLP workshop (2025)

Structuring Security: A Survey of Cybersecurity Ontologies, Semantic Log Processing, and LLMs Application

Bruno Lourenço, Pedro Adão, João F. Ferreira, Mario Monteiro Marques, Cátia Vaz

专题命中 复杂问题求解 ：reasoning(abstract)

Code Execution as Grounded Supervision for LLM Reasoning

Dongwon Jung, Wenxuan Zhou, Muhao Chen

机构 * University of California, Davis（加州大学戴维斯分校） ； University of Southern California（南加州大学） 纠错

专题命中 推理评测 ：reasoning(title,abstract)；chain-of-thought(abstract)；CoT(abstract)；分类 cs.CL、cs.AI

Comments EMNLP 2025

Max It or Miss It: Benchmarking LLM On Solving Extremal Problems

Binxin Gao, Jingjun Han

机构 * University of Maryland（马里兰大学） ； Fudan University（复旦大学） 纠错

专题命中 推理评测 ：reasoning(abstract)；chain-of-thought(abstract)；CoT(abstract)；planning(abstract)

Comments Our benchmark dataset is available at https://huggingface.co/datasets/binxingao/extrem-bench

Deep Self-Evolving Reasoning

Zihan Liu, Shun Zheng, Xumeng Wen, Yang Wang, Jiang Bian, Mao Yang

机构 * Peking University（北京大学） ； Microsoft Research Asia（微软亚洲研究院） 纠错

专题命中 推理评测 ：reasoning(title,abstract)；chain-of-thought(abstract)；分类 cs.CL

Multilingual Text-to-Image Person Retrieval via Bidirectional Relation Reasoning and Aligning

Min Cao, Xinyu Zhou, Ding Jiang, Bo Du, Mang Ye, Min Zhang

机构 * School of Computer Science and Technology, Soochow University（苏州大学计算机科学与技术学院） ； School of Computer Science, Wuhan University（武汉大学计算机学院） ； Key Laboratory of New Generation Artificial Intelligence Technology & Its Interdisciplinary Applications (Southeast University), Ministry of Education, China（新 generation 人工智能技术及交叉应用重点实验室（东南大学），教育部，中国） 纠错

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

Comments Final version published in IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI). Xplore link: https://ieeexplore.ieee.org/document/11199360

Reliability of Large Language Model Generated Clinical Reasoning in Assisted Reproductive Technology: Blinded Comparative Evaluation Study

Dou Liu, Ying Long, Sophia Zuoqiu, Di Liu, Kang Li, Yiting Lin, Hanyi Liu, Rong Yin, Tian Tang

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