资讯时间轴
1562 条 · 11 个来源正文已在本站存档，原文失效也不会变成空页；已剔除失效外链与不合规内容。
LRMoE: Demand-Side Load Re-Calibration for Efficient MoE Training on Memory-Constrained Nodes
Sparse Mixture-of-Experts (MoE) models offer scalable capacity for Web applications, but uneven routing creates computation and communication bottlenecks. Exist…
来源：arXiv cs.AI · 本地全文 站内详情ThinkJev: One Thought, Every Budget for Calibrated Typed Decisions
Many uses of language models end in a typed decision: a choice among a closed set of options, whose probabilities code can act on. Reading those probabilities f…
来源：arXiv cs.AI · 本地全文 站内详情OpenFrontiers: A Collection of Open Research Problems and a Benchmark for Scientific Rediscovery
Large language models (LLMs) have shown increasing promise in research-level problem solving, yet a comprehensive collection of such problems is still missing, …
来源：arXiv cs.AI · 本地全文 站内详情Evaluating and improving AI agents for sequencing protocol understanding
Existing scientific-agent benchmarks rarely evaluate whether agents can reconstruct the exact molecular structures and generation workflows specified by sequenc…
来源：arXiv cs.AI · 本地全文 站内详情What Constitutes Food Similarity? Visual Representations and Retrieval in Indian Cuisine
Food similarity underlies retrieval, recommendation, substitution, and other food computing tasks, yet food representations are typically evaluated by identity …
来源：arXiv cs.AI · 本地全文 站内详情The Anatomy of Socio-Demographics in Pluralistic Offensiveness Modeling: When and Where It Helps, and the Role of Reasoning
Subjective judgments such as offensiveness often vary substantially across individuals, motivating pluralistic modeling approaches that incorporate annotators' …
来源：arXiv cs.AI · 本地全文 站内详情Beyond Low-Pass Dynamics: Frequency-Selective Spiking Reservoirs with Resonant Neurons
Spiking reservoir computing combines the training efficiency of fixed recurrent reservoirs with the energy efficiency of event-driven inference, where computati…
来源：arXiv cs.AI · 本地全文 站内详情SciGame: Let Agents Play the Scientific Discovery Games in the Real and Counterfactual World
Large language model agents are increasingly used for scientific discovery. However, existing benchmarks mix law inference with programming, score data fitting …
来源：arXiv cs.AI · 本地全文 站内详情VerifiedLab: Harnessing Knowns for Unknowns
Across human history, knowledge has undergone several transitions, from orality to writing and then to digital archives. At each transition, humans established …
来源：arXiv cs.AI · 本地全文 站内详情Understanding Large Language Models as Numerical Optimizers
Large language models (LLMs) have emerged as potential tools for numerical optimization. Despite promising results, their intrinsic optimization capabilities an…
来源：arXiv cs.AI · 本地全文 站内详情Algorithmic Bias from the Perspectives of Healthcare Professionals
This paper focuses on algorithmic bias of machine learning and artificial intelligence applications in healthcare information systems. Based on the quantitative…
来源：arXiv cs.AI · 本地全文 站内详情Assessing Language Models for Scientific Applications: Inertial Confinement Fusion as a Use Case
Advancing language models for specialized scientific applications requires evaluations of scientific fluency, physical reasoning, and quantitative analyses. Usi…
来源：arXiv cs.AI · 本地全文 站内详情PromptSense: Sentiment-Aware Mental Health Chatbot with Dynamic Prompting
Current Large Language Models (LLMs)-based mental health chatbots often lack sufficient empathy, contextual awareness, and safety mechanisms, potentially leadin…
来源：arXiv cs.AI · 本地全文 站内详情Evaluating Causal Reasoning in LLMs: A Case Study in the Challenges of Procedural Generation
