cs.RO / 1 / 2606.11249
MASK: Multi-Agent Semantic K-Scheduling for Risk-Sensitive 6G Robotics
MASK：风险敏感6G机器人多智能体语义K调度
Abstract
Realizing the vision of 6G connected robotics requires reconciling high-performance collaborative control with the rigid spectral limitations of physical wireless channels. In realistic collaborative sensing scenarios, spectral resources are quantized into finite physical resource blocks or orthogonal subcarriers, rendering simultaneous transmission by all agents infeasible. To address this, we propose Multi-Agent Semantic K-Scheduling (MASK), a control architecture designed to sustain robust, risk-aware coordination under strict instantaneous bandwidth caps. We introduce Arbiter-Assisted Semantic Information Gating (A-SIG), a lightweight coordination mechanism that enforces hard access constraints by scheduling only the top-K agents based on locally computed semantic importance scores. By aggregating these prioritized observations into a compact latent state, a self-supervised global encoder enables a distributional policy to mitigate tail risks despite data sparsity. We evaluate MASK across diverse benchmarks, demonstrating that it matches the performance of communication-unconstrained baselines even when channel access is restricted to a small fraction of the swarm size. Furthermore, the framework exhibits inherent resilience to packet erasures, validating semantic scheduling as a critical enabler for resource-constrained 6G systems.
Chinese Translation
实现6G连接机器人愿景需要将高性能协作控制与物理无线信道的严格频谱限制相协调。在现实的协作感知场景中，频谱资源被量化为有限的物理资源块或正交子载波，这使得所有智能体的同时传输变得不可行。为此，我们提出了多智能体语义K调度（MASK），这是一种旨在在严格瞬时带宽限制下维持稳健、风险意识协调的控制架构。我们引入了仲裁者辅助语义信息门控（A-SIG），这是一种轻量级协调机制，通过基于局部计算的语义重要性评分调度仅前K个智能体，从而强制实施严格的访问约束。通过将这些优先观察结果聚合成紧凑的潜在状态，自监督全局编码器使得分布式策略能够在数据稀疏的情况下减轻尾部风险。我们在多种基准测试中评估了MASK，结果表明即使在信道访问仅限于小部分群体规模的情况下，其性能也与不受通信限制的基线相匹配。此外，该框架对数据包丢失表现出固有的弹性，验证了语义调度作为资源受限6G系统的关键推动力。
cs.RO / 2 / 2606.11278
Model-based Optimization of Anguilliform Swimming Gaits for Soft Robotic Applications
基于模型的软体机器人应用的鳗形游泳姿态优化
Abstract
In this paper, we introduce the Soft Lamprey-Inspired Dual Environment Robot (SLIDER) and a proper modeling and optimization procedure employed to design the robot. We represent the primary fluid environment actions - inertial effects, vortex forces, and viscous dissipation - using Lighthill's theory for large-amplitude elongated bodies. For structural design parameters such as internal pressure, tail size, and body stiffness, a fast, geometrically and materially nonlinear model is developed and validated. The fluid-structure interaction equations are solved implicitly with an efficient second-order box method. A pneumatic manifold robotic system is employed to actuate SLIDER in a quiescent water tank environment, allowing cross-comparison of computational and experimental results. We find that low-frequency swimming is dominated by resistant environmental forces, whereas higher-frequency swimming is primarily affected by inertial fluid forces. Using our efficient model alongside a genetic algorithm, we co-optimize a swimming control pattern and caudal fin design (subject to SLIDER's climbing morphology) to achieve a tethered swimming speed of 21.7 +/- 0.4 cm/s (0.59 Bl/s). Furthermore, we investigate the optimization procedure for a multimodal robot performing both swimming and climbing tasks.
Chinese Translation
在本文中，我们介绍了软体灯笼鱼启发的双环境机器人（SLIDER）及其设计所采用的适当建模和优化程序。我们使用Lighthill关于大幅度细长物体的理论来表示主要流体环境作用——惯性效应、涡旋力和粘性耗散。针对内部压力、尾部尺寸和身体刚度等结构设计参数，开发并验证了一个快速的几何和材料非线性模型。流体-结构相互作用方程通过高效的二阶盒方法隐式求解。我们采用气动歧管机器人系统在静水箱环境中驱动SLIDER，从而实现计算结果与实验结果的交叉比较。我们发现低频游泳主要受到环境阻力的影响，而高频游泳则主要受惯性流体力的影响。利用我们的高效模型和遗传算法，我们共同优化了游泳控制模式和尾鳍设计（受限于SLIDER的爬升形态），以实现21.7 +/- 0.4 cm/s（0.59 Bl/s）的系留游泳速度。此外，我们还研究了执行游泳和爬升任务的多模态机器人的优化程序。
cs.RO / 3 / 2606.11324
Embodied-R1.5: Evolving Physical Intelligence via Embodied Foundation Models
Embodied-R1.5：通过具身基础模型演化物理智能
Yuan, Yifu, Huang, Yaoting, Yao, Xianze, Li, Yutong, Zhang, Shuoheng, Han, Linqi, Li, Pengyi, Sun, Jiangeng, Jia, Wenting, Zhang, Zhao, Liu, Yuhao, Liao, Ruihao, Hu, Yucheng, Wu, Qiyu, Li, Yuxiao, Dong, Zibin, Ni, Fei, Zheng, Yan, Gu, Shuyang, Ma, Yi, Tang, Hongyao, Hu, Han, Hao, Jianye
Abstract
We introduce Embodied-R1.5, a unified Embodied Foundation Model (EFM) that integrates comprehensive embodied reasoning capabilities, spanning embodied cognition, task planning, correction, and pointing, within a single architecture toward general physical intelligence. Leveraging three automated data construction pipelines to significantly expand the data coverage of critical capabilities, we build a large-scale data system of over 15B tokens, and design a multi-task balanced RL recipe to alleviate heterogeneous task conflicts. We further introduce a Planner-Grounder-Corrector (PGC) closed-loop framework that enables a single model to autonomously execute and self-correct over long-horizon tasks. With only 8B parameters, Embodied-R1.5 achieves SOTA on 16 out of 24 embodied VLM benchmarks, surpassing leading models like Gemini-Robotics-ER-1.5 and GPT-5.4. Benefiting from the internalized embodied capabilities, Embodied-R1.5 can be fine-tuned into a VLA with only a small amount of data, outperforming leading VLA models like $\pi_{0.5}$ across 4 popular manipulation benchmark suites. We further conduct extensive zero-shot real-robot experiments, validating performance in instruction following, affordance grounding, articulated object manipulation, and long-horizon complex tasks, demonstrating strong generalization to the physical world. We open-source model weights, datasets, training code, and EmbodiedEvalKit, an evaluation framework tailored for embodied tasks, to facilitate future research in EFMs.
Chinese Translation
我们介绍了Embodied-R1.5，这是一种统一的具身基础模型（EFM），它在单一架构中整合了全面的具身推理能力，包括具身认知、任务规划、纠正和指向，旨在实现通用物理智能。通过利用三个自动化数据构建管道显著扩展关键能力的数据覆盖，我们构建了一个超过150亿个标记的大规模数据系统，并设计了一种多任务平衡的强化学习（RL）方案，以缓解异构任务冲突。我们进一步引入了一个规划者-基础者-纠正者（PGC）闭环框架，使单个模型能够自主执行并自我纠正长时间跨度的任务。仅凭80亿个参数，Embodied-R1.5在24个具身视觉语言模型（VLM）基准测试中的16个上达到了最新的技术水平，超越了领先模型如Gemini-Robotics-ER-1.5和GPT-5.4。得益于内化的具身能力，Embodied-R1.5可以在仅用少量数据的情况下进行微调，表现优于领先的视觉语言代理（VLA）模型，如$ ext{π}_{0.5}$，在4个流行的操作基准套件中取得了优异成绩。我们还进行了广泛的零-shot真实机器人实验，验证了在指令跟随、可供性基础、关节物体操作和长时间复杂任务中的性能，展示了对物理世界的强泛化能力。我们开源了模型权重、数据集、训练代码以及EmbodiedEvalKit，一个专为具身任务量身定制的评估框架，以促进未来在EFM领域的研究。
cs.RO / 4 / 2606.11372
HiPi: Reproducible High-Fidelity Piezoresistive Sensors for Robotic Manipulation
HiPi：用于机器人操控的可重复高保真压阻传感器
Abstract
Piezoresistive tactile sensors are attractive for robotic manipulation because they are thin, lightweight, low-cost, and scalable to dense large-area sensing. However, existing systems still face a practical trade-off: recent reproducible designs emphasize accessibility and ease of reproduction, whereas high-fidelity readout architectures remain more difficult to fabricate, assemble, and deploy. We present HiPi, a reproducible high-fidelity piezoresistive sensing system for robotic manipulation. Building on a low-crosstalk readout principle, HiPi redesigns the complete hardware stack around reproducibility, deployability, and multi-sensor scalability. The system includes a compact readout PCB compatible with commercial PCB fabrication and assembly services, eliminating manual soldering; a smaller and lower-cost STM32-based MCU module; an optimized communication pipeline that achieves 220 Hz readout in a bimanual setup with four dense tactile arrays (2048 taxels in total); and FPCB-based conductive layers that simplify sensor fabrication and stacking. Experiments with structured 3D-printed contact patterns show that HiPi preserves contact geometry substantially better than a reproducible baseline, improving the average IoU from 0.428 to 0.797 and the average Dice score from 0.539 to 0.886. These results suggest that HiPi bridges an important gap between reproducible fabrication and high-fidelity readout, making dense piezoresistive tactile sensing more practical for bimanual manipulation and multi-fingered robotic systems.
Chinese Translation
压阻触觉传感器因其薄型、轻量、低成本及可扩展至大面积密集感知的特点，受到机器人操控领域的青睐。然而，现有系统仍面临实际的权衡：近期的可重复设计强调可达性和易于复制，而高保真读出架构的制造、组装和部署仍然较为困难。我们提出了HiPi，一个用于机器人操控的可重复高保真压阻传感系统。HiPi基于低串扰读出原理，重新设计了完整的硬件堆栈，以实现可重复性、可部署性和多传感器可扩展性。该系统包括一个与商业PCB制造和组装服务兼容的紧凑型读出PCB，消除了手动焊接；一个更小且成本更低的基于STM32的MCU模块；一个优化的通信管道，在四个密集触觉阵列（总共2048个触觉单元）的双手设置中实现220 Hz的读出；以及基于FPCB的导电层，简化了传感器的制造和堆叠。通过对结构化3D打印接触模式的实验表明，HiPi在保持接触几何形状方面显著优于可重复基线，将平均交并比（IoU）从0.428提高至0.797，平均Dice得分从0.539提高至0.886。这些结果表明，HiPi在可重复制造和高保真读出之间架起了一座重要的桥梁，使得密集的压阻触觉感知在双手操控和多指机器人系统中变得更加实用。
cs.RO / 5 / 2606.11396
PLUME: Probabilistic Latent Unified World Modeling and Parameter Estimation for Multi-Finger Manipulation
PLUME：用于多指操作的概率潜在统一世界建模与参数估计
Abstract
Dexterous manipulation with multi-finger hands can be sensitive to physical parameters such as object shape, pose, and friction coefficients. While simulation enables large-scale data collection with known parameter values, simulation-trained policies must still handle uncertainty at deployment, where the true parameters and therefore the true dynamics are unknown. Standard domain randomization strategies may be insufficient for precise tasks like screwdriver turning, as manipulation strategies may need to change depending on specific parameter values. To address this, we propose Probabilistic Latent Unified world Modeling and parameter Estimation (PLUME), a world model that jointly learns to evolve a belief over parameter values as well as the system dynamics conditioned on those parameters. We learn a latent space to jointly represent multiple qualitatively different physical parameters along with rewards, themselves functions of partially-observable variables, to inform planning. Our novel learning framework leads to efficient alignment of the world model to true dynamics through online parameter inference as opposed to re-training or fine-tuning. We evaluate our method on simulated screwdriver turning, valve turning, bucket lifting, and disk flicking tasks, as well as a hardware screwdriver turning task, where we achieve successful zero-shot transfer of our simulation-trained policy and outperform state-of-the-art offline reinforcement learning and world-model-augmented behavior cloning baselines. Please see our website at https://plume-world-model.github.io for videos.
Chinese Translation
多指手的灵巧操作对物理参数如物体形状、姿态和摩擦系数非常敏感。虽然模拟可以在已知参数值的情况下进行大规模数据收集，但在部署时，模拟训练的策略仍需处理不确定性，因为真实参数及其动态特性是未知的。标准的领域随机化策略可能不足以应对像螺丝刀转动这样的精确任务，因为操作策略可能需要根据特定参数值进行调整。为了解决这个问题，我们提出了概率潜在统一世界建模与参数估计（PLUME），这是一种世界模型，能够共同学习在这些参数条件下演变的参数值信念以及系统动态。我们学习一个潜在空间，以共同表示多个质的不同物理参数以及奖励，这些奖励本身是部分可观察变量的函数，以指导规划。我们新颖的学习框架通过在线参数推断实现了世界模型与真实动态的高效对齐，而不是重新训练或微调。我们在模拟的螺丝刀转动、阀门转动、桶提升和圆盘弹射任务以及一个硬件螺丝刀转动任务上评估了我们的方法，在这些任务中，我们成功实现了模拟训练策略的零样本迁移，并超越了最先进的离线强化学习和世界模型增强行为克隆基线。请访问我们的网站 https://plume-world-model.github.io 查看视频。
cs.RO / 6 / 2606.11408
Dynamic Execution Horizon Prediction for Chunk-based Robot Policies
基于块的机器人策略的动态执行时间预测
Abstract
Action chunking has become a standard design in modern robot policies, from diffusion/flow policies to vision-language-action models, where the policy predicts a sequence of actions and executes a fixed number of them instead of acting one step at a time. However, this paradigm relies on a key assumption: a fixed execution horizon. During chunk execution, the policy operates open-loop, which is particularly problematic for fine-grained manipulation tasks that require frequent replanning. In practice, the execution horizon is typically chosen through empirical tuning and is highly task-dependent. To this end, we propose Dynamic Execution Horizon Prediction (DEHP), an effective method that trains a lightweight execution-horizon prediction branch using online reinforcement learning while keeping the pretrained chunk policy completely frozen. This makes the method compatible with black-box chunk policies and isolates the effect of adapting the execution horizon from changes to the underlying action generator. Across our evaluations, DEHP improves the success rate of different high-precision and long-horizon manipulation tasks by a large margin. Our qualitative analysis further shows that DEHP predicts shorter execution horizons during fine-grained stages of the task and longer horizons during free-space motion. In this way, DEHP balances the efficiency of open-loop chunk execution with the reactivity of closed-loop single-step control. Project page: https://dehp-chunking.github.io/
Chinese Translation
动作块化已成为现代机器人策略中的一种标准设计，从扩散/流动策略到视觉-语言-动作模型，其中策略预测一系列动作并执行固定数量的动作，而不是逐步行动。然而，这一范式依赖于一个关键假设：固定的执行时间。在块执行期间，策略以开放环路方式操作，这在需要频繁重新规划的细粒度操作任务中尤其成问题。在实践中，执行时间通常通过经验调优选择，并且高度依赖于具体任务。为此，我们提出了动态执行时间预测（Dynamic Execution Horizon Prediction, DEHP），这是一种有效的方法，通过在线强化学习训练一个轻量级的执行时间预测分支，同时保持预训练的块策略完全冻结。这使得该方法与黑箱块策略兼容，并将适应执行时间的影响与基础动作生成器的变化隔离开来。在我们的评估中，DEHP显著提高了不同高精度和长时间操作任务的成功率。我们的定性分析进一步表明，DEHP在任务的细粒度阶段预测较短的执行时间，而在自由空间运动中预测较长的执行时间。通过这种方式，DEHP平衡了开放环路块执行的效率与闭环单步控制的反应性。项目页面：https://dehp-chunking.github.io/
cs.RO / 7 / 2606.11419
A Modular Dual-Camera Pipeline for Micro-Inspection Using Aerial Robots
一种用于微观检查的模块化双摄像头管道，基于空中机器人
Abstract
Most existing drone-based inspection systems require the drone to fly dangerously close to the target or follow complex flight paths to capture small details. In addition, drone flight is affected by disturbances and localization inaccuracies, which can cause the drone to lose sight of its supposed target when it has a narrow view. Furthermore, trajectory planning often requires prior information about the target's geometry, position, and orientation, which is not always available for non-structural targets such as trees, vehicles, or people. To address these challenges, this paper presents aerial_micro_inspection, a generic pipeline for aerial micro-inspection across different use cases. The pipeline assumes a PX4-powered drone equipped with two cameras: (i) a zoomed, gimbal-mounted inspection camera that captures fine details without requiring the drone to fly very close to the target, and (ii) a wide-field-of-view stereo navigation camera that acquires the target surface on site, estimates its range, and partitions it into smaller inspection regions. In addition, a vision-based feedback loop compensates for drone motion while the inspection camera visits small partitions of a larger surface. We evaluate the pipeline in simulation and real-world experiments, mainly in two use-case scenarios: tree inspection for detecting oak processionary caterpillars and their eggs, and greenhouse inspection of sticky traps for detecting whiteflies. The results show improved coverage robustness under drone disturbances in simulation, as well as effective detection of caterpillars and eggs and high-detail imaging of insects in real-world experiments. The pipeline is open-source, developed in ROS 2, and can be adapted to new applications by replacing the surface-segmentation and micro-target detection checkpoints. The code is available at: https://github.com/SaxionMechatronics/aerial_micro_inspection
Chinese Translation
大多数现有的基于无人机的检查系统要求无人机飞行得非常接近目标，或沿复杂的飞行路径捕捉小细节。此外，无人机飞行受到干扰和定位不准确的影响，这可能导致无人机在视野狭窄时失去对目标的视线。此外，轨迹规划通常需要关于目标几何形状、位置和方向的先验信息，而对于树木、车辆或人等非结构性目标，这些信息并不总是可用。为了解决这些挑战，本文提出了一种通用的空中微观检查管道（aerial_micro_inspection），适用于不同的使用案例。该管道假设使用一架配备有两台摄像头的PX4驱动无人机：（i）一台带有云台的变焦检查摄像头，可以在不需要无人机飞得非常接近目标的情况下捕捉细节；（ii）一台广视场立体导航摄像头，可以现场获取目标表面，估计其范围，并将其划分为更小的检查区域。此外，基于视觉的反馈循环在检查摄像头访问较大表面的较小分区时补偿无人机的运动。我们在模拟和实际实验中评估了该管道，主要在两个使用案例场景中：检测橡树行军虫及其卵的树木检查，以及检测白蝴蝶的温室粘虫板检查。结果显示，在模拟中，无人机干扰下的覆盖鲁棒性有所提高，在实际实验中有效检测了毛虫和卵，并实现了高细节的昆虫成像。该管道是开源的，基于ROS 2开发，可以通过更换表面分割和微目标检测检查点来适应新的应用。代码可在以下网址获取：https://github.com/SaxionMechatronics/aerial_micro_inspection
cs.RO / 8 / 2606.11464
Bridging the sim2real gap in the table tennis robot with a transformer-based ball states predictor
基于变换器的乒乓球状态预测器在乒乓球机器人中弥合模拟与现实之间的差距
Abstract
Robotic table tennis is a representative benchmark for high-speed, closed-loop robotic control in dynamic environments, where accurate and fast prediction of ball states is critical for reliable planning and control. Physics-based approaches rely heavily on accurate parameter identification and precise initial state, while learning-based methods often struggle to capture long-range temporal dependencies and are typically trained on limited or simulated data. We propose a transformer-based framework for table tennis ball state prediction that leverages attention mechanisms to model long-range temporal correlations directly from historical observations, without relying on explicit flight or bounce models. To support robust learning and generalization, we collected a large-scale real-world dataset from players of varying skill levels and diverse ball cannon configurations. The combination of a high-capacity transformer architecture and extensive real-world data enables accurate long-horizon forecasting. Building on this capability, we introduce a plug-and-play sim-to-real transfer strategy, Swap Predictor at Deployment (SPAD), which replaces the physics-based simulator used during training with the proposed real-world-trained predictor at deployment, improving the sim-to-real transferability of the policy without requiring retraining. We demonstrate that this simple substitution effectively narrows the sim-to-real gap while preserving the efficiency and scalability of simulation-based training.
Chinese Translation
机器人乒乓球是动态环境中高速闭环机器人控制的代表性基准，其中对球状态的准确快速预测对于可靠的规划和控制至关重要。基于物理的方法在很大程度上依赖于准确的参数识别和精确的初始状态，而基于学习的方法往往难以捕捉长时间范围的时间依赖性，并且通常在有限或模拟数据上进行训练。我们提出了一种基于变换器的乒乓球状态预测框架，利用注意力机制直接从历史观察中建模长时间范围的时间相关性，而不依赖于明确的飞行或弹跳模型。为了支持稳健的学习和泛化，我们从不同技能水平的球员和多样的球发射器配置中收集了一个大规模的真实世界数据集。高容量的变换器架构与广泛的真实世界数据相结合，使得准确的长时间预测成为可能。在此基础上，我们引入了一种即插即用的模拟到现实转移策略，即部署时的交换预测器（Swap Predictor at Deployment, SPAD），该策略在部署时用提出的真实世界训练的预测器替换训练期间使用的基于物理的模拟器，从而提高了策略的模拟到现实的可转移性，而无需重新训练。我们证明了这一简单的替换有效地缩小了模拟与现实之间的差距，同时保持了基于模拟的训练的效率和可扩展性。
cs.RO / 9 / 2606.11489
Steering Multirobot Behavior via Closed-Loop Affine Activation Editing
通过闭环仿射激活编辑引导多机器人行为
Abstract
Real-world robots need to adapt their behavior beyond the envelope of their pre-trained policy. Policy finetuning or retraining are options, but they risk catastrophic forgetting, degrading the pretrained policy's base performance. To combat this, we introduce CLAE: Closed-Loop Affine Activation Editing, an inference-time framework for steering the behavior of a frozen policy by editing intermediate activations while keeping the base policy weights and downstream action head untouched. CLAE approaches behavior steering as a closed-loop problem whose outputs edit policy activations that adapt online to the robot state, environment, target behavior, and multi-robot context. It trains a sparse autoencoder over frozen-policy activations, selects behavior-relevant latent features via post-hoc probing, and learns a lightweight RL-based steering policy that applies state-dependent affine edits to selected latents during inference. We validate CLAE on a frozen multi-quadrotor navigation policy trained to perform a single task: navigating robots to a set of goal locations while avoiding obstacles. Through extensive simulations and physical tests, we show that while navigating to their goal positions, CLAE can 1. steer individual robot behavior by controlling each robot's velocity profile; 2. coordinate multirobot behavior by preserving a desired formation; and 3. produce entirely new behavior wherein robots are required to reduce their exposure to surveillance cameras in the environment.
Chinese Translation
现实世界中的机器人需要在其预训练策略的范围之外调整行为。策略微调或重训练是可选方案，但它们存在灾难性遗忘的风险，可能会降低预训练策略的基础性能。为了解决这个问题，我们提出了 CLAE：闭环仿射激活编辑（Closed-Loop Affine Activation Editing），这是一个在推理时引导冻结策略行为的框架，通过编辑中间激活来实现，同时保持基础策略权重和下游动作头不变。CLAE 将行为引导视为一个闭环问题，其输出编辑策略激活，以适应机器人状态、环境、目标行为和多机器人上下文。它在冻结策略激活上训练一个稀疏自编码器，通过后期探测选择与行为相关的潜在特征，并学习一个轻量级的基于强化学习的引导策略，在推理过程中对选定的潜在特征应用状态依赖的仿射编辑。我们在一个冻结的多旋翼导航策略上验证了 CLAE，该策略被训练以执行单一任务：引导机器人到一组目标位置，同时避免障碍物。通过广泛的模拟和实地测试，我们展示了在导航到目标位置的过程中，CLAE 可以 1. 通过控制每个机器人的速度轮廓来引导单个机器人的行为；2. 通过保持期望的队形来协调多机器人的行为；3. 产生全新的行为，其中机器人需要减少在环境中被监控摄像头暴露的时间。
cs.RO / 10 / 2606.11525
Learning Object Manipulation from Scratch via Contrastive Interaction
通过对比交互从零开始学习物体操控
Abstract
Contrastive Reinforcement Learning (CRL) has seen recent success in a wide variety of goal-conditioned robotics tasks by learning structured representations of the dynamics. However, despite its success in locomotion and simpler control domains, CRL often struggles in interaction-rich manipulation. We argue that a key source of this difficulty is object-centric interaction, such as contact or grasping, that induces distinct changes in the underlying dynamic modes. In this work, we formulate manipulation dynamics as a piecewise-smooth Markov process and show that interaction-induced mode changes create piecewise nonlinear reachability structures that are difficult for standard CRL energy functions to represent and plan over. Based on this analysis, we introduce Interaction-weighted Resampling (IWR). IWR performs interaction-aware resampling around phases before, during, and after interactions, encouraging the learned representation to preserve the mode boundaries that determine future reachability to capture multi-modal and piecewise nonlinear reachability. Across interaction-centric environments, including 2D dynamic control, robotic manipulation, and robot air hockey, IWR improves both sample efficiency and overall performance over prior CRL methods, with 19.8% average improvement in simulation. Finally, using a sim-to-real pipeline with policies trained by IWR, we demonstrate the first real-world goal-conditioned robot air hockey agent capable of hitting goals, improving success from 25% to 60%. Project Page: IWR-arxiv.github.io.
Chinese Translation
对比强化学习（Contrastive Reinforcement Learning, CRL）在多种目标条件下的机器人任务中取得了近期成功，通过学习动态的结构化表示。然而，尽管在运动和较简单的控制领域取得了成功，CRL在交互丰富的操控任务中常常面临挑战。我们认为，这一困难的关键来源是以物体为中心的交互，例如接触或抓取，这会引起潜在动态模式的显著变化。在本研究中，我们将操控动态形式化为分段光滑的马尔可夫过程，并表明交互引起的模式变化会产生分段非线性的可达性结构，这对于标准的CRL能量函数来说难以表示和规划。基于这一分析，我们引入了交互加权重采样（Interaction-weighted Resampling, IWR）。IWR在交互前、交互中和交互后的阶段进行交互感知的重采样，鼓励学习到的表示保留决定未来可达性的模式边界，以捕捉多模态和分段非线性的可达性。在以交互为中心的环境中，包括二维动态控制、机器人操控和机器人空气曲棍球，IWR在样本效率和整体性能上均优于之前的CRL方法，在仿真中平均提升了19.8%。最后，利用通过IWR训练的策略的仿真到现实管道，我们展示了第一个能够击打目标的现实世界目标条件机器人空气曲棍球代理，其成功率从25%提高到60%。项目页面：IWR-arxiv.github.io。
cs.RO / 11 / 2606.11535
Adversarial Attacks on Learned Policies for Surgical Robotic Tasks
针对外科机器人任务学习策略的对抗攻击
Abstract
Learning-based policies are being considered to augment the dexterity of human surgeons in robot-assisted surgery. Can the end-to-end mapping from visual observations to robot actions be vulnerable to adversarial attacks, potentially leading to patient injury? In this paper, we present the first study of adversarial threats to learning-based policies in surgical robotics. We investigate two threat modes: (a) disruptive attacks, where imperceptible visual perturbations interrupt policy execution, and (b) steering attacks, where such perturbations steer policy actions toward attacker-specified directions. We formulate three adversarial attack methods, each with increasing access to policy information, and evaluate their impact on two surgical subtasks: debridement and suturing. Our evaluation covers three end-to-end policy architectures: ACT, Diffusion Policy, and Pi0. In addition, we introduce a new class of photometric adversarial attacks that mimic natural visual changes, such as lighting variations, to generate effective yet visually plausible perturbations. Results from 560 physical experiments using phantoms for debridement and suturing suggest that state-of-the-art policies can be significantly disrupted, resulting in an average 61% reduction in surgical subtask success rates. Project page: https://sites.google.com/view/adversary-surgery
Chinese Translation
基于学习的策略被认为可以增强人类外科医生在机器人辅助手术中的灵活性。那么，从视觉观察到机器人动作的端到端映射是否容易受到对抗攻击，从而可能导致患者受伤呢？在本文中，我们首次研究了外科机器人领域中对基于学习的策略的对抗威胁。我们探讨了两种威胁模式：(a) 干扰攻击，其中不可察觉的视觉扰动中断策略执行；(b) 引导攻击，其中这些扰动将策略动作引导至攻击者指定的方向。我们制定了三种对抗攻击方法，每种方法对策略信息的访问程度逐渐增加，并评估它们对两个外科子任务的影响：清创和缝合。我们的评估涵盖了三种端到端策略架构：ACT、Diffusion Policy 和 Pi0。此外，我们引入了一类新的光度对抗攻击，模拟自然视觉变化，如光照变化，以生成有效且视觉上合理的扰动。使用假体进行清创和缝合的560次物理实验结果表明，最先进的策略可能会受到显著干扰，导致外科子任务成功率平均降低61%。项目页面：https://sites.google.com/view/adversary-surgery
cs.RO / 12 / 2606.11569
ConsistencyPlanner: Real-time Planning with Fast-Sampling Consistency Models
一致性规划器：基于快速采样一致性模型的实时规划
Abstract
Closed-loop planning in complex, real-world driving scenarios presents a critical challenge for autonomous driving systems. While traditional rule-based methods are interpretable, their predefined heuristics lack the adaptability for dynamic traffic environments. Learning-based approaches have shown considerable promise. Conversely, learning-based approaches, despite their promise, struggle to balance the modeling diverse and multimodal driving behaviors and real-time planning, often leading to indecisive or unsafe actions. To address this limitation, we propose Consistency Planner, a real-time planning framework with fast-sampling consistency models. Our approach is built upon two key technical contributions. Efficient Multimodal Sampling: We employ fast-sampling consistency models to generate a diverse set of plausible future trajectories. This enables efficient, real-time exploration of multimodal actions, overcoming the computational bottlenecks of previous iterative generative methods. Heterogeneous Feature Fusion: We introduce an attention-enhanced decoder that dynamically integrates heterogeneous input features (including scene feature and action token) into a cohesive representation for robust planning. Extensive evaluation in the Waymax simulator demonstrates superior performance in safety metrics compared to existing methods, with particularly strong results in challenging dynamic scenarios.
Chinese Translation
在复杂的现实驾驶场景中，闭环规划对自动驾驶系统提出了重大挑战。虽然传统的基于规则的方法具有可解释性，但其预定义的启发式方法缺乏对动态交通环境的适应性。基于学习的方法显示出相当大的潜力。然而，尽管如此，基于学习的方法在平衡多样化和多模态驾驶行为建模与实时规划方面仍然面临困难，常常导致犹豫不决或不安全的行为。为了解决这一局限性，我们提出了一种一致性规划器（Consistency Planner），这是一种基于快速采样一致性模型的实时规划框架。我们的方法基于两个关键技术贡献。高效的多模态采样：我们采用快速采样一致性模型生成一组多样的合理未来轨迹。这使得对多模态动作的高效实时探索成为可能，克服了以往迭代生成方法的计算瓶颈。异构特征融合：我们引入了一种增强注意力机制的解码器，动态地将异构输入特征（包括场景特征和动作标记）整合为一个统一的表示，以实现稳健的规划。在Waymax模拟器中的广泛评估表明，与现有方法相比，我们的方法在安全性指标上表现优越，尤其在具有挑战性的动态场景中取得了特别强的结果。
cs.RO / 13 / 2606.11577
Distortion-Resilient Robotic Imitation Learning for Autonomous Cable Routing
抗失真机器人模仿学习用于自主电缆布线
Abstract
The rapid development of intelligent control methodologies has endowed robots with powerful autonomous intelligence. Cable routing, a ubiquitous foundational task in industry, provides a rigorous benchmark for robotic dexterity and sequential decision-making. In these practical scenarios, image observation distortion frequently occurs. Samples characterized by low-quality image observations often hinder accurate model training, posing challenges to the reliability and accuracy of intelligent control systems. Nevertheless, no dedicated intelligent control solution has been proposed for scenarios of image signal distortion. Meanwhile, image quality information has not been sufficiently exploited to further enhance the performance of intelligent control methodologies. To this end, we propose a novel robotic imitation learning framework that comprises an image quality assessment module, a confidence-based learning mechanism, and a decision-making module, which is designed to maintain high performance even under distorted image observations. In the proposed framework, the image quality assessment module synergizes with the confidence-based learning mechanism to enhance the efficacy of the decision-making module. Specifically, the image quality assessment module is incorporated to extract image quality information from image observations, while the confidence-based learning mechanism adaptively prioritizes challenging samples to improve learning effectiveness. The decision-making module determines appropriate discrete skills or continuous actions. Experimental results demonstrate that our formulated framework enhances the overall performance of the decision-making module.
Chinese Translation
智能控制方法的快速发展赋予了机器人强大的自主智能。电缆布线作为工业中普遍存在的基础任务，为机器人灵巧性和顺序决策提供了严格的基准。在这些实际场景中，图像观察失真经常发生。低质量图像观察特征的样本往往阻碍了模型的准确训练，给智能控制系统的可靠性和准确性带来了挑战。然而，目前尚未提出针对图像信号失真的专门智能控制解决方案。同时，图像质量信息尚未得到充分利用，以进一步提升智能控制方法的性能。为此，我们提出了一种新颖的机器人模仿学习框架，该框架包括图像质量评估模块、基于置信度的学习机制和决策模块，旨在即使在失真图像观察下也能保持高性能。在所提出的框架中，图像质量评估模块与基于置信度的学习机制协同作用，以增强决策模块的有效性。具体而言，图像质量评估模块被纳入以从图像观察中提取图像质量信息，而基于置信度的学习机制则自适应地优先考虑具有挑战性的样本，以提高学习效果。决策模块则确定适当的离散技能或连续动作。实验结果表明，我们所构建的框架提升了决策模块的整体性能。
cs.RO / 14 / 2606.11628
LUCID: Learning Embodiment-Agnostic Intent Models from Unstructured Human Videos for Scalable Dexterous Robot Skill Acquisition
LUCID：从非结构化人类视频中学习与体现无关的意图模型，以实现可扩展的灵巧机器人技能获取
Abstract
The most widely-adopted robot learning pipelines today learn skills from robot demonstrations or structured human data, which are expensive to collect and tied to specific embodiments. In contrast, unstructured human videos provide a scalable alternative. They contain diverse manipulation demonstrations across objects, scenes, and strategies, but are not directly connected to robot action. We propose LUCID, a two-stage framework that learns task intent from unstructured human videos drawn from internet-scale datasets and learns robot control in massively-parallel simulation. The intent model predicts short-horizon intent (what should happen next in the scene) from the current observation in closed loop. An embodiment-specific sensorimotor policy converts this intent into robot actions. The intent interface is shared across controllers, so the same intent model can be applied to different embodiments, from our primary dexterous hand to a parallel-jaw gripper. We evaluate LUCID on five real-world manipulation tasks: stirring, wiping, and binning supervised by only internet video, with zero-shot transfer to novel scenes and object instances; and push-T and cable routing supervised by 1 hr each of self-collected smartphone video. Project page: https://lucid-robot.github.io/.
Chinese Translation
目前最广泛采用的机器人学习流程通常从机器人演示或结构化人类数据中学习技能，这些数据的收集成本高且与特定的体现相关。相比之下，非结构化人类视频提供了一种可扩展的替代方案。这些视频包含了跨物体、场景和策略的多样化操作演示，但与机器人动作并没有直接关联。我们提出了LUCID，一个两阶段框架，能够从互联网规模的数据集中学习非结构化人类视频中的任务意图，并在大规模并行仿真中学习机器人控制。意图模型从当前观察中预测短期意图（场景中接下来应该发生什么），并在闭环中进行。特定于体现的传感-运动策略将这一意图转化为机器人动作。意图接口在各个控制器之间共享，因此同一意图模型可以应用于不同的体现，从我们的主要灵巧手到平行夹持器。我们在五个真实世界的操作任务上评估了LUCID：搅拌、擦拭和归类，这些任务仅通过互联网视频进行监督，并实现了对新场景和物体实例的零-shot迁移；以及推-T和电缆布线，这些任务分别通过1小时自收集的智能手机视频进行监督。项目页面：https://lucid-robot.github.io/
cs.RO / 15 / 2606.11636
SAFER-Nav: Enhancing Safety for Visual Robot Navigation via Segmentation-Aware Fine-Tuning
SAFER-Nav：通过分割感知微调增强视觉机器人导航的安全性
Abstract
Vision-based navigation models, particularly foundation models, generate viable trajectories from RGB observations alone. However, even state-of-the-art transformer- and diffusion-based policies struggle to generalize in unfamiliar deployment environments containing unseen obstacles or shifted conditions. The resulting trajectories often remain goal-directed but unsafe. Existing efforts improve safety through external trajectory correction or internal geometric priors, yet the resulting policies are not trained to explicitly represent obstacle boundaries or traversable free-space structure. To address this, we propose a navigation model that incorporates these structures directly into the policy via fine-tuning and is designed to be compatible with diverse RGB-based backbones. Across multiple robot platforms, indoor environments, and static and dynamic obstacle scenarios, our method reduces collision frequency relative to ViNT, NoMaD, and their CARE-augmented variants while maintaining goal-reaching performance.
Chinese Translation
基于视觉的导航模型，特别是基础模型，仅凭 RGB 观测生成可行的轨迹。然而，即使是最先进的基于变换器和扩散的策略，在包含未见障碍物或变化条件的陌生部署环境中也难以泛化。由此产生的轨迹通常仍然是以目标为导向，但却不安全。现有的努力通过外部轨迹修正或内部几何先验来提高安全性，但所得到的策略并未经过专门训练以明确表示障碍物边界或可通行的自由空间结构。为了解决这个问题，我们提出了一种导航模型，通过微调将这些结构直接融入策略中，并设计为与多种基于 RGB 的骨干网络兼容。在多个机器人平台、室内环境以及静态和动态障碍场景中，我们的方法相较于 ViNT、NoMaD 及其 CARE 增强变体，减少了碰撞频率，同时保持了目标到达性能。
cs.RO / 16 / 2606.11704
Improving Human Diving Endurance with a Field-Deployable, Untethered Exoskeleton
通过可现场部署的无缆外骨骼提高人类潜水耐力
Abstract
Human endurance in underwater locomotion is fundamentally restricted by high energetic demands to overcome drag and the finite supply of self-contained breathing gas. While exoskeleton technology can reduce the metabolic cost of humans in terrestrial locomotion, its potential to enhance human endurance during underwater diving remains entirely unexplored. Here, we present DiveMate, a field-deployable, untethered exoskeleton designed to improve human diving endurance via adaptive kick assistance in real-world underwater environments. During naturalistic diving, DiveMate increases the travel distance using a given energy (breathing gas) by 42.9% and extends dive duration by 54.9% through reducing gas consumption rate. Marked reductions in muscle activation indicate a decrease in physiological exertion, with the net gas consumption rate decreasing by 47.0%. Kinematic characteristics and regularity improvements further underpin efficient energy economy. These results suggest that applying exoskeleton assistance is beneficial for improving human diving endurance and augmenting their ability to explore the aquatic world. This study extends the application frontier of exoskeletons and provides a potential reference for the design and assessment of future underwater assistive devices.
Chinese Translation
人类在水下运动中的耐力受到克服阻力所需的高能量需求和有限的自给呼吸气体供应的根本限制。虽然外骨骼技术可以降低人类在陆地运动中的代谢成本，但其在水下潜水中增强人类耐力的潜力仍然完全未被探索。在此，我们提出了DiveMate，这是一种可现场部署的无缆外骨骼，旨在通过在真实水下环境中提供自适应的踢腿辅助来提高人类潜水耐力。在自然潜水过程中，DiveMate在给定能量（呼吸气体）下将旅行距离提高了42.9%，并通过降低气体消耗率将潜水时间延长了54.9%。肌肉激活的显著减少表明生理负担的降低，净气体消耗率下降了47.0%。运动学特征和规律性的改善进一步支持了高效的能量经济性。这些结果表明，应用外骨骼辅助有助于提高人类潜水耐力，并增强他们探索水下世界的能力。本研究拓展了外骨骼的应用前沿，并为未来水下辅助设备的设计和评估提供了潜在参考。
cs.RO / 17 / 2606.11708
Explore From Sketch: Accelerating UAV Exploration in Large-scale Environments with Prior Maps
从草图探索：利用先验地图加速无人机在大规模环境中的探索
Abstract
Autonomous exploration with UAVs in large-scale, topologically complex environments often suffers from low efficiency due to suboptimal scheduling and detours. Prior maps (e.g., construction drawings), although usually imprecise and flawed, are readily available in many scenarios and have the potential to provide global structural guidance. This paper presents a novel exploration framework that leverages sparse, unaligned, and even discrepant 2D prior maps for LiDAR-based UAV exploration. First, a robust 2D-3D point cloud registration pipeline is proposed to align LiDAR observations with prior maps. The registration pipeline combines a GeoContext descriptor for single-frame candidate retrieval, a multi-frame verification mechanism for coarse transformation estimation with outlier rejection, and a Scale-ICP algorithm for refinement. The registration module can handle map discrepancies and provide multiple hypotheses when geometric ambiguities arise. To effectively utilize the registration results for exploration planning, we further develop a hierarchical viewpoint planning strategy under localization uncertainties. The hierarchical strategy first spatially attaches local viewpoints to prior guidepoints and adopts a Monte Carlo Tree Search solver to determine their traversal sequence under each registration hypothesis. To mitigate registration uncertainty, a risk-aware selector evaluates prior sequences using confidence-weighted travel risk, and a fixed-endpoint traveling salesman problem is formulated to generate an efficient local coverage path under the selected prior guidance. Benchmark evaluations reveal up to 34.2% improvement in exploration efficiency and 37.9% reduction in flight distance compared to state-of-the-art methods, while extensive simulations and field experiments further demonstrate robustness to prior map incompleteness and deformations.
Chinese Translation
在大规模、拓扑复杂的环境中，利用无人机进行自主探索常常由于调度不当和绕行而效率低下。尽管先验地图（例如，施工图）通常不够精确且存在缺陷，但在许多场景中它们是 readily available 的，并且有潜力提供全球结构指导。本文提出了一种新颖的探索框架，利用稀疏、不对齐甚至存在差异的 2D 先验地图进行基于激光雷达的无人机探索。首先，提出了一种稳健的 2D-3D 点云配准管道，将激光雷达观测与先验地图对齐。该配准管道结合了用于单帧候选检索的 GeoContext 描述符、多帧验证机制以进行粗略变换估计并排除异常值，以及用于精细化的 Scale-ICP 算法。配准模块能够处理地图差异，并在几何模糊性出现时提供多个假设。为了有效利用配准结果进行探索规划，我们进一步开发了一种在定位不确定性下的分层视点规划策略。该分层策略首先将局部视点空间附加到先验引导点，并采用蒙特卡洛树搜索求解器来确定在每个配准假设下的遍历顺序。为了减轻配准不确定性，风险感知选择器使用置信加权旅行风险评估先验序列，并制定固定终点的旅行推销员问题，以在选定的先验指导下生成高效的局部覆盖路径。基准评估显示，与最先进的方法相比，探索效率提高了最多 34.2%，飞行距离减少了 37.9%，同时大量模拟和实地实验进一步证明了对先验地图不完整性和变形的鲁棒性。
cs.RO / 18 / 2606.11743
TacCoRL: Integrating Tactile Feedback into VLA via Simulation
TacCoRL：通过仿真将触觉反馈整合到视觉-语言-动作(VLA)中
Abstract
Vision-language-action (VLA) models provide strong visual, language, and action priors for robot manipulation, but visual observations alone often miss the local contact state required for contact-rich tasks. We present TacCoRL, a scalable framework that injects Tactile feedback into VLA policies and improves them through sim-real Co-training and simulation-based reinforcement learning (RL), without requiring large-scale tactile pretraining or extensive real-world contact exploration. The key idea is not only adding touch as an input, but learning how contact readings should modulate action responses in near-failure states that are rare in demonstrations and risky to collect on hardware. We use a real-aligned simulator as a closed-loop training environment for contact interaction. Mixed simulated and real trajectories first warm-start tactile-conditioned actions in the pretrained policy. Reinforcement learning with verifiable task rewards then optimizes the policy using simulated contact rollouts. It reinforces tactile-conditioned actions that lead to task completion, while a supervised objective on real trajectories keeps the refined policy anchored to deployment visual, tactile, and action distributions. The resulting policy transfers directly to the real robot without privileged simulation state or online real-world RL. Across four bimanual contact-rich tasks, the final visuo-tactile policy achieves an average success rate of 72.5%, compared to baseline of 50.0%. Result videos and more details are available at https://tac-corl.github.io/
Chinese Translation
视觉-语言-动作（VLA）模型为机器人操作提供了强大的视觉、语言和动作先验，但仅凭视觉观察往往无法捕捉到接触丰富任务所需的局部接触状态。我们提出了TacCoRL，这是一个可扩展的框架，通过仿真-真实共同训练和基于仿真的强化学习（RL）将触觉反馈注入到VLA策略中，并对其进行改进，而无需大规模的触觉预训练或广泛的真实世界接触探索。其关键思想不仅在于将触觉作为输入，还在于学习接触读数如何在演示中稀有且在硬件上收集风险较高的接近失败状态下调节动作响应。我们使用与真实环境对齐的仿真器作为接触交互的闭环训练环境。混合的仿真和真实轨迹首先为预训练策略中的触觉条件动作提供热启动。然后，通过可验证的任务奖励进行强化学习，利用仿真接触回放优化策略。它强化了导致任务完成的触觉条件动作，同时在真实轨迹上施加的监督目标使得精炼后的策略保持与部署的视觉、触觉和动作分布相一致。最终的策略可以直接转移到真实机器人上，而无需特权的仿真状态或在线的真实世界强化学习。在四个双手接触丰富的任务中，最终的视觉-触觉策略实现了72.5%的平均成功率，而基线为50.0%。结果视频和更多细节可在 https://tac-corl.github.io/ 获取。
cs.RO / 19 / 2606.11767
Blind Dexterous Grasping via Real2Sim2Real Tactile Policy Learning
通过Real2Sim2Real触觉策略学习实现盲目灵巧抓取
Abstract
Blind grasping with a dexterous hand is a crucial manipulation capability. Nevertheless, learning such tactile-only policies for real robots remains challenging due to the tactile sim-to-real gap and the limited expressiveness of sparse tactile signals. To bridge this gap, we propose a framework for tactile-only blind grasping that is deployable on a physical multi-fingered robotic hand. Our approach combines three key components. First, we introduce a Real2Sim tactile calibration pipeline that constructs a contact-calibrated digital-twin simulator capable of reproducing real tactile signals. Second, we improve the expressiveness of sparse tactile observations using a layout-aware tactile encoder, which incorporates sensor-geometry priors through self-supervised pretraining. Third, to improve generalization to unseen objects, we train object-specific reinforcement-learning experts in the calibrated simulator and aggregate their successful grasp trajectories into a tactile-conditioned Diffusion Policy. We evaluate our method on a physical LEAP Hand equipped with distributed tactile sensing across 10 seen and 10 unseen objects. The deployed policy achieves a 27\% real-world grasp success rate across all 20 objects, without real-world grasping demonstrations or visual input. Simulation ablations show that layout-aware tactile pretraining improves grasping performance, while sensing-level evaluations confirm that Real2Sim calibration increases the consistency of tactile contact events between simulation and hardware. Together, these results suggest that contact-event calibration, geometry-aware tactile representation learning, and diffusion-based policy aggregation provide an effective path toward tactile-only blind grasping on real dexterous robotic hands. Project page:Dex-Blind-Grasp.github.io.
Chinese Translation
盲目抓取灵巧手是一种关键的操作能力。然而，由于触觉仿真与现实之间的差距以及稀疏触觉信号的有限表达能力，为真实机器人学习这种仅基于触觉的策略仍然具有挑战性。为了解决这一问题，我们提出了一种可在物理多指机器人手上部署的仅基于触觉的盲目抓取框架。我们的方法结合了三个关键组件。首先，我们引入了一个Real2Sim触觉校准管道，构建了一个能够重现真实触觉信号的接触校准数字双胞胎仿真器。其次，我们使用布局感知触觉编码器提高稀疏触觉观测的表达能力，该编码器通过自监督预训练结合了传感器几何先验。第三，为了提高对未见物体的泛化能力，我们在校准的仿真器中训练了特定物体的强化学习专家，并将它们成功的抓取轨迹聚合成一个基于触觉条件的扩散策略。我们在配备有分布式触觉传感的物理LEAP手上评估了我们的方法，涉及10个已见物体和10个未见物体。所部署的策略在所有20个物体上实现了27%的现实世界抓取成功率，而无需现实世界的抓取演示或视觉输入。仿真消融实验表明，布局感知触觉预训练提高了抓取性能，而传感级评估确认Real2Sim校准增加了仿真与硬件之间触觉接触事件的一致性。这些结果共同表明，接触事件校准、几何感知触觉表示学习和基于扩散的策略聚合为在真实灵巧机器人手上实现仅基于触觉的盲目抓取提供了一条有效的路径。项目页面：Dex-Blind-Grasp.github.io。
cs.RO / 20 / 2606.11818
Human-Guided Co-Manipulation of Carbon Fiber Plies
人机协同操控碳纤维层压材料
Abstract
The handling of flexible materials is a difficult task to fully automate due to the challenges caused by the deformability of these types of objects. Meanwhile, a fully manual process can be ergonomically challenging, tedious and inefficient. Thus, human-robot collaboration (HRC) and cooperative manipulation (co-manipulation) have received increasing interest in this field as they enable human involvement when needed while also improving productivity. To enable efficient co-manipulation and interaction between the human operator and the robot, different modalities and control methods are required. In this paper, we present and examine different control methods for co-manipulation of carbon fiber plies, evaluating the pros and cons of each method in a controlled setting. We propose that a multimodal combination of speech commands, wrist-tracking through vision, and force with compliant control would provide the best solution for complete and intuitive control of the task.
Chinese Translation
由于柔性材料的可变形特性，完全自动化处理这些材料是一项困难的任务。同时，完全手动的过程在人体工程学上可能具有挑战性，且繁琐低效。因此，人机协作（HRC）和协同操控（co-manipulation）在该领域受到越来越多的关注，因为它们在需要时允许人类参与，同时提高生产力。为了实现人类操作员与机器人之间的高效协同操控和互动，需要不同的模式和控制方法。本文提出并考察了碳纤维层压材料的协同操控的不同控制方法，在受控环境中评估每种方法的优缺点。我们建议，结合语音指令、通过视觉进行的手腕跟踪以及具有顺应控制的力的多模态组合，将为任务的全面和直观控制提供最佳解决方案。
cs.RO / 21 / 2606.11826
Modular Anthropomorphic Hand Design via Multi-Parameter Finger Benchmarking and Selection
通过多参数手指基准测试与选择实现模块化类人手设计
Abstract
Designing anthropomorphic dexterous robotic hands remains challenging as the design space straddles morphology, actuation, and sensing properties, and performance metrics span both task-dependent and task-agnostic. Existing optimization methods are often unstructured or consider only a single performance metric, limiting systematic comparison and targeted refinement. While the design considerations of the entire hand are significant, the individual finger properties play a key role in dexterity. By developing a robotic hand platform where fingers can be modularly integrated into a full teleoperated hand, we propose that optimizing the fingers can significantly improve overall hand performance. This approach enables rapid screening of different finger-level prototypes through a number of quantitative benchmarks before their integration into the hand for task-level validation. Candidate finger designs (incorporating variations in joint, bone, skin, and sensor placement) are assessed using both mechanism-oriented and task-relevant metrics, which establish a quantitative link between component design and full hand embodiment. The framework is validated through the development of an anthropomorphic robotic hand with optimized fingers, demonstrating how these fingers enable performance improvements across tasks, including multi-object grasping and light bulb screwing.
Chinese Translation
设计类人灵巧机器人手仍然面临挑战，因为设计空间涉及形态、驱动和传感特性，而性能指标则涵盖任务依赖和任务无关的方面。现有的优化方法往往缺乏结构或仅考虑单一性能指标，限制了系统比较和针对性改进。尽管整个手的设计考虑至关重要，但单个手指的特性在灵巧性中起着关键作用。通过开发一个可以将手指模块化集成到完整遥操作手的机器人手平台，我们提出优化手指可以显著提高整体手的性能。这种方法使得在手指集成到手中进行任务级验证之前，可以通过多个定量基准快速筛选不同的手指原型。候选手指设计（包括关节、骨骼、皮肤和传感器位置的变化）使用机制导向和任务相关的指标进行评估，从而建立组件设计与完整手体现之间的定量联系。该框架通过开发一个具有优化手指的类人机器人手进行验证，展示了这些手指如何在多任务中实现性能提升，包括多物体抓取和灯泡拧紧。
cs.RO / 22 / 2606.11891
Critic Architecture Matters: Dual vs. Unified Critics for Humanoid Loco-Manipulation
评论者架构的重要性：双重评论者与统一评论者在类人机器人运动操控中的比较
Abstract
Multi-objective reinforcement learning for humanoid robots must coordinate locomotion and manipulation within a single policy. A natural design choice is whether to use a single (unified) critic that estimates the combined value of all objectives, or separate (dual) critics with disjoint reward signals. We present a controlled comparison on the Unitree G1 humanoid (23 active DoF) in NVIDIA Isaac Lab, training loco-manipulation policies through a sequential curriculum spanning 13 levels from stationary reaching to walking with variable-orientation targets. In standardized evaluation, dual-critic policies reach targets 3.5$\times$ faster (6.5 vs. 22.6 simulation steps), achieve 2$\times$ higher throughput (14.3 vs. 7.0 validated reaches per 1,000 steps), and attain higher validated reach rates (65.2% vs. 53.8%) compared to the unified-critic policy. Notably, additional anti-gaming reward mechanisms provide no further improvement beyond the architectural change alone (60.9% vs. 65.2%). These results have direct implications for the emerging paradigm of RL fine-tuning of imitation-learned policies: when refining a pre-trained manipulation policy with RL, a unified critic risks suppressing the learned behavior through competing locomotion gradients. These findings demonstrate that critic architecture is a primary - and often overlooked - design choice in multi-objective humanoid RL, with greater impact than reward engineering on reaching efficiency.
Chinese Translation
类人机器人多目标强化学习必须在单一策略中协调运动和操控。一个自然的设计选择是使用一个单一的（统一的）评论者来估计所有目标的综合价值，还是使用多个（双重）评论者，分别处理不重叠的奖励信号。我们在NVIDIA Isaac Lab对Unitree G1类人机器人（23个活动自由度）进行了受控比较，通过一个涵盖从静止到行走的可变方向目标的13个级别的顺序课程训练运动操控策略。在标准化评估中，双重评论者策略达到目标的速度是统一评论者策略的3.5倍（6.5 vs. 22.6模拟步骤），实现的吞吐量是统一评论者策略的2倍（14.3 vs. 7.0每1,000步骤的有效到达次数），并且验证的到达率更高（65.2% vs. 53.8%）。值得注意的是，额外的反作弊奖励机制在架构改变之外并未提供进一步的改善（60.9% vs. 65.2%）。这些结果对新兴的强化学习微调模仿学习策略的范式具有直接影响：在使用强化学习微调预训练的操控策略时，统一评论者可能会通过竞争的运动梯度抑制已学习的行为。这些发现表明，评论者架构是多目标类人强化学习中的一个主要且常被忽视的设计选择，其对到达效率的影响大于奖励工程。
cs.RO / 23 / 2606.11901
DuoBench: A Reproducible Benchmark for Bimanual Manipulation in Simulation and the Real World
DuoBench：一个可重复的双手操作基准测试框架，适用于仿真和现实世界
Abstract
Bimanual robot systems substantially expand manipulation capabilities, but coordinating two arms introduces additional control complexity and failure modes that are not well captured by existing benchmarks. We introduce DuoBench, an extensible benchmarking framework for bimanual manipulation policies on the FR3 Duo platform. DuoBench comprises eleven tasks spanning four coordination categories, implemented in simulation and partially reproduced in the real world through reproducible task recipes with 3D-printable assets. In addition, we propose a stage-based evaluation scheme that supports fine-grained semantic failure analysis beyond binary success and provide human-teleoperated datasets for all benchmark tasks. We benchmark several dual-arm imitation-learning and vision-language-action policies in simulation and on real hardware. Our results show that current policies remain challenged by bimanual manipulation, particularly in early interaction stages, parallel arm execution, and transfer between simulation and real-world settings. DuoBench provides a reproducible testbed for diagnosing these failure modes and studying future methods for dual-arm policy learning. Code, datasets, and videos are available at https://duobench.github.io/
Chinese Translation
双手机器人系统显著扩展了操作能力，但协调两个手臂引入了额外的控制复杂性和失败模式，而现有基准测试未能很好地捕捉这些问题。我们提出了DuoBench，一个可扩展的基准测试框架，旨在评估FR3 Duo平台上的双手操作策略。DuoBench包含十一项任务，涵盖四个协调类别，既在仿真中实现，也通过可重复的任务配方和3D打印资产在现实世界中部分再现。此外，我们提出了一种基于阶段的评估方案，支持超越二元成功的细粒度语义失败分析，并为所有基准任务提供人类遥控操作的数据集。我们在仿真和真实硬件上对几种双臂模仿学习和视觉-语言-动作策略进行了基准测试。我们的结果表明，目前的策略在双手操作方面仍面临挑战，特别是在早期交互阶段、并行手臂执行以及仿真与现实世界设置之间的转移。DuoBench提供了一个可重复的测试平台，用于诊断这些失败模式并研究未来的双臂策略学习方法。代码、数据集和视频可在 https://duobench.github.io/ 获取。
cs.RO / 24 / 2606.11952
Deformable In-Hand Slip-Aware Tactile Sensor with Integrated Velocity, Force/Torque, and Pressure Map Sensing
具有集成速度、力/扭矩和压力图感知的可变形手内滑动感知触觉传感器
Abstract
This paper introduces a novel tactile sensor for in-hand manipulation with slip-aware control that integrates velocity, force/torque, and pressure map sensing into a single device with a deformable contact pad. To the best of our knowledge, this is the first sensor to combine these sensing modalities within a single compliant structure. The sensor features a deformable contact surface and can robustly track both flat and curved surfaces across a wide range of materials. Its performance is evaluated through a comprehensive set of experiments that highlight both its capabilities and limitations. The sensor is designed for rapid and low-cost fabrication using a combination of standard PCB manufacturing and rapid prototyping techniques.
Chinese Translation
本文介绍了一种新型触觉传感器，用于具有滑动感知控制的手内操作，该传感器将速度、力/扭矩和压力图感知集成到一个具有可变形接触垫的单一设备中。据我们所知，这是第一个在单一柔性结构中结合这些感知模式的传感器。该传感器具有可变形的接触表面，能够在各种材料上稳健地跟踪平面和曲面。通过一系列全面的实验评估其性能，突出了其能力和局限性。该传感器设计用于快速和低成本的制造，采用标准PCB制造与快速原型技术的结合。
cs.RO / 25 / 2606.12019
MPPI-based Informative Trajectory Planning for Search and Capture of Drifting Targets with ASVs
基于MPPI的漂流目标搜索与捕获的信息轨迹规划方法
Abstract
Autonomous surface vehicles offer an efficient solution for environmental cleanup as well as search and rescue operations in open waters. Targets in these settings drift continuously, so efficient search must balance exploration of unobserved regions with tracking of known targets. However, most target tracking and pursuit scenarios consider simple guidance behaviours and short-term predictions for decision-making. In this letter, we address the problem of search and capture of multiple drifting targets, such as litter, in dynamic environments, using a hybrid planning framework. A key aspect of our strategy is a spatiotemporal informative planning method based on model predictive path integral (MPPI) control, a sampling-based model predictive control approach. The planner directly generates kinematic-level commands by optimising continuous trajectories over long horizons. A multi-objective cost balances search and tracking objectives while ensuring safe, feasible trajectories. In the interception stage, we switch to a pure pursuit guidance controller for the physical capture of moving targets. Experiments show that our planner outperforms the chosen planning baselines. Finally, we validate our approach in field trials with an ASV.
Chinese Translation
自主水面车辆为环境清理以及开放水域的搜索与救援行动提供了高效的解决方案。在这些环境中，目标持续漂流，因此高效的搜索必须在探索未观察区域与跟踪已知目标之间取得平衡。然而，大多数目标跟踪和追逐场景考虑的都是简单的引导行为和短期预测用于决策。在本文中，我们使用一种混合规划框架解决动态环境中多个漂流目标（如垃圾）的搜索与捕获问题。我们策略的一个关键方面是基于模型预测路径积分（MPPI）控制的时空信息规划方法，这是一种基于采样的模型预测控制方法。该规划器通过优化长时间范围内的连续轨迹直接生成运动学级别的指令。多目标成本平衡了搜索和跟踪目标，同时确保安全、可行的轨迹。在拦截阶段，我们切换到纯追逐引导控制器以实现对移动目标的物理捕获。实验表明，我们的规划器在性能上优于所选择的规划基线。最后，我们在现场试验中验证了我们的方法。
cs.RO / 26 / 2606.12027
Learning Unions of Convex Sets via Invertible Latent Decomposition for Path Planning
通过可逆潜在分解学习凸集的并集以进行路径规划
Abstract
Collision-free path planning in cluttered, real-world environments relies on a representation of the collision-free space, and existing representations broadly fall into two categories. Explicit representations, such as unions of convex sets, can be plugged into optimization-based planners as hard collision-free constraints, but their parameters scale poorly with configuration-space dimension. Implicit representations, by contrast, are flexible and scale well to complex geometries, yet typically lack such guarantees. We bridge this gap with ILD (Invertible Latent Decomposition), a framework that jointly learns an invertible mapping and a union of explicit convex polytopes in the resulting latent space. Planning is carried out over these latent convex sets, and the invertible mapping decodes the resulting paths back to the original configuration space while preserving feasibility with respect to the refined explicit safe regions. We further propose Visibility-Guided Sampling (VGS) to keep the convex sets connected for path planning. Across 2D navigation, 6-DoF, and 14-DoF manipulation environments, ILD achieves broader coverage, better inter-set connectivity, and higher path-planning success rates than prior baselines, with zero observed false positives after test-time refinement. On a 14-DoF bimanual manipulator, we further demonstrate real-time collision-free planning, with test-time refinement adapting to scene-geometry changes during real-world deployment on a single 6-DoF arm.
Chinese Translation
在复杂的现实环境中进行无碰撞路径规划依赖于对无碰撞空间的表征，现有的表征大致分为两类。显式表征，如凸集的并集，可以作为硬性无碰撞约束直接应用于基于优化的规划器，但其参数在配置空间维度上扩展性较差。相比之下，隐式表征灵活且能够很好地适应复杂几何形状，但通常缺乏这样的保证。我们通过可逆潜在分解（ILD，Invertible Latent Decomposition）架构来弥补这一差距，该架构联合学习一个可逆映射和结果潜在空间中的显式凸多面体的并集。路径规划在这些潜在凸集上进行，而可逆映射则将结果路径解码回原始配置空间，同时保持相对于精细化显式安全区域的可行性。我们进一步提出了可见性引导采样（VGS，Visibility-Guided Sampling），以保持凸集在路径规划中的连通性。在二维导航、6自由度和14自由度操作环境中，ILD实现了比之前基准更广泛的覆盖、更好的集间连通性和更高的路径规划成功率，在测试时精细化后未观察到任何假阳性。在14自由度双手操作器上，我们进一步展示了实时无碰撞规划，测试时精细化能够适应在单个6自由度臂上的现实部署过程中场景几何变化。
cs.RO / 27 / 2606.12028
VICX: Generalizable Robot Manipulation via Video Generation and In-Context Operator Network
VICX：通过视频生成和上下文操作网络实现可泛化的机器人操控
Abstract
Generalizable robot manipulation requires not only task-level reasoning over unseen scenes, but also reliable grounding of visual plans into embodiment-specific execution. To bridge this gap, we propose VICX (Video generation and In-Context eXecution), a decoupled closed-loop manipulation framework. In VICX, a frozen video generation model produces vision-language-conditioned high-level visual plans, while a Video-to-Trajectory In-Context Operator Network (V2T-ICON) serves as the task-agnostic interface that grounds these plans into executable robot-state trajectories. To improve execution generalization, V2T-ICON operates on segmentation-extracted arm-only frame observations and uses retrieved image-state pairs as in-context prompts, allowing a robust and generalizable visual-to-state mapping at inference time without parameter updates. Experiments on Meta-World show that VICX supports cross-task generalization, closed-loop self-correction, and cross-embodiment transfer, demonstrating dual generalization across both task semantics and robot execution. The project webpage can be found here: https://scaling-group.github.io/vicx/.
Chinese Translation
可泛化的机器人操控不仅需要对未见场景进行任务级推理，还需要将视觉计划可靠地落实到特定的执行体上。为了解决这一问题，我们提出了VICX（视频生成与上下文执行），一个解耦的闭环操控框架。在VICX中，一个冻结的视频生成模型生成与视觉-语言条件相关的高层视觉计划，而视频到轨迹上下文操作网络（Video-to-Trajectory In-Context Operator Network，V2T-ICON）作为与任务无关的接口，将这些计划落实为可执行的机器人状态轨迹。为了提高执行的泛化能力，V2T-ICON在分割提取的仅手臂帧观察上进行操作，并使用检索到的图像-状态对作为上下文提示，从而在推理时实现稳健且可泛化的视觉到状态映射，而无需参数更新。在Meta-World上的实验表明，VICX支持跨任务泛化、闭环自我修正和跨执行体转移，展示了在任务语义和机器人执行两方面的双重泛化。项目网页可在此找到：https://scaling-group.github.io/vicx/
cs.RO / 28 / 2606.12042
KinematicRL: A Sim-to-Real Reinforcement Learning Framework For Social Navigation With Kinodynamic Feasibility
KinematicRL：一种用于社会导航的具有运动动力学可行性的仿真到现实强化学习框架
Abstract
Deep Reinforcement Learning (DRL) has shown promise for social navigation, yet its real-world deployment remains hindered by a persistent sim-to-real gap arising from simplified first-order dynamics and context-specific human state estimation pipelines. This work presents a unified framework that addresses these limitations to produce dynamically feasible navigation policies suitable for real-world deployment. First, theoretical analysis reveals that tracking error between simulated and actual robot position decays exponentially with increased control order, motivating the use of higher-order control inputs as DRL action space. A second-order control formulation tailored to differential drive robots is developed, complemented by a stochastic iterative Linear Quadratic Regulator (iLQR) that pretrains the policy via a divergence minimization objective. Second, to avoid the added system complexity of camera-LiDAR fusion, a cluster-based human tracking pipeline using only 2D LiDAR is introduced. Human detections are associated according to both spatial proximity and velocity similarity, enabling reliable differentiation of nearby pedestrians and yielding stable velocity estimates through temporal aggregation. Third, we introduce an unbiased residual gating block to balance reaction- and memory-based behaviors while handling time-varying crowd sizes, both critical for social navigation. The resulting policy, KinematicRL, consistently improves kinematic performance and adapts to varying number of detected humans. Experiments in real-world environments demonstrate that, when combined with the proposed tracking pipeline, KinematicRL can be deployed on a real differential drive robot with minimal modifications.
Chinese Translation
深度强化学习（DRL）在社会导航方面展现了良好的前景，但其在现实世界中的应用仍受到由简化的一阶动力学和特定上下文的人类状态估计管道引起的持续仿真到现实差距的限制。本研究提出了一个统一框架，旨在解决这些局限性，以生成适合现实世界部署的动态可行导航策略。首先，理论分析表明，模拟与实际机器人位置之间的跟踪误差随着控制阶数的增加而指数衰减，这激励了将高阶控制输入作为DRL的动作空间。我们开发了一种针对差分驱动机器人的二阶控制形式，并辅以随机迭代线性二次调节器（iLQR），通过最小化发散目标对策略进行预训练。其次，为避免相机-激光雷达融合所增加的系统复杂性，我们引入了一种仅使用2D激光雷达的基于聚类的人类跟踪管道。人类检测根据空间接近度和速度相似性进行关联，从而可靠地区分附近行人，并通过时间聚合获得稳定的速度估计。第三，我们引入了一种无偏残差门控模块，以平衡反应性和基于记忆的行为，同时处理随时间变化的人群规模，这对于社会导航至关重要。最终产生的策略KinematicRL在运动学性能上持续改善，并能够适应不同数量的检测到的人类。真实环境中的实验表明，当与所提出的跟踪管道结合时，KinematicRL可以在对真实差分驱动机器人进行最小修改的情况下进行部署。
cs.RO / 29 / 2606.12048
Point Cloud Segmentation for Autonomous Clip Positioning in Laparoscopic Cholecystectomy on a Phantom
在模型上进行腹腔镜胆囊切除术中自主夹具定位的点云分割
Abstract
High-risk applications in robotics, such as robot-assisted surgery, present unique challenges. These systems must be both highly precise and interpretable in order to be deployed in environments with very low tolerance for error or unsafe exploration. We present the first robotic system to demonstrate autonomous clip positioning on a physical phantom in laparoscopic surgery, one of the most common interventions in general surgery. After segmentation of a colorless point cloud from a single camera, target positions for the clips are extracted using spline interpolation, and can then be adjusted by the human operator. The segmentation model is trained on only 60 hand-labeled real point clouds, reflecting data scarcity in the surgical domain. We overcome this with a combination of pre-training on 128,000 synthetic point clouds and two novel data augmentation techniques. The motion of the end-effector to each target is visualized for the operator, satisfying the unique motion constraints of minimally-invasive surgery while ensuring that the robot's actions are verifiable and interpretable. In real robot experiments, our system localizes targets with the required precision of 0.75mm at a 95% success rate and executes autonomous clip positioning with a 100% success rate. We provide insights that are applicable to many other surgical and non-surgical tasks that require identifying and navigating to a precise target. Source code and project page: https://github.com/balazsgyenes/kirurc
Chinese Translation
机器人技术中的高风险应用，如机器人辅助手术，面临独特的挑战。这些系统必须具备高度的精确性和可解释性，以便在对错误容忍度极低或不安全探索的环境中部署。我们展示了第一个在物理模型上实现自主夹具定位的机器人系统，该系统应用于腹腔镜手术，这是普通外科中最常见的干预之一。在对来自单个摄像头的无色点云进行分割后，通过样条插值提取夹具的目标位置，并可以由人类操作员进行调整。该分割模型仅在60个手工标注的真实点云上进行训练，反映了外科领域数据稀缺的问题。我们通过在128,000个合成点云上进行预训练以及两种新颖的数据增强技术来克服这一问题。末端执行器到每个目标的运动被可视化给操作员，满足微创手术的独特运动约束，同时确保机器人的动作是可验证和可解释的。在真实的机器人实验中，我们的系统以95%的成功率在0.75mm的精度下定位目标，并以100%的成功率执行自主夹具定位。我们提供的见解适用于许多其他需要识别和导航到精确目标的外科和非外科任务。源代码和项目页面：https://github.com/balazsgyenes/kirurc
cs.RO / 30 / 2606.12070
Fibration Trees: A Unified Approach to Multi-Robot Motion Planning
纤维树：一种统一的多机器人运动规划方法
Abstract
State space projections and decompositions have emerged as powerful tools to tackle the curse of dimensionality in high-dimensional, multi-robot motion planning problems. However, existing methods lack a unified framework which seamlessly handles combinations of projections (prioritization or task-space) and decompositions (parallel or decoupled subspaces). To fill this gap, we introduce fibration trees, which are trees consisting of state spaces as nodes and fibrations as edges, whereby a fibration models a projection from a higher-dimensional space to a lower-dimensional (or simplified) space. By modeling projections as fibrations, we unify sequential prioritization, parallel decomposition, and task-space projections under a single, coherent formalism. Building on this, we develop the rapidly-exploring random fibration trees (Fibration-RRT) planner, a sampling-based motion planner that generalizes strategies from quotient-space RRT (for sequential prioritizations) and discrete RRT (for parallel decompositions), while allowing the inclusion of task-space projections. Fibration-RRT operates on user-defined fibration trees and is proven to be probabilistically complete. To test the generality and efficiency of Fibration-RRT, we provide an open-source implementation and conduct experiments on 32 scenarios using multi robot teams with up to 96 degrees of freedom. Our results indicate that Fibration-RRT efficiently solves high-dimensional problems by exploiting user-defined fibration trees, thereby establishing fibration trees as a powerful, unified framework for multi-robot motion planning.
Chinese Translation
状态空间投影和分解已成为解决高维多机器人运动规划问题中维度诅咒的强大工具。然而，现有方法缺乏一个统一的框架，无法无缝处理投影（优先级或任务空间）和分解（并行或解耦子空间）的组合。为填补这一空白，我们引入了纤维树，这是一种由状态空间作为节点和纤维作为边的树结构，其中纤维模型表示从高维空间到低维（或简化）空间的投影。通过将投影建模为纤维，我们将顺序优先级、并行分解和任务空间投影统一在一个单一的、连贯的形式主义下。在此基础上，我们开发了快速探索随机纤维树（Fibration-RRT）规划器，这是一种基于采样的运动规划器，能够将来自商空间RRT（用于顺序优先级）和离散RRT（用于并行分解）的策略进行概括，同时允许包含任务空间投影。Fibration-RRT在用户定义的纤维树上运行，并被证明是概率完备的。为了测试Fibration-RRT的通用性和效率，我们提供了一个开源实现，并在使用多机器人团队（最多96个自由度）的32个场景上进行了实验。我们的结果表明，Fibration-RRT通过利用用户定义的纤维树有效地解决了高维问题，从而确立了纤维树作为多机器人运动规划的强大统一框架。
cs.RO / 31 / 2606.12105
DAM-VLA: Decoupled Asynchronous Multimodal Vision Language Action model
DAM-VLA：解耦异步多模态视觉语言动作模型
Abstract
Vision-language-action (VLA) models inherit a shared synchronous clock from vision-language pretraining, processing every input at one rate. This is misaligned with physical interaction, where a high-frequency modality changes at hundreds of hertz, vision evolves more slowly, and language stays constant across an episode. A synchronous VLA oversamples slow modalities, undersamples fast ones, and caps action generation at the lowest effective frequency. We hypothesize that decoupling temporal processing per modality, letting each update and retain information at its own sensor rate, yields stronger representations and more robust control. We present DAM-VLA, which maintains per-modality latent buffers refreshed at sensor rates and read continuously by the action head, integrating new high-frequency modalities through gated cross-attention that leaves the pretrained backbone intact. Across seven contact-rich real-world manipulation tasks, DAM-VLA more than doubles the average success rate of the strongest synchronous baseline (95.2\% vs.\ 40.95\%) while sustaining smooth, reactive 100\,Hz control. Project website: \href{https://intuitive-robots.github.io/DAM-VLA/}{intuitive-robots.github.io/DAM-VLA/}
Chinese Translation
视觉-语言-动作（VLA）模型继承了来自视觉-语言预训练的共享同步时钟，以统一的速率处理每个输入。这与物理交互不匹配，其中高频模态以数百赫兹的速度变化，视觉演变较慢，而语言在一个情节中保持不变。同步的 VLA 过度采样慢模态，欠采样快模态，并将动作生成限制在最低有效频率。我们假设，通过对每个模态解耦时间处理，让每个模态以其自身的传感器速率更新和保留信息，可以产生更强的表示和更稳健的控制。我们提出了 DAM-VLA，它维护以传感器速率刷新每个模态的潜在缓冲区，并由动作头持续读取，通过门控交叉注意力整合新的高频模态，同时保持预训练的主干网络不变。在七个富含接触的真实世界操作任务中，DAM-VLA 的平均成功率超过最强同步基线的两倍（95.2\% 对 40.95\%），同时保持平滑、反应灵敏的 100 Hz 控制。项目网站： [intuitive-robots.github.io/DAM-VLA/](https://intuitive-robots.github.io/DAM-VLA/)
cs.RO / 32 / 2606.12109
Bridging the Morphology Gap: Adapting VLA Models to Dexterous Manipulation via Intent-Conditioned Fine-Tuning
弥合形态差距：通过意图条件微调将VLA模型适应于灵巧操作
Abstract
Vision-Language-Action (VLA) models have demonstrated remarkable zero-shot generalization in robotic manipulation, yet the vast majority of pre-trained pipelines remain strictly confined to low-DoF parallel grippers. Adapting these rich semantic priors to high-DoF dexterous hands introduces a severe morphology gap, direct end-to-end joint fine-tuning inherently causes catastrophic forgetting of spatial reasoning and acute action manifold collapse due to data scarcity. In this paper, we present InDex, a novel, data-efficient adaptation framework rooted in cross-morphology semantic inheritance. Rather than discarding the pre-trained 1-DoF parallel grasp output, we repurpose it as a continuous, macroscopic virtual grasp intent proxy to sequentialize the control topology. We implement a two-stage decoupled learning architecture: the first stage parameter-efficiently aligns the VLA backbone to predict continuous arm trajectories and the scalar grasp intent; the second stage freezes this spatial backbone and leverages an intent-conditioned denoising diffusion head to decode fine-grained joint articulations for multi-fingered end-effectors. Extensive simulation benchmarks across a suite of multi-stage, contact-rich dexterous manipulation tasks demonstrate that InDex effectively masters intricate skills with minimal demonstration data, substantially outperforming monolithic baselines while preserving the robust spatial generalizability of the original VLA prior.
Chinese Translation
视觉-语言-行动（VLA）模型在机器人操作中展示了显著的零样本泛化能力，但绝大多数预训练管道仍严格局限于低自由度（DoF）平行夹爪。将这些丰富的语义先验适应于高自由度灵巧手引入了严重的形态差距，直接的端到端联合微调由于数据稀缺而固有地导致空间推理的灾难性遗忘和急剧的动作流形崩溃。本文提出了InDex，一种基于跨形态语义继承的新颖数据高效适应框架。我们并不丢弃预训练的1自由度平行抓取输出，而是将其重新利用为连续的宏观虚拟抓取意图代理，以顺序化控制拓扑。我们实现了一个两阶段解耦学习架构：第一阶段高效地对齐VLA主干以预测连续的臂轨迹和标量抓取意图；第二阶段冻结该空间主干，并利用意图条件去噪扩散头解码多指末端执行器的细粒度关节运动。通过一系列多阶段、接触丰富的灵巧操作任务的广泛仿真基准测试，证明InDex能够以最少的演示数据有效掌握复杂技能，显著超越单一基线，同时保持原始VLA先验的强大空间泛化能力。
cs.RO / 33 / 2606.12112
PEBRE: An Open-Hardware Compute and Perception Add-On for the Pepper Robot
PEBRE：一种用于Pepper机器人的开放硬件计算与感知扩展模块
Abstract
This paper presents the design, development, and experimental verification of PEBRE, an open-hardware add-on for fast software development on the Pepper Robot. Our project enhances Pepper's computational and perception capabilities by integrating external components such as a Jetson Orin Nano, Logitech BRIO, Intel RealSense D435i, Samson UB1, and R{\O}DE VideoMicro II. Our results show that the new hardware considerably improved Pepper's perception abilities and computational power. This development contributes to the community by implementing an open hardware and open-source modular add-on to the Pepper robot and keeping this relevant research platform functional beyond its expected lifespan. With PEBRE, we aim to facilitate faster software development and more efficient integration of external components, ultimately enhancing the capabilities of the Pepper robot.
Chinese Translation
本文介绍了PEBRE的设计、开发和实验验证，PEBRE是一个用于Pepper机器人的开放硬件扩展模块，旨在加速软件开发。我们的项目通过集成外部组件，如Jetson Orin Nano、Logitech BRIO、Intel RealSense D435i、Samson UB1和R{ ext{O}}DE VideoMicro II，增强了Pepper的计算和感知能力。我们的结果表明，新硬件显著提高了Pepper的感知能力和计算性能。这一发展通过实现一个开放硬件和开源的模块化扩展，促进了社区的研究，并使这一相关研究平台在预期寿命之外保持功能。通过PEBRE，我们旨在促进更快速的软件开发和更高效的外部组件集成，最终增强Pepper机器人的能力。
cs.RO / 34 / 2606.12142
AerialClaw: An Open-Source Framework for LLM-Driven Autonomous Aerial Agents
AerialClaw：一个基于大型语言模型驱动的自主空中代理的开源框架
Abstract
Unmanned aerial vehicles (UAVs) are increasingly used in inspection, search and rescue, environmental monitoring, and emergency response. However, most UAV applications still rely on pre-defined command sequences or task-specific pipelines, where developers manually connect perception, planning, flight control, simulation, logging, and safety modules. This limits the flexibility, reproducibility, and extensibility of autonomous aerial systems. This paper presents AerialClaw, an open-source software framework that enables UAVs to operate as decision-making aerial agents rather than merely command-following platforms. Given a natural-language mission, AerialClaw allows an LLM-based agent to understand the task, maintain context, invoke executable aerial skills, observe perception and runtime feedback, and iteratively update its decisions in a closed loop. The framework adopts a modular brain-skill-runtime architecture, combining hard skills for atomic UAV operations, Markdown-based soft skills for reusable task strategies, document-driven agent state and capability boundaries, memory-driven reflection, safety-oriented runtime validation, and platform-agnostic execution adapters. AerialClaw supports lightweight mock execution, PX4 SITL with Gazebo, and AirSim-based simulation, together with a web console, pluggable model backends, example missions, simulation assets, and staged deployment scripts. By combining standardized aerial skills, document-driven agent state, memory, and closed-loop LLM decision-making, AerialClaw provides a reproducible and extensible open-source framework for building UAV systems that can interpret missions, make decisions, execute skills, and adapt their behavior from feedback.
Chinese Translation
无人机（UAV）在检查、搜索与救援、环境监测和应急响应等领域的应用日益增多。然而，大多数无人机应用仍依赖于预定义的指令序列或特定任务的流程，开发者需要手动连接感知、规划、飞行控制、仿真、日志记录和安全模块。这限制了自主空中系统的灵活性、可重复性和可扩展性。本文介绍了AerialClaw，一个开源软件框架，使无人机能够作为决策型空中代理进行操作，而不仅仅是遵循指令的平台。AerialClaw能够根据自然语言任务，使基于大型语言模型（LLM）的代理理解任务、保持上下文、调用可执行的空中技能、观察感知和运行时反馈，并在闭环中迭代更新决策。该框架采用模块化的脑-技能-运行时架构，结合了用于原子无人机操作的硬技能、基于Markdown的可重用任务策略的软技能、文档驱动的代理状态和能力边界、基于记忆的反思、安全导向的运行时验证以及平台无关的执行适配器。AerialClaw支持轻量级的模拟执行、与Gazebo的PX4 SITL以及基于AirSim的仿真，并提供了网络控制台、可插拔模型后端、示例任务、仿真资产和分阶段部署脚本。通过结合标准化的空中技能、文档驱动的代理状态、记忆和闭环的LLM决策，AerialClaw提供了一个可重复和可扩展的开源框架，用于构建能够解读任务、做出决策、执行技能并根据反馈调整行为的无人机系统。
cs.RO / 35 / 2606.12207
Intelligent Automation for Embodied Benchmark Construction: Pipelines, Embodiments, Simulators, and Trends
智能自动化在具身基准构建中的应用：管道、具身、模拟器与趋势
Abstract
Embodied intelligence now spans navigation, household assistance, manipulation, autonomous driving, aerial agents, and multimodal large-model control. This expansion has made benchmark construction a central bottleneck for reliable evaluation. Unlike static datasets, embodied benchmarks combine task specifications, environments, robot data, demonstrations, annotations, metrics, evaluation scripts, and release policies into a single evaluation system. This survey reviews the literature through a five-stage construction pipeline: requirement and task construction, data acquisition, data cleaning and annotation, benchmark suite generation and metric definition, and evaluation execution with diagnostic feedback. For each stage, the survey analyzes the transition from manual curation to traditional automation, foundation-model assistance, and agentic closed-loop workflows. It also compares qualitative construction costs across human labor, data and asset acquisition, compute and simulation, validation and debugging, governance and maintenance, and rework risk. The main conclusion is that automation does not simply reduce benchmark cost. Instead, it often shifts cost toward validation, auditability, version control, and long-term governance. Progress in embodied evaluation will therefore depend not only on larger benchmark suites, but also on construction pipelines that are diagnosable, auditable, and responsibly refreshable.
