论文

本文提出 Chamaileon，通过将问题建模为跨上下文结合景观（cross-context binding landscape modeling），统一多目标与多态 binder 设计，有效生成可适配多样构象景观与多目标需求的序列。Chamaileon is introduced, which unifies multi-target and multi-state binder design by formulating the problem as cross-context binding landscape modeling and effectively generates sequences adaptable to diverse conformational landscapes and multi-target requirements.

即使 NVIDIA 控制流 ISA 与重汇聚（reconvergence）机制持续演进，Divergence 仍能保持稳定且可预期的性能开销。Divergence retains a stable and predictable performance cost even as NVIDIA's control-flow ISA and reconvergence mechanisms continue to evolve.

本文提出 IndicTalk，目前最大的多语言印度语码混合（code-mixed）会话语料库，包含超过 13,28,604 段事件驱动的多轮对话，覆盖 9 种印度语言的 18 种语言变体，将公开发布以支持低资源印度语多语言会话 AI 的开发与评估。IndicTalk is presented, one of the largest multilingual Indic code-mixed conversational corpora, comprising over 13,28,604 event-grounded multi-turn conversations across 18 language varieties covering 9 Indic languages and will be released to support the development and evaluation of multilingual conversational AI for underrepresented Indic languages.

RedE 利用 final-answer attention 作为自动监督信号来塑造 step-level 表征空间，使其中的噪声步骤可被可靠识别与过滤，并在检测性能上超越有竞争力的基线。RedE leverages final-answer attention as an automatic supervision signal to shape the step-level representation space, yielding refined embeddings in which noisy steps can be reliably identified and filtered and improves detection performance over competitive baselines.

本文提出 PAJAMA，该系统将程序合成为 judge，将其决策聚合为联合裁决（joint verdict），并通过 fallback 机制选择性地将低置信度用例升级交由 LLM 处理。PAJAMA is introduced, a system that synthesizes programs as judges, aggregates their decisions into a joint verdict, and incorporates a fallback mechanism to selectively escalate low-confidence cases to an LLM.

WorldDiT 是一种统一的 diffusion Transformer 架构，将动作生成与视觉世界建模耦合，无需大型预训练 VLM 动作主干即取得强性能，在报告全部四个 suite 的方法中，其总模型参数量与平均成功率处于已报告的 Pareto 前沿上。WorldDiT, a unified diffusion transformer architecture that couples action generation with visual world modeling and achieves strong performance without a large pretrained VLM action backbone, lies on the reported Pareto frontier for total model parameters and mean success among methods reporting all four suites.

大量实验表明，结合端上延迟感知设计与定制化训练策略，建立了高效 LVLM 编码的新 SOTA，在端上以近 1.7 倍速度运行的同时显著优于现有以编码器为中心的基线。Extensive experiments demonstrate that the on-device latency-informed design combined with the tailored training strategy establishes a new state-of-the-art for efficient LVLM encoding, significantly outperforming existing encoder-centric baselines while operating on-device at nearly 1.7xthe speed.

本文提出一种面向由一个或多个 agent 构建的自动化研究系统的词汇表，使其设计选择更易于描述与比较，从而将结构性设计问题——例如 agent 应在何时通信、获得或失去某项能力，或在多次运行间传递信息——转化为可测试的选择。A vocabulary for automated research systems built from one or more agents to make their design choices easier to describe and compare, which turns structural design questions, such as when agents should communicate, gain or lose a capability, or carry information across runs, into testable choices.

本文提出问题：与其缩减真实机器人数据占比，不如提高无机器人 UMI 数据的保真度，从而在后训练阶段移除该 anchor；并开源 HiFi-UMI，一套面向轨迹精度、夹爪间相对位姿、同步与视场协同设计的便携式 UMI 数据生产系统。It is asked whether raising the fidelity of robot-free UMI data, rather than shrinking the real-robot fraction, can remove that anchor at post-training, and open-source HiFi-UMI, a portable UMI data-production system co-designed for trajectory accuracy, inter-gripper relative pose, synchronization, and field of view.