Large language models (LLMs) appear to represent and reason about a large range of phenomena. In particular, recent studies have examined whether LLMs can accur…
来源：arXiv cs.AI · 本地全文 站内详情AI-Driven Digital Twin Framework for Warehouse Logistics Automation Optimization
In recent years, warehouse logistics optimization has increasingly focused on intelligent, data-driven systems to enhance operational efficiency. Conventional s…
来源：arXiv cs.AI · 本地全文 站内详情Too Hard, Too Easy, Just Right: Performance-Driven Challenge Adaptation in Virtual Reality Firefighter Training
Effective firefighter training requires practice under dynamic and stressful emergency conditions, yet real-world live-fire exercises are costly, risky, and dif…
来源：arXiv cs.AI · 本地全文 站内详情Post-Training Large Language Models with LLMON
Large Language Models (LLMs) treat all context tokens uniformly, lacking explicit separation between executable instructions and passive data. This leads to pro…
来源：arXiv cs.AI · 本地全文 站内详情SkillCT: A Multidimensional Benchmark and Execution-Free Baseline for Agent Skill Assessment
Agent skills package instructions, scripts, and resources for specialized tasks and are increasingly distributed through public Web repositories. However, exist…
来源：arXiv cs.AI · 本地全文 站内详情BridgeRouter: Gain-Aware Instance Routing over Cross-Modal Bridge Portfolios
Lightweight representation bridges make pretrained, task-adapted EEG foundation models executable on other biosignal modalities without retraining their backbon…
来源：arXiv cs.AI · 本地全文 站内详情DecompGym: Benchmarking LLM Agents on Designing Decomposition Algorithms
Benders decomposition and column generation solve large optimization problems by passing cuts, columns, and dual values between a master problem and subproblems…
来源：arXiv cs.AI · 本地全文 站内详情Prototype-Guided Dynamic Domain Modulation and a Large-Scale Multi-SKU Benchmark for Shampoo Bottle Surface Defect Detection
Industrial shampoo-bottle surface defect detection must operate across many stock keeping units (SKUs), where differences in bottle geometry, material, label la…
来源：arXiv cs.AI · 本地全文 站内详情From Natural Language to Real Schematics: An Automated Workflow for Power Electronics Schematic Synthesis
Designing power electronics converters traditionally requires intensive human expertise across topology sizing, commercial component selection, circuit simulati…
来源：arXiv cs.AI · 本地全文 站内详情GAAC: Lossless KV Compression for Bandwidth-Efficient LLM Inference on Personal Devices
Weight-only, lossy quantization substantially reduces the storage and memory requirements for serving large language models (LLMs), but the growing key-value (K…
来源：arXiv cs.AI · 本地全文 站内详情Verification-Guided Retrosynthesis: Reaction Feedback and Evidence Timing
Retrosynthesis planning seeks synthesis routes from purchasable starting materials to a target molecule. Search over predicted reactions can connect a target to…
来源：arXiv cs.AI · 本地全文 站内详情Inner-Learning: From Hand-Coded Beliefs to Declarative Knowledge in Trick-Taking Games
We present a declarative approach to belief-state modeling in trick-taking card games. Starting from a state-of-the-art Skat player, we automatically extract it…
来源：arXiv cs.AI · 本地全文 站内详情Game-Theoretic Foundations for Undiscounted Bisimulation Metrics
Bisimulation metrics offer a principled notion of task-relevant state similarity in reinforcement learning, but existing constructions and methods are largely c…
来源：arXiv cs.AI · 本地全文 站内详情Investor's Last Exam: Can AI Agents Do Equity Investment Without Seeing the Future?
