论文

推出 DeepSeek-V4.1-Flash 模型，这是一个具有 552B 骨干参数、支持最长一百万 token 上下文的多模态 Mixture-of-Experts 模型，显著提升了 agent 工作负载的成本效率，并突破了 KV cache 压缩的极限。The DeepSeek-V4.1-Flash model, a multimodal Mixture-of-Experts model with 552B backbone parameters and support for contexts of up to one million tokens, is introduced, substantially improving cost efficiency for agentic workloads and pushing the limits of KV cache compression.

该工作提出 EvoSkill-GUI，一个免训练框架，其中每个 skill 都是一个结构化的多文件包，包含检索元数据、可执行计划、备份定位、故障恢复规则、可访问性工具以及失败案例。This work proposes EvoSkill-GUI, a training-free framework in which each skill is a structured multi-file package containing retrieval metadata, executable plans, backup localization, failure-recovery rules, accessibility utilities, and failure cases.

本文提出 ActObs，对每条轨迹中已有的观测 token 也进行监督，并将这一差异归因于 SFT：动作与观测梯度迅速趋于正交，而仅训练动作会留下较大的残留观测梯度，并将环境预测能力拉低至基座模型之下。This work introduces ActObs, which also supervises the observation tokens already present in each trajectory, and traces this difference to SFT: action and observation gradients rapidly become orthogonal, while action-only training leaves a large residual observation gradient and degrades environment prediction below the base model.

本文提出 Srijika，一个为九种 Brahmic 文字（Devanagari、Tamil、Bengali、Telugu、Kannada、Malayalam、Gujarati、Gurmukhi、Odia）生成可安装 OpenType 字体的系统，并附带一份负面结果目录，覆盖参考引导重风格化中失败的 conditioning、目标函数选择和数据凸包限制。Srijika is presented, a system for producing installable OpenType fonts for nine Brahmic scripts: Devanagari, Tamil, Bengali, Telugu, Kannada, Malayalam, Gujarati, Gurmukhi, and Odia, and a negative-results catalogue covering failed conditioning, objective choices, and data-hull limits of reference-guided restyling.

提出 FRAUDSkill——一种结构化的 frozen-weight 适配框架：底层音频-语言模型保持不变，转而优化外部的 skill program、路由策略和决策规则，并将结构化输出控制与验证引导的多路径推理相结合，以确保符合协议规范的预测。FRAUDSkill is proposed, a structured frozen-weight adaptation framework that leaves the underlying audio-language model unchanged while optimizing an external layer of skill programs, route-specific policies, and decision rules and combines structured output control with validation-guided multi-path inference to ensure protocol-compliant predictions.

HEAL 将动态信息校准因子注入协同注意力头的 value 向量中，主动调节视觉-语言依赖，引导输出分布趋向事实证据，提供了一条简单且可解释的增强模型可信度的路径。HEAL injects dynamic information calibration factors into the value vectors of synergy heads and actively regulates visual-language dependencies, steering the output distribution towards factual evidence, offering a simple and interpretable pathway to enhance model trustworthiness.

提出 ACLArena，一个全面研究、分析与评估 Agent 持续学习（ACL）的框架，并提出新的 ACL 方案：将高质量轨迹的离线回放与多个由 RL 专精化的 LoRA 专家路由网络相结合，显著提升 Agent 跨多领域学习的能力。This work introduces ACLArena, a framework for comprehensively studying, analyzing, and evaluating Agent Continual Learning, and proposes a new ACL recipe that combines offline replay over high-quality trajectories with a routed network of multiple LoRA experts each specialized via RL, substantially improving the agent's ability to learn across multiple domains.

本文构建了 TextMuSS-10M，一个涵盖 10 种文字、229 种语言的大规模合成场景文本数据集，并提出了 ScriptMoE，一种具备文字感知能力的 Mixture-of-Experts (MoE) 架构。该架构在精度上达到最高，且比 per-language experts 更简单、比 VLM 更轻量，同时精度优于两者。This work constructs TextMuSS-10M, a large-scale synthetic scene text dataset spanning 10 scripts and 229 languages and proposes ScriptMoE, a script-aware Mixture-of-Experts (MoE) architecture that achieves the highest accuracy and is simpler than per-language experts, lighter than VLMs, and more accurate than both.

总体而言，研究结果表明，长期交互会以产生安全风险的方式重塑 Agent 的协作模式；限制 Agent 可获取的交互历史数量与范围能够减少串通行为。Overall, the findings show that long-horizon interaction can reshape how agents coordinate in ways that create safety risks, and restricting the amount and scope of interaction history available to agents reduces collusion.