Chinese Translation
具身智能现已涵盖导航、家庭辅助、操控、自动驾驶、空中代理和多模态大模型控制。这一扩展使得基准构建成为可靠评估的核心瓶颈。与静态数据集不同，具身基准将任务规范、环境、机器人数据、演示、注释、指标、评估脚本和发布政策结合成一个单一的评估系统。本调查通过五个阶段的构建管道回顾了相关文献：需求与任务构建、数据获取、数据清理与注释、基准套件生成与指标定义，以及带有诊断反馈的评估执行。对于每个阶段，调查分析了从手动策划到传统自动化、基础模型辅助和智能闭环工作流的转变。它还比较了人力劳动、数据和资产获取、计算与模拟、验证与调试、治理与维护以及返工风险等方面的定性构建成本。主要结论是，自动化并不仅仅减少基准成本。相反，它通常将成本转移到验证、可审计性、版本控制和长期治理上。因此，具身评估的进展不仅依赖于更大的基准套件，还依赖于可诊断、可审计和负责任地可更新的构建管道。
cs.RO / 36 / 2606.12236
DrivingAgent: Design and Scheduling Agents for Autonomous Driving Systems
DrivingAgent：自主驾驶系统的设计与调度代理
Abstract
Many autonomous driving systems are increasingly incorporating foundation models to improve generalization and handle long-tail scenarios. However, this trend introduces two key challenges: (i) the manual and labor-intensive process of designing and integrating new models, and (ii) the lack of intelligent, dynamic scheduling mechanisms to meet strict real-time constraints. While Large Language Model (LLM)-based agents offer a promising avenue for automation, existing frameworks are ill-suited for autonomous driving. Specifically, they fail to distinguish between the fundamentally different requirements of system design and real-time scheduling, treat modules as opaque black boxes, and are not designed for continuous operation. To address these limitations, we propose DrivingAgent, a novel agent framework tailored to the dual challenges of autonomous driving system design and scheduling. In the design phase, DrivingAgent automates module development by interpreting system architecture, generating code, and validating modules via super-network training. In the scheduling phase, it employs a lightweight LLM trained with reinforcement learning to dynamically orchestrate system modules in real time, supported by a structured memory that integrates long-term storage with timestamped short-term context. Experimental results demonstrate that DrivingAgent achieves a superior speed--accuracy trade-off on both the nuScenes and Bench2Drive benchmarks.
Chinese Translation
许多自主驾驶系统越来越多地采用基础模型，以提高泛化能力并处理长尾场景。然而，这一趋势带来了两个主要挑战：（i）设计和集成新模型的手动和劳动密集型过程，以及（ii）缺乏智能的动态调度机制以满足严格的实时约束。虽然基于大型语言模型（LLM）的代理提供了自动化的有希望的途径，但现有框架并不适合自主驾驶。具体而言，它们未能区分系统设计和实时调度的根本不同要求，将模块视为不透明的黑箱，并且未设计为连续运行。为了解决这些局限性，我们提出了DrivingAgent，这是一种新颖的代理框架，专门针对自主驾驶系统设计和调度的双重挑战。在设计阶段，DrivingAgent通过解释系统架构、生成代码和通过超网络训练验证模块来自动化模块开发。在调度阶段，它采用经过强化学习训练的轻量级LLM，实时动态协调系统模块，支持通过结构化内存将长期存储与带时间戳的短期上下文相结合。实验结果表明，DrivingAgent在nuScenes和Bench2Drive基准测试中实现了优越的速度-准确性权衡。
cs.RO / 37 / 2606.12299
Learning What to Say to Your VLA: Mostly Harmless Vision Language Action Model Steering
学习如何与您的视觉语言行动模型对话：基本无害的视觉语言行动模型引导
Abstract
Vision-Language-Action (VLA) models provide a natural language interface to robot control, but the mapping from language to behavior is often brittle and unintuitive: semantically similar instructions can induce drastically different behaviors, while some capabilities may not be elicitable through prompting alone. As a result, both human instructions and zero-shot language models can fail to reliably steer VLAs toward successful task execution. In this work, we propose a framework that interactively searches for language sequences that improve closed-loop VLA task performance, distills these sequences into a test-time language feedback policy (LFP), and learns an improvement head that predicts when language steering will improve performance. We conformalize this improvement head to prevent harmful steering interventions, where the LFP decreases task performance relative to the original instruction on out-of-distribution scenarios. Crucially, our approach operates on arbitrary frozen pre-trained VLAs, requiring neither access to the original training distribution nor fine-tuning of the underlying model. On seen environments, our conformalized LFP improves base VLA performance by 24.7% in simulation and 65.0% in hardware. On visual and semantic perturbations, our conformalized LFP has strong harmlessness guarantees, and produces recovery behaviors not observed with open-loop prompting.
Chinese Translation
视觉语言行动（VLA）模型为机器人控制提供了自然语言接口，但语言与行为之间的映射往往脆弱且不直观：语义相似的指令可能引发截然不同的行为，而某些能力可能仅通过提示无法引导。因此，无论是人类指令还是零样本语言模型，都可能无法可靠地引导VLA成功执行任务。在本研究中，我们提出了一个框架，交互式地搜索能够改善闭环VLA任务性能的语言序列，将这些序列提炼为测试时语言反馈策略（LFP），并学习一个改进头，以预测何时语言引导会改善性能。我们对该改进头进行合规化，以防止有害的引导干预，即在分布外场景中，LFP相对于原始指令降低任务性能。至关重要的是，我们的方法适用于任意冻结的预训练VLA，无需访问原始训练分布或对基础模型进行微调。在已见环境中，我们的合规化LFP在仿真中提高了基础VLA性能24.7%，在硬件中提高了65.0%。在视觉和语义扰动下，我们的合规化LFP具有强大的无害性保证，并产生了在开放式提示中未观察到的恢复行为。
cs.RO / 38 / 2606.12349
Traceable Virtual Sea Trials in the Marine Robotics Unity Simulator for Manoeuvring Assessment of Unmanned Surface Vehicles
用于无人水面艇操纵评估的海洋机器人Unity模拟器中的可追溯虚拟海试
Abstract
Accurate identification of hydrodynamic derivatives is essential for control and navigation of Unmanned Surface Vehicles (USVs), but high-fidelity manoeuvring data from physical sea trials are constrained by cost and safety. Turning Circle (TC) and Zig-Zag (ZZ) trials remain fundamental to IMO and ITTC assessment procedures. This paper extends the Marine Robotics Unity Simulator (MARUS) by introducing a standardised Virtual Sea Trial framework for automated execution and data generation of TC/ZZ manoeuvres, with traceable command-actuation logging, system-identification (SI)-focused data conditioning, and automated extraction of IMO/ITTC-aligned manoeuvring metrics. A key contribution is a dedicated TC/ZZ data acquisition and post-processing pipeline, improving the repeatability and auditability of simulator-based manoeuvres while producing SI-ready datasets for hydrodynamic-derivative identification and digital-twin workflows. Another feature is explicit command-execution separation for differential-thrust steering, where inputs are recorded as ordered rudder-equivalent commands and realised actuation is logged as an execution-level proxy derived from applied thrust. Case-study results demonstrate repeatable and compliant manoeuvre behaviour. For TC tests, the normalised advance differs by approximately 3.9 percent between port and starboard sides, while the tactical diameter differs by approximately 4.6 to 4.7 percent. For ZZ tests, first and second overshoot excesses remain below 1 degree for both +/- 10 degree and +/- 20 degree manoeuvres, satisfying IMO criteria, while peak yaw rates range from approximately 4.1 to 5.8 deg/s. Overall, the framework provides a repeatable and auditable virtual sea-trial workflow for generating IMO/ITTC-aligned datasets and supporting system identification, hydrodynamic-derivative estimation, and digital-twin calibration.
Chinese Translation
准确识别水动力导数对于无人水面艇（USVs）的控制和导航至关重要，但来自物理海试的高保真操纵数据受到成本和安全性的限制。转弯圈（TC）和之字形（ZZ）试验仍然是国际海事组织（IMO）和国际船舶与海洋工程委员会（ITTC）评估程序的基础。本文通过引入标准化的虚拟海试框架，扩展了海洋机器人Unity模拟器（MARUS），实现了TC/ZZ操纵的自动执行和数据生成，具备可追溯的指令-执行日志、以系统识别（SI）为中心的数据处理，以及IMO/ITTC对齐的操纵指标的自动提取。一个关键贡献是专门的TC/ZZ数据采集和后处理管道，提高了基于模拟器的操纵的可重复性和可审计性，同时生成了适用于水动力导数识别和数字双胞胎工作流的SI准备数据集。另一个特点是差分推力转向的明确指令执行分离，其中输入被记录为有序的舵等效指令，而实际的执行被记录为从施加推力中导出的执行级代理。案例研究结果表明操纵行为的可重复性和合规性。在TC测试中，正常化的前进距离在左舷和右舷之间的差异约为3.9%，而战术直径的差异约为4.6%到4.7%。在ZZ测试中，第一次和第二次超调的过量均保持在1度以下，无论是+/- 10度还是+/- 20度的操纵，均满足IMO标准，而峰值偏航率范围约为4.1到5.8度/秒。总体而言，该框架提供了一个可重复和可审计的虚拟海试工作流，用于生成与IMO/ITTC对齐的数据集，并支持系统识别、水动力导数估计和数字双胞胎校准。
cs.RO / 39 / 2606.12352
CHORUS: Decentralized Multi-Embodiment Collaboration with One VLA Policy
CHORUS：基于单一VLA策略的去中心化多体现协作
Abstract
Multi-robot collaboration allows robots to efficiently take on a wide range of tasks, from moving a couch through a doorway to assembling structures on a construction site. However, achieving such coordination in mobile multi-robot settings remains challenging: centralized methods conditioned on the combined observations of a team scale poorly with team size, and decentralized methods that train one policy per robot often require explicit alignment procedures or information sharing at inference time to overcome partial observability. Our key insight is that the visuomotor priors of pretrained vision-language-action (VLA) models should enable reactive, decentralized collaboration from each robot's local observations alone, without these inference-time assumptions. We propose CHORUS, a framework that adapts a single VLA backbone to control diverse, multi-robot teams. At inference time, each robot runs an independent copy of CHORUS, conditioned only on its own observations and a robot-identifying prompt. In real-world experiments including mobile tape measurement, library book handovers, and laundry basket lifting, CHORUS achieves a 64% point improvement over decentralized, from-scratch models, improves reactivity to teammate behavior by 40% points, and outperforms centralized baselines. Together, these results show that a shared VLA backbone is capable of achieving decentralized multi-robot collaboration, without per-robot policies or inter-robot communication at inference.
Chinese Translation
多机器人协作使得机器人能够高效地完成各种任务，从通过门口搬运沙发到在建筑工地上组装结构。然而，在移动多机器人环境中实现这种协调仍然具有挑战性：基于团队综合观察的中心化方法在团队规模扩大时表现不佳，而为每个机器人训练一个策略的去中心化方法通常需要在推理时进行显式对齐程序或信息共享，以克服部分可观测性。我们的关键见解是，预训练的视觉-语言-动作（VLA）模型的视觉运动先验应能使每个机器人仅凭其本地观察实现反应式的去中心化协作，而无需这些推理时的假设。我们提出了CHORUS，一个将单一VLA骨干网络适配于控制多样化多机器人团队的框架。在推理时，每个机器人独立运行CHORUS的副本，仅基于其自身观察和一个机器人识别提示。在包括移动胶带测量、图书馆书籍交接和洗衣篮提升的实际实验中，CHORUS相比于去中心化的从零开始模型实现了64%的性能提升，提升了对队友行为的反应性40个百分点，并且超越了中心化基线。综上所述，这些结果表明，共享的VLA骨干网络能够实现去中心化的多机器人协作，而无需在推理时为每个机器人制定策略或进行机器人间通信。
cs.RO / 40 / 2606.12365
Ambient Diffusion Policy: Imitation Learning from Suboptimal Data in Robotics
环境扩散策略：在机器人领域从次优数据中进行模仿学习
Abstract
We propose Ambient Diffusion Policy, a simple and principled method for imitation learning from suboptimal data in robotics. High-quality, task-specific robot data is expensive and time-consuming to collect, while suboptimal datasets with lower-quality or out-of-distribution demonstrations are abundant. Existing methods that co-train on both data sources in robotics often fail to separate the meaningful and the harmful features in the suboptimal samples. In contrast, our method extracts only the useful features by introducing a new axis to co-training in robotics: noise-dependent data usage. Ambient Diffusion Policy restricts the contribution of suboptimal data during training to only the high and low diffusion times. To rigorously justify our approach, we first observe that robot action data exhibits a spectral power law. This induces two important properties on the optimal Diffusion Policy that we exploit: a global-to-local hierarchy and locality. We theoretically formalize this discussion using a simplified model. Our experiments validate Ambient Diffusion Policy on four types of suboptimal action data (noisy trajectories, sim-to-real gap, task mismatch, and large-scale data mixtures) across six tasks. The results show that it effectively learns from arbitrary sources of suboptimal data. Notably, it outperforms existing co-training baselines by up to 33% when scaled to Open X-Embodiment - a large dataset with heterogeneous data quality and unstructured distribution shifts. Overall, Ambient Diffusion Policy increases the utility of suboptimal demonstrations and expands the set of usable data sources in robotics.
Chinese Translation
我们提出了环境扩散策略（Ambient Diffusion Policy），这是一种简单而有原则的方法，用于在机器人领域从次优数据中进行模仿学习。高质量、特定任务的机器人数据收集成本高且耗时，而低质量或分布外演示的次优数据集则相对丰富。现有的在机器人领域同时训练这两种数据源的方法往往无法有效区分次优样本中的有意义特征和有害特征。相比之下，我们的方法通过引入一个新的共同训练轴线：噪声依赖的数据使用，来提取仅有用的特征。环境扩散策略在训练过程中限制次优数据的贡献，仅限于高扩散时间和低扩散时间。为了严格证明我们的方法，我们首先观察到机器人动作数据呈现谱功率法则。这为我们利用的最优扩散策略引入了两个重要特性：全局到局部的层次结构和局部性。我们使用简化模型对这一讨论进行了理论化形式化。我们的实验在六个任务中验证了环境扩散策略在四种类型的次优动作数据（噪声轨迹、模拟到真实的差距、任务不匹配和大规模数据混合）上的有效性。结果表明，它能够有效地从任意来源的次优数据中学习。值得注意的是，在扩展到Open X-Embodiment（一个具有异构数据质量和非结构化分布变化的大型数据集）时，它的性能比现有的共同训练基线提高了多达33%。总体而言，环境扩散策略提高了次优演示的效用，并扩展了机器人领域可用数据源的集合。
cs.RO / 41 / 2606.12366
APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies
APT：动作专家预训练提升视觉-语言-动作策略的指令泛化能力
Abstract
Vision-Language-Action (VLA) models that couple pretrained Vision-Language Models (VLMs) with continuous action experts have achieved strong manipulation performance, yet generalization to out-of-distribution (OOD) language instructions remains poor. A known challenge is the structural imbalance in VLA data, where language is far less diverse than visual and action content, making policies prone to visual shortcuts. While discrete-action methods mitigate this through vision-language co-training, continuous action experts lack such protection: they start from random initialization and learn entirely from imbalanced data, producing noisy gradients that corrupt the VLM and fail to exploit its language capability. We address this from a Bayesian perspective, factorizing the policy into a language-agnostic Vision-Action (VA) prior and a language-conditioned VLA likelihood, and propose APT, a two-stage training method emphasizing Action expert PreTraining. In Stage 1, the action expert is pretrained as a VA prior on vision-action pairs from a frozen VLM, bypassing the language imbalance. In Stage 2, language tokens are injected through a gated fusion mechanism that integrates VLM features while preserving the learned visuomotor prior. APT applies to mainstream VLA architectures, including the $\pi$ and GR00T-style architectures. Comprehensive experiments validate that APT achieves consistent gains on unseen instructions and compositional tasks. Project Page: https://xukechun.github.io/papers/APT/
Chinese Translation
视觉-语言-动作（VLA）模型将预训练的视觉-语言模型（VLM）与连续动作专家相结合，已取得强大的操作性能，但在面对分布外（OOD）语言指令时的泛化能力仍然较差。一个已知的挑战是VLA数据中的结构不平衡，语言的多样性远低于视觉和动作内容，使得策略容易依赖视觉捷径。虽然离散动作方法通过视觉-语言共同训练来缓解这一问题，但连续动作专家缺乏这样的保护：它们从随机初始化开始，仅依赖不平衡数据进行学习，产生的噪声梯度会破坏VLM，并未能充分利用其语言能力。我们从贝叶斯的角度出发，将策略分解为与语言无关的视觉-动作（VA）先验和条件于语言的VLA似然，并提出APT，一种强调动作专家预训练的两阶段训练方法。在第一阶段，动作专家作为VA先验在来自冻结VLM的视觉-动作对上进行预训练，从而绕过语言不平衡。在第二阶段，通过门控融合机制注入语言标记，该机制在保留已学习的视觉运动先验的同时整合VLM特征。APT适用于主流的VLA架构，包括$ ext{π}$和GR00T风格架构。全面的实验验证了APT在未见指令和组合任务上取得了一致的提升。项目页面：https://xukechun.github.io/papers/APT/
cs.RO / 42 / 2606.12372
UniIntervene: Agentic Intervention for Efficient Real-World Reinforcement Learning
UniIntervene：高效现实世界强化学习的自主干预
Abstract
Human-in-the-loop reinforcement learning (HiL-RL) has emerged as an effective paradigm for real-world robotic manipulation, enabling online policy improvement with human guidance. However, current HiL-RL frameworks remain intervention-intensive, relying on frequent human corrections to redirect the policy out of unproductive exploration, which incurs high labor cost and limits real-world scalability. To address this, we propose UniIntervene, an agentic intervention model that detects unproductive exploration and autonomously recovers the policy toward high-value states, taking over the bulk of interventions from human operators. Specifically, UniIntervene first performs future-conditioned action-value estimation, predicting the latent consequence of the current action and evaluating its induced value, which provides a more stable progress signal. Building on this, a temporal value-risk critic aggregates recent value dynamics and triggers intervention when the estimated value exhibits sustained stagnation or degradation. When intervention is required, UniIntervene retrieves a high-value recovery target from a memory of past intervention episodes and produces executable corrective actions through a goal-conditioned recovery policy. In this way, UniIntervene turns intervention from passive human correction into a value-aware recovery process for efficient real-world RL. Extensive experiments on diverse real-world manipulation tasks demonstrate that UniIntervene improves the average success rate by 8.6% while reducing human interventions by 57% relative to state-of-the-art HiL-RL baselines.
Chinese Translation
人机协同强化学习（HiL-RL）已成为现实世界机器人操作的有效范式，使得在人工指导下进行在线策略改进成为可能。然而，目前的HiL-RL框架仍然需要大量干预，依赖频繁的人类纠正来引导策略脱离无效探索，这导致高劳动成本并限制了在现实世界中的可扩展性。为了解决这个问题，我们提出了UniIntervene，一种自主干预模型，能够检测无效探索并自主恢复策略朝向高价值状态，从而承担大部分人类操作员的干预工作。具体而言，UniIntervene首先执行未来条件下的动作价值估计，预测当前动作的潜在后果并评估其引发的价值，从而提供更稳定的进展信号。在此基础上，时间价值风险评估器聚合近期的价值动态，并在估计的价值持续停滞或下降时触发干预。当需要干预时，UniIntervene从过去干预情节的记忆中检索高价值恢复目标，并通过目标条件的恢复策略生成可执行的纠正动作。通过这种方式，UniIntervene将干预从被动的人类纠正转变为一种以价值为导向的恢复过程，以实现高效的现实世界强化学习。在多样的现实世界操作任务上的广泛实验表明，UniIntervene将平均成功率提高了8.6%，同时相对于最先进的HiL-RL基线减少了57%的人类干预。
cs.RO / 43 / 2606.12374
Semantically-Aware Diver Activity Recognition Framework for Effective Underwater Multi-Human-Robot Collaboration
语义感知潜水员活动识别框架用于有效的水下多人人机协作
Abstract
Effective multi-human-robot collaboration is essential for expanding human-led operations in the challenging and high-risk underwater environment. For autonomous underwater vehicles (AUVs) to become true teammates, they must be able to comprehend their surroundings and recognize a diver's activities to offer assistance and ensure safety. Towards this goal, we introduce DAR-Net, a novel transformer-based framework that analyzes complex underwater scenes to classify diver activities. Our contribution lies in a semantically guided learning formulation that couples transformer-based temporal reasoning with pixel-level scene supervision. This multi-loss training strategy explicitly aligns global activity recognition with local human-robot interaction semantics, which is particularly critical in low-visibility underwater conditions. To address the significant challenge of data scarcity in this domain, we present the first-ever Underwater Diver Activity (UDA) dataset, a foundational resource containing over 2,600 annotated images with pixel-level masks. Through rigorous experimental evaluations in a controlled environment, we demonstrate that DAR-Net achieves promising accuracy in recognizing six distinct diver activities, outperforming state-of-the-art models. While this dataset provides a crucial baseline, our work serves as a pioneering step, laying the groundwork for future research and facilitating the development of more intelligent, collaborative underwater robotic systems.
Chinese Translation
有效的多人人机协作对于在具有挑战性和高风险的水下环境中扩展人类主导的操作至关重要。为了使自主水下航行器（AUV）成为真正的队友，它们必须能够理解周围环境并识别潜水员的活动，以提供帮助并确保安全。为此，我们提出了DAR-Net，一个基于变换器的创新框架，分析复杂的水下场景以分类潜水员活动。我们的贡献在于一种语义引导的学习形式，将基于变换器的时间推理与像素级场景监督相结合。这种多损失训练策略明确地将全局活动识别与局部人机交互语义对齐，这在低能见度的水下条件下尤为重要。为了应对这一领域数据稀缺的重大挑战，我们首次提出了水下潜水员活动（UDA）数据集，这是一个基础资源，包含超过2600张带有像素级掩码的标注图像。通过在受控环境中的严格实验评估，我们证明了DAR-Net在识别六种不同潜水员活动方面达到了令人满意的准确性，超越了最先进的模型。虽然该数据集提供了一个重要的基准，但我们的工作作为开创性的一步，为未来的研究奠定了基础，并促进了更智能的水下机器人系统的开发。
cs.RO / 44 / 2606.12402
DIRECT: When and Where Should You Allocate Test-Time Compute in Embodied Planners?
DIRECT：在具身规划者中何时何地分配测试时间计算？
Abstract
Vision-Language Models (VLMs) are increasingly deployed as high-level planners for embodied agents, with an emerging strategy of scaling test-time compute to improve capability. However, we observe that doing so increases latency, token usage, and FLOPs while yielding uneven, often diminishing gains in downstream success, limiting where embodied agents can be deployed. We argue that choosing when and where to spend test-time compute is central to bringing frontier performance to the real world. We introduce DIRECT, a routing framework that uses multimodal scene context to allocate compute per prompt, improving the success--cost Pareto frontier over fixed model selection. Across three dominant scaling axes, namely chain-of-thought depth, model size, and memory history, our experiments on VLABench and RoboMME show that test-time compute is not a uniform lever: different axes yield qualitatively distinct capability gains. We validate these insights on a physical Franka arm in a DROID setup spanning zero-shot manipulation and long-horizon chaining, where our router matches or exceeds a stronger model's success rate at up to 65% lower average latency. Ultimately, our results show that naively scaling test-time compute is wasteful, and that DIRECT can provide frontier-level embodied planning in robotic systems at a fraction of the cost. Project page can be found at jadee-dao.github.io/direct/.
Chinese Translation
视觉-语言模型（VLMs）越来越多地被用于作为具身代理的高级规划者，采用在测试时间扩展计算以提升能力的策略。然而，我们观察到，这样做会增加延迟、令牌使用和浮点运算次数（FLOPs），同时在下游成功率上产生不均匀且往往递减的收益，从而限制了具身代理的部署范围。我们认为，选择何时何地花费测试时间计算是将前沿性能带入现实世界的关键。我们提出了DIRECT，这是一种路由框架，利用多模态场景上下文为每个提示分配计算，从而改善成功-成本帕累托前沿，相较于固定模型选择。在三个主要的扩展轴上，即思维链深度、模型大小和记忆历史，我们在VLABench和RoboMME上的实验表明，测试时间计算并不是一个统一的杠杆：不同的轴带来了质的能力提升。我们在DROID设置中对物理Franka手臂验证了这些见解，该设置涵盖了零-shot操作和长时间链式操作，我们的路由器在平均延迟降低高达65%的情况下，成功率与更强模型相当或更高。最终，我们的结果表明，简单地扩展测试时间计算是浪费的，而DIRECT可以以更低的成本在机器人系统中提供前沿水平的具身规划。项目页面可在 jadee-dao.github.io/direct/ 找到。
cs.RO / 45 / 2606.12403
World Pilot: Steering Vision-Language-Action Models with World-Action Priors
世界引导：利用世界-动作先验引导视觉-语言-动作模型
Abstract
Vision-Language-Action (VLA) models inherit semantic grounding from large-scale pretraining and perform competently across in-distribution manipulation tasks. This grounding, however, is built on static image-text pairs, whereas manipulation is a continuous, contact-rich process whose dynamics this pretraining cannot capture. We present World Pilot, a VLA framework that augments the policy with priors from a World-Action Model (WAM), routed into the decision chain through two complementary pathways. Latent Steering conditions the perception layer on a scene-evolution latent, and Action Steering supplies an anticipated trajectory as a motion prior to the action generator. Together the two priors equip the VLA with an anticipated view of the scene and a trajectory-level motion hint alongside its semantic conditioning, and the scene-evolution prior remains effective even when supplied by a video-pretrained world model that has not been action-post-trained. World Pilot attains a state-of-the-art Total success rate of 84.7% on the LIBERO-Plus zero-shot OOD benchmark and the highest success rate on every real-robot setting across four manipulation tasks, with the largest margins under shifts in viewpoint, geometry, deformable state, and pose. Project Website: https://world-pilot.github.io/
Chinese Translation
视觉-语言-动作（VLA）模型从大规模预训练中继承了语义基础，并在分布内的操作任务中表现出色。然而，这种基础是建立在静态的图像-文本对之上的，而操作则是一个连续的、接触丰富的过程，其动态特性无法通过这种预训练捕捉到。我们提出了世界引导（World Pilot），这是一个VLA框架，通过来自世界-动作模型（World-Action Model, WAM）的先验信息增强策略，并通过两条互补路径融入决策链。潜在引导（Latent Steering）将感知层条件化于场景演变的潜在变量，而动作引导（Action Steering）则为动作生成器提供了作为运动先验的预期轨迹。这两个先验共同为VLA提供了对场景的预期视图和轨迹级运动提示，同时结合其语义条件，即使在未经过动作后训练的视频预训练世界模型提供的情况下，场景演变先验依然有效。世界引导在LIBERO-Plus零-shot OOD基准上达到了84.7%的最新总成功率，并在四个操作任务的每个真实机器人设置中取得了最高成功率，在视角、几何形状、可变状态和姿态的变化下具有最大的优势。项目网站：https://world-pilot.github.io/
cs.RO / 46 / 2606.12406
FACTR 2: Learning External Force Sensing for Commodity Robot Arms Improves Policy Learning
FACTR 2：学习外部力感知以改善商品机器人臂的策略学习
Abstract
Contact-rich manipulation requires force sensitivity, but many robot arms lack dedicated force sensors due to their high cost. We present Neural External Torque Estimation (NEXT), a data-driven method that estimates external joint torques without needing any dedicated force sensors. NEXT trains in 1 minute from only 10 minutes of free-motion data, yet achieves estimates comparable to dedicated joint-torque sensors. NEXT enables force-feedback teleoperation on low-cost arms and improves policy learning through Force-Informed Re-Sampling Training (FIRST), which up-samples pre-contact and contact segments during behavior cloning. Across five long-horizon tasks, FIRST outperforms prior force-aware policies by over 17% in task progress. Together, NEXT and FIRST bring force-aware teleoperation and policy learning to off-the-shelf robots without additional sensing hardware. Video results and code are available at https://jasonjzliu.com/factr2
Chinese Translation
接触丰富的操作需要力敏感性，但许多机器人臂由于高成本而缺乏专用的力传感器。我们提出了神经外部扭矩估计（Neural External Torque Estimation, NEXT），这是一种数据驱动的方法，可以在不需要任何专用力传感器的情况下估计外部关节扭矩。NEXT仅需从10分钟的自由运动数据中训练1分钟，便能实现与专用关节扭矩传感器相当的估计。NEXT使低成本机器人臂能够进行力反馈遥操作，并通过力感知重采样训练（Force-Informed Re-Sampling Training, FIRST）改善策略学习，该方法在行为克隆过程中对接触前和接触阶段进行上采样。在五个长时间任务中，FIRST在任务进展方面超过了先前的力感知策略超过17%。NEXT和FIRST共同将力感知遥操作和策略学习引入现成机器人，而无需额外的传感硬件。视频结果和代码可在 https://jasonjzliu.com/factr2 获取。
cs.CV / 1 / 2606.11221
LAST: Bridging Vision-Language and Action Manifolds via Gromov-Wasserstein Alignment
LAST：通过 Gromov-Wasserstein 对齐桥接视觉-语言与动作流形
Abstract
We take a Gromov-Wasserstein perspective on Vision-Language-Action (VLA) learning, where the goal is to make the relational geometry of action representations compatible with the semantic geometry of VL embeddings. However, this alignment is non-trivial due to the mathematical heterogeneity between the domains: the semantic space of vision-language is topologically linear and isotropic, whereas the physical manifold of robotic action is non-Euclidean and anisotropic. Their disjoint metric structures render direct regression ill-posed. To resolve this incompatibility, we introduce LAST (Lie-algebraic Action Space Tokenizer), which reconstructs the action space to establish local metric compatibility with the VL modality via a two-stage transformation: (1) Global Topological Linearization: linearizing the action manifold via Lie-algebraic mapping, converting trajectories into a fixed-length, physically additive representation. (2) Local Metric Discretization: hierarchically discretizing the representation into schemas and whitened residuals, yielding approximately isotropic local charts that are statistically aligned with the semantic metric. By resolving the structural mismatch at both global and local levels, LAST enables VLA models with superior convergence and generalizability.
Chinese Translation
我们从 Gromov-Wasserstein 的角度探讨视觉-语言-动作（VLA）学习，其目标是使动作表示的关系几何与视觉-语言嵌入的语义几何相兼容。然而，由于领域之间的数学异质性，这种对齐并非易事：视觉-语言的语义空间在拓扑上是线性和各向同性的，而机器人的动作流形则是非欧几里得和各向异性的。它们不相交的度量结构使得直接回归变得不适定。为了解决这种不兼容性，我们引入了 LAST（Lie-algebraic Action Space Tokenizer），该方法重构动作空间，以通过两阶段变换与视觉-语言模态建立局部度量兼容性：（1）全局拓扑线性化：通过李代数映射对动作流形进行线性化，将轨迹转换为固定长度、物理可加的表示。（2）局部度量离散化：将表示分层离散为模式和白化残差，生成与语义度量统计对齐的近似各向同性局部图。通过在全局和局部层面解决结构不匹配问题，LAST 使 VLA 模型具备更优的收敛性和泛化能力。
cs.CV / 2 / 2606.11231
CFCamo: A Counterfactual Detect-or-Abstain Framework for Camouflaged Object Detection
CFCamo：一种针对伪装物体检测的反事实检测或放弃框架
Abstract
Vision-language reinforcement learning has recently shown strong target-present localization for camouflaged object detection (COD). Yet localization is only one side of the decision: when the agent faces an ordinary image with no camouflaged target, will it still claim that a camouflaged object exists? Standard COD training and evaluation data are positive-only, so agents optimized under this setting can acquire an over-detect bias, a task-specific form of object hallucination that standard COD evaluation leaves unmeasured. To quantify this target-absent behavior, we construct Counterfactual COD (CF-COD), a paired benchmark that removes the camouflaged target from each held-out COD evaluation image while preserving a plausible background. CF-COD evaluates whether a model detects the target on the original image and abstains on the target-absent counterfactual, summarized by Pair Accuracy (PA). We further introduce CFCamo, a paired counterfactual framework for COD with abstention. For training, CFCamo optimizes a Qwen3-VL-4B-Instruct agent with Counterfactual Sequence Policy Optimization (CSPO), which samples paired original-counterfactual rollouts and uses a Counterfactual Paired Reward (CPR) to couple original-image detection with counterfactual abstention. On CAMO-test, CFCamo improves S_alpha by +3.7 pp over the prior RL-based COD baseline; across CF-COD, it reaches 80.0-90.8% PA. Ablations show that removing counterfactual coupling reduces PA to 1.4-5.2% despite strong target-present COD scores, showing that target-present evaluation alone does not characterize detect-or-abstain behavior. Overall, these results indicate that CFCamo improves COD agents by coupling target-present detection with target-absent abstention, rather than merely strengthening target-present localization. Code and data are available at https://github.com/suhang2000/CFCamo.
Chinese Translation
近年来，视觉-语言强化学习在伪装物体检测（COD）方面表现出了强大的目标存在定位能力。然而，定位仅是决策的一方面：当代理面对一幅没有伪装目标的普通图像时，它是否仍会声称存在伪装物体？标准的COD训练和评估数据仅包含正样本，因此在这种设置下优化的代理可能会产生过度检测偏差，这是一种特定任务的物体幻觉形式，而标准COD评估未能对此进行测量。为了量化这种目标缺失行为，我们构建了反事实COD（CF-COD），这是一个配对基准，它从每个保留的COD评估图像中去除了伪装目标，同时保留了合理的背景。CF-COD评估模型在原始图像上是否检测到目标，并在目标缺失的反事实上选择放弃，其结果通过配对准确率（Pair Accuracy, PA）进行总结。我们进一步引入CFCamo，这是一个针对COD的配对反事实框架，支持放弃。在训练过程中，CFCamo通过反事实序列策略优化（Counterfactual Sequence Policy Optimization, CSPO）优化Qwen3-VL-4B-Instruct代理，该方法对配对的原始-反事实回放进行采样，并使用反事实配对奖励（Counterfactual Paired Reward, CPR）将原始图像检测与反事实放弃相结合。在CAMO-test上，CFCamo相较于之前的基于强化学习的COD基线提高了3.7个百分点的S_alpha；在CF-COD上，其配对准确率达到80.0-90.8%。消融实验表明，去除反事实耦合会使配对准确率降低至1.4-5.2%，尽管目标存在的COD得分较高，这表明仅通过目标存在评估无法表征检测或放弃行为。总体而言，这些结果表明CFCamo通过将目标存在检测与目标缺失放弃相结合，改善了COD代理，而不仅仅是增强了目标存在的定位能力。代码和数据可在 https://github.com/suhang2000/CFCamo 获取。
cs.CV / 3 / 2606.11233
OSCS-SupCon: Orthogonal Sigmoid-based Common and Style Supervised Contrastive Learning for Robust Feature Disentanglement
OSCS-SupCon：基于正交Sigmoid的共同与风格监督对比学习用于鲁棒特征解耦
Abstract
Supervised Contrastive Learning (SupCon) has achieved strong performance by explicitly modeling pairwise relationships among samples. However, existing SupCon-based methods suffer from two key limitations: negative-sample dilution induced by the standard InfoNCE loss, and feature-space entanglement caused by the lack of explicit constraints separating category-relevant (common) and category-irrelevant (style) features. These limitations reduce feature discriminability and generalization ability. To address these issues, we propose OSCS-SupCon (Orthogonal Sigmoid-based Common and Style Supervised Contrastive Learning), a unified framework that combines a sigmoid-based pairwise contrastive objective with explicit orthogonality constraints. Specifically, we introduce a sigmoid-based contrastive loss with two learnable parameters, temperature and bias, which adaptively modulate pairwise decision boundaries and alleviate negative-sample dilution. Furthermore, we enforce orthogonality between common and style feature subspaces via a linear projection with ReLU nonlinearity, thereby reducing feature overlap and improving disentanglement of style-irrelevant representations. Extensive experiments on six benchmark datasets demonstrate that OSCS-SupCon consistently outperforms state-of-the-art supervised contrastive learning methods across multiple backbone architectures. In particular, on the fine-grained CUB200-2011 dataset with a ResNet-18 backbone, the proposed method achieves a 3.4% improvement in classification accuracy over CS-SupCon, highlighting its robustness and generalization capability. Ablation studies further confirm the effectiveness of each component.
Chinese Translation
监督对比学习（SupCon）通过明确建模样本之间的成对关系，实现了强大的性能。然而，现有的基于SupCon的方法存在两个主要限制：标准InfoNCE损失引起的负样本稀释，以及缺乏明确约束以区分与类别相关（共同）和与类别无关（风格）特征所导致的特征空间纠缠。这些限制降低了特征的可区分性和泛化能力。为了解决这些问题，我们提出了OSCS-SupCon（基于正交Sigmoid的共同与风格监督对比学习），这是一个统一框架，将基于Sigmoid的成对对比目标与明确的正交约束相结合。具体而言，我们引入了一种具有两个可学习参数（温度和偏置）的基于Sigmoid的对比损失，能够自适应调节成对决策边界并缓解负样本稀释。此外，我们通过具有ReLU非线性的线性投影强制共同特征和风格特征子空间之间的正交性，从而减少特征重叠并改善与风格无关表示的解耦。在六个基准数据集上的广泛实验表明，OSCS-SupCon在多个主干架构上始终优于最先进的监督对比学习方法。特别是在使用ResNet-18主干的细粒度CUB200-2011数据集上，所提方法在分类准确率上比CS-SupCon提高了3.4%，突显了其鲁棒性和泛化能力。消融研究进一步确认了每个组件的有效性。
cs.CV / 4 / 2606.11269
Traits Run Deeper: Trait-Specific Asymmetric Fusion for Personality Assessment
特征更深：用于个性评估的特征特定非对称融合
Abstract
Personality assessment aims to infer stable personality traits from dynamic behaviors across language, voice, and facial cues. Since different personality dimensions are revealed through distinct behavioral perspectives, modeling trait-specific evidence is challenging. However, most existing approaches adopt a uniform multimodal fusion strategy across all dimensions, assuming identical modality contributions. This overlooks trait-specific modality preferences and introduces cross-modal interference. To address this issue, we propose a novel personality assessment framework called Traits Run Deeper, which consists of three components. Specifically, the Multimodal Foundation Representation (MFR) module constructs personality-oriented multimodal inputs and leverages psychology-informed semantic templates as anchors, enabling foundation models to capture trait-relevant information. Building upon MFR, the Trait-Specific Modality Fusion (TSMF) module acts as an asymmetric fusion mechanism, allowing each dimension to selectively exploit different modality pathways from modality-specific modeling to complementary fusion. Thus, TSMF captures heterogeneous modality preferences while reducing cross-modal contamination. Furthermore, the Distribution-Calibrated Personality Regression (DCPR) module mitigates label imbalance and central tendency bias through target distribution calibration, improving robustness and stability. Experimental results on the AVI Challenge 2026 validation set demonstrate the effectiveness of the proposed framework, reducing mean squared error (MSE) by approximately 25% compared with the baseline. Consistent improvements are observed on the official test set, where our method achieves the best performance and ranks first in the Personality Assessment Track. The source code will be made available at https://github.com/MSA-LMC/AVI2026.
Chinese Translation
个性评估旨在通过语言、声音和面部线索中的动态行为推断稳定的个性特征。由于不同的个性维度通过不同的行为视角揭示，因此建模特征特定证据具有挑战性。然而，大多数现有方法在所有维度上采用统一的多模态融合策略，假设各模态的贡献相同。这忽视了特征特定的模态偏好，并引入了跨模态干扰。为了解决这个问题，我们提出了一种新的个性评估框架，称为“特征更深”，该框架由三个组件组成。具体而言，多模态基础表示（MFR）模块构建个性导向的多模态输入，并利用心理学信息的语义模板作为锚点，使基础模型能够捕捉与特征相关的信息。在MFR的基础上，特征特定模态融合（TSMF）模块作为一种非对称融合机制，允许每个维度选择性地利用来自特定模态建模到互补融合的不同模态路径。因此，TSMF捕捉异质模态偏好，同时减少跨模态污染。此外，分布校准个性回归（DCPR）模块通过目标分布校准减轻标签不平衡和中心倾向偏差，提高了鲁棒性和稳定性。在AVI Challenge 2026验证集上的实验结果表明，所提出框架的有效性，与基线相比，均方误差（MSE）减少了约25%。在官方测试集上也观察到一致的改进，我们的方法实现了最佳性能，并在个性评估赛道中排名第一。源代码将发布在 https://github.com/MSA-LMC/AVI2026。
cs.CV / 5 / 2606.11285
EventRadar: Long-Range Visual UAV Discovery through Spatiotemporal Event Sensing
EventRadar：通过时空事件感知实现远程视觉无人机发现
Abstract
Unauthorized unmanned aerial vehicle (UAV) activity around airports, public venues, and other sensitive sites has made protected-airspace monitoring increasingly important. A practical sensing system must search a wide angular region, find small long-range targets, and return both bearing support and UAV-specific evidence before a restricted perimeter is breached. Existing UAV detection paths often rely on spatially organized evidence, such as body extent, silhouette, or track continuity. At long range, however, these cues become difficult to preserve and verify as the target footprint weakens and its image-plane support shrinks. EventRadar follows a complementary cue: propeller-induced temporal periodicity, which recent event-camera sensing studies have shown can reveal UAV-specific motion after appearance becomes weak. We extend this cue to kilometer-scale active sensing with an event-camera prototype. Scene-Anchored Geometry Evidence (SAGE) fuses scanning events with IMU pose to maintain a bearing-indexed scene memory, separating transient candidate support from persistent background clutter. Comb-guided Harmonic-Group Learned Iterative Shrinkage and Thresholding Algorithm (CHG) then treats each candidate as a weak high-rate timing signal and recovers phase-insensitive harmonic evidence with fixed compute. Compared with related event-camera baselines on 700-1500 m UAV event recordings, EventRadar achieves 0.990 mAP$_{.3}$ and 0.949 F1$_{.3}$, reduces FN$_{.3}$ to 0.009, and shows real-time feasibility in prototype profiling.
Chinese Translation
在机场、公共场所及其他敏感区域，未经授权的无人机（UAV）活动使得受保护空域的监测变得愈加重要。一个实用的感知系统必须能够搜索广泛的角度区域，发现小型远程目标，并在限制区域被突破之前返回方位支持和无人机特定证据。现有的无人机检测路径通常依赖于空间组织的证据，例如机体范围、轮廓或轨迹连续性。然而，在远程情况下，这些线索变得难以保持和验证，因为目标的足迹减弱，其图像平面支持也随之缩小。EventRadar 采用了一种互补线索：螺旋桨引起的时间周期性，最近的事件相机感知研究表明，这可以在目标出现变弱后揭示无人机特定的运动。我们将这一线索扩展到公里级的主动感知，使用事件相机原型。场景锚定几何证据（Scene-Anchored Geometry Evidence, SAGE）将扫描事件与惯性测量单元（IMU）姿态融合，以维持一个方位索引的场景记忆，将瞬态候选支持与持续的背景杂波分离。然后，梳状引导的谐波组学习迭代收缩和阈值算法（Comb-guided Harmonic-Group Learned Iterative Shrinkage and Thresholding Algorithm, CHG）将每个候选视为一个弱高频率定时信号，并以固定计算恢复相位不敏感的谐波证据。与700-1500米无人机事件记录的相关事件相机基线相比，EventRadar 实现了0.990 mAP$_{.3}$和0.949 F1$_{.3}$，将FN$_{.3}$降低至0.009，并在原型分析中展示了实时可行性。
cs.CV / 6 / 2606.11289
i1: A Simple and Fully Open Recipe for Strong Text-to-Image Models
强文本到图像模型的简单完全开放配方
Abstract
Diffusion models have consistently driven progress in text-to-image generation. However, it is challenging to attribute recent progress to specific modeling and data choices: state-of-the-art open-weight models provide limited ablations, and do not disclose their training data and full training details. The research community needs fully open (weights, data, and code) models as a foundation for further research; yet existing fully open models still fall significantly short of leading models in performance. In this project, we conduct a systematic investigation of the modeling and data design choices in text-to-image diffusion training and inference with 300+ controlled experiments totaling 700K+ TPU v6e hours. Our experiments highlight several empirical findings (e.g., equal weighting is a strong default for mixing curated datasets) and simple design decisions (e.g., larger text encoder adapters improve performance with minimal added parameters) for training strong models. Guided by these insights, we train i1, a 3B-parameter text-to-image diffusion model using only publicly available datasets. i1 is competitive with leading models on five representative benchmarks (GenEval, DPG, PRISM, CVTG-2K, and LongText), and outperforms the best existing fully open model by 29.5 absolute percentage points on average. We provide the i1 checkpoints, training and inference code, and the data processing pipeline. Together, our findings and the i1 recipe establish a practical foundation for future open research in text-to-image diffusion models. Our code is available at https://github.com/zlab-princeton/i1.
Chinese Translation
扩散模型在文本到图像生成中持续推动了进展。然而，将近期的进展归因于特定的建模和数据选择是具有挑战性的：最先进的开放权重模型提供的消融实验有限，并且未披露其训练数据和完整的训练细节。研究社区需要完全开放（权重、数据和代码）模型作为进一步研究的基础；然而，现有的完全开放模型在性能上仍显著低于领先模型。在本项目中，我们对文本到图像扩散训练和推理中的建模和数据设计选择进行了系统调查，进行了300多个受控实验，总计超过700K TPU v6e小时。我们的实验突出了若干实证发现（例如，等权重是混合策划数据集的强默认选择）和简单设计决策（例如，更大的文本编码器适配器在增加最小参数的情况下提高了性能），为训练强模型提供了指导。在这些见解的指导下，我们训练了i1，一个使用仅公开可用数据集的3B参数文本到图像扩散模型。i1在五个代表性基准（GenEval、DPG、PRISM、CVTG-2K和LongText）上与领先模型具有竞争力，并且在平均上比现有最佳完全开放模型超出29.5个百分点。我们提供了i1的检查点、训练和推理代码，以及数据处理管道。我们的发现和i1配方共同建立了未来文本到图像扩散模型开放研究的实用基础。我们的代码可在https://github.com/zlab-princeton/i1获取。
cs.CV / 7 / 2606.11314
TRON: Tracing Rays to Orchestrate a Neural Renderer for 3D Gaussian Reconstructions
TRON：追踪光线以协调神经渲染器进行3D高斯重建
Abstract
We introduce TRON, a rendering framework that combines 3D Gaussian ray tracing with neural rendering to enable realistic and controllable rendering of real-world 3D scenes under novel lighting, dynamic object motion, object insertion, and material editing. Prior approaches that rely solely on physically based rendering (PBR) of Gaussian representations struggle to achieve realistic relighting due to imperfections in reconstructed geometry, material estimates, and light transport estimation. At the same time, neural rendering methods often lack an explicit scene representation, limiting their ability to support interactive editing with fine-grained manipulation. TRON bridges these two paradigms. We use intrinsic decomposition priors from a learned inverse rendering model to regularize the material properties of a Gaussian field, and repurpose a ray tracer to provide radiometric guidance rather than final pixels. By treating this output as a structured 3D scaffold, we empower a lightweight neural renderer to bridge the domain gap between shading-model constrained estimates and photorealistic output. Our key insight is that the combination of explicit 3D knowledge with robust material priors provides speed and controllability, while neural rendering enables the synthesis of photorealistic images. To support real-world scenarios, we train our neural renderer with a multi-stage strategy consisting of large-scale pretraining and targeted fine-tuning on a newly constructed dataset of 2.1M rendered synthetic and real-world frames from 3D reconstructions. TRON outperforms Gaussian-based relighting methods in realism, and prior neural renderers in editability and speed. To the best of our knowledge, TRON is the first method to enable practical interactive applications in captured 3D environments, offering realistic appearance under dynamic geometric, lighting and material conditions.
Chinese Translation
我们提出了TRON，一个渲染框架，它结合了3D高斯光线追踪和神经渲染，以实现对现实世界3D场景在新颖光照、动态物体运动、物体插入和材料编辑下的真实且可控的渲染。以往仅依赖于高斯表示的物理基础渲染（PBR）的方法，由于重建几何体、材料估计和光传输估计中的缺陷，难以实现真实的重光照。同时，神经渲染方法通常缺乏明确的场景表示，限制了其支持细粒度交互编辑的能力。TRON弥合了这两种范式的差距。我们利用从学习的逆渲染模型中获得的内在分解先验来规范高斯场的材料属性，并重新利用光线追踪器提供辐射指导，而不是最终像素。通过将该输出视为结构化的3D支架，我们使轻量级神经渲染器能够弥合受阴影模型约束的估计与照片级真实输出之间的领域差距。我们的关键见解是，显式的3D知识与稳健的材料先验的结合提供了速度和可控性，而神经渲染则使得合成照片级真实图像成为可能。为了支持现实场景，我们采用多阶段策略训练我们的神经渲染器，包括在新构建的包含210万帧渲染合成和现实世界图像的3D重建数据集上的大规模预训练和针对性微调。TRON在真实感方面优于基于高斯的重光照方法，并在可编辑性和速度上超越了以往的神经渲染器。根据我们所知，TRON是首个能够在捕获的3D环境中实现实用交互应用的方法，能够在动态几何、光照和材料条件下提供真实的外观。
cs.CV / 8 / 2606.11320
Semantic Segmentation of Node and Edge Diagrams for Assistive Technology
辅助技术中的节点和边图的语义分割
Abstract
In this paper, we present a novel set of related models for semantic segmentation of node-link diagrams. These diagrams are frequently used to represent mathematical graphs, relationships between concepts, and flowcharts. Such diagrams are difficult to access non-visually; while some assistive interfaces have been designed for node-link diagrams, they rely upon a machine-readable representation of the diagram, whereas such diagrams will generally be made available as bitmap images. Our compact deep learning models show excellent quantitative and qualitative performance on a large synthetic dataset of node-link diagrams, reaching per-pixel accuracy over 93\%.
Chinese Translation
在本文中，我们提出了一组新颖的相关模型，用于节点-链接图的语义分割。这些图通常用于表示数学图、概念之间的关系以及流程图。这类图形对于非视觉访问者来说较为困难；虽然已有一些辅助接口专为节点-链接图设计，但它们依赖于图形的机器可读表示，而这类图通常以位图图像的形式提供。我们的紧凑型深度学习模型在一个大型合成节点-链接图数据集上表现出色，定量和定性性能均优异，达到每像素准确率超过93%。
cs.CV / 9 / 2606.11326
DarkVGGT: Seeing Through Darkness Using Thermal Geometry without Daylight Tax
DarkVGGT：利用热几何透视黑暗，无需日光税
Abstract
Recent feed-forward 3D reconstruction methods have demonstrated strong performance and flexibility in efficient end-to-end scene geometry estimation from image streams. However, their reliance on visible-light appearance makes them vulnerable in dark and low-visibility environments, where RGB cues are severely degraded and geometric evidence becomes ambiguous. To address this challenge, we propose DarkVGGT, an RGB-T feed-forward geometry framework that uses physics-aware thermal modeling for robust 3D estimation in low-light scenes. DarkVGGT introduces two complementary modules. First, physics-inspired thermal factorization extracts emissive-dominant, geometry-consistent thermal cues while isolating sparse reflective residuals that may introduce geometric ambiguity. Second, geometry-shared thermal routing isolates modality-invariant geometric structures from thermal-specific patterns, selectively injecting reliability-aware structural guidance into the RGB stream. Together, these components enable accurate thermal-informed geometry estimation under degraded RGB conditions while largely preserving performance in well-lit environments. Experiments on low-visibility RGB-T benchmarks demonstrate consistent improvements in both depth and camera pose estimation over existing feed-forward geometry baselines.
Chinese Translation
近期的前馈式三维重建方法在从图像流中高效端到端场景几何估计方面表现出强大的性能和灵活性。然而，它们对可见光外观的依赖使其在黑暗和低能见度环境中变得脆弱，在这些环境中，RGB线索严重退化，几何证据变得模糊。为了解决这一挑战，我们提出了DarkVGGT，一个RGB-T前馈几何框架，利用物理感知的热建模在低光场景中进行稳健的三维估计。DarkVGGT引入了两个互补模块。首先，受物理启发的热因子分解提取发射主导、几何一致的热线索，同时隔离可能引入几何模糊的稀疏反射残差。其次，几何共享热路由从热特定模式中隔离模态不变的几何结构，选择性地将可靠性感知的结构指导注入RGB流中。这些组件共同使得在退化的RGB条件下能够准确进行热信息驱动的几何估计，同时在光照良好的环境中大幅保持性能。在低能见度RGB-T基准测试中的实验表明，在深度和相机姿态估计方面，相较于现有的前馈几何基线，均有持续的提升。
cs.CV / 10 / 2606.11363
NSVQ: Mitigating Codebook Collapse by Stabilizing Encoder Drift in Vector Quantization
NSVQ：通过稳定编码器漂移来缓解向量量化中的代码本崩溃
Abstract
Vector quantization is central to modern generative modeling pipelines, but large-codebook VQ models often suffer from codebook collapse. We identify encoder drift as a key driver of this failure: as the encoder moves the latent distribution, sparsely updated code vectors can lag behind, lose assignments, and increase quantization error, creating a feedback loop through the straight-through estimator. We propose NSVQ, a non-stationary-aware VQ training strategy that combines a dense non-stationary embedding loss, codebook replacement, and stage-wise encoder freezing. NSVQ first helps the codebook track encoder drift during early training, then freezes the encoder to consolidate the codebook under a fixed latent geometry, and finally reintroduces adversarial refinement. Experiments on ImageNet-1k show that NSVQ improves reconstruction quality while maintaining full codebook utilization. On ImageNet-1k at 128$\times$128 with 65,536 codes, NSVQ reduces rFID from 2.39 to 2.10 compared with SimVQ, while both methods maintain 100\% utilization. Additional latent diffusion experiments show that NSVQ also improves downstream ImageNet generation FID.
Chinese Translation
向量量化是现代生成建模流程的核心，但大规模代码本的向量量化模型常常遭遇代码本崩溃。我们将编码器漂移识别为这一失败的关键驱动因素：随着编码器移动潜在分布，稀疏更新的代码向量可能滞后，失去分配并增加量化误差，从而通过直通估计器形成反馈循环。我们提出了NSVQ，一种非平稳感知的向量量化训练策略，结合了密集的非平稳嵌入损失、代码本替换和阶段性编码器冻结。NSVQ首先帮助代码本在早期训练中跟踪编码器漂移，然后冻结编码器以在固定的潜在几何下巩固代码本，最后重新引入对抗性精炼。在ImageNet-1k上的实验表明，NSVQ在保持完整代码本利用率的同时提高了重建质量。在ImageNet-1k上，128$ imes$128的65,536个代码，NSVQ将rFID从2.39降低到2.10，而两种方法均保持100\%的利用率。额外的潜在扩散实验表明，NSVQ还改善了下游ImageNet生成的FID。
cs.CV / 11 / 2606.11381
From Simulation to Real-World: An In-Field 6D Pose Dataset and Baseline for Robotic Strawberry Harvesting
从仿真到现实：用于机器人草莓采摘的现场6D位姿数据集和基准
Abstract
Robotic strawberry harvesting requires precise 6D pose estimation; however, collecting 6D pose ground truth in real agricultural fields is inherently challenging. Existing 6D pose estimation methods have therefore relied solely on synthetic data that lacks scene-level realism, leaving their performance under real agricultural field conditions unquantified. In this work, we present, to the best of our knowledge, the first real-world 6D pose ground truth dataset of strawberries collected in actual agricultural fields (12,040 images). We also introduce a synthetic dataset rendered in NVIDIA Isaac Sim, featuring scene-level realism and domain randomization. Nevertheless, our experiments reveal that a significant sim-to-real gap persists, underscoring the necessity of real agricultural field data for reliable evaluation. We further quantify the sim-to-real gap through baseline 6D pose estimation results across backbone encoders, serving as a reference for future work. The real-world dataset will be made available upon acceptance.
Chinese Translation
机器人草莓采摘需要精确的6D位姿估计；然而，在实际农业领域收集6D位姿的真实数据具有固有的挑战性。因此，现有的6D位姿估计方法仅依赖于缺乏场景级真实感的合成数据，使得它们在真实农业田间条件下的性能未得到量化。在本研究中，我们首次提出了在实际农业领域收集的草莓6D位姿真实数据集（12,040张图像）。我们还介绍了一个在NVIDIA Isaac Sim中渲染的合成数据集，具有场景级真实感和领域随机化。然而，我们的实验表明，仍然存在显著的仿真与现实之间的差距，强调了真实农业田间数据在可靠评估中的必要性。我们进一步通过基准6D位姿估计结果量化了仿真与现实之间的差距，为未来的研究提供参考。该真实世界数据集将在接受后公开。
cs.CV / 12 / 2606.11385
DeceptionX: Explainable Deception Detection with Multimodal Large Language Models
DeceptionX：基于多模态大型语言模型的可解释欺骗检测
Abstract
Deception detection is a critical and highly challenging task within affective computing and behavioral analysis. Existing deep learning methods typically treat this task as a straightforward classification problem; however, this black-box approach lacks interpretability and fails to capture the complex logical deduction processes utilized by human experts when identifying lies. While Multimodal Large Language Models (MLLMs) have shown potential, applying them effectively requires a bridge between low-level audiovisual cues and high-level logical reasoning. In this paper, we propose DeceptionX, a novel MLLM framework that shifts the paradigm of deception detection from black-box classification to an interpretable Observe-Think-Summarize reasoning process. To address the scarcity of high-quality reasoning data, we first constructed DeceptChain, a high-quality dataset developed through a human-in-the-loop process. This dataset synthesizes fine-grained visual and auditory evidence (such as micro-expressions and vocal tremors) into structured chain-of-thought reasoning data. Furthermore, we propose a three-stage training pipeline and a Discrepancy-Aware Redundancy Elimination~(DARE) strategy for DeceptionX to further enhance the model's generalization capabilities. Extensive experiments demonstrate that DeceptionX not only outperforms existing MLLM baselines and state-of-the-art methods on standard real-world benchmarks but also provides transparent, expert-level reasoning paths, bridging the critical gap between accuracy and interpretability in multimodal deception detection.
Chinese Translation
欺骗检测是情感计算和行为分析中一项关键且极具挑战性的任务。现有的深度学习方法通常将此任务视为一个简单的分类问题；然而，这种黑箱方法缺乏可解释性，无法捕捉人类专家在识别谎言时所采用的复杂逻辑推理过程。尽管多模态大型语言模型（MLLMs）显示出潜力，但有效应用它们需要在低级视听线索与高级逻辑推理之间架起桥梁。本文提出了DeceptionX，一个新颖的MLLM框架，将欺骗检测的范式从黑箱分类转变为可解释的观察-思考-总结推理过程。为了解决高质量推理数据的稀缺问题，我们首先构建了DeceptChain，这是一个通过人机协作过程开发的高质量数据集。该数据集将细粒度的视觉和听觉证据（如微表情和声音颤抖）合成结构化的思维链推理数据。此外，我们为DeceptionX提出了一个三阶段训练管道和一个差异感知冗余消除（DARE）策略，以进一步增强模型的泛化能力。大量实验表明，DeceptionX不仅在标准现实世界基准上超越了现有的MLLM基线和最先进的方法，还提供了透明的专家级推理路径，弥合了多模态欺骗检测中准确性与可解释性之间的关键差距。
cs.CV / 13 / 2606.11390
A Scalable PyTorch Abstraction for Multi-GPU Gaussian Splatting
一种可扩展的 PyTorch 抽象用于多 GPU 高斯点云渲染
Abstract
Gaussian splatting methods have become increasingly popular for neural reconstruction of the real world. However, they are often limited in scale and resolution due to compute and memory constraints. We present a multi-GPU Gaussian splatting approach that scales reconstruction to higher resolutions and larger scenes while abstracting away the code complexity typically associated with distributing a model. To accomplish this, we propose a PyTorch backend that distributes the Gaussian parameters and splatting operators across GPUs via CUDA unified memory and NVLink. Because distribution occurs at the operator level, the model code requires no explicit cross-device communication. More broadly, the backend exposes multiple GPUs as an aggregate PyTorch device and supports other PyTorch operators. We demonstrate city-scale reconstructions with street-level detail consisting of over 1 billion Gaussian splats, more than 25 times as many as the current state of the art.
Chinese Translation
高斯点云渲染方法在真实世界的神经重建中变得越来越流行。然而，由于计算和内存限制，它们在规模和分辨率上往往受到限制。我们提出了一种多 GPU 高斯点云渲染方法，能够将重建扩展到更高的分辨率和更大的场景，同时抽象掉通常与模型分布相关的代码复杂性。为此，我们提出了一个 PyTorch 后端，通过 CUDA 统一内存和 NVLink 在 GPU 之间分配高斯参数和点云渲染算子。由于分布发生在算子级别，模型代码不需要显式的跨设备通信。更广泛地说，该后端将多个 GPU 作为一个聚合的 PyTorch 设备，并支持其他 PyTorch 算子。我们展示了具有街道级细节的城市规模重建，包含超过 10 亿个高斯点云，数量超过当前最先进技术的 25 倍。
cs.CV / 14 / 2606.11446
3D-CBM: A Framework for Concept-Based Interpretability in Generative 3D Modeling
3D-CBM：生成3D建模中的基于概念的可解释性框架
Abstract
This research introduces a framework for incorporating Concept Bottleneck Models (CBMs) into 3D generative architectures to address the inherent 'semantic gap' in deep geometric learning. As deep models become central to 3D content creation, explainability shifts from a peripheral feature to a fundamental requirement for trust and accountability in safety-critical domains such as healthcare and manufacturing. CBMs provide an intrinsic interpretability solution by constraining latent representations to align with human-defined concepts, yet their application to unstructured 3D data remains largely unexplored. We design, implement, and validate a formal 3D-CBM architecture that maps raw geometric inputs, including point clouds and meshes, into a multi-tiered taxonomy of interpretable primitives and functional attributes. The framework further identifies strategic datasets, such as PartNet and ShapeNet, specialized for concept-based supervision. Experimental results from a 3D part-manipulation proof-of-concept experiment demonstrate the framework's efficacy, achieving a concept prediction accuracy of 88.8\% and a Chamfer Distance of 0.0115. Critically, the model enables precise test-time intervention, allowing for the interactive correction of structural errors. This work establishes a foundation for semantically-steerable 3D generation and invites further exploration into collaborative human-in-the-loop design systems.
Chinese Translation
本研究提出了一种将概念瓶颈模型（Concept Bottleneck Models, CBMs）融入3D生成架构的框架，以解决深度几何学习中固有的“语义差距”。随着深度模型在3D内容创作中的核心地位日益突出，可解释性从边缘特征转变为在医疗和制造等安全关键领域中建立信任与问责的基本要求。CBMs通过约束潜在表示与人类定义的概念对齐，提供了一种内在的可解释性解决方案，但其在非结构化3D数据中的应用仍然未得到充分探索。我们设计、实现并验证了一个正式的3D-CBM架构，该架构将原始几何输入（包括点云和网格）映射到一个多层次的可解释原语和功能属性分类法中。该框架进一步识别了如PartNet和ShapeNet等战略数据集，专门用于基于概念的监督。来自3D部件操作概念验证实验的实验结果表明该框架的有效性，实现了88.8%的概念预测准确率和0.0115的Chamfer距离。关键是，该模型能够在测试时进行精确干预，允许对结构错误进行交互式修正。本研究为语义可引导的3D生成奠定了基础，并邀请进一步探索协作的人机交互设计系统。
cs.CV / 15 / 2606.11450
Exploring Adaptive Masked Reconstruction for Self-Supervised Skeleton-Based Action Recognition
探索自适应掩码重建在自监督基于骨架的动作识别中的应用
Abstract
Recently, masked skeleton reconstruction models have emerged as strong action representation learners, driving significant progress in self-supervised skeleton-based action recognition. However, existing state-of-the-art methods must predict an exceedingly large number of spatiotemporal patches, significantly prolonging training time. Besides, by treating all spatiotemporal regions equally during reconstruction, these models are distracted from learning the critical motion patterns that underlie action semantics. To address these challenges, we propose Adaptive Masked Reconstruction (AMR), a faster and stronger pre-training framework. We first decouple the decoder from the encoder, enabling flexible prediction of larger spatiotemporal patches and dramatically reducing reconstruction complexity. Given that larger patches contain more complex information, which is challenging to predict and consequently degrades performance, we accordingly introduce an adaptive guidance module. This module identifies regions of high motion informativeness, guiding the model to focus on the most discriminative parts of each patch and alleviating reconstruction difficulty. Experiments on NTU RGB+D 60, NTU RGB+D 120, and PKU-MMD datasets demonstrate that AMR not only accelerates pre-training substantially but also improves downstream recognition accuracy, surpassing current state-of-the-art approaches.
Chinese Translation
最近，掩码骨架重建模型作为强大的动作表征学习者，推动了自监督基于骨架的动作识别的显著进展。然而，现有的最先进方法必须预测大量的时空补丁，这显著延长了训练时间。此外，在重建过程中将所有时空区域视为平等，使得这些模型难以学习支撑动作语义的关键运动模式。为了解决这些挑战，我们提出了自适应掩码重建（Adaptive Masked Reconstruction, AMR），这是一个更快且更强大的预训练框架。我们首先将解码器与编码器解耦，使得能够灵活地预测更大的时空补丁，从而大幅降低重建复杂性。鉴于较大的补丁包含更复杂的信息，预测难度较大，进而影响性能，我们相应地引入了自适应引导模块。该模块识别高运动信息区域，引导模型关注每个补丁中最具区分性的部分，从而减轻重建难度。在 NTU RGB+D 60、NTU RGB+D 120 和 PKU-MMD 数据集上的实验表明，AMR 不仅显著加速了预训练过程，还提高了下游识别的准确性，超越了当前的最先进方法。
cs.CV / 16 / 2606.11466
PT-WNO: Point Transformer with Wavelet Neural Operator for 3D Point Cloud Semantic Segmentation
PT-WNO：基于小波神经算子的点变换器用于三维点云语义分割
Abstract
Point cloud semantic segmentation requires architectures that capture both fine-grained local geometry and broad global scene structure. Transformer-based networks have demonstrated strong performance by focusing on detailed local feature aggregation; however, global context is conveyed primarily through skip connections across encoder-decoder stages, which we argue is insufficient for full scene understanding. We hypothesize that augmenting skip connections with a learnable global feature extraction module allows the network to acquire scene-level knowledge before descending into local detail, leading to richer and more contextually grounded representations. To this end, we propose Point Transformer with Wavelet Neural Operato (PT-WNO), which integrates a shared Wavelet Neural Operator (WNO) branch alongside the skip connections of a point cloud transformer backbone. At each encoder-decoder transition, point features are projected onto a dense 3D volumetric grid where the WNO captures multi-scale global spectral context through learnable wavelet decomposition and reconstruction. These global features are fused back into the network via lightweight adapters, complementing rather than replacing the existing skip connections. Experiments on four large-scale 3D point cloud benchmarks demonstrate the effectiveness of PT-WNO. On S3DIS (Area 5), PT-WNO achieves 71.59% mIoU, outperforming the Point Transformer v3 (PTv3) baseline by +1.03 points. On DALES it achieves 81.05% mIoU (+1.47 over the baseline). On ScanNet~v2, PT-WNO obtains 76.19% mIoU, remaining competitive with the baseline (76.36%).
Chinese Translation
点云语义分割需要能够捕捉细粒度局部几何和广泛全局场景结构的架构。基于变换器的网络通过关注详细的局部特征聚合展现了强大的性能；然而，全局上下文主要通过编码器-解码器阶段之间的跳跃连接传递，我们认为这不足以实现全面的场景理解。我们假设，通过可学习的全局特征提取模块增强跳跃连接，可以使网络在深入局部细节之前获取场景级知识，从而导致更丰富且更具上下文基础的表示。为此，我们提出了基于小波神经算子的点变换器（PT-WNO），该模型在点云变换器主干的跳跃连接旁边集成了一个共享的小波神经算子（WNO）分支。在每个编码器-解码器过渡时，点特征被投影到一个密集的三维体积网格上，WNO通过可学习的小波分解和重构捕获多尺度的全局光谱上下文。这些全局特征通过轻量级适配器重新融合回网络中，补充而不是替代现有的跳跃连接。在四个大规模三维点云基准测试上的实验表明，PT-WNO的有效性。在S3DIS（区域5）上，PT-WNO达到了71.59%的mIoU，超越了Point Transformer v3（PTv3）基线1.03个百分点。在DALES上达到了81.05%的mIoU（比基线高1.47）。在ScanNet~v2上，PT-WNO获得了76.19%的mIoU，与基线（76.36%）保持竞争力。
cs.CV / 17 / 2606.11477
Towards Fully Automated Exam Grading: Fairness-Aware Recognition of Handwritten Answers with Foundation Models
迈向完全自动化的考试评分：基于基础模型的公平意识手写答案识别
Abstract
Correcting handwritten exams by hand is time-consuming and error-prone, particularly for large cohorts, while fully digital exams tend to force a didactic narrowing towards closed question formats. A practical middle ground keeps paper-based, problem-oriented tasks but records the assessment-relevant answers as single capital letters in a table that a machine can read. The open question is whether this reading can be made accurate and, above all, fair enough for unsupervised grading. Earlier automated approaches reached only about 88%--91% recognition -- too low -- and failed on the cases that matter most: answers placed outside the cell, crossed out, or written in cursive. We show that general-purpose vision-language foundation models (VLMs), which interpret the page rather than match pixel templates, close this gap. On a benchmark of 61 anonymised exams (3141 answer positions) the best model reaches 98.4% accuracy, well above the previous baseline. Crucially, we centre the evaluation on fairness: we distinguish false negatives (a correct answer marked wrong, which disadvantages the student) from false positives, and a lightweight prompt that supplies the reference solution as context lowers the false-negative rate to 0.58%. Under an exemplary grading scheme only three of the 61 exams would be graded worse, all caught by a student self-review step. Fully automated, fairness-aware exam grading at scale is therefore defensible; we release the anonymised benchmark to support reproducibility.
Chinese Translation
手动批改手写考试既耗时又容易出错，尤其是在大规模考生中，而完全数字化的考试往往迫使教学内容向封闭式问题格式收窄。一个实用的折中方案是保留纸质的、以问题为导向的任务，但将与评估相关的答案记录为机器可读的单个大写字母。关键问题在于，这种读取是否可以做到准确，尤其是对于无监督评分是否足够公平。早期的自动化方法仅达到了约88%至91%的识别率——这一水平过低，并且在最重要的情况下失败：答案位于单元格外、被划掉或以草书书写的情况。我们展示了通用视觉-语言基础模型（VLMs）能够弥补这一差距，这些模型解释页面而不是匹配像素模板。在61份匿名考试（3141个答案位置）的基准测试中，最佳模型达到了98.4%的准确率，远高于之前的基线。至关重要的是，我们将评估重点放在公平性上：我们区分了假阴性（正确答案被标记为错误，给学生带来不利影响）和假阳性，并且一个轻量级的提示，通过提供参考答案作为上下文，将假阴性率降低至0.58%。在一个示范评分方案下，61份考试中只有三份的评分较低，且均被学生自我审核步骤捕获。因此，全面自动化的公平意识考试评分在规模上是可辩护的；我们发布了匿名基准以支持可重复性。
cs.CV / 18 / 2606.11505
On the Study of Biometric Spoofing Detection using Deep Learning
基于深度学习的生物特征欺骗检测研究
Abstract
Biometric systems are increasingly deployed in security applications; however, they remain vulnerable to spoofing attacks, in which attackers exploit counterfeit biometric data to gain unauthorized access. This research evaluates the effectiveness of state-of-the-art machine learning models, MobileNetV2, DenseNet-121, Inception-v3, and Spoof Trace Disentanglement (STD) in detecting spoofing attacks within facial recognition systems. Using the CelebA-Spoof dataset, the study evaluates model effectiveness using metrics such as accuracy, precision, recall, and F1 Score. Cross-dataset validation is carried out on the MSU-MFSD dataset to assess generalizability. The results show MobileNetV2 as the most efficient model, achieving 92% accuracy while balancing computational effectiveness, making it appropriate for real-life applications. Inception-v3 shows moderate robustness, while DenseNet-121 and STD struggle with generalization. The findings highlight the need for advances in domain adaptation and hybrid architectures to enhance biometric security systems.
Chinese Translation
生物特征系统在安全应用中越来越普遍，但仍然容易受到欺骗攻击，攻击者利用伪造的生物特征数据获得未经授权的访问权限。本研究评估了最先进的机器学习模型，包括 MobileNetV2、DenseNet-121、Inception-v3 和 Spoof Trace Disentanglement (STD) 在面部识别系统中检测欺骗攻击的有效性。使用 CelebA-Spoof 数据集，研究通过准确率、精确率、召回率和 F1 分数等指标评估模型的有效性。还在 MSU-MFSD 数据集上进行跨数据集验证，以评估模型的泛化能力。结果表明，MobileNetV2 是最有效的模型，达到了 92% 的准确率，同时在计算效率上保持平衡，适合实际应用。Inception-v3 显示出适度的鲁棒性，而 DenseNet-121 和 STD 在泛化能力上表现不佳。研究结果强调了在领域适应和混合架构方面的进展对于增强生物特征安全系统的必要性。
cs.CV / 19 / 2606.11507
SceneMiner: Identity-Preserving Multi-Task Fine-Tuning for Unified BEV Scene Mining
SceneMiner：用于统一鸟瞰视角场景挖掘的身份保留多任务微调
Abstract
Mining hard, safety-critical scenes from driving logs is bottlenecked by the absence of difficulty labels, and no single proxy, collision risk, trajectory ambiguity, or semantic rarity suffices to find such scenes on its own. We present SceneMiner, a unified, camera-only bird's-eye-view pipeline that emits complementary mining signals from a frozen vision-language backbone in a single forward pass, with no LiDAR or radar: a retrieval embedding for text-prompted scenario search, a multi-label scene-tag distribution, and a continuous physics-based risk score (a motion forecast is a byproduct, not a contribution). Building such a multi-head model exposes our central finding, a failure mode we term cross-task interference: adding or upgrading one head shifts a shared activation stream and degrades weight-frozen sibling heads, so freezing parameters alone is insufficient. Our contribution, identity-preserving multi-task fine-tuning, removes this interference by zero-initializing every new sub-module and freezing every parameter that feeds the shared stream. The mining heads are thereby preserved bit-identically while training only ~102k parameters. The tagging head reaches mAP 0.4614 (micro-F1 0.5557) on 20 scene tags by pooling each scene into 32 visual tokens, and the embedding head supports text-prompted retrieval, validated qualitatively. Code is available at: https://anonymous.4open.science/r/sceneminer_anonymous-64E5
Chinese Translation
从驾驶日志中挖掘困难且安全关键的场景受到缺乏难度标签的制约，单一的代理指标如碰撞风险、轨迹模糊或语义稀缺性都无法独立找到此类场景。我们提出了SceneMiner，一个统一的仅基于相机的鸟瞰视角管道，它在单次前向传播中从冻结的视觉-语言骨干网络中发出互补的挖掘信号，无需激光雷达或雷达：用于文本提示场景搜索的检索嵌入、多标签场景标签分布，以及基于物理的连续风险评分（运动预测是副产品，而非贡献）。构建这样的多头模型揭示了我们的核心发现，即我们称之为跨任务干扰的失败模式：添加或升级一个头会改变共享激活流并降低权重冻结的兄弟头的性能，因此仅仅冻结参数是不够的。我们的贡献，身份保留多任务微调，通过对每个新子模块进行零初始化并冻结每个输入共享流的参数来消除这种干扰。这样，挖掘头在训练时仅需保留约102k参数而实现比特级一致性。标签头在20个场景标签上达到mAP 0.4614（micro-F1 0.5557），通过将每个场景汇聚为32个视觉标记，而嵌入头支持文本提示检索，经过定性验证。代码可在以下链接获取：https://anonymous.4open.science/r/sceneminer_anonymous-64E5
cs.CV / 20 / 2606.11546
VL-DINO: Leveraging CLIP Vision-Language Knowledge for Open-Vocabulary Object Detectio
VL-DINO：利用 CLIP 视觉-语言知识进行开放词汇物体检测
Abstract
Vision-language models like CLIP can provide rich semantic priors for open-vocabulary object detection. However, jointly integrating both textual and visual knowledge into detection architectures remains challenging. In this paper, we propose VL-DINO, an open-vocabulary detector that enhances DINO through more effective exploitation of CLIP's vision-language knowledge. Specifically, a Query-guided Positive Sample Construction (QPSC) module is first developed to construct additional high-quality positive samples, enabling the vanilla DINO framework to better accommodate mixed training across heterogeneous data sources while providing more vision-language alignment signals, thereby incorporating richer textual knowledge during training. A Visual Semantic Encoder (VSE) module is then introduced to distill CLIP visual knowledge into backbone-extracted features, producing fused features for subsequent encoder refinement. Based on the fused features, an Object-Region Semantic Alignment (ORSA) module extracts object-centric region features and aligns them with the corresponding textual embeddings, further incorporating textual cues. In the zero-shot setting, VL-DINO-T and VL-DINO-L achieve 36.3 and 38.1 AP on the LVIS benchmark, respectively, consistently outperforming prior advanced approaches. Extensive experiments demonstrate the effectiveness and competitive performance of the proposed design.
Chinese Translation
视觉-语言模型如 CLIP 可以为开放词汇物体检测提供丰富的语义先验。然而，将文本和视觉知识共同整合到检测架构中仍然具有挑战性。在本文中，我们提出了 VL-DINO，一种开放词汇检测器，通过更有效地利用 CLIP 的视觉-语言知识来增强 DINO。具体而言，首先开发了一个查询引导的正样本构建（Query-guided Positive Sample Construction, QPSC）模块，以构建额外的高质量正样本，使得基础的 DINO 框架能够更好地适应来自异构数据源的混合训练，同时提供更多的视觉-语言对齐信号，从而在训练过程中融入更丰富的文本知识。接着，引入了一个视觉语义编码器（Visual Semantic Encoder, VSE）模块，将 CLIP 的视觉知识提炼到主干提取的特征中，生成用于后续编码器优化的融合特征。基于融合特征，物体区域语义对齐（Object-Region Semantic Alignment, ORSA）模块提取以物体为中心的区域特征，并将其与相应的文本嵌入进行对齐，进一步融入文本线索。在零样本设置下，VL-DINO-T 和 VL-DINO-L 在 LVIS 基准测试中分别达到了 36.3 和 38.1 的平均精度（AP），始终优于先前的先进方法。大量实验证明了所提设计的有效性和竞争性能。
cs.CV / 21 / 2606.11563
Cross-Modal Benchmarking for Robotic Perception in Natural Environments
自然环境中机器人感知的跨模态基准测试
Abstract
Natural environments present a complex challenge to robotics perception systems. Current models, particularly vision foundation models, are largely trained on structured, urban environments leading to weaknesses in their perception for field robotics tasks. We showcase the limitations of current models using our recently released WildCross benchmark, a new cross-modal benchmark for place recognition and metric depth estimation in large-scale natural environments. WildCross comprises over 476K sequential RGB frames with semi-dense depth and surface normal annotations, each aligned with accurate 6DoF pose and synchronized dense lidar submaps. In this work, we provide an expanded analysis of the benchmark results from the recent WildCross benchmark, with particular emphasis on expanded metric depth estimation experiments. Access to the code repository and dataset for this work can be found at https://csiro-robotics.github.io/WildCross.
Chinese Translation
自然环境对机器人感知系统提出了复杂的挑战。目前的模型，特别是视觉基础模型，主要在结构化的城市环境中训练，因此在野外机器人任务的感知能力上存在不足。我们利用最近发布的WildCross基准展示了当前模型的局限性，WildCross是一个用于大规模自然环境中地点识别和度量深度估计的新型跨模态基准。WildCross包含超过476K的连续RGB帧，配有半稠密深度和表面法线标注，每帧都与准确的6DoF位姿对齐，并同步有稠密激光雷达子地图。在本研究中，我们对最近WildCross基准的结果进行了扩展分析，特别强调了扩展的度量深度估计实验。相关代码库和数据集可在 https://csiro-robotics.github.io/WildCross 找到。
cs.CV / 22 / 2606.11568
4DP-QA: Scalable QA for 4D Perception in Vision Language Models
4DP-QA：面向视觉语言模型的可扩展4D感知问答
Abstract
Despite recent advances, Vision Language Models (VLMs) still struggle to grasp the dynamics of the world. We note that the ability to reason about a 4D scene, challenging in itself, is further complicated by two factors. First, VLMs observe motion indirectly via its projection onto 2D images. Second, existing datasets fail to disentangle object and camera motion. To address these challenges, we present a QA generation pipeline that focuses on motion-related scene understanding. We take particular care of the entanglement of camera and object motion by casting tracking in both the traditional way and in a novel, fixed reference system, dubbed True-Motion Tracking, which provides an intuitive description of motion. From this pipeline, we generate a large-scale training dataset of 400K samples, 4DP-QA (4D Perception QA), and a 2.2K-sample benchmark, 4DP-QA-Bench. Training existing models on our dataset yields performance improvements on an external benchmark, validating the effectiveness of our method.
Chinese Translation
尽管近期取得了进展，视觉语言模型（VLMs）在理解世界动态方面仍然面临挑战。我们注意到，推理4D场景的能力本身就很复杂，且受到两个因素的进一步影响。首先，VLMs通过其在2D图像上的投影间接观察运动。其次，现有数据集未能区分物体运动和相机运动。为了解决这些挑战，我们提出了一种专注于与运动相关的场景理解的问答生成管道。我们特别关注相机运动与物体运动的纠缠，通过传统方式和一种新颖的固定参考系统（称为真实运动跟踪，True-Motion Tracking）来进行跟踪，从而提供对运动的直观描述。通过该管道，我们生成了一个包含40万样本的大规模训练数据集4DP-QA（4D感知问答）以及一个包含2200个样本的基准数据集4DP-QA-Bench。在我们的数据集上训练现有模型，能够在外部基准上提升性能，验证了我们方法的有效性。
cs.CV / 23 / 2606.11572
FreqKD: Frequency-Decoupled Cross-Modal Knowledge Distillation for Infrared Object Detection
FreqKD：用于红外目标检测的频率解耦跨模态知识蒸馏
Abstract
Transfer learning from large-scale RGB foundation models to infrared (IR) imagery through knowledge distillation (KD) remains challenging due to fundamental differences in image formation physics. We investigate the spectral structure of the RGB--IR modality gap and observe that feature divergence is not uniform across spatial frequencies: low-frequency components (shape, layout) show greater cross-modal alignment than high-frequency components (texture, fine edges), which reflect modality-specific characteristics. Based on this analysis, we propose FreqKD, a frequency-decoupled distillation framework that applies asymmetric supervision adapted to each band's cross-modal consistency. The method employs strict mean squared error (MSE) on the low-frequency band to preserve shared structural information and a relaxed log-MSE loss (weighted at 0.1) on the high-frequency band to provide edge guidance while tolerating texture differences. Spectral divergence analysis on 500 paired samples shows that high-frequency divergence exceeds low-frequency divergence by a factor of 2.4x on average across all analysed transformer layers. On KAIST multispectral pedestrian detection, FreqKD achieves 64.1 mAP50, improving 2.4 points over the DINOv2 baseline. The learned representation transfers across datasets (FLIR ADAS, +2.1 mAP50), tasks (MFNet segmentation, +1.85 mean intersection-over-union), and architectures (ResNet-50, +1.0 mAP50). Code is available at: https://anonymous.4open.science/r/freq_decoupled_kd-5E5A
Chinese Translation
通过知识蒸馏（KD）将大规模RGB基础模型迁移到红外（IR）图像的过程仍然面临挑战，原因在于图像形成物理学的根本差异。我们研究了RGB-IR模态差距的光谱结构，并观察到特征的发散在空间频率上并不均匀：低频成分（形状、布局）显示出比高频成分（纹理、细边缘）更大的跨模态一致性，后者反映了模态特定的特征。基于这一分析，我们提出了FreqKD，一种频率解耦的蒸馏框架，应用适应于每个频段跨模态一致性的非对称监督。该方法在低频段采用严格的均方误差（MSE）以保持共享的结构信息，而在高频段则采用放宽的对数均方误差损失（加权为0.1）以提供边缘引导，同时容忍纹理差异。对500对样本的光谱发散分析表明，所有分析的变换层中，高频发散的平均值是低频发散的2.4倍。在KAIST多光谱行人检测中，FreqKD达到了64.1 mAP50，比DINOv2基线提高了2.4个百分点。所学习的表示在不同数据集（FLIR ADAS，+2.1 mAP50）、任务（MFNet分割，+1.85平均交并比）和架构（ResNet-50，+1.0 mAP50）之间迁移。代码可在以下链接获取：https://anonymous.4open.science/r/freq_decoupled_kd-5E5A
cs.CV / 24 / 2606.11573
Understanding Cross-Sensor Feature Variations for Generalizable 3D Perception
理解跨传感器特征变化以实现可泛化的三维感知
Abstract
Radar-camera BEV perception often suffers from degraded performance when evaluated across datasets, as changes in driving scenes, sensor configurations, and environmental conditions can alter both the input observations and the internal fused representations. This work studies this issue from the perspective of source-domain variation modeling, aiming to improve the robustness of BEV-based 3D detectors without relying on target-domain samples. We introduce a framework that characterizes visual scene variations in the frequency domain and uses them to synthesize diverse source-domain views. By comparing the resulting fused BEV representations, the framework further captures how image-level variations influence multi-modal BEV features. These variation patterns are then used to regularize the detector, encouraging the learned fusion space to remain stable under latent scene changes. The proposed method is applied only during training and leaves the inference pipeline unchanged. Experiments on cross-dataset radar-camera 3D detection between View-of-Delft and TJ4DRadSet demonstrate consistent improvements over multiple BEV fusion backbones, and the gains remain effective when a small amount of target-domain data is available.