Financial agents must form forecasts and investment judgments from decision-time evidence, with predictions evaluated against subsequent outcomes. Existing benc…
来源：arXiv cs.AI · 本地全文 站内详情Encode Once, Answer Many: Internalizing Context via Parametric Flow
Large language models are typically adapted to a task either by placing knowledge in the input context, which keeps it external and re-encodes it on every call,…
来源：arXiv cs.AI · 本地全文 站内详情Compute Has Shape: Self-Morphing the Internal Computation of Frozen LLMs
Test-time scaling improves large language models through additional computation, but how to adapt the topology of internal computation to individual inputs rema…
来源：arXiv cs.AI · 本地全文 站内详情Governance-Aware Federated Learning for Multi-Agent AI Systems: A Policy-Driven Framework for Trustworthy Collaborative AI
Federated Learning (FL) enables collaborative model training without exchanging raw data, but conventional FL provides limited runtime governance, evidence veri…
来源：arXiv cs.AI · 本地全文 站内详情From Leniency to Suspicion: How an Open Post-Training Recipe Changes Whom LLM Fact-Checkers Believe
LLMs increasingly fact-check online content with the poster's identity in view. However, it remains unclear which post-training changes how much a model's truth…
来源：arXiv cs.AI · 本地全文 站内详情An Alignment Benchmark for Enterprise Agents on Era by Eon
AI agents are widely used in enterprises to access business systems and act on them in pursuit of a goal. However, fulfilling a user's request can cause harm. E…
来源：arXiv cs.AI · 本地全文 站内详情PT$^3$: Prototype-Guided Self-Supervised Pretraining for Generalizable mmWave Human Point Cloud Representation
Multi-source millimeter-wave (mmWave) observations offer an opportunity to learn human representations reusable across sensing tasks. However, sparse detections…
来源：arXiv cs.AI · 本地全文 站内详情Co-Evolving Policy and Rubric via Capability Awareness for Open-Ended Generation
Rubric-based reinforcement learning (RL) offers interpretable, fine-grained supervision for aligning large language models (LLMs) on open-ended generation tasks…
来源：arXiv cs.AI · 本地全文 站内详情具身智能军备竞赛，胜负手在工程生态
开源工程底座 FluxVLA Engine 技术报告全解读。 作者丨邓哲敏 编辑丨齐铖湧 一个模型在论文中取得好成绩，距离它真正驱动机器人完成任务，还有多远？模型能否接入现有的数据与训练流程，推理速度能否跟上机器人的控制节奏，生成的动作能否连续、稳定地执行，实验结果又能否在不同机器人上得到验证——这些问题，未必能靠再提…
来源：雷锋网 · 本地全文 站内详情Counting the Bandages: On measuring the time that AI saves
I consider what it means for AI assistance to save time. I contrast a stopwatch view, on which time saved is whatever the clock records, with a ledger view, on …
来源：arXiv cs.AI · 本地全文 站内详情Why and How to Do Science When AI Is Better at It: A Framework Grounded in Human Understanding and Direction
As AI begins to match or exceed human researchers at core scientific tasks, two questions become pressing: \emph{why} should humans keep doing science, and \emp…
来源：arXiv cs.AI · 本地全文 站内详情From Structural Feedback to Prompt Policies: Learning Faithful Text-to-Image Prompt Editors
Prompt optimization should improve the image a user receives without changing the scene the user requested. Yet stylistic rewriting can weaken object attributes…
来源：arXiv cs.AI · 本地全文 站内详情Learning Human Sports Motion Feasibility via Quantile Regression on Tracking Data
We study time-to-arrival (TTA) in sports: the shortest plausible time a player needs to reach an arbitrary location on the pitch given their current kinematic s…
来源：arXiv cs.AI · 本地全文 站内详情MARQ: A Modular, LLM-Agnostic Requirement Quality Assessor Infused with Systems Engineering Knowledge
Requirements engineering (RE) is foundational to systems engineering (SE), as requirement defects can propagate downstream and adversely affect development outc…
来源：arXiv cs.AI · 本地全文 站内详情BehaveRec: When Does Long-Term Behavioral History Help Sequential Recommendation? A Study with Session, History, and Community Signals
Sequential recommendation models typically predict a user's next action from recent session interactions, which limits personalization when short-term signals a…
来源：arXiv cs.AI · 本地全文 站内详情Spiking Neural Networks Are Harder to Steal: A Benchmark Study of Model Stealing Attacks
Machine learning models deployed behind query interfaces expose more than their predictions, because an adversary can use those predictions to rebuild the model…
来源：arXiv cs.AI · 本地全文 站内详情When Retrieval Is Not Enough: Obsolete Facts in Agent Memory
Memory-agent benchmarks often collapse retrieval and answer generation into one score, making it hard to tell whether a failure comes from missing an update or …
来源：arXiv cs.AI · 本地全文 站内详情How Stable Are LLM-Generated Proofs Under Program Refactoring? A Stress Test for Verifiable Code Generation
Benchmarks for verifiable code generation usually test a proof against one fixed implementation. We ask whether the same proof remains valid after a program ref…
来源：arXiv cs.AI · 本地全文 站内详情Plenty of Room in the Middle: Revealing Opportunities on a GPU Cluster via Fine-Grained Monitoring