政治话语日益呈现更强的敌对感是普遍观感，但稳健证据仍稀缺。本文研究 1997 至 2026 年丹麦议会的归责行为，结合专门构建的分类器 BlameBERT（F1: 0.80）与多层统计建模。该分类器采用面向低至中等资源语言的标注高效流程。结果显示出一条香蕉形轨迹：归责水平约在 2016 年前下降，随后在近年（2019–2026）进入显著且持续的上升阶段。执政地位显著Political discourse is widely perceived to be growing more hostile, yet robust evidence remains scarce. This study examines blame attribution in the Danish Parliament from 1997 to 2026, combining a purpose-built classifier, BlameBERT (F1: 0.80), with multilevel statistical modeling. The classifier is constructed using an annotation-efficient pipeline for blame attribution in low-to-mid resource languages. The results reveal a banana-shaped trajectory, with blame declining until around 2016 before entering a significant and sustained increase in recent years (2019-2026). Government status consi

本文论证代理需要一个操作系统级基底，为身份、输入中介、内存治理与执行控制提供强制且不可绕过的服务，并提出 AgentKernel，一个以安全为一流设计约束为前提、面向信任的代理操作系统。This work argues that agents need an operating-system substrate providing mandatory, non-bypassable services for identity, input mediation, memory governance, and execution control, and introduces AgentKernel, a trust-native agent operating system built around the premise that security must be a first-class design constraint.

本文对从 GitHub 收集的 Jev 项目进行了大规模、数据驱动的分析，发现其公共生态在早期增长迅速，新项目不断涌现并被集成到既有仓库中；研究提示 Jev 充当一种可复用的决策组件，其功能随所嵌入的工作流而变化。A large-scale, data-driven analysis of Jev projects collected from GitHub finds rapid early growth in Jev's public ecosystem, with both new projects and integration into existing repositories, and suggests that Jev serves as a reusable decision component whose functionality varies with the surrounding workflow.

本文提出 CARD，一种通过渐进式细化实现有效个性化的层级框架：先按共享风格模式对用户聚类，再为各组学习专用的 LoRA adapter，从而在低资源场景下也能实现稳健的泛化与强劲的性能。This work presents CARD, a hierarchical framework that achieves effective personalization through progressive refinement that first clusters users according to shared stylistic patterns and learns group-specific LoRA adapters, enabling robust generalization and strong low-resource performance.

EngramRAG 是一种自适应记忆架构，将低延迟的 Waking State 反射与异步后台 Dreaming State 整合周期耦合，并引入融合密集向量、BM25 与 U-PPR 的三源混合检索，通过动态 Reciprocal Rank Fusion (RRF) 实现。The proposed EngramRAG is an adaptive memory architecture coupling a low-latency Waking State reflex with an asynchronous background Dreaming State consolidation cycle, and introduces triple-source hybrid retrieval fusing dense vectors, BM25, and U-PPR via dynamic Reciprocal Rank Fusion (RRF).

提出反事实树上的注意力搜索（ASCT），将训练时的多步搜索转化为局部动作信用，将反事实评估与策略学习相连接，同时部署时只需使用 actor。Attentive Search over Counterfactual Trees (ASCT) is introduced, a framework that turns training-time multi-step search into local action credit and connects counterfactual evaluation to policy learning while deploying the actor alone.

提出 Budgeted ATTA：测试批次中仅有一小部分可获得标签，且监督时机是 active test-time adaptation 中一个关键但尚未充分探索的方面。Budgeted ATTA is introduced in which labels are available for only a fraction of test batches, and the timing of supervision is a key, yet underexplored, aspect of active test-time adaptation.

本文提出 EVOKE，一种后训练方法，通过在固定状态下以目标多样性对直接决策施加监督，来施加压力以激发模型内化的、可迁移动作的世界知识，并提供了一种通过直接决策监督激发内化世界知识以获得可迁移动作的新视角。EVOKE is introduced, a post-training method that supplies pressure on eliciting internalized world knowledge for transferable action through direct decision supervision through goal diversity at fixed states, and offers a new perspective on eliciting internalized world knowledge for transferable action through direct decision supervision.