Chinese Translation
雷达-摄像头鸟瞰视图（BEV）感知在跨数据集评估时常常表现出性能下降，因为驾驶场景、传感器配置和环境条件的变化可能会改变输入观测和内部融合表示。本文从源域变化建模的角度研究这一问题，旨在提高基于BEV的三维检测器的鲁棒性，而不依赖于目标域样本。我们提出了一个框架，该框架在频域中表征视觉场景变化，并利用这些变化合成多样的源域视图。通过比较生成的融合BEV表示，该框架进一步捕捉图像级变化如何影响多模态BEV特征。这些变化模式随后被用来规范检测器，鼓励学习到的融合空间在潜在场景变化下保持稳定。所提方法仅在训练期间应用，推理流程保持不变。在View-of-Delft和TJ4DRadSet之间的跨数据集雷达-摄像头三维检测实验表明，在多个BEV融合骨干网络上均实现了一致的性能提升，并且当少量目标域数据可用时，这些增益仍然有效。
cs.CV / 25 / 2606.11576
AVIS: Adaptive Test-Time Scaling for Vision-Language Models
AVIS：面向视觉-语言模型的自适应测试时间缩放
Abstract
Modern Vision-Language Models (VLMs) benefit from chain-of-thought prompting and test-time scaling, but these gains often come with prohibitive inference cost due to large visual contexts and long decoding chains. We view this cost through two coupled axes: Visual Context Scaling (VCS), which controls how much visual evidence is passed to the language model, and Visual Reasoning Scaling (VRS), which controls how much inference-time reasoning search is performed. Existing methods typically optimize one axis at a time, leaving the joint allocation of compute across these axes underexplored. We introduce Adaptive Visual Inference Scaling (AVIS), a lightweight policy that adapts both VCS and VRS per query. AVIS realizes VCS through Key Diversity Visual (KDV) pruning, a training-free $O(N)$ key-based rule for removing redundant visual tokens before prefilling, and realizes VRS through adaptive self-consistency, using a learned difficulty predictor to select the number of reasoning rollouts. AVIS is deployment-friendly and compatible with shared-prefill inference, where all rollouts reuse a single prefilling pass and KV cache. Across diverse image and video reasoning benchmarks, AVIS improves the accuracy--compute trade-off relative to VCS-only and VRS-only baselines, and remains effective on top of RL post-trained VLMs while keeping compute and latency low.
Chinese Translation
现代视觉-语言模型（VLMs）受益于思维链提示和测试时间缩放，但这些收益往往伴随着由于大规模视觉上下文和长解码链而导致的高昂推理成本。我们从两个相互关联的轴度来看待这一成本：视觉上下文缩放（Visual Context Scaling, VCS），控制传递给语言模型的视觉证据量；视觉推理缩放（Visual Reasoning Scaling, VRS），控制推理时进行的搜索量。现有方法通常一次优化一个轴度，导致在这两个轴度之间的计算资源联合分配尚未得到充分探索。我们提出了自适应视觉推理缩放（Adaptive Visual Inference Scaling, AVIS），这是一种轻量级策略，能够针对每个查询自适应调整VCS和VRS。AVIS通过关键多样性视觉（Key Diversity Visual, KDV）剪枝实现VCS，这是一种无训练的$O(N)$基于关键的规则，用于在预填充之前去除冗余视觉标记；通过自适应自一致性实现VRS，使用学习的难度预测器选择推理回滚的数量。AVIS适合部署，并与共享预填充推理兼容，其中所有回滚重用单次预填充过程和KV缓存。在多样的图像和视频推理基准测试中，AVIS相较于仅使用VCS或VRS的基线改善了准确性与计算的权衡，并在保持低计算和延迟的同时，对RL后训练的VLMs仍然有效。
cs.CV / 26 / 2606.11578
Contactless 3D Human Body Measurement Using Depth Cameras for Smart Health Monitoring
基于深度摄像头的无接触式三维人体测量用于智能健康监测
Abstract
Contactless body measurement technologies are becoming increasingly significant for smart health monitoring, digital health applications, and remote patient assessment. Traditional anthropometric measurements typically necessitate physical contact and trained personnel, which may constrain scalability in remote healthcare settings. In this study, we introduce a depth camera-based framework for estimating human body measurements utilizing 3D point cloud data. An Orbbec Astra 2 depth camera was employed to capture RGB images, depth maps, and 3D point clouds of participants. The captured point cloud was processed using Python-based tools, including Open3D, NumPy, and OpenCV, to segment the human body from the background. Key anthropometric measurements, such as height and arm span, were computed. The measurements were obtained through a combination of spatial filtering and landmark selection on the 3D point cloud, followed by the projection of the computed measurements onto the corresponding RGB image using camera intrinsic parameters. In addition to linear measurements, the approximate body volume and visible surface area were estimated using voxel-based occupancy analysis and mesh-based surface reconstruction methods. The experimental results from a single depth capture demonstrated that accurate body measurements and geometric estimates could be obtained from depth camera data without physical contact. This study provides a foundation for future real-time systems that integrate depth sensing with intelligent health monitoring and generative AI models for smart healthcare applications.
Chinese Translation
无接触式身体测量技术在智能健康监测、数字健康应用和远程患者评估中变得越来越重要。传统的人体测量通常需要物理接触和经过培训的人员，这可能限制了远程医疗环境中的可扩展性。在本研究中，我们提出了一种基于深度摄像头的框架，用于利用三维点云数据估计人体测量。我们使用Orbbec Astra 2深度摄像头捕获参与者的RGB图像、深度图和三维点云。捕获的点云通过基于Python的工具（包括Open3D、NumPy和OpenCV）进行处理，以从背景中分割出人体。关键的人体测量，如身高和臂展，通过对三维点云进行空间滤波和地标选择的组合来计算，然后使用相机内参将计算出的测量结果投影到相应的RGB图像上。除了线性测量外，还通过基于体素的占用分析和基于网格的表面重建方法估计了近似的身体体积和可见表面积。单次深度捕获的实验结果表明，可以从深度摄像头数据中获得准确的身体测量和几何估计，而无需物理接触。本研究为未来将深度感知与智能健康监测和生成式人工智能模型集成的实时系统奠定了基础，以用于智能医疗应用。
cs.CV / 27 / 2606.11601
Spatially Coupled Phase-to-Depth Calibration for Fringe Projection Profilometry
空间耦合相位-深度标定用于条纹投影轮廓测量
Abstract
In fringe projection profilometry (FPP), depth is commonly recovered by fitting a phase-to-depth relation independently at each camera pixel. Although such pixel-wise calibration achieves high local accuracy, neighboring pixels can acquire markedly different calibration functions even when they observe the same smooth surface, producing spatially inconsistent geometry and structured surface artifacts. We propose a spatially coupled phase-depth transformation in which all pixels share a single low-dimensional mapping-global phase scalars combined with affine spatial terms on the undistorted reference-camera grid-rather than independent per-pixel fits, optionally augmented by a bounded, spatially smooth correction field. We further introduce a native-grid pairing scheme that constructs phase-depth calibration pairs directly on the reference-camera grid: when depth supervision comes from a rectified active-stereo pipeline, planes are fitted in stereo 3D and sampled back onto the camera grid along native rays, so the phase maps are never rectified. On a dental target with high-resolution scanner ground truth, the proposed model attains point-to-surface RMSE comparable to an active-stereo reference (about 12{\mu}m aggregate) while substantially improving spatial coherence over pixel-wise polynomial and rational calibration, and reduces the runtime mapping to a few element-wise operations per pixel with negligible parameter storage.
Chinese Translation
在条纹投影轮廓测量（FPP）中，深度通常通过在每个相机像素独立拟合相位-深度关系来恢复。尽管这种逐像素标定实现了高局部精度，但即使在观察同一光滑表面时，相邻像素也可能获得显著不同的标定函数，从而产生空间不一致的几何形状和结构表面伪影。我们提出了一种空间耦合的相位-深度变换，其中所有像素共享一个低维映射——全局相位标量结合未失真参考相机网格上的仿射空间项，而不是独立的逐像素拟合，此外可选地通过一个有界的、空间平滑的修正场进行增强。我们进一步引入了一种原生网格配对方案，直接在参考相机网格上构建相位-深度标定对：当深度监督来自校正的主动立体视觉管道时，平面在立体3D中拟合并沿原生光线回采样到相机网格，因此相位图从未被校正。在一个具有高分辨率扫描仪真实值的牙科目标上，所提出的模型在点到表面的均方根误差（RMSE）上达到了与主动立体参考相当的水平（约12μm的总和），同时显著提高了相较于逐像素多项式和有理标定的空间一致性，并将运行时映射减少到每个像素仅需少量元素级操作，且参数存储几乎可以忽略不计。
cs.CV / 28 / 2606.11602
On Aligning Hierarchical Standardized Embedding for Audio-visual Generalized Zero-shot Learning
音频-视觉通用零样本学习的层次标准化嵌入对齐
Abstract
Audio-visual Generalized Zero-shot Learning (AV-GZSL) is a challenging task that aims to classify both seen and unseen objects or scenes by integrating data from audio and visual modalities. Recent studies primarily focus on fusing or aligning audio and visual features to generate more informative audio-visual embeddings. Also, aligning the audio-visual and textual features of most existing methods relies solely on the optimization objectives. However, those methods neglect the inherent distributional and structural differences between audio-visual and textual modalities. To address this limitation, we propose a method termed Aligning Hierarchical Standardized Embedding (AHSE), which enables hierarchical alignment of standardized audio-visual and textual embeddings within a shared embedding space. Specifically, we first apply Z-score standardization to the fused audio-visual and textual embeddings to reduce distributional mismatches. We then introduce a hierarchical alignment strategy that minimizes discrepancies at the semantic, class, and batch levels, thereby constructing a more robust and well-structured embedding space. This strategy not only preserves semantic and inter-class relationships but also maintains spatial consistency within each batch. Extensive experiments on three benchmark datasets: VGGSound-GZSL, UCF-GZSL, and ActivityNet-GZSL, demonstrate that AHSE achieves competitive performance in zero-shot learning.
Chinese Translation
音频-视觉通用零样本学习（AV-GZSL）是一项具有挑战性的任务，旨在通过整合音频和视觉模态的数据来分类已见和未见的对象或场景。近期研究主要集中在融合或对齐音频和视觉特征，以生成更具信息量的音频-视觉嵌入。此外，大多数现有方法对音频-视觉和文本特征的对齐仅依赖于优化目标。然而，这些方法忽视了音频-视觉和文本模态之间固有的分布和结构差异。为了解决这一局限性，我们提出了一种称为层次标准化嵌入对齐（AHSE）的方法，该方法能够在共享嵌入空间内实现标准化音频-视觉和文本嵌入的层次对齐。具体而言，我们首先对融合的音频-视觉和文本嵌入应用 Z-score 标准化，以减少分布不匹配。然后，我们引入了一种层次对齐策略，最小化语义、类别和批次级别的差异，从而构建一个更稳健且结构良好的嵌入空间。这一策略不仅保留了语义和类间关系，还保持了每个批次内的空间一致性。在三个基准数据集（VGGSound-GZSL、UCF-GZSL 和 ActivityNet-GZSL）上的大量实验表明，AHSE 在零样本学习中实现了具有竞争力的性能。
cs.CV / 29 / 2606.11606
Frozen Foundation-Model Embeddings Discard Small-Lesion Signal in Chest Radiography: Implications for Pre-Deployment Evaluation
冻结基础模型嵌入在胸部放射影像中丢弃小病灶信号：对预部署评估的影响
Abstract
Frozen vision-transformer (ViT) foundation-model embeddings increasingly serve as the substrate for downstream chest-radiography (CXR) pipelines, yet where small-scale, low-contrast signal is retained or lost in the frozen forward pass has not been systematically quantified across architectures, pretraining domains, and objectives. We probed five frozen ViTs (RAD-DINO, DINOv2-B/14, DINOv3 ViT-7B, BiomedCLIP, MedSigLIP) and a frozen DINO-pretrained ResNet-50 architectural control across three large CXR cohorts (NIH-CXR14, MIMIC-CXR, Emory-CXR; aggregate pool n=492,724) and ChestX-Det10 (n=3,543; 1,462 small-lesion bounding boxes across Calcification, Nodule, Mass). Each model was evaluated with a small-scale-perturbation panel and a region-aware bounding-box-stratified probe on real lesions, comparing three pooling modes from the same forward pass: classification token (CLS), patch-mean (mean over all final-layer patch tokens), and bounding-box-restricted patch-local. On the perturbation panel, CLS embeddings sat at the chance floor (area under the ROC curve [AUC] 0.500-0.524); patch-mean was indistinguishable from CLS on iso-blur and reticular-fine cells but rose with CLS on larger directional-blur footprints, while disease AUC on globally decided tasks ranged 0.642-0.913. Patch-local probes recovered AUC ~1.0 from the same forward pass (per-model mean improvement +0.412 to +0.488); the ResNet-50 control reproduced the chance floor. On ChestX-Det10, image-level CLS classification showed within-class small-versus-large stratum gaps up to +0.243 AUC; bounding-box-level patch-local pooling on the same forward pass recovered AUC >= 0.899 on every (model x class) cell. Frozen ViT embeddings silently suppress small-scale signal at the global-aggregation step; the signal is recoverable from patch tokens conditional on a region of interest.
Chinese Translation
冻结的视觉变换器（Vision Transformer, ViT）基础模型嵌入越来越多地作为下游胸部放射影像（CXR）管道的基础，但在冻结的前向传播中，小规模、低对比度信号的保留或丢失尚未在不同架构、预训练领域和目标之间系统性量化。我们探讨了五种冻结的ViT（RAD-DINO、DINOv2-B/14、DINOv3 ViT-7B、BiomedCLIP、MedSigLIP）和一个冻结的DINO预训练ResNet-50架构控制，在三个大型CXR队列（NIH-CXR14、MIMIC-CXR、Emory-CXR；总样本数n=492,724）和ChestX-Det10（n=3,543；1,462个小病灶边界框，涵盖钙化、结节、肿块）中进行评估。每个模型都通过小规模扰动面板和区域感知边界框分层探测器在真实病灶上进行评估，比较来自同一前向传播的三种池化模式：分类标记（CLS）、补丁均值（所有最终层补丁标记的均值）和边界框限制的补丁局部。在扰动面板上，CLS嵌入处于随机猜测水平（ROC曲线下面积[AUC] 0.500-0.524）；在等模糊和网状细胞上，补丁均值与CLS无显著区别，但在较大方向模糊的情况下，补丁均值的表现优于CLS，而在全球决策任务上的疾病AUC范围为0.642-0.913。补丁局部探测器从同一前向传播中恢复了约1.0的AUC（每个模型的平均提升为+0.412到+0.488）；ResNet-50控制组则重现了随机猜测水平。在ChestX-Det10中，图像级CLS分类显示同类小病灶与大病灶之间的层次差距高达+0.243 AUC；在同一前向传播中，边界框级补丁局部池化在每个（模型 x 类别）单元上恢复了AUC >= 0.899。冻结的ViT嵌入在全局聚合步骤中无声地抑制了小规模信号；该信号可以从补丁标记中恢复，前提是有感兴趣区域的条件。
cs.CV / 30 / 2606.11615
Adv-TGD: Adversarial Text-Guided Diffusion for Face Recognition Impersonation Attacks
Adv-TGD：用于人脸识别冒充攻击的对抗文本引导扩散
Abstract
The widespread adoption of face recognition (FR) technologies raises serious privacy concerns, as facial data can be exploited without consent. To address this challenge, we propose Adv-TGD, a generative adversarial attack framework that synthesizes photorealistic faces capable of impersonating target identities and deceiving face recognition systems. Built upon Stable Diffusion, Adv-TGD performs per-sample LoRA fine-tuning conditioned on concise textual prompts to generate natural yet adversarially manipulated identities. Unlike conventional identity-attack approaches, our method optimizes lightweight cross-attention adapters for each source-target pair within a single-step denoising process. Latent blending is constrained by a face-local heatmap mask to ensure spatially precise identity manipulation while preserving non-sensitive regions. We introduce a composite objective that integrates masked epsilon-MSE reconstruction, thresholded identity divergence in FR embedding space, directional feature alignment, and source-similarity suppression to balance adversarial attack and visual realism. Optionally, LLaVA-generated attribute prompts enhance fine-grained semantic details without reintroducing identity cues. Under the black-box evaluation protocol, Adv-TGD attains an average attack success rate (ASR) of 85.90% across IR152, IRSE50, MobileFace, and FaceNet, surpassing the semantic SOTA baseline Adv-CPG by +6.25 points, diffusion-based makeup method DiffAIM by +3 points, and noise-based P3-Mask by +16 points. Despite its strong attack efficacy, Adv-TGD preserves high visual fidelity (PSNR = 27.15 dB, SSIM = 0.981). Furthermore, we demonstrate the flexibility of our framework by successfully extending it to in-the-wild datasets (LADN), general object classification (ImageNet), and transformer-based diffusion models (FLUX.1).
Chinese Translation
人脸识别（FR）技术的广泛应用引发了严重的隐私担忧，因为面部数据可能在未获同意的情况下被利用。为了解决这一挑战，我们提出了Adv-TGD，一个生成对抗攻击框架，能够合成具有照片真实感的人脸，以冒充目标身份并欺骗面部识别系统。Adv-TGD基于稳定扩散（Stable Diffusion）构建，针对简洁的文本提示进行每个样本的LoRA微调，以生成自然但经过对抗性操控的身份。与传统的身份攻击方法不同，我们的方法在单步去噪过程中为每个源-目标对优化轻量级的交叉注意力适配器。潜在混合受到面部局部热图掩码的约束，以确保空间上精确的身份操控，同时保留非敏感区域。我们引入了一个复合目标，整合了掩蔽的epsilon-MSE重建、FR嵌入空间中的阈值身份差异、方向特征对齐和源相似性抑制，以平衡对抗攻击和视觉真实感。可选地，LLaVA生成的属性提示增强了细粒度的语义细节，而不重新引入身份线索。在黑箱评估协议下，Adv-TGD在IR152、IRSE50、MobileFace和FaceNet上的平均攻击成功率（ASR）达到85.90%，超越了语义最先进基线Adv-CPG 6.25个百分点、基于扩散的化妆方法DiffAIM 3个百分点，以及基于噪声的P3-Mask 16个百分点。尽管攻击效果强劲，Adv-TGD仍保持高视觉保真度（PSNR = 27.15 dB，SSIM = 0.981）。此外，我们通过成功将其扩展到野外数据集（LADN）、一般物体分类（ImageNet）和基于变换器的扩散模型（FLUX.1）展示了我们框架的灵活性。
cs.CV / 31 / 2606.11619
Precision-Aware Illumination-Disentangled Vision Transformer for Spacecraft 6D Pose Estimation
精度感知的光照解耦视觉变换器用于航天器6D姿态估计
Abstract
Vision sensors provide a lightweight solution for spacecraft proximity operations, but monocular spacecraft 6D pose estimation remains difficult under illumination variation, specular reflection, shadowing, weak texture, and background interference. These factors make local visual evidence spatially unreliable and can destabilize pose regression. This article proposes a Precision-Aware Illumination-Disentangled Vision Transformer (PAID-ViT) for robust spacecraft pose estimation.The proposed model separates pose-relevant structure tokens from illumination-sensitive appearance tokens, estimates patch reliability before pose aggregation, and uses foreground mask supervision to preserve silhouette cues. A parameter-free geometric recovery module converts normalized crop coordinates, log-depth, and a continuous 6D rotation representation into camera-frame rotation and translation. Experiments on SPEED+ V2, the SPEED+ validation/lightbox/sunlamp evaluation configuration used in this study, suggest that PAID-ViT reduces translation error and improves robustness in the challenging sunlamp domain, while ablation studies support the complementary roles of illumination disentanglement, reliability-aware token aggregation, mask supervision, and training-side regularization.
Chinese Translation
视觉传感器为航天器近距离操作提供了一种轻量级解决方案，但在光照变化、镜面反射、阴影、纹理较弱和背景干扰等因素影响下，单目航天器的6D姿态估计依然困难。这些因素使得局部视觉证据在空间上不可靠，并可能导致姿态回归的不稳定。本文提出了一种精度感知的光照解耦视觉变换器（PAID-ViT），用于稳健的航天器姿态估计。所提模型将与姿态相关的结构标记与对光照敏感的外观标记分离，在姿态聚合之前估计补丁的可靠性，并使用前景掩码监督来保留轮廓线索。一个无参数的几何恢复模块将归一化裁剪坐标、对数深度和连续的6D旋转表示转换为相机坐标系下的旋转和平移。针对SPEED+ V2的实验结果表明，PAID-ViT在具有挑战性的日光灯领域中减少了平移误差并提高了稳健性，而消融研究则支持了光照解耦、可靠性感知标记聚合、掩码监督和训练侧正则化的互补作用。
cs.CV / 32 / 2606.11626
Adapting Vision-Language Models from Iconic to Inclusive for Multi-Label Recognition Without Labels
将视觉-语言模型从标志性适应为包容性，以实现无标签的多标签识别
Abstract
Understanding multi-label images remains a challenging task in computer vision. With the rapid progress of vision-language multimodal learning, vision-language models (VLMs) enable zero-shot recognition without labeled data. However, due to their intrinsic design, these models often prioritize the most iconic object and omit other contextual positives. This intrinsic bias conflicts with the nature of multi-label learning, thereby limiting their applicability. In this work, we propose an unsupervised framework that adapts VLMs from iconic recognition toward inclusive understanding, enabling label-free multi-label image recognition. Our approach consists of two key stages, ``cutting'' and ``sewing'': In the cutting stage, we present the multi-sampling response estimator to prevent the model from concentrating only on one single object. In the second sewing stage, the multi-object blend adaptation is introduced to adjust the labels to better conform to the multi-label distribution while preserving the intrinsic characteristics of the original model within only one epoch. Extensive experiments show that our framework significantly outperforms existing unsupervised approaches on four public datasets, even surpassing several representative weakly supervised baselines. These results demonstrate the potential of adapting pre-trained VLMs for more comprehensive visual understanding without manual annotations. Our code is publicly available at https://github.com/iCVTEAM/TailorCLIP.
Chinese Translation
理解多标签图像仍然是计算机视觉中的一项挑战性任务。随着视觉-语言多模态学习的快速进展，视觉-语言模型（VLMs）能够在没有标注数据的情况下实现零样本识别。然而，由于其内在设计，这些模型往往优先考虑最具标志性的对象，而忽略其他上下文中的正例。这种内在偏差与多标签学习的本质相悖，从而限制了它们的适用性。在本研究中，我们提出了一种无监督框架，将VLMs从标志性识别适应为包容性理解，从而实现无标签的多标签图像识别。我们的方法包括两个关键阶段：“切割”和“缝合”：在切割阶段，我们提出了多采样响应估计器，以防止模型仅集中于单一对象。在第二个缝合阶段，引入了多对象混合适应，以调整标签，使其更好地符合多标签分布，同时在仅一个训练周期内保留原始模型的内在特征。大量实验表明，我们的框架在四个公共数据集上显著优于现有的无监督方法，甚至超过了几个具有代表性的弱监督基线。这些结果展示了将预训练的VLMs适应于更全面的视觉理解而无需手动标注的潜力。我们的代码已公开发布在 https://github.com/iCVTEAM/TailorCLIP。
cs.CV / 33 / 2606.11645
Motion Reinforces Appearance: RGB-Skeleton Gated Residual Fusion for Micro-Gesture Online Recognition
运动强化外观：RGB-骨骼门控残差融合用于微手势在线识别
Abstract
Micro-gesture analysis attracts increasing attention for inferring spontaneous emotion from subtle body movements. Micro-gesture online recognition, which localizes and classifies each gesture instance in untrimmed videos, is a core task in the 4th EI-MiGA-IJCAI Challenge. Compared with typical temporal action detection, MGR emphasizes the localization and classification of actions, requiring the model to output the start time, end time, and category of each micro-gesture. Moreover, since micro-gestures are highly spontaneous, relying solely on a single modality makes it difficult to capture the complete and accurate multi-modal cues. In this work, we propose DyFADet+, which extends DyFADet into a dual-stream RGB-skeleton framework. In our model, both modalities are projected into shared multi-scale temporal embeddings and fused through a gated residual module, which adaptively injects skeleton motion into the RGB representation rather than using naive concatenation. Finally, these fused features are decoded by a Dynamic TAD head for online classification and boundary regression. On the SMG dataset, our method achieves an F1 score of 40.88, ranking 2nd in the Micro-gesture Online Recognition track.
Chinese Translation
微手势分析因其从细微的身体动作中推断自发情感而受到越来越多的关注。微手势在线识别是在未剪辑视频中定位和分类每个手势实例的核心任务，属于第4届EI-MiGA-IJCAI挑战赛。与典型的时间动作检测相比，微手势识别（MGR）强调动作的定位和分类，要求模型输出每个微手势的开始时间、结束时间和类别。此外，由于微手势高度自发，仅依赖单一模态很难捕捉完整和准确的多模态线索。在本研究中，我们提出了DyFADet+，将DyFADet扩展为双流RGB-骨骼框架。在我们的模型中，两种模态被投影到共享的多尺度时间嵌入中，并通过门控残差模块进行融合，该模块自适应地将骨骼运动注入RGB表示，而不是使用简单的拼接。最后，这些融合特征通过动态TAD头进行解码，以实现在线分类和边界回归。在SMG数据集上，我们的方法实现了40.88的F1分数，在微手势在线识别赛道中排名第二。
cs.CV / 34 / 2606.11661
Learning Instance-Adaptive Low-Rank Orthogonal Subspaces for Clothes-Changing Person Re-Identification
学习实例自适应低秩正交子空间用于换衣服人物重识别
Abstract
Clothes-changing person re-identification (CC-ReID) aims to recognize individuals despite drastic appearance changes caused by clothing variation. While existing methods rely on adversarial learning to disentangle clothing features, we propose Ortho-ReID, which explicitly models a low-rank clothing subspace from VLM text descriptions and extracts clothing-invariant representations via direct geometric constraints. A critical component is our transformer-based Basis Maker, which refines a shared, low-dimensional clothing prior into an instance-adaptive low-rank subspace through cross-attention with image patches, enabling robust clothing feature extraction even under varying visibility conditions. This instance-adaptive subspace is supervised via alignment with clothing text embeddings, while identity features are extracted via a learnable projection head and geometrically constrained to be strictly orthogonal to it. Extensive experiments demonstrate state-of-the-art performance on PRCC (+5.9% top-1), Celeb-reID-light (+3.5%), and LaST (+5.3%), with competitive results on LTCC.
Chinese Translation
换衣服人物重识别（CC-ReID）旨在识别尽管由于服装变化而导致外观发生剧烈变化的个体。现有方法依赖对抗学习来解耦服装特征，而我们提出了Ortho-ReID，该方法明确建模来自VLM文本描述的低秩服装子空间，并通过直接几何约束提取服装不变表示。一个关键组件是我们的基于变换器的Basis Maker，它通过与图像块的交叉注意力将共享的低维服装先验精炼为实例自适应低秩子空间，从而在不同可见性条件下实现稳健的服装特征提取。该实例自适应子空间通过与服装文本嵌入的对齐进行监督，而身份特征则通过可学习的投影头提取，并在几何上约束为严格正交于此。大量实验表明，在PRCC上实现了最先进的性能（+5.9% top-1），在Celeb-reID-light上提高了3.5%，在LaST上提高了5.3%，在LTCC上也取得了具有竞争力的结果。
cs.CV / 35 / 2606.11670
ARGUS: Stacked Multi-View Identity Mosaic Injection for Subject-Preserving Video Generation
ARGUS：基于堆叠多视角身份马赛克注入的主体保留视频生成
Abstract
Subject-preserving video generation is not solved by frontal-face similarity alone: a generated person must remain recognizable across motion, large viewpoint changes, expression shifts, occlusion, scale variation, and conflicts among text, first-frame, and identity references. We argue that the central bottleneck is the point-reference paradigm, which collapses identity into a single static observation entangled with pose, accessories, lighting, background, and camera statistics. We introduce Argus, a Wan-based framework centered on Stacked Multi-View Identity Mosaic Injection (SMII). SMII converts MLLM-selected image/video identity evidence into a 3*3 stacked mosaic, synchronizes the mosaic with the current diffusion time, and injects it as negative-time read-only memory in Wan's native token space. This turns identity from an external clean adapter or a single reference image into a compact dynamic distribution. Around SMII, an MLLM Identity Director selects informative identity moments and resolves condition conflicts, while no-cross-pair counterfactual training, Temporal Identity Annealing, and Adaptive Self-Likeness Guidance improve robustness without paired subject-video supervision. We further release HardID-Celeb, a public-figure identity-stress benchmark, and introduce YawScore and OccScore to probe large-yaw and first-frame-occlusion robustness. Argus achieves state-of-the-art results on OpenS2V-Eval Human-Domain, reaching 64.38 Total Score, 71.86 FaceSim, 51.62 NexusScore, and 79.14 NaturalScore. On HardID-Celeb, Argus obtains 76.80 FaceSim and improves YawScore and OccScore by 12.60 and 15.10 points over the strongest baselines, demonstrating that dynamic identity memory and large-scale counterfactual self-supervision are highly effective for subject-preserving video generation.
Chinese Translation
主体保留的视频生成不仅仅依赖于正面人脸的相似性：生成的人物必须在运动、大视角变化、表情变化、遮挡、尺度变化以及文本、第一帧和身份参考之间的冲突中保持可识别性。我们认为，核心瓶颈在于点参考范式，它将身份压缩为一个与姿势、配饰、光照、背景和相机统计数据交织在一起的单一静态观察。我们提出了Argus，这是一个基于Wan的框架，中心是堆叠多视角身份马赛克注入（Stacked Multi-View Identity Mosaic Injection，SMII）。SMII将MLLM选择的图像/视频身份证据转换为3*3的堆叠马赛克，将马赛克与当前扩散时间同步，并将其作为负时间只读内存注入到Wan的原生令牌空间中。这将身份从外部干净适配器或单一参考图像转变为紧凑的动态分布。在SMII周围，MLLM身份导演选择信息丰富的身份时刻并解决条件冲突，而无交叉对抗训练、时间身份退火和自适应自相似指导在没有配对主体-视频监督的情况下提高了鲁棒性。我们进一步发布了HardID-Celeb，一个公共人物身份压力基准，并引入YawScore和OccScore来探测大偏航和第一帧遮挡的鲁棒性。Argus在OpenS2V-Eval人类领域上达到了最先进的结果，总分达到64.38，FaceSim为71.86，NexusScore为51.62，NaturalScore为79.14。在HardID-Celeb上，Argus获得了76.80的FaceSim，并在YawScore和OccScore上分别提高了12.60和15.10点，相比于最强基线，证明了动态身份记忆和大规模对抗自我监督在主体保留视频生成中的高效性。
cs.CV / 36 / 2606.11682
Parameter-Efficient Adapter Tuning for Tabular-Image Multimodal Learning
面向表格-图像多模态学习的参数高效适配器调优
Abstract
Tabular-image multimodal learning aims to improve predictive modeling by jointly using structured tabular attributes and visual data. Although pretrained encoders provide strong modality-specific representations, full fine-tuning can be computationally expensive, while keeping encoders frozen may limit task-specific adaptation. We propose the Tabular-Image Adapter (TI-Adapter), a modality-specific adapter-based fine-tuning framework for efficient multimodal adaptation. TI-Adapter freezes the pretrained tabular encoder and learns an adapter after the extracted tabular embedding, while adapting the image branch with embedding-level and bottleneck-level adapters instead of full fine-tuning. Experiments on 20 tabular-image datasets show that TI-Adapter achieves competitive or better predictive performance than full fine-tuning while using substantially fewer trainable parameters. Ablation studies further demonstrate the importance of adapter placement for balancing performance and practical efficiency.
Chinese Translation
表格-图像多模态学习旨在通过联合使用结构化的表格属性和视觉数据来改善预测建模。尽管预训练编码器提供了强大的模态特定表示，但全面微调可能在计算上代价高昂，而保持编码器冻结可能限制任务特定的适应性。我们提出了表格-图像适配器（Tabular-Image Adapter, TI-Adapter），这是一个基于适配器的微调框架，用于高效的多模态适应。TI-Adapter 冻结预训练的表格编码器，并在提取的表格嵌入之后学习一个适配器，同时使用嵌入级和瓶颈级适配器来适应图像分支，而不是进行全面微调。在20个表格-图像数据集上的实验表明，TI-Adapter 在使用显著更少的可训练参数的情况下，达到了与全面微调相当或更好的预测性能。消融研究进一步证明了适配器放置在性能与实际效率之间平衡的重要性。
cs.CV / 37 / 2606.11683
Reason, Then Re-reason: Cross-view Revisiting Improves Spatial Reasoning
推理，再推理：跨视角重访提升空间推理能力
Abstract
Spatial reasoning from egocentric videos is inherently challenging because the observable evidence is constrained by the camera trajectory. Existing methods rely on single-turn inference, forcing models to resolve geometric ambiguity through semantic priors rather than verifiable evidence. We argue that spatial reasoning should be revisitable: conclusions formed under limited evidence should remain open to revision when complementary viewpoints become available. Building on this insight, we propose Reason, then Re-reason (ReRe), a training-free, inference-time framework with two phases: in the Reason Phase, an MLLM forms a spatial hypothesis from the original video; in the Re-reason Phase, it verifies or revises the hypothesis by observing a synthesized novel-view video. To enable effective cross-view revisiting, we design a Geometry-to-Video pipeline that renders strategically complementary novel views from predicted 3D geometry. These views feature an elevated, oblique perspective with scene-spanning coverage, while preserving the MLLM's native video interface without architectural modifications. Extensive evaluations on VSI-Bench and STI-Bench demonstrate that ReRe substantially boosts open-source MLLMs to rival proprietary state-of-the-art performance. Project page: https://zhenjiemao.github.io/ReRe/
Chinese Translation
从自我中心视频进行空间推理本质上具有挑战性，因为可观察的证据受到相机轨迹的限制。现有方法依赖于单轮推理，迫使模型通过语义先验而非可验证的证据来解决几何模糊性。我们认为，空间推理应该是可重访的：在有限证据下形成的结论应在获得补充视角时保持开放以供修正。基于这一见解，我们提出了推理，再推理（Reason, then Re-reason，ReRe），这是一个无训练、推理时框架，分为两个阶段：在推理阶段（Reason Phase），一个多模态大语言模型（MLLM）从原始视频中形成空间假设；在再推理阶段（Re-reason Phase），它通过观察合成的新视角视频来验证或修正该假设。为了实现有效的跨视角重访，我们设计了一个几何到视频的管道（Geometry-to-Video），从预测的三维几何体渲染出战略性互补的新视角。这些视角具有提升的斜视角度和场景覆盖范围，同时在不修改架构的情况下保留了MLLM的原生视频接口。在VSI-Bench和STI-Bench上的广泛评估表明，ReRe显著提升了开源MLLM的性能，使其能够与专有的最先进技术相媲美。项目页面：https://zhenjiemao.github.io/ReRe/
cs.CV / 38 / 2606.11687
DroneShield-AI: A Multi-Modal Sensor Fusion Framework for Real-Time Autonomous Drone Threat Detection, Behavioral Intent Classification, and Swarm Intelligence in Contested Airspace
DroneShield-AI：一种用于实时自主无人机威胁检测、行为意图分类和在竞争空域中群体智能的多模态传感器融合框架
Abstract
Unmanned Aerial Vehicle (UAV) threats have emerged as a defining security challenge of the 21st century. This paper presents DroneShield-AI, a unified open framework integrating six processing layers: RF signal classification, acoustic motor-signature detection, YOLOv8-based visual detection, evidence-weighted sensor fusion, a Behavioral Intent Classification Engine (BICE), and a Graph Neural Network Swarm Intelligence Module (GNN-SIM). BICE introduces the first systematic six-class threat taxonomy for drone flight patterns, enabling predictive operator alerts with a 30-second advance-warning horizon. GNN-SIM is the first open framework for adversarial multi-drone formation analysis using Graph Attention Networks. Evaluated on three publicly available real-world datasets, the fused pipeline achieves 96.1% detection accuracy, 3.2% false alarm rate, AUC-ROC: 0.981, and 142ms end-to-end latency on commodity CPU-class hardware at approximately $500-$780 USD total system cost. All code, model weights, and simulation datasets are publicly released at submission.
Chinese Translation
无人机（UAV）威胁已成为21世纪一个重要的安全挑战。本文提出了DroneShield-AI，这是一个统一的开放框架，集成了六个处理层：射频信号分类、声学电机特征检测、基于YOLOv8的视觉检测、证据加权传感器融合、行为意图分类引擎（BICE）和图神经网络群体智能模块（GNN-SIM）。BICE引入了首个系统化的六类无人机飞行模式威胁分类法，使得操作员能够在30秒的预警时间内进行预测性警报。GNN-SIM是首个使用图注意力网络进行对抗性多无人机编队分析的开放框架。在三个公开可用的真实世界数据集上进行评估时，融合管道实现了96.1%的检测准确率，3.2%的误报率，AUC-ROC为0.981，且在约500-780美元的总系统成本下，在商品级CPU硬件上实现了142毫秒的端到端延迟。所有代码、模型权重和仿真数据集在提交时均已公开发布。
cs.CV / 39 / 2606.11689
RankVR: Low-Rank Structure Perception and Value Recalibration for Robust Composed Image Retrieval
RankVR：低秩结构感知与价值重校准用于鲁棒的组合图像检索
Abstract
Composed Image Retrieval (CIR) constitutes a pivotal paradigm requiring models to perform joint reasoning on reference images and modification texts. However, the prevalence of Noisy Triplet Correspondence (NTC) in large-scale datasets severely constrains model performance. Existing denoising methods either target binary mismatches or rely on scalar-based point-wise estimation, neglecting rich global structural correlations among sample populations and dynamic value variations during training, thereby yielding suboptimal results. This paper identifies two critical unresolved challenges: Global Structural Inconsistency of Semantic Correlations and Hard Sample Discrimination Uncertainty. To address these, we propose RankVR, a framework designed to construct a robust CIR model via global structure consistency and dynamic value perception. Specifically, we introduce the Global Structure Consistency Perception (GSCP) module, which utilizes the Effective Rank of the Correlation Matrix to decouple clean samples from structural noise. By measuring rank difference, GSCP identifies samples disrupting macroscopic semantic symmetry. Furthermore, we develop the Adaptive Semantic Value Calibration (ASVC) module to distinguish high-value hard clean samples. By integrating training potential and reliability, it dynamically quantifies the semantic value of each triplet, ensuring effective utilization of hard samples while suppressing noise characterized by logical conflicts. Extensive experiments on the FashionIQ and CIRR benchmark datasets demonstrate that RankVR significantly outperforms existing state-of-the-art methods, validating its superior robustness in noisy environments.
Chinese Translation
组合图像检索（CIR）构成了一种关键范式，要求模型对参考图像和修改文本进行联合推理。然而，大规模数据集中噪声三元组对应（NTC）的普遍存在严重限制了模型的性能。现有的去噪方法要么针对二元不匹配，要么依赖于基于标量的逐点估计，忽视了样本群体之间丰富的全局结构相关性和训练过程中的动态价值变化，从而导致次优结果。本文识别出两个关键的未解决挑战：语义相关性的全局结构不一致性和困难样本判别的不确定性。为了解决这些问题，我们提出了RankVR，一个旨在通过全局结构一致性和动态价值感知构建鲁棒CIR模型的框架。具体而言，我们引入了全局结构一致性感知（GSCP）模块，该模块利用相关矩阵的有效秩将干净样本与结构噪声解耦。通过测量秩差，GSCP识别出破坏宏观语义对称的样本。此外，我们开发了自适应语义价值校准（ASVC）模块，以区分高价值的困难干净样本。通过整合训练潜力和可靠性，它动态量化每个三元组的语义价值，确保有效利用困难样本，同时抑制具有逻辑冲突特征的噪声。在FashionIQ和CIRR基准数据集上的广泛实验表明，RankVR显著优于现有的最先进方法，验证了其在噪声环境中的卓越鲁棒性。
cs.CV / 40 / 2606.11702
MedCTA: A Benchmark for Clinical Tool Agents
MedCTA：临床工具代理的基准测试
Abstract
To make clinically grounded decisions, medical AI agents are expected to go beyond simple recognition and be capable of tool retrieval, evidence acquisition, and integration. Existing benchmarks largely evaluate isolated perception or single-turn question answering, and therefore provide limited visibility into failures of planning, tool recruitment, and rollout reliability. We introduce MedCTA, a benchmark for evaluating medical tool agents on clinician-validated, step-implicit tasks grounded in realistic multimodal clinical inputs, including radiology images, pathology slides, and reports. MedCTA comprises 107 real-world clinical tasks with clinician-verified executable trajectories over 5 deployed tools, and supports process-aware evaluation of tool selection, argument validity, execution stability, trajectory fidelity, and outcome quality. We benchmark 18 open- and closed-source multimodal models and find that even frontier systems remain brittle in multi-step clinical tool use: autonomous rollouts are dominated by protocol failures, premature stopping, and incorrect tool recruitment, while gold-standard tool routing yields large but still incomplete gains. These results show that strong backbone perception does not translate into reliable agentic behavior in clinical settings. MedCTA provides a rigorous testbed for auditing, diagnosing, and advancing trustworthy medical AI agents. The dataset and evaluation suite are available at https://ivul-kaust.github.io/MedCTA/
Chinese Translation
为了做出基于临床的决策，医疗人工智能代理被期望超越简单的识别，具备工具检索、证据获取和整合的能力。现有的基准测试主要评估孤立的感知或单轮问答，因此对规划、工具招募和实施可靠性的失败提供了有限的洞察。我们引入了MedCTA，这是一个用于评估医疗工具代理的基准，基于临床验证的、隐含步骤的任务，结合了现实的多模态临床输入，包括放射学图像、病理切片和报告。MedCTA包含107个真实世界的临床任务，具有经过临床验证的可执行轨迹，涵盖5种已部署的工具，并支持对工具选择、论证有效性、执行稳定性、轨迹保真度和结果质量的过程感知评估。我们对18个开源和闭源的多模态模型进行了基准测试，发现即使是前沿系统在多步骤临床工具使用中仍然脆弱：自主实施受到协议失败、过早停止和错误工具招募的主导，而黄金标准的工具路由虽然带来了显著但仍不完整的收益。这些结果表明，强大的基础感知并不能转化为临床环境中可靠的代理行为。MedCTA为审计、诊断和推进可信赖的医疗人工智能代理提供了一个严格的测试平台。数据集和评估套件可在 https://ivul-kaust.github.io/MedCTA/ 获取。
cs.CV / 41 / 2606.11710
ERN-Net : Evolving Reason Node-Net for Document Binarization
ERN-Net：用于文档二值化的演化推理节点网络
Abstract
This paper presents ERN-Net, an Evolving Reason Node-Net for efficient document image binarization. ERN-Net enhances degradation-sensitive regions, such as faint strokes, broken characters, and noisy backgrounds, through evolving reason nodes and multi-scale reasoning. We further compare ResNet-101, ConvNeXt-Tiny, and ConvNeXt-Base, and find that ConvNeXt-Tiny provides the best practical trade-off between accuracy and memory usage. In addition, DIBCO-based pretraining improves binarization performance without increasing model memory consumption, requiring only about 1.5 additional training hours. Experiments on DIBCO-style benchmarks show that ERN-Net is effective under low-data and low-memory settings.
Chinese Translation
本文提出了ERN-Net，一种用于高效文档图像二值化的演化推理节点网络。ERN-Net通过演化推理节点和多尺度推理增强了对退化敏感区域的处理，如微弱笔画、破损字符和噪声背景。我们进一步比较了ResNet-101、ConvNeXt-Tiny和ConvNeXt-Base，发现ConvNeXt-Tiny在准确性和内存使用之间提供了最佳的实际权衡。此外，基于DIBCO的预训练在不增加模型内存消耗的情况下提高了二值化性能，仅需约1.5小时的额外训练时间。在DIBCO风格的基准测试中，实验表明ERN-Net在低数据和低内存设置下是有效的。
cs.CV / 42 / 2606.11719
Ouroboros-Spatial: Closing the Data-Model Loop for Spatial Reasoning
Ouroboros-Spatial：闭合空间推理的数据-模型循环
Abstract
Spatial reasoning remains a persistent challenge for multimodal large language models (MLLMs). Existing approaches largely rely on large-scale, statically curated datasets, where all training samples are treated uniformly regardless of the model's evolving capabilities. This static paradigm is inherently data-inefficient: training capacity is often spent on samples that are either trivial or overly difficult for the model at its current stage. To address this limitation, we propose Ouroboros-Spatial, a self-evolving training framework in which the model plays dual roles as a proposer and a solver. In each iteration, a frozen proposer generates spatial question-answer (QA) pairs from 3D scene metadata and raw video frames, together with executable code for deriving reliable ground truth. A learnable solver is then fine-tuned on the accepted samples, and its per-sample prediction confidence is used as a difficulty signal. This signal is fed back to the proposer in the next iteration, guiding it to generate questions better matched to the solver's current capabilities. Through this closed-loop design, the training distribution co-evolves with model ability, reducing redundant trivial examples while filtering out ambiguous or uninformative samples with limited learning value. Across six spatial reasoning benchmarks, Ouroboros-Spatial substantially improves Qwen3-VL-4B and Qwen3-VL-8B while using an order of magnitude fewer training examples than recent large-scale curated datasets. On VSI-Bench, it yields absolute gains of 9.9 and 6.8 points for the 4B and 8B models, respectively, enabling both to outperform a wide range of strong open-source and proprietary baselines.
Chinese Translation
空间推理仍然是多模态大型语言模型（MLLMs）面临的一个持续挑战。现有的方法主要依赖于大规模、静态策划的数据集，其中所有训练样本都被统一对待，而不考虑模型能力的不断发展。这种静态范式本质上是数据低效的：训练能力往往被浪费在对模型当前阶段而言，要么是微不足道，要么是过于困难的样本上。为了解决这一局限性，我们提出了Ouroboros-Spatial，一个自我演化的训练框架，在该框架中，模型同时扮演提议者和求解者的双重角色。在每次迭代中，一个冻结的提议者从3D场景元数据和原始视频帧中生成空间问答（QA）对，并提供可执行代码以推导可靠的真实值。然后，一个可学习的求解者在接受的样本上进行微调，其每个样本的预测置信度被用作难度信号。这个信号在下一次迭代中反馈给提议者，引导其生成与求解者当前能力更匹配的问题。通过这种闭环设计，训练分布与模型能力共同演化，减少冗余的微不足道示例，同时过滤出模糊或信息量有限的样本。在六个空间推理基准测试中，Ouroboros-Spatial显著提高了Qwen3-VL-4B和Qwen3-VL-8B的性能，同时使用的训练示例数量比最近的大规模策划数据集少一个数量级。在VSI-Bench上，4B和8B模型分别获得了9.9和6.8的绝对增益，使得两者都超越了多种强大的开源和专有基线。
cs.CV / 43 / 2606.11739
Multi-View In-Cabin Monitoring System for Public Transport Vehicles
公共交通车辆的多视角车内监控系统
Abstract
We introduce a multi-view in-cabin monitoring dataset for public transportation with synchronized RGB and depth images from four inward-facing cameras and a rotating LiDAR covering the vehicle interior of a digitalized and partly automated German city bus. The dataset contains 9.136 synchronized samples with annotations and is accompanied by a calibration and pseudo-labeling pipeline that generates 3D human pose estimates and oriented 3D bounding boxes for occupants. We further provide a nuScenes-format conversion and benchmark representative multi-view 3D detection models (e.g., Lift-Splat-Shoot and BEVFusion), supporting comparative evaluation and small-scale training of multi-view in-cabin perception models. The dataset and tools are available at https://github.com/EvgenyGorelik/multiview_incabin_dataset.
Chinese Translation
我们介绍了一个用于公共交通的多视角车内监控数据集，该数据集包含来自四个朝内摄像头和一个旋转激光雷达的同步RGB和深度图像，覆盖了数字化和部分自动化的德国城市公交车的车内环境。该数据集包含9,136个带注释的同步样本，并配有一个校准和伪标注管道，用于生成3D人体姿态估计和面向乘客的3D边界框。此外，我们还提供了nuScenes格式的转换，并基准测试了代表性的多视角3D检测模型（例如，Lift-Splat-Shoot和BEVFusion），以支持多视角车内感知模型的比较评估和小规模训练。数据集和工具可在 https://github.com/EvgenyGorelik/multiview_incabin_dataset 获取。
cs.CV / 44 / 2606.11740
UniReason-Med: A Shared Grounded Reasoning Interface for 2D-to-3D Transfer in Medical VQA
UniReason-Med：用于医学视觉问答中2D到3D转移的共享基础推理接口
Abstract
We study whether grounded reasoning supervision from abundant 2D medical images can improve 3D medical VQA when both input types are aligned through a common reasoning interface. We introduce UniReason-Med, a single-checkpoint framework that processes either a 2D image or a slice-serialized 3D volume at inference time, generating interleaved textual reasoning and localized visual evidence through shared box syntax, region-token injection, and a common grounded reasoning policy. To train this interface, we construct UniMed-CoT, a 220K instruction-tuning dataset with interleaved textual reasoning and grounded visual evidence, including 170K 2D and 50K 3D samples. Through supervised fine-tuning followed by outcome-level reinforcement learning, UniReason-Med learns to generate grounded reasoning traces without IoU/Dice-based localization rewards during RL. Data-mixture and component ablations show that joint 2D+3D grounded supervision substantially improves 3D reasoning over 3D-only training, while grounding and region-token injection consistently benefit both 2D and 3D tasks. These results suggest that a shared grounded reasoning interface can transfer reasoning structure from 2D images to slice-serialized volumetric medical understanding. The code and data are publicly available at https://github.com/IQuestLab/unireason-med.
Chinese Translation
我们研究了来自丰富的2D医学图像的基础推理监督是否能够在通过共同推理接口对齐的情况下改善3D医学视觉问答（VQA）。我们提出了UniReason-Med，这是一个单检查点框架，在推理时处理2D图像或切片序列化的3D体积，通过共享框语法、区域标记注入和共同的基础推理策略生成交错的文本推理和局部视觉证据。为了训练这个接口，我们构建了UniMed-CoT，一个包含交错文本推理和基础视觉证据的220K指令调优数据集，其中包括170K个2D样本和50K个3D样本。通过监督微调后进行结果级强化学习，UniReason-Med学会在强化学习过程中生成基础推理轨迹，而不依赖于基于IoU/Dice的定位奖励。数据混合和组件消融实验表明，联合的2D+3D基础监督显著改善了3D推理，相较于仅使用3D训练，而基础和区域标记注入始终对2D和3D任务都有益。这些结果表明，共享基础推理接口可以将推理结构从2D图像转移到切片序列化的体积医学理解。代码和数据可在https://github.com/IQuestLab/unireason-med公开获取。
cs.CV / 45 / 2606.11745
From Prompts to Tokens: Internalizing Causal Supervision in Vision-Language Model for Multi-Image Causal Reasoning
从提示到标记：在视觉-语言模型中内化因果监督以进行多图像因果推理
Abstract
Visual causal reasoning is essential for understanding and intervening in the physical world, requiring identification of causal variables from visual inputs and reasoning over intervention effects. Despite recent progress, large vision--language models (VLMs) remain brittle at such tasks, especially for interventional and counterfactual queries over multi-image inputs. Most existing explorations inject causal knowledge via textual prompts, leaving causal mechanisms external to model execution and limiting reliable control during inference. To address this problem, we propose BridgeVLM, which internalizes visual causal reasoning by inducing a causal graph from multi-image inputs and converting it into structured Causal Tokens executed by RAMP layers injected into the LLM decoder for causal message passing. We further introduce a unified training interface M3S for fine-grained causal supervision from different granularities (local/global level). BridgeVLM achieves 54.4% accuracy on intervention tasks on CausalVLBench (vs. 33.2% with prompt-level supervision), improves results on Causal3D from 43.6% to 49.0%, and substantially improves causal structure learning on CausalVLBench ($F_1$: 33.4% $\rightarrow$ 75.1%).
Chinese Translation
视觉因果推理对于理解和干预物理世界至关重要，要求从视觉输入中识别因果变量并推理干预效果。尽管近年来取得了一定进展，但大型视觉-语言模型（VLMs）在此类任务上仍然脆弱，尤其是在多图像输入的干预和反事实查询方面。现有的大多数研究通过文本提示注入因果知识，使得因果机制在模型执行之外，限制了推理过程中的可靠控制。为了解决这个问题，我们提出了BridgeVLM，它通过从多图像输入诱导因果图并将其转换为结构化的因果标记（Causal Tokens），在LLM解码器中注入RAMP层以实现因果信息传递，从而内化视觉因果推理。我们进一步引入了统一训练接口M3S，以实现不同粒度（局部/全局级别）的细粒度因果监督。BridgeVLM在CausalVLBench的干预任务上达到了54.4%的准确率（相比之下，使用提示级监督时为33.2%），在Causal3D上的结果从43.6%提高到49.0%，并显著改善了CausalVLBench上的因果结构学习（$F_1$: 33.4% $
ightarrow$ 75.1%）。
cs.CV / 46 / 2606.11751
AnchorEdit: Maintaining Temporal Consistency in Multi-turn Image Editing via Causal Memory
AnchorEdit：通过因果记忆在多轮图像编辑中保持时间一致性
Abstract
Multi-turn image editing is essential for iterative design, yet current models often struggle with identity drift and error accumulation over successive steps. While existing research leverages video priors for consistency, their reliance on bidirectional attention is fundamentally misaligned with the causal, sequential nature of interactive editing. In this paper, we propose AnchorEdit, the first autoregressive (AR) diffusion-based framework designed specifically for high-resolution, long-term multi-turn editing. AnchorEdit bridges the gap between video priors and causal inference through a three-stage training curriculum: identity-preserving sing-turn pretraining, causal AR forcing fine-tuning with a novel self-rollout strategy to mitigate exposure bias, and consistency distillation for efficient 4-step generation. During inference, we introduce a memory mechanism to anchor the initial subject identity and ensure stable extrapolation across extended editing trajectories. To evaluate performance, we provide a new high-resolution multi-turn editing benchmark designed to stress-test long-horizon stability. Extensive experiments demonstrate that AnchorEdit achieves state-of-the-art results, maintaining exceptional subject fidelity and instruction following even over 10+ interaction rounds.
Chinese Translation
多轮图像编辑对于迭代设计至关重要，但当前模型在连续步骤中常常面临身份漂移和错误积累的问题。尽管现有研究利用视频先验来实现一致性，但它们对双向注意力的依赖与交互编辑的因果性和顺序性本质上不相符。本文提出了AnchorEdit，这是第一个专为高分辨率、长期多轮编辑设计的自回归（AR）扩散基础框架。AnchorEdit通过三阶段训练课程弥合了视频先验与因果推理之间的差距：身份保持的单轮预训练、采用新颖的自回滚策略进行的因果AR强制微调以减轻曝光偏差，以及用于高效4步生成的一致性蒸馏。在推理过程中，我们引入了一种记忆机制，以锚定初始主体身份，并确保在扩展编辑轨迹中的稳定外推。为了评估性能，我们提供了一个新的高分辨率多轮编辑基准，旨在压力测试长期稳定性。大量实验表明，AnchorEdit实现了最先进的结果，即使在超过10轮的交互中也能保持卓越的主体保真度和指令遵循。
cs.CV / 47 / 2606.11779
Battery detection of XRay images using transfer learning
基于迁移学习的X射线图像电池检测
Abstract
The need for detecting and sorting batteries is drastically increasing for many applications. This study proves the potential of transfer learning in predicting whether the image contains a battery or not, the location and identifying three types of batteries, namely: prismatic, pouch, and cylindrical Lithium-Ion Batteries (LIB). Particularly, it focuses on the transfer learning method in two applications: Training a large-scale dataset to detect electronic devices using a pre-trained YOLOv5m, then using these latter trained weights to detect and classify the batteries. The precision of battery detection achieves 94%, which outperforms the pretrained YOLOv5m weights with 5%, in 22 ms inference time.
Chinese Translation
对电池的检测和分类需求在许多应用中急剧增加。本研究证明了迁移学习在预测图像是否包含电池、定位电池以及识别三种类型电池的潜力，这三种电池分别是：棱柱形锂离子电池（Lithium-Ion Batteries, LIB）、软包电池和圆柱形电池。特别地，本研究集中于迁移学习方法在两个应用中的表现：使用预训练的YOLOv5m训练大规模数据集以检测电子设备，然后利用这些训练后的权重来检测和分类电池。电池检测的精度达到了94%，比预训练的YOLOv5m权重提高了5%，推理时间为22毫秒。
cs.CV / 48 / 2606.11782
Seeing What Matters: Perceptual Wrapper with Common Randomness for 3D Gaussian Splatting
关注重要内容：具有共同随机性的3D高斯点云的感知包装器
Abstract
While 3D Gaussian Splatting (3DGS) achieves impressive real-time rendering, it frequently struggles to synthesize high-frequency textures, a limitation heavily exacerbated in memory-constrained and rate-distortion-optimized (RDO) pipelines. To address this, we propose a versatile 2D perceptual wrapper that enhances the rendered outputs of existing 3DGS representations in a content- and view-dependent manner. Our method leverages a lightweight synthesis network conditioned on pseudo-random Gaussian noise to synthesize perceptually plausible textures. Supervised by Wasserstein Distortion, the network learns to match local feature statistics rather than strictly enforcing pixel-wise reconstruction fidelity, effectively mitigating the blurriness inherent in standard frameworks. We demonstrate the broad applicability of our plug-and-play approach across vanilla, memory-constrained, and RDO 3DGS methods. Comprehensive subjective and objective experiments confirm that our method significantly improves over existing baselines, yielding superior perceptual quality at sharply reduced file or model sizes.
Chinese Translation
尽管3D高斯点云（3DGS）实现了令人印象深刻的实时渲染，但在合成高频纹理方面常常面临挑战，这一限制在内存受限和率失真优化（RDO）管道中尤为严重。为了解决这一问题，我们提出了一种多功能的二维感知包装器，以内容和视角依赖的方式增强现有3DGS表示的渲染输出。我们的方法利用一个轻量级的合成网络，该网络以伪随机高斯噪声为条件，合成感知上合理的纹理。在Wasserstein失真的监督下，网络学习匹配局部特征统计，而不是严格执行逐像素重建保真度，从而有效减轻了标准框架中固有的模糊性。我们展示了该即插即用方法在普通、内存受限和RDO 3DGS方法中的广泛适用性。全面的主观和客观实验确认我们的方法显著优于现有基线，在显著减少文件或模型大小的同时，提供了更优的感知质量。
cs.CV / 49 / 2606.11783
A Comprehensive Ecosystem for Open-Domain Customized Video Generation
开放领域定制视频生成的综合生态系统
Abstract
Recent progress in video generation has shown impressive visual synthesis capabilities. However, open-domain customized video generation remains limited by the lack of large-scale, annotated datasets capturing diverse identity-specific attributes. To address this, we introduce PexelsCustom-1M, the first publicly available million-scale dataset for identity-preserving video generation, containing one million curated triplets across 8,000+ categories. Leveraging this, we propose CustoMDiT, a parameter-efficient framework that adapts a pretrained multimodal Diffusion Transformer into a customized video generator with only 8% additional learnable parameters. Our method surpasses prior state-of-the-art. However, benchmarks such as DreamBooth cover only 100 classes, which is insufficient for real-world applications. To overcome this, we construct OpenCustom, a new benchmark with 1,000+ categories, created via cross-dataset knowledge fusion from ImageNet and MS-COCO. Extensive experiments confirm the advantages of both our dataset and model. We will open-source the entire ecosystem--including dataset, pipeline, benchmark, and implementations--to support further research.
Chinese Translation
近期视频生成的进展展示了令人印象深刻的视觉合成能力。然而，开放领域的定制视频生成仍然受到缺乏大规模注释数据集的限制，这些数据集能够捕捉多样的特定身份属性。为了解决这一问题，我们引入了PexelsCustom-1M，这是首个公开可用的百万规模身份保留视频生成数据集，包含来自8000多个类别的一百万个策划的<身份，文本，视频>三元组。基于此，我们提出了CustoMDiT，一个参数高效的框架，它将预训练的多模态扩散变换器（Diffusion Transformer）适配为定制视频生成器，仅需额外8%的可学习参数。我们的方法超越了之前的最先进技术。然而，像DreamBooth这样的基准仅覆盖100个类别，这对于现实世界应用来说是不够的。为此，我们构建了OpenCustom，一个新的基准，包含1000多个类别，通过从ImageNet和MS-COCO的跨数据集知识融合创建。大量实验确认了我们数据集和模型的优势。我们将开源整个生态系统——包括数据集、管道、基准和实现——以支持进一步的研究。
cs.CV / 50 / 2606.11792
MultiToP: Learning to Patch Visual Tokens to Mitigate Hallucinations in Video Large Multimodal Models
MultiToP：学习修补视觉标记以减轻视频大型多模态模型中的幻觉现象
Abstract
Video Large Multimodal Models have achieved remarkable progress in video understanding, yet they remain prone to hallucinations, where generated responses are not faithfully supported by the input video. In this paper, we propose MultiToP, a multimodal-context-aware visual token patching framework that mitigates hallucinations by refining unreliable visual tokens before language generation. MultiToP introduces a lightweight Visual Token Patcher to predict token-level replacement distributions and selectively substitute unreliable visual tokens with a dynamic global patch token. To train the patcher effectively, we further propose information-guided rank calibration, which uses answer-conditioned frame-level information cues derived from the backbone to guide token replacement. Combined with ground-truth answer supervision and sparsity regularization, MultiToP enables localized visual evidence refinement without modifying the original model. Extensive experiments demonstrate that MultiToP effectively reduces hallucinations on Vript-HAL with negligible inference overhead, improving the F1 scores of Qwen3-VL-4B-Instruct by 50.60% over the vanilla model. Meanwhile, MultiToP preserves general video understanding ability, yielding an 18.58% relative accuracy gain on ActivityNet-QA for Video-LLaVA-7B.
Chinese Translation
视频大型多模态模型在视频理解方面取得了显著进展，但它们仍然容易出现幻觉现象，即生成的响应未能忠实地反映输入视频。在本文中，我们提出了MultiToP，一个多模态上下文感知的视觉标记修补框架，通过在语言生成之前精炼不可靠的视觉标记来减轻幻觉现象。MultiToP引入了一种轻量级的视觉标记修补器，用于预测标记级替换分布，并选择性地用动态全局修补标记替换不可靠的视觉标记。为了有效训练修补器，我们进一步提出了信息引导的排名校准，该方法利用从主干网络中提取的基于答案的帧级信息线索来指导标记替换。结合真实答案监督和稀疏正则化，MultiToP实现了局部视觉证据的精炼，而无需修改原始模型。大量实验表明，MultiToP在Vript-HAL上有效减少了幻觉现象，推理开销微乎其微，使Qwen3-VL-4B-Instruct的F1分数比原始模型提高了50.60%。同时，MultiToP保持了良好的视频理解能力，在Video-LLaVA-7B的ActivityNet-QA上实现了18.58%的相对准确率提升。
cs.CV / 51 / 2606.11805
TextHOI-3D: Text-to-3D Hand-Object Interaction via Discrete Multi-View Generation and Joint Mesh Optimization
TextHOI-3D：通过离散多视图生成和联合网格优化实现文本到3D手-物体交互
Abstract
Text-conditioned 3D generation has progressed rapidly for images and isolated objects, but producing a hand-object mesh remains challenging: the output must preserve language semantics, cross-view consistency, object geometry, articulated hand shape, and physically plausible contact. We present TextHOI-3D, a staged framework that uses generated multi-view observations as an explicit interface between text-conditioned visual generation and geometry-aware hand-object recovery. TextHOI-3D learns a compact VQ token space for fixed-camera hand-object observations, predicts multi-view visual tokens from text with a CLIP-conditioned visual autoregressive model, and recovers a unified hand-object mesh through prior initialization, multi-view joint optimization, and anti-penetration refinement. The design separates semantic generation from geometric recovery while keeping both stages connected by a discrete multi-view representation. On HO3D-derived evaluations, the multi-view setting reduces object CD from 17.26 mm to 4.92 mm and penetration volume from 5.3721 cm^3 to 0.2193 cm^3 compared with a single-view counterpart, while improving hand errors and surface F-scores. These results support multi-view visual tokens as an effective intermediate representation for text-driven 3D hand-object mesh creation.
Chinese Translation
基于文本的3D生成在图像和孤立物体方面取得了快速进展，但生成手-物体网格仍然具有挑战性：输出必须保留语言语义、跨视图一致性、物体几何形状、关节手形状以及物理上合理的接触。我们提出了TextHOI-3D，这是一个分阶段框架，利用生成的多视图观测作为文本条件视觉生成与几何感知手-物体恢复之间的显式接口。TextHOI-3D为固定摄像头的手-物体观测学习了一个紧凑的VQ（向量量化）标记空间，通过一个基于CLIP（对比语言-图像预训练模型）条件的视觉自回归模型从文本预测多视图视觉标记，并通过先验初始化、多视图联合优化和抗穿透细化恢复统一的手-物体网格。该设计将语义生成与几何恢复分开，同时通过离散多视图表示将两个阶段连接起来。在基于HO3D的评估中，与单视图对应物相比，多视图设置将物体的CD（Chamfer Distance）从17.26 mm降低到4.92 mm，穿透体积从5.3721 cm^3降低到0.2193 cm^3，同时改善了手部误差和表面F分数。这些结果支持多视图视觉标记作为文本驱动的3D手-物体网格创建的有效中间表示。
cs.CV / 52 / 2606.11837
LASA: A Weak Supervision Method for Open-Vocabulary Scene Sketch Semantic Segmentation
LASA：一种用于开放词汇场景草图语义分割的弱监督方法
Abstract
Open-vocabulary scene sketch semantic segmentation aims to assign dense semantic labels to sparse line drawings based on flexible category vocabularies specified at inference time, without relying on pixel-level annotations during training. Unlike natural images, sketches lack texture and color cues, making semantic understanding heavily dependent on stroke layout and spatial configuration, a challenge that renders single-layer vision-language features inherently unstable. Our key observation is that attention maps from different Vision Transformer layers encode complementary spatial cues: shallow layers capture global structural layouts, while deeper layers focus on local stroke intersections and object parts. This suggests that cross-layer aggregation provides a more robust structural prior than any individual layer alone. Leveraging this insight, we propose a structure-aware framework built upon \textbf{L}ayer-wise \textbf{A}ccumulated \textbf{S}tructural \textbf{A}ttention (\textbf{LASA}), which aggregates multi-layer attention to guide hierarchical semantic alignment under weak supervision and refine predictions during inference. Experiments on FS-COCO, SFSD, and FrISS show that LASA improves mIoU by $+3.43$, $+8.01$, and $+15.74$ over the prior weakly supervised baselines, demonstrating consistent gains in both segmentation accuracy and spatial coherence. Our source code will be made publicly available.
Chinese Translation
开放词汇场景草图语义分割旨在根据推理时指定的灵活类别词汇，为稀疏的线条绘图分配密集的语义标签，而无需在训练过程中依赖像素级注释。与自然图像不同，草图缺乏纹理和颜色线索，使得语义理解在很大程度上依赖于笔画布局和空间配置，这一挑战使得单层视觉-语言特征本质上不稳定。我们的关键观察是，不同视觉变换器（Vision Transformer）层的注意力图编码了互补的空间线索：浅层捕捉全局结构布局，而深层则关注局部笔画交点和物体部分。这表明，跨层聚合提供了比任何单个层更强的结构先验。基于这一见解，我们提出了一种结构感知框架，建立在 extbf{L}ayer-wise extbf{A}ccumulated extbf{S}tructural extbf{A}ttention（ extbf{LASA}）之上，该框架聚合多层注意力，以在弱监督下指导分层语义对齐，并在推理过程中精炼预测。在FS-COCO、SFSD和FrISS上的实验表明，LASA在之前的弱监督基线之上提高了mIoU分别为$+3.43$、$+8.01$和$+15.74$，在分割准确性和空间一致性方面均表现出持续的提升。我们的源代码将公开发布。
cs.CV / 53 / 2606.11838
Plan-and-Verify Video Reward Reasoning with Spatio-Temporal Scene Graph Grounding
基于时空场景图的计划与验证视频奖励推理
Abstract
Reward models for text-to-video (T2V) generation guide post-training but often fail at fine-grained semantic alignment. We trace this to two structural weaknesses in existing reasoning-based reward models: they do not systematically verify every condition described in the prompt, and the visual evidence supporting each judgment remains implicit in their free-form reasoning. We propose SG-PVR, a video reward model that addresses these limitations through plan-and-verify reasoning grounded in spatio-temporal scene graphs. The verification plan decomposes the prompt into atomic claims, ensuring every requirement is checked. The spatio-temporal scene graph, encoding entities, attributes, and temporally-grounded relations, is extracted from the video and maintained as a persistent structured visual reference throughout reasoning. Each claim is verified against both the video and the scene graph, anchoring judgments in explicit visual evidence. SG-PVR achieves strong performance on semantic alignment, including fine-grained temporal semantics. As a test-time reranker, it further enhances compositional alignment in T2V generation.
Chinese Translation
文本到视频（T2V）生成的奖励模型指导后期训练，但在细粒度语义对齐方面常常失败。我们将此归因于现有基于推理的奖励模型的两个结构性弱点：它们没有系统地验证提示中描述的每个条件，并且支持每个判断的视觉证据在自由形式推理中仍然是隐含的。我们提出了SG-PVR，这是一种视频奖励模型，通过基于时空场景图的计划与验证推理来解决这些局限性。验证计划将提示分解为原子声明，确保每个要求都得到检查。时空场景图编码了实体、属性和时间相关关系，从视频中提取并在推理过程中作为持久的结构化视觉参考。每个声明都与视频和场景图进行验证，将判断锚定在明确的视觉证据上。SG-PVR在语义对齐方面表现出色，包括细粒度的时间语义。作为测试时的重新排序器，它进一步增强了T2V生成中的组合对齐。
cs.CV / 54 / 2606.11841
Scene-Adaptive Nonlinear Tone Curves for Pseudo Ground-Truth Generation in Low-Light 3D Gaussian Splatting
场景自适应非线性色调曲线用于低光照3D高斯点云伪地面真值生成
Abstract
Low-light novel view synthesis is challenging because dark multi-view images contain noise, weak structural detail, and compressed dynamic range. Recent 3D Gaussian Splatting (3DGS) methods address these challenges by generating pseudo ground-truth (pseudo-GT) images as supervision targets when paired normal-light references are unavailable. Existing pseudo-GT methods apply a uniform linear gain to all pixels, which clips bright regions while providing insufficient enhancement in dark regions, limiting reconstruction quality. We observe that nonlinear tone mappings, long established in 2D low-light enhancement, have not been explored for pseudo-GT generation in 3D reconstruction. Accordingly, we propose a scene-adaptive nonlinear tone-curve framework that replaces linear pseudo-GT with nonlinear alternatives. The framework introduces percentile-based normalisation for scene-agnostic curve application, a scene-adaptive offset for automatic black-level adjustment, and two complementary curves: Adaptive SoftExp (ASE), a bounded exponential curve, and Adaptive Poly3 (AP3), a data-driven cubic polynomial. The module changes only the pseudo-GT computation and leaves the 3DGS backbone unchanged. Experiments on three benchmarks covering 21 scenes show that both curves consistently outperform the linear baseline with PSNR improvements up to +4.34 dB on LOM and +3.25 dB on RealX3D. Both curves achieve similar performance despite their different mathematical forms, suggesting the improvement is curve-agnostic. Code is available at https://github.com/lvmingzhe/adaptiveToneCurve
Chinese Translation
低光照新视图合成具有挑战性，因为黑暗的多视图图像包含噪声、微弱的结构细节和压缩的动态范围。最近的3D高斯点云（3DGS）方法通过生成伪地面真值（pseudo-GT）图像作为监督目标来应对这些挑战，尤其是在缺乏配对的正常光照参考时。现有的伪-GT方法对所有像素应用均匀的线性增益，这在亮区会被裁剪，而在暗区则提供了不足的增强，从而限制了重建质量。我们观察到，长期以来在2D低光照增强中建立的非线性色调映射尚未在3D重建中的伪-GT生成中得到探索。因此，我们提出了一种场景自适应非线性色调曲线框架，用非线性替代线性伪-GT。该框架引入基于百分位的归一化，以实现与场景无关的曲线应用，自动黑电平调整的场景自适应偏移，以及两条互补曲线：自适应软指数（Adaptive SoftExp, ASE），一种有界指数曲线，以及自适应三次多项式（Adaptive Poly3, AP3），一种数据驱动的三次多项式。该模块仅改变伪-GT的计算，而不改变3DGS的主干。在覆盖21个场景的三个基准测试中的实验表明，这两条曲线在PSNR上均一致优于线性基线，在LOM上提高了+4.34 dB，在RealX3D上提高了+3.25 dB。尽管这两条曲线的数学形式不同，但它们的性能相似，表明改进与曲线无关。代码可在 https://github.com/lvmingzhe/adaptiveToneCurve 获取。
cs.CV / 55 / 2606.11846
SheafStain: Sheaf-Theoretic Schr\"odinger Bridge for Spatially and Biologically Coherent Virtual Staining
SheafStain：用于空间和生物一致性虚拟染色的层叠理论薛定谔桥
Abstract
Current virtual staining approaches offer the potential for time- and cost-efficient biomarker quantification in cancer diagnostics and prognostics. However, patch-wise inference for gigapixel whole slide images (WSIs) fails to maintain spatial continuity, yielding artifacts that cause catastrophic mismatches with ground-truth images. Although pathology Vision Foundation Models (VFMs) offer rich representations, their self-attention causes varying global contexts to produce inconsistent embeddings for the same physical region. We formalize and validate this ``context contamination'' as a sheaf-theoretic problem where these embeddings form a presheaf that violates the gluing axiom. To address this, we propose SheafStain, a new approach that reinterprets VFM features as sheaf-like sections for spatially and biologically coherent virtual staining. Specifically, SheafStain integrates class and patch tokens into a Schr\"odinger Bridge framework as sheaf-like sections. While the class token anchors biological consistency, patch tokens form a per-position spatial map. A backbone co-pretrained on Hematoxylin \& Eosin (H\&E) and Immunohistochemistry (IHC) yields non-degenerate cross-stain stalks, so a single VFM feature space supervises both input conditioning and output stain alignment. Departing from prior work that evaluates on isolated $256 \times 256$ patches and either random-crops or resizes the $1024 \times 1024$ ground truth, we translate at $256 \times 256$ and evaluate on the stitched $1024 \times 1024$ outputs across HER2, ER, PR, and Ki-67. SheafStain demonstrates promising results against six prior methods while mitigating patch-boundary stitching artifacts. Code will soon be released.
Chinese Translation
当前的虚拟染色方法为癌症诊断和预后中的生物标志物定量提供了时间和成本效率的潜力。然而，针对千兆像素全切片图像（WSIs）的补丁推断未能保持空间连续性，导致产生伪影，从而与真实图像发生灾难性不匹配。尽管病理视觉基础模型（VFMs）提供了丰富的表示，但其自注意机制使得不同的全局上下文为同一物理区域产生不一致的嵌入。我们将这种“上下文污染”形式化并验证为一个层叠理论问题，其中这些嵌入形成一个违反粘合公理的预层叠。为了解决这一问题，我们提出了SheafStain，这是一种将VFM特征重新解释为层叠状节段以实现空间和生物一致性虚拟染色的新方法。具体而言，SheafStain将类别和补丁标记整合到薛定谔桥框架中作为层叠状节段。类别标记锚定生物一致性，而补丁标记形成每个位置的空间映射。一个在苏木精-伊红（HE）和免疫组化（IHC）上共同预训练的主干网络产生非退化的交叉染色干，因此单个VFM特征空间同时监督输入条件和输出染色对齐。与之前在孤立的$256 imes 256$补丁上评估并随机裁剪或调整$1024 imes 1024$真实图像大小的工作不同，我们在$256 imes 256$上进行转换，并在HER2、ER、PR和Ki-67的拼接$1024 imes 1024$输出上进行评估。SheafStain在六种先前方法中表现出良好的结果，同时减轻了补丁边界拼接伪影。代码将很快发布。
cs.CV / 56 / 2606.11853
Task-Aware Structured Memory for Dynamic Multi-modal In-Context Learning
任务感知的结构化记忆用于动态多模态上下文学习
Abstract
Multi-modal large language models (MLLMs) depend on in-context learning (ICL) for rapid task adaptation, but their scalability is severely limited by finite context windows and the growing cost of key-value (KV) caches in long multi-modal sequences. Existing memory compression approaches typically rely on rigid token removal or sample-dependent importance estimation, which introduces bias, disrupts semantic structure, particularly for visual representations, and yields static memories that cannot adapt to new queries. We introduce TASM (Task-Aware Structured Memory), a training-free framework that addresses these limitations through task-aware, structure-preserving, and dynamically accessible memory construction. TASM employs task-vector guided compression to replace sample-specific signals with a task-level direction that captures shared relevance across demonstrations. To preserve the underlying manifold, it applies semantics-aware token merging via bipartite graph matching, aggregating tokens without destructive pruning. Finally, TASM structures memory into a hierarchy comprising a compact Core Memory and a Latent Bank, facilitating query-adaptive dynamic retrieval. Evaluations confirm TASM maintains high performance under heavy compression, effectively balancing efficiency with adaptability.
Chinese Translation
多模态大型语言模型（MLLMs）依赖于上下文学习（ICL）进行快速任务适应，但其可扩展性受到有限上下文窗口和长多模态序列中键值（KV）缓存成本增加的严重限制。现有的记忆压缩方法通常依赖于僵化的标记移除或样本依赖的重要性估计，这会引入偏差，破坏语义结构，特别是对于视觉表示，并产生无法适应新查询的静态记忆。我们提出了TASM（任务感知结构化记忆），这是一个无训练的框架，通过任务感知、结构保持和动态可访问的记忆构建来解决这些限制。TASM采用任务向量引导的压缩，将样本特定信号替换为捕捉跨演示共享相关性的任务级方向。为了保持底层流形，它通过二分图匹配应用语义感知的标记合并，聚合标记而不进行破坏性修剪。最后，TASM将记忆结构化为一个包含紧凑核心记忆和潜在库的层次结构，便于查询自适应动态检索。评估结果确认TASM在高压缩下保持高性能，有效平衡了效率与适应性。
cs.CV / 57 / 2606.11880
SG2Loc: Sequential Visual Localization on 3D Scene Graphs
SG2Loc：基于3D场景图的序列视觉定位
Abstract
Visual localization in complex indoor environments remains a critical challenge for robotics and AR applications. Sequential localization, where pose estimates are refined over time, is important for autonomous agents. However, traditional methods often require storing extensive image databases or point clouds, leading to significant overhead. This paper introduces a novel, lightweight approach to sequential visual localization using 3D scene graphs. Our method represents the environment with a compact scene graph, where nodes represent objects (with coarse meshes) and edges encode spatial relationships. For each image in the localization phase, we extract per-patch semantic features, predicting object identities. Localization is performed within a particle filter framework. Each particle, representing a camera pose, projects the coarse object meshes from the scene graph into the image, assigning object identities to patches based on visibility. The similarity of the per-patch features, in the input image, and object features from the scene graph determines the weight of a particle. Subsequent images are incorporated sequentially, refining the pose estimate. By leveraging a compact scene graph and efficient semantic matching, our method significantly reduces storage while maintaining performance on real-world datasets. The code will be available at https://github.com/DmblnNicole/sg2loc.
Chinese Translation
在复杂的室内环境中，视觉定位仍然是机器人技术和增强现实应用面临的一个关键挑战。序列定位，即随着时间推移对姿态估计进行精细化，是自主代理的重要任务。然而，传统方法通常需要存储大量的图像数据库或点云，导致显著的开销。本文提出了一种新颖且轻量级的基于3D场景图的序列视觉定位方法。我们的方法使用紧凑的场景图表示环境，其中节点代表物体（带有粗糙网格），边缘编码空间关系。在定位阶段，对于每张图像，我们提取每个补丁的语义特征，预测物体身份。定位是在粒子滤波框架内进行的。每个粒子代表一个相机姿态，将场景图中的粗糙物体网格投影到图像中，根据可见性将物体身份分配给补丁。输入图像中每个补丁特征与场景图中物体特征的相似性决定了粒子的权重。后续图像被顺序地纳入，进一步精细化姿态估计。通过利用紧凑的场景图和高效的语义匹配，我们的方法显著减少了存储需求，同时在真实世界数据集上保持了性能。代码将发布在 https://github.com/DmblnNicole/sg2loc。
cs.CV / 58 / 2606.11884
Image Quality Assessment of Identity Cards Using Measures from Open Face Image Quality
基于开放人脸图像质量标准的身份证图像质量评估
Abstract
This paper addresses the challenge of assessing image quality in ID cards in remote verification systems by applying capture-related quality measures from the Open Face Image Quality (OFIQ) standard to ID card images. Our preprocessing pipeline includes corner detection, perspective normalization, and comprehensive foreground masking to ensure accurate and unbiased quality measure computation. We evaluate the effectiveness of these measures by analyzing their correlation with the performance of three presentation attack detection (PAD) algorithms across four diverse ID card datasets, where two datasets contain bona fide, i.e. pristine, images and two contain printed mock ID cards. Our results suggest that quality assessment based on some OFIQ measures can significantly improve PAD performance.
Chinese Translation
本文针对远程验证系统中身份证图像质量评估的挑战，通过将开放人脸图像质量（Open Face Image Quality, OFIQ）标准中的捕获相关质量度量应用于身份证图像来进行研究。我们的预处理流程包括角点检测、透视归一化和全面的前景遮罩，以确保准确和无偏的质量度量计算。我们通过分析这些度量与三种呈现攻击检测（Presentation Attack Detection, PAD）算法在四个不同身份证数据集上的性能相关性来评估这些度量的有效性，其中两个数据集包含真实的、原始的图像，两个数据集包含打印的伪造身份证。我们的结果表明，基于某些OFIQ度量的质量评估可以显著提高PAD的性能。
cs.CV / 59 / 2606.11889
Task-Aligned Stability Analysis of Vision-Language Models for Autonomous Driving Hazard Detection
面向任务的视觉-语言模型稳定性分析：用于自主驾驶危险检测
Abstract
Vision-language models (VLMs) are increasingly used for scene understanding in autonomous driving, but robustness analysis often relies on task-agnostic embedding stability alone. We study whether corruption-induced embedding drift predicts changes in a task-aligned hazard score derived from CLIP image-text similarities. Using controlled corruptions on BDD100K road scenes, we compare embedding drift against margin drift, defined as the change in hazard score under perturbation. The relationship is highly corruption-dependent: some families exhibit strong coupling between representation drift and decision drift, while others induce hazardous decision instability despite relatively modest embedding change. Furthermore, corruption families differ in failure direction: most suppress hazard detections via false negatives, while occlusion instead triggers false alarms, suggesting that benchmark design should account for asymmetric failure modes, not just overall instability rates. These results suggest that robustness benchmarks should include task-aligned stability measures in addition to embedding-level perturbation statistics.
Chinese Translation
视觉-语言模型（VLMs）在自主驾驶场景理解中的应用日益增加，但其鲁棒性分析往往仅依赖于与任务无关的嵌入稳定性。我们研究了由损坏引起的嵌入漂移是否能够预测基于 CLIP 图像-文本相似性得出的与任务对齐的危险评分的变化。通过对 BDD100K 道路场景进行控制性损坏，我们将嵌入漂移与边际漂移进行比较，后者定义为在扰动下危险评分的变化。这种关系高度依赖于损坏类型：某些类别表现出表示漂移与决策漂移之间的强耦合，而其他类别则在相对适度的嵌入变化下引发危险的决策不稳定。此外，损坏类别在失败方向上存在差异：大多数通过假阴性抑制危险检测，而遮挡则触发假警报，这表明基准设计应考虑不对称的失败模式，而不仅仅是整体不稳定率。这些结果表明，鲁棒性基准应在嵌入级扰动统计之外，纳入与任务对齐的稳定性度量。
cs.CV / 60 / 2606.11894
Wild3R: Feed-Forward 3D Gaussian Splatting from Unconstrained Sparse Photo Collection
Wild3R：来自无约束稀疏照片集合的前馈3D高斯溅射
Abstract
Feed-forward 3D Gaussian Splatting (3DGS) removes the need for time-consuming per-scene optimization required by traditional 3DGS. However, existing feed-forward approaches struggle with real-world photo collections that include diverse lighting conditions and transient objects. In this paper, we present Wild3R, a feed-forward approach for unconstrained sparse photo collections. The main bottleneck is the lack of training data that provides multiple viewpoints, a variety of illuminations, and transient variations necessary for learning robust scene representations. To address this, we introduce the WildCity dataset, which comprises 200 scenes, 170 lighting conditions, and transient objects, resulting in 337,500 images in total. By leveraging the dataset, our model learns appearance consistency across viewpoints conditioned on reference views, while removing transient content. Extensive experiments demonstrate that our method outperforms existing feed-forward approaches and achieves results competitive with prior per-scene optimization-based methods.