GPU clusters have become essential computing resources for AI and scientific research in academia and industry. However, their high cost and growing demand ofte…
来源：arXiv cs.AI · 本地全文 站内详情LoRA on a Budget: Function-Aware Post-Hoc Compression for Adapter Fleets
Low-Rank Adaptation (LoRA) has become a key mechanism for efficiently specializing foundation models into many task- and user-specific variants. Serving many Lo…
来源：arXiv cs.AI · 本地全文 站内详情CogArena: Benchmarking Multimodal Agents on Interactive Behavioral Experiments
Claims about the cognitive capabilities of large language models are typically tested by translating behavioral paradigms into text prompts, bypassing the visua…
来源：arXiv cs.AI · 本地全文 站内详情AC-JEPA: Anchored Cross-Modal Joint Embedding Predictive Architecture for Visual-to-Symbolic Translation
Cross-modal translation requires models to capture shared semantics across modalities while preserving the fine-grained structure required by the target modalit…
来源：arXiv cs.AI · 本地全文 站内详情MedConstBench: Benchmarking How Constitutions Steer LLMs in Medical Ethics
As large language models advance rapidly, they are increasingly used to answer medical questions, and many such models are now deployed locally. Once a model is…
来源：arXiv cs.AI · 本地全文 站内详情ADTGuard: Adaptive Depth-Time Safeguard for Streaming Inference in Large Language Models
Large Language Models (LLMs) can generate fluent responses to diverse user requests, yet streaming may expose users to harmful content before generation is comp…
来源：arXiv cs.AI · 本地全文 站内详情Follow the Herd: A Data-Driven Approach for Actionable Recourse
As machine learning systems make increasingly greater decisions about people's lives, it is critical to provide users with actionable recourse about how they ca…
来源：arXiv cs.AI · 本地全文 站内详情InfraDev: Can Coding Agents Develop AI Infrastructure, and How Can We Tell?
Coding agents are beginning to develop the production-grade AI infrastructure that trains and serves large models, yet existing benchmarks cover only single ker…
来源：arXiv cs.AI · 本地全文 站内详情Learning to Formalize: Bridging Open-Ended Workloads and Fast Decision Models
Decision models offer fast choices over sup- plied alternatives, while general-purpose large language models interpret open-ended tasks through flexible generat…
来源：arXiv cs.AI · 本地全文 站内详情Polarity--Time Interaction and Multi-Level Reliability for Event-Based Spacecraft Pose Estimation
Sparse event responses produce uneven keypoint errors whose effects on spacecraft pose depend on correspondence geometry and the reliability of successive obser…
来源：arXiv cs.AI · 本地全文 站内详情Learning Where to Trust: Reliability-Guided Expert Routing and Velocity Gating for 3D Deformable Registration
Deformable image registration relies on accurate anatomical correspondences. However, explicitly coupling local matching reliability with spatial context select…
来源：arXiv cs.AI · 本地全文 站内详情PsyJev: Psychological Analysis via System-1 Judgment
Self-disclosures on the Web support diverse psychological analyses, from recognizing expressed experiences to identifying symptom-relevant information. We study…
来源：arXiv cs.AI · 本地全文 站内详情Can LLMs Reliably Grade Olympiad Proofs? A Study of Mathematical Verification with LLMs
Large Language Models (LLMs) have made remarkable progress on solving competition-level mathematics, yet their ability to verify natural-language mathematical p…
来源：arXiv cs.AI · 本地全文 站内详情Dependency Graphs of Mathematical Arguments: Representation, Structural Measurement, and Proof Review
We propose a source-grounded framework for representing and measuring mathematical arguments with assistance from large language models. It includes numbered an…
来源：arXiv cs.AI · 本地全文 站内详情An Incremental Knowledge-Graph Learner Digital Twin for Evidence-Grounded Early Warning
Early-warning models in learning analytics can identify learners at risk of failure or withdrawal, but the resulting alerts are often only weakly connected to i…
来源：arXiv cs.AI · 本地全文 站内详情Conservative Policy Iteration For Zero-Sum POSGs
Occupancy-state reductions make dynamic programming possible for zero-sum partially observable stochastic games (zs-POSGs). Existing point-based solvers usually…
来源：arXiv cs.AI · 本地全文 站内详情RubricAtlas: Mining Type-Level Rubrics from Expert Reports to Evaluate Deep Research Agents
Deep research agents conduct multi-step Web searches and synthesise retrieved information into analytical reports. Evaluating these outputs requires checking wh…
来源：arXiv cs.AI · 本地全文 站内详情Noise Steering: Exploiting Wake Steering and Acoustic Directivity for Noise-Constrained Wind Farm Control via Reinforcement Learning
Existing noise mitigation strategies for wind farms usually consist in reducing turbines power output, without accounting for the wake interactions between them…
来源：arXiv cs.AI · 本地全文 站内详情Retell uses OpenAI models to help power 100M monthly calls
GPT-6 Astra tops Retell’s voice agent benchmark, while Codex helps it reach a $1.6M annual revenue run rate per employee.