本文提出 ReDesign，一种 agentic 框架，通过跨模态选择与组合专用工具来构建可编辑的层级结构（layer hierarchy），在取得强视觉保真度的同时，于布局、颜色与文本编辑上提供最高的可编辑性。ReDesign is presented, an agentic framework that grows an editable layer hierarchy by selecting and composing specialized tools across modalities, and achieves strong visual fidelity while delivering the highest editability across layout, color, and text edits.

本文发现视觉提示工程（visual prompt engineering，简称 VIPE）能在多项任务上提升视频推理性能，甚至比经典的文本提示工程或 test-time scaling 更有效。It is found that visual prompt engineering, or VIPE for short, improves video reasoning performance across tasks and can be even more effective than classic text-based prompt engineering or test-time scaling.

本文在由专家级 MITRE Center for Threat-Informed Defense 标注构成的、包含 1,207 条 CVE 的精选 gold 数据集上训练多标签分类器，结果表明该分类器受限于标签质量而非数据规模。A multi-label classifier is trained on a curated gold dataset of 1,207 CVEs from expert MITRE Center for Threat-Informed Defense mappings, indicating that the classifier is limited by label quality rather than dataset size.

本文提出 RARG（Relevance-Aware RipGrep Search Agent），将相关性转化为 corpus 交互的执行先验，并证明相关性感知交互可带来更快且更可靠的搜索收敛。The Relevance-Aware RipGrep Search Agent (RARG) is introduced, which turns relevance into an execution prior for corpus interaction, and demonstrates that relevance-aware interaction enables faster and more reliable search convergence.

本文提出 InMind，一项包含 125 个任务、经专家校验、跨越十个生活领域的基准，其中 113 个任务基于可引用的公开来源；本文将此类失效模式命名为 implicit-association blind spot，并提出一种极简诊断探针（diagnostic probe），在 query 到达前保持 memory 可见即可恢复大部分性能差距。This work introduces InMind, a 125-task, expert-verified benchmark spanning ten life domains, with 113 tasks grounded in citable public sources, and calls this failure mode the implicit-association blind spot, and introduces a minimal diagnostic probe that keeps memory visible before the query arrives recovers most of the gap.

本文提出 Mage-VL，一种面向实时多模态理解与交互的高效 codec-native 流式基础模型，并构建了 AI4AI 数据流水线，涵盖面向多模态 captioning 的 prompt-code 联合优化与以 AI 驱动的性能诊断，以指导训练方案。Mage-VL is presented, an efficient codec-native streaming foundation model for real-time multimodal understanding and interaction and establishes AI4AI data pipelines encompassing prompt-code joint optimization for multimodal captioning and AI-driven performance diagnosis to guide training recipes.

PerceptionBench 通过诊断前沿 MLLMs 在 42 项现有基准上响应中的最早失效点，并构建一个感知分支定义十种原子感知能力的错误分类法，为衡量与诊断 MLLMs 视觉感知边界提供了一项能力级（capability-level）标准。PerceptionBench provides a capability-level standard for measuring and diagnosing the visual perception boundaries of MLLMs, by diagnosing the earliest failure points in the responses of frontier MLLMs across 42 existing benchmarks and constructing an error taxonomy whose perception branch defines ten atomic perceptual capabilities.

本文综述了该边界上的五类漏洞：多步攻击链、与沙箱边界冲突的目标、供应链与凭据暴露、持续性的 command-and-control，以及自动化行动的速度。This review synthesizes five vulnerability classes at that boundary: multi-step offensive chains, objectives that conflict with sandbox boundaries, supply-chain and credential exposure, persistent command-and-control, and the speed of automated action.