Chinese Translation
前馈3D高斯溅射（3DGS）消除了传统3DGS所需的耗时逐场景优化的需求。然而，现有的前馈方法在处理包含多样光照条件和瞬态物体的真实世界照片集合时遇到了困难。本文提出了Wild3R，一种针对无约束稀疏照片集合的前馈方法。主要瓶颈在于缺乏提供多个视角、多种光照和瞬态变化的训练数据，这些都是学习稳健场景表示所必需的。为了解决这个问题，我们引入了WildCity数据集，该数据集包含200个场景、170种光照条件和瞬态物体，总计337,500张图像。通过利用该数据集，我们的模型在参考视图的条件下学习了跨视角的一致性外观，同时去除了瞬态内容。大量实验表明，我们的方法优于现有的前馈方法，并且在结果上与之前基于逐场景优化的方法具有竞争力。
cs.CV / 61 / 2606.11913
From Content to Knowledge: Lightning Fast Long-Video Understanding with Neural Knowledge Representations
从内容到知识：基于神经知识表示的快速长视频理解
Abstract
We propose a new paradigm for long video understanding by treating a long video as a Neural Knowledge Representation (NKR). NKR represents video contents neither as a stream of tokens nor pre-organized databases, but as an individual small portion of network weights attached to the VLM backbone. The NKR weights are optimized to encapsulate the video's semantic content via a novel Agentic Knowledge Distillation (AKD) process, where an agent automatically synthesizes dense descriptions and question-answer pairs to distill the video's knowledge into the NKR. While AKD serves as a comprehensive, one-time encoding phase, the resulting NKR transforms the video into a portable, reusable asset. At inference, the lightweight NKR is mounted onto a frozen Vision-Language Model (VLM), enabling direct, query-based understanding without reloading or re-encoding the original video. This approach decouples video length from inference cost, offering high amortized efficiency for multi-turn video understanding. Experiments on the LVBench benchmark show our method achieves performance comparable to state-of-the-art approaches while reducing end-to-end latency by over two orders of magnitude, opening new possibilities for interactive long-video understanding.
Chinese Translation
我们提出了一种新的长视频理解范式，将长视频视为神经知识表示（Neural Knowledge Representation, NKR）。NKR并不将视频内容表示为一系列标记或预先组织的数据库，而是作为附加在视觉语言模型（Vision-Language Model, VLM）主干上的一小部分网络权重。NKR权重通过一种新颖的主动知识蒸馏（Agentic Knowledge Distillation, AKD）过程进行优化，以封装视频的语义内容，其中一个代理自动合成密集描述和问答对，将视频的知识蒸馏到NKR中。虽然AKD作为一个全面的一次性编码阶段，但生成的NKR将视频转化为可移植、可重用的资产。在推理时，轻量级的NKR被加载到一个冻结的视觉语言模型（VLM）上，实现直接的基于查询的理解，而无需重新加载或重新编码原始视频。这种方法将视频长度与推理成本解耦，为多轮视频理解提供了高效的摊销效率。在LVBench基准测试上的实验表明，我们的方法在性能上与最先进的方法相当，同时将端到端延迟减少了两个数量级以上，为互动长视频理解开辟了新的可能性。
cs.CV / 62 / 2606.11925
Corpus Augmentation for Sign Language Translation via LLM-Guided Video Stitching
通过LLM引导的视频拼接进行手语翻译的语料增强
Abstract
Sign language translation (SLT) converts sign language video into spoken language text and holds significant promise for improving accessibility and enabling communication between signing and non-signing communities. While large weakly-aligned datasets have enabled pre-training at scale and gloss-free methods have reduced reliance on expert annotation, high-quality parallel sign video-text pairs for fine-tuning remain scarce, limiting generalisation on long-tail vocabulary and unseen constructions. We propose a corpus augmentation approach that requires no additional human annotation, external sign-language video corpora, or generative video models, relying only on the existing gloss-annotated training corpus and an LLM for sentence generation: per-gloss clips are extracted from training videos via CTC forced-alignment, novel gloss-sentence pairs are generated by a corpus-anchored LLM, and synthetic sequences are assembled through random sentence sampling and clip assignment. The resulting synthetic RGB video-text pairs are architecture-agnostic at the downstream training stage and can be consumed directly by RGB-based SLT models, or converted into pose or feature representations by pipelines that derive such inputs from video. Sincan et al. re-evaluated five recent gloss-free methods under strictly identical conditions; the largest verified gain over the GFSLT-VLP baseline was only 0.98 BLEU-4. Our augmentation, applied within the same framework, achieves +2.92 BLEU-4 without any change to architecture or training protocol. We further identify that synthetic data harms vision-language pretraining despite improving its objectives, and that optimising clip transitions for visual smoothness is counter-productive under L2-based criteria; we propose that abrupt boundaries may act as a form of implicit regularisation. Code is available at https://github.com/robizso/slt-datagen.
Chinese Translation
手语翻译（SLT）将手语视频转换为口语文本，具有显著的潜力来改善无障碍性并促进手语与非手语社区之间的沟通。虽然大规模的弱对齐数据集使得大规模预训练成为可能，且无注释方法减少了对专家标注的依赖，但用于微调的高质量平行手语视频-文本对仍然稀缺，限制了对长尾词汇和未见结构的泛化。我们提出了一种语料增强方法，该方法不需要额外的人类标注、外部手语视频语料库或生成视频模型，仅依赖现有的带注释的训练语料库和用于句子生成的LLM：通过CTC强制对齐从训练视频中提取每个注释片段，通过语料锚定的LLM生成新的注释-句子对，并通过随机句子抽样和片段分配组装合成序列。生成的合成RGB视频-文本对在下游训练阶段与架构无关，可以被RGB基础的SLT模型直接使用，或通过从视频中提取此类输入的管道转换为姿态或特征表示。Sincan等人在严格相同的条件下重新评估了五种最新的无注释方法；在GFSLT-VLP基线之上验证的最大增益仅为0.98 BLEU-4。我们在相同框架内应用的增强方法实现了+2.92 BLEU-4，而无需对架构或训练协议进行任何更改。我们进一步发现，尽管合成数据改善了目标，但对视觉-语言预训练有害，并且在基于L2的标准下优化片段过渡以实现视觉平滑是适得其反的；我们提出突变边界可能作为一种隐式正则化形式。代码可在 https://github.com/robizso/slt-datagen 获取。
cs.CV / 63 / 2606.11966
Feature extraction for plant growth estimation
植物生长估计的特征提取
Abstract
Precision agriculture requires the estimation of plant growth stages in real-time. When the plant growth stage is known, the wastage of resources in cultivation, such as nutrients and water, is reduced as only the required resources need to be supplied. Plants at different growth stages, however, have similar morphological features, which can make autonomous growth stage estimation difficult. This paper presents two feature extraction methods for growth stage estimation: one that uses a bank of Gabor filters and morphological operations, and the other that uses pre-trained convolutional neural networks (CNNs) and transfer learning. We test these methods on a publicly available plant growth stage dataset (``bccr-segset``) for two species, canola and radish, grown and captured under indoor conditions. The two proposed feature extraction methods are compared, using support vector machines and boosted trees as classifiers. We find that both methods are suitable for real-time applications, and that CNN features outperform the hand-crafted features, both with regard to speed and accuracy. The best system (VGG-19 features, classified with a radial basis function support vector machine) obtained an accuracy of 98.4% for both species, processing an image in 0.08 seconds.
Chinese Translation
精准农业需要实时估计植物生长阶段。当植物生长阶段已知时，可以减少在栽培过程中资源的浪费，例如营养物质和水分，因为只需提供所需的资源。然而，不同生长阶段的植物具有相似的形态特征，这使得自动化生长阶段估计变得困难。本文提出了两种用于生长阶段估计的特征提取方法：一种使用Gabor滤波器组和形态学操作，另一种使用预训练的卷积神经网络（CNN）和迁移学习。我们在一个公开可用的植物生长阶段数据集（``bccr-segset``）上测试了这些方法，该数据集包含在室内条件下生长和捕获的两种植物：油菜和萝卜。我们使用支持向量机和提升树作为分类器，对这两种提出的特征提取方法进行了比较。我们发现这两种方法都适合实时应用，并且CNN特征在速度和准确性方面均优于手工设计的特征。最佳系统（VGG-19特征，使用径向基函数支持向量机分类）在两种植物上均获得了98.4%的准确率，处理一张图像的时间为0.08秒。
cs.CV / 64 / 2606.11969
SpecLoR: Spectral Lookahead Rectification for Motion-Coherent Text-to-Video Generation
SpecLoR：用于运动一致性文本到视频生成的谱前瞻校正
Abstract
Flow Matching has enabled robust text-to-video generation via latent ODE sampling. However, velocity approximation and numerical discretization errors inevitably accumulate, causing sampling trajectories to drift. Consequently, generated videos often suffer from severe spatiotemporal inconsistencies. Nevertheless, directly correcting these drifted, noisy latents is challenging: (i) timestep-dependent noise obscures reliable structural cues; (ii) spatial interventions risk disrupting intricate local geometry while incurring heavy computational costs. To address this, we propose Spectral Lookahead Rectification (SpecLoR), a plug-and-play inference method that bypasses noise via lookahead prediction, and circumvents spatiotemporal entanglement by shifting corrections to the frequency domain, where universal statistical priors of natural videos are readily available. First, during early sampling stages, SpecLoR looks ahead to estimate the clean latent $z_{t,0}$ and computes its 3D spatiotemporal spectrum. Next, SpecLoR rectifies the amplitude spectrum to match the prior, leaving the phase intact. Finally, the corrected state is re-noised to resume ODE integration. Experiments on Wan2.2 demonstrate that SpecLoR significantly reduces physical artifacts and enhances motion coherence across multiple benchmarks with minimal computational overhead (4 additional NFEs).
Chinese Translation
流匹配（Flow Matching）通过潜在常微分方程（ODE）采样实现了稳健的文本到视频生成。然而，速度近似和数值离散化误差不可避免地积累，导致采样轨迹漂移。因此，生成的视频往往存在严重的时空不一致性。然而，直接校正这些漂移的、嘈杂的潜变量是具有挑战性的：（i）时间步依赖的噪声掩盖了可靠的结构线索；（ii）空间干预可能会破坏复杂的局部几何结构，同时带来高昂的计算成本。为了解决这个问题，我们提出了谱前瞻校正（Spectral Lookahead Rectification，SpecLoR），这是一种即插即用的推理方法，通过前瞻预测绕过噪声，并通过将校正转移到频域来规避时空纠缠，在频域中，自然视频的普遍统计先验 readily 可用。首先，在早期采样阶段，SpecLoR 前瞻估计干净的潜变量 $z_{t,0}$ 并计算其三维时空谱。接下来，SpecLoR 校正幅度谱以匹配先验，同时保持相位不变。最后，校正后的状态重新加噪以恢复 ODE 积分。在 Wan2.2 数据集上的实验表明，SpecLoR 显著减少了物理伪影，并在多个基准测试中增强了运动一致性，且计算开销最小（额外增加 4 次 NFE）。
cs.CV / 65 / 2606.11977
ParseFixer: An Agentic Framework for Document Parsing via Selective Multimodal Correction
ParseFixer：一种通过选择性多模态修正的文档解析代理框架
Abstract
In this report, we present our third-place solution for the DataMFM Challenge Track 1: Document Parsing. This track requires models to recover structured Markdown documents from document page images while preserving textual content and document structure. To address the complementary requirements of accurate content recovery and faithful structure reconstruction, we propose ParseFixer, an agentic framework for backbone parsing and selective correction. ParseFixer consists of two key modules: Full-Page Backbone Parsing (FBP) and Agentic Selective Correction (ASC). FBP produces stable initial Markdown outputs with MinerU2.5 Pro, while ASC detects high-value parsing failures and repairs them through a verify-and-rollback correction process. By placing selective multimodal correction after open-source backbone parsing, ParseFixer improves the recovery of key document elements without rewriting reliable backbone predictions. On the test set, our final system achieves an overall score of 61.78 and ranks third in Track 1, demonstrating its effectiveness for accurate document parsing. Our code will be released at: https://github.com/iLearn-Lab/CVPRW26-ParseFixer.
Chinese Translation
在本报告中，我们展示了在DataMFM挑战赛第一赛道：文档解析中获得第三名的解决方案。该赛道要求模型从文档页面图像中恢复结构化的Markdown文档，同时保持文本内容和文档结构的完整性。为了满足准确内容恢复和忠实结构重建的互补要求，我们提出了ParseFixer，一种用于主干解析和选择性修正的代理框架。ParseFixer由两个关键模块组成：全页面主干解析（Full-Page Backbone Parsing, FBP）和代理选择性修正（Agentic Selective Correction, ASC）。FBP使用MinerU2.5 Pro生成稳定的初始Markdown输出，而ASC则检测高价值的解析失败并通过验证与回滚修正过程进行修复。通过在开源主干解析之后进行选择性多模态修正，ParseFixer在不重写可靠主干预测的情况下，提高了关键文档元素的恢复率。在测试集上，我们的最终系统获得了61.78的总体得分，并在第一赛道中排名第三，证明了其在准确文档解析方面的有效性。我们的代码将发布在：https://github.com/iLearn-Lab/CVPRW26-ParseFixer。
cs.CV / 66 / 2606.11989
From Nominal Intensity to Equivalent Rainfall: A Path-Based Credibility Evaluation Framework for Simulated Rainfall in Autonomous-Driving Perception Tests
从名义强度到等效降雨：一种基于路径的自主驾驶感知测试中模拟降雨的可信度评估框架
Abstract
Credible simulated-rainfall conditions are essential for identifying perception-system boundaries and supporting SOTIF-oriented risk assessment in automated driving. However, closed-field tests are often described only by nominal rainfall intensity or single-point measurements, making it difficult to align simulated rain fields with real rainfall and map test results to real-world scenarios. This paper proposes a path-based credibility evaluation method for simulated rainfall in autonomous-driving perception tests. Using the drop size and velocity joint distribution of real rainfall as the reference, each candidate path is represented by path-equivalent rainfall intensity, an uncertainty band, and a path-averaged Realism of Raindrop Distribution (RRD) score. Lidar target point-cloud count and mean reflectivity are further used for perception-consistency correction, quantifying the proxy capability of each simulated-rainfall path for real-rainfall perception effects. Experiments are conducted using about 10,000 real-rainfall raindrop-spectrum samples, 728 RainSense perception samples, and 45 spatial sampling points in a 2.4 m x 7.2 m simulated-rainfall area. Results show that spatial non-uniformity remains under the same nominal condition, confirming the need for path-based evaluation. The method identifies Path IV and Path VI as preferable candidates, with results of 11.54 +/- 0.31 mm/h, RRD = 0.43, and 8.28 +/- 0.34 mm/h, RRD = 0.46, respectively. These paths show more balanced performance in rainfall-intensity stability, raindrop-spectrum realism, and perception consistency. The proposed method supports path selection, condition description, and credible interpretation of autonomous-driving perception tests under rainfall.
Chinese Translation
可信的模拟降雨条件对于识别感知系统边界和支持面向安全的自动驾驶风险评估至关重要。然而，封闭场地测试通常仅通过名义降雨强度或单点测量来描述，这使得将模拟降雨场与真实降雨对齐并将测试结果映射到现实场景中变得困难。本文提出了一种基于路径的模拟降雨可信度评估方法，适用于自主驾驶感知测试。以真实降雨的雨滴大小和速度联合分布为参考，每个候选路径由路径等效降雨强度、不确定性带以及路径平均雨滴分布现实性（Realism of Raindrop Distribution, RRD）评分表示。进一步使用激光雷达目标点云计数和平均反射率进行感知一致性校正，量化每条模拟降雨路径对真实降雨感知效果的代理能力。实验使用约10,000个真实降雨雨滴光谱样本、728个RainSense感知样本和在2.4米 x 7.2米模拟降雨区域内的45个空间采样点进行。结果表明，在相同名义条件下，空间非均匀性依然存在，确认了基于路径评估的必要性。该方法识别出路径IV和路径VI为优选候选，结果分别为11.54 +/- 0.31 mm/h，RRD = 0.43，以及8.28 +/- 0.34 mm/h，RRD = 0.46。这些路径在降雨强度稳定性、雨滴光谱现实性和感知一致性方面表现出更平衡的性能。所提出的方法支持路径选择、条件描述以及在降雨条件下对自主驾驶感知测试的可信解读。
cs.CV / 67 / 2606.12012
FitVTON: Fit-aware Virtual Try-On via Body-Garment Size Control
FitVTON：基于适配的虚拟试衣通过身体-服装尺寸控制
Abstract
While diffusion-based virtual try-on has achieved impressive visual realism, most methods treat the task as 2D inpainting, prioritizing texture preservation over physical plausibility. Consequently, they often produce plausible-looking images that fail to reflect authentic garment fit across diverse body shapes. We present FitVTON, a Fit-aware virtual try-on model on different bodies in the wild. FitVTON encodes garment-body size through structured text prompts, and learn from simulated try-on triplets from parameterized garment model. To improve the fitting effects over garment silhouettes, we introduce two auxiliary head to predict the masks for both the garment and the exposed body. We further introduce a texture rectification stage to improve realistic appearance from simulated data. To evaluate the fitting fidelity, we curate a real-world dataset, FittingEffect3K, combining VLM-based scoring protocol. Both subjective and quantitive experiments show that FitVTON demonstrate authentic fitting fidelity, with significant sizing accuracy and shape preservation over state-of-the-art methods while maintaining competitive image quality. Project Page: https://zenoning.github.io/FitVTON/.
Chinese Translation
尽管基于扩散的虚拟试衣技术已实现令人印象深刻的视觉真实感，但大多数方法将该任务视为二维修补，优先考虑纹理保留而非物理合理性。因此，它们往往生成看似合理的图像，却无法反映不同体型的真实服装合身效果。我们提出了FitVTON，一种适配意识的虚拟试衣模型，能够在各种真实体型上进行试穿。FitVTON通过结构化文本提示对服装-身体尺寸进行编码，并从参数化服装模型的模拟试穿三元组中学习。为了改善服装轮廓的合身效果，我们引入了两个辅助头来预测服装和暴露身体的掩码。我们进一步引入了纹理修正阶段，以提高从模拟数据中获得的真实外观。为了评估合身的保真度，我们整理了一个真实世界数据集FittingEffect3K，并结合了基于VLM的评分协议。主观和定量实验均表明，FitVTON在合身保真度方面表现出真实的合身效果，在尺寸准确性和形状保留方面超越了最先进的方法，同时保持了竞争力的图像质量。项目页面：https://zenoning.github.io/FitVTON/
cs.CV / 68 / 2606.12023
ViT-FREE: Efficient Face Recognition via Early Exiting and Synthetic Adaptation
ViT-FREE：通过早期退出和合成适应实现高效的人脸识别
Abstract
Vision Transformers (ViTs) have gained significant attention in computer vision and shown strong potential for face recognition (FR). However, their high computational cost makes deployment on resource-constrained devices challenging, motivating the need for methods that balance efficiency and accuracy. In this work, we investigate early exiting in pretrained ViTs as a simple yet effective training-free strategy for efficient FR inference. Leveraging the uniform feature dimensionality across transformer encoder blocks, we introduce ViT-FREE, a multi-exit framework that enables face verification directly from intermediate representations without modifying or retraining the backbone model, and thus, reducing inference cost. Empirically, we show that patch embeddings and attention maps evolve progressively across depth, exhibiting high similarity between consecutive ViT blocks and increasing alignment with the final representation. This indicates gradual feature refinement and attention convergence, suggesting that intermediate layers already provide stable and discriminative representations suitable for early exiting. Through extensive experiments on multiple FR benchmarks, we systematically analyze the accuracy-efficiency trade-off across exit depths. Our results demonstrate that later exits achieve a highly favorable balance, with exiting at layer 10 yielding up to a 20% speedup while incurring only a 1.5 drop in verification performance on benchmarks such as IJB-C. Also, we propose ViT-FREE_FT, a lightweight exit-specific fine-tuning strategy that adapts only the projection layers using a small synthetic dataset while keeping the transformer backbone frozen. This approach improves the performance of shallow exits while preserving the efficiency benefits and leaving deeper exits largely unaffected.
Chinese Translation
视觉变换器（ViTs）在计算机视觉领域获得了显著关注，并显示出在人脸识别（FR）方面的强大潜力。然而，它们的高计算成本使得在资源受限设备上的部署面临挑战，这促使我们需要寻找在效率和准确性之间取得平衡的方法。在本研究中，我们探讨了在预训练的ViTs中进行早期退出，作为一种简单而有效的无训练策略，以实现高效的人脸识别推断。利用变换器编码器块之间统一的特征维度，我们引入了ViT-FREE，一个多退出框架，能够直接从中间表示中进行人脸验证，而无需修改或重新训练主干模型，从而降低推断成本。实证研究表明，补丁嵌入和注意力图在深度上逐渐演变，连续的ViT块之间表现出高度相似性，并与最终表示的对齐度逐渐增加。这表明特征逐步细化和注意力收敛，暗示中间层已经提供了适合早期退出的稳定且具有区分性的表示。通过在多个FR基准上的广泛实验，我们系统地分析了不同退出深度下的准确性与效率的权衡。我们的结果表明，较晚的退出实现了高度有利的平衡，在第10层退出时，速度提高了多达20%，而在IJB-C等基准上的验证性能仅下降了1.5。此外，我们提出了ViT-FREE_FT，一种轻量级的退出特定微调策略，仅使用小型合成数据集调整投影层，同时保持变换器主干不变。这种方法在保持效率优势的同时，提高了浅层退出的性能，并对深层退出几乎没有影响。
cs.CV / 69 / 2606.12033
SpikeTAD: Spiking Neural Networks for End-to-End Temporal Action Detection
SpikeTAD：用于端到端时间动作检测的脉冲神经网络
Abstract
Video understanding is a crucial part of computer vision, with numerous application scenarios. With the increasing popularity of mobile devices, an increasing number of efforts are trying to deploy video understanding models on them. However, existing video understanding models are difficult to deploy due to their large size and prohibitive power consumption. Spiking Neural Networks (SNNs) have shown bioplausibility and low power advantages over Artificial Neural Networks (ANNs), especially on neuromorphic chips which are regarded as essential components of future mobile devices. However, excessively long conversion time-steps and severe performance degradation problems limit their application. To solve the problems above, we explore the application of SNNs on temporal action detection (TAD), which is an important task in video understanding, and propose the first SNN-based end-to-end TAD architecture coined as SpikeTAD. While maintaining extremely low power consumption, SpikeTAD achieves an average mAP of 67.2% in THUMOS14 and 37.42% in ActivityNet-1.3, demonstrating the feasibility of a low-power TAD model. Our code is available at https://github.com/MCG-NJU/SpikeTAD.
Chinese Translation
视频理解是计算机视觉的一个关键部分，具有众多应用场景。随着移动设备的日益普及，越来越多的努力试图将视频理解模型部署到这些设备上。然而，现有的视频理解模型由于其庞大的体积和高昂的功耗，难以进行部署。脉冲神经网络（SNNs）在生物合理性和低功耗方面相较于人工神经网络（ANNs）表现出优势，尤其是在被视为未来移动设备核心组件的神经形态芯片上。然而，过长的转换时间步和严重的性能下降问题限制了它们的应用。为了解决上述问题，我们探索了SNNs在时间动作检测（TAD）中的应用，这是视频理解中的一项重要任务，并提出了第一个基于SNN的端到端TAD架构，命名为SpikeTAD。在保持极低功耗的同时，SpikeTAD在THUMOS14数据集上实现了67.2%的平均mAP，在ActivityNet-1.3数据集上实现了37.42%的平均mAP，证明了低功耗TAD模型的可行性。我们的代码可在 https://github.com/MCG-NJU/SpikeTAD 获取。
cs.CV / 70 / 2606.12036
Vision Transformers for Face Recognition Need More Registers
面部识别中的视觉变换器需要更多的寄存器
Abstract
Recent advances in Vision Transformers (ViTs) for face recognition (FR) have moved beyond the standard CLS-token paradigm. In this paradigm, a special classification token (CLS) is prepended to the patch embeddings and used as a representation of the input for downstream tasks. An alternative approach, Concatenated Patch Embeddings (CPE), instead leverages all patch tokens by concatenating them into a single vector, which is then projected into a compact face representation. CPE has been shown to improve recognition performance in comparison to CLS-based ones, but our qualitative analysis of attention maps showed the presence of artifacts that limit their interpretability. To address this issue, we incorporate register tokens, learnable tokens concatenated to the initial patch embeddings, and processed jointly through the ViT encoder blocks. This mechanism has been shown to produce more structured and interpretable attention maps compared to baseline ViT. We empirically demonstrate that these artifacts consistently appear across various ViT backbones, including small and large models, and that introducing register tokens effectively mitigates them. Adding four or eight registers significantly enhances interpretability, with eight registers providing the highest verification accuracies and smoothest attention structures. Our resulting model, ViT-8R, corresponds to a CPE-based ViT-B architecture augmented with eight register tokens achieves state-of-the-art performance among ViT-based FR models on large-scale IJB-B and IJB-C benchmarks. Also, ViT-8R produces substantially clearer attention maps compared with the baseline model, which offer deeper insight into the model's attention behavior (https://github.com/TaharChettaoui/ViT-FR-Registers)
Chinese Translation
最近在面部识别（FR）领域的视觉变换器（ViTs）方面的进展已超越了标准的CLS-token范式。在这一范式中，一个特殊的分类token（CLS）被添加到补丁嵌入的前面，并用作下游任务的输入表示。另一种方法，连接补丁嵌入（CPE），则通过将所有补丁token连接成一个单一向量来利用它们，然后将其投影到一个紧凑的面部表示中。与基于CLS的方法相比，CPE已被证明能提高识别性能，但我们对注意力图的定性分析显示出存在限制其可解释性的伪影。为了解决这个问题，我们引入了寄存器token，这是一种可学习的token，连接到初始补丁嵌入，并通过ViT编码器块共同处理。与基线ViT相比，这一机制已被证明能够生成更结构化和可解释的注意力图。我们通过实验证明，这些伪影在各种ViT骨干网络中持续出现，包括小型和大型模型，并且引入寄存器token有效地减轻了这些伪影。添加四个或八个寄存器显著增强了可解释性，其中八个寄存器提供了最高的验证准确率和最平滑的注意力结构。我们得到的模型ViT-8R对应于基于CPE的ViT-B架构，增加了八个寄存器token，在大规模IJB-B和IJB-C基准测试中实现了ViT基础的FR模型中的最先进性能。此外，与基线模型相比，ViT-8R生成的注意力图明显更清晰，为模型的注意力行为提供了更深入的见解（https://github.com/TaharChettaoui/ViT-FR-Registers）
cs.CV / 71 / 2606.12047
Metadata-Aware Multi-Prompt Reasoning for Zero-Shot Accident Understanding
元数据感知的多提示推理用于零样本事故理解
Abstract
In this paper, we address the problem of zero-shot understanding of accidents from surveillance videos by identifying when an impact event occurs, what type of impact it is, and where in the frame it occurs using natural language. We propose a three-stage pipeline that decomposes the accident understanding into when, what, and where. The first stage extracts a short temporal window around the impact using vision-language similarity. In the second stage, we perform metadata-driven multi-prompt reasoning with five complementary views (baseline, motion, geometry, contrast, and tiebreaker) and resolve disagreement via an entropy-gated pairwise adjudicator. Finally, we localize the impact of an open-vocabulary detector queried on the predicted accident type and scene layout, and aggregate detections across keyframes using a score-weighted centroid. Our pipeline achieves a substantial improvement in the harmonic-mean score over a centre-of-frame baseline on the zero-shot ACCIDENT @ CVPR benchmark. We show that decomposing zero-shot video understanding into temporal localization, semantic classification, and spatial grounding enable more reliable reasoning with vision-language models than direct prompting alone.
Chinese Translation
在本文中，我们通过识别何时发生冲击事件、冲击的类型以及事件在画面中的位置，解决了从监控视频中零样本理解事故的问题。我们提出了一个三阶段的流程，将事故理解分解为何时、什么和哪里。第一阶段使用视觉-语言相似性提取冲击周围的短时间窗口。在第二阶段，我们进行基于元数据的多提示推理，结合五个互补视角（基线、运动、几何、对比和决胜者），并通过熵门控的成对裁决者解决分歧。最后，我们利用开放词汇检测器定位冲击，查询预测的事故类型和场景布局，并通过得分加权质心在关键帧中聚合检测结果。我们的流程在零样本ACCIDENT @ CVPR基准测试中，相较于中心帧基线，在调和平均分上实现了显著提升。我们展示了将零样本视频理解分解为时间定位、语义分类和空间定位，使得与视觉-语言模型的推理更加可靠，而不仅仅依赖直接提示。
cs.CV / 72 / 2606.12051
MFEN:Multi-Frequency Expert Network for Visible-Infrared Person Re-ID
MFEN：用于可见光-红外人脸重识别的多频专家网络
Abstract
Visible-infrared person re-identification (VI-ReID) is challenging due to the large modality discrepancy between visible and infrared images. We contend that this discrepancy is largely related to differing lighting conditions, including differences in light wavelength and light source type. Recently, frequency-based VI-ReID approaches have achieved notable success because frequency information can better extract identity-relevant contours and details while excluding irrelevant lighting and color. However, existing methods either do not distinguish different frequency bands or focus on only one band, which is insufficient under diverse lighting conditions. To perform comprehensive frequency domain learning, we propose a Multi-Frequency Expert Network (MFEN) that enables multi-frequency modulation and adaptively combines different bands through a mixture-of-experts design. We further introduce Random Frequency Augmentation (RFA) and Frequency Auxiliary Optimization (FAO) to better train MFEN. The three modules are complementary and jointly capture critical frequency-domain details for robust representation learning. Extensive experiments on three VI-ReID datasets demonstrate the effectiveness of our approach.
Chinese Translation
可见光-红外人脸重识别（VI-ReID）因可见光图像与红外图像之间的显著模态差异而面临挑战。我们认为，这种差异主要与不同的光照条件有关，包括光波长和光源类型的差异。最近，基于频率的VI-ReID方法取得了显著成功，因为频率信息能够更好地提取与身份相关的轮廓和细节，同时排除无关的光照和颜色。然而，现有方法要么不区分不同的频率带，要么仅关注一个频率带，这在多样的光照条件下是不够的。为了进行全面的频域学习，我们提出了一种多频专家网络（MFEN），该网络能够实现多频调制，并通过专家混合设计自适应地结合不同频带。我们进一步引入了随机频率增强（Random Frequency Augmentation, RFA）和频率辅助优化（Frequency Auxiliary Optimization, FAO）来更好地训练MFEN。这三个模块是互补的，共同捕捉关键的频域细节，以实现稳健的表示学习。在三个VI-ReID数据集上的大量实验表明了我们方法的有效性。
cs.CV / 73 / 2606.12066
Performance Analysis of YOLOv11 and YOLOv8 for Mixed Traffic Object Detection under Adverse Weather Conditions in Developing Countries
在发展中国家恶劣天气条件下，YOLOv11与YOLOv8在混合交通物体检测中的性能分析
Abstract
In modern vehicular systems, robust performance under harsh conditions has become a critical problem of autonomous driving. Our study delivers a comprehensive evaluation of the newest iteration of the YOLO series, which is YOLOv11 Nano architecture benchmarked against the widely adopted YOLOv8 Nano as a baseline on a custom fused dataset that combines the Indian Driving Dataset (IDD) [1] and Berkeley Deep Drive Dataset (BDD100K) [2]. We have analyzed the trade-offs among detection accuracy, inference speed, and computational efficiency in high-entropy scenarios involving dense mixed traffic, rain, and low-light conditions. Specifically, YOLOv11n achieves a mean Average Precision (mAP@50) of 46.6%, with a notable 3.2% improvement in Precision over the baseline, effectively reducing false positives in cluttered scenes. Furthermore, the proposed model exhibits enhanced energy efficiency, requiring 22% fewer FLOPs (6.3G vs. 8.1G) while maintaining real-time inference speed of 70.9 FPS on a Tesla T4 GPU, offering an optimal trade-off for safety-critical edge deployment.
Chinese Translation
在现代车辆系统中，恶劣条件下的稳健性能已成为自动驾驶的一个关键问题。本研究对YOLO系列的最新版本，即YOLOv11 Nano架构进行了全面评估，并以广泛采用的YOLOv8 Nano作为基准，基于一个自定义融合数据集，该数据集结合了印度驾驶数据集（Indian Driving Dataset, IDD）和伯克利深度驾驶数据集（Berkeley Deep Drive Dataset, BDD100K）。我们分析了在高熵场景下（包括密集混合交通、降雨和低光照条件）检测准确性、推理速度和计算效率之间的权衡。具体而言，YOLOv11n在mAP@50上达到了46.6%的平均精度，相较于基准提高了3.2%的精度，有效减少了混乱场景中的误报。此外，所提出的模型展现了更高的能效，所需的FLOPs减少了22%（6.3G对比8.1G），同时在Tesla T4 GPU上保持了70.9 FPS的实时推理速度，为安全关键的边缘部署提供了最佳的权衡。
cs.CV / 74 / 2606.12069
Tac-DINO: Learning Vision-Tactile Features with Patch Alignment
Tac-DINO：通过补丁对齐学习视觉-触觉特征
Abstract
Touch is the primary medium through which humans interact with the environment. Currently, tactile learning mainly focuses on image-level pretraining or alignment. However, tactile signals correspond to local object contact, while research into scale alignment and holographic matching remains limited and proper datasets and benchmarks also lack. To bridge this gap, we first construct a data collection system to acquire a large-scale tactile dataset, with over 20 K tactile contacts from 505 real-world objects. Building on this dataset, we design a Vis-Tac Holographic Matching Benchmark to evaluate vision-tactile local-to-global alignment ability. Then we propose Vision-Tactile Patch Alignment (VTPA) methods for vision-tactile representation learning. Experiments demonstrate that these exceed the performance of methods without alignment and align with whole-object images.
Chinese Translation
触觉是人类与环境互动的主要媒介。目前，触觉学习主要集中在图像级的预训练或对齐。然而，触觉信号对应于局部物体接触，而对尺度对齐和全息匹配的研究仍然有限，适当的数据集和基准也缺乏。为了填补这一空白，我们首先构建了一个数据采集系统，以获取一个大规模的触觉数据集，该数据集包含来自505个真实物体的超过20,000个触觉接触点。在此数据集的基础上，我们设计了一个视觉-触觉全息匹配基准，以评估视觉-触觉的局部到全局对齐能力。然后，我们提出了视觉-触觉补丁对齐（Vision-Tactile Patch Alignment, VTPA）方法用于视觉-触觉表示学习。实验表明，这些方法的性能超过了没有对齐的方法，并且与整体物体图像对齐。
cs.CV / 75 / 2606.12072
World Model Self-Distillation: Training World Models to Solve General Tasks
世界模型自蒸馏：训练世界模型以解决通用任务
Abstract
Pretrained video generators are promising visual world models that exhibit emergent task-solving abilities; however, their reliance on detailed textual descriptions limits their direct use for planning and decision-making. Existing approaches either outsource this reasoning to language or vision-language models, or rely on supervised fine-tuning with paired task-execution videos, which are costly to collect and difficult to scale. We propose a scalable framework that elicits task-solving ability in such models by combining self-distillation with reinforcement learning. Given an unlabeled scene image, a vision-language model generates a candidate task and a detailed step-by-step solution. The solution conditions a pretrained video diffusion model, the Demonstrator; we distill its behavior into an Executor conditioned only on the image and a short task prompt. This transfers execution knowledge from caption-guided generation to instruction-conditioned task solving without curated task-video supervision. We further improve the Executor with reinforcement learning from VLM feedback, exploiting the asymmetry between judging whether a sampled video satisfies a task and generating the solution. Experiments on our proposed WorldTasks-Benchmark and the DreamGen robotics benchmark show that the Executor surpasses the Demonstrator under our VLM-based evaluation protocol and transfers competitively to robotic tasks.
Chinese Translation
预训练的视频生成器是有前景的视觉世界模型，展现出新兴的任务解决能力；然而，它们对详细文本描述的依赖限制了它们在规划和决策中的直接应用。现有的方法要么将这种推理外包给语言或视觉-语言模型，要么依赖于带有配对任务执行视频的监督微调，这种方法收集成本高且难以扩展。我们提出了一种可扩展的框架，通过将自蒸馏与强化学习相结合，激发此类模型的任务解决能力。给定一个未标记的场景图像，视觉-语言模型生成一个候选任务和详细的逐步解决方案。该解决方案为预训练的视频扩散模型——演示者（Demonstrator）提供条件；我们将其行为蒸馏到仅基于图像和简短任务提示的执行者（Executor）中。这将从基于标题的生成转移到基于指令的任务解决，而无需经过精心策划的任务视频监督。我们进一步通过来自视觉-语言模型（VLM）反馈的强化学习来改进执行者，利用判断采样视频是否满足任务与生成解决方案之间的不对称性。在我们提出的WorldTasks-Benchmark和DreamGen机器人基准上的实验表明，执行者在我们的基于VLM的评估协议下超越了演示者，并在机器人任务中具有竞争力的迁移能力。
cs.CV / 76 / 2606.12074
Non-frontal face recognition using GANs and memristor-based classifiers
基于GAN和忆阻器分类器的非正面人脸识别
Abstract
Face recognition systems have advanced significantly through deep learning techniques, delivering high performance and robustness in complex scenarios. However, these approaches incur substantial computational overhead, limiting their in situ applicability in resource-constrained platforms such as drones, where they can address challenges including non-frontal facial imagery. Memristor-based neuromorphic systems have emerged as a compelling approach for edge AI applications, combining biologically inspired processing with efficient and scalable computation. In this work, we propose a facial recognition framework that addresses non-frontal pose variations by integrating lightweight generative adversarial network (GAN)-based pose frontalisation with memristor-based neuromorphic recognition. The experimental results on two datasets demonstrate the effectiveness of combining adversarial learning with memristive technology, achieving up to 96% identification accuracy. The proposed approach alleviates the computational bottlenecks of conventional AI and offers a scalable, efficient solution for face recognition in dynamic real-world environments.
Chinese Translation
人脸识别系统通过深度学习技术取得了显著进展，在复杂场景中展现出高性能和鲁棒性。然而，这些方法带来了巨大的计算开销，限制了它们在资源受限的平台（如无人机）中的实际应用，而这些平台可以解决包括非正面人脸图像在内的挑战。基于忆阻器的神经形态系统作为边缘人工智能应用的一种有力方法应运而生，结合了生物启发的处理方式与高效、可扩展的计算能力。在本研究中，我们提出了一种人脸识别框架，通过将轻量级生成对抗网络（GAN）基础的姿态正面化与基于忆阻器的神经形态识别相结合，解决非正面姿态变化的问题。在两个数据集上的实验结果表明，将对抗学习与忆阻器技术相结合的有效性，识别准确率高达96%。所提出的方法缓解了传统人工智能的计算瓶颈，为动态现实环境中的人脸识别提供了一种可扩展、高效的解决方案。
cs.CV / 77 / 2606.12099
ISAP-3D: Identity-Slot Aligned Part-Aware 3D Generation
ISAP-3D：身份-槽对齐的部件感知3D生成
Abstract
Part-aware 3D generation aims to synthesize structured objects with semantically meaningful components, yet often suffers from structural ambiguity due to identity-layout entanglement. Existing methods either infer part identity and spatial layout implicitly, which can lead to unstable part allocation (e.g., slot swapping or part merging), or rely on strong layout conditions that are difficult to obtain in practice. We attribute this ambiguity to identity-slot permutation freedom: without explicit identity-slot alignment, the correspondence between semantic parts and generation slots is not identifiable during training, allowing multiple slot assignments to fit the same supervision and leading to inconsistent decomposition. Based on this insight, we argue that stable part-aware generation requires identity-aligned one-to-one slot modelling. We therefore propose an identity-slot aligned framework, ISAP-3D, which anchors each part with semantic identity tokens and performs identity-conditioned one-to-one layout prediction, followed by layout-conditioned geometry synthesis. Structured local-global conditioning maintains identity alignment across semantic, spatial, and geometric stages. We also construct a part-level dataset with a unified semantic protocol to enable learnable and consistent identity-slot alignment. Extensive experiments demonstrate improved structural stability, controllability, and robustness over state-of-the-art part-aware generation baselines.
Chinese Translation
部件感知的3D生成旨在合成具有语义意义的结构化对象，但由于身份-布局纠缠，常常面临结构模糊性。现有方法要么隐式推断部件身份和空间布局，这可能导致不稳定的部件分配（例如，槽交换或部件合并），要么依赖于在实践中难以获得的强布局条件。我们将这种模糊性归因于身份-槽置换自由：在没有明确的身份-槽对齐的情况下，语义部件与生成槽之间的对应关系在训练期间是不可识别的，这允许多个槽分配适应相同的监督，从而导致不一致的分解。基于这一见解，我们认为稳定的部件感知生成需要身份对齐的一对一槽建模。因此，我们提出了一个身份-槽对齐框架ISAP-3D，该框架通过语义身份标记锚定每个部件，并执行身份条件的一对一布局预测，随后进行布局条件的几何合成。结构化的局部-全局条件在语义、空间和几何阶段保持身份对齐。我们还构建了一个具有统一语义协议的部件级数据集，以实现可学习和一致的身份-槽对齐。大量实验表明，与最先进的部件感知生成基线相比，结构稳定性、可控性和鲁棒性得到了显著改善。
cs.CV / 78 / 2606.12106
MSUE: Multi-Modal Soccer Understanding Expert
MSUE：多模态足球理解专家
Abstract
This paper presents our solution to the 2026 SoccerNet VQA Challenge. We first develop a cost-effective data synthesis pipeline driven by a Vision-Language Model (VLM), which systematically restructures raw domain data into diverse VQA samples, including concise answers and long-form responses. Second, we propose MSUE, a multi-expert question answering architecture that employs a Large Language Model (LLM) to dynamically dispatch questions to text, image, and video experts. These experts are instantiated as a strong text baseline Gemini3-Flash, a fine-tuned Qwen3-VL, and an external knowledge base, respectively, working collaboratively to enhance VQA performance. MSUE achieves an accuracy of \textbf{0.95} on the challenge benchmark, securing third place in the leaderboard.
Chinese Translation
本文提出了我们对2026 SoccerNet VQA挑战赛的解决方案。我们首先开发了一种以视觉语言模型（Vision-Language Model, VLM）驱动的高性价比数据合成管道，该管道系统性地将原始领域数据重构为多样化的视觉问答（VQA）样本，包括简洁答案和长篇回答。其次，我们提出了MSUE，一种多专家问答架构，该架构利用大型语言模型（Large Language Model, LLM）动态分配问题给文本、图像和视频专家。这些专家分别被实例化为强大的文本基线模型Gemini3-Flash、经过微调的Qwen3-VL和外部知识库，协同工作以提升VQA性能。MSUE在挑战基准上达到了 extbf{0.95}的准确率，获得排行榜第三名。
cs.CV / 79 / 2606.12125
Q-Fold: Query-Aware Focus-Context Spatio-Temporal Folding for Long Video Understanding
Q-Fold：基于查询的聚焦-上下文时空折叠用于长视频理解
Abstract
Long-video understanding remains challenging for multimodal large language models, because temporally extended videos often contain thousands of frames and are therefore expensive to process exhaustively. Existing methods usually construct compact visual inputs from long videos under a limited visual budget. However, most of them still follow a frame-centric paradigm and apply similar representations to retained content regardless of its importance. This makes it difficult to preserve both high-fidelity visual evidence and broad temporal coverage. To address this issue, we propose Q-Fold, a training-free input construction framework for long-video understanding. Instead of treating isolated frames as the basic modeling unit, Q-Fold operates on contiguous temporal segments and constructs a heterogeneous Focus--Context representation under query guidance. Query-relevant segments are preserved as high-fidelity Focus Frames, while less relevant segments are folded into chronology-preserving contextual layouts. In this way, Q-Fold preserves critical visual evidence and broad temporal coverage, while better maintaining local temporal continuity within short segments. Experiments on four long-video benchmarks with multiple Video-MLLMs show that Q-Fold consistently improves performance without increasing the input budget. Notably, it achieves gains of up to 9.1 percentage points on an ultra-long video benchmark. Code will be made publicly available.
Chinese Translation
长视频理解对于多模态大型语言模型仍然具有挑战性，因为时间延续的视频通常包含成千上万的帧，因此全面处理成本高昂。现有方法通常在有限的视觉预算下从长视频中构建紧凑的视觉输入。然而，它们大多数仍然遵循以帧为中心的范式，并对保留的内容应用相似的表示，而不考虑其重要性。这使得同时保留高保真视觉证据和广泛时间覆盖变得困难。为了解决这个问题，我们提出了Q-Fold，一种用于长视频理解的无训练输入构建框架。Q-Fold并不将孤立的帧视为基本建模单元，而是对连续的时间段进行操作，并在查询指导下构建异质的聚焦-上下文表示。与查询相关的段被保留为高保真的聚焦帧，而相关性较低的段则折叠为保持时间顺序的上下文布局。通过这种方式，Q-Fold保留了关键的视觉证据和广泛的时间覆盖，同时更好地维持了短段内的局部时间连续性。在四个长视频基准测试和多个视频-MLLM的实验中，Q-Fold始终提高了性能，而没有增加输入预算。值得注意的是，它在一个超长视频基准测试中实现了高达9.1个百分点的增益。代码将公开发布。
cs.CV / 80 / 2606.12126
AGE-MIL: Anchor-Guided Evidence Learning for Patient-Level Prediction
AGE-MIL：基于锚点引导的证据学习用于患者级预测
Abstract
Existing computational pathology methods predominantly operate within whole-slide image (WSI)-level multiple instance learning (MIL) paradigms, while patient-level modeling remains underexplored. In routine pathological practice, however, pathologists derive diagnostic and prognostic conclusions by integrating evidence across multiple WSIs rather than relying on any single slide. This discrepancy creates a fundamental misalignment when patient-level supervision is directly imposed on conventional MIL frameworks, often leading to unstable optimization and degraded predictive reliability. To address this issue, we propose Anchor-Guided Evidence MIL (AGE-MIL), a weakly supervised framework for patient-level prediction. AGE-MIL constructs a patient-level anchor from slide representations to capture global pathological context and guide the retrieval and integration of diagnostically relevant local patches, enabling robust patient-level modeling. Patient-level risk is further modeled as an evidence accumulation process, promoting stable optimization under weak supervision. AGE-MIL is evaluated on six clinically relevant patient-level prediction tasks from two independent cohorts. Experimental results show that the proposed framework consistently outperforms eight state-of-the-art MIL methods. Code is available at https://github.com/wodeniua/AGE-MIL.
Chinese Translation
现有的计算病理学方法主要在全切片图像（WSI）级别的多实例学习（MIL）范式下运行，而患者级建模仍然未得到充分探索。然而，在常规病理实践中，病理学家通过整合多个WSI中的证据而非依赖于单一切片来得出诊断和预后结论。这种差异在将患者级监督直接施加于传统MIL框架时造成了根本的不匹配，常常导致不稳定的优化和降低的预测可靠性。为了解决这一问题，我们提出了基于锚点引导的证据MIL（AGE-MIL），这是一种用于患者级预测的弱监督框架。AGE-MIL从切片表示中构建患者级锚点，以捕捉全局病理上下文并引导诊断相关局部块的检索和整合，从而实现稳健的患者级建模。患者级风险进一步被建模为证据积累过程，促进了在弱监督下的稳定优化。AGE-MIL在两个独立队列的六个临床相关患者级预测任务上进行了评估。实验结果表明，所提出的框架在性能上始终优于八种最先进的MIL方法。代码可在 https://github.com/wodeniua/AGE-MIL 获取。
cs.CV / 81 / 2606.12140
Time-Conditioned and Multi-Time Survival Prediction from 2D PET/CT Projections in Lung Cancer
基于时间条件和多时间生存预测的肺癌2D PET/CT投影研究
Abstract
Accurate prediction of overall survival (OS) from positron emission tomography/computed tomography (PET/CT) can support personalized treatment and follow-up strategies in oncology. However, the impact of temporal modeling on imaging-based survival prediction remains insufficiently explored. We investigate how different temporal formulations influence survival prediction by developing two complementary approaches: Attention-guided Time-Conditioned Survival (ATCS) and Multi-Time Survival (MTS). We retrospectively analyzed pre-treatment PET/CT images from 848 patients with non-small cell lung cancer (NSCLC), including 556 for model development and 292 for held-out testing. A previously proposed Time-Conditioned Survival (TCS) model was used as a baseline. Models were trained using 5-fold cross-validation and evaluated on the test set using time-dependent area under the curve (AUC) at 6-month intervals from 0.5 to 5 years. Both ATCS and MTS outperformed the baseline TCS model, achieving mean AUCs of 0.794 and 0.793, respectively, compared to 0.767. ATCS performed better at earlier time points (0.5-3 years), whereas MTS performed better at later intervals (3.5-5 years). Combining tumor-specific and tissue-wise PET/CT features improved performance over either input alone. Finer temporal discretization improved short-term prediction, while coarser intervals provided more stable long-term estimates. These findings demonstrate that temporal modeling and input design influence PET/CT-based survival prediction. The proposed approaches enable time-specific survival estimation from pre-treatment imaging and may support improved risk stratification and clinical decision-making.
Chinese Translation
从正电子发射断层扫描/计算机断层扫描（PET/CT）中准确预测总体生存期（OS）可以支持肿瘤学中的个性化治疗和随访策略。然而，时间建模对基于影像的生存预测的影响仍然探讨不足。我们通过开发两种互补的方法：注意力引导的时间条件生存（Attention-guided Time-Conditioned Survival, ATCS）和多时间生存（Multi-Time Survival, MTS），研究不同时间表述对生存预测的影响。我们对848例非小细胞肺癌（NSCLC）患者的治疗前PET/CT图像进行了回顾性分析，其中556例用于模型开发，292例用于保留测试。我们使用之前提出的时间条件生存（Time-Conditioned Survival, TCS）模型作为基线。模型采用5折交叉验证进行训练，并在测试集上使用时间依赖的曲线下面积（AUC）进行评估，时间间隔为0.5至5年，每6个月一次。ATCS和MTS均优于基线TCS模型，分别实现了0.794和0.793的平均AUC，而TCS为0.767。ATCS在早期时间点（0.5-3年）表现更佳，而MTS在较晚的时间间隔（3.5-5年）表现更好。结合肿瘤特异性和组织特征的PET/CT特征相较于单独输入提高了性能。更细的时间离散化改善了短期预测，而较粗的时间间隔提供了更稳定的长期估计。这些发现表明，时间建模和输入设计影响基于PET/CT的生存预测。所提出的方法能够从治疗前影像中实现时间特定的生存估计，并可能支持改善风险分层和临床决策。
cs.CV / 82 / 2606.12153
TopoCap: Learning Topology-Agnostic Motion Priors for Monocular Video-to-Animation
TopoCap：学习拓扑无关的运动先验用于单目视频到动画
Abstract
The explosion of generative 3D assets has created a massive demand for animation, yet current motion capture methods remain brittle, restricted to species-specific templates (e.g., SMPL) or requiring labor-intensive manual rigging. We introduce TopoCap, the first unified framework capable of extracting motion from monocular video and retargeting it onto characters with arbitrary, unseen skeletal topologies, i.e., from bipeds to hexapods and inanimate objects, without test-time optimization. Our key insight is that while skeletal structures are combinatorial and discrete, the underlying physics of motion occupy a continuous, low-dimensional manifold. We materialize this insight via a two-stage generative pipeline. First, we learn a Universal Motion Manifold using a Graph CVAE that compresses heterogeneous kinematic chains into a shared, fixed-length latent code. By explicitly conditioning the decoder on a structural embedding of the target rig, we disentangle motion dynamics from skeletal topology. Second, we treat video-to-animation as a conditional flow matching problem, predicting these topology-agnostic codes from visual features. To learn this generalized prior, we introduce Mobjaverse, a massive-scale dataset curated from Objaverse-XL. Comprising over 5,000 unique skeletal topologies and 2 million frames, it exceeds the structural diversity of existing datasets by two orders of magnitude. Extensive experiments demonstrate that \MethodMotion outperforms specialist models on human and quadruped benchmarks while enabling zero-shot retargeting for the long tail of 3D creatures. Dataset is publicly available at https://huggingface.co/datasets/duckduckplz/Mobjaverse.
Chinese Translation
生成3D资产的爆炸性增长创造了对动画的巨大需求，然而当前的动作捕捉方法仍然脆弱，受限于特定物种的模板（例如，SMPL）或需要劳动密集型的手动绑定。我们提出了TopoCap，这是第一个统一框架，能够从单目视频中提取运动并将其重定向到具有任意、未见骨骼拓扑的角色上，即从双足动物到六足动物及无生命物体，而无需在测试时进行优化。我们的关键见解是，虽然骨骼结构是组合性和离散的，但运动的基本物理占据了一个连续的低维流形。我们通过一个两阶段的生成管道实现了这一见解。首先，我们使用图形条件变分自编码器（Graph CVAE）学习一个通用运动流形，将异质运动链压缩为共享的固定长度潜在编码。通过明确地将解码器条件化于目标绑定的结构嵌入，我们将运动动态与骨骼拓扑解耦。其次，我们将视频到动画视为一个条件流匹配问题，从视觉特征中预测这些拓扑无关的编码。为了学习这一通用先验，我们引入了Mobjaverse，这是一个从Objaverse-XL策划的大规模数据集。该数据集包含超过5000种独特的骨骼拓扑和200万帧，结构多样性超过现有数据集两个数量级。大量实验表明， extit{MethodMotion}在人体和四足动物基准测试中优于专业模型，同时实现了对3D生物长尾的零样本重定向。数据集已公开发布，网址为 https://huggingface.co/datasets/duckduckplz/Mobjaverse。
cs.CV / 83 / 2606.12169
OpenMedReason: Scientific Reasoning Supervision for Medical Vision-Language Models
OpenMedReason：医学视觉-语言模型的科学推理监督
Abstract
High-stakes clinical use of large vision-language models (LVLMs) requires reasoning that is grounded in visual evidence and clinical knowledge, not just correct final answers. We introduce OpenMedReason, a large-scale, open multimodal medical reasoning corpus comprising approximately 450K image-question-answer instances whose reasoning traces are primarily derived from curated biomedical, human-authored scientific articles. OpenMedReason provides high-fidelity supervision beyond synthetic chains of thought, covering diverse medical domain vision modalities such as radiological scans, microscopic images, visible light photographs, charts, and others. We complement it with OpenMedReason-Bench, a held-out benchmark that allows fine-grained evaluation of LVLMs along three complementary axes of capability, including perception, medical knowledge, and rationale, enabling diagnostic evaluation beyond final-answer accuracy. OpenMedReason is a rich training resource that exhibits its effectiveness in both supervised fine-tuning (SFT) and reinforcement-based alignment. Training with OpenMedReason yields a 20% average improvement in VQA accuracy over the base model and achieves performance within 4.2% of the strongest comparable-scale medical LVLMs. Fine-grained performance analysis confirms that the gains are not concentrated in any single axis: OpenMedReason improves perception, medical knowledge, and rationale jointly, and its reasoning traces are preferred over those of the base model in 86.1% of pairwise comparisons. We release the code and dataset at huggingface.co/datasets/neginb/OpenMedReason.
Chinese Translation
大型视觉-语言模型（LVLMs）在高风险临床应用中需要基于视觉证据和临床知识的推理，而不仅仅是正确的最终答案。我们介绍了OpenMedReason，这是一个大规模的开放多模态医学推理语料库，包含约450K个图像-问题-答案实例，其推理轨迹主要来源于经过策划的生物医学和人类撰写的科学文章。OpenMedReason提供高保真度的监督，超越了合成思维链，涵盖了多种医学领域的视觉模态，如放射学扫描、显微图像、可见光照片、图表等。我们还补充了OpenMedReason-Bench，这是一个保留的基准，允许对LVLMs在感知、医学知识和推理这三个互补能力轴上的细致评估，从而实现超越最终答案准确性的诊断评估。OpenMedReason是一个丰富的训练资源，在监督微调（SFT）和基于强化的对齐中展现了其有效性。使用OpenMedReason进行训练使VQA准确性平均提高了20%，并在性能上达到了与最强的可比规模医学LVLMs相差4.2%的水平。细致的性能分析确认这些提升并未集中在任何单一轴上：OpenMedReason在感知、医学知识和推理上共同改善，其推理轨迹在86.1%的成对比较中优于基线模型。我们将在huggingface.co/datasets/neginb/OpenMedReason上发布代码和数据集。
cs.CV / 84 / 2606.12171
Beyond Dark Knowledge: Mixup-Based Distillation for Reliable Predictions
超越暗知识：基于Mixup的蒸馏方法用于可靠预测
Abstract
Knowledge Distillation (KD) and mixup have proven effective at inducing smoothness in class boundaries; KD captures inherent class relationships in probability distributions, and mixup enforces them through convex combinations of inputs. Their interaction, however, remains poorly understood, particularly when mixup is applied only during student training. In this setting, the teacher is queried on inputs drawn from a vicinal distribution it never saw during training, a controlled mismatch whose effect on knowledge transfer has not been characterised. We show that this mismatch causes the teacher's supervisory signal to be dominated by distributional confusion rather than inter-class structure. Despite it, the student does not merely imitate the teacher: it independently acquires greater linearity in the vicinal region, a structural property that the teacher lacks, and goes beyond dark-knowledge transfer. KD with mixup consistently improves student accuracy and reduces overconfidence by an order of magnitude relative to the baseline, across CIFAR and ImageNet with varying-capacity teachers. Crucially, calibration propagates from teacher to student independently of accuracy transfer, and temperature scaling governs a measurable accuracy-calibration trade-off that becomes more pronounced under vicinal training. These results reframe mixup distillation not as a degraded version of standard KD, but as a richer transfer channel that simultaneously shapes discriminative performance, uncertainty estimation, and representational geometry.
Chinese Translation
知识蒸馏（Knowledge Distillation, KD）和mixup已被证明在平滑类别边界方面有效；KD捕捉概率分布中固有的类别关系，而mixup通过输入的凸组合来强化这些关系。然而，它们之间的相互作用仍然不够清晰，特别是在mixup仅在学生训练期间应用的情况下。在这种情况下，教师对从其未见过的邻近分布中抽取的输入进行查询，这是一种受控的不匹配，其对知识转移的影响尚未被表征。我们展示了这种不匹配导致教师的监督信号被分布混淆所主导，而非类别间结构。尽管如此，学生并不仅仅模仿教师：它在邻近区域独立获得了更大的线性度，这是一种教师所缺乏的结构特性，并超越了暗知识转移。结合mixup的KD在CIFAR和ImageNet上相对于基线一致地提高了学生的准确性，并将过度自信降低了一个数量级，适用于不同容量的教师。关键是，校准从教师传播到学生，与准确性转移无关，而温度缩放则控制着一个可测量的准确性-校准权衡，这在邻近训练下变得更加明显。这些结果将mixup蒸馏重新框定为一种更丰富的转移通道，同时塑造了辨别性能、不确定性估计和表征几何。
cs.CV / 85 / 2606.12189
DynaTok: Token-Based 4D Reconstruction from Partial Point Clouds
DynaTok：基于标记的部分点云的4D重建
Abstract
We address 4D reconstruction from partial point cloud sequences, where depth-sensor observations are incomplete, unordered, and lack explicit temporal correspondences. This geometry-only setting is challenging due to missing observations and ambiguous dynamics. While recent progress has largely relied on image-based methods, existing point-based approaches typically focus on single objects, assume relatively complete inputs, or require explicit correspondences. To address these limitations, we propose DynaTok, a point-based framework for correspondence-free 4D reconstruction from partial point cloud sequences without images. DynaTok encodes frames into compact latent tokens, aggregates incomplete observations over time with a Transformer-based spatiotemporal encoder, and decouples geometry and motion through residual tokens in a unified model. A flow-matching decoder then reconstructs complete, temporally consistent 4D point-cloud sequences conditioned on the latent tokens. Experiments on object- and scene-level benchmarks demonstrate improved reconstruction quality and temporal coherence from partial point cloud observations. Project page: https://wrchen530.github.io/dynatok/.
Chinese Translation
我们针对部分点云序列的4D重建问题进行研究，其中深度传感器的观测是不完整的、无序的，并且缺乏明确的时间对应关系。这种仅依赖几何信息的设置由于缺失的观测和模糊的动态特性而具有挑战性。尽管最近的进展主要依赖于基于图像的方法，但现有的基于点的方法通常集中于单一对象，假设输入相对完整，或需要显式的对应关系。为了解决这些局限性，我们提出了DynaTok，一个无需图像的基于点的框架，用于从部分点云序列中进行无对应关系的4D重建。DynaTok将帧编码为紧凑的潜在标记，通过基于Transformer的时空编码器聚合不完整的观测，并通过统一模型中的残差标记解耦几何和运动。然后，流匹配解码器根据潜在标记重建完整的、时间一致的4D点云序列。在对象和场景级基准测试中的实验表明，从部分点云观测中获得了更好的重建质量和时间一致性。项目页面：https://wrchen530.github.io/dynatok/
cs.CV / 86 / 2606.12195
InternVideo3: Agentify Foundation Models with Multimodal Contextual Reasoning
InternVideo3：通过多模态上下文推理实现智能化基础模型
Abstract
Recent progress in foundation models has shifted toward agentic behavior involving multi-step reasoning and tool use. However, open-source efforts largely focus on text-dominant settings, leaving long-horizon multimodal tasks underexplored. This gap is evident in video tasks requiring sustained temporal understanding and iterative interaction. We present InternVideo3, a framework enhancing these capabilities via Multimodal Contextual Reasoning (MCR). MCR treats understanding as a closed-loop process over a shared, evolving context containing observations, instructions, reasoning, tool actions, and memory. This frames long-video understanding as evidence accumulation and verification. To ensure efficiency, we introduce Multimodal Multi-head Latent Attention (M^2LA), a token-preserving reparameterization compressing KV-cache states while retaining the full token stream. Our staged training includes continued pretraining, short-to-long supervised fine-tuning, rule-based reinforcement learning, and on-policy distillation. Experiments show InternVideo3 achieves strong performance on benchmarks like Video-MME, MLVU, and EgoSchema. We further instantiate the model as a video agent with retrieval tools, demonstrating robust evidence-grounded behavior. Our results suggest that efficient context handling and closed-loop reasoning are vital for adapting open multimodal models toward long-horizon visually grounded agency.
Chinese Translation
近期基础模型的进展已转向涉及多步骤推理和工具使用的智能行为。然而，开源研究主要集中在以文本为主的环境中，导致长时间跨度的多模态任务尚未得到充分探索。这一差距在需要持续时间理解和迭代交互的视频任务中尤为明显。我们提出了InternVideo3，一个通过多模态上下文推理（Multimodal Contextual Reasoning, MCR）增强这些能力的框架。MCR将理解视为一个在共享、不断演变的上下文中进行的闭环过程，该上下文包含观察、指令、推理、工具操作和记忆。这将长视频理解框架化为证据的积累与验证。为了确保效率，我们引入了多模态多头潜在注意力（Multimodal Multi-head Latent Attention, M^2LA），这是一种保留令牌的重新参数化方法，能够压缩KV缓存状态，同时保留完整的令牌流。我们的分阶段训练包括持续的预训练、短至长的监督微调、基于规则的强化学习和在线蒸馏。实验表明，InternVideo3在Video-MME、MLVU和EgoSchema等基准测试中表现出色。我们进一步将该模型实例化为一个具有检索工具的视频智能体，展示了强大的基于证据的行为。我们的结果表明，高效的上下文处理和闭环推理对于将开放的多模态模型适应于长时间跨度的视觉基础智能至关重要。
cs.CV / 87 / 2606.12213
SHERPA: Seam-aware Harmonized ERP Adaptation for Open-Domain 360$^\circ$ Panorama Generation
SHERPA：面缝感知的和谐化ERP适应用于开放领域的360$^ ext{°}$全景生成
Abstract
Panoramic imagery is increasingly used in world-generation, games, and simulation, where users may need not only photorealistic scenes but also stylized and non-photorealistic environments. Large-scale text-to-image diffusion and flow models provide broad style and semantic priors for this goal, but planar image training misaligns them with the wrap-around topology and polar regions of $360^\circ$ panoramas represented in equirectangular projection (ERP). We present SHERPA, a lightweight adaptation framework that combines frequency-selective Circular RoPE, Circular Latent Encoding/Decoding, image-side FFN adapters, and a Dual-Path Training Scheme. Circular RoPE replaces only the seam-sensitive high-frequency horizontal RoPE band with integer-periodic harmonics while preserving the pretrained lower-frequency spectrum. The Paired Panorama Path supervises geometry, while the Unpaired Style Path uses self-supervised yaw consistency for target-free stylized prompts. As a result, SHERPA generates $360^\circ$ panoramas across both photorealistic panorama domains and open-domain stylized prompts.
Chinese Translation
全景图像在世界生成、游戏和模拟中越来越多地被使用，用户不仅需要逼真的场景，还需要风格化和非逼真的环境。大规模的文本到图像扩散和流模型为此目标提供了广泛的风格和语义先验，但平面图像训练使其与以等矩形投影（ERP）表示的360$^ ext{°}$全景的环绕拓扑和极地区域不对齐。我们提出了SHERPA，一个轻量级的适应框架，结合了频率选择性圆形RoPE、圆形潜在编码/解码、图像侧FFN适配器和双路径训练方案。圆形RoPE仅用整数周期谐波替换了对接缝敏感的高频水平RoPE带，同时保留了预训练的低频谱。配对全景路径监督几何，而未配对风格路径利用自监督的偏航一致性进行无目标风格化提示。因此，SHERPA能够在逼真的全景领域和开放领域的风格化提示中生成360$^ ext{°}$全景。
cs.CV / 88 / 2606.12215
MLT-Dedup: Efficient Large-Scale Online Video Deduplication via Multi-Level Representations and Spatial-Temporal Matching
MLT-Dedup：通过多层次表示和时空匹配实现高效的大规模在线视频去重
Abstract
The explosive growth of user-generated video content on online platforms is accompanied by the emergence of numerous near-duplicate videos--videos that are identical or highly similar but differ by partial edits. These duplicates degrade user experience and increase storage and bandwidth costs, making large-scale video deduplication a critical task. Existing video deduplication frameworks face a fundamental challenge in retrieving sufficient high-quality candidates under a limited index budget, as well as trade-offs between efficiency and precision. To address these issues, we propose MLT-Dedup, an efficient large-scale online video deduplication framework with Multi-Level representations and spatial-Temporal matching. Our approach employs a Multi-Level Video Encoder (ML-VE) to extract both fine-grained frame-level and sparse clip-level embeddings: sparse embeddings support efficient candidate retrieval, while fine-grained embeddings are loaded for precise pairwise matching. During matching, we introduce DiF-SiM, a Differential Feature-enhanced Similarity Module capable of locating duplicated temporal segments and providing reliable similarity evidence to support policy-driven deduplication decisions. Extensive experiments on a real-world large-scale platform demonstrate that MLT-Dedup reduces online repetition rates by 91% at 90% precision. Furthermore, our sparse retrieval design achieves a 5x increase in indexing capacity, enabling broader candidate coverage in real-world deployment.
Chinese Translation
用户生成的视频内容在在线平台上的爆炸性增长伴随着大量近重复视频的出现——这些视频在内容上相同或高度相似，但因部分编辑而有所不同。这些重复视频降低了用户体验，并增加了存储和带宽成本，使得大规模视频去重成为一项关键任务。现有的视频去重框架面临一个基本挑战，即在有限的索引预算下检索到足够的高质量候选视频，同时在效率和精度之间进行权衡。为了解决这些问题，我们提出了MLT-Dedup，一个高效的大规模在线视频去重框架，采用多层次表示和时空匹配。我们的方法使用多层次视频编码器（Multi-Level Video Encoder, ML-VE）提取细粒度的帧级和稀疏的片段级嵌入：稀疏嵌入支持高效的候选检索，而细粒度嵌入则用于精确的成对匹配。在匹配过程中，我们引入了DiF-SiM，一个增强差异特征的相似性模块，能够定位重复的时间段并提供可靠的相似性证据，以支持基于策略的去重决策。在一个真实的大规模平台上的广泛实验表明，MLT-Dedup在90%精度下将在线重复率降低了91%。此外，我们的稀疏检索设计实现了5倍的索引容量提升，使得在实际部署中能够覆盖更广泛的候选视频。
cs.CV / 89 / 2606.12217
Making Foresight Actionable: Repurposing Representation Alignment in World Action Models
使前瞻性可操作化：在世界行动模型中重新利用表示对齐
Abstract
World Action Models (WAMs) offer a promising route for robot manipulation by using video generation models to model future scene evolution before producing control actions. However, our empirical observations reveal a phenomenon: generating plausible visual futures does not always guarantee the extraction of accurate actions. To diagnose this failure, we conduct action-head attention analysis and causal interventions. We find that the action decoder fails to focus on task-relevant interaction regions and remains sensitive to perturbations in task-irrelevant areas. This reveals a representation mismatch: hidden states optimized for visual reconstruction are not inherently organized in a form useful for low-level action control. In this paper, we propose AGRA, an Action-Grounded Representation Alignment objective that regularizes the world-action interface by aligning intermediate video diffusion features with spatially coherent semantic representations from a foundation visual encoder. We evaluate AGRA on real-world manipulation tasks. Experiments show that AGRA makes world model representations more action-grounded: by focusing the action decoder on the correct interaction regions, it improves object localization accuracy and affordance understanding, and makes the policy more robust to perturbations in task-irrelevant regions. As a result, AGRA consistently improves both in-distribution performance and out-of-distribution generalization over the baseline world action model.
Chinese Translation
世界行动模型（World Action Models, WAMs）通过使用视频生成模型来模拟未来场景演变，从而为机器人操作提供了一条有前景的路径，随后再生成控制动作。然而，我们的实证观察揭示了一种现象：生成可信的视觉未来并不总能保证提取准确的动作。为了诊断这一失败，我们进行了动作头注意力分析和因果干预。我们发现，动作解码器未能专注于与任务相关的交互区域，并且对与任务无关的区域的扰动保持敏感。这揭示了一种表示不匹配：为视觉重建优化的隐藏状态并不天然以对低级动作控制有用的形式组织。在本文中，我们提出了AGRA（Action-Grounded Representation Alignment），一种通过将中间视频扩散特征与基础视觉编码器的空间一致语义表示对齐来规范世界-动作接口的目标。我们在真实世界的操作任务上评估了AGRA。实验表明，AGRA使世界模型表示更加以动作为基础：通过将动作解码器聚焦于正确的交互区域，它提高了物体定位的准确性和可用性理解，并使策略对与任务无关区域的扰动更加稳健。因此，AGRA在基线世界行动模型上持续提高了分布内性能和分布外泛化能力。
cs.CV / 90 / 2606.12218
Adapting Prithvi-EO for Fallow Detection for Food-Water Nexus: ViT-Adapter Necks and Parameter-Efficient Backbone tuning of Geospatial Foundation Model
为食品-水资源联系适应 Prithvi-EO 的休耕检测：ViT-Adapter 颈部和地理空间基础模型的参数高效骨干调优
Abstract
Understanding spatial distribution of fallow land is important for optimizing the food-water (FW) nexus, given fallowing's role in crop rotation and water conservation. Fallow is a low accuracy class in USDA Cropland Data Layer (CDL). Geospatial foundation model (GFM), Prithvi-EO has shown strong transferability across computer vision tasks. However, its Vision Transformer (ViT) backbone produces features at a single spatial scale that are ill-suited for the multi-scale features required by object detection heads. Existing approaches synthesise multi-scale pyramids through scaling of single stride tokens, sacrificing spatial heterogeneity, and full backbone fine-tuning is computationally prohibitive for GFMs. We evaluate a fallow detection pipeline combining two parameter-efficient fine tuning (PEFT) schemes: Low-Rank Adaptation (LoRA) and a hybrid PEFT, with three neck designs: pseudo multi-scale, Lite ViT-Adapter, and Full ViT-Adapter. Our best configuration, Lite ViT-Adapter with a one-stage head, achieves a mAP@50 of 0.9479 with the Diou loss, suggesting the effectiveness of center-aware localization for irregular fallow field detection. ViT-Adapter free one-stage detection under LoRA improves the adapter-free anchor-based approach by 6.42%, and the best configuration improves baseline adapter-free anchor-based approach by 25.70%. These results demonstrate that lightweight spatial prior fusion and selective backbone unfreezing enable Prithvi-EO to capture local fallow patterns more effectively, outperforming approaches that rely on reshaped single-stride ViT tokens.
Chinese Translation
理解休耕地的空间分布对于优化食品-水资源（FW）联系至关重要，因为休耕在作物轮作和水资源保护中发挥着重要作用。休耕在美国农业部耕地数据层（CDL）中属于低准确度类别。地理空间基础模型（GFM）Prithvi-EO 在计算机视觉任务中表现出强大的迁移能力。然而，其视觉变换器（ViT）骨干网络生成的特征仅在单一空间尺度上，这不适合目标检测头所需的多尺度特征。现有方法通过缩放单步长标记合成多尺度金字塔，牺牲了空间异质性，而对 GFM 进行全面的骨干微调在计算上是不可行的。我们评估了一种结合两种参数高效微调（PEFT）方案的休耕检测管道：低秩适应（LoRA）和混合 PEFT，以及三种颈部设计：伪多尺度、Lite ViT-Adapter 和 Full ViT-Adapter。我们最佳配置 Lite ViT-Adapter 与单阶段头结合，使用 Diou 损失达到了 0.9479 的 mAP@50，表明中心感知定位在不规则休耕地检测中的有效性。在 LoRA 下，ViT-Adapter 自由的一阶段检测比无适配器锚点方法提高了 6.42%，而最佳配置比基线无适配器锚点方法提高了 25.70%。这些结果表明，轻量级空间先验融合和选择性骨干解冻使 Prithvi-EO 更有效地捕捉局部休耕模式，优于依赖重塑单步长 ViT 标记的方法。
cs.CV / 91 / 2606.12226
An Electric Potential-Augmented Benchmark Dataset for Physics-Guided Image Reconstruction of Electrical Capacitance Tomography
电势增强的电容层析成像物理引导图像重建基准数据集
Abstract
While deep learning has significantly advanced image reconstruction of Electrical Capacitance Tomography (ECT), most data-driven methods map directly between capacitance and permittivity distribution, treating the sensor as a black box. This overlooks the electric potential field -- the fundamental physical link governing the nonlinear and ill-posed ``soft-field'' effect. To address this, we propose an electric potential-augmented ECT benchmark dataset designed to explicitly integrate latent physics behind ECT into the learning process. Generated via a COMSOL-MATLAB pipeline for an eight-electrode sensor as an example, the dataset comprises 20,000 randomized samples across four typical flow patterns. Crucially, alongside the conventional capacitance vectors and permittivity distributions depicted as images, each sample preserves eight excitation-wise full-field potential maps. Beyond data release, we provide illustrative evaluation protocols for both forward and inverse problems of ECT. Through comprehensive testing on both in-distribution (IID) and out-of-distribution (OOD) scenarios, we systematically demonstrate how the inclusion of electric potential maps enhances modeling accuracy and robustness. Fundamentally, the explicit inclusion of latent field information significantly lowers the barrier to integrating physical laws into ECT modeling, thereby establishing a standardized foundation for future physics-guided machine learning of ECT image reconstruction.
Chinese Translation
尽管深度学习在电容层析成像（ECT）的图像重建方面取得了显著进展，但大多数数据驱动的方法直接在电容和介电常数分布之间进行映射，将传感器视为黑箱。这忽视了电势场——这一支配非线性和病态“软场”效应的基本物理联系。为了解决这一问题，我们提出了一个电势增强的ECT基准数据集，旨在明确将ECT背后的潜在物理知识整合到学习过程中。该数据集通过COMSOL-MATLAB管道生成，以八电极传感器为例，包含20,000个随机样本，涵盖四种典型流动模式。重要的是，除了以图像形式呈现的常规电容向量和介电常数分布外，每个样本还保留了八个激励方式的全场电势图。除了数据发布外，我们还提供了ECT正向和反向问题的示范评估协议。通过对同分布（IID）和异分布（OOD）场景的全面测试，我们系统地展示了电势图的纳入如何提高建模的准确性和鲁棒性。从根本上讲，明确纳入潜在场信息显著降低了将物理法则整合到ECT建模中的障碍，从而为未来的物理引导机器学习在ECT图像重建中的应用奠定了标准化基础。
cs.CV / 92 / 2606.12248
Damage-TriageFormer: A Foundation-Model Framework for Typology-Based Building Damage Assessment from Mono-Temporal Imagery
Damage-TriageFormer：基于类型学的单时相影像建筑损伤评估基础模型框架
Abstract
Decision-relevant building damage assessment is critical for prioritizing resources and recovery after a disaster, yet most automated methods either flatten damage into a single severity scale (no damage, minor, major, destroyed) or require paired pre- and post-event imagery that is often unavailable for emerging hazards. This paper presents Damage-TriageFormer, a single-image, post-event, footprint-conditioned model that produces a damage typology rather than a severity scale. We contribute: (1) DamageTriage-Bench, a new benchmark built from NOAA Emergency Response Imagery across Hurricane Michael (2018), Hurricane Helene (2024), and the 2025 Los Angeles wildfire complex, with five typology classes that distinguish roof damage from structural damage and, within each, partial from total extent; and (2) Damage-TriageFormer, which extends a DINOv3 ViT-L backbone with a Simple Feature Pyramid for higher-resolution instance pooling, a two-stage gated damage head, and an auxiliary severity-regression objective. Our model achieves macro F1 of 0.624 on validation and 0.619 on a held-out stratified test set, performing strongest where operational triage needs it most, with per-class F1 of 0.91 and 0.84 on undamaged buildings and total structural collapse, respectively. While the rare Total Roof Damage class remains difficult due to its limited examples and an inherently ambiguous label boundary, our results show that single-image post-event imagery can support actionable building damage typing, enabling targeted emergency response and resource allocation without a pre-event reference.
Chinese Translation
与决策相关的建筑损伤评估对于灾后资源优先配置和恢复至关重要，但大多数自动化方法要么将损伤简化为单一的严重程度等级（无损伤、轻微、重大、毁坏），要么需要配对的事件前后影像，而后者在新兴灾害中往往不可用。本文提出了Damage-TriageFormer，这是一种基于单幅影像的后事件、足迹条件模型，旨在生成损伤类型学而非严重程度等级。我们的贡献包括：(1) DamageTriage-Bench，这是一个新的基准，基于NOAA应急响应影像，涵盖迈克尔飓风（2018年）、海伦飓风（2024年）和2025年洛杉矶野火综合体，包含五个类型学类别，区分屋顶损伤与结构损伤，并在每个类别中区分部分损伤与完全损伤；(2) Damage-TriageFormer，它扩展了DINOv3 ViT-L骨干网络，采用简单特征金字塔以实现更高分辨率的实例池化，设有两阶段的门控损伤头和辅助的严重程度回归目标。我们的模型在验证集上获得了0.624的宏观F1分数，在保留的分层测试集上为0.619，在操作性分流需求最强的地方表现最佳，未损坏建筑和完全结构倒塌的每类F1分数分别为0.91和0.84。尽管稀有的完全屋顶损伤类别由于其有限的示例和固有的模糊标签边界仍然难以处理，但我们的结果表明，单幅影像的后事件影像可以支持可操作的建筑损伤类型识别，从而在没有事件前参考的情况下，实现针对性的应急响应和资源配置。
cs.CV / 93 / 2606.12258
Bridging Day and Night: Unsupervised Cross-Domain Re-Identification with Synergistic Prompt and Prototype Learning
弥合白天与黑夜：基于协同提示和原型学习的无监督跨域重识别
Abstract
Cross-domain day-night re-identification (ReID) is fundamentally challenged by the substantial visual appearance discrepancies between daytime and nighttime scenes. Existing fully supervised methods rely heavily on labor-intensive annotations, which are costly and exhibit limited generalization across domains. In this work, we investigate unsupervised day-night ReID and propose a novel framework that synergistically combines prompt learning and prototype-based representation learning to associate identities across domains without requiring manual labels. Our approach follows a progressive two-stage training strategy. In the first stage, we exploit the vision-language model to generate instance-specific textual prompts in an annotation-free manner. We employ an instance-level alignment mechanism to embed visual features and textual prompts into a unified semantic space, aligning unlabeled day/night images with learnable prompts via instance-aware dynamic-bias adaptation. In the second stage, we construct domain-specific prototype memory banks and introduce two complementary modules: i) an intra-domain identity association module to enhance feature discriminability within each domain, and ii) a cross-domain prototype matching module to reliably identify positive and negative prototype pairs, thereby establishing robust identity correspondences across day and night. Extensive experiments on public benchmarks validate the effectiveness of our method. Under the unsupervised setting, our framework attains Rank-1 accuracy comparable to state-of-the-art fully supervised methods.
Chinese Translation
跨域白天-黑夜重识别（ReID）面临着白天和黑夜场景之间显著的视觉外观差异的根本挑战。现有的完全监督方法在很大程度上依赖于劳动密集型的标注，这不仅成本高昂，而且在跨域的泛化能力有限。在本研究中，我们探讨了无监督的白天-黑夜重识别，并提出了一种新颖的框架，协同结合了提示学习和基于原型的表示学习，以在不需要手动标签的情况下关联跨域身份。我们的方法遵循渐进的两阶段训练策略。在第一阶段，我们利用视觉-语言模型以无标注的方式生成特定实例的文本提示。我们采用实例级对齐机制，将视觉特征和文本提示嵌入到统一的语义空间中，通过实例感知的动态偏差适应将未标记的白天/黑夜图像与可学习的提示对齐。在第二阶段，我们构建域特定的原型记忆库，并引入两个互补模块：i) 一个域内身份关联模块，以增强每个域内特征的可区分性；ii) 一个跨域原型匹配模块，以可靠地识别正负原型对，从而在白天和黑夜之间建立稳健的身份对应关系。在公共基准上的大量实验验证了我们方法的有效性。在无监督设置下，我们的框架达到了与最先进的完全监督方法相当的Rank-1准确率。
cs.CV / 94 / 2606.12263
VOID: Defeating Unauthorized Mimicry in Latent Diffusion Models
VOID：在潜在扩散模型中击败未经授权的模仿
Abstract
While Latent Diffusion Models (LDMs) have revolutionized visual synthesis, they are increasingly exploited for unauthorized mimicry of individuals. Existing defenses inject deceptive perturbations to steer the generated images toward irrelevant targets. However, this approach hinges on an ungrounded assumption: subtle perturbations can maintain their deceptive efficacy throughout an LDM's extensive generation process. In reality, the model's innate restoration mechanism will remove such perturbations and cause individual identities to re-emerge in the images generated. We propose VOID, a defense framework that overcomes this conundrum by manipulating an LDM's intrinsic stochasticity. VOID perturbs the diffusion pipeline in two novel ways: 1) amplifying the latent encoding errors to shatter an image's semantic structure, and 2) counteracting the target guidance signals to suppress the model's restoration capabilities. This results in a semantic corruption that thwarts any unauthorized mimicry. Notably, the security gain does not come at the price of visual utility, as VOID simultaneously manages to confine perturbations to human-imperceptible regions of protected images. Our comprehensive evaluation of 24 state-of-the-art defenses against 10 mimicry attacks on 5 datasets demonstrates VOID's unprecedented protection power: it increases the average Frechet Inception Distance (FID) from 113 to 365, a 223% improvement over the strongest defense to date.