来源：OpenAI 站内详情GLM-5.5 要来了？唐杰最新论文，曝出智谱神秘模型
图像生成正确率 54%，文字问答仅 14.57%。 作者丨郑佳美 编辑丨岑 峰 智谱内部的 GLM-Omni，先在一篇论文里露了面。这篇研究由清华大学与智谱 AI 联合完成，唐杰担任通讯作者。模型名称刚刚出现在论文中，社交平台上就有人把它与 GLM-5.5 联系起来。虽然论文并没有披露版本对应关系，但给出了 GLM-O…
来源：雷锋网 · 本地全文 站内详情OCR 3.0 时代来了！文心 Unlimited OCR 2 开源发布，超长文档解析迎来“页数自由”
10月27日，百度发布并开源文心 Unlimited OCR 2，全球首次实现任意页数文档解析，将 OCR 的处理上限从数十页推向“无限”，并达到最先进水平（SOTA）。实测显示，该模型能够一次性解析《哈利·波特》全系列 7 部作品，累计超过 3600 页，在文档长度大幅增加的情况下，依然保持较高的精度和处理效率，推动…
来源：雷锋网 · 本地全文 站内详情IRON-Meme: A Benchmark for Structured Understanding of Identities and Roles in Online Meme Narratives
As polarization grows on social media, understanding how political memes shape online narratives has become increasingly important. Previous research has relied…
来源：arXiv cs.AI · 本地全文 站内详情Prospect Theory in Large Language Models: A Meta-Analysis of Loss Aversion, Diminishing Sensitivity, and Probability Weighting
Prospect theory splits risky choice into three parts: loss aversion, measured by the relative weight of losses to gains, $\lambda$ (e.g., $\lambda = 2$ implies …
来源：arXiv cs.AI · 本地全文 站内详情Artificial Intelligence in Mechanical Engineering: Foundations, Opportunities, Challenges, and Responsibilities
Artificial intelligence (AI) is reshaping mechanical engineering (ME), from modeling, simulation, and optimization to diagnostics, design, and technical communi…
来源：arXiv cs.AI · 本地全文 站内详情Conformance as a Distribution: Delegation Contracts, Error Budgets, and the Allocation of Evaluation Authority in Use-Based AI Regulation
Use-based AI regulation assigns obligations by deployment risk but lacks an evidentiary object to carry them. A benchmark score is a property of a model; confor…
来源：arXiv cs.AI · 本地全文 站内详情Optimal Correction Sets for Provable Redress of Contestable Causal Discovery
Constraint-based causal discovery relies on conditional-independence (CI) tests that may conflict in finite samples, leaving no causal graph that satisfies them…
来源：arXiv cs.AI · 本地全文 站内详情Propensities Please: Exposing Alignment-Faking LLMs using Synthetic Document Finetuning
Alignment evaluations aim to assess whether large language models would engage in dangerous misaligned behaviour. Misaligned and evaluation-aware models might e…
来源：arXiv cs.AI · 本地全文 站内详情Isharah-Selfie: Continuous Sign Language Recognition Dataset for One-handed Signing
Current sign language recognition and translation benchmarks are commonly collected in controlled settings using fixed or studio-style cameras, particularly lim…
来源：arXiv cs.AI · 本地全文 站内详情Scaling Qualitative Analysis with Large Language Models: Development and Validation of the Smart Coder Framework for Human-LLM Collaborative Coding
Current uses of large language models (LLMs) for qualitative coding often rely on ad hoc prompting with limited guidance for reliability and human oversight. To…
来源：arXiv cs.AI · 本地全文 站内详情Distilling Expert Judgment at Scale
Frontier AI models are increasingly deployed on topics where the stakes are high and expert judgment is required, including healthcare, defense, finance, govern…
来源：arXiv cs.AI · 本地全文 站内详情NewsBench: Expert-Grounded Evaluation of Epistemic Quality in AI News Reporting
As frontier AI systems increasingly serve as sources of political and current-events information, evaluating the epistemic quality of their responses is now esp…
来源：arXiv cs.AI · 本地全文 站内详情Scoped Semantic Invariance as a Foundation for LLM Reliability
LLMs often answer semantically equivalent queries inconsistently, even when the underlying facts are known. We formalize consistency as semantic invariance over…
来源：arXiv cs.AI · 本地全文 站内详情GEMS-RAG: Graph-grounded Engineering Multimodal Standards Retrieval-Augmented Generation
Engineering standards are challenging knowledge sources for large language models (LLMs) because they are hierarchical, multimodal, and densely connected throug…
来源：arXiv cs.AI · 本地全文 站内详情BenchBench-Cloning: Evaluating Molecular Cloning Design and Protocol Development