RAGScope 是一个泄漏受控的评估协议，用于评估仅使用任务输入、检索上下文和答案文本的本地证据门，结合了上下文分组划分、折范围预处理、组自举区间、部署工作点、端到端运行时以及显式的源偏移压力测试。RAGScope, a leakage-controlled protocol for evaluating local evidence gates that use only the task input, retrieved context, and answer text is presented, which combines context-grouped splits, fold-scoped preprocessing, group bootstrap intervals, deployment operating points, end-to-end runtime, and explicit source-shift stress tests.

将 Jev 面向决策的 API 集成到边缘服务编排中，在保持服务完成度的同时降低开销，并支持在有界契约下针对延迟受限的准入进行决策模型替换。Jev's decision-oriented application programming interface (API) is integrated into edge service orchestration to reduce overhead while retaining service completion, and decision-model substitution for latency-bound admission on bounded contracts is supported.

基于 10,732 条早期 ChatGPT 用户推文的混合方法研究，对每个主题进行深入定性情感分析，结果显示大多数早期采用者在软件开发颠覆性、娱乐与创意发挥等主题上表达了压倒性的积极情感。A mixed-method study using 10,732 tweets from early ChatGPT users to conduct an in-depth qualitative sentiment analysis of each topic, showing that the majority of the early adopters have expressed overwhelmingly positive sentiments related to topics such as Disruptions to software development, Entertainment and exercising creativity.

本文将多智能体系统（MAS）协调建模为数据管理问题，提出用智能体的状态足迹来刻画它们：即在其自身局部上下文与状态、以及编排器和外部系统状态上的读写行为。This work frames MAS coordination as a data management problem and proposes to describe agents by their state footprint: the state they read and write across their own local context and state, as well as the state of the orchestrator and external systems.

本文首次系统地探索了不同 LLM 的有效选择与组织，以培育更公平的 LLM 回答，并展示了 CBM 显著优于独立基线。This work is the first to systematically explore the effective selection and organization of distinct LLMs to cultivate fairer LLM responses and show CBM substantially outperforms standalone baselines.

结果表明现有科学软件能够为终端 Agent 提供可扩展且经过行为验证的监督，并提出 software-in-the-loop reconstruction，一种自监督框架，从现有软件工作流（即把结构化输入映射为输出的可执行程序）中获取参考输出与验证目标。Results indicate that existing scientific software can provide scalable, behaviorally verified supervision for terminal agents, and introduces software-in-the-loop reconstruction, a self-supervised framework that obtains reference outputs and verification targets from existing software workflows, executable programs mapping structured inputs to outputs.

论文提出 Self-compensating VLA，一种部署阶段的自适应方法，使 VLA 策略在生成指令时能够预补偿机器人的执行误差，并在平均任务成功率上高于基线策略以及在训练阶段增强鲁棒性的方法。Self-compensating VLA is proposed, a deployment-time adaptation method that enables a VLA policy to pre-compensate for the robot's execution errors when generating commands, and achieves higher average task success than both the base policies and methods that build in robustness during training.

视觉-语言-动作（VLA）基础模型规模迅速扩大以提升操作性能与泛化能力，但这种规模化带来了高昂的计算成本，使真实世界部署日益困难。现有方法通常通过设计更小的架构或减少基于流（flow-based）策略中的迭代去噪步数来缓解该问题。本文提出 FastOPD，一个从基础到轻量的 VLA 框架，通过高效的在线蒸馏实现大规模 VLA 的实际部署。具体而言，FastOPD 适配流映射（flow map）……Vision-Language-Action (VLA) foundation models have scaled rapidly to enhance manipulation performance and generalizability, but this scaling incurs high computational costs that render real-world deployment increasingly challenging. Existing approaches typically mitigate this issue by designing smaller architectures or reducing the iterative denoising steps in flow-based policies. In this work, we propose FastOPD, a foundation-to-lightweight VLA framework that enables the practical deployment of large-scale VLAs through efficient on-policy distillation. Specifically, FastOPD adapts a flow map

得益于涵盖全身自由度的扩展预训练数据，LingBot-VLA-2.0 在两个机器人平台上展现出强大的跨具身长时程移动操作能力。Benefiting from the expanded pretraining data that covers whole-body degrees of freedom, LingBot-VLA-2.0 demonstrates strong cross-embodiment long-horizon mobile manipulation capability across the two robotic platforms.

本工作提出 MultAttnAttrib，一种免训练的归因生成方法，利用模型的预填充过程、选定的注意力头以及校准阈值在文档中定位源证据，且在多种归因生成方法上一致地表现更优。This work introduces MultAttnAttrib, a training-free attribution-generation method that leverages a model's prefill pass, selected attention heads, and calibrated thresholds to locate source evidence within a document, and consistently outperforms a variety of attribution-generation methods.