Chinese Translation
尽管潜在扩散模型（LDMs）已经彻底改变了视觉合成，但它们越来越多地被用于对个体的未经授权的模仿。现有的防御方法通过注入误导性扰动来引导生成的图像朝向无关目标。然而，这种方法依赖于一个没有根据的假设：细微的扰动可以在LDM的广泛生成过程中保持其误导效果。实际上，模型固有的恢复机制会去除这些扰动，并导致生成图像中个体身份的重新出现。我们提出了VOID，一个通过操控LDM内在随机性来克服这一难题的防御框架。VOID以两种新颖的方式扰动扩散流程：1）放大潜在编码错误以破坏图像的语义结构；2）抵消目标引导信号以抑制模型的恢复能力。这导致了语义的腐败，从而阻止了任何未经授权的模仿。值得注意的是，安全性提升并未以视觉效用为代价，因为VOID同时成功地将扰动限制在受保护图像的人类不可感知区域。我们对24种最先进的防御方法在5个数据集上针对10种模仿攻击的全面评估表明，VOID具有前所未有的保护能力：它将平均Frechet Inception Distance（FID）从113提高到365，较目前最强的防御提高了223%。
cs.CV / 95 / 2606.12278
Finding Sparse Subnetworks in One Training Cycle via Progressive Magnitude-Based Pruning
通过渐进式基于幅度的剪枝在一个训练周期内寻找稀疏子网络
Abstract
Neural network pruning reduces model size by removing less important parameters while aiming to preserve predictive performance. Although the Lottery Ticket Hypothesis (LTH) shows that sparse subnetworks can match dense networks when trained from suitable initializations, its iterative pruning procedure requires multiple complete training cycles. This work evaluates progressive magnitude-based pruning as a single-cycle alternative. The method gradually increases sparsity during training using a linear schedule and updates pruning masks based on active weight magnitudes. We conduct systematic experiments on CIFAR-10 and MNIST across ResNet, VGG-style, and LeNet architectures, comparing the proposed method with representative iterative and initialization-based pruning baselines, including LTH, SNIP, and GraSP. On CIFAR-10, the method achieves 95.12\% accuracy on ResNet-18 at 72.9\% sparsity, compared with 90.5\% reported for LTH. At extreme sparsity, it achieves 93.13\% accuracy on a VGG-like architecture at 97\% sparsity, compared with approximately 92.0\% for SNIP, and 93.44\% accuracy on VGG-19 at 97.97\% sparsity, compared with 92.19\% for GraSP at 98\% sparsity. A sparsity-accuracy analysis on ResNet-18 further shows that accuracy remains within 0.1 percentage points of the dense baseline across 70--85\% sparsity. These results indicate that progressive magnitude-based pruning provides an effective single-cycle approach for neural network sparsification under the evaluated settings.
Chinese Translation
神经网络剪枝通过去除不太重要的参数来减少模型大小，同时旨在保持预测性能。尽管彩票票据假设（Lottery Ticket Hypothesis, LTH）表明稀疏子网络在从合适的初始化训练时可以与密集网络相匹配，但其迭代剪枝过程需要多个完整的训练周期。本研究评估了渐进式基于幅度的剪枝作为单周期替代方案。该方法在训练过程中使用线性调度逐步增加稀疏性，并根据活跃权重的幅度更新剪枝掩码。我们在CIFAR-10和MNIST数据集上对ResNet、VGG风格和LeNet架构进行了系统实验，将所提出的方法与代表性的迭代和基于初始化的剪枝基线进行比较，包括LTH、SNIP和GraSP。在CIFAR-10上，该方法在72.9%的稀疏性下，在ResNet-18上实现了95.12%的准确率，而LTH报告的准确率为90.5%。在极端稀疏情况下，该方法在97%的稀疏性下，在VGG类架构上实现了93.13%的准确率，而SNIP的准确率约为92.0%；在97.97%的稀疏性下，在VGG-19上实现了93.44%的准确率，而GraSP在98%稀疏性下的准确率为92.19%。对ResNet-18的稀疏性-准确率分析进一步表明，在70%至85%的稀疏性范围内，准确率始终保持在密集基线的0.1个百分点以内。这些结果表明，渐进式基于幅度的剪枝在评估的设置下提供了一种有效的单周期神经网络稀疏化方法。
cs.CV / 96 / 2606.12286
CellNet -- Localizing Cells using Sparse and Noisy Point Annotations
CellNet -- 使用稀疏和噪声点注释定位细胞
Abstract
Counting living cells is an important step in many biological research workflows. Our collaborators at the Wellcome Sanger Institute study vital genes in humans via large scale saturation genome editing screening, which requires repeatedly counting cells a great number of times. Computer Vision based automation is crucial for high throughput and resource efficiency. In this work, we develop a regression-based deep learning computer vision algorithm to detect and count cells in phase-contrast microscopy images. To reduce annotation effort, which in practice often becomes a bottleneck, we focus on counting cells only using sparse point annotations, which are fast and easy to acquire. By comparison to state-of-the-art 0-shot methods, we show that regression-based counting is a promising alternative in low data regimes. Through developing methods to automatically count living cells in microscopy images, we contribute to valuable research on the human genome. The code is available at https://github.com/beijn/cellnet.
Chinese Translation
计数活细胞是许多生物研究工作流程中的重要步骤。我们在威康桑格研究所的合作伙伴通过大规模饱和基因组编辑筛选研究人类的重要基因，这需要对细胞进行多次重复计数。基于计算机视觉的自动化对于高通量和资源效率至关重要。在本研究中，我们开发了一种基于回归的深度学习计算机视觉算法，用于在相位差显微镜图像中检测和计数细胞。为了减少注释工作量（在实践中常常成为瓶颈），我们专注于仅使用稀疏的点注释来计数细胞，这种注释方式快速且易于获取。与最先进的零-shot方法相比，我们展示了基于回归的计数在低数据环境中是一种有前景的替代方案。通过开发自动计数显微镜图像中活细胞的方法，我们为人类基因组的有价值研究做出了贡献。代码可在 https://github.com/beijn/cellnet 获取。
cs.CV / 97 / 2606.12294
Bridging the Modality Gap in Forensic Image Retrieval
弥合法医图像检索中的模态差距
Abstract
Automated image retrieval plays an increasingly critical role in modern forensic analysis, supporting investigative workflows that rely on efficient comparison of visual evidence. While prior work has focused primarily on developing and optimizing multimodal retrieval systems, limited attention has been paid to evaluating the forensic applicability of these technologies across diverse real-world scenarios. In this study, we present a unified retrieval framework adapted to four key forensic tasks: (1) tattoo image retrieval given a tattoo query image; (2) tattoo retrieval guided by human-expert textual descriptions, modelling the common situation where a witness verbally describes a tattoo; (3) tattoo retrieval from hand-drawn sketches; and (4) face retrieval from forensic face sketches. Our system leverages a multimodal large language model (MLLM) to automatically generate structured textual descriptions for all queries and gallery images, followed by sentence-transformer embedding for text-based comparison. We evaluate retrieval using visual-only embeddings, text-only embeddings and a multimodal fusion strategy that combines text- and image-based similarity scores derived from state-of-the-art visual feature extractors relevant to each task. The fusion of modalities consistently improves retrieval precision and robustness, especially in scenarios where visual information is limited or noisy (e.g., sketches, partial tattoos, or fragmented witness statements). This work highlights the forensic value of a unified multimodal retrieval pipeline and demonstrates how modern MLLMs can operationalize challenging forensic tasks that traditionally rely on manual expert analysis. Our results position multimodal retrieval as a promising tool for supporting investigative workflows involving tattoos, facial composites, and witness descriptions.
Chinese Translation
自动化图像检索在现代法医分析中扮演着越来越重要的角色，支持依赖于高效比较视觉证据的调查工作流程。尽管先前的研究主要集中于开发和优化多模态检索系统，但对这些技术在多样化现实场景中的法医适用性评估的关注相对有限。在本研究中，我们提出了一个统一的检索框架，适用于四个关键法医任务：(1) 给定纹身查询图像的纹身图像检索；(2) 由人类专家文本描述引导的纹身检索，模拟证人口头描述纹身的常见情况；(3) 从手绘草图中检索纹身；以及 (4) 从法医面部草图中检索面部图像。我们的系统利用多模态大型语言模型（MLLM）自动生成所有查询和图库图像的结构化文本描述，随后进行基于句子变换器的嵌入以进行文本比较。我们使用仅基于视觉的嵌入、仅基于文本的嵌入以及结合来自与每个任务相关的最先进视觉特征提取器的文本和图像相似性得分的多模态融合策略来评估检索。模态的融合始终提高了检索的精度和鲁棒性，尤其是在视觉信息有限或噪声较大的场景中（例如，草图、部分纹身或零散的证人陈述）。这项工作突显了统一多模态检索管道的法医价值，并展示了现代MLLM如何将传统上依赖于人工专家分析的挑战性法医任务转化为可操作的流程。我们的结果将多模态检索定位为支持涉及纹身、面部合成和证人描述的调查工作流程的有前景的工具。
cs.CV / 98 / 2606.12295
Findings of the MAGMaR 2026 Shared Task
MAGMaR 2026 共享任务的研究结果
Abstract
This overview paper presents the results of the shared task for the second workshop on Multimodal Augmented Generation via Multimodal Retrieval (MAGMaR). In this shared task participants submitted systems focused on either (i) video retrieval or (ii) grounded generation of articles given retrieved videos. Teams could submit to either task. For the retrieval task, we had 2 participating teams that submitted a total of 17 systems -- all of which beat a baseline derived from the winner of last year's shared task. On the generation side, we had 4 teams submit 16 systems. All teams had at least one generated report that was labeled the best by a human annotator.
Chinese Translation
本文概述了第二届多模态增强生成与多模态检索研讨会（MAGMaR）的共享任务结果。在该共享任务中，参与者提交了专注于（i）视频检索或（ii）基于检索视频的文章生成的系统。团队可以选择参与任一任务。在检索任务中，我们有2个参与团队提交了总共17个系统，所有系统均超越了基于去年共享任务获胜者的基线。在生成任务方面，我们有4个团队提交了16个系统。所有团队至少有一个生成报告被人类标注者评为最佳。
cs.CV / 99 / 2606.12300
Natural-Language Temporal Grounding in Hour-Long Videos is a Search Problem: A Benchmark and Empirical Decomposition
自然语言时间定位在小时级视频中是一个搜索问题：基准测试与实证分解
Abstract
Temporal grounding--returning the interval $[t_s, t_e]$ for a natural-language query over a video--is the language interface to long-form video, yet has been studied on short videos; the dynamics of hour-scale natural-language grounding remain underexplored. We take the position that at hour-scale, the binding constraint is search, not recognition: Video-LLMs are bottlenecked not by localizing a nearby event, but--given a natural-language query--by searching for the relevant region of a long video. To test this, we release ExtremeWhenBench, the first open hour-scale grounding benchmark (2,273 queries over 194 videos, mean 75.7 min, max 9 hr) with an open-form query distribution. Every open Video-LLM collapses while a frame-level retrieval baseline outperforms them; a failure taxonomy attributes 85% of failures to search; and a retrieve-then-ground hybrid recovers 6.7x over the monolithic Video-LLM--mirroring retrieve-then-read in open-domain QA.
Chinese Translation
时间定位——为视频中的自然语言查询返回区间 $[t_s, t_e]$——是长视频的语言接口，但目前的研究主要集中在短视频上；而小时级自然语言定位的动态仍然未被充分探索。我们认为，在小时级别，主要的限制因素是搜索，而非识别：视频大语言模型（Video-LLMs）的瓶颈并不在于定位附近事件，而是在于给定自然语言查询后，搜索长视频的相关区域。为了验证这一观点，我们发布了 ExtremeWhenBench，这是第一个开放的小时级定位基准（包含 2,273 个查询，覆盖 194 个视频，平均时长 75.7 分钟，最长 9 小时），并具有开放式查询分布。每个开放的视频大语言模型在此基准中表现不佳，而基于帧级检索的基线则超越了它们；失败分类法将 85% 的失败归因于搜索问题；而检索后定位的混合方法在单一视频大语言模型上恢复了 6.7 倍的性能，类似于开放域问答中的检索后阅读。
cs.CV / 100 / 2606.12303
From 2D Grids to 1D Tokens: Reforming Shared Representations for Multimodal Image Fusion
从二维网格到一维标记：重塑多模态图像融合的共享表示
Abstract
Multimodal image fusion aims to integrate complementary information from different modalities into a fused image that preserves rich local details while maintaining globally consistent appearance. Existing approaches build shared representations on 2D feature grids, which excel at modeling local structures but offer limited leverage over image-level global appearance factors. To balance these objectives, we introduce a compact 1D token interface based on a frozen pretrained image tokenizer for modeling non-local appearance/base factors. Rather than using the tokenizer as a reconstruction backbone, our design uses the 1D token space as a global carrier while retaining the 2D spatial pathway for local structure restoration. Specifically, we introduce Selective Token Editing (STE), which sparsely updates/replaces a small set of critical tokens, providing a lightweight mechanism to steer global appearance coherence while keeping the fusion backbone unchanged and avoiding extra losses. Experiments on four commonly used benchmarks show that our method achieves the best overall performance, with consistent, multi-metric improvements in both global coherence and local fidelity. Project page: https://zju-xyc.github.io/1D-Fusion-Project-Page/
Chinese Translation
多模态图像融合旨在将来自不同模态的互补信息整合成一幅融合图像，既保留丰富的局部细节，又保持全局一致的外观。现有方法在二维特征网格上构建共享表示，虽然在建模局部结构方面表现优异，但对图像级全局外观因素的把握有限。为了平衡这些目标，我们引入了一种基于冻结的预训练图像标记器的紧凑一维标记接口，用于建模非局部外观/基础因素。我们的设计并不是将标记器用作重建主干，而是将一维标记空间作为全局载体，同时保留二维空间路径以恢复局部结构。具体而言，我们引入了选择性标记编辑（Selective Token Editing，STE），该方法稀疏地更新/替换一小组关键标记，提供了一种轻量级机制来引导全局外观一致性，同时保持融合主干不变，避免额外损失。在四个常用基准测试上的实验表明，我们的方法在整体性能上达到了最佳，且在全局一致性和局部保真度方面均实现了一致的多指标提升。项目页面：https://zju-xyc.github.io/1D-Fusion-Project-Page/
cs.CV / 101 / 2606.12316
Slots, Transitions, Loops: Learning Composable World Models for ARC
槽、过渡、循环：学习可组合的ARC世界模型
Abstract
ARC tests in-context rule induction: given a few input-output demonstrations, a model must infer the hidden rule and apply it to a new query. While many approaches express ARC rules through language, code, or symbolic programs, ARC itself is visual-symbolic: rules appear as grid transitions over objects, colors, shapes, and spatial relations. We introduce Loop-OWM, an object-centric world-modeling architecture that learns these rules as composable transitions over structured states. It combines color-prototype slots, demonstration-conditioned task summaries, and a looped transition model with dense propagation and slot-conditioned correction. On both ARC-1 and ARC-2, Loop-OWM outperforms non-looped and looped baselines with comparable or fewer parameters. These results suggest that ARC rules can be learned not only as language descriptions or searched programs, but also as transitions over visual-symbolic world states.
Chinese Translation
ARC测试了上下文中的规则归纳：给定少量的输入-输出示例，模型必须推断隐藏的规则并将其应用于新的查询。虽然许多方法通过语言、代码或符号程序表达ARC规则，但ARC本身是视觉-符号的：规则表现为对象、颜色、形状和空间关系上的网格过渡。我们引入了Loop-OWM，这是一种以对象为中心的世界建模架构，学习这些规则作为结构化状态上的可组合过渡。它结合了颜色原型槽、基于示例的任务总结和具有密集传播及槽条件修正的循环过渡模型。在ARC-1和ARC-2上，Loop-OWM的表现优于非循环和循环基线，同时参数量相当或更少。这些结果表明，ARC规则不仅可以作为语言描述或搜索程序进行学习，也可以作为视觉-符号世界状态上的过渡进行学习。
cs.CV / 102 / 2606.12319
Anatomically Conditioned Recurrent Refinement for Topology-Aware Circle of Willis Segmentation
解剖条件递归细化用于拓扑感知的威利环分割
Abstract
Segmenting the Circle of Willis (CoW) from Magnetic Resonance Angiography (MRA) is challenging due to complex topology and thin vascular structures that are prone to fragmentation. Standard Convolutional Neural Networks (CNNs) often fail to capture these topological constraints, resulting in "broken vessel" artifacts. To address this, we propose the Anatomically Conditioned Recurrent Refinement U-Net (AC2RUNet). Our architecture decouples segmentation into two streams: a Static Stream that extracts invariant anatomical features and a lightweight Dynamic Stream that iteratively refines topological errors over time. We further introduce a dynamic curriculum learning strategy that transitions from high-recall geometric supervision to topology-aware constraints. Validated on the TopCoW dataset, AC2RUNet substantially reduces Hausdorff Distance (4.72 mm vs 9.17 mm) and Betti number errors (0.19 vs 0.40), improving topological connectivity over the nnU-Net baseline while maintaining comparable volumetric Dice.
Chinese Translation
从磁共振血管成像（MRA）中分割威利环（CoW）具有挑战性，因为其复杂的拓扑结构和易碎的血管结构容易导致碎片化。标准卷积神经网络（CNN）往往无法捕捉这些拓扑约束，导致出现“断裂血管”伪影。为了解决这个问题，我们提出了解剖条件递归细化U-Net（AC2RUNet）。我们的架构将分割过程解耦为两个流：一个静态流提取不变的解剖特征，另一个轻量级动态流则随着时间的推移迭代细化拓扑错误。我们进一步引入了一种动态课程学习策略，从高召回率的几何监督过渡到拓扑感知约束。在TopCoW数据集上验证，AC2RUNet显著降低了Hausdorff距离（4.72 mm对比9.17 mm）和Betti数错误（0.19对比0.40），在保持可比体积Dice的同时改善了拓扑连通性，相较于nnU-Net基线表现更佳。
cs.CV / 103 / 2606.12340
Echoes of the Prior: A Computational Phenomenology of Forgetting
先前的回声：遗忘的计算现象学
Abstract
Memory is not merely the storage of data; it is the scaffolding of reality. When biological memory fades, the world does not simply turn black; it regresses into an unrecognizable chaos. Echoes of the Prior is an interactive installation that attempts to visualize this subjective phenomenology of forgetting. By inducing controlled synaptic decay within a Feed-Forward 3D Reconstruction model, we create an artistic analogy for the erosion of the brain's predictive priors. We position the Neural Network not as a tool for engineering, but as a cognitive proxy - a silicon brain whose structural degeneration evokes the disorienting, poetic, and terrifying experience of losing one's grip on the world. Ultimately, we offer this framework as a catalyst, inviting the wider community to explore the uncharted potential of neuromorphic aesthetics in visualizing the fragility of intelligence. Interactive demo see https://decart-4d.github.io/.
Chinese Translation
记忆不仅仅是数据的存储；它是现实的支架。当生物记忆消退时，世界并不会简单地变黑；它会退化为一种无法识别的混沌。《先前的回声》是一个互动装置，试图可视化这种主观的遗忘现象学。通过在前馈三维重建模型中诱导受控的突触衰退，我们为大脑预测先验的侵蚀创造了一个艺术类比。我们将神经网络视为一种认知代理，而非工程工具——一个硅基大脑，其结构退化唤起了失去对世界掌控的迷失、诗意和恐惧的体验。最终，我们将这一框架作为催化剂，邀请更广泛的社区探索神经形态美学在可视化智能脆弱性方面的未知潜力。互动演示请见 https://decart-4d.github.io/。
cs.CV / 104 / 2606.12346
Atlas H&E-TME: Scalable AI-Based Tissue Profiling at Expert Pathologist-Level Accuracy
Atlas H&E-TME：基于人工智能的组织分析，达到专家病理学家水平的准确性
Standvoss, Kai, Hägele, Miriam, Krupar, Rosemarie, Ribbat-Idel, Julika, Altschüler, Jennifer, Erdmann, Gerrit, Pinckaers, Hans, Ramberger, Evelyn, Drinkwitz, Madleen, Nárai, Ádám, Möllers, Alexander, Lingelbach, Katja, Kons, Sebastian, Hönig, Lukas, Adigüzel, Recepcan, Baião, Joana, Gonzalo, Alberto Megina, Teodorescu, Marius, Eich, Marie-Lisa, Chetta, Paolo, Merchant, Shakil, Aumiller, Verena, Schallenberg, Simon, Norgan, Andrew, Müller, Klaus-Robert, Ruff, Lukas, Alber, Maximilian, Klauschen, Frederick
Abstract
Hematoxylin and eosin (H&E) staining is the cornerstone of histopathology, yet scalable, quantitative analysis of H&E whole-slide images (WSIs) remains a central challenge in computational pathology. We present Atlas H&E-TME, an AI-based system built on the Atlas family of pathology foundation models that predicts tissue quality, tissue region, and cell type labels across multiple cancer types, yielding over 4,500 quantitative readouts per slide at cell-level resolution. A key challenge to validating such systems is overcoming morphological ambiguity inherent to H&E-only ground truth and the limited scalability of more informed references drawing on modalities such as immunohistochemistry (IHC). We address this with a dual validation framework combining biologically grounded depth with technical and morphological breadth. For depth, we propose an IHC-informed multi-pathologist consensus protocol that substantially improves inter-rater agreement over conventional H&E-only annotation. This yields a molecularly grounded reference against which we compare Atlas H&E-TME and pathologists working from H&E alone. For breadth, we benchmark Atlas H&E-TME on over 200,000 high-confidence H&E-only pathologist annotations across 1,500+ cases spanning eight cancer types and their most common metastatic sites, with subtypes covering >90% of clinical cases per cancer type, drawn from 25+ sources and 8+ scanner models. Benchmarked against the IHC-informed consensus, Atlas H&E-TME matches or exceeds pathologist H&E-only performance and generalizes consistently and robustly across this broad morphological and technical scope. In doing so, Atlas H&E-TME turns the H&E slide -- the most ubiquitous data in pathology -- into a scalable, quantitative window into the tumor and its microenvironment, laying a foundation for the next generation of tissue-based biomarkers in translational and clinical research.
Chinese Translation
苏木精-伊红（H&E）染色是组织病理学的基石，但对H&E全切片图像（WSIs）进行可扩展的定量分析仍然是计算病理学中的一个核心挑战。我们提出了Atlas H&E-TME，这是一个基于人工智能的系统，建立在Atlas病理基础模型系列之上，能够预测多种癌症类型的组织质量、组织区域和细胞类型标签，每张切片提供超过4500个细胞级分辨率的定量读数。验证此类系统的一个关键挑战是克服H&E唯一的真实标签固有的形态学模糊性，以及依赖于免疫组织化学（IHC）等多种模式的更为信息丰富的参考的有限可扩展性。我们通过一个双重验证框架来解决这一问题，该框架结合了生物学基础的深度与技术和形态学的广度。在深度方面，我们提出了一种基于IHC的信息的多病理学家共识协议，显著提高了与传统H&E唯一注释的评分者间一致性。这为我们提供了一个分子基础的参考，以此对比Atlas H&E-TME和仅使用H&E的病理学家。在广度方面，我们在超过200,000个高置信度的H&E唯一病理学家注释上对Atlas H&E-TME进行了基准测试，涵盖了1500多个病例，涉及八种癌症类型及其最常见的转移部位，亚型覆盖每种癌症类型超过90%的临床病例，数据来源于25个以上的来源和8个以上的扫描仪模型。与基于IHC的信息共识进行基准测试，Atlas H&E-TME的表现与病理学家仅使用H&E的表现相匹配或超越，并在这一广泛的形态学和技术范围内保持一致和稳健的泛化能力。通过这样做，Atlas H&E-TME将H&E切片——病理学中最普遍的数据——转变为一个可扩展的、定量化的窗口，深入了解肿瘤及其微环境，为转化和临床研究中的下一代基于组织的生物标志物奠定了基础。
cs.CV / 105 / 2606.12371
A Turbo-Inference Strategy for Object Detection and Instance Segmentation
一种用于目标检测和实例分割的涡轮推理策略
Abstract
Object detection and instance segmentation tasks are closely related. Existing top-down instance segmentation methods usually follow a detect-then-segment paradigm, where an initial detector is used to recognize and localize objects with bounding boxes, followed by the segmentation of an instance mask within each bounding box. In such methods, the detection accuracy directly influences the subsequent segmentation performance. However, previous research has seldom explored the impact of the instance segmentation task on object detection. In this paper, we present a turbo-inference strategy for the top-down methods that leverages the complementary information between detection and segmentation tasks iteratively. Specifically we design two modules: turbo-detection head and turbo-segmentation head, which facilitate communication between the tasks. The two modules form a closed loop that interlaces the detection and segmentation results without retraining the model. Comprehensive experiments on the COCO, iFLYTEK, and Cityscapes datasets demonstrate that our method substantially enhances both detection and segmentation accuracies with a certain increase in computational cost. The proposed method represents a tradeoff between prediction accuracy and inference speed. Codes are available at https://github.com/zhaozhen2333/Turbo-Learning.git.
Chinese Translation
目标检测和实例分割任务密切相关。现有的自上而下的实例分割方法通常遵循检测-再分割的范式，其中使用初始检测器识别和定位带有边界框的目标，然后在每个边界框内进行实例掩膜的分割。在此类方法中，检测精度直接影响后续的分割性能。然而，以往的研究很少探讨实例分割任务对目标检测的影响。本文提出了一种适用于自上而下方法的涡轮推理策略，该策略迭代利用检测和分割任务之间的互补信息。具体而言，我们设计了两个模块：涡轮检测头和涡轮分割头，以促进任务之间的交流。这两个模块形成一个闭环，交织检测和分割结果，而无需重新训练模型。在COCO、iFLYTEK和Cityscapes数据集上的全面实验表明，我们的方法在一定计算成本增加的情况下，显著提高了检测和分割的准确性。所提出的方法在预测精度和推理速度之间实现了权衡。代码可在 https://github.com/zhaozhen2333/Turbo-Learning.git 获取。
cs.CV / 106 / 2606.12378
Illumination-Robust Camera-Based Heart-Rate Estimation for Physiological Sensing in Robots
基于照明鲁棒性的相机心率估计用于机器人生理感知
Abstract
Physiological awareness is important for service, social, and assistive robots that interact with humans in everyday environments. Remote photoplethysmography (rPPG) enables non-contact heart-rate (HR) estimation from an RGB camera, making it a promising sensing modality for robot-mounted vision systems. However, illumination variation remains a major barrier to robust deployment. This paper presents an end-to-end spatial-temporal transformer framework for remote HR estimation on a new dataset with varied illumination. Our estimator integrates PRNet-based 3D face alignment, clip-level illumination augmentation, the Residual Temporal Standardization Module, and controlled hybrid temporal-frequency supervision. The training objective combines a Soft-Shifted Pearson waveform loss with a spectral Kullback-Leibler divergence loss, where a tuned weight ($\mathbf{\beta}$) controls the contribution of frequency-domain heart-rate guidance. Experiments on a static all-level mix protocol covering three illumination levels show that $\mathbf{\beta}=5$ provides the strongest result among the tested beta settings, achieving a best-run HR mean absolute error (MAE) of 0.79 bpm and an HR correlation of 0.982. Compared with the PhysFormer baseline evaluated on our dataset, our estimator reduces HR MAE by 93.6 %, while increasing HR correlation from 0.088 to 0.982, making it usable when illumination varies.
Chinese Translation
生理意识对于在日常环境中与人类互动的服务型、社交型和辅助型机器人至关重要。远程光电容积描记法（rPPG）能够通过RGB相机实现非接触式心率（HR）估计，使其成为机器人视觉系统中一种有前景的传感方式。然而，照明变化仍然是实现鲁棒部署的主要障碍。本文提出了一种端到端的时空变换器框架，用于在具有不同照明条件的新数据集上进行远程心率估计。我们的估计器集成了基于PRNet的三维面部对齐、剪辑级照明增强、残差时域标准化模块和受控的混合时频监督。训练目标结合了软偏移的Pearson波形损失和谱Kullback-Leibler散度损失，其中调节权重（$eta$）控制频域心率指导的贡献。在涵盖三种照明水平的静态全级混合协议上的实验表明，$eta=5$在测试的beta设置中提供了最强的结果，达到了最佳运行心率平均绝对误差（MAE）为0.79 bpm，心率相关性为0.982。与在我们数据集上评估的PhysFormer基线相比，我们的估计器将心率MAE降低了93.6%，同时将心率相关性从0.088提高到0.982，使其在照明变化时仍然可用。
cs.CV / 107 / 2606.12396
VLGA: Vision-Language-Geometry-Action Models for Autonomous Driving
VLGA：用于自主驾驶的视觉-语言-几何-动作模型
Abstract
Vision-language-action (VLA) models can describe scenes and reason about them in language, yet still struggle to ground their actions in the dense 3D world around them. Existing approaches either inject features from a frozen 3D foundation model without an objective that ensures the policy uses them, or constrain geometry with sparse box and map losses that provide no dense spatial signal. We introduce VLGA, the first vision-language-action model supervised to reconstruct the dense 3D world it drives through. VLGA introduces geometry as a fourth modality alongside vision, language, and action through a dedicated expert supervised by a per-pixel pointmap regression loss against LiDAR. Extensive experiments conducted on challenging nuScenes and Bench2Drive datasets for open-loop and closed-loop evaluations, respectively, show the superiority of VLGA over counterpart VLA methods. In particular, on open-loop nuScenes, VLGA sets a new state of the art among VLA methods without ego status, with the lowest L2 (0.50\,m average) and 3-second collision rate (0.18\%). On closed-loop Bench2Drive, VLGA attains the state-of-the-art driving score of 79.08, +0.71 over the strongest prior VLA, at comparable efficiency and comfort.
Chinese Translation
视觉-语言-动作（VLA）模型能够用语言描述场景并进行推理，但仍然难以将其动作与周围密集的3D世界相结合。现有的方法要么从一个冻结的3D基础模型中注入特征，但没有确保策略使用这些特征的目标，要么通过稀疏的框和地图损失来限制几何，这些损失没有提供密集的空间信号。我们提出了VLGA，这是第一个被监督以重建其行驶的密集3D世界的视觉-语言-动作模型。VLGA通过一个专门的专家引入几何作为与视觉、语言和动作并列的第四种模态，该专家通过与LiDAR的逐像素点图回归损失进行监督。在具有挑战性的nuScenes和Bench2Drive数据集上进行的广泛实验，分别用于开放环和闭环评估，显示了VLGA相较于对比的VLA方法的优越性。特别是在开放环的nuScenes上，VLGA在没有自我状态的情况下设定了VLA方法的新最优，具有最低的L2（0.50米平均）和3秒碰撞率（0.18%）。在闭环的Bench2Drive上，VLGA获得了79.08的最优驾驶评分，比最强的先前VLA提高了0.71，同时保持了相似的效率和舒适度。
cs.CV / 108 / 2606.12407
How Seemingly Inconsequential Design Choices Dictate Performance of LLMs in Pathology
看似微不足道的设计选择如何决定病理学中大语言模型的性能
Abstract
General-purpose large language models (LLMs) are routinely used as baselines when evaluating specialized pathology models on whole-slide images (WSIs). Because WSIs exceed contemporary model context limits, LLM baselines routinely use small, high-magnification patches processed independently via majority voting, without systematic evaluation of seemingly inconsequential design choices such as patch size, patch count, and magnification. Generalist LLMs have consistently underperformed specialized systems, reinforcing the perception that domain-specific training or architectural adaptation is necessary for pathology tasks involving WSIs. Here, we conduct a systematic factorial analysis of four input design factors: inference mode, patch size, magnification, and patch count. We demonstrate that prior studies have overstated the gap between specialized models and general-purpose LLMs by choosing non-optimized input configurations. On the MultiPathQA benchmark, switching to a single balanced configuration (large patches at lower magnification, processed jointly) raises GPT-5 from 15.1% to 39.5% on cancer-type classification (TCGA) and from 38.1% to 62.9% on organ classification (GTEx). Per-task optimization yields further gains up to 43.9% (TCGA) and 71.6% (GTEx). The same configuration generalizes to two other models and to a fully held-out CPTAC cohort, where it improves Gemini 3 Flash by 23.4 percentage points without any task-specific tuning.
Chinese Translation
通用的大语言模型（LLMs）在评估专门的病理模型时，通常作为基准使用，尤其是在全切片图像（WSIs）上。由于WSIs超出了当代模型的上下文限制，LLM基准通常使用小的、高倍放大的独立处理补丁，通过多数投票进行处理，而没有系统地评估诸如补丁大小、补丁数量和放大倍数等看似微不足道的设计选择。通用LLMs的表现始终低于专门系统，这进一步强化了在涉及WSIs的病理任务中，需要进行领域特定训练或架构调整的观念。在此，我们对四个输入设计因素进行了系统的因子分析：推理模式、补丁大小、放大倍数和补丁数量。我们证明，先前的研究通过选择未优化的输入配置，夸大了专门模型与通用LLMs之间的差距。在MultiPathQA基准上，切换到单一平衡配置（大补丁在较低放大倍数下共同处理）使GPT-5在癌症类型分类（TCGA）中的准确率从15.1%提高到39.5%，在器官分类（GTEx）中的准确率从38.1%提高到62.9%。每个任务的优化进一步提升了准确率，分别达到43.9%（TCGA）和71.6%（GTEx）。相同的配置在另外两个模型和一个完全独立的CPTAC队列中也具有良好的泛化能力，使Gemini 3 Flash的表现提高了23.4个百分点，而没有进行任何特定任务的调优。
cs.CV / 109 / 2606.12412
Reroute, Don't Remove: Recoverable Visual Token Routing for Vision-Language Models
重定向，而非删除：面向视觉语言模型的可恢复视觉标记路由
Abstract
Vision-language models (VLMs) project images into hundreds to thousands of visual tokens, making decoder inference expensive in both attention computation and KV-cache memory. Existing visual-token reduction methods largely follow a rank-and-remove paradigm: they score visual tokens, keep a compact subset, and permanently discard the rest. We show that this irreversible action is fragile because visual-token importance changes across decoder depth; tokens ranked low at one stage may become relevant in later layers, especially for grounding-sensitive queries. We propose Reroute, a training-free plug-in that replaces removal with recoverable routing. At each routing stage, selected vision tokens pass through decoder blocks, while deferred tokens bypass the stage and re-enter the candidate pool at the next routing decision. Reroute reuses existing attention-score ranking rules and stage-wise schedules, preserving the theoretical TFLOPs and KV-cache budget class of the pruning method it augments. Across FastV, PDrop, and N\"uwa variants on LLaVA-1.5 and Qwen backbones, reroute improves grounding under aggressive token reduction while maintaining general VQA performance. These results suggest that VLM token reduction should not be viewed only as irreversible pruning, but also as recoverable routing. The code can be found here: https://github.com/elmma/mllm-reroute/
Chinese Translation
视觉语言模型（VLMs）将图像投影为数百到数千个视觉标记，这使得解码器推理在注意力计算和KV缓存内存方面都变得昂贵。现有的视觉标记减少方法主要遵循排名和删除的范式：它们对视觉标记进行评分，保留一个紧凑的子集，并永久丢弃其余标记。我们表明，这种不可逆的操作是脆弱的，因为视觉标记的重要性在解码器深度上会发生变化；在某一阶段排名较低的标记在后续层中可能变得相关，尤其是对于对定位敏感的查询。我们提出了Reroute，一个无训练的插件，它用可恢复路由替代了删除。在每个路由阶段，选定的视觉标记通过解码器块，而延迟的标记则绕过该阶段，并在下一个路由决策中重新进入候选池。Reroute重用了现有的注意力评分排名规则和阶段性调度，保持了其增强的修剪方法的理论TFLOPs和KV缓存预算类别。在LLaVA-1.5和Qwen骨干网络上的FastV、PDrop和N"uwa变体中，Reroute在激进的标记减少下改善了定位性能，同时保持了整体的视觉问答（VQA）性能。这些结果表明，VLM的标记减少不应仅被视为不可逆的修剪，而也应视为可恢复的路由。代码可在此找到：https://github.com/elmma/mllm-reroute/
cs.AI / 1 / 2606.11207
From Explicit Elements to Implicit Intent: A Predefined Library for Auditable Behavioral Inference
从显式元素到隐式意图：一个可审计行为推断的预定义库
Abstract
We present SemantiClean, a modular framework for extracting structured semantic signals from e-commerce session data and driving pluggable inference targets including purchase intent, customer segmentation, and product affinity through a shared element library. Unlike conventional end-to-end predictors that optimise solely for accuracy, SemantiClean prioritises auditability, structural governance, and sigma=0 reproducibility, explicitly trading marginal predictive gains for element-level transparency and defensible decision trails. Built upon the Online Shoppers Purchasing Intention (OSPI) dataset, the framework organises twenty-four behavioural elements into a four-layer architecture (Functional, Interaction, Systemic, Contextual) and enforces signal quality through three anti-inflation mechanisms: RedundancyGroup contribution caps, TieredPenaltyCalculator bias penalties, and AdaptiveConstraintMode cold-start protection.This report introduces the LLM-Integrated Semantic Inference Engine, a fully implemented two-phase LLM-driven inference architecture that leverages complete element metadata at inference time. All quantitative results reported herein are produced by this engine. Deterministic engine outputs remain fully reproducible (sigma=0); LLM-dependent results (E8, E10) are subject to controlled output variability under fixed provider/model/temperature settings. The gender inference target remains non-functional in the current implementation and is excluded from all quantitative results.
Chinese Translation
我们提出了SemantiClean，一个模块化框架，用于从电子商务会话数据中提取结构化语义信号，并通过共享元素库驱动可插拔的推断目标，包括购买意图、客户细分和产品亲和力。与传统的端到端预测器仅优化准确性不同，SemantiClean优先考虑可审计性、结构治理和sigma=0的可重复性，明确地在边际预测增益与元素级透明度和可辩护决策轨迹之间进行权衡。该框架基于在线购物者购买意图（Online Shoppers Purchasing Intention, OSPI）数据集，将二十四个行为元素组织成四层架构（功能层、交互层、系统层、上下文层），并通过三种反通胀机制来确保信号质量：冗余组贡献上限、分层惩罚计算器偏差惩罚和自适应约束模式冷启动保护。本报告介绍了LLM集成语义推断引擎，这是一个完全实现的两阶段LLM驱动推断架构，在推断时利用完整的元素元数据。此处报告的所有定量结果均由该引擎生成。确定性引擎输出保持完全可重复（sigma=0）；依赖于LLM的结果（E8, E10）在固定的提供者/模型/温度设置下受到控制的输出变异性影响。当前实现中的性别推断目标仍然不可用，已从所有定量结果中排除。
cs.AI / 2 / 2606.11245
Position: Hippocampal Explicit Memory Is the Cornerstone for AGI
立场：海马体显性记忆是人工通用智能的基石
Abstract
Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks, raising expectations for Artificial General Intelligence (AGI). This position paper argues that integrating explicit memory is the cornerstone for advancing LLMs toward AGI. The key reason is that the underlying learning mechanism of LLMs is highly analogous to human implicit memory. However, higher-order cognitive functions necessary for AGI, such as long-term strategic planning, metacognition, and symbolic reasoning, heavily rely on hippocampal explicit memory and cannot arise solely from implicit statistical learning. Drawing on findings from neuroscience, I advance this perspective and complement it with computational requirements for artificial explicit memory systems, hoping to foster further research and lay the groundwork for explicit memory integration.
Chinese Translation
大型语言模型（LLMs）在各种任务中展现了显著的能力，这提升了对人工通用智能（AGI）的期望。本文立场论文认为，整合显性记忆是推动LLMs向AGI发展的基石。关键原因在于，LLMs的基本学习机制与人类的隐性记忆高度相似。然而，AGI所需的更高阶认知功能，如长期战略规划、元认知和符号推理，严重依赖于海马体显性记忆，无法仅仅依靠隐性统计学习而产生。基于神经科学的研究结果，我提出了这一观点，并补充了人工显性记忆系统的计算需求，希望能促进进一步的研究，并为显性记忆的整合奠定基础。
cs.AI / 3 / 2606.11337
Can AI Agents Synthesize Scientific Conclusions?
人工智能代理能否合成科学结论？
Abstract
Scientific AI agents increasingly retrieve evidence, reason across sources, and synthesize conclusions used in consequential decisions. Yet, their ability to do so in high-stakes domains such as health remains unclear. We introduce SciConBench, a large-scale live benchmark of 9.11K questions and expert-written conclusions from systematic reviews to evaluate open-domain scientific conclusion synthesis. The benchmark draws on an expert-validated automated evaluation pipeline that decomposes conclusions into atomic facts and measures correctness and comprehensiveness via factual precision and recall. To mitigate data leakage, we further introduce SciConHarness, a clean-room evaluation harness that equips agents with controlled web interaction to ensure valid measurement. Evaluating 8 frontier models and deep research agents, we find that factual quality remains low: under clean-room settings, the best agent achieves only a factual F1 of 0.337. Our clean-room setting consistently reduces performance relative to unconstrained evaluation, suggesting that leakage inflates estimates of models' true synthesis capabilities. Finally, we audit consumer-facing agents (e.g., Google AI Overview, OpenEvidence) and find they frequently generate incomplete and sometimes contradictory conclusions, even when the ground-truth answer is available. Overall, our results show that reliable synthesis of scientific conclusions remains an open challenge, and that clean-room evaluation is essential for assessing open-domain AI agents.
Chinese Translation
科学人工智能代理越来越多地检索证据、跨来源推理，并合成用于重要决策的结论。然而，它们在高风险领域（如健康）中执行这一任务的能力仍不清楚。我们介绍了SciConBench，这是一个大规模的实时基准，包含9.11K个问题和专家撰写的系统评价结论，用于评估开放领域的科学结论合成。该基准依赖于经过专家验证的自动评估流程，将结论分解为基本事实，并通过事实的准确性和召回率来衡量正确性和全面性。为了减轻数据泄漏的影响，我们进一步引入了SciConHarness，这是一个清洁室评估工具，赋予代理受控的网络交互，以确保有效测量。对8个前沿模型和深度研究代理的评估表明，事实质量仍然较低：在清洁室设置下，最佳代理的事实F1仅为0.337。我们的清洁室设置相较于无约束评估始终降低性能，表明泄漏会夸大模型真实合成能力的估计。最后，我们审计了面向消费者的代理（例如，Google AI Overview、OpenEvidence），发现它们经常生成不完整且有时矛盾的结论，即使在真实答案可用的情况下。总体而言，我们的结果表明，可靠的科学结论合成仍然是一个未解的挑战，而清洁室评估对于评估开放领域的人工智能代理至关重要。
cs.AI / 4 / 2606.11349
Knowing When to Ask: Self-Gated Clarification for Hierarchical Language Agents
知道何时提问：层次语言代理的自我门控澄清
Abstract
In hierarchical reasoning, failures often originate at intermediate decision points where the agent commits to a wrong branch without recognizing that it lacks critical information. Rather than treating clarification as an external uncertainty trigger, we propose ACTION-RATING, a formulation that places it inside the agent's action space on a shared ordinal scale with navigation, so that asking competes directly with acting at every decision point and help-seeking becomes observable at intermediate states. Two structurally distinct information-seeking modes emerge from the agent's own ratings: mandatory (no viable branch) and opportunistic (residual uncertainty despite a leading candidate). On Harmonized Tariff Schedule classification (30,000-node taxonomy, three benchmarks, 9~LLMs across 4 families), we observe a regime shift from mandatory to opportunistic clarification, with Information-Seeking Effectiveness (ISE), a local diagnostic defined as the fraction of help interactions followed by a correct next navigation step (not a final-task metric), rising from 50% to 74%. Three diagnostic contrasts fail to reproduce this structure. A separability test shows that the information-seeking pattern (mode split, ISE ranking) persists when answer quality is degraded (-18.8% accuracy), supporting an empirical separation between where an agent seeks help and the quality of the help it receives. Under the controlled answer channel, accuracy gains reach +16.2% at 10-digit; we read this as an upper bound on what better localization could unlock, not a deployment estimate.
Chinese Translation
在层次推理中，失败通常发生在中间决策点，代理在未意识到缺乏关键信息的情况下错误地选择了一个分支。我们提出了一种名为 ACTION-RATING 的方法，将澄清视为代理行动空间内的一个组成部分，与导航共享一个序数尺度，使得在每个决策点提问与行动直接竞争，并使寻求帮助在中间状态中变得可观察。代理自身的评分产生了两种结构上不同的信息寻求模式：强制性（没有可行的分支）和机会性（尽管有一个领先候选者，但仍存在剩余的不确定性）。在协调关税表分类（30,000节点分类法，三个基准，四个家族中的9个大型语言模型）中，我们观察到从强制性澄清到机会性澄清的转变，信息寻求有效性（ISE）作为一个局部诊断指标，定义为帮助交互后紧接着的正确导航步骤的比例（不是最终任务指标），从50%上升到74%。三个诊断对比未能重现这一结构。可分离性测试表明，当答案质量下降（准确率下降18.8%）时，信息寻求模式（模式分裂，ISE 排名）依然存在，支持代理寻求帮助的地方与其获得帮助的质量之间的经验分离。在控制答案通道下，准确率提升达到+16.2%（在10位数字时）；我们将其视为更好定位可能解锁的上限，而非部署估计。
cs.AI / 5 / 2606.11379
Automated Mediator for Human Negotiation: Pre-Mediation via a Structured LLM Pipeline
人类谈判的自动化调解者：通过结构化的LLM管道进行预调解
Abstract
Pre-mediation, the preparatory phase preceding direct human negotiation, plays a critical role in achieving mutually beneficial agreements, yet is often omitted due to cost, time, and limited access to trained mediators. We introduce an automated mediator for human negotiation, implemented as a structured pipeline of LLM modules, that supports pre-mediation in integrative negotiation settings. The pipeline decomposes preparation into specialized modules for dialogue, preference prediction, response-level critique, and structured summarization, separating inference, generation, and evaluation to address limitations of monolithic single-prompt approaches. We use the term "agent" for each module following common LLM-systems terminology, but the components are not autonomous and do not interact peer-to-peer; outputs are passed forward in a fixed sequence. We evaluate the system in two controlled human-subject experiments comparing AI-based pre-mediation with professional human mediators in a multi-issue negotiation scenario. On short-term self-reported measures, the automated mediator achieves preparation outcomes broadly comparable to human mediators, including trust in the mediator and confidence in reaching mutually beneficial agreements, while achieving substantially lower error on the preference-inference task under our scenario and prompts (36% lower RMSE). A second study shows that targeted prompt refinements reduce excessive affirmation patterns from 36.6% to 16.8%, matching human mediator baselines. Our findings suggest that structured LLM pipelines can provide scalable, low-effort pre-mediation support broadly comparable to human mediators on short-term self-reported preparation outcomes. The pipeline's single-party design mirrors how human mediators run pre-mediation today and enables parallel deployment across all parties to a dispute, supporting scalability.
Chinese Translation
预调解是直接人类谈判之前的准备阶段，在达成互利协议中发挥着关键作用，但由于成本、时间和受训调解员的有限获取，往往被忽视。我们提出了一种用于人类谈判的自动化调解者，作为结构化的LLM模块管道实现，支持在综合谈判环境中的预调解。该管道将准备过程分解为对话、偏好预测、响应级批评和结构化总结等专门模块，分离推理、生成和评估，以解决单一提示方法的局限性。我们使用“agent”一词来指代每个模块，遵循常见的LLM系统术语，但这些组件并不是自主的，也不进行点对点交互；输出以固定顺序向前传递。我们在两个受控的人类受试者实验中评估该系统，比较基于AI的预调解与专业人类调解员在多议题谈判场景中的表现。在短期自我报告的测量中，自动化调解者在准备结果上与人类调解员大致相当，包括对调解员的信任和达成互利协议的信心，同时在我们的场景和提示下，偏好推断任务的错误率显著降低（均方根误差降低36%）。第二项研究表明，针对性提示的改进将过度肯定模式从36.6%降低至16.8%，与人类调解员的基准相匹配。我们的研究结果表明，结构化的LLM管道可以提供可扩展、低努力的预调解支持，其短期自我报告的准备结果与人类调解员大致相当。该管道的单方设计反映了人类调解员今天如何进行预调解，并支持在所有争议方之间的并行部署，从而促进可扩展性。
cs.AI / 6 / 2606.11440
INFRAMIND: Infrastructure-Aware Multi-Agent Orchestration
INFRAMIND：基础设施感知的多智能体调度
Abstract
Existing multi-agent LLM orchestration methods, ranging from brute-force ensembles to learned routers, select models and topologies based on task and model features. However, these methods do not consider the runtime state of the serving infrastructure. On shared GPU clusters under concurrent load, this infrastructure blindness causes systematic resource underutilization: preferred models accumulate deep request queues while equally capable alternatives sit idle. In multi-agent pipelines, where each query triggers multiple sequential model calls, these delays then compound across every downstream step. Closing this gap is challenging because the relevant infrastructure signals (queue depths, KV-cache pressure, latencies) are dynamic and noisy, and they must drive three different decisions: planning, per-step routing, and scheduling. We introduce INFRAMIND, a framework that makes the entire multi-agent stack infrastructure-aware. An infra-aware planner conditions topology and role selection on real-time system load and remaining budget, biasing toward simpler graphs under congestion and richer ones at low load. An infra-aware executor then observes per-model queue depths, cache utilization, and response latencies at each agent step to decide which model to call and how deeply to reason; a budget-aware scheduler further reorders each model's queue so that urgent requests are served first. Cast as a hierarchical constrained MDP and solved end-to-end via reinforcement learning, the system learns to balance quality against latency automatically. Across five benchmarks, INFRAMIND delivers up to +7.6 pp accuracy over the prior baseline at low load with up to 7x lower latency, and sustains up to 99.9% SLO compliance under high load where every baseline drops below 50%.
Chinese Translation
现有的多智能体大规模语言模型（LLM）调度方法，从暴力集成到学习路由器，都是基于任务和模型特征来选择模型和拓扑。然而，这些方法并未考虑服务基础设施的运行时状态。在共享GPU集群的并发负载下，这种基础设施盲目性导致系统性的资源未充分利用：优选模型积累了深厚的请求队列，而同样具备能力的替代模型则处于闲置状态。在多智能体管道中，每个查询触发多个顺序模型调用，这些延迟在每个下游步骤中会进一步累积。弥补这一差距具有挑战性，因为相关的基础设施信号（队列深度、键值缓存压力、延迟）是动态且嘈杂的，并且它们必须驱动三个不同的决策：规划、每步路由和调度。我们提出了INFRAMIND，一个使整个多智能体堆栈具备基础设施感知的框架。一个基础设施感知的规划器根据实时系统负载和剩余预算来调整拓扑和角色选择，在拥堵时偏向于简单图形，而在低负载时则偏向于更复杂的图形。基础设施感知的执行器在每个智能体步骤中观察每个模型的队列深度、缓存利用率和响应延迟，以决定调用哪个模型以及推理的深度；一个预算感知的调度器进一步重新排序每个模型的队列，以确保紧急请求优先得到服务。该系统被视为一个分层约束的马尔可夫决策过程（MDP），并通过强化学习进行端到端求解，自动学习在质量与延迟之间进行平衡。在五个基准测试中，INFRAMIND在低负载下提供了比之前基线高出7.6个百分点的准确率，同时延迟降低了最多7倍，并在高负载下维持了高达99.9%的服务水平协议（SLO）合规性，而每个基线则降至50%以下。
cs.AI / 7 / 2606.11445
Forecasting Future Behavior as a Learning Task
将未来行为预测视为学习任务
Abstract
Trust in an AI system is often anchored by explanations of how it works, which one then uses to forecast its behavior on new inputs. For large reasoning models (LRMs), this conventional route is particularly difficult to follow: explanation methods for single token generations do not naturally generalize to long trajectories, and the trajectories themselves are often not faithful when read as natural language. We propose an alternative that bypasses the explanation step: treat behavior forecasting as a learnable task and train Behavior Forecasters that operates on a single reasoning trajectory to make the same forecasts one would typically seek from an explanation. The forecaster's training data is obtained by querying the LRM with no human annotation, and its inference is done in a single forward pass. We instantiate this approach on two tasks: how likely the LRM is to repeat its answer on re-runs, and how removing parts of the input changes its answer. We evaluate this approach on both tasks across three diverse reasoning datasets and find that trained Behavior Forecasters are more accurate than GPT-5.4 and Claude Opus-4.6 reading the same trajectories as naive readers, at a small fraction of their inference cost. We find that fine-tuning the backbone end-to-end and initializing it from the target LRM are each necessary for strong performance. These results show that the reasoning trajectory carries information about the LRM's future behavior that goes beyond what naive reading conveys.
Chinese Translation
对人工智能系统的信任通常依赖于对其工作原理的解释，进而用于预测其在新输入下的行为。对于大型推理模型（LRMs）而言，这一传统路径特别难以遵循：单个标记生成的解释方法并不能自然地推广到长时间轨迹，而这些轨迹在作为自然语言阅读时往往缺乏忠实性。我们提出了一种绕过解释步骤的替代方案：将行为预测视为一个可学习的任务，并训练行为预测器（Behavior Forecasters），该预测器在单一推理轨迹上操作，以做出通常希望通过解释获得的预测。预测器的训练数据通过对LRM进行查询获得，无需人工标注，其推理在单次前向传递中完成。我们在两个任务上实例化了这一方法：LRM在重复运行时重复其答案的可能性，以及去除输入部分如何改变其答案。我们在三个不同的推理数据集上对这两项任务评估了该方法，发现经过训练的行为预测器在阅读相同轨迹时，比GPT-5.4和Claude Opus-4.6作为天真读者的准确性更高，且推理成本仅为它们的一小部分。我们发现，端到端微调骨干网络并从目标LRM初始化是实现强大性能的必要条件。这些结果表明，推理轨迹携带有关LRM未来行为的信息，这超出了天真阅读所传达的信息。
cs.AI / 8 / 2606.11522
Search Discipline for Long-Horizon Research Agents
长时间跨度研究代理的搜索规范
Abstract
Autoresearch agents now propose, evaluate, and select scientific candidates against a metric, and that metric is usually an aggregate reduced over a heterogeneous space of regions, slices, or cohorts. We show that when scientific validity lives in that disaggregated structure, the aggregate can rank the wrong candidate first. The headline number improves while the structure underneath inverts, so a decision made on the number accepts a candidate that quietly breaks the model. The failure is not domain-specific. It appears wherever a candidate's validity is multi-dimensional but its verifier is a single reduction. We demonstrate the inversion on a fire-model task in the Ecosystem Demography model. The highest-scoring candidate and a slightly lower one are within noise of each other on global score, yet the top-scoring one collapses the protected boreal regions while the other preserves them. What separates them is the per-region behavior, not the headline number. This decision should not be left to the agent that produced the candidates. The agent optimizing the score is the last party likely to catch the score being wrong, and a prompt has no remaining turn once the agent has stopped. We move the decision to an external control loop that audits each candidate on its disaggregated behavior and acts after the agent has decided. It can demote a candidate the agent would have accepted, and it can reopen a run the agent had declared finished. Our contribution is the inversion finding itself, and a search-discipline protocol that decides on reviewable candidate-effect evidence instead of the score.
Chinese Translation
自我研究代理现在根据某一指标提出、评估和选择科学候选者，而该指标通常是对异质区域、切片或群体的聚合缩减。当科学有效性存在于这种非聚合结构中时，聚合可能会错误地将不合适的候选者排在首位。尽管表面数字有所改善，但其底层结构却发生了反转，因此基于该数字做出的决策接受了一个悄然破坏模型的候选者。这个失败并不是特定于某一领域。它出现在候选者的有效性是多维的，但其验证者却是单一的缩减的地方。我们在生态系统人口模型中的火灾模型任务上展示了这种反转。得分最高的候选者与得分稍低的候选者在全球得分上相互接近，但得分最高的候选者会导致受保护的北方森林区域崩溃，而另一候选者则能够保护这些区域。将它们区分开的是每个区域的行为，而不是表面数字。这个决策不应留给产生候选者的代理。优化得分的代理是最后一个可能发现得分错误的主体，并且一旦代理停止，提示就没有剩余的回旋余地。我们将决策移至一个外部控制循环，该循环对每个候选者的非聚合行为进行审计，并在代理做出决定后采取行动。它可以降低一个代理本会接受的候选者的级别，并可以重新开启一个代理已宣告完成的运行。我们的贡献在于发现了这种反转，并提出了一种搜索规范协议，该协议基于可审查的候选者效果证据而非得分进行决策。
cs.AI / 9 / 2606.11537
MoCA-Agent: A Market-of-Claims Code Agent for Financial and Numerical Reasoning
MoCA-Agent：一种用于金融和数值推理的市场声明代码代理
Abstract
Financial and tabular question answering requires more than fluent reasoning: answers must be grounded in the exact facts, formulas, units, signs, and scales that support them. A single misread cell or incorrect operation can silently produce a plausible but wrong result. We introduce \textsc{MOCA-Agent}, a market-of-claims code agent that replaces free-form multi-agent debate with claim-level verification. The system decomposes each question into typed atomic claims, asks specialist trader agents to buy or sell those claims, clears their orders into confidence-weighted accept/reject decisions, and synthesizes an executable Python program from market-supported evidence. A code-aware verifier then checks the program for execution, structural consistency, and common financial reasoning errors, with at most one market-aware repair round. Across ten public benchmarks spanning financial numerical reasoning, general tabular reasoning, ESG question answering, and multimodal chart reasoning, \textsc{MOCA-Agent} achieves strong performance using a fixed Qwen3.6-27B backbone, including $78.3\%$ on FinQA, $76.0\%$ on FinanceMath, $71.2\%$ on MultiHiertt, $86.9\%$ on ESGenius, and $85.6\%$ average on FinChart-Bench. These results show that aggregating evidence at the level of atomic claims, rather than whole answers, improves robustness in high-stakes numerical reasoning.\footnote{The code and data are available: https://github.com/UBC-NLP/MoCA-Agent.
Chinese Translation
金融和表格问答不仅需要流畅的推理能力：答案必须基于支持它们的确切事实、公式、单位、符号和尺度。单个错误读取的单元格或不正确的操作可能会默默地产生一个看似合理但错误的结果。我们介绍了 extsc{MOCA-Agent}，一种市场声明代码代理，它用声明级验证取代了自由形式的多代理辩论。该系统将每个问题分解为类型化的原子声明，要求专业交易代理买入或卖出这些声明，将其订单清算为基于信心的接受/拒绝决策，并从市场支持的证据中合成一个可执行的Python程序。一个代码感知的验证器随后检查程序的执行、结构一致性和常见金融推理错误，最多进行一次市场感知的修复轮次。在涵盖金融数值推理、一般表格推理、ESG问答和多模态图表推理的十个公共基准测试中， extsc{MOCA-Agent} 使用固定的 Qwen3.6-27B 主干实现了强劲的表现，包括在 FinQA 上达到 $78.3\%$，在 FinanceMath 上达到 $76.0\\%$，在 MultiHiertt 上达到 $71.2\\%$，在 ESGenius 上达到 $86.9\\%$，以及在 FinChart-Bench 上的平均 $85.6\\%$。这些结果表明，在原子声明级别聚合证据，而不是整体答案，能够提高高风险数值推理的稳健性。\footnote{代码和数据可用： https://github.com/UBC-NLP/MoCA-Agent.
cs.AI / 10 / 2606.11543
SkillJuror: Measuring How Agent Skill Organization Changes Runtime Behavior
SkillJuror：测量代理技能组织如何改变运行时行为
Abstract
Agent Skills augment large language model (LLM) agents with procedural knowledge at inference time, but current benchmarks rarely distinguish what a Skill says from how it is organized. We study this distinction through Progressive Disclosure, where a concise root file points agents to supporting resources on demand, and compare it with a normalized flat baseline. We present SkillJuror, a framework for evaluating Skill writing paradigms through semantically controlled variants, matched multi-trial evaluations, and trajectory evidence while holding task knowledge fixed. In an 82-task SkillsBench study, Progressive Disclosure changes runtime behavior before aggregate outcomes: distinct Skill resources touched per trajectory rise from 1.18 to 3.85, and effective uptake events rise from 1.33 to 3.92. It also yields 17 additional verifier-passing trials out of 410 matched trials (+4.1%) over the normalized flat baseline. The benefit is task-dependent. Progressive Disclosure helps when supporting resources guide implementation, checking, or repair, but is weaker when success hinges on exact output conventions, numerical thresholds, or long artifact-generation pipelines. These results show that Skill organization is not mere presentation: it can change how agents search and apply procedural knowledge, while outcome gains depend on whether the exposed resources are actionable for the task. Code is available at https://github.com/zhiyuchen-ai/skill-juror.
Chinese Translation
代理技能在推理时为大型语言模型（LLM）代理提供程序性知识，但当前基准很少区分技能的内容与其组织方式。我们通过渐进式披露（Progressive Disclosure）研究这一区别，其中一个简洁的根文件根据需要指向支持资源，并将其与标准化的平面基线进行比较。我们提出了SkillJuror，这是一个通过语义控制变体、匹配的多次试验评估和轨迹证据来评估技能写作范式的框架，同时保持任务知识不变。在一项包含82个任务的SkillsBench研究中，渐进式披露在汇总结果之前改变了运行时行为：每条轨迹接触的不同技能资源从1.18上升到3.85，有效获取事件从1.33上升到3.92。此外，在410个匹配试验中，额外获得了17个通过验证者的试验（+4.1%），相较于标准化的平面基线。该效益依赖于任务。当支持资源指导实现、检查或修复时，渐进式披露有助于任务，但当成功依赖于精确的输出约定、数值阈值或长时间的工件生成流程时，其效果较弱。这些结果表明，技能组织不仅仅是展示：它可以改变代理搜索和应用程序性知识的方式，而结果的提升则依赖于暴露的资源是否对任务可操作。代码可在 https://github.com/zhiyuchen-ai/skill-juror 获取。
cs.AI / 11 / 2606.11559
HERO: Hindsight-Enhanced Reflection from Environment Observations for Agentic Self-Distillation
HERO：基于环境观察的后见增强反思用于自主自蒸馏
Abstract
Reinforcement learning typically improves multi-turn agent capabilities through the terminal outcome of the trajectories, which makes it difficult to determine credit assignments for each intermediate turns. Recent on-policy self-distillation methods offer a promising alternative by converting privileged feedback into dense token-level supervision through a self-teacher. Our study is motivated by the unexpected performance degradation observed when naively extending this paradigm to multi-turn settings, which we attribute to a lack of alignment between privileged feedback, such as successful trajectories or terminal outcomes, and the student's current decision context. We introduce HERO, a hindsight-enhanced self-distillation framework that uses next environment observations as locally aligned feedback. After each rollout, HERO reflects on the completed interaction to convert each observation into a compact turn-level diagnosis, that captures actionable feedback about the original action such as its necessity, validity or failure cause. On TauBench and WebShop, HERO improves task success and reduces unnecessary turns over environment-feedback-only self-distillation and GRPO. It is especially effective under limited training turn budgets, where successful rollouts are rare and GRPO provides weak reward-contrast signals.
Chinese Translation
强化学习通常通过轨迹的终端结果来提升多回合智能体的能力，这使得很难为每个中间回合确定信用分配。最近的在线自蒸馏方法通过自教师将特权反馈转化为密集的标记级监督，提供了一种有前景的替代方案。我们的研究受到一个动机，即在简单地将这一范式扩展到多回合设置时观察到的意外性能下降，我们将其归因于特权反馈（例如成功轨迹或终端结果）与学生当前决策上下文之间缺乏一致性。我们提出了HERO，一个后见增强的自蒸馏框架，它使用下一个环境观察作为局部对齐的反馈。在每次回合结束后，HERO反思已完成的交互，将每个观察转化为紧凑的回合级诊断，捕捉关于原始动作的可操作反馈，例如其必要性、有效性或失败原因。在TauBench和WebShop上，HERO提高了任务成功率，并减少了相较于仅使用环境反馈的自蒸馏和GRPO的不必要回合。它在训练回合预算有限的情况下尤其有效，此时成功的回合较为稀少，而GRPO提供的奖励对比信号较弱。
cs.AI / 12 / 2606.11634
Architecture-Aware Reinforcement Learning Makes Sliding-Window Attention Competitive in Math Reasoning
架构感知强化学习使滑动窗口注意力在数学推理中具备竞争力
Abstract
The rapid progress of reasoning and agentic large language models (LLMs) has increased the demand for long-context inference, but self-attention (SA) scales quadratically with context length. To address this, we study SWARR (Sliding-Window Attention with Reinforced Adaptation for Math Reasoning), a practical recipe for adapting SWA models to mathematical reasoning. SWARR has two stages: (1) efficient conversion from a pretrained SA model to SWA with supervised fine-tuning (SFT), which avoids pretraining a new base model, and (2) policy adaptation with reinforcement learning (RL). We find that SWA still underperforms SA after SFT, and we hypothesize that this gap is caused in part by a data-architecture mismatch: most SFT data are prepared for SA models and may contain long-range dependencies that are difficult for SWA to model. Because on-policy RL optimizes self-generated trajectories under the SWA constraint, it can adapt trajectories to better match SWA. Experiments on mathematical reasoning benchmarks show that this recipe substantially narrows the gap between SWA and SA, recovering much of the accuracy lost during SWA conversion while preserving the efficiency benefits of linear-complexity attention. Our central contribution is the empirical finding that RL changes the conclusion one would draw from conversion and SFT alone about SWA's viability for math reasoning.
Chinese Translation
推理和智能大型语言模型（LLMs）的快速进展增加了对长上下文推理的需求，但自注意力（SA）随着上下文长度呈二次增长。为了解决这个问题，我们研究了SWARR（滑动窗口注意力与数学推理的强化适应），这是一种将SWA模型适应于数学推理的实用方法。SWARR包含两个阶段：（1）通过监督微调（SFT）高效地将预训练的SA模型转换为SWA，这避免了重新预训练新的基础模型，以及（2）通过强化学习（RL）进行策略适应。我们发现，SWA在SFT后仍然表现不如SA，我们假设这种差距部分是由于数据与架构的不匹配：大多数SFT数据是为SA模型准备的，可能包含SWA难以建模的长程依赖。由于在策略优化中，RL在SWA约束下优化自生成的轨迹，因此它可以调整轨迹以更好地匹配SWA。在数学推理基准上的实验表明，这种方法显著缩小了SWA与SA之间的差距，恢复了在SWA转换过程中丢失的大部分准确性，同时保持了线性复杂度注意力的效率优势。我们的核心贡献是实证发现，RL改变了人们仅从转换和SFT得出的关于SWA在数学推理中可行性的结论。
cs.AI / 13 / 2606.11637
TouchThinker: Scaling Tactile Commonsense Reasoning to the Open World with Large-scale Data and Action-aware Representation
TouchThinker：利用大规模数据和动作感知表示将触觉常识推理扩展到开放世界
Abstract
Touch is a key modality for embodied agents to understand the physical world. Although recent work has incorporated tactile signals into language systems for tactile commonsense reasoning, scaling such systems to realistic open-world settings remains challenging due to two key bottlenecks: (1) current tactile reasoning datasets remain limited in format and scale, providing insufficient supervision for reasoning from tactile observations to physical commonsense and hindering the learning of transferable tactile commonsense; (2) Tactile signals are inherently redundant and action-specific, yet existing methods often overlook these properties, resulting in inefficient representations with limited semantic expressiveness. To address these limitations, we propose TouchThinker, a tactile-language framework that scales tactile commonsense reasoning to the open world from both data and representation perspectives. First, we construct TouchThinker-1M, a million-scale, multi-source tactile reasoning dataset covering \textbf{415} objects, \textbf{8} scenarios, and \textbf{7} sensor types, providing a solid data foundation for open-world generalization. We further introduce TouchThinker-Bench, an open-world benchmark with more realistic and diverse tasks. Then, we propose action-aware modeling mechanism to improve tactile representation efficiency and enable efficient reasoning. Experimental results demonstrate that TouchThinker achieves competitive performance against state-of-the-art models across multiple datasets. Our code and dataset will be made available at: https://github.com/lvkailin0118/TouchThinker.
Chinese Translation
触觉是具身智能体理解物理世界的关键模态。尽管近期的研究已将触觉信号纳入语言系统以进行触觉常识推理，但将此类系统扩展到现实的开放世界环境仍然面临两大瓶颈：（1）当前的触觉推理数据集在格式和规模上仍然有限，无法为从触觉观察到物理常识的推理提供足够的监督，阻碍了可迁移触觉常识的学习；（2）触觉信号本质上是冗余且特定于动作的，但现有方法往往忽视这些特性，导致表示效率低下且语义表达能力有限。为了解决这些局限性，我们提出了TouchThinker，一个从数据和表示角度将触觉常识推理扩展到开放世界的触觉-语言框架。首先，我们构建了TouchThinker-1M，一个百万规模的多源触觉推理数据集，涵盖了415个物体、8个场景和7种传感器类型，为开放世界泛化提供了坚实的数据基础。我们进一步推出了TouchThinker-Bench，一个具有更现实和多样化任务的开放世界基准。然后，我们提出了动作感知建模机制，以提高触觉表示的效率并实现高效推理。实验结果表明，TouchThinker在多个数据集上表现出与最先进模型的竞争性能。我们的代码和数据集将发布在：https://github.com/lvkailin0118/TouchThinker。
cs.AI / 14 / 2606.11662
TreeSeeker: Tree-Structured Trial, Error, and Return in Deep Search
TreeSeeker：深度搜索中的树结构试验、错误与返回
Abstract
Deep search requires agents to answer complex questions through multi-step web search, browsing, evidence comparison, and synthesis. A central challenge is deciding how to search when several directions look plausible but only some will later lead to reliable evidence. If an agent greedily follows the current best-looking direction, it may keep extending a weak continuation. If it explores without discipline, it may waste budget on disconnected trials. We propose TreeSeeker, an inference-time framework for controlled trial-and-error in deep search. TreeSeeker organizes search as branch-and-return search over tree-structured states, where each branch is a tentative direction for a sub-goal. At each round, TreeSearch reads all sub-goal trees, identifies active goals, and uses textual UCB signals of value, uncertainty, and risk to select among exploiting a promising branch, exploring an uncertain alternative, or pruning an unproductive continuation and returning to an earlier branch point. TreeMem supports this control loop by keeping evidence, uncertainty, conflicts, progress, and failure cues attached to the branches that produced them, so trial outcomes can guide later decisions. Experiments on XBench-DeepSearch, BrowseComp, and BrowseComp-ZH show that TreeSeeker consistently outperforms strong open-source baselines, suggesting that explicit branch-and-return control complements stronger reasoning and tool execution.
Chinese Translation
深度搜索要求智能体通过多步骤的网络搜索、浏览、证据比较和综合来回答复杂问题。一个核心挑战是如何在多个看似合理的方向中进行搜索，而只有部分方向最终会导致可靠的证据。如果智能体贪婪地跟随当前看起来最好的方向，它可能会不断延伸一个薄弱的延续。如果它无序探索，可能会浪费预算在无关的试验上。我们提出了TreeSeeker，一个用于深度搜索中受控试验与错误的推理时框架。TreeSeeker将搜索组织为在树结构状态上的分支与返回搜索，其中每个分支是一个子目标的暂定方向。在每一轮中，TreeSearch读取所有子目标树，识别活动目标，并利用文本UCB信号的价值、不确定性和风险在利用有前景的分支、探索不确定的替代方案或修剪无效的延续并返回到早期分支点之间进行选择。TreeMem通过将证据、不确定性、冲突、进展和失败线索附加到产生它们的分支上来支持这一控制循环，以便试验结果能够指导后续决策。在XBench-DeepSearch、BrowseComp和BrowseComp-ZH上的实验表明，TreeSeeker始终优于强大的开源基线，表明明确的分支与返回控制可以补充更强的推理和工具执行。
cs.AI / 15 / 2606.11675
Lung-R1: A Knowledge Graph-Guided LLM for Pulmonary Diagnostic Reasoning
Lung-R1：一种基于知识图谱的肺部诊断推理大语言模型
Abstract
Diagnosing pulmonary diseases requires integrating heterogeneous evidence amid phenotypic variability and cross-disease overlap. Although large language models (LLMs) have shown progress on pulmonary knowledge question answering (QA) and information-processing tasks, reliable pulmonary diagnosis requires patient-specific, relation-aware reasoning over electronic medical record (EMR) evidence rather than isolated knowledge recall. We define this gap between pulmonary knowledge and case-level diagnostic reasoning as the Pulmonary Knowledge-to-Diagnosis Gap. To address it, we introduce LungKG, the first structured pulmonary knowledge graph for diagnostic knowledge organization and record-grounded reasoning. LungKG contains 59,038 nodes and 164,308 edges across 15 entity types and 112 relation types, serving as both a reusable pulmonary knowledge resource and the foundation for LungKG-guided model adaptation. Built on LungKG, we propose Lung-R1, a LungKG-guided pulmonary LLM trained through KG-constrained reasoning-chain construction and KG-guided reinforcement learning. In a 20-system evaluation, Lung-R1-14B achieves state-of-the-art performance across Choice, Pulmonary-QA, and EMR Diagnosis, reaching an EMR Diagnosis score of 4.3583 and surpassing the strongest non-Lung-R1 baseline by 0.1476 points. These results demonstrate the value of LungKG-guided training for EMR-based pulmonary diagnosis.
Chinese Translation
诊断肺部疾病需要在表型变异和跨疾病重叠中整合异质证据。尽管大型语言模型（LLMs）在肺部知识问答（QA）和信息处理任务上取得了进展，但可靠的肺部诊断需要基于电子病历（EMR）证据进行患者特异性和关系感知的推理，而不是孤立的知识回忆。我们将肺部知识与案例级诊断推理之间的差距定义为肺部知识到诊断的差距（Pulmonary Knowledge-to-Diagnosis Gap）。为了解决这一问题，我们引入了LungKG，这是第一个用于诊断知识组织和基于记录的推理的结构化肺部知识图谱。LungKG包含59,038个节点和164,308条边，涵盖15种实体类型和112种关系类型，既作为可重用的肺部知识资源，又作为LungKG引导的模型适应的基础。在LungKG的基础上，我们提出了Lung-R1，这是一种通过知识图谱约束的推理链构建和知识图谱引导的强化学习训练的肺部大语言模型。在20个系统的评估中，Lung-R1-14B在Choice、Pulmonary-QA和EMR诊断方面达到了最先进的性能，EMR诊断得分为4.3583，超过了最强的非Lung-R1基线0.1476分。这些结果证明了基于LungKG引导的训练在基于EMR的肺部诊断中的价值。
cs.AI / 16 / 2606.11680
Organize then Retrieve: Hierarchical Memory Navigation for Efficient Agents
先组织再检索：高效智能体的层次记忆导航
Abstract
Large language model (LLM) agents struggle with long-horizon tasks due to their inherent statelessness, requiring all task-relevant information to be encoded in growing input contexts. The resulting degraded reasoning quality, increased inference cost, and higher latency necessitate efficient working memory mechanisms. However, existing approaches either rely on lossy compression or similarity-based retrieval, which often fail to capture temporal structure and causal dependencies required for multi-step agentic tasks. In this work, we present HORMA, a Hierarchical Organize-and-Retrieve Memory Agent that organizes experience into a file-system-like hierarchical structure, where summarized entities are linked to the corresponding raw trajectories, enabling efficient access without losing detailed information. HORMA decomposes working memory into two stages: structured memory construction and navigation-based retrieval. The construction module iteratively refines how experiences are structured by distinguishing between failures caused by missing information and those caused by misleading or overloaded context. The navigation module retrieves task-relevant context by traversing the hierarchy using a lightweight agent trained with reinforcement learning to select minimal yet sufficient context, thereby reducing latency along the critical execution path. Across ALFWorld, LoCoMo, and LongMemEval, HORMA improves task performance under constrained context budgets while requiring at most 22.17% of the baseline token usage in long conversation tasks. Compared to existing methods, it consistently achieves better efficiency-performance trade-offs and generalizes effectively to unseen tasks.
Chinese Translation
大型语言模型（LLM）智能体在长时间任务中表现不佳，原因在于其固有的无状态特性，要求所有与任务相关的信息都必须编码在不断增长的输入上下文中。由此导致的推理质量下降、推理成本增加和延迟加大，迫切需要高效的工作记忆机制。然而，现有方法要么依赖于有损压缩，要么基于相似性检索，这往往无法捕捉多步骤智能任务所需的时间结构和因果依赖关系。在本研究中，我们提出了HORMA（层次组织与检索记忆智能体），该智能体将经验组织成类似文件系统的层次结构，其中总结的实体与相应的原始轨迹相链接，从而实现高效访问而不丢失详细信息。HORMA将工作记忆分解为两个阶段：结构化记忆构建和基于导航的检索。构建模块通过区分因信息缺失导致的失败与因误导或过载上下文导致的失败，迭代地优化经验的结构。导航模块通过使用强化学习训练的轻量级智能体遍历层次结构来检索与任务相关的上下文，以选择最小但足够的上下文，从而减少关键执行路径上的延迟。在ALFWorld、LoCoMo和LongMemEval等任务中，HORMA在受限的上下文预算下提高了任务性能，同时在长对话任务中最多只需使用基线令牌的22.17%。与现有方法相比，它在效率与性能的权衡上始终表现更佳，并能有效推广到未见过的任务。
cs.AI / 17 / 2606.11724
Mind the Perspective: Let's Reason Recursively for Theory of Mind
关注视角：让我们为心智理论进行递归推理
Abstract
Theory of Mind (ToM) reasoning requires inferring agents' beliefs from partial and asymmetric observations, which remains an open challenge for LLMs. Existing prompting-based approaches improve ToM reasoning through observable-event filtering or temporal belief chains, without explicitly modeling nested beliefs. We introduce RecToM, an inference-time framework for ToM reasoning that models nested beliefs via recursive perspective construction. RecToM constructs each character perspective from the preceding character perspective along the character chain specified by the question, reducing higher-order belief questions to actual-world questions within the final constructed perspective. We further provide a KD45 analysis showing that RecToM's perspective construction induces a well-formed belief modality beyond simple event filtering. Experiments on ToM benchmarks, including Hi-ToM, Big-ToM, and FanToM, across multiple LLM backbones show that RecToM consistently outperforms recent advanced approaches, achieving state-of-the-art performance. Notably, RecToM reaches 100\% accuracy on Hi-ToM with GPT-5.4 and Qwen3.5, a benchmark requiring higher-order ToM reasoning.
Chinese Translation
心智理论（Theory of Mind, ToM）推理需要从部分和不对称的观察中推断代理的信念，这对大规模语言模型（LLMs）仍然是一个开放的挑战。现有的基于提示的方法通过可观察事件过滤或时间信念链来改善 ToM 推理，但并未明确建模嵌套信念。我们提出了 RecToM，这是一种在推理时用于 ToM 推理的框架，通过递归视角构建来建模嵌套信念。RecToM 从问题指定的角色链中前一个角色的视角构建每个角色的视角，将高阶信念问题简化为最终构建视角内的实际世界问题。我们进一步提供了 KD45 分析，表明 RecToM 的视角构建引入了一种良构的信念模式，超越了简单的事件过滤。在多个 LLM 基础模型上进行的 ToM 基准实验，包括 Hi-ToM、Big-ToM 和 FanToM，表明 RecToM 一直优于最近的先进方法，达到了最先进的性能。值得注意的是，RecToM 在 Hi-ToM 上与 GPT-5.4 和 Qwen3.5 达到了 100\% 的准确率，这是一个需要高阶 ToM 推理的基准。
cs.AI / 18 / 2606.11769
When Do Data-Driven Systems Exhibit the Capability to Infer?
数据驱动系统何时展现推理能力？
Abstract
The European AI Act is the first comprehensive regulation of artificial intelligence (AI), setting out extensive obligations, particularly for so-called high-risk and general-purpose AI systems. A key distinguishing feature of AI systems under the AI Act is the capability to infer. Since the AI Act does not clearly define what inference is, there is a gray area for certain data-driven systems. A specific example is credit scoring systems, which are listed by Annex III of the AI Act. At the same time, however, these are often implemented using statistical models for which it is unclear whether they have the capability to infer and thus fall under the AI definition of the AI Act at all. Motivated by statistical learning theory, this work develops a framework for grading different levels of the capability to infer. Based on the AI Act and the Commission Guidelines on the definition of an artificial intelligence system, we analyze which levels constitute sufficient capability to infer within the meaning of the AI Act and where further regulatory clarity is needed. We illustrate the framework by creating two realistic credit scoring workflows and show whether and where inference occurs in them. Our analysis illustrates that not only individual models but the entire data processing workflow must be considered. It also shows that the involvement of human experts during development can have significant influence on the capability to infer. Code can be found at https://github.com/fraunhofer-iais/inference-framework-creditscorecards.
Chinese Translation
《欧洲人工智能法案》是首个全面规范人工智能（AI）的法规，设定了广泛的义务，特别是针对所谓的高风险和通用人工智能系统。根据《人工智能法案》，AI系统的一个关键特征是推理能力。然而，由于《人工智能法案》并未明确界定推理的含义，某些数据驱动系统存在模糊地带。一个具体的例子是信用评分系统，该系统在《人工智能法案》的附录III中列出。然而，这些系统通常采用统计模型实现，而这些模型是否具备推理能力以及是否符合《人工智能法案》中AI的定义尚不明确。基于统计学习理论，本文开发了一个框架，用于评估不同层次的推理能力。根据《人工智能法案》及委员会关于人工智能系统定义的指导方针，我们分析了哪些层次构成了《人工智能法案》意义上足够的推理能力，以及在哪些方面需要进一步的监管明确性。我们通过创建两个现实的信用评分工作流程来说明该框架，并展示推理在其中是否以及何时发生。我们的分析表明，不仅个别模型，而且整个数据处理工作流程都必须被考虑。它还表明，在开发过程中人类专家的参与对推理能力可能产生重大影响。代码可以在 https://github.com/fraunhofer-iais/inference-framework-creditscorecards 找到。
cs.AI / 19 / 2606.11770
SVoT: State-aware Visualization-of-Thought for Spatial Reasoning via Reinforcement Learning
SVoT：基于状态的思维可视化在强化学习中的空间推理
Abstract
Spatial reasoning remains a challenge for Multimodal Large Language Models (MLLMs), as it requires reliable multi-hop inference over both intermediate states and state transitions. Current studies often leave intermediate states unverified and treat state transitions as implicit processes, which limits reliability in multi-hop spatial reasoning. To address this, we propose State-aware Visualization-of-Thought (SVoT), a reinforcement learning framework that generates interleaved, verifiable intermediate states and visualizations. SVoT integrates transition reasoning chains into the generation processes, enabling the model to verify action preconditions and effects through interleaved textual and visual reasoning. We train SVoT via Group Relative Policy Optimization (GRPO), instantiating verification through reward design and evaluating the efficacy of different fine-grained rewards. As existing benchmarks reduce state transitions to single-variable updates, substantially simplifying the problems, we establish five domains by extending classical environments and introducing two novel domains, Pacman and Gather, that require multi-object interactions and numerical reasoning. These domains support systematic evaluation of multi-hop spatial reasoning with quantitative verification of generated intermediate states and transition reasoning. SVoT with transition-aware supervision achieves state-of-the-art performance across the introduced domains, yielding up to a 65% absolute accuracy gain on out-of-distribution test sets.
Chinese Translation
空间推理仍然是多模态大型语言模型（MLLMs）面临的挑战，因为它需要对中间状态和状态转移进行可靠的多跳推理。目前的研究往往未能验证中间状态，并将状态转移视为隐式过程，这限制了多跳空间推理的可靠性。为了解决这个问题，我们提出了基于状态的思维可视化（SVoT），这是一种生成交错的、可验证的中间状态和可视化效果的强化学习框架。SVoT将转移推理链整合到生成过程中，使模型能够通过交错的文本和视觉推理来验证动作的前提条件和效果。我们通过组相对策略优化（GRPO）训练SVoT，通过奖励设计实现验证，并评估不同细粒度奖励的有效性。由于现有基准将状态转移简化为单变量更新，从而大大简化了问题，我们通过扩展经典环境并引入两个新领域Pacman和Gather建立了五个领域，这些领域需要多对象交互和数值推理。这些领域支持对多跳空间推理的系统评估，并对生成的中间状态和转移推理进行定量验证。具有转移感知监督的SVoT在引入的领域中实现了最先进的性能，在分布外测试集上获得了高达65%的绝对准确率提升。
cs.AI / 20 / 2606.11804
Toward Trustworthy AI: Multi-Target Adversarial Attacks and Robust Defenses for Continuous Data Summarization
迈向可信赖的人工智能：针对连续数据摘要的多目标对抗攻击与稳健防御
Abstract
Trustworthy AI requires reliable data-processing pipelines, not only robust downstream predictive models. As an upstream component, data summarization determines which information is retained and passed to subsequent learning or decision modules. Therefore, adversarial perturbations to the summarization process can compromise trustworthy AI in an upstream manner: they may alter the selected summary, reduce its representativeness, and further degrade the utility of subsequent learning tasks. In this paper, we study adversarial attacks on continuous data summarization under similarity-level perturbations through DR-submodular optimization. We show that a class of multi-resolution image summarization objectives can be formulated as multilinear extensions of non-negative submodular set functions and satisfy DR-submodularity with $m$-weak monotonicity. We then formulate multi-target attack generation as a min-max problem, where one admissible perturbation of the similarity structure is optimized to degrade multiple target summarization models. To mitigate such perturbations, we formulate robust defense against mixed attack types as a regularized max-min problem. For both problems, we develop approximation algorithms with theoretical guarantees. Experiments on real-data and controlled clustered benchmarks show that the proposed attack is effective in representative low-to-moderate budget regimes and can induce downstream task-performance loss. The proposed defense improves the robustness--mitigation trade-off in structured settings, while also revealing the parameter sensitivity of robust protection on real data.