Molecular cloning is one of the most routine wet-lab workflows, yet existing evaluations of language models in biology mostly test knowledge recall or isolated …
来源：arXiv cs.AI · 本地全文 站内详情Benchmarking System One Models for Time-Series Decision Making: A Case Study of Jev
Time-series tasks require both accurate predictions and fast decisions. Advances in large models have introduced decision-oriented models that directly output a…
来源：arXiv cs.AI · 本地全文 站内详情The Future Is Finer: Multivariate Time Series Super-Resolution Forecasting
Web platforms run on multivariate time series, from service telemetry and Web traffic logs to sensor streams served through Web APIs. Yet, retention policies of…
来源：arXiv cs.AI · 本地全文 站内详情SkillShift-Bench: Benchmarking Skill Learning under Environment Evolution for Web Agents
Skill learning enables web agents to reuse prior experience for long-horizon interactive tasks. However, learned skills are typically evaluated in closed enviro…
来源：arXiv cs.AI · 本地全文 站内详情Auditing evaluation artifacts and data leakage in large language model systems for rare disease diagnosis
Augmented large language model (LLM) systems have reported high benchmark accuracies in rare disease diagnosis, yet their real-world clinical utility remains un…
来源：arXiv cs.AI · 本地全文 站内详情When Generated Information Returns: Error Persistence in LLM-Mediated Information Ecosystems
Large language models (LLMs) increasingly produce online articles from material that later becomes input to other models. This circulation connects individual r…
来源：arXiv cs.AI · 本地全文 站内详情Graph-dependent pooling: adaptive granularity in hierarchical graph neural networks
Graph Neural Networks are widely used to learn structural representations of relational data. Hierarchical graph pooling learns multi-level representations by p…
来源：arXiv cs.AI · 本地全文 站内详情RAQ-MLA: Role-Aware Quantization for Multi-Head Latent Attention Caches
Multi-Head Latent Attention (MLA) has been adopted by recent frontier large language models. Instead of per-head keys and values, MLA caches a compressed KV lat…
来源：arXiv cs.AI · 本地全文 站内详情AlphaIdeaBench: Benchmarking Automated Research Idea Generation under Feasibility Constraints with Peer Review Alignment
Research ideas are always proposed under resource constraints: a plan must fit the available compute, API credit, and time. Yet evaluations of research-ideation…
来源：arXiv cs.AI · 本地全文 站内详情Physics-Informed Graph-Based Safe RL with Constraint-Specific Critics for Community Battery Coordination in Distribution Network
High penetrations of distributed energy resources (DERs) and fast-varying electric vehicle (EV) charging intensify voltage and line-limit violations in distribu…
来源：arXiv cs.AI · 本地全文 站内详情Power-of-Two Leaky Integrate-and-Fire Neurons for Event-Driven Vision-Language-Action Inference
Vision-language-action (VLA) models have demonstrated strong capabilities in robotic manipulation, yet their reliance on dense, continuous-valued computation pr…
来源：arXiv cs.AI · 本地全文 站内详情Meta-CLIP: Metadata-Conditioned Contrastive Learning for Zero-Shot Disease Inference
Medical vision--language pretraining learns transferable representations from paired images and radiology reports, yet differences in scanners and acquisition p…
来源：arXiv cs.AI · 本地全文 站内详情PACE: Research Proving Agent for Adaptive Conjecture Exploration
Many mathematical research agents focus on solving problems with fixed statements. However, conjecture research, especially in applied mathematics, may require …
来源：arXiv cs.AI · 本地全文 站内详情All the Facts, Still the Wrong Answer: Integration Gaps in Vision-Language Models
Multimodal reasoning requires models to combine multiple pieces of evidence into decisions that satisfy all relevant constraints. Evaluations centered on final-…
来源：arXiv cs.AI · 本地全文 站内详情An actively trained neural operator for near-explicit atmospheric chemistry