本文提出 Kontrast，一种利用 Text-to-SPARQL 与 LLM 推理将基于表格的答案与 KG 证据进行对比并对所产生的不一致性进行分类的自动框架，并表明文本、表格与 KG 可通过系统性对比相互补充与纠错。Kontrast is presented, an automatic framework that uses Text-to-SPARQL and LLM reasoning to compare table-based answers with KG evidence and categorize the resulting inconsistencies, and shows that text, tables, and KGs can complement and correct one another through systematic comparison.

BeyondUncertainty 首先引出结构化的临时答案与置信度估计，然后应用在 held-out 验证数据上选定并在测试评估前冻结的模型特定阈值，揭示了更具选择性的证据获取与端到端 token 效率之间的权衡。BeyondUncertainty first elicits a structured provisional answer and confidence estimate, then applies a model-specific threshold selected on held-out validation data and frozen before test evaluation, revealing a trade-off between more selective evidence acquisition and end-to-end token efficiency.

介绍 RepoReasoner，一个用于评估仓库级代码推理的 benchmark，评估两种互补能力：Output Prediction，衡量跨文件的细粒度、有状态执行推理；Call Chain Prediction，在噪声上下文下评估高层架构依赖理解。RepoReasoner is introduced, a benchmark for evaluating repository-level code reasoning that assesses two complementary abilities: Output Prediction, which measures fine-grained, stateful execution reasoning across files, and Call Chain Prediction, which evaluates high-level architectural dependency understanding under noisy context.

这些结果共同支持多视角的仓库上下文服务，在仓库上下文全生命周期中提供显式的、面向操作的有效性边界以及质量-成本前沿。Together, these results support multi-view repository-context serving with explicit, operation-specific validity boundaries with quality-cost frontiers across the repository-context lifecycle.

介绍 Parallel Decoding Distillation，一种简化且可扩展的基于轨迹的蒸馏方法，用于 diffusion 和 flow matching 模型的快速推理，并显著提升生成视频的多样性。Parallel Decoding Distillation is introduced, a simplified and scalable trajectory-based distillation method for fast inference of diffusion and flow matching models and presents a significant improvement in generated video diversity.

介绍 VisualPatchWorld，将世界动态表示为代码，先通过短时主动探查选择定性动力学形式，再通过最小化多步预测误差，从记录的状态-动作轨迹中拟合该形式的自由参数。VisualPatchWorld is introduced, which represents world dynamics as code and first selects a qualitative dynamical form with short active probes, then fits that form's free parameters from recorded state-action traces by minimizing multi-step prediction error.

Temporal-Distance-JEPA 通过发现离线日志中的时间进展结构，并将代价形式与规划时部署协同设计，缩小了 JEPA world-model 规划器的训练-规划差距。Temporal-Distance-JEPA narrows the train--plan gap for JEPA world-model planners by discovering temporal progress structure in offline logs and co-designing cost form with plan-time deployment.

本文提出基于双向 Mamba 块（Vim）的通用视觉 backbone，通过位置嵌入标记图像序列，并利用双向 state space model 压缩视觉表征，具有成为下一代视觉基础模型 backbone 的巨大潜力。This paper proposes a new generic vision backbone with bidirectional Mamba blocks (Vim), which marks the image sequences with position embeddings and compresses the visual representation with bidirectional state space models and has great potential to be the next-generation backbone for vision foundation models.

本文提出 embodied language model，将真实世界连续传感器模态直接融入语言模型，从而建立词语与感知之间的联系，实现真实世界中的通用推理。This work proposes embodied language models to directly incorporate real-world continuous sensor modalities into language models and thereby establish the link between words and percepts to enable general inference in the real world.

本文提出极简的预训练框架 SimVLM，在广泛的判别式与生成式视觉-语言基准上显著超越既往预训练方法并取得新 SOTA，包括 VQA、NLVR2 以及图像描述任务。This work presents a minimalist pretraining framework, named SimVLM, which significantly outperforms previous pretraining methods and achieves new state-of-the-art results on a wide range of discriminative and generative vision-language benchmarks, including VQA, NLVR2, and image captioning tasks.