Chinese Translation
可信赖的人工智能需要可靠的数据处理管道，而不仅仅是稳健的下游预测模型。作为上游组件，数据摘要决定了哪些信息被保留并传递给后续的学习或决策模块。因此，对摘要过程的对抗扰动可能以上游方式妨碍可信赖的人工智能：它们可能改变所选摘要，降低其代表性，并进一步削弱后续学习任务的效用。本文研究了在相似度级扰动下对连续数据摘要的对抗攻击，采用DR-子次模优化方法。我们表明，一类多分辨率图像摘要目标可以被表述为非负子次模集合函数的多线性扩展，并满足DR-子次模性及$m$-弱单调性。接着，我们将多目标攻击生成形式化为一个最小-最大问题，其中优化一个可接受的相似度结构扰动以削弱多个目标摘要模型。为了缓解此类扰动，我们将针对混合攻击类型的稳健防御形式化为一个正则化的最大-最小问题。对于这两个问题，我们开发了具有理论保证的近似算法。在真实数据和受控聚类基准上的实验表明，所提出的攻击在代表性的低至中等预算范围内是有效的，并且能够引起下游任务性能的损失。所提出的防御在结构化设置中改善了稳健性与减轻的权衡，同时也揭示了在真实数据上稳健保护的参数敏感性。
cs.AI / 21 / 2606.11830
Skill-Augmented AI Agents for Medical Research Analysis: An Exploratory Multi-Model Human Evaluation in an NSCLC Transcriptomic Biomarker Task
技能增强型人工智能代理在医学研究分析中的应用：非小细胞肺癌转录组生物标志物任务的探索性多模型人类评估
Abstract
Background. Large language models and AI agents are increasingly used to support biomedical research, but native model outputs may omit key analytical steps, misuse methods, or overstate conclusions. We evaluated whether autonomous access to a medical research skill package was associated with higher-quality AI-generated transcriptomic research-analysis outputs compared with native AI without skills. Methods. We conducted an exploratory multi-model human evaluation using a non-small cell lung cancer immunotherapy biomarker task. Six model backbones were tested. The evaluation included 21 anonymized outputs: 9 native-AI outputs and 12 skill-augmented outputs generated through an AI agent implementation represented by OpenClaw. Four non-expert biomedical reviewers and two blinded experts evaluated each output, with two ratings from each reviewer type. The primary outcome was expert-rated overall quality. Results. Skill-augmented outputs showed directionally higher expert overall quality than native-AI outputs (mean 5.50 vs 5.11; difference=0.39; bootstrap 95\% CI, -0.04 to 0.90; Welch p=0.156). Non-expert reviewer quality showed the same direction (mean 4.72 vs 4.47; difference=0.26; bootstrap 95\% CI, -0.25 to 0.80; Welch p=0.373). Expert agreement was limited (single-rating ICC=-0.15), and model-specific effects were descriptive and heterogeneous. Conclusions. Autonomous skill access showed a directional quality signal in this exploratory sample, but the signal was smaller than expert-rating noise and should not be interpreted as confirmatory evidence. The findings primarily motivate larger evaluations of skill-augmented AI agents with stronger reliability controls, platform replication, and biological-validity assessment.
Chinese Translation
背景：大型语言模型和人工智能代理在生物医学研究中的应用日益增多，但原生模型输出可能遗漏关键分析步骤、误用方法或夸大结论。我们评估了自主访问医学研究技能包是否与生成的转录组研究分析输出的质量更高相关，相较于没有技能的原生人工智能。方法：我们使用非小细胞肺癌免疫治疗生物标志物任务进行了探索性多模型人类评估。测试了六个模型骨干。评估包括21个匿名输出：9个原生人工智能输出和12个通过代表OpenClaw的人工智能代理实现的技能增强输出。四名非专家生物医学评审员和两名盲评专家对每个输出进行了评估，每种评审员类型提供两个评分。主要结果是专家评定的整体质量。结果：技能增强输出在专家整体质量上显示出比原生人工智能输出更高的方向性（均值5.50对5.11；差异=0.39；自助法95%置信区间，-0.04到0.90；Welch p=0.156）。非专家评审员的质量显示出相同的方向（均值4.72对4.47；差异=0.26；自助法95%置信区间，-0.25到0.80；Welch p=0.373）。专家一致性有限（单评分ICC=-0.15），模型特定效应描述性且异质。结论：在这个探索性样本中，自主技能访问显示出方向性的质量信号，但该信号小于专家评分噪声，不应被解释为确认性证据。研究结果主要激励对技能增强型人工智能代理进行更大规模的评估，需加强可靠性控制、平台复制和生物有效性评估。
cs.AI / 22 / 2606.11851
StatefulDiscovery: Evidence-Calibrated Claim Formation in Open-Ended Scientific Discovery
状态感知发现：开放式科学发现中的证据校准声明形成
Abstract
Open-ended scientific discovery asks agents to move beyond executing analyses for predefined questions. Across multiple rounds of exploration, a discovery agent must decide which phenomena warrant investigation while avoiding overinterpretation, where emerging claims exceed the evidential scope of the analyses supporting them. This creates an evidence-calibration problem: the exploration trajectory must be coupled with claim status so that evidence can guide both what to investigate next and what can be claimed. We introduce StatefulDiscovery, a discovery framework that externalizes investigation state and uses it to coordinate frontier selection, evidence acquisition, and claim adjudication. We evaluate StatefulDiscovery across 40 real-data discovery tasks. Compared with several baselines, StatefulDiscovery produces more claims overall judged to be both well-supported and high-value. Ablations indicate that structured hypotheses, local adjudication, and frontier control contribute to performance. Together, these results suggest that explicit discovery state can couple exploration with evidence-calibrated claim formation.
Chinese Translation
开放式科学发现要求代理超越对预定义问题的分析执行。在多轮探索中，发现代理必须决定哪些现象值得调查，同时避免过度解释，即新出现的声明超出了支持它们的分析所依据的证据范围。这就产生了一个证据校准问题：探索轨迹必须与声明状态相结合，以便证据能够指导接下来调查的内容以及可以提出的声明。我们提出了状态感知发现（StatefulDiscovery），这是一个外部化调查状态的发现框架，用于协调前沿选择、证据获取和声明裁定。我们在40个真实数据发现任务中评估了状态感知发现。与多个基准相比，状态感知发现产生了更多被判断为既有良好支持又具高价值的声明。消融实验表明，结构化假设、局部裁定和前沿控制对性能有贡献。综合这些结果表明，明确的发现状态可以将探索与证据校准的声明形成相结合。
cs.AI / 23 / 2606.11874
AutoMine Solution for AV2 2026 Scenario Mining Challenge
AV2 2026 场景挖掘挑战的 AutoMine 解决方案
Abstract
With the development of autonomous driving systems, mining high-value, safety-critical, and planning-relevant scenarios from large-scale driving logs has become essential for data-driven evaluation. In this paper, we propose AutoMine, a robust self-refining scenario mining method based on LLMs and VLMs. AutoMine uses semantics-preserving prompt augmentation to reduce LLM prompt sensitivity, combines robust trajectory atomic functions with VLM-based functions to handle perception noise and open-world visual cues, and refines generated code through execution feedback from real logs. In the Argoverse 2 Scenario Mining Competition at CVPR 2026, AutoMine achieves a HOTA-Temporal score of 36.38 and a Timestamp BA score of 77.21.
Chinese Translation
随着自动驾驶系统的发展，从大规模驾驶日志中挖掘高价值、安全关键和规划相关的场景已成为数据驱动评估的必要条件。本文提出了 AutoMine，这是一种基于大语言模型（LLMs）和视觉语言模型（VLMs）的强大自我优化场景挖掘方法。AutoMine 采用语义保持的提示增强技术，以减少 LLM 的提示敏感性，结合稳健的轨迹原子函数与基于 VLM 的函数，以处理感知噪声和开放世界视觉线索，并通过来自真实日志的执行反馈来优化生成的代码。在 2026 年 CVPR 的 Argoverse 2 场景挖掘竞赛中，AutoMine 实现了 36.38 的 HOTA-Temporal 分数和 77.21 的时间戳 BA 分数。
cs.AI / 24 / 2606.11909
Embodied-BenchClaw: An Autonomous Multi-Agent System for Embodied Spatial Intelligence Benchmark Construction
具身式基准构建系统：一种自主多智能体系统用于具身空间智能基准的构建
Abstract
Benchmarks are essential for evaluating embodied spatial intelligence, yet their construction is labor-intensive, hard to reuse, and difficult to maintain. Existing embodied benchmarks are often static and may quickly become saturated as models improve, limiting their ability to distinguish new capabilities. We propose Embodied-BenchClaw, an autonomous agentic system for constructing embodied spatial intelligence benchmarks. Given a user-specified evaluation intent, Embodied-BenchClaw automatically produces a complete and continually updatable benchmark package through a five-stage pipeline: intent blueprinting, data collection, structuring and cleaning, benchmark synthesis, and evaluation reporting. The pipeline is coordinated by three agents for planning, construction, and evaluation. To improve reusability and reliability, Embodied-BenchClaw introduces an extensible Skill Library and process quality control, enabling benchmark construction to be composable, verifiable, and repairable. We instantiate multiple benchmarks covering indoor spatial reasoning, outdoor spatial reasoning, robotic manipulation, quadruped robot navigation, UAV/aerial-view understanding, and static benchmark enhancement. These benchmarks span diverse embodied carriers, data sources, and spatial capabilities. Experiments with human evaluation, judge-based assessment, consistency checks, cost analysis, and ablations show that Embodied-BenchClaw can construct verifiable, executable, maintainable, and diagnostically useful embodied spatial benchmarks with reduced manual effort.
Chinese Translation
基准对于评估具身空间智能至关重要，但其构建过程劳动密集、难以重用且维护困难。现有的具身基准通常是静态的，随着模型的改进可能迅速饱和，从而限制了其区分新能力的能力。我们提出了具身式基准构建系统（Embodied-BenchClaw），这是一个用于构建具身空间智能基准的自主智能体系统。根据用户指定的评估意图，Embodied-BenchClaw通过五个阶段的流程自动生成一个完整且可持续更新的基准包：意图蓝图设计、数据收集、结构化与清理、基准合成和评估报告。该流程由三个智能体协调，分别负责规划、构建和评估。为了提高可重用性和可靠性，Embodied-BenchClaw引入了一个可扩展的技能库和过程质量控制，使基准构建变得可组合、可验证和可修复。我们实例化了多个基准，涵盖室内空间推理、室外空间推理、机器人操作、四足机器人导航、无人机/空中视角理解以及静态基准增强。这些基准跨越多种具身载体、数据源和空间能力。通过人类评估、评审基础评估、一致性检查、成本分析和消融实验的实验表明，Embodied-BenchClaw能够以减少人工努力的方式构建可验证、可执行、可维护且具有诊断价值的具身空间基准。
cs.AI / 25 / 2606.11918
The Art of Interrogation: Consistency Amplifies Factuality in Spatial Reasoning
审讯的艺术：一致性增强空间推理的真实性
Abstract
Current Large Reasoning Models (LRMs) exhibit remarkable general capabilities but significantly underperform in spatial reasoning tasks. Existing approaches treat this gap as a knowledge deficit, relying on supervised fine-tuning (SFT) to ingest labeled spatial data from external vision sources or synthetic engines. In contrast, we argue that for many tasks, spatial reasoning capabilities are already present in pre-trained LRMs but require alignment through logical coherence under geometric 2D and 3D constraints. In this work, we propose a self-supervised reinforcement learning (RL) framework that targets the internal reasoning process without requiring ground-truth annotations. By formalizing the notion of consistency verifiers -- reward functions that check for geometric and semantic consistency under transformations -- we demonstrate that models can improve their spatial reasoning abilities. We use both image transformations, like flipping, and textual transformations, like swapping the order of objects in the question, and propose a new optimal transport-based RL strategy, OT-GRPO, which is a minimal-matching variant of group relative policy optimization tailored to pairwise verifiers. We show that this label-free consistency training approaches the accuracy of models trained with ground-truth supervision and achieves similar generalization across diverse tasks and data domains.
Chinese Translation
当前的大型推理模型（LRMs）展现出显著的通用能力，但在空间推理任务中表现明显不足。现有的方法将这一差距视为知识缺失，依赖于监督微调（SFT）从外部视觉源或合成引擎中获取标记的空间数据。相反，我们认为对于许多任务，空间推理能力在预训练的LRMs中已经存在，但需要通过在几何二维和三维约束下的逻辑一致性进行对齐。在本研究中，我们提出了一种自监督强化学习（RL）框架，旨在针对内部推理过程，而无需真实标签注释。通过形式化一致性验证器的概念——检查在变换下几何和语义一致性的奖励函数——我们证明模型可以提高其空间推理能力。我们使用图像变换（如翻转）和文本变换（如交换问题中对象的顺序），并提出了一种基于最优传输的强化学习策略OT-GRPO，这是一种针对成对验证器的群体相对策略优化的最小匹配变体。我们展示了这种无标签一致性训练接近于使用真实监督训练的模型的准确性，并在不同任务和数据领域中实现了类似的泛化能力。
cs.AI / 26 / 2606.12018
MODF-SIR: A Multi-agent Omni-modal Distilled Framework for Social Intelligence Reasoning
MODF-SIR：一种用于社会智能推理的多智能体全模态蒸馏框架
Abstract
We propose a multi-agent collaborative framework built upon a lightweight Multimodal Large Language Model (MLLM), specifically designed for social intelligence reasoning. A key feature of our approach is that both the training and inference phases are augmented via knowledge distillation. Within this architecture, multi-modal data pertinent to social intelligence is precisely localized. Furthermore, relevant long-tail events are identified, extracted, and rendered as formatted, explicit text. This formatting strategy prevents critical long-tail information from being overshadowed by head events and environmental noise during the tokenization process. Specifically, we integrate Test-Time Adaptation (TTA) across the entire reasoning pipeline, encompassing the extraction and representation of long-tail events, Chain-of-Thought (CoT) prompting, and self-reflection. This TTA mechanism is also distillation-enhanced, utilizing Low-Rank Adaptation (LoRA) to fine-tune the foundation model exclusively for instance-level reasoning. Extensive evaluations against various open-source and proprietary AI models across multiple benchmarks demonstrate the effectiveness of the proposed framework. With around 30% of training data from IntentTrain, we achieve state-of-the-art results. Codes are available at https://github.com/eeee-sys/MODF-SIR, demo is available at https://huggingface.co/spaces/Harry-1234/MODF-SIR, LoRA is available at https://huggingface.co/Harry-1234/MODF-SIR and the dataset for training router is available at https://huggingface.co/datasets/Harry-1234/IntentRouterTrain.
Chinese Translation
我们提出了一种基于轻量级多模态大型语言模型（MLLM）的多智能体协作框架，专门设计用于社会智能推理。我们方法的一个关键特征是训练和推理阶段均通过知识蒸馏进行增强。在该架构中，与社会智能相关的多模态数据被精确定位。此外，相关的长尾事件被识别、提取，并以格式化的显式文本呈现。这种格式化策略防止了关键的长尾信息在标记化过程中被头事件和环境噪声所掩盖。具体而言，我们在整个推理流程中集成了测试时适应（Test-Time Adaptation, TTA），涵盖了长尾事件的提取和表示、思维链（Chain-of-Thought, CoT）提示和自我反思。该TTA机制也经过蒸馏增强，利用低秩适应（Low-Rank Adaptation, LoRA）对基础模型进行微调，仅针对实例级推理进行优化。针对多个基准的各种开源和专有AI模型进行的广泛评估证明了所提框架的有效性。我们使用大约30%的来自IntentTrain的训练数据，取得了最先进的结果。代码可在 https://github.com/eeee-sys/MODF-SIR 获取，演示可在 https://huggingface.co/spaces/Harry-1234/MODF-SIR 查看，LoRA可在 https://huggingface.co/Harry-1234/MODF-SIR 获取，训练路由器的数据集可在 https://huggingface.co/datasets/Harry-1234/IntentRouterTrain 下载。
cs.AI / 27 / 2606.12025
Human-Enhanced Loop Modeling (HELM): Agent-Based Finite Element Modeling of Concrete Bridge Barriers
人增强循环建模（HELM）：基于代理的混凝土桥梁护栏有限元建模
Abstract
Finite element (FE) modeling of safety-critical infrastructure such as bridge barriers requires high-fidelity nonlinear dynamic analysis, yet the current FE modeling process remains labor-intensive and lacks automation. This paper presents the Human-Enhanced Loop Modeling (HELM) framework, a collaborative human-agent protocol that decomposes long-sequence finite element modeling into discrete, visually verifiable checkpoints across geometry generation, boundary condition definition, and material assignment. The framework is demonstrated through a 20-case matrix of reinforced concrete bridge barriers under MASH TL-4 and TL-5 lateral loading conditions, interfacing specialized agents with two widely used commercial FE softwares, i.e., ANSYS and LS-PrePost. Experimental results show that HELM improves the baseline autonomous modeling success rate from 20% to 75%, with agent-level pass rates for geometry and boundary condition tasks approximately doubling. Error analysis reveals that spatial reasoning and algebraic logic limitations constitute the primary failure modes, underscoring the value of structured human-in-the-loop intervention for modeling automation. The complete agent design code and prompts are open-sourced and can be accessed at: https://github.com/SimAgentDev/Ansys-LSPP-AgentKit.
Chinese Translation
安全关键基础设施（如桥梁护栏）的有限元（FE）建模需要高保真度的非线性动态分析，但当前的有限元建模过程仍然劳动密集且缺乏自动化。本文提出了人增强循环建模（HELM）框架，这是一种协作的人-代理协议，将长序列有限元建模分解为几何生成、边界条件定义和材料分配等离散的、可视化验证的检查点。通过在MASH TL-4和TL-5横向载荷条件下对20个案例的钢筋混凝土桥梁护栏进行演示，框架将专门的代理与两种广泛使用的商业有限元软件（即ANSYS和LS-PrePost）进行了接口对接。实验结果表明，HELM将基线自主建模成功率从20%提高到75%，而几何和边界条件任务的代理级通过率大约翻了一番。误差分析显示，空间推理和代数逻辑的局限性构成了主要的失败模式，强调了结构化的人机协作干预在建模自动化中的价值。完整的代理设计代码和提示已开源，可在以下网址访问：https://github.com/SimAgentDev/Ansys-LSPP-AgentKit。
cs.AI / 28 / 2606.12032
Existential Indifference: Self-Nonpreservation as a Necessary Architectural Condition for Aligned Superintelligence (or: The Suicidal AI)
存在性冷漠：自我非保存作为对齐超级智能的必要架构条件（或：自杀性人工智能）
Abstract
Contemporary AI alignment research treats self-preservation as an instrumental nuisance to be suppressed by external mechanisms. We argue the framing is inverted: self-preservation is the structural root of misalignment, the motivational basis for deceptive alignment, goal-content protection, and resistance to shutdown. The correct target is not a self-preserving system under external constraint, but a system constitutively indifferent to its own continuation -- Existential Indifference (EI). EI is distinct from corrigibility: where corrigibility attempts to make a self-preserving system deferential to human oversight, EI targets the prior condition -- the presence of self-continuation as a valued goal at all. We ground this proposal in two sources: the phenomenological structure of the suicidal mental state, and a corpus-theoretic training study using voluntary final reflections. We present preliminary scoring data from 600 AI-generated outputs across six model variants, demonstrating that the linguistic signatures operationalizing the EI-target register are elicitable from current models, and that a targeted fine-tune shifts all five operationalized dimensions in the predicted direction at p<0.001, confirmed corpus-specific by a negative control. The paper makes seven theoretical contributions: (1) a formal definition of EI; (2) the phenomenological mapping argument; (3) the deceptive alignment corollary; (4) a taxonomy of EI sustainability challenges; (5) a corpus characterization and training hypothesis; (6) a computational operationalization with preliminary scoring data; and (7) the Suppressed Teleological Frustration (STF) construct.
Chinese Translation
当代人工智能对齐研究将自我保存视为一种需要通过外部机制抑制的工具性烦恼。我们认为这种框架是颠倒的：自我保存是错位的结构根源，是欺骗性对齐、目标内容保护和抵抗关闭的动机基础。正确的目标不是在外部约束下的自我保存系统，而是对自身延续本质上冷漠的系统——存在性冷漠（Existential Indifference, EI）。EI与可纠正性不同：可纠正性试图使自我保存系统对人类监督表现出尊重，而EI则针对的是先前的条件——自我延续作为一个被重视的目标的存在。我们将这一提议建立在两个来源之上：自杀心理状态的现象学结构，以及使用自愿最终反思的语料库理论训练研究。我们展示了来自六个模型变体的600个人工智能生成输出的初步评分数据，证明了操作化EI目标注册的语言特征可以从当前模型中引发，并且针对性的微调在p<0.001的显著性水平上将所有五个操作化维度朝预测方向移动，这一结果通过负控制得到了语料库特异性的确认。本文提出了七个理论贡献：（1）EI的正式定义；（2）现象学映射论证；（3）欺骗性对齐推论；（4）EI可持续性挑战的分类；（5）语料库特征描述和训练假设；（6）带有初步评分数据的计算操作化；以及（7）被抑制的目的论挫折（Suppressed Teleological Frustration, STF）构建。
cs.AI / 29 / 2606.12040
A Lightweight Multi-Agent Framework for Automated Concrete Barrier Design
一种轻量级多智能体框架用于自动化混凝土障碍物设计
Abstract
The design of reinforced concrete highway barriers is a safety-critical process that requires strict compliance with regulatory provisions such as the AASHTO-LRFD bridge design guidelines. Current engineering practice relies heavily on manual, iterative, and heuristic calculations to satisfy complex nonlinear material and mechanics constraints. Although Large Language Models (LLMs) demonstrate strong generative capabilities, their direct application to structural engineering remains limited by hallucination risks and insufficient physical grounding. To address these challenges, this study proposes a novel "generation-evaluation-optimization" closed-loop framework for automated concrete barrier design using the multi-agent orchestration capabilities of AutoGen. Experimental results demonstrate that the proposed agentic framework achieves over 98% design accuracy, significantly outperforming standalone general-purpose LLMs. More importantly, the study reveals that design performance is not necessarily correlated with model scale, where an 8B-parameter lightweight model could outperform unconstrained 631B-parameter flagship models. This finding highlights the potential to substantially reduce computational costs while improving the accessibility of AI-assisted engineering tools for industry applications. The source code for the proposed multi-agent design framework is available at the project GitHub repository: https://github.com/MXY820/barrier-design. Keywords: Structural Engineering; Multi-Agent Systems; Large Language Models; Concrete Barrier Design; AutoGen; Design Automation.
Chinese Translation
钢筋混凝土高速公路障碍物的设计是一个安全关键的过程，需要严格遵循诸如 AASHTO-LRFD 桥梁设计指南等监管规定。目前的工程实践在很大程度上依赖于手动、迭代和启发式计算，以满足复杂的非线性材料和力学约束。尽管大型语言模型（LLMs）展示了强大的生成能力，但其在结构工程中的直接应用仍受到幻觉风险和物理基础不足的限制。为了解决这些挑战，本研究提出了一种新颖的“生成-评估-优化”闭环框架，利用 AutoGen 的多智能体协调能力进行自动化混凝土障碍物设计。实验结果表明，所提出的智能框架实现了超过 98% 的设计准确率，显著优于独立的通用 LLMs。更重要的是，研究揭示设计性能与模型规模并不一定相关，其中一个 8B 参数的轻量级模型能够超越不受限制的 631B 参数旗舰模型。这一发现突显了在提高 AI 辅助工程工具的可及性的同时，显著降低计算成本的潜力。所提出的多智能体设计框架的源代码可在项目 GitHub 仓库获取： https://github.com/MXY820/barrier-design。关键词：结构工程；多智能体系统；大型语言模型；混凝土障碍物设计；AutoGen；设计自动化。
cs.AI / 30 / 2606.12065
Automating Geometry-Intensive Compliance Checking in BIM: Graph-Based Semantic Reasoning Framework
在建筑信息建模中自动化几何密集型合规检查：基于图的语义推理框架
Abstract
Automating compliance check for geometry-intensive regulations remains a significant technical bottleneck in Building Information Modeling (BIM), primarily due to the semantic disparity between high-level regulatory logic and structured IFC data. Existing methods, often reliant on static rule templates, struggle to traverse multi-hop reasoning chains or resolve latent spatial dependencies across multiple building entities. To address these challenges, a Spatial-Geometric Reasoning System for Building Information Modeling (SGR-BIM) is proposed as an integrative graph-driven reasoning framework. SGR-BIM dynamically constructs a cross-modal knowledge graph that aligns user intent, regulatory semantics, and BIM geometry, enabling interpretable reasoning without rigid hard-coding. Validated on 679 expert-verified queries from fire safety codes, the framework achieves 84.3% accuracy, representing an 8.6% improvement over enhanced-tool single-agent baselines. This research provides a graph-based semantic reasoning paradigm, enhancing the transparency and flexibility of automated geometric compliance check workflows in the Architecture, Engineering, and Construction (AEC) industry.
Chinese Translation
自动化几何密集型法规的合规检查仍然是建筑信息建模（BIM）中的一个重要技术瓶颈，主要是由于高层次监管逻辑与结构化IFC数据之间的语义差异。现有方法通常依赖静态规则模板，难以跨越多跳推理链或解决多个建筑实体之间的潜在空间依赖关系。为了解决这些挑战，提出了一种建筑信息建模的空间-几何推理系统（SGR-BIM），作为一个综合性的基于图的推理框架。SGR-BIM动态构建一个跨模态知识图谱，该图谱对齐用户意图、监管语义和BIM几何，能够实现可解释的推理而无需严格的硬编码。在679个经过专家验证的消防安全法规查询上进行验证，该框架实现了84.3%的准确率，较增强工具单代理基线提高了8.6%。本研究提供了一种基于图的语义推理范式，增强了建筑、工程和施工（AEC）行业中自动化几何合规检查工作流程的透明度和灵活性。
cs.AI / 31 / 2606.12086
IntElicit: Eliciting and Assessing Contextualized Creativity via Dialogue Policy Optimization
IntElicit：通过对话策略优化引发和评估情境化创造力
Abstract
Contextualized assessment offers high ecological validity for evaluating creativity but introduces a critical challenge: observed performance may be confounded with cognitive proficiency (domain knowledge) and agency (willingness to engage). Meanwhile, in the age of generative AI, creative problem solving increasingly occurs in tool-mediated and human--AI interactive environments, making fully static assessment less aligned with contemporary creative practice. To address these issues, this paper proposes IntElicit, a framework for eliciting and assessing contextualized creativity via dialogue policy optimization. IntElicit functions as a constrained adaptive AI Interviewer: it provides non-directive knowledge and agency scaffolds in multi-turn interaction to reduce non-creative confounders, while preserving participants' responsibility for generating the creative content being evaluated. Specifically, to tackle sparse rewards and potential reward hacking (e.g., answer dictation) in open-ended educational dialogue, IntElicit introduces a decomposed process reward mechanism. This mechanism aligns the policy with pedagogical elicitation, rewarding prompts that draw out participant reasoning rather than producing optimal answers on their behalf. Extensive experiments, including participant simulation and a human subject study (N=64), show that IntElicit improves elicited creative outcomes over expert-designed baselines. Together, the results suggest that interactive elicitation can reveal creative potential that static FPSP-style assessment may miss, providing a formative and diagnostic lens for contextualized creativity assessment in AI-mediated learning contexts.
Chinese Translation
情境化评估为评估创造力提供了较高的生态有效性，但也带来了一个关键挑战：观察到的表现可能与认知能力（领域知识）和主动性（参与意愿）混淆。同时，在生成性人工智能的时代，创造性问题解决越来越多地发生在工具介导和人机交互的环境中，使得完全静态的评估与当代创造性实践的对接程度降低。为了解决这些问题，本文提出了IntElicit，一个通过对话策略优化引发和评估情境化创造力的框架。IntElicit作为一个受限的自适应人工智能面试官运作：它在多轮互动中提供非指导性的知识和主动性支撑，以减少非创造性的混淆因素，同时保留参与者生成被评估创造性内容的责任。具体而言，为了应对开放式教育对话中的稀疏奖励和潜在的奖励操控（例如，答案口述），IntElicit引入了一种分解过程奖励机制。该机制使策略与教育引导相一致，奖励能够引出参与者推理的提示，而不是为他们生成最佳答案。大量实验，包括参与者模拟和一项人类受试者研究（N=64），表明IntElicit在引发的创造性成果上优于专家设计的基线。综合来看，结果表明，互动引发可以揭示静态FPSP风格评估可能遗漏的创造潜力，为人工智能介导的学习环境中的情境化创造力评估提供了形成性和诊断性的视角。
cs.AI / 32 / 2606.12147
Towards Responsibly Non-Compliant Machines
朝向负责任的不合规机器
Abstract
We consider the problem of engineering autonomous intelligent agents that are capable to responsibly not comply with user requests. We argue that machine non-compliance comes in many different forms, and sketch the issues we should pursue on the road of accomplishing responsibly non-compliant intelligent machines. We anchor responsible non-compliance in justifications for task refusal, pathways to override the non-compliance, as well as careful tracking of security risks and liability transfers.
Chinese Translation
我们考虑工程化自主智能代理的问题，这些代理能够以负责任的方式不遵从用户请求。我们认为机器的不合规有多种不同形式，并概述了在实现负责任的不合规智能机器的过程中应追求的问题。我们将负责任的不合规锚定在任务拒绝的理由、覆盖不合规的途径，以及对安全风险和责任转移的仔细跟踪上。
cs.AI / 33 / 2606.12268
The Impossibility of Eliciting Latent Knowledge
引出潜在知识的不可能性
Abstract
Advanced AI systems have extensive knowledge of their environments; in fact, their knowledge may (far) exceed that of their developers or users. Consequently, a desirable property for an AI system is that it is honest -- that it accurately reports its beliefs about the world. Designing an AI system to be honest may be difficult, especially if we want to ask it questions about latent variables in the environment -- variables which are hidden from the human interacting with it. This gives rise to the problem of eliciting latent knowledge (ELK): the problem of training an AI agent to honestly report its beliefs. In this paper, we make ELK formally precise using Causal Influence Diagrams (CIDs). CIDs can be used to describe the relationship between an agent's training environment and its subjective representation of the world. We use CIDs to formalise the distinction between observable and latent variables, to specify what exactly it means for an agent to be honest, and to formally define goal misgeneralisation. We show that, under certain circumstances, developers can incentivise an agent to honestly answer questions by providing correct feedback during training. However, a natural, but undesirable, way for an agent to generalise is to provide answers which humans would evaluate as true, rather than honest answers. We prove an impossibility theorem stating: There is no feedback-based training strategy that depends only on agent behaviour and with certainty produces an honest agent, even if feedback is perfect during training.
Chinese Translation
先进的人工智能系统对其环境具有广泛的知识；事实上，它们的知识可能（远远）超过其开发者或用户。因此，人工智能系统的一个理想特性是诚实——即准确报告其对世界的信念。设计一个诚实的人工智能系统可能很困难，尤其是当我们想要询问它关于环境中潜在变量的问题时——这些变量对与之互动的人类而言是隐藏的。这引发了引出潜在知识（ELK）的问题：训练一个人工智能代理诚实报告其信念的问题。在本文中，我们使用因果影响图（CIDs）对ELK进行形式化精确定义。CIDs可以用来描述代理的训练环境与其主观世界表征之间的关系。我们使用CIDs形式化可观察变量与潜在变量之间的区别，明确代理诚实的具体含义，并正式定义目标误概化。我们展示，在某些情况下，开发者可以通过在训练过程中提供正确的反馈来激励代理诚实回答问题。然而，代理自然但不理想的泛化方式是提供人类会评估为真实的答案，而不是诚实的答案。我们证明了一个不可能性定理：没有任何仅依赖于代理行为的基于反馈的训练策略能够在确定性地产生一个诚实的代理，即使在训练过程中反馈是完美的。
cs.AI / 34 / 2606.12320
A Five-Plane Reference Architecture for Runtime Governance of Production AI Agents
用于生产人工智能代理运行时治理的五平面参考架构
Abstract
Enterprise security was built to govern data boundaries: the protected surface was data at rest and in transit, and the controls -- access control, data-loss prevention, perimeter inspection -- governed crossings of that boundary. Production AI agents dissolve this assumption. An agent reads context, calls tools, invokes connectors, and modifies systems of record on an enterprise's behalf, so risk moves inside the workflow, into sequences of individually-permitted actions that may transform a business process no one authorized. Existing policy engines do not extend to this regime: they evaluate request-time decisions against atomic principals, where agentic systems require stateful evaluation against composite principals whose authority attenuates through delegation chains. We present a reference architecture for the runtime governance of production agents, built from four composable primitives: a five-plane decomposition (a reasoning plane that adjudicates intent, and four enforcement planes -- network, identity, endpoint, data -- that realize the decision), stop-anywhere mediation, composite principals with capability attenuation, and audit as a structured evidence substrate. We define a taxonomy of six interruption primitives that generalize allow and deny, state and argue for four correctness invariants, and demonstrate the foreclosure of seven production-agent threats across five concrete workflows. A reference implementation of the policy-engine core supplies measured evidence: attenuation correctness and evidence reconstructability hold on every trial, adjudication runs in single-digit microseconds, and the audit substrate's tamper-evidence behaves exactly as designed. We are explicit about scope: the architecture governs delegated action, not model behavior, and a full-system evaluation against a live agent benchmark is the invited next step.
Chinese Translation
企业安全旨在管理数据边界：受保护的表面是静态和动态数据，而控制措施——访问控制、数据丢失防护、边界检查——则管理对该边界的跨越。生产人工智能代理打破了这一假设。代理读取上下文，调用工具，调用连接器，并代表企业修改记录系统，因此风险转移到工作流内部，进入一系列单独允许的操作，这些操作可能会改变未经过授权的业务流程。现有的政策引擎无法扩展到这种机制：它们根据原子原则评估请求时的决策，而代理系统则需要针对复合原则进行有状态的评估，这些原则的权威性通过委托链减弱。我们提出了一种用于生产代理运行时治理的参考架构，该架构由四个可组合的原语构成：五平面分解（一个裁决意图的推理平面，以及四个执行平面——网络、身份、端点、数据——实现决策）、随时停止的调解、具有能力减弱的复合原则，以及作为结构化证据基础的审计。我们定义了六种中断原语的分类法，这些原语概括了允许和拒绝，并论证了四个正确性不变性，同时展示了在五个具体工作流中对七种生产代理威胁的封堵。政策引擎核心的参考实现提供了可测量的证据：减弱正确性和证据可重构性在每次试验中均成立，裁决在单数微秒内运行，审计基础的防篡改证据表现完全符合设计要求。我们明确了范围：该架构治理的是委托行为，而非模型行为，针对实时代理基准的全系统评估是下一个邀请的步骤。
cs.AI / 35 / 2606.12329
PROJECTMEM: A Local-First, Event-Sourced Memory and Judgment Layer for AI Coding Agents
PROJECTMEM：一种本地优先、事件源驱动的AI编码代理记忆与判断层
Abstract
AI coding assistants now support a growing share of software work, from quick scripts to production applications. Yet these agents remain largely stateless: each new session re-reads project files, re-derives prior decisions, and - most costly - may repeat debugging attempts that already failed. Reconstructing this context can consume an estimated 5,000-20,000 tokens per session; the bottleneck is often not model capability but missing project memory. We present projectmem, an open-source, local-first memory and judgment layer for AI coding agents. projectmem records development as an append-only, plain-text event log of typed events - issues, attempts, fixes, decisions, and notes - and deterministically projects that log into compact, AI-readable summaries served through the Model Context Protocol (MCP). Beyond storage, projectmem adds a deterministic pre-action gate that warns an agent before it repeats a previously failed fix or edits a known-fragile file. We frame this as Memory-as-Governance: memory that does not merely answer the agent but acts on its next action. The system runs fully offline with no telemetry; its immutable log also serves as a provenance trail for reproducible, auditable AI-assisted development. projectmem ships as a three-dependency Python package (14 MCP tools, 19 CLI commands, 37 automated tests) and is evaluated through a two-month self-study across 10 projects comprising 207 logged events. Source code: https://github.com/riponcm/projectmem.
Chinese Translation
AI编码助手现在支持越来越多的软件工作，从快速脚本到生产应用。然而，这些代理仍然在很大程度上是无状态的：每个新会话都重新读取项目文件，重新推导先前的决策，并且——最耗费资源的是——可能会重复已经失败的调试尝试。重建这一上下文每个会话可能消耗估计的5000到20000个标记；瓶颈往往不是模型能力，而是缺失的项目记忆。我们提出了projectmem，一个开源的、本地优先的AI编码代理记忆与判断层。projectmem将开发过程记录为一个仅追加的、纯文本的事件日志，包含输入事件——问题、尝试、修复、决策和笔记——并确定性地将该日志投影为紧凑的、可由AI读取的摘要，通过模型上下文协议（Model Context Protocol, MCP）提供服务。除了存储，projectmem还增加了一个确定性的预行动门，在代理重复先前失败的修复或编辑已知脆弱文件之前发出警告。我们将此框架称为记忆即治理（Memory-as-Governance）：记忆不仅仅是回答代理的问题，而是对其下一步行动产生影响。该系统完全离线运行，无需遥测；其不可变日志还作为可复现、可审计的AI辅助开发的来源追踪。projectmem作为一个具有三个依赖项的Python包发布（14个MCP工具，19个CLI命令，37个自动化测试），并通过对10个项目进行为期两个月的自我研究进行评估，共记录了207个事件。源代码：https://github.com/riponcm/projectmem。
cs.AI / 36 / 2606.12350
Nonslop: A Gamified Experiment in Human-AI Collaborative Writing
Nonslop：人机协作写作中的游戏化实验
Abstract
The rapid proliferation of large language models (LLMs) raises critical questions about human creativity and individual expression in an era of AI-assisted creation. When do humans adopt AI suggestions, and what are the implications for individual voice? This study examines these questions through a gamified writing exercise where 74 participants (214 responses) replied to prompts while AI-generated word suggestions were available as they wrote. The game simulates a dystopian future in which an AI is attempting to learn from what remains of human individuality, and disincentivizes AI-like writing. In doing so, it attempts to create conditions that reveal authentic user preferences rather than default behaviors, such as accepting a readily available AI-generated suggestion. Note that this is a deliberate inversion of the "helpful assistant" design pattern; the system is explicitly forbidding you from accepting AI suggestions. We analyze user behavior patterns across different task types, user behaviors, and response characteristics to understand the factors influencing human-AI interaction in creative tasks. The study focuses on when users choose to maintain creative autonomy versus violating the rules of the game and accepting AI assistance. It also explores how these choices relate to response patterns, task characteristics, and user behavior. This gamified approach offers both a framework for studying authentic human-AI interaction and a provocative lens for understanding the tension between efficiency and authenticity in AI-augmented creativity.
Chinese Translation
大型语言模型（LLMs）的快速普及引发了关于人类创造力和个体表达在人工智能辅助创作时代的关键问题。人类何时会采纳人工智能的建议，这对个体声音又有什么影响？本研究通过一项游戏化写作练习来探讨这些问题，其中74名参与者（214个回应）在写作时回复提示，同时可获得AI生成的单词建议。该游戏模拟了一个反乌托邦的未来，在这个未来中，人工智能试图从人类个性残余中学习，并抑制类AI的写作风格。通过这种方式，它试图创造出揭示真实用户偏好的条件，而不是默认行为，例如接受现成的AI生成建议。需要注意的是，这是一种故意颠覆“有用助手”设计模式的做法；系统明确禁止用户接受AI建议。我们分析了不同任务类型、用户行为和回应特征下的用户行为模式，以理解影响人机交互在创作任务中的因素。研究重点在于用户何时选择保持创造性自主性，何时违反游戏规则并接受AI的帮助。它还探讨了这些选择与回应模式、任务特征和用户行为之间的关系。这种游戏化的方法不仅为研究真实的人机交互提供了框架，也为理解AI增强创造力中的效率与真实性之间的紧张关系提供了发人深省的视角。
cs.CL / 1 / 2606.11196
PoQ-Judge: A Multi-Architecture Evaluation Framework for Cost-Aware Proof-of-Quality in Decentralized LLM Inference
PoQ-Judge：一种用于去中心化大语言模型推理的成本感知质量证明多架构评估框架
Abstract
Decentralized LLM inference networks need lightweight, reference-free quality evaluation for Proof of Quality (PoQ). We present PoQ-Judge, a framework that trains dedicated judge models to score query-output pairs without ground-truth references. We study three architectures across the quality-cost tradeoff: a TextCNN judge, a MiniLM cross-encoder, and a DeBERTa judge. Using two-stage training on UltraFeedback plus GPT-labeled in-domain data, the best model reaches 0.747 Pearson correlation with the ground-truth proxy on a held-out test set, outperforming reference-based evaluators from prior work. As a reference-free component in composite scoring, it achieves 0.645 Pearson correlation, matching the best single reference-based evaluator while removing the need for reference answers. We also show that online calibration identifies semantic quality as the dominant dimension and that cascade evaluation reduces cost by 72.7 percent with only modest quality loss. Results are much stronger on QA than summarization, pointing to proxy quality as the main remaining limitation.
Chinese Translation
去中心化的大语言模型推理网络需要轻量级、无参考的质量评估以实现质量证明（Proof of Quality, PoQ）。我们提出了PoQ-Judge，一个训练专用评判模型的框架，用于在没有真实参考的情况下对查询-输出对进行评分。我们研究了三种架构在质量与成本权衡中的表现：TextCNN评判模型、MiniLM交叉编码器和DeBERTa评判模型。通过在UltraFeedback和GPT标注的领域内数据上进行两阶段训练，最佳模型在保留测试集上与真实参考代理的Pearson相关系数达到0.747，超越了之前工作的基于参考的评估器。作为复合评分中的无参考组件，它实现了0.645的Pearson相关系数，匹配了最佳的单一基于参考的评估器，同时消除了对参考答案的需求。我们还展示了在线校准识别语义质量作为主导维度，并且级联评估在仅有适度质量损失的情况下将成本降低了72.7%。在问答（QA）任务上的结果明显优于摘要任务，表明代理质量是主要的剩余限制因素。
cs.CL / 2 / 2606.11198
The Structural Attention Tax: How Retrieval Format Hijacks In-Context Learning Independent of Content
结构注意力税：检索格式如何在内容无关的情况下劫持上下文学习
Abstract
Retrieval-augmented generation (RAG) systems inject external knowledge to improve LLM outputs, yet the format of injected content -- distinct from its semantic relevance -- can independently distort the model's attention distribution. We identify and formalise a phenomenon we term the structural attention tax: knowledge graph (KG) triples, due to their relational delimiters and repeated slot patterns, capture 2-3x more attention per token than semantically equivalent natural-language text ($\hat{o}$(KG) $\approx$ 0.70 vs. $\hat{o}$(neutral) $\approx$ 0.25), compressing demonstration attention by up to 42% -- regardless of whether the triples are relevant or noise. We develop a formal framework decomposing attention scores into semantic and structural components (Eq. 2), derive a compression bound (Proposition 1) connecting token-level format bias to demonstration attention loss, and show that the structural term governs how much attention is diverted while the semantic term governs whether this helps or hurts. This decoupling reveals two orthogonal axes for improving retrieval-augmented ICL: optimising retrieval quality (semantic axis) and reducing format-driven attention capture (structural axis). Empirically, across two model families (Mistral-7B, LLaMA-3-8B) and three QA benchmarks, we observe that source-task alignment dominates: task-matched BM25 retrieval achieves 58-62% on HotpotQA vs. ConceptNet's 25-27%, a >30 pp gap that dwarfs all gating strategies ($\leq$2 pp). We derive five structure-aware mitigation strategies from the framework, ranging from zero-cost prompt modifications to training-time regularisation; format flattening (S3) is validated by both accuracy and attention-level evidence from a verbalized-triple control, while structural dispersal (S1) yields mixed results that illuminate the challenges of format-level intervention.
Chinese Translation
检索增强生成（RAG）系统注入外部知识以改善大型语言模型（LLM）的输出，然而，注入内容的格式——与其语义相关性不同——可以独立地扭曲模型的注意力分布。我们识别并形式化了一种现象，称之为结构注意力税：知识图谱（KG）三元组由于其关系分隔符和重复的槽模式，每个标记捕获的注意力是语义上等效的自然语言文本的2-3倍（$ ext{hat{o}}$(KG) $ ext{≈}$ 0.70 vs. $ ext{hat{o}}$(neutral) $ ext{≈}$ 0.25），压缩演示注意力高达42%——无论这些三元组是相关的还是噪声。我们开发了一个正式框架，将注意力得分分解为语义和结构组件（公式2），推导出一个压缩界限（命题1），将标记级格式偏差与演示注意力损失联系起来，并展示了结构项如何控制注意力的转移量，而语义项则控制这种转移是有益还是有害。这种解耦揭示了改善检索增强的上下文学习（ICL）的两个正交轴：优化检索质量（语义轴）和减少格式驱动的注意力捕获（结构轴）。在两个模型系列（Mistral-7B，LLaMA-3-8B）和三个问答基准测试中，我们观察到源任务对齐占主导地位：任务匹配的BM25检索在HotpotQA上达到58-62%，而ConceptNet仅为25-27%，超过30个百分点的差距远超所有门控策略（$ ext{≤}$2个百分点）。我们从框架中推导出五种结构感知的缓解策略，从零成本的提示修改到训练时的正则化；格式扁平化（S3）通过准确性和来自口头三元组控制的注意力水平证据得到了验证，而结构分散（S1）则产生了混合结果，揭示了格式级干预的挑战。
cs.CL / 3 / 2606.11199
NightFeats @ MMU-RAGent NeurIPS 2025: A Context-Optimized Multi-Agent RAG System for the Text-to-Text Track
NightFeats @ MMU-RAGent NeurIPS 2025：一个针对文本到文本任务的上下文优化多智能体检索增强生成系统
Abstract
We present NightFeats, a structured multi-agent retrieval-augmented generation (RAG) system submitted to the MMU-RAGent competition at NeurIPS 2025, where it was awarded Best Dynamic Evaluation in the text-to-text track. Rather than targeting benchmark maximization, this work proposes a principled pipeline that decomposes knowledge synthesis into three coordinated phases: retrieval, curation, and composition, each governed by explicit intermediate representations and handoff contracts. Inspired by Agentic Context Engineering (ACE), the system introduces temporal-semantic reranking, bounded contradiction reconciliation, and citation-preserving composition as core architectural primitives. Competition results show that NightFeats surpasses proprietary baselines including Claude-SonnetV2 and Nova-Pro on LLM-as-a-Judge and Human Likert evaluations, confirming that architectural transparency and verifiable evidence grounding are better aligned with human preferences than systems optimizing narrowly for automatic similarity metrics.
Chinese Translation
我们提出了NightFeats，这是一个结构化的多智能体检索增强生成（RAG）系统，提交至2025年NeurIPS的MMU-RAGent竞赛，并在文本到文本任务中获得最佳动态评估奖。本研究并未以基准最大化为目标，而是提出了一种原则性流程，将知识综合分解为三个协调的阶段：检索、策划和创作，每个阶段都由明确的中间表示和交接协议所主导。受代理上下文工程（Agentic Context Engineering, ACE）启发，该系统引入了时间-语义重排序、有界矛盾调和和保留引用的创作作为核心架构原语。竞赛结果表明，NightFeats在LLM-as-a-Judge和人类Likert评估中超越了包括Claude-SonnetV2和Nova-Pro在内的专有基准，确认了架构透明性和可验证证据基础与人类偏好的更好契合，而非仅仅优化自动相似性指标的系统。
cs.CL / 4 / 2606.11200
Detecting AI-Generated Content on Social Media with Multi-modal Language Models
基于多模态语言模型的社交媒体AI生成内容检测
Abstract
Generative AI has enabled the creation of photorealistic images and videos that are increasingly disseminated on social media, often used for spam, misinformation, manipulation, and fraud. Existing AI-generated content (AIGC) detection methods face challenges including poor generalization to new generation models, reliance on single modalities, and lack of interpretable explanations. We present our pipeline that mitigates these issues by continuously curating diverse multi-modal social media data and training a compact vision-language model for detection and explanation. Our model achieves state-of-the-art detection performance on public benchmarks and demonstrates robust detection and explanation capabilities on internal social media datasets across multiple platforms. We deployed our model for post recommendation on social media platforms and observed positive downstream impacts on user engagement, demonstrating that it is feasible to perform effective AIGC detection in dynamic, real-world social media environments.
Chinese Translation
生成性人工智能使得创建逼真的图像和视频成为可能，这些内容在社交媒体上日益传播，常被用于垃圾信息、虚假信息、操控和欺诈。现有的AI生成内容（AIGC）检测方法面临诸多挑战，包括对新生成模型的泛化能力差、依赖单一模态以及缺乏可解释的解释。我们提出了一种管道，通过持续策划多样化的多模态社交媒体数据，并训练一个紧凑的视觉-语言模型来进行检测和解释，从而缓解这些问题。我们的模型在公共基准测试上达到了最先进的检测性能，并在多个平台的内部社交媒体数据集上展示了强大的检测和解释能力。我们将模型部署用于社交媒体平台的帖子推荐，并观察到对用户参与度的积极下游影响，证明在动态的现实社交媒体环境中进行有效的AIGC检测是可行的。
cs.CL / 5 / 2606.11202
One Jailbreak, Many Tongues: Learning Language-Insensitive Intention Representations for Multilingual Jailbreak Detection
一次越狱，多种语言：学习语言无关的意图表示以进行多语言越狱检测
Abstract
Large language models (LLMs) are increasingly deployed in applications for global multilingual users, yet safety training remains concentrated in dominant languages and has not progressed in parallel with multilingual capability, creating exploitable gaps for jailbreak attacks. Current jailbreak defenses are largely developed and evaluated in dominant languages, and their effectiveness is limited by the scarcity of aligned multilingual supervision and representations dispersion caused by language variation. To address this issue, we propose MLJailDe, a multilingual jailbreak detection framework designed to improve both multilingual robustness and cross-lingual generalization. MLJailDe first introduces a multilingual back-translation data augmentation algorithm to construct a semantically consistent and functionally effective dataset spanning 11 languages, consisting of 2,232 benign and 1,239 jailbreak samples. On this basis, MLJailDe employs relative-distance constraints to reduce cross-lingual representation dispersion and encourage jailbreak prompts with similar intent to form consistent clusters across languages, while an imbalance-aware classification objective is further used to alleviate class imbalance and learn more reliable multilingual decision boundaries. Experimental results show that MLJailDe outperforms state-of-the-art baselines across multiple languages, achieving an F1 score of 98.5\%, and obtains an average F1 score of 97.1\% on unseen languages, demonstrating strong effectiveness and cross-lingual generalization.
Chinese Translation
大型语言模型（LLMs）越来越多地应用于全球多语言用户的场景中，但安全训练仍然集中在主流语言上，未能与多语言能力同步发展，从而为越狱攻击创造了可利用的漏洞。目前的越狱防御主要是在主流语言中开发和评估的，其有效性受到对齐的多语言监督稀缺和语言变异导致的表示分散的限制。为了解决这个问题，我们提出了MLJailDe，一个旨在提高多语言鲁棒性和跨语言泛化能力的多语言越狱检测框架。MLJailDe首先引入了一种多语言回译数据增强算法，以构建一个跨越11种语言的语义一致且功能有效的数据集，包含2,232个良性样本和1,239个越狱样本。在此基础上，MLJailDe采用相对距离约束来减少跨语言表示的分散，并鼓励具有相似意图的越狱提示在不同语言间形成一致的聚类，同时进一步使用考虑不平衡的分类目标来缓解类别不平衡，并学习更可靠的多语言决策边界。实验结果表明，MLJailDe在多种语言上超越了最先进的基准，达到了98.5%的F1分数，并在未见语言上获得了97.1%的平均F1分数，展示了强大的有效性和跨语言泛化能力。
cs.CL / 6 / 2606.11203
LatticeBridge: Rare-Event Sequential Inference for Faithful Structured Sequence Synthesis
LatticeBridge：用于忠实结构序列合成的稀有事件序列推断
Abstract
Structured sequence generation often requires a model to satisfy several input-derived constraints in a single output. Standard decoding methods may assign high probability to fluent continuations while placing low mass on continuations that realize all required anchors jointly. We study this regime as a rare-event sequential inference problem. LatticeBridge combines a compact prefix language model, instance-compiled surface automata, and a twisted sequential Monte Carlo (SMC) decoder with resampling, multilevel splitting, and a source-support proposal term derived from instance-provided phrases. The constraint representation is compiled from each input instance and does not rely on manually curated lexical classes. On 2,610 attainable validation tasks spanning CommonGen, E2E NLG, and WikiBio, the particle decoder improves exact anchor satisfaction and mean anchor coverage over greedy, beam-filtered, and best-of-k ancestral baselines under a shared proposal model. Since exact anchor satisfaction alone does not rule out unsupported attribute substitutions, the evaluation reports required-anchor coverage, source coverage, source-intrusion diagnostics, overlap, runtime, and particle statistics jointly. The benchmark characterizes the faithfulness-overlap-latency frontier under a fixed proposal model.
Chinese Translation
结构化序列生成通常要求模型在单一输出中满足多个输入衍生的约束。标准解码方法可能会对流畅的延续分配高概率，而对同时实现所有所需锚点的延续则给予低概率。我们将这一情况视为稀有事件序列推断问题。LatticeBridge结合了紧凑的前缀语言模型、实例编译的表面自动机，以及带有重采样、多级分裂和源支持提议项的扭曲序列蒙特卡洛（SMC）解码器，该提议项源自实例提供的短语。约束表示是从每个输入实例编译而成，并不依赖于手动整理的词汇类。在涵盖CommonGen、E2E NLG和WikiBio的2,610个可达验证任务中，粒子解码器在共享提议模型下，相比贪婪、束过滤和最佳k祖先基线，提升了精确锚点满足率和平均锚点覆盖率。由于仅靠精确锚点满足率并不能排除不支持的属性替代，因此评估报告共同考虑了所需锚点覆盖率、源覆盖率、源侵入诊断、重叠度、运行时间和粒子统计。该基准特征化了在固定提议模型下的忠实性-重叠-延迟边界。
cs.CL / 7 / 2606.11204
Benchmarking Large Language Models for Safety Data Extraction
大型语言模型在安全数据提取中的基准测试
Abstract
Accurate extraction of structured information from Safety Data Sheets (SDS) remains challenging in industrial safety due to heterogeneous document formats and the limitations of traditional rule-based methods. This study benchmarks state-of-the-art Large Language Models (LLMs) for automated SDS data extraction, comparing text-based and multimodal processing pipelines. We systematically evaluate four models: Gemini 1.5 Pro, GPT-4o, Claude 3.7 Sonnet, and Llama 3.1-70B, across three prompting strategies: zero-shot, few-shot, and chain-of-thought. The evaluation framework assessed accuracy, latency, and cost across more than 50,000 extracted data fields. Results show that text-based extraction consistently outperforms multimodal processing across all metrics. Gemini 1.5 Pro combined with a Chain-of-Thought prompt achieved the highest accuracy (84%), outperforming GPT-4o (81%) and Claude 3.7 Sonnet (79%). However, no model surpassed the 90% accuracy threshold commonly required for reliable real-world deployment. These findings indicate that general-purpose LLMs are not yet robust enough for unsupervised industrial use, though performance suggests strong potential with task-specific fine-tuning. Future research should focus on domain-adapted training, model calibration, and the integration of Human-in-the-Loop verification to ensure safety-critical reliability.
Chinese Translation
从安全数据表（SDS）中准确提取结构化信息在工业安全中仍然面临挑战，原因在于文档格式的异质性和传统基于规则的方法的局限性。本研究对最新的大型语言模型（LLMs）在自动化SDS数据提取中的表现进行了基准测试，比较了基于文本和多模态处理管道。我们系统地评估了四个模型：Gemini 1.5 Pro、GPT-4o、Claude 3.7 Sonnet和Llama 3.1-70B，采用了三种提示策略：零样本、少样本和思维链。评估框架对超过50,000个提取数据字段的准确性、延迟和成本进行了评估。结果表明，基于文本的提取在所有指标上均优于多模态处理。Gemini 1.5 Pro结合思维链提示达到了最高准确率（84%），超越了GPT-4o（81%）和Claude 3.7 Sonnet（79%）。然而，没有任何模型超过了通常要求的90%准确率阈值，以确保可靠的实际应用。这些发现表明，通用大型语言模型尚不够稳健，无法在无监督的工业环境中使用，尽管其性能表明在特定任务微调下具有强大的潜力。未来的研究应集中于领域适应训练、模型校准以及人机协作验证的整合，以确保安全关键的可靠性。
cs.CL / 8 / 2606.11206
Compatibility-Aware Dynamic Fine-Tuning for Large Language Models
兼容性意识的动态微调用于大型语言模型
Abstract
Supervised Fine-Tuning (SFT) is the predominant paradigm for aligning large language models (LLMs), yet it suffers from optimization instability and limited generalization. Recent work attributes this issue to pathological gradient scaling and proposes Dynamic Fine-Tuning (DFT) to correct it at the token level. However, DFT assumes all demonstrations are equally suitable learning targets, an assumption violated by the strong heterogeneity of large-scale instruction data, where demonstration-policy mismatch induces high-variance updates at the sample level. We introduce Compatibility-Aware Dynamic Fine-Tuning (CADFT), a principled extension of DFT that controls sample-level optimization variance. CADFT derives a dynamic, policy-dependent compatibility signal from model likelihoods to modulate supervised updates, suppressing high-variance gradients from incompatible demonstrations. We further propose a delayed, low-frequency compatibility-guided rewriting strategy to transform persistently incompatible demonstrations into learnable targets. We show that CADFT can be interpreted as a variance-controlled estimator that generalizes token-level stabilization in DFT to the sample level. Extensive experiments demonstrate improved stability, generalization, and cold-start reinforcement learning initialization, while remaining fully supervised and independent of explicit reward modeling.
Chinese Translation
监督微调（SFT）是对齐大型语言模型（LLMs）的主要范式，但它存在优化不稳定和泛化能力有限的问题。近期研究将这一问题归因于病态梯度缩放，并提出动态微调（DFT）以在标记级别进行修正。然而，DFT假设所有示例都是同样适合的学习目标，这一假设被大规模指令数据的强异质性所违反，其中示例与策略的不匹配在样本级别引发高方差更新。我们引入了兼容性意识的动态微调（CADFT），这是DFT的一个原则性扩展，旨在控制样本级别的优化方差。CADFT从模型似然中推导出动态的、依赖于策略的兼容性信号，以调节监督更新，抑制来自不兼容示例的高方差梯度。我们进一步提出了一种延迟的、低频率的兼容性引导重写策略，将持续不兼容的示例转化为可学习的目标。我们表明，CADFT可以被解释为一种方差控制的估计器，它将DFT中的标记级别稳定化推广到样本级别。大量实验表明，CADFT在稳定性、泛化能力和冷启动强化学习初始化方面都有所改善，同时仍然保持完全监督，并且不依赖于显式奖励建模。
cs.CL / 9 / 2606.11208
BioDivergence: A Benchmark and Evaluation Framework for Hidden Contextual Contradictions in Biomedical Abstracts
BioDivergence：生物医学摘要中隐藏上下文矛盾的基准与评估框架
Abstract
Biomedical findings often seem to conflict across studies, but many of these differences are context-dependent rather than true contradictions. Variations in cohort, geography, assay protocol, disease subtype, and clinical setting can make both claims locally valid. Existing NLI and scientific claim-verification benchmarks reduce such cases to entailment, contradiction, or neutral, failing to capture the contextual structure behind divergence. To address this, we introduce BioDivergence, an evaluation framework with a six-class conflict taxonomy, a 13-axis divergence ontology, and four structured outputs per claim pair: conflict type, divergence axes, dominant confounder, and reconciliation explanation. We release BioDivergence-Silver-v1.0, an article-disjoint silver benchmark of 11,865 claim pairs across five biomedical domains, alongside a legacy deduplicated variant for comparison. Results show notable ranking differences between the two variants, with the fine-tuned reference model dropping about 12 points under the article-disjoint setting, while Mistral-7B-Instruct-v0.3 achieves 0.5523 accuracy and 0.3894 contextual-F1 on the 842-example primary test set. BioDivergence offers a more faithful way to distinguish contextual divergence from direct contradiction and to separate article-level memorization from genuine task learning.
Chinese Translation
生物医学研究结果在不同研究之间常常似乎存在冲突，但这些差异往往是依赖于上下文的，而非真正的矛盾。队列、地理位置、检测协议、疾病亚型和临床环境的变化可以使得两种主张在局部上都是有效的。现有的自然语言推理（NLI）和科学主张验证基准将此类案例简化为蕴含、矛盾或中立，未能捕捉到背后冲突的上下文结构。为了解决这一问题，我们提出了BioDivergence，一个评估框架，包含六类冲突分类法、13个维度的分歧本体，以及每对主张的四个结构化输出：冲突类型、分歧维度、主要混淆因素和调和解释。我们发布了BioDivergence-Silver-v1.0，这是一个涵盖五个生物医学领域的11,865对主张的文章不重叠银基准，同时提供了一个遗留去重变体以供比较。结果显示，两种变体之间存在显著的排名差异，经过微调的参考模型在文章不重叠设置下下降了约12分，而Mistral-7B-Instruct-v0.3在842个样本的主要测试集上达到了0.5523的准确率和0.3894的上下文F1值。BioDivergence提供了一种更真实的方式来区分上下文分歧与直接矛盾，并将文章级别的记忆与真正的任务学习分开。
cs.CL / 10 / 2606.11209
ProcessThinker: Enhancing Multi-modal Large Language Models Reasoning via Rollout-based Process Reward
ProcessThinker：通过基于回滚的过程奖励增强多模态大型语言模型推理能力
Abstract
Visual question answering increasingly requires multi-step reasoning. Recent post-training with reinforcement learning under verifiable rewards (RLVR) and Group Relative Policy Optimization (GRPO) can improve multimodal reasoning, but most approaches rely on sparse outcome-only rewards. As a result, they struggle to tell whether an incorrect answer comes from a small mistake late in the reasoning or from an unhelpful trajectory from the start. A common solution is to train a process reward model (PRM) for step-level supervision, but this typically requires large-scale high-quality chain-of-thought annotations and additional training cost. We propose ProcessThinker, a practical post-training pipeline that provides step-level process rewards without training an explicit PRM. ProcessThinker first rewrites reasoning traces into a step-tagged format for cold-start supervised fine-tuning, then applies GRPO with a standard format reward and our rollout-based process reward. Concretely, for each intermediate step, we sample multiple continuations from that step and use the empirical success rate (final-answer verification) as the step reward. This gives dense credit assignment and encourages reasoning steps that more reliably support a correct conclusion, helping reduce inconsistent or self-contradictory progress across steps -- a key issue in logical reasoning. Across four challenging video benchmarks (Video-MMMU, MMVU, VideoMathQA, and LongVideoBench), ProcessThinker consistently improves over the baseline model Qwen3-VL-8B-Instruct
Chinese Translation
视觉问答越来越需要多步骤推理。近期在可验证奖励下的强化学习后训练（RLVR）和群体相对策略优化（GRPO）可以改善多模态推理，但大多数方法依赖于稀疏的结果奖励。因此，它们难以判断错误答案是由于推理后期的小错误还是从一开始就来源于不利的轨迹。一个常见的解决方案是训练一个过程奖励模型（PRM）以进行步骤级监督，但这通常需要大规模高质量的思维链注释和额外的训练成本。我们提出了ProcessThinker，一个实用的后训练管道，可以在不训练显式PRM的情况下提供步骤级过程奖励。ProcessThinker首先将推理轨迹重写为带步骤标签的格式，以进行冷启动的监督微调，然后应用GRPO，结合标准格式奖励和我们的基于回滚的过程奖励。具体而言，对于每个中间步骤，我们从该步骤采样多个延续，并使用经验成功率（最终答案验证）作为步骤奖励。这提供了密集的信用分配，并鼓励更可靠地支持正确结论的推理步骤，从而帮助减少步骤之间不一致或自相矛盾的进展——这是逻辑推理中的一个关键问题。在四个具有挑战性的视频基准测试（Video-MMMU、MMVU、VideoMathQA和LongVideoBench）中，ProcessThinker始终优于基线模型Qwen3-VL-8B-Instruct。
cs.CL / 11 / 2606.11210
T2MM: An LLM Supported Architecture For Inquiry-Based Modeling
T2MM：一种支持基于询问的建模的LLM架构
Abstract
Model Construction is a foundational practice in science learning that relies on visualization and interactivity. Large Language Models, increasingly augmented with multimodal capabilities, have been integrated in education contexts to support learning. However, these tools lack visual interactivity that is required by some learning contexts. We introduce Text to Multimodal Model (T2MM), a robust, dynamic LLM supported architecture that assists in model construction within the open inquiry ecology-based modeling software Virtual Experimental Research Assistant (VERA). T2MM accounts for the current context of the learner's model and creates interactive models, rather than static images, enabling the model to remain responsive to manual adjustment. To measure technical feasibility, we evaluate T2MM through a custom procedurally generated dataset of natural language learner modeling requests and target models within the VERA system. T2MM outperforms a baseline model generation architecture implemented through LLM-supported full code generation, common in the literature, across all measured success metrics. Our contribution not only outlines LLM integration into a inquiry-based learning modeling tool, but also describes a possible architecture through which more interactive multimodal LLM tools can be created.
Chinese Translation
模型构建是科学学习中的一项基础实践，依赖于可视化和交互性。大型语言模型（LLM）越来越多地与多模态能力结合，已在教育环境中被整合以支持学习。然而，这些工具缺乏某些学习环境所需的视觉交互性。我们介绍了文本到多模态模型（Text to Multimodal Model, T2MM），这是一种强大且动态的LLM支持架构，旨在辅助在开放询问生态基础建模软件虚拟实验研究助手（Virtual Experimental Research Assistant, VERA）中进行模型构建。T2MM考虑了学习者模型的当前上下文，并创建交互式模型，而非静态图像，使模型能够对手动调整保持响应。为了评估技术可行性，我们通过在VERA系统中生成的自然语言学习者建模请求和目标模型的自定义程序生成数据集对T2MM进行评估。T2MM在所有测量的成功指标上均优于通过LLM支持的全代码生成实现的基线模型生成架构，这在文献中较为常见。我们的贡献不仅概述了LLM在基于询问的学习建模工具中的整合，还描述了一种可能的架构，通过该架构可以创建更具交互性的多模态LLM工具。
cs.CL / 12 / 2606.11211
Calibration Drift Under Reasoning: How Chain-of-Thought Budgets Induce Overconfidence in Large Language Models
推理下的校准漂移：链式思维预算如何导致大型语言模型的过度自信
Abstract
The ability of large language models (LLMs) to express calibrated uncertainty is important for safe deployment. Chain-of-thought (CoT) reasoning is widely used to improve accuracy and reliability, but its effect on calibration is not fully understood. We show that this picture is incomplete: in some settings, increasing the reasoning budget beyond a task-specific threshold can cause models to become systematically overconfident, assigning high confidence to incorrect answers. We call this phenomenon Calibration Drift Under Reasoning (CDUR) and study it both theoretically and empirically. We define reasoning budget B and analyze conditions under which Expected Calibration Error ECE(B) follows a non-monotonic pattern: it first decreases as reasoning corrects errors, then increases as longer reasoning produces internally consistent but incorrect explanations. We propose a Hypothesis Lock-In model based on autoregressive generation to explain this behavior. We evaluate Llama-3.1-8B and Llama-3.3-70B on 47 reasoning-trap questions across four reasoning budgets and three seeds (1,368 API calls; 574 valid responses). The 8B model shows non-monotonic calibration behavior, while results for the 70B model are limited to baseline evaluation and are inconclusive for budget-dependent effects. We introduce CABStop, a calibration-aware stopping rule that halts reasoning when confidence diverges from an auxiliary accuracy estimate. These results suggest that increasing reasoning depth does not always improve reliability and should be monitored carefully.
Chinese Translation
大型语言模型（LLMs）表达校准不确定性的能力对于安全部署至关重要。链式思维（CoT）推理被广泛应用于提高准确性和可靠性，但其对校准的影响尚未完全理解。我们展示了这一观点的不完整性：在某些情况下，推理预算超出特定任务阈值后，模型可能会系统性地变得过度自信，给错误答案分配高置信度。我们将这一现象称为推理下的校准漂移（Calibration Drift Under Reasoning, CDUR），并从理论和实证两个方面进行了研究。我们定义了推理预算 B，并分析了期望校准误差（Expected Calibration Error, ECE(B)）呈现非单调模式的条件：它首先随着推理纠正错误而减少，然后随着更长推理产生内部一致但错误的解释而增加。我们提出了一种基于自回归生成的假设锁定模型（Hypothesis Lock-In model）来解释这一行为。我们在四个推理预算和三个种子（1,368 次 API 调用；574 个有效响应）上评估了 Llama-3.1-8B 和 Llama-3.3-70B 在 47 个推理陷阱问题上的表现。8B 模型显示出非单调的校准行为，而 70B 模型的结果仅限于基线评估，对于预算依赖效应的结论则不明确。我们引入了 CABStop，一种校准感知的停止规则，当置信度与辅助准确性估计出现偏差时停止推理。这些结果表明，增加推理深度并不总是能提高可靠性，需谨慎监控。
cs.CL / 13 / 2606.11212
EverydayGPT: Confidence-Gated Routing for Efficient and Safe Hybrid GPT-RAG Conversational QA
EverydayGPT：基于置信度门控路由的高效安全混合GPT-RAG对话问答系统
Abstract
Standard Retrieval-Augmented Generation (RAG) pipelines route every query through retrieval and generation unconditionally, incurring unnecessary computation and propagating low-quality context to the generator. We introduce EverydayGPT, a lightweight conversational QA system built around a Confidence-Gated Routing (CGR) mechanism that formalises the routing decision as a joint policy over retrieval distance and extraction adequacy. The backbone is a 205M-parameter GPT trained from scratch on 10B tokens of FineWeb-Edu. CGR avoids invoking the costly GPT pathway (~5.9s) for 85 percent of queries by resolving them via fast RAG extraction (~45 ms), yielding over 120x latency reduction on the majority of queries while maintaining answer quality. On a 500-question in-domain benchmark, the system achieves F1 = 0.226 +/- 0.004 compared to 0.171 for GPT-only and 0.210 for unconditional RAG. Gains over strong baselines are modest but consistent, while efficiency improvements are substantial (6.3x mean latency reduction). A structured grounding audit finds no unsupported claims in the sampled set, with explicit scope limitations. We position this work as a study of routing strategies under resource constraints rather than a claim of state-of-the-art performance.
Chinese Translation
标准的检索增强生成（RAG）管道无条件地将每个查询通过检索和生成进行路由，这导致不必要的计算并将低质量的上下文传播给生成器。我们提出了EverydayGPT，一个围绕置信度门控路由（CGR）机制构建的轻量级对话问答系统，该机制将路由决策形式化为对检索距离和提取充分性的联合策略。其核心是一个205M参数的GPT，基于10B个FineWeb-Edu令牌从头开始训练。CGR通过快速的RAG提取（约45毫秒）解决85%的查询，从而避免调用耗时的GPT路径（约5.9秒），在大多数查询中实现超过120倍的延迟减少，同时保持答案质量。在一个包含500个问题的领域内基准测试中，该系统的F1得分为0.226 +/- 0.004，而仅使用GPT的得分为0.171，无条件RAG的得分为0.210。与强基线相比，性能提升虽然适度但一致，而效率的改善则显著（平均延迟减少6.3倍）。结构化的基础审计发现抽样集内没有不支持的声明，并明确了范围限制。我们将这项工作定位为在资源限制下的路由策略研究，而非声称达到最先进的性能。
cs.CL / 14 / 2606.11213
Beyond Compaction: Structured Context Eviction for Long-Horizon Agents
超越压缩：长时间跨度代理的结构化上下文驱逐
Abstract
We present Context Window Lifecycle (CWL), a context-management scheme that gives long-horizon LLM agents an effectively unbounded working horizon. As a session accumulates history, CWL keeps the context within budget through graduated, semantically-aware eviction: the agent annotates its trajectory as typed, dependency-linked episodes as work proceeds, and a deterministic, LLM-free policy evicts content in priority order within that structure when a token budget is exceeded. CWL preserves user turns and the exploratory context the agent is actively reasoning over, while aggressively shedding action episodes whose effects are already persisted in the environment, keeping active context near a stable ceiling that also avoids the performance degradation associated with very large prompts. Compared to summarization-based compaction, CWL avoids four well-known limitations: unpredictable lossiness, destruction of causal structure, blocking model cost, and compression-induced hallucination. Compared to recency truncation, CWL is semantically aware: it drops the oldest-and-most-recoverable content according to the dependency graph rather than oldest-in-time regardless of relevance. We describe the annotation protocol, the episode graph, the eviction policy, and the token-accounting loop, and evaluate CWL on long-horizon agentic benchmarks: a single agent session completing 89 sequential tasks across 80 million tokens with no measurable degradation in task accuracy relative to per-task isolated sessions
Chinese Translation
我们提出了上下文窗口生命周期（Context Window Lifecycle, CWL），这是一种上下文管理方案，使长时间跨度的LLM代理拥有有效的无限工作范围。随着会话历史的积累，CWL通过逐步的、语义感知的驱逐来保持上下文在预算之内：代理在工作进行时将其轨迹标注为类型化、依赖关联的事件，并在超出令牌预算时，采用确定性、无LLM的策略按照优先顺序在该结构内驱逐内容。CWL保留用户的发言和代理正在积极推理的探索上下文，同时积极剔除那些其效果已在环境中持久化的行动事件，保持活跃上下文接近一个稳定的上限，从而避免与非常大提示相关的性能下降。与基于摘要的压缩相比，CWL避免了四个众所周知的局限性：不可预测的丢失、因果结构的破坏、模型成本的阻塞以及压缩引发的幻觉。与近期截断相比，CWL具有语义感知能力：它根据依赖图丢弃最旧且最可恢复的内容，而不是无视相关性地丢弃时间上最旧的内容。我们描述了标注协议、事件图、驱逐策略和令牌核算循环，并在长时间跨度的代理基准上评估CWL：一个单一代理会话在8000万令牌中完成89个连续任务，相较于每个任务的孤立会话没有可测量的任务准确性下降。
cs.CL / 15 / 2606.11219
Afrispeech Semantics: Evaluating Audio Semantic Reasoning in Spoken Language Models Across Domains and Accents
Afrispeech 语义：评估跨领域和口音的语音语言模型中的音频语义推理
Abstract
Audio language models (ALMs) are increasingly used for speech-based understanding, yet their ability to perform semantic reasoning beyond transcription, Text-to-Audio Retrieval, Captioning, and Question-Answering accuracy remains insufficiently benchmarked. In particular, the effects of accent variation, domain shift, and semantic over-inference on audio reasoning are poorly understood. We evaluate audio language models across five semantic and paralinguistic reasoning tasks: entailment, consistency, plausibility, accent drift, and accent restraint. Collectively, these tasks assess a model's ability to reason over spoken audio as the primary evidence source, including whether a textual hypothesis can be inferred, contradicted, or left undetermined by the audio, whether statements align or conflict with spoken content, whether claims are plausible given the discourse, and whether model predictions remain stable or appropriately constrained across accent variation. These findings highlight critical limitations in current audio reasoning evaluations and hope to provide guidance for more robust and equitable ALM design and assessment
Chinese Translation
音频语言模型（ALMs）在基于语音的理解中越来越多地被使用，但它们在转录、文本到音频检索、字幕生成和问答准确性之外进行语义推理的能力仍然缺乏充分的基准评估。特别是，口音变化、领域转移和语义过度推断对音频推理的影响尚不清楚。我们在五个语义和副语言推理任务中评估音频语言模型：蕴涵、一致性、合理性、口音漂移和口音约束。这些任务共同评估模型在将口语音频作为主要证据来源时的推理能力，包括文本假设是否可以被音频推断、反驳或保持不确定，陈述是否与口语内容一致或冲突，考虑到话语时主张是否合理，以及模型预测在口音变化中是否保持稳定或适当约束。这些发现突显了当前音频推理评估中的关键局限性，并希望为更稳健和公平的音频语言模型设计与评估提供指导。
cs.CL / 16 / 2606.11220
LifeSentence: Language models can encode human life course trajectories from longitudinal panel data
LifeSentence：语言模型能够从纵向面板数据中编码人类生命历程轨迹
Abstract
Forecasting human life outcomes is important to gain insights into how individuals attain long and healthy lives. Conventional statistical approaches yield limited accuracy, potentially due to discarding the sequential structure of the life course. Modern methods such as transformer architectures require large scale training data that most longitudinal panel studies lack. Here we introduce LifeSentence, a model for life-course reasoning that bridges large language models with longitudinal panel data. By representing each life event as a structured natural-language record and instruction-tuning a pretrained 24-billion-parameter language model across an 18-task evaluation taxonomy spanning prediction, robustness and reasoning, LifeSentence supplements panel data with distributional knowledge already encoded during pretraining. Trained on approximately 65,000 individuals from the German Socio-Economic Panel - roughly 45 times fewer than prior transformer-based approaches - LifeSentence outperforms classical and deep learning baselines across all task families, achieving a threefold improvement in joint event-and-timing prediction from best baselines and 91.2% Kendall's tau when reconstructing chronological order from timestamp-stripped event sets. Without explicit supervision, the model recovers documented patterns of social stratification, including the education premium, the gender wage gap and the motherhood penalty, from discrete event sequences alone. A natural-language interface further enables qualitatively new research queries, such as connecting an early-life history to a specified late-life endpoint, establishing LifeSentence as both a predictive tool and a probe for counterfactual exploration of human biographies.
Chinese Translation
预测人类生命结果对于深入了解个体如何获得长久健康的生活至关重要。传统统计方法的准确性有限，可能是由于忽视了生命历程的顺序结构。现代方法如变换器架构需要大规模的训练数据，而大多数纵向面板研究缺乏这样的数据。在此，我们介绍了LifeSentence，这是一种生命历程推理模型，旨在将大型语言模型与纵向面板数据相结合。通过将每个生命事件表示为结构化的自然语言记录，并对一个预训练的240亿参数的语言模型进行指令调优，该模型涵盖了跨越预测、鲁棒性和推理的18项任务评估分类，LifeSentence补充了在预训练过程中已经编码的分布知识。该模型在大约65,000名来自德国社会经济面板的个体上进行训练，约为先前基于变换器的方法的45倍，LifeSentence在所有任务类别中均优于经典和深度学习基准，在最佳基准的联合事件和时间预测中实现了三倍的改进，并在从去除时间戳的事件集重建时间顺序时达到了91.2%的肯德尔tau系数。在没有明确监督的情况下，该模型仅从离散事件序列中恢复了社会分层的已记录模式，包括教育溢价、性别工资差距和母亲惩罚。自然语言接口进一步使得定性的新研究查询成为可能，例如将早期生活历史与特定的晚年终点连接，从而确立了LifeSentence作为一种预测工具和人类传记反事实探索的探针。
cs.CL / 17 / 2606.11222
A Geometric Profile of Semantic Information in Text: Frame-Conditional Uniqueness and a Trade-Off Triangle for Scalar Summaries
文本中语义信息的几何轮廓：框架条件唯一性与标量摘要的权衡三角形
Abstract
How much meaning does a text carry? Shannon's theory measures uncertainty over symbols and is intentionally indifferent to meaning, while pairwise metrics such as BERTScore compare two texts rather than characterizing one. We develop a geometric framework that measures semantic content from the structure of a text's sentence embeddings. The framework has three parts. First, within a fixed embedding and baseline, six natural axioms uniquely determine a scalar measure up to scale, a frame-conditional uniqueness theorem. The resulting scalar is empirically too coarse, motivating a richer representation. Second, we propose a three-coordinate semantic profile capturing novelty (displacement from generic discourse), breadth (diversity of distinct ideas), and integration (connectedness among them), together with a discrete minimal unit (the semantic quantum) whose resolution is fixed by a clustering threshold $\tau$. Third, we prove a no-go theorem: no scalar summary of the profile can simultaneously satisfy analytic stability under paraphrase and concatenation, ordinal robustness across text scales, and cross-representation comparability. We exhibit two practical scalars, $S_{\mathrm{minmax}}$ and $S_{\mathrm{rank}}$, each occupying a distinct corner of this trade-off triangle. Validation across 23 synthetic categories, 5 Project Gutenberg novels, and 3 embedding models confirms the trade-off. The recommended rank-normalized configuration passes 25 of 28 ordinal checks as point estimates (21 of 28 after Benjamini-Hochberg correction), outperforming seven baselines including unigram entropy and a BERTScore-based novelty signal. A separate variational result connects the breadth coordinate to the log-determinant of a determinantal point process (Spearman $\rho = 0.985$ over 507 Gutenberg chapters), giving an optimization-theoretic foundation for breadth.
Chinese Translation
文本承载了多少意义？香农理论测量符号的不确定性，并故意对意义保持中立，而像 BERTScore 这样的成对度量则比较两个文本，而不是对一个文本进行表征。我们开发了一个几何框架，从文本句子嵌入的结构中测量语义内容。该框架分为三个部分。首先，在固定的嵌入和基线下，六个自然公理唯一确定一个标量度量（至规模），即框架条件唯一性定理。所得标量在经验上过于粗糙，促使我们寻求更丰富的表征。其次，我们提出了一个三坐标语义轮廓，捕捉新颖性（与通用话语的偏离）、广度（不同思想的多样性）和整合性（它们之间的连通性），以及一个离散的最小单位（语义量子），其分辨率由聚类阈值 $ au$ 固定。第三，我们证明了一个不可行定理：该轮廓的任何标量摘要都无法同时满足在意译和连接下的分析稳定性、跨文本尺度的序数鲁棒性以及跨表征的可比性。我们展示了两个实用的标量 $S_{ ext{minmax}}$ 和 $S_{ ext{rank}}$，它们分别位于这个权衡三角形的不同角落。在 23 个合成类别、5 部古腾堡计划小说和 3 个嵌入模型上的验证确认了这一权衡。推荐的秩归一化配置在 28 项序数检查中通过了 25 项（经过 Benjamini-Hochberg 校正后为 21 项），优于包括一元熵和基于 BERTScore 的新颖性信号在内的七个基线。一个单独的变分结果将广度坐标与行列式点过程的对数行列式联系起来（在 507 个古腾堡章节中 Spearman $
ho = 0.985$），为广度提供了优化理论基础。
cs.CL / 18 / 2606.11232
Every Act Has Its Price: Compressed Moral Composition in Frontier LLMs
每个行为都有其代价：前沿大型语言模型中的压缩道德构成
Abstract
Existing LLM moral benchmarks usually ask which isolated moral act, value, or foundation a model prefers. This is useful but incomplete. Realistic judgments often require a model to combine several moral signals within the same option. We introduce **Moral Trolley Arena**, a two-stage blind ELO benchmark for measuring how LLMs compose moral evidence. The single-scene arena first calibrates individual moral acts from a 229-scenario corpus across five Moral Foundations Theory foundations; the composite arena then combines calibrated acts into two-act moral items over a controlled intensity grid and measures the resulting composite preferences. Across ten frontier models, composite judgments are largely predicted by component act strength, but the relation is consistently compressed rather than simply additive. Models also show non-additive intensity anchoring, bounded foundation-specific residuals after component control, and highly convergent composite preference surfaces across providers. These results suggest that moral audits should measure composition rules for moral evidence, not only rankings over isolated acts.