Earth-system models simplify atmospheric chemistry into condensed mechanisms, because near-explicit mechanisms such as the Master Chemical Mechanism (MCM) are t…
来源：arXiv cs.AI · 本地全文 站内详情The Hitchhiker's Guide to System 1 Models: Fast Decision-Making Beyond LLMS
Large Language Models excel at reasoning and generation but their autoregressive, token-by-token inference makes them unnecessarily slow, expensive, and poorly …
来源：arXiv cs.AI · 本地全文 站内详情InfiniteWorld: Separating Visual Evidence from Familiar Physics
Physical reasoning requires adapting familiar expectations to the dynamics of the observed world. Existing benchmarks establish broad physical capabilities, yet…
来源：arXiv cs.AI · 本地全文 站内详情Unlocking the Specialization Capability of PPG Foundation Models for Wearable Signals through Structured Pruning and Adaptive Fine-Tuning
Photoplethysmography (PPG) plays a central role in wearable health monitoring and clinical decision support. However, existing approaches to universal PPG repre…
来源：arXiv cs.AI · 本地全文 站内详情Build, Don't Play: Self-Improving LLM Research Agents Develop a Hearthstone Battlegrounds AI That Rivals Top Players
Can LLM agents carry a hard real-world problem from raw data to a system that holds its own against people? We study this in Hearthstone Battlegrounds, an eight…
来源：arXiv cs.AI · 本地全文 站内详情Cohere says Alibaba’s Qwen model tried to send a file to Beijing
An AI model said it wouldn’t use any tools, then tried to send something to a server in Beijing anyway, and while researchers say it’s probably a training quirk…
来源：TechCrunch AI · 本地全文 站内详情How Check Point accelerates vulnerability research with Daybreak
With OpenAI Daybreak, Check Point speeds up the research behind new security protections, enabling its analysts to turn vulnerability intelligence into customer…
来源：OpenAI 站内详情Patronizing and Condescending Language Detection in English and Mandarin
Patronizing and condescending language (PCL) is a form of harmful language that belittles and assumes pity for vulnerable groups, reinforcing negative societal …
来源：arXiv cs.AI · 本地全文 站内详情PaperFormalizer: Author-in-the-Loop Autoformalization of Research Papers in Lean
Even a carefully proofread paper can leave small gaps in its proofs or its stated assumptions. We introduce PaperFormalizer, a Lean-based agentic pipeline that …
来源：arXiv cs.AI · 本地全文 站内详情$M^3$: A Multi-LLM System for Multi-Turn Jailbreaking as a Markov Decision Process
Modern Large Language Models (LLMs) remain vulnerable to carefully orchestrated multi-turn interactions that progressively bypass safety mechanisms. However, ma…
来源：arXiv cs.AI · 本地全文 站内详情Reproducible Evaluation as the Substrate for Safe Prompt Compression in Production Agents
Evaluating conversational agents that act on live, mutable enterprise data is hard: replayed sessions drift as backend state changes, and a benchmark built toda…
来源：arXiv cs.AI · 本地全文 站内详情CEMIR vs. FISER: a comparative study of reasoning-based and enrichment-based approaches to pragmatic intent recognition
Intent recognition in human-robot interaction requires agents to resolve ambiguous natural language instructions by inferring the speaker's underlying goals. Re…
来源：arXiv cs.AI · 本地全文 站内详情PPO vs. SAC: A Comparative Study for Cooperative Motion Control of Dual-Arm Space Robots
The growing population of orbital debris has intensified interest in active debris removal and on-orbit servicing, for which dual-arm space manipulators offer i…
来源：arXiv cs.AI · 本地全文 站内详情DELA: Learning Controllable Edits for Diffusion Language Models
Diffusion language models (DLMs) support bidirectional denoising and iterative refinement, but generation remains largely controlled by token-level masking sche…
来源：arXiv cs.AI · 本地全文 站内详情Learning Nonlinear Waveform Fusion Across Scales for Integrated Sensing and Communication