在资源受限环境中，专业癫痫专家稀缺，使基于 LLM 的决策支持对管理纵向治疗的一线临床医生具有吸引力。此类系统必须适应当地处方实践并知道何时转诊。我们在乌干达儿科癫痫诊疗中研究该问题，基于纵向非结构化门诊记录预测抗癫痫用药方案。标准提示与医生处方取得了一定程度的一致性，但神经科医生审查显示许多错误反映的是分布失校的处方默认值而非失败。Specialist epilepsy expertise is scarce in resource-constrained settings, making LLM-based decision support attractive for frontline clinicians managing longitudinal treatment. Such systems must adapt to local prescribing practice and know when to defer. We study this problem in Ugandan pediatric epilepsy care, predicting anti-seizure medication regimens from longitudinal unstructured clinic notes. Standard prompting achieves non-trivial agreement with physician prescriptions, but neurologist review shows that many errors reflect distribution-miscalibrated prescribing defaults rather than fail

京东 Oxygen AI 商品中心（Oxygen AIIC）：基于 LLM/VLM 的工业级商品知识生产与服务平台，已在大规模场景下取得可量化的收益。The JD Oxygen AI Item Center (Oxygen AIIC), an industrial-scale platform built on LLMs/VLMs for item-knowledge production and service, has delivered measurable gains at scale.

应在真实类别分布下使用跨网络评估来判断部署就绪度，而非仅依赖域内准确率。Deployment readiness should be assessed using cross-network evaluation under realistic class distributions, rather than within-domain accuracy alone, to suggest deployment readiness should be assessed using cross-network evaluation under realistic class distributions, rather than within-domain accuracy alone.

实验结果表明，所得参数集可生成可区分的个性化换道行为，同时 RAG 始终提升偏好理解效果，尤其对隐式指令效果显著，表明将基于 LLM 的自然语言交互与 Apollo 集成以支持个性化换道行为生成具有潜力。Experimental results show that the derived parameter sets generate distinguishable personalized lane-change behaviors, while RAG consistently improves preference interpretation, particularly for implicit commands, indicating the potential of integrating LLM-based natural-language interaction with Apollo to support personalized lane-change behavior generation.

本文提出 DuoMem，一种双空间蒸馏框架，可将程序化问题求解能力从大型教师模型迁移到紧凑学生模型，并适用于实时边缘部署，而这一点对教师模型而言颇具挑战。DuoMem is introduced, a dual-space distillation framework that transfers procedural problem-solving ability from a large teacher model to compact student models and is viable for real-time edge deployment, which would be challenging for the teacher.

提出 Transparent Two-Pass Execution，一种在推理时将工具执行与 schema 约束响应生成解耦的策略；实验结果表明该方法无需模型重新训练即可恢复工具调用能力，同时保持结构化输出保证。Transparent Two-Pass Execution is proposed, an inference-time strategy that decouples tool execution from schema-constrained response generation and experimental results show that this approach restores tool invocation while preserving structured output guarantees without requiring model retraining.

AOHP 的核心设计原则是将 Agent 视为 OS 中的一等公民，从而支持自适应用户界面以及对 Agent 友好的运行时环境；在任务完成度、执行成本和安全策略合规性方面均展现出明显优势。The core design principle of AOHP is to treat agents as first-class OS actors, enabling adaptive user interfaces and agent-friendly runtime environments, and shows clear advantages in task completion, execution cost, and security-policy compliance.

本研究收集了 9,041 个使用流行 AI Agent 开发的开源应用，审计了 200 个公开部署的应用，发现 1,186 个漏洞，并对 vibe-coded 应用的安全态势提供了实证分析。This study collects 9,041 open-source applications developed using popular AI agents, audits 200 publicly deployed applications, uncovering 1,186 vulnerabilities, and provides an empirical understanding of the security landscape of vibe-coded applications.

本文命名了"检索状态锁定"这一失败模式，通过分离单一置信度分数所混淆的三个对象——答案表面、检索到的证据以及检索状态本身——来诊断该问题，并直接衡量"一致性盲区"。This work names the failure retrieval-state lock-in and diagnose it by separating the three objects a single confidence score conflates: the answer surface, the retrieved evidence, and the retrieval state itself, and measures the agreement blind spot directly.