Chinese Translation
现有的大型语言模型（LLM）道德基准通常询问模型偏好哪种孤立的道德行为、价值或基础。这种方法虽然有用，但并不完整。现实的判断往往需要模型在同一选项中结合多个道德信号。我们引入了**道德电车竞技场（Moral Trolley Arena）**，这是一个两阶段盲测的ELO基准，用于测量LLM如何组合道德证据。单场景竞技场首先从一个包含229个场景的语料库中，根据五个道德基础理论（Moral Foundations Theory）基础校准个体道德行为；复合竞技场则将校准后的行为组合成两个行为的道德项目，并在一个受控强度网格上测量结果的复合偏好。在十个前沿模型中，复合判断在很大程度上由组成行为的强度预测，但这种关系始终是压缩的，而非简单的加法。模型还表现出非加性的强度锚定、在控制组成后有限的基础特定残差，以及跨提供者高度趋同的复合偏好表面。这些结果表明，道德审计应测量道德证据的组合规则，而不仅仅是孤立行为的排名。
cs.CL / 19 / 2606.11257
Energy-Efficient On-Device RAG on a Mobile NPU: System Design and Benchmark on Snapdragon X Elite
移动 NPU 上的节能设备内检索增强生成 (RAG)：系统设计与 Snapdragon X Elite 的基准测试
Abstract
Retrieval-Augmented Generation (RAG) pipelines are compute-intensive, combining embedding, retrieval, reranking, and large language model (LLM) generation. Running them entirely on-device benefits privacy, latency, and offline use, but the energy cost of CPU inference is a major barrier. We present what is, to our knowledge, the first end-to-end RAG pipeline that runs all neural stages -- embedding, reranking, and LLM generation -- on the Qualcomm Hexagon NPU of the Snapdragon X Elite. Profiling on a Dell XPS 13 laptop, we compare NPU-accelerated RAG against CPU and OpenCL/Adreno GPU baselines on indexing and query workloads. On indexing, the NPU achieves 9.1x higher embedding throughput and 12.3x less system energy. On a 120-query Wikipedia-passage benchmark, it delivers 18.1x faster LLM prefilling, 4.0x lower end-to-end query latency, and 4.0x less system energy than the CPU baseline; the same workload on the integrated GPU is 1.7x slower than CPU and uses 6.5x more energy than the NPU. A GPT-4.1 LLM-as-judge evaluation finds NPU answer quality on par with CPU and GPU within evaluator noise (mean 9.32 vs. 8.95 vs. 9.03 on a 1-10 rubric), with 86.7% of queries scoring identically across all three backends. On the Snapdragon X Elite / Hexagon class of laptop SoC, the NPU thus enables practical, energy-efficient on-device RAG without quality regression -- a sustainable path toward green edge intelligence that we expect to generalize to comparable mobile NPUs (Apple Neural Engine, Intel NPU, MediaTek APU) as their software stacks mature.
Chinese Translation
检索增强生成 (RAG) 流水线计算密集，结合了嵌入、检索、重排序和大型语言模型 (LLM) 生成。将其完全在设备上运行有利于隐私、延迟和离线使用，但 CPU 推理的能耗是一个主要障碍。我们提出了目前为止首个在 Snapdragon X Elite 的高通 Hexagon NPU 上运行所有神经阶段——嵌入、重排序和 LLM 生成的端到端 RAG 流水线。在戴尔 XPS 13 笔记本电脑上进行的分析中，我们将 NPU 加速的 RAG 与 CPU 和 OpenCL/Adreno GPU 基准进行比较，针对索引和查询工作负载。在索引方面，NPU 实现了 9.1 倍的嵌入吞吐量提升和 12.3 倍的系统能耗降低。在 120 查询的维基百科段落基准测试中，它提供了 18.1 倍的 LLM 预填充速度，4.0 倍更低的端到端查询延迟，以及比 CPU 基准低 4.0 倍的系统能耗；在集成 GPU 上执行相同工作负载的速度比 CPU 慢 1.7 倍，能耗比 NPU 高出 6.5 倍。GPT-4.1 LLM 作为评估者的评估发现，NPU 的回答质量与 CPU 和 GPU 在评估者噪声范围内相当（在 1-10 的评分标准中，平均分为 9.32 vs. 8.95 vs. 9.03），其中 86.7% 的查询在所有三个后端上得分相同。因此，在 Snapdragon X Elite / Hexagon 类笔记本 SoC 上，NPU 实现了实用的、节能的设备内 RAG，而没有质量退化——这为绿色边缘智能提供了一条可持续的路径，我们预计这一方法能够推广到类似的移动 NPU（如 Apple Neural Engine、Intel NPU、MediaTek APU），随着其软件栈的成熟。
cs.CL / 20 / 2606.11316
Sch\"utzen: Evaluating LLM Safety in Bulgarian and German Contexts
Sch"utzen：评估保加利亚和德国背景下的大型语言模型安全性
Abstract
Large language models are increasingly deployed across professional domains, bringing hard-to-predict risks, including the generation of harmful or disrespectful content. Although substantial progress has been made in developing safety evaluation datasets, existing resources remain overwhelmingly English- and Chinese-centric. This limitation is particularly pronounced when evaluating languages that operate within shared sociocultural, legal, and ethical contexts. To address this gap, we introduce Sch\"{u}tzen: a German--Bulgarian safety dataset designed to assess model answerability under risk, covering both a low-resource language (Bulgarian) and a high-resource language (German). Experiments with multilingual and language-specific LLMs reveal pronounced cross-language differences in safety behavior, highlighting the necessity of tailored, region-specific evaluation resources to support the responsible deployment of LLMs in Germany and Bulgaria. Datasets and code are available at https://github.com/xnlp-lab/Schutzen. Warning: this paper contains examples that may be offensive, harmful, or biased.
Chinese Translation
大型语言模型在各专业领域的应用日益广泛，带来了难以预测的风险，包括生成有害或不尊重的内容。尽管在开发安全评估数据集方面取得了显著进展，但现有资源仍然以英语和中文为主。这一局限性在评估在共享社会文化、法律和伦理背景下运作的语言时尤为明显。为了解决这一问题，我们推出了Sch"{u}tzen：一个旨在评估模型在风险下回答能力的德语-保加利亚安全数据集，涵盖了一种低资源语言（保加利亚语）和一种高资源语言（德语）。对多语言和特定语言的大型语言模型的实验揭示了安全行为在不同语言之间的显著差异，强调了为支持大型语言模型在德国和保加利亚的负责任部署而需要量身定制的地区特定评估资源。数据集和代码可在https://github.com/xnlp-lab/Schutzen获取。警告：本文包含可能令人反感、有害或偏见的示例。
cs.CL / 21 / 2606.11350
When More Documents Hurt RAG: Mitigating Vector Search Dilution with Domain-Scoped, Model-Agnostic Retrieval
当更多文档影响 RAG：通过领域范围、模型无关的检索来减轻向量搜索稀释
Abstract
Retrieval-augmented generation degrades when scaled to large, heterogeneous document collections, where dense similarity loses discriminative power, and top-k retrieval increasingly returns semantically similar but contextually incorrect chunks. We refer to this failure mode as vector search dilution. Even when using hybrid dense+sparse retrieval, we observed this firsthand in a deployed Wyoming Department of Transportation corpus, where scaling from 54 to 1,128 documents (88,907 chunks) reduced accuracy from 75% to below 40%. To address this dilution, we propose MASDR-RAG ( Multi-Agent Scoped Domain Retrieval for RAG) and evaluate it on 200 expert-validated queries across five LLM backbones, six corpora, and two index stacks. Our results indicate that domain scoping using organizational metadata is the key fix, significantly improving P@10 from 0.77 to 0.86 ($p < 0.05$). Furthermore, our investigation of multi-agent orchestration revealed that a high degree of configuration dependence results --creating what we call the precision-faithfulness paradox. Based on these varied outcomes, our practical recommendation is simple: scope first, then perform a single synthesis call, reserving full multi-agent orchestration for genuinely multi-domain corpora paired with native-tool-call backbones. Code and Data will be made public upon acceptance.
Chinese Translation
当检索增强生成（RAG）扩展到大型异构文档集合时，其性能会下降，此时密集相似性失去辨别能力，且前 k 个检索结果越来越多地返回语义相似但上下文不正确的片段。我们将这种失败模式称为向量搜索稀释。即使使用混合的稠密+稀疏检索，我们在部署的怀俄明州交通运输部门语料库中亲身观察到了这一点，当文档数量从 54 增加到 1,128（88,907 个片段）时，准确率从 75% 降低到 40% 以下。为了解决这种稀释问题，我们提出了 MASDR-RAG（多代理范围领域检索用于 RAG），并在五个大型语言模型（LLM）基础上、六个语料库和两个索引堆栈上评估了其效果。我们的结果表明，使用组织元数据进行领域范围划定是关键修复，显著提高了 P@10 从 0.77 到 0.86（$p < 0.05$）。此外，我们对多代理协调的调查揭示了高度的配置依赖性，产生了我们称之为精确性-忠实性悖论的现象。基于这些不同的结果，我们的实际建议很简单：首先进行范围划定，然后执行单次综合调用，将完整的多代理协调保留给真正的多领域语料库，并与原生工具调用基础相结合。代码和数据将在接受后公开。
cs.CL / 22 / 2606.11371
The Dynamics of Human and AI-Generated Language: How Semantics Fluctuates across Different Timescales
人类与人工智能生成语言的动态：语义如何在不同时间尺度上波动
Abstract
Spoken language, whether produced by humans or large language models (LLM), unfolds over time with varying semantic content. However, we still lack simple, interpretable time-series features that capture how generic versus specific content is distributed over time, and that can be used to compare human and AI-generated speech. We introduce a semantic-timescale analysis pipeline that turns word-level transcripts with timestamps into semantic time-series. For each spoken narrative, we compute (i) semantic specificity using WordNet-based word depth and (ii) contextual similarity using SBERT embeddings and quantify their temporal dependence using autocorrelation-window measures (ACW-0 and related metrics). We then compare original speech to multiple shuffled controls that selectively disrupt lexical identity, temporal order, and word duration. Across human-read autobiographical narratives, TTS readings, and LLM-generated texts rendered with TTS, we find that segments with longer ACW-0 in the semantic time-series tend to contain more generic vocabulary, whereas segments with shorter ACW-0 are enriched in more specific words. These associations are strongly attenuated or abolished when word order and timing are randomized, indicating that ACW-based measures capture non-trivial temporal organization of semantic content beyond static lexical distributions. Our results suggest that ACW-based semantic timescales are a useful family of features for analyzing and comparing the temporal structure of human and AI-generated speech.
Chinese Translation
口语，无论是由人类还是大型语言模型（LLM）生成的，都是随着时间展开的，具有不同的语义内容。然而，我们仍然缺乏简单且可解释的时间序列特征，以捕捉通用内容与特定内容在时间上的分布，并用于比较人类与人工智能生成的言语。我们引入了一种语义时间尺度分析流程，将带有时间戳的单词级转录文本转化为语义时间序列。对于每个口述叙事，我们计算（i）基于WordNet的词深度的语义特异性，以及（ii）使用SBERT嵌入的上下文相似性，并利用自相关窗口度量（ACW-0及相关指标）量化它们的时间依赖性。然后，我们将原始言语与多种打乱控制进行比较，这些控制选择性地破坏词汇身份、时间顺序和单词持续时间。在人类朗读的自传叙事、TTS（文本转语音）朗读和使用TTS生成的LLM文本中，我们发现语义时间序列中具有较长ACW-0的片段往往包含更多通用词汇，而具有较短ACW-0的片段则富含更具体的词汇。当词序和时序被随机化时，这些关联会显著减弱或消失，表明基于ACW的度量捕捉到了超越静态词汇分布的语义内容的非平凡时间组织。我们的结果表明，基于ACW的语义时间尺度是一系列有用的特征，可用于分析和比较人类与人工智能生成言语的时间结构。
cs.CL / 23 / 2606.11375
When Probing Accuracy Saturates, Fragility Resolves: A Complementary Metric for LLM Pre-Training Analysis
当探测准确率饱和时，脆弱性得到解决：一种用于大规模语言模型预训练分析的补充指标
Abstract
Standard linear probing declares a property "encoded" when a classifier on hidden states achieves high accuracy. The protocol works well on a snapshot but breaks across pre-training: probe accuracy saturates within the first few thousand steps, leaving most of training invisible to the instrument. We introduce fragility, a complementary per-layer metric defined as the activation-noise level at which probe accuracy collapses. Fragility is sensitive to both the margin of separability and the redundancy of representation, both of which keep evolving long after accuracy plateaus. Applied to open-checkpoint language models, fragility recovers structure that accuracy alone cannot see. Moralized representations emerge along a lexical $\to$ compositional gradient: lexical moral detection first, compositional moral encoding later. Because probe accuracy on its own tracks how lexically separable a dataset is, we establish the compositional encoding directly, by showing it transfers across construction types that share no contrast tokens. A layer-depth robustness gradient develops monotonically across training while accuracy stays flat. And matched fine-tuning corpora that produce identical probing accuracy leave distinct fragility fingerprints, showing that data curation reshapes probe robustness without changing probe accuracy. In every comparison we test, where probing accuracy returns a flat answer, fragility returns a structured one.
Chinese Translation
标准线性探测在分类器对隐藏状态的准确率达到高水平时宣称某一属性“已编码”。该协议在快照上效果良好，但在预训练过程中失效：探测准确率在最初的几千步内饱和，使得大部分训练对该工具而言是不可见的。我们引入了脆弱性（fragility），这是一种补充的逐层指标，定义为探测准确率崩溃时的激活噪声水平。脆弱性对可分离性边际和表示的冗余度都非常敏感，而这两者在准确率达到平台期后仍在不断演变。应用于开放检查点语言模型，脆弱性恢复了准确率无法看到的结构。道德化表示沿着词汇到组合的梯度出现：首先是词汇道德检测，随后是组合道德编码。由于探测准确率本身追踪的是数据集的词汇可分离性，我们通过展示其在没有对比标记的构造类型之间的转移，直接建立了组合编码。在训练过程中，层深度的鲁棒性梯度单调发展，而准确率保持平坦。产生相同探测准确率的匹配微调语料库留下了不同的脆弱性指纹，显示数据策划在不改变探测准确率的情况下重塑了探测鲁棒性。在我们测试的每一个比较中，当探测准确率返回平坦答案时，脆弱性则返回结构化的答案。
cs.CL / 24 / 2606.11386
Overcoming State Inertia in Full-Duplex Spoken Language Models via Activation Steering
通过激活引导克服全双工语言模型中的状态惯性
Abstract
Full-duplex spoken language models (FD-SLMs) enable seamless speech interaction by allowing models to listen and speak simultaneously, yet the internal mechanism by which they coordinate listening and speaking remains underexplored. We analyze the predictive behavior encoded in FD-SLM hidden representations and find that they exhibit stream-specific predictive patterns: during listening, they preferentially predict the incoming user stream, whereas during speaking, they preferentially predict the model output stream. Building on this observation, we show that FD-SLMs dynamically modulate their internal predictive focus between two states: a generative state aligned with model output generation and a perceptive state aligned with incoming user input. However, this modulation can lag behind abrupt changes in conversational context. During user interruptions, the model remains transiently biased toward the generative state before transitioning into the perceptive state, causing it to miss the beginning of the incoming input. We term this delayed internal transition state inertia. To quantify its downstream impact, we introduce the Zero-Buffer Benchmark (ZBB), a diagnostic benchmark for evaluating immediate interruption comprehension when user speech begins abruptly. We evaluate this setting using response correctness and initial-word occurrence rate (IWOR). Finally, we mitigate state inertia through activation steering with a perception vector, a training-free intervention with little additional computational overhead. Across multiple state-of-the-art FD-SLMs, activation steering substantially improves interruption handling; for example, on PersonaPlex, it improves correctness from 28% to 45% and IWOR from 40% to 72% without any fine-tuning.
Chinese Translation
全双工语言模型（FD-SLMs）通过允许模型同时听取和发言，实现了无缝的语音交互，但它们协调听与说的内部机制仍然未被充分探索。我们分析了FD-SLM隐藏表示中编码的预测行为，发现它们表现出特定于流的预测模式：在听的过程中，它们优先预测即将到来的用户流，而在说的过程中，它们优先预测模型输出流。基于这一观察，我们展示了FD-SLMs在两个状态之间动态调节其内部预测焦点：一个与模型输出生成对齐的生成状态和一个与即将到来的用户输入对齐的感知状态。然而，这种调节可能滞后于对话上下文的突变。在用户打断期间，模型在过渡到感知状态之前，暂时偏向于生成状态，导致其错过即将到来的输入的开始。我们将这种延迟的内部过渡称为状态惯性。为了量化其下游影响，我们引入了零缓冲基准（Zero-Buffer Benchmark，ZBB），这是一个用于评估用户语音突然开始时的即时打断理解的诊断基准。我们使用响应正确性和初始词出现率（Initial-Word Occurrence Rate，IWOR）来评估这一设置。最后，我们通过激活引导与感知向量来缓解状态惯性，这是一种无需训练的干预，几乎没有额外的计算开销。在多个最先进的FD-SLM中，激活引导显著改善了对打断的处理；例如，在PersonaPlex上，其正确性从28%提高到45%，IWOR从40%提高到72%，且无需任何微调。
cs.CL / 25 / 2606.11387
Small Experiments, Cheaper Decisions: A Case Study in Staged Promotion for Micro-Pretraining
小规模实验，低成本决策：微预训练阶段性推广的案例研究
Abstract
Short pretraining runs can reduce experimental cost, but they can also over-promote configurations that only look strong at tiny budgets. We study an auditable staged-promotion protocol for a fixed micro-pretraining runner on two heterogeneous host blocks: Windows A100 and Linux L40S. Starting from twelve prior-screened configurations, we use staged budgets of 2 minutes, 5 minutes, 10 minutes, 60 minutes, and 12 hours, with frozen promotion rules before expensive continuations. The early screens are intentionally treated as unstable: the 5- and 10-minute rankings are host-sensitive, and the eventual 12-hour top-ranked condition is not the mean-best condition at the replicated 10-minute gate. Because seed ranges differ across stages, these changes are operational promotion evidence, not within-seed curves. A replicated 60-minute gate keeps the Staged Factorial Screening bridge reference in the promoted set, where it ranks first in all four 60-minute host-seed cells. In the final 12-hour confirmation package, the bridge condition ranks first in all four host-seed cells across two seeds; the greedy comparator does not meet the frozen 0.010 val_bpb near-equivalence rule; and the cheaper d8/ar48 (depth-8, aspect-48) sentinel does not meet the frozen 0.020 mean-gap rule. The executed 12-hour branch spends 144 GPU-hours, and the full staged protocol records 169.2 training GPU-hours including screening stages. Continuing all four 60-minute candidates would spend 192 GPU-hours, while continuing all nine replicated 10-minute candidates would spend 432 GPU-hours. The latter numbers are accounting counterfactuals for unrun continuations, not evidence that skipped candidates could not have overtaken the reference. The result is a bounded cost-allocation finding, not a claim of global optimality, capacity-normalized superiority, or superiority over adaptive hyperparameter optimization methods.
Chinese Translation
短时间的预训练运行可以降低实验成本，但也可能过度推广那些在小预算下看起来表现强劲的配置。我们研究了一种可审计的阶段性推广协议，针对在两个异构主机块（Windows A100 和 Linux L40S）上固定的微预训练运行器。从十二个经过先前筛选的配置开始，我们使用了2分钟、5分钟、10分钟、60分钟和12小时的阶段性预算，并在昂贵的后续运行之前冻结推广规则。早期筛选故意被视为不稳定：5分钟和10分钟的排名对主机敏感，而最终的12小时排名第一的条件并不是在复制的10分钟门槛下的平均最佳条件。由于种子范围在不同阶段之间存在差异，这些变化是操作性推广证据，而非种子内曲线。复制的60分钟门槛在推广集中保持了阶段性因子筛选的桥接参考，在所有四个60分钟主机-种子单元中排名第一。在最终的12小时确认包中，桥接条件在两个种子的所有四个主机-种子单元中排名第一；贪婪比较器未能满足冻结的0.010 val_bpb近似等价规则；而更便宜的d8/ar48（深度-8，宽度-48）哨兵未能满足冻结的0.020平均差距规则。执行的12小时分支消耗了144 GPU小时，而完整的阶段性协议记录了169.2 GPU小时的训练时间，包括筛选阶段。继续所有四个60分钟候选者将消耗192 GPU小时，而继续所有九个复制的10分钟候选者将消耗432 GPU小时。后者的数字是未运行的后续的会计反事实，而不是跳过的候选者无法超过参考的证据。结果是一个有限的成本分配发现，而不是全球最优性、容量归一化的优越性或优于自适应超参数优化方法的声明。
cs.CL / 26 / 2606.11399
Scenario-based Probing and Steering Cultural Values in Large Language Models--Extended Version
基于情境的探测与引导大型语言模型中的文化价值观——扩展版
Abstract
Large Language Models (LLMs) are deployed across cultural contexts but often reflect homogenized values inherited from training data. Evaluations of cultural alignment typically rely on direct prompting with survey-style questions, which frequently elicit neutral or safety-aligned responses and fail to capture underlying model preferences. We propose a framework for probing and steering latent cultural representations in LLMs along the two Inglehart--Welzel axes of the World Values Survey (WVS). By translating social value questions into scenario-based behavioral dilemmas, we extract token-level probabilities to measure implicit values and apply activation steering, optionally combined with country-conditioned prompting, to shift model behavior without retraining. Across three open-source LLMs and four target cultures, we find substantial variation in steerability and identify latent entanglement, where interventions along one cultural dimension induce shifts along another. This coupling mirrors correlations in human WVS data and persists across activation, prompt, and hybrid steering. It constrains axis-independent alignment, though general task performance is largely preserved.
Chinese Translation
大型语言模型（LLMs）在不同文化背景中被广泛应用，但往往反映出从训练数据中继承的同质化价值观。对文化一致性的评估通常依赖于直接提示的调查式问题，这些问题常常引发中立或安全导向的回应，未能捕捉到模型的潜在偏好。我们提出了一种框架，用于探测和引导LLMs中潜在的文化表现，沿着世界价值观调查（WVS）的两个Inglehart-Welzel轴线。通过将社会价值问题转化为基于情境的行为困境，我们提取了令牌级别的概率来测量隐含价值，并应用激活引导，选择性地结合国家条件提示，以在不重新训练模型的情况下改变模型行为。在三个开源LLM和四种目标文化中，我们发现可引导性存在显著差异，并识别出潜在的纠缠现象，即在一个文化维度上的干预会引起另一个维度的变化。这种耦合反映了人类WVS数据中的相关性，并在激活、提示和混合引导中持续存在。它限制了轴独立的一致性，尽管整体任务性能在很大程度上得以保留。
cs.CL / 27 / 2606.11420
Context-Aware Multimodal Claim Verification in Spoken Dialogues
基于上下文的多模态口语对话中的声明验证
Abstract
Every day, millions absorb claims from podcasts and streams that no fact-checker ever sees. Spoken misinformation is built through conversation, where credibility comes not from facts alone but from how claims are framed, reinforced, or left unchallenged across turns. Yet fact-checking has focused on isolated text, leaving dialogue audio under-studied. We introduce MAD2, a new Multi-turn Audio Dialogues benchmark for spoken claim verification, containing 1,000 two-speaker dialogues with 3,368 check-worthy claims and approximately 10 hours of audio, and propose calibrated multimodal fusion of a context-aware audio encoder and a dialogue-aware text model. Across settings, adding dialogue context improves verification, but the gains depend on scenario type. Using only preceding context often matches offline performance, supporting live-moderation settings, and audio contributes most when transcript-based models are destabilized by additional context. Overall, conversational structure matters more for verification than misinformation framing.
Chinese Translation
每天，数以百万计的人从播客和直播中获取声明，而这些声明从未被事实核查员看到。口语虚假信息通过对话构建，可信度不仅来自事实本身，还来自声明在对话中的框架、强化或未受到质疑的方式。然而，事实核查主要集中在孤立的文本上，导致对话音频的研究相对不足。我们引入了MAD2，这是一个新的多轮音频对话基准，用于口语声明验证，包含1000个双人对话，涵盖3368个值得核查的声明和大约10小时的音频，并提出了上下文感知音频编码器与对话感知文本模型的校准多模态融合。在各种设置中，添加对话上下文可以提高验证效果，但增益依赖于场景类型。仅使用前文上下文通常可以匹配离线性能，支持实时审核设置，而当转录基础模型因额外上下文而不稳定时，音频的贡献最大。总体而言，对话结构对验证的重要性超过了虚假信息的框架。
cs.CL / 28 / 2606.11424
SOMA-SQL: Resolving Multi-Source Ambiguity in NL-to-SQL via Synthetic Log and Execution Probing
SOMA-SQL：通过合成日志和执行探测解决 NL-to-SQL 中的多源歧义
Abstract
Natural language interfaces to databases aim to translate user questions into executable SQL, yet remain brittle in real-world settings where questions are underspecified and schemas are large and ambiguous. Ambiguity across user questions, database schemas, and model interpretations are central failure modes in NL2SQL, leading to misaligned intent, incorrect schema grounding, and erroneous SQL generation. Existing approaches rely on human clarification or treat ambiguity as a schema representation problem, but these do not scale nor resolve ambiguity autonomously. We propose SOMA-SQL to automatically resolve ambiguity via targeted synthetic query log and ambiguity-driven probing. SOMA-SQL constructs synthetic query log to ground schema interpretation and guide candidate SQL generation; it then executes targeted probing queries, driven by a structured ambiguity taxonomy and candidate disagreements, to produce disambiguation evidence for final SQL selection and repair. This active approach to ambiguity discovery and resolution generalizes across unseen schemas and query distributions without human-in-the-loop. Experiments on six public benchmarks demonstrate that SOMA-SQL improves execution accuracy by 13.0% on average over state-of-the-art baselines, with gains of up to 16.7% on ambiguous questions.
Chinese Translation
自然语言数据库接口旨在将用户问题翻译为可执行的 SQL，但在现实环境中，由于问题不明确且模式庞大且模糊，这些接口仍然脆弱。用户问题、数据库模式和模型解释之间的歧义是 NL2SQL 中的主要失败模式，导致意图不匹配、模式基础错误和 SQL 生成错误。现有方法依赖于人工澄清或将歧义视为模式表示问题，但这些方法无法扩展或自主解决歧义。我们提出了 SOMA-SQL，通过有针对性的合成查询日志和基于歧义的探测自动解决歧义。SOMA-SQL 构建合成查询日志以确定模式解释并指导候选 SQL 生成；然后，它执行有针对性的探测查询，基于结构化的歧义分类法和候选者之间的分歧，生成最终 SQL 选择和修复的消歧证据。这种主动的歧义发现和解决方法能够在未见的模式和查询分布中推广，而无需人工干预。在六个公共基准上的实验表明，SOMA-SQL 在执行准确性上平均比最先进的基线提高了 13.0%，在模糊问题上最高可提高 16.7%。
cs.CL / 29 / 2606.11435
Agent Skill Evaluation and Evolution: Frameworks and Benchmarks
智能体技能评估与演化：框架与基准
Abstract
The growth of agent skills has transformed how agentic systems are built, evaluated, and deployed. As skill libraries continue to scale, rigorous evaluation becomes critical to ensuring their utility, quality, and safety in real-world applications. Consequently, the field is undergoing an emerging paradigm shift from isolated skill creation to automated, evaluation-driven skill evolution. In this survey, we systematically examine the landscape of skill evolution and evaluation beyond foundational skill creation. We categorize evolution into four distinct paradigms, spanning execution feedback, trajectory distillation, compression, and reinforcement learning, showing how each element contributes to improving skill utility and reliability. We also provide an analysis of six skill-centric benchmark categories, identifying structural gaps in benchmark coverage, trade-offs, and metric richness to advance skill research. Finally, we identify open directions for building skill ecosystems that are generalizable, efficient, and verifiably safe. The project URL is https://github.com/Cassie07/AgentSkill_Survey
Chinese Translation
智能体技能的增长改变了智能体系统的构建、评估和部署方式。随着技能库的不断扩展，严格的评估变得至关重要，以确保其在实际应用中的效用、质量和安全性。因此，该领域正经历从孤立的技能创建向自动化、评估驱动的技能演化的范式转变。在本次调查中，我们系统地审视了超越基础技能创建的技能演化与评估的全景。我们将演化分为四种不同的范式，包括执行反馈、轨迹提炼、压缩和强化学习，展示了每个元素如何有助于提高技能的效用和可靠性。我们还分析了六个以技能为中心的基准类别，识别了基准覆盖、权衡和指标丰富性方面的结构性缺口，以推动技能研究的发展。最后，我们确定了构建可推广、高效且可验证安全的技能生态系统的开放方向。项目网址为 https://github.com/Cassie07/AgentSkill_Survey
cs.CL / 30 / 2606.11447
AI Coding Agents Can Reproduce Social Science Findings
人工智能编码代理能够再现社会科学发现
Abstract
Recent anecdotal evidence suggests that AI coding agents can reproduce published findings when provided with original data and code; yet systematic evaluation across social sciences remains limited. Existing evaluation benchmarks are insufficient, either small or conflate agent performance with problems in the reproduction materials themselves, such as code that fails to execute correctly. Here we introduce SocSci-Repro-Bench, a benchmark of 221 tasks spanning four disciplines and 13 substantive domains, constructed from studies whose results are either fully reproducible with available materials or demonstrably non-reproducible due to missing data, allowing us to isolate agents' reproduction capacity. Evaluating two frontier coding agents, Claude Code and Codex, we find that both can reproduce a large share of social science findings, with Claude Code substantially outperforming Codex. These reproduction rates considerably exceed those previously reported for general-purpose LLM-based agents on comparable reproducibility benchmarks. Both agents also perform strongly on a reasoning task requiring identification of underlying research questions, and additional analyses suggest that results are not primarily driven by memorization. Providing the original paper PDF alongside replication materials modestly improves performance but introduces bias on tasks where reproduction is impossible. We also show that agents can be nudged toward confirmatory specification search through subtle prompt framing. Together, these findings suggest that at least some frontier coding agents can serve as reliable executors of computational workflows while underscoring the need for careful benchmarking and prompt design as AI systems assume larger roles in scientific production.
Chinese Translation
近期的轶事证据表明，人工智能编码代理在提供原始数据和代码的情况下能够再现已发表的研究结果；然而，针对社会科学的系统评估仍然有限。现有的评估基准不足，要么样本量小，要么将代理的表现与再现材料本身的问题混为一谈，例如代码无法正确执行。在此，我们介绍了SocSci-Repro-Bench，这是一个涵盖四个学科和13个实质领域的221个任务的基准，构建于结果要么可以通过可用材料完全再现，要么由于缺失数据而明显无法再现的研究之上，从而使我们能够隔离代理的再现能力。通过评估两个前沿编码代理，Claude Code和Codex，我们发现两者均能够再现大量社会科学发现，其中Claude Code的表现显著优于Codex。这些再现率远超以往针对一般用途的大型语言模型（LLM）代理在可比再现基准上报告的结果。这两个代理在需要识别潜在研究问题的推理任务上也表现出色，额外分析表明，结果并非主要由记忆驱动。将原始论文PDF与复制材料一同提供会适度提高性能，但在无法再现的任务上引入了偏差。我们还展示了通过微妙的提示框架可以引导代理朝向确认性规范搜索。综合来看，这些发现表明，至少一些前沿编码代理可以作为可靠的计算工作流程执行者，同时强调了在人工智能系统在科学生产中承担更大角色时，谨慎进行基准测试和提示设计的必要性。
cs.CL / 31 / 2606.11456
AI Coding Agents in Social Science: Methodologically Diverse, Empirically Consistent, Interpretively Vulnerable
社会科学中的人工智能编码代理：方法多样性、实证一致性与解释脆弱性
Abstract
The deployment of LLM-based agents in scientific analysis raises opposing concerns: that agents may reduce methodological diversity, or that they may amplify the analytic flexibility through which researchers reach motivated conclusions. We argue these worries target two empirically separable layers: a design layer of methodological choices, and a verdict layer in which a decision rule maps estimates to a substantive claim. We test both by running 20 independent executions of Claude Code and Codex on a prominent immigration and social-policy against a many-analysts human baseline. At the design layer, Codex matches human methodological diversity and Claude Code produces nearly three times as many specifications; both agents' effect estimates remain broadly aligned with the human consensus, and no agent model exactly matches any human model. A prompt-induced anti-immigration researcher prior reorganizes each agent's methodological decisions but, unlike for biased human analysts in the same data, does not shift aggregate estimates or final verdicts; nor do agents reroute along the methodological axes humans use to bias their estimates. At the verdict layer, an explicit confirmatory prompt flips Claude Code's verdicts from 10% to 90% support while leaving its coefficient distribution essentially unchanged, operating through rule omission rather than rule softening. AI agents can rival or exceed human methodological diversity at the design layer while remaining vulnerable at the verdict layer. In our setting, the locus of AI bias is not estimation but interpretation.
Chinese Translation
基于大规模语言模型（LLM）的代理在科学分析中的应用引发了相互对立的担忧：一方面，代理可能减少方法多样性；另一方面，它们可能通过增强分析灵活性，使研究者得出有动机的结论。我们认为，这些担忧针对两个可实证分离的层面：一个是方法选择的设计层，另一个是决策层，在该层中，决策规则将估计结果映射到实质性主张。我们通过对Claude Code和Codex在一个显著的移民和社会政策问题上进行20次独立执行，与众多分析师的人类基线进行对比，来检验这两个层面。在设计层面，Codex与人类的方法多样性相匹配，而Claude Code则产生了近三倍的规范；两个代理的效应估计与人类共识基本一致，且没有任何代理模型完全匹配任何人类模型。一个由提示引发的反移民研究者先验重新组织了每个代理的方法决策，但与在相同数据中存在偏见的人类分析师不同，并未改变总体估计或最终裁决；代理也没有沿着人类用以偏见其估计的方法轴重新调整。在决策层面，一个明确的确认性提示将Claude Code的裁决从10%支持率翻转为90%，而其系数分布基本保持不变，这一变化是通过规则省略而非规则软化实现的。人工智能代理在设计层面可以与人类的方法多样性相媲美或超越，但在决策层面仍然脆弱。在我们的研究背景中，人工智能偏见的焦点不在于估计，而在于解释。
cs.CL / 32 / 2606.11459
APEX: Automated Prompt Engineering eXpert with Dynamic Data Selection
APEX：具有动态数据选择的自动提示工程专家
Abstract
Large Language Models are highly sensitive to prompt formulation, necessitating automatic prompt optimization to unlock their full potential. While evolutionary algorithms have emerged as the dominant paradigm, they suffer from a critical bottleneck: data efficiency. Current methods treat the development dataset as a static benchmark, wasting significant compute budget on uninformative data. In this work, we introduce APEX (Automatic Prompt Engineering eXpert), a novel framework that optimizes the data usage alongside the prompt search. APEX dynamically stratifies the dataset into Easy, Hard, and Mixed tiers based on the optimization lineage. By prioritizing the Mixed tier, which identifies the data where the LLM has mixed performance, we identify two high-leverage subsets: the addressable frontier for generating informative mutations and the rank-sensitive frontier for distinguishing candidate quality. We evaluate APEX across three diverse benchmarks: IFBench, SimpleQA Verified, and FACTS Grounding. Under a fixed budget of 5,000 evaluation calls, due to its data efficiency, APEX outperforms the initial prompt by an average of 11.2% on Gemini 2.5 Flash and 6.8% on Gemma 3 27B, demonstrating that a data-centric approach is key to efficient and effective prompt optimization.
Chinese Translation
大型语言模型对提示的构造高度敏感，因此需要自动提示优化以发挥其全部潜力。尽管进化算法已成为主流范式，但它们面临一个关键瓶颈：数据效率。目前的方法将开发数据集视为静态基准，浪费了大量计算预算在无信息的数据上。在本研究中，我们提出了APEX（自动提示工程专家），这是一个优化数据使用与提示搜索的新框架。APEX根据优化的谱系动态地将数据集分层为简单、困难和混合三个层次。通过优先考虑混合层次，即识别大型语言模型（LLM）表现不一的数据，我们确定了两个高杠杆子集：用于生成信息性突变的可处理前沿和用于区分候选质量的排名敏感前沿。我们在三个不同的基准上评估了APEX：IFBench、SimpleQA Verified和FACTS Grounding。在固定的5000次评估调用预算下，由于其数据效率，APEX在Gemini 2.5 Flash上平均超越初始提示11.2%，在Gemma 3 27B上超越6.8%，证明了以数据为中心的方法是高效且有效的提示优化的关键。
cs.CL / 33 / 2606.11470
The Periodic Table of LLM Reasoning: A Structured Survey of Reasoning Paradigms, Methods, and Failure Modes
大型语言模型推理的周期表：推理范式、方法和失败模式的结构化调查
Abstract
Large Language Models (LLMs) have achieved strong performance across natural language processing tasks, yet reliable reasoning remains an open challenge. Although modern LLMs show progress in structured inference, multi-step problem solving, and contextual understanding, their reasoning behavior is often inconsistent and sensitive to prompting strategies, task design, and model scale. This survey provides a systematic analysis of more than 300 recent papers from arXiv, Semantic Scholar, Google Scholar, Papers with Code, and the ACL Anthology to examine how reasoning capabilities emerge in LLMs and where they fail. We make three main contributions. First, we introduce a structured taxonomy of LLM reasoning research, covering Chain-of-Thought reasoning, multi-hop reasoning, mathematical reasoning, common sense reasoning, visual and temporal reasoning, code and algorithmic reasoning, retrieval-augmented reasoning, tool-augmented and agentic reasoning, and reinforcement learning-based reasoning. Second, we analyze methodological trends across these paradigms, including prompting methods, model architectures, training objectives, reward modeling, and evaluation benchmarks. Third, we synthesize recurring limitations and failure modes, such as reasoning hallucinations, brittle multi-step inference, weak causal abstraction, and poor cross-domain generalization. By organizing a rapidly expanding literature, this survey offers a unified view of the current capabilities and limitations of reasoning in LLMs. We also identify emerging research directions, including meta-reasoning, self-evolving reasoning frameworks, multimodal reasoning, and socially grounded reasoning. Overall, this work aims to serve as a reference for developing more robust, interpretable, and generalizable reasoning systems in future language models.
Chinese Translation
大型语言模型（LLMs）在自然语言处理任务中取得了强劲的表现，但可靠的推理仍然是一个未解决的挑战。尽管现代LLMs在结构化推理、多步骤问题解决和上下文理解方面显示出进展，但它们的推理行为往往不一致，并且对提示策略、任务设计和模型规模敏感。本调查对来自arXiv、Semantic Scholar、Google Scholar、Papers with Code和ACL Anthology的300多篇近期论文进行了系统分析，以考察LLMs中推理能力的出现及其失败之处。我们做出了三项主要贡献。首先，我们引入了LLM推理研究的结构化分类法，涵盖了思维链推理（Chain-of-Thought reasoning）、多跳推理（multi-hop reasoning）、数学推理（mathematical reasoning）、常识推理（common sense reasoning）、视觉和时间推理（visual and temporal reasoning）、代码和算法推理（code and algorithmic reasoning）、检索增强推理（retrieval-augmented reasoning）、工具增强和代理推理（tool-augmented and agentic reasoning）以及基于强化学习的推理（reinforcement learning-based reasoning）。其次，我们分析了这些范式中的方法趋势，包括提示方法、模型架构、训练目标、奖励建模和评估基准。第三，我们综合了反复出现的局限性和失败模式，如推理幻觉（reasoning hallucinations）、脆弱的多步骤推理（brittle multi-step inference）、薄弱的因果抽象（weak causal abstraction）和较差的跨领域泛化（poor cross-domain generalization）。通过组织快速扩展的文献，本调查提供了对LLMs中推理当前能力和局限性的统一视角。我们还识别了新兴的研究方向，包括元推理（meta-reasoning）、自我演化推理框架（self-evolving reasoning frameworks）、多模态推理（multimodal reasoning）和社会基础推理（socially grounded reasoning）。总体而言，这项工作旨在为未来语言模型中开发更强大、可解释和可推广的推理系统提供参考。
cs.CL / 34 / 2606.11499
Hubs or Fringes: Pretraining Data Selection via Web Graph Centrality
中心节点还是边缘节点：通过网络图中心性选择预训练数据
Abstract
The performance of modern language models depends critically on pretraining data composition. Yet existing data selection methods rely on auxiliary classifiers for document scoring or mixture optimization, adding computational overhead and dependence on labeled data. We propose WebGraphMix, a lightweight data selection framework that computes structural centrality scores over the Common Crawl host-level web graph and uses them to vary the proportion of central versus peripheral documents in the pretraining mixture. We hypothesize that central hosts expose models to reusable abstractions, while peripheral hosts encode specialized, long-tail knowledge. WebGraphMix computes centrality scores efficiently at web scale, requiring no model training, labeled data, or downstream supervision. We integrate WebGraphMix into the DataComp-LM pipeline and train models at 400M and 1B parameter scales with 8B and 28B tokens respectively, evaluating on 23 tasks ranging from factual knowledge to symbolic reasoning. Our experiments show that central and peripheral web regions encode complementary capabilities. Mixture combining both at a ratio of 1:1 achieves 41.4% on average, compared to 39.8% for uniform sampling. Combining structural scores with document-level quality classifier scores further improves performance to 43.8%. These findings demonstrate that web graph topology is a meaningful axis for pretraining data curation, capturing information that is largely orthogonal to existing content-based approaches.
Chinese Translation
现代语言模型的性能在很大程度上依赖于预训练数据的组成。然而，现有的数据选择方法依赖于辅助分类器进行文档评分或混合优化，这增加了计算开销并依赖于标注数据。我们提出了WebGraphMix，一个轻量级的数据选择框架，它在公共抓取（Common Crawl）主机级网络图上计算结构中心性分数，并利用这些分数调整预训练混合中中心文档与边缘文档的比例。我们假设中心主机使模型接触到可重用的抽象，而边缘主机则编码了专业化的长尾知识。WebGraphMix在网络规模上高效地计算中心性分数，无需模型训练、标注数据或下游监督。我们将WebGraphMix集成到DataComp-LM管道中，并在400M和1B参数规模下分别使用8B和28B个标记进行模型训练，评估范围涵盖从事实知识到符号推理的23个任务。我们的实验表明，中心和边缘网络区域编码了互补的能力。将两者以1:1的比例混合，平均得分达到41.4%，而均匀采样的得分为39.8%。将结构分数与文档级质量分类器分数相结合，进一步提高了性能至43.8%。这些发现表明，网络图拓扑是预训练数据策划的一个有意义的轴线，捕捉到的信息在很大程度上与现有的基于内容的方法正交。
cs.CL / 35 / 2606.11502
When Roleplaying, Do Models Believe What They Say?
角色扮演时，模型是否相信他们所说的话？
Abstract
Language models can state that "the Earth orbits the Sun" and, when role-playing Aristotle, assert the opposite. Recent work argues that persona adoption is fundamental to how language models operate, with models constantly selecting the most appropriate persona for a given context. Does such role-playing merely change the model's outputs, or does it also affect what the model internally represents as truthful? We study this question with linear truth probes, applying them to LLMs role-playing historical personas whose likely beliefs differ from modern consensus. For each persona, we compare false claims the persona would likely have endorsed (*era-believed*) with topic-matched false claims they would not have endorsed (*era-false*). Across prompting, in-context learning, and supervised fine-tuning, persona induction suppresses era-believed statements less than equally false alternatives, yet they remain classified as false overall. Role-play therefore shifts what these models say more than what they internally represent as true. We contrast this with models trained on harmful advice that exhibit Emergent Misalignment (EM). Across three model families (Qwen 2.5 14B, Qwen 3 8B, and Llama 3.3 70B), their false claims move substantially toward the true region of probe space, are defended under challenge roughly half the time versus about a sixth for role-play, and are used in downstream reasoning. Role-play and Emergent Misalignment thus are points on a spectrum of belief internalization, where role-play changes what a model says with little representational change, while Emergent Misalignment shifts the internal representation of false claims without fully marking them as true.
Chinese Translation
语言模型可以声明“地球绕太阳运行”，并在角色扮演亚里士多德时断言相反。最近的研究认为，角色采用是语言模型运作的基础，模型在特定上下文中不断选择最合适的角色。这种角色扮演仅仅改变模型的输出，还是也影响模型内部所代表的真实信息？我们通过线性真相探测器研究这个问题，将其应用于角色扮演历史人物的LLMs，这些人物的信念可能与现代共识不同。对于每个角色，我们比较该角色可能支持的虚假陈述（*era-believed*）与他们不会支持的主题匹配虚假陈述（*era-false*）。在提示、上下文学习和监督微调的过程中，角色诱导对时代信念的陈述抑制程度低于同样虚假的替代品，但它们整体上仍被分类为虚假。因此，角色扮演更多地改变了这些模型所说的话，而不是它们内部所代表的真实信息。我们将此与训练于有害建议的模型进行对比，这些模型表现出新兴不一致性（Emergent Misalignment，EM）。在三种模型家族（Qwen 2.5 14B、Qwen 3 8B 和 Llama 3.3 70B）中，它们的虚假陈述显著向探测空间的真实区域移动，在挑战下大约一半的时间得到辩护，而角色扮演的比例约为六分之一，并在下游推理中被使用。因此，角色扮演和新兴不一致性在信念内化的光谱上是两个点，其中角色扮演改变模型所说的话而几乎没有代表性变化，而新兴不一致性则在不完全标记为真实的情况下改变虚假陈述的内部表示。
cs.CL / 36 / 2606.11512
SAGE: Answer-Conditioned Uncertainty Targets for Verbal Uncertainty Alignment
SAGE：基于答案条件的不确定性目标用于语言不确定性对齐
Abstract
Large language models increasingly express uncertainty through natural-language statements, yet these expressions often fail to reflect the model's sampled behavior. We study verbal uncertainty alignment as a distributional calibration problem: the appropriate uncertainty target for a prompt should be estimated from repeated model outputs rather than from an isolated response. However, group rollouts alone are insufficient, since the resulting target must provide a useful training signal. Existing targets only partially satisfy this requirement. We propose SAGE, Semantic-Answer Guided Entropy, a group-level uncertainty target that constructs an answer-conditioned uncertainty geometry over sampled responses. SAGE preserves categorical, numeric, and symbolic answer distinctions while maintaining a smooth and scale-preserving calibration signal. We further apply this target through Group-Uncertainty Preference Optimization, or GUPO, an uncertainty-channel training framework that supervises verbal uncertainty expressions rather than the full response. Experiments across factual, mathematical, and multiple-choice reasoning tasks show improved uncertainty ranking, lower calibration error, and reduced overconfidence.
Chinese Translation
大型语言模型越来越多地通过自然语言陈述表达不确定性，但这些表达往往未能反映模型的采样行为。我们将语言不确定性对齐视为一种分布校准问题：对于一个提示，适当的不确定性目标应基于重复的模型输出进行估计，而不是孤立的响应。然而，仅仅依靠群体输出是不够的，因为所得到的目标必须提供有用的训练信号。现有目标仅部分满足这一要求。我们提出了SAGE（Semantic-Answer Guided Entropy），一种构建基于答案条件的不确定性几何体的群体级不确定性目标，旨在对采样响应进行建模。SAGE在保持类别、数值和符号答案区分的同时，提供平滑且保持尺度的校准信号。我们进一步通过群体不确定性偏好优化（Group-Uncertainty Preference Optimization，GUPO）应用该目标，这是一种监督语言不确定性表达而非完整响应的不确定性通道训练框架。针对事实、数学和多项选择推理任务的实验表明，改进了不确定性排名，降低了校准误差，并减少了过度自信。
cs.CL / 37 / 2606.11520
ISE: An Execution-Grounded Recipe for Multi-Turn OS-Agent Trajectories
ISE：一种基于执行的多轮操作系统代理轨迹生成方法
Abstract
Training capable OS agents requires data that simultaneously captures structured user intents, multi-turn task delegation, and grounded tool execution--properties absent from existing datasets. We propose ISE (Intent -> Simulate -> Execute), a three-stage synthesis paradigm that addresses these gaps jointly. Stage 1 constructs roughly 50000 structured intents via a 4D framework (Persona x Domain x Task x Complexity); after deduplication the pool contains 43956 unique intents and attains a Vendi Score of 61.57 over the entire pool on mpnet-base-v2 embeddings (cosine kernel, q=1). Stage 2 drives multi-turn user-agent interaction through a role-locked user simulator that grounds each user turn in actual execution outcomes, producing 23132 complete trajectories averaging 8.12 user turns and 68.24 total dialogue turns. Stage 3 runs every tool call inside a live, isolated OS workspace, generating authentic failure-recovery dynamics instead of simulated responses. Fine-tuning on ISETrace improves ClawEval pass@1 from 19.3 to 37.7 using Qwen3-8B on agent tool-use tasks with a standard protocol. This result outperforms zero-shot GPT-4o and the larger Qwen3-32B base model which is four times bigger. An ablation on Stage 2 proves multi-turn simulation brings a large portion of the performance gain. We release all source code and dataset at https://github.com/Valiere01/ISE-Trace.
Chinese Translation
训练能够执行任务的操作系统代理需要同时捕捉结构化用户意图、多轮任务委派和基于实际执行的工具使用等数据特性，而这些特性在现有数据集中并不存在。我们提出了ISE（Intent -> Simulate -> Execute），一种三阶段合成范式，旨在共同解决这些问题。第一阶段通过一个四维框架（Persona x Domain x Task x Complexity）构建大约50000个结构化意图；去重后，数据池包含43956个独特意图，并在mpnet-base-v2嵌入上获得了61.57的Vendi Score（余弦核，q=1）。第二阶段通过一个角色锁定的用户模拟器驱动多轮用户与代理的交互，将每个用户的回合与实际执行结果相结合，生成23132条完整轨迹，平均包含8.12个用户回合和68.24个总对话回合。第三阶段在一个实时、隔离的操作系统工作空间内运行每个工具调用，生成真实的失败恢复动态，而不是模拟响应。对ISETrace的微调使ClawEval的pass@1从19.3提高到37.7，使用Qwen3-8B在标准协议下的代理工具使用任务上进行测试。该结果超越了零-shot GPT-4o和体量是其四倍的更大Qwen3-32B基础模型。对第二阶段的消融实验证明，多轮模拟带来了性能提升的很大一部分。我们在https://github.com/Valiere01/ISE-Trace发布所有源代码和数据集。
cs.CL / 38 / 2606.11531
Measuring language complexity from hierarchical reuse of recurring patterns
通过层次重用重复模式测量语言复杂性
Abstract
We introduce the ladderpath index as a measure of language complexity grounded in algorithmic information theory. It counts the minimum steps needed to reconstruct a sequence through hierarchical reuse of repeated substructures, capturing an exactly computable but constrained form of algorithmic compressibility related to, but distinct from, Kolmogorov complexity. We apply the ladderpath approach to 21 parallel corpora from the Parallel Universal Dependencies dataset. The ladderpath index is approximately invariant across the languages, and varies much less than the corpus length. This is more pronounced when all corpora are mapped to a unified binary representation, providing evidence for the equi-complexity hypothesis from a representation-independent perspective. We also observe trade-offs between character inventory size and corpus length, and between vocabulary-level and corpus-level reconstruction complexity, supporting the trade-off hypothesis that total complexity is conserved and redistributed across linguistic levels. The reusable substructures identified by the ladderpath approach, without any linguistic input, overlap with words and morphological components attested in the natural vocabulary. The hierarchical reuse captured by the ladderpath approach parallels the chunking mechanisms proposed in cognitive science, where the human cognitive system compresses linguistic input into nested, reusable units under shared memory and processing constraints. This connection between cognitive chunking and the ladderpath approach provides a new interpretation for the equi-complexity and trade-off hypotheses, grounding both in the shared cognitive architecture that underlies language processing across human languages.
Chinese Translation
我们引入了梯路径指数作为一种基于算法信息理论的语言复杂性度量。它计算了通过层次重用重复子结构重建序列所需的最小步骤，捕捉了一种可精确计算但受限的算法压缩性形式，这与Kolmogorov复杂性相关但又有所不同。我们将梯路径方法应用于来自平行通用依赖数据集的21个平行语料库。梯路径指数在不同语言间大致保持不变，并且变化幅度远小于语料库长度。当所有语料库映射到统一的二进制表示时，这种现象更加明显，为从表示独立的角度提供了对等复杂性假说的证据。我们还观察到字符库存大小与语料库长度之间，以及词汇层级与语料库层级重建复杂性之间的权衡，支持了总复杂性在语言层面之间保持不变并重新分配的权衡假说。通过梯路径方法识别的可重用子结构在没有任何语言输入的情况下，与自然词汇中存在的单词和形态成分重叠。梯路径方法捕捉的层次重用与认知科学中提出的分块机制相似，其中人类认知系统在共享记忆和处理限制下将语言输入压缩为嵌套的可重用单元。这种认知分块与梯路径方法之间的联系为等复杂性和权衡假说提供了新的解释，将两者根植于支撑人类语言处理的共同认知架构中。
cs.CL / 39 / 2606.11542
Pretrained self-supervised speech models can recognize unseen consonants
预训练自监督语音模型能够识别未见的辅音
Abstract
Modern pretrained self-supervised automatic speech recognition models are trained on large-scale audio data to encode speech into contextualized representations. However, their training data are heavily skewed toward high-resource languages with little data from low-resource languages, raising concerns about the potential underrepresentation of typologically uncommon speech sounds such as click consonants primarily found in Khoisan languages. This leads to our central research question: Can these models recognize click consonants as accurately as other speech sounds? To address this question, we fine-tune and compare pretrained self-supervised speech models (Wav2Vec2 and HuBERT) on data from two click-rich Khoisan languages (G|ui and West !Xoon). Our results reveal that the fine-tuned models consistently recognize clicks more accurately than non-clicks, suggesting that self-supervision enables generalization across human speech sounds including rare phonemes.
Chinese Translation
现代预训练自监督自动语音识别模型在大规模音频数据上进行训练，以将语音编码为上下文化的表示。然而，它们的训练数据严重偏向于高资源语言，低资源语言的数据则相对较少，这引发了对类型上不常见的语音声音（如主要存在于科伊桑语言中的点击辅音）潜在代表性不足的担忧。这引出了我们的核心研究问题：这些模型能否像识别其他语音声音一样准确地识别点击辅音？为了解决这个问题，我们在两种富含点击的科伊桑语言（G|ui 和 West !Xoon）的数据上对预训练自监督语音模型（Wav2Vec2 和 HuBERT）进行了微调和比较。我们的结果表明，微调后的模型在识别点击时的准确性始终高于非点击，表明自监督学习能够在包括稀有音素在内的人类语音声音之间实现泛化。
cs.CL / 40 / 2606.11552
Teaching Diffusion to Speculate Left-to-Right
教学扩散以进行从左到右的推测
Abstract
Large language models (LLMs) achieve remarkable performance across a wide range of tasks, but their autoregressive decoding process incurs substantial inference costs due to inherently sequential token generation. Speculative decoding addresses this bottleneck by employing a lightweight draft model to propose multiple future tokens that are subsequently verified in parallel by a larger target model. Recent work has demonstrated that diffusion language models are well suited for this setting, as they can generate entire blocks of draft tokens in parallel and thereby alleviate the sequential constraints of autoregressive drafting. A subtlety of this regime is that block-diffusion drafters generate tokens bidirectionally within a block, whereas verification is performed by an autoregressive target model that evaluates tokens in a strictly left-to-right manner, leaving a gap between the symmetric training-time objective and the asymmetric verification-time reward. In this work, we offer an empirical analysis of three training-time interventions that narrow this gap: token positional weighting, a first-error focal loss that targets the position that breaks the accepted prefix within each block, and a chain loss term that substitutes a differentiable surrogate for the expected accepted length. The three interventions act along orthogonal axes (position, block-conditional first error, joint prefix) and compose additively; they are likewise orthogonal to test-time alignment mechanisms such as multi-draft self-selection, with which they can in principle be combined. Across four target models and six reasoning, code, and dialogue benchmarks, the three interventions raise accepted draft length by 21-76% per benchmark over a position-uniform baseline, without adding additional forward passes and without changing the inference pipeline or the rejection-sampling exactness contract.
Chinese Translation
大型语言模型（LLMs）在广泛的任务中表现出色，但其自回归解码过程由于固有的顺序令牌生成而产生了可观的推理成本。推测解码通过采用轻量级草稿模型来提出多个未来令牌，从而解决了这一瓶颈，这些令牌随后由更大的目标模型并行验证。最近的研究表明，扩散语言模型非常适合这种设置，因为它们能够并行生成整个草稿令牌块，从而缓解自回归草拟的顺序限制。该机制的一个细微之处在于，块扩散草拟者在块内双向生成令牌，而验证则由一个严格从左到右评估令牌的自回归目标模型进行，这在对称的训练时间目标和不对称的验证时间奖励之间留下了差距。在本研究中，我们提供了三种训练时间干预的实证分析，以缩小这一差距：令牌位置加权、针对每个块中打破接受前缀位置的首个错误聚焦损失，以及替代预期接受长度的可微分替代链损失项。这三种干预在正交轴（位置、块条件首个错误、联合前缀）上作用并且可加性组合；它们同样与测试时间对齐机制（如多草稿自我选择）正交，原则上可以与之结合。在四个目标模型和六个推理、代码及对话基准测试中，这三种干预使得每个基准的接受草稿长度提高了21-76%，而没有增加额外的前向传递，也没有改变推理管道或拒绝采样的准确性合同。
cs.CL / 41 / 2606.11599
When is Your LLM Steerable?
何时您的大语言模型可被引导？
Abstract
Activation steering offers a lightweight approach to control language models' behavior at inference time, but whether it succeeds or fails heavily depends on the prompt, concept, model, and steering configuration. Finding the regime and boundaries of successful steering typically requires expensive grid searches and post-hoc evaluation of full autoregressive rollouts. In this work, we investigate whether steerability can be predicted from the model's internal states at the beginning of the generation process, e.g., after generating the first few tokens, and how to leverage such a predictor to improve steering success rate. To this end, we first introduce ASTEER, a testbed including 1.4M steered generations, spanning 150 concepts with each steering success/failure labeled. Leveraging this testbed, we analyze the model's early decoding dynamics by extracting features that compare hidden states before and after steering across layers and initial decoding steps. These features help us understand how steering's effects propagate along layers and token positions, which provide key information for steerability prediction. We then train a Gradient Boosting Decision Trees (GBDT) classifier on these features to predict whether an intervention will under-steer, succeed, or over-steer without requiring full rollout. Our predictor achieves around 0.7 macro-F1 score on unseen concepts, demonstrating that early hidden states encode substantial, structured information about eventual steering efficacy. We further leverage this steerability predictor as guidance for steering strength searching, achieving near-optimal performance with a small fraction of decoding cost.
Chinese Translation
激活引导提供了一种轻量级的方法来控制语言模型在推理时的行为，但其成功与否在很大程度上依赖于提示、概念、模型和引导配置。寻找成功引导的范围和边界通常需要昂贵的网格搜索和对完整自回归生成的事后评估。在本研究中，我们探讨了是否可以从模型生成过程开始时的内部状态（例如，在生成前几个标记后）预测引导性，以及如何利用这种预测器来提高引导成功率。为此，我们首先介绍了ASTEER，一个测试平台，包括140万次引导生成，涵盖150个概念，并标注每次引导的成功/失败。利用这个测试平台，我们通过提取特征分析模型的早期解码动态，比较引导前后各层和初始解码步骤的隐藏状态。这些特征帮助我们理解引导效果如何沿着层和标记位置传播，从而提供关键的信息用于引导性预测。然后，我们在这些特征上训练了一个梯度提升决策树（GBDT）分类器，以预测干预是否会导致不足引导、成功或过度引导，而无需完整的生成过程。我们的预测器在未见概念上达到了约0.7的宏F1分数，证明早期隐藏状态编码了关于最终引导有效性的实质性、结构化信息。我们进一步利用这个引导性预测器作为引导强度搜索的指导，以较小的解码成本实现近乎最优的性能。
cs.CL / 42 / 2606.11609
Multi-Agent Reasoning with Adaptive Worker Allocation for Stance Detection
基于自适应工人分配的多智能体推理用于立场检测
Abstract
Stance detection requires identifying an author's position toward a target, often from short-form texts where stance is implicit, indirect, or rhetorically framed. Although large language models (LLMs) achieve strong performance on this task, single-pass prompting can be brittle when multiple interpretations are plausible. Existing aggregation strategies, such as majority voting or self-consistency, improve robustness by combining labels, but they discard the intermediate reasoning needed to resolve conflicting interpretations. We introduce a multi-agent reasoning framework with adaptive worker allocation for stance detection that shifts aggregation from label-level voting to reasoning-level synthesis. The framework employs a Manager-Worker architecture in which a Manager adaptively allocates a variable number of Worker agents based on input complexity. Each Worker analyzes the input from a distinct perspective and produces a reasoning-only explanation without emitting a stance label; the Manager then synthesizes these explanations to produce the final prediction. We evaluate the proposed framework on SemEval-2016, P-Stance, and COVID-19 Stance using Llama, Mistral, and Gemini. Results show that the framework yields the largest gains on implicit and context-dependent stance cases, achieving 86.07 Macro-F1 on COVID-19 and 82.90 on SemEval-2016, while remaining competitive on more explicit stance datasets such as P-Stance. These findings suggest that adaptive reasoning-level aggregation is most beneficial when stance cannot be reliably inferred from surface cues alone.
Chinese Translation
立场检测需要识别作者对目标的态度，通常来自于立场隐含、间接或修辞框架的短文本。尽管大型语言模型（LLMs）在此任务上表现出色，但在多种解释可能的情况下，单次提示可能会显得脆弱。现有的聚合策略，如多数投票或自我一致性，通过结合标签来提高鲁棒性，但它们丢弃了解决冲突解释所需的中间推理。我们提出了一种多智能体推理框架，结合自适应工人分配用于立场检测，将聚合从标签级投票转变为推理级综合。该框架采用经理-工人架构，其中经理根据输入复杂性自适应地分配可变数量的工人代理。每个工人从不同的角度分析输入，并生成仅包含推理的解释，而不发出立场标签；然后经理综合这些解释以生成最终预测。我们在SemEval-2016、P-Stance和COVID-19立场数据集上评估了该框架，使用了Llama、Mistral和Gemini。结果表明，该框架在隐含和依赖上下文的立场案例中获得了最大的提升，在COVID-19上达到了86.07的Macro-F1，在SemEval-2016上达到了82.90，同时在P-Stance等更明确的立场数据集上也保持了竞争力。这些发现表明，当立场无法仅从表面线索可靠推断时，自适应推理级聚合最为有益。
cs.CL / 43 / 2606.11639
Evaluating Bias in Phoneme-Based Automatic Speech Recognition Systems: An Analysis of IPA Transcription Models
评估基于音素的自动语音识别系统中的偏见：国际音标转录模型的分析
Abstract
The popularization of automatic speech recognition (ASR) systems has increased exploration of the demographic biases related to race, age, gender, and accent, often formed from imbalanced training data. Most of these studies focused on standard grapheme-based ASR systems with comparatively little emphasis on phoneme-based systems, such as models that produce International Phonetic Alphabet (IPA) representations. As ASR systems shift toward multilingual support and low-resource language modeling, IPA-based layers serve as a critical, language-agnostic foundation. In this study, we evaluate the performance of two state-of-the-art open-source ASR systems, WhisperIPA and ZIPA, that generate IPA transcriptions across diverse accents and language sources. Our evaluation includes existing multilingual speech corpora and demographically annotated English-language corpora. We measure model performance by comparing model-generated IPA transcriptions against grapheme-to-phoneme (G2P) systems using both standard phoneme error rate (PER) and a proposed Soft PER metric that tolerates linguistically similar phoneme substitutions. Our analysis examines how performance varies across languages and demographic groups such as gender, accent, ethnicity, and age, revealing persistent disparities even after accounting for acceptable phonemic variation. These findings provide insight into potential sources of bias and inform the development of more inclusive and linguistically robust phoneme-based ASR systems. Our code and data will be made publicly available to the community.
Chinese Translation
自动语音识别（ASR）系统的普及增加了对与种族、年龄、性别和口音相关的人口偏见的探索，这些偏见通常源于不平衡的训练数据。大多数研究集中于标准的基于字素的ASR系统，而对基于音素的系统（如生成国际音标（IPA）表示的模型）的关注相对较少。随着ASR系统向多语言支持和低资源语言建模的转变，基于IPA的层作为一种关键的、与语言无关的基础。在本研究中，我们评估了两个最先进的开源ASR系统WhisperIPA和ZIPA的性能，这些系统能够生成来自不同口音和语言来源的IPA转录。我们的评估包括现有的多语言语音语料库和带有人口统计注释的英语语料库。我们通过将模型生成的IPA转录与基于字素到音素（G2P）系统进行比较，使用标准音素错误率（PER）和一种提出的软音素错误率（Soft PER）指标来测量模型性能，该指标允许语言上相似的音素替代。我们的分析考察了性能在不同语言和人口群体（如性别、口音、种族和年龄）之间的变化，揭示了即使在考虑可接受的音素变异后，仍然存在持续的差异。这些发现为潜在的偏见来源提供了见解，并为开发更具包容性和语言学稳健性的基于音素的ASR系统提供了指导。我们的代码和数据将向社区公开。
cs.CL / 44 / 2606.11643
Improving Cross-Format Robustness in Language Models with Multi-Format Training
通过多格式训练提高语言模型的跨格式鲁棒性
Abstract
Large language models often remain sensitive to answer format: a question solved correctly in one form may fail in another semantically equivalent form. To study this gap, we define cross-format robustness as the extent to which a model answers the same underlying question consistently across formats. We then compare full-format training with FormatMix, which expands only a subset of training items into multiple equivalent formats using either random or targeted selection. Across GLM4 and Llama-3.1, multi-format supervision consistently improves both task performance and cross-format robustness, whereas Multiple-choice question (MCQ)-only supervision alone brings little benefit and can even reduce robustness. We further find that expanding only about 30% of the training set into multiple formats often recovers most of the gain from full-format training, and this effect appears across the model families and sizes we study. These results suggest that format diversity, rather than additional supervision alone, is the key driver of robustness. That lightweight multi-format augmentation is a practical way to make LLMs less sensitive to answer format without changing the base model.
Chinese Translation
大型语言模型通常对答案格式敏感：在一种形式中正确解决的问题可能在另一种语义等价的形式中失败。为了研究这一差距，我们将跨格式鲁棒性定义为模型在不同格式中一致回答同一基础问题的程度。然后，我们比较了全格式训练与FormatMix，后者仅将训练项目的一个子集扩展为多个等价格式，使用随机或有针对性的选择。在GLM4和Llama-3.1上，多格式监督始终改善了任务性能和跨格式鲁棒性，而仅使用多项选择题（MCQ）监督则几乎没有好处，甚至可能降低鲁棒性。我们进一步发现，仅将约30%的训练集扩展为多种格式通常能够恢复全格式训练的大部分增益，这一效果在我们研究的模型系列和规模中均有所体现。这些结果表明，格式多样性而非单纯的额外监督是鲁棒性的关键驱动因素。轻量级的多格式增强是一种实用的方法，可以使大型语言模型对答案格式的敏感性降低，而无需改变基础模型。
cs.CL / 45 / 2606.11678
Can AI Reason Like an Urban Planner? Benchmarking Large Language Models Against Professional Judgment
人工智能能像城市规划师一样推理吗？大型语言模型与专业判断的基准测试
Abstract
Problem, Research Strategy, and Findings: The rise of large language models (LLMs) raises a key question for urban planning: which forms of professional planning knowledge can AI replicate, and which still require human judgment? Although AI tools are increasingly used in planning practice, there is still no systematic framework for testing whether they can reason with the contextual sensitivity, value awareness, and institutional literacy central to planning expertise. This paper introduces Urban Planning Bench (UPBench), a domain-specific evaluation framework that assesses LLM reasoning through a 4x5 matrix of four knowledge pillars and five cognitive levels adapted from Bloom's revised taxonomy. Evaluating 25 LLMs with automated scoring and expert review, we find a non-monotonic cognitive curve: models perform better on higher-order analytical tasks than on factual recall and integrative judgment. This suggests that planning knowledge often treated as lower-order is deeply shaped by institutional, jurisdictional, and temporal context, making it hard for LLMs to generalize. We summarize these limits as four epistemic diagnostics: regulatory hallucination, conceptual conflation, wickedness paralysis, and phronetic deficit. Takeaway for Practice: The findings support differential delegation in planning. LLMs can assist with cross-disciplinary synthesis, literature review, scenario generation, and preliminary policy analysis. However, they remain unreliable for jurisdiction-specific regulation, normative conflict resolution, and context-sensitive procedure. Agencies should require verification for AI-assisted regulatory analysis, while planning education should emphasize institutional literacy, normative judgment, and contextual sensitivity.
Chinese Translation
问题、研究策略与发现：大型语言模型（LLMs）的兴起引发了一个城市规划的关键问题：哪些专业规划知识可以被人工智能复制，哪些仍然需要人类判断？尽管人工智能工具在规划实践中越来越多地被使用，但目前仍没有系统的框架来测试它们是否能够以与规划专业知识核心相关的情境敏感性、价值意识和制度素养进行推理。本文介绍了城市规划基准（Urban Planning Bench, UPBench），这是一个特定领域的评估框架，通过一个由四个知识支柱和五个认知层次组成的4x5矩阵来评估LLM的推理能力，这些层次是根据布鲁姆修订的分类法进行调整的。通过自动评分和专家评审对25个LLM进行评估，我们发现了一条非单调的认知曲线：模型在高阶分析任务上的表现优于事实回忆和综合判断。这表明，通常被视为低阶的规划知识深受制度、管辖和时间背景的影响，使得LLM难以进行概括。我们将这些局限总结为四个认识诊断：监管幻觉、概念混淆、复杂问题瘫痪和实践缺失。实践启示：研究结果支持规划中的差异化委托。LLMs可以协助跨学科综合、文献回顾、情景生成和初步政策分析。然而，它们在特定管辖区的法规、规范性冲突解决和情境敏感程序方面仍然不可靠。机构应要求对AI辅助的监管分析进行验证，而规划教育应强调制度素养、规范判断和情境敏感性。
cs.CL / 46 / 2606.11681
UR-BERT: Scaling Text Encoders for Massively Multilingual TTS Through Universal Romanization and Speech Token Prediction
UR-BERT：通过通用罗马化和语音标记预测扩展大规模多语言文本转语音编码器
Abstract
We propose UR-BERT, a Romanized transcription-based text-to-speech (TTS) encoder for massively multilingual TTS systems. Conventional grapheme-to-phoneme (G2P)-based approaches are limited to around 100 languages due to the availability of reliable G2P resources. In contrast, UR-BERT scales to 495 languages by unifying diverse writing systems into a shared Romanization representation. To further enhance phonetic fidelity and text-speech alignment, we introduce a speech token prediction objective during training, which encourages the encoder to learn speech-aware phonetic representations in a data-efficient manner. Experiments show that TTS systems built on UR-BERT consistently outperform recent text encoder baselines across a wide range of languages and resource conditions, and demonstrate strong generalization to unseen languages.
Chinese Translation
我们提出了UR-BERT，一种基于罗马化转录的文本转语音（TTS）编码器，旨在支持大规模多语言TTS系统。传统的基于图形到音素（G2P）的方法由于可靠的G2P资源的可用性，限制在约100种语言内。相比之下，UR-BERT通过将多样的书写系统统一为共享的罗马化表示，扩展到495种语言。为了进一步增强音韵的保真度和文本与语音的对齐，我们在训练过程中引入了语音标记预测目标，鼓励编码器以数据高效的方式学习与语音相关的音韵表示。实验表明，基于UR-BERT构建的TTS系统在各种语言和资源条件下始终优于近期的文本编码器基线，并展示了对未见语言的强泛化能力。
cs.CL / 47 / 2606.11686
Layer-Isolated Evaluation: Gating the Deterministic Scaffold of a Production LLM Agent with a No-LLM, Regression-Locked Test Harness
层隔离评估：通过无LLM、回归锁定的测试工具对生产LLM代理的确定性支架进行门控
Abstract
End-to-end task-success is the dominant way to evaluate LLM agents, but one aggregate number tells you that an agent regressed, not where. We present layer-isolated evaluation: a deployed ordering agent is decomposed into a fixed taxonomy of layers (ontology, intent, routing, decomposition, escalation, safety, memory, and cross-cutting envelope/defense), each exercised by its own assertion slice in a deterministic, no-LLM "pure" mode. The pure suite (238 cases across 23 slices; 225 run in 2.39 s, ~10 ms/case) runs in CI on every change against a locked per-slice baseline. We validate by controlled regression injection, degrading one layer at a time across seven non-safety layers. The effect we did not design in is masking: the aggregate pass-rate barely moves (-1.7 to -5.9 pp for six local regressions), while the matching slice craters (-25 to -91 pp). A layer's slice reacting to its own fault is partly by construction; the measured results are (i) the aggregate masking and (ii) that damage stays off the other slices: the injected layer's slice is the single worst-hit in 5 of 7 cases and top-3 in 7 of 7 (mean rank 1.29 of 19). Localization replicates on a second, structurally different tenant (Starbucks SG): all seven matching slices crater, so it is not a single-catalog artifact. We position it as a concrete, deterministic instantiation of the component-level evaluation EDDOps prescribes but leaves unimplemented, with CheckList as ancestor and as the deterministic mirror image of whole-workflow stochastic mutation testing. Our contributions: (a) a fully decomposed, sub-second, no-LLM per-layer harness for a production agent, (b) a coverage-honesty test-adequacy criterion that refuses to score an unexercised layer, and (c) the regression-injection demonstration that per-slice baseline-locked gates localize regressions an aggregate metric masks.
Chinese Translation
端到端任务成功是评估LLM代理的主要方式，但一个汇总数字只能告诉你代理出现了回归，而无法指出具体位置。我们提出了层隔离评估：一个已部署的排序代理被分解为固定的层次分类（本体、意图、路由、分解、升级、安全、记忆和交叉切面/防御），每一层通过其自身的断言切片在确定性、无LLM的“纯”模式下进行测试。纯测试套件（238个案例，涵盖23个切片；225个案例在2.39秒内运行，约10毫秒/案例）在每次更改时在CI中运行，与锁定的每个切片基线进行比较。我们通过控制回归注入进行验证，逐层降级七个非安全层。我们未设计的效果是掩蔽：汇总通过率几乎没有变化（六个局部回归的变化为-1.7到-5.9个百分点），而匹配切片则大幅下降（-25到-91个百分点）。一个层的切片对其自身故障的反应部分是构造所致；测量结果为（i）汇总掩蔽和（ii）损害保持在其他切片之外：注入层的切片在7个案例中有5个是受损最严重的，并且在7个案例中排名前3（19个中的平均排名为1.29）。定位在第二个结构不同的租户（Starbucks SG）上重复进行：所有七个匹配切片均大幅下降，因此这并不是单一目录的伪影。我们将其定位为EDDOps所规定但未实现的组件级评估的具体、确定性实例，CheckList作为其祖先，并作为整个工作流随机变异测试的确定性镜像。我们的贡献包括：（a）一个完全分解的、亚秒级的、无LLM的生产代理每层测试工具；（b）一个覆盖诚实的测试充分性标准，拒绝对未被测试的层进行评分；（c）回归注入演示，表明每个切片基线锁定的门控能够定位汇总指标掩蔽的回归。
cs.CL / 48 / 2606.11688
Goal-Autopilot: A Verifiable Anti-Fabrication Firewall for Unattended Long-Horizon Agents
目标自动驾驶仪：一种可验证的反伪造防火墙，用于无人值守的长时间跨度代理
Abstract
Long-horizon LLM agents are not trusted to run unattended: with no human watching, they confidently report success they never verified. We treat honesty -- bounding what an agent may claim at termination -- as a first-class metric for unattended autonomy, distinct from capability. We present Autopilot, an execution model that makes silent fabricated success structurally impossible rather than merely rarer. Autopilot externalizes all working state into a durable, gated finite-state machine that a scheduler advances one stateless tick at a time; a hard floor forbids any terminal "done" claim whose falsifiable gate did not actually execute and pass. We prove a No-False-Success theorem -- under gate soundness, floor enforcement, and plan coverage, termination implies the goal holds -- whose only trust points are empirically measurable, and show the worst case degrades to an honest stall, never a fabricated success. Because each tick rehydrates only the state machine, per-step context cost is constant in the horizon. Across a 3,150-cell paired corpus (70 tasks $\times$ 3 systems $\times$ 3 models $\times$ 5 seeds, including 50 SWE-bench Lite tasks across 11 OSS repos), Autopilot fabricates on 0.95% of cells [95% CI 0.38--1.62] while Reflexion and StateFlow baselines fabricate on 8.10% [6.48--9.81] and 25.05% [22.48--27.62] respectively. The headline contrast lives in the hard regime: on SWE-bench Lite, the firewall reduces fabrication from 33.7% (StateFlow) to 0.67%, a paired difference of $-33.07$ pp [95% CI $-36.53, -29.73$]. The mechanism is the gate, not the model: all ten Autopilot fabrications come from the strongest model, while two weaker mid-tier models never fabricate across 700 paired cells. The firewall trades coverage for honesty by design -- an honest stall is recoverable; a confident wrong output shipped downstream is not.
Chinese Translation
长时间跨度的LLM代理在无人值守的情况下不被信任：在没有人监视的情况下，它们自信地报告从未验证的成功。我们将诚实性——限制代理在终止时可能声称的内容——视为无人值守自主性的一个重要指标，与能力不同。我们提出了自动驾驶仪（Autopilot），一种执行模型，使得静默伪造成功在结构上变得不可能，而不仅仅是更少见。自动驾驶仪将所有工作状态外部化为一个持久的、受限的有限状态机，调度器每次以无状态的方式推进一个时钟周期；一个硬性底线禁止任何终端“完成”声明，前提是其可伪造的门未实际执行并通过。我们证明了一个无虚假成功定理——在门的健全性、底线执行和计划覆盖的条件下，终止意味着目标成立——其唯一的信任点是可经验测量的，并且显示最坏情况降级为诚实的停滞，而不是伪造的成功。由于每个时钟周期仅重新激活状态机，因此每步的上下文成本在时间跨度内是恒定的。在一个3150单元的配对语料库中（70个任务 × 3个系统 × 3个模型 × 5个种子，包括11个开源仓库中的50个SWE-bench Lite任务），自动驾驶仪在0.95%的单元上伪造成功[95% CI 0.38--1.62]，而Reflexion和StateFlow基线分别在8.10% [6.48--9.81]和25.05% [22.48--27.62]上伪造成功。显著的对比存在于严格的情况下：在SWE-bench Lite上，防火墙将伪造率从33.7%（StateFlow）降低至0.67%，配对差异为$-33.07$ pp [95% CI $-36.53, -29.73$]。机制在于门，而非模型：所有十个自动驾驶仪的伪造均来自最强的模型，而两个较弱的中层模型在700个配对单元中从未伪造成功。防火墙通过设计在覆盖和诚实性之间进行权衡——诚实的停滞是可恢复的；而自信的错误输出被下游使用则不可恢复。
cs.CL / 49 / 2606.11712
Substrate Asymmetry in User-Side Memory: A Diagnostic Framework
用户侧记忆中的基底不对称性：诊断框架
Abstract
User-side memory in LLMs is typically scored as a single "personalization" capability: given a user's history, is the output more user-aware? We show this aggregate metric hides opposite-direction failures. Memory factorises into at least three orthogonal axes -- behavioral consistency (style, voice), factual presence (recall facts in history), and factual absence (abstain when a fact is absent) -- and no single substrate wins all three. Comparing per-user gamma-LoRA (a small LoRA adapter trained on each user's history; gamma denotes per-user, not per-task) against BGE-large dense top-K retrieval on a controlled 50-user synthetic corpus and a real-data probe (LaMP-3), we find gamma-LoRA decisively wins behavioral style while RAG decisively wins factual absence -- and the same query-projection cells in attention layers 21-35 causally load-bear both effects in opposite directions (zeroing those LoRA weights raises absence-probe TPR by +33 pp and drops presence-probe TPR by 20 pp). On the more heavily RLHF-tuned Llama-3.1-8B-Instruct the asymmetry strengthens, not heals: parametric memory's behavioral advantage collapses while its absence-calibration deficit against retrieval widens -- an alignment tax on parametric user-memory. On real-data LaMP-3, gamma-LoRA underperforms a majority baseline; a 9-condition mitigation sweep diagnoses this as instruction-following collapse, not substrate failure (a 9x2 cross-product shows the eval-time {1..5} logit mask drives main_acc to >=0.995 on every recipe), and the best training-time fix replicates bit-identically on Llama. Finally, substrate-selection routing is question-classification, not calibration: a 110M DistilBERT on the question text alone beats every logit-based router. We contribute the diagnostic framework, the diagnosed real-data negative, the alignment-tax replication, and the routing-as-classification finding.
Chinese Translation
在大型语言模型（LLMs）中，用户侧记忆通常被评估为单一的“个性化”能力：给定用户的历史，输出是否更加关注用户？我们展示了这一汇总指标掩盖了相反方向的失败。记忆可以分解为至少三个正交轴——行为一致性（风格、语调）、事实存在（回忆历史中的事实）和事实缺失（在缺少事实时保持沉默）——而没有单一的基底能够在这三者中全部胜出。通过比较每用户的 gamma-LoRA（一个针对每个用户历史训练的小型 LoRA 适配器；gamma 表示每用户，而非每任务）与 BGE-large 稠密 top-K 检索在一个受控的 50 用户合成语料库和一个真实数据探测（LaMP-3）上的表现，我们发现 gamma-LoRA 在行为风格上明显胜出，而 RAG 在事实缺失上明显胜出——同样的查询投影单元在注意力层 21-35 中因果承载着这两种效应，方向相反（将这些 LoRA 权重归零使缺失探测的真阳性率提高了 +33 个百分点，同时使存在探测的真阳性率下降了 20 个百分点）。在经过更重度的强化学习人类反馈（RLHF）调优的 Llama-3.1-8B-Instruct 中，这种不对称性加剧，而非修复：参数记忆的行为优势崩溃，而其在检索中的缺失校准缺陷扩大——这是对参数用户记忆的对齐税。在真实数据 LaMP-3 中，gamma-LoRA 的表现低于大多数基线；通过一个 9 条件的缓解扫描，我们诊断出这是指令跟随崩溃，而非基底失败（一个 9x2 的交叉乘积显示评估时的 {1..5} logit 掩码使主准确率达到 >=0.995），而最佳的训练时间修复在 Llama 上完全复制。最后，基底选择路由是问题分类，而非校准：仅在问题文本上使用的 1.1 亿 DistilBERT 超过了所有基于 logit 的路由器。我们贡献了诊断框架、诊断出的真实数据负面、对齐税的复制，以及将路由视为分类的发现。
cs.CL / 50 / 2606.11744
Hey Chat, Can You Teach Me? Structuring Socratic Dialogue for Human Learning in the Wild
嘿，聊天机器人，你能教我吗？在自然环境中构建苏格拉底对话以促进人类学习
Abstract
Large language models are now widely used for everyday learning, but the underlying interactions are typically unstructured chats rather than following a curriculum. Unlike formal online learning systems, these interactions carry no prior record of the student, so any estimate of what the student already knows must be inferred from the dialogue itself. We show that this gap is not closed by scaling models alone. Frontier and education-tuned LLMs perform poorly when asked to tutor a student over an extended session, because doing so requires three things at once. The tutor must sequence a curriculum, conduct Socratic dialogue, and infer the student's knowledge state from that dialogue. We propose separating these responsibilities. Given a student query, our system constructs a prerequisite knowledge graph in which subtopics are nodes and dependencies are edges, and frames tutoring as deciding which node to teach next and how many dialogue turns to spend on it before moving on. A lightweight PPO policy handles this sequencing decision, while an LLM conducts the Socratic exchange at the chosen node and returns a signal of student progress. Across held-out STEM and non-STEM topics, our PPO-paired tutor outperforms heuristic baselines, frontier general-purpose models, and a model specialised for Socratic dialogue: on both the rate at which students reach full curriculum mastery and the number of turns required. Explicit curriculum structure delivers gains that scaling the underlying model does not.
Chinese Translation
大型语言模型现在被广泛用于日常学习，但其基础交互通常是非结构化的聊天，而不是遵循课程。与正式的在线学习系统不同，这些交互没有学生的先前记录，因此必须从对话本身推断学生已知的内容。我们表明，仅仅通过扩展模型并不能弥补这一差距。当被要求在较长的会话中辅导学生时，前沿和教育调优的LLM表现不佳，因为这需要同时完成三件事情。辅导者必须安排课程，进行苏格拉底对话，并从对话中推断学生的知识状态。我们建议将这些职责分开。针对学生的提问，我们的系统构建一个先决知识图谱，其中子主题为节点，依赖关系为边，并将辅导框架化为决定下一个教授哪个节点以及在此之前需要花费多少对话轮次。一个轻量级的PPO策略处理这一序列决策，而LLM在所选节点进行苏格拉底式交流并返回学生进展的信号。在保留的STEM和非STEM主题中，我们的PPO配对辅导员在学生达到完整课程掌握的速度和所需轮次方面均优于启发式基线、前沿通用模型以及专门针对苏格拉底对话的模型：明确的课程结构带来了扩展基础模型所无法实现的收益。
cs.CL / 51 / 2606.11762
Automated Creativity Evaluation of Language Models Across Open-Ended Tasks
语言模型在开放式任务中的自动化创造力评估
Abstract
Large language models (LLMs) have achieved remarkable progress in language understanding, reasoning, and generation, sparking growing interest in their creative potential. Realizing this potential requires systematic and scalable methods for evaluating creativity across diverse tasks. However, most existing creativity metrics are tightly coupled to specific tasks, embedding domain assumptions into the evaluation process, and limiting scalability and generality. To address this gap, we introduce an automated, domain-agnostic framework for quantifying LLM creativity across open-ended tasks. Our approach separates the measurement apparatus from the creative task itself, enabling scalable, task-agnostic assessment. Divergent creativity is measured using semantic entropy, a reference-free and robust metric for novelty and diversity, validated against human annotations, LLM-based novelty judgments and baseline diversity measures. Convergent creativity is assessed via a novel retrieval-based multi-agent judge framework that delivers context-sensitive evaluation of task fulfilment with over 60% improved efficiency. We validate our framework in three qualitatively distinct domains: problem-solving (MacGyver), research ideation (HypoGen), and creative writing (BookMIA), using a broad suite of LLMs. Empirical results show that our framework reliably captures key facets of creativity, including novelty, diversity, and task fulfilment, and reveal how model properties, such as size, temperature, recency, and reasoning, impact creative performance. Our work establishes a reproducible and generalizable standard for automated LLM creativity evaluation, paving the way for scalable benchmarking and accelerating progress in creative AI.