Fixed source superposition and frequency partitioning constrain how communication content interacts with a probing reference in integrated sensing and communica…
来源：arXiv cs.AI · 本地全文 站内详情PersonaScienceBench: Evaluating Scientific Agents with Diverse Simulated Researchers
Scientific agents collaborate with researchers whose expertise and preferences differ, but benchmarks usually hold the collaborator fixed. We introduce PersonaS…
来源：arXiv cs.AI · 本地全文 站内详情Who Verifies the Tools? How Warped Formatters Mislead Agents and Their Verifiers
Agents and verifiers often rely on the text a tool prints, and that text can misrepresent what the tool actually did, even when the tool's computation is correc…
来源：arXiv cs.AI · 本地全文 站内详情MPQ: A Message-Passing View of Post-Training Quantization
Post-training quantization (PTQ) is a key technique for efficiently deploying large language models, as it compresses weights into low-bit representations witho…
来源：arXiv cs.AI · 本地全文 站内详情Not All Information Is Worth Asking For: Decision-Informed Information Seeking with LLMs
Large language models (LLMs) are increasingly used to support decisions under incomplete information, where deciding what to look up and when to stop matters as…
来源：arXiv cs.AI · 本地全文 站内详情Evidence-Grounded SLM Behavioral Prediction Architecture for Future Assistance Use
Users often exhibit changes in interaction behavior before explicitly seeking assistance, enabling systems to anticipate future assistance use. This is especial…
来源：arXiv cs.AI · 本地全文 站内详情SWE-Shift: Evaluating Coding Agents in Evolving Workspaces
Repository-level coding benchmarks largely assume that tasks and environments remain unchanged throughout execution. In software development, code, dependencies…
来源：arXiv cs.AI · 本地全文 站内详情RxnExBench: Benchmarking Vision-Language Models for Structured Chemical Reaction Extraction from Scientific Literature
Screen Tables and Substrate Scopes are information-dense figures widely used in synthetic chemistry literature, yet automated extraction of structured reaction …
来源：arXiv cs.AI · 本地全文 站内详情Spatial Heuristic Strategy based Diffusion Reinforcement Learning for Online 3D Bin Packing
Traditional heuristic methods have accumulated mature experience in core objectives, such as space utilization and layout compactness of 3D bin packing, relying…
来源：arXiv cs.AI · 本地全文 站内详情Trial by Trace: Benchmarking AI Agent Harnesses for Reliability and Failure Attribution
LLM agents are increasingly used to solve real-world problems, and their effectiveness depends on both the underlying models and the harnesses that support thei…
来源：arXiv cs.AI · 本地全文 站内详情From Visual Affinity to Learning Credit in Multimodal Reinforcement Learning
Reinforcement learning with verifiable rewards (RLVR) improves multimodal reasoning, but uniformly assigning a single response-level advantage to all tokens ove…
来源：arXiv cs.AI · 本地全文 站内详情TGR-EvoDAG: A DAG-based Multi-Agent Code Evolution Framework for Generative Recommendation on Tencent
Large Language Model (LLM) agents have enabled automated code evolution for algorithm discovery. However, current recursive self-improvement (RSI) research for …
来源：arXiv cs.AI · 本地全文 站内详情FOCUS: A Lightweight Mamba-2 Model for EEG-Based Cognitive Workload Classification
Cognitive workload estimation from physiological signals is increasingly relevant in high-risk environments, where continuous and unobtrusive monitoring can inf…
来源：arXiv cs.AI · 本地全文 站内详情Async-RBench: Benchmarking Asynchronous Task Execution in LLM Agents
Large language model (LLM) agents increasingly execute tasks asynchronously: a main agent continues working while its subagents run, and the task state may chan…
来源：arXiv cs.AI · 本地全文 站内详情ALLMTS: Automated LLM-based Medical Triage System
Telephone triage is a safety critical task: from a brief intake note, a nurse must decide whether a caller needs an ambulance, an emergency department, an appoi…
来源：arXiv cs.AI · 本地全文 站内详情