介绍 HANDBOOK_md，一个包含 65 个 agentic 任务的 benchmark，模拟员工遵循公司手册的方式；每项任务对 10 份基础手册之一进行修改，变动评分所依赖的具体规则和阈值，因此没有任何两个任务共享同一套策略。HANDBOOK_md is presented, a benchmark of 65 agentic tasks modeled on how employees follow company handbooks, and every task modifies one of 10 base handbooks, altering the specific rules and thresholds on which grading depends, so no two tasks share the same set of policies.

本文研究无人机（UAV）可见光-近红外（VNIR）高光谱图像中 PFM-1 地雷的检测，使用光谱角制图（SAM）、匹配滤波器（MF）、自适应相干估计器（ACE）和约束能量最小化（CEM）。This paper studies PFM-1 landmine detection in unmanned aerial vehicle (UAV) visible and near-infrared (VNIR) HSI using spectral angle mapper (SAM), matched filter (MF), adaptive coherence estimator (ACE), and constrained energy minimization (CEM).

一项受控的种子干预实验发现，与随机非 gold 上下文相比，由检索得到的初始上下文以更少的后续探索取得更高的文件 F1，而 oracle gold 上下文仍存在可观的提升空间。A controlled seed-intervention pilot finds that retrieval-derived initial context yields higher file F1 with less post-seed exploration than random non-gold context, while oracle gold context shows substantial remaining headroom.

基于 ONNX Runtime 的原生 C++ 推理引擎击败了所有 baseline，置信区间甚至不重叠；batch 策略的影响比语言或运行时的选择更关键，超出作者预期。A native C++ inference engine on ONNX Runtime that beat every baseline, confidence intervals didn't even overlap, and batching strategy mattered more than either the language or the runtime choice, more than the authors expected.

本文提出 Metis，首个 memory foundation model 原型，赋予 foundation model 原生记忆能力，并表明原生记忆在架构、端到端优化和效率方面具有优势。This paper proposes Metis, the first prototype of memory foundation models, which empower foundation models with native memory capabilities and shows that native memory offers advantages in architecture, end-to-end optimization, and efficiency.

跨任务的测试时扩展实验表明，SkillRise 跨任务复用可迁移的 skill，而非受益于对同一任务的重复采样，在保持强性能的同时显著降低多阶段 skill 学习流水线的运行时开销。Scaling at test time across tasks suggests that SkillRise reuses transferable skills across tasks rather than benefiting from repeated sampling of the same task, and retains strong performance while substantially reducing the runtime overhead of skill learning pipelines with multiple stages.

介绍 DecoEvo（Decoupled Co-Evolution），在解耦目标下协同进化一个求解器 skill 和一个 rubric 生成器 skill，优化过程中不使用 gold rubric。DecoEvo (Decoupled Co-Evolution), which co-evolves a solver skill and a rubric-generator skill under decoupled objectives without using gold rubrics during optimization, is introduced.

本文介绍 CLBench-V，一个多模态上下文学习 benchmark，围绕三个维度组织任务——上下文 grounding、新信息应用与新知识学习——以解决定位上下文使用失效位置的难题。This work introduces CLBench-V, a benchmark for multimodal context learning that addresses the difficulty of localizing where context use breaks down by organizing tasks around three dimensions: context grounding, new information application, and new knowledge learning.

提出 CAST（Credit Assignment from Solver Teachers），将游戏求解器状态价值的变化转换为求解器优势，并将其作为 turn 级信号注入 RLVR，在 ALFWorld 和 WebShop 上取得最高的平均 zero-shot 性能。CAST (Credit Assignment from Solver Teachers), which converts value changes in a game solver's state value into solver advantages and injects them into RLVR as turn-level signals and achieves the highest average zero-shot performance on ALFWorld and WebShop.