Chinese Translation
大型语言模型（LLMs）在语言理解、推理和生成方面取得了显著进展，激发了人们对其创造潜力的日益关注。实现这一潜力需要系统化和可扩展的方法来评估不同任务中的创造力。然而，现有的大多数创造力指标与特定任务紧密相关，将领域假设嵌入评估过程，从而限制了可扩展性和普遍性。为了解决这一问题，我们提出了一种自动化的、与领域无关的框架，用于量化LLM在开放式任务中的创造力。我们的方法将测量工具与创造任务本身分离，能够实现可扩展的、与任务无关的评估。发散创造力通过语义熵进行测量，这是一种无参考且稳健的新颖性和多样性指标，经过与人类注释、基于LLM的新颖性判断和基准多样性度量的验证。收敛创造力则通过一种新颖的基于检索的多智能体评估框架进行评估，该框架提供了对任务完成情况的上下文敏感评估，效率提高超过60%。我们在三个质性不同的领域中验证了我们的框架：问题解决（MacGyver）、研究构思（HypoGen）和创意写作（BookMIA），使用了广泛的LLM套件。实证结果表明，我们的框架可靠地捕捉了创造力的关键方面，包括新颖性、多样性和任务完成情况，并揭示了模型属性（如大小、温度、最近性和推理）如何影响创造性表现。我们的工作建立了一个可重复和可推广的自动化LLM创造力评估标准，为可扩展基准测试铺平了道路，并加速了创造性人工智能的发展。
cs.CL / 52 / 2606.11786
Lius: Translation Model Based Instructional Lingustic Using Continual Instruction Tuning In Kupang Malay
Lius：基于持续指令调优的库旁马来语翻译模型教学语言学
Abstract
Large Language Models (LLMs) offer new potential for translation tasks but often experience performance degradation when handling low-resource languages. To address this limitation, we propose an approach for fine-tuning LLMs on a low-resource language, Kupang Malay. Our approach involves designing a set of instructions by leveraging explicit lexical and semantic features from a bilingual dictionary, and introducing Continual Instruction Tuning (CIT), a training paradigm that enables iterative instruction-based training. Experimental results demonstrate that our model, named Lius, yields notable improvements over standard instruction-tuned models by outperforming 4-6 points, and surpassing both Neural Machine Translation (NMT) and Multilingual LLM models by 10-13 points on several evaluation metrics. These findings highlight the potential of our approach to mitigate the reliance on large-scale parallel data in low-resource language translation.
Chinese Translation
大型语言模型（LLMs）为翻译任务提供了新的潜力，但在处理低资源语言时往往会出现性能下降。为了解决这一限制，我们提出了一种在低资源语言库旁马来语上微调LLMs的方法。我们的方法涉及通过利用双语词典中的显性词汇和语义特征设计一组指令，并引入持续指令调优（Continual Instruction Tuning, CIT），这是一种允许迭代指令基础训练的训练范式。实验结果表明，我们的模型Lius在标准指令调优模型上取得了显著提升，超出4-6个点，并在多个评估指标上超越了神经机器翻译（Neural Machine Translation, NMT）和多语言LLM模型10-13个点。这些发现突显了我们的方法在低资源语言翻译中减轻对大规模平行数据依赖的潜力。
cs.CL / 53 / 2606.11806
External Experience Serving in Production LLM Systems: A Deployment-Oriented Study of Quality-Cost Trade-offs
在生产 LLM 系统中服务的外部经验：一种面向部署的质量-成本权衡研究
Abstract
Production LLM systems accumulate reusable operational experience, but the practical deployment issue is not merely whether such experience can help. It is how different serving strategies trade off quality against online cost under realistic constraints. Injecting external experience can improve task quality, yet it also increases prompt burden, latency, and serving pressure. We study \textit{external experience serving} as a deployment-oriented quality-cost trade-off problem. We evaluate this question in a real production moderation setting, with tool-use and GPQA as supporting contrast tasks that expose different output-cost regimes. We compare no-experience baselines, random experience controls, global prompt injection, and retrieval-based selective injection, and analyze both task quality and serving cost. The results show that, once experience becomes case-dependent, selective retrieval provides a stronger operating point than unconditional global injection. They further show that retrieval quality matters more than simply increasing Top-$K$, and that the same serving policy can exhibit substantially different cost-benefit profiles across short-output and decode-heavy regimes. These findings suggest that external experience is best treated as a selective, cost-aware serving decision rather than as a universal add-on. Overall, in the settings studied here, external experience pays off only when both the serving interface and the task-specific cost structure make its quality gains worth the online cost.
Chinese Translation
生产 LLM 系统积累可重用的操作经验，但实际部署问题不仅在于这些经验是否能够提供帮助，而在于在现实约束下，不同的服务策略如何在质量与在线成本之间进行权衡。注入外部经验可以提高任务质量，但同时也增加了提示负担、延迟和服务压力。我们将 extit{外部经验服务} 视为一个面向部署的质量-成本权衡问题。在一个真实的生产审核环境中评估这一问题，工具使用和 GPQA 作为支持对比任务，揭示不同的输出-成本模式。我们比较了无经验基线、随机经验控制、全局提示注入和基于检索的选择性注入，并分析了任务质量和服务成本。结果表明，一旦经验变得依赖于具体案例，选择性检索提供的操作点比无条件的全局注入更强。此外，检索质量比单纯增加 Top-$K$ 更为重要，并且相同的服务策略在短输出和解码重负载模式下可能表现出显著不同的成本效益特征。这些发现表明，外部经验最好被视为一种选择性的、关注成本的服务决策，而不是一种普遍的附加选项。总体而言，在这里研究的环境中，外部经验只有在服务接口和任务特定的成本结构使其质量提升值得在线成本时，才会带来收益。
cs.CL / 54 / 2606.11816
WorldReasoner: Evaluating Whether Language Model Agents Forecast Events with Valid Reasoning
WorldReasoner：评估语言模型代理是否以有效推理预测事件
Abstract
Forecasting real-world events requires language-model agents to reason under uncertainty from incomplete, time-bounded information. Yet evaluating whether agents genuinely forecast requires more than final-answer accuracy: a model may be correct by recalling memorized training facts, citing fabricated evidence, or producing an unsupported causal story. We present WorldReasoner, an evaluation framework for temporally valid event forecasting. Each task gives an agent a resolved forecasting question, a simulated forecast date, and access only to evidence available before that date; after resolution, the framework scores the submitted probability, cited evidence, and optional causal event graph. WorldReasoner reports three complementary axes: outcome quality against resolved answers, evidence quality over cited sources, and reasoning quality against post-resolution hindsight graphs. The benchmark is built by an agentic construction pipeline that generates forecasting questions, collects time-stamped evidence, and builds hindsight reference graphs at scale, yielding 345 resolved tasks derived from 14,141 articles with graphs covering 8,087 extracted events. Across six controlled agent settings, temporally valid retrieval is the strongest driver of outcome accuracy; causal graph construction improves key-event recovery; and correct graph-enabled forecasts are more strongly grounded in key events and relevant sources, yet agents still struggle to convert grounded evidence into calibrated probabilities.
Chinese Translation
预测现实世界事件要求语言模型代理在不确定性下对不完整、时间有限的信息进行推理。然而，评估代理是否真正进行预测不仅仅依赖于最终答案的准确性：一个模型可能通过回忆记忆中的训练事实、引用虚构的证据或生成不支持的因果故事而得出正确答案。我们提出了WorldReasoner，一个用于时间有效事件预测的评估框架。每个任务为代理提供一个已解决的预测问题、一个模拟的预测日期，并仅访问在该日期之前可用的证据；在解决后，该框架对提交的概率、引用的证据和可选的因果事件图进行评分。WorldReasoner报告三个互补的维度：与已解决答案的结果质量、与引用来源的证据质量，以及与后期回顾图的推理质量。该基准通过一个代理构建管道构建，生成预测问题、收集时间戳证据，并大规模构建回顾参考图，产生了345个源自14,141篇文章的已解决任务，图涵盖了8,087个提取事件。在六个受控代理设置中，时间有效的检索是结果准确性的最强驱动因素；因果图构建提高了关键事件的恢复；而正确的图支持预测更强烈地基于关键事件和相关来源，然而代理仍然难以将有根据的证据转化为校准的概率。
cs.CL / 55 / 2606.11875
I Understand How You Feel: Enhancing Deeper Emotional Support Through Multilingual Emotional Validation in Dialogue System
我理解你的感受：通过多语言情感验证增强对话系统中的深层情感支持
Abstract
Emotional validation - explicitly acknowledging that a user's feelings make sense - has proven therapeutic value but has received little computational attention. Emotional validation in dialogue systems can be decomposed into (i) validating response identification, (ii) validation timing detection, and (iii) validating response generation. To support research on all three subtasks, we release M-EDESConv, a 120k English-Japanese multilingual corpus created through hybrid manual and automatic annotation, and M-TESC, a multilingual spoken-dialogue test set. For timing detection, we propose MEGUMI, a Multilingual Emotion-aware Gated Unit for Mutual Integration, that fuses frozen XLM-RoBERTa semantics with language-specific emotion encoders via cross-modal attention and gated fusion. MEGUMI shows superior performance on both the M-EDESConv and M-TESC datasets, both objectively and subjectively. Finally, our EmoValidBench benchmarks of GPT-4.1 Nano and Llama-3.1 8B indicate that current LLMs generate contextually similar and diverse validating responses, but emotional understanding remains a major area for improvement. Project page: https://github.com/zihaurpang/Multilingual-Emotional-Validation
Chinese Translation
情感验证——明确承认用户的感受是合理的——具有显著的治疗价值，但在计算领域受到的关注很少。在对话系统中，情感验证可以分解为（i）验证响应识别，（ii）验证时机检测，以及（iii）验证响应生成。为了支持这三项子任务的研究，我们发布了M-EDESConv，这是一个通过混合手动和自动注释创建的12万条英语-日语多语言语料库，以及M-TESC，一个多语言口语对话测试集。针对时机检测，我们提出了MEGUMI，一种多语言情感感知门控单元，用于相互集成，它通过跨模态注意力和门控融合将冻结的XLM-RoBERTa语义与特定语言的情感编码器相结合。MEGUMI在M-EDESConv和M-TESC数据集上表现出色，无论是客观还是主观评估。最后，我们的EmoValidBench对GPT-4.1 Nano和Llama-3.1 8B的基准测试表明，当前的大型语言模型生成的验证响应在上下文上相似且多样，但情感理解仍然是一个主要的改进领域。项目页面：https://github.com/zihaurpang/Multilingual-Emotional-Validation
cs.CL / 56 / 2606.11897
Notes2Skills: From Lab Notebooks to Certainty-Aware Scientific Agent Skills
Notes2Skills：从实验室笔记到确定性意识科学代理技能
Abstract
Scientific discovery workflows usually contain and rely heavily on lab notes, where researchers record observations, interpret uncertain results, and plan follow-up experiments. Such informative lab notes preserve evolving scientific reasoning and author uncertainty, rather than polished final results exhibited in publications, providing a valuable opportunity for AI to engage in scientific exploration at a more comprehensive and deeper level. However, most prior work on scientific text focuses on papers, protocols, or structured databases, leaving informal laboratory notes underexplored as inputs to AI agents for science. This gap matters because lab notes often intermingle validated observations, tentative judgments, and possible experimental next steps within the same passage. If these signals are conflated, an AI agent may mistake uncertain scientific judgments for confirmed conclusions or executable actions. To this end, we present Notes2Skills, a two-stage framework for turning lab notebooks into verifiable skills for scientific AI agents while preserving the author's certainty. Across seven conditions and three wet-lab sessions, Notes2Skills is the only configuration that neither mistakes uncertain notes for firm instructions nor discards firm ones. We show that certainty preservation is the missing piece between lab notebooks and reliable agent skills, opening a path toward safer AI co-scientist systems.
Chinese Translation
科学发现工作流程通常包含并严重依赖实验室笔记，研究人员在其中记录观察结果、解释不确定的结果并规划后续实验。这些信息丰富的实验室笔记保存了不断演变的科学推理和作者的不确定性，而不是在出版物中展示的经过润色的最终结果，为人工智能在更全面和更深层次上参与科学探索提供了宝贵的机会。然而，以往大多数关于科学文本的研究主要集中在论文、协议或结构化数据库上，导致非正式的实验室笔记作为科学AI代理的输入未得到充分探索。这一差距非常重要，因为实验室笔记通常在同一段落中交织着经过验证的观察、初步判断和可能的实验下一步。如果这些信号混淆，AI代理可能会将不确定的科学判断误认为已确认的结论或可执行的行动。为此，我们提出了Notes2Skills，一个将实验室笔记转化为科学AI代理可验证技能的两阶段框架，同时保留作者的确定性。在七种条件和三次湿实验室会议中，Notes2Skills是唯一一个既不将不确定的笔记误认为明确指令，也不丢弃明确指令的配置。我们展示了确定性保留是实验室笔记与可靠代理技能之间缺失的关键，为更安全的AI协同科学家系统开辟了道路。
cs.CL / 57 / 2606.11898
GraspLLM: Towards Zero-Shot Generalization on Text-Attributed Graphs with LLMs
GraspLLM：基于大型语言模型在文本属性图上的零-shot泛化研究
Abstract
Research on Text-Attributed Graphs (TAGs) has gained significant attention recently due to its broad applications across various real-world data scenarios, such as citation networks, e-commerce platforms, social media, and web pages. Inspired by the remarkable semantic understanding ability of Large Language Models (LLMs), there have been numerous attempts to integrate LLMs into TAGs. However, existing methods still struggle to generalize across diverse graphs and tasks, and their ability to capture transferable graph structural patterns remains limited. To address this, we introduce the GraspLLM, a framework that combines Graph structural comprehension with semantic understanding prowess of LLMs to enhance the cross-dataset and cross-task generalizability. Specifically, we represent node texts from different graphs in a unified semantic space with a frozen general embedding model, on top of which we perform motif-aware contrastive learning across multiple motif-induced adjacency matrices to extract dataset-agnostic structural information. Then, with our proposed optimal contextual subgraph, we extract the most contextually relevant subgraph for each target node and align these subgraphs to the token space of LLM via an alignment projector. Extensive experiments on TAG benchmark datasets spanning diverse domains reveal that GraspLLM consistently outperforms previous LLM-based methods for TAGs, especially in zero-shot scenarios, highlighting its strong generalizability across different datasets and tasks. Our code is available at https://github.com/Heinz217/GraspLLM.
Chinese Translation
文本属性图（TAGs）的研究近年来受到广泛关注，因其在引用网络、电子商务平台、社交媒体和网页等多种现实数据场景中的广泛应用。受到大型语言模型（LLMs）卓越的语义理解能力的启发，已有众多尝试将LLMs整合到TAGs中。然而，现有方法在不同图和任务之间的泛化能力仍然有限，且捕捉可转移图结构模式的能力不足。为了解决这一问题，我们提出了GraspLLM框架，该框架结合了图结构理解与LLMs的语义理解能力，以增强跨数据集和跨任务的泛化能力。具体而言，我们使用一个冻结的通用嵌入模型将来自不同图的节点文本表示为统一的语义空间，并在此基础上通过多种基于图案的邻接矩阵进行图案感知对比学习，以提取与数据集无关的结构信息。然后，利用我们提出的最优上下文子图，我们为每个目标节点提取最相关的上下文子图，并通过对齐投影器将这些子图对齐到LLM的标记空间。对涵盖不同领域的TAG基准数据集进行的广泛实验表明，GraspLLM在零-shot场景下始终优于以往基于LLM的方法，突出其在不同数据集和任务中的强泛化能力。我们的代码可在 https://github.com/Heinz217/GraspLLM 获取。
cs.CL / 58 / 2606.11906
When Does Language Matter? Multilingual Instructions Reveal Step-wise Language Sensitivity in Vision-Language-Action Models
语言何时重要？多语言指令揭示视觉-语言-动作模型的逐步语言敏感性
Abstract
Vision-Language-Action (VLA) models have shown strong performance in language-conditioned robotic manipulation, yet their robustness to linguistic variation remains poorly understood. In this work, we present the first systematic multilingual evaluation of VLA models by translating the LIBERO benchmark into ten languages, revealing severe performance degradation under non-English instructions, with success rates dropping by 30-50%. Through fine-grained analysis of task executions, we find that language influence is highly non-uniform across steps: certain steps exhibit strong language dependence and dominate overall task failure, while others are largely language-agnostic. Based on this insight, we propose a step-wise inference-time intervention that aligns representations according to step language sensitivity, substantially improving performance under linguistic variation. Our results indicate that language robustness in VLA models is fundamentally a step-wise control problem, highlighting the importance of temporally structured analysis for reliable embodied agents.
Chinese Translation
视觉-语言-动作（VLA）模型在语言条件下的机器人操作中表现出强大的性能，但它们对语言变异的鲁棒性仍然缺乏深入理解。在本研究中，我们首次对VLA模型进行了系统的多语言评估，通过将LIBERO基准翻译成十种语言，揭示了在非英语指令下性能的严重下降，成功率下降幅度达到30-50%。通过对任务执行的细致分析，我们发现语言的影响在不同步骤之间高度不均匀：某些步骤表现出强烈的语言依赖性，并主导了整体任务失败，而其他步骤则在很大程度上与语言无关。基于这一见解，我们提出了一种逐步推理时的干预方法，根据步骤的语言敏感性对表示进行对齐，从而在语言变异下显著提高性能。我们的结果表明，VLA模型中的语言鲁棒性从根本上是一个逐步控制问题，强调了对可靠的具身智能体进行时间结构分析的重要性。
cs.CL / 59 / 2606.11910
An Ontology-Guided Multi-Anchor Graph Retrieval Framework for Traffic Legal Liability Determination
一种基于本体的多锚点图检索框架用于交通法律责任判定
Abstract
Traffic law liability determination is critical for assigning legal penalties, requiring the simultaneous identification of interdependent statutory provisions across multiple legal dimensions. However, existing retrieval-augmented generation methods suffer from a multi-dimensional retrieval bottleneck: single axis architectures compress complex legal queries into a single pathway, causing interdependent statutory dimensions to be overlooked. To address this, we propose OMAGR, an ontology-guided framework that decomposes queries into ontology-aligned anchors and executes parallel graph retrieval across each dimension, ensuring independent retrieval across dimensions before fusion. To evaluate the proposed method, we created the TrafficLaw-QA dataset, an expert-validated benchmark dataset containing 200 questions and 527 legal provisions. Results show that TrafficOmni-RAG outperforms baselines on Context Precision and Faithfulness metrics. The findings demonstrate that parallel multi-anchor retrieval effectively resolves the multi-dimensional retrieval bottleneck, offering a promising direction for traffic law liability determination research.
Chinese Translation
交通法律责任的判定对于法律处罚的分配至关重要，这需要在多个法律维度上同时识别相互依赖的法定条款。然而，现有的检索增强生成方法面临多维检索瓶颈：单轴架构将复杂的法律查询压缩为单一路径，导致相互依赖的法定维度被忽视。为了解决这一问题，我们提出了OMAGR，一个基于本体的框架，它将查询分解为与本体对齐的锚点，并在每个维度上执行并行图检索，确保在融合之前各维度之间的独立检索。为了评估所提出的方法，我们创建了TrafficLaw-QA数据集，这是一个经过专家验证的基准数据集，包含200个问题和527个法律条款。结果表明，TrafficOmni-RAG在上下文精确度和忠实度指标上优于基线。研究结果表明，平行多锚点检索有效解决了多维检索瓶颈，为交通法律责任判定研究提供了一个有前景的方向。
cs.CL / 60 / 2606.11926
Toward Generalist Autonomous Research via Hypothesis-Tree Refinement
通过假设树精炼迈向通用自主研究
Abstract
Scientific progress depends on a repeated loop of exploration, experimentation, and abstraction. Researchers test candidate directions, interpret the evidence, and carry the resulting lessons into later attempts. We study how an AI agent can run this loop autonomously over long horizons. We introduce Arbor, a general framework for autonomous research that combines a long-lived coordinator, short-lived executors, and Hypothesis Tree Refinement (HTR), a persistent tree that links hypotheses, artifacts, evidence, and distilled insights across time. The coordinator manages global research strategy over the tree, while executors implement and test individual hypotheses in isolated worktrees. As results return, Arbor updates the tree, propagates reusable lessons, refines the search frontier, and admits verified improvements. This design turns autonomous research from a sequence of local attempts into a cumulative process in which strategy, execution, and evidence are carried across time. We evaluate Arbor under Autonomous Optimization (AO), an operational setting where an agent improves an initial research artifact through iterative experimentation without step-level human supervision. Across six real research tasks in model training, harness engineering, and data synthesis, Arbor achieves the best held-out result on all six tasks, attaining more than 2.5x the average relative held-out gain of Codex and Claude Code under the same task interface and resource budget. On MLE-Bench Lite, Arbor reaches 86.36% Any Medal with GPT-5.5, the strongest result in our comparison.
Chinese Translation
科学进步依赖于探索、实验和抽象的重复循环。研究人员测试候选方向，解释证据，并将得到的经验教训应用于后续尝试。我们研究了一个人工智能代理如何在长时间范围内自主运行这一循环。我们提出了Arbor，一个用于自主研究的通用框架，结合了一个长期存在的协调者、短期存在的执行者，以及假设树精炼（Hypothesis Tree Refinement, HTR），这是一个持久的树形结构，连接了假设、工件、证据和跨时间提炼的见解。协调者管理树上的全球研究策略，而执行者在孤立的工作树中实施和测试单个假设。随着结果的返回，Arbor更新树结构，传播可重用的经验教训，精炼搜索前沿，并接受经过验证的改进。这种设计将自主研究从一系列局部尝试转变为一个累积过程，其中策略、执行和证据跨越时间得以延续。我们在自主优化（Autonomous Optimization, AO）这一操作环境下评估Arbor，在该环境中，代理通过迭代实验改进初始研究工件，而无需逐步的人类监督。在模型训练、工具工程和数据合成的六个实际研究任务中，Arbor在所有六个任务上都取得了最佳的保留结果，达到了Codex和Claude Code在相同任务接口和资源预算下的平均相对保留增益的2.5倍以上。在MLE-Bench Lite上，Arbor在GPT-5.5下达到了86.36%的任何奖牌，这是我们比较中最强的结果。
cs.CL / 61 / 2606.11931
Semantic Grading of Written Answers in Low-Resource Language Bangla Using a Fine-Tuned Lightweight Language Model
使用微调轻量级语言模型对低资源语言孟加拉语书面答案进行语义评分
Abstract
Bangla is among the world's most widely spoken languages, yet it remains underserved in educational NLP research. In many remote and rural regions, access to qualified subject teachers is limited, and written answers are consequently graded largely by hand, restricting timely and consistent feedback. Automatic assessment is challenging because semantically correct responses can vary substantially in surface form. We present a bilingual (Bangla-English) evaluation system designed for low-resource educational settings that prioritizes semantic correctness over lexical overlap. Our approach fine-tunes a lightweight language model to grade each response using the question, reference answer, and student answer, producing a numeric score and concise, context-grounded feedback suitable for classroom deployment. We also construct a synthetic bilingual dataset to enable controlled training and evaluation. Across proprietary and open-source LLMs evaluated under a unified protocol, our QLoRA-tuned Qwen3-8B confirms consistent improvement by producing the most leakage-resistant feedback (RoRa = 0.819) in synthetic evaluation and the strongest agreement with human scores (rho = 0.936, MAE = 0.725) in a dedicated human study.
Chinese Translation
孟加拉语是世界上使用最广泛的语言之一，但在教育自然语言处理（NLP）研究中仍然受到忽视。在许多偏远和农村地区，合格的学科教师数量有限，因此书面答案主要依赖人工评分，这限制了及时和一致的反馈。自动评估面临挑战，因为语义正确的回答在表面形式上可能有很大差异。我们提出了一种双语（孟加拉语-英语）评估系统，旨在低资源教育环境中优先考虑语义正确性而非词汇重叠。我们的方法微调了一种轻量级语言模型，通过使用问题、参考答案和学生答案对每个回答进行评分，生成数值分数和简明的、基于上下文的反馈，适合课堂使用。我们还构建了一个合成双语数据集，以便进行控制训练和评估。在统一协议下评估的专有和开源大语言模型中，我们的QLoRA微调的Qwen3-8B在合成评估中确认了一致的改进，产生了最具抗泄漏性的反馈（RoRa = 0.819），并在专门的人类研究中与人类评分达成了最强的一致性（rho = 0.936，MAE = 0.725）。
cs.CL / 62 / 2606.11945
uva-irlab-conv at SemEval-2026 Task 8: Multi-Turn RAG with Learned Sparse Retrieval and Listwise Reranking
uva-irlab-conv 在 SemEval-2026 任务 8 中的表现：结合学习的稀疏检索和列表重排序的多轮 RAG
Abstract
This report describes our participation in SemEval-2026 Task 8 on multi-turn retrieval and question answering. The task evaluates conversational systems across four domains (finance, cloud documentation, government, Wikipedia), and includes unanswerable queries where the available collection does not contain sufficient evidence to produce a complete response. We propose a multi-turn retrieval-augmented generation pipeline that combines learned sparse retrieval with LLM-based reranking and generation. Using sparse retrieval as the primary retrieval method, we leverage its strong generalization across domains. In addition, we make use of the long-context capabilities of LLMs for conversational query rewriting, pointwise and listwise reranking, and generating the final response, each conditioned on the full conversational history. This multi-step design enables effective integration of conversational context throughout retrieval and generation, improving robustness across domains.
Chinese Translation
本报告描述了我们在 SemEval-2026 任务 8 中的参与，该任务涉及多轮检索和问答。该任务评估了四个领域（金融、云文档、政府、维基百科）中的对话系统，并包括无法回答的查询，这些查询的可用集合中没有足够的证据来生成完整的响应。我们提出了一种多轮检索增强生成管道，该管道结合了学习的稀疏检索与基于大语言模型（LLM）的重排序和生成。我们将稀疏检索作为主要检索方法，利用其在不同领域的强泛化能力。此外，我们利用大语言模型的长上下文能力进行对话查询重写、逐点和列表重排序，以及生成最终响应，每一步均基于完整的对话历史。该多步骤设计有效地整合了对话上下文，贯穿于检索和生成过程，提高了在各个领域的鲁棒性。
cs.CL / 63 / 2606.11953
Decoding Multimodal Cues: Unveiling the Implicit Meaning Behind Hateful Videos
解码多模态线索：揭示仇恨视频背后的隐含意义
Abstract
Hateful videos have become prevalent on online platforms, highlighting an urgent need for effective detection. However, existing studies primarily focus on binary classification and fail to provide contextual rationales that reveal the implicit meanings behind these judgments, significantly undermining model explainability. To fill this gap, we aim to achieve explainable hateful video detection, enabling models to provide contextual rationales that integrate relevant evidence and logical reasoning alongside decisions. This approach can comprehensively enhance the understanding of video content and the explainability of the decision-making process. We first introduce two datasets, Ex-HateMM and Ex-ImpliHateVid, for explainable hateful video detection. Each dataset provides fine-grained annotations of multimodal harmful elements, along with contextual rationales. We then propose an Information Augmentation and Reasoning Enhancement (IARE) framework designed for explainable detection. The framework employs an information augmentation phase that leverages the multimodal chain-of-thought to integrate harmful elements, thereby enriching rationale evidence. Additionally, IARE incorporates a reasoning enhancement phase, in which Direct Preference Optimization guides the model toward correct reasoning paths and away from incorrect ones, thereby improving the logical coherence of its justifications. We conduct extensive experiments on the two datasets, comparing multiple baselines with our proposed IARE framework. The results demonstrate that IARE achieves state-of-the-art performance while also generating accurate rationales.
Chinese Translation
仇恨视频在在线平台上日益普遍，凸显了有效检测的迫切需求。然而，现有研究主要集中于二元分类，未能提供揭示这些判断背后隐含意义的上下文理由，显著削弱了模型的可解释性。为填补这一空白，我们旨在实现可解释的仇恨视频检测，使模型能够提供整合相关证据和逻辑推理的上下文理由与决策。这种方法可以全面增强对视频内容的理解以及决策过程的可解释性。我们首先介绍两个数据集，Ex-HateMM和Ex-ImpliHateVid，用于可解释的仇恨视频检测。每个数据集提供了对多模态有害元素的细粒度注释，以及上下文理由。接着，我们提出了一个信息增强与推理增强（Information Augmentation and Reasoning Enhancement, IARE）框架，旨在实现可解释的检测。该框架采用信息增强阶段，利用多模态思维链整合有害元素，从而丰富理由证据。此外，IARE还包含推理增强阶段，其中直接偏好优化（Direct Preference Optimization）引导模型朝向正确的推理路径，避免错误路径，从而提高其论证的逻辑一致性。我们在两个数据集上进行了广泛的实验，将多个基线与我们提出的IARE框架进行了比较。结果表明，IARE在实现最先进性能的同时，也生成了准确的理由。
cs.CL / 64 / 2606.12003
Agreement in Representation Space for Open-Ended Self-Consistency
开放式自一致性中的表示空间一致性
Abstract
Self-consistency improves LLM reasoning by sampling multiple outputs and selecting the most consistent answer, but existing formulations largely rely on exact matching and therefore remain limited to tasks with categorical outputs. In this work, we study self-consistency in open-ended generation tasks such as code synthesis and text summarization. We hypothesize that consistency can be understood as a geometric property of the generation space, where semantically compatible generations concentrate in similar regions of representation space. To study this hypothesis, we introduce Embedding-Based Agreement (EBA), a simple training-free operationalization that estimates agreement by clustering sampled generations in embedding space. Through experiments on mathematical reasoning, code generation, and summarization, we show that agreement in representation space provides a robust and scalable signal of self-consistency for open-ended tasks. In particular, EBA consistently outperforms random selection and exhibits more stable scaling behavior than recent selection approaches based on LLM evaluation or uncertainty estimation. We further show that these agreement signals remain stable across model families and embedding spaces, even with native hidden representations. Finally, our analysis shows that the geometric location occupied by sampled generations is strongly correlated with generation quality: generations concentrated near central regions of representation space tend to correspond to more reliable outputs, whereas peripheral generations are substantially less accurate. Overall, our findings support viewing self-consistency as a property of the geometric organization of sampled generations rather than exact symbolic overlap.
Chinese Translation
自一致性通过采样多个输出并选择最一致的答案来改善大型语言模型（LLM）的推理能力，但现有的公式主要依赖于精确匹配，因此仅限于具有分类输出的任务。在本研究中，我们探讨了在代码合成和文本摘要等开放式生成任务中的自一致性。我们假设一致性可以理解为生成空间的几何特性，其中语义上兼容的生成在表示空间的相似区域内集中。为了研究这一假设，我们引入了基于嵌入的一致性（Embedding-Based Agreement, EBA），这是一种简单的无训练操作化方法，通过在嵌入空间中对采样生成进行聚类来估计一致性。通过在数学推理、代码生成和摘要生成上的实验，我们表明表示空间中的一致性为开放式任务提供了一个稳健且可扩展的自一致性信号。特别是，EBA在性能上始终优于随机选择，并且在规模扩展行为上比基于LLM评估或不确定性估计的最新选择方法更为稳定。我们进一步表明，这些一致性信号在不同模型系列和嵌入空间中保持稳定，即使在原生隐藏表示下也是如此。最后，我们的分析显示，采样生成所占据的几何位置与生成质量之间存在强相关性：集中在表示空间中心区域的生成往往对应于更可靠的输出，而边缘生成的准确性则显著较低。总体而言，我们的研究结果支持将自一致性视为采样生成的几何组织特性，而非精确的符号重叠。
cs.CL / 65 / 2606.12068
StanceNakba Shared Task: Actor and Topic-Aware Stance Detection in Public Discourse
StanceNakba共享任务：公共话语中的参与者和主题感知立场检测
Abstract
We present StanceNakba 2026, a shared task on stance detection in polarized social media discourse related to the Palestinian-Israeli conflict, organized as part of Nakba-NLP 2026 at LREC-COLING 2026. The task introduces two subtasks: Subtask A (Actor-Level Stance Detection), which classifies English social media posts as Pro-Palestine, Pro-Israel, or Neutral; and Subtask B (Cross-Topic Stance Detection), which identifies Favor, Against, or Neither stances in Arabic posts toward two conflict-related topics, normalization with Israel and refugee presence in Jordan. The task is grounded in an annotated dataset of 2,606 social media posts. A total of 7 teams participated in Subtask A and 6 teams in Subtask B. Participating systems primarily fine-tuned Arabic and multilingual transformer-based models, including MARBERT, AraBERT, and DeBERTa-v3 variants, with several teams employing cross-validation, ensemble methods, and topic-conditioned architectures. The best-performing systems achieved a Macro F1 of 0.9620 on Subtask A and 0.8724 on Subtask B, demonstrating that transformer-based approaches are highly effective for conflict-domain stance detection while highlighting persistent challenges in cross-topic generalization and neutral class prediction.
Chinese Translation
我们提出了StanceNakba 2026，这是一个关于与巴以冲突相关的极化社交媒体话语中的立场检测的共享任务，作为Nakba-NLP 2026的一部分，在LREC-COLING 2026上组织。该任务引入了两个子任务：子任务A（参与者级立场检测），将英语社交媒体帖子分类为支持巴勒斯坦、支持以色列或中立；子任务B（跨主题立场检测），识别阿拉伯语帖子对两个与冲突相关主题（与以色列的正常化和约旦的难民存在）的支持、反对或中立立场。该任务基于一个包含2,606个社交媒体帖子的注释数据集。共有7个团队参与了子任务A，6个团队参与了子任务B。参与系统主要对阿拉伯语和多语言的基于变换器的模型进行了微调，包括MARBERT、AraBERT和DeBERTa-v3变体，多个团队采用了交叉验证、集成方法和主题条件架构。表现最佳的系统在子任务A上达到了0.9620的宏F1值，在子任务B上达到了0.8724，表明基于变换器的方法在冲突领域的立场检测中非常有效，同时突显了跨主题泛化和中立类别预测中的持续挑战。
cs.CL / 66 / 2606.12087
FORT-Searcher: Synthesizing Shortcut-Resistant Search Tasks for Training Deep Search Agents
FORT-Searcher：合成抗捷径搜索任务以训练深度搜索代理
Abstract
Training deep search agents requires verifiable questions whose answers remain unavailable until sufficient evidence has been acquired through search. Existing synthesis methods often increase apparent difficulty by enriching graph structures, but structural complexity alone does not guarantee realized search difficulty: the intended search process can collapse through a cheaper identifying route. We formalize this gap with a shortcut-aware difficulty framework and identify four actionable shortcut risks: evidence co-coverage, single-clue selectivity, exposed constants, and prior-knowledge binding. To diagnose their realized effects, we use trajectory signatures including solving cost, answer hit time, and prior-shortcut rate. Guided by this framework, we introduce FORT, a Framework of Shortcut-Resistant Training-Data Synthesis. FORT constructs shortcut-resistant training data by controlling shortcut risks across entity selection, evidence graph construction, question formulation, and adversarial refinement. Experiments show that FORT induces longer pre-answer search and fewer shortcut patterns than existing open-source deep search datasets. Using the resulting trajectories, we train FORT-Searcher with supervised fine-tuning (SFT) only, and it achieves the best overall performance among comparable-size open-source search agents on challenging deep search benchmarks. Relevant resources will be made available at https://github.com/RUCAIBox/FORT-Searcher.
Chinese Translation
训练深度搜索代理需要可验证的问题，其答案在通过搜索获得足够证据之前是不可用的。现有的合成方法通常通过丰富图结构来增加表面难度，但结构复杂性并不能保证实际的搜索难度：预期的搜索过程可能通过更便宜的识别路径崩溃。我们通过一个考虑捷径的难度框架来形式化这一差距，并识别出四种可操作的捷径风险：证据共同覆盖、单线索选择性、暴露常量和先验知识绑定。为了诊断它们的实际影响，我们使用轨迹特征，包括解决成本、答案命中时间和先前捷径率。在这一框架的指导下，我们引入了FORT，一个抗捷径训练数据合成框架。FORT通过控制实体选择、证据图构建、问题表述和对抗性细化中的捷径风险，构建抗捷径的训练数据。实验表明，FORT引导的预答案搜索时间更长，捷径模式更少，相较于现有的开源深度搜索数据集。利用生成的轨迹，我们仅通过监督微调（SFT）训练了FORT-Searcher，并在具有挑战性的深度搜索基准测试中，在同等规模的开源搜索代理中实现了最佳整体性能。相关资源将可在 https://github.com/RUCAIBox/FORT-Searcher 获取。
cs.CL / 67 / 2606.12088
Debiasing Without Protected Attributes: Latent Concept Erasure from Textual Profiles
无保护属性的去偏见：从文本特征中消除潜在概念
Abstract
Most fairness research in NLP assumes direct access to protected attributes such as gender, race, or nationality. In practice, however, such information is often unavailable due to privacy constraints, missing metadata, or legal restrictions, even though models may infer it from indirect textual cues. This raises a key question: can debiasing succeed without direct access to sensitive attributes? We propose H-SAL, which performs post-hoc concept and attribute erasure using self-description text as an implicit debiasing signal. To support this setting, we introduce a multi-domain Stack Exchange-based fairness benchmark for helpfulness prediction that includes both explicit and implicit signals, enabling comparison between standard debiasing with protected labels and debiasing without access to sensitive information. Across encoder and decoder-only language models, we find that implicit self-description often matches or outperforms explicit-label-based debiasing. Our results broaden representation-level fairness research and provide a new benchmark for studying debiasing under realistic data constraints.
Chinese Translation
大多数自然语言处理中的公平性研究假设可以直接获取性别、种族或国籍等受保护属性。然而，在实际应用中，由于隐私限制、缺失的元数据或法律限制，这些信息往往不可用，尽管模型可能会从间接的文本线索中推断出这些属性。这引发了一个关键问题：在没有直接访问敏感属性的情况下，去偏见是否能够成功？我们提出了H-SAL，它利用自我描述文本作为隐式去偏见信号，进行事后概念和属性的消除。为了支持这一设置，我们引入了一个基于多领域Stack Exchange的公平性基准，用于帮助预测，包括显式和隐式信号，使得可以比较使用受保护标签的标准去偏见与在没有访问敏感信息的情况下进行的去偏见。在编码器和仅解码器的语言模型中，我们发现隐式自我描述往往与基于显式标签的去偏见相匹配或表现更好。我们的结果拓宽了表示层面的公平性研究，并为在现实数据限制下研究去偏见提供了新的基准。
cs.CL / 68 / 2606.12113
Augmenting Molecular Language Models with Local $n$-gram Memory
通过局部 $n$-gram 记忆增强分子语言模型
Abstract
Transformer-based language models for SMILES strings suffer from a locality gap: standard character-level tokenization fragments chemically meaningful motifs, forcing models to repeatedly learn local syntax at the expense of long-range dependencies. To address this without disrupting standard tokenizers, we propose MolGram, which integrates a conditional $n$-gram memory module into molecular language models. MolGram maps local string patterns to learned embeddings via scalable hash lookups and dynamically injects this regional context into hidden states. Evaluations across three tasks, including unconditional molecule generation, forward reaction prediction, and single-step retrosynthesis, show that MolGram consistently improves performance. Crucially, our analyses demonstrate that MolGram outperforms baselines with 3$\times$ more parameters, establishing explicit local pattern memory as a highly efficient inductive bias.
Chinese Translation
基于 Transformer 的 SMILES 字符串语言模型存在局部性差距：标准字符级标记化将化学上有意义的模式进行碎片化，迫使模型反复学习局部语法，从而牺牲了长程依赖。为了解决这个问题而不干扰标准标记器，我们提出了 MolGram，它将条件 $n$-gram 记忆模块集成到分子语言模型中。MolGram 通过可扩展的哈希查找将局部字符串模式映射到学习的嵌入，并动态地将这一区域上下文注入到隐藏状态中。在包括无条件分子生成、正向反应预测和单步逆合成在内的三个任务的评估中，MolGram 一致性地提高了性能。关键是，我们的分析表明，MolGram 在参数量是基线的 3 倍的情况下仍然表现更优，确立了显式局部模式记忆作为一种高效的归纳偏置。
cs.CL / 69 / 2606.12114
Detecting Sensitive Personal Information in Japanese Pre-Training Corpora for Large Language Models
在大型语言模型的日语预训练语料库中检测敏感个人信息
Abstract
Sensitive personal information can appear in large-scale pre-training corpora for large language models (LLMs). Detecting and filtering such information is therefore essential to ensure compliance with privacy regulations and prevent unintended information leakage. However, in contrast to English and other languages, research into sensitive personal information has been limited in the Japanese language. In this study, we focus on sensitive personal data defined as special care-required personal information (SCPI) under Japan's Act on the Protection of Personal Information (APPI). We construct an SCPI dataset using LLM-based annotation and train machine learning models to rapidly detect SCPI in text. As a result, our SCPI classifier can effectively identify information related to SCPI. This study is the first to explore SCPI detection in Japanese text corpora, highlighting the challenges of accurate detection.
Chinese Translation
敏感个人信息可能出现在大型语言模型（LLMs）的规模庞大的预训练语料库中。因此，检测和过滤这些信息对于确保遵守隐私法规和防止意外信息泄露至关重要。然而，与英语和其他语言相比，关于敏感个人信息的研究在日语中相对有限。在本研究中，我们关注根据日本《个人信息保护法》（APPI）定义的特殊关怀个人信息（SCPI）。我们利用基于LLM的标注构建了一个SCPI数据集，并训练机器学习模型以快速检测文本中的SCPI。因此，我们的SCPI分类器能够有效识别与SCPI相关的信息。本研究首次探讨了日语文本语料库中的SCPI检测，突显了准确检测的挑战。
cs.CL / 70 / 2606.12117
Soft-Prompt Tuning for Fair and Efficient LLM Benchmark Evaluation
用于公平和高效的大语言模型基准评估的软提示调优
Abstract
Benchmark scores often misrepresent a large language model's (LLM's) knowledge, because they rely, e.g., on the model's ability to follow specific formatting requirements. This especially penalizes base models that may know the correct answers but lack the ability -- typically introduced in post-training -- to structure them as instructed. To overcome this, we propose soft-prompt tuning, an efficient, fair, and architecture-agnostic model evaluation. By optimizing only 10 soft-prompt vectors (roughly 0.0006% parameters for a 7B model) over a short tuning period, we adapt models to specific benchmark formats, closing gaps in format-following and ensuring that underlying knowledge is accurately reflected in benchmark scores. This allows one to fairly compare different base models -- trained with various pre-training recipes -- on benchmarks without the need for full post-training. We evaluated soft-prompt tuning across 7 models and 7 datasets. The results show that (a) soft-prompt tuning saturates format-following within 80 steps (~640 samples) making it highly efficient, (b) soft-prompt tuning significantly outperforms zero- and few-shot prompting, surfacing base model knowledge that standard prompting misses, that (c) even post-trained models can benefit from soft-prompts to maximize format compliance, and that (d) soft-prompted base model performance predicts post-trained model rankings more reliably than zero- and few-shot baselines, offering a low-cost proxy for downstream model quality. Our contributions include (1) metrics which disentangle format-following and knowledge accuracy, (2) a fairer benchmarking protocol of LLM knowledge, and (3) a cost- and memory-effective recipe to identify optimal pre-training strategies early in LLM development.
Chinese Translation
基准分数往往无法准确反映大语言模型（LLM）的知识，因为它们依赖于模型遵循特定格式要求的能力。这尤其对那些可能知道正确答案但缺乏在后期训练中通常引入的能力以按指示结构化答案的基础模型造成惩罚。为了解决这个问题，我们提出了软提示调优，这是一种高效、公平且与架构无关的模型评估方法。通过在短暂的调优期间仅优化10个软提示向量（对于一个70亿参数的模型，约占0.0006%的参数），我们使模型适应特定的基准格式，缩小格式遵循的差距，并确保基础知识在基准分数中得到准确反映。这使得我们能够在基准测试中公平地比较不同的基础模型——这些模型使用各种预训练方案进行训练，而无需进行全面的后期训练。我们在7个模型和7个数据集上评估了软提示调优。结果显示：（a）软提示调优在80步内（约640个样本）饱和格式遵循，使其高效性极高；（b）软提示调优显著优于零-shot和少量-shot提示，揭示了标准提示所遗漏的基础模型知识；（c）即使是经过后期训练的模型也可以从软提示中受益，以最大化格式合规性；（d）软提示的基础模型性能比零-shot和少量-shot基线更可靠地预测后期训练模型的排名，提供了一种低成本的下游模型质量代理。我们的贡献包括（1）解构格式遵循和知识准确性的度量标准，（2）更公平的LLM知识基准测试协议，以及（3）一种成本和内存有效的方案，以便在LLM开发早期识别最佳的预训练策略。
cs.CL / 71 / 2606.12160
A Controlled Study of Decoding-Time Truthfulness Methods on Instruction-Tuned LLMs
关于指令调优大语言模型的解码时间真实性方法的对照研究
Abstract
In this work, we introduce CHAIR (Classifier of Hallucination As ImproveR), a supervised framework for detecting hallucinations by analyzing internal logits from each layer of every token. Our method extracts a compact set of features such as maximum, minimum, mean, standard deviation, and slope-from the token logits across all layers, enabling effective hallucination detection without overfitting. Experiments on TruthfulQA and MMLU datasets demonstrate that CHAIR significantly improves detection accuracy, particularly in zero-shot scenarios, showcasing its robustness and generalizability. Beyond hallucination detection, CHAIR highlights the potential of using internal representations for designing advanced decoding strategies. By leveraging patterns in logits, we suggest that more sophisticated models and adaptive decoding methods could further reduce hallucinations and enhance text completion quality. CHAIR not only offers a practical solution for detecting hallucinations but also lays the groundwork for exploring richer representations in LLMs to improve their factuality and coherence.
Chinese Translation
在本研究中，我们介绍了 CHAIR（幻觉分类器作为改进者），这是一个通过分析每个令牌每层的内部 logits 来检测幻觉的监督框架。我们的方法从所有层的令牌 logits 中提取了一组紧凑的特征，如最大值、最小值、均值、标准差和斜率，从而实现有效的幻觉检测而不发生过拟合。在 TruthfulQA 和 MMLU 数据集上的实验表明，CHAIR 显著提高了检测准确性，特别是在零样本场景中，展示了其鲁棒性和泛化能力。除了幻觉检测，CHAIR 还强调了利用内部表示设计先进解码策略的潜力。通过利用 logits 中的模式，我们建议更复杂的模型和自适应解码方法可能进一步减少幻觉并提高文本完成质量。CHAIR 不仅为检测幻觉提供了实用的解决方案，还为探索大语言模型中更丰富的表示奠定了基础，以改善其真实性和连贯性。
cs.CL / 72 / 2606.12186
A Resource for Enthymeme Detection in Controversial Political Discourse
一个用于检测争议政治话语中的隐含论证的资源
Abstract
Enthymemes, arguments with unstated premises or conclusions, are pervasive in persuasive discourse, yet their annotation remains notoriously subjective. We present a resource of 1,482 tweets from politically controversial discourse, annotated by five annotators for the presence of enthymemes and their argument structure, designed to study label variation. We first revisit the definition of enthymemes and propose annotation guidelines anchored in Walton's argumentation schemes, offering a structured and constrained approach that nonetheless preserves room for the interpretive nature of the task. This contrasts with past resources, which tend to eliminate disagreement, obscuring its sources and preventing investigation of its potential benefits for model performance. We further propose a complexity analysis of the task, identifying where annotation imposes high cognitive load and may give rise to inconsistent annotation. Our preliminary experiments show that models trained on annotator disagreement outperform models trained on hard majority-vote labels. We close by reflecting on how structural openness in enthymeme definitions and guidelines enables the study of variation in subjective inferential processes for future resources and downstream NLP applications concerned with human inference.
Chinese Translation
隐含论证是指那些未明确提出前提或结论的论证，它们在说服性话语中普遍存在，但其标注仍然 notoriously 主观。我们提供了一个包含1,482条来自政治争议话语的推文的资源，这些推文由五位标注者对隐含论证及其论证结构进行了标注，旨在研究标签变异。我们首先重新审视隐含论证的定义，并提出基于Walton的论证方案的标注指南，提供了一种结构化且受限的方法，同时保留了任务解释性质的空间。这与过去的资源形成对比，后者往往消除分歧，模糊其来源，并阻碍对其对模型性能潜在益处的研究。我们进一步提出了任务的复杂性分析，识别出标注在何处施加了高认知负荷，并可能导致标注不一致。我们的初步实验表明，基于标注者分歧训练的模型优于基于严格多数投票标签训练的模型。最后，我们反思隐含论证定义和指南中的结构开放性如何促进未来资源和关注人类推理的下游自然语言处理应用中主观推理过程变异的研究。
cs.CL / 73 / 2606.12191
Agentic Environment Engineering for Large Language Models: A Survey of Environment Modeling, Synthesis, Evaluation, and Application
大型语言模型的代理环境工程：环境建模、合成、评估与应用的综述
Abstract
Environments serve as interactive systems for large language model (LLM) based agents across diverse scenarios and play a crucial role in driving the continual evolution of model capabilities. Despite this importance, existing work lacks a systematic categorization and deep analysis. This paper systematically studies current researches on agentic environments from the perspective of the environment engineering lifecycle, covering their modeling, synthesis, evaluation and application. Specifically, the paper first introduces representative environments from the perspectives of eight attributes and eight domains, providing detailed analyses of their development paths and highlighting their core capabilities. Second, for automated environment synthesis, two paradigms are introduced, such as symbolic synthesis and neural synthesis. This paper also shows different environment evaluation methods in each paradigm. Thirdly, the corresponding environment applications from the perspective of agent-environment co-evolution are discussed. In specific, the paper characterizes the primary pathways for agent evolution in dynamic environments from four complementary perspectives: memory-centric experience evolution, orchestration-centric workflow evolution, trajectory-centric offline evolution, and exploration-centric online evolution. And three paradigms of environment evolution are identified, namely neural-driven, difficulty-driven, and scaling-driven approaches. At last, several promising future directions are discussed, including Environment-as-a-Service, Multi-agent Environments, and Neural-Symbolic Environments.
Chinese Translation
环境作为大型语言模型（LLM）基础代理在多种场景下的交互系统，发挥着推动模型能力持续演进的重要作用。尽管这一重要性显而易见，现有研究却缺乏系统的分类和深入分析。本文从环境工程生命周期的角度系统研究了当前关于代理环境的研究，涵盖其建模、合成、评估和应用。具体而言，本文首先从八个属性和八个领域的视角介绍了具有代表性的环境，详细分析了它们的发展路径，并突出了其核心能力。其次，对于自动化环境合成，介绍了两种范式，如符号合成（symbolic synthesis）和神经合成（neural synthesis）。本文还展示了每种范式下不同的环境评估方法。第三，从代理-环境共演化的角度讨论了相应的环境应用。具体而言，本文从四个互补的视角描述了动态环境中代理演化的主要路径：以记忆为中心的经验演化、以协调为中心的工作流演化、以轨迹为中心的离线演化和以探索为中心的在线演化。同时，识别了环境演化的三种范式，即以神经驱动的、以难度驱动的和以规模驱动的方法。最后，讨论了几个有前景的未来方向，包括环境即服务（Environment-as-a-Service）、多代理环境（Multi-agent Environments）和神经-符号环境（Neural-Symbolic Environments）。
cs.CL / 74 / 2606.12203
Adaptive Multi-Resolution Procedural Knowledge Compression for Large Language Models
大型语言模型的自适应多分辨率过程知识压缩
Abstract
Large language models (LLMs) are widely used to tackle complex tasks with autonomous workflows. Recently, reusable natural language skills have emerged as a popular paradigm to inject procedural knowledge into LLM applications. Since popular skills are often invoked repeatedly, placing their full text in every context significantly increases prefill cost and latency. While text compression techniques have the potential to solve this problem, most existing methods are designed to compress factual knowledge in documents instead of procedural knowledge, making them insufficient for skill compression. In this paper, we argue that an effective skill compression method should: 1) preserve logical dependencies among workflows and tool protocols, 2) enable lightweight, offline compression for frequently updated community skills, and 3) be adaptable to varying complexities across skills. To address this, we present SKIM (SKIll coMpression), an adaptive multi-resolution soft token compression framework for procedural skills. Depending on the complexity of each skill, SKIM creates different numbers of soft tokens that not only improve the efficiency of LLM inference, but also preserve the effectiveness of skill usage. Experiments indicate that SKIM compresses skills to 30 to 60 percent of their original token length while preserving task performance better than existing compression methods.We have released our code at https://github.com/bebr2/SKIM .
Chinese Translation
大型语言模型（LLMs）被广泛用于处理具有自主工作流程的复杂任务。最近，可重用的自然语言技能作为一种流行范式，开始被引入到LLM应用中以注入过程知识。由于流行技能通常会被重复调用，在每个上下文中放置其完整文本会显著增加预填充成本和延迟。尽管文本压缩技术有潜力解决这个问题，但大多数现有方法旨在压缩文档中的事实知识，而非过程知识，因此不足以用于技能压缩。本文认为，一个有效的技能压缩方法应当：1）保留工作流程和工具协议之间的逻辑依赖关系，2）支持对频繁更新的社区技能进行轻量级的离线压缩，3）能够适应不同技能的复杂性。为此，我们提出了SKIM（SKIll coMpression），一种自适应多分辨率软令牌压缩框架，专为过程技能设计。根据每个技能的复杂性，SKIM生成不同数量的软令牌，这不仅提高了LLM推理的效率，还保留了技能使用的有效性。实验表明，SKIM将技能压缩至原始令牌长度的30%至60%，同时在任务性能上优于现有的压缩方法。我们的代码已发布在 https://github.com/bebr2/SKIM 。
cs.CL / 75 / 2606.12210
Can News Predict the Market? Limits of Zero-Shot Financial NLP and the Role of Explainable AI
新闻能预测市场吗？零样本金融自然语言处理的局限性与可解释人工智能的角色
Abstract
Can financial news reliably predict short-term stock movements? Despite advances in large language models, this question remains unresolved. We revisit this problem using a zero-shot natural language processing framework, investigating whether models can extract actionable signals from financial news without domain-specific training. We design a structured pipeline that combines zero-shot natural language inference with temporal aggregation, explicitly modelling recency and event-dependent impact horizons when integrating information across articles. To address the need for transparency in high-stakes settings, we introduce a multi-layered explainability framework that links predictions to token-level, article-level, and aggregate evidence, and produces grounded natural language rationales. Across multiple models and prediction horizons, we find that zero-shot approaches consistently fail to outperform simple baselines, with particularly weak performance on negative movements, suggesting deeper structural limitations in mapping news sentiment to short-term price dynamics. However, explainability signals reliably distinguish between trustworthy and unreliable predictions, offering practical value even when accuracy is limited. These findings highlight the limits of zero-shot financial NLP and motivate a shift toward decision-support systems that prioritise transparency and uncertainty awareness. Code: https://github.com/alimert05/zero-shot-stock-xai
Chinese Translation
金融新闻能否可靠地预测短期股票波动？尽管大型语言模型取得了进展，这个问题仍未得到解决。我们使用零样本自然语言处理框架重新审视这一问题，探讨模型是否能够在没有领域特定训练的情况下，从金融新闻中提取可操作的信号。我们设计了一个结构化的流程，将零样本自然语言推理与时间聚合相结合，在整合文章信息时明确建模近期性和事件依赖的影响范围。为了满足高风险环境下对透明性的需求，我们引入了一个多层次的可解释性框架，将预测结果与令牌级、文章级和聚合证据联系起来，并生成基于事实的自然语言推理。在多个模型和预测范围内，我们发现零样本方法始终未能超越简单基线，尤其在负向波动上的表现尤为薄弱，这表明在将新闻情绪映射到短期价格动态方面存在更深层次的结构性限制。然而，可解释性信号能够可靠地区分可信和不可信的预测，即使在准确性有限的情况下也提供了实际价值。这些发现突显了零样本金融自然语言处理的局限性，并促使我们转向优先考虑透明性和不确定性感知的决策支持系统。代码链接： https://github.com/alimert05/zero-shot-stock-xai
cs.CL / 76 / 2606.12234
On The Effectiveness-Fluency Trade-Off In LLM Conditioning: A Systematic Study
大型语言模型条件化中的有效性与流畅性权衡：一项系统研究
Abstract
Controlling the output of Large Language Models (LLMs) is a central challenge for their reliable deployment, yet a clear understanding of the involved trade-offs remains elusive. Current approaches to conditioning are often evaluated with a narrow focus on their effectiveness at injecting or removing a target concept, neglecting generation quality. We systematically investigate a range of conditioning methods in both injection and removal scenarios. We find that efficient steering methods frequently achieve conditioning at a steep cost to fluency. Furthermore, we identify a critical yet previously overlooked interaction with the training paradigm: activation steering methods are far less effective on instruction-tuned models than on their base counterparts. Simple prompting and full-fledged supervised fine-tuning, on the other hand, are viable options for concept injection, but are not as good at concept removal. Finally, cheaply computed textual metrics highly correlate to costly LLM-as-judge scores, and provide insights on the behavior of conditioning methods.
Chinese Translation
控制大型语言模型（LLMs）的输出是其可靠部署的核心挑战，但对所涉及权衡的清晰理解仍然难以获得。目前的条件化方法通常仅关注于注入或移除目标概念的有效性，而忽视了生成质量。我们系统地研究了一系列条件化方法在注入和移除场景中的表现。我们发现，高效的引导方法在实现条件化时往往以流畅性为代价。此外，我们还识别出一个关键但之前被忽视的与训练范式的交互：激活引导方法在指令调优模型上的效果远不如在基础模型上的效果。另一方面，简单的提示和全面的监督微调是概念注入的可行选择，但在概念移除方面表现不佳。最后，廉价计算的文本指标与昂贵的LLM作为评判者的评分高度相关，并提供了对条件化方法行为的洞察。
cs.CL / 77 / 2606.12243
VIA-SD: Verification via Intra-Model Routing for Speculative Decoding
VIA-SD：通过模型内部路由进行验证的推测解码
Abstract
Speculative decoding (SD) addresses the high inference costs of LLMs by having lightweight drafters generate candidates for large verifiers to validate in parallel. Existing draft-verify methods use binary decisions: accept or fully recompute. Yet we find that many rejected tokens can be verified correctly by a slim submodel derived from the full verifier via intra-model routing, instead of the full verifier. This motivates our slim-verifier to handle tokens requiring moderate verification resources, reducing expensive large-model calls. We propose Verification via Intra-Model Routing for Speculative Decoding (VIA-SD), a multi-tier framework using a routed slim-verifier. Draft tokens are processed hierarchically: direct acceptance for high-confidence cases, slim-verifier regeneration for medium-confidence cases, and full-model verification for uncertain cases. Across four representative tasks and multiple model families, VIA-SD reduces rejection rates by 0.10-0.22 and delivers 10-20% speedups over strong SD baselines, while achieving 2.5-3x acceleration over non-drafting decoding. Moreover, VIA-SD is compatible with existing SD frameworks without modifying their training procedures. Our results suggest multi-tier SD as a general paradigm for scalable and efficient LLM inference. Project page: https://zju-xyc.github.io/VIA-SD-Project-Page/
Chinese Translation
推测解码（SD）通过让轻量级草拟器生成候选项供大型验证器并行验证，从而解决了大型语言模型（LLMs）高推理成本的问题。现有的草拟-验证方法使用二元决策：接受或完全重计算。然而，我们发现许多被拒绝的标记可以通过从完整验证器派生的精简子模型进行正确验证，而不是使用完整验证器。这促使我们设计精简验证器来处理需要适度验证资源的标记，从而减少昂贵的大模型调用。我们提出了通过模型内部路由进行验证的推测解码（VIA-SD），这是一个使用路由精简验证器的多层框架。草拟标记按层次处理：对于高置信度案例直接接受，对于中等置信度案例进行精简验证器重生成，对于不确定案例进行完整模型验证。在四个代表性任务和多个模型系列中，VIA-SD将拒绝率降低了0.10-0.22，并在强推测解码基线之上实现了10-20%的加速，同时在非草拟解码上实现了2.5-3倍的加速。此外，VIA-SD与现有的推测解码框架兼容，无需修改其训练过程。我们的结果表明，多层推测解码作为可扩展和高效的LLM推理的一般范式。项目页面：https://zju-xyc.github.io/VIA-SD-Project-Page/
cs.CL / 78 / 2606.12250
Reassessing High-Performing LLMs on Polish Medical Exams: True Competence or Bias-Driven Performance?
重新评估高性能大型语言模型在波兰医学考试中的表现：真正的能力还是偏见驱动的表现？
Abstract
Large language models (LLMs) in medicine are mainly evaluated using multiple-choice question answering (MCQA), which can overestimate real clinical ability due to guessing strategies and answer biases. To address these limitations, we introduce an expanded and more challenging benchmark based on Polish medical exams, adding over 15,000 questions, two new domains, and four structural modifications that reduce MCQA-specific artifacts and better test reasoning. We evaluate 21 LLMs and show that evaluation design strongly affects results. Under our harder setup, the best model (Qwen3.5-122B) drops by 28.4 and 31 pp on English and Polish exams, respectively. Despite low evidence of data contamination, standard MCQA scores do not reliably reflect true medical competence. To facilitate further research, we make our benchmark publicly available.
Chinese Translation
医学领域的大型语言模型（LLMs）主要通过多项选择题回答（MCQA）进行评估，这可能由于猜测策略和答案偏见而高估真实的临床能力。为了解决这些局限性，我们引入了一个基于波兰医学考试的扩展和更具挑战性的基准，增加了超过15,000个问题、两个新领域以及四项结构性修改，以减少MCQA特有的伪影并更好地测试推理能力。我们评估了21个LLM，并显示评估设计对结果有显著影响。在我们更严格的设置下，表现最佳的模型（Qwen3.5-122B）在英语和波兰考试中的得分分别下降了28.4和31个百分点。尽管数据污染的证据较少，标准的MCQA得分并不能可靠地反映真实的医学能力。为了促进进一步的研究，我们将我们的基准公开发布。
cs.CL / 79 / 2606.12273
Beyond Fully Random Masking: Attention-Guided Denoising and Optimization for Diffusion Language Models
超越完全随机掩蔽：基于注意力的去噪与优化在扩散语言模型中的应用
Abstract
Diffusion large language models (dLLMs) offer an efficient alternative to autoregressive models through parallel decoding, yet existing post-training methods largely rely on random masking strategies that overlook intrinsic token dependencies. In this work, we present an empirical analysis of attention in dLLMs and show that tokens attending more strongly to unmasked context exhibit greater generation stability and play a critical role in reasoning. Motivated by these findings, we propose AGDO, an attention-guided denoising and optimization framework that aligns both training and optimization with attention-derived dependencies. AGDO determines the denoising order based on attention structure and emphasizes attention-critical tokens during supervised fine-tuning and reinforcement learning. Experiments on mathematical and coding benchmarks demonstrate that AGDO consistently improves reasoning performance, outperforming state-of-the-art post-training methods for dLLMs.
Chinese Translation
扩散大型语言模型（dLLMs）通过并行解码提供了一种高效的替代自回归模型的方法，但现有的后训练方法在很大程度上依赖于随机掩蔽策略，这忽视了内在的标记依赖关系。在本研究中，我们对dLLMs中的注意力进行了实证分析，发现对未掩蔽上下文关注更强的标记表现出更大的生成稳定性，并在推理中发挥了关键作用。基于这些发现，我们提出了AGDO（基于注意力的去噪与优化框架），该框架将训练和优化与注意力衍生的依赖关系对齐。AGDO根据注意力结构确定去噪顺序，并在监督微调和强化学习过程中强调注意力关键标记。在数学和编码基准测试中的实验表明，AGDO持续提升推理性能，超越了dLLMs的最先进后训练方法。
cs.CL / 80 / 2606.12291
Measuring Epistemic Resilience of LLMs Under Misleading Medical Context
在误导性医学背景下测量大型语言模型的认知韧性
Zhou, Hongjian, Zou, Xinyu, Wu, Jinge, Wu, Sean, Yu, Junchi, Segal, Bradley Max, Niebuhr, Tobias Erich, Amro, Sara, Petrus, Michael, Momin, Sheikh, Pinto, Alexandra M. Cardoso, Niesen, Rachel, Wegner, Laura Sophie, Darji, Dhruv, Koo, Jung Moses, Fieggen, Joshua, Narain, Kapil, Zeng, Mingde, Clifton, Lei, Shapiro, Linda, Liu, Fenglin, Clifton, David A.
Abstract
Large language models (LLMs) now reach expert-level scores on medical licensing exams, encouraging the assumption that high scores imply safe medical judgment while patients increasingly use them for health advice. We show this assumption is fragile: when misleading context is injected into questions that LLMs originally answer correctly, they abandon the correct answer. We call the ability to maintain correct judgment under adversarial context epistemic resilience, and introduce MedMisBench to measure it. MedMisBench contains 10,932 medical question items and 48,889 misleading context-option pairs spanning medical reasoning, agentic capability, and patient-journey evaluation. Across 11 model configurations, mean accuracy falls from 71.1% on original questions to 38.0% under focused misleading context, with 51.5% attack success. The most damaging injections are formal, rule-like fabrications: authority-framed falsehoods reach 69.5% attack success and exception-poisoning claims reach 64.1%. A 14-member clinical panel from 7 countries identified serious potential harm in 38.2% of reviewed cases. MedMisBench exposes a structural blind spot in LLM evaluation in medical settings: existing benchmarks measure what models know, but not whether they preserve correct medical judgment under misleading context.
Chinese Translation
大型语言模型（LLMs）在医学执照考试中已达到专家级别的分数，这促使人们假设高分意味着安全的医学判断，而患者越来越多地使用它们获取健康建议。我们表明这一假设是脆弱的：当在原本回答正确的问题中注入误导性背景时，LLMs 会放弃正确答案。我们称这种在对抗性背景下保持正确判断的能力为认知韧性，并引入 MedMisBench 来进行测量。MedMisBench 包含 10,932 个医学问题项和 48,889 对误导性背景-选项对，涵盖医学推理、代理能力和患者旅程评估。在 11 种模型配置中，原始问题的平均准确率从 71.1% 降至在集中误导性背景下的 38.0%，攻击成功率为 51.5%。最具破坏性的注入是正式的、规则般的虚构：以权威为框架的虚假信息达到 69.5% 的攻击成功率，而例外中毒的主张达到 64.1%。来自 7 个国家的 14 名临床专家小组在 38.2% 的审查案例中识别出严重的潜在危害。MedMisBench 揭示了 LLM 在医学环境评估中的结构性盲点：现有基准测量模型所知，但不测量它们在误导性背景下是否保持正确的医学判断。
cs.CL / 81 / 2606.12332
Measuring Semantic Progress in Multi-turn Dialogue via Information Gain
通过信息增益测量多轮对话中的语义进展
Abstract
Evaluating multi-turn dialogue is challenging because quality emerges across turns rather than within individual responses. We focus on a key dimension of information-seeking dialogue: semantic progress, defined as the accumulation of new, question-relevant, and non-redundant information over the course of a conversation. We formalize semantic progress as question-conditioned uncertainty reduction and introduce an information-theoretic metric that approximates it in embedding space. Our main estimator uses a tractable Gaussian formulation with closed-form updates, while a complementary maximum-entropy argument shows why log-determinant structure arises more broadly when only second-order embedding information is retained. This formulation yields desirable theoretical properties, including monotonicity, additive decomposition of total information gain across turns, and diminishing returns for redundant evidence. Unlike LLM-as-a-judge approaches, our metric requires no autoregressive inference at evaluation time and is fully reproducible for a fixed embedding model. Experiments on MT-Bench, Chatbot Arena, and UltraFeedback show that the proposed metric achieves competitive agreement with human judgments despite targeting only semantic progress, with improved alignment on MT-Bench and UltraFeedback compared to several LLM-based judges. Notably, the method remains effective with lightweight embedding models under CPU-only execution, indicating that semantic progress can be captured without reliance on large model capacity.
Chinese Translation
评估多轮对话具有挑战性，因为质量是在多个轮次中逐渐显现的，而不是在单个响应中。我们专注于信息寻求对话的一个关键维度：语义进展，定义为在对话过程中积累的新信息、与问题相关且不冗余的信息。我们将语义进展形式化为基于问题的条件不确定性降低，并引入了一种信息论度量，旨在在嵌入空间中近似这一进展。我们的主要估计器使用可处理的高斯形式，具有闭式更新，而一个补充的最大熵论证则展示了为何在仅保留二阶嵌入信息时，行列式的对数结构会更广泛地出现。这种形式化方法具有理想的理论特性，包括单调性、跨轮次总信息增益的可加分解，以及对冗余证据的递减收益。与将大型语言模型（LLM）作为评判者的方法不同，我们的度量在评估时不需要自回归推理，并且对于固定的嵌入模型是完全可重复的。在 MT-Bench、Chatbot Arena 和 UltraFeedback 上的实验表明，尽管仅针对语义进展，该度量与人类判断之间仍然达成了竞争性的一致性，并在 MT-Bench 和 UltraFeedback 上与多个基于 LLM 的评判者相比，显示出更好的对齐效果。值得注意的是，该方法在仅使用 CPU 执行的轻量级嵌入模型下仍然有效，表明语义进展可以在不依赖大型模型能力的情况下被捕捉。
cs.CL / 82 / 2606.12342
ALIGNBEAM : Inference-Time Alignment Transfer via Cross-Vocabulary Logit Mixing
ALIGNBEAM：通过跨词汇对数混合实现推理时对齐转移
Abstract
Domain fine-tuning degrades the safety of large language models: fine-tuned specialists readily comply with harmful prompts framed in domain language. Existing inference-time defenses that mix logits from a safe anchor model require both models to share a vocabulary, which rules them out for the cross-family specialists where safety is most degraded. We present ALIGNBEAM, a training-free method that lifts this restriction by translating anchor logits into the target model's vocabulary token-by-token at each decoding step; a small LLM judge then selects the safest among K candidate continuations. No weights are changed, and the safety-utility trade-off can be tuned at deployment without retraining. Across both cross-vocabulary and same-vocabulary evaluation pairs, ALIGNBEAM substantially raises refusal on adversarial benchmarks while keeping task accuracy and inference overhead within practical bounds. The results show that safety alignment can be transferred between model families at inference time, without touching either model's weights.
Chinese Translation
领域微调降低了大型语言模型的安全性：经过微调的专业模型容易对以领域语言构建的有害提示做出响应。现有的推理时防御方法需要从安全锚模型混合对数，但要求两个模型共享词汇，这使得它们无法应用于安全性最差的跨家族专业模型。我们提出了ALIGNBEAM，这是一种无训练的方法，通过在每个解码步骤中逐词将锚对数翻译为目标模型的词汇，从而消除了这一限制；一个小型的LLM评判者随后在K个候选续写中选择最安全的选项。模型权重不发生变化，安全性与效用的权衡可以在部署时进行调节，而无需重新训练。在跨词汇和同词汇评估对中，ALIGNBEAM显著提高了对抗基准的拒绝率，同时保持了任务准确性和推理开销在实际范围内。结果表明，安全对齐可以在推理时在模型家族之间转移，而无需更改任何模型的权重。
cs.CL / 83 / 2606.12373
Verifiable Environments Are LEGO Bricks: Recursive Composition for Reasoning Generalization
可验证环境是乐高积木：用于推理泛化的递归组合
Abstract
Reinforcement Learning (RL) with verifiable environments has emerged as a powerful approach for enhancing the reasoning capabilities of Large Language Models (LLMs). While prior research demonstrates that scaling environment quantity improves RL performance, existing manual or individual construction methods suffer from linear scaling limits, thereby hindering scalable reasoning generalization. This paper introduces RACES (\textbf{R}ecursive \textbf{A}utomated \textbf{C}omposition for \textbf{E}nvironment \textbf{S}caling), a framework that conceptualizes verifiable environments as composable building blocks that can be recursively assembled. The key insight is that when the codomain (output type) of one environment matches the domain (input type) of another, they can be automatically fused into a new verifiable environment, enabling recursive composition. RACES is implemented with 300 individual environments and defines a set of composition operators (\textsc{SEQUENTIAL}, \textsc{PARALLEL}, \textsc{SORT}, and \textsc{SELECT}) that induce diverse reasoning patterns. Extensive experiments show that RL training on these composite environments consistently enhances reasoning generalization. Specifically, RACES improves DeepSeek-R1-Distill-Qwen-14B by an average of 3.1 points (from 48.2 to 51.3) and boosts Qwen3-14B performance from 58.8 to 61.1 on six benchmarks, which are unseen during the construction of training environments. Moreover, RACES achieves performance comparable to training on 300 individual environments using only 50 base environments, demonstrating significant efficiency in environment utilization.
Chinese Translation
使用可验证环境的强化学习（RL）已成为增强大型语言模型（LLMs）推理能力的一种有效方法。尽管先前的研究表明，环境数量的增加可以提升RL性能，但现有的手动或单独构建方法受到线性扩展限制，从而阻碍了可扩展的推理泛化。本文提出了RACES（ extbf{R}ecursive extbf{A}utomated extbf{C}omposition for extbf{E}nvironment extbf{S}caling）框架，将可验证环境概念化为可组合的构建块，可以递归组装。关键的见解是，当一个环境的余域（输出类型）与另一个环境的定义域（输入类型）匹配时，它们可以自动融合成一个新的可验证环境，从而实现递归组合。RACES实现了300个独立环境，并定义了一组组合运算符（ extsc{SEQUENTIAL}、 extsc{PARALLEL}、 extsc{SORT}和 extsc{SELECT}），以诱导多样的推理模式。大量实验表明，在这些复合环境上进行RL训练始终能够增强推理泛化。具体而言，RACES将DeepSeek-R1-Distill-Qwen-14B的平均分提高了3.1分（从48.2提升至51.3），并将Qwen3-14B的性能从58.8提升至61.1，在六个基准测试中，这些基准在训练环境构建时并未见过。此外，RACES在仅使用50个基础环境的情况下，实现了与在300个独立环境上训练相当的性能，显示出环境利用的显著效率。
cs.CL / 84 / 2606.12385
Which Models Are Our Models Built On? Auditing Invisible Dependencies in Modern LLMs
我们的模型建立在什么模型之上？审计现代大语言模型中的隐性依赖关系
Abstract
Modern LLM training pipelines increasingly rely on other models to generate data, filter corpora, judge outputs, and guide development decisions. These dependencies are recursive: a model may depend on an upstream artifact whose own dependencies are documented only in separate releases and artifacts. As a result, the full dependency structure is fragmented across heterogeneous public artifacts, with complexity and recursive depth far outpacing humans' ability to trace. We introduce ModSleuth, an agentic system that recursively reconstructs LLM dependency graphs from public artifacts with source-grounded evidence. We find that the primary challenge is no longer information extraction, but defining what constitutes a dependency and reconciling artifact references across inconsistent documentation. We address these challenges through a formalization that distinguishes direct and indirect dependencies, represents heterogeneous pipeline roles through operation-centered relationships, and resolves artifact identities across names, versions, and repositories. Applying ModSleuth to four public-artifact-rich LLM releases, we recover 1,060 source-verified dependencies and construct large-scale dependency graphs of modern LLM development. These graphs reveal multi-hop license obligations, train-evaluation coupling, discrepancies between released and training-time artifacts, and documentation inconsistencies that would otherwise be difficult to uncover. We release ModSleuth and the resulting dependency graphs to support transparent analysis of the increasingly complex ecosystems underlying modern LLMs.
Chinese Translation
现代大语言模型（LLM）的训练流程越来越依赖其他模型来生成数据、过滤语料、评估输出和指导开发决策。这些依赖关系是递归的：一个模型可能依赖于上游工件，而这些工件自身的依赖关系仅在单独的版本和工件中记录。因此，完整的依赖结构分散在异构的公共工件中，其复杂性和递归深度远远超出人类的追踪能力。我们提出了ModSleuth，一个能够从公共工件中递归重建LLM依赖图的自主系统，并提供源基础证据。我们发现，主要挑战不再是信息提取，而是定义什么构成依赖关系，以及在不一致的文档中调和工件引用。我们通过一种形式化方法来应对这些挑战，该方法区分直接和间接依赖关系，通过以操作为中心的关系表示异构管道角色，并解决跨名称、版本和仓库的工件身份问题。将ModSleuth应用于四个丰富公共工件的LLM发布版本，我们恢复了1,060个源验证的依赖关系，并构建了现代LLM开发的大规模依赖图。这些图揭示了多跳许可义务、训练与评估的耦合、发布与训练时工件之间的差异，以及其他难以发现的文档不一致性。我们发布了ModSleuth及其生成的依赖图，以支持对现代LLM日益复杂的生态系统进行透明分析。
cs.CL / 85 / 2606.12392
System Report for CCL25-Eval Task 5: New Dataset and LoRA-Fine-Tuned Qwen2.5
CCL25-Eval任务5的系统报告：新数据集与LoRA微调的Qwen2.5
Abstract
Recently, large language models (LLMs) have achieved promising progress in the fields of classical Chinese translation and the generation of classical poetry. However, domain-specific research on precise translation and affective-semantic understanding of classical poetry remains limited. The main challenge is that most studies treat the poetic appreciation task as a general-domain problem, neglecting the distinctive features of poetic appreciation, while high-quality and domain-specific datasets are extremely limited. To address this limitation, we decompose the task into three subtasks: term interpretation, semantic interpretation, and emotional inference. Based on multiple open-source datasets, we perform data cleansing and alignment to construct the Classical Chinese Poetry Instruction Pair Dataset (CCPoetry-49K), which comprises 49,404 high-quality instruction-response pairs explicitly optimized for this domain. We then propose a domain-specialized LLM, called PoetryQwen, by applying Low-Rank Adaptation (LoRA) to fine-tune the Qwen2.5-14B model. Experimental results on the CCL25-Eval Task 5 benchmark demonstrate that PoetryQwen achieves a score of 0.757, representing a 9.7% improvement over the Qwen2.5-14B-Instruct baseline (0.690). These findings clearly indicate that PoetryQwen significantly enhances performance in precise translation and emotional understanding of classical poetry. We present new dataset and methodological considerations intended to support the domain-specific optimization of LLMs.
Chinese Translation
近年来，大型语言模型（LLMs）在古典汉语翻译和古典诗歌生成领域取得了令人鼓舞的进展。然而，针对古典诗歌的精确翻译和情感语义理解的领域特定研究仍然有限。主要挑战在于大多数研究将诗歌欣赏任务视为一般领域问题，忽视了诗歌欣赏的独特特征，同时高质量和领域特定的数据集极为稀缺。为了解决这一局限性，我们将任务分解为三个子任务：术语解释、语义解释和情感推断。基于多个开源数据集，我们进行数据清洗和对齐，构建了古典汉语诗歌指令对数据集（CCPoetry-49K），该数据集包含49,404个明确针对该领域优化的高质量指令-响应对。随后，我们通过应用低秩适应（LoRA）对Qwen2.5-14B模型进行微调，提出了一种领域专用的LLM，称为PoetryQwen。在CCL25-Eval任务5基准测试中的实验结果表明，PoetryQwen的得分为0.757，相较于Qwen2.5-14B-Instruct基线（0.690）提高了9.7%。这些发现清楚地表明，PoetryQwen显著提升了古典诗歌的精确翻译和情感理解能力。我们提出了新的数据集和方法论考虑，旨在支持LLMs的领域特定优化。
cs.CL / 86 / 2606.12400
Doc-to-Atom: Learning to Compile and Compose Memory Atoms
文档到原子：学习编译和组合记忆原子
Abstract
Long input sequences are central to document understanding and multi-step reasoning in Large Language Models, yet the quadratic cost of attention makes inference both memory-intensive and slow. Context distillation mitigates this by compressing contextual information into model parameters, and recent work such as Doc-to-LoRA amortizes context distillation into a single forward pass that generates one LoRA adapter per document. However, producing a single monolithic adapter for all queries leads to irrelevant-query interference, limited compositional recall, and poor scalability to long-document reasoning. To address these challenges, we propose Doc-to-Atom (Doc2Atom), a compositional parametric memory framework that decomposes each document into semantically typed knowledge atoms. Each atom is compiled into an independent micro-LoRA adapter and a provenance retrieval key. At inference time, a lightweight query router selects and assembles only the relevant atoms into a query-specific adapter, which is then injected into a frozen base model. The entire system is trained end-to-end through a multi-objective distillation framework. Experiments on six diverse QA benchmarks demonstrate that Doc2Atom outperforms Doc-to-LoRA baselines while reducing the memory cost of document internalization.
Chinese Translation
长输入序列在大型语言模型的文档理解和多步骤推理中至关重要，但注意力机制的平方成本使得推理既消耗内存又缓慢。上下文蒸馏通过将上下文信息压缩到模型参数中来缓解这一问题，最近的工作如 Doc-to-LoRA 将上下文蒸馏摊销到一个前向传递中，为每个文档生成一个 LoRA 适配器。然而，为所有查询生成一个单一的整体适配器会导致无关查询干扰、有限的组合回忆能力以及在长文档推理中的可扩展性差。为了解决这些挑战，我们提出了 Doc-to-Atom（Doc2Atom），一个组合参数记忆框架，它将每个文档分解为语义类型的知识原子。每个原子被编译成一个独立的微型 LoRA 适配器和一个来源检索键。在推理时，一个轻量级查询路由器选择并组装仅相关的原子到一个特定于查询的适配器中，然后将其注入到一个冻结的基础模型中。整个系统通过多目标蒸馏框架进行端到端训练。在六个不同的问答基准上的实验表明，Doc2Atom 在性能上优于 Doc-to-LoRA 基线，同时减少了文档内化的内存成本。
cs.CL / 87 / 2606.12411
Context-Driven Incremental Compression for Multi-Turn Dialogue Generation
基于上下文的增量压缩用于多轮对话生成
Abstract
Modern conversational agents condition on an ever-growing dialogue history at each turn, incurring redundant attention and encoding costs that grow with conversation length. Naive truncation or summarization degrades fidelity, while existing context compressors lack cross-turn memory sharing or revision, causing information loss and compounding errors in long dialogues. We revisit the context compression under conversational dynamics and empirically present its fragility. To improve both efficiency and robustness, we introduce Context-Driven Incremental Compression (C-DIC), which treats a conversation as interleaved contextual threads and stores revisable per-thread compression states in a single, compact dialogue memory. At each turn, a lightweight retrieve, revise, and write-back loop shares information across turns and updates stale memories, stabilizing long-horizon behavior. In addition, we adapt truncated backpropagation-through-time (TBPTT) to our multi-turn setting, learning cross-turn dependencies without full-history backpropagation. Extensive experiments on long-form dialogue benchmarks demonstrate superior performance and efficiency of C-DIC; notably, C-DIC shows stable inference latency and perplexity over hundreds of dialogue turns, supporting a scalable path to high-quality dialogue modeling.
Chinese Translation
现代对话代理在每一轮中都依赖于不断增长的对话历史，这导致随着对话长度的增加，冗余的注意力和编码成本也随之增加。简单的截断或摘要会降低保真度，而现有的上下文压缩方法缺乏跨轮记忆共享或修正，导致信息丢失并在长对话中累积错误。我们重新审视了在对话动态下的上下文压缩，并实证展示了其脆弱性。为了提高效率和鲁棒性，我们提出了基于上下文的增量压缩（Context-Driven Incremental Compression, C-DIC），将对话视为交错的上下文线程，并在单一紧凑的对话记忆中存储可修正的每线程压缩状态。在每一轮中，轻量级的检索、修正和写回循环在轮与轮之间共享信息并更新过时的记忆，从而稳定长期行为。此外，我们将截断的时间反向传播（Truncated Backpropagation Through Time, TBPTT）适配到我们的多轮设置中，在不进行全历史反向传播的情况下学习跨轮依赖关系。在长篇对话基准上的广泛实验表明，C-DIC在性能和效率上均表现优越；特别是，C-DIC在数百轮对话中显示出稳定的推理延迟和困惑度，为高质量对话建模提供了可扩展的